{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.665671482646479, "eval_steps": 500, "global_step": 50000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 5.331342965292957e-05, "grad_norm": 8.960769653320312, "learning_rate": 1e-05, "loss": 1.4485, "step": 1 }, { "epoch": 0.00010662685930585914, "grad_norm": 9.378430366516113, "learning_rate": 9.9998e-06, "loss": 1.417, "step": 2 }, { "epoch": 0.00015994028895878872, "grad_norm": 8.152729034423828, "learning_rate": 9.9996e-06, "loss": 1.3459, "step": 3 }, { "epoch": 0.00021325371861171828, "grad_norm": 7.727672100067139, "learning_rate": 9.999400000000001e-06, "loss": 1.3083, "step": 4 }, { "epoch": 0.00026656714826464784, "grad_norm": 5.791541576385498, "learning_rate": 9.9992e-06, "loss": 1.1531, "step": 5 }, { "epoch": 0.00031988057791757743, "grad_norm": 5.233710765838623, "learning_rate": 9.999e-06, "loss": 1.0865, "step": 6 }, { "epoch": 0.000373194007570507, "grad_norm": 3.5417187213897705, "learning_rate": 9.998800000000002e-06, "loss": 1.1122, "step": 7 }, { "epoch": 0.00042650743722343656, "grad_norm": 2.2198212146759033, "learning_rate": 9.998600000000001e-06, "loss": 0.9976, "step": 8 }, { "epoch": 0.00047982086687636615, "grad_norm": 1.1936628818511963, "learning_rate": 9.9984e-06, "loss": 0.978, "step": 9 }, { "epoch": 0.0005331342965292957, "grad_norm": 2.9824702739715576, "learning_rate": 9.998200000000002e-06, "loss": 1.0291, "step": 10 }, { "epoch": 0.0005864477261822253, "grad_norm": 2.3760929107666016, "learning_rate": 9.998000000000002e-06, "loss": 1.001, "step": 11 }, { "epoch": 0.0006397611558351549, "grad_norm": 1.3715907335281372, "learning_rate": 9.997800000000001e-06, "loss": 0.9985, "step": 12 }, { "epoch": 0.0006930745854880845, "grad_norm": 1.9275307655334473, "learning_rate": 9.9976e-06, "loss": 1.0107, "step": 13 }, { "epoch": 0.000746388015141014, "grad_norm": 0.6184626817703247, "learning_rate": 9.9974e-06, "loss": 0.9656, "step": 14 }, { "epoch": 0.0007997014447939436, "grad_norm": 0.7435579895973206, "learning_rate": 9.997200000000001e-06, "loss": 0.9896, "step": 15 }, { "epoch": 0.0008530148744468731, "grad_norm": 1.1701269149780273, "learning_rate": 9.997000000000001e-06, "loss": 0.9028, "step": 16 }, { "epoch": 0.0009063283040998027, "grad_norm": 2.3373043537139893, "learning_rate": 9.9968e-06, "loss": 0.8719, "step": 17 }, { "epoch": 0.0009596417337527323, "grad_norm": 2.0897562503814697, "learning_rate": 9.9966e-06, "loss": 1.0126, "step": 18 }, { "epoch": 0.001012955163405662, "grad_norm": 0.944312572479248, "learning_rate": 9.996400000000001e-06, "loss": 0.9561, "step": 19 }, { "epoch": 0.0010662685930585914, "grad_norm": 1.821272373199463, "learning_rate": 9.996200000000001e-06, "loss": 1.0148, "step": 20 }, { "epoch": 0.001119582022711521, "grad_norm": 0.6787191033363342, "learning_rate": 9.996e-06, "loss": 0.9258, "step": 21 }, { "epoch": 0.0011728954523644506, "grad_norm": 0.7113598585128784, "learning_rate": 9.995800000000002e-06, "loss": 0.9136, "step": 22 }, { "epoch": 0.0012262088820173803, "grad_norm": 1.029117226600647, "learning_rate": 9.995600000000001e-06, "loss": 0.9282, "step": 23 }, { "epoch": 0.0012795223116703097, "grad_norm": 2.274350643157959, "learning_rate": 9.995400000000001e-06, "loss": 0.9684, "step": 24 }, { "epoch": 0.0013328357413232392, "grad_norm": 2.092135190963745, "learning_rate": 9.9952e-06, "loss": 0.9464, "step": 25 }, { "epoch": 0.001386149170976169, "grad_norm": 1.0660781860351562, "learning_rate": 9.995000000000002e-06, "loss": 0.9144, "step": 26 }, { "epoch": 0.0014394626006290984, "grad_norm": 0.9635916948318481, "learning_rate": 9.994800000000001e-06, "loss": 0.9579, "step": 27 }, { "epoch": 0.001492776030282028, "grad_norm": 1.215332269668579, "learning_rate": 9.9946e-06, "loss": 0.9614, "step": 28 }, { "epoch": 0.0015460894599349576, "grad_norm": 1.0763520002365112, "learning_rate": 9.9944e-06, "loss": 0.9288, "step": 29 }, { "epoch": 0.0015994028895878873, "grad_norm": 1.9046753644943237, "learning_rate": 9.9942e-06, "loss": 0.8972, "step": 30 }, { "epoch": 0.0016527163192408168, "grad_norm": 1.1674866676330566, "learning_rate": 9.994000000000001e-06, "loss": 1.0262, "step": 31 }, { "epoch": 0.0017060297488937462, "grad_norm": 1.286513090133667, "learning_rate": 9.9938e-06, "loss": 0.9773, "step": 32 }, { "epoch": 0.001759343178546676, "grad_norm": 1.7154263257980347, "learning_rate": 9.9936e-06, "loss": 0.8848, "step": 33 }, { "epoch": 0.0018126566081996054, "grad_norm": 1.096118688583374, "learning_rate": 9.993400000000002e-06, "loss": 0.9186, "step": 34 }, { "epoch": 0.0018659700378525351, "grad_norm": 1.696483850479126, "learning_rate": 9.993200000000001e-06, "loss": 0.8531, "step": 35 }, { "epoch": 0.0019192834675054646, "grad_norm": 1.487643837928772, "learning_rate": 9.993e-06, "loss": 0.9138, "step": 36 }, { "epoch": 0.001972596897158394, "grad_norm": 1.266534686088562, "learning_rate": 9.9928e-06, "loss": 0.8472, "step": 37 }, { "epoch": 0.002025910326811324, "grad_norm": 2.731980085372925, "learning_rate": 9.992600000000002e-06, "loss": 0.9778, "step": 38 }, { "epoch": 0.0020792237564642535, "grad_norm": 2.323108434677124, "learning_rate": 9.992400000000001e-06, "loss": 0.8687, "step": 39 }, { "epoch": 0.0021325371861171828, "grad_norm": 2.7741963863372803, "learning_rate": 9.9922e-06, "loss": 0.8156, "step": 40 }, { "epoch": 0.0021858506157701124, "grad_norm": 5.107997894287109, "learning_rate": 9.992e-06, "loss": 0.9294, "step": 41 }, { "epoch": 0.002239164045423042, "grad_norm": 1.627322793006897, "learning_rate": 9.9918e-06, "loss": 0.8756, "step": 42 }, { "epoch": 0.002292477475075972, "grad_norm": 4.631983280181885, "learning_rate": 9.991600000000001e-06, "loss": 0.8113, "step": 43 }, { "epoch": 0.002345790904728901, "grad_norm": 3.424898386001587, "learning_rate": 9.9914e-06, "loss": 0.8513, "step": 44 }, { "epoch": 0.002399104334381831, "grad_norm": 7.246274948120117, "learning_rate": 9.9912e-06, "loss": 0.8439, "step": 45 }, { "epoch": 0.0024524177640347605, "grad_norm": 2.228410243988037, "learning_rate": 9.991000000000001e-06, "loss": 0.8296, "step": 46 }, { "epoch": 0.0025057311936876898, "grad_norm": 1.6254878044128418, "learning_rate": 9.990800000000001e-06, "loss": 0.8226, "step": 47 }, { "epoch": 0.0025590446233406195, "grad_norm": 7.365884780883789, "learning_rate": 9.9906e-06, "loss": 0.874, "step": 48 }, { "epoch": 0.002612358052993549, "grad_norm": 2.186091661453247, "learning_rate": 9.990400000000002e-06, "loss": 0.8431, "step": 49 }, { "epoch": 0.0026656714826464784, "grad_norm": 2.8249261379241943, "learning_rate": 9.990200000000001e-06, "loss": 0.8146, "step": 50 }, { "epoch": 0.002718984912299408, "grad_norm": 3.9206881523132324, "learning_rate": 9.990000000000001e-06, "loss": 0.8085, "step": 51 }, { "epoch": 0.002772298341952338, "grad_norm": 2.595458984375, "learning_rate": 9.9898e-06, "loss": 0.7847, "step": 52 }, { "epoch": 0.0028256117716052675, "grad_norm": 4.955793857574463, "learning_rate": 9.9896e-06, "loss": 0.8065, "step": 53 }, { "epoch": 0.002878925201258197, "grad_norm": 5.257761478424072, "learning_rate": 9.989400000000001e-06, "loss": 0.8646, "step": 54 }, { "epoch": 0.0029322386309111265, "grad_norm": 11.239571571350098, "learning_rate": 9.989200000000001e-06, "loss": 0.8439, "step": 55 }, { "epoch": 0.002985552060564056, "grad_norm": 5.9935784339904785, "learning_rate": 9.989e-06, "loss": 0.7861, "step": 56 }, { "epoch": 0.0030388654902169855, "grad_norm": 2.360766887664795, "learning_rate": 9.9888e-06, "loss": 0.7275, "step": 57 }, { "epoch": 0.003092178919869915, "grad_norm": 15.1712646484375, "learning_rate": 9.988600000000001e-06, "loss": 0.8671, "step": 58 }, { "epoch": 0.003145492349522845, "grad_norm": 10.194405555725098, "learning_rate": 9.9884e-06, "loss": 0.7985, "step": 59 }, { "epoch": 0.0031988057791757746, "grad_norm": 5.825827121734619, "learning_rate": 9.988200000000002e-06, "loss": 0.7161, "step": 60 }, { "epoch": 0.003252119208828704, "grad_norm": 14.847115516662598, "learning_rate": 9.988000000000002e-06, "loss": 0.7838, "step": 61 }, { "epoch": 0.0033054326384816335, "grad_norm": 8.92076587677002, "learning_rate": 9.987800000000001e-06, "loss": 0.7372, "step": 62 }, { "epoch": 0.0033587460681345632, "grad_norm": 2.9107677936553955, "learning_rate": 9.9876e-06, "loss": 0.7225, "step": 63 }, { "epoch": 0.0034120594977874925, "grad_norm": 11.428593635559082, "learning_rate": 9.9874e-06, "loss": 0.7781, "step": 64 }, { "epoch": 0.003465372927440422, "grad_norm": 11.074180603027344, "learning_rate": 9.987200000000002e-06, "loss": 0.7349, "step": 65 }, { "epoch": 0.003518686357093352, "grad_norm": 5.752894401550293, "learning_rate": 9.987000000000001e-06, "loss": 0.7396, "step": 66 }, { "epoch": 0.0035719997867462816, "grad_norm": 2.738863945007324, "learning_rate": 9.9868e-06, "loss": 0.6911, "step": 67 }, { "epoch": 0.003625313216399211, "grad_norm": 15.88848876953125, "learning_rate": 9.9866e-06, "loss": 0.7715, "step": 68 }, { "epoch": 0.0036786266460521405, "grad_norm": 6.2538743019104, "learning_rate": 9.9864e-06, "loss": 0.7336, "step": 69 }, { "epoch": 0.0037319400757050702, "grad_norm": 3.0735692977905273, "learning_rate": 9.986200000000001e-06, "loss": 0.6523, "step": 70 }, { "epoch": 0.0037852535053579995, "grad_norm": 7.122920036315918, "learning_rate": 9.986e-06, "loss": 0.7149, "step": 71 }, { "epoch": 0.003838566935010929, "grad_norm": 8.70314884185791, "learning_rate": 9.985800000000002e-06, "loss": 0.6115, "step": 72 }, { "epoch": 0.003891880364663859, "grad_norm": 6.233135223388672, "learning_rate": 9.985600000000001e-06, "loss": 0.679, "step": 73 }, { "epoch": 0.003945193794316788, "grad_norm": 9.093994140625, "learning_rate": 9.985400000000001e-06, "loss": 0.6084, "step": 74 }, { "epoch": 0.003998507223969718, "grad_norm": 9.739960670471191, "learning_rate": 9.9852e-06, "loss": 0.6675, "step": 75 }, { "epoch": 0.004051820653622648, "grad_norm": 9.097187042236328, "learning_rate": 9.985000000000002e-06, "loss": 0.6245, "step": 76 }, { "epoch": 0.004105134083275577, "grad_norm": 9.99565315246582, "learning_rate": 9.984800000000001e-06, "loss": 0.6329, "step": 77 }, { "epoch": 0.004158447512928507, "grad_norm": 20.56688117980957, "learning_rate": 9.984600000000001e-06, "loss": 0.7286, "step": 78 }, { "epoch": 0.004211760942581437, "grad_norm": 3.4001340866088867, "learning_rate": 9.9844e-06, "loss": 0.6397, "step": 79 }, { "epoch": 0.0042650743722343655, "grad_norm": 10.634171485900879, "learning_rate": 9.9842e-06, "loss": 0.6199, "step": 80 }, { "epoch": 0.004318387801887295, "grad_norm": 9.930357933044434, "learning_rate": 9.984e-06, "loss": 0.6079, "step": 81 }, { "epoch": 0.004371701231540225, "grad_norm": 3.820631265640259, "learning_rate": 9.983800000000001e-06, "loss": 0.585, "step": 82 }, { "epoch": 0.004425014661193155, "grad_norm": 5.867438316345215, "learning_rate": 9.9836e-06, "loss": 0.5701, "step": 83 }, { "epoch": 0.004478328090846084, "grad_norm": 10.247586250305176, "learning_rate": 9.983400000000002e-06, "loss": 0.5618, "step": 84 }, { "epoch": 0.004531641520499014, "grad_norm": 11.052252769470215, "learning_rate": 9.983200000000001e-06, "loss": 0.5802, "step": 85 }, { "epoch": 0.004584954950151944, "grad_norm": 4.375123977661133, "learning_rate": 9.983e-06, "loss": 0.5104, "step": 86 }, { "epoch": 0.0046382683798048725, "grad_norm": 9.940057754516602, "learning_rate": 9.9828e-06, "loss": 0.5716, "step": 87 }, { "epoch": 0.004691581809457802, "grad_norm": 3.6178975105285645, "learning_rate": 9.982600000000002e-06, "loss": 0.5241, "step": 88 }, { "epoch": 0.004744895239110732, "grad_norm": 9.060243606567383, "learning_rate": 9.982400000000001e-06, "loss": 0.5055, "step": 89 }, { "epoch": 0.004798208668763662, "grad_norm": 6.420359134674072, "learning_rate": 9.9822e-06, "loss": 0.5212, "step": 90 }, { "epoch": 0.004851522098416591, "grad_norm": 4.6098952293396, "learning_rate": 9.982e-06, "loss": 0.5353, "step": 91 }, { "epoch": 0.004904835528069521, "grad_norm": 3.6437056064605713, "learning_rate": 9.9818e-06, "loss": 0.5139, "step": 92 }, { "epoch": 0.00495814895772245, "grad_norm": 6.449934482574463, "learning_rate": 9.981600000000001e-06, "loss": 0.4983, "step": 93 }, { "epoch": 0.0050114623873753796, "grad_norm": 10.350210189819336, "learning_rate": 9.9814e-06, "loss": 0.5295, "step": 94 }, { "epoch": 0.005064775817028309, "grad_norm": 7.641249656677246, "learning_rate": 9.9812e-06, "loss": 0.5181, "step": 95 }, { "epoch": 0.005118089246681239, "grad_norm": 4.831916332244873, "learning_rate": 9.981000000000002e-06, "loss": 0.4736, "step": 96 }, { "epoch": 0.005171402676334169, "grad_norm": 5.021637916564941, "learning_rate": 9.980800000000001e-06, "loss": 0.5544, "step": 97 }, { "epoch": 0.005224716105987098, "grad_norm": 8.085831642150879, "learning_rate": 9.9806e-06, "loss": 0.5212, "step": 98 }, { "epoch": 0.005278029535640028, "grad_norm": 9.884246826171875, "learning_rate": 9.980400000000002e-06, "loss": 0.453, "step": 99 }, { "epoch": 0.005331342965292957, "grad_norm": 3.568842649459839, "learning_rate": 9.980200000000001e-06, "loss": 0.4933, "step": 100 }, { "epoch": 0.005384656394945887, "grad_norm": 11.712270736694336, "learning_rate": 9.980000000000001e-06, "loss": 0.5194, "step": 101 }, { "epoch": 0.005437969824598816, "grad_norm": 10.94614315032959, "learning_rate": 9.9798e-06, "loss": 0.501, "step": 102 }, { "epoch": 0.005491283254251746, "grad_norm": 8.019649505615234, "learning_rate": 9.9796e-06, "loss": 0.5267, "step": 103 }, { "epoch": 0.005544596683904676, "grad_norm": 9.073199272155762, "learning_rate": 9.979400000000001e-06, "loss": 0.4547, "step": 104 }, { "epoch": 0.005597910113557605, "grad_norm": 8.235196113586426, "learning_rate": 9.979200000000001e-06, "loss": 0.4296, "step": 105 }, { "epoch": 0.005651223543210535, "grad_norm": 13.54327392578125, "learning_rate": 9.979e-06, "loss": 0.4431, "step": 106 }, { "epoch": 0.005704536972863464, "grad_norm": 8.48166561126709, "learning_rate": 9.9788e-06, "loss": 0.4432, "step": 107 }, { "epoch": 0.005757850402516394, "grad_norm": 12.27595329284668, "learning_rate": 9.978600000000001e-06, "loss": 0.4396, "step": 108 }, { "epoch": 0.005811163832169323, "grad_norm": 7.436861038208008, "learning_rate": 9.978400000000001e-06, "loss": 0.4012, "step": 109 }, { "epoch": 0.005864477261822253, "grad_norm": 28.089189529418945, "learning_rate": 9.9782e-06, "loss": 0.5226, "step": 110 }, { "epoch": 0.005917790691475183, "grad_norm": 7.074982643127441, "learning_rate": 9.978000000000002e-06, "loss": 0.4331, "step": 111 }, { "epoch": 0.005971104121128112, "grad_norm": 9.185254096984863, "learning_rate": 9.977800000000001e-06, "loss": 0.4612, "step": 112 }, { "epoch": 0.006024417550781042, "grad_norm": 26.553255081176758, "learning_rate": 9.9776e-06, "loss": 0.4015, "step": 113 }, { "epoch": 0.006077730980433971, "grad_norm": 8.240938186645508, "learning_rate": 9.9774e-06, "loss": 0.3384, "step": 114 }, { "epoch": 0.006131044410086901, "grad_norm": 36.16035842895508, "learning_rate": 9.977200000000002e-06, "loss": 0.5062, "step": 115 }, { "epoch": 0.00618435783973983, "grad_norm": 15.88504409790039, "learning_rate": 9.977000000000001e-06, "loss": 0.43, "step": 116 }, { "epoch": 0.00623767126939276, "grad_norm": 20.842378616333008, "learning_rate": 9.9768e-06, "loss": 0.3949, "step": 117 }, { "epoch": 0.00629098469904569, "grad_norm": 25.730052947998047, "learning_rate": 9.9766e-06, "loss": 0.5082, "step": 118 }, { "epoch": 0.006344298128698619, "grad_norm": 19.322406768798828, "learning_rate": 9.9764e-06, "loss": 0.4439, "step": 119 }, { "epoch": 0.006397611558351549, "grad_norm": 26.656232833862305, "learning_rate": 9.976200000000001e-06, "loss": 0.5222, "step": 120 }, { "epoch": 0.006450924988004478, "grad_norm": 31.868391036987305, "learning_rate": 9.976e-06, "loss": 0.4962, "step": 121 }, { "epoch": 0.006504238417657408, "grad_norm": 6.350840091705322, "learning_rate": 9.9758e-06, "loss": 0.4125, "step": 122 }, { "epoch": 0.006557551847310337, "grad_norm": 17.68107795715332, "learning_rate": 9.975600000000002e-06, "loss": 0.476, "step": 123 }, { "epoch": 0.006610865276963267, "grad_norm": 22.557373046875, "learning_rate": 9.975400000000001e-06, "loss": 0.5437, "step": 124 }, { "epoch": 0.006664178706616197, "grad_norm": 15.382554054260254, "learning_rate": 9.9752e-06, "loss": 0.4461, "step": 125 }, { "epoch": 0.0067174921362691264, "grad_norm": 10.060270309448242, "learning_rate": 9.975000000000002e-06, "loss": 0.4479, "step": 126 }, { "epoch": 0.006770805565922056, "grad_norm": 15.5154390335083, "learning_rate": 9.974800000000001e-06, "loss": 0.4106, "step": 127 }, { "epoch": 0.006824118995574985, "grad_norm": 16.320899963378906, "learning_rate": 9.974600000000001e-06, "loss": 0.4388, "step": 128 }, { "epoch": 0.006877432425227915, "grad_norm": 12.772472381591797, "learning_rate": 9.9744e-06, "loss": 0.4343, "step": 129 }, { "epoch": 0.006930745854880844, "grad_norm": 9.817267417907715, "learning_rate": 9.9742e-06, "loss": 0.4247, "step": 130 }, { "epoch": 0.006984059284533774, "grad_norm": 12.083002090454102, "learning_rate": 9.974e-06, "loss": 0.344, "step": 131 }, { "epoch": 0.007037372714186704, "grad_norm": 5.698822975158691, "learning_rate": 9.973800000000001e-06, "loss": 0.3991, "step": 132 }, { "epoch": 0.0070906861438396335, "grad_norm": 7.929882049560547, "learning_rate": 9.9736e-06, "loss": 0.3587, "step": 133 }, { "epoch": 0.007143999573492563, "grad_norm": 10.856049537658691, "learning_rate": 9.9734e-06, "loss": 0.4469, "step": 134 }, { "epoch": 0.007197313003145492, "grad_norm": 10.964094161987305, "learning_rate": 9.973200000000001e-06, "loss": 0.3968, "step": 135 }, { "epoch": 0.007250626432798422, "grad_norm": 5.751630783081055, "learning_rate": 9.973000000000001e-06, "loss": 0.3669, "step": 136 }, { "epoch": 0.007303939862451351, "grad_norm": 5.3411335945129395, "learning_rate": 9.972800000000002e-06, "loss": 0.4006, "step": 137 }, { "epoch": 0.007357253292104281, "grad_norm": 4.073331832885742, "learning_rate": 9.972600000000002e-06, "loss": 0.3728, "step": 138 }, { "epoch": 0.007410566721757211, "grad_norm": 8.653517723083496, "learning_rate": 9.972400000000001e-06, "loss": 0.4002, "step": 139 }, { "epoch": 0.0074638801514101405, "grad_norm": 11.276178359985352, "learning_rate": 9.9722e-06, "loss": 0.4226, "step": 140 }, { "epoch": 0.00751719358106307, "grad_norm": 4.543489933013916, "learning_rate": 9.972e-06, "loss": 0.3551, "step": 141 }, { "epoch": 0.007570507010715999, "grad_norm": 6.407944679260254, "learning_rate": 9.9718e-06, "loss": 0.3796, "step": 142 }, { "epoch": 0.007623820440368929, "grad_norm": 6.27612829208374, "learning_rate": 9.971600000000001e-06, "loss": 0.386, "step": 143 }, { "epoch": 0.007677133870021858, "grad_norm": 5.771403789520264, "learning_rate": 9.9714e-06, "loss": 0.3407, "step": 144 }, { "epoch": 0.007730447299674788, "grad_norm": 5.066903114318848, "learning_rate": 9.9712e-06, "loss": 0.349, "step": 145 }, { "epoch": 0.007783760729327718, "grad_norm": 15.177400588989258, "learning_rate": 9.971e-06, "loss": 0.3468, "step": 146 }, { "epoch": 0.007837074158980648, "grad_norm": 4.856639385223389, "learning_rate": 9.970800000000001e-06, "loss": 0.3618, "step": 147 }, { "epoch": 0.007890387588633576, "grad_norm": 6.517629623413086, "learning_rate": 9.9706e-06, "loss": 0.3717, "step": 148 }, { "epoch": 0.007943701018286507, "grad_norm": 10.877967834472656, "learning_rate": 9.970400000000002e-06, "loss": 0.3599, "step": 149 }, { "epoch": 0.007997014447939436, "grad_norm": 17.01068878173828, "learning_rate": 9.970200000000002e-06, "loss": 0.4262, "step": 150 }, { "epoch": 0.008050327877592366, "grad_norm": 6.703619480133057, "learning_rate": 9.970000000000001e-06, "loss": 0.3305, "step": 151 }, { "epoch": 0.008103641307245295, "grad_norm": 5.432844161987305, "learning_rate": 9.9698e-06, "loss": 0.3546, "step": 152 }, { "epoch": 0.008156954736898224, "grad_norm": 6.486485004425049, "learning_rate": 9.9696e-06, "loss": 0.3193, "step": 153 }, { "epoch": 0.008210268166551155, "grad_norm": 6.453185081481934, "learning_rate": 9.969400000000001e-06, "loss": 0.3437, "step": 154 }, { "epoch": 0.008263581596204083, "grad_norm": 9.767376899719238, "learning_rate": 9.969200000000001e-06, "loss": 0.3554, "step": 155 }, { "epoch": 0.008316895025857014, "grad_norm": 7.990174293518066, "learning_rate": 9.969e-06, "loss": 0.3295, "step": 156 }, { "epoch": 0.008370208455509943, "grad_norm": 10.181550979614258, "learning_rate": 9.9688e-06, "loss": 0.3848, "step": 157 }, { "epoch": 0.008423521885162873, "grad_norm": 11.115480422973633, "learning_rate": 9.9686e-06, "loss": 0.3277, "step": 158 }, { "epoch": 0.008476835314815802, "grad_norm": 6.343972206115723, "learning_rate": 9.968400000000001e-06, "loss": 0.3668, "step": 159 }, { "epoch": 0.008530148744468731, "grad_norm": 9.042622566223145, "learning_rate": 9.9682e-06, "loss": 0.3492, "step": 160 }, { "epoch": 0.008583462174121662, "grad_norm": 8.00826358795166, "learning_rate": 9.968000000000002e-06, "loss": 0.3697, "step": 161 }, { "epoch": 0.00863677560377459, "grad_norm": 4.4533772468566895, "learning_rate": 9.967800000000001e-06, "loss": 0.3032, "step": 162 }, { "epoch": 0.008690089033427521, "grad_norm": 12.264516830444336, "learning_rate": 9.967600000000001e-06, "loss": 0.3026, "step": 163 }, { "epoch": 0.00874340246308045, "grad_norm": 17.262577056884766, "learning_rate": 9.9674e-06, "loss": 0.3703, "step": 164 }, { "epoch": 0.00879671589273338, "grad_norm": 15.658928871154785, "learning_rate": 9.967200000000002e-06, "loss": 0.2972, "step": 165 }, { "epoch": 0.00885002932238631, "grad_norm": 9.16011905670166, "learning_rate": 9.967000000000001e-06, "loss": 0.3289, "step": 166 }, { "epoch": 0.008903342752039238, "grad_norm": 6.221110820770264, "learning_rate": 9.966800000000001e-06, "loss": 0.3208, "step": 167 }, { "epoch": 0.008956656181692169, "grad_norm": 17.28338623046875, "learning_rate": 9.9666e-06, "loss": 0.3229, "step": 168 }, { "epoch": 0.009009969611345097, "grad_norm": 12.275278091430664, "learning_rate": 9.9664e-06, "loss": 0.3519, "step": 169 }, { "epoch": 0.009063283040998028, "grad_norm": 5.8157429695129395, "learning_rate": 9.9662e-06, "loss": 0.3457, "step": 170 }, { "epoch": 0.009116596470650957, "grad_norm": 7.837258338928223, "learning_rate": 9.966e-06, "loss": 0.312, "step": 171 }, { "epoch": 0.009169909900303887, "grad_norm": 9.54043197631836, "learning_rate": 9.9658e-06, "loss": 0.3575, "step": 172 }, { "epoch": 0.009223223329956816, "grad_norm": 8.46560001373291, "learning_rate": 9.965600000000002e-06, "loss": 0.3679, "step": 173 }, { "epoch": 0.009276536759609745, "grad_norm": 13.822833061218262, "learning_rate": 9.965400000000001e-06, "loss": 0.3798, "step": 174 }, { "epoch": 0.009329850189262676, "grad_norm": 19.51679801940918, "learning_rate": 9.9652e-06, "loss": 0.3891, "step": 175 }, { "epoch": 0.009383163618915604, "grad_norm": 7.3764495849609375, "learning_rate": 9.965000000000002e-06, "loss": 0.309, "step": 176 }, { "epoch": 0.009436477048568535, "grad_norm": 21.531034469604492, "learning_rate": 9.964800000000002e-06, "loss": 0.309, "step": 177 }, { "epoch": 0.009489790478221464, "grad_norm": 7.089829444885254, "learning_rate": 9.964600000000001e-06, "loss": 0.3275, "step": 178 }, { "epoch": 0.009543103907874394, "grad_norm": 17.899534225463867, "learning_rate": 9.9644e-06, "loss": 0.3392, "step": 179 }, { "epoch": 0.009596417337527323, "grad_norm": 4.702307224273682, "learning_rate": 9.9642e-06, "loss": 0.3006, "step": 180 }, { "epoch": 0.009649730767180252, "grad_norm": 11.620116233825684, "learning_rate": 9.964e-06, "loss": 0.2623, "step": 181 }, { "epoch": 0.009703044196833183, "grad_norm": 16.661701202392578, "learning_rate": 9.963800000000001e-06, "loss": 0.3625, "step": 182 }, { "epoch": 0.009756357626486111, "grad_norm": 8.764348983764648, "learning_rate": 9.9636e-06, "loss": 0.2974, "step": 183 }, { "epoch": 0.009809671056139042, "grad_norm": 10.701285362243652, "learning_rate": 9.9634e-06, "loss": 0.3403, "step": 184 }, { "epoch": 0.009862984485791971, "grad_norm": 11.372420310974121, "learning_rate": 9.963200000000001e-06, "loss": 0.2948, "step": 185 }, { "epoch": 0.0099162979154449, "grad_norm": 23.542570114135742, "learning_rate": 9.963000000000001e-06, "loss": 0.3492, "step": 186 }, { "epoch": 0.00996961134509783, "grad_norm": 11.822953224182129, "learning_rate": 9.9628e-06, "loss": 0.3019, "step": 187 }, { "epoch": 0.010022924774750759, "grad_norm": 15.600472450256348, "learning_rate": 9.962600000000002e-06, "loss": 0.3353, "step": 188 }, { "epoch": 0.01007623820440369, "grad_norm": 11.266243934631348, "learning_rate": 9.962400000000001e-06, "loss": 0.2935, "step": 189 }, { "epoch": 0.010129551634056619, "grad_norm": 14.055030822753906, "learning_rate": 9.962200000000001e-06, "loss": 0.2972, "step": 190 }, { "epoch": 0.010182865063709549, "grad_norm": 14.92376708984375, "learning_rate": 9.962e-06, "loss": 0.3599, "step": 191 }, { "epoch": 0.010236178493362478, "grad_norm": 9.725488662719727, "learning_rate": 9.9618e-06, "loss": 0.2677, "step": 192 }, { "epoch": 0.010289491923015407, "grad_norm": 5.820506572723389, "learning_rate": 9.961600000000001e-06, "loss": 0.26, "step": 193 }, { "epoch": 0.010342805352668337, "grad_norm": 26.39034080505371, "learning_rate": 9.961400000000001e-06, "loss": 0.3613, "step": 194 }, { "epoch": 0.010396118782321266, "grad_norm": 11.661584854125977, "learning_rate": 9.9612e-06, "loss": 0.3292, "step": 195 }, { "epoch": 0.010449432211974197, "grad_norm": 47.88852310180664, "learning_rate": 9.961e-06, "loss": 0.4494, "step": 196 }, { "epoch": 0.010502745641627126, "grad_norm": 8.1890230178833, "learning_rate": 9.960800000000001e-06, "loss": 0.2756, "step": 197 }, { "epoch": 0.010556059071280056, "grad_norm": 16.66854476928711, "learning_rate": 9.9606e-06, "loss": 0.337, "step": 198 }, { "epoch": 0.010609372500932985, "grad_norm": 15.370323181152344, "learning_rate": 9.9604e-06, "loss": 0.324, "step": 199 }, { "epoch": 0.010662685930585914, "grad_norm": 7.980715274810791, "learning_rate": 9.960200000000002e-06, "loss": 0.3066, "step": 200 }, { "epoch": 0.010715999360238844, "grad_norm": 21.35443878173828, "learning_rate": 9.960000000000001e-06, "loss": 0.3258, "step": 201 }, { "epoch": 0.010769312789891773, "grad_norm": 15.11815357208252, "learning_rate": 9.9598e-06, "loss": 0.3219, "step": 202 }, { "epoch": 0.010822626219544704, "grad_norm": 4.536488056182861, "learning_rate": 9.9596e-06, "loss": 0.2729, "step": 203 }, { "epoch": 0.010875939649197633, "grad_norm": 12.71891975402832, "learning_rate": 9.959400000000002e-06, "loss": 0.2857, "step": 204 }, { "epoch": 0.010929253078850563, "grad_norm": 10.881317138671875, "learning_rate": 9.959200000000001e-06, "loss": 0.2884, "step": 205 }, { "epoch": 0.010982566508503492, "grad_norm": 7.440808296203613, "learning_rate": 9.959e-06, "loss": 0.3221, "step": 206 }, { "epoch": 0.01103587993815642, "grad_norm": 10.445682525634766, "learning_rate": 9.9588e-06, "loss": 0.3192, "step": 207 }, { "epoch": 0.011089193367809351, "grad_norm": 11.490154266357422, "learning_rate": 9.9586e-06, "loss": 0.2948, "step": 208 }, { "epoch": 0.01114250679746228, "grad_norm": 10.946991920471191, "learning_rate": 9.958400000000001e-06, "loss": 0.2976, "step": 209 }, { "epoch": 0.01119582022711521, "grad_norm": 13.7036771774292, "learning_rate": 9.9582e-06, "loss": 0.2878, "step": 210 }, { "epoch": 0.01124913365676814, "grad_norm": 8.218505859375, "learning_rate": 9.958e-06, "loss": 0.2628, "step": 211 }, { "epoch": 0.01130244708642107, "grad_norm": 11.552397727966309, "learning_rate": 9.957800000000001e-06, "loss": 0.3118, "step": 212 }, { "epoch": 0.011355760516073999, "grad_norm": 17.38397789001465, "learning_rate": 9.957600000000001e-06, "loss": 0.3114, "step": 213 }, { "epoch": 0.011409073945726928, "grad_norm": 6.5189971923828125, "learning_rate": 9.9574e-06, "loss": 0.3111, "step": 214 }, { "epoch": 0.011462387375379858, "grad_norm": 13.340921401977539, "learning_rate": 9.957200000000002e-06, "loss": 0.2901, "step": 215 }, { "epoch": 0.011515700805032787, "grad_norm": 8.023892402648926, "learning_rate": 9.957000000000001e-06, "loss": 0.3183, "step": 216 }, { "epoch": 0.011569014234685718, "grad_norm": 4.147246837615967, "learning_rate": 9.956800000000001e-06, "loss": 0.2591, "step": 217 }, { "epoch": 0.011622327664338647, "grad_norm": 13.035296440124512, "learning_rate": 9.9566e-06, "loss": 0.2681, "step": 218 }, { "epoch": 0.011675641093991577, "grad_norm": 5.66063117980957, "learning_rate": 9.9564e-06, "loss": 0.2512, "step": 219 }, { "epoch": 0.011728954523644506, "grad_norm": 6.592710971832275, "learning_rate": 9.9562e-06, "loss": 0.294, "step": 220 }, { "epoch": 0.011782267953297435, "grad_norm": 6.720879077911377, "learning_rate": 9.956000000000001e-06, "loss": 0.2895, "step": 221 }, { "epoch": 0.011835581382950365, "grad_norm": 10.122069358825684, "learning_rate": 9.9558e-06, "loss": 0.3006, "step": 222 }, { "epoch": 0.011888894812603294, "grad_norm": 12.531805038452148, "learning_rate": 9.9556e-06, "loss": 0.2191, "step": 223 }, { "epoch": 0.011942208242256225, "grad_norm": 8.668679237365723, "learning_rate": 9.955400000000001e-06, "loss": 0.2619, "step": 224 }, { "epoch": 0.011995521671909154, "grad_norm": 25.5927791595459, "learning_rate": 9.9552e-06, "loss": 0.3488, "step": 225 }, { "epoch": 0.012048835101562084, "grad_norm": 5.684465408325195, "learning_rate": 9.955000000000002e-06, "loss": 0.2827, "step": 226 }, { "epoch": 0.012102148531215013, "grad_norm": 12.707826614379883, "learning_rate": 9.954800000000002e-06, "loss": 0.3133, "step": 227 }, { "epoch": 0.012155461960867942, "grad_norm": 25.86747932434082, "learning_rate": 9.954600000000001e-06, "loss": 0.3861, "step": 228 }, { "epoch": 0.012208775390520872, "grad_norm": 6.91982364654541, "learning_rate": 9.9544e-06, "loss": 0.2255, "step": 229 }, { "epoch": 0.012262088820173801, "grad_norm": 10.230962753295898, "learning_rate": 9.9542e-06, "loss": 0.1909, "step": 230 }, { "epoch": 0.012315402249826732, "grad_norm": 17.22847557067871, "learning_rate": 9.954e-06, "loss": 0.3168, "step": 231 }, { "epoch": 0.01236871567947966, "grad_norm": 13.448250770568848, "learning_rate": 9.953800000000001e-06, "loss": 0.2623, "step": 232 }, { "epoch": 0.012422029109132591, "grad_norm": 13.120665550231934, "learning_rate": 9.9536e-06, "loss": 0.292, "step": 233 }, { "epoch": 0.01247534253878552, "grad_norm": 19.719730377197266, "learning_rate": 9.9534e-06, "loss": 0.3266, "step": 234 }, { "epoch": 0.012528655968438449, "grad_norm": 14.156998634338379, "learning_rate": 9.9532e-06, "loss": 0.3172, "step": 235 }, { "epoch": 0.01258196939809138, "grad_norm": 13.704869270324707, "learning_rate": 9.953000000000001e-06, "loss": 0.3274, "step": 236 }, { "epoch": 0.012635282827744308, "grad_norm": 23.832517623901367, "learning_rate": 9.9528e-06, "loss": 0.2598, "step": 237 }, { "epoch": 0.012688596257397239, "grad_norm": 11.530607223510742, "learning_rate": 9.952600000000002e-06, "loss": 0.3185, "step": 238 }, { "epoch": 0.012741909687050168, "grad_norm": 8.370346069335938, "learning_rate": 9.952400000000001e-06, "loss": 0.2796, "step": 239 }, { "epoch": 0.012795223116703098, "grad_norm": 13.898874282836914, "learning_rate": 9.952200000000001e-06, "loss": 0.2558, "step": 240 }, { "epoch": 0.012848536546356027, "grad_norm": 12.4537935256958, "learning_rate": 9.952e-06, "loss": 0.2954, "step": 241 }, { "epoch": 0.012901849976008956, "grad_norm": 7.1912007331848145, "learning_rate": 9.9518e-06, "loss": 0.2483, "step": 242 }, { "epoch": 0.012955163405661886, "grad_norm": 9.21025562286377, "learning_rate": 9.951600000000001e-06, "loss": 0.2514, "step": 243 }, { "epoch": 0.013008476835314815, "grad_norm": 12.977232933044434, "learning_rate": 9.951400000000001e-06, "loss": 0.3157, "step": 244 }, { "epoch": 0.013061790264967746, "grad_norm": 10.453651428222656, "learning_rate": 9.9512e-06, "loss": 0.2441, "step": 245 }, { "epoch": 0.013115103694620675, "grad_norm": 12.600842475891113, "learning_rate": 9.951e-06, "loss": 0.2651, "step": 246 }, { "epoch": 0.013168417124273605, "grad_norm": 18.62221336364746, "learning_rate": 9.9508e-06, "loss": 0.2654, "step": 247 }, { "epoch": 0.013221730553926534, "grad_norm": 5.203212261199951, "learning_rate": 9.950600000000001e-06, "loss": 0.3095, "step": 248 }, { "epoch": 0.013275043983579463, "grad_norm": 20.73604965209961, "learning_rate": 9.9504e-06, "loss": 0.2901, "step": 249 }, { "epoch": 0.013328357413232393, "grad_norm": 14.512863159179688, "learning_rate": 9.950200000000002e-06, "loss": 0.2993, "step": 250 }, { "epoch": 0.013381670842885322, "grad_norm": 5.0119242668151855, "learning_rate": 9.950000000000001e-06, "loss": 0.2463, "step": 251 }, { "epoch": 0.013434984272538253, "grad_norm": 18.551042556762695, "learning_rate": 9.949800000000001e-06, "loss": 0.2533, "step": 252 }, { "epoch": 0.013488297702191182, "grad_norm": 9.902993202209473, "learning_rate": 9.9496e-06, "loss": 0.2254, "step": 253 }, { "epoch": 0.013541611131844112, "grad_norm": 8.619606018066406, "learning_rate": 9.949400000000002e-06, "loss": 0.3126, "step": 254 }, { "epoch": 0.013594924561497041, "grad_norm": 22.993881225585938, "learning_rate": 9.949200000000001e-06, "loss": 0.2934, "step": 255 }, { "epoch": 0.01364823799114997, "grad_norm": 8.330218315124512, "learning_rate": 9.949e-06, "loss": 0.2325, "step": 256 }, { "epoch": 0.0137015514208029, "grad_norm": 19.40289306640625, "learning_rate": 9.9488e-06, "loss": 0.3093, "step": 257 }, { "epoch": 0.01375486485045583, "grad_norm": 14.655821800231934, "learning_rate": 9.9486e-06, "loss": 0.3287, "step": 258 }, { "epoch": 0.01380817828010876, "grad_norm": 14.651030540466309, "learning_rate": 9.9484e-06, "loss": 0.269, "step": 259 }, { "epoch": 0.013861491709761689, "grad_norm": 12.725882530212402, "learning_rate": 9.9482e-06, "loss": 0.3119, "step": 260 }, { "epoch": 0.01391480513941462, "grad_norm": 5.293174743652344, "learning_rate": 9.948e-06, "loss": 0.2361, "step": 261 }, { "epoch": 0.013968118569067548, "grad_norm": 7.544923782348633, "learning_rate": 9.947800000000002e-06, "loss": 0.2394, "step": 262 }, { "epoch": 0.014021431998720477, "grad_norm": 17.697507858276367, "learning_rate": 9.947600000000001e-06, "loss": 0.2438, "step": 263 }, { "epoch": 0.014074745428373408, "grad_norm": 16.432771682739258, "learning_rate": 9.9474e-06, "loss": 0.3416, "step": 264 }, { "epoch": 0.014128058858026336, "grad_norm": 11.94946575164795, "learning_rate": 9.947200000000002e-06, "loss": 0.2671, "step": 265 }, { "epoch": 0.014181372287679267, "grad_norm": 36.980247497558594, "learning_rate": 9.947000000000001e-06, "loss": 0.2948, "step": 266 }, { "epoch": 0.014234685717332196, "grad_norm": 5.412839412689209, "learning_rate": 9.946800000000001e-06, "loss": 0.2517, "step": 267 }, { "epoch": 0.014287999146985126, "grad_norm": 15.159749031066895, "learning_rate": 9.9466e-06, "loss": 0.2096, "step": 268 }, { "epoch": 0.014341312576638055, "grad_norm": 18.910934448242188, "learning_rate": 9.9464e-06, "loss": 0.2849, "step": 269 }, { "epoch": 0.014394626006290984, "grad_norm": 9.852517127990723, "learning_rate": 9.9462e-06, "loss": 0.3, "step": 270 }, { "epoch": 0.014447939435943915, "grad_norm": 6.503200054168701, "learning_rate": 9.946000000000001e-06, "loss": 0.2915, "step": 271 }, { "epoch": 0.014501252865596843, "grad_norm": 23.362966537475586, "learning_rate": 9.9458e-06, "loss": 0.2738, "step": 272 }, { "epoch": 0.014554566295249774, "grad_norm": 14.968941688537598, "learning_rate": 9.9456e-06, "loss": 0.3215, "step": 273 }, { "epoch": 0.014607879724902703, "grad_norm": 4.799715042114258, "learning_rate": 9.945400000000001e-06, "loss": 0.2634, "step": 274 }, { "epoch": 0.014661193154555633, "grad_norm": 11.061326026916504, "learning_rate": 9.945200000000001e-06, "loss": 0.3018, "step": 275 }, { "epoch": 0.014714506584208562, "grad_norm": 11.138480186462402, "learning_rate": 9.945e-06, "loss": 0.262, "step": 276 }, { "epoch": 0.014767820013861491, "grad_norm": 7.519979476928711, "learning_rate": 9.944800000000002e-06, "loss": 0.2266, "step": 277 }, { "epoch": 0.014821133443514422, "grad_norm": 8.837923049926758, "learning_rate": 9.944600000000001e-06, "loss": 0.2321, "step": 278 }, { "epoch": 0.01487444687316735, "grad_norm": 6.5573883056640625, "learning_rate": 9.944400000000001e-06, "loss": 0.2516, "step": 279 }, { "epoch": 0.014927760302820281, "grad_norm": 4.131873607635498, "learning_rate": 9.9442e-06, "loss": 0.2437, "step": 280 }, { "epoch": 0.01498107373247321, "grad_norm": 8.214289665222168, "learning_rate": 9.944e-06, "loss": 0.2639, "step": 281 }, { "epoch": 0.01503438716212614, "grad_norm": 9.781641960144043, "learning_rate": 9.943800000000001e-06, "loss": 0.2136, "step": 282 }, { "epoch": 0.01508770059177907, "grad_norm": 8.170719146728516, "learning_rate": 9.9436e-06, "loss": 0.2345, "step": 283 }, { "epoch": 0.015141014021431998, "grad_norm": 6.730399131774902, "learning_rate": 9.9434e-06, "loss": 0.2382, "step": 284 }, { "epoch": 0.015194327451084929, "grad_norm": 9.278285026550293, "learning_rate": 9.9432e-06, "loss": 0.24, "step": 285 }, { "epoch": 0.015247640880737857, "grad_norm": 7.0752153396606445, "learning_rate": 9.943000000000001e-06, "loss": 0.284, "step": 286 }, { "epoch": 0.015300954310390788, "grad_norm": 5.817159652709961, "learning_rate": 9.9428e-06, "loss": 0.2193, "step": 287 }, { "epoch": 0.015354267740043717, "grad_norm": 8.240462303161621, "learning_rate": 9.9426e-06, "loss": 0.2797, "step": 288 }, { "epoch": 0.015407581169696647, "grad_norm": 7.104747295379639, "learning_rate": 9.942400000000002e-06, "loss": 0.2364, "step": 289 }, { "epoch": 0.015460894599349576, "grad_norm": 13.189618110656738, "learning_rate": 9.942200000000001e-06, "loss": 0.2602, "step": 290 }, { "epoch": 0.015514208029002505, "grad_norm": 5.314778804779053, "learning_rate": 9.942e-06, "loss": 0.2279, "step": 291 }, { "epoch": 0.015567521458655436, "grad_norm": 13.315853118896484, "learning_rate": 9.9418e-06, "loss": 0.2121, "step": 292 }, { "epoch": 0.015620834888308364, "grad_norm": 14.059967041015625, "learning_rate": 9.941600000000002e-06, "loss": 0.2307, "step": 293 }, { "epoch": 0.015674148317961295, "grad_norm": 8.354838371276855, "learning_rate": 9.941400000000001e-06, "loss": 0.2341, "step": 294 }, { "epoch": 0.015727461747614226, "grad_norm": 8.91091537475586, "learning_rate": 9.9412e-06, "loss": 0.2405, "step": 295 }, { "epoch": 0.015780775177267153, "grad_norm": 20.775468826293945, "learning_rate": 9.941e-06, "loss": 0.2734, "step": 296 }, { "epoch": 0.015834088606920083, "grad_norm": 11.591252326965332, "learning_rate": 9.9408e-06, "loss": 0.2165, "step": 297 }, { "epoch": 0.015887402036573014, "grad_norm": 22.34743309020996, "learning_rate": 9.940600000000001e-06, "loss": 0.2915, "step": 298 }, { "epoch": 0.01594071546622594, "grad_norm": 6.865456581115723, "learning_rate": 9.9404e-06, "loss": 0.2617, "step": 299 }, { "epoch": 0.01599402889587887, "grad_norm": 6.076998233795166, "learning_rate": 9.9402e-06, "loss": 0.1755, "step": 300 }, { "epoch": 0.016047342325531802, "grad_norm": 20.07436752319336, "learning_rate": 9.940000000000001e-06, "loss": 0.2324, "step": 301 }, { "epoch": 0.016100655755184733, "grad_norm": 12.766711235046387, "learning_rate": 9.939800000000001e-06, "loss": 0.2056, "step": 302 }, { "epoch": 0.01615396918483766, "grad_norm": 11.50462818145752, "learning_rate": 9.9396e-06, "loss": 0.3042, "step": 303 }, { "epoch": 0.01620728261449059, "grad_norm": 19.004003524780273, "learning_rate": 9.939400000000002e-06, "loss": 0.2568, "step": 304 }, { "epoch": 0.01626059604414352, "grad_norm": 7.083500385284424, "learning_rate": 9.939200000000001e-06, "loss": 0.2206, "step": 305 }, { "epoch": 0.016313909473796448, "grad_norm": 9.873062133789062, "learning_rate": 9.939000000000001e-06, "loss": 0.2443, "step": 306 }, { "epoch": 0.01636722290344938, "grad_norm": 18.123722076416016, "learning_rate": 9.9388e-06, "loss": 0.2916, "step": 307 }, { "epoch": 0.01642053633310231, "grad_norm": 9.712362289428711, "learning_rate": 9.9386e-06, "loss": 0.2342, "step": 308 }, { "epoch": 0.01647384976275524, "grad_norm": 9.807356834411621, "learning_rate": 9.9384e-06, "loss": 0.2532, "step": 309 }, { "epoch": 0.016527163192408167, "grad_norm": 7.764070510864258, "learning_rate": 9.9382e-06, "loss": 0.2108, "step": 310 }, { "epoch": 0.016580476622061097, "grad_norm": 13.797320365905762, "learning_rate": 9.938e-06, "loss": 0.1864, "step": 311 }, { "epoch": 0.016633790051714028, "grad_norm": 8.883193016052246, "learning_rate": 9.9378e-06, "loss": 0.2205, "step": 312 }, { "epoch": 0.016687103481366955, "grad_norm": 12.21835994720459, "learning_rate": 9.937600000000001e-06, "loss": 0.2017, "step": 313 }, { "epoch": 0.016740416911019886, "grad_norm": 22.727840423583984, "learning_rate": 9.9374e-06, "loss": 0.2575, "step": 314 }, { "epoch": 0.016793730340672816, "grad_norm": 14.096231460571289, "learning_rate": 9.937200000000002e-06, "loss": 0.2573, "step": 315 }, { "epoch": 0.016847043770325747, "grad_norm": 12.74871826171875, "learning_rate": 9.937000000000002e-06, "loss": 0.2288, "step": 316 }, { "epoch": 0.016900357199978674, "grad_norm": 10.305940628051758, "learning_rate": 9.936800000000001e-06, "loss": 0.2422, "step": 317 }, { "epoch": 0.016953670629631604, "grad_norm": 13.874433517456055, "learning_rate": 9.9366e-06, "loss": 0.1937, "step": 318 }, { "epoch": 0.017006984059284535, "grad_norm": 8.267145156860352, "learning_rate": 9.9364e-06, "loss": 0.2344, "step": 319 }, { "epoch": 0.017060297488937462, "grad_norm": 11.073529243469238, "learning_rate": 9.9362e-06, "loss": 0.1856, "step": 320 }, { "epoch": 0.017113610918590393, "grad_norm": 23.48326873779297, "learning_rate": 9.936000000000001e-06, "loss": 0.2477, "step": 321 }, { "epoch": 0.017166924348243323, "grad_norm": 24.487173080444336, "learning_rate": 9.9358e-06, "loss": 0.2742, "step": 322 }, { "epoch": 0.017220237777896254, "grad_norm": 18.0654354095459, "learning_rate": 9.9356e-06, "loss": 0.225, "step": 323 }, { "epoch": 0.01727355120754918, "grad_norm": 25.583032608032227, "learning_rate": 9.9354e-06, "loss": 0.2483, "step": 324 }, { "epoch": 0.01732686463720211, "grad_norm": 19.47852325439453, "learning_rate": 9.935200000000001e-06, "loss": 0.3079, "step": 325 }, { "epoch": 0.017380178066855042, "grad_norm": 14.916329383850098, "learning_rate": 9.935e-06, "loss": 0.243, "step": 326 }, { "epoch": 0.01743349149650797, "grad_norm": 29.543941497802734, "learning_rate": 9.934800000000002e-06, "loss": 0.2788, "step": 327 }, { "epoch": 0.0174868049261609, "grad_norm": 14.277339935302734, "learning_rate": 9.934600000000001e-06, "loss": 0.2077, "step": 328 }, { "epoch": 0.01754011835581383, "grad_norm": 11.945015907287598, "learning_rate": 9.934400000000001e-06, "loss": 0.2153, "step": 329 }, { "epoch": 0.01759343178546676, "grad_norm": 17.023391723632812, "learning_rate": 9.9342e-06, "loss": 0.2586, "step": 330 }, { "epoch": 0.017646745215119688, "grad_norm": 18.176998138427734, "learning_rate": 9.934e-06, "loss": 0.2624, "step": 331 }, { "epoch": 0.01770005864477262, "grad_norm": 13.893528938293457, "learning_rate": 9.933800000000001e-06, "loss": 0.2108, "step": 332 }, { "epoch": 0.01775337207442555, "grad_norm": 19.838157653808594, "learning_rate": 9.933600000000001e-06, "loss": 0.2684, "step": 333 }, { "epoch": 0.017806685504078476, "grad_norm": 31.05731773376465, "learning_rate": 9.9334e-06, "loss": 0.2798, "step": 334 }, { "epoch": 0.017859998933731407, "grad_norm": 9.584748268127441, "learning_rate": 9.9332e-06, "loss": 0.2075, "step": 335 }, { "epoch": 0.017913312363384337, "grad_norm": 26.844858169555664, "learning_rate": 9.933e-06, "loss": 0.3196, "step": 336 }, { "epoch": 0.017966625793037268, "grad_norm": 20.090789794921875, "learning_rate": 9.932800000000001e-06, "loss": 0.2605, "step": 337 }, { "epoch": 0.018019939222690195, "grad_norm": 5.949544906616211, "learning_rate": 9.9326e-06, "loss": 0.2445, "step": 338 }, { "epoch": 0.018073252652343125, "grad_norm": 29.637222290039062, "learning_rate": 9.932400000000002e-06, "loss": 0.3211, "step": 339 }, { "epoch": 0.018126566081996056, "grad_norm": 24.945533752441406, "learning_rate": 9.932200000000001e-06, "loss": 0.2803, "step": 340 }, { "epoch": 0.018179879511648983, "grad_norm": 7.887670040130615, "learning_rate": 9.932e-06, "loss": 0.3104, "step": 341 }, { "epoch": 0.018233192941301914, "grad_norm": 15.485922813415527, "learning_rate": 9.9318e-06, "loss": 0.1618, "step": 342 }, { "epoch": 0.018286506370954844, "grad_norm": 9.495627403259277, "learning_rate": 9.931600000000002e-06, "loss": 0.1887, "step": 343 }, { "epoch": 0.018339819800607775, "grad_norm": 21.045164108276367, "learning_rate": 9.931400000000001e-06, "loss": 0.2557, "step": 344 }, { "epoch": 0.018393133230260702, "grad_norm": 7.1868720054626465, "learning_rate": 9.9312e-06, "loss": 0.2031, "step": 345 }, { "epoch": 0.018446446659913632, "grad_norm": 8.929556846618652, "learning_rate": 9.931e-06, "loss": 0.2134, "step": 346 }, { "epoch": 0.018499760089566563, "grad_norm": 36.67235565185547, "learning_rate": 9.9308e-06, "loss": 0.3133, "step": 347 }, { "epoch": 0.01855307351921949, "grad_norm": 15.285879135131836, "learning_rate": 9.930600000000001e-06, "loss": 0.2334, "step": 348 }, { "epoch": 0.01860638694887242, "grad_norm": 13.207486152648926, "learning_rate": 9.9304e-06, "loss": 0.2655, "step": 349 }, { "epoch": 0.01865970037852535, "grad_norm": 26.1778621673584, "learning_rate": 9.9302e-06, "loss": 0.2818, "step": 350 }, { "epoch": 0.018713013808178282, "grad_norm": 8.703028678894043, "learning_rate": 9.930000000000001e-06, "loss": 0.2023, "step": 351 }, { "epoch": 0.01876632723783121, "grad_norm": 5.238168239593506, "learning_rate": 9.929800000000001e-06, "loss": 0.2805, "step": 352 }, { "epoch": 0.01881964066748414, "grad_norm": 8.990622520446777, "learning_rate": 9.9296e-06, "loss": 0.2785, "step": 353 }, { "epoch": 0.01887295409713707, "grad_norm": 9.299652099609375, "learning_rate": 9.929400000000002e-06, "loss": 0.2108, "step": 354 }, { "epoch": 0.018926267526789997, "grad_norm": 12.493908882141113, "learning_rate": 9.929200000000001e-06, "loss": 0.232, "step": 355 }, { "epoch": 0.018979580956442928, "grad_norm": 4.801002502441406, "learning_rate": 9.929000000000001e-06, "loss": 0.2085, "step": 356 }, { "epoch": 0.019032894386095858, "grad_norm": 6.52285623550415, "learning_rate": 9.9288e-06, "loss": 0.271, "step": 357 }, { "epoch": 0.01908620781574879, "grad_norm": 4.7100372314453125, "learning_rate": 9.9286e-06, "loss": 0.1872, "step": 358 }, { "epoch": 0.019139521245401716, "grad_norm": 10.36832332611084, "learning_rate": 9.9284e-06, "loss": 0.199, "step": 359 }, { "epoch": 0.019192834675054646, "grad_norm": 9.390624046325684, "learning_rate": 9.928200000000001e-06, "loss": 0.1996, "step": 360 }, { "epoch": 0.019246148104707577, "grad_norm": 4.9809980392456055, "learning_rate": 9.928e-06, "loss": 0.1877, "step": 361 }, { "epoch": 0.019299461534360504, "grad_norm": 12.096366882324219, "learning_rate": 9.9278e-06, "loss": 0.2116, "step": 362 }, { "epoch": 0.019352774964013435, "grad_norm": 18.05756187438965, "learning_rate": 9.927600000000001e-06, "loss": 0.252, "step": 363 }, { "epoch": 0.019406088393666365, "grad_norm": 10.770271301269531, "learning_rate": 9.927400000000001e-06, "loss": 0.2179, "step": 364 }, { "epoch": 0.019459401823319296, "grad_norm": 13.77939224243164, "learning_rate": 9.927200000000002e-06, "loss": 0.2637, "step": 365 }, { "epoch": 0.019512715252972223, "grad_norm": 6.8288187980651855, "learning_rate": 9.927000000000002e-06, "loss": 0.1965, "step": 366 }, { "epoch": 0.019566028682625154, "grad_norm": 5.072197437286377, "learning_rate": 9.926800000000001e-06, "loss": 0.2026, "step": 367 }, { "epoch": 0.019619342112278084, "grad_norm": 6.164743423461914, "learning_rate": 9.9266e-06, "loss": 0.1903, "step": 368 }, { "epoch": 0.01967265554193101, "grad_norm": 6.500518798828125, "learning_rate": 9.9264e-06, "loss": 0.2623, "step": 369 }, { "epoch": 0.019725968971583942, "grad_norm": 10.953998565673828, "learning_rate": 9.9262e-06, "loss": 0.1775, "step": 370 }, { "epoch": 0.019779282401236872, "grad_norm": 5.778527736663818, "learning_rate": 9.926000000000001e-06, "loss": 0.2027, "step": 371 }, { "epoch": 0.0198325958308898, "grad_norm": 9.027677536010742, "learning_rate": 9.9258e-06, "loss": 0.219, "step": 372 }, { "epoch": 0.01988590926054273, "grad_norm": 6.11890172958374, "learning_rate": 9.9256e-06, "loss": 0.2084, "step": 373 }, { "epoch": 0.01993922269019566, "grad_norm": 8.623141288757324, "learning_rate": 9.9254e-06, "loss": 0.1971, "step": 374 }, { "epoch": 0.01999253611984859, "grad_norm": 5.6168413162231445, "learning_rate": 9.925200000000001e-06, "loss": 0.1625, "step": 375 }, { "epoch": 0.020045849549501518, "grad_norm": 11.184049606323242, "learning_rate": 9.925e-06, "loss": 0.2296, "step": 376 }, { "epoch": 0.02009916297915445, "grad_norm": 4.705050468444824, "learning_rate": 9.924800000000002e-06, "loss": 0.1989, "step": 377 }, { "epoch": 0.02015247640880738, "grad_norm": 6.306752681732178, "learning_rate": 9.924600000000002e-06, "loss": 0.1746, "step": 378 }, { "epoch": 0.020205789838460306, "grad_norm": 8.04523754119873, "learning_rate": 9.924400000000001e-06, "loss": 0.2131, "step": 379 }, { "epoch": 0.020259103268113237, "grad_norm": 5.88600492477417, "learning_rate": 9.9242e-06, "loss": 0.2065, "step": 380 }, { "epoch": 0.020312416697766168, "grad_norm": 9.545868873596191, "learning_rate": 9.924e-06, "loss": 0.2048, "step": 381 }, { "epoch": 0.020365730127419098, "grad_norm": 7.787680625915527, "learning_rate": 9.923800000000001e-06, "loss": 0.203, "step": 382 }, { "epoch": 0.020419043557072025, "grad_norm": 8.101371765136719, "learning_rate": 9.923600000000001e-06, "loss": 0.2459, "step": 383 }, { "epoch": 0.020472356986724956, "grad_norm": 10.87304401397705, "learning_rate": 9.9234e-06, "loss": 0.2217, "step": 384 }, { "epoch": 0.020525670416377886, "grad_norm": 12.322413444519043, "learning_rate": 9.9232e-06, "loss": 0.1586, "step": 385 }, { "epoch": 0.020578983846030813, "grad_norm": 6.973572731018066, "learning_rate": 9.923e-06, "loss": 0.2184, "step": 386 }, { "epoch": 0.020632297275683744, "grad_norm": 7.168830871582031, "learning_rate": 9.922800000000001e-06, "loss": 0.2049, "step": 387 }, { "epoch": 0.020685610705336675, "grad_norm": 16.868324279785156, "learning_rate": 9.9226e-06, "loss": 0.2127, "step": 388 }, { "epoch": 0.020738924134989605, "grad_norm": 7.487894058227539, "learning_rate": 9.922400000000002e-06, "loss": 0.2286, "step": 389 }, { "epoch": 0.020792237564642532, "grad_norm": 8.767257690429688, "learning_rate": 9.922200000000001e-06, "loss": 0.2103, "step": 390 }, { "epoch": 0.020845550994295463, "grad_norm": 15.711902618408203, "learning_rate": 9.922000000000001e-06, "loss": 0.2472, "step": 391 }, { "epoch": 0.020898864423948393, "grad_norm": 14.54548454284668, "learning_rate": 9.9218e-06, "loss": 0.177, "step": 392 }, { "epoch": 0.02095217785360132, "grad_norm": 12.40511703491211, "learning_rate": 9.921600000000002e-06, "loss": 0.2108, "step": 393 }, { "epoch": 0.02100549128325425, "grad_norm": 19.235774993896484, "learning_rate": 9.921400000000001e-06, "loss": 0.2462, "step": 394 }, { "epoch": 0.02105880471290718, "grad_norm": 9.475895881652832, "learning_rate": 9.9212e-06, "loss": 0.202, "step": 395 }, { "epoch": 0.021112118142560112, "grad_norm": 7.723922252655029, "learning_rate": 9.921e-06, "loss": 0.1443, "step": 396 }, { "epoch": 0.02116543157221304, "grad_norm": 16.233407974243164, "learning_rate": 9.9208e-06, "loss": 0.2261, "step": 397 }, { "epoch": 0.02121874500186597, "grad_norm": 10.493538856506348, "learning_rate": 9.920600000000001e-06, "loss": 0.1988, "step": 398 }, { "epoch": 0.0212720584315189, "grad_norm": 9.558208465576172, "learning_rate": 9.9204e-06, "loss": 0.2392, "step": 399 }, { "epoch": 0.021325371861171828, "grad_norm": 9.851112365722656, "learning_rate": 9.9202e-06, "loss": 0.1748, "step": 400 }, { "epoch": 0.021378685290824758, "grad_norm": 5.727317810058594, "learning_rate": 9.920000000000002e-06, "loss": 0.1641, "step": 401 }, { "epoch": 0.02143199872047769, "grad_norm": 7.507072925567627, "learning_rate": 9.919800000000001e-06, "loss": 0.1707, "step": 402 }, { "epoch": 0.02148531215013062, "grad_norm": 7.301167011260986, "learning_rate": 9.9196e-06, "loss": 0.1966, "step": 403 }, { "epoch": 0.021538625579783546, "grad_norm": 9.8295316696167, "learning_rate": 9.919400000000002e-06, "loss": 0.2041, "step": 404 }, { "epoch": 0.021591939009436477, "grad_norm": 5.426194667816162, "learning_rate": 9.919200000000002e-06, "loss": 0.1912, "step": 405 }, { "epoch": 0.021645252439089407, "grad_norm": 7.241537570953369, "learning_rate": 9.919000000000001e-06, "loss": 0.1513, "step": 406 }, { "epoch": 0.021698565868742335, "grad_norm": 8.999641418457031, "learning_rate": 9.9188e-06, "loss": 0.2128, "step": 407 }, { "epoch": 0.021751879298395265, "grad_norm": 7.038762092590332, "learning_rate": 9.9186e-06, "loss": 0.1662, "step": 408 }, { "epoch": 0.021805192728048196, "grad_norm": 15.00442886352539, "learning_rate": 9.918400000000001e-06, "loss": 0.2435, "step": 409 }, { "epoch": 0.021858506157701126, "grad_norm": 7.820398330688477, "learning_rate": 9.918200000000001e-06, "loss": 0.1505, "step": 410 }, { "epoch": 0.021911819587354053, "grad_norm": 26.265918731689453, "learning_rate": 9.918e-06, "loss": 0.2661, "step": 411 }, { "epoch": 0.021965133017006984, "grad_norm": 10.420687675476074, "learning_rate": 9.9178e-06, "loss": 0.184, "step": 412 }, { "epoch": 0.022018446446659914, "grad_norm": 22.2409610748291, "learning_rate": 9.917600000000001e-06, "loss": 0.2076, "step": 413 }, { "epoch": 0.02207175987631284, "grad_norm": 11.62652587890625, "learning_rate": 9.917400000000001e-06, "loss": 0.176, "step": 414 }, { "epoch": 0.022125073305965772, "grad_norm": 7.3519463539123535, "learning_rate": 9.9172e-06, "loss": 0.1998, "step": 415 }, { "epoch": 0.022178386735618703, "grad_norm": 7.843620777130127, "learning_rate": 9.917000000000002e-06, "loss": 0.2326, "step": 416 }, { "epoch": 0.022231700165271633, "grad_norm": 5.327004432678223, "learning_rate": 9.916800000000001e-06, "loss": 0.1441, "step": 417 }, { "epoch": 0.02228501359492456, "grad_norm": 9.52043628692627, "learning_rate": 9.916600000000001e-06, "loss": 0.1378, "step": 418 }, { "epoch": 0.02233832702457749, "grad_norm": 5.433986186981201, "learning_rate": 9.9164e-06, "loss": 0.1958, "step": 419 }, { "epoch": 0.02239164045423042, "grad_norm": 6.975714683532715, "learning_rate": 9.9162e-06, "loss": 0.2056, "step": 420 }, { "epoch": 0.02244495388388335, "grad_norm": 16.48089027404785, "learning_rate": 9.916000000000001e-06, "loss": 0.2264, "step": 421 }, { "epoch": 0.02249826731353628, "grad_norm": 6.250727653503418, "learning_rate": 9.9158e-06, "loss": 0.1669, "step": 422 }, { "epoch": 0.02255158074318921, "grad_norm": 8.614611625671387, "learning_rate": 9.9156e-06, "loss": 0.2326, "step": 423 }, { "epoch": 0.02260489417284214, "grad_norm": 13.455598831176758, "learning_rate": 9.9154e-06, "loss": 0.2397, "step": 424 }, { "epoch": 0.022658207602495067, "grad_norm": 6.962201118469238, "learning_rate": 9.915200000000001e-06, "loss": 0.2107, "step": 425 }, { "epoch": 0.022711521032147998, "grad_norm": 9.837480545043945, "learning_rate": 9.915e-06, "loss": 0.2138, "step": 426 }, { "epoch": 0.02276483446180093, "grad_norm": 15.121448516845703, "learning_rate": 9.9148e-06, "loss": 0.1997, "step": 427 }, { "epoch": 0.022818147891453856, "grad_norm": 10.514009475708008, "learning_rate": 9.914600000000002e-06, "loss": 0.1737, "step": 428 }, { "epoch": 0.022871461321106786, "grad_norm": 21.827289581298828, "learning_rate": 9.914400000000001e-06, "loss": 0.2399, "step": 429 }, { "epoch": 0.022924774750759717, "grad_norm": 8.360135078430176, "learning_rate": 9.9142e-06, "loss": 0.2102, "step": 430 }, { "epoch": 0.022978088180412647, "grad_norm": 11.886600494384766, "learning_rate": 9.914e-06, "loss": 0.1989, "step": 431 }, { "epoch": 0.023031401610065574, "grad_norm": 14.851537704467773, "learning_rate": 9.913800000000002e-06, "loss": 0.2039, "step": 432 }, { "epoch": 0.023084715039718505, "grad_norm": 10.598621368408203, "learning_rate": 9.913600000000001e-06, "loss": 0.1804, "step": 433 }, { "epoch": 0.023138028469371436, "grad_norm": 13.62662410736084, "learning_rate": 9.9134e-06, "loss": 0.158, "step": 434 }, { "epoch": 0.023191341899024363, "grad_norm": 10.798352241516113, "learning_rate": 9.9132e-06, "loss": 0.2204, "step": 435 }, { "epoch": 0.023244655328677293, "grad_norm": 14.311994552612305, "learning_rate": 9.913e-06, "loss": 0.2076, "step": 436 }, { "epoch": 0.023297968758330224, "grad_norm": 9.439507484436035, "learning_rate": 9.912800000000001e-06, "loss": 0.1704, "step": 437 }, { "epoch": 0.023351282187983154, "grad_norm": 16.214595794677734, "learning_rate": 9.9126e-06, "loss": 0.1918, "step": 438 }, { "epoch": 0.02340459561763608, "grad_norm": 24.53922462463379, "learning_rate": 9.9124e-06, "loss": 0.2546, "step": 439 }, { "epoch": 0.023457909047289012, "grad_norm": 9.099987983703613, "learning_rate": 9.912200000000001e-06, "loss": 0.2004, "step": 440 }, { "epoch": 0.023511222476941943, "grad_norm": 31.12079429626465, "learning_rate": 9.912000000000001e-06, "loss": 0.2734, "step": 441 }, { "epoch": 0.02356453590659487, "grad_norm": 7.606357574462891, "learning_rate": 9.9118e-06, "loss": 0.1773, "step": 442 }, { "epoch": 0.0236178493362478, "grad_norm": 5.902374267578125, "learning_rate": 9.911600000000002e-06, "loss": 0.1781, "step": 443 }, { "epoch": 0.02367116276590073, "grad_norm": 16.79217529296875, "learning_rate": 9.911400000000001e-06, "loss": 0.2002, "step": 444 }, { "epoch": 0.02372447619555366, "grad_norm": 10.810759544372559, "learning_rate": 9.911200000000001e-06, "loss": 0.2419, "step": 445 }, { "epoch": 0.02377778962520659, "grad_norm": 17.800338745117188, "learning_rate": 9.911e-06, "loss": 0.238, "step": 446 }, { "epoch": 0.02383110305485952, "grad_norm": 6.151447296142578, "learning_rate": 9.9108e-06, "loss": 0.1734, "step": 447 }, { "epoch": 0.02388441648451245, "grad_norm": 11.19926643371582, "learning_rate": 9.910600000000001e-06, "loss": 0.1852, "step": 448 }, { "epoch": 0.023937729914165377, "grad_norm": 7.05264949798584, "learning_rate": 9.910400000000001e-06, "loss": 0.1684, "step": 449 }, { "epoch": 0.023991043343818307, "grad_norm": 10.246603965759277, "learning_rate": 9.9102e-06, "loss": 0.208, "step": 450 }, { "epoch": 0.024044356773471238, "grad_norm": 5.439992427825928, "learning_rate": 9.91e-06, "loss": 0.1638, "step": 451 }, { "epoch": 0.02409767020312417, "grad_norm": 17.240453720092773, "learning_rate": 9.909800000000001e-06, "loss": 0.1779, "step": 452 }, { "epoch": 0.024150983632777095, "grad_norm": 8.71756649017334, "learning_rate": 9.9096e-06, "loss": 0.1624, "step": 453 }, { "epoch": 0.024204297062430026, "grad_norm": 7.379639148712158, "learning_rate": 9.909400000000002e-06, "loss": 0.1615, "step": 454 }, { "epoch": 0.024257610492082957, "grad_norm": 7.91209077835083, "learning_rate": 9.909200000000002e-06, "loss": 0.1947, "step": 455 }, { "epoch": 0.024310923921735884, "grad_norm": 10.57120132446289, "learning_rate": 9.909000000000001e-06, "loss": 0.1688, "step": 456 }, { "epoch": 0.024364237351388814, "grad_norm": 10.081754684448242, "learning_rate": 9.9088e-06, "loss": 0.1687, "step": 457 }, { "epoch": 0.024417550781041745, "grad_norm": 6.4855427742004395, "learning_rate": 9.9086e-06, "loss": 0.2205, "step": 458 }, { "epoch": 0.024470864210694675, "grad_norm": 6.089195251464844, "learning_rate": 9.908400000000002e-06, "loss": 0.1674, "step": 459 }, { "epoch": 0.024524177640347603, "grad_norm": 10.791252136230469, "learning_rate": 9.908200000000001e-06, "loss": 0.1819, "step": 460 }, { "epoch": 0.024577491070000533, "grad_norm": 14.363564491271973, "learning_rate": 9.908e-06, "loss": 0.214, "step": 461 }, { "epoch": 0.024630804499653464, "grad_norm": 14.792388916015625, "learning_rate": 9.9078e-06, "loss": 0.184, "step": 462 }, { "epoch": 0.02468411792930639, "grad_norm": 11.6745023727417, "learning_rate": 9.9076e-06, "loss": 0.1585, "step": 463 }, { "epoch": 0.02473743135895932, "grad_norm": 15.255436897277832, "learning_rate": 9.907400000000001e-06, "loss": 0.2062, "step": 464 }, { "epoch": 0.024790744788612252, "grad_norm": 13.190953254699707, "learning_rate": 9.9072e-06, "loss": 0.2289, "step": 465 }, { "epoch": 0.024844058218265182, "grad_norm": 15.681346893310547, "learning_rate": 9.907000000000002e-06, "loss": 0.1737, "step": 466 }, { "epoch": 0.02489737164791811, "grad_norm": 10.646458625793457, "learning_rate": 9.906800000000001e-06, "loss": 0.1662, "step": 467 }, { "epoch": 0.02495068507757104, "grad_norm": 12.593668937683105, "learning_rate": 9.906600000000001e-06, "loss": 0.135, "step": 468 }, { "epoch": 0.02500399850722397, "grad_norm": 10.603588104248047, "learning_rate": 9.9064e-06, "loss": 0.172, "step": 469 }, { "epoch": 0.025057311936876898, "grad_norm": 14.535835266113281, "learning_rate": 9.9062e-06, "loss": 0.1948, "step": 470 }, { "epoch": 0.02511062536652983, "grad_norm": 9.17468547821045, "learning_rate": 9.906000000000001e-06, "loss": 0.1656, "step": 471 }, { "epoch": 0.02516393879618276, "grad_norm": 18.66618537902832, "learning_rate": 9.905800000000001e-06, "loss": 0.2523, "step": 472 }, { "epoch": 0.02521725222583569, "grad_norm": 14.34012222290039, "learning_rate": 9.9056e-06, "loss": 0.2092, "step": 473 }, { "epoch": 0.025270565655488617, "grad_norm": 19.823936462402344, "learning_rate": 9.9054e-06, "loss": 0.1877, "step": 474 }, { "epoch": 0.025323879085141547, "grad_norm": 29.03982925415039, "learning_rate": 9.9052e-06, "loss": 0.2419, "step": 475 }, { "epoch": 0.025377192514794478, "grad_norm": 7.1466827392578125, "learning_rate": 9.905000000000001e-06, "loss": 0.1672, "step": 476 }, { "epoch": 0.025430505944447405, "grad_norm": 16.154417037963867, "learning_rate": 9.9048e-06, "loss": 0.1942, "step": 477 }, { "epoch": 0.025483819374100335, "grad_norm": 23.658166885375977, "learning_rate": 9.904600000000002e-06, "loss": 0.2083, "step": 478 }, { "epoch": 0.025537132803753266, "grad_norm": 7.780315399169922, "learning_rate": 9.904400000000001e-06, "loss": 0.1517, "step": 479 }, { "epoch": 0.025590446233406196, "grad_norm": 22.860408782958984, "learning_rate": 9.9042e-06, "loss": 0.2289, "step": 480 }, { "epoch": 0.025643759663059124, "grad_norm": 25.359655380249023, "learning_rate": 9.904e-06, "loss": 0.2426, "step": 481 }, { "epoch": 0.025697073092712054, "grad_norm": 9.403043746948242, "learning_rate": 9.903800000000002e-06, "loss": 0.1721, "step": 482 }, { "epoch": 0.025750386522364985, "grad_norm": 10.473834991455078, "learning_rate": 9.903600000000001e-06, "loss": 0.1613, "step": 483 }, { "epoch": 0.025803699952017912, "grad_norm": 13.6873779296875, "learning_rate": 9.9034e-06, "loss": 0.2035, "step": 484 }, { "epoch": 0.025857013381670842, "grad_norm": 10.148738861083984, "learning_rate": 9.9032e-06, "loss": 0.1732, "step": 485 }, { "epoch": 0.025910326811323773, "grad_norm": 6.555778980255127, "learning_rate": 9.903e-06, "loss": 0.2185, "step": 486 }, { "epoch": 0.025963640240976704, "grad_norm": 23.75136947631836, "learning_rate": 9.902800000000001e-06, "loss": 0.2451, "step": 487 }, { "epoch": 0.02601695367062963, "grad_norm": 12.536886215209961, "learning_rate": 9.9026e-06, "loss": 0.2111, "step": 488 }, { "epoch": 0.02607026710028256, "grad_norm": 5.341726303100586, "learning_rate": 9.9024e-06, "loss": 0.144, "step": 489 }, { "epoch": 0.02612358052993549, "grad_norm": 12.502069473266602, "learning_rate": 9.902200000000002e-06, "loss": 0.1084, "step": 490 }, { "epoch": 0.02617689395958842, "grad_norm": 18.617919921875, "learning_rate": 9.902000000000001e-06, "loss": 0.2514, "step": 491 }, { "epoch": 0.02623020738924135, "grad_norm": 10.881681442260742, "learning_rate": 9.9018e-06, "loss": 0.1584, "step": 492 }, { "epoch": 0.02628352081889428, "grad_norm": 9.260968208312988, "learning_rate": 9.901600000000002e-06, "loss": 0.2396, "step": 493 }, { "epoch": 0.02633683424854721, "grad_norm": 23.961179733276367, "learning_rate": 9.901400000000001e-06, "loss": 0.2422, "step": 494 }, { "epoch": 0.026390147678200138, "grad_norm": 16.69447898864746, "learning_rate": 9.901200000000001e-06, "loss": 0.1866, "step": 495 }, { "epoch": 0.026443461107853068, "grad_norm": 4.907302379608154, "learning_rate": 9.901e-06, "loss": 0.1447, "step": 496 }, { "epoch": 0.026496774537506, "grad_norm": 6.821753025054932, "learning_rate": 9.9008e-06, "loss": 0.1532, "step": 497 }, { "epoch": 0.026550087967158926, "grad_norm": 14.216325759887695, "learning_rate": 9.900600000000001e-06, "loss": 0.1984, "step": 498 }, { "epoch": 0.026603401396811856, "grad_norm": 15.019767761230469, "learning_rate": 9.900400000000001e-06, "loss": 0.2114, "step": 499 }, { "epoch": 0.026656714826464787, "grad_norm": 7.638628005981445, "learning_rate": 9.9002e-06, "loss": 0.1639, "step": 500 }, { "epoch": 0.026710028256117718, "grad_norm": 12.375450134277344, "learning_rate": 9.9e-06, "loss": 0.1843, "step": 501 }, { "epoch": 0.026763341685770645, "grad_norm": 9.336312294006348, "learning_rate": 9.899800000000001e-06, "loss": 0.1571, "step": 502 }, { "epoch": 0.026816655115423575, "grad_norm": 6.20499324798584, "learning_rate": 9.899600000000001e-06, "loss": 0.1682, "step": 503 }, { "epoch": 0.026869968545076506, "grad_norm": 3.924057722091675, "learning_rate": 9.8994e-06, "loss": 0.1234, "step": 504 }, { "epoch": 0.026923281974729433, "grad_norm": 6.126389503479004, "learning_rate": 9.899200000000002e-06, "loss": 0.1842, "step": 505 }, { "epoch": 0.026976595404382363, "grad_norm": 9.103784561157227, "learning_rate": 9.899000000000001e-06, "loss": 0.2166, "step": 506 }, { "epoch": 0.027029908834035294, "grad_norm": 10.937093734741211, "learning_rate": 9.8988e-06, "loss": 0.1753, "step": 507 }, { "epoch": 0.027083222263688225, "grad_norm": 10.883543014526367, "learning_rate": 9.8986e-06, "loss": 0.2034, "step": 508 }, { "epoch": 0.02713653569334115, "grad_norm": 7.4324631690979, "learning_rate": 9.898400000000002e-06, "loss": 0.2113, "step": 509 }, { "epoch": 0.027189849122994082, "grad_norm": 11.533535957336426, "learning_rate": 9.898200000000001e-06, "loss": 0.1538, "step": 510 }, { "epoch": 0.027243162552647013, "grad_norm": 7.24744987487793, "learning_rate": 9.898e-06, "loss": 0.1382, "step": 511 }, { "epoch": 0.02729647598229994, "grad_norm": 7.064036846160889, "learning_rate": 9.8978e-06, "loss": 0.1888, "step": 512 }, { "epoch": 0.02734978941195287, "grad_norm": 17.964956283569336, "learning_rate": 9.8976e-06, "loss": 0.1676, "step": 513 }, { "epoch": 0.0274031028416058, "grad_norm": 10.405848503112793, "learning_rate": 9.897400000000001e-06, "loss": 0.173, "step": 514 }, { "epoch": 0.02745641627125873, "grad_norm": 11.129328727722168, "learning_rate": 9.8972e-06, "loss": 0.1654, "step": 515 }, { "epoch": 0.02750972970091166, "grad_norm": 14.82622241973877, "learning_rate": 9.897e-06, "loss": 0.1897, "step": 516 }, { "epoch": 0.02756304313056459, "grad_norm": 9.824042320251465, "learning_rate": 9.896800000000002e-06, "loss": 0.1548, "step": 517 }, { "epoch": 0.02761635656021752, "grad_norm": 11.869132995605469, "learning_rate": 9.896600000000001e-06, "loss": 0.1463, "step": 518 }, { "epoch": 0.027669669989870447, "grad_norm": 5.148487567901611, "learning_rate": 9.8964e-06, "loss": 0.147, "step": 519 }, { "epoch": 0.027722983419523377, "grad_norm": 14.932727813720703, "learning_rate": 9.896200000000002e-06, "loss": 0.1599, "step": 520 }, { "epoch": 0.027776296849176308, "grad_norm": 14.1984281539917, "learning_rate": 9.896000000000001e-06, "loss": 0.2001, "step": 521 }, { "epoch": 0.02782961027882924, "grad_norm": 7.049540042877197, "learning_rate": 9.895800000000001e-06, "loss": 0.1465, "step": 522 }, { "epoch": 0.027882923708482166, "grad_norm": 7.3323516845703125, "learning_rate": 9.8956e-06, "loss": 0.1698, "step": 523 }, { "epoch": 0.027936237138135096, "grad_norm": 15.871715545654297, "learning_rate": 9.8954e-06, "loss": 0.1714, "step": 524 }, { "epoch": 0.027989550567788027, "grad_norm": 8.236368179321289, "learning_rate": 9.8952e-06, "loss": 0.1589, "step": 525 }, { "epoch": 0.028042863997440954, "grad_norm": 11.231840133666992, "learning_rate": 9.895000000000001e-06, "loss": 0.1585, "step": 526 }, { "epoch": 0.028096177427093885, "grad_norm": 14.317397117614746, "learning_rate": 9.8948e-06, "loss": 0.1265, "step": 527 }, { "epoch": 0.028149490856746815, "grad_norm": 5.634779930114746, "learning_rate": 9.8946e-06, "loss": 0.1147, "step": 528 }, { "epoch": 0.028202804286399746, "grad_norm": 14.145242691040039, "learning_rate": 9.894400000000001e-06, "loss": 0.2037, "step": 529 }, { "epoch": 0.028256117716052673, "grad_norm": 8.087326049804688, "learning_rate": 9.894200000000001e-06, "loss": 0.1273, "step": 530 }, { "epoch": 0.028309431145705603, "grad_norm": 7.978499412536621, "learning_rate": 9.894e-06, "loss": 0.1773, "step": 531 }, { "epoch": 0.028362744575358534, "grad_norm": 15.137093544006348, "learning_rate": 9.893800000000002e-06, "loss": 0.2244, "step": 532 }, { "epoch": 0.02841605800501146, "grad_norm": 10.766651153564453, "learning_rate": 9.893600000000001e-06, "loss": 0.1648, "step": 533 }, { "epoch": 0.02846937143466439, "grad_norm": 10.418318748474121, "learning_rate": 9.893400000000001e-06, "loss": 0.1945, "step": 534 }, { "epoch": 0.028522684864317322, "grad_norm": 5.754423141479492, "learning_rate": 9.8932e-06, "loss": 0.1225, "step": 535 }, { "epoch": 0.028575998293970253, "grad_norm": 8.100637435913086, "learning_rate": 9.893e-06, "loss": 0.1495, "step": 536 }, { "epoch": 0.02862931172362318, "grad_norm": 7.332054615020752, "learning_rate": 9.892800000000001e-06, "loss": 0.2008, "step": 537 }, { "epoch": 0.02868262515327611, "grad_norm": 10.919559478759766, "learning_rate": 9.8926e-06, "loss": 0.1805, "step": 538 }, { "epoch": 0.02873593858292904, "grad_norm": 15.231783866882324, "learning_rate": 9.8924e-06, "loss": 0.1484, "step": 539 }, { "epoch": 0.028789252012581968, "grad_norm": 12.580184936523438, "learning_rate": 9.8922e-06, "loss": 0.1565, "step": 540 }, { "epoch": 0.0288425654422349, "grad_norm": 6.550548076629639, "learning_rate": 9.892000000000001e-06, "loss": 0.1388, "step": 541 }, { "epoch": 0.02889587887188783, "grad_norm": 8.80394172668457, "learning_rate": 9.8918e-06, "loss": 0.1915, "step": 542 }, { "epoch": 0.02894919230154076, "grad_norm": 6.812580108642578, "learning_rate": 9.891600000000002e-06, "loss": 0.1058, "step": 543 }, { "epoch": 0.029002505731193687, "grad_norm": 7.070510387420654, "learning_rate": 9.891400000000002e-06, "loss": 0.1915, "step": 544 }, { "epoch": 0.029055819160846617, "grad_norm": 8.753482818603516, "learning_rate": 9.891200000000001e-06, "loss": 0.1393, "step": 545 }, { "epoch": 0.029109132590499548, "grad_norm": 8.173974990844727, "learning_rate": 9.891e-06, "loss": 0.1841, "step": 546 }, { "epoch": 0.029162446020152475, "grad_norm": 8.423271179199219, "learning_rate": 9.8908e-06, "loss": 0.187, "step": 547 }, { "epoch": 0.029215759449805406, "grad_norm": 7.987669467926025, "learning_rate": 9.890600000000001e-06, "loss": 0.1863, "step": 548 }, { "epoch": 0.029269072879458336, "grad_norm": 7.499307155609131, "learning_rate": 9.890400000000001e-06, "loss": 0.2142, "step": 549 }, { "epoch": 0.029322386309111267, "grad_norm": 10.673420906066895, "learning_rate": 9.8902e-06, "loss": 0.1564, "step": 550 }, { "epoch": 0.029375699738764194, "grad_norm": 9.800189971923828, "learning_rate": 9.89e-06, "loss": 0.1616, "step": 551 }, { "epoch": 0.029429013168417124, "grad_norm": 5.86992073059082, "learning_rate": 9.8898e-06, "loss": 0.1639, "step": 552 }, { "epoch": 0.029482326598070055, "grad_norm": 11.958218574523926, "learning_rate": 9.889600000000001e-06, "loss": 0.1605, "step": 553 }, { "epoch": 0.029535640027722982, "grad_norm": 7.8236260414123535, "learning_rate": 9.8894e-06, "loss": 0.1777, "step": 554 }, { "epoch": 0.029588953457375913, "grad_norm": 7.044162273406982, "learning_rate": 9.889200000000002e-06, "loss": 0.1073, "step": 555 }, { "epoch": 0.029642266887028843, "grad_norm": 7.3043928146362305, "learning_rate": 9.889000000000001e-06, "loss": 0.1377, "step": 556 }, { "epoch": 0.029695580316681774, "grad_norm": 14.635580062866211, "learning_rate": 9.888800000000001e-06, "loss": 0.1979, "step": 557 }, { "epoch": 0.0297488937463347, "grad_norm": 7.815776824951172, "learning_rate": 9.8886e-06, "loss": 0.1356, "step": 558 }, { "epoch": 0.02980220717598763, "grad_norm": 28.05089569091797, "learning_rate": 9.888400000000002e-06, "loss": 0.2815, "step": 559 }, { "epoch": 0.029855520605640562, "grad_norm": 13.914690017700195, "learning_rate": 9.888200000000001e-06, "loss": 0.1991, "step": 560 }, { "epoch": 0.02990883403529349, "grad_norm": 13.530871391296387, "learning_rate": 9.888000000000001e-06, "loss": 0.173, "step": 561 }, { "epoch": 0.02996214746494642, "grad_norm": 24.471820831298828, "learning_rate": 9.8878e-06, "loss": 0.2105, "step": 562 }, { "epoch": 0.03001546089459935, "grad_norm": 13.496963500976562, "learning_rate": 9.8876e-06, "loss": 0.1415, "step": 563 }, { "epoch": 0.03006877432425228, "grad_norm": 7.130620956420898, "learning_rate": 9.8874e-06, "loss": 0.2102, "step": 564 }, { "epoch": 0.030122087753905208, "grad_norm": 12.020750999450684, "learning_rate": 9.8872e-06, "loss": 0.1869, "step": 565 }, { "epoch": 0.03017540118355814, "grad_norm": 8.079834938049316, "learning_rate": 9.887e-06, "loss": 0.1865, "step": 566 }, { "epoch": 0.03022871461321107, "grad_norm": 12.537734985351562, "learning_rate": 9.886800000000002e-06, "loss": 0.1728, "step": 567 }, { "epoch": 0.030282028042863996, "grad_norm": 8.578534126281738, "learning_rate": 9.886600000000001e-06, "loss": 0.1435, "step": 568 }, { "epoch": 0.030335341472516927, "grad_norm": 15.632637023925781, "learning_rate": 9.8864e-06, "loss": 0.196, "step": 569 }, { "epoch": 0.030388654902169857, "grad_norm": 14.611496925354004, "learning_rate": 9.886200000000002e-06, "loss": 0.1714, "step": 570 }, { "epoch": 0.030441968331822788, "grad_norm": 6.365734577178955, "learning_rate": 9.886000000000002e-06, "loss": 0.1231, "step": 571 }, { "epoch": 0.030495281761475715, "grad_norm": 10.07763957977295, "learning_rate": 9.885800000000001e-06, "loss": 0.1717, "step": 572 }, { "epoch": 0.030548595191128645, "grad_norm": 8.870412826538086, "learning_rate": 9.8856e-06, "loss": 0.1788, "step": 573 }, { "epoch": 0.030601908620781576, "grad_norm": 10.848949432373047, "learning_rate": 9.8854e-06, "loss": 0.1695, "step": 574 }, { "epoch": 0.030655222050434503, "grad_norm": 6.548939228057861, "learning_rate": 9.8852e-06, "loss": 0.1498, "step": 575 }, { "epoch": 0.030708535480087434, "grad_norm": 18.4104061126709, "learning_rate": 9.885000000000001e-06, "loss": 0.1716, "step": 576 }, { "epoch": 0.030761848909740364, "grad_norm": 8.68282413482666, "learning_rate": 9.8848e-06, "loss": 0.1805, "step": 577 }, { "epoch": 0.030815162339393295, "grad_norm": 12.635462760925293, "learning_rate": 9.8846e-06, "loss": 0.1904, "step": 578 }, { "epoch": 0.030868475769046222, "grad_norm": 6.826554298400879, "learning_rate": 9.884400000000001e-06, "loss": 0.1887, "step": 579 }, { "epoch": 0.030921789198699152, "grad_norm": 15.172776222229004, "learning_rate": 9.884200000000001e-06, "loss": 0.1895, "step": 580 }, { "epoch": 0.030975102628352083, "grad_norm": 10.058018684387207, "learning_rate": 9.884e-06, "loss": 0.1502, "step": 581 }, { "epoch": 0.03102841605800501, "grad_norm": 6.999688625335693, "learning_rate": 9.883800000000002e-06, "loss": 0.1463, "step": 582 }, { "epoch": 0.03108172948765794, "grad_norm": 5.440664768218994, "learning_rate": 9.883600000000001e-06, "loss": 0.1645, "step": 583 }, { "epoch": 0.03113504291731087, "grad_norm": 11.549346923828125, "learning_rate": 9.883400000000001e-06, "loss": 0.1652, "step": 584 }, { "epoch": 0.031188356346963802, "grad_norm": 5.881464004516602, "learning_rate": 9.8832e-06, "loss": 0.1874, "step": 585 }, { "epoch": 0.03124166977661673, "grad_norm": 6.0882792472839355, "learning_rate": 9.883e-06, "loss": 0.1342, "step": 586 }, { "epoch": 0.03129498320626966, "grad_norm": 11.371223449707031, "learning_rate": 9.882800000000001e-06, "loss": 0.1733, "step": 587 }, { "epoch": 0.03134829663592259, "grad_norm": 10.309940338134766, "learning_rate": 9.882600000000001e-06, "loss": 0.1205, "step": 588 }, { "epoch": 0.03140161006557552, "grad_norm": 7.784701347351074, "learning_rate": 9.8824e-06, "loss": 0.133, "step": 589 }, { "epoch": 0.03145492349522845, "grad_norm": 12.97317886352539, "learning_rate": 9.8822e-06, "loss": 0.1196, "step": 590 }, { "epoch": 0.031508236924881375, "grad_norm": 8.26850414276123, "learning_rate": 9.882000000000001e-06, "loss": 0.1508, "step": 591 }, { "epoch": 0.031561550354534305, "grad_norm": 11.802139282226562, "learning_rate": 9.8818e-06, "loss": 0.1466, "step": 592 }, { "epoch": 0.031614863784187236, "grad_norm": 8.949982643127441, "learning_rate": 9.8816e-06, "loss": 0.141, "step": 593 }, { "epoch": 0.03166817721384017, "grad_norm": 12.226090431213379, "learning_rate": 9.881400000000002e-06, "loss": 0.1384, "step": 594 }, { "epoch": 0.0317214906434931, "grad_norm": 14.786581993103027, "learning_rate": 9.881200000000001e-06, "loss": 0.1792, "step": 595 }, { "epoch": 0.03177480407314603, "grad_norm": 16.6456298828125, "learning_rate": 9.881e-06, "loss": 0.185, "step": 596 }, { "epoch": 0.03182811750279896, "grad_norm": 6.666614532470703, "learning_rate": 9.8808e-06, "loss": 0.1341, "step": 597 }, { "epoch": 0.03188143093245188, "grad_norm": 12.436727523803711, "learning_rate": 9.880600000000002e-06, "loss": 0.1587, "step": 598 }, { "epoch": 0.03193474436210481, "grad_norm": 9.265117645263672, "learning_rate": 9.880400000000001e-06, "loss": 0.1654, "step": 599 }, { "epoch": 0.03198805779175774, "grad_norm": 11.589969635009766, "learning_rate": 9.8802e-06, "loss": 0.192, "step": 600 }, { "epoch": 0.032041371221410674, "grad_norm": 15.572157859802246, "learning_rate": 9.88e-06, "loss": 0.1705, "step": 601 }, { "epoch": 0.032094684651063604, "grad_norm": 10.859305381774902, "learning_rate": 9.8798e-06, "loss": 0.138, "step": 602 }, { "epoch": 0.032147998080716535, "grad_norm": 10.069561004638672, "learning_rate": 9.879600000000001e-06, "loss": 0.1617, "step": 603 }, { "epoch": 0.032201311510369465, "grad_norm": 8.20162582397461, "learning_rate": 9.8794e-06, "loss": 0.1862, "step": 604 }, { "epoch": 0.03225462494002239, "grad_norm": 7.278533458709717, "learning_rate": 9.8792e-06, "loss": 0.1346, "step": 605 }, { "epoch": 0.03230793836967532, "grad_norm": 10.60849666595459, "learning_rate": 9.879000000000001e-06, "loss": 0.1629, "step": 606 }, { "epoch": 0.03236125179932825, "grad_norm": 8.72110652923584, "learning_rate": 9.878800000000001e-06, "loss": 0.1962, "step": 607 }, { "epoch": 0.03241456522898118, "grad_norm": 6.028036117553711, "learning_rate": 9.8786e-06, "loss": 0.1681, "step": 608 }, { "epoch": 0.03246787865863411, "grad_norm": 5.27310848236084, "learning_rate": 9.878400000000002e-06, "loss": 0.1294, "step": 609 }, { "epoch": 0.03252119208828704, "grad_norm": 7.7068352699279785, "learning_rate": 9.878200000000001e-06, "loss": 0.1901, "step": 610 }, { "epoch": 0.03257450551793997, "grad_norm": 17.37053680419922, "learning_rate": 9.878000000000001e-06, "loss": 0.2035, "step": 611 }, { "epoch": 0.032627818947592896, "grad_norm": 7.640080451965332, "learning_rate": 9.8778e-06, "loss": 0.1608, "step": 612 }, { "epoch": 0.032681132377245826, "grad_norm": 9.801114082336426, "learning_rate": 9.8776e-06, "loss": 0.1438, "step": 613 }, { "epoch": 0.03273444580689876, "grad_norm": 17.605623245239258, "learning_rate": 9.8774e-06, "loss": 0.1854, "step": 614 }, { "epoch": 0.03278775923655169, "grad_norm": 12.462841987609863, "learning_rate": 9.877200000000001e-06, "loss": 0.1446, "step": 615 }, { "epoch": 0.03284107266620462, "grad_norm": 7.629948616027832, "learning_rate": 9.877e-06, "loss": 0.1212, "step": 616 }, { "epoch": 0.03289438609585755, "grad_norm": 30.88214683532715, "learning_rate": 9.8768e-06, "loss": 0.2568, "step": 617 }, { "epoch": 0.03294769952551048, "grad_norm": 8.404494285583496, "learning_rate": 9.876600000000001e-06, "loss": 0.1558, "step": 618 }, { "epoch": 0.0330010129551634, "grad_norm": 9.332806587219238, "learning_rate": 9.876400000000001e-06, "loss": 0.1479, "step": 619 }, { "epoch": 0.033054326384816334, "grad_norm": 11.240091323852539, "learning_rate": 9.876200000000002e-06, "loss": 0.1553, "step": 620 }, { "epoch": 0.033107639814469264, "grad_norm": 6.362691402435303, "learning_rate": 9.876000000000002e-06, "loss": 0.0983, "step": 621 }, { "epoch": 0.033160953244122195, "grad_norm": 12.063401222229004, "learning_rate": 9.875800000000001e-06, "loss": 0.2006, "step": 622 }, { "epoch": 0.033214266673775125, "grad_norm": 8.505529403686523, "learning_rate": 9.8756e-06, "loss": 0.1836, "step": 623 }, { "epoch": 0.033267580103428056, "grad_norm": 13.395551681518555, "learning_rate": 9.8754e-06, "loss": 0.1292, "step": 624 }, { "epoch": 0.033320893533080986, "grad_norm": 14.243236541748047, "learning_rate": 9.8752e-06, "loss": 0.1595, "step": 625 }, { "epoch": 0.03337420696273391, "grad_norm": 5.673371315002441, "learning_rate": 9.875000000000001e-06, "loss": 0.1542, "step": 626 }, { "epoch": 0.03342752039238684, "grad_norm": 12.703034400939941, "learning_rate": 9.8748e-06, "loss": 0.2071, "step": 627 }, { "epoch": 0.03348083382203977, "grad_norm": 11.252038955688477, "learning_rate": 9.8746e-06, "loss": 0.1222, "step": 628 }, { "epoch": 0.0335341472516927, "grad_norm": 10.247370719909668, "learning_rate": 9.8744e-06, "loss": 0.1769, "step": 629 }, { "epoch": 0.03358746068134563, "grad_norm": 8.671151161193848, "learning_rate": 9.874200000000001e-06, "loss": 0.1724, "step": 630 }, { "epoch": 0.03364077411099856, "grad_norm": 19.068775177001953, "learning_rate": 9.874e-06, "loss": 0.2551, "step": 631 }, { "epoch": 0.03369408754065149, "grad_norm": 9.859038352966309, "learning_rate": 9.873800000000002e-06, "loss": 0.1177, "step": 632 }, { "epoch": 0.03374740097030442, "grad_norm": 6.223719596862793, "learning_rate": 9.873600000000001e-06, "loss": 0.1812, "step": 633 }, { "epoch": 0.03380071439995735, "grad_norm": 8.366297721862793, "learning_rate": 9.873400000000001e-06, "loss": 0.1325, "step": 634 }, { "epoch": 0.03385402782961028, "grad_norm": 21.118745803833008, "learning_rate": 9.8732e-06, "loss": 0.2145, "step": 635 }, { "epoch": 0.03390734125926321, "grad_norm": 9.763642311096191, "learning_rate": 9.873e-06, "loss": 0.1391, "step": 636 }, { "epoch": 0.03396065468891614, "grad_norm": 14.08376407623291, "learning_rate": 9.872800000000001e-06, "loss": 0.247, "step": 637 }, { "epoch": 0.03401396811856907, "grad_norm": 13.829219818115234, "learning_rate": 9.872600000000001e-06, "loss": 0.1487, "step": 638 }, { "epoch": 0.034067281548222, "grad_norm": 10.43845272064209, "learning_rate": 9.8724e-06, "loss": 0.1607, "step": 639 }, { "epoch": 0.034120594977874924, "grad_norm": 18.2194881439209, "learning_rate": 9.8722e-06, "loss": 0.2131, "step": 640 }, { "epoch": 0.034173908407527855, "grad_norm": 9.800444602966309, "learning_rate": 9.872e-06, "loss": 0.1295, "step": 641 }, { "epoch": 0.034227221837180785, "grad_norm": 10.966073036193848, "learning_rate": 9.871800000000001e-06, "loss": 0.1624, "step": 642 }, { "epoch": 0.034280535266833716, "grad_norm": 9.76494026184082, "learning_rate": 9.8716e-06, "loss": 0.1344, "step": 643 }, { "epoch": 0.034333848696486646, "grad_norm": 10.118760108947754, "learning_rate": 9.871400000000002e-06, "loss": 0.1449, "step": 644 }, { "epoch": 0.03438716212613958, "grad_norm": 17.64834213256836, "learning_rate": 9.871200000000001e-06, "loss": 0.162, "step": 645 }, { "epoch": 0.03444047555579251, "grad_norm": 4.83218240737915, "learning_rate": 9.871000000000001e-06, "loss": 0.1205, "step": 646 }, { "epoch": 0.03449378898544543, "grad_norm": 11.43222713470459, "learning_rate": 9.8708e-06, "loss": 0.1702, "step": 647 }, { "epoch": 0.03454710241509836, "grad_norm": 4.821547031402588, "learning_rate": 9.870600000000002e-06, "loss": 0.1215, "step": 648 }, { "epoch": 0.03460041584475129, "grad_norm": 11.206171989440918, "learning_rate": 9.870400000000001e-06, "loss": 0.1834, "step": 649 }, { "epoch": 0.03465372927440422, "grad_norm": 10.265257835388184, "learning_rate": 9.8702e-06, "loss": 0.1596, "step": 650 }, { "epoch": 0.03470704270405715, "grad_norm": 21.00145149230957, "learning_rate": 9.87e-06, "loss": 0.1864, "step": 651 }, { "epoch": 0.034760356133710084, "grad_norm": 8.05295467376709, "learning_rate": 9.8698e-06, "loss": 0.1349, "step": 652 }, { "epoch": 0.034813669563363014, "grad_norm": 5.041062831878662, "learning_rate": 9.8696e-06, "loss": 0.1147, "step": 653 }, { "epoch": 0.03486698299301594, "grad_norm": 5.663896560668945, "learning_rate": 9.8694e-06, "loss": 0.1316, "step": 654 }, { "epoch": 0.03492029642266887, "grad_norm": 8.933537483215332, "learning_rate": 9.8692e-06, "loss": 0.1414, "step": 655 }, { "epoch": 0.0349736098523218, "grad_norm": 5.089033603668213, "learning_rate": 9.869000000000002e-06, "loss": 0.1033, "step": 656 }, { "epoch": 0.03502692328197473, "grad_norm": 10.968677520751953, "learning_rate": 9.868800000000001e-06, "loss": 0.1646, "step": 657 }, { "epoch": 0.03508023671162766, "grad_norm": 11.452962875366211, "learning_rate": 9.8686e-06, "loss": 0.1377, "step": 658 }, { "epoch": 0.03513355014128059, "grad_norm": 7.144217491149902, "learning_rate": 9.868400000000002e-06, "loss": 0.1324, "step": 659 }, { "epoch": 0.03518686357093352, "grad_norm": 7.523974895477295, "learning_rate": 9.868200000000002e-06, "loss": 0.1604, "step": 660 }, { "epoch": 0.035240177000586445, "grad_norm": 10.773066520690918, "learning_rate": 9.868000000000001e-06, "loss": 0.1577, "step": 661 }, { "epoch": 0.035293490430239376, "grad_norm": 7.7653093338012695, "learning_rate": 9.8678e-06, "loss": 0.1916, "step": 662 }, { "epoch": 0.035346803859892306, "grad_norm": 8.651836395263672, "learning_rate": 9.8676e-06, "loss": 0.1406, "step": 663 }, { "epoch": 0.03540011728954524, "grad_norm": 9.596959114074707, "learning_rate": 9.8674e-06, "loss": 0.1587, "step": 664 }, { "epoch": 0.03545343071919817, "grad_norm": 6.952032089233398, "learning_rate": 9.867200000000001e-06, "loss": 0.1275, "step": 665 }, { "epoch": 0.0355067441488511, "grad_norm": 7.567192077636719, "learning_rate": 9.867e-06, "loss": 0.1799, "step": 666 }, { "epoch": 0.03556005757850403, "grad_norm": 15.743369102478027, "learning_rate": 9.8668e-06, "loss": 0.2066, "step": 667 }, { "epoch": 0.03561337100815695, "grad_norm": 13.490986824035645, "learning_rate": 9.866600000000001e-06, "loss": 0.1986, "step": 668 }, { "epoch": 0.03566668443780988, "grad_norm": 6.767172336578369, "learning_rate": 9.866400000000001e-06, "loss": 0.1308, "step": 669 }, { "epoch": 0.03571999786746281, "grad_norm": 25.243499755859375, "learning_rate": 9.8662e-06, "loss": 0.1699, "step": 670 }, { "epoch": 0.035773311297115744, "grad_norm": 21.12540054321289, "learning_rate": 9.866000000000002e-06, "loss": 0.188, "step": 671 }, { "epoch": 0.035826624726768674, "grad_norm": 7.818428039550781, "learning_rate": 9.865800000000001e-06, "loss": 0.1389, "step": 672 }, { "epoch": 0.035879938156421605, "grad_norm": 19.670835494995117, "learning_rate": 9.865600000000001e-06, "loss": 0.2037, "step": 673 }, { "epoch": 0.035933251586074536, "grad_norm": 18.912912368774414, "learning_rate": 9.8654e-06, "loss": 0.1452, "step": 674 }, { "epoch": 0.03598656501572746, "grad_norm": 13.858796119689941, "learning_rate": 9.8652e-06, "loss": 0.1603, "step": 675 }, { "epoch": 0.03603987844538039, "grad_norm": 11.570991516113281, "learning_rate": 9.865000000000001e-06, "loss": 0.1605, "step": 676 }, { "epoch": 0.03609319187503332, "grad_norm": 17.67969512939453, "learning_rate": 9.8648e-06, "loss": 0.1655, "step": 677 }, { "epoch": 0.03614650530468625, "grad_norm": 16.726680755615234, "learning_rate": 9.8646e-06, "loss": 0.1749, "step": 678 }, { "epoch": 0.03619981873433918, "grad_norm": 13.14844036102295, "learning_rate": 9.8644e-06, "loss": 0.1807, "step": 679 }, { "epoch": 0.03625313216399211, "grad_norm": 13.429925918579102, "learning_rate": 9.864200000000001e-06, "loss": 0.1591, "step": 680 }, { "epoch": 0.03630644559364504, "grad_norm": 19.2500057220459, "learning_rate": 9.864e-06, "loss": 0.187, "step": 681 }, { "epoch": 0.036359759023297966, "grad_norm": 9.26648998260498, "learning_rate": 9.8638e-06, "loss": 0.1247, "step": 682 }, { "epoch": 0.0364130724529509, "grad_norm": 7.552212238311768, "learning_rate": 9.863600000000002e-06, "loss": 0.1402, "step": 683 }, { "epoch": 0.03646638588260383, "grad_norm": 8.142904281616211, "learning_rate": 9.863400000000001e-06, "loss": 0.1584, "step": 684 }, { "epoch": 0.03651969931225676, "grad_norm": 12.102801322937012, "learning_rate": 9.8632e-06, "loss": 0.1383, "step": 685 }, { "epoch": 0.03657301274190969, "grad_norm": 22.135438919067383, "learning_rate": 9.863e-06, "loss": 0.181, "step": 686 }, { "epoch": 0.03662632617156262, "grad_norm": 4.892940044403076, "learning_rate": 9.862800000000002e-06, "loss": 0.1318, "step": 687 }, { "epoch": 0.03667963960121555, "grad_norm": 13.095974922180176, "learning_rate": 9.862600000000001e-06, "loss": 0.1685, "step": 688 }, { "epoch": 0.03673295303086847, "grad_norm": 25.77857780456543, "learning_rate": 9.8624e-06, "loss": 0.2334, "step": 689 }, { "epoch": 0.036786266460521404, "grad_norm": 6.979545593261719, "learning_rate": 9.8622e-06, "loss": 0.1287, "step": 690 }, { "epoch": 0.036839579890174334, "grad_norm": 7.192805767059326, "learning_rate": 9.862e-06, "loss": 0.1481, "step": 691 }, { "epoch": 0.036892893319827265, "grad_norm": 15.506162643432617, "learning_rate": 9.861800000000001e-06, "loss": 0.1723, "step": 692 }, { "epoch": 0.036946206749480195, "grad_norm": 17.46891212463379, "learning_rate": 9.8616e-06, "loss": 0.1676, "step": 693 }, { "epoch": 0.036999520179133126, "grad_norm": 9.919028282165527, "learning_rate": 9.8614e-06, "loss": 0.1772, "step": 694 }, { "epoch": 0.03705283360878606, "grad_norm": 14.324204444885254, "learning_rate": 9.861200000000001e-06, "loss": 0.1697, "step": 695 }, { "epoch": 0.03710614703843898, "grad_norm": 15.536725044250488, "learning_rate": 9.861000000000001e-06, "loss": 0.1698, "step": 696 }, { "epoch": 0.03715946046809191, "grad_norm": 12.882040977478027, "learning_rate": 9.8608e-06, "loss": 0.1049, "step": 697 }, { "epoch": 0.03721277389774484, "grad_norm": 5.65929651260376, "learning_rate": 9.860600000000002e-06, "loss": 0.1036, "step": 698 }, { "epoch": 0.03726608732739777, "grad_norm": 17.63448143005371, "learning_rate": 9.860400000000001e-06, "loss": 0.1367, "step": 699 }, { "epoch": 0.0373194007570507, "grad_norm": 18.922258377075195, "learning_rate": 9.860200000000001e-06, "loss": 0.1575, "step": 700 }, { "epoch": 0.03737271418670363, "grad_norm": 5.622760772705078, "learning_rate": 9.86e-06, "loss": 0.1362, "step": 701 }, { "epoch": 0.037426027616356564, "grad_norm": 5.475443363189697, "learning_rate": 9.8598e-06, "loss": 0.1383, "step": 702 }, { "epoch": 0.03747934104600949, "grad_norm": 15.910670280456543, "learning_rate": 9.8596e-06, "loss": 0.1899, "step": 703 }, { "epoch": 0.03753265447566242, "grad_norm": 11.7489652633667, "learning_rate": 9.8594e-06, "loss": 0.169, "step": 704 }, { "epoch": 0.03758596790531535, "grad_norm": 6.566220283508301, "learning_rate": 9.8592e-06, "loss": 0.1635, "step": 705 }, { "epoch": 0.03763928133496828, "grad_norm": 8.469725608825684, "learning_rate": 9.859e-06, "loss": 0.1445, "step": 706 }, { "epoch": 0.03769259476462121, "grad_norm": 11.743858337402344, "learning_rate": 9.858800000000001e-06, "loss": 0.1609, "step": 707 }, { "epoch": 0.03774590819427414, "grad_norm": 14.182398796081543, "learning_rate": 9.8586e-06, "loss": 0.1303, "step": 708 }, { "epoch": 0.03779922162392707, "grad_norm": 5.900699138641357, "learning_rate": 9.858400000000002e-06, "loss": 0.1431, "step": 709 }, { "epoch": 0.037852535053579994, "grad_norm": 5.3469462394714355, "learning_rate": 9.858200000000002e-06, "loss": 0.1297, "step": 710 }, { "epoch": 0.037905848483232925, "grad_norm": 10.524170875549316, "learning_rate": 9.858000000000001e-06, "loss": 0.1592, "step": 711 }, { "epoch": 0.037959161912885855, "grad_norm": 8.054459571838379, "learning_rate": 9.8578e-06, "loss": 0.1923, "step": 712 }, { "epoch": 0.038012475342538786, "grad_norm": 8.8856201171875, "learning_rate": 9.8576e-06, "loss": 0.1429, "step": 713 }, { "epoch": 0.038065788772191717, "grad_norm": 8.775568962097168, "learning_rate": 9.8574e-06, "loss": 0.1199, "step": 714 }, { "epoch": 0.03811910220184465, "grad_norm": 8.770369529724121, "learning_rate": 9.857200000000001e-06, "loss": 0.1663, "step": 715 }, { "epoch": 0.03817241563149758, "grad_norm": 9.581830024719238, "learning_rate": 9.857e-06, "loss": 0.1328, "step": 716 }, { "epoch": 0.0382257290611505, "grad_norm": 11.665273666381836, "learning_rate": 9.8568e-06, "loss": 0.1595, "step": 717 }, { "epoch": 0.03827904249080343, "grad_norm": 5.104519367218018, "learning_rate": 9.8566e-06, "loss": 0.1275, "step": 718 }, { "epoch": 0.03833235592045636, "grad_norm": 7.785119533538818, "learning_rate": 9.856400000000001e-06, "loss": 0.1796, "step": 719 }, { "epoch": 0.03838566935010929, "grad_norm": 10.97070026397705, "learning_rate": 9.8562e-06, "loss": 0.1698, "step": 720 }, { "epoch": 0.038438982779762224, "grad_norm": 6.975160598754883, "learning_rate": 9.856000000000002e-06, "loss": 0.1585, "step": 721 }, { "epoch": 0.038492296209415154, "grad_norm": 6.480014801025391, "learning_rate": 9.855800000000001e-06, "loss": 0.124, "step": 722 }, { "epoch": 0.038545609639068085, "grad_norm": 8.662135124206543, "learning_rate": 9.855600000000001e-06, "loss": 0.1494, "step": 723 }, { "epoch": 0.03859892306872101, "grad_norm": 12.88713264465332, "learning_rate": 9.8554e-06, "loss": 0.1232, "step": 724 }, { "epoch": 0.03865223649837394, "grad_norm": 7.104633808135986, "learning_rate": 9.8552e-06, "loss": 0.1173, "step": 725 }, { "epoch": 0.03870554992802687, "grad_norm": 5.314774990081787, "learning_rate": 9.855000000000001e-06, "loss": 0.111, "step": 726 }, { "epoch": 0.0387588633576798, "grad_norm": 6.036491394042969, "learning_rate": 9.854800000000001e-06, "loss": 0.1022, "step": 727 }, { "epoch": 0.03881217678733273, "grad_norm": 7.7759809494018555, "learning_rate": 9.8546e-06, "loss": 0.1512, "step": 728 }, { "epoch": 0.03886549021698566, "grad_norm": 8.126455307006836, "learning_rate": 9.8544e-06, "loss": 0.1511, "step": 729 }, { "epoch": 0.03891880364663859, "grad_norm": 8.781806945800781, "learning_rate": 9.8542e-06, "loss": 0.1039, "step": 730 }, { "epoch": 0.038972117076291515, "grad_norm": 11.33977222442627, "learning_rate": 9.854000000000001e-06, "loss": 0.1378, "step": 731 }, { "epoch": 0.039025430505944446, "grad_norm": 5.809812068939209, "learning_rate": 9.8538e-06, "loss": 0.1083, "step": 732 }, { "epoch": 0.039078743935597376, "grad_norm": 9.979913711547852, "learning_rate": 9.853600000000002e-06, "loss": 0.1397, "step": 733 }, { "epoch": 0.03913205736525031, "grad_norm": 7.510994911193848, "learning_rate": 9.853400000000001e-06, "loss": 0.1651, "step": 734 }, { "epoch": 0.03918537079490324, "grad_norm": 6.563389778137207, "learning_rate": 9.8532e-06, "loss": 0.0978, "step": 735 }, { "epoch": 0.03923868422455617, "grad_norm": 11.384450912475586, "learning_rate": 9.853e-06, "loss": 0.1634, "step": 736 }, { "epoch": 0.0392919976542091, "grad_norm": 6.81398868560791, "learning_rate": 9.852800000000002e-06, "loss": 0.1663, "step": 737 }, { "epoch": 0.03934531108386202, "grad_norm": 10.026250839233398, "learning_rate": 9.852600000000001e-06, "loss": 0.0811, "step": 738 }, { "epoch": 0.03939862451351495, "grad_norm": 10.820765495300293, "learning_rate": 9.8524e-06, "loss": 0.1701, "step": 739 }, { "epoch": 0.039451937943167883, "grad_norm": 5.794950485229492, "learning_rate": 9.8522e-06, "loss": 0.1167, "step": 740 }, { "epoch": 0.039505251372820814, "grad_norm": 5.703094482421875, "learning_rate": 9.852e-06, "loss": 0.1017, "step": 741 }, { "epoch": 0.039558564802473745, "grad_norm": 7.268002033233643, "learning_rate": 9.8518e-06, "loss": 0.1359, "step": 742 }, { "epoch": 0.039611878232126675, "grad_norm": 10.416629791259766, "learning_rate": 9.8516e-06, "loss": 0.1474, "step": 743 }, { "epoch": 0.0396651916617796, "grad_norm": 8.337390899658203, "learning_rate": 9.8514e-06, "loss": 0.1551, "step": 744 }, { "epoch": 0.03971850509143253, "grad_norm": 5.6382927894592285, "learning_rate": 9.851200000000001e-06, "loss": 0.0974, "step": 745 }, { "epoch": 0.03977181852108546, "grad_norm": 14.053211212158203, "learning_rate": 9.851000000000001e-06, "loss": 0.182, "step": 746 }, { "epoch": 0.03982513195073839, "grad_norm": 15.184633255004883, "learning_rate": 9.8508e-06, "loss": 0.1564, "step": 747 }, { "epoch": 0.03987844538039132, "grad_norm": 11.050532341003418, "learning_rate": 9.850600000000002e-06, "loss": 0.1743, "step": 748 }, { "epoch": 0.03993175881004425, "grad_norm": 11.493996620178223, "learning_rate": 9.850400000000001e-06, "loss": 0.141, "step": 749 }, { "epoch": 0.03998507223969718, "grad_norm": 28.420669555664062, "learning_rate": 9.850200000000001e-06, "loss": 0.2237, "step": 750 }, { "epoch": 0.040038385669350106, "grad_norm": 14.627198219299316, "learning_rate": 9.85e-06, "loss": 0.1444, "step": 751 }, { "epoch": 0.040091699099003036, "grad_norm": 11.451884269714355, "learning_rate": 9.8498e-06, "loss": 0.1818, "step": 752 }, { "epoch": 0.04014501252865597, "grad_norm": 13.404989242553711, "learning_rate": 9.8496e-06, "loss": 0.1655, "step": 753 }, { "epoch": 0.0401983259583089, "grad_norm": 20.123502731323242, "learning_rate": 9.849400000000001e-06, "loss": 0.1899, "step": 754 }, { "epoch": 0.04025163938796183, "grad_norm": 4.480038166046143, "learning_rate": 9.8492e-06, "loss": 0.1298, "step": 755 }, { "epoch": 0.04030495281761476, "grad_norm": 8.81839370727539, "learning_rate": 9.849e-06, "loss": 0.1385, "step": 756 }, { "epoch": 0.04035826624726769, "grad_norm": 10.424925804138184, "learning_rate": 9.848800000000001e-06, "loss": 0.1786, "step": 757 }, { "epoch": 0.04041157967692061, "grad_norm": 16.4183292388916, "learning_rate": 9.848600000000001e-06, "loss": 0.1534, "step": 758 }, { "epoch": 0.04046489310657354, "grad_norm": 6.593957901000977, "learning_rate": 9.848400000000002e-06, "loss": 0.1587, "step": 759 }, { "epoch": 0.040518206536226474, "grad_norm": 7.721549987792969, "learning_rate": 9.848200000000002e-06, "loss": 0.1466, "step": 760 }, { "epoch": 0.040571519965879405, "grad_norm": 8.815919876098633, "learning_rate": 9.848000000000001e-06, "loss": 0.119, "step": 761 }, { "epoch": 0.040624833395532335, "grad_norm": 11.193181037902832, "learning_rate": 9.8478e-06, "loss": 0.1398, "step": 762 }, { "epoch": 0.040678146825185266, "grad_norm": 8.226603507995605, "learning_rate": 9.8476e-06, "loss": 0.1278, "step": 763 }, { "epoch": 0.040731460254838196, "grad_norm": 9.067766189575195, "learning_rate": 9.8474e-06, "loss": 0.1808, "step": 764 }, { "epoch": 0.04078477368449112, "grad_norm": 7.074794769287109, "learning_rate": 9.847200000000001e-06, "loss": 0.1255, "step": 765 }, { "epoch": 0.04083808711414405, "grad_norm": 6.360619068145752, "learning_rate": 9.847e-06, "loss": 0.1533, "step": 766 }, { "epoch": 0.04089140054379698, "grad_norm": 17.11484718322754, "learning_rate": 9.8468e-06, "loss": 0.1789, "step": 767 }, { "epoch": 0.04094471397344991, "grad_norm": 11.702899932861328, "learning_rate": 9.8466e-06, "loss": 0.1446, "step": 768 }, { "epoch": 0.04099802740310284, "grad_norm": 7.214894771575928, "learning_rate": 9.846400000000001e-06, "loss": 0.1438, "step": 769 }, { "epoch": 0.04105134083275577, "grad_norm": 9.746734619140625, "learning_rate": 9.8462e-06, "loss": 0.1765, "step": 770 }, { "epoch": 0.0411046542624087, "grad_norm": 6.518383026123047, "learning_rate": 9.846000000000002e-06, "loss": 0.1025, "step": 771 }, { "epoch": 0.04115796769206163, "grad_norm": 8.176795959472656, "learning_rate": 9.845800000000002e-06, "loss": 0.1013, "step": 772 }, { "epoch": 0.04121128112171456, "grad_norm": 3.8721113204956055, "learning_rate": 9.845600000000001e-06, "loss": 0.1083, "step": 773 }, { "epoch": 0.04126459455136749, "grad_norm": 6.452846050262451, "learning_rate": 9.8454e-06, "loss": 0.1501, "step": 774 }, { "epoch": 0.04131790798102042, "grad_norm": 13.588703155517578, "learning_rate": 9.8452e-06, "loss": 0.1419, "step": 775 }, { "epoch": 0.04137122141067335, "grad_norm": 8.225605010986328, "learning_rate": 9.845000000000001e-06, "loss": 0.1312, "step": 776 }, { "epoch": 0.04142453484032628, "grad_norm": 9.411690711975098, "learning_rate": 9.844800000000001e-06, "loss": 0.1542, "step": 777 }, { "epoch": 0.04147784826997921, "grad_norm": 5.455034255981445, "learning_rate": 9.8446e-06, "loss": 0.1286, "step": 778 }, { "epoch": 0.041531161699632134, "grad_norm": 4.878549098968506, "learning_rate": 9.8444e-06, "loss": 0.1057, "step": 779 }, { "epoch": 0.041584475129285065, "grad_norm": 8.638808250427246, "learning_rate": 9.8442e-06, "loss": 0.141, "step": 780 }, { "epoch": 0.041637788558937995, "grad_norm": 5.992186069488525, "learning_rate": 9.844000000000001e-06, "loss": 0.123, "step": 781 }, { "epoch": 0.041691101988590926, "grad_norm": 7.194914817810059, "learning_rate": 9.8438e-06, "loss": 0.1483, "step": 782 }, { "epoch": 0.041744415418243856, "grad_norm": 8.430578231811523, "learning_rate": 9.843600000000002e-06, "loss": 0.0979, "step": 783 }, { "epoch": 0.04179772884789679, "grad_norm": 5.388896465301514, "learning_rate": 9.843400000000001e-06, "loss": 0.121, "step": 784 }, { "epoch": 0.04185104227754972, "grad_norm": 7.6692657470703125, "learning_rate": 9.843200000000001e-06, "loss": 0.1241, "step": 785 }, { "epoch": 0.04190435570720264, "grad_norm": 5.395076751708984, "learning_rate": 9.843e-06, "loss": 0.1376, "step": 786 }, { "epoch": 0.04195766913685557, "grad_norm": 9.474895477294922, "learning_rate": 9.842800000000002e-06, "loss": 0.2066, "step": 787 }, { "epoch": 0.0420109825665085, "grad_norm": 7.2836012840271, "learning_rate": 9.842600000000001e-06, "loss": 0.0885, "step": 788 }, { "epoch": 0.04206429599616143, "grad_norm": 9.998619079589844, "learning_rate": 9.8424e-06, "loss": 0.1608, "step": 789 }, { "epoch": 0.04211760942581436, "grad_norm": 6.095553398132324, "learning_rate": 9.8422e-06, "loss": 0.1484, "step": 790 }, { "epoch": 0.042170922855467294, "grad_norm": 7.892253875732422, "learning_rate": 9.842e-06, "loss": 0.1364, "step": 791 }, { "epoch": 0.042224236285120224, "grad_norm": 5.146937370300293, "learning_rate": 9.841800000000001e-06, "loss": 0.098, "step": 792 }, { "epoch": 0.04227754971477315, "grad_norm": 14.321866035461426, "learning_rate": 9.8416e-06, "loss": 0.1231, "step": 793 }, { "epoch": 0.04233086314442608, "grad_norm": 5.309294700622559, "learning_rate": 9.8414e-06, "loss": 0.1201, "step": 794 }, { "epoch": 0.04238417657407901, "grad_norm": 10.359244346618652, "learning_rate": 9.841200000000002e-06, "loss": 0.1082, "step": 795 }, { "epoch": 0.04243749000373194, "grad_norm": 19.52107048034668, "learning_rate": 9.841000000000001e-06, "loss": 0.1491, "step": 796 }, { "epoch": 0.04249080343338487, "grad_norm": 14.072019577026367, "learning_rate": 9.8408e-06, "loss": 0.1612, "step": 797 }, { "epoch": 0.0425441168630378, "grad_norm": 8.404019355773926, "learning_rate": 9.840600000000002e-06, "loss": 0.116, "step": 798 }, { "epoch": 0.04259743029269073, "grad_norm": 18.09164810180664, "learning_rate": 9.840400000000002e-06, "loss": 0.1685, "step": 799 }, { "epoch": 0.042650743722343655, "grad_norm": 18.36307144165039, "learning_rate": 9.840200000000001e-06, "loss": 0.1964, "step": 800 }, { "epoch": 0.042704057151996586, "grad_norm": 7.926852226257324, "learning_rate": 9.84e-06, "loss": 0.1378, "step": 801 }, { "epoch": 0.042757370581649516, "grad_norm": 16.50916862487793, "learning_rate": 9.8398e-06, "loss": 0.1033, "step": 802 }, { "epoch": 0.04281068401130245, "grad_norm": 13.34975814819336, "learning_rate": 9.8396e-06, "loss": 0.1558, "step": 803 }, { "epoch": 0.04286399744095538, "grad_norm": 10.186304092407227, "learning_rate": 9.839400000000001e-06, "loss": 0.1632, "step": 804 }, { "epoch": 0.04291731087060831, "grad_norm": 11.507972717285156, "learning_rate": 9.8392e-06, "loss": 0.1387, "step": 805 }, { "epoch": 0.04297062430026124, "grad_norm": 6.065525054931641, "learning_rate": 9.839e-06, "loss": 0.1258, "step": 806 }, { "epoch": 0.04302393772991416, "grad_norm": 9.818326950073242, "learning_rate": 9.838800000000001e-06, "loss": 0.1517, "step": 807 }, { "epoch": 0.04307725115956709, "grad_norm": 11.665841102600098, "learning_rate": 9.838600000000001e-06, "loss": 0.098, "step": 808 }, { "epoch": 0.04313056458922002, "grad_norm": 6.3654961585998535, "learning_rate": 9.8384e-06, "loss": 0.1481, "step": 809 }, { "epoch": 0.043183878018872954, "grad_norm": 8.87704086303711, "learning_rate": 9.838200000000002e-06, "loss": 0.1034, "step": 810 }, { "epoch": 0.043237191448525884, "grad_norm": 8.40369701385498, "learning_rate": 9.838000000000001e-06, "loss": 0.1231, "step": 811 }, { "epoch": 0.043290504878178815, "grad_norm": 5.364266872406006, "learning_rate": 9.837800000000001e-06, "loss": 0.0906, "step": 812 }, { "epoch": 0.043343818307831745, "grad_norm": 5.400266170501709, "learning_rate": 9.8376e-06, "loss": 0.0958, "step": 813 }, { "epoch": 0.04339713173748467, "grad_norm": 10.411415100097656, "learning_rate": 9.8374e-06, "loss": 0.1048, "step": 814 }, { "epoch": 0.0434504451671376, "grad_norm": 9.503265380859375, "learning_rate": 9.837200000000001e-06, "loss": 0.098, "step": 815 }, { "epoch": 0.04350375859679053, "grad_norm": 16.86316680908203, "learning_rate": 9.837000000000001e-06, "loss": 0.1521, "step": 816 }, { "epoch": 0.04355707202644346, "grad_norm": 7.156310081481934, "learning_rate": 9.8368e-06, "loss": 0.1368, "step": 817 }, { "epoch": 0.04361038545609639, "grad_norm": 10.563570976257324, "learning_rate": 9.8366e-06, "loss": 0.1491, "step": 818 }, { "epoch": 0.04366369888574932, "grad_norm": 8.851905822753906, "learning_rate": 9.836400000000001e-06, "loss": 0.1848, "step": 819 }, { "epoch": 0.04371701231540225, "grad_norm": 6.478482723236084, "learning_rate": 9.8362e-06, "loss": 0.1449, "step": 820 }, { "epoch": 0.043770325745055176, "grad_norm": 12.022374153137207, "learning_rate": 9.836e-06, "loss": 0.0995, "step": 821 }, { "epoch": 0.04382363917470811, "grad_norm": 17.303314208984375, "learning_rate": 9.835800000000002e-06, "loss": 0.1942, "step": 822 }, { "epoch": 0.04387695260436104, "grad_norm": 7.035817623138428, "learning_rate": 9.835600000000001e-06, "loss": 0.1423, "step": 823 }, { "epoch": 0.04393026603401397, "grad_norm": 7.615351676940918, "learning_rate": 9.8354e-06, "loss": 0.1165, "step": 824 }, { "epoch": 0.0439835794636669, "grad_norm": 10.3641939163208, "learning_rate": 9.8352e-06, "loss": 0.1087, "step": 825 }, { "epoch": 0.04403689289331983, "grad_norm": 26.339590072631836, "learning_rate": 9.835000000000002e-06, "loss": 0.1455, "step": 826 }, { "epoch": 0.04409020632297276, "grad_norm": 5.865658283233643, "learning_rate": 9.834800000000001e-06, "loss": 0.1041, "step": 827 }, { "epoch": 0.04414351975262568, "grad_norm": 10.97970962524414, "learning_rate": 9.8346e-06, "loss": 0.1561, "step": 828 }, { "epoch": 0.044196833182278614, "grad_norm": 15.764129638671875, "learning_rate": 9.8344e-06, "loss": 0.1931, "step": 829 }, { "epoch": 0.044250146611931544, "grad_norm": 7.483834266662598, "learning_rate": 9.8342e-06, "loss": 0.0964, "step": 830 }, { "epoch": 0.044303460041584475, "grad_norm": 7.374724388122559, "learning_rate": 9.834000000000001e-06, "loss": 0.0781, "step": 831 }, { "epoch": 0.044356773471237405, "grad_norm": 5.116653919219971, "learning_rate": 9.8338e-06, "loss": 0.1089, "step": 832 }, { "epoch": 0.044410086900890336, "grad_norm": 13.740887641906738, "learning_rate": 9.8336e-06, "loss": 0.1768, "step": 833 }, { "epoch": 0.044463400330543267, "grad_norm": 20.29261589050293, "learning_rate": 9.833400000000001e-06, "loss": 0.1768, "step": 834 }, { "epoch": 0.04451671376019619, "grad_norm": 11.68218994140625, "learning_rate": 9.833200000000001e-06, "loss": 0.1184, "step": 835 }, { "epoch": 0.04457002718984912, "grad_norm": 7.319336414337158, "learning_rate": 9.833e-06, "loss": 0.1096, "step": 836 }, { "epoch": 0.04462334061950205, "grad_norm": 8.017800331115723, "learning_rate": 9.832800000000002e-06, "loss": 0.1146, "step": 837 }, { "epoch": 0.04467665404915498, "grad_norm": 21.37534523010254, "learning_rate": 9.832600000000001e-06, "loss": 0.2155, "step": 838 }, { "epoch": 0.04472996747880791, "grad_norm": 9.415364265441895, "learning_rate": 9.832400000000001e-06, "loss": 0.1381, "step": 839 }, { "epoch": 0.04478328090846084, "grad_norm": 5.71196174621582, "learning_rate": 9.8322e-06, "loss": 0.0947, "step": 840 }, { "epoch": 0.044836594338113774, "grad_norm": 10.641891479492188, "learning_rate": 9.832e-06, "loss": 0.1439, "step": 841 }, { "epoch": 0.0448899077677667, "grad_norm": 14.238262176513672, "learning_rate": 9.831800000000001e-06, "loss": 0.163, "step": 842 }, { "epoch": 0.04494322119741963, "grad_norm": 13.50958251953125, "learning_rate": 9.831600000000001e-06, "loss": 0.1311, "step": 843 }, { "epoch": 0.04499653462707256, "grad_norm": 6.112170219421387, "learning_rate": 9.8314e-06, "loss": 0.1136, "step": 844 }, { "epoch": 0.04504984805672549, "grad_norm": 10.587234497070312, "learning_rate": 9.8312e-06, "loss": 0.1235, "step": 845 }, { "epoch": 0.04510316148637842, "grad_norm": 7.172475814819336, "learning_rate": 9.831000000000001e-06, "loss": 0.1126, "step": 846 }, { "epoch": 0.04515647491603135, "grad_norm": 12.17352294921875, "learning_rate": 9.8308e-06, "loss": 0.103, "step": 847 }, { "epoch": 0.04520978834568428, "grad_norm": 9.749699592590332, "learning_rate": 9.830600000000002e-06, "loss": 0.124, "step": 848 }, { "epoch": 0.045263101775337204, "grad_norm": 5.610562324523926, "learning_rate": 9.830400000000002e-06, "loss": 0.1394, "step": 849 }, { "epoch": 0.045316415204990135, "grad_norm": 32.798946380615234, "learning_rate": 9.830200000000001e-06, "loss": 0.1821, "step": 850 }, { "epoch": 0.045369728634643065, "grad_norm": 16.20484161376953, "learning_rate": 9.83e-06, "loss": 0.1232, "step": 851 }, { "epoch": 0.045423042064295996, "grad_norm": 7.544561862945557, "learning_rate": 9.8298e-06, "loss": 0.1413, "step": 852 }, { "epoch": 0.045476355493948926, "grad_norm": 15.482837677001953, "learning_rate": 9.8296e-06, "loss": 0.1428, "step": 853 }, { "epoch": 0.04552966892360186, "grad_norm": 14.368240356445312, "learning_rate": 9.829400000000001e-06, "loss": 0.1478, "step": 854 }, { "epoch": 0.04558298235325479, "grad_norm": 11.523109436035156, "learning_rate": 9.8292e-06, "loss": 0.1645, "step": 855 }, { "epoch": 0.04563629578290771, "grad_norm": 7.093719959259033, "learning_rate": 9.829e-06, "loss": 0.113, "step": 856 }, { "epoch": 0.04568960921256064, "grad_norm": 20.180252075195312, "learning_rate": 9.8288e-06, "loss": 0.1575, "step": 857 }, { "epoch": 0.04574292264221357, "grad_norm": 17.790803909301758, "learning_rate": 9.828600000000001e-06, "loss": 0.2466, "step": 858 }, { "epoch": 0.0457962360718665, "grad_norm": 13.82046127319336, "learning_rate": 9.8284e-06, "loss": 0.1289, "step": 859 }, { "epoch": 0.045849549501519433, "grad_norm": 13.279053688049316, "learning_rate": 9.828200000000002e-06, "loss": 0.1121, "step": 860 }, { "epoch": 0.045902862931172364, "grad_norm": 14.9869384765625, "learning_rate": 9.828000000000001e-06, "loss": 0.1198, "step": 861 }, { "epoch": 0.045956176360825295, "grad_norm": 7.622185707092285, "learning_rate": 9.827800000000001e-06, "loss": 0.1513, "step": 862 }, { "epoch": 0.04600948979047822, "grad_norm": 12.314472198486328, "learning_rate": 9.8276e-06, "loss": 0.1833, "step": 863 }, { "epoch": 0.04606280322013115, "grad_norm": 15.01047134399414, "learning_rate": 9.8274e-06, "loss": 0.2174, "step": 864 }, { "epoch": 0.04611611664978408, "grad_norm": 15.44912338256836, "learning_rate": 9.827200000000001e-06, "loss": 0.1432, "step": 865 }, { "epoch": 0.04616943007943701, "grad_norm": 5.560319423675537, "learning_rate": 9.827000000000001e-06, "loss": 0.0991, "step": 866 }, { "epoch": 0.04622274350908994, "grad_norm": 7.815497398376465, "learning_rate": 9.8268e-06, "loss": 0.1431, "step": 867 }, { "epoch": 0.04627605693874287, "grad_norm": 7.344640254974365, "learning_rate": 9.8266e-06, "loss": 0.1021, "step": 868 }, { "epoch": 0.0463293703683958, "grad_norm": 17.928546905517578, "learning_rate": 9.8264e-06, "loss": 0.1213, "step": 869 }, { "epoch": 0.046382683798048725, "grad_norm": 17.35573387145996, "learning_rate": 9.826200000000001e-06, "loss": 0.1933, "step": 870 }, { "epoch": 0.046435997227701656, "grad_norm": 13.161453247070312, "learning_rate": 9.826e-06, "loss": 0.1736, "step": 871 }, { "epoch": 0.046489310657354586, "grad_norm": 6.046862602233887, "learning_rate": 9.825800000000002e-06, "loss": 0.1469, "step": 872 }, { "epoch": 0.04654262408700752, "grad_norm": 10.816205978393555, "learning_rate": 9.825600000000001e-06, "loss": 0.1049, "step": 873 }, { "epoch": 0.04659593751666045, "grad_norm": 17.492551803588867, "learning_rate": 9.8254e-06, "loss": 0.147, "step": 874 }, { "epoch": 0.04664925094631338, "grad_norm": 21.013256072998047, "learning_rate": 9.8252e-06, "loss": 0.1737, "step": 875 }, { "epoch": 0.04670256437596631, "grad_norm": 5.338113307952881, "learning_rate": 9.825000000000002e-06, "loss": 0.1136, "step": 876 }, { "epoch": 0.04675587780561923, "grad_norm": 13.15307903289795, "learning_rate": 9.824800000000001e-06, "loss": 0.1393, "step": 877 }, { "epoch": 0.04680919123527216, "grad_norm": 9.9210786819458, "learning_rate": 9.8246e-06, "loss": 0.1527, "step": 878 }, { "epoch": 0.04686250466492509, "grad_norm": 12.640941619873047, "learning_rate": 9.8244e-06, "loss": 0.1588, "step": 879 }, { "epoch": 0.046915818094578024, "grad_norm": 9.145071029663086, "learning_rate": 9.8242e-06, "loss": 0.1111, "step": 880 }, { "epoch": 0.046969131524230955, "grad_norm": 9.670758247375488, "learning_rate": 9.824000000000001e-06, "loss": 0.184, "step": 881 }, { "epoch": 0.047022444953883885, "grad_norm": 19.325942993164062, "learning_rate": 9.8238e-06, "loss": 0.2038, "step": 882 }, { "epoch": 0.047075758383536816, "grad_norm": 12.713282585144043, "learning_rate": 9.8236e-06, "loss": 0.138, "step": 883 }, { "epoch": 0.04712907181318974, "grad_norm": 7.738716125488281, "learning_rate": 9.823400000000002e-06, "loss": 0.1343, "step": 884 }, { "epoch": 0.04718238524284267, "grad_norm": 5.280039310455322, "learning_rate": 9.823200000000001e-06, "loss": 0.1138, "step": 885 }, { "epoch": 0.0472356986724956, "grad_norm": 25.37277603149414, "learning_rate": 9.823e-06, "loss": 0.1894, "step": 886 }, { "epoch": 0.04728901210214853, "grad_norm": 5.800923824310303, "learning_rate": 9.822800000000002e-06, "loss": 0.1032, "step": 887 }, { "epoch": 0.04734232553180146, "grad_norm": 5.5029120445251465, "learning_rate": 9.822600000000001e-06, "loss": 0.1541, "step": 888 }, { "epoch": 0.04739563896145439, "grad_norm": 7.640103816986084, "learning_rate": 9.822400000000001e-06, "loss": 0.1112, "step": 889 }, { "epoch": 0.04744895239110732, "grad_norm": 6.719459056854248, "learning_rate": 9.8222e-06, "loss": 0.1322, "step": 890 }, { "epoch": 0.047502265820760246, "grad_norm": 4.949357509613037, "learning_rate": 9.822e-06, "loss": 0.1075, "step": 891 }, { "epoch": 0.04755557925041318, "grad_norm": 24.056442260742188, "learning_rate": 9.821800000000001e-06, "loss": 0.1897, "step": 892 }, { "epoch": 0.04760889268006611, "grad_norm": 13.449034690856934, "learning_rate": 9.821600000000001e-06, "loss": 0.1214, "step": 893 }, { "epoch": 0.04766220610971904, "grad_norm": 8.798563957214355, "learning_rate": 9.8214e-06, "loss": 0.173, "step": 894 }, { "epoch": 0.04771551953937197, "grad_norm": 5.011709690093994, "learning_rate": 9.8212e-06, "loss": 0.0863, "step": 895 }, { "epoch": 0.0477688329690249, "grad_norm": 6.8700103759765625, "learning_rate": 9.821000000000001e-06, "loss": 0.1327, "step": 896 }, { "epoch": 0.04782214639867783, "grad_norm": 8.13134765625, "learning_rate": 9.820800000000001e-06, "loss": 0.0943, "step": 897 }, { "epoch": 0.04787545982833075, "grad_norm": 18.423908233642578, "learning_rate": 9.8206e-06, "loss": 0.184, "step": 898 }, { "epoch": 0.047928773257983684, "grad_norm": 10.482346534729004, "learning_rate": 9.820400000000002e-06, "loss": 0.121, "step": 899 }, { "epoch": 0.047982086687636614, "grad_norm": 12.433959007263184, "learning_rate": 9.820200000000001e-06, "loss": 0.2148, "step": 900 }, { "epoch": 0.048035400117289545, "grad_norm": 9.731371879577637, "learning_rate": 9.820000000000001e-06, "loss": 0.1327, "step": 901 }, { "epoch": 0.048088713546942476, "grad_norm": 7.054946422576904, "learning_rate": 9.8198e-06, "loss": 0.1102, "step": 902 }, { "epoch": 0.048142026976595406, "grad_norm": 6.816934108734131, "learning_rate": 9.819600000000002e-06, "loss": 0.0919, "step": 903 }, { "epoch": 0.04819534040624834, "grad_norm": 19.587528228759766, "learning_rate": 9.819400000000001e-06, "loss": 0.2181, "step": 904 }, { "epoch": 0.04824865383590126, "grad_norm": 8.388558387756348, "learning_rate": 9.8192e-06, "loss": 0.134, "step": 905 }, { "epoch": 0.04830196726555419, "grad_norm": 8.535911560058594, "learning_rate": 9.819e-06, "loss": 0.1572, "step": 906 }, { "epoch": 0.04835528069520712, "grad_norm": 10.742433547973633, "learning_rate": 9.8188e-06, "loss": 0.1892, "step": 907 }, { "epoch": 0.04840859412486005, "grad_norm": 7.04490327835083, "learning_rate": 9.818600000000001e-06, "loss": 0.116, "step": 908 }, { "epoch": 0.04846190755451298, "grad_norm": 9.743851661682129, "learning_rate": 9.8184e-06, "loss": 0.1227, "step": 909 }, { "epoch": 0.04851522098416591, "grad_norm": 6.144331455230713, "learning_rate": 9.8182e-06, "loss": 0.1209, "step": 910 }, { "epoch": 0.048568534413818844, "grad_norm": 5.847157001495361, "learning_rate": 9.818000000000002e-06, "loss": 0.083, "step": 911 }, { "epoch": 0.04862184784347177, "grad_norm": 7.85462760925293, "learning_rate": 9.817800000000001e-06, "loss": 0.1127, "step": 912 }, { "epoch": 0.0486751612731247, "grad_norm": 8.946028709411621, "learning_rate": 9.8176e-06, "loss": 0.1338, "step": 913 }, { "epoch": 0.04872847470277763, "grad_norm": 5.5174150466918945, "learning_rate": 9.8174e-06, "loss": 0.1001, "step": 914 }, { "epoch": 0.04878178813243056, "grad_norm": 6.193253993988037, "learning_rate": 9.817200000000001e-06, "loss": 0.1047, "step": 915 }, { "epoch": 0.04883510156208349, "grad_norm": 7.827485084533691, "learning_rate": 9.817000000000001e-06, "loss": 0.1535, "step": 916 }, { "epoch": 0.04888841499173642, "grad_norm": 11.272909164428711, "learning_rate": 9.8168e-06, "loss": 0.1458, "step": 917 }, { "epoch": 0.04894172842138935, "grad_norm": 7.058994293212891, "learning_rate": 9.8166e-06, "loss": 0.1257, "step": 918 }, { "epoch": 0.048995041851042274, "grad_norm": 5.538515567779541, "learning_rate": 9.8164e-06, "loss": 0.1107, "step": 919 }, { "epoch": 0.049048355280695205, "grad_norm": 4.496594429016113, "learning_rate": 9.816200000000001e-06, "loss": 0.084, "step": 920 }, { "epoch": 0.049101668710348136, "grad_norm": 9.399484634399414, "learning_rate": 9.816e-06, "loss": 0.1343, "step": 921 }, { "epoch": 0.049154982140001066, "grad_norm": 7.475164890289307, "learning_rate": 9.8158e-06, "loss": 0.1164, "step": 922 }, { "epoch": 0.049208295569654, "grad_norm": 5.608774662017822, "learning_rate": 9.815600000000001e-06, "loss": 0.1459, "step": 923 }, { "epoch": 0.04926160899930693, "grad_norm": 7.37686014175415, "learning_rate": 9.815400000000001e-06, "loss": 0.1404, "step": 924 }, { "epoch": 0.04931492242895986, "grad_norm": 8.524681091308594, "learning_rate": 9.8152e-06, "loss": 0.134, "step": 925 }, { "epoch": 0.04936823585861278, "grad_norm": 5.7618021965026855, "learning_rate": 9.815000000000002e-06, "loss": 0.092, "step": 926 }, { "epoch": 0.04942154928826571, "grad_norm": 6.267151832580566, "learning_rate": 9.814800000000001e-06, "loss": 0.1016, "step": 927 }, { "epoch": 0.04947486271791864, "grad_norm": 16.277433395385742, "learning_rate": 9.814600000000001e-06, "loss": 0.1441, "step": 928 }, { "epoch": 0.04952817614757157, "grad_norm": 5.04347038269043, "learning_rate": 9.8144e-06, "loss": 0.085, "step": 929 }, { "epoch": 0.049581489577224504, "grad_norm": 6.048042297363281, "learning_rate": 9.8142e-06, "loss": 0.1209, "step": 930 }, { "epoch": 0.049634803006877434, "grad_norm": 5.424015998840332, "learning_rate": 9.814000000000001e-06, "loss": 0.1103, "step": 931 }, { "epoch": 0.049688116436530365, "grad_norm": 7.428493022918701, "learning_rate": 9.8138e-06, "loss": 0.1583, "step": 932 }, { "epoch": 0.04974142986618329, "grad_norm": 5.365775108337402, "learning_rate": 9.8136e-06, "loss": 0.1029, "step": 933 }, { "epoch": 0.04979474329583622, "grad_norm": 8.188775062561035, "learning_rate": 9.8134e-06, "loss": 0.0888, "step": 934 }, { "epoch": 0.04984805672548915, "grad_norm": 4.583959102630615, "learning_rate": 9.813200000000001e-06, "loss": 0.1104, "step": 935 }, { "epoch": 0.04990137015514208, "grad_norm": 6.955982685089111, "learning_rate": 9.813e-06, "loss": 0.1274, "step": 936 }, { "epoch": 0.04995468358479501, "grad_norm": 9.598678588867188, "learning_rate": 9.812800000000002e-06, "loss": 0.1516, "step": 937 }, { "epoch": 0.05000799701444794, "grad_norm": 6.83724308013916, "learning_rate": 9.812600000000002e-06, "loss": 0.0966, "step": 938 }, { "epoch": 0.05006131044410087, "grad_norm": 5.26918363571167, "learning_rate": 9.812400000000001e-06, "loss": 0.1011, "step": 939 }, { "epoch": 0.050114623873753796, "grad_norm": 11.910554885864258, "learning_rate": 9.8122e-06, "loss": 0.1019, "step": 940 }, { "epoch": 0.050167937303406726, "grad_norm": 6.165811538696289, "learning_rate": 9.812e-06, "loss": 0.0907, "step": 941 }, { "epoch": 0.05022125073305966, "grad_norm": 10.116252899169922, "learning_rate": 9.811800000000002e-06, "loss": 0.1498, "step": 942 }, { "epoch": 0.05027456416271259, "grad_norm": 9.357147216796875, "learning_rate": 9.811600000000001e-06, "loss": 0.146, "step": 943 }, { "epoch": 0.05032787759236552, "grad_norm": 7.986752510070801, "learning_rate": 9.8114e-06, "loss": 0.105, "step": 944 }, { "epoch": 0.05038119102201845, "grad_norm": 5.703375339508057, "learning_rate": 9.8112e-06, "loss": 0.0929, "step": 945 }, { "epoch": 0.05043450445167138, "grad_norm": 15.0906343460083, "learning_rate": 9.811e-06, "loss": 0.1241, "step": 946 }, { "epoch": 0.0504878178813243, "grad_norm": 5.633539199829102, "learning_rate": 9.810800000000001e-06, "loss": 0.0828, "step": 947 }, { "epoch": 0.05054113131097723, "grad_norm": 6.781215667724609, "learning_rate": 9.8106e-06, "loss": 0.1408, "step": 948 }, { "epoch": 0.050594444740630164, "grad_norm": 9.250749588012695, "learning_rate": 9.810400000000002e-06, "loss": 0.0832, "step": 949 }, { "epoch": 0.050647758170283094, "grad_norm": 5.535115718841553, "learning_rate": 9.810200000000001e-06, "loss": 0.1007, "step": 950 }, { "epoch": 0.050701071599936025, "grad_norm": 6.847747802734375, "learning_rate": 9.810000000000001e-06, "loss": 0.1695, "step": 951 }, { "epoch": 0.050754385029588955, "grad_norm": 9.282713890075684, "learning_rate": 9.8098e-06, "loss": 0.1292, "step": 952 }, { "epoch": 0.050807698459241886, "grad_norm": 11.37580680847168, "learning_rate": 9.809600000000002e-06, "loss": 0.1129, "step": 953 }, { "epoch": 0.05086101188889481, "grad_norm": 11.85297966003418, "learning_rate": 9.809400000000001e-06, "loss": 0.1463, "step": 954 }, { "epoch": 0.05091432531854774, "grad_norm": 6.965335369110107, "learning_rate": 9.809200000000001e-06, "loss": 0.0773, "step": 955 }, { "epoch": 0.05096763874820067, "grad_norm": 5.236762046813965, "learning_rate": 9.809e-06, "loss": 0.0862, "step": 956 }, { "epoch": 0.0510209521778536, "grad_norm": 10.139512062072754, "learning_rate": 9.8088e-06, "loss": 0.1738, "step": 957 }, { "epoch": 0.05107426560750653, "grad_norm": 5.763803005218506, "learning_rate": 9.8086e-06, "loss": 0.1047, "step": 958 }, { "epoch": 0.05112757903715946, "grad_norm": 11.31713581085205, "learning_rate": 9.8084e-06, "loss": 0.0791, "step": 959 }, { "epoch": 0.05118089246681239, "grad_norm": 6.773923397064209, "learning_rate": 9.8082e-06, "loss": 0.1395, "step": 960 }, { "epoch": 0.05123420589646532, "grad_norm": 5.462005138397217, "learning_rate": 9.808000000000002e-06, "loss": 0.0554, "step": 961 }, { "epoch": 0.05128751932611825, "grad_norm": 14.950578689575195, "learning_rate": 9.807800000000001e-06, "loss": 0.1593, "step": 962 }, { "epoch": 0.05134083275577118, "grad_norm": 13.173341751098633, "learning_rate": 9.8076e-06, "loss": 0.1337, "step": 963 }, { "epoch": 0.05139414618542411, "grad_norm": 5.886234760284424, "learning_rate": 9.8074e-06, "loss": 0.0746, "step": 964 }, { "epoch": 0.05144745961507704, "grad_norm": 12.396269798278809, "learning_rate": 9.807200000000002e-06, "loss": 0.1169, "step": 965 }, { "epoch": 0.05150077304472997, "grad_norm": 15.748106956481934, "learning_rate": 9.807000000000001e-06, "loss": 0.1127, "step": 966 }, { "epoch": 0.0515540864743829, "grad_norm": 13.264166831970215, "learning_rate": 9.8068e-06, "loss": 0.1052, "step": 967 }, { "epoch": 0.051607399904035824, "grad_norm": 11.339934349060059, "learning_rate": 9.8066e-06, "loss": 0.2185, "step": 968 }, { "epoch": 0.051660713333688754, "grad_norm": 11.580388069152832, "learning_rate": 9.8064e-06, "loss": 0.1423, "step": 969 }, { "epoch": 0.051714026763341685, "grad_norm": 10.669148445129395, "learning_rate": 9.806200000000001e-06, "loss": 0.1106, "step": 970 }, { "epoch": 0.051767340192994615, "grad_norm": 6.499359607696533, "learning_rate": 9.806e-06, "loss": 0.1107, "step": 971 }, { "epoch": 0.051820653622647546, "grad_norm": 7.969882488250732, "learning_rate": 9.8058e-06, "loss": 0.1042, "step": 972 }, { "epoch": 0.051873967052300476, "grad_norm": 26.07274627685547, "learning_rate": 9.805600000000001e-06, "loss": 0.1524, "step": 973 }, { "epoch": 0.05192728048195341, "grad_norm": 5.5433430671691895, "learning_rate": 9.805400000000001e-06, "loss": 0.0707, "step": 974 }, { "epoch": 0.05198059391160633, "grad_norm": 6.608109474182129, "learning_rate": 9.8052e-06, "loss": 0.0977, "step": 975 }, { "epoch": 0.05203390734125926, "grad_norm": 6.960915565490723, "learning_rate": 9.805000000000002e-06, "loss": 0.1179, "step": 976 }, { "epoch": 0.05208722077091219, "grad_norm": 9.760954856872559, "learning_rate": 9.804800000000001e-06, "loss": 0.1173, "step": 977 }, { "epoch": 0.05214053420056512, "grad_norm": 11.320228576660156, "learning_rate": 9.804600000000001e-06, "loss": 0.1437, "step": 978 }, { "epoch": 0.05219384763021805, "grad_norm": 5.7564520835876465, "learning_rate": 9.8044e-06, "loss": 0.0859, "step": 979 }, { "epoch": 0.05224716105987098, "grad_norm": 7.4913411140441895, "learning_rate": 9.8042e-06, "loss": 0.17, "step": 980 }, { "epoch": 0.052300474489523914, "grad_norm": 8.187118530273438, "learning_rate": 9.804000000000001e-06, "loss": 0.164, "step": 981 }, { "epoch": 0.05235378791917684, "grad_norm": 7.809186935424805, "learning_rate": 9.803800000000001e-06, "loss": 0.1299, "step": 982 }, { "epoch": 0.05240710134882977, "grad_norm": 12.420278549194336, "learning_rate": 9.8036e-06, "loss": 0.1285, "step": 983 }, { "epoch": 0.0524604147784827, "grad_norm": 8.129678726196289, "learning_rate": 9.8034e-06, "loss": 0.1262, "step": 984 }, { "epoch": 0.05251372820813563, "grad_norm": 5.637617111206055, "learning_rate": 9.803200000000001e-06, "loss": 0.0798, "step": 985 }, { "epoch": 0.05256704163778856, "grad_norm": 6.607453346252441, "learning_rate": 9.803e-06, "loss": 0.1169, "step": 986 }, { "epoch": 0.05262035506744149, "grad_norm": 17.807804107666016, "learning_rate": 9.8028e-06, "loss": 0.1641, "step": 987 }, { "epoch": 0.05267366849709442, "grad_norm": 10.51431941986084, "learning_rate": 9.802600000000002e-06, "loss": 0.0999, "step": 988 }, { "epoch": 0.052726981926747345, "grad_norm": 6.6077470779418945, "learning_rate": 9.802400000000001e-06, "loss": 0.1405, "step": 989 }, { "epoch": 0.052780295356400275, "grad_norm": 10.250544548034668, "learning_rate": 9.8022e-06, "loss": 0.1061, "step": 990 }, { "epoch": 0.052833608786053206, "grad_norm": 14.524435043334961, "learning_rate": 9.802e-06, "loss": 0.1118, "step": 991 }, { "epoch": 0.052886922215706136, "grad_norm": 5.907712936401367, "learning_rate": 9.801800000000002e-06, "loss": 0.1206, "step": 992 }, { "epoch": 0.05294023564535907, "grad_norm": 6.061426639556885, "learning_rate": 9.801600000000001e-06, "loss": 0.1039, "step": 993 }, { "epoch": 0.052993549075012, "grad_norm": 4.4374823570251465, "learning_rate": 9.8014e-06, "loss": 0.1328, "step": 994 }, { "epoch": 0.05304686250466493, "grad_norm": 7.991711139678955, "learning_rate": 9.8012e-06, "loss": 0.15, "step": 995 }, { "epoch": 0.05310017593431785, "grad_norm": 8.692255020141602, "learning_rate": 9.801e-06, "loss": 0.1335, "step": 996 }, { "epoch": 0.05315348936397078, "grad_norm": 4.215648174285889, "learning_rate": 9.800800000000001e-06, "loss": 0.0912, "step": 997 }, { "epoch": 0.05320680279362371, "grad_norm": 6.671176433563232, "learning_rate": 9.8006e-06, "loss": 0.0787, "step": 998 }, { "epoch": 0.05326011622327664, "grad_norm": 8.280072212219238, "learning_rate": 9.8004e-06, "loss": 0.1252, "step": 999 }, { "epoch": 0.053313429652929574, "grad_norm": 5.1779351234436035, "learning_rate": 9.800200000000001e-06, "loss": 0.0769, "step": 1000 }, { "epoch": 0.053366743082582505, "grad_norm": 10.897695541381836, "learning_rate": 9.800000000000001e-06, "loss": 0.0906, "step": 1001 }, { "epoch": 0.053420056512235435, "grad_norm": 5.633361339569092, "learning_rate": 9.7998e-06, "loss": 0.1138, "step": 1002 }, { "epoch": 0.05347336994188836, "grad_norm": 5.7273054122924805, "learning_rate": 9.799600000000002e-06, "loss": 0.128, "step": 1003 }, { "epoch": 0.05352668337154129, "grad_norm": 11.209517478942871, "learning_rate": 9.799400000000001e-06, "loss": 0.128, "step": 1004 }, { "epoch": 0.05357999680119422, "grad_norm": 8.34424877166748, "learning_rate": 9.799200000000001e-06, "loss": 0.0746, "step": 1005 }, { "epoch": 0.05363331023084715, "grad_norm": 9.524312973022461, "learning_rate": 9.799e-06, "loss": 0.1273, "step": 1006 }, { "epoch": 0.05368662366050008, "grad_norm": 9.096494674682617, "learning_rate": 9.7988e-06, "loss": 0.1193, "step": 1007 }, { "epoch": 0.05373993709015301, "grad_norm": 7.334498882293701, "learning_rate": 9.7986e-06, "loss": 0.1181, "step": 1008 }, { "epoch": 0.05379325051980594, "grad_norm": 5.290680408477783, "learning_rate": 9.798400000000001e-06, "loss": 0.1037, "step": 1009 }, { "epoch": 0.053846563949458866, "grad_norm": 11.417533874511719, "learning_rate": 9.7982e-06, "loss": 0.1073, "step": 1010 }, { "epoch": 0.053899877379111796, "grad_norm": 11.296675682067871, "learning_rate": 9.798e-06, "loss": 0.1291, "step": 1011 }, { "epoch": 0.05395319080876473, "grad_norm": 5.226722717285156, "learning_rate": 9.797800000000001e-06, "loss": 0.0866, "step": 1012 }, { "epoch": 0.05400650423841766, "grad_norm": 7.8430376052856445, "learning_rate": 9.797600000000001e-06, "loss": 0.1054, "step": 1013 }, { "epoch": 0.05405981766807059, "grad_norm": 5.952463150024414, "learning_rate": 9.797400000000002e-06, "loss": 0.1163, "step": 1014 }, { "epoch": 0.05411313109772352, "grad_norm": 8.73465633392334, "learning_rate": 9.797200000000002e-06, "loss": 0.0658, "step": 1015 }, { "epoch": 0.05416644452737645, "grad_norm": 6.839545249938965, "learning_rate": 9.797000000000001e-06, "loss": 0.1013, "step": 1016 }, { "epoch": 0.05421975795702937, "grad_norm": 5.614004135131836, "learning_rate": 9.7968e-06, "loss": 0.0952, "step": 1017 }, { "epoch": 0.0542730713866823, "grad_norm": 7.057100296020508, "learning_rate": 9.7966e-06, "loss": 0.1362, "step": 1018 }, { "epoch": 0.054326384816335234, "grad_norm": 10.707517623901367, "learning_rate": 9.7964e-06, "loss": 0.1622, "step": 1019 }, { "epoch": 0.054379698245988164, "grad_norm": 13.2088623046875, "learning_rate": 9.796200000000001e-06, "loss": 0.0855, "step": 1020 }, { "epoch": 0.054433011675641095, "grad_norm": 6.768912315368652, "learning_rate": 9.796e-06, "loss": 0.1346, "step": 1021 }, { "epoch": 0.054486325105294026, "grad_norm": 6.986632347106934, "learning_rate": 9.7958e-06, "loss": 0.099, "step": 1022 }, { "epoch": 0.054539638534946956, "grad_norm": 12.661738395690918, "learning_rate": 9.7956e-06, "loss": 0.1081, "step": 1023 }, { "epoch": 0.05459295196459988, "grad_norm": 7.974583625793457, "learning_rate": 9.795400000000001e-06, "loss": 0.1022, "step": 1024 }, { "epoch": 0.05464626539425281, "grad_norm": 10.229658126831055, "learning_rate": 9.7952e-06, "loss": 0.1525, "step": 1025 }, { "epoch": 0.05469957882390574, "grad_norm": 17.165699005126953, "learning_rate": 9.795000000000002e-06, "loss": 0.1725, "step": 1026 }, { "epoch": 0.05475289225355867, "grad_norm": 9.03791618347168, "learning_rate": 9.794800000000001e-06, "loss": 0.1275, "step": 1027 }, { "epoch": 0.0548062056832116, "grad_norm": 6.025590419769287, "learning_rate": 9.794600000000001e-06, "loss": 0.0884, "step": 1028 }, { "epoch": 0.05485951911286453, "grad_norm": 6.089366912841797, "learning_rate": 9.7944e-06, "loss": 0.1031, "step": 1029 }, { "epoch": 0.05491283254251746, "grad_norm": 12.155848503112793, "learning_rate": 9.7942e-06, "loss": 0.1375, "step": 1030 }, { "epoch": 0.05496614597217039, "grad_norm": 9.693799018859863, "learning_rate": 9.794000000000001e-06, "loss": 0.1487, "step": 1031 }, { "epoch": 0.05501945940182332, "grad_norm": 4.542647838592529, "learning_rate": 9.793800000000001e-06, "loss": 0.0772, "step": 1032 }, { "epoch": 0.05507277283147625, "grad_norm": 8.389686584472656, "learning_rate": 9.7936e-06, "loss": 0.1047, "step": 1033 }, { "epoch": 0.05512608626112918, "grad_norm": 7.412461280822754, "learning_rate": 9.7934e-06, "loss": 0.1142, "step": 1034 }, { "epoch": 0.05517939969078211, "grad_norm": 19.51192855834961, "learning_rate": 9.7932e-06, "loss": 0.1551, "step": 1035 }, { "epoch": 0.05523271312043504, "grad_norm": 12.374509811401367, "learning_rate": 9.793000000000001e-06, "loss": 0.1874, "step": 1036 }, { "epoch": 0.05528602655008797, "grad_norm": 8.590258598327637, "learning_rate": 9.7928e-06, "loss": 0.1274, "step": 1037 }, { "epoch": 0.055339339979740894, "grad_norm": 7.316140651702881, "learning_rate": 9.792600000000002e-06, "loss": 0.1205, "step": 1038 }, { "epoch": 0.055392653409393824, "grad_norm": 6.747189044952393, "learning_rate": 9.792400000000001e-06, "loss": 0.0899, "step": 1039 }, { "epoch": 0.055445966839046755, "grad_norm": 7.860826015472412, "learning_rate": 9.792200000000001e-06, "loss": 0.1682, "step": 1040 }, { "epoch": 0.055499280268699686, "grad_norm": 5.3710036277771, "learning_rate": 9.792e-06, "loss": 0.1215, "step": 1041 }, { "epoch": 0.055552593698352616, "grad_norm": 5.287220001220703, "learning_rate": 9.791800000000002e-06, "loss": 0.1215, "step": 1042 }, { "epoch": 0.05560590712800555, "grad_norm": 5.27832555770874, "learning_rate": 9.791600000000001e-06, "loss": 0.0829, "step": 1043 }, { "epoch": 0.05565922055765848, "grad_norm": 8.29938793182373, "learning_rate": 9.7914e-06, "loss": 0.1342, "step": 1044 }, { "epoch": 0.0557125339873114, "grad_norm": 6.933365821838379, "learning_rate": 9.7912e-06, "loss": 0.0988, "step": 1045 }, { "epoch": 0.05576584741696433, "grad_norm": 5.366513729095459, "learning_rate": 9.791e-06, "loss": 0.0959, "step": 1046 }, { "epoch": 0.05581916084661726, "grad_norm": 6.992581367492676, "learning_rate": 9.7908e-06, "loss": 0.1397, "step": 1047 }, { "epoch": 0.05587247427627019, "grad_norm": 8.606806755065918, "learning_rate": 9.7906e-06, "loss": 0.0954, "step": 1048 }, { "epoch": 0.05592578770592312, "grad_norm": 9.32618522644043, "learning_rate": 9.7904e-06, "loss": 0.1363, "step": 1049 }, { "epoch": 0.055979101135576054, "grad_norm": 5.266575336456299, "learning_rate": 9.790200000000002e-06, "loss": 0.0929, "step": 1050 }, { "epoch": 0.056032414565228984, "grad_norm": 9.593696594238281, "learning_rate": 9.790000000000001e-06, "loss": 0.1413, "step": 1051 }, { "epoch": 0.05608572799488191, "grad_norm": 12.466054916381836, "learning_rate": 9.7898e-06, "loss": 0.1005, "step": 1052 }, { "epoch": 0.05613904142453484, "grad_norm": 10.53577709197998, "learning_rate": 9.789600000000002e-06, "loss": 0.1235, "step": 1053 }, { "epoch": 0.05619235485418777, "grad_norm": 7.918834209442139, "learning_rate": 9.789400000000002e-06, "loss": 0.1395, "step": 1054 }, { "epoch": 0.0562456682838407, "grad_norm": 12.902227401733398, "learning_rate": 9.789200000000001e-06, "loss": 0.1153, "step": 1055 }, { "epoch": 0.05629898171349363, "grad_norm": 11.396700859069824, "learning_rate": 9.789e-06, "loss": 0.1159, "step": 1056 }, { "epoch": 0.05635229514314656, "grad_norm": 6.821463584899902, "learning_rate": 9.7888e-06, "loss": 0.1145, "step": 1057 }, { "epoch": 0.05640560857279949, "grad_norm": 15.836729049682617, "learning_rate": 9.7886e-06, "loss": 0.1118, "step": 1058 }, { "epoch": 0.056458922002452415, "grad_norm": 13.327064514160156, "learning_rate": 9.788400000000001e-06, "loss": 0.1257, "step": 1059 }, { "epoch": 0.056512235432105346, "grad_norm": 6.439247131347656, "learning_rate": 9.7882e-06, "loss": 0.0666, "step": 1060 }, { "epoch": 0.056565548861758276, "grad_norm": 6.047985076904297, "learning_rate": 9.788e-06, "loss": 0.1078, "step": 1061 }, { "epoch": 0.05661886229141121, "grad_norm": 5.508490562438965, "learning_rate": 9.787800000000001e-06, "loss": 0.0989, "step": 1062 }, { "epoch": 0.05667217572106414, "grad_norm": 12.623931884765625, "learning_rate": 9.787600000000001e-06, "loss": 0.1321, "step": 1063 }, { "epoch": 0.05672548915071707, "grad_norm": 12.773183822631836, "learning_rate": 9.7874e-06, "loss": 0.1457, "step": 1064 }, { "epoch": 0.05677880258037, "grad_norm": 10.96330738067627, "learning_rate": 9.787200000000002e-06, "loss": 0.1533, "step": 1065 }, { "epoch": 0.05683211601002292, "grad_norm": 14.074156761169434, "learning_rate": 9.787000000000001e-06, "loss": 0.1097, "step": 1066 }, { "epoch": 0.05688542943967585, "grad_norm": 9.608171463012695, "learning_rate": 9.786800000000001e-06, "loss": 0.1634, "step": 1067 }, { "epoch": 0.05693874286932878, "grad_norm": 8.118597030639648, "learning_rate": 9.7866e-06, "loss": 0.1237, "step": 1068 }, { "epoch": 0.056992056298981714, "grad_norm": 8.663446426391602, "learning_rate": 9.7864e-06, "loss": 0.1172, "step": 1069 }, { "epoch": 0.057045369728634644, "grad_norm": 8.024710655212402, "learning_rate": 9.786200000000001e-06, "loss": 0.0943, "step": 1070 }, { "epoch": 0.057098683158287575, "grad_norm": 8.432111740112305, "learning_rate": 9.786e-06, "loss": 0.0994, "step": 1071 }, { "epoch": 0.057151996587940505, "grad_norm": 15.210427284240723, "learning_rate": 9.7858e-06, "loss": 0.1005, "step": 1072 }, { "epoch": 0.05720531001759343, "grad_norm": 9.190542221069336, "learning_rate": 9.7856e-06, "loss": 0.0865, "step": 1073 }, { "epoch": 0.05725862344724636, "grad_norm": 10.782073974609375, "learning_rate": 9.785400000000001e-06, "loss": 0.1194, "step": 1074 }, { "epoch": 0.05731193687689929, "grad_norm": 10.662333488464355, "learning_rate": 9.7852e-06, "loss": 0.1281, "step": 1075 }, { "epoch": 0.05736525030655222, "grad_norm": 13.125216484069824, "learning_rate": 9.785e-06, "loss": 0.119, "step": 1076 }, { "epoch": 0.05741856373620515, "grad_norm": 7.937605857849121, "learning_rate": 9.784800000000002e-06, "loss": 0.1016, "step": 1077 }, { "epoch": 0.05747187716585808, "grad_norm": 10.237606048583984, "learning_rate": 9.784600000000001e-06, "loss": 0.1077, "step": 1078 }, { "epoch": 0.05752519059551101, "grad_norm": 4.26400899887085, "learning_rate": 9.7844e-06, "loss": 0.0857, "step": 1079 }, { "epoch": 0.057578504025163936, "grad_norm": 6.910892963409424, "learning_rate": 9.7842e-06, "loss": 0.0861, "step": 1080 }, { "epoch": 0.05763181745481687, "grad_norm": 8.257102966308594, "learning_rate": 9.784000000000002e-06, "loss": 0.1372, "step": 1081 }, { "epoch": 0.0576851308844698, "grad_norm": 8.799578666687012, "learning_rate": 9.783800000000001e-06, "loss": 0.1194, "step": 1082 }, { "epoch": 0.05773844431412273, "grad_norm": 5.179149627685547, "learning_rate": 9.7836e-06, "loss": 0.0709, "step": 1083 }, { "epoch": 0.05779175774377566, "grad_norm": 6.299641132354736, "learning_rate": 9.7834e-06, "loss": 0.1087, "step": 1084 }, { "epoch": 0.05784507117342859, "grad_norm": 14.879990577697754, "learning_rate": 9.7832e-06, "loss": 0.127, "step": 1085 }, { "epoch": 0.05789838460308152, "grad_norm": 14.642558097839355, "learning_rate": 9.783000000000001e-06, "loss": 0.0981, "step": 1086 }, { "epoch": 0.05795169803273444, "grad_norm": 11.005169868469238, "learning_rate": 9.7828e-06, "loss": 0.1467, "step": 1087 }, { "epoch": 0.058005011462387374, "grad_norm": 16.447284698486328, "learning_rate": 9.7826e-06, "loss": 0.1441, "step": 1088 }, { "epoch": 0.058058324892040304, "grad_norm": 15.096861839294434, "learning_rate": 9.782400000000001e-06, "loss": 0.1559, "step": 1089 }, { "epoch": 0.058111638321693235, "grad_norm": 5.329514980316162, "learning_rate": 9.782200000000001e-06, "loss": 0.1212, "step": 1090 }, { "epoch": 0.058164951751346165, "grad_norm": 15.124671936035156, "learning_rate": 9.782e-06, "loss": 0.1509, "step": 1091 }, { "epoch": 0.058218265180999096, "grad_norm": 13.349143981933594, "learning_rate": 9.781800000000002e-06, "loss": 0.0902, "step": 1092 }, { "epoch": 0.058271578610652026, "grad_norm": 16.01079750061035, "learning_rate": 9.781600000000001e-06, "loss": 0.1277, "step": 1093 }, { "epoch": 0.05832489204030495, "grad_norm": 9.267243385314941, "learning_rate": 9.781400000000001e-06, "loss": 0.1392, "step": 1094 }, { "epoch": 0.05837820546995788, "grad_norm": 11.409904479980469, "learning_rate": 9.7812e-06, "loss": 0.1137, "step": 1095 }, { "epoch": 0.05843151889961081, "grad_norm": 9.357407569885254, "learning_rate": 9.781e-06, "loss": 0.0906, "step": 1096 }, { "epoch": 0.05848483232926374, "grad_norm": 9.125494956970215, "learning_rate": 9.7808e-06, "loss": 0.1102, "step": 1097 }, { "epoch": 0.05853814575891667, "grad_norm": 7.620985984802246, "learning_rate": 9.780600000000001e-06, "loss": 0.1606, "step": 1098 }, { "epoch": 0.0585914591885696, "grad_norm": 5.774655342102051, "learning_rate": 9.7804e-06, "loss": 0.1175, "step": 1099 }, { "epoch": 0.05864477261822253, "grad_norm": 18.298341751098633, "learning_rate": 9.7802e-06, "loss": 0.162, "step": 1100 }, { "epoch": 0.05869808604787546, "grad_norm": 18.03533935546875, "learning_rate": 9.780000000000001e-06, "loss": 0.1753, "step": 1101 }, { "epoch": 0.05875139947752839, "grad_norm": 9.094078063964844, "learning_rate": 9.7798e-06, "loss": 0.1062, "step": 1102 }, { "epoch": 0.05880471290718132, "grad_norm": 7.453545093536377, "learning_rate": 9.779600000000002e-06, "loss": 0.1065, "step": 1103 }, { "epoch": 0.05885802633683425, "grad_norm": 15.489660263061523, "learning_rate": 9.779400000000002e-06, "loss": 0.1674, "step": 1104 }, { "epoch": 0.05891133976648718, "grad_norm": 13.40820026397705, "learning_rate": 9.779200000000001e-06, "loss": 0.1199, "step": 1105 }, { "epoch": 0.05896465319614011, "grad_norm": 9.695240020751953, "learning_rate": 9.779e-06, "loss": 0.1244, "step": 1106 }, { "epoch": 0.05901796662579304, "grad_norm": 6.875389575958252, "learning_rate": 9.7788e-06, "loss": 0.1032, "step": 1107 }, { "epoch": 0.059071280055445964, "grad_norm": 13.111930847167969, "learning_rate": 9.7786e-06, "loss": 0.1376, "step": 1108 }, { "epoch": 0.059124593485098895, "grad_norm": 14.416450500488281, "learning_rate": 9.778400000000001e-06, "loss": 0.0976, "step": 1109 }, { "epoch": 0.059177906914751825, "grad_norm": 14.595574378967285, "learning_rate": 9.7782e-06, "loss": 0.1428, "step": 1110 }, { "epoch": 0.059231220344404756, "grad_norm": 13.81889820098877, "learning_rate": 9.778e-06, "loss": 0.1684, "step": 1111 }, { "epoch": 0.059284533774057686, "grad_norm": 7.368175983428955, "learning_rate": 9.7778e-06, "loss": 0.0938, "step": 1112 }, { "epoch": 0.05933784720371062, "grad_norm": 13.340516090393066, "learning_rate": 9.777600000000001e-06, "loss": 0.155, "step": 1113 }, { "epoch": 0.05939116063336355, "grad_norm": 8.658504486083984, "learning_rate": 9.7774e-06, "loss": 0.1348, "step": 1114 }, { "epoch": 0.05944447406301647, "grad_norm": 5.317590236663818, "learning_rate": 9.777200000000002e-06, "loss": 0.1018, "step": 1115 }, { "epoch": 0.0594977874926694, "grad_norm": 11.7642183303833, "learning_rate": 9.777000000000001e-06, "loss": 0.1267, "step": 1116 }, { "epoch": 0.05955110092232233, "grad_norm": 9.103870391845703, "learning_rate": 9.776800000000001e-06, "loss": 0.1302, "step": 1117 }, { "epoch": 0.05960441435197526, "grad_norm": 9.369601249694824, "learning_rate": 9.7766e-06, "loss": 0.1335, "step": 1118 }, { "epoch": 0.05965772778162819, "grad_norm": 4.719306468963623, "learning_rate": 9.7764e-06, "loss": 0.0858, "step": 1119 }, { "epoch": 0.059711041211281124, "grad_norm": 4.376290321350098, "learning_rate": 9.776200000000001e-06, "loss": 0.0955, "step": 1120 }, { "epoch": 0.059764354640934055, "grad_norm": 9.573092460632324, "learning_rate": 9.776000000000001e-06, "loss": 0.1349, "step": 1121 }, { "epoch": 0.05981766807058698, "grad_norm": 9.886490821838379, "learning_rate": 9.7758e-06, "loss": 0.0981, "step": 1122 }, { "epoch": 0.05987098150023991, "grad_norm": 12.640137672424316, "learning_rate": 9.7756e-06, "loss": 0.1559, "step": 1123 }, { "epoch": 0.05992429492989284, "grad_norm": 9.768352508544922, "learning_rate": 9.7754e-06, "loss": 0.102, "step": 1124 }, { "epoch": 0.05997760835954577, "grad_norm": 6.306491374969482, "learning_rate": 9.775200000000001e-06, "loss": 0.1154, "step": 1125 }, { "epoch": 0.0600309217891987, "grad_norm": 11.509134292602539, "learning_rate": 9.775e-06, "loss": 0.1073, "step": 1126 }, { "epoch": 0.06008423521885163, "grad_norm": 5.398394584655762, "learning_rate": 9.774800000000002e-06, "loss": 0.0886, "step": 1127 }, { "epoch": 0.06013754864850456, "grad_norm": 6.478825569152832, "learning_rate": 9.774600000000001e-06, "loss": 0.1054, "step": 1128 }, { "epoch": 0.060190862078157485, "grad_norm": 6.729395389556885, "learning_rate": 9.7744e-06, "loss": 0.1235, "step": 1129 }, { "epoch": 0.060244175507810416, "grad_norm": 9.116671562194824, "learning_rate": 9.7742e-06, "loss": 0.0913, "step": 1130 }, { "epoch": 0.060297488937463346, "grad_norm": 10.148269653320312, "learning_rate": 9.774000000000002e-06, "loss": 0.1085, "step": 1131 }, { "epoch": 0.06035080236711628, "grad_norm": 6.956345081329346, "learning_rate": 9.773800000000001e-06, "loss": 0.1324, "step": 1132 }, { "epoch": 0.06040411579676921, "grad_norm": 9.077391624450684, "learning_rate": 9.7736e-06, "loss": 0.0822, "step": 1133 }, { "epoch": 0.06045742922642214, "grad_norm": 9.857474327087402, "learning_rate": 9.7734e-06, "loss": 0.1174, "step": 1134 }, { "epoch": 0.06051074265607507, "grad_norm": 7.5036516189575195, "learning_rate": 9.7732e-06, "loss": 0.1313, "step": 1135 }, { "epoch": 0.06056405608572799, "grad_norm": 7.7395429611206055, "learning_rate": 9.773e-06, "loss": 0.15, "step": 1136 }, { "epoch": 0.06061736951538092, "grad_norm": 4.763211727142334, "learning_rate": 9.7728e-06, "loss": 0.0686, "step": 1137 }, { "epoch": 0.06067068294503385, "grad_norm": 8.134703636169434, "learning_rate": 9.7726e-06, "loss": 0.118, "step": 1138 }, { "epoch": 0.060723996374686784, "grad_norm": 6.766697883605957, "learning_rate": 9.772400000000002e-06, "loss": 0.0994, "step": 1139 }, { "epoch": 0.060777309804339714, "grad_norm": 4.634028434753418, "learning_rate": 9.772200000000001e-06, "loss": 0.0788, "step": 1140 }, { "epoch": 0.060830623233992645, "grad_norm": 7.6737236976623535, "learning_rate": 9.772e-06, "loss": 0.1046, "step": 1141 }, { "epoch": 0.060883936663645576, "grad_norm": 5.324032306671143, "learning_rate": 9.771800000000002e-06, "loss": 0.0798, "step": 1142 }, { "epoch": 0.0609372500932985, "grad_norm": 6.090784072875977, "learning_rate": 9.771600000000001e-06, "loss": 0.1303, "step": 1143 }, { "epoch": 0.06099056352295143, "grad_norm": 6.249975681304932, "learning_rate": 9.771400000000001e-06, "loss": 0.1155, "step": 1144 }, { "epoch": 0.06104387695260436, "grad_norm": 9.070443153381348, "learning_rate": 9.7712e-06, "loss": 0.093, "step": 1145 }, { "epoch": 0.06109719038225729, "grad_norm": 4.159156322479248, "learning_rate": 9.771e-06, "loss": 0.074, "step": 1146 }, { "epoch": 0.06115050381191022, "grad_norm": 10.892524719238281, "learning_rate": 9.7708e-06, "loss": 0.1709, "step": 1147 }, { "epoch": 0.06120381724156315, "grad_norm": 10.840734481811523, "learning_rate": 9.770600000000001e-06, "loss": 0.1036, "step": 1148 }, { "epoch": 0.06125713067121608, "grad_norm": 6.976202011108398, "learning_rate": 9.7704e-06, "loss": 0.1372, "step": 1149 }, { "epoch": 0.061310444100869006, "grad_norm": 6.760790824890137, "learning_rate": 9.7702e-06, "loss": 0.1397, "step": 1150 }, { "epoch": 0.06136375753052194, "grad_norm": 6.169122219085693, "learning_rate": 9.770000000000001e-06, "loss": 0.1112, "step": 1151 }, { "epoch": 0.06141707096017487, "grad_norm": 8.099852561950684, "learning_rate": 9.769800000000001e-06, "loss": 0.1644, "step": 1152 }, { "epoch": 0.0614703843898278, "grad_norm": 8.042125701904297, "learning_rate": 9.769600000000002e-06, "loss": 0.1057, "step": 1153 }, { "epoch": 0.06152369781948073, "grad_norm": 6.868786811828613, "learning_rate": 9.769400000000002e-06, "loss": 0.0742, "step": 1154 }, { "epoch": 0.06157701124913366, "grad_norm": 8.292021751403809, "learning_rate": 9.769200000000001e-06, "loss": 0.1295, "step": 1155 }, { "epoch": 0.06163032467878659, "grad_norm": 8.726435661315918, "learning_rate": 9.769e-06, "loss": 0.1077, "step": 1156 }, { "epoch": 0.06168363810843951, "grad_norm": 5.629456043243408, "learning_rate": 9.7688e-06, "loss": 0.0726, "step": 1157 }, { "epoch": 0.061736951538092444, "grad_norm": 5.828682899475098, "learning_rate": 9.7686e-06, "loss": 0.0849, "step": 1158 }, { "epoch": 0.061790264967745374, "grad_norm": 4.582000732421875, "learning_rate": 9.768400000000001e-06, "loss": 0.0744, "step": 1159 }, { "epoch": 0.061843578397398305, "grad_norm": 5.856855392456055, "learning_rate": 9.7682e-06, "loss": 0.0682, "step": 1160 }, { "epoch": 0.061896891827051236, "grad_norm": 4.801296710968018, "learning_rate": 9.768e-06, "loss": 0.1047, "step": 1161 }, { "epoch": 0.061950205256704166, "grad_norm": 12.809771537780762, "learning_rate": 9.7678e-06, "loss": 0.1464, "step": 1162 }, { "epoch": 0.0620035186863571, "grad_norm": 6.84320592880249, "learning_rate": 9.767600000000001e-06, "loss": 0.0988, "step": 1163 }, { "epoch": 0.06205683211601002, "grad_norm": 6.355743408203125, "learning_rate": 9.7674e-06, "loss": 0.1067, "step": 1164 }, { "epoch": 0.06211014554566295, "grad_norm": 4.228799343109131, "learning_rate": 9.767200000000002e-06, "loss": 0.0606, "step": 1165 }, { "epoch": 0.06216345897531588, "grad_norm": 8.9925537109375, "learning_rate": 9.767000000000002e-06, "loss": 0.0707, "step": 1166 }, { "epoch": 0.06221677240496881, "grad_norm": 13.632147789001465, "learning_rate": 9.766800000000001e-06, "loss": 0.097, "step": 1167 }, { "epoch": 0.06227008583462174, "grad_norm": 7.306240558624268, "learning_rate": 9.7666e-06, "loss": 0.0805, "step": 1168 }, { "epoch": 0.06232339926427467, "grad_norm": 6.420141220092773, "learning_rate": 9.7664e-06, "loss": 0.0957, "step": 1169 }, { "epoch": 0.062376712693927604, "grad_norm": 10.252357482910156, "learning_rate": 9.766200000000001e-06, "loss": 0.1238, "step": 1170 }, { "epoch": 0.06243002612358053, "grad_norm": 8.942436218261719, "learning_rate": 9.766000000000001e-06, "loss": 0.103, "step": 1171 }, { "epoch": 0.06248333955323346, "grad_norm": 4.829742908477783, "learning_rate": 9.7658e-06, "loss": 0.0764, "step": 1172 }, { "epoch": 0.0625366529828864, "grad_norm": 5.811575412750244, "learning_rate": 9.7656e-06, "loss": 0.0875, "step": 1173 }, { "epoch": 0.06258996641253932, "grad_norm": 7.581599712371826, "learning_rate": 9.7654e-06, "loss": 0.0958, "step": 1174 }, { "epoch": 0.06264327984219224, "grad_norm": 6.851376056671143, "learning_rate": 9.765200000000001e-06, "loss": 0.0994, "step": 1175 }, { "epoch": 0.06269659327184518, "grad_norm": 11.871820449829102, "learning_rate": 9.765e-06, "loss": 0.1099, "step": 1176 }, { "epoch": 0.0627499067014981, "grad_norm": 6.935635566711426, "learning_rate": 9.764800000000002e-06, "loss": 0.0949, "step": 1177 }, { "epoch": 0.06280322013115104, "grad_norm": 7.536258220672607, "learning_rate": 9.764600000000001e-06, "loss": 0.0997, "step": 1178 }, { "epoch": 0.06285653356080396, "grad_norm": 10.788915634155273, "learning_rate": 9.764400000000001e-06, "loss": 0.1205, "step": 1179 }, { "epoch": 0.0629098469904569, "grad_norm": 7.847055912017822, "learning_rate": 9.7642e-06, "loss": 0.1212, "step": 1180 }, { "epoch": 0.06296316042010983, "grad_norm": 8.24731159210205, "learning_rate": 9.764000000000002e-06, "loss": 0.0798, "step": 1181 }, { "epoch": 0.06301647384976275, "grad_norm": 8.675071716308594, "learning_rate": 9.763800000000001e-06, "loss": 0.0778, "step": 1182 }, { "epoch": 0.06306978727941569, "grad_norm": 5.611132621765137, "learning_rate": 9.7636e-06, "loss": 0.092, "step": 1183 }, { "epoch": 0.06312310070906861, "grad_norm": 10.437002182006836, "learning_rate": 9.7634e-06, "loss": 0.0875, "step": 1184 }, { "epoch": 0.06317641413872155, "grad_norm": 5.111097812652588, "learning_rate": 9.7632e-06, "loss": 0.0749, "step": 1185 }, { "epoch": 0.06322972756837447, "grad_norm": 7.129400253295898, "learning_rate": 9.763e-06, "loss": 0.0741, "step": 1186 }, { "epoch": 0.06328304099802741, "grad_norm": 5.957415580749512, "learning_rate": 9.7628e-06, "loss": 0.0987, "step": 1187 }, { "epoch": 0.06333635442768033, "grad_norm": 9.959009170532227, "learning_rate": 9.7626e-06, "loss": 0.1728, "step": 1188 }, { "epoch": 0.06338966785733326, "grad_norm": 4.651831150054932, "learning_rate": 9.762400000000002e-06, "loss": 0.0667, "step": 1189 }, { "epoch": 0.0634429812869862, "grad_norm": 6.65704345703125, "learning_rate": 9.762200000000001e-06, "loss": 0.1083, "step": 1190 }, { "epoch": 0.06349629471663912, "grad_norm": 7.8797430992126465, "learning_rate": 9.762e-06, "loss": 0.0756, "step": 1191 }, { "epoch": 0.06354960814629206, "grad_norm": 5.64208984375, "learning_rate": 9.761800000000002e-06, "loss": 0.0843, "step": 1192 }, { "epoch": 0.06360292157594498, "grad_norm": 13.97037124633789, "learning_rate": 9.761600000000002e-06, "loss": 0.0903, "step": 1193 }, { "epoch": 0.06365623500559792, "grad_norm": 12.643730163574219, "learning_rate": 9.761400000000001e-06, "loss": 0.1448, "step": 1194 }, { "epoch": 0.06370954843525084, "grad_norm": 5.9476213455200195, "learning_rate": 9.7612e-06, "loss": 0.0805, "step": 1195 }, { "epoch": 0.06376286186490376, "grad_norm": 5.775064945220947, "learning_rate": 9.761e-06, "loss": 0.0774, "step": 1196 }, { "epoch": 0.0638161752945567, "grad_norm": 10.783669471740723, "learning_rate": 9.7608e-06, "loss": 0.1333, "step": 1197 }, { "epoch": 0.06386948872420962, "grad_norm": 5.309019565582275, "learning_rate": 9.760600000000001e-06, "loss": 0.0907, "step": 1198 }, { "epoch": 0.06392280215386256, "grad_norm": 12.059395790100098, "learning_rate": 9.7604e-06, "loss": 0.154, "step": 1199 }, { "epoch": 0.06397611558351549, "grad_norm": 13.7621488571167, "learning_rate": 9.7602e-06, "loss": 0.1235, "step": 1200 }, { "epoch": 0.06402942901316842, "grad_norm": 10.800841331481934, "learning_rate": 9.760000000000001e-06, "loss": 0.1083, "step": 1201 }, { "epoch": 0.06408274244282135, "grad_norm": 5.174660682678223, "learning_rate": 9.759800000000001e-06, "loss": 0.0753, "step": 1202 }, { "epoch": 0.06413605587247427, "grad_norm": 5.015161991119385, "learning_rate": 9.7596e-06, "loss": 0.0711, "step": 1203 }, { "epoch": 0.06418936930212721, "grad_norm": 6.147846221923828, "learning_rate": 9.759400000000002e-06, "loss": 0.0993, "step": 1204 }, { "epoch": 0.06424268273178013, "grad_norm": 11.210344314575195, "learning_rate": 9.759200000000001e-06, "loss": 0.1228, "step": 1205 }, { "epoch": 0.06429599616143307, "grad_norm": 16.43347930908203, "learning_rate": 9.759000000000001e-06, "loss": 0.1097, "step": 1206 }, { "epoch": 0.06434930959108599, "grad_norm": 8.100709915161133, "learning_rate": 9.7588e-06, "loss": 0.1147, "step": 1207 }, { "epoch": 0.06440262302073893, "grad_norm": 10.718738555908203, "learning_rate": 9.7586e-06, "loss": 0.1277, "step": 1208 }, { "epoch": 0.06445593645039185, "grad_norm": 8.002325057983398, "learning_rate": 9.758400000000001e-06, "loss": 0.0936, "step": 1209 }, { "epoch": 0.06450924988004478, "grad_norm": 18.905654907226562, "learning_rate": 9.758200000000001e-06, "loss": 0.1516, "step": 1210 }, { "epoch": 0.06456256330969772, "grad_norm": 13.459768295288086, "learning_rate": 9.758e-06, "loss": 0.1226, "step": 1211 }, { "epoch": 0.06461587673935064, "grad_norm": 9.208149909973145, "learning_rate": 9.7578e-06, "loss": 0.086, "step": 1212 }, { "epoch": 0.06466919016900358, "grad_norm": 11.466038703918457, "learning_rate": 9.757600000000001e-06, "loss": 0.0987, "step": 1213 }, { "epoch": 0.0647225035986565, "grad_norm": 5.982420444488525, "learning_rate": 9.7574e-06, "loss": 0.0804, "step": 1214 }, { "epoch": 0.06477581702830944, "grad_norm": 11.346817970275879, "learning_rate": 9.7572e-06, "loss": 0.11, "step": 1215 }, { "epoch": 0.06482913045796236, "grad_norm": 8.779989242553711, "learning_rate": 9.757000000000002e-06, "loss": 0.0894, "step": 1216 }, { "epoch": 0.06488244388761528, "grad_norm": 5.398651123046875, "learning_rate": 9.756800000000001e-06, "loss": 0.0942, "step": 1217 }, { "epoch": 0.06493575731726822, "grad_norm": 12.336498260498047, "learning_rate": 9.7566e-06, "loss": 0.1296, "step": 1218 }, { "epoch": 0.06498907074692115, "grad_norm": 7.239138603210449, "learning_rate": 9.7564e-06, "loss": 0.0743, "step": 1219 }, { "epoch": 0.06504238417657408, "grad_norm": 9.931772232055664, "learning_rate": 9.756200000000002e-06, "loss": 0.1066, "step": 1220 }, { "epoch": 0.06509569760622701, "grad_norm": 8.60368537902832, "learning_rate": 9.756000000000001e-06, "loss": 0.1379, "step": 1221 }, { "epoch": 0.06514901103587994, "grad_norm": 6.311507225036621, "learning_rate": 9.7558e-06, "loss": 0.1004, "step": 1222 }, { "epoch": 0.06520232446553287, "grad_norm": 5.63808536529541, "learning_rate": 9.7556e-06, "loss": 0.1126, "step": 1223 }, { "epoch": 0.06525563789518579, "grad_norm": 8.504964828491211, "learning_rate": 9.7554e-06, "loss": 0.1421, "step": 1224 }, { "epoch": 0.06530895132483873, "grad_norm": 7.556707859039307, "learning_rate": 9.755200000000001e-06, "loss": 0.1137, "step": 1225 }, { "epoch": 0.06536226475449165, "grad_norm": 7.730994701385498, "learning_rate": 9.755e-06, "loss": 0.0906, "step": 1226 }, { "epoch": 0.06541557818414459, "grad_norm": 6.14415979385376, "learning_rate": 9.7548e-06, "loss": 0.1082, "step": 1227 }, { "epoch": 0.06546889161379751, "grad_norm": 3.9023799896240234, "learning_rate": 9.754600000000001e-06, "loss": 0.0596, "step": 1228 }, { "epoch": 0.06552220504345045, "grad_norm": 11.564648628234863, "learning_rate": 9.754400000000001e-06, "loss": 0.1348, "step": 1229 }, { "epoch": 0.06557551847310338, "grad_norm": 3.61331844329834, "learning_rate": 9.7542e-06, "loss": 0.0753, "step": 1230 }, { "epoch": 0.0656288319027563, "grad_norm": 8.16926097869873, "learning_rate": 9.754000000000002e-06, "loss": 0.0925, "step": 1231 }, { "epoch": 0.06568214533240924, "grad_norm": 9.646261215209961, "learning_rate": 9.753800000000001e-06, "loss": 0.1112, "step": 1232 }, { "epoch": 0.06573545876206216, "grad_norm": 5.589472770690918, "learning_rate": 9.753600000000001e-06, "loss": 0.0926, "step": 1233 }, { "epoch": 0.0657887721917151, "grad_norm": 8.25328540802002, "learning_rate": 9.7534e-06, "loss": 0.1176, "step": 1234 }, { "epoch": 0.06584208562136802, "grad_norm": 10.388134956359863, "learning_rate": 9.7532e-06, "loss": 0.1055, "step": 1235 }, { "epoch": 0.06589539905102096, "grad_norm": 7.327758312225342, "learning_rate": 9.753e-06, "loss": 0.1231, "step": 1236 }, { "epoch": 0.06594871248067388, "grad_norm": 7.079362392425537, "learning_rate": 9.752800000000001e-06, "loss": 0.1051, "step": 1237 }, { "epoch": 0.0660020259103268, "grad_norm": 6.850011348724365, "learning_rate": 9.7526e-06, "loss": 0.1338, "step": 1238 }, { "epoch": 0.06605533933997974, "grad_norm": 5.360888957977295, "learning_rate": 9.7524e-06, "loss": 0.0939, "step": 1239 }, { "epoch": 0.06610865276963267, "grad_norm": 9.447250366210938, "learning_rate": 9.752200000000001e-06, "loss": 0.181, "step": 1240 }, { "epoch": 0.0661619661992856, "grad_norm": 4.327243328094482, "learning_rate": 9.752e-06, "loss": 0.077, "step": 1241 }, { "epoch": 0.06621527962893853, "grad_norm": 4.345074653625488, "learning_rate": 9.751800000000002e-06, "loss": 0.0696, "step": 1242 }, { "epoch": 0.06626859305859147, "grad_norm": 4.930066108703613, "learning_rate": 9.751600000000002e-06, "loss": 0.091, "step": 1243 }, { "epoch": 0.06632190648824439, "grad_norm": 8.714357376098633, "learning_rate": 9.751400000000001e-06, "loss": 0.1113, "step": 1244 }, { "epoch": 0.06637521991789731, "grad_norm": 10.488165855407715, "learning_rate": 9.7512e-06, "loss": 0.1299, "step": 1245 }, { "epoch": 0.06642853334755025, "grad_norm": 8.804488182067871, "learning_rate": 9.751e-06, "loss": 0.1275, "step": 1246 }, { "epoch": 0.06648184677720317, "grad_norm": 7.566505432128906, "learning_rate": 9.7508e-06, "loss": 0.1253, "step": 1247 }, { "epoch": 0.06653516020685611, "grad_norm": 7.855594635009766, "learning_rate": 9.750600000000001e-06, "loss": 0.1304, "step": 1248 }, { "epoch": 0.06658847363650904, "grad_norm": 5.784225940704346, "learning_rate": 9.7504e-06, "loss": 0.1197, "step": 1249 }, { "epoch": 0.06664178706616197, "grad_norm": 6.98615026473999, "learning_rate": 9.7502e-06, "loss": 0.0747, "step": 1250 }, { "epoch": 0.0666951004958149, "grad_norm": 10.278362274169922, "learning_rate": 9.75e-06, "loss": 0.1201, "step": 1251 }, { "epoch": 0.06674841392546782, "grad_norm": 6.885603904724121, "learning_rate": 9.749800000000001e-06, "loss": 0.0904, "step": 1252 }, { "epoch": 0.06680172735512076, "grad_norm": 5.31528902053833, "learning_rate": 9.7496e-06, "loss": 0.0775, "step": 1253 }, { "epoch": 0.06685504078477368, "grad_norm": 9.280621528625488, "learning_rate": 9.749400000000002e-06, "loss": 0.0907, "step": 1254 }, { "epoch": 0.06690835421442662, "grad_norm": 8.39372444152832, "learning_rate": 9.749200000000001e-06, "loss": 0.1046, "step": 1255 }, { "epoch": 0.06696166764407954, "grad_norm": 11.307162284851074, "learning_rate": 9.749000000000001e-06, "loss": 0.084, "step": 1256 }, { "epoch": 0.06701498107373248, "grad_norm": 6.219311237335205, "learning_rate": 9.7488e-06, "loss": 0.0624, "step": 1257 }, { "epoch": 0.0670682945033854, "grad_norm": 12.532402992248535, "learning_rate": 9.7486e-06, "loss": 0.1219, "step": 1258 }, { "epoch": 0.06712160793303833, "grad_norm": 9.733393669128418, "learning_rate": 9.748400000000001e-06, "loss": 0.0822, "step": 1259 }, { "epoch": 0.06717492136269126, "grad_norm": 7.835059642791748, "learning_rate": 9.748200000000001e-06, "loss": 0.0739, "step": 1260 }, { "epoch": 0.06722823479234419, "grad_norm": 6.461076736450195, "learning_rate": 9.748e-06, "loss": 0.0695, "step": 1261 }, { "epoch": 0.06728154822199713, "grad_norm": 10.1917142868042, "learning_rate": 9.7478e-06, "loss": 0.094, "step": 1262 }, { "epoch": 0.06733486165165005, "grad_norm": 10.934144020080566, "learning_rate": 9.7476e-06, "loss": 0.0894, "step": 1263 }, { "epoch": 0.06738817508130299, "grad_norm": 15.071439743041992, "learning_rate": 9.747400000000001e-06, "loss": 0.1053, "step": 1264 }, { "epoch": 0.06744148851095591, "grad_norm": 5.478629112243652, "learning_rate": 9.7472e-06, "loss": 0.0668, "step": 1265 }, { "epoch": 0.06749480194060883, "grad_norm": 13.762388229370117, "learning_rate": 9.747000000000002e-06, "loss": 0.1361, "step": 1266 }, { "epoch": 0.06754811537026177, "grad_norm": 9.680427551269531, "learning_rate": 9.746800000000001e-06, "loss": 0.1142, "step": 1267 }, { "epoch": 0.0676014287999147, "grad_norm": 8.159688949584961, "learning_rate": 9.7466e-06, "loss": 0.0779, "step": 1268 }, { "epoch": 0.06765474222956763, "grad_norm": 5.697932243347168, "learning_rate": 9.7464e-06, "loss": 0.0716, "step": 1269 }, { "epoch": 0.06770805565922056, "grad_norm": 7.651941299438477, "learning_rate": 9.746200000000002e-06, "loss": 0.1417, "step": 1270 }, { "epoch": 0.0677613690888735, "grad_norm": 5.067052841186523, "learning_rate": 9.746000000000001e-06, "loss": 0.0781, "step": 1271 }, { "epoch": 0.06781468251852642, "grad_norm": 5.294726371765137, "learning_rate": 9.7458e-06, "loss": 0.1107, "step": 1272 }, { "epoch": 0.06786799594817934, "grad_norm": 5.599640846252441, "learning_rate": 9.7456e-06, "loss": 0.063, "step": 1273 }, { "epoch": 0.06792130937783228, "grad_norm": 8.487263679504395, "learning_rate": 9.7454e-06, "loss": 0.1488, "step": 1274 }, { "epoch": 0.0679746228074852, "grad_norm": 4.703219890594482, "learning_rate": 9.745200000000001e-06, "loss": 0.0686, "step": 1275 }, { "epoch": 0.06802793623713814, "grad_norm": 4.361145973205566, "learning_rate": 9.745e-06, "loss": 0.0902, "step": 1276 }, { "epoch": 0.06808124966679106, "grad_norm": 3.5699105262756348, "learning_rate": 9.7448e-06, "loss": 0.0553, "step": 1277 }, { "epoch": 0.068134563096444, "grad_norm": 8.15126895904541, "learning_rate": 9.744600000000002e-06, "loss": 0.1023, "step": 1278 }, { "epoch": 0.06818787652609692, "grad_norm": 6.904513835906982, "learning_rate": 9.744400000000001e-06, "loss": 0.0668, "step": 1279 }, { "epoch": 0.06824118995574985, "grad_norm": 10.809986114501953, "learning_rate": 9.7442e-06, "loss": 0.113, "step": 1280 }, { "epoch": 0.06829450338540279, "grad_norm": 6.726937294006348, "learning_rate": 9.744000000000002e-06, "loss": 0.065, "step": 1281 }, { "epoch": 0.06834781681505571, "grad_norm": 11.331320762634277, "learning_rate": 9.743800000000001e-06, "loss": 0.0773, "step": 1282 }, { "epoch": 0.06840113024470865, "grad_norm": 8.923863410949707, "learning_rate": 9.743600000000001e-06, "loss": 0.0865, "step": 1283 }, { "epoch": 0.06845444367436157, "grad_norm": 9.627092361450195, "learning_rate": 9.7434e-06, "loss": 0.1236, "step": 1284 }, { "epoch": 0.06850775710401451, "grad_norm": 6.741084575653076, "learning_rate": 9.7432e-06, "loss": 0.0844, "step": 1285 }, { "epoch": 0.06856107053366743, "grad_norm": 9.056588172912598, "learning_rate": 9.743000000000001e-06, "loss": 0.0636, "step": 1286 }, { "epoch": 0.06861438396332036, "grad_norm": 12.607742309570312, "learning_rate": 9.742800000000001e-06, "loss": 0.1327, "step": 1287 }, { "epoch": 0.06866769739297329, "grad_norm": 4.534584045410156, "learning_rate": 9.7426e-06, "loss": 0.0887, "step": 1288 }, { "epoch": 0.06872101082262622, "grad_norm": 6.17387580871582, "learning_rate": 9.7424e-06, "loss": 0.0937, "step": 1289 }, { "epoch": 0.06877432425227915, "grad_norm": 6.340063095092773, "learning_rate": 9.742200000000001e-06, "loss": 0.1111, "step": 1290 }, { "epoch": 0.06882763768193208, "grad_norm": 7.863002777099609, "learning_rate": 9.742000000000001e-06, "loss": 0.1164, "step": 1291 }, { "epoch": 0.06888095111158501, "grad_norm": 12.41604232788086, "learning_rate": 9.7418e-06, "loss": 0.1208, "step": 1292 }, { "epoch": 0.06893426454123794, "grad_norm": 4.895146369934082, "learning_rate": 9.741600000000002e-06, "loss": 0.0936, "step": 1293 }, { "epoch": 0.06898757797089086, "grad_norm": 13.354392051696777, "learning_rate": 9.741400000000001e-06, "loss": 0.1205, "step": 1294 }, { "epoch": 0.0690408914005438, "grad_norm": 4.781411647796631, "learning_rate": 9.741200000000001e-06, "loss": 0.1007, "step": 1295 }, { "epoch": 0.06909420483019672, "grad_norm": 7.584843158721924, "learning_rate": 9.741e-06, "loss": 0.0918, "step": 1296 }, { "epoch": 0.06914751825984966, "grad_norm": 6.6266374588012695, "learning_rate": 9.7408e-06, "loss": 0.0966, "step": 1297 }, { "epoch": 0.06920083168950258, "grad_norm": 4.0819830894470215, "learning_rate": 9.740600000000001e-06, "loss": 0.0437, "step": 1298 }, { "epoch": 0.06925414511915552, "grad_norm": 10.970120429992676, "learning_rate": 9.7404e-06, "loss": 0.1041, "step": 1299 }, { "epoch": 0.06930745854880845, "grad_norm": 5.771185398101807, "learning_rate": 9.7402e-06, "loss": 0.0629, "step": 1300 }, { "epoch": 0.06936077197846137, "grad_norm": 8.906908988952637, "learning_rate": 9.74e-06, "loss": 0.0899, "step": 1301 }, { "epoch": 0.0694140854081143, "grad_norm": 5.58319616317749, "learning_rate": 9.739800000000001e-06, "loss": 0.069, "step": 1302 }, { "epoch": 0.06946739883776723, "grad_norm": 6.133340358734131, "learning_rate": 9.7396e-06, "loss": 0.1103, "step": 1303 }, { "epoch": 0.06952071226742017, "grad_norm": 6.789834022521973, "learning_rate": 9.7394e-06, "loss": 0.1063, "step": 1304 }, { "epoch": 0.06957402569707309, "grad_norm": 9.455716133117676, "learning_rate": 9.739200000000002e-06, "loss": 0.1235, "step": 1305 }, { "epoch": 0.06962733912672603, "grad_norm": 8.133991241455078, "learning_rate": 9.739000000000001e-06, "loss": 0.0884, "step": 1306 }, { "epoch": 0.06968065255637895, "grad_norm": 6.872854709625244, "learning_rate": 9.7388e-06, "loss": 0.1289, "step": 1307 }, { "epoch": 0.06973396598603188, "grad_norm": 6.147084712982178, "learning_rate": 9.7386e-06, "loss": 0.0915, "step": 1308 }, { "epoch": 0.06978727941568481, "grad_norm": 6.03138542175293, "learning_rate": 9.738400000000001e-06, "loss": 0.0899, "step": 1309 }, { "epoch": 0.06984059284533774, "grad_norm": 5.741815090179443, "learning_rate": 9.738200000000001e-06, "loss": 0.0469, "step": 1310 }, { "epoch": 0.06989390627499067, "grad_norm": 7.075740337371826, "learning_rate": 9.738e-06, "loss": 0.1257, "step": 1311 }, { "epoch": 0.0699472197046436, "grad_norm": 7.569990634918213, "learning_rate": 9.7378e-06, "loss": 0.0893, "step": 1312 }, { "epoch": 0.07000053313429654, "grad_norm": 13.395947456359863, "learning_rate": 9.7376e-06, "loss": 0.1078, "step": 1313 }, { "epoch": 0.07005384656394946, "grad_norm": 13.680778503417969, "learning_rate": 9.737400000000001e-06, "loss": 0.1309, "step": 1314 }, { "epoch": 0.07010715999360238, "grad_norm": 5.936118125915527, "learning_rate": 9.7372e-06, "loss": 0.0878, "step": 1315 }, { "epoch": 0.07016047342325532, "grad_norm": 7.371811389923096, "learning_rate": 9.737e-06, "loss": 0.1064, "step": 1316 }, { "epoch": 0.07021378685290824, "grad_norm": 8.244985580444336, "learning_rate": 9.736800000000001e-06, "loss": 0.1113, "step": 1317 }, { "epoch": 0.07026710028256118, "grad_norm": 14.515878677368164, "learning_rate": 9.736600000000001e-06, "loss": 0.0782, "step": 1318 }, { "epoch": 0.0703204137122141, "grad_norm": 13.713237762451172, "learning_rate": 9.7364e-06, "loss": 0.1309, "step": 1319 }, { "epoch": 0.07037372714186704, "grad_norm": 14.123847961425781, "learning_rate": 9.736200000000002e-06, "loss": 0.1245, "step": 1320 }, { "epoch": 0.07042704057151997, "grad_norm": 6.957522392272949, "learning_rate": 9.736000000000001e-06, "loss": 0.0964, "step": 1321 }, { "epoch": 0.07048035400117289, "grad_norm": 5.800316333770752, "learning_rate": 9.735800000000001e-06, "loss": 0.0807, "step": 1322 }, { "epoch": 0.07053366743082583, "grad_norm": 12.341042518615723, "learning_rate": 9.7356e-06, "loss": 0.1084, "step": 1323 }, { "epoch": 0.07058698086047875, "grad_norm": 9.54378890991211, "learning_rate": 9.7354e-06, "loss": 0.0909, "step": 1324 }, { "epoch": 0.07064029429013169, "grad_norm": 7.148706436157227, "learning_rate": 9.735200000000001e-06, "loss": 0.0789, "step": 1325 }, { "epoch": 0.07069360771978461, "grad_norm": 6.549839496612549, "learning_rate": 9.735e-06, "loss": 0.1242, "step": 1326 }, { "epoch": 0.07074692114943755, "grad_norm": 9.372785568237305, "learning_rate": 9.7348e-06, "loss": 0.1206, "step": 1327 }, { "epoch": 0.07080023457909047, "grad_norm": 7.053905487060547, "learning_rate": 9.7346e-06, "loss": 0.0854, "step": 1328 }, { "epoch": 0.0708535480087434, "grad_norm": 8.992955207824707, "learning_rate": 9.734400000000001e-06, "loss": 0.0487, "step": 1329 }, { "epoch": 0.07090686143839633, "grad_norm": 4.460865020751953, "learning_rate": 9.7342e-06, "loss": 0.098, "step": 1330 }, { "epoch": 0.07096017486804926, "grad_norm": 4.958316326141357, "learning_rate": 9.734000000000002e-06, "loss": 0.0797, "step": 1331 }, { "epoch": 0.0710134882977022, "grad_norm": 6.900412082672119, "learning_rate": 9.733800000000002e-06, "loss": 0.1012, "step": 1332 }, { "epoch": 0.07106680172735512, "grad_norm": 4.592297077178955, "learning_rate": 9.733600000000001e-06, "loss": 0.0741, "step": 1333 }, { "epoch": 0.07112011515700806, "grad_norm": 6.24752950668335, "learning_rate": 9.7334e-06, "loss": 0.1062, "step": 1334 }, { "epoch": 0.07117342858666098, "grad_norm": 11.984127044677734, "learning_rate": 9.7332e-06, "loss": 0.1381, "step": 1335 }, { "epoch": 0.0712267420163139, "grad_norm": 11.814473152160645, "learning_rate": 9.733000000000002e-06, "loss": 0.099, "step": 1336 }, { "epoch": 0.07128005544596684, "grad_norm": 6.5316267013549805, "learning_rate": 9.732800000000001e-06, "loss": 0.11, "step": 1337 }, { "epoch": 0.07133336887561977, "grad_norm": 9.282065391540527, "learning_rate": 9.7326e-06, "loss": 0.124, "step": 1338 }, { "epoch": 0.0713866823052727, "grad_norm": 7.570818901062012, "learning_rate": 9.7324e-06, "loss": 0.105, "step": 1339 }, { "epoch": 0.07143999573492563, "grad_norm": 6.405334949493408, "learning_rate": 9.7322e-06, "loss": 0.0983, "step": 1340 }, { "epoch": 0.07149330916457856, "grad_norm": 6.420897960662842, "learning_rate": 9.732000000000001e-06, "loss": 0.1422, "step": 1341 }, { "epoch": 0.07154662259423149, "grad_norm": 19.783220291137695, "learning_rate": 9.7318e-06, "loss": 0.1183, "step": 1342 }, { "epoch": 0.07159993602388441, "grad_norm": 8.535715103149414, "learning_rate": 9.731600000000002e-06, "loss": 0.0551, "step": 1343 }, { "epoch": 0.07165324945353735, "grad_norm": 7.330718040466309, "learning_rate": 9.731400000000001e-06, "loss": 0.0797, "step": 1344 }, { "epoch": 0.07170656288319027, "grad_norm": 6.764856338500977, "learning_rate": 9.731200000000001e-06, "loss": 0.114, "step": 1345 }, { "epoch": 0.07175987631284321, "grad_norm": 7.23422908782959, "learning_rate": 9.731e-06, "loss": 0.1074, "step": 1346 }, { "epoch": 0.07181318974249613, "grad_norm": 4.583471298217773, "learning_rate": 9.7308e-06, "loss": 0.0752, "step": 1347 }, { "epoch": 0.07186650317214907, "grad_norm": 9.015926361083984, "learning_rate": 9.730600000000001e-06, "loss": 0.0889, "step": 1348 }, { "epoch": 0.071919816601802, "grad_norm": 14.277802467346191, "learning_rate": 9.730400000000001e-06, "loss": 0.1166, "step": 1349 }, { "epoch": 0.07197313003145492, "grad_norm": 7.066869258880615, "learning_rate": 9.7302e-06, "loss": 0.1258, "step": 1350 }, { "epoch": 0.07202644346110786, "grad_norm": 5.221611976623535, "learning_rate": 9.73e-06, "loss": 0.1184, "step": 1351 }, { "epoch": 0.07207975689076078, "grad_norm": 6.213799953460693, "learning_rate": 9.7298e-06, "loss": 0.0971, "step": 1352 }, { "epoch": 0.07213307032041372, "grad_norm": 9.464666366577148, "learning_rate": 9.7296e-06, "loss": 0.1088, "step": 1353 }, { "epoch": 0.07218638375006664, "grad_norm": 6.459017276763916, "learning_rate": 9.7294e-06, "loss": 0.1156, "step": 1354 }, { "epoch": 0.07223969717971958, "grad_norm": 11.228836059570312, "learning_rate": 9.729200000000002e-06, "loss": 0.0962, "step": 1355 }, { "epoch": 0.0722930106093725, "grad_norm": 9.134706497192383, "learning_rate": 9.729000000000001e-06, "loss": 0.0754, "step": 1356 }, { "epoch": 0.07234632403902543, "grad_norm": 8.293281555175781, "learning_rate": 9.7288e-06, "loss": 0.144, "step": 1357 }, { "epoch": 0.07239963746867836, "grad_norm": 6.527919292449951, "learning_rate": 9.7286e-06, "loss": 0.1151, "step": 1358 }, { "epoch": 0.07245295089833129, "grad_norm": 6.18101692199707, "learning_rate": 9.728400000000002e-06, "loss": 0.1139, "step": 1359 }, { "epoch": 0.07250626432798422, "grad_norm": 7.157834529876709, "learning_rate": 9.728200000000001e-06, "loss": 0.1214, "step": 1360 }, { "epoch": 0.07255957775763715, "grad_norm": 4.036426067352295, "learning_rate": 9.728e-06, "loss": 0.0732, "step": 1361 }, { "epoch": 0.07261289118729009, "grad_norm": 7.137887001037598, "learning_rate": 9.7278e-06, "loss": 0.0758, "step": 1362 }, { "epoch": 0.07266620461694301, "grad_norm": 9.5221586227417, "learning_rate": 9.7276e-06, "loss": 0.0908, "step": 1363 }, { "epoch": 0.07271951804659593, "grad_norm": 8.058477401733398, "learning_rate": 9.727400000000001e-06, "loss": 0.0974, "step": 1364 }, { "epoch": 0.07277283147624887, "grad_norm": 10.134167671203613, "learning_rate": 9.7272e-06, "loss": 0.116, "step": 1365 }, { "epoch": 0.0728261449059018, "grad_norm": 10.108951568603516, "learning_rate": 9.727e-06, "loss": 0.0774, "step": 1366 }, { "epoch": 0.07287945833555473, "grad_norm": 4.354427814483643, "learning_rate": 9.726800000000001e-06, "loss": 0.0649, "step": 1367 }, { "epoch": 0.07293277176520765, "grad_norm": 3.758195638656616, "learning_rate": 9.726600000000001e-06, "loss": 0.0676, "step": 1368 }, { "epoch": 0.07298608519486059, "grad_norm": 4.549787521362305, "learning_rate": 9.7264e-06, "loss": 0.0829, "step": 1369 }, { "epoch": 0.07303939862451352, "grad_norm": 4.182229042053223, "learning_rate": 9.726200000000002e-06, "loss": 0.0696, "step": 1370 }, { "epoch": 0.07309271205416644, "grad_norm": 7.184840202331543, "learning_rate": 9.726000000000001e-06, "loss": 0.1011, "step": 1371 }, { "epoch": 0.07314602548381938, "grad_norm": 6.466052532196045, "learning_rate": 9.725800000000001e-06, "loss": 0.0946, "step": 1372 }, { "epoch": 0.0731993389134723, "grad_norm": 7.960040092468262, "learning_rate": 9.7256e-06, "loss": 0.1106, "step": 1373 }, { "epoch": 0.07325265234312524, "grad_norm": 5.317131519317627, "learning_rate": 9.7254e-06, "loss": 0.0877, "step": 1374 }, { "epoch": 0.07330596577277816, "grad_norm": 5.133398056030273, "learning_rate": 9.725200000000001e-06, "loss": 0.0761, "step": 1375 }, { "epoch": 0.0733592792024311, "grad_norm": 7.9382123947143555, "learning_rate": 9.725000000000001e-06, "loss": 0.1153, "step": 1376 }, { "epoch": 0.07341259263208402, "grad_norm": 7.642828941345215, "learning_rate": 9.7248e-06, "loss": 0.0578, "step": 1377 }, { "epoch": 0.07346590606173695, "grad_norm": 5.874636173248291, "learning_rate": 9.7246e-06, "loss": 0.1086, "step": 1378 }, { "epoch": 0.07351921949138988, "grad_norm": 7.214466094970703, "learning_rate": 9.724400000000001e-06, "loss": 0.0961, "step": 1379 }, { "epoch": 0.07357253292104281, "grad_norm": 5.8495635986328125, "learning_rate": 9.724200000000001e-06, "loss": 0.0685, "step": 1380 }, { "epoch": 0.07362584635069575, "grad_norm": 5.901286602020264, "learning_rate": 9.724e-06, "loss": 0.0842, "step": 1381 }, { "epoch": 0.07367915978034867, "grad_norm": 12.021254539489746, "learning_rate": 9.723800000000002e-06, "loss": 0.1007, "step": 1382 }, { "epoch": 0.0737324732100016, "grad_norm": 11.65130615234375, "learning_rate": 9.723600000000001e-06, "loss": 0.109, "step": 1383 }, { "epoch": 0.07378578663965453, "grad_norm": 14.193418502807617, "learning_rate": 9.7234e-06, "loss": 0.121, "step": 1384 }, { "epoch": 0.07383910006930745, "grad_norm": 5.676482677459717, "learning_rate": 9.7232e-06, "loss": 0.082, "step": 1385 }, { "epoch": 0.07389241349896039, "grad_norm": 11.860734939575195, "learning_rate": 9.723000000000002e-06, "loss": 0.0995, "step": 1386 }, { "epoch": 0.07394572692861331, "grad_norm": 7.007334232330322, "learning_rate": 9.722800000000001e-06, "loss": 0.1122, "step": 1387 }, { "epoch": 0.07399904035826625, "grad_norm": 6.28340482711792, "learning_rate": 9.7226e-06, "loss": 0.0815, "step": 1388 }, { "epoch": 0.07405235378791918, "grad_norm": 4.872664451599121, "learning_rate": 9.7224e-06, "loss": 0.069, "step": 1389 }, { "epoch": 0.07410566721757211, "grad_norm": 4.971083641052246, "learning_rate": 9.7222e-06, "loss": 0.0798, "step": 1390 }, { "epoch": 0.07415898064722504, "grad_norm": 7.022758483886719, "learning_rate": 9.722000000000001e-06, "loss": 0.1077, "step": 1391 }, { "epoch": 0.07421229407687796, "grad_norm": 7.200530529022217, "learning_rate": 9.7218e-06, "loss": 0.1046, "step": 1392 }, { "epoch": 0.0742656075065309, "grad_norm": 3.5102767944335938, "learning_rate": 9.7216e-06, "loss": 0.0665, "step": 1393 }, { "epoch": 0.07431892093618382, "grad_norm": 8.012306213378906, "learning_rate": 9.721400000000001e-06, "loss": 0.1102, "step": 1394 }, { "epoch": 0.07437223436583676, "grad_norm": 9.475774765014648, "learning_rate": 9.721200000000001e-06, "loss": 0.119, "step": 1395 }, { "epoch": 0.07442554779548968, "grad_norm": 9.549339294433594, "learning_rate": 9.721e-06, "loss": 0.1153, "step": 1396 }, { "epoch": 0.07447886122514262, "grad_norm": 7.162965297698975, "learning_rate": 9.720800000000002e-06, "loss": 0.0506, "step": 1397 }, { "epoch": 0.07453217465479554, "grad_norm": 6.299543380737305, "learning_rate": 9.720600000000001e-06, "loss": 0.0942, "step": 1398 }, { "epoch": 0.07458548808444847, "grad_norm": 5.358593463897705, "learning_rate": 9.720400000000001e-06, "loss": 0.0761, "step": 1399 }, { "epoch": 0.0746388015141014, "grad_norm": 6.6221160888671875, "learning_rate": 9.7202e-06, "loss": 0.0813, "step": 1400 }, { "epoch": 0.07469211494375433, "grad_norm": 6.584346294403076, "learning_rate": 9.72e-06, "loss": 0.0771, "step": 1401 }, { "epoch": 0.07474542837340727, "grad_norm": 9.66886043548584, "learning_rate": 9.7198e-06, "loss": 0.0759, "step": 1402 }, { "epoch": 0.07479874180306019, "grad_norm": 5.424915313720703, "learning_rate": 9.719600000000001e-06, "loss": 0.0682, "step": 1403 }, { "epoch": 0.07485205523271313, "grad_norm": 13.066831588745117, "learning_rate": 9.7194e-06, "loss": 0.1289, "step": 1404 }, { "epoch": 0.07490536866236605, "grad_norm": 6.4952168464660645, "learning_rate": 9.7192e-06, "loss": 0.0674, "step": 1405 }, { "epoch": 0.07495868209201897, "grad_norm": 6.671270847320557, "learning_rate": 9.719000000000001e-06, "loss": 0.0721, "step": 1406 }, { "epoch": 0.07501199552167191, "grad_norm": 5.7286763191223145, "learning_rate": 9.718800000000001e-06, "loss": 0.0732, "step": 1407 }, { "epoch": 0.07506530895132484, "grad_norm": 7.806839942932129, "learning_rate": 9.7186e-06, "loss": 0.0674, "step": 1408 }, { "epoch": 0.07511862238097777, "grad_norm": 13.762643814086914, "learning_rate": 9.718400000000002e-06, "loss": 0.1325, "step": 1409 }, { "epoch": 0.0751719358106307, "grad_norm": 6.760550498962402, "learning_rate": 9.718200000000001e-06, "loss": 0.0986, "step": 1410 }, { "epoch": 0.07522524924028363, "grad_norm": 10.939187049865723, "learning_rate": 9.718e-06, "loss": 0.1087, "step": 1411 }, { "epoch": 0.07527856266993656, "grad_norm": 11.401494979858398, "learning_rate": 9.7178e-06, "loss": 0.1474, "step": 1412 }, { "epoch": 0.07533187609958948, "grad_norm": 11.722922325134277, "learning_rate": 9.7176e-06, "loss": 0.117, "step": 1413 }, { "epoch": 0.07538518952924242, "grad_norm": 10.875382423400879, "learning_rate": 9.717400000000001e-06, "loss": 0.0698, "step": 1414 }, { "epoch": 0.07543850295889534, "grad_norm": 17.14560317993164, "learning_rate": 9.7172e-06, "loss": 0.1453, "step": 1415 }, { "epoch": 0.07549181638854828, "grad_norm": 6.183603763580322, "learning_rate": 9.717e-06, "loss": 0.0418, "step": 1416 }, { "epoch": 0.0755451298182012, "grad_norm": 12.062458038330078, "learning_rate": 9.7168e-06, "loss": 0.1538, "step": 1417 }, { "epoch": 0.07559844324785414, "grad_norm": 8.310050010681152, "learning_rate": 9.716600000000001e-06, "loss": 0.087, "step": 1418 }, { "epoch": 0.07565175667750706, "grad_norm": 11.880721092224121, "learning_rate": 9.7164e-06, "loss": 0.1637, "step": 1419 }, { "epoch": 0.07570507010715999, "grad_norm": 5.805741310119629, "learning_rate": 9.716200000000002e-06, "loss": 0.0633, "step": 1420 }, { "epoch": 0.07575838353681293, "grad_norm": 8.066566467285156, "learning_rate": 9.716000000000002e-06, "loss": 0.1225, "step": 1421 }, { "epoch": 0.07581169696646585, "grad_norm": 9.09829044342041, "learning_rate": 9.715800000000001e-06, "loss": 0.1058, "step": 1422 }, { "epoch": 0.07586501039611879, "grad_norm": 8.552882194519043, "learning_rate": 9.7156e-06, "loss": 0.1111, "step": 1423 }, { "epoch": 0.07591832382577171, "grad_norm": 6.582526683807373, "learning_rate": 9.7154e-06, "loss": 0.1282, "step": 1424 }, { "epoch": 0.07597163725542465, "grad_norm": 7.962866306304932, "learning_rate": 9.715200000000001e-06, "loss": 0.1069, "step": 1425 }, { "epoch": 0.07602495068507757, "grad_norm": 8.613215446472168, "learning_rate": 9.715000000000001e-06, "loss": 0.0964, "step": 1426 }, { "epoch": 0.0760782641147305, "grad_norm": 5.64771842956543, "learning_rate": 9.7148e-06, "loss": 0.098, "step": 1427 }, { "epoch": 0.07613157754438343, "grad_norm": 5.206039905548096, "learning_rate": 9.7146e-06, "loss": 0.0921, "step": 1428 }, { "epoch": 0.07618489097403636, "grad_norm": 4.944587707519531, "learning_rate": 9.7144e-06, "loss": 0.0781, "step": 1429 }, { "epoch": 0.0762382044036893, "grad_norm": 9.593969345092773, "learning_rate": 9.714200000000001e-06, "loss": 0.0956, "step": 1430 }, { "epoch": 0.07629151783334222, "grad_norm": 9.809225082397461, "learning_rate": 9.714e-06, "loss": 0.1437, "step": 1431 }, { "epoch": 0.07634483126299516, "grad_norm": 5.685484886169434, "learning_rate": 9.713800000000002e-06, "loss": 0.1057, "step": 1432 }, { "epoch": 0.07639814469264808, "grad_norm": 6.451715469360352, "learning_rate": 9.713600000000001e-06, "loss": 0.0813, "step": 1433 }, { "epoch": 0.076451458122301, "grad_norm": 5.8756184577941895, "learning_rate": 9.713400000000001e-06, "loss": 0.0789, "step": 1434 }, { "epoch": 0.07650477155195394, "grad_norm": 8.73114013671875, "learning_rate": 9.7132e-06, "loss": 0.0986, "step": 1435 }, { "epoch": 0.07655808498160686, "grad_norm": 7.438831806182861, "learning_rate": 9.713000000000002e-06, "loss": 0.0895, "step": 1436 }, { "epoch": 0.0766113984112598, "grad_norm": 4.120545387268066, "learning_rate": 9.712800000000001e-06, "loss": 0.0656, "step": 1437 }, { "epoch": 0.07666471184091272, "grad_norm": 5.340901851654053, "learning_rate": 9.7126e-06, "loss": 0.0961, "step": 1438 }, { "epoch": 0.07671802527056566, "grad_norm": 12.54427719116211, "learning_rate": 9.7124e-06, "loss": 0.0968, "step": 1439 }, { "epoch": 0.07677133870021859, "grad_norm": 9.89846420288086, "learning_rate": 9.7122e-06, "loss": 0.0727, "step": 1440 }, { "epoch": 0.07682465212987151, "grad_norm": 5.979603290557861, "learning_rate": 9.712e-06, "loss": 0.0774, "step": 1441 }, { "epoch": 0.07687796555952445, "grad_norm": 8.80598258972168, "learning_rate": 9.7118e-06, "loss": 0.112, "step": 1442 }, { "epoch": 0.07693127898917737, "grad_norm": 9.793399810791016, "learning_rate": 9.7116e-06, "loss": 0.1053, "step": 1443 }, { "epoch": 0.07698459241883031, "grad_norm": 10.985679626464844, "learning_rate": 9.711400000000002e-06, "loss": 0.1044, "step": 1444 }, { "epoch": 0.07703790584848323, "grad_norm": 5.231333255767822, "learning_rate": 9.711200000000001e-06, "loss": 0.0822, "step": 1445 }, { "epoch": 0.07709121927813617, "grad_norm": 6.3495192527771, "learning_rate": 9.711e-06, "loss": 0.0985, "step": 1446 }, { "epoch": 0.07714453270778909, "grad_norm": 13.894281387329102, "learning_rate": 9.710800000000002e-06, "loss": 0.1174, "step": 1447 }, { "epoch": 0.07719784613744202, "grad_norm": 15.216400146484375, "learning_rate": 9.710600000000002e-06, "loss": 0.1036, "step": 1448 }, { "epoch": 0.07725115956709495, "grad_norm": 5.260962009429932, "learning_rate": 9.710400000000001e-06, "loss": 0.1119, "step": 1449 }, { "epoch": 0.07730447299674788, "grad_norm": 10.437612533569336, "learning_rate": 9.7102e-06, "loss": 0.0979, "step": 1450 }, { "epoch": 0.07735778642640082, "grad_norm": 8.81939697265625, "learning_rate": 9.71e-06, "loss": 0.088, "step": 1451 }, { "epoch": 0.07741109985605374, "grad_norm": 8.980902671813965, "learning_rate": 9.7098e-06, "loss": 0.1166, "step": 1452 }, { "epoch": 0.07746441328570668, "grad_norm": 8.17115592956543, "learning_rate": 9.709600000000001e-06, "loss": 0.1379, "step": 1453 }, { "epoch": 0.0775177267153596, "grad_norm": 6.28400182723999, "learning_rate": 9.7094e-06, "loss": 0.0935, "step": 1454 }, { "epoch": 0.07757104014501252, "grad_norm": 9.784645080566406, "learning_rate": 9.7092e-06, "loss": 0.1072, "step": 1455 }, { "epoch": 0.07762435357466546, "grad_norm": 11.145628929138184, "learning_rate": 9.709000000000001e-06, "loss": 0.1227, "step": 1456 }, { "epoch": 0.07767766700431838, "grad_norm": 7.2091851234436035, "learning_rate": 9.708800000000001e-06, "loss": 0.1018, "step": 1457 }, { "epoch": 0.07773098043397132, "grad_norm": 6.184233665466309, "learning_rate": 9.7086e-06, "loss": 0.097, "step": 1458 }, { "epoch": 0.07778429386362425, "grad_norm": 10.771005630493164, "learning_rate": 9.708400000000002e-06, "loss": 0.1125, "step": 1459 }, { "epoch": 0.07783760729327718, "grad_norm": 10.539566993713379, "learning_rate": 9.708200000000001e-06, "loss": 0.0962, "step": 1460 }, { "epoch": 0.07789092072293011, "grad_norm": 9.195686340332031, "learning_rate": 9.708000000000001e-06, "loss": 0.1195, "step": 1461 }, { "epoch": 0.07794423415258303, "grad_norm": 6.0836873054504395, "learning_rate": 9.7078e-06, "loss": 0.0866, "step": 1462 }, { "epoch": 0.07799754758223597, "grad_norm": 9.670282363891602, "learning_rate": 9.7076e-06, "loss": 0.094, "step": 1463 }, { "epoch": 0.07805086101188889, "grad_norm": 5.940077781677246, "learning_rate": 9.707400000000001e-06, "loss": 0.0853, "step": 1464 }, { "epoch": 0.07810417444154183, "grad_norm": 9.235711097717285, "learning_rate": 9.7072e-06, "loss": 0.139, "step": 1465 }, { "epoch": 0.07815748787119475, "grad_norm": 8.194954872131348, "learning_rate": 9.707e-06, "loss": 0.0724, "step": 1466 }, { "epoch": 0.07821080130084769, "grad_norm": 6.908810615539551, "learning_rate": 9.7068e-06, "loss": 0.0871, "step": 1467 }, { "epoch": 0.07826411473050061, "grad_norm": 7.003875255584717, "learning_rate": 9.706600000000001e-06, "loss": 0.0786, "step": 1468 }, { "epoch": 0.07831742816015354, "grad_norm": 8.655607223510742, "learning_rate": 9.7064e-06, "loss": 0.11, "step": 1469 }, { "epoch": 0.07837074158980648, "grad_norm": 6.422076225280762, "learning_rate": 9.7062e-06, "loss": 0.1164, "step": 1470 }, { "epoch": 0.0784240550194594, "grad_norm": 4.511866092681885, "learning_rate": 9.706000000000002e-06, "loss": 0.0733, "step": 1471 }, { "epoch": 0.07847736844911234, "grad_norm": 8.5906982421875, "learning_rate": 9.705800000000001e-06, "loss": 0.1166, "step": 1472 }, { "epoch": 0.07853068187876526, "grad_norm": 4.609722137451172, "learning_rate": 9.7056e-06, "loss": 0.0964, "step": 1473 }, { "epoch": 0.0785839953084182, "grad_norm": 6.352220058441162, "learning_rate": 9.7054e-06, "loss": 0.0673, "step": 1474 }, { "epoch": 0.07863730873807112, "grad_norm": 5.373413562774658, "learning_rate": 9.705200000000002e-06, "loss": 0.087, "step": 1475 }, { "epoch": 0.07869062216772404, "grad_norm": 5.3209123611450195, "learning_rate": 9.705000000000001e-06, "loss": 0.0911, "step": 1476 }, { "epoch": 0.07874393559737698, "grad_norm": 6.7292799949646, "learning_rate": 9.7048e-06, "loss": 0.0584, "step": 1477 }, { "epoch": 0.0787972490270299, "grad_norm": 9.91110610961914, "learning_rate": 9.7046e-06, "loss": 0.1099, "step": 1478 }, { "epoch": 0.07885056245668284, "grad_norm": 5.2569899559021, "learning_rate": 9.7044e-06, "loss": 0.0963, "step": 1479 }, { "epoch": 0.07890387588633577, "grad_norm": 6.35845422744751, "learning_rate": 9.704200000000001e-06, "loss": 0.0877, "step": 1480 }, { "epoch": 0.07895718931598869, "grad_norm": 5.037757396697998, "learning_rate": 9.704e-06, "loss": 0.0613, "step": 1481 }, { "epoch": 0.07901050274564163, "grad_norm": 4.155538558959961, "learning_rate": 9.7038e-06, "loss": 0.0633, "step": 1482 }, { "epoch": 0.07906381617529455, "grad_norm": 10.47472095489502, "learning_rate": 9.703600000000001e-06, "loss": 0.0953, "step": 1483 }, { "epoch": 0.07911712960494749, "grad_norm": 8.092464447021484, "learning_rate": 9.703400000000001e-06, "loss": 0.1205, "step": 1484 }, { "epoch": 0.07917044303460041, "grad_norm": 9.086042404174805, "learning_rate": 9.7032e-06, "loss": 0.1228, "step": 1485 }, { "epoch": 0.07922375646425335, "grad_norm": 12.267413139343262, "learning_rate": 9.703000000000002e-06, "loss": 0.106, "step": 1486 }, { "epoch": 0.07927706989390627, "grad_norm": 5.901854991912842, "learning_rate": 9.702800000000001e-06, "loss": 0.1094, "step": 1487 }, { "epoch": 0.0793303833235592, "grad_norm": 3.670807361602783, "learning_rate": 9.702600000000001e-06, "loss": 0.0554, "step": 1488 }, { "epoch": 0.07938369675321214, "grad_norm": 5.8567023277282715, "learning_rate": 9.7024e-06, "loss": 0.0726, "step": 1489 }, { "epoch": 0.07943701018286506, "grad_norm": 10.072736740112305, "learning_rate": 9.7022e-06, "loss": 0.0974, "step": 1490 }, { "epoch": 0.079490323612518, "grad_norm": 6.1171040534973145, "learning_rate": 9.702e-06, "loss": 0.1194, "step": 1491 }, { "epoch": 0.07954363704217092, "grad_norm": 8.62149715423584, "learning_rate": 9.701800000000001e-06, "loss": 0.0787, "step": 1492 }, { "epoch": 0.07959695047182386, "grad_norm": 5.298532009124756, "learning_rate": 9.7016e-06, "loss": 0.0767, "step": 1493 }, { "epoch": 0.07965026390147678, "grad_norm": 8.557493209838867, "learning_rate": 9.7014e-06, "loss": 0.0817, "step": 1494 }, { "epoch": 0.0797035773311297, "grad_norm": 5.490959644317627, "learning_rate": 9.701200000000001e-06, "loss": 0.0762, "step": 1495 }, { "epoch": 0.07975689076078264, "grad_norm": 6.147198677062988, "learning_rate": 9.701e-06, "loss": 0.074, "step": 1496 }, { "epoch": 0.07981020419043557, "grad_norm": 11.036715507507324, "learning_rate": 9.700800000000002e-06, "loss": 0.0703, "step": 1497 }, { "epoch": 0.0798635176200885, "grad_norm": 4.24611759185791, "learning_rate": 9.700600000000002e-06, "loss": 0.0721, "step": 1498 }, { "epoch": 0.07991683104974143, "grad_norm": 12.064371109008789, "learning_rate": 9.700400000000001e-06, "loss": 0.1005, "step": 1499 }, { "epoch": 0.07997014447939436, "grad_norm": 6.782639026641846, "learning_rate": 9.7002e-06, "loss": 0.1113, "step": 1500 }, { "epoch": 0.08002345790904729, "grad_norm": 6.705643177032471, "learning_rate": 9.7e-06, "loss": 0.0951, "step": 1501 }, { "epoch": 0.08007677133870021, "grad_norm": 6.373834133148193, "learning_rate": 9.6998e-06, "loss": 0.1073, "step": 1502 }, { "epoch": 0.08013008476835315, "grad_norm": 5.431877613067627, "learning_rate": 9.699600000000001e-06, "loss": 0.0816, "step": 1503 }, { "epoch": 0.08018339819800607, "grad_norm": 8.418456077575684, "learning_rate": 9.6994e-06, "loss": 0.0799, "step": 1504 }, { "epoch": 0.08023671162765901, "grad_norm": 4.840212345123291, "learning_rate": 9.6992e-06, "loss": 0.0912, "step": 1505 }, { "epoch": 0.08029002505731193, "grad_norm": 12.366071701049805, "learning_rate": 9.699e-06, "loss": 0.1212, "step": 1506 }, { "epoch": 0.08034333848696487, "grad_norm": 7.238495349884033, "learning_rate": 9.698800000000001e-06, "loss": 0.0849, "step": 1507 }, { "epoch": 0.0803966519166178, "grad_norm": 13.02888298034668, "learning_rate": 9.6986e-06, "loss": 0.1285, "step": 1508 }, { "epoch": 0.08044996534627072, "grad_norm": 10.12817096710205, "learning_rate": 9.698400000000002e-06, "loss": 0.0906, "step": 1509 }, { "epoch": 0.08050327877592366, "grad_norm": 7.689854145050049, "learning_rate": 9.698200000000001e-06, "loss": 0.1001, "step": 1510 }, { "epoch": 0.08055659220557658, "grad_norm": 7.73274040222168, "learning_rate": 9.698000000000001e-06, "loss": 0.0938, "step": 1511 }, { "epoch": 0.08060990563522952, "grad_norm": 6.666817665100098, "learning_rate": 9.6978e-06, "loss": 0.0736, "step": 1512 }, { "epoch": 0.08066321906488244, "grad_norm": 4.933631420135498, "learning_rate": 9.6976e-06, "loss": 0.0773, "step": 1513 }, { "epoch": 0.08071653249453538, "grad_norm": 6.693402290344238, "learning_rate": 9.697400000000001e-06, "loss": 0.1024, "step": 1514 }, { "epoch": 0.0807698459241883, "grad_norm": 6.136996269226074, "learning_rate": 9.697200000000001e-06, "loss": 0.0904, "step": 1515 }, { "epoch": 0.08082315935384123, "grad_norm": 5.378804683685303, "learning_rate": 9.697e-06, "loss": 0.0999, "step": 1516 }, { "epoch": 0.08087647278349416, "grad_norm": 13.952888488769531, "learning_rate": 9.6968e-06, "loss": 0.0757, "step": 1517 }, { "epoch": 0.08092978621314709, "grad_norm": 5.850429534912109, "learning_rate": 9.6966e-06, "loss": 0.1048, "step": 1518 }, { "epoch": 0.08098309964280002, "grad_norm": 4.2012939453125, "learning_rate": 9.696400000000001e-06, "loss": 0.0804, "step": 1519 }, { "epoch": 0.08103641307245295, "grad_norm": 3.950129985809326, "learning_rate": 9.6962e-06, "loss": 0.0834, "step": 1520 }, { "epoch": 0.08108972650210589, "grad_norm": 5.369688987731934, "learning_rate": 9.696000000000002e-06, "loss": 0.076, "step": 1521 }, { "epoch": 0.08114303993175881, "grad_norm": 12.030744552612305, "learning_rate": 9.695800000000001e-06, "loss": 0.0852, "step": 1522 }, { "epoch": 0.08119635336141173, "grad_norm": 7.762292861938477, "learning_rate": 9.6956e-06, "loss": 0.1061, "step": 1523 }, { "epoch": 0.08124966679106467, "grad_norm": 9.969987869262695, "learning_rate": 9.6954e-06, "loss": 0.1021, "step": 1524 }, { "epoch": 0.0813029802207176, "grad_norm": 15.065942764282227, "learning_rate": 9.695200000000002e-06, "loss": 0.1572, "step": 1525 }, { "epoch": 0.08135629365037053, "grad_norm": 7.744112014770508, "learning_rate": 9.695000000000001e-06, "loss": 0.0842, "step": 1526 }, { "epoch": 0.08140960708002346, "grad_norm": 5.240216255187988, "learning_rate": 9.6948e-06, "loss": 0.0992, "step": 1527 }, { "epoch": 0.08146292050967639, "grad_norm": 8.018842697143555, "learning_rate": 9.6946e-06, "loss": 0.1054, "step": 1528 }, { "epoch": 0.08151623393932932, "grad_norm": 7.2317891120910645, "learning_rate": 9.6944e-06, "loss": 0.0814, "step": 1529 }, { "epoch": 0.08156954736898224, "grad_norm": 14.592001914978027, "learning_rate": 9.6942e-06, "loss": 0.1127, "step": 1530 }, { "epoch": 0.08162286079863518, "grad_norm": 8.342447280883789, "learning_rate": 9.694e-06, "loss": 0.0686, "step": 1531 }, { "epoch": 0.0816761742282881, "grad_norm": 9.002666473388672, "learning_rate": 9.6938e-06, "loss": 0.1202, "step": 1532 }, { "epoch": 0.08172948765794104, "grad_norm": 14.133782386779785, "learning_rate": 9.693600000000002e-06, "loss": 0.1612, "step": 1533 }, { "epoch": 0.08178280108759396, "grad_norm": 7.513899803161621, "learning_rate": 9.693400000000001e-06, "loss": 0.1471, "step": 1534 }, { "epoch": 0.0818361145172469, "grad_norm": 5.595844745635986, "learning_rate": 9.6932e-06, "loss": 0.0603, "step": 1535 }, { "epoch": 0.08188942794689982, "grad_norm": 7.566453456878662, "learning_rate": 9.693000000000002e-06, "loss": 0.0912, "step": 1536 }, { "epoch": 0.08194274137655275, "grad_norm": 6.851400852203369, "learning_rate": 9.692800000000001e-06, "loss": 0.0891, "step": 1537 }, { "epoch": 0.08199605480620568, "grad_norm": 16.62845802307129, "learning_rate": 9.692600000000001e-06, "loss": 0.1286, "step": 1538 }, { "epoch": 0.08204936823585861, "grad_norm": 5.224331855773926, "learning_rate": 9.6924e-06, "loss": 0.0991, "step": 1539 }, { "epoch": 0.08210268166551155, "grad_norm": 10.417040824890137, "learning_rate": 9.6922e-06, "loss": 0.1173, "step": 1540 }, { "epoch": 0.08215599509516447, "grad_norm": 10.153863906860352, "learning_rate": 9.692e-06, "loss": 0.0903, "step": 1541 }, { "epoch": 0.0822093085248174, "grad_norm": 5.821232795715332, "learning_rate": 9.691800000000001e-06, "loss": 0.1096, "step": 1542 }, { "epoch": 0.08226262195447033, "grad_norm": 3.7797951698303223, "learning_rate": 9.6916e-06, "loss": 0.0553, "step": 1543 }, { "epoch": 0.08231593538412325, "grad_norm": 12.12804889678955, "learning_rate": 9.6914e-06, "loss": 0.0946, "step": 1544 }, { "epoch": 0.08236924881377619, "grad_norm": 10.466700553894043, "learning_rate": 9.691200000000001e-06, "loss": 0.1032, "step": 1545 }, { "epoch": 0.08242256224342911, "grad_norm": 9.189945220947266, "learning_rate": 9.691000000000001e-06, "loss": 0.0937, "step": 1546 }, { "epoch": 0.08247587567308205, "grad_norm": 5.08502197265625, "learning_rate": 9.690800000000002e-06, "loss": 0.0892, "step": 1547 }, { "epoch": 0.08252918910273498, "grad_norm": 7.698066234588623, "learning_rate": 9.690600000000002e-06, "loss": 0.0909, "step": 1548 }, { "epoch": 0.08258250253238791, "grad_norm": 6.503532409667969, "learning_rate": 9.690400000000001e-06, "loss": 0.0969, "step": 1549 }, { "epoch": 0.08263581596204084, "grad_norm": 3.9736642837524414, "learning_rate": 9.6902e-06, "loss": 0.0695, "step": 1550 }, { "epoch": 0.08268912939169376, "grad_norm": 4.705900192260742, "learning_rate": 9.69e-06, "loss": 0.0865, "step": 1551 }, { "epoch": 0.0827424428213467, "grad_norm": 6.559675693511963, "learning_rate": 9.6898e-06, "loss": 0.0698, "step": 1552 }, { "epoch": 0.08279575625099962, "grad_norm": 13.826254844665527, "learning_rate": 9.689600000000001e-06, "loss": 0.0965, "step": 1553 }, { "epoch": 0.08284906968065256, "grad_norm": 7.577064037322998, "learning_rate": 9.6894e-06, "loss": 0.0799, "step": 1554 }, { "epoch": 0.08290238311030548, "grad_norm": 5.127850532531738, "learning_rate": 9.6892e-06, "loss": 0.0685, "step": 1555 }, { "epoch": 0.08295569653995842, "grad_norm": 9.125129699707031, "learning_rate": 9.689e-06, "loss": 0.1191, "step": 1556 }, { "epoch": 0.08300900996961134, "grad_norm": 5.636542797088623, "learning_rate": 9.688800000000001e-06, "loss": 0.0918, "step": 1557 }, { "epoch": 0.08306232339926427, "grad_norm": 5.322159767150879, "learning_rate": 9.6886e-06, "loss": 0.0608, "step": 1558 }, { "epoch": 0.0831156368289172, "grad_norm": 7.259111404418945, "learning_rate": 9.688400000000002e-06, "loss": 0.066, "step": 1559 }, { "epoch": 0.08316895025857013, "grad_norm": 6.103092670440674, "learning_rate": 9.688200000000002e-06, "loss": 0.0752, "step": 1560 }, { "epoch": 0.08322226368822307, "grad_norm": 6.295994758605957, "learning_rate": 9.688000000000001e-06, "loss": 0.078, "step": 1561 }, { "epoch": 0.08327557711787599, "grad_norm": 6.063397407531738, "learning_rate": 9.6878e-06, "loss": 0.0799, "step": 1562 }, { "epoch": 0.08332889054752893, "grad_norm": 5.565957546234131, "learning_rate": 9.6876e-06, "loss": 0.0751, "step": 1563 }, { "epoch": 0.08338220397718185, "grad_norm": 4.9365715980529785, "learning_rate": 9.687400000000001e-06, "loss": 0.0572, "step": 1564 }, { "epoch": 0.08343551740683477, "grad_norm": 3.6374318599700928, "learning_rate": 9.687200000000001e-06, "loss": 0.0473, "step": 1565 }, { "epoch": 0.08348883083648771, "grad_norm": 7.046898365020752, "learning_rate": 9.687e-06, "loss": 0.0707, "step": 1566 }, { "epoch": 0.08354214426614064, "grad_norm": 8.784867286682129, "learning_rate": 9.6868e-06, "loss": 0.0783, "step": 1567 }, { "epoch": 0.08359545769579357, "grad_norm": 5.502035140991211, "learning_rate": 9.6866e-06, "loss": 0.0653, "step": 1568 }, { "epoch": 0.0836487711254465, "grad_norm": 5.065954208374023, "learning_rate": 9.686400000000001e-06, "loss": 0.0643, "step": 1569 }, { "epoch": 0.08370208455509943, "grad_norm": 9.187886238098145, "learning_rate": 9.6862e-06, "loss": 0.1018, "step": 1570 }, { "epoch": 0.08375539798475236, "grad_norm": 7.422583103179932, "learning_rate": 9.686000000000002e-06, "loss": 0.1124, "step": 1571 }, { "epoch": 0.08380871141440528, "grad_norm": 6.746932029724121, "learning_rate": 9.685800000000001e-06, "loss": 0.0918, "step": 1572 }, { "epoch": 0.08386202484405822, "grad_norm": 4.971617221832275, "learning_rate": 9.685600000000001e-06, "loss": 0.0535, "step": 1573 }, { "epoch": 0.08391533827371114, "grad_norm": 6.474730968475342, "learning_rate": 9.6854e-06, "loss": 0.0934, "step": 1574 }, { "epoch": 0.08396865170336408, "grad_norm": 7.961223602294922, "learning_rate": 9.685200000000002e-06, "loss": 0.0848, "step": 1575 }, { "epoch": 0.084021965133017, "grad_norm": 6.483239650726318, "learning_rate": 9.685000000000001e-06, "loss": 0.0786, "step": 1576 }, { "epoch": 0.08407527856266994, "grad_norm": 5.9457502365112305, "learning_rate": 9.684800000000001e-06, "loss": 0.0859, "step": 1577 }, { "epoch": 0.08412859199232287, "grad_norm": 8.826349258422852, "learning_rate": 9.6846e-06, "loss": 0.0925, "step": 1578 }, { "epoch": 0.08418190542197579, "grad_norm": 7.825564861297607, "learning_rate": 9.6844e-06, "loss": 0.0753, "step": 1579 }, { "epoch": 0.08423521885162873, "grad_norm": 6.71626091003418, "learning_rate": 9.6842e-06, "loss": 0.083, "step": 1580 }, { "epoch": 0.08428853228128165, "grad_norm": 8.142829895019531, "learning_rate": 9.684e-06, "loss": 0.1155, "step": 1581 }, { "epoch": 0.08434184571093459, "grad_norm": 5.9116973876953125, "learning_rate": 9.6838e-06, "loss": 0.0979, "step": 1582 }, { "epoch": 0.08439515914058751, "grad_norm": 10.347718238830566, "learning_rate": 9.683600000000002e-06, "loss": 0.1068, "step": 1583 }, { "epoch": 0.08444847257024045, "grad_norm": 3.6952714920043945, "learning_rate": 9.683400000000001e-06, "loss": 0.0388, "step": 1584 }, { "epoch": 0.08450178599989337, "grad_norm": 7.986570358276367, "learning_rate": 9.6832e-06, "loss": 0.0629, "step": 1585 }, { "epoch": 0.0845550994295463, "grad_norm": 8.547807693481445, "learning_rate": 9.683000000000002e-06, "loss": 0.0875, "step": 1586 }, { "epoch": 0.08460841285919923, "grad_norm": 6.800418376922607, "learning_rate": 9.682800000000002e-06, "loss": 0.1204, "step": 1587 }, { "epoch": 0.08466172628885216, "grad_norm": 4.740189075469971, "learning_rate": 9.682600000000001e-06, "loss": 0.0881, "step": 1588 }, { "epoch": 0.0847150397185051, "grad_norm": 7.2401204109191895, "learning_rate": 9.6824e-06, "loss": 0.1275, "step": 1589 }, { "epoch": 0.08476835314815802, "grad_norm": 8.123764991760254, "learning_rate": 9.6822e-06, "loss": 0.1024, "step": 1590 }, { "epoch": 0.08482166657781096, "grad_norm": 6.854065418243408, "learning_rate": 9.682e-06, "loss": 0.1073, "step": 1591 }, { "epoch": 0.08487498000746388, "grad_norm": 3.7772231101989746, "learning_rate": 9.681800000000001e-06, "loss": 0.0894, "step": 1592 }, { "epoch": 0.0849282934371168, "grad_norm": 6.345895290374756, "learning_rate": 9.6816e-06, "loss": 0.0728, "step": 1593 }, { "epoch": 0.08498160686676974, "grad_norm": 6.035253524780273, "learning_rate": 9.6814e-06, "loss": 0.0747, "step": 1594 }, { "epoch": 0.08503492029642266, "grad_norm": 7.069423675537109, "learning_rate": 9.681200000000001e-06, "loss": 0.1254, "step": 1595 }, { "epoch": 0.0850882337260756, "grad_norm": 7.476380825042725, "learning_rate": 9.681000000000001e-06, "loss": 0.0851, "step": 1596 }, { "epoch": 0.08514154715572853, "grad_norm": 7.060535430908203, "learning_rate": 9.6808e-06, "loss": 0.12, "step": 1597 }, { "epoch": 0.08519486058538146, "grad_norm": 6.988683223724365, "learning_rate": 9.680600000000002e-06, "loss": 0.1388, "step": 1598 }, { "epoch": 0.08524817401503439, "grad_norm": 11.194375991821289, "learning_rate": 9.680400000000001e-06, "loss": 0.154, "step": 1599 }, { "epoch": 0.08530148744468731, "grad_norm": 7.803961753845215, "learning_rate": 9.680200000000001e-06, "loss": 0.1003, "step": 1600 }, { "epoch": 0.08535480087434025, "grad_norm": 5.9653801918029785, "learning_rate": 9.68e-06, "loss": 0.105, "step": 1601 }, { "epoch": 0.08540811430399317, "grad_norm": 6.190005779266357, "learning_rate": 9.6798e-06, "loss": 0.0796, "step": 1602 }, { "epoch": 0.08546142773364611, "grad_norm": 4.885382175445557, "learning_rate": 9.679600000000001e-06, "loss": 0.0703, "step": 1603 }, { "epoch": 0.08551474116329903, "grad_norm": 5.687381267547607, "learning_rate": 9.679400000000001e-06, "loss": 0.0766, "step": 1604 }, { "epoch": 0.08556805459295197, "grad_norm": 4.676196575164795, "learning_rate": 9.6792e-06, "loss": 0.0541, "step": 1605 }, { "epoch": 0.0856213680226049, "grad_norm": 8.580108642578125, "learning_rate": 9.679e-06, "loss": 0.0591, "step": 1606 }, { "epoch": 0.08567468145225782, "grad_norm": 4.740568161010742, "learning_rate": 9.678800000000001e-06, "loss": 0.0947, "step": 1607 }, { "epoch": 0.08572799488191075, "grad_norm": 6.236904144287109, "learning_rate": 9.6786e-06, "loss": 0.0942, "step": 1608 }, { "epoch": 0.08578130831156368, "grad_norm": 5.921385288238525, "learning_rate": 9.6784e-06, "loss": 0.0967, "step": 1609 }, { "epoch": 0.08583462174121662, "grad_norm": 7.589230537414551, "learning_rate": 9.678200000000002e-06, "loss": 0.1087, "step": 1610 }, { "epoch": 0.08588793517086954, "grad_norm": 8.973947525024414, "learning_rate": 9.678000000000001e-06, "loss": 0.1027, "step": 1611 }, { "epoch": 0.08594124860052248, "grad_norm": 8.600285530090332, "learning_rate": 9.6778e-06, "loss": 0.1051, "step": 1612 }, { "epoch": 0.0859945620301754, "grad_norm": 4.988226413726807, "learning_rate": 9.6776e-06, "loss": 0.0696, "step": 1613 }, { "epoch": 0.08604787545982832, "grad_norm": 7.920315265655518, "learning_rate": 9.677400000000002e-06, "loss": 0.0806, "step": 1614 }, { "epoch": 0.08610118888948126, "grad_norm": 8.26094913482666, "learning_rate": 9.677200000000001e-06, "loss": 0.0825, "step": 1615 }, { "epoch": 0.08615450231913419, "grad_norm": 7.205849647521973, "learning_rate": 9.677e-06, "loss": 0.1045, "step": 1616 }, { "epoch": 0.08620781574878712, "grad_norm": 9.900589942932129, "learning_rate": 9.6768e-06, "loss": 0.0978, "step": 1617 }, { "epoch": 0.08626112917844005, "grad_norm": 4.411928653717041, "learning_rate": 9.6766e-06, "loss": 0.0702, "step": 1618 }, { "epoch": 0.08631444260809298, "grad_norm": 6.876716613769531, "learning_rate": 9.676400000000001e-06, "loss": 0.0865, "step": 1619 }, { "epoch": 0.08636775603774591, "grad_norm": 4.2123799324035645, "learning_rate": 9.6762e-06, "loss": 0.0583, "step": 1620 }, { "epoch": 0.08642106946739883, "grad_norm": 4.512383460998535, "learning_rate": 9.676e-06, "loss": 0.1041, "step": 1621 }, { "epoch": 0.08647438289705177, "grad_norm": 4.956210613250732, "learning_rate": 9.675800000000001e-06, "loss": 0.1078, "step": 1622 }, { "epoch": 0.08652769632670469, "grad_norm": 6.418862342834473, "learning_rate": 9.675600000000001e-06, "loss": 0.0951, "step": 1623 }, { "epoch": 0.08658100975635763, "grad_norm": 5.58465051651001, "learning_rate": 9.6754e-06, "loss": 0.0683, "step": 1624 }, { "epoch": 0.08663432318601055, "grad_norm": 5.233416557312012, "learning_rate": 9.675200000000002e-06, "loss": 0.0673, "step": 1625 }, { "epoch": 0.08668763661566349, "grad_norm": 4.987259864807129, "learning_rate": 9.675000000000001e-06, "loss": 0.0727, "step": 1626 }, { "epoch": 0.08674095004531641, "grad_norm": 5.448206901550293, "learning_rate": 9.674800000000001e-06, "loss": 0.1052, "step": 1627 }, { "epoch": 0.08679426347496934, "grad_norm": 6.49127721786499, "learning_rate": 9.6746e-06, "loss": 0.059, "step": 1628 }, { "epoch": 0.08684757690462228, "grad_norm": 4.213440418243408, "learning_rate": 9.6744e-06, "loss": 0.0625, "step": 1629 }, { "epoch": 0.0869008903342752, "grad_norm": 6.451070308685303, "learning_rate": 9.6742e-06, "loss": 0.0929, "step": 1630 }, { "epoch": 0.08695420376392814, "grad_norm": 4.6461663246154785, "learning_rate": 9.674000000000001e-06, "loss": 0.0619, "step": 1631 }, { "epoch": 0.08700751719358106, "grad_norm": 8.20478630065918, "learning_rate": 9.6738e-06, "loss": 0.0952, "step": 1632 }, { "epoch": 0.087060830623234, "grad_norm": 3.7276525497436523, "learning_rate": 9.6736e-06, "loss": 0.0495, "step": 1633 }, { "epoch": 0.08711414405288692, "grad_norm": 9.742413520812988, "learning_rate": 9.673400000000001e-06, "loss": 0.0935, "step": 1634 }, { "epoch": 0.08716745748253985, "grad_norm": 4.278339862823486, "learning_rate": 9.6732e-06, "loss": 0.0561, "step": 1635 }, { "epoch": 0.08722077091219278, "grad_norm": 6.201415538787842, "learning_rate": 9.673000000000002e-06, "loss": 0.0641, "step": 1636 }, { "epoch": 0.0872740843418457, "grad_norm": 9.166988372802734, "learning_rate": 9.672800000000002e-06, "loss": 0.1024, "step": 1637 }, { "epoch": 0.08732739777149864, "grad_norm": 7.325932502746582, "learning_rate": 9.672600000000001e-06, "loss": 0.0709, "step": 1638 }, { "epoch": 0.08738071120115157, "grad_norm": 6.409237861633301, "learning_rate": 9.6724e-06, "loss": 0.096, "step": 1639 }, { "epoch": 0.0874340246308045, "grad_norm": 8.82451057434082, "learning_rate": 9.6722e-06, "loss": 0.0969, "step": 1640 }, { "epoch": 0.08748733806045743, "grad_norm": 9.460980415344238, "learning_rate": 9.672e-06, "loss": 0.0938, "step": 1641 }, { "epoch": 0.08754065149011035, "grad_norm": 6.281120300292969, "learning_rate": 9.671800000000001e-06, "loss": 0.1023, "step": 1642 }, { "epoch": 0.08759396491976329, "grad_norm": 9.445292472839355, "learning_rate": 9.6716e-06, "loss": 0.1091, "step": 1643 }, { "epoch": 0.08764727834941621, "grad_norm": 12.329477310180664, "learning_rate": 9.6714e-06, "loss": 0.0927, "step": 1644 }, { "epoch": 0.08770059177906915, "grad_norm": 7.679685115814209, "learning_rate": 9.6712e-06, "loss": 0.0639, "step": 1645 }, { "epoch": 0.08775390520872207, "grad_norm": 7.60012674331665, "learning_rate": 9.671000000000001e-06, "loss": 0.0953, "step": 1646 }, { "epoch": 0.08780721863837501, "grad_norm": 6.614062786102295, "learning_rate": 9.6708e-06, "loss": 0.1077, "step": 1647 }, { "epoch": 0.08786053206802794, "grad_norm": 5.573602676391602, "learning_rate": 9.670600000000002e-06, "loss": 0.0823, "step": 1648 }, { "epoch": 0.08791384549768086, "grad_norm": 6.890989303588867, "learning_rate": 9.670400000000001e-06, "loss": 0.0608, "step": 1649 }, { "epoch": 0.0879671589273338, "grad_norm": 5.244322776794434, "learning_rate": 9.670200000000001e-06, "loss": 0.0823, "step": 1650 }, { "epoch": 0.08802047235698672, "grad_norm": 9.89881420135498, "learning_rate": 9.67e-06, "loss": 0.0777, "step": 1651 }, { "epoch": 0.08807378578663966, "grad_norm": 4.666301727294922, "learning_rate": 9.6698e-06, "loss": 0.0821, "step": 1652 }, { "epoch": 0.08812709921629258, "grad_norm": 4.895158767700195, "learning_rate": 9.669600000000001e-06, "loss": 0.0927, "step": 1653 }, { "epoch": 0.08818041264594552, "grad_norm": 4.902342796325684, "learning_rate": 9.669400000000001e-06, "loss": 0.093, "step": 1654 }, { "epoch": 0.08823372607559844, "grad_norm": 5.294573783874512, "learning_rate": 9.6692e-06, "loss": 0.0528, "step": 1655 }, { "epoch": 0.08828703950525137, "grad_norm": 7.138918399810791, "learning_rate": 9.669e-06, "loss": 0.0891, "step": 1656 }, { "epoch": 0.0883403529349043, "grad_norm": 4.489591598510742, "learning_rate": 9.6688e-06, "loss": 0.0465, "step": 1657 }, { "epoch": 0.08839366636455723, "grad_norm": 5.170538425445557, "learning_rate": 9.668600000000001e-06, "loss": 0.1179, "step": 1658 }, { "epoch": 0.08844697979421016, "grad_norm": 5.325382232666016, "learning_rate": 9.6684e-06, "loss": 0.0613, "step": 1659 }, { "epoch": 0.08850029322386309, "grad_norm": 5.12196683883667, "learning_rate": 9.668200000000002e-06, "loss": 0.065, "step": 1660 }, { "epoch": 0.08855360665351603, "grad_norm": 5.197511196136475, "learning_rate": 9.668000000000001e-06, "loss": 0.0642, "step": 1661 }, { "epoch": 0.08860692008316895, "grad_norm": 4.9933905601501465, "learning_rate": 9.667800000000001e-06, "loss": 0.0409, "step": 1662 }, { "epoch": 0.08866023351282187, "grad_norm": 6.408590793609619, "learning_rate": 9.6676e-06, "loss": 0.0957, "step": 1663 }, { "epoch": 0.08871354694247481, "grad_norm": 15.710732460021973, "learning_rate": 9.667400000000002e-06, "loss": 0.1204, "step": 1664 }, { "epoch": 0.08876686037212773, "grad_norm": 11.226034164428711, "learning_rate": 9.667200000000001e-06, "loss": 0.161, "step": 1665 }, { "epoch": 0.08882017380178067, "grad_norm": 6.052797794342041, "learning_rate": 9.667e-06, "loss": 0.0635, "step": 1666 }, { "epoch": 0.0888734872314336, "grad_norm": 7.998771667480469, "learning_rate": 9.6668e-06, "loss": 0.0776, "step": 1667 }, { "epoch": 0.08892680066108653, "grad_norm": 7.7266058921813965, "learning_rate": 9.6666e-06, "loss": 0.0586, "step": 1668 }, { "epoch": 0.08898011409073946, "grad_norm": 6.744964599609375, "learning_rate": 9.666400000000001e-06, "loss": 0.139, "step": 1669 }, { "epoch": 0.08903342752039238, "grad_norm": 7.309122085571289, "learning_rate": 9.6662e-06, "loss": 0.104, "step": 1670 }, { "epoch": 0.08908674095004532, "grad_norm": 5.917224407196045, "learning_rate": 9.666e-06, "loss": 0.0891, "step": 1671 }, { "epoch": 0.08914005437969824, "grad_norm": 7.996727466583252, "learning_rate": 9.665800000000002e-06, "loss": 0.0981, "step": 1672 }, { "epoch": 0.08919336780935118, "grad_norm": 5.455536365509033, "learning_rate": 9.665600000000001e-06, "loss": 0.1041, "step": 1673 }, { "epoch": 0.0892466812390041, "grad_norm": 7.127310752868652, "learning_rate": 9.6654e-06, "loss": 0.0744, "step": 1674 }, { "epoch": 0.08929999466865704, "grad_norm": 5.1853742599487305, "learning_rate": 9.665200000000002e-06, "loss": 0.1087, "step": 1675 }, { "epoch": 0.08935330809830996, "grad_norm": 6.671501636505127, "learning_rate": 9.665000000000001e-06, "loss": 0.1034, "step": 1676 }, { "epoch": 0.08940662152796289, "grad_norm": 5.452585697174072, "learning_rate": 9.664800000000001e-06, "loss": 0.0999, "step": 1677 }, { "epoch": 0.08945993495761582, "grad_norm": 4.436854839324951, "learning_rate": 9.6646e-06, "loss": 0.0777, "step": 1678 }, { "epoch": 0.08951324838726875, "grad_norm": 4.761751651763916, "learning_rate": 9.6644e-06, "loss": 0.0646, "step": 1679 }, { "epoch": 0.08956656181692169, "grad_norm": 5.45147180557251, "learning_rate": 9.6642e-06, "loss": 0.1013, "step": 1680 }, { "epoch": 0.08961987524657461, "grad_norm": 3.809370994567871, "learning_rate": 9.664000000000001e-06, "loss": 0.0467, "step": 1681 }, { "epoch": 0.08967318867622755, "grad_norm": 6.302578449249268, "learning_rate": 9.6638e-06, "loss": 0.1354, "step": 1682 }, { "epoch": 0.08972650210588047, "grad_norm": 4.479809284210205, "learning_rate": 9.6636e-06, "loss": 0.0699, "step": 1683 }, { "epoch": 0.0897798155355334, "grad_norm": 6.800408363342285, "learning_rate": 9.663400000000001e-06, "loss": 0.1289, "step": 1684 }, { "epoch": 0.08983312896518633, "grad_norm": 6.342606544494629, "learning_rate": 9.663200000000001e-06, "loss": 0.0789, "step": 1685 }, { "epoch": 0.08988644239483926, "grad_norm": 9.352279663085938, "learning_rate": 9.663e-06, "loss": 0.083, "step": 1686 }, { "epoch": 0.08993975582449219, "grad_norm": 4.825088977813721, "learning_rate": 9.662800000000002e-06, "loss": 0.0828, "step": 1687 }, { "epoch": 0.08999306925414512, "grad_norm": 9.490744590759277, "learning_rate": 9.662600000000001e-06, "loss": 0.11, "step": 1688 }, { "epoch": 0.09004638268379805, "grad_norm": 5.3550801277160645, "learning_rate": 9.662400000000001e-06, "loss": 0.0714, "step": 1689 }, { "epoch": 0.09009969611345098, "grad_norm": 9.414670944213867, "learning_rate": 9.6622e-06, "loss": 0.0894, "step": 1690 }, { "epoch": 0.0901530095431039, "grad_norm": 8.278051376342773, "learning_rate": 9.662e-06, "loss": 0.0838, "step": 1691 }, { "epoch": 0.09020632297275684, "grad_norm": 10.996475219726562, "learning_rate": 9.661800000000001e-06, "loss": 0.0925, "step": 1692 }, { "epoch": 0.09025963640240976, "grad_norm": 3.3068885803222656, "learning_rate": 9.6616e-06, "loss": 0.042, "step": 1693 }, { "epoch": 0.0903129498320627, "grad_norm": 7.795329570770264, "learning_rate": 9.6614e-06, "loss": 0.0998, "step": 1694 }, { "epoch": 0.09036626326171562, "grad_norm": 16.24544334411621, "learning_rate": 9.6612e-06, "loss": 0.1476, "step": 1695 }, { "epoch": 0.09041957669136856, "grad_norm": 6.2334794998168945, "learning_rate": 9.661000000000001e-06, "loss": 0.0524, "step": 1696 }, { "epoch": 0.09047289012102148, "grad_norm": 6.42195987701416, "learning_rate": 9.6608e-06, "loss": 0.0754, "step": 1697 }, { "epoch": 0.09052620355067441, "grad_norm": 15.917716026306152, "learning_rate": 9.6606e-06, "loss": 0.1075, "step": 1698 }, { "epoch": 0.09057951698032735, "grad_norm": 6.290809631347656, "learning_rate": 9.660400000000002e-06, "loss": 0.0918, "step": 1699 }, { "epoch": 0.09063283040998027, "grad_norm": 9.3389310836792, "learning_rate": 9.660200000000001e-06, "loss": 0.0905, "step": 1700 }, { "epoch": 0.09068614383963321, "grad_norm": 3.6861867904663086, "learning_rate": 9.66e-06, "loss": 0.0629, "step": 1701 }, { "epoch": 0.09073945726928613, "grad_norm": 6.555775165557861, "learning_rate": 9.6598e-06, "loss": 0.1142, "step": 1702 }, { "epoch": 0.09079277069893907, "grad_norm": 6.278690338134766, "learning_rate": 9.659600000000002e-06, "loss": 0.0877, "step": 1703 }, { "epoch": 0.09084608412859199, "grad_norm": 5.355186462402344, "learning_rate": 9.659400000000001e-06, "loss": 0.0734, "step": 1704 }, { "epoch": 0.09089939755824492, "grad_norm": 5.296138763427734, "learning_rate": 9.6592e-06, "loss": 0.0911, "step": 1705 }, { "epoch": 0.09095271098789785, "grad_norm": 4.554244518280029, "learning_rate": 9.659e-06, "loss": 0.0581, "step": 1706 }, { "epoch": 0.09100602441755078, "grad_norm": 5.929847717285156, "learning_rate": 9.6588e-06, "loss": 0.137, "step": 1707 }, { "epoch": 0.09105933784720371, "grad_norm": 5.050040245056152, "learning_rate": 9.658600000000001e-06, "loss": 0.1079, "step": 1708 }, { "epoch": 0.09111265127685664, "grad_norm": 6.378777503967285, "learning_rate": 9.6584e-06, "loss": 0.0704, "step": 1709 }, { "epoch": 0.09116596470650958, "grad_norm": 7.36391019821167, "learning_rate": 9.6582e-06, "loss": 0.0978, "step": 1710 }, { "epoch": 0.0912192781361625, "grad_norm": 5.065075397491455, "learning_rate": 9.658000000000001e-06, "loss": 0.0812, "step": 1711 }, { "epoch": 0.09127259156581542, "grad_norm": 8.5441312789917, "learning_rate": 9.657800000000001e-06, "loss": 0.1479, "step": 1712 }, { "epoch": 0.09132590499546836, "grad_norm": 6.901577472686768, "learning_rate": 9.6576e-06, "loss": 0.0954, "step": 1713 }, { "epoch": 0.09137921842512128, "grad_norm": 8.171467781066895, "learning_rate": 9.657400000000002e-06, "loss": 0.086, "step": 1714 }, { "epoch": 0.09143253185477422, "grad_norm": 5.257411003112793, "learning_rate": 9.657200000000001e-06, "loss": 0.0511, "step": 1715 }, { "epoch": 0.09148584528442714, "grad_norm": 5.203339099884033, "learning_rate": 9.657000000000001e-06, "loss": 0.0697, "step": 1716 }, { "epoch": 0.09153915871408008, "grad_norm": 7.511033535003662, "learning_rate": 9.6568e-06, "loss": 0.1114, "step": 1717 }, { "epoch": 0.091592472143733, "grad_norm": 4.483554840087891, "learning_rate": 9.6566e-06, "loss": 0.0683, "step": 1718 }, { "epoch": 0.09164578557338593, "grad_norm": 4.286118030548096, "learning_rate": 9.656400000000001e-06, "loss": 0.0777, "step": 1719 }, { "epoch": 0.09169909900303887, "grad_norm": 5.506274223327637, "learning_rate": 9.6562e-06, "loss": 0.0882, "step": 1720 }, { "epoch": 0.09175241243269179, "grad_norm": 6.633643627166748, "learning_rate": 9.656e-06, "loss": 0.1075, "step": 1721 }, { "epoch": 0.09180572586234473, "grad_norm": 2.7963268756866455, "learning_rate": 9.6558e-06, "loss": 0.0355, "step": 1722 }, { "epoch": 0.09185903929199765, "grad_norm": 6.256730556488037, "learning_rate": 9.655600000000001e-06, "loss": 0.0998, "step": 1723 }, { "epoch": 0.09191235272165059, "grad_norm": 8.360176086425781, "learning_rate": 9.6554e-06, "loss": 0.0962, "step": 1724 }, { "epoch": 0.09196566615130351, "grad_norm": 8.899641036987305, "learning_rate": 9.655200000000002e-06, "loss": 0.1379, "step": 1725 }, { "epoch": 0.09201897958095644, "grad_norm": 16.619964599609375, "learning_rate": 9.655000000000002e-06, "loss": 0.1046, "step": 1726 }, { "epoch": 0.09207229301060937, "grad_norm": 5.128309726715088, "learning_rate": 9.654800000000001e-06, "loss": 0.0429, "step": 1727 }, { "epoch": 0.0921256064402623, "grad_norm": 11.584901809692383, "learning_rate": 9.6546e-06, "loss": 0.1082, "step": 1728 }, { "epoch": 0.09217891986991524, "grad_norm": 12.095208168029785, "learning_rate": 9.6544e-06, "loss": 0.1116, "step": 1729 }, { "epoch": 0.09223223329956816, "grad_norm": 6.331319332122803, "learning_rate": 9.6542e-06, "loss": 0.0788, "step": 1730 }, { "epoch": 0.0922855467292211, "grad_norm": 5.999672889709473, "learning_rate": 9.654000000000001e-06, "loss": 0.0718, "step": 1731 }, { "epoch": 0.09233886015887402, "grad_norm": 10.177460670471191, "learning_rate": 9.6538e-06, "loss": 0.1033, "step": 1732 }, { "epoch": 0.09239217358852694, "grad_norm": 6.381886959075928, "learning_rate": 9.6536e-06, "loss": 0.0705, "step": 1733 }, { "epoch": 0.09244548701817988, "grad_norm": 4.501198768615723, "learning_rate": 9.6534e-06, "loss": 0.0774, "step": 1734 }, { "epoch": 0.0924988004478328, "grad_norm": 6.8598408699035645, "learning_rate": 9.653200000000001e-06, "loss": 0.1132, "step": 1735 }, { "epoch": 0.09255211387748574, "grad_norm": 6.150145530700684, "learning_rate": 9.653e-06, "loss": 0.0943, "step": 1736 }, { "epoch": 0.09260542730713867, "grad_norm": 7.001084327697754, "learning_rate": 9.652800000000002e-06, "loss": 0.1111, "step": 1737 }, { "epoch": 0.0926587407367916, "grad_norm": 6.661590576171875, "learning_rate": 9.652600000000001e-06, "loss": 0.1134, "step": 1738 }, { "epoch": 0.09271205416644453, "grad_norm": 5.635904312133789, "learning_rate": 9.652400000000001e-06, "loss": 0.0721, "step": 1739 }, { "epoch": 0.09276536759609745, "grad_norm": 4.359408378601074, "learning_rate": 9.6522e-06, "loss": 0.067, "step": 1740 }, { "epoch": 0.09281868102575039, "grad_norm": 11.453203201293945, "learning_rate": 9.652e-06, "loss": 0.1243, "step": 1741 }, { "epoch": 0.09287199445540331, "grad_norm": 8.663905143737793, "learning_rate": 9.651800000000001e-06, "loss": 0.0923, "step": 1742 }, { "epoch": 0.09292530788505625, "grad_norm": 4.382359027862549, "learning_rate": 9.651600000000001e-06, "loss": 0.0894, "step": 1743 }, { "epoch": 0.09297862131470917, "grad_norm": 7.270852088928223, "learning_rate": 9.6514e-06, "loss": 0.0771, "step": 1744 }, { "epoch": 0.09303193474436211, "grad_norm": 5.81149435043335, "learning_rate": 9.6512e-06, "loss": 0.0595, "step": 1745 }, { "epoch": 0.09308524817401503, "grad_norm": 6.063785552978516, "learning_rate": 9.651e-06, "loss": 0.1204, "step": 1746 }, { "epoch": 0.09313856160366796, "grad_norm": 4.715359210968018, "learning_rate": 9.6508e-06, "loss": 0.0735, "step": 1747 }, { "epoch": 0.0931918750333209, "grad_norm": 4.732358932495117, "learning_rate": 9.6506e-06, "loss": 0.0751, "step": 1748 }, { "epoch": 0.09324518846297382, "grad_norm": 10.364123344421387, "learning_rate": 9.650400000000002e-06, "loss": 0.0828, "step": 1749 }, { "epoch": 0.09329850189262676, "grad_norm": 10.072189331054688, "learning_rate": 9.650200000000001e-06, "loss": 0.076, "step": 1750 }, { "epoch": 0.09335181532227968, "grad_norm": 3.546238422393799, "learning_rate": 9.65e-06, "loss": 0.06, "step": 1751 }, { "epoch": 0.09340512875193262, "grad_norm": 4.546898365020752, "learning_rate": 9.6498e-06, "loss": 0.0785, "step": 1752 }, { "epoch": 0.09345844218158554, "grad_norm": 7.09552001953125, "learning_rate": 9.649600000000002e-06, "loss": 0.0622, "step": 1753 }, { "epoch": 0.09351175561123846, "grad_norm": 7.95562219619751, "learning_rate": 9.649400000000001e-06, "loss": 0.0832, "step": 1754 }, { "epoch": 0.0935650690408914, "grad_norm": 5.467335224151611, "learning_rate": 9.6492e-06, "loss": 0.0696, "step": 1755 }, { "epoch": 0.09361838247054433, "grad_norm": 5.775519847869873, "learning_rate": 9.649e-06, "loss": 0.0907, "step": 1756 }, { "epoch": 0.09367169590019726, "grad_norm": 8.073285102844238, "learning_rate": 9.6488e-06, "loss": 0.0907, "step": 1757 }, { "epoch": 0.09372500932985019, "grad_norm": 5.740871429443359, "learning_rate": 9.648600000000001e-06, "loss": 0.069, "step": 1758 }, { "epoch": 0.09377832275950312, "grad_norm": 6.523553371429443, "learning_rate": 9.6484e-06, "loss": 0.0925, "step": 1759 }, { "epoch": 0.09383163618915605, "grad_norm": 7.873080730438232, "learning_rate": 9.6482e-06, "loss": 0.0643, "step": 1760 }, { "epoch": 0.09388494961880897, "grad_norm": 9.501702308654785, "learning_rate": 9.648000000000001e-06, "loss": 0.056, "step": 1761 }, { "epoch": 0.09393826304846191, "grad_norm": 4.290858268737793, "learning_rate": 9.647800000000001e-06, "loss": 0.0562, "step": 1762 }, { "epoch": 0.09399157647811483, "grad_norm": 7.91655969619751, "learning_rate": 9.6476e-06, "loss": 0.0757, "step": 1763 }, { "epoch": 0.09404488990776777, "grad_norm": 7.158578395843506, "learning_rate": 9.647400000000002e-06, "loss": 0.0704, "step": 1764 }, { "epoch": 0.0940982033374207, "grad_norm": 8.028894424438477, "learning_rate": 9.647200000000001e-06, "loss": 0.1354, "step": 1765 }, { "epoch": 0.09415151676707363, "grad_norm": 6.944857120513916, "learning_rate": 9.647000000000001e-06, "loss": 0.0786, "step": 1766 }, { "epoch": 0.09420483019672656, "grad_norm": 7.1781907081604, "learning_rate": 9.6468e-06, "loss": 0.056, "step": 1767 }, { "epoch": 0.09425814362637948, "grad_norm": 11.555838584899902, "learning_rate": 9.6466e-06, "loss": 0.1201, "step": 1768 }, { "epoch": 0.09431145705603242, "grad_norm": 9.516561508178711, "learning_rate": 9.646400000000001e-06, "loss": 0.0689, "step": 1769 }, { "epoch": 0.09436477048568534, "grad_norm": 4.2000651359558105, "learning_rate": 9.646200000000001e-06, "loss": 0.055, "step": 1770 }, { "epoch": 0.09441808391533828, "grad_norm": 3.598726987838745, "learning_rate": 9.646e-06, "loss": 0.0566, "step": 1771 }, { "epoch": 0.0944713973449912, "grad_norm": 5.002058029174805, "learning_rate": 9.6458e-06, "loss": 0.0553, "step": 1772 }, { "epoch": 0.09452471077464414, "grad_norm": 7.0874810218811035, "learning_rate": 9.645600000000001e-06, "loss": 0.0703, "step": 1773 }, { "epoch": 0.09457802420429706, "grad_norm": 6.183320045471191, "learning_rate": 9.645400000000001e-06, "loss": 0.0373, "step": 1774 }, { "epoch": 0.09463133763394999, "grad_norm": 4.001706600189209, "learning_rate": 9.6452e-06, "loss": 0.0573, "step": 1775 }, { "epoch": 0.09468465106360292, "grad_norm": 6.806792259216309, "learning_rate": 9.645000000000002e-06, "loss": 0.0822, "step": 1776 }, { "epoch": 0.09473796449325585, "grad_norm": 7.27153205871582, "learning_rate": 9.644800000000001e-06, "loss": 0.0746, "step": 1777 }, { "epoch": 0.09479127792290878, "grad_norm": 11.584488868713379, "learning_rate": 9.6446e-06, "loss": 0.0664, "step": 1778 }, { "epoch": 0.09484459135256171, "grad_norm": 5.487268447875977, "learning_rate": 9.6444e-06, "loss": 0.0729, "step": 1779 }, { "epoch": 0.09489790478221465, "grad_norm": 6.767502784729004, "learning_rate": 9.644200000000002e-06, "loss": 0.0509, "step": 1780 }, { "epoch": 0.09495121821186757, "grad_norm": 4.498756408691406, "learning_rate": 9.644000000000001e-06, "loss": 0.0482, "step": 1781 }, { "epoch": 0.09500453164152049, "grad_norm": 5.353474140167236, "learning_rate": 9.6438e-06, "loss": 0.0543, "step": 1782 }, { "epoch": 0.09505784507117343, "grad_norm": 5.283037185668945, "learning_rate": 9.6436e-06, "loss": 0.0835, "step": 1783 }, { "epoch": 0.09511115850082635, "grad_norm": 8.094268798828125, "learning_rate": 9.6434e-06, "loss": 0.122, "step": 1784 }, { "epoch": 0.09516447193047929, "grad_norm": 5.239691257476807, "learning_rate": 9.643200000000001e-06, "loss": 0.0653, "step": 1785 }, { "epoch": 0.09521778536013221, "grad_norm": 4.093641757965088, "learning_rate": 9.643e-06, "loss": 0.0966, "step": 1786 }, { "epoch": 0.09527109878978515, "grad_norm": 7.818663120269775, "learning_rate": 9.6428e-06, "loss": 0.1248, "step": 1787 }, { "epoch": 0.09532441221943808, "grad_norm": 9.518045425415039, "learning_rate": 9.642600000000001e-06, "loss": 0.12, "step": 1788 }, { "epoch": 0.095377725649091, "grad_norm": 10.957491874694824, "learning_rate": 9.642400000000001e-06, "loss": 0.075, "step": 1789 }, { "epoch": 0.09543103907874394, "grad_norm": 6.194788932800293, "learning_rate": 9.6422e-06, "loss": 0.0479, "step": 1790 }, { "epoch": 0.09548435250839686, "grad_norm": 6.47603702545166, "learning_rate": 9.642e-06, "loss": 0.0691, "step": 1791 }, { "epoch": 0.0955376659380498, "grad_norm": 12.064579010009766, "learning_rate": 9.641800000000001e-06, "loss": 0.1262, "step": 1792 }, { "epoch": 0.09559097936770272, "grad_norm": 5.0280327796936035, "learning_rate": 9.641600000000001e-06, "loss": 0.0643, "step": 1793 }, { "epoch": 0.09564429279735566, "grad_norm": 5.638038158416748, "learning_rate": 9.6414e-06, "loss": 0.0703, "step": 1794 }, { "epoch": 0.09569760622700858, "grad_norm": 6.001587867736816, "learning_rate": 9.6412e-06, "loss": 0.0614, "step": 1795 }, { "epoch": 0.0957509196566615, "grad_norm": 5.977590560913086, "learning_rate": 9.641e-06, "loss": 0.0721, "step": 1796 }, { "epoch": 0.09580423308631444, "grad_norm": 5.022111415863037, "learning_rate": 9.640800000000001e-06, "loss": 0.0828, "step": 1797 }, { "epoch": 0.09585754651596737, "grad_norm": 9.867451667785645, "learning_rate": 9.6406e-06, "loss": 0.0969, "step": 1798 }, { "epoch": 0.0959108599456203, "grad_norm": 5.901682376861572, "learning_rate": 9.6404e-06, "loss": 0.0938, "step": 1799 }, { "epoch": 0.09596417337527323, "grad_norm": 5.571822643280029, "learning_rate": 9.640200000000001e-06, "loss": 0.0521, "step": 1800 }, { "epoch": 0.09601748680492617, "grad_norm": 6.900478839874268, "learning_rate": 9.640000000000001e-06, "loss": 0.0828, "step": 1801 }, { "epoch": 0.09607080023457909, "grad_norm": 5.554034233093262, "learning_rate": 9.6398e-06, "loss": 0.0602, "step": 1802 }, { "epoch": 0.09612411366423201, "grad_norm": 8.502317428588867, "learning_rate": 9.639600000000002e-06, "loss": 0.095, "step": 1803 }, { "epoch": 0.09617742709388495, "grad_norm": 10.463675498962402, "learning_rate": 9.639400000000001e-06, "loss": 0.0961, "step": 1804 }, { "epoch": 0.09623074052353787, "grad_norm": 3.882549285888672, "learning_rate": 9.6392e-06, "loss": 0.0595, "step": 1805 }, { "epoch": 0.09628405395319081, "grad_norm": 9.39457893371582, "learning_rate": 9.639e-06, "loss": 0.0704, "step": 1806 }, { "epoch": 0.09633736738284374, "grad_norm": 11.789401054382324, "learning_rate": 9.6388e-06, "loss": 0.0709, "step": 1807 }, { "epoch": 0.09639068081249667, "grad_norm": 4.416109561920166, "learning_rate": 9.638600000000001e-06, "loss": 0.0721, "step": 1808 }, { "epoch": 0.0964439942421496, "grad_norm": 5.473197937011719, "learning_rate": 9.6384e-06, "loss": 0.1125, "step": 1809 }, { "epoch": 0.09649730767180252, "grad_norm": 7.211398601531982, "learning_rate": 9.6382e-06, "loss": 0.0776, "step": 1810 }, { "epoch": 0.09655062110145546, "grad_norm": 9.82418441772461, "learning_rate": 9.638e-06, "loss": 0.0958, "step": 1811 }, { "epoch": 0.09660393453110838, "grad_norm": 5.5680036544799805, "learning_rate": 9.637800000000001e-06, "loss": 0.059, "step": 1812 }, { "epoch": 0.09665724796076132, "grad_norm": 6.15327787399292, "learning_rate": 9.6376e-06, "loss": 0.0644, "step": 1813 }, { "epoch": 0.09671056139041424, "grad_norm": 5.96987247467041, "learning_rate": 9.637400000000002e-06, "loss": 0.0751, "step": 1814 }, { "epoch": 0.09676387482006718, "grad_norm": 5.079755783081055, "learning_rate": 9.637200000000002e-06, "loss": 0.0732, "step": 1815 }, { "epoch": 0.0968171882497201, "grad_norm": 4.693946838378906, "learning_rate": 9.637000000000001e-06, "loss": 0.0867, "step": 1816 }, { "epoch": 0.09687050167937303, "grad_norm": 9.28245735168457, "learning_rate": 9.6368e-06, "loss": 0.0623, "step": 1817 }, { "epoch": 0.09692381510902597, "grad_norm": 5.600433349609375, "learning_rate": 9.6366e-06, "loss": 0.0725, "step": 1818 }, { "epoch": 0.09697712853867889, "grad_norm": 6.89528226852417, "learning_rate": 9.636400000000001e-06, "loss": 0.0717, "step": 1819 }, { "epoch": 0.09703044196833183, "grad_norm": 8.8065767288208, "learning_rate": 9.636200000000001e-06, "loss": 0.1705, "step": 1820 }, { "epoch": 0.09708375539798475, "grad_norm": 6.2719831466674805, "learning_rate": 9.636e-06, "loss": 0.0691, "step": 1821 }, { "epoch": 0.09713706882763769, "grad_norm": 9.993330001831055, "learning_rate": 9.6358e-06, "loss": 0.0834, "step": 1822 }, { "epoch": 0.09719038225729061, "grad_norm": 4.747758865356445, "learning_rate": 9.6356e-06, "loss": 0.065, "step": 1823 }, { "epoch": 0.09724369568694353, "grad_norm": 6.988314151763916, "learning_rate": 9.635400000000001e-06, "loss": 0.0877, "step": 1824 }, { "epoch": 0.09729700911659647, "grad_norm": 9.661299705505371, "learning_rate": 9.6352e-06, "loss": 0.109, "step": 1825 }, { "epoch": 0.0973503225462494, "grad_norm": 9.436357498168945, "learning_rate": 9.635000000000002e-06, "loss": 0.1293, "step": 1826 }, { "epoch": 0.09740363597590233, "grad_norm": 5.301903247833252, "learning_rate": 9.634800000000001e-06, "loss": 0.0859, "step": 1827 }, { "epoch": 0.09745694940555526, "grad_norm": 8.486394882202148, "learning_rate": 9.634600000000001e-06, "loss": 0.0655, "step": 1828 }, { "epoch": 0.0975102628352082, "grad_norm": 11.203658103942871, "learning_rate": 9.6344e-06, "loss": 0.0755, "step": 1829 }, { "epoch": 0.09756357626486112, "grad_norm": 7.60983943939209, "learning_rate": 9.634200000000002e-06, "loss": 0.0984, "step": 1830 }, { "epoch": 0.09761688969451404, "grad_norm": 8.751531600952148, "learning_rate": 9.634000000000001e-06, "loss": 0.0882, "step": 1831 }, { "epoch": 0.09767020312416698, "grad_norm": 6.029045581817627, "learning_rate": 9.6338e-06, "loss": 0.0638, "step": 1832 }, { "epoch": 0.0977235165538199, "grad_norm": 4.504543781280518, "learning_rate": 9.6336e-06, "loss": 0.0653, "step": 1833 }, { "epoch": 0.09777682998347284, "grad_norm": 6.256695747375488, "learning_rate": 9.6334e-06, "loss": 0.0606, "step": 1834 }, { "epoch": 0.09783014341312576, "grad_norm": 9.07538890838623, "learning_rate": 9.6332e-06, "loss": 0.0652, "step": 1835 }, { "epoch": 0.0978834568427787, "grad_norm": 6.668236255645752, "learning_rate": 9.633e-06, "loss": 0.0901, "step": 1836 }, { "epoch": 0.09793677027243163, "grad_norm": 8.502459526062012, "learning_rate": 9.6328e-06, "loss": 0.144, "step": 1837 }, { "epoch": 0.09799008370208455, "grad_norm": 2.683502674102783, "learning_rate": 9.632600000000002e-06, "loss": 0.0304, "step": 1838 }, { "epoch": 0.09804339713173749, "grad_norm": 5.506189346313477, "learning_rate": 9.632400000000001e-06, "loss": 0.0647, "step": 1839 }, { "epoch": 0.09809671056139041, "grad_norm": 6.012531757354736, "learning_rate": 9.6322e-06, "loss": 0.0939, "step": 1840 }, { "epoch": 0.09815002399104335, "grad_norm": 7.532972812652588, "learning_rate": 9.632e-06, "loss": 0.1246, "step": 1841 }, { "epoch": 0.09820333742069627, "grad_norm": 11.146320343017578, "learning_rate": 9.631800000000002e-06, "loss": 0.0858, "step": 1842 }, { "epoch": 0.09825665085034921, "grad_norm": 10.902961730957031, "learning_rate": 9.631600000000001e-06, "loss": 0.1071, "step": 1843 }, { "epoch": 0.09830996428000213, "grad_norm": 4.08650016784668, "learning_rate": 9.6314e-06, "loss": 0.07, "step": 1844 }, { "epoch": 0.09836327770965506, "grad_norm": 5.236686706542969, "learning_rate": 9.6312e-06, "loss": 0.0603, "step": 1845 }, { "epoch": 0.098416591139308, "grad_norm": 7.920804977416992, "learning_rate": 9.631e-06, "loss": 0.08, "step": 1846 }, { "epoch": 0.09846990456896092, "grad_norm": 9.919190406799316, "learning_rate": 9.630800000000001e-06, "loss": 0.0851, "step": 1847 }, { "epoch": 0.09852321799861385, "grad_norm": 6.6669921875, "learning_rate": 9.6306e-06, "loss": 0.0882, "step": 1848 }, { "epoch": 0.09857653142826678, "grad_norm": 12.194072723388672, "learning_rate": 9.6304e-06, "loss": 0.0794, "step": 1849 }, { "epoch": 0.09862984485791972, "grad_norm": 5.07260274887085, "learning_rate": 9.630200000000001e-06, "loss": 0.0701, "step": 1850 }, { "epoch": 0.09868315828757264, "grad_norm": 7.5373969078063965, "learning_rate": 9.630000000000001e-06, "loss": 0.1087, "step": 1851 }, { "epoch": 0.09873647171722556, "grad_norm": 5.2321600914001465, "learning_rate": 9.6298e-06, "loss": 0.0719, "step": 1852 }, { "epoch": 0.0987897851468785, "grad_norm": 6.845871925354004, "learning_rate": 9.629600000000002e-06, "loss": 0.0771, "step": 1853 }, { "epoch": 0.09884309857653142, "grad_norm": 6.313289165496826, "learning_rate": 9.629400000000001e-06, "loss": 0.0625, "step": 1854 }, { "epoch": 0.09889641200618436, "grad_norm": 9.320115089416504, "learning_rate": 9.629200000000001e-06, "loss": 0.1088, "step": 1855 }, { "epoch": 0.09894972543583729, "grad_norm": 6.689497947692871, "learning_rate": 9.629e-06, "loss": 0.0487, "step": 1856 }, { "epoch": 0.09900303886549022, "grad_norm": 6.411299705505371, "learning_rate": 9.6288e-06, "loss": 0.1018, "step": 1857 }, { "epoch": 0.09905635229514315, "grad_norm": 6.57082986831665, "learning_rate": 9.628600000000001e-06, "loss": 0.0745, "step": 1858 }, { "epoch": 0.09910966572479607, "grad_norm": 4.3888421058654785, "learning_rate": 9.628400000000001e-06, "loss": 0.0616, "step": 1859 }, { "epoch": 0.09916297915444901, "grad_norm": 6.757158279418945, "learning_rate": 9.6282e-06, "loss": 0.0994, "step": 1860 }, { "epoch": 0.09921629258410193, "grad_norm": 8.014549255371094, "learning_rate": 9.628e-06, "loss": 0.0829, "step": 1861 }, { "epoch": 0.09926960601375487, "grad_norm": 6.935234546661377, "learning_rate": 9.627800000000001e-06, "loss": 0.0492, "step": 1862 }, { "epoch": 0.09932291944340779, "grad_norm": 5.575379371643066, "learning_rate": 9.6276e-06, "loss": 0.0687, "step": 1863 }, { "epoch": 0.09937623287306073, "grad_norm": 5.262087821960449, "learning_rate": 9.6274e-06, "loss": 0.0682, "step": 1864 }, { "epoch": 0.09942954630271365, "grad_norm": 6.59810209274292, "learning_rate": 9.627200000000002e-06, "loss": 0.0883, "step": 1865 }, { "epoch": 0.09948285973236658, "grad_norm": 3.7932398319244385, "learning_rate": 9.627000000000001e-06, "loss": 0.041, "step": 1866 }, { "epoch": 0.09953617316201951, "grad_norm": 7.930037498474121, "learning_rate": 9.6268e-06, "loss": 0.0721, "step": 1867 }, { "epoch": 0.09958948659167244, "grad_norm": 5.306077480316162, "learning_rate": 9.6266e-06, "loss": 0.0675, "step": 1868 }, { "epoch": 0.09964280002132538, "grad_norm": 5.01995325088501, "learning_rate": 9.626400000000002e-06, "loss": 0.0472, "step": 1869 }, { "epoch": 0.0996961134509783, "grad_norm": 5.824914455413818, "learning_rate": 9.626200000000001e-06, "loss": 0.0699, "step": 1870 }, { "epoch": 0.09974942688063124, "grad_norm": 5.721524715423584, "learning_rate": 9.626e-06, "loss": 0.0689, "step": 1871 }, { "epoch": 0.09980274031028416, "grad_norm": 5.1678876876831055, "learning_rate": 9.6258e-06, "loss": 0.0724, "step": 1872 }, { "epoch": 0.09985605373993708, "grad_norm": 10.776748657226562, "learning_rate": 9.6256e-06, "loss": 0.0782, "step": 1873 }, { "epoch": 0.09990936716959002, "grad_norm": 11.11640453338623, "learning_rate": 9.625400000000001e-06, "loss": 0.0902, "step": 1874 }, { "epoch": 0.09996268059924295, "grad_norm": 6.278106212615967, "learning_rate": 9.6252e-06, "loss": 0.0581, "step": 1875 }, { "epoch": 0.10001599402889588, "grad_norm": 7.229669570922852, "learning_rate": 9.625e-06, "loss": 0.0878, "step": 1876 }, { "epoch": 0.1000693074585488, "grad_norm": 5.863879203796387, "learning_rate": 9.624800000000001e-06, "loss": 0.0568, "step": 1877 }, { "epoch": 0.10012262088820174, "grad_norm": 8.166887283325195, "learning_rate": 9.624600000000001e-06, "loss": 0.0737, "step": 1878 }, { "epoch": 0.10017593431785467, "grad_norm": 6.8063812255859375, "learning_rate": 9.6244e-06, "loss": 0.0864, "step": 1879 }, { "epoch": 0.10022924774750759, "grad_norm": 4.726382732391357, "learning_rate": 9.624200000000002e-06, "loss": 0.0634, "step": 1880 }, { "epoch": 0.10028256117716053, "grad_norm": 7.842065811157227, "learning_rate": 9.624000000000001e-06, "loss": 0.0723, "step": 1881 }, { "epoch": 0.10033587460681345, "grad_norm": 8.062018394470215, "learning_rate": 9.623800000000001e-06, "loss": 0.1437, "step": 1882 }, { "epoch": 0.10038918803646639, "grad_norm": 5.830296516418457, "learning_rate": 9.6236e-06, "loss": 0.0689, "step": 1883 }, { "epoch": 0.10044250146611931, "grad_norm": 8.087141036987305, "learning_rate": 9.6234e-06, "loss": 0.0932, "step": 1884 }, { "epoch": 0.10049581489577225, "grad_norm": 7.1936187744140625, "learning_rate": 9.6232e-06, "loss": 0.0561, "step": 1885 }, { "epoch": 0.10054912832542517, "grad_norm": 7.6173577308654785, "learning_rate": 9.623000000000001e-06, "loss": 0.1026, "step": 1886 }, { "epoch": 0.1006024417550781, "grad_norm": 5.053377151489258, "learning_rate": 9.6228e-06, "loss": 0.0632, "step": 1887 }, { "epoch": 0.10065575518473104, "grad_norm": 6.07616662979126, "learning_rate": 9.6226e-06, "loss": 0.1009, "step": 1888 }, { "epoch": 0.10070906861438396, "grad_norm": 4.822805881500244, "learning_rate": 9.622400000000001e-06, "loss": 0.0848, "step": 1889 }, { "epoch": 0.1007623820440369, "grad_norm": 6.505133628845215, "learning_rate": 9.6222e-06, "loss": 0.0727, "step": 1890 }, { "epoch": 0.10081569547368982, "grad_norm": 7.332902431488037, "learning_rate": 9.622000000000002e-06, "loss": 0.0852, "step": 1891 }, { "epoch": 0.10086900890334276, "grad_norm": 6.866931438446045, "learning_rate": 9.621800000000002e-06, "loss": 0.0684, "step": 1892 }, { "epoch": 0.10092232233299568, "grad_norm": 7.023245811462402, "learning_rate": 9.621600000000001e-06, "loss": 0.1106, "step": 1893 }, { "epoch": 0.1009756357626486, "grad_norm": 6.000154495239258, "learning_rate": 9.6214e-06, "loss": 0.1019, "step": 1894 }, { "epoch": 0.10102894919230154, "grad_norm": 7.315256118774414, "learning_rate": 9.6212e-06, "loss": 0.1217, "step": 1895 }, { "epoch": 0.10108226262195447, "grad_norm": 8.520294189453125, "learning_rate": 9.621e-06, "loss": 0.1148, "step": 1896 }, { "epoch": 0.1011355760516074, "grad_norm": 4.721472263336182, "learning_rate": 9.620800000000001e-06, "loss": 0.0926, "step": 1897 }, { "epoch": 0.10118888948126033, "grad_norm": 6.181365013122559, "learning_rate": 9.6206e-06, "loss": 0.0786, "step": 1898 }, { "epoch": 0.10124220291091326, "grad_norm": 2.944666624069214, "learning_rate": 9.6204e-06, "loss": 0.0344, "step": 1899 }, { "epoch": 0.10129551634056619, "grad_norm": 4.148434638977051, "learning_rate": 9.6202e-06, "loss": 0.0665, "step": 1900 }, { "epoch": 0.10134882977021911, "grad_norm": 8.192036628723145, "learning_rate": 9.620000000000001e-06, "loss": 0.0843, "step": 1901 }, { "epoch": 0.10140214319987205, "grad_norm": 8.239766120910645, "learning_rate": 9.6198e-06, "loss": 0.0795, "step": 1902 }, { "epoch": 0.10145545662952497, "grad_norm": 7.083098888397217, "learning_rate": 9.619600000000002e-06, "loss": 0.0705, "step": 1903 }, { "epoch": 0.10150877005917791, "grad_norm": 3.70839786529541, "learning_rate": 9.619400000000001e-06, "loss": 0.0627, "step": 1904 }, { "epoch": 0.10156208348883083, "grad_norm": 11.369545936584473, "learning_rate": 9.619200000000001e-06, "loss": 0.1126, "step": 1905 }, { "epoch": 0.10161539691848377, "grad_norm": 9.84957218170166, "learning_rate": 9.619e-06, "loss": 0.0814, "step": 1906 }, { "epoch": 0.1016687103481367, "grad_norm": 6.303634166717529, "learning_rate": 9.6188e-06, "loss": 0.0766, "step": 1907 }, { "epoch": 0.10172202377778962, "grad_norm": 4.739428997039795, "learning_rate": 9.618600000000001e-06, "loss": 0.0617, "step": 1908 }, { "epoch": 0.10177533720744256, "grad_norm": 8.75710391998291, "learning_rate": 9.618400000000001e-06, "loss": 0.0738, "step": 1909 }, { "epoch": 0.10182865063709548, "grad_norm": 11.325014114379883, "learning_rate": 9.6182e-06, "loss": 0.1289, "step": 1910 }, { "epoch": 0.10188196406674842, "grad_norm": 4.99437141418457, "learning_rate": 9.618e-06, "loss": 0.0794, "step": 1911 }, { "epoch": 0.10193527749640134, "grad_norm": 10.629424095153809, "learning_rate": 9.6178e-06, "loss": 0.063, "step": 1912 }, { "epoch": 0.10198859092605428, "grad_norm": 9.019408226013184, "learning_rate": 9.617600000000001e-06, "loss": 0.0851, "step": 1913 }, { "epoch": 0.1020419043557072, "grad_norm": 6.116448402404785, "learning_rate": 9.6174e-06, "loss": 0.1069, "step": 1914 }, { "epoch": 0.10209521778536013, "grad_norm": 5.853928565979004, "learning_rate": 9.617200000000002e-06, "loss": 0.0681, "step": 1915 }, { "epoch": 0.10214853121501306, "grad_norm": 5.111221790313721, "learning_rate": 9.617000000000001e-06, "loss": 0.0685, "step": 1916 }, { "epoch": 0.10220184464466599, "grad_norm": 5.510978698730469, "learning_rate": 9.6168e-06, "loss": 0.0854, "step": 1917 }, { "epoch": 0.10225515807431892, "grad_norm": 14.177327156066895, "learning_rate": 9.6166e-06, "loss": 0.0959, "step": 1918 }, { "epoch": 0.10230847150397185, "grad_norm": 4.152467727661133, "learning_rate": 9.616400000000002e-06, "loss": 0.0588, "step": 1919 }, { "epoch": 0.10236178493362479, "grad_norm": 7.040280342102051, "learning_rate": 9.616200000000001e-06, "loss": 0.0819, "step": 1920 }, { "epoch": 0.10241509836327771, "grad_norm": 4.786077976226807, "learning_rate": 9.616e-06, "loss": 0.0519, "step": 1921 }, { "epoch": 0.10246841179293063, "grad_norm": 10.382857322692871, "learning_rate": 9.6158e-06, "loss": 0.1213, "step": 1922 }, { "epoch": 0.10252172522258357, "grad_norm": 7.587361812591553, "learning_rate": 9.6156e-06, "loss": 0.0855, "step": 1923 }, { "epoch": 0.1025750386522365, "grad_norm": 4.129538059234619, "learning_rate": 9.6154e-06, "loss": 0.0712, "step": 1924 }, { "epoch": 0.10262835208188943, "grad_norm": 5.416398525238037, "learning_rate": 9.6152e-06, "loss": 0.0595, "step": 1925 }, { "epoch": 0.10268166551154236, "grad_norm": 4.951369285583496, "learning_rate": 9.615e-06, "loss": 0.0788, "step": 1926 }, { "epoch": 0.10273497894119529, "grad_norm": 12.848220825195312, "learning_rate": 9.614800000000002e-06, "loss": 0.1058, "step": 1927 }, { "epoch": 0.10278829237084822, "grad_norm": 5.4425482749938965, "learning_rate": 9.614600000000001e-06, "loss": 0.077, "step": 1928 }, { "epoch": 0.10284160580050114, "grad_norm": 4.078721523284912, "learning_rate": 9.6144e-06, "loss": 0.0742, "step": 1929 }, { "epoch": 0.10289491923015408, "grad_norm": 9.6710786819458, "learning_rate": 9.614200000000002e-06, "loss": 0.0782, "step": 1930 }, { "epoch": 0.102948232659807, "grad_norm": 8.149613380432129, "learning_rate": 9.614000000000001e-06, "loss": 0.0759, "step": 1931 }, { "epoch": 0.10300154608945994, "grad_norm": 6.324500560760498, "learning_rate": 9.613800000000001e-06, "loss": 0.0993, "step": 1932 }, { "epoch": 0.10305485951911286, "grad_norm": 7.041213512420654, "learning_rate": 9.6136e-06, "loss": 0.0809, "step": 1933 }, { "epoch": 0.1031081729487658, "grad_norm": 4.720857620239258, "learning_rate": 9.6134e-06, "loss": 0.0605, "step": 1934 }, { "epoch": 0.10316148637841872, "grad_norm": 3.65556001663208, "learning_rate": 9.6132e-06, "loss": 0.028, "step": 1935 }, { "epoch": 0.10321479980807165, "grad_norm": 10.380606651306152, "learning_rate": 9.613000000000001e-06, "loss": 0.1213, "step": 1936 }, { "epoch": 0.10326811323772458, "grad_norm": 8.033411979675293, "learning_rate": 9.6128e-06, "loss": 0.1048, "step": 1937 }, { "epoch": 0.10332142666737751, "grad_norm": 5.867807865142822, "learning_rate": 9.6126e-06, "loss": 0.0839, "step": 1938 }, { "epoch": 0.10337474009703045, "grad_norm": 5.2310028076171875, "learning_rate": 9.612400000000001e-06, "loss": 0.0793, "step": 1939 }, { "epoch": 0.10342805352668337, "grad_norm": 13.530454635620117, "learning_rate": 9.612200000000001e-06, "loss": 0.1418, "step": 1940 }, { "epoch": 0.10348136695633631, "grad_norm": 10.0425386428833, "learning_rate": 9.612000000000002e-06, "loss": 0.0916, "step": 1941 }, { "epoch": 0.10353468038598923, "grad_norm": 6.940279006958008, "learning_rate": 9.611800000000002e-06, "loss": 0.1078, "step": 1942 }, { "epoch": 0.10358799381564215, "grad_norm": 6.962823867797852, "learning_rate": 9.611600000000001e-06, "loss": 0.0816, "step": 1943 }, { "epoch": 0.10364130724529509, "grad_norm": 6.059780120849609, "learning_rate": 9.6114e-06, "loss": 0.1073, "step": 1944 }, { "epoch": 0.10369462067494802, "grad_norm": 6.174068927764893, "learning_rate": 9.6112e-06, "loss": 0.0835, "step": 1945 }, { "epoch": 0.10374793410460095, "grad_norm": 8.655949592590332, "learning_rate": 9.611e-06, "loss": 0.1122, "step": 1946 }, { "epoch": 0.10380124753425388, "grad_norm": 4.746423721313477, "learning_rate": 9.610800000000001e-06, "loss": 0.0716, "step": 1947 }, { "epoch": 0.10385456096390681, "grad_norm": 4.506649017333984, "learning_rate": 9.6106e-06, "loss": 0.0786, "step": 1948 }, { "epoch": 0.10390787439355974, "grad_norm": 7.469048976898193, "learning_rate": 9.6104e-06, "loss": 0.0935, "step": 1949 }, { "epoch": 0.10396118782321266, "grad_norm": 5.085237979888916, "learning_rate": 9.6102e-06, "loss": 0.0806, "step": 1950 }, { "epoch": 0.1040145012528656, "grad_norm": 5.914918899536133, "learning_rate": 9.610000000000001e-06, "loss": 0.0715, "step": 1951 }, { "epoch": 0.10406781468251852, "grad_norm": 5.06016731262207, "learning_rate": 9.6098e-06, "loss": 0.0719, "step": 1952 }, { "epoch": 0.10412112811217146, "grad_norm": 11.207536697387695, "learning_rate": 9.609600000000002e-06, "loss": 0.117, "step": 1953 }, { "epoch": 0.10417444154182438, "grad_norm": 9.110957145690918, "learning_rate": 9.609400000000002e-06, "loss": 0.1246, "step": 1954 }, { "epoch": 0.10422775497147732, "grad_norm": 5.043302536010742, "learning_rate": 9.609200000000001e-06, "loss": 0.057, "step": 1955 }, { "epoch": 0.10428106840113024, "grad_norm": 3.6694087982177734, "learning_rate": 9.609e-06, "loss": 0.0631, "step": 1956 }, { "epoch": 0.10433438183078317, "grad_norm": 6.1675872802734375, "learning_rate": 9.6088e-06, "loss": 0.0784, "step": 1957 }, { "epoch": 0.1043876952604361, "grad_norm": 6.214999198913574, "learning_rate": 9.608600000000001e-06, "loss": 0.1029, "step": 1958 }, { "epoch": 0.10444100869008903, "grad_norm": 14.500714302062988, "learning_rate": 9.608400000000001e-06, "loss": 0.0969, "step": 1959 }, { "epoch": 0.10449432211974197, "grad_norm": 9.232666015625, "learning_rate": 9.6082e-06, "loss": 0.111, "step": 1960 }, { "epoch": 0.10454763554939489, "grad_norm": 6.9177093505859375, "learning_rate": 9.608e-06, "loss": 0.1185, "step": 1961 }, { "epoch": 0.10460094897904783, "grad_norm": 10.302338600158691, "learning_rate": 9.6078e-06, "loss": 0.1352, "step": 1962 }, { "epoch": 0.10465426240870075, "grad_norm": 5.678474426269531, "learning_rate": 9.607600000000001e-06, "loss": 0.074, "step": 1963 }, { "epoch": 0.10470757583835368, "grad_norm": 3.949104070663452, "learning_rate": 9.6074e-06, "loss": 0.0616, "step": 1964 }, { "epoch": 0.10476088926800661, "grad_norm": 9.405009269714355, "learning_rate": 9.607200000000002e-06, "loss": 0.105, "step": 1965 }, { "epoch": 0.10481420269765954, "grad_norm": 10.888212203979492, "learning_rate": 9.607000000000001e-06, "loss": 0.0764, "step": 1966 }, { "epoch": 0.10486751612731247, "grad_norm": 4.70135498046875, "learning_rate": 9.606800000000001e-06, "loss": 0.0948, "step": 1967 }, { "epoch": 0.1049208295569654, "grad_norm": 4.188379287719727, "learning_rate": 9.6066e-06, "loss": 0.0599, "step": 1968 }, { "epoch": 0.10497414298661834, "grad_norm": 7.5144758224487305, "learning_rate": 9.606400000000002e-06, "loss": 0.0721, "step": 1969 }, { "epoch": 0.10502745641627126, "grad_norm": 5.0025200843811035, "learning_rate": 9.606200000000001e-06, "loss": 0.0777, "step": 1970 }, { "epoch": 0.10508076984592418, "grad_norm": 4.9312968254089355, "learning_rate": 9.606000000000001e-06, "loss": 0.0753, "step": 1971 }, { "epoch": 0.10513408327557712, "grad_norm": 4.198721408843994, "learning_rate": 9.6058e-06, "loss": 0.0907, "step": 1972 }, { "epoch": 0.10518739670523004, "grad_norm": 9.901311874389648, "learning_rate": 9.6056e-06, "loss": 0.0722, "step": 1973 }, { "epoch": 0.10524071013488298, "grad_norm": 3.4258604049682617, "learning_rate": 9.6054e-06, "loss": 0.0582, "step": 1974 }, { "epoch": 0.1052940235645359, "grad_norm": 4.246060371398926, "learning_rate": 9.6052e-06, "loss": 0.0617, "step": 1975 }, { "epoch": 0.10534733699418884, "grad_norm": 6.701199054718018, "learning_rate": 9.605e-06, "loss": 0.0905, "step": 1976 }, { "epoch": 0.10540065042384177, "grad_norm": 6.316847324371338, "learning_rate": 9.604800000000002e-06, "loss": 0.1158, "step": 1977 }, { "epoch": 0.10545396385349469, "grad_norm": 9.980586051940918, "learning_rate": 9.604600000000001e-06, "loss": 0.0879, "step": 1978 }, { "epoch": 0.10550727728314763, "grad_norm": 6.536570072174072, "learning_rate": 9.6044e-06, "loss": 0.0604, "step": 1979 }, { "epoch": 0.10556059071280055, "grad_norm": 4.728451251983643, "learning_rate": 9.604200000000002e-06, "loss": 0.0663, "step": 1980 }, { "epoch": 0.10561390414245349, "grad_norm": 12.387752532958984, "learning_rate": 9.604000000000002e-06, "loss": 0.0759, "step": 1981 }, { "epoch": 0.10566721757210641, "grad_norm": 5.977807521820068, "learning_rate": 9.603800000000001e-06, "loss": 0.0556, "step": 1982 }, { "epoch": 0.10572053100175935, "grad_norm": 13.923665046691895, "learning_rate": 9.6036e-06, "loss": 0.072, "step": 1983 }, { "epoch": 0.10577384443141227, "grad_norm": 12.990683555603027, "learning_rate": 9.6034e-06, "loss": 0.1024, "step": 1984 }, { "epoch": 0.1058271578610652, "grad_norm": 8.14808464050293, "learning_rate": 9.6032e-06, "loss": 0.0936, "step": 1985 }, { "epoch": 0.10588047129071813, "grad_norm": 5.361259937286377, "learning_rate": 9.603000000000001e-06, "loss": 0.068, "step": 1986 }, { "epoch": 0.10593378472037106, "grad_norm": 6.402388572692871, "learning_rate": 9.6028e-06, "loss": 0.0854, "step": 1987 }, { "epoch": 0.105987098150024, "grad_norm": 8.726178169250488, "learning_rate": 9.6026e-06, "loss": 0.0814, "step": 1988 }, { "epoch": 0.10604041157967692, "grad_norm": 7.307845592498779, "learning_rate": 9.602400000000001e-06, "loss": 0.1103, "step": 1989 }, { "epoch": 0.10609372500932986, "grad_norm": 11.759932518005371, "learning_rate": 9.602200000000001e-06, "loss": 0.1, "step": 1990 }, { "epoch": 0.10614703843898278, "grad_norm": 5.7840166091918945, "learning_rate": 9.602e-06, "loss": 0.0669, "step": 1991 }, { "epoch": 0.1062003518686357, "grad_norm": 7.836947917938232, "learning_rate": 9.601800000000002e-06, "loss": 0.0804, "step": 1992 }, { "epoch": 0.10625366529828864, "grad_norm": 6.8966546058654785, "learning_rate": 9.601600000000001e-06, "loss": 0.0943, "step": 1993 }, { "epoch": 0.10630697872794156, "grad_norm": 8.1943359375, "learning_rate": 9.601400000000001e-06, "loss": 0.1335, "step": 1994 }, { "epoch": 0.1063602921575945, "grad_norm": 16.024578094482422, "learning_rate": 9.6012e-06, "loss": 0.1396, "step": 1995 }, { "epoch": 0.10641360558724743, "grad_norm": 3.9471755027770996, "learning_rate": 9.601e-06, "loss": 0.0594, "step": 1996 }, { "epoch": 0.10646691901690036, "grad_norm": 5.47201681137085, "learning_rate": 9.600800000000001e-06, "loss": 0.1172, "step": 1997 }, { "epoch": 0.10652023244655329, "grad_norm": 7.324556827545166, "learning_rate": 9.600600000000001e-06, "loss": 0.1004, "step": 1998 }, { "epoch": 0.10657354587620621, "grad_norm": 5.734264373779297, "learning_rate": 9.6004e-06, "loss": 0.0938, "step": 1999 }, { "epoch": 0.10662685930585915, "grad_norm": 6.0439324378967285, "learning_rate": 9.6002e-06, "loss": 0.0796, "step": 2000 }, { "epoch": 0.10668017273551207, "grad_norm": 12.011953353881836, "learning_rate": 9.600000000000001e-06, "loss": 0.1028, "step": 2001 }, { "epoch": 0.10673348616516501, "grad_norm": 5.390696048736572, "learning_rate": 9.5998e-06, "loss": 0.0426, "step": 2002 }, { "epoch": 0.10678679959481793, "grad_norm": 11.212321281433105, "learning_rate": 9.5996e-06, "loss": 0.1162, "step": 2003 }, { "epoch": 0.10684011302447087, "grad_norm": 5.481504917144775, "learning_rate": 9.599400000000002e-06, "loss": 0.0933, "step": 2004 }, { "epoch": 0.1068934264541238, "grad_norm": 8.598372459411621, "learning_rate": 9.599200000000001e-06, "loss": 0.1234, "step": 2005 }, { "epoch": 0.10694673988377672, "grad_norm": 9.973339080810547, "learning_rate": 9.599e-06, "loss": 0.1189, "step": 2006 }, { "epoch": 0.10700005331342965, "grad_norm": 5.032235145568848, "learning_rate": 9.5988e-06, "loss": 0.0323, "step": 2007 }, { "epoch": 0.10705336674308258, "grad_norm": 5.2733378410339355, "learning_rate": 9.598600000000002e-06, "loss": 0.0616, "step": 2008 }, { "epoch": 0.10710668017273552, "grad_norm": 6.520573139190674, "learning_rate": 9.598400000000001e-06, "loss": 0.0815, "step": 2009 }, { "epoch": 0.10715999360238844, "grad_norm": 6.156379699707031, "learning_rate": 9.5982e-06, "loss": 0.0658, "step": 2010 }, { "epoch": 0.10721330703204138, "grad_norm": 20.468364715576172, "learning_rate": 9.598e-06, "loss": 0.1511, "step": 2011 }, { "epoch": 0.1072666204616943, "grad_norm": 5.014009475708008, "learning_rate": 9.5978e-06, "loss": 0.073, "step": 2012 }, { "epoch": 0.10731993389134722, "grad_norm": 4.853428840637207, "learning_rate": 9.597600000000001e-06, "loss": 0.1058, "step": 2013 }, { "epoch": 0.10737324732100016, "grad_norm": 5.878428936004639, "learning_rate": 9.5974e-06, "loss": 0.0672, "step": 2014 }, { "epoch": 0.10742656075065309, "grad_norm": 6.249721050262451, "learning_rate": 9.5972e-06, "loss": 0.0746, "step": 2015 }, { "epoch": 0.10747987418030602, "grad_norm": 15.037679672241211, "learning_rate": 9.597000000000001e-06, "loss": 0.1309, "step": 2016 }, { "epoch": 0.10753318760995895, "grad_norm": 8.743050575256348, "learning_rate": 9.596800000000001e-06, "loss": 0.0879, "step": 2017 }, { "epoch": 0.10758650103961188, "grad_norm": 6.187231540679932, "learning_rate": 9.5966e-06, "loss": 0.0741, "step": 2018 }, { "epoch": 0.10763981446926481, "grad_norm": 5.068152904510498, "learning_rate": 9.596400000000002e-06, "loss": 0.056, "step": 2019 }, { "epoch": 0.10769312789891773, "grad_norm": 5.067633152008057, "learning_rate": 9.596200000000001e-06, "loss": 0.0546, "step": 2020 }, { "epoch": 0.10774644132857067, "grad_norm": 6.4599928855896, "learning_rate": 9.596000000000001e-06, "loss": 0.0899, "step": 2021 }, { "epoch": 0.10779975475822359, "grad_norm": 4.768054008483887, "learning_rate": 9.5958e-06, "loss": 0.0587, "step": 2022 }, { "epoch": 0.10785306818787653, "grad_norm": 11.664144515991211, "learning_rate": 9.5956e-06, "loss": 0.0543, "step": 2023 }, { "epoch": 0.10790638161752945, "grad_norm": 3.74751353263855, "learning_rate": 9.5954e-06, "loss": 0.0711, "step": 2024 }, { "epoch": 0.10795969504718239, "grad_norm": 5.071041107177734, "learning_rate": 9.595200000000001e-06, "loss": 0.0563, "step": 2025 }, { "epoch": 0.10801300847683531, "grad_norm": 9.135416030883789, "learning_rate": 9.595e-06, "loss": 0.094, "step": 2026 }, { "epoch": 0.10806632190648824, "grad_norm": 13.10383415222168, "learning_rate": 9.5948e-06, "loss": 0.1487, "step": 2027 }, { "epoch": 0.10811963533614118, "grad_norm": 5.172210216522217, "learning_rate": 9.594600000000001e-06, "loss": 0.1036, "step": 2028 }, { "epoch": 0.1081729487657941, "grad_norm": 10.908750534057617, "learning_rate": 9.5944e-06, "loss": 0.072, "step": 2029 }, { "epoch": 0.10822626219544704, "grad_norm": 15.025901794433594, "learning_rate": 9.594200000000002e-06, "loss": 0.0973, "step": 2030 }, { "epoch": 0.10827957562509996, "grad_norm": 3.856684923171997, "learning_rate": 9.594000000000002e-06, "loss": 0.0488, "step": 2031 }, { "epoch": 0.1083328890547529, "grad_norm": 5.559016704559326, "learning_rate": 9.593800000000001e-06, "loss": 0.0647, "step": 2032 }, { "epoch": 0.10838620248440582, "grad_norm": 7.2325029373168945, "learning_rate": 9.5936e-06, "loss": 0.0882, "step": 2033 }, { "epoch": 0.10843951591405875, "grad_norm": 6.821028709411621, "learning_rate": 9.5934e-06, "loss": 0.0901, "step": 2034 }, { "epoch": 0.10849282934371168, "grad_norm": 6.706064701080322, "learning_rate": 9.5932e-06, "loss": 0.1093, "step": 2035 }, { "epoch": 0.1085461427733646, "grad_norm": 6.353436470031738, "learning_rate": 9.593000000000001e-06, "loss": 0.1063, "step": 2036 }, { "epoch": 0.10859945620301754, "grad_norm": 4.704007625579834, "learning_rate": 9.5928e-06, "loss": 0.0803, "step": 2037 }, { "epoch": 0.10865276963267047, "grad_norm": 8.130114555358887, "learning_rate": 9.5926e-06, "loss": 0.0734, "step": 2038 }, { "epoch": 0.1087060830623234, "grad_norm": 5.536386966705322, "learning_rate": 9.5924e-06, "loss": 0.0538, "step": 2039 }, { "epoch": 0.10875939649197633, "grad_norm": 6.858710765838623, "learning_rate": 9.592200000000001e-06, "loss": 0.0939, "step": 2040 }, { "epoch": 0.10881270992162925, "grad_norm": 7.9393439292907715, "learning_rate": 9.592e-06, "loss": 0.1036, "step": 2041 }, { "epoch": 0.10886602335128219, "grad_norm": 10.952796936035156, "learning_rate": 9.591800000000002e-06, "loss": 0.1127, "step": 2042 }, { "epoch": 0.10891933678093511, "grad_norm": 5.348071098327637, "learning_rate": 9.591600000000001e-06, "loss": 0.0919, "step": 2043 }, { "epoch": 0.10897265021058805, "grad_norm": 3.715816020965576, "learning_rate": 9.591400000000001e-06, "loss": 0.0506, "step": 2044 }, { "epoch": 0.10902596364024097, "grad_norm": 8.095680236816406, "learning_rate": 9.5912e-06, "loss": 0.1268, "step": 2045 }, { "epoch": 0.10907927706989391, "grad_norm": 3.8046576976776123, "learning_rate": 9.591e-06, "loss": 0.0676, "step": 2046 }, { "epoch": 0.10913259049954684, "grad_norm": 5.4483795166015625, "learning_rate": 9.590800000000001e-06, "loss": 0.0844, "step": 2047 }, { "epoch": 0.10918590392919976, "grad_norm": 5.819815635681152, "learning_rate": 9.590600000000001e-06, "loss": 0.0853, "step": 2048 }, { "epoch": 0.1092392173588527, "grad_norm": 5.263369560241699, "learning_rate": 9.5904e-06, "loss": 0.0611, "step": 2049 }, { "epoch": 0.10929253078850562, "grad_norm": 3.9662795066833496, "learning_rate": 9.5902e-06, "loss": 0.0605, "step": 2050 }, { "epoch": 0.10934584421815856, "grad_norm": 7.034587860107422, "learning_rate": 9.59e-06, "loss": 0.0854, "step": 2051 }, { "epoch": 0.10939915764781148, "grad_norm": 6.105611801147461, "learning_rate": 9.589800000000001e-06, "loss": 0.0647, "step": 2052 }, { "epoch": 0.10945247107746442, "grad_norm": 4.948035717010498, "learning_rate": 9.5896e-06, "loss": 0.0919, "step": 2053 }, { "epoch": 0.10950578450711734, "grad_norm": 2.9768526554107666, "learning_rate": 9.589400000000002e-06, "loss": 0.0446, "step": 2054 }, { "epoch": 0.10955909793677027, "grad_norm": 4.398472785949707, "learning_rate": 9.589200000000001e-06, "loss": 0.0584, "step": 2055 }, { "epoch": 0.1096124113664232, "grad_norm": 7.940127849578857, "learning_rate": 9.589000000000001e-06, "loss": 0.0559, "step": 2056 }, { "epoch": 0.10966572479607613, "grad_norm": 5.255151271820068, "learning_rate": 9.5888e-06, "loss": 0.0699, "step": 2057 }, { "epoch": 0.10971903822572907, "grad_norm": 5.1350321769714355, "learning_rate": 9.588600000000002e-06, "loss": 0.0451, "step": 2058 }, { "epoch": 0.10977235165538199, "grad_norm": 5.498278617858887, "learning_rate": 9.588400000000001e-06, "loss": 0.0722, "step": 2059 }, { "epoch": 0.10982566508503493, "grad_norm": 2.867957592010498, "learning_rate": 9.5882e-06, "loss": 0.0326, "step": 2060 }, { "epoch": 0.10987897851468785, "grad_norm": 2.8157293796539307, "learning_rate": 9.588e-06, "loss": 0.0532, "step": 2061 }, { "epoch": 0.10993229194434077, "grad_norm": 5.588266849517822, "learning_rate": 9.5878e-06, "loss": 0.0808, "step": 2062 }, { "epoch": 0.10998560537399371, "grad_norm": 4.764220714569092, "learning_rate": 9.5876e-06, "loss": 0.0689, "step": 2063 }, { "epoch": 0.11003891880364663, "grad_norm": 4.920010089874268, "learning_rate": 9.5874e-06, "loss": 0.0698, "step": 2064 }, { "epoch": 0.11009223223329957, "grad_norm": 6.913272857666016, "learning_rate": 9.5872e-06, "loss": 0.0889, "step": 2065 }, { "epoch": 0.1101455456629525, "grad_norm": 5.593337535858154, "learning_rate": 9.587000000000002e-06, "loss": 0.0781, "step": 2066 }, { "epoch": 0.11019885909260543, "grad_norm": 6.227756977081299, "learning_rate": 9.586800000000001e-06, "loss": 0.1058, "step": 2067 }, { "epoch": 0.11025217252225836, "grad_norm": 5.88455867767334, "learning_rate": 9.5866e-06, "loss": 0.0383, "step": 2068 }, { "epoch": 0.11030548595191128, "grad_norm": 6.869051933288574, "learning_rate": 9.586400000000002e-06, "loss": 0.0854, "step": 2069 }, { "epoch": 0.11035879938156422, "grad_norm": 10.45270824432373, "learning_rate": 9.586200000000001e-06, "loss": 0.059, "step": 2070 }, { "epoch": 0.11041211281121714, "grad_norm": 7.2219014167785645, "learning_rate": 9.586000000000001e-06, "loss": 0.0803, "step": 2071 }, { "epoch": 0.11046542624087008, "grad_norm": 6.602633953094482, "learning_rate": 9.5858e-06, "loss": 0.0778, "step": 2072 }, { "epoch": 0.110518739670523, "grad_norm": 4.6374430656433105, "learning_rate": 9.5856e-06, "loss": 0.0332, "step": 2073 }, { "epoch": 0.11057205310017594, "grad_norm": 7.421484470367432, "learning_rate": 9.5854e-06, "loss": 0.0973, "step": 2074 }, { "epoch": 0.11062536652982886, "grad_norm": 3.696913957595825, "learning_rate": 9.585200000000001e-06, "loss": 0.022, "step": 2075 }, { "epoch": 0.11067867995948179, "grad_norm": 6.3135247230529785, "learning_rate": 9.585e-06, "loss": 0.0707, "step": 2076 }, { "epoch": 0.11073199338913473, "grad_norm": 6.495188236236572, "learning_rate": 9.5848e-06, "loss": 0.0993, "step": 2077 }, { "epoch": 0.11078530681878765, "grad_norm": 7.676464557647705, "learning_rate": 9.584600000000001e-06, "loss": 0.0722, "step": 2078 }, { "epoch": 0.11083862024844059, "grad_norm": 7.305606365203857, "learning_rate": 9.584400000000001e-06, "loss": 0.0703, "step": 2079 }, { "epoch": 0.11089193367809351, "grad_norm": 9.883872985839844, "learning_rate": 9.5842e-06, "loss": 0.0516, "step": 2080 }, { "epoch": 0.11094524710774645, "grad_norm": 10.722813606262207, "learning_rate": 9.584000000000002e-06, "loss": 0.0574, "step": 2081 }, { "epoch": 0.11099856053739937, "grad_norm": 5.22703218460083, "learning_rate": 9.583800000000001e-06, "loss": 0.0706, "step": 2082 }, { "epoch": 0.1110518739670523, "grad_norm": 5.27661657333374, "learning_rate": 9.583600000000001e-06, "loss": 0.0695, "step": 2083 }, { "epoch": 0.11110518739670523, "grad_norm": 4.728772163391113, "learning_rate": 9.5834e-06, "loss": 0.0538, "step": 2084 }, { "epoch": 0.11115850082635816, "grad_norm": 5.959503650665283, "learning_rate": 9.5832e-06, "loss": 0.0658, "step": 2085 }, { "epoch": 0.1112118142560111, "grad_norm": 4.868648529052734, "learning_rate": 9.583000000000001e-06, "loss": 0.0715, "step": 2086 }, { "epoch": 0.11126512768566402, "grad_norm": 6.173496723175049, "learning_rate": 9.5828e-06, "loss": 0.0697, "step": 2087 }, { "epoch": 0.11131844111531695, "grad_norm": 6.332114219665527, "learning_rate": 9.5826e-06, "loss": 0.0849, "step": 2088 }, { "epoch": 0.11137175454496988, "grad_norm": 6.71475887298584, "learning_rate": 9.5824e-06, "loss": 0.0758, "step": 2089 }, { "epoch": 0.1114250679746228, "grad_norm": 8.527728080749512, "learning_rate": 9.582200000000001e-06, "loss": 0.0846, "step": 2090 }, { "epoch": 0.11147838140427574, "grad_norm": 6.050198554992676, "learning_rate": 9.582e-06, "loss": 0.0877, "step": 2091 }, { "epoch": 0.11153169483392866, "grad_norm": 4.331725120544434, "learning_rate": 9.5818e-06, "loss": 0.0532, "step": 2092 }, { "epoch": 0.1115850082635816, "grad_norm": 5.456384181976318, "learning_rate": 9.581600000000002e-06, "loss": 0.0855, "step": 2093 }, { "epoch": 0.11163832169323452, "grad_norm": 6.987585544586182, "learning_rate": 9.581400000000001e-06, "loss": 0.0861, "step": 2094 }, { "epoch": 0.11169163512288746, "grad_norm": 5.117955684661865, "learning_rate": 9.5812e-06, "loss": 0.09, "step": 2095 }, { "epoch": 0.11174494855254039, "grad_norm": 7.930166721343994, "learning_rate": 9.581e-06, "loss": 0.0993, "step": 2096 }, { "epoch": 0.11179826198219331, "grad_norm": 5.326883792877197, "learning_rate": 9.580800000000002e-06, "loss": 0.0491, "step": 2097 }, { "epoch": 0.11185157541184625, "grad_norm": 6.086287498474121, "learning_rate": 9.580600000000001e-06, "loss": 0.1111, "step": 2098 }, { "epoch": 0.11190488884149917, "grad_norm": 5.238565444946289, "learning_rate": 9.5804e-06, "loss": 0.0924, "step": 2099 }, { "epoch": 0.11195820227115211, "grad_norm": 4.838528633117676, "learning_rate": 9.5802e-06, "loss": 0.0557, "step": 2100 }, { "epoch": 0.11201151570080503, "grad_norm": 3.204996109008789, "learning_rate": 9.58e-06, "loss": 0.0389, "step": 2101 }, { "epoch": 0.11206482913045797, "grad_norm": 4.633418560028076, "learning_rate": 9.579800000000001e-06, "loss": 0.052, "step": 2102 }, { "epoch": 0.11211814256011089, "grad_norm": 5.745978355407715, "learning_rate": 9.5796e-06, "loss": 0.0737, "step": 2103 }, { "epoch": 0.11217145598976382, "grad_norm": 6.102120399475098, "learning_rate": 9.5794e-06, "loss": 0.0431, "step": 2104 }, { "epoch": 0.11222476941941675, "grad_norm": 5.8819580078125, "learning_rate": 9.579200000000001e-06, "loss": 0.047, "step": 2105 }, { "epoch": 0.11227808284906968, "grad_norm": 4.774788856506348, "learning_rate": 9.579000000000001e-06, "loss": 0.0567, "step": 2106 }, { "epoch": 0.11233139627872261, "grad_norm": 3.489744186401367, "learning_rate": 9.5788e-06, "loss": 0.0554, "step": 2107 }, { "epoch": 0.11238470970837554, "grad_norm": 6.175748348236084, "learning_rate": 9.578600000000002e-06, "loss": 0.0517, "step": 2108 }, { "epoch": 0.11243802313802848, "grad_norm": 10.720958709716797, "learning_rate": 9.578400000000001e-06, "loss": 0.1108, "step": 2109 }, { "epoch": 0.1124913365676814, "grad_norm": 4.260913372039795, "learning_rate": 9.578200000000001e-06, "loss": 0.0649, "step": 2110 }, { "epoch": 0.11254464999733432, "grad_norm": 6.448254108428955, "learning_rate": 9.578e-06, "loss": 0.0789, "step": 2111 }, { "epoch": 0.11259796342698726, "grad_norm": 6.5243000984191895, "learning_rate": 9.5778e-06, "loss": 0.0584, "step": 2112 }, { "epoch": 0.11265127685664018, "grad_norm": 13.441194534301758, "learning_rate": 9.5776e-06, "loss": 0.1076, "step": 2113 }, { "epoch": 0.11270459028629312, "grad_norm": 10.32314682006836, "learning_rate": 9.5774e-06, "loss": 0.1078, "step": 2114 }, { "epoch": 0.11275790371594605, "grad_norm": 6.544495105743408, "learning_rate": 9.5772e-06, "loss": 0.0488, "step": 2115 }, { "epoch": 0.11281121714559898, "grad_norm": 10.671783447265625, "learning_rate": 9.577e-06, "loss": 0.1212, "step": 2116 }, { "epoch": 0.1128645305752519, "grad_norm": 6.497642517089844, "learning_rate": 9.576800000000001e-06, "loss": 0.0588, "step": 2117 }, { "epoch": 0.11291784400490483, "grad_norm": 9.300955772399902, "learning_rate": 9.5766e-06, "loss": 0.1588, "step": 2118 }, { "epoch": 0.11297115743455777, "grad_norm": 8.410069465637207, "learning_rate": 9.576400000000002e-06, "loss": 0.1023, "step": 2119 }, { "epoch": 0.11302447086421069, "grad_norm": 8.310739517211914, "learning_rate": 9.576200000000002e-06, "loss": 0.0909, "step": 2120 }, { "epoch": 0.11307778429386363, "grad_norm": 5.216659069061279, "learning_rate": 9.576000000000001e-06, "loss": 0.0444, "step": 2121 }, { "epoch": 0.11313109772351655, "grad_norm": 7.819812774658203, "learning_rate": 9.5758e-06, "loss": 0.0738, "step": 2122 }, { "epoch": 0.11318441115316949, "grad_norm": 16.377582550048828, "learning_rate": 9.5756e-06, "loss": 0.1031, "step": 2123 }, { "epoch": 0.11323772458282241, "grad_norm": 4.957370281219482, "learning_rate": 9.5754e-06, "loss": 0.0748, "step": 2124 }, { "epoch": 0.11329103801247534, "grad_norm": 5.070677280426025, "learning_rate": 9.575200000000001e-06, "loss": 0.105, "step": 2125 }, { "epoch": 0.11334435144212827, "grad_norm": 4.124678611755371, "learning_rate": 9.575e-06, "loss": 0.0497, "step": 2126 }, { "epoch": 0.1133976648717812, "grad_norm": 10.219698905944824, "learning_rate": 9.5748e-06, "loss": 0.1051, "step": 2127 }, { "epoch": 0.11345097830143414, "grad_norm": 5.301743984222412, "learning_rate": 9.5746e-06, "loss": 0.0594, "step": 2128 }, { "epoch": 0.11350429173108706, "grad_norm": 4.080896377563477, "learning_rate": 9.574400000000001e-06, "loss": 0.0559, "step": 2129 }, { "epoch": 0.11355760516074, "grad_norm": 4.428734302520752, "learning_rate": 9.5742e-06, "loss": 0.056, "step": 2130 }, { "epoch": 0.11361091859039292, "grad_norm": 6.7861785888671875, "learning_rate": 9.574000000000002e-06, "loss": 0.1011, "step": 2131 }, { "epoch": 0.11366423202004584, "grad_norm": 4.986473560333252, "learning_rate": 9.573800000000001e-06, "loss": 0.0659, "step": 2132 }, { "epoch": 0.11371754544969878, "grad_norm": 9.578454971313477, "learning_rate": 9.573600000000001e-06, "loss": 0.0766, "step": 2133 }, { "epoch": 0.1137708588793517, "grad_norm": 5.056352615356445, "learning_rate": 9.5734e-06, "loss": 0.0712, "step": 2134 }, { "epoch": 0.11382417230900464, "grad_norm": 4.253161430358887, "learning_rate": 9.5732e-06, "loss": 0.0684, "step": 2135 }, { "epoch": 0.11387748573865757, "grad_norm": 4.3786540031433105, "learning_rate": 9.573000000000001e-06, "loss": 0.0818, "step": 2136 }, { "epoch": 0.1139307991683105, "grad_norm": 9.569689750671387, "learning_rate": 9.572800000000001e-06, "loss": 0.0964, "step": 2137 }, { "epoch": 0.11398411259796343, "grad_norm": 11.385228157043457, "learning_rate": 9.5726e-06, "loss": 0.0858, "step": 2138 }, { "epoch": 0.11403742602761635, "grad_norm": 4.362950325012207, "learning_rate": 9.5724e-06, "loss": 0.0687, "step": 2139 }, { "epoch": 0.11409073945726929, "grad_norm": 4.8428635597229, "learning_rate": 9.5722e-06, "loss": 0.0869, "step": 2140 }, { "epoch": 0.11414405288692221, "grad_norm": 3.922499179840088, "learning_rate": 9.572000000000001e-06, "loss": 0.063, "step": 2141 }, { "epoch": 0.11419736631657515, "grad_norm": 5.309403896331787, "learning_rate": 9.5718e-06, "loss": 0.0321, "step": 2142 }, { "epoch": 0.11425067974622807, "grad_norm": 10.159601211547852, "learning_rate": 9.571600000000002e-06, "loss": 0.1038, "step": 2143 }, { "epoch": 0.11430399317588101, "grad_norm": 8.947741508483887, "learning_rate": 9.571400000000001e-06, "loss": 0.079, "step": 2144 }, { "epoch": 0.11435730660553393, "grad_norm": 4.359866619110107, "learning_rate": 9.5712e-06, "loss": 0.061, "step": 2145 }, { "epoch": 0.11441062003518686, "grad_norm": 3.788299798965454, "learning_rate": 9.571e-06, "loss": 0.0538, "step": 2146 }, { "epoch": 0.1144639334648398, "grad_norm": 8.201562881469727, "learning_rate": 9.570800000000002e-06, "loss": 0.0877, "step": 2147 }, { "epoch": 0.11451724689449272, "grad_norm": 7.798890113830566, "learning_rate": 9.570600000000001e-06, "loss": 0.0651, "step": 2148 }, { "epoch": 0.11457056032414566, "grad_norm": 5.260587692260742, "learning_rate": 9.5704e-06, "loss": 0.0731, "step": 2149 }, { "epoch": 0.11462387375379858, "grad_norm": 8.28142261505127, "learning_rate": 9.5702e-06, "loss": 0.0932, "step": 2150 }, { "epoch": 0.11467718718345152, "grad_norm": 3.5982120037078857, "learning_rate": 9.57e-06, "loss": 0.0384, "step": 2151 }, { "epoch": 0.11473050061310444, "grad_norm": 6.413087844848633, "learning_rate": 9.569800000000001e-06, "loss": 0.0839, "step": 2152 }, { "epoch": 0.11478381404275736, "grad_norm": 8.650176048278809, "learning_rate": 9.5696e-06, "loss": 0.1513, "step": 2153 }, { "epoch": 0.1148371274724103, "grad_norm": 11.188520431518555, "learning_rate": 9.5694e-06, "loss": 0.0811, "step": 2154 }, { "epoch": 0.11489044090206323, "grad_norm": 6.364480018615723, "learning_rate": 9.569200000000001e-06, "loss": 0.0737, "step": 2155 }, { "epoch": 0.11494375433171616, "grad_norm": 4.278763294219971, "learning_rate": 9.569000000000001e-06, "loss": 0.0382, "step": 2156 }, { "epoch": 0.11499706776136909, "grad_norm": 7.1144256591796875, "learning_rate": 9.5688e-06, "loss": 0.0595, "step": 2157 }, { "epoch": 0.11505038119102202, "grad_norm": 5.8004302978515625, "learning_rate": 9.568600000000002e-06, "loss": 0.0679, "step": 2158 }, { "epoch": 0.11510369462067495, "grad_norm": 6.021104335784912, "learning_rate": 9.568400000000001e-06, "loss": 0.0393, "step": 2159 }, { "epoch": 0.11515700805032787, "grad_norm": 6.685469150543213, "learning_rate": 9.568200000000001e-06, "loss": 0.0576, "step": 2160 }, { "epoch": 0.11521032147998081, "grad_norm": 8.503915786743164, "learning_rate": 9.568e-06, "loss": 0.0793, "step": 2161 }, { "epoch": 0.11526363490963373, "grad_norm": 4.534307956695557, "learning_rate": 9.5678e-06, "loss": 0.0351, "step": 2162 }, { "epoch": 0.11531694833928667, "grad_norm": 5.234809398651123, "learning_rate": 9.567600000000001e-06, "loss": 0.0855, "step": 2163 }, { "epoch": 0.1153702617689396, "grad_norm": 7.63584566116333, "learning_rate": 9.567400000000001e-06, "loss": 0.062, "step": 2164 }, { "epoch": 0.11542357519859253, "grad_norm": 8.882722854614258, "learning_rate": 9.5672e-06, "loss": 0.0568, "step": 2165 }, { "epoch": 0.11547688862824546, "grad_norm": 5.28769063949585, "learning_rate": 9.567e-06, "loss": 0.0725, "step": 2166 }, { "epoch": 0.11553020205789838, "grad_norm": 6.8863749504089355, "learning_rate": 9.566800000000001e-06, "loss": 0.1005, "step": 2167 }, { "epoch": 0.11558351548755132, "grad_norm": 6.666110515594482, "learning_rate": 9.566600000000001e-06, "loss": 0.0926, "step": 2168 }, { "epoch": 0.11563682891720424, "grad_norm": 3.921898603439331, "learning_rate": 9.5664e-06, "loss": 0.0414, "step": 2169 }, { "epoch": 0.11569014234685718, "grad_norm": 9.742759704589844, "learning_rate": 9.566200000000002e-06, "loss": 0.0997, "step": 2170 }, { "epoch": 0.1157434557765101, "grad_norm": 4.409698486328125, "learning_rate": 9.566000000000001e-06, "loss": 0.0687, "step": 2171 }, { "epoch": 0.11579676920616304, "grad_norm": 3.7588188648223877, "learning_rate": 9.5658e-06, "loss": 0.0664, "step": 2172 }, { "epoch": 0.11585008263581596, "grad_norm": 8.273776054382324, "learning_rate": 9.5656e-06, "loss": 0.0668, "step": 2173 }, { "epoch": 0.11590339606546889, "grad_norm": 4.946328639984131, "learning_rate": 9.5654e-06, "loss": 0.1004, "step": 2174 }, { "epoch": 0.11595670949512182, "grad_norm": 10.181373596191406, "learning_rate": 9.565200000000001e-06, "loss": 0.0986, "step": 2175 }, { "epoch": 0.11601002292477475, "grad_norm": 3.6216654777526855, "learning_rate": 9.565e-06, "loss": 0.0382, "step": 2176 }, { "epoch": 0.11606333635442768, "grad_norm": 5.068856716156006, "learning_rate": 9.5648e-06, "loss": 0.0631, "step": 2177 }, { "epoch": 0.11611664978408061, "grad_norm": 7.452508449554443, "learning_rate": 9.5646e-06, "loss": 0.0712, "step": 2178 }, { "epoch": 0.11616996321373355, "grad_norm": 9.375558853149414, "learning_rate": 9.564400000000001e-06, "loss": 0.113, "step": 2179 }, { "epoch": 0.11622327664338647, "grad_norm": 4.757625102996826, "learning_rate": 9.5642e-06, "loss": 0.09, "step": 2180 }, { "epoch": 0.1162765900730394, "grad_norm": 5.08009672164917, "learning_rate": 9.564e-06, "loss": 0.0932, "step": 2181 }, { "epoch": 0.11632990350269233, "grad_norm": 6.355001926422119, "learning_rate": 9.563800000000002e-06, "loss": 0.0771, "step": 2182 }, { "epoch": 0.11638321693234525, "grad_norm": 4.171897888183594, "learning_rate": 9.563600000000001e-06, "loss": 0.056, "step": 2183 }, { "epoch": 0.11643653036199819, "grad_norm": 4.6937432289123535, "learning_rate": 9.5634e-06, "loss": 0.0773, "step": 2184 }, { "epoch": 0.11648984379165112, "grad_norm": 5.597710609436035, "learning_rate": 9.5632e-06, "loss": 0.0499, "step": 2185 }, { "epoch": 0.11654315722130405, "grad_norm": 5.46769905090332, "learning_rate": 9.563000000000001e-06, "loss": 0.0601, "step": 2186 }, { "epoch": 0.11659647065095698, "grad_norm": 7.126342296600342, "learning_rate": 9.562800000000001e-06, "loss": 0.0839, "step": 2187 }, { "epoch": 0.1166497840806099, "grad_norm": 5.549083709716797, "learning_rate": 9.5626e-06, "loss": 0.0785, "step": 2188 }, { "epoch": 0.11670309751026284, "grad_norm": 4.793149948120117, "learning_rate": 9.5624e-06, "loss": 0.07, "step": 2189 }, { "epoch": 0.11675641093991576, "grad_norm": 5.052064895629883, "learning_rate": 9.5622e-06, "loss": 0.0726, "step": 2190 }, { "epoch": 0.1168097243695687, "grad_norm": 7.7711873054504395, "learning_rate": 9.562000000000001e-06, "loss": 0.1184, "step": 2191 }, { "epoch": 0.11686303779922162, "grad_norm": 4.41152286529541, "learning_rate": 9.5618e-06, "loss": 0.0762, "step": 2192 }, { "epoch": 0.11691635122887456, "grad_norm": 6.987242221832275, "learning_rate": 9.5616e-06, "loss": 0.0769, "step": 2193 }, { "epoch": 0.11696966465852748, "grad_norm": 5.776454925537109, "learning_rate": 9.561400000000001e-06, "loss": 0.1274, "step": 2194 }, { "epoch": 0.11702297808818041, "grad_norm": 6.91465950012207, "learning_rate": 9.561200000000001e-06, "loss": 0.0897, "step": 2195 }, { "epoch": 0.11707629151783334, "grad_norm": 5.209548473358154, "learning_rate": 9.561e-06, "loss": 0.0499, "step": 2196 }, { "epoch": 0.11712960494748627, "grad_norm": 5.045783996582031, "learning_rate": 9.560800000000002e-06, "loss": 0.0664, "step": 2197 }, { "epoch": 0.1171829183771392, "grad_norm": 5.407670497894287, "learning_rate": 9.560600000000001e-06, "loss": 0.0847, "step": 2198 }, { "epoch": 0.11723623180679213, "grad_norm": 4.372446537017822, "learning_rate": 9.5604e-06, "loss": 0.066, "step": 2199 }, { "epoch": 0.11728954523644507, "grad_norm": 3.841125726699829, "learning_rate": 9.5602e-06, "loss": 0.0475, "step": 2200 }, { "epoch": 0.11734285866609799, "grad_norm": 3.645167112350464, "learning_rate": 9.56e-06, "loss": 0.0697, "step": 2201 }, { "epoch": 0.11739617209575091, "grad_norm": 5.048239707946777, "learning_rate": 9.559800000000001e-06, "loss": 0.0695, "step": 2202 }, { "epoch": 0.11744948552540385, "grad_norm": 5.630209922790527, "learning_rate": 9.5596e-06, "loss": 0.0765, "step": 2203 }, { "epoch": 0.11750279895505678, "grad_norm": 4.637487411499023, "learning_rate": 9.5594e-06, "loss": 0.0798, "step": 2204 }, { "epoch": 0.11755611238470971, "grad_norm": 3.295311450958252, "learning_rate": 9.5592e-06, "loss": 0.0416, "step": 2205 }, { "epoch": 0.11760942581436264, "grad_norm": 4.7546257972717285, "learning_rate": 9.559000000000001e-06, "loss": 0.0718, "step": 2206 }, { "epoch": 0.11766273924401557, "grad_norm": 5.073644161224365, "learning_rate": 9.5588e-06, "loss": 0.1006, "step": 2207 }, { "epoch": 0.1177160526736685, "grad_norm": 3.7882168292999268, "learning_rate": 9.558600000000002e-06, "loss": 0.0991, "step": 2208 }, { "epoch": 0.11776936610332142, "grad_norm": 4.976717472076416, "learning_rate": 9.558400000000002e-06, "loss": 0.0679, "step": 2209 }, { "epoch": 0.11782267953297436, "grad_norm": 6.0805768966674805, "learning_rate": 9.558200000000001e-06, "loss": 0.0718, "step": 2210 }, { "epoch": 0.11787599296262728, "grad_norm": 3.461153268814087, "learning_rate": 9.558e-06, "loss": 0.0604, "step": 2211 }, { "epoch": 0.11792930639228022, "grad_norm": 4.625609874725342, "learning_rate": 9.5578e-06, "loss": 0.0619, "step": 2212 }, { "epoch": 0.11798261982193314, "grad_norm": 4.558797359466553, "learning_rate": 9.557600000000001e-06, "loss": 0.0559, "step": 2213 }, { "epoch": 0.11803593325158608, "grad_norm": 5.747365474700928, "learning_rate": 9.557400000000001e-06, "loss": 0.0682, "step": 2214 }, { "epoch": 0.118089246681239, "grad_norm": 4.825963020324707, "learning_rate": 9.5572e-06, "loss": 0.0567, "step": 2215 }, { "epoch": 0.11814256011089193, "grad_norm": 4.9642767906188965, "learning_rate": 9.557e-06, "loss": 0.0578, "step": 2216 }, { "epoch": 0.11819587354054487, "grad_norm": 5.062757968902588, "learning_rate": 9.5568e-06, "loss": 0.0495, "step": 2217 }, { "epoch": 0.11824918697019779, "grad_norm": 6.478240013122559, "learning_rate": 9.556600000000001e-06, "loss": 0.0995, "step": 2218 }, { "epoch": 0.11830250039985073, "grad_norm": 5.229343891143799, "learning_rate": 9.5564e-06, "loss": 0.0787, "step": 2219 }, { "epoch": 0.11835581382950365, "grad_norm": 4.123178958892822, "learning_rate": 9.556200000000002e-06, "loss": 0.0619, "step": 2220 }, { "epoch": 0.11840912725915659, "grad_norm": 5.9605326652526855, "learning_rate": 9.556000000000001e-06, "loss": 0.0522, "step": 2221 }, { "epoch": 0.11846244068880951, "grad_norm": 11.281333923339844, "learning_rate": 9.555800000000001e-06, "loss": 0.1269, "step": 2222 }, { "epoch": 0.11851575411846244, "grad_norm": 9.181845664978027, "learning_rate": 9.5556e-06, "loss": 0.0699, "step": 2223 }, { "epoch": 0.11856906754811537, "grad_norm": 7.046482563018799, "learning_rate": 9.5554e-06, "loss": 0.0807, "step": 2224 }, { "epoch": 0.1186223809777683, "grad_norm": 8.416838645935059, "learning_rate": 9.555200000000001e-06, "loss": 0.1249, "step": 2225 }, { "epoch": 0.11867569440742123, "grad_norm": 9.412920951843262, "learning_rate": 9.555e-06, "loss": 0.0919, "step": 2226 }, { "epoch": 0.11872900783707416, "grad_norm": 7.998930931091309, "learning_rate": 9.5548e-06, "loss": 0.0849, "step": 2227 }, { "epoch": 0.1187823212667271, "grad_norm": 6.732646465301514, "learning_rate": 9.5546e-06, "loss": 0.0909, "step": 2228 }, { "epoch": 0.11883563469638002, "grad_norm": 5.392120361328125, "learning_rate": 9.5544e-06, "loss": 0.0712, "step": 2229 }, { "epoch": 0.11888894812603294, "grad_norm": 8.968955039978027, "learning_rate": 9.5542e-06, "loss": 0.1048, "step": 2230 }, { "epoch": 0.11894226155568588, "grad_norm": 7.592615604400635, "learning_rate": 9.554e-06, "loss": 0.0608, "step": 2231 }, { "epoch": 0.1189955749853388, "grad_norm": 6.150416851043701, "learning_rate": 9.553800000000002e-06, "loss": 0.0652, "step": 2232 }, { "epoch": 0.11904888841499174, "grad_norm": 6.792389869689941, "learning_rate": 9.553600000000001e-06, "loss": 0.0662, "step": 2233 }, { "epoch": 0.11910220184464466, "grad_norm": 4.45244836807251, "learning_rate": 9.5534e-06, "loss": 0.0671, "step": 2234 }, { "epoch": 0.1191555152742976, "grad_norm": 5.01438570022583, "learning_rate": 9.5532e-06, "loss": 0.0913, "step": 2235 }, { "epoch": 0.11920882870395053, "grad_norm": 8.163198471069336, "learning_rate": 9.553000000000002e-06, "loss": 0.0777, "step": 2236 }, { "epoch": 0.11926214213360345, "grad_norm": 4.2277913093566895, "learning_rate": 9.552800000000001e-06, "loss": 0.0683, "step": 2237 }, { "epoch": 0.11931545556325639, "grad_norm": 4.3608784675598145, "learning_rate": 9.5526e-06, "loss": 0.0493, "step": 2238 }, { "epoch": 0.11936876899290931, "grad_norm": 4.938601493835449, "learning_rate": 9.5524e-06, "loss": 0.0895, "step": 2239 }, { "epoch": 0.11942208242256225, "grad_norm": 5.931239128112793, "learning_rate": 9.5522e-06, "loss": 0.0895, "step": 2240 }, { "epoch": 0.11947539585221517, "grad_norm": 4.263851165771484, "learning_rate": 9.552000000000001e-06, "loss": 0.0551, "step": 2241 }, { "epoch": 0.11952870928186811, "grad_norm": 6.099824905395508, "learning_rate": 9.5518e-06, "loss": 0.106, "step": 2242 }, { "epoch": 0.11958202271152103, "grad_norm": 4.536717891693115, "learning_rate": 9.5516e-06, "loss": 0.0554, "step": 2243 }, { "epoch": 0.11963533614117396, "grad_norm": 7.50793981552124, "learning_rate": 9.551400000000001e-06, "loss": 0.0893, "step": 2244 }, { "epoch": 0.1196886495708269, "grad_norm": 5.179062843322754, "learning_rate": 9.551200000000001e-06, "loss": 0.0557, "step": 2245 }, { "epoch": 0.11974196300047982, "grad_norm": 5.254506587982178, "learning_rate": 9.551e-06, "loss": 0.0917, "step": 2246 }, { "epoch": 0.11979527643013275, "grad_norm": 6.098150253295898, "learning_rate": 9.550800000000002e-06, "loss": 0.0695, "step": 2247 }, { "epoch": 0.11984858985978568, "grad_norm": 6.661954879760742, "learning_rate": 9.550600000000001e-06, "loss": 0.0886, "step": 2248 }, { "epoch": 0.11990190328943862, "grad_norm": 3.9432685375213623, "learning_rate": 9.550400000000001e-06, "loss": 0.0388, "step": 2249 }, { "epoch": 0.11995521671909154, "grad_norm": 6.163379669189453, "learning_rate": 9.5502e-06, "loss": 0.0751, "step": 2250 }, { "epoch": 0.12000853014874446, "grad_norm": 4.483510494232178, "learning_rate": 9.55e-06, "loss": 0.0728, "step": 2251 }, { "epoch": 0.1200618435783974, "grad_norm": 8.151628494262695, "learning_rate": 9.549800000000001e-06, "loss": 0.0625, "step": 2252 }, { "epoch": 0.12011515700805032, "grad_norm": 7.90641975402832, "learning_rate": 9.549600000000001e-06, "loss": 0.1155, "step": 2253 }, { "epoch": 0.12016847043770326, "grad_norm": 5.255874156951904, "learning_rate": 9.5494e-06, "loss": 0.0815, "step": 2254 }, { "epoch": 0.12022178386735619, "grad_norm": 4.866916179656982, "learning_rate": 9.5492e-06, "loss": 0.0497, "step": 2255 }, { "epoch": 0.12027509729700912, "grad_norm": 4.060431480407715, "learning_rate": 9.549000000000001e-06, "loss": 0.0648, "step": 2256 }, { "epoch": 0.12032841072666205, "grad_norm": 3.758545398712158, "learning_rate": 9.5488e-06, "loss": 0.0383, "step": 2257 }, { "epoch": 0.12038172415631497, "grad_norm": 5.825056552886963, "learning_rate": 9.5486e-06, "loss": 0.082, "step": 2258 }, { "epoch": 0.12043503758596791, "grad_norm": 6.927016735076904, "learning_rate": 9.548400000000002e-06, "loss": 0.0854, "step": 2259 }, { "epoch": 0.12048835101562083, "grad_norm": 4.036416530609131, "learning_rate": 9.548200000000001e-06, "loss": 0.0731, "step": 2260 }, { "epoch": 0.12054166444527377, "grad_norm": 4.934053421020508, "learning_rate": 9.548e-06, "loss": 0.0658, "step": 2261 }, { "epoch": 0.12059497787492669, "grad_norm": 7.838344097137451, "learning_rate": 9.5478e-06, "loss": 0.0481, "step": 2262 }, { "epoch": 0.12064829130457963, "grad_norm": 4.696733474731445, "learning_rate": 9.547600000000002e-06, "loss": 0.0584, "step": 2263 }, { "epoch": 0.12070160473423255, "grad_norm": 9.248741149902344, "learning_rate": 9.547400000000001e-06, "loss": 0.0825, "step": 2264 }, { "epoch": 0.12075491816388548, "grad_norm": 6.446761131286621, "learning_rate": 9.5472e-06, "loss": 0.0586, "step": 2265 }, { "epoch": 0.12080823159353841, "grad_norm": 4.418895244598389, "learning_rate": 9.547e-06, "loss": 0.0633, "step": 2266 }, { "epoch": 0.12086154502319134, "grad_norm": 7.765792369842529, "learning_rate": 9.5468e-06, "loss": 0.0526, "step": 2267 }, { "epoch": 0.12091485845284428, "grad_norm": 15.935559272766113, "learning_rate": 9.546600000000001e-06, "loss": 0.1826, "step": 2268 }, { "epoch": 0.1209681718824972, "grad_norm": 4.734261512756348, "learning_rate": 9.5464e-06, "loss": 0.0686, "step": 2269 }, { "epoch": 0.12102148531215014, "grad_norm": 5.517232418060303, "learning_rate": 9.5462e-06, "loss": 0.0634, "step": 2270 }, { "epoch": 0.12107479874180306, "grad_norm": 17.91840934753418, "learning_rate": 9.546000000000001e-06, "loss": 0.1382, "step": 2271 }, { "epoch": 0.12112811217145598, "grad_norm": 3.596242666244507, "learning_rate": 9.545800000000001e-06, "loss": 0.0613, "step": 2272 }, { "epoch": 0.12118142560110892, "grad_norm": 4.29654598236084, "learning_rate": 9.5456e-06, "loss": 0.0501, "step": 2273 }, { "epoch": 0.12123473903076185, "grad_norm": 5.615097999572754, "learning_rate": 9.545400000000002e-06, "loss": 0.0677, "step": 2274 }, { "epoch": 0.12128805246041478, "grad_norm": 4.616471290588379, "learning_rate": 9.545200000000001e-06, "loss": 0.0713, "step": 2275 }, { "epoch": 0.1213413658900677, "grad_norm": 6.761290550231934, "learning_rate": 9.545000000000001e-06, "loss": 0.0829, "step": 2276 }, { "epoch": 0.12139467931972064, "grad_norm": 10.02217960357666, "learning_rate": 9.5448e-06, "loss": 0.073, "step": 2277 }, { "epoch": 0.12144799274937357, "grad_norm": 5.269980430603027, "learning_rate": 9.5446e-06, "loss": 0.0928, "step": 2278 }, { "epoch": 0.12150130617902649, "grad_norm": 5.127298355102539, "learning_rate": 9.5444e-06, "loss": 0.0519, "step": 2279 }, { "epoch": 0.12155461960867943, "grad_norm": 6.972429275512695, "learning_rate": 9.544200000000001e-06, "loss": 0.1071, "step": 2280 }, { "epoch": 0.12160793303833235, "grad_norm": 11.826943397521973, "learning_rate": 9.544e-06, "loss": 0.117, "step": 2281 }, { "epoch": 0.12166124646798529, "grad_norm": 3.9435791969299316, "learning_rate": 9.5438e-06, "loss": 0.0788, "step": 2282 }, { "epoch": 0.12171455989763821, "grad_norm": 6.285158157348633, "learning_rate": 9.543600000000001e-06, "loss": 0.0653, "step": 2283 }, { "epoch": 0.12176787332729115, "grad_norm": 10.813741683959961, "learning_rate": 9.5434e-06, "loss": 0.0567, "step": 2284 }, { "epoch": 0.12182118675694407, "grad_norm": 7.996917247772217, "learning_rate": 9.5432e-06, "loss": 0.0705, "step": 2285 }, { "epoch": 0.121874500186597, "grad_norm": 5.81641960144043, "learning_rate": 9.543000000000002e-06, "loss": 0.0488, "step": 2286 }, { "epoch": 0.12192781361624994, "grad_norm": 3.6899962425231934, "learning_rate": 9.542800000000001e-06, "loss": 0.0229, "step": 2287 }, { "epoch": 0.12198112704590286, "grad_norm": 7.936707019805908, "learning_rate": 9.5426e-06, "loss": 0.1169, "step": 2288 }, { "epoch": 0.1220344404755558, "grad_norm": 6.3347673416137695, "learning_rate": 9.5424e-06, "loss": 0.066, "step": 2289 }, { "epoch": 0.12208775390520872, "grad_norm": 10.343145370483398, "learning_rate": 9.5422e-06, "loss": 0.1494, "step": 2290 }, { "epoch": 0.12214106733486166, "grad_norm": 17.66998863220215, "learning_rate": 9.542000000000001e-06, "loss": 0.1029, "step": 2291 }, { "epoch": 0.12219438076451458, "grad_norm": 9.303488731384277, "learning_rate": 9.5418e-06, "loss": 0.0614, "step": 2292 }, { "epoch": 0.1222476941941675, "grad_norm": 2.7536933422088623, "learning_rate": 9.5416e-06, "loss": 0.0391, "step": 2293 }, { "epoch": 0.12230100762382044, "grad_norm": 6.692880153656006, "learning_rate": 9.5414e-06, "loss": 0.0609, "step": 2294 }, { "epoch": 0.12235432105347337, "grad_norm": 11.112269401550293, "learning_rate": 9.541200000000001e-06, "loss": 0.1097, "step": 2295 }, { "epoch": 0.1224076344831263, "grad_norm": 5.3651838302612305, "learning_rate": 9.541e-06, "loss": 0.0554, "step": 2296 }, { "epoch": 0.12246094791277923, "grad_norm": 6.546874046325684, "learning_rate": 9.540800000000002e-06, "loss": 0.069, "step": 2297 }, { "epoch": 0.12251426134243217, "grad_norm": 8.917288780212402, "learning_rate": 9.540600000000001e-06, "loss": 0.0772, "step": 2298 }, { "epoch": 0.12256757477208509, "grad_norm": 7.1584367752075195, "learning_rate": 9.540400000000001e-06, "loss": 0.0842, "step": 2299 }, { "epoch": 0.12262088820173801, "grad_norm": 4.060824871063232, "learning_rate": 9.5402e-06, "loss": 0.0675, "step": 2300 }, { "epoch": 0.12267420163139095, "grad_norm": 11.049906730651855, "learning_rate": 9.54e-06, "loss": 0.1264, "step": 2301 }, { "epoch": 0.12272751506104387, "grad_norm": 3.852480411529541, "learning_rate": 9.539800000000001e-06, "loss": 0.0441, "step": 2302 }, { "epoch": 0.12278082849069681, "grad_norm": 5.478687763214111, "learning_rate": 9.539600000000001e-06, "loss": 0.0775, "step": 2303 }, { "epoch": 0.12283414192034973, "grad_norm": 4.106463432312012, "learning_rate": 9.5394e-06, "loss": 0.0614, "step": 2304 }, { "epoch": 0.12288745535000267, "grad_norm": 6.203771114349365, "learning_rate": 9.5392e-06, "loss": 0.068, "step": 2305 }, { "epoch": 0.1229407687796556, "grad_norm": 15.321916580200195, "learning_rate": 9.539e-06, "loss": 0.1138, "step": 2306 }, { "epoch": 0.12299408220930852, "grad_norm": 8.841817855834961, "learning_rate": 9.538800000000001e-06, "loss": 0.0622, "step": 2307 }, { "epoch": 0.12304739563896146, "grad_norm": 10.099228858947754, "learning_rate": 9.5386e-06, "loss": 0.0677, "step": 2308 }, { "epoch": 0.12310070906861438, "grad_norm": 7.2656474113464355, "learning_rate": 9.538400000000002e-06, "loss": 0.0493, "step": 2309 }, { "epoch": 0.12315402249826732, "grad_norm": 6.929571151733398, "learning_rate": 9.538200000000001e-06, "loss": 0.1068, "step": 2310 }, { "epoch": 0.12320733592792024, "grad_norm": 5.59106969833374, "learning_rate": 9.538e-06, "loss": 0.0728, "step": 2311 }, { "epoch": 0.12326064935757318, "grad_norm": 5.3973541259765625, "learning_rate": 9.5378e-06, "loss": 0.0638, "step": 2312 }, { "epoch": 0.1233139627872261, "grad_norm": 9.221360206604004, "learning_rate": 9.537600000000002e-06, "loss": 0.095, "step": 2313 }, { "epoch": 0.12336727621687903, "grad_norm": 16.072126388549805, "learning_rate": 9.537400000000001e-06, "loss": 0.1123, "step": 2314 }, { "epoch": 0.12342058964653196, "grad_norm": 9.58655071258545, "learning_rate": 9.5372e-06, "loss": 0.106, "step": 2315 }, { "epoch": 0.12347390307618489, "grad_norm": 6.166843414306641, "learning_rate": 9.537e-06, "loss": 0.078, "step": 2316 }, { "epoch": 0.12352721650583783, "grad_norm": 8.019460678100586, "learning_rate": 9.5368e-06, "loss": 0.0786, "step": 2317 }, { "epoch": 0.12358052993549075, "grad_norm": 7.170016765594482, "learning_rate": 9.5366e-06, "loss": 0.0631, "step": 2318 }, { "epoch": 0.12363384336514369, "grad_norm": 6.220587253570557, "learning_rate": 9.5364e-06, "loss": 0.0599, "step": 2319 }, { "epoch": 0.12368715679479661, "grad_norm": 4.137763023376465, "learning_rate": 9.5362e-06, "loss": 0.0502, "step": 2320 }, { "epoch": 0.12374047022444953, "grad_norm": 3.509716272354126, "learning_rate": 9.536000000000002e-06, "loss": 0.033, "step": 2321 }, { "epoch": 0.12379378365410247, "grad_norm": 8.407846450805664, "learning_rate": 9.535800000000001e-06, "loss": 0.1142, "step": 2322 }, { "epoch": 0.1238470970837554, "grad_norm": 19.823707580566406, "learning_rate": 9.5356e-06, "loss": 0.1263, "step": 2323 }, { "epoch": 0.12390041051340833, "grad_norm": 4.727784633636475, "learning_rate": 9.535400000000002e-06, "loss": 0.0663, "step": 2324 }, { "epoch": 0.12395372394306126, "grad_norm": 3.649014949798584, "learning_rate": 9.535200000000001e-06, "loss": 0.0565, "step": 2325 }, { "epoch": 0.1240070373727142, "grad_norm": 7.192704200744629, "learning_rate": 9.535000000000001e-06, "loss": 0.1312, "step": 2326 }, { "epoch": 0.12406035080236712, "grad_norm": 19.213220596313477, "learning_rate": 9.5348e-06, "loss": 0.1033, "step": 2327 }, { "epoch": 0.12411366423202004, "grad_norm": 11.668816566467285, "learning_rate": 9.5346e-06, "loss": 0.0821, "step": 2328 }, { "epoch": 0.12416697766167298, "grad_norm": 5.205768585205078, "learning_rate": 9.5344e-06, "loss": 0.1048, "step": 2329 }, { "epoch": 0.1242202910913259, "grad_norm": 4.603944301605225, "learning_rate": 9.534200000000001e-06, "loss": 0.0552, "step": 2330 }, { "epoch": 0.12427360452097884, "grad_norm": 14.216465950012207, "learning_rate": 9.534e-06, "loss": 0.1649, "step": 2331 }, { "epoch": 0.12432691795063176, "grad_norm": 12.625772476196289, "learning_rate": 9.5338e-06, "loss": 0.1107, "step": 2332 }, { "epoch": 0.1243802313802847, "grad_norm": 5.430022239685059, "learning_rate": 9.533600000000001e-06, "loss": 0.0761, "step": 2333 }, { "epoch": 0.12443354480993762, "grad_norm": 7.986271381378174, "learning_rate": 9.533400000000001e-06, "loss": 0.1118, "step": 2334 }, { "epoch": 0.12448685823959055, "grad_norm": 9.119190216064453, "learning_rate": 9.5332e-06, "loss": 0.1034, "step": 2335 }, { "epoch": 0.12454017166924349, "grad_norm": 15.019940376281738, "learning_rate": 9.533000000000002e-06, "loss": 0.1391, "step": 2336 }, { "epoch": 0.12459348509889641, "grad_norm": 7.4068684577941895, "learning_rate": 9.532800000000001e-06, "loss": 0.0781, "step": 2337 }, { "epoch": 0.12464679852854935, "grad_norm": 6.792425632476807, "learning_rate": 9.532600000000001e-06, "loss": 0.1088, "step": 2338 }, { "epoch": 0.12470011195820227, "grad_norm": 6.746087074279785, "learning_rate": 9.5324e-06, "loss": 0.061, "step": 2339 }, { "epoch": 0.12475342538785521, "grad_norm": 6.486972808837891, "learning_rate": 9.5322e-06, "loss": 0.0568, "step": 2340 }, { "epoch": 0.12480673881750813, "grad_norm": 7.7053608894348145, "learning_rate": 9.532000000000001e-06, "loss": 0.0663, "step": 2341 }, { "epoch": 0.12486005224716105, "grad_norm": 5.986505508422852, "learning_rate": 9.5318e-06, "loss": 0.0922, "step": 2342 }, { "epoch": 0.12491336567681399, "grad_norm": 3.279001235961914, "learning_rate": 9.5316e-06, "loss": 0.0332, "step": 2343 }, { "epoch": 0.12496667910646692, "grad_norm": 4.752469062805176, "learning_rate": 9.5314e-06, "loss": 0.0575, "step": 2344 }, { "epoch": 0.12501999253611984, "grad_norm": 6.487397193908691, "learning_rate": 9.531200000000001e-06, "loss": 0.0809, "step": 2345 }, { "epoch": 0.1250733059657728, "grad_norm": 5.370238780975342, "learning_rate": 9.531e-06, "loss": 0.0641, "step": 2346 }, { "epoch": 0.12512661939542571, "grad_norm": 6.539247512817383, "learning_rate": 9.530800000000002e-06, "loss": 0.0694, "step": 2347 }, { "epoch": 0.12517993282507864, "grad_norm": 4.877710819244385, "learning_rate": 9.530600000000002e-06, "loss": 0.0767, "step": 2348 }, { "epoch": 0.12523324625473156, "grad_norm": 8.489631652832031, "learning_rate": 9.530400000000001e-06, "loss": 0.0737, "step": 2349 }, { "epoch": 0.12528655968438449, "grad_norm": 7.548209190368652, "learning_rate": 9.5302e-06, "loss": 0.0721, "step": 2350 }, { "epoch": 0.12533987311403744, "grad_norm": 3.8056440353393555, "learning_rate": 9.53e-06, "loss": 0.0815, "step": 2351 }, { "epoch": 0.12539318654369036, "grad_norm": 4.381932735443115, "learning_rate": 9.529800000000001e-06, "loss": 0.0576, "step": 2352 }, { "epoch": 0.12544649997334328, "grad_norm": 5.073395729064941, "learning_rate": 9.529600000000001e-06, "loss": 0.074, "step": 2353 }, { "epoch": 0.1254998134029962, "grad_norm": 4.727437496185303, "learning_rate": 9.5294e-06, "loss": 0.0888, "step": 2354 }, { "epoch": 0.12555312683264913, "grad_norm": 3.7023069858551025, "learning_rate": 9.5292e-06, "loss": 0.0678, "step": 2355 }, { "epoch": 0.12560644026230208, "grad_norm": 5.0693464279174805, "learning_rate": 9.529e-06, "loss": 0.0639, "step": 2356 }, { "epoch": 0.125659753691955, "grad_norm": 2.75319766998291, "learning_rate": 9.528800000000001e-06, "loss": 0.0473, "step": 2357 }, { "epoch": 0.12571306712160793, "grad_norm": 7.029744625091553, "learning_rate": 9.5286e-06, "loss": 0.0591, "step": 2358 }, { "epoch": 0.12576638055126085, "grad_norm": 3.4667351245880127, "learning_rate": 9.528400000000002e-06, "loss": 0.0529, "step": 2359 }, { "epoch": 0.1258196939809138, "grad_norm": 10.908360481262207, "learning_rate": 9.528200000000001e-06, "loss": 0.0818, "step": 2360 }, { "epoch": 0.12587300741056673, "grad_norm": 6.525012493133545, "learning_rate": 9.528000000000001e-06, "loss": 0.1288, "step": 2361 }, { "epoch": 0.12592632084021965, "grad_norm": 6.522982597351074, "learning_rate": 9.5278e-06, "loss": 0.0576, "step": 2362 }, { "epoch": 0.12597963426987258, "grad_norm": 4.063587188720703, "learning_rate": 9.527600000000002e-06, "loss": 0.0769, "step": 2363 }, { "epoch": 0.1260329476995255, "grad_norm": 8.583587646484375, "learning_rate": 9.527400000000001e-06, "loss": 0.1073, "step": 2364 }, { "epoch": 0.12608626112917845, "grad_norm": 4.822509288787842, "learning_rate": 9.527200000000001e-06, "loss": 0.0619, "step": 2365 }, { "epoch": 0.12613957455883137, "grad_norm": 5.286661148071289, "learning_rate": 9.527e-06, "loss": 0.065, "step": 2366 }, { "epoch": 0.1261928879884843, "grad_norm": 4.8939666748046875, "learning_rate": 9.5268e-06, "loss": 0.0631, "step": 2367 }, { "epoch": 0.12624620141813722, "grad_norm": 7.293911933898926, "learning_rate": 9.5266e-06, "loss": 0.0622, "step": 2368 }, { "epoch": 0.12629951484779015, "grad_norm": 9.448556900024414, "learning_rate": 9.5264e-06, "loss": 0.0992, "step": 2369 }, { "epoch": 0.1263528282774431, "grad_norm": 4.014091491699219, "learning_rate": 9.5262e-06, "loss": 0.0557, "step": 2370 }, { "epoch": 0.12640614170709602, "grad_norm": 4.498177528381348, "learning_rate": 9.526000000000002e-06, "loss": 0.0491, "step": 2371 }, { "epoch": 0.12645945513674894, "grad_norm": 4.975451469421387, "learning_rate": 9.525800000000001e-06, "loss": 0.0993, "step": 2372 }, { "epoch": 0.12651276856640187, "grad_norm": 6.689369201660156, "learning_rate": 9.5256e-06, "loss": 0.0805, "step": 2373 }, { "epoch": 0.12656608199605482, "grad_norm": 8.041094779968262, "learning_rate": 9.525400000000002e-06, "loss": 0.0942, "step": 2374 }, { "epoch": 0.12661939542570774, "grad_norm": 4.325109481811523, "learning_rate": 9.525200000000002e-06, "loss": 0.0526, "step": 2375 }, { "epoch": 0.12667270885536067, "grad_norm": 4.010102272033691, "learning_rate": 9.525000000000001e-06, "loss": 0.0635, "step": 2376 }, { "epoch": 0.1267260222850136, "grad_norm": 6.032255172729492, "learning_rate": 9.5248e-06, "loss": 0.0644, "step": 2377 }, { "epoch": 0.1267793357146665, "grad_norm": 4.6433258056640625, "learning_rate": 9.5246e-06, "loss": 0.0677, "step": 2378 }, { "epoch": 0.12683264914431946, "grad_norm": 7.466030120849609, "learning_rate": 9.5244e-06, "loss": 0.0886, "step": 2379 }, { "epoch": 0.1268859625739724, "grad_norm": 4.798100471496582, "learning_rate": 9.524200000000001e-06, "loss": 0.0728, "step": 2380 }, { "epoch": 0.1269392760036253, "grad_norm": 7.991889476776123, "learning_rate": 9.524e-06, "loss": 0.0733, "step": 2381 }, { "epoch": 0.12699258943327824, "grad_norm": 9.090821266174316, "learning_rate": 9.5238e-06, "loss": 0.0782, "step": 2382 }, { "epoch": 0.12704590286293116, "grad_norm": 6.8022780418396, "learning_rate": 9.523600000000001e-06, "loss": 0.0771, "step": 2383 }, { "epoch": 0.1270992162925841, "grad_norm": 3.645318031311035, "learning_rate": 9.523400000000001e-06, "loss": 0.0802, "step": 2384 }, { "epoch": 0.12715252972223703, "grad_norm": 5.298220157623291, "learning_rate": 9.5232e-06, "loss": 0.0691, "step": 2385 }, { "epoch": 0.12720584315188996, "grad_norm": 6.0981340408325195, "learning_rate": 9.523000000000002e-06, "loss": 0.1022, "step": 2386 }, { "epoch": 0.12725915658154288, "grad_norm": 10.63562297821045, "learning_rate": 9.522800000000001e-06, "loss": 0.1439, "step": 2387 }, { "epoch": 0.12731247001119583, "grad_norm": 11.030632019042969, "learning_rate": 9.522600000000001e-06, "loss": 0.1248, "step": 2388 }, { "epoch": 0.12736578344084876, "grad_norm": 4.644731044769287, "learning_rate": 9.5224e-06, "loss": 0.0597, "step": 2389 }, { "epoch": 0.12741909687050168, "grad_norm": 5.078250885009766, "learning_rate": 9.5222e-06, "loss": 0.0705, "step": 2390 }, { "epoch": 0.1274724103001546, "grad_norm": 11.138772010803223, "learning_rate": 9.522000000000001e-06, "loss": 0.1259, "step": 2391 }, { "epoch": 0.12752572372980753, "grad_norm": 11.215058326721191, "learning_rate": 9.521800000000001e-06, "loss": 0.1101, "step": 2392 }, { "epoch": 0.12757903715946048, "grad_norm": 11.248831748962402, "learning_rate": 9.5216e-06, "loss": 0.0895, "step": 2393 }, { "epoch": 0.1276323505891134, "grad_norm": 4.39060640335083, "learning_rate": 9.5214e-06, "loss": 0.0493, "step": 2394 }, { "epoch": 0.12768566401876633, "grad_norm": 7.655284881591797, "learning_rate": 9.521200000000001e-06, "loss": 0.0982, "step": 2395 }, { "epoch": 0.12773897744841925, "grad_norm": 12.843854904174805, "learning_rate": 9.521e-06, "loss": 0.0992, "step": 2396 }, { "epoch": 0.12779229087807217, "grad_norm": 9.047322273254395, "learning_rate": 9.5208e-06, "loss": 0.0753, "step": 2397 }, { "epoch": 0.12784560430772512, "grad_norm": 5.689648151397705, "learning_rate": 9.520600000000002e-06, "loss": 0.0782, "step": 2398 }, { "epoch": 0.12789891773737805, "grad_norm": 6.521282196044922, "learning_rate": 9.520400000000001e-06, "loss": 0.1065, "step": 2399 }, { "epoch": 0.12795223116703097, "grad_norm": 6.161636829376221, "learning_rate": 9.5202e-06, "loss": 0.0871, "step": 2400 }, { "epoch": 0.1280055445966839, "grad_norm": 11.152929306030273, "learning_rate": 9.52e-06, "loss": 0.0783, "step": 2401 }, { "epoch": 0.12805885802633685, "grad_norm": 14.912379264831543, "learning_rate": 9.519800000000002e-06, "loss": 0.1086, "step": 2402 }, { "epoch": 0.12811217145598977, "grad_norm": 8.11755084991455, "learning_rate": 9.519600000000001e-06, "loss": 0.0924, "step": 2403 }, { "epoch": 0.1281654848856427, "grad_norm": 2.6386759281158447, "learning_rate": 9.5194e-06, "loss": 0.0466, "step": 2404 }, { "epoch": 0.12821879831529562, "grad_norm": 6.97644567489624, "learning_rate": 9.5192e-06, "loss": 0.0748, "step": 2405 }, { "epoch": 0.12827211174494854, "grad_norm": 11.560741424560547, "learning_rate": 9.519e-06, "loss": 0.095, "step": 2406 }, { "epoch": 0.1283254251746015, "grad_norm": 8.043791770935059, "learning_rate": 9.518800000000001e-06, "loss": 0.0521, "step": 2407 }, { "epoch": 0.12837873860425442, "grad_norm": 4.457624435424805, "learning_rate": 9.5186e-06, "loss": 0.047, "step": 2408 }, { "epoch": 0.12843205203390734, "grad_norm": 3.8943088054656982, "learning_rate": 9.5184e-06, "loss": 0.0465, "step": 2409 }, { "epoch": 0.12848536546356026, "grad_norm": 5.221862316131592, "learning_rate": 9.518200000000001e-06, "loss": 0.0584, "step": 2410 }, { "epoch": 0.1285386788932132, "grad_norm": 4.886537075042725, "learning_rate": 9.518000000000001e-06, "loss": 0.0539, "step": 2411 }, { "epoch": 0.12859199232286614, "grad_norm": 8.800799369812012, "learning_rate": 9.5178e-06, "loss": 0.0981, "step": 2412 }, { "epoch": 0.12864530575251906, "grad_norm": 7.495689392089844, "learning_rate": 9.517600000000002e-06, "loss": 0.0673, "step": 2413 }, { "epoch": 0.12869861918217199, "grad_norm": 4.999023914337158, "learning_rate": 9.517400000000001e-06, "loss": 0.0662, "step": 2414 }, { "epoch": 0.1287519326118249, "grad_norm": 6.154995918273926, "learning_rate": 9.517200000000001e-06, "loss": 0.052, "step": 2415 }, { "epoch": 0.12880524604147786, "grad_norm": 4.069422245025635, "learning_rate": 9.517e-06, "loss": 0.0509, "step": 2416 }, { "epoch": 0.12885855947113078, "grad_norm": 6.665944576263428, "learning_rate": 9.5168e-06, "loss": 0.0646, "step": 2417 }, { "epoch": 0.1289118729007837, "grad_norm": 11.031054496765137, "learning_rate": 9.5166e-06, "loss": 0.1376, "step": 2418 }, { "epoch": 0.12896518633043663, "grad_norm": 7.225711345672607, "learning_rate": 9.516400000000001e-06, "loss": 0.0857, "step": 2419 }, { "epoch": 0.12901849976008956, "grad_norm": 4.590983867645264, "learning_rate": 9.5162e-06, "loss": 0.109, "step": 2420 }, { "epoch": 0.1290718131897425, "grad_norm": 4.916673183441162, "learning_rate": 9.516e-06, "loss": 0.0476, "step": 2421 }, { "epoch": 0.12912512661939543, "grad_norm": 5.26749849319458, "learning_rate": 9.515800000000001e-06, "loss": 0.0847, "step": 2422 }, { "epoch": 0.12917844004904835, "grad_norm": 11.154993057250977, "learning_rate": 9.515600000000001e-06, "loss": 0.0785, "step": 2423 }, { "epoch": 0.12923175347870128, "grad_norm": 5.99039888381958, "learning_rate": 9.515400000000002e-06, "loss": 0.0886, "step": 2424 }, { "epoch": 0.1292850669083542, "grad_norm": 4.3894524574279785, "learning_rate": 9.515200000000002e-06, "loss": 0.0794, "step": 2425 }, { "epoch": 0.12933838033800715, "grad_norm": 4.273664474487305, "learning_rate": 9.515000000000001e-06, "loss": 0.049, "step": 2426 }, { "epoch": 0.12939169376766008, "grad_norm": 4.018242835998535, "learning_rate": 9.5148e-06, "loss": 0.0719, "step": 2427 }, { "epoch": 0.129445007197313, "grad_norm": 7.240436553955078, "learning_rate": 9.5146e-06, "loss": 0.0816, "step": 2428 }, { "epoch": 0.12949832062696592, "grad_norm": 4.817876815795898, "learning_rate": 9.5144e-06, "loss": 0.0564, "step": 2429 }, { "epoch": 0.12955163405661888, "grad_norm": 5.479523181915283, "learning_rate": 9.514200000000001e-06, "loss": 0.0832, "step": 2430 }, { "epoch": 0.1296049474862718, "grad_norm": 5.431707859039307, "learning_rate": 9.514e-06, "loss": 0.083, "step": 2431 }, { "epoch": 0.12965826091592472, "grad_norm": 4.17360258102417, "learning_rate": 9.5138e-06, "loss": 0.0739, "step": 2432 }, { "epoch": 0.12971157434557765, "grad_norm": 8.821337699890137, "learning_rate": 9.5136e-06, "loss": 0.083, "step": 2433 }, { "epoch": 0.12976488777523057, "grad_norm": 7.383172512054443, "learning_rate": 9.513400000000001e-06, "loss": 0.0565, "step": 2434 }, { "epoch": 0.12981820120488352, "grad_norm": 3.446603775024414, "learning_rate": 9.5132e-06, "loss": 0.0429, "step": 2435 }, { "epoch": 0.12987151463453644, "grad_norm": 6.280208587646484, "learning_rate": 9.513000000000002e-06, "loss": 0.0671, "step": 2436 }, { "epoch": 0.12992482806418937, "grad_norm": 5.968794822692871, "learning_rate": 9.512800000000001e-06, "loss": 0.0889, "step": 2437 }, { "epoch": 0.1299781414938423, "grad_norm": 7.075199604034424, "learning_rate": 9.512600000000001e-06, "loss": 0.058, "step": 2438 }, { "epoch": 0.13003145492349522, "grad_norm": 6.196951866149902, "learning_rate": 9.5124e-06, "loss": 0.0621, "step": 2439 }, { "epoch": 0.13008476835314817, "grad_norm": 4.547242164611816, "learning_rate": 9.5122e-06, "loss": 0.0682, "step": 2440 }, { "epoch": 0.1301380817828011, "grad_norm": 6.994466304779053, "learning_rate": 9.512000000000001e-06, "loss": 0.0961, "step": 2441 }, { "epoch": 0.13019139521245401, "grad_norm": 8.579412460327148, "learning_rate": 9.511800000000001e-06, "loss": 0.1169, "step": 2442 }, { "epoch": 0.13024470864210694, "grad_norm": 9.793623924255371, "learning_rate": 9.5116e-06, "loss": 0.0785, "step": 2443 }, { "epoch": 0.1302980220717599, "grad_norm": 5.454773902893066, "learning_rate": 9.5114e-06, "loss": 0.0708, "step": 2444 }, { "epoch": 0.1303513355014128, "grad_norm": 4.151304244995117, "learning_rate": 9.5112e-06, "loss": 0.0482, "step": 2445 }, { "epoch": 0.13040464893106574, "grad_norm": 3.6420648097991943, "learning_rate": 9.511000000000001e-06, "loss": 0.0378, "step": 2446 }, { "epoch": 0.13045796236071866, "grad_norm": 3.85981822013855, "learning_rate": 9.5108e-06, "loss": 0.0656, "step": 2447 }, { "epoch": 0.13051127579037158, "grad_norm": 8.183548927307129, "learning_rate": 9.510600000000002e-06, "loss": 0.0788, "step": 2448 }, { "epoch": 0.13056458922002454, "grad_norm": 5.733021259307861, "learning_rate": 9.510400000000001e-06, "loss": 0.0486, "step": 2449 }, { "epoch": 0.13061790264967746, "grad_norm": 12.294954299926758, "learning_rate": 9.510200000000001e-06, "loss": 0.1049, "step": 2450 }, { "epoch": 0.13067121607933038, "grad_norm": 6.317641258239746, "learning_rate": 9.51e-06, "loss": 0.0447, "step": 2451 }, { "epoch": 0.1307245295089833, "grad_norm": 7.513126850128174, "learning_rate": 9.509800000000002e-06, "loss": 0.07, "step": 2452 }, { "epoch": 0.13077784293863623, "grad_norm": 7.214866638183594, "learning_rate": 9.509600000000001e-06, "loss": 0.0717, "step": 2453 }, { "epoch": 0.13083115636828918, "grad_norm": 7.025999069213867, "learning_rate": 9.5094e-06, "loss": 0.0824, "step": 2454 }, { "epoch": 0.1308844697979421, "grad_norm": 3.768169403076172, "learning_rate": 9.5092e-06, "loss": 0.0483, "step": 2455 }, { "epoch": 0.13093778322759503, "grad_norm": 6.90902853012085, "learning_rate": 9.509e-06, "loss": 0.0637, "step": 2456 }, { "epoch": 0.13099109665724795, "grad_norm": 3.2013323307037354, "learning_rate": 9.5088e-06, "loss": 0.0476, "step": 2457 }, { "epoch": 0.1310444100869009, "grad_norm": 10.6218900680542, "learning_rate": 9.5086e-06, "loss": 0.077, "step": 2458 }, { "epoch": 0.13109772351655383, "grad_norm": 5.748955249786377, "learning_rate": 9.5084e-06, "loss": 0.0551, "step": 2459 }, { "epoch": 0.13115103694620675, "grad_norm": 4.740905284881592, "learning_rate": 9.508200000000002e-06, "loss": 0.0526, "step": 2460 }, { "epoch": 0.13120435037585967, "grad_norm": 8.7058687210083, "learning_rate": 9.508000000000001e-06, "loss": 0.0906, "step": 2461 }, { "epoch": 0.1312576638055126, "grad_norm": 5.039515018463135, "learning_rate": 9.5078e-06, "loss": 0.0602, "step": 2462 }, { "epoch": 0.13131097723516555, "grad_norm": 4.576518535614014, "learning_rate": 9.507600000000002e-06, "loss": 0.0598, "step": 2463 }, { "epoch": 0.13136429066481847, "grad_norm": 6.209156036376953, "learning_rate": 9.507400000000002e-06, "loss": 0.0582, "step": 2464 }, { "epoch": 0.1314176040944714, "grad_norm": 7.193066596984863, "learning_rate": 9.507200000000001e-06, "loss": 0.0531, "step": 2465 }, { "epoch": 0.13147091752412432, "grad_norm": 4.152843952178955, "learning_rate": 9.507e-06, "loss": 0.041, "step": 2466 }, { "epoch": 0.13152423095377724, "grad_norm": 6.660765171051025, "learning_rate": 9.5068e-06, "loss": 0.1087, "step": 2467 }, { "epoch": 0.1315775443834302, "grad_norm": 4.845313549041748, "learning_rate": 9.5066e-06, "loss": 0.0452, "step": 2468 }, { "epoch": 0.13163085781308312, "grad_norm": 4.9347100257873535, "learning_rate": 9.506400000000001e-06, "loss": 0.0579, "step": 2469 }, { "epoch": 0.13168417124273604, "grad_norm": 4.209217071533203, "learning_rate": 9.5062e-06, "loss": 0.0371, "step": 2470 }, { "epoch": 0.13173748467238897, "grad_norm": 7.445100784301758, "learning_rate": 9.506e-06, "loss": 0.0958, "step": 2471 }, { "epoch": 0.13179079810204192, "grad_norm": 5.7105913162231445, "learning_rate": 9.505800000000001e-06, "loss": 0.0658, "step": 2472 }, { "epoch": 0.13184411153169484, "grad_norm": 5.998685836791992, "learning_rate": 9.505600000000001e-06, "loss": 0.0879, "step": 2473 }, { "epoch": 0.13189742496134776, "grad_norm": 7.164468288421631, "learning_rate": 9.5054e-06, "loss": 0.0914, "step": 2474 }, { "epoch": 0.1319507383910007, "grad_norm": 11.005521774291992, "learning_rate": 9.505200000000002e-06, "loss": 0.0686, "step": 2475 }, { "epoch": 0.1320040518206536, "grad_norm": 8.61903190612793, "learning_rate": 9.505000000000001e-06, "loss": 0.0768, "step": 2476 }, { "epoch": 0.13205736525030656, "grad_norm": 5.104544639587402, "learning_rate": 9.504800000000001e-06, "loss": 0.0584, "step": 2477 }, { "epoch": 0.1321106786799595, "grad_norm": 7.428233623504639, "learning_rate": 9.5046e-06, "loss": 0.0832, "step": 2478 }, { "epoch": 0.1321639921096124, "grad_norm": 6.108643054962158, "learning_rate": 9.5044e-06, "loss": 0.0757, "step": 2479 }, { "epoch": 0.13221730553926533, "grad_norm": 6.061413288116455, "learning_rate": 9.504200000000001e-06, "loss": 0.0916, "step": 2480 }, { "epoch": 0.13227061896891826, "grad_norm": 4.112868785858154, "learning_rate": 9.504e-06, "loss": 0.0506, "step": 2481 }, { "epoch": 0.1323239323985712, "grad_norm": 3.323298692703247, "learning_rate": 9.5038e-06, "loss": 0.0384, "step": 2482 }, { "epoch": 0.13237724582822413, "grad_norm": 6.2104902267456055, "learning_rate": 9.5036e-06, "loss": 0.0886, "step": 2483 }, { "epoch": 0.13243055925787706, "grad_norm": 4.37217378616333, "learning_rate": 9.503400000000001e-06, "loss": 0.0897, "step": 2484 }, { "epoch": 0.13248387268752998, "grad_norm": 9.164115905761719, "learning_rate": 9.5032e-06, "loss": 0.0809, "step": 2485 }, { "epoch": 0.13253718611718293, "grad_norm": 5.238144874572754, "learning_rate": 9.503e-06, "loss": 0.0553, "step": 2486 }, { "epoch": 0.13259049954683585, "grad_norm": 7.116396903991699, "learning_rate": 9.502800000000002e-06, "loss": 0.0747, "step": 2487 }, { "epoch": 0.13264381297648878, "grad_norm": 9.053709030151367, "learning_rate": 9.502600000000001e-06, "loss": 0.0775, "step": 2488 }, { "epoch": 0.1326971264061417, "grad_norm": 4.0396647453308105, "learning_rate": 9.5024e-06, "loss": 0.0442, "step": 2489 }, { "epoch": 0.13275043983579463, "grad_norm": 3.3413193225860596, "learning_rate": 9.5022e-06, "loss": 0.0483, "step": 2490 }, { "epoch": 0.13280375326544758, "grad_norm": 5.569638252258301, "learning_rate": 9.502000000000002e-06, "loss": 0.0839, "step": 2491 }, { "epoch": 0.1328570666951005, "grad_norm": 10.56816291809082, "learning_rate": 9.501800000000001e-06, "loss": 0.1166, "step": 2492 }, { "epoch": 0.13291038012475342, "grad_norm": 7.351048469543457, "learning_rate": 9.5016e-06, "loss": 0.0526, "step": 2493 }, { "epoch": 0.13296369355440635, "grad_norm": 3.823554754257202, "learning_rate": 9.5014e-06, "loss": 0.0287, "step": 2494 }, { "epoch": 0.13301700698405927, "grad_norm": 4.44343900680542, "learning_rate": 9.5012e-06, "loss": 0.085, "step": 2495 }, { "epoch": 0.13307032041371222, "grad_norm": 13.262333869934082, "learning_rate": 9.501000000000001e-06, "loss": 0.0968, "step": 2496 }, { "epoch": 0.13312363384336515, "grad_norm": 15.606441497802734, "learning_rate": 9.5008e-06, "loss": 0.1032, "step": 2497 }, { "epoch": 0.13317694727301807, "grad_norm": 3.671945571899414, "learning_rate": 9.5006e-06, "loss": 0.0683, "step": 2498 }, { "epoch": 0.133230260702671, "grad_norm": 5.412996768951416, "learning_rate": 9.500400000000001e-06, "loss": 0.0555, "step": 2499 }, { "epoch": 0.13328357413232395, "grad_norm": 4.891412258148193, "learning_rate": 9.500200000000001e-06, "loss": 0.0321, "step": 2500 }, { "epoch": 0.13333688756197687, "grad_norm": 5.56758975982666, "learning_rate": 9.5e-06, "loss": 0.0509, "step": 2501 }, { "epoch": 0.1333902009916298, "grad_norm": 5.496169567108154, "learning_rate": 9.499800000000002e-06, "loss": 0.08, "step": 2502 }, { "epoch": 0.13344351442128272, "grad_norm": 4.370581150054932, "learning_rate": 9.499600000000001e-06, "loss": 0.0509, "step": 2503 }, { "epoch": 0.13349682785093564, "grad_norm": 5.887495517730713, "learning_rate": 9.499400000000001e-06, "loss": 0.0989, "step": 2504 }, { "epoch": 0.1335501412805886, "grad_norm": 3.205383062362671, "learning_rate": 9.4992e-06, "loss": 0.0495, "step": 2505 }, { "epoch": 0.13360345471024151, "grad_norm": 2.9679677486419678, "learning_rate": 9.499e-06, "loss": 0.0496, "step": 2506 }, { "epoch": 0.13365676813989444, "grad_norm": 11.5258150100708, "learning_rate": 9.4988e-06, "loss": 0.0854, "step": 2507 }, { "epoch": 0.13371008156954736, "grad_norm": 13.214839935302734, "learning_rate": 9.4986e-06, "loss": 0.0888, "step": 2508 }, { "epoch": 0.13376339499920029, "grad_norm": 5.19655704498291, "learning_rate": 9.4984e-06, "loss": 0.0723, "step": 2509 }, { "epoch": 0.13381670842885324, "grad_norm": 6.57230806350708, "learning_rate": 9.4982e-06, "loss": 0.0785, "step": 2510 }, { "epoch": 0.13387002185850616, "grad_norm": 5.855205059051514, "learning_rate": 9.498000000000001e-06, "loss": 0.0571, "step": 2511 }, { "epoch": 0.13392333528815908, "grad_norm": 8.118059158325195, "learning_rate": 9.4978e-06, "loss": 0.0916, "step": 2512 }, { "epoch": 0.133976648717812, "grad_norm": 2.4443435668945312, "learning_rate": 9.497600000000002e-06, "loss": 0.0325, "step": 2513 }, { "epoch": 0.13402996214746496, "grad_norm": 6.792413711547852, "learning_rate": 9.497400000000002e-06, "loss": 0.0572, "step": 2514 }, { "epoch": 0.13408327557711788, "grad_norm": 3.957848072052002, "learning_rate": 9.497200000000001e-06, "loss": 0.0563, "step": 2515 }, { "epoch": 0.1341365890067708, "grad_norm": 9.025157928466797, "learning_rate": 9.497e-06, "loss": 0.1259, "step": 2516 }, { "epoch": 0.13418990243642373, "grad_norm": 6.642300605773926, "learning_rate": 9.4968e-06, "loss": 0.1074, "step": 2517 }, { "epoch": 0.13424321586607665, "grad_norm": 6.271884441375732, "learning_rate": 9.4966e-06, "loss": 0.0477, "step": 2518 }, { "epoch": 0.1342965292957296, "grad_norm": 3.674980878829956, "learning_rate": 9.496400000000001e-06, "loss": 0.0682, "step": 2519 }, { "epoch": 0.13434984272538253, "grad_norm": 4.915855407714844, "learning_rate": 9.4962e-06, "loss": 0.1002, "step": 2520 }, { "epoch": 0.13440315615503545, "grad_norm": 4.668018341064453, "learning_rate": 9.496e-06, "loss": 0.0638, "step": 2521 }, { "epoch": 0.13445646958468838, "grad_norm": 4.924293518066406, "learning_rate": 9.4958e-06, "loss": 0.0478, "step": 2522 }, { "epoch": 0.1345097830143413, "grad_norm": 3.905712842941284, "learning_rate": 9.495600000000001e-06, "loss": 0.0446, "step": 2523 }, { "epoch": 0.13456309644399425, "grad_norm": 3.2948663234710693, "learning_rate": 9.4954e-06, "loss": 0.0759, "step": 2524 }, { "epoch": 0.13461640987364717, "grad_norm": 3.1691079139709473, "learning_rate": 9.495200000000002e-06, "loss": 0.0537, "step": 2525 }, { "epoch": 0.1346697233033001, "grad_norm": 4.899756908416748, "learning_rate": 9.495000000000001e-06, "loss": 0.0583, "step": 2526 }, { "epoch": 0.13472303673295302, "grad_norm": 4.7249369621276855, "learning_rate": 9.494800000000001e-06, "loss": 0.0707, "step": 2527 }, { "epoch": 0.13477635016260597, "grad_norm": 6.678049564361572, "learning_rate": 9.4946e-06, "loss": 0.1186, "step": 2528 }, { "epoch": 0.1348296635922589, "grad_norm": 6.271153450012207, "learning_rate": 9.4944e-06, "loss": 0.0749, "step": 2529 }, { "epoch": 0.13488297702191182, "grad_norm": 3.0777204036712646, "learning_rate": 9.494200000000001e-06, "loss": 0.0367, "step": 2530 }, { "epoch": 0.13493629045156474, "grad_norm": 3.3961946964263916, "learning_rate": 9.494000000000001e-06, "loss": 0.0475, "step": 2531 }, { "epoch": 0.13498960388121767, "grad_norm": 4.630449295043945, "learning_rate": 9.4938e-06, "loss": 0.0959, "step": 2532 }, { "epoch": 0.13504291731087062, "grad_norm": 4.582997798919678, "learning_rate": 9.4936e-06, "loss": 0.0394, "step": 2533 }, { "epoch": 0.13509623074052354, "grad_norm": 2.3754830360412598, "learning_rate": 9.4934e-06, "loss": 0.0485, "step": 2534 }, { "epoch": 0.13514954417017647, "grad_norm": 6.229498863220215, "learning_rate": 9.493200000000001e-06, "loss": 0.0599, "step": 2535 }, { "epoch": 0.1352028575998294, "grad_norm": 6.044164657592773, "learning_rate": 9.493e-06, "loss": 0.0434, "step": 2536 }, { "epoch": 0.1352561710294823, "grad_norm": 5.830554485321045, "learning_rate": 9.492800000000002e-06, "loss": 0.1052, "step": 2537 }, { "epoch": 0.13530948445913527, "grad_norm": 3.0680811405181885, "learning_rate": 9.492600000000001e-06, "loss": 0.0422, "step": 2538 }, { "epoch": 0.1353627978887882, "grad_norm": 3.673069715499878, "learning_rate": 9.4924e-06, "loss": 0.0387, "step": 2539 }, { "epoch": 0.1354161113184411, "grad_norm": 6.763548851013184, "learning_rate": 9.4922e-06, "loss": 0.0841, "step": 2540 }, { "epoch": 0.13546942474809404, "grad_norm": 4.487274646759033, "learning_rate": 9.492000000000002e-06, "loss": 0.0571, "step": 2541 }, { "epoch": 0.135522738177747, "grad_norm": 5.804204940795898, "learning_rate": 9.491800000000001e-06, "loss": 0.091, "step": 2542 }, { "epoch": 0.1355760516073999, "grad_norm": 6.537950038909912, "learning_rate": 9.4916e-06, "loss": 0.063, "step": 2543 }, { "epoch": 0.13562936503705283, "grad_norm": 5.443797588348389, "learning_rate": 9.4914e-06, "loss": 0.0749, "step": 2544 }, { "epoch": 0.13568267846670576, "grad_norm": 7.788443565368652, "learning_rate": 9.4912e-06, "loss": 0.0728, "step": 2545 }, { "epoch": 0.13573599189635868, "grad_norm": 9.65627670288086, "learning_rate": 9.491000000000001e-06, "loss": 0.0758, "step": 2546 }, { "epoch": 0.13578930532601163, "grad_norm": 5.242218494415283, "learning_rate": 9.4908e-06, "loss": 0.0503, "step": 2547 }, { "epoch": 0.13584261875566456, "grad_norm": 6.958974361419678, "learning_rate": 9.4906e-06, "loss": 0.0796, "step": 2548 }, { "epoch": 0.13589593218531748, "grad_norm": 5.30303955078125, "learning_rate": 9.490400000000001e-06, "loss": 0.0642, "step": 2549 }, { "epoch": 0.1359492456149704, "grad_norm": 8.981188774108887, "learning_rate": 9.490200000000001e-06, "loss": 0.0712, "step": 2550 }, { "epoch": 0.13600255904462333, "grad_norm": 4.918367385864258, "learning_rate": 9.49e-06, "loss": 0.0634, "step": 2551 }, { "epoch": 0.13605587247427628, "grad_norm": 3.752230405807495, "learning_rate": 9.489800000000002e-06, "loss": 0.0435, "step": 2552 }, { "epoch": 0.1361091859039292, "grad_norm": 5.567627906799316, "learning_rate": 9.489600000000001e-06, "loss": 0.0728, "step": 2553 }, { "epoch": 0.13616249933358213, "grad_norm": 6.263791561126709, "learning_rate": 9.489400000000001e-06, "loss": 0.1114, "step": 2554 }, { "epoch": 0.13621581276323505, "grad_norm": 4.590132236480713, "learning_rate": 9.4892e-06, "loss": 0.0689, "step": 2555 }, { "epoch": 0.136269126192888, "grad_norm": 5.905659198760986, "learning_rate": 9.489e-06, "loss": 0.0752, "step": 2556 }, { "epoch": 0.13632243962254093, "grad_norm": 10.75754451751709, "learning_rate": 9.4888e-06, "loss": 0.0725, "step": 2557 }, { "epoch": 0.13637575305219385, "grad_norm": 4.623749732971191, "learning_rate": 9.488600000000001e-06, "loss": 0.0483, "step": 2558 }, { "epoch": 0.13642906648184677, "grad_norm": 5.101014614105225, "learning_rate": 9.4884e-06, "loss": 0.0771, "step": 2559 }, { "epoch": 0.1364823799114997, "grad_norm": 8.84290885925293, "learning_rate": 9.4882e-06, "loss": 0.0906, "step": 2560 }, { "epoch": 0.13653569334115265, "grad_norm": 7.314949035644531, "learning_rate": 9.488000000000001e-06, "loss": 0.0992, "step": 2561 }, { "epoch": 0.13658900677080557, "grad_norm": 4.584573268890381, "learning_rate": 9.487800000000001e-06, "loss": 0.0517, "step": 2562 }, { "epoch": 0.1366423202004585, "grad_norm": 4.725846767425537, "learning_rate": 9.4876e-06, "loss": 0.0548, "step": 2563 }, { "epoch": 0.13669563363011142, "grad_norm": 6.045044422149658, "learning_rate": 9.487400000000002e-06, "loss": 0.0898, "step": 2564 }, { "epoch": 0.13674894705976434, "grad_norm": 7.526721954345703, "learning_rate": 9.487200000000001e-06, "loss": 0.0906, "step": 2565 }, { "epoch": 0.1368022604894173, "grad_norm": 4.521383762359619, "learning_rate": 9.487e-06, "loss": 0.0634, "step": 2566 }, { "epoch": 0.13685557391907022, "grad_norm": 6.484198570251465, "learning_rate": 9.4868e-06, "loss": 0.0613, "step": 2567 }, { "epoch": 0.13690888734872314, "grad_norm": 8.800575256347656, "learning_rate": 9.4866e-06, "loss": 0.1143, "step": 2568 }, { "epoch": 0.13696220077837606, "grad_norm": 4.667126655578613, "learning_rate": 9.486400000000001e-06, "loss": 0.0955, "step": 2569 }, { "epoch": 0.13701551420802902, "grad_norm": 5.537447452545166, "learning_rate": 9.4862e-06, "loss": 0.0887, "step": 2570 }, { "epoch": 0.13706882763768194, "grad_norm": 4.383976459503174, "learning_rate": 9.486e-06, "loss": 0.0843, "step": 2571 }, { "epoch": 0.13712214106733486, "grad_norm": 4.369630336761475, "learning_rate": 9.4858e-06, "loss": 0.0753, "step": 2572 }, { "epoch": 0.1371754544969878, "grad_norm": 5.205353736877441, "learning_rate": 9.485600000000001e-06, "loss": 0.0762, "step": 2573 }, { "epoch": 0.1372287679266407, "grad_norm": 4.723837375640869, "learning_rate": 9.4854e-06, "loss": 0.0755, "step": 2574 }, { "epoch": 0.13728208135629366, "grad_norm": 9.71926498413086, "learning_rate": 9.4852e-06, "loss": 0.1035, "step": 2575 }, { "epoch": 0.13733539478594659, "grad_norm": 4.444847583770752, "learning_rate": 9.485000000000002e-06, "loss": 0.0444, "step": 2576 }, { "epoch": 0.1373887082155995, "grad_norm": 2.686579704284668, "learning_rate": 9.484800000000001e-06, "loss": 0.0459, "step": 2577 }, { "epoch": 0.13744202164525243, "grad_norm": 5.0355377197265625, "learning_rate": 9.4846e-06, "loss": 0.0502, "step": 2578 }, { "epoch": 0.13749533507490536, "grad_norm": 4.586061477661133, "learning_rate": 9.4844e-06, "loss": 0.0551, "step": 2579 }, { "epoch": 0.1375486485045583, "grad_norm": 3.2147557735443115, "learning_rate": 9.484200000000001e-06, "loss": 0.0388, "step": 2580 }, { "epoch": 0.13760196193421123, "grad_norm": 4.404341220855713, "learning_rate": 9.484000000000001e-06, "loss": 0.0483, "step": 2581 }, { "epoch": 0.13765527536386415, "grad_norm": 5.747556209564209, "learning_rate": 9.4838e-06, "loss": 0.0961, "step": 2582 }, { "epoch": 0.13770858879351708, "grad_norm": 4.439338684082031, "learning_rate": 9.4836e-06, "loss": 0.042, "step": 2583 }, { "epoch": 0.13776190222317003, "grad_norm": 4.679214954376221, "learning_rate": 9.4834e-06, "loss": 0.0647, "step": 2584 }, { "epoch": 0.13781521565282295, "grad_norm": 7.929930686950684, "learning_rate": 9.483200000000001e-06, "loss": 0.095, "step": 2585 }, { "epoch": 0.13786852908247588, "grad_norm": 7.612556457519531, "learning_rate": 9.483e-06, "loss": 0.0736, "step": 2586 }, { "epoch": 0.1379218425121288, "grad_norm": 5.256961345672607, "learning_rate": 9.4828e-06, "loss": 0.0967, "step": 2587 }, { "epoch": 0.13797515594178172, "grad_norm": 7.9519524574279785, "learning_rate": 9.482600000000001e-06, "loss": 0.0785, "step": 2588 }, { "epoch": 0.13802846937143468, "grad_norm": 6.787533760070801, "learning_rate": 9.482400000000001e-06, "loss": 0.059, "step": 2589 }, { "epoch": 0.1380817828010876, "grad_norm": 3.5975265502929688, "learning_rate": 9.4822e-06, "loss": 0.0497, "step": 2590 }, { "epoch": 0.13813509623074052, "grad_norm": 4.009990692138672, "learning_rate": 9.482000000000002e-06, "loss": 0.0658, "step": 2591 }, { "epoch": 0.13818840966039345, "grad_norm": 6.357887268066406, "learning_rate": 9.481800000000001e-06, "loss": 0.1088, "step": 2592 }, { "epoch": 0.13824172309004637, "grad_norm": 7.076717853546143, "learning_rate": 9.4816e-06, "loss": 0.0791, "step": 2593 }, { "epoch": 0.13829503651969932, "grad_norm": 4.322475910186768, "learning_rate": 9.4814e-06, "loss": 0.0585, "step": 2594 }, { "epoch": 0.13834834994935225, "grad_norm": 3.882840394973755, "learning_rate": 9.4812e-06, "loss": 0.0475, "step": 2595 }, { "epoch": 0.13840166337900517, "grad_norm": 4.316417217254639, "learning_rate": 9.481000000000001e-06, "loss": 0.0597, "step": 2596 }, { "epoch": 0.1384549768086581, "grad_norm": 4.287191867828369, "learning_rate": 9.4808e-06, "loss": 0.0625, "step": 2597 }, { "epoch": 0.13850829023831104, "grad_norm": 3.920620918273926, "learning_rate": 9.4806e-06, "loss": 0.0469, "step": 2598 }, { "epoch": 0.13856160366796397, "grad_norm": 5.768825054168701, "learning_rate": 9.4804e-06, "loss": 0.1278, "step": 2599 }, { "epoch": 0.1386149170976169, "grad_norm": 4.389445781707764, "learning_rate": 9.480200000000001e-06, "loss": 0.0771, "step": 2600 }, { "epoch": 0.13866823052726981, "grad_norm": 6.164566993713379, "learning_rate": 9.48e-06, "loss": 0.096, "step": 2601 }, { "epoch": 0.13872154395692274, "grad_norm": 4.61724853515625, "learning_rate": 9.479800000000002e-06, "loss": 0.035, "step": 2602 }, { "epoch": 0.1387748573865757, "grad_norm": 7.472314834594727, "learning_rate": 9.479600000000002e-06, "loss": 0.0842, "step": 2603 }, { "epoch": 0.1388281708162286, "grad_norm": 5.891531467437744, "learning_rate": 9.479400000000001e-06, "loss": 0.0803, "step": 2604 }, { "epoch": 0.13888148424588154, "grad_norm": 5.525386810302734, "learning_rate": 9.4792e-06, "loss": 0.0766, "step": 2605 }, { "epoch": 0.13893479767553446, "grad_norm": 5.666056156158447, "learning_rate": 9.479e-06, "loss": 0.0627, "step": 2606 }, { "epoch": 0.13898811110518738, "grad_norm": 7.582596778869629, "learning_rate": 9.4788e-06, "loss": 0.1124, "step": 2607 }, { "epoch": 0.13904142453484034, "grad_norm": 4.30995512008667, "learning_rate": 9.478600000000001e-06, "loss": 0.0575, "step": 2608 }, { "epoch": 0.13909473796449326, "grad_norm": 6.207901954650879, "learning_rate": 9.4784e-06, "loss": 0.0711, "step": 2609 }, { "epoch": 0.13914805139414618, "grad_norm": 5.172885417938232, "learning_rate": 9.4782e-06, "loss": 0.0994, "step": 2610 }, { "epoch": 0.1392013648237991, "grad_norm": 10.402176856994629, "learning_rate": 9.478e-06, "loss": 0.0726, "step": 2611 }, { "epoch": 0.13925467825345206, "grad_norm": 3.8313050270080566, "learning_rate": 9.477800000000001e-06, "loss": 0.0735, "step": 2612 }, { "epoch": 0.13930799168310498, "grad_norm": 2.729245662689209, "learning_rate": 9.4776e-06, "loss": 0.0481, "step": 2613 }, { "epoch": 0.1393613051127579, "grad_norm": 6.002742290496826, "learning_rate": 9.477400000000002e-06, "loss": 0.1055, "step": 2614 }, { "epoch": 0.13941461854241083, "grad_norm": 4.978211402893066, "learning_rate": 9.477200000000001e-06, "loss": 0.0574, "step": 2615 }, { "epoch": 0.13946793197206375, "grad_norm": 3.1781749725341797, "learning_rate": 9.477000000000001e-06, "loss": 0.0405, "step": 2616 }, { "epoch": 0.1395212454017167, "grad_norm": 4.127377986907959, "learning_rate": 9.4768e-06, "loss": 0.0752, "step": 2617 }, { "epoch": 0.13957455883136963, "grad_norm": 4.362404823303223, "learning_rate": 9.4766e-06, "loss": 0.0511, "step": 2618 }, { "epoch": 0.13962787226102255, "grad_norm": 7.077073097229004, "learning_rate": 9.476400000000001e-06, "loss": 0.1012, "step": 2619 }, { "epoch": 0.13968118569067547, "grad_norm": 8.758520126342773, "learning_rate": 9.476200000000001e-06, "loss": 0.0718, "step": 2620 }, { "epoch": 0.1397344991203284, "grad_norm": 4.26593542098999, "learning_rate": 9.476e-06, "loss": 0.058, "step": 2621 }, { "epoch": 0.13978781254998135, "grad_norm": 5.892333507537842, "learning_rate": 9.4758e-06, "loss": 0.0713, "step": 2622 }, { "epoch": 0.13984112597963427, "grad_norm": 3.839790105819702, "learning_rate": 9.4756e-06, "loss": 0.0402, "step": 2623 }, { "epoch": 0.1398944394092872, "grad_norm": 4.749054908752441, "learning_rate": 9.4754e-06, "loss": 0.0613, "step": 2624 }, { "epoch": 0.13994775283894012, "grad_norm": 10.193136215209961, "learning_rate": 9.4752e-06, "loss": 0.0683, "step": 2625 }, { "epoch": 0.14000106626859307, "grad_norm": 9.654473304748535, "learning_rate": 9.475000000000002e-06, "loss": 0.0903, "step": 2626 }, { "epoch": 0.140054379698246, "grad_norm": 2.792113780975342, "learning_rate": 9.474800000000001e-06, "loss": 0.0328, "step": 2627 }, { "epoch": 0.14010769312789892, "grad_norm": 3.3850409984588623, "learning_rate": 9.4746e-06, "loss": 0.0459, "step": 2628 }, { "epoch": 0.14016100655755184, "grad_norm": 5.170468807220459, "learning_rate": 9.4744e-06, "loss": 0.0529, "step": 2629 }, { "epoch": 0.14021431998720477, "grad_norm": 4.847060203552246, "learning_rate": 9.474200000000002e-06, "loss": 0.041, "step": 2630 }, { "epoch": 0.14026763341685772, "grad_norm": 9.356976509094238, "learning_rate": 9.474000000000001e-06, "loss": 0.1093, "step": 2631 }, { "epoch": 0.14032094684651064, "grad_norm": 3.4563798904418945, "learning_rate": 9.4738e-06, "loss": 0.0397, "step": 2632 }, { "epoch": 0.14037426027616356, "grad_norm": 4.063190460205078, "learning_rate": 9.4736e-06, "loss": 0.0518, "step": 2633 }, { "epoch": 0.1404275737058165, "grad_norm": 4.696131229400635, "learning_rate": 9.4734e-06, "loss": 0.0729, "step": 2634 }, { "epoch": 0.1404808871354694, "grad_norm": 5.54452657699585, "learning_rate": 9.473200000000001e-06, "loss": 0.0703, "step": 2635 }, { "epoch": 0.14053420056512236, "grad_norm": 9.55959701538086, "learning_rate": 9.473e-06, "loss": 0.0834, "step": 2636 }, { "epoch": 0.1405875139947753, "grad_norm": 6.507941722869873, "learning_rate": 9.4728e-06, "loss": 0.1183, "step": 2637 }, { "epoch": 0.1406408274244282, "grad_norm": 7.760159492492676, "learning_rate": 9.472600000000001e-06, "loss": 0.0802, "step": 2638 }, { "epoch": 0.14069414085408113, "grad_norm": 4.937872886657715, "learning_rate": 9.472400000000001e-06, "loss": 0.0633, "step": 2639 }, { "epoch": 0.14074745428373409, "grad_norm": 5.195076942443848, "learning_rate": 9.4722e-06, "loss": 0.0406, "step": 2640 }, { "epoch": 0.140800767713387, "grad_norm": 5.621351718902588, "learning_rate": 9.472000000000002e-06, "loss": 0.0362, "step": 2641 }, { "epoch": 0.14085408114303993, "grad_norm": 11.180052757263184, "learning_rate": 9.471800000000001e-06, "loss": 0.1373, "step": 2642 }, { "epoch": 0.14090739457269286, "grad_norm": 5.742346286773682, "learning_rate": 9.471600000000001e-06, "loss": 0.0589, "step": 2643 }, { "epoch": 0.14096070800234578, "grad_norm": 4.4842047691345215, "learning_rate": 9.4714e-06, "loss": 0.0748, "step": 2644 }, { "epoch": 0.14101402143199873, "grad_norm": 6.2374982833862305, "learning_rate": 9.4712e-06, "loss": 0.0654, "step": 2645 }, { "epoch": 0.14106733486165166, "grad_norm": 8.753718376159668, "learning_rate": 9.471000000000001e-06, "loss": 0.0983, "step": 2646 }, { "epoch": 0.14112064829130458, "grad_norm": 7.7765913009643555, "learning_rate": 9.470800000000001e-06, "loss": 0.0675, "step": 2647 }, { "epoch": 0.1411739617209575, "grad_norm": 5.1624603271484375, "learning_rate": 9.4706e-06, "loss": 0.076, "step": 2648 }, { "epoch": 0.14122727515061043, "grad_norm": 3.567322015762329, "learning_rate": 9.4704e-06, "loss": 0.0402, "step": 2649 }, { "epoch": 0.14128058858026338, "grad_norm": 6.879333019256592, "learning_rate": 9.470200000000001e-06, "loss": 0.0566, "step": 2650 }, { "epoch": 0.1413339020099163, "grad_norm": 6.056056976318359, "learning_rate": 9.47e-06, "loss": 0.0763, "step": 2651 }, { "epoch": 0.14138721543956922, "grad_norm": 2.5234315395355225, "learning_rate": 9.4698e-06, "loss": 0.0433, "step": 2652 }, { "epoch": 0.14144052886922215, "grad_norm": 6.072152614593506, "learning_rate": 9.469600000000002e-06, "loss": 0.0754, "step": 2653 }, { "epoch": 0.1414938422988751, "grad_norm": 4.154623031616211, "learning_rate": 9.469400000000001e-06, "loss": 0.0476, "step": 2654 }, { "epoch": 0.14154715572852802, "grad_norm": 3.6816630363464355, "learning_rate": 9.4692e-06, "loss": 0.0584, "step": 2655 }, { "epoch": 0.14160046915818095, "grad_norm": 7.069291114807129, "learning_rate": 9.469e-06, "loss": 0.0891, "step": 2656 }, { "epoch": 0.14165378258783387, "grad_norm": 9.63979721069336, "learning_rate": 9.468800000000002e-06, "loss": 0.0705, "step": 2657 }, { "epoch": 0.1417070960174868, "grad_norm": 7.157464504241943, "learning_rate": 9.468600000000001e-06, "loss": 0.0655, "step": 2658 }, { "epoch": 0.14176040944713975, "grad_norm": 5.839478969573975, "learning_rate": 9.4684e-06, "loss": 0.0863, "step": 2659 }, { "epoch": 0.14181372287679267, "grad_norm": 5.344696521759033, "learning_rate": 9.4682e-06, "loss": 0.0981, "step": 2660 }, { "epoch": 0.1418670363064456, "grad_norm": 6.888393878936768, "learning_rate": 9.468e-06, "loss": 0.082, "step": 2661 }, { "epoch": 0.14192034973609852, "grad_norm": 4.208876609802246, "learning_rate": 9.467800000000001e-06, "loss": 0.054, "step": 2662 }, { "epoch": 0.14197366316575144, "grad_norm": 5.447206497192383, "learning_rate": 9.4676e-06, "loss": 0.0574, "step": 2663 }, { "epoch": 0.1420269765954044, "grad_norm": 8.86874771118164, "learning_rate": 9.4674e-06, "loss": 0.0789, "step": 2664 }, { "epoch": 0.14208029002505732, "grad_norm": 4.297821521759033, "learning_rate": 9.467200000000001e-06, "loss": 0.0704, "step": 2665 }, { "epoch": 0.14213360345471024, "grad_norm": 6.01923131942749, "learning_rate": 9.467000000000001e-06, "loss": 0.1037, "step": 2666 }, { "epoch": 0.14218691688436316, "grad_norm": 8.557138442993164, "learning_rate": 9.4668e-06, "loss": 0.074, "step": 2667 }, { "epoch": 0.14224023031401611, "grad_norm": 5.893817901611328, "learning_rate": 9.4666e-06, "loss": 0.0518, "step": 2668 }, { "epoch": 0.14229354374366904, "grad_norm": 8.29339599609375, "learning_rate": 9.466400000000001e-06, "loss": 0.0921, "step": 2669 }, { "epoch": 0.14234685717332196, "grad_norm": 7.168618202209473, "learning_rate": 9.466200000000001e-06, "loss": 0.0874, "step": 2670 }, { "epoch": 0.14240017060297488, "grad_norm": 5.127285480499268, "learning_rate": 9.466e-06, "loss": 0.0633, "step": 2671 }, { "epoch": 0.1424534840326278, "grad_norm": 4.4165215492248535, "learning_rate": 9.4658e-06, "loss": 0.0539, "step": 2672 }, { "epoch": 0.14250679746228076, "grad_norm": 3.1383659839630127, "learning_rate": 9.4656e-06, "loss": 0.0421, "step": 2673 }, { "epoch": 0.14256011089193368, "grad_norm": 4.3979082107543945, "learning_rate": 9.465400000000001e-06, "loss": 0.0373, "step": 2674 }, { "epoch": 0.1426134243215866, "grad_norm": 3.700718879699707, "learning_rate": 9.4652e-06, "loss": 0.0418, "step": 2675 }, { "epoch": 0.14266673775123953, "grad_norm": 5.895987033843994, "learning_rate": 9.465e-06, "loss": 0.0617, "step": 2676 }, { "epoch": 0.14272005118089245, "grad_norm": 4.8594183921813965, "learning_rate": 9.464800000000001e-06, "loss": 0.0589, "step": 2677 }, { "epoch": 0.1427733646105454, "grad_norm": 5.502606391906738, "learning_rate": 9.4646e-06, "loss": 0.0952, "step": 2678 }, { "epoch": 0.14282667804019833, "grad_norm": 7.629941463470459, "learning_rate": 9.4644e-06, "loss": 0.0976, "step": 2679 }, { "epoch": 0.14287999146985125, "grad_norm": 8.559369087219238, "learning_rate": 9.464200000000002e-06, "loss": 0.1112, "step": 2680 }, { "epoch": 0.14293330489950418, "grad_norm": 6.421077251434326, "learning_rate": 9.464000000000001e-06, "loss": 0.0363, "step": 2681 }, { "epoch": 0.14298661832915713, "grad_norm": 4.4159746170043945, "learning_rate": 9.4638e-06, "loss": 0.0474, "step": 2682 }, { "epoch": 0.14303993175881005, "grad_norm": 5.689158916473389, "learning_rate": 9.4636e-06, "loss": 0.0848, "step": 2683 }, { "epoch": 0.14309324518846298, "grad_norm": 5.8943586349487305, "learning_rate": 9.4634e-06, "loss": 0.0602, "step": 2684 }, { "epoch": 0.1431465586181159, "grad_norm": 2.434946060180664, "learning_rate": 9.463200000000001e-06, "loss": 0.0329, "step": 2685 }, { "epoch": 0.14319987204776882, "grad_norm": 4.857457637786865, "learning_rate": 9.463e-06, "loss": 0.0504, "step": 2686 }, { "epoch": 0.14325318547742177, "grad_norm": 3.981539726257324, "learning_rate": 9.4628e-06, "loss": 0.0556, "step": 2687 }, { "epoch": 0.1433064989070747, "grad_norm": 3.0387656688690186, "learning_rate": 9.4626e-06, "loss": 0.0393, "step": 2688 }, { "epoch": 0.14335981233672762, "grad_norm": 4.5383734703063965, "learning_rate": 9.462400000000001e-06, "loss": 0.0324, "step": 2689 }, { "epoch": 0.14341312576638054, "grad_norm": 5.166879653930664, "learning_rate": 9.4622e-06, "loss": 0.082, "step": 2690 }, { "epoch": 0.14346643919603347, "grad_norm": 5.362340927124023, "learning_rate": 9.462000000000002e-06, "loss": 0.0689, "step": 2691 }, { "epoch": 0.14351975262568642, "grad_norm": 5.4324164390563965, "learning_rate": 9.461800000000001e-06, "loss": 0.074, "step": 2692 }, { "epoch": 0.14357306605533934, "grad_norm": 4.60175895690918, "learning_rate": 9.461600000000001e-06, "loss": 0.0913, "step": 2693 }, { "epoch": 0.14362637948499227, "grad_norm": 9.982446670532227, "learning_rate": 9.4614e-06, "loss": 0.1051, "step": 2694 }, { "epoch": 0.1436796929146452, "grad_norm": 3.8568949699401855, "learning_rate": 9.4612e-06, "loss": 0.0596, "step": 2695 }, { "epoch": 0.14373300634429814, "grad_norm": 6.304455757141113, "learning_rate": 9.461000000000001e-06, "loss": 0.0754, "step": 2696 }, { "epoch": 0.14378631977395107, "grad_norm": 8.161599159240723, "learning_rate": 9.460800000000001e-06, "loss": 0.1045, "step": 2697 }, { "epoch": 0.143839633203604, "grad_norm": 6.839254856109619, "learning_rate": 9.4606e-06, "loss": 0.089, "step": 2698 }, { "epoch": 0.1438929466332569, "grad_norm": 6.566006660461426, "learning_rate": 9.4604e-06, "loss": 0.0572, "step": 2699 }, { "epoch": 0.14394626006290984, "grad_norm": 4.950960636138916, "learning_rate": 9.4602e-06, "loss": 0.0756, "step": 2700 }, { "epoch": 0.1439995734925628, "grad_norm": 5.574655055999756, "learning_rate": 9.460000000000001e-06, "loss": 0.0734, "step": 2701 }, { "epoch": 0.1440528869222157, "grad_norm": 5.8678879737854, "learning_rate": 9.4598e-06, "loss": 0.0644, "step": 2702 }, { "epoch": 0.14410620035186864, "grad_norm": 4.833932399749756, "learning_rate": 9.459600000000002e-06, "loss": 0.0636, "step": 2703 }, { "epoch": 0.14415951378152156, "grad_norm": 6.0644683837890625, "learning_rate": 9.459400000000001e-06, "loss": 0.0639, "step": 2704 }, { "epoch": 0.14421282721117448, "grad_norm": 3.878267288208008, "learning_rate": 9.4592e-06, "loss": 0.054, "step": 2705 }, { "epoch": 0.14426614064082743, "grad_norm": 4.815384864807129, "learning_rate": 9.459e-06, "loss": 0.0775, "step": 2706 }, { "epoch": 0.14431945407048036, "grad_norm": 4.464125156402588, "learning_rate": 9.458800000000002e-06, "loss": 0.0425, "step": 2707 }, { "epoch": 0.14437276750013328, "grad_norm": 1.7627477645874023, "learning_rate": 9.458600000000001e-06, "loss": 0.013, "step": 2708 }, { "epoch": 0.1444260809297862, "grad_norm": 6.021356105804443, "learning_rate": 9.4584e-06, "loss": 0.0384, "step": 2709 }, { "epoch": 0.14447939435943916, "grad_norm": 5.139030456542969, "learning_rate": 9.4582e-06, "loss": 0.0639, "step": 2710 }, { "epoch": 0.14453270778909208, "grad_norm": 7.470303535461426, "learning_rate": 9.458e-06, "loss": 0.0891, "step": 2711 }, { "epoch": 0.144586021218745, "grad_norm": 4.099822998046875, "learning_rate": 9.4578e-06, "loss": 0.0608, "step": 2712 }, { "epoch": 0.14463933464839793, "grad_norm": 4.790823936462402, "learning_rate": 9.4576e-06, "loss": 0.0532, "step": 2713 }, { "epoch": 0.14469264807805085, "grad_norm": 4.702938079833984, "learning_rate": 9.4574e-06, "loss": 0.0494, "step": 2714 }, { "epoch": 0.1447459615077038, "grad_norm": 5.911701679229736, "learning_rate": 9.457200000000002e-06, "loss": 0.0618, "step": 2715 }, { "epoch": 0.14479927493735673, "grad_norm": 8.767314910888672, "learning_rate": 9.457000000000001e-06, "loss": 0.0656, "step": 2716 }, { "epoch": 0.14485258836700965, "grad_norm": 5.057553768157959, "learning_rate": 9.4568e-06, "loss": 0.077, "step": 2717 }, { "epoch": 0.14490590179666257, "grad_norm": 4.730592250823975, "learning_rate": 9.4566e-06, "loss": 0.0639, "step": 2718 }, { "epoch": 0.1449592152263155, "grad_norm": 4.912346363067627, "learning_rate": 9.456400000000001e-06, "loss": 0.0685, "step": 2719 }, { "epoch": 0.14501252865596845, "grad_norm": 4.194802761077881, "learning_rate": 9.456200000000001e-06, "loss": 0.0654, "step": 2720 }, { "epoch": 0.14506584208562137, "grad_norm": 5.558197498321533, "learning_rate": 9.456e-06, "loss": 0.0963, "step": 2721 }, { "epoch": 0.1451191555152743, "grad_norm": 6.809539318084717, "learning_rate": 9.4558e-06, "loss": 0.052, "step": 2722 }, { "epoch": 0.14517246894492722, "grad_norm": 6.442348957061768, "learning_rate": 9.4556e-06, "loss": 0.0579, "step": 2723 }, { "epoch": 0.14522578237458017, "grad_norm": 3.0223684310913086, "learning_rate": 9.455400000000001e-06, "loss": 0.039, "step": 2724 }, { "epoch": 0.1452790958042331, "grad_norm": 3.562251091003418, "learning_rate": 9.4552e-06, "loss": 0.1023, "step": 2725 }, { "epoch": 0.14533240923388602, "grad_norm": 7.431546211242676, "learning_rate": 9.455e-06, "loss": 0.0759, "step": 2726 }, { "epoch": 0.14538572266353894, "grad_norm": 4.28421688079834, "learning_rate": 9.454800000000001e-06, "loss": 0.0623, "step": 2727 }, { "epoch": 0.14543903609319186, "grad_norm": 6.120974063873291, "learning_rate": 9.454600000000001e-06, "loss": 0.0553, "step": 2728 }, { "epoch": 0.14549234952284482, "grad_norm": 7.513151168823242, "learning_rate": 9.4544e-06, "loss": 0.0856, "step": 2729 }, { "epoch": 0.14554566295249774, "grad_norm": 7.4048380851745605, "learning_rate": 9.454200000000002e-06, "loss": 0.0599, "step": 2730 }, { "epoch": 0.14559897638215066, "grad_norm": 5.041337966918945, "learning_rate": 9.454000000000001e-06, "loss": 0.0629, "step": 2731 }, { "epoch": 0.1456522898118036, "grad_norm": 12.483588218688965, "learning_rate": 9.453800000000001e-06, "loss": 0.1445, "step": 2732 }, { "epoch": 0.1457056032414565, "grad_norm": 13.408435821533203, "learning_rate": 9.4536e-06, "loss": 0.1137, "step": 2733 }, { "epoch": 0.14575891667110946, "grad_norm": 4.376750946044922, "learning_rate": 9.4534e-06, "loss": 0.0943, "step": 2734 }, { "epoch": 0.14581223010076239, "grad_norm": 12.105297088623047, "learning_rate": 9.453200000000001e-06, "loss": 0.0712, "step": 2735 }, { "epoch": 0.1458655435304153, "grad_norm": 7.929429054260254, "learning_rate": 9.453e-06, "loss": 0.0803, "step": 2736 }, { "epoch": 0.14591885696006823, "grad_norm": 10.300190925598145, "learning_rate": 9.4528e-06, "loss": 0.1006, "step": 2737 }, { "epoch": 0.14597217038972118, "grad_norm": 2.938096523284912, "learning_rate": 9.4526e-06, "loss": 0.04, "step": 2738 }, { "epoch": 0.1460254838193741, "grad_norm": 8.822652816772461, "learning_rate": 9.452400000000001e-06, "loss": 0.0849, "step": 2739 }, { "epoch": 0.14607879724902703, "grad_norm": 5.364218235015869, "learning_rate": 9.4522e-06, "loss": 0.0626, "step": 2740 }, { "epoch": 0.14613211067867996, "grad_norm": 7.676126956939697, "learning_rate": 9.452000000000002e-06, "loss": 0.0712, "step": 2741 }, { "epoch": 0.14618542410833288, "grad_norm": 19.742156982421875, "learning_rate": 9.451800000000002e-06, "loss": 0.1325, "step": 2742 }, { "epoch": 0.14623873753798583, "grad_norm": 4.87999153137207, "learning_rate": 9.451600000000001e-06, "loss": 0.0555, "step": 2743 }, { "epoch": 0.14629205096763875, "grad_norm": 6.98639440536499, "learning_rate": 9.4514e-06, "loss": 0.0655, "step": 2744 }, { "epoch": 0.14634536439729168, "grad_norm": 5.029386520385742, "learning_rate": 9.4512e-06, "loss": 0.0571, "step": 2745 }, { "epoch": 0.1463986778269446, "grad_norm": 5.19363260269165, "learning_rate": 9.451000000000002e-06, "loss": 0.065, "step": 2746 }, { "epoch": 0.14645199125659752, "grad_norm": 8.254658699035645, "learning_rate": 9.450800000000001e-06, "loss": 0.1327, "step": 2747 }, { "epoch": 0.14650530468625048, "grad_norm": 13.292832374572754, "learning_rate": 9.4506e-06, "loss": 0.0647, "step": 2748 }, { "epoch": 0.1465586181159034, "grad_norm": 7.2846174240112305, "learning_rate": 9.4504e-06, "loss": 0.0674, "step": 2749 }, { "epoch": 0.14661193154555632, "grad_norm": 5.210937023162842, "learning_rate": 9.4502e-06, "loss": 0.0662, "step": 2750 }, { "epoch": 0.14666524497520925, "grad_norm": 4.7004618644714355, "learning_rate": 9.450000000000001e-06, "loss": 0.0539, "step": 2751 }, { "epoch": 0.1467185584048622, "grad_norm": 6.74406099319458, "learning_rate": 9.4498e-06, "loss": 0.0855, "step": 2752 }, { "epoch": 0.14677187183451512, "grad_norm": 4.107334613800049, "learning_rate": 9.449600000000002e-06, "loss": 0.0835, "step": 2753 }, { "epoch": 0.14682518526416805, "grad_norm": 4.916776657104492, "learning_rate": 9.449400000000001e-06, "loss": 0.0798, "step": 2754 }, { "epoch": 0.14687849869382097, "grad_norm": 5.008604049682617, "learning_rate": 9.449200000000001e-06, "loss": 0.075, "step": 2755 }, { "epoch": 0.1469318121234739, "grad_norm": 4.2650675773620605, "learning_rate": 9.449e-06, "loss": 0.0574, "step": 2756 }, { "epoch": 0.14698512555312684, "grad_norm": 4.4116692543029785, "learning_rate": 9.448800000000002e-06, "loss": 0.051, "step": 2757 }, { "epoch": 0.14703843898277977, "grad_norm": 6.327103137969971, "learning_rate": 9.448600000000001e-06, "loss": 0.0672, "step": 2758 }, { "epoch": 0.1470917524124327, "grad_norm": 3.2601664066314697, "learning_rate": 9.448400000000001e-06, "loss": 0.033, "step": 2759 }, { "epoch": 0.14714506584208562, "grad_norm": 5.413403034210205, "learning_rate": 9.4482e-06, "loss": 0.1246, "step": 2760 }, { "epoch": 0.14719837927173854, "grad_norm": 7.20847749710083, "learning_rate": 9.448e-06, "loss": 0.0676, "step": 2761 }, { "epoch": 0.1472516927013915, "grad_norm": 5.399909496307373, "learning_rate": 9.4478e-06, "loss": 0.0777, "step": 2762 }, { "epoch": 0.1473050061310444, "grad_norm": 1.9989153146743774, "learning_rate": 9.4476e-06, "loss": 0.0349, "step": 2763 }, { "epoch": 0.14735831956069734, "grad_norm": 8.961457252502441, "learning_rate": 9.4474e-06, "loss": 0.0756, "step": 2764 }, { "epoch": 0.14741163299035026, "grad_norm": 4.064888000488281, "learning_rate": 9.447200000000002e-06, "loss": 0.0722, "step": 2765 }, { "epoch": 0.1474649464200032, "grad_norm": 3.6893041133880615, "learning_rate": 9.447000000000001e-06, "loss": 0.0446, "step": 2766 }, { "epoch": 0.14751825984965614, "grad_norm": 4.578953266143799, "learning_rate": 9.4468e-06, "loss": 0.0444, "step": 2767 }, { "epoch": 0.14757157327930906, "grad_norm": 4.170060157775879, "learning_rate": 9.446600000000002e-06, "loss": 0.0885, "step": 2768 }, { "epoch": 0.14762488670896198, "grad_norm": 6.721785068511963, "learning_rate": 9.446400000000002e-06, "loss": 0.0562, "step": 2769 }, { "epoch": 0.1476782001386149, "grad_norm": 4.721146583557129, "learning_rate": 9.446200000000001e-06, "loss": 0.0836, "step": 2770 }, { "epoch": 0.14773151356826786, "grad_norm": 5.298158645629883, "learning_rate": 9.446e-06, "loss": 0.0765, "step": 2771 }, { "epoch": 0.14778482699792078, "grad_norm": 3.888401746749878, "learning_rate": 9.4458e-06, "loss": 0.0292, "step": 2772 }, { "epoch": 0.1478381404275737, "grad_norm": 3.1887314319610596, "learning_rate": 9.4456e-06, "loss": 0.0524, "step": 2773 }, { "epoch": 0.14789145385722663, "grad_norm": 5.555722713470459, "learning_rate": 9.445400000000001e-06, "loss": 0.0855, "step": 2774 }, { "epoch": 0.14794476728687955, "grad_norm": 4.679891109466553, "learning_rate": 9.4452e-06, "loss": 0.0358, "step": 2775 }, { "epoch": 0.1479980807165325, "grad_norm": 4.860391616821289, "learning_rate": 9.445e-06, "loss": 0.0889, "step": 2776 }, { "epoch": 0.14805139414618543, "grad_norm": 6.0797200202941895, "learning_rate": 9.444800000000001e-06, "loss": 0.0929, "step": 2777 }, { "epoch": 0.14810470757583835, "grad_norm": 9.63415241241455, "learning_rate": 9.444600000000001e-06, "loss": 0.0709, "step": 2778 }, { "epoch": 0.14815802100549127, "grad_norm": 7.916779041290283, "learning_rate": 9.4444e-06, "loss": 0.0509, "step": 2779 }, { "epoch": 0.14821133443514423, "grad_norm": 2.0615525245666504, "learning_rate": 9.444200000000002e-06, "loss": 0.0148, "step": 2780 }, { "epoch": 0.14826464786479715, "grad_norm": 8.805608749389648, "learning_rate": 9.444000000000001e-06, "loss": 0.1154, "step": 2781 }, { "epoch": 0.14831796129445007, "grad_norm": 10.095833778381348, "learning_rate": 9.443800000000001e-06, "loss": 0.1018, "step": 2782 }, { "epoch": 0.148371274724103, "grad_norm": 5.889451503753662, "learning_rate": 9.4436e-06, "loss": 0.0686, "step": 2783 }, { "epoch": 0.14842458815375592, "grad_norm": 3.636303424835205, "learning_rate": 9.4434e-06, "loss": 0.0428, "step": 2784 }, { "epoch": 0.14847790158340887, "grad_norm": 6.189029693603516, "learning_rate": 9.443200000000001e-06, "loss": 0.0812, "step": 2785 }, { "epoch": 0.1485312150130618, "grad_norm": 4.9097819328308105, "learning_rate": 9.443000000000001e-06, "loss": 0.0529, "step": 2786 }, { "epoch": 0.14858452844271472, "grad_norm": 6.97888708114624, "learning_rate": 9.4428e-06, "loss": 0.0591, "step": 2787 }, { "epoch": 0.14863784187236764, "grad_norm": 7.819008827209473, "learning_rate": 9.4426e-06, "loss": 0.0804, "step": 2788 }, { "epoch": 0.14869115530202057, "grad_norm": 3.664022445678711, "learning_rate": 9.442400000000001e-06, "loss": 0.0424, "step": 2789 }, { "epoch": 0.14874446873167352, "grad_norm": 3.4143736362457275, "learning_rate": 9.4422e-06, "loss": 0.0538, "step": 2790 }, { "epoch": 0.14879778216132644, "grad_norm": 6.376624584197998, "learning_rate": 9.442e-06, "loss": 0.0612, "step": 2791 }, { "epoch": 0.14885109559097937, "grad_norm": 5.950148105621338, "learning_rate": 9.441800000000002e-06, "loss": 0.0697, "step": 2792 }, { "epoch": 0.1489044090206323, "grad_norm": 7.227529048919678, "learning_rate": 9.441600000000001e-06, "loss": 0.0696, "step": 2793 }, { "epoch": 0.14895772245028524, "grad_norm": 3.4277918338775635, "learning_rate": 9.4414e-06, "loss": 0.0424, "step": 2794 }, { "epoch": 0.14901103587993816, "grad_norm": 6.559664249420166, "learning_rate": 9.4412e-06, "loss": 0.0848, "step": 2795 }, { "epoch": 0.1490643493095911, "grad_norm": 3.8672475814819336, "learning_rate": 9.441000000000002e-06, "loss": 0.0597, "step": 2796 }, { "epoch": 0.149117662739244, "grad_norm": 5.031593322753906, "learning_rate": 9.440800000000001e-06, "loss": 0.0666, "step": 2797 }, { "epoch": 0.14917097616889693, "grad_norm": 6.142082691192627, "learning_rate": 9.4406e-06, "loss": 0.072, "step": 2798 }, { "epoch": 0.1492242895985499, "grad_norm": 6.194141387939453, "learning_rate": 9.4404e-06, "loss": 0.0659, "step": 2799 }, { "epoch": 0.1492776030282028, "grad_norm": 11.865872383117676, "learning_rate": 9.4402e-06, "loss": 0.0782, "step": 2800 }, { "epoch": 0.14933091645785573, "grad_norm": 5.667440891265869, "learning_rate": 9.440000000000001e-06, "loss": 0.0404, "step": 2801 }, { "epoch": 0.14938422988750866, "grad_norm": 8.163309097290039, "learning_rate": 9.4398e-06, "loss": 0.0605, "step": 2802 }, { "epoch": 0.14943754331716158, "grad_norm": 5.6505537033081055, "learning_rate": 9.4396e-06, "loss": 0.0553, "step": 2803 }, { "epoch": 0.14949085674681453, "grad_norm": 6.945629596710205, "learning_rate": 9.439400000000001e-06, "loss": 0.0524, "step": 2804 }, { "epoch": 0.14954417017646746, "grad_norm": 7.787411689758301, "learning_rate": 9.439200000000001e-06, "loss": 0.0779, "step": 2805 }, { "epoch": 0.14959748360612038, "grad_norm": 6.3307366371154785, "learning_rate": 9.439e-06, "loss": 0.0578, "step": 2806 }, { "epoch": 0.1496507970357733, "grad_norm": 3.503530740737915, "learning_rate": 9.438800000000002e-06, "loss": 0.0439, "step": 2807 }, { "epoch": 0.14970411046542625, "grad_norm": 4.266392230987549, "learning_rate": 9.438600000000001e-06, "loss": 0.059, "step": 2808 }, { "epoch": 0.14975742389507918, "grad_norm": 10.361377716064453, "learning_rate": 9.438400000000001e-06, "loss": 0.0806, "step": 2809 }, { "epoch": 0.1498107373247321, "grad_norm": 8.439335823059082, "learning_rate": 9.4382e-06, "loss": 0.0886, "step": 2810 }, { "epoch": 0.14986405075438503, "grad_norm": 4.638167858123779, "learning_rate": 9.438e-06, "loss": 0.0593, "step": 2811 }, { "epoch": 0.14991736418403795, "grad_norm": 3.3890225887298584, "learning_rate": 9.4378e-06, "loss": 0.0545, "step": 2812 }, { "epoch": 0.1499706776136909, "grad_norm": 5.2601494789123535, "learning_rate": 9.437600000000001e-06, "loss": 0.083, "step": 2813 }, { "epoch": 0.15002399104334382, "grad_norm": 4.950198650360107, "learning_rate": 9.4374e-06, "loss": 0.0503, "step": 2814 }, { "epoch": 0.15007730447299675, "grad_norm": 5.179244518280029, "learning_rate": 9.4372e-06, "loss": 0.0561, "step": 2815 }, { "epoch": 0.15013061790264967, "grad_norm": 5.090068340301514, "learning_rate": 9.437000000000001e-06, "loss": 0.044, "step": 2816 }, { "epoch": 0.1501839313323026, "grad_norm": 5.143197536468506, "learning_rate": 9.436800000000001e-06, "loss": 0.0376, "step": 2817 }, { "epoch": 0.15023724476195555, "grad_norm": 4.120787620544434, "learning_rate": 9.436600000000002e-06, "loss": 0.0504, "step": 2818 }, { "epoch": 0.15029055819160847, "grad_norm": 4.696982383728027, "learning_rate": 9.436400000000002e-06, "loss": 0.0982, "step": 2819 }, { "epoch": 0.1503438716212614, "grad_norm": 7.407324314117432, "learning_rate": 9.436200000000001e-06, "loss": 0.0699, "step": 2820 }, { "epoch": 0.15039718505091432, "grad_norm": 14.21636962890625, "learning_rate": 9.436e-06, "loss": 0.094, "step": 2821 }, { "epoch": 0.15045049848056727, "grad_norm": 3.22304630279541, "learning_rate": 9.4358e-06, "loss": 0.0251, "step": 2822 }, { "epoch": 0.1505038119102202, "grad_norm": 4.811312198638916, "learning_rate": 9.4356e-06, "loss": 0.0517, "step": 2823 }, { "epoch": 0.15055712533987312, "grad_norm": 8.194008827209473, "learning_rate": 9.435400000000001e-06, "loss": 0.0685, "step": 2824 }, { "epoch": 0.15061043876952604, "grad_norm": 4.328136444091797, "learning_rate": 9.4352e-06, "loss": 0.0359, "step": 2825 }, { "epoch": 0.15066375219917896, "grad_norm": 9.773053169250488, "learning_rate": 9.435e-06, "loss": 0.0562, "step": 2826 }, { "epoch": 0.15071706562883191, "grad_norm": 6.4395294189453125, "learning_rate": 9.4348e-06, "loss": 0.07, "step": 2827 }, { "epoch": 0.15077037905848484, "grad_norm": 6.962773323059082, "learning_rate": 9.434600000000001e-06, "loss": 0.0774, "step": 2828 }, { "epoch": 0.15082369248813776, "grad_norm": 6.218657970428467, "learning_rate": 9.4344e-06, "loss": 0.0443, "step": 2829 }, { "epoch": 0.15087700591779069, "grad_norm": 11.439976692199707, "learning_rate": 9.434200000000002e-06, "loss": 0.0788, "step": 2830 }, { "epoch": 0.1509303193474436, "grad_norm": 8.22903823852539, "learning_rate": 9.434000000000001e-06, "loss": 0.0671, "step": 2831 }, { "epoch": 0.15098363277709656, "grad_norm": 9.449559211730957, "learning_rate": 9.433800000000001e-06, "loss": 0.0641, "step": 2832 }, { "epoch": 0.15103694620674948, "grad_norm": 6.437077522277832, "learning_rate": 9.4336e-06, "loss": 0.0648, "step": 2833 }, { "epoch": 0.1510902596364024, "grad_norm": 5.783801555633545, "learning_rate": 9.4334e-06, "loss": 0.0552, "step": 2834 }, { "epoch": 0.15114357306605533, "grad_norm": 5.780570983886719, "learning_rate": 9.433200000000001e-06, "loss": 0.0398, "step": 2835 }, { "epoch": 0.15119688649570828, "grad_norm": 5.741219997406006, "learning_rate": 9.433000000000001e-06, "loss": 0.0482, "step": 2836 }, { "epoch": 0.1512501999253612, "grad_norm": 5.406022548675537, "learning_rate": 9.4328e-06, "loss": 0.045, "step": 2837 }, { "epoch": 0.15130351335501413, "grad_norm": 8.692877769470215, "learning_rate": 9.4326e-06, "loss": 0.0381, "step": 2838 }, { "epoch": 0.15135682678466705, "grad_norm": 4.319143295288086, "learning_rate": 9.4324e-06, "loss": 0.0434, "step": 2839 }, { "epoch": 0.15141014021431998, "grad_norm": 4.851274013519287, "learning_rate": 9.432200000000001e-06, "loss": 0.0595, "step": 2840 }, { "epoch": 0.15146345364397293, "grad_norm": 8.97496223449707, "learning_rate": 9.432e-06, "loss": 0.0817, "step": 2841 }, { "epoch": 0.15151676707362585, "grad_norm": 8.802092552185059, "learning_rate": 9.431800000000002e-06, "loss": 0.1454, "step": 2842 }, { "epoch": 0.15157008050327878, "grad_norm": 4.8312883377075195, "learning_rate": 9.431600000000001e-06, "loss": 0.0456, "step": 2843 }, { "epoch": 0.1516233939329317, "grad_norm": 3.9706108570098877, "learning_rate": 9.431400000000001e-06, "loss": 0.0353, "step": 2844 }, { "epoch": 0.15167670736258462, "grad_norm": 7.041614055633545, "learning_rate": 9.4312e-06, "loss": 0.0527, "step": 2845 }, { "epoch": 0.15173002079223757, "grad_norm": 7.6234660148620605, "learning_rate": 9.431000000000002e-06, "loss": 0.0743, "step": 2846 }, { "epoch": 0.1517833342218905, "grad_norm": 5.166258811950684, "learning_rate": 9.430800000000001e-06, "loss": 0.0612, "step": 2847 }, { "epoch": 0.15183664765154342, "grad_norm": 4.623849868774414, "learning_rate": 9.4306e-06, "loss": 0.0766, "step": 2848 }, { "epoch": 0.15188996108119635, "grad_norm": 5.815498352050781, "learning_rate": 9.4304e-06, "loss": 0.0445, "step": 2849 }, { "epoch": 0.1519432745108493, "grad_norm": 6.092007160186768, "learning_rate": 9.4302e-06, "loss": 0.0601, "step": 2850 }, { "epoch": 0.15199658794050222, "grad_norm": 6.2241058349609375, "learning_rate": 9.43e-06, "loss": 0.0823, "step": 2851 }, { "epoch": 0.15204990137015514, "grad_norm": 8.041537284851074, "learning_rate": 9.4298e-06, "loss": 0.0975, "step": 2852 }, { "epoch": 0.15210321479980807, "grad_norm": 6.571162223815918, "learning_rate": 9.4296e-06, "loss": 0.0804, "step": 2853 }, { "epoch": 0.152156528229461, "grad_norm": 5.513949394226074, "learning_rate": 9.429400000000002e-06, "loss": 0.0973, "step": 2854 }, { "epoch": 0.15220984165911394, "grad_norm": 6.228139400482178, "learning_rate": 9.429200000000001e-06, "loss": 0.0789, "step": 2855 }, { "epoch": 0.15226315508876687, "grad_norm": 4.864426136016846, "learning_rate": 9.429e-06, "loss": 0.0866, "step": 2856 }, { "epoch": 0.1523164685184198, "grad_norm": 7.688011646270752, "learning_rate": 9.428800000000002e-06, "loss": 0.061, "step": 2857 }, { "epoch": 0.1523697819480727, "grad_norm": 5.737064838409424, "learning_rate": 9.428600000000002e-06, "loss": 0.0718, "step": 2858 }, { "epoch": 0.15242309537772564, "grad_norm": 3.443568229675293, "learning_rate": 9.428400000000001e-06, "loss": 0.0517, "step": 2859 }, { "epoch": 0.1524764088073786, "grad_norm": 9.842276573181152, "learning_rate": 9.4282e-06, "loss": 0.1032, "step": 2860 }, { "epoch": 0.1525297222370315, "grad_norm": 4.372702598571777, "learning_rate": 9.428e-06, "loss": 0.0489, "step": 2861 }, { "epoch": 0.15258303566668444, "grad_norm": 9.045947074890137, "learning_rate": 9.4278e-06, "loss": 0.0816, "step": 2862 }, { "epoch": 0.15263634909633736, "grad_norm": 4.331857204437256, "learning_rate": 9.427600000000001e-06, "loss": 0.0466, "step": 2863 }, { "epoch": 0.1526896625259903, "grad_norm": 4.925385475158691, "learning_rate": 9.4274e-06, "loss": 0.09, "step": 2864 }, { "epoch": 0.15274297595564323, "grad_norm": 6.796831130981445, "learning_rate": 9.4272e-06, "loss": 0.1106, "step": 2865 }, { "epoch": 0.15279628938529616, "grad_norm": 6.952811241149902, "learning_rate": 9.427000000000001e-06, "loss": 0.0721, "step": 2866 }, { "epoch": 0.15284960281494908, "grad_norm": 6.655405044555664, "learning_rate": 9.426800000000001e-06, "loss": 0.0622, "step": 2867 }, { "epoch": 0.152902916244602, "grad_norm": 3.8772408962249756, "learning_rate": 9.4266e-06, "loss": 0.0454, "step": 2868 }, { "epoch": 0.15295622967425496, "grad_norm": 4.34799861907959, "learning_rate": 9.426400000000002e-06, "loss": 0.0396, "step": 2869 }, { "epoch": 0.15300954310390788, "grad_norm": 3.8169875144958496, "learning_rate": 9.426200000000001e-06, "loss": 0.055, "step": 2870 }, { "epoch": 0.1530628565335608, "grad_norm": 4.655226230621338, "learning_rate": 9.426000000000001e-06, "loss": 0.0591, "step": 2871 }, { "epoch": 0.15311616996321373, "grad_norm": 6.737799644470215, "learning_rate": 9.4258e-06, "loss": 0.0569, "step": 2872 }, { "epoch": 0.15316948339286665, "grad_norm": 4.036988735198975, "learning_rate": 9.4256e-06, "loss": 0.0357, "step": 2873 }, { "epoch": 0.1532227968225196, "grad_norm": 4.3672685623168945, "learning_rate": 9.425400000000001e-06, "loss": 0.0517, "step": 2874 }, { "epoch": 0.15327611025217253, "grad_norm": 5.318243503570557, "learning_rate": 9.4252e-06, "loss": 0.0966, "step": 2875 }, { "epoch": 0.15332942368182545, "grad_norm": 3.739140510559082, "learning_rate": 9.425e-06, "loss": 0.0505, "step": 2876 }, { "epoch": 0.15338273711147837, "grad_norm": 7.332160949707031, "learning_rate": 9.4248e-06, "loss": 0.0893, "step": 2877 }, { "epoch": 0.15343605054113132, "grad_norm": 5.917294502258301, "learning_rate": 9.424600000000001e-06, "loss": 0.0507, "step": 2878 }, { "epoch": 0.15348936397078425, "grad_norm": 6.514030933380127, "learning_rate": 9.4244e-06, "loss": 0.0762, "step": 2879 }, { "epoch": 0.15354267740043717, "grad_norm": 2.928903102874756, "learning_rate": 9.4242e-06, "loss": 0.0339, "step": 2880 }, { "epoch": 0.1535959908300901, "grad_norm": 3.9528274536132812, "learning_rate": 9.424000000000002e-06, "loss": 0.0583, "step": 2881 }, { "epoch": 0.15364930425974302, "grad_norm": 5.503884315490723, "learning_rate": 9.423800000000001e-06, "loss": 0.0687, "step": 2882 }, { "epoch": 0.15370261768939597, "grad_norm": 3.4028666019439697, "learning_rate": 9.4236e-06, "loss": 0.0516, "step": 2883 }, { "epoch": 0.1537559311190489, "grad_norm": 6.619061470031738, "learning_rate": 9.4234e-06, "loss": 0.0748, "step": 2884 }, { "epoch": 0.15380924454870182, "grad_norm": 5.414682388305664, "learning_rate": 9.423200000000002e-06, "loss": 0.0757, "step": 2885 }, { "epoch": 0.15386255797835474, "grad_norm": 5.770467281341553, "learning_rate": 9.423000000000001e-06, "loss": 0.0574, "step": 2886 }, { "epoch": 0.15391587140800767, "grad_norm": 4.078252792358398, "learning_rate": 9.4228e-06, "loss": 0.0697, "step": 2887 }, { "epoch": 0.15396918483766062, "grad_norm": 4.596065044403076, "learning_rate": 9.4226e-06, "loss": 0.0669, "step": 2888 }, { "epoch": 0.15402249826731354, "grad_norm": 5.567879676818848, "learning_rate": 9.4224e-06, "loss": 0.0563, "step": 2889 }, { "epoch": 0.15407581169696646, "grad_norm": 6.073062419891357, "learning_rate": 9.422200000000001e-06, "loss": 0.0628, "step": 2890 }, { "epoch": 0.1541291251266194, "grad_norm": 6.8943963050842285, "learning_rate": 9.422e-06, "loss": 0.0804, "step": 2891 }, { "epoch": 0.15418243855627234, "grad_norm": 5.083795070648193, "learning_rate": 9.4218e-06, "loss": 0.0982, "step": 2892 }, { "epoch": 0.15423575198592526, "grad_norm": 6.587345123291016, "learning_rate": 9.421600000000001e-06, "loss": 0.0759, "step": 2893 }, { "epoch": 0.15428906541557819, "grad_norm": 4.160222053527832, "learning_rate": 9.421400000000001e-06, "loss": 0.0724, "step": 2894 }, { "epoch": 0.1543423788452311, "grad_norm": 2.639996290206909, "learning_rate": 9.4212e-06, "loss": 0.0459, "step": 2895 }, { "epoch": 0.15439569227488403, "grad_norm": 5.4834418296813965, "learning_rate": 9.421000000000002e-06, "loss": 0.0528, "step": 2896 }, { "epoch": 0.15444900570453698, "grad_norm": 5.15147590637207, "learning_rate": 9.420800000000001e-06, "loss": 0.074, "step": 2897 }, { "epoch": 0.1545023191341899, "grad_norm": 7.492550373077393, "learning_rate": 9.420600000000001e-06, "loss": 0.0703, "step": 2898 }, { "epoch": 0.15455563256384283, "grad_norm": 4.980106830596924, "learning_rate": 9.4204e-06, "loss": 0.0529, "step": 2899 }, { "epoch": 0.15460894599349576, "grad_norm": 5.422240734100342, "learning_rate": 9.4202e-06, "loss": 0.0814, "step": 2900 }, { "epoch": 0.15466225942314868, "grad_norm": 5.720141887664795, "learning_rate": 9.42e-06, "loss": 0.089, "step": 2901 }, { "epoch": 0.15471557285280163, "grad_norm": 4.867556095123291, "learning_rate": 9.419800000000001e-06, "loss": 0.0573, "step": 2902 }, { "epoch": 0.15476888628245455, "grad_norm": 4.843560695648193, "learning_rate": 9.4196e-06, "loss": 0.0744, "step": 2903 }, { "epoch": 0.15482219971210748, "grad_norm": 3.654561758041382, "learning_rate": 9.4194e-06, "loss": 0.0827, "step": 2904 }, { "epoch": 0.1548755131417604, "grad_norm": 3.5775513648986816, "learning_rate": 9.419200000000001e-06, "loss": 0.0494, "step": 2905 }, { "epoch": 0.15492882657141335, "grad_norm": 4.785416603088379, "learning_rate": 9.419e-06, "loss": 0.0516, "step": 2906 }, { "epoch": 0.15498214000106628, "grad_norm": 2.968437433242798, "learning_rate": 9.418800000000002e-06, "loss": 0.0648, "step": 2907 }, { "epoch": 0.1550354534307192, "grad_norm": 3.940521478652954, "learning_rate": 9.418600000000002e-06, "loss": 0.0438, "step": 2908 }, { "epoch": 0.15508876686037212, "grad_norm": 2.984675884246826, "learning_rate": 9.418400000000001e-06, "loss": 0.0406, "step": 2909 }, { "epoch": 0.15514208029002505, "grad_norm": 3.7214903831481934, "learning_rate": 9.4182e-06, "loss": 0.0616, "step": 2910 }, { "epoch": 0.155195393719678, "grad_norm": 4.919248580932617, "learning_rate": 9.418e-06, "loss": 0.0903, "step": 2911 }, { "epoch": 0.15524870714933092, "grad_norm": 3.5949885845184326, "learning_rate": 9.4178e-06, "loss": 0.0375, "step": 2912 }, { "epoch": 0.15530202057898385, "grad_norm": 6.0716447830200195, "learning_rate": 9.417600000000001e-06, "loss": 0.0517, "step": 2913 }, { "epoch": 0.15535533400863677, "grad_norm": 4.178379535675049, "learning_rate": 9.4174e-06, "loss": 0.0458, "step": 2914 }, { "epoch": 0.1554086474382897, "grad_norm": 3.3685836791992188, "learning_rate": 9.4172e-06, "loss": 0.0397, "step": 2915 }, { "epoch": 0.15546196086794264, "grad_norm": 5.6163225173950195, "learning_rate": 9.417e-06, "loss": 0.0337, "step": 2916 }, { "epoch": 0.15551527429759557, "grad_norm": 5.681313514709473, "learning_rate": 9.416800000000001e-06, "loss": 0.0599, "step": 2917 }, { "epoch": 0.1555685877272485, "grad_norm": 5.017063140869141, "learning_rate": 9.4166e-06, "loss": 0.0808, "step": 2918 }, { "epoch": 0.15562190115690142, "grad_norm": 4.459233283996582, "learning_rate": 9.416400000000002e-06, "loss": 0.0547, "step": 2919 }, { "epoch": 0.15567521458655437, "grad_norm": 4.8990583419799805, "learning_rate": 9.416200000000001e-06, "loss": 0.0835, "step": 2920 }, { "epoch": 0.1557285280162073, "grad_norm": 2.3104426860809326, "learning_rate": 9.416000000000001e-06, "loss": 0.0279, "step": 2921 }, { "epoch": 0.15578184144586021, "grad_norm": 5.230677127838135, "learning_rate": 9.4158e-06, "loss": 0.062, "step": 2922 }, { "epoch": 0.15583515487551314, "grad_norm": 5.443619251251221, "learning_rate": 9.4156e-06, "loss": 0.0722, "step": 2923 }, { "epoch": 0.15588846830516606, "grad_norm": 11.876770973205566, "learning_rate": 9.415400000000001e-06, "loss": 0.084, "step": 2924 }, { "epoch": 0.155941781734819, "grad_norm": 3.506492853164673, "learning_rate": 9.415200000000001e-06, "loss": 0.0549, "step": 2925 }, { "epoch": 0.15599509516447194, "grad_norm": 4.463400840759277, "learning_rate": 9.415e-06, "loss": 0.053, "step": 2926 }, { "epoch": 0.15604840859412486, "grad_norm": 5.749514579772949, "learning_rate": 9.4148e-06, "loss": 0.0726, "step": 2927 }, { "epoch": 0.15610172202377778, "grad_norm": 3.83614444732666, "learning_rate": 9.4146e-06, "loss": 0.0451, "step": 2928 }, { "epoch": 0.1561550354534307, "grad_norm": 4.772350788116455, "learning_rate": 9.414400000000001e-06, "loss": 0.0883, "step": 2929 }, { "epoch": 0.15620834888308366, "grad_norm": 6.316212177276611, "learning_rate": 9.4142e-06, "loss": 0.1437, "step": 2930 }, { "epoch": 0.15626166231273658, "grad_norm": 5.717012405395508, "learning_rate": 9.414000000000002e-06, "loss": 0.0557, "step": 2931 }, { "epoch": 0.1563149757423895, "grad_norm": 4.15798807144165, "learning_rate": 9.413800000000001e-06, "loss": 0.0651, "step": 2932 }, { "epoch": 0.15636828917204243, "grad_norm": 2.7733113765716553, "learning_rate": 9.4136e-06, "loss": 0.0362, "step": 2933 }, { "epoch": 0.15642160260169538, "grad_norm": 2.9619123935699463, "learning_rate": 9.4134e-06, "loss": 0.0567, "step": 2934 }, { "epoch": 0.1564749160313483, "grad_norm": 6.867477893829346, "learning_rate": 9.413200000000002e-06, "loss": 0.0523, "step": 2935 }, { "epoch": 0.15652822946100123, "grad_norm": 3.7852296829223633, "learning_rate": 9.413000000000001e-06, "loss": 0.0636, "step": 2936 }, { "epoch": 0.15658154289065415, "grad_norm": 3.8077597618103027, "learning_rate": 9.4128e-06, "loss": 0.0628, "step": 2937 }, { "epoch": 0.15663485632030708, "grad_norm": 7.517792701721191, "learning_rate": 9.4126e-06, "loss": 0.0807, "step": 2938 }, { "epoch": 0.15668816974996003, "grad_norm": 5.9268059730529785, "learning_rate": 9.4124e-06, "loss": 0.0817, "step": 2939 }, { "epoch": 0.15674148317961295, "grad_norm": 6.497961044311523, "learning_rate": 9.4122e-06, "loss": 0.1366, "step": 2940 }, { "epoch": 0.15679479660926587, "grad_norm": 9.820429801940918, "learning_rate": 9.412e-06, "loss": 0.0655, "step": 2941 }, { "epoch": 0.1568481100389188, "grad_norm": 4.68546199798584, "learning_rate": 9.4118e-06, "loss": 0.0755, "step": 2942 }, { "epoch": 0.15690142346857172, "grad_norm": 7.492652416229248, "learning_rate": 9.411600000000002e-06, "loss": 0.0652, "step": 2943 }, { "epoch": 0.15695473689822467, "grad_norm": 4.532747745513916, "learning_rate": 9.411400000000001e-06, "loss": 0.0766, "step": 2944 }, { "epoch": 0.1570080503278776, "grad_norm": 10.6109619140625, "learning_rate": 9.4112e-06, "loss": 0.0701, "step": 2945 }, { "epoch": 0.15706136375753052, "grad_norm": 3.0305705070495605, "learning_rate": 9.411000000000002e-06, "loss": 0.0228, "step": 2946 }, { "epoch": 0.15711467718718344, "grad_norm": 5.612420558929443, "learning_rate": 9.410800000000001e-06, "loss": 0.049, "step": 2947 }, { "epoch": 0.1571679906168364, "grad_norm": 3.67867112159729, "learning_rate": 9.410600000000001e-06, "loss": 0.0598, "step": 2948 }, { "epoch": 0.15722130404648932, "grad_norm": 6.28702974319458, "learning_rate": 9.4104e-06, "loss": 0.0719, "step": 2949 }, { "epoch": 0.15727461747614224, "grad_norm": 5.297260284423828, "learning_rate": 9.4102e-06, "loss": 0.0421, "step": 2950 }, { "epoch": 0.15732793090579517, "grad_norm": 8.275559425354004, "learning_rate": 9.41e-06, "loss": 0.0607, "step": 2951 }, { "epoch": 0.1573812443354481, "grad_norm": 7.054596424102783, "learning_rate": 9.409800000000001e-06, "loss": 0.0867, "step": 2952 }, { "epoch": 0.15743455776510104, "grad_norm": 5.489792823791504, "learning_rate": 9.4096e-06, "loss": 0.0336, "step": 2953 }, { "epoch": 0.15748787119475396, "grad_norm": 4.112180233001709, "learning_rate": 9.4094e-06, "loss": 0.0611, "step": 2954 }, { "epoch": 0.1575411846244069, "grad_norm": 2.722827196121216, "learning_rate": 9.409200000000001e-06, "loss": 0.0347, "step": 2955 }, { "epoch": 0.1575944980540598, "grad_norm": 7.3304972648620605, "learning_rate": 9.409000000000001e-06, "loss": 0.0968, "step": 2956 }, { "epoch": 0.15764781148371274, "grad_norm": 4.689087390899658, "learning_rate": 9.4088e-06, "loss": 0.0608, "step": 2957 }, { "epoch": 0.1577011249133657, "grad_norm": 6.3346734046936035, "learning_rate": 9.408600000000002e-06, "loss": 0.0765, "step": 2958 }, { "epoch": 0.1577544383430186, "grad_norm": 2.375246524810791, "learning_rate": 9.408400000000001e-06, "loss": 0.0158, "step": 2959 }, { "epoch": 0.15780775177267153, "grad_norm": 8.435893058776855, "learning_rate": 9.4082e-06, "loss": 0.1077, "step": 2960 }, { "epoch": 0.15786106520232446, "grad_norm": 3.169236660003662, "learning_rate": 9.408e-06, "loss": 0.0397, "step": 2961 }, { "epoch": 0.15791437863197738, "grad_norm": 7.508915424346924, "learning_rate": 9.4078e-06, "loss": 0.0857, "step": 2962 }, { "epoch": 0.15796769206163033, "grad_norm": 10.426424026489258, "learning_rate": 9.407600000000001e-06, "loss": 0.0877, "step": 2963 }, { "epoch": 0.15802100549128326, "grad_norm": 6.302271842956543, "learning_rate": 9.4074e-06, "loss": 0.0558, "step": 2964 }, { "epoch": 0.15807431892093618, "grad_norm": 4.858465671539307, "learning_rate": 9.4072e-06, "loss": 0.0683, "step": 2965 }, { "epoch": 0.1581276323505891, "grad_norm": 2.736876964569092, "learning_rate": 9.407e-06, "loss": 0.0256, "step": 2966 }, { "epoch": 0.15818094578024205, "grad_norm": 5.942720890045166, "learning_rate": 9.406800000000001e-06, "loss": 0.0935, "step": 2967 }, { "epoch": 0.15823425920989498, "grad_norm": 5.627699375152588, "learning_rate": 9.4066e-06, "loss": 0.0607, "step": 2968 }, { "epoch": 0.1582875726395479, "grad_norm": 2.7366855144500732, "learning_rate": 9.4064e-06, "loss": 0.0566, "step": 2969 }, { "epoch": 0.15834088606920083, "grad_norm": 3.531477451324463, "learning_rate": 9.406200000000002e-06, "loss": 0.0397, "step": 2970 }, { "epoch": 0.15839419949885375, "grad_norm": 3.2326278686523438, "learning_rate": 9.406000000000001e-06, "loss": 0.0519, "step": 2971 }, { "epoch": 0.1584475129285067, "grad_norm": 4.534372329711914, "learning_rate": 9.4058e-06, "loss": 0.0597, "step": 2972 }, { "epoch": 0.15850082635815962, "grad_norm": 3.847489595413208, "learning_rate": 9.4056e-06, "loss": 0.0284, "step": 2973 }, { "epoch": 0.15855413978781255, "grad_norm": 3.213167190551758, "learning_rate": 9.405400000000001e-06, "loss": 0.0546, "step": 2974 }, { "epoch": 0.15860745321746547, "grad_norm": 4.412073135375977, "learning_rate": 9.405200000000001e-06, "loss": 0.0513, "step": 2975 }, { "epoch": 0.1586607666471184, "grad_norm": 5.955959320068359, "learning_rate": 9.405e-06, "loss": 0.0983, "step": 2976 }, { "epoch": 0.15871408007677135, "grad_norm": 9.330915451049805, "learning_rate": 9.4048e-06, "loss": 0.108, "step": 2977 }, { "epoch": 0.15876739350642427, "grad_norm": 6.483301162719727, "learning_rate": 9.4046e-06, "loss": 0.0571, "step": 2978 }, { "epoch": 0.1588207069360772, "grad_norm": 6.906846523284912, "learning_rate": 9.404400000000001e-06, "loss": 0.0636, "step": 2979 }, { "epoch": 0.15887402036573012, "grad_norm": 8.10815715789795, "learning_rate": 9.4042e-06, "loss": 0.0572, "step": 2980 }, { "epoch": 0.15892733379538307, "grad_norm": 4.355531692504883, "learning_rate": 9.404e-06, "loss": 0.0651, "step": 2981 }, { "epoch": 0.158980647225036, "grad_norm": 4.640595436096191, "learning_rate": 9.403800000000001e-06, "loss": 0.0689, "step": 2982 }, { "epoch": 0.15903396065468892, "grad_norm": 7.771763801574707, "learning_rate": 9.403600000000001e-06, "loss": 0.0741, "step": 2983 }, { "epoch": 0.15908727408434184, "grad_norm": 4.242342472076416, "learning_rate": 9.4034e-06, "loss": 0.0551, "step": 2984 }, { "epoch": 0.15914058751399476, "grad_norm": 5.131932258605957, "learning_rate": 9.403200000000002e-06, "loss": 0.0631, "step": 2985 }, { "epoch": 0.15919390094364771, "grad_norm": 3.7450058460235596, "learning_rate": 9.403000000000001e-06, "loss": 0.0599, "step": 2986 }, { "epoch": 0.15924721437330064, "grad_norm": 4.3213958740234375, "learning_rate": 9.4028e-06, "loss": 0.0747, "step": 2987 }, { "epoch": 0.15930052780295356, "grad_norm": 2.438004970550537, "learning_rate": 9.4026e-06, "loss": 0.0237, "step": 2988 }, { "epoch": 0.15935384123260649, "grad_norm": 4.108456611633301, "learning_rate": 9.4024e-06, "loss": 0.0695, "step": 2989 }, { "epoch": 0.1594071546622594, "grad_norm": 3.5719058513641357, "learning_rate": 9.4022e-06, "loss": 0.0548, "step": 2990 }, { "epoch": 0.15946046809191236, "grad_norm": 5.308999538421631, "learning_rate": 9.402e-06, "loss": 0.0766, "step": 2991 }, { "epoch": 0.15951378152156528, "grad_norm": 3.78385591506958, "learning_rate": 9.4018e-06, "loss": 0.0555, "step": 2992 }, { "epoch": 0.1595670949512182, "grad_norm": 3.6168549060821533, "learning_rate": 9.4016e-06, "loss": 0.0488, "step": 2993 }, { "epoch": 0.15962040838087113, "grad_norm": 3.3473403453826904, "learning_rate": 9.401400000000001e-06, "loss": 0.0663, "step": 2994 }, { "epoch": 0.15967372181052408, "grad_norm": 7.6847710609436035, "learning_rate": 9.4012e-06, "loss": 0.1187, "step": 2995 }, { "epoch": 0.159727035240177, "grad_norm": 3.7268214225769043, "learning_rate": 9.401000000000002e-06, "loss": 0.061, "step": 2996 }, { "epoch": 0.15978034866982993, "grad_norm": 3.00578236579895, "learning_rate": 9.400800000000002e-06, "loss": 0.0482, "step": 2997 }, { "epoch": 0.15983366209948285, "grad_norm": 4.366797924041748, "learning_rate": 9.400600000000001e-06, "loss": 0.0763, "step": 2998 }, { "epoch": 0.15988697552913578, "grad_norm": 5.094300746917725, "learning_rate": 9.4004e-06, "loss": 0.0694, "step": 2999 }, { "epoch": 0.15994028895878873, "grad_norm": 5.3519816398620605, "learning_rate": 9.4002e-06, "loss": 0.0683, "step": 3000 }, { "epoch": 0.15999360238844165, "grad_norm": 5.019913196563721, "learning_rate": 9.4e-06, "loss": 0.0582, "step": 3001 }, { "epoch": 0.16004691581809458, "grad_norm": 5.338108539581299, "learning_rate": 9.399800000000001e-06, "loss": 0.0467, "step": 3002 }, { "epoch": 0.1601002292477475, "grad_norm": 6.1398491859436035, "learning_rate": 9.3996e-06, "loss": 0.075, "step": 3003 }, { "epoch": 0.16015354267740042, "grad_norm": 3.812051296234131, "learning_rate": 9.3994e-06, "loss": 0.0615, "step": 3004 }, { "epoch": 0.16020685610705337, "grad_norm": 6.570279598236084, "learning_rate": 9.3992e-06, "loss": 0.0718, "step": 3005 }, { "epoch": 0.1602601695367063, "grad_norm": 4.632203102111816, "learning_rate": 9.399000000000001e-06, "loss": 0.0611, "step": 3006 }, { "epoch": 0.16031348296635922, "grad_norm": 4.487908840179443, "learning_rate": 9.3988e-06, "loss": 0.0576, "step": 3007 }, { "epoch": 0.16036679639601215, "grad_norm": 14.73709774017334, "learning_rate": 9.398600000000002e-06, "loss": 0.0956, "step": 3008 }, { "epoch": 0.1604201098256651, "grad_norm": 11.15608024597168, "learning_rate": 9.398400000000001e-06, "loss": 0.0716, "step": 3009 }, { "epoch": 0.16047342325531802, "grad_norm": 4.532166481018066, "learning_rate": 9.398200000000001e-06, "loss": 0.0771, "step": 3010 }, { "epoch": 0.16052673668497094, "grad_norm": 6.086196422576904, "learning_rate": 9.398e-06, "loss": 0.0808, "step": 3011 }, { "epoch": 0.16058005011462387, "grad_norm": 10.481849670410156, "learning_rate": 9.3978e-06, "loss": 0.0987, "step": 3012 }, { "epoch": 0.1606333635442768, "grad_norm": 8.400382995605469, "learning_rate": 9.397600000000001e-06, "loss": 0.1011, "step": 3013 }, { "epoch": 0.16068667697392974, "grad_norm": 3.8719944953918457, "learning_rate": 9.397400000000001e-06, "loss": 0.047, "step": 3014 }, { "epoch": 0.16073999040358267, "grad_norm": 4.52311897277832, "learning_rate": 9.3972e-06, "loss": 0.0692, "step": 3015 }, { "epoch": 0.1607933038332356, "grad_norm": 5.868561267852783, "learning_rate": 9.397e-06, "loss": 0.0689, "step": 3016 }, { "epoch": 0.1608466172628885, "grad_norm": 6.038958549499512, "learning_rate": 9.3968e-06, "loss": 0.0557, "step": 3017 }, { "epoch": 0.16089993069254144, "grad_norm": 10.232202529907227, "learning_rate": 9.3966e-06, "loss": 0.0887, "step": 3018 }, { "epoch": 0.1609532441221944, "grad_norm": 5.643657207489014, "learning_rate": 9.3964e-06, "loss": 0.0383, "step": 3019 }, { "epoch": 0.1610065575518473, "grad_norm": 4.7559027671813965, "learning_rate": 9.396200000000002e-06, "loss": 0.086, "step": 3020 }, { "epoch": 0.16105987098150024, "grad_norm": 3.2593517303466797, "learning_rate": 9.396000000000001e-06, "loss": 0.061, "step": 3021 }, { "epoch": 0.16111318441115316, "grad_norm": 4.523990631103516, "learning_rate": 9.3958e-06, "loss": 0.0657, "step": 3022 }, { "epoch": 0.1611664978408061, "grad_norm": 8.309072494506836, "learning_rate": 9.3956e-06, "loss": 0.1408, "step": 3023 }, { "epoch": 0.16121981127045903, "grad_norm": 7.7797698974609375, "learning_rate": 9.395400000000002e-06, "loss": 0.0907, "step": 3024 }, { "epoch": 0.16127312470011196, "grad_norm": 4.853447437286377, "learning_rate": 9.395200000000001e-06, "loss": 0.0686, "step": 3025 }, { "epoch": 0.16132643812976488, "grad_norm": 4.863089084625244, "learning_rate": 9.395e-06, "loss": 0.0654, "step": 3026 }, { "epoch": 0.1613797515594178, "grad_norm": 5.968469619750977, "learning_rate": 9.3948e-06, "loss": 0.0555, "step": 3027 }, { "epoch": 0.16143306498907076, "grad_norm": 4.489852428436279, "learning_rate": 9.3946e-06, "loss": 0.0346, "step": 3028 }, { "epoch": 0.16148637841872368, "grad_norm": 5.907983303070068, "learning_rate": 9.394400000000001e-06, "loss": 0.0723, "step": 3029 }, { "epoch": 0.1615396918483766, "grad_norm": 6.353515148162842, "learning_rate": 9.3942e-06, "loss": 0.0717, "step": 3030 }, { "epoch": 0.16159300527802953, "grad_norm": 3.3740413188934326, "learning_rate": 9.394e-06, "loss": 0.0642, "step": 3031 }, { "epoch": 0.16164631870768245, "grad_norm": 3.5130722522735596, "learning_rate": 9.393800000000001e-06, "loss": 0.0563, "step": 3032 }, { "epoch": 0.1616996321373354, "grad_norm": 2.990960121154785, "learning_rate": 9.393600000000001e-06, "loss": 0.0303, "step": 3033 }, { "epoch": 0.16175294556698833, "grad_norm": 3.981581449508667, "learning_rate": 9.3934e-06, "loss": 0.0596, "step": 3034 }, { "epoch": 0.16180625899664125, "grad_norm": 2.563014030456543, "learning_rate": 9.393200000000002e-06, "loss": 0.0425, "step": 3035 }, { "epoch": 0.16185957242629417, "grad_norm": 6.146775245666504, "learning_rate": 9.393000000000001e-06, "loss": 0.0906, "step": 3036 }, { "epoch": 0.16191288585594713, "grad_norm": 3.807898759841919, "learning_rate": 9.392800000000001e-06, "loss": 0.0547, "step": 3037 }, { "epoch": 0.16196619928560005, "grad_norm": 4.349599838256836, "learning_rate": 9.3926e-06, "loss": 0.0666, "step": 3038 }, { "epoch": 0.16201951271525297, "grad_norm": 3.724963426589966, "learning_rate": 9.3924e-06, "loss": 0.0422, "step": 3039 }, { "epoch": 0.1620728261449059, "grad_norm": 6.06813383102417, "learning_rate": 9.392200000000001e-06, "loss": 0.0812, "step": 3040 }, { "epoch": 0.16212613957455882, "grad_norm": 2.976289749145508, "learning_rate": 9.392000000000001e-06, "loss": 0.0728, "step": 3041 }, { "epoch": 0.16217945300421177, "grad_norm": 6.562687397003174, "learning_rate": 9.3918e-06, "loss": 0.0578, "step": 3042 }, { "epoch": 0.1622327664338647, "grad_norm": 4.405706405639648, "learning_rate": 9.3916e-06, "loss": 0.0512, "step": 3043 }, { "epoch": 0.16228607986351762, "grad_norm": 4.132866382598877, "learning_rate": 9.391400000000001e-06, "loss": 0.0862, "step": 3044 }, { "epoch": 0.16233939329317054, "grad_norm": 3.2612192630767822, "learning_rate": 9.3912e-06, "loss": 0.0542, "step": 3045 }, { "epoch": 0.16239270672282347, "grad_norm": 7.207956790924072, "learning_rate": 9.391e-06, "loss": 0.0757, "step": 3046 }, { "epoch": 0.16244602015247642, "grad_norm": 4.918108940124512, "learning_rate": 9.390800000000002e-06, "loss": 0.0747, "step": 3047 }, { "epoch": 0.16249933358212934, "grad_norm": 2.497318983078003, "learning_rate": 9.390600000000001e-06, "loss": 0.0578, "step": 3048 }, { "epoch": 0.16255264701178226, "grad_norm": 3.143688440322876, "learning_rate": 9.3904e-06, "loss": 0.0536, "step": 3049 }, { "epoch": 0.1626059604414352, "grad_norm": 4.091588020324707, "learning_rate": 9.3902e-06, "loss": 0.0469, "step": 3050 }, { "epoch": 0.16265927387108814, "grad_norm": 4.124161243438721, "learning_rate": 9.39e-06, "loss": 0.046, "step": 3051 }, { "epoch": 0.16271258730074106, "grad_norm": 3.816958427429199, "learning_rate": 9.389800000000001e-06, "loss": 0.0536, "step": 3052 }, { "epoch": 0.162765900730394, "grad_norm": 3.043745279312134, "learning_rate": 9.3896e-06, "loss": 0.0474, "step": 3053 }, { "epoch": 0.1628192141600469, "grad_norm": 3.408560276031494, "learning_rate": 9.3894e-06, "loss": 0.0572, "step": 3054 }, { "epoch": 0.16287252758969983, "grad_norm": 3.9740638732910156, "learning_rate": 9.3892e-06, "loss": 0.0629, "step": 3055 }, { "epoch": 0.16292584101935279, "grad_norm": 3.934079647064209, "learning_rate": 9.389000000000001e-06, "loss": 0.0406, "step": 3056 }, { "epoch": 0.1629791544490057, "grad_norm": 2.533613681793213, "learning_rate": 9.3888e-06, "loss": 0.0295, "step": 3057 }, { "epoch": 0.16303246787865863, "grad_norm": 4.014244556427002, "learning_rate": 9.3886e-06, "loss": 0.0597, "step": 3058 }, { "epoch": 0.16308578130831156, "grad_norm": 4.190340042114258, "learning_rate": 9.388400000000001e-06, "loss": 0.075, "step": 3059 }, { "epoch": 0.16313909473796448, "grad_norm": 6.202320098876953, "learning_rate": 9.388200000000001e-06, "loss": 0.0822, "step": 3060 }, { "epoch": 0.16319240816761743, "grad_norm": 4.1718244552612305, "learning_rate": 9.388e-06, "loss": 0.0688, "step": 3061 }, { "epoch": 0.16324572159727035, "grad_norm": 3.545893907546997, "learning_rate": 9.3878e-06, "loss": 0.0343, "step": 3062 }, { "epoch": 0.16329903502692328, "grad_norm": 5.553671360015869, "learning_rate": 9.387600000000001e-06, "loss": 0.0768, "step": 3063 }, { "epoch": 0.1633523484565762, "grad_norm": 6.684047698974609, "learning_rate": 9.387400000000001e-06, "loss": 0.08, "step": 3064 }, { "epoch": 0.16340566188622915, "grad_norm": 6.022050857543945, "learning_rate": 9.3872e-06, "loss": 0.0815, "step": 3065 }, { "epoch": 0.16345897531588208, "grad_norm": 4.683667182922363, "learning_rate": 9.387e-06, "loss": 0.0852, "step": 3066 }, { "epoch": 0.163512288745535, "grad_norm": 10.174221992492676, "learning_rate": 9.3868e-06, "loss": 0.106, "step": 3067 }, { "epoch": 0.16356560217518792, "grad_norm": 6.829997539520264, "learning_rate": 9.386600000000001e-06, "loss": 0.1061, "step": 3068 }, { "epoch": 0.16361891560484085, "grad_norm": 3.8656933307647705, "learning_rate": 9.3864e-06, "loss": 0.0442, "step": 3069 }, { "epoch": 0.1636722290344938, "grad_norm": 6.323308944702148, "learning_rate": 9.3862e-06, "loss": 0.087, "step": 3070 }, { "epoch": 0.16372554246414672, "grad_norm": 4.4489426612854, "learning_rate": 9.386000000000001e-06, "loss": 0.0436, "step": 3071 }, { "epoch": 0.16377885589379965, "grad_norm": 3.0149734020233154, "learning_rate": 9.3858e-06, "loss": 0.0312, "step": 3072 }, { "epoch": 0.16383216932345257, "grad_norm": 5.109731674194336, "learning_rate": 9.3856e-06, "loss": 0.0757, "step": 3073 }, { "epoch": 0.1638854827531055, "grad_norm": 3.038268566131592, "learning_rate": 9.385400000000002e-06, "loss": 0.057, "step": 3074 }, { "epoch": 0.16393879618275845, "grad_norm": 5.376433372497559, "learning_rate": 9.385200000000001e-06, "loss": 0.0875, "step": 3075 }, { "epoch": 0.16399210961241137, "grad_norm": 6.017337322235107, "learning_rate": 9.385e-06, "loss": 0.0679, "step": 3076 }, { "epoch": 0.1640454230420643, "grad_norm": 4.094588279724121, "learning_rate": 9.3848e-06, "loss": 0.0454, "step": 3077 }, { "epoch": 0.16409873647171722, "grad_norm": 5.508284091949463, "learning_rate": 9.3846e-06, "loss": 0.0712, "step": 3078 }, { "epoch": 0.16415204990137017, "grad_norm": 6.9985032081604, "learning_rate": 9.384400000000001e-06, "loss": 0.0599, "step": 3079 }, { "epoch": 0.1642053633310231, "grad_norm": 4.667842388153076, "learning_rate": 9.3842e-06, "loss": 0.0665, "step": 3080 }, { "epoch": 0.16425867676067601, "grad_norm": 3.8015949726104736, "learning_rate": 9.384e-06, "loss": 0.059, "step": 3081 }, { "epoch": 0.16431199019032894, "grad_norm": 5.978672027587891, "learning_rate": 9.3838e-06, "loss": 0.0949, "step": 3082 }, { "epoch": 0.16436530361998186, "grad_norm": 3.454212188720703, "learning_rate": 9.383600000000001e-06, "loss": 0.0378, "step": 3083 }, { "epoch": 0.1644186170496348, "grad_norm": 4.15841817855835, "learning_rate": 9.3834e-06, "loss": 0.064, "step": 3084 }, { "epoch": 0.16447193047928774, "grad_norm": 5.302603244781494, "learning_rate": 9.383200000000002e-06, "loss": 0.0862, "step": 3085 }, { "epoch": 0.16452524390894066, "grad_norm": 5.9816718101501465, "learning_rate": 9.383000000000001e-06, "loss": 0.0728, "step": 3086 }, { "epoch": 0.16457855733859358, "grad_norm": 4.57951545715332, "learning_rate": 9.382800000000001e-06, "loss": 0.0603, "step": 3087 }, { "epoch": 0.1646318707682465, "grad_norm": 3.58044695854187, "learning_rate": 9.3826e-06, "loss": 0.042, "step": 3088 }, { "epoch": 0.16468518419789946, "grad_norm": 4.4961442947387695, "learning_rate": 9.3824e-06, "loss": 0.062, "step": 3089 }, { "epoch": 0.16473849762755238, "grad_norm": 5.381777763366699, "learning_rate": 9.382200000000001e-06, "loss": 0.1087, "step": 3090 }, { "epoch": 0.1647918110572053, "grad_norm": 5.355443477630615, "learning_rate": 9.382000000000001e-06, "loss": 0.1041, "step": 3091 }, { "epoch": 0.16484512448685823, "grad_norm": 5.3655171394348145, "learning_rate": 9.3818e-06, "loss": 0.0442, "step": 3092 }, { "epoch": 0.16489843791651118, "grad_norm": 3.6949634552001953, "learning_rate": 9.3816e-06, "loss": 0.0403, "step": 3093 }, { "epoch": 0.1649517513461641, "grad_norm": 3.482149124145508, "learning_rate": 9.381400000000001e-06, "loss": 0.0732, "step": 3094 }, { "epoch": 0.16500506477581703, "grad_norm": 3.4663643836975098, "learning_rate": 9.381200000000001e-06, "loss": 0.066, "step": 3095 }, { "epoch": 0.16505837820546995, "grad_norm": 3.7171688079833984, "learning_rate": 9.381e-06, "loss": 0.0531, "step": 3096 }, { "epoch": 0.16511169163512288, "grad_norm": 3.107775926589966, "learning_rate": 9.380800000000002e-06, "loss": 0.0281, "step": 3097 }, { "epoch": 0.16516500506477583, "grad_norm": 5.1805644035339355, "learning_rate": 9.380600000000001e-06, "loss": 0.059, "step": 3098 }, { "epoch": 0.16521831849442875, "grad_norm": 4.2788615226745605, "learning_rate": 9.380400000000001e-06, "loss": 0.0477, "step": 3099 }, { "epoch": 0.16527163192408167, "grad_norm": 6.619399070739746, "learning_rate": 9.3802e-06, "loss": 0.1114, "step": 3100 }, { "epoch": 0.1653249453537346, "grad_norm": 5.554587364196777, "learning_rate": 9.38e-06, "loss": 0.046, "step": 3101 }, { "epoch": 0.16537825878338752, "grad_norm": 3.5189192295074463, "learning_rate": 9.379800000000001e-06, "loss": 0.0634, "step": 3102 }, { "epoch": 0.16543157221304047, "grad_norm": 5.259426593780518, "learning_rate": 9.3796e-06, "loss": 0.0383, "step": 3103 }, { "epoch": 0.1654848856426934, "grad_norm": 5.020683288574219, "learning_rate": 9.3794e-06, "loss": 0.0508, "step": 3104 }, { "epoch": 0.16553819907234632, "grad_norm": 7.316666126251221, "learning_rate": 9.3792e-06, "loss": 0.0804, "step": 3105 }, { "epoch": 0.16559151250199924, "grad_norm": 5.981879234313965, "learning_rate": 9.379000000000001e-06, "loss": 0.1017, "step": 3106 }, { "epoch": 0.1656448259316522, "grad_norm": 5.135242462158203, "learning_rate": 9.3788e-06, "loss": 0.0538, "step": 3107 }, { "epoch": 0.16569813936130512, "grad_norm": 3.70548152923584, "learning_rate": 9.3786e-06, "loss": 0.0474, "step": 3108 }, { "epoch": 0.16575145279095804, "grad_norm": 4.420739650726318, "learning_rate": 9.378400000000002e-06, "loss": 0.0527, "step": 3109 }, { "epoch": 0.16580476622061097, "grad_norm": 4.04107141494751, "learning_rate": 9.378200000000001e-06, "loss": 0.0454, "step": 3110 }, { "epoch": 0.1658580796502639, "grad_norm": 4.615190505981445, "learning_rate": 9.378e-06, "loss": 0.0444, "step": 3111 }, { "epoch": 0.16591139307991684, "grad_norm": 4.91542387008667, "learning_rate": 9.3778e-06, "loss": 0.0948, "step": 3112 }, { "epoch": 0.16596470650956976, "grad_norm": 6.4368133544921875, "learning_rate": 9.377600000000001e-06, "loss": 0.077, "step": 3113 }, { "epoch": 0.1660180199392227, "grad_norm": 5.131424903869629, "learning_rate": 9.377400000000001e-06, "loss": 0.0599, "step": 3114 }, { "epoch": 0.1660713333688756, "grad_norm": 6.25350284576416, "learning_rate": 9.3772e-06, "loss": 0.0881, "step": 3115 }, { "epoch": 0.16612464679852854, "grad_norm": 4.706294536590576, "learning_rate": 9.377e-06, "loss": 0.0621, "step": 3116 }, { "epoch": 0.1661779602281815, "grad_norm": 3.2017154693603516, "learning_rate": 9.3768e-06, "loss": 0.0454, "step": 3117 }, { "epoch": 0.1662312736578344, "grad_norm": 7.089242935180664, "learning_rate": 9.376600000000001e-06, "loss": 0.0807, "step": 3118 }, { "epoch": 0.16628458708748733, "grad_norm": 4.33846378326416, "learning_rate": 9.3764e-06, "loss": 0.0652, "step": 3119 }, { "epoch": 0.16633790051714026, "grad_norm": 6.0840911865234375, "learning_rate": 9.3762e-06, "loss": 0.0452, "step": 3120 }, { "epoch": 0.1663912139467932, "grad_norm": 4.67572021484375, "learning_rate": 9.376000000000001e-06, "loss": 0.0823, "step": 3121 }, { "epoch": 0.16644452737644613, "grad_norm": 3.529067039489746, "learning_rate": 9.375800000000001e-06, "loss": 0.054, "step": 3122 }, { "epoch": 0.16649784080609906, "grad_norm": 4.0757598876953125, "learning_rate": 9.3756e-06, "loss": 0.0603, "step": 3123 }, { "epoch": 0.16655115423575198, "grad_norm": 2.842533826828003, "learning_rate": 9.375400000000002e-06, "loss": 0.0535, "step": 3124 }, { "epoch": 0.1666044676654049, "grad_norm": 4.0754899978637695, "learning_rate": 9.375200000000001e-06, "loss": 0.0712, "step": 3125 }, { "epoch": 0.16665778109505786, "grad_norm": 3.90552020072937, "learning_rate": 9.375000000000001e-06, "loss": 0.0567, "step": 3126 }, { "epoch": 0.16671109452471078, "grad_norm": 3.2882163524627686, "learning_rate": 9.3748e-06, "loss": 0.0605, "step": 3127 }, { "epoch": 0.1667644079543637, "grad_norm": 6.018831253051758, "learning_rate": 9.3746e-06, "loss": 0.0807, "step": 3128 }, { "epoch": 0.16681772138401663, "grad_norm": 8.803914070129395, "learning_rate": 9.374400000000001e-06, "loss": 0.0761, "step": 3129 }, { "epoch": 0.16687103481366955, "grad_norm": 2.93315052986145, "learning_rate": 9.3742e-06, "loss": 0.0249, "step": 3130 }, { "epoch": 0.1669243482433225, "grad_norm": 6.007487773895264, "learning_rate": 9.374e-06, "loss": 0.084, "step": 3131 }, { "epoch": 0.16697766167297542, "grad_norm": 8.392476081848145, "learning_rate": 9.3738e-06, "loss": 0.0714, "step": 3132 }, { "epoch": 0.16703097510262835, "grad_norm": 5.223421573638916, "learning_rate": 9.373600000000001e-06, "loss": 0.1079, "step": 3133 }, { "epoch": 0.16708428853228127, "grad_norm": 5.930133819580078, "learning_rate": 9.3734e-06, "loss": 0.0355, "step": 3134 }, { "epoch": 0.16713760196193422, "grad_norm": 6.331439971923828, "learning_rate": 9.373200000000002e-06, "loss": 0.0596, "step": 3135 }, { "epoch": 0.16719091539158715, "grad_norm": 2.7104806900024414, "learning_rate": 9.373000000000002e-06, "loss": 0.0334, "step": 3136 }, { "epoch": 0.16724422882124007, "grad_norm": 1.8987324237823486, "learning_rate": 9.372800000000001e-06, "loss": 0.0303, "step": 3137 }, { "epoch": 0.167297542250893, "grad_norm": 4.347182273864746, "learning_rate": 9.3726e-06, "loss": 0.0538, "step": 3138 }, { "epoch": 0.16735085568054592, "grad_norm": 11.40890884399414, "learning_rate": 9.3724e-06, "loss": 0.0673, "step": 3139 }, { "epoch": 0.16740416911019887, "grad_norm": 8.55290412902832, "learning_rate": 9.372200000000002e-06, "loss": 0.0816, "step": 3140 }, { "epoch": 0.1674574825398518, "grad_norm": 3.1462512016296387, "learning_rate": 9.372000000000001e-06, "loss": 0.0507, "step": 3141 }, { "epoch": 0.16751079596950472, "grad_norm": 4.902497291564941, "learning_rate": 9.3718e-06, "loss": 0.0709, "step": 3142 }, { "epoch": 0.16756410939915764, "grad_norm": 6.907037258148193, "learning_rate": 9.3716e-06, "loss": 0.0529, "step": 3143 }, { "epoch": 0.16761742282881056, "grad_norm": 7.407793998718262, "learning_rate": 9.3714e-06, "loss": 0.0767, "step": 3144 }, { "epoch": 0.16767073625846352, "grad_norm": 10.749975204467773, "learning_rate": 9.371200000000001e-06, "loss": 0.0804, "step": 3145 }, { "epoch": 0.16772404968811644, "grad_norm": 5.421565055847168, "learning_rate": 9.371e-06, "loss": 0.0781, "step": 3146 }, { "epoch": 0.16777736311776936, "grad_norm": 4.720961570739746, "learning_rate": 9.370800000000002e-06, "loss": 0.0633, "step": 3147 }, { "epoch": 0.1678306765474223, "grad_norm": 10.661517143249512, "learning_rate": 9.370600000000001e-06, "loss": 0.147, "step": 3148 }, { "epoch": 0.16788398997707524, "grad_norm": 16.403812408447266, "learning_rate": 9.370400000000001e-06, "loss": 0.1129, "step": 3149 }, { "epoch": 0.16793730340672816, "grad_norm": 11.795564651489258, "learning_rate": 9.3702e-06, "loss": 0.0951, "step": 3150 }, { "epoch": 0.16799061683638108, "grad_norm": 3.01601505279541, "learning_rate": 9.370000000000002e-06, "loss": 0.0358, "step": 3151 }, { "epoch": 0.168043930266034, "grad_norm": 7.760706901550293, "learning_rate": 9.369800000000001e-06, "loss": 0.1324, "step": 3152 }, { "epoch": 0.16809724369568693, "grad_norm": 10.705961227416992, "learning_rate": 9.369600000000001e-06, "loss": 0.0981, "step": 3153 }, { "epoch": 0.16815055712533988, "grad_norm": 7.204439163208008, "learning_rate": 9.3694e-06, "loss": 0.0877, "step": 3154 }, { "epoch": 0.1682038705549928, "grad_norm": 6.202849388122559, "learning_rate": 9.3692e-06, "loss": 0.0679, "step": 3155 }, { "epoch": 0.16825718398464573, "grad_norm": 4.305963039398193, "learning_rate": 9.369e-06, "loss": 0.0695, "step": 3156 }, { "epoch": 0.16831049741429865, "grad_norm": 7.218565940856934, "learning_rate": 9.3688e-06, "loss": 0.0536, "step": 3157 }, { "epoch": 0.16836381084395158, "grad_norm": 6.6476006507873535, "learning_rate": 9.3686e-06, "loss": 0.0666, "step": 3158 }, { "epoch": 0.16841712427360453, "grad_norm": 3.0557541847229004, "learning_rate": 9.368400000000002e-06, "loss": 0.0405, "step": 3159 }, { "epoch": 0.16847043770325745, "grad_norm": 6.668039321899414, "learning_rate": 9.368200000000001e-06, "loss": 0.0624, "step": 3160 }, { "epoch": 0.16852375113291038, "grad_norm": 4.657038688659668, "learning_rate": 9.368e-06, "loss": 0.0677, "step": 3161 }, { "epoch": 0.1685770645625633, "grad_norm": 4.815465450286865, "learning_rate": 9.3678e-06, "loss": 0.0611, "step": 3162 }, { "epoch": 0.16863037799221625, "grad_norm": 6.153390884399414, "learning_rate": 9.367600000000002e-06, "loss": 0.079, "step": 3163 }, { "epoch": 0.16868369142186918, "grad_norm": 5.133151531219482, "learning_rate": 9.367400000000001e-06, "loss": 0.0497, "step": 3164 }, { "epoch": 0.1687370048515221, "grad_norm": 2.5040228366851807, "learning_rate": 9.3672e-06, "loss": 0.0308, "step": 3165 }, { "epoch": 0.16879031828117502, "grad_norm": 5.355856418609619, "learning_rate": 9.367e-06, "loss": 0.0868, "step": 3166 }, { "epoch": 0.16884363171082795, "grad_norm": 5.70566463470459, "learning_rate": 9.3668e-06, "loss": 0.0338, "step": 3167 }, { "epoch": 0.1688969451404809, "grad_norm": 7.4953179359436035, "learning_rate": 9.366600000000001e-06, "loss": 0.058, "step": 3168 }, { "epoch": 0.16895025857013382, "grad_norm": 4.7359938621521, "learning_rate": 9.3664e-06, "loss": 0.0622, "step": 3169 }, { "epoch": 0.16900357199978674, "grad_norm": 5.898888111114502, "learning_rate": 9.3662e-06, "loss": 0.0547, "step": 3170 }, { "epoch": 0.16905688542943967, "grad_norm": 3.2327613830566406, "learning_rate": 9.366000000000001e-06, "loss": 0.0494, "step": 3171 }, { "epoch": 0.1691101988590926, "grad_norm": 3.8971030712127686, "learning_rate": 9.365800000000001e-06, "loss": 0.0728, "step": 3172 }, { "epoch": 0.16916351228874554, "grad_norm": 2.676077127456665, "learning_rate": 9.3656e-06, "loss": 0.0255, "step": 3173 }, { "epoch": 0.16921682571839847, "grad_norm": 4.452286243438721, "learning_rate": 9.365400000000002e-06, "loss": 0.0635, "step": 3174 }, { "epoch": 0.1692701391480514, "grad_norm": 6.69461727142334, "learning_rate": 9.365200000000001e-06, "loss": 0.0582, "step": 3175 }, { "epoch": 0.16932345257770431, "grad_norm": 3.297454595565796, "learning_rate": 9.365000000000001e-06, "loss": 0.0445, "step": 3176 }, { "epoch": 0.16937676600735727, "grad_norm": 5.597596168518066, "learning_rate": 9.3648e-06, "loss": 0.0561, "step": 3177 }, { "epoch": 0.1694300794370102, "grad_norm": 5.103105545043945, "learning_rate": 9.3646e-06, "loss": 0.0696, "step": 3178 }, { "epoch": 0.1694833928666631, "grad_norm": 4.323822498321533, "learning_rate": 9.364400000000001e-06, "loss": 0.0711, "step": 3179 }, { "epoch": 0.16953670629631604, "grad_norm": 2.577958345413208, "learning_rate": 9.364200000000001e-06, "loss": 0.0325, "step": 3180 }, { "epoch": 0.16959001972596896, "grad_norm": 4.642903804779053, "learning_rate": 9.364e-06, "loss": 0.0396, "step": 3181 }, { "epoch": 0.1696433331556219, "grad_norm": 5.375071048736572, "learning_rate": 9.3638e-06, "loss": 0.0702, "step": 3182 }, { "epoch": 0.16969664658527484, "grad_norm": 6.289055824279785, "learning_rate": 9.363600000000001e-06, "loss": 0.0773, "step": 3183 }, { "epoch": 0.16974996001492776, "grad_norm": 3.4730384349823, "learning_rate": 9.363400000000001e-06, "loss": 0.0183, "step": 3184 }, { "epoch": 0.16980327344458068, "grad_norm": 5.368181228637695, "learning_rate": 9.3632e-06, "loss": 0.0908, "step": 3185 }, { "epoch": 0.1698565868742336, "grad_norm": 7.145420074462891, "learning_rate": 9.363000000000002e-06, "loss": 0.1054, "step": 3186 }, { "epoch": 0.16990990030388656, "grad_norm": 3.0311808586120605, "learning_rate": 9.362800000000001e-06, "loss": 0.051, "step": 3187 }, { "epoch": 0.16996321373353948, "grad_norm": 4.075430870056152, "learning_rate": 9.3626e-06, "loss": 0.077, "step": 3188 }, { "epoch": 0.1700165271631924, "grad_norm": 6.426331043243408, "learning_rate": 9.3624e-06, "loss": 0.0595, "step": 3189 }, { "epoch": 0.17006984059284533, "grad_norm": 5.855385780334473, "learning_rate": 9.362200000000002e-06, "loss": 0.0668, "step": 3190 }, { "epoch": 0.17012315402249828, "grad_norm": 3.0624096393585205, "learning_rate": 9.362000000000001e-06, "loss": 0.0396, "step": 3191 }, { "epoch": 0.1701764674521512, "grad_norm": 3.7934741973876953, "learning_rate": 9.3618e-06, "loss": 0.0555, "step": 3192 }, { "epoch": 0.17022978088180413, "grad_norm": 4.338766098022461, "learning_rate": 9.3616e-06, "loss": 0.0477, "step": 3193 }, { "epoch": 0.17028309431145705, "grad_norm": 4.055190563201904, "learning_rate": 9.3614e-06, "loss": 0.0509, "step": 3194 }, { "epoch": 0.17033640774110997, "grad_norm": 5.850281715393066, "learning_rate": 9.361200000000001e-06, "loss": 0.0475, "step": 3195 }, { "epoch": 0.17038972117076293, "grad_norm": 3.366203546524048, "learning_rate": 9.361e-06, "loss": 0.0363, "step": 3196 }, { "epoch": 0.17044303460041585, "grad_norm": 4.493179798126221, "learning_rate": 9.3608e-06, "loss": 0.0614, "step": 3197 }, { "epoch": 0.17049634803006877, "grad_norm": 4.747391700744629, "learning_rate": 9.360600000000001e-06, "loss": 0.0656, "step": 3198 }, { "epoch": 0.1705496614597217, "grad_norm": 3.5865304470062256, "learning_rate": 9.360400000000001e-06, "loss": 0.023, "step": 3199 }, { "epoch": 0.17060297488937462, "grad_norm": 4.692368507385254, "learning_rate": 9.3602e-06, "loss": 0.0648, "step": 3200 }, { "epoch": 0.17065628831902757, "grad_norm": 3.913331985473633, "learning_rate": 9.360000000000002e-06, "loss": 0.072, "step": 3201 }, { "epoch": 0.1707096017486805, "grad_norm": 5.400357723236084, "learning_rate": 9.359800000000001e-06, "loss": 0.0794, "step": 3202 }, { "epoch": 0.17076291517833342, "grad_norm": 2.765411853790283, "learning_rate": 9.359600000000001e-06, "loss": 0.0253, "step": 3203 }, { "epoch": 0.17081622860798634, "grad_norm": 5.019941806793213, "learning_rate": 9.3594e-06, "loss": 0.0723, "step": 3204 }, { "epoch": 0.1708695420376393, "grad_norm": 3.9229331016540527, "learning_rate": 9.3592e-06, "loss": 0.0499, "step": 3205 }, { "epoch": 0.17092285546729222, "grad_norm": 2.8898074626922607, "learning_rate": 9.359e-06, "loss": 0.0274, "step": 3206 }, { "epoch": 0.17097616889694514, "grad_norm": 4.828232288360596, "learning_rate": 9.358800000000001e-06, "loss": 0.0554, "step": 3207 }, { "epoch": 0.17102948232659806, "grad_norm": 5.546638011932373, "learning_rate": 9.3586e-06, "loss": 0.0543, "step": 3208 }, { "epoch": 0.171082795756251, "grad_norm": 2.9433975219726562, "learning_rate": 9.3584e-06, "loss": 0.0471, "step": 3209 }, { "epoch": 0.17113610918590394, "grad_norm": 8.94835090637207, "learning_rate": 9.358200000000001e-06, "loss": 0.0761, "step": 3210 }, { "epoch": 0.17118942261555686, "grad_norm": 7.738071441650391, "learning_rate": 9.358000000000001e-06, "loss": 0.0635, "step": 3211 }, { "epoch": 0.1712427360452098, "grad_norm": 5.166869640350342, "learning_rate": 9.3578e-06, "loss": 0.1156, "step": 3212 }, { "epoch": 0.1712960494748627, "grad_norm": 5.7327704429626465, "learning_rate": 9.357600000000002e-06, "loss": 0.077, "step": 3213 }, { "epoch": 0.17134936290451563, "grad_norm": 5.733272552490234, "learning_rate": 9.357400000000001e-06, "loss": 0.0758, "step": 3214 }, { "epoch": 0.17140267633416859, "grad_norm": 7.423706531524658, "learning_rate": 9.3572e-06, "loss": 0.0819, "step": 3215 }, { "epoch": 0.1714559897638215, "grad_norm": 3.2275304794311523, "learning_rate": 9.357e-06, "loss": 0.0318, "step": 3216 }, { "epoch": 0.17150930319347443, "grad_norm": 8.051230430603027, "learning_rate": 9.3568e-06, "loss": 0.0808, "step": 3217 }, { "epoch": 0.17156261662312736, "grad_norm": 14.026009559631348, "learning_rate": 9.356600000000001e-06, "loss": 0.1352, "step": 3218 }, { "epoch": 0.1716159300527803, "grad_norm": 4.850341796875, "learning_rate": 9.3564e-06, "loss": 0.0571, "step": 3219 }, { "epoch": 0.17166924348243323, "grad_norm": 6.911867618560791, "learning_rate": 9.3562e-06, "loss": 0.0664, "step": 3220 }, { "epoch": 0.17172255691208616, "grad_norm": 4.714668273925781, "learning_rate": 9.356e-06, "loss": 0.0599, "step": 3221 }, { "epoch": 0.17177587034173908, "grad_norm": 5.596923351287842, "learning_rate": 9.355800000000001e-06, "loss": 0.0777, "step": 3222 }, { "epoch": 0.171829183771392, "grad_norm": 9.637683868408203, "learning_rate": 9.3556e-06, "loss": 0.0866, "step": 3223 }, { "epoch": 0.17188249720104495, "grad_norm": 3.266140937805176, "learning_rate": 9.355400000000002e-06, "loss": 0.0487, "step": 3224 }, { "epoch": 0.17193581063069788, "grad_norm": 2.6922597885131836, "learning_rate": 9.355200000000002e-06, "loss": 0.0411, "step": 3225 }, { "epoch": 0.1719891240603508, "grad_norm": 3.112750291824341, "learning_rate": 9.355000000000001e-06, "loss": 0.0413, "step": 3226 }, { "epoch": 0.17204243749000372, "grad_norm": 4.453759670257568, "learning_rate": 9.3548e-06, "loss": 0.0708, "step": 3227 }, { "epoch": 0.17209575091965665, "grad_norm": 3.2285168170928955, "learning_rate": 9.3546e-06, "loss": 0.0376, "step": 3228 }, { "epoch": 0.1721490643493096, "grad_norm": 4.55444860458374, "learning_rate": 9.354400000000001e-06, "loss": 0.0805, "step": 3229 }, { "epoch": 0.17220237777896252, "grad_norm": 4.15724515914917, "learning_rate": 9.354200000000001e-06, "loss": 0.0558, "step": 3230 }, { "epoch": 0.17225569120861545, "grad_norm": 4.987331390380859, "learning_rate": 9.354e-06, "loss": 0.0551, "step": 3231 }, { "epoch": 0.17230900463826837, "grad_norm": 7.618710994720459, "learning_rate": 9.3538e-06, "loss": 0.1128, "step": 3232 }, { "epoch": 0.17236231806792132, "grad_norm": 2.3773086071014404, "learning_rate": 9.3536e-06, "loss": 0.0225, "step": 3233 }, { "epoch": 0.17241563149757425, "grad_norm": 4.523777008056641, "learning_rate": 9.353400000000001e-06, "loss": 0.0574, "step": 3234 }, { "epoch": 0.17246894492722717, "grad_norm": 4.241950988769531, "learning_rate": 9.3532e-06, "loss": 0.0433, "step": 3235 }, { "epoch": 0.1725222583568801, "grad_norm": 5.6548261642456055, "learning_rate": 9.353000000000002e-06, "loss": 0.0444, "step": 3236 }, { "epoch": 0.17257557178653302, "grad_norm": 3.2355260848999023, "learning_rate": 9.352800000000001e-06, "loss": 0.055, "step": 3237 }, { "epoch": 0.17262888521618597, "grad_norm": 4.157188892364502, "learning_rate": 9.352600000000001e-06, "loss": 0.0554, "step": 3238 }, { "epoch": 0.1726821986458389, "grad_norm": 4.323822975158691, "learning_rate": 9.3524e-06, "loss": 0.0395, "step": 3239 }, { "epoch": 0.17273551207549181, "grad_norm": 2.6124017238616943, "learning_rate": 9.352200000000002e-06, "loss": 0.0383, "step": 3240 }, { "epoch": 0.17278882550514474, "grad_norm": 4.152798175811768, "learning_rate": 9.352000000000001e-06, "loss": 0.0503, "step": 3241 }, { "epoch": 0.17284213893479766, "grad_norm": 3.192692518234253, "learning_rate": 9.3518e-06, "loss": 0.0464, "step": 3242 }, { "epoch": 0.1728954523644506, "grad_norm": 4.3439483642578125, "learning_rate": 9.3516e-06, "loss": 0.0548, "step": 3243 }, { "epoch": 0.17294876579410354, "grad_norm": 5.72455358505249, "learning_rate": 9.3514e-06, "loss": 0.0535, "step": 3244 }, { "epoch": 0.17300207922375646, "grad_norm": 5.583871364593506, "learning_rate": 9.3512e-06, "loss": 0.0662, "step": 3245 }, { "epoch": 0.17305539265340938, "grad_norm": 4.630636215209961, "learning_rate": 9.351e-06, "loss": 0.0564, "step": 3246 }, { "epoch": 0.17310870608306234, "grad_norm": 3.855680465698242, "learning_rate": 9.3508e-06, "loss": 0.0352, "step": 3247 }, { "epoch": 0.17316201951271526, "grad_norm": 4.928130149841309, "learning_rate": 9.350600000000002e-06, "loss": 0.0874, "step": 3248 }, { "epoch": 0.17321533294236818, "grad_norm": 5.252429962158203, "learning_rate": 9.350400000000001e-06, "loss": 0.0546, "step": 3249 }, { "epoch": 0.1732686463720211, "grad_norm": 5.875580310821533, "learning_rate": 9.3502e-06, "loss": 0.0714, "step": 3250 }, { "epoch": 0.17332195980167403, "grad_norm": 5.284673690795898, "learning_rate": 9.350000000000002e-06, "loss": 0.0482, "step": 3251 }, { "epoch": 0.17337527323132698, "grad_norm": 5.2548604011535645, "learning_rate": 9.349800000000002e-06, "loss": 0.0585, "step": 3252 }, { "epoch": 0.1734285866609799, "grad_norm": 6.336286544799805, "learning_rate": 9.349600000000001e-06, "loss": 0.0677, "step": 3253 }, { "epoch": 0.17348190009063283, "grad_norm": 1.9910762310028076, "learning_rate": 9.3494e-06, "loss": 0.0258, "step": 3254 }, { "epoch": 0.17353521352028575, "grad_norm": 4.263818264007568, "learning_rate": 9.3492e-06, "loss": 0.0384, "step": 3255 }, { "epoch": 0.17358852694993868, "grad_norm": 3.585859775543213, "learning_rate": 9.349e-06, "loss": 0.0459, "step": 3256 }, { "epoch": 0.17364184037959163, "grad_norm": 4.262221336364746, "learning_rate": 9.348800000000001e-06, "loss": 0.0613, "step": 3257 }, { "epoch": 0.17369515380924455, "grad_norm": 5.645179748535156, "learning_rate": 9.3486e-06, "loss": 0.0809, "step": 3258 }, { "epoch": 0.17374846723889747, "grad_norm": 6.128369331359863, "learning_rate": 9.3484e-06, "loss": 0.0957, "step": 3259 }, { "epoch": 0.1738017806685504, "grad_norm": 4.983671188354492, "learning_rate": 9.348200000000001e-06, "loss": 0.0944, "step": 3260 }, { "epoch": 0.17385509409820335, "grad_norm": 4.178412914276123, "learning_rate": 9.348000000000001e-06, "loss": 0.0462, "step": 3261 }, { "epoch": 0.17390840752785627, "grad_norm": 4.386481761932373, "learning_rate": 9.3478e-06, "loss": 0.0753, "step": 3262 }, { "epoch": 0.1739617209575092, "grad_norm": 6.007906913757324, "learning_rate": 9.347600000000002e-06, "loss": 0.0464, "step": 3263 }, { "epoch": 0.17401503438716212, "grad_norm": 9.402705192565918, "learning_rate": 9.347400000000001e-06, "loss": 0.0877, "step": 3264 }, { "epoch": 0.17406834781681504, "grad_norm": 3.806426763534546, "learning_rate": 9.347200000000001e-06, "loss": 0.0312, "step": 3265 }, { "epoch": 0.174121661246468, "grad_norm": 3.9747986793518066, "learning_rate": 9.347e-06, "loss": 0.0475, "step": 3266 }, { "epoch": 0.17417497467612092, "grad_norm": 8.824905395507812, "learning_rate": 9.3468e-06, "loss": 0.0934, "step": 3267 }, { "epoch": 0.17422828810577384, "grad_norm": 8.98042106628418, "learning_rate": 9.346600000000001e-06, "loss": 0.0805, "step": 3268 }, { "epoch": 0.17428160153542677, "grad_norm": 3.2357358932495117, "learning_rate": 9.3464e-06, "loss": 0.0473, "step": 3269 }, { "epoch": 0.1743349149650797, "grad_norm": 2.697659492492676, "learning_rate": 9.3462e-06, "loss": 0.0221, "step": 3270 }, { "epoch": 0.17438822839473264, "grad_norm": 5.091615676879883, "learning_rate": 9.346e-06, "loss": 0.0438, "step": 3271 }, { "epoch": 0.17444154182438557, "grad_norm": 11.210750579833984, "learning_rate": 9.345800000000001e-06, "loss": 0.0983, "step": 3272 }, { "epoch": 0.1744948552540385, "grad_norm": 8.388339042663574, "learning_rate": 9.3456e-06, "loss": 0.0761, "step": 3273 }, { "epoch": 0.1745481686836914, "grad_norm": 4.320592403411865, "learning_rate": 9.3454e-06, "loss": 0.0567, "step": 3274 }, { "epoch": 0.17460148211334436, "grad_norm": 4.265121936798096, "learning_rate": 9.345200000000002e-06, "loss": 0.0504, "step": 3275 }, { "epoch": 0.1746547955429973, "grad_norm": 4.687986850738525, "learning_rate": 9.345000000000001e-06, "loss": 0.062, "step": 3276 }, { "epoch": 0.1747081089726502, "grad_norm": 10.77160358428955, "learning_rate": 9.3448e-06, "loss": 0.0823, "step": 3277 }, { "epoch": 0.17476142240230313, "grad_norm": 8.061090469360352, "learning_rate": 9.3446e-06, "loss": 0.0591, "step": 3278 }, { "epoch": 0.17481473583195606, "grad_norm": 4.835594654083252, "learning_rate": 9.344400000000002e-06, "loss": 0.0665, "step": 3279 }, { "epoch": 0.174868049261609, "grad_norm": 6.6030449867248535, "learning_rate": 9.344200000000001e-06, "loss": 0.0777, "step": 3280 }, { "epoch": 0.17492136269126193, "grad_norm": 7.132630348205566, "learning_rate": 9.344e-06, "loss": 0.0715, "step": 3281 }, { "epoch": 0.17497467612091486, "grad_norm": 11.094738006591797, "learning_rate": 9.3438e-06, "loss": 0.0934, "step": 3282 }, { "epoch": 0.17502798955056778, "grad_norm": 5.30325174331665, "learning_rate": 9.3436e-06, "loss": 0.0711, "step": 3283 }, { "epoch": 0.1750813029802207, "grad_norm": 2.431004047393799, "learning_rate": 9.343400000000001e-06, "loss": 0.0338, "step": 3284 }, { "epoch": 0.17513461640987366, "grad_norm": 2.6689329147338867, "learning_rate": 9.3432e-06, "loss": 0.0318, "step": 3285 }, { "epoch": 0.17518792983952658, "grad_norm": 4.314306259155273, "learning_rate": 9.343e-06, "loss": 0.0298, "step": 3286 }, { "epoch": 0.1752412432691795, "grad_norm": 5.724691390991211, "learning_rate": 9.342800000000001e-06, "loss": 0.0725, "step": 3287 }, { "epoch": 0.17529455669883243, "grad_norm": 7.6410322189331055, "learning_rate": 9.342600000000001e-06, "loss": 0.0935, "step": 3288 }, { "epoch": 0.17534787012848538, "grad_norm": 4.59070348739624, "learning_rate": 9.3424e-06, "loss": 0.0505, "step": 3289 }, { "epoch": 0.1754011835581383, "grad_norm": 8.412013053894043, "learning_rate": 9.342200000000002e-06, "loss": 0.0425, "step": 3290 }, { "epoch": 0.17545449698779123, "grad_norm": 7.879114627838135, "learning_rate": 9.342000000000001e-06, "loss": 0.0573, "step": 3291 }, { "epoch": 0.17550781041744415, "grad_norm": 7.319638252258301, "learning_rate": 9.341800000000001e-06, "loss": 0.0507, "step": 3292 }, { "epoch": 0.17556112384709707, "grad_norm": 11.956052780151367, "learning_rate": 9.3416e-06, "loss": 0.0943, "step": 3293 }, { "epoch": 0.17561443727675002, "grad_norm": 3.5209529399871826, "learning_rate": 9.3414e-06, "loss": 0.0432, "step": 3294 }, { "epoch": 0.17566775070640295, "grad_norm": 4.974749565124512, "learning_rate": 9.3412e-06, "loss": 0.0633, "step": 3295 }, { "epoch": 0.17572106413605587, "grad_norm": 5.113019943237305, "learning_rate": 9.341000000000001e-06, "loss": 0.0701, "step": 3296 }, { "epoch": 0.1757743775657088, "grad_norm": 7.24384880065918, "learning_rate": 9.3408e-06, "loss": 0.0731, "step": 3297 }, { "epoch": 0.17582769099536172, "grad_norm": 4.392817497253418, "learning_rate": 9.3406e-06, "loss": 0.0511, "step": 3298 }, { "epoch": 0.17588100442501467, "grad_norm": 10.351607322692871, "learning_rate": 9.340400000000001e-06, "loss": 0.12, "step": 3299 }, { "epoch": 0.1759343178546676, "grad_norm": 5.076275825500488, "learning_rate": 9.3402e-06, "loss": 0.0383, "step": 3300 }, { "epoch": 0.17598763128432052, "grad_norm": 6.395242214202881, "learning_rate": 9.340000000000002e-06, "loss": 0.0758, "step": 3301 }, { "epoch": 0.17604094471397344, "grad_norm": 3.7144196033477783, "learning_rate": 9.339800000000002e-06, "loss": 0.0688, "step": 3302 }, { "epoch": 0.1760942581436264, "grad_norm": 5.932956695556641, "learning_rate": 9.339600000000001e-06, "loss": 0.0589, "step": 3303 }, { "epoch": 0.17614757157327932, "grad_norm": 4.96278190612793, "learning_rate": 9.3394e-06, "loss": 0.0665, "step": 3304 }, { "epoch": 0.17620088500293224, "grad_norm": 3.9939839839935303, "learning_rate": 9.3392e-06, "loss": 0.0463, "step": 3305 }, { "epoch": 0.17625419843258516, "grad_norm": 4.641748905181885, "learning_rate": 9.339e-06, "loss": 0.0471, "step": 3306 }, { "epoch": 0.1763075118622381, "grad_norm": 5.053072452545166, "learning_rate": 9.338800000000001e-06, "loss": 0.0676, "step": 3307 }, { "epoch": 0.17636082529189104, "grad_norm": 6.553955554962158, "learning_rate": 9.3386e-06, "loss": 0.1169, "step": 3308 }, { "epoch": 0.17641413872154396, "grad_norm": 4.140974998474121, "learning_rate": 9.3384e-06, "loss": 0.0486, "step": 3309 }, { "epoch": 0.17646745215119689, "grad_norm": 6.726321697235107, "learning_rate": 9.3382e-06, "loss": 0.1012, "step": 3310 }, { "epoch": 0.1765207655808498, "grad_norm": 2.440056085586548, "learning_rate": 9.338000000000001e-06, "loss": 0.0218, "step": 3311 }, { "epoch": 0.17657407901050273, "grad_norm": 3.045363187789917, "learning_rate": 9.3378e-06, "loss": 0.0334, "step": 3312 }, { "epoch": 0.17662739244015568, "grad_norm": 5.200494766235352, "learning_rate": 9.337600000000002e-06, "loss": 0.0568, "step": 3313 }, { "epoch": 0.1766807058698086, "grad_norm": 5.276790618896484, "learning_rate": 9.337400000000001e-06, "loss": 0.0899, "step": 3314 }, { "epoch": 0.17673401929946153, "grad_norm": 3.9972498416900635, "learning_rate": 9.337200000000001e-06, "loss": 0.061, "step": 3315 }, { "epoch": 0.17678733272911445, "grad_norm": 4.220996856689453, "learning_rate": 9.337e-06, "loss": 0.0946, "step": 3316 }, { "epoch": 0.1768406461587674, "grad_norm": 5.225404739379883, "learning_rate": 9.3368e-06, "loss": 0.0447, "step": 3317 }, { "epoch": 0.17689395958842033, "grad_norm": 4.278402805328369, "learning_rate": 9.336600000000001e-06, "loss": 0.0544, "step": 3318 }, { "epoch": 0.17694727301807325, "grad_norm": 2.6974472999572754, "learning_rate": 9.336400000000001e-06, "loss": 0.0328, "step": 3319 }, { "epoch": 0.17700058644772618, "grad_norm": 4.063782215118408, "learning_rate": 9.3362e-06, "loss": 0.0464, "step": 3320 }, { "epoch": 0.1770538998773791, "grad_norm": 3.765160322189331, "learning_rate": 9.336e-06, "loss": 0.0348, "step": 3321 }, { "epoch": 0.17710721330703205, "grad_norm": 3.1881113052368164, "learning_rate": 9.3358e-06, "loss": 0.0452, "step": 3322 }, { "epoch": 0.17716052673668498, "grad_norm": 5.094831943511963, "learning_rate": 9.335600000000001e-06, "loss": 0.0403, "step": 3323 }, { "epoch": 0.1772138401663379, "grad_norm": 3.8648886680603027, "learning_rate": 9.3354e-06, "loss": 0.0355, "step": 3324 }, { "epoch": 0.17726715359599082, "grad_norm": 8.461472511291504, "learning_rate": 9.335200000000002e-06, "loss": 0.0778, "step": 3325 }, { "epoch": 0.17732046702564375, "grad_norm": 7.965189456939697, "learning_rate": 9.335000000000001e-06, "loss": 0.0681, "step": 3326 }, { "epoch": 0.1773737804552967, "grad_norm": 3.4042298793792725, "learning_rate": 9.3348e-06, "loss": 0.0423, "step": 3327 }, { "epoch": 0.17742709388494962, "grad_norm": 6.650249004364014, "learning_rate": 9.3346e-06, "loss": 0.0867, "step": 3328 }, { "epoch": 0.17748040731460255, "grad_norm": 3.5240366458892822, "learning_rate": 9.334400000000002e-06, "loss": 0.0522, "step": 3329 }, { "epoch": 0.17753372074425547, "grad_norm": 5.665248870849609, "learning_rate": 9.334200000000001e-06, "loss": 0.0716, "step": 3330 }, { "epoch": 0.17758703417390842, "grad_norm": 4.85902214050293, "learning_rate": 9.334e-06, "loss": 0.0643, "step": 3331 }, { "epoch": 0.17764034760356134, "grad_norm": 5.124114513397217, "learning_rate": 9.3338e-06, "loss": 0.0329, "step": 3332 }, { "epoch": 0.17769366103321427, "grad_norm": 5.990568161010742, "learning_rate": 9.3336e-06, "loss": 0.0703, "step": 3333 }, { "epoch": 0.1777469744628672, "grad_norm": 4.32629919052124, "learning_rate": 9.3334e-06, "loss": 0.0666, "step": 3334 }, { "epoch": 0.17780028789252011, "grad_norm": 10.16152286529541, "learning_rate": 9.3332e-06, "loss": 0.0544, "step": 3335 }, { "epoch": 0.17785360132217307, "grad_norm": 8.277653694152832, "learning_rate": 9.333e-06, "loss": 0.0797, "step": 3336 }, { "epoch": 0.177906914751826, "grad_norm": 4.607676982879639, "learning_rate": 9.332800000000002e-06, "loss": 0.0653, "step": 3337 }, { "epoch": 0.1779602281814789, "grad_norm": 4.7047343254089355, "learning_rate": 9.332600000000001e-06, "loss": 0.0696, "step": 3338 }, { "epoch": 0.17801354161113184, "grad_norm": 6.159671306610107, "learning_rate": 9.3324e-06, "loss": 0.0591, "step": 3339 }, { "epoch": 0.17806685504078476, "grad_norm": 6.775638580322266, "learning_rate": 9.332200000000002e-06, "loss": 0.0849, "step": 3340 }, { "epoch": 0.1781201684704377, "grad_norm": 3.859644651412964, "learning_rate": 9.332000000000001e-06, "loss": 0.0574, "step": 3341 }, { "epoch": 0.17817348190009064, "grad_norm": 6.636796951293945, "learning_rate": 9.331800000000001e-06, "loss": 0.0743, "step": 3342 }, { "epoch": 0.17822679532974356, "grad_norm": 5.084325790405273, "learning_rate": 9.3316e-06, "loss": 0.0591, "step": 3343 }, { "epoch": 0.17828010875939648, "grad_norm": 6.668359279632568, "learning_rate": 9.3314e-06, "loss": 0.1017, "step": 3344 }, { "epoch": 0.17833342218904943, "grad_norm": 5.325651168823242, "learning_rate": 9.3312e-06, "loss": 0.0811, "step": 3345 }, { "epoch": 0.17838673561870236, "grad_norm": 4.440792083740234, "learning_rate": 9.331000000000001e-06, "loss": 0.05, "step": 3346 }, { "epoch": 0.17844004904835528, "grad_norm": 3.445486545562744, "learning_rate": 9.3308e-06, "loss": 0.0399, "step": 3347 }, { "epoch": 0.1784933624780082, "grad_norm": 3.9053709506988525, "learning_rate": 9.3306e-06, "loss": 0.0757, "step": 3348 }, { "epoch": 0.17854667590766113, "grad_norm": 3.96760892868042, "learning_rate": 9.330400000000001e-06, "loss": 0.0607, "step": 3349 }, { "epoch": 0.17859998933731408, "grad_norm": 4.269350528717041, "learning_rate": 9.330200000000001e-06, "loss": 0.0441, "step": 3350 }, { "epoch": 0.178653302766967, "grad_norm": 3.735238552093506, "learning_rate": 9.33e-06, "loss": 0.0638, "step": 3351 }, { "epoch": 0.17870661619661993, "grad_norm": 3.099919080734253, "learning_rate": 9.329800000000002e-06, "loss": 0.0555, "step": 3352 }, { "epoch": 0.17875992962627285, "grad_norm": 8.663366317749023, "learning_rate": 9.329600000000001e-06, "loss": 0.0718, "step": 3353 }, { "epoch": 0.17881324305592577, "grad_norm": 3.9371397495269775, "learning_rate": 9.3294e-06, "loss": 0.0412, "step": 3354 }, { "epoch": 0.17886655648557873, "grad_norm": 3.9197041988372803, "learning_rate": 9.3292e-06, "loss": 0.0487, "step": 3355 }, { "epoch": 0.17891986991523165, "grad_norm": 5.331196308135986, "learning_rate": 9.329e-06, "loss": 0.0477, "step": 3356 }, { "epoch": 0.17897318334488457, "grad_norm": 3.3184683322906494, "learning_rate": 9.328800000000001e-06, "loss": 0.0438, "step": 3357 }, { "epoch": 0.1790264967745375, "grad_norm": 6.049184322357178, "learning_rate": 9.3286e-06, "loss": 0.0905, "step": 3358 }, { "epoch": 0.17907981020419045, "grad_norm": 4.625940799713135, "learning_rate": 9.3284e-06, "loss": 0.07, "step": 3359 }, { "epoch": 0.17913312363384337, "grad_norm": 6.373401165008545, "learning_rate": 9.3282e-06, "loss": 0.1012, "step": 3360 }, { "epoch": 0.1791864370634963, "grad_norm": 9.185760498046875, "learning_rate": 9.328000000000001e-06, "loss": 0.0882, "step": 3361 }, { "epoch": 0.17923975049314922, "grad_norm": 7.246307849884033, "learning_rate": 9.3278e-06, "loss": 0.0762, "step": 3362 }, { "epoch": 0.17929306392280214, "grad_norm": 5.7457146644592285, "learning_rate": 9.3276e-06, "loss": 0.1135, "step": 3363 }, { "epoch": 0.1793463773524551, "grad_norm": 10.53426742553711, "learning_rate": 9.327400000000002e-06, "loss": 0.0463, "step": 3364 }, { "epoch": 0.17939969078210802, "grad_norm": 7.537827014923096, "learning_rate": 9.327200000000001e-06, "loss": 0.0643, "step": 3365 }, { "epoch": 0.17945300421176094, "grad_norm": 4.70295524597168, "learning_rate": 9.327e-06, "loss": 0.0561, "step": 3366 }, { "epoch": 0.17950631764141387, "grad_norm": 2.9870853424072266, "learning_rate": 9.3268e-06, "loss": 0.0406, "step": 3367 }, { "epoch": 0.1795596310710668, "grad_norm": 3.2665903568267822, "learning_rate": 9.326600000000001e-06, "loss": 0.0323, "step": 3368 }, { "epoch": 0.17961294450071974, "grad_norm": 5.508352279663086, "learning_rate": 9.326400000000001e-06, "loss": 0.06, "step": 3369 }, { "epoch": 0.17966625793037266, "grad_norm": 5.814552307128906, "learning_rate": 9.3262e-06, "loss": 0.0977, "step": 3370 }, { "epoch": 0.1797195713600256, "grad_norm": 9.179100036621094, "learning_rate": 9.326e-06, "loss": 0.0629, "step": 3371 }, { "epoch": 0.1797728847896785, "grad_norm": 2.9606266021728516, "learning_rate": 9.3258e-06, "loss": 0.0363, "step": 3372 }, { "epoch": 0.17982619821933146, "grad_norm": 2.7225968837738037, "learning_rate": 9.325600000000001e-06, "loss": 0.0354, "step": 3373 }, { "epoch": 0.17987951164898439, "grad_norm": 5.02419376373291, "learning_rate": 9.3254e-06, "loss": 0.0957, "step": 3374 }, { "epoch": 0.1799328250786373, "grad_norm": 4.174210548400879, "learning_rate": 9.3252e-06, "loss": 0.0992, "step": 3375 }, { "epoch": 0.17998613850829023, "grad_norm": 4.654776573181152, "learning_rate": 9.325000000000001e-06, "loss": 0.0797, "step": 3376 }, { "epoch": 0.18003945193794316, "grad_norm": 3.2609827518463135, "learning_rate": 9.324800000000001e-06, "loss": 0.0716, "step": 3377 }, { "epoch": 0.1800927653675961, "grad_norm": 4.030890464782715, "learning_rate": 9.3246e-06, "loss": 0.0485, "step": 3378 }, { "epoch": 0.18014607879724903, "grad_norm": 3.604591131210327, "learning_rate": 9.324400000000002e-06, "loss": 0.0505, "step": 3379 }, { "epoch": 0.18019939222690196, "grad_norm": 4.758306980133057, "learning_rate": 9.324200000000001e-06, "loss": 0.0671, "step": 3380 }, { "epoch": 0.18025270565655488, "grad_norm": 6.3611040115356445, "learning_rate": 9.324000000000001e-06, "loss": 0.103, "step": 3381 }, { "epoch": 0.1803060190862078, "grad_norm": 6.873837947845459, "learning_rate": 9.3238e-06, "loss": 0.0679, "step": 3382 }, { "epoch": 0.18035933251586075, "grad_norm": 9.406601905822754, "learning_rate": 9.3236e-06, "loss": 0.0908, "step": 3383 }, { "epoch": 0.18041264594551368, "grad_norm": 3.711642265319824, "learning_rate": 9.3234e-06, "loss": 0.0657, "step": 3384 }, { "epoch": 0.1804659593751666, "grad_norm": 4.075189590454102, "learning_rate": 9.3232e-06, "loss": 0.0454, "step": 3385 }, { "epoch": 0.18051927280481952, "grad_norm": 3.6655056476593018, "learning_rate": 9.323e-06, "loss": 0.0485, "step": 3386 }, { "epoch": 0.18057258623447248, "grad_norm": 4.009129047393799, "learning_rate": 9.3228e-06, "loss": 0.0655, "step": 3387 }, { "epoch": 0.1806258996641254, "grad_norm": 7.272761821746826, "learning_rate": 9.322600000000001e-06, "loss": 0.1236, "step": 3388 }, { "epoch": 0.18067921309377832, "grad_norm": 3.6578867435455322, "learning_rate": 9.3224e-06, "loss": 0.0586, "step": 3389 }, { "epoch": 0.18073252652343125, "grad_norm": 12.785931587219238, "learning_rate": 9.322200000000002e-06, "loss": 0.0644, "step": 3390 }, { "epoch": 0.18078583995308417, "grad_norm": 5.306786060333252, "learning_rate": 9.322000000000002e-06, "loss": 0.0634, "step": 3391 }, { "epoch": 0.18083915338273712, "grad_norm": 4.400528907775879, "learning_rate": 9.321800000000001e-06, "loss": 0.0641, "step": 3392 }, { "epoch": 0.18089246681239005, "grad_norm": 4.304744720458984, "learning_rate": 9.3216e-06, "loss": 0.061, "step": 3393 }, { "epoch": 0.18094578024204297, "grad_norm": 6.913015365600586, "learning_rate": 9.3214e-06, "loss": 0.0424, "step": 3394 }, { "epoch": 0.1809990936716959, "grad_norm": 3.9494926929473877, "learning_rate": 9.3212e-06, "loss": 0.04, "step": 3395 }, { "epoch": 0.18105240710134882, "grad_norm": 5.264172077178955, "learning_rate": 9.321000000000001e-06, "loss": 0.0709, "step": 3396 }, { "epoch": 0.18110572053100177, "grad_norm": 4.154259204864502, "learning_rate": 9.3208e-06, "loss": 0.0447, "step": 3397 }, { "epoch": 0.1811590339606547, "grad_norm": 5.433553218841553, "learning_rate": 9.3206e-06, "loss": 0.0637, "step": 3398 }, { "epoch": 0.18121234739030762, "grad_norm": 3.028797149658203, "learning_rate": 9.3204e-06, "loss": 0.0433, "step": 3399 }, { "epoch": 0.18126566081996054, "grad_norm": 7.825049877166748, "learning_rate": 9.320200000000001e-06, "loss": 0.085, "step": 3400 }, { "epoch": 0.1813189742496135, "grad_norm": 2.6755640506744385, "learning_rate": 9.32e-06, "loss": 0.0724, "step": 3401 }, { "epoch": 0.18137228767926641, "grad_norm": 5.550328731536865, "learning_rate": 9.319800000000002e-06, "loss": 0.0889, "step": 3402 }, { "epoch": 0.18142560110891934, "grad_norm": 3.12338924407959, "learning_rate": 9.319600000000001e-06, "loss": 0.0467, "step": 3403 }, { "epoch": 0.18147891453857226, "grad_norm": 4.681898593902588, "learning_rate": 9.319400000000001e-06, "loss": 0.0542, "step": 3404 }, { "epoch": 0.18153222796822518, "grad_norm": 3.208216428756714, "learning_rate": 9.3192e-06, "loss": 0.025, "step": 3405 }, { "epoch": 0.18158554139787814, "grad_norm": 6.4125752449035645, "learning_rate": 9.319e-06, "loss": 0.0515, "step": 3406 }, { "epoch": 0.18163885482753106, "grad_norm": 3.7544424533843994, "learning_rate": 9.318800000000001e-06, "loss": 0.064, "step": 3407 }, { "epoch": 0.18169216825718398, "grad_norm": 6.2188544273376465, "learning_rate": 9.318600000000001e-06, "loss": 0.052, "step": 3408 }, { "epoch": 0.1817454816868369, "grad_norm": 4.205999851226807, "learning_rate": 9.3184e-06, "loss": 0.0895, "step": 3409 }, { "epoch": 0.18179879511648983, "grad_norm": 2.3108344078063965, "learning_rate": 9.3182e-06, "loss": 0.0258, "step": 3410 }, { "epoch": 0.18185210854614278, "grad_norm": 4.953460216522217, "learning_rate": 9.318e-06, "loss": 0.0539, "step": 3411 }, { "epoch": 0.1819054219757957, "grad_norm": 3.9633660316467285, "learning_rate": 9.3178e-06, "loss": 0.061, "step": 3412 }, { "epoch": 0.18195873540544863, "grad_norm": 5.807484149932861, "learning_rate": 9.3176e-06, "loss": 0.0439, "step": 3413 }, { "epoch": 0.18201204883510155, "grad_norm": 2.0929908752441406, "learning_rate": 9.317400000000002e-06, "loss": 0.0265, "step": 3414 }, { "epoch": 0.1820653622647545, "grad_norm": 2.330881357192993, "learning_rate": 9.317200000000001e-06, "loss": 0.0217, "step": 3415 }, { "epoch": 0.18211867569440743, "grad_norm": 5.105229377746582, "learning_rate": 9.317e-06, "loss": 0.0669, "step": 3416 }, { "epoch": 0.18217198912406035, "grad_norm": 2.6296226978302, "learning_rate": 9.3168e-06, "loss": 0.0222, "step": 3417 }, { "epoch": 0.18222530255371328, "grad_norm": 5.786065578460693, "learning_rate": 9.316600000000002e-06, "loss": 0.0614, "step": 3418 }, { "epoch": 0.1822786159833662, "grad_norm": 2.9899532794952393, "learning_rate": 9.316400000000001e-06, "loss": 0.0479, "step": 3419 }, { "epoch": 0.18233192941301915, "grad_norm": 4.111888885498047, "learning_rate": 9.3162e-06, "loss": 0.0305, "step": 3420 }, { "epoch": 0.18238524284267207, "grad_norm": 3.4230713844299316, "learning_rate": 9.316e-06, "loss": 0.0206, "step": 3421 }, { "epoch": 0.182438556272325, "grad_norm": 8.130460739135742, "learning_rate": 9.3158e-06, "loss": 0.0934, "step": 3422 }, { "epoch": 0.18249186970197792, "grad_norm": 1.9349404573440552, "learning_rate": 9.315600000000001e-06, "loss": 0.0313, "step": 3423 }, { "epoch": 0.18254518313163084, "grad_norm": 5.04518461227417, "learning_rate": 9.3154e-06, "loss": 0.0606, "step": 3424 }, { "epoch": 0.1825984965612838, "grad_norm": 4.412247657775879, "learning_rate": 9.3152e-06, "loss": 0.0323, "step": 3425 }, { "epoch": 0.18265180999093672, "grad_norm": 12.305939674377441, "learning_rate": 9.315000000000001e-06, "loss": 0.0749, "step": 3426 }, { "epoch": 0.18270512342058964, "grad_norm": 5.056240558624268, "learning_rate": 9.314800000000001e-06, "loss": 0.0626, "step": 3427 }, { "epoch": 0.18275843685024257, "grad_norm": 7.352794647216797, "learning_rate": 9.3146e-06, "loss": 0.0766, "step": 3428 }, { "epoch": 0.18281175027989552, "grad_norm": 4.196517467498779, "learning_rate": 9.314400000000002e-06, "loss": 0.0184, "step": 3429 }, { "epoch": 0.18286506370954844, "grad_norm": 5.1922197341918945, "learning_rate": 9.314200000000001e-06, "loss": 0.0721, "step": 3430 }, { "epoch": 0.18291837713920137, "grad_norm": 6.356308937072754, "learning_rate": 9.314000000000001e-06, "loss": 0.0843, "step": 3431 }, { "epoch": 0.1829716905688543, "grad_norm": 3.84732723236084, "learning_rate": 9.3138e-06, "loss": 0.0311, "step": 3432 }, { "epoch": 0.1830250039985072, "grad_norm": 6.312093734741211, "learning_rate": 9.3136e-06, "loss": 0.0836, "step": 3433 }, { "epoch": 0.18307831742816016, "grad_norm": 10.499442100524902, "learning_rate": 9.3134e-06, "loss": 0.0763, "step": 3434 }, { "epoch": 0.1831316308578131, "grad_norm": 3.718324661254883, "learning_rate": 9.313200000000001e-06, "loss": 0.0455, "step": 3435 }, { "epoch": 0.183184944287466, "grad_norm": 5.810435771942139, "learning_rate": 9.313e-06, "loss": 0.0351, "step": 3436 }, { "epoch": 0.18323825771711894, "grad_norm": 5.685726165771484, "learning_rate": 9.3128e-06, "loss": 0.0806, "step": 3437 }, { "epoch": 0.18329157114677186, "grad_norm": 3.5902628898620605, "learning_rate": 9.312600000000001e-06, "loss": 0.0514, "step": 3438 }, { "epoch": 0.1833448845764248, "grad_norm": 3.1062753200531006, "learning_rate": 9.3124e-06, "loss": 0.0261, "step": 3439 }, { "epoch": 0.18339819800607773, "grad_norm": 3.5068023204803467, "learning_rate": 9.312200000000002e-06, "loss": 0.0615, "step": 3440 }, { "epoch": 0.18345151143573066, "grad_norm": 3.310709238052368, "learning_rate": 9.312000000000002e-06, "loss": 0.0308, "step": 3441 }, { "epoch": 0.18350482486538358, "grad_norm": 3.040520668029785, "learning_rate": 9.311800000000001e-06, "loss": 0.0316, "step": 3442 }, { "epoch": 0.18355813829503653, "grad_norm": 4.0171051025390625, "learning_rate": 9.3116e-06, "loss": 0.0305, "step": 3443 }, { "epoch": 0.18361145172468946, "grad_norm": 5.488437175750732, "learning_rate": 9.3114e-06, "loss": 0.098, "step": 3444 }, { "epoch": 0.18366476515434238, "grad_norm": 2.4440383911132812, "learning_rate": 9.3112e-06, "loss": 0.0436, "step": 3445 }, { "epoch": 0.1837180785839953, "grad_norm": 5.9580888748168945, "learning_rate": 9.311000000000001e-06, "loss": 0.104, "step": 3446 }, { "epoch": 0.18377139201364823, "grad_norm": 8.603740692138672, "learning_rate": 9.3108e-06, "loss": 0.0837, "step": 3447 }, { "epoch": 0.18382470544330118, "grad_norm": 4.9456787109375, "learning_rate": 9.3106e-06, "loss": 0.0549, "step": 3448 }, { "epoch": 0.1838780188729541, "grad_norm": 4.0104804039001465, "learning_rate": 9.3104e-06, "loss": 0.0336, "step": 3449 }, { "epoch": 0.18393133230260703, "grad_norm": 3.2699623107910156, "learning_rate": 9.310200000000001e-06, "loss": 0.0382, "step": 3450 }, { "epoch": 0.18398464573225995, "grad_norm": 4.4048237800598145, "learning_rate": 9.31e-06, "loss": 0.0555, "step": 3451 }, { "epoch": 0.18403795916191287, "grad_norm": 4.7771830558776855, "learning_rate": 9.309800000000002e-06, "loss": 0.0836, "step": 3452 }, { "epoch": 0.18409127259156582, "grad_norm": 4.458744049072266, "learning_rate": 9.309600000000001e-06, "loss": 0.0594, "step": 3453 }, { "epoch": 0.18414458602121875, "grad_norm": 6.792351245880127, "learning_rate": 9.309400000000001e-06, "loss": 0.0541, "step": 3454 }, { "epoch": 0.18419789945087167, "grad_norm": 5.275461196899414, "learning_rate": 9.3092e-06, "loss": 0.0571, "step": 3455 }, { "epoch": 0.1842512128805246, "grad_norm": 3.857646942138672, "learning_rate": 9.309e-06, "loss": 0.0625, "step": 3456 }, { "epoch": 0.18430452631017755, "grad_norm": 5.444296836853027, "learning_rate": 9.308800000000001e-06, "loss": 0.0575, "step": 3457 }, { "epoch": 0.18435783973983047, "grad_norm": 4.461390972137451, "learning_rate": 9.308600000000001e-06, "loss": 0.054, "step": 3458 }, { "epoch": 0.1844111531694834, "grad_norm": 2.2989046573638916, "learning_rate": 9.3084e-06, "loss": 0.0327, "step": 3459 }, { "epoch": 0.18446446659913632, "grad_norm": 6.366151332855225, "learning_rate": 9.3082e-06, "loss": 0.0766, "step": 3460 }, { "epoch": 0.18451778002878924, "grad_norm": 4.192171573638916, "learning_rate": 9.308e-06, "loss": 0.0685, "step": 3461 }, { "epoch": 0.1845710934584422, "grad_norm": 2.732325315475464, "learning_rate": 9.307800000000001e-06, "loss": 0.0322, "step": 3462 }, { "epoch": 0.18462440688809512, "grad_norm": 3.6427128314971924, "learning_rate": 9.3076e-06, "loss": 0.0719, "step": 3463 }, { "epoch": 0.18467772031774804, "grad_norm": 2.6487677097320557, "learning_rate": 9.307400000000002e-06, "loss": 0.029, "step": 3464 }, { "epoch": 0.18473103374740096, "grad_norm": 3.555777072906494, "learning_rate": 9.307200000000001e-06, "loss": 0.0441, "step": 3465 }, { "epoch": 0.1847843471770539, "grad_norm": 4.317951679229736, "learning_rate": 9.307e-06, "loss": 0.0344, "step": 3466 }, { "epoch": 0.18483766060670684, "grad_norm": 5.927742004394531, "learning_rate": 9.3068e-06, "loss": 0.0812, "step": 3467 }, { "epoch": 0.18489097403635976, "grad_norm": 5.375673770904541, "learning_rate": 9.306600000000002e-06, "loss": 0.0452, "step": 3468 }, { "epoch": 0.18494428746601269, "grad_norm": 4.488911151885986, "learning_rate": 9.306400000000001e-06, "loss": 0.0436, "step": 3469 }, { "epoch": 0.1849976008956656, "grad_norm": 3.5146753787994385, "learning_rate": 9.3062e-06, "loss": 0.0702, "step": 3470 }, { "epoch": 0.18505091432531856, "grad_norm": 6.55459451675415, "learning_rate": 9.306e-06, "loss": 0.0725, "step": 3471 }, { "epoch": 0.18510422775497148, "grad_norm": 4.528399467468262, "learning_rate": 9.3058e-06, "loss": 0.0439, "step": 3472 }, { "epoch": 0.1851575411846244, "grad_norm": 7.382299900054932, "learning_rate": 9.305600000000001e-06, "loss": 0.0843, "step": 3473 }, { "epoch": 0.18521085461427733, "grad_norm": 3.582430601119995, "learning_rate": 9.3054e-06, "loss": 0.054, "step": 3474 }, { "epoch": 0.18526416804393026, "grad_norm": 5.016281604766846, "learning_rate": 9.3052e-06, "loss": 0.0439, "step": 3475 }, { "epoch": 0.1853174814735832, "grad_norm": 3.4395010471343994, "learning_rate": 9.305000000000002e-06, "loss": 0.0765, "step": 3476 }, { "epoch": 0.18537079490323613, "grad_norm": 4.0141282081604, "learning_rate": 9.304800000000001e-06, "loss": 0.0653, "step": 3477 }, { "epoch": 0.18542410833288905, "grad_norm": 3.4690535068511963, "learning_rate": 9.3046e-06, "loss": 0.0398, "step": 3478 }, { "epoch": 0.18547742176254198, "grad_norm": 2.186103105545044, "learning_rate": 9.304400000000002e-06, "loss": 0.0312, "step": 3479 }, { "epoch": 0.1855307351921949, "grad_norm": 7.672717094421387, "learning_rate": 9.304200000000001e-06, "loss": 0.113, "step": 3480 }, { "epoch": 0.18558404862184785, "grad_norm": 3.767104387283325, "learning_rate": 9.304000000000001e-06, "loss": 0.067, "step": 3481 }, { "epoch": 0.18563736205150078, "grad_norm": 3.610393524169922, "learning_rate": 9.3038e-06, "loss": 0.0496, "step": 3482 }, { "epoch": 0.1856906754811537, "grad_norm": 4.672531604766846, "learning_rate": 9.3036e-06, "loss": 0.0659, "step": 3483 }, { "epoch": 0.18574398891080662, "grad_norm": 2.958441972732544, "learning_rate": 9.3034e-06, "loss": 0.04, "step": 3484 }, { "epoch": 0.18579730234045957, "grad_norm": 5.480247497558594, "learning_rate": 9.303200000000001e-06, "loss": 0.0689, "step": 3485 }, { "epoch": 0.1858506157701125, "grad_norm": 6.426791191101074, "learning_rate": 9.303e-06, "loss": 0.0404, "step": 3486 }, { "epoch": 0.18590392919976542, "grad_norm": 3.733219861984253, "learning_rate": 9.3028e-06, "loss": 0.0632, "step": 3487 }, { "epoch": 0.18595724262941835, "grad_norm": 4.8188796043396, "learning_rate": 9.302600000000001e-06, "loss": 0.0749, "step": 3488 }, { "epoch": 0.18601055605907127, "grad_norm": 6.858316898345947, "learning_rate": 9.302400000000001e-06, "loss": 0.0667, "step": 3489 }, { "epoch": 0.18606386948872422, "grad_norm": 3.964688777923584, "learning_rate": 9.3022e-06, "loss": 0.0402, "step": 3490 }, { "epoch": 0.18611718291837714, "grad_norm": 3.2817485332489014, "learning_rate": 9.302000000000002e-06, "loss": 0.0535, "step": 3491 }, { "epoch": 0.18617049634803007, "grad_norm": 7.792030334472656, "learning_rate": 9.301800000000001e-06, "loss": 0.0657, "step": 3492 }, { "epoch": 0.186223809777683, "grad_norm": 4.592111587524414, "learning_rate": 9.301600000000001e-06, "loss": 0.0343, "step": 3493 }, { "epoch": 0.18627712320733592, "grad_norm": 4.671570777893066, "learning_rate": 9.3014e-06, "loss": 0.0418, "step": 3494 }, { "epoch": 0.18633043663698887, "grad_norm": 2.9266576766967773, "learning_rate": 9.3012e-06, "loss": 0.0722, "step": 3495 }, { "epoch": 0.1863837500666418, "grad_norm": 4.969672679901123, "learning_rate": 9.301000000000001e-06, "loss": 0.0618, "step": 3496 }, { "epoch": 0.1864370634962947, "grad_norm": 4.680408477783203, "learning_rate": 9.3008e-06, "loss": 0.0642, "step": 3497 }, { "epoch": 0.18649037692594764, "grad_norm": 3.9345600605010986, "learning_rate": 9.3006e-06, "loss": 0.054, "step": 3498 }, { "epoch": 0.1865436903556006, "grad_norm": 5.3131327629089355, "learning_rate": 9.3004e-06, "loss": 0.0434, "step": 3499 }, { "epoch": 0.1865970037852535, "grad_norm": 2.2174534797668457, "learning_rate": 9.300200000000001e-06, "loss": 0.0195, "step": 3500 }, { "epoch": 0.18665031721490644, "grad_norm": 3.617914915084839, "learning_rate": 9.3e-06, "loss": 0.0372, "step": 3501 }, { "epoch": 0.18670363064455936, "grad_norm": 5.066535949707031, "learning_rate": 9.2998e-06, "loss": 0.065, "step": 3502 }, { "epoch": 0.18675694407421228, "grad_norm": 6.12184476852417, "learning_rate": 9.299600000000002e-06, "loss": 0.0629, "step": 3503 }, { "epoch": 0.18681025750386523, "grad_norm": 3.4785029888153076, "learning_rate": 9.299400000000001e-06, "loss": 0.0664, "step": 3504 }, { "epoch": 0.18686357093351816, "grad_norm": 3.882660388946533, "learning_rate": 9.2992e-06, "loss": 0.0563, "step": 3505 }, { "epoch": 0.18691688436317108, "grad_norm": 3.1254544258117676, "learning_rate": 9.299e-06, "loss": 0.0477, "step": 3506 }, { "epoch": 0.186970197792824, "grad_norm": 6.279579162597656, "learning_rate": 9.298800000000002e-06, "loss": 0.0948, "step": 3507 }, { "epoch": 0.18702351122247693, "grad_norm": 5.712896823883057, "learning_rate": 9.298600000000001e-06, "loss": 0.0723, "step": 3508 }, { "epoch": 0.18707682465212988, "grad_norm": 4.050152778625488, "learning_rate": 9.2984e-06, "loss": 0.0549, "step": 3509 }, { "epoch": 0.1871301380817828, "grad_norm": 3.342302083969116, "learning_rate": 9.2982e-06, "loss": 0.0418, "step": 3510 }, { "epoch": 0.18718345151143573, "grad_norm": 3.928560972213745, "learning_rate": 9.298e-06, "loss": 0.0659, "step": 3511 }, { "epoch": 0.18723676494108865, "grad_norm": 3.7810630798339844, "learning_rate": 9.297800000000001e-06, "loss": 0.0679, "step": 3512 }, { "epoch": 0.1872900783707416, "grad_norm": 7.625146865844727, "learning_rate": 9.2976e-06, "loss": 0.0519, "step": 3513 }, { "epoch": 0.18734339180039453, "grad_norm": 7.130128860473633, "learning_rate": 9.2974e-06, "loss": 0.0691, "step": 3514 }, { "epoch": 0.18739670523004745, "grad_norm": 8.463590621948242, "learning_rate": 9.297200000000001e-06, "loss": 0.0764, "step": 3515 }, { "epoch": 0.18745001865970037, "grad_norm": 4.527881145477295, "learning_rate": 9.297000000000001e-06, "loss": 0.0538, "step": 3516 }, { "epoch": 0.1875033320893533, "grad_norm": 5.40554666519165, "learning_rate": 9.2968e-06, "loss": 0.0571, "step": 3517 }, { "epoch": 0.18755664551900625, "grad_norm": 4.859543800354004, "learning_rate": 9.296600000000002e-06, "loss": 0.0701, "step": 3518 }, { "epoch": 0.18760995894865917, "grad_norm": 4.526221752166748, "learning_rate": 9.296400000000001e-06, "loss": 0.0675, "step": 3519 }, { "epoch": 0.1876632723783121, "grad_norm": 3.716721773147583, "learning_rate": 9.296200000000001e-06, "loss": 0.0645, "step": 3520 }, { "epoch": 0.18771658580796502, "grad_norm": 7.279275894165039, "learning_rate": 9.296e-06, "loss": 0.1027, "step": 3521 }, { "epoch": 0.18776989923761794, "grad_norm": 5.696093559265137, "learning_rate": 9.2958e-06, "loss": 0.0514, "step": 3522 }, { "epoch": 0.1878232126672709, "grad_norm": 3.460000514984131, "learning_rate": 9.295600000000001e-06, "loss": 0.0566, "step": 3523 }, { "epoch": 0.18787652609692382, "grad_norm": 3.694241523742676, "learning_rate": 9.2954e-06, "loss": 0.0494, "step": 3524 }, { "epoch": 0.18792983952657674, "grad_norm": 4.024970531463623, "learning_rate": 9.2952e-06, "loss": 0.0436, "step": 3525 }, { "epoch": 0.18798315295622967, "grad_norm": 4.621187210083008, "learning_rate": 9.295e-06, "loss": 0.0347, "step": 3526 }, { "epoch": 0.18803646638588262, "grad_norm": 2.9767050743103027, "learning_rate": 9.294800000000001e-06, "loss": 0.0495, "step": 3527 }, { "epoch": 0.18808977981553554, "grad_norm": 2.8394505977630615, "learning_rate": 9.2946e-06, "loss": 0.0431, "step": 3528 }, { "epoch": 0.18814309324518846, "grad_norm": 4.22358512878418, "learning_rate": 9.294400000000002e-06, "loss": 0.0732, "step": 3529 }, { "epoch": 0.1881964066748414, "grad_norm": 4.35360050201416, "learning_rate": 9.294200000000002e-06, "loss": 0.0639, "step": 3530 }, { "epoch": 0.1882497201044943, "grad_norm": 3.5105884075164795, "learning_rate": 9.294000000000001e-06, "loss": 0.044, "step": 3531 }, { "epoch": 0.18830303353414726, "grad_norm": 4.232689380645752, "learning_rate": 9.2938e-06, "loss": 0.0581, "step": 3532 }, { "epoch": 0.1883563469638002, "grad_norm": 3.7241575717926025, "learning_rate": 9.2936e-06, "loss": 0.0503, "step": 3533 }, { "epoch": 0.1884096603934531, "grad_norm": 7.378397464752197, "learning_rate": 9.293400000000002e-06, "loss": 0.1107, "step": 3534 }, { "epoch": 0.18846297382310603, "grad_norm": 2.930659532546997, "learning_rate": 9.293200000000001e-06, "loss": 0.0301, "step": 3535 }, { "epoch": 0.18851628725275896, "grad_norm": 2.732334852218628, "learning_rate": 9.293e-06, "loss": 0.0388, "step": 3536 }, { "epoch": 0.1885696006824119, "grad_norm": 6.019008159637451, "learning_rate": 9.2928e-06, "loss": 0.045, "step": 3537 }, { "epoch": 0.18862291411206483, "grad_norm": 4.476070404052734, "learning_rate": 9.2926e-06, "loss": 0.0937, "step": 3538 }, { "epoch": 0.18867622754171776, "grad_norm": 3.5096657276153564, "learning_rate": 9.292400000000001e-06, "loss": 0.0534, "step": 3539 }, { "epoch": 0.18872954097137068, "grad_norm": 4.525259971618652, "learning_rate": 9.2922e-06, "loss": 0.0645, "step": 3540 }, { "epoch": 0.18878285440102363, "grad_norm": 6.306429386138916, "learning_rate": 9.292000000000002e-06, "loss": 0.097, "step": 3541 }, { "epoch": 0.18883616783067655, "grad_norm": 5.204219341278076, "learning_rate": 9.291800000000001e-06, "loss": 0.0551, "step": 3542 }, { "epoch": 0.18888948126032948, "grad_norm": 3.596482038497925, "learning_rate": 9.291600000000001e-06, "loss": 0.065, "step": 3543 }, { "epoch": 0.1889427946899824, "grad_norm": 2.458303451538086, "learning_rate": 9.2914e-06, "loss": 0.0273, "step": 3544 }, { "epoch": 0.18899610811963533, "grad_norm": 3.1821539402008057, "learning_rate": 9.2912e-06, "loss": 0.0231, "step": 3545 }, { "epoch": 0.18904942154928828, "grad_norm": 3.6686782836914062, "learning_rate": 9.291000000000001e-06, "loss": 0.0513, "step": 3546 }, { "epoch": 0.1891027349789412, "grad_norm": 5.11789083480835, "learning_rate": 9.290800000000001e-06, "loss": 0.0679, "step": 3547 }, { "epoch": 0.18915604840859412, "grad_norm": 5.649054527282715, "learning_rate": 9.2906e-06, "loss": 0.0876, "step": 3548 }, { "epoch": 0.18920936183824705, "grad_norm": 8.884258270263672, "learning_rate": 9.2904e-06, "loss": 0.0553, "step": 3549 }, { "epoch": 0.18926267526789997, "grad_norm": 5.472231388092041, "learning_rate": 9.2902e-06, "loss": 0.0864, "step": 3550 }, { "epoch": 0.18931598869755292, "grad_norm": 6.459191799163818, "learning_rate": 9.29e-06, "loss": 0.0516, "step": 3551 }, { "epoch": 0.18936930212720585, "grad_norm": 2.7460367679595947, "learning_rate": 9.2898e-06, "loss": 0.0386, "step": 3552 }, { "epoch": 0.18942261555685877, "grad_norm": 6.234304904937744, "learning_rate": 9.289600000000002e-06, "loss": 0.0561, "step": 3553 }, { "epoch": 0.1894759289865117, "grad_norm": 4.50050687789917, "learning_rate": 9.289400000000001e-06, "loss": 0.0288, "step": 3554 }, { "epoch": 0.18952924241616464, "grad_norm": 5.173280239105225, "learning_rate": 9.2892e-06, "loss": 0.0514, "step": 3555 }, { "epoch": 0.18958255584581757, "grad_norm": 4.327754020690918, "learning_rate": 9.289e-06, "loss": 0.0322, "step": 3556 }, { "epoch": 0.1896358692754705, "grad_norm": 2.8540244102478027, "learning_rate": 9.288800000000002e-06, "loss": 0.024, "step": 3557 }, { "epoch": 0.18968918270512342, "grad_norm": 4.480246543884277, "learning_rate": 9.288600000000001e-06, "loss": 0.0856, "step": 3558 }, { "epoch": 0.18974249613477634, "grad_norm": 5.266228199005127, "learning_rate": 9.2884e-06, "loss": 0.0588, "step": 3559 }, { "epoch": 0.1897958095644293, "grad_norm": 7.515175819396973, "learning_rate": 9.2882e-06, "loss": 0.103, "step": 3560 }, { "epoch": 0.18984912299408221, "grad_norm": 4.550137519836426, "learning_rate": 9.288e-06, "loss": 0.0477, "step": 3561 }, { "epoch": 0.18990243642373514, "grad_norm": 4.718032360076904, "learning_rate": 9.287800000000001e-06, "loss": 0.0658, "step": 3562 }, { "epoch": 0.18995574985338806, "grad_norm": 4.051750183105469, "learning_rate": 9.2876e-06, "loss": 0.062, "step": 3563 }, { "epoch": 0.19000906328304099, "grad_norm": 4.685146808624268, "learning_rate": 9.2874e-06, "loss": 0.0892, "step": 3564 }, { "epoch": 0.19006237671269394, "grad_norm": 5.174282073974609, "learning_rate": 9.287200000000001e-06, "loss": 0.0503, "step": 3565 }, { "epoch": 0.19011569014234686, "grad_norm": 4.148860454559326, "learning_rate": 9.287000000000001e-06, "loss": 0.066, "step": 3566 }, { "epoch": 0.19016900357199978, "grad_norm": 5.992599964141846, "learning_rate": 9.2868e-06, "loss": 0.0905, "step": 3567 }, { "epoch": 0.1902223170016527, "grad_norm": 6.512564182281494, "learning_rate": 9.286600000000002e-06, "loss": 0.0895, "step": 3568 }, { "epoch": 0.19027563043130566, "grad_norm": 6.04111385345459, "learning_rate": 9.286400000000001e-06, "loss": 0.0878, "step": 3569 }, { "epoch": 0.19032894386095858, "grad_norm": 5.705407619476318, "learning_rate": 9.286200000000001e-06, "loss": 0.0553, "step": 3570 }, { "epoch": 0.1903822572906115, "grad_norm": 5.180878639221191, "learning_rate": 9.286e-06, "loss": 0.0943, "step": 3571 }, { "epoch": 0.19043557072026443, "grad_norm": 5.890937328338623, "learning_rate": 9.2858e-06, "loss": 0.0672, "step": 3572 }, { "epoch": 0.19048888414991735, "grad_norm": 2.9623773097991943, "learning_rate": 9.285600000000001e-06, "loss": 0.0597, "step": 3573 }, { "epoch": 0.1905421975795703, "grad_norm": 6.305797100067139, "learning_rate": 9.285400000000001e-06, "loss": 0.0696, "step": 3574 }, { "epoch": 0.19059551100922323, "grad_norm": 3.8377647399902344, "learning_rate": 9.2852e-06, "loss": 0.0394, "step": 3575 }, { "epoch": 0.19064882443887615, "grad_norm": 4.099860668182373, "learning_rate": 9.285e-06, "loss": 0.0471, "step": 3576 }, { "epoch": 0.19070213786852908, "grad_norm": 9.798615455627441, "learning_rate": 9.284800000000001e-06, "loss": 0.0787, "step": 3577 }, { "epoch": 0.190755451298182, "grad_norm": 3.6084272861480713, "learning_rate": 9.284600000000001e-06, "loss": 0.0415, "step": 3578 }, { "epoch": 0.19080876472783495, "grad_norm": 7.142980575561523, "learning_rate": 9.2844e-06, "loss": 0.115, "step": 3579 }, { "epoch": 0.19086207815748787, "grad_norm": 4.0516037940979, "learning_rate": 9.284200000000002e-06, "loss": 0.0731, "step": 3580 }, { "epoch": 0.1909153915871408, "grad_norm": 4.646738052368164, "learning_rate": 9.284000000000001e-06, "loss": 0.0462, "step": 3581 }, { "epoch": 0.19096870501679372, "grad_norm": 5.376834869384766, "learning_rate": 9.2838e-06, "loss": 0.0582, "step": 3582 }, { "epoch": 0.19102201844644667, "grad_norm": 4.029207706451416, "learning_rate": 9.2836e-06, "loss": 0.0507, "step": 3583 }, { "epoch": 0.1910753318760996, "grad_norm": 3.648789167404175, "learning_rate": 9.283400000000002e-06, "loss": 0.0549, "step": 3584 }, { "epoch": 0.19112864530575252, "grad_norm": 5.056656837463379, "learning_rate": 9.283200000000001e-06, "loss": 0.0685, "step": 3585 }, { "epoch": 0.19118195873540544, "grad_norm": 5.442882061004639, "learning_rate": 9.283e-06, "loss": 0.0674, "step": 3586 }, { "epoch": 0.19123527216505837, "grad_norm": 2.7976932525634766, "learning_rate": 9.2828e-06, "loss": 0.0363, "step": 3587 }, { "epoch": 0.19128858559471132, "grad_norm": 3.6938912868499756, "learning_rate": 9.2826e-06, "loss": 0.0389, "step": 3588 }, { "epoch": 0.19134189902436424, "grad_norm": 7.8998026847839355, "learning_rate": 9.282400000000001e-06, "loss": 0.0996, "step": 3589 }, { "epoch": 0.19139521245401717, "grad_norm": 6.349306106567383, "learning_rate": 9.2822e-06, "loss": 0.0639, "step": 3590 }, { "epoch": 0.1914485258836701, "grad_norm": 3.1646225452423096, "learning_rate": 9.282e-06, "loss": 0.0339, "step": 3591 }, { "epoch": 0.191501839313323, "grad_norm": 3.897843599319458, "learning_rate": 9.281800000000001e-06, "loss": 0.0545, "step": 3592 }, { "epoch": 0.19155515274297596, "grad_norm": 3.2728238105773926, "learning_rate": 9.281600000000001e-06, "loss": 0.0506, "step": 3593 }, { "epoch": 0.1916084661726289, "grad_norm": 4.746016502380371, "learning_rate": 9.2814e-06, "loss": 0.0457, "step": 3594 }, { "epoch": 0.1916617796022818, "grad_norm": 13.784689903259277, "learning_rate": 9.2812e-06, "loss": 0.1179, "step": 3595 }, { "epoch": 0.19171509303193474, "grad_norm": 5.487696647644043, "learning_rate": 9.281000000000001e-06, "loss": 0.0672, "step": 3596 }, { "epoch": 0.1917684064615877, "grad_norm": 3.5419416427612305, "learning_rate": 9.280800000000001e-06, "loss": 0.0418, "step": 3597 }, { "epoch": 0.1918217198912406, "grad_norm": 1.9153183698654175, "learning_rate": 9.2806e-06, "loss": 0.0247, "step": 3598 }, { "epoch": 0.19187503332089353, "grad_norm": 6.441226959228516, "learning_rate": 9.2804e-06, "loss": 0.079, "step": 3599 }, { "epoch": 0.19192834675054646, "grad_norm": 5.459754467010498, "learning_rate": 9.2802e-06, "loss": 0.0533, "step": 3600 }, { "epoch": 0.19198166018019938, "grad_norm": 6.89984130859375, "learning_rate": 9.280000000000001e-06, "loss": 0.0697, "step": 3601 }, { "epoch": 0.19203497360985233, "grad_norm": 5.368167877197266, "learning_rate": 9.2798e-06, "loss": 0.0435, "step": 3602 }, { "epoch": 0.19208828703950526, "grad_norm": 5.311424732208252, "learning_rate": 9.2796e-06, "loss": 0.0738, "step": 3603 }, { "epoch": 0.19214160046915818, "grad_norm": 3.2893149852752686, "learning_rate": 9.279400000000001e-06, "loss": 0.053, "step": 3604 }, { "epoch": 0.1921949138988111, "grad_norm": 11.777488708496094, "learning_rate": 9.279200000000001e-06, "loss": 0.0819, "step": 3605 }, { "epoch": 0.19224822732846403, "grad_norm": 9.825127601623535, "learning_rate": 9.279e-06, "loss": 0.1, "step": 3606 }, { "epoch": 0.19230154075811698, "grad_norm": 5.6702399253845215, "learning_rate": 9.278800000000002e-06, "loss": 0.0623, "step": 3607 }, { "epoch": 0.1923548541877699, "grad_norm": 4.350867748260498, "learning_rate": 9.278600000000001e-06, "loss": 0.0567, "step": 3608 }, { "epoch": 0.19240816761742283, "grad_norm": 5.529909610748291, "learning_rate": 9.2784e-06, "loss": 0.1012, "step": 3609 }, { "epoch": 0.19246148104707575, "grad_norm": 4.320470809936523, "learning_rate": 9.2782e-06, "loss": 0.0687, "step": 3610 }, { "epoch": 0.1925147944767287, "grad_norm": 8.026424407958984, "learning_rate": 9.278e-06, "loss": 0.0871, "step": 3611 }, { "epoch": 0.19256810790638162, "grad_norm": 2.37432861328125, "learning_rate": 9.277800000000001e-06, "loss": 0.0269, "step": 3612 }, { "epoch": 0.19262142133603455, "grad_norm": 4.776724338531494, "learning_rate": 9.2776e-06, "loss": 0.0686, "step": 3613 }, { "epoch": 0.19267473476568747, "grad_norm": 4.989001274108887, "learning_rate": 9.2774e-06, "loss": 0.0437, "step": 3614 }, { "epoch": 0.1927280481953404, "grad_norm": 6.737872123718262, "learning_rate": 9.2772e-06, "loss": 0.1049, "step": 3615 }, { "epoch": 0.19278136162499335, "grad_norm": 3.8937795162200928, "learning_rate": 9.277000000000001e-06, "loss": 0.0455, "step": 3616 }, { "epoch": 0.19283467505464627, "grad_norm": 6.117433071136475, "learning_rate": 9.2768e-06, "loss": 0.1083, "step": 3617 }, { "epoch": 0.1928879884842992, "grad_norm": 2.6919755935668945, "learning_rate": 9.276600000000002e-06, "loss": 0.0398, "step": 3618 }, { "epoch": 0.19294130191395212, "grad_norm": 4.073519706726074, "learning_rate": 9.276400000000002e-06, "loss": 0.0554, "step": 3619 }, { "epoch": 0.19299461534360504, "grad_norm": 2.9890522956848145, "learning_rate": 9.276200000000001e-06, "loss": 0.0364, "step": 3620 }, { "epoch": 0.193047928773258, "grad_norm": 6.434784412384033, "learning_rate": 9.276e-06, "loss": 0.0532, "step": 3621 }, { "epoch": 0.19310124220291092, "grad_norm": 2.4579079151153564, "learning_rate": 9.2758e-06, "loss": 0.0262, "step": 3622 }, { "epoch": 0.19315455563256384, "grad_norm": 2.4908926486968994, "learning_rate": 9.275600000000001e-06, "loss": 0.039, "step": 3623 }, { "epoch": 0.19320786906221676, "grad_norm": 4.721054553985596, "learning_rate": 9.275400000000001e-06, "loss": 0.0497, "step": 3624 }, { "epoch": 0.19326118249186972, "grad_norm": 7.887257099151611, "learning_rate": 9.2752e-06, "loss": 0.0535, "step": 3625 }, { "epoch": 0.19331449592152264, "grad_norm": 4.963194847106934, "learning_rate": 9.275e-06, "loss": 0.0862, "step": 3626 }, { "epoch": 0.19336780935117556, "grad_norm": 5.91717529296875, "learning_rate": 9.2748e-06, "loss": 0.0377, "step": 3627 }, { "epoch": 0.19342112278082849, "grad_norm": 3.5741395950317383, "learning_rate": 9.274600000000001e-06, "loss": 0.0338, "step": 3628 }, { "epoch": 0.1934744362104814, "grad_norm": 2.861024856567383, "learning_rate": 9.2744e-06, "loss": 0.0493, "step": 3629 }, { "epoch": 0.19352774964013436, "grad_norm": 4.146917819976807, "learning_rate": 9.274200000000002e-06, "loss": 0.0646, "step": 3630 }, { "epoch": 0.19358106306978728, "grad_norm": 4.382960319519043, "learning_rate": 9.274000000000001e-06, "loss": 0.0556, "step": 3631 }, { "epoch": 0.1936343764994402, "grad_norm": 4.085736274719238, "learning_rate": 9.273800000000001e-06, "loss": 0.0643, "step": 3632 }, { "epoch": 0.19368768992909313, "grad_norm": 4.029188632965088, "learning_rate": 9.2736e-06, "loss": 0.0516, "step": 3633 }, { "epoch": 0.19374100335874606, "grad_norm": 4.5857720375061035, "learning_rate": 9.273400000000002e-06, "loss": 0.0618, "step": 3634 }, { "epoch": 0.193794316788399, "grad_norm": 3.679624080657959, "learning_rate": 9.273200000000001e-06, "loss": 0.0412, "step": 3635 }, { "epoch": 0.19384763021805193, "grad_norm": 5.553173542022705, "learning_rate": 9.273e-06, "loss": 0.0403, "step": 3636 }, { "epoch": 0.19390094364770485, "grad_norm": 4.376071929931641, "learning_rate": 9.2728e-06, "loss": 0.0369, "step": 3637 }, { "epoch": 0.19395425707735778, "grad_norm": 4.38112735748291, "learning_rate": 9.2726e-06, "loss": 0.065, "step": 3638 }, { "epoch": 0.19400757050701073, "grad_norm": 5.896853446960449, "learning_rate": 9.2724e-06, "loss": 0.0675, "step": 3639 }, { "epoch": 0.19406088393666365, "grad_norm": 6.52122163772583, "learning_rate": 9.2722e-06, "loss": 0.0614, "step": 3640 }, { "epoch": 0.19411419736631658, "grad_norm": 5.04568338394165, "learning_rate": 9.272e-06, "loss": 0.0482, "step": 3641 }, { "epoch": 0.1941675107959695, "grad_norm": 3.827677011489868, "learning_rate": 9.271800000000002e-06, "loss": 0.0732, "step": 3642 }, { "epoch": 0.19422082422562242, "grad_norm": 5.83779239654541, "learning_rate": 9.271600000000001e-06, "loss": 0.06, "step": 3643 }, { "epoch": 0.19427413765527538, "grad_norm": 5.110632419586182, "learning_rate": 9.2714e-06, "loss": 0.0275, "step": 3644 }, { "epoch": 0.1943274510849283, "grad_norm": 2.8058035373687744, "learning_rate": 9.271200000000002e-06, "loss": 0.0343, "step": 3645 }, { "epoch": 0.19438076451458122, "grad_norm": 3.4726545810699463, "learning_rate": 9.271000000000002e-06, "loss": 0.0616, "step": 3646 }, { "epoch": 0.19443407794423415, "grad_norm": 3.8737261295318604, "learning_rate": 9.270800000000001e-06, "loss": 0.0511, "step": 3647 }, { "epoch": 0.19448739137388707, "grad_norm": 4.852327346801758, "learning_rate": 9.2706e-06, "loss": 0.0382, "step": 3648 }, { "epoch": 0.19454070480354002, "grad_norm": 5.388936519622803, "learning_rate": 9.2704e-06, "loss": 0.0694, "step": 3649 }, { "epoch": 0.19459401823319294, "grad_norm": 3.003664970397949, "learning_rate": 9.2702e-06, "loss": 0.0492, "step": 3650 }, { "epoch": 0.19464733166284587, "grad_norm": 2.1933276653289795, "learning_rate": 9.270000000000001e-06, "loss": 0.0256, "step": 3651 }, { "epoch": 0.1947006450924988, "grad_norm": 3.0711183547973633, "learning_rate": 9.2698e-06, "loss": 0.0197, "step": 3652 }, { "epoch": 0.19475395852215174, "grad_norm": 4.296849250793457, "learning_rate": 9.2696e-06, "loss": 0.0576, "step": 3653 }, { "epoch": 0.19480727195180467, "grad_norm": 3.203984260559082, "learning_rate": 9.269400000000001e-06, "loss": 0.0548, "step": 3654 }, { "epoch": 0.1948605853814576, "grad_norm": 2.3952996730804443, "learning_rate": 9.269200000000001e-06, "loss": 0.0489, "step": 3655 }, { "epoch": 0.19491389881111051, "grad_norm": 4.224237442016602, "learning_rate": 9.269e-06, "loss": 0.0368, "step": 3656 }, { "epoch": 0.19496721224076344, "grad_norm": 5.748708248138428, "learning_rate": 9.268800000000002e-06, "loss": 0.0817, "step": 3657 }, { "epoch": 0.1950205256704164, "grad_norm": 7.1706695556640625, "learning_rate": 9.268600000000001e-06, "loss": 0.0927, "step": 3658 }, { "epoch": 0.1950738391000693, "grad_norm": 4.351480484008789, "learning_rate": 9.268400000000001e-06, "loss": 0.0528, "step": 3659 }, { "epoch": 0.19512715252972224, "grad_norm": 6.977167129516602, "learning_rate": 9.2682e-06, "loss": 0.0617, "step": 3660 }, { "epoch": 0.19518046595937516, "grad_norm": 3.4644017219543457, "learning_rate": 9.268e-06, "loss": 0.0518, "step": 3661 }, { "epoch": 0.19523377938902808, "grad_norm": 4.062305927276611, "learning_rate": 9.267800000000001e-06, "loss": 0.0564, "step": 3662 }, { "epoch": 0.19528709281868104, "grad_norm": 5.013349533081055, "learning_rate": 9.267600000000001e-06, "loss": 0.0501, "step": 3663 }, { "epoch": 0.19534040624833396, "grad_norm": 4.4874372482299805, "learning_rate": 9.2674e-06, "loss": 0.0571, "step": 3664 }, { "epoch": 0.19539371967798688, "grad_norm": 3.028820514678955, "learning_rate": 9.2672e-06, "loss": 0.0453, "step": 3665 }, { "epoch": 0.1954470331076398, "grad_norm": 2.8741230964660645, "learning_rate": 9.267000000000001e-06, "loss": 0.0438, "step": 3666 }, { "epoch": 0.19550034653729276, "grad_norm": 4.593432426452637, "learning_rate": 9.2668e-06, "loss": 0.0667, "step": 3667 }, { "epoch": 0.19555365996694568, "grad_norm": 6.711014747619629, "learning_rate": 9.2666e-06, "loss": 0.0481, "step": 3668 }, { "epoch": 0.1956069733965986, "grad_norm": 5.094631195068359, "learning_rate": 9.266400000000002e-06, "loss": 0.0478, "step": 3669 }, { "epoch": 0.19566028682625153, "grad_norm": 5.8865132331848145, "learning_rate": 9.266200000000001e-06, "loss": 0.047, "step": 3670 }, { "epoch": 0.19571360025590445, "grad_norm": 4.054696083068848, "learning_rate": 9.266e-06, "loss": 0.0533, "step": 3671 }, { "epoch": 0.1957669136855574, "grad_norm": 5.731663703918457, "learning_rate": 9.2658e-06, "loss": 0.0638, "step": 3672 }, { "epoch": 0.19582022711521033, "grad_norm": 5.470729827880859, "learning_rate": 9.265600000000002e-06, "loss": 0.0942, "step": 3673 }, { "epoch": 0.19587354054486325, "grad_norm": 3.9908323287963867, "learning_rate": 9.265400000000001e-06, "loss": 0.0804, "step": 3674 }, { "epoch": 0.19592685397451617, "grad_norm": 6.234920501708984, "learning_rate": 9.2652e-06, "loss": 0.043, "step": 3675 }, { "epoch": 0.1959801674041691, "grad_norm": 5.603939533233643, "learning_rate": 9.265e-06, "loss": 0.0422, "step": 3676 }, { "epoch": 0.19603348083382205, "grad_norm": 4.664346218109131, "learning_rate": 9.2648e-06, "loss": 0.0406, "step": 3677 }, { "epoch": 0.19608679426347497, "grad_norm": 5.644549369812012, "learning_rate": 9.264600000000001e-06, "loss": 0.0755, "step": 3678 }, { "epoch": 0.1961401076931279, "grad_norm": 2.78810453414917, "learning_rate": 9.2644e-06, "loss": 0.0394, "step": 3679 }, { "epoch": 0.19619342112278082, "grad_norm": 3.8903229236602783, "learning_rate": 9.2642e-06, "loss": 0.0486, "step": 3680 }, { "epoch": 0.19624673455243377, "grad_norm": 4.771225929260254, "learning_rate": 9.264000000000001e-06, "loss": 0.0421, "step": 3681 }, { "epoch": 0.1963000479820867, "grad_norm": 6.817392349243164, "learning_rate": 9.263800000000001e-06, "loss": 0.0425, "step": 3682 }, { "epoch": 0.19635336141173962, "grad_norm": 4.230677127838135, "learning_rate": 9.2636e-06, "loss": 0.0408, "step": 3683 }, { "epoch": 0.19640667484139254, "grad_norm": 4.939212799072266, "learning_rate": 9.263400000000002e-06, "loss": 0.0743, "step": 3684 }, { "epoch": 0.19645998827104547, "grad_norm": 4.023117542266846, "learning_rate": 9.263200000000001e-06, "loss": 0.0548, "step": 3685 }, { "epoch": 0.19651330170069842, "grad_norm": 4.99441385269165, "learning_rate": 9.263000000000001e-06, "loss": 0.05, "step": 3686 }, { "epoch": 0.19656661513035134, "grad_norm": 3.891279697418213, "learning_rate": 9.2628e-06, "loss": 0.0433, "step": 3687 }, { "epoch": 0.19661992856000426, "grad_norm": 8.634451866149902, "learning_rate": 9.2626e-06, "loss": 0.0772, "step": 3688 }, { "epoch": 0.1966732419896572, "grad_norm": 4.703487396240234, "learning_rate": 9.2624e-06, "loss": 0.0411, "step": 3689 }, { "epoch": 0.1967265554193101, "grad_norm": 6.909204006195068, "learning_rate": 9.262200000000001e-06, "loss": 0.0444, "step": 3690 }, { "epoch": 0.19677986884896306, "grad_norm": 5.14766788482666, "learning_rate": 9.262e-06, "loss": 0.0683, "step": 3691 }, { "epoch": 0.196833182278616, "grad_norm": 3.3679616451263428, "learning_rate": 9.2618e-06, "loss": 0.0431, "step": 3692 }, { "epoch": 0.1968864957082689, "grad_norm": 2.456927537918091, "learning_rate": 9.261600000000001e-06, "loss": 0.0454, "step": 3693 }, { "epoch": 0.19693980913792183, "grad_norm": 2.932650089263916, "learning_rate": 9.2614e-06, "loss": 0.0426, "step": 3694 }, { "epoch": 0.19699312256757479, "grad_norm": 3.932548761367798, "learning_rate": 9.261200000000002e-06, "loss": 0.0456, "step": 3695 }, { "epoch": 0.1970464359972277, "grad_norm": 8.97085952758789, "learning_rate": 9.261000000000002e-06, "loss": 0.1106, "step": 3696 }, { "epoch": 0.19709974942688063, "grad_norm": 7.71624231338501, "learning_rate": 9.260800000000001e-06, "loss": 0.0832, "step": 3697 }, { "epoch": 0.19715306285653356, "grad_norm": 7.92278528213501, "learning_rate": 9.2606e-06, "loss": 0.0817, "step": 3698 }, { "epoch": 0.19720637628618648, "grad_norm": 10.446487426757812, "learning_rate": 9.2604e-06, "loss": 0.1049, "step": 3699 }, { "epoch": 0.19725968971583943, "grad_norm": 10.883957862854004, "learning_rate": 9.2602e-06, "loss": 0.0532, "step": 3700 }, { "epoch": 0.19731300314549235, "grad_norm": 7.297445774078369, "learning_rate": 9.260000000000001e-06, "loss": 0.0638, "step": 3701 }, { "epoch": 0.19736631657514528, "grad_norm": 3.442225456237793, "learning_rate": 9.2598e-06, "loss": 0.0439, "step": 3702 }, { "epoch": 0.1974196300047982, "grad_norm": 3.1088027954101562, "learning_rate": 9.2596e-06, "loss": 0.0365, "step": 3703 }, { "epoch": 0.19747294343445113, "grad_norm": 3.690473794937134, "learning_rate": 9.2594e-06, "loss": 0.0604, "step": 3704 }, { "epoch": 0.19752625686410408, "grad_norm": 4.0376458168029785, "learning_rate": 9.259200000000001e-06, "loss": 0.0495, "step": 3705 }, { "epoch": 0.197579570293757, "grad_norm": 9.7305326461792, "learning_rate": 9.259e-06, "loss": 0.0594, "step": 3706 }, { "epoch": 0.19763288372340992, "grad_norm": 7.73766565322876, "learning_rate": 9.258800000000002e-06, "loss": 0.0829, "step": 3707 }, { "epoch": 0.19768619715306285, "grad_norm": 5.549054145812988, "learning_rate": 9.258600000000001e-06, "loss": 0.0633, "step": 3708 }, { "epoch": 0.1977395105827158, "grad_norm": 4.260230541229248, "learning_rate": 9.258400000000001e-06, "loss": 0.0616, "step": 3709 }, { "epoch": 0.19779282401236872, "grad_norm": 8.309752464294434, "learning_rate": 9.2582e-06, "loss": 0.0976, "step": 3710 }, { "epoch": 0.19784613744202165, "grad_norm": 3.798435926437378, "learning_rate": 9.258e-06, "loss": 0.0607, "step": 3711 }, { "epoch": 0.19789945087167457, "grad_norm": 5.727705955505371, "learning_rate": 9.257800000000001e-06, "loss": 0.062, "step": 3712 }, { "epoch": 0.1979527643013275, "grad_norm": 3.203200101852417, "learning_rate": 9.257600000000001e-06, "loss": 0.0504, "step": 3713 }, { "epoch": 0.19800607773098045, "grad_norm": 4.392217636108398, "learning_rate": 9.2574e-06, "loss": 0.0522, "step": 3714 }, { "epoch": 0.19805939116063337, "grad_norm": 4.542960166931152, "learning_rate": 9.2572e-06, "loss": 0.072, "step": 3715 }, { "epoch": 0.1981127045902863, "grad_norm": 3.202455997467041, "learning_rate": 9.257e-06, "loss": 0.0358, "step": 3716 }, { "epoch": 0.19816601801993922, "grad_norm": 3.6028153896331787, "learning_rate": 9.256800000000001e-06, "loss": 0.0607, "step": 3717 }, { "epoch": 0.19821933144959214, "grad_norm": 4.460388660430908, "learning_rate": 9.2566e-06, "loss": 0.0704, "step": 3718 }, { "epoch": 0.1982726448792451, "grad_norm": 3.227388620376587, "learning_rate": 9.256400000000002e-06, "loss": 0.0499, "step": 3719 }, { "epoch": 0.19832595830889801, "grad_norm": 11.10088062286377, "learning_rate": 9.256200000000001e-06, "loss": 0.0788, "step": 3720 }, { "epoch": 0.19837927173855094, "grad_norm": 3.805443286895752, "learning_rate": 9.256e-06, "loss": 0.0451, "step": 3721 }, { "epoch": 0.19843258516820386, "grad_norm": 7.0081095695495605, "learning_rate": 9.2558e-06, "loss": 0.0612, "step": 3722 }, { "epoch": 0.1984858985978568, "grad_norm": 3.6596620082855225, "learning_rate": 9.255600000000002e-06, "loss": 0.0673, "step": 3723 }, { "epoch": 0.19853921202750974, "grad_norm": 4.773427486419678, "learning_rate": 9.255400000000001e-06, "loss": 0.0811, "step": 3724 }, { "epoch": 0.19859252545716266, "grad_norm": 3.5389535427093506, "learning_rate": 9.2552e-06, "loss": 0.029, "step": 3725 }, { "epoch": 0.19864583888681558, "grad_norm": 4.55683708190918, "learning_rate": 9.255e-06, "loss": 0.0641, "step": 3726 }, { "epoch": 0.1986991523164685, "grad_norm": 5.604912757873535, "learning_rate": 9.2548e-06, "loss": 0.0679, "step": 3727 }, { "epoch": 0.19875246574612146, "grad_norm": 3.757016181945801, "learning_rate": 9.2546e-06, "loss": 0.0537, "step": 3728 }, { "epoch": 0.19880577917577438, "grad_norm": 3.7407186031341553, "learning_rate": 9.2544e-06, "loss": 0.0436, "step": 3729 }, { "epoch": 0.1988590926054273, "grad_norm": 3.7651100158691406, "learning_rate": 9.2542e-06, "loss": 0.0456, "step": 3730 }, { "epoch": 0.19891240603508023, "grad_norm": 4.307519435882568, "learning_rate": 9.254000000000002e-06, "loss": 0.0362, "step": 3731 }, { "epoch": 0.19896571946473315, "grad_norm": 3.0457916259765625, "learning_rate": 9.253800000000001e-06, "loss": 0.0324, "step": 3732 }, { "epoch": 0.1990190328943861, "grad_norm": 4.034231185913086, "learning_rate": 9.2536e-06, "loss": 0.0671, "step": 3733 }, { "epoch": 0.19907234632403903, "grad_norm": 3.4579005241394043, "learning_rate": 9.253400000000002e-06, "loss": 0.0427, "step": 3734 }, { "epoch": 0.19912565975369195, "grad_norm": 4.811919689178467, "learning_rate": 9.253200000000001e-06, "loss": 0.044, "step": 3735 }, { "epoch": 0.19917897318334488, "grad_norm": 2.7892863750457764, "learning_rate": 9.253000000000001e-06, "loss": 0.0495, "step": 3736 }, { "epoch": 0.19923228661299783, "grad_norm": 4.419168472290039, "learning_rate": 9.2528e-06, "loss": 0.052, "step": 3737 }, { "epoch": 0.19928560004265075, "grad_norm": 13.387453079223633, "learning_rate": 9.2526e-06, "loss": 0.1024, "step": 3738 }, { "epoch": 0.19933891347230367, "grad_norm": 4.287874698638916, "learning_rate": 9.2524e-06, "loss": 0.058, "step": 3739 }, { "epoch": 0.1993922269019566, "grad_norm": 1.9441536664962769, "learning_rate": 9.252200000000001e-06, "loss": 0.0346, "step": 3740 }, { "epoch": 0.19944554033160952, "grad_norm": 5.306351184844971, "learning_rate": 9.252e-06, "loss": 0.0393, "step": 3741 }, { "epoch": 0.19949885376126247, "grad_norm": 8.714079856872559, "learning_rate": 9.2518e-06, "loss": 0.1125, "step": 3742 }, { "epoch": 0.1995521671909154, "grad_norm": 10.885123252868652, "learning_rate": 9.251600000000001e-06, "loss": 0.0724, "step": 3743 }, { "epoch": 0.19960548062056832, "grad_norm": 7.221037864685059, "learning_rate": 9.251400000000001e-06, "loss": 0.0753, "step": 3744 }, { "epoch": 0.19965879405022124, "grad_norm": 3.0374343395233154, "learning_rate": 9.2512e-06, "loss": 0.0332, "step": 3745 }, { "epoch": 0.19971210747987417, "grad_norm": 5.133808612823486, "learning_rate": 9.251000000000002e-06, "loss": 0.0786, "step": 3746 }, { "epoch": 0.19976542090952712, "grad_norm": 4.7593560218811035, "learning_rate": 9.250800000000001e-06, "loss": 0.0555, "step": 3747 }, { "epoch": 0.19981873433918004, "grad_norm": 7.302270889282227, "learning_rate": 9.2506e-06, "loss": 0.1026, "step": 3748 }, { "epoch": 0.19987204776883297, "grad_norm": 10.60986614227295, "learning_rate": 9.2504e-06, "loss": 0.0752, "step": 3749 }, { "epoch": 0.1999253611984859, "grad_norm": 5.218132495880127, "learning_rate": 9.2502e-06, "loss": 0.0727, "step": 3750 }, { "epoch": 0.19997867462813884, "grad_norm": 5.830289363861084, "learning_rate": 9.250000000000001e-06, "loss": 0.0762, "step": 3751 }, { "epoch": 0.20003198805779177, "grad_norm": 5.457309722900391, "learning_rate": 9.2498e-06, "loss": 0.0938, "step": 3752 }, { "epoch": 0.2000853014874447, "grad_norm": 7.448395729064941, "learning_rate": 9.2496e-06, "loss": 0.0427, "step": 3753 }, { "epoch": 0.2001386149170976, "grad_norm": 11.87810230255127, "learning_rate": 9.2494e-06, "loss": 0.0761, "step": 3754 }, { "epoch": 0.20019192834675054, "grad_norm": 6.560561656951904, "learning_rate": 9.249200000000001e-06, "loss": 0.0327, "step": 3755 }, { "epoch": 0.2002452417764035, "grad_norm": 6.482290267944336, "learning_rate": 9.249e-06, "loss": 0.0761, "step": 3756 }, { "epoch": 0.2002985552060564, "grad_norm": 3.9256677627563477, "learning_rate": 9.2488e-06, "loss": 0.0412, "step": 3757 }, { "epoch": 0.20035186863570933, "grad_norm": 2.8237483501434326, "learning_rate": 9.248600000000002e-06, "loss": 0.0512, "step": 3758 }, { "epoch": 0.20040518206536226, "grad_norm": 6.052872657775879, "learning_rate": 9.248400000000001e-06, "loss": 0.0391, "step": 3759 }, { "epoch": 0.20045849549501518, "grad_norm": 4.38685417175293, "learning_rate": 9.2482e-06, "loss": 0.0542, "step": 3760 }, { "epoch": 0.20051180892466813, "grad_norm": 11.86379623413086, "learning_rate": 9.248e-06, "loss": 0.1024, "step": 3761 }, { "epoch": 0.20056512235432106, "grad_norm": 5.588171482086182, "learning_rate": 9.247800000000001e-06, "loss": 0.0577, "step": 3762 }, { "epoch": 0.20061843578397398, "grad_norm": 4.102270126342773, "learning_rate": 9.247600000000001e-06, "loss": 0.0351, "step": 3763 }, { "epoch": 0.2006717492136269, "grad_norm": 3.5927014350891113, "learning_rate": 9.2474e-06, "loss": 0.0709, "step": 3764 }, { "epoch": 0.20072506264327986, "grad_norm": 2.087951183319092, "learning_rate": 9.2472e-06, "loss": 0.0393, "step": 3765 }, { "epoch": 0.20077837607293278, "grad_norm": 5.305668830871582, "learning_rate": 9.247e-06, "loss": 0.0798, "step": 3766 }, { "epoch": 0.2008316895025857, "grad_norm": 4.6068220138549805, "learning_rate": 9.246800000000001e-06, "loss": 0.0686, "step": 3767 }, { "epoch": 0.20088500293223863, "grad_norm": 3.500169515609741, "learning_rate": 9.2466e-06, "loss": 0.0602, "step": 3768 }, { "epoch": 0.20093831636189155, "grad_norm": 4.300352096557617, "learning_rate": 9.2464e-06, "loss": 0.0387, "step": 3769 }, { "epoch": 0.2009916297915445, "grad_norm": 4.603560447692871, "learning_rate": 9.246200000000001e-06, "loss": 0.0451, "step": 3770 }, { "epoch": 0.20104494322119743, "grad_norm": 2.907989978790283, "learning_rate": 9.246000000000001e-06, "loss": 0.0432, "step": 3771 }, { "epoch": 0.20109825665085035, "grad_norm": 4.181288719177246, "learning_rate": 9.2458e-06, "loss": 0.0684, "step": 3772 }, { "epoch": 0.20115157008050327, "grad_norm": 2.8265910148620605, "learning_rate": 9.245600000000002e-06, "loss": 0.0285, "step": 3773 }, { "epoch": 0.2012048835101562, "grad_norm": 2.32973051071167, "learning_rate": 9.245400000000001e-06, "loss": 0.0243, "step": 3774 }, { "epoch": 0.20125819693980915, "grad_norm": 4.430824279785156, "learning_rate": 9.245200000000001e-06, "loss": 0.0499, "step": 3775 }, { "epoch": 0.20131151036946207, "grad_norm": 3.231748342514038, "learning_rate": 9.245e-06, "loss": 0.037, "step": 3776 }, { "epoch": 0.201364823799115, "grad_norm": 6.134253978729248, "learning_rate": 9.2448e-06, "loss": 0.0824, "step": 3777 }, { "epoch": 0.20141813722876792, "grad_norm": 5.340794086456299, "learning_rate": 9.2446e-06, "loss": 0.039, "step": 3778 }, { "epoch": 0.20147145065842087, "grad_norm": 4.719887733459473, "learning_rate": 9.2444e-06, "loss": 0.0531, "step": 3779 }, { "epoch": 0.2015247640880738, "grad_norm": 6.034505844116211, "learning_rate": 9.2442e-06, "loss": 0.0499, "step": 3780 }, { "epoch": 0.20157807751772672, "grad_norm": 4.4790730476379395, "learning_rate": 9.244e-06, "loss": 0.0822, "step": 3781 }, { "epoch": 0.20163139094737964, "grad_norm": 2.5572071075439453, "learning_rate": 9.243800000000001e-06, "loss": 0.0457, "step": 3782 }, { "epoch": 0.20168470437703256, "grad_norm": 2.710160493850708, "learning_rate": 9.2436e-06, "loss": 0.0264, "step": 3783 }, { "epoch": 0.20173801780668552, "grad_norm": 4.827866554260254, "learning_rate": 9.243400000000002e-06, "loss": 0.0503, "step": 3784 }, { "epoch": 0.20179133123633844, "grad_norm": 3.734079360961914, "learning_rate": 9.243200000000002e-06, "loss": 0.035, "step": 3785 }, { "epoch": 0.20184464466599136, "grad_norm": 5.507885456085205, "learning_rate": 9.243000000000001e-06, "loss": 0.0487, "step": 3786 }, { "epoch": 0.2018979580956443, "grad_norm": 2.4364774227142334, "learning_rate": 9.2428e-06, "loss": 0.0341, "step": 3787 }, { "epoch": 0.2019512715252972, "grad_norm": 5.554754257202148, "learning_rate": 9.2426e-06, "loss": 0.0737, "step": 3788 }, { "epoch": 0.20200458495495016, "grad_norm": 5.274206161499023, "learning_rate": 9.2424e-06, "loss": 0.0711, "step": 3789 }, { "epoch": 0.20205789838460309, "grad_norm": 6.707823276519775, "learning_rate": 9.242200000000001e-06, "loss": 0.0719, "step": 3790 }, { "epoch": 0.202111211814256, "grad_norm": 2.752843141555786, "learning_rate": 9.242e-06, "loss": 0.0253, "step": 3791 }, { "epoch": 0.20216452524390893, "grad_norm": 6.153560638427734, "learning_rate": 9.2418e-06, "loss": 0.0965, "step": 3792 }, { "epoch": 0.20221783867356188, "grad_norm": 4.130615711212158, "learning_rate": 9.2416e-06, "loss": 0.064, "step": 3793 }, { "epoch": 0.2022711521032148, "grad_norm": 3.725513219833374, "learning_rate": 9.241400000000001e-06, "loss": 0.0294, "step": 3794 }, { "epoch": 0.20232446553286773, "grad_norm": 7.3921637535095215, "learning_rate": 9.2412e-06, "loss": 0.1072, "step": 3795 }, { "epoch": 0.20237777896252065, "grad_norm": 2.219024896621704, "learning_rate": 9.241000000000002e-06, "loss": 0.0182, "step": 3796 }, { "epoch": 0.20243109239217358, "grad_norm": 3.311013698577881, "learning_rate": 9.240800000000001e-06, "loss": 0.0463, "step": 3797 }, { "epoch": 0.20248440582182653, "grad_norm": 3.1482744216918945, "learning_rate": 9.240600000000001e-06, "loss": 0.0441, "step": 3798 }, { "epoch": 0.20253771925147945, "grad_norm": 3.641695261001587, "learning_rate": 9.2404e-06, "loss": 0.0476, "step": 3799 }, { "epoch": 0.20259103268113238, "grad_norm": 3.2516043186187744, "learning_rate": 9.2402e-06, "loss": 0.0525, "step": 3800 }, { "epoch": 0.2026443461107853, "grad_norm": 7.66985559463501, "learning_rate": 9.240000000000001e-06, "loss": 0.125, "step": 3801 }, { "epoch": 0.20269765954043822, "grad_norm": 3.941779136657715, "learning_rate": 9.239800000000001e-06, "loss": 0.0547, "step": 3802 }, { "epoch": 0.20275097297009118, "grad_norm": 3.9968860149383545, "learning_rate": 9.2396e-06, "loss": 0.0849, "step": 3803 }, { "epoch": 0.2028042863997441, "grad_norm": 4.142568111419678, "learning_rate": 9.2394e-06, "loss": 0.0635, "step": 3804 }, { "epoch": 0.20285759982939702, "grad_norm": 4.9272565841674805, "learning_rate": 9.2392e-06, "loss": 0.0544, "step": 3805 }, { "epoch": 0.20291091325904995, "grad_norm": 5.316062927246094, "learning_rate": 9.239e-06, "loss": 0.0395, "step": 3806 }, { "epoch": 0.2029642266887029, "grad_norm": 3.8013250827789307, "learning_rate": 9.2388e-06, "loss": 0.0463, "step": 3807 }, { "epoch": 0.20301754011835582, "grad_norm": 4.758338928222656, "learning_rate": 9.238600000000002e-06, "loss": 0.0742, "step": 3808 }, { "epoch": 0.20307085354800875, "grad_norm": 2.3924179077148438, "learning_rate": 9.238400000000001e-06, "loss": 0.0351, "step": 3809 }, { "epoch": 0.20312416697766167, "grad_norm": 5.377312660217285, "learning_rate": 9.2382e-06, "loss": 0.0501, "step": 3810 }, { "epoch": 0.2031774804073146, "grad_norm": 3.8538525104522705, "learning_rate": 9.238e-06, "loss": 0.0739, "step": 3811 }, { "epoch": 0.20323079383696754, "grad_norm": 2.835181474685669, "learning_rate": 9.237800000000002e-06, "loss": 0.0533, "step": 3812 }, { "epoch": 0.20328410726662047, "grad_norm": 3.954650402069092, "learning_rate": 9.237600000000001e-06, "loss": 0.0522, "step": 3813 }, { "epoch": 0.2033374206962734, "grad_norm": 2.912071704864502, "learning_rate": 9.2374e-06, "loss": 0.0446, "step": 3814 }, { "epoch": 0.20339073412592631, "grad_norm": 2.7825939655303955, "learning_rate": 9.2372e-06, "loss": 0.0436, "step": 3815 }, { "epoch": 0.20344404755557924, "grad_norm": 8.549111366271973, "learning_rate": 9.237e-06, "loss": 0.0934, "step": 3816 }, { "epoch": 0.2034973609852322, "grad_norm": 4.000372409820557, "learning_rate": 9.2368e-06, "loss": 0.0781, "step": 3817 }, { "epoch": 0.2035506744148851, "grad_norm": 2.725205183029175, "learning_rate": 9.2366e-06, "loss": 0.0438, "step": 3818 }, { "epoch": 0.20360398784453804, "grad_norm": 6.246260643005371, "learning_rate": 9.2364e-06, "loss": 0.0611, "step": 3819 }, { "epoch": 0.20365730127419096, "grad_norm": 7.936162948608398, "learning_rate": 9.236200000000001e-06, "loss": 0.0769, "step": 3820 }, { "epoch": 0.2037106147038439, "grad_norm": 3.4424400329589844, "learning_rate": 9.236000000000001e-06, "loss": 0.038, "step": 3821 }, { "epoch": 0.20376392813349684, "grad_norm": 3.1566720008850098, "learning_rate": 9.2358e-06, "loss": 0.0436, "step": 3822 }, { "epoch": 0.20381724156314976, "grad_norm": 3.398958921432495, "learning_rate": 9.235600000000002e-06, "loss": 0.0481, "step": 3823 }, { "epoch": 0.20387055499280268, "grad_norm": 3.4665353298187256, "learning_rate": 9.235400000000001e-06, "loss": 0.0313, "step": 3824 }, { "epoch": 0.2039238684224556, "grad_norm": 3.369814872741699, "learning_rate": 9.235200000000001e-06, "loss": 0.0584, "step": 3825 }, { "epoch": 0.20397718185210856, "grad_norm": 3.629335880279541, "learning_rate": 9.235e-06, "loss": 0.0343, "step": 3826 }, { "epoch": 0.20403049528176148, "grad_norm": 5.285756587982178, "learning_rate": 9.2348e-06, "loss": 0.1214, "step": 3827 }, { "epoch": 0.2040838087114144, "grad_norm": 5.8630690574646, "learning_rate": 9.2346e-06, "loss": 0.0711, "step": 3828 }, { "epoch": 0.20413712214106733, "grad_norm": 3.825690746307373, "learning_rate": 9.234400000000001e-06, "loss": 0.0475, "step": 3829 }, { "epoch": 0.20419043557072025, "grad_norm": 3.577045440673828, "learning_rate": 9.2342e-06, "loss": 0.0444, "step": 3830 }, { "epoch": 0.2042437490003732, "grad_norm": 3.9810924530029297, "learning_rate": 9.234e-06, "loss": 0.0559, "step": 3831 }, { "epoch": 0.20429706243002613, "grad_norm": 5.421412944793701, "learning_rate": 9.233800000000001e-06, "loss": 0.0662, "step": 3832 }, { "epoch": 0.20435037585967905, "grad_norm": 2.816617965698242, "learning_rate": 9.2336e-06, "loss": 0.03, "step": 3833 }, { "epoch": 0.20440368928933197, "grad_norm": 2.3753139972686768, "learning_rate": 9.233400000000002e-06, "loss": 0.0231, "step": 3834 }, { "epoch": 0.20445700271898493, "grad_norm": 4.138454914093018, "learning_rate": 9.233200000000002e-06, "loss": 0.0392, "step": 3835 }, { "epoch": 0.20451031614863785, "grad_norm": 2.5226831436157227, "learning_rate": 9.233000000000001e-06, "loss": 0.0391, "step": 3836 }, { "epoch": 0.20456362957829077, "grad_norm": 3.952315330505371, "learning_rate": 9.2328e-06, "loss": 0.0421, "step": 3837 }, { "epoch": 0.2046169430079437, "grad_norm": 2.82348370552063, "learning_rate": 9.2326e-06, "loss": 0.0383, "step": 3838 }, { "epoch": 0.20467025643759662, "grad_norm": 2.7417612075805664, "learning_rate": 9.2324e-06, "loss": 0.0686, "step": 3839 }, { "epoch": 0.20472356986724957, "grad_norm": 4.643013000488281, "learning_rate": 9.232200000000001e-06, "loss": 0.062, "step": 3840 }, { "epoch": 0.2047768832969025, "grad_norm": 2.7338926792144775, "learning_rate": 9.232e-06, "loss": 0.0601, "step": 3841 }, { "epoch": 0.20483019672655542, "grad_norm": 3.0602004528045654, "learning_rate": 9.2318e-06, "loss": 0.0532, "step": 3842 }, { "epoch": 0.20488351015620834, "grad_norm": 2.780085563659668, "learning_rate": 9.2316e-06, "loss": 0.0469, "step": 3843 }, { "epoch": 0.20493682358586127, "grad_norm": 4.14665412902832, "learning_rate": 9.231400000000001e-06, "loss": 0.0726, "step": 3844 }, { "epoch": 0.20499013701551422, "grad_norm": 4.95853328704834, "learning_rate": 9.2312e-06, "loss": 0.0419, "step": 3845 }, { "epoch": 0.20504345044516714, "grad_norm": 6.008606433868408, "learning_rate": 9.231000000000002e-06, "loss": 0.0699, "step": 3846 }, { "epoch": 0.20509676387482006, "grad_norm": 3.4426732063293457, "learning_rate": 9.230800000000001e-06, "loss": 0.0407, "step": 3847 }, { "epoch": 0.205150077304473, "grad_norm": 4.174365043640137, "learning_rate": 9.230600000000001e-06, "loss": 0.0305, "step": 3848 }, { "epoch": 0.20520339073412594, "grad_norm": 2.7560782432556152, "learning_rate": 9.2304e-06, "loss": 0.0317, "step": 3849 }, { "epoch": 0.20525670416377886, "grad_norm": 9.046004295349121, "learning_rate": 9.2302e-06, "loss": 0.0857, "step": 3850 }, { "epoch": 0.2053100175934318, "grad_norm": 4.287853240966797, "learning_rate": 9.230000000000001e-06, "loss": 0.0509, "step": 3851 }, { "epoch": 0.2053633310230847, "grad_norm": 4.924942970275879, "learning_rate": 9.229800000000001e-06, "loss": 0.0403, "step": 3852 }, { "epoch": 0.20541664445273763, "grad_norm": 10.338829040527344, "learning_rate": 9.2296e-06, "loss": 0.0956, "step": 3853 }, { "epoch": 0.20546995788239059, "grad_norm": 4.785990238189697, "learning_rate": 9.2294e-06, "loss": 0.0505, "step": 3854 }, { "epoch": 0.2055232713120435, "grad_norm": 6.036256790161133, "learning_rate": 9.2292e-06, "loss": 0.0415, "step": 3855 }, { "epoch": 0.20557658474169643, "grad_norm": 4.361359119415283, "learning_rate": 9.229000000000001e-06, "loss": 0.0551, "step": 3856 }, { "epoch": 0.20562989817134936, "grad_norm": 3.188086986541748, "learning_rate": 9.2288e-06, "loss": 0.0446, "step": 3857 }, { "epoch": 0.20568321160100228, "grad_norm": 6.764862537384033, "learning_rate": 9.228600000000002e-06, "loss": 0.0738, "step": 3858 }, { "epoch": 0.20573652503065523, "grad_norm": 10.864121437072754, "learning_rate": 9.228400000000001e-06, "loss": 0.1193, "step": 3859 }, { "epoch": 0.20578983846030816, "grad_norm": 2.7636795043945312, "learning_rate": 9.228200000000001e-06, "loss": 0.0349, "step": 3860 }, { "epoch": 0.20584315188996108, "grad_norm": 3.2718100547790527, "learning_rate": 9.228e-06, "loss": 0.077, "step": 3861 }, { "epoch": 0.205896465319614, "grad_norm": 4.32207727432251, "learning_rate": 9.227800000000002e-06, "loss": 0.0518, "step": 3862 }, { "epoch": 0.20594977874926695, "grad_norm": 9.360406875610352, "learning_rate": 9.227600000000001e-06, "loss": 0.081, "step": 3863 }, { "epoch": 0.20600309217891988, "grad_norm": 4.0974531173706055, "learning_rate": 9.2274e-06, "loss": 0.05, "step": 3864 }, { "epoch": 0.2060564056085728, "grad_norm": 3.3445377349853516, "learning_rate": 9.2272e-06, "loss": 0.0575, "step": 3865 }, { "epoch": 0.20610971903822572, "grad_norm": 4.012585639953613, "learning_rate": 9.227e-06, "loss": 0.0517, "step": 3866 }, { "epoch": 0.20616303246787865, "grad_norm": 4.1633195877075195, "learning_rate": 9.2268e-06, "loss": 0.0653, "step": 3867 }, { "epoch": 0.2062163458975316, "grad_norm": 2.921135187149048, "learning_rate": 9.2266e-06, "loss": 0.0403, "step": 3868 }, { "epoch": 0.20626965932718452, "grad_norm": 3.192434549331665, "learning_rate": 9.2264e-06, "loss": 0.0494, "step": 3869 }, { "epoch": 0.20632297275683745, "grad_norm": 2.773054838180542, "learning_rate": 9.226200000000002e-06, "loss": 0.0382, "step": 3870 }, { "epoch": 0.20637628618649037, "grad_norm": 2.9560329914093018, "learning_rate": 9.226000000000001e-06, "loss": 0.0375, "step": 3871 }, { "epoch": 0.2064295996161433, "grad_norm": 2.1396865844726562, "learning_rate": 9.2258e-06, "loss": 0.0284, "step": 3872 }, { "epoch": 0.20648291304579625, "grad_norm": 3.028299570083618, "learning_rate": 9.225600000000002e-06, "loss": 0.0333, "step": 3873 }, { "epoch": 0.20653622647544917, "grad_norm": 4.9887824058532715, "learning_rate": 9.225400000000001e-06, "loss": 0.0585, "step": 3874 }, { "epoch": 0.2065895399051021, "grad_norm": 3.2394275665283203, "learning_rate": 9.225200000000001e-06, "loss": 0.0406, "step": 3875 }, { "epoch": 0.20664285333475502, "grad_norm": 5.323122978210449, "learning_rate": 9.225e-06, "loss": 0.0404, "step": 3876 }, { "epoch": 0.20669616676440797, "grad_norm": 4.3613457679748535, "learning_rate": 9.2248e-06, "loss": 0.0752, "step": 3877 }, { "epoch": 0.2067494801940609, "grad_norm": 4.833776473999023, "learning_rate": 9.2246e-06, "loss": 0.0453, "step": 3878 }, { "epoch": 0.20680279362371382, "grad_norm": 3.1486244201660156, "learning_rate": 9.224400000000001e-06, "loss": 0.0279, "step": 3879 }, { "epoch": 0.20685610705336674, "grad_norm": 4.275340557098389, "learning_rate": 9.2242e-06, "loss": 0.0493, "step": 3880 }, { "epoch": 0.20690942048301966, "grad_norm": 5.005548000335693, "learning_rate": 9.224e-06, "loss": 0.0632, "step": 3881 }, { "epoch": 0.20696273391267261, "grad_norm": 2.823493480682373, "learning_rate": 9.223800000000001e-06, "loss": 0.0469, "step": 3882 }, { "epoch": 0.20701604734232554, "grad_norm": 7.227517604827881, "learning_rate": 9.223600000000001e-06, "loss": 0.0479, "step": 3883 }, { "epoch": 0.20706936077197846, "grad_norm": 3.8649885654449463, "learning_rate": 9.2234e-06, "loss": 0.0712, "step": 3884 }, { "epoch": 0.20712267420163138, "grad_norm": 4.991369247436523, "learning_rate": 9.223200000000002e-06, "loss": 0.0463, "step": 3885 }, { "epoch": 0.2071759876312843, "grad_norm": 6.031644821166992, "learning_rate": 9.223000000000001e-06, "loss": 0.082, "step": 3886 }, { "epoch": 0.20722930106093726, "grad_norm": 6.455522060394287, "learning_rate": 9.222800000000001e-06, "loss": 0.0422, "step": 3887 }, { "epoch": 0.20728261449059018, "grad_norm": 3.7685329914093018, "learning_rate": 9.2226e-06, "loss": 0.0619, "step": 3888 }, { "epoch": 0.2073359279202431, "grad_norm": 4.07440185546875, "learning_rate": 9.2224e-06, "loss": 0.0245, "step": 3889 }, { "epoch": 0.20738924134989603, "grad_norm": 2.6669771671295166, "learning_rate": 9.222200000000001e-06, "loss": 0.0355, "step": 3890 }, { "epoch": 0.20744255477954898, "grad_norm": 3.5745418071746826, "learning_rate": 9.222e-06, "loss": 0.0368, "step": 3891 }, { "epoch": 0.2074958682092019, "grad_norm": 7.615664005279541, "learning_rate": 9.2218e-06, "loss": 0.0987, "step": 3892 }, { "epoch": 0.20754918163885483, "grad_norm": 4.134769439697266, "learning_rate": 9.2216e-06, "loss": 0.0566, "step": 3893 }, { "epoch": 0.20760249506850775, "grad_norm": 5.866201877593994, "learning_rate": 9.221400000000001e-06, "loss": 0.0766, "step": 3894 }, { "epoch": 0.20765580849816068, "grad_norm": 3.4006130695343018, "learning_rate": 9.2212e-06, "loss": 0.0416, "step": 3895 }, { "epoch": 0.20770912192781363, "grad_norm": 2.650639295578003, "learning_rate": 9.221e-06, "loss": 0.0182, "step": 3896 }, { "epoch": 0.20776243535746655, "grad_norm": 3.3157601356506348, "learning_rate": 9.220800000000002e-06, "loss": 0.0322, "step": 3897 }, { "epoch": 0.20781574878711948, "grad_norm": 3.1867401599884033, "learning_rate": 9.220600000000001e-06, "loss": 0.0261, "step": 3898 }, { "epoch": 0.2078690622167724, "grad_norm": 4.00164794921875, "learning_rate": 9.2204e-06, "loss": 0.0711, "step": 3899 }, { "epoch": 0.20792237564642532, "grad_norm": 7.840385437011719, "learning_rate": 9.2202e-06, "loss": 0.0359, "step": 3900 }, { "epoch": 0.20797568907607827, "grad_norm": 2.0941436290740967, "learning_rate": 9.220000000000002e-06, "loss": 0.0363, "step": 3901 }, { "epoch": 0.2080290025057312, "grad_norm": 6.0279436111450195, "learning_rate": 9.219800000000001e-06, "loss": 0.0593, "step": 3902 }, { "epoch": 0.20808231593538412, "grad_norm": 3.1392157077789307, "learning_rate": 9.2196e-06, "loss": 0.0383, "step": 3903 }, { "epoch": 0.20813562936503704, "grad_norm": 5.325164794921875, "learning_rate": 9.2194e-06, "loss": 0.0696, "step": 3904 }, { "epoch": 0.20818894279469, "grad_norm": 2.846278667449951, "learning_rate": 9.2192e-06, "loss": 0.0305, "step": 3905 }, { "epoch": 0.20824225622434292, "grad_norm": 4.306515693664551, "learning_rate": 9.219000000000001e-06, "loss": 0.0793, "step": 3906 }, { "epoch": 0.20829556965399584, "grad_norm": 5.221813678741455, "learning_rate": 9.2188e-06, "loss": 0.0791, "step": 3907 }, { "epoch": 0.20834888308364877, "grad_norm": 3.448808193206787, "learning_rate": 9.2186e-06, "loss": 0.0402, "step": 3908 }, { "epoch": 0.2084021965133017, "grad_norm": 2.7778403759002686, "learning_rate": 9.218400000000001e-06, "loss": 0.047, "step": 3909 }, { "epoch": 0.20845550994295464, "grad_norm": 5.319943428039551, "learning_rate": 9.218200000000001e-06, "loss": 0.0379, "step": 3910 }, { "epoch": 0.20850882337260757, "grad_norm": 4.256800174713135, "learning_rate": 9.218e-06, "loss": 0.0636, "step": 3911 }, { "epoch": 0.2085621368022605, "grad_norm": 5.319637298583984, "learning_rate": 9.217800000000002e-06, "loss": 0.0481, "step": 3912 }, { "epoch": 0.2086154502319134, "grad_norm": 5.10732364654541, "learning_rate": 9.217600000000001e-06, "loss": 0.0459, "step": 3913 }, { "epoch": 0.20866876366156634, "grad_norm": 4.318622589111328, "learning_rate": 9.217400000000001e-06, "loss": 0.075, "step": 3914 }, { "epoch": 0.2087220770912193, "grad_norm": 4.440527439117432, "learning_rate": 9.2172e-06, "loss": 0.0728, "step": 3915 }, { "epoch": 0.2087753905208722, "grad_norm": 3.2889907360076904, "learning_rate": 9.217e-06, "loss": 0.0429, "step": 3916 }, { "epoch": 0.20882870395052514, "grad_norm": 2.196174383163452, "learning_rate": 9.216800000000001e-06, "loss": 0.0496, "step": 3917 }, { "epoch": 0.20888201738017806, "grad_norm": 3.5477888584136963, "learning_rate": 9.2166e-06, "loss": 0.0504, "step": 3918 }, { "epoch": 0.208935330809831, "grad_norm": 4.7415008544921875, "learning_rate": 9.2164e-06, "loss": 0.0694, "step": 3919 }, { "epoch": 0.20898864423948393, "grad_norm": 4.3617844581604, "learning_rate": 9.2162e-06, "loss": 0.0669, "step": 3920 }, { "epoch": 0.20904195766913686, "grad_norm": 3.7498669624328613, "learning_rate": 9.216000000000001e-06, "loss": 0.0551, "step": 3921 }, { "epoch": 0.20909527109878978, "grad_norm": 3.850452184677124, "learning_rate": 9.2158e-06, "loss": 0.0624, "step": 3922 }, { "epoch": 0.2091485845284427, "grad_norm": 5.224884510040283, "learning_rate": 9.215600000000002e-06, "loss": 0.0631, "step": 3923 }, { "epoch": 0.20920189795809566, "grad_norm": 3.5646464824676514, "learning_rate": 9.215400000000002e-06, "loss": 0.075, "step": 3924 }, { "epoch": 0.20925521138774858, "grad_norm": 3.291691541671753, "learning_rate": 9.215200000000001e-06, "loss": 0.0543, "step": 3925 }, { "epoch": 0.2093085248174015, "grad_norm": 6.052486896514893, "learning_rate": 9.215e-06, "loss": 0.0454, "step": 3926 }, { "epoch": 0.20936183824705443, "grad_norm": 4.017720699310303, "learning_rate": 9.2148e-06, "loss": 0.0425, "step": 3927 }, { "epoch": 0.20941515167670735, "grad_norm": 2.669447660446167, "learning_rate": 9.2146e-06, "loss": 0.0178, "step": 3928 }, { "epoch": 0.2094684651063603, "grad_norm": 7.681480884552002, "learning_rate": 9.214400000000001e-06, "loss": 0.0853, "step": 3929 }, { "epoch": 0.20952177853601323, "grad_norm": 4.97586727142334, "learning_rate": 9.2142e-06, "loss": 0.0689, "step": 3930 }, { "epoch": 0.20957509196566615, "grad_norm": 2.525432586669922, "learning_rate": 9.214e-06, "loss": 0.0278, "step": 3931 }, { "epoch": 0.20962840539531907, "grad_norm": 4.352752208709717, "learning_rate": 9.2138e-06, "loss": 0.0299, "step": 3932 }, { "epoch": 0.20968171882497202, "grad_norm": 4.406512260437012, "learning_rate": 9.213600000000001e-06, "loss": 0.0455, "step": 3933 }, { "epoch": 0.20973503225462495, "grad_norm": 2.7409040927886963, "learning_rate": 9.2134e-06, "loss": 0.0581, "step": 3934 }, { "epoch": 0.20978834568427787, "grad_norm": 3.4992451667785645, "learning_rate": 9.213200000000002e-06, "loss": 0.0339, "step": 3935 }, { "epoch": 0.2098416591139308, "grad_norm": 5.083933353424072, "learning_rate": 9.213000000000001e-06, "loss": 0.087, "step": 3936 }, { "epoch": 0.20989497254358372, "grad_norm": 6.528805732727051, "learning_rate": 9.212800000000001e-06, "loss": 0.1234, "step": 3937 }, { "epoch": 0.20994828597323667, "grad_norm": 6.110654354095459, "learning_rate": 9.2126e-06, "loss": 0.069, "step": 3938 }, { "epoch": 0.2100015994028896, "grad_norm": 5.617485046386719, "learning_rate": 9.2124e-06, "loss": 0.0456, "step": 3939 }, { "epoch": 0.21005491283254252, "grad_norm": 6.372354030609131, "learning_rate": 9.212200000000001e-06, "loss": 0.059, "step": 3940 }, { "epoch": 0.21010822626219544, "grad_norm": 3.733677864074707, "learning_rate": 9.212000000000001e-06, "loss": 0.0941, "step": 3941 }, { "epoch": 0.21016153969184836, "grad_norm": 4.304776191711426, "learning_rate": 9.2118e-06, "loss": 0.0561, "step": 3942 }, { "epoch": 0.21021485312150132, "grad_norm": 2.858966588973999, "learning_rate": 9.2116e-06, "loss": 0.0226, "step": 3943 }, { "epoch": 0.21026816655115424, "grad_norm": 3.4856529235839844, "learning_rate": 9.2114e-06, "loss": 0.0495, "step": 3944 }, { "epoch": 0.21032147998080716, "grad_norm": 1.3714408874511719, "learning_rate": 9.211200000000001e-06, "loss": 0.0107, "step": 3945 }, { "epoch": 0.2103747934104601, "grad_norm": 3.697030544281006, "learning_rate": 9.211e-06, "loss": 0.0385, "step": 3946 }, { "epoch": 0.21042810684011304, "grad_norm": 2.242522716522217, "learning_rate": 9.210800000000002e-06, "loss": 0.0203, "step": 3947 }, { "epoch": 0.21048142026976596, "grad_norm": 3.137598991394043, "learning_rate": 9.210600000000001e-06, "loss": 0.0443, "step": 3948 }, { "epoch": 0.21053473369941889, "grad_norm": 3.973924160003662, "learning_rate": 9.2104e-06, "loss": 0.0601, "step": 3949 }, { "epoch": 0.2105880471290718, "grad_norm": 2.839646100997925, "learning_rate": 9.2102e-06, "loss": 0.0248, "step": 3950 }, { "epoch": 0.21064136055872473, "grad_norm": 3.343135118484497, "learning_rate": 9.210000000000002e-06, "loss": 0.0728, "step": 3951 }, { "epoch": 0.21069467398837768, "grad_norm": 3.9118833541870117, "learning_rate": 9.209800000000001e-06, "loss": 0.044, "step": 3952 }, { "epoch": 0.2107479874180306, "grad_norm": 5.1226630210876465, "learning_rate": 9.2096e-06, "loss": 0.0638, "step": 3953 }, { "epoch": 0.21080130084768353, "grad_norm": 3.351517915725708, "learning_rate": 9.2094e-06, "loss": 0.0331, "step": 3954 }, { "epoch": 0.21085461427733646, "grad_norm": 4.414608955383301, "learning_rate": 9.2092e-06, "loss": 0.0495, "step": 3955 }, { "epoch": 0.21090792770698938, "grad_norm": 4.655821323394775, "learning_rate": 9.209000000000001e-06, "loss": 0.0448, "step": 3956 }, { "epoch": 0.21096124113664233, "grad_norm": 4.536909103393555, "learning_rate": 9.2088e-06, "loss": 0.0392, "step": 3957 }, { "epoch": 0.21101455456629525, "grad_norm": 2.756530523300171, "learning_rate": 9.2086e-06, "loss": 0.0305, "step": 3958 }, { "epoch": 0.21106786799594818, "grad_norm": 3.889251232147217, "learning_rate": 9.208400000000001e-06, "loss": 0.0734, "step": 3959 }, { "epoch": 0.2111211814256011, "grad_norm": 5.322640419006348, "learning_rate": 9.208200000000001e-06, "loss": 0.0363, "step": 3960 }, { "epoch": 0.21117449485525402, "grad_norm": 7.148506164550781, "learning_rate": 9.208e-06, "loss": 0.0514, "step": 3961 }, { "epoch": 0.21122780828490698, "grad_norm": 5.3037238121032715, "learning_rate": 9.207800000000002e-06, "loss": 0.033, "step": 3962 }, { "epoch": 0.2112811217145599, "grad_norm": 4.197885036468506, "learning_rate": 9.207600000000001e-06, "loss": 0.0532, "step": 3963 }, { "epoch": 0.21133443514421282, "grad_norm": 4.261761665344238, "learning_rate": 9.207400000000001e-06, "loss": 0.0289, "step": 3964 }, { "epoch": 0.21138774857386575, "grad_norm": 7.077730178833008, "learning_rate": 9.2072e-06, "loss": 0.0642, "step": 3965 }, { "epoch": 0.2114410620035187, "grad_norm": 5.809256553649902, "learning_rate": 9.207e-06, "loss": 0.0531, "step": 3966 }, { "epoch": 0.21149437543317162, "grad_norm": 6.2617597579956055, "learning_rate": 9.206800000000001e-06, "loss": 0.0781, "step": 3967 }, { "epoch": 0.21154768886282455, "grad_norm": 4.366542339324951, "learning_rate": 9.206600000000001e-06, "loss": 0.0686, "step": 3968 }, { "epoch": 0.21160100229247747, "grad_norm": 5.190414905548096, "learning_rate": 9.2064e-06, "loss": 0.056, "step": 3969 }, { "epoch": 0.2116543157221304, "grad_norm": 6.262243270874023, "learning_rate": 9.2062e-06, "loss": 0.0449, "step": 3970 }, { "epoch": 0.21170762915178334, "grad_norm": 13.937419891357422, "learning_rate": 9.206000000000001e-06, "loss": 0.095, "step": 3971 }, { "epoch": 0.21176094258143627, "grad_norm": 6.9753007888793945, "learning_rate": 9.205800000000001e-06, "loss": 0.0977, "step": 3972 }, { "epoch": 0.2118142560110892, "grad_norm": 4.179386138916016, "learning_rate": 9.2056e-06, "loss": 0.0524, "step": 3973 }, { "epoch": 0.21186756944074212, "grad_norm": 5.534088611602783, "learning_rate": 9.205400000000002e-06, "loss": 0.0775, "step": 3974 }, { "epoch": 0.21192088287039504, "grad_norm": 5.557461261749268, "learning_rate": 9.205200000000001e-06, "loss": 0.0579, "step": 3975 }, { "epoch": 0.211974196300048, "grad_norm": 2.1042473316192627, "learning_rate": 9.205e-06, "loss": 0.0286, "step": 3976 }, { "epoch": 0.2120275097297009, "grad_norm": 4.256715774536133, "learning_rate": 9.2048e-06, "loss": 0.0437, "step": 3977 }, { "epoch": 0.21208082315935384, "grad_norm": 3.2886064052581787, "learning_rate": 9.2046e-06, "loss": 0.0344, "step": 3978 }, { "epoch": 0.21213413658900676, "grad_norm": 3.5777082443237305, "learning_rate": 9.204400000000001e-06, "loss": 0.0615, "step": 3979 }, { "epoch": 0.2121874500186597, "grad_norm": 4.901087760925293, "learning_rate": 9.2042e-06, "loss": 0.0672, "step": 3980 }, { "epoch": 0.21224076344831264, "grad_norm": 5.8008646965026855, "learning_rate": 9.204e-06, "loss": 0.0514, "step": 3981 }, { "epoch": 0.21229407687796556, "grad_norm": 3.5560178756713867, "learning_rate": 9.2038e-06, "loss": 0.0448, "step": 3982 }, { "epoch": 0.21234739030761848, "grad_norm": 3.5411500930786133, "learning_rate": 9.203600000000001e-06, "loss": 0.0427, "step": 3983 }, { "epoch": 0.2124007037372714, "grad_norm": 3.3499865531921387, "learning_rate": 9.2034e-06, "loss": 0.0384, "step": 3984 }, { "epoch": 0.21245401716692436, "grad_norm": 4.268802642822266, "learning_rate": 9.2032e-06, "loss": 0.0691, "step": 3985 }, { "epoch": 0.21250733059657728, "grad_norm": 5.248834133148193, "learning_rate": 9.203000000000002e-06, "loss": 0.0363, "step": 3986 }, { "epoch": 0.2125606440262302, "grad_norm": 3.431241273880005, "learning_rate": 9.202800000000001e-06, "loss": 0.0433, "step": 3987 }, { "epoch": 0.21261395745588313, "grad_norm": 5.548766136169434, "learning_rate": 9.2026e-06, "loss": 0.0841, "step": 3988 }, { "epoch": 0.21266727088553605, "grad_norm": 3.1832692623138428, "learning_rate": 9.2024e-06, "loss": 0.0288, "step": 3989 }, { "epoch": 0.212720584315189, "grad_norm": 3.4831886291503906, "learning_rate": 9.202200000000001e-06, "loss": 0.0653, "step": 3990 }, { "epoch": 0.21277389774484193, "grad_norm": 5.131575107574463, "learning_rate": 9.202000000000001e-06, "loss": 0.0686, "step": 3991 }, { "epoch": 0.21282721117449485, "grad_norm": 3.761888265609741, "learning_rate": 9.2018e-06, "loss": 0.0418, "step": 3992 }, { "epoch": 0.21288052460414778, "grad_norm": 4.372527599334717, "learning_rate": 9.2016e-06, "loss": 0.0688, "step": 3993 }, { "epoch": 0.21293383803380073, "grad_norm": 4.222292423248291, "learning_rate": 9.2014e-06, "loss": 0.0551, "step": 3994 }, { "epoch": 0.21298715146345365, "grad_norm": 3.355456829071045, "learning_rate": 9.201200000000001e-06, "loss": 0.0582, "step": 3995 }, { "epoch": 0.21304046489310657, "grad_norm": 5.37815523147583, "learning_rate": 9.201e-06, "loss": 0.0844, "step": 3996 }, { "epoch": 0.2130937783227595, "grad_norm": 4.182397365570068, "learning_rate": 9.2008e-06, "loss": 0.0706, "step": 3997 }, { "epoch": 0.21314709175241242, "grad_norm": 4.047147750854492, "learning_rate": 9.200600000000001e-06, "loss": 0.0425, "step": 3998 }, { "epoch": 0.21320040518206537, "grad_norm": 4.278081893920898, "learning_rate": 9.200400000000001e-06, "loss": 0.0488, "step": 3999 }, { "epoch": 0.2132537186117183, "grad_norm": 6.775119304656982, "learning_rate": 9.2002e-06, "loss": 0.0922, "step": 4000 }, { "epoch": 0.21330703204137122, "grad_norm": 4.130957126617432, "learning_rate": 9.200000000000002e-06, "loss": 0.0847, "step": 4001 }, { "epoch": 0.21336034547102414, "grad_norm": 4.500328540802002, "learning_rate": 9.199800000000001e-06, "loss": 0.0554, "step": 4002 }, { "epoch": 0.21341365890067707, "grad_norm": 3.108257293701172, "learning_rate": 9.1996e-06, "loss": 0.032, "step": 4003 }, { "epoch": 0.21346697233033002, "grad_norm": 5.051377296447754, "learning_rate": 9.1994e-06, "loss": 0.076, "step": 4004 }, { "epoch": 0.21352028575998294, "grad_norm": 3.2949626445770264, "learning_rate": 9.1992e-06, "loss": 0.0446, "step": 4005 }, { "epoch": 0.21357359918963587, "grad_norm": 7.192322254180908, "learning_rate": 9.199000000000001e-06, "loss": 0.0389, "step": 4006 }, { "epoch": 0.2136269126192888, "grad_norm": 5.931868553161621, "learning_rate": 9.1988e-06, "loss": 0.0471, "step": 4007 }, { "epoch": 0.21368022604894174, "grad_norm": 5.202200412750244, "learning_rate": 9.1986e-06, "loss": 0.0469, "step": 4008 }, { "epoch": 0.21373353947859466, "grad_norm": 5.597779273986816, "learning_rate": 9.1984e-06, "loss": 0.087, "step": 4009 }, { "epoch": 0.2137868529082476, "grad_norm": 4.103443145751953, "learning_rate": 9.198200000000001e-06, "loss": 0.0444, "step": 4010 }, { "epoch": 0.2138401663379005, "grad_norm": 4.967184543609619, "learning_rate": 9.198e-06, "loss": 0.0753, "step": 4011 }, { "epoch": 0.21389347976755343, "grad_norm": 6.665450096130371, "learning_rate": 9.197800000000002e-06, "loss": 0.0557, "step": 4012 }, { "epoch": 0.2139467931972064, "grad_norm": 3.4157328605651855, "learning_rate": 9.197600000000002e-06, "loss": 0.0373, "step": 4013 }, { "epoch": 0.2140001066268593, "grad_norm": 5.648249626159668, "learning_rate": 9.197400000000001e-06, "loss": 0.0639, "step": 4014 }, { "epoch": 0.21405342005651223, "grad_norm": 3.4827880859375, "learning_rate": 9.1972e-06, "loss": 0.0472, "step": 4015 }, { "epoch": 0.21410673348616516, "grad_norm": 5.6238274574279785, "learning_rate": 9.197e-06, "loss": 0.0661, "step": 4016 }, { "epoch": 0.21416004691581808, "grad_norm": 4.447406768798828, "learning_rate": 9.196800000000001e-06, "loss": 0.0662, "step": 4017 }, { "epoch": 0.21421336034547103, "grad_norm": 5.326404094696045, "learning_rate": 9.196600000000001e-06, "loss": 0.071, "step": 4018 }, { "epoch": 0.21426667377512396, "grad_norm": 6.440304756164551, "learning_rate": 9.1964e-06, "loss": 0.0855, "step": 4019 }, { "epoch": 0.21431998720477688, "grad_norm": 4.216289043426514, "learning_rate": 9.1962e-06, "loss": 0.0505, "step": 4020 }, { "epoch": 0.2143733006344298, "grad_norm": 4.309494972229004, "learning_rate": 9.196e-06, "loss": 0.037, "step": 4021 }, { "epoch": 0.21442661406408275, "grad_norm": 4.311898231506348, "learning_rate": 9.195800000000001e-06, "loss": 0.0503, "step": 4022 }, { "epoch": 0.21447992749373568, "grad_norm": 3.0356619358062744, "learning_rate": 9.1956e-06, "loss": 0.0365, "step": 4023 }, { "epoch": 0.2145332409233886, "grad_norm": 3.52860426902771, "learning_rate": 9.195400000000002e-06, "loss": 0.0263, "step": 4024 }, { "epoch": 0.21458655435304153, "grad_norm": 5.145632743835449, "learning_rate": 9.195200000000001e-06, "loss": 0.0856, "step": 4025 }, { "epoch": 0.21463986778269445, "grad_norm": 4.8159871101379395, "learning_rate": 9.195000000000001e-06, "loss": 0.0605, "step": 4026 }, { "epoch": 0.2146931812123474, "grad_norm": 4.0841569900512695, "learning_rate": 9.1948e-06, "loss": 0.0591, "step": 4027 }, { "epoch": 0.21474649464200032, "grad_norm": 2.666654109954834, "learning_rate": 9.194600000000002e-06, "loss": 0.0465, "step": 4028 }, { "epoch": 0.21479980807165325, "grad_norm": 3.410804271697998, "learning_rate": 9.194400000000001e-06, "loss": 0.047, "step": 4029 }, { "epoch": 0.21485312150130617, "grad_norm": 6.5789899826049805, "learning_rate": 9.1942e-06, "loss": 0.113, "step": 4030 }, { "epoch": 0.2149064349309591, "grad_norm": 5.140225410461426, "learning_rate": 9.194e-06, "loss": 0.0548, "step": 4031 }, { "epoch": 0.21495974836061205, "grad_norm": 5.383633613586426, "learning_rate": 9.1938e-06, "loss": 0.0753, "step": 4032 }, { "epoch": 0.21501306179026497, "grad_norm": 4.463693618774414, "learning_rate": 9.1936e-06, "loss": 0.0634, "step": 4033 }, { "epoch": 0.2150663752199179, "grad_norm": 3.160724639892578, "learning_rate": 9.1934e-06, "loss": 0.0374, "step": 4034 }, { "epoch": 0.21511968864957082, "grad_norm": 3.339658260345459, "learning_rate": 9.1932e-06, "loss": 0.0641, "step": 4035 }, { "epoch": 0.21517300207922377, "grad_norm": 4.055910110473633, "learning_rate": 9.193000000000002e-06, "loss": 0.0814, "step": 4036 }, { "epoch": 0.2152263155088767, "grad_norm": 4.724562644958496, "learning_rate": 9.192800000000001e-06, "loss": 0.0664, "step": 4037 }, { "epoch": 0.21527962893852962, "grad_norm": 4.325861930847168, "learning_rate": 9.1926e-06, "loss": 0.0588, "step": 4038 }, { "epoch": 0.21533294236818254, "grad_norm": 4.615197658538818, "learning_rate": 9.1924e-06, "loss": 0.0611, "step": 4039 }, { "epoch": 0.21538625579783546, "grad_norm": 4.039312362670898, "learning_rate": 9.192200000000002e-06, "loss": 0.0369, "step": 4040 }, { "epoch": 0.21543956922748841, "grad_norm": 5.634718418121338, "learning_rate": 9.192000000000001e-06, "loss": 0.0403, "step": 4041 }, { "epoch": 0.21549288265714134, "grad_norm": 8.018804550170898, "learning_rate": 9.1918e-06, "loss": 0.0347, "step": 4042 }, { "epoch": 0.21554619608679426, "grad_norm": 6.005666255950928, "learning_rate": 9.1916e-06, "loss": 0.0537, "step": 4043 }, { "epoch": 0.21559950951644719, "grad_norm": 3.954848289489746, "learning_rate": 9.1914e-06, "loss": 0.0464, "step": 4044 }, { "epoch": 0.2156528229461001, "grad_norm": 3.2547829151153564, "learning_rate": 9.191200000000001e-06, "loss": 0.0309, "step": 4045 }, { "epoch": 0.21570613637575306, "grad_norm": 4.823047637939453, "learning_rate": 9.191e-06, "loss": 0.0455, "step": 4046 }, { "epoch": 0.21575944980540598, "grad_norm": 3.9840288162231445, "learning_rate": 9.1908e-06, "loss": 0.0527, "step": 4047 }, { "epoch": 0.2158127632350589, "grad_norm": 8.014357566833496, "learning_rate": 9.190600000000001e-06, "loss": 0.0606, "step": 4048 }, { "epoch": 0.21586607666471183, "grad_norm": 4.949170112609863, "learning_rate": 9.190400000000001e-06, "loss": 0.0601, "step": 4049 }, { "epoch": 0.21591939009436478, "grad_norm": 3.49098539352417, "learning_rate": 9.1902e-06, "loss": 0.039, "step": 4050 }, { "epoch": 0.2159727035240177, "grad_norm": 5.113686561584473, "learning_rate": 9.190000000000002e-06, "loss": 0.0325, "step": 4051 }, { "epoch": 0.21602601695367063, "grad_norm": 5.546542644500732, "learning_rate": 9.189800000000001e-06, "loss": 0.1248, "step": 4052 }, { "epoch": 0.21607933038332355, "grad_norm": 4.909501075744629, "learning_rate": 9.189600000000001e-06, "loss": 0.0491, "step": 4053 }, { "epoch": 0.21613264381297648, "grad_norm": 6.96661901473999, "learning_rate": 9.1894e-06, "loss": 0.073, "step": 4054 }, { "epoch": 0.21618595724262943, "grad_norm": 5.892750263214111, "learning_rate": 9.1892e-06, "loss": 0.0517, "step": 4055 }, { "epoch": 0.21623927067228235, "grad_norm": 3.2232587337493896, "learning_rate": 9.189000000000001e-06, "loss": 0.0469, "step": 4056 }, { "epoch": 0.21629258410193528, "grad_norm": 2.3667514324188232, "learning_rate": 9.188800000000001e-06, "loss": 0.0393, "step": 4057 }, { "epoch": 0.2163458975315882, "grad_norm": 6.193928241729736, "learning_rate": 9.1886e-06, "loss": 0.0487, "step": 4058 }, { "epoch": 0.21639921096124112, "grad_norm": 9.429533004760742, "learning_rate": 9.1884e-06, "loss": 0.1063, "step": 4059 }, { "epoch": 0.21645252439089407, "grad_norm": 4.500209808349609, "learning_rate": 9.188200000000001e-06, "loss": 0.0593, "step": 4060 }, { "epoch": 0.216505837820547, "grad_norm": 3.213632106781006, "learning_rate": 9.188e-06, "loss": 0.0535, "step": 4061 }, { "epoch": 0.21655915125019992, "grad_norm": 11.246296882629395, "learning_rate": 9.1878e-06, "loss": 0.0933, "step": 4062 }, { "epoch": 0.21661246467985285, "grad_norm": 10.525626182556152, "learning_rate": 9.187600000000002e-06, "loss": 0.0636, "step": 4063 }, { "epoch": 0.2166657781095058, "grad_norm": 7.0493340492248535, "learning_rate": 9.187400000000001e-06, "loss": 0.0674, "step": 4064 }, { "epoch": 0.21671909153915872, "grad_norm": 5.075588226318359, "learning_rate": 9.1872e-06, "loss": 0.1057, "step": 4065 }, { "epoch": 0.21677240496881164, "grad_norm": 6.151618957519531, "learning_rate": 9.187e-06, "loss": 0.0916, "step": 4066 }, { "epoch": 0.21682571839846457, "grad_norm": 4.740996837615967, "learning_rate": 9.186800000000002e-06, "loss": 0.0524, "step": 4067 }, { "epoch": 0.2168790318281175, "grad_norm": 4.4820027351379395, "learning_rate": 9.186600000000001e-06, "loss": 0.083, "step": 4068 }, { "epoch": 0.21693234525777044, "grad_norm": 7.983625411987305, "learning_rate": 9.1864e-06, "loss": 0.0702, "step": 4069 }, { "epoch": 0.21698565868742337, "grad_norm": 2.9916300773620605, "learning_rate": 9.1862e-06, "loss": 0.05, "step": 4070 }, { "epoch": 0.2170389721170763, "grad_norm": 7.1074442863464355, "learning_rate": 9.186e-06, "loss": 0.0863, "step": 4071 }, { "epoch": 0.2170922855467292, "grad_norm": 2.9941582679748535, "learning_rate": 9.185800000000001e-06, "loss": 0.0276, "step": 4072 }, { "epoch": 0.21714559897638214, "grad_norm": 3.105499029159546, "learning_rate": 9.1856e-06, "loss": 0.0408, "step": 4073 }, { "epoch": 0.2171989124060351, "grad_norm": 2.0913949012756348, "learning_rate": 9.1854e-06, "loss": 0.0234, "step": 4074 }, { "epoch": 0.217252225835688, "grad_norm": 4.643030166625977, "learning_rate": 9.185200000000001e-06, "loss": 0.0874, "step": 4075 }, { "epoch": 0.21730553926534094, "grad_norm": 3.6125829219818115, "learning_rate": 9.185000000000001e-06, "loss": 0.0602, "step": 4076 }, { "epoch": 0.21735885269499386, "grad_norm": 5.513272285461426, "learning_rate": 9.1848e-06, "loss": 0.0696, "step": 4077 }, { "epoch": 0.2174121661246468, "grad_norm": 7.376022815704346, "learning_rate": 9.184600000000002e-06, "loss": 0.1004, "step": 4078 }, { "epoch": 0.21746547955429973, "grad_norm": 3.3115062713623047, "learning_rate": 9.184400000000001e-06, "loss": 0.0417, "step": 4079 }, { "epoch": 0.21751879298395266, "grad_norm": 5.347006797790527, "learning_rate": 9.184200000000001e-06, "loss": 0.0995, "step": 4080 }, { "epoch": 0.21757210641360558, "grad_norm": 3.0721347332000732, "learning_rate": 9.184e-06, "loss": 0.0606, "step": 4081 }, { "epoch": 0.2176254198432585, "grad_norm": 4.277502536773682, "learning_rate": 9.1838e-06, "loss": 0.0793, "step": 4082 }, { "epoch": 0.21767873327291146, "grad_norm": 3.415104389190674, "learning_rate": 9.1836e-06, "loss": 0.0518, "step": 4083 }, { "epoch": 0.21773204670256438, "grad_norm": 3.508646011352539, "learning_rate": 9.183400000000001e-06, "loss": 0.0495, "step": 4084 }, { "epoch": 0.2177853601322173, "grad_norm": 3.6535396575927734, "learning_rate": 9.1832e-06, "loss": 0.0416, "step": 4085 }, { "epoch": 0.21783867356187023, "grad_norm": 2.915414333343506, "learning_rate": 9.183e-06, "loss": 0.0514, "step": 4086 }, { "epoch": 0.21789198699152315, "grad_norm": 3.6825413703918457, "learning_rate": 9.182800000000001e-06, "loss": 0.0463, "step": 4087 }, { "epoch": 0.2179453004211761, "grad_norm": 4.108119964599609, "learning_rate": 9.1826e-06, "loss": 0.0505, "step": 4088 }, { "epoch": 0.21799861385082903, "grad_norm": 3.3314430713653564, "learning_rate": 9.1824e-06, "loss": 0.0295, "step": 4089 }, { "epoch": 0.21805192728048195, "grad_norm": 8.219949722290039, "learning_rate": 9.182200000000002e-06, "loss": 0.078, "step": 4090 }, { "epoch": 0.21810524071013487, "grad_norm": 8.873302459716797, "learning_rate": 9.182000000000001e-06, "loss": 0.0534, "step": 4091 }, { "epoch": 0.21815855413978782, "grad_norm": 2.7396082878112793, "learning_rate": 9.1818e-06, "loss": 0.0531, "step": 4092 }, { "epoch": 0.21821186756944075, "grad_norm": 3.6786999702453613, "learning_rate": 9.1816e-06, "loss": 0.0536, "step": 4093 }, { "epoch": 0.21826518099909367, "grad_norm": 3.070739984512329, "learning_rate": 9.1814e-06, "loss": 0.0301, "step": 4094 }, { "epoch": 0.2183184944287466, "grad_norm": 5.373664855957031, "learning_rate": 9.181200000000001e-06, "loss": 0.0369, "step": 4095 }, { "epoch": 0.21837180785839952, "grad_norm": 3.852201223373413, "learning_rate": 9.181e-06, "loss": 0.0355, "step": 4096 }, { "epoch": 0.21842512128805247, "grad_norm": 3.7031092643737793, "learning_rate": 9.1808e-06, "loss": 0.0503, "step": 4097 }, { "epoch": 0.2184784347177054, "grad_norm": 3.748011827468872, "learning_rate": 9.1806e-06, "loss": 0.0203, "step": 4098 }, { "epoch": 0.21853174814735832, "grad_norm": 4.603901386260986, "learning_rate": 9.180400000000001e-06, "loss": 0.0573, "step": 4099 }, { "epoch": 0.21858506157701124, "grad_norm": 3.301766872406006, "learning_rate": 9.1802e-06, "loss": 0.0446, "step": 4100 }, { "epoch": 0.21863837500666417, "grad_norm": 3.430511236190796, "learning_rate": 9.180000000000002e-06, "loss": 0.0328, "step": 4101 }, { "epoch": 0.21869168843631712, "grad_norm": 7.804294586181641, "learning_rate": 9.179800000000001e-06, "loss": 0.0605, "step": 4102 }, { "epoch": 0.21874500186597004, "grad_norm": 2.895531415939331, "learning_rate": 9.179600000000001e-06, "loss": 0.029, "step": 4103 }, { "epoch": 0.21879831529562296, "grad_norm": 4.260056972503662, "learning_rate": 9.1794e-06, "loss": 0.0457, "step": 4104 }, { "epoch": 0.2188516287252759, "grad_norm": 5.754096031188965, "learning_rate": 9.1792e-06, "loss": 0.0643, "step": 4105 }, { "epoch": 0.21890494215492884, "grad_norm": 3.365367889404297, "learning_rate": 9.179000000000001e-06, "loss": 0.0699, "step": 4106 }, { "epoch": 0.21895825558458176, "grad_norm": 3.217426300048828, "learning_rate": 9.178800000000001e-06, "loss": 0.0468, "step": 4107 }, { "epoch": 0.21901156901423469, "grad_norm": 2.016228437423706, "learning_rate": 9.1786e-06, "loss": 0.0128, "step": 4108 }, { "epoch": 0.2190648824438876, "grad_norm": 6.219063758850098, "learning_rate": 9.1784e-06, "loss": 0.0636, "step": 4109 }, { "epoch": 0.21911819587354053, "grad_norm": 3.756486415863037, "learning_rate": 9.1782e-06, "loss": 0.0358, "step": 4110 }, { "epoch": 0.21917150930319348, "grad_norm": 2.7913007736206055, "learning_rate": 9.178000000000001e-06, "loss": 0.0288, "step": 4111 }, { "epoch": 0.2192248227328464, "grad_norm": 2.813504695892334, "learning_rate": 9.1778e-06, "loss": 0.0231, "step": 4112 }, { "epoch": 0.21927813616249933, "grad_norm": 3.0100717544555664, "learning_rate": 9.177600000000002e-06, "loss": 0.0246, "step": 4113 }, { "epoch": 0.21933144959215226, "grad_norm": 3.3465633392333984, "learning_rate": 9.177400000000001e-06, "loss": 0.0458, "step": 4114 }, { "epoch": 0.21938476302180518, "grad_norm": 4.585386276245117, "learning_rate": 9.1772e-06, "loss": 0.0478, "step": 4115 }, { "epoch": 0.21943807645145813, "grad_norm": 6.902108192443848, "learning_rate": 9.177e-06, "loss": 0.0518, "step": 4116 }, { "epoch": 0.21949138988111105, "grad_norm": 4.175037384033203, "learning_rate": 9.176800000000002e-06, "loss": 0.0772, "step": 4117 }, { "epoch": 0.21954470331076398, "grad_norm": 4.237269878387451, "learning_rate": 9.176600000000001e-06, "loss": 0.0503, "step": 4118 }, { "epoch": 0.2195980167404169, "grad_norm": 3.891566514968872, "learning_rate": 9.1764e-06, "loss": 0.0497, "step": 4119 }, { "epoch": 0.21965133017006985, "grad_norm": 4.656401634216309, "learning_rate": 9.1762e-06, "loss": 0.0551, "step": 4120 }, { "epoch": 0.21970464359972278, "grad_norm": 3.334772825241089, "learning_rate": 9.176e-06, "loss": 0.0386, "step": 4121 }, { "epoch": 0.2197579570293757, "grad_norm": 2.7231173515319824, "learning_rate": 9.1758e-06, "loss": 0.0336, "step": 4122 }, { "epoch": 0.21981127045902862, "grad_norm": 2.694253444671631, "learning_rate": 9.1756e-06, "loss": 0.0429, "step": 4123 }, { "epoch": 0.21986458388868155, "grad_norm": 4.555934429168701, "learning_rate": 9.1754e-06, "loss": 0.0385, "step": 4124 }, { "epoch": 0.2199178973183345, "grad_norm": 4.038819789886475, "learning_rate": 9.175200000000002e-06, "loss": 0.0744, "step": 4125 }, { "epoch": 0.21997121074798742, "grad_norm": 3.9092390537261963, "learning_rate": 9.175000000000001e-06, "loss": 0.0459, "step": 4126 }, { "epoch": 0.22002452417764035, "grad_norm": 7.869515419006348, "learning_rate": 9.1748e-06, "loss": 0.0974, "step": 4127 }, { "epoch": 0.22007783760729327, "grad_norm": 6.2959394454956055, "learning_rate": 9.174600000000002e-06, "loss": 0.0368, "step": 4128 }, { "epoch": 0.2201311510369462, "grad_norm": 3.6507325172424316, "learning_rate": 9.174400000000001e-06, "loss": 0.0395, "step": 4129 }, { "epoch": 0.22018446446659914, "grad_norm": 2.4468278884887695, "learning_rate": 9.174200000000001e-06, "loss": 0.0311, "step": 4130 }, { "epoch": 0.22023777789625207, "grad_norm": 3.357179641723633, "learning_rate": 9.174e-06, "loss": 0.0519, "step": 4131 }, { "epoch": 0.220291091325905, "grad_norm": 3.8867223262786865, "learning_rate": 9.1738e-06, "loss": 0.0393, "step": 4132 }, { "epoch": 0.22034440475555792, "grad_norm": 6.0048747062683105, "learning_rate": 9.1736e-06, "loss": 0.0662, "step": 4133 }, { "epoch": 0.22039771818521087, "grad_norm": 2.0030980110168457, "learning_rate": 9.173400000000001e-06, "loss": 0.0451, "step": 4134 }, { "epoch": 0.2204510316148638, "grad_norm": 4.172176361083984, "learning_rate": 9.1732e-06, "loss": 0.0433, "step": 4135 }, { "epoch": 0.22050434504451671, "grad_norm": 4.941365718841553, "learning_rate": 9.173e-06, "loss": 0.0617, "step": 4136 }, { "epoch": 0.22055765847416964, "grad_norm": 2.722987413406372, "learning_rate": 9.172800000000001e-06, "loss": 0.0462, "step": 4137 }, { "epoch": 0.22061097190382256, "grad_norm": 4.492512226104736, "learning_rate": 9.172600000000001e-06, "loss": 0.0675, "step": 4138 }, { "epoch": 0.2206642853334755, "grad_norm": 2.5637943744659424, "learning_rate": 9.1724e-06, "loss": 0.0471, "step": 4139 }, { "epoch": 0.22071759876312844, "grad_norm": 3.877595901489258, "learning_rate": 9.172200000000002e-06, "loss": 0.0422, "step": 4140 }, { "epoch": 0.22077091219278136, "grad_norm": 2.8524858951568604, "learning_rate": 9.172000000000001e-06, "loss": 0.0348, "step": 4141 }, { "epoch": 0.22082422562243428, "grad_norm": 6.07757568359375, "learning_rate": 9.171800000000001e-06, "loss": 0.0547, "step": 4142 }, { "epoch": 0.2208775390520872, "grad_norm": 4.972906589508057, "learning_rate": 9.1716e-06, "loss": 0.0443, "step": 4143 }, { "epoch": 0.22093085248174016, "grad_norm": 4.304509162902832, "learning_rate": 9.1714e-06, "loss": 0.1019, "step": 4144 }, { "epoch": 0.22098416591139308, "grad_norm": 4.649209976196289, "learning_rate": 9.171200000000001e-06, "loss": 0.093, "step": 4145 }, { "epoch": 0.221037479341046, "grad_norm": 4.319967746734619, "learning_rate": 9.171e-06, "loss": 0.0388, "step": 4146 }, { "epoch": 0.22109079277069893, "grad_norm": 4.930304527282715, "learning_rate": 9.1708e-06, "loss": 0.06, "step": 4147 }, { "epoch": 0.22114410620035188, "grad_norm": 2.71954607963562, "learning_rate": 9.1706e-06, "loss": 0.0336, "step": 4148 }, { "epoch": 0.2211974196300048, "grad_norm": 3.8516340255737305, "learning_rate": 9.170400000000001e-06, "loss": 0.0467, "step": 4149 }, { "epoch": 0.22125073305965773, "grad_norm": 4.234728813171387, "learning_rate": 9.1702e-06, "loss": 0.0922, "step": 4150 }, { "epoch": 0.22130404648931065, "grad_norm": 3.0923147201538086, "learning_rate": 9.17e-06, "loss": 0.0472, "step": 4151 }, { "epoch": 0.22135735991896358, "grad_norm": 3.4253146648406982, "learning_rate": 9.169800000000002e-06, "loss": 0.0545, "step": 4152 }, { "epoch": 0.22141067334861653, "grad_norm": 4.799880027770996, "learning_rate": 9.169600000000001e-06, "loss": 0.0433, "step": 4153 }, { "epoch": 0.22146398677826945, "grad_norm": 4.011843681335449, "learning_rate": 9.1694e-06, "loss": 0.0499, "step": 4154 }, { "epoch": 0.22151730020792237, "grad_norm": 3.394418239593506, "learning_rate": 9.1692e-06, "loss": 0.0457, "step": 4155 }, { "epoch": 0.2215706136375753, "grad_norm": 6.064443588256836, "learning_rate": 9.169000000000001e-06, "loss": 0.0469, "step": 4156 }, { "epoch": 0.22162392706722822, "grad_norm": 4.7801833152771, "learning_rate": 9.168800000000001e-06, "loss": 0.069, "step": 4157 }, { "epoch": 0.22167724049688117, "grad_norm": 3.5717012882232666, "learning_rate": 9.1686e-06, "loss": 0.0577, "step": 4158 }, { "epoch": 0.2217305539265341, "grad_norm": 3.8593504428863525, "learning_rate": 9.1684e-06, "loss": 0.0977, "step": 4159 }, { "epoch": 0.22178386735618702, "grad_norm": 4.721794128417969, "learning_rate": 9.1682e-06, "loss": 0.0727, "step": 4160 }, { "epoch": 0.22183718078583994, "grad_norm": 5.381872177124023, "learning_rate": 9.168000000000001e-06, "loss": 0.0492, "step": 4161 }, { "epoch": 0.2218904942154929, "grad_norm": 4.350890636444092, "learning_rate": 9.1678e-06, "loss": 0.0452, "step": 4162 }, { "epoch": 0.22194380764514582, "grad_norm": 3.3864307403564453, "learning_rate": 9.1676e-06, "loss": 0.0419, "step": 4163 }, { "epoch": 0.22199712107479874, "grad_norm": 2.529707670211792, "learning_rate": 9.167400000000001e-06, "loss": 0.0205, "step": 4164 }, { "epoch": 0.22205043450445167, "grad_norm": 4.0292816162109375, "learning_rate": 9.167200000000001e-06, "loss": 0.0235, "step": 4165 }, { "epoch": 0.2221037479341046, "grad_norm": 6.830253601074219, "learning_rate": 9.167e-06, "loss": 0.0793, "step": 4166 }, { "epoch": 0.22215706136375754, "grad_norm": 4.655019760131836, "learning_rate": 9.166800000000002e-06, "loss": 0.0422, "step": 4167 }, { "epoch": 0.22221037479341046, "grad_norm": 2.9639949798583984, "learning_rate": 9.166600000000001e-06, "loss": 0.0419, "step": 4168 }, { "epoch": 0.2222636882230634, "grad_norm": 4.740772724151611, "learning_rate": 9.166400000000001e-06, "loss": 0.0569, "step": 4169 }, { "epoch": 0.2223170016527163, "grad_norm": 3.4004063606262207, "learning_rate": 9.1662e-06, "loss": 0.0383, "step": 4170 }, { "epoch": 0.22237031508236924, "grad_norm": 2.396312952041626, "learning_rate": 9.166e-06, "loss": 0.031, "step": 4171 }, { "epoch": 0.2224236285120222, "grad_norm": 4.439006805419922, "learning_rate": 9.1658e-06, "loss": 0.0848, "step": 4172 }, { "epoch": 0.2224769419416751, "grad_norm": 5.862430572509766, "learning_rate": 9.1656e-06, "loss": 0.0682, "step": 4173 }, { "epoch": 0.22253025537132803, "grad_norm": 2.5383429527282715, "learning_rate": 9.1654e-06, "loss": 0.0273, "step": 4174 }, { "epoch": 0.22258356880098096, "grad_norm": 3.6789772510528564, "learning_rate": 9.1652e-06, "loss": 0.0367, "step": 4175 }, { "epoch": 0.2226368822306339, "grad_norm": 4.039163589477539, "learning_rate": 9.165000000000001e-06, "loss": 0.0526, "step": 4176 }, { "epoch": 0.22269019566028683, "grad_norm": 2.911167621612549, "learning_rate": 9.1648e-06, "loss": 0.0505, "step": 4177 }, { "epoch": 0.22274350908993976, "grad_norm": 3.1471993923187256, "learning_rate": 9.164600000000002e-06, "loss": 0.024, "step": 4178 }, { "epoch": 0.22279682251959268, "grad_norm": 3.227928400039673, "learning_rate": 9.164400000000002e-06, "loss": 0.0446, "step": 4179 }, { "epoch": 0.2228501359492456, "grad_norm": 7.4627604484558105, "learning_rate": 9.164200000000001e-06, "loss": 0.0787, "step": 4180 }, { "epoch": 0.22290344937889855, "grad_norm": 2.2141003608703613, "learning_rate": 9.164e-06, "loss": 0.0215, "step": 4181 }, { "epoch": 0.22295676280855148, "grad_norm": 3.9989817142486572, "learning_rate": 9.1638e-06, "loss": 0.0463, "step": 4182 }, { "epoch": 0.2230100762382044, "grad_norm": 2.896888017654419, "learning_rate": 9.1636e-06, "loss": 0.037, "step": 4183 }, { "epoch": 0.22306338966785733, "grad_norm": 2.2513010501861572, "learning_rate": 9.163400000000001e-06, "loss": 0.0563, "step": 4184 }, { "epoch": 0.22311670309751025, "grad_norm": 4.4970703125, "learning_rate": 9.1632e-06, "loss": 0.0322, "step": 4185 }, { "epoch": 0.2231700165271632, "grad_norm": 6.214661121368408, "learning_rate": 9.163e-06, "loss": 0.0922, "step": 4186 }, { "epoch": 0.22322332995681612, "grad_norm": 3.930891275405884, "learning_rate": 9.1628e-06, "loss": 0.0407, "step": 4187 }, { "epoch": 0.22327664338646905, "grad_norm": 4.782918453216553, "learning_rate": 9.162600000000001e-06, "loss": 0.0609, "step": 4188 }, { "epoch": 0.22332995681612197, "grad_norm": 3.2756142616271973, "learning_rate": 9.1624e-06, "loss": 0.0276, "step": 4189 }, { "epoch": 0.22338327024577492, "grad_norm": 4.044316291809082, "learning_rate": 9.162200000000002e-06, "loss": 0.0642, "step": 4190 }, { "epoch": 0.22343658367542785, "grad_norm": 4.363856792449951, "learning_rate": 9.162000000000001e-06, "loss": 0.0592, "step": 4191 }, { "epoch": 0.22348989710508077, "grad_norm": 4.142923355102539, "learning_rate": 9.161800000000001e-06, "loss": 0.0468, "step": 4192 }, { "epoch": 0.2235432105347337, "grad_norm": 4.036378860473633, "learning_rate": 9.1616e-06, "loss": 0.0458, "step": 4193 }, { "epoch": 0.22359652396438662, "grad_norm": 3.8321263790130615, "learning_rate": 9.1614e-06, "loss": 0.0409, "step": 4194 }, { "epoch": 0.22364983739403957, "grad_norm": 3.9333250522613525, "learning_rate": 9.161200000000001e-06, "loss": 0.061, "step": 4195 }, { "epoch": 0.2237031508236925, "grad_norm": 3.3161532878875732, "learning_rate": 9.161000000000001e-06, "loss": 0.0318, "step": 4196 }, { "epoch": 0.22375646425334542, "grad_norm": 4.583881378173828, "learning_rate": 9.1608e-06, "loss": 0.0481, "step": 4197 }, { "epoch": 0.22380977768299834, "grad_norm": 5.103100776672363, "learning_rate": 9.1606e-06, "loss": 0.0463, "step": 4198 }, { "epoch": 0.22386309111265126, "grad_norm": 4.77508544921875, "learning_rate": 9.1604e-06, "loss": 0.0564, "step": 4199 }, { "epoch": 0.22391640454230421, "grad_norm": 4.38510274887085, "learning_rate": 9.1602e-06, "loss": 0.0443, "step": 4200 }, { "epoch": 0.22396971797195714, "grad_norm": 5.304172039031982, "learning_rate": 9.16e-06, "loss": 0.0481, "step": 4201 }, { "epoch": 0.22402303140161006, "grad_norm": 3.8221733570098877, "learning_rate": 9.159800000000002e-06, "loss": 0.0432, "step": 4202 }, { "epoch": 0.22407634483126299, "grad_norm": 5.09731912612915, "learning_rate": 9.159600000000001e-06, "loss": 0.0466, "step": 4203 }, { "epoch": 0.22412965826091594, "grad_norm": 4.27243709564209, "learning_rate": 9.1594e-06, "loss": 0.046, "step": 4204 }, { "epoch": 0.22418297169056886, "grad_norm": 4.078075408935547, "learning_rate": 9.1592e-06, "loss": 0.0324, "step": 4205 }, { "epoch": 0.22423628512022178, "grad_norm": 6.387048244476318, "learning_rate": 9.159000000000002e-06, "loss": 0.0596, "step": 4206 }, { "epoch": 0.2242895985498747, "grad_norm": 5.579100131988525, "learning_rate": 9.158800000000001e-06, "loss": 0.0749, "step": 4207 }, { "epoch": 0.22434291197952763, "grad_norm": 6.324828624725342, "learning_rate": 9.1586e-06, "loss": 0.0635, "step": 4208 }, { "epoch": 0.22439622540918058, "grad_norm": 11.695353507995605, "learning_rate": 9.1584e-06, "loss": 0.0884, "step": 4209 }, { "epoch": 0.2244495388388335, "grad_norm": 3.617420196533203, "learning_rate": 9.1582e-06, "loss": 0.0606, "step": 4210 }, { "epoch": 0.22450285226848643, "grad_norm": 7.917542457580566, "learning_rate": 9.158e-06, "loss": 0.041, "step": 4211 }, { "epoch": 0.22455616569813935, "grad_norm": 2.8964977264404297, "learning_rate": 9.1578e-06, "loss": 0.0577, "step": 4212 }, { "epoch": 0.22460947912779228, "grad_norm": 4.085171699523926, "learning_rate": 9.1576e-06, "loss": 0.0603, "step": 4213 }, { "epoch": 0.22466279255744523, "grad_norm": 6.64133358001709, "learning_rate": 9.157400000000001e-06, "loss": 0.0882, "step": 4214 }, { "epoch": 0.22471610598709815, "grad_norm": 6.442428112030029, "learning_rate": 9.157200000000001e-06, "loss": 0.0727, "step": 4215 }, { "epoch": 0.22476941941675108, "grad_norm": 3.266510486602783, "learning_rate": 9.157e-06, "loss": 0.0522, "step": 4216 }, { "epoch": 0.224822732846404, "grad_norm": 4.853969573974609, "learning_rate": 9.156800000000002e-06, "loss": 0.0762, "step": 4217 }, { "epoch": 0.22487604627605695, "grad_norm": 4.858145713806152, "learning_rate": 9.156600000000001e-06, "loss": 0.0432, "step": 4218 }, { "epoch": 0.22492935970570987, "grad_norm": 4.977133750915527, "learning_rate": 9.156400000000001e-06, "loss": 0.0592, "step": 4219 }, { "epoch": 0.2249826731353628, "grad_norm": 7.234360694885254, "learning_rate": 9.1562e-06, "loss": 0.0672, "step": 4220 }, { "epoch": 0.22503598656501572, "grad_norm": 6.1078290939331055, "learning_rate": 9.156e-06, "loss": 0.0759, "step": 4221 }, { "epoch": 0.22508929999466865, "grad_norm": 6.996977806091309, "learning_rate": 9.1558e-06, "loss": 0.1012, "step": 4222 }, { "epoch": 0.2251426134243216, "grad_norm": 5.082017421722412, "learning_rate": 9.155600000000001e-06, "loss": 0.0624, "step": 4223 }, { "epoch": 0.22519592685397452, "grad_norm": 3.2146668434143066, "learning_rate": 9.1554e-06, "loss": 0.0474, "step": 4224 }, { "epoch": 0.22524924028362744, "grad_norm": 4.131708145141602, "learning_rate": 9.1552e-06, "loss": 0.0643, "step": 4225 }, { "epoch": 0.22530255371328037, "grad_norm": 3.2959604263305664, "learning_rate": 9.155000000000001e-06, "loss": 0.0514, "step": 4226 }, { "epoch": 0.2253558671429333, "grad_norm": 3.139209747314453, "learning_rate": 9.1548e-06, "loss": 0.039, "step": 4227 }, { "epoch": 0.22540918057258624, "grad_norm": 6.969472885131836, "learning_rate": 9.154600000000002e-06, "loss": 0.0668, "step": 4228 }, { "epoch": 0.22546249400223917, "grad_norm": 4.897603511810303, "learning_rate": 9.154400000000002e-06, "loss": 0.0694, "step": 4229 }, { "epoch": 0.2255158074318921, "grad_norm": 4.7323126792907715, "learning_rate": 9.154200000000001e-06, "loss": 0.0573, "step": 4230 }, { "epoch": 0.225569120861545, "grad_norm": 5.177385330200195, "learning_rate": 9.154e-06, "loss": 0.0609, "step": 4231 }, { "epoch": 0.22562243429119797, "grad_norm": 7.2464752197265625, "learning_rate": 9.1538e-06, "loss": 0.0945, "step": 4232 }, { "epoch": 0.2256757477208509, "grad_norm": 2.1538338661193848, "learning_rate": 9.1536e-06, "loss": 0.0228, "step": 4233 }, { "epoch": 0.2257290611505038, "grad_norm": 4.477114677429199, "learning_rate": 9.153400000000001e-06, "loss": 0.0379, "step": 4234 }, { "epoch": 0.22578237458015674, "grad_norm": 5.130542755126953, "learning_rate": 9.1532e-06, "loss": 0.0494, "step": 4235 }, { "epoch": 0.22583568800980966, "grad_norm": 2.9841201305389404, "learning_rate": 9.153e-06, "loss": 0.053, "step": 4236 }, { "epoch": 0.2258890014394626, "grad_norm": 4.652614116668701, "learning_rate": 9.1528e-06, "loss": 0.0527, "step": 4237 }, { "epoch": 0.22594231486911553, "grad_norm": 7.094602584838867, "learning_rate": 9.152600000000001e-06, "loss": 0.0888, "step": 4238 }, { "epoch": 0.22599562829876846, "grad_norm": 3.1914286613464355, "learning_rate": 9.1524e-06, "loss": 0.0484, "step": 4239 }, { "epoch": 0.22604894172842138, "grad_norm": 10.12666130065918, "learning_rate": 9.152200000000002e-06, "loss": 0.0608, "step": 4240 }, { "epoch": 0.2261022551580743, "grad_norm": 3.886434316635132, "learning_rate": 9.152000000000001e-06, "loss": 0.0438, "step": 4241 }, { "epoch": 0.22615556858772726, "grad_norm": 5.026529788970947, "learning_rate": 9.151800000000001e-06, "loss": 0.0633, "step": 4242 }, { "epoch": 0.22620888201738018, "grad_norm": 5.13212251663208, "learning_rate": 9.1516e-06, "loss": 0.054, "step": 4243 }, { "epoch": 0.2262621954470331, "grad_norm": 6.841622829437256, "learning_rate": 9.1514e-06, "loss": 0.0937, "step": 4244 }, { "epoch": 0.22631550887668603, "grad_norm": 4.900943756103516, "learning_rate": 9.151200000000001e-06, "loss": 0.0796, "step": 4245 }, { "epoch": 0.22636882230633898, "grad_norm": 4.9131646156311035, "learning_rate": 9.151000000000001e-06, "loss": 0.0748, "step": 4246 }, { "epoch": 0.2264221357359919, "grad_norm": 2.6738674640655518, "learning_rate": 9.1508e-06, "loss": 0.0217, "step": 4247 }, { "epoch": 0.22647544916564483, "grad_norm": 2.780460834503174, "learning_rate": 9.1506e-06, "loss": 0.0355, "step": 4248 }, { "epoch": 0.22652876259529775, "grad_norm": 6.321931838989258, "learning_rate": 9.1504e-06, "loss": 0.1141, "step": 4249 }, { "epoch": 0.22658207602495067, "grad_norm": 4.264806270599365, "learning_rate": 9.150200000000001e-06, "loss": 0.0453, "step": 4250 }, { "epoch": 0.22663538945460363, "grad_norm": 6.245548248291016, "learning_rate": 9.15e-06, "loss": 0.0867, "step": 4251 }, { "epoch": 0.22668870288425655, "grad_norm": 6.008468151092529, "learning_rate": 9.149800000000002e-06, "loss": 0.0583, "step": 4252 }, { "epoch": 0.22674201631390947, "grad_norm": 6.63062858581543, "learning_rate": 9.149600000000001e-06, "loss": 0.0637, "step": 4253 }, { "epoch": 0.2267953297435624, "grad_norm": 2.753161668777466, "learning_rate": 9.149400000000001e-06, "loss": 0.0412, "step": 4254 }, { "epoch": 0.22684864317321532, "grad_norm": 5.004485607147217, "learning_rate": 9.1492e-06, "loss": 0.0671, "step": 4255 }, { "epoch": 0.22690195660286827, "grad_norm": 3.7938218116760254, "learning_rate": 9.149000000000002e-06, "loss": 0.0657, "step": 4256 }, { "epoch": 0.2269552700325212, "grad_norm": 4.406217575073242, "learning_rate": 9.148800000000001e-06, "loss": 0.0736, "step": 4257 }, { "epoch": 0.22700858346217412, "grad_norm": 5.510732650756836, "learning_rate": 9.1486e-06, "loss": 0.0517, "step": 4258 }, { "epoch": 0.22706189689182704, "grad_norm": 4.079010963439941, "learning_rate": 9.1484e-06, "loss": 0.0434, "step": 4259 }, { "epoch": 0.22711521032148, "grad_norm": 4.966255187988281, "learning_rate": 9.1482e-06, "loss": 0.0934, "step": 4260 }, { "epoch": 0.22716852375113292, "grad_norm": 8.41355037689209, "learning_rate": 9.148e-06, "loss": 0.0809, "step": 4261 }, { "epoch": 0.22722183718078584, "grad_norm": 6.960259437561035, "learning_rate": 9.1478e-06, "loss": 0.0813, "step": 4262 }, { "epoch": 0.22727515061043876, "grad_norm": 4.3494133949279785, "learning_rate": 9.1476e-06, "loss": 0.0697, "step": 4263 }, { "epoch": 0.2273284640400917, "grad_norm": 6.161351203918457, "learning_rate": 9.147400000000002e-06, "loss": 0.0657, "step": 4264 }, { "epoch": 0.22738177746974464, "grad_norm": 4.522912979125977, "learning_rate": 9.147200000000001e-06, "loss": 0.0683, "step": 4265 }, { "epoch": 0.22743509089939756, "grad_norm": 5.089977264404297, "learning_rate": 9.147e-06, "loss": 0.0426, "step": 4266 }, { "epoch": 0.2274884043290505, "grad_norm": 8.3170804977417, "learning_rate": 9.146800000000002e-06, "loss": 0.0829, "step": 4267 }, { "epoch": 0.2275417177587034, "grad_norm": 5.464208126068115, "learning_rate": 9.146600000000002e-06, "loss": 0.0716, "step": 4268 }, { "epoch": 0.22759503118835633, "grad_norm": 2.9292008876800537, "learning_rate": 9.146400000000001e-06, "loss": 0.0526, "step": 4269 }, { "epoch": 0.22764834461800929, "grad_norm": 5.073414325714111, "learning_rate": 9.1462e-06, "loss": 0.0827, "step": 4270 }, { "epoch": 0.2277016580476622, "grad_norm": 4.862511157989502, "learning_rate": 9.146e-06, "loss": 0.0676, "step": 4271 }, { "epoch": 0.22775497147731513, "grad_norm": 2.7789015769958496, "learning_rate": 9.1458e-06, "loss": 0.0365, "step": 4272 }, { "epoch": 0.22780828490696806, "grad_norm": 3.2090249061584473, "learning_rate": 9.145600000000001e-06, "loss": 0.0498, "step": 4273 }, { "epoch": 0.227861598336621, "grad_norm": 3.59535551071167, "learning_rate": 9.1454e-06, "loss": 0.0504, "step": 4274 }, { "epoch": 0.22791491176627393, "grad_norm": 3.268352508544922, "learning_rate": 9.1452e-06, "loss": 0.0691, "step": 4275 }, { "epoch": 0.22796822519592685, "grad_norm": 3.5793516635894775, "learning_rate": 9.145000000000001e-06, "loss": 0.0709, "step": 4276 }, { "epoch": 0.22802153862557978, "grad_norm": 3.9705822467803955, "learning_rate": 9.144800000000001e-06, "loss": 0.038, "step": 4277 }, { "epoch": 0.2280748520552327, "grad_norm": 5.688128471374512, "learning_rate": 9.1446e-06, "loss": 0.0462, "step": 4278 }, { "epoch": 0.22812816548488565, "grad_norm": 5.989963531494141, "learning_rate": 9.144400000000002e-06, "loss": 0.0831, "step": 4279 }, { "epoch": 0.22818147891453858, "grad_norm": 3.597306489944458, "learning_rate": 9.144200000000001e-06, "loss": 0.0668, "step": 4280 }, { "epoch": 0.2282347923441915, "grad_norm": 2.984323501586914, "learning_rate": 9.144000000000001e-06, "loss": 0.0345, "step": 4281 }, { "epoch": 0.22828810577384442, "grad_norm": 4.293782711029053, "learning_rate": 9.1438e-06, "loss": 0.0431, "step": 4282 }, { "epoch": 0.22834141920349735, "grad_norm": 6.6184773445129395, "learning_rate": 9.1436e-06, "loss": 0.094, "step": 4283 }, { "epoch": 0.2283947326331503, "grad_norm": 3.273362874984741, "learning_rate": 9.143400000000001e-06, "loss": 0.0648, "step": 4284 }, { "epoch": 0.22844804606280322, "grad_norm": 4.124316692352295, "learning_rate": 9.1432e-06, "loss": 0.0436, "step": 4285 }, { "epoch": 0.22850135949245615, "grad_norm": 5.019550800323486, "learning_rate": 9.143e-06, "loss": 0.0518, "step": 4286 }, { "epoch": 0.22855467292210907, "grad_norm": 8.893257141113281, "learning_rate": 9.1428e-06, "loss": 0.0795, "step": 4287 }, { "epoch": 0.22860798635176202, "grad_norm": 3.0446884632110596, "learning_rate": 9.142600000000001e-06, "loss": 0.0436, "step": 4288 }, { "epoch": 0.22866129978141495, "grad_norm": 4.259616374969482, "learning_rate": 9.1424e-06, "loss": 0.043, "step": 4289 }, { "epoch": 0.22871461321106787, "grad_norm": 3.2838494777679443, "learning_rate": 9.1422e-06, "loss": 0.075, "step": 4290 }, { "epoch": 0.2287679266407208, "grad_norm": 4.009182453155518, "learning_rate": 9.142000000000002e-06, "loss": 0.0837, "step": 4291 }, { "epoch": 0.22882124007037372, "grad_norm": 3.409883737564087, "learning_rate": 9.141800000000001e-06, "loss": 0.0497, "step": 4292 }, { "epoch": 0.22887455350002667, "grad_norm": 3.395571231842041, "learning_rate": 9.1416e-06, "loss": 0.0415, "step": 4293 }, { "epoch": 0.2289278669296796, "grad_norm": 3.876152276992798, "learning_rate": 9.1414e-06, "loss": 0.0382, "step": 4294 }, { "epoch": 0.22898118035933251, "grad_norm": 5.518331527709961, "learning_rate": 9.141200000000002e-06, "loss": 0.0811, "step": 4295 }, { "epoch": 0.22903449378898544, "grad_norm": 5.352883338928223, "learning_rate": 9.141000000000001e-06, "loss": 0.0624, "step": 4296 }, { "epoch": 0.22908780721863836, "grad_norm": 3.649716377258301, "learning_rate": 9.1408e-06, "loss": 0.0355, "step": 4297 }, { "epoch": 0.2291411206482913, "grad_norm": 2.613292694091797, "learning_rate": 9.1406e-06, "loss": 0.0375, "step": 4298 }, { "epoch": 0.22919443407794424, "grad_norm": 2.346785306930542, "learning_rate": 9.1404e-06, "loss": 0.0224, "step": 4299 }, { "epoch": 0.22924774750759716, "grad_norm": 5.175352096557617, "learning_rate": 9.140200000000001e-06, "loss": 0.1107, "step": 4300 }, { "epoch": 0.22930106093725008, "grad_norm": 1.880063772201538, "learning_rate": 9.14e-06, "loss": 0.0389, "step": 4301 }, { "epoch": 0.22935437436690304, "grad_norm": 3.5489420890808105, "learning_rate": 9.1398e-06, "loss": 0.0348, "step": 4302 }, { "epoch": 0.22940768779655596, "grad_norm": 5.495047092437744, "learning_rate": 9.139600000000001e-06, "loss": 0.0824, "step": 4303 }, { "epoch": 0.22946100122620888, "grad_norm": 5.231942176818848, "learning_rate": 9.139400000000001e-06, "loss": 0.0694, "step": 4304 }, { "epoch": 0.2295143146558618, "grad_norm": 2.7161521911621094, "learning_rate": 9.1392e-06, "loss": 0.0299, "step": 4305 }, { "epoch": 0.22956762808551473, "grad_norm": 2.625386953353882, "learning_rate": 9.139000000000002e-06, "loss": 0.034, "step": 4306 }, { "epoch": 0.22962094151516768, "grad_norm": 5.234643936157227, "learning_rate": 9.138800000000001e-06, "loss": 0.0746, "step": 4307 }, { "epoch": 0.2296742549448206, "grad_norm": 5.004977226257324, "learning_rate": 9.138600000000001e-06, "loss": 0.0457, "step": 4308 }, { "epoch": 0.22972756837447353, "grad_norm": 2.752946615219116, "learning_rate": 9.1384e-06, "loss": 0.0277, "step": 4309 }, { "epoch": 0.22978088180412645, "grad_norm": 11.083215713500977, "learning_rate": 9.1382e-06, "loss": 0.1472, "step": 4310 }, { "epoch": 0.22983419523377938, "grad_norm": 3.4866108894348145, "learning_rate": 9.138e-06, "loss": 0.068, "step": 4311 }, { "epoch": 0.22988750866343233, "grad_norm": 3.933105945587158, "learning_rate": 9.1378e-06, "loss": 0.0795, "step": 4312 }, { "epoch": 0.22994082209308525, "grad_norm": 4.879090785980225, "learning_rate": 9.1376e-06, "loss": 0.0596, "step": 4313 }, { "epoch": 0.22999413552273817, "grad_norm": 5.610250473022461, "learning_rate": 9.1374e-06, "loss": 0.0718, "step": 4314 }, { "epoch": 0.2300474489523911, "grad_norm": 2.777010679244995, "learning_rate": 9.137200000000001e-06, "loss": 0.0509, "step": 4315 }, { "epoch": 0.23010076238204405, "grad_norm": 6.449737548828125, "learning_rate": 9.137e-06, "loss": 0.067, "step": 4316 }, { "epoch": 0.23015407581169697, "grad_norm": 3.489668130874634, "learning_rate": 9.136800000000002e-06, "loss": 0.0597, "step": 4317 }, { "epoch": 0.2302073892413499, "grad_norm": 3.3130016326904297, "learning_rate": 9.136600000000002e-06, "loss": 0.0405, "step": 4318 }, { "epoch": 0.23026070267100282, "grad_norm": 3.2620623111724854, "learning_rate": 9.136400000000001e-06, "loss": 0.0401, "step": 4319 }, { "epoch": 0.23031401610065574, "grad_norm": 6.954593658447266, "learning_rate": 9.1362e-06, "loss": 0.0776, "step": 4320 }, { "epoch": 0.2303673295303087, "grad_norm": 6.476220607757568, "learning_rate": 9.136e-06, "loss": 0.0637, "step": 4321 }, { "epoch": 0.23042064295996162, "grad_norm": 2.9703757762908936, "learning_rate": 9.1358e-06, "loss": 0.0356, "step": 4322 }, { "epoch": 0.23047395638961454, "grad_norm": 2.2211899757385254, "learning_rate": 9.135600000000001e-06, "loss": 0.0313, "step": 4323 }, { "epoch": 0.23052726981926747, "grad_norm": 2.8829712867736816, "learning_rate": 9.1354e-06, "loss": 0.0407, "step": 4324 }, { "epoch": 0.2305805832489204, "grad_norm": 3.3758652210235596, "learning_rate": 9.1352e-06, "loss": 0.0399, "step": 4325 }, { "epoch": 0.23063389667857334, "grad_norm": 2.850083112716675, "learning_rate": 9.135e-06, "loss": 0.0443, "step": 4326 }, { "epoch": 0.23068721010822626, "grad_norm": 4.881485462188721, "learning_rate": 9.134800000000001e-06, "loss": 0.0396, "step": 4327 }, { "epoch": 0.2307405235378792, "grad_norm": 2.761291980743408, "learning_rate": 9.1346e-06, "loss": 0.0441, "step": 4328 }, { "epoch": 0.2307938369675321, "grad_norm": 3.259735584259033, "learning_rate": 9.134400000000002e-06, "loss": 0.0313, "step": 4329 }, { "epoch": 0.23084715039718506, "grad_norm": 4.7856950759887695, "learning_rate": 9.134200000000001e-06, "loss": 0.0449, "step": 4330 }, { "epoch": 0.230900463826838, "grad_norm": 1.7890006303787231, "learning_rate": 9.134000000000001e-06, "loss": 0.0315, "step": 4331 }, { "epoch": 0.2309537772564909, "grad_norm": 7.596116542816162, "learning_rate": 9.1338e-06, "loss": 0.0695, "step": 4332 }, { "epoch": 0.23100709068614383, "grad_norm": 4.979065418243408, "learning_rate": 9.1336e-06, "loss": 0.0475, "step": 4333 }, { "epoch": 0.23106040411579676, "grad_norm": 3.890355348587036, "learning_rate": 9.133400000000001e-06, "loss": 0.0601, "step": 4334 }, { "epoch": 0.2311137175454497, "grad_norm": 1.2059731483459473, "learning_rate": 9.133200000000001e-06, "loss": 0.0139, "step": 4335 }, { "epoch": 0.23116703097510263, "grad_norm": 4.006425380706787, "learning_rate": 9.133e-06, "loss": 0.0638, "step": 4336 }, { "epoch": 0.23122034440475556, "grad_norm": 5.935459613800049, "learning_rate": 9.1328e-06, "loss": 0.0603, "step": 4337 }, { "epoch": 0.23127365783440848, "grad_norm": 3.05287766456604, "learning_rate": 9.1326e-06, "loss": 0.0598, "step": 4338 }, { "epoch": 0.2313269712640614, "grad_norm": 5.012997150421143, "learning_rate": 9.132400000000001e-06, "loss": 0.0624, "step": 4339 }, { "epoch": 0.23138028469371436, "grad_norm": 6.757143974304199, "learning_rate": 9.1322e-06, "loss": 0.0409, "step": 4340 }, { "epoch": 0.23143359812336728, "grad_norm": 4.316866874694824, "learning_rate": 9.132000000000002e-06, "loss": 0.0807, "step": 4341 }, { "epoch": 0.2314869115530202, "grad_norm": 4.151782512664795, "learning_rate": 9.131800000000001e-06, "loss": 0.0488, "step": 4342 }, { "epoch": 0.23154022498267313, "grad_norm": 4.330130577087402, "learning_rate": 9.1316e-06, "loss": 0.0385, "step": 4343 }, { "epoch": 0.23159353841232608, "grad_norm": 3.9551124572753906, "learning_rate": 9.1314e-06, "loss": 0.0391, "step": 4344 }, { "epoch": 0.231646851841979, "grad_norm": 2.5440452098846436, "learning_rate": 9.131200000000002e-06, "loss": 0.0212, "step": 4345 }, { "epoch": 0.23170016527163192, "grad_norm": 3.766050338745117, "learning_rate": 9.131000000000001e-06, "loss": 0.0345, "step": 4346 }, { "epoch": 0.23175347870128485, "grad_norm": 3.706016778945923, "learning_rate": 9.1308e-06, "loss": 0.0497, "step": 4347 }, { "epoch": 0.23180679213093777, "grad_norm": 3.318603754043579, "learning_rate": 9.1306e-06, "loss": 0.029, "step": 4348 }, { "epoch": 0.23186010556059072, "grad_norm": 3.429884433746338, "learning_rate": 9.1304e-06, "loss": 0.0541, "step": 4349 }, { "epoch": 0.23191341899024365, "grad_norm": 4.690652370452881, "learning_rate": 9.130200000000001e-06, "loss": 0.0617, "step": 4350 }, { "epoch": 0.23196673241989657, "grad_norm": 2.3672521114349365, "learning_rate": 9.13e-06, "loss": 0.0409, "step": 4351 }, { "epoch": 0.2320200458495495, "grad_norm": 5.076229095458984, "learning_rate": 9.1298e-06, "loss": 0.0673, "step": 4352 }, { "epoch": 0.23207335927920242, "grad_norm": 5.73406982421875, "learning_rate": 9.129600000000001e-06, "loss": 0.0713, "step": 4353 }, { "epoch": 0.23212667270885537, "grad_norm": 4.871267795562744, "learning_rate": 9.129400000000001e-06, "loss": 0.098, "step": 4354 }, { "epoch": 0.2321799861385083, "grad_norm": 4.900058269500732, "learning_rate": 9.1292e-06, "loss": 0.0354, "step": 4355 }, { "epoch": 0.23223329956816122, "grad_norm": 4.494074821472168, "learning_rate": 9.129000000000002e-06, "loss": 0.0621, "step": 4356 }, { "epoch": 0.23228661299781414, "grad_norm": 4.979936122894287, "learning_rate": 9.128800000000001e-06, "loss": 0.0586, "step": 4357 }, { "epoch": 0.2323399264274671, "grad_norm": 7.631080150604248, "learning_rate": 9.128600000000001e-06, "loss": 0.0904, "step": 4358 }, { "epoch": 0.23239323985712002, "grad_norm": 3.488413095474243, "learning_rate": 9.1284e-06, "loss": 0.0259, "step": 4359 }, { "epoch": 0.23244655328677294, "grad_norm": 3.814549207687378, "learning_rate": 9.1282e-06, "loss": 0.0945, "step": 4360 }, { "epoch": 0.23249986671642586, "grad_norm": 2.0024302005767822, "learning_rate": 9.128e-06, "loss": 0.0335, "step": 4361 }, { "epoch": 0.2325531801460788, "grad_norm": 3.3936939239501953, "learning_rate": 9.127800000000001e-06, "loss": 0.0353, "step": 4362 }, { "epoch": 0.23260649357573174, "grad_norm": 5.378135681152344, "learning_rate": 9.1276e-06, "loss": 0.0601, "step": 4363 }, { "epoch": 0.23265980700538466, "grad_norm": 3.6170520782470703, "learning_rate": 9.1274e-06, "loss": 0.0333, "step": 4364 }, { "epoch": 0.23271312043503758, "grad_norm": 3.201812267303467, "learning_rate": 9.127200000000001e-06, "loss": 0.0275, "step": 4365 }, { "epoch": 0.2327664338646905, "grad_norm": 3.847334384918213, "learning_rate": 9.127000000000001e-06, "loss": 0.0375, "step": 4366 }, { "epoch": 0.23281974729434343, "grad_norm": 2.370171070098877, "learning_rate": 9.1268e-06, "loss": 0.0463, "step": 4367 }, { "epoch": 0.23287306072399638, "grad_norm": 2.862269639968872, "learning_rate": 9.126600000000002e-06, "loss": 0.0518, "step": 4368 }, { "epoch": 0.2329263741536493, "grad_norm": 2.6560916900634766, "learning_rate": 9.126400000000001e-06, "loss": 0.0345, "step": 4369 }, { "epoch": 0.23297968758330223, "grad_norm": 5.599900245666504, "learning_rate": 9.1262e-06, "loss": 0.0748, "step": 4370 }, { "epoch": 0.23303300101295515, "grad_norm": 2.7724838256835938, "learning_rate": 9.126e-06, "loss": 0.0373, "step": 4371 }, { "epoch": 0.2330863144426081, "grad_norm": 4.58217191696167, "learning_rate": 9.1258e-06, "loss": 0.0396, "step": 4372 }, { "epoch": 0.23313962787226103, "grad_norm": 3.444310426712036, "learning_rate": 9.125600000000001e-06, "loss": 0.0489, "step": 4373 }, { "epoch": 0.23319294130191395, "grad_norm": 3.023763418197632, "learning_rate": 9.1254e-06, "loss": 0.0442, "step": 4374 }, { "epoch": 0.23324625473156688, "grad_norm": 2.8230135440826416, "learning_rate": 9.1252e-06, "loss": 0.0454, "step": 4375 }, { "epoch": 0.2332995681612198, "grad_norm": 4.8400397300720215, "learning_rate": 9.125e-06, "loss": 0.0389, "step": 4376 }, { "epoch": 0.23335288159087275, "grad_norm": 4.584663391113281, "learning_rate": 9.124800000000001e-06, "loss": 0.0785, "step": 4377 }, { "epoch": 0.23340619502052568, "grad_norm": 3.404067277908325, "learning_rate": 9.1246e-06, "loss": 0.0517, "step": 4378 }, { "epoch": 0.2334595084501786, "grad_norm": 5.126509666442871, "learning_rate": 9.1244e-06, "loss": 0.0743, "step": 4379 }, { "epoch": 0.23351282187983152, "grad_norm": 2.770484447479248, "learning_rate": 9.124200000000002e-06, "loss": 0.0497, "step": 4380 }, { "epoch": 0.23356613530948445, "grad_norm": 2.9046988487243652, "learning_rate": 9.124000000000001e-06, "loss": 0.0435, "step": 4381 }, { "epoch": 0.2336194487391374, "grad_norm": 7.708545684814453, "learning_rate": 9.1238e-06, "loss": 0.0527, "step": 4382 }, { "epoch": 0.23367276216879032, "grad_norm": 3.4162817001342773, "learning_rate": 9.1236e-06, "loss": 0.0566, "step": 4383 }, { "epoch": 0.23372607559844324, "grad_norm": 8.248064041137695, "learning_rate": 9.123400000000001e-06, "loss": 0.0763, "step": 4384 }, { "epoch": 0.23377938902809617, "grad_norm": 3.4702980518341064, "learning_rate": 9.123200000000001e-06, "loss": 0.0319, "step": 4385 }, { "epoch": 0.23383270245774912, "grad_norm": 3.274564266204834, "learning_rate": 9.123e-06, "loss": 0.0664, "step": 4386 }, { "epoch": 0.23388601588740204, "grad_norm": 5.641150951385498, "learning_rate": 9.1228e-06, "loss": 0.1143, "step": 4387 }, { "epoch": 0.23393932931705497, "grad_norm": 4.2130208015441895, "learning_rate": 9.1226e-06, "loss": 0.0552, "step": 4388 }, { "epoch": 0.2339926427467079, "grad_norm": 5.8560967445373535, "learning_rate": 9.122400000000001e-06, "loss": 0.0448, "step": 4389 }, { "epoch": 0.23404595617636081, "grad_norm": 4.622434616088867, "learning_rate": 9.1222e-06, "loss": 0.0561, "step": 4390 }, { "epoch": 0.23409926960601377, "grad_norm": 5.079837799072266, "learning_rate": 9.122e-06, "loss": 0.0612, "step": 4391 }, { "epoch": 0.2341525830356667, "grad_norm": 4.666459560394287, "learning_rate": 9.121800000000001e-06, "loss": 0.0607, "step": 4392 }, { "epoch": 0.2342058964653196, "grad_norm": 3.8032901287078857, "learning_rate": 9.121600000000001e-06, "loss": 0.0649, "step": 4393 }, { "epoch": 0.23425920989497254, "grad_norm": 2.736264944076538, "learning_rate": 9.1214e-06, "loss": 0.0377, "step": 4394 }, { "epoch": 0.23431252332462546, "grad_norm": 3.7793829441070557, "learning_rate": 9.121200000000002e-06, "loss": 0.0608, "step": 4395 }, { "epoch": 0.2343658367542784, "grad_norm": 9.421759605407715, "learning_rate": 9.121000000000001e-06, "loss": 0.0657, "step": 4396 }, { "epoch": 0.23441915018393134, "grad_norm": 3.632037401199341, "learning_rate": 9.1208e-06, "loss": 0.0668, "step": 4397 }, { "epoch": 0.23447246361358426, "grad_norm": 3.2947423458099365, "learning_rate": 9.1206e-06, "loss": 0.058, "step": 4398 }, { "epoch": 0.23452577704323718, "grad_norm": 4.133000373840332, "learning_rate": 9.1204e-06, "loss": 0.044, "step": 4399 }, { "epoch": 0.23457909047289013, "grad_norm": 3.725700616836548, "learning_rate": 9.120200000000001e-06, "loss": 0.0331, "step": 4400 }, { "epoch": 0.23463240390254306, "grad_norm": 5.09810209274292, "learning_rate": 9.12e-06, "loss": 0.078, "step": 4401 }, { "epoch": 0.23468571733219598, "grad_norm": 5.3273210525512695, "learning_rate": 9.1198e-06, "loss": 0.0505, "step": 4402 }, { "epoch": 0.2347390307618489, "grad_norm": 2.3984017372131348, "learning_rate": 9.1196e-06, "loss": 0.0334, "step": 4403 }, { "epoch": 0.23479234419150183, "grad_norm": 13.410295486450195, "learning_rate": 9.119400000000001e-06, "loss": 0.0996, "step": 4404 }, { "epoch": 0.23484565762115478, "grad_norm": 2.121405839920044, "learning_rate": 9.1192e-06, "loss": 0.0228, "step": 4405 }, { "epoch": 0.2348989710508077, "grad_norm": 5.04721212387085, "learning_rate": 9.119000000000002e-06, "loss": 0.0444, "step": 4406 }, { "epoch": 0.23495228448046063, "grad_norm": 3.727958917617798, "learning_rate": 9.118800000000002e-06, "loss": 0.0538, "step": 4407 }, { "epoch": 0.23500559791011355, "grad_norm": 9.829797744750977, "learning_rate": 9.118600000000001e-06, "loss": 0.1594, "step": 4408 }, { "epoch": 0.23505891133976647, "grad_norm": 3.731304168701172, "learning_rate": 9.1184e-06, "loss": 0.0548, "step": 4409 }, { "epoch": 0.23511222476941943, "grad_norm": 5.9034647941589355, "learning_rate": 9.1182e-06, "loss": 0.0775, "step": 4410 }, { "epoch": 0.23516553819907235, "grad_norm": 5.729189395904541, "learning_rate": 9.118000000000001e-06, "loss": 0.0798, "step": 4411 }, { "epoch": 0.23521885162872527, "grad_norm": 13.777924537658691, "learning_rate": 9.117800000000001e-06, "loss": 0.0779, "step": 4412 }, { "epoch": 0.2352721650583782, "grad_norm": 7.000422954559326, "learning_rate": 9.1176e-06, "loss": 0.071, "step": 4413 }, { "epoch": 0.23532547848803115, "grad_norm": 4.0470476150512695, "learning_rate": 9.1174e-06, "loss": 0.0358, "step": 4414 }, { "epoch": 0.23537879191768407, "grad_norm": 4.1258697509765625, "learning_rate": 9.1172e-06, "loss": 0.0644, "step": 4415 }, { "epoch": 0.235432105347337, "grad_norm": 6.839288711547852, "learning_rate": 9.117000000000001e-06, "loss": 0.0835, "step": 4416 }, { "epoch": 0.23548541877698992, "grad_norm": 9.641347885131836, "learning_rate": 9.1168e-06, "loss": 0.0827, "step": 4417 }, { "epoch": 0.23553873220664284, "grad_norm": 10.341816902160645, "learning_rate": 9.116600000000002e-06, "loss": 0.0868, "step": 4418 }, { "epoch": 0.2355920456362958, "grad_norm": 2.746389627456665, "learning_rate": 9.116400000000001e-06, "loss": 0.0392, "step": 4419 }, { "epoch": 0.23564535906594872, "grad_norm": 3.4437789916992188, "learning_rate": 9.116200000000001e-06, "loss": 0.0317, "step": 4420 }, { "epoch": 0.23569867249560164, "grad_norm": 2.4399454593658447, "learning_rate": 9.116e-06, "loss": 0.0373, "step": 4421 }, { "epoch": 0.23575198592525456, "grad_norm": 2.7043187618255615, "learning_rate": 9.1158e-06, "loss": 0.0486, "step": 4422 }, { "epoch": 0.2358052993549075, "grad_norm": 9.645702362060547, "learning_rate": 9.115600000000001e-06, "loss": 0.1117, "step": 4423 }, { "epoch": 0.23585861278456044, "grad_norm": 7.4173407554626465, "learning_rate": 9.115400000000001e-06, "loss": 0.0805, "step": 4424 }, { "epoch": 0.23591192621421336, "grad_norm": 6.1596598625183105, "learning_rate": 9.1152e-06, "loss": 0.0503, "step": 4425 }, { "epoch": 0.2359652396438663, "grad_norm": 4.5177836418151855, "learning_rate": 9.115e-06, "loss": 0.0533, "step": 4426 }, { "epoch": 0.2360185530735192, "grad_norm": 7.1287150382995605, "learning_rate": 9.1148e-06, "loss": 0.0464, "step": 4427 }, { "epoch": 0.23607186650317216, "grad_norm": 3.594045400619507, "learning_rate": 9.1146e-06, "loss": 0.068, "step": 4428 }, { "epoch": 0.23612517993282509, "grad_norm": 3.933706760406494, "learning_rate": 9.1144e-06, "loss": 0.0347, "step": 4429 }, { "epoch": 0.236178493362478, "grad_norm": 3.5864405632019043, "learning_rate": 9.114200000000002e-06, "loss": 0.0327, "step": 4430 }, { "epoch": 0.23623180679213093, "grad_norm": 4.331587791442871, "learning_rate": 9.114000000000001e-06, "loss": 0.0692, "step": 4431 }, { "epoch": 0.23628512022178386, "grad_norm": 5.713861465454102, "learning_rate": 9.1138e-06, "loss": 0.0631, "step": 4432 }, { "epoch": 0.2363384336514368, "grad_norm": 6.168315887451172, "learning_rate": 9.1136e-06, "loss": 0.057, "step": 4433 }, { "epoch": 0.23639174708108973, "grad_norm": 1.8854092359542847, "learning_rate": 9.113400000000002e-06, "loss": 0.0145, "step": 4434 }, { "epoch": 0.23644506051074266, "grad_norm": 5.407592296600342, "learning_rate": 9.113200000000001e-06, "loss": 0.0615, "step": 4435 }, { "epoch": 0.23649837394039558, "grad_norm": 3.5029022693634033, "learning_rate": 9.113e-06, "loss": 0.0463, "step": 4436 }, { "epoch": 0.2365516873700485, "grad_norm": 3.9621410369873047, "learning_rate": 9.1128e-06, "loss": 0.0501, "step": 4437 }, { "epoch": 0.23660500079970145, "grad_norm": 6.145812034606934, "learning_rate": 9.1126e-06, "loss": 0.108, "step": 4438 }, { "epoch": 0.23665831422935438, "grad_norm": 4.601186752319336, "learning_rate": 9.112400000000001e-06, "loss": 0.0573, "step": 4439 }, { "epoch": 0.2367116276590073, "grad_norm": 4.015514850616455, "learning_rate": 9.1122e-06, "loss": 0.0601, "step": 4440 }, { "epoch": 0.23676494108866022, "grad_norm": 6.925798416137695, "learning_rate": 9.112e-06, "loss": 0.0668, "step": 4441 }, { "epoch": 0.23681825451831318, "grad_norm": 2.7132184505462646, "learning_rate": 9.111800000000001e-06, "loss": 0.039, "step": 4442 }, { "epoch": 0.2368715679479661, "grad_norm": 4.119056701660156, "learning_rate": 9.111600000000001e-06, "loss": 0.0486, "step": 4443 }, { "epoch": 0.23692488137761902, "grad_norm": 3.6335701942443848, "learning_rate": 9.1114e-06, "loss": 0.0326, "step": 4444 }, { "epoch": 0.23697819480727195, "grad_norm": 4.699387550354004, "learning_rate": 9.111200000000002e-06, "loss": 0.0452, "step": 4445 }, { "epoch": 0.23703150823692487, "grad_norm": 2.42925763130188, "learning_rate": 9.111000000000001e-06, "loss": 0.0461, "step": 4446 }, { "epoch": 0.23708482166657782, "grad_norm": 3.681856155395508, "learning_rate": 9.110800000000001e-06, "loss": 0.0439, "step": 4447 }, { "epoch": 0.23713813509623075, "grad_norm": 5.384443283081055, "learning_rate": 9.1106e-06, "loss": 0.0899, "step": 4448 }, { "epoch": 0.23719144852588367, "grad_norm": 3.289461135864258, "learning_rate": 9.1104e-06, "loss": 0.0483, "step": 4449 }, { "epoch": 0.2372447619555366, "grad_norm": 3.615483522415161, "learning_rate": 9.110200000000001e-06, "loss": 0.0438, "step": 4450 }, { "epoch": 0.23729807538518952, "grad_norm": 4.256819248199463, "learning_rate": 9.110000000000001e-06, "loss": 0.0637, "step": 4451 }, { "epoch": 0.23735138881484247, "grad_norm": 4.985501289367676, "learning_rate": 9.1098e-06, "loss": 0.0836, "step": 4452 }, { "epoch": 0.2374047022444954, "grad_norm": 5.55941104888916, "learning_rate": 9.1096e-06, "loss": 0.0593, "step": 4453 }, { "epoch": 0.23745801567414832, "grad_norm": 4.8799967765808105, "learning_rate": 9.109400000000001e-06, "loss": 0.0652, "step": 4454 }, { "epoch": 0.23751132910380124, "grad_norm": 4.100747108459473, "learning_rate": 9.1092e-06, "loss": 0.0362, "step": 4455 }, { "epoch": 0.2375646425334542, "grad_norm": 2.680856704711914, "learning_rate": 9.109e-06, "loss": 0.0418, "step": 4456 }, { "epoch": 0.2376179559631071, "grad_norm": 3.068399429321289, "learning_rate": 9.108800000000002e-06, "loss": 0.0491, "step": 4457 }, { "epoch": 0.23767126939276004, "grad_norm": 4.053610324859619, "learning_rate": 9.108600000000001e-06, "loss": 0.0612, "step": 4458 }, { "epoch": 0.23772458282241296, "grad_norm": 3.2196826934814453, "learning_rate": 9.1084e-06, "loss": 0.0271, "step": 4459 }, { "epoch": 0.23777789625206588, "grad_norm": 5.120584964752197, "learning_rate": 9.1082e-06, "loss": 0.0921, "step": 4460 }, { "epoch": 0.23783120968171884, "grad_norm": 4.824377536773682, "learning_rate": 9.108000000000002e-06, "loss": 0.0613, "step": 4461 }, { "epoch": 0.23788452311137176, "grad_norm": 5.22797966003418, "learning_rate": 9.107800000000001e-06, "loss": 0.0317, "step": 4462 }, { "epoch": 0.23793783654102468, "grad_norm": 8.333134651184082, "learning_rate": 9.1076e-06, "loss": 0.0757, "step": 4463 }, { "epoch": 0.2379911499706776, "grad_norm": 5.25455904006958, "learning_rate": 9.1074e-06, "loss": 0.0628, "step": 4464 }, { "epoch": 0.23804446340033053, "grad_norm": 4.456216335296631, "learning_rate": 9.1072e-06, "loss": 0.052, "step": 4465 }, { "epoch": 0.23809777682998348, "grad_norm": 3.664275646209717, "learning_rate": 9.107000000000001e-06, "loss": 0.0673, "step": 4466 }, { "epoch": 0.2381510902596364, "grad_norm": 4.84494161605835, "learning_rate": 9.1068e-06, "loss": 0.0643, "step": 4467 }, { "epoch": 0.23820440368928933, "grad_norm": 4.279231071472168, "learning_rate": 9.1066e-06, "loss": 0.067, "step": 4468 }, { "epoch": 0.23825771711894225, "grad_norm": 4.035869121551514, "learning_rate": 9.106400000000001e-06, "loss": 0.0387, "step": 4469 }, { "epoch": 0.2383110305485952, "grad_norm": 4.726657867431641, "learning_rate": 9.106200000000001e-06, "loss": 0.054, "step": 4470 }, { "epoch": 0.23836434397824813, "grad_norm": 4.042085647583008, "learning_rate": 9.106e-06, "loss": 0.0626, "step": 4471 }, { "epoch": 0.23841765740790105, "grad_norm": 5.27855920791626, "learning_rate": 9.1058e-06, "loss": 0.0351, "step": 4472 }, { "epoch": 0.23847097083755397, "grad_norm": 2.22416615486145, "learning_rate": 9.105600000000001e-06, "loss": 0.0293, "step": 4473 }, { "epoch": 0.2385242842672069, "grad_norm": 2.2465884685516357, "learning_rate": 9.105400000000001e-06, "loss": 0.0421, "step": 4474 }, { "epoch": 0.23857759769685985, "grad_norm": 3.4526171684265137, "learning_rate": 9.1052e-06, "loss": 0.0319, "step": 4475 }, { "epoch": 0.23863091112651277, "grad_norm": 4.313796043395996, "learning_rate": 9.105e-06, "loss": 0.0404, "step": 4476 }, { "epoch": 0.2386842245561657, "grad_norm": 4.25489616394043, "learning_rate": 9.1048e-06, "loss": 0.0672, "step": 4477 }, { "epoch": 0.23873753798581862, "grad_norm": 3.155155658721924, "learning_rate": 9.104600000000001e-06, "loss": 0.0583, "step": 4478 }, { "epoch": 0.23879085141547154, "grad_norm": 4.407638072967529, "learning_rate": 9.1044e-06, "loss": 0.0604, "step": 4479 }, { "epoch": 0.2388441648451245, "grad_norm": 4.985938549041748, "learning_rate": 9.1042e-06, "loss": 0.1213, "step": 4480 }, { "epoch": 0.23889747827477742, "grad_norm": 4.51473331451416, "learning_rate": 9.104000000000001e-06, "loss": 0.0552, "step": 4481 }, { "epoch": 0.23895079170443034, "grad_norm": 3.981020212173462, "learning_rate": 9.1038e-06, "loss": 0.0322, "step": 4482 }, { "epoch": 0.23900410513408327, "grad_norm": 3.5441925525665283, "learning_rate": 9.1036e-06, "loss": 0.0385, "step": 4483 }, { "epoch": 0.23905741856373622, "grad_norm": 3.420520067214966, "learning_rate": 9.103400000000002e-06, "loss": 0.0375, "step": 4484 }, { "epoch": 0.23911073199338914, "grad_norm": 3.330859422683716, "learning_rate": 9.103200000000001e-06, "loss": 0.0332, "step": 4485 }, { "epoch": 0.23916404542304207, "grad_norm": 4.210722923278809, "learning_rate": 9.103e-06, "loss": 0.0507, "step": 4486 }, { "epoch": 0.239217358852695, "grad_norm": 4.636994361877441, "learning_rate": 9.1028e-06, "loss": 0.0481, "step": 4487 }, { "epoch": 0.2392706722823479, "grad_norm": 6.766992092132568, "learning_rate": 9.1026e-06, "loss": 0.0502, "step": 4488 }, { "epoch": 0.23932398571200086, "grad_norm": 3.261826992034912, "learning_rate": 9.102400000000001e-06, "loss": 0.0377, "step": 4489 }, { "epoch": 0.2393772991416538, "grad_norm": 4.767095565795898, "learning_rate": 9.1022e-06, "loss": 0.0522, "step": 4490 }, { "epoch": 0.2394306125713067, "grad_norm": 3.6430153846740723, "learning_rate": 9.102e-06, "loss": 0.0444, "step": 4491 }, { "epoch": 0.23948392600095963, "grad_norm": 3.275264263153076, "learning_rate": 9.1018e-06, "loss": 0.0329, "step": 4492 }, { "epoch": 0.23953723943061256, "grad_norm": 4.346240520477295, "learning_rate": 9.101600000000001e-06, "loss": 0.0634, "step": 4493 }, { "epoch": 0.2395905528602655, "grad_norm": 3.195162296295166, "learning_rate": 9.1014e-06, "loss": 0.0404, "step": 4494 }, { "epoch": 0.23964386628991843, "grad_norm": 3.7428205013275146, "learning_rate": 9.101200000000002e-06, "loss": 0.0477, "step": 4495 }, { "epoch": 0.23969717971957136, "grad_norm": 4.716318130493164, "learning_rate": 9.101000000000001e-06, "loss": 0.0505, "step": 4496 }, { "epoch": 0.23975049314922428, "grad_norm": 5.509292125701904, "learning_rate": 9.100800000000001e-06, "loss": 0.0519, "step": 4497 }, { "epoch": 0.23980380657887723, "grad_norm": 6.553747177124023, "learning_rate": 9.1006e-06, "loss": 0.0667, "step": 4498 }, { "epoch": 0.23985712000853016, "grad_norm": 4.222931861877441, "learning_rate": 9.1004e-06, "loss": 0.0316, "step": 4499 }, { "epoch": 0.23991043343818308, "grad_norm": 2.973330497741699, "learning_rate": 9.100200000000001e-06, "loss": 0.0285, "step": 4500 }, { "epoch": 0.239963746867836, "grad_norm": 9.8207368850708, "learning_rate": 9.100000000000001e-06, "loss": 0.0987, "step": 4501 }, { "epoch": 0.24001706029748893, "grad_norm": 2.485694408416748, "learning_rate": 9.0998e-06, "loss": 0.0213, "step": 4502 }, { "epoch": 0.24007037372714188, "grad_norm": 4.394402980804443, "learning_rate": 9.0996e-06, "loss": 0.063, "step": 4503 }, { "epoch": 0.2401236871567948, "grad_norm": 3.144497871398926, "learning_rate": 9.0994e-06, "loss": 0.0538, "step": 4504 }, { "epoch": 0.24017700058644773, "grad_norm": 1.8306597471237183, "learning_rate": 9.099200000000001e-06, "loss": 0.0243, "step": 4505 }, { "epoch": 0.24023031401610065, "grad_norm": 3.7484889030456543, "learning_rate": 9.099e-06, "loss": 0.044, "step": 4506 }, { "epoch": 0.24028362744575357, "grad_norm": 4.333733558654785, "learning_rate": 9.098800000000002e-06, "loss": 0.0502, "step": 4507 }, { "epoch": 0.24033694087540652, "grad_norm": 7.613016128540039, "learning_rate": 9.098600000000001e-06, "loss": 0.0543, "step": 4508 }, { "epoch": 0.24039025430505945, "grad_norm": 3.963592052459717, "learning_rate": 9.0984e-06, "loss": 0.0519, "step": 4509 }, { "epoch": 0.24044356773471237, "grad_norm": 3.3695175647735596, "learning_rate": 9.0982e-06, "loss": 0.0626, "step": 4510 }, { "epoch": 0.2404968811643653, "grad_norm": 2.9403865337371826, "learning_rate": 9.098000000000002e-06, "loss": 0.0549, "step": 4511 }, { "epoch": 0.24055019459401825, "grad_norm": 4.586994647979736, "learning_rate": 9.097800000000001e-06, "loss": 0.0699, "step": 4512 }, { "epoch": 0.24060350802367117, "grad_norm": 1.857216477394104, "learning_rate": 9.0976e-06, "loss": 0.0199, "step": 4513 }, { "epoch": 0.2406568214533241, "grad_norm": 2.6391048431396484, "learning_rate": 9.0974e-06, "loss": 0.0714, "step": 4514 }, { "epoch": 0.24071013488297702, "grad_norm": 3.519660234451294, "learning_rate": 9.0972e-06, "loss": 0.0584, "step": 4515 }, { "epoch": 0.24076344831262994, "grad_norm": 3.322774648666382, "learning_rate": 9.097e-06, "loss": 0.0359, "step": 4516 }, { "epoch": 0.2408167617422829, "grad_norm": 2.532680034637451, "learning_rate": 9.0968e-06, "loss": 0.029, "step": 4517 }, { "epoch": 0.24087007517193582, "grad_norm": 1.6835834980010986, "learning_rate": 9.0966e-06, "loss": 0.0153, "step": 4518 }, { "epoch": 0.24092338860158874, "grad_norm": 4.007778644561768, "learning_rate": 9.096400000000002e-06, "loss": 0.0686, "step": 4519 }, { "epoch": 0.24097670203124166, "grad_norm": 4.538104057312012, "learning_rate": 9.096200000000001e-06, "loss": 0.0664, "step": 4520 }, { "epoch": 0.2410300154608946, "grad_norm": 4.317060947418213, "learning_rate": 9.096e-06, "loss": 0.0598, "step": 4521 }, { "epoch": 0.24108332889054754, "grad_norm": 5.231680393218994, "learning_rate": 9.095800000000002e-06, "loss": 0.0689, "step": 4522 }, { "epoch": 0.24113664232020046, "grad_norm": 4.529483795166016, "learning_rate": 9.095600000000001e-06, "loss": 0.069, "step": 4523 }, { "epoch": 0.24118995574985339, "grad_norm": 8.027828216552734, "learning_rate": 9.095400000000001e-06, "loss": 0.0749, "step": 4524 }, { "epoch": 0.2412432691795063, "grad_norm": 3.222670078277588, "learning_rate": 9.0952e-06, "loss": 0.0498, "step": 4525 }, { "epoch": 0.24129658260915926, "grad_norm": 2.8542466163635254, "learning_rate": 9.095e-06, "loss": 0.0158, "step": 4526 }, { "epoch": 0.24134989603881218, "grad_norm": 6.18346643447876, "learning_rate": 9.0948e-06, "loss": 0.0853, "step": 4527 }, { "epoch": 0.2414032094684651, "grad_norm": 4.079136848449707, "learning_rate": 9.094600000000001e-06, "loss": 0.0942, "step": 4528 }, { "epoch": 0.24145652289811803, "grad_norm": 5.716500759124756, "learning_rate": 9.0944e-06, "loss": 0.0526, "step": 4529 }, { "epoch": 0.24150983632777095, "grad_norm": 3.932544708251953, "learning_rate": 9.0942e-06, "loss": 0.0604, "step": 4530 }, { "epoch": 0.2415631497574239, "grad_norm": 5.416529655456543, "learning_rate": 9.094000000000001e-06, "loss": 0.0459, "step": 4531 }, { "epoch": 0.24161646318707683, "grad_norm": 3.4210245609283447, "learning_rate": 9.093800000000001e-06, "loss": 0.0612, "step": 4532 }, { "epoch": 0.24166977661672975, "grad_norm": 3.0827691555023193, "learning_rate": 9.0936e-06, "loss": 0.0449, "step": 4533 }, { "epoch": 0.24172309004638268, "grad_norm": 2.42728328704834, "learning_rate": 9.093400000000002e-06, "loss": 0.0313, "step": 4534 }, { "epoch": 0.2417764034760356, "grad_norm": 3.061094045639038, "learning_rate": 9.093200000000001e-06, "loss": 0.0364, "step": 4535 }, { "epoch": 0.24182971690568855, "grad_norm": 3.4538681507110596, "learning_rate": 9.093000000000001e-06, "loss": 0.0538, "step": 4536 }, { "epoch": 0.24188303033534148, "grad_norm": 4.546746730804443, "learning_rate": 9.0928e-06, "loss": 0.0666, "step": 4537 }, { "epoch": 0.2419363437649944, "grad_norm": 4.4986572265625, "learning_rate": 9.0926e-06, "loss": 0.0633, "step": 4538 }, { "epoch": 0.24198965719464732, "grad_norm": 5.4492268562316895, "learning_rate": 9.092400000000001e-06, "loss": 0.091, "step": 4539 }, { "epoch": 0.24204297062430027, "grad_norm": 2.4400458335876465, "learning_rate": 9.0922e-06, "loss": 0.0371, "step": 4540 }, { "epoch": 0.2420962840539532, "grad_norm": 4.60254430770874, "learning_rate": 9.092e-06, "loss": 0.0503, "step": 4541 }, { "epoch": 0.24214959748360612, "grad_norm": 3.8808586597442627, "learning_rate": 9.0918e-06, "loss": 0.0364, "step": 4542 }, { "epoch": 0.24220291091325905, "grad_norm": 3.6410069465637207, "learning_rate": 9.091600000000001e-06, "loss": 0.0438, "step": 4543 }, { "epoch": 0.24225622434291197, "grad_norm": 7.974427223205566, "learning_rate": 9.0914e-06, "loss": 0.0703, "step": 4544 }, { "epoch": 0.24230953777256492, "grad_norm": 4.585545063018799, "learning_rate": 9.0912e-06, "loss": 0.0624, "step": 4545 }, { "epoch": 0.24236285120221784, "grad_norm": 3.8215646743774414, "learning_rate": 9.091000000000002e-06, "loss": 0.0521, "step": 4546 }, { "epoch": 0.24241616463187077, "grad_norm": 3.2390449047088623, "learning_rate": 9.090800000000001e-06, "loss": 0.06, "step": 4547 }, { "epoch": 0.2424694780615237, "grad_norm": 4.843366622924805, "learning_rate": 9.0906e-06, "loss": 0.0453, "step": 4548 }, { "epoch": 0.24252279149117661, "grad_norm": 2.979407787322998, "learning_rate": 9.0904e-06, "loss": 0.0307, "step": 4549 }, { "epoch": 0.24257610492082957, "grad_norm": 3.2468903064727783, "learning_rate": 9.090200000000001e-06, "loss": 0.0379, "step": 4550 }, { "epoch": 0.2426294183504825, "grad_norm": 3.870774745941162, "learning_rate": 9.090000000000001e-06, "loss": 0.0381, "step": 4551 }, { "epoch": 0.2426827317801354, "grad_norm": 2.6395840644836426, "learning_rate": 9.0898e-06, "loss": 0.0271, "step": 4552 }, { "epoch": 0.24273604520978834, "grad_norm": 4.127944469451904, "learning_rate": 9.0896e-06, "loss": 0.0898, "step": 4553 }, { "epoch": 0.2427893586394413, "grad_norm": 6.184939861297607, "learning_rate": 9.0894e-06, "loss": 0.0712, "step": 4554 }, { "epoch": 0.2428426720690942, "grad_norm": 4.2402191162109375, "learning_rate": 9.089200000000001e-06, "loss": 0.0563, "step": 4555 }, { "epoch": 0.24289598549874714, "grad_norm": 4.695191860198975, "learning_rate": 9.089e-06, "loss": 0.0652, "step": 4556 }, { "epoch": 0.24294929892840006, "grad_norm": 2.2910640239715576, "learning_rate": 9.0888e-06, "loss": 0.0294, "step": 4557 }, { "epoch": 0.24300261235805298, "grad_norm": 4.531547546386719, "learning_rate": 9.088600000000001e-06, "loss": 0.0721, "step": 4558 }, { "epoch": 0.24305592578770593, "grad_norm": 3.427561044692993, "learning_rate": 9.088400000000001e-06, "loss": 0.0553, "step": 4559 }, { "epoch": 0.24310923921735886, "grad_norm": 2.946542501449585, "learning_rate": 9.0882e-06, "loss": 0.0505, "step": 4560 }, { "epoch": 0.24316255264701178, "grad_norm": 4.284055709838867, "learning_rate": 9.088000000000002e-06, "loss": 0.0674, "step": 4561 }, { "epoch": 0.2432158660766647, "grad_norm": 5.949391841888428, "learning_rate": 9.087800000000001e-06, "loss": 0.0584, "step": 4562 }, { "epoch": 0.24326917950631763, "grad_norm": 3.5013880729675293, "learning_rate": 9.087600000000001e-06, "loss": 0.0521, "step": 4563 }, { "epoch": 0.24332249293597058, "grad_norm": 3.2207651138305664, "learning_rate": 9.0874e-06, "loss": 0.0535, "step": 4564 }, { "epoch": 0.2433758063656235, "grad_norm": 3.428118944168091, "learning_rate": 9.0872e-06, "loss": 0.0466, "step": 4565 }, { "epoch": 0.24342911979527643, "grad_norm": 4.839649677276611, "learning_rate": 9.087e-06, "loss": 0.0468, "step": 4566 }, { "epoch": 0.24348243322492935, "grad_norm": 3.5328383445739746, "learning_rate": 9.0868e-06, "loss": 0.0589, "step": 4567 }, { "epoch": 0.2435357466545823, "grad_norm": 3.829347610473633, "learning_rate": 9.0866e-06, "loss": 0.0388, "step": 4568 }, { "epoch": 0.24358906008423523, "grad_norm": 2.720430374145508, "learning_rate": 9.0864e-06, "loss": 0.0454, "step": 4569 }, { "epoch": 0.24364237351388815, "grad_norm": 2.4929049015045166, "learning_rate": 9.086200000000001e-06, "loss": 0.0362, "step": 4570 }, { "epoch": 0.24369568694354107, "grad_norm": 2.9786386489868164, "learning_rate": 9.086e-06, "loss": 0.0483, "step": 4571 }, { "epoch": 0.243749000373194, "grad_norm": 3.4458651542663574, "learning_rate": 9.085800000000002e-06, "loss": 0.0345, "step": 4572 }, { "epoch": 0.24380231380284695, "grad_norm": 3.2711338996887207, "learning_rate": 9.085600000000002e-06, "loss": 0.0646, "step": 4573 }, { "epoch": 0.24385562723249987, "grad_norm": 3.157843589782715, "learning_rate": 9.085400000000001e-06, "loss": 0.0533, "step": 4574 }, { "epoch": 0.2439089406621528, "grad_norm": 2.3085994720458984, "learning_rate": 9.0852e-06, "loss": 0.0382, "step": 4575 }, { "epoch": 0.24396225409180572, "grad_norm": 4.877254486083984, "learning_rate": 9.085e-06, "loss": 0.0632, "step": 4576 }, { "epoch": 0.24401556752145864, "grad_norm": 4.4051947593688965, "learning_rate": 9.0848e-06, "loss": 0.0567, "step": 4577 }, { "epoch": 0.2440688809511116, "grad_norm": 3.799919843673706, "learning_rate": 9.084600000000001e-06, "loss": 0.0818, "step": 4578 }, { "epoch": 0.24412219438076452, "grad_norm": 2.3694851398468018, "learning_rate": 9.0844e-06, "loss": 0.0169, "step": 4579 }, { "epoch": 0.24417550781041744, "grad_norm": 5.2666015625, "learning_rate": 9.0842e-06, "loss": 0.0753, "step": 4580 }, { "epoch": 0.24422882124007037, "grad_norm": 6.366913795471191, "learning_rate": 9.084e-06, "loss": 0.0642, "step": 4581 }, { "epoch": 0.24428213466972332, "grad_norm": 3.8298869132995605, "learning_rate": 9.083800000000001e-06, "loss": 0.0551, "step": 4582 }, { "epoch": 0.24433544809937624, "grad_norm": 4.335777759552002, "learning_rate": 9.0836e-06, "loss": 0.0692, "step": 4583 }, { "epoch": 0.24438876152902916, "grad_norm": 4.8201518058776855, "learning_rate": 9.083400000000002e-06, "loss": 0.0538, "step": 4584 }, { "epoch": 0.2444420749586821, "grad_norm": 3.958176612854004, "learning_rate": 9.083200000000001e-06, "loss": 0.06, "step": 4585 }, { "epoch": 0.244495388388335, "grad_norm": 2.8305504322052, "learning_rate": 9.083000000000001e-06, "loss": 0.0335, "step": 4586 }, { "epoch": 0.24454870181798796, "grad_norm": 2.892691135406494, "learning_rate": 9.0828e-06, "loss": 0.037, "step": 4587 }, { "epoch": 0.24460201524764089, "grad_norm": 6.896413326263428, "learning_rate": 9.0826e-06, "loss": 0.0749, "step": 4588 }, { "epoch": 0.2446553286772938, "grad_norm": 3.8096182346343994, "learning_rate": 9.082400000000001e-06, "loss": 0.0488, "step": 4589 }, { "epoch": 0.24470864210694673, "grad_norm": 3.875786304473877, "learning_rate": 9.082200000000001e-06, "loss": 0.0631, "step": 4590 }, { "epoch": 0.24476195553659966, "grad_norm": 5.973961353302002, "learning_rate": 9.082e-06, "loss": 0.0556, "step": 4591 }, { "epoch": 0.2448152689662526, "grad_norm": 7.098196506500244, "learning_rate": 9.0818e-06, "loss": 0.0578, "step": 4592 }, { "epoch": 0.24486858239590553, "grad_norm": 2.1329505443573, "learning_rate": 9.0816e-06, "loss": 0.0353, "step": 4593 }, { "epoch": 0.24492189582555846, "grad_norm": 2.264775514602661, "learning_rate": 9.0814e-06, "loss": 0.0205, "step": 4594 }, { "epoch": 0.24497520925521138, "grad_norm": 4.816076755523682, "learning_rate": 9.0812e-06, "loss": 0.0719, "step": 4595 }, { "epoch": 0.24502852268486433, "grad_norm": 3.0788967609405518, "learning_rate": 9.081000000000002e-06, "loss": 0.0608, "step": 4596 }, { "epoch": 0.24508183611451725, "grad_norm": 2.209014415740967, "learning_rate": 9.080800000000001e-06, "loss": 0.021, "step": 4597 }, { "epoch": 0.24513514954417018, "grad_norm": 2.734238862991333, "learning_rate": 9.0806e-06, "loss": 0.0392, "step": 4598 }, { "epoch": 0.2451884629738231, "grad_norm": 4.230685234069824, "learning_rate": 9.0804e-06, "loss": 0.0413, "step": 4599 }, { "epoch": 0.24524177640347603, "grad_norm": 3.4035348892211914, "learning_rate": 9.080200000000002e-06, "loss": 0.0533, "step": 4600 }, { "epoch": 0.24529508983312898, "grad_norm": 9.252511024475098, "learning_rate": 9.080000000000001e-06, "loss": 0.0792, "step": 4601 }, { "epoch": 0.2453484032627819, "grad_norm": 4.148297309875488, "learning_rate": 9.0798e-06, "loss": 0.0508, "step": 4602 }, { "epoch": 0.24540171669243482, "grad_norm": 2.9415783882141113, "learning_rate": 9.0796e-06, "loss": 0.0388, "step": 4603 }, { "epoch": 0.24545503012208775, "grad_norm": 3.163543701171875, "learning_rate": 9.0794e-06, "loss": 0.0354, "step": 4604 }, { "epoch": 0.24550834355174067, "grad_norm": 3.0544066429138184, "learning_rate": 9.0792e-06, "loss": 0.0218, "step": 4605 }, { "epoch": 0.24556165698139362, "grad_norm": 2.607691764831543, "learning_rate": 9.079e-06, "loss": 0.035, "step": 4606 }, { "epoch": 0.24561497041104655, "grad_norm": 4.322522163391113, "learning_rate": 9.0788e-06, "loss": 0.0576, "step": 4607 }, { "epoch": 0.24566828384069947, "grad_norm": 3.5032877922058105, "learning_rate": 9.078600000000001e-06, "loss": 0.0426, "step": 4608 }, { "epoch": 0.2457215972703524, "grad_norm": 4.042177200317383, "learning_rate": 9.078400000000001e-06, "loss": 0.0283, "step": 4609 }, { "epoch": 0.24577491070000534, "grad_norm": 5.811893463134766, "learning_rate": 9.0782e-06, "loss": 0.0826, "step": 4610 }, { "epoch": 0.24582822412965827, "grad_norm": 5.269220352172852, "learning_rate": 9.078000000000002e-06, "loss": 0.0707, "step": 4611 }, { "epoch": 0.2458815375593112, "grad_norm": 4.152690410614014, "learning_rate": 9.077800000000001e-06, "loss": 0.0434, "step": 4612 }, { "epoch": 0.24593485098896412, "grad_norm": 3.0041706562042236, "learning_rate": 9.077600000000001e-06, "loss": 0.0291, "step": 4613 }, { "epoch": 0.24598816441861704, "grad_norm": 5.32097053527832, "learning_rate": 9.0774e-06, "loss": 0.0395, "step": 4614 }, { "epoch": 0.24604147784827, "grad_norm": 2.7631752490997314, "learning_rate": 9.0772e-06, "loss": 0.0339, "step": 4615 }, { "epoch": 0.24609479127792291, "grad_norm": 5.247844219207764, "learning_rate": 9.077e-06, "loss": 0.0444, "step": 4616 }, { "epoch": 0.24614810470757584, "grad_norm": 6.69490909576416, "learning_rate": 9.076800000000001e-06, "loss": 0.0661, "step": 4617 }, { "epoch": 0.24620141813722876, "grad_norm": 3.3832027912139893, "learning_rate": 9.0766e-06, "loss": 0.0398, "step": 4618 }, { "epoch": 0.24625473156688168, "grad_norm": 1.9311091899871826, "learning_rate": 9.0764e-06, "loss": 0.0161, "step": 4619 }, { "epoch": 0.24630804499653464, "grad_norm": 6.29435396194458, "learning_rate": 9.076200000000001e-06, "loss": 0.0898, "step": 4620 }, { "epoch": 0.24636135842618756, "grad_norm": 3.0908031463623047, "learning_rate": 9.076000000000001e-06, "loss": 0.0631, "step": 4621 }, { "epoch": 0.24641467185584048, "grad_norm": 5.690043926239014, "learning_rate": 9.075800000000002e-06, "loss": 0.0349, "step": 4622 }, { "epoch": 0.2464679852854934, "grad_norm": 3.060553550720215, "learning_rate": 9.075600000000002e-06, "loss": 0.0421, "step": 4623 }, { "epoch": 0.24652129871514636, "grad_norm": 4.750607490539551, "learning_rate": 9.075400000000001e-06, "loss": 0.0733, "step": 4624 }, { "epoch": 0.24657461214479928, "grad_norm": 4.940503120422363, "learning_rate": 9.0752e-06, "loss": 0.0435, "step": 4625 }, { "epoch": 0.2466279255744522, "grad_norm": 2.014859437942505, "learning_rate": 9.075e-06, "loss": 0.0289, "step": 4626 }, { "epoch": 0.24668123900410513, "grad_norm": 4.762179851531982, "learning_rate": 9.0748e-06, "loss": 0.0391, "step": 4627 }, { "epoch": 0.24673455243375805, "grad_norm": 4.491787910461426, "learning_rate": 9.074600000000001e-06, "loss": 0.0658, "step": 4628 }, { "epoch": 0.246787865863411, "grad_norm": 7.070427894592285, "learning_rate": 9.0744e-06, "loss": 0.1113, "step": 4629 }, { "epoch": 0.24684117929306393, "grad_norm": 4.581657886505127, "learning_rate": 9.0742e-06, "loss": 0.0449, "step": 4630 }, { "epoch": 0.24689449272271685, "grad_norm": 5.1856279373168945, "learning_rate": 9.074e-06, "loss": 0.0826, "step": 4631 }, { "epoch": 0.24694780615236978, "grad_norm": 6.647148609161377, "learning_rate": 9.073800000000001e-06, "loss": 0.0537, "step": 4632 }, { "epoch": 0.2470011195820227, "grad_norm": 3.664876699447632, "learning_rate": 9.0736e-06, "loss": 0.038, "step": 4633 }, { "epoch": 0.24705443301167565, "grad_norm": 5.633475303649902, "learning_rate": 9.073400000000002e-06, "loss": 0.0694, "step": 4634 }, { "epoch": 0.24710774644132857, "grad_norm": 3.2813565731048584, "learning_rate": 9.073200000000001e-06, "loss": 0.0697, "step": 4635 }, { "epoch": 0.2471610598709815, "grad_norm": 3.3665926456451416, "learning_rate": 9.073000000000001e-06, "loss": 0.0504, "step": 4636 }, { "epoch": 0.24721437330063442, "grad_norm": 4.346756935119629, "learning_rate": 9.0728e-06, "loss": 0.0498, "step": 4637 }, { "epoch": 0.24726768673028737, "grad_norm": 5.228225231170654, "learning_rate": 9.0726e-06, "loss": 0.0681, "step": 4638 }, { "epoch": 0.2473210001599403, "grad_norm": 4.734785079956055, "learning_rate": 9.072400000000001e-06, "loss": 0.06, "step": 4639 }, { "epoch": 0.24737431358959322, "grad_norm": 3.039179801940918, "learning_rate": 9.072200000000001e-06, "loss": 0.0584, "step": 4640 }, { "epoch": 0.24742762701924614, "grad_norm": 4.414852619171143, "learning_rate": 9.072e-06, "loss": 0.0655, "step": 4641 }, { "epoch": 0.24748094044889907, "grad_norm": 3.359257698059082, "learning_rate": 9.0718e-06, "loss": 0.0401, "step": 4642 }, { "epoch": 0.24753425387855202, "grad_norm": 2.7804861068725586, "learning_rate": 9.0716e-06, "loss": 0.0277, "step": 4643 }, { "epoch": 0.24758756730820494, "grad_norm": 6.517780780792236, "learning_rate": 9.071400000000001e-06, "loss": 0.1233, "step": 4644 }, { "epoch": 0.24764088073785787, "grad_norm": 4.938013076782227, "learning_rate": 9.0712e-06, "loss": 0.1036, "step": 4645 }, { "epoch": 0.2476941941675108, "grad_norm": 2.9706201553344727, "learning_rate": 9.071000000000002e-06, "loss": 0.0646, "step": 4646 }, { "epoch": 0.2477475075971637, "grad_norm": 3.5608816146850586, "learning_rate": 9.070800000000001e-06, "loss": 0.0793, "step": 4647 }, { "epoch": 0.24780082102681666, "grad_norm": 3.371649742126465, "learning_rate": 9.070600000000001e-06, "loss": 0.0394, "step": 4648 }, { "epoch": 0.2478541344564696, "grad_norm": 3.9044973850250244, "learning_rate": 9.0704e-06, "loss": 0.024, "step": 4649 }, { "epoch": 0.2479074478861225, "grad_norm": 4.680046558380127, "learning_rate": 9.070200000000002e-06, "loss": 0.0592, "step": 4650 }, { "epoch": 0.24796076131577544, "grad_norm": 3.622802495956421, "learning_rate": 9.070000000000001e-06, "loss": 0.0516, "step": 4651 }, { "epoch": 0.2480140747454284, "grad_norm": 4.81026029586792, "learning_rate": 9.0698e-06, "loss": 0.0741, "step": 4652 }, { "epoch": 0.2480673881750813, "grad_norm": 3.411871910095215, "learning_rate": 9.0696e-06, "loss": 0.0428, "step": 4653 }, { "epoch": 0.24812070160473423, "grad_norm": 5.936515808105469, "learning_rate": 9.0694e-06, "loss": 0.0673, "step": 4654 }, { "epoch": 0.24817401503438716, "grad_norm": 1.666016936302185, "learning_rate": 9.0692e-06, "loss": 0.0248, "step": 4655 }, { "epoch": 0.24822732846404008, "grad_norm": 3.870640754699707, "learning_rate": 9.069e-06, "loss": 0.0443, "step": 4656 }, { "epoch": 0.24828064189369303, "grad_norm": 2.2603321075439453, "learning_rate": 9.0688e-06, "loss": 0.0372, "step": 4657 }, { "epoch": 0.24833395532334596, "grad_norm": 1.3596065044403076, "learning_rate": 9.068600000000002e-06, "loss": 0.0163, "step": 4658 }, { "epoch": 0.24838726875299888, "grad_norm": 3.8355603218078613, "learning_rate": 9.068400000000001e-06, "loss": 0.0447, "step": 4659 }, { "epoch": 0.2484405821826518, "grad_norm": 4.059698581695557, "learning_rate": 9.0682e-06, "loss": 0.0722, "step": 4660 }, { "epoch": 0.24849389561230473, "grad_norm": 3.806769609451294, "learning_rate": 9.068000000000002e-06, "loss": 0.0437, "step": 4661 }, { "epoch": 0.24854720904195768, "grad_norm": 3.404475212097168, "learning_rate": 9.067800000000002e-06, "loss": 0.0519, "step": 4662 }, { "epoch": 0.2486005224716106, "grad_norm": 2.4022774696350098, "learning_rate": 9.067600000000001e-06, "loss": 0.0339, "step": 4663 }, { "epoch": 0.24865383590126353, "grad_norm": 3.2493367195129395, "learning_rate": 9.0674e-06, "loss": 0.0789, "step": 4664 }, { "epoch": 0.24870714933091645, "grad_norm": 4.000176429748535, "learning_rate": 9.0672e-06, "loss": 0.0694, "step": 4665 }, { "epoch": 0.2487604627605694, "grad_norm": 3.165297031402588, "learning_rate": 9.067e-06, "loss": 0.057, "step": 4666 }, { "epoch": 0.24881377619022232, "grad_norm": 5.509983539581299, "learning_rate": 9.066800000000001e-06, "loss": 0.0478, "step": 4667 }, { "epoch": 0.24886708961987525, "grad_norm": 6.113201141357422, "learning_rate": 9.0666e-06, "loss": 0.048, "step": 4668 }, { "epoch": 0.24892040304952817, "grad_norm": 6.458226203918457, "learning_rate": 9.0664e-06, "loss": 0.0778, "step": 4669 }, { "epoch": 0.2489737164791811, "grad_norm": 2.839329957962036, "learning_rate": 9.066200000000001e-06, "loss": 0.0284, "step": 4670 }, { "epoch": 0.24902702990883405, "grad_norm": 3.439511299133301, "learning_rate": 9.066000000000001e-06, "loss": 0.0528, "step": 4671 }, { "epoch": 0.24908034333848697, "grad_norm": 4.291935443878174, "learning_rate": 9.0658e-06, "loss": 0.0553, "step": 4672 }, { "epoch": 0.2491336567681399, "grad_norm": 2.3336422443389893, "learning_rate": 9.065600000000002e-06, "loss": 0.0224, "step": 4673 }, { "epoch": 0.24918697019779282, "grad_norm": 3.478749990463257, "learning_rate": 9.065400000000001e-06, "loss": 0.0321, "step": 4674 }, { "epoch": 0.24924028362744574, "grad_norm": 4.755202293395996, "learning_rate": 9.065200000000001e-06, "loss": 0.052, "step": 4675 }, { "epoch": 0.2492935970570987, "grad_norm": 4.353216171264648, "learning_rate": 9.065e-06, "loss": 0.0675, "step": 4676 }, { "epoch": 0.24934691048675162, "grad_norm": 2.4301583766937256, "learning_rate": 9.0648e-06, "loss": 0.0335, "step": 4677 }, { "epoch": 0.24940022391640454, "grad_norm": 3.633913993835449, "learning_rate": 9.064600000000001e-06, "loss": 0.0306, "step": 4678 }, { "epoch": 0.24945353734605746, "grad_norm": 4.47554874420166, "learning_rate": 9.0644e-06, "loss": 0.0748, "step": 4679 }, { "epoch": 0.24950685077571041, "grad_norm": 7.955990791320801, "learning_rate": 9.0642e-06, "loss": 0.0935, "step": 4680 }, { "epoch": 0.24956016420536334, "grad_norm": 4.104987621307373, "learning_rate": 9.064e-06, "loss": 0.0495, "step": 4681 }, { "epoch": 0.24961347763501626, "grad_norm": 3.091081142425537, "learning_rate": 9.063800000000001e-06, "loss": 0.0481, "step": 4682 }, { "epoch": 0.24966679106466919, "grad_norm": 3.908034086227417, "learning_rate": 9.0636e-06, "loss": 0.0581, "step": 4683 }, { "epoch": 0.2497201044943221, "grad_norm": 3.2766125202178955, "learning_rate": 9.0634e-06, "loss": 0.0609, "step": 4684 }, { "epoch": 0.24977341792397506, "grad_norm": 4.340527057647705, "learning_rate": 9.063200000000002e-06, "loss": 0.0774, "step": 4685 }, { "epoch": 0.24982673135362798, "grad_norm": 3.273616075515747, "learning_rate": 9.063000000000001e-06, "loss": 0.0229, "step": 4686 }, { "epoch": 0.2498800447832809, "grad_norm": 4.904613494873047, "learning_rate": 9.0628e-06, "loss": 0.0568, "step": 4687 }, { "epoch": 0.24993335821293383, "grad_norm": 3.2099573612213135, "learning_rate": 9.0626e-06, "loss": 0.0665, "step": 4688 }, { "epoch": 0.24998667164258676, "grad_norm": 6.348764419555664, "learning_rate": 9.062400000000002e-06, "loss": 0.08, "step": 4689 }, { "epoch": 0.2500399850722397, "grad_norm": 3.1323046684265137, "learning_rate": 9.062200000000001e-06, "loss": 0.0553, "step": 4690 }, { "epoch": 0.2500932985018926, "grad_norm": 3.341318130493164, "learning_rate": 9.062e-06, "loss": 0.0351, "step": 4691 }, { "epoch": 0.2501466119315456, "grad_norm": 4.165842533111572, "learning_rate": 9.0618e-06, "loss": 0.0622, "step": 4692 }, { "epoch": 0.2501999253611985, "grad_norm": 1.8694562911987305, "learning_rate": 9.0616e-06, "loss": 0.0125, "step": 4693 }, { "epoch": 0.25025323879085143, "grad_norm": 3.5033528804779053, "learning_rate": 9.061400000000001e-06, "loss": 0.0275, "step": 4694 }, { "epoch": 0.25030655222050435, "grad_norm": 4.527076721191406, "learning_rate": 9.0612e-06, "loss": 0.0516, "step": 4695 }, { "epoch": 0.2503598656501573, "grad_norm": 4.532998561859131, "learning_rate": 9.061e-06, "loss": 0.0457, "step": 4696 }, { "epoch": 0.2504131790798102, "grad_norm": 3.093611240386963, "learning_rate": 9.060800000000001e-06, "loss": 0.0501, "step": 4697 }, { "epoch": 0.2504664925094631, "grad_norm": 2.6638450622558594, "learning_rate": 9.060600000000001e-06, "loss": 0.056, "step": 4698 }, { "epoch": 0.25051980593911605, "grad_norm": 6.906111240386963, "learning_rate": 9.0604e-06, "loss": 0.0612, "step": 4699 }, { "epoch": 0.25057311936876897, "grad_norm": 4.463908672332764, "learning_rate": 9.060200000000002e-06, "loss": 0.0619, "step": 4700 }, { "epoch": 0.25062643279842195, "grad_norm": 5.216240882873535, "learning_rate": 9.060000000000001e-06, "loss": 0.069, "step": 4701 }, { "epoch": 0.2506797462280749, "grad_norm": 3.583604574203491, "learning_rate": 9.059800000000001e-06, "loss": 0.0487, "step": 4702 }, { "epoch": 0.2507330596577278, "grad_norm": 3.873690605163574, "learning_rate": 9.0596e-06, "loss": 0.0355, "step": 4703 }, { "epoch": 0.2507863730873807, "grad_norm": 3.6857521533966064, "learning_rate": 9.0594e-06, "loss": 0.0543, "step": 4704 }, { "epoch": 0.25083968651703364, "grad_norm": 3.3881707191467285, "learning_rate": 9.0592e-06, "loss": 0.0383, "step": 4705 }, { "epoch": 0.25089299994668657, "grad_norm": 4.832564830780029, "learning_rate": 9.059000000000001e-06, "loss": 0.064, "step": 4706 }, { "epoch": 0.2509463133763395, "grad_norm": 2.43813419342041, "learning_rate": 9.0588e-06, "loss": 0.0297, "step": 4707 }, { "epoch": 0.2509996268059924, "grad_norm": 2.681684970855713, "learning_rate": 9.0586e-06, "loss": 0.0182, "step": 4708 }, { "epoch": 0.25105294023564534, "grad_norm": 4.166341304779053, "learning_rate": 9.058400000000001e-06, "loss": 0.0589, "step": 4709 }, { "epoch": 0.25110625366529826, "grad_norm": 3.6820144653320312, "learning_rate": 9.0582e-06, "loss": 0.0262, "step": 4710 }, { "epoch": 0.25115956709495124, "grad_norm": 3.3974757194519043, "learning_rate": 9.058000000000002e-06, "loss": 0.0519, "step": 4711 }, { "epoch": 0.25121288052460417, "grad_norm": 4.819483280181885, "learning_rate": 9.057800000000002e-06, "loss": 0.0742, "step": 4712 }, { "epoch": 0.2512661939542571, "grad_norm": 4.13148307800293, "learning_rate": 9.057600000000001e-06, "loss": 0.0542, "step": 4713 }, { "epoch": 0.25131950738391, "grad_norm": 3.8971683979034424, "learning_rate": 9.0574e-06, "loss": 0.0659, "step": 4714 }, { "epoch": 0.25137282081356294, "grad_norm": 4.5911760330200195, "learning_rate": 9.0572e-06, "loss": 0.0546, "step": 4715 }, { "epoch": 0.25142613424321586, "grad_norm": 4.525530815124512, "learning_rate": 9.057e-06, "loss": 0.0821, "step": 4716 }, { "epoch": 0.2514794476728688, "grad_norm": 5.5381855964660645, "learning_rate": 9.056800000000001e-06, "loss": 0.0625, "step": 4717 }, { "epoch": 0.2515327611025217, "grad_norm": 6.878273963928223, "learning_rate": 9.0566e-06, "loss": 0.0726, "step": 4718 }, { "epoch": 0.25158607453217463, "grad_norm": 5.9802937507629395, "learning_rate": 9.0564e-06, "loss": 0.0634, "step": 4719 }, { "epoch": 0.2516393879618276, "grad_norm": 5.59482479095459, "learning_rate": 9.0562e-06, "loss": 0.0168, "step": 4720 }, { "epoch": 0.25169270139148053, "grad_norm": 1.6413509845733643, "learning_rate": 9.056000000000001e-06, "loss": 0.0263, "step": 4721 }, { "epoch": 0.25174601482113346, "grad_norm": 5.620181560516357, "learning_rate": 9.0558e-06, "loss": 0.0633, "step": 4722 }, { "epoch": 0.2517993282507864, "grad_norm": 4.255200386047363, "learning_rate": 9.055600000000002e-06, "loss": 0.0557, "step": 4723 }, { "epoch": 0.2518526416804393, "grad_norm": 3.8836491107940674, "learning_rate": 9.055400000000001e-06, "loss": 0.0628, "step": 4724 }, { "epoch": 0.25190595511009223, "grad_norm": 3.2431557178497314, "learning_rate": 9.055200000000001e-06, "loss": 0.0331, "step": 4725 }, { "epoch": 0.25195926853974515, "grad_norm": 3.169649600982666, "learning_rate": 9.055e-06, "loss": 0.0497, "step": 4726 }, { "epoch": 0.2520125819693981, "grad_norm": 1.894792079925537, "learning_rate": 9.0548e-06, "loss": 0.0201, "step": 4727 }, { "epoch": 0.252065895399051, "grad_norm": 2.3855416774749756, "learning_rate": 9.054600000000001e-06, "loss": 0.0265, "step": 4728 }, { "epoch": 0.252119208828704, "grad_norm": 5.816556930541992, "learning_rate": 9.054400000000001e-06, "loss": 0.0814, "step": 4729 }, { "epoch": 0.2521725222583569, "grad_norm": 3.985375165939331, "learning_rate": 9.0542e-06, "loss": 0.0968, "step": 4730 }, { "epoch": 0.2522258356880098, "grad_norm": 8.18417739868164, "learning_rate": 9.054e-06, "loss": 0.0646, "step": 4731 }, { "epoch": 0.25227914911766275, "grad_norm": 7.435092449188232, "learning_rate": 9.0538e-06, "loss": 0.0886, "step": 4732 }, { "epoch": 0.2523324625473157, "grad_norm": 4.578146457672119, "learning_rate": 9.053600000000001e-06, "loss": 0.041, "step": 4733 }, { "epoch": 0.2523857759769686, "grad_norm": 3.6674675941467285, "learning_rate": 9.0534e-06, "loss": 0.0321, "step": 4734 }, { "epoch": 0.2524390894066215, "grad_norm": 3.0222890377044678, "learning_rate": 9.053200000000002e-06, "loss": 0.0585, "step": 4735 }, { "epoch": 0.25249240283627444, "grad_norm": 4.389321327209473, "learning_rate": 9.053000000000001e-06, "loss": 0.079, "step": 4736 }, { "epoch": 0.25254571626592737, "grad_norm": 3.202758550643921, "learning_rate": 9.0528e-06, "loss": 0.0322, "step": 4737 }, { "epoch": 0.2525990296955803, "grad_norm": 4.746048450469971, "learning_rate": 9.0526e-06, "loss": 0.0798, "step": 4738 }, { "epoch": 0.25265234312523327, "grad_norm": 4.900179862976074, "learning_rate": 9.052400000000002e-06, "loss": 0.0722, "step": 4739 }, { "epoch": 0.2527056565548862, "grad_norm": 4.699301719665527, "learning_rate": 9.052200000000001e-06, "loss": 0.0256, "step": 4740 }, { "epoch": 0.2527589699845391, "grad_norm": 2.3178915977478027, "learning_rate": 9.052e-06, "loss": 0.0467, "step": 4741 }, { "epoch": 0.25281228341419204, "grad_norm": 3.313344955444336, "learning_rate": 9.0518e-06, "loss": 0.0439, "step": 4742 }, { "epoch": 0.25286559684384496, "grad_norm": 3.1357500553131104, "learning_rate": 9.0516e-06, "loss": 0.0457, "step": 4743 }, { "epoch": 0.2529189102734979, "grad_norm": 2.820446014404297, "learning_rate": 9.0514e-06, "loss": 0.0524, "step": 4744 }, { "epoch": 0.2529722237031508, "grad_norm": 4.819199085235596, "learning_rate": 9.0512e-06, "loss": 0.0521, "step": 4745 }, { "epoch": 0.25302553713280374, "grad_norm": 2.0266525745391846, "learning_rate": 9.051e-06, "loss": 0.0344, "step": 4746 }, { "epoch": 0.25307885056245666, "grad_norm": 3.005028247833252, "learning_rate": 9.050800000000002e-06, "loss": 0.0227, "step": 4747 }, { "epoch": 0.25313216399210964, "grad_norm": 2.3548455238342285, "learning_rate": 9.050600000000001e-06, "loss": 0.0332, "step": 4748 }, { "epoch": 0.25318547742176256, "grad_norm": 2.006972074508667, "learning_rate": 9.0504e-06, "loss": 0.0325, "step": 4749 }, { "epoch": 0.2532387908514155, "grad_norm": 2.671659231185913, "learning_rate": 9.050200000000002e-06, "loss": 0.0508, "step": 4750 }, { "epoch": 0.2532921042810684, "grad_norm": 2.502986431121826, "learning_rate": 9.050000000000001e-06, "loss": 0.0632, "step": 4751 }, { "epoch": 0.25334541771072133, "grad_norm": 4.146180152893066, "learning_rate": 9.049800000000001e-06, "loss": 0.056, "step": 4752 }, { "epoch": 0.25339873114037426, "grad_norm": 4.994761943817139, "learning_rate": 9.0496e-06, "loss": 0.0782, "step": 4753 }, { "epoch": 0.2534520445700272, "grad_norm": 4.804893493652344, "learning_rate": 9.0494e-06, "loss": 0.0298, "step": 4754 }, { "epoch": 0.2535053579996801, "grad_norm": 3.1048572063446045, "learning_rate": 9.0492e-06, "loss": 0.038, "step": 4755 }, { "epoch": 0.253558671429333, "grad_norm": 6.46009635925293, "learning_rate": 9.049000000000001e-06, "loss": 0.062, "step": 4756 }, { "epoch": 0.253611984858986, "grad_norm": 8.612924575805664, "learning_rate": 9.0488e-06, "loss": 0.0546, "step": 4757 }, { "epoch": 0.25366529828863893, "grad_norm": 4.859718322753906, "learning_rate": 9.0486e-06, "loss": 0.0346, "step": 4758 }, { "epoch": 0.25371861171829185, "grad_norm": 3.9445135593414307, "learning_rate": 9.048400000000001e-06, "loss": 0.0572, "step": 4759 }, { "epoch": 0.2537719251479448, "grad_norm": 3.360989570617676, "learning_rate": 9.048200000000001e-06, "loss": 0.0407, "step": 4760 }, { "epoch": 0.2538252385775977, "grad_norm": 1.350484848022461, "learning_rate": 9.048e-06, "loss": 0.0111, "step": 4761 }, { "epoch": 0.2538785520072506, "grad_norm": 4.9351325035095215, "learning_rate": 9.047800000000002e-06, "loss": 0.0535, "step": 4762 }, { "epoch": 0.25393186543690355, "grad_norm": 5.10786247253418, "learning_rate": 9.047600000000001e-06, "loss": 0.0407, "step": 4763 }, { "epoch": 0.25398517886655647, "grad_norm": 3.325815200805664, "learning_rate": 9.0474e-06, "loss": 0.0322, "step": 4764 }, { "epoch": 0.2540384922962094, "grad_norm": 5.112375259399414, "learning_rate": 9.0472e-06, "loss": 0.0549, "step": 4765 }, { "epoch": 0.2540918057258623, "grad_norm": 6.33311128616333, "learning_rate": 9.047e-06, "loss": 0.1096, "step": 4766 }, { "epoch": 0.2541451191555153, "grad_norm": 2.7815675735473633, "learning_rate": 9.046800000000001e-06, "loss": 0.0255, "step": 4767 }, { "epoch": 0.2541984325851682, "grad_norm": 6.250075340270996, "learning_rate": 9.0466e-06, "loss": 0.0581, "step": 4768 }, { "epoch": 0.25425174601482115, "grad_norm": 6.170403003692627, "learning_rate": 9.0464e-06, "loss": 0.0618, "step": 4769 }, { "epoch": 0.25430505944447407, "grad_norm": 5.50147819519043, "learning_rate": 9.0462e-06, "loss": 0.0457, "step": 4770 }, { "epoch": 0.254358372874127, "grad_norm": 8.102713584899902, "learning_rate": 9.046000000000001e-06, "loss": 0.0502, "step": 4771 }, { "epoch": 0.2544116863037799, "grad_norm": 4.3526458740234375, "learning_rate": 9.0458e-06, "loss": 0.0505, "step": 4772 }, { "epoch": 0.25446499973343284, "grad_norm": 4.758033752441406, "learning_rate": 9.0456e-06, "loss": 0.043, "step": 4773 }, { "epoch": 0.25451831316308576, "grad_norm": 2.4074103832244873, "learning_rate": 9.045400000000002e-06, "loss": 0.0293, "step": 4774 }, { "epoch": 0.2545716265927387, "grad_norm": 5.763688564300537, "learning_rate": 9.045200000000001e-06, "loss": 0.0538, "step": 4775 }, { "epoch": 0.25462494002239167, "grad_norm": 4.884819984436035, "learning_rate": 9.045e-06, "loss": 0.0716, "step": 4776 }, { "epoch": 0.2546782534520446, "grad_norm": 6.059475898742676, "learning_rate": 9.0448e-06, "loss": 0.0551, "step": 4777 }, { "epoch": 0.2547315668816975, "grad_norm": 3.3567850589752197, "learning_rate": 9.044600000000001e-06, "loss": 0.0376, "step": 4778 }, { "epoch": 0.25478488031135044, "grad_norm": 3.4629364013671875, "learning_rate": 9.044400000000001e-06, "loss": 0.0368, "step": 4779 }, { "epoch": 0.25483819374100336, "grad_norm": 3.4746880531311035, "learning_rate": 9.0442e-06, "loss": 0.0704, "step": 4780 }, { "epoch": 0.2548915071706563, "grad_norm": 4.96652364730835, "learning_rate": 9.044e-06, "loss": 0.0312, "step": 4781 }, { "epoch": 0.2549448206003092, "grad_norm": 3.6525967121124268, "learning_rate": 9.0438e-06, "loss": 0.0422, "step": 4782 }, { "epoch": 0.25499813402996213, "grad_norm": 3.0312235355377197, "learning_rate": 9.043600000000001e-06, "loss": 0.0442, "step": 4783 }, { "epoch": 0.25505144745961505, "grad_norm": 4.032959461212158, "learning_rate": 9.0434e-06, "loss": 0.0562, "step": 4784 }, { "epoch": 0.25510476088926803, "grad_norm": 2.3535449504852295, "learning_rate": 9.0432e-06, "loss": 0.0216, "step": 4785 }, { "epoch": 0.25515807431892096, "grad_norm": 4.158348083496094, "learning_rate": 9.043000000000001e-06, "loss": 0.0472, "step": 4786 }, { "epoch": 0.2552113877485739, "grad_norm": 2.848188638687134, "learning_rate": 9.042800000000001e-06, "loss": 0.0614, "step": 4787 }, { "epoch": 0.2552647011782268, "grad_norm": 3.887661933898926, "learning_rate": 9.0426e-06, "loss": 0.0391, "step": 4788 }, { "epoch": 0.25531801460787973, "grad_norm": 4.487307548522949, "learning_rate": 9.042400000000002e-06, "loss": 0.0505, "step": 4789 }, { "epoch": 0.25537132803753265, "grad_norm": 2.9148828983306885, "learning_rate": 9.042200000000001e-06, "loss": 0.0388, "step": 4790 }, { "epoch": 0.2554246414671856, "grad_norm": 3.1242895126342773, "learning_rate": 9.042e-06, "loss": 0.0451, "step": 4791 }, { "epoch": 0.2554779548968385, "grad_norm": 3.0785367488861084, "learning_rate": 9.0418e-06, "loss": 0.0488, "step": 4792 }, { "epoch": 0.2555312683264914, "grad_norm": 2.3803911209106445, "learning_rate": 9.0416e-06, "loss": 0.0296, "step": 4793 }, { "epoch": 0.25558458175614435, "grad_norm": 7.655568599700928, "learning_rate": 9.041400000000001e-06, "loss": 0.0772, "step": 4794 }, { "epoch": 0.2556378951857973, "grad_norm": 3.260114908218384, "learning_rate": 9.0412e-06, "loss": 0.0315, "step": 4795 }, { "epoch": 0.25569120861545025, "grad_norm": 3.4608333110809326, "learning_rate": 9.041e-06, "loss": 0.0485, "step": 4796 }, { "epoch": 0.2557445220451032, "grad_norm": 2.3366925716400146, "learning_rate": 9.0408e-06, "loss": 0.0319, "step": 4797 }, { "epoch": 0.2557978354747561, "grad_norm": 3.3864381313323975, "learning_rate": 9.040600000000001e-06, "loss": 0.0536, "step": 4798 }, { "epoch": 0.255851148904409, "grad_norm": 4.9391703605651855, "learning_rate": 9.0404e-06, "loss": 0.0691, "step": 4799 }, { "epoch": 0.25590446233406194, "grad_norm": 2.4743027687072754, "learning_rate": 9.040200000000002e-06, "loss": 0.0267, "step": 4800 }, { "epoch": 0.25595777576371487, "grad_norm": 3.632333993911743, "learning_rate": 9.040000000000002e-06, "loss": 0.0526, "step": 4801 }, { "epoch": 0.2560110891933678, "grad_norm": 3.3920412063598633, "learning_rate": 9.039800000000001e-06, "loss": 0.0352, "step": 4802 }, { "epoch": 0.2560644026230207, "grad_norm": 2.337094306945801, "learning_rate": 9.0396e-06, "loss": 0.0286, "step": 4803 }, { "epoch": 0.2561177160526737, "grad_norm": 5.508491039276123, "learning_rate": 9.0394e-06, "loss": 0.0692, "step": 4804 }, { "epoch": 0.2561710294823266, "grad_norm": 3.995781183242798, "learning_rate": 9.0392e-06, "loss": 0.063, "step": 4805 }, { "epoch": 0.25622434291197954, "grad_norm": 4.8468499183654785, "learning_rate": 9.039000000000001e-06, "loss": 0.0862, "step": 4806 }, { "epoch": 0.25627765634163246, "grad_norm": 3.2690296173095703, "learning_rate": 9.0388e-06, "loss": 0.0557, "step": 4807 }, { "epoch": 0.2563309697712854, "grad_norm": 2.1335856914520264, "learning_rate": 9.0386e-06, "loss": 0.031, "step": 4808 }, { "epoch": 0.2563842832009383, "grad_norm": 2.7456159591674805, "learning_rate": 9.0384e-06, "loss": 0.0359, "step": 4809 }, { "epoch": 0.25643759663059124, "grad_norm": 3.654106855392456, "learning_rate": 9.038200000000001e-06, "loss": 0.0444, "step": 4810 }, { "epoch": 0.25649091006024416, "grad_norm": 3.4882853031158447, "learning_rate": 9.038e-06, "loss": 0.0547, "step": 4811 }, { "epoch": 0.2565442234898971, "grad_norm": 7.178729057312012, "learning_rate": 9.037800000000002e-06, "loss": 0.1246, "step": 4812 }, { "epoch": 0.25659753691955006, "grad_norm": 2.0475502014160156, "learning_rate": 9.037600000000001e-06, "loss": 0.0244, "step": 4813 }, { "epoch": 0.256650850349203, "grad_norm": 2.6763784885406494, "learning_rate": 9.037400000000001e-06, "loss": 0.0436, "step": 4814 }, { "epoch": 0.2567041637788559, "grad_norm": 6.488926887512207, "learning_rate": 9.0372e-06, "loss": 0.0785, "step": 4815 }, { "epoch": 0.25675747720850883, "grad_norm": 8.184958457946777, "learning_rate": 9.037e-06, "loss": 0.0758, "step": 4816 }, { "epoch": 0.25681079063816176, "grad_norm": 4.872811794281006, "learning_rate": 9.036800000000001e-06, "loss": 0.0925, "step": 4817 }, { "epoch": 0.2568641040678147, "grad_norm": 9.034634590148926, "learning_rate": 9.036600000000001e-06, "loss": 0.1006, "step": 4818 }, { "epoch": 0.2569174174974676, "grad_norm": 2.6061408519744873, "learning_rate": 9.0364e-06, "loss": 0.0334, "step": 4819 }, { "epoch": 0.2569707309271205, "grad_norm": 3.9324285984039307, "learning_rate": 9.0362e-06, "loss": 0.0754, "step": 4820 }, { "epoch": 0.25702404435677345, "grad_norm": 6.044460773468018, "learning_rate": 9.036e-06, "loss": 0.0522, "step": 4821 }, { "epoch": 0.2570773577864264, "grad_norm": 7.6061482429504395, "learning_rate": 9.0358e-06, "loss": 0.0704, "step": 4822 }, { "epoch": 0.25713067121607935, "grad_norm": 5.015057563781738, "learning_rate": 9.0356e-06, "loss": 0.0574, "step": 4823 }, { "epoch": 0.2571839846457323, "grad_norm": 6.308739185333252, "learning_rate": 9.035400000000002e-06, "loss": 0.0764, "step": 4824 }, { "epoch": 0.2572372980753852, "grad_norm": 3.4447178840637207, "learning_rate": 9.035200000000001e-06, "loss": 0.0399, "step": 4825 }, { "epoch": 0.2572906115050381, "grad_norm": 4.9746856689453125, "learning_rate": 9.035e-06, "loss": 0.0835, "step": 4826 }, { "epoch": 0.25734392493469105, "grad_norm": 7.484307765960693, "learning_rate": 9.0348e-06, "loss": 0.0728, "step": 4827 }, { "epoch": 0.25739723836434397, "grad_norm": 12.221722602844238, "learning_rate": 9.034600000000002e-06, "loss": 0.1099, "step": 4828 }, { "epoch": 0.2574505517939969, "grad_norm": 5.0273661613464355, "learning_rate": 9.034400000000001e-06, "loss": 0.0618, "step": 4829 }, { "epoch": 0.2575038652236498, "grad_norm": 2.9631481170654297, "learning_rate": 9.0342e-06, "loss": 0.0325, "step": 4830 }, { "epoch": 0.25755717865330274, "grad_norm": 3.1687018871307373, "learning_rate": 9.034e-06, "loss": 0.0345, "step": 4831 }, { "epoch": 0.2576104920829557, "grad_norm": 5.397350311279297, "learning_rate": 9.0338e-06, "loss": 0.0649, "step": 4832 }, { "epoch": 0.25766380551260865, "grad_norm": 4.461808681488037, "learning_rate": 9.033600000000001e-06, "loss": 0.0515, "step": 4833 }, { "epoch": 0.25771711894226157, "grad_norm": 4.928049564361572, "learning_rate": 9.0334e-06, "loss": 0.0603, "step": 4834 }, { "epoch": 0.2577704323719145, "grad_norm": 3.259634494781494, "learning_rate": 9.0332e-06, "loss": 0.0378, "step": 4835 }, { "epoch": 0.2578237458015674, "grad_norm": 2.4723305702209473, "learning_rate": 9.033000000000001e-06, "loss": 0.0265, "step": 4836 }, { "epoch": 0.25787705923122034, "grad_norm": 3.7423057556152344, "learning_rate": 9.032800000000001e-06, "loss": 0.0623, "step": 4837 }, { "epoch": 0.25793037266087326, "grad_norm": 5.2584733963012695, "learning_rate": 9.0326e-06, "loss": 0.0318, "step": 4838 }, { "epoch": 0.2579836860905262, "grad_norm": 4.68380069732666, "learning_rate": 9.032400000000002e-06, "loss": 0.0434, "step": 4839 }, { "epoch": 0.2580369995201791, "grad_norm": 4.504749298095703, "learning_rate": 9.032200000000001e-06, "loss": 0.051, "step": 4840 }, { "epoch": 0.2580903129498321, "grad_norm": 2.790586233139038, "learning_rate": 9.032000000000001e-06, "loss": 0.0672, "step": 4841 }, { "epoch": 0.258143626379485, "grad_norm": 3.455080509185791, "learning_rate": 9.0318e-06, "loss": 0.0475, "step": 4842 }, { "epoch": 0.25819693980913794, "grad_norm": 4.6004638671875, "learning_rate": 9.0316e-06, "loss": 0.0307, "step": 4843 }, { "epoch": 0.25825025323879086, "grad_norm": 3.564694404602051, "learning_rate": 9.031400000000001e-06, "loss": 0.0696, "step": 4844 }, { "epoch": 0.2583035666684438, "grad_norm": 4.532865524291992, "learning_rate": 9.031200000000001e-06, "loss": 0.0495, "step": 4845 }, { "epoch": 0.2583568800980967, "grad_norm": 4.911703586578369, "learning_rate": 9.031e-06, "loss": 0.0463, "step": 4846 }, { "epoch": 0.25841019352774963, "grad_norm": 4.125251293182373, "learning_rate": 9.0308e-06, "loss": 0.0607, "step": 4847 }, { "epoch": 0.25846350695740256, "grad_norm": 5.77139139175415, "learning_rate": 9.030600000000001e-06, "loss": 0.0322, "step": 4848 }, { "epoch": 0.2585168203870555, "grad_norm": 4.791807174682617, "learning_rate": 9.0304e-06, "loss": 0.0358, "step": 4849 }, { "epoch": 0.2585701338167084, "grad_norm": 3.3484230041503906, "learning_rate": 9.0302e-06, "loss": 0.0359, "step": 4850 }, { "epoch": 0.2586234472463614, "grad_norm": 4.897939205169678, "learning_rate": 9.030000000000002e-06, "loss": 0.0561, "step": 4851 }, { "epoch": 0.2586767606760143, "grad_norm": 3.5861716270446777, "learning_rate": 9.029800000000001e-06, "loss": 0.0373, "step": 4852 }, { "epoch": 0.25873007410566723, "grad_norm": 6.314844131469727, "learning_rate": 9.0296e-06, "loss": 0.0611, "step": 4853 }, { "epoch": 0.25878338753532015, "grad_norm": 1.6835414171218872, "learning_rate": 9.0294e-06, "loss": 0.0207, "step": 4854 }, { "epoch": 0.2588367009649731, "grad_norm": 4.918824672698975, "learning_rate": 9.0292e-06, "loss": 0.0566, "step": 4855 }, { "epoch": 0.258890014394626, "grad_norm": 3.8764030933380127, "learning_rate": 9.029000000000001e-06, "loss": 0.0493, "step": 4856 }, { "epoch": 0.2589433278242789, "grad_norm": 4.486584186553955, "learning_rate": 9.0288e-06, "loss": 0.059, "step": 4857 }, { "epoch": 0.25899664125393185, "grad_norm": 6.642289161682129, "learning_rate": 9.0286e-06, "loss": 0.0992, "step": 4858 }, { "epoch": 0.25904995468358477, "grad_norm": 5.000884532928467, "learning_rate": 9.0284e-06, "loss": 0.0314, "step": 4859 }, { "epoch": 0.25910326811323775, "grad_norm": 3.0171539783477783, "learning_rate": 9.028200000000001e-06, "loss": 0.0258, "step": 4860 }, { "epoch": 0.2591565815428907, "grad_norm": 5.867378234863281, "learning_rate": 9.028e-06, "loss": 0.0765, "step": 4861 }, { "epoch": 0.2592098949725436, "grad_norm": 2.9077770709991455, "learning_rate": 9.0278e-06, "loss": 0.0314, "step": 4862 }, { "epoch": 0.2592632084021965, "grad_norm": 2.415680408477783, "learning_rate": 9.027600000000001e-06, "loss": 0.0215, "step": 4863 }, { "epoch": 0.25931652183184944, "grad_norm": 4.148387908935547, "learning_rate": 9.027400000000001e-06, "loss": 0.062, "step": 4864 }, { "epoch": 0.25936983526150237, "grad_norm": 3.0485305786132812, "learning_rate": 9.0272e-06, "loss": 0.0666, "step": 4865 }, { "epoch": 0.2594231486911553, "grad_norm": 3.483219861984253, "learning_rate": 9.027e-06, "loss": 0.0503, "step": 4866 }, { "epoch": 0.2594764621208082, "grad_norm": 4.177730560302734, "learning_rate": 9.026800000000001e-06, "loss": 0.0335, "step": 4867 }, { "epoch": 0.25952977555046114, "grad_norm": 6.34296989440918, "learning_rate": 9.026600000000001e-06, "loss": 0.0745, "step": 4868 }, { "epoch": 0.2595830889801141, "grad_norm": 2.49369740486145, "learning_rate": 9.0264e-06, "loss": 0.0598, "step": 4869 }, { "epoch": 0.25963640240976704, "grad_norm": 4.209351062774658, "learning_rate": 9.0262e-06, "loss": 0.064, "step": 4870 }, { "epoch": 0.25968971583941997, "grad_norm": 3.0283899307250977, "learning_rate": 9.026e-06, "loss": 0.0552, "step": 4871 }, { "epoch": 0.2597430292690729, "grad_norm": 3.089423418045044, "learning_rate": 9.025800000000001e-06, "loss": 0.0598, "step": 4872 }, { "epoch": 0.2597963426987258, "grad_norm": 3.299717664718628, "learning_rate": 9.0256e-06, "loss": 0.0282, "step": 4873 }, { "epoch": 0.25984965612837874, "grad_norm": 5.2852253913879395, "learning_rate": 9.0254e-06, "loss": 0.0439, "step": 4874 }, { "epoch": 0.25990296955803166, "grad_norm": 3.976658821105957, "learning_rate": 9.025200000000001e-06, "loss": 0.0537, "step": 4875 }, { "epoch": 0.2599562829876846, "grad_norm": 2.853942632675171, "learning_rate": 9.025e-06, "loss": 0.0475, "step": 4876 }, { "epoch": 0.2600095964173375, "grad_norm": 5.313129425048828, "learning_rate": 9.0248e-06, "loss": 0.059, "step": 4877 }, { "epoch": 0.26006290984699043, "grad_norm": 3.4835102558135986, "learning_rate": 9.024600000000002e-06, "loss": 0.0349, "step": 4878 }, { "epoch": 0.2601162232766434, "grad_norm": 4.6567769050598145, "learning_rate": 9.024400000000001e-06, "loss": 0.0508, "step": 4879 }, { "epoch": 0.26016953670629633, "grad_norm": 11.93474292755127, "learning_rate": 9.0242e-06, "loss": 0.0913, "step": 4880 }, { "epoch": 0.26022285013594926, "grad_norm": 7.516478538513184, "learning_rate": 9.024e-06, "loss": 0.0595, "step": 4881 }, { "epoch": 0.2602761635656022, "grad_norm": 4.862811088562012, "learning_rate": 9.0238e-06, "loss": 0.0557, "step": 4882 }, { "epoch": 0.2603294769952551, "grad_norm": 3.4004135131835938, "learning_rate": 9.023600000000001e-06, "loss": 0.0748, "step": 4883 }, { "epoch": 0.26038279042490803, "grad_norm": 5.53292989730835, "learning_rate": 9.0234e-06, "loss": 0.0698, "step": 4884 }, { "epoch": 0.26043610385456095, "grad_norm": 7.418461799621582, "learning_rate": 9.0232e-06, "loss": 0.1064, "step": 4885 }, { "epoch": 0.2604894172842139, "grad_norm": 5.25362491607666, "learning_rate": 9.023e-06, "loss": 0.0541, "step": 4886 }, { "epoch": 0.2605427307138668, "grad_norm": 3.7488956451416016, "learning_rate": 9.022800000000001e-06, "loss": 0.0406, "step": 4887 }, { "epoch": 0.2605960441435198, "grad_norm": 7.149594306945801, "learning_rate": 9.0226e-06, "loss": 0.0934, "step": 4888 }, { "epoch": 0.2606493575731727, "grad_norm": 6.305069923400879, "learning_rate": 9.022400000000002e-06, "loss": 0.0475, "step": 4889 }, { "epoch": 0.2607026710028256, "grad_norm": 5.1041107177734375, "learning_rate": 9.022200000000001e-06, "loss": 0.045, "step": 4890 }, { "epoch": 0.26075598443247855, "grad_norm": 3.7662012577056885, "learning_rate": 9.022000000000001e-06, "loss": 0.0587, "step": 4891 }, { "epoch": 0.2608092978621315, "grad_norm": 3.0267183780670166, "learning_rate": 9.0218e-06, "loss": 0.0472, "step": 4892 }, { "epoch": 0.2608626112917844, "grad_norm": 3.3888845443725586, "learning_rate": 9.0216e-06, "loss": 0.0374, "step": 4893 }, { "epoch": 0.2609159247214373, "grad_norm": 3.705705165863037, "learning_rate": 9.021400000000001e-06, "loss": 0.0419, "step": 4894 }, { "epoch": 0.26096923815109024, "grad_norm": 3.59771466255188, "learning_rate": 9.021200000000001e-06, "loss": 0.0418, "step": 4895 }, { "epoch": 0.26102255158074317, "grad_norm": 4.451777935028076, "learning_rate": 9.021e-06, "loss": 0.0683, "step": 4896 }, { "epoch": 0.26107586501039615, "grad_norm": 2.831284761428833, "learning_rate": 9.0208e-06, "loss": 0.0404, "step": 4897 }, { "epoch": 0.26112917844004907, "grad_norm": 3.5664570331573486, "learning_rate": 9.0206e-06, "loss": 0.0513, "step": 4898 }, { "epoch": 0.261182491869702, "grad_norm": 5.555919170379639, "learning_rate": 9.020400000000001e-06, "loss": 0.0707, "step": 4899 }, { "epoch": 0.2612358052993549, "grad_norm": 2.86521053314209, "learning_rate": 9.0202e-06, "loss": 0.0228, "step": 4900 }, { "epoch": 0.26128911872900784, "grad_norm": 3.2419981956481934, "learning_rate": 9.020000000000002e-06, "loss": 0.0612, "step": 4901 }, { "epoch": 0.26134243215866076, "grad_norm": 3.2415897846221924, "learning_rate": 9.019800000000001e-06, "loss": 0.0703, "step": 4902 }, { "epoch": 0.2613957455883137, "grad_norm": 3.9610910415649414, "learning_rate": 9.019600000000001e-06, "loss": 0.0675, "step": 4903 }, { "epoch": 0.2614490590179666, "grad_norm": 5.4804911613464355, "learning_rate": 9.0194e-06, "loss": 0.0728, "step": 4904 }, { "epoch": 0.26150237244761954, "grad_norm": 4.531431674957275, "learning_rate": 9.019200000000002e-06, "loss": 0.0225, "step": 4905 }, { "epoch": 0.26155568587727246, "grad_norm": 9.567025184631348, "learning_rate": 9.019000000000001e-06, "loss": 0.0987, "step": 4906 }, { "epoch": 0.26160899930692544, "grad_norm": 4.61315393447876, "learning_rate": 9.0188e-06, "loss": 0.0334, "step": 4907 }, { "epoch": 0.26166231273657836, "grad_norm": 5.466941833496094, "learning_rate": 9.0186e-06, "loss": 0.0878, "step": 4908 }, { "epoch": 0.2617156261662313, "grad_norm": 5.429835796356201, "learning_rate": 9.0184e-06, "loss": 0.0564, "step": 4909 }, { "epoch": 0.2617689395958842, "grad_norm": 4.297543525695801, "learning_rate": 9.0182e-06, "loss": 0.0903, "step": 4910 }, { "epoch": 0.26182225302553713, "grad_norm": 5.714200019836426, "learning_rate": 9.018e-06, "loss": 0.0764, "step": 4911 }, { "epoch": 0.26187556645519006, "grad_norm": 2.1972994804382324, "learning_rate": 9.0178e-06, "loss": 0.0208, "step": 4912 }, { "epoch": 0.261928879884843, "grad_norm": 3.1642794609069824, "learning_rate": 9.017600000000002e-06, "loss": 0.0534, "step": 4913 }, { "epoch": 0.2619821933144959, "grad_norm": 3.3743109703063965, "learning_rate": 9.017400000000001e-06, "loss": 0.0379, "step": 4914 }, { "epoch": 0.2620355067441488, "grad_norm": 4.159863471984863, "learning_rate": 9.0172e-06, "loss": 0.0544, "step": 4915 }, { "epoch": 0.2620888201738018, "grad_norm": 5.518507480621338, "learning_rate": 9.017e-06, "loss": 0.0646, "step": 4916 }, { "epoch": 0.26214213360345473, "grad_norm": 4.801395416259766, "learning_rate": 9.016800000000001e-06, "loss": 0.0348, "step": 4917 }, { "epoch": 0.26219544703310765, "grad_norm": 2.310901403427124, "learning_rate": 9.016600000000001e-06, "loss": 0.0345, "step": 4918 }, { "epoch": 0.2622487604627606, "grad_norm": 2.2336537837982178, "learning_rate": 9.0164e-06, "loss": 0.0301, "step": 4919 }, { "epoch": 0.2623020738924135, "grad_norm": 3.537200689315796, "learning_rate": 9.0162e-06, "loss": 0.0402, "step": 4920 }, { "epoch": 0.2623553873220664, "grad_norm": 3.6646523475646973, "learning_rate": 9.016e-06, "loss": 0.095, "step": 4921 }, { "epoch": 0.26240870075171935, "grad_norm": 3.2425808906555176, "learning_rate": 9.015800000000001e-06, "loss": 0.0494, "step": 4922 }, { "epoch": 0.26246201418137227, "grad_norm": 2.415858030319214, "learning_rate": 9.0156e-06, "loss": 0.0482, "step": 4923 }, { "epoch": 0.2625153276110252, "grad_norm": 2.140718698501587, "learning_rate": 9.0154e-06, "loss": 0.0297, "step": 4924 }, { "epoch": 0.2625686410406782, "grad_norm": 2.8745076656341553, "learning_rate": 9.015200000000001e-06, "loss": 0.056, "step": 4925 }, { "epoch": 0.2626219544703311, "grad_norm": 2.400200128555298, "learning_rate": 9.015000000000001e-06, "loss": 0.0273, "step": 4926 }, { "epoch": 0.262675267899984, "grad_norm": 2.383547782897949, "learning_rate": 9.0148e-06, "loss": 0.0207, "step": 4927 }, { "epoch": 0.26272858132963695, "grad_norm": 1.9540458917617798, "learning_rate": 9.014600000000002e-06, "loss": 0.0198, "step": 4928 }, { "epoch": 0.26278189475928987, "grad_norm": 3.346087694168091, "learning_rate": 9.014400000000001e-06, "loss": 0.0534, "step": 4929 }, { "epoch": 0.2628352081889428, "grad_norm": 2.203650712966919, "learning_rate": 9.014200000000001e-06, "loss": 0.0339, "step": 4930 }, { "epoch": 0.2628885216185957, "grad_norm": 3.5558407306671143, "learning_rate": 9.014e-06, "loss": 0.0325, "step": 4931 }, { "epoch": 0.26294183504824864, "grad_norm": 3.664766550064087, "learning_rate": 9.0138e-06, "loss": 0.0554, "step": 4932 }, { "epoch": 0.26299514847790156, "grad_norm": 4.684883117675781, "learning_rate": 9.013600000000001e-06, "loss": 0.0694, "step": 4933 }, { "epoch": 0.2630484619075545, "grad_norm": 3.3522729873657227, "learning_rate": 9.0134e-06, "loss": 0.053, "step": 4934 }, { "epoch": 0.26310177533720747, "grad_norm": 2.740079641342163, "learning_rate": 9.0132e-06, "loss": 0.0338, "step": 4935 }, { "epoch": 0.2631550887668604, "grad_norm": 1.4306389093399048, "learning_rate": 9.013e-06, "loss": 0.024, "step": 4936 }, { "epoch": 0.2632084021965133, "grad_norm": 3.3103978633880615, "learning_rate": 9.012800000000001e-06, "loss": 0.0406, "step": 4937 }, { "epoch": 0.26326171562616624, "grad_norm": 3.563109874725342, "learning_rate": 9.0126e-06, "loss": 0.0338, "step": 4938 }, { "epoch": 0.26331502905581916, "grad_norm": 4.430733680725098, "learning_rate": 9.0124e-06, "loss": 0.0573, "step": 4939 }, { "epoch": 0.2633683424854721, "grad_norm": 2.5925376415252686, "learning_rate": 9.012200000000002e-06, "loss": 0.0378, "step": 4940 }, { "epoch": 0.263421655915125, "grad_norm": 2.660806179046631, "learning_rate": 9.012000000000001e-06, "loss": 0.0472, "step": 4941 }, { "epoch": 0.26347496934477793, "grad_norm": 3.1226110458374023, "learning_rate": 9.0118e-06, "loss": 0.0484, "step": 4942 }, { "epoch": 0.26352828277443086, "grad_norm": 2.8800370693206787, "learning_rate": 9.0116e-06, "loss": 0.0568, "step": 4943 }, { "epoch": 0.26358159620408383, "grad_norm": 2.597177267074585, "learning_rate": 9.011400000000002e-06, "loss": 0.0425, "step": 4944 }, { "epoch": 0.26363490963373676, "grad_norm": 2.1888625621795654, "learning_rate": 9.011200000000001e-06, "loss": 0.0175, "step": 4945 }, { "epoch": 0.2636882230633897, "grad_norm": 3.2314915657043457, "learning_rate": 9.011e-06, "loss": 0.0534, "step": 4946 }, { "epoch": 0.2637415364930426, "grad_norm": 4.1667304039001465, "learning_rate": 9.0108e-06, "loss": 0.0698, "step": 4947 }, { "epoch": 0.26379484992269553, "grad_norm": 3.122715473175049, "learning_rate": 9.0106e-06, "loss": 0.0338, "step": 4948 }, { "epoch": 0.26384816335234845, "grad_norm": 3.983666181564331, "learning_rate": 9.010400000000001e-06, "loss": 0.0599, "step": 4949 }, { "epoch": 0.2639014767820014, "grad_norm": 3.5187530517578125, "learning_rate": 9.0102e-06, "loss": 0.0278, "step": 4950 }, { "epoch": 0.2639547902116543, "grad_norm": 4.1455488204956055, "learning_rate": 9.01e-06, "loss": 0.0571, "step": 4951 }, { "epoch": 0.2640081036413072, "grad_norm": 6.1457672119140625, "learning_rate": 9.009800000000001e-06, "loss": 0.0438, "step": 4952 }, { "epoch": 0.26406141707096015, "grad_norm": 4.175992012023926, "learning_rate": 9.009600000000001e-06, "loss": 0.0382, "step": 4953 }, { "epoch": 0.2641147305006131, "grad_norm": 2.862694263458252, "learning_rate": 9.0094e-06, "loss": 0.0427, "step": 4954 }, { "epoch": 0.26416804393026605, "grad_norm": 2.906139612197876, "learning_rate": 9.009200000000002e-06, "loss": 0.0293, "step": 4955 }, { "epoch": 0.264221357359919, "grad_norm": 6.43183708190918, "learning_rate": 9.009000000000001e-06, "loss": 0.0545, "step": 4956 }, { "epoch": 0.2642746707895719, "grad_norm": 4.209745407104492, "learning_rate": 9.008800000000001e-06, "loss": 0.0497, "step": 4957 }, { "epoch": 0.2643279842192248, "grad_norm": 5.847285270690918, "learning_rate": 9.0086e-06, "loss": 0.0833, "step": 4958 }, { "epoch": 0.26438129764887774, "grad_norm": 7.011894702911377, "learning_rate": 9.0084e-06, "loss": 0.0712, "step": 4959 }, { "epoch": 0.26443461107853067, "grad_norm": 4.077628135681152, "learning_rate": 9.0082e-06, "loss": 0.0333, "step": 4960 }, { "epoch": 0.2644879245081836, "grad_norm": 4.918948173522949, "learning_rate": 9.008e-06, "loss": 0.0625, "step": 4961 }, { "epoch": 0.2645412379378365, "grad_norm": 2.9663093090057373, "learning_rate": 9.0078e-06, "loss": 0.0336, "step": 4962 }, { "epoch": 0.2645945513674895, "grad_norm": 6.046238899230957, "learning_rate": 9.0076e-06, "loss": 0.0361, "step": 4963 }, { "epoch": 0.2646478647971424, "grad_norm": 2.719503164291382, "learning_rate": 9.007400000000001e-06, "loss": 0.0734, "step": 4964 }, { "epoch": 0.26470117822679534, "grad_norm": 3.8458139896392822, "learning_rate": 9.0072e-06, "loss": 0.0219, "step": 4965 }, { "epoch": 0.26475449165644827, "grad_norm": 3.563398599624634, "learning_rate": 9.007e-06, "loss": 0.0301, "step": 4966 }, { "epoch": 0.2648078050861012, "grad_norm": 3.5442187786102295, "learning_rate": 9.006800000000002e-06, "loss": 0.0278, "step": 4967 }, { "epoch": 0.2648611185157541, "grad_norm": 2.4337117671966553, "learning_rate": 9.006600000000001e-06, "loss": 0.0383, "step": 4968 }, { "epoch": 0.26491443194540704, "grad_norm": 7.935568809509277, "learning_rate": 9.0064e-06, "loss": 0.0748, "step": 4969 }, { "epoch": 0.26496774537505996, "grad_norm": 4.203366756439209, "learning_rate": 9.0062e-06, "loss": 0.1093, "step": 4970 }, { "epoch": 0.2650210588047129, "grad_norm": 2.8946001529693604, "learning_rate": 9.006e-06, "loss": 0.0526, "step": 4971 }, { "epoch": 0.26507437223436586, "grad_norm": 4.705783843994141, "learning_rate": 9.005800000000001e-06, "loss": 0.0486, "step": 4972 }, { "epoch": 0.2651276856640188, "grad_norm": 5.386800765991211, "learning_rate": 9.0056e-06, "loss": 0.0542, "step": 4973 }, { "epoch": 0.2651809990936717, "grad_norm": 6.1801438331604, "learning_rate": 9.0054e-06, "loss": 0.0608, "step": 4974 }, { "epoch": 0.26523431252332463, "grad_norm": 5.375235557556152, "learning_rate": 9.0052e-06, "loss": 0.0765, "step": 4975 }, { "epoch": 0.26528762595297756, "grad_norm": 3.489168882369995, "learning_rate": 9.005000000000001e-06, "loss": 0.0319, "step": 4976 }, { "epoch": 0.2653409393826305, "grad_norm": 3.8685638904571533, "learning_rate": 9.0048e-06, "loss": 0.0352, "step": 4977 }, { "epoch": 0.2653942528122834, "grad_norm": 4.8566813468933105, "learning_rate": 9.004600000000002e-06, "loss": 0.069, "step": 4978 }, { "epoch": 0.26544756624193633, "grad_norm": 2.586132287979126, "learning_rate": 9.004400000000001e-06, "loss": 0.0484, "step": 4979 }, { "epoch": 0.26550087967158925, "grad_norm": 4.36729621887207, "learning_rate": 9.004200000000001e-06, "loss": 0.0241, "step": 4980 }, { "epoch": 0.2655541931012422, "grad_norm": 1.8631268739700317, "learning_rate": 9.004e-06, "loss": 0.026, "step": 4981 }, { "epoch": 0.26560750653089515, "grad_norm": 2.2293918132781982, "learning_rate": 9.0038e-06, "loss": 0.024, "step": 4982 }, { "epoch": 0.2656608199605481, "grad_norm": 4.028402328491211, "learning_rate": 9.003600000000001e-06, "loss": 0.0511, "step": 4983 }, { "epoch": 0.265714133390201, "grad_norm": 4.241849899291992, "learning_rate": 9.003400000000001e-06, "loss": 0.0812, "step": 4984 }, { "epoch": 0.2657674468198539, "grad_norm": 5.828392505645752, "learning_rate": 9.0032e-06, "loss": 0.0577, "step": 4985 }, { "epoch": 0.26582076024950685, "grad_norm": 3.6094791889190674, "learning_rate": 9.003e-06, "loss": 0.0816, "step": 4986 }, { "epoch": 0.2658740736791598, "grad_norm": 3.878413200378418, "learning_rate": 9.0028e-06, "loss": 0.094, "step": 4987 }, { "epoch": 0.2659273871088127, "grad_norm": 5.584622383117676, "learning_rate": 9.0026e-06, "loss": 0.0691, "step": 4988 }, { "epoch": 0.2659807005384656, "grad_norm": 7.9283342361450195, "learning_rate": 9.0024e-06, "loss": 0.0837, "step": 4989 }, { "epoch": 0.26603401396811854, "grad_norm": 6.3657050132751465, "learning_rate": 9.002200000000002e-06, "loss": 0.0766, "step": 4990 }, { "epoch": 0.2660873273977715, "grad_norm": 3.126708507537842, "learning_rate": 9.002000000000001e-06, "loss": 0.0421, "step": 4991 }, { "epoch": 0.26614064082742445, "grad_norm": 3.8515732288360596, "learning_rate": 9.0018e-06, "loss": 0.062, "step": 4992 }, { "epoch": 0.26619395425707737, "grad_norm": 12.865378379821777, "learning_rate": 9.0016e-06, "loss": 0.1057, "step": 4993 }, { "epoch": 0.2662472676867303, "grad_norm": 4.139520168304443, "learning_rate": 9.001400000000002e-06, "loss": 0.0436, "step": 4994 }, { "epoch": 0.2663005811163832, "grad_norm": 6.974618434906006, "learning_rate": 9.001200000000001e-06, "loss": 0.0798, "step": 4995 }, { "epoch": 0.26635389454603614, "grad_norm": 4.914154529571533, "learning_rate": 9.001e-06, "loss": 0.0706, "step": 4996 }, { "epoch": 0.26640720797568906, "grad_norm": 3.768308162689209, "learning_rate": 9.0008e-06, "loss": 0.0452, "step": 4997 }, { "epoch": 0.266460521405342, "grad_norm": 2.3525404930114746, "learning_rate": 9.0006e-06, "loss": 0.0367, "step": 4998 }, { "epoch": 0.2665138348349949, "grad_norm": 4.814731597900391, "learning_rate": 9.0004e-06, "loss": 0.0705, "step": 4999 }, { "epoch": 0.2665671482646479, "grad_norm": 2.8361446857452393, "learning_rate": 9.0002e-06, "loss": 0.0514, "step": 5000 }, { "epoch": 0.2666204616943008, "grad_norm": 5.060916423797607, "learning_rate": 9e-06, "loss": 0.0814, "step": 5001 }, { "epoch": 0.26667377512395374, "grad_norm": 3.1713855266571045, "learning_rate": 8.999800000000001e-06, "loss": 0.054, "step": 5002 }, { "epoch": 0.26672708855360666, "grad_norm": 3.2080180644989014, "learning_rate": 8.999600000000001e-06, "loss": 0.0314, "step": 5003 }, { "epoch": 0.2667804019832596, "grad_norm": 3.452730417251587, "learning_rate": 8.9994e-06, "loss": 0.0581, "step": 5004 }, { "epoch": 0.2668337154129125, "grad_norm": 5.609930992126465, "learning_rate": 8.999200000000002e-06, "loss": 0.0695, "step": 5005 }, { "epoch": 0.26688702884256543, "grad_norm": 6.333728313446045, "learning_rate": 8.999000000000001e-06, "loss": 0.0889, "step": 5006 }, { "epoch": 0.26694034227221836, "grad_norm": 3.5692954063415527, "learning_rate": 8.998800000000001e-06, "loss": 0.052, "step": 5007 }, { "epoch": 0.2669936557018713, "grad_norm": 1.9077365398406982, "learning_rate": 8.9986e-06, "loss": 0.031, "step": 5008 }, { "epoch": 0.2670469691315242, "grad_norm": 4.13592004776001, "learning_rate": 8.9984e-06, "loss": 0.0738, "step": 5009 }, { "epoch": 0.2671002825611772, "grad_norm": 3.031796455383301, "learning_rate": 8.9982e-06, "loss": 0.0495, "step": 5010 }, { "epoch": 0.2671535959908301, "grad_norm": 3.0757179260253906, "learning_rate": 8.998000000000001e-06, "loss": 0.0498, "step": 5011 }, { "epoch": 0.26720690942048303, "grad_norm": 2.79972767829895, "learning_rate": 8.9978e-06, "loss": 0.0324, "step": 5012 }, { "epoch": 0.26726022285013595, "grad_norm": 2.464641809463501, "learning_rate": 8.9976e-06, "loss": 0.0387, "step": 5013 }, { "epoch": 0.2673135362797889, "grad_norm": 4.310692310333252, "learning_rate": 8.997400000000001e-06, "loss": 0.04, "step": 5014 }, { "epoch": 0.2673668497094418, "grad_norm": 3.9350316524505615, "learning_rate": 8.997200000000001e-06, "loss": 0.0525, "step": 5015 }, { "epoch": 0.2674201631390947, "grad_norm": 6.570262432098389, "learning_rate": 8.997000000000002e-06, "loss": 0.0703, "step": 5016 }, { "epoch": 0.26747347656874765, "grad_norm": 2.8677978515625, "learning_rate": 8.996800000000002e-06, "loss": 0.0469, "step": 5017 }, { "epoch": 0.26752678999840057, "grad_norm": 4.222312927246094, "learning_rate": 8.996600000000001e-06, "loss": 0.0332, "step": 5018 }, { "epoch": 0.26758010342805355, "grad_norm": 2.9571099281311035, "learning_rate": 8.9964e-06, "loss": 0.0398, "step": 5019 }, { "epoch": 0.2676334168577065, "grad_norm": 3.34100604057312, "learning_rate": 8.9962e-06, "loss": 0.0207, "step": 5020 }, { "epoch": 0.2676867302873594, "grad_norm": 3.227504253387451, "learning_rate": 8.996e-06, "loss": 0.0262, "step": 5021 }, { "epoch": 0.2677400437170123, "grad_norm": 4.464113712310791, "learning_rate": 8.995800000000001e-06, "loss": 0.0495, "step": 5022 }, { "epoch": 0.26779335714666525, "grad_norm": 4.547257900238037, "learning_rate": 8.9956e-06, "loss": 0.0253, "step": 5023 }, { "epoch": 0.26784667057631817, "grad_norm": 4.046042442321777, "learning_rate": 8.9954e-06, "loss": 0.0475, "step": 5024 }, { "epoch": 0.2678999840059711, "grad_norm": 2.7949955463409424, "learning_rate": 8.9952e-06, "loss": 0.0431, "step": 5025 }, { "epoch": 0.267953297435624, "grad_norm": 5.354994297027588, "learning_rate": 8.995000000000001e-06, "loss": 0.0674, "step": 5026 }, { "epoch": 0.26800661086527694, "grad_norm": 2.3615288734436035, "learning_rate": 8.9948e-06, "loss": 0.0169, "step": 5027 }, { "epoch": 0.2680599242949299, "grad_norm": 4.177478790283203, "learning_rate": 8.994600000000002e-06, "loss": 0.0666, "step": 5028 }, { "epoch": 0.26811323772458284, "grad_norm": 5.87453031539917, "learning_rate": 8.994400000000002e-06, "loss": 0.0785, "step": 5029 }, { "epoch": 0.26816655115423577, "grad_norm": 2.9100382328033447, "learning_rate": 8.994200000000001e-06, "loss": 0.039, "step": 5030 }, { "epoch": 0.2682198645838887, "grad_norm": 3.7002382278442383, "learning_rate": 8.994e-06, "loss": 0.0506, "step": 5031 }, { "epoch": 0.2682731780135416, "grad_norm": 6.25092077255249, "learning_rate": 8.9938e-06, "loss": 0.1164, "step": 5032 }, { "epoch": 0.26832649144319454, "grad_norm": 6.04270601272583, "learning_rate": 8.993600000000001e-06, "loss": 0.0447, "step": 5033 }, { "epoch": 0.26837980487284746, "grad_norm": 5.496007919311523, "learning_rate": 8.993400000000001e-06, "loss": 0.0444, "step": 5034 }, { "epoch": 0.2684331183025004, "grad_norm": 3.3506112098693848, "learning_rate": 8.9932e-06, "loss": 0.0334, "step": 5035 }, { "epoch": 0.2684864317321533, "grad_norm": 2.2603068351745605, "learning_rate": 8.993e-06, "loss": 0.017, "step": 5036 }, { "epoch": 0.26853974516180623, "grad_norm": 2.308047294616699, "learning_rate": 8.9928e-06, "loss": 0.0288, "step": 5037 }, { "epoch": 0.2685930585914592, "grad_norm": 5.543563365936279, "learning_rate": 8.992600000000001e-06, "loss": 0.0826, "step": 5038 }, { "epoch": 0.26864637202111213, "grad_norm": 4.5310540199279785, "learning_rate": 8.9924e-06, "loss": 0.0556, "step": 5039 }, { "epoch": 0.26869968545076506, "grad_norm": 6.839753150939941, "learning_rate": 8.992200000000002e-06, "loss": 0.0696, "step": 5040 }, { "epoch": 0.268752998880418, "grad_norm": 3.99835205078125, "learning_rate": 8.992000000000001e-06, "loss": 0.0694, "step": 5041 }, { "epoch": 0.2688063123100709, "grad_norm": 4.831282138824463, "learning_rate": 8.991800000000001e-06, "loss": 0.0284, "step": 5042 }, { "epoch": 0.26885962573972383, "grad_norm": 4.749800205230713, "learning_rate": 8.9916e-06, "loss": 0.0613, "step": 5043 }, { "epoch": 0.26891293916937675, "grad_norm": 3.7218525409698486, "learning_rate": 8.991400000000002e-06, "loss": 0.0438, "step": 5044 }, { "epoch": 0.2689662525990297, "grad_norm": 2.8786940574645996, "learning_rate": 8.991200000000001e-06, "loss": 0.0588, "step": 5045 }, { "epoch": 0.2690195660286826, "grad_norm": 4.616518974304199, "learning_rate": 8.991e-06, "loss": 0.0691, "step": 5046 }, { "epoch": 0.2690728794583356, "grad_norm": 3.5737733840942383, "learning_rate": 8.9908e-06, "loss": 0.037, "step": 5047 }, { "epoch": 0.2691261928879885, "grad_norm": 3.657095193862915, "learning_rate": 8.9906e-06, "loss": 0.0678, "step": 5048 }, { "epoch": 0.2691795063176414, "grad_norm": 4.7217183113098145, "learning_rate": 8.9904e-06, "loss": 0.0388, "step": 5049 }, { "epoch": 0.26923281974729435, "grad_norm": 3.352222442626953, "learning_rate": 8.9902e-06, "loss": 0.0586, "step": 5050 }, { "epoch": 0.2692861331769473, "grad_norm": 3.162360191345215, "learning_rate": 8.99e-06, "loss": 0.031, "step": 5051 }, { "epoch": 0.2693394466066002, "grad_norm": 3.200432300567627, "learning_rate": 8.989800000000002e-06, "loss": 0.0542, "step": 5052 }, { "epoch": 0.2693927600362531, "grad_norm": 4.630084991455078, "learning_rate": 8.989600000000001e-06, "loss": 0.0312, "step": 5053 }, { "epoch": 0.26944607346590604, "grad_norm": 2.337872266769409, "learning_rate": 8.9894e-06, "loss": 0.0275, "step": 5054 }, { "epoch": 0.26949938689555897, "grad_norm": 6.405411720275879, "learning_rate": 8.989200000000002e-06, "loss": 0.069, "step": 5055 }, { "epoch": 0.26955270032521195, "grad_norm": 2.415085554122925, "learning_rate": 8.989000000000002e-06, "loss": 0.0329, "step": 5056 }, { "epoch": 0.26960601375486487, "grad_norm": 2.619760751724243, "learning_rate": 8.988800000000001e-06, "loss": 0.0298, "step": 5057 }, { "epoch": 0.2696593271845178, "grad_norm": 6.365194797515869, "learning_rate": 8.9886e-06, "loss": 0.0793, "step": 5058 }, { "epoch": 0.2697126406141707, "grad_norm": 5.78893518447876, "learning_rate": 8.9884e-06, "loss": 0.0803, "step": 5059 }, { "epoch": 0.26976595404382364, "grad_norm": 5.5550761222839355, "learning_rate": 8.9882e-06, "loss": 0.079, "step": 5060 }, { "epoch": 0.26981926747347657, "grad_norm": 3.055960178375244, "learning_rate": 8.988000000000001e-06, "loss": 0.0265, "step": 5061 }, { "epoch": 0.2698725809031295, "grad_norm": 2.6940033435821533, "learning_rate": 8.9878e-06, "loss": 0.0263, "step": 5062 }, { "epoch": 0.2699258943327824, "grad_norm": 2.16087007522583, "learning_rate": 8.9876e-06, "loss": 0.0377, "step": 5063 }, { "epoch": 0.26997920776243534, "grad_norm": 3.6833388805389404, "learning_rate": 8.987400000000001e-06, "loss": 0.0529, "step": 5064 }, { "epoch": 0.27003252119208826, "grad_norm": 4.059699058532715, "learning_rate": 8.987200000000001e-06, "loss": 0.0553, "step": 5065 }, { "epoch": 0.27008583462174124, "grad_norm": 3.1537461280822754, "learning_rate": 8.987e-06, "loss": 0.0442, "step": 5066 }, { "epoch": 0.27013914805139416, "grad_norm": 3.8023147583007812, "learning_rate": 8.986800000000002e-06, "loss": 0.0503, "step": 5067 }, { "epoch": 0.2701924614810471, "grad_norm": 3.6606945991516113, "learning_rate": 8.986600000000001e-06, "loss": 0.062, "step": 5068 }, { "epoch": 0.2702457749107, "grad_norm": 1.7917200326919556, "learning_rate": 8.986400000000001e-06, "loss": 0.0344, "step": 5069 }, { "epoch": 0.27029908834035293, "grad_norm": 5.6993489265441895, "learning_rate": 8.9862e-06, "loss": 0.0521, "step": 5070 }, { "epoch": 0.27035240177000586, "grad_norm": 3.675428628921509, "learning_rate": 8.986e-06, "loss": 0.0399, "step": 5071 }, { "epoch": 0.2704057151996588, "grad_norm": 3.578493356704712, "learning_rate": 8.985800000000001e-06, "loss": 0.0932, "step": 5072 }, { "epoch": 0.2704590286293117, "grad_norm": 6.435462951660156, "learning_rate": 8.9856e-06, "loss": 0.0614, "step": 5073 }, { "epoch": 0.2705123420589646, "grad_norm": 3.2671010494232178, "learning_rate": 8.9854e-06, "loss": 0.029, "step": 5074 }, { "epoch": 0.2705656554886176, "grad_norm": 3.5316343307495117, "learning_rate": 8.9852e-06, "loss": 0.0441, "step": 5075 }, { "epoch": 0.27061896891827053, "grad_norm": 4.720334053039551, "learning_rate": 8.985000000000001e-06, "loss": 0.0691, "step": 5076 }, { "epoch": 0.27067228234792345, "grad_norm": 2.741532325744629, "learning_rate": 8.9848e-06, "loss": 0.0442, "step": 5077 }, { "epoch": 0.2707255957775764, "grad_norm": 2.7613167762756348, "learning_rate": 8.9846e-06, "loss": 0.0578, "step": 5078 }, { "epoch": 0.2707789092072293, "grad_norm": 5.71161413192749, "learning_rate": 8.984400000000002e-06, "loss": 0.0674, "step": 5079 }, { "epoch": 0.2708322226368822, "grad_norm": 2.064751386642456, "learning_rate": 8.984200000000001e-06, "loss": 0.0144, "step": 5080 }, { "epoch": 0.27088553606653515, "grad_norm": 2.780984878540039, "learning_rate": 8.984e-06, "loss": 0.0264, "step": 5081 }, { "epoch": 0.27093884949618807, "grad_norm": 3.0073251724243164, "learning_rate": 8.9838e-06, "loss": 0.0392, "step": 5082 }, { "epoch": 0.270992162925841, "grad_norm": 3.682408332824707, "learning_rate": 8.983600000000002e-06, "loss": 0.0268, "step": 5083 }, { "epoch": 0.271045476355494, "grad_norm": 4.158533096313477, "learning_rate": 8.983400000000001e-06, "loss": 0.0722, "step": 5084 }, { "epoch": 0.2710987897851469, "grad_norm": 2.6539056301116943, "learning_rate": 8.9832e-06, "loss": 0.0592, "step": 5085 }, { "epoch": 0.2711521032147998, "grad_norm": 2.7114827632904053, "learning_rate": 8.983e-06, "loss": 0.0484, "step": 5086 }, { "epoch": 0.27120541664445275, "grad_norm": 3.477975845336914, "learning_rate": 8.9828e-06, "loss": 0.0343, "step": 5087 }, { "epoch": 0.27125873007410567, "grad_norm": 5.0936126708984375, "learning_rate": 8.982600000000001e-06, "loss": 0.0691, "step": 5088 }, { "epoch": 0.2713120435037586, "grad_norm": 2.8056962490081787, "learning_rate": 8.9824e-06, "loss": 0.0382, "step": 5089 }, { "epoch": 0.2713653569334115, "grad_norm": 2.225726366043091, "learning_rate": 8.9822e-06, "loss": 0.0173, "step": 5090 }, { "epoch": 0.27141867036306444, "grad_norm": 5.493151664733887, "learning_rate": 8.982000000000001e-06, "loss": 0.0577, "step": 5091 }, { "epoch": 0.27147198379271736, "grad_norm": 4.634810924530029, "learning_rate": 8.981800000000001e-06, "loss": 0.0697, "step": 5092 }, { "epoch": 0.2715252972223703, "grad_norm": 3.4915947914123535, "learning_rate": 8.9816e-06, "loss": 0.0527, "step": 5093 }, { "epoch": 0.27157861065202327, "grad_norm": 4.681921005249023, "learning_rate": 8.981400000000002e-06, "loss": 0.0565, "step": 5094 }, { "epoch": 0.2716319240816762, "grad_norm": 4.473388195037842, "learning_rate": 8.981200000000001e-06, "loss": 0.0667, "step": 5095 }, { "epoch": 0.2716852375113291, "grad_norm": 2.525848388671875, "learning_rate": 8.981000000000001e-06, "loss": 0.0358, "step": 5096 }, { "epoch": 0.27173855094098204, "grad_norm": 3.067416191101074, "learning_rate": 8.9808e-06, "loss": 0.0514, "step": 5097 }, { "epoch": 0.27179186437063496, "grad_norm": 10.690455436706543, "learning_rate": 8.9806e-06, "loss": 0.0753, "step": 5098 }, { "epoch": 0.2718451778002879, "grad_norm": 4.849803447723389, "learning_rate": 8.9804e-06, "loss": 0.0413, "step": 5099 }, { "epoch": 0.2718984912299408, "grad_norm": 5.992323875427246, "learning_rate": 8.980200000000001e-06, "loss": 0.0572, "step": 5100 }, { "epoch": 0.27195180465959373, "grad_norm": 3.713949203491211, "learning_rate": 8.98e-06, "loss": 0.047, "step": 5101 }, { "epoch": 0.27200511808924666, "grad_norm": 3.689840793609619, "learning_rate": 8.9798e-06, "loss": 0.0606, "step": 5102 }, { "epoch": 0.27205843151889963, "grad_norm": 2.9174046516418457, "learning_rate": 8.979600000000001e-06, "loss": 0.0412, "step": 5103 }, { "epoch": 0.27211174494855256, "grad_norm": 3.37290620803833, "learning_rate": 8.9794e-06, "loss": 0.0623, "step": 5104 }, { "epoch": 0.2721650583782055, "grad_norm": 3.3871147632598877, "learning_rate": 8.979200000000002e-06, "loss": 0.0534, "step": 5105 }, { "epoch": 0.2722183718078584, "grad_norm": 6.820534706115723, "learning_rate": 8.979000000000002e-06, "loss": 0.0825, "step": 5106 }, { "epoch": 0.27227168523751133, "grad_norm": 6.7399516105651855, "learning_rate": 8.978800000000001e-06, "loss": 0.0579, "step": 5107 }, { "epoch": 0.27232499866716425, "grad_norm": 3.87471604347229, "learning_rate": 8.9786e-06, "loss": 0.051, "step": 5108 }, { "epoch": 0.2723783120968172, "grad_norm": 2.286313056945801, "learning_rate": 8.9784e-06, "loss": 0.022, "step": 5109 }, { "epoch": 0.2724316255264701, "grad_norm": 3.608494997024536, "learning_rate": 8.9782e-06, "loss": 0.0494, "step": 5110 }, { "epoch": 0.272484938956123, "grad_norm": 3.478907346725464, "learning_rate": 8.978000000000001e-06, "loss": 0.049, "step": 5111 }, { "epoch": 0.272538252385776, "grad_norm": 4.480889320373535, "learning_rate": 8.9778e-06, "loss": 0.0518, "step": 5112 }, { "epoch": 0.2725915658154289, "grad_norm": 4.975453853607178, "learning_rate": 8.9776e-06, "loss": 0.0653, "step": 5113 }, { "epoch": 0.27264487924508185, "grad_norm": 2.1941492557525635, "learning_rate": 8.9774e-06, "loss": 0.0403, "step": 5114 }, { "epoch": 0.2726981926747348, "grad_norm": 3.4647557735443115, "learning_rate": 8.977200000000001e-06, "loss": 0.0903, "step": 5115 }, { "epoch": 0.2727515061043877, "grad_norm": 3.0714447498321533, "learning_rate": 8.977e-06, "loss": 0.0447, "step": 5116 }, { "epoch": 0.2728048195340406, "grad_norm": 3.03157377243042, "learning_rate": 8.976800000000002e-06, "loss": 0.0341, "step": 5117 }, { "epoch": 0.27285813296369354, "grad_norm": 3.909381628036499, "learning_rate": 8.976600000000001e-06, "loss": 0.0491, "step": 5118 }, { "epoch": 0.27291144639334647, "grad_norm": 5.167801380157471, "learning_rate": 8.976400000000001e-06, "loss": 0.089, "step": 5119 }, { "epoch": 0.2729647598229994, "grad_norm": 4.25054931640625, "learning_rate": 8.9762e-06, "loss": 0.0504, "step": 5120 }, { "epoch": 0.2730180732526523, "grad_norm": 6.027379035949707, "learning_rate": 8.976e-06, "loss": 0.0467, "step": 5121 }, { "epoch": 0.2730713866823053, "grad_norm": 4.9793548583984375, "learning_rate": 8.975800000000001e-06, "loss": 0.0478, "step": 5122 }, { "epoch": 0.2731247001119582, "grad_norm": 4.130747318267822, "learning_rate": 8.975600000000001e-06, "loss": 0.0426, "step": 5123 }, { "epoch": 0.27317801354161114, "grad_norm": 5.226565361022949, "learning_rate": 8.9754e-06, "loss": 0.0686, "step": 5124 }, { "epoch": 0.27323132697126407, "grad_norm": 3.2723093032836914, "learning_rate": 8.9752e-06, "loss": 0.0729, "step": 5125 }, { "epoch": 0.273284640400917, "grad_norm": 4.427443981170654, "learning_rate": 8.975e-06, "loss": 0.0562, "step": 5126 }, { "epoch": 0.2733379538305699, "grad_norm": 3.8533310890197754, "learning_rate": 8.974800000000001e-06, "loss": 0.0449, "step": 5127 }, { "epoch": 0.27339126726022284, "grad_norm": 2.840904474258423, "learning_rate": 8.9746e-06, "loss": 0.0429, "step": 5128 }, { "epoch": 0.27344458068987576, "grad_norm": 6.880736827850342, "learning_rate": 8.974400000000002e-06, "loss": 0.0493, "step": 5129 }, { "epoch": 0.2734978941195287, "grad_norm": 4.700404644012451, "learning_rate": 8.974200000000001e-06, "loss": 0.0628, "step": 5130 }, { "epoch": 0.27355120754918166, "grad_norm": 3.566908121109009, "learning_rate": 8.974e-06, "loss": 0.0807, "step": 5131 }, { "epoch": 0.2736045209788346, "grad_norm": 1.9792968034744263, "learning_rate": 8.9738e-06, "loss": 0.0381, "step": 5132 }, { "epoch": 0.2736578344084875, "grad_norm": 3.160051107406616, "learning_rate": 8.973600000000002e-06, "loss": 0.034, "step": 5133 }, { "epoch": 0.27371114783814043, "grad_norm": 2.654930830001831, "learning_rate": 8.973400000000001e-06, "loss": 0.0247, "step": 5134 }, { "epoch": 0.27376446126779336, "grad_norm": 3.998967170715332, "learning_rate": 8.9732e-06, "loss": 0.065, "step": 5135 }, { "epoch": 0.2738177746974463, "grad_norm": 2.348132610321045, "learning_rate": 8.973e-06, "loss": 0.0246, "step": 5136 }, { "epoch": 0.2738710881270992, "grad_norm": 2.6507623195648193, "learning_rate": 8.9728e-06, "loss": 0.0257, "step": 5137 }, { "epoch": 0.27392440155675213, "grad_norm": 4.405205726623535, "learning_rate": 8.9726e-06, "loss": 0.0919, "step": 5138 }, { "epoch": 0.27397771498640505, "grad_norm": 3.0861315727233887, "learning_rate": 8.9724e-06, "loss": 0.0401, "step": 5139 }, { "epoch": 0.27403102841605803, "grad_norm": 2.2964916229248047, "learning_rate": 8.9722e-06, "loss": 0.0428, "step": 5140 }, { "epoch": 0.27408434184571095, "grad_norm": 3.8483850955963135, "learning_rate": 8.972000000000002e-06, "loss": 0.0427, "step": 5141 }, { "epoch": 0.2741376552753639, "grad_norm": 3.1074254512786865, "learning_rate": 8.971800000000001e-06, "loss": 0.0348, "step": 5142 }, { "epoch": 0.2741909687050168, "grad_norm": 5.219662666320801, "learning_rate": 8.9716e-06, "loss": 0.0651, "step": 5143 }, { "epoch": 0.2742442821346697, "grad_norm": 2.5993454456329346, "learning_rate": 8.971400000000002e-06, "loss": 0.0382, "step": 5144 }, { "epoch": 0.27429759556432265, "grad_norm": 2.9628140926361084, "learning_rate": 8.971200000000001e-06, "loss": 0.0418, "step": 5145 }, { "epoch": 0.2743509089939756, "grad_norm": 3.354796886444092, "learning_rate": 8.971000000000001e-06, "loss": 0.0357, "step": 5146 }, { "epoch": 0.2744042224236285, "grad_norm": 3.078143358230591, "learning_rate": 8.9708e-06, "loss": 0.0485, "step": 5147 }, { "epoch": 0.2744575358532814, "grad_norm": 3.1093783378601074, "learning_rate": 8.9706e-06, "loss": 0.0499, "step": 5148 }, { "epoch": 0.27451084928293434, "grad_norm": 2.74318528175354, "learning_rate": 8.9704e-06, "loss": 0.0538, "step": 5149 }, { "epoch": 0.2745641627125873, "grad_norm": 5.340903282165527, "learning_rate": 8.970200000000001e-06, "loss": 0.0816, "step": 5150 }, { "epoch": 0.27461747614224025, "grad_norm": 1.7920576333999634, "learning_rate": 8.97e-06, "loss": 0.0315, "step": 5151 }, { "epoch": 0.27467078957189317, "grad_norm": 2.9902615547180176, "learning_rate": 8.9698e-06, "loss": 0.0605, "step": 5152 }, { "epoch": 0.2747241030015461, "grad_norm": 1.5291075706481934, "learning_rate": 8.969600000000001e-06, "loss": 0.0188, "step": 5153 }, { "epoch": 0.274777416431199, "grad_norm": 2.912806749343872, "learning_rate": 8.969400000000001e-06, "loss": 0.0416, "step": 5154 }, { "epoch": 0.27483072986085194, "grad_norm": 1.829641580581665, "learning_rate": 8.9692e-06, "loss": 0.0208, "step": 5155 }, { "epoch": 0.27488404329050486, "grad_norm": 4.93087911605835, "learning_rate": 8.969000000000002e-06, "loss": 0.0578, "step": 5156 }, { "epoch": 0.2749373567201578, "grad_norm": 3.6602931022644043, "learning_rate": 8.968800000000001e-06, "loss": 0.0759, "step": 5157 }, { "epoch": 0.2749906701498107, "grad_norm": 4.618592739105225, "learning_rate": 8.9686e-06, "loss": 0.0513, "step": 5158 }, { "epoch": 0.2750439835794637, "grad_norm": 3.5539634227752686, "learning_rate": 8.9684e-06, "loss": 0.0461, "step": 5159 }, { "epoch": 0.2750972970091166, "grad_norm": 4.115421295166016, "learning_rate": 8.9682e-06, "loss": 0.0339, "step": 5160 }, { "epoch": 0.27515061043876954, "grad_norm": 3.6396403312683105, "learning_rate": 8.968000000000001e-06, "loss": 0.0517, "step": 5161 }, { "epoch": 0.27520392386842246, "grad_norm": 3.625431537628174, "learning_rate": 8.9678e-06, "loss": 0.054, "step": 5162 }, { "epoch": 0.2752572372980754, "grad_norm": 3.959505796432495, "learning_rate": 8.9676e-06, "loss": 0.058, "step": 5163 }, { "epoch": 0.2753105507277283, "grad_norm": 5.709029197692871, "learning_rate": 8.9674e-06, "loss": 0.0381, "step": 5164 }, { "epoch": 0.27536386415738123, "grad_norm": 3.597966194152832, "learning_rate": 8.967200000000001e-06, "loss": 0.0583, "step": 5165 }, { "epoch": 0.27541717758703416, "grad_norm": 3.459254503250122, "learning_rate": 8.967e-06, "loss": 0.0349, "step": 5166 }, { "epoch": 0.2754704910166871, "grad_norm": 8.141888618469238, "learning_rate": 8.9668e-06, "loss": 0.0536, "step": 5167 }, { "epoch": 0.27552380444634006, "grad_norm": 7.655948162078857, "learning_rate": 8.966600000000002e-06, "loss": 0.0554, "step": 5168 }, { "epoch": 0.275577117875993, "grad_norm": 5.802751541137695, "learning_rate": 8.966400000000001e-06, "loss": 0.0946, "step": 5169 }, { "epoch": 0.2756304313056459, "grad_norm": 2.261186361312866, "learning_rate": 8.9662e-06, "loss": 0.0203, "step": 5170 }, { "epoch": 0.27568374473529883, "grad_norm": 2.2115769386291504, "learning_rate": 8.966e-06, "loss": 0.0482, "step": 5171 }, { "epoch": 0.27573705816495175, "grad_norm": 2.052971124649048, "learning_rate": 8.965800000000001e-06, "loss": 0.0191, "step": 5172 }, { "epoch": 0.2757903715946047, "grad_norm": 3.143880605697632, "learning_rate": 8.965600000000001e-06, "loss": 0.0335, "step": 5173 }, { "epoch": 0.2758436850242576, "grad_norm": 3.542320489883423, "learning_rate": 8.9654e-06, "loss": 0.055, "step": 5174 }, { "epoch": 0.2758969984539105, "grad_norm": 10.288379669189453, "learning_rate": 8.9652e-06, "loss": 0.0783, "step": 5175 }, { "epoch": 0.27595031188356345, "grad_norm": 3.066218852996826, "learning_rate": 8.965e-06, "loss": 0.0397, "step": 5176 }, { "epoch": 0.27600362531321637, "grad_norm": 5.451569557189941, "learning_rate": 8.964800000000001e-06, "loss": 0.1096, "step": 5177 }, { "epoch": 0.27605693874286935, "grad_norm": 4.846907615661621, "learning_rate": 8.9646e-06, "loss": 0.0516, "step": 5178 }, { "epoch": 0.2761102521725223, "grad_norm": 4.548272609710693, "learning_rate": 8.9644e-06, "loss": 0.0541, "step": 5179 }, { "epoch": 0.2761635656021752, "grad_norm": 3.5976345539093018, "learning_rate": 8.964200000000001e-06, "loss": 0.0419, "step": 5180 }, { "epoch": 0.2762168790318281, "grad_norm": 3.3396713733673096, "learning_rate": 8.964000000000001e-06, "loss": 0.0631, "step": 5181 }, { "epoch": 0.27627019246148105, "grad_norm": 9.76309871673584, "learning_rate": 8.9638e-06, "loss": 0.0605, "step": 5182 }, { "epoch": 0.27632350589113397, "grad_norm": 5.097068786621094, "learning_rate": 8.963600000000002e-06, "loss": 0.084, "step": 5183 }, { "epoch": 0.2763768193207869, "grad_norm": 4.524988651275635, "learning_rate": 8.963400000000001e-06, "loss": 0.0527, "step": 5184 }, { "epoch": 0.2764301327504398, "grad_norm": 3.0923328399658203, "learning_rate": 8.963200000000001e-06, "loss": 0.0457, "step": 5185 }, { "epoch": 0.27648344618009274, "grad_norm": 3.659513473510742, "learning_rate": 8.963e-06, "loss": 0.0538, "step": 5186 }, { "epoch": 0.2765367596097457, "grad_norm": 3.3676462173461914, "learning_rate": 8.9628e-06, "loss": 0.03, "step": 5187 }, { "epoch": 0.27659007303939864, "grad_norm": 6.8135833740234375, "learning_rate": 8.9626e-06, "loss": 0.0666, "step": 5188 }, { "epoch": 0.27664338646905157, "grad_norm": 5.51185941696167, "learning_rate": 8.9624e-06, "loss": 0.0594, "step": 5189 }, { "epoch": 0.2766966998987045, "grad_norm": 4.143584251403809, "learning_rate": 8.9622e-06, "loss": 0.0365, "step": 5190 }, { "epoch": 0.2767500133283574, "grad_norm": 1.8788940906524658, "learning_rate": 8.962e-06, "loss": 0.0229, "step": 5191 }, { "epoch": 0.27680332675801034, "grad_norm": 3.2630741596221924, "learning_rate": 8.961800000000001e-06, "loss": 0.0314, "step": 5192 }, { "epoch": 0.27685664018766326, "grad_norm": 4.752302646636963, "learning_rate": 8.9616e-06, "loss": 0.0337, "step": 5193 }, { "epoch": 0.2769099536173162, "grad_norm": 5.4770379066467285, "learning_rate": 8.961400000000002e-06, "loss": 0.0363, "step": 5194 }, { "epoch": 0.2769632670469691, "grad_norm": 4.7295379638671875, "learning_rate": 8.961200000000002e-06, "loss": 0.0517, "step": 5195 }, { "epoch": 0.2770165804766221, "grad_norm": 3.4857637882232666, "learning_rate": 8.961000000000001e-06, "loss": 0.0331, "step": 5196 }, { "epoch": 0.277069893906275, "grad_norm": 3.3513731956481934, "learning_rate": 8.9608e-06, "loss": 0.052, "step": 5197 }, { "epoch": 0.27712320733592793, "grad_norm": 2.978583574295044, "learning_rate": 8.9606e-06, "loss": 0.0698, "step": 5198 }, { "epoch": 0.27717652076558086, "grad_norm": 6.453673362731934, "learning_rate": 8.9604e-06, "loss": 0.0573, "step": 5199 }, { "epoch": 0.2772298341952338, "grad_norm": 3.14337420463562, "learning_rate": 8.960200000000001e-06, "loss": 0.0444, "step": 5200 }, { "epoch": 0.2772831476248867, "grad_norm": 2.5005078315734863, "learning_rate": 8.96e-06, "loss": 0.0394, "step": 5201 }, { "epoch": 0.27733646105453963, "grad_norm": 3.2535154819488525, "learning_rate": 8.9598e-06, "loss": 0.0299, "step": 5202 }, { "epoch": 0.27738977448419255, "grad_norm": 4.362800121307373, "learning_rate": 8.9596e-06, "loss": 0.0526, "step": 5203 }, { "epoch": 0.2774430879138455, "grad_norm": 4.026961803436279, "learning_rate": 8.959400000000001e-06, "loss": 0.0516, "step": 5204 }, { "epoch": 0.2774964013434984, "grad_norm": 3.626204252243042, "learning_rate": 8.9592e-06, "loss": 0.0371, "step": 5205 }, { "epoch": 0.2775497147731514, "grad_norm": 5.436481475830078, "learning_rate": 8.959000000000002e-06, "loss": 0.0298, "step": 5206 }, { "epoch": 0.2776030282028043, "grad_norm": 6.500696182250977, "learning_rate": 8.958800000000001e-06, "loss": 0.0482, "step": 5207 }, { "epoch": 0.2776563416324572, "grad_norm": 4.348647117614746, "learning_rate": 8.958600000000001e-06, "loss": 0.0447, "step": 5208 }, { "epoch": 0.27770965506211015, "grad_norm": 4.5174055099487305, "learning_rate": 8.9584e-06, "loss": 0.0592, "step": 5209 }, { "epoch": 0.2777629684917631, "grad_norm": 3.761747121810913, "learning_rate": 8.9582e-06, "loss": 0.0328, "step": 5210 }, { "epoch": 0.277816281921416, "grad_norm": 2.5875935554504395, "learning_rate": 8.958000000000001e-06, "loss": 0.0252, "step": 5211 }, { "epoch": 0.2778695953510689, "grad_norm": 6.267485618591309, "learning_rate": 8.957800000000001e-06, "loss": 0.069, "step": 5212 }, { "epoch": 0.27792290878072184, "grad_norm": 3.0575578212738037, "learning_rate": 8.9576e-06, "loss": 0.0593, "step": 5213 }, { "epoch": 0.27797622221037477, "grad_norm": 4.271108150482178, "learning_rate": 8.9574e-06, "loss": 0.0707, "step": 5214 }, { "epoch": 0.27802953564002775, "grad_norm": 4.049891471862793, "learning_rate": 8.9572e-06, "loss": 0.04, "step": 5215 }, { "epoch": 0.27808284906968067, "grad_norm": 1.727791666984558, "learning_rate": 8.957e-06, "loss": 0.024, "step": 5216 }, { "epoch": 0.2781361624993336, "grad_norm": 2.8207926750183105, "learning_rate": 8.9568e-06, "loss": 0.0384, "step": 5217 }, { "epoch": 0.2781894759289865, "grad_norm": 4.631971836090088, "learning_rate": 8.956600000000002e-06, "loss": 0.0458, "step": 5218 }, { "epoch": 0.27824278935863944, "grad_norm": 2.3885276317596436, "learning_rate": 8.956400000000001e-06, "loss": 0.0221, "step": 5219 }, { "epoch": 0.27829610278829237, "grad_norm": 3.3242714405059814, "learning_rate": 8.9562e-06, "loss": 0.0395, "step": 5220 }, { "epoch": 0.2783494162179453, "grad_norm": 4.197261810302734, "learning_rate": 8.956e-06, "loss": 0.0638, "step": 5221 }, { "epoch": 0.2784027296475982, "grad_norm": 5.669502258300781, "learning_rate": 8.955800000000002e-06, "loss": 0.0708, "step": 5222 }, { "epoch": 0.27845604307725114, "grad_norm": 5.103144645690918, "learning_rate": 8.955600000000001e-06, "loss": 0.0314, "step": 5223 }, { "epoch": 0.2785093565069041, "grad_norm": 3.502972364425659, "learning_rate": 8.9554e-06, "loss": 0.0494, "step": 5224 }, { "epoch": 0.27856266993655704, "grad_norm": 2.51789927482605, "learning_rate": 8.9552e-06, "loss": 0.0633, "step": 5225 }, { "epoch": 0.27861598336620996, "grad_norm": 4.295873165130615, "learning_rate": 8.955e-06, "loss": 0.0596, "step": 5226 }, { "epoch": 0.2786692967958629, "grad_norm": 2.7386741638183594, "learning_rate": 8.954800000000001e-06, "loss": 0.0275, "step": 5227 }, { "epoch": 0.2787226102255158, "grad_norm": 2.5732574462890625, "learning_rate": 8.9546e-06, "loss": 0.0396, "step": 5228 }, { "epoch": 0.27877592365516873, "grad_norm": 2.2279696464538574, "learning_rate": 8.9544e-06, "loss": 0.0205, "step": 5229 }, { "epoch": 0.27882923708482166, "grad_norm": 4.32731819152832, "learning_rate": 8.954200000000001e-06, "loss": 0.0597, "step": 5230 }, { "epoch": 0.2788825505144746, "grad_norm": 3.753453493118286, "learning_rate": 8.954000000000001e-06, "loss": 0.0512, "step": 5231 }, { "epoch": 0.2789358639441275, "grad_norm": 5.219347953796387, "learning_rate": 8.9538e-06, "loss": 0.0561, "step": 5232 }, { "epoch": 0.27898917737378043, "grad_norm": 5.121862888336182, "learning_rate": 8.953600000000002e-06, "loss": 0.069, "step": 5233 }, { "epoch": 0.2790424908034334, "grad_norm": 4.519096851348877, "learning_rate": 8.953400000000001e-06, "loss": 0.0738, "step": 5234 }, { "epoch": 0.27909580423308633, "grad_norm": 3.3952579498291016, "learning_rate": 8.953200000000001e-06, "loss": 0.04, "step": 5235 }, { "epoch": 0.27914911766273925, "grad_norm": 9.86619758605957, "learning_rate": 8.953e-06, "loss": 0.0686, "step": 5236 }, { "epoch": 0.2792024310923922, "grad_norm": 2.984231948852539, "learning_rate": 8.9528e-06, "loss": 0.054, "step": 5237 }, { "epoch": 0.2792557445220451, "grad_norm": 3.770432472229004, "learning_rate": 8.9526e-06, "loss": 0.0828, "step": 5238 }, { "epoch": 0.279309057951698, "grad_norm": 3.3572657108306885, "learning_rate": 8.952400000000001e-06, "loss": 0.042, "step": 5239 }, { "epoch": 0.27936237138135095, "grad_norm": 3.0545287132263184, "learning_rate": 8.9522e-06, "loss": 0.043, "step": 5240 }, { "epoch": 0.2794156848110039, "grad_norm": 3.4667446613311768, "learning_rate": 8.952e-06, "loss": 0.044, "step": 5241 }, { "epoch": 0.2794689982406568, "grad_norm": 2.368865728378296, "learning_rate": 8.951800000000001e-06, "loss": 0.0442, "step": 5242 }, { "epoch": 0.2795223116703098, "grad_norm": 4.9770636558532715, "learning_rate": 8.9516e-06, "loss": 0.0639, "step": 5243 }, { "epoch": 0.2795756250999627, "grad_norm": 3.152298927307129, "learning_rate": 8.9514e-06, "loss": 0.0296, "step": 5244 }, { "epoch": 0.2796289385296156, "grad_norm": 3.855530023574829, "learning_rate": 8.951200000000002e-06, "loss": 0.0448, "step": 5245 }, { "epoch": 0.27968225195926855, "grad_norm": 2.9807488918304443, "learning_rate": 8.951000000000001e-06, "loss": 0.0541, "step": 5246 }, { "epoch": 0.27973556538892147, "grad_norm": 3.2078723907470703, "learning_rate": 8.9508e-06, "loss": 0.0305, "step": 5247 }, { "epoch": 0.2797888788185744, "grad_norm": 4.36043643951416, "learning_rate": 8.9506e-06, "loss": 0.0462, "step": 5248 }, { "epoch": 0.2798421922482273, "grad_norm": 2.517826795578003, "learning_rate": 8.9504e-06, "loss": 0.0364, "step": 5249 }, { "epoch": 0.27989550567788024, "grad_norm": 3.435502529144287, "learning_rate": 8.950200000000001e-06, "loss": 0.057, "step": 5250 }, { "epoch": 0.27994881910753316, "grad_norm": 3.5878283977508545, "learning_rate": 8.95e-06, "loss": 0.0499, "step": 5251 }, { "epoch": 0.28000213253718614, "grad_norm": 3.7341411113739014, "learning_rate": 8.9498e-06, "loss": 0.0386, "step": 5252 }, { "epoch": 0.28005544596683907, "grad_norm": 2.675262689590454, "learning_rate": 8.9496e-06, "loss": 0.0352, "step": 5253 }, { "epoch": 0.280108759396492, "grad_norm": 4.271298885345459, "learning_rate": 8.949400000000001e-06, "loss": 0.0611, "step": 5254 }, { "epoch": 0.2801620728261449, "grad_norm": 3.120873212814331, "learning_rate": 8.9492e-06, "loss": 0.0472, "step": 5255 }, { "epoch": 0.28021538625579784, "grad_norm": 2.495795726776123, "learning_rate": 8.949e-06, "loss": 0.0259, "step": 5256 }, { "epoch": 0.28026869968545076, "grad_norm": 3.7220358848571777, "learning_rate": 8.948800000000001e-06, "loss": 0.0587, "step": 5257 }, { "epoch": 0.2803220131151037, "grad_norm": 3.3194057941436768, "learning_rate": 8.948600000000001e-06, "loss": 0.0594, "step": 5258 }, { "epoch": 0.2803753265447566, "grad_norm": 3.5418004989624023, "learning_rate": 8.9484e-06, "loss": 0.0414, "step": 5259 }, { "epoch": 0.28042863997440953, "grad_norm": 4.415144443511963, "learning_rate": 8.9482e-06, "loss": 0.029, "step": 5260 }, { "epoch": 0.28048195340406246, "grad_norm": 6.625975608825684, "learning_rate": 8.948000000000001e-06, "loss": 0.0884, "step": 5261 }, { "epoch": 0.28053526683371544, "grad_norm": 1.6851080656051636, "learning_rate": 8.947800000000001e-06, "loss": 0.025, "step": 5262 }, { "epoch": 0.28058858026336836, "grad_norm": 2.9542298316955566, "learning_rate": 8.9476e-06, "loss": 0.0408, "step": 5263 }, { "epoch": 0.2806418936930213, "grad_norm": 6.175050258636475, "learning_rate": 8.9474e-06, "loss": 0.0549, "step": 5264 }, { "epoch": 0.2806952071226742, "grad_norm": 3.4177513122558594, "learning_rate": 8.9472e-06, "loss": 0.0557, "step": 5265 }, { "epoch": 0.28074852055232713, "grad_norm": 4.225581169128418, "learning_rate": 8.947000000000001e-06, "loss": 0.0399, "step": 5266 }, { "epoch": 0.28080183398198005, "grad_norm": 2.5615789890289307, "learning_rate": 8.9468e-06, "loss": 0.0334, "step": 5267 }, { "epoch": 0.280855147411633, "grad_norm": 10.352811813354492, "learning_rate": 8.9466e-06, "loss": 0.1057, "step": 5268 }, { "epoch": 0.2809084608412859, "grad_norm": 2.4854538440704346, "learning_rate": 8.946400000000001e-06, "loss": 0.0362, "step": 5269 }, { "epoch": 0.2809617742709388, "grad_norm": 3.158649444580078, "learning_rate": 8.9462e-06, "loss": 0.0584, "step": 5270 }, { "epoch": 0.2810150877005918, "grad_norm": 2.7630696296691895, "learning_rate": 8.946e-06, "loss": 0.0568, "step": 5271 }, { "epoch": 0.2810684011302447, "grad_norm": 3.695784091949463, "learning_rate": 8.945800000000002e-06, "loss": 0.0385, "step": 5272 }, { "epoch": 0.28112171455989765, "grad_norm": 5.526832580566406, "learning_rate": 8.945600000000001e-06, "loss": 0.055, "step": 5273 }, { "epoch": 0.2811750279895506, "grad_norm": 8.814876556396484, "learning_rate": 8.9454e-06, "loss": 0.0627, "step": 5274 }, { "epoch": 0.2812283414192035, "grad_norm": 3.39255428314209, "learning_rate": 8.9452e-06, "loss": 0.0522, "step": 5275 }, { "epoch": 0.2812816548488564, "grad_norm": 2.8464765548706055, "learning_rate": 8.945e-06, "loss": 0.0481, "step": 5276 }, { "epoch": 0.28133496827850935, "grad_norm": 2.631981134414673, "learning_rate": 8.944800000000001e-06, "loss": 0.0453, "step": 5277 }, { "epoch": 0.28138828170816227, "grad_norm": 3.9303619861602783, "learning_rate": 8.9446e-06, "loss": 0.053, "step": 5278 }, { "epoch": 0.2814415951378152, "grad_norm": 2.596339225769043, "learning_rate": 8.9444e-06, "loss": 0.0323, "step": 5279 }, { "epoch": 0.28149490856746817, "grad_norm": 4.14792013168335, "learning_rate": 8.9442e-06, "loss": 0.035, "step": 5280 }, { "epoch": 0.2815482219971211, "grad_norm": 3.0168893337249756, "learning_rate": 8.944000000000001e-06, "loss": 0.0425, "step": 5281 }, { "epoch": 0.281601535426774, "grad_norm": 3.7774569988250732, "learning_rate": 8.9438e-06, "loss": 0.0433, "step": 5282 }, { "epoch": 0.28165484885642694, "grad_norm": 7.580154895782471, "learning_rate": 8.943600000000002e-06, "loss": 0.0855, "step": 5283 }, { "epoch": 0.28170816228607987, "grad_norm": 5.762319564819336, "learning_rate": 8.943400000000001e-06, "loss": 0.0471, "step": 5284 }, { "epoch": 0.2817614757157328, "grad_norm": 2.722069025039673, "learning_rate": 8.943200000000001e-06, "loss": 0.0392, "step": 5285 }, { "epoch": 0.2818147891453857, "grad_norm": 3.76802921295166, "learning_rate": 8.943e-06, "loss": 0.0576, "step": 5286 }, { "epoch": 0.28186810257503864, "grad_norm": 2.3172543048858643, "learning_rate": 8.9428e-06, "loss": 0.0382, "step": 5287 }, { "epoch": 0.28192141600469156, "grad_norm": 2.699451446533203, "learning_rate": 8.942600000000001e-06, "loss": 0.0392, "step": 5288 }, { "epoch": 0.2819747294343445, "grad_norm": 2.8122966289520264, "learning_rate": 8.942400000000001e-06, "loss": 0.0365, "step": 5289 }, { "epoch": 0.28202804286399746, "grad_norm": 2.727497100830078, "learning_rate": 8.9422e-06, "loss": 0.0415, "step": 5290 }, { "epoch": 0.2820813562936504, "grad_norm": 2.6732075214385986, "learning_rate": 8.942e-06, "loss": 0.0459, "step": 5291 }, { "epoch": 0.2821346697233033, "grad_norm": 3.7587339878082275, "learning_rate": 8.9418e-06, "loss": 0.0356, "step": 5292 }, { "epoch": 0.28218798315295623, "grad_norm": 2.9096217155456543, "learning_rate": 8.941600000000001e-06, "loss": 0.042, "step": 5293 }, { "epoch": 0.28224129658260916, "grad_norm": 3.412428617477417, "learning_rate": 8.9414e-06, "loss": 0.0305, "step": 5294 }, { "epoch": 0.2822946100122621, "grad_norm": 3.3983912467956543, "learning_rate": 8.941200000000002e-06, "loss": 0.0311, "step": 5295 }, { "epoch": 0.282347923441915, "grad_norm": 4.963376522064209, "learning_rate": 8.941000000000001e-06, "loss": 0.114, "step": 5296 }, { "epoch": 0.28240123687156793, "grad_norm": 3.0897271633148193, "learning_rate": 8.940800000000001e-06, "loss": 0.0354, "step": 5297 }, { "epoch": 0.28245455030122085, "grad_norm": 3.679586887359619, "learning_rate": 8.9406e-06, "loss": 0.0435, "step": 5298 }, { "epoch": 0.28250786373087383, "grad_norm": 5.96331787109375, "learning_rate": 8.9404e-06, "loss": 0.056, "step": 5299 }, { "epoch": 0.28256117716052676, "grad_norm": 5.806448936462402, "learning_rate": 8.940200000000001e-06, "loss": 0.0586, "step": 5300 }, { "epoch": 0.2826144905901797, "grad_norm": 2.388753652572632, "learning_rate": 8.94e-06, "loss": 0.0233, "step": 5301 }, { "epoch": 0.2826678040198326, "grad_norm": 2.6755168437957764, "learning_rate": 8.9398e-06, "loss": 0.0308, "step": 5302 }, { "epoch": 0.2827211174494855, "grad_norm": 4.224398612976074, "learning_rate": 8.9396e-06, "loss": 0.0493, "step": 5303 }, { "epoch": 0.28277443087913845, "grad_norm": 4.559255599975586, "learning_rate": 8.9394e-06, "loss": 0.0698, "step": 5304 }, { "epoch": 0.2828277443087914, "grad_norm": 3.5410304069519043, "learning_rate": 8.9392e-06, "loss": 0.0329, "step": 5305 }, { "epoch": 0.2828810577384443, "grad_norm": 2.7685506343841553, "learning_rate": 8.939e-06, "loss": 0.0234, "step": 5306 }, { "epoch": 0.2829343711680972, "grad_norm": 2.779975175857544, "learning_rate": 8.938800000000002e-06, "loss": 0.0357, "step": 5307 }, { "epoch": 0.2829876845977502, "grad_norm": 3.3703184127807617, "learning_rate": 8.938600000000001e-06, "loss": 0.044, "step": 5308 }, { "epoch": 0.2830409980274031, "grad_norm": 3.896184206008911, "learning_rate": 8.9384e-06, "loss": 0.0458, "step": 5309 }, { "epoch": 0.28309431145705605, "grad_norm": 4.398073673248291, "learning_rate": 8.9382e-06, "loss": 0.0579, "step": 5310 }, { "epoch": 0.28314762488670897, "grad_norm": 3.8475759029388428, "learning_rate": 8.938000000000001e-06, "loss": 0.0625, "step": 5311 }, { "epoch": 0.2832009383163619, "grad_norm": 2.6344707012176514, "learning_rate": 8.937800000000001e-06, "loss": 0.0413, "step": 5312 }, { "epoch": 0.2832542517460148, "grad_norm": 1.5869239568710327, "learning_rate": 8.9376e-06, "loss": 0.0167, "step": 5313 }, { "epoch": 0.28330756517566774, "grad_norm": 3.608414888381958, "learning_rate": 8.9374e-06, "loss": 0.0416, "step": 5314 }, { "epoch": 0.28336087860532067, "grad_norm": 4.578672885894775, "learning_rate": 8.9372e-06, "loss": 0.0615, "step": 5315 }, { "epoch": 0.2834141920349736, "grad_norm": 3.4030330181121826, "learning_rate": 8.937000000000001e-06, "loss": 0.0476, "step": 5316 }, { "epoch": 0.2834675054646265, "grad_norm": 3.4436304569244385, "learning_rate": 8.9368e-06, "loss": 0.0669, "step": 5317 }, { "epoch": 0.2835208188942795, "grad_norm": 3.872748613357544, "learning_rate": 8.9366e-06, "loss": 0.0589, "step": 5318 }, { "epoch": 0.2835741323239324, "grad_norm": 3.468071222305298, "learning_rate": 8.936400000000001e-06, "loss": 0.0593, "step": 5319 }, { "epoch": 0.28362744575358534, "grad_norm": 1.7952553033828735, "learning_rate": 8.936200000000001e-06, "loss": 0.0253, "step": 5320 }, { "epoch": 0.28368075918323826, "grad_norm": 2.9844629764556885, "learning_rate": 8.936e-06, "loss": 0.0447, "step": 5321 }, { "epoch": 0.2837340726128912, "grad_norm": 2.4521420001983643, "learning_rate": 8.935800000000002e-06, "loss": 0.0277, "step": 5322 }, { "epoch": 0.2837873860425441, "grad_norm": 4.068515300750732, "learning_rate": 8.935600000000001e-06, "loss": 0.0576, "step": 5323 }, { "epoch": 0.28384069947219703, "grad_norm": 3.6468677520751953, "learning_rate": 8.935400000000001e-06, "loss": 0.0542, "step": 5324 }, { "epoch": 0.28389401290184996, "grad_norm": 1.7338452339172363, "learning_rate": 8.9352e-06, "loss": 0.0226, "step": 5325 }, { "epoch": 0.2839473263315029, "grad_norm": 3.861359119415283, "learning_rate": 8.935e-06, "loss": 0.0659, "step": 5326 }, { "epoch": 0.28400063976115586, "grad_norm": 4.709593772888184, "learning_rate": 8.934800000000001e-06, "loss": 0.0542, "step": 5327 }, { "epoch": 0.2840539531908088, "grad_norm": 2.3076186180114746, "learning_rate": 8.9346e-06, "loss": 0.0495, "step": 5328 }, { "epoch": 0.2841072666204617, "grad_norm": 11.707414627075195, "learning_rate": 8.9344e-06, "loss": 0.1019, "step": 5329 }, { "epoch": 0.28416058005011463, "grad_norm": 3.7152116298675537, "learning_rate": 8.9342e-06, "loss": 0.0287, "step": 5330 }, { "epoch": 0.28421389347976755, "grad_norm": 2.798888921737671, "learning_rate": 8.934000000000001e-06, "loss": 0.0175, "step": 5331 }, { "epoch": 0.2842672069094205, "grad_norm": 6.653524875640869, "learning_rate": 8.9338e-06, "loss": 0.0503, "step": 5332 }, { "epoch": 0.2843205203390734, "grad_norm": 5.651182651519775, "learning_rate": 8.9336e-06, "loss": 0.091, "step": 5333 }, { "epoch": 0.2843738337687263, "grad_norm": 5.044902801513672, "learning_rate": 8.933400000000002e-06, "loss": 0.0628, "step": 5334 }, { "epoch": 0.28442714719837925, "grad_norm": 10.628700256347656, "learning_rate": 8.933200000000001e-06, "loss": 0.0956, "step": 5335 }, { "epoch": 0.28448046062803223, "grad_norm": 3.9917409420013428, "learning_rate": 8.933e-06, "loss": 0.0652, "step": 5336 }, { "epoch": 0.28453377405768515, "grad_norm": 7.4599738121032715, "learning_rate": 8.9328e-06, "loss": 0.0702, "step": 5337 }, { "epoch": 0.2845870874873381, "grad_norm": 10.906377792358398, "learning_rate": 8.932600000000002e-06, "loss": 0.0734, "step": 5338 }, { "epoch": 0.284640400916991, "grad_norm": 7.894202709197998, "learning_rate": 8.932400000000001e-06, "loss": 0.0517, "step": 5339 }, { "epoch": 0.2846937143466439, "grad_norm": 6.4340715408325195, "learning_rate": 8.9322e-06, "loss": 0.0931, "step": 5340 }, { "epoch": 0.28474702777629685, "grad_norm": 3.056766986846924, "learning_rate": 8.932e-06, "loss": 0.0405, "step": 5341 }, { "epoch": 0.28480034120594977, "grad_norm": 3.3187668323516846, "learning_rate": 8.9318e-06, "loss": 0.0597, "step": 5342 }, { "epoch": 0.2848536546356027, "grad_norm": 3.204028606414795, "learning_rate": 8.931600000000001e-06, "loss": 0.0459, "step": 5343 }, { "epoch": 0.2849069680652556, "grad_norm": 2.332758665084839, "learning_rate": 8.9314e-06, "loss": 0.0215, "step": 5344 }, { "epoch": 0.28496028149490854, "grad_norm": 5.480348587036133, "learning_rate": 8.9312e-06, "loss": 0.0299, "step": 5345 }, { "epoch": 0.2850135949245615, "grad_norm": 7.810227394104004, "learning_rate": 8.931000000000001e-06, "loss": 0.0658, "step": 5346 }, { "epoch": 0.28506690835421444, "grad_norm": 2.332225799560547, "learning_rate": 8.930800000000001e-06, "loss": 0.0238, "step": 5347 }, { "epoch": 0.28512022178386737, "grad_norm": 4.964149475097656, "learning_rate": 8.9306e-06, "loss": 0.0827, "step": 5348 }, { "epoch": 0.2851735352135203, "grad_norm": 3.3038060665130615, "learning_rate": 8.9304e-06, "loss": 0.0697, "step": 5349 }, { "epoch": 0.2852268486431732, "grad_norm": 6.015600681304932, "learning_rate": 8.930200000000001e-06, "loss": 0.0345, "step": 5350 }, { "epoch": 0.28528016207282614, "grad_norm": 2.308889627456665, "learning_rate": 8.930000000000001e-06, "loss": 0.026, "step": 5351 }, { "epoch": 0.28533347550247906, "grad_norm": 3.66080379486084, "learning_rate": 8.9298e-06, "loss": 0.0337, "step": 5352 }, { "epoch": 0.285386788932132, "grad_norm": 7.005263328552246, "learning_rate": 8.9296e-06, "loss": 0.0547, "step": 5353 }, { "epoch": 0.2854401023617849, "grad_norm": 3.8293049335479736, "learning_rate": 8.9294e-06, "loss": 0.0421, "step": 5354 }, { "epoch": 0.2854934157914379, "grad_norm": 3.618804454803467, "learning_rate": 8.9292e-06, "loss": 0.0499, "step": 5355 }, { "epoch": 0.2855467292210908, "grad_norm": 3.0527708530426025, "learning_rate": 8.929e-06, "loss": 0.0366, "step": 5356 }, { "epoch": 0.28560004265074374, "grad_norm": 3.014014720916748, "learning_rate": 8.9288e-06, "loss": 0.0332, "step": 5357 }, { "epoch": 0.28565335608039666, "grad_norm": 12.683980941772461, "learning_rate": 8.928600000000001e-06, "loss": 0.0675, "step": 5358 }, { "epoch": 0.2857066695100496, "grad_norm": 2.5976295471191406, "learning_rate": 8.9284e-06, "loss": 0.0238, "step": 5359 }, { "epoch": 0.2857599829397025, "grad_norm": 4.027119159698486, "learning_rate": 8.9282e-06, "loss": 0.0463, "step": 5360 }, { "epoch": 0.28581329636935543, "grad_norm": 2.62845778465271, "learning_rate": 8.928000000000002e-06, "loss": 0.0231, "step": 5361 }, { "epoch": 0.28586660979900835, "grad_norm": 4.017064571380615, "learning_rate": 8.927800000000001e-06, "loss": 0.0463, "step": 5362 }, { "epoch": 0.2859199232286613, "grad_norm": 9.257384300231934, "learning_rate": 8.9276e-06, "loss": 0.0667, "step": 5363 }, { "epoch": 0.28597323665831426, "grad_norm": 9.029706954956055, "learning_rate": 8.9274e-06, "loss": 0.0303, "step": 5364 }, { "epoch": 0.2860265500879672, "grad_norm": 3.2658910751342773, "learning_rate": 8.9272e-06, "loss": 0.03, "step": 5365 }, { "epoch": 0.2860798635176201, "grad_norm": 4.490231513977051, "learning_rate": 8.927000000000001e-06, "loss": 0.0512, "step": 5366 }, { "epoch": 0.286133176947273, "grad_norm": 7.2007036209106445, "learning_rate": 8.9268e-06, "loss": 0.0697, "step": 5367 }, { "epoch": 0.28618649037692595, "grad_norm": 3.1958096027374268, "learning_rate": 8.9266e-06, "loss": 0.0429, "step": 5368 }, { "epoch": 0.2862398038065789, "grad_norm": 2.0347163677215576, "learning_rate": 8.9264e-06, "loss": 0.0272, "step": 5369 }, { "epoch": 0.2862931172362318, "grad_norm": 5.624851703643799, "learning_rate": 8.926200000000001e-06, "loss": 0.0742, "step": 5370 }, { "epoch": 0.2863464306658847, "grad_norm": 2.209841012954712, "learning_rate": 8.926e-06, "loss": 0.0208, "step": 5371 }, { "epoch": 0.28639974409553765, "grad_norm": 3.186380624771118, "learning_rate": 8.925800000000002e-06, "loss": 0.0264, "step": 5372 }, { "epoch": 0.28645305752519057, "grad_norm": 6.293044090270996, "learning_rate": 8.925600000000001e-06, "loss": 0.0459, "step": 5373 }, { "epoch": 0.28650637095484355, "grad_norm": 4.326900959014893, "learning_rate": 8.925400000000001e-06, "loss": 0.0486, "step": 5374 }, { "epoch": 0.28655968438449647, "grad_norm": 3.510275363922119, "learning_rate": 8.9252e-06, "loss": 0.0185, "step": 5375 }, { "epoch": 0.2866129978141494, "grad_norm": 4.283905506134033, "learning_rate": 8.925e-06, "loss": 0.0634, "step": 5376 }, { "epoch": 0.2866663112438023, "grad_norm": 3.384751558303833, "learning_rate": 8.924800000000001e-06, "loss": 0.0187, "step": 5377 }, { "epoch": 0.28671962467345524, "grad_norm": 4.393059730529785, "learning_rate": 8.924600000000001e-06, "loss": 0.0579, "step": 5378 }, { "epoch": 0.28677293810310817, "grad_norm": 3.6582579612731934, "learning_rate": 8.9244e-06, "loss": 0.046, "step": 5379 }, { "epoch": 0.2868262515327611, "grad_norm": 6.764665603637695, "learning_rate": 8.9242e-06, "loss": 0.0352, "step": 5380 }, { "epoch": 0.286879564962414, "grad_norm": 7.124831676483154, "learning_rate": 8.924e-06, "loss": 0.0319, "step": 5381 }, { "epoch": 0.28693287839206694, "grad_norm": 4.893610954284668, "learning_rate": 8.923800000000001e-06, "loss": 0.028, "step": 5382 }, { "epoch": 0.2869861918217199, "grad_norm": 2.4732041358947754, "learning_rate": 8.9236e-06, "loss": 0.0222, "step": 5383 }, { "epoch": 0.28703950525137284, "grad_norm": 6.552478790283203, "learning_rate": 8.923400000000002e-06, "loss": 0.0779, "step": 5384 }, { "epoch": 0.28709281868102576, "grad_norm": 5.061934947967529, "learning_rate": 8.923200000000001e-06, "loss": 0.0467, "step": 5385 }, { "epoch": 0.2871461321106787, "grad_norm": 6.230727672576904, "learning_rate": 8.923e-06, "loss": 0.0606, "step": 5386 }, { "epoch": 0.2871994455403316, "grad_norm": 3.4471302032470703, "learning_rate": 8.9228e-06, "loss": 0.0579, "step": 5387 }, { "epoch": 0.28725275896998453, "grad_norm": 6.31633186340332, "learning_rate": 8.922600000000002e-06, "loss": 0.0551, "step": 5388 }, { "epoch": 0.28730607239963746, "grad_norm": 4.450409412384033, "learning_rate": 8.922400000000001e-06, "loss": 0.0376, "step": 5389 }, { "epoch": 0.2873593858292904, "grad_norm": 2.255420207977295, "learning_rate": 8.9222e-06, "loss": 0.0267, "step": 5390 }, { "epoch": 0.2874126992589433, "grad_norm": 3.1579108238220215, "learning_rate": 8.922e-06, "loss": 0.0292, "step": 5391 }, { "epoch": 0.2874660126885963, "grad_norm": 3.515782594680786, "learning_rate": 8.9218e-06, "loss": 0.0308, "step": 5392 }, { "epoch": 0.2875193261182492, "grad_norm": 5.166193962097168, "learning_rate": 8.9216e-06, "loss": 0.0714, "step": 5393 }, { "epoch": 0.28757263954790213, "grad_norm": 3.3076043128967285, "learning_rate": 8.9214e-06, "loss": 0.0292, "step": 5394 }, { "epoch": 0.28762595297755505, "grad_norm": 3.7686493396759033, "learning_rate": 8.9212e-06, "loss": 0.0602, "step": 5395 }, { "epoch": 0.287679266407208, "grad_norm": 2.2083089351654053, "learning_rate": 8.921000000000001e-06, "loss": 0.0265, "step": 5396 }, { "epoch": 0.2877325798368609, "grad_norm": 3.9681928157806396, "learning_rate": 8.920800000000001e-06, "loss": 0.0595, "step": 5397 }, { "epoch": 0.2877858932665138, "grad_norm": 3.088772773742676, "learning_rate": 8.9206e-06, "loss": 0.0536, "step": 5398 }, { "epoch": 0.28783920669616675, "grad_norm": 3.8727192878723145, "learning_rate": 8.920400000000002e-06, "loss": 0.0453, "step": 5399 }, { "epoch": 0.2878925201258197, "grad_norm": 2.177232265472412, "learning_rate": 8.920200000000001e-06, "loss": 0.0207, "step": 5400 }, { "epoch": 0.2879458335554726, "grad_norm": 3.3341822624206543, "learning_rate": 8.920000000000001e-06, "loss": 0.0546, "step": 5401 }, { "epoch": 0.2879991469851256, "grad_norm": 6.371274948120117, "learning_rate": 8.9198e-06, "loss": 0.0477, "step": 5402 }, { "epoch": 0.2880524604147785, "grad_norm": 3.2304577827453613, "learning_rate": 8.9196e-06, "loss": 0.0657, "step": 5403 }, { "epoch": 0.2881057738444314, "grad_norm": 2.0966312885284424, "learning_rate": 8.9194e-06, "loss": 0.0327, "step": 5404 }, { "epoch": 0.28815908727408435, "grad_norm": 3.9487195014953613, "learning_rate": 8.919200000000001e-06, "loss": 0.0362, "step": 5405 }, { "epoch": 0.28821240070373727, "grad_norm": 2.3235604763031006, "learning_rate": 8.919e-06, "loss": 0.0618, "step": 5406 }, { "epoch": 0.2882657141333902, "grad_norm": 4.5306477546691895, "learning_rate": 8.9188e-06, "loss": 0.0562, "step": 5407 }, { "epoch": 0.2883190275630431, "grad_norm": 3.253701686859131, "learning_rate": 8.918600000000001e-06, "loss": 0.0301, "step": 5408 }, { "epoch": 0.28837234099269604, "grad_norm": 3.357527732849121, "learning_rate": 8.918400000000001e-06, "loss": 0.037, "step": 5409 }, { "epoch": 0.28842565442234896, "grad_norm": 2.812370777130127, "learning_rate": 8.9182e-06, "loss": 0.0356, "step": 5410 }, { "epoch": 0.28847896785200194, "grad_norm": 2.281675338745117, "learning_rate": 8.918000000000002e-06, "loss": 0.0265, "step": 5411 }, { "epoch": 0.28853228128165487, "grad_norm": 5.138279438018799, "learning_rate": 8.917800000000001e-06, "loss": 0.0744, "step": 5412 }, { "epoch": 0.2885855947113078, "grad_norm": 4.140498161315918, "learning_rate": 8.9176e-06, "loss": 0.0335, "step": 5413 }, { "epoch": 0.2886389081409607, "grad_norm": 4.218619346618652, "learning_rate": 8.9174e-06, "loss": 0.0646, "step": 5414 }, { "epoch": 0.28869222157061364, "grad_norm": 2.4908366203308105, "learning_rate": 8.9172e-06, "loss": 0.0179, "step": 5415 }, { "epoch": 0.28874553500026656, "grad_norm": 2.826230049133301, "learning_rate": 8.917000000000001e-06, "loss": 0.0696, "step": 5416 }, { "epoch": 0.2887988484299195, "grad_norm": 1.7370069026947021, "learning_rate": 8.9168e-06, "loss": 0.0262, "step": 5417 }, { "epoch": 0.2888521618595724, "grad_norm": 2.8277790546417236, "learning_rate": 8.9166e-06, "loss": 0.0522, "step": 5418 }, { "epoch": 0.28890547528922533, "grad_norm": 2.9401369094848633, "learning_rate": 8.9164e-06, "loss": 0.0446, "step": 5419 }, { "epoch": 0.2889587887188783, "grad_norm": 3.128035545349121, "learning_rate": 8.916200000000001e-06, "loss": 0.0367, "step": 5420 }, { "epoch": 0.28901210214853124, "grad_norm": 3.7752528190612793, "learning_rate": 8.916e-06, "loss": 0.0504, "step": 5421 }, { "epoch": 0.28906541557818416, "grad_norm": 2.399139881134033, "learning_rate": 8.915800000000002e-06, "loss": 0.0231, "step": 5422 }, { "epoch": 0.2891187290078371, "grad_norm": 2.3697686195373535, "learning_rate": 8.915600000000002e-06, "loss": 0.0355, "step": 5423 }, { "epoch": 0.28917204243749, "grad_norm": 4.7613115310668945, "learning_rate": 8.915400000000001e-06, "loss": 0.0408, "step": 5424 }, { "epoch": 0.28922535586714293, "grad_norm": 4.212809085845947, "learning_rate": 8.9152e-06, "loss": 0.0675, "step": 5425 }, { "epoch": 0.28927866929679585, "grad_norm": 1.95116126537323, "learning_rate": 8.915e-06, "loss": 0.0321, "step": 5426 }, { "epoch": 0.2893319827264488, "grad_norm": 1.8051303625106812, "learning_rate": 8.914800000000001e-06, "loss": 0.0157, "step": 5427 }, { "epoch": 0.2893852961561017, "grad_norm": 3.7577226161956787, "learning_rate": 8.914600000000001e-06, "loss": 0.0325, "step": 5428 }, { "epoch": 0.2894386095857546, "grad_norm": 3.8953239917755127, "learning_rate": 8.9144e-06, "loss": 0.0641, "step": 5429 }, { "epoch": 0.2894919230154076, "grad_norm": 3.0009284019470215, "learning_rate": 8.9142e-06, "loss": 0.0453, "step": 5430 }, { "epoch": 0.2895452364450605, "grad_norm": 7.195056915283203, "learning_rate": 8.914e-06, "loss": 0.1013, "step": 5431 }, { "epoch": 0.28959854987471345, "grad_norm": 4.064537048339844, "learning_rate": 8.913800000000001e-06, "loss": 0.04, "step": 5432 }, { "epoch": 0.2896518633043664, "grad_norm": 1.891005277633667, "learning_rate": 8.9136e-06, "loss": 0.0254, "step": 5433 }, { "epoch": 0.2897051767340193, "grad_norm": 3.1983306407928467, "learning_rate": 8.913400000000002e-06, "loss": 0.0257, "step": 5434 }, { "epoch": 0.2897584901636722, "grad_norm": 3.324730634689331, "learning_rate": 8.913200000000001e-06, "loss": 0.0282, "step": 5435 }, { "epoch": 0.28981180359332515, "grad_norm": 3.4106388092041016, "learning_rate": 8.913000000000001e-06, "loss": 0.0422, "step": 5436 }, { "epoch": 0.28986511702297807, "grad_norm": 4.8930840492248535, "learning_rate": 8.9128e-06, "loss": 0.0777, "step": 5437 }, { "epoch": 0.289918430452631, "grad_norm": 5.267186641693115, "learning_rate": 8.912600000000002e-06, "loss": 0.061, "step": 5438 }, { "epoch": 0.28997174388228397, "grad_norm": 5.67596960067749, "learning_rate": 8.912400000000001e-06, "loss": 0.0742, "step": 5439 }, { "epoch": 0.2900250573119369, "grad_norm": 3.5086493492126465, "learning_rate": 8.9122e-06, "loss": 0.0613, "step": 5440 }, { "epoch": 0.2900783707415898, "grad_norm": 4.091922760009766, "learning_rate": 8.912e-06, "loss": 0.0595, "step": 5441 }, { "epoch": 0.29013168417124274, "grad_norm": 3.8612194061279297, "learning_rate": 8.9118e-06, "loss": 0.0314, "step": 5442 }, { "epoch": 0.29018499760089567, "grad_norm": 7.3807196617126465, "learning_rate": 8.9116e-06, "loss": 0.073, "step": 5443 }, { "epoch": 0.2902383110305486, "grad_norm": 3.294379472732544, "learning_rate": 8.9114e-06, "loss": 0.0493, "step": 5444 }, { "epoch": 0.2902916244602015, "grad_norm": 5.457437992095947, "learning_rate": 8.9112e-06, "loss": 0.0542, "step": 5445 }, { "epoch": 0.29034493788985444, "grad_norm": 4.824381351470947, "learning_rate": 8.911000000000002e-06, "loss": 0.0607, "step": 5446 }, { "epoch": 0.29039825131950736, "grad_norm": 2.883530616760254, "learning_rate": 8.910800000000001e-06, "loss": 0.067, "step": 5447 }, { "epoch": 0.29045156474916034, "grad_norm": 3.800142526626587, "learning_rate": 8.9106e-06, "loss": 0.0509, "step": 5448 }, { "epoch": 0.29050487817881326, "grad_norm": 5.153197765350342, "learning_rate": 8.910400000000002e-06, "loss": 0.0704, "step": 5449 }, { "epoch": 0.2905581916084662, "grad_norm": 4.059353828430176, "learning_rate": 8.910200000000002e-06, "loss": 0.07, "step": 5450 }, { "epoch": 0.2906115050381191, "grad_norm": 3.0049846172332764, "learning_rate": 8.910000000000001e-06, "loss": 0.0361, "step": 5451 }, { "epoch": 0.29066481846777203, "grad_norm": 4.231979846954346, "learning_rate": 8.9098e-06, "loss": 0.0681, "step": 5452 }, { "epoch": 0.29071813189742496, "grad_norm": 4.766856670379639, "learning_rate": 8.9096e-06, "loss": 0.0719, "step": 5453 }, { "epoch": 0.2907714453270779, "grad_norm": 2.9432590007781982, "learning_rate": 8.9094e-06, "loss": 0.04, "step": 5454 }, { "epoch": 0.2908247587567308, "grad_norm": 3.023799419403076, "learning_rate": 8.909200000000001e-06, "loss": 0.0585, "step": 5455 }, { "epoch": 0.29087807218638373, "grad_norm": 4.6854777336120605, "learning_rate": 8.909e-06, "loss": 0.0623, "step": 5456 }, { "epoch": 0.29093138561603665, "grad_norm": 4.130236625671387, "learning_rate": 8.9088e-06, "loss": 0.066, "step": 5457 }, { "epoch": 0.29098469904568963, "grad_norm": 2.1163084506988525, "learning_rate": 8.908600000000001e-06, "loss": 0.0473, "step": 5458 }, { "epoch": 0.29103801247534256, "grad_norm": 2.6777353286743164, "learning_rate": 8.908400000000001e-06, "loss": 0.0597, "step": 5459 }, { "epoch": 0.2910913259049955, "grad_norm": 3.372404098510742, "learning_rate": 8.9082e-06, "loss": 0.0517, "step": 5460 }, { "epoch": 0.2911446393346484, "grad_norm": 3.289771318435669, "learning_rate": 8.908000000000002e-06, "loss": 0.0514, "step": 5461 }, { "epoch": 0.2911979527643013, "grad_norm": 1.9008314609527588, "learning_rate": 8.907800000000001e-06, "loss": 0.0338, "step": 5462 }, { "epoch": 0.29125126619395425, "grad_norm": 4.823404788970947, "learning_rate": 8.907600000000001e-06, "loss": 0.075, "step": 5463 }, { "epoch": 0.2913045796236072, "grad_norm": 1.5792887210845947, "learning_rate": 8.9074e-06, "loss": 0.0147, "step": 5464 }, { "epoch": 0.2913578930532601, "grad_norm": 3.8084030151367188, "learning_rate": 8.9072e-06, "loss": 0.05, "step": 5465 }, { "epoch": 0.291411206482913, "grad_norm": 4.106741905212402, "learning_rate": 8.907000000000001e-06, "loss": 0.0685, "step": 5466 }, { "epoch": 0.291464519912566, "grad_norm": 3.3801987171173096, "learning_rate": 8.906800000000001e-06, "loss": 0.045, "step": 5467 }, { "epoch": 0.2915178333422189, "grad_norm": 4.4174699783325195, "learning_rate": 8.9066e-06, "loss": 0.038, "step": 5468 }, { "epoch": 0.29157114677187185, "grad_norm": 2.422309398651123, "learning_rate": 8.9064e-06, "loss": 0.0483, "step": 5469 }, { "epoch": 0.29162446020152477, "grad_norm": 2.847241163253784, "learning_rate": 8.906200000000001e-06, "loss": 0.0238, "step": 5470 }, { "epoch": 0.2916777736311777, "grad_norm": 4.038010597229004, "learning_rate": 8.906e-06, "loss": 0.0321, "step": 5471 }, { "epoch": 0.2917310870608306, "grad_norm": 4.612678527832031, "learning_rate": 8.9058e-06, "loss": 0.0607, "step": 5472 }, { "epoch": 0.29178440049048354, "grad_norm": 4.58881139755249, "learning_rate": 8.905600000000002e-06, "loss": 0.0841, "step": 5473 }, { "epoch": 0.29183771392013647, "grad_norm": 2.796992063522339, "learning_rate": 8.905400000000001e-06, "loss": 0.0441, "step": 5474 }, { "epoch": 0.2918910273497894, "grad_norm": 4.243381023406982, "learning_rate": 8.9052e-06, "loss": 0.0743, "step": 5475 }, { "epoch": 0.29194434077944237, "grad_norm": 2.9380574226379395, "learning_rate": 8.905e-06, "loss": 0.0378, "step": 5476 }, { "epoch": 0.2919976542090953, "grad_norm": 2.1147773265838623, "learning_rate": 8.904800000000002e-06, "loss": 0.0312, "step": 5477 }, { "epoch": 0.2920509676387482, "grad_norm": 1.5457831621170044, "learning_rate": 8.904600000000001e-06, "loss": 0.028, "step": 5478 }, { "epoch": 0.29210428106840114, "grad_norm": 1.3535575866699219, "learning_rate": 8.9044e-06, "loss": 0.0242, "step": 5479 }, { "epoch": 0.29215759449805406, "grad_norm": 2.0749967098236084, "learning_rate": 8.9042e-06, "loss": 0.028, "step": 5480 }, { "epoch": 0.292210907927707, "grad_norm": 3.037935495376587, "learning_rate": 8.904e-06, "loss": 0.038, "step": 5481 }, { "epoch": 0.2922642213573599, "grad_norm": 3.521991729736328, "learning_rate": 8.903800000000001e-06, "loss": 0.0712, "step": 5482 }, { "epoch": 0.29231753478701283, "grad_norm": 2.0661983489990234, "learning_rate": 8.9036e-06, "loss": 0.0295, "step": 5483 }, { "epoch": 0.29237084821666576, "grad_norm": 4.169869422912598, "learning_rate": 8.9034e-06, "loss": 0.0397, "step": 5484 }, { "epoch": 0.2924241616463187, "grad_norm": 3.6364948749542236, "learning_rate": 8.903200000000001e-06, "loss": 0.0523, "step": 5485 }, { "epoch": 0.29247747507597166, "grad_norm": 3.5818612575531006, "learning_rate": 8.903000000000001e-06, "loss": 0.0368, "step": 5486 }, { "epoch": 0.2925307885056246, "grad_norm": 3.946856737136841, "learning_rate": 8.9028e-06, "loss": 0.0472, "step": 5487 }, { "epoch": 0.2925841019352775, "grad_norm": 5.237720489501953, "learning_rate": 8.902600000000002e-06, "loss": 0.0775, "step": 5488 }, { "epoch": 0.29263741536493043, "grad_norm": 3.739894390106201, "learning_rate": 8.902400000000001e-06, "loss": 0.0741, "step": 5489 }, { "epoch": 0.29269072879458335, "grad_norm": 2.4268603324890137, "learning_rate": 8.902200000000001e-06, "loss": 0.0338, "step": 5490 }, { "epoch": 0.2927440422242363, "grad_norm": 1.7539831399917603, "learning_rate": 8.902e-06, "loss": 0.0125, "step": 5491 }, { "epoch": 0.2927973556538892, "grad_norm": 2.817394971847534, "learning_rate": 8.9018e-06, "loss": 0.0515, "step": 5492 }, { "epoch": 0.2928506690835421, "grad_norm": 3.580901622772217, "learning_rate": 8.9016e-06, "loss": 0.051, "step": 5493 }, { "epoch": 0.29290398251319505, "grad_norm": 3.7681665420532227, "learning_rate": 8.901400000000001e-06, "loss": 0.0277, "step": 5494 }, { "epoch": 0.29295729594284803, "grad_norm": 1.5850573778152466, "learning_rate": 8.9012e-06, "loss": 0.0166, "step": 5495 }, { "epoch": 0.29301060937250095, "grad_norm": 3.7369720935821533, "learning_rate": 8.901e-06, "loss": 0.0375, "step": 5496 }, { "epoch": 0.2930639228021539, "grad_norm": 2.2407002449035645, "learning_rate": 8.900800000000001e-06, "loss": 0.0311, "step": 5497 }, { "epoch": 0.2931172362318068, "grad_norm": 2.0645599365234375, "learning_rate": 8.9006e-06, "loss": 0.0355, "step": 5498 }, { "epoch": 0.2931705496614597, "grad_norm": 3.2192254066467285, "learning_rate": 8.900400000000002e-06, "loss": 0.0633, "step": 5499 }, { "epoch": 0.29322386309111265, "grad_norm": 3.373002290725708, "learning_rate": 8.900200000000002e-06, "loss": 0.0792, "step": 5500 }, { "epoch": 0.29327717652076557, "grad_norm": 1.991923213005066, "learning_rate": 8.900000000000001e-06, "loss": 0.0216, "step": 5501 }, { "epoch": 0.2933304899504185, "grad_norm": 1.927420973777771, "learning_rate": 8.8998e-06, "loss": 0.0378, "step": 5502 }, { "epoch": 0.2933838033800714, "grad_norm": 3.0035312175750732, "learning_rate": 8.8996e-06, "loss": 0.0564, "step": 5503 }, { "epoch": 0.2934371168097244, "grad_norm": 1.797820806503296, "learning_rate": 8.8994e-06, "loss": 0.0202, "step": 5504 }, { "epoch": 0.2934904302393773, "grad_norm": 4.419416427612305, "learning_rate": 8.899200000000001e-06, "loss": 0.046, "step": 5505 }, { "epoch": 0.29354374366903024, "grad_norm": 3.047152519226074, "learning_rate": 8.899e-06, "loss": 0.0244, "step": 5506 }, { "epoch": 0.29359705709868317, "grad_norm": 2.4869608879089355, "learning_rate": 8.8988e-06, "loss": 0.0223, "step": 5507 }, { "epoch": 0.2936503705283361, "grad_norm": 3.59682559967041, "learning_rate": 8.8986e-06, "loss": 0.0351, "step": 5508 }, { "epoch": 0.293703683957989, "grad_norm": 3.640902042388916, "learning_rate": 8.898400000000001e-06, "loss": 0.0475, "step": 5509 }, { "epoch": 0.29375699738764194, "grad_norm": 4.017417907714844, "learning_rate": 8.8982e-06, "loss": 0.0485, "step": 5510 }, { "epoch": 0.29381031081729486, "grad_norm": 3.5288453102111816, "learning_rate": 8.898000000000002e-06, "loss": 0.0393, "step": 5511 }, { "epoch": 0.2938636242469478, "grad_norm": 2.1140854358673096, "learning_rate": 8.897800000000001e-06, "loss": 0.0346, "step": 5512 }, { "epoch": 0.2939169376766007, "grad_norm": 4.210577964782715, "learning_rate": 8.897600000000001e-06, "loss": 0.0583, "step": 5513 }, { "epoch": 0.2939702511062537, "grad_norm": 2.9983065128326416, "learning_rate": 8.8974e-06, "loss": 0.0479, "step": 5514 }, { "epoch": 0.2940235645359066, "grad_norm": 4.3928608894348145, "learning_rate": 8.8972e-06, "loss": 0.0546, "step": 5515 }, { "epoch": 0.29407687796555954, "grad_norm": 5.023612976074219, "learning_rate": 8.897000000000001e-06, "loss": 0.046, "step": 5516 }, { "epoch": 0.29413019139521246, "grad_norm": 2.8349602222442627, "learning_rate": 8.896800000000001e-06, "loss": 0.0244, "step": 5517 }, { "epoch": 0.2941835048248654, "grad_norm": 5.297980785369873, "learning_rate": 8.8966e-06, "loss": 0.029, "step": 5518 }, { "epoch": 0.2942368182545183, "grad_norm": 2.9117050170898438, "learning_rate": 8.8964e-06, "loss": 0.0294, "step": 5519 }, { "epoch": 0.29429013168417123, "grad_norm": 6.551631927490234, "learning_rate": 8.8962e-06, "loss": 0.0766, "step": 5520 }, { "epoch": 0.29434344511382415, "grad_norm": 5.282371997833252, "learning_rate": 8.896000000000001e-06, "loss": 0.11, "step": 5521 }, { "epoch": 0.2943967585434771, "grad_norm": 7.540379047393799, "learning_rate": 8.8958e-06, "loss": 0.0527, "step": 5522 }, { "epoch": 0.29445007197313006, "grad_norm": 2.3936703205108643, "learning_rate": 8.895600000000002e-06, "loss": 0.0227, "step": 5523 }, { "epoch": 0.294503385402783, "grad_norm": 2.0153141021728516, "learning_rate": 8.895400000000001e-06, "loss": 0.0232, "step": 5524 }, { "epoch": 0.2945566988324359, "grad_norm": 5.536219120025635, "learning_rate": 8.8952e-06, "loss": 0.0358, "step": 5525 }, { "epoch": 0.2946100122620888, "grad_norm": 3.984910726547241, "learning_rate": 8.895e-06, "loss": 0.0496, "step": 5526 }, { "epoch": 0.29466332569174175, "grad_norm": 4.636905670166016, "learning_rate": 8.894800000000002e-06, "loss": 0.0434, "step": 5527 }, { "epoch": 0.2947166391213947, "grad_norm": 2.6841721534729004, "learning_rate": 8.894600000000001e-06, "loss": 0.0386, "step": 5528 }, { "epoch": 0.2947699525510476, "grad_norm": 3.1111297607421875, "learning_rate": 8.8944e-06, "loss": 0.0379, "step": 5529 }, { "epoch": 0.2948232659807005, "grad_norm": 5.34640645980835, "learning_rate": 8.8942e-06, "loss": 0.0496, "step": 5530 }, { "epoch": 0.29487657941035345, "grad_norm": 4.67938756942749, "learning_rate": 8.894e-06, "loss": 0.0466, "step": 5531 }, { "epoch": 0.2949298928400064, "grad_norm": 2.732764720916748, "learning_rate": 8.8938e-06, "loss": 0.0354, "step": 5532 }, { "epoch": 0.29498320626965935, "grad_norm": 2.966172218322754, "learning_rate": 8.8936e-06, "loss": 0.0399, "step": 5533 }, { "epoch": 0.29503651969931227, "grad_norm": 2.994596481323242, "learning_rate": 8.8934e-06, "loss": 0.022, "step": 5534 }, { "epoch": 0.2950898331289652, "grad_norm": 2.795454978942871, "learning_rate": 8.893200000000002e-06, "loss": 0.023, "step": 5535 }, { "epoch": 0.2951431465586181, "grad_norm": 7.966578006744385, "learning_rate": 8.893000000000001e-06, "loss": 0.0542, "step": 5536 }, { "epoch": 0.29519645998827104, "grad_norm": 2.3685085773468018, "learning_rate": 8.8928e-06, "loss": 0.0365, "step": 5537 }, { "epoch": 0.29524977341792397, "grad_norm": 2.721726894378662, "learning_rate": 8.892600000000002e-06, "loss": 0.0621, "step": 5538 }, { "epoch": 0.2953030868475769, "grad_norm": 2.6955742835998535, "learning_rate": 8.892400000000001e-06, "loss": 0.0396, "step": 5539 }, { "epoch": 0.2953564002772298, "grad_norm": 3.574896812438965, "learning_rate": 8.892200000000001e-06, "loss": 0.0323, "step": 5540 }, { "epoch": 0.29540971370688274, "grad_norm": 2.788459300994873, "learning_rate": 8.892e-06, "loss": 0.022, "step": 5541 }, { "epoch": 0.2954630271365357, "grad_norm": 4.391505718231201, "learning_rate": 8.8918e-06, "loss": 0.0419, "step": 5542 }, { "epoch": 0.29551634056618864, "grad_norm": 5.70393705368042, "learning_rate": 8.8916e-06, "loss": 0.0677, "step": 5543 }, { "epoch": 0.29556965399584156, "grad_norm": 3.2925970554351807, "learning_rate": 8.891400000000001e-06, "loss": 0.0168, "step": 5544 }, { "epoch": 0.2956229674254945, "grad_norm": 5.558798789978027, "learning_rate": 8.8912e-06, "loss": 0.0732, "step": 5545 }, { "epoch": 0.2956762808551474, "grad_norm": 1.8819236755371094, "learning_rate": 8.891e-06, "loss": 0.03, "step": 5546 }, { "epoch": 0.29572959428480033, "grad_norm": 2.149977445602417, "learning_rate": 8.890800000000001e-06, "loss": 0.0233, "step": 5547 }, { "epoch": 0.29578290771445326, "grad_norm": 2.6398544311523438, "learning_rate": 8.890600000000001e-06, "loss": 0.0454, "step": 5548 }, { "epoch": 0.2958362211441062, "grad_norm": 1.4063647985458374, "learning_rate": 8.8904e-06, "loss": 0.0104, "step": 5549 }, { "epoch": 0.2958895345737591, "grad_norm": 2.9799439907073975, "learning_rate": 8.890200000000002e-06, "loss": 0.0478, "step": 5550 }, { "epoch": 0.2959428480034121, "grad_norm": 3.2995071411132812, "learning_rate": 8.890000000000001e-06, "loss": 0.0291, "step": 5551 }, { "epoch": 0.295996161433065, "grad_norm": 3.84957218170166, "learning_rate": 8.8898e-06, "loss": 0.0242, "step": 5552 }, { "epoch": 0.29604947486271793, "grad_norm": 3.624598264694214, "learning_rate": 8.8896e-06, "loss": 0.0485, "step": 5553 }, { "epoch": 0.29610278829237086, "grad_norm": 3.588499069213867, "learning_rate": 8.8894e-06, "loss": 0.0451, "step": 5554 }, { "epoch": 0.2961561017220238, "grad_norm": 3.2730867862701416, "learning_rate": 8.889200000000001e-06, "loss": 0.0275, "step": 5555 }, { "epoch": 0.2962094151516767, "grad_norm": 3.300067663192749, "learning_rate": 8.889e-06, "loss": 0.026, "step": 5556 }, { "epoch": 0.2962627285813296, "grad_norm": 3.3007123470306396, "learning_rate": 8.8888e-06, "loss": 0.0364, "step": 5557 }, { "epoch": 0.29631604201098255, "grad_norm": 4.554710865020752, "learning_rate": 8.8886e-06, "loss": 0.0866, "step": 5558 }, { "epoch": 0.2963693554406355, "grad_norm": 4.975395679473877, "learning_rate": 8.888400000000001e-06, "loss": 0.0397, "step": 5559 }, { "epoch": 0.29642266887028845, "grad_norm": 3.782303810119629, "learning_rate": 8.8882e-06, "loss": 0.0468, "step": 5560 }, { "epoch": 0.2964759822999414, "grad_norm": 3.759066343307495, "learning_rate": 8.888e-06, "loss": 0.0547, "step": 5561 }, { "epoch": 0.2965292957295943, "grad_norm": 10.552390098571777, "learning_rate": 8.887800000000002e-06, "loss": 0.0428, "step": 5562 }, { "epoch": 0.2965826091592472, "grad_norm": 4.337838172912598, "learning_rate": 8.887600000000001e-06, "loss": 0.046, "step": 5563 }, { "epoch": 0.29663592258890015, "grad_norm": 2.999290943145752, "learning_rate": 8.8874e-06, "loss": 0.0308, "step": 5564 }, { "epoch": 0.29668923601855307, "grad_norm": 2.9879353046417236, "learning_rate": 8.8872e-06, "loss": 0.042, "step": 5565 }, { "epoch": 0.296742549448206, "grad_norm": 4.193004608154297, "learning_rate": 8.887000000000001e-06, "loss": 0.027, "step": 5566 }, { "epoch": 0.2967958628778589, "grad_norm": 6.804745197296143, "learning_rate": 8.886800000000001e-06, "loss": 0.0377, "step": 5567 }, { "epoch": 0.29684917630751184, "grad_norm": 3.554372549057007, "learning_rate": 8.8866e-06, "loss": 0.047, "step": 5568 }, { "epoch": 0.29690248973716477, "grad_norm": 3.7070624828338623, "learning_rate": 8.8864e-06, "loss": 0.0282, "step": 5569 }, { "epoch": 0.29695580316681774, "grad_norm": 5.204434871673584, "learning_rate": 8.8862e-06, "loss": 0.0768, "step": 5570 }, { "epoch": 0.29700911659647067, "grad_norm": 4.345863342285156, "learning_rate": 8.886000000000001e-06, "loss": 0.0664, "step": 5571 }, { "epoch": 0.2970624300261236, "grad_norm": 9.248699188232422, "learning_rate": 8.8858e-06, "loss": 0.0759, "step": 5572 }, { "epoch": 0.2971157434557765, "grad_norm": 2.998258590698242, "learning_rate": 8.8856e-06, "loss": 0.0721, "step": 5573 }, { "epoch": 0.29716905688542944, "grad_norm": 4.534618377685547, "learning_rate": 8.885400000000001e-06, "loss": 0.0431, "step": 5574 }, { "epoch": 0.29722237031508236, "grad_norm": 2.0492184162139893, "learning_rate": 8.885200000000001e-06, "loss": 0.0276, "step": 5575 }, { "epoch": 0.2972756837447353, "grad_norm": 4.793741226196289, "learning_rate": 8.885e-06, "loss": 0.0464, "step": 5576 }, { "epoch": 0.2973289971743882, "grad_norm": 4.900995254516602, "learning_rate": 8.884800000000002e-06, "loss": 0.0571, "step": 5577 }, { "epoch": 0.29738231060404113, "grad_norm": 4.0065693855285645, "learning_rate": 8.884600000000001e-06, "loss": 0.0395, "step": 5578 }, { "epoch": 0.2974356240336941, "grad_norm": 10.170304298400879, "learning_rate": 8.884400000000001e-06, "loss": 0.0501, "step": 5579 }, { "epoch": 0.29748893746334704, "grad_norm": 4.893664360046387, "learning_rate": 8.8842e-06, "loss": 0.0617, "step": 5580 }, { "epoch": 0.29754225089299996, "grad_norm": 3.386892557144165, "learning_rate": 8.884e-06, "loss": 0.0415, "step": 5581 }, { "epoch": 0.2975955643226529, "grad_norm": 2.3417015075683594, "learning_rate": 8.8838e-06, "loss": 0.0595, "step": 5582 }, { "epoch": 0.2976488777523058, "grad_norm": 2.8493545055389404, "learning_rate": 8.8836e-06, "loss": 0.0438, "step": 5583 }, { "epoch": 0.29770219118195873, "grad_norm": 3.2388665676116943, "learning_rate": 8.8834e-06, "loss": 0.0341, "step": 5584 }, { "epoch": 0.29775550461161165, "grad_norm": 7.745791912078857, "learning_rate": 8.8832e-06, "loss": 0.0721, "step": 5585 }, { "epoch": 0.2978088180412646, "grad_norm": 3.2132837772369385, "learning_rate": 8.883000000000001e-06, "loss": 0.0261, "step": 5586 }, { "epoch": 0.2978621314709175, "grad_norm": 3.452157497406006, "learning_rate": 8.8828e-06, "loss": 0.0516, "step": 5587 }, { "epoch": 0.2979154449005705, "grad_norm": 3.923637628555298, "learning_rate": 8.882600000000002e-06, "loss": 0.0519, "step": 5588 }, { "epoch": 0.2979687583302234, "grad_norm": 6.141470909118652, "learning_rate": 8.882400000000002e-06, "loss": 0.0626, "step": 5589 }, { "epoch": 0.29802207175987633, "grad_norm": 3.551281213760376, "learning_rate": 8.882200000000001e-06, "loss": 0.0522, "step": 5590 }, { "epoch": 0.29807538518952925, "grad_norm": 5.027553558349609, "learning_rate": 8.882e-06, "loss": 0.0666, "step": 5591 }, { "epoch": 0.2981286986191822, "grad_norm": 4.095149517059326, "learning_rate": 8.8818e-06, "loss": 0.0618, "step": 5592 }, { "epoch": 0.2981820120488351, "grad_norm": 3.900831460952759, "learning_rate": 8.8816e-06, "loss": 0.0447, "step": 5593 }, { "epoch": 0.298235325478488, "grad_norm": 6.222609996795654, "learning_rate": 8.881400000000001e-06, "loss": 0.0687, "step": 5594 }, { "epoch": 0.29828863890814095, "grad_norm": 2.424478769302368, "learning_rate": 8.8812e-06, "loss": 0.0558, "step": 5595 }, { "epoch": 0.29834195233779387, "grad_norm": 2.994774103164673, "learning_rate": 8.881e-06, "loss": 0.0273, "step": 5596 }, { "epoch": 0.2983952657674468, "grad_norm": 3.2526607513427734, "learning_rate": 8.8808e-06, "loss": 0.0349, "step": 5597 }, { "epoch": 0.2984485791970998, "grad_norm": 4.162513256072998, "learning_rate": 8.880600000000001e-06, "loss": 0.0897, "step": 5598 }, { "epoch": 0.2985018926267527, "grad_norm": 2.1688380241394043, "learning_rate": 8.8804e-06, "loss": 0.0301, "step": 5599 }, { "epoch": 0.2985552060564056, "grad_norm": 4.400182247161865, "learning_rate": 8.880200000000002e-06, "loss": 0.0397, "step": 5600 }, { "epoch": 0.29860851948605854, "grad_norm": 5.418437957763672, "learning_rate": 8.880000000000001e-06, "loss": 0.0957, "step": 5601 }, { "epoch": 0.29866183291571147, "grad_norm": 3.967818021774292, "learning_rate": 8.879800000000001e-06, "loss": 0.0639, "step": 5602 }, { "epoch": 0.2987151463453644, "grad_norm": 3.5747945308685303, "learning_rate": 8.8796e-06, "loss": 0.0715, "step": 5603 }, { "epoch": 0.2987684597750173, "grad_norm": 3.1540377140045166, "learning_rate": 8.8794e-06, "loss": 0.0537, "step": 5604 }, { "epoch": 0.29882177320467024, "grad_norm": 5.697232246398926, "learning_rate": 8.879200000000001e-06, "loss": 0.0482, "step": 5605 }, { "epoch": 0.29887508663432316, "grad_norm": 1.7797048091888428, "learning_rate": 8.879000000000001e-06, "loss": 0.0184, "step": 5606 }, { "epoch": 0.29892840006397614, "grad_norm": 3.1797983646392822, "learning_rate": 8.8788e-06, "loss": 0.0598, "step": 5607 }, { "epoch": 0.29898171349362906, "grad_norm": 6.629063129425049, "learning_rate": 8.8786e-06, "loss": 0.0567, "step": 5608 }, { "epoch": 0.299035026923282, "grad_norm": 2.9746100902557373, "learning_rate": 8.8784e-06, "loss": 0.0301, "step": 5609 }, { "epoch": 0.2990883403529349, "grad_norm": 3.88354754447937, "learning_rate": 8.8782e-06, "loss": 0.054, "step": 5610 }, { "epoch": 0.29914165378258784, "grad_norm": 4.208117485046387, "learning_rate": 8.878e-06, "loss": 0.0519, "step": 5611 }, { "epoch": 0.29919496721224076, "grad_norm": 6.208467483520508, "learning_rate": 8.877800000000002e-06, "loss": 0.0883, "step": 5612 }, { "epoch": 0.2992482806418937, "grad_norm": 2.859754800796509, "learning_rate": 8.877600000000001e-06, "loss": 0.032, "step": 5613 }, { "epoch": 0.2993015940715466, "grad_norm": 4.692954063415527, "learning_rate": 8.8774e-06, "loss": 0.0611, "step": 5614 }, { "epoch": 0.29935490750119953, "grad_norm": 3.3975987434387207, "learning_rate": 8.8772e-06, "loss": 0.0537, "step": 5615 }, { "epoch": 0.2994082209308525, "grad_norm": 2.532482624053955, "learning_rate": 8.877000000000002e-06, "loss": 0.0564, "step": 5616 }, { "epoch": 0.29946153436050543, "grad_norm": 4.057301998138428, "learning_rate": 8.876800000000001e-06, "loss": 0.0347, "step": 5617 }, { "epoch": 0.29951484779015836, "grad_norm": 4.3415069580078125, "learning_rate": 8.8766e-06, "loss": 0.06, "step": 5618 }, { "epoch": 0.2995681612198113, "grad_norm": 2.4087533950805664, "learning_rate": 8.8764e-06, "loss": 0.0402, "step": 5619 }, { "epoch": 0.2996214746494642, "grad_norm": 2.793565511703491, "learning_rate": 8.8762e-06, "loss": 0.054, "step": 5620 }, { "epoch": 0.2996747880791171, "grad_norm": 2.435750722885132, "learning_rate": 8.876e-06, "loss": 0.044, "step": 5621 }, { "epoch": 0.29972810150877005, "grad_norm": 3.7454495429992676, "learning_rate": 8.8758e-06, "loss": 0.0626, "step": 5622 }, { "epoch": 0.299781414938423, "grad_norm": 2.5394744873046875, "learning_rate": 8.8756e-06, "loss": 0.0711, "step": 5623 }, { "epoch": 0.2998347283680759, "grad_norm": 3.054453134536743, "learning_rate": 8.875400000000001e-06, "loss": 0.0432, "step": 5624 }, { "epoch": 0.2998880417977288, "grad_norm": 2.500612258911133, "learning_rate": 8.875200000000001e-06, "loss": 0.0545, "step": 5625 }, { "epoch": 0.2999413552273818, "grad_norm": 2.121216297149658, "learning_rate": 8.875e-06, "loss": 0.0352, "step": 5626 }, { "epoch": 0.2999946686570347, "grad_norm": 3.0214455127716064, "learning_rate": 8.874800000000002e-06, "loss": 0.0336, "step": 5627 }, { "epoch": 0.30004798208668765, "grad_norm": 5.956537246704102, "learning_rate": 8.874600000000001e-06, "loss": 0.0424, "step": 5628 }, { "epoch": 0.30010129551634057, "grad_norm": 3.542142391204834, "learning_rate": 8.874400000000001e-06, "loss": 0.044, "step": 5629 }, { "epoch": 0.3001546089459935, "grad_norm": 2.4568920135498047, "learning_rate": 8.8742e-06, "loss": 0.038, "step": 5630 }, { "epoch": 0.3002079223756464, "grad_norm": 3.6517555713653564, "learning_rate": 8.874e-06, "loss": 0.0509, "step": 5631 }, { "epoch": 0.30026123580529934, "grad_norm": 2.249191999435425, "learning_rate": 8.8738e-06, "loss": 0.0331, "step": 5632 }, { "epoch": 0.30031454923495227, "grad_norm": 4.02691650390625, "learning_rate": 8.873600000000001e-06, "loss": 0.0523, "step": 5633 }, { "epoch": 0.3003678626646052, "grad_norm": 2.9865376949310303, "learning_rate": 8.8734e-06, "loss": 0.043, "step": 5634 }, { "epoch": 0.30042117609425817, "grad_norm": 3.4922714233398438, "learning_rate": 8.8732e-06, "loss": 0.036, "step": 5635 }, { "epoch": 0.3004744895239111, "grad_norm": 2.267138719558716, "learning_rate": 8.873000000000001e-06, "loss": 0.0355, "step": 5636 }, { "epoch": 0.300527802953564, "grad_norm": 2.295654535293579, "learning_rate": 8.8728e-06, "loss": 0.0332, "step": 5637 }, { "epoch": 0.30058111638321694, "grad_norm": 3.502267599105835, "learning_rate": 8.8726e-06, "loss": 0.0374, "step": 5638 }, { "epoch": 0.30063442981286986, "grad_norm": 3.3637115955352783, "learning_rate": 8.872400000000002e-06, "loss": 0.0343, "step": 5639 }, { "epoch": 0.3006877432425228, "grad_norm": 3.180302858352661, "learning_rate": 8.872200000000001e-06, "loss": 0.0494, "step": 5640 }, { "epoch": 0.3007410566721757, "grad_norm": 1.5696961879730225, "learning_rate": 8.872e-06, "loss": 0.0145, "step": 5641 }, { "epoch": 0.30079437010182863, "grad_norm": 4.427831649780273, "learning_rate": 8.8718e-06, "loss": 0.0585, "step": 5642 }, { "epoch": 0.30084768353148156, "grad_norm": 2.7759475708007812, "learning_rate": 8.8716e-06, "loss": 0.0356, "step": 5643 }, { "epoch": 0.30090099696113454, "grad_norm": 3.224695920944214, "learning_rate": 8.871400000000001e-06, "loss": 0.0513, "step": 5644 }, { "epoch": 0.30095431039078746, "grad_norm": 2.6976964473724365, "learning_rate": 8.8712e-06, "loss": 0.0283, "step": 5645 }, { "epoch": 0.3010076238204404, "grad_norm": 4.045102596282959, "learning_rate": 8.871e-06, "loss": 0.0735, "step": 5646 }, { "epoch": 0.3010609372500933, "grad_norm": 6.9969587326049805, "learning_rate": 8.8708e-06, "loss": 0.0753, "step": 5647 }, { "epoch": 0.30111425067974623, "grad_norm": 4.914448261260986, "learning_rate": 8.870600000000001e-06, "loss": 0.0526, "step": 5648 }, { "epoch": 0.30116756410939916, "grad_norm": 2.736746072769165, "learning_rate": 8.8704e-06, "loss": 0.0333, "step": 5649 }, { "epoch": 0.3012208775390521, "grad_norm": 3.4078333377838135, "learning_rate": 8.8702e-06, "loss": 0.0597, "step": 5650 }, { "epoch": 0.301274190968705, "grad_norm": 4.163862228393555, "learning_rate": 8.870000000000001e-06, "loss": 0.0403, "step": 5651 }, { "epoch": 0.3013275043983579, "grad_norm": 3.0192301273345947, "learning_rate": 8.869800000000001e-06, "loss": 0.0383, "step": 5652 }, { "epoch": 0.30138081782801085, "grad_norm": 4.5116167068481445, "learning_rate": 8.8696e-06, "loss": 0.0608, "step": 5653 }, { "epoch": 0.30143413125766383, "grad_norm": 1.6894479990005493, "learning_rate": 8.8694e-06, "loss": 0.0334, "step": 5654 }, { "epoch": 0.30148744468731675, "grad_norm": 3.1684157848358154, "learning_rate": 8.869200000000001e-06, "loss": 0.0507, "step": 5655 }, { "epoch": 0.3015407581169697, "grad_norm": 4.41007137298584, "learning_rate": 8.869000000000001e-06, "loss": 0.0401, "step": 5656 }, { "epoch": 0.3015940715466226, "grad_norm": 2.6916866302490234, "learning_rate": 8.8688e-06, "loss": 0.0564, "step": 5657 }, { "epoch": 0.3016473849762755, "grad_norm": 3.1309056282043457, "learning_rate": 8.8686e-06, "loss": 0.0357, "step": 5658 }, { "epoch": 0.30170069840592845, "grad_norm": 2.260890483856201, "learning_rate": 8.8684e-06, "loss": 0.0574, "step": 5659 }, { "epoch": 0.30175401183558137, "grad_norm": 5.492054462432861, "learning_rate": 8.868200000000001e-06, "loss": 0.0531, "step": 5660 }, { "epoch": 0.3018073252652343, "grad_norm": 3.2739150524139404, "learning_rate": 8.868e-06, "loss": 0.0541, "step": 5661 }, { "epoch": 0.3018606386948872, "grad_norm": 6.05333948135376, "learning_rate": 8.8678e-06, "loss": 0.0421, "step": 5662 }, { "epoch": 0.3019139521245402, "grad_norm": 3.5797178745269775, "learning_rate": 8.867600000000001e-06, "loss": 0.0405, "step": 5663 }, { "epoch": 0.3019672655541931, "grad_norm": 2.856877088546753, "learning_rate": 8.867400000000001e-06, "loss": 0.0323, "step": 5664 }, { "epoch": 0.30202057898384604, "grad_norm": 4.293397426605225, "learning_rate": 8.8672e-06, "loss": 0.069, "step": 5665 }, { "epoch": 0.30207389241349897, "grad_norm": 2.931407928466797, "learning_rate": 8.867000000000002e-06, "loss": 0.0634, "step": 5666 }, { "epoch": 0.3021272058431519, "grad_norm": 2.3784689903259277, "learning_rate": 8.866800000000001e-06, "loss": 0.0463, "step": 5667 }, { "epoch": 0.3021805192728048, "grad_norm": 5.197876930236816, "learning_rate": 8.8666e-06, "loss": 0.053, "step": 5668 }, { "epoch": 0.30223383270245774, "grad_norm": 3.7252068519592285, "learning_rate": 8.8664e-06, "loss": 0.0498, "step": 5669 }, { "epoch": 0.30228714613211066, "grad_norm": 2.8588192462921143, "learning_rate": 8.8662e-06, "loss": 0.0504, "step": 5670 }, { "epoch": 0.3023404595617636, "grad_norm": 3.49545955657959, "learning_rate": 8.866000000000001e-06, "loss": 0.0504, "step": 5671 }, { "epoch": 0.30239377299141657, "grad_norm": 2.7342371940612793, "learning_rate": 8.8658e-06, "loss": 0.059, "step": 5672 }, { "epoch": 0.3024470864210695, "grad_norm": 3.2138891220092773, "learning_rate": 8.8656e-06, "loss": 0.0509, "step": 5673 }, { "epoch": 0.3025003998507224, "grad_norm": 4.274609565734863, "learning_rate": 8.8654e-06, "loss": 0.0523, "step": 5674 }, { "epoch": 0.30255371328037534, "grad_norm": 6.908424377441406, "learning_rate": 8.865200000000001e-06, "loss": 0.0526, "step": 5675 }, { "epoch": 0.30260702671002826, "grad_norm": 3.998760938644409, "learning_rate": 8.865e-06, "loss": 0.04, "step": 5676 }, { "epoch": 0.3026603401396812, "grad_norm": 5.08120584487915, "learning_rate": 8.864800000000002e-06, "loss": 0.0357, "step": 5677 }, { "epoch": 0.3027136535693341, "grad_norm": 3.2536282539367676, "learning_rate": 8.864600000000001e-06, "loss": 0.0649, "step": 5678 }, { "epoch": 0.30276696699898703, "grad_norm": 5.300818920135498, "learning_rate": 8.864400000000001e-06, "loss": 0.0467, "step": 5679 }, { "epoch": 0.30282028042863995, "grad_norm": 5.025247573852539, "learning_rate": 8.8642e-06, "loss": 0.0755, "step": 5680 }, { "epoch": 0.3028735938582929, "grad_norm": 4.335057258605957, "learning_rate": 8.864e-06, "loss": 0.045, "step": 5681 }, { "epoch": 0.30292690728794586, "grad_norm": 4.121288776397705, "learning_rate": 8.8638e-06, "loss": 0.0442, "step": 5682 }, { "epoch": 0.3029802207175988, "grad_norm": 3.070875644683838, "learning_rate": 8.863600000000001e-06, "loss": 0.044, "step": 5683 }, { "epoch": 0.3030335341472517, "grad_norm": 3.2947700023651123, "learning_rate": 8.8634e-06, "loss": 0.0391, "step": 5684 }, { "epoch": 0.3030868475769046, "grad_norm": 4.1568732261657715, "learning_rate": 8.8632e-06, "loss": 0.0665, "step": 5685 }, { "epoch": 0.30314016100655755, "grad_norm": 2.7395806312561035, "learning_rate": 8.863e-06, "loss": 0.0381, "step": 5686 }, { "epoch": 0.3031934744362105, "grad_norm": 6.492772579193115, "learning_rate": 8.862800000000001e-06, "loss": 0.073, "step": 5687 }, { "epoch": 0.3032467878658634, "grad_norm": 9.305519104003906, "learning_rate": 8.8626e-06, "loss": 0.077, "step": 5688 }, { "epoch": 0.3033001012955163, "grad_norm": 3.947373151779175, "learning_rate": 8.862400000000002e-06, "loss": 0.0465, "step": 5689 }, { "epoch": 0.30335341472516925, "grad_norm": 2.2140398025512695, "learning_rate": 8.862200000000001e-06, "loss": 0.0307, "step": 5690 }, { "epoch": 0.3034067281548222, "grad_norm": 4.457790374755859, "learning_rate": 8.862000000000001e-06, "loss": 0.0524, "step": 5691 }, { "epoch": 0.30346004158447515, "grad_norm": 3.3606925010681152, "learning_rate": 8.8618e-06, "loss": 0.0606, "step": 5692 }, { "epoch": 0.3035133550141281, "grad_norm": 3.444378614425659, "learning_rate": 8.8616e-06, "loss": 0.0314, "step": 5693 }, { "epoch": 0.303566668443781, "grad_norm": 5.231302261352539, "learning_rate": 8.861400000000001e-06, "loss": 0.054, "step": 5694 }, { "epoch": 0.3036199818734339, "grad_norm": 3.3843865394592285, "learning_rate": 8.8612e-06, "loss": 0.048, "step": 5695 }, { "epoch": 0.30367329530308684, "grad_norm": 5.08392858505249, "learning_rate": 8.861e-06, "loss": 0.0691, "step": 5696 }, { "epoch": 0.30372660873273977, "grad_norm": 4.514339447021484, "learning_rate": 8.8608e-06, "loss": 0.0563, "step": 5697 }, { "epoch": 0.3037799221623927, "grad_norm": 2.287405252456665, "learning_rate": 8.8606e-06, "loss": 0.0417, "step": 5698 }, { "epoch": 0.3038332355920456, "grad_norm": 6.881963729858398, "learning_rate": 8.8604e-06, "loss": 0.0471, "step": 5699 }, { "epoch": 0.3038865490216986, "grad_norm": 4.032498836517334, "learning_rate": 8.8602e-06, "loss": 0.0569, "step": 5700 }, { "epoch": 0.3039398624513515, "grad_norm": 3.5562734603881836, "learning_rate": 8.860000000000002e-06, "loss": 0.0431, "step": 5701 }, { "epoch": 0.30399317588100444, "grad_norm": 5.755902290344238, "learning_rate": 8.859800000000001e-06, "loss": 0.0518, "step": 5702 }, { "epoch": 0.30404648931065736, "grad_norm": 9.096096992492676, "learning_rate": 8.8596e-06, "loss": 0.0717, "step": 5703 }, { "epoch": 0.3040998027403103, "grad_norm": 5.432896137237549, "learning_rate": 8.8594e-06, "loss": 0.0519, "step": 5704 }, { "epoch": 0.3041531161699632, "grad_norm": 3.45257306098938, "learning_rate": 8.859200000000002e-06, "loss": 0.0538, "step": 5705 }, { "epoch": 0.30420642959961613, "grad_norm": 3.9643499851226807, "learning_rate": 8.859000000000001e-06, "loss": 0.0621, "step": 5706 }, { "epoch": 0.30425974302926906, "grad_norm": 5.6778059005737305, "learning_rate": 8.8588e-06, "loss": 0.0794, "step": 5707 }, { "epoch": 0.304313056458922, "grad_norm": 6.142612457275391, "learning_rate": 8.8586e-06, "loss": 0.0766, "step": 5708 }, { "epoch": 0.3043663698885749, "grad_norm": 3.276487112045288, "learning_rate": 8.8584e-06, "loss": 0.0333, "step": 5709 }, { "epoch": 0.3044196833182279, "grad_norm": 3.2514846324920654, "learning_rate": 8.858200000000001e-06, "loss": 0.0415, "step": 5710 }, { "epoch": 0.3044729967478808, "grad_norm": 1.642209529876709, "learning_rate": 8.858e-06, "loss": 0.0176, "step": 5711 }, { "epoch": 0.30452631017753373, "grad_norm": 4.976584434509277, "learning_rate": 8.8578e-06, "loss": 0.0526, "step": 5712 }, { "epoch": 0.30457962360718666, "grad_norm": 8.189492225646973, "learning_rate": 8.857600000000001e-06, "loss": 0.0508, "step": 5713 }, { "epoch": 0.3046329370368396, "grad_norm": 2.586913824081421, "learning_rate": 8.857400000000001e-06, "loss": 0.0423, "step": 5714 }, { "epoch": 0.3046862504664925, "grad_norm": 3.057753086090088, "learning_rate": 8.8572e-06, "loss": 0.0628, "step": 5715 }, { "epoch": 0.3047395638961454, "grad_norm": 5.521394729614258, "learning_rate": 8.857000000000002e-06, "loss": 0.074, "step": 5716 }, { "epoch": 0.30479287732579835, "grad_norm": 1.4840480089187622, "learning_rate": 8.856800000000001e-06, "loss": 0.0138, "step": 5717 }, { "epoch": 0.3048461907554513, "grad_norm": 2.4080464839935303, "learning_rate": 8.856600000000001e-06, "loss": 0.0601, "step": 5718 }, { "epoch": 0.30489950418510425, "grad_norm": 4.753239631652832, "learning_rate": 8.8564e-06, "loss": 0.0646, "step": 5719 }, { "epoch": 0.3049528176147572, "grad_norm": 2.571429491043091, "learning_rate": 8.8562e-06, "loss": 0.0352, "step": 5720 }, { "epoch": 0.3050061310444101, "grad_norm": 3.340083599090576, "learning_rate": 8.856000000000001e-06, "loss": 0.0552, "step": 5721 }, { "epoch": 0.305059444474063, "grad_norm": 2.2554495334625244, "learning_rate": 8.8558e-06, "loss": 0.0306, "step": 5722 }, { "epoch": 0.30511275790371595, "grad_norm": 1.982591986656189, "learning_rate": 8.8556e-06, "loss": 0.0345, "step": 5723 }, { "epoch": 0.30516607133336887, "grad_norm": 4.249853134155273, "learning_rate": 8.8554e-06, "loss": 0.0332, "step": 5724 }, { "epoch": 0.3052193847630218, "grad_norm": 4.2295660972595215, "learning_rate": 8.855200000000001e-06, "loss": 0.0376, "step": 5725 }, { "epoch": 0.3052726981926747, "grad_norm": 4.494891166687012, "learning_rate": 8.855e-06, "loss": 0.0628, "step": 5726 }, { "epoch": 0.30532601162232764, "grad_norm": 4.314882755279541, "learning_rate": 8.8548e-06, "loss": 0.0862, "step": 5727 }, { "epoch": 0.3053793250519806, "grad_norm": 2.0288383960723877, "learning_rate": 8.854600000000002e-06, "loss": 0.0344, "step": 5728 }, { "epoch": 0.30543263848163354, "grad_norm": 3.878657341003418, "learning_rate": 8.854400000000001e-06, "loss": 0.054, "step": 5729 }, { "epoch": 0.30548595191128647, "grad_norm": 4.551731586456299, "learning_rate": 8.8542e-06, "loss": 0.0871, "step": 5730 }, { "epoch": 0.3055392653409394, "grad_norm": 3.6560001373291016, "learning_rate": 8.854e-06, "loss": 0.0393, "step": 5731 }, { "epoch": 0.3055925787705923, "grad_norm": 2.0023434162139893, "learning_rate": 8.8538e-06, "loss": 0.0141, "step": 5732 }, { "epoch": 0.30564589220024524, "grad_norm": 3.623535633087158, "learning_rate": 8.853600000000001e-06, "loss": 0.0351, "step": 5733 }, { "epoch": 0.30569920562989816, "grad_norm": 3.558955192565918, "learning_rate": 8.8534e-06, "loss": 0.0532, "step": 5734 }, { "epoch": 0.3057525190595511, "grad_norm": 2.175985813140869, "learning_rate": 8.8532e-06, "loss": 0.0361, "step": 5735 }, { "epoch": 0.305805832489204, "grad_norm": 2.1476612091064453, "learning_rate": 8.853e-06, "loss": 0.0225, "step": 5736 }, { "epoch": 0.30585914591885693, "grad_norm": 3.9594080448150635, "learning_rate": 8.852800000000001e-06, "loss": 0.0393, "step": 5737 }, { "epoch": 0.3059124593485099, "grad_norm": 2.9609997272491455, "learning_rate": 8.8526e-06, "loss": 0.0326, "step": 5738 }, { "epoch": 0.30596577277816284, "grad_norm": 2.369875192642212, "learning_rate": 8.8524e-06, "loss": 0.0274, "step": 5739 }, { "epoch": 0.30601908620781576, "grad_norm": 3.6906232833862305, "learning_rate": 8.852200000000001e-06, "loss": 0.0416, "step": 5740 }, { "epoch": 0.3060723996374687, "grad_norm": 3.7157704830169678, "learning_rate": 8.852000000000001e-06, "loss": 0.0797, "step": 5741 }, { "epoch": 0.3061257130671216, "grad_norm": 3.53938627243042, "learning_rate": 8.8518e-06, "loss": 0.0322, "step": 5742 }, { "epoch": 0.30617902649677453, "grad_norm": 1.4459547996520996, "learning_rate": 8.8516e-06, "loss": 0.0158, "step": 5743 }, { "epoch": 0.30623233992642745, "grad_norm": 3.615952968597412, "learning_rate": 8.851400000000001e-06, "loss": 0.0321, "step": 5744 }, { "epoch": 0.3062856533560804, "grad_norm": 5.681549549102783, "learning_rate": 8.851200000000001e-06, "loss": 0.0901, "step": 5745 }, { "epoch": 0.3063389667857333, "grad_norm": 3.465440511703491, "learning_rate": 8.851e-06, "loss": 0.0491, "step": 5746 }, { "epoch": 0.3063922802153863, "grad_norm": 2.6617698669433594, "learning_rate": 8.8508e-06, "loss": 0.0325, "step": 5747 }, { "epoch": 0.3064455936450392, "grad_norm": 3.3138415813446045, "learning_rate": 8.8506e-06, "loss": 0.0441, "step": 5748 }, { "epoch": 0.30649890707469213, "grad_norm": 2.732839822769165, "learning_rate": 8.850400000000001e-06, "loss": 0.046, "step": 5749 }, { "epoch": 0.30655222050434505, "grad_norm": 3.465747594833374, "learning_rate": 8.8502e-06, "loss": 0.0629, "step": 5750 }, { "epoch": 0.306605533933998, "grad_norm": 1.8375149965286255, "learning_rate": 8.85e-06, "loss": 0.0307, "step": 5751 }, { "epoch": 0.3066588473636509, "grad_norm": 4.930338382720947, "learning_rate": 8.849800000000001e-06, "loss": 0.0612, "step": 5752 }, { "epoch": 0.3067121607933038, "grad_norm": 2.2192728519439697, "learning_rate": 8.8496e-06, "loss": 0.0148, "step": 5753 }, { "epoch": 0.30676547422295675, "grad_norm": 2.9180285930633545, "learning_rate": 8.8494e-06, "loss": 0.0441, "step": 5754 }, { "epoch": 0.30681878765260967, "grad_norm": 4.568373680114746, "learning_rate": 8.849200000000002e-06, "loss": 0.0687, "step": 5755 }, { "epoch": 0.30687210108226265, "grad_norm": 4.441539764404297, "learning_rate": 8.849000000000001e-06, "loss": 0.027, "step": 5756 }, { "epoch": 0.3069254145119156, "grad_norm": 2.9108777046203613, "learning_rate": 8.8488e-06, "loss": 0.0237, "step": 5757 }, { "epoch": 0.3069787279415685, "grad_norm": 4.127772331237793, "learning_rate": 8.8486e-06, "loss": 0.0506, "step": 5758 }, { "epoch": 0.3070320413712214, "grad_norm": 1.2882798910140991, "learning_rate": 8.8484e-06, "loss": 0.0118, "step": 5759 }, { "epoch": 0.30708535480087434, "grad_norm": 3.1686480045318604, "learning_rate": 8.848200000000001e-06, "loss": 0.0443, "step": 5760 }, { "epoch": 0.30713866823052727, "grad_norm": 1.4902284145355225, "learning_rate": 8.848e-06, "loss": 0.0226, "step": 5761 }, { "epoch": 0.3071919816601802, "grad_norm": 5.102294445037842, "learning_rate": 8.8478e-06, "loss": 0.0544, "step": 5762 }, { "epoch": 0.3072452950898331, "grad_norm": 2.821032762527466, "learning_rate": 8.8476e-06, "loss": 0.0567, "step": 5763 }, { "epoch": 0.30729860851948604, "grad_norm": 4.63505220413208, "learning_rate": 8.847400000000001e-06, "loss": 0.0548, "step": 5764 }, { "epoch": 0.30735192194913896, "grad_norm": 5.264856815338135, "learning_rate": 8.8472e-06, "loss": 0.0507, "step": 5765 }, { "epoch": 0.30740523537879194, "grad_norm": 6.306783676147461, "learning_rate": 8.847000000000002e-06, "loss": 0.0524, "step": 5766 }, { "epoch": 0.30745854880844486, "grad_norm": 3.368741512298584, "learning_rate": 8.846800000000001e-06, "loss": 0.04, "step": 5767 }, { "epoch": 0.3075118622380978, "grad_norm": 4.467993259429932, "learning_rate": 8.846600000000001e-06, "loss": 0.047, "step": 5768 }, { "epoch": 0.3075651756677507, "grad_norm": 2.6200644969940186, "learning_rate": 8.8464e-06, "loss": 0.0396, "step": 5769 }, { "epoch": 0.30761848909740364, "grad_norm": 3.926907539367676, "learning_rate": 8.8462e-06, "loss": 0.057, "step": 5770 }, { "epoch": 0.30767180252705656, "grad_norm": 4.605762004852295, "learning_rate": 8.846000000000001e-06, "loss": 0.0895, "step": 5771 }, { "epoch": 0.3077251159567095, "grad_norm": 3.1883718967437744, "learning_rate": 8.845800000000001e-06, "loss": 0.0269, "step": 5772 }, { "epoch": 0.3077784293863624, "grad_norm": 4.736939907073975, "learning_rate": 8.8456e-06, "loss": 0.0544, "step": 5773 }, { "epoch": 0.30783174281601533, "grad_norm": 1.8053250312805176, "learning_rate": 8.8454e-06, "loss": 0.0149, "step": 5774 }, { "epoch": 0.3078850562456683, "grad_norm": 5.976571083068848, "learning_rate": 8.8452e-06, "loss": 0.0457, "step": 5775 }, { "epoch": 0.30793836967532123, "grad_norm": 4.103928089141846, "learning_rate": 8.845000000000001e-06, "loss": 0.0588, "step": 5776 }, { "epoch": 0.30799168310497416, "grad_norm": 1.2664947509765625, "learning_rate": 8.8448e-06, "loss": 0.0282, "step": 5777 }, { "epoch": 0.3080449965346271, "grad_norm": 3.8745501041412354, "learning_rate": 8.844600000000002e-06, "loss": 0.0619, "step": 5778 }, { "epoch": 0.30809830996428, "grad_norm": 4.07175874710083, "learning_rate": 8.844400000000001e-06, "loss": 0.0509, "step": 5779 }, { "epoch": 0.3081516233939329, "grad_norm": 2.2819712162017822, "learning_rate": 8.8442e-06, "loss": 0.0341, "step": 5780 }, { "epoch": 0.30820493682358585, "grad_norm": 4.767197608947754, "learning_rate": 8.844e-06, "loss": 0.0386, "step": 5781 }, { "epoch": 0.3082582502532388, "grad_norm": 4.952181816101074, "learning_rate": 8.843800000000002e-06, "loss": 0.0462, "step": 5782 }, { "epoch": 0.3083115636828917, "grad_norm": 4.082140922546387, "learning_rate": 8.843600000000001e-06, "loss": 0.0371, "step": 5783 }, { "epoch": 0.3083648771125447, "grad_norm": 5.002349376678467, "learning_rate": 8.8434e-06, "loss": 0.0471, "step": 5784 }, { "epoch": 0.3084181905421976, "grad_norm": 4.785473823547363, "learning_rate": 8.8432e-06, "loss": 0.0493, "step": 5785 }, { "epoch": 0.3084715039718505, "grad_norm": 5.736403465270996, "learning_rate": 8.843e-06, "loss": 0.0597, "step": 5786 }, { "epoch": 0.30852481740150345, "grad_norm": 4.1172776222229, "learning_rate": 8.8428e-06, "loss": 0.0273, "step": 5787 }, { "epoch": 0.30857813083115637, "grad_norm": 1.6361576318740845, "learning_rate": 8.8426e-06, "loss": 0.0221, "step": 5788 }, { "epoch": 0.3086314442608093, "grad_norm": 6.140713691711426, "learning_rate": 8.8424e-06, "loss": 0.0563, "step": 5789 }, { "epoch": 0.3086847576904622, "grad_norm": 2.621941566467285, "learning_rate": 8.842200000000002e-06, "loss": 0.0618, "step": 5790 }, { "epoch": 0.30873807112011514, "grad_norm": 3.5325653553009033, "learning_rate": 8.842000000000001e-06, "loss": 0.0462, "step": 5791 }, { "epoch": 0.30879138454976807, "grad_norm": 5.006819725036621, "learning_rate": 8.8418e-06, "loss": 0.0521, "step": 5792 }, { "epoch": 0.308844697979421, "grad_norm": 2.660531997680664, "learning_rate": 8.8416e-06, "loss": 0.0394, "step": 5793 }, { "epoch": 0.30889801140907397, "grad_norm": 4.293129920959473, "learning_rate": 8.841400000000001e-06, "loss": 0.0451, "step": 5794 }, { "epoch": 0.3089513248387269, "grad_norm": 5.543213367462158, "learning_rate": 8.841200000000001e-06, "loss": 0.0918, "step": 5795 }, { "epoch": 0.3090046382683798, "grad_norm": 3.847705841064453, "learning_rate": 8.841e-06, "loss": 0.0442, "step": 5796 }, { "epoch": 0.30905795169803274, "grad_norm": 3.18493914604187, "learning_rate": 8.8408e-06, "loss": 0.0549, "step": 5797 }, { "epoch": 0.30911126512768566, "grad_norm": 4.365265846252441, "learning_rate": 8.8406e-06, "loss": 0.0399, "step": 5798 }, { "epoch": 0.3091645785573386, "grad_norm": 5.163261413574219, "learning_rate": 8.840400000000001e-06, "loss": 0.0641, "step": 5799 }, { "epoch": 0.3092178919869915, "grad_norm": 2.531738519668579, "learning_rate": 8.8402e-06, "loss": 0.0278, "step": 5800 }, { "epoch": 0.30927120541664443, "grad_norm": 2.806845188140869, "learning_rate": 8.84e-06, "loss": 0.0363, "step": 5801 }, { "epoch": 0.30932451884629736, "grad_norm": 2.781156301498413, "learning_rate": 8.839800000000001e-06, "loss": 0.0259, "step": 5802 }, { "epoch": 0.30937783227595034, "grad_norm": 1.8615468740463257, "learning_rate": 8.839600000000001e-06, "loss": 0.0438, "step": 5803 }, { "epoch": 0.30943114570560326, "grad_norm": 6.258791923522949, "learning_rate": 8.8394e-06, "loss": 0.0876, "step": 5804 }, { "epoch": 0.3094844591352562, "grad_norm": 2.364880323410034, "learning_rate": 8.839200000000002e-06, "loss": 0.0358, "step": 5805 }, { "epoch": 0.3095377725649091, "grad_norm": 3.669936180114746, "learning_rate": 8.839000000000001e-06, "loss": 0.0839, "step": 5806 }, { "epoch": 0.30959108599456203, "grad_norm": 2.427076578140259, "learning_rate": 8.8388e-06, "loss": 0.036, "step": 5807 }, { "epoch": 0.30964439942421496, "grad_norm": 6.210628032684326, "learning_rate": 8.8386e-06, "loss": 0.0882, "step": 5808 }, { "epoch": 0.3096977128538679, "grad_norm": 4.313497066497803, "learning_rate": 8.8384e-06, "loss": 0.0606, "step": 5809 }, { "epoch": 0.3097510262835208, "grad_norm": 3.5035221576690674, "learning_rate": 8.838200000000001e-06, "loss": 0.0459, "step": 5810 }, { "epoch": 0.3098043397131737, "grad_norm": 3.32749080657959, "learning_rate": 8.838e-06, "loss": 0.0677, "step": 5811 }, { "epoch": 0.3098576531428267, "grad_norm": 4.678511142730713, "learning_rate": 8.8378e-06, "loss": 0.035, "step": 5812 }, { "epoch": 0.30991096657247963, "grad_norm": 2.1272878646850586, "learning_rate": 8.8376e-06, "loss": 0.0196, "step": 5813 }, { "epoch": 0.30996428000213255, "grad_norm": 2.777646780014038, "learning_rate": 8.837400000000001e-06, "loss": 0.0219, "step": 5814 }, { "epoch": 0.3100175934317855, "grad_norm": 3.8392558097839355, "learning_rate": 8.8372e-06, "loss": 0.0395, "step": 5815 }, { "epoch": 0.3100709068614384, "grad_norm": 3.535118818283081, "learning_rate": 8.837000000000002e-06, "loss": 0.0334, "step": 5816 }, { "epoch": 0.3101242202910913, "grad_norm": 4.638618469238281, "learning_rate": 8.836800000000002e-06, "loss": 0.0517, "step": 5817 }, { "epoch": 0.31017753372074425, "grad_norm": 3.462322950363159, "learning_rate": 8.836600000000001e-06, "loss": 0.066, "step": 5818 }, { "epoch": 0.31023084715039717, "grad_norm": 2.1191647052764893, "learning_rate": 8.8364e-06, "loss": 0.0443, "step": 5819 }, { "epoch": 0.3102841605800501, "grad_norm": 4.7270636558532715, "learning_rate": 8.8362e-06, "loss": 0.057, "step": 5820 }, { "epoch": 0.310337474009703, "grad_norm": 3.2876267433166504, "learning_rate": 8.836000000000001e-06, "loss": 0.0367, "step": 5821 }, { "epoch": 0.310390787439356, "grad_norm": 4.435173988342285, "learning_rate": 8.835800000000001e-06, "loss": 0.0686, "step": 5822 }, { "epoch": 0.3104441008690089, "grad_norm": 5.652251243591309, "learning_rate": 8.8356e-06, "loss": 0.0557, "step": 5823 }, { "epoch": 0.31049741429866184, "grad_norm": 6.750976085662842, "learning_rate": 8.8354e-06, "loss": 0.0461, "step": 5824 }, { "epoch": 0.31055072772831477, "grad_norm": 3.4695615768432617, "learning_rate": 8.8352e-06, "loss": 0.0197, "step": 5825 }, { "epoch": 0.3106040411579677, "grad_norm": 3.192753553390503, "learning_rate": 8.835000000000001e-06, "loss": 0.0373, "step": 5826 }, { "epoch": 0.3106573545876206, "grad_norm": 2.1181089878082275, "learning_rate": 8.8348e-06, "loss": 0.0316, "step": 5827 }, { "epoch": 0.31071066801727354, "grad_norm": 2.395930767059326, "learning_rate": 8.834600000000002e-06, "loss": 0.0458, "step": 5828 }, { "epoch": 0.31076398144692646, "grad_norm": 2.750838279724121, "learning_rate": 8.834400000000001e-06, "loss": 0.0277, "step": 5829 }, { "epoch": 0.3108172948765794, "grad_norm": 3.351787567138672, "learning_rate": 8.834200000000001e-06, "loss": 0.0603, "step": 5830 }, { "epoch": 0.31087060830623237, "grad_norm": 4.843334197998047, "learning_rate": 8.834e-06, "loss": 0.0845, "step": 5831 }, { "epoch": 0.3109239217358853, "grad_norm": 3.3756935596466064, "learning_rate": 8.833800000000002e-06, "loss": 0.0418, "step": 5832 }, { "epoch": 0.3109772351655382, "grad_norm": 3.2918202877044678, "learning_rate": 8.833600000000001e-06, "loss": 0.0551, "step": 5833 }, { "epoch": 0.31103054859519114, "grad_norm": 6.278693675994873, "learning_rate": 8.8334e-06, "loss": 0.0744, "step": 5834 }, { "epoch": 0.31108386202484406, "grad_norm": 4.448496341705322, "learning_rate": 8.8332e-06, "loss": 0.0573, "step": 5835 }, { "epoch": 0.311137175454497, "grad_norm": 4.523166179656982, "learning_rate": 8.833e-06, "loss": 0.0531, "step": 5836 }, { "epoch": 0.3111904888841499, "grad_norm": 4.941734790802002, "learning_rate": 8.8328e-06, "loss": 0.0511, "step": 5837 }, { "epoch": 0.31124380231380283, "grad_norm": 2.479322671890259, "learning_rate": 8.8326e-06, "loss": 0.0342, "step": 5838 }, { "epoch": 0.31129711574345575, "grad_norm": 3.4456369876861572, "learning_rate": 8.8324e-06, "loss": 0.0232, "step": 5839 }, { "epoch": 0.31135042917310873, "grad_norm": 1.024474859237671, "learning_rate": 8.832200000000002e-06, "loss": 0.0103, "step": 5840 }, { "epoch": 0.31140374260276166, "grad_norm": 4.272007465362549, "learning_rate": 8.832000000000001e-06, "loss": 0.0697, "step": 5841 }, { "epoch": 0.3114570560324146, "grad_norm": 2.8119020462036133, "learning_rate": 8.8318e-06, "loss": 0.0396, "step": 5842 }, { "epoch": 0.3115103694620675, "grad_norm": 2.9809675216674805, "learning_rate": 8.8316e-06, "loss": 0.0563, "step": 5843 }, { "epoch": 0.31156368289172043, "grad_norm": 3.384784460067749, "learning_rate": 8.831400000000002e-06, "loss": 0.0331, "step": 5844 }, { "epoch": 0.31161699632137335, "grad_norm": 2.7417829036712646, "learning_rate": 8.831200000000001e-06, "loss": 0.0508, "step": 5845 }, { "epoch": 0.3116703097510263, "grad_norm": 3.233243703842163, "learning_rate": 8.831e-06, "loss": 0.0437, "step": 5846 }, { "epoch": 0.3117236231806792, "grad_norm": 6.546213626861572, "learning_rate": 8.8308e-06, "loss": 0.0523, "step": 5847 }, { "epoch": 0.3117769366103321, "grad_norm": 1.909013271331787, "learning_rate": 8.8306e-06, "loss": 0.0172, "step": 5848 }, { "epoch": 0.31183025003998505, "grad_norm": 2.9521000385284424, "learning_rate": 8.830400000000001e-06, "loss": 0.0253, "step": 5849 }, { "epoch": 0.311883563469638, "grad_norm": 4.147167682647705, "learning_rate": 8.8302e-06, "loss": 0.0471, "step": 5850 }, { "epoch": 0.31193687689929095, "grad_norm": 2.848283052444458, "learning_rate": 8.83e-06, "loss": 0.0356, "step": 5851 }, { "epoch": 0.3119901903289439, "grad_norm": 3.6241049766540527, "learning_rate": 8.829800000000001e-06, "loss": 0.0422, "step": 5852 }, { "epoch": 0.3120435037585968, "grad_norm": 4.219566345214844, "learning_rate": 8.829600000000001e-06, "loss": 0.0657, "step": 5853 }, { "epoch": 0.3120968171882497, "grad_norm": 2.2864832878112793, "learning_rate": 8.8294e-06, "loss": 0.0346, "step": 5854 }, { "epoch": 0.31215013061790264, "grad_norm": 4.128800392150879, "learning_rate": 8.829200000000002e-06, "loss": 0.0755, "step": 5855 }, { "epoch": 0.31220344404755557, "grad_norm": 3.5952401161193848, "learning_rate": 8.829000000000001e-06, "loss": 0.0612, "step": 5856 }, { "epoch": 0.3122567574772085, "grad_norm": 4.755633354187012, "learning_rate": 8.828800000000001e-06, "loss": 0.0492, "step": 5857 }, { "epoch": 0.3123100709068614, "grad_norm": 4.4767584800720215, "learning_rate": 8.8286e-06, "loss": 0.0597, "step": 5858 }, { "epoch": 0.3123633843365144, "grad_norm": 6.325934410095215, "learning_rate": 8.8284e-06, "loss": 0.0802, "step": 5859 }, { "epoch": 0.3124166977661673, "grad_norm": 2.372246503829956, "learning_rate": 8.828200000000001e-06, "loss": 0.0262, "step": 5860 }, { "epoch": 0.31247001119582024, "grad_norm": 4.0412187576293945, "learning_rate": 8.828000000000001e-06, "loss": 0.0376, "step": 5861 }, { "epoch": 0.31252332462547316, "grad_norm": 3.932035207748413, "learning_rate": 8.8278e-06, "loss": 0.0776, "step": 5862 }, { "epoch": 0.3125766380551261, "grad_norm": 6.680678367614746, "learning_rate": 8.8276e-06, "loss": 0.0716, "step": 5863 }, { "epoch": 0.312629951484779, "grad_norm": 3.366637945175171, "learning_rate": 8.827400000000001e-06, "loss": 0.0469, "step": 5864 }, { "epoch": 0.31268326491443194, "grad_norm": 5.513449668884277, "learning_rate": 8.8272e-06, "loss": 0.0644, "step": 5865 }, { "epoch": 0.31273657834408486, "grad_norm": 4.365952968597412, "learning_rate": 8.827e-06, "loss": 0.0375, "step": 5866 }, { "epoch": 0.3127898917737378, "grad_norm": 4.307469367980957, "learning_rate": 8.826800000000002e-06, "loss": 0.0431, "step": 5867 }, { "epoch": 0.31284320520339076, "grad_norm": 3.600027322769165, "learning_rate": 8.826600000000001e-06, "loss": 0.0574, "step": 5868 }, { "epoch": 0.3128965186330437, "grad_norm": 4.729079723358154, "learning_rate": 8.8264e-06, "loss": 0.0755, "step": 5869 }, { "epoch": 0.3129498320626966, "grad_norm": 2.482255220413208, "learning_rate": 8.8262e-06, "loss": 0.0386, "step": 5870 }, { "epoch": 0.31300314549234953, "grad_norm": 3.041396141052246, "learning_rate": 8.826000000000002e-06, "loss": 0.0499, "step": 5871 }, { "epoch": 0.31305645892200246, "grad_norm": 3.047670602798462, "learning_rate": 8.825800000000001e-06, "loss": 0.0261, "step": 5872 }, { "epoch": 0.3131097723516554, "grad_norm": 1.3247209787368774, "learning_rate": 8.8256e-06, "loss": 0.014, "step": 5873 }, { "epoch": 0.3131630857813083, "grad_norm": 4.491729259490967, "learning_rate": 8.8254e-06, "loss": 0.0828, "step": 5874 }, { "epoch": 0.3132163992109612, "grad_norm": 1.4654525518417358, "learning_rate": 8.8252e-06, "loss": 0.0364, "step": 5875 }, { "epoch": 0.31326971264061415, "grad_norm": 2.6685357093811035, "learning_rate": 8.825000000000001e-06, "loss": 0.037, "step": 5876 }, { "epoch": 0.3133230260702671, "grad_norm": 4.154197692871094, "learning_rate": 8.8248e-06, "loss": 0.0695, "step": 5877 }, { "epoch": 0.31337633949992005, "grad_norm": 2.8769805431365967, "learning_rate": 8.8246e-06, "loss": 0.0329, "step": 5878 }, { "epoch": 0.313429652929573, "grad_norm": 3.2640938758850098, "learning_rate": 8.824400000000001e-06, "loss": 0.0391, "step": 5879 }, { "epoch": 0.3134829663592259, "grad_norm": 6.552751064300537, "learning_rate": 8.824200000000001e-06, "loss": 0.0778, "step": 5880 }, { "epoch": 0.3135362797888788, "grad_norm": 6.992452144622803, "learning_rate": 8.824e-06, "loss": 0.0613, "step": 5881 }, { "epoch": 0.31358959321853175, "grad_norm": 2.0567452907562256, "learning_rate": 8.823800000000002e-06, "loss": 0.0165, "step": 5882 }, { "epoch": 0.31364290664818467, "grad_norm": 2.43170428276062, "learning_rate": 8.823600000000001e-06, "loss": 0.03, "step": 5883 }, { "epoch": 0.3136962200778376, "grad_norm": 4.825310230255127, "learning_rate": 8.823400000000001e-06, "loss": 0.0931, "step": 5884 }, { "epoch": 0.3137495335074905, "grad_norm": 4.618715286254883, "learning_rate": 8.8232e-06, "loss": 0.0613, "step": 5885 }, { "epoch": 0.31380284693714344, "grad_norm": 3.7502946853637695, "learning_rate": 8.823e-06, "loss": 0.0516, "step": 5886 }, { "epoch": 0.3138561603667964, "grad_norm": 3.1657917499542236, "learning_rate": 8.8228e-06, "loss": 0.0632, "step": 5887 }, { "epoch": 0.31390947379644935, "grad_norm": 4.3614935874938965, "learning_rate": 8.822600000000001e-06, "loss": 0.0432, "step": 5888 }, { "epoch": 0.31396278722610227, "grad_norm": 4.704006195068359, "learning_rate": 8.8224e-06, "loss": 0.0421, "step": 5889 }, { "epoch": 0.3140161006557552, "grad_norm": 2.1161575317382812, "learning_rate": 8.8222e-06, "loss": 0.0279, "step": 5890 }, { "epoch": 0.3140694140854081, "grad_norm": 2.827521800994873, "learning_rate": 8.822000000000001e-06, "loss": 0.0388, "step": 5891 }, { "epoch": 0.31412272751506104, "grad_norm": 3.105557441711426, "learning_rate": 8.8218e-06, "loss": 0.0314, "step": 5892 }, { "epoch": 0.31417604094471396, "grad_norm": 3.5559439659118652, "learning_rate": 8.821600000000002e-06, "loss": 0.0439, "step": 5893 }, { "epoch": 0.3142293543743669, "grad_norm": 2.911346673965454, "learning_rate": 8.821400000000002e-06, "loss": 0.0488, "step": 5894 }, { "epoch": 0.3142826678040198, "grad_norm": 3.5192880630493164, "learning_rate": 8.821200000000001e-06, "loss": 0.0476, "step": 5895 }, { "epoch": 0.3143359812336728, "grad_norm": 3.0129382610321045, "learning_rate": 8.821e-06, "loss": 0.0231, "step": 5896 }, { "epoch": 0.3143892946633257, "grad_norm": 2.898820400238037, "learning_rate": 8.8208e-06, "loss": 0.0425, "step": 5897 }, { "epoch": 0.31444260809297864, "grad_norm": 3.5176141262054443, "learning_rate": 8.8206e-06, "loss": 0.064, "step": 5898 }, { "epoch": 0.31449592152263156, "grad_norm": 4.055974006652832, "learning_rate": 8.820400000000001e-06, "loss": 0.0573, "step": 5899 }, { "epoch": 0.3145492349522845, "grad_norm": 10.938155174255371, "learning_rate": 8.8202e-06, "loss": 0.1068, "step": 5900 }, { "epoch": 0.3146025483819374, "grad_norm": 3.5500433444976807, "learning_rate": 8.82e-06, "loss": 0.0373, "step": 5901 }, { "epoch": 0.31465586181159033, "grad_norm": 5.005517482757568, "learning_rate": 8.8198e-06, "loss": 0.0327, "step": 5902 }, { "epoch": 0.31470917524124326, "grad_norm": 5.261180400848389, "learning_rate": 8.819600000000001e-06, "loss": 0.0463, "step": 5903 }, { "epoch": 0.3147624886708962, "grad_norm": 9.14316177368164, "learning_rate": 8.8194e-06, "loss": 0.069, "step": 5904 }, { "epoch": 0.3148158021005491, "grad_norm": 3.0292484760284424, "learning_rate": 8.819200000000002e-06, "loss": 0.0549, "step": 5905 }, { "epoch": 0.3148691155302021, "grad_norm": 3.41127872467041, "learning_rate": 8.819000000000001e-06, "loss": 0.041, "step": 5906 }, { "epoch": 0.314922428959855, "grad_norm": 1.9173431396484375, "learning_rate": 8.818800000000001e-06, "loss": 0.0219, "step": 5907 }, { "epoch": 0.31497574238950793, "grad_norm": 2.8011980056762695, "learning_rate": 8.8186e-06, "loss": 0.0442, "step": 5908 }, { "epoch": 0.31502905581916085, "grad_norm": 5.42430305480957, "learning_rate": 8.8184e-06, "loss": 0.0553, "step": 5909 }, { "epoch": 0.3150823692488138, "grad_norm": 2.4984207153320312, "learning_rate": 8.818200000000001e-06, "loss": 0.0276, "step": 5910 }, { "epoch": 0.3151356826784667, "grad_norm": 6.922295093536377, "learning_rate": 8.818000000000001e-06, "loss": 0.0366, "step": 5911 }, { "epoch": 0.3151889961081196, "grad_norm": 3.264775514602661, "learning_rate": 8.8178e-06, "loss": 0.0319, "step": 5912 }, { "epoch": 0.31524230953777255, "grad_norm": 4.792294979095459, "learning_rate": 8.8176e-06, "loss": 0.0658, "step": 5913 }, { "epoch": 0.31529562296742547, "grad_norm": 2.6893606185913086, "learning_rate": 8.8174e-06, "loss": 0.0383, "step": 5914 }, { "epoch": 0.31534893639707845, "grad_norm": 3.705495834350586, "learning_rate": 8.817200000000001e-06, "loss": 0.0337, "step": 5915 }, { "epoch": 0.3154022498267314, "grad_norm": 4.17851448059082, "learning_rate": 8.817e-06, "loss": 0.0339, "step": 5916 }, { "epoch": 0.3154555632563843, "grad_norm": 7.222530364990234, "learning_rate": 8.816800000000002e-06, "loss": 0.0781, "step": 5917 }, { "epoch": 0.3155088766860372, "grad_norm": 4.528781414031982, "learning_rate": 8.816600000000001e-06, "loss": 0.0502, "step": 5918 }, { "epoch": 0.31556219011569014, "grad_norm": 2.772531509399414, "learning_rate": 8.8164e-06, "loss": 0.0595, "step": 5919 }, { "epoch": 0.31561550354534307, "grad_norm": 1.9896208047866821, "learning_rate": 8.8162e-06, "loss": 0.0265, "step": 5920 }, { "epoch": 0.315668816974996, "grad_norm": 3.5500500202178955, "learning_rate": 8.816000000000002e-06, "loss": 0.0566, "step": 5921 }, { "epoch": 0.3157221304046489, "grad_norm": 1.4912326335906982, "learning_rate": 8.815800000000001e-06, "loss": 0.0242, "step": 5922 }, { "epoch": 0.31577544383430184, "grad_norm": 3.310903549194336, "learning_rate": 8.8156e-06, "loss": 0.0373, "step": 5923 }, { "epoch": 0.31582875726395476, "grad_norm": 6.372254848480225, "learning_rate": 8.8154e-06, "loss": 0.058, "step": 5924 }, { "epoch": 0.31588207069360774, "grad_norm": 2.497929573059082, "learning_rate": 8.8152e-06, "loss": 0.0346, "step": 5925 }, { "epoch": 0.31593538412326067, "grad_norm": 7.715010166168213, "learning_rate": 8.815e-06, "loss": 0.0653, "step": 5926 }, { "epoch": 0.3159886975529136, "grad_norm": 3.030949115753174, "learning_rate": 8.8148e-06, "loss": 0.0521, "step": 5927 }, { "epoch": 0.3160420109825665, "grad_norm": 4.32387113571167, "learning_rate": 8.8146e-06, "loss": 0.0355, "step": 5928 }, { "epoch": 0.31609532441221944, "grad_norm": 5.173714637756348, "learning_rate": 8.814400000000002e-06, "loss": 0.0512, "step": 5929 }, { "epoch": 0.31614863784187236, "grad_norm": 3.4702324867248535, "learning_rate": 8.814200000000001e-06, "loss": 0.048, "step": 5930 }, { "epoch": 0.3162019512715253, "grad_norm": 3.660731554031372, "learning_rate": 8.814e-06, "loss": 0.0417, "step": 5931 }, { "epoch": 0.3162552647011782, "grad_norm": 3.682767629623413, "learning_rate": 8.813800000000002e-06, "loss": 0.0559, "step": 5932 }, { "epoch": 0.31630857813083113, "grad_norm": 2.4691872596740723, "learning_rate": 8.813600000000001e-06, "loss": 0.0406, "step": 5933 }, { "epoch": 0.3163618915604841, "grad_norm": 4.800954818725586, "learning_rate": 8.813400000000001e-06, "loss": 0.0653, "step": 5934 }, { "epoch": 0.31641520499013703, "grad_norm": 7.167446136474609, "learning_rate": 8.8132e-06, "loss": 0.0841, "step": 5935 }, { "epoch": 0.31646851841978996, "grad_norm": 4.164644718170166, "learning_rate": 8.813e-06, "loss": 0.0382, "step": 5936 }, { "epoch": 0.3165218318494429, "grad_norm": 2.5951521396636963, "learning_rate": 8.8128e-06, "loss": 0.0235, "step": 5937 }, { "epoch": 0.3165751452790958, "grad_norm": 2.441134214401245, "learning_rate": 8.812600000000001e-06, "loss": 0.0439, "step": 5938 }, { "epoch": 0.31662845870874873, "grad_norm": 2.5798394680023193, "learning_rate": 8.8124e-06, "loss": 0.0333, "step": 5939 }, { "epoch": 0.31668177213840165, "grad_norm": 2.4296956062316895, "learning_rate": 8.8122e-06, "loss": 0.0301, "step": 5940 }, { "epoch": 0.3167350855680546, "grad_norm": 2.9386892318725586, "learning_rate": 8.812000000000001e-06, "loss": 0.0437, "step": 5941 }, { "epoch": 0.3167883989977075, "grad_norm": 4.669655799865723, "learning_rate": 8.811800000000001e-06, "loss": 0.0372, "step": 5942 }, { "epoch": 0.3168417124273605, "grad_norm": 6.526830196380615, "learning_rate": 8.8116e-06, "loss": 0.0834, "step": 5943 }, { "epoch": 0.3168950258570134, "grad_norm": 2.4366300106048584, "learning_rate": 8.811400000000002e-06, "loss": 0.0202, "step": 5944 }, { "epoch": 0.3169483392866663, "grad_norm": 4.184478759765625, "learning_rate": 8.811200000000001e-06, "loss": 0.0628, "step": 5945 }, { "epoch": 0.31700165271631925, "grad_norm": 4.33701753616333, "learning_rate": 8.811000000000001e-06, "loss": 0.0641, "step": 5946 }, { "epoch": 0.3170549661459722, "grad_norm": 3.58682918548584, "learning_rate": 8.8108e-06, "loss": 0.052, "step": 5947 }, { "epoch": 0.3171082795756251, "grad_norm": 4.687028408050537, "learning_rate": 8.8106e-06, "loss": 0.0656, "step": 5948 }, { "epoch": 0.317161593005278, "grad_norm": 2.745565414428711, "learning_rate": 8.810400000000001e-06, "loss": 0.0456, "step": 5949 }, { "epoch": 0.31721490643493094, "grad_norm": 5.160040378570557, "learning_rate": 8.8102e-06, "loss": 0.0686, "step": 5950 }, { "epoch": 0.31726821986458387, "grad_norm": 2.552107572555542, "learning_rate": 8.81e-06, "loss": 0.0261, "step": 5951 }, { "epoch": 0.3173215332942368, "grad_norm": 4.194762229919434, "learning_rate": 8.8098e-06, "loss": 0.0322, "step": 5952 }, { "epoch": 0.31737484672388977, "grad_norm": 3.4537909030914307, "learning_rate": 8.809600000000001e-06, "loss": 0.0473, "step": 5953 }, { "epoch": 0.3174281601535427, "grad_norm": 1.8723666667938232, "learning_rate": 8.8094e-06, "loss": 0.0243, "step": 5954 }, { "epoch": 0.3174814735831956, "grad_norm": 2.6262383460998535, "learning_rate": 8.8092e-06, "loss": 0.0239, "step": 5955 }, { "epoch": 0.31753478701284854, "grad_norm": 2.908310651779175, "learning_rate": 8.809000000000002e-06, "loss": 0.0234, "step": 5956 }, { "epoch": 0.31758810044250146, "grad_norm": 1.8946396112442017, "learning_rate": 8.808800000000001e-06, "loss": 0.0295, "step": 5957 }, { "epoch": 0.3176414138721544, "grad_norm": 3.596226692199707, "learning_rate": 8.8086e-06, "loss": 0.0528, "step": 5958 }, { "epoch": 0.3176947273018073, "grad_norm": 6.359073162078857, "learning_rate": 8.8084e-06, "loss": 0.0215, "step": 5959 }, { "epoch": 0.31774804073146024, "grad_norm": 5.1427202224731445, "learning_rate": 8.808200000000001e-06, "loss": 0.0787, "step": 5960 }, { "epoch": 0.31780135416111316, "grad_norm": 4.030384063720703, "learning_rate": 8.808000000000001e-06, "loss": 0.0465, "step": 5961 }, { "epoch": 0.31785466759076614, "grad_norm": 3.8443915843963623, "learning_rate": 8.8078e-06, "loss": 0.0297, "step": 5962 }, { "epoch": 0.31790798102041906, "grad_norm": 4.502450942993164, "learning_rate": 8.8076e-06, "loss": 0.0649, "step": 5963 }, { "epoch": 0.317961294450072, "grad_norm": 4.2275848388671875, "learning_rate": 8.8074e-06, "loss": 0.0474, "step": 5964 }, { "epoch": 0.3180146078797249, "grad_norm": 3.6306471824645996, "learning_rate": 8.807200000000001e-06, "loss": 0.0311, "step": 5965 }, { "epoch": 0.31806792130937783, "grad_norm": 3.0077555179595947, "learning_rate": 8.807e-06, "loss": 0.0356, "step": 5966 }, { "epoch": 0.31812123473903076, "grad_norm": 3.8644769191741943, "learning_rate": 8.8068e-06, "loss": 0.0429, "step": 5967 }, { "epoch": 0.3181745481686837, "grad_norm": 7.08347749710083, "learning_rate": 8.806600000000001e-06, "loss": 0.0437, "step": 5968 }, { "epoch": 0.3182278615983366, "grad_norm": 4.436577796936035, "learning_rate": 8.806400000000001e-06, "loss": 0.0633, "step": 5969 }, { "epoch": 0.3182811750279895, "grad_norm": 5.329939365386963, "learning_rate": 8.8062e-06, "loss": 0.0615, "step": 5970 }, { "epoch": 0.3183344884576425, "grad_norm": 4.081998348236084, "learning_rate": 8.806000000000002e-06, "loss": 0.0419, "step": 5971 }, { "epoch": 0.31838780188729543, "grad_norm": 4.034533500671387, "learning_rate": 8.805800000000001e-06, "loss": 0.0394, "step": 5972 }, { "epoch": 0.31844111531694835, "grad_norm": 4.552614688873291, "learning_rate": 8.805600000000001e-06, "loss": 0.0687, "step": 5973 }, { "epoch": 0.3184944287466013, "grad_norm": 2.054865837097168, "learning_rate": 8.8054e-06, "loss": 0.0189, "step": 5974 }, { "epoch": 0.3185477421762542, "grad_norm": 3.059443473815918, "learning_rate": 8.8052e-06, "loss": 0.0255, "step": 5975 }, { "epoch": 0.3186010556059071, "grad_norm": 5.789629936218262, "learning_rate": 8.805e-06, "loss": 0.0629, "step": 5976 }, { "epoch": 0.31865436903556005, "grad_norm": 2.0673720836639404, "learning_rate": 8.8048e-06, "loss": 0.0442, "step": 5977 }, { "epoch": 0.31870768246521297, "grad_norm": 3.33443021774292, "learning_rate": 8.8046e-06, "loss": 0.0152, "step": 5978 }, { "epoch": 0.3187609958948659, "grad_norm": 2.385266065597534, "learning_rate": 8.8044e-06, "loss": 0.0271, "step": 5979 }, { "epoch": 0.3188143093245188, "grad_norm": 2.3057470321655273, "learning_rate": 8.804200000000001e-06, "loss": 0.0286, "step": 5980 }, { "epoch": 0.3188676227541718, "grad_norm": 2.3593649864196777, "learning_rate": 8.804e-06, "loss": 0.0153, "step": 5981 }, { "epoch": 0.3189209361838247, "grad_norm": 5.350549697875977, "learning_rate": 8.803800000000002e-06, "loss": 0.0581, "step": 5982 }, { "epoch": 0.31897424961347765, "grad_norm": 1.792648434638977, "learning_rate": 8.803600000000002e-06, "loss": 0.0342, "step": 5983 }, { "epoch": 0.31902756304313057, "grad_norm": 2.426640033721924, "learning_rate": 8.803400000000001e-06, "loss": 0.0357, "step": 5984 }, { "epoch": 0.3190808764727835, "grad_norm": 5.249068737030029, "learning_rate": 8.8032e-06, "loss": 0.0629, "step": 5985 }, { "epoch": 0.3191341899024364, "grad_norm": 3.1927175521850586, "learning_rate": 8.803e-06, "loss": 0.0342, "step": 5986 }, { "epoch": 0.31918750333208934, "grad_norm": 4.658799171447754, "learning_rate": 8.8028e-06, "loss": 0.0251, "step": 5987 }, { "epoch": 0.31924081676174226, "grad_norm": 3.018174648284912, "learning_rate": 8.802600000000001e-06, "loss": 0.0487, "step": 5988 }, { "epoch": 0.3192941301913952, "grad_norm": 3.5665669441223145, "learning_rate": 8.8024e-06, "loss": 0.0431, "step": 5989 }, { "epoch": 0.31934744362104817, "grad_norm": 6.283122539520264, "learning_rate": 8.8022e-06, "loss": 0.085, "step": 5990 }, { "epoch": 0.3194007570507011, "grad_norm": 3.4270153045654297, "learning_rate": 8.802e-06, "loss": 0.0232, "step": 5991 }, { "epoch": 0.319454070480354, "grad_norm": 3.6976585388183594, "learning_rate": 8.801800000000001e-06, "loss": 0.0389, "step": 5992 }, { "epoch": 0.31950738391000694, "grad_norm": 4.499184608459473, "learning_rate": 8.8016e-06, "loss": 0.0305, "step": 5993 }, { "epoch": 0.31956069733965986, "grad_norm": 4.31993293762207, "learning_rate": 8.801400000000002e-06, "loss": 0.043, "step": 5994 }, { "epoch": 0.3196140107693128, "grad_norm": 3.0285115242004395, "learning_rate": 8.801200000000001e-06, "loss": 0.0427, "step": 5995 }, { "epoch": 0.3196673241989657, "grad_norm": 3.5668389797210693, "learning_rate": 8.801000000000001e-06, "loss": 0.03, "step": 5996 }, { "epoch": 0.31972063762861863, "grad_norm": 5.933322906494141, "learning_rate": 8.8008e-06, "loss": 0.0875, "step": 5997 }, { "epoch": 0.31977395105827155, "grad_norm": 6.124509811401367, "learning_rate": 8.8006e-06, "loss": 0.0782, "step": 5998 }, { "epoch": 0.31982726448792453, "grad_norm": 1.7007231712341309, "learning_rate": 8.800400000000001e-06, "loss": 0.0203, "step": 5999 }, { "epoch": 0.31988057791757746, "grad_norm": 2.704414129257202, "learning_rate": 8.800200000000001e-06, "loss": 0.0532, "step": 6000 }, { "epoch": 0.3199338913472304, "grad_norm": 5.380077838897705, "learning_rate": 8.8e-06, "loss": 0.0647, "step": 6001 }, { "epoch": 0.3199872047768833, "grad_norm": 3.4175050258636475, "learning_rate": 8.7998e-06, "loss": 0.0525, "step": 6002 }, { "epoch": 0.32004051820653623, "grad_norm": 4.262737274169922, "learning_rate": 8.7996e-06, "loss": 0.0472, "step": 6003 }, { "epoch": 0.32009383163618915, "grad_norm": 6.9748005867004395, "learning_rate": 8.7994e-06, "loss": 0.0482, "step": 6004 }, { "epoch": 0.3201471450658421, "grad_norm": 7.7261552810668945, "learning_rate": 8.7992e-06, "loss": 0.1156, "step": 6005 }, { "epoch": 0.320200458495495, "grad_norm": 3.660006523132324, "learning_rate": 8.799000000000002e-06, "loss": 0.0461, "step": 6006 }, { "epoch": 0.3202537719251479, "grad_norm": 3.3989946842193604, "learning_rate": 8.798800000000001e-06, "loss": 0.0291, "step": 6007 }, { "epoch": 0.32030708535480085, "grad_norm": 4.271456241607666, "learning_rate": 8.7986e-06, "loss": 0.0576, "step": 6008 }, { "epoch": 0.3203603987844538, "grad_norm": 1.8170650005340576, "learning_rate": 8.7984e-06, "loss": 0.0144, "step": 6009 }, { "epoch": 0.32041371221410675, "grad_norm": 1.6659941673278809, "learning_rate": 8.798200000000002e-06, "loss": 0.0185, "step": 6010 }, { "epoch": 0.3204670256437597, "grad_norm": 5.490015029907227, "learning_rate": 8.798000000000001e-06, "loss": 0.0863, "step": 6011 }, { "epoch": 0.3205203390734126, "grad_norm": 2.349757432937622, "learning_rate": 8.7978e-06, "loss": 0.0312, "step": 6012 }, { "epoch": 0.3205736525030655, "grad_norm": 3.8605616092681885, "learning_rate": 8.7976e-06, "loss": 0.0669, "step": 6013 }, { "epoch": 0.32062696593271844, "grad_norm": 2.3472049236297607, "learning_rate": 8.7974e-06, "loss": 0.0273, "step": 6014 }, { "epoch": 0.32068027936237137, "grad_norm": 4.302341461181641, "learning_rate": 8.7972e-06, "loss": 0.0446, "step": 6015 }, { "epoch": 0.3207335927920243, "grad_norm": 3.2421605587005615, "learning_rate": 8.797e-06, "loss": 0.05, "step": 6016 }, { "epoch": 0.3207869062216772, "grad_norm": 2.3208463191986084, "learning_rate": 8.7968e-06, "loss": 0.0161, "step": 6017 }, { "epoch": 0.3208402196513302, "grad_norm": 4.021702766418457, "learning_rate": 8.796600000000001e-06, "loss": 0.0613, "step": 6018 }, { "epoch": 0.3208935330809831, "grad_norm": 3.1536247730255127, "learning_rate": 8.796400000000001e-06, "loss": 0.0543, "step": 6019 }, { "epoch": 0.32094684651063604, "grad_norm": 1.736437201499939, "learning_rate": 8.7962e-06, "loss": 0.0212, "step": 6020 }, { "epoch": 0.32100015994028896, "grad_norm": 3.2989895343780518, "learning_rate": 8.796000000000002e-06, "loss": 0.0516, "step": 6021 }, { "epoch": 0.3210534733699419, "grad_norm": 2.7784128189086914, "learning_rate": 8.795800000000001e-06, "loss": 0.0362, "step": 6022 }, { "epoch": 0.3211067867995948, "grad_norm": 2.599382162094116, "learning_rate": 8.795600000000001e-06, "loss": 0.0421, "step": 6023 }, { "epoch": 0.32116010022924774, "grad_norm": 4.575443267822266, "learning_rate": 8.7954e-06, "loss": 0.0574, "step": 6024 }, { "epoch": 0.32121341365890066, "grad_norm": 4.668063163757324, "learning_rate": 8.7952e-06, "loss": 0.091, "step": 6025 }, { "epoch": 0.3212667270885536, "grad_norm": 2.1152257919311523, "learning_rate": 8.795e-06, "loss": 0.0263, "step": 6026 }, { "epoch": 0.32132004051820656, "grad_norm": 4.218199729919434, "learning_rate": 8.794800000000001e-06, "loss": 0.0432, "step": 6027 }, { "epoch": 0.3213733539478595, "grad_norm": 7.59244441986084, "learning_rate": 8.7946e-06, "loss": 0.0564, "step": 6028 }, { "epoch": 0.3214266673775124, "grad_norm": 5.1862101554870605, "learning_rate": 8.7944e-06, "loss": 0.0524, "step": 6029 }, { "epoch": 0.32147998080716533, "grad_norm": 3.401874542236328, "learning_rate": 8.794200000000001e-06, "loss": 0.0774, "step": 6030 }, { "epoch": 0.32153329423681826, "grad_norm": 2.47340726852417, "learning_rate": 8.794e-06, "loss": 0.0463, "step": 6031 }, { "epoch": 0.3215866076664712, "grad_norm": 3.4678869247436523, "learning_rate": 8.7938e-06, "loss": 0.02, "step": 6032 }, { "epoch": 0.3216399210961241, "grad_norm": 4.380055904388428, "learning_rate": 8.793600000000002e-06, "loss": 0.0596, "step": 6033 }, { "epoch": 0.321693234525777, "grad_norm": 1.8725078105926514, "learning_rate": 8.793400000000001e-06, "loss": 0.0179, "step": 6034 }, { "epoch": 0.32174654795542995, "grad_norm": 2.5985114574432373, "learning_rate": 8.7932e-06, "loss": 0.0242, "step": 6035 }, { "epoch": 0.3217998613850829, "grad_norm": 3.8481507301330566, "learning_rate": 8.793e-06, "loss": 0.0427, "step": 6036 }, { "epoch": 0.32185317481473585, "grad_norm": 2.5593783855438232, "learning_rate": 8.7928e-06, "loss": 0.0505, "step": 6037 }, { "epoch": 0.3219064882443888, "grad_norm": 3.9668784141540527, "learning_rate": 8.792600000000001e-06, "loss": 0.026, "step": 6038 }, { "epoch": 0.3219598016740417, "grad_norm": 1.5161174535751343, "learning_rate": 8.7924e-06, "loss": 0.0201, "step": 6039 }, { "epoch": 0.3220131151036946, "grad_norm": 4.205483913421631, "learning_rate": 8.7922e-06, "loss": 0.0806, "step": 6040 }, { "epoch": 0.32206642853334755, "grad_norm": 3.928971529006958, "learning_rate": 8.792e-06, "loss": 0.0561, "step": 6041 }, { "epoch": 0.32211974196300047, "grad_norm": 5.254323482513428, "learning_rate": 8.791800000000001e-06, "loss": 0.0416, "step": 6042 }, { "epoch": 0.3221730553926534, "grad_norm": 2.697099447250366, "learning_rate": 8.7916e-06, "loss": 0.0306, "step": 6043 }, { "epoch": 0.3222263688223063, "grad_norm": 3.0106542110443115, "learning_rate": 8.7914e-06, "loss": 0.0466, "step": 6044 }, { "epoch": 0.32227968225195924, "grad_norm": 2.2875027656555176, "learning_rate": 8.791200000000001e-06, "loss": 0.0389, "step": 6045 }, { "epoch": 0.3223329956816122, "grad_norm": 3.8864564895629883, "learning_rate": 8.791000000000001e-06, "loss": 0.0322, "step": 6046 }, { "epoch": 0.32238630911126515, "grad_norm": 2.409151077270508, "learning_rate": 8.7908e-06, "loss": 0.0454, "step": 6047 }, { "epoch": 0.32243962254091807, "grad_norm": 3.5187957286834717, "learning_rate": 8.7906e-06, "loss": 0.0461, "step": 6048 }, { "epoch": 0.322492935970571, "grad_norm": 5.132418632507324, "learning_rate": 8.790400000000001e-06, "loss": 0.0395, "step": 6049 }, { "epoch": 0.3225462494002239, "grad_norm": 5.427480220794678, "learning_rate": 8.790200000000001e-06, "loss": 0.0714, "step": 6050 }, { "epoch": 0.32259956282987684, "grad_norm": 1.535745620727539, "learning_rate": 8.79e-06, "loss": 0.0178, "step": 6051 }, { "epoch": 0.32265287625952976, "grad_norm": 2.520423650741577, "learning_rate": 8.7898e-06, "loss": 0.0251, "step": 6052 }, { "epoch": 0.3227061896891827, "grad_norm": 4.263155937194824, "learning_rate": 8.7896e-06, "loss": 0.0483, "step": 6053 }, { "epoch": 0.3227595031188356, "grad_norm": 3.785493850708008, "learning_rate": 8.789400000000001e-06, "loss": 0.0676, "step": 6054 }, { "epoch": 0.3228128165484886, "grad_norm": 5.762596130371094, "learning_rate": 8.7892e-06, "loss": 0.0641, "step": 6055 }, { "epoch": 0.3228661299781415, "grad_norm": 2.228135347366333, "learning_rate": 8.789e-06, "loss": 0.0362, "step": 6056 }, { "epoch": 0.32291944340779444, "grad_norm": 4.067672252655029, "learning_rate": 8.788800000000001e-06, "loss": 0.0479, "step": 6057 }, { "epoch": 0.32297275683744736, "grad_norm": 6.362580299377441, "learning_rate": 8.788600000000001e-06, "loss": 0.0681, "step": 6058 }, { "epoch": 0.3230260702671003, "grad_norm": 2.335814952850342, "learning_rate": 8.7884e-06, "loss": 0.0257, "step": 6059 }, { "epoch": 0.3230793836967532, "grad_norm": 2.3497414588928223, "learning_rate": 8.788200000000002e-06, "loss": 0.0202, "step": 6060 }, { "epoch": 0.32313269712640613, "grad_norm": 3.615036964416504, "learning_rate": 8.788000000000001e-06, "loss": 0.0374, "step": 6061 }, { "epoch": 0.32318601055605906, "grad_norm": 4.21600341796875, "learning_rate": 8.7878e-06, "loss": 0.0452, "step": 6062 }, { "epoch": 0.323239323985712, "grad_norm": 3.851950168609619, "learning_rate": 8.7876e-06, "loss": 0.0631, "step": 6063 }, { "epoch": 0.3232926374153649, "grad_norm": 4.787542819976807, "learning_rate": 8.7874e-06, "loss": 0.0679, "step": 6064 }, { "epoch": 0.3233459508450179, "grad_norm": 3.6108765602111816, "learning_rate": 8.7872e-06, "loss": 0.0496, "step": 6065 }, { "epoch": 0.3233992642746708, "grad_norm": 5.614781379699707, "learning_rate": 8.787e-06, "loss": 0.0858, "step": 6066 }, { "epoch": 0.32345257770432373, "grad_norm": 4.137618541717529, "learning_rate": 8.7868e-06, "loss": 0.0474, "step": 6067 }, { "epoch": 0.32350589113397665, "grad_norm": 2.1277823448181152, "learning_rate": 8.7866e-06, "loss": 0.0232, "step": 6068 }, { "epoch": 0.3235592045636296, "grad_norm": 2.3361456394195557, "learning_rate": 8.786400000000001e-06, "loss": 0.0301, "step": 6069 }, { "epoch": 0.3236125179932825, "grad_norm": 2.94368839263916, "learning_rate": 8.7862e-06, "loss": 0.0351, "step": 6070 }, { "epoch": 0.3236658314229354, "grad_norm": 5.798233509063721, "learning_rate": 8.786000000000002e-06, "loss": 0.0607, "step": 6071 }, { "epoch": 0.32371914485258835, "grad_norm": 2.4999616146087646, "learning_rate": 8.785800000000001e-06, "loss": 0.0297, "step": 6072 }, { "epoch": 0.32377245828224127, "grad_norm": 2.9874236583709717, "learning_rate": 8.785600000000001e-06, "loss": 0.0463, "step": 6073 }, { "epoch": 0.32382577171189425, "grad_norm": 3.663142681121826, "learning_rate": 8.7854e-06, "loss": 0.0468, "step": 6074 }, { "epoch": 0.3238790851415472, "grad_norm": 2.387929916381836, "learning_rate": 8.7852e-06, "loss": 0.0594, "step": 6075 }, { "epoch": 0.3239323985712001, "grad_norm": 1.7998239994049072, "learning_rate": 8.785e-06, "loss": 0.0291, "step": 6076 }, { "epoch": 0.323985712000853, "grad_norm": 2.980222463607788, "learning_rate": 8.784800000000001e-06, "loss": 0.0493, "step": 6077 }, { "epoch": 0.32403902543050594, "grad_norm": 1.6197400093078613, "learning_rate": 8.7846e-06, "loss": 0.0314, "step": 6078 }, { "epoch": 0.32409233886015887, "grad_norm": 1.5436067581176758, "learning_rate": 8.7844e-06, "loss": 0.0347, "step": 6079 }, { "epoch": 0.3241456522898118, "grad_norm": 4.257120609283447, "learning_rate": 8.7842e-06, "loss": 0.0485, "step": 6080 }, { "epoch": 0.3241989657194647, "grad_norm": 3.1306071281433105, "learning_rate": 8.784000000000001e-06, "loss": 0.0524, "step": 6081 }, { "epoch": 0.32425227914911764, "grad_norm": 2.930849313735962, "learning_rate": 8.7838e-06, "loss": 0.0404, "step": 6082 }, { "epoch": 0.3243055925787706, "grad_norm": 3.2125403881073, "learning_rate": 8.783600000000002e-06, "loss": 0.057, "step": 6083 }, { "epoch": 0.32435890600842354, "grad_norm": 3.1992037296295166, "learning_rate": 8.783400000000001e-06, "loss": 0.0564, "step": 6084 }, { "epoch": 0.32441221943807647, "grad_norm": 2.961367130279541, "learning_rate": 8.783200000000001e-06, "loss": 0.0387, "step": 6085 }, { "epoch": 0.3244655328677294, "grad_norm": 4.92563533782959, "learning_rate": 8.783e-06, "loss": 0.027, "step": 6086 }, { "epoch": 0.3245188462973823, "grad_norm": 2.1272759437561035, "learning_rate": 8.7828e-06, "loss": 0.0291, "step": 6087 }, { "epoch": 0.32457215972703524, "grad_norm": 2.799276113510132, "learning_rate": 8.782600000000001e-06, "loss": 0.0523, "step": 6088 }, { "epoch": 0.32462547315668816, "grad_norm": 3.971900463104248, "learning_rate": 8.7824e-06, "loss": 0.048, "step": 6089 }, { "epoch": 0.3246787865863411, "grad_norm": 2.418686628341675, "learning_rate": 8.7822e-06, "loss": 0.0225, "step": 6090 }, { "epoch": 0.324732100015994, "grad_norm": 2.2499706745147705, "learning_rate": 8.782e-06, "loss": 0.0215, "step": 6091 }, { "epoch": 0.32478541344564693, "grad_norm": 2.4800190925598145, "learning_rate": 8.7818e-06, "loss": 0.0448, "step": 6092 }, { "epoch": 0.3248387268752999, "grad_norm": 2.181163787841797, "learning_rate": 8.7816e-06, "loss": 0.0344, "step": 6093 }, { "epoch": 0.32489204030495283, "grad_norm": 4.132284164428711, "learning_rate": 8.7814e-06, "loss": 0.0404, "step": 6094 }, { "epoch": 0.32494535373460576, "grad_norm": 3.121425151824951, "learning_rate": 8.781200000000002e-06, "loss": 0.0332, "step": 6095 }, { "epoch": 0.3249986671642587, "grad_norm": 2.5222020149230957, "learning_rate": 8.781000000000001e-06, "loss": 0.0366, "step": 6096 }, { "epoch": 0.3250519805939116, "grad_norm": 2.7209062576293945, "learning_rate": 8.7808e-06, "loss": 0.0247, "step": 6097 }, { "epoch": 0.32510529402356453, "grad_norm": 2.8648316860198975, "learning_rate": 8.7806e-06, "loss": 0.0183, "step": 6098 }, { "epoch": 0.32515860745321745, "grad_norm": 3.3081467151641846, "learning_rate": 8.780400000000002e-06, "loss": 0.0204, "step": 6099 }, { "epoch": 0.3252119208828704, "grad_norm": 5.327517032623291, "learning_rate": 8.780200000000001e-06, "loss": 0.0516, "step": 6100 }, { "epoch": 0.3252652343125233, "grad_norm": 5.424770832061768, "learning_rate": 8.78e-06, "loss": 0.0374, "step": 6101 }, { "epoch": 0.3253185477421763, "grad_norm": 1.950871467590332, "learning_rate": 8.7798e-06, "loss": 0.0163, "step": 6102 }, { "epoch": 0.3253718611718292, "grad_norm": 1.656998634338379, "learning_rate": 8.7796e-06, "loss": 0.0105, "step": 6103 }, { "epoch": 0.3254251746014821, "grad_norm": 3.292043685913086, "learning_rate": 8.779400000000001e-06, "loss": 0.0346, "step": 6104 }, { "epoch": 0.32547848803113505, "grad_norm": 3.2111990451812744, "learning_rate": 8.7792e-06, "loss": 0.0301, "step": 6105 }, { "epoch": 0.325531801460788, "grad_norm": 3.9524312019348145, "learning_rate": 8.779e-06, "loss": 0.0444, "step": 6106 }, { "epoch": 0.3255851148904409, "grad_norm": 3.447554349899292, "learning_rate": 8.778800000000001e-06, "loss": 0.0332, "step": 6107 }, { "epoch": 0.3256384283200938, "grad_norm": 6.69636869430542, "learning_rate": 8.778600000000001e-06, "loss": 0.0833, "step": 6108 }, { "epoch": 0.32569174174974674, "grad_norm": 6.136996746063232, "learning_rate": 8.7784e-06, "loss": 0.0606, "step": 6109 }, { "epoch": 0.32574505517939967, "grad_norm": 3.44973087310791, "learning_rate": 8.778200000000002e-06, "loss": 0.0456, "step": 6110 }, { "epoch": 0.32579836860905265, "grad_norm": 5.415493011474609, "learning_rate": 8.778000000000001e-06, "loss": 0.0418, "step": 6111 }, { "epoch": 0.32585168203870557, "grad_norm": 2.7596161365509033, "learning_rate": 8.777800000000001e-06, "loss": 0.0169, "step": 6112 }, { "epoch": 0.3259049954683585, "grad_norm": 4.151577949523926, "learning_rate": 8.7776e-06, "loss": 0.0538, "step": 6113 }, { "epoch": 0.3259583088980114, "grad_norm": 6.352422714233398, "learning_rate": 8.7774e-06, "loss": 0.0752, "step": 6114 }, { "epoch": 0.32601162232766434, "grad_norm": 3.811666965484619, "learning_rate": 8.7772e-06, "loss": 0.0642, "step": 6115 }, { "epoch": 0.32606493575731726, "grad_norm": 3.976985454559326, "learning_rate": 8.777e-06, "loss": 0.0399, "step": 6116 }, { "epoch": 0.3261182491869702, "grad_norm": 4.1776862144470215, "learning_rate": 8.7768e-06, "loss": 0.0431, "step": 6117 }, { "epoch": 0.3261715626166231, "grad_norm": 3.7123465538024902, "learning_rate": 8.7766e-06, "loss": 0.0439, "step": 6118 }, { "epoch": 0.32622487604627604, "grad_norm": 3.386948347091675, "learning_rate": 8.776400000000001e-06, "loss": 0.071, "step": 6119 }, { "epoch": 0.32627818947592896, "grad_norm": 4.009329795837402, "learning_rate": 8.7762e-06, "loss": 0.0453, "step": 6120 }, { "epoch": 0.32633150290558194, "grad_norm": 3.447096347808838, "learning_rate": 8.776e-06, "loss": 0.0494, "step": 6121 }, { "epoch": 0.32638481633523486, "grad_norm": 4.609971523284912, "learning_rate": 8.775800000000002e-06, "loss": 0.0721, "step": 6122 }, { "epoch": 0.3264381297648878, "grad_norm": 3.1463825702667236, "learning_rate": 8.775600000000001e-06, "loss": 0.044, "step": 6123 }, { "epoch": 0.3264914431945407, "grad_norm": 6.463651180267334, "learning_rate": 8.7754e-06, "loss": 0.0593, "step": 6124 }, { "epoch": 0.32654475662419363, "grad_norm": 4.163881778717041, "learning_rate": 8.7752e-06, "loss": 0.042, "step": 6125 }, { "epoch": 0.32659807005384656, "grad_norm": 6.067870140075684, "learning_rate": 8.775e-06, "loss": 0.0783, "step": 6126 }, { "epoch": 0.3266513834834995, "grad_norm": 4.016909599304199, "learning_rate": 8.774800000000001e-06, "loss": 0.0468, "step": 6127 }, { "epoch": 0.3267046969131524, "grad_norm": 5.8555707931518555, "learning_rate": 8.7746e-06, "loss": 0.044, "step": 6128 }, { "epoch": 0.3267580103428053, "grad_norm": 2.432100772857666, "learning_rate": 8.7744e-06, "loss": 0.0468, "step": 6129 }, { "epoch": 0.3268113237724583, "grad_norm": 2.6861119270324707, "learning_rate": 8.7742e-06, "loss": 0.0633, "step": 6130 }, { "epoch": 0.32686463720211123, "grad_norm": 3.70672869682312, "learning_rate": 8.774000000000001e-06, "loss": 0.0499, "step": 6131 }, { "epoch": 0.32691795063176415, "grad_norm": 5.916614532470703, "learning_rate": 8.7738e-06, "loss": 0.0936, "step": 6132 }, { "epoch": 0.3269712640614171, "grad_norm": 3.010655164718628, "learning_rate": 8.7736e-06, "loss": 0.0282, "step": 6133 }, { "epoch": 0.32702457749107, "grad_norm": 4.176937580108643, "learning_rate": 8.773400000000001e-06, "loss": 0.0546, "step": 6134 }, { "epoch": 0.3270778909207229, "grad_norm": 6.762862682342529, "learning_rate": 8.773200000000001e-06, "loss": 0.0609, "step": 6135 }, { "epoch": 0.32713120435037585, "grad_norm": 5.007803440093994, "learning_rate": 8.773e-06, "loss": 0.0502, "step": 6136 }, { "epoch": 0.32718451778002877, "grad_norm": 4.829556941986084, "learning_rate": 8.7728e-06, "loss": 0.05, "step": 6137 }, { "epoch": 0.3272378312096817, "grad_norm": 2.136600971221924, "learning_rate": 8.772600000000001e-06, "loss": 0.0339, "step": 6138 }, { "epoch": 0.3272911446393347, "grad_norm": 7.158109188079834, "learning_rate": 8.772400000000001e-06, "loss": 0.054, "step": 6139 }, { "epoch": 0.3273444580689876, "grad_norm": 4.853822231292725, "learning_rate": 8.7722e-06, "loss": 0.0696, "step": 6140 }, { "epoch": 0.3273977714986405, "grad_norm": 7.63421106338501, "learning_rate": 8.772e-06, "loss": 0.0385, "step": 6141 }, { "epoch": 0.32745108492829345, "grad_norm": 4.037654876708984, "learning_rate": 8.7718e-06, "loss": 0.0351, "step": 6142 }, { "epoch": 0.32750439835794637, "grad_norm": 2.8536136150360107, "learning_rate": 8.771600000000001e-06, "loss": 0.045, "step": 6143 }, { "epoch": 0.3275577117875993, "grad_norm": 3.0882139205932617, "learning_rate": 8.7714e-06, "loss": 0.05, "step": 6144 }, { "epoch": 0.3276110252172522, "grad_norm": 3.07289981842041, "learning_rate": 8.7712e-06, "loss": 0.0536, "step": 6145 }, { "epoch": 0.32766433864690514, "grad_norm": 5.258673191070557, "learning_rate": 8.771000000000001e-06, "loss": 0.0791, "step": 6146 }, { "epoch": 0.32771765207655806, "grad_norm": 4.167630195617676, "learning_rate": 8.7708e-06, "loss": 0.0555, "step": 6147 }, { "epoch": 0.327770965506211, "grad_norm": 3.634843111038208, "learning_rate": 8.7706e-06, "loss": 0.0326, "step": 6148 }, { "epoch": 0.32782427893586397, "grad_norm": 2.4849045276641846, "learning_rate": 8.770400000000002e-06, "loss": 0.0276, "step": 6149 }, { "epoch": 0.3278775923655169, "grad_norm": 5.645739555358887, "learning_rate": 8.770200000000001e-06, "loss": 0.0951, "step": 6150 }, { "epoch": 0.3279309057951698, "grad_norm": 3.6659352779388428, "learning_rate": 8.77e-06, "loss": 0.0283, "step": 6151 }, { "epoch": 0.32798421922482274, "grad_norm": 2.4459779262542725, "learning_rate": 8.7698e-06, "loss": 0.0284, "step": 6152 }, { "epoch": 0.32803753265447566, "grad_norm": 8.02523422241211, "learning_rate": 8.7696e-06, "loss": 0.0927, "step": 6153 }, { "epoch": 0.3280908460841286, "grad_norm": 1.928033471107483, "learning_rate": 8.769400000000001e-06, "loss": 0.0301, "step": 6154 }, { "epoch": 0.3281441595137815, "grad_norm": 3.0011658668518066, "learning_rate": 8.7692e-06, "loss": 0.046, "step": 6155 }, { "epoch": 0.32819747294343443, "grad_norm": 3.5437662601470947, "learning_rate": 8.769e-06, "loss": 0.0661, "step": 6156 }, { "epoch": 0.32825078637308736, "grad_norm": 2.455793857574463, "learning_rate": 8.7688e-06, "loss": 0.0453, "step": 6157 }, { "epoch": 0.32830409980274033, "grad_norm": 3.989464282989502, "learning_rate": 8.768600000000001e-06, "loss": 0.0499, "step": 6158 }, { "epoch": 0.32835741323239326, "grad_norm": 2.2152113914489746, "learning_rate": 8.7684e-06, "loss": 0.0332, "step": 6159 }, { "epoch": 0.3284107266620462, "grad_norm": 9.726213455200195, "learning_rate": 8.768200000000002e-06, "loss": 0.0403, "step": 6160 }, { "epoch": 0.3284640400916991, "grad_norm": 3.3630688190460205, "learning_rate": 8.768000000000001e-06, "loss": 0.0479, "step": 6161 }, { "epoch": 0.32851735352135203, "grad_norm": 4.172132968902588, "learning_rate": 8.767800000000001e-06, "loss": 0.0652, "step": 6162 }, { "epoch": 0.32857066695100495, "grad_norm": 3.782984972000122, "learning_rate": 8.7676e-06, "loss": 0.0619, "step": 6163 }, { "epoch": 0.3286239803806579, "grad_norm": 3.0730857849121094, "learning_rate": 8.7674e-06, "loss": 0.0398, "step": 6164 }, { "epoch": 0.3286772938103108, "grad_norm": 3.3330001831054688, "learning_rate": 8.767200000000001e-06, "loss": 0.0542, "step": 6165 }, { "epoch": 0.3287306072399637, "grad_norm": 4.799734592437744, "learning_rate": 8.767000000000001e-06, "loss": 0.0353, "step": 6166 }, { "epoch": 0.3287839206696167, "grad_norm": 2.370346784591675, "learning_rate": 8.7668e-06, "loss": 0.0289, "step": 6167 }, { "epoch": 0.3288372340992696, "grad_norm": 3.4754345417022705, "learning_rate": 8.7666e-06, "loss": 0.0253, "step": 6168 }, { "epoch": 0.32889054752892255, "grad_norm": 2.7918038368225098, "learning_rate": 8.7664e-06, "loss": 0.0431, "step": 6169 }, { "epoch": 0.3289438609585755, "grad_norm": 1.834357738494873, "learning_rate": 8.766200000000001e-06, "loss": 0.0357, "step": 6170 }, { "epoch": 0.3289971743882284, "grad_norm": 5.216276168823242, "learning_rate": 8.766e-06, "loss": 0.0534, "step": 6171 }, { "epoch": 0.3290504878178813, "grad_norm": 7.771127700805664, "learning_rate": 8.765800000000002e-06, "loss": 0.0422, "step": 6172 }, { "epoch": 0.32910380124753424, "grad_norm": 4.4882378578186035, "learning_rate": 8.765600000000001e-06, "loss": 0.0381, "step": 6173 }, { "epoch": 0.32915711467718717, "grad_norm": 3.4891366958618164, "learning_rate": 8.7654e-06, "loss": 0.037, "step": 6174 }, { "epoch": 0.3292104281068401, "grad_norm": 4.399557113647461, "learning_rate": 8.7652e-06, "loss": 0.0699, "step": 6175 }, { "epoch": 0.329263741536493, "grad_norm": 3.748715877532959, "learning_rate": 8.765e-06, "loss": 0.0474, "step": 6176 }, { "epoch": 0.329317054966146, "grad_norm": 3.9601786136627197, "learning_rate": 8.764800000000001e-06, "loss": 0.0433, "step": 6177 }, { "epoch": 0.3293703683957989, "grad_norm": 7.181678771972656, "learning_rate": 8.7646e-06, "loss": 0.0911, "step": 6178 }, { "epoch": 0.32942368182545184, "grad_norm": 2.037696361541748, "learning_rate": 8.7644e-06, "loss": 0.0276, "step": 6179 }, { "epoch": 0.32947699525510477, "grad_norm": 2.94531512260437, "learning_rate": 8.7642e-06, "loss": 0.0396, "step": 6180 }, { "epoch": 0.3295303086847577, "grad_norm": 11.588677406311035, "learning_rate": 8.764e-06, "loss": 0.0768, "step": 6181 }, { "epoch": 0.3295836221144106, "grad_norm": 2.0525686740875244, "learning_rate": 8.7638e-06, "loss": 0.0324, "step": 6182 }, { "epoch": 0.32963693554406354, "grad_norm": 2.7055158615112305, "learning_rate": 8.7636e-06, "loss": 0.0662, "step": 6183 }, { "epoch": 0.32969024897371646, "grad_norm": 4.6818742752075195, "learning_rate": 8.763400000000002e-06, "loss": 0.059, "step": 6184 }, { "epoch": 0.3297435624033694, "grad_norm": 5.551112651824951, "learning_rate": 8.763200000000001e-06, "loss": 0.0781, "step": 6185 }, { "epoch": 0.32979687583302236, "grad_norm": 1.5155601501464844, "learning_rate": 8.763e-06, "loss": 0.0534, "step": 6186 }, { "epoch": 0.3298501892626753, "grad_norm": 2.159806728363037, "learning_rate": 8.7628e-06, "loss": 0.0323, "step": 6187 }, { "epoch": 0.3299035026923282, "grad_norm": 2.680354118347168, "learning_rate": 8.762600000000001e-06, "loss": 0.0377, "step": 6188 }, { "epoch": 0.32995681612198113, "grad_norm": 2.250183343887329, "learning_rate": 8.762400000000001e-06, "loss": 0.0395, "step": 6189 }, { "epoch": 0.33001012955163406, "grad_norm": 5.510083198547363, "learning_rate": 8.7622e-06, "loss": 0.0783, "step": 6190 }, { "epoch": 0.330063442981287, "grad_norm": 2.685537576675415, "learning_rate": 8.762e-06, "loss": 0.0198, "step": 6191 }, { "epoch": 0.3301167564109399, "grad_norm": 7.424177646636963, "learning_rate": 8.7618e-06, "loss": 0.0718, "step": 6192 }, { "epoch": 0.33017006984059283, "grad_norm": 8.415295600891113, "learning_rate": 8.761600000000001e-06, "loss": 0.1372, "step": 6193 }, { "epoch": 0.33022338327024575, "grad_norm": 2.64292573928833, "learning_rate": 8.7614e-06, "loss": 0.0324, "step": 6194 }, { "epoch": 0.33027669669989873, "grad_norm": 2.867600679397583, "learning_rate": 8.7612e-06, "loss": 0.0401, "step": 6195 }, { "epoch": 0.33033001012955165, "grad_norm": 4.213593006134033, "learning_rate": 8.761000000000001e-06, "loss": 0.0471, "step": 6196 }, { "epoch": 0.3303833235592046, "grad_norm": 3.40952730178833, "learning_rate": 8.760800000000001e-06, "loss": 0.041, "step": 6197 }, { "epoch": 0.3304366369888575, "grad_norm": 4.686985015869141, "learning_rate": 8.7606e-06, "loss": 0.0484, "step": 6198 }, { "epoch": 0.3304899504185104, "grad_norm": 2.707258701324463, "learning_rate": 8.760400000000002e-06, "loss": 0.0288, "step": 6199 }, { "epoch": 0.33054326384816335, "grad_norm": 2.974494218826294, "learning_rate": 8.760200000000001e-06, "loss": 0.0437, "step": 6200 }, { "epoch": 0.3305965772778163, "grad_norm": 3.18827223777771, "learning_rate": 8.76e-06, "loss": 0.0643, "step": 6201 }, { "epoch": 0.3306498907074692, "grad_norm": 7.529760360717773, "learning_rate": 8.7598e-06, "loss": 0.0505, "step": 6202 }, { "epoch": 0.3307032041371221, "grad_norm": 5.460042476654053, "learning_rate": 8.7596e-06, "loss": 0.0556, "step": 6203 }, { "epoch": 0.33075651756677504, "grad_norm": 3.462355852127075, "learning_rate": 8.759400000000001e-06, "loss": 0.048, "step": 6204 }, { "epoch": 0.330809830996428, "grad_norm": 2.042534112930298, "learning_rate": 8.7592e-06, "loss": 0.0144, "step": 6205 }, { "epoch": 0.33086314442608095, "grad_norm": 1.7277450561523438, "learning_rate": 8.759e-06, "loss": 0.0145, "step": 6206 }, { "epoch": 0.33091645785573387, "grad_norm": 3.2722384929656982, "learning_rate": 8.7588e-06, "loss": 0.0253, "step": 6207 }, { "epoch": 0.3309697712853868, "grad_norm": 2.738163948059082, "learning_rate": 8.758600000000001e-06, "loss": 0.0267, "step": 6208 }, { "epoch": 0.3310230847150397, "grad_norm": 2.485991954803467, "learning_rate": 8.7584e-06, "loss": 0.0439, "step": 6209 }, { "epoch": 0.33107639814469264, "grad_norm": 2.7096312046051025, "learning_rate": 8.758200000000002e-06, "loss": 0.0353, "step": 6210 }, { "epoch": 0.33112971157434556, "grad_norm": 2.8029181957244873, "learning_rate": 8.758000000000002e-06, "loss": 0.0376, "step": 6211 }, { "epoch": 0.3311830250039985, "grad_norm": 3.724618434906006, "learning_rate": 8.757800000000001e-06, "loss": 0.0442, "step": 6212 }, { "epoch": 0.3312363384336514, "grad_norm": 2.0493316650390625, "learning_rate": 8.7576e-06, "loss": 0.0254, "step": 6213 }, { "epoch": 0.3312896518633044, "grad_norm": 3.6103339195251465, "learning_rate": 8.7574e-06, "loss": 0.0389, "step": 6214 }, { "epoch": 0.3313429652929573, "grad_norm": 1.9011495113372803, "learning_rate": 8.757200000000001e-06, "loss": 0.0256, "step": 6215 }, { "epoch": 0.33139627872261024, "grad_norm": 2.5091140270233154, "learning_rate": 8.757000000000001e-06, "loss": 0.0291, "step": 6216 }, { "epoch": 0.33144959215226316, "grad_norm": 3.51649808883667, "learning_rate": 8.7568e-06, "loss": 0.041, "step": 6217 }, { "epoch": 0.3315029055819161, "grad_norm": 6.38361930847168, "learning_rate": 8.7566e-06, "loss": 0.04, "step": 6218 }, { "epoch": 0.331556219011569, "grad_norm": 3.3827779293060303, "learning_rate": 8.7564e-06, "loss": 0.0548, "step": 6219 }, { "epoch": 0.33160953244122193, "grad_norm": 2.6900970935821533, "learning_rate": 8.756200000000001e-06, "loss": 0.0368, "step": 6220 }, { "epoch": 0.33166284587087486, "grad_norm": 2.925569772720337, "learning_rate": 8.756e-06, "loss": 0.0312, "step": 6221 }, { "epoch": 0.3317161593005278, "grad_norm": 4.794036388397217, "learning_rate": 8.755800000000002e-06, "loss": 0.0795, "step": 6222 }, { "epoch": 0.33176947273018076, "grad_norm": 2.690269708633423, "learning_rate": 8.755600000000001e-06, "loss": 0.0358, "step": 6223 }, { "epoch": 0.3318227861598337, "grad_norm": 3.338703155517578, "learning_rate": 8.755400000000001e-06, "loss": 0.034, "step": 6224 }, { "epoch": 0.3318760995894866, "grad_norm": 2.9201152324676514, "learning_rate": 8.7552e-06, "loss": 0.0618, "step": 6225 }, { "epoch": 0.33192941301913953, "grad_norm": 1.5288455486297607, "learning_rate": 8.755e-06, "loss": 0.0159, "step": 6226 }, { "epoch": 0.33198272644879245, "grad_norm": 2.573218822479248, "learning_rate": 8.754800000000001e-06, "loss": 0.0295, "step": 6227 }, { "epoch": 0.3320360398784454, "grad_norm": 2.7202670574188232, "learning_rate": 8.754600000000001e-06, "loss": 0.051, "step": 6228 }, { "epoch": 0.3320893533080983, "grad_norm": 4.039168357849121, "learning_rate": 8.7544e-06, "loss": 0.0748, "step": 6229 }, { "epoch": 0.3321426667377512, "grad_norm": 4.517536163330078, "learning_rate": 8.7542e-06, "loss": 0.0323, "step": 6230 }, { "epoch": 0.33219598016740415, "grad_norm": 4.441738605499268, "learning_rate": 8.754e-06, "loss": 0.0391, "step": 6231 }, { "epoch": 0.33224929359705707, "grad_norm": 6.519414901733398, "learning_rate": 8.7538e-06, "loss": 0.0899, "step": 6232 }, { "epoch": 0.33230260702671005, "grad_norm": 3.23201060295105, "learning_rate": 8.7536e-06, "loss": 0.046, "step": 6233 }, { "epoch": 0.332355920456363, "grad_norm": 5.187926769256592, "learning_rate": 8.753400000000002e-06, "loss": 0.0552, "step": 6234 }, { "epoch": 0.3324092338860159, "grad_norm": 2.452059268951416, "learning_rate": 8.753200000000001e-06, "loss": 0.0251, "step": 6235 }, { "epoch": 0.3324625473156688, "grad_norm": 4.582291126251221, "learning_rate": 8.753e-06, "loss": 0.0485, "step": 6236 }, { "epoch": 0.33251586074532175, "grad_norm": 5.131900787353516, "learning_rate": 8.7528e-06, "loss": 0.057, "step": 6237 }, { "epoch": 0.33256917417497467, "grad_norm": 2.628570318222046, "learning_rate": 8.752600000000002e-06, "loss": 0.0292, "step": 6238 }, { "epoch": 0.3326224876046276, "grad_norm": 3.9098665714263916, "learning_rate": 8.752400000000001e-06, "loss": 0.0501, "step": 6239 }, { "epoch": 0.3326758010342805, "grad_norm": 3.4055674076080322, "learning_rate": 8.7522e-06, "loss": 0.0425, "step": 6240 }, { "epoch": 0.33272911446393344, "grad_norm": 3.9439496994018555, "learning_rate": 8.752e-06, "loss": 0.0336, "step": 6241 }, { "epoch": 0.3327824278935864, "grad_norm": 2.7088491916656494, "learning_rate": 8.7518e-06, "loss": 0.027, "step": 6242 }, { "epoch": 0.33283574132323934, "grad_norm": 1.9675343036651611, "learning_rate": 8.751600000000001e-06, "loss": 0.0249, "step": 6243 }, { "epoch": 0.33288905475289227, "grad_norm": 2.9598565101623535, "learning_rate": 8.7514e-06, "loss": 0.0386, "step": 6244 }, { "epoch": 0.3329423681825452, "grad_norm": 2.601088047027588, "learning_rate": 8.7512e-06, "loss": 0.0406, "step": 6245 }, { "epoch": 0.3329956816121981, "grad_norm": 2.3051559925079346, "learning_rate": 8.751000000000001e-06, "loss": 0.0313, "step": 6246 }, { "epoch": 0.33304899504185104, "grad_norm": 3.4826390743255615, "learning_rate": 8.750800000000001e-06, "loss": 0.0566, "step": 6247 }, { "epoch": 0.33310230847150396, "grad_norm": 2.6151790618896484, "learning_rate": 8.7506e-06, "loss": 0.0416, "step": 6248 }, { "epoch": 0.3331556219011569, "grad_norm": 2.672581434249878, "learning_rate": 8.750400000000002e-06, "loss": 0.0375, "step": 6249 }, { "epoch": 0.3332089353308098, "grad_norm": 2.5567498207092285, "learning_rate": 8.750200000000001e-06, "loss": 0.0387, "step": 6250 }, { "epoch": 0.3332622487604628, "grad_norm": 2.187793254852295, "learning_rate": 8.750000000000001e-06, "loss": 0.0273, "step": 6251 }, { "epoch": 0.3333155621901157, "grad_norm": 2.9002280235290527, "learning_rate": 8.7498e-06, "loss": 0.0346, "step": 6252 }, { "epoch": 0.33336887561976863, "grad_norm": 3.454084634780884, "learning_rate": 8.7496e-06, "loss": 0.0382, "step": 6253 }, { "epoch": 0.33342218904942156, "grad_norm": 3.62746524810791, "learning_rate": 8.749400000000001e-06, "loss": 0.0474, "step": 6254 }, { "epoch": 0.3334755024790745, "grad_norm": 11.327073097229004, "learning_rate": 8.749200000000001e-06, "loss": 0.0788, "step": 6255 }, { "epoch": 0.3335288159087274, "grad_norm": 4.394981861114502, "learning_rate": 8.749e-06, "loss": 0.0267, "step": 6256 }, { "epoch": 0.33358212933838033, "grad_norm": 3.4925525188446045, "learning_rate": 8.7488e-06, "loss": 0.0448, "step": 6257 }, { "epoch": 0.33363544276803325, "grad_norm": 2.854186534881592, "learning_rate": 8.748600000000001e-06, "loss": 0.0436, "step": 6258 }, { "epoch": 0.3336887561976862, "grad_norm": 2.494385004043579, "learning_rate": 8.7484e-06, "loss": 0.0311, "step": 6259 }, { "epoch": 0.3337420696273391, "grad_norm": 3.5653042793273926, "learning_rate": 8.7482e-06, "loss": 0.0378, "step": 6260 }, { "epoch": 0.3337953830569921, "grad_norm": 2.6425397396087646, "learning_rate": 8.748000000000002e-06, "loss": 0.0488, "step": 6261 }, { "epoch": 0.333848696486645, "grad_norm": 5.2874555587768555, "learning_rate": 8.747800000000001e-06, "loss": 0.0533, "step": 6262 }, { "epoch": 0.3339020099162979, "grad_norm": 3.3814687728881836, "learning_rate": 8.7476e-06, "loss": 0.067, "step": 6263 }, { "epoch": 0.33395532334595085, "grad_norm": 4.4140496253967285, "learning_rate": 8.7474e-06, "loss": 0.0582, "step": 6264 }, { "epoch": 0.3340086367756038, "grad_norm": 3.796487808227539, "learning_rate": 8.747200000000002e-06, "loss": 0.0663, "step": 6265 }, { "epoch": 0.3340619502052567, "grad_norm": 3.0350005626678467, "learning_rate": 8.747000000000001e-06, "loss": 0.031, "step": 6266 }, { "epoch": 0.3341152636349096, "grad_norm": 6.693719863891602, "learning_rate": 8.7468e-06, "loss": 0.0655, "step": 6267 }, { "epoch": 0.33416857706456254, "grad_norm": 6.814669132232666, "learning_rate": 8.7466e-06, "loss": 0.0934, "step": 6268 }, { "epoch": 0.33422189049421547, "grad_norm": 5.184483528137207, "learning_rate": 8.7464e-06, "loss": 0.0695, "step": 6269 }, { "epoch": 0.33427520392386845, "grad_norm": 2.5600500106811523, "learning_rate": 8.746200000000001e-06, "loss": 0.0385, "step": 6270 }, { "epoch": 0.33432851735352137, "grad_norm": 2.0163819789886475, "learning_rate": 8.746e-06, "loss": 0.0382, "step": 6271 }, { "epoch": 0.3343818307831743, "grad_norm": 3.0099234580993652, "learning_rate": 8.7458e-06, "loss": 0.056, "step": 6272 }, { "epoch": 0.3344351442128272, "grad_norm": 4.852395534515381, "learning_rate": 8.745600000000001e-06, "loss": 0.0518, "step": 6273 }, { "epoch": 0.33448845764248014, "grad_norm": 2.3491084575653076, "learning_rate": 8.745400000000001e-06, "loss": 0.0283, "step": 6274 }, { "epoch": 0.33454177107213307, "grad_norm": 1.7331247329711914, "learning_rate": 8.7452e-06, "loss": 0.0178, "step": 6275 }, { "epoch": 0.334595084501786, "grad_norm": 2.516118288040161, "learning_rate": 8.745000000000002e-06, "loss": 0.039, "step": 6276 }, { "epoch": 0.3346483979314389, "grad_norm": 3.7880027294158936, "learning_rate": 8.744800000000001e-06, "loss": 0.0665, "step": 6277 }, { "epoch": 0.33470171136109184, "grad_norm": 3.286668062210083, "learning_rate": 8.744600000000001e-06, "loss": 0.0284, "step": 6278 }, { "epoch": 0.3347550247907448, "grad_norm": 2.393474578857422, "learning_rate": 8.7444e-06, "loss": 0.0328, "step": 6279 }, { "epoch": 0.33480833822039774, "grad_norm": 5.462867736816406, "learning_rate": 8.7442e-06, "loss": 0.0672, "step": 6280 }, { "epoch": 0.33486165165005066, "grad_norm": 3.3182854652404785, "learning_rate": 8.744e-06, "loss": 0.0447, "step": 6281 }, { "epoch": 0.3349149650797036, "grad_norm": 2.225569725036621, "learning_rate": 8.743800000000001e-06, "loss": 0.0408, "step": 6282 }, { "epoch": 0.3349682785093565, "grad_norm": 3.3327887058258057, "learning_rate": 8.7436e-06, "loss": 0.0647, "step": 6283 }, { "epoch": 0.33502159193900943, "grad_norm": 3.413933277130127, "learning_rate": 8.7434e-06, "loss": 0.0393, "step": 6284 }, { "epoch": 0.33507490536866236, "grad_norm": 2.875887632369995, "learning_rate": 8.743200000000001e-06, "loss": 0.0454, "step": 6285 }, { "epoch": 0.3351282187983153, "grad_norm": 2.682851552963257, "learning_rate": 8.743e-06, "loss": 0.0578, "step": 6286 }, { "epoch": 0.3351815322279682, "grad_norm": 1.7448320388793945, "learning_rate": 8.7428e-06, "loss": 0.0267, "step": 6287 }, { "epoch": 0.3352348456576211, "grad_norm": 1.7648361921310425, "learning_rate": 8.742600000000002e-06, "loss": 0.0224, "step": 6288 }, { "epoch": 0.3352881590872741, "grad_norm": 3.3816449642181396, "learning_rate": 8.742400000000001e-06, "loss": 0.0251, "step": 6289 }, { "epoch": 0.33534147251692703, "grad_norm": 5.730880260467529, "learning_rate": 8.7422e-06, "loss": 0.0584, "step": 6290 }, { "epoch": 0.33539478594657995, "grad_norm": 2.714617967605591, "learning_rate": 8.742e-06, "loss": 0.0238, "step": 6291 }, { "epoch": 0.3354480993762329, "grad_norm": 3.426652669906616, "learning_rate": 8.7418e-06, "loss": 0.0249, "step": 6292 }, { "epoch": 0.3355014128058858, "grad_norm": 1.5383110046386719, "learning_rate": 8.741600000000001e-06, "loss": 0.0158, "step": 6293 }, { "epoch": 0.3355547262355387, "grad_norm": 3.831590414047241, "learning_rate": 8.7414e-06, "loss": 0.0566, "step": 6294 }, { "epoch": 0.33560803966519165, "grad_norm": 2.4129669666290283, "learning_rate": 8.7412e-06, "loss": 0.0235, "step": 6295 }, { "epoch": 0.3356613530948446, "grad_norm": 3.012915849685669, "learning_rate": 8.741e-06, "loss": 0.0495, "step": 6296 }, { "epoch": 0.3357146665244975, "grad_norm": 2.8908543586730957, "learning_rate": 8.740800000000001e-06, "loss": 0.0414, "step": 6297 }, { "epoch": 0.3357679799541505, "grad_norm": 4.218776702880859, "learning_rate": 8.7406e-06, "loss": 0.055, "step": 6298 }, { "epoch": 0.3358212933838034, "grad_norm": 4.953854084014893, "learning_rate": 8.740400000000002e-06, "loss": 0.0456, "step": 6299 }, { "epoch": 0.3358746068134563, "grad_norm": 2.6016759872436523, "learning_rate": 8.740200000000001e-06, "loss": 0.0275, "step": 6300 }, { "epoch": 0.33592792024310925, "grad_norm": 2.8569159507751465, "learning_rate": 8.740000000000001e-06, "loss": 0.0185, "step": 6301 }, { "epoch": 0.33598123367276217, "grad_norm": 5.225261211395264, "learning_rate": 8.7398e-06, "loss": 0.0604, "step": 6302 }, { "epoch": 0.3360345471024151, "grad_norm": 2.069056272506714, "learning_rate": 8.7396e-06, "loss": 0.0217, "step": 6303 }, { "epoch": 0.336087860532068, "grad_norm": 3.9169774055480957, "learning_rate": 8.739400000000001e-06, "loss": 0.0606, "step": 6304 }, { "epoch": 0.33614117396172094, "grad_norm": 2.8180019855499268, "learning_rate": 8.739200000000001e-06, "loss": 0.0547, "step": 6305 }, { "epoch": 0.33619448739137386, "grad_norm": 3.6599886417388916, "learning_rate": 8.739e-06, "loss": 0.0712, "step": 6306 }, { "epoch": 0.33624780082102684, "grad_norm": 3.2844033241271973, "learning_rate": 8.7388e-06, "loss": 0.0228, "step": 6307 }, { "epoch": 0.33630111425067977, "grad_norm": 5.0949506759643555, "learning_rate": 8.7386e-06, "loss": 0.0579, "step": 6308 }, { "epoch": 0.3363544276803327, "grad_norm": 1.7015272378921509, "learning_rate": 8.738400000000001e-06, "loss": 0.0115, "step": 6309 }, { "epoch": 0.3364077411099856, "grad_norm": 3.361262321472168, "learning_rate": 8.7382e-06, "loss": 0.0453, "step": 6310 }, { "epoch": 0.33646105453963854, "grad_norm": 4.605474472045898, "learning_rate": 8.738000000000002e-06, "loss": 0.0553, "step": 6311 }, { "epoch": 0.33651436796929146, "grad_norm": 3.0979652404785156, "learning_rate": 8.737800000000001e-06, "loss": 0.0343, "step": 6312 }, { "epoch": 0.3365676813989444, "grad_norm": 2.532360792160034, "learning_rate": 8.7376e-06, "loss": 0.0467, "step": 6313 }, { "epoch": 0.3366209948285973, "grad_norm": 2.235241174697876, "learning_rate": 8.7374e-06, "loss": 0.0364, "step": 6314 }, { "epoch": 0.33667430825825023, "grad_norm": 3.5756287574768066, "learning_rate": 8.737200000000002e-06, "loss": 0.0507, "step": 6315 }, { "epoch": 0.33672762168790316, "grad_norm": 2.1876766681671143, "learning_rate": 8.737000000000001e-06, "loss": 0.0348, "step": 6316 }, { "epoch": 0.33678093511755613, "grad_norm": 3.4748117923736572, "learning_rate": 8.7368e-06, "loss": 0.0425, "step": 6317 }, { "epoch": 0.33683424854720906, "grad_norm": 2.687349319458008, "learning_rate": 8.7366e-06, "loss": 0.025, "step": 6318 }, { "epoch": 0.336887561976862, "grad_norm": 2.0729727745056152, "learning_rate": 8.7364e-06, "loss": 0.0169, "step": 6319 }, { "epoch": 0.3369408754065149, "grad_norm": 2.636000871658325, "learning_rate": 8.7362e-06, "loss": 0.038, "step": 6320 }, { "epoch": 0.33699418883616783, "grad_norm": 3.4471683502197266, "learning_rate": 8.736e-06, "loss": 0.0342, "step": 6321 }, { "epoch": 0.33704750226582075, "grad_norm": 4.1305646896362305, "learning_rate": 8.7358e-06, "loss": 0.0459, "step": 6322 }, { "epoch": 0.3371008156954737, "grad_norm": 3.88226318359375, "learning_rate": 8.735600000000002e-06, "loss": 0.0631, "step": 6323 }, { "epoch": 0.3371541291251266, "grad_norm": 3.3311920166015625, "learning_rate": 8.735400000000001e-06, "loss": 0.0624, "step": 6324 }, { "epoch": 0.3372074425547795, "grad_norm": 3.6202821731567383, "learning_rate": 8.7352e-06, "loss": 0.0347, "step": 6325 }, { "epoch": 0.3372607559844325, "grad_norm": 1.9918485879898071, "learning_rate": 8.735000000000002e-06, "loss": 0.0173, "step": 6326 }, { "epoch": 0.3373140694140854, "grad_norm": 4.07207727432251, "learning_rate": 8.734800000000001e-06, "loss": 0.0554, "step": 6327 }, { "epoch": 0.33736738284373835, "grad_norm": 2.8900249004364014, "learning_rate": 8.734600000000001e-06, "loss": 0.0286, "step": 6328 }, { "epoch": 0.3374206962733913, "grad_norm": 2.3909220695495605, "learning_rate": 8.7344e-06, "loss": 0.0251, "step": 6329 }, { "epoch": 0.3374740097030442, "grad_norm": 3.4307236671447754, "learning_rate": 8.7342e-06, "loss": 0.0287, "step": 6330 }, { "epoch": 0.3375273231326971, "grad_norm": 2.9134654998779297, "learning_rate": 8.734e-06, "loss": 0.0315, "step": 6331 }, { "epoch": 0.33758063656235004, "grad_norm": 4.264942646026611, "learning_rate": 8.733800000000001e-06, "loss": 0.0571, "step": 6332 }, { "epoch": 0.33763394999200297, "grad_norm": 6.68358850479126, "learning_rate": 8.7336e-06, "loss": 0.129, "step": 6333 }, { "epoch": 0.3376872634216559, "grad_norm": 2.334207773208618, "learning_rate": 8.7334e-06, "loss": 0.016, "step": 6334 }, { "epoch": 0.33774057685130887, "grad_norm": 2.3210039138793945, "learning_rate": 8.733200000000001e-06, "loss": 0.0396, "step": 6335 }, { "epoch": 0.3377938902809618, "grad_norm": 3.4975931644439697, "learning_rate": 8.733000000000001e-06, "loss": 0.0496, "step": 6336 }, { "epoch": 0.3378472037106147, "grad_norm": 5.362027645111084, "learning_rate": 8.7328e-06, "loss": 0.0596, "step": 6337 }, { "epoch": 0.33790051714026764, "grad_norm": 5.9414167404174805, "learning_rate": 8.732600000000002e-06, "loss": 0.0541, "step": 6338 }, { "epoch": 0.33795383056992057, "grad_norm": 3.5606982707977295, "learning_rate": 8.732400000000001e-06, "loss": 0.0762, "step": 6339 }, { "epoch": 0.3380071439995735, "grad_norm": 3.599684238433838, "learning_rate": 8.732200000000001e-06, "loss": 0.0321, "step": 6340 }, { "epoch": 0.3380604574292264, "grad_norm": 2.489656448364258, "learning_rate": 8.732e-06, "loss": 0.0235, "step": 6341 }, { "epoch": 0.33811377085887934, "grad_norm": 3.486818313598633, "learning_rate": 8.7318e-06, "loss": 0.0156, "step": 6342 }, { "epoch": 0.33816708428853226, "grad_norm": 2.125314950942993, "learning_rate": 8.731600000000001e-06, "loss": 0.0204, "step": 6343 }, { "epoch": 0.3382203977181852, "grad_norm": 3.3914027214050293, "learning_rate": 8.7314e-06, "loss": 0.0352, "step": 6344 }, { "epoch": 0.33827371114783816, "grad_norm": 3.741488218307495, "learning_rate": 8.7312e-06, "loss": 0.0426, "step": 6345 }, { "epoch": 0.3383270245774911, "grad_norm": 2.2209644317626953, "learning_rate": 8.731e-06, "loss": 0.021, "step": 6346 }, { "epoch": 0.338380338007144, "grad_norm": 2.552483320236206, "learning_rate": 8.730800000000001e-06, "loss": 0.0236, "step": 6347 }, { "epoch": 0.33843365143679693, "grad_norm": 4.1388115882873535, "learning_rate": 8.7306e-06, "loss": 0.0569, "step": 6348 }, { "epoch": 0.33848696486644986, "grad_norm": 2.383631467819214, "learning_rate": 8.7304e-06, "loss": 0.0223, "step": 6349 }, { "epoch": 0.3385402782961028, "grad_norm": 3.274789571762085, "learning_rate": 8.730200000000002e-06, "loss": 0.0554, "step": 6350 }, { "epoch": 0.3385935917257557, "grad_norm": 2.4700229167938232, "learning_rate": 8.730000000000001e-06, "loss": 0.0446, "step": 6351 }, { "epoch": 0.33864690515540863, "grad_norm": 3.6256556510925293, "learning_rate": 8.7298e-06, "loss": 0.023, "step": 6352 }, { "epoch": 0.33870021858506155, "grad_norm": 3.13289475440979, "learning_rate": 8.7296e-06, "loss": 0.0304, "step": 6353 }, { "epoch": 0.33875353201471453, "grad_norm": 3.8335764408111572, "learning_rate": 8.729400000000001e-06, "loss": 0.0216, "step": 6354 }, { "epoch": 0.33880684544436745, "grad_norm": 3.589168071746826, "learning_rate": 8.729200000000001e-06, "loss": 0.0423, "step": 6355 }, { "epoch": 0.3388601588740204, "grad_norm": 2.8984107971191406, "learning_rate": 8.729e-06, "loss": 0.0393, "step": 6356 }, { "epoch": 0.3389134723036733, "grad_norm": 4.404190540313721, "learning_rate": 8.7288e-06, "loss": 0.0547, "step": 6357 }, { "epoch": 0.3389667857333262, "grad_norm": 2.9625141620635986, "learning_rate": 8.7286e-06, "loss": 0.0474, "step": 6358 }, { "epoch": 0.33902009916297915, "grad_norm": 3.8604588508605957, "learning_rate": 8.728400000000001e-06, "loss": 0.0527, "step": 6359 }, { "epoch": 0.3390734125926321, "grad_norm": 3.259941339492798, "learning_rate": 8.7282e-06, "loss": 0.0378, "step": 6360 }, { "epoch": 0.339126726022285, "grad_norm": 2.6304593086242676, "learning_rate": 8.728e-06, "loss": 0.0308, "step": 6361 }, { "epoch": 0.3391800394519379, "grad_norm": 3.022347927093506, "learning_rate": 8.727800000000001e-06, "loss": 0.0424, "step": 6362 }, { "epoch": 0.3392333528815909, "grad_norm": 3.764104127883911, "learning_rate": 8.727600000000001e-06, "loss": 0.0315, "step": 6363 }, { "epoch": 0.3392866663112438, "grad_norm": 5.197044849395752, "learning_rate": 8.7274e-06, "loss": 0.0482, "step": 6364 }, { "epoch": 0.33933997974089675, "grad_norm": 3.5410830974578857, "learning_rate": 8.727200000000002e-06, "loss": 0.0368, "step": 6365 }, { "epoch": 0.33939329317054967, "grad_norm": 5.0965399742126465, "learning_rate": 8.727000000000001e-06, "loss": 0.0896, "step": 6366 }, { "epoch": 0.3394466066002026, "grad_norm": 2.7028088569641113, "learning_rate": 8.726800000000001e-06, "loss": 0.0373, "step": 6367 }, { "epoch": 0.3394999200298555, "grad_norm": 3.6277036666870117, "learning_rate": 8.7266e-06, "loss": 0.03, "step": 6368 }, { "epoch": 0.33955323345950844, "grad_norm": 3.654622793197632, "learning_rate": 8.7264e-06, "loss": 0.0457, "step": 6369 }, { "epoch": 0.33960654688916136, "grad_norm": 3.2566609382629395, "learning_rate": 8.7262e-06, "loss": 0.0148, "step": 6370 }, { "epoch": 0.3396598603188143, "grad_norm": 3.6740453243255615, "learning_rate": 8.726e-06, "loss": 0.056, "step": 6371 }, { "epoch": 0.3397131737484672, "grad_norm": 4.810910224914551, "learning_rate": 8.7258e-06, "loss": 0.08, "step": 6372 }, { "epoch": 0.3397664871781202, "grad_norm": 2.931997537612915, "learning_rate": 8.7256e-06, "loss": 0.0652, "step": 6373 }, { "epoch": 0.3398198006077731, "grad_norm": 3.9921560287475586, "learning_rate": 8.725400000000001e-06, "loss": 0.0468, "step": 6374 }, { "epoch": 0.33987311403742604, "grad_norm": 2.7623636722564697, "learning_rate": 8.7252e-06, "loss": 0.0414, "step": 6375 }, { "epoch": 0.33992642746707896, "grad_norm": 2.763920307159424, "learning_rate": 8.725000000000002e-06, "loss": 0.0317, "step": 6376 }, { "epoch": 0.3399797408967319, "grad_norm": 3.608020067214966, "learning_rate": 8.724800000000002e-06, "loss": 0.0458, "step": 6377 }, { "epoch": 0.3400330543263848, "grad_norm": 3.469421625137329, "learning_rate": 8.724600000000001e-06, "loss": 0.0353, "step": 6378 }, { "epoch": 0.34008636775603773, "grad_norm": 5.992349624633789, "learning_rate": 8.7244e-06, "loss": 0.0687, "step": 6379 }, { "epoch": 0.34013968118569066, "grad_norm": 6.60371732711792, "learning_rate": 8.7242e-06, "loss": 0.0459, "step": 6380 }, { "epoch": 0.3401929946153436, "grad_norm": 5.221349716186523, "learning_rate": 8.724e-06, "loss": 0.0642, "step": 6381 }, { "epoch": 0.34024630804499656, "grad_norm": 4.113179683685303, "learning_rate": 8.723800000000001e-06, "loss": 0.0418, "step": 6382 }, { "epoch": 0.3402996214746495, "grad_norm": 2.903637409210205, "learning_rate": 8.7236e-06, "loss": 0.0527, "step": 6383 }, { "epoch": 0.3403529349043024, "grad_norm": 2.9162724018096924, "learning_rate": 8.7234e-06, "loss": 0.0456, "step": 6384 }, { "epoch": 0.34040624833395533, "grad_norm": 4.463282108306885, "learning_rate": 8.7232e-06, "loss": 0.048, "step": 6385 }, { "epoch": 0.34045956176360825, "grad_norm": 4.369755744934082, "learning_rate": 8.723000000000001e-06, "loss": 0.0328, "step": 6386 }, { "epoch": 0.3405128751932612, "grad_norm": 4.240911960601807, "learning_rate": 8.7228e-06, "loss": 0.0422, "step": 6387 }, { "epoch": 0.3405661886229141, "grad_norm": 2.474452257156372, "learning_rate": 8.722600000000002e-06, "loss": 0.0392, "step": 6388 }, { "epoch": 0.340619502052567, "grad_norm": 1.8168888092041016, "learning_rate": 8.722400000000001e-06, "loss": 0.0365, "step": 6389 }, { "epoch": 0.34067281548221995, "grad_norm": 3.2654871940612793, "learning_rate": 8.722200000000001e-06, "loss": 0.045, "step": 6390 }, { "epoch": 0.3407261289118729, "grad_norm": 4.23785400390625, "learning_rate": 8.722e-06, "loss": 0.0546, "step": 6391 }, { "epoch": 0.34077944234152585, "grad_norm": 4.899878025054932, "learning_rate": 8.7218e-06, "loss": 0.0707, "step": 6392 }, { "epoch": 0.3408327557711788, "grad_norm": 3.8448972702026367, "learning_rate": 8.721600000000001e-06, "loss": 0.0528, "step": 6393 }, { "epoch": 0.3408860692008317, "grad_norm": 2.8490471839904785, "learning_rate": 8.721400000000001e-06, "loss": 0.0212, "step": 6394 }, { "epoch": 0.3409393826304846, "grad_norm": 6.858052730560303, "learning_rate": 8.7212e-06, "loss": 0.0474, "step": 6395 }, { "epoch": 0.34099269606013755, "grad_norm": 3.8179666996002197, "learning_rate": 8.721e-06, "loss": 0.0551, "step": 6396 }, { "epoch": 0.34104600948979047, "grad_norm": 1.6849263906478882, "learning_rate": 8.7208e-06, "loss": 0.0199, "step": 6397 }, { "epoch": 0.3410993229194434, "grad_norm": 2.6041252613067627, "learning_rate": 8.7206e-06, "loss": 0.0482, "step": 6398 }, { "epoch": 0.3411526363490963, "grad_norm": 6.486670017242432, "learning_rate": 8.7204e-06, "loss": 0.0808, "step": 6399 }, { "epoch": 0.34120594977874924, "grad_norm": 5.854592800140381, "learning_rate": 8.720200000000002e-06, "loss": 0.0615, "step": 6400 }, { "epoch": 0.3412592632084022, "grad_norm": 6.8809895515441895, "learning_rate": 8.720000000000001e-06, "loss": 0.0626, "step": 6401 }, { "epoch": 0.34131257663805514, "grad_norm": 2.087369203567505, "learning_rate": 8.7198e-06, "loss": 0.0316, "step": 6402 }, { "epoch": 0.34136589006770807, "grad_norm": 2.9829025268554688, "learning_rate": 8.7196e-06, "loss": 0.0374, "step": 6403 }, { "epoch": 0.341419203497361, "grad_norm": 7.299148082733154, "learning_rate": 8.719400000000002e-06, "loss": 0.059, "step": 6404 }, { "epoch": 0.3414725169270139, "grad_norm": 4.905500411987305, "learning_rate": 8.719200000000001e-06, "loss": 0.0373, "step": 6405 }, { "epoch": 0.34152583035666684, "grad_norm": 4.0263752937316895, "learning_rate": 8.719e-06, "loss": 0.0371, "step": 6406 }, { "epoch": 0.34157914378631976, "grad_norm": 4.196747779846191, "learning_rate": 8.7188e-06, "loss": 0.0377, "step": 6407 }, { "epoch": 0.3416324572159727, "grad_norm": 5.648545265197754, "learning_rate": 8.7186e-06, "loss": 0.0509, "step": 6408 }, { "epoch": 0.3416857706456256, "grad_norm": 2.418743133544922, "learning_rate": 8.7184e-06, "loss": 0.0282, "step": 6409 }, { "epoch": 0.3417390840752786, "grad_norm": 5.137548923492432, "learning_rate": 8.7182e-06, "loss": 0.0811, "step": 6410 }, { "epoch": 0.3417923975049315, "grad_norm": 2.6496095657348633, "learning_rate": 8.718e-06, "loss": 0.0297, "step": 6411 }, { "epoch": 0.34184571093458443, "grad_norm": 2.015059471130371, "learning_rate": 8.717800000000001e-06, "loss": 0.0257, "step": 6412 }, { "epoch": 0.34189902436423736, "grad_norm": 5.187122344970703, "learning_rate": 8.717600000000001e-06, "loss": 0.0816, "step": 6413 }, { "epoch": 0.3419523377938903, "grad_norm": 2.580930233001709, "learning_rate": 8.7174e-06, "loss": 0.0337, "step": 6414 }, { "epoch": 0.3420056512235432, "grad_norm": 2.5372908115386963, "learning_rate": 8.717200000000002e-06, "loss": 0.0415, "step": 6415 }, { "epoch": 0.34205896465319613, "grad_norm": 2.9087698459625244, "learning_rate": 8.717000000000001e-06, "loss": 0.0437, "step": 6416 }, { "epoch": 0.34211227808284905, "grad_norm": 2.286715030670166, "learning_rate": 8.716800000000001e-06, "loss": 0.028, "step": 6417 }, { "epoch": 0.342165591512502, "grad_norm": 4.705673694610596, "learning_rate": 8.7166e-06, "loss": 0.0471, "step": 6418 }, { "epoch": 0.34221890494215496, "grad_norm": 1.5457533597946167, "learning_rate": 8.7164e-06, "loss": 0.0189, "step": 6419 }, { "epoch": 0.3422722183718079, "grad_norm": 4.971792697906494, "learning_rate": 8.7162e-06, "loss": 0.047, "step": 6420 }, { "epoch": 0.3423255318014608, "grad_norm": 6.643716335296631, "learning_rate": 8.716000000000001e-06, "loss": 0.0705, "step": 6421 }, { "epoch": 0.3423788452311137, "grad_norm": 6.339252471923828, "learning_rate": 8.7158e-06, "loss": 0.0722, "step": 6422 }, { "epoch": 0.34243215866076665, "grad_norm": 2.6940150260925293, "learning_rate": 8.7156e-06, "loss": 0.0279, "step": 6423 }, { "epoch": 0.3424854720904196, "grad_norm": 3.6590631008148193, "learning_rate": 8.715400000000001e-06, "loss": 0.0553, "step": 6424 }, { "epoch": 0.3425387855200725, "grad_norm": 3.2144694328308105, "learning_rate": 8.715200000000001e-06, "loss": 0.0525, "step": 6425 }, { "epoch": 0.3425920989497254, "grad_norm": 3.7723240852355957, "learning_rate": 8.715e-06, "loss": 0.0377, "step": 6426 }, { "epoch": 0.34264541237937834, "grad_norm": 3.732363224029541, "learning_rate": 8.714800000000002e-06, "loss": 0.0446, "step": 6427 }, { "epoch": 0.34269872580903127, "grad_norm": 5.520730018615723, "learning_rate": 8.714600000000001e-06, "loss": 0.0659, "step": 6428 }, { "epoch": 0.34275203923868425, "grad_norm": 2.422868490219116, "learning_rate": 8.7144e-06, "loss": 0.0485, "step": 6429 }, { "epoch": 0.34280535266833717, "grad_norm": 2.102557897567749, "learning_rate": 8.7142e-06, "loss": 0.0155, "step": 6430 }, { "epoch": 0.3428586660979901, "grad_norm": 1.6362144947052002, "learning_rate": 8.714e-06, "loss": 0.0316, "step": 6431 }, { "epoch": 0.342911979527643, "grad_norm": 3.4207773208618164, "learning_rate": 8.713800000000001e-06, "loss": 0.0377, "step": 6432 }, { "epoch": 0.34296529295729594, "grad_norm": 2.288095712661743, "learning_rate": 8.7136e-06, "loss": 0.0235, "step": 6433 }, { "epoch": 0.34301860638694887, "grad_norm": 3.2417514324188232, "learning_rate": 8.7134e-06, "loss": 0.0567, "step": 6434 }, { "epoch": 0.3430719198166018, "grad_norm": 3.420114755630493, "learning_rate": 8.7132e-06, "loss": 0.0482, "step": 6435 }, { "epoch": 0.3431252332462547, "grad_norm": 3.6006128787994385, "learning_rate": 8.713000000000001e-06, "loss": 0.0447, "step": 6436 }, { "epoch": 0.34317854667590764, "grad_norm": 3.5425519943237305, "learning_rate": 8.7128e-06, "loss": 0.0478, "step": 6437 }, { "epoch": 0.3432318601055606, "grad_norm": 3.0825252532958984, "learning_rate": 8.7126e-06, "loss": 0.0398, "step": 6438 }, { "epoch": 0.34328517353521354, "grad_norm": 2.0599000453948975, "learning_rate": 8.712400000000001e-06, "loss": 0.028, "step": 6439 }, { "epoch": 0.34333848696486646, "grad_norm": 4.039651870727539, "learning_rate": 8.712200000000001e-06, "loss": 0.0478, "step": 6440 }, { "epoch": 0.3433918003945194, "grad_norm": 4.328583717346191, "learning_rate": 8.712e-06, "loss": 0.0398, "step": 6441 }, { "epoch": 0.3434451138241723, "grad_norm": 2.7193076610565186, "learning_rate": 8.7118e-06, "loss": 0.0499, "step": 6442 }, { "epoch": 0.34349842725382523, "grad_norm": 4.238212585449219, "learning_rate": 8.711600000000001e-06, "loss": 0.056, "step": 6443 }, { "epoch": 0.34355174068347816, "grad_norm": 4.402920722961426, "learning_rate": 8.711400000000001e-06, "loss": 0.0412, "step": 6444 }, { "epoch": 0.3436050541131311, "grad_norm": 3.127131462097168, "learning_rate": 8.7112e-06, "loss": 0.0337, "step": 6445 }, { "epoch": 0.343658367542784, "grad_norm": 5.584523677825928, "learning_rate": 8.711e-06, "loss": 0.055, "step": 6446 }, { "epoch": 0.343711680972437, "grad_norm": 2.4252843856811523, "learning_rate": 8.7108e-06, "loss": 0.0257, "step": 6447 }, { "epoch": 0.3437649944020899, "grad_norm": 3.849644660949707, "learning_rate": 8.710600000000001e-06, "loss": 0.0478, "step": 6448 }, { "epoch": 0.34381830783174283, "grad_norm": 3.9956088066101074, "learning_rate": 8.7104e-06, "loss": 0.0693, "step": 6449 }, { "epoch": 0.34387162126139575, "grad_norm": 4.016221523284912, "learning_rate": 8.7102e-06, "loss": 0.0378, "step": 6450 }, { "epoch": 0.3439249346910487, "grad_norm": 4.129312992095947, "learning_rate": 8.710000000000001e-06, "loss": 0.039, "step": 6451 }, { "epoch": 0.3439782481207016, "grad_norm": 2.256220579147339, "learning_rate": 8.709800000000001e-06, "loss": 0.0264, "step": 6452 }, { "epoch": 0.3440315615503545, "grad_norm": 4.937298774719238, "learning_rate": 8.7096e-06, "loss": 0.0587, "step": 6453 }, { "epoch": 0.34408487498000745, "grad_norm": 2.5545198917388916, "learning_rate": 8.709400000000002e-06, "loss": 0.0445, "step": 6454 }, { "epoch": 0.3441381884096604, "grad_norm": 3.2230939865112305, "learning_rate": 8.709200000000001e-06, "loss": 0.0264, "step": 6455 }, { "epoch": 0.3441915018393133, "grad_norm": 2.6747357845306396, "learning_rate": 8.709e-06, "loss": 0.0354, "step": 6456 }, { "epoch": 0.3442448152689663, "grad_norm": 2.2736058235168457, "learning_rate": 8.7088e-06, "loss": 0.0115, "step": 6457 }, { "epoch": 0.3442981286986192, "grad_norm": 2.842771530151367, "learning_rate": 8.7086e-06, "loss": 0.0422, "step": 6458 }, { "epoch": 0.3443514421282721, "grad_norm": 4.028233051300049, "learning_rate": 8.7084e-06, "loss": 0.0518, "step": 6459 }, { "epoch": 0.34440475555792505, "grad_norm": 2.017223834991455, "learning_rate": 8.7082e-06, "loss": 0.0213, "step": 6460 }, { "epoch": 0.34445806898757797, "grad_norm": 3.8933663368225098, "learning_rate": 8.708e-06, "loss": 0.0608, "step": 6461 }, { "epoch": 0.3445113824172309, "grad_norm": 4.84832763671875, "learning_rate": 8.7078e-06, "loss": 0.0354, "step": 6462 }, { "epoch": 0.3445646958468838, "grad_norm": 3.214953899383545, "learning_rate": 8.707600000000001e-06, "loss": 0.057, "step": 6463 }, { "epoch": 0.34461800927653674, "grad_norm": 3.1050729751586914, "learning_rate": 8.7074e-06, "loss": 0.0668, "step": 6464 }, { "epoch": 0.34467132270618966, "grad_norm": 3.83083438873291, "learning_rate": 8.707200000000002e-06, "loss": 0.0817, "step": 6465 }, { "epoch": 0.34472463613584264, "grad_norm": 1.9675331115722656, "learning_rate": 8.707000000000002e-06, "loss": 0.0327, "step": 6466 }, { "epoch": 0.34477794956549557, "grad_norm": 2.8411638736724854, "learning_rate": 8.706800000000001e-06, "loss": 0.0485, "step": 6467 }, { "epoch": 0.3448312629951485, "grad_norm": 2.1216464042663574, "learning_rate": 8.7066e-06, "loss": 0.0302, "step": 6468 }, { "epoch": 0.3448845764248014, "grad_norm": 2.1450469493865967, "learning_rate": 8.7064e-06, "loss": 0.0235, "step": 6469 }, { "epoch": 0.34493788985445434, "grad_norm": 3.252016544342041, "learning_rate": 8.7062e-06, "loss": 0.0429, "step": 6470 }, { "epoch": 0.34499120328410726, "grad_norm": 3.7521815299987793, "learning_rate": 8.706000000000001e-06, "loss": 0.0528, "step": 6471 }, { "epoch": 0.3450445167137602, "grad_norm": 2.0555872917175293, "learning_rate": 8.7058e-06, "loss": 0.0247, "step": 6472 }, { "epoch": 0.3450978301434131, "grad_norm": 3.1259419918060303, "learning_rate": 8.7056e-06, "loss": 0.0353, "step": 6473 }, { "epoch": 0.34515114357306603, "grad_norm": 2.447558641433716, "learning_rate": 8.7054e-06, "loss": 0.0237, "step": 6474 }, { "epoch": 0.345204457002719, "grad_norm": 2.7037110328674316, "learning_rate": 8.705200000000001e-06, "loss": 0.0372, "step": 6475 }, { "epoch": 0.34525777043237194, "grad_norm": 3.804978132247925, "learning_rate": 8.705e-06, "loss": 0.0384, "step": 6476 }, { "epoch": 0.34531108386202486, "grad_norm": 4.3046770095825195, "learning_rate": 8.704800000000002e-06, "loss": 0.0393, "step": 6477 }, { "epoch": 0.3453643972916778, "grad_norm": 3.377735137939453, "learning_rate": 8.704600000000001e-06, "loss": 0.0259, "step": 6478 }, { "epoch": 0.3454177107213307, "grad_norm": 2.0735974311828613, "learning_rate": 8.704400000000001e-06, "loss": 0.024, "step": 6479 }, { "epoch": 0.34547102415098363, "grad_norm": 0.9550482034683228, "learning_rate": 8.7042e-06, "loss": 0.008, "step": 6480 }, { "epoch": 0.34552433758063655, "grad_norm": 3.441889762878418, "learning_rate": 8.704e-06, "loss": 0.0386, "step": 6481 }, { "epoch": 0.3455776510102895, "grad_norm": 4.009936809539795, "learning_rate": 8.703800000000001e-06, "loss": 0.0599, "step": 6482 }, { "epoch": 0.3456309644399424, "grad_norm": 3.832531213760376, "learning_rate": 8.7036e-06, "loss": 0.0889, "step": 6483 }, { "epoch": 0.3456842778695953, "grad_norm": 4.77934455871582, "learning_rate": 8.7034e-06, "loss": 0.0749, "step": 6484 }, { "epoch": 0.3457375912992483, "grad_norm": 3.325566530227661, "learning_rate": 8.7032e-06, "loss": 0.0487, "step": 6485 }, { "epoch": 0.3457909047289012, "grad_norm": 2.8847081661224365, "learning_rate": 8.703e-06, "loss": 0.0602, "step": 6486 }, { "epoch": 0.34584421815855415, "grad_norm": 4.1577301025390625, "learning_rate": 8.7028e-06, "loss": 0.0651, "step": 6487 }, { "epoch": 0.3458975315882071, "grad_norm": 2.864990234375, "learning_rate": 8.7026e-06, "loss": 0.0331, "step": 6488 }, { "epoch": 0.34595084501786, "grad_norm": 2.9807183742523193, "learning_rate": 8.702400000000002e-06, "loss": 0.0422, "step": 6489 }, { "epoch": 0.3460041584475129, "grad_norm": 3.0154528617858887, "learning_rate": 8.702200000000001e-06, "loss": 0.0603, "step": 6490 }, { "epoch": 0.34605747187716585, "grad_norm": 3.127258777618408, "learning_rate": 8.702e-06, "loss": 0.0534, "step": 6491 }, { "epoch": 0.34611078530681877, "grad_norm": 4.124026775360107, "learning_rate": 8.7018e-06, "loss": 0.0453, "step": 6492 }, { "epoch": 0.3461640987364717, "grad_norm": 3.53733229637146, "learning_rate": 8.701600000000002e-06, "loss": 0.0636, "step": 6493 }, { "epoch": 0.34621741216612467, "grad_norm": 4.193565368652344, "learning_rate": 8.701400000000001e-06, "loss": 0.0818, "step": 6494 }, { "epoch": 0.3462707255957776, "grad_norm": 1.6037148237228394, "learning_rate": 8.7012e-06, "loss": 0.03, "step": 6495 }, { "epoch": 0.3463240390254305, "grad_norm": 2.954594373703003, "learning_rate": 8.701e-06, "loss": 0.0633, "step": 6496 }, { "epoch": 0.34637735245508344, "grad_norm": 4.218498229980469, "learning_rate": 8.7008e-06, "loss": 0.0724, "step": 6497 }, { "epoch": 0.34643066588473637, "grad_norm": 2.2140653133392334, "learning_rate": 8.7006e-06, "loss": 0.0272, "step": 6498 }, { "epoch": 0.3464839793143893, "grad_norm": 2.321164608001709, "learning_rate": 8.7004e-06, "loss": 0.0365, "step": 6499 }, { "epoch": 0.3465372927440422, "grad_norm": 3.4253549575805664, "learning_rate": 8.7002e-06, "loss": 0.0491, "step": 6500 }, { "epoch": 0.34659060617369514, "grad_norm": 1.594071388244629, "learning_rate": 8.700000000000001e-06, "loss": 0.016, "step": 6501 }, { "epoch": 0.34664391960334806, "grad_norm": 4.200340270996094, "learning_rate": 8.699800000000001e-06, "loss": 0.0416, "step": 6502 }, { "epoch": 0.34669723303300104, "grad_norm": 3.1073596477508545, "learning_rate": 8.6996e-06, "loss": 0.0658, "step": 6503 }, { "epoch": 0.34675054646265396, "grad_norm": 4.53737735748291, "learning_rate": 8.699400000000002e-06, "loss": 0.0677, "step": 6504 }, { "epoch": 0.3468038598923069, "grad_norm": 3.5337953567504883, "learning_rate": 8.699200000000001e-06, "loss": 0.0438, "step": 6505 }, { "epoch": 0.3468571733219598, "grad_norm": 2.1783816814422607, "learning_rate": 8.699000000000001e-06, "loss": 0.0404, "step": 6506 }, { "epoch": 0.34691048675161273, "grad_norm": 4.648090839385986, "learning_rate": 8.6988e-06, "loss": 0.0745, "step": 6507 }, { "epoch": 0.34696380018126566, "grad_norm": 3.2444705963134766, "learning_rate": 8.6986e-06, "loss": 0.0414, "step": 6508 }, { "epoch": 0.3470171136109186, "grad_norm": 4.5937933921813965, "learning_rate": 8.6984e-06, "loss": 0.0286, "step": 6509 }, { "epoch": 0.3470704270405715, "grad_norm": 3.8019232749938965, "learning_rate": 8.698200000000001e-06, "loss": 0.0528, "step": 6510 }, { "epoch": 0.34712374047022443, "grad_norm": 1.719130277633667, "learning_rate": 8.698e-06, "loss": 0.0167, "step": 6511 }, { "epoch": 0.34717705389987735, "grad_norm": 2.481963872909546, "learning_rate": 8.6978e-06, "loss": 0.0344, "step": 6512 }, { "epoch": 0.34723036732953033, "grad_norm": 1.5353684425354004, "learning_rate": 8.697600000000001e-06, "loss": 0.0427, "step": 6513 }, { "epoch": 0.34728368075918326, "grad_norm": 3.1973700523376465, "learning_rate": 8.6974e-06, "loss": 0.0438, "step": 6514 }, { "epoch": 0.3473369941888362, "grad_norm": 4.085268020629883, "learning_rate": 8.697200000000002e-06, "loss": 0.0859, "step": 6515 }, { "epoch": 0.3473903076184891, "grad_norm": 2.284015417098999, "learning_rate": 8.697000000000002e-06, "loss": 0.0182, "step": 6516 }, { "epoch": 0.347443621048142, "grad_norm": 2.3996737003326416, "learning_rate": 8.696800000000001e-06, "loss": 0.0316, "step": 6517 }, { "epoch": 0.34749693447779495, "grad_norm": 3.8100180625915527, "learning_rate": 8.6966e-06, "loss": 0.0632, "step": 6518 }, { "epoch": 0.3475502479074479, "grad_norm": 3.0826849937438965, "learning_rate": 8.6964e-06, "loss": 0.0431, "step": 6519 }, { "epoch": 0.3476035613371008, "grad_norm": 2.998323678970337, "learning_rate": 8.6962e-06, "loss": 0.0588, "step": 6520 }, { "epoch": 0.3476568747667537, "grad_norm": 2.812058687210083, "learning_rate": 8.696000000000001e-06, "loss": 0.0411, "step": 6521 }, { "epoch": 0.3477101881964067, "grad_norm": 6.822702407836914, "learning_rate": 8.6958e-06, "loss": 0.0316, "step": 6522 }, { "epoch": 0.3477635016260596, "grad_norm": 3.690701961517334, "learning_rate": 8.6956e-06, "loss": 0.0432, "step": 6523 }, { "epoch": 0.34781681505571255, "grad_norm": 2.3791017532348633, "learning_rate": 8.6954e-06, "loss": 0.0509, "step": 6524 }, { "epoch": 0.34787012848536547, "grad_norm": 3.983503580093384, "learning_rate": 8.695200000000001e-06, "loss": 0.0437, "step": 6525 }, { "epoch": 0.3479234419150184, "grad_norm": 2.3877811431884766, "learning_rate": 8.695e-06, "loss": 0.0273, "step": 6526 }, { "epoch": 0.3479767553446713, "grad_norm": 2.3495843410491943, "learning_rate": 8.694800000000002e-06, "loss": 0.054, "step": 6527 }, { "epoch": 0.34803006877432424, "grad_norm": 3.3564841747283936, "learning_rate": 8.694600000000001e-06, "loss": 0.0416, "step": 6528 }, { "epoch": 0.34808338220397717, "grad_norm": 2.2347726821899414, "learning_rate": 8.694400000000001e-06, "loss": 0.0219, "step": 6529 }, { "epoch": 0.3481366956336301, "grad_norm": 1.6126121282577515, "learning_rate": 8.6942e-06, "loss": 0.0228, "step": 6530 }, { "epoch": 0.34819000906328307, "grad_norm": 3.6144120693206787, "learning_rate": 8.694e-06, "loss": 0.0803, "step": 6531 }, { "epoch": 0.348243322492936, "grad_norm": 2.980243682861328, "learning_rate": 8.693800000000001e-06, "loss": 0.0382, "step": 6532 }, { "epoch": 0.3482966359225889, "grad_norm": 3.7440667152404785, "learning_rate": 8.693600000000001e-06, "loss": 0.0443, "step": 6533 }, { "epoch": 0.34834994935224184, "grad_norm": 4.238636016845703, "learning_rate": 8.6934e-06, "loss": 0.0737, "step": 6534 }, { "epoch": 0.34840326278189476, "grad_norm": 3.2716894149780273, "learning_rate": 8.6932e-06, "loss": 0.0572, "step": 6535 }, { "epoch": 0.3484565762115477, "grad_norm": 5.845613956451416, "learning_rate": 8.693e-06, "loss": 0.0817, "step": 6536 }, { "epoch": 0.3485098896412006, "grad_norm": 1.621200680732727, "learning_rate": 8.692800000000001e-06, "loss": 0.0139, "step": 6537 }, { "epoch": 0.34856320307085353, "grad_norm": 4.674535751342773, "learning_rate": 8.6926e-06, "loss": 0.0534, "step": 6538 }, { "epoch": 0.34861651650050646, "grad_norm": 3.8309054374694824, "learning_rate": 8.692400000000002e-06, "loss": 0.0276, "step": 6539 }, { "epoch": 0.3486698299301594, "grad_norm": 3.077784299850464, "learning_rate": 8.692200000000001e-06, "loss": 0.0293, "step": 6540 }, { "epoch": 0.34872314335981236, "grad_norm": 3.9311447143554688, "learning_rate": 8.692e-06, "loss": 0.0563, "step": 6541 }, { "epoch": 0.3487764567894653, "grad_norm": 2.584240198135376, "learning_rate": 8.6918e-06, "loss": 0.0288, "step": 6542 }, { "epoch": 0.3488297702191182, "grad_norm": 3.940087080001831, "learning_rate": 8.691600000000002e-06, "loss": 0.0398, "step": 6543 }, { "epoch": 0.34888308364877113, "grad_norm": 3.9365131855010986, "learning_rate": 8.691400000000001e-06, "loss": 0.0465, "step": 6544 }, { "epoch": 0.34893639707842405, "grad_norm": 3.088461399078369, "learning_rate": 8.6912e-06, "loss": 0.0415, "step": 6545 }, { "epoch": 0.348989710508077, "grad_norm": 4.43233585357666, "learning_rate": 8.691e-06, "loss": 0.0672, "step": 6546 }, { "epoch": 0.3490430239377299, "grad_norm": 2.5947139263153076, "learning_rate": 8.6908e-06, "loss": 0.0316, "step": 6547 }, { "epoch": 0.3490963373673828, "grad_norm": 3.5945615768432617, "learning_rate": 8.690600000000001e-06, "loss": 0.0372, "step": 6548 }, { "epoch": 0.34914965079703575, "grad_norm": 1.1897802352905273, "learning_rate": 8.6904e-06, "loss": 0.0131, "step": 6549 }, { "epoch": 0.34920296422668873, "grad_norm": 5.486077785491943, "learning_rate": 8.6902e-06, "loss": 0.0509, "step": 6550 }, { "epoch": 0.34925627765634165, "grad_norm": 5.6699910163879395, "learning_rate": 8.690000000000002e-06, "loss": 0.0874, "step": 6551 }, { "epoch": 0.3493095910859946, "grad_norm": 3.437262773513794, "learning_rate": 8.689800000000001e-06, "loss": 0.0724, "step": 6552 }, { "epoch": 0.3493629045156475, "grad_norm": 1.669832706451416, "learning_rate": 8.6896e-06, "loss": 0.0381, "step": 6553 }, { "epoch": 0.3494162179453004, "grad_norm": 2.522353410720825, "learning_rate": 8.689400000000002e-06, "loss": 0.052, "step": 6554 }, { "epoch": 0.34946953137495335, "grad_norm": 3.1411943435668945, "learning_rate": 8.689200000000001e-06, "loss": 0.0402, "step": 6555 }, { "epoch": 0.34952284480460627, "grad_norm": 2.617079973220825, "learning_rate": 8.689000000000001e-06, "loss": 0.0334, "step": 6556 }, { "epoch": 0.3495761582342592, "grad_norm": 2.7325046062469482, "learning_rate": 8.6888e-06, "loss": 0.0397, "step": 6557 }, { "epoch": 0.3496294716639121, "grad_norm": 2.962414264678955, "learning_rate": 8.6886e-06, "loss": 0.0294, "step": 6558 }, { "epoch": 0.3496827850935651, "grad_norm": 4.471792221069336, "learning_rate": 8.6884e-06, "loss": 0.0541, "step": 6559 }, { "epoch": 0.349736098523218, "grad_norm": 3.0253660678863525, "learning_rate": 8.688200000000001e-06, "loss": 0.0289, "step": 6560 }, { "epoch": 0.34978941195287094, "grad_norm": 2.448425531387329, "learning_rate": 8.688e-06, "loss": 0.0458, "step": 6561 }, { "epoch": 0.34984272538252387, "grad_norm": 2.294752597808838, "learning_rate": 8.6878e-06, "loss": 0.0407, "step": 6562 }, { "epoch": 0.3498960388121768, "grad_norm": 1.6771225929260254, "learning_rate": 8.687600000000001e-06, "loss": 0.0257, "step": 6563 }, { "epoch": 0.3499493522418297, "grad_norm": 2.4334182739257812, "learning_rate": 8.687400000000001e-06, "loss": 0.039, "step": 6564 }, { "epoch": 0.35000266567148264, "grad_norm": 3.1106245517730713, "learning_rate": 8.6872e-06, "loss": 0.0579, "step": 6565 }, { "epoch": 0.35005597910113556, "grad_norm": 3.2288777828216553, "learning_rate": 8.687000000000002e-06, "loss": 0.0348, "step": 6566 }, { "epoch": 0.3501092925307885, "grad_norm": 3.069936513900757, "learning_rate": 8.686800000000001e-06, "loss": 0.071, "step": 6567 }, { "epoch": 0.3501626059604414, "grad_norm": 2.727678060531616, "learning_rate": 8.6866e-06, "loss": 0.0261, "step": 6568 }, { "epoch": 0.3502159193900944, "grad_norm": 3.2554867267608643, "learning_rate": 8.6864e-06, "loss": 0.0642, "step": 6569 }, { "epoch": 0.3502692328197473, "grad_norm": 3.013612747192383, "learning_rate": 8.6862e-06, "loss": 0.0496, "step": 6570 }, { "epoch": 0.35032254624940024, "grad_norm": 3.704342842102051, "learning_rate": 8.686000000000001e-06, "loss": 0.0354, "step": 6571 }, { "epoch": 0.35037585967905316, "grad_norm": 2.594977855682373, "learning_rate": 8.6858e-06, "loss": 0.0415, "step": 6572 }, { "epoch": 0.3504291731087061, "grad_norm": 7.04206657409668, "learning_rate": 8.6856e-06, "loss": 0.0875, "step": 6573 }, { "epoch": 0.350482486538359, "grad_norm": 2.512885093688965, "learning_rate": 8.6854e-06, "loss": 0.031, "step": 6574 }, { "epoch": 0.35053579996801193, "grad_norm": 2.333899736404419, "learning_rate": 8.685200000000001e-06, "loss": 0.0256, "step": 6575 }, { "epoch": 0.35058911339766485, "grad_norm": 2.545588493347168, "learning_rate": 8.685e-06, "loss": 0.0491, "step": 6576 }, { "epoch": 0.3506424268273178, "grad_norm": 2.0134828090667725, "learning_rate": 8.6848e-06, "loss": 0.024, "step": 6577 }, { "epoch": 0.35069574025697076, "grad_norm": 2.75311541557312, "learning_rate": 8.684600000000002e-06, "loss": 0.0397, "step": 6578 }, { "epoch": 0.3507490536866237, "grad_norm": 1.4830639362335205, "learning_rate": 8.684400000000001e-06, "loss": 0.0173, "step": 6579 }, { "epoch": 0.3508023671162766, "grad_norm": 4.777902126312256, "learning_rate": 8.6842e-06, "loss": 0.0441, "step": 6580 }, { "epoch": 0.3508556805459295, "grad_norm": 3.4719889163970947, "learning_rate": 8.684e-06, "loss": 0.069, "step": 6581 }, { "epoch": 0.35090899397558245, "grad_norm": 4.192148685455322, "learning_rate": 8.683800000000001e-06, "loss": 0.0593, "step": 6582 }, { "epoch": 0.3509623074052354, "grad_norm": 3.237745761871338, "learning_rate": 8.683600000000001e-06, "loss": 0.0476, "step": 6583 }, { "epoch": 0.3510156208348883, "grad_norm": 2.4786500930786133, "learning_rate": 8.6834e-06, "loss": 0.0375, "step": 6584 }, { "epoch": 0.3510689342645412, "grad_norm": 4.036406517028809, "learning_rate": 8.6832e-06, "loss": 0.0665, "step": 6585 }, { "epoch": 0.35112224769419415, "grad_norm": 2.090715169906616, "learning_rate": 8.683e-06, "loss": 0.0252, "step": 6586 }, { "epoch": 0.3511755611238471, "grad_norm": 1.9017326831817627, "learning_rate": 8.682800000000001e-06, "loss": 0.0234, "step": 6587 }, { "epoch": 0.35122887455350005, "grad_norm": 2.638880968093872, "learning_rate": 8.6826e-06, "loss": 0.0378, "step": 6588 }, { "epoch": 0.35128218798315297, "grad_norm": 4.052395343780518, "learning_rate": 8.6824e-06, "loss": 0.0692, "step": 6589 }, { "epoch": 0.3513355014128059, "grad_norm": 2.308579206466675, "learning_rate": 8.682200000000001e-06, "loss": 0.048, "step": 6590 }, { "epoch": 0.3513888148424588, "grad_norm": 2.4427692890167236, "learning_rate": 8.682000000000001e-06, "loss": 0.0385, "step": 6591 }, { "epoch": 0.35144212827211174, "grad_norm": 4.333404541015625, "learning_rate": 8.6818e-06, "loss": 0.0554, "step": 6592 }, { "epoch": 0.35149544170176467, "grad_norm": 4.037243366241455, "learning_rate": 8.681600000000002e-06, "loss": 0.0586, "step": 6593 }, { "epoch": 0.3515487551314176, "grad_norm": 3.993990421295166, "learning_rate": 8.681400000000001e-06, "loss": 0.0468, "step": 6594 }, { "epoch": 0.3516020685610705, "grad_norm": 2.863539695739746, "learning_rate": 8.6812e-06, "loss": 0.057, "step": 6595 }, { "epoch": 0.35165538199072344, "grad_norm": 3.7665441036224365, "learning_rate": 8.681e-06, "loss": 0.0477, "step": 6596 }, { "epoch": 0.3517086954203764, "grad_norm": 3.502667188644409, "learning_rate": 8.6808e-06, "loss": 0.0705, "step": 6597 }, { "epoch": 0.35176200885002934, "grad_norm": 3.2033755779266357, "learning_rate": 8.680600000000001e-06, "loss": 0.0416, "step": 6598 }, { "epoch": 0.35181532227968226, "grad_norm": 1.1256508827209473, "learning_rate": 8.6804e-06, "loss": 0.011, "step": 6599 }, { "epoch": 0.3518686357093352, "grad_norm": 2.0277040004730225, "learning_rate": 8.6802e-06, "loss": 0.0247, "step": 6600 }, { "epoch": 0.3519219491389881, "grad_norm": 2.065870523452759, "learning_rate": 8.68e-06, "loss": 0.0106, "step": 6601 }, { "epoch": 0.35197526256864103, "grad_norm": 3.942598581314087, "learning_rate": 8.679800000000001e-06, "loss": 0.0622, "step": 6602 }, { "epoch": 0.35202857599829396, "grad_norm": 3.2630209922790527, "learning_rate": 8.6796e-06, "loss": 0.0571, "step": 6603 }, { "epoch": 0.3520818894279469, "grad_norm": 5.064792633056641, "learning_rate": 8.679400000000002e-06, "loss": 0.0407, "step": 6604 }, { "epoch": 0.3521352028575998, "grad_norm": 4.8570027351379395, "learning_rate": 8.679200000000002e-06, "loss": 0.051, "step": 6605 }, { "epoch": 0.3521885162872528, "grad_norm": 5.506720066070557, "learning_rate": 8.679000000000001e-06, "loss": 0.0827, "step": 6606 }, { "epoch": 0.3522418297169057, "grad_norm": 2.746037721633911, "learning_rate": 8.6788e-06, "loss": 0.0365, "step": 6607 }, { "epoch": 0.35229514314655863, "grad_norm": 4.170828342437744, "learning_rate": 8.6786e-06, "loss": 0.0674, "step": 6608 }, { "epoch": 0.35234845657621156, "grad_norm": 2.897207498550415, "learning_rate": 8.6784e-06, "loss": 0.0412, "step": 6609 }, { "epoch": 0.3524017700058645, "grad_norm": 1.97072434425354, "learning_rate": 8.678200000000001e-06, "loss": 0.0189, "step": 6610 }, { "epoch": 0.3524550834355174, "grad_norm": 2.3688812255859375, "learning_rate": 8.678e-06, "loss": 0.0418, "step": 6611 }, { "epoch": 0.3525083968651703, "grad_norm": 4.060761451721191, "learning_rate": 8.6778e-06, "loss": 0.0769, "step": 6612 }, { "epoch": 0.35256171029482325, "grad_norm": 5.014840126037598, "learning_rate": 8.6776e-06, "loss": 0.0463, "step": 6613 }, { "epoch": 0.3526150237244762, "grad_norm": 2.450645923614502, "learning_rate": 8.677400000000001e-06, "loss": 0.03, "step": 6614 }, { "epoch": 0.35266833715412915, "grad_norm": 3.2181577682495117, "learning_rate": 8.6772e-06, "loss": 0.0285, "step": 6615 }, { "epoch": 0.3527216505837821, "grad_norm": 3.0797882080078125, "learning_rate": 8.677000000000002e-06, "loss": 0.0502, "step": 6616 }, { "epoch": 0.352774964013435, "grad_norm": 3.7743825912475586, "learning_rate": 8.676800000000001e-06, "loss": 0.0631, "step": 6617 }, { "epoch": 0.3528282774430879, "grad_norm": 2.118518590927124, "learning_rate": 8.676600000000001e-06, "loss": 0.03, "step": 6618 }, { "epoch": 0.35288159087274085, "grad_norm": 1.5313423871994019, "learning_rate": 8.6764e-06, "loss": 0.0378, "step": 6619 }, { "epoch": 0.35293490430239377, "grad_norm": 2.988175630569458, "learning_rate": 8.6762e-06, "loss": 0.0661, "step": 6620 }, { "epoch": 0.3529882177320467, "grad_norm": 2.9416747093200684, "learning_rate": 8.676000000000001e-06, "loss": 0.0448, "step": 6621 }, { "epoch": 0.3530415311616996, "grad_norm": 2.5446276664733887, "learning_rate": 8.675800000000001e-06, "loss": 0.035, "step": 6622 }, { "epoch": 0.35309484459135254, "grad_norm": 2.9994349479675293, "learning_rate": 8.6756e-06, "loss": 0.0504, "step": 6623 }, { "epoch": 0.35314815802100546, "grad_norm": 2.1116528511047363, "learning_rate": 8.6754e-06, "loss": 0.0482, "step": 6624 }, { "epoch": 0.35320147145065844, "grad_norm": 3.1380388736724854, "learning_rate": 8.6752e-06, "loss": 0.0267, "step": 6625 }, { "epoch": 0.35325478488031137, "grad_norm": 2.7332956790924072, "learning_rate": 8.675e-06, "loss": 0.0349, "step": 6626 }, { "epoch": 0.3533080983099643, "grad_norm": 2.4912538528442383, "learning_rate": 8.6748e-06, "loss": 0.0377, "step": 6627 }, { "epoch": 0.3533614117396172, "grad_norm": 3.9125194549560547, "learning_rate": 8.674600000000002e-06, "loss": 0.0344, "step": 6628 }, { "epoch": 0.35341472516927014, "grad_norm": 2.4794015884399414, "learning_rate": 8.674400000000001e-06, "loss": 0.0469, "step": 6629 }, { "epoch": 0.35346803859892306, "grad_norm": 2.586005926132202, "learning_rate": 8.6742e-06, "loss": 0.0474, "step": 6630 }, { "epoch": 0.353521352028576, "grad_norm": 3.6025876998901367, "learning_rate": 8.674e-06, "loss": 0.026, "step": 6631 }, { "epoch": 0.3535746654582289, "grad_norm": 4.513936519622803, "learning_rate": 8.673800000000002e-06, "loss": 0.0568, "step": 6632 }, { "epoch": 0.35362797888788183, "grad_norm": 5.339632034301758, "learning_rate": 8.673600000000001e-06, "loss": 0.0498, "step": 6633 }, { "epoch": 0.3536812923175348, "grad_norm": 4.552511692047119, "learning_rate": 8.6734e-06, "loss": 0.0562, "step": 6634 }, { "epoch": 0.35373460574718774, "grad_norm": 3.6123011112213135, "learning_rate": 8.6732e-06, "loss": 0.0521, "step": 6635 }, { "epoch": 0.35378791917684066, "grad_norm": 2.0986297130584717, "learning_rate": 8.673e-06, "loss": 0.0132, "step": 6636 }, { "epoch": 0.3538412326064936, "grad_norm": 2.4534366130828857, "learning_rate": 8.672800000000001e-06, "loss": 0.0389, "step": 6637 }, { "epoch": 0.3538945460361465, "grad_norm": 2.590766668319702, "learning_rate": 8.6726e-06, "loss": 0.0458, "step": 6638 }, { "epoch": 0.35394785946579943, "grad_norm": 4.174480438232422, "learning_rate": 8.6724e-06, "loss": 0.0457, "step": 6639 }, { "epoch": 0.35400117289545235, "grad_norm": 3.1695752143859863, "learning_rate": 8.672200000000001e-06, "loss": 0.0514, "step": 6640 }, { "epoch": 0.3540544863251053, "grad_norm": 2.621910333633423, "learning_rate": 8.672000000000001e-06, "loss": 0.0207, "step": 6641 }, { "epoch": 0.3541077997547582, "grad_norm": 1.094234585762024, "learning_rate": 8.6718e-06, "loss": 0.0109, "step": 6642 }, { "epoch": 0.3541611131844112, "grad_norm": 2.2869162559509277, "learning_rate": 8.671600000000002e-06, "loss": 0.0372, "step": 6643 }, { "epoch": 0.3542144266140641, "grad_norm": 3.8036091327667236, "learning_rate": 8.671400000000001e-06, "loss": 0.0407, "step": 6644 }, { "epoch": 0.354267740043717, "grad_norm": 2.0219316482543945, "learning_rate": 8.671200000000001e-06, "loss": 0.0202, "step": 6645 }, { "epoch": 0.35432105347336995, "grad_norm": 2.0009853839874268, "learning_rate": 8.671e-06, "loss": 0.021, "step": 6646 }, { "epoch": 0.3543743669030229, "grad_norm": 2.753291130065918, "learning_rate": 8.6708e-06, "loss": 0.0383, "step": 6647 }, { "epoch": 0.3544276803326758, "grad_norm": 5.203151702880859, "learning_rate": 8.670600000000001e-06, "loss": 0.0705, "step": 6648 }, { "epoch": 0.3544809937623287, "grad_norm": 4.793611526489258, "learning_rate": 8.670400000000001e-06, "loss": 0.0703, "step": 6649 }, { "epoch": 0.35453430719198165, "grad_norm": 1.844651699066162, "learning_rate": 8.6702e-06, "loss": 0.04, "step": 6650 }, { "epoch": 0.35458762062163457, "grad_norm": 6.308538913726807, "learning_rate": 8.67e-06, "loss": 0.0509, "step": 6651 }, { "epoch": 0.3546409340512875, "grad_norm": 4.1015119552612305, "learning_rate": 8.669800000000001e-06, "loss": 0.0691, "step": 6652 }, { "epoch": 0.35469424748094047, "grad_norm": 2.4308788776397705, "learning_rate": 8.6696e-06, "loss": 0.0213, "step": 6653 }, { "epoch": 0.3547475609105934, "grad_norm": 4.091341018676758, "learning_rate": 8.6694e-06, "loss": 0.0389, "step": 6654 }, { "epoch": 0.3548008743402463, "grad_norm": 2.8855056762695312, "learning_rate": 8.669200000000002e-06, "loss": 0.0194, "step": 6655 }, { "epoch": 0.35485418776989924, "grad_norm": 1.792884349822998, "learning_rate": 8.669000000000001e-06, "loss": 0.0151, "step": 6656 }, { "epoch": 0.35490750119955217, "grad_norm": 5.197785377502441, "learning_rate": 8.6688e-06, "loss": 0.0542, "step": 6657 }, { "epoch": 0.3549608146292051, "grad_norm": 1.937870979309082, "learning_rate": 8.6686e-06, "loss": 0.046, "step": 6658 }, { "epoch": 0.355014128058858, "grad_norm": 2.2766737937927246, "learning_rate": 8.668400000000002e-06, "loss": 0.0482, "step": 6659 }, { "epoch": 0.35506744148851094, "grad_norm": 3.3601646423339844, "learning_rate": 8.668200000000001e-06, "loss": 0.0547, "step": 6660 }, { "epoch": 0.35512075491816386, "grad_norm": 3.517738103866577, "learning_rate": 8.668e-06, "loss": 0.051, "step": 6661 }, { "epoch": 0.35517406834781684, "grad_norm": 4.503619194030762, "learning_rate": 8.6678e-06, "loss": 0.0633, "step": 6662 }, { "epoch": 0.35522738177746976, "grad_norm": 1.8225140571594238, "learning_rate": 8.6676e-06, "loss": 0.0132, "step": 6663 }, { "epoch": 0.3552806952071227, "grad_norm": 4.313034534454346, "learning_rate": 8.667400000000001e-06, "loss": 0.0513, "step": 6664 }, { "epoch": 0.3553340086367756, "grad_norm": 9.375164031982422, "learning_rate": 8.6672e-06, "loss": 0.0662, "step": 6665 }, { "epoch": 0.35538732206642853, "grad_norm": 5.127685546875, "learning_rate": 8.667e-06, "loss": 0.0582, "step": 6666 }, { "epoch": 0.35544063549608146, "grad_norm": 3.011824131011963, "learning_rate": 8.666800000000001e-06, "loss": 0.0447, "step": 6667 }, { "epoch": 0.3554939489257344, "grad_norm": 4.043291091918945, "learning_rate": 8.666600000000001e-06, "loss": 0.0275, "step": 6668 }, { "epoch": 0.3555472623553873, "grad_norm": 3.279527425765991, "learning_rate": 8.6664e-06, "loss": 0.0347, "step": 6669 }, { "epoch": 0.35560057578504023, "grad_norm": 3.8965678215026855, "learning_rate": 8.6662e-06, "loss": 0.0425, "step": 6670 }, { "epoch": 0.3556538892146932, "grad_norm": 4.541014671325684, "learning_rate": 8.666000000000001e-06, "loss": 0.0935, "step": 6671 }, { "epoch": 0.35570720264434613, "grad_norm": 9.77603530883789, "learning_rate": 8.665800000000001e-06, "loss": 0.0557, "step": 6672 }, { "epoch": 0.35576051607399906, "grad_norm": 2.5551180839538574, "learning_rate": 8.6656e-06, "loss": 0.0364, "step": 6673 }, { "epoch": 0.355813829503652, "grad_norm": 6.305482864379883, "learning_rate": 8.6654e-06, "loss": 0.0378, "step": 6674 }, { "epoch": 0.3558671429333049, "grad_norm": 3.460958242416382, "learning_rate": 8.6652e-06, "loss": 0.0528, "step": 6675 }, { "epoch": 0.3559204563629578, "grad_norm": 2.4135098457336426, "learning_rate": 8.665000000000001e-06, "loss": 0.036, "step": 6676 }, { "epoch": 0.35597376979261075, "grad_norm": 6.225327491760254, "learning_rate": 8.6648e-06, "loss": 0.0613, "step": 6677 }, { "epoch": 0.3560270832222637, "grad_norm": 4.438592910766602, "learning_rate": 8.6646e-06, "loss": 0.0254, "step": 6678 }, { "epoch": 0.3560803966519166, "grad_norm": 5.276154041290283, "learning_rate": 8.664400000000001e-06, "loss": 0.0489, "step": 6679 }, { "epoch": 0.3561337100815695, "grad_norm": 3.139953136444092, "learning_rate": 8.6642e-06, "loss": 0.0338, "step": 6680 }, { "epoch": 0.3561870235112225, "grad_norm": 4.173604965209961, "learning_rate": 8.664e-06, "loss": 0.0545, "step": 6681 }, { "epoch": 0.3562403369408754, "grad_norm": 4.331448078155518, "learning_rate": 8.663800000000002e-06, "loss": 0.0858, "step": 6682 }, { "epoch": 0.35629365037052835, "grad_norm": 7.469580173492432, "learning_rate": 8.663600000000001e-06, "loss": 0.0617, "step": 6683 }, { "epoch": 0.35634696380018127, "grad_norm": 3.023527145385742, "learning_rate": 8.6634e-06, "loss": 0.0424, "step": 6684 }, { "epoch": 0.3564002772298342, "grad_norm": 3.352362632751465, "learning_rate": 8.6632e-06, "loss": 0.0638, "step": 6685 }, { "epoch": 0.3564535906594871, "grad_norm": 6.116374969482422, "learning_rate": 8.663e-06, "loss": 0.0816, "step": 6686 }, { "epoch": 0.35650690408914004, "grad_norm": 2.988477945327759, "learning_rate": 8.662800000000001e-06, "loss": 0.0307, "step": 6687 }, { "epoch": 0.35656021751879297, "grad_norm": 5.327790260314941, "learning_rate": 8.6626e-06, "loss": 0.0631, "step": 6688 }, { "epoch": 0.3566135309484459, "grad_norm": 3.86687970161438, "learning_rate": 8.6624e-06, "loss": 0.0479, "step": 6689 }, { "epoch": 0.35666684437809887, "grad_norm": 3.1045777797698975, "learning_rate": 8.6622e-06, "loss": 0.0332, "step": 6690 }, { "epoch": 0.3567201578077518, "grad_norm": 1.876712441444397, "learning_rate": 8.662000000000001e-06, "loss": 0.0281, "step": 6691 }, { "epoch": 0.3567734712374047, "grad_norm": 2.1280126571655273, "learning_rate": 8.6618e-06, "loss": 0.0315, "step": 6692 }, { "epoch": 0.35682678466705764, "grad_norm": 2.69067645072937, "learning_rate": 8.661600000000002e-06, "loss": 0.0186, "step": 6693 }, { "epoch": 0.35688009809671056, "grad_norm": 3.973257064819336, "learning_rate": 8.661400000000001e-06, "loss": 0.0583, "step": 6694 }, { "epoch": 0.3569334115263635, "grad_norm": 4.039155006408691, "learning_rate": 8.661200000000001e-06, "loss": 0.0326, "step": 6695 }, { "epoch": 0.3569867249560164, "grad_norm": 3.5466771125793457, "learning_rate": 8.661e-06, "loss": 0.0475, "step": 6696 }, { "epoch": 0.35704003838566933, "grad_norm": 2.630350112915039, "learning_rate": 8.6608e-06, "loss": 0.0272, "step": 6697 }, { "epoch": 0.35709335181532226, "grad_norm": 3.028505563735962, "learning_rate": 8.660600000000001e-06, "loss": 0.0309, "step": 6698 }, { "epoch": 0.35714666524497524, "grad_norm": 3.756223440170288, "learning_rate": 8.660400000000001e-06, "loss": 0.0265, "step": 6699 }, { "epoch": 0.35719997867462816, "grad_norm": 6.287943363189697, "learning_rate": 8.6602e-06, "loss": 0.0656, "step": 6700 }, { "epoch": 0.3572532921042811, "grad_norm": 3.724522352218628, "learning_rate": 8.66e-06, "loss": 0.0449, "step": 6701 }, { "epoch": 0.357306605533934, "grad_norm": 3.1341891288757324, "learning_rate": 8.6598e-06, "loss": 0.0455, "step": 6702 }, { "epoch": 0.35735991896358693, "grad_norm": 3.700188636779785, "learning_rate": 8.659600000000001e-06, "loss": 0.0499, "step": 6703 }, { "epoch": 0.35741323239323985, "grad_norm": 3.152574062347412, "learning_rate": 8.6594e-06, "loss": 0.0738, "step": 6704 }, { "epoch": 0.3574665458228928, "grad_norm": 2.2157182693481445, "learning_rate": 8.659200000000002e-06, "loss": 0.0223, "step": 6705 }, { "epoch": 0.3575198592525457, "grad_norm": 6.114505290985107, "learning_rate": 8.659000000000001e-06, "loss": 0.0638, "step": 6706 }, { "epoch": 0.3575731726821986, "grad_norm": 4.182809352874756, "learning_rate": 8.658800000000001e-06, "loss": 0.0463, "step": 6707 }, { "epoch": 0.35762648611185155, "grad_norm": 3.2826364040374756, "learning_rate": 8.6586e-06, "loss": 0.0352, "step": 6708 }, { "epoch": 0.35767979954150453, "grad_norm": 5.106380939483643, "learning_rate": 8.658400000000002e-06, "loss": 0.0394, "step": 6709 }, { "epoch": 0.35773311297115745, "grad_norm": 3.785836696624756, "learning_rate": 8.658200000000001e-06, "loss": 0.0427, "step": 6710 }, { "epoch": 0.3577864264008104, "grad_norm": 3.7734615802764893, "learning_rate": 8.658e-06, "loss": 0.0375, "step": 6711 }, { "epoch": 0.3578397398304633, "grad_norm": 2.724350690841675, "learning_rate": 8.6578e-06, "loss": 0.027, "step": 6712 }, { "epoch": 0.3578930532601162, "grad_norm": 2.0310099124908447, "learning_rate": 8.6576e-06, "loss": 0.0259, "step": 6713 }, { "epoch": 0.35794636668976915, "grad_norm": 2.368544340133667, "learning_rate": 8.6574e-06, "loss": 0.0518, "step": 6714 }, { "epoch": 0.35799968011942207, "grad_norm": 3.4112823009490967, "learning_rate": 8.6572e-06, "loss": 0.0292, "step": 6715 }, { "epoch": 0.358052993549075, "grad_norm": 2.1366467475891113, "learning_rate": 8.657e-06, "loss": 0.0381, "step": 6716 }, { "epoch": 0.3581063069787279, "grad_norm": 3.3628182411193848, "learning_rate": 8.656800000000002e-06, "loss": 0.048, "step": 6717 }, { "epoch": 0.3581596204083809, "grad_norm": 1.8235983848571777, "learning_rate": 8.656600000000001e-06, "loss": 0.029, "step": 6718 }, { "epoch": 0.3582129338380338, "grad_norm": 1.359545111656189, "learning_rate": 8.6564e-06, "loss": 0.0118, "step": 6719 }, { "epoch": 0.35826624726768674, "grad_norm": 3.9530673027038574, "learning_rate": 8.6562e-06, "loss": 0.0324, "step": 6720 }, { "epoch": 0.35831956069733967, "grad_norm": 2.0971744060516357, "learning_rate": 8.656000000000001e-06, "loss": 0.0291, "step": 6721 }, { "epoch": 0.3583728741269926, "grad_norm": 2.6134822368621826, "learning_rate": 8.655800000000001e-06, "loss": 0.0355, "step": 6722 }, { "epoch": 0.3584261875566455, "grad_norm": 6.210853576660156, "learning_rate": 8.6556e-06, "loss": 0.047, "step": 6723 }, { "epoch": 0.35847950098629844, "grad_norm": 2.0571954250335693, "learning_rate": 8.6554e-06, "loss": 0.0366, "step": 6724 }, { "epoch": 0.35853281441595136, "grad_norm": 3.926931858062744, "learning_rate": 8.6552e-06, "loss": 0.0443, "step": 6725 }, { "epoch": 0.3585861278456043, "grad_norm": 4.915724277496338, "learning_rate": 8.655000000000001e-06, "loss": 0.0676, "step": 6726 }, { "epoch": 0.35863944127525726, "grad_norm": 4.597232341766357, "learning_rate": 8.6548e-06, "loss": 0.0456, "step": 6727 }, { "epoch": 0.3586927547049102, "grad_norm": 3.4559926986694336, "learning_rate": 8.6546e-06, "loss": 0.0677, "step": 6728 }, { "epoch": 0.3587460681345631, "grad_norm": 2.0875084400177, "learning_rate": 8.654400000000001e-06, "loss": 0.0355, "step": 6729 }, { "epoch": 0.35879938156421604, "grad_norm": 2.0257649421691895, "learning_rate": 8.654200000000001e-06, "loss": 0.0281, "step": 6730 }, { "epoch": 0.35885269499386896, "grad_norm": 6.165207386016846, "learning_rate": 8.654e-06, "loss": 0.0791, "step": 6731 }, { "epoch": 0.3589060084235219, "grad_norm": 5.1966118812561035, "learning_rate": 8.653800000000002e-06, "loss": 0.059, "step": 6732 }, { "epoch": 0.3589593218531748, "grad_norm": 2.8112971782684326, "learning_rate": 8.653600000000001e-06, "loss": 0.042, "step": 6733 }, { "epoch": 0.35901263528282773, "grad_norm": 4.415555477142334, "learning_rate": 8.653400000000001e-06, "loss": 0.0322, "step": 6734 }, { "epoch": 0.35906594871248065, "grad_norm": 2.606133460998535, "learning_rate": 8.6532e-06, "loss": 0.0387, "step": 6735 }, { "epoch": 0.3591192621421336, "grad_norm": 2.488640069961548, "learning_rate": 8.653e-06, "loss": 0.0238, "step": 6736 }, { "epoch": 0.35917257557178656, "grad_norm": 3.6008710861206055, "learning_rate": 8.652800000000001e-06, "loss": 0.0339, "step": 6737 }, { "epoch": 0.3592258890014395, "grad_norm": 3.4691410064697266, "learning_rate": 8.6526e-06, "loss": 0.0298, "step": 6738 }, { "epoch": 0.3592792024310924, "grad_norm": 5.1600775718688965, "learning_rate": 8.6524e-06, "loss": 0.0613, "step": 6739 }, { "epoch": 0.3593325158607453, "grad_norm": 2.213996648788452, "learning_rate": 8.6522e-06, "loss": 0.0344, "step": 6740 }, { "epoch": 0.35938582929039825, "grad_norm": 6.224246501922607, "learning_rate": 8.652000000000001e-06, "loss": 0.0392, "step": 6741 }, { "epoch": 0.3594391427200512, "grad_norm": 3.464125156402588, "learning_rate": 8.6518e-06, "loss": 0.0394, "step": 6742 }, { "epoch": 0.3594924561497041, "grad_norm": 3.498321771621704, "learning_rate": 8.6516e-06, "loss": 0.0389, "step": 6743 }, { "epoch": 0.359545769579357, "grad_norm": 2.6766927242279053, "learning_rate": 8.651400000000002e-06, "loss": 0.0352, "step": 6744 }, { "epoch": 0.35959908300900995, "grad_norm": 3.481226682662964, "learning_rate": 8.651200000000001e-06, "loss": 0.0579, "step": 6745 }, { "epoch": 0.3596523964386629, "grad_norm": 4.35962438583374, "learning_rate": 8.651e-06, "loss": 0.0529, "step": 6746 }, { "epoch": 0.35970570986831585, "grad_norm": 2.4383883476257324, "learning_rate": 8.6508e-06, "loss": 0.0304, "step": 6747 }, { "epoch": 0.35975902329796877, "grad_norm": 7.982935905456543, "learning_rate": 8.650600000000002e-06, "loss": 0.0358, "step": 6748 }, { "epoch": 0.3598123367276217, "grad_norm": 3.3891642093658447, "learning_rate": 8.650400000000001e-06, "loss": 0.0616, "step": 6749 }, { "epoch": 0.3598656501572746, "grad_norm": 3.7507517337799072, "learning_rate": 8.6502e-06, "loss": 0.0233, "step": 6750 }, { "epoch": 0.35991896358692754, "grad_norm": 4.369295597076416, "learning_rate": 8.65e-06, "loss": 0.0647, "step": 6751 }, { "epoch": 0.35997227701658047, "grad_norm": 5.276477336883545, "learning_rate": 8.6498e-06, "loss": 0.0529, "step": 6752 }, { "epoch": 0.3600255904462334, "grad_norm": 1.6295419931411743, "learning_rate": 8.649600000000001e-06, "loss": 0.0144, "step": 6753 }, { "epoch": 0.3600789038758863, "grad_norm": 5.22939920425415, "learning_rate": 8.6494e-06, "loss": 0.0706, "step": 6754 }, { "epoch": 0.3601322173055393, "grad_norm": 3.381258726119995, "learning_rate": 8.6492e-06, "loss": 0.0485, "step": 6755 }, { "epoch": 0.3601855307351922, "grad_norm": 2.64654278755188, "learning_rate": 8.649000000000001e-06, "loss": 0.0311, "step": 6756 }, { "epoch": 0.36023884416484514, "grad_norm": 1.7071664333343506, "learning_rate": 8.648800000000001e-06, "loss": 0.0184, "step": 6757 }, { "epoch": 0.36029215759449806, "grad_norm": 6.2508392333984375, "learning_rate": 8.6486e-06, "loss": 0.0528, "step": 6758 }, { "epoch": 0.360345471024151, "grad_norm": 4.323482513427734, "learning_rate": 8.648400000000002e-06, "loss": 0.0634, "step": 6759 }, { "epoch": 0.3603987844538039, "grad_norm": 2.526580810546875, "learning_rate": 8.648200000000001e-06, "loss": 0.0412, "step": 6760 }, { "epoch": 0.36045209788345683, "grad_norm": 3.151905059814453, "learning_rate": 8.648000000000001e-06, "loss": 0.0466, "step": 6761 }, { "epoch": 0.36050541131310976, "grad_norm": 4.418805122375488, "learning_rate": 8.6478e-06, "loss": 0.0516, "step": 6762 }, { "epoch": 0.3605587247427627, "grad_norm": 5.582616806030273, "learning_rate": 8.6476e-06, "loss": 0.0675, "step": 6763 }, { "epoch": 0.3606120381724156, "grad_norm": 2.347465991973877, "learning_rate": 8.6474e-06, "loss": 0.0212, "step": 6764 }, { "epoch": 0.3606653516020686, "grad_norm": 3.5201332569122314, "learning_rate": 8.6472e-06, "loss": 0.0404, "step": 6765 }, { "epoch": 0.3607186650317215, "grad_norm": 4.980655670166016, "learning_rate": 8.647e-06, "loss": 0.0343, "step": 6766 }, { "epoch": 0.36077197846137443, "grad_norm": 2.5895915031433105, "learning_rate": 8.6468e-06, "loss": 0.0163, "step": 6767 }, { "epoch": 0.36082529189102736, "grad_norm": 2.751516342163086, "learning_rate": 8.646600000000001e-06, "loss": 0.0558, "step": 6768 }, { "epoch": 0.3608786053206803, "grad_norm": 6.08681583404541, "learning_rate": 8.6464e-06, "loss": 0.0807, "step": 6769 }, { "epoch": 0.3609319187503332, "grad_norm": 2.254629135131836, "learning_rate": 8.646200000000002e-06, "loss": 0.0159, "step": 6770 }, { "epoch": 0.3609852321799861, "grad_norm": 2.666405439376831, "learning_rate": 8.646000000000002e-06, "loss": 0.0261, "step": 6771 }, { "epoch": 0.36103854560963905, "grad_norm": 3.1223578453063965, "learning_rate": 8.645800000000001e-06, "loss": 0.0485, "step": 6772 }, { "epoch": 0.361091859039292, "grad_norm": 6.903550148010254, "learning_rate": 8.6456e-06, "loss": 0.105, "step": 6773 }, { "epoch": 0.36114517246894495, "grad_norm": 2.413614273071289, "learning_rate": 8.6454e-06, "loss": 0.0573, "step": 6774 }, { "epoch": 0.3611984858985979, "grad_norm": 3.017331838607788, "learning_rate": 8.6452e-06, "loss": 0.031, "step": 6775 }, { "epoch": 0.3612517993282508, "grad_norm": 6.354657173156738, "learning_rate": 8.645000000000001e-06, "loss": 0.0468, "step": 6776 }, { "epoch": 0.3613051127579037, "grad_norm": 6.4181036949157715, "learning_rate": 8.6448e-06, "loss": 0.0469, "step": 6777 }, { "epoch": 0.36135842618755665, "grad_norm": 4.20194149017334, "learning_rate": 8.6446e-06, "loss": 0.0429, "step": 6778 }, { "epoch": 0.36141173961720957, "grad_norm": 2.6267340183258057, "learning_rate": 8.6444e-06, "loss": 0.02, "step": 6779 }, { "epoch": 0.3614650530468625, "grad_norm": 2.3032824993133545, "learning_rate": 8.644200000000001e-06, "loss": 0.0207, "step": 6780 }, { "epoch": 0.3615183664765154, "grad_norm": 3.102754592895508, "learning_rate": 8.644e-06, "loss": 0.0343, "step": 6781 }, { "epoch": 0.36157167990616834, "grad_norm": 6.513744354248047, "learning_rate": 8.643800000000002e-06, "loss": 0.0909, "step": 6782 }, { "epoch": 0.3616249933358213, "grad_norm": 3.068770408630371, "learning_rate": 8.643600000000001e-06, "loss": 0.0736, "step": 6783 }, { "epoch": 0.36167830676547424, "grad_norm": 3.736541748046875, "learning_rate": 8.643400000000001e-06, "loss": 0.057, "step": 6784 }, { "epoch": 0.36173162019512717, "grad_norm": 5.650110721588135, "learning_rate": 8.6432e-06, "loss": 0.0795, "step": 6785 }, { "epoch": 0.3617849336247801, "grad_norm": 1.9959088563919067, "learning_rate": 8.643e-06, "loss": 0.0376, "step": 6786 }, { "epoch": 0.361838247054433, "grad_norm": 4.454514980316162, "learning_rate": 8.642800000000001e-06, "loss": 0.0899, "step": 6787 }, { "epoch": 0.36189156048408594, "grad_norm": 5.549509048461914, "learning_rate": 8.642600000000001e-06, "loss": 0.0702, "step": 6788 }, { "epoch": 0.36194487391373886, "grad_norm": 4.280496120452881, "learning_rate": 8.6424e-06, "loss": 0.0474, "step": 6789 }, { "epoch": 0.3619981873433918, "grad_norm": 1.9714733362197876, "learning_rate": 8.6422e-06, "loss": 0.0178, "step": 6790 }, { "epoch": 0.3620515007730447, "grad_norm": 2.503344774246216, "learning_rate": 8.642e-06, "loss": 0.0246, "step": 6791 }, { "epoch": 0.36210481420269763, "grad_norm": 2.9960172176361084, "learning_rate": 8.6418e-06, "loss": 0.0509, "step": 6792 }, { "epoch": 0.3621581276323506, "grad_norm": 2.3161447048187256, "learning_rate": 8.6416e-06, "loss": 0.04, "step": 6793 }, { "epoch": 0.36221144106200354, "grad_norm": 2.8221778869628906, "learning_rate": 8.641400000000002e-06, "loss": 0.0562, "step": 6794 }, { "epoch": 0.36226475449165646, "grad_norm": 4.008866310119629, "learning_rate": 8.641200000000001e-06, "loss": 0.0868, "step": 6795 }, { "epoch": 0.3623180679213094, "grad_norm": 2.1063363552093506, "learning_rate": 8.641e-06, "loss": 0.0325, "step": 6796 }, { "epoch": 0.3623713813509623, "grad_norm": 2.413684129714966, "learning_rate": 8.6408e-06, "loss": 0.0456, "step": 6797 }, { "epoch": 0.36242469478061523, "grad_norm": 2.178725242614746, "learning_rate": 8.640600000000002e-06, "loss": 0.0261, "step": 6798 }, { "epoch": 0.36247800821026815, "grad_norm": 1.7717857360839844, "learning_rate": 8.640400000000001e-06, "loss": 0.0396, "step": 6799 }, { "epoch": 0.3625313216399211, "grad_norm": 4.885164737701416, "learning_rate": 8.6402e-06, "loss": 0.0411, "step": 6800 }, { "epoch": 0.362584635069574, "grad_norm": 2.6178903579711914, "learning_rate": 8.64e-06, "loss": 0.0459, "step": 6801 }, { "epoch": 0.362637948499227, "grad_norm": 2.2827835083007812, "learning_rate": 8.6398e-06, "loss": 0.0481, "step": 6802 }, { "epoch": 0.3626912619288799, "grad_norm": 3.243177890777588, "learning_rate": 8.6396e-06, "loss": 0.0381, "step": 6803 }, { "epoch": 0.36274457535853283, "grad_norm": 2.354797601699829, "learning_rate": 8.6394e-06, "loss": 0.0589, "step": 6804 }, { "epoch": 0.36279788878818575, "grad_norm": 3.163078784942627, "learning_rate": 8.6392e-06, "loss": 0.0368, "step": 6805 }, { "epoch": 0.3628512022178387, "grad_norm": 2.6556317806243896, "learning_rate": 8.639000000000001e-06, "loss": 0.0657, "step": 6806 }, { "epoch": 0.3629045156474916, "grad_norm": 1.8851203918457031, "learning_rate": 8.638800000000001e-06, "loss": 0.0308, "step": 6807 }, { "epoch": 0.3629578290771445, "grad_norm": 1.5793286561965942, "learning_rate": 8.6386e-06, "loss": 0.0245, "step": 6808 }, { "epoch": 0.36301114250679745, "grad_norm": 2.2253756523132324, "learning_rate": 8.638400000000002e-06, "loss": 0.0297, "step": 6809 }, { "epoch": 0.36306445593645037, "grad_norm": 2.546670436859131, "learning_rate": 8.638200000000001e-06, "loss": 0.0225, "step": 6810 }, { "epoch": 0.36311776936610335, "grad_norm": 3.4953925609588623, "learning_rate": 8.638000000000001e-06, "loss": 0.0351, "step": 6811 }, { "epoch": 0.3631710827957563, "grad_norm": 3.2383382320404053, "learning_rate": 8.6378e-06, "loss": 0.0378, "step": 6812 }, { "epoch": 0.3632243962254092, "grad_norm": 3.5146172046661377, "learning_rate": 8.6376e-06, "loss": 0.0536, "step": 6813 }, { "epoch": 0.3632777096550621, "grad_norm": 3.093170166015625, "learning_rate": 8.6374e-06, "loss": 0.0483, "step": 6814 }, { "epoch": 0.36333102308471504, "grad_norm": 3.6070964336395264, "learning_rate": 8.637200000000001e-06, "loss": 0.0671, "step": 6815 }, { "epoch": 0.36338433651436797, "grad_norm": 3.4047586917877197, "learning_rate": 8.637e-06, "loss": 0.0466, "step": 6816 }, { "epoch": 0.3634376499440209, "grad_norm": 1.9343785047531128, "learning_rate": 8.6368e-06, "loss": 0.0148, "step": 6817 }, { "epoch": 0.3634909633736738, "grad_norm": 1.9471060037612915, "learning_rate": 8.636600000000001e-06, "loss": 0.0228, "step": 6818 }, { "epoch": 0.36354427680332674, "grad_norm": 4.091455459594727, "learning_rate": 8.636400000000001e-06, "loss": 0.0726, "step": 6819 }, { "epoch": 0.36359759023297966, "grad_norm": 1.9851963520050049, "learning_rate": 8.6362e-06, "loss": 0.0484, "step": 6820 }, { "epoch": 0.36365090366263264, "grad_norm": 3.342336654663086, "learning_rate": 8.636000000000002e-06, "loss": 0.052, "step": 6821 }, { "epoch": 0.36370421709228556, "grad_norm": 3.330819606781006, "learning_rate": 8.635800000000001e-06, "loss": 0.0493, "step": 6822 }, { "epoch": 0.3637575305219385, "grad_norm": 3.9187710285186768, "learning_rate": 8.6356e-06, "loss": 0.0488, "step": 6823 }, { "epoch": 0.3638108439515914, "grad_norm": 3.688227415084839, "learning_rate": 8.6354e-06, "loss": 0.046, "step": 6824 }, { "epoch": 0.36386415738124434, "grad_norm": 2.8953042030334473, "learning_rate": 8.6352e-06, "loss": 0.0554, "step": 6825 }, { "epoch": 0.36391747081089726, "grad_norm": 3.0950980186462402, "learning_rate": 8.635000000000001e-06, "loss": 0.0362, "step": 6826 }, { "epoch": 0.3639707842405502, "grad_norm": 2.694568157196045, "learning_rate": 8.6348e-06, "loss": 0.0317, "step": 6827 }, { "epoch": 0.3640240976702031, "grad_norm": 3.134216547012329, "learning_rate": 8.6346e-06, "loss": 0.032, "step": 6828 }, { "epoch": 0.36407741109985603, "grad_norm": 2.8181912899017334, "learning_rate": 8.6344e-06, "loss": 0.0439, "step": 6829 }, { "epoch": 0.364130724529509, "grad_norm": 4.7899394035339355, "learning_rate": 8.634200000000001e-06, "loss": 0.0954, "step": 6830 }, { "epoch": 0.36418403795916193, "grad_norm": 2.3748371601104736, "learning_rate": 8.634e-06, "loss": 0.0549, "step": 6831 }, { "epoch": 0.36423735138881486, "grad_norm": 3.0069713592529297, "learning_rate": 8.6338e-06, "loss": 0.0525, "step": 6832 }, { "epoch": 0.3642906648184678, "grad_norm": 2.242798328399658, "learning_rate": 8.633600000000001e-06, "loss": 0.0204, "step": 6833 }, { "epoch": 0.3643439782481207, "grad_norm": 4.113816738128662, "learning_rate": 8.633400000000001e-06, "loss": 0.0373, "step": 6834 }, { "epoch": 0.3643972916777736, "grad_norm": 2.4636199474334717, "learning_rate": 8.6332e-06, "loss": 0.0358, "step": 6835 }, { "epoch": 0.36445060510742655, "grad_norm": 2.443028450012207, "learning_rate": 8.633e-06, "loss": 0.0244, "step": 6836 }, { "epoch": 0.3645039185370795, "grad_norm": 4.434462070465088, "learning_rate": 8.632800000000001e-06, "loss": 0.0483, "step": 6837 }, { "epoch": 0.3645572319667324, "grad_norm": 2.5746853351593018, "learning_rate": 8.632600000000001e-06, "loss": 0.0247, "step": 6838 }, { "epoch": 0.3646105453963854, "grad_norm": 2.7223973274230957, "learning_rate": 8.6324e-06, "loss": 0.0261, "step": 6839 }, { "epoch": 0.3646638588260383, "grad_norm": 3.4587292671203613, "learning_rate": 8.6322e-06, "loss": 0.0484, "step": 6840 }, { "epoch": 0.3647171722556912, "grad_norm": 2.8399393558502197, "learning_rate": 8.632e-06, "loss": 0.0554, "step": 6841 }, { "epoch": 0.36477048568534415, "grad_norm": 5.669736385345459, "learning_rate": 8.631800000000001e-06, "loss": 0.0621, "step": 6842 }, { "epoch": 0.36482379911499707, "grad_norm": 14.864871978759766, "learning_rate": 8.6316e-06, "loss": 0.0947, "step": 6843 }, { "epoch": 0.36487711254465, "grad_norm": 7.701630592346191, "learning_rate": 8.6314e-06, "loss": 0.0839, "step": 6844 }, { "epoch": 0.3649304259743029, "grad_norm": 2.404367208480835, "learning_rate": 8.631200000000001e-06, "loss": 0.0521, "step": 6845 }, { "epoch": 0.36498373940395584, "grad_norm": 3.1853511333465576, "learning_rate": 8.631000000000001e-06, "loss": 0.0317, "step": 6846 }, { "epoch": 0.36503705283360877, "grad_norm": 5.030832290649414, "learning_rate": 8.6308e-06, "loss": 0.0564, "step": 6847 }, { "epoch": 0.3650903662632617, "grad_norm": 7.836820602416992, "learning_rate": 8.630600000000002e-06, "loss": 0.0426, "step": 6848 }, { "epoch": 0.36514367969291467, "grad_norm": 2.643336772918701, "learning_rate": 8.630400000000001e-06, "loss": 0.0416, "step": 6849 }, { "epoch": 0.3651969931225676, "grad_norm": 4.23140287399292, "learning_rate": 8.6302e-06, "loss": 0.061, "step": 6850 }, { "epoch": 0.3652503065522205, "grad_norm": 4.426443099975586, "learning_rate": 8.63e-06, "loss": 0.0663, "step": 6851 }, { "epoch": 0.36530361998187344, "grad_norm": 2.9080045223236084, "learning_rate": 8.6298e-06, "loss": 0.0636, "step": 6852 }, { "epoch": 0.36535693341152636, "grad_norm": 2.3773396015167236, "learning_rate": 8.6296e-06, "loss": 0.0399, "step": 6853 }, { "epoch": 0.3654102468411793, "grad_norm": 5.228131294250488, "learning_rate": 8.6294e-06, "loss": 0.053, "step": 6854 }, { "epoch": 0.3654635602708322, "grad_norm": 4.516989707946777, "learning_rate": 8.6292e-06, "loss": 0.0493, "step": 6855 }, { "epoch": 0.36551687370048513, "grad_norm": 3.5624353885650635, "learning_rate": 8.629e-06, "loss": 0.0407, "step": 6856 }, { "epoch": 0.36557018713013806, "grad_norm": 3.684124231338501, "learning_rate": 8.628800000000001e-06, "loss": 0.0807, "step": 6857 }, { "epoch": 0.36562350055979104, "grad_norm": 3.891511917114258, "learning_rate": 8.6286e-06, "loss": 0.0742, "step": 6858 }, { "epoch": 0.36567681398944396, "grad_norm": 3.3879528045654297, "learning_rate": 8.628400000000002e-06, "loss": 0.0474, "step": 6859 }, { "epoch": 0.3657301274190969, "grad_norm": 1.7727580070495605, "learning_rate": 8.628200000000002e-06, "loss": 0.037, "step": 6860 }, { "epoch": 0.3657834408487498, "grad_norm": 2.3175888061523438, "learning_rate": 8.628000000000001e-06, "loss": 0.0411, "step": 6861 }, { "epoch": 0.36583675427840273, "grad_norm": 3.0885441303253174, "learning_rate": 8.6278e-06, "loss": 0.0486, "step": 6862 }, { "epoch": 0.36589006770805566, "grad_norm": 1.576698899269104, "learning_rate": 8.6276e-06, "loss": 0.0306, "step": 6863 }, { "epoch": 0.3659433811377086, "grad_norm": 3.446298599243164, "learning_rate": 8.6274e-06, "loss": 0.0617, "step": 6864 }, { "epoch": 0.3659966945673615, "grad_norm": 2.1356284618377686, "learning_rate": 8.627200000000001e-06, "loss": 0.0245, "step": 6865 }, { "epoch": 0.3660500079970144, "grad_norm": 2.8523049354553223, "learning_rate": 8.627e-06, "loss": 0.0415, "step": 6866 }, { "epoch": 0.3661033214266674, "grad_norm": 1.1132296323776245, "learning_rate": 8.6268e-06, "loss": 0.0107, "step": 6867 }, { "epoch": 0.36615663485632033, "grad_norm": 3.0523762702941895, "learning_rate": 8.6266e-06, "loss": 0.0437, "step": 6868 }, { "epoch": 0.36620994828597325, "grad_norm": 2.1244566440582275, "learning_rate": 8.626400000000001e-06, "loss": 0.034, "step": 6869 }, { "epoch": 0.3662632617156262, "grad_norm": 2.7099268436431885, "learning_rate": 8.6262e-06, "loss": 0.0242, "step": 6870 }, { "epoch": 0.3663165751452791, "grad_norm": 3.55061411857605, "learning_rate": 8.626000000000002e-06, "loss": 0.0956, "step": 6871 }, { "epoch": 0.366369888574932, "grad_norm": 3.0020077228546143, "learning_rate": 8.625800000000001e-06, "loss": 0.0453, "step": 6872 }, { "epoch": 0.36642320200458495, "grad_norm": 4.537899971008301, "learning_rate": 8.625600000000001e-06, "loss": 0.0571, "step": 6873 }, { "epoch": 0.36647651543423787, "grad_norm": 2.368844747543335, "learning_rate": 8.6254e-06, "loss": 0.0354, "step": 6874 }, { "epoch": 0.3665298288638908, "grad_norm": 2.3037450313568115, "learning_rate": 8.6252e-06, "loss": 0.0214, "step": 6875 }, { "epoch": 0.3665831422935437, "grad_norm": 2.742283582687378, "learning_rate": 8.625000000000001e-06, "loss": 0.0323, "step": 6876 }, { "epoch": 0.3666364557231967, "grad_norm": 5.021969795227051, "learning_rate": 8.6248e-06, "loss": 0.0567, "step": 6877 }, { "epoch": 0.3666897691528496, "grad_norm": 2.564528226852417, "learning_rate": 8.6246e-06, "loss": 0.0402, "step": 6878 }, { "epoch": 0.36674308258250254, "grad_norm": 3.8831787109375, "learning_rate": 8.6244e-06, "loss": 0.0729, "step": 6879 }, { "epoch": 0.36679639601215547, "grad_norm": 4.303605556488037, "learning_rate": 8.6242e-06, "loss": 0.0483, "step": 6880 }, { "epoch": 0.3668497094418084, "grad_norm": 4.697910308837891, "learning_rate": 8.624e-06, "loss": 0.0665, "step": 6881 }, { "epoch": 0.3669030228714613, "grad_norm": 4.405210018157959, "learning_rate": 8.6238e-06, "loss": 0.0715, "step": 6882 }, { "epoch": 0.36695633630111424, "grad_norm": 4.105605125427246, "learning_rate": 8.623600000000002e-06, "loss": 0.0696, "step": 6883 }, { "epoch": 0.36700964973076716, "grad_norm": 6.636093616485596, "learning_rate": 8.623400000000001e-06, "loss": 0.0789, "step": 6884 }, { "epoch": 0.3670629631604201, "grad_norm": 4.405509948730469, "learning_rate": 8.6232e-06, "loss": 0.0684, "step": 6885 }, { "epoch": 0.36711627659007307, "grad_norm": 4.325534343719482, "learning_rate": 8.623e-06, "loss": 0.0485, "step": 6886 }, { "epoch": 0.367169590019726, "grad_norm": 1.8751498460769653, "learning_rate": 8.622800000000002e-06, "loss": 0.0405, "step": 6887 }, { "epoch": 0.3672229034493789, "grad_norm": 3.041105031967163, "learning_rate": 8.622600000000001e-06, "loss": 0.0674, "step": 6888 }, { "epoch": 0.36727621687903184, "grad_norm": 3.818765878677368, "learning_rate": 8.6224e-06, "loss": 0.0373, "step": 6889 }, { "epoch": 0.36732953030868476, "grad_norm": 2.5210890769958496, "learning_rate": 8.6222e-06, "loss": 0.0221, "step": 6890 }, { "epoch": 0.3673828437383377, "grad_norm": 4.005397796630859, "learning_rate": 8.622e-06, "loss": 0.0326, "step": 6891 }, { "epoch": 0.3674361571679906, "grad_norm": 2.6946511268615723, "learning_rate": 8.6218e-06, "loss": 0.0329, "step": 6892 }, { "epoch": 0.36748947059764353, "grad_norm": 3.562349557876587, "learning_rate": 8.6216e-06, "loss": 0.0741, "step": 6893 }, { "epoch": 0.36754278402729645, "grad_norm": 2.1432440280914307, "learning_rate": 8.6214e-06, "loss": 0.0262, "step": 6894 }, { "epoch": 0.36759609745694943, "grad_norm": 2.5718579292297363, "learning_rate": 8.621200000000001e-06, "loss": 0.0411, "step": 6895 }, { "epoch": 0.36764941088660236, "grad_norm": 2.731694221496582, "learning_rate": 8.621000000000001e-06, "loss": 0.037, "step": 6896 }, { "epoch": 0.3677027243162553, "grad_norm": 2.251966953277588, "learning_rate": 8.6208e-06, "loss": 0.0368, "step": 6897 }, { "epoch": 0.3677560377459082, "grad_norm": 3.5099058151245117, "learning_rate": 8.620600000000002e-06, "loss": 0.0324, "step": 6898 }, { "epoch": 0.36780935117556113, "grad_norm": 4.37968111038208, "learning_rate": 8.620400000000001e-06, "loss": 0.0958, "step": 6899 }, { "epoch": 0.36786266460521405, "grad_norm": 2.361947774887085, "learning_rate": 8.620200000000001e-06, "loss": 0.0401, "step": 6900 }, { "epoch": 0.367915978034867, "grad_norm": 4.400248050689697, "learning_rate": 8.62e-06, "loss": 0.0675, "step": 6901 }, { "epoch": 0.3679692914645199, "grad_norm": 2.560459613800049, "learning_rate": 8.6198e-06, "loss": 0.0361, "step": 6902 }, { "epoch": 0.3680226048941728, "grad_norm": 3.153636932373047, "learning_rate": 8.6196e-06, "loss": 0.028, "step": 6903 }, { "epoch": 0.36807591832382575, "grad_norm": 2.1488471031188965, "learning_rate": 8.619400000000001e-06, "loss": 0.0319, "step": 6904 }, { "epoch": 0.3681292317534787, "grad_norm": 3.9237558841705322, "learning_rate": 8.6192e-06, "loss": 0.0343, "step": 6905 }, { "epoch": 0.36818254518313165, "grad_norm": 3.6344895362854004, "learning_rate": 8.619e-06, "loss": 0.0609, "step": 6906 }, { "epoch": 0.3682358586127846, "grad_norm": 1.8974186182022095, "learning_rate": 8.618800000000001e-06, "loss": 0.0258, "step": 6907 }, { "epoch": 0.3682891720424375, "grad_norm": 2.4148337841033936, "learning_rate": 8.6186e-06, "loss": 0.053, "step": 6908 }, { "epoch": 0.3683424854720904, "grad_norm": 4.017190456390381, "learning_rate": 8.618400000000002e-06, "loss": 0.06, "step": 6909 }, { "epoch": 0.36839579890174334, "grad_norm": 2.8453071117401123, "learning_rate": 8.618200000000002e-06, "loss": 0.0281, "step": 6910 }, { "epoch": 0.36844911233139627, "grad_norm": 2.295156240463257, "learning_rate": 8.618000000000001e-06, "loss": 0.0428, "step": 6911 }, { "epoch": 0.3685024257610492, "grad_norm": 3.567760705947876, "learning_rate": 8.6178e-06, "loss": 0.0794, "step": 6912 }, { "epoch": 0.3685557391907021, "grad_norm": 4.177418231964111, "learning_rate": 8.6176e-06, "loss": 0.0225, "step": 6913 }, { "epoch": 0.3686090526203551, "grad_norm": 3.3975350856781006, "learning_rate": 8.6174e-06, "loss": 0.0381, "step": 6914 }, { "epoch": 0.368662366050008, "grad_norm": 1.9072242975234985, "learning_rate": 8.617200000000001e-06, "loss": 0.0132, "step": 6915 }, { "epoch": 0.36871567947966094, "grad_norm": 5.073875904083252, "learning_rate": 8.617e-06, "loss": 0.0849, "step": 6916 }, { "epoch": 0.36876899290931386, "grad_norm": 3.1118152141571045, "learning_rate": 8.6168e-06, "loss": 0.0232, "step": 6917 }, { "epoch": 0.3688223063389668, "grad_norm": 4.706408977508545, "learning_rate": 8.6166e-06, "loss": 0.0783, "step": 6918 }, { "epoch": 0.3688756197686197, "grad_norm": 1.3218903541564941, "learning_rate": 8.616400000000001e-06, "loss": 0.0187, "step": 6919 }, { "epoch": 0.36892893319827263, "grad_norm": 3.137620449066162, "learning_rate": 8.6162e-06, "loss": 0.0425, "step": 6920 }, { "epoch": 0.36898224662792556, "grad_norm": 3.1421594619750977, "learning_rate": 8.616000000000002e-06, "loss": 0.0505, "step": 6921 }, { "epoch": 0.3690355600575785, "grad_norm": 2.5431201457977295, "learning_rate": 8.615800000000001e-06, "loss": 0.0398, "step": 6922 }, { "epoch": 0.3690888734872314, "grad_norm": 2.876039505004883, "learning_rate": 8.615600000000001e-06, "loss": 0.0209, "step": 6923 }, { "epoch": 0.3691421869168844, "grad_norm": 4.051978588104248, "learning_rate": 8.6154e-06, "loss": 0.0416, "step": 6924 }, { "epoch": 0.3691955003465373, "grad_norm": 5.601072311401367, "learning_rate": 8.6152e-06, "loss": 0.0345, "step": 6925 }, { "epoch": 0.36924881377619023, "grad_norm": 4.455698013305664, "learning_rate": 8.615000000000001e-06, "loss": 0.0347, "step": 6926 }, { "epoch": 0.36930212720584316, "grad_norm": 4.266068935394287, "learning_rate": 8.614800000000001e-06, "loss": 0.076, "step": 6927 }, { "epoch": 0.3693554406354961, "grad_norm": 1.6051918268203735, "learning_rate": 8.6146e-06, "loss": 0.0185, "step": 6928 }, { "epoch": 0.369408754065149, "grad_norm": 7.498939037322998, "learning_rate": 8.6144e-06, "loss": 0.0816, "step": 6929 }, { "epoch": 0.3694620674948019, "grad_norm": 4.182642459869385, "learning_rate": 8.6142e-06, "loss": 0.0656, "step": 6930 }, { "epoch": 0.36951538092445485, "grad_norm": 2.288569927215576, "learning_rate": 8.614000000000001e-06, "loss": 0.0228, "step": 6931 }, { "epoch": 0.3695686943541078, "grad_norm": 4.085175037384033, "learning_rate": 8.6138e-06, "loss": 0.059, "step": 6932 }, { "epoch": 0.36962200778376075, "grad_norm": 2.8147456645965576, "learning_rate": 8.613600000000002e-06, "loss": 0.0197, "step": 6933 }, { "epoch": 0.3696753212134137, "grad_norm": 5.082434177398682, "learning_rate": 8.613400000000001e-06, "loss": 0.0576, "step": 6934 }, { "epoch": 0.3697286346430666, "grad_norm": 1.4683120250701904, "learning_rate": 8.6132e-06, "loss": 0.0192, "step": 6935 }, { "epoch": 0.3697819480727195, "grad_norm": 4.469171047210693, "learning_rate": 8.613e-06, "loss": 0.036, "step": 6936 }, { "epoch": 0.36983526150237245, "grad_norm": 2.582608938217163, "learning_rate": 8.612800000000002e-06, "loss": 0.0478, "step": 6937 }, { "epoch": 0.36988857493202537, "grad_norm": 4.223018169403076, "learning_rate": 8.612600000000001e-06, "loss": 0.0592, "step": 6938 }, { "epoch": 0.3699418883616783, "grad_norm": 3.379281520843506, "learning_rate": 8.6124e-06, "loss": 0.0575, "step": 6939 }, { "epoch": 0.3699952017913312, "grad_norm": 2.953181743621826, "learning_rate": 8.6122e-06, "loss": 0.0283, "step": 6940 }, { "epoch": 0.37004851522098414, "grad_norm": 1.7241853475570679, "learning_rate": 8.612e-06, "loss": 0.0295, "step": 6941 }, { "epoch": 0.3701018286506371, "grad_norm": 4.5094895362854, "learning_rate": 8.6118e-06, "loss": 0.1265, "step": 6942 }, { "epoch": 0.37015514208029004, "grad_norm": 2.3662219047546387, "learning_rate": 8.6116e-06, "loss": 0.0261, "step": 6943 }, { "epoch": 0.37020845550994297, "grad_norm": 1.8264882564544678, "learning_rate": 8.6114e-06, "loss": 0.0326, "step": 6944 }, { "epoch": 0.3702617689395959, "grad_norm": 1.4750304222106934, "learning_rate": 8.611200000000002e-06, "loss": 0.0127, "step": 6945 }, { "epoch": 0.3703150823692488, "grad_norm": 4.030642986297607, "learning_rate": 8.611000000000001e-06, "loss": 0.0513, "step": 6946 }, { "epoch": 0.37036839579890174, "grad_norm": 3.1874420642852783, "learning_rate": 8.6108e-06, "loss": 0.0639, "step": 6947 }, { "epoch": 0.37042170922855466, "grad_norm": 2.571293830871582, "learning_rate": 8.610600000000002e-06, "loss": 0.0557, "step": 6948 }, { "epoch": 0.3704750226582076, "grad_norm": 2.709371566772461, "learning_rate": 8.610400000000001e-06, "loss": 0.0287, "step": 6949 }, { "epoch": 0.3705283360878605, "grad_norm": 2.035402774810791, "learning_rate": 8.610200000000001e-06, "loss": 0.024, "step": 6950 }, { "epoch": 0.37058164951751343, "grad_norm": 2.0018296241760254, "learning_rate": 8.61e-06, "loss": 0.0193, "step": 6951 }, { "epoch": 0.3706349629471664, "grad_norm": 2.920781373977661, "learning_rate": 8.6098e-06, "loss": 0.0384, "step": 6952 }, { "epoch": 0.37068827637681934, "grad_norm": 1.4102864265441895, "learning_rate": 8.6096e-06, "loss": 0.0218, "step": 6953 }, { "epoch": 0.37074158980647226, "grad_norm": 2.97605037689209, "learning_rate": 8.609400000000001e-06, "loss": 0.057, "step": 6954 }, { "epoch": 0.3707949032361252, "grad_norm": 2.6448614597320557, "learning_rate": 8.6092e-06, "loss": 0.0206, "step": 6955 }, { "epoch": 0.3708482166657781, "grad_norm": 2.8251497745513916, "learning_rate": 8.609e-06, "loss": 0.0362, "step": 6956 }, { "epoch": 0.37090153009543103, "grad_norm": 4.031785011291504, "learning_rate": 8.608800000000001e-06, "loss": 0.0325, "step": 6957 }, { "epoch": 0.37095484352508395, "grad_norm": 2.7427971363067627, "learning_rate": 8.608600000000001e-06, "loss": 0.0504, "step": 6958 }, { "epoch": 0.3710081569547369, "grad_norm": 2.4816224575042725, "learning_rate": 8.6084e-06, "loss": 0.0492, "step": 6959 }, { "epoch": 0.3710614703843898, "grad_norm": 3.8656952381134033, "learning_rate": 8.608200000000002e-06, "loss": 0.0297, "step": 6960 }, { "epoch": 0.3711147838140428, "grad_norm": 4.85365104675293, "learning_rate": 8.608000000000001e-06, "loss": 0.0249, "step": 6961 }, { "epoch": 0.3711680972436957, "grad_norm": 2.8984434604644775, "learning_rate": 8.6078e-06, "loss": 0.0286, "step": 6962 }, { "epoch": 0.37122141067334863, "grad_norm": 4.1534743309021, "learning_rate": 8.6076e-06, "loss": 0.031, "step": 6963 }, { "epoch": 0.37127472410300155, "grad_norm": 1.4018954038619995, "learning_rate": 8.6074e-06, "loss": 0.0126, "step": 6964 }, { "epoch": 0.3713280375326545, "grad_norm": 3.0572211742401123, "learning_rate": 8.607200000000001e-06, "loss": 0.0727, "step": 6965 }, { "epoch": 0.3713813509623074, "grad_norm": 2.0525097846984863, "learning_rate": 8.607e-06, "loss": 0.0219, "step": 6966 }, { "epoch": 0.3714346643919603, "grad_norm": 1.66464102268219, "learning_rate": 8.6068e-06, "loss": 0.0331, "step": 6967 }, { "epoch": 0.37148797782161325, "grad_norm": 3.215674877166748, "learning_rate": 8.6066e-06, "loss": 0.0244, "step": 6968 }, { "epoch": 0.37154129125126617, "grad_norm": 1.3008793592453003, "learning_rate": 8.606400000000001e-06, "loss": 0.0098, "step": 6969 }, { "epoch": 0.37159460468091915, "grad_norm": 2.011746883392334, "learning_rate": 8.6062e-06, "loss": 0.0377, "step": 6970 }, { "epoch": 0.3716479181105721, "grad_norm": 3.5920724868774414, "learning_rate": 8.606e-06, "loss": 0.0336, "step": 6971 }, { "epoch": 0.371701231540225, "grad_norm": 4.890716075897217, "learning_rate": 8.605800000000002e-06, "loss": 0.0328, "step": 6972 }, { "epoch": 0.3717545449698779, "grad_norm": 4.11118221282959, "learning_rate": 8.605600000000001e-06, "loss": 0.0407, "step": 6973 }, { "epoch": 0.37180785839953084, "grad_norm": 1.7090318202972412, "learning_rate": 8.6054e-06, "loss": 0.0183, "step": 6974 }, { "epoch": 0.37186117182918377, "grad_norm": 2.6835217475891113, "learning_rate": 8.6052e-06, "loss": 0.0188, "step": 6975 }, { "epoch": 0.3719144852588367, "grad_norm": 3.0919950008392334, "learning_rate": 8.605000000000001e-06, "loss": 0.0279, "step": 6976 }, { "epoch": 0.3719677986884896, "grad_norm": 2.7113819122314453, "learning_rate": 8.604800000000001e-06, "loss": 0.0285, "step": 6977 }, { "epoch": 0.37202111211814254, "grad_norm": 3.421149253845215, "learning_rate": 8.6046e-06, "loss": 0.0428, "step": 6978 }, { "epoch": 0.37207442554779546, "grad_norm": 3.4113564491271973, "learning_rate": 8.6044e-06, "loss": 0.0529, "step": 6979 }, { "epoch": 0.37212773897744844, "grad_norm": 2.226229190826416, "learning_rate": 8.6042e-06, "loss": 0.0394, "step": 6980 }, { "epoch": 0.37218105240710136, "grad_norm": 4.790716648101807, "learning_rate": 8.604000000000001e-06, "loss": 0.045, "step": 6981 }, { "epoch": 0.3722343658367543, "grad_norm": 3.1026418209075928, "learning_rate": 8.6038e-06, "loss": 0.0342, "step": 6982 }, { "epoch": 0.3722876792664072, "grad_norm": 2.5855562686920166, "learning_rate": 8.6036e-06, "loss": 0.0228, "step": 6983 }, { "epoch": 0.37234099269606014, "grad_norm": 1.6016572713851929, "learning_rate": 8.603400000000001e-06, "loss": 0.0212, "step": 6984 }, { "epoch": 0.37239430612571306, "grad_norm": 3.4963574409484863, "learning_rate": 8.603200000000001e-06, "loss": 0.0613, "step": 6985 }, { "epoch": 0.372447619555366, "grad_norm": 3.3451054096221924, "learning_rate": 8.603e-06, "loss": 0.0379, "step": 6986 }, { "epoch": 0.3725009329850189, "grad_norm": 5.0933966636657715, "learning_rate": 8.602800000000002e-06, "loss": 0.0467, "step": 6987 }, { "epoch": 0.37255424641467183, "grad_norm": 3.1767497062683105, "learning_rate": 8.602600000000001e-06, "loss": 0.0375, "step": 6988 }, { "epoch": 0.3726075598443248, "grad_norm": 2.9323530197143555, "learning_rate": 8.602400000000001e-06, "loss": 0.0316, "step": 6989 }, { "epoch": 0.37266087327397773, "grad_norm": 1.9867768287658691, "learning_rate": 8.6022e-06, "loss": 0.0326, "step": 6990 }, { "epoch": 0.37271418670363066, "grad_norm": 4.152577877044678, "learning_rate": 8.602e-06, "loss": 0.0319, "step": 6991 }, { "epoch": 0.3727675001332836, "grad_norm": 3.371826171875, "learning_rate": 8.6018e-06, "loss": 0.0348, "step": 6992 }, { "epoch": 0.3728208135629365, "grad_norm": 3.5882551670074463, "learning_rate": 8.6016e-06, "loss": 0.0537, "step": 6993 }, { "epoch": 0.3728741269925894, "grad_norm": 3.12793231010437, "learning_rate": 8.6014e-06, "loss": 0.0361, "step": 6994 }, { "epoch": 0.37292744042224235, "grad_norm": 2.760664463043213, "learning_rate": 8.6012e-06, "loss": 0.0432, "step": 6995 }, { "epoch": 0.3729807538518953, "grad_norm": 4.2254180908203125, "learning_rate": 8.601000000000001e-06, "loss": 0.0394, "step": 6996 }, { "epoch": 0.3730340672815482, "grad_norm": 2.6062660217285156, "learning_rate": 8.6008e-06, "loss": 0.0396, "step": 6997 }, { "epoch": 0.3730873807112012, "grad_norm": 2.48371958732605, "learning_rate": 8.600600000000002e-06, "loss": 0.0281, "step": 6998 }, { "epoch": 0.3731406941408541, "grad_norm": 4.493300914764404, "learning_rate": 8.600400000000002e-06, "loss": 0.0552, "step": 6999 }, { "epoch": 0.373194007570507, "grad_norm": 2.6021947860717773, "learning_rate": 8.600200000000001e-06, "loss": 0.0306, "step": 7000 }, { "epoch": 0.37324732100015995, "grad_norm": 5.604491710662842, "learning_rate": 8.6e-06, "loss": 0.0725, "step": 7001 }, { "epoch": 0.37330063442981287, "grad_norm": 8.265474319458008, "learning_rate": 8.5998e-06, "loss": 0.0519, "step": 7002 }, { "epoch": 0.3733539478594658, "grad_norm": 3.7126708030700684, "learning_rate": 8.5996e-06, "loss": 0.0689, "step": 7003 }, { "epoch": 0.3734072612891187, "grad_norm": 2.34246826171875, "learning_rate": 8.599400000000001e-06, "loss": 0.0211, "step": 7004 }, { "epoch": 0.37346057471877164, "grad_norm": 3.181763172149658, "learning_rate": 8.5992e-06, "loss": 0.0275, "step": 7005 }, { "epoch": 0.37351388814842457, "grad_norm": 1.4401123523712158, "learning_rate": 8.599e-06, "loss": 0.0353, "step": 7006 }, { "epoch": 0.3735672015780775, "grad_norm": 2.161611318588257, "learning_rate": 8.5988e-06, "loss": 0.0544, "step": 7007 }, { "epoch": 0.37362051500773047, "grad_norm": 4.271481037139893, "learning_rate": 8.598600000000001e-06, "loss": 0.0867, "step": 7008 }, { "epoch": 0.3736738284373834, "grad_norm": 2.7557756900787354, "learning_rate": 8.5984e-06, "loss": 0.0391, "step": 7009 }, { "epoch": 0.3737271418670363, "grad_norm": 2.221001386642456, "learning_rate": 8.598200000000002e-06, "loss": 0.0742, "step": 7010 }, { "epoch": 0.37378045529668924, "grad_norm": 2.3793487548828125, "learning_rate": 8.598000000000001e-06, "loss": 0.0397, "step": 7011 }, { "epoch": 0.37383376872634216, "grad_norm": 1.686620831489563, "learning_rate": 8.597800000000001e-06, "loss": 0.0345, "step": 7012 }, { "epoch": 0.3738870821559951, "grad_norm": 2.256281852722168, "learning_rate": 8.5976e-06, "loss": 0.0345, "step": 7013 }, { "epoch": 0.373940395585648, "grad_norm": 2.099377393722534, "learning_rate": 8.5974e-06, "loss": 0.0343, "step": 7014 }, { "epoch": 0.37399370901530093, "grad_norm": 4.110326766967773, "learning_rate": 8.597200000000001e-06, "loss": 0.0689, "step": 7015 }, { "epoch": 0.37404702244495386, "grad_norm": 2.5505690574645996, "learning_rate": 8.597000000000001e-06, "loss": 0.0383, "step": 7016 }, { "epoch": 0.37410033587460684, "grad_norm": 3.939802408218384, "learning_rate": 8.5968e-06, "loss": 0.0543, "step": 7017 }, { "epoch": 0.37415364930425976, "grad_norm": 2.853482961654663, "learning_rate": 8.5966e-06, "loss": 0.0594, "step": 7018 }, { "epoch": 0.3742069627339127, "grad_norm": 3.5353474617004395, "learning_rate": 8.5964e-06, "loss": 0.0566, "step": 7019 }, { "epoch": 0.3742602761635656, "grad_norm": 2.515500545501709, "learning_rate": 8.5962e-06, "loss": 0.0543, "step": 7020 }, { "epoch": 0.37431358959321853, "grad_norm": 2.4769890308380127, "learning_rate": 8.596e-06, "loss": 0.0288, "step": 7021 }, { "epoch": 0.37436690302287146, "grad_norm": 3.4239020347595215, "learning_rate": 8.595800000000002e-06, "loss": 0.0453, "step": 7022 }, { "epoch": 0.3744202164525244, "grad_norm": 2.576237916946411, "learning_rate": 8.595600000000001e-06, "loss": 0.0159, "step": 7023 }, { "epoch": 0.3744735298821773, "grad_norm": 2.9739773273468018, "learning_rate": 8.5954e-06, "loss": 0.0312, "step": 7024 }, { "epoch": 0.3745268433118302, "grad_norm": 1.8798624277114868, "learning_rate": 8.5952e-06, "loss": 0.0345, "step": 7025 }, { "epoch": 0.3745801567414832, "grad_norm": 2.03987717628479, "learning_rate": 8.595000000000002e-06, "loss": 0.0383, "step": 7026 }, { "epoch": 0.37463347017113613, "grad_norm": 2.747591972351074, "learning_rate": 8.594800000000001e-06, "loss": 0.025, "step": 7027 }, { "epoch": 0.37468678360078905, "grad_norm": 2.0039725303649902, "learning_rate": 8.5946e-06, "loss": 0.0383, "step": 7028 }, { "epoch": 0.374740097030442, "grad_norm": 4.604848861694336, "learning_rate": 8.5944e-06, "loss": 0.0564, "step": 7029 }, { "epoch": 0.3747934104600949, "grad_norm": 1.9376510381698608, "learning_rate": 8.5942e-06, "loss": 0.0197, "step": 7030 }, { "epoch": 0.3748467238897478, "grad_norm": 2.3142480850219727, "learning_rate": 8.594000000000001e-06, "loss": 0.0243, "step": 7031 }, { "epoch": 0.37490003731940075, "grad_norm": 1.9095630645751953, "learning_rate": 8.5938e-06, "loss": 0.0299, "step": 7032 }, { "epoch": 0.37495335074905367, "grad_norm": 1.900635004043579, "learning_rate": 8.5936e-06, "loss": 0.0172, "step": 7033 }, { "epoch": 0.3750066641787066, "grad_norm": 5.978532791137695, "learning_rate": 8.593400000000001e-06, "loss": 0.0681, "step": 7034 }, { "epoch": 0.3750599776083595, "grad_norm": 3.941420793533325, "learning_rate": 8.593200000000001e-06, "loss": 0.065, "step": 7035 }, { "epoch": 0.3751132910380125, "grad_norm": 1.9007444381713867, "learning_rate": 8.593e-06, "loss": 0.0284, "step": 7036 }, { "epoch": 0.3751666044676654, "grad_norm": 2.9524312019348145, "learning_rate": 8.592800000000002e-06, "loss": 0.0355, "step": 7037 }, { "epoch": 0.37521991789731834, "grad_norm": 5.096805572509766, "learning_rate": 8.592600000000001e-06, "loss": 0.0525, "step": 7038 }, { "epoch": 0.37527323132697127, "grad_norm": 4.255024433135986, "learning_rate": 8.592400000000001e-06, "loss": 0.0398, "step": 7039 }, { "epoch": 0.3753265447566242, "grad_norm": 3.1195948123931885, "learning_rate": 8.5922e-06, "loss": 0.0427, "step": 7040 }, { "epoch": 0.3753798581862771, "grad_norm": 3.1891980171203613, "learning_rate": 8.592e-06, "loss": 0.0491, "step": 7041 }, { "epoch": 0.37543317161593004, "grad_norm": 4.128945350646973, "learning_rate": 8.591800000000001e-06, "loss": 0.0714, "step": 7042 }, { "epoch": 0.37548648504558296, "grad_norm": 2.2198269367218018, "learning_rate": 8.591600000000001e-06, "loss": 0.0155, "step": 7043 }, { "epoch": 0.3755397984752359, "grad_norm": 2.1000280380249023, "learning_rate": 8.5914e-06, "loss": 0.0269, "step": 7044 }, { "epoch": 0.37559311190488887, "grad_norm": 3.4035606384277344, "learning_rate": 8.5912e-06, "loss": 0.0371, "step": 7045 }, { "epoch": 0.3756464253345418, "grad_norm": 2.223257303237915, "learning_rate": 8.591000000000001e-06, "loss": 0.0188, "step": 7046 }, { "epoch": 0.3756997387641947, "grad_norm": 7.833162307739258, "learning_rate": 8.5908e-06, "loss": 0.0734, "step": 7047 }, { "epoch": 0.37575305219384764, "grad_norm": 3.082984685897827, "learning_rate": 8.5906e-06, "loss": 0.0436, "step": 7048 }, { "epoch": 0.37580636562350056, "grad_norm": 1.0619221925735474, "learning_rate": 8.590400000000002e-06, "loss": 0.0257, "step": 7049 }, { "epoch": 0.3758596790531535, "grad_norm": 2.6424314975738525, "learning_rate": 8.590200000000001e-06, "loss": 0.0534, "step": 7050 }, { "epoch": 0.3759129924828064, "grad_norm": 4.183761119842529, "learning_rate": 8.59e-06, "loss": 0.0281, "step": 7051 }, { "epoch": 0.37596630591245933, "grad_norm": 6.8253397941589355, "learning_rate": 8.5898e-06, "loss": 0.0566, "step": 7052 }, { "epoch": 0.37601961934211225, "grad_norm": 4.5217719078063965, "learning_rate": 8.5896e-06, "loss": 0.0334, "step": 7053 }, { "epoch": 0.37607293277176523, "grad_norm": 3.253451108932495, "learning_rate": 8.589400000000001e-06, "loss": 0.0518, "step": 7054 }, { "epoch": 0.37612624620141816, "grad_norm": 5.180943965911865, "learning_rate": 8.5892e-06, "loss": 0.0571, "step": 7055 }, { "epoch": 0.3761795596310711, "grad_norm": 1.968234658241272, "learning_rate": 8.589e-06, "loss": 0.0375, "step": 7056 }, { "epoch": 0.376232873060724, "grad_norm": 2.21327543258667, "learning_rate": 8.5888e-06, "loss": 0.0179, "step": 7057 }, { "epoch": 0.37628618649037693, "grad_norm": 3.0536701679229736, "learning_rate": 8.588600000000001e-06, "loss": 0.0514, "step": 7058 }, { "epoch": 0.37633949992002985, "grad_norm": 4.408967018127441, "learning_rate": 8.5884e-06, "loss": 0.0865, "step": 7059 }, { "epoch": 0.3763928133496828, "grad_norm": 1.9035731554031372, "learning_rate": 8.5882e-06, "loss": 0.0213, "step": 7060 }, { "epoch": 0.3764461267793357, "grad_norm": 3.374596357345581, "learning_rate": 8.588000000000001e-06, "loss": 0.0382, "step": 7061 }, { "epoch": 0.3764994402089886, "grad_norm": 3.4807682037353516, "learning_rate": 8.587800000000001e-06, "loss": 0.0524, "step": 7062 }, { "epoch": 0.37655275363864155, "grad_norm": 3.7571630477905273, "learning_rate": 8.5876e-06, "loss": 0.0674, "step": 7063 }, { "epoch": 0.3766060670682945, "grad_norm": 2.2310001850128174, "learning_rate": 8.5874e-06, "loss": 0.0494, "step": 7064 }, { "epoch": 0.37665938049794745, "grad_norm": 7.283260345458984, "learning_rate": 8.587200000000001e-06, "loss": 0.0626, "step": 7065 }, { "epoch": 0.3767126939276004, "grad_norm": 3.5256786346435547, "learning_rate": 8.587000000000001e-06, "loss": 0.0417, "step": 7066 }, { "epoch": 0.3767660073572533, "grad_norm": 5.074280738830566, "learning_rate": 8.5868e-06, "loss": 0.0433, "step": 7067 }, { "epoch": 0.3768193207869062, "grad_norm": 3.886103630065918, "learning_rate": 8.5866e-06, "loss": 0.0369, "step": 7068 }, { "epoch": 0.37687263421655914, "grad_norm": 1.9382590055465698, "learning_rate": 8.5864e-06, "loss": 0.0244, "step": 7069 }, { "epoch": 0.37692594764621207, "grad_norm": 4.2046356201171875, "learning_rate": 8.586200000000001e-06, "loss": 0.0547, "step": 7070 }, { "epoch": 0.376979261075865, "grad_norm": 3.4411847591400146, "learning_rate": 8.586e-06, "loss": 0.0349, "step": 7071 }, { "epoch": 0.3770325745055179, "grad_norm": 2.3756141662597656, "learning_rate": 8.5858e-06, "loss": 0.0345, "step": 7072 }, { "epoch": 0.3770858879351709, "grad_norm": 3.1059134006500244, "learning_rate": 8.585600000000001e-06, "loss": 0.0636, "step": 7073 }, { "epoch": 0.3771392013648238, "grad_norm": 3.577484130859375, "learning_rate": 8.5854e-06, "loss": 0.049, "step": 7074 }, { "epoch": 0.37719251479447674, "grad_norm": 2.961815357208252, "learning_rate": 8.5852e-06, "loss": 0.0494, "step": 7075 }, { "epoch": 0.37724582822412966, "grad_norm": 2.7523889541625977, "learning_rate": 8.585000000000002e-06, "loss": 0.0354, "step": 7076 }, { "epoch": 0.3772991416537826, "grad_norm": 5.307783603668213, "learning_rate": 8.584800000000001e-06, "loss": 0.0769, "step": 7077 }, { "epoch": 0.3773524550834355, "grad_norm": 2.395975351333618, "learning_rate": 8.5846e-06, "loss": 0.0279, "step": 7078 }, { "epoch": 0.37740576851308844, "grad_norm": 3.401130437850952, "learning_rate": 8.5844e-06, "loss": 0.0562, "step": 7079 }, { "epoch": 0.37745908194274136, "grad_norm": 5.232913017272949, "learning_rate": 8.5842e-06, "loss": 0.0453, "step": 7080 }, { "epoch": 0.3775123953723943, "grad_norm": 2.0944812297821045, "learning_rate": 8.584000000000001e-06, "loss": 0.0202, "step": 7081 }, { "epoch": 0.37756570880204726, "grad_norm": 2.9423205852508545, "learning_rate": 8.5838e-06, "loss": 0.0326, "step": 7082 }, { "epoch": 0.3776190222317002, "grad_norm": 4.562292098999023, "learning_rate": 8.5836e-06, "loss": 0.0616, "step": 7083 }, { "epoch": 0.3776723356613531, "grad_norm": 4.0649871826171875, "learning_rate": 8.5834e-06, "loss": 0.0357, "step": 7084 }, { "epoch": 0.37772564909100603, "grad_norm": 2.372121572494507, "learning_rate": 8.583200000000001e-06, "loss": 0.0465, "step": 7085 }, { "epoch": 0.37777896252065896, "grad_norm": 2.291922092437744, "learning_rate": 8.583e-06, "loss": 0.0532, "step": 7086 }, { "epoch": 0.3778322759503119, "grad_norm": 2.4235870838165283, "learning_rate": 8.582800000000002e-06, "loss": 0.032, "step": 7087 }, { "epoch": 0.3778855893799648, "grad_norm": 2.526122570037842, "learning_rate": 8.582600000000001e-06, "loss": 0.033, "step": 7088 }, { "epoch": 0.3779389028096177, "grad_norm": 4.559231281280518, "learning_rate": 8.582400000000001e-06, "loss": 0.0534, "step": 7089 }, { "epoch": 0.37799221623927065, "grad_norm": 2.5436086654663086, "learning_rate": 8.5822e-06, "loss": 0.0282, "step": 7090 }, { "epoch": 0.3780455296689236, "grad_norm": 3.3957276344299316, "learning_rate": 8.582e-06, "loss": 0.0351, "step": 7091 }, { "epoch": 0.37809884309857655, "grad_norm": 3.908795118331909, "learning_rate": 8.581800000000001e-06, "loss": 0.0395, "step": 7092 }, { "epoch": 0.3781521565282295, "grad_norm": 3.4973514080047607, "learning_rate": 8.581600000000001e-06, "loss": 0.0354, "step": 7093 }, { "epoch": 0.3782054699578824, "grad_norm": 3.4214515686035156, "learning_rate": 8.5814e-06, "loss": 0.0361, "step": 7094 }, { "epoch": 0.3782587833875353, "grad_norm": 8.063328742980957, "learning_rate": 8.5812e-06, "loss": 0.0678, "step": 7095 }, { "epoch": 0.37831209681718825, "grad_norm": 2.603252649307251, "learning_rate": 8.581e-06, "loss": 0.0332, "step": 7096 }, { "epoch": 0.37836541024684117, "grad_norm": 3.785918951034546, "learning_rate": 8.580800000000001e-06, "loss": 0.0437, "step": 7097 }, { "epoch": 0.3784187236764941, "grad_norm": 4.45440673828125, "learning_rate": 8.5806e-06, "loss": 0.0522, "step": 7098 }, { "epoch": 0.378472037106147, "grad_norm": 5.055835723876953, "learning_rate": 8.580400000000002e-06, "loss": 0.0488, "step": 7099 }, { "epoch": 0.37852535053579994, "grad_norm": 4.988241672515869, "learning_rate": 8.580200000000001e-06, "loss": 0.0617, "step": 7100 }, { "epoch": 0.3785786639654529, "grad_norm": 4.914175033569336, "learning_rate": 8.580000000000001e-06, "loss": 0.0614, "step": 7101 }, { "epoch": 0.37863197739510585, "grad_norm": 2.070152759552002, "learning_rate": 8.5798e-06, "loss": 0.0286, "step": 7102 }, { "epoch": 0.37868529082475877, "grad_norm": 4.91052770614624, "learning_rate": 8.5796e-06, "loss": 0.0915, "step": 7103 }, { "epoch": 0.3787386042544117, "grad_norm": 4.123298168182373, "learning_rate": 8.579400000000001e-06, "loss": 0.0532, "step": 7104 }, { "epoch": 0.3787919176840646, "grad_norm": 3.861253499984741, "learning_rate": 8.5792e-06, "loss": 0.055, "step": 7105 }, { "epoch": 0.37884523111371754, "grad_norm": 3.663029193878174, "learning_rate": 8.579e-06, "loss": 0.0578, "step": 7106 }, { "epoch": 0.37889854454337046, "grad_norm": 4.933523178100586, "learning_rate": 8.5788e-06, "loss": 0.0334, "step": 7107 }, { "epoch": 0.3789518579730234, "grad_norm": 7.49627685546875, "learning_rate": 8.5786e-06, "loss": 0.0967, "step": 7108 }, { "epoch": 0.3790051714026763, "grad_norm": 2.8953967094421387, "learning_rate": 8.5784e-06, "loss": 0.0242, "step": 7109 }, { "epoch": 0.3790584848323293, "grad_norm": 3.8702664375305176, "learning_rate": 8.5782e-06, "loss": 0.0231, "step": 7110 }, { "epoch": 0.3791117982619822, "grad_norm": 4.855426788330078, "learning_rate": 8.578000000000002e-06, "loss": 0.0592, "step": 7111 }, { "epoch": 0.37916511169163514, "grad_norm": 2.7374958992004395, "learning_rate": 8.577800000000001e-06, "loss": 0.0332, "step": 7112 }, { "epoch": 0.37921842512128806, "grad_norm": 4.225501537322998, "learning_rate": 8.5776e-06, "loss": 0.1008, "step": 7113 }, { "epoch": 0.379271738550941, "grad_norm": 6.944490909576416, "learning_rate": 8.5774e-06, "loss": 0.1543, "step": 7114 }, { "epoch": 0.3793250519805939, "grad_norm": 4.5680251121521, "learning_rate": 8.577200000000001e-06, "loss": 0.0487, "step": 7115 }, { "epoch": 0.37937836541024683, "grad_norm": 4.078990459442139, "learning_rate": 8.577000000000001e-06, "loss": 0.0801, "step": 7116 }, { "epoch": 0.37943167883989976, "grad_norm": 2.258683919906616, "learning_rate": 8.5768e-06, "loss": 0.0416, "step": 7117 }, { "epoch": 0.3794849922695527, "grad_norm": 1.8910026550292969, "learning_rate": 8.5766e-06, "loss": 0.0364, "step": 7118 }, { "epoch": 0.3795383056992056, "grad_norm": 5.495851039886475, "learning_rate": 8.5764e-06, "loss": 0.0454, "step": 7119 }, { "epoch": 0.3795916191288586, "grad_norm": 11.081949234008789, "learning_rate": 8.576200000000001e-06, "loss": 0.0738, "step": 7120 }, { "epoch": 0.3796449325585115, "grad_norm": 5.409010410308838, "learning_rate": 8.576e-06, "loss": 0.0455, "step": 7121 }, { "epoch": 0.37969824598816443, "grad_norm": 4.245100021362305, "learning_rate": 8.5758e-06, "loss": 0.0324, "step": 7122 }, { "epoch": 0.37975155941781735, "grad_norm": 2.669065237045288, "learning_rate": 8.575600000000001e-06, "loss": 0.046, "step": 7123 }, { "epoch": 0.3798048728474703, "grad_norm": 2.015395402908325, "learning_rate": 8.575400000000001e-06, "loss": 0.0374, "step": 7124 }, { "epoch": 0.3798581862771232, "grad_norm": 4.40616512298584, "learning_rate": 8.5752e-06, "loss": 0.0667, "step": 7125 }, { "epoch": 0.3799114997067761, "grad_norm": 5.239363193511963, "learning_rate": 8.575000000000002e-06, "loss": 0.1026, "step": 7126 }, { "epoch": 0.37996481313642905, "grad_norm": 3.907318115234375, "learning_rate": 8.574800000000001e-06, "loss": 0.0648, "step": 7127 }, { "epoch": 0.38001812656608197, "grad_norm": 1.8430392742156982, "learning_rate": 8.574600000000001e-06, "loss": 0.0298, "step": 7128 }, { "epoch": 0.38007143999573495, "grad_norm": 1.3714903593063354, "learning_rate": 8.5744e-06, "loss": 0.0345, "step": 7129 }, { "epoch": 0.3801247534253879, "grad_norm": 1.6412410736083984, "learning_rate": 8.5742e-06, "loss": 0.0221, "step": 7130 }, { "epoch": 0.3801780668550408, "grad_norm": 3.555041790008545, "learning_rate": 8.574000000000001e-06, "loss": 0.0342, "step": 7131 }, { "epoch": 0.3802313802846937, "grad_norm": 2.1391735076904297, "learning_rate": 8.5738e-06, "loss": 0.0397, "step": 7132 }, { "epoch": 0.38028469371434664, "grad_norm": 1.8358798027038574, "learning_rate": 8.5736e-06, "loss": 0.0373, "step": 7133 }, { "epoch": 0.38033800714399957, "grad_norm": 3.975062847137451, "learning_rate": 8.5734e-06, "loss": 0.0602, "step": 7134 }, { "epoch": 0.3803913205736525, "grad_norm": 2.4527506828308105, "learning_rate": 8.573200000000001e-06, "loss": 0.0312, "step": 7135 }, { "epoch": 0.3804446340033054, "grad_norm": 7.448643684387207, "learning_rate": 8.573e-06, "loss": 0.0562, "step": 7136 }, { "epoch": 0.38049794743295834, "grad_norm": 3.418766975402832, "learning_rate": 8.5728e-06, "loss": 0.0569, "step": 7137 }, { "epoch": 0.3805512608626113, "grad_norm": 4.1254777908325195, "learning_rate": 8.572600000000002e-06, "loss": 0.0517, "step": 7138 }, { "epoch": 0.38060457429226424, "grad_norm": 2.3881349563598633, "learning_rate": 8.572400000000001e-06, "loss": 0.0312, "step": 7139 }, { "epoch": 0.38065788772191717, "grad_norm": 4.930458068847656, "learning_rate": 8.5722e-06, "loss": 0.0255, "step": 7140 }, { "epoch": 0.3807112011515701, "grad_norm": 3.3108773231506348, "learning_rate": 8.572e-06, "loss": 0.0465, "step": 7141 }, { "epoch": 0.380764514581223, "grad_norm": 3.5833492279052734, "learning_rate": 8.571800000000002e-06, "loss": 0.0307, "step": 7142 }, { "epoch": 0.38081782801087594, "grad_norm": 1.7861133813858032, "learning_rate": 8.571600000000001e-06, "loss": 0.031, "step": 7143 }, { "epoch": 0.38087114144052886, "grad_norm": 6.298113822937012, "learning_rate": 8.5714e-06, "loss": 0.045, "step": 7144 }, { "epoch": 0.3809244548701818, "grad_norm": 3.5578572750091553, "learning_rate": 8.5712e-06, "loss": 0.0337, "step": 7145 }, { "epoch": 0.3809777682998347, "grad_norm": 2.603030204772949, "learning_rate": 8.571e-06, "loss": 0.0323, "step": 7146 }, { "epoch": 0.38103108172948763, "grad_norm": 4.279669284820557, "learning_rate": 8.570800000000001e-06, "loss": 0.0267, "step": 7147 }, { "epoch": 0.3810843951591406, "grad_norm": 2.890721559524536, "learning_rate": 8.5706e-06, "loss": 0.0628, "step": 7148 }, { "epoch": 0.38113770858879353, "grad_norm": 4.117905139923096, "learning_rate": 8.5704e-06, "loss": 0.0365, "step": 7149 }, { "epoch": 0.38119102201844646, "grad_norm": 3.1528942584991455, "learning_rate": 8.570200000000001e-06, "loss": 0.0302, "step": 7150 }, { "epoch": 0.3812443354480994, "grad_norm": 1.956010341644287, "learning_rate": 8.570000000000001e-06, "loss": 0.0228, "step": 7151 }, { "epoch": 0.3812976488777523, "grad_norm": 4.291037082672119, "learning_rate": 8.5698e-06, "loss": 0.0702, "step": 7152 }, { "epoch": 0.38135096230740523, "grad_norm": 3.075965404510498, "learning_rate": 8.569600000000002e-06, "loss": 0.0446, "step": 7153 }, { "epoch": 0.38140427573705815, "grad_norm": 7.201079845428467, "learning_rate": 8.569400000000001e-06, "loss": 0.1047, "step": 7154 }, { "epoch": 0.3814575891667111, "grad_norm": 2.236720561981201, "learning_rate": 8.569200000000001e-06, "loss": 0.0224, "step": 7155 }, { "epoch": 0.381510902596364, "grad_norm": 1.3296180963516235, "learning_rate": 8.569e-06, "loss": 0.0157, "step": 7156 }, { "epoch": 0.381564216026017, "grad_norm": 2.8983538150787354, "learning_rate": 8.5688e-06, "loss": 0.0286, "step": 7157 }, { "epoch": 0.3816175294556699, "grad_norm": 4.78933572769165, "learning_rate": 8.5686e-06, "loss": 0.0545, "step": 7158 }, { "epoch": 0.3816708428853228, "grad_norm": 2.3833258152008057, "learning_rate": 8.5684e-06, "loss": 0.0454, "step": 7159 }, { "epoch": 0.38172415631497575, "grad_norm": 2.8854291439056396, "learning_rate": 8.5682e-06, "loss": 0.056, "step": 7160 }, { "epoch": 0.3817774697446287, "grad_norm": 3.9232563972473145, "learning_rate": 8.568e-06, "loss": 0.0723, "step": 7161 }, { "epoch": 0.3818307831742816, "grad_norm": 1.2683216333389282, "learning_rate": 8.567800000000001e-06, "loss": 0.0118, "step": 7162 }, { "epoch": 0.3818840966039345, "grad_norm": 2.308415412902832, "learning_rate": 8.5676e-06, "loss": 0.0231, "step": 7163 }, { "epoch": 0.38193741003358744, "grad_norm": 1.797039270401001, "learning_rate": 8.5674e-06, "loss": 0.0258, "step": 7164 }, { "epoch": 0.38199072346324037, "grad_norm": 2.208139181137085, "learning_rate": 8.567200000000002e-06, "loss": 0.0486, "step": 7165 }, { "epoch": 0.38204403689289335, "grad_norm": 2.6974637508392334, "learning_rate": 8.567000000000001e-06, "loss": 0.0547, "step": 7166 }, { "epoch": 0.38209735032254627, "grad_norm": 3.2782063484191895, "learning_rate": 8.5668e-06, "loss": 0.0662, "step": 7167 }, { "epoch": 0.3821506637521992, "grad_norm": 4.5263824462890625, "learning_rate": 8.5666e-06, "loss": 0.0856, "step": 7168 }, { "epoch": 0.3822039771818521, "grad_norm": 3.1228325366973877, "learning_rate": 8.5664e-06, "loss": 0.0407, "step": 7169 }, { "epoch": 0.38225729061150504, "grad_norm": 9.270323753356934, "learning_rate": 8.566200000000001e-06, "loss": 0.0677, "step": 7170 }, { "epoch": 0.38231060404115796, "grad_norm": 5.107796669006348, "learning_rate": 8.566e-06, "loss": 0.0531, "step": 7171 }, { "epoch": 0.3823639174708109, "grad_norm": 5.339487552642822, "learning_rate": 8.5658e-06, "loss": 0.0396, "step": 7172 }, { "epoch": 0.3824172309004638, "grad_norm": 2.4252970218658447, "learning_rate": 8.5656e-06, "loss": 0.0323, "step": 7173 }, { "epoch": 0.38247054433011674, "grad_norm": 2.3355281352996826, "learning_rate": 8.565400000000001e-06, "loss": 0.013, "step": 7174 }, { "epoch": 0.38252385775976966, "grad_norm": 3.961400270462036, "learning_rate": 8.5652e-06, "loss": 0.0371, "step": 7175 }, { "epoch": 0.38257717118942264, "grad_norm": 3.469846725463867, "learning_rate": 8.565000000000002e-06, "loss": 0.0205, "step": 7176 }, { "epoch": 0.38263048461907556, "grad_norm": 5.143335819244385, "learning_rate": 8.564800000000001e-06, "loss": 0.0523, "step": 7177 }, { "epoch": 0.3826837980487285, "grad_norm": 4.209366321563721, "learning_rate": 8.564600000000001e-06, "loss": 0.0519, "step": 7178 }, { "epoch": 0.3827371114783814, "grad_norm": 2.2785985469818115, "learning_rate": 8.5644e-06, "loss": 0.0386, "step": 7179 }, { "epoch": 0.38279042490803433, "grad_norm": 3.1789116859436035, "learning_rate": 8.5642e-06, "loss": 0.0746, "step": 7180 }, { "epoch": 0.38284373833768726, "grad_norm": 2.9170029163360596, "learning_rate": 8.564000000000001e-06, "loss": 0.0459, "step": 7181 }, { "epoch": 0.3828970517673402, "grad_norm": 4.1119914054870605, "learning_rate": 8.563800000000001e-06, "loss": 0.063, "step": 7182 }, { "epoch": 0.3829503651969931, "grad_norm": 5.240184783935547, "learning_rate": 8.5636e-06, "loss": 0.0704, "step": 7183 }, { "epoch": 0.383003678626646, "grad_norm": 2.431716203689575, "learning_rate": 8.5634e-06, "loss": 0.0209, "step": 7184 }, { "epoch": 0.383056992056299, "grad_norm": 2.284787178039551, "learning_rate": 8.5632e-06, "loss": 0.0356, "step": 7185 }, { "epoch": 0.38311030548595193, "grad_norm": 3.10626482963562, "learning_rate": 8.563000000000001e-06, "loss": 0.0275, "step": 7186 }, { "epoch": 0.38316361891560485, "grad_norm": 3.158222198486328, "learning_rate": 8.5628e-06, "loss": 0.0542, "step": 7187 }, { "epoch": 0.3832169323452578, "grad_norm": 2.3041234016418457, "learning_rate": 8.562600000000002e-06, "loss": 0.0375, "step": 7188 }, { "epoch": 0.3832702457749107, "grad_norm": 1.6011619567871094, "learning_rate": 8.562400000000001e-06, "loss": 0.0227, "step": 7189 }, { "epoch": 0.3833235592045636, "grad_norm": 2.546862840652466, "learning_rate": 8.5622e-06, "loss": 0.0401, "step": 7190 }, { "epoch": 0.38337687263421655, "grad_norm": 1.4003610610961914, "learning_rate": 8.562e-06, "loss": 0.0275, "step": 7191 }, { "epoch": 0.38343018606386947, "grad_norm": 2.329176187515259, "learning_rate": 8.561800000000002e-06, "loss": 0.0319, "step": 7192 }, { "epoch": 0.3834834994935224, "grad_norm": 2.5597939491271973, "learning_rate": 8.561600000000001e-06, "loss": 0.0262, "step": 7193 }, { "epoch": 0.3835368129231754, "grad_norm": 2.4357171058654785, "learning_rate": 8.5614e-06, "loss": 0.0257, "step": 7194 }, { "epoch": 0.3835901263528283, "grad_norm": 3.5177090167999268, "learning_rate": 8.5612e-06, "loss": 0.0497, "step": 7195 }, { "epoch": 0.3836434397824812, "grad_norm": 2.5532758235931396, "learning_rate": 8.561e-06, "loss": 0.0585, "step": 7196 }, { "epoch": 0.38369675321213415, "grad_norm": 1.2359665632247925, "learning_rate": 8.5608e-06, "loss": 0.0095, "step": 7197 }, { "epoch": 0.38375006664178707, "grad_norm": 3.0315489768981934, "learning_rate": 8.5606e-06, "loss": 0.0243, "step": 7198 }, { "epoch": 0.38380338007144, "grad_norm": 2.843597650527954, "learning_rate": 8.5604e-06, "loss": 0.0261, "step": 7199 }, { "epoch": 0.3838566935010929, "grad_norm": 2.3229305744171143, "learning_rate": 8.560200000000001e-06, "loss": 0.0469, "step": 7200 }, { "epoch": 0.38391000693074584, "grad_norm": 3.6921212673187256, "learning_rate": 8.560000000000001e-06, "loss": 0.0267, "step": 7201 }, { "epoch": 0.38396332036039876, "grad_norm": 3.2820749282836914, "learning_rate": 8.5598e-06, "loss": 0.0507, "step": 7202 }, { "epoch": 0.3840166337900517, "grad_norm": 3.810145139694214, "learning_rate": 8.559600000000002e-06, "loss": 0.052, "step": 7203 }, { "epoch": 0.38406994721970467, "grad_norm": 3.1101021766662598, "learning_rate": 8.559400000000001e-06, "loss": 0.0228, "step": 7204 }, { "epoch": 0.3841232606493576, "grad_norm": 4.494278907775879, "learning_rate": 8.559200000000001e-06, "loss": 0.0429, "step": 7205 }, { "epoch": 0.3841765740790105, "grad_norm": 4.846410751342773, "learning_rate": 8.559e-06, "loss": 0.0449, "step": 7206 }, { "epoch": 0.38422988750866344, "grad_norm": 2.477738857269287, "learning_rate": 8.5588e-06, "loss": 0.0416, "step": 7207 }, { "epoch": 0.38428320093831636, "grad_norm": 3.4896278381347656, "learning_rate": 8.5586e-06, "loss": 0.0238, "step": 7208 }, { "epoch": 0.3843365143679693, "grad_norm": 5.89023494720459, "learning_rate": 8.558400000000001e-06, "loss": 0.0713, "step": 7209 }, { "epoch": 0.3843898277976222, "grad_norm": 6.047091484069824, "learning_rate": 8.5582e-06, "loss": 0.0613, "step": 7210 }, { "epoch": 0.38444314122727513, "grad_norm": 2.720050811767578, "learning_rate": 8.558e-06, "loss": 0.0211, "step": 7211 }, { "epoch": 0.38449645465692806, "grad_norm": 5.545614242553711, "learning_rate": 8.557800000000001e-06, "loss": 0.0394, "step": 7212 }, { "epoch": 0.38454976808658103, "grad_norm": 1.9217181205749512, "learning_rate": 8.557600000000001e-06, "loss": 0.0415, "step": 7213 }, { "epoch": 0.38460308151623396, "grad_norm": 2.6591713428497314, "learning_rate": 8.5574e-06, "loss": 0.0365, "step": 7214 }, { "epoch": 0.3846563949458869, "grad_norm": 3.5755083560943604, "learning_rate": 8.557200000000002e-06, "loss": 0.0671, "step": 7215 }, { "epoch": 0.3847097083755398, "grad_norm": 3.401632070541382, "learning_rate": 8.557000000000001e-06, "loss": 0.0831, "step": 7216 }, { "epoch": 0.38476302180519273, "grad_norm": 3.4225239753723145, "learning_rate": 8.5568e-06, "loss": 0.052, "step": 7217 }, { "epoch": 0.38481633523484565, "grad_norm": 6.055980205535889, "learning_rate": 8.5566e-06, "loss": 0.0858, "step": 7218 }, { "epoch": 0.3848696486644986, "grad_norm": 6.015074253082275, "learning_rate": 8.5564e-06, "loss": 0.0855, "step": 7219 }, { "epoch": 0.3849229620941515, "grad_norm": 5.310850620269775, "learning_rate": 8.556200000000001e-06, "loss": 0.0354, "step": 7220 }, { "epoch": 0.3849762755238044, "grad_norm": 3.839916706085205, "learning_rate": 8.556e-06, "loss": 0.0811, "step": 7221 }, { "epoch": 0.3850295889534574, "grad_norm": 5.812201499938965, "learning_rate": 8.5558e-06, "loss": 0.0468, "step": 7222 }, { "epoch": 0.3850829023831103, "grad_norm": 6.928668975830078, "learning_rate": 8.5556e-06, "loss": 0.0698, "step": 7223 }, { "epoch": 0.38513621581276325, "grad_norm": 4.129201889038086, "learning_rate": 8.555400000000001e-06, "loss": 0.0679, "step": 7224 }, { "epoch": 0.3851895292424162, "grad_norm": 2.839897394180298, "learning_rate": 8.5552e-06, "loss": 0.0444, "step": 7225 }, { "epoch": 0.3852428426720691, "grad_norm": 2.6201233863830566, "learning_rate": 8.555e-06, "loss": 0.0362, "step": 7226 }, { "epoch": 0.385296156101722, "grad_norm": 5.16197395324707, "learning_rate": 8.554800000000002e-06, "loss": 0.0385, "step": 7227 }, { "epoch": 0.38534946953137494, "grad_norm": 3.6947364807128906, "learning_rate": 8.554600000000001e-06, "loss": 0.0307, "step": 7228 }, { "epoch": 0.38540278296102787, "grad_norm": 4.50845193862915, "learning_rate": 8.5544e-06, "loss": 0.0729, "step": 7229 }, { "epoch": 0.3854560963906808, "grad_norm": 2.826915740966797, "learning_rate": 8.5542e-06, "loss": 0.0267, "step": 7230 }, { "epoch": 0.3855094098203337, "grad_norm": 5.465648174285889, "learning_rate": 8.554000000000001e-06, "loss": 0.0652, "step": 7231 }, { "epoch": 0.3855627232499867, "grad_norm": 6.626134872436523, "learning_rate": 8.553800000000001e-06, "loss": 0.0818, "step": 7232 }, { "epoch": 0.3856160366796396, "grad_norm": 2.508641481399536, "learning_rate": 8.5536e-06, "loss": 0.0285, "step": 7233 }, { "epoch": 0.38566935010929254, "grad_norm": 3.133908987045288, "learning_rate": 8.5534e-06, "loss": 0.0592, "step": 7234 }, { "epoch": 0.38572266353894546, "grad_norm": 2.998528003692627, "learning_rate": 8.5532e-06, "loss": 0.0262, "step": 7235 }, { "epoch": 0.3857759769685984, "grad_norm": 5.243727207183838, "learning_rate": 8.553000000000001e-06, "loss": 0.0619, "step": 7236 }, { "epoch": 0.3858292903982513, "grad_norm": 3.498159646987915, "learning_rate": 8.5528e-06, "loss": 0.028, "step": 7237 }, { "epoch": 0.38588260382790424, "grad_norm": 2.3963840007781982, "learning_rate": 8.5526e-06, "loss": 0.0339, "step": 7238 }, { "epoch": 0.38593591725755716, "grad_norm": 3.195312261581421, "learning_rate": 8.552400000000001e-06, "loss": 0.0227, "step": 7239 }, { "epoch": 0.3859892306872101, "grad_norm": 3.643221378326416, "learning_rate": 8.552200000000001e-06, "loss": 0.0745, "step": 7240 }, { "epoch": 0.38604254411686306, "grad_norm": 3.0867533683776855, "learning_rate": 8.552e-06, "loss": 0.046, "step": 7241 }, { "epoch": 0.386095857546516, "grad_norm": 2.56549334526062, "learning_rate": 8.551800000000002e-06, "loss": 0.0309, "step": 7242 }, { "epoch": 0.3861491709761689, "grad_norm": 2.7974491119384766, "learning_rate": 8.551600000000001e-06, "loss": 0.0179, "step": 7243 }, { "epoch": 0.38620248440582183, "grad_norm": 1.0236800909042358, "learning_rate": 8.5514e-06, "loss": 0.0079, "step": 7244 }, { "epoch": 0.38625579783547476, "grad_norm": 2.7330498695373535, "learning_rate": 8.5512e-06, "loss": 0.0359, "step": 7245 }, { "epoch": 0.3863091112651277, "grad_norm": 3.1768693923950195, "learning_rate": 8.551e-06, "loss": 0.039, "step": 7246 }, { "epoch": 0.3863624246947806, "grad_norm": 1.612544059753418, "learning_rate": 8.5508e-06, "loss": 0.0306, "step": 7247 }, { "epoch": 0.3864157381244335, "grad_norm": 3.439269781112671, "learning_rate": 8.5506e-06, "loss": 0.0283, "step": 7248 }, { "epoch": 0.38646905155408645, "grad_norm": 3.68917179107666, "learning_rate": 8.5504e-06, "loss": 0.0364, "step": 7249 }, { "epoch": 0.38652236498373943, "grad_norm": 2.1103341579437256, "learning_rate": 8.5502e-06, "loss": 0.0115, "step": 7250 }, { "epoch": 0.38657567841339235, "grad_norm": 3.257169723510742, "learning_rate": 8.550000000000001e-06, "loss": 0.0544, "step": 7251 }, { "epoch": 0.3866289918430453, "grad_norm": 2.516392946243286, "learning_rate": 8.5498e-06, "loss": 0.0291, "step": 7252 }, { "epoch": 0.3866823052726982, "grad_norm": 3.1023895740509033, "learning_rate": 8.549600000000002e-06, "loss": 0.0335, "step": 7253 }, { "epoch": 0.3867356187023511, "grad_norm": 3.9465491771698, "learning_rate": 8.549400000000002e-06, "loss": 0.0378, "step": 7254 }, { "epoch": 0.38678893213200405, "grad_norm": 3.8443336486816406, "learning_rate": 8.549200000000001e-06, "loss": 0.0548, "step": 7255 }, { "epoch": 0.38684224556165697, "grad_norm": 3.077704429626465, "learning_rate": 8.549e-06, "loss": 0.0395, "step": 7256 }, { "epoch": 0.3868955589913099, "grad_norm": 6.9168548583984375, "learning_rate": 8.5488e-06, "loss": 0.0574, "step": 7257 }, { "epoch": 0.3869488724209628, "grad_norm": 2.9978768825531006, "learning_rate": 8.5486e-06, "loss": 0.045, "step": 7258 }, { "epoch": 0.38700218585061574, "grad_norm": 1.5789579153060913, "learning_rate": 8.548400000000001e-06, "loss": 0.0116, "step": 7259 }, { "epoch": 0.3870554992802687, "grad_norm": 4.2148942947387695, "learning_rate": 8.5482e-06, "loss": 0.0714, "step": 7260 }, { "epoch": 0.38710881270992165, "grad_norm": 4.9373369216918945, "learning_rate": 8.548e-06, "loss": 0.0771, "step": 7261 }, { "epoch": 0.38716212613957457, "grad_norm": 4.557689666748047, "learning_rate": 8.5478e-06, "loss": 0.0392, "step": 7262 }, { "epoch": 0.3872154395692275, "grad_norm": 2.714077949523926, "learning_rate": 8.547600000000001e-06, "loss": 0.0249, "step": 7263 }, { "epoch": 0.3872687529988804, "grad_norm": 3.566877603530884, "learning_rate": 8.5474e-06, "loss": 0.0207, "step": 7264 }, { "epoch": 0.38732206642853334, "grad_norm": 5.360912322998047, "learning_rate": 8.547200000000002e-06, "loss": 0.047, "step": 7265 }, { "epoch": 0.38737537985818626, "grad_norm": 4.156881332397461, "learning_rate": 8.547000000000001e-06, "loss": 0.0479, "step": 7266 }, { "epoch": 0.3874286932878392, "grad_norm": 5.124048233032227, "learning_rate": 8.546800000000001e-06, "loss": 0.0349, "step": 7267 }, { "epoch": 0.3874820067174921, "grad_norm": 4.556978225708008, "learning_rate": 8.5466e-06, "loss": 0.0493, "step": 7268 }, { "epoch": 0.3875353201471451, "grad_norm": 3.728424549102783, "learning_rate": 8.5464e-06, "loss": 0.0508, "step": 7269 }, { "epoch": 0.387588633576798, "grad_norm": 2.1668362617492676, "learning_rate": 8.546200000000001e-06, "loss": 0.0276, "step": 7270 }, { "epoch": 0.38764194700645094, "grad_norm": 1.969949722290039, "learning_rate": 8.546000000000001e-06, "loss": 0.0352, "step": 7271 }, { "epoch": 0.38769526043610386, "grad_norm": 3.196296453475952, "learning_rate": 8.5458e-06, "loss": 0.0329, "step": 7272 }, { "epoch": 0.3877485738657568, "grad_norm": 3.4116225242614746, "learning_rate": 8.5456e-06, "loss": 0.0446, "step": 7273 }, { "epoch": 0.3878018872954097, "grad_norm": 4.3858866691589355, "learning_rate": 8.5454e-06, "loss": 0.0626, "step": 7274 }, { "epoch": 0.38785520072506263, "grad_norm": 2.029806137084961, "learning_rate": 8.5452e-06, "loss": 0.028, "step": 7275 }, { "epoch": 0.38790851415471556, "grad_norm": 2.5782296657562256, "learning_rate": 8.545e-06, "loss": 0.0537, "step": 7276 }, { "epoch": 0.3879618275843685, "grad_norm": 2.7684967517852783, "learning_rate": 8.544800000000002e-06, "loss": 0.0246, "step": 7277 }, { "epoch": 0.38801514101402146, "grad_norm": 4.037350177764893, "learning_rate": 8.544600000000001e-06, "loss": 0.0714, "step": 7278 }, { "epoch": 0.3880684544436744, "grad_norm": 6.464323997497559, "learning_rate": 8.5444e-06, "loss": 0.0549, "step": 7279 }, { "epoch": 0.3881217678733273, "grad_norm": 2.796175003051758, "learning_rate": 8.5442e-06, "loss": 0.0234, "step": 7280 }, { "epoch": 0.38817508130298023, "grad_norm": 2.9229960441589355, "learning_rate": 8.544000000000002e-06, "loss": 0.0693, "step": 7281 }, { "epoch": 0.38822839473263315, "grad_norm": 3.714519739151001, "learning_rate": 8.543800000000001e-06, "loss": 0.0613, "step": 7282 }, { "epoch": 0.3882817081622861, "grad_norm": 3.5886263847351074, "learning_rate": 8.5436e-06, "loss": 0.0287, "step": 7283 }, { "epoch": 0.388335021591939, "grad_norm": 5.740673542022705, "learning_rate": 8.5434e-06, "loss": 0.0673, "step": 7284 }, { "epoch": 0.3883883350215919, "grad_norm": 2.823543071746826, "learning_rate": 8.5432e-06, "loss": 0.0675, "step": 7285 }, { "epoch": 0.38844164845124485, "grad_norm": 2.376195192337036, "learning_rate": 8.543e-06, "loss": 0.0383, "step": 7286 }, { "epoch": 0.38849496188089777, "grad_norm": 3.479252815246582, "learning_rate": 8.5428e-06, "loss": 0.0603, "step": 7287 }, { "epoch": 0.38854827531055075, "grad_norm": 3.4542603492736816, "learning_rate": 8.5426e-06, "loss": 0.0529, "step": 7288 }, { "epoch": 0.3886015887402037, "grad_norm": 0.880453884601593, "learning_rate": 8.542400000000001e-06, "loss": 0.009, "step": 7289 }, { "epoch": 0.3886549021698566, "grad_norm": 1.8979520797729492, "learning_rate": 8.542200000000001e-06, "loss": 0.0274, "step": 7290 }, { "epoch": 0.3887082155995095, "grad_norm": 1.3719598054885864, "learning_rate": 8.542e-06, "loss": 0.0159, "step": 7291 }, { "epoch": 0.38876152902916244, "grad_norm": 3.2525529861450195, "learning_rate": 8.541800000000002e-06, "loss": 0.0432, "step": 7292 }, { "epoch": 0.38881484245881537, "grad_norm": 2.8955225944519043, "learning_rate": 8.541600000000001e-06, "loss": 0.053, "step": 7293 }, { "epoch": 0.3888681558884683, "grad_norm": 2.613879680633545, "learning_rate": 8.541400000000001e-06, "loss": 0.0374, "step": 7294 }, { "epoch": 0.3889214693181212, "grad_norm": 2.9810373783111572, "learning_rate": 8.5412e-06, "loss": 0.0249, "step": 7295 }, { "epoch": 0.38897478274777414, "grad_norm": 2.8139264583587646, "learning_rate": 8.541e-06, "loss": 0.0265, "step": 7296 }, { "epoch": 0.3890280961774271, "grad_norm": 4.15229606628418, "learning_rate": 8.5408e-06, "loss": 0.046, "step": 7297 }, { "epoch": 0.38908140960708004, "grad_norm": 2.175691843032837, "learning_rate": 8.540600000000001e-06, "loss": 0.0368, "step": 7298 }, { "epoch": 0.38913472303673297, "grad_norm": 3.4659860134124756, "learning_rate": 8.5404e-06, "loss": 0.0572, "step": 7299 }, { "epoch": 0.3891880364663859, "grad_norm": 3.3957061767578125, "learning_rate": 8.5402e-06, "loss": 0.0376, "step": 7300 }, { "epoch": 0.3892413498960388, "grad_norm": 3.343050479888916, "learning_rate": 8.540000000000001e-06, "loss": 0.0176, "step": 7301 }, { "epoch": 0.38929466332569174, "grad_norm": 2.890786647796631, "learning_rate": 8.5398e-06, "loss": 0.0482, "step": 7302 }, { "epoch": 0.38934797675534466, "grad_norm": 4.319984436035156, "learning_rate": 8.539600000000002e-06, "loss": 0.0454, "step": 7303 }, { "epoch": 0.3894012901849976, "grad_norm": 2.637399911880493, "learning_rate": 8.539400000000002e-06, "loss": 0.0326, "step": 7304 }, { "epoch": 0.3894546036146505, "grad_norm": 2.4481799602508545, "learning_rate": 8.539200000000001e-06, "loss": 0.0501, "step": 7305 }, { "epoch": 0.3895079170443035, "grad_norm": 1.7582814693450928, "learning_rate": 8.539e-06, "loss": 0.02, "step": 7306 }, { "epoch": 0.3895612304739564, "grad_norm": 4.6295061111450195, "learning_rate": 8.5388e-06, "loss": 0.0766, "step": 7307 }, { "epoch": 0.38961454390360933, "grad_norm": 3.88926362991333, "learning_rate": 8.5386e-06, "loss": 0.0448, "step": 7308 }, { "epoch": 0.38966785733326226, "grad_norm": 2.382620096206665, "learning_rate": 8.538400000000001e-06, "loss": 0.052, "step": 7309 }, { "epoch": 0.3897211707629152, "grad_norm": 3.310490608215332, "learning_rate": 8.5382e-06, "loss": 0.0556, "step": 7310 }, { "epoch": 0.3897744841925681, "grad_norm": 3.0743868350982666, "learning_rate": 8.538e-06, "loss": 0.0301, "step": 7311 }, { "epoch": 0.38982779762222103, "grad_norm": 3.0490589141845703, "learning_rate": 8.5378e-06, "loss": 0.0358, "step": 7312 }, { "epoch": 0.38988111105187395, "grad_norm": 1.5565810203552246, "learning_rate": 8.537600000000001e-06, "loss": 0.0112, "step": 7313 }, { "epoch": 0.3899344244815269, "grad_norm": 3.4562952518463135, "learning_rate": 8.5374e-06, "loss": 0.045, "step": 7314 }, { "epoch": 0.3899877379111798, "grad_norm": 1.39336359500885, "learning_rate": 8.537200000000002e-06, "loss": 0.021, "step": 7315 }, { "epoch": 0.3900410513408328, "grad_norm": 1.0053491592407227, "learning_rate": 8.537000000000001e-06, "loss": 0.011, "step": 7316 }, { "epoch": 0.3900943647704857, "grad_norm": 2.193852424621582, "learning_rate": 8.536800000000001e-06, "loss": 0.0432, "step": 7317 }, { "epoch": 0.3901476782001386, "grad_norm": 3.286846399307251, "learning_rate": 8.5366e-06, "loss": 0.0334, "step": 7318 }, { "epoch": 0.39020099162979155, "grad_norm": 2.0910184383392334, "learning_rate": 8.5364e-06, "loss": 0.0297, "step": 7319 }, { "epoch": 0.3902543050594445, "grad_norm": 3.708937883377075, "learning_rate": 8.536200000000001e-06, "loss": 0.0468, "step": 7320 }, { "epoch": 0.3903076184890974, "grad_norm": 2.0130796432495117, "learning_rate": 8.536000000000001e-06, "loss": 0.0233, "step": 7321 }, { "epoch": 0.3903609319187503, "grad_norm": 2.674368143081665, "learning_rate": 8.5358e-06, "loss": 0.0412, "step": 7322 }, { "epoch": 0.39041424534840324, "grad_norm": 1.9135178327560425, "learning_rate": 8.5356e-06, "loss": 0.0299, "step": 7323 }, { "epoch": 0.39046755877805617, "grad_norm": 2.0482406616210938, "learning_rate": 8.5354e-06, "loss": 0.0302, "step": 7324 }, { "epoch": 0.39052087220770915, "grad_norm": 2.387979030609131, "learning_rate": 8.535200000000001e-06, "loss": 0.0394, "step": 7325 }, { "epoch": 0.39057418563736207, "grad_norm": 3.9357047080993652, "learning_rate": 8.535e-06, "loss": 0.0882, "step": 7326 }, { "epoch": 0.390627499067015, "grad_norm": 2.217674970626831, "learning_rate": 8.534800000000002e-06, "loss": 0.0458, "step": 7327 }, { "epoch": 0.3906808124966679, "grad_norm": 2.160938262939453, "learning_rate": 8.534600000000001e-06, "loss": 0.046, "step": 7328 }, { "epoch": 0.39073412592632084, "grad_norm": 2.2823128700256348, "learning_rate": 8.5344e-06, "loss": 0.0552, "step": 7329 }, { "epoch": 0.39078743935597376, "grad_norm": 2.656599998474121, "learning_rate": 8.5342e-06, "loss": 0.0514, "step": 7330 }, { "epoch": 0.3908407527856267, "grad_norm": 5.3003764152526855, "learning_rate": 8.534000000000002e-06, "loss": 0.066, "step": 7331 }, { "epoch": 0.3908940662152796, "grad_norm": 1.7804138660430908, "learning_rate": 8.533800000000001e-06, "loss": 0.0246, "step": 7332 }, { "epoch": 0.39094737964493254, "grad_norm": 1.183660626411438, "learning_rate": 8.5336e-06, "loss": 0.0158, "step": 7333 }, { "epoch": 0.3910006930745855, "grad_norm": 3.66813325881958, "learning_rate": 8.5334e-06, "loss": 0.0313, "step": 7334 }, { "epoch": 0.39105400650423844, "grad_norm": 1.9138308763504028, "learning_rate": 8.5332e-06, "loss": 0.0368, "step": 7335 }, { "epoch": 0.39110731993389136, "grad_norm": 4.362485408782959, "learning_rate": 8.533e-06, "loss": 0.0424, "step": 7336 }, { "epoch": 0.3911606333635443, "grad_norm": 1.398478388786316, "learning_rate": 8.5328e-06, "loss": 0.0186, "step": 7337 }, { "epoch": 0.3912139467931972, "grad_norm": 2.2751717567443848, "learning_rate": 8.5326e-06, "loss": 0.0333, "step": 7338 }, { "epoch": 0.39126726022285013, "grad_norm": 2.2952828407287598, "learning_rate": 8.532400000000002e-06, "loss": 0.0196, "step": 7339 }, { "epoch": 0.39132057365250306, "grad_norm": 2.2813706398010254, "learning_rate": 8.532200000000001e-06, "loss": 0.0368, "step": 7340 }, { "epoch": 0.391373887082156, "grad_norm": 4.44993782043457, "learning_rate": 8.532e-06, "loss": 0.0601, "step": 7341 }, { "epoch": 0.3914272005118089, "grad_norm": 5.513818264007568, "learning_rate": 8.531800000000002e-06, "loss": 0.0853, "step": 7342 }, { "epoch": 0.3914805139414618, "grad_norm": 1.7698822021484375, "learning_rate": 8.531600000000001e-06, "loss": 0.036, "step": 7343 }, { "epoch": 0.3915338273711148, "grad_norm": 1.8095550537109375, "learning_rate": 8.531400000000001e-06, "loss": 0.0254, "step": 7344 }, { "epoch": 0.39158714080076773, "grad_norm": 6.252951622009277, "learning_rate": 8.5312e-06, "loss": 0.0459, "step": 7345 }, { "epoch": 0.39164045423042065, "grad_norm": 3.263871192932129, "learning_rate": 8.531e-06, "loss": 0.0334, "step": 7346 }, { "epoch": 0.3916937676600736, "grad_norm": 1.501718282699585, "learning_rate": 8.5308e-06, "loss": 0.0345, "step": 7347 }, { "epoch": 0.3917470810897265, "grad_norm": 2.1181375980377197, "learning_rate": 8.530600000000001e-06, "loss": 0.0213, "step": 7348 }, { "epoch": 0.3918003945193794, "grad_norm": 2.846330165863037, "learning_rate": 8.5304e-06, "loss": 0.0298, "step": 7349 }, { "epoch": 0.39185370794903235, "grad_norm": 5.101343154907227, "learning_rate": 8.5302e-06, "loss": 0.0647, "step": 7350 }, { "epoch": 0.39190702137868527, "grad_norm": 4.55341911315918, "learning_rate": 8.530000000000001e-06, "loss": 0.0582, "step": 7351 }, { "epoch": 0.3919603348083382, "grad_norm": 2.1848185062408447, "learning_rate": 8.529800000000001e-06, "loss": 0.0159, "step": 7352 }, { "epoch": 0.3920136482379912, "grad_norm": 4.25601863861084, "learning_rate": 8.5296e-06, "loss": 0.0681, "step": 7353 }, { "epoch": 0.3920669616676441, "grad_norm": 3.149951934814453, "learning_rate": 8.529400000000002e-06, "loss": 0.0253, "step": 7354 }, { "epoch": 0.392120275097297, "grad_norm": 3.499694585800171, "learning_rate": 8.529200000000001e-06, "loss": 0.0384, "step": 7355 }, { "epoch": 0.39217358852694995, "grad_norm": 2.567922353744507, "learning_rate": 8.529e-06, "loss": 0.0298, "step": 7356 }, { "epoch": 0.39222690195660287, "grad_norm": 3.801119565963745, "learning_rate": 8.5288e-06, "loss": 0.0503, "step": 7357 }, { "epoch": 0.3922802153862558, "grad_norm": 5.516757965087891, "learning_rate": 8.5286e-06, "loss": 0.0663, "step": 7358 }, { "epoch": 0.3923335288159087, "grad_norm": 2.1574931144714355, "learning_rate": 8.528400000000001e-06, "loss": 0.0323, "step": 7359 }, { "epoch": 0.39238684224556164, "grad_norm": 3.915977954864502, "learning_rate": 8.5282e-06, "loss": 0.0415, "step": 7360 }, { "epoch": 0.39244015567521456, "grad_norm": 5.215554237365723, "learning_rate": 8.528e-06, "loss": 0.0697, "step": 7361 }, { "epoch": 0.39249346910486754, "grad_norm": 3.931697130203247, "learning_rate": 8.5278e-06, "loss": 0.0789, "step": 7362 }, { "epoch": 0.39254678253452047, "grad_norm": 2.326533317565918, "learning_rate": 8.527600000000001e-06, "loss": 0.035, "step": 7363 }, { "epoch": 0.3926000959641734, "grad_norm": 2.984341859817505, "learning_rate": 8.5274e-06, "loss": 0.0358, "step": 7364 }, { "epoch": 0.3926534093938263, "grad_norm": 2.6611275672912598, "learning_rate": 8.5272e-06, "loss": 0.0395, "step": 7365 }, { "epoch": 0.39270672282347924, "grad_norm": 3.0752575397491455, "learning_rate": 8.527000000000002e-06, "loss": 0.0203, "step": 7366 }, { "epoch": 0.39276003625313216, "grad_norm": 1.8933207988739014, "learning_rate": 8.526800000000001e-06, "loss": 0.0392, "step": 7367 }, { "epoch": 0.3928133496827851, "grad_norm": 2.240922689437866, "learning_rate": 8.5266e-06, "loss": 0.0253, "step": 7368 }, { "epoch": 0.392866663112438, "grad_norm": 2.3755180835723877, "learning_rate": 8.5264e-06, "loss": 0.0226, "step": 7369 }, { "epoch": 0.39291997654209093, "grad_norm": 3.097012519836426, "learning_rate": 8.526200000000001e-06, "loss": 0.0406, "step": 7370 }, { "epoch": 0.39297328997174386, "grad_norm": 1.588475227355957, "learning_rate": 8.526000000000001e-06, "loss": 0.0164, "step": 7371 }, { "epoch": 0.39302660340139683, "grad_norm": 2.371262550354004, "learning_rate": 8.5258e-06, "loss": 0.0301, "step": 7372 }, { "epoch": 0.39307991683104976, "grad_norm": 2.8211872577667236, "learning_rate": 8.5256e-06, "loss": 0.0465, "step": 7373 }, { "epoch": 0.3931332302607027, "grad_norm": 2.9361464977264404, "learning_rate": 8.5254e-06, "loss": 0.0444, "step": 7374 }, { "epoch": 0.3931865436903556, "grad_norm": 3.430769681930542, "learning_rate": 8.525200000000001e-06, "loss": 0.0167, "step": 7375 }, { "epoch": 0.39323985712000853, "grad_norm": 2.7419965267181396, "learning_rate": 8.525e-06, "loss": 0.0278, "step": 7376 }, { "epoch": 0.39329317054966145, "grad_norm": 5.352997303009033, "learning_rate": 8.5248e-06, "loss": 0.0296, "step": 7377 }, { "epoch": 0.3933464839793144, "grad_norm": 3.544313907623291, "learning_rate": 8.524600000000001e-06, "loss": 0.0419, "step": 7378 }, { "epoch": 0.3933997974089673, "grad_norm": 3.5560615062713623, "learning_rate": 8.524400000000001e-06, "loss": 0.0285, "step": 7379 }, { "epoch": 0.3934531108386202, "grad_norm": 4.337634563446045, "learning_rate": 8.5242e-06, "loss": 0.0514, "step": 7380 }, { "epoch": 0.3935064242682732, "grad_norm": 8.056051254272461, "learning_rate": 8.524000000000002e-06, "loss": 0.0398, "step": 7381 }, { "epoch": 0.3935597376979261, "grad_norm": 4.558933258056641, "learning_rate": 8.523800000000001e-06, "loss": 0.0673, "step": 7382 }, { "epoch": 0.39361305112757905, "grad_norm": 3.2573959827423096, "learning_rate": 8.523600000000001e-06, "loss": 0.0883, "step": 7383 }, { "epoch": 0.393666364557232, "grad_norm": 4.023209095001221, "learning_rate": 8.5234e-06, "loss": 0.0713, "step": 7384 }, { "epoch": 0.3937196779868849, "grad_norm": 2.395920991897583, "learning_rate": 8.5232e-06, "loss": 0.0371, "step": 7385 }, { "epoch": 0.3937729914165378, "grad_norm": 6.12778377532959, "learning_rate": 8.523e-06, "loss": 0.0586, "step": 7386 }, { "epoch": 0.39382630484619074, "grad_norm": 3.7815804481506348, "learning_rate": 8.5228e-06, "loss": 0.0847, "step": 7387 }, { "epoch": 0.39387961827584367, "grad_norm": 4.968860626220703, "learning_rate": 8.5226e-06, "loss": 0.0753, "step": 7388 }, { "epoch": 0.3939329317054966, "grad_norm": 3.7383482456207275, "learning_rate": 8.5224e-06, "loss": 0.0387, "step": 7389 }, { "epoch": 0.39398624513514957, "grad_norm": 7.618319034576416, "learning_rate": 8.522200000000001e-06, "loss": 0.0779, "step": 7390 }, { "epoch": 0.3940395585648025, "grad_norm": 3.374387741088867, "learning_rate": 8.522e-06, "loss": 0.0492, "step": 7391 }, { "epoch": 0.3940928719944554, "grad_norm": 5.465934753417969, "learning_rate": 8.521800000000002e-06, "loss": 0.0939, "step": 7392 }, { "epoch": 0.39414618542410834, "grad_norm": 2.5762853622436523, "learning_rate": 8.521600000000002e-06, "loss": 0.0325, "step": 7393 }, { "epoch": 0.39419949885376127, "grad_norm": 1.9242286682128906, "learning_rate": 8.521400000000001e-06, "loss": 0.0228, "step": 7394 }, { "epoch": 0.3942528122834142, "grad_norm": 2.3387341499328613, "learning_rate": 8.5212e-06, "loss": 0.0328, "step": 7395 }, { "epoch": 0.3943061257130671, "grad_norm": 3.9435837268829346, "learning_rate": 8.521e-06, "loss": 0.0569, "step": 7396 }, { "epoch": 0.39435943914272004, "grad_norm": 2.31884503364563, "learning_rate": 8.5208e-06, "loss": 0.0415, "step": 7397 }, { "epoch": 0.39441275257237296, "grad_norm": 2.6661574840545654, "learning_rate": 8.520600000000001e-06, "loss": 0.0239, "step": 7398 }, { "epoch": 0.3944660660020259, "grad_norm": 3.9814646244049072, "learning_rate": 8.5204e-06, "loss": 0.0803, "step": 7399 }, { "epoch": 0.39451937943167886, "grad_norm": 1.1490274667739868, "learning_rate": 8.5202e-06, "loss": 0.0121, "step": 7400 }, { "epoch": 0.3945726928613318, "grad_norm": 2.071538209915161, "learning_rate": 8.52e-06, "loss": 0.0264, "step": 7401 }, { "epoch": 0.3946260062909847, "grad_norm": 3.198594331741333, "learning_rate": 8.519800000000001e-06, "loss": 0.0497, "step": 7402 }, { "epoch": 0.39467931972063763, "grad_norm": 2.58685302734375, "learning_rate": 8.5196e-06, "loss": 0.0246, "step": 7403 }, { "epoch": 0.39473263315029056, "grad_norm": 1.8671263456344604, "learning_rate": 8.519400000000002e-06, "loss": 0.0189, "step": 7404 }, { "epoch": 0.3947859465799435, "grad_norm": 3.9734890460968018, "learning_rate": 8.519200000000001e-06, "loss": 0.0234, "step": 7405 }, { "epoch": 0.3948392600095964, "grad_norm": 3.1691670417785645, "learning_rate": 8.519000000000001e-06, "loss": 0.0658, "step": 7406 }, { "epoch": 0.39489257343924933, "grad_norm": 1.9716105461120605, "learning_rate": 8.5188e-06, "loss": 0.0195, "step": 7407 }, { "epoch": 0.39494588686890225, "grad_norm": 3.0896337032318115, "learning_rate": 8.5186e-06, "loss": 0.0525, "step": 7408 }, { "epoch": 0.39499920029855523, "grad_norm": 1.8653568029403687, "learning_rate": 8.518400000000001e-06, "loss": 0.024, "step": 7409 }, { "epoch": 0.39505251372820815, "grad_norm": 5.60355806350708, "learning_rate": 8.518200000000001e-06, "loss": 0.082, "step": 7410 }, { "epoch": 0.3951058271578611, "grad_norm": 3.946025848388672, "learning_rate": 8.518e-06, "loss": 0.0398, "step": 7411 }, { "epoch": 0.395159140587514, "grad_norm": 1.8030468225479126, "learning_rate": 8.5178e-06, "loss": 0.0417, "step": 7412 }, { "epoch": 0.3952124540171669, "grad_norm": 2.4059619903564453, "learning_rate": 8.5176e-06, "loss": 0.0371, "step": 7413 }, { "epoch": 0.39526576744681985, "grad_norm": 4.199915885925293, "learning_rate": 8.5174e-06, "loss": 0.0715, "step": 7414 }, { "epoch": 0.3953190808764728, "grad_norm": 4.090373992919922, "learning_rate": 8.5172e-06, "loss": 0.082, "step": 7415 }, { "epoch": 0.3953723943061257, "grad_norm": 1.6157417297363281, "learning_rate": 8.517000000000002e-06, "loss": 0.0232, "step": 7416 }, { "epoch": 0.3954257077357786, "grad_norm": 1.592361569404602, "learning_rate": 8.516800000000001e-06, "loss": 0.0242, "step": 7417 }, { "epoch": 0.3954790211654316, "grad_norm": 4.560198783874512, "learning_rate": 8.5166e-06, "loss": 0.0794, "step": 7418 }, { "epoch": 0.3955323345950845, "grad_norm": 5.4658894538879395, "learning_rate": 8.5164e-06, "loss": 0.0433, "step": 7419 }, { "epoch": 0.39558564802473745, "grad_norm": 4.183846473693848, "learning_rate": 8.516200000000002e-06, "loss": 0.0333, "step": 7420 }, { "epoch": 0.39563896145439037, "grad_norm": 3.451608896255493, "learning_rate": 8.516000000000001e-06, "loss": 0.0312, "step": 7421 }, { "epoch": 0.3956922748840433, "grad_norm": 2.4602584838867188, "learning_rate": 8.5158e-06, "loss": 0.0348, "step": 7422 }, { "epoch": 0.3957455883136962, "grad_norm": 3.551269292831421, "learning_rate": 8.5156e-06, "loss": 0.0829, "step": 7423 }, { "epoch": 0.39579890174334914, "grad_norm": 2.8468472957611084, "learning_rate": 8.5154e-06, "loss": 0.0212, "step": 7424 }, { "epoch": 0.39585221517300206, "grad_norm": 3.559417247772217, "learning_rate": 8.515200000000001e-06, "loss": 0.0349, "step": 7425 }, { "epoch": 0.395905528602655, "grad_norm": 3.4945948123931885, "learning_rate": 8.515e-06, "loss": 0.0292, "step": 7426 }, { "epoch": 0.3959588420323079, "grad_norm": 5.283374786376953, "learning_rate": 8.5148e-06, "loss": 0.0263, "step": 7427 }, { "epoch": 0.3960121554619609, "grad_norm": 2.6288492679595947, "learning_rate": 8.514600000000001e-06, "loss": 0.0377, "step": 7428 }, { "epoch": 0.3960654688916138, "grad_norm": 1.7919334173202515, "learning_rate": 8.514400000000001e-06, "loss": 0.0193, "step": 7429 }, { "epoch": 0.39611878232126674, "grad_norm": 3.1682980060577393, "learning_rate": 8.5142e-06, "loss": 0.0328, "step": 7430 }, { "epoch": 0.39617209575091966, "grad_norm": 1.8861967325210571, "learning_rate": 8.514000000000002e-06, "loss": 0.0304, "step": 7431 }, { "epoch": 0.3962254091805726, "grad_norm": 5.339797496795654, "learning_rate": 8.513800000000001e-06, "loss": 0.0549, "step": 7432 }, { "epoch": 0.3962787226102255, "grad_norm": 4.2484660148620605, "learning_rate": 8.513600000000001e-06, "loss": 0.0697, "step": 7433 }, { "epoch": 0.39633203603987843, "grad_norm": 5.205305099487305, "learning_rate": 8.5134e-06, "loss": 0.0844, "step": 7434 }, { "epoch": 0.39638534946953136, "grad_norm": 5.179315567016602, "learning_rate": 8.5132e-06, "loss": 0.0693, "step": 7435 }, { "epoch": 0.3964386628991843, "grad_norm": 1.2964308261871338, "learning_rate": 8.513e-06, "loss": 0.0108, "step": 7436 }, { "epoch": 0.39649197632883726, "grad_norm": 1.9486709833145142, "learning_rate": 8.512800000000001e-06, "loss": 0.0216, "step": 7437 }, { "epoch": 0.3965452897584902, "grad_norm": 1.6084041595458984, "learning_rate": 8.5126e-06, "loss": 0.0224, "step": 7438 }, { "epoch": 0.3965986031881431, "grad_norm": 6.362576007843018, "learning_rate": 8.5124e-06, "loss": 0.0588, "step": 7439 }, { "epoch": 0.39665191661779603, "grad_norm": 5.3615193367004395, "learning_rate": 8.512200000000001e-06, "loss": 0.0594, "step": 7440 }, { "epoch": 0.39670523004744895, "grad_norm": 3.452375650405884, "learning_rate": 8.512e-06, "loss": 0.0584, "step": 7441 }, { "epoch": 0.3967585434771019, "grad_norm": 1.3186924457550049, "learning_rate": 8.5118e-06, "loss": 0.0098, "step": 7442 }, { "epoch": 0.3968118569067548, "grad_norm": 2.5025579929351807, "learning_rate": 8.511600000000002e-06, "loss": 0.0267, "step": 7443 }, { "epoch": 0.3968651703364077, "grad_norm": 3.067913055419922, "learning_rate": 8.511400000000001e-06, "loss": 0.0254, "step": 7444 }, { "epoch": 0.39691848376606065, "grad_norm": 3.2603201866149902, "learning_rate": 8.5112e-06, "loss": 0.0334, "step": 7445 }, { "epoch": 0.3969717971957136, "grad_norm": 4.147237777709961, "learning_rate": 8.511e-06, "loss": 0.0599, "step": 7446 }, { "epoch": 0.39702511062536655, "grad_norm": 3.0944135189056396, "learning_rate": 8.5108e-06, "loss": 0.0262, "step": 7447 }, { "epoch": 0.3970784240550195, "grad_norm": 3.055297374725342, "learning_rate": 8.510600000000001e-06, "loss": 0.0473, "step": 7448 }, { "epoch": 0.3971317374846724, "grad_norm": 2.3906519412994385, "learning_rate": 8.5104e-06, "loss": 0.0202, "step": 7449 }, { "epoch": 0.3971850509143253, "grad_norm": 3.0927011966705322, "learning_rate": 8.5102e-06, "loss": 0.042, "step": 7450 }, { "epoch": 0.39723836434397825, "grad_norm": 3.308053731918335, "learning_rate": 8.51e-06, "loss": 0.0386, "step": 7451 }, { "epoch": 0.39729167777363117, "grad_norm": 3.333786725997925, "learning_rate": 8.509800000000001e-06, "loss": 0.0542, "step": 7452 }, { "epoch": 0.3973449912032841, "grad_norm": 5.057849407196045, "learning_rate": 8.5096e-06, "loss": 0.0807, "step": 7453 }, { "epoch": 0.397398304632937, "grad_norm": 2.7263593673706055, "learning_rate": 8.5094e-06, "loss": 0.0322, "step": 7454 }, { "epoch": 0.39745161806258994, "grad_norm": 1.9377259016036987, "learning_rate": 8.509200000000001e-06, "loss": 0.0148, "step": 7455 }, { "epoch": 0.3975049314922429, "grad_norm": 2.2839932441711426, "learning_rate": 8.509000000000001e-06, "loss": 0.0161, "step": 7456 }, { "epoch": 0.39755824492189584, "grad_norm": 3.863553047180176, "learning_rate": 8.5088e-06, "loss": 0.0634, "step": 7457 }, { "epoch": 0.39761155835154877, "grad_norm": 2.709780693054199, "learning_rate": 8.5086e-06, "loss": 0.0339, "step": 7458 }, { "epoch": 0.3976648717812017, "grad_norm": 1.9173755645751953, "learning_rate": 8.508400000000001e-06, "loss": 0.0186, "step": 7459 }, { "epoch": 0.3977181852108546, "grad_norm": 2.6136879920959473, "learning_rate": 8.508200000000001e-06, "loss": 0.0201, "step": 7460 }, { "epoch": 0.39777149864050754, "grad_norm": 3.9894137382507324, "learning_rate": 8.508e-06, "loss": 0.0325, "step": 7461 }, { "epoch": 0.39782481207016046, "grad_norm": 2.9245445728302, "learning_rate": 8.5078e-06, "loss": 0.0833, "step": 7462 }, { "epoch": 0.3978781254998134, "grad_norm": 3.4305036067962646, "learning_rate": 8.5076e-06, "loss": 0.0518, "step": 7463 }, { "epoch": 0.3979314389294663, "grad_norm": 2.981837272644043, "learning_rate": 8.507400000000001e-06, "loss": 0.0351, "step": 7464 }, { "epoch": 0.3979847523591193, "grad_norm": 6.020973205566406, "learning_rate": 8.5072e-06, "loss": 0.0619, "step": 7465 }, { "epoch": 0.3980380657887722, "grad_norm": 4.015032768249512, "learning_rate": 8.507e-06, "loss": 0.0488, "step": 7466 }, { "epoch": 0.39809137921842513, "grad_norm": 1.636239767074585, "learning_rate": 8.506800000000001e-06, "loss": 0.0304, "step": 7467 }, { "epoch": 0.39814469264807806, "grad_norm": 3.4863312244415283, "learning_rate": 8.506600000000001e-06, "loss": 0.0465, "step": 7468 }, { "epoch": 0.398198006077731, "grad_norm": 3.209995746612549, "learning_rate": 8.5064e-06, "loss": 0.0503, "step": 7469 }, { "epoch": 0.3982513195073839, "grad_norm": 1.6075479984283447, "learning_rate": 8.506200000000002e-06, "loss": 0.0344, "step": 7470 }, { "epoch": 0.39830463293703683, "grad_norm": 4.426925182342529, "learning_rate": 8.506000000000001e-06, "loss": 0.0791, "step": 7471 }, { "epoch": 0.39835794636668975, "grad_norm": 2.3193678855895996, "learning_rate": 8.5058e-06, "loss": 0.0647, "step": 7472 }, { "epoch": 0.3984112597963427, "grad_norm": 1.6753923892974854, "learning_rate": 8.5056e-06, "loss": 0.0149, "step": 7473 }, { "epoch": 0.39846457322599566, "grad_norm": 2.2331316471099854, "learning_rate": 8.5054e-06, "loss": 0.0293, "step": 7474 }, { "epoch": 0.3985178866556486, "grad_norm": 3.2516918182373047, "learning_rate": 8.505200000000001e-06, "loss": 0.0658, "step": 7475 }, { "epoch": 0.3985712000853015, "grad_norm": 1.574217677116394, "learning_rate": 8.505e-06, "loss": 0.0298, "step": 7476 }, { "epoch": 0.3986245135149544, "grad_norm": 3.1923282146453857, "learning_rate": 8.5048e-06, "loss": 0.0342, "step": 7477 }, { "epoch": 0.39867782694460735, "grad_norm": 4.708161354064941, "learning_rate": 8.5046e-06, "loss": 0.0696, "step": 7478 }, { "epoch": 0.3987311403742603, "grad_norm": 3.23565936088562, "learning_rate": 8.504400000000001e-06, "loss": 0.0318, "step": 7479 }, { "epoch": 0.3987844538039132, "grad_norm": 2.255586862564087, "learning_rate": 8.5042e-06, "loss": 0.0272, "step": 7480 }, { "epoch": 0.3988377672335661, "grad_norm": 2.5279436111450195, "learning_rate": 8.504000000000002e-06, "loss": 0.0626, "step": 7481 }, { "epoch": 0.39889108066321904, "grad_norm": 2.436067581176758, "learning_rate": 8.503800000000001e-06, "loss": 0.0275, "step": 7482 }, { "epoch": 0.39894439409287197, "grad_norm": 2.8121557235717773, "learning_rate": 8.503600000000001e-06, "loss": 0.0202, "step": 7483 }, { "epoch": 0.39899770752252495, "grad_norm": 1.7973167896270752, "learning_rate": 8.5034e-06, "loss": 0.0314, "step": 7484 }, { "epoch": 0.39905102095217787, "grad_norm": 2.3346219062805176, "learning_rate": 8.5032e-06, "loss": 0.0394, "step": 7485 }, { "epoch": 0.3991043343818308, "grad_norm": 1.9186270236968994, "learning_rate": 8.503e-06, "loss": 0.0425, "step": 7486 }, { "epoch": 0.3991576478114837, "grad_norm": 2.5492918491363525, "learning_rate": 8.502800000000001e-06, "loss": 0.0461, "step": 7487 }, { "epoch": 0.39921096124113664, "grad_norm": 2.444777011871338, "learning_rate": 8.5026e-06, "loss": 0.0602, "step": 7488 }, { "epoch": 0.39926427467078957, "grad_norm": 2.229509115219116, "learning_rate": 8.5024e-06, "loss": 0.0264, "step": 7489 }, { "epoch": 0.3993175881004425, "grad_norm": 2.6000046730041504, "learning_rate": 8.5022e-06, "loss": 0.0383, "step": 7490 }, { "epoch": 0.3993709015300954, "grad_norm": 2.221950054168701, "learning_rate": 8.502000000000001e-06, "loss": 0.038, "step": 7491 }, { "epoch": 0.39942421495974834, "grad_norm": 1.9735419750213623, "learning_rate": 8.5018e-06, "loss": 0.0232, "step": 7492 }, { "epoch": 0.3994775283894013, "grad_norm": 2.917217969894409, "learning_rate": 8.501600000000002e-06, "loss": 0.038, "step": 7493 }, { "epoch": 0.39953084181905424, "grad_norm": 3.062537670135498, "learning_rate": 8.501400000000001e-06, "loss": 0.0477, "step": 7494 }, { "epoch": 0.39958415524870716, "grad_norm": 3.74326753616333, "learning_rate": 8.501200000000001e-06, "loss": 0.0643, "step": 7495 }, { "epoch": 0.3996374686783601, "grad_norm": 5.433439254760742, "learning_rate": 8.501e-06, "loss": 0.0892, "step": 7496 }, { "epoch": 0.399690782108013, "grad_norm": 3.055845022201538, "learning_rate": 8.5008e-06, "loss": 0.0383, "step": 7497 }, { "epoch": 0.39974409553766593, "grad_norm": 1.8050267696380615, "learning_rate": 8.500600000000001e-06, "loss": 0.0377, "step": 7498 }, { "epoch": 0.39979740896731886, "grad_norm": 2.525101661682129, "learning_rate": 8.5004e-06, "loss": 0.0469, "step": 7499 }, { "epoch": 0.3998507223969718, "grad_norm": 3.3984932899475098, "learning_rate": 8.5002e-06, "loss": 0.0552, "step": 7500 }, { "epoch": 0.3999040358266247, "grad_norm": 2.1060707569122314, "learning_rate": 8.5e-06, "loss": 0.0224, "step": 7501 }, { "epoch": 0.3999573492562777, "grad_norm": 2.8420917987823486, "learning_rate": 8.4998e-06, "loss": 0.0293, "step": 7502 }, { "epoch": 0.4000106626859306, "grad_norm": 5.1835856437683105, "learning_rate": 8.4996e-06, "loss": 0.032, "step": 7503 }, { "epoch": 0.40006397611558353, "grad_norm": 3.0913262367248535, "learning_rate": 8.4994e-06, "loss": 0.0663, "step": 7504 }, { "epoch": 0.40011728954523645, "grad_norm": 3.166316270828247, "learning_rate": 8.499200000000002e-06, "loss": 0.0338, "step": 7505 }, { "epoch": 0.4001706029748894, "grad_norm": 2.6045217514038086, "learning_rate": 8.499000000000001e-06, "loss": 0.0262, "step": 7506 }, { "epoch": 0.4002239164045423, "grad_norm": 1.7236592769622803, "learning_rate": 8.4988e-06, "loss": 0.0273, "step": 7507 }, { "epoch": 0.4002772298341952, "grad_norm": 2.99310040473938, "learning_rate": 8.4986e-06, "loss": 0.0253, "step": 7508 }, { "epoch": 0.40033054326384815, "grad_norm": 1.9085476398468018, "learning_rate": 8.498400000000002e-06, "loss": 0.0199, "step": 7509 }, { "epoch": 0.4003838566935011, "grad_norm": 3.6357381343841553, "learning_rate": 8.498200000000001e-06, "loss": 0.0406, "step": 7510 }, { "epoch": 0.400437170123154, "grad_norm": 1.7835637331008911, "learning_rate": 8.498e-06, "loss": 0.018, "step": 7511 }, { "epoch": 0.400490483552807, "grad_norm": 4.276575565338135, "learning_rate": 8.4978e-06, "loss": 0.0384, "step": 7512 }, { "epoch": 0.4005437969824599, "grad_norm": 4.043800354003906, "learning_rate": 8.4976e-06, "loss": 0.0546, "step": 7513 }, { "epoch": 0.4005971104121128, "grad_norm": 4.140431880950928, "learning_rate": 8.497400000000001e-06, "loss": 0.0607, "step": 7514 }, { "epoch": 0.40065042384176575, "grad_norm": 2.899360418319702, "learning_rate": 8.4972e-06, "loss": 0.0583, "step": 7515 }, { "epoch": 0.40070373727141867, "grad_norm": 2.0737619400024414, "learning_rate": 8.497e-06, "loss": 0.0206, "step": 7516 }, { "epoch": 0.4007570507010716, "grad_norm": 2.760646104812622, "learning_rate": 8.496800000000001e-06, "loss": 0.0441, "step": 7517 }, { "epoch": 0.4008103641307245, "grad_norm": 3.644944190979004, "learning_rate": 8.496600000000001e-06, "loss": 0.0464, "step": 7518 }, { "epoch": 0.40086367756037744, "grad_norm": 3.05696964263916, "learning_rate": 8.4964e-06, "loss": 0.0327, "step": 7519 }, { "epoch": 0.40091699099003036, "grad_norm": 2.6395554542541504, "learning_rate": 8.496200000000002e-06, "loss": 0.0172, "step": 7520 }, { "epoch": 0.40097030441968334, "grad_norm": 3.051624059677124, "learning_rate": 8.496000000000001e-06, "loss": 0.0459, "step": 7521 }, { "epoch": 0.40102361784933627, "grad_norm": 3.5714497566223145, "learning_rate": 8.495800000000001e-06, "loss": 0.047, "step": 7522 }, { "epoch": 0.4010769312789892, "grad_norm": 3.1057934761047363, "learning_rate": 8.4956e-06, "loss": 0.0267, "step": 7523 }, { "epoch": 0.4011302447086421, "grad_norm": 3.7019741535186768, "learning_rate": 8.4954e-06, "loss": 0.0532, "step": 7524 }, { "epoch": 0.40118355813829504, "grad_norm": 1.826816439628601, "learning_rate": 8.495200000000001e-06, "loss": 0.039, "step": 7525 }, { "epoch": 0.40123687156794796, "grad_norm": 5.518064975738525, "learning_rate": 8.495e-06, "loss": 0.0688, "step": 7526 }, { "epoch": 0.4012901849976009, "grad_norm": 3.2584762573242188, "learning_rate": 8.4948e-06, "loss": 0.0513, "step": 7527 }, { "epoch": 0.4013434984272538, "grad_norm": 3.0601487159729004, "learning_rate": 8.4946e-06, "loss": 0.0371, "step": 7528 }, { "epoch": 0.40139681185690673, "grad_norm": 1.9924790859222412, "learning_rate": 8.494400000000001e-06, "loss": 0.0161, "step": 7529 }, { "epoch": 0.4014501252865597, "grad_norm": 4.9456071853637695, "learning_rate": 8.4942e-06, "loss": 0.0632, "step": 7530 }, { "epoch": 0.40150343871621264, "grad_norm": 2.93884539604187, "learning_rate": 8.494e-06, "loss": 0.0298, "step": 7531 }, { "epoch": 0.40155675214586556, "grad_norm": 6.515923500061035, "learning_rate": 8.493800000000002e-06, "loss": 0.048, "step": 7532 }, { "epoch": 0.4016100655755185, "grad_norm": 4.3581414222717285, "learning_rate": 8.493600000000001e-06, "loss": 0.0478, "step": 7533 }, { "epoch": 0.4016633790051714, "grad_norm": 3.4866018295288086, "learning_rate": 8.4934e-06, "loss": 0.0919, "step": 7534 }, { "epoch": 0.40171669243482433, "grad_norm": 3.2042431831359863, "learning_rate": 8.4932e-06, "loss": 0.0307, "step": 7535 }, { "epoch": 0.40177000586447725, "grad_norm": 3.5980758666992188, "learning_rate": 8.493000000000002e-06, "loss": 0.0492, "step": 7536 }, { "epoch": 0.4018233192941302, "grad_norm": 2.0508906841278076, "learning_rate": 8.492800000000001e-06, "loss": 0.0302, "step": 7537 }, { "epoch": 0.4018766327237831, "grad_norm": 3.4562795162200928, "learning_rate": 8.4926e-06, "loss": 0.0671, "step": 7538 }, { "epoch": 0.401929946153436, "grad_norm": 2.1134514808654785, "learning_rate": 8.4924e-06, "loss": 0.0198, "step": 7539 }, { "epoch": 0.401983259583089, "grad_norm": 2.4601476192474365, "learning_rate": 8.4922e-06, "loss": 0.0659, "step": 7540 }, { "epoch": 0.4020365730127419, "grad_norm": 2.4269118309020996, "learning_rate": 8.492000000000001e-06, "loss": 0.0443, "step": 7541 }, { "epoch": 0.40208988644239485, "grad_norm": 3.1355109214782715, "learning_rate": 8.4918e-06, "loss": 0.0521, "step": 7542 }, { "epoch": 0.4021431998720478, "grad_norm": 2.9710750579833984, "learning_rate": 8.4916e-06, "loss": 0.0252, "step": 7543 }, { "epoch": 0.4021965133017007, "grad_norm": 4.063722133636475, "learning_rate": 8.491400000000001e-06, "loss": 0.0485, "step": 7544 }, { "epoch": 0.4022498267313536, "grad_norm": 4.06224250793457, "learning_rate": 8.491200000000001e-06, "loss": 0.0461, "step": 7545 }, { "epoch": 0.40230314016100654, "grad_norm": 5.912726879119873, "learning_rate": 8.491e-06, "loss": 0.066, "step": 7546 }, { "epoch": 0.40235645359065947, "grad_norm": 2.9164373874664307, "learning_rate": 8.4908e-06, "loss": 0.0511, "step": 7547 }, { "epoch": 0.4024097670203124, "grad_norm": 2.499955415725708, "learning_rate": 8.490600000000001e-06, "loss": 0.0558, "step": 7548 }, { "epoch": 0.40246308044996537, "grad_norm": 3.295034885406494, "learning_rate": 8.490400000000001e-06, "loss": 0.0717, "step": 7549 }, { "epoch": 0.4025163938796183, "grad_norm": 2.651226758956909, "learning_rate": 8.4902e-06, "loss": 0.05, "step": 7550 }, { "epoch": 0.4025697073092712, "grad_norm": 1.2831372022628784, "learning_rate": 8.49e-06, "loss": 0.0172, "step": 7551 }, { "epoch": 0.40262302073892414, "grad_norm": 3.526592493057251, "learning_rate": 8.4898e-06, "loss": 0.0572, "step": 7552 }, { "epoch": 0.40267633416857707, "grad_norm": 3.3328628540039062, "learning_rate": 8.4896e-06, "loss": 0.0611, "step": 7553 }, { "epoch": 0.40272964759823, "grad_norm": 2.5465261936187744, "learning_rate": 8.4894e-06, "loss": 0.0387, "step": 7554 }, { "epoch": 0.4027829610278829, "grad_norm": 1.6341478824615479, "learning_rate": 8.4892e-06, "loss": 0.0119, "step": 7555 }, { "epoch": 0.40283627445753584, "grad_norm": 2.9040417671203613, "learning_rate": 8.489000000000001e-06, "loss": 0.0537, "step": 7556 }, { "epoch": 0.40288958788718876, "grad_norm": 2.1380908489227295, "learning_rate": 8.4888e-06, "loss": 0.0394, "step": 7557 }, { "epoch": 0.40294290131684174, "grad_norm": 2.84091854095459, "learning_rate": 8.4886e-06, "loss": 0.0301, "step": 7558 }, { "epoch": 0.40299621474649466, "grad_norm": 2.200148105621338, "learning_rate": 8.488400000000002e-06, "loss": 0.035, "step": 7559 }, { "epoch": 0.4030495281761476, "grad_norm": 4.120403289794922, "learning_rate": 8.488200000000001e-06, "loss": 0.0586, "step": 7560 }, { "epoch": 0.4031028416058005, "grad_norm": 1.3364874124526978, "learning_rate": 8.488e-06, "loss": 0.0189, "step": 7561 }, { "epoch": 0.40315615503545343, "grad_norm": 2.5132439136505127, "learning_rate": 8.4878e-06, "loss": 0.0357, "step": 7562 }, { "epoch": 0.40320946846510636, "grad_norm": 2.964599847793579, "learning_rate": 8.4876e-06, "loss": 0.0253, "step": 7563 }, { "epoch": 0.4032627818947593, "grad_norm": 3.576324462890625, "learning_rate": 8.487400000000001e-06, "loss": 0.0297, "step": 7564 }, { "epoch": 0.4033160953244122, "grad_norm": 3.787343740463257, "learning_rate": 8.4872e-06, "loss": 0.032, "step": 7565 }, { "epoch": 0.40336940875406513, "grad_norm": 2.5798630714416504, "learning_rate": 8.487e-06, "loss": 0.0222, "step": 7566 }, { "epoch": 0.40342272218371805, "grad_norm": 1.8303735256195068, "learning_rate": 8.4868e-06, "loss": 0.0153, "step": 7567 }, { "epoch": 0.40347603561337103, "grad_norm": 4.759398937225342, "learning_rate": 8.486600000000001e-06, "loss": 0.0613, "step": 7568 }, { "epoch": 0.40352934904302395, "grad_norm": 2.7887516021728516, "learning_rate": 8.4864e-06, "loss": 0.0185, "step": 7569 }, { "epoch": 0.4035826624726769, "grad_norm": 1.8057399988174438, "learning_rate": 8.486200000000002e-06, "loss": 0.027, "step": 7570 }, { "epoch": 0.4036359759023298, "grad_norm": 2.223787307739258, "learning_rate": 8.486000000000001e-06, "loss": 0.0295, "step": 7571 }, { "epoch": 0.4036892893319827, "grad_norm": 2.5009207725524902, "learning_rate": 8.485800000000001e-06, "loss": 0.0309, "step": 7572 }, { "epoch": 0.40374260276163565, "grad_norm": 1.5983688831329346, "learning_rate": 8.4856e-06, "loss": 0.0241, "step": 7573 }, { "epoch": 0.4037959161912886, "grad_norm": 2.5533344745635986, "learning_rate": 8.4854e-06, "loss": 0.0608, "step": 7574 }, { "epoch": 0.4038492296209415, "grad_norm": 3.099501848220825, "learning_rate": 8.485200000000001e-06, "loss": 0.0406, "step": 7575 }, { "epoch": 0.4039025430505944, "grad_norm": 2.1349599361419678, "learning_rate": 8.485000000000001e-06, "loss": 0.0228, "step": 7576 }, { "epoch": 0.4039558564802474, "grad_norm": 3.6282198429107666, "learning_rate": 8.4848e-06, "loss": 0.0279, "step": 7577 }, { "epoch": 0.4040091699099003, "grad_norm": 5.902492523193359, "learning_rate": 8.4846e-06, "loss": 0.0553, "step": 7578 }, { "epoch": 0.40406248333955325, "grad_norm": 4.104339122772217, "learning_rate": 8.4844e-06, "loss": 0.0199, "step": 7579 }, { "epoch": 0.40411579676920617, "grad_norm": 2.4095797538757324, "learning_rate": 8.484200000000001e-06, "loss": 0.0285, "step": 7580 }, { "epoch": 0.4041691101988591, "grad_norm": 3.64302396774292, "learning_rate": 8.484e-06, "loss": 0.0425, "step": 7581 }, { "epoch": 0.404222423628512, "grad_norm": 3.325169324874878, "learning_rate": 8.483800000000002e-06, "loss": 0.0464, "step": 7582 }, { "epoch": 0.40427573705816494, "grad_norm": 3.1761474609375, "learning_rate": 8.483600000000001e-06, "loss": 0.0538, "step": 7583 }, { "epoch": 0.40432905048781786, "grad_norm": 2.6716933250427246, "learning_rate": 8.4834e-06, "loss": 0.0244, "step": 7584 }, { "epoch": 0.4043823639174708, "grad_norm": 4.174764156341553, "learning_rate": 8.4832e-06, "loss": 0.0615, "step": 7585 }, { "epoch": 0.40443567734712377, "grad_norm": 4.179331302642822, "learning_rate": 8.483000000000002e-06, "loss": 0.0541, "step": 7586 }, { "epoch": 0.4044889907767767, "grad_norm": 4.750006198883057, "learning_rate": 8.482800000000001e-06, "loss": 0.0416, "step": 7587 }, { "epoch": 0.4045423042064296, "grad_norm": 3.8285391330718994, "learning_rate": 8.4826e-06, "loss": 0.0366, "step": 7588 }, { "epoch": 0.40459561763608254, "grad_norm": 4.7161030769348145, "learning_rate": 8.4824e-06, "loss": 0.0334, "step": 7589 }, { "epoch": 0.40464893106573546, "grad_norm": 4.9087700843811035, "learning_rate": 8.4822e-06, "loss": 0.0334, "step": 7590 }, { "epoch": 0.4047022444953884, "grad_norm": 4.953991889953613, "learning_rate": 8.482e-06, "loss": 0.0473, "step": 7591 }, { "epoch": 0.4047555579250413, "grad_norm": 7.571132183074951, "learning_rate": 8.4818e-06, "loss": 0.0433, "step": 7592 }, { "epoch": 0.40480887135469423, "grad_norm": 1.5195462703704834, "learning_rate": 8.4816e-06, "loss": 0.0259, "step": 7593 }, { "epoch": 0.40486218478434716, "grad_norm": 3.5170493125915527, "learning_rate": 8.481400000000002e-06, "loss": 0.0226, "step": 7594 }, { "epoch": 0.4049154982140001, "grad_norm": 3.071708917617798, "learning_rate": 8.481200000000001e-06, "loss": 0.0535, "step": 7595 }, { "epoch": 0.40496881164365306, "grad_norm": 2.1951398849487305, "learning_rate": 8.481e-06, "loss": 0.0311, "step": 7596 }, { "epoch": 0.405022125073306, "grad_norm": 2.1021080017089844, "learning_rate": 8.4808e-06, "loss": 0.0279, "step": 7597 }, { "epoch": 0.4050754385029589, "grad_norm": 3.1175742149353027, "learning_rate": 8.480600000000001e-06, "loss": 0.0454, "step": 7598 }, { "epoch": 0.40512875193261183, "grad_norm": 5.961543560028076, "learning_rate": 8.480400000000001e-06, "loss": 0.0439, "step": 7599 }, { "epoch": 0.40518206536226475, "grad_norm": 8.549420356750488, "learning_rate": 8.4802e-06, "loss": 0.049, "step": 7600 }, { "epoch": 0.4052353787919177, "grad_norm": 3.5810210704803467, "learning_rate": 8.48e-06, "loss": 0.0633, "step": 7601 }, { "epoch": 0.4052886922215706, "grad_norm": 4.172428131103516, "learning_rate": 8.4798e-06, "loss": 0.0461, "step": 7602 }, { "epoch": 0.4053420056512235, "grad_norm": 4.247969150543213, "learning_rate": 8.479600000000001e-06, "loss": 0.0533, "step": 7603 }, { "epoch": 0.40539531908087645, "grad_norm": 3.584381580352783, "learning_rate": 8.4794e-06, "loss": 0.0451, "step": 7604 }, { "epoch": 0.4054486325105294, "grad_norm": 2.150355339050293, "learning_rate": 8.4792e-06, "loss": 0.0204, "step": 7605 }, { "epoch": 0.40550194594018235, "grad_norm": 4.097981929779053, "learning_rate": 8.479000000000001e-06, "loss": 0.048, "step": 7606 }, { "epoch": 0.4055552593698353, "grad_norm": 3.083301305770874, "learning_rate": 8.478800000000001e-06, "loss": 0.0446, "step": 7607 }, { "epoch": 0.4056085727994882, "grad_norm": 6.513078689575195, "learning_rate": 8.4786e-06, "loss": 0.0591, "step": 7608 }, { "epoch": 0.4056618862291411, "grad_norm": 4.800381183624268, "learning_rate": 8.478400000000002e-06, "loss": 0.0608, "step": 7609 }, { "epoch": 0.40571519965879405, "grad_norm": 2.7723593711853027, "learning_rate": 8.478200000000001e-06, "loss": 0.046, "step": 7610 }, { "epoch": 0.40576851308844697, "grad_norm": 2.877606153488159, "learning_rate": 8.478e-06, "loss": 0.0249, "step": 7611 }, { "epoch": 0.4058218265180999, "grad_norm": 2.9020025730133057, "learning_rate": 8.4778e-06, "loss": 0.0281, "step": 7612 }, { "epoch": 0.4058751399477528, "grad_norm": 3.332674980163574, "learning_rate": 8.4776e-06, "loss": 0.0677, "step": 7613 }, { "epoch": 0.4059284533774058, "grad_norm": 3.285399913787842, "learning_rate": 8.477400000000001e-06, "loss": 0.0349, "step": 7614 }, { "epoch": 0.4059817668070587, "grad_norm": 3.083862066268921, "learning_rate": 8.4772e-06, "loss": 0.0353, "step": 7615 }, { "epoch": 0.40603508023671164, "grad_norm": 4.242992877960205, "learning_rate": 8.477e-06, "loss": 0.0655, "step": 7616 }, { "epoch": 0.40608839366636457, "grad_norm": 2.3190255165100098, "learning_rate": 8.4768e-06, "loss": 0.0262, "step": 7617 }, { "epoch": 0.4061417070960175, "grad_norm": 3.3702309131622314, "learning_rate": 8.476600000000001e-06, "loss": 0.0405, "step": 7618 }, { "epoch": 0.4061950205256704, "grad_norm": 3.224224090576172, "learning_rate": 8.4764e-06, "loss": 0.0489, "step": 7619 }, { "epoch": 0.40624833395532334, "grad_norm": 3.0271637439727783, "learning_rate": 8.4762e-06, "loss": 0.0296, "step": 7620 }, { "epoch": 0.40630164738497626, "grad_norm": 4.504424571990967, "learning_rate": 8.476000000000002e-06, "loss": 0.0404, "step": 7621 }, { "epoch": 0.4063549608146292, "grad_norm": 3.06742262840271, "learning_rate": 8.475800000000001e-06, "loss": 0.0341, "step": 7622 }, { "epoch": 0.4064082742442821, "grad_norm": 2.3290936946868896, "learning_rate": 8.4756e-06, "loss": 0.0167, "step": 7623 }, { "epoch": 0.4064615876739351, "grad_norm": 2.8056609630584717, "learning_rate": 8.4754e-06, "loss": 0.0557, "step": 7624 }, { "epoch": 0.406514901103588, "grad_norm": 3.502345323562622, "learning_rate": 8.475200000000001e-06, "loss": 0.0705, "step": 7625 }, { "epoch": 0.40656821453324093, "grad_norm": 2.870114803314209, "learning_rate": 8.475000000000001e-06, "loss": 0.0448, "step": 7626 }, { "epoch": 0.40662152796289386, "grad_norm": 5.543553829193115, "learning_rate": 8.4748e-06, "loss": 0.0344, "step": 7627 }, { "epoch": 0.4066748413925468, "grad_norm": 2.374009609222412, "learning_rate": 8.4746e-06, "loss": 0.0562, "step": 7628 }, { "epoch": 0.4067281548221997, "grad_norm": 4.662389755249023, "learning_rate": 8.4744e-06, "loss": 0.0226, "step": 7629 }, { "epoch": 0.40678146825185263, "grad_norm": 4.510319232940674, "learning_rate": 8.474200000000001e-06, "loss": 0.0863, "step": 7630 }, { "epoch": 0.40683478168150555, "grad_norm": 2.2524001598358154, "learning_rate": 8.474e-06, "loss": 0.0279, "step": 7631 }, { "epoch": 0.4068880951111585, "grad_norm": 2.7949535846710205, "learning_rate": 8.4738e-06, "loss": 0.0501, "step": 7632 }, { "epoch": 0.40694140854081146, "grad_norm": 4.417653560638428, "learning_rate": 8.473600000000001e-06, "loss": 0.0419, "step": 7633 }, { "epoch": 0.4069947219704644, "grad_norm": 2.5187833309173584, "learning_rate": 8.473400000000001e-06, "loss": 0.0572, "step": 7634 }, { "epoch": 0.4070480354001173, "grad_norm": 3.2338507175445557, "learning_rate": 8.4732e-06, "loss": 0.03, "step": 7635 }, { "epoch": 0.4071013488297702, "grad_norm": 2.505544662475586, "learning_rate": 8.473000000000002e-06, "loss": 0.0204, "step": 7636 }, { "epoch": 0.40715466225942315, "grad_norm": 3.058170795440674, "learning_rate": 8.472800000000001e-06, "loss": 0.0857, "step": 7637 }, { "epoch": 0.4072079756890761, "grad_norm": 1.2678256034851074, "learning_rate": 8.4726e-06, "loss": 0.0141, "step": 7638 }, { "epoch": 0.407261289118729, "grad_norm": 3.915658950805664, "learning_rate": 8.4724e-06, "loss": 0.054, "step": 7639 }, { "epoch": 0.4073146025483819, "grad_norm": 4.311903953552246, "learning_rate": 8.4722e-06, "loss": 0.0601, "step": 7640 }, { "epoch": 0.40736791597803484, "grad_norm": 1.9115508794784546, "learning_rate": 8.472e-06, "loss": 0.0209, "step": 7641 }, { "epoch": 0.4074212294076878, "grad_norm": 2.6434762477874756, "learning_rate": 8.4718e-06, "loss": 0.0453, "step": 7642 }, { "epoch": 0.40747454283734075, "grad_norm": 1.6969707012176514, "learning_rate": 8.4716e-06, "loss": 0.0203, "step": 7643 }, { "epoch": 0.40752785626699367, "grad_norm": 2.9316885471343994, "learning_rate": 8.4714e-06, "loss": 0.0369, "step": 7644 }, { "epoch": 0.4075811696966466, "grad_norm": 1.678707242012024, "learning_rate": 8.471200000000001e-06, "loss": 0.0197, "step": 7645 }, { "epoch": 0.4076344831262995, "grad_norm": 2.322242498397827, "learning_rate": 8.471e-06, "loss": 0.047, "step": 7646 }, { "epoch": 0.40768779655595244, "grad_norm": 4.613234043121338, "learning_rate": 8.470800000000002e-06, "loss": 0.0604, "step": 7647 }, { "epoch": 0.40774110998560537, "grad_norm": 3.9466910362243652, "learning_rate": 8.470600000000002e-06, "loss": 0.0735, "step": 7648 }, { "epoch": 0.4077944234152583, "grad_norm": 3.724201202392578, "learning_rate": 8.470400000000001e-06, "loss": 0.0411, "step": 7649 }, { "epoch": 0.4078477368449112, "grad_norm": 1.9708123207092285, "learning_rate": 8.4702e-06, "loss": 0.0182, "step": 7650 }, { "epoch": 0.40790105027456414, "grad_norm": 2.2658233642578125, "learning_rate": 8.47e-06, "loss": 0.0304, "step": 7651 }, { "epoch": 0.4079543637042171, "grad_norm": 3.9704880714416504, "learning_rate": 8.4698e-06, "loss": 0.0553, "step": 7652 }, { "epoch": 0.40800767713387004, "grad_norm": 4.575718879699707, "learning_rate": 8.469600000000001e-06, "loss": 0.0706, "step": 7653 }, { "epoch": 0.40806099056352296, "grad_norm": 4.403769493103027, "learning_rate": 8.4694e-06, "loss": 0.048, "step": 7654 }, { "epoch": 0.4081143039931759, "grad_norm": 3.1288156509399414, "learning_rate": 8.4692e-06, "loss": 0.0351, "step": 7655 }, { "epoch": 0.4081676174228288, "grad_norm": 5.210864543914795, "learning_rate": 8.469e-06, "loss": 0.0256, "step": 7656 }, { "epoch": 0.40822093085248173, "grad_norm": 3.1578478813171387, "learning_rate": 8.468800000000001e-06, "loss": 0.0596, "step": 7657 }, { "epoch": 0.40827424428213466, "grad_norm": 1.9585038423538208, "learning_rate": 8.4686e-06, "loss": 0.0178, "step": 7658 }, { "epoch": 0.4083275577117876, "grad_norm": 4.921911239624023, "learning_rate": 8.468400000000002e-06, "loss": 0.0506, "step": 7659 }, { "epoch": 0.4083808711414405, "grad_norm": 2.653731107711792, "learning_rate": 8.468200000000001e-06, "loss": 0.0403, "step": 7660 }, { "epoch": 0.4084341845710935, "grad_norm": 2.031127691268921, "learning_rate": 8.468000000000001e-06, "loss": 0.0472, "step": 7661 }, { "epoch": 0.4084874980007464, "grad_norm": 3.8189697265625, "learning_rate": 8.4678e-06, "loss": 0.0599, "step": 7662 }, { "epoch": 0.40854081143039933, "grad_norm": 2.53964900970459, "learning_rate": 8.4676e-06, "loss": 0.0181, "step": 7663 }, { "epoch": 0.40859412486005225, "grad_norm": 2.070650339126587, "learning_rate": 8.467400000000001e-06, "loss": 0.036, "step": 7664 }, { "epoch": 0.4086474382897052, "grad_norm": 3.888676881790161, "learning_rate": 8.467200000000001e-06, "loss": 0.0486, "step": 7665 }, { "epoch": 0.4087007517193581, "grad_norm": 3.369110584259033, "learning_rate": 8.467e-06, "loss": 0.0428, "step": 7666 }, { "epoch": 0.408754065149011, "grad_norm": 3.1094212532043457, "learning_rate": 8.4668e-06, "loss": 0.0467, "step": 7667 }, { "epoch": 0.40880737857866395, "grad_norm": 1.5333404541015625, "learning_rate": 8.4666e-06, "loss": 0.018, "step": 7668 }, { "epoch": 0.4088606920083169, "grad_norm": 3.8557803630828857, "learning_rate": 8.4664e-06, "loss": 0.0298, "step": 7669 }, { "epoch": 0.40891400543796985, "grad_norm": 3.285351037979126, "learning_rate": 8.4662e-06, "loss": 0.0568, "step": 7670 }, { "epoch": 0.4089673188676228, "grad_norm": 2.134697198867798, "learning_rate": 8.466000000000002e-06, "loss": 0.0263, "step": 7671 }, { "epoch": 0.4090206322972757, "grad_norm": 3.367800235748291, "learning_rate": 8.465800000000001e-06, "loss": 0.0377, "step": 7672 }, { "epoch": 0.4090739457269286, "grad_norm": 2.0130138397216797, "learning_rate": 8.4656e-06, "loss": 0.0179, "step": 7673 }, { "epoch": 0.40912725915658155, "grad_norm": 3.125795602798462, "learning_rate": 8.4654e-06, "loss": 0.0476, "step": 7674 }, { "epoch": 0.40918057258623447, "grad_norm": 4.693226337432861, "learning_rate": 8.465200000000002e-06, "loss": 0.041, "step": 7675 }, { "epoch": 0.4092338860158874, "grad_norm": 2.9351677894592285, "learning_rate": 8.465000000000001e-06, "loss": 0.0701, "step": 7676 }, { "epoch": 0.4092871994455403, "grad_norm": 1.1801775693893433, "learning_rate": 8.4648e-06, "loss": 0.0186, "step": 7677 }, { "epoch": 0.40934051287519324, "grad_norm": 3.392892599105835, "learning_rate": 8.4646e-06, "loss": 0.0637, "step": 7678 }, { "epoch": 0.40939382630484616, "grad_norm": 3.5793914794921875, "learning_rate": 8.4644e-06, "loss": 0.0499, "step": 7679 }, { "epoch": 0.40944713973449914, "grad_norm": 1.6430442333221436, "learning_rate": 8.4642e-06, "loss": 0.0263, "step": 7680 }, { "epoch": 0.40950045316415207, "grad_norm": 3.487586736679077, "learning_rate": 8.464e-06, "loss": 0.0613, "step": 7681 }, { "epoch": 0.409553766593805, "grad_norm": 2.2815816402435303, "learning_rate": 8.4638e-06, "loss": 0.0345, "step": 7682 }, { "epoch": 0.4096070800234579, "grad_norm": 4.304603099822998, "learning_rate": 8.463600000000001e-06, "loss": 0.0687, "step": 7683 }, { "epoch": 0.40966039345311084, "grad_norm": 2.7134735584259033, "learning_rate": 8.463400000000001e-06, "loss": 0.0381, "step": 7684 }, { "epoch": 0.40971370688276376, "grad_norm": 4.263282775878906, "learning_rate": 8.4632e-06, "loss": 0.0441, "step": 7685 }, { "epoch": 0.4097670203124167, "grad_norm": 2.2912449836730957, "learning_rate": 8.463000000000002e-06, "loss": 0.0222, "step": 7686 }, { "epoch": 0.4098203337420696, "grad_norm": 2.3904783725738525, "learning_rate": 8.462800000000001e-06, "loss": 0.0509, "step": 7687 }, { "epoch": 0.40987364717172253, "grad_norm": 2.707188129425049, "learning_rate": 8.462600000000001e-06, "loss": 0.0321, "step": 7688 }, { "epoch": 0.4099269606013755, "grad_norm": 3.0539894104003906, "learning_rate": 8.4624e-06, "loss": 0.0552, "step": 7689 }, { "epoch": 0.40998027403102844, "grad_norm": 1.6461820602416992, "learning_rate": 8.4622e-06, "loss": 0.0242, "step": 7690 }, { "epoch": 0.41003358746068136, "grad_norm": 4.293005466461182, "learning_rate": 8.462e-06, "loss": 0.0268, "step": 7691 }, { "epoch": 0.4100869008903343, "grad_norm": 6.109586238861084, "learning_rate": 8.461800000000001e-06, "loss": 0.0392, "step": 7692 }, { "epoch": 0.4101402143199872, "grad_norm": 2.7371883392333984, "learning_rate": 8.4616e-06, "loss": 0.0608, "step": 7693 }, { "epoch": 0.41019352774964013, "grad_norm": 3.537766218185425, "learning_rate": 8.4614e-06, "loss": 0.0532, "step": 7694 }, { "epoch": 0.41024684117929305, "grad_norm": 2.591531276702881, "learning_rate": 8.461200000000001e-06, "loss": 0.0418, "step": 7695 }, { "epoch": 0.410300154608946, "grad_norm": 2.6208693981170654, "learning_rate": 8.461e-06, "loss": 0.0494, "step": 7696 }, { "epoch": 0.4103534680385989, "grad_norm": 2.2025976181030273, "learning_rate": 8.460800000000002e-06, "loss": 0.0338, "step": 7697 }, { "epoch": 0.4104067814682519, "grad_norm": 5.468449592590332, "learning_rate": 8.460600000000002e-06, "loss": 0.0645, "step": 7698 }, { "epoch": 0.4104600948979048, "grad_norm": 4.152406692504883, "learning_rate": 8.460400000000001e-06, "loss": 0.0859, "step": 7699 }, { "epoch": 0.4105134083275577, "grad_norm": 2.352048397064209, "learning_rate": 8.4602e-06, "loss": 0.032, "step": 7700 }, { "epoch": 0.41056672175721065, "grad_norm": 3.8644349575042725, "learning_rate": 8.46e-06, "loss": 0.0333, "step": 7701 }, { "epoch": 0.4106200351868636, "grad_norm": 3.0326600074768066, "learning_rate": 8.4598e-06, "loss": 0.0452, "step": 7702 }, { "epoch": 0.4106733486165165, "grad_norm": 2.092050790786743, "learning_rate": 8.459600000000001e-06, "loss": 0.0353, "step": 7703 }, { "epoch": 0.4107266620461694, "grad_norm": 4.054507732391357, "learning_rate": 8.4594e-06, "loss": 0.0293, "step": 7704 }, { "epoch": 0.41077997547582235, "grad_norm": 3.921717405319214, "learning_rate": 8.4592e-06, "loss": 0.0304, "step": 7705 }, { "epoch": 0.41083328890547527, "grad_norm": 1.9770002365112305, "learning_rate": 8.459e-06, "loss": 0.0392, "step": 7706 }, { "epoch": 0.4108866023351282, "grad_norm": 1.7453806400299072, "learning_rate": 8.458800000000001e-06, "loss": 0.0578, "step": 7707 }, { "epoch": 0.41093991576478117, "grad_norm": 2.919088840484619, "learning_rate": 8.4586e-06, "loss": 0.0489, "step": 7708 }, { "epoch": 0.4109932291944341, "grad_norm": 3.4872493743896484, "learning_rate": 8.458400000000002e-06, "loss": 0.0444, "step": 7709 }, { "epoch": 0.411046542624087, "grad_norm": 2.184343099594116, "learning_rate": 8.458200000000001e-06, "loss": 0.0303, "step": 7710 }, { "epoch": 0.41109985605373994, "grad_norm": 3.021857738494873, "learning_rate": 8.458000000000001e-06, "loss": 0.046, "step": 7711 }, { "epoch": 0.41115316948339287, "grad_norm": 2.8732895851135254, "learning_rate": 8.4578e-06, "loss": 0.0628, "step": 7712 }, { "epoch": 0.4112064829130458, "grad_norm": 2.2726454734802246, "learning_rate": 8.4576e-06, "loss": 0.0196, "step": 7713 }, { "epoch": 0.4112597963426987, "grad_norm": 3.0209054946899414, "learning_rate": 8.457400000000001e-06, "loss": 0.0527, "step": 7714 }, { "epoch": 0.41131310977235164, "grad_norm": 3.2641761302948, "learning_rate": 8.457200000000001e-06, "loss": 0.0535, "step": 7715 }, { "epoch": 0.41136642320200456, "grad_norm": 2.4824130535125732, "learning_rate": 8.457e-06, "loss": 0.0417, "step": 7716 }, { "epoch": 0.41141973663165754, "grad_norm": 2.354888439178467, "learning_rate": 8.4568e-06, "loss": 0.027, "step": 7717 }, { "epoch": 0.41147305006131046, "grad_norm": 4.6057538986206055, "learning_rate": 8.4566e-06, "loss": 0.0546, "step": 7718 }, { "epoch": 0.4115263634909634, "grad_norm": 3.2920033931732178, "learning_rate": 8.456400000000001e-06, "loss": 0.0241, "step": 7719 }, { "epoch": 0.4115796769206163, "grad_norm": 4.215709209442139, "learning_rate": 8.4562e-06, "loss": 0.0571, "step": 7720 }, { "epoch": 0.41163299035026923, "grad_norm": 2.655855178833008, "learning_rate": 8.456000000000002e-06, "loss": 0.0422, "step": 7721 }, { "epoch": 0.41168630377992216, "grad_norm": 2.422792434692383, "learning_rate": 8.455800000000001e-06, "loss": 0.0276, "step": 7722 }, { "epoch": 0.4117396172095751, "grad_norm": 4.116455554962158, "learning_rate": 8.4556e-06, "loss": 0.0688, "step": 7723 }, { "epoch": 0.411792930639228, "grad_norm": 3.239720582962036, "learning_rate": 8.4554e-06, "loss": 0.0381, "step": 7724 }, { "epoch": 0.41184624406888093, "grad_norm": 2.6702229976654053, "learning_rate": 8.455200000000002e-06, "loss": 0.0505, "step": 7725 }, { "epoch": 0.4118995574985339, "grad_norm": 1.8422267436981201, "learning_rate": 8.455000000000001e-06, "loss": 0.0146, "step": 7726 }, { "epoch": 0.41195287092818683, "grad_norm": 1.587646484375, "learning_rate": 8.4548e-06, "loss": 0.0247, "step": 7727 }, { "epoch": 0.41200618435783976, "grad_norm": 3.6893491744995117, "learning_rate": 8.4546e-06, "loss": 0.0611, "step": 7728 }, { "epoch": 0.4120594977874927, "grad_norm": 4.036632061004639, "learning_rate": 8.4544e-06, "loss": 0.0473, "step": 7729 }, { "epoch": 0.4121128112171456, "grad_norm": 2.557845115661621, "learning_rate": 8.4542e-06, "loss": 0.0264, "step": 7730 }, { "epoch": 0.4121661246467985, "grad_norm": 3.7333390712738037, "learning_rate": 8.454e-06, "loss": 0.0407, "step": 7731 }, { "epoch": 0.41221943807645145, "grad_norm": 9.322192192077637, "learning_rate": 8.4538e-06, "loss": 0.0684, "step": 7732 }, { "epoch": 0.4122727515061044, "grad_norm": 2.976118803024292, "learning_rate": 8.453600000000002e-06, "loss": 0.0225, "step": 7733 }, { "epoch": 0.4123260649357573, "grad_norm": 3.440847158432007, "learning_rate": 8.453400000000001e-06, "loss": 0.0225, "step": 7734 }, { "epoch": 0.4123793783654102, "grad_norm": 2.2982168197631836, "learning_rate": 8.4532e-06, "loss": 0.0237, "step": 7735 }, { "epoch": 0.4124326917950632, "grad_norm": 3.802610397338867, "learning_rate": 8.453000000000002e-06, "loss": 0.0394, "step": 7736 }, { "epoch": 0.4124860052247161, "grad_norm": 3.4334399700164795, "learning_rate": 8.452800000000001e-06, "loss": 0.0261, "step": 7737 }, { "epoch": 0.41253931865436905, "grad_norm": 4.782835006713867, "learning_rate": 8.452600000000001e-06, "loss": 0.0601, "step": 7738 }, { "epoch": 0.41259263208402197, "grad_norm": 3.3811137676239014, "learning_rate": 8.4524e-06, "loss": 0.0345, "step": 7739 }, { "epoch": 0.4126459455136749, "grad_norm": 1.9481265544891357, "learning_rate": 8.4522e-06, "loss": 0.0197, "step": 7740 }, { "epoch": 0.4126992589433278, "grad_norm": 3.653383255004883, "learning_rate": 8.452e-06, "loss": 0.0269, "step": 7741 }, { "epoch": 0.41275257237298074, "grad_norm": 2.0103065967559814, "learning_rate": 8.451800000000001e-06, "loss": 0.0159, "step": 7742 }, { "epoch": 0.41280588580263367, "grad_norm": 5.016267776489258, "learning_rate": 8.4516e-06, "loss": 0.0641, "step": 7743 }, { "epoch": 0.4128591992322866, "grad_norm": 3.3119006156921387, "learning_rate": 8.4514e-06, "loss": 0.0324, "step": 7744 }, { "epoch": 0.41291251266193957, "grad_norm": 2.0641250610351562, "learning_rate": 8.451200000000001e-06, "loss": 0.0214, "step": 7745 }, { "epoch": 0.4129658260915925, "grad_norm": 3.0096006393432617, "learning_rate": 8.451000000000001e-06, "loss": 0.0323, "step": 7746 }, { "epoch": 0.4130191395212454, "grad_norm": 5.398148536682129, "learning_rate": 8.4508e-06, "loss": 0.0504, "step": 7747 }, { "epoch": 0.41307245295089834, "grad_norm": 4.341707229614258, "learning_rate": 8.450600000000002e-06, "loss": 0.0515, "step": 7748 }, { "epoch": 0.41312576638055126, "grad_norm": 2.722750425338745, "learning_rate": 8.450400000000001e-06, "loss": 0.0359, "step": 7749 }, { "epoch": 0.4131790798102042, "grad_norm": 2.712559461593628, "learning_rate": 8.450200000000001e-06, "loss": 0.0399, "step": 7750 }, { "epoch": 0.4132323932398571, "grad_norm": 2.1184122562408447, "learning_rate": 8.45e-06, "loss": 0.015, "step": 7751 }, { "epoch": 0.41328570666951003, "grad_norm": 3.0808980464935303, "learning_rate": 8.4498e-06, "loss": 0.0537, "step": 7752 }, { "epoch": 0.41333902009916296, "grad_norm": 3.449032783508301, "learning_rate": 8.449600000000001e-06, "loss": 0.0461, "step": 7753 }, { "epoch": 0.41339233352881594, "grad_norm": 4.485919952392578, "learning_rate": 8.4494e-06, "loss": 0.0525, "step": 7754 }, { "epoch": 0.41344564695846886, "grad_norm": 4.72426700592041, "learning_rate": 8.4492e-06, "loss": 0.0413, "step": 7755 }, { "epoch": 0.4134989603881218, "grad_norm": 3.5460293292999268, "learning_rate": 8.449e-06, "loss": 0.0494, "step": 7756 }, { "epoch": 0.4135522738177747, "grad_norm": 2.203171491622925, "learning_rate": 8.448800000000001e-06, "loss": 0.0237, "step": 7757 }, { "epoch": 0.41360558724742763, "grad_norm": 2.48715877532959, "learning_rate": 8.4486e-06, "loss": 0.0554, "step": 7758 }, { "epoch": 0.41365890067708055, "grad_norm": 2.940314531326294, "learning_rate": 8.4484e-06, "loss": 0.056, "step": 7759 }, { "epoch": 0.4137122141067335, "grad_norm": 3.5700223445892334, "learning_rate": 8.448200000000002e-06, "loss": 0.0412, "step": 7760 }, { "epoch": 0.4137655275363864, "grad_norm": 3.884673833847046, "learning_rate": 8.448000000000001e-06, "loss": 0.0675, "step": 7761 }, { "epoch": 0.4138188409660393, "grad_norm": 3.2364299297332764, "learning_rate": 8.4478e-06, "loss": 0.0256, "step": 7762 }, { "epoch": 0.41387215439569225, "grad_norm": 1.8331873416900635, "learning_rate": 8.4476e-06, "loss": 0.046, "step": 7763 }, { "epoch": 0.41392546782534523, "grad_norm": 3.8625295162200928, "learning_rate": 8.447400000000001e-06, "loss": 0.0305, "step": 7764 }, { "epoch": 0.41397878125499815, "grad_norm": 2.4075369834899902, "learning_rate": 8.447200000000001e-06, "loss": 0.0216, "step": 7765 }, { "epoch": 0.4140320946846511, "grad_norm": 3.909663438796997, "learning_rate": 8.447e-06, "loss": 0.0347, "step": 7766 }, { "epoch": 0.414085408114304, "grad_norm": 2.4599857330322266, "learning_rate": 8.4468e-06, "loss": 0.0575, "step": 7767 }, { "epoch": 0.4141387215439569, "grad_norm": 5.56900691986084, "learning_rate": 8.4466e-06, "loss": 0.0582, "step": 7768 }, { "epoch": 0.41419203497360985, "grad_norm": 5.384970664978027, "learning_rate": 8.446400000000001e-06, "loss": 0.0771, "step": 7769 }, { "epoch": 0.41424534840326277, "grad_norm": 3.7255020141601562, "learning_rate": 8.4462e-06, "loss": 0.039, "step": 7770 }, { "epoch": 0.4142986618329157, "grad_norm": 4.266979694366455, "learning_rate": 8.446e-06, "loss": 0.0435, "step": 7771 }, { "epoch": 0.4143519752625686, "grad_norm": 4.249077796936035, "learning_rate": 8.445800000000001e-06, "loss": 0.0264, "step": 7772 }, { "epoch": 0.4144052886922216, "grad_norm": 4.429710865020752, "learning_rate": 8.445600000000001e-06, "loss": 0.0859, "step": 7773 }, { "epoch": 0.4144586021218745, "grad_norm": 1.7304836511611938, "learning_rate": 8.4454e-06, "loss": 0.0111, "step": 7774 }, { "epoch": 0.41451191555152744, "grad_norm": 3.105933666229248, "learning_rate": 8.445200000000002e-06, "loss": 0.0473, "step": 7775 }, { "epoch": 0.41456522898118037, "grad_norm": 2.3621826171875, "learning_rate": 8.445000000000001e-06, "loss": 0.0341, "step": 7776 }, { "epoch": 0.4146185424108333, "grad_norm": 6.4712748527526855, "learning_rate": 8.444800000000001e-06, "loss": 0.0306, "step": 7777 }, { "epoch": 0.4146718558404862, "grad_norm": 2.310739517211914, "learning_rate": 8.4446e-06, "loss": 0.0451, "step": 7778 }, { "epoch": 0.41472516927013914, "grad_norm": 4.125908374786377, "learning_rate": 8.4444e-06, "loss": 0.0273, "step": 7779 }, { "epoch": 0.41477848269979206, "grad_norm": 4.029835224151611, "learning_rate": 8.4442e-06, "loss": 0.0364, "step": 7780 }, { "epoch": 0.414831796129445, "grad_norm": 2.2481021881103516, "learning_rate": 8.444e-06, "loss": 0.0346, "step": 7781 }, { "epoch": 0.41488510955909796, "grad_norm": 4.733692646026611, "learning_rate": 8.4438e-06, "loss": 0.0439, "step": 7782 }, { "epoch": 0.4149384229887509, "grad_norm": 1.4865578413009644, "learning_rate": 8.4436e-06, "loss": 0.0143, "step": 7783 }, { "epoch": 0.4149917364184038, "grad_norm": 1.013290524482727, "learning_rate": 8.443400000000001e-06, "loss": 0.0099, "step": 7784 }, { "epoch": 0.41504504984805674, "grad_norm": 3.3360302448272705, "learning_rate": 8.4432e-06, "loss": 0.0607, "step": 7785 }, { "epoch": 0.41509836327770966, "grad_norm": 2.252037525177002, "learning_rate": 8.443000000000002e-06, "loss": 0.0165, "step": 7786 }, { "epoch": 0.4151516767073626, "grad_norm": 1.6274851560592651, "learning_rate": 8.442800000000002e-06, "loss": 0.0082, "step": 7787 }, { "epoch": 0.4152049901370155, "grad_norm": 3.2876501083374023, "learning_rate": 8.442600000000001e-06, "loss": 0.062, "step": 7788 }, { "epoch": 0.41525830356666843, "grad_norm": 2.1480906009674072, "learning_rate": 8.4424e-06, "loss": 0.0513, "step": 7789 }, { "epoch": 0.41531161699632135, "grad_norm": 4.9319915771484375, "learning_rate": 8.4422e-06, "loss": 0.0245, "step": 7790 }, { "epoch": 0.4153649304259743, "grad_norm": 3.4883780479431152, "learning_rate": 8.442e-06, "loss": 0.0275, "step": 7791 }, { "epoch": 0.41541824385562726, "grad_norm": 1.922278642654419, "learning_rate": 8.441800000000001e-06, "loss": 0.0164, "step": 7792 }, { "epoch": 0.4154715572852802, "grad_norm": 5.136935234069824, "learning_rate": 8.4416e-06, "loss": 0.079, "step": 7793 }, { "epoch": 0.4155248707149331, "grad_norm": 3.783151388168335, "learning_rate": 8.4414e-06, "loss": 0.0218, "step": 7794 }, { "epoch": 0.415578184144586, "grad_norm": 3.508596897125244, "learning_rate": 8.4412e-06, "loss": 0.0476, "step": 7795 }, { "epoch": 0.41563149757423895, "grad_norm": 3.9171643257141113, "learning_rate": 8.441000000000001e-06, "loss": 0.0339, "step": 7796 }, { "epoch": 0.4156848110038919, "grad_norm": 2.580143451690674, "learning_rate": 8.4408e-06, "loss": 0.0164, "step": 7797 }, { "epoch": 0.4157381244335448, "grad_norm": 2.931252956390381, "learning_rate": 8.440600000000002e-06, "loss": 0.0265, "step": 7798 }, { "epoch": 0.4157914378631977, "grad_norm": 3.4580798149108887, "learning_rate": 8.440400000000001e-06, "loss": 0.0578, "step": 7799 }, { "epoch": 0.41584475129285065, "grad_norm": 2.4764819145202637, "learning_rate": 8.440200000000001e-06, "loss": 0.0751, "step": 7800 }, { "epoch": 0.4158980647225036, "grad_norm": 3.252117395401001, "learning_rate": 8.44e-06, "loss": 0.0516, "step": 7801 }, { "epoch": 0.41595137815215655, "grad_norm": 2.2668464183807373, "learning_rate": 8.4398e-06, "loss": 0.0357, "step": 7802 }, { "epoch": 0.41600469158180947, "grad_norm": 1.3508472442626953, "learning_rate": 8.439600000000001e-06, "loss": 0.0195, "step": 7803 }, { "epoch": 0.4160580050114624, "grad_norm": 1.066972017288208, "learning_rate": 8.439400000000001e-06, "loss": 0.0261, "step": 7804 }, { "epoch": 0.4161113184411153, "grad_norm": 3.6885743141174316, "learning_rate": 8.4392e-06, "loss": 0.0424, "step": 7805 }, { "epoch": 0.41616463187076824, "grad_norm": 2.880436658859253, "learning_rate": 8.439e-06, "loss": 0.0322, "step": 7806 }, { "epoch": 0.41621794530042117, "grad_norm": 2.8943541049957275, "learning_rate": 8.4388e-06, "loss": 0.0262, "step": 7807 }, { "epoch": 0.4162712587300741, "grad_norm": 3.193936586380005, "learning_rate": 8.4386e-06, "loss": 0.0601, "step": 7808 }, { "epoch": 0.416324572159727, "grad_norm": 2.4927730560302734, "learning_rate": 8.4384e-06, "loss": 0.0287, "step": 7809 }, { "epoch": 0.41637788558938, "grad_norm": 2.860858917236328, "learning_rate": 8.438200000000002e-06, "loss": 0.0759, "step": 7810 }, { "epoch": 0.4164311990190329, "grad_norm": 2.0255210399627686, "learning_rate": 8.438000000000001e-06, "loss": 0.0325, "step": 7811 }, { "epoch": 0.41648451244868584, "grad_norm": 3.8515546321868896, "learning_rate": 8.4378e-06, "loss": 0.0404, "step": 7812 }, { "epoch": 0.41653782587833876, "grad_norm": 7.494121074676514, "learning_rate": 8.4376e-06, "loss": 0.0711, "step": 7813 }, { "epoch": 0.4165911393079917, "grad_norm": 3.575632095336914, "learning_rate": 8.437400000000002e-06, "loss": 0.0246, "step": 7814 }, { "epoch": 0.4166444527376446, "grad_norm": 1.9394385814666748, "learning_rate": 8.437200000000001e-06, "loss": 0.0135, "step": 7815 }, { "epoch": 0.41669776616729753, "grad_norm": 2.5161564350128174, "learning_rate": 8.437e-06, "loss": 0.0353, "step": 7816 }, { "epoch": 0.41675107959695046, "grad_norm": 2.571870803833008, "learning_rate": 8.4368e-06, "loss": 0.0516, "step": 7817 }, { "epoch": 0.4168043930266034, "grad_norm": 2.0696775913238525, "learning_rate": 8.4366e-06, "loss": 0.0304, "step": 7818 }, { "epoch": 0.4168577064562563, "grad_norm": 2.656259775161743, "learning_rate": 8.4364e-06, "loss": 0.0194, "step": 7819 }, { "epoch": 0.4169110198859093, "grad_norm": 2.8253934383392334, "learning_rate": 8.4362e-06, "loss": 0.0436, "step": 7820 }, { "epoch": 0.4169643333155622, "grad_norm": 3.7086925506591797, "learning_rate": 8.436e-06, "loss": 0.0301, "step": 7821 }, { "epoch": 0.41701764674521513, "grad_norm": 1.8137322664260864, "learning_rate": 8.435800000000001e-06, "loss": 0.0366, "step": 7822 }, { "epoch": 0.41707096017486806, "grad_norm": 3.242173194885254, "learning_rate": 8.435600000000001e-06, "loss": 0.0512, "step": 7823 }, { "epoch": 0.417124273604521, "grad_norm": 4.320403099060059, "learning_rate": 8.4354e-06, "loss": 0.066, "step": 7824 }, { "epoch": 0.4171775870341739, "grad_norm": 2.932511568069458, "learning_rate": 8.435200000000002e-06, "loss": 0.0622, "step": 7825 }, { "epoch": 0.4172309004638268, "grad_norm": 1.8304470777511597, "learning_rate": 8.435000000000001e-06, "loss": 0.031, "step": 7826 }, { "epoch": 0.41728421389347975, "grad_norm": 1.9936736822128296, "learning_rate": 8.434800000000001e-06, "loss": 0.0494, "step": 7827 }, { "epoch": 0.4173375273231327, "grad_norm": 2.8124098777770996, "learning_rate": 8.4346e-06, "loss": 0.0308, "step": 7828 }, { "epoch": 0.41739084075278565, "grad_norm": 3.508223056793213, "learning_rate": 8.4344e-06, "loss": 0.0545, "step": 7829 }, { "epoch": 0.4174441541824386, "grad_norm": 2.1041524410247803, "learning_rate": 8.4342e-06, "loss": 0.0206, "step": 7830 }, { "epoch": 0.4174974676120915, "grad_norm": 3.483402967453003, "learning_rate": 8.434000000000001e-06, "loss": 0.0287, "step": 7831 }, { "epoch": 0.4175507810417444, "grad_norm": 1.6374940872192383, "learning_rate": 8.4338e-06, "loss": 0.0153, "step": 7832 }, { "epoch": 0.41760409447139735, "grad_norm": 2.0913708209991455, "learning_rate": 8.4336e-06, "loss": 0.0365, "step": 7833 }, { "epoch": 0.41765740790105027, "grad_norm": 3.282911777496338, "learning_rate": 8.433400000000001e-06, "loss": 0.0557, "step": 7834 }, { "epoch": 0.4177107213307032, "grad_norm": 3.841449022293091, "learning_rate": 8.4332e-06, "loss": 0.0489, "step": 7835 }, { "epoch": 0.4177640347603561, "grad_norm": 2.2611544132232666, "learning_rate": 8.433e-06, "loss": 0.0246, "step": 7836 }, { "epoch": 0.41781734819000904, "grad_norm": 3.0660693645477295, "learning_rate": 8.432800000000002e-06, "loss": 0.0308, "step": 7837 }, { "epoch": 0.417870661619662, "grad_norm": 1.4999403953552246, "learning_rate": 8.432600000000001e-06, "loss": 0.0167, "step": 7838 }, { "epoch": 0.41792397504931494, "grad_norm": 2.858285903930664, "learning_rate": 8.4324e-06, "loss": 0.0701, "step": 7839 }, { "epoch": 0.41797728847896787, "grad_norm": 3.6796202659606934, "learning_rate": 8.4322e-06, "loss": 0.0431, "step": 7840 }, { "epoch": 0.4180306019086208, "grad_norm": 2.131589889526367, "learning_rate": 8.432e-06, "loss": 0.0283, "step": 7841 }, { "epoch": 0.4180839153382737, "grad_norm": 2.4386990070343018, "learning_rate": 8.431800000000001e-06, "loss": 0.0244, "step": 7842 }, { "epoch": 0.41813722876792664, "grad_norm": 4.125922203063965, "learning_rate": 8.4316e-06, "loss": 0.0828, "step": 7843 }, { "epoch": 0.41819054219757956, "grad_norm": 1.1911369562149048, "learning_rate": 8.4314e-06, "loss": 0.0099, "step": 7844 }, { "epoch": 0.4182438556272325, "grad_norm": 2.981415271759033, "learning_rate": 8.4312e-06, "loss": 0.0347, "step": 7845 }, { "epoch": 0.4182971690568854, "grad_norm": 4.077301979064941, "learning_rate": 8.431000000000001e-06, "loss": 0.0423, "step": 7846 }, { "epoch": 0.41835048248653833, "grad_norm": 3.026550531387329, "learning_rate": 8.4308e-06, "loss": 0.0434, "step": 7847 }, { "epoch": 0.4184037959161913, "grad_norm": 3.6949501037597656, "learning_rate": 8.4306e-06, "loss": 0.0313, "step": 7848 }, { "epoch": 0.41845710934584424, "grad_norm": 3.491511344909668, "learning_rate": 8.430400000000001e-06, "loss": 0.0416, "step": 7849 }, { "epoch": 0.41851042277549716, "grad_norm": 1.0943121910095215, "learning_rate": 8.430200000000001e-06, "loss": 0.0072, "step": 7850 }, { "epoch": 0.4185637362051501, "grad_norm": 2.230595827102661, "learning_rate": 8.43e-06, "loss": 0.0531, "step": 7851 }, { "epoch": 0.418617049634803, "grad_norm": 3.4190211296081543, "learning_rate": 8.4298e-06, "loss": 0.0298, "step": 7852 }, { "epoch": 0.41867036306445593, "grad_norm": 2.3817830085754395, "learning_rate": 8.429600000000001e-06, "loss": 0.0286, "step": 7853 }, { "epoch": 0.41872367649410885, "grad_norm": 1.3679813146591187, "learning_rate": 8.429400000000001e-06, "loss": 0.0178, "step": 7854 }, { "epoch": 0.4187769899237618, "grad_norm": 5.245944499969482, "learning_rate": 8.4292e-06, "loss": 0.0394, "step": 7855 }, { "epoch": 0.4188303033534147, "grad_norm": 2.3984789848327637, "learning_rate": 8.429e-06, "loss": 0.0193, "step": 7856 }, { "epoch": 0.4188836167830677, "grad_norm": 2.9933080673217773, "learning_rate": 8.4288e-06, "loss": 0.047, "step": 7857 }, { "epoch": 0.4189369302127206, "grad_norm": 4.526039123535156, "learning_rate": 8.428600000000001e-06, "loss": 0.0504, "step": 7858 }, { "epoch": 0.4189902436423735, "grad_norm": 1.7205374240875244, "learning_rate": 8.4284e-06, "loss": 0.0212, "step": 7859 }, { "epoch": 0.41904355707202645, "grad_norm": 2.9856083393096924, "learning_rate": 8.4282e-06, "loss": 0.0401, "step": 7860 }, { "epoch": 0.4190968705016794, "grad_norm": 7.274533271789551, "learning_rate": 8.428000000000001e-06, "loss": 0.0693, "step": 7861 }, { "epoch": 0.4191501839313323, "grad_norm": 4.427825450897217, "learning_rate": 8.427800000000001e-06, "loss": 0.0257, "step": 7862 }, { "epoch": 0.4192034973609852, "grad_norm": 2.637855291366577, "learning_rate": 8.4276e-06, "loss": 0.028, "step": 7863 }, { "epoch": 0.41925681079063815, "grad_norm": 2.2252752780914307, "learning_rate": 8.427400000000002e-06, "loss": 0.0155, "step": 7864 }, { "epoch": 0.41931012422029107, "grad_norm": 3.5833401679992676, "learning_rate": 8.427200000000001e-06, "loss": 0.0436, "step": 7865 }, { "epoch": 0.41936343764994405, "grad_norm": 3.8068463802337646, "learning_rate": 8.427e-06, "loss": 0.0599, "step": 7866 }, { "epoch": 0.41941675107959697, "grad_norm": 1.4938035011291504, "learning_rate": 8.4268e-06, "loss": 0.0134, "step": 7867 }, { "epoch": 0.4194700645092499, "grad_norm": 2.1173157691955566, "learning_rate": 8.4266e-06, "loss": 0.0342, "step": 7868 }, { "epoch": 0.4195233779389028, "grad_norm": 4.096808433532715, "learning_rate": 8.4264e-06, "loss": 0.0673, "step": 7869 }, { "epoch": 0.41957669136855574, "grad_norm": 1.9429842233657837, "learning_rate": 8.4262e-06, "loss": 0.0236, "step": 7870 }, { "epoch": 0.41963000479820867, "grad_norm": 2.7405385971069336, "learning_rate": 8.426e-06, "loss": 0.0221, "step": 7871 }, { "epoch": 0.4196833182278616, "grad_norm": 6.9729180335998535, "learning_rate": 8.4258e-06, "loss": 0.0504, "step": 7872 }, { "epoch": 0.4197366316575145, "grad_norm": 2.385256290435791, "learning_rate": 8.425600000000001e-06, "loss": 0.0313, "step": 7873 }, { "epoch": 0.41978994508716744, "grad_norm": 1.4409953355789185, "learning_rate": 8.4254e-06, "loss": 0.0116, "step": 7874 }, { "epoch": 0.41984325851682036, "grad_norm": 2.7736330032348633, "learning_rate": 8.425200000000002e-06, "loss": 0.0351, "step": 7875 }, { "epoch": 0.41989657194647334, "grad_norm": 4.140346527099609, "learning_rate": 8.425000000000001e-06, "loss": 0.0642, "step": 7876 }, { "epoch": 0.41994988537612626, "grad_norm": 4.910403728485107, "learning_rate": 8.424800000000001e-06, "loss": 0.0348, "step": 7877 }, { "epoch": 0.4200031988057792, "grad_norm": 1.9812904596328735, "learning_rate": 8.4246e-06, "loss": 0.0263, "step": 7878 }, { "epoch": 0.4200565122354321, "grad_norm": 2.303819179534912, "learning_rate": 8.4244e-06, "loss": 0.038, "step": 7879 }, { "epoch": 0.42010982566508503, "grad_norm": 3.9878478050231934, "learning_rate": 8.4242e-06, "loss": 0.0799, "step": 7880 }, { "epoch": 0.42016313909473796, "grad_norm": 2.6458568572998047, "learning_rate": 8.424000000000001e-06, "loss": 0.0371, "step": 7881 }, { "epoch": 0.4202164525243909, "grad_norm": 3.505385398864746, "learning_rate": 8.4238e-06, "loss": 0.0571, "step": 7882 }, { "epoch": 0.4202697659540438, "grad_norm": 2.3098418712615967, "learning_rate": 8.4236e-06, "loss": 0.0266, "step": 7883 }, { "epoch": 0.42032307938369673, "grad_norm": 2.6569712162017822, "learning_rate": 8.4234e-06, "loss": 0.045, "step": 7884 }, { "epoch": 0.4203763928133497, "grad_norm": 6.0835161209106445, "learning_rate": 8.423200000000001e-06, "loss": 0.0901, "step": 7885 }, { "epoch": 0.42042970624300263, "grad_norm": 3.5428433418273926, "learning_rate": 8.423e-06, "loss": 0.0374, "step": 7886 }, { "epoch": 0.42048301967265556, "grad_norm": 4.562068462371826, "learning_rate": 8.422800000000002e-06, "loss": 0.049, "step": 7887 }, { "epoch": 0.4205363331023085, "grad_norm": 2.136758327484131, "learning_rate": 8.422600000000001e-06, "loss": 0.0429, "step": 7888 }, { "epoch": 0.4205896465319614, "grad_norm": 5.045431137084961, "learning_rate": 8.422400000000001e-06, "loss": 0.1012, "step": 7889 }, { "epoch": 0.4206429599616143, "grad_norm": 2.9057540893554688, "learning_rate": 8.4222e-06, "loss": 0.0492, "step": 7890 }, { "epoch": 0.42069627339126725, "grad_norm": 3.057832717895508, "learning_rate": 8.422e-06, "loss": 0.0546, "step": 7891 }, { "epoch": 0.4207495868209202, "grad_norm": 2.559865951538086, "learning_rate": 8.421800000000001e-06, "loss": 0.0384, "step": 7892 }, { "epoch": 0.4208029002505731, "grad_norm": 3.412992238998413, "learning_rate": 8.4216e-06, "loss": 0.0401, "step": 7893 }, { "epoch": 0.4208562136802261, "grad_norm": 2.449331521987915, "learning_rate": 8.4214e-06, "loss": 0.0342, "step": 7894 }, { "epoch": 0.420909527109879, "grad_norm": 2.6256158351898193, "learning_rate": 8.4212e-06, "loss": 0.0413, "step": 7895 }, { "epoch": 0.4209628405395319, "grad_norm": 2.928035020828247, "learning_rate": 8.421e-06, "loss": 0.0471, "step": 7896 }, { "epoch": 0.42101615396918485, "grad_norm": 1.6544554233551025, "learning_rate": 8.4208e-06, "loss": 0.0211, "step": 7897 }, { "epoch": 0.42106946739883777, "grad_norm": 1.2068724632263184, "learning_rate": 8.4206e-06, "loss": 0.0108, "step": 7898 }, { "epoch": 0.4211227808284907, "grad_norm": 1.3538426160812378, "learning_rate": 8.420400000000002e-06, "loss": 0.0129, "step": 7899 }, { "epoch": 0.4211760942581436, "grad_norm": 2.2670576572418213, "learning_rate": 8.420200000000001e-06, "loss": 0.0491, "step": 7900 }, { "epoch": 0.42122940768779654, "grad_norm": 4.465100288391113, "learning_rate": 8.42e-06, "loss": 0.0771, "step": 7901 }, { "epoch": 0.42128272111744947, "grad_norm": 2.5029125213623047, "learning_rate": 8.4198e-06, "loss": 0.037, "step": 7902 }, { "epoch": 0.4213360345471024, "grad_norm": 2.2325150966644287, "learning_rate": 8.419600000000002e-06, "loss": 0.0181, "step": 7903 }, { "epoch": 0.42138934797675537, "grad_norm": 4.450076580047607, "learning_rate": 8.419400000000001e-06, "loss": 0.0336, "step": 7904 }, { "epoch": 0.4214426614064083, "grad_norm": 3.52195405960083, "learning_rate": 8.4192e-06, "loss": 0.0321, "step": 7905 }, { "epoch": 0.4214959748360612, "grad_norm": 1.866247534751892, "learning_rate": 8.419e-06, "loss": 0.0351, "step": 7906 }, { "epoch": 0.42154928826571414, "grad_norm": 2.705831527709961, "learning_rate": 8.4188e-06, "loss": 0.0554, "step": 7907 }, { "epoch": 0.42160260169536706, "grad_norm": 3.6614246368408203, "learning_rate": 8.418600000000001e-06, "loss": 0.0499, "step": 7908 }, { "epoch": 0.42165591512502, "grad_norm": 2.7010114192962646, "learning_rate": 8.4184e-06, "loss": 0.0423, "step": 7909 }, { "epoch": 0.4217092285546729, "grad_norm": 1.7260328531265259, "learning_rate": 8.4182e-06, "loss": 0.0099, "step": 7910 }, { "epoch": 0.42176254198432583, "grad_norm": 3.0455482006073, "learning_rate": 8.418000000000001e-06, "loss": 0.0725, "step": 7911 }, { "epoch": 0.42181585541397876, "grad_norm": 1.8227851390838623, "learning_rate": 8.417800000000001e-06, "loss": 0.0451, "step": 7912 }, { "epoch": 0.42186916884363174, "grad_norm": 1.6817352771759033, "learning_rate": 8.4176e-06, "loss": 0.016, "step": 7913 }, { "epoch": 0.42192248227328466, "grad_norm": 2.527830123901367, "learning_rate": 8.417400000000002e-06, "loss": 0.0725, "step": 7914 }, { "epoch": 0.4219757957029376, "grad_norm": 2.099721908569336, "learning_rate": 8.417200000000001e-06, "loss": 0.0307, "step": 7915 }, { "epoch": 0.4220291091325905, "grad_norm": 2.4138505458831787, "learning_rate": 8.417000000000001e-06, "loss": 0.0448, "step": 7916 }, { "epoch": 0.42208242256224343, "grad_norm": 3.571699619293213, "learning_rate": 8.4168e-06, "loss": 0.0341, "step": 7917 }, { "epoch": 0.42213573599189635, "grad_norm": 2.979957342147827, "learning_rate": 8.4166e-06, "loss": 0.0382, "step": 7918 }, { "epoch": 0.4221890494215493, "grad_norm": 2.0894153118133545, "learning_rate": 8.416400000000001e-06, "loss": 0.0371, "step": 7919 }, { "epoch": 0.4222423628512022, "grad_norm": 3.234614610671997, "learning_rate": 8.4162e-06, "loss": 0.054, "step": 7920 }, { "epoch": 0.4222956762808551, "grad_norm": 2.693601369857788, "learning_rate": 8.416e-06, "loss": 0.023, "step": 7921 }, { "epoch": 0.42234898971050805, "grad_norm": 1.818206548690796, "learning_rate": 8.4158e-06, "loss": 0.0193, "step": 7922 }, { "epoch": 0.42240230314016103, "grad_norm": 3.4590837955474854, "learning_rate": 8.415600000000001e-06, "loss": 0.0594, "step": 7923 }, { "epoch": 0.42245561656981395, "grad_norm": 3.5612261295318604, "learning_rate": 8.4154e-06, "loss": 0.0726, "step": 7924 }, { "epoch": 0.4225089299994669, "grad_norm": 2.391582489013672, "learning_rate": 8.4152e-06, "loss": 0.0513, "step": 7925 }, { "epoch": 0.4225622434291198, "grad_norm": 1.5972750186920166, "learning_rate": 8.415000000000002e-06, "loss": 0.0234, "step": 7926 }, { "epoch": 0.4226155568587727, "grad_norm": 3.824786424636841, "learning_rate": 8.414800000000001e-06, "loss": 0.0614, "step": 7927 }, { "epoch": 0.42266887028842565, "grad_norm": 2.833531141281128, "learning_rate": 8.4146e-06, "loss": 0.0368, "step": 7928 }, { "epoch": 0.42272218371807857, "grad_norm": 3.1485989093780518, "learning_rate": 8.4144e-06, "loss": 0.0372, "step": 7929 }, { "epoch": 0.4227754971477315, "grad_norm": 2.865652322769165, "learning_rate": 8.4142e-06, "loss": 0.023, "step": 7930 }, { "epoch": 0.4228288105773844, "grad_norm": 3.7102904319763184, "learning_rate": 8.414000000000001e-06, "loss": 0.0537, "step": 7931 }, { "epoch": 0.4228821240070374, "grad_norm": 5.107373237609863, "learning_rate": 8.4138e-06, "loss": 0.069, "step": 7932 }, { "epoch": 0.4229354374366903, "grad_norm": 2.904048442840576, "learning_rate": 8.4136e-06, "loss": 0.0598, "step": 7933 }, { "epoch": 0.42298875086634324, "grad_norm": 1.9293612241744995, "learning_rate": 8.4134e-06, "loss": 0.0338, "step": 7934 }, { "epoch": 0.42304206429599617, "grad_norm": 4.131194591522217, "learning_rate": 8.413200000000001e-06, "loss": 0.0741, "step": 7935 }, { "epoch": 0.4230953777256491, "grad_norm": 2.68864369392395, "learning_rate": 8.413e-06, "loss": 0.0268, "step": 7936 }, { "epoch": 0.423148691155302, "grad_norm": 1.6557918787002563, "learning_rate": 8.4128e-06, "loss": 0.031, "step": 7937 }, { "epoch": 0.42320200458495494, "grad_norm": 2.600348949432373, "learning_rate": 8.412600000000001e-06, "loss": 0.0499, "step": 7938 }, { "epoch": 0.42325531801460786, "grad_norm": 2.4290671348571777, "learning_rate": 8.412400000000001e-06, "loss": 0.042, "step": 7939 }, { "epoch": 0.4233086314442608, "grad_norm": 1.6793665885925293, "learning_rate": 8.4122e-06, "loss": 0.0358, "step": 7940 }, { "epoch": 0.42336194487391376, "grad_norm": 2.622736692428589, "learning_rate": 8.412e-06, "loss": 0.0507, "step": 7941 }, { "epoch": 0.4234152583035667, "grad_norm": 3.2328147888183594, "learning_rate": 8.411800000000001e-06, "loss": 0.0504, "step": 7942 }, { "epoch": 0.4234685717332196, "grad_norm": 3.6199944019317627, "learning_rate": 8.411600000000001e-06, "loss": 0.0583, "step": 7943 }, { "epoch": 0.42352188516287254, "grad_norm": 7.167890548706055, "learning_rate": 8.4114e-06, "loss": 0.0856, "step": 7944 }, { "epoch": 0.42357519859252546, "grad_norm": 1.2505967617034912, "learning_rate": 8.4112e-06, "loss": 0.0116, "step": 7945 }, { "epoch": 0.4236285120221784, "grad_norm": 1.2670243978500366, "learning_rate": 8.411e-06, "loss": 0.0188, "step": 7946 }, { "epoch": 0.4236818254518313, "grad_norm": 1.5233843326568604, "learning_rate": 8.410800000000001e-06, "loss": 0.0361, "step": 7947 }, { "epoch": 0.42373513888148423, "grad_norm": 3.6410043239593506, "learning_rate": 8.4106e-06, "loss": 0.0735, "step": 7948 }, { "epoch": 0.42378845231113715, "grad_norm": 1.997290849685669, "learning_rate": 8.4104e-06, "loss": 0.0222, "step": 7949 }, { "epoch": 0.4238417657407901, "grad_norm": 2.358440399169922, "learning_rate": 8.410200000000001e-06, "loss": 0.0205, "step": 7950 }, { "epoch": 0.42389507917044306, "grad_norm": 5.039011478424072, "learning_rate": 8.41e-06, "loss": 0.0884, "step": 7951 }, { "epoch": 0.423948392600096, "grad_norm": 2.6483559608459473, "learning_rate": 8.4098e-06, "loss": 0.0268, "step": 7952 }, { "epoch": 0.4240017060297489, "grad_norm": 2.1991329193115234, "learning_rate": 8.409600000000002e-06, "loss": 0.0315, "step": 7953 }, { "epoch": 0.4240550194594018, "grad_norm": 3.4119319915771484, "learning_rate": 8.409400000000001e-06, "loss": 0.0491, "step": 7954 }, { "epoch": 0.42410833288905475, "grad_norm": 1.8328903913497925, "learning_rate": 8.4092e-06, "loss": 0.031, "step": 7955 }, { "epoch": 0.4241616463187077, "grad_norm": 2.970663547515869, "learning_rate": 8.409e-06, "loss": 0.0237, "step": 7956 }, { "epoch": 0.4242149597483606, "grad_norm": 2.251882553100586, "learning_rate": 8.4088e-06, "loss": 0.041, "step": 7957 }, { "epoch": 0.4242682731780135, "grad_norm": 3.7415857315063477, "learning_rate": 8.408600000000001e-06, "loss": 0.0368, "step": 7958 }, { "epoch": 0.42432158660766645, "grad_norm": 2.1881890296936035, "learning_rate": 8.4084e-06, "loss": 0.0534, "step": 7959 }, { "epoch": 0.4243749000373194, "grad_norm": 2.363919258117676, "learning_rate": 8.4082e-06, "loss": 0.039, "step": 7960 }, { "epoch": 0.42442821346697235, "grad_norm": 3.4949400424957275, "learning_rate": 8.408e-06, "loss": 0.058, "step": 7961 }, { "epoch": 0.42448152689662527, "grad_norm": 2.1659207344055176, "learning_rate": 8.407800000000001e-06, "loss": 0.042, "step": 7962 }, { "epoch": 0.4245348403262782, "grad_norm": 3.1117985248565674, "learning_rate": 8.4076e-06, "loss": 0.0403, "step": 7963 }, { "epoch": 0.4245881537559311, "grad_norm": 2.2078020572662354, "learning_rate": 8.407400000000002e-06, "loss": 0.0301, "step": 7964 }, { "epoch": 0.42464146718558404, "grad_norm": 2.6533901691436768, "learning_rate": 8.407200000000001e-06, "loss": 0.0288, "step": 7965 }, { "epoch": 0.42469478061523697, "grad_norm": 3.2481532096862793, "learning_rate": 8.407000000000001e-06, "loss": 0.0517, "step": 7966 }, { "epoch": 0.4247480940448899, "grad_norm": 4.111559867858887, "learning_rate": 8.4068e-06, "loss": 0.0484, "step": 7967 }, { "epoch": 0.4248014074745428, "grad_norm": 3.7404139041900635, "learning_rate": 8.4066e-06, "loss": 0.0466, "step": 7968 }, { "epoch": 0.4248547209041958, "grad_norm": 1.3563998937606812, "learning_rate": 8.406400000000001e-06, "loss": 0.0164, "step": 7969 }, { "epoch": 0.4249080343338487, "grad_norm": 2.42036509513855, "learning_rate": 8.406200000000001e-06, "loss": 0.0231, "step": 7970 }, { "epoch": 0.42496134776350164, "grad_norm": 2.5976080894470215, "learning_rate": 8.406e-06, "loss": 0.046, "step": 7971 }, { "epoch": 0.42501466119315456, "grad_norm": 4.398369312286377, "learning_rate": 8.4058e-06, "loss": 0.0357, "step": 7972 }, { "epoch": 0.4250679746228075, "grad_norm": 2.853281259536743, "learning_rate": 8.4056e-06, "loss": 0.0705, "step": 7973 }, { "epoch": 0.4251212880524604, "grad_norm": 3.314018726348877, "learning_rate": 8.405400000000001e-06, "loss": 0.0462, "step": 7974 }, { "epoch": 0.42517460148211333, "grad_norm": 2.3486714363098145, "learning_rate": 8.4052e-06, "loss": 0.0475, "step": 7975 }, { "epoch": 0.42522791491176626, "grad_norm": 2.238553524017334, "learning_rate": 8.405000000000002e-06, "loss": 0.0455, "step": 7976 }, { "epoch": 0.4252812283414192, "grad_norm": 1.7462446689605713, "learning_rate": 8.404800000000001e-06, "loss": 0.0182, "step": 7977 }, { "epoch": 0.4253345417710721, "grad_norm": 3.042095899581909, "learning_rate": 8.4046e-06, "loss": 0.0301, "step": 7978 }, { "epoch": 0.4253878552007251, "grad_norm": 2.554701089859009, "learning_rate": 8.4044e-06, "loss": 0.0583, "step": 7979 }, { "epoch": 0.425441168630378, "grad_norm": 2.4067163467407227, "learning_rate": 8.4042e-06, "loss": 0.0478, "step": 7980 }, { "epoch": 0.42549448206003093, "grad_norm": 2.91965389251709, "learning_rate": 8.404000000000001e-06, "loss": 0.0303, "step": 7981 }, { "epoch": 0.42554779548968386, "grad_norm": 2.907801628112793, "learning_rate": 8.4038e-06, "loss": 0.0293, "step": 7982 }, { "epoch": 0.4256011089193368, "grad_norm": 2.4424712657928467, "learning_rate": 8.4036e-06, "loss": 0.0309, "step": 7983 }, { "epoch": 0.4256544223489897, "grad_norm": 5.082684516906738, "learning_rate": 8.4034e-06, "loss": 0.0769, "step": 7984 }, { "epoch": 0.4257077357786426, "grad_norm": 3.122678279876709, "learning_rate": 8.4032e-06, "loss": 0.0431, "step": 7985 }, { "epoch": 0.42576104920829555, "grad_norm": 2.591284990310669, "learning_rate": 8.403e-06, "loss": 0.03, "step": 7986 }, { "epoch": 0.4258143626379485, "grad_norm": 2.8295910358428955, "learning_rate": 8.4028e-06, "loss": 0.0216, "step": 7987 }, { "epoch": 0.42586767606760145, "grad_norm": 2.14508056640625, "learning_rate": 8.402600000000002e-06, "loss": 0.0279, "step": 7988 }, { "epoch": 0.4259209894972544, "grad_norm": 2.3702750205993652, "learning_rate": 8.402400000000001e-06, "loss": 0.05, "step": 7989 }, { "epoch": 0.4259743029269073, "grad_norm": 2.0694315433502197, "learning_rate": 8.4022e-06, "loss": 0.0306, "step": 7990 }, { "epoch": 0.4260276163565602, "grad_norm": 4.556219100952148, "learning_rate": 8.402e-06, "loss": 0.0276, "step": 7991 }, { "epoch": 0.42608092978621315, "grad_norm": 4.1180267333984375, "learning_rate": 8.401800000000001e-06, "loss": 0.0545, "step": 7992 }, { "epoch": 0.42613424321586607, "grad_norm": 2.239419937133789, "learning_rate": 8.401600000000001e-06, "loss": 0.0357, "step": 7993 }, { "epoch": 0.426187556645519, "grad_norm": 2.6448051929473877, "learning_rate": 8.4014e-06, "loss": 0.0348, "step": 7994 }, { "epoch": 0.4262408700751719, "grad_norm": 2.515509843826294, "learning_rate": 8.4012e-06, "loss": 0.0294, "step": 7995 }, { "epoch": 0.42629418350482484, "grad_norm": 1.612112283706665, "learning_rate": 8.401e-06, "loss": 0.0102, "step": 7996 }, { "epoch": 0.4263474969344778, "grad_norm": 3.5209314823150635, "learning_rate": 8.400800000000001e-06, "loss": 0.0473, "step": 7997 }, { "epoch": 0.42640081036413074, "grad_norm": 1.9577800035476685, "learning_rate": 8.4006e-06, "loss": 0.021, "step": 7998 }, { "epoch": 0.42645412379378367, "grad_norm": 6.274689197540283, "learning_rate": 8.4004e-06, "loss": 0.0534, "step": 7999 }, { "epoch": 0.4265074372234366, "grad_norm": 2.0015740394592285, "learning_rate": 8.400200000000001e-06, "loss": 0.0253, "step": 8000 }, { "epoch": 0.4265607506530895, "grad_norm": 2.143651008605957, "learning_rate": 8.400000000000001e-06, "loss": 0.0332, "step": 8001 }, { "epoch": 0.42661406408274244, "grad_norm": 2.8007009029388428, "learning_rate": 8.3998e-06, "loss": 0.0459, "step": 8002 }, { "epoch": 0.42666737751239536, "grad_norm": 1.6729053258895874, "learning_rate": 8.399600000000002e-06, "loss": 0.0188, "step": 8003 }, { "epoch": 0.4267206909420483, "grad_norm": 1.642519235610962, "learning_rate": 8.399400000000001e-06, "loss": 0.0241, "step": 8004 }, { "epoch": 0.4267740043717012, "grad_norm": 1.3174233436584473, "learning_rate": 8.3992e-06, "loss": 0.0203, "step": 8005 }, { "epoch": 0.42682731780135413, "grad_norm": 2.8349239826202393, "learning_rate": 8.399e-06, "loss": 0.0418, "step": 8006 }, { "epoch": 0.4268806312310071, "grad_norm": 2.6780076026916504, "learning_rate": 8.3988e-06, "loss": 0.035, "step": 8007 }, { "epoch": 0.42693394466066004, "grad_norm": 2.6100010871887207, "learning_rate": 8.398600000000001e-06, "loss": 0.0467, "step": 8008 }, { "epoch": 0.42698725809031296, "grad_norm": 2.1140196323394775, "learning_rate": 8.3984e-06, "loss": 0.0491, "step": 8009 }, { "epoch": 0.4270405715199659, "grad_norm": 4.34821891784668, "learning_rate": 8.3982e-06, "loss": 0.0225, "step": 8010 }, { "epoch": 0.4270938849496188, "grad_norm": 2.603785753250122, "learning_rate": 8.398e-06, "loss": 0.0611, "step": 8011 }, { "epoch": 0.42714719837927173, "grad_norm": 8.09742259979248, "learning_rate": 8.397800000000001e-06, "loss": 0.0661, "step": 8012 }, { "epoch": 0.42720051180892465, "grad_norm": 3.1029980182647705, "learning_rate": 8.3976e-06, "loss": 0.0226, "step": 8013 }, { "epoch": 0.4272538252385776, "grad_norm": 5.339808464050293, "learning_rate": 8.3974e-06, "loss": 0.0675, "step": 8014 }, { "epoch": 0.4273071386682305, "grad_norm": 1.6908409595489502, "learning_rate": 8.397200000000002e-06, "loss": 0.0211, "step": 8015 }, { "epoch": 0.4273604520978835, "grad_norm": 3.010240077972412, "learning_rate": 8.397000000000001e-06, "loss": 0.0516, "step": 8016 }, { "epoch": 0.4274137655275364, "grad_norm": 3.6552083492279053, "learning_rate": 8.3968e-06, "loss": 0.0592, "step": 8017 }, { "epoch": 0.42746707895718933, "grad_norm": 3.693747043609619, "learning_rate": 8.3966e-06, "loss": 0.0661, "step": 8018 }, { "epoch": 0.42752039238684225, "grad_norm": 3.0040409564971924, "learning_rate": 8.396400000000001e-06, "loss": 0.0416, "step": 8019 }, { "epoch": 0.4275737058164952, "grad_norm": 2.799429416656494, "learning_rate": 8.396200000000001e-06, "loss": 0.0225, "step": 8020 }, { "epoch": 0.4276270192461481, "grad_norm": 3.6014885902404785, "learning_rate": 8.396e-06, "loss": 0.0749, "step": 8021 }, { "epoch": 0.427680332675801, "grad_norm": 4.112411022186279, "learning_rate": 8.3958e-06, "loss": 0.0456, "step": 8022 }, { "epoch": 0.42773364610545395, "grad_norm": 4.428885459899902, "learning_rate": 8.3956e-06, "loss": 0.0355, "step": 8023 }, { "epoch": 0.42778695953510687, "grad_norm": 5.684564590454102, "learning_rate": 8.395400000000001e-06, "loss": 0.1024, "step": 8024 }, { "epoch": 0.42784027296475985, "grad_norm": 4.2109479904174805, "learning_rate": 8.3952e-06, "loss": 0.0385, "step": 8025 }, { "epoch": 0.4278935863944128, "grad_norm": 2.7735824584960938, "learning_rate": 8.395e-06, "loss": 0.0419, "step": 8026 }, { "epoch": 0.4279468998240657, "grad_norm": 4.86625337600708, "learning_rate": 8.394800000000001e-06, "loss": 0.0876, "step": 8027 }, { "epoch": 0.4280002132537186, "grad_norm": 2.4759838581085205, "learning_rate": 8.394600000000001e-06, "loss": 0.0347, "step": 8028 }, { "epoch": 0.42805352668337154, "grad_norm": 0.9863107204437256, "learning_rate": 8.3944e-06, "loss": 0.0129, "step": 8029 }, { "epoch": 0.42810684011302447, "grad_norm": 1.7444729804992676, "learning_rate": 8.394200000000002e-06, "loss": 0.0253, "step": 8030 }, { "epoch": 0.4281601535426774, "grad_norm": 4.248649597167969, "learning_rate": 8.394000000000001e-06, "loss": 0.0507, "step": 8031 }, { "epoch": 0.4282134669723303, "grad_norm": 4.003203868865967, "learning_rate": 8.393800000000001e-06, "loss": 0.0447, "step": 8032 }, { "epoch": 0.42826678040198324, "grad_norm": 1.6484163999557495, "learning_rate": 8.3936e-06, "loss": 0.034, "step": 8033 }, { "epoch": 0.42832009383163616, "grad_norm": 2.715907335281372, "learning_rate": 8.3934e-06, "loss": 0.0341, "step": 8034 }, { "epoch": 0.42837340726128914, "grad_norm": 2.750978946685791, "learning_rate": 8.3932e-06, "loss": 0.0547, "step": 8035 }, { "epoch": 0.42842672069094206, "grad_norm": 2.8950350284576416, "learning_rate": 8.393e-06, "loss": 0.0637, "step": 8036 }, { "epoch": 0.428480034120595, "grad_norm": 2.6965861320495605, "learning_rate": 8.3928e-06, "loss": 0.0708, "step": 8037 }, { "epoch": 0.4285333475502479, "grad_norm": 3.395681142807007, "learning_rate": 8.3926e-06, "loss": 0.0388, "step": 8038 }, { "epoch": 0.42858666097990084, "grad_norm": 2.73626971244812, "learning_rate": 8.392400000000001e-06, "loss": 0.048, "step": 8039 }, { "epoch": 0.42863997440955376, "grad_norm": 3.4062819480895996, "learning_rate": 8.3922e-06, "loss": 0.0702, "step": 8040 }, { "epoch": 0.4286932878392067, "grad_norm": 5.115875720977783, "learning_rate": 8.392e-06, "loss": 0.0572, "step": 8041 }, { "epoch": 0.4287466012688596, "grad_norm": 7.233651161193848, "learning_rate": 8.391800000000002e-06, "loss": 0.0499, "step": 8042 }, { "epoch": 0.42879991469851253, "grad_norm": 3.7055790424346924, "learning_rate": 8.391600000000001e-06, "loss": 0.0393, "step": 8043 }, { "epoch": 0.4288532281281655, "grad_norm": 2.572075843811035, "learning_rate": 8.3914e-06, "loss": 0.0733, "step": 8044 }, { "epoch": 0.42890654155781843, "grad_norm": 3.1377835273742676, "learning_rate": 8.3912e-06, "loss": 0.0543, "step": 8045 }, { "epoch": 0.42895985498747136, "grad_norm": 4.278900146484375, "learning_rate": 8.391e-06, "loss": 0.0669, "step": 8046 }, { "epoch": 0.4290131684171243, "grad_norm": 3.0546889305114746, "learning_rate": 8.390800000000001e-06, "loss": 0.0344, "step": 8047 }, { "epoch": 0.4290664818467772, "grad_norm": 2.005431652069092, "learning_rate": 8.3906e-06, "loss": 0.0379, "step": 8048 }, { "epoch": 0.4291197952764301, "grad_norm": 2.568880558013916, "learning_rate": 8.3904e-06, "loss": 0.0317, "step": 8049 }, { "epoch": 0.42917310870608305, "grad_norm": 4.143054962158203, "learning_rate": 8.3902e-06, "loss": 0.0503, "step": 8050 }, { "epoch": 0.429226422135736, "grad_norm": 2.2146224975585938, "learning_rate": 8.390000000000001e-06, "loss": 0.0457, "step": 8051 }, { "epoch": 0.4292797355653889, "grad_norm": 1.8631187677383423, "learning_rate": 8.3898e-06, "loss": 0.0352, "step": 8052 }, { "epoch": 0.4293330489950419, "grad_norm": 2.4221010208129883, "learning_rate": 8.389600000000002e-06, "loss": 0.0323, "step": 8053 }, { "epoch": 0.4293863624246948, "grad_norm": 2.9012997150421143, "learning_rate": 8.389400000000001e-06, "loss": 0.0247, "step": 8054 }, { "epoch": 0.4294396758543477, "grad_norm": 1.9899846315383911, "learning_rate": 8.389200000000001e-06, "loss": 0.0434, "step": 8055 }, { "epoch": 0.42949298928400065, "grad_norm": 7.082215785980225, "learning_rate": 8.389e-06, "loss": 0.0595, "step": 8056 }, { "epoch": 0.42954630271365357, "grad_norm": 3.5567400455474854, "learning_rate": 8.3888e-06, "loss": 0.0319, "step": 8057 }, { "epoch": 0.4295996161433065, "grad_norm": 2.1666245460510254, "learning_rate": 8.388600000000001e-06, "loss": 0.0206, "step": 8058 }, { "epoch": 0.4296529295729594, "grad_norm": 4.776373863220215, "learning_rate": 8.388400000000001e-06, "loss": 0.0729, "step": 8059 }, { "epoch": 0.42970624300261234, "grad_norm": 2.9446346759796143, "learning_rate": 8.3882e-06, "loss": 0.0565, "step": 8060 }, { "epoch": 0.42975955643226527, "grad_norm": 1.235876441001892, "learning_rate": 8.388e-06, "loss": 0.0283, "step": 8061 }, { "epoch": 0.4298128698619182, "grad_norm": 2.3147878646850586, "learning_rate": 8.3878e-06, "loss": 0.0383, "step": 8062 }, { "epoch": 0.42986618329157117, "grad_norm": 4.1322784423828125, "learning_rate": 8.3876e-06, "loss": 0.0573, "step": 8063 }, { "epoch": 0.4299194967212241, "grad_norm": 2.2011401653289795, "learning_rate": 8.3874e-06, "loss": 0.0308, "step": 8064 }, { "epoch": 0.429972810150877, "grad_norm": 2.08528470993042, "learning_rate": 8.387200000000002e-06, "loss": 0.0345, "step": 8065 }, { "epoch": 0.43002612358052994, "grad_norm": 4.061593055725098, "learning_rate": 8.387000000000001e-06, "loss": 0.0888, "step": 8066 }, { "epoch": 0.43007943701018286, "grad_norm": 3.278024911880493, "learning_rate": 8.3868e-06, "loss": 0.032, "step": 8067 }, { "epoch": 0.4301327504398358, "grad_norm": 1.9907376766204834, "learning_rate": 8.3866e-06, "loss": 0.0358, "step": 8068 }, { "epoch": 0.4301860638694887, "grad_norm": 1.5598338842391968, "learning_rate": 8.386400000000002e-06, "loss": 0.0282, "step": 8069 }, { "epoch": 0.43023937729914163, "grad_norm": 1.9260691404342651, "learning_rate": 8.386200000000001e-06, "loss": 0.0163, "step": 8070 }, { "epoch": 0.43029269072879456, "grad_norm": 2.9701621532440186, "learning_rate": 8.386e-06, "loss": 0.0599, "step": 8071 }, { "epoch": 0.43034600415844754, "grad_norm": 1.3823843002319336, "learning_rate": 8.3858e-06, "loss": 0.0226, "step": 8072 }, { "epoch": 0.43039931758810046, "grad_norm": 3.519808769226074, "learning_rate": 8.3856e-06, "loss": 0.0484, "step": 8073 }, { "epoch": 0.4304526310177534, "grad_norm": 2.475759267807007, "learning_rate": 8.3854e-06, "loss": 0.0368, "step": 8074 }, { "epoch": 0.4305059444474063, "grad_norm": 4.256880283355713, "learning_rate": 8.3852e-06, "loss": 0.0652, "step": 8075 }, { "epoch": 0.43055925787705923, "grad_norm": 1.8124886751174927, "learning_rate": 8.385e-06, "loss": 0.0201, "step": 8076 }, { "epoch": 0.43061257130671216, "grad_norm": 1.8471410274505615, "learning_rate": 8.384800000000001e-06, "loss": 0.0384, "step": 8077 }, { "epoch": 0.4306658847363651, "grad_norm": 3.8668484687805176, "learning_rate": 8.384600000000001e-06, "loss": 0.0251, "step": 8078 }, { "epoch": 0.430719198166018, "grad_norm": 1.3079710006713867, "learning_rate": 8.3844e-06, "loss": 0.0206, "step": 8079 }, { "epoch": 0.4307725115956709, "grad_norm": 6.739748001098633, "learning_rate": 8.384200000000002e-06, "loss": 0.0604, "step": 8080 }, { "epoch": 0.4308258250253239, "grad_norm": 2.2922444343566895, "learning_rate": 8.384000000000001e-06, "loss": 0.0221, "step": 8081 }, { "epoch": 0.43087913845497683, "grad_norm": 2.277026414871216, "learning_rate": 8.383800000000001e-06, "loss": 0.0165, "step": 8082 }, { "epoch": 0.43093245188462975, "grad_norm": 2.7457523345947266, "learning_rate": 8.3836e-06, "loss": 0.0269, "step": 8083 }, { "epoch": 0.4309857653142827, "grad_norm": 3.5765340328216553, "learning_rate": 8.3834e-06, "loss": 0.0273, "step": 8084 }, { "epoch": 0.4310390787439356, "grad_norm": 3.8056535720825195, "learning_rate": 8.3832e-06, "loss": 0.0657, "step": 8085 }, { "epoch": 0.4310923921735885, "grad_norm": 1.3551137447357178, "learning_rate": 8.383000000000001e-06, "loss": 0.0114, "step": 8086 }, { "epoch": 0.43114570560324145, "grad_norm": 6.032891273498535, "learning_rate": 8.3828e-06, "loss": 0.0651, "step": 8087 }, { "epoch": 0.43119901903289437, "grad_norm": 3.6441729068756104, "learning_rate": 8.3826e-06, "loss": 0.0745, "step": 8088 }, { "epoch": 0.4312523324625473, "grad_norm": 3.300565481185913, "learning_rate": 8.382400000000001e-06, "loss": 0.0515, "step": 8089 }, { "epoch": 0.4313056458922002, "grad_norm": 1.1367759704589844, "learning_rate": 8.3822e-06, "loss": 0.0095, "step": 8090 }, { "epoch": 0.4313589593218532, "grad_norm": 1.4171204566955566, "learning_rate": 8.382e-06, "loss": 0.0209, "step": 8091 }, { "epoch": 0.4314122727515061, "grad_norm": 2.474546194076538, "learning_rate": 8.381800000000002e-06, "loss": 0.0434, "step": 8092 }, { "epoch": 0.43146558618115904, "grad_norm": 2.6487250328063965, "learning_rate": 8.381600000000001e-06, "loss": 0.0371, "step": 8093 }, { "epoch": 0.43151889961081197, "grad_norm": 3.4634454250335693, "learning_rate": 8.3814e-06, "loss": 0.0482, "step": 8094 }, { "epoch": 0.4315722130404649, "grad_norm": 1.9917924404144287, "learning_rate": 8.3812e-06, "loss": 0.0335, "step": 8095 }, { "epoch": 0.4316255264701178, "grad_norm": 2.870506763458252, "learning_rate": 8.381e-06, "loss": 0.0266, "step": 8096 }, { "epoch": 0.43167883989977074, "grad_norm": 4.414074897766113, "learning_rate": 8.380800000000001e-06, "loss": 0.0555, "step": 8097 }, { "epoch": 0.43173215332942366, "grad_norm": 2.2136542797088623, "learning_rate": 8.3806e-06, "loss": 0.0272, "step": 8098 }, { "epoch": 0.4317854667590766, "grad_norm": 4.396102428436279, "learning_rate": 8.3804e-06, "loss": 0.0934, "step": 8099 }, { "epoch": 0.43183878018872957, "grad_norm": 2.4874489307403564, "learning_rate": 8.3802e-06, "loss": 0.0349, "step": 8100 }, { "epoch": 0.4318920936183825, "grad_norm": 2.2606992721557617, "learning_rate": 8.380000000000001e-06, "loss": 0.0246, "step": 8101 }, { "epoch": 0.4319454070480354, "grad_norm": 1.3134815692901611, "learning_rate": 8.3798e-06, "loss": 0.0354, "step": 8102 }, { "epoch": 0.43199872047768834, "grad_norm": 3.444309711456299, "learning_rate": 8.379600000000002e-06, "loss": 0.054, "step": 8103 }, { "epoch": 0.43205203390734126, "grad_norm": 2.9257473945617676, "learning_rate": 8.379400000000001e-06, "loss": 0.0573, "step": 8104 }, { "epoch": 0.4321053473369942, "grad_norm": 3.200124979019165, "learning_rate": 8.379200000000001e-06, "loss": 0.0452, "step": 8105 }, { "epoch": 0.4321586607666471, "grad_norm": 2.1471824645996094, "learning_rate": 8.379e-06, "loss": 0.0256, "step": 8106 }, { "epoch": 0.43221197419630003, "grad_norm": 7.8351569175720215, "learning_rate": 8.3788e-06, "loss": 0.0536, "step": 8107 }, { "epoch": 0.43226528762595295, "grad_norm": 4.643676280975342, "learning_rate": 8.378600000000001e-06, "loss": 0.0634, "step": 8108 }, { "epoch": 0.43231860105560593, "grad_norm": 3.000429391860962, "learning_rate": 8.378400000000001e-06, "loss": 0.0497, "step": 8109 }, { "epoch": 0.43237191448525886, "grad_norm": 3.2268872261047363, "learning_rate": 8.3782e-06, "loss": 0.0417, "step": 8110 }, { "epoch": 0.4324252279149118, "grad_norm": 2.3196322917938232, "learning_rate": 8.378e-06, "loss": 0.0402, "step": 8111 }, { "epoch": 0.4324785413445647, "grad_norm": 2.0280370712280273, "learning_rate": 8.3778e-06, "loss": 0.0422, "step": 8112 }, { "epoch": 0.43253185477421763, "grad_norm": 2.4751944541931152, "learning_rate": 8.377600000000001e-06, "loss": 0.0407, "step": 8113 }, { "epoch": 0.43258516820387055, "grad_norm": 5.561569690704346, "learning_rate": 8.3774e-06, "loss": 0.0395, "step": 8114 }, { "epoch": 0.4326384816335235, "grad_norm": 2.1653504371643066, "learning_rate": 8.377200000000002e-06, "loss": 0.0168, "step": 8115 }, { "epoch": 0.4326917950631764, "grad_norm": 3.6479594707489014, "learning_rate": 8.377000000000001e-06, "loss": 0.0386, "step": 8116 }, { "epoch": 0.4327451084928293, "grad_norm": 3.164565324783325, "learning_rate": 8.3768e-06, "loss": 0.056, "step": 8117 }, { "epoch": 0.43279842192248225, "grad_norm": 2.0069363117218018, "learning_rate": 8.3766e-06, "loss": 0.0275, "step": 8118 }, { "epoch": 0.4328517353521352, "grad_norm": 2.6068239212036133, "learning_rate": 8.376400000000002e-06, "loss": 0.0241, "step": 8119 }, { "epoch": 0.43290504878178815, "grad_norm": 1.1824603080749512, "learning_rate": 8.376200000000001e-06, "loss": 0.0222, "step": 8120 }, { "epoch": 0.4329583622114411, "grad_norm": 4.345384120941162, "learning_rate": 8.376e-06, "loss": 0.0648, "step": 8121 }, { "epoch": 0.433011675641094, "grad_norm": 3.030240297317505, "learning_rate": 8.3758e-06, "loss": 0.0431, "step": 8122 }, { "epoch": 0.4330649890707469, "grad_norm": 5.561174392700195, "learning_rate": 8.3756e-06, "loss": 0.0451, "step": 8123 }, { "epoch": 0.43311830250039984, "grad_norm": 4.4665632247924805, "learning_rate": 8.3754e-06, "loss": 0.0433, "step": 8124 }, { "epoch": 0.43317161593005277, "grad_norm": 2.1297028064727783, "learning_rate": 8.3752e-06, "loss": 0.0358, "step": 8125 }, { "epoch": 0.4332249293597057, "grad_norm": 1.4964606761932373, "learning_rate": 8.375e-06, "loss": 0.0145, "step": 8126 }, { "epoch": 0.4332782427893586, "grad_norm": 1.3285895586013794, "learning_rate": 8.374800000000002e-06, "loss": 0.0251, "step": 8127 }, { "epoch": 0.4333315562190116, "grad_norm": 3.1969738006591797, "learning_rate": 8.374600000000001e-06, "loss": 0.0665, "step": 8128 }, { "epoch": 0.4333848696486645, "grad_norm": 2.6074514389038086, "learning_rate": 8.3744e-06, "loss": 0.0335, "step": 8129 }, { "epoch": 0.43343818307831744, "grad_norm": 5.090755939483643, "learning_rate": 8.374200000000002e-06, "loss": 0.0419, "step": 8130 }, { "epoch": 0.43349149650797036, "grad_norm": 3.136601448059082, "learning_rate": 8.374000000000001e-06, "loss": 0.0694, "step": 8131 }, { "epoch": 0.4335448099376233, "grad_norm": 1.6697978973388672, "learning_rate": 8.373800000000001e-06, "loss": 0.0183, "step": 8132 }, { "epoch": 0.4335981233672762, "grad_norm": 2.696838140487671, "learning_rate": 8.3736e-06, "loss": 0.0277, "step": 8133 }, { "epoch": 0.43365143679692914, "grad_norm": 2.279538631439209, "learning_rate": 8.3734e-06, "loss": 0.0296, "step": 8134 }, { "epoch": 0.43370475022658206, "grad_norm": 2.36983323097229, "learning_rate": 8.3732e-06, "loss": 0.0331, "step": 8135 }, { "epoch": 0.433758063656235, "grad_norm": 2.9894938468933105, "learning_rate": 8.373000000000001e-06, "loss": 0.0602, "step": 8136 }, { "epoch": 0.43381137708588796, "grad_norm": 2.1686689853668213, "learning_rate": 8.3728e-06, "loss": 0.033, "step": 8137 }, { "epoch": 0.4338646905155409, "grad_norm": 2.167778253555298, "learning_rate": 8.3726e-06, "loss": 0.0301, "step": 8138 }, { "epoch": 0.4339180039451938, "grad_norm": 3.408090829849243, "learning_rate": 8.372400000000001e-06, "loss": 0.0533, "step": 8139 }, { "epoch": 0.43397131737484673, "grad_norm": 1.328560709953308, "learning_rate": 8.372200000000001e-06, "loss": 0.0153, "step": 8140 }, { "epoch": 0.43402463080449966, "grad_norm": 1.3174245357513428, "learning_rate": 8.372e-06, "loss": 0.02, "step": 8141 }, { "epoch": 0.4340779442341526, "grad_norm": 1.2405650615692139, "learning_rate": 8.371800000000002e-06, "loss": 0.0187, "step": 8142 }, { "epoch": 0.4341312576638055, "grad_norm": 2.4996581077575684, "learning_rate": 8.371600000000001e-06, "loss": 0.0375, "step": 8143 }, { "epoch": 0.4341845710934584, "grad_norm": 1.385549545288086, "learning_rate": 8.371400000000001e-06, "loss": 0.0165, "step": 8144 }, { "epoch": 0.43423788452311135, "grad_norm": 3.1138112545013428, "learning_rate": 8.3712e-06, "loss": 0.0321, "step": 8145 }, { "epoch": 0.4342911979527643, "grad_norm": 1.6579115390777588, "learning_rate": 8.371e-06, "loss": 0.0255, "step": 8146 }, { "epoch": 0.43434451138241725, "grad_norm": 1.9830808639526367, "learning_rate": 8.370800000000001e-06, "loss": 0.0428, "step": 8147 }, { "epoch": 0.4343978248120702, "grad_norm": 1.6759144067764282, "learning_rate": 8.3706e-06, "loss": 0.0125, "step": 8148 }, { "epoch": 0.4344511382417231, "grad_norm": 2.7021279335021973, "learning_rate": 8.3704e-06, "loss": 0.0411, "step": 8149 }, { "epoch": 0.434504451671376, "grad_norm": 3.0900306701660156, "learning_rate": 8.3702e-06, "loss": 0.0547, "step": 8150 }, { "epoch": 0.43455776510102895, "grad_norm": 3.156862497329712, "learning_rate": 8.370000000000001e-06, "loss": 0.0293, "step": 8151 }, { "epoch": 0.43461107853068187, "grad_norm": 3.238689422607422, "learning_rate": 8.3698e-06, "loss": 0.03, "step": 8152 }, { "epoch": 0.4346643919603348, "grad_norm": 2.605280876159668, "learning_rate": 8.3696e-06, "loss": 0.0342, "step": 8153 }, { "epoch": 0.4347177053899877, "grad_norm": 2.1162919998168945, "learning_rate": 8.369400000000002e-06, "loss": 0.0383, "step": 8154 }, { "epoch": 0.43477101881964064, "grad_norm": 3.3399453163146973, "learning_rate": 8.369200000000001e-06, "loss": 0.0312, "step": 8155 }, { "epoch": 0.4348243322492936, "grad_norm": 2.610776901245117, "learning_rate": 8.369e-06, "loss": 0.024, "step": 8156 }, { "epoch": 0.43487764567894654, "grad_norm": 3.9510884284973145, "learning_rate": 8.3688e-06, "loss": 0.0579, "step": 8157 }, { "epoch": 0.43493095910859947, "grad_norm": 4.951785087585449, "learning_rate": 8.368600000000001e-06, "loss": 0.0467, "step": 8158 }, { "epoch": 0.4349842725382524, "grad_norm": 5.262027740478516, "learning_rate": 8.368400000000001e-06, "loss": 0.0445, "step": 8159 }, { "epoch": 0.4350375859679053, "grad_norm": 3.492640972137451, "learning_rate": 8.3682e-06, "loss": 0.0331, "step": 8160 }, { "epoch": 0.43509089939755824, "grad_norm": 2.5634448528289795, "learning_rate": 8.368e-06, "loss": 0.0316, "step": 8161 }, { "epoch": 0.43514421282721116, "grad_norm": 1.8913931846618652, "learning_rate": 8.3678e-06, "loss": 0.0161, "step": 8162 }, { "epoch": 0.4351975262568641, "grad_norm": 2.178582191467285, "learning_rate": 8.367600000000001e-06, "loss": 0.0408, "step": 8163 }, { "epoch": 0.435250839686517, "grad_norm": 3.5737781524658203, "learning_rate": 8.3674e-06, "loss": 0.033, "step": 8164 }, { "epoch": 0.43530415311617, "grad_norm": 3.859320878982544, "learning_rate": 8.3672e-06, "loss": 0.049, "step": 8165 }, { "epoch": 0.4353574665458229, "grad_norm": 3.139516592025757, "learning_rate": 8.367000000000001e-06, "loss": 0.0247, "step": 8166 }, { "epoch": 0.43541077997547584, "grad_norm": 2.242666482925415, "learning_rate": 8.366800000000001e-06, "loss": 0.0216, "step": 8167 }, { "epoch": 0.43546409340512876, "grad_norm": 3.0327630043029785, "learning_rate": 8.3666e-06, "loss": 0.0501, "step": 8168 }, { "epoch": 0.4355174068347817, "grad_norm": 2.657883882522583, "learning_rate": 8.366400000000002e-06, "loss": 0.0247, "step": 8169 }, { "epoch": 0.4355707202644346, "grad_norm": 3.528981924057007, "learning_rate": 8.366200000000001e-06, "loss": 0.0377, "step": 8170 }, { "epoch": 0.43562403369408753, "grad_norm": 3.484218120574951, "learning_rate": 8.366000000000001e-06, "loss": 0.0689, "step": 8171 }, { "epoch": 0.43567734712374045, "grad_norm": 3.8242969512939453, "learning_rate": 8.3658e-06, "loss": 0.0465, "step": 8172 }, { "epoch": 0.4357306605533934, "grad_norm": 2.8515238761901855, "learning_rate": 8.3656e-06, "loss": 0.0313, "step": 8173 }, { "epoch": 0.4357839739830463, "grad_norm": 2.898522138595581, "learning_rate": 8.3654e-06, "loss": 0.0246, "step": 8174 }, { "epoch": 0.4358372874126993, "grad_norm": 4.692660331726074, "learning_rate": 8.3652e-06, "loss": 0.0308, "step": 8175 }, { "epoch": 0.4358906008423522, "grad_norm": 2.1350908279418945, "learning_rate": 8.365e-06, "loss": 0.0428, "step": 8176 }, { "epoch": 0.43594391427200513, "grad_norm": 2.3773276805877686, "learning_rate": 8.3648e-06, "loss": 0.0271, "step": 8177 }, { "epoch": 0.43599722770165805, "grad_norm": 6.037754535675049, "learning_rate": 8.364600000000001e-06, "loss": 0.0716, "step": 8178 }, { "epoch": 0.436050541131311, "grad_norm": 3.990841865539551, "learning_rate": 8.3644e-06, "loss": 0.0903, "step": 8179 }, { "epoch": 0.4361038545609639, "grad_norm": 2.0197463035583496, "learning_rate": 8.364200000000002e-06, "loss": 0.0364, "step": 8180 }, { "epoch": 0.4361571679906168, "grad_norm": 4.438655853271484, "learning_rate": 8.364000000000002e-06, "loss": 0.0503, "step": 8181 }, { "epoch": 0.43621048142026975, "grad_norm": 2.320661783218384, "learning_rate": 8.363800000000001e-06, "loss": 0.0585, "step": 8182 }, { "epoch": 0.43626379484992267, "grad_norm": 3.0330095291137695, "learning_rate": 8.3636e-06, "loss": 0.0495, "step": 8183 }, { "epoch": 0.43631710827957565, "grad_norm": 3.141434907913208, "learning_rate": 8.3634e-06, "loss": 0.0402, "step": 8184 }, { "epoch": 0.4363704217092286, "grad_norm": 2.2746849060058594, "learning_rate": 8.3632e-06, "loss": 0.022, "step": 8185 }, { "epoch": 0.4364237351388815, "grad_norm": 3.7720248699188232, "learning_rate": 8.363000000000001e-06, "loss": 0.035, "step": 8186 }, { "epoch": 0.4364770485685344, "grad_norm": 2.066959857940674, "learning_rate": 8.3628e-06, "loss": 0.0309, "step": 8187 }, { "epoch": 0.43653036199818734, "grad_norm": 2.7884573936462402, "learning_rate": 8.3626e-06, "loss": 0.0461, "step": 8188 }, { "epoch": 0.43658367542784027, "grad_norm": 4.143524646759033, "learning_rate": 8.3624e-06, "loss": 0.0615, "step": 8189 }, { "epoch": 0.4366369888574932, "grad_norm": 3.297091007232666, "learning_rate": 8.362200000000001e-06, "loss": 0.0393, "step": 8190 }, { "epoch": 0.4366903022871461, "grad_norm": 5.6128106117248535, "learning_rate": 8.362e-06, "loss": 0.0391, "step": 8191 }, { "epoch": 0.43674361571679904, "grad_norm": 1.8323233127593994, "learning_rate": 8.361800000000002e-06, "loss": 0.0343, "step": 8192 }, { "epoch": 0.436796929146452, "grad_norm": 1.8303064107894897, "learning_rate": 8.361600000000001e-06, "loss": 0.0359, "step": 8193 }, { "epoch": 0.43685024257610494, "grad_norm": 2.1308279037475586, "learning_rate": 8.361400000000001e-06, "loss": 0.0324, "step": 8194 }, { "epoch": 0.43690355600575786, "grad_norm": 3.9170114994049072, "learning_rate": 8.3612e-06, "loss": 0.0745, "step": 8195 }, { "epoch": 0.4369568694354108, "grad_norm": 3.4392101764678955, "learning_rate": 8.361e-06, "loss": 0.0339, "step": 8196 }, { "epoch": 0.4370101828650637, "grad_norm": 1.7833940982818604, "learning_rate": 8.360800000000001e-06, "loss": 0.0291, "step": 8197 }, { "epoch": 0.43706349629471664, "grad_norm": 3.498629093170166, "learning_rate": 8.360600000000001e-06, "loss": 0.0237, "step": 8198 }, { "epoch": 0.43711680972436956, "grad_norm": 2.9703917503356934, "learning_rate": 8.3604e-06, "loss": 0.05, "step": 8199 }, { "epoch": 0.4371701231540225, "grad_norm": 1.7333513498306274, "learning_rate": 8.3602e-06, "loss": 0.0444, "step": 8200 }, { "epoch": 0.4372234365836754, "grad_norm": 3.3370614051818848, "learning_rate": 8.36e-06, "loss": 0.0335, "step": 8201 }, { "epoch": 0.43727675001332833, "grad_norm": 2.1504974365234375, "learning_rate": 8.3598e-06, "loss": 0.0563, "step": 8202 }, { "epoch": 0.4373300634429813, "grad_norm": 3.987502336502075, "learning_rate": 8.3596e-06, "loss": 0.0719, "step": 8203 }, { "epoch": 0.43738337687263423, "grad_norm": 3.3838722705841064, "learning_rate": 8.359400000000002e-06, "loss": 0.0369, "step": 8204 }, { "epoch": 0.43743669030228716, "grad_norm": 2.157947301864624, "learning_rate": 8.359200000000001e-06, "loss": 0.0242, "step": 8205 }, { "epoch": 0.4374900037319401, "grad_norm": 3.9823403358459473, "learning_rate": 8.359e-06, "loss": 0.0587, "step": 8206 }, { "epoch": 0.437543317161593, "grad_norm": 2.959195375442505, "learning_rate": 8.3588e-06, "loss": 0.0365, "step": 8207 }, { "epoch": 0.4375966305912459, "grad_norm": 3.889540672302246, "learning_rate": 8.358600000000002e-06, "loss": 0.0555, "step": 8208 }, { "epoch": 0.43764994402089885, "grad_norm": 4.4497809410095215, "learning_rate": 8.358400000000001e-06, "loss": 0.0783, "step": 8209 }, { "epoch": 0.4377032574505518, "grad_norm": 4.687913417816162, "learning_rate": 8.3582e-06, "loss": 0.049, "step": 8210 }, { "epoch": 0.4377565708802047, "grad_norm": 2.333336114883423, "learning_rate": 8.358e-06, "loss": 0.0315, "step": 8211 }, { "epoch": 0.4378098843098577, "grad_norm": 2.9753336906433105, "learning_rate": 8.3578e-06, "loss": 0.0306, "step": 8212 }, { "epoch": 0.4378631977395106, "grad_norm": 3.2232062816619873, "learning_rate": 8.3576e-06, "loss": 0.0679, "step": 8213 }, { "epoch": 0.4379165111691635, "grad_norm": 3.8281774520874023, "learning_rate": 8.3574e-06, "loss": 0.0673, "step": 8214 }, { "epoch": 0.43796982459881645, "grad_norm": 4.3964338302612305, "learning_rate": 8.3572e-06, "loss": 0.0502, "step": 8215 }, { "epoch": 0.43802313802846937, "grad_norm": 2.1137473583221436, "learning_rate": 8.357000000000001e-06, "loss": 0.0326, "step": 8216 }, { "epoch": 0.4380764514581223, "grad_norm": 2.723607063293457, "learning_rate": 8.356800000000001e-06, "loss": 0.0212, "step": 8217 }, { "epoch": 0.4381297648877752, "grad_norm": 3.531168222427368, "learning_rate": 8.3566e-06, "loss": 0.0471, "step": 8218 }, { "epoch": 0.43818307831742814, "grad_norm": 2.4072675704956055, "learning_rate": 8.356400000000002e-06, "loss": 0.0444, "step": 8219 }, { "epoch": 0.43823639174708107, "grad_norm": 2.446791648864746, "learning_rate": 8.356200000000001e-06, "loss": 0.0659, "step": 8220 }, { "epoch": 0.43828970517673405, "grad_norm": 2.767916440963745, "learning_rate": 8.356000000000001e-06, "loss": 0.0652, "step": 8221 }, { "epoch": 0.43834301860638697, "grad_norm": 2.7595062255859375, "learning_rate": 8.3558e-06, "loss": 0.0558, "step": 8222 }, { "epoch": 0.4383963320360399, "grad_norm": 3.208272933959961, "learning_rate": 8.3556e-06, "loss": 0.0417, "step": 8223 }, { "epoch": 0.4384496454656928, "grad_norm": 1.98964524269104, "learning_rate": 8.3554e-06, "loss": 0.0316, "step": 8224 }, { "epoch": 0.43850295889534574, "grad_norm": 3.875138759613037, "learning_rate": 8.355200000000001e-06, "loss": 0.0462, "step": 8225 }, { "epoch": 0.43855627232499866, "grad_norm": 2.9940829277038574, "learning_rate": 8.355e-06, "loss": 0.0376, "step": 8226 }, { "epoch": 0.4386095857546516, "grad_norm": 2.284743070602417, "learning_rate": 8.3548e-06, "loss": 0.0386, "step": 8227 }, { "epoch": 0.4386628991843045, "grad_norm": 1.6212587356567383, "learning_rate": 8.354600000000001e-06, "loss": 0.0185, "step": 8228 }, { "epoch": 0.43871621261395743, "grad_norm": 1.8025883436203003, "learning_rate": 8.354400000000001e-06, "loss": 0.0413, "step": 8229 }, { "epoch": 0.43876952604361036, "grad_norm": 1.6169320344924927, "learning_rate": 8.3542e-06, "loss": 0.0141, "step": 8230 }, { "epoch": 0.43882283947326334, "grad_norm": 2.018670082092285, "learning_rate": 8.354000000000002e-06, "loss": 0.0358, "step": 8231 }, { "epoch": 0.43887615290291626, "grad_norm": 5.3020124435424805, "learning_rate": 8.353800000000001e-06, "loss": 0.0378, "step": 8232 }, { "epoch": 0.4389294663325692, "grad_norm": 4.712358474731445, "learning_rate": 8.3536e-06, "loss": 0.0603, "step": 8233 }, { "epoch": 0.4389827797622221, "grad_norm": 1.9338449239730835, "learning_rate": 8.3534e-06, "loss": 0.0452, "step": 8234 }, { "epoch": 0.43903609319187503, "grad_norm": 2.360031843185425, "learning_rate": 8.3532e-06, "loss": 0.048, "step": 8235 }, { "epoch": 0.43908940662152796, "grad_norm": 1.4196022748947144, "learning_rate": 8.353000000000001e-06, "loss": 0.0231, "step": 8236 }, { "epoch": 0.4391427200511809, "grad_norm": 3.831825017929077, "learning_rate": 8.3528e-06, "loss": 0.0504, "step": 8237 }, { "epoch": 0.4391960334808338, "grad_norm": 2.0939064025878906, "learning_rate": 8.3526e-06, "loss": 0.0258, "step": 8238 }, { "epoch": 0.4392493469104867, "grad_norm": 2.563839912414551, "learning_rate": 8.3524e-06, "loss": 0.0312, "step": 8239 }, { "epoch": 0.4393026603401397, "grad_norm": 1.84627366065979, "learning_rate": 8.352200000000001e-06, "loss": 0.0256, "step": 8240 }, { "epoch": 0.43935597376979263, "grad_norm": 1.3201874494552612, "learning_rate": 8.352e-06, "loss": 0.0167, "step": 8241 }, { "epoch": 0.43940928719944555, "grad_norm": 1.793765664100647, "learning_rate": 8.3518e-06, "loss": 0.0262, "step": 8242 }, { "epoch": 0.4394626006290985, "grad_norm": 1.9929585456848145, "learning_rate": 8.351600000000001e-06, "loss": 0.0273, "step": 8243 }, { "epoch": 0.4395159140587514, "grad_norm": 1.894404411315918, "learning_rate": 8.351400000000001e-06, "loss": 0.0434, "step": 8244 }, { "epoch": 0.4395692274884043, "grad_norm": 3.138174533843994, "learning_rate": 8.3512e-06, "loss": 0.0735, "step": 8245 }, { "epoch": 0.43962254091805725, "grad_norm": 3.277228593826294, "learning_rate": 8.351e-06, "loss": 0.0578, "step": 8246 }, { "epoch": 0.43967585434771017, "grad_norm": 2.078779458999634, "learning_rate": 8.350800000000001e-06, "loss": 0.0571, "step": 8247 }, { "epoch": 0.4397291677773631, "grad_norm": 1.4180586338043213, "learning_rate": 8.350600000000001e-06, "loss": 0.0255, "step": 8248 }, { "epoch": 0.4397824812070161, "grad_norm": 2.309401035308838, "learning_rate": 8.3504e-06, "loss": 0.0416, "step": 8249 }, { "epoch": 0.439835794636669, "grad_norm": 3.358264446258545, "learning_rate": 8.3502e-06, "loss": 0.0381, "step": 8250 }, { "epoch": 0.4398891080663219, "grad_norm": 3.479295492172241, "learning_rate": 8.35e-06, "loss": 0.0547, "step": 8251 }, { "epoch": 0.43994242149597484, "grad_norm": 2.647516965866089, "learning_rate": 8.349800000000001e-06, "loss": 0.0372, "step": 8252 }, { "epoch": 0.43999573492562777, "grad_norm": 3.442603588104248, "learning_rate": 8.3496e-06, "loss": 0.0551, "step": 8253 }, { "epoch": 0.4400490483552807, "grad_norm": 1.986000418663025, "learning_rate": 8.3494e-06, "loss": 0.0154, "step": 8254 }, { "epoch": 0.4401023617849336, "grad_norm": 2.7164037227630615, "learning_rate": 8.349200000000001e-06, "loss": 0.0401, "step": 8255 }, { "epoch": 0.44015567521458654, "grad_norm": 5.148115158081055, "learning_rate": 8.349000000000001e-06, "loss": 0.0245, "step": 8256 }, { "epoch": 0.44020898864423946, "grad_norm": 3.2888381481170654, "learning_rate": 8.3488e-06, "loss": 0.0554, "step": 8257 }, { "epoch": 0.4402623020738924, "grad_norm": 1.264635682106018, "learning_rate": 8.348600000000002e-06, "loss": 0.0117, "step": 8258 }, { "epoch": 0.44031561550354537, "grad_norm": 3.361205816268921, "learning_rate": 8.348400000000001e-06, "loss": 0.0374, "step": 8259 }, { "epoch": 0.4403689289331983, "grad_norm": 2.5223169326782227, "learning_rate": 8.3482e-06, "loss": 0.0255, "step": 8260 }, { "epoch": 0.4404222423628512, "grad_norm": 2.576869010925293, "learning_rate": 8.348e-06, "loss": 0.0614, "step": 8261 }, { "epoch": 0.44047555579250414, "grad_norm": 2.3957576751708984, "learning_rate": 8.3478e-06, "loss": 0.0494, "step": 8262 }, { "epoch": 0.44052886922215706, "grad_norm": 2.721787929534912, "learning_rate": 8.3476e-06, "loss": 0.0422, "step": 8263 }, { "epoch": 0.44058218265181, "grad_norm": 2.294384002685547, "learning_rate": 8.3474e-06, "loss": 0.0602, "step": 8264 }, { "epoch": 0.4406354960814629, "grad_norm": 2.322331428527832, "learning_rate": 8.3472e-06, "loss": 0.0305, "step": 8265 }, { "epoch": 0.44068880951111583, "grad_norm": 2.8352558612823486, "learning_rate": 8.347e-06, "loss": 0.0311, "step": 8266 }, { "epoch": 0.44074212294076875, "grad_norm": 3.0289652347564697, "learning_rate": 8.346800000000001e-06, "loss": 0.0588, "step": 8267 }, { "epoch": 0.44079543637042173, "grad_norm": 2.441068410873413, "learning_rate": 8.3466e-06, "loss": 0.0561, "step": 8268 }, { "epoch": 0.44084874980007466, "grad_norm": 2.2295961380004883, "learning_rate": 8.346400000000002e-06, "loss": 0.0422, "step": 8269 }, { "epoch": 0.4409020632297276, "grad_norm": 3.053187370300293, "learning_rate": 8.346200000000002e-06, "loss": 0.0389, "step": 8270 }, { "epoch": 0.4409553766593805, "grad_norm": 1.036980390548706, "learning_rate": 8.346000000000001e-06, "loss": 0.0197, "step": 8271 }, { "epoch": 0.44100869008903343, "grad_norm": 3.4746334552764893, "learning_rate": 8.3458e-06, "loss": 0.0423, "step": 8272 }, { "epoch": 0.44106200351868635, "grad_norm": 2.8545033931732178, "learning_rate": 8.3456e-06, "loss": 0.0331, "step": 8273 }, { "epoch": 0.4411153169483393, "grad_norm": 1.543868899345398, "learning_rate": 8.3454e-06, "loss": 0.013, "step": 8274 }, { "epoch": 0.4411686303779922, "grad_norm": 3.7941060066223145, "learning_rate": 8.345200000000001e-06, "loss": 0.0424, "step": 8275 }, { "epoch": 0.4412219438076451, "grad_norm": 3.6504666805267334, "learning_rate": 8.345e-06, "loss": 0.0662, "step": 8276 }, { "epoch": 0.4412752572372981, "grad_norm": 3.4995875358581543, "learning_rate": 8.3448e-06, "loss": 0.0633, "step": 8277 }, { "epoch": 0.441328570666951, "grad_norm": 2.566206216812134, "learning_rate": 8.3446e-06, "loss": 0.027, "step": 8278 }, { "epoch": 0.44138188409660395, "grad_norm": 2.6653857231140137, "learning_rate": 8.344400000000001e-06, "loss": 0.0507, "step": 8279 }, { "epoch": 0.4414351975262569, "grad_norm": 2.100973606109619, "learning_rate": 8.3442e-06, "loss": 0.0314, "step": 8280 }, { "epoch": 0.4414885109559098, "grad_norm": 2.4136106967926025, "learning_rate": 8.344000000000002e-06, "loss": 0.0545, "step": 8281 }, { "epoch": 0.4415418243855627, "grad_norm": 2.9166433811187744, "learning_rate": 8.343800000000001e-06, "loss": 0.0429, "step": 8282 }, { "epoch": 0.44159513781521564, "grad_norm": 5.75805139541626, "learning_rate": 8.343600000000001e-06, "loss": 0.0737, "step": 8283 }, { "epoch": 0.44164845124486857, "grad_norm": 7.3694915771484375, "learning_rate": 8.3434e-06, "loss": 0.0636, "step": 8284 }, { "epoch": 0.4417017646745215, "grad_norm": 2.1838018894195557, "learning_rate": 8.3432e-06, "loss": 0.0343, "step": 8285 }, { "epoch": 0.4417550781041744, "grad_norm": 2.6624672412872314, "learning_rate": 8.343000000000001e-06, "loss": 0.037, "step": 8286 }, { "epoch": 0.4418083915338274, "grad_norm": 1.3737542629241943, "learning_rate": 8.3428e-06, "loss": 0.0226, "step": 8287 }, { "epoch": 0.4418617049634803, "grad_norm": 1.595401406288147, "learning_rate": 8.3426e-06, "loss": 0.0233, "step": 8288 }, { "epoch": 0.44191501839313324, "grad_norm": 3.424048900604248, "learning_rate": 8.3424e-06, "loss": 0.0684, "step": 8289 }, { "epoch": 0.44196833182278616, "grad_norm": 4.098211288452148, "learning_rate": 8.3422e-06, "loss": 0.066, "step": 8290 }, { "epoch": 0.4420216452524391, "grad_norm": 3.257185935974121, "learning_rate": 8.342e-06, "loss": 0.0483, "step": 8291 }, { "epoch": 0.442074958682092, "grad_norm": 2.3038065433502197, "learning_rate": 8.3418e-06, "loss": 0.0408, "step": 8292 }, { "epoch": 0.44212827211174494, "grad_norm": 2.7490172386169434, "learning_rate": 8.341600000000002e-06, "loss": 0.0727, "step": 8293 }, { "epoch": 0.44218158554139786, "grad_norm": 2.8700709342956543, "learning_rate": 8.341400000000001e-06, "loss": 0.0248, "step": 8294 }, { "epoch": 0.4422348989710508, "grad_norm": 3.62912917137146, "learning_rate": 8.3412e-06, "loss": 0.0508, "step": 8295 }, { "epoch": 0.44228821240070376, "grad_norm": 4.137305736541748, "learning_rate": 8.341e-06, "loss": 0.0707, "step": 8296 }, { "epoch": 0.4423415258303567, "grad_norm": 4.726568698883057, "learning_rate": 8.340800000000002e-06, "loss": 0.0464, "step": 8297 }, { "epoch": 0.4423948392600096, "grad_norm": 3.1619720458984375, "learning_rate": 8.340600000000001e-06, "loss": 0.0317, "step": 8298 }, { "epoch": 0.44244815268966253, "grad_norm": 4.04552698135376, "learning_rate": 8.3404e-06, "loss": 0.045, "step": 8299 }, { "epoch": 0.44250146611931546, "grad_norm": 1.6090173721313477, "learning_rate": 8.3402e-06, "loss": 0.0227, "step": 8300 }, { "epoch": 0.4425547795489684, "grad_norm": 2.706601142883301, "learning_rate": 8.34e-06, "loss": 0.053, "step": 8301 }, { "epoch": 0.4426080929786213, "grad_norm": 2.8260159492492676, "learning_rate": 8.339800000000001e-06, "loss": 0.0386, "step": 8302 }, { "epoch": 0.4426614064082742, "grad_norm": 3.2831788063049316, "learning_rate": 8.3396e-06, "loss": 0.043, "step": 8303 }, { "epoch": 0.44271471983792715, "grad_norm": 3.278175115585327, "learning_rate": 8.3394e-06, "loss": 0.053, "step": 8304 }, { "epoch": 0.44276803326758013, "grad_norm": 5.176849842071533, "learning_rate": 8.339200000000001e-06, "loss": 0.0694, "step": 8305 }, { "epoch": 0.44282134669723305, "grad_norm": 4.8444294929504395, "learning_rate": 8.339000000000001e-06, "loss": 0.0823, "step": 8306 }, { "epoch": 0.442874660126886, "grad_norm": 2.124307632446289, "learning_rate": 8.3388e-06, "loss": 0.0314, "step": 8307 }, { "epoch": 0.4429279735565389, "grad_norm": 1.4677332639694214, "learning_rate": 8.338600000000002e-06, "loss": 0.0215, "step": 8308 }, { "epoch": 0.4429812869861918, "grad_norm": 3.187187433242798, "learning_rate": 8.338400000000001e-06, "loss": 0.0259, "step": 8309 }, { "epoch": 0.44303460041584475, "grad_norm": 3.9123072624206543, "learning_rate": 8.338200000000001e-06, "loss": 0.0398, "step": 8310 }, { "epoch": 0.44308791384549767, "grad_norm": 6.047586917877197, "learning_rate": 8.338e-06, "loss": 0.0686, "step": 8311 }, { "epoch": 0.4431412272751506, "grad_norm": 4.064803123474121, "learning_rate": 8.3378e-06, "loss": 0.0414, "step": 8312 }, { "epoch": 0.4431945407048035, "grad_norm": 2.671210765838623, "learning_rate": 8.3376e-06, "loss": 0.0392, "step": 8313 }, { "epoch": 0.44324785413445644, "grad_norm": 1.545985221862793, "learning_rate": 8.3374e-06, "loss": 0.0399, "step": 8314 }, { "epoch": 0.4433011675641094, "grad_norm": 2.18176531791687, "learning_rate": 8.3372e-06, "loss": 0.0191, "step": 8315 }, { "epoch": 0.44335448099376235, "grad_norm": 3.5188097953796387, "learning_rate": 8.337e-06, "loss": 0.073, "step": 8316 }, { "epoch": 0.44340779442341527, "grad_norm": 2.8546793460845947, "learning_rate": 8.336800000000001e-06, "loss": 0.029, "step": 8317 }, { "epoch": 0.4434611078530682, "grad_norm": 3.234523057937622, "learning_rate": 8.3366e-06, "loss": 0.0213, "step": 8318 }, { "epoch": 0.4435144212827211, "grad_norm": 2.053530216217041, "learning_rate": 8.3364e-06, "loss": 0.036, "step": 8319 }, { "epoch": 0.44356773471237404, "grad_norm": 1.7027122974395752, "learning_rate": 8.336200000000002e-06, "loss": 0.0228, "step": 8320 }, { "epoch": 0.44362104814202696, "grad_norm": 3.0693094730377197, "learning_rate": 8.336000000000001e-06, "loss": 0.0265, "step": 8321 }, { "epoch": 0.4436743615716799, "grad_norm": 1.9592632055282593, "learning_rate": 8.3358e-06, "loss": 0.0258, "step": 8322 }, { "epoch": 0.4437276750013328, "grad_norm": 3.7617952823638916, "learning_rate": 8.3356e-06, "loss": 0.0741, "step": 8323 }, { "epoch": 0.4437809884309858, "grad_norm": 2.8435862064361572, "learning_rate": 8.3354e-06, "loss": 0.032, "step": 8324 }, { "epoch": 0.4438343018606387, "grad_norm": 1.8675832748413086, "learning_rate": 8.335200000000001e-06, "loss": 0.04, "step": 8325 }, { "epoch": 0.44388761529029164, "grad_norm": 3.467386484146118, "learning_rate": 8.335e-06, "loss": 0.0496, "step": 8326 }, { "epoch": 0.44394092871994456, "grad_norm": 1.3846319913864136, "learning_rate": 8.3348e-06, "loss": 0.0219, "step": 8327 }, { "epoch": 0.4439942421495975, "grad_norm": 1.8418688774108887, "learning_rate": 8.3346e-06, "loss": 0.0218, "step": 8328 }, { "epoch": 0.4440475555792504, "grad_norm": 2.2543487548828125, "learning_rate": 8.334400000000001e-06, "loss": 0.0229, "step": 8329 }, { "epoch": 0.44410086900890333, "grad_norm": 8.179606437683105, "learning_rate": 8.3342e-06, "loss": 0.0607, "step": 8330 }, { "epoch": 0.44415418243855626, "grad_norm": 2.5113980770111084, "learning_rate": 8.334e-06, "loss": 0.0205, "step": 8331 }, { "epoch": 0.4442074958682092, "grad_norm": 5.785793781280518, "learning_rate": 8.333800000000001e-06, "loss": 0.026, "step": 8332 }, { "epoch": 0.44426080929786216, "grad_norm": 3.531644344329834, "learning_rate": 8.333600000000001e-06, "loss": 0.0475, "step": 8333 }, { "epoch": 0.4443141227275151, "grad_norm": 3.117568254470825, "learning_rate": 8.3334e-06, "loss": 0.0512, "step": 8334 }, { "epoch": 0.444367436157168, "grad_norm": 5.1380743980407715, "learning_rate": 8.3332e-06, "loss": 0.0433, "step": 8335 }, { "epoch": 0.44442074958682093, "grad_norm": 7.0030717849731445, "learning_rate": 8.333000000000001e-06, "loss": 0.0661, "step": 8336 }, { "epoch": 0.44447406301647385, "grad_norm": 2.691054582595825, "learning_rate": 8.332800000000001e-06, "loss": 0.0362, "step": 8337 }, { "epoch": 0.4445273764461268, "grad_norm": 2.17463755607605, "learning_rate": 8.3326e-06, "loss": 0.0599, "step": 8338 }, { "epoch": 0.4445806898757797, "grad_norm": 3.7759547233581543, "learning_rate": 8.3324e-06, "loss": 0.0731, "step": 8339 }, { "epoch": 0.4446340033054326, "grad_norm": 1.4814568758010864, "learning_rate": 8.3322e-06, "loss": 0.0114, "step": 8340 }, { "epoch": 0.44468731673508555, "grad_norm": 2.7715041637420654, "learning_rate": 8.332000000000001e-06, "loss": 0.0569, "step": 8341 }, { "epoch": 0.44474063016473847, "grad_norm": 1.4814174175262451, "learning_rate": 8.3318e-06, "loss": 0.0244, "step": 8342 }, { "epoch": 0.44479394359439145, "grad_norm": 2.269552707672119, "learning_rate": 8.3316e-06, "loss": 0.026, "step": 8343 }, { "epoch": 0.4448472570240444, "grad_norm": 3.147174596786499, "learning_rate": 8.331400000000001e-06, "loss": 0.0369, "step": 8344 }, { "epoch": 0.4449005704536973, "grad_norm": 3.340325117111206, "learning_rate": 8.3312e-06, "loss": 0.0573, "step": 8345 }, { "epoch": 0.4449538838833502, "grad_norm": 3.7646148204803467, "learning_rate": 8.331e-06, "loss": 0.055, "step": 8346 }, { "epoch": 0.44500719731300314, "grad_norm": 4.193481922149658, "learning_rate": 8.330800000000002e-06, "loss": 0.0571, "step": 8347 }, { "epoch": 0.44506051074265607, "grad_norm": 2.646042585372925, "learning_rate": 8.330600000000001e-06, "loss": 0.0214, "step": 8348 }, { "epoch": 0.445113824172309, "grad_norm": 0.9111601114273071, "learning_rate": 8.3304e-06, "loss": 0.0103, "step": 8349 }, { "epoch": 0.4451671376019619, "grad_norm": 1.3372763395309448, "learning_rate": 8.3302e-06, "loss": 0.0228, "step": 8350 }, { "epoch": 0.44522045103161484, "grad_norm": 2.287134885787964, "learning_rate": 8.33e-06, "loss": 0.0278, "step": 8351 }, { "epoch": 0.4452737644612678, "grad_norm": 1.96882164478302, "learning_rate": 8.329800000000001e-06, "loss": 0.0508, "step": 8352 }, { "epoch": 0.44532707789092074, "grad_norm": 1.7463568449020386, "learning_rate": 8.3296e-06, "loss": 0.0211, "step": 8353 }, { "epoch": 0.44538039132057367, "grad_norm": 1.9099611043930054, "learning_rate": 8.3294e-06, "loss": 0.0345, "step": 8354 }, { "epoch": 0.4454337047502266, "grad_norm": 2.426882266998291, "learning_rate": 8.3292e-06, "loss": 0.044, "step": 8355 }, { "epoch": 0.4454870181798795, "grad_norm": 3.8756182193756104, "learning_rate": 8.329000000000001e-06, "loss": 0.0433, "step": 8356 }, { "epoch": 0.44554033160953244, "grad_norm": 3.598747730255127, "learning_rate": 8.3288e-06, "loss": 0.0607, "step": 8357 }, { "epoch": 0.44559364503918536, "grad_norm": 4.528536796569824, "learning_rate": 8.328600000000002e-06, "loss": 0.0354, "step": 8358 }, { "epoch": 0.4456469584688383, "grad_norm": 1.845137119293213, "learning_rate": 8.328400000000001e-06, "loss": 0.0354, "step": 8359 }, { "epoch": 0.4457002718984912, "grad_norm": 2.1308200359344482, "learning_rate": 8.328200000000001e-06, "loss": 0.035, "step": 8360 }, { "epoch": 0.4457535853281442, "grad_norm": 3.6243577003479004, "learning_rate": 8.328e-06, "loss": 0.0329, "step": 8361 }, { "epoch": 0.4458068987577971, "grad_norm": 4.598294258117676, "learning_rate": 8.3278e-06, "loss": 0.0473, "step": 8362 }, { "epoch": 0.44586021218745003, "grad_norm": 3.4748477935791016, "learning_rate": 8.3276e-06, "loss": 0.0314, "step": 8363 }, { "epoch": 0.44591352561710296, "grad_norm": 3.951195478439331, "learning_rate": 8.327400000000001e-06, "loss": 0.0265, "step": 8364 }, { "epoch": 0.4459668390467559, "grad_norm": 1.581218957901001, "learning_rate": 8.3272e-06, "loss": 0.0298, "step": 8365 }, { "epoch": 0.4460201524764088, "grad_norm": 2.9099390506744385, "learning_rate": 8.327e-06, "loss": 0.0501, "step": 8366 }, { "epoch": 0.44607346590606173, "grad_norm": 1.7564384937286377, "learning_rate": 8.3268e-06, "loss": 0.0172, "step": 8367 }, { "epoch": 0.44612677933571465, "grad_norm": 3.036344289779663, "learning_rate": 8.326600000000001e-06, "loss": 0.0531, "step": 8368 }, { "epoch": 0.4461800927653676, "grad_norm": 3.151431083679199, "learning_rate": 8.3264e-06, "loss": 0.0435, "step": 8369 }, { "epoch": 0.4462334061950205, "grad_norm": 3.7525837421417236, "learning_rate": 8.326200000000002e-06, "loss": 0.0487, "step": 8370 }, { "epoch": 0.4462867196246735, "grad_norm": 2.870908498764038, "learning_rate": 8.326000000000001e-06, "loss": 0.0315, "step": 8371 }, { "epoch": 0.4463400330543264, "grad_norm": 2.303901195526123, "learning_rate": 8.3258e-06, "loss": 0.0374, "step": 8372 }, { "epoch": 0.4463933464839793, "grad_norm": 3.2823092937469482, "learning_rate": 8.3256e-06, "loss": 0.0526, "step": 8373 }, { "epoch": 0.44644665991363225, "grad_norm": 1.697100281715393, "learning_rate": 8.3254e-06, "loss": 0.0155, "step": 8374 }, { "epoch": 0.4464999733432852, "grad_norm": 2.0300381183624268, "learning_rate": 8.325200000000001e-06, "loss": 0.039, "step": 8375 }, { "epoch": 0.4465532867729381, "grad_norm": 3.43697190284729, "learning_rate": 8.325e-06, "loss": 0.042, "step": 8376 }, { "epoch": 0.446606600202591, "grad_norm": 4.984198093414307, "learning_rate": 8.3248e-06, "loss": 0.0344, "step": 8377 }, { "epoch": 0.44665991363224394, "grad_norm": 2.7072901725769043, "learning_rate": 8.3246e-06, "loss": 0.0212, "step": 8378 }, { "epoch": 0.44671322706189687, "grad_norm": 3.2215230464935303, "learning_rate": 8.3244e-06, "loss": 0.0544, "step": 8379 }, { "epoch": 0.44676654049154985, "grad_norm": 2.4354755878448486, "learning_rate": 8.3242e-06, "loss": 0.0354, "step": 8380 }, { "epoch": 0.44681985392120277, "grad_norm": 2.2981255054473877, "learning_rate": 8.324e-06, "loss": 0.0194, "step": 8381 }, { "epoch": 0.4468731673508557, "grad_norm": 2.636479377746582, "learning_rate": 8.323800000000002e-06, "loss": 0.0506, "step": 8382 }, { "epoch": 0.4469264807805086, "grad_norm": 2.4193994998931885, "learning_rate": 8.323600000000001e-06, "loss": 0.0337, "step": 8383 }, { "epoch": 0.44697979421016154, "grad_norm": 3.0913617610931396, "learning_rate": 8.3234e-06, "loss": 0.0494, "step": 8384 }, { "epoch": 0.44703310763981446, "grad_norm": 2.443920850753784, "learning_rate": 8.3232e-06, "loss": 0.0269, "step": 8385 }, { "epoch": 0.4470864210694674, "grad_norm": 2.2088098526000977, "learning_rate": 8.323000000000001e-06, "loss": 0.017, "step": 8386 }, { "epoch": 0.4471397344991203, "grad_norm": 3.0628466606140137, "learning_rate": 8.322800000000001e-06, "loss": 0.0423, "step": 8387 }, { "epoch": 0.44719304792877324, "grad_norm": 3.2637882232666016, "learning_rate": 8.3226e-06, "loss": 0.0559, "step": 8388 }, { "epoch": 0.4472463613584262, "grad_norm": 6.323810577392578, "learning_rate": 8.3224e-06, "loss": 0.0665, "step": 8389 }, { "epoch": 0.44729967478807914, "grad_norm": 3.518815040588379, "learning_rate": 8.3222e-06, "loss": 0.0283, "step": 8390 }, { "epoch": 0.44735298821773206, "grad_norm": 2.5154900550842285, "learning_rate": 8.322000000000001e-06, "loss": 0.0167, "step": 8391 }, { "epoch": 0.447406301647385, "grad_norm": 3.2989931106567383, "learning_rate": 8.3218e-06, "loss": 0.0446, "step": 8392 }, { "epoch": 0.4474596150770379, "grad_norm": 2.3481385707855225, "learning_rate": 8.3216e-06, "loss": 0.0386, "step": 8393 }, { "epoch": 0.44751292850669083, "grad_norm": 1.4517978429794312, "learning_rate": 8.321400000000001e-06, "loss": 0.03, "step": 8394 }, { "epoch": 0.44756624193634376, "grad_norm": 2.8207318782806396, "learning_rate": 8.321200000000001e-06, "loss": 0.052, "step": 8395 }, { "epoch": 0.4476195553659967, "grad_norm": 1.7273985147476196, "learning_rate": 8.321e-06, "loss": 0.0284, "step": 8396 }, { "epoch": 0.4476728687956496, "grad_norm": 4.601938247680664, "learning_rate": 8.320800000000002e-06, "loss": 0.0647, "step": 8397 }, { "epoch": 0.4477261822253025, "grad_norm": 2.496528387069702, "learning_rate": 8.320600000000001e-06, "loss": 0.0435, "step": 8398 }, { "epoch": 0.4477794956549555, "grad_norm": 2.791288375854492, "learning_rate": 8.3204e-06, "loss": 0.035, "step": 8399 }, { "epoch": 0.44783280908460843, "grad_norm": 3.5604522228240967, "learning_rate": 8.3202e-06, "loss": 0.0404, "step": 8400 }, { "epoch": 0.44788612251426135, "grad_norm": 1.659324288368225, "learning_rate": 8.32e-06, "loss": 0.0275, "step": 8401 }, { "epoch": 0.4479394359439143, "grad_norm": 1.500111699104309, "learning_rate": 8.319800000000001e-06, "loss": 0.0342, "step": 8402 }, { "epoch": 0.4479927493735672, "grad_norm": 3.0216307640075684, "learning_rate": 8.3196e-06, "loss": 0.0498, "step": 8403 }, { "epoch": 0.4480460628032201, "grad_norm": 2.085836887359619, "learning_rate": 8.3194e-06, "loss": 0.0232, "step": 8404 }, { "epoch": 0.44809937623287305, "grad_norm": 3.7400267124176025, "learning_rate": 8.3192e-06, "loss": 0.0285, "step": 8405 }, { "epoch": 0.44815268966252597, "grad_norm": 6.079769134521484, "learning_rate": 8.319000000000001e-06, "loss": 0.0496, "step": 8406 }, { "epoch": 0.4482060030921789, "grad_norm": 6.387768268585205, "learning_rate": 8.3188e-06, "loss": 0.061, "step": 8407 }, { "epoch": 0.4482593165218319, "grad_norm": 2.24079966545105, "learning_rate": 8.3186e-06, "loss": 0.0216, "step": 8408 }, { "epoch": 0.4483126299514848, "grad_norm": 2.480901002883911, "learning_rate": 8.318400000000002e-06, "loss": 0.037, "step": 8409 }, { "epoch": 0.4483659433811377, "grad_norm": 1.5131219625473022, "learning_rate": 8.318200000000001e-06, "loss": 0.0304, "step": 8410 }, { "epoch": 0.44841925681079065, "grad_norm": 2.8909764289855957, "learning_rate": 8.318e-06, "loss": 0.0727, "step": 8411 }, { "epoch": 0.44847257024044357, "grad_norm": 3.8509252071380615, "learning_rate": 8.3178e-06, "loss": 0.0401, "step": 8412 }, { "epoch": 0.4485258836700965, "grad_norm": 4.540992259979248, "learning_rate": 8.317600000000001e-06, "loss": 0.0393, "step": 8413 }, { "epoch": 0.4485791970997494, "grad_norm": 3.193894147872925, "learning_rate": 8.317400000000001e-06, "loss": 0.0482, "step": 8414 }, { "epoch": 0.44863251052940234, "grad_norm": 3.012427806854248, "learning_rate": 8.3172e-06, "loss": 0.0425, "step": 8415 }, { "epoch": 0.44868582395905526, "grad_norm": 3.772188663482666, "learning_rate": 8.317e-06, "loss": 0.0628, "step": 8416 }, { "epoch": 0.44873913738870824, "grad_norm": 1.9385563135147095, "learning_rate": 8.3168e-06, "loss": 0.0211, "step": 8417 }, { "epoch": 0.44879245081836117, "grad_norm": 6.206301689147949, "learning_rate": 8.316600000000001e-06, "loss": 0.0535, "step": 8418 }, { "epoch": 0.4488457642480141, "grad_norm": 7.134018421173096, "learning_rate": 8.3164e-06, "loss": 0.0644, "step": 8419 }, { "epoch": 0.448899077677667, "grad_norm": 6.148384094238281, "learning_rate": 8.3162e-06, "loss": 0.0449, "step": 8420 }, { "epoch": 0.44895239110731994, "grad_norm": 3.4385852813720703, "learning_rate": 8.316000000000001e-06, "loss": 0.0548, "step": 8421 }, { "epoch": 0.44900570453697286, "grad_norm": 2.26837158203125, "learning_rate": 8.315800000000001e-06, "loss": 0.0448, "step": 8422 }, { "epoch": 0.4490590179666258, "grad_norm": 2.4741177558898926, "learning_rate": 8.3156e-06, "loss": 0.0575, "step": 8423 }, { "epoch": 0.4491123313962787, "grad_norm": 3.803778648376465, "learning_rate": 8.3154e-06, "loss": 0.071, "step": 8424 }, { "epoch": 0.44916564482593163, "grad_norm": 1.8485087156295776, "learning_rate": 8.315200000000001e-06, "loss": 0.0361, "step": 8425 }, { "epoch": 0.44921895825558456, "grad_norm": 2.1006228923797607, "learning_rate": 8.315000000000001e-06, "loss": 0.0239, "step": 8426 }, { "epoch": 0.44927227168523753, "grad_norm": 1.4979134798049927, "learning_rate": 8.3148e-06, "loss": 0.0256, "step": 8427 }, { "epoch": 0.44932558511489046, "grad_norm": 2.4115283489227295, "learning_rate": 8.3146e-06, "loss": 0.0291, "step": 8428 }, { "epoch": 0.4493788985445434, "grad_norm": 3.9185237884521484, "learning_rate": 8.3144e-06, "loss": 0.0464, "step": 8429 }, { "epoch": 0.4494322119741963, "grad_norm": 1.1249384880065918, "learning_rate": 8.3142e-06, "loss": 0.0183, "step": 8430 }, { "epoch": 0.44948552540384923, "grad_norm": 2.587709665298462, "learning_rate": 8.314e-06, "loss": 0.03, "step": 8431 }, { "epoch": 0.44953883883350215, "grad_norm": 1.6264851093292236, "learning_rate": 8.3138e-06, "loss": 0.0272, "step": 8432 }, { "epoch": 0.4495921522631551, "grad_norm": 1.4429045915603638, "learning_rate": 8.313600000000001e-06, "loss": 0.0198, "step": 8433 }, { "epoch": 0.449645465692808, "grad_norm": 1.9460982084274292, "learning_rate": 8.3134e-06, "loss": 0.0233, "step": 8434 }, { "epoch": 0.4496987791224609, "grad_norm": 4.535577774047852, "learning_rate": 8.3132e-06, "loss": 0.0448, "step": 8435 }, { "epoch": 0.4497520925521139, "grad_norm": 2.3548967838287354, "learning_rate": 8.313000000000002e-06, "loss": 0.0466, "step": 8436 }, { "epoch": 0.4498054059817668, "grad_norm": 3.394737482070923, "learning_rate": 8.312800000000001e-06, "loss": 0.0253, "step": 8437 }, { "epoch": 0.44985871941141975, "grad_norm": 2.4947245121002197, "learning_rate": 8.3126e-06, "loss": 0.0336, "step": 8438 }, { "epoch": 0.4499120328410727, "grad_norm": 2.8641529083251953, "learning_rate": 8.3124e-06, "loss": 0.0388, "step": 8439 }, { "epoch": 0.4499653462707256, "grad_norm": 4.2757673263549805, "learning_rate": 8.3122e-06, "loss": 0.0358, "step": 8440 }, { "epoch": 0.4500186597003785, "grad_norm": 2.9769206047058105, "learning_rate": 8.312000000000001e-06, "loss": 0.0451, "step": 8441 }, { "epoch": 0.45007197313003144, "grad_norm": 2.149726152420044, "learning_rate": 8.3118e-06, "loss": 0.0467, "step": 8442 }, { "epoch": 0.45012528655968437, "grad_norm": 1.3535969257354736, "learning_rate": 8.3116e-06, "loss": 0.0196, "step": 8443 }, { "epoch": 0.4501785999893373, "grad_norm": 2.7757809162139893, "learning_rate": 8.3114e-06, "loss": 0.0276, "step": 8444 }, { "epoch": 0.45023191341899027, "grad_norm": 4.35052490234375, "learning_rate": 8.311200000000001e-06, "loss": 0.0497, "step": 8445 }, { "epoch": 0.4502852268486432, "grad_norm": 2.532155752182007, "learning_rate": 8.311e-06, "loss": 0.0461, "step": 8446 }, { "epoch": 0.4503385402782961, "grad_norm": 3.6953539848327637, "learning_rate": 8.310800000000002e-06, "loss": 0.0249, "step": 8447 }, { "epoch": 0.45039185370794904, "grad_norm": 1.1541496515274048, "learning_rate": 8.310600000000001e-06, "loss": 0.0129, "step": 8448 }, { "epoch": 0.45044516713760197, "grad_norm": 3.8786067962646484, "learning_rate": 8.310400000000001e-06, "loss": 0.0751, "step": 8449 }, { "epoch": 0.4504984805672549, "grad_norm": 2.2409214973449707, "learning_rate": 8.3102e-06, "loss": 0.0295, "step": 8450 }, { "epoch": 0.4505517939969078, "grad_norm": 4.497528076171875, "learning_rate": 8.31e-06, "loss": 0.0431, "step": 8451 }, { "epoch": 0.45060510742656074, "grad_norm": 2.5668208599090576, "learning_rate": 8.309800000000001e-06, "loss": 0.0299, "step": 8452 }, { "epoch": 0.45065842085621366, "grad_norm": 2.7907257080078125, "learning_rate": 8.309600000000001e-06, "loss": 0.0701, "step": 8453 }, { "epoch": 0.4507117342858666, "grad_norm": 1.7905529737472534, "learning_rate": 8.3094e-06, "loss": 0.0429, "step": 8454 }, { "epoch": 0.45076504771551956, "grad_norm": 2.733867645263672, "learning_rate": 8.3092e-06, "loss": 0.0303, "step": 8455 }, { "epoch": 0.4508183611451725, "grad_norm": 3.2049148082733154, "learning_rate": 8.309e-06, "loss": 0.0341, "step": 8456 }, { "epoch": 0.4508716745748254, "grad_norm": 2.9587981700897217, "learning_rate": 8.3088e-06, "loss": 0.043, "step": 8457 }, { "epoch": 0.45092498800447833, "grad_norm": 2.56498122215271, "learning_rate": 8.3086e-06, "loss": 0.0515, "step": 8458 }, { "epoch": 0.45097830143413126, "grad_norm": 2.5535314083099365, "learning_rate": 8.308400000000002e-06, "loss": 0.0338, "step": 8459 }, { "epoch": 0.4510316148637842, "grad_norm": 0.9561753869056702, "learning_rate": 8.308200000000001e-06, "loss": 0.0112, "step": 8460 }, { "epoch": 0.4510849282934371, "grad_norm": 2.122971773147583, "learning_rate": 8.308e-06, "loss": 0.0129, "step": 8461 }, { "epoch": 0.45113824172309, "grad_norm": 2.9790961742401123, "learning_rate": 8.3078e-06, "loss": 0.0155, "step": 8462 }, { "epoch": 0.45119155515274295, "grad_norm": 4.6757941246032715, "learning_rate": 8.307600000000002e-06, "loss": 0.0796, "step": 8463 }, { "epoch": 0.45124486858239593, "grad_norm": 1.7616560459136963, "learning_rate": 8.307400000000001e-06, "loss": 0.0161, "step": 8464 }, { "epoch": 0.45129818201204885, "grad_norm": 1.3657200336456299, "learning_rate": 8.3072e-06, "loss": 0.0222, "step": 8465 }, { "epoch": 0.4513514954417018, "grad_norm": 3.974256753921509, "learning_rate": 8.307e-06, "loss": 0.0487, "step": 8466 }, { "epoch": 0.4514048088713547, "grad_norm": 1.8243285417556763, "learning_rate": 8.3068e-06, "loss": 0.0392, "step": 8467 }, { "epoch": 0.4514581223010076, "grad_norm": 1.9559403657913208, "learning_rate": 8.3066e-06, "loss": 0.0234, "step": 8468 }, { "epoch": 0.45151143573066055, "grad_norm": 1.965814471244812, "learning_rate": 8.3064e-06, "loss": 0.018, "step": 8469 }, { "epoch": 0.45156474916031347, "grad_norm": 8.260003089904785, "learning_rate": 8.3062e-06, "loss": 0.1016, "step": 8470 }, { "epoch": 0.4516180625899664, "grad_norm": 2.5282862186431885, "learning_rate": 8.306000000000001e-06, "loss": 0.0226, "step": 8471 }, { "epoch": 0.4516713760196193, "grad_norm": 1.3737198114395142, "learning_rate": 8.305800000000001e-06, "loss": 0.0116, "step": 8472 }, { "epoch": 0.4517246894492723, "grad_norm": 2.8663463592529297, "learning_rate": 8.3056e-06, "loss": 0.05, "step": 8473 }, { "epoch": 0.4517780028789252, "grad_norm": 2.55890154838562, "learning_rate": 8.3054e-06, "loss": 0.0321, "step": 8474 }, { "epoch": 0.45183131630857815, "grad_norm": 3.113136053085327, "learning_rate": 8.305200000000001e-06, "loss": 0.0487, "step": 8475 }, { "epoch": 0.45188462973823107, "grad_norm": 1.9382611513137817, "learning_rate": 8.305000000000001e-06, "loss": 0.0273, "step": 8476 }, { "epoch": 0.451937943167884, "grad_norm": 1.2494372129440308, "learning_rate": 8.3048e-06, "loss": 0.0248, "step": 8477 }, { "epoch": 0.4519912565975369, "grad_norm": 2.5149009227752686, "learning_rate": 8.3046e-06, "loss": 0.0441, "step": 8478 }, { "epoch": 0.45204457002718984, "grad_norm": 1.8900272846221924, "learning_rate": 8.3044e-06, "loss": 0.0233, "step": 8479 }, { "epoch": 0.45209788345684276, "grad_norm": 1.9982752799987793, "learning_rate": 8.304200000000001e-06, "loss": 0.0354, "step": 8480 }, { "epoch": 0.4521511968864957, "grad_norm": 2.7993505001068115, "learning_rate": 8.304e-06, "loss": 0.0298, "step": 8481 }, { "epoch": 0.4522045103161486, "grad_norm": 1.7241355180740356, "learning_rate": 8.3038e-06, "loss": 0.0386, "step": 8482 }, { "epoch": 0.4522578237458016, "grad_norm": 2.8773839473724365, "learning_rate": 8.303600000000001e-06, "loss": 0.0314, "step": 8483 }, { "epoch": 0.4523111371754545, "grad_norm": 3.13378643989563, "learning_rate": 8.3034e-06, "loss": 0.041, "step": 8484 }, { "epoch": 0.45236445060510744, "grad_norm": 4.61952543258667, "learning_rate": 8.3032e-06, "loss": 0.0602, "step": 8485 }, { "epoch": 0.45241776403476036, "grad_norm": 2.8279833793640137, "learning_rate": 8.303000000000002e-06, "loss": 0.0377, "step": 8486 }, { "epoch": 0.4524710774644133, "grad_norm": 1.6355217695236206, "learning_rate": 8.302800000000001e-06, "loss": 0.0178, "step": 8487 }, { "epoch": 0.4525243908940662, "grad_norm": 4.170669078826904, "learning_rate": 8.3026e-06, "loss": 0.0365, "step": 8488 }, { "epoch": 0.45257770432371913, "grad_norm": 2.9722626209259033, "learning_rate": 8.3024e-06, "loss": 0.0168, "step": 8489 }, { "epoch": 0.45263101775337206, "grad_norm": 2.971527576446533, "learning_rate": 8.3022e-06, "loss": 0.0403, "step": 8490 }, { "epoch": 0.452684331183025, "grad_norm": 3.131147861480713, "learning_rate": 8.302000000000001e-06, "loss": 0.0547, "step": 8491 }, { "epoch": 0.45273764461267796, "grad_norm": 3.308819532394409, "learning_rate": 8.3018e-06, "loss": 0.053, "step": 8492 }, { "epoch": 0.4527909580423309, "grad_norm": 2.5348737239837646, "learning_rate": 8.3016e-06, "loss": 0.0387, "step": 8493 }, { "epoch": 0.4528442714719838, "grad_norm": 7.302124977111816, "learning_rate": 8.3014e-06, "loss": 0.0764, "step": 8494 }, { "epoch": 0.45289758490163673, "grad_norm": 1.7461906671524048, "learning_rate": 8.301200000000001e-06, "loss": 0.0243, "step": 8495 }, { "epoch": 0.45295089833128965, "grad_norm": 2.9355881214141846, "learning_rate": 8.301e-06, "loss": 0.031, "step": 8496 }, { "epoch": 0.4530042117609426, "grad_norm": 1.4901515245437622, "learning_rate": 8.300800000000002e-06, "loss": 0.0327, "step": 8497 }, { "epoch": 0.4530575251905955, "grad_norm": 2.3861587047576904, "learning_rate": 8.300600000000001e-06, "loss": 0.0354, "step": 8498 }, { "epoch": 0.4531108386202484, "grad_norm": 2.6686830520629883, "learning_rate": 8.300400000000001e-06, "loss": 0.0225, "step": 8499 }, { "epoch": 0.45316415204990135, "grad_norm": 3.6867873668670654, "learning_rate": 8.3002e-06, "loss": 0.0362, "step": 8500 }, { "epoch": 0.4532174654795543, "grad_norm": 2.0196733474731445, "learning_rate": 8.3e-06, "loss": 0.0195, "step": 8501 }, { "epoch": 0.45327077890920725, "grad_norm": 2.5125584602355957, "learning_rate": 8.299800000000001e-06, "loss": 0.0401, "step": 8502 }, { "epoch": 0.4533240923388602, "grad_norm": 1.873447299003601, "learning_rate": 8.299600000000001e-06, "loss": 0.0122, "step": 8503 }, { "epoch": 0.4533774057685131, "grad_norm": 2.8596572875976562, "learning_rate": 8.2994e-06, "loss": 0.0656, "step": 8504 }, { "epoch": 0.453430719198166, "grad_norm": 1.7898527383804321, "learning_rate": 8.2992e-06, "loss": 0.0212, "step": 8505 }, { "epoch": 0.45348403262781894, "grad_norm": 1.8069108724594116, "learning_rate": 8.299e-06, "loss": 0.0263, "step": 8506 }, { "epoch": 0.45353734605747187, "grad_norm": 2.3757882118225098, "learning_rate": 8.298800000000001e-06, "loss": 0.0434, "step": 8507 }, { "epoch": 0.4535906594871248, "grad_norm": 5.098515033721924, "learning_rate": 8.2986e-06, "loss": 0.0825, "step": 8508 }, { "epoch": 0.4536439729167777, "grad_norm": 4.645260810852051, "learning_rate": 8.298400000000002e-06, "loss": 0.0647, "step": 8509 }, { "epoch": 0.45369728634643064, "grad_norm": 3.1290884017944336, "learning_rate": 8.298200000000001e-06, "loss": 0.0368, "step": 8510 }, { "epoch": 0.4537505997760836, "grad_norm": 3.394796133041382, "learning_rate": 8.298000000000001e-06, "loss": 0.0394, "step": 8511 }, { "epoch": 0.45380391320573654, "grad_norm": 4.168704509735107, "learning_rate": 8.2978e-06, "loss": 0.0575, "step": 8512 }, { "epoch": 0.45385722663538947, "grad_norm": 4.216591835021973, "learning_rate": 8.297600000000002e-06, "loss": 0.0321, "step": 8513 }, { "epoch": 0.4539105400650424, "grad_norm": 2.481827974319458, "learning_rate": 8.297400000000001e-06, "loss": 0.0429, "step": 8514 }, { "epoch": 0.4539638534946953, "grad_norm": 3.140573501586914, "learning_rate": 8.2972e-06, "loss": 0.0502, "step": 8515 }, { "epoch": 0.45401716692434824, "grad_norm": 3.558847427368164, "learning_rate": 8.297e-06, "loss": 0.0417, "step": 8516 }, { "epoch": 0.45407048035400116, "grad_norm": 3.238311290740967, "learning_rate": 8.2968e-06, "loss": 0.0388, "step": 8517 }, { "epoch": 0.4541237937836541, "grad_norm": 2.992525577545166, "learning_rate": 8.2966e-06, "loss": 0.0385, "step": 8518 }, { "epoch": 0.454177107213307, "grad_norm": 1.384198546409607, "learning_rate": 8.2964e-06, "loss": 0.0137, "step": 8519 }, { "epoch": 0.45423042064296, "grad_norm": 3.594938039779663, "learning_rate": 8.2962e-06, "loss": 0.0302, "step": 8520 }, { "epoch": 0.4542837340726129, "grad_norm": 2.899366855621338, "learning_rate": 8.296000000000002e-06, "loss": 0.033, "step": 8521 }, { "epoch": 0.45433704750226583, "grad_norm": 3.1672604084014893, "learning_rate": 8.295800000000001e-06, "loss": 0.0481, "step": 8522 }, { "epoch": 0.45439036093191876, "grad_norm": 1.963823676109314, "learning_rate": 8.2956e-06, "loss": 0.0349, "step": 8523 }, { "epoch": 0.4544436743615717, "grad_norm": 1.7809951305389404, "learning_rate": 8.295400000000002e-06, "loss": 0.0234, "step": 8524 }, { "epoch": 0.4544969877912246, "grad_norm": 1.6236543655395508, "learning_rate": 8.295200000000001e-06, "loss": 0.033, "step": 8525 }, { "epoch": 0.45455030122087753, "grad_norm": 3.8352367877960205, "learning_rate": 8.295000000000001e-06, "loss": 0.0612, "step": 8526 }, { "epoch": 0.45460361465053045, "grad_norm": 2.6676671504974365, "learning_rate": 8.2948e-06, "loss": 0.0337, "step": 8527 }, { "epoch": 0.4546569280801834, "grad_norm": 2.950155019760132, "learning_rate": 8.2946e-06, "loss": 0.041, "step": 8528 }, { "epoch": 0.45471024150983635, "grad_norm": 2.874100685119629, "learning_rate": 8.2944e-06, "loss": 0.0441, "step": 8529 }, { "epoch": 0.4547635549394893, "grad_norm": 1.698062777519226, "learning_rate": 8.294200000000001e-06, "loss": 0.0216, "step": 8530 }, { "epoch": 0.4548168683691422, "grad_norm": 2.5776963233947754, "learning_rate": 8.294e-06, "loss": 0.029, "step": 8531 }, { "epoch": 0.4548701817987951, "grad_norm": 3.4383888244628906, "learning_rate": 8.2938e-06, "loss": 0.0547, "step": 8532 }, { "epoch": 0.45492349522844805, "grad_norm": 2.14719295501709, "learning_rate": 8.293600000000001e-06, "loss": 0.038, "step": 8533 }, { "epoch": 0.454976808658101, "grad_norm": 2.5781807899475098, "learning_rate": 8.293400000000001e-06, "loss": 0.0484, "step": 8534 }, { "epoch": 0.4550301220877539, "grad_norm": 1.519331455230713, "learning_rate": 8.2932e-06, "loss": 0.0243, "step": 8535 }, { "epoch": 0.4550834355174068, "grad_norm": 1.7986621856689453, "learning_rate": 8.293000000000002e-06, "loss": 0.03, "step": 8536 }, { "epoch": 0.45513674894705974, "grad_norm": 2.833828926086426, "learning_rate": 8.292800000000001e-06, "loss": 0.0533, "step": 8537 }, { "epoch": 0.45519006237671267, "grad_norm": 2.8686153888702393, "learning_rate": 8.292600000000001e-06, "loss": 0.0471, "step": 8538 }, { "epoch": 0.45524337580636565, "grad_norm": 2.7572994232177734, "learning_rate": 8.2924e-06, "loss": 0.0724, "step": 8539 }, { "epoch": 0.45529668923601857, "grad_norm": 1.7971161603927612, "learning_rate": 8.2922e-06, "loss": 0.0249, "step": 8540 }, { "epoch": 0.4553500026656715, "grad_norm": 2.1365561485290527, "learning_rate": 8.292000000000001e-06, "loss": 0.0353, "step": 8541 }, { "epoch": 0.4554033160953244, "grad_norm": 5.995145797729492, "learning_rate": 8.2918e-06, "loss": 0.0419, "step": 8542 }, { "epoch": 0.45545662952497734, "grad_norm": 2.310995101928711, "learning_rate": 8.2916e-06, "loss": 0.0336, "step": 8543 }, { "epoch": 0.45550994295463026, "grad_norm": 2.3178398609161377, "learning_rate": 8.2914e-06, "loss": 0.0295, "step": 8544 }, { "epoch": 0.4555632563842832, "grad_norm": 1.7551658153533936, "learning_rate": 8.291200000000001e-06, "loss": 0.0303, "step": 8545 }, { "epoch": 0.4556165698139361, "grad_norm": 2.059253692626953, "learning_rate": 8.291e-06, "loss": 0.0226, "step": 8546 }, { "epoch": 0.45566988324358904, "grad_norm": 2.0591325759887695, "learning_rate": 8.2908e-06, "loss": 0.0303, "step": 8547 }, { "epoch": 0.455723196673242, "grad_norm": 10.164265632629395, "learning_rate": 8.290600000000002e-06, "loss": 0.0304, "step": 8548 }, { "epoch": 0.45577651010289494, "grad_norm": 2.7416861057281494, "learning_rate": 8.290400000000001e-06, "loss": 0.0288, "step": 8549 }, { "epoch": 0.45582982353254786, "grad_norm": 4.2868194580078125, "learning_rate": 8.2902e-06, "loss": 0.0471, "step": 8550 }, { "epoch": 0.4558831369622008, "grad_norm": 4.8236002922058105, "learning_rate": 8.29e-06, "loss": 0.0601, "step": 8551 }, { "epoch": 0.4559364503918537, "grad_norm": 1.4896360635757446, "learning_rate": 8.289800000000002e-06, "loss": 0.0116, "step": 8552 }, { "epoch": 0.45598976382150663, "grad_norm": 6.1999735832214355, "learning_rate": 8.289600000000001e-06, "loss": 0.0697, "step": 8553 }, { "epoch": 0.45604307725115956, "grad_norm": 4.965113639831543, "learning_rate": 8.2894e-06, "loss": 0.055, "step": 8554 }, { "epoch": 0.4560963906808125, "grad_norm": 5.820131778717041, "learning_rate": 8.2892e-06, "loss": 0.0209, "step": 8555 }, { "epoch": 0.4561497041104654, "grad_norm": 2.674844741821289, "learning_rate": 8.289e-06, "loss": 0.0311, "step": 8556 }, { "epoch": 0.4562030175401184, "grad_norm": 4.099145889282227, "learning_rate": 8.288800000000001e-06, "loss": 0.0714, "step": 8557 }, { "epoch": 0.4562563309697713, "grad_norm": 1.338225245475769, "learning_rate": 8.2886e-06, "loss": 0.0141, "step": 8558 }, { "epoch": 0.45630964439942423, "grad_norm": 2.675145149230957, "learning_rate": 8.2884e-06, "loss": 0.0384, "step": 8559 }, { "epoch": 0.45636295782907715, "grad_norm": 2.0256710052490234, "learning_rate": 8.288200000000001e-06, "loss": 0.0262, "step": 8560 }, { "epoch": 0.4564162712587301, "grad_norm": 3.2443795204162598, "learning_rate": 8.288000000000001e-06, "loss": 0.0362, "step": 8561 }, { "epoch": 0.456469584688383, "grad_norm": 7.590441703796387, "learning_rate": 8.2878e-06, "loss": 0.0368, "step": 8562 }, { "epoch": 0.4565228981180359, "grad_norm": 9.436836242675781, "learning_rate": 8.287600000000002e-06, "loss": 0.1004, "step": 8563 }, { "epoch": 0.45657621154768885, "grad_norm": 4.060388088226318, "learning_rate": 8.287400000000001e-06, "loss": 0.0292, "step": 8564 }, { "epoch": 0.45662952497734177, "grad_norm": 1.9652700424194336, "learning_rate": 8.287200000000001e-06, "loss": 0.0166, "step": 8565 }, { "epoch": 0.4566828384069947, "grad_norm": 4.285469055175781, "learning_rate": 8.287e-06, "loss": 0.0481, "step": 8566 }, { "epoch": 0.4567361518366477, "grad_norm": 2.241272449493408, "learning_rate": 8.2868e-06, "loss": 0.0283, "step": 8567 }, { "epoch": 0.4567894652663006, "grad_norm": 3.546616315841675, "learning_rate": 8.2866e-06, "loss": 0.0586, "step": 8568 }, { "epoch": 0.4568427786959535, "grad_norm": 0.6240847110748291, "learning_rate": 8.2864e-06, "loss": 0.005, "step": 8569 }, { "epoch": 0.45689609212560645, "grad_norm": 1.5704716444015503, "learning_rate": 8.2862e-06, "loss": 0.017, "step": 8570 }, { "epoch": 0.45694940555525937, "grad_norm": 7.207503318786621, "learning_rate": 8.286e-06, "loss": 0.0422, "step": 8571 }, { "epoch": 0.4570027189849123, "grad_norm": 3.656317710876465, "learning_rate": 8.285800000000001e-06, "loss": 0.0543, "step": 8572 }, { "epoch": 0.4570560324145652, "grad_norm": 3.225731372833252, "learning_rate": 8.2856e-06, "loss": 0.0248, "step": 8573 }, { "epoch": 0.45710934584421814, "grad_norm": 2.8488588333129883, "learning_rate": 8.285400000000002e-06, "loss": 0.0309, "step": 8574 }, { "epoch": 0.45716265927387106, "grad_norm": 3.4963254928588867, "learning_rate": 8.285200000000002e-06, "loss": 0.0311, "step": 8575 }, { "epoch": 0.45721597270352404, "grad_norm": 2.115798234939575, "learning_rate": 8.285000000000001e-06, "loss": 0.058, "step": 8576 }, { "epoch": 0.45726928613317697, "grad_norm": 3.893777370452881, "learning_rate": 8.2848e-06, "loss": 0.0491, "step": 8577 }, { "epoch": 0.4573225995628299, "grad_norm": 9.220407485961914, "learning_rate": 8.2846e-06, "loss": 0.0888, "step": 8578 }, { "epoch": 0.4573759129924828, "grad_norm": 1.5550425052642822, "learning_rate": 8.2844e-06, "loss": 0.0315, "step": 8579 }, { "epoch": 0.45742922642213574, "grad_norm": 2.0912561416625977, "learning_rate": 8.284200000000001e-06, "loss": 0.0358, "step": 8580 }, { "epoch": 0.45748253985178866, "grad_norm": 2.0046863555908203, "learning_rate": 8.284e-06, "loss": 0.0211, "step": 8581 }, { "epoch": 0.4575358532814416, "grad_norm": 2.291503429412842, "learning_rate": 8.2838e-06, "loss": 0.0369, "step": 8582 }, { "epoch": 0.4575891667110945, "grad_norm": 5.376947402954102, "learning_rate": 8.2836e-06, "loss": 0.0821, "step": 8583 }, { "epoch": 0.45764248014074743, "grad_norm": 2.1329243183135986, "learning_rate": 8.283400000000001e-06, "loss": 0.0239, "step": 8584 }, { "epoch": 0.4576957935704004, "grad_norm": 2.1997060775756836, "learning_rate": 8.2832e-06, "loss": 0.0251, "step": 8585 }, { "epoch": 0.45774910700005333, "grad_norm": 2.6698265075683594, "learning_rate": 8.283000000000002e-06, "loss": 0.0305, "step": 8586 }, { "epoch": 0.45780242042970626, "grad_norm": 2.61381196975708, "learning_rate": 8.282800000000001e-06, "loss": 0.0447, "step": 8587 }, { "epoch": 0.4578557338593592, "grad_norm": 2.6862268447875977, "learning_rate": 8.282600000000001e-06, "loss": 0.0384, "step": 8588 }, { "epoch": 0.4579090472890121, "grad_norm": 5.02671480178833, "learning_rate": 8.2824e-06, "loss": 0.0813, "step": 8589 }, { "epoch": 0.45796236071866503, "grad_norm": 4.0958123207092285, "learning_rate": 8.2822e-06, "loss": 0.0427, "step": 8590 }, { "epoch": 0.45801567414831795, "grad_norm": 2.6049532890319824, "learning_rate": 8.282000000000001e-06, "loss": 0.0546, "step": 8591 }, { "epoch": 0.4580689875779709, "grad_norm": 3.2422540187835693, "learning_rate": 8.281800000000001e-06, "loss": 0.0502, "step": 8592 }, { "epoch": 0.4581223010076238, "grad_norm": 4.139871120452881, "learning_rate": 8.2816e-06, "loss": 0.0344, "step": 8593 }, { "epoch": 0.4581756144372767, "grad_norm": 1.9970989227294922, "learning_rate": 8.2814e-06, "loss": 0.0166, "step": 8594 }, { "epoch": 0.4582289278669297, "grad_norm": 3.4545657634735107, "learning_rate": 8.2812e-06, "loss": 0.0655, "step": 8595 }, { "epoch": 0.4582822412965826, "grad_norm": 4.173430442810059, "learning_rate": 8.281e-06, "loss": 0.0516, "step": 8596 }, { "epoch": 0.45833555472623555, "grad_norm": 3.8160877227783203, "learning_rate": 8.2808e-06, "loss": 0.0449, "step": 8597 }, { "epoch": 0.4583888681558885, "grad_norm": 2.0533411502838135, "learning_rate": 8.280600000000002e-06, "loss": 0.0246, "step": 8598 }, { "epoch": 0.4584421815855414, "grad_norm": 2.6217310428619385, "learning_rate": 8.280400000000001e-06, "loss": 0.0573, "step": 8599 }, { "epoch": 0.4584954950151943, "grad_norm": 1.8831232786178589, "learning_rate": 8.2802e-06, "loss": 0.0406, "step": 8600 }, { "epoch": 0.45854880844484724, "grad_norm": 3.2735939025878906, "learning_rate": 8.28e-06, "loss": 0.0427, "step": 8601 }, { "epoch": 0.45860212187450017, "grad_norm": 1.7248181104660034, "learning_rate": 8.279800000000002e-06, "loss": 0.032, "step": 8602 }, { "epoch": 0.4586554353041531, "grad_norm": 0.7819593548774719, "learning_rate": 8.279600000000001e-06, "loss": 0.0096, "step": 8603 }, { "epoch": 0.45870874873380607, "grad_norm": 2.119323492050171, "learning_rate": 8.2794e-06, "loss": 0.0451, "step": 8604 }, { "epoch": 0.458762062163459, "grad_norm": 4.121568202972412, "learning_rate": 8.2792e-06, "loss": 0.0894, "step": 8605 }, { "epoch": 0.4588153755931119, "grad_norm": 2.0949807167053223, "learning_rate": 8.279e-06, "loss": 0.0458, "step": 8606 }, { "epoch": 0.45886868902276484, "grad_norm": 1.8541679382324219, "learning_rate": 8.2788e-06, "loss": 0.0491, "step": 8607 }, { "epoch": 0.45892200245241777, "grad_norm": 5.8329596519470215, "learning_rate": 8.2786e-06, "loss": 0.0699, "step": 8608 }, { "epoch": 0.4589753158820707, "grad_norm": 6.403766632080078, "learning_rate": 8.2784e-06, "loss": 0.0511, "step": 8609 }, { "epoch": 0.4590286293117236, "grad_norm": 3.319173812866211, "learning_rate": 8.278200000000001e-06, "loss": 0.0528, "step": 8610 }, { "epoch": 0.45908194274137654, "grad_norm": 2.955031633377075, "learning_rate": 8.278000000000001e-06, "loss": 0.0509, "step": 8611 }, { "epoch": 0.45913525617102946, "grad_norm": 2.550413131713867, "learning_rate": 8.2778e-06, "loss": 0.0272, "step": 8612 }, { "epoch": 0.45918856960068244, "grad_norm": 2.2732863426208496, "learning_rate": 8.277600000000002e-06, "loss": 0.0384, "step": 8613 }, { "epoch": 0.45924188303033536, "grad_norm": 0.8812933564186096, "learning_rate": 8.277400000000001e-06, "loss": 0.0104, "step": 8614 }, { "epoch": 0.4592951964599883, "grad_norm": 3.18319034576416, "learning_rate": 8.277200000000001e-06, "loss": 0.0466, "step": 8615 }, { "epoch": 0.4593485098896412, "grad_norm": 1.3939149379730225, "learning_rate": 8.277e-06, "loss": 0.0229, "step": 8616 }, { "epoch": 0.45940182331929413, "grad_norm": 2.0082662105560303, "learning_rate": 8.2768e-06, "loss": 0.0287, "step": 8617 }, { "epoch": 0.45945513674894706, "grad_norm": 2.6919074058532715, "learning_rate": 8.2766e-06, "loss": 0.0353, "step": 8618 }, { "epoch": 0.4595084501786, "grad_norm": 3.612496852874756, "learning_rate": 8.276400000000001e-06, "loss": 0.0228, "step": 8619 }, { "epoch": 0.4595617636082529, "grad_norm": 3.3531785011291504, "learning_rate": 8.2762e-06, "loss": 0.0457, "step": 8620 }, { "epoch": 0.45961507703790583, "grad_norm": 2.3026609420776367, "learning_rate": 8.276e-06, "loss": 0.0338, "step": 8621 }, { "epoch": 0.45966839046755875, "grad_norm": 2.6012258529663086, "learning_rate": 8.275800000000001e-06, "loss": 0.0423, "step": 8622 }, { "epoch": 0.45972170389721173, "grad_norm": 2.245387315750122, "learning_rate": 8.275600000000001e-06, "loss": 0.0247, "step": 8623 }, { "epoch": 0.45977501732686465, "grad_norm": 2.713031768798828, "learning_rate": 8.2754e-06, "loss": 0.0461, "step": 8624 }, { "epoch": 0.4598283307565176, "grad_norm": 2.386470079421997, "learning_rate": 8.275200000000002e-06, "loss": 0.0235, "step": 8625 }, { "epoch": 0.4598816441861705, "grad_norm": 3.0130369663238525, "learning_rate": 8.275000000000001e-06, "loss": 0.0235, "step": 8626 }, { "epoch": 0.4599349576158234, "grad_norm": 3.0873804092407227, "learning_rate": 8.2748e-06, "loss": 0.0661, "step": 8627 }, { "epoch": 0.45998827104547635, "grad_norm": 2.7665657997131348, "learning_rate": 8.2746e-06, "loss": 0.0228, "step": 8628 }, { "epoch": 0.4600415844751293, "grad_norm": 2.4193177223205566, "learning_rate": 8.2744e-06, "loss": 0.0226, "step": 8629 }, { "epoch": 0.4600948979047822, "grad_norm": 1.916569471359253, "learning_rate": 8.274200000000001e-06, "loss": 0.038, "step": 8630 }, { "epoch": 0.4601482113344351, "grad_norm": 1.9400385618209839, "learning_rate": 8.274e-06, "loss": 0.0347, "step": 8631 }, { "epoch": 0.4602015247640881, "grad_norm": 3.9428107738494873, "learning_rate": 8.2738e-06, "loss": 0.0412, "step": 8632 }, { "epoch": 0.460254838193741, "grad_norm": 2.2572920322418213, "learning_rate": 8.2736e-06, "loss": 0.0283, "step": 8633 }, { "epoch": 0.46030815162339395, "grad_norm": 2.695361375808716, "learning_rate": 8.273400000000001e-06, "loss": 0.0558, "step": 8634 }, { "epoch": 0.46036146505304687, "grad_norm": 2.5593721866607666, "learning_rate": 8.2732e-06, "loss": 0.0529, "step": 8635 }, { "epoch": 0.4604147784826998, "grad_norm": 2.579563617706299, "learning_rate": 8.273e-06, "loss": 0.0195, "step": 8636 }, { "epoch": 0.4604680919123527, "grad_norm": 2.1707565784454346, "learning_rate": 8.272800000000001e-06, "loss": 0.033, "step": 8637 }, { "epoch": 0.46052140534200564, "grad_norm": 3.9486210346221924, "learning_rate": 8.272600000000001e-06, "loss": 0.0603, "step": 8638 }, { "epoch": 0.46057471877165856, "grad_norm": 1.1332175731658936, "learning_rate": 8.2724e-06, "loss": 0.0113, "step": 8639 }, { "epoch": 0.4606280322013115, "grad_norm": 2.8667683601379395, "learning_rate": 8.2722e-06, "loss": 0.0428, "step": 8640 }, { "epoch": 0.46068134563096447, "grad_norm": 4.757730007171631, "learning_rate": 8.272000000000001e-06, "loss": 0.042, "step": 8641 }, { "epoch": 0.4607346590606174, "grad_norm": 1.8699896335601807, "learning_rate": 8.271800000000001e-06, "loss": 0.0193, "step": 8642 }, { "epoch": 0.4607879724902703, "grad_norm": 2.9746615886688232, "learning_rate": 8.2716e-06, "loss": 0.0411, "step": 8643 }, { "epoch": 0.46084128591992324, "grad_norm": 2.0052382946014404, "learning_rate": 8.2714e-06, "loss": 0.021, "step": 8644 }, { "epoch": 0.46089459934957616, "grad_norm": 2.3387575149536133, "learning_rate": 8.2712e-06, "loss": 0.0161, "step": 8645 }, { "epoch": 0.4609479127792291, "grad_norm": 2.409733533859253, "learning_rate": 8.271000000000001e-06, "loss": 0.0274, "step": 8646 }, { "epoch": 0.461001226208882, "grad_norm": 1.212067723274231, "learning_rate": 8.2708e-06, "loss": 0.0064, "step": 8647 }, { "epoch": 0.46105453963853493, "grad_norm": 1.32540762424469, "learning_rate": 8.2706e-06, "loss": 0.0352, "step": 8648 }, { "epoch": 0.46110785306818786, "grad_norm": 3.2566730976104736, "learning_rate": 8.270400000000001e-06, "loss": 0.0509, "step": 8649 }, { "epoch": 0.4611611664978408, "grad_norm": 2.4831182956695557, "learning_rate": 8.270200000000001e-06, "loss": 0.0294, "step": 8650 }, { "epoch": 0.46121447992749376, "grad_norm": 1.7098816633224487, "learning_rate": 8.27e-06, "loss": 0.0273, "step": 8651 }, { "epoch": 0.4612677933571467, "grad_norm": 3.987917423248291, "learning_rate": 8.269800000000002e-06, "loss": 0.0331, "step": 8652 }, { "epoch": 0.4613211067867996, "grad_norm": 3.573885202407837, "learning_rate": 8.269600000000001e-06, "loss": 0.0819, "step": 8653 }, { "epoch": 0.46137442021645253, "grad_norm": 4.549803733825684, "learning_rate": 8.2694e-06, "loss": 0.0405, "step": 8654 }, { "epoch": 0.46142773364610545, "grad_norm": 3.897846221923828, "learning_rate": 8.2692e-06, "loss": 0.0827, "step": 8655 }, { "epoch": 0.4614810470757584, "grad_norm": 1.8740781545639038, "learning_rate": 8.269e-06, "loss": 0.0203, "step": 8656 }, { "epoch": 0.4615343605054113, "grad_norm": 4.248481750488281, "learning_rate": 8.2688e-06, "loss": 0.0667, "step": 8657 }, { "epoch": 0.4615876739350642, "grad_norm": 1.5579402446746826, "learning_rate": 8.2686e-06, "loss": 0.026, "step": 8658 }, { "epoch": 0.46164098736471715, "grad_norm": 1.5306919813156128, "learning_rate": 8.2684e-06, "loss": 0.0174, "step": 8659 }, { "epoch": 0.4616943007943701, "grad_norm": 1.6536790132522583, "learning_rate": 8.2682e-06, "loss": 0.0255, "step": 8660 }, { "epoch": 0.46174761422402305, "grad_norm": 3.6906065940856934, "learning_rate": 8.268000000000001e-06, "loss": 0.0465, "step": 8661 }, { "epoch": 0.461800927653676, "grad_norm": 2.263298749923706, "learning_rate": 8.2678e-06, "loss": 0.0387, "step": 8662 }, { "epoch": 0.4618542410833289, "grad_norm": 2.119062662124634, "learning_rate": 8.267600000000002e-06, "loss": 0.039, "step": 8663 }, { "epoch": 0.4619075545129818, "grad_norm": 2.6363651752471924, "learning_rate": 8.267400000000002e-06, "loss": 0.0184, "step": 8664 }, { "epoch": 0.46196086794263475, "grad_norm": 1.31069815158844, "learning_rate": 8.267200000000001e-06, "loss": 0.0239, "step": 8665 }, { "epoch": 0.46201418137228767, "grad_norm": 3.401134490966797, "learning_rate": 8.267e-06, "loss": 0.0463, "step": 8666 }, { "epoch": 0.4620674948019406, "grad_norm": 2.0741894245147705, "learning_rate": 8.2668e-06, "loss": 0.0566, "step": 8667 }, { "epoch": 0.4621208082315935, "grad_norm": 2.252984046936035, "learning_rate": 8.2666e-06, "loss": 0.0349, "step": 8668 }, { "epoch": 0.4621741216612465, "grad_norm": 2.3369662761688232, "learning_rate": 8.266400000000001e-06, "loss": 0.0477, "step": 8669 }, { "epoch": 0.4622274350908994, "grad_norm": 1.0001829862594604, "learning_rate": 8.2662e-06, "loss": 0.0097, "step": 8670 }, { "epoch": 0.46228074852055234, "grad_norm": 2.6627590656280518, "learning_rate": 8.266e-06, "loss": 0.03, "step": 8671 }, { "epoch": 0.46233406195020527, "grad_norm": 1.9021252393722534, "learning_rate": 8.2658e-06, "loss": 0.0475, "step": 8672 }, { "epoch": 0.4623873753798582, "grad_norm": 3.3038454055786133, "learning_rate": 8.265600000000001e-06, "loss": 0.0493, "step": 8673 }, { "epoch": 0.4624406888095111, "grad_norm": 1.1269639730453491, "learning_rate": 8.2654e-06, "loss": 0.0234, "step": 8674 }, { "epoch": 0.46249400223916404, "grad_norm": 1.2751190662384033, "learning_rate": 8.265200000000002e-06, "loss": 0.0197, "step": 8675 }, { "epoch": 0.46254731566881696, "grad_norm": 2.1918392181396484, "learning_rate": 8.265000000000001e-06, "loss": 0.0323, "step": 8676 }, { "epoch": 0.4626006290984699, "grad_norm": 1.704032063484192, "learning_rate": 8.264800000000001e-06, "loss": 0.0285, "step": 8677 }, { "epoch": 0.4626539425281228, "grad_norm": 2.6273105144500732, "learning_rate": 8.2646e-06, "loss": 0.0319, "step": 8678 }, { "epoch": 0.4627072559577758, "grad_norm": 1.8644132614135742, "learning_rate": 8.2644e-06, "loss": 0.0327, "step": 8679 }, { "epoch": 0.4627605693874287, "grad_norm": 2.1543500423431396, "learning_rate": 8.264200000000001e-06, "loss": 0.0263, "step": 8680 }, { "epoch": 0.46281388281708163, "grad_norm": 2.5636966228485107, "learning_rate": 8.264e-06, "loss": 0.0187, "step": 8681 }, { "epoch": 0.46286719624673456, "grad_norm": 4.075875282287598, "learning_rate": 8.2638e-06, "loss": 0.0592, "step": 8682 }, { "epoch": 0.4629205096763875, "grad_norm": 1.1907238960266113, "learning_rate": 8.2636e-06, "loss": 0.0302, "step": 8683 }, { "epoch": 0.4629738231060404, "grad_norm": 2.2228200435638428, "learning_rate": 8.2634e-06, "loss": 0.045, "step": 8684 }, { "epoch": 0.46302713653569333, "grad_norm": 4.22467565536499, "learning_rate": 8.2632e-06, "loss": 0.0335, "step": 8685 }, { "epoch": 0.46308044996534625, "grad_norm": 3.2776808738708496, "learning_rate": 8.263e-06, "loss": 0.0451, "step": 8686 }, { "epoch": 0.4631337633949992, "grad_norm": 1.2499847412109375, "learning_rate": 8.262800000000002e-06, "loss": 0.0108, "step": 8687 }, { "epoch": 0.46318707682465216, "grad_norm": 5.557831764221191, "learning_rate": 8.262600000000001e-06, "loss": 0.0303, "step": 8688 }, { "epoch": 0.4632403902543051, "grad_norm": 3.1149067878723145, "learning_rate": 8.2624e-06, "loss": 0.0598, "step": 8689 }, { "epoch": 0.463293703683958, "grad_norm": 4.403390407562256, "learning_rate": 8.2622e-06, "loss": 0.0529, "step": 8690 }, { "epoch": 0.4633470171136109, "grad_norm": 1.9110522270202637, "learning_rate": 8.262000000000002e-06, "loss": 0.0146, "step": 8691 }, { "epoch": 0.46340033054326385, "grad_norm": 2.647148847579956, "learning_rate": 8.261800000000001e-06, "loss": 0.0497, "step": 8692 }, { "epoch": 0.4634536439729168, "grad_norm": 2.22636342048645, "learning_rate": 8.2616e-06, "loss": 0.0346, "step": 8693 }, { "epoch": 0.4635069574025697, "grad_norm": 1.618349552154541, "learning_rate": 8.2614e-06, "loss": 0.0119, "step": 8694 }, { "epoch": 0.4635602708322226, "grad_norm": 5.82906436920166, "learning_rate": 8.2612e-06, "loss": 0.0398, "step": 8695 }, { "epoch": 0.46361358426187554, "grad_norm": 4.003154277801514, "learning_rate": 8.261e-06, "loss": 0.0683, "step": 8696 }, { "epoch": 0.4636668976915285, "grad_norm": 4.402738571166992, "learning_rate": 8.2608e-06, "loss": 0.0553, "step": 8697 }, { "epoch": 0.46372021112118145, "grad_norm": 1.7695300579071045, "learning_rate": 8.2606e-06, "loss": 0.0323, "step": 8698 }, { "epoch": 0.46377352455083437, "grad_norm": 2.7775607109069824, "learning_rate": 8.260400000000001e-06, "loss": 0.0378, "step": 8699 }, { "epoch": 0.4638268379804873, "grad_norm": 3.097672462463379, "learning_rate": 8.260200000000001e-06, "loss": 0.0842, "step": 8700 }, { "epoch": 0.4638801514101402, "grad_norm": 1.8535792827606201, "learning_rate": 8.26e-06, "loss": 0.0256, "step": 8701 }, { "epoch": 0.46393346483979314, "grad_norm": 4.6475443840026855, "learning_rate": 8.259800000000002e-06, "loss": 0.0681, "step": 8702 }, { "epoch": 0.46398677826944607, "grad_norm": 2.09904146194458, "learning_rate": 8.259600000000001e-06, "loss": 0.0202, "step": 8703 }, { "epoch": 0.464040091699099, "grad_norm": 2.5136704444885254, "learning_rate": 8.259400000000001e-06, "loss": 0.0347, "step": 8704 }, { "epoch": 0.4640934051287519, "grad_norm": 2.2876553535461426, "learning_rate": 8.2592e-06, "loss": 0.0476, "step": 8705 }, { "epoch": 0.46414671855840484, "grad_norm": 1.5795220136642456, "learning_rate": 8.259e-06, "loss": 0.0239, "step": 8706 }, { "epoch": 0.4642000319880578, "grad_norm": 3.4756975173950195, "learning_rate": 8.2588e-06, "loss": 0.0376, "step": 8707 }, { "epoch": 0.46425334541771074, "grad_norm": 4.968441009521484, "learning_rate": 8.258600000000001e-06, "loss": 0.0626, "step": 8708 }, { "epoch": 0.46430665884736366, "grad_norm": 4.0862650871276855, "learning_rate": 8.2584e-06, "loss": 0.0441, "step": 8709 }, { "epoch": 0.4643599722770166, "grad_norm": 1.8639036417007446, "learning_rate": 8.2582e-06, "loss": 0.0271, "step": 8710 }, { "epoch": 0.4644132857066695, "grad_norm": 3.3017578125, "learning_rate": 8.258000000000001e-06, "loss": 0.0556, "step": 8711 }, { "epoch": 0.46446659913632243, "grad_norm": 1.6781107187271118, "learning_rate": 8.2578e-06, "loss": 0.0191, "step": 8712 }, { "epoch": 0.46451991256597536, "grad_norm": 2.0495688915252686, "learning_rate": 8.2576e-06, "loss": 0.0258, "step": 8713 }, { "epoch": 0.4645732259956283, "grad_norm": 2.3070738315582275, "learning_rate": 8.257400000000002e-06, "loss": 0.0413, "step": 8714 }, { "epoch": 0.4646265394252812, "grad_norm": 2.2638275623321533, "learning_rate": 8.257200000000001e-06, "loss": 0.0303, "step": 8715 }, { "epoch": 0.4646798528549342, "grad_norm": 2.551581382751465, "learning_rate": 8.257e-06, "loss": 0.0177, "step": 8716 }, { "epoch": 0.4647331662845871, "grad_norm": 3.3908607959747314, "learning_rate": 8.2568e-06, "loss": 0.0474, "step": 8717 }, { "epoch": 0.46478647971424003, "grad_norm": 3.5340819358825684, "learning_rate": 8.2566e-06, "loss": 0.0561, "step": 8718 }, { "epoch": 0.46483979314389295, "grad_norm": 1.5073857307434082, "learning_rate": 8.256400000000001e-06, "loss": 0.0289, "step": 8719 }, { "epoch": 0.4648931065735459, "grad_norm": 3.3793742656707764, "learning_rate": 8.2562e-06, "loss": 0.0477, "step": 8720 }, { "epoch": 0.4649464200031988, "grad_norm": 2.3803696632385254, "learning_rate": 8.256e-06, "loss": 0.0472, "step": 8721 }, { "epoch": 0.4649997334328517, "grad_norm": 2.13136887550354, "learning_rate": 8.2558e-06, "loss": 0.0442, "step": 8722 }, { "epoch": 0.46505304686250465, "grad_norm": 2.753493070602417, "learning_rate": 8.255600000000001e-06, "loss": 0.0327, "step": 8723 }, { "epoch": 0.4651063602921576, "grad_norm": 3.0345284938812256, "learning_rate": 8.2554e-06, "loss": 0.0489, "step": 8724 }, { "epoch": 0.46515967372181055, "grad_norm": 2.5649378299713135, "learning_rate": 8.2552e-06, "loss": 0.034, "step": 8725 }, { "epoch": 0.4652129871514635, "grad_norm": 2.010969638824463, "learning_rate": 8.255000000000001e-06, "loss": 0.026, "step": 8726 }, { "epoch": 0.4652663005811164, "grad_norm": 3.1247730255126953, "learning_rate": 8.254800000000001e-06, "loss": 0.035, "step": 8727 }, { "epoch": 0.4653196140107693, "grad_norm": 1.761031985282898, "learning_rate": 8.2546e-06, "loss": 0.0214, "step": 8728 }, { "epoch": 0.46537292744042225, "grad_norm": 6.6495795249938965, "learning_rate": 8.2544e-06, "loss": 0.0531, "step": 8729 }, { "epoch": 0.46542624087007517, "grad_norm": 4.717263698577881, "learning_rate": 8.254200000000001e-06, "loss": 0.0478, "step": 8730 }, { "epoch": 0.4654795542997281, "grad_norm": 1.134368658065796, "learning_rate": 8.254000000000001e-06, "loss": 0.0323, "step": 8731 }, { "epoch": 0.465532867729381, "grad_norm": 4.86345911026001, "learning_rate": 8.2538e-06, "loss": 0.08, "step": 8732 }, { "epoch": 0.46558618115903394, "grad_norm": 3.289907217025757, "learning_rate": 8.2536e-06, "loss": 0.0497, "step": 8733 }, { "epoch": 0.46563949458868686, "grad_norm": 2.4692864418029785, "learning_rate": 8.2534e-06, "loss": 0.0359, "step": 8734 }, { "epoch": 0.46569280801833984, "grad_norm": 2.8032195568084717, "learning_rate": 8.253200000000001e-06, "loss": 0.0518, "step": 8735 }, { "epoch": 0.46574612144799277, "grad_norm": 5.28057861328125, "learning_rate": 8.253e-06, "loss": 0.0487, "step": 8736 }, { "epoch": 0.4657994348776457, "grad_norm": 2.315462827682495, "learning_rate": 8.2528e-06, "loss": 0.0349, "step": 8737 }, { "epoch": 0.4658527483072986, "grad_norm": 2.2614383697509766, "learning_rate": 8.252600000000001e-06, "loss": 0.049, "step": 8738 }, { "epoch": 0.46590606173695154, "grad_norm": 4.275557041168213, "learning_rate": 8.2524e-06, "loss": 0.0231, "step": 8739 }, { "epoch": 0.46595937516660446, "grad_norm": 2.335094690322876, "learning_rate": 8.2522e-06, "loss": 0.0442, "step": 8740 }, { "epoch": 0.4660126885962574, "grad_norm": 2.0185353755950928, "learning_rate": 8.252000000000002e-06, "loss": 0.0172, "step": 8741 }, { "epoch": 0.4660660020259103, "grad_norm": 2.6915769577026367, "learning_rate": 8.251800000000001e-06, "loss": 0.0557, "step": 8742 }, { "epoch": 0.46611931545556323, "grad_norm": 1.8970293998718262, "learning_rate": 8.2516e-06, "loss": 0.0378, "step": 8743 }, { "epoch": 0.4661726288852162, "grad_norm": 5.651613235473633, "learning_rate": 8.2514e-06, "loss": 0.0455, "step": 8744 }, { "epoch": 0.46622594231486914, "grad_norm": 2.3254756927490234, "learning_rate": 8.2512e-06, "loss": 0.0329, "step": 8745 }, { "epoch": 0.46627925574452206, "grad_norm": 1.984316349029541, "learning_rate": 8.251e-06, "loss": 0.0249, "step": 8746 }, { "epoch": 0.466332569174175, "grad_norm": 2.380920886993408, "learning_rate": 8.2508e-06, "loss": 0.0237, "step": 8747 }, { "epoch": 0.4663858826038279, "grad_norm": 5.109742164611816, "learning_rate": 8.2506e-06, "loss": 0.0484, "step": 8748 }, { "epoch": 0.46643919603348083, "grad_norm": 3.2448902130126953, "learning_rate": 8.2504e-06, "loss": 0.054, "step": 8749 }, { "epoch": 0.46649250946313375, "grad_norm": 1.9134615659713745, "learning_rate": 8.250200000000001e-06, "loss": 0.0232, "step": 8750 }, { "epoch": 0.4665458228927867, "grad_norm": 1.3014649152755737, "learning_rate": 8.25e-06, "loss": 0.0123, "step": 8751 }, { "epoch": 0.4665991363224396, "grad_norm": 3.263287305831909, "learning_rate": 8.249800000000002e-06, "loss": 0.0617, "step": 8752 }, { "epoch": 0.4666524497520926, "grad_norm": 1.8900185823440552, "learning_rate": 8.249600000000001e-06, "loss": 0.0273, "step": 8753 }, { "epoch": 0.4667057631817455, "grad_norm": 3.715477705001831, "learning_rate": 8.249400000000001e-06, "loss": 0.0565, "step": 8754 }, { "epoch": 0.4667590766113984, "grad_norm": 2.3155007362365723, "learning_rate": 8.2492e-06, "loss": 0.0149, "step": 8755 }, { "epoch": 0.46681239004105135, "grad_norm": 3.846705913543701, "learning_rate": 8.249e-06, "loss": 0.0383, "step": 8756 }, { "epoch": 0.4668657034707043, "grad_norm": 3.799334764480591, "learning_rate": 8.2488e-06, "loss": 0.0205, "step": 8757 }, { "epoch": 0.4669190169003572, "grad_norm": 6.61381196975708, "learning_rate": 8.248600000000001e-06, "loss": 0.0868, "step": 8758 }, { "epoch": 0.4669723303300101, "grad_norm": 2.487691879272461, "learning_rate": 8.2484e-06, "loss": 0.0096, "step": 8759 }, { "epoch": 0.46702564375966304, "grad_norm": 3.4389612674713135, "learning_rate": 8.2482e-06, "loss": 0.0409, "step": 8760 }, { "epoch": 0.46707895718931597, "grad_norm": 2.4677491188049316, "learning_rate": 8.248e-06, "loss": 0.0287, "step": 8761 }, { "epoch": 0.4671322706189689, "grad_norm": 2.7336907386779785, "learning_rate": 8.247800000000001e-06, "loss": 0.04, "step": 8762 }, { "epoch": 0.46718558404862187, "grad_norm": 4.9128336906433105, "learning_rate": 8.2476e-06, "loss": 0.0753, "step": 8763 }, { "epoch": 0.4672388974782748, "grad_norm": 1.4069818258285522, "learning_rate": 8.247400000000002e-06, "loss": 0.0221, "step": 8764 }, { "epoch": 0.4672922109079277, "grad_norm": 4.0972394943237305, "learning_rate": 8.247200000000001e-06, "loss": 0.0863, "step": 8765 }, { "epoch": 0.46734552433758064, "grad_norm": 2.263624429702759, "learning_rate": 8.247e-06, "loss": 0.0467, "step": 8766 }, { "epoch": 0.46739883776723357, "grad_norm": 2.03991961479187, "learning_rate": 8.2468e-06, "loss": 0.0197, "step": 8767 }, { "epoch": 0.4674521511968865, "grad_norm": 2.048006534576416, "learning_rate": 8.2466e-06, "loss": 0.0235, "step": 8768 }, { "epoch": 0.4675054646265394, "grad_norm": 3.1850647926330566, "learning_rate": 8.246400000000001e-06, "loss": 0.0442, "step": 8769 }, { "epoch": 0.46755877805619234, "grad_norm": 3.9741873741149902, "learning_rate": 8.2462e-06, "loss": 0.0258, "step": 8770 }, { "epoch": 0.46761209148584526, "grad_norm": 3.4336555004119873, "learning_rate": 8.246e-06, "loss": 0.0541, "step": 8771 }, { "epoch": 0.46766540491549824, "grad_norm": 3.359225034713745, "learning_rate": 8.2458e-06, "loss": 0.0546, "step": 8772 }, { "epoch": 0.46771871834515116, "grad_norm": 4.397634506225586, "learning_rate": 8.2456e-06, "loss": 0.0362, "step": 8773 }, { "epoch": 0.4677720317748041, "grad_norm": 3.5486032962799072, "learning_rate": 8.2454e-06, "loss": 0.0714, "step": 8774 }, { "epoch": 0.467825345204457, "grad_norm": 2.8199219703674316, "learning_rate": 8.2452e-06, "loss": 0.0452, "step": 8775 }, { "epoch": 0.46787865863410993, "grad_norm": 2.300629138946533, "learning_rate": 8.245000000000002e-06, "loss": 0.0339, "step": 8776 }, { "epoch": 0.46793197206376286, "grad_norm": 1.972190260887146, "learning_rate": 8.244800000000001e-06, "loss": 0.0511, "step": 8777 }, { "epoch": 0.4679852854934158, "grad_norm": 2.8530046939849854, "learning_rate": 8.2446e-06, "loss": 0.0516, "step": 8778 }, { "epoch": 0.4680385989230687, "grad_norm": 3.1950860023498535, "learning_rate": 8.2444e-06, "loss": 0.0278, "step": 8779 }, { "epoch": 0.46809191235272163, "grad_norm": 1.8885117769241333, "learning_rate": 8.244200000000001e-06, "loss": 0.0115, "step": 8780 }, { "epoch": 0.4681452257823746, "grad_norm": 2.4335720539093018, "learning_rate": 8.244000000000001e-06, "loss": 0.0238, "step": 8781 }, { "epoch": 0.46819853921202753, "grad_norm": 2.1587107181549072, "learning_rate": 8.2438e-06, "loss": 0.0337, "step": 8782 }, { "epoch": 0.46825185264168045, "grad_norm": 1.889670968055725, "learning_rate": 8.2436e-06, "loss": 0.0315, "step": 8783 }, { "epoch": 0.4683051660713334, "grad_norm": 3.0231049060821533, "learning_rate": 8.2434e-06, "loss": 0.0218, "step": 8784 }, { "epoch": 0.4683584795009863, "grad_norm": 3.552457809448242, "learning_rate": 8.243200000000001e-06, "loss": 0.0709, "step": 8785 }, { "epoch": 0.4684117929306392, "grad_norm": 3.243394136428833, "learning_rate": 8.243e-06, "loss": 0.0442, "step": 8786 }, { "epoch": 0.46846510636029215, "grad_norm": 2.6275954246520996, "learning_rate": 8.2428e-06, "loss": 0.0326, "step": 8787 }, { "epoch": 0.4685184197899451, "grad_norm": 3.6986730098724365, "learning_rate": 8.242600000000001e-06, "loss": 0.055, "step": 8788 }, { "epoch": 0.468571733219598, "grad_norm": 3.103181838989258, "learning_rate": 8.242400000000001e-06, "loss": 0.0434, "step": 8789 }, { "epoch": 0.4686250466492509, "grad_norm": 3.5424909591674805, "learning_rate": 8.2422e-06, "loss": 0.06, "step": 8790 }, { "epoch": 0.4686783600789039, "grad_norm": 4.338690280914307, "learning_rate": 8.242000000000002e-06, "loss": 0.0627, "step": 8791 }, { "epoch": 0.4687316735085568, "grad_norm": 4.986711025238037, "learning_rate": 8.241800000000001e-06, "loss": 0.0513, "step": 8792 }, { "epoch": 0.46878498693820975, "grad_norm": 5.217329502105713, "learning_rate": 8.241600000000001e-06, "loss": 0.0852, "step": 8793 }, { "epoch": 0.46883830036786267, "grad_norm": 2.1841087341308594, "learning_rate": 8.2414e-06, "loss": 0.0239, "step": 8794 }, { "epoch": 0.4688916137975156, "grad_norm": 3.9854884147644043, "learning_rate": 8.2412e-06, "loss": 0.0334, "step": 8795 }, { "epoch": 0.4689449272271685, "grad_norm": 1.6523076295852661, "learning_rate": 8.241000000000001e-06, "loss": 0.024, "step": 8796 }, { "epoch": 0.46899824065682144, "grad_norm": 3.0505423545837402, "learning_rate": 8.2408e-06, "loss": 0.0369, "step": 8797 }, { "epoch": 0.46905155408647436, "grad_norm": 3.6244168281555176, "learning_rate": 8.2406e-06, "loss": 0.0414, "step": 8798 }, { "epoch": 0.4691048675161273, "grad_norm": 2.113548994064331, "learning_rate": 8.2404e-06, "loss": 0.0454, "step": 8799 }, { "epoch": 0.46915818094578027, "grad_norm": 1.6963132619857788, "learning_rate": 8.240200000000001e-06, "loss": 0.0216, "step": 8800 }, { "epoch": 0.4692114943754332, "grad_norm": 1.5808457136154175, "learning_rate": 8.24e-06, "loss": 0.0363, "step": 8801 }, { "epoch": 0.4692648078050861, "grad_norm": 3.185258388519287, "learning_rate": 8.2398e-06, "loss": 0.0493, "step": 8802 }, { "epoch": 0.46931812123473904, "grad_norm": 3.395502805709839, "learning_rate": 8.239600000000002e-06, "loss": 0.0448, "step": 8803 }, { "epoch": 0.46937143466439196, "grad_norm": 2.715665102005005, "learning_rate": 8.239400000000001e-06, "loss": 0.0401, "step": 8804 }, { "epoch": 0.4694247480940449, "grad_norm": 3.917137861251831, "learning_rate": 8.2392e-06, "loss": 0.0485, "step": 8805 }, { "epoch": 0.4694780615236978, "grad_norm": 2.390392303466797, "learning_rate": 8.239e-06, "loss": 0.0486, "step": 8806 }, { "epoch": 0.46953137495335073, "grad_norm": 3.6613447666168213, "learning_rate": 8.2388e-06, "loss": 0.027, "step": 8807 }, { "epoch": 0.46958468838300366, "grad_norm": 3.7303168773651123, "learning_rate": 8.238600000000001e-06, "loss": 0.0576, "step": 8808 }, { "epoch": 0.46963800181265664, "grad_norm": 3.593942880630493, "learning_rate": 8.2384e-06, "loss": 0.0848, "step": 8809 }, { "epoch": 0.46969131524230956, "grad_norm": 3.4317939281463623, "learning_rate": 8.2382e-06, "loss": 0.0378, "step": 8810 }, { "epoch": 0.4697446286719625, "grad_norm": 1.7855430841445923, "learning_rate": 8.238e-06, "loss": 0.0293, "step": 8811 }, { "epoch": 0.4697979421016154, "grad_norm": 1.8970651626586914, "learning_rate": 8.237800000000001e-06, "loss": 0.0328, "step": 8812 }, { "epoch": 0.46985125553126833, "grad_norm": 3.573458671569824, "learning_rate": 8.2376e-06, "loss": 0.0714, "step": 8813 }, { "epoch": 0.46990456896092125, "grad_norm": 2.6360394954681396, "learning_rate": 8.2374e-06, "loss": 0.0241, "step": 8814 }, { "epoch": 0.4699578823905742, "grad_norm": 3.178982734680176, "learning_rate": 8.237200000000001e-06, "loss": 0.0276, "step": 8815 }, { "epoch": 0.4700111958202271, "grad_norm": 3.084311008453369, "learning_rate": 8.237000000000001e-06, "loss": 0.0434, "step": 8816 }, { "epoch": 0.47006450924988, "grad_norm": 2.259674072265625, "learning_rate": 8.2368e-06, "loss": 0.0518, "step": 8817 }, { "epoch": 0.47011782267953295, "grad_norm": 3.0163238048553467, "learning_rate": 8.2366e-06, "loss": 0.0539, "step": 8818 }, { "epoch": 0.4701711361091859, "grad_norm": 2.0255846977233887, "learning_rate": 8.236400000000001e-06, "loss": 0.0124, "step": 8819 }, { "epoch": 0.47022444953883885, "grad_norm": 2.3910529613494873, "learning_rate": 8.236200000000001e-06, "loss": 0.0283, "step": 8820 }, { "epoch": 0.4702777629684918, "grad_norm": 1.9678970575332642, "learning_rate": 8.236e-06, "loss": 0.0163, "step": 8821 }, { "epoch": 0.4703310763981447, "grad_norm": 1.9208486080169678, "learning_rate": 8.2358e-06, "loss": 0.0274, "step": 8822 }, { "epoch": 0.4703843898277976, "grad_norm": 1.8898355960845947, "learning_rate": 8.2356e-06, "loss": 0.0282, "step": 8823 }, { "epoch": 0.47043770325745055, "grad_norm": 1.685348391532898, "learning_rate": 8.2354e-06, "loss": 0.0201, "step": 8824 }, { "epoch": 0.47049101668710347, "grad_norm": 1.9380948543548584, "learning_rate": 8.2352e-06, "loss": 0.0465, "step": 8825 }, { "epoch": 0.4705443301167564, "grad_norm": 4.018935203552246, "learning_rate": 8.235e-06, "loss": 0.0432, "step": 8826 }, { "epoch": 0.4705976435464093, "grad_norm": 3.327972888946533, "learning_rate": 8.234800000000001e-06, "loss": 0.0503, "step": 8827 }, { "epoch": 0.4706509569760623, "grad_norm": 2.808441638946533, "learning_rate": 8.2346e-06, "loss": 0.0254, "step": 8828 }, { "epoch": 0.4707042704057152, "grad_norm": 2.460829734802246, "learning_rate": 8.2344e-06, "loss": 0.0234, "step": 8829 }, { "epoch": 0.47075758383536814, "grad_norm": 4.934524059295654, "learning_rate": 8.234200000000002e-06, "loss": 0.0398, "step": 8830 }, { "epoch": 0.47081089726502107, "grad_norm": 3.4145543575286865, "learning_rate": 8.234000000000001e-06, "loss": 0.0437, "step": 8831 }, { "epoch": 0.470864210694674, "grad_norm": 1.4825739860534668, "learning_rate": 8.2338e-06, "loss": 0.0292, "step": 8832 }, { "epoch": 0.4709175241243269, "grad_norm": 2.8058230876922607, "learning_rate": 8.2336e-06, "loss": 0.0849, "step": 8833 }, { "epoch": 0.47097083755397984, "grad_norm": 5.976889133453369, "learning_rate": 8.2334e-06, "loss": 0.0341, "step": 8834 }, { "epoch": 0.47102415098363276, "grad_norm": 2.666229248046875, "learning_rate": 8.233200000000001e-06, "loss": 0.0462, "step": 8835 }, { "epoch": 0.4710774644132857, "grad_norm": 3.2564573287963867, "learning_rate": 8.233e-06, "loss": 0.0642, "step": 8836 }, { "epoch": 0.47113077784293866, "grad_norm": 3.1777493953704834, "learning_rate": 8.2328e-06, "loss": 0.063, "step": 8837 }, { "epoch": 0.4711840912725916, "grad_norm": 3.0609748363494873, "learning_rate": 8.2326e-06, "loss": 0.0369, "step": 8838 }, { "epoch": 0.4712374047022445, "grad_norm": 2.299084424972534, "learning_rate": 8.232400000000001e-06, "loss": 0.0273, "step": 8839 }, { "epoch": 0.47129071813189743, "grad_norm": 1.6118531227111816, "learning_rate": 8.2322e-06, "loss": 0.027, "step": 8840 }, { "epoch": 0.47134403156155036, "grad_norm": 3.1299281120300293, "learning_rate": 8.232000000000002e-06, "loss": 0.0306, "step": 8841 }, { "epoch": 0.4713973449912033, "grad_norm": 4.214879035949707, "learning_rate": 8.231800000000001e-06, "loss": 0.0847, "step": 8842 }, { "epoch": 0.4714506584208562, "grad_norm": 4.176661968231201, "learning_rate": 8.231600000000001e-06, "loss": 0.06, "step": 8843 }, { "epoch": 0.47150397185050913, "grad_norm": 4.6556596755981445, "learning_rate": 8.2314e-06, "loss": 0.0493, "step": 8844 }, { "epoch": 0.47155728528016205, "grad_norm": 2.184842824935913, "learning_rate": 8.2312e-06, "loss": 0.0338, "step": 8845 }, { "epoch": 0.471610598709815, "grad_norm": 2.724921703338623, "learning_rate": 8.231000000000001e-06, "loss": 0.0467, "step": 8846 }, { "epoch": 0.47166391213946796, "grad_norm": 5.464670658111572, "learning_rate": 8.230800000000001e-06, "loss": 0.0635, "step": 8847 }, { "epoch": 0.4717172255691209, "grad_norm": 4.567408561706543, "learning_rate": 8.2306e-06, "loss": 0.0545, "step": 8848 }, { "epoch": 0.4717705389987738, "grad_norm": 2.0642971992492676, "learning_rate": 8.2304e-06, "loss": 0.0429, "step": 8849 }, { "epoch": 0.4718238524284267, "grad_norm": 5.030055999755859, "learning_rate": 8.2302e-06, "loss": 0.0581, "step": 8850 }, { "epoch": 0.47187716585807965, "grad_norm": 1.450728178024292, "learning_rate": 8.23e-06, "loss": 0.0168, "step": 8851 }, { "epoch": 0.4719304792877326, "grad_norm": 1.886206030845642, "learning_rate": 8.2298e-06, "loss": 0.0273, "step": 8852 }, { "epoch": 0.4719837927173855, "grad_norm": 2.5292952060699463, "learning_rate": 8.229600000000002e-06, "loss": 0.0457, "step": 8853 }, { "epoch": 0.4720371061470384, "grad_norm": 2.159233808517456, "learning_rate": 8.229400000000001e-06, "loss": 0.0461, "step": 8854 }, { "epoch": 0.47209041957669134, "grad_norm": 3.0370795726776123, "learning_rate": 8.2292e-06, "loss": 0.0466, "step": 8855 }, { "epoch": 0.4721437330063443, "grad_norm": 1.9870388507843018, "learning_rate": 8.229e-06, "loss": 0.0373, "step": 8856 }, { "epoch": 0.47219704643599725, "grad_norm": 2.0011701583862305, "learning_rate": 8.2288e-06, "loss": 0.0278, "step": 8857 }, { "epoch": 0.47225035986565017, "grad_norm": 4.247040271759033, "learning_rate": 8.228600000000001e-06, "loss": 0.0705, "step": 8858 }, { "epoch": 0.4723036732953031, "grad_norm": 2.953218936920166, "learning_rate": 8.2284e-06, "loss": 0.053, "step": 8859 }, { "epoch": 0.472356986724956, "grad_norm": 2.3020544052124023, "learning_rate": 8.2282e-06, "loss": 0.0447, "step": 8860 }, { "epoch": 0.47241030015460894, "grad_norm": 3.7827396392822266, "learning_rate": 8.228e-06, "loss": 0.0535, "step": 8861 }, { "epoch": 0.47246361358426187, "grad_norm": 2.6673765182495117, "learning_rate": 8.2278e-06, "loss": 0.0701, "step": 8862 }, { "epoch": 0.4725169270139148, "grad_norm": 2.7744460105895996, "learning_rate": 8.2276e-06, "loss": 0.0311, "step": 8863 }, { "epoch": 0.4725702404435677, "grad_norm": 2.274303913116455, "learning_rate": 8.2274e-06, "loss": 0.0399, "step": 8864 }, { "epoch": 0.4726235538732207, "grad_norm": 1.6860326528549194, "learning_rate": 8.227200000000001e-06, "loss": 0.0376, "step": 8865 }, { "epoch": 0.4726768673028736, "grad_norm": 1.2250221967697144, "learning_rate": 8.227000000000001e-06, "loss": 0.018, "step": 8866 }, { "epoch": 0.47273018073252654, "grad_norm": 2.1287448406219482, "learning_rate": 8.2268e-06, "loss": 0.0415, "step": 8867 }, { "epoch": 0.47278349416217946, "grad_norm": 2.8343923091888428, "learning_rate": 8.2266e-06, "loss": 0.0404, "step": 8868 }, { "epoch": 0.4728368075918324, "grad_norm": 2.6215600967407227, "learning_rate": 8.226400000000001e-06, "loss": 0.0591, "step": 8869 }, { "epoch": 0.4728901210214853, "grad_norm": 3.408261775970459, "learning_rate": 8.226200000000001e-06, "loss": 0.0637, "step": 8870 }, { "epoch": 0.47294343445113823, "grad_norm": 3.2340266704559326, "learning_rate": 8.226e-06, "loss": 0.0604, "step": 8871 }, { "epoch": 0.47299674788079116, "grad_norm": 3.6010212898254395, "learning_rate": 8.2258e-06, "loss": 0.0438, "step": 8872 }, { "epoch": 0.4730500613104441, "grad_norm": 5.093661785125732, "learning_rate": 8.2256e-06, "loss": 0.0277, "step": 8873 }, { "epoch": 0.473103374740097, "grad_norm": 2.752126932144165, "learning_rate": 8.225400000000001e-06, "loss": 0.0385, "step": 8874 }, { "epoch": 0.47315668816975, "grad_norm": 1.9731875658035278, "learning_rate": 8.2252e-06, "loss": 0.0298, "step": 8875 }, { "epoch": 0.4732100015994029, "grad_norm": 2.076565980911255, "learning_rate": 8.225e-06, "loss": 0.0372, "step": 8876 }, { "epoch": 0.47326331502905583, "grad_norm": 2.7159550189971924, "learning_rate": 8.224800000000001e-06, "loss": 0.0411, "step": 8877 }, { "epoch": 0.47331662845870875, "grad_norm": 3.3052873611450195, "learning_rate": 8.2246e-06, "loss": 0.0248, "step": 8878 }, { "epoch": 0.4733699418883617, "grad_norm": 3.257476329803467, "learning_rate": 8.2244e-06, "loss": 0.0418, "step": 8879 }, { "epoch": 0.4734232553180146, "grad_norm": 3.6612067222595215, "learning_rate": 8.224200000000002e-06, "loss": 0.0626, "step": 8880 }, { "epoch": 0.4734765687476675, "grad_norm": 2.788424253463745, "learning_rate": 8.224000000000001e-06, "loss": 0.0382, "step": 8881 }, { "epoch": 0.47352988217732045, "grad_norm": 4.010462284088135, "learning_rate": 8.2238e-06, "loss": 0.0581, "step": 8882 }, { "epoch": 0.4735831956069734, "grad_norm": 2.399106502532959, "learning_rate": 8.2236e-06, "loss": 0.0404, "step": 8883 }, { "epoch": 0.47363650903662635, "grad_norm": 2.070202350616455, "learning_rate": 8.2234e-06, "loss": 0.0161, "step": 8884 }, { "epoch": 0.4736898224662793, "grad_norm": 3.5500969886779785, "learning_rate": 8.223200000000001e-06, "loss": 0.0536, "step": 8885 }, { "epoch": 0.4737431358959322, "grad_norm": 3.3511850833892822, "learning_rate": 8.223e-06, "loss": 0.0295, "step": 8886 }, { "epoch": 0.4737964493255851, "grad_norm": 2.2045629024505615, "learning_rate": 8.2228e-06, "loss": 0.037, "step": 8887 }, { "epoch": 0.47384976275523805, "grad_norm": 1.9491456747055054, "learning_rate": 8.2226e-06, "loss": 0.0191, "step": 8888 }, { "epoch": 0.47390307618489097, "grad_norm": 3.2646069526672363, "learning_rate": 8.222400000000001e-06, "loss": 0.0439, "step": 8889 }, { "epoch": 0.4739563896145439, "grad_norm": 4.330398082733154, "learning_rate": 8.2222e-06, "loss": 0.0254, "step": 8890 }, { "epoch": 0.4740097030441968, "grad_norm": 3.8691370487213135, "learning_rate": 8.222000000000002e-06, "loss": 0.0866, "step": 8891 }, { "epoch": 0.47406301647384974, "grad_norm": 2.622554302215576, "learning_rate": 8.221800000000001e-06, "loss": 0.0575, "step": 8892 }, { "epoch": 0.47411632990350266, "grad_norm": 3.184807538986206, "learning_rate": 8.221600000000001e-06, "loss": 0.0287, "step": 8893 }, { "epoch": 0.47416964333315564, "grad_norm": 2.2267556190490723, "learning_rate": 8.2214e-06, "loss": 0.0486, "step": 8894 }, { "epoch": 0.47422295676280857, "grad_norm": 4.479766845703125, "learning_rate": 8.2212e-06, "loss": 0.0268, "step": 8895 }, { "epoch": 0.4742762701924615, "grad_norm": 2.6494884490966797, "learning_rate": 8.221000000000001e-06, "loss": 0.0186, "step": 8896 }, { "epoch": 0.4743295836221144, "grad_norm": 2.7346184253692627, "learning_rate": 8.220800000000001e-06, "loss": 0.024, "step": 8897 }, { "epoch": 0.47438289705176734, "grad_norm": 2.7408769130706787, "learning_rate": 8.2206e-06, "loss": 0.0473, "step": 8898 }, { "epoch": 0.47443621048142026, "grad_norm": 2.4349305629730225, "learning_rate": 8.2204e-06, "loss": 0.0316, "step": 8899 }, { "epoch": 0.4744895239110732, "grad_norm": 2.5667338371276855, "learning_rate": 8.2202e-06, "loss": 0.0382, "step": 8900 }, { "epoch": 0.4745428373407261, "grad_norm": 2.6421070098876953, "learning_rate": 8.220000000000001e-06, "loss": 0.0453, "step": 8901 }, { "epoch": 0.47459615077037903, "grad_norm": 4.0245208740234375, "learning_rate": 8.2198e-06, "loss": 0.0533, "step": 8902 }, { "epoch": 0.474649464200032, "grad_norm": 2.3726491928100586, "learning_rate": 8.219600000000002e-06, "loss": 0.0384, "step": 8903 }, { "epoch": 0.47470277762968494, "grad_norm": 1.9985893964767456, "learning_rate": 8.219400000000001e-06, "loss": 0.0349, "step": 8904 }, { "epoch": 0.47475609105933786, "grad_norm": 2.0753047466278076, "learning_rate": 8.219200000000001e-06, "loss": 0.0449, "step": 8905 }, { "epoch": 0.4748094044889908, "grad_norm": 2.069347620010376, "learning_rate": 8.219e-06, "loss": 0.0339, "step": 8906 }, { "epoch": 0.4748627179186437, "grad_norm": 1.858189582824707, "learning_rate": 8.218800000000002e-06, "loss": 0.0286, "step": 8907 }, { "epoch": 0.47491603134829663, "grad_norm": 2.595280408859253, "learning_rate": 8.218600000000001e-06, "loss": 0.041, "step": 8908 }, { "epoch": 0.47496934477794955, "grad_norm": 2.3032174110412598, "learning_rate": 8.2184e-06, "loss": 0.024, "step": 8909 }, { "epoch": 0.4750226582076025, "grad_norm": 2.8918917179107666, "learning_rate": 8.2182e-06, "loss": 0.0528, "step": 8910 }, { "epoch": 0.4750759716372554, "grad_norm": 2.682584285736084, "learning_rate": 8.218e-06, "loss": 0.0213, "step": 8911 }, { "epoch": 0.4751292850669084, "grad_norm": 2.5050885677337646, "learning_rate": 8.2178e-06, "loss": 0.0304, "step": 8912 }, { "epoch": 0.4751825984965613, "grad_norm": 3.9149210453033447, "learning_rate": 8.2176e-06, "loss": 0.0609, "step": 8913 }, { "epoch": 0.4752359119262142, "grad_norm": 1.7711468935012817, "learning_rate": 8.2174e-06, "loss": 0.0372, "step": 8914 }, { "epoch": 0.47528922535586715, "grad_norm": 2.5640952587127686, "learning_rate": 8.217200000000002e-06, "loss": 0.0236, "step": 8915 }, { "epoch": 0.4753425387855201, "grad_norm": 3.181270122528076, "learning_rate": 8.217000000000001e-06, "loss": 0.0633, "step": 8916 }, { "epoch": 0.475395852215173, "grad_norm": 2.877782106399536, "learning_rate": 8.2168e-06, "loss": 0.0364, "step": 8917 }, { "epoch": 0.4754491656448259, "grad_norm": 2.883896589279175, "learning_rate": 8.2166e-06, "loss": 0.0232, "step": 8918 }, { "epoch": 0.47550247907447885, "grad_norm": 3.111405611038208, "learning_rate": 8.216400000000001e-06, "loss": 0.0594, "step": 8919 }, { "epoch": 0.47555579250413177, "grad_norm": 2.92341947555542, "learning_rate": 8.216200000000001e-06, "loss": 0.0274, "step": 8920 }, { "epoch": 0.4756091059337847, "grad_norm": 1.9834249019622803, "learning_rate": 8.216e-06, "loss": 0.048, "step": 8921 }, { "epoch": 0.47566241936343767, "grad_norm": 2.9510550498962402, "learning_rate": 8.2158e-06, "loss": 0.045, "step": 8922 }, { "epoch": 0.4757157327930906, "grad_norm": 4.379228591918945, "learning_rate": 8.2156e-06, "loss": 0.0839, "step": 8923 }, { "epoch": 0.4757690462227435, "grad_norm": 4.268551349639893, "learning_rate": 8.215400000000001e-06, "loss": 0.0506, "step": 8924 }, { "epoch": 0.47582235965239644, "grad_norm": 4.3950653076171875, "learning_rate": 8.2152e-06, "loss": 0.0509, "step": 8925 }, { "epoch": 0.47587567308204937, "grad_norm": 2.9370622634887695, "learning_rate": 8.215e-06, "loss": 0.0212, "step": 8926 }, { "epoch": 0.4759289865117023, "grad_norm": 2.276155710220337, "learning_rate": 8.214800000000001e-06, "loss": 0.0424, "step": 8927 }, { "epoch": 0.4759822999413552, "grad_norm": 3.156447172164917, "learning_rate": 8.214600000000001e-06, "loss": 0.0358, "step": 8928 }, { "epoch": 0.47603561337100814, "grad_norm": 2.5962939262390137, "learning_rate": 8.2144e-06, "loss": 0.0402, "step": 8929 }, { "epoch": 0.47608892680066106, "grad_norm": 2.289649724960327, "learning_rate": 8.214200000000002e-06, "loss": 0.0525, "step": 8930 }, { "epoch": 0.47614224023031404, "grad_norm": 2.1199862957000732, "learning_rate": 8.214000000000001e-06, "loss": 0.0266, "step": 8931 }, { "epoch": 0.47619555365996696, "grad_norm": 5.154075622558594, "learning_rate": 8.213800000000001e-06, "loss": 0.0797, "step": 8932 }, { "epoch": 0.4762488670896199, "grad_norm": 5.784501552581787, "learning_rate": 8.2136e-06, "loss": 0.0393, "step": 8933 }, { "epoch": 0.4763021805192728, "grad_norm": 2.4071156978607178, "learning_rate": 8.2134e-06, "loss": 0.0311, "step": 8934 }, { "epoch": 0.47635549394892573, "grad_norm": 1.7119979858398438, "learning_rate": 8.213200000000001e-06, "loss": 0.0334, "step": 8935 }, { "epoch": 0.47640880737857866, "grad_norm": 1.9862598180770874, "learning_rate": 8.213e-06, "loss": 0.0366, "step": 8936 }, { "epoch": 0.4764621208082316, "grad_norm": 3.5808894634246826, "learning_rate": 8.2128e-06, "loss": 0.0579, "step": 8937 }, { "epoch": 0.4765154342378845, "grad_norm": 2.0894618034362793, "learning_rate": 8.2126e-06, "loss": 0.0176, "step": 8938 }, { "epoch": 0.47656874766753743, "grad_norm": 3.966552495956421, "learning_rate": 8.212400000000001e-06, "loss": 0.0482, "step": 8939 }, { "epoch": 0.4766220610971904, "grad_norm": 1.8933055400848389, "learning_rate": 8.2122e-06, "loss": 0.0146, "step": 8940 }, { "epoch": 0.47667537452684333, "grad_norm": 3.6950364112854004, "learning_rate": 8.212e-06, "loss": 0.0419, "step": 8941 }, { "epoch": 0.47672868795649626, "grad_norm": 1.7034257650375366, "learning_rate": 8.211800000000002e-06, "loss": 0.0168, "step": 8942 }, { "epoch": 0.4767820013861492, "grad_norm": 2.3340811729431152, "learning_rate": 8.211600000000001e-06, "loss": 0.033, "step": 8943 }, { "epoch": 0.4768353148158021, "grad_norm": 2.5390942096710205, "learning_rate": 8.2114e-06, "loss": 0.0386, "step": 8944 }, { "epoch": 0.476888628245455, "grad_norm": 2.3120486736297607, "learning_rate": 8.2112e-06, "loss": 0.0386, "step": 8945 }, { "epoch": 0.47694194167510795, "grad_norm": 1.9990590810775757, "learning_rate": 8.211000000000002e-06, "loss": 0.0379, "step": 8946 }, { "epoch": 0.4769952551047609, "grad_norm": 1.0862071514129639, "learning_rate": 8.210800000000001e-06, "loss": 0.0176, "step": 8947 }, { "epoch": 0.4770485685344138, "grad_norm": 1.2942639589309692, "learning_rate": 8.2106e-06, "loss": 0.0106, "step": 8948 }, { "epoch": 0.4771018819640667, "grad_norm": 6.162423133850098, "learning_rate": 8.2104e-06, "loss": 0.0495, "step": 8949 }, { "epoch": 0.4771551953937197, "grad_norm": 2.9294090270996094, "learning_rate": 8.2102e-06, "loss": 0.0465, "step": 8950 }, { "epoch": 0.4772085088233726, "grad_norm": 1.8959368467330933, "learning_rate": 8.210000000000001e-06, "loss": 0.0357, "step": 8951 }, { "epoch": 0.47726182225302555, "grad_norm": 2.0106098651885986, "learning_rate": 8.2098e-06, "loss": 0.0629, "step": 8952 }, { "epoch": 0.47731513568267847, "grad_norm": 2.315397024154663, "learning_rate": 8.2096e-06, "loss": 0.0337, "step": 8953 }, { "epoch": 0.4773684491123314, "grad_norm": 3.1004977226257324, "learning_rate": 8.209400000000001e-06, "loss": 0.0169, "step": 8954 }, { "epoch": 0.4774217625419843, "grad_norm": 1.3206180334091187, "learning_rate": 8.209200000000001e-06, "loss": 0.0199, "step": 8955 }, { "epoch": 0.47747507597163724, "grad_norm": 3.156766414642334, "learning_rate": 8.209e-06, "loss": 0.0348, "step": 8956 }, { "epoch": 0.47752838940129017, "grad_norm": 4.752356052398682, "learning_rate": 8.208800000000002e-06, "loss": 0.0665, "step": 8957 }, { "epoch": 0.4775817028309431, "grad_norm": 3.37424898147583, "learning_rate": 8.208600000000001e-06, "loss": 0.0216, "step": 8958 }, { "epoch": 0.47763501626059607, "grad_norm": 3.2124838829040527, "learning_rate": 8.208400000000001e-06, "loss": 0.0167, "step": 8959 }, { "epoch": 0.477688329690249, "grad_norm": 4.249020099639893, "learning_rate": 8.2082e-06, "loss": 0.0249, "step": 8960 }, { "epoch": 0.4777416431199019, "grad_norm": 1.9658246040344238, "learning_rate": 8.208e-06, "loss": 0.0324, "step": 8961 }, { "epoch": 0.47779495654955484, "grad_norm": 2.753805637359619, "learning_rate": 8.2078e-06, "loss": 0.0288, "step": 8962 }, { "epoch": 0.47784826997920776, "grad_norm": 3.1671946048736572, "learning_rate": 8.2076e-06, "loss": 0.0361, "step": 8963 }, { "epoch": 0.4779015834088607, "grad_norm": 4.258285999298096, "learning_rate": 8.2074e-06, "loss": 0.037, "step": 8964 }, { "epoch": 0.4779548968385136, "grad_norm": 1.5948920249938965, "learning_rate": 8.2072e-06, "loss": 0.0217, "step": 8965 }, { "epoch": 0.47800821026816653, "grad_norm": 4.019404411315918, "learning_rate": 8.207000000000001e-06, "loss": 0.0641, "step": 8966 }, { "epoch": 0.47806152369781946, "grad_norm": 5.127399921417236, "learning_rate": 8.2068e-06, "loss": 0.0624, "step": 8967 }, { "epoch": 0.47811483712747244, "grad_norm": 3.126986265182495, "learning_rate": 8.2066e-06, "loss": 0.0565, "step": 8968 }, { "epoch": 0.47816815055712536, "grad_norm": 4.414942741394043, "learning_rate": 8.206400000000002e-06, "loss": 0.0655, "step": 8969 }, { "epoch": 0.4782214639867783, "grad_norm": 4.48086404800415, "learning_rate": 8.206200000000001e-06, "loss": 0.0385, "step": 8970 }, { "epoch": 0.4782747774164312, "grad_norm": 2.885453224182129, "learning_rate": 8.206e-06, "loss": 0.055, "step": 8971 }, { "epoch": 0.47832809084608413, "grad_norm": 3.2163522243499756, "learning_rate": 8.2058e-06, "loss": 0.0282, "step": 8972 }, { "epoch": 0.47838140427573705, "grad_norm": 3.9699082374572754, "learning_rate": 8.2056e-06, "loss": 0.072, "step": 8973 }, { "epoch": 0.47843471770539, "grad_norm": 3.3523693084716797, "learning_rate": 8.205400000000001e-06, "loss": 0.0183, "step": 8974 }, { "epoch": 0.4784880311350429, "grad_norm": 1.1943917274475098, "learning_rate": 8.2052e-06, "loss": 0.0087, "step": 8975 }, { "epoch": 0.4785413445646958, "grad_norm": 2.0220534801483154, "learning_rate": 8.205e-06, "loss": 0.0192, "step": 8976 }, { "epoch": 0.47859465799434875, "grad_norm": 3.9047160148620605, "learning_rate": 8.2048e-06, "loss": 0.0847, "step": 8977 }, { "epoch": 0.47864797142400173, "grad_norm": 2.4837207794189453, "learning_rate": 8.204600000000001e-06, "loss": 0.0526, "step": 8978 }, { "epoch": 0.47870128485365465, "grad_norm": 3.152461528778076, "learning_rate": 8.2044e-06, "loss": 0.0763, "step": 8979 }, { "epoch": 0.4787545982833076, "grad_norm": 2.6425952911376953, "learning_rate": 8.204200000000002e-06, "loss": 0.0355, "step": 8980 }, { "epoch": 0.4788079117129605, "grad_norm": 2.7147419452667236, "learning_rate": 8.204000000000001e-06, "loss": 0.0426, "step": 8981 }, { "epoch": 0.4788612251426134, "grad_norm": 3.6935665607452393, "learning_rate": 8.203800000000001e-06, "loss": 0.0581, "step": 8982 }, { "epoch": 0.47891453857226635, "grad_norm": 1.7207767963409424, "learning_rate": 8.2036e-06, "loss": 0.0307, "step": 8983 }, { "epoch": 0.47896785200191927, "grad_norm": 2.0497329235076904, "learning_rate": 8.2034e-06, "loss": 0.0475, "step": 8984 }, { "epoch": 0.4790211654315722, "grad_norm": 1.9164552688598633, "learning_rate": 8.203200000000001e-06, "loss": 0.0362, "step": 8985 }, { "epoch": 0.4790744788612251, "grad_norm": 2.0103518962860107, "learning_rate": 8.203000000000001e-06, "loss": 0.0223, "step": 8986 }, { "epoch": 0.4791277922908781, "grad_norm": 3.0172414779663086, "learning_rate": 8.2028e-06, "loss": 0.0613, "step": 8987 }, { "epoch": 0.479181105720531, "grad_norm": 2.219801664352417, "learning_rate": 8.2026e-06, "loss": 0.0204, "step": 8988 }, { "epoch": 0.47923441915018394, "grad_norm": 1.5974377393722534, "learning_rate": 8.2024e-06, "loss": 0.0333, "step": 8989 }, { "epoch": 0.47928773257983687, "grad_norm": 1.3325103521347046, "learning_rate": 8.202200000000001e-06, "loss": 0.0252, "step": 8990 }, { "epoch": 0.4793410460094898, "grad_norm": 1.8552812337875366, "learning_rate": 8.202e-06, "loss": 0.0205, "step": 8991 }, { "epoch": 0.4793943594391427, "grad_norm": 1.8657212257385254, "learning_rate": 8.201800000000002e-06, "loss": 0.037, "step": 8992 }, { "epoch": 0.47944767286879564, "grad_norm": 3.7708394527435303, "learning_rate": 8.201600000000001e-06, "loss": 0.0696, "step": 8993 }, { "epoch": 0.47950098629844856, "grad_norm": 1.7076905965805054, "learning_rate": 8.2014e-06, "loss": 0.0257, "step": 8994 }, { "epoch": 0.4795542997281015, "grad_norm": 1.3954778909683228, "learning_rate": 8.2012e-06, "loss": 0.0197, "step": 8995 }, { "epoch": 0.47960761315775446, "grad_norm": 2.2308766841888428, "learning_rate": 8.201000000000002e-06, "loss": 0.0283, "step": 8996 }, { "epoch": 0.4796609265874074, "grad_norm": 2.42807936668396, "learning_rate": 8.200800000000001e-06, "loss": 0.0536, "step": 8997 }, { "epoch": 0.4797142400170603, "grad_norm": 3.555238962173462, "learning_rate": 8.2006e-06, "loss": 0.044, "step": 8998 }, { "epoch": 0.47976755344671324, "grad_norm": 1.6272342205047607, "learning_rate": 8.2004e-06, "loss": 0.0136, "step": 8999 }, { "epoch": 0.47982086687636616, "grad_norm": 1.805322527885437, "learning_rate": 8.2002e-06, "loss": 0.0295, "step": 9000 }, { "epoch": 0.4798741803060191, "grad_norm": 1.6403402090072632, "learning_rate": 8.2e-06, "loss": 0.023, "step": 9001 }, { "epoch": 0.479927493735672, "grad_norm": 4.0205302238464355, "learning_rate": 8.1998e-06, "loss": 0.0705, "step": 9002 }, { "epoch": 0.47998080716532493, "grad_norm": 2.3143842220306396, "learning_rate": 8.1996e-06, "loss": 0.0509, "step": 9003 }, { "epoch": 0.48003412059497785, "grad_norm": 2.6856465339660645, "learning_rate": 8.199400000000001e-06, "loss": 0.0628, "step": 9004 }, { "epoch": 0.4800874340246308, "grad_norm": 2.4919204711914062, "learning_rate": 8.199200000000001e-06, "loss": 0.0371, "step": 9005 }, { "epoch": 0.48014074745428376, "grad_norm": 2.4968104362487793, "learning_rate": 8.199e-06, "loss": 0.0503, "step": 9006 }, { "epoch": 0.4801940608839367, "grad_norm": 1.9463326930999756, "learning_rate": 8.198800000000002e-06, "loss": 0.0241, "step": 9007 }, { "epoch": 0.4802473743135896, "grad_norm": 1.6418225765228271, "learning_rate": 8.198600000000001e-06, "loss": 0.0381, "step": 9008 }, { "epoch": 0.4803006877432425, "grad_norm": 2.005028009414673, "learning_rate": 8.198400000000001e-06, "loss": 0.0343, "step": 9009 }, { "epoch": 0.48035400117289545, "grad_norm": 2.6306183338165283, "learning_rate": 8.1982e-06, "loss": 0.0501, "step": 9010 }, { "epoch": 0.4804073146025484, "grad_norm": 2.6071765422821045, "learning_rate": 8.198e-06, "loss": 0.0564, "step": 9011 }, { "epoch": 0.4804606280322013, "grad_norm": 1.3031742572784424, "learning_rate": 8.1978e-06, "loss": 0.0118, "step": 9012 }, { "epoch": 0.4805139414618542, "grad_norm": 4.249506950378418, "learning_rate": 8.197600000000001e-06, "loss": 0.0394, "step": 9013 }, { "epoch": 0.48056725489150715, "grad_norm": 2.0636231899261475, "learning_rate": 8.1974e-06, "loss": 0.035, "step": 9014 }, { "epoch": 0.4806205683211601, "grad_norm": 1.7832305431365967, "learning_rate": 8.1972e-06, "loss": 0.0311, "step": 9015 }, { "epoch": 0.48067388175081305, "grad_norm": 3.1289114952087402, "learning_rate": 8.197000000000001e-06, "loss": 0.0418, "step": 9016 }, { "epoch": 0.48072719518046597, "grad_norm": 1.9017770290374756, "learning_rate": 8.196800000000001e-06, "loss": 0.0231, "step": 9017 }, { "epoch": 0.4807805086101189, "grad_norm": 1.68905508518219, "learning_rate": 8.1966e-06, "loss": 0.024, "step": 9018 }, { "epoch": 0.4808338220397718, "grad_norm": 1.628890872001648, "learning_rate": 8.196400000000002e-06, "loss": 0.0208, "step": 9019 }, { "epoch": 0.48088713546942474, "grad_norm": 1.8847017288208008, "learning_rate": 8.196200000000001e-06, "loss": 0.0355, "step": 9020 }, { "epoch": 0.48094044889907767, "grad_norm": 3.1255064010620117, "learning_rate": 8.196e-06, "loss": 0.0776, "step": 9021 }, { "epoch": 0.4809937623287306, "grad_norm": 2.566804885864258, "learning_rate": 8.1958e-06, "loss": 0.0358, "step": 9022 }, { "epoch": 0.4810470757583835, "grad_norm": 1.8060883283615112, "learning_rate": 8.1956e-06, "loss": 0.0351, "step": 9023 }, { "epoch": 0.4811003891880365, "grad_norm": 1.7021862268447876, "learning_rate": 8.195400000000001e-06, "loss": 0.0198, "step": 9024 }, { "epoch": 0.4811537026176894, "grad_norm": 2.8179407119750977, "learning_rate": 8.1952e-06, "loss": 0.0228, "step": 9025 }, { "epoch": 0.48120701604734234, "grad_norm": 1.9489026069641113, "learning_rate": 8.195e-06, "loss": 0.0346, "step": 9026 }, { "epoch": 0.48126032947699526, "grad_norm": 1.4079350233078003, "learning_rate": 8.1948e-06, "loss": 0.0129, "step": 9027 }, { "epoch": 0.4813136429066482, "grad_norm": 2.9499759674072266, "learning_rate": 8.194600000000001e-06, "loss": 0.0544, "step": 9028 }, { "epoch": 0.4813669563363011, "grad_norm": 5.414252281188965, "learning_rate": 8.1944e-06, "loss": 0.0582, "step": 9029 }, { "epoch": 0.48142026976595403, "grad_norm": 2.25070858001709, "learning_rate": 8.1942e-06, "loss": 0.0297, "step": 9030 }, { "epoch": 0.48147358319560696, "grad_norm": 2.9235267639160156, "learning_rate": 8.194000000000002e-06, "loss": 0.0422, "step": 9031 }, { "epoch": 0.4815268966252599, "grad_norm": 1.8936634063720703, "learning_rate": 8.193800000000001e-06, "loss": 0.0205, "step": 9032 }, { "epoch": 0.4815802100549128, "grad_norm": 3.794929265975952, "learning_rate": 8.1936e-06, "loss": 0.0773, "step": 9033 }, { "epoch": 0.4816335234845658, "grad_norm": 3.283898115158081, "learning_rate": 8.1934e-06, "loss": 0.0339, "step": 9034 }, { "epoch": 0.4816868369142187, "grad_norm": 4.146723747253418, "learning_rate": 8.193200000000001e-06, "loss": 0.0852, "step": 9035 }, { "epoch": 0.48174015034387163, "grad_norm": 2.552983522415161, "learning_rate": 8.193000000000001e-06, "loss": 0.0194, "step": 9036 }, { "epoch": 0.48179346377352456, "grad_norm": 2.151688814163208, "learning_rate": 8.1928e-06, "loss": 0.0317, "step": 9037 }, { "epoch": 0.4818467772031775, "grad_norm": 2.310262441635132, "learning_rate": 8.1926e-06, "loss": 0.0448, "step": 9038 }, { "epoch": 0.4819000906328304, "grad_norm": 0.9088316559791565, "learning_rate": 8.1924e-06, "loss": 0.0082, "step": 9039 }, { "epoch": 0.4819534040624833, "grad_norm": 3.602900505065918, "learning_rate": 8.192200000000001e-06, "loss": 0.044, "step": 9040 }, { "epoch": 0.48200671749213625, "grad_norm": 1.8797646760940552, "learning_rate": 8.192e-06, "loss": 0.0275, "step": 9041 }, { "epoch": 0.4820600309217892, "grad_norm": 2.816840648651123, "learning_rate": 8.1918e-06, "loss": 0.0425, "step": 9042 }, { "epoch": 0.48211334435144215, "grad_norm": 3.347832202911377, "learning_rate": 8.191600000000001e-06, "loss": 0.0543, "step": 9043 }, { "epoch": 0.4821666577810951, "grad_norm": 3.5675809383392334, "learning_rate": 8.191400000000001e-06, "loss": 0.0228, "step": 9044 }, { "epoch": 0.482219971210748, "grad_norm": 3.515641689300537, "learning_rate": 8.1912e-06, "loss": 0.0298, "step": 9045 }, { "epoch": 0.4822732846404009, "grad_norm": 1.337717890739441, "learning_rate": 8.191000000000002e-06, "loss": 0.0111, "step": 9046 }, { "epoch": 0.48232659807005385, "grad_norm": 2.3327693939208984, "learning_rate": 8.190800000000001e-06, "loss": 0.0384, "step": 9047 }, { "epoch": 0.48237991149970677, "grad_norm": 3.3363430500030518, "learning_rate": 8.1906e-06, "loss": 0.0455, "step": 9048 }, { "epoch": 0.4824332249293597, "grad_norm": 2.631272554397583, "learning_rate": 8.1904e-06, "loss": 0.0386, "step": 9049 }, { "epoch": 0.4824865383590126, "grad_norm": 1.903005838394165, "learning_rate": 8.1902e-06, "loss": 0.0172, "step": 9050 }, { "epoch": 0.48253985178866554, "grad_norm": 3.3136274814605713, "learning_rate": 8.19e-06, "loss": 0.015, "step": 9051 }, { "epoch": 0.4825931652183185, "grad_norm": 1.9252310991287231, "learning_rate": 8.1898e-06, "loss": 0.0354, "step": 9052 }, { "epoch": 0.48264647864797144, "grad_norm": 4.449656963348389, "learning_rate": 8.1896e-06, "loss": 0.0547, "step": 9053 }, { "epoch": 0.48269979207762437, "grad_norm": 2.923811674118042, "learning_rate": 8.1894e-06, "loss": 0.0341, "step": 9054 }, { "epoch": 0.4827531055072773, "grad_norm": 3.4447755813598633, "learning_rate": 8.189200000000001e-06, "loss": 0.0352, "step": 9055 }, { "epoch": 0.4828064189369302, "grad_norm": 3.7983264923095703, "learning_rate": 8.189e-06, "loss": 0.0507, "step": 9056 }, { "epoch": 0.48285973236658314, "grad_norm": 5.629607677459717, "learning_rate": 8.188800000000002e-06, "loss": 0.0478, "step": 9057 }, { "epoch": 0.48291304579623606, "grad_norm": 1.58505380153656, "learning_rate": 8.188600000000002e-06, "loss": 0.0351, "step": 9058 }, { "epoch": 0.482966359225889, "grad_norm": 2.3789937496185303, "learning_rate": 8.188400000000001e-06, "loss": 0.0376, "step": 9059 }, { "epoch": 0.4830196726555419, "grad_norm": 3.325716972351074, "learning_rate": 8.1882e-06, "loss": 0.0378, "step": 9060 }, { "epoch": 0.48307298608519483, "grad_norm": 2.7318437099456787, "learning_rate": 8.188e-06, "loss": 0.0466, "step": 9061 }, { "epoch": 0.4831262995148478, "grad_norm": 2.808838367462158, "learning_rate": 8.1878e-06, "loss": 0.0133, "step": 9062 }, { "epoch": 0.48317961294450074, "grad_norm": 4.1846795082092285, "learning_rate": 8.187600000000001e-06, "loss": 0.0479, "step": 9063 }, { "epoch": 0.48323292637415366, "grad_norm": 3.5095529556274414, "learning_rate": 8.1874e-06, "loss": 0.0822, "step": 9064 }, { "epoch": 0.4832862398038066, "grad_norm": 2.063891887664795, "learning_rate": 8.1872e-06, "loss": 0.0358, "step": 9065 }, { "epoch": 0.4833395532334595, "grad_norm": 4.581962585449219, "learning_rate": 8.187e-06, "loss": 0.0669, "step": 9066 }, { "epoch": 0.48339286666311243, "grad_norm": 2.5126588344573975, "learning_rate": 8.186800000000001e-06, "loss": 0.0323, "step": 9067 }, { "epoch": 0.48344618009276535, "grad_norm": 2.985987424850464, "learning_rate": 8.1866e-06, "loss": 0.0237, "step": 9068 }, { "epoch": 0.4834994935224183, "grad_norm": 4.789019584655762, "learning_rate": 8.186400000000002e-06, "loss": 0.044, "step": 9069 }, { "epoch": 0.4835528069520712, "grad_norm": 1.9931046962738037, "learning_rate": 8.186200000000001e-06, "loss": 0.0219, "step": 9070 }, { "epoch": 0.4836061203817242, "grad_norm": 2.2298619747161865, "learning_rate": 8.186000000000001e-06, "loss": 0.0525, "step": 9071 }, { "epoch": 0.4836594338113771, "grad_norm": 3.583220958709717, "learning_rate": 8.1858e-06, "loss": 0.0366, "step": 9072 }, { "epoch": 0.48371274724103, "grad_norm": 4.108738899230957, "learning_rate": 8.1856e-06, "loss": 0.0407, "step": 9073 }, { "epoch": 0.48376606067068295, "grad_norm": 4.8345489501953125, "learning_rate": 8.185400000000001e-06, "loss": 0.0379, "step": 9074 }, { "epoch": 0.4838193741003359, "grad_norm": 3.1231026649475098, "learning_rate": 8.1852e-06, "loss": 0.0452, "step": 9075 }, { "epoch": 0.4838726875299888, "grad_norm": 2.1397557258605957, "learning_rate": 8.185e-06, "loss": 0.0213, "step": 9076 }, { "epoch": 0.4839260009596417, "grad_norm": 4.946019649505615, "learning_rate": 8.1848e-06, "loss": 0.0671, "step": 9077 }, { "epoch": 0.48397931438929465, "grad_norm": 4.907201766967773, "learning_rate": 8.1846e-06, "loss": 0.0719, "step": 9078 }, { "epoch": 0.48403262781894757, "grad_norm": 3.106257677078247, "learning_rate": 8.1844e-06, "loss": 0.0703, "step": 9079 }, { "epoch": 0.48408594124860055, "grad_norm": 1.370007872581482, "learning_rate": 8.1842e-06, "loss": 0.0125, "step": 9080 }, { "epoch": 0.4841392546782535, "grad_norm": 3.247133493423462, "learning_rate": 8.184000000000002e-06, "loss": 0.0529, "step": 9081 }, { "epoch": 0.4841925681079064, "grad_norm": 2.691318988800049, "learning_rate": 8.183800000000001e-06, "loss": 0.0242, "step": 9082 }, { "epoch": 0.4842458815375593, "grad_norm": 1.8551360368728638, "learning_rate": 8.1836e-06, "loss": 0.0313, "step": 9083 }, { "epoch": 0.48429919496721224, "grad_norm": 1.258612036705017, "learning_rate": 8.1834e-06, "loss": 0.0121, "step": 9084 }, { "epoch": 0.48435250839686517, "grad_norm": 1.6970194578170776, "learning_rate": 8.183200000000002e-06, "loss": 0.0231, "step": 9085 }, { "epoch": 0.4844058218265181, "grad_norm": 2.86560320854187, "learning_rate": 8.183000000000001e-06, "loss": 0.0395, "step": 9086 }, { "epoch": 0.484459135256171, "grad_norm": 3.1573305130004883, "learning_rate": 8.1828e-06, "loss": 0.0298, "step": 9087 }, { "epoch": 0.48451244868582394, "grad_norm": 3.7269957065582275, "learning_rate": 8.1826e-06, "loss": 0.0449, "step": 9088 }, { "epoch": 0.48456576211547686, "grad_norm": 3.1279332637786865, "learning_rate": 8.1824e-06, "loss": 0.0531, "step": 9089 }, { "epoch": 0.48461907554512984, "grad_norm": 1.6694142818450928, "learning_rate": 8.1822e-06, "loss": 0.0179, "step": 9090 }, { "epoch": 0.48467238897478276, "grad_norm": 6.1644110679626465, "learning_rate": 8.182e-06, "loss": 0.0771, "step": 9091 }, { "epoch": 0.4847257024044357, "grad_norm": 2.3233978748321533, "learning_rate": 8.1818e-06, "loss": 0.0295, "step": 9092 }, { "epoch": 0.4847790158340886, "grad_norm": 2.4670705795288086, "learning_rate": 8.181600000000001e-06, "loss": 0.0242, "step": 9093 }, { "epoch": 0.48483232926374153, "grad_norm": 2.1515703201293945, "learning_rate": 8.181400000000001e-06, "loss": 0.0412, "step": 9094 }, { "epoch": 0.48488564269339446, "grad_norm": 2.6008429527282715, "learning_rate": 8.1812e-06, "loss": 0.0159, "step": 9095 }, { "epoch": 0.4849389561230474, "grad_norm": 2.731457233428955, "learning_rate": 8.181000000000002e-06, "loss": 0.0424, "step": 9096 }, { "epoch": 0.4849922695527003, "grad_norm": 2.4045071601867676, "learning_rate": 8.180800000000001e-06, "loss": 0.0402, "step": 9097 }, { "epoch": 0.48504558298235323, "grad_norm": 2.441932201385498, "learning_rate": 8.180600000000001e-06, "loss": 0.037, "step": 9098 }, { "epoch": 0.4850988964120062, "grad_norm": 2.654737710952759, "learning_rate": 8.1804e-06, "loss": 0.0279, "step": 9099 }, { "epoch": 0.48515220984165913, "grad_norm": 2.3710625171661377, "learning_rate": 8.1802e-06, "loss": 0.016, "step": 9100 }, { "epoch": 0.48520552327131206, "grad_norm": 2.6262385845184326, "learning_rate": 8.18e-06, "loss": 0.0269, "step": 9101 }, { "epoch": 0.485258836700965, "grad_norm": 3.739511728286743, "learning_rate": 8.179800000000001e-06, "loss": 0.0459, "step": 9102 }, { "epoch": 0.4853121501306179, "grad_norm": 3.141859292984009, "learning_rate": 8.1796e-06, "loss": 0.0473, "step": 9103 }, { "epoch": 0.4853654635602708, "grad_norm": 1.4925769567489624, "learning_rate": 8.1794e-06, "loss": 0.0117, "step": 9104 }, { "epoch": 0.48541877698992375, "grad_norm": 2.7539806365966797, "learning_rate": 8.179200000000001e-06, "loss": 0.0609, "step": 9105 }, { "epoch": 0.4854720904195767, "grad_norm": 2.254338264465332, "learning_rate": 8.179e-06, "loss": 0.0425, "step": 9106 }, { "epoch": 0.4855254038492296, "grad_norm": 3.5507025718688965, "learning_rate": 8.1788e-06, "loss": 0.0318, "step": 9107 }, { "epoch": 0.4855787172788826, "grad_norm": 2.414205312728882, "learning_rate": 8.178600000000002e-06, "loss": 0.0219, "step": 9108 }, { "epoch": 0.4856320307085355, "grad_norm": 2.646488904953003, "learning_rate": 8.178400000000001e-06, "loss": 0.0368, "step": 9109 }, { "epoch": 0.4856853441381884, "grad_norm": 2.466735363006592, "learning_rate": 8.1782e-06, "loss": 0.0396, "step": 9110 }, { "epoch": 0.48573865756784135, "grad_norm": 3.608999252319336, "learning_rate": 8.178e-06, "loss": 0.0721, "step": 9111 }, { "epoch": 0.48579197099749427, "grad_norm": 4.600935935974121, "learning_rate": 8.1778e-06, "loss": 0.0492, "step": 9112 }, { "epoch": 0.4858452844271472, "grad_norm": 2.5464653968811035, "learning_rate": 8.177600000000001e-06, "loss": 0.0357, "step": 9113 }, { "epoch": 0.4858985978568001, "grad_norm": 3.293787717819214, "learning_rate": 8.1774e-06, "loss": 0.0433, "step": 9114 }, { "epoch": 0.48595191128645304, "grad_norm": 5.945652008056641, "learning_rate": 8.1772e-06, "loss": 0.0442, "step": 9115 }, { "epoch": 0.48600522471610597, "grad_norm": 3.0608291625976562, "learning_rate": 8.177e-06, "loss": 0.0485, "step": 9116 }, { "epoch": 0.4860585381457589, "grad_norm": 2.835671901702881, "learning_rate": 8.176800000000001e-06, "loss": 0.0459, "step": 9117 }, { "epoch": 0.48611185157541187, "grad_norm": 3.803041696548462, "learning_rate": 8.1766e-06, "loss": 0.0504, "step": 9118 }, { "epoch": 0.4861651650050648, "grad_norm": 8.406722068786621, "learning_rate": 8.1764e-06, "loss": 0.0562, "step": 9119 }, { "epoch": 0.4862184784347177, "grad_norm": 3.3336119651794434, "learning_rate": 8.176200000000001e-06, "loss": 0.0274, "step": 9120 }, { "epoch": 0.48627179186437064, "grad_norm": 1.5864449739456177, "learning_rate": 8.176000000000001e-06, "loss": 0.0299, "step": 9121 }, { "epoch": 0.48632510529402356, "grad_norm": 2.387000322341919, "learning_rate": 8.1758e-06, "loss": 0.0182, "step": 9122 }, { "epoch": 0.4863784187236765, "grad_norm": 1.9381715059280396, "learning_rate": 8.1756e-06, "loss": 0.0602, "step": 9123 }, { "epoch": 0.4864317321533294, "grad_norm": 3.0303795337677, "learning_rate": 8.175400000000001e-06, "loss": 0.0278, "step": 9124 }, { "epoch": 0.48648504558298233, "grad_norm": 5.483022212982178, "learning_rate": 8.175200000000001e-06, "loss": 0.0317, "step": 9125 }, { "epoch": 0.48653835901263526, "grad_norm": 2.370408773422241, "learning_rate": 8.175e-06, "loss": 0.0353, "step": 9126 }, { "epoch": 0.48659167244228824, "grad_norm": 4.155689716339111, "learning_rate": 8.1748e-06, "loss": 0.0506, "step": 9127 }, { "epoch": 0.48664498587194116, "grad_norm": 1.4271658658981323, "learning_rate": 8.1746e-06, "loss": 0.0184, "step": 9128 }, { "epoch": 0.4866982993015941, "grad_norm": 2.5373454093933105, "learning_rate": 8.174400000000001e-06, "loss": 0.0359, "step": 9129 }, { "epoch": 0.486751612731247, "grad_norm": 4.2825608253479, "learning_rate": 8.1742e-06, "loss": 0.0447, "step": 9130 }, { "epoch": 0.48680492616089993, "grad_norm": 1.377557635307312, "learning_rate": 8.174e-06, "loss": 0.0257, "step": 9131 }, { "epoch": 0.48685823959055285, "grad_norm": 3.783775806427002, "learning_rate": 8.173800000000001e-06, "loss": 0.0508, "step": 9132 }, { "epoch": 0.4869115530202058, "grad_norm": 2.608930826187134, "learning_rate": 8.1736e-06, "loss": 0.0242, "step": 9133 }, { "epoch": 0.4869648664498587, "grad_norm": 2.3722853660583496, "learning_rate": 8.1734e-06, "loss": 0.0222, "step": 9134 }, { "epoch": 0.4870181798795116, "grad_norm": 4.64044713973999, "learning_rate": 8.173200000000002e-06, "loss": 0.0422, "step": 9135 }, { "epoch": 0.4870714933091646, "grad_norm": 2.1056361198425293, "learning_rate": 8.173000000000001e-06, "loss": 0.0342, "step": 9136 }, { "epoch": 0.48712480673881753, "grad_norm": 0.6675088405609131, "learning_rate": 8.1728e-06, "loss": 0.0062, "step": 9137 }, { "epoch": 0.48717812016847045, "grad_norm": 3.3489131927490234, "learning_rate": 8.1726e-06, "loss": 0.0268, "step": 9138 }, { "epoch": 0.4872314335981234, "grad_norm": 2.337164878845215, "learning_rate": 8.1724e-06, "loss": 0.0298, "step": 9139 }, { "epoch": 0.4872847470277763, "grad_norm": 3.5354530811309814, "learning_rate": 8.1722e-06, "loss": 0.0373, "step": 9140 }, { "epoch": 0.4873380604574292, "grad_norm": 4.664783000946045, "learning_rate": 8.172e-06, "loss": 0.0621, "step": 9141 }, { "epoch": 0.48739137388708215, "grad_norm": 2.9013171195983887, "learning_rate": 8.1718e-06, "loss": 0.0617, "step": 9142 }, { "epoch": 0.48744468731673507, "grad_norm": 2.703378677368164, "learning_rate": 8.1716e-06, "loss": 0.0574, "step": 9143 }, { "epoch": 0.487498000746388, "grad_norm": 3.6598193645477295, "learning_rate": 8.171400000000001e-06, "loss": 0.066, "step": 9144 }, { "epoch": 0.4875513141760409, "grad_norm": 2.082426071166992, "learning_rate": 8.1712e-06, "loss": 0.0238, "step": 9145 }, { "epoch": 0.4876046276056939, "grad_norm": 3.10364031791687, "learning_rate": 8.171000000000002e-06, "loss": 0.0273, "step": 9146 }, { "epoch": 0.4876579410353468, "grad_norm": 2.8276350498199463, "learning_rate": 8.170800000000001e-06, "loss": 0.0321, "step": 9147 }, { "epoch": 0.48771125446499974, "grad_norm": 3.6571428775787354, "learning_rate": 8.170600000000001e-06, "loss": 0.0321, "step": 9148 }, { "epoch": 0.48776456789465267, "grad_norm": 3.651560068130493, "learning_rate": 8.1704e-06, "loss": 0.0322, "step": 9149 }, { "epoch": 0.4878178813243056, "grad_norm": 2.4687516689300537, "learning_rate": 8.1702e-06, "loss": 0.055, "step": 9150 }, { "epoch": 0.4878711947539585, "grad_norm": 3.8949413299560547, "learning_rate": 8.17e-06, "loss": 0.0703, "step": 9151 }, { "epoch": 0.48792450818361144, "grad_norm": 1.9285080432891846, "learning_rate": 8.169800000000001e-06, "loss": 0.0402, "step": 9152 }, { "epoch": 0.48797782161326436, "grad_norm": 3.26725435256958, "learning_rate": 8.1696e-06, "loss": 0.064, "step": 9153 }, { "epoch": 0.4880311350429173, "grad_norm": 2.769909143447876, "learning_rate": 8.1694e-06, "loss": 0.0513, "step": 9154 }, { "epoch": 0.48808444847257026, "grad_norm": 4.207695484161377, "learning_rate": 8.1692e-06, "loss": 0.0451, "step": 9155 }, { "epoch": 0.4881377619022232, "grad_norm": 1.7367151975631714, "learning_rate": 8.169000000000001e-06, "loss": 0.019, "step": 9156 }, { "epoch": 0.4881910753318761, "grad_norm": 2.6302850246429443, "learning_rate": 8.1688e-06, "loss": 0.0206, "step": 9157 }, { "epoch": 0.48824438876152904, "grad_norm": 4.275787353515625, "learning_rate": 8.168600000000002e-06, "loss": 0.0339, "step": 9158 }, { "epoch": 0.48829770219118196, "grad_norm": 2.884936809539795, "learning_rate": 8.168400000000001e-06, "loss": 0.0304, "step": 9159 }, { "epoch": 0.4883510156208349, "grad_norm": 1.7655155658721924, "learning_rate": 8.1682e-06, "loss": 0.0383, "step": 9160 }, { "epoch": 0.4884043290504878, "grad_norm": 2.420416831970215, "learning_rate": 8.168e-06, "loss": 0.0366, "step": 9161 }, { "epoch": 0.48845764248014073, "grad_norm": 3.6360113620758057, "learning_rate": 8.1678e-06, "loss": 0.0375, "step": 9162 }, { "epoch": 0.48851095590979365, "grad_norm": 2.9223883152008057, "learning_rate": 8.167600000000001e-06, "loss": 0.04, "step": 9163 }, { "epoch": 0.48856426933944663, "grad_norm": 5.017273902893066, "learning_rate": 8.1674e-06, "loss": 0.1107, "step": 9164 }, { "epoch": 0.48861758276909956, "grad_norm": 3.581974744796753, "learning_rate": 8.1672e-06, "loss": 0.0645, "step": 9165 }, { "epoch": 0.4886708961987525, "grad_norm": 1.6380841732025146, "learning_rate": 8.167e-06, "loss": 0.0208, "step": 9166 }, { "epoch": 0.4887242096284054, "grad_norm": 2.0413920879364014, "learning_rate": 8.1668e-06, "loss": 0.0306, "step": 9167 }, { "epoch": 0.4887775230580583, "grad_norm": 3.7492525577545166, "learning_rate": 8.1666e-06, "loss": 0.0523, "step": 9168 }, { "epoch": 0.48883083648771125, "grad_norm": 4.600712299346924, "learning_rate": 8.1664e-06, "loss": 0.0421, "step": 9169 }, { "epoch": 0.4888841499173642, "grad_norm": 5.175541400909424, "learning_rate": 8.166200000000002e-06, "loss": 0.0588, "step": 9170 }, { "epoch": 0.4889374633470171, "grad_norm": 4.261321067810059, "learning_rate": 8.166000000000001e-06, "loss": 0.0426, "step": 9171 }, { "epoch": 0.48899077677667, "grad_norm": 2.687018871307373, "learning_rate": 8.1658e-06, "loss": 0.0518, "step": 9172 }, { "epoch": 0.48904409020632295, "grad_norm": 1.4886138439178467, "learning_rate": 8.1656e-06, "loss": 0.0286, "step": 9173 }, { "epoch": 0.4890974036359759, "grad_norm": 3.045618772506714, "learning_rate": 8.165400000000001e-06, "loss": 0.056, "step": 9174 }, { "epoch": 0.48915071706562885, "grad_norm": 1.4166321754455566, "learning_rate": 8.165200000000001e-06, "loss": 0.0459, "step": 9175 }, { "epoch": 0.48920403049528177, "grad_norm": 3.7010550498962402, "learning_rate": 8.165e-06, "loss": 0.0487, "step": 9176 }, { "epoch": 0.4892573439249347, "grad_norm": 2.0562655925750732, "learning_rate": 8.1648e-06, "loss": 0.0181, "step": 9177 }, { "epoch": 0.4893106573545876, "grad_norm": 2.876600980758667, "learning_rate": 8.1646e-06, "loss": 0.0638, "step": 9178 }, { "epoch": 0.48936397078424054, "grad_norm": 2.385267496109009, "learning_rate": 8.164400000000001e-06, "loss": 0.0372, "step": 9179 }, { "epoch": 0.48941728421389347, "grad_norm": 3.4275264739990234, "learning_rate": 8.1642e-06, "loss": 0.0481, "step": 9180 }, { "epoch": 0.4894705976435464, "grad_norm": 2.7889599800109863, "learning_rate": 8.164e-06, "loss": 0.0261, "step": 9181 }, { "epoch": 0.4895239110731993, "grad_norm": 2.201237201690674, "learning_rate": 8.163800000000001e-06, "loss": 0.0491, "step": 9182 }, { "epoch": 0.4895772245028523, "grad_norm": 1.4775182008743286, "learning_rate": 8.163600000000001e-06, "loss": 0.0377, "step": 9183 }, { "epoch": 0.4896305379325052, "grad_norm": 1.84212064743042, "learning_rate": 8.1634e-06, "loss": 0.0302, "step": 9184 }, { "epoch": 0.48968385136215814, "grad_norm": 2.5204434394836426, "learning_rate": 8.163200000000002e-06, "loss": 0.032, "step": 9185 }, { "epoch": 0.48973716479181106, "grad_norm": 1.8261085748672485, "learning_rate": 8.163000000000001e-06, "loss": 0.0479, "step": 9186 }, { "epoch": 0.489790478221464, "grad_norm": 3.8400115966796875, "learning_rate": 8.162800000000001e-06, "loss": 0.039, "step": 9187 }, { "epoch": 0.4898437916511169, "grad_norm": 2.3034956455230713, "learning_rate": 8.1626e-06, "loss": 0.0413, "step": 9188 }, { "epoch": 0.48989710508076983, "grad_norm": 2.4551796913146973, "learning_rate": 8.1624e-06, "loss": 0.0267, "step": 9189 }, { "epoch": 0.48995041851042276, "grad_norm": 2.9331934452056885, "learning_rate": 8.1622e-06, "loss": 0.033, "step": 9190 }, { "epoch": 0.4900037319400757, "grad_norm": 2.989027738571167, "learning_rate": 8.162e-06, "loss": 0.0449, "step": 9191 }, { "epoch": 0.49005704536972866, "grad_norm": 3.7388875484466553, "learning_rate": 8.1618e-06, "loss": 0.0903, "step": 9192 }, { "epoch": 0.4901103587993816, "grad_norm": 2.5576388835906982, "learning_rate": 8.1616e-06, "loss": 0.0261, "step": 9193 }, { "epoch": 0.4901636722290345, "grad_norm": 1.6347250938415527, "learning_rate": 8.161400000000001e-06, "loss": 0.026, "step": 9194 }, { "epoch": 0.49021698565868743, "grad_norm": 2.49912428855896, "learning_rate": 8.1612e-06, "loss": 0.0367, "step": 9195 }, { "epoch": 0.49027029908834036, "grad_norm": 2.2489190101623535, "learning_rate": 8.161e-06, "loss": 0.0273, "step": 9196 }, { "epoch": 0.4903236125179933, "grad_norm": 4.185606002807617, "learning_rate": 8.160800000000002e-06, "loss": 0.0725, "step": 9197 }, { "epoch": 0.4903769259476462, "grad_norm": 2.7386221885681152, "learning_rate": 8.160600000000001e-06, "loss": 0.0458, "step": 9198 }, { "epoch": 0.4904302393772991, "grad_norm": 2.208494186401367, "learning_rate": 8.1604e-06, "loss": 0.0401, "step": 9199 }, { "epoch": 0.49048355280695205, "grad_norm": 3.8634679317474365, "learning_rate": 8.1602e-06, "loss": 0.0581, "step": 9200 }, { "epoch": 0.490536866236605, "grad_norm": 3.0813965797424316, "learning_rate": 8.16e-06, "loss": 0.029, "step": 9201 }, { "epoch": 0.49059017966625795, "grad_norm": 2.085188865661621, "learning_rate": 8.159800000000001e-06, "loss": 0.0625, "step": 9202 }, { "epoch": 0.4906434930959109, "grad_norm": 4.170675277709961, "learning_rate": 8.1596e-06, "loss": 0.0551, "step": 9203 }, { "epoch": 0.4906968065255638, "grad_norm": 1.4971022605895996, "learning_rate": 8.1594e-06, "loss": 0.0253, "step": 9204 }, { "epoch": 0.4907501199552167, "grad_norm": 2.4521749019622803, "learning_rate": 8.1592e-06, "loss": 0.039, "step": 9205 }, { "epoch": 0.49080343338486965, "grad_norm": 1.7274272441864014, "learning_rate": 8.159000000000001e-06, "loss": 0.0336, "step": 9206 }, { "epoch": 0.49085674681452257, "grad_norm": 2.1001172065734863, "learning_rate": 8.1588e-06, "loss": 0.0288, "step": 9207 }, { "epoch": 0.4909100602441755, "grad_norm": 1.6051275730133057, "learning_rate": 8.1586e-06, "loss": 0.0357, "step": 9208 }, { "epoch": 0.4909633736738284, "grad_norm": 4.503185749053955, "learning_rate": 8.158400000000001e-06, "loss": 0.0507, "step": 9209 }, { "epoch": 0.49101668710348134, "grad_norm": 3.384561538696289, "learning_rate": 8.158200000000001e-06, "loss": 0.0581, "step": 9210 }, { "epoch": 0.4910700005331343, "grad_norm": 5.163660526275635, "learning_rate": 8.158e-06, "loss": 0.0352, "step": 9211 }, { "epoch": 0.49112331396278724, "grad_norm": 2.7542526721954346, "learning_rate": 8.1578e-06, "loss": 0.036, "step": 9212 }, { "epoch": 0.49117662739244017, "grad_norm": 3.1496191024780273, "learning_rate": 8.157600000000001e-06, "loss": 0.0582, "step": 9213 }, { "epoch": 0.4912299408220931, "grad_norm": 1.6154676675796509, "learning_rate": 8.157400000000001e-06, "loss": 0.0258, "step": 9214 }, { "epoch": 0.491283254251746, "grad_norm": 1.706459403038025, "learning_rate": 8.1572e-06, "loss": 0.0229, "step": 9215 }, { "epoch": 0.49133656768139894, "grad_norm": 2.3182780742645264, "learning_rate": 8.157e-06, "loss": 0.0353, "step": 9216 }, { "epoch": 0.49138988111105186, "grad_norm": 2.0862069129943848, "learning_rate": 8.1568e-06, "loss": 0.017, "step": 9217 }, { "epoch": 0.4914431945407048, "grad_norm": 1.6403454542160034, "learning_rate": 8.1566e-06, "loss": 0.0246, "step": 9218 }, { "epoch": 0.4914965079703577, "grad_norm": 6.3363037109375, "learning_rate": 8.1564e-06, "loss": 0.0793, "step": 9219 }, { "epoch": 0.4915498214000107, "grad_norm": 5.386277675628662, "learning_rate": 8.1562e-06, "loss": 0.0572, "step": 9220 }, { "epoch": 0.4916031348296636, "grad_norm": 4.357233047485352, "learning_rate": 8.156000000000001e-06, "loss": 0.072, "step": 9221 }, { "epoch": 0.49165644825931654, "grad_norm": 3.6975455284118652, "learning_rate": 8.1558e-06, "loss": 0.0465, "step": 9222 }, { "epoch": 0.49170976168896946, "grad_norm": 1.2884624004364014, "learning_rate": 8.1556e-06, "loss": 0.0184, "step": 9223 }, { "epoch": 0.4917630751186224, "grad_norm": 1.8221937417984009, "learning_rate": 8.155400000000002e-06, "loss": 0.0216, "step": 9224 }, { "epoch": 0.4918163885482753, "grad_norm": 3.9772861003875732, "learning_rate": 8.155200000000001e-06, "loss": 0.0375, "step": 9225 }, { "epoch": 0.49186970197792823, "grad_norm": 3.605069160461426, "learning_rate": 8.155e-06, "loss": 0.0571, "step": 9226 }, { "epoch": 0.49192301540758115, "grad_norm": 2.3793482780456543, "learning_rate": 8.1548e-06, "loss": 0.0491, "step": 9227 }, { "epoch": 0.4919763288372341, "grad_norm": 4.855785369873047, "learning_rate": 8.1546e-06, "loss": 0.0417, "step": 9228 }, { "epoch": 0.492029642266887, "grad_norm": 2.9984679222106934, "learning_rate": 8.154400000000001e-06, "loss": 0.0458, "step": 9229 }, { "epoch": 0.49208295569654, "grad_norm": 2.1343600749969482, "learning_rate": 8.1542e-06, "loss": 0.0353, "step": 9230 }, { "epoch": 0.4921362691261929, "grad_norm": 2.898223400115967, "learning_rate": 8.154e-06, "loss": 0.0273, "step": 9231 }, { "epoch": 0.49218958255584583, "grad_norm": 1.8459150791168213, "learning_rate": 8.1538e-06, "loss": 0.0305, "step": 9232 }, { "epoch": 0.49224289598549875, "grad_norm": 1.8837194442749023, "learning_rate": 8.153600000000001e-06, "loss": 0.0269, "step": 9233 }, { "epoch": 0.4922962094151517, "grad_norm": 1.0988086462020874, "learning_rate": 8.1534e-06, "loss": 0.0186, "step": 9234 }, { "epoch": 0.4923495228448046, "grad_norm": 4.542226314544678, "learning_rate": 8.153200000000002e-06, "loss": 0.0502, "step": 9235 }, { "epoch": 0.4924028362744575, "grad_norm": 1.866589903831482, "learning_rate": 8.153000000000001e-06, "loss": 0.0199, "step": 9236 }, { "epoch": 0.49245614970411045, "grad_norm": 3.4574339389801025, "learning_rate": 8.152800000000001e-06, "loss": 0.0744, "step": 9237 }, { "epoch": 0.49250946313376337, "grad_norm": 1.8880809545516968, "learning_rate": 8.1526e-06, "loss": 0.0458, "step": 9238 }, { "epoch": 0.49256277656341635, "grad_norm": 2.323669672012329, "learning_rate": 8.1524e-06, "loss": 0.0134, "step": 9239 }, { "epoch": 0.4926160899930693, "grad_norm": 3.4629461765289307, "learning_rate": 8.1522e-06, "loss": 0.044, "step": 9240 }, { "epoch": 0.4926694034227222, "grad_norm": 2.4843239784240723, "learning_rate": 8.152000000000001e-06, "loss": 0.038, "step": 9241 }, { "epoch": 0.4927227168523751, "grad_norm": 2.0119400024414062, "learning_rate": 8.1518e-06, "loss": 0.0457, "step": 9242 }, { "epoch": 0.49277603028202804, "grad_norm": 2.699878692626953, "learning_rate": 8.1516e-06, "loss": 0.0291, "step": 9243 }, { "epoch": 0.49282934371168097, "grad_norm": 3.1962995529174805, "learning_rate": 8.1514e-06, "loss": 0.0678, "step": 9244 }, { "epoch": 0.4928826571413339, "grad_norm": 2.1342976093292236, "learning_rate": 8.1512e-06, "loss": 0.0621, "step": 9245 }, { "epoch": 0.4929359705709868, "grad_norm": 2.586379051208496, "learning_rate": 8.151e-06, "loss": 0.0582, "step": 9246 }, { "epoch": 0.49298928400063974, "grad_norm": 2.3304660320281982, "learning_rate": 8.150800000000002e-06, "loss": 0.0239, "step": 9247 }, { "epoch": 0.4930425974302927, "grad_norm": 4.059202194213867, "learning_rate": 8.150600000000001e-06, "loss": 0.0432, "step": 9248 }, { "epoch": 0.49309591085994564, "grad_norm": 3.5575318336486816, "learning_rate": 8.1504e-06, "loss": 0.054, "step": 9249 }, { "epoch": 0.49314922428959856, "grad_norm": 2.411142110824585, "learning_rate": 8.1502e-06, "loss": 0.0235, "step": 9250 }, { "epoch": 0.4932025377192515, "grad_norm": 1.8609120845794678, "learning_rate": 8.15e-06, "loss": 0.0223, "step": 9251 }, { "epoch": 0.4932558511489044, "grad_norm": 2.2390244007110596, "learning_rate": 8.149800000000001e-06, "loss": 0.0445, "step": 9252 }, { "epoch": 0.49330916457855734, "grad_norm": 2.8110718727111816, "learning_rate": 8.1496e-06, "loss": 0.0615, "step": 9253 }, { "epoch": 0.49336247800821026, "grad_norm": 1.307706594467163, "learning_rate": 8.1494e-06, "loss": 0.0233, "step": 9254 }, { "epoch": 0.4934157914378632, "grad_norm": 1.4295156002044678, "learning_rate": 8.1492e-06, "loss": 0.014, "step": 9255 }, { "epoch": 0.4934691048675161, "grad_norm": 3.3200843334198, "learning_rate": 8.149e-06, "loss": 0.0528, "step": 9256 }, { "epoch": 0.49352241829716903, "grad_norm": 2.535951614379883, "learning_rate": 8.1488e-06, "loss": 0.0298, "step": 9257 }, { "epoch": 0.493575731726822, "grad_norm": 1.7638843059539795, "learning_rate": 8.1486e-06, "loss": 0.0282, "step": 9258 }, { "epoch": 0.49362904515647493, "grad_norm": 2.5570178031921387, "learning_rate": 8.148400000000001e-06, "loss": 0.0371, "step": 9259 }, { "epoch": 0.49368235858612786, "grad_norm": 1.5027049779891968, "learning_rate": 8.148200000000001e-06, "loss": 0.0406, "step": 9260 }, { "epoch": 0.4937356720157808, "grad_norm": 2.6531083583831787, "learning_rate": 8.148e-06, "loss": 0.0335, "step": 9261 }, { "epoch": 0.4937889854454337, "grad_norm": 2.101264476776123, "learning_rate": 8.1478e-06, "loss": 0.0317, "step": 9262 }, { "epoch": 0.4938422988750866, "grad_norm": 3.5684473514556885, "learning_rate": 8.147600000000001e-06, "loss": 0.0321, "step": 9263 }, { "epoch": 0.49389561230473955, "grad_norm": 2.5999248027801514, "learning_rate": 8.147400000000001e-06, "loss": 0.0366, "step": 9264 }, { "epoch": 0.4939489257343925, "grad_norm": 3.3841371536254883, "learning_rate": 8.1472e-06, "loss": 0.0497, "step": 9265 }, { "epoch": 0.4940022391640454, "grad_norm": 3.2406179904937744, "learning_rate": 8.147e-06, "loss": 0.0268, "step": 9266 }, { "epoch": 0.4940555525936984, "grad_norm": 3.73201322555542, "learning_rate": 8.1468e-06, "loss": 0.0737, "step": 9267 }, { "epoch": 0.4941088660233513, "grad_norm": 4.144035816192627, "learning_rate": 8.146600000000001e-06, "loss": 0.0479, "step": 9268 }, { "epoch": 0.4941621794530042, "grad_norm": 3.159642219543457, "learning_rate": 8.1464e-06, "loss": 0.0403, "step": 9269 }, { "epoch": 0.49421549288265715, "grad_norm": 1.900780439376831, "learning_rate": 8.1462e-06, "loss": 0.0348, "step": 9270 }, { "epoch": 0.49426880631231007, "grad_norm": 2.3540213108062744, "learning_rate": 8.146000000000001e-06, "loss": 0.0216, "step": 9271 }, { "epoch": 0.494322119741963, "grad_norm": 2.949204683303833, "learning_rate": 8.145800000000001e-06, "loss": 0.0277, "step": 9272 }, { "epoch": 0.4943754331716159, "grad_norm": 3.219517946243286, "learning_rate": 8.1456e-06, "loss": 0.0289, "step": 9273 }, { "epoch": 0.49442874660126884, "grad_norm": 1.138891339302063, "learning_rate": 8.145400000000002e-06, "loss": 0.0247, "step": 9274 }, { "epoch": 0.49448206003092177, "grad_norm": 2.2927663326263428, "learning_rate": 8.145200000000001e-06, "loss": 0.0304, "step": 9275 }, { "epoch": 0.49453537346057475, "grad_norm": 2.99459171295166, "learning_rate": 8.145e-06, "loss": 0.041, "step": 9276 }, { "epoch": 0.49458868689022767, "grad_norm": 2.0269007682800293, "learning_rate": 8.1448e-06, "loss": 0.035, "step": 9277 }, { "epoch": 0.4946420003198806, "grad_norm": 1.7433892488479614, "learning_rate": 8.1446e-06, "loss": 0.0316, "step": 9278 }, { "epoch": 0.4946953137495335, "grad_norm": 2.52597975730896, "learning_rate": 8.144400000000001e-06, "loss": 0.0491, "step": 9279 }, { "epoch": 0.49474862717918644, "grad_norm": 2.0526905059814453, "learning_rate": 8.1442e-06, "loss": 0.022, "step": 9280 }, { "epoch": 0.49480194060883936, "grad_norm": 3.705134153366089, "learning_rate": 8.144e-06, "loss": 0.0549, "step": 9281 }, { "epoch": 0.4948552540384923, "grad_norm": 3.5533995628356934, "learning_rate": 8.1438e-06, "loss": 0.0427, "step": 9282 }, { "epoch": 0.4949085674681452, "grad_norm": 2.767359495162964, "learning_rate": 8.143600000000001e-06, "loss": 0.043, "step": 9283 }, { "epoch": 0.49496188089779813, "grad_norm": 3.4604251384735107, "learning_rate": 8.1434e-06, "loss": 0.059, "step": 9284 }, { "epoch": 0.49501519432745106, "grad_norm": 3.359055519104004, "learning_rate": 8.143200000000002e-06, "loss": 0.0401, "step": 9285 }, { "epoch": 0.49506850775710404, "grad_norm": 2.591710329055786, "learning_rate": 8.143000000000001e-06, "loss": 0.048, "step": 9286 }, { "epoch": 0.49512182118675696, "grad_norm": 2.558816909790039, "learning_rate": 8.142800000000001e-06, "loss": 0.0366, "step": 9287 }, { "epoch": 0.4951751346164099, "grad_norm": 4.272306442260742, "learning_rate": 8.1426e-06, "loss": 0.0499, "step": 9288 }, { "epoch": 0.4952284480460628, "grad_norm": 4.718417167663574, "learning_rate": 8.1424e-06, "loss": 0.0837, "step": 9289 }, { "epoch": 0.49528176147571573, "grad_norm": 1.9450064897537231, "learning_rate": 8.142200000000001e-06, "loss": 0.0469, "step": 9290 }, { "epoch": 0.49533507490536866, "grad_norm": 3.946582078933716, "learning_rate": 8.142000000000001e-06, "loss": 0.048, "step": 9291 }, { "epoch": 0.4953883883350216, "grad_norm": 2.402855157852173, "learning_rate": 8.1418e-06, "loss": 0.0463, "step": 9292 }, { "epoch": 0.4954417017646745, "grad_norm": 1.6042917966842651, "learning_rate": 8.1416e-06, "loss": 0.0409, "step": 9293 }, { "epoch": 0.4954950151943274, "grad_norm": 1.9142812490463257, "learning_rate": 8.1414e-06, "loss": 0.0281, "step": 9294 }, { "epoch": 0.4955483286239804, "grad_norm": 1.486526370048523, "learning_rate": 8.141200000000001e-06, "loss": 0.0291, "step": 9295 }, { "epoch": 0.49560164205363333, "grad_norm": 2.0165505409240723, "learning_rate": 8.141e-06, "loss": 0.0263, "step": 9296 }, { "epoch": 0.49565495548328625, "grad_norm": 2.142824411392212, "learning_rate": 8.140800000000002e-06, "loss": 0.0245, "step": 9297 }, { "epoch": 0.4957082689129392, "grad_norm": 1.4552961587905884, "learning_rate": 8.140600000000001e-06, "loss": 0.0298, "step": 9298 }, { "epoch": 0.4957615823425921, "grad_norm": 1.1387397050857544, "learning_rate": 8.140400000000001e-06, "loss": 0.0244, "step": 9299 }, { "epoch": 0.495814895772245, "grad_norm": 1.1771502494812012, "learning_rate": 8.1402e-06, "loss": 0.0177, "step": 9300 }, { "epoch": 0.49586820920189795, "grad_norm": 1.8261345624923706, "learning_rate": 8.14e-06, "loss": 0.0385, "step": 9301 }, { "epoch": 0.49592152263155087, "grad_norm": 2.4689881801605225, "learning_rate": 8.139800000000001e-06, "loss": 0.0308, "step": 9302 }, { "epoch": 0.4959748360612038, "grad_norm": 3.762394428253174, "learning_rate": 8.1396e-06, "loss": 0.0782, "step": 9303 }, { "epoch": 0.4960281494908568, "grad_norm": 1.6558971405029297, "learning_rate": 8.1394e-06, "loss": 0.0393, "step": 9304 }, { "epoch": 0.4960814629205097, "grad_norm": 1.9943994283676147, "learning_rate": 8.1392e-06, "loss": 0.0376, "step": 9305 }, { "epoch": 0.4961347763501626, "grad_norm": 1.9194482564926147, "learning_rate": 8.139e-06, "loss": 0.0335, "step": 9306 }, { "epoch": 0.49618808977981554, "grad_norm": 2.7708370685577393, "learning_rate": 8.1388e-06, "loss": 0.0403, "step": 9307 }, { "epoch": 0.49624140320946847, "grad_norm": 2.6426122188568115, "learning_rate": 8.1386e-06, "loss": 0.0449, "step": 9308 }, { "epoch": 0.4962947166391214, "grad_norm": 4.790609836578369, "learning_rate": 8.138400000000002e-06, "loss": 0.0593, "step": 9309 }, { "epoch": 0.4963480300687743, "grad_norm": 2.78782057762146, "learning_rate": 8.138200000000001e-06, "loss": 0.0246, "step": 9310 }, { "epoch": 0.49640134349842724, "grad_norm": 2.5514581203460693, "learning_rate": 8.138e-06, "loss": 0.0334, "step": 9311 }, { "epoch": 0.49645465692808016, "grad_norm": 1.2738220691680908, "learning_rate": 8.1378e-06, "loss": 0.0173, "step": 9312 }, { "epoch": 0.4965079703577331, "grad_norm": 1.1680971384048462, "learning_rate": 8.137600000000002e-06, "loss": 0.031, "step": 9313 }, { "epoch": 0.49656128378738607, "grad_norm": 2.496971607208252, "learning_rate": 8.137400000000001e-06, "loss": 0.0307, "step": 9314 }, { "epoch": 0.496614597217039, "grad_norm": 2.7343950271606445, "learning_rate": 8.1372e-06, "loss": 0.0687, "step": 9315 }, { "epoch": 0.4966679106466919, "grad_norm": 2.377502918243408, "learning_rate": 8.137e-06, "loss": 0.055, "step": 9316 }, { "epoch": 0.49672122407634484, "grad_norm": 1.3500187397003174, "learning_rate": 8.1368e-06, "loss": 0.015, "step": 9317 }, { "epoch": 0.49677453750599776, "grad_norm": 3.9692978858947754, "learning_rate": 8.136600000000001e-06, "loss": 0.0725, "step": 9318 }, { "epoch": 0.4968278509356507, "grad_norm": 2.6126859188079834, "learning_rate": 8.1364e-06, "loss": 0.0386, "step": 9319 }, { "epoch": 0.4968811643653036, "grad_norm": 4.559717655181885, "learning_rate": 8.1362e-06, "loss": 0.0449, "step": 9320 }, { "epoch": 0.49693447779495653, "grad_norm": 2.1071839332580566, "learning_rate": 8.136000000000001e-06, "loss": 0.0282, "step": 9321 }, { "epoch": 0.49698779122460945, "grad_norm": 2.006138324737549, "learning_rate": 8.135800000000001e-06, "loss": 0.0229, "step": 9322 }, { "epoch": 0.49704110465426243, "grad_norm": 2.7809832096099854, "learning_rate": 8.1356e-06, "loss": 0.045, "step": 9323 }, { "epoch": 0.49709441808391536, "grad_norm": 1.8108489513397217, "learning_rate": 8.135400000000002e-06, "loss": 0.0237, "step": 9324 }, { "epoch": 0.4971477315135683, "grad_norm": 4.183950901031494, "learning_rate": 8.135200000000001e-06, "loss": 0.063, "step": 9325 }, { "epoch": 0.4972010449432212, "grad_norm": 1.8666068315505981, "learning_rate": 8.135000000000001e-06, "loss": 0.028, "step": 9326 }, { "epoch": 0.49725435837287413, "grad_norm": 3.535576105117798, "learning_rate": 8.1348e-06, "loss": 0.0524, "step": 9327 }, { "epoch": 0.49730767180252705, "grad_norm": 2.1436095237731934, "learning_rate": 8.1346e-06, "loss": 0.0196, "step": 9328 }, { "epoch": 0.49736098523218, "grad_norm": 3.1472413539886475, "learning_rate": 8.134400000000001e-06, "loss": 0.0342, "step": 9329 }, { "epoch": 0.4974142986618329, "grad_norm": 2.532640218734741, "learning_rate": 8.1342e-06, "loss": 0.0112, "step": 9330 }, { "epoch": 0.4974676120914858, "grad_norm": 2.0718190670013428, "learning_rate": 8.134e-06, "loss": 0.0324, "step": 9331 }, { "epoch": 0.4975209255211388, "grad_norm": 2.028564214706421, "learning_rate": 8.1338e-06, "loss": 0.0289, "step": 9332 }, { "epoch": 0.4975742389507917, "grad_norm": 2.7852158546447754, "learning_rate": 8.133600000000001e-06, "loss": 0.0364, "step": 9333 }, { "epoch": 0.49762755238044465, "grad_norm": 2.9387059211730957, "learning_rate": 8.1334e-06, "loss": 0.0482, "step": 9334 }, { "epoch": 0.4976808658100976, "grad_norm": 2.563556432723999, "learning_rate": 8.1332e-06, "loss": 0.0376, "step": 9335 }, { "epoch": 0.4977341792397505, "grad_norm": 3.4548428058624268, "learning_rate": 8.133000000000002e-06, "loss": 0.0479, "step": 9336 }, { "epoch": 0.4977874926694034, "grad_norm": 1.0149303674697876, "learning_rate": 8.132800000000001e-06, "loss": 0.014, "step": 9337 }, { "epoch": 0.49784080609905634, "grad_norm": 3.1607635021209717, "learning_rate": 8.1326e-06, "loss": 0.0482, "step": 9338 }, { "epoch": 0.49789411952870927, "grad_norm": 2.8169169425964355, "learning_rate": 8.1324e-06, "loss": 0.0323, "step": 9339 }, { "epoch": 0.4979474329583622, "grad_norm": 3.0933640003204346, "learning_rate": 8.132200000000002e-06, "loss": 0.058, "step": 9340 }, { "epoch": 0.4980007463880151, "grad_norm": 4.519920349121094, "learning_rate": 8.132000000000001e-06, "loss": 0.0484, "step": 9341 }, { "epoch": 0.4980540598176681, "grad_norm": 1.8538883924484253, "learning_rate": 8.1318e-06, "loss": 0.0223, "step": 9342 }, { "epoch": 0.498107373247321, "grad_norm": 1.3962146043777466, "learning_rate": 8.1316e-06, "loss": 0.0349, "step": 9343 }, { "epoch": 0.49816068667697394, "grad_norm": 1.635050654411316, "learning_rate": 8.1314e-06, "loss": 0.044, "step": 9344 }, { "epoch": 0.49821400010662686, "grad_norm": 2.271226406097412, "learning_rate": 8.131200000000001e-06, "loss": 0.028, "step": 9345 }, { "epoch": 0.4982673135362798, "grad_norm": 3.0905542373657227, "learning_rate": 8.131e-06, "loss": 0.0427, "step": 9346 }, { "epoch": 0.4983206269659327, "grad_norm": 1.2197473049163818, "learning_rate": 8.1308e-06, "loss": 0.0163, "step": 9347 }, { "epoch": 0.49837394039558564, "grad_norm": 3.061509370803833, "learning_rate": 8.130600000000001e-06, "loss": 0.0618, "step": 9348 }, { "epoch": 0.49842725382523856, "grad_norm": 2.0433149337768555, "learning_rate": 8.130400000000001e-06, "loss": 0.0302, "step": 9349 }, { "epoch": 0.4984805672548915, "grad_norm": 1.4110033512115479, "learning_rate": 8.1302e-06, "loss": 0.0136, "step": 9350 }, { "epoch": 0.49853388068454446, "grad_norm": 5.605788230895996, "learning_rate": 8.13e-06, "loss": 0.0537, "step": 9351 }, { "epoch": 0.4985871941141974, "grad_norm": 3.9893767833709717, "learning_rate": 8.129800000000001e-06, "loss": 0.0398, "step": 9352 }, { "epoch": 0.4986405075438503, "grad_norm": 1.695548415184021, "learning_rate": 8.129600000000001e-06, "loss": 0.0244, "step": 9353 }, { "epoch": 0.49869382097350323, "grad_norm": 1.764426350593567, "learning_rate": 8.1294e-06, "loss": 0.0137, "step": 9354 }, { "epoch": 0.49874713440315616, "grad_norm": 7.769952774047852, "learning_rate": 8.1292e-06, "loss": 0.0634, "step": 9355 }, { "epoch": 0.4988004478328091, "grad_norm": 2.571671724319458, "learning_rate": 8.129e-06, "loss": 0.0404, "step": 9356 }, { "epoch": 0.498853761262462, "grad_norm": 5.546229362487793, "learning_rate": 8.1288e-06, "loss": 0.0667, "step": 9357 }, { "epoch": 0.4989070746921149, "grad_norm": 2.8451051712036133, "learning_rate": 8.1286e-06, "loss": 0.0659, "step": 9358 }, { "epoch": 0.49896038812176785, "grad_norm": 1.6582003831863403, "learning_rate": 8.1284e-06, "loss": 0.0243, "step": 9359 }, { "epoch": 0.49901370155142083, "grad_norm": 2.844344139099121, "learning_rate": 8.128200000000001e-06, "loss": 0.0605, "step": 9360 }, { "epoch": 0.49906701498107375, "grad_norm": 4.088986396789551, "learning_rate": 8.128e-06, "loss": 0.0538, "step": 9361 }, { "epoch": 0.4991203284107267, "grad_norm": 2.290743827819824, "learning_rate": 8.1278e-06, "loss": 0.0439, "step": 9362 }, { "epoch": 0.4991736418403796, "grad_norm": 2.6381654739379883, "learning_rate": 8.127600000000002e-06, "loss": 0.0406, "step": 9363 }, { "epoch": 0.4992269552700325, "grad_norm": 5.375274658203125, "learning_rate": 8.127400000000001e-06, "loss": 0.0847, "step": 9364 }, { "epoch": 0.49928026869968545, "grad_norm": 2.0841715335845947, "learning_rate": 8.1272e-06, "loss": 0.0335, "step": 9365 }, { "epoch": 0.49933358212933837, "grad_norm": 5.715031623840332, "learning_rate": 8.127e-06, "loss": 0.0358, "step": 9366 }, { "epoch": 0.4993868955589913, "grad_norm": 4.152835369110107, "learning_rate": 8.1268e-06, "loss": 0.0417, "step": 9367 }, { "epoch": 0.4994402089886442, "grad_norm": 2.4548628330230713, "learning_rate": 8.126600000000001e-06, "loss": 0.0396, "step": 9368 }, { "epoch": 0.49949352241829714, "grad_norm": 2.5164031982421875, "learning_rate": 8.1264e-06, "loss": 0.03, "step": 9369 }, { "epoch": 0.4995468358479501, "grad_norm": 3.535346269607544, "learning_rate": 8.1262e-06, "loss": 0.0617, "step": 9370 }, { "epoch": 0.49960014927760305, "grad_norm": 1.286658525466919, "learning_rate": 8.126e-06, "loss": 0.0194, "step": 9371 }, { "epoch": 0.49965346270725597, "grad_norm": 1.5322575569152832, "learning_rate": 8.125800000000001e-06, "loss": 0.0357, "step": 9372 }, { "epoch": 0.4997067761369089, "grad_norm": 1.1313039064407349, "learning_rate": 8.1256e-06, "loss": 0.0123, "step": 9373 }, { "epoch": 0.4997600895665618, "grad_norm": 2.1503779888153076, "learning_rate": 8.125400000000002e-06, "loss": 0.0252, "step": 9374 }, { "epoch": 0.49981340299621474, "grad_norm": 1.3370622396469116, "learning_rate": 8.125200000000001e-06, "loss": 0.0251, "step": 9375 }, { "epoch": 0.49986671642586766, "grad_norm": 1.9543977975845337, "learning_rate": 8.125000000000001e-06, "loss": 0.0418, "step": 9376 }, { "epoch": 0.4999200298555206, "grad_norm": 2.147817611694336, "learning_rate": 8.1248e-06, "loss": 0.026, "step": 9377 }, { "epoch": 0.4999733432851735, "grad_norm": 2.9130074977874756, "learning_rate": 8.1246e-06, "loss": 0.0584, "step": 9378 }, { "epoch": 0.5000266567148265, "grad_norm": 3.066216230392456, "learning_rate": 8.124400000000001e-06, "loss": 0.0413, "step": 9379 }, { "epoch": 0.5000799701444794, "grad_norm": 1.8451322317123413, "learning_rate": 8.124200000000001e-06, "loss": 0.0303, "step": 9380 }, { "epoch": 0.5001332835741323, "grad_norm": 3.161881446838379, "learning_rate": 8.124e-06, "loss": 0.0617, "step": 9381 }, { "epoch": 0.5001865970037852, "grad_norm": 2.0474843978881836, "learning_rate": 8.1238e-06, "loss": 0.0313, "step": 9382 }, { "epoch": 0.5002399104334382, "grad_norm": 2.0109071731567383, "learning_rate": 8.1236e-06, "loss": 0.0536, "step": 9383 }, { "epoch": 0.5002932238630912, "grad_norm": 2.3747055530548096, "learning_rate": 8.123400000000001e-06, "loss": 0.0178, "step": 9384 }, { "epoch": 0.500346537292744, "grad_norm": 1.6198039054870605, "learning_rate": 8.1232e-06, "loss": 0.0187, "step": 9385 }, { "epoch": 0.500399850722397, "grad_norm": 1.5158051252365112, "learning_rate": 8.123000000000002e-06, "loss": 0.0182, "step": 9386 }, { "epoch": 0.5004531641520499, "grad_norm": 2.093146324157715, "learning_rate": 8.122800000000001e-06, "loss": 0.0448, "step": 9387 }, { "epoch": 0.5005064775817029, "grad_norm": 5.8183746337890625, "learning_rate": 8.1226e-06, "loss": 0.0658, "step": 9388 }, { "epoch": 0.5005597910113557, "grad_norm": 3.180481195449829, "learning_rate": 8.1224e-06, "loss": 0.0316, "step": 9389 }, { "epoch": 0.5006131044410087, "grad_norm": 3.3024439811706543, "learning_rate": 8.122200000000002e-06, "loss": 0.0329, "step": 9390 }, { "epoch": 0.5006664178706616, "grad_norm": 2.6937568187713623, "learning_rate": 8.122000000000001e-06, "loss": 0.0232, "step": 9391 }, { "epoch": 0.5007197313003146, "grad_norm": 2.321883201599121, "learning_rate": 8.1218e-06, "loss": 0.0408, "step": 9392 }, { "epoch": 0.5007730447299675, "grad_norm": 3.2233476638793945, "learning_rate": 8.1216e-06, "loss": 0.064, "step": 9393 }, { "epoch": 0.5008263581596204, "grad_norm": 2.7941770553588867, "learning_rate": 8.1214e-06, "loss": 0.0615, "step": 9394 }, { "epoch": 0.5008796715892734, "grad_norm": 3.778311252593994, "learning_rate": 8.1212e-06, "loss": 0.0541, "step": 9395 }, { "epoch": 0.5009329850189262, "grad_norm": 4.7030134201049805, "learning_rate": 8.121e-06, "loss": 0.0635, "step": 9396 }, { "epoch": 0.5009862984485792, "grad_norm": 2.279940128326416, "learning_rate": 8.1208e-06, "loss": 0.042, "step": 9397 }, { "epoch": 0.5010396118782321, "grad_norm": 1.2921562194824219, "learning_rate": 8.120600000000001e-06, "loss": 0.0223, "step": 9398 }, { "epoch": 0.5010929253078851, "grad_norm": 3.178799629211426, "learning_rate": 8.120400000000001e-06, "loss": 0.0546, "step": 9399 }, { "epoch": 0.5011462387375379, "grad_norm": 2.9631993770599365, "learning_rate": 8.1202e-06, "loss": 0.026, "step": 9400 }, { "epoch": 0.5011995521671909, "grad_norm": 4.409402847290039, "learning_rate": 8.120000000000002e-06, "loss": 0.0372, "step": 9401 }, { "epoch": 0.5012528655968439, "grad_norm": 3.262192726135254, "learning_rate": 8.119800000000001e-06, "loss": 0.041, "step": 9402 }, { "epoch": 0.5013061790264968, "grad_norm": 2.811225175857544, "learning_rate": 8.119600000000001e-06, "loss": 0.0196, "step": 9403 }, { "epoch": 0.5013594924561497, "grad_norm": 2.197566270828247, "learning_rate": 8.1194e-06, "loss": 0.0345, "step": 9404 }, { "epoch": 0.5014128058858026, "grad_norm": 2.7892909049987793, "learning_rate": 8.1192e-06, "loss": 0.0352, "step": 9405 }, { "epoch": 0.5014661193154556, "grad_norm": 2.203596353530884, "learning_rate": 8.119e-06, "loss": 0.0187, "step": 9406 }, { "epoch": 0.5015194327451085, "grad_norm": 1.5195815563201904, "learning_rate": 8.118800000000001e-06, "loss": 0.0106, "step": 9407 }, { "epoch": 0.5015727461747614, "grad_norm": 4.857480049133301, "learning_rate": 8.1186e-06, "loss": 0.078, "step": 9408 }, { "epoch": 0.5016260596044143, "grad_norm": 1.1256097555160522, "learning_rate": 8.1184e-06, "loss": 0.0092, "step": 9409 }, { "epoch": 0.5016793730340673, "grad_norm": 2.6750288009643555, "learning_rate": 8.118200000000001e-06, "loss": 0.0669, "step": 9410 }, { "epoch": 0.5017326864637202, "grad_norm": 2.6134538650512695, "learning_rate": 8.118000000000001e-06, "loss": 0.0407, "step": 9411 }, { "epoch": 0.5017859998933731, "grad_norm": 2.4572126865386963, "learning_rate": 8.1178e-06, "loss": 0.0216, "step": 9412 }, { "epoch": 0.5018393133230261, "grad_norm": 2.741259813308716, "learning_rate": 8.117600000000002e-06, "loss": 0.0595, "step": 9413 }, { "epoch": 0.501892626752679, "grad_norm": 4.611198902130127, "learning_rate": 8.117400000000001e-06, "loss": 0.0423, "step": 9414 }, { "epoch": 0.501945940182332, "grad_norm": 2.3875510692596436, "learning_rate": 8.1172e-06, "loss": 0.0605, "step": 9415 }, { "epoch": 0.5019992536119848, "grad_norm": 3.2522387504577637, "learning_rate": 8.117e-06, "loss": 0.0477, "step": 9416 }, { "epoch": 0.5020525670416378, "grad_norm": 4.027730464935303, "learning_rate": 8.1168e-06, "loss": 0.0498, "step": 9417 }, { "epoch": 0.5021058804712907, "grad_norm": 2.0489537715911865, "learning_rate": 8.116600000000001e-06, "loss": 0.0208, "step": 9418 }, { "epoch": 0.5021591939009437, "grad_norm": 2.2490854263305664, "learning_rate": 8.1164e-06, "loss": 0.0258, "step": 9419 }, { "epoch": 0.5022125073305965, "grad_norm": 2.471677303314209, "learning_rate": 8.1162e-06, "loss": 0.0316, "step": 9420 }, { "epoch": 0.5022658207602495, "grad_norm": 5.077786445617676, "learning_rate": 8.116e-06, "loss": 0.0401, "step": 9421 }, { "epoch": 0.5023191341899025, "grad_norm": 2.044679641723633, "learning_rate": 8.115800000000001e-06, "loss": 0.0424, "step": 9422 }, { "epoch": 0.5023724476195554, "grad_norm": 2.083052635192871, "learning_rate": 8.1156e-06, "loss": 0.039, "step": 9423 }, { "epoch": 0.5024257610492083, "grad_norm": 1.4772287607192993, "learning_rate": 8.1154e-06, "loss": 0.0126, "step": 9424 }, { "epoch": 0.5024790744788612, "grad_norm": 2.7046754360198975, "learning_rate": 8.115200000000002e-06, "loss": 0.0212, "step": 9425 }, { "epoch": 0.5025323879085142, "grad_norm": 2.8000640869140625, "learning_rate": 8.115000000000001e-06, "loss": 0.0384, "step": 9426 }, { "epoch": 0.502585701338167, "grad_norm": 4.589207172393799, "learning_rate": 8.1148e-06, "loss": 0.0734, "step": 9427 }, { "epoch": 0.50263901476782, "grad_norm": 2.4723942279815674, "learning_rate": 8.1146e-06, "loss": 0.0147, "step": 9428 }, { "epoch": 0.5026923281974729, "grad_norm": 1.1219708919525146, "learning_rate": 8.114400000000001e-06, "loss": 0.0172, "step": 9429 }, { "epoch": 0.5027456416271259, "grad_norm": 2.907592535018921, "learning_rate": 8.114200000000001e-06, "loss": 0.0591, "step": 9430 }, { "epoch": 0.5027989550567789, "grad_norm": 4.634849548339844, "learning_rate": 8.114e-06, "loss": 0.068, "step": 9431 }, { "epoch": 0.5028522684864317, "grad_norm": 2.9041600227355957, "learning_rate": 8.1138e-06, "loss": 0.0802, "step": 9432 }, { "epoch": 0.5029055819160847, "grad_norm": 3.017468214035034, "learning_rate": 8.1136e-06, "loss": 0.0511, "step": 9433 }, { "epoch": 0.5029588953457376, "grad_norm": 2.9923148155212402, "learning_rate": 8.113400000000001e-06, "loss": 0.0307, "step": 9434 }, { "epoch": 0.5030122087753905, "grad_norm": 2.300063133239746, "learning_rate": 8.1132e-06, "loss": 0.046, "step": 9435 }, { "epoch": 0.5030655222050434, "grad_norm": 1.710443139076233, "learning_rate": 8.113e-06, "loss": 0.0241, "step": 9436 }, { "epoch": 0.5031188356346964, "grad_norm": 2.586737871170044, "learning_rate": 8.112800000000001e-06, "loss": 0.0422, "step": 9437 }, { "epoch": 0.5031721490643493, "grad_norm": 2.539879322052002, "learning_rate": 8.112600000000001e-06, "loss": 0.0187, "step": 9438 }, { "epoch": 0.5032254624940022, "grad_norm": 3.3884241580963135, "learning_rate": 8.1124e-06, "loss": 0.0358, "step": 9439 }, { "epoch": 0.5032787759236552, "grad_norm": 2.115978240966797, "learning_rate": 8.112200000000002e-06, "loss": 0.0299, "step": 9440 }, { "epoch": 0.5033320893533081, "grad_norm": 2.76743221282959, "learning_rate": 8.112000000000001e-06, "loss": 0.0289, "step": 9441 }, { "epoch": 0.5033854027829611, "grad_norm": 2.4296932220458984, "learning_rate": 8.1118e-06, "loss": 0.0437, "step": 9442 }, { "epoch": 0.5034387162126139, "grad_norm": 2.808661699295044, "learning_rate": 8.1116e-06, "loss": 0.0477, "step": 9443 }, { "epoch": 0.5034920296422669, "grad_norm": 2.4876022338867188, "learning_rate": 8.1114e-06, "loss": 0.0547, "step": 9444 }, { "epoch": 0.5035453430719198, "grad_norm": 3.780733823776245, "learning_rate": 8.1112e-06, "loss": 0.0331, "step": 9445 }, { "epoch": 0.5035986565015728, "grad_norm": 2.009216785430908, "learning_rate": 8.111e-06, "loss": 0.0267, "step": 9446 }, { "epoch": 0.5036519699312256, "grad_norm": 3.028240919113159, "learning_rate": 8.1108e-06, "loss": 0.0483, "step": 9447 }, { "epoch": 0.5037052833608786, "grad_norm": 3.18331241607666, "learning_rate": 8.1106e-06, "loss": 0.0427, "step": 9448 }, { "epoch": 0.5037585967905316, "grad_norm": 2.785759687423706, "learning_rate": 8.110400000000001e-06, "loss": 0.0409, "step": 9449 }, { "epoch": 0.5038119102201845, "grad_norm": 1.815380573272705, "learning_rate": 8.1102e-06, "loss": 0.0361, "step": 9450 }, { "epoch": 0.5038652236498374, "grad_norm": 2.661792516708374, "learning_rate": 8.110000000000002e-06, "loss": 0.0382, "step": 9451 }, { "epoch": 0.5039185370794903, "grad_norm": 2.495163679122925, "learning_rate": 8.109800000000002e-06, "loss": 0.0252, "step": 9452 }, { "epoch": 0.5039718505091433, "grad_norm": 4.4478840827941895, "learning_rate": 8.109600000000001e-06, "loss": 0.0678, "step": 9453 }, { "epoch": 0.5040251639387962, "grad_norm": 1.1561752557754517, "learning_rate": 8.1094e-06, "loss": 0.0105, "step": 9454 }, { "epoch": 0.5040784773684491, "grad_norm": 2.5480129718780518, "learning_rate": 8.1092e-06, "loss": 0.0361, "step": 9455 }, { "epoch": 0.504131790798102, "grad_norm": 2.5499558448791504, "learning_rate": 8.109e-06, "loss": 0.0205, "step": 9456 }, { "epoch": 0.504185104227755, "grad_norm": 1.0821533203125, "learning_rate": 8.108800000000001e-06, "loss": 0.0114, "step": 9457 }, { "epoch": 0.504238417657408, "grad_norm": 2.1757864952087402, "learning_rate": 8.1086e-06, "loss": 0.0377, "step": 9458 }, { "epoch": 0.5042917310870608, "grad_norm": 2.730238199234009, "learning_rate": 8.1084e-06, "loss": 0.0363, "step": 9459 }, { "epoch": 0.5043450445167138, "grad_norm": 3.9063055515289307, "learning_rate": 8.1082e-06, "loss": 0.0551, "step": 9460 }, { "epoch": 0.5043983579463667, "grad_norm": 2.2198925018310547, "learning_rate": 8.108000000000001e-06, "loss": 0.0192, "step": 9461 }, { "epoch": 0.5044516713760197, "grad_norm": 2.9229624271392822, "learning_rate": 8.1078e-06, "loss": 0.072, "step": 9462 }, { "epoch": 0.5045049848056725, "grad_norm": 1.7458829879760742, "learning_rate": 8.107600000000002e-06, "loss": 0.0379, "step": 9463 }, { "epoch": 0.5045582982353255, "grad_norm": 3.5248117446899414, "learning_rate": 8.107400000000001e-06, "loss": 0.0419, "step": 9464 }, { "epoch": 0.5046116116649784, "grad_norm": 2.5097334384918213, "learning_rate": 8.107200000000001e-06, "loss": 0.0482, "step": 9465 }, { "epoch": 0.5046649250946313, "grad_norm": 2.082502841949463, "learning_rate": 8.107e-06, "loss": 0.0536, "step": 9466 }, { "epoch": 0.5047182385242842, "grad_norm": 5.615779876708984, "learning_rate": 8.1068e-06, "loss": 0.0648, "step": 9467 }, { "epoch": 0.5047715519539372, "grad_norm": 1.453136920928955, "learning_rate": 8.106600000000001e-06, "loss": 0.0191, "step": 9468 }, { "epoch": 0.5048248653835902, "grad_norm": 2.370075225830078, "learning_rate": 8.106400000000001e-06, "loss": 0.0512, "step": 9469 }, { "epoch": 0.504878178813243, "grad_norm": 1.6314127445220947, "learning_rate": 8.1062e-06, "loss": 0.0181, "step": 9470 }, { "epoch": 0.504931492242896, "grad_norm": 1.7298099994659424, "learning_rate": 8.106e-06, "loss": 0.0196, "step": 9471 }, { "epoch": 0.5049848056725489, "grad_norm": 2.6494455337524414, "learning_rate": 8.1058e-06, "loss": 0.0248, "step": 9472 }, { "epoch": 0.5050381191022019, "grad_norm": 2.5705759525299072, "learning_rate": 8.1056e-06, "loss": 0.0228, "step": 9473 }, { "epoch": 0.5050914325318547, "grad_norm": 2.2490499019622803, "learning_rate": 8.1054e-06, "loss": 0.032, "step": 9474 }, { "epoch": 0.5051447459615077, "grad_norm": 3.028909683227539, "learning_rate": 8.105200000000002e-06, "loss": 0.028, "step": 9475 }, { "epoch": 0.5051980593911606, "grad_norm": 2.3373305797576904, "learning_rate": 8.105000000000001e-06, "loss": 0.0366, "step": 9476 }, { "epoch": 0.5052513728208136, "grad_norm": 2.9118499755859375, "learning_rate": 8.1048e-06, "loss": 0.0854, "step": 9477 }, { "epoch": 0.5053046862504665, "grad_norm": 1.071202039718628, "learning_rate": 8.1046e-06, "loss": 0.01, "step": 9478 }, { "epoch": 0.5053579996801194, "grad_norm": 1.7323691844940186, "learning_rate": 8.104400000000002e-06, "loss": 0.0255, "step": 9479 }, { "epoch": 0.5054113131097724, "grad_norm": 3.3844621181488037, "learning_rate": 8.104200000000001e-06, "loss": 0.0666, "step": 9480 }, { "epoch": 0.5054646265394253, "grad_norm": 4.083924770355225, "learning_rate": 8.104e-06, "loss": 0.0457, "step": 9481 }, { "epoch": 0.5055179399690782, "grad_norm": 4.098482131958008, "learning_rate": 8.1038e-06, "loss": 0.0534, "step": 9482 }, { "epoch": 0.5055712533987311, "grad_norm": 2.247044563293457, "learning_rate": 8.1036e-06, "loss": 0.0277, "step": 9483 }, { "epoch": 0.5056245668283841, "grad_norm": 5.039497375488281, "learning_rate": 8.1034e-06, "loss": 0.0326, "step": 9484 }, { "epoch": 0.505677880258037, "grad_norm": 1.6987303495407104, "learning_rate": 8.1032e-06, "loss": 0.0183, "step": 9485 }, { "epoch": 0.5057311936876899, "grad_norm": 2.4930858612060547, "learning_rate": 8.103e-06, "loss": 0.0298, "step": 9486 }, { "epoch": 0.5057845071173429, "grad_norm": 1.7524175643920898, "learning_rate": 8.102800000000001e-06, "loss": 0.0322, "step": 9487 }, { "epoch": 0.5058378205469958, "grad_norm": 3.162571907043457, "learning_rate": 8.102600000000001e-06, "loss": 0.031, "step": 9488 }, { "epoch": 0.5058911339766488, "grad_norm": 3.3049283027648926, "learning_rate": 8.1024e-06, "loss": 0.0306, "step": 9489 }, { "epoch": 0.5059444474063016, "grad_norm": 1.6521564722061157, "learning_rate": 8.102200000000002e-06, "loss": 0.0158, "step": 9490 }, { "epoch": 0.5059977608359546, "grad_norm": 2.34621262550354, "learning_rate": 8.102000000000001e-06, "loss": 0.0325, "step": 9491 }, { "epoch": 0.5060510742656075, "grad_norm": 2.124246120452881, "learning_rate": 8.101800000000001e-06, "loss": 0.016, "step": 9492 }, { "epoch": 0.5061043876952604, "grad_norm": 3.8621013164520264, "learning_rate": 8.1016e-06, "loss": 0.0267, "step": 9493 }, { "epoch": 0.5061577011249133, "grad_norm": 1.844099521636963, "learning_rate": 8.1014e-06, "loss": 0.0116, "step": 9494 }, { "epoch": 0.5062110145545663, "grad_norm": 1.9495627880096436, "learning_rate": 8.1012e-06, "loss": 0.0266, "step": 9495 }, { "epoch": 0.5062643279842193, "grad_norm": 3.805077314376831, "learning_rate": 8.101000000000001e-06, "loss": 0.0258, "step": 9496 }, { "epoch": 0.5063176414138721, "grad_norm": 2.900299310684204, "learning_rate": 8.1008e-06, "loss": 0.0213, "step": 9497 }, { "epoch": 0.5063709548435251, "grad_norm": 3.0738184452056885, "learning_rate": 8.1006e-06, "loss": 0.0448, "step": 9498 }, { "epoch": 0.506424268273178, "grad_norm": 2.9767613410949707, "learning_rate": 8.100400000000001e-06, "loss": 0.0338, "step": 9499 }, { "epoch": 0.506477581702831, "grad_norm": 4.182558059692383, "learning_rate": 8.1002e-06, "loss": 0.0346, "step": 9500 }, { "epoch": 0.5065308951324838, "grad_norm": 3.098646402359009, "learning_rate": 8.1e-06, "loss": 0.0524, "step": 9501 }, { "epoch": 0.5065842085621368, "grad_norm": 2.178009271621704, "learning_rate": 8.099800000000002e-06, "loss": 0.0129, "step": 9502 }, { "epoch": 0.5066375219917897, "grad_norm": 3.672302722930908, "learning_rate": 8.099600000000001e-06, "loss": 0.0406, "step": 9503 }, { "epoch": 0.5066908354214427, "grad_norm": 2.8239636421203613, "learning_rate": 8.0994e-06, "loss": 0.0454, "step": 9504 }, { "epoch": 0.5067441488510956, "grad_norm": 2.856341600418091, "learning_rate": 8.0992e-06, "loss": 0.0362, "step": 9505 }, { "epoch": 0.5067974622807485, "grad_norm": 3.1528260707855225, "learning_rate": 8.099e-06, "loss": 0.0131, "step": 9506 }, { "epoch": 0.5068507757104015, "grad_norm": 2.0829358100891113, "learning_rate": 8.098800000000001e-06, "loss": 0.0355, "step": 9507 }, { "epoch": 0.5069040891400544, "grad_norm": 4.466075420379639, "learning_rate": 8.0986e-06, "loss": 0.0489, "step": 9508 }, { "epoch": 0.5069574025697073, "grad_norm": 3.3594937324523926, "learning_rate": 8.0984e-06, "loss": 0.0507, "step": 9509 }, { "epoch": 0.5070107159993602, "grad_norm": 1.2681852579116821, "learning_rate": 8.0982e-06, "loss": 0.0171, "step": 9510 }, { "epoch": 0.5070640294290132, "grad_norm": 4.110269546508789, "learning_rate": 8.098000000000001e-06, "loss": 0.0756, "step": 9511 }, { "epoch": 0.507117342858666, "grad_norm": 2.619126081466675, "learning_rate": 8.0978e-06, "loss": 0.0454, "step": 9512 }, { "epoch": 0.507170656288319, "grad_norm": 2.65348219871521, "learning_rate": 8.0976e-06, "loss": 0.0333, "step": 9513 }, { "epoch": 0.507223969717972, "grad_norm": 1.7999945878982544, "learning_rate": 8.097400000000001e-06, "loss": 0.0203, "step": 9514 }, { "epoch": 0.5072772831476249, "grad_norm": 2.90818190574646, "learning_rate": 8.097200000000001e-06, "loss": 0.0412, "step": 9515 }, { "epoch": 0.5073305965772779, "grad_norm": 2.9242336750030518, "learning_rate": 8.097e-06, "loss": 0.0802, "step": 9516 }, { "epoch": 0.5073839100069307, "grad_norm": 1.457483172416687, "learning_rate": 8.0968e-06, "loss": 0.0141, "step": 9517 }, { "epoch": 0.5074372234365837, "grad_norm": 2.119537115097046, "learning_rate": 8.096600000000001e-06, "loss": 0.0251, "step": 9518 }, { "epoch": 0.5074905368662366, "grad_norm": 3.4082822799682617, "learning_rate": 8.096400000000001e-06, "loss": 0.0247, "step": 9519 }, { "epoch": 0.5075438502958896, "grad_norm": 1.9350430965423584, "learning_rate": 8.0962e-06, "loss": 0.0454, "step": 9520 }, { "epoch": 0.5075971637255424, "grad_norm": 3.2557430267333984, "learning_rate": 8.096e-06, "loss": 0.0336, "step": 9521 }, { "epoch": 0.5076504771551954, "grad_norm": 2.1304707527160645, "learning_rate": 8.0958e-06, "loss": 0.029, "step": 9522 }, { "epoch": 0.5077037905848483, "grad_norm": 2.922360897064209, "learning_rate": 8.095600000000001e-06, "loss": 0.0506, "step": 9523 }, { "epoch": 0.5077571040145012, "grad_norm": 2.4600095748901367, "learning_rate": 8.0954e-06, "loss": 0.0291, "step": 9524 }, { "epoch": 0.5078104174441542, "grad_norm": 2.2811992168426514, "learning_rate": 8.0952e-06, "loss": 0.0483, "step": 9525 }, { "epoch": 0.5078637308738071, "grad_norm": 2.716576099395752, "learning_rate": 8.095000000000001e-06, "loss": 0.0841, "step": 9526 }, { "epoch": 0.5079170443034601, "grad_norm": 1.5546058416366577, "learning_rate": 8.0948e-06, "loss": 0.0256, "step": 9527 }, { "epoch": 0.5079703577331129, "grad_norm": 4.1118388175964355, "learning_rate": 8.0946e-06, "loss": 0.0331, "step": 9528 }, { "epoch": 0.5080236711627659, "grad_norm": 2.5024008750915527, "learning_rate": 8.094400000000002e-06, "loss": 0.0321, "step": 9529 }, { "epoch": 0.5080769845924188, "grad_norm": 2.5190091133117676, "learning_rate": 8.094200000000001e-06, "loss": 0.0383, "step": 9530 }, { "epoch": 0.5081302980220718, "grad_norm": 2.5222175121307373, "learning_rate": 8.094e-06, "loss": 0.0343, "step": 9531 }, { "epoch": 0.5081836114517246, "grad_norm": 2.5576725006103516, "learning_rate": 8.0938e-06, "loss": 0.0568, "step": 9532 }, { "epoch": 0.5082369248813776, "grad_norm": 2.343440532684326, "learning_rate": 8.0936e-06, "loss": 0.0198, "step": 9533 }, { "epoch": 0.5082902383110306, "grad_norm": 1.7004733085632324, "learning_rate": 8.0934e-06, "loss": 0.0225, "step": 9534 }, { "epoch": 0.5083435517406835, "grad_norm": 3.3676302433013916, "learning_rate": 8.0932e-06, "loss": 0.0338, "step": 9535 }, { "epoch": 0.5083968651703364, "grad_norm": 2.78568959236145, "learning_rate": 8.093e-06, "loss": 0.0397, "step": 9536 }, { "epoch": 0.5084501785999893, "grad_norm": 4.397684574127197, "learning_rate": 8.0928e-06, "loss": 0.0787, "step": 9537 }, { "epoch": 0.5085034920296423, "grad_norm": 1.4061521291732788, "learning_rate": 8.092600000000001e-06, "loss": 0.0143, "step": 9538 }, { "epoch": 0.5085568054592952, "grad_norm": 1.8168870210647583, "learning_rate": 8.0924e-06, "loss": 0.0403, "step": 9539 }, { "epoch": 0.5086101188889481, "grad_norm": 1.9910457134246826, "learning_rate": 8.092200000000002e-06, "loss": 0.0173, "step": 9540 }, { "epoch": 0.508663432318601, "grad_norm": 2.3848648071289062, "learning_rate": 8.092000000000001e-06, "loss": 0.0261, "step": 9541 }, { "epoch": 0.508716745748254, "grad_norm": 2.070446252822876, "learning_rate": 8.091800000000001e-06, "loss": 0.0167, "step": 9542 }, { "epoch": 0.508770059177907, "grad_norm": 4.459072113037109, "learning_rate": 8.0916e-06, "loss": 0.0439, "step": 9543 }, { "epoch": 0.5088233726075598, "grad_norm": 2.3232526779174805, "learning_rate": 8.0914e-06, "loss": 0.0345, "step": 9544 }, { "epoch": 0.5088766860372128, "grad_norm": 2.1052684783935547, "learning_rate": 8.0912e-06, "loss": 0.0283, "step": 9545 }, { "epoch": 0.5089299994668657, "grad_norm": 2.4702584743499756, "learning_rate": 8.091000000000001e-06, "loss": 0.0528, "step": 9546 }, { "epoch": 0.5089833128965187, "grad_norm": 1.9905366897583008, "learning_rate": 8.0908e-06, "loss": 0.0279, "step": 9547 }, { "epoch": 0.5090366263261715, "grad_norm": 3.8298251628875732, "learning_rate": 8.0906e-06, "loss": 0.019, "step": 9548 }, { "epoch": 0.5090899397558245, "grad_norm": 3.5637362003326416, "learning_rate": 8.0904e-06, "loss": 0.0316, "step": 9549 }, { "epoch": 0.5091432531854774, "grad_norm": 1.535068392753601, "learning_rate": 8.090200000000001e-06, "loss": 0.0459, "step": 9550 }, { "epoch": 0.5091965666151304, "grad_norm": 1.969464659690857, "learning_rate": 8.09e-06, "loss": 0.0519, "step": 9551 }, { "epoch": 0.5092498800447833, "grad_norm": 2.3278133869171143, "learning_rate": 8.089800000000002e-06, "loss": 0.0258, "step": 9552 }, { "epoch": 0.5093031934744362, "grad_norm": 1.749024748802185, "learning_rate": 8.089600000000001e-06, "loss": 0.0347, "step": 9553 }, { "epoch": 0.5093565069040892, "grad_norm": 2.0717740058898926, "learning_rate": 8.089400000000001e-06, "loss": 0.0305, "step": 9554 }, { "epoch": 0.509409820333742, "grad_norm": 3.6348793506622314, "learning_rate": 8.0892e-06, "loss": 0.0312, "step": 9555 }, { "epoch": 0.509463133763395, "grad_norm": 2.920342206954956, "learning_rate": 8.089e-06, "loss": 0.0573, "step": 9556 }, { "epoch": 0.5095164471930479, "grad_norm": 4.143182754516602, "learning_rate": 8.088800000000001e-06, "loss": 0.0432, "step": 9557 }, { "epoch": 0.5095697606227009, "grad_norm": 1.9782401323318481, "learning_rate": 8.0886e-06, "loss": 0.0461, "step": 9558 }, { "epoch": 0.5096230740523537, "grad_norm": 1.4516370296478271, "learning_rate": 8.0884e-06, "loss": 0.0235, "step": 9559 }, { "epoch": 0.5096763874820067, "grad_norm": 2.5405592918395996, "learning_rate": 8.0882e-06, "loss": 0.0294, "step": 9560 }, { "epoch": 0.5097297009116597, "grad_norm": 1.6471381187438965, "learning_rate": 8.088e-06, "loss": 0.0302, "step": 9561 }, { "epoch": 0.5097830143413126, "grad_norm": 2.6865785121917725, "learning_rate": 8.0878e-06, "loss": 0.0648, "step": 9562 }, { "epoch": 0.5098363277709655, "grad_norm": 2.7601826190948486, "learning_rate": 8.0876e-06, "loss": 0.0427, "step": 9563 }, { "epoch": 0.5098896412006184, "grad_norm": 2.108419418334961, "learning_rate": 8.087400000000002e-06, "loss": 0.0151, "step": 9564 }, { "epoch": 0.5099429546302714, "grad_norm": 2.940873146057129, "learning_rate": 8.087200000000001e-06, "loss": 0.0592, "step": 9565 }, { "epoch": 0.5099962680599243, "grad_norm": 1.8761042356491089, "learning_rate": 8.087e-06, "loss": 0.019, "step": 9566 }, { "epoch": 0.5100495814895772, "grad_norm": 2.8130550384521484, "learning_rate": 8.0868e-06, "loss": 0.0198, "step": 9567 }, { "epoch": 0.5101028949192301, "grad_norm": 3.218613624572754, "learning_rate": 8.086600000000001e-06, "loss": 0.0629, "step": 9568 }, { "epoch": 0.5101562083488831, "grad_norm": 4.2724609375, "learning_rate": 8.086400000000001e-06, "loss": 0.0534, "step": 9569 }, { "epoch": 0.5102095217785361, "grad_norm": 2.7806220054626465, "learning_rate": 8.0862e-06, "loss": 0.0442, "step": 9570 }, { "epoch": 0.5102628352081889, "grad_norm": 2.9801042079925537, "learning_rate": 8.086e-06, "loss": 0.0379, "step": 9571 }, { "epoch": 0.5103161486378419, "grad_norm": 1.2876087427139282, "learning_rate": 8.0858e-06, "loss": 0.0332, "step": 9572 }, { "epoch": 0.5103694620674948, "grad_norm": 2.602621078491211, "learning_rate": 8.0856e-06, "loss": 0.0176, "step": 9573 }, { "epoch": 0.5104227754971478, "grad_norm": 1.777134895324707, "learning_rate": 8.0854e-06, "loss": 0.0363, "step": 9574 }, { "epoch": 0.5104760889268006, "grad_norm": 1.578629970550537, "learning_rate": 8.0852e-06, "loss": 0.0201, "step": 9575 }, { "epoch": 0.5105294023564536, "grad_norm": 3.313457489013672, "learning_rate": 8.085000000000001e-06, "loss": 0.0504, "step": 9576 }, { "epoch": 0.5105827157861065, "grad_norm": 3.818453073501587, "learning_rate": 8.084800000000001e-06, "loss": 0.0418, "step": 9577 }, { "epoch": 0.5106360292157595, "grad_norm": 2.637065887451172, "learning_rate": 8.0846e-06, "loss": 0.0346, "step": 9578 }, { "epoch": 0.5106893426454123, "grad_norm": 2.8296594619750977, "learning_rate": 8.084400000000002e-06, "loss": 0.0611, "step": 9579 }, { "epoch": 0.5107426560750653, "grad_norm": 1.9912316799163818, "learning_rate": 8.084200000000001e-06, "loss": 0.0254, "step": 9580 }, { "epoch": 0.5107959695047183, "grad_norm": 2.9384894371032715, "learning_rate": 8.084000000000001e-06, "loss": 0.0424, "step": 9581 }, { "epoch": 0.5108492829343712, "grad_norm": 2.1611406803131104, "learning_rate": 8.0838e-06, "loss": 0.0387, "step": 9582 }, { "epoch": 0.5109025963640241, "grad_norm": 2.017876386642456, "learning_rate": 8.0836e-06, "loss": 0.0375, "step": 9583 }, { "epoch": 0.510955909793677, "grad_norm": 1.8954683542251587, "learning_rate": 8.0834e-06, "loss": 0.0407, "step": 9584 }, { "epoch": 0.51100922322333, "grad_norm": 2.6650235652923584, "learning_rate": 8.0832e-06, "loss": 0.0497, "step": 9585 }, { "epoch": 0.5110625366529828, "grad_norm": 1.8855314254760742, "learning_rate": 8.083e-06, "loss": 0.03, "step": 9586 }, { "epoch": 0.5111158500826358, "grad_norm": 2.760634422302246, "learning_rate": 8.0828e-06, "loss": 0.048, "step": 9587 }, { "epoch": 0.5111691635122887, "grad_norm": 1.7503823041915894, "learning_rate": 8.082600000000001e-06, "loss": 0.0211, "step": 9588 }, { "epoch": 0.5112224769419417, "grad_norm": 2.396698474884033, "learning_rate": 8.0824e-06, "loss": 0.0335, "step": 9589 }, { "epoch": 0.5112757903715947, "grad_norm": 4.6037702560424805, "learning_rate": 8.082200000000002e-06, "loss": 0.048, "step": 9590 }, { "epoch": 0.5113291038012475, "grad_norm": 2.802532911300659, "learning_rate": 8.082000000000002e-06, "loss": 0.0165, "step": 9591 }, { "epoch": 0.5113824172309005, "grad_norm": 1.3023301362991333, "learning_rate": 8.081800000000001e-06, "loss": 0.0143, "step": 9592 }, { "epoch": 0.5114357306605534, "grad_norm": 3.2073981761932373, "learning_rate": 8.0816e-06, "loss": 0.0403, "step": 9593 }, { "epoch": 0.5114890440902063, "grad_norm": 2.6384387016296387, "learning_rate": 8.0814e-06, "loss": 0.0353, "step": 9594 }, { "epoch": 0.5115423575198592, "grad_norm": 2.5164639949798584, "learning_rate": 8.0812e-06, "loss": 0.0543, "step": 9595 }, { "epoch": 0.5115956709495122, "grad_norm": 3.807528495788574, "learning_rate": 8.081000000000001e-06, "loss": 0.0276, "step": 9596 }, { "epoch": 0.5116489843791651, "grad_norm": 2.8582255840301514, "learning_rate": 8.0808e-06, "loss": 0.0392, "step": 9597 }, { "epoch": 0.511702297808818, "grad_norm": 3.1698625087738037, "learning_rate": 8.0806e-06, "loss": 0.0347, "step": 9598 }, { "epoch": 0.511755611238471, "grad_norm": 1.8456449508666992, "learning_rate": 8.0804e-06, "loss": 0.0251, "step": 9599 }, { "epoch": 0.5118089246681239, "grad_norm": 2.748790979385376, "learning_rate": 8.080200000000001e-06, "loss": 0.0165, "step": 9600 }, { "epoch": 0.5118622380977769, "grad_norm": 2.8559179306030273, "learning_rate": 8.08e-06, "loss": 0.0418, "step": 9601 }, { "epoch": 0.5119155515274297, "grad_norm": 0.877949595451355, "learning_rate": 8.079800000000002e-06, "loss": 0.0122, "step": 9602 }, { "epoch": 0.5119688649570827, "grad_norm": 2.428192615509033, "learning_rate": 8.079600000000001e-06, "loss": 0.0458, "step": 9603 }, { "epoch": 0.5120221783867356, "grad_norm": 2.3100063800811768, "learning_rate": 8.079400000000001e-06, "loss": 0.0448, "step": 9604 }, { "epoch": 0.5120754918163886, "grad_norm": 1.0635303258895874, "learning_rate": 8.0792e-06, "loss": 0.0091, "step": 9605 }, { "epoch": 0.5121288052460414, "grad_norm": 2.542970657348633, "learning_rate": 8.079e-06, "loss": 0.0557, "step": 9606 }, { "epoch": 0.5121821186756944, "grad_norm": 5.887134552001953, "learning_rate": 8.078800000000001e-06, "loss": 0.0227, "step": 9607 }, { "epoch": 0.5122354321053474, "grad_norm": 4.628597259521484, "learning_rate": 8.078600000000001e-06, "loss": 0.0465, "step": 9608 }, { "epoch": 0.5122887455350003, "grad_norm": 3.1793408393859863, "learning_rate": 8.0784e-06, "loss": 0.0375, "step": 9609 }, { "epoch": 0.5123420589646532, "grad_norm": 4.17946720123291, "learning_rate": 8.0782e-06, "loss": 0.0543, "step": 9610 }, { "epoch": 0.5123953723943061, "grad_norm": 2.7920913696289062, "learning_rate": 8.078e-06, "loss": 0.0256, "step": 9611 }, { "epoch": 0.5124486858239591, "grad_norm": 1.887574553489685, "learning_rate": 8.0778e-06, "loss": 0.0223, "step": 9612 }, { "epoch": 0.512501999253612, "grad_norm": 4.409056186676025, "learning_rate": 8.0776e-06, "loss": 0.0269, "step": 9613 }, { "epoch": 0.5125553126832649, "grad_norm": 0.6678928136825562, "learning_rate": 8.077400000000002e-06, "loss": 0.0047, "step": 9614 }, { "epoch": 0.5126086261129178, "grad_norm": 2.854905366897583, "learning_rate": 8.077200000000001e-06, "loss": 0.0527, "step": 9615 }, { "epoch": 0.5126619395425708, "grad_norm": 1.9380102157592773, "learning_rate": 8.077e-06, "loss": 0.0444, "step": 9616 }, { "epoch": 0.5127152529722238, "grad_norm": 1.3336137533187866, "learning_rate": 8.0768e-06, "loss": 0.0173, "step": 9617 }, { "epoch": 0.5127685664018766, "grad_norm": 2.7317354679107666, "learning_rate": 8.076600000000002e-06, "loss": 0.02, "step": 9618 }, { "epoch": 0.5128218798315296, "grad_norm": 4.47303581237793, "learning_rate": 8.076400000000001e-06, "loss": 0.0653, "step": 9619 }, { "epoch": 0.5128751932611825, "grad_norm": 3.351344585418701, "learning_rate": 8.0762e-06, "loss": 0.0692, "step": 9620 }, { "epoch": 0.5129285066908355, "grad_norm": 2.4146275520324707, "learning_rate": 8.076e-06, "loss": 0.0414, "step": 9621 }, { "epoch": 0.5129818201204883, "grad_norm": 1.9187365770339966, "learning_rate": 8.0758e-06, "loss": 0.0221, "step": 9622 }, { "epoch": 0.5130351335501413, "grad_norm": 2.258711814880371, "learning_rate": 8.0756e-06, "loss": 0.0188, "step": 9623 }, { "epoch": 0.5130884469797942, "grad_norm": 3.53867769241333, "learning_rate": 8.0754e-06, "loss": 0.0596, "step": 9624 }, { "epoch": 0.5131417604094471, "grad_norm": 3.6944570541381836, "learning_rate": 8.0752e-06, "loss": 0.0215, "step": 9625 }, { "epoch": 0.5131950738391001, "grad_norm": 3.785025119781494, "learning_rate": 8.075000000000001e-06, "loss": 0.0454, "step": 9626 }, { "epoch": 0.513248387268753, "grad_norm": 2.41433048248291, "learning_rate": 8.074800000000001e-06, "loss": 0.036, "step": 9627 }, { "epoch": 0.513301700698406, "grad_norm": 3.2247836589813232, "learning_rate": 8.0746e-06, "loss": 0.0392, "step": 9628 }, { "epoch": 0.5133550141280588, "grad_norm": 2.970996379852295, "learning_rate": 8.074400000000002e-06, "loss": 0.0708, "step": 9629 }, { "epoch": 0.5134083275577118, "grad_norm": 1.9065574407577515, "learning_rate": 8.074200000000001e-06, "loss": 0.0224, "step": 9630 }, { "epoch": 0.5134616409873647, "grad_norm": 1.6885050535202026, "learning_rate": 8.074000000000001e-06, "loss": 0.0298, "step": 9631 }, { "epoch": 0.5135149544170177, "grad_norm": 2.0205492973327637, "learning_rate": 8.0738e-06, "loss": 0.0339, "step": 9632 }, { "epoch": 0.5135682678466705, "grad_norm": 1.5403884649276733, "learning_rate": 8.0736e-06, "loss": 0.0171, "step": 9633 }, { "epoch": 0.5136215812763235, "grad_norm": 3.3353500366210938, "learning_rate": 8.0734e-06, "loss": 0.0278, "step": 9634 }, { "epoch": 0.5136748947059764, "grad_norm": 1.2688713073730469, "learning_rate": 8.073200000000001e-06, "loss": 0.0145, "step": 9635 }, { "epoch": 0.5137282081356294, "grad_norm": 2.114793300628662, "learning_rate": 8.073e-06, "loss": 0.0233, "step": 9636 }, { "epoch": 0.5137815215652823, "grad_norm": 1.7518147230148315, "learning_rate": 8.0728e-06, "loss": 0.0266, "step": 9637 }, { "epoch": 0.5138348349949352, "grad_norm": 3.98004412651062, "learning_rate": 8.072600000000001e-06, "loss": 0.0609, "step": 9638 }, { "epoch": 0.5138881484245882, "grad_norm": 2.6481521129608154, "learning_rate": 8.0724e-06, "loss": 0.0472, "step": 9639 }, { "epoch": 0.513941461854241, "grad_norm": 2.5427770614624023, "learning_rate": 8.0722e-06, "loss": 0.0297, "step": 9640 }, { "epoch": 0.513994775283894, "grad_norm": 2.0268514156341553, "learning_rate": 8.072000000000002e-06, "loss": 0.0347, "step": 9641 }, { "epoch": 0.5140480887135469, "grad_norm": 1.9822602272033691, "learning_rate": 8.071800000000001e-06, "loss": 0.0272, "step": 9642 }, { "epoch": 0.5141014021431999, "grad_norm": 1.332802414894104, "learning_rate": 8.0716e-06, "loss": 0.026, "step": 9643 }, { "epoch": 0.5141547155728527, "grad_norm": 2.1079752445220947, "learning_rate": 8.0714e-06, "loss": 0.0383, "step": 9644 }, { "epoch": 0.5142080290025057, "grad_norm": 1.596307635307312, "learning_rate": 8.0712e-06, "loss": 0.0314, "step": 9645 }, { "epoch": 0.5142613424321587, "grad_norm": 1.1796212196350098, "learning_rate": 8.071000000000001e-06, "loss": 0.0121, "step": 9646 }, { "epoch": 0.5143146558618116, "grad_norm": 3.1224887371063232, "learning_rate": 8.0708e-06, "loss": 0.018, "step": 9647 }, { "epoch": 0.5143679692914646, "grad_norm": 3.7065372467041016, "learning_rate": 8.0706e-06, "loss": 0.0613, "step": 9648 }, { "epoch": 0.5144212827211174, "grad_norm": 2.1187710762023926, "learning_rate": 8.0704e-06, "loss": 0.0203, "step": 9649 }, { "epoch": 0.5144745961507704, "grad_norm": 3.255025863647461, "learning_rate": 8.070200000000001e-06, "loss": 0.043, "step": 9650 }, { "epoch": 0.5145279095804233, "grad_norm": 3.074805974960327, "learning_rate": 8.07e-06, "loss": 0.0558, "step": 9651 }, { "epoch": 0.5145812230100762, "grad_norm": 2.511914014816284, "learning_rate": 8.0698e-06, "loss": 0.0351, "step": 9652 }, { "epoch": 0.5146345364397291, "grad_norm": 2.568881034851074, "learning_rate": 8.069600000000001e-06, "loss": 0.0418, "step": 9653 }, { "epoch": 0.5146878498693821, "grad_norm": 3.3194093704223633, "learning_rate": 8.069400000000001e-06, "loss": 0.0598, "step": 9654 }, { "epoch": 0.5147411632990351, "grad_norm": 2.9816906452178955, "learning_rate": 8.0692e-06, "loss": 0.0611, "step": 9655 }, { "epoch": 0.5147944767286879, "grad_norm": 1.4969391822814941, "learning_rate": 8.069e-06, "loss": 0.0201, "step": 9656 }, { "epoch": 0.5148477901583409, "grad_norm": 3.6478800773620605, "learning_rate": 8.068800000000001e-06, "loss": 0.0435, "step": 9657 }, { "epoch": 0.5149011035879938, "grad_norm": 3.1023788452148438, "learning_rate": 8.068600000000001e-06, "loss": 0.037, "step": 9658 }, { "epoch": 0.5149544170176468, "grad_norm": 1.5727523565292358, "learning_rate": 8.0684e-06, "loss": 0.0187, "step": 9659 }, { "epoch": 0.5150077304472996, "grad_norm": 2.5303795337677, "learning_rate": 8.0682e-06, "loss": 0.0494, "step": 9660 }, { "epoch": 0.5150610438769526, "grad_norm": 1.5947896242141724, "learning_rate": 8.068e-06, "loss": 0.022, "step": 9661 }, { "epoch": 0.5151143573066055, "grad_norm": 3.5078790187835693, "learning_rate": 8.067800000000001e-06, "loss": 0.0271, "step": 9662 }, { "epoch": 0.5151676707362585, "grad_norm": 1.5318800210952759, "learning_rate": 8.0676e-06, "loss": 0.0167, "step": 9663 }, { "epoch": 0.5152209841659114, "grad_norm": 2.6351065635681152, "learning_rate": 8.0674e-06, "loss": 0.0257, "step": 9664 }, { "epoch": 0.5152742975955643, "grad_norm": 2.2323288917541504, "learning_rate": 8.067200000000001e-06, "loss": 0.0434, "step": 9665 }, { "epoch": 0.5153276110252173, "grad_norm": 1.8364536762237549, "learning_rate": 8.067000000000001e-06, "loss": 0.0254, "step": 9666 }, { "epoch": 0.5153809244548702, "grad_norm": 1.8224191665649414, "learning_rate": 8.0668e-06, "loss": 0.0373, "step": 9667 }, { "epoch": 0.5154342378845231, "grad_norm": 4.655062198638916, "learning_rate": 8.066600000000002e-06, "loss": 0.0505, "step": 9668 }, { "epoch": 0.515487551314176, "grad_norm": 1.3914437294006348, "learning_rate": 8.066400000000001e-06, "loss": 0.0201, "step": 9669 }, { "epoch": 0.515540864743829, "grad_norm": 1.6402219533920288, "learning_rate": 8.0662e-06, "loss": 0.0217, "step": 9670 }, { "epoch": 0.5155941781734819, "grad_norm": 1.9617292881011963, "learning_rate": 8.066e-06, "loss": 0.0265, "step": 9671 }, { "epoch": 0.5156474916031348, "grad_norm": 1.710381031036377, "learning_rate": 8.0658e-06, "loss": 0.0211, "step": 9672 }, { "epoch": 0.5157008050327878, "grad_norm": 1.837642788887024, "learning_rate": 8.065600000000001e-06, "loss": 0.0272, "step": 9673 }, { "epoch": 0.5157541184624407, "grad_norm": 3.596031904220581, "learning_rate": 8.0654e-06, "loss": 0.0327, "step": 9674 }, { "epoch": 0.5158074318920937, "grad_norm": 1.5291202068328857, "learning_rate": 8.0652e-06, "loss": 0.0284, "step": 9675 }, { "epoch": 0.5158607453217465, "grad_norm": 2.5978035926818848, "learning_rate": 8.065e-06, "loss": 0.018, "step": 9676 }, { "epoch": 0.5159140587513995, "grad_norm": 1.9356275796890259, "learning_rate": 8.064800000000001e-06, "loss": 0.0574, "step": 9677 }, { "epoch": 0.5159673721810524, "grad_norm": 2.0281097888946533, "learning_rate": 8.0646e-06, "loss": 0.0208, "step": 9678 }, { "epoch": 0.5160206856107054, "grad_norm": 3.3459434509277344, "learning_rate": 8.064400000000002e-06, "loss": 0.0341, "step": 9679 }, { "epoch": 0.5160739990403582, "grad_norm": 4.458475112915039, "learning_rate": 8.064200000000001e-06, "loss": 0.0353, "step": 9680 }, { "epoch": 0.5161273124700112, "grad_norm": 2.85433292388916, "learning_rate": 8.064000000000001e-06, "loss": 0.053, "step": 9681 }, { "epoch": 0.5161806258996642, "grad_norm": 2.2046096324920654, "learning_rate": 8.0638e-06, "loss": 0.0241, "step": 9682 }, { "epoch": 0.516233939329317, "grad_norm": 3.0185964107513428, "learning_rate": 8.0636e-06, "loss": 0.0464, "step": 9683 }, { "epoch": 0.51628725275897, "grad_norm": 1.7904311418533325, "learning_rate": 8.0634e-06, "loss": 0.0145, "step": 9684 }, { "epoch": 0.5163405661886229, "grad_norm": 2.178483724594116, "learning_rate": 8.063200000000001e-06, "loss": 0.0511, "step": 9685 }, { "epoch": 0.5163938796182759, "grad_norm": 2.6336727142333984, "learning_rate": 8.063e-06, "loss": 0.0366, "step": 9686 }, { "epoch": 0.5164471930479287, "grad_norm": 2.694883108139038, "learning_rate": 8.0628e-06, "loss": 0.0208, "step": 9687 }, { "epoch": 0.5165005064775817, "grad_norm": 2.249084711074829, "learning_rate": 8.0626e-06, "loss": 0.0307, "step": 9688 }, { "epoch": 0.5165538199072346, "grad_norm": 2.6073760986328125, "learning_rate": 8.062400000000001e-06, "loss": 0.0351, "step": 9689 }, { "epoch": 0.5166071333368876, "grad_norm": 2.7307984828948975, "learning_rate": 8.0622e-06, "loss": 0.0271, "step": 9690 }, { "epoch": 0.5166604467665404, "grad_norm": 2.273146867752075, "learning_rate": 8.062000000000002e-06, "loss": 0.0317, "step": 9691 }, { "epoch": 0.5167137601961934, "grad_norm": 2.286693572998047, "learning_rate": 8.061800000000001e-06, "loss": 0.017, "step": 9692 }, { "epoch": 0.5167670736258464, "grad_norm": 3.099994421005249, "learning_rate": 8.061600000000001e-06, "loss": 0.0409, "step": 9693 }, { "epoch": 0.5168203870554993, "grad_norm": 3.918299436569214, "learning_rate": 8.0614e-06, "loss": 0.0348, "step": 9694 }, { "epoch": 0.5168737004851522, "grad_norm": 1.431490182876587, "learning_rate": 8.0612e-06, "loss": 0.0147, "step": 9695 }, { "epoch": 0.5169270139148051, "grad_norm": 4.314356327056885, "learning_rate": 8.061000000000001e-06, "loss": 0.0496, "step": 9696 }, { "epoch": 0.5169803273444581, "grad_norm": 1.2505930662155151, "learning_rate": 8.0608e-06, "loss": 0.0137, "step": 9697 }, { "epoch": 0.517033640774111, "grad_norm": 3.7341511249542236, "learning_rate": 8.0606e-06, "loss": 0.0887, "step": 9698 }, { "epoch": 0.5170869542037639, "grad_norm": 1.9324510097503662, "learning_rate": 8.0604e-06, "loss": 0.028, "step": 9699 }, { "epoch": 0.5171402676334168, "grad_norm": 1.8069401979446411, "learning_rate": 8.0602e-06, "loss": 0.0263, "step": 9700 }, { "epoch": 0.5171935810630698, "grad_norm": 2.6253132820129395, "learning_rate": 8.06e-06, "loss": 0.0576, "step": 9701 }, { "epoch": 0.5172468944927228, "grad_norm": 2.443310022354126, "learning_rate": 8.0598e-06, "loss": 0.0271, "step": 9702 }, { "epoch": 0.5173002079223756, "grad_norm": 3.6008639335632324, "learning_rate": 8.059600000000002e-06, "loss": 0.0292, "step": 9703 }, { "epoch": 0.5173535213520286, "grad_norm": 3.8948261737823486, "learning_rate": 8.059400000000001e-06, "loss": 0.0806, "step": 9704 }, { "epoch": 0.5174068347816815, "grad_norm": 1.8898744583129883, "learning_rate": 8.0592e-06, "loss": 0.0111, "step": 9705 }, { "epoch": 0.5174601482113345, "grad_norm": 2.0768203735351562, "learning_rate": 8.059e-06, "loss": 0.0258, "step": 9706 }, { "epoch": 0.5175134616409873, "grad_norm": 2.890146493911743, "learning_rate": 8.058800000000002e-06, "loss": 0.0256, "step": 9707 }, { "epoch": 0.5175667750706403, "grad_norm": 2.3157408237457275, "learning_rate": 8.058600000000001e-06, "loss": 0.0388, "step": 9708 }, { "epoch": 0.5176200885002932, "grad_norm": 2.4027836322784424, "learning_rate": 8.0584e-06, "loss": 0.0195, "step": 9709 }, { "epoch": 0.5176734019299462, "grad_norm": 4.318781852722168, "learning_rate": 8.0582e-06, "loss": 0.0353, "step": 9710 }, { "epoch": 0.5177267153595991, "grad_norm": 2.6027920246124268, "learning_rate": 8.058e-06, "loss": 0.0494, "step": 9711 }, { "epoch": 0.517780028789252, "grad_norm": 1.5514267683029175, "learning_rate": 8.057800000000001e-06, "loss": 0.0252, "step": 9712 }, { "epoch": 0.517833342218905, "grad_norm": 3.459246873855591, "learning_rate": 8.0576e-06, "loss": 0.0391, "step": 9713 }, { "epoch": 0.5178866556485578, "grad_norm": 3.0659921169281006, "learning_rate": 8.0574e-06, "loss": 0.043, "step": 9714 }, { "epoch": 0.5179399690782108, "grad_norm": 2.033514976501465, "learning_rate": 8.057200000000001e-06, "loss": 0.0359, "step": 9715 }, { "epoch": 0.5179932825078637, "grad_norm": 4.073425769805908, "learning_rate": 8.057000000000001e-06, "loss": 0.089, "step": 9716 }, { "epoch": 0.5180465959375167, "grad_norm": 4.055098056793213, "learning_rate": 8.0568e-06, "loss": 0.0673, "step": 9717 }, { "epoch": 0.5180999093671695, "grad_norm": 1.820970892906189, "learning_rate": 8.056600000000002e-06, "loss": 0.0183, "step": 9718 }, { "epoch": 0.5181532227968225, "grad_norm": 2.459449052810669, "learning_rate": 8.056400000000001e-06, "loss": 0.0162, "step": 9719 }, { "epoch": 0.5182065362264755, "grad_norm": 1.9662448167800903, "learning_rate": 8.056200000000001e-06, "loss": 0.0306, "step": 9720 }, { "epoch": 0.5182598496561284, "grad_norm": 3.6609368324279785, "learning_rate": 8.056e-06, "loss": 0.0363, "step": 9721 }, { "epoch": 0.5183131630857813, "grad_norm": 1.9373054504394531, "learning_rate": 8.0558e-06, "loss": 0.03, "step": 9722 }, { "epoch": 0.5183664765154342, "grad_norm": 3.139566421508789, "learning_rate": 8.055600000000001e-06, "loss": 0.054, "step": 9723 }, { "epoch": 0.5184197899450872, "grad_norm": 3.7779839038848877, "learning_rate": 8.0554e-06, "loss": 0.0414, "step": 9724 }, { "epoch": 0.5184731033747401, "grad_norm": 2.056626081466675, "learning_rate": 8.0552e-06, "loss": 0.0263, "step": 9725 }, { "epoch": 0.518526416804393, "grad_norm": 1.458464503288269, "learning_rate": 8.055e-06, "loss": 0.0208, "step": 9726 }, { "epoch": 0.5185797302340459, "grad_norm": 2.1857051849365234, "learning_rate": 8.054800000000001e-06, "loss": 0.0318, "step": 9727 }, { "epoch": 0.5186330436636989, "grad_norm": 1.3308621644973755, "learning_rate": 8.0546e-06, "loss": 0.0293, "step": 9728 }, { "epoch": 0.5186863570933519, "grad_norm": 3.209907293319702, "learning_rate": 8.0544e-06, "loss": 0.0519, "step": 9729 }, { "epoch": 0.5187396705230047, "grad_norm": 1.2569935321807861, "learning_rate": 8.054200000000002e-06, "loss": 0.0221, "step": 9730 }, { "epoch": 0.5187929839526577, "grad_norm": 3.5412330627441406, "learning_rate": 8.054000000000001e-06, "loss": 0.0214, "step": 9731 }, { "epoch": 0.5188462973823106, "grad_norm": 2.5978612899780273, "learning_rate": 8.0538e-06, "loss": 0.0309, "step": 9732 }, { "epoch": 0.5188996108119636, "grad_norm": 1.0590009689331055, "learning_rate": 8.0536e-06, "loss": 0.0091, "step": 9733 }, { "epoch": 0.5189529242416164, "grad_norm": 3.7618916034698486, "learning_rate": 8.0534e-06, "loss": 0.0503, "step": 9734 }, { "epoch": 0.5190062376712694, "grad_norm": 4.379506587982178, "learning_rate": 8.053200000000001e-06, "loss": 0.0611, "step": 9735 }, { "epoch": 0.5190595511009223, "grad_norm": 2.4479033946990967, "learning_rate": 8.053e-06, "loss": 0.0429, "step": 9736 }, { "epoch": 0.5191128645305753, "grad_norm": 3.121448278427124, "learning_rate": 8.0528e-06, "loss": 0.0367, "step": 9737 }, { "epoch": 0.5191661779602282, "grad_norm": 3.1740078926086426, "learning_rate": 8.0526e-06, "loss": 0.0278, "step": 9738 }, { "epoch": 0.5192194913898811, "grad_norm": 6.614207744598389, "learning_rate": 8.052400000000001e-06, "loss": 0.0351, "step": 9739 }, { "epoch": 0.5192728048195341, "grad_norm": 2.119349479675293, "learning_rate": 8.0522e-06, "loss": 0.0246, "step": 9740 }, { "epoch": 0.519326118249187, "grad_norm": 4.1746745109558105, "learning_rate": 8.052e-06, "loss": 0.0415, "step": 9741 }, { "epoch": 0.5193794316788399, "grad_norm": 1.8332098722457886, "learning_rate": 8.051800000000001e-06, "loss": 0.0194, "step": 9742 }, { "epoch": 0.5194327451084928, "grad_norm": 3.0176477432250977, "learning_rate": 8.051600000000001e-06, "loss": 0.0948, "step": 9743 }, { "epoch": 0.5194860585381458, "grad_norm": 2.1341750621795654, "learning_rate": 8.0514e-06, "loss": 0.0201, "step": 9744 }, { "epoch": 0.5195393719677986, "grad_norm": 1.5802887678146362, "learning_rate": 8.0512e-06, "loss": 0.021, "step": 9745 }, { "epoch": 0.5195926853974516, "grad_norm": 2.642158031463623, "learning_rate": 8.051000000000001e-06, "loss": 0.0834, "step": 9746 }, { "epoch": 0.5196459988271045, "grad_norm": 2.203160285949707, "learning_rate": 8.050800000000001e-06, "loss": 0.0543, "step": 9747 }, { "epoch": 0.5196993122567575, "grad_norm": 4.233396530151367, "learning_rate": 8.0506e-06, "loss": 0.0418, "step": 9748 }, { "epoch": 0.5197526256864105, "grad_norm": 1.2990498542785645, "learning_rate": 8.0504e-06, "loss": 0.029, "step": 9749 }, { "epoch": 0.5198059391160633, "grad_norm": 6.08804988861084, "learning_rate": 8.0502e-06, "loss": 0.0817, "step": 9750 }, { "epoch": 0.5198592525457163, "grad_norm": 1.9896478652954102, "learning_rate": 8.050000000000001e-06, "loss": 0.0272, "step": 9751 }, { "epoch": 0.5199125659753692, "grad_norm": 1.6384066343307495, "learning_rate": 8.0498e-06, "loss": 0.022, "step": 9752 }, { "epoch": 0.5199658794050221, "grad_norm": 1.8192329406738281, "learning_rate": 8.0496e-06, "loss": 0.0356, "step": 9753 }, { "epoch": 0.520019192834675, "grad_norm": 4.166981220245361, "learning_rate": 8.049400000000001e-06, "loss": 0.047, "step": 9754 }, { "epoch": 0.520072506264328, "grad_norm": 4.202893257141113, "learning_rate": 8.0492e-06, "loss": 0.0669, "step": 9755 }, { "epoch": 0.5201258196939809, "grad_norm": 3.1847152709960938, "learning_rate": 8.049e-06, "loss": 0.0461, "step": 9756 }, { "epoch": 0.5201791331236338, "grad_norm": 1.4690803289413452, "learning_rate": 8.048800000000002e-06, "loss": 0.0277, "step": 9757 }, { "epoch": 0.5202324465532868, "grad_norm": 2.1734201908111572, "learning_rate": 8.048600000000001e-06, "loss": 0.0444, "step": 9758 }, { "epoch": 0.5202857599829397, "grad_norm": 1.5374990701675415, "learning_rate": 8.0484e-06, "loss": 0.0423, "step": 9759 }, { "epoch": 0.5203390734125927, "grad_norm": 3.7487809658050537, "learning_rate": 8.0482e-06, "loss": 0.0406, "step": 9760 }, { "epoch": 0.5203923868422455, "grad_norm": 1.8508812189102173, "learning_rate": 8.048e-06, "loss": 0.0413, "step": 9761 }, { "epoch": 0.5204457002718985, "grad_norm": 2.0598230361938477, "learning_rate": 8.047800000000001e-06, "loss": 0.0295, "step": 9762 }, { "epoch": 0.5204990137015514, "grad_norm": 3.6385512351989746, "learning_rate": 8.0476e-06, "loss": 0.0535, "step": 9763 }, { "epoch": 0.5205523271312044, "grad_norm": 1.357675552368164, "learning_rate": 8.0474e-06, "loss": 0.0117, "step": 9764 }, { "epoch": 0.5206056405608572, "grad_norm": 3.348273515701294, "learning_rate": 8.0472e-06, "loss": 0.0348, "step": 9765 }, { "epoch": 0.5206589539905102, "grad_norm": 1.5554341077804565, "learning_rate": 8.047000000000001e-06, "loss": 0.0126, "step": 9766 }, { "epoch": 0.5207122674201632, "grad_norm": 3.1258766651153564, "learning_rate": 8.0468e-06, "loss": 0.0625, "step": 9767 }, { "epoch": 0.5207655808498161, "grad_norm": 2.611124277114868, "learning_rate": 8.046600000000002e-06, "loss": 0.031, "step": 9768 }, { "epoch": 0.520818894279469, "grad_norm": 1.4579243659973145, "learning_rate": 8.046400000000001e-06, "loss": 0.0192, "step": 9769 }, { "epoch": 0.5208722077091219, "grad_norm": 2.3688647747039795, "learning_rate": 8.046200000000001e-06, "loss": 0.0249, "step": 9770 }, { "epoch": 0.5209255211387749, "grad_norm": 2.9823269844055176, "learning_rate": 8.046e-06, "loss": 0.0426, "step": 9771 }, { "epoch": 0.5209788345684278, "grad_norm": 2.155916929244995, "learning_rate": 8.0458e-06, "loss": 0.0375, "step": 9772 }, { "epoch": 0.5210321479980807, "grad_norm": 3.1306309700012207, "learning_rate": 8.045600000000001e-06, "loss": 0.0687, "step": 9773 }, { "epoch": 0.5210854614277336, "grad_norm": 1.9520164728164673, "learning_rate": 8.045400000000001e-06, "loss": 0.0326, "step": 9774 }, { "epoch": 0.5211387748573866, "grad_norm": 2.0575308799743652, "learning_rate": 8.0452e-06, "loss": 0.0253, "step": 9775 }, { "epoch": 0.5211920882870396, "grad_norm": 5.236303806304932, "learning_rate": 8.045e-06, "loss": 0.0595, "step": 9776 }, { "epoch": 0.5212454017166924, "grad_norm": 3.8582980632781982, "learning_rate": 8.0448e-06, "loss": 0.0302, "step": 9777 }, { "epoch": 0.5212987151463454, "grad_norm": 1.6269433498382568, "learning_rate": 8.044600000000001e-06, "loss": 0.0159, "step": 9778 }, { "epoch": 0.5213520285759983, "grad_norm": 1.8217214345932007, "learning_rate": 8.0444e-06, "loss": 0.0449, "step": 9779 }, { "epoch": 0.5214053420056513, "grad_norm": 2.9182982444763184, "learning_rate": 8.044200000000002e-06, "loss": 0.0524, "step": 9780 }, { "epoch": 0.5214586554353041, "grad_norm": 1.3829432725906372, "learning_rate": 8.044000000000001e-06, "loss": 0.0153, "step": 9781 }, { "epoch": 0.5215119688649571, "grad_norm": 1.666575312614441, "learning_rate": 8.0438e-06, "loss": 0.0248, "step": 9782 }, { "epoch": 0.52156528229461, "grad_norm": 1.3872355222702026, "learning_rate": 8.0436e-06, "loss": 0.0311, "step": 9783 }, { "epoch": 0.521618595724263, "grad_norm": 1.1417732238769531, "learning_rate": 8.043400000000002e-06, "loss": 0.0181, "step": 9784 }, { "epoch": 0.5216719091539159, "grad_norm": 0.8992058634757996, "learning_rate": 8.043200000000001e-06, "loss": 0.013, "step": 9785 }, { "epoch": 0.5217252225835688, "grad_norm": 2.494311809539795, "learning_rate": 8.043e-06, "loss": 0.0387, "step": 9786 }, { "epoch": 0.5217785360132218, "grad_norm": 4.194471836090088, "learning_rate": 8.0428e-06, "loss": 0.0581, "step": 9787 }, { "epoch": 0.5218318494428746, "grad_norm": 2.858215570449829, "learning_rate": 8.0426e-06, "loss": 0.0321, "step": 9788 }, { "epoch": 0.5218851628725276, "grad_norm": 3.346285820007324, "learning_rate": 8.0424e-06, "loss": 0.0471, "step": 9789 }, { "epoch": 0.5219384763021805, "grad_norm": 1.7874912023544312, "learning_rate": 8.0422e-06, "loss": 0.0277, "step": 9790 }, { "epoch": 0.5219917897318335, "grad_norm": 1.997818946838379, "learning_rate": 8.042e-06, "loss": 0.0258, "step": 9791 }, { "epoch": 0.5220451031614863, "grad_norm": 2.370276689529419, "learning_rate": 8.041800000000002e-06, "loss": 0.039, "step": 9792 }, { "epoch": 0.5220984165911393, "grad_norm": 1.1695334911346436, "learning_rate": 8.041600000000001e-06, "loss": 0.0129, "step": 9793 }, { "epoch": 0.5221517300207923, "grad_norm": 6.184828281402588, "learning_rate": 8.0414e-06, "loss": 0.0823, "step": 9794 }, { "epoch": 0.5222050434504452, "grad_norm": 4.792304515838623, "learning_rate": 8.0412e-06, "loss": 0.0671, "step": 9795 }, { "epoch": 0.5222583568800981, "grad_norm": 3.1139512062072754, "learning_rate": 8.041000000000001e-06, "loss": 0.024, "step": 9796 }, { "epoch": 0.522311670309751, "grad_norm": 3.2178573608398438, "learning_rate": 8.040800000000001e-06, "loss": 0.0322, "step": 9797 }, { "epoch": 0.522364983739404, "grad_norm": 2.128073215484619, "learning_rate": 8.0406e-06, "loss": 0.0205, "step": 9798 }, { "epoch": 0.5224182971690569, "grad_norm": 2.0889463424682617, "learning_rate": 8.0404e-06, "loss": 0.0176, "step": 9799 }, { "epoch": 0.5224716105987098, "grad_norm": 3.3298850059509277, "learning_rate": 8.0402e-06, "loss": 0.06, "step": 9800 }, { "epoch": 0.5225249240283627, "grad_norm": 2.6902832984924316, "learning_rate": 8.040000000000001e-06, "loss": 0.036, "step": 9801 }, { "epoch": 0.5225782374580157, "grad_norm": 1.1207666397094727, "learning_rate": 8.0398e-06, "loss": 0.0108, "step": 9802 }, { "epoch": 0.5226315508876686, "grad_norm": 1.5194356441497803, "learning_rate": 8.0396e-06, "loss": 0.0123, "step": 9803 }, { "epoch": 0.5226848643173215, "grad_norm": 2.7984182834625244, "learning_rate": 8.039400000000001e-06, "loss": 0.0405, "step": 9804 }, { "epoch": 0.5227381777469745, "grad_norm": 3.881120204925537, "learning_rate": 8.039200000000001e-06, "loss": 0.0997, "step": 9805 }, { "epoch": 0.5227914911766274, "grad_norm": 2.729626178741455, "learning_rate": 8.039e-06, "loss": 0.0433, "step": 9806 }, { "epoch": 0.5228448046062804, "grad_norm": 1.7722136974334717, "learning_rate": 8.038800000000002e-06, "loss": 0.0232, "step": 9807 }, { "epoch": 0.5228981180359332, "grad_norm": 2.2375314235687256, "learning_rate": 8.038600000000001e-06, "loss": 0.0343, "step": 9808 }, { "epoch": 0.5229514314655862, "grad_norm": 1.7057702541351318, "learning_rate": 8.0384e-06, "loss": 0.0094, "step": 9809 }, { "epoch": 0.5230047448952391, "grad_norm": 2.710322618484497, "learning_rate": 8.0382e-06, "loss": 0.0307, "step": 9810 }, { "epoch": 0.523058058324892, "grad_norm": 3.760641098022461, "learning_rate": 8.038e-06, "loss": 0.0497, "step": 9811 }, { "epoch": 0.5231113717545449, "grad_norm": 2.4754135608673096, "learning_rate": 8.037800000000001e-06, "loss": 0.0232, "step": 9812 }, { "epoch": 0.5231646851841979, "grad_norm": 2.857193946838379, "learning_rate": 8.0376e-06, "loss": 0.0236, "step": 9813 }, { "epoch": 0.5232179986138509, "grad_norm": 3.39845871925354, "learning_rate": 8.0374e-06, "loss": 0.0425, "step": 9814 }, { "epoch": 0.5232713120435037, "grad_norm": 2.377774953842163, "learning_rate": 8.0372e-06, "loss": 0.0276, "step": 9815 }, { "epoch": 0.5233246254731567, "grad_norm": 2.1079983711242676, "learning_rate": 8.037000000000001e-06, "loss": 0.026, "step": 9816 }, { "epoch": 0.5233779389028096, "grad_norm": 2.735104560852051, "learning_rate": 8.0368e-06, "loss": 0.0546, "step": 9817 }, { "epoch": 0.5234312523324626, "grad_norm": 2.1142513751983643, "learning_rate": 8.0366e-06, "loss": 0.0377, "step": 9818 }, { "epoch": 0.5234845657621154, "grad_norm": 2.7694091796875, "learning_rate": 8.036400000000002e-06, "loss": 0.0575, "step": 9819 }, { "epoch": 0.5235378791917684, "grad_norm": 1.822422981262207, "learning_rate": 8.036200000000001e-06, "loss": 0.0425, "step": 9820 }, { "epoch": 0.5235911926214213, "grad_norm": 2.598630666732788, "learning_rate": 8.036e-06, "loss": 0.0309, "step": 9821 }, { "epoch": 0.5236445060510743, "grad_norm": 1.6309722661972046, "learning_rate": 8.0358e-06, "loss": 0.0157, "step": 9822 }, { "epoch": 0.5236978194807272, "grad_norm": 2.462702751159668, "learning_rate": 8.035600000000001e-06, "loss": 0.0475, "step": 9823 }, { "epoch": 0.5237511329103801, "grad_norm": 2.9056601524353027, "learning_rate": 8.035400000000001e-06, "loss": 0.0315, "step": 9824 }, { "epoch": 0.5238044463400331, "grad_norm": 2.2934210300445557, "learning_rate": 8.0352e-06, "loss": 0.024, "step": 9825 }, { "epoch": 0.523857759769686, "grad_norm": 2.5638904571533203, "learning_rate": 8.035e-06, "loss": 0.0326, "step": 9826 }, { "epoch": 0.5239110731993389, "grad_norm": 2.6060409545898438, "learning_rate": 8.0348e-06, "loss": 0.0235, "step": 9827 }, { "epoch": 0.5239643866289918, "grad_norm": 2.398766040802002, "learning_rate": 8.034600000000001e-06, "loss": 0.0229, "step": 9828 }, { "epoch": 0.5240177000586448, "grad_norm": 2.6176629066467285, "learning_rate": 8.0344e-06, "loss": 0.0335, "step": 9829 }, { "epoch": 0.5240710134882977, "grad_norm": 2.225040912628174, "learning_rate": 8.0342e-06, "loss": 0.037, "step": 9830 }, { "epoch": 0.5241243269179506, "grad_norm": 3.313096523284912, "learning_rate": 8.034000000000001e-06, "loss": 0.0532, "step": 9831 }, { "epoch": 0.5241776403476036, "grad_norm": 2.8489232063293457, "learning_rate": 8.033800000000001e-06, "loss": 0.0576, "step": 9832 }, { "epoch": 0.5242309537772565, "grad_norm": 3.699207067489624, "learning_rate": 8.0336e-06, "loss": 0.0493, "step": 9833 }, { "epoch": 0.5242842672069095, "grad_norm": 1.5346503257751465, "learning_rate": 8.033400000000002e-06, "loss": 0.0184, "step": 9834 }, { "epoch": 0.5243375806365623, "grad_norm": 7.5367279052734375, "learning_rate": 8.033200000000001e-06, "loss": 0.0505, "step": 9835 }, { "epoch": 0.5243908940662153, "grad_norm": 2.657773733139038, "learning_rate": 8.033e-06, "loss": 0.0452, "step": 9836 }, { "epoch": 0.5244442074958682, "grad_norm": 3.471513271331787, "learning_rate": 8.0328e-06, "loss": 0.0378, "step": 9837 }, { "epoch": 0.5244975209255212, "grad_norm": 4.368317127227783, "learning_rate": 8.0326e-06, "loss": 0.0595, "step": 9838 }, { "epoch": 0.524550834355174, "grad_norm": 2.864171028137207, "learning_rate": 8.0324e-06, "loss": 0.0398, "step": 9839 }, { "epoch": 0.524604147784827, "grad_norm": 3.0739543437957764, "learning_rate": 8.0322e-06, "loss": 0.0333, "step": 9840 }, { "epoch": 0.52465746121448, "grad_norm": 2.5880136489868164, "learning_rate": 8.032e-06, "loss": 0.0231, "step": 9841 }, { "epoch": 0.5247107746441328, "grad_norm": 2.688103675842285, "learning_rate": 8.0318e-06, "loss": 0.028, "step": 9842 }, { "epoch": 0.5247640880737858, "grad_norm": 1.672615647315979, "learning_rate": 8.031600000000001e-06, "loss": 0.0302, "step": 9843 }, { "epoch": 0.5248174015034387, "grad_norm": 2.6246395111083984, "learning_rate": 8.0314e-06, "loss": 0.0408, "step": 9844 }, { "epoch": 0.5248707149330917, "grad_norm": 2.686680316925049, "learning_rate": 8.0312e-06, "loss": 0.0392, "step": 9845 }, { "epoch": 0.5249240283627445, "grad_norm": 2.868809938430786, "learning_rate": 8.031000000000002e-06, "loss": 0.0408, "step": 9846 }, { "epoch": 0.5249773417923975, "grad_norm": 3.8359787464141846, "learning_rate": 8.030800000000001e-06, "loss": 0.0644, "step": 9847 }, { "epoch": 0.5250306552220504, "grad_norm": 3.131448984146118, "learning_rate": 8.0306e-06, "loss": 0.0128, "step": 9848 }, { "epoch": 0.5250839686517034, "grad_norm": 2.2649755477905273, "learning_rate": 8.0304e-06, "loss": 0.0536, "step": 9849 }, { "epoch": 0.5251372820813563, "grad_norm": 3.5794825553894043, "learning_rate": 8.0302e-06, "loss": 0.0299, "step": 9850 }, { "epoch": 0.5251905955110092, "grad_norm": 1.8760279417037964, "learning_rate": 8.030000000000001e-06, "loss": 0.0296, "step": 9851 }, { "epoch": 0.5252439089406622, "grad_norm": 2.6531503200531006, "learning_rate": 8.0298e-06, "loss": 0.0375, "step": 9852 }, { "epoch": 0.5252972223703151, "grad_norm": 2.3345420360565186, "learning_rate": 8.0296e-06, "loss": 0.0368, "step": 9853 }, { "epoch": 0.525350535799968, "grad_norm": 1.4260562658309937, "learning_rate": 8.0294e-06, "loss": 0.0212, "step": 9854 }, { "epoch": 0.5254038492296209, "grad_norm": 3.2027933597564697, "learning_rate": 8.029200000000001e-06, "loss": 0.0413, "step": 9855 }, { "epoch": 0.5254571626592739, "grad_norm": 3.8043088912963867, "learning_rate": 8.029e-06, "loss": 0.0343, "step": 9856 }, { "epoch": 0.5255104760889268, "grad_norm": 2.0937023162841797, "learning_rate": 8.028800000000002e-06, "loss": 0.0225, "step": 9857 }, { "epoch": 0.5255637895185797, "grad_norm": 3.8968448638916016, "learning_rate": 8.028600000000001e-06, "loss": 0.0591, "step": 9858 }, { "epoch": 0.5256171029482326, "grad_norm": 2.052964925765991, "learning_rate": 8.028400000000001e-06, "loss": 0.046, "step": 9859 }, { "epoch": 0.5256704163778856, "grad_norm": 2.668910264968872, "learning_rate": 8.0282e-06, "loss": 0.0598, "step": 9860 }, { "epoch": 0.5257237298075386, "grad_norm": 1.119861364364624, "learning_rate": 8.028e-06, "loss": 0.0142, "step": 9861 }, { "epoch": 0.5257770432371914, "grad_norm": 1.527608036994934, "learning_rate": 8.027800000000001e-06, "loss": 0.0205, "step": 9862 }, { "epoch": 0.5258303566668444, "grad_norm": 2.028047561645508, "learning_rate": 8.027600000000001e-06, "loss": 0.0437, "step": 9863 }, { "epoch": 0.5258836700964973, "grad_norm": 3.4962475299835205, "learning_rate": 8.0274e-06, "loss": 0.0548, "step": 9864 }, { "epoch": 0.5259369835261503, "grad_norm": 1.2343579530715942, "learning_rate": 8.0272e-06, "loss": 0.0179, "step": 9865 }, { "epoch": 0.5259902969558031, "grad_norm": 1.671390175819397, "learning_rate": 8.027e-06, "loss": 0.0198, "step": 9866 }, { "epoch": 0.5260436103854561, "grad_norm": 3.348088026046753, "learning_rate": 8.0268e-06, "loss": 0.0418, "step": 9867 }, { "epoch": 0.526096923815109, "grad_norm": 2.527509927749634, "learning_rate": 8.0266e-06, "loss": 0.0264, "step": 9868 }, { "epoch": 0.526150237244762, "grad_norm": 2.4766781330108643, "learning_rate": 8.026400000000002e-06, "loss": 0.0306, "step": 9869 }, { "epoch": 0.5262035506744149, "grad_norm": 1.1974223852157593, "learning_rate": 8.026200000000001e-06, "loss": 0.014, "step": 9870 }, { "epoch": 0.5262568641040678, "grad_norm": 2.320014238357544, "learning_rate": 8.026e-06, "loss": 0.0308, "step": 9871 }, { "epoch": 0.5263101775337208, "grad_norm": 1.6388274431228638, "learning_rate": 8.0258e-06, "loss": 0.0343, "step": 9872 }, { "epoch": 0.5263634909633736, "grad_norm": 3.9165499210357666, "learning_rate": 8.025600000000002e-06, "loss": 0.0371, "step": 9873 }, { "epoch": 0.5264168043930266, "grad_norm": 1.7105650901794434, "learning_rate": 8.025400000000001e-06, "loss": 0.0267, "step": 9874 }, { "epoch": 0.5264701178226795, "grad_norm": 1.5808888673782349, "learning_rate": 8.0252e-06, "loss": 0.0212, "step": 9875 }, { "epoch": 0.5265234312523325, "grad_norm": 2.400452136993408, "learning_rate": 8.025e-06, "loss": 0.0509, "step": 9876 }, { "epoch": 0.5265767446819853, "grad_norm": 2.6731812953948975, "learning_rate": 8.0248e-06, "loss": 0.0558, "step": 9877 }, { "epoch": 0.5266300581116383, "grad_norm": 1.7762224674224854, "learning_rate": 8.0246e-06, "loss": 0.0155, "step": 9878 }, { "epoch": 0.5266833715412913, "grad_norm": 2.0998337268829346, "learning_rate": 8.0244e-06, "loss": 0.0423, "step": 9879 }, { "epoch": 0.5267366849709442, "grad_norm": 2.0194132328033447, "learning_rate": 8.0242e-06, "loss": 0.0291, "step": 9880 }, { "epoch": 0.5267899984005971, "grad_norm": 3.6075148582458496, "learning_rate": 8.024000000000001e-06, "loss": 0.0448, "step": 9881 }, { "epoch": 0.52684331183025, "grad_norm": 1.6377428770065308, "learning_rate": 8.023800000000001e-06, "loss": 0.0268, "step": 9882 }, { "epoch": 0.526896625259903, "grad_norm": 2.794881820678711, "learning_rate": 8.0236e-06, "loss": 0.0264, "step": 9883 }, { "epoch": 0.5269499386895559, "grad_norm": 4.129805088043213, "learning_rate": 8.023400000000002e-06, "loss": 0.0441, "step": 9884 }, { "epoch": 0.5270032521192088, "grad_norm": 2.5091824531555176, "learning_rate": 8.023200000000001e-06, "loss": 0.0184, "step": 9885 }, { "epoch": 0.5270565655488617, "grad_norm": 4.263108730316162, "learning_rate": 8.023000000000001e-06, "loss": 0.0437, "step": 9886 }, { "epoch": 0.5271098789785147, "grad_norm": 3.87747859954834, "learning_rate": 8.0228e-06, "loss": 0.0471, "step": 9887 }, { "epoch": 0.5271631924081677, "grad_norm": 2.8024959564208984, "learning_rate": 8.0226e-06, "loss": 0.0359, "step": 9888 }, { "epoch": 0.5272165058378205, "grad_norm": 2.1412155628204346, "learning_rate": 8.0224e-06, "loss": 0.0377, "step": 9889 }, { "epoch": 0.5272698192674735, "grad_norm": 3.3591766357421875, "learning_rate": 8.022200000000001e-06, "loss": 0.0251, "step": 9890 }, { "epoch": 0.5273231326971264, "grad_norm": 3.9240102767944336, "learning_rate": 8.022e-06, "loss": 0.0381, "step": 9891 }, { "epoch": 0.5273764461267794, "grad_norm": 4.548863887786865, "learning_rate": 8.0218e-06, "loss": 0.0327, "step": 9892 }, { "epoch": 0.5274297595564322, "grad_norm": 2.8519506454467773, "learning_rate": 8.021600000000001e-06, "loss": 0.0409, "step": 9893 }, { "epoch": 0.5274830729860852, "grad_norm": 1.9101399183273315, "learning_rate": 8.0214e-06, "loss": 0.0345, "step": 9894 }, { "epoch": 0.5275363864157381, "grad_norm": 3.1033151149749756, "learning_rate": 8.0212e-06, "loss": 0.0255, "step": 9895 }, { "epoch": 0.5275896998453911, "grad_norm": 7.402624130249023, "learning_rate": 8.021000000000002e-06, "loss": 0.0515, "step": 9896 }, { "epoch": 0.527643013275044, "grad_norm": 8.18564510345459, "learning_rate": 8.020800000000001e-06, "loss": 0.0321, "step": 9897 }, { "epoch": 0.5276963267046969, "grad_norm": 1.890203595161438, "learning_rate": 8.0206e-06, "loss": 0.0107, "step": 9898 }, { "epoch": 0.5277496401343499, "grad_norm": 2.2373905181884766, "learning_rate": 8.0204e-06, "loss": 0.0251, "step": 9899 }, { "epoch": 0.5278029535640028, "grad_norm": 3.7721750736236572, "learning_rate": 8.0202e-06, "loss": 0.0231, "step": 9900 }, { "epoch": 0.5278562669936557, "grad_norm": 3.1728148460388184, "learning_rate": 8.020000000000001e-06, "loss": 0.0395, "step": 9901 }, { "epoch": 0.5279095804233086, "grad_norm": 2.210845947265625, "learning_rate": 8.0198e-06, "loss": 0.0345, "step": 9902 }, { "epoch": 0.5279628938529616, "grad_norm": 5.059890270233154, "learning_rate": 8.0196e-06, "loss": 0.014, "step": 9903 }, { "epoch": 0.5280162072826144, "grad_norm": 4.625617504119873, "learning_rate": 8.0194e-06, "loss": 0.0491, "step": 9904 }, { "epoch": 0.5280695207122674, "grad_norm": 0.7411012649536133, "learning_rate": 8.019200000000001e-06, "loss": 0.0044, "step": 9905 }, { "epoch": 0.5281228341419203, "grad_norm": 3.5983498096466064, "learning_rate": 8.019e-06, "loss": 0.0342, "step": 9906 }, { "epoch": 0.5281761475715733, "grad_norm": 4.439246654510498, "learning_rate": 8.0188e-06, "loss": 0.0568, "step": 9907 }, { "epoch": 0.5282294610012263, "grad_norm": 4.030791759490967, "learning_rate": 8.018600000000001e-06, "loss": 0.0351, "step": 9908 }, { "epoch": 0.5282827744308791, "grad_norm": 3.2139060497283936, "learning_rate": 8.018400000000001e-06, "loss": 0.0461, "step": 9909 }, { "epoch": 0.5283360878605321, "grad_norm": 2.93814754486084, "learning_rate": 8.0182e-06, "loss": 0.0303, "step": 9910 }, { "epoch": 0.528389401290185, "grad_norm": 3.273350954055786, "learning_rate": 8.018e-06, "loss": 0.0445, "step": 9911 }, { "epoch": 0.528442714719838, "grad_norm": 6.406154632568359, "learning_rate": 8.017800000000001e-06, "loss": 0.0812, "step": 9912 }, { "epoch": 0.5284960281494908, "grad_norm": 2.5307633876800537, "learning_rate": 8.017600000000001e-06, "loss": 0.0157, "step": 9913 }, { "epoch": 0.5285493415791438, "grad_norm": 2.3125100135803223, "learning_rate": 8.0174e-06, "loss": 0.0491, "step": 9914 }, { "epoch": 0.5286026550087967, "grad_norm": 2.197838306427002, "learning_rate": 8.0172e-06, "loss": 0.0478, "step": 9915 }, { "epoch": 0.5286559684384496, "grad_norm": 2.7504348754882812, "learning_rate": 8.017e-06, "loss": 0.0325, "step": 9916 }, { "epoch": 0.5287092818681026, "grad_norm": 5.180997848510742, "learning_rate": 8.016800000000001e-06, "loss": 0.0726, "step": 9917 }, { "epoch": 0.5287625952977555, "grad_norm": 3.1005146503448486, "learning_rate": 8.0166e-06, "loss": 0.0573, "step": 9918 }, { "epoch": 0.5288159087274085, "grad_norm": 3.985337257385254, "learning_rate": 8.0164e-06, "loss": 0.0368, "step": 9919 }, { "epoch": 0.5288692221570613, "grad_norm": 3.0467641353607178, "learning_rate": 8.016200000000001e-06, "loss": 0.0487, "step": 9920 }, { "epoch": 0.5289225355867143, "grad_norm": 3.8001201152801514, "learning_rate": 8.016e-06, "loss": 0.0325, "step": 9921 }, { "epoch": 0.5289758490163672, "grad_norm": 2.7118170261383057, "learning_rate": 8.0158e-06, "loss": 0.0317, "step": 9922 }, { "epoch": 0.5290291624460202, "grad_norm": 4.044283866882324, "learning_rate": 8.015600000000002e-06, "loss": 0.0625, "step": 9923 }, { "epoch": 0.529082475875673, "grad_norm": 2.7099528312683105, "learning_rate": 8.015400000000001e-06, "loss": 0.0513, "step": 9924 }, { "epoch": 0.529135789305326, "grad_norm": 3.4854254722595215, "learning_rate": 8.0152e-06, "loss": 0.0431, "step": 9925 }, { "epoch": 0.529189102734979, "grad_norm": 1.6655287742614746, "learning_rate": 8.015e-06, "loss": 0.0326, "step": 9926 }, { "epoch": 0.5292424161646319, "grad_norm": 2.3405251502990723, "learning_rate": 8.0148e-06, "loss": 0.0363, "step": 9927 }, { "epoch": 0.5292957295942848, "grad_norm": 4.266508102416992, "learning_rate": 8.0146e-06, "loss": 0.0713, "step": 9928 }, { "epoch": 0.5293490430239377, "grad_norm": 1.3251352310180664, "learning_rate": 8.0144e-06, "loss": 0.0247, "step": 9929 }, { "epoch": 0.5294023564535907, "grad_norm": 2.950202703475952, "learning_rate": 8.0142e-06, "loss": 0.028, "step": 9930 }, { "epoch": 0.5294556698832436, "grad_norm": 1.947582721710205, "learning_rate": 8.014e-06, "loss": 0.0344, "step": 9931 }, { "epoch": 0.5295089833128965, "grad_norm": 2.70192289352417, "learning_rate": 8.013800000000001e-06, "loss": 0.0417, "step": 9932 }, { "epoch": 0.5295622967425494, "grad_norm": 2.977498769760132, "learning_rate": 8.0136e-06, "loss": 0.0671, "step": 9933 }, { "epoch": 0.5296156101722024, "grad_norm": 1.6433898210525513, "learning_rate": 8.013400000000002e-06, "loss": 0.0261, "step": 9934 }, { "epoch": 0.5296689236018554, "grad_norm": 3.5522632598876953, "learning_rate": 8.013200000000001e-06, "loss": 0.06, "step": 9935 }, { "epoch": 0.5297222370315082, "grad_norm": 2.5701053142547607, "learning_rate": 8.013000000000001e-06, "loss": 0.0262, "step": 9936 }, { "epoch": 0.5297755504611612, "grad_norm": 1.610290288925171, "learning_rate": 8.0128e-06, "loss": 0.036, "step": 9937 }, { "epoch": 0.5298288638908141, "grad_norm": 3.3284993171691895, "learning_rate": 8.0126e-06, "loss": 0.0387, "step": 9938 }, { "epoch": 0.529882177320467, "grad_norm": 1.7622451782226562, "learning_rate": 8.0124e-06, "loss": 0.018, "step": 9939 }, { "epoch": 0.5299354907501199, "grad_norm": 2.588270425796509, "learning_rate": 8.012200000000001e-06, "loss": 0.0441, "step": 9940 }, { "epoch": 0.5299888041797729, "grad_norm": 2.374387264251709, "learning_rate": 8.012e-06, "loss": 0.0294, "step": 9941 }, { "epoch": 0.5300421176094258, "grad_norm": 3.9897470474243164, "learning_rate": 8.0118e-06, "loss": 0.0649, "step": 9942 }, { "epoch": 0.5300954310390787, "grad_norm": 2.5647053718566895, "learning_rate": 8.0116e-06, "loss": 0.0319, "step": 9943 }, { "epoch": 0.5301487444687317, "grad_norm": 2.2504959106445312, "learning_rate": 8.011400000000001e-06, "loss": 0.0243, "step": 9944 }, { "epoch": 0.5302020578983846, "grad_norm": 1.7335772514343262, "learning_rate": 8.0112e-06, "loss": 0.0393, "step": 9945 }, { "epoch": 0.5302553713280376, "grad_norm": 1.5682255029678345, "learning_rate": 8.011000000000002e-06, "loss": 0.0446, "step": 9946 }, { "epoch": 0.5303086847576904, "grad_norm": 3.981510639190674, "learning_rate": 8.010800000000001e-06, "loss": 0.0626, "step": 9947 }, { "epoch": 0.5303619981873434, "grad_norm": 1.7906489372253418, "learning_rate": 8.010600000000001e-06, "loss": 0.0229, "step": 9948 }, { "epoch": 0.5304153116169963, "grad_norm": 3.8685033321380615, "learning_rate": 8.0104e-06, "loss": 0.0628, "step": 9949 }, { "epoch": 0.5304686250466493, "grad_norm": 5.602297782897949, "learning_rate": 8.0102e-06, "loss": 0.0764, "step": 9950 }, { "epoch": 0.5305219384763021, "grad_norm": 1.894432544708252, "learning_rate": 8.010000000000001e-06, "loss": 0.0316, "step": 9951 }, { "epoch": 0.5305752519059551, "grad_norm": 1.9218242168426514, "learning_rate": 8.0098e-06, "loss": 0.0302, "step": 9952 }, { "epoch": 0.5306285653356081, "grad_norm": 3.4379565715789795, "learning_rate": 8.0096e-06, "loss": 0.0623, "step": 9953 }, { "epoch": 0.530681878765261, "grad_norm": 2.2992172241210938, "learning_rate": 8.0094e-06, "loss": 0.0381, "step": 9954 }, { "epoch": 0.5307351921949139, "grad_norm": 3.252438545227051, "learning_rate": 8.0092e-06, "loss": 0.063, "step": 9955 }, { "epoch": 0.5307885056245668, "grad_norm": 1.4819061756134033, "learning_rate": 8.009e-06, "loss": 0.0307, "step": 9956 }, { "epoch": 0.5308418190542198, "grad_norm": 2.43013334274292, "learning_rate": 8.0088e-06, "loss": 0.0263, "step": 9957 }, { "epoch": 0.5308951324838727, "grad_norm": 1.6397435665130615, "learning_rate": 8.008600000000002e-06, "loss": 0.0311, "step": 9958 }, { "epoch": 0.5309484459135256, "grad_norm": 2.09173846244812, "learning_rate": 8.008400000000001e-06, "loss": 0.048, "step": 9959 }, { "epoch": 0.5310017593431785, "grad_norm": 3.511838436126709, "learning_rate": 8.0082e-06, "loss": 0.0578, "step": 9960 }, { "epoch": 0.5310550727728315, "grad_norm": 4.093258857727051, "learning_rate": 8.008e-06, "loss": 0.0466, "step": 9961 }, { "epoch": 0.5311083862024844, "grad_norm": 2.2572522163391113, "learning_rate": 8.007800000000001e-06, "loss": 0.0416, "step": 9962 }, { "epoch": 0.5311616996321373, "grad_norm": 2.149906873703003, "learning_rate": 8.007600000000001e-06, "loss": 0.0329, "step": 9963 }, { "epoch": 0.5312150130617903, "grad_norm": 2.196113348007202, "learning_rate": 8.0074e-06, "loss": 0.0434, "step": 9964 }, { "epoch": 0.5312683264914432, "grad_norm": 1.7266535758972168, "learning_rate": 8.0072e-06, "loss": 0.0377, "step": 9965 }, { "epoch": 0.5313216399210962, "grad_norm": 4.160599231719971, "learning_rate": 8.007e-06, "loss": 0.0765, "step": 9966 }, { "epoch": 0.531374953350749, "grad_norm": 2.770679473876953, "learning_rate": 8.0068e-06, "loss": 0.0394, "step": 9967 }, { "epoch": 0.531428266780402, "grad_norm": 2.232304811477661, "learning_rate": 8.0066e-06, "loss": 0.027, "step": 9968 }, { "epoch": 0.5314815802100549, "grad_norm": 1.5868115425109863, "learning_rate": 8.0064e-06, "loss": 0.0267, "step": 9969 }, { "epoch": 0.5315348936397079, "grad_norm": 4.20068883895874, "learning_rate": 8.006200000000001e-06, "loss": 0.0273, "step": 9970 }, { "epoch": 0.5315882070693607, "grad_norm": 1.8032293319702148, "learning_rate": 8.006000000000001e-06, "loss": 0.0329, "step": 9971 }, { "epoch": 0.5316415204990137, "grad_norm": 1.3526065349578857, "learning_rate": 8.0058e-06, "loss": 0.0191, "step": 9972 }, { "epoch": 0.5316948339286667, "grad_norm": 2.176872968673706, "learning_rate": 8.005600000000002e-06, "loss": 0.0388, "step": 9973 }, { "epoch": 0.5317481473583195, "grad_norm": 2.2264363765716553, "learning_rate": 8.005400000000001e-06, "loss": 0.048, "step": 9974 }, { "epoch": 0.5318014607879725, "grad_norm": 2.2634665966033936, "learning_rate": 8.005200000000001e-06, "loss": 0.0634, "step": 9975 }, { "epoch": 0.5318547742176254, "grad_norm": 2.9565539360046387, "learning_rate": 8.005e-06, "loss": 0.0197, "step": 9976 }, { "epoch": 0.5319080876472784, "grad_norm": 1.9356770515441895, "learning_rate": 8.0048e-06, "loss": 0.0326, "step": 9977 }, { "epoch": 0.5319614010769312, "grad_norm": 2.3440017700195312, "learning_rate": 8.0046e-06, "loss": 0.04, "step": 9978 }, { "epoch": 0.5320147145065842, "grad_norm": 1.7124085426330566, "learning_rate": 8.0044e-06, "loss": 0.0385, "step": 9979 }, { "epoch": 0.5320680279362371, "grad_norm": 1.6388238668441772, "learning_rate": 8.0042e-06, "loss": 0.0264, "step": 9980 }, { "epoch": 0.5321213413658901, "grad_norm": 1.2944862842559814, "learning_rate": 8.004e-06, "loss": 0.0146, "step": 9981 }, { "epoch": 0.532174654795543, "grad_norm": 2.4663829803466797, "learning_rate": 8.003800000000001e-06, "loss": 0.0289, "step": 9982 }, { "epoch": 0.5322279682251959, "grad_norm": 4.3050217628479, "learning_rate": 8.0036e-06, "loss": 0.0275, "step": 9983 }, { "epoch": 0.5322812816548489, "grad_norm": 2.0859487056732178, "learning_rate": 8.003400000000002e-06, "loss": 0.0344, "step": 9984 }, { "epoch": 0.5323345950845018, "grad_norm": 2.990229368209839, "learning_rate": 8.003200000000002e-06, "loss": 0.0503, "step": 9985 }, { "epoch": 0.5323879085141547, "grad_norm": 2.248995780944824, "learning_rate": 8.003000000000001e-06, "loss": 0.0289, "step": 9986 }, { "epoch": 0.5324412219438076, "grad_norm": 0.9057332873344421, "learning_rate": 8.0028e-06, "loss": 0.011, "step": 9987 }, { "epoch": 0.5324945353734606, "grad_norm": 1.603933572769165, "learning_rate": 8.0026e-06, "loss": 0.0319, "step": 9988 }, { "epoch": 0.5325478488031135, "grad_norm": 4.652328014373779, "learning_rate": 8.0024e-06, "loss": 0.0962, "step": 9989 }, { "epoch": 0.5326011622327664, "grad_norm": 2.256359100341797, "learning_rate": 8.002200000000001e-06, "loss": 0.0198, "step": 9990 }, { "epoch": 0.5326544756624194, "grad_norm": 1.444089412689209, "learning_rate": 8.002e-06, "loss": 0.0217, "step": 9991 }, { "epoch": 0.5327077890920723, "grad_norm": 1.9301506280899048, "learning_rate": 8.0018e-06, "loss": 0.0143, "step": 9992 }, { "epoch": 0.5327611025217253, "grad_norm": 1.3789036273956299, "learning_rate": 8.0016e-06, "loss": 0.0163, "step": 9993 }, { "epoch": 0.5328144159513781, "grad_norm": 2.0994932651519775, "learning_rate": 8.001400000000001e-06, "loss": 0.0474, "step": 9994 }, { "epoch": 0.5328677293810311, "grad_norm": 2.5952792167663574, "learning_rate": 8.0012e-06, "loss": 0.0393, "step": 9995 }, { "epoch": 0.532921042810684, "grad_norm": 4.827663898468018, "learning_rate": 8.001000000000002e-06, "loss": 0.0448, "step": 9996 }, { "epoch": 0.532974356240337, "grad_norm": 2.416595220565796, "learning_rate": 8.000800000000001e-06, "loss": 0.0383, "step": 9997 }, { "epoch": 0.5330276696699898, "grad_norm": 1.834247350692749, "learning_rate": 8.000600000000001e-06, "loss": 0.0259, "step": 9998 }, { "epoch": 0.5330809830996428, "grad_norm": 2.7501304149627686, "learning_rate": 8.0004e-06, "loss": 0.0539, "step": 9999 }, { "epoch": 0.5331342965292958, "grad_norm": 3.045637845993042, "learning_rate": 8.0002e-06, "loss": 0.03, "step": 10000 }, { "epoch": 0.5331876099589486, "grad_norm": 3.0923025608062744, "learning_rate": 8.000000000000001e-06, "loss": 0.047, "step": 10001 }, { "epoch": 0.5332409233886016, "grad_norm": 1.4652258157730103, "learning_rate": 7.999800000000001e-06, "loss": 0.0171, "step": 10002 }, { "epoch": 0.5332942368182545, "grad_norm": 1.3105387687683105, "learning_rate": 7.9996e-06, "loss": 0.0222, "step": 10003 }, { "epoch": 0.5333475502479075, "grad_norm": 2.526184558868408, "learning_rate": 7.9994e-06, "loss": 0.0465, "step": 10004 }, { "epoch": 0.5334008636775603, "grad_norm": 1.118943214416504, "learning_rate": 7.9992e-06, "loss": 0.0132, "step": 10005 }, { "epoch": 0.5334541771072133, "grad_norm": 2.2726337909698486, "learning_rate": 7.999e-06, "loss": 0.0268, "step": 10006 }, { "epoch": 0.5335074905368662, "grad_norm": 1.2320044040679932, "learning_rate": 7.9988e-06, "loss": 0.0098, "step": 10007 }, { "epoch": 0.5335608039665192, "grad_norm": 2.0416364669799805, "learning_rate": 7.998600000000002e-06, "loss": 0.0228, "step": 10008 }, { "epoch": 0.5336141173961721, "grad_norm": 2.7603683471679688, "learning_rate": 7.998400000000001e-06, "loss": 0.0515, "step": 10009 }, { "epoch": 0.533667430825825, "grad_norm": 6.084752559661865, "learning_rate": 7.9982e-06, "loss": 0.0522, "step": 10010 }, { "epoch": 0.533720744255478, "grad_norm": 4.264558792114258, "learning_rate": 7.998e-06, "loss": 0.02, "step": 10011 }, { "epoch": 0.5337740576851309, "grad_norm": 2.852775812149048, "learning_rate": 7.997800000000002e-06, "loss": 0.0511, "step": 10012 }, { "epoch": 0.5338273711147838, "grad_norm": 3.444397211074829, "learning_rate": 7.997600000000001e-06, "loss": 0.0747, "step": 10013 }, { "epoch": 0.5338806845444367, "grad_norm": 1.6144051551818848, "learning_rate": 7.9974e-06, "loss": 0.0209, "step": 10014 }, { "epoch": 0.5339339979740897, "grad_norm": 1.8960835933685303, "learning_rate": 7.9972e-06, "loss": 0.0285, "step": 10015 }, { "epoch": 0.5339873114037426, "grad_norm": 2.2102413177490234, "learning_rate": 7.997e-06, "loss": 0.0387, "step": 10016 }, { "epoch": 0.5340406248333955, "grad_norm": 2.0444908142089844, "learning_rate": 7.9968e-06, "loss": 0.0172, "step": 10017 }, { "epoch": 0.5340939382630484, "grad_norm": 2.290541648864746, "learning_rate": 7.9966e-06, "loss": 0.0301, "step": 10018 }, { "epoch": 0.5341472516927014, "grad_norm": 1.7289397716522217, "learning_rate": 7.9964e-06, "loss": 0.0263, "step": 10019 }, { "epoch": 0.5342005651223544, "grad_norm": 2.7504560947418213, "learning_rate": 7.996200000000001e-06, "loss": 0.024, "step": 10020 }, { "epoch": 0.5342538785520072, "grad_norm": 1.8566303253173828, "learning_rate": 7.996000000000001e-06, "loss": 0.0112, "step": 10021 }, { "epoch": 0.5343071919816602, "grad_norm": 2.22878360748291, "learning_rate": 7.9958e-06, "loss": 0.0213, "step": 10022 }, { "epoch": 0.5343605054113131, "grad_norm": 2.219224691390991, "learning_rate": 7.995600000000002e-06, "loss": 0.0493, "step": 10023 }, { "epoch": 0.5344138188409661, "grad_norm": 2.613129138946533, "learning_rate": 7.995400000000001e-06, "loss": 0.0556, "step": 10024 }, { "epoch": 0.5344671322706189, "grad_norm": 1.8214994668960571, "learning_rate": 7.995200000000001e-06, "loss": 0.0333, "step": 10025 }, { "epoch": 0.5345204457002719, "grad_norm": 1.424757719039917, "learning_rate": 7.995e-06, "loss": 0.0121, "step": 10026 }, { "epoch": 0.5345737591299248, "grad_norm": 2.8679492473602295, "learning_rate": 7.9948e-06, "loss": 0.0225, "step": 10027 }, { "epoch": 0.5346270725595778, "grad_norm": 2.2184174060821533, "learning_rate": 7.9946e-06, "loss": 0.031, "step": 10028 }, { "epoch": 0.5346803859892307, "grad_norm": 1.690091848373413, "learning_rate": 7.994400000000001e-06, "loss": 0.0394, "step": 10029 }, { "epoch": 0.5347336994188836, "grad_norm": 2.3531036376953125, "learning_rate": 7.9942e-06, "loss": 0.0366, "step": 10030 }, { "epoch": 0.5347870128485366, "grad_norm": 2.2891969680786133, "learning_rate": 7.994e-06, "loss": 0.0307, "step": 10031 }, { "epoch": 0.5348403262781894, "grad_norm": 3.1397833824157715, "learning_rate": 7.993800000000001e-06, "loss": 0.0475, "step": 10032 }, { "epoch": 0.5348936397078424, "grad_norm": 2.3735501766204834, "learning_rate": 7.993600000000001e-06, "loss": 0.0328, "step": 10033 }, { "epoch": 0.5349469531374953, "grad_norm": 3.9650983810424805, "learning_rate": 7.9934e-06, "loss": 0.0434, "step": 10034 }, { "epoch": 0.5350002665671483, "grad_norm": 3.214905261993408, "learning_rate": 7.993200000000002e-06, "loss": 0.0343, "step": 10035 }, { "epoch": 0.5350535799968011, "grad_norm": 4.184406280517578, "learning_rate": 7.993000000000001e-06, "loss": 0.0583, "step": 10036 }, { "epoch": 0.5351068934264541, "grad_norm": 2.5732226371765137, "learning_rate": 7.9928e-06, "loss": 0.0518, "step": 10037 }, { "epoch": 0.5351602068561071, "grad_norm": 2.9898853302001953, "learning_rate": 7.9926e-06, "loss": 0.029, "step": 10038 }, { "epoch": 0.53521352028576, "grad_norm": 3.4630751609802246, "learning_rate": 7.9924e-06, "loss": 0.032, "step": 10039 }, { "epoch": 0.535266833715413, "grad_norm": 3.3453423976898193, "learning_rate": 7.992200000000001e-06, "loss": 0.0398, "step": 10040 }, { "epoch": 0.5353201471450658, "grad_norm": 1.312094807624817, "learning_rate": 7.992e-06, "loss": 0.0299, "step": 10041 }, { "epoch": 0.5353734605747188, "grad_norm": 1.6440982818603516, "learning_rate": 7.9918e-06, "loss": 0.0335, "step": 10042 }, { "epoch": 0.5354267740043717, "grad_norm": 3.1799864768981934, "learning_rate": 7.9916e-06, "loss": 0.0509, "step": 10043 }, { "epoch": 0.5354800874340246, "grad_norm": 4.004801273345947, "learning_rate": 7.991400000000001e-06, "loss": 0.0596, "step": 10044 }, { "epoch": 0.5355334008636775, "grad_norm": 1.3296773433685303, "learning_rate": 7.9912e-06, "loss": 0.0216, "step": 10045 }, { "epoch": 0.5355867142933305, "grad_norm": 3.7496135234832764, "learning_rate": 7.991e-06, "loss": 0.079, "step": 10046 }, { "epoch": 0.5356400277229835, "grad_norm": 4.986497402191162, "learning_rate": 7.990800000000001e-06, "loss": 0.0307, "step": 10047 }, { "epoch": 0.5356933411526363, "grad_norm": 2.0920846462249756, "learning_rate": 7.990600000000001e-06, "loss": 0.0188, "step": 10048 }, { "epoch": 0.5357466545822893, "grad_norm": 2.2424628734588623, "learning_rate": 7.9904e-06, "loss": 0.0278, "step": 10049 }, { "epoch": 0.5357999680119422, "grad_norm": 1.5802165269851685, "learning_rate": 7.9902e-06, "loss": 0.0234, "step": 10050 }, { "epoch": 0.5358532814415952, "grad_norm": 1.1244677305221558, "learning_rate": 7.990000000000001e-06, "loss": 0.0239, "step": 10051 }, { "epoch": 0.535906594871248, "grad_norm": 1.7370376586914062, "learning_rate": 7.989800000000001e-06, "loss": 0.0482, "step": 10052 }, { "epoch": 0.535959908300901, "grad_norm": 4.149655818939209, "learning_rate": 7.9896e-06, "loss": 0.0761, "step": 10053 }, { "epoch": 0.5360132217305539, "grad_norm": 3.4218928813934326, "learning_rate": 7.9894e-06, "loss": 0.0196, "step": 10054 }, { "epoch": 0.5360665351602069, "grad_norm": 1.7158468961715698, "learning_rate": 7.9892e-06, "loss": 0.0341, "step": 10055 }, { "epoch": 0.5361198485898598, "grad_norm": 1.577844262123108, "learning_rate": 7.989000000000001e-06, "loss": 0.0263, "step": 10056 }, { "epoch": 0.5361731620195127, "grad_norm": 2.702749252319336, "learning_rate": 7.9888e-06, "loss": 0.0257, "step": 10057 }, { "epoch": 0.5362264754491657, "grad_norm": 3.304647922515869, "learning_rate": 7.9886e-06, "loss": 0.0267, "step": 10058 }, { "epoch": 0.5362797888788186, "grad_norm": 1.6905264854431152, "learning_rate": 7.988400000000001e-06, "loss": 0.0182, "step": 10059 }, { "epoch": 0.5363331023084715, "grad_norm": 2.8512284755706787, "learning_rate": 7.988200000000001e-06, "loss": 0.0431, "step": 10060 }, { "epoch": 0.5363864157381244, "grad_norm": 2.561112880706787, "learning_rate": 7.988e-06, "loss": 0.0422, "step": 10061 }, { "epoch": 0.5364397291677774, "grad_norm": 3.5299434661865234, "learning_rate": 7.987800000000002e-06, "loss": 0.0586, "step": 10062 }, { "epoch": 0.5364930425974302, "grad_norm": 2.3627026081085205, "learning_rate": 7.987600000000001e-06, "loss": 0.0455, "step": 10063 }, { "epoch": 0.5365463560270832, "grad_norm": 2.0273401737213135, "learning_rate": 7.9874e-06, "loss": 0.0308, "step": 10064 }, { "epoch": 0.5365996694567362, "grad_norm": 2.9225618839263916, "learning_rate": 7.9872e-06, "loss": 0.0317, "step": 10065 }, { "epoch": 0.5366529828863891, "grad_norm": 2.7442548274993896, "learning_rate": 7.987e-06, "loss": 0.0693, "step": 10066 }, { "epoch": 0.536706296316042, "grad_norm": 2.6602511405944824, "learning_rate": 7.9868e-06, "loss": 0.0191, "step": 10067 }, { "epoch": 0.5367596097456949, "grad_norm": 2.7262606620788574, "learning_rate": 7.9866e-06, "loss": 0.0491, "step": 10068 }, { "epoch": 0.5368129231753479, "grad_norm": 1.742788314819336, "learning_rate": 7.9864e-06, "loss": 0.0176, "step": 10069 }, { "epoch": 0.5368662366050008, "grad_norm": 2.1018755435943604, "learning_rate": 7.9862e-06, "loss": 0.0365, "step": 10070 }, { "epoch": 0.5369195500346537, "grad_norm": 2.592048168182373, "learning_rate": 7.986000000000001e-06, "loss": 0.0457, "step": 10071 }, { "epoch": 0.5369728634643066, "grad_norm": 2.5940189361572266, "learning_rate": 7.9858e-06, "loss": 0.0308, "step": 10072 }, { "epoch": 0.5370261768939596, "grad_norm": 2.399315118789673, "learning_rate": 7.985600000000002e-06, "loss": 0.0492, "step": 10073 }, { "epoch": 0.5370794903236125, "grad_norm": 3.0939018726348877, "learning_rate": 7.985400000000002e-06, "loss": 0.0623, "step": 10074 }, { "epoch": 0.5371328037532654, "grad_norm": 3.6743276119232178, "learning_rate": 7.985200000000001e-06, "loss": 0.031, "step": 10075 }, { "epoch": 0.5371861171829184, "grad_norm": 1.719353199005127, "learning_rate": 7.985e-06, "loss": 0.0232, "step": 10076 }, { "epoch": 0.5372394306125713, "grad_norm": 2.5442087650299072, "learning_rate": 7.9848e-06, "loss": 0.0551, "step": 10077 }, { "epoch": 0.5372927440422243, "grad_norm": 2.433931350708008, "learning_rate": 7.9846e-06, "loss": 0.0217, "step": 10078 }, { "epoch": 0.5373460574718771, "grad_norm": 2.3031532764434814, "learning_rate": 7.984400000000001e-06, "loss": 0.0159, "step": 10079 }, { "epoch": 0.5373993709015301, "grad_norm": 4.05746603012085, "learning_rate": 7.9842e-06, "loss": 0.0654, "step": 10080 }, { "epoch": 0.537452684331183, "grad_norm": 6.624824047088623, "learning_rate": 7.984e-06, "loss": 0.0571, "step": 10081 }, { "epoch": 0.537505997760836, "grad_norm": 2.834994316101074, "learning_rate": 7.9838e-06, "loss": 0.033, "step": 10082 }, { "epoch": 0.5375593111904888, "grad_norm": 4.021297454833984, "learning_rate": 7.983600000000001e-06, "loss": 0.0839, "step": 10083 }, { "epoch": 0.5376126246201418, "grad_norm": 1.7992935180664062, "learning_rate": 7.9834e-06, "loss": 0.0159, "step": 10084 }, { "epoch": 0.5376659380497948, "grad_norm": 4.647357940673828, "learning_rate": 7.983200000000002e-06, "loss": 0.0303, "step": 10085 }, { "epoch": 0.5377192514794477, "grad_norm": 2.6338398456573486, "learning_rate": 7.983000000000001e-06, "loss": 0.0325, "step": 10086 }, { "epoch": 0.5377725649091006, "grad_norm": 8.912745475769043, "learning_rate": 7.982800000000001e-06, "loss": 0.0653, "step": 10087 }, { "epoch": 0.5378258783387535, "grad_norm": 3.9940185546875, "learning_rate": 7.9826e-06, "loss": 0.0309, "step": 10088 }, { "epoch": 0.5378791917684065, "grad_norm": 1.7145261764526367, "learning_rate": 7.9824e-06, "loss": 0.0487, "step": 10089 }, { "epoch": 0.5379325051980594, "grad_norm": 1.1144503355026245, "learning_rate": 7.982200000000001e-06, "loss": 0.024, "step": 10090 }, { "epoch": 0.5379858186277123, "grad_norm": 2.0732638835906982, "learning_rate": 7.982e-06, "loss": 0.0351, "step": 10091 }, { "epoch": 0.5380391320573652, "grad_norm": 1.5957062244415283, "learning_rate": 7.9818e-06, "loss": 0.0367, "step": 10092 }, { "epoch": 0.5380924454870182, "grad_norm": 0.9488283395767212, "learning_rate": 7.9816e-06, "loss": 0.0107, "step": 10093 }, { "epoch": 0.5381457589166712, "grad_norm": 1.7835133075714111, "learning_rate": 7.9814e-06, "loss": 0.0356, "step": 10094 }, { "epoch": 0.538199072346324, "grad_norm": 1.568657398223877, "learning_rate": 7.9812e-06, "loss": 0.0166, "step": 10095 }, { "epoch": 0.538252385775977, "grad_norm": 10.135357856750488, "learning_rate": 7.981e-06, "loss": 0.0908, "step": 10096 }, { "epoch": 0.5383056992056299, "grad_norm": 4.842031478881836, "learning_rate": 7.980800000000002e-06, "loss": 0.0824, "step": 10097 }, { "epoch": 0.5383590126352829, "grad_norm": 5.4182963371276855, "learning_rate": 7.980600000000001e-06, "loss": 0.043, "step": 10098 }, { "epoch": 0.5384123260649357, "grad_norm": 3.9575459957122803, "learning_rate": 7.9804e-06, "loss": 0.0328, "step": 10099 }, { "epoch": 0.5384656394945887, "grad_norm": 3.045300006866455, "learning_rate": 7.9802e-06, "loss": 0.046, "step": 10100 }, { "epoch": 0.5385189529242416, "grad_norm": 1.8782517910003662, "learning_rate": 7.980000000000002e-06, "loss": 0.0498, "step": 10101 }, { "epoch": 0.5385722663538945, "grad_norm": 1.3971689939498901, "learning_rate": 7.979800000000001e-06, "loss": 0.0236, "step": 10102 }, { "epoch": 0.5386255797835475, "grad_norm": 2.5117452144622803, "learning_rate": 7.9796e-06, "loss": 0.0542, "step": 10103 }, { "epoch": 0.5386788932132004, "grad_norm": 2.4217164516448975, "learning_rate": 7.9794e-06, "loss": 0.03, "step": 10104 }, { "epoch": 0.5387322066428534, "grad_norm": 2.541832685470581, "learning_rate": 7.9792e-06, "loss": 0.0482, "step": 10105 }, { "epoch": 0.5387855200725062, "grad_norm": 3.6562814712524414, "learning_rate": 7.979000000000001e-06, "loss": 0.0366, "step": 10106 }, { "epoch": 0.5388388335021592, "grad_norm": 2.7959399223327637, "learning_rate": 7.9788e-06, "loss": 0.0428, "step": 10107 }, { "epoch": 0.5388921469318121, "grad_norm": 2.245872735977173, "learning_rate": 7.9786e-06, "loss": 0.042, "step": 10108 }, { "epoch": 0.5389454603614651, "grad_norm": 2.593249797821045, "learning_rate": 7.978400000000001e-06, "loss": 0.0511, "step": 10109 }, { "epoch": 0.5389987737911179, "grad_norm": 2.3918771743774414, "learning_rate": 7.978200000000001e-06, "loss": 0.0418, "step": 10110 }, { "epoch": 0.5390520872207709, "grad_norm": 1.447704553604126, "learning_rate": 7.978e-06, "loss": 0.0359, "step": 10111 }, { "epoch": 0.5391054006504239, "grad_norm": 4.636207580566406, "learning_rate": 7.977800000000002e-06, "loss": 0.0353, "step": 10112 }, { "epoch": 0.5391587140800768, "grad_norm": 1.8140066862106323, "learning_rate": 7.977600000000001e-06, "loss": 0.0665, "step": 10113 }, { "epoch": 0.5392120275097297, "grad_norm": 2.159980297088623, "learning_rate": 7.977400000000001e-06, "loss": 0.0385, "step": 10114 }, { "epoch": 0.5392653409393826, "grad_norm": 2.3583991527557373, "learning_rate": 7.9772e-06, "loss": 0.0429, "step": 10115 }, { "epoch": 0.5393186543690356, "grad_norm": 2.2832064628601074, "learning_rate": 7.977e-06, "loss": 0.0496, "step": 10116 }, { "epoch": 0.5393719677986885, "grad_norm": 2.201895236968994, "learning_rate": 7.9768e-06, "loss": 0.0322, "step": 10117 }, { "epoch": 0.5394252812283414, "grad_norm": 1.0680633783340454, "learning_rate": 7.9766e-06, "loss": 0.0163, "step": 10118 }, { "epoch": 0.5394785946579943, "grad_norm": 3.907174587249756, "learning_rate": 7.9764e-06, "loss": 0.0307, "step": 10119 }, { "epoch": 0.5395319080876473, "grad_norm": 3.04162859916687, "learning_rate": 7.9762e-06, "loss": 0.0514, "step": 10120 }, { "epoch": 0.5395852215173003, "grad_norm": 1.6359403133392334, "learning_rate": 7.976000000000001e-06, "loss": 0.0213, "step": 10121 }, { "epoch": 0.5396385349469531, "grad_norm": 2.5746419429779053, "learning_rate": 7.9758e-06, "loss": 0.0281, "step": 10122 }, { "epoch": 0.5396918483766061, "grad_norm": 2.526419162750244, "learning_rate": 7.9756e-06, "loss": 0.0392, "step": 10123 }, { "epoch": 0.539745161806259, "grad_norm": 1.6667112112045288, "learning_rate": 7.975400000000002e-06, "loss": 0.0308, "step": 10124 }, { "epoch": 0.539798475235912, "grad_norm": 2.407972574234009, "learning_rate": 7.975200000000001e-06, "loss": 0.0324, "step": 10125 }, { "epoch": 0.5398517886655648, "grad_norm": 2.8814175128936768, "learning_rate": 7.975e-06, "loss": 0.0278, "step": 10126 }, { "epoch": 0.5399051020952178, "grad_norm": 2.4295308589935303, "learning_rate": 7.9748e-06, "loss": 0.0633, "step": 10127 }, { "epoch": 0.5399584155248707, "grad_norm": 1.9349292516708374, "learning_rate": 7.9746e-06, "loss": 0.0361, "step": 10128 }, { "epoch": 0.5400117289545237, "grad_norm": 2.028151750564575, "learning_rate": 7.974400000000001e-06, "loss": 0.0187, "step": 10129 }, { "epoch": 0.5400650423841765, "grad_norm": 3.8288893699645996, "learning_rate": 7.9742e-06, "loss": 0.0266, "step": 10130 }, { "epoch": 0.5401183558138295, "grad_norm": 4.265861988067627, "learning_rate": 7.974e-06, "loss": 0.0414, "step": 10131 }, { "epoch": 0.5401716692434825, "grad_norm": 3.309640407562256, "learning_rate": 7.9738e-06, "loss": 0.0879, "step": 10132 }, { "epoch": 0.5402249826731353, "grad_norm": 2.9380061626434326, "learning_rate": 7.973600000000001e-06, "loss": 0.0598, "step": 10133 }, { "epoch": 0.5402782961027883, "grad_norm": 3.026832103729248, "learning_rate": 7.9734e-06, "loss": 0.0498, "step": 10134 }, { "epoch": 0.5403316095324412, "grad_norm": 2.941025733947754, "learning_rate": 7.9732e-06, "loss": 0.0228, "step": 10135 }, { "epoch": 0.5403849229620942, "grad_norm": 1.3974841833114624, "learning_rate": 7.973000000000001e-06, "loss": 0.0375, "step": 10136 }, { "epoch": 0.540438236391747, "grad_norm": 1.2765684127807617, "learning_rate": 7.972800000000001e-06, "loss": 0.0109, "step": 10137 }, { "epoch": 0.5404915498214, "grad_norm": 1.713598608970642, "learning_rate": 7.9726e-06, "loss": 0.017, "step": 10138 }, { "epoch": 0.5405448632510529, "grad_norm": 1.6710842847824097, "learning_rate": 7.9724e-06, "loss": 0.0206, "step": 10139 }, { "epoch": 0.5405981766807059, "grad_norm": 2.61196231842041, "learning_rate": 7.972200000000001e-06, "loss": 0.0275, "step": 10140 }, { "epoch": 0.5406514901103588, "grad_norm": 2.375898838043213, "learning_rate": 7.972000000000001e-06, "loss": 0.021, "step": 10141 }, { "epoch": 0.5407048035400117, "grad_norm": 2.09269118309021, "learning_rate": 7.9718e-06, "loss": 0.0405, "step": 10142 }, { "epoch": 0.5407581169696647, "grad_norm": 6.8060712814331055, "learning_rate": 7.9716e-06, "loss": 0.0743, "step": 10143 }, { "epoch": 0.5408114303993176, "grad_norm": 2.2935070991516113, "learning_rate": 7.9714e-06, "loss": 0.0355, "step": 10144 }, { "epoch": 0.5408647438289705, "grad_norm": 2.014195442199707, "learning_rate": 7.971200000000001e-06, "loss": 0.0383, "step": 10145 }, { "epoch": 0.5409180572586234, "grad_norm": 1.146187424659729, "learning_rate": 7.971e-06, "loss": 0.0111, "step": 10146 }, { "epoch": 0.5409713706882764, "grad_norm": 2.034271478652954, "learning_rate": 7.9708e-06, "loss": 0.0521, "step": 10147 }, { "epoch": 0.5410246841179293, "grad_norm": 3.633934497833252, "learning_rate": 7.970600000000001e-06, "loss": 0.0841, "step": 10148 }, { "epoch": 0.5410779975475822, "grad_norm": 2.963829517364502, "learning_rate": 7.9704e-06, "loss": 0.0446, "step": 10149 }, { "epoch": 0.5411313109772352, "grad_norm": 3.339719295501709, "learning_rate": 7.9702e-06, "loss": 0.0384, "step": 10150 }, { "epoch": 0.5411846244068881, "grad_norm": 2.6613147258758545, "learning_rate": 7.970000000000002e-06, "loss": 0.0295, "step": 10151 }, { "epoch": 0.5412379378365411, "grad_norm": 1.8047884702682495, "learning_rate": 7.969800000000001e-06, "loss": 0.0193, "step": 10152 }, { "epoch": 0.5412912512661939, "grad_norm": 1.7331295013427734, "learning_rate": 7.9696e-06, "loss": 0.0247, "step": 10153 }, { "epoch": 0.5413445646958469, "grad_norm": 3.7431461811065674, "learning_rate": 7.9694e-06, "loss": 0.0468, "step": 10154 }, { "epoch": 0.5413978781254998, "grad_norm": 0.7717735171318054, "learning_rate": 7.9692e-06, "loss": 0.0075, "step": 10155 }, { "epoch": 0.5414511915551528, "grad_norm": 3.4927029609680176, "learning_rate": 7.969000000000001e-06, "loss": 0.0366, "step": 10156 }, { "epoch": 0.5415045049848056, "grad_norm": 3.6476078033447266, "learning_rate": 7.9688e-06, "loss": 0.0299, "step": 10157 }, { "epoch": 0.5415578184144586, "grad_norm": 2.914396286010742, "learning_rate": 7.9686e-06, "loss": 0.0319, "step": 10158 }, { "epoch": 0.5416111318441116, "grad_norm": 3.4074320793151855, "learning_rate": 7.9684e-06, "loss": 0.0237, "step": 10159 }, { "epoch": 0.5416644452737644, "grad_norm": 3.6299426555633545, "learning_rate": 7.968200000000001e-06, "loss": 0.0313, "step": 10160 }, { "epoch": 0.5417177587034174, "grad_norm": 2.966679811477661, "learning_rate": 7.968e-06, "loss": 0.0587, "step": 10161 }, { "epoch": 0.5417710721330703, "grad_norm": 0.7322874069213867, "learning_rate": 7.967800000000002e-06, "loss": 0.0069, "step": 10162 }, { "epoch": 0.5418243855627233, "grad_norm": 1.1980358362197876, "learning_rate": 7.967600000000001e-06, "loss": 0.0097, "step": 10163 }, { "epoch": 0.5418776989923761, "grad_norm": 3.086277484893799, "learning_rate": 7.967400000000001e-06, "loss": 0.0563, "step": 10164 }, { "epoch": 0.5419310124220291, "grad_norm": 3.9971060752868652, "learning_rate": 7.9672e-06, "loss": 0.0504, "step": 10165 }, { "epoch": 0.541984325851682, "grad_norm": 3.6086626052856445, "learning_rate": 7.967e-06, "loss": 0.0882, "step": 10166 }, { "epoch": 0.542037639281335, "grad_norm": 1.897565484046936, "learning_rate": 7.966800000000001e-06, "loss": 0.0295, "step": 10167 }, { "epoch": 0.542090952710988, "grad_norm": 1.2271463871002197, "learning_rate": 7.966600000000001e-06, "loss": 0.018, "step": 10168 }, { "epoch": 0.5421442661406408, "grad_norm": 3.668989896774292, "learning_rate": 7.9664e-06, "loss": 0.0739, "step": 10169 }, { "epoch": 0.5421975795702938, "grad_norm": 2.5947470664978027, "learning_rate": 7.9662e-06, "loss": 0.0429, "step": 10170 }, { "epoch": 0.5422508929999467, "grad_norm": 2.8793869018554688, "learning_rate": 7.966e-06, "loss": 0.0353, "step": 10171 }, { "epoch": 0.5423042064295996, "grad_norm": 2.5229296684265137, "learning_rate": 7.965800000000001e-06, "loss": 0.028, "step": 10172 }, { "epoch": 0.5423575198592525, "grad_norm": 1.4429407119750977, "learning_rate": 7.9656e-06, "loss": 0.0119, "step": 10173 }, { "epoch": 0.5424108332889055, "grad_norm": 4.295469760894775, "learning_rate": 7.965400000000002e-06, "loss": 0.0527, "step": 10174 }, { "epoch": 0.5424641467185584, "grad_norm": 1.7859714031219482, "learning_rate": 7.965200000000001e-06, "loss": 0.02, "step": 10175 }, { "epoch": 0.5425174601482113, "grad_norm": 5.113926887512207, "learning_rate": 7.965e-06, "loss": 0.0533, "step": 10176 }, { "epoch": 0.5425707735778643, "grad_norm": 1.5437990427017212, "learning_rate": 7.9648e-06, "loss": 0.0172, "step": 10177 }, { "epoch": 0.5426240870075172, "grad_norm": 2.1867141723632812, "learning_rate": 7.9646e-06, "loss": 0.0414, "step": 10178 }, { "epoch": 0.5426774004371702, "grad_norm": 1.1549755334854126, "learning_rate": 7.964400000000001e-06, "loss": 0.024, "step": 10179 }, { "epoch": 0.542730713866823, "grad_norm": 3.014561176300049, "learning_rate": 7.9642e-06, "loss": 0.0683, "step": 10180 }, { "epoch": 0.542784027296476, "grad_norm": 3.0837390422821045, "learning_rate": 7.964e-06, "loss": 0.0555, "step": 10181 }, { "epoch": 0.5428373407261289, "grad_norm": 2.4668118953704834, "learning_rate": 7.9638e-06, "loss": 0.035, "step": 10182 }, { "epoch": 0.5428906541557819, "grad_norm": 2.7897789478302, "learning_rate": 7.9636e-06, "loss": 0.0412, "step": 10183 }, { "epoch": 0.5429439675854347, "grad_norm": 3.5097153186798096, "learning_rate": 7.9634e-06, "loss": 0.0753, "step": 10184 }, { "epoch": 0.5429972810150877, "grad_norm": 2.6836564540863037, "learning_rate": 7.9632e-06, "loss": 0.0684, "step": 10185 }, { "epoch": 0.5430505944447406, "grad_norm": 2.2241199016571045, "learning_rate": 7.963000000000002e-06, "loss": 0.0236, "step": 10186 }, { "epoch": 0.5431039078743936, "grad_norm": 1.6441172361373901, "learning_rate": 7.962800000000001e-06, "loss": 0.0271, "step": 10187 }, { "epoch": 0.5431572213040465, "grad_norm": 3.3102633953094482, "learning_rate": 7.9626e-06, "loss": 0.0709, "step": 10188 }, { "epoch": 0.5432105347336994, "grad_norm": 2.7330007553100586, "learning_rate": 7.9624e-06, "loss": 0.0291, "step": 10189 }, { "epoch": 0.5432638481633524, "grad_norm": 2.2414939403533936, "learning_rate": 7.962200000000001e-06, "loss": 0.0271, "step": 10190 }, { "epoch": 0.5433171615930052, "grad_norm": 4.056769847869873, "learning_rate": 7.962000000000001e-06, "loss": 0.0525, "step": 10191 }, { "epoch": 0.5433704750226582, "grad_norm": 3.6833648681640625, "learning_rate": 7.9618e-06, "loss": 0.0428, "step": 10192 }, { "epoch": 0.5434237884523111, "grad_norm": 1.907818078994751, "learning_rate": 7.9616e-06, "loss": 0.0182, "step": 10193 }, { "epoch": 0.5434771018819641, "grad_norm": 2.389237403869629, "learning_rate": 7.9614e-06, "loss": 0.0279, "step": 10194 }, { "epoch": 0.5435304153116169, "grad_norm": 2.4518563747406006, "learning_rate": 7.961200000000001e-06, "loss": 0.056, "step": 10195 }, { "epoch": 0.5435837287412699, "grad_norm": 1.7939536571502686, "learning_rate": 7.961e-06, "loss": 0.0251, "step": 10196 }, { "epoch": 0.5436370421709229, "grad_norm": 2.9530444145202637, "learning_rate": 7.9608e-06, "loss": 0.027, "step": 10197 }, { "epoch": 0.5436903556005758, "grad_norm": 1.3767985105514526, "learning_rate": 7.960600000000001e-06, "loss": 0.0091, "step": 10198 }, { "epoch": 0.5437436690302287, "grad_norm": 2.2999649047851562, "learning_rate": 7.960400000000001e-06, "loss": 0.0339, "step": 10199 }, { "epoch": 0.5437969824598816, "grad_norm": 1.4464573860168457, "learning_rate": 7.9602e-06, "loss": 0.0284, "step": 10200 }, { "epoch": 0.5438502958895346, "grad_norm": 2.305065155029297, "learning_rate": 7.960000000000002e-06, "loss": 0.025, "step": 10201 }, { "epoch": 0.5439036093191875, "grad_norm": 1.6081806421279907, "learning_rate": 7.959800000000001e-06, "loss": 0.0473, "step": 10202 }, { "epoch": 0.5439569227488404, "grad_norm": 3.343895196914673, "learning_rate": 7.9596e-06, "loss": 0.0519, "step": 10203 }, { "epoch": 0.5440102361784933, "grad_norm": 2.7094762325286865, "learning_rate": 7.9594e-06, "loss": 0.021, "step": 10204 }, { "epoch": 0.5440635496081463, "grad_norm": 3.2860517501831055, "learning_rate": 7.9592e-06, "loss": 0.0394, "step": 10205 }, { "epoch": 0.5441168630377993, "grad_norm": 9.164619445800781, "learning_rate": 7.959000000000001e-06, "loss": 0.0782, "step": 10206 }, { "epoch": 0.5441701764674521, "grad_norm": 2.403074264526367, "learning_rate": 7.9588e-06, "loss": 0.0494, "step": 10207 }, { "epoch": 0.5442234898971051, "grad_norm": 2.8539438247680664, "learning_rate": 7.9586e-06, "loss": 0.0483, "step": 10208 }, { "epoch": 0.544276803326758, "grad_norm": 1.885091781616211, "learning_rate": 7.9584e-06, "loss": 0.0211, "step": 10209 }, { "epoch": 0.544330116756411, "grad_norm": 2.5792489051818848, "learning_rate": 7.958200000000001e-06, "loss": 0.0668, "step": 10210 }, { "epoch": 0.5443834301860638, "grad_norm": 2.2222843170166016, "learning_rate": 7.958e-06, "loss": 0.0489, "step": 10211 }, { "epoch": 0.5444367436157168, "grad_norm": 2.480433702468872, "learning_rate": 7.9578e-06, "loss": 0.0322, "step": 10212 }, { "epoch": 0.5444900570453697, "grad_norm": 2.5140459537506104, "learning_rate": 7.957600000000002e-06, "loss": 0.0524, "step": 10213 }, { "epoch": 0.5445433704750227, "grad_norm": 2.2707817554473877, "learning_rate": 7.957400000000001e-06, "loss": 0.0316, "step": 10214 }, { "epoch": 0.5445966839046756, "grad_norm": 4.2706708908081055, "learning_rate": 7.9572e-06, "loss": 0.0602, "step": 10215 }, { "epoch": 0.5446499973343285, "grad_norm": 3.5118961334228516, "learning_rate": 7.957e-06, "loss": 0.0437, "step": 10216 }, { "epoch": 0.5447033107639815, "grad_norm": 3.2172842025756836, "learning_rate": 7.956800000000001e-06, "loss": 0.0484, "step": 10217 }, { "epoch": 0.5447566241936344, "grad_norm": 2.6348562240600586, "learning_rate": 7.956600000000001e-06, "loss": 0.0343, "step": 10218 }, { "epoch": 0.5448099376232873, "grad_norm": 2.3957464694976807, "learning_rate": 7.9564e-06, "loss": 0.0494, "step": 10219 }, { "epoch": 0.5448632510529402, "grad_norm": 1.9373406171798706, "learning_rate": 7.9562e-06, "loss": 0.0247, "step": 10220 }, { "epoch": 0.5449165644825932, "grad_norm": 1.7368720769882202, "learning_rate": 7.956e-06, "loss": 0.0285, "step": 10221 }, { "epoch": 0.544969877912246, "grad_norm": 1.7649887800216675, "learning_rate": 7.955800000000001e-06, "loss": 0.0386, "step": 10222 }, { "epoch": 0.545023191341899, "grad_norm": 1.3733760118484497, "learning_rate": 7.9556e-06, "loss": 0.0287, "step": 10223 }, { "epoch": 0.545076504771552, "grad_norm": 3.7284560203552246, "learning_rate": 7.9554e-06, "loss": 0.0363, "step": 10224 }, { "epoch": 0.5451298182012049, "grad_norm": 3.019458532333374, "learning_rate": 7.955200000000001e-06, "loss": 0.0431, "step": 10225 }, { "epoch": 0.5451831316308579, "grad_norm": 2.1077017784118652, "learning_rate": 7.955000000000001e-06, "loss": 0.0526, "step": 10226 }, { "epoch": 0.5452364450605107, "grad_norm": 1.417346715927124, "learning_rate": 7.9548e-06, "loss": 0.0122, "step": 10227 }, { "epoch": 0.5452897584901637, "grad_norm": 2.1525845527648926, "learning_rate": 7.9546e-06, "loss": 0.018, "step": 10228 }, { "epoch": 0.5453430719198166, "grad_norm": 3.1458628177642822, "learning_rate": 7.954400000000001e-06, "loss": 0.0463, "step": 10229 }, { "epoch": 0.5453963853494695, "grad_norm": 1.9950801134109497, "learning_rate": 7.954200000000001e-06, "loss": 0.0341, "step": 10230 }, { "epoch": 0.5454496987791224, "grad_norm": 2.8004844188690186, "learning_rate": 7.954e-06, "loss": 0.0238, "step": 10231 }, { "epoch": 0.5455030122087754, "grad_norm": 2.7681620121002197, "learning_rate": 7.9538e-06, "loss": 0.0403, "step": 10232 }, { "epoch": 0.5455563256384284, "grad_norm": 3.710846185684204, "learning_rate": 7.9536e-06, "loss": 0.0442, "step": 10233 }, { "epoch": 0.5456096390680812, "grad_norm": 0.9534610509872437, "learning_rate": 7.9534e-06, "loss": 0.0174, "step": 10234 }, { "epoch": 0.5456629524977342, "grad_norm": 3.2025816440582275, "learning_rate": 7.9532e-06, "loss": 0.0422, "step": 10235 }, { "epoch": 0.5457162659273871, "grad_norm": 1.901772141456604, "learning_rate": 7.953e-06, "loss": 0.0378, "step": 10236 }, { "epoch": 0.5457695793570401, "grad_norm": 3.466134548187256, "learning_rate": 7.952800000000001e-06, "loss": 0.0246, "step": 10237 }, { "epoch": 0.5458228927866929, "grad_norm": 1.9491190910339355, "learning_rate": 7.9526e-06, "loss": 0.0303, "step": 10238 }, { "epoch": 0.5458762062163459, "grad_norm": 2.7493653297424316, "learning_rate": 7.9524e-06, "loss": 0.0394, "step": 10239 }, { "epoch": 0.5459295196459988, "grad_norm": 2.839376926422119, "learning_rate": 7.952200000000002e-06, "loss": 0.0367, "step": 10240 }, { "epoch": 0.5459828330756518, "grad_norm": 3.6222116947174072, "learning_rate": 7.952000000000001e-06, "loss": 0.0318, "step": 10241 }, { "epoch": 0.5460361465053046, "grad_norm": 4.218357563018799, "learning_rate": 7.9518e-06, "loss": 0.0444, "step": 10242 }, { "epoch": 0.5460894599349576, "grad_norm": 3.7527871131896973, "learning_rate": 7.9516e-06, "loss": 0.0555, "step": 10243 }, { "epoch": 0.5461427733646106, "grad_norm": 2.686239004135132, "learning_rate": 7.9514e-06, "loss": 0.0423, "step": 10244 }, { "epoch": 0.5461960867942635, "grad_norm": 4.601635456085205, "learning_rate": 7.951200000000001e-06, "loss": 0.0461, "step": 10245 }, { "epoch": 0.5462494002239164, "grad_norm": 3.686584234237671, "learning_rate": 7.951e-06, "loss": 0.0563, "step": 10246 }, { "epoch": 0.5463027136535693, "grad_norm": 3.3200066089630127, "learning_rate": 7.9508e-06, "loss": 0.0363, "step": 10247 }, { "epoch": 0.5463560270832223, "grad_norm": 1.665188193321228, "learning_rate": 7.9506e-06, "loss": 0.0302, "step": 10248 }, { "epoch": 0.5464093405128752, "grad_norm": 1.525686264038086, "learning_rate": 7.950400000000001e-06, "loss": 0.0437, "step": 10249 }, { "epoch": 0.5464626539425281, "grad_norm": 3.0138375759124756, "learning_rate": 7.9502e-06, "loss": 0.0282, "step": 10250 }, { "epoch": 0.546515967372181, "grad_norm": 2.479123115539551, "learning_rate": 7.950000000000002e-06, "loss": 0.0445, "step": 10251 }, { "epoch": 0.546569280801834, "grad_norm": 1.9051926136016846, "learning_rate": 7.949800000000001e-06, "loss": 0.0286, "step": 10252 }, { "epoch": 0.546622594231487, "grad_norm": 2.871567487716675, "learning_rate": 7.949600000000001e-06, "loss": 0.0396, "step": 10253 }, { "epoch": 0.5466759076611398, "grad_norm": 1.960391640663147, "learning_rate": 7.9494e-06, "loss": 0.0344, "step": 10254 }, { "epoch": 0.5467292210907928, "grad_norm": 2.1793031692504883, "learning_rate": 7.9492e-06, "loss": 0.0311, "step": 10255 }, { "epoch": 0.5467825345204457, "grad_norm": 1.8237988948822021, "learning_rate": 7.949000000000001e-06, "loss": 0.0389, "step": 10256 }, { "epoch": 0.5468358479500987, "grad_norm": 1.7775452136993408, "learning_rate": 7.948800000000001e-06, "loss": 0.0271, "step": 10257 }, { "epoch": 0.5468891613797515, "grad_norm": 1.8987958431243896, "learning_rate": 7.9486e-06, "loss": 0.0191, "step": 10258 }, { "epoch": 0.5469424748094045, "grad_norm": 2.7481677532196045, "learning_rate": 7.9484e-06, "loss": 0.0804, "step": 10259 }, { "epoch": 0.5469957882390574, "grad_norm": 2.779747486114502, "learning_rate": 7.9482e-06, "loss": 0.0441, "step": 10260 }, { "epoch": 0.5470491016687103, "grad_norm": 2.5087051391601562, "learning_rate": 7.948e-06, "loss": 0.0314, "step": 10261 }, { "epoch": 0.5471024150983633, "grad_norm": 1.6445568799972534, "learning_rate": 7.9478e-06, "loss": 0.0226, "step": 10262 }, { "epoch": 0.5471557285280162, "grad_norm": 5.48806095123291, "learning_rate": 7.947600000000002e-06, "loss": 0.0359, "step": 10263 }, { "epoch": 0.5472090419576692, "grad_norm": 2.335965394973755, "learning_rate": 7.947400000000001e-06, "loss": 0.0259, "step": 10264 }, { "epoch": 0.547262355387322, "grad_norm": 1.9221093654632568, "learning_rate": 7.9472e-06, "loss": 0.0399, "step": 10265 }, { "epoch": 0.547315668816975, "grad_norm": 1.541417121887207, "learning_rate": 7.947e-06, "loss": 0.0186, "step": 10266 }, { "epoch": 0.5473689822466279, "grad_norm": 1.3258932828903198, "learning_rate": 7.946800000000002e-06, "loss": 0.0186, "step": 10267 }, { "epoch": 0.5474222956762809, "grad_norm": 3.4593567848205566, "learning_rate": 7.946600000000001e-06, "loss": 0.0593, "step": 10268 }, { "epoch": 0.5474756091059337, "grad_norm": 4.02166748046875, "learning_rate": 7.9464e-06, "loss": 0.067, "step": 10269 }, { "epoch": 0.5475289225355867, "grad_norm": 1.9863290786743164, "learning_rate": 7.9462e-06, "loss": 0.0186, "step": 10270 }, { "epoch": 0.5475822359652397, "grad_norm": 4.245052337646484, "learning_rate": 7.946e-06, "loss": 0.0877, "step": 10271 }, { "epoch": 0.5476355493948926, "grad_norm": 1.6491847038269043, "learning_rate": 7.9458e-06, "loss": 0.0251, "step": 10272 }, { "epoch": 0.5476888628245455, "grad_norm": 5.494404315948486, "learning_rate": 7.9456e-06, "loss": 0.0655, "step": 10273 }, { "epoch": 0.5477421762541984, "grad_norm": 2.705254554748535, "learning_rate": 7.9454e-06, "loss": 0.0266, "step": 10274 }, { "epoch": 0.5477954896838514, "grad_norm": 2.032533645629883, "learning_rate": 7.945200000000001e-06, "loss": 0.0309, "step": 10275 }, { "epoch": 0.5478488031135043, "grad_norm": 1.4882959127426147, "learning_rate": 7.945000000000001e-06, "loss": 0.0136, "step": 10276 }, { "epoch": 0.5479021165431572, "grad_norm": 1.7772003412246704, "learning_rate": 7.9448e-06, "loss": 0.0215, "step": 10277 }, { "epoch": 0.5479554299728101, "grad_norm": 4.315176486968994, "learning_rate": 7.944600000000002e-06, "loss": 0.0212, "step": 10278 }, { "epoch": 0.5480087434024631, "grad_norm": 1.9686187505722046, "learning_rate": 7.944400000000001e-06, "loss": 0.0169, "step": 10279 }, { "epoch": 0.5480620568321161, "grad_norm": 3.3487062454223633, "learning_rate": 7.944200000000001e-06, "loss": 0.0478, "step": 10280 }, { "epoch": 0.5481153702617689, "grad_norm": 3.0869569778442383, "learning_rate": 7.944e-06, "loss": 0.05, "step": 10281 }, { "epoch": 0.5481686836914219, "grad_norm": 1.5424540042877197, "learning_rate": 7.9438e-06, "loss": 0.0137, "step": 10282 }, { "epoch": 0.5482219971210748, "grad_norm": 1.768015742301941, "learning_rate": 7.9436e-06, "loss": 0.0286, "step": 10283 }, { "epoch": 0.5482753105507278, "grad_norm": 1.255145788192749, "learning_rate": 7.943400000000001e-06, "loss": 0.0219, "step": 10284 }, { "epoch": 0.5483286239803806, "grad_norm": 2.8361644744873047, "learning_rate": 7.9432e-06, "loss": 0.0218, "step": 10285 }, { "epoch": 0.5483819374100336, "grad_norm": 4.646643161773682, "learning_rate": 7.943e-06, "loss": 0.0881, "step": 10286 }, { "epoch": 0.5484352508396865, "grad_norm": 1.735141396522522, "learning_rate": 7.942800000000001e-06, "loss": 0.039, "step": 10287 }, { "epoch": 0.5484885642693395, "grad_norm": 2.0784173011779785, "learning_rate": 7.9426e-06, "loss": 0.0291, "step": 10288 }, { "epoch": 0.5485418776989924, "grad_norm": 2.2585337162017822, "learning_rate": 7.9424e-06, "loss": 0.0434, "step": 10289 }, { "epoch": 0.5485951911286453, "grad_norm": 4.76536226272583, "learning_rate": 7.942200000000002e-06, "loss": 0.0603, "step": 10290 }, { "epoch": 0.5486485045582983, "grad_norm": 2.3104701042175293, "learning_rate": 7.942000000000001e-06, "loss": 0.0275, "step": 10291 }, { "epoch": 0.5487018179879511, "grad_norm": 3.5052154064178467, "learning_rate": 7.9418e-06, "loss": 0.0199, "step": 10292 }, { "epoch": 0.5487551314176041, "grad_norm": 4.0161542892456055, "learning_rate": 7.9416e-06, "loss": 0.0339, "step": 10293 }, { "epoch": 0.548808444847257, "grad_norm": 1.6622602939605713, "learning_rate": 7.9414e-06, "loss": 0.0312, "step": 10294 }, { "epoch": 0.54886175827691, "grad_norm": 3.6300692558288574, "learning_rate": 7.941200000000001e-06, "loss": 0.0703, "step": 10295 }, { "epoch": 0.5489150717065628, "grad_norm": 3.124528646469116, "learning_rate": 7.941e-06, "loss": 0.0278, "step": 10296 }, { "epoch": 0.5489683851362158, "grad_norm": 2.4281201362609863, "learning_rate": 7.9408e-06, "loss": 0.0608, "step": 10297 }, { "epoch": 0.5490216985658687, "grad_norm": 3.1436524391174316, "learning_rate": 7.9406e-06, "loss": 0.0442, "step": 10298 }, { "epoch": 0.5490750119955217, "grad_norm": 3.0664865970611572, "learning_rate": 7.940400000000001e-06, "loss": 0.0413, "step": 10299 }, { "epoch": 0.5491283254251746, "grad_norm": 0.7479102611541748, "learning_rate": 7.9402e-06, "loss": 0.0087, "step": 10300 }, { "epoch": 0.5491816388548275, "grad_norm": 4.150445461273193, "learning_rate": 7.94e-06, "loss": 0.0457, "step": 10301 }, { "epoch": 0.5492349522844805, "grad_norm": 2.6706135272979736, "learning_rate": 7.939800000000001e-06, "loss": 0.0541, "step": 10302 }, { "epoch": 0.5492882657141334, "grad_norm": 2.3740553855895996, "learning_rate": 7.939600000000001e-06, "loss": 0.0477, "step": 10303 }, { "epoch": 0.5493415791437863, "grad_norm": 1.972792625427246, "learning_rate": 7.9394e-06, "loss": 0.0328, "step": 10304 }, { "epoch": 0.5493948925734392, "grad_norm": 2.1389777660369873, "learning_rate": 7.9392e-06, "loss": 0.0354, "step": 10305 }, { "epoch": 0.5494482060030922, "grad_norm": 3.609215497970581, "learning_rate": 7.939000000000001e-06, "loss": 0.043, "step": 10306 }, { "epoch": 0.5495015194327451, "grad_norm": 2.845453977584839, "learning_rate": 7.938800000000001e-06, "loss": 0.0395, "step": 10307 }, { "epoch": 0.549554832862398, "grad_norm": 2.810492753982544, "learning_rate": 7.9386e-06, "loss": 0.0239, "step": 10308 }, { "epoch": 0.549608146292051, "grad_norm": 2.473342180252075, "learning_rate": 7.9384e-06, "loss": 0.0343, "step": 10309 }, { "epoch": 0.5496614597217039, "grad_norm": 1.713112235069275, "learning_rate": 7.9382e-06, "loss": 0.0259, "step": 10310 }, { "epoch": 0.5497147731513569, "grad_norm": 3.8350887298583984, "learning_rate": 7.938000000000001e-06, "loss": 0.0442, "step": 10311 }, { "epoch": 0.5497680865810097, "grad_norm": 1.9918084144592285, "learning_rate": 7.9378e-06, "loss": 0.0406, "step": 10312 }, { "epoch": 0.5498214000106627, "grad_norm": 1.3740817308425903, "learning_rate": 7.9376e-06, "loss": 0.0168, "step": 10313 }, { "epoch": 0.5498747134403156, "grad_norm": 2.5710954666137695, "learning_rate": 7.937400000000001e-06, "loss": 0.0482, "step": 10314 }, { "epoch": 0.5499280268699686, "grad_norm": 2.84896183013916, "learning_rate": 7.937200000000001e-06, "loss": 0.0313, "step": 10315 }, { "epoch": 0.5499813402996214, "grad_norm": 1.4906275272369385, "learning_rate": 7.937e-06, "loss": 0.0286, "step": 10316 }, { "epoch": 0.5500346537292744, "grad_norm": 2.979787826538086, "learning_rate": 7.936800000000002e-06, "loss": 0.0307, "step": 10317 }, { "epoch": 0.5500879671589274, "grad_norm": 1.8714933395385742, "learning_rate": 7.936600000000001e-06, "loss": 0.0509, "step": 10318 }, { "epoch": 0.5501412805885803, "grad_norm": 1.945353627204895, "learning_rate": 7.9364e-06, "loss": 0.0317, "step": 10319 }, { "epoch": 0.5501945940182332, "grad_norm": 0.7791306972503662, "learning_rate": 7.9362e-06, "loss": 0.0105, "step": 10320 }, { "epoch": 0.5502479074478861, "grad_norm": 6.454505443572998, "learning_rate": 7.936e-06, "loss": 0.0622, "step": 10321 }, { "epoch": 0.5503012208775391, "grad_norm": 1.651558756828308, "learning_rate": 7.9358e-06, "loss": 0.0238, "step": 10322 }, { "epoch": 0.550354534307192, "grad_norm": 2.528259038925171, "learning_rate": 7.9356e-06, "loss": 0.0345, "step": 10323 }, { "epoch": 0.5504078477368449, "grad_norm": 1.6453148126602173, "learning_rate": 7.9354e-06, "loss": 0.0312, "step": 10324 }, { "epoch": 0.5504611611664978, "grad_norm": 1.557350993156433, "learning_rate": 7.9352e-06, "loss": 0.0214, "step": 10325 }, { "epoch": 0.5505144745961508, "grad_norm": 2.1789443492889404, "learning_rate": 7.935000000000001e-06, "loss": 0.0358, "step": 10326 }, { "epoch": 0.5505677880258038, "grad_norm": 2.917949676513672, "learning_rate": 7.9348e-06, "loss": 0.0522, "step": 10327 }, { "epoch": 0.5506211014554566, "grad_norm": 3.726477861404419, "learning_rate": 7.934600000000002e-06, "loss": 0.0358, "step": 10328 }, { "epoch": 0.5506744148851096, "grad_norm": 1.9293267726898193, "learning_rate": 7.934400000000001e-06, "loss": 0.0364, "step": 10329 }, { "epoch": 0.5507277283147625, "grad_norm": 1.5376596450805664, "learning_rate": 7.934200000000001e-06, "loss": 0.0354, "step": 10330 }, { "epoch": 0.5507810417444154, "grad_norm": 1.066602110862732, "learning_rate": 7.934e-06, "loss": 0.0146, "step": 10331 }, { "epoch": 0.5508343551740683, "grad_norm": 1.8139511346817017, "learning_rate": 7.9338e-06, "loss": 0.0213, "step": 10332 }, { "epoch": 0.5508876686037213, "grad_norm": 2.0661895275115967, "learning_rate": 7.9336e-06, "loss": 0.0274, "step": 10333 }, { "epoch": 0.5509409820333742, "grad_norm": 3.072312116622925, "learning_rate": 7.933400000000001e-06, "loss": 0.0302, "step": 10334 }, { "epoch": 0.5509942954630271, "grad_norm": 2.2326841354370117, "learning_rate": 7.9332e-06, "loss": 0.059, "step": 10335 }, { "epoch": 0.5510476088926801, "grad_norm": 2.095388412475586, "learning_rate": 7.933e-06, "loss": 0.0349, "step": 10336 }, { "epoch": 0.551100922322333, "grad_norm": 1.9867517948150635, "learning_rate": 7.9328e-06, "loss": 0.0505, "step": 10337 }, { "epoch": 0.551154235751986, "grad_norm": 1.066158413887024, "learning_rate": 7.932600000000001e-06, "loss": 0.0231, "step": 10338 }, { "epoch": 0.5512075491816388, "grad_norm": 2.450425624847412, "learning_rate": 7.9324e-06, "loss": 0.0267, "step": 10339 }, { "epoch": 0.5512608626112918, "grad_norm": 4.239301681518555, "learning_rate": 7.932200000000002e-06, "loss": 0.0863, "step": 10340 }, { "epoch": 0.5513141760409447, "grad_norm": 1.884633183479309, "learning_rate": 7.932000000000001e-06, "loss": 0.0264, "step": 10341 }, { "epoch": 0.5513674894705977, "grad_norm": 4.98289155960083, "learning_rate": 7.931800000000001e-06, "loss": 0.0462, "step": 10342 }, { "epoch": 0.5514208029002505, "grad_norm": 1.5265439748764038, "learning_rate": 7.9316e-06, "loss": 0.0122, "step": 10343 }, { "epoch": 0.5514741163299035, "grad_norm": 5.558093070983887, "learning_rate": 7.9314e-06, "loss": 0.0476, "step": 10344 }, { "epoch": 0.5515274297595565, "grad_norm": 5.363339424133301, "learning_rate": 7.931200000000001e-06, "loss": 0.0794, "step": 10345 }, { "epoch": 0.5515807431892094, "grad_norm": 4.971897125244141, "learning_rate": 7.931e-06, "loss": 0.0517, "step": 10346 }, { "epoch": 0.5516340566188623, "grad_norm": 1.9828842878341675, "learning_rate": 7.9308e-06, "loss": 0.0265, "step": 10347 }, { "epoch": 0.5516873700485152, "grad_norm": 2.1174633502960205, "learning_rate": 7.9306e-06, "loss": 0.0381, "step": 10348 }, { "epoch": 0.5517406834781682, "grad_norm": 3.274106979370117, "learning_rate": 7.9304e-06, "loss": 0.0651, "step": 10349 }, { "epoch": 0.551793996907821, "grad_norm": 1.515688419342041, "learning_rate": 7.9302e-06, "loss": 0.0153, "step": 10350 }, { "epoch": 0.551847310337474, "grad_norm": 3.6511173248291016, "learning_rate": 7.93e-06, "loss": 0.0343, "step": 10351 }, { "epoch": 0.5519006237671269, "grad_norm": 3.005450487136841, "learning_rate": 7.929800000000002e-06, "loss": 0.0229, "step": 10352 }, { "epoch": 0.5519539371967799, "grad_norm": 3.6101057529449463, "learning_rate": 7.929600000000001e-06, "loss": 0.0557, "step": 10353 }, { "epoch": 0.5520072506264327, "grad_norm": 2.315217971801758, "learning_rate": 7.9294e-06, "loss": 0.0443, "step": 10354 }, { "epoch": 0.5520605640560857, "grad_norm": 1.790588140487671, "learning_rate": 7.9292e-06, "loss": 0.024, "step": 10355 }, { "epoch": 0.5521138774857387, "grad_norm": 2.045457601547241, "learning_rate": 7.929000000000002e-06, "loss": 0.0309, "step": 10356 }, { "epoch": 0.5521671909153916, "grad_norm": 1.3905702829360962, "learning_rate": 7.928800000000001e-06, "loss": 0.0121, "step": 10357 }, { "epoch": 0.5522205043450445, "grad_norm": 3.5023605823516846, "learning_rate": 7.9286e-06, "loss": 0.0496, "step": 10358 }, { "epoch": 0.5522738177746974, "grad_norm": 2.574428081512451, "learning_rate": 7.9284e-06, "loss": 0.0337, "step": 10359 }, { "epoch": 0.5523271312043504, "grad_norm": 2.6041181087493896, "learning_rate": 7.9282e-06, "loss": 0.0428, "step": 10360 }, { "epoch": 0.5523804446340033, "grad_norm": 3.1394708156585693, "learning_rate": 7.928e-06, "loss": 0.0517, "step": 10361 }, { "epoch": 0.5524337580636562, "grad_norm": 0.7948697805404663, "learning_rate": 7.9278e-06, "loss": 0.0081, "step": 10362 }, { "epoch": 0.5524870714933091, "grad_norm": 1.1548140048980713, "learning_rate": 7.9276e-06, "loss": 0.0068, "step": 10363 }, { "epoch": 0.5525403849229621, "grad_norm": 2.671581983566284, "learning_rate": 7.927400000000001e-06, "loss": 0.0466, "step": 10364 }, { "epoch": 0.5525936983526151, "grad_norm": 2.947840929031372, "learning_rate": 7.927200000000001e-06, "loss": 0.0629, "step": 10365 }, { "epoch": 0.5526470117822679, "grad_norm": 1.7064040899276733, "learning_rate": 7.927e-06, "loss": 0.0204, "step": 10366 }, { "epoch": 0.5527003252119209, "grad_norm": 1.708300232887268, "learning_rate": 7.926800000000002e-06, "loss": 0.0114, "step": 10367 }, { "epoch": 0.5527536386415738, "grad_norm": 2.8227012157440186, "learning_rate": 7.926600000000001e-06, "loss": 0.043, "step": 10368 }, { "epoch": 0.5528069520712268, "grad_norm": 2.281822443008423, "learning_rate": 7.926400000000001e-06, "loss": 0.0369, "step": 10369 }, { "epoch": 0.5528602655008796, "grad_norm": 3.3399415016174316, "learning_rate": 7.9262e-06, "loss": 0.0311, "step": 10370 }, { "epoch": 0.5529135789305326, "grad_norm": 2.352487087249756, "learning_rate": 7.926e-06, "loss": 0.0324, "step": 10371 }, { "epoch": 0.5529668923601855, "grad_norm": 3.327472686767578, "learning_rate": 7.9258e-06, "loss": 0.0404, "step": 10372 }, { "epoch": 0.5530202057898385, "grad_norm": 2.7180747985839844, "learning_rate": 7.9256e-06, "loss": 0.044, "step": 10373 }, { "epoch": 0.5530735192194914, "grad_norm": 2.6355466842651367, "learning_rate": 7.9254e-06, "loss": 0.0513, "step": 10374 }, { "epoch": 0.5531268326491443, "grad_norm": 2.194915294647217, "learning_rate": 7.9252e-06, "loss": 0.0264, "step": 10375 }, { "epoch": 0.5531801460787973, "grad_norm": 3.0791499614715576, "learning_rate": 7.925000000000001e-06, "loss": 0.0403, "step": 10376 }, { "epoch": 0.5532334595084502, "grad_norm": 1.0413434505462646, "learning_rate": 7.9248e-06, "loss": 0.0158, "step": 10377 }, { "epoch": 0.5532867729381031, "grad_norm": 3.373427391052246, "learning_rate": 7.924600000000002e-06, "loss": 0.0414, "step": 10378 }, { "epoch": 0.553340086367756, "grad_norm": 3.0147130489349365, "learning_rate": 7.924400000000002e-06, "loss": 0.0362, "step": 10379 }, { "epoch": 0.553393399797409, "grad_norm": 1.1030460596084595, "learning_rate": 7.924200000000001e-06, "loss": 0.0075, "step": 10380 }, { "epoch": 0.5534467132270618, "grad_norm": 2.9612913131713867, "learning_rate": 7.924e-06, "loss": 0.0261, "step": 10381 }, { "epoch": 0.5535000266567148, "grad_norm": 3.5173139572143555, "learning_rate": 7.9238e-06, "loss": 0.0781, "step": 10382 }, { "epoch": 0.5535533400863678, "grad_norm": 2.9455223083496094, "learning_rate": 7.9236e-06, "loss": 0.0384, "step": 10383 }, { "epoch": 0.5536066535160207, "grad_norm": 1.1390615701675415, "learning_rate": 7.923400000000001e-06, "loss": 0.0139, "step": 10384 }, { "epoch": 0.5536599669456737, "grad_norm": 0.8585545420646667, "learning_rate": 7.9232e-06, "loss": 0.0102, "step": 10385 }, { "epoch": 0.5537132803753265, "grad_norm": 1.7680425643920898, "learning_rate": 7.923e-06, "loss": 0.0235, "step": 10386 }, { "epoch": 0.5537665938049795, "grad_norm": 2.6925933361053467, "learning_rate": 7.9228e-06, "loss": 0.0414, "step": 10387 }, { "epoch": 0.5538199072346324, "grad_norm": 1.4131468534469604, "learning_rate": 7.922600000000001e-06, "loss": 0.0124, "step": 10388 }, { "epoch": 0.5538732206642853, "grad_norm": 3.5800158977508545, "learning_rate": 7.9224e-06, "loss": 0.0268, "step": 10389 }, { "epoch": 0.5539265340939382, "grad_norm": 2.9712178707122803, "learning_rate": 7.922200000000002e-06, "loss": 0.0342, "step": 10390 }, { "epoch": 0.5539798475235912, "grad_norm": 1.7420151233673096, "learning_rate": 7.922000000000001e-06, "loss": 0.0302, "step": 10391 }, { "epoch": 0.5540331609532442, "grad_norm": 1.3739750385284424, "learning_rate": 7.921800000000001e-06, "loss": 0.0254, "step": 10392 }, { "epoch": 0.554086474382897, "grad_norm": 1.8939405679702759, "learning_rate": 7.9216e-06, "loss": 0.0257, "step": 10393 }, { "epoch": 0.55413978781255, "grad_norm": 2.741241693496704, "learning_rate": 7.9214e-06, "loss": 0.0223, "step": 10394 }, { "epoch": 0.5541931012422029, "grad_norm": 2.6008527278900146, "learning_rate": 7.921200000000001e-06, "loss": 0.0247, "step": 10395 }, { "epoch": 0.5542464146718559, "grad_norm": 3.277181625366211, "learning_rate": 7.921000000000001e-06, "loss": 0.0456, "step": 10396 }, { "epoch": 0.5542997281015087, "grad_norm": 3.1503303050994873, "learning_rate": 7.9208e-06, "loss": 0.0323, "step": 10397 }, { "epoch": 0.5543530415311617, "grad_norm": 4.288330078125, "learning_rate": 7.9206e-06, "loss": 0.0409, "step": 10398 }, { "epoch": 0.5544063549608146, "grad_norm": 3.09072208404541, "learning_rate": 7.9204e-06, "loss": 0.0194, "step": 10399 }, { "epoch": 0.5544596683904676, "grad_norm": 3.460418224334717, "learning_rate": 7.9202e-06, "loss": 0.0671, "step": 10400 }, { "epoch": 0.5545129818201205, "grad_norm": 1.39155912399292, "learning_rate": 7.92e-06, "loss": 0.0146, "step": 10401 }, { "epoch": 0.5545662952497734, "grad_norm": 3.586787223815918, "learning_rate": 7.919800000000002e-06, "loss": 0.0527, "step": 10402 }, { "epoch": 0.5546196086794264, "grad_norm": 2.224915027618408, "learning_rate": 7.919600000000001e-06, "loss": 0.0236, "step": 10403 }, { "epoch": 0.5546729221090793, "grad_norm": 3.1390790939331055, "learning_rate": 7.9194e-06, "loss": 0.0636, "step": 10404 }, { "epoch": 0.5547262355387322, "grad_norm": 2.2573189735412598, "learning_rate": 7.9192e-06, "loss": 0.0437, "step": 10405 }, { "epoch": 0.5547795489683851, "grad_norm": 3.6572883129119873, "learning_rate": 7.919000000000002e-06, "loss": 0.0276, "step": 10406 }, { "epoch": 0.5548328623980381, "grad_norm": 2.0690324306488037, "learning_rate": 7.918800000000001e-06, "loss": 0.0358, "step": 10407 }, { "epoch": 0.554886175827691, "grad_norm": 2.2426035404205322, "learning_rate": 7.9186e-06, "loss": 0.0279, "step": 10408 }, { "epoch": 0.5549394892573439, "grad_norm": 2.1386430263519287, "learning_rate": 7.9184e-06, "loss": 0.0212, "step": 10409 }, { "epoch": 0.5549928026869968, "grad_norm": 4.223090171813965, "learning_rate": 7.9182e-06, "loss": 0.0449, "step": 10410 }, { "epoch": 0.5550461161166498, "grad_norm": 4.280144691467285, "learning_rate": 7.918e-06, "loss": 0.0373, "step": 10411 }, { "epoch": 0.5550994295463028, "grad_norm": 2.923236131668091, "learning_rate": 7.9178e-06, "loss": 0.0484, "step": 10412 }, { "epoch": 0.5551527429759556, "grad_norm": 3.8588972091674805, "learning_rate": 7.9176e-06, "loss": 0.0378, "step": 10413 }, { "epoch": 0.5552060564056086, "grad_norm": 2.682990074157715, "learning_rate": 7.917400000000001e-06, "loss": 0.0219, "step": 10414 }, { "epoch": 0.5552593698352615, "grad_norm": 2.4925951957702637, "learning_rate": 7.917200000000001e-06, "loss": 0.0237, "step": 10415 }, { "epoch": 0.5553126832649145, "grad_norm": 1.285701036453247, "learning_rate": 7.917e-06, "loss": 0.0261, "step": 10416 }, { "epoch": 0.5553659966945673, "grad_norm": 2.015026092529297, "learning_rate": 7.916800000000002e-06, "loss": 0.0338, "step": 10417 }, { "epoch": 0.5554193101242203, "grad_norm": 1.4970448017120361, "learning_rate": 7.916600000000001e-06, "loss": 0.0124, "step": 10418 }, { "epoch": 0.5554726235538732, "grad_norm": 1.6702795028686523, "learning_rate": 7.916400000000001e-06, "loss": 0.0297, "step": 10419 }, { "epoch": 0.5555259369835261, "grad_norm": 1.6920413970947266, "learning_rate": 7.9162e-06, "loss": 0.027, "step": 10420 }, { "epoch": 0.5555792504131791, "grad_norm": 2.0256338119506836, "learning_rate": 7.916e-06, "loss": 0.0183, "step": 10421 }, { "epoch": 0.555632563842832, "grad_norm": 1.5727283954620361, "learning_rate": 7.9158e-06, "loss": 0.0164, "step": 10422 }, { "epoch": 0.555685877272485, "grad_norm": 2.932218074798584, "learning_rate": 7.915600000000001e-06, "loss": 0.0598, "step": 10423 }, { "epoch": 0.5557391907021378, "grad_norm": 3.586329460144043, "learning_rate": 7.9154e-06, "loss": 0.0616, "step": 10424 }, { "epoch": 0.5557925041317908, "grad_norm": 1.8892611265182495, "learning_rate": 7.9152e-06, "loss": 0.0267, "step": 10425 }, { "epoch": 0.5558458175614437, "grad_norm": 1.7729042768478394, "learning_rate": 7.915000000000001e-06, "loss": 0.0215, "step": 10426 }, { "epoch": 0.5558991309910967, "grad_norm": 2.497899055480957, "learning_rate": 7.914800000000001e-06, "loss": 0.0534, "step": 10427 }, { "epoch": 0.5559524444207495, "grad_norm": 2.9995758533477783, "learning_rate": 7.9146e-06, "loss": 0.0524, "step": 10428 }, { "epoch": 0.5560057578504025, "grad_norm": 2.056553363800049, "learning_rate": 7.914400000000002e-06, "loss": 0.0457, "step": 10429 }, { "epoch": 0.5560590712800555, "grad_norm": 2.459404945373535, "learning_rate": 7.914200000000001e-06, "loss": 0.06, "step": 10430 }, { "epoch": 0.5561123847097084, "grad_norm": 3.935002088546753, "learning_rate": 7.914e-06, "loss": 0.0322, "step": 10431 }, { "epoch": 0.5561656981393613, "grad_norm": 4.732156276702881, "learning_rate": 7.9138e-06, "loss": 0.0461, "step": 10432 }, { "epoch": 0.5562190115690142, "grad_norm": 3.594132661819458, "learning_rate": 7.9136e-06, "loss": 0.0354, "step": 10433 }, { "epoch": 0.5562723249986672, "grad_norm": 1.1785290241241455, "learning_rate": 7.913400000000001e-06, "loss": 0.0135, "step": 10434 }, { "epoch": 0.5563256384283201, "grad_norm": 2.7926101684570312, "learning_rate": 7.9132e-06, "loss": 0.0185, "step": 10435 }, { "epoch": 0.556378951857973, "grad_norm": 2.642540693283081, "learning_rate": 7.913e-06, "loss": 0.0309, "step": 10436 }, { "epoch": 0.5564322652876259, "grad_norm": 1.2387213706970215, "learning_rate": 7.9128e-06, "loss": 0.0168, "step": 10437 }, { "epoch": 0.5564855787172789, "grad_norm": 4.250747203826904, "learning_rate": 7.912600000000001e-06, "loss": 0.0541, "step": 10438 }, { "epoch": 0.5565388921469319, "grad_norm": 3.8128888607025146, "learning_rate": 7.9124e-06, "loss": 0.0587, "step": 10439 }, { "epoch": 0.5565922055765847, "grad_norm": 3.0044169425964355, "learning_rate": 7.9122e-06, "loss": 0.0335, "step": 10440 }, { "epoch": 0.5566455190062377, "grad_norm": 2.237771987915039, "learning_rate": 7.912000000000001e-06, "loss": 0.0485, "step": 10441 }, { "epoch": 0.5566988324358906, "grad_norm": 1.6844124794006348, "learning_rate": 7.911800000000001e-06, "loss": 0.0181, "step": 10442 }, { "epoch": 0.5567521458655436, "grad_norm": 1.6743338108062744, "learning_rate": 7.9116e-06, "loss": 0.0241, "step": 10443 }, { "epoch": 0.5568054592951964, "grad_norm": 2.4160523414611816, "learning_rate": 7.9114e-06, "loss": 0.0241, "step": 10444 }, { "epoch": 0.5568587727248494, "grad_norm": 5.122101783752441, "learning_rate": 7.911200000000001e-06, "loss": 0.0429, "step": 10445 }, { "epoch": 0.5569120861545023, "grad_norm": 1.7090270519256592, "learning_rate": 7.911000000000001e-06, "loss": 0.0196, "step": 10446 }, { "epoch": 0.5569653995841553, "grad_norm": 4.594321250915527, "learning_rate": 7.9108e-06, "loss": 0.0496, "step": 10447 }, { "epoch": 0.5570187130138082, "grad_norm": 4.4920806884765625, "learning_rate": 7.9106e-06, "loss": 0.0898, "step": 10448 }, { "epoch": 0.5570720264434611, "grad_norm": 2.7338147163391113, "learning_rate": 7.9104e-06, "loss": 0.0705, "step": 10449 }, { "epoch": 0.5571253398731141, "grad_norm": 3.934821844100952, "learning_rate": 7.910200000000001e-06, "loss": 0.0367, "step": 10450 }, { "epoch": 0.557178653302767, "grad_norm": 1.264529824256897, "learning_rate": 7.91e-06, "loss": 0.0162, "step": 10451 }, { "epoch": 0.5572319667324199, "grad_norm": 2.601306200027466, "learning_rate": 7.9098e-06, "loss": 0.0593, "step": 10452 }, { "epoch": 0.5572852801620728, "grad_norm": 3.0681605339050293, "learning_rate": 7.909600000000001e-06, "loss": 0.0346, "step": 10453 }, { "epoch": 0.5573385935917258, "grad_norm": 1.782678246498108, "learning_rate": 7.909400000000001e-06, "loss": 0.0336, "step": 10454 }, { "epoch": 0.5573919070213786, "grad_norm": 3.708057165145874, "learning_rate": 7.9092e-06, "loss": 0.025, "step": 10455 }, { "epoch": 0.5574452204510316, "grad_norm": 2.8634135723114014, "learning_rate": 7.909000000000002e-06, "loss": 0.041, "step": 10456 }, { "epoch": 0.5574985338806846, "grad_norm": 3.847321033477783, "learning_rate": 7.908800000000001e-06, "loss": 0.0417, "step": 10457 }, { "epoch": 0.5575518473103375, "grad_norm": 3.8076140880584717, "learning_rate": 7.9086e-06, "loss": 0.0458, "step": 10458 }, { "epoch": 0.5576051607399904, "grad_norm": 1.0379072427749634, "learning_rate": 7.9084e-06, "loss": 0.0203, "step": 10459 }, { "epoch": 0.5576584741696433, "grad_norm": 2.570833683013916, "learning_rate": 7.9082e-06, "loss": 0.0385, "step": 10460 }, { "epoch": 0.5577117875992963, "grad_norm": 2.7680630683898926, "learning_rate": 7.908e-06, "loss": 0.0373, "step": 10461 }, { "epoch": 0.5577651010289492, "grad_norm": 4.170571804046631, "learning_rate": 7.9078e-06, "loss": 0.0949, "step": 10462 }, { "epoch": 0.5578184144586021, "grad_norm": 1.9650379419326782, "learning_rate": 7.9076e-06, "loss": 0.0314, "step": 10463 }, { "epoch": 0.557871727888255, "grad_norm": 2.1253902912139893, "learning_rate": 7.9074e-06, "loss": 0.0446, "step": 10464 }, { "epoch": 0.557925041317908, "grad_norm": 1.6072319746017456, "learning_rate": 7.907200000000001e-06, "loss": 0.0197, "step": 10465 }, { "epoch": 0.5579783547475609, "grad_norm": 1.1017320156097412, "learning_rate": 7.907e-06, "loss": 0.0252, "step": 10466 }, { "epoch": 0.5580316681772138, "grad_norm": 2.852654218673706, "learning_rate": 7.906800000000002e-06, "loss": 0.052, "step": 10467 }, { "epoch": 0.5580849816068668, "grad_norm": 1.7221401929855347, "learning_rate": 7.906600000000002e-06, "loss": 0.0358, "step": 10468 }, { "epoch": 0.5581382950365197, "grad_norm": 2.3502936363220215, "learning_rate": 7.906400000000001e-06, "loss": 0.0738, "step": 10469 }, { "epoch": 0.5581916084661727, "grad_norm": 1.5448487997055054, "learning_rate": 7.9062e-06, "loss": 0.0282, "step": 10470 }, { "epoch": 0.5582449218958255, "grad_norm": 2.5842413902282715, "learning_rate": 7.906e-06, "loss": 0.0434, "step": 10471 }, { "epoch": 0.5582982353254785, "grad_norm": 2.1025073528289795, "learning_rate": 7.9058e-06, "loss": 0.0341, "step": 10472 }, { "epoch": 0.5583515487551314, "grad_norm": 2.193530559539795, "learning_rate": 7.905600000000001e-06, "loss": 0.0343, "step": 10473 }, { "epoch": 0.5584048621847844, "grad_norm": 2.2404849529266357, "learning_rate": 7.9054e-06, "loss": 0.0469, "step": 10474 }, { "epoch": 0.5584581756144372, "grad_norm": 1.0722672939300537, "learning_rate": 7.9052e-06, "loss": 0.0171, "step": 10475 }, { "epoch": 0.5585114890440902, "grad_norm": 2.441887855529785, "learning_rate": 7.905e-06, "loss": 0.0589, "step": 10476 }, { "epoch": 0.5585648024737432, "grad_norm": 3.616325616836548, "learning_rate": 7.904800000000001e-06, "loss": 0.0511, "step": 10477 }, { "epoch": 0.558618115903396, "grad_norm": 1.8428726196289062, "learning_rate": 7.9046e-06, "loss": 0.0325, "step": 10478 }, { "epoch": 0.558671429333049, "grad_norm": 5.524747848510742, "learning_rate": 7.904400000000002e-06, "loss": 0.0585, "step": 10479 }, { "epoch": 0.5587247427627019, "grad_norm": 2.2213220596313477, "learning_rate": 7.904200000000001e-06, "loss": 0.0263, "step": 10480 }, { "epoch": 0.5587780561923549, "grad_norm": 3.2051429748535156, "learning_rate": 7.904000000000001e-06, "loss": 0.0404, "step": 10481 }, { "epoch": 0.5588313696220077, "grad_norm": 1.3012279272079468, "learning_rate": 7.9038e-06, "loss": 0.0164, "step": 10482 }, { "epoch": 0.5588846830516607, "grad_norm": 1.367978572845459, "learning_rate": 7.9036e-06, "loss": 0.0231, "step": 10483 }, { "epoch": 0.5589379964813136, "grad_norm": 1.7412810325622559, "learning_rate": 7.903400000000001e-06, "loss": 0.0201, "step": 10484 }, { "epoch": 0.5589913099109666, "grad_norm": 1.5474777221679688, "learning_rate": 7.9032e-06, "loss": 0.0283, "step": 10485 }, { "epoch": 0.5590446233406196, "grad_norm": 2.775513172149658, "learning_rate": 7.903e-06, "loss": 0.0513, "step": 10486 }, { "epoch": 0.5590979367702724, "grad_norm": 3.710859537124634, "learning_rate": 7.9028e-06, "loss": 0.0595, "step": 10487 }, { "epoch": 0.5591512501999254, "grad_norm": 2.686570167541504, "learning_rate": 7.9026e-06, "loss": 0.0389, "step": 10488 }, { "epoch": 0.5592045636295783, "grad_norm": 1.9188898801803589, "learning_rate": 7.9024e-06, "loss": 0.0345, "step": 10489 }, { "epoch": 0.5592578770592312, "grad_norm": 1.9772390127182007, "learning_rate": 7.9022e-06, "loss": 0.03, "step": 10490 }, { "epoch": 0.5593111904888841, "grad_norm": 4.308084964752197, "learning_rate": 7.902000000000002e-06, "loss": 0.046, "step": 10491 }, { "epoch": 0.5593645039185371, "grad_norm": 1.3480113744735718, "learning_rate": 7.901800000000001e-06, "loss": 0.0242, "step": 10492 }, { "epoch": 0.55941781734819, "grad_norm": 3.4308574199676514, "learning_rate": 7.9016e-06, "loss": 0.0368, "step": 10493 }, { "epoch": 0.5594711307778429, "grad_norm": 2.079709768295288, "learning_rate": 7.9014e-06, "loss": 0.0314, "step": 10494 }, { "epoch": 0.5595244442074959, "grad_norm": 3.245985507965088, "learning_rate": 7.901200000000002e-06, "loss": 0.0317, "step": 10495 }, { "epoch": 0.5595777576371488, "grad_norm": 2.9692575931549072, "learning_rate": 7.901000000000001e-06, "loss": 0.0708, "step": 10496 }, { "epoch": 0.5596310710668018, "grad_norm": 1.3229767084121704, "learning_rate": 7.9008e-06, "loss": 0.0112, "step": 10497 }, { "epoch": 0.5596843844964546, "grad_norm": 2.0960803031921387, "learning_rate": 7.9006e-06, "loss": 0.0445, "step": 10498 }, { "epoch": 0.5597376979261076, "grad_norm": 0.9467750191688538, "learning_rate": 7.9004e-06, "loss": 0.0105, "step": 10499 }, { "epoch": 0.5597910113557605, "grad_norm": 2.9930500984191895, "learning_rate": 7.9002e-06, "loss": 0.0327, "step": 10500 }, { "epoch": 0.5598443247854135, "grad_norm": 1.9092416763305664, "learning_rate": 7.9e-06, "loss": 0.0289, "step": 10501 }, { "epoch": 0.5598976382150663, "grad_norm": 1.4707908630371094, "learning_rate": 7.8998e-06, "loss": 0.0173, "step": 10502 }, { "epoch": 0.5599509516447193, "grad_norm": 2.3222739696502686, "learning_rate": 7.899600000000001e-06, "loss": 0.0306, "step": 10503 }, { "epoch": 0.5600042650743723, "grad_norm": 2.588547945022583, "learning_rate": 7.899400000000001e-06, "loss": 0.0293, "step": 10504 }, { "epoch": 0.5600575785040252, "grad_norm": 1.8881161212921143, "learning_rate": 7.8992e-06, "loss": 0.0227, "step": 10505 }, { "epoch": 0.5601108919336781, "grad_norm": 5.866500377655029, "learning_rate": 7.899000000000002e-06, "loss": 0.0679, "step": 10506 }, { "epoch": 0.560164205363331, "grad_norm": 1.5726088285446167, "learning_rate": 7.898800000000001e-06, "loss": 0.0101, "step": 10507 }, { "epoch": 0.560217518792984, "grad_norm": 1.5866667032241821, "learning_rate": 7.898600000000001e-06, "loss": 0.0189, "step": 10508 }, { "epoch": 0.5602708322226368, "grad_norm": 3.2898452281951904, "learning_rate": 7.8984e-06, "loss": 0.0201, "step": 10509 }, { "epoch": 0.5603241456522898, "grad_norm": 2.4349441528320312, "learning_rate": 7.8982e-06, "loss": 0.0502, "step": 10510 }, { "epoch": 0.5603774590819427, "grad_norm": 2.5576884746551514, "learning_rate": 7.898e-06, "loss": 0.037, "step": 10511 }, { "epoch": 0.5604307725115957, "grad_norm": 2.950251817703247, "learning_rate": 7.897800000000001e-06, "loss": 0.0228, "step": 10512 }, { "epoch": 0.5604840859412487, "grad_norm": 1.9127446413040161, "learning_rate": 7.8976e-06, "loss": 0.0209, "step": 10513 }, { "epoch": 0.5605373993709015, "grad_norm": 3.8664441108703613, "learning_rate": 7.8974e-06, "loss": 0.0604, "step": 10514 }, { "epoch": 0.5605907128005545, "grad_norm": 2.8337924480438232, "learning_rate": 7.897200000000001e-06, "loss": 0.0389, "step": 10515 }, { "epoch": 0.5606440262302074, "grad_norm": 2.086979866027832, "learning_rate": 7.897e-06, "loss": 0.0332, "step": 10516 }, { "epoch": 0.5606973396598603, "grad_norm": 2.6218318939208984, "learning_rate": 7.8968e-06, "loss": 0.0289, "step": 10517 }, { "epoch": 0.5607506530895132, "grad_norm": 1.0306798219680786, "learning_rate": 7.896600000000002e-06, "loss": 0.0069, "step": 10518 }, { "epoch": 0.5608039665191662, "grad_norm": 2.0745437145233154, "learning_rate": 7.896400000000001e-06, "loss": 0.0146, "step": 10519 }, { "epoch": 0.5608572799488191, "grad_norm": 2.6986141204833984, "learning_rate": 7.8962e-06, "loss": 0.0321, "step": 10520 }, { "epoch": 0.560910593378472, "grad_norm": 3.4320363998413086, "learning_rate": 7.896e-06, "loss": 0.0452, "step": 10521 }, { "epoch": 0.5609639068081249, "grad_norm": 2.287391424179077, "learning_rate": 7.8958e-06, "loss": 0.0296, "step": 10522 }, { "epoch": 0.5610172202377779, "grad_norm": 1.0818763971328735, "learning_rate": 7.895600000000001e-06, "loss": 0.0083, "step": 10523 }, { "epoch": 0.5610705336674309, "grad_norm": 2.6487910747528076, "learning_rate": 7.8954e-06, "loss": 0.0352, "step": 10524 }, { "epoch": 0.5611238470970837, "grad_norm": 1.9126709699630737, "learning_rate": 7.8952e-06, "loss": 0.0189, "step": 10525 }, { "epoch": 0.5611771605267367, "grad_norm": 1.9459576606750488, "learning_rate": 7.895e-06, "loss": 0.0339, "step": 10526 }, { "epoch": 0.5612304739563896, "grad_norm": 2.6361167430877686, "learning_rate": 7.894800000000001e-06, "loss": 0.0291, "step": 10527 }, { "epoch": 0.5612837873860426, "grad_norm": 2.1329071521759033, "learning_rate": 7.8946e-06, "loss": 0.0213, "step": 10528 }, { "epoch": 0.5613371008156954, "grad_norm": 1.7015697956085205, "learning_rate": 7.8944e-06, "loss": 0.0217, "step": 10529 }, { "epoch": 0.5613904142453484, "grad_norm": 1.8346452713012695, "learning_rate": 7.894200000000001e-06, "loss": 0.023, "step": 10530 }, { "epoch": 0.5614437276750013, "grad_norm": 2.6648170948028564, "learning_rate": 7.894000000000001e-06, "loss": 0.0292, "step": 10531 }, { "epoch": 0.5614970411046543, "grad_norm": 2.6055800914764404, "learning_rate": 7.8938e-06, "loss": 0.0436, "step": 10532 }, { "epoch": 0.5615503545343072, "grad_norm": 3.3473124504089355, "learning_rate": 7.8936e-06, "loss": 0.037, "step": 10533 }, { "epoch": 0.5616036679639601, "grad_norm": 2.1393139362335205, "learning_rate": 7.893400000000001e-06, "loss": 0.0166, "step": 10534 }, { "epoch": 0.5616569813936131, "grad_norm": 0.4546482264995575, "learning_rate": 7.893200000000001e-06, "loss": 0.0048, "step": 10535 }, { "epoch": 0.561710294823266, "grad_norm": 4.370682716369629, "learning_rate": 7.893e-06, "loss": 0.037, "step": 10536 }, { "epoch": 0.5617636082529189, "grad_norm": 1.6991204023361206, "learning_rate": 7.8928e-06, "loss": 0.0181, "step": 10537 }, { "epoch": 0.5618169216825718, "grad_norm": 1.5239484310150146, "learning_rate": 7.8926e-06, "loss": 0.0271, "step": 10538 }, { "epoch": 0.5618702351122248, "grad_norm": 1.9808528423309326, "learning_rate": 7.892400000000001e-06, "loss": 0.0108, "step": 10539 }, { "epoch": 0.5619235485418776, "grad_norm": 1.6107319593429565, "learning_rate": 7.8922e-06, "loss": 0.0133, "step": 10540 }, { "epoch": 0.5619768619715306, "grad_norm": 2.1780359745025635, "learning_rate": 7.892e-06, "loss": 0.0186, "step": 10541 }, { "epoch": 0.5620301754011836, "grad_norm": 3.6073386669158936, "learning_rate": 7.891800000000001e-06, "loss": 0.0519, "step": 10542 }, { "epoch": 0.5620834888308365, "grad_norm": 2.3611626625061035, "learning_rate": 7.8916e-06, "loss": 0.0201, "step": 10543 }, { "epoch": 0.5621368022604895, "grad_norm": 4.826295375823975, "learning_rate": 7.8914e-06, "loss": 0.0562, "step": 10544 }, { "epoch": 0.5621901156901423, "grad_norm": 4.132631301879883, "learning_rate": 7.891200000000002e-06, "loss": 0.0323, "step": 10545 }, { "epoch": 0.5622434291197953, "grad_norm": 1.2856545448303223, "learning_rate": 7.891000000000001e-06, "loss": 0.0104, "step": 10546 }, { "epoch": 0.5622967425494482, "grad_norm": 1.488101840019226, "learning_rate": 7.8908e-06, "loss": 0.0298, "step": 10547 }, { "epoch": 0.5623500559791011, "grad_norm": 2.741823434829712, "learning_rate": 7.8906e-06, "loss": 0.0401, "step": 10548 }, { "epoch": 0.562403369408754, "grad_norm": 1.2030270099639893, "learning_rate": 7.8904e-06, "loss": 0.0246, "step": 10549 }, { "epoch": 0.562456682838407, "grad_norm": 1.4547622203826904, "learning_rate": 7.890200000000001e-06, "loss": 0.0239, "step": 10550 }, { "epoch": 0.56250999626806, "grad_norm": 2.1545262336730957, "learning_rate": 7.89e-06, "loss": 0.0147, "step": 10551 }, { "epoch": 0.5625633096977128, "grad_norm": 3.140350103378296, "learning_rate": 7.8898e-06, "loss": 0.0374, "step": 10552 }, { "epoch": 0.5626166231273658, "grad_norm": 3.6699392795562744, "learning_rate": 7.8896e-06, "loss": 0.049, "step": 10553 }, { "epoch": 0.5626699365570187, "grad_norm": 0.8968491554260254, "learning_rate": 7.889400000000001e-06, "loss": 0.0246, "step": 10554 }, { "epoch": 0.5627232499866717, "grad_norm": 2.397118091583252, "learning_rate": 7.8892e-06, "loss": 0.0266, "step": 10555 }, { "epoch": 0.5627765634163245, "grad_norm": 2.5685949325561523, "learning_rate": 7.889000000000002e-06, "loss": 0.0382, "step": 10556 }, { "epoch": 0.5628298768459775, "grad_norm": 2.9835286140441895, "learning_rate": 7.888800000000001e-06, "loss": 0.0343, "step": 10557 }, { "epoch": 0.5628831902756304, "grad_norm": 5.530223846435547, "learning_rate": 7.888600000000001e-06, "loss": 0.1083, "step": 10558 }, { "epoch": 0.5629365037052834, "grad_norm": 2.753401756286621, "learning_rate": 7.8884e-06, "loss": 0.0548, "step": 10559 }, { "epoch": 0.5629898171349363, "grad_norm": 6.874622344970703, "learning_rate": 7.8882e-06, "loss": 0.09, "step": 10560 }, { "epoch": 0.5630431305645892, "grad_norm": 7.681033611297607, "learning_rate": 7.888e-06, "loss": 0.0471, "step": 10561 }, { "epoch": 0.5630964439942422, "grad_norm": 3.7358591556549072, "learning_rate": 7.887800000000001e-06, "loss": 0.0674, "step": 10562 }, { "epoch": 0.5631497574238951, "grad_norm": 3.3089663982391357, "learning_rate": 7.8876e-06, "loss": 0.0858, "step": 10563 }, { "epoch": 0.563203070853548, "grad_norm": 2.2784645557403564, "learning_rate": 7.8874e-06, "loss": 0.0324, "step": 10564 }, { "epoch": 0.5632563842832009, "grad_norm": 2.0572097301483154, "learning_rate": 7.8872e-06, "loss": 0.0247, "step": 10565 }, { "epoch": 0.5633096977128539, "grad_norm": 1.8007384538650513, "learning_rate": 7.887000000000001e-06, "loss": 0.0305, "step": 10566 }, { "epoch": 0.5633630111425068, "grad_norm": 4.342413902282715, "learning_rate": 7.8868e-06, "loss": 0.0564, "step": 10567 }, { "epoch": 0.5634163245721597, "grad_norm": 3.0711355209350586, "learning_rate": 7.886600000000002e-06, "loss": 0.0545, "step": 10568 }, { "epoch": 0.5634696380018127, "grad_norm": 3.0874598026275635, "learning_rate": 7.886400000000001e-06, "loss": 0.0595, "step": 10569 }, { "epoch": 0.5635229514314656, "grad_norm": 3.378216028213501, "learning_rate": 7.8862e-06, "loss": 0.0693, "step": 10570 }, { "epoch": 0.5635762648611186, "grad_norm": 3.282912015914917, "learning_rate": 7.886e-06, "loss": 0.0654, "step": 10571 }, { "epoch": 0.5636295782907714, "grad_norm": 6.454256534576416, "learning_rate": 7.8858e-06, "loss": 0.0729, "step": 10572 }, { "epoch": 0.5636828917204244, "grad_norm": 2.5708231925964355, "learning_rate": 7.885600000000001e-06, "loss": 0.0389, "step": 10573 }, { "epoch": 0.5637362051500773, "grad_norm": 4.445218563079834, "learning_rate": 7.8854e-06, "loss": 0.0598, "step": 10574 }, { "epoch": 0.5637895185797303, "grad_norm": 3.196054220199585, "learning_rate": 7.8852e-06, "loss": 0.0437, "step": 10575 }, { "epoch": 0.5638428320093831, "grad_norm": 2.055938720703125, "learning_rate": 7.885e-06, "loss": 0.0378, "step": 10576 }, { "epoch": 0.5638961454390361, "grad_norm": 2.08935284614563, "learning_rate": 7.8848e-06, "loss": 0.0382, "step": 10577 }, { "epoch": 0.563949458868689, "grad_norm": 1.312900185585022, "learning_rate": 7.8846e-06, "loss": 0.0284, "step": 10578 }, { "epoch": 0.564002772298342, "grad_norm": 3.3384532928466797, "learning_rate": 7.8844e-06, "loss": 0.065, "step": 10579 }, { "epoch": 0.5640560857279949, "grad_norm": 2.4274189472198486, "learning_rate": 7.884200000000002e-06, "loss": 0.0361, "step": 10580 }, { "epoch": 0.5641093991576478, "grad_norm": 3.1963706016540527, "learning_rate": 7.884000000000001e-06, "loss": 0.0327, "step": 10581 }, { "epoch": 0.5641627125873008, "grad_norm": 3.6451914310455322, "learning_rate": 7.8838e-06, "loss": 0.0346, "step": 10582 }, { "epoch": 0.5642160260169536, "grad_norm": 1.3939017057418823, "learning_rate": 7.8836e-06, "loss": 0.0084, "step": 10583 }, { "epoch": 0.5642693394466066, "grad_norm": 2.2737958431243896, "learning_rate": 7.883400000000001e-06, "loss": 0.0348, "step": 10584 }, { "epoch": 0.5643226528762595, "grad_norm": 3.2209720611572266, "learning_rate": 7.883200000000001e-06, "loss": 0.0454, "step": 10585 }, { "epoch": 0.5643759663059125, "grad_norm": 2.057943820953369, "learning_rate": 7.883e-06, "loss": 0.0334, "step": 10586 }, { "epoch": 0.5644292797355653, "grad_norm": 3.5999152660369873, "learning_rate": 7.8828e-06, "loss": 0.0568, "step": 10587 }, { "epoch": 0.5644825931652183, "grad_norm": 1.94642972946167, "learning_rate": 7.8826e-06, "loss": 0.0282, "step": 10588 }, { "epoch": 0.5645359065948713, "grad_norm": 3.1764941215515137, "learning_rate": 7.882400000000001e-06, "loss": 0.0458, "step": 10589 }, { "epoch": 0.5645892200245242, "grad_norm": 3.6340696811676025, "learning_rate": 7.8822e-06, "loss": 0.0233, "step": 10590 }, { "epoch": 0.5646425334541771, "grad_norm": 4.6860671043396, "learning_rate": 7.882e-06, "loss": 0.0582, "step": 10591 }, { "epoch": 0.56469584688383, "grad_norm": 4.803212642669678, "learning_rate": 7.881800000000001e-06, "loss": 0.0653, "step": 10592 }, { "epoch": 0.564749160313483, "grad_norm": 2.453977346420288, "learning_rate": 7.881600000000001e-06, "loss": 0.0319, "step": 10593 }, { "epoch": 0.5648024737431359, "grad_norm": 4.0623321533203125, "learning_rate": 7.8814e-06, "loss": 0.0549, "step": 10594 }, { "epoch": 0.5648557871727888, "grad_norm": 3.3748035430908203, "learning_rate": 7.881200000000002e-06, "loss": 0.0412, "step": 10595 }, { "epoch": 0.5649091006024417, "grad_norm": 2.864344596862793, "learning_rate": 7.881000000000001e-06, "loss": 0.0267, "step": 10596 }, { "epoch": 0.5649624140320947, "grad_norm": 2.735724687576294, "learning_rate": 7.8808e-06, "loss": 0.0593, "step": 10597 }, { "epoch": 0.5650157274617477, "grad_norm": 3.217552661895752, "learning_rate": 7.8806e-06, "loss": 0.0641, "step": 10598 }, { "epoch": 0.5650690408914005, "grad_norm": 0.9971041679382324, "learning_rate": 7.8804e-06, "loss": 0.0089, "step": 10599 }, { "epoch": 0.5651223543210535, "grad_norm": 1.535678744316101, "learning_rate": 7.880200000000001e-06, "loss": 0.0253, "step": 10600 }, { "epoch": 0.5651756677507064, "grad_norm": 4.5267486572265625, "learning_rate": 7.88e-06, "loss": 0.0836, "step": 10601 }, { "epoch": 0.5652289811803594, "grad_norm": 2.8860325813293457, "learning_rate": 7.8798e-06, "loss": 0.0772, "step": 10602 }, { "epoch": 0.5652822946100122, "grad_norm": 2.0867116451263428, "learning_rate": 7.8796e-06, "loss": 0.0499, "step": 10603 }, { "epoch": 0.5653356080396652, "grad_norm": 2.1962931156158447, "learning_rate": 7.879400000000001e-06, "loss": 0.0215, "step": 10604 }, { "epoch": 0.5653889214693181, "grad_norm": 1.3817894458770752, "learning_rate": 7.8792e-06, "loss": 0.0323, "step": 10605 }, { "epoch": 0.565442234898971, "grad_norm": 2.2917983531951904, "learning_rate": 7.879e-06, "loss": 0.0343, "step": 10606 }, { "epoch": 0.565495548328624, "grad_norm": 2.1922390460968018, "learning_rate": 7.878800000000002e-06, "loss": 0.0323, "step": 10607 }, { "epoch": 0.5655488617582769, "grad_norm": 2.608795642852783, "learning_rate": 7.878600000000001e-06, "loss": 0.0293, "step": 10608 }, { "epoch": 0.5656021751879299, "grad_norm": 0.9591084718704224, "learning_rate": 7.8784e-06, "loss": 0.0113, "step": 10609 }, { "epoch": 0.5656554886175827, "grad_norm": 2.013218879699707, "learning_rate": 7.8782e-06, "loss": 0.0234, "step": 10610 }, { "epoch": 0.5657088020472357, "grad_norm": 2.397536516189575, "learning_rate": 7.878e-06, "loss": 0.0182, "step": 10611 }, { "epoch": 0.5657621154768886, "grad_norm": 1.4726310968399048, "learning_rate": 7.877800000000001e-06, "loss": 0.021, "step": 10612 }, { "epoch": 0.5658154289065416, "grad_norm": 3.400402069091797, "learning_rate": 7.8776e-06, "loss": 0.0226, "step": 10613 }, { "epoch": 0.5658687423361944, "grad_norm": 3.2699944972991943, "learning_rate": 7.8774e-06, "loss": 0.0391, "step": 10614 }, { "epoch": 0.5659220557658474, "grad_norm": 4.49130916595459, "learning_rate": 7.8772e-06, "loss": 0.0864, "step": 10615 }, { "epoch": 0.5659753691955004, "grad_norm": 2.305131435394287, "learning_rate": 7.877000000000001e-06, "loss": 0.063, "step": 10616 }, { "epoch": 0.5660286826251533, "grad_norm": 1.9215656518936157, "learning_rate": 7.8768e-06, "loss": 0.0269, "step": 10617 }, { "epoch": 0.5660819960548062, "grad_norm": 1.7026262283325195, "learning_rate": 7.8766e-06, "loss": 0.0241, "step": 10618 }, { "epoch": 0.5661353094844591, "grad_norm": 3.448704242706299, "learning_rate": 7.876400000000001e-06, "loss": 0.0494, "step": 10619 }, { "epoch": 0.5661886229141121, "grad_norm": 2.453238010406494, "learning_rate": 7.876200000000001e-06, "loss": 0.0225, "step": 10620 }, { "epoch": 0.566241936343765, "grad_norm": 1.745042085647583, "learning_rate": 7.876e-06, "loss": 0.0306, "step": 10621 }, { "epoch": 0.5662952497734179, "grad_norm": 3.0502665042877197, "learning_rate": 7.8758e-06, "loss": 0.0394, "step": 10622 }, { "epoch": 0.5663485632030708, "grad_norm": 2.8666720390319824, "learning_rate": 7.875600000000001e-06, "loss": 0.0372, "step": 10623 }, { "epoch": 0.5664018766327238, "grad_norm": 1.9184187650680542, "learning_rate": 7.875400000000001e-06, "loss": 0.0218, "step": 10624 }, { "epoch": 0.5664551900623768, "grad_norm": 4.15635347366333, "learning_rate": 7.8752e-06, "loss": 0.0524, "step": 10625 }, { "epoch": 0.5665085034920296, "grad_norm": 1.7189384698867798, "learning_rate": 7.875e-06, "loss": 0.0248, "step": 10626 }, { "epoch": 0.5665618169216826, "grad_norm": 2.9245786666870117, "learning_rate": 7.8748e-06, "loss": 0.032, "step": 10627 }, { "epoch": 0.5666151303513355, "grad_norm": 2.9059534072875977, "learning_rate": 7.8746e-06, "loss": 0.0549, "step": 10628 }, { "epoch": 0.5666684437809885, "grad_norm": 2.7094480991363525, "learning_rate": 7.8744e-06, "loss": 0.0144, "step": 10629 }, { "epoch": 0.5667217572106413, "grad_norm": 2.1600141525268555, "learning_rate": 7.8742e-06, "loss": 0.0259, "step": 10630 }, { "epoch": 0.5667750706402943, "grad_norm": 3.6786861419677734, "learning_rate": 7.874000000000001e-06, "loss": 0.0397, "step": 10631 }, { "epoch": 0.5668283840699472, "grad_norm": 4.242504596710205, "learning_rate": 7.8738e-06, "loss": 0.0371, "step": 10632 }, { "epoch": 0.5668816974996002, "grad_norm": 2.8424606323242188, "learning_rate": 7.8736e-06, "loss": 0.0571, "step": 10633 }, { "epoch": 0.566935010929253, "grad_norm": 3.6618404388427734, "learning_rate": 7.873400000000002e-06, "loss": 0.0671, "step": 10634 }, { "epoch": 0.566988324358906, "grad_norm": 2.0895285606384277, "learning_rate": 7.873200000000001e-06, "loss": 0.0485, "step": 10635 }, { "epoch": 0.567041637788559, "grad_norm": 1.639001727104187, "learning_rate": 7.873e-06, "loss": 0.0381, "step": 10636 }, { "epoch": 0.5670949512182119, "grad_norm": 1.6674138307571411, "learning_rate": 7.8728e-06, "loss": 0.0162, "step": 10637 }, { "epoch": 0.5671482646478648, "grad_norm": 2.051788568496704, "learning_rate": 7.8726e-06, "loss": 0.0223, "step": 10638 }, { "epoch": 0.5672015780775177, "grad_norm": 2.8163106441497803, "learning_rate": 7.872400000000001e-06, "loss": 0.0339, "step": 10639 }, { "epoch": 0.5672548915071707, "grad_norm": 2.5499894618988037, "learning_rate": 7.8722e-06, "loss": 0.0288, "step": 10640 }, { "epoch": 0.5673082049368235, "grad_norm": 1.313408374786377, "learning_rate": 7.872e-06, "loss": 0.0137, "step": 10641 }, { "epoch": 0.5673615183664765, "grad_norm": 2.1646013259887695, "learning_rate": 7.8718e-06, "loss": 0.0472, "step": 10642 }, { "epoch": 0.5674148317961294, "grad_norm": 2.5977418422698975, "learning_rate": 7.871600000000001e-06, "loss": 0.0331, "step": 10643 }, { "epoch": 0.5674681452257824, "grad_norm": 2.484513282775879, "learning_rate": 7.8714e-06, "loss": 0.0319, "step": 10644 }, { "epoch": 0.5675214586554354, "grad_norm": 2.9008522033691406, "learning_rate": 7.871200000000002e-06, "loss": 0.0255, "step": 10645 }, { "epoch": 0.5675747720850882, "grad_norm": 1.3442810773849487, "learning_rate": 7.871000000000001e-06, "loss": 0.02, "step": 10646 }, { "epoch": 0.5676280855147412, "grad_norm": 2.7349932193756104, "learning_rate": 7.870800000000001e-06, "loss": 0.027, "step": 10647 }, { "epoch": 0.5676813989443941, "grad_norm": 1.1988335847854614, "learning_rate": 7.8706e-06, "loss": 0.0184, "step": 10648 }, { "epoch": 0.567734712374047, "grad_norm": 2.965989828109741, "learning_rate": 7.8704e-06, "loss": 0.0389, "step": 10649 }, { "epoch": 0.5677880258036999, "grad_norm": 1.3696714639663696, "learning_rate": 7.870200000000001e-06, "loss": 0.0169, "step": 10650 }, { "epoch": 0.5678413392333529, "grad_norm": 1.04841947555542, "learning_rate": 7.870000000000001e-06, "loss": 0.0136, "step": 10651 }, { "epoch": 0.5678946526630058, "grad_norm": 1.8110218048095703, "learning_rate": 7.8698e-06, "loss": 0.046, "step": 10652 }, { "epoch": 0.5679479660926587, "grad_norm": 1.9317783117294312, "learning_rate": 7.8696e-06, "loss": 0.0241, "step": 10653 }, { "epoch": 0.5680012795223117, "grad_norm": 1.5094258785247803, "learning_rate": 7.8694e-06, "loss": 0.016, "step": 10654 }, { "epoch": 0.5680545929519646, "grad_norm": 3.27535343170166, "learning_rate": 7.8692e-06, "loss": 0.0338, "step": 10655 }, { "epoch": 0.5681079063816176, "grad_norm": 1.5800395011901855, "learning_rate": 7.869e-06, "loss": 0.0225, "step": 10656 }, { "epoch": 0.5681612198112704, "grad_norm": 3.1444411277770996, "learning_rate": 7.868800000000002e-06, "loss": 0.0266, "step": 10657 }, { "epoch": 0.5682145332409234, "grad_norm": 4.6460466384887695, "learning_rate": 7.868600000000001e-06, "loss": 0.0308, "step": 10658 }, { "epoch": 0.5682678466705763, "grad_norm": 3.398674726486206, "learning_rate": 7.8684e-06, "loss": 0.0271, "step": 10659 }, { "epoch": 0.5683211601002293, "grad_norm": 2.3461902141571045, "learning_rate": 7.8682e-06, "loss": 0.0263, "step": 10660 }, { "epoch": 0.5683744735298821, "grad_norm": 4.0047502517700195, "learning_rate": 7.868000000000002e-06, "loss": 0.0265, "step": 10661 }, { "epoch": 0.5684277869595351, "grad_norm": 4.414963722229004, "learning_rate": 7.867800000000001e-06, "loss": 0.0885, "step": 10662 }, { "epoch": 0.5684811003891881, "grad_norm": 2.440521717071533, "learning_rate": 7.8676e-06, "loss": 0.0171, "step": 10663 }, { "epoch": 0.568534413818841, "grad_norm": 0.7943317890167236, "learning_rate": 7.8674e-06, "loss": 0.0042, "step": 10664 }, { "epoch": 0.5685877272484939, "grad_norm": 2.5825209617614746, "learning_rate": 7.8672e-06, "loss": 0.0273, "step": 10665 }, { "epoch": 0.5686410406781468, "grad_norm": 1.1921837329864502, "learning_rate": 7.867e-06, "loss": 0.0133, "step": 10666 }, { "epoch": 0.5686943541077998, "grad_norm": 4.428239345550537, "learning_rate": 7.8668e-06, "loss": 0.0402, "step": 10667 }, { "epoch": 0.5687476675374527, "grad_norm": 2.328155517578125, "learning_rate": 7.8666e-06, "loss": 0.0373, "step": 10668 }, { "epoch": 0.5688009809671056, "grad_norm": 3.446256160736084, "learning_rate": 7.866400000000001e-06, "loss": 0.0518, "step": 10669 }, { "epoch": 0.5688542943967585, "grad_norm": 3.05521821975708, "learning_rate": 7.866200000000001e-06, "loss": 0.0252, "step": 10670 }, { "epoch": 0.5689076078264115, "grad_norm": 2.7054851055145264, "learning_rate": 7.866e-06, "loss": 0.0657, "step": 10671 }, { "epoch": 0.5689609212560645, "grad_norm": 3.037872314453125, "learning_rate": 7.8658e-06, "loss": 0.0555, "step": 10672 }, { "epoch": 0.5690142346857173, "grad_norm": 4.070452690124512, "learning_rate": 7.865600000000001e-06, "loss": 0.0292, "step": 10673 }, { "epoch": 0.5690675481153703, "grad_norm": 2.208096742630005, "learning_rate": 7.865400000000001e-06, "loss": 0.0555, "step": 10674 }, { "epoch": 0.5691208615450232, "grad_norm": 1.8902043104171753, "learning_rate": 7.8652e-06, "loss": 0.0273, "step": 10675 }, { "epoch": 0.5691741749746762, "grad_norm": 0.967769980430603, "learning_rate": 7.865e-06, "loss": 0.0103, "step": 10676 }, { "epoch": 0.569227488404329, "grad_norm": 3.4319682121276855, "learning_rate": 7.8648e-06, "loss": 0.0488, "step": 10677 }, { "epoch": 0.569280801833982, "grad_norm": 1.9986774921417236, "learning_rate": 7.864600000000001e-06, "loss": 0.025, "step": 10678 }, { "epoch": 0.5693341152636349, "grad_norm": 2.8768811225891113, "learning_rate": 7.8644e-06, "loss": 0.051, "step": 10679 }, { "epoch": 0.5693874286932878, "grad_norm": 2.646756172180176, "learning_rate": 7.8642e-06, "loss": 0.0592, "step": 10680 }, { "epoch": 0.5694407421229408, "grad_norm": 7.177898406982422, "learning_rate": 7.864000000000001e-06, "loss": 0.065, "step": 10681 }, { "epoch": 0.5694940555525937, "grad_norm": 2.45005202293396, "learning_rate": 7.8638e-06, "loss": 0.0249, "step": 10682 }, { "epoch": 0.5695473689822467, "grad_norm": 1.9379916191101074, "learning_rate": 7.8636e-06, "loss": 0.0353, "step": 10683 }, { "epoch": 0.5696006824118995, "grad_norm": 0.9416369199752808, "learning_rate": 7.863400000000002e-06, "loss": 0.0167, "step": 10684 }, { "epoch": 0.5696539958415525, "grad_norm": 1.6213799715042114, "learning_rate": 7.863200000000001e-06, "loss": 0.0368, "step": 10685 }, { "epoch": 0.5697073092712054, "grad_norm": 7.498775005340576, "learning_rate": 7.863e-06, "loss": 0.0732, "step": 10686 }, { "epoch": 0.5697606227008584, "grad_norm": 3.047438859939575, "learning_rate": 7.8628e-06, "loss": 0.0177, "step": 10687 }, { "epoch": 0.5698139361305112, "grad_norm": 3.0304698944091797, "learning_rate": 7.8626e-06, "loss": 0.0673, "step": 10688 }, { "epoch": 0.5698672495601642, "grad_norm": 2.1720829010009766, "learning_rate": 7.862400000000001e-06, "loss": 0.0408, "step": 10689 }, { "epoch": 0.5699205629898171, "grad_norm": 2.5462334156036377, "learning_rate": 7.8622e-06, "loss": 0.0362, "step": 10690 }, { "epoch": 0.5699738764194701, "grad_norm": 1.8901407718658447, "learning_rate": 7.862e-06, "loss": 0.041, "step": 10691 }, { "epoch": 0.570027189849123, "grad_norm": 1.46170175075531, "learning_rate": 7.8618e-06, "loss": 0.0178, "step": 10692 }, { "epoch": 0.5700805032787759, "grad_norm": 1.9295310974121094, "learning_rate": 7.861600000000001e-06, "loss": 0.016, "step": 10693 }, { "epoch": 0.5701338167084289, "grad_norm": 3.0472500324249268, "learning_rate": 7.8614e-06, "loss": 0.0515, "step": 10694 }, { "epoch": 0.5701871301380818, "grad_norm": 2.2511723041534424, "learning_rate": 7.8612e-06, "loss": 0.048, "step": 10695 }, { "epoch": 0.5702404435677347, "grad_norm": 1.6612147092819214, "learning_rate": 7.861000000000001e-06, "loss": 0.0181, "step": 10696 }, { "epoch": 0.5702937569973876, "grad_norm": 1.8767142295837402, "learning_rate": 7.860800000000001e-06, "loss": 0.0207, "step": 10697 }, { "epoch": 0.5703470704270406, "grad_norm": 1.243099570274353, "learning_rate": 7.8606e-06, "loss": 0.0152, "step": 10698 }, { "epoch": 0.5704003838566934, "grad_norm": 4.2247724533081055, "learning_rate": 7.8604e-06, "loss": 0.0361, "step": 10699 }, { "epoch": 0.5704536972863464, "grad_norm": 1.244107723236084, "learning_rate": 7.860200000000001e-06, "loss": 0.0144, "step": 10700 }, { "epoch": 0.5705070107159994, "grad_norm": 2.672982692718506, "learning_rate": 7.860000000000001e-06, "loss": 0.0419, "step": 10701 }, { "epoch": 0.5705603241456523, "grad_norm": 1.0682984590530396, "learning_rate": 7.8598e-06, "loss": 0.0093, "step": 10702 }, { "epoch": 0.5706136375753053, "grad_norm": 2.669853448867798, "learning_rate": 7.8596e-06, "loss": 0.0649, "step": 10703 }, { "epoch": 0.5706669510049581, "grad_norm": 2.5198750495910645, "learning_rate": 7.8594e-06, "loss": 0.0258, "step": 10704 }, { "epoch": 0.5707202644346111, "grad_norm": 1.8395392894744873, "learning_rate": 7.859200000000001e-06, "loss": 0.0481, "step": 10705 }, { "epoch": 0.570773577864264, "grad_norm": 2.4101266860961914, "learning_rate": 7.859e-06, "loss": 0.0416, "step": 10706 }, { "epoch": 0.570826891293917, "grad_norm": 3.7433390617370605, "learning_rate": 7.8588e-06, "loss": 0.0229, "step": 10707 }, { "epoch": 0.5708802047235698, "grad_norm": 2.5236411094665527, "learning_rate": 7.858600000000001e-06, "loss": 0.0298, "step": 10708 }, { "epoch": 0.5709335181532228, "grad_norm": 2.978255271911621, "learning_rate": 7.858400000000001e-06, "loss": 0.0281, "step": 10709 }, { "epoch": 0.5709868315828758, "grad_norm": 1.4353206157684326, "learning_rate": 7.8582e-06, "loss": 0.0127, "step": 10710 }, { "epoch": 0.5710401450125286, "grad_norm": 6.984438896179199, "learning_rate": 7.858000000000002e-06, "loss": 0.0662, "step": 10711 }, { "epoch": 0.5710934584421816, "grad_norm": 5.090353488922119, "learning_rate": 7.857800000000001e-06, "loss": 0.057, "step": 10712 }, { "epoch": 0.5711467718718345, "grad_norm": 1.8937904834747314, "learning_rate": 7.8576e-06, "loss": 0.0147, "step": 10713 }, { "epoch": 0.5712000853014875, "grad_norm": 1.1564557552337646, "learning_rate": 7.8574e-06, "loss": 0.0141, "step": 10714 }, { "epoch": 0.5712533987311403, "grad_norm": 2.7502174377441406, "learning_rate": 7.8572e-06, "loss": 0.0244, "step": 10715 }, { "epoch": 0.5713067121607933, "grad_norm": 2.065598726272583, "learning_rate": 7.857e-06, "loss": 0.04, "step": 10716 }, { "epoch": 0.5713600255904462, "grad_norm": 4.352067947387695, "learning_rate": 7.8568e-06, "loss": 0.0717, "step": 10717 }, { "epoch": 0.5714133390200992, "grad_norm": 1.8467278480529785, "learning_rate": 7.8566e-06, "loss": 0.0149, "step": 10718 }, { "epoch": 0.5714666524497521, "grad_norm": 3.5433261394500732, "learning_rate": 7.8564e-06, "loss": 0.0652, "step": 10719 }, { "epoch": 0.571519965879405, "grad_norm": 1.6133908033370972, "learning_rate": 7.856200000000001e-06, "loss": 0.0333, "step": 10720 }, { "epoch": 0.571573279309058, "grad_norm": 2.891467571258545, "learning_rate": 7.856e-06, "loss": 0.0354, "step": 10721 }, { "epoch": 0.5716265927387109, "grad_norm": 2.2221453189849854, "learning_rate": 7.8558e-06, "loss": 0.0302, "step": 10722 }, { "epoch": 0.5716799061683638, "grad_norm": 1.8277785778045654, "learning_rate": 7.855600000000001e-06, "loss": 0.0235, "step": 10723 }, { "epoch": 0.5717332195980167, "grad_norm": 2.2513303756713867, "learning_rate": 7.855400000000001e-06, "loss": 0.0222, "step": 10724 }, { "epoch": 0.5717865330276697, "grad_norm": 1.1577239036560059, "learning_rate": 7.8552e-06, "loss": 0.0157, "step": 10725 }, { "epoch": 0.5718398464573226, "grad_norm": 6.4806928634643555, "learning_rate": 7.855e-06, "loss": 0.0259, "step": 10726 }, { "epoch": 0.5718931598869755, "grad_norm": 1.8159340620040894, "learning_rate": 7.8548e-06, "loss": 0.0189, "step": 10727 }, { "epoch": 0.5719464733166285, "grad_norm": 2.3390471935272217, "learning_rate": 7.854600000000001e-06, "loss": 0.0211, "step": 10728 }, { "epoch": 0.5719997867462814, "grad_norm": 2.5183801651000977, "learning_rate": 7.8544e-06, "loss": 0.031, "step": 10729 }, { "epoch": 0.5720531001759344, "grad_norm": 2.3659284114837646, "learning_rate": 7.8542e-06, "loss": 0.0469, "step": 10730 }, { "epoch": 0.5721064136055872, "grad_norm": 3.906266689300537, "learning_rate": 7.854e-06, "loss": 0.0554, "step": 10731 }, { "epoch": 0.5721597270352402, "grad_norm": 3.5129406452178955, "learning_rate": 7.853800000000001e-06, "loss": 0.0591, "step": 10732 }, { "epoch": 0.5722130404648931, "grad_norm": 1.4008598327636719, "learning_rate": 7.8536e-06, "loss": 0.034, "step": 10733 }, { "epoch": 0.572266353894546, "grad_norm": 2.871626615524292, "learning_rate": 7.853400000000002e-06, "loss": 0.0254, "step": 10734 }, { "epoch": 0.5723196673241989, "grad_norm": 2.827733278274536, "learning_rate": 7.853200000000001e-06, "loss": 0.0089, "step": 10735 }, { "epoch": 0.5723729807538519, "grad_norm": 1.6730799674987793, "learning_rate": 7.853000000000001e-06, "loss": 0.0357, "step": 10736 }, { "epoch": 0.5724262941835049, "grad_norm": 2.245194435119629, "learning_rate": 7.8528e-06, "loss": 0.0165, "step": 10737 }, { "epoch": 0.5724796076131577, "grad_norm": 1.825016975402832, "learning_rate": 7.8526e-06, "loss": 0.0237, "step": 10738 }, { "epoch": 0.5725329210428107, "grad_norm": 3.4500324726104736, "learning_rate": 7.852400000000001e-06, "loss": 0.0583, "step": 10739 }, { "epoch": 0.5725862344724636, "grad_norm": 2.560912847518921, "learning_rate": 7.8522e-06, "loss": 0.0573, "step": 10740 }, { "epoch": 0.5726395479021166, "grad_norm": 4.556049346923828, "learning_rate": 7.852e-06, "loss": 0.033, "step": 10741 }, { "epoch": 0.5726928613317694, "grad_norm": 2.7803428173065186, "learning_rate": 7.8518e-06, "loss": 0.032, "step": 10742 }, { "epoch": 0.5727461747614224, "grad_norm": 2.194650173187256, "learning_rate": 7.8516e-06, "loss": 0.0549, "step": 10743 }, { "epoch": 0.5727994881910753, "grad_norm": 2.590857744216919, "learning_rate": 7.8514e-06, "loss": 0.0428, "step": 10744 }, { "epoch": 0.5728528016207283, "grad_norm": 3.7193009853363037, "learning_rate": 7.8512e-06, "loss": 0.037, "step": 10745 }, { "epoch": 0.5729061150503811, "grad_norm": 3.1130621433258057, "learning_rate": 7.851000000000002e-06, "loss": 0.0241, "step": 10746 }, { "epoch": 0.5729594284800341, "grad_norm": 1.691810131072998, "learning_rate": 7.850800000000001e-06, "loss": 0.0431, "step": 10747 }, { "epoch": 0.5730127419096871, "grad_norm": 1.960829734802246, "learning_rate": 7.8506e-06, "loss": 0.0316, "step": 10748 }, { "epoch": 0.57306605533934, "grad_norm": 2.1718549728393555, "learning_rate": 7.8504e-06, "loss": 0.0409, "step": 10749 }, { "epoch": 0.5731193687689929, "grad_norm": 3.248002767562866, "learning_rate": 7.850200000000002e-06, "loss": 0.0559, "step": 10750 }, { "epoch": 0.5731726821986458, "grad_norm": 3.5345425605773926, "learning_rate": 7.850000000000001e-06, "loss": 0.0576, "step": 10751 }, { "epoch": 0.5732259956282988, "grad_norm": 3.07051157951355, "learning_rate": 7.8498e-06, "loss": 0.037, "step": 10752 }, { "epoch": 0.5732793090579517, "grad_norm": 2.404632806777954, "learning_rate": 7.8496e-06, "loss": 0.0527, "step": 10753 }, { "epoch": 0.5733326224876046, "grad_norm": 2.7292532920837402, "learning_rate": 7.8494e-06, "loss": 0.0622, "step": 10754 }, { "epoch": 0.5733859359172575, "grad_norm": 2.213228940963745, "learning_rate": 7.8492e-06, "loss": 0.0349, "step": 10755 }, { "epoch": 0.5734392493469105, "grad_norm": 2.678252935409546, "learning_rate": 7.849e-06, "loss": 0.0298, "step": 10756 }, { "epoch": 0.5734925627765635, "grad_norm": 2.1842212677001953, "learning_rate": 7.8488e-06, "loss": 0.0473, "step": 10757 }, { "epoch": 0.5735458762062163, "grad_norm": 1.3037008047103882, "learning_rate": 7.848600000000001e-06, "loss": 0.0321, "step": 10758 }, { "epoch": 0.5735991896358693, "grad_norm": 2.809361219406128, "learning_rate": 7.848400000000001e-06, "loss": 0.0312, "step": 10759 }, { "epoch": 0.5736525030655222, "grad_norm": 1.824768304824829, "learning_rate": 7.8482e-06, "loss": 0.0258, "step": 10760 }, { "epoch": 0.5737058164951752, "grad_norm": 2.3734254837036133, "learning_rate": 7.848000000000002e-06, "loss": 0.0348, "step": 10761 }, { "epoch": 0.573759129924828, "grad_norm": 3.229255437850952, "learning_rate": 7.847800000000001e-06, "loss": 0.027, "step": 10762 }, { "epoch": 0.573812443354481, "grad_norm": 1.3339923620224, "learning_rate": 7.847600000000001e-06, "loss": 0.0114, "step": 10763 }, { "epoch": 0.5738657567841339, "grad_norm": 3.3375751972198486, "learning_rate": 7.8474e-06, "loss": 0.061, "step": 10764 }, { "epoch": 0.5739190702137869, "grad_norm": 3.8472657203674316, "learning_rate": 7.8472e-06, "loss": 0.0304, "step": 10765 }, { "epoch": 0.5739723836434398, "grad_norm": 2.2707865238189697, "learning_rate": 7.847e-06, "loss": 0.0435, "step": 10766 }, { "epoch": 0.5740256970730927, "grad_norm": 2.661163568496704, "learning_rate": 7.8468e-06, "loss": 0.0263, "step": 10767 }, { "epoch": 0.5740790105027457, "grad_norm": 2.3176121711730957, "learning_rate": 7.8466e-06, "loss": 0.0269, "step": 10768 }, { "epoch": 0.5741323239323985, "grad_norm": 3.3658018112182617, "learning_rate": 7.8464e-06, "loss": 0.0464, "step": 10769 }, { "epoch": 0.5741856373620515, "grad_norm": 2.343618392944336, "learning_rate": 7.846200000000001e-06, "loss": 0.0254, "step": 10770 }, { "epoch": 0.5742389507917044, "grad_norm": 1.494771957397461, "learning_rate": 7.846e-06, "loss": 0.0129, "step": 10771 }, { "epoch": 0.5742922642213574, "grad_norm": 1.678537130355835, "learning_rate": 7.845800000000002e-06, "loss": 0.0241, "step": 10772 }, { "epoch": 0.5743455776510102, "grad_norm": 1.7549408674240112, "learning_rate": 7.845600000000002e-06, "loss": 0.0222, "step": 10773 }, { "epoch": 0.5743988910806632, "grad_norm": 1.8642388582229614, "learning_rate": 7.845400000000001e-06, "loss": 0.0128, "step": 10774 }, { "epoch": 0.5744522045103162, "grad_norm": 2.783622980117798, "learning_rate": 7.8452e-06, "loss": 0.0562, "step": 10775 }, { "epoch": 0.5745055179399691, "grad_norm": 2.0694732666015625, "learning_rate": 7.845e-06, "loss": 0.0273, "step": 10776 }, { "epoch": 0.574558831369622, "grad_norm": 3.6608400344848633, "learning_rate": 7.8448e-06, "loss": 0.03, "step": 10777 }, { "epoch": 0.5746121447992749, "grad_norm": 2.1696395874023438, "learning_rate": 7.844600000000001e-06, "loss": 0.032, "step": 10778 }, { "epoch": 0.5746654582289279, "grad_norm": 1.0539599657058716, "learning_rate": 7.8444e-06, "loss": 0.0087, "step": 10779 }, { "epoch": 0.5747187716585808, "grad_norm": 0.843058168888092, "learning_rate": 7.8442e-06, "loss": 0.006, "step": 10780 }, { "epoch": 0.5747720850882337, "grad_norm": 6.795835018157959, "learning_rate": 7.844e-06, "loss": 0.0426, "step": 10781 }, { "epoch": 0.5748253985178866, "grad_norm": 3.4274990558624268, "learning_rate": 7.843800000000001e-06, "loss": 0.0647, "step": 10782 }, { "epoch": 0.5748787119475396, "grad_norm": 2.8539154529571533, "learning_rate": 7.8436e-06, "loss": 0.0347, "step": 10783 }, { "epoch": 0.5749320253771926, "grad_norm": 1.7333415746688843, "learning_rate": 7.843400000000002e-06, "loss": 0.0124, "step": 10784 }, { "epoch": 0.5749853388068454, "grad_norm": 3.3355872631073, "learning_rate": 7.843200000000001e-06, "loss": 0.0253, "step": 10785 }, { "epoch": 0.5750386522364984, "grad_norm": 1.9656596183776855, "learning_rate": 7.843000000000001e-06, "loss": 0.0359, "step": 10786 }, { "epoch": 0.5750919656661513, "grad_norm": 2.013035535812378, "learning_rate": 7.8428e-06, "loss": 0.021, "step": 10787 }, { "epoch": 0.5751452790958043, "grad_norm": 3.0735509395599365, "learning_rate": 7.8426e-06, "loss": 0.031, "step": 10788 }, { "epoch": 0.5751985925254571, "grad_norm": 2.531310796737671, "learning_rate": 7.842400000000001e-06, "loss": 0.019, "step": 10789 }, { "epoch": 0.5752519059551101, "grad_norm": 1.691462755203247, "learning_rate": 7.842200000000001e-06, "loss": 0.021, "step": 10790 }, { "epoch": 0.575305219384763, "grad_norm": 2.5557193756103516, "learning_rate": 7.842e-06, "loss": 0.0445, "step": 10791 }, { "epoch": 0.575358532814416, "grad_norm": 2.2847583293914795, "learning_rate": 7.8418e-06, "loss": 0.0216, "step": 10792 }, { "epoch": 0.5754118462440689, "grad_norm": 2.8664028644561768, "learning_rate": 7.8416e-06, "loss": 0.0697, "step": 10793 }, { "epoch": 0.5754651596737218, "grad_norm": 2.2600929737091064, "learning_rate": 7.841400000000001e-06, "loss": 0.024, "step": 10794 }, { "epoch": 0.5755184731033748, "grad_norm": 2.9028048515319824, "learning_rate": 7.8412e-06, "loss": 0.0642, "step": 10795 }, { "epoch": 0.5755717865330277, "grad_norm": 2.475353717803955, "learning_rate": 7.841000000000002e-06, "loss": 0.0238, "step": 10796 }, { "epoch": 0.5756250999626806, "grad_norm": 2.6352157592773438, "learning_rate": 7.840800000000001e-06, "loss": 0.0329, "step": 10797 }, { "epoch": 0.5756784133923335, "grad_norm": 3.034388303756714, "learning_rate": 7.8406e-06, "loss": 0.0185, "step": 10798 }, { "epoch": 0.5757317268219865, "grad_norm": 3.2505569458007812, "learning_rate": 7.8404e-06, "loss": 0.0323, "step": 10799 }, { "epoch": 0.5757850402516393, "grad_norm": 2.5514190196990967, "learning_rate": 7.840200000000002e-06, "loss": 0.0399, "step": 10800 }, { "epoch": 0.5758383536812923, "grad_norm": 1.6810591220855713, "learning_rate": 7.840000000000001e-06, "loss": 0.0234, "step": 10801 }, { "epoch": 0.5758916671109452, "grad_norm": 2.9378724098205566, "learning_rate": 7.8398e-06, "loss": 0.0245, "step": 10802 }, { "epoch": 0.5759449805405982, "grad_norm": 1.3257802724838257, "learning_rate": 7.8396e-06, "loss": 0.0242, "step": 10803 }, { "epoch": 0.5759982939702512, "grad_norm": 2.3019492626190186, "learning_rate": 7.8394e-06, "loss": 0.0239, "step": 10804 }, { "epoch": 0.576051607399904, "grad_norm": 0.9746189713478088, "learning_rate": 7.8392e-06, "loss": 0.0076, "step": 10805 }, { "epoch": 0.576104920829557, "grad_norm": 2.281348466873169, "learning_rate": 7.839e-06, "loss": 0.0362, "step": 10806 }, { "epoch": 0.5761582342592099, "grad_norm": 2.2161624431610107, "learning_rate": 7.8388e-06, "loss": 0.0472, "step": 10807 }, { "epoch": 0.5762115476888628, "grad_norm": 2.5122921466827393, "learning_rate": 7.838600000000001e-06, "loss": 0.0419, "step": 10808 }, { "epoch": 0.5762648611185157, "grad_norm": 1.6465686559677124, "learning_rate": 7.838400000000001e-06, "loss": 0.0114, "step": 10809 }, { "epoch": 0.5763181745481687, "grad_norm": 2.8815231323242188, "learning_rate": 7.8382e-06, "loss": 0.0668, "step": 10810 }, { "epoch": 0.5763714879778216, "grad_norm": 4.299543857574463, "learning_rate": 7.838000000000002e-06, "loss": 0.0718, "step": 10811 }, { "epoch": 0.5764248014074745, "grad_norm": 2.6804873943328857, "learning_rate": 7.837800000000001e-06, "loss": 0.0248, "step": 10812 }, { "epoch": 0.5764781148371275, "grad_norm": 3.8709471225738525, "learning_rate": 7.837600000000001e-06, "loss": 0.0526, "step": 10813 }, { "epoch": 0.5765314282667804, "grad_norm": 4.152965545654297, "learning_rate": 7.8374e-06, "loss": 0.0475, "step": 10814 }, { "epoch": 0.5765847416964334, "grad_norm": 2.3063626289367676, "learning_rate": 7.8372e-06, "loss": 0.0188, "step": 10815 }, { "epoch": 0.5766380551260862, "grad_norm": 2.1780529022216797, "learning_rate": 7.837e-06, "loss": 0.0167, "step": 10816 }, { "epoch": 0.5766913685557392, "grad_norm": 2.361809492111206, "learning_rate": 7.836800000000001e-06, "loss": 0.0382, "step": 10817 }, { "epoch": 0.5767446819853921, "grad_norm": 2.4939017295837402, "learning_rate": 7.8366e-06, "loss": 0.0297, "step": 10818 }, { "epoch": 0.5767979954150451, "grad_norm": 3.2260732650756836, "learning_rate": 7.8364e-06, "loss": 0.0245, "step": 10819 }, { "epoch": 0.5768513088446979, "grad_norm": 1.245178461074829, "learning_rate": 7.836200000000001e-06, "loss": 0.0186, "step": 10820 }, { "epoch": 0.5769046222743509, "grad_norm": 2.938979148864746, "learning_rate": 7.836000000000001e-06, "loss": 0.0501, "step": 10821 }, { "epoch": 0.5769579357040039, "grad_norm": 2.0232973098754883, "learning_rate": 7.8358e-06, "loss": 0.0411, "step": 10822 }, { "epoch": 0.5770112491336568, "grad_norm": 2.665972948074341, "learning_rate": 7.835600000000002e-06, "loss": 0.0258, "step": 10823 }, { "epoch": 0.5770645625633097, "grad_norm": 3.2249085903167725, "learning_rate": 7.835400000000001e-06, "loss": 0.0603, "step": 10824 }, { "epoch": 0.5771178759929626, "grad_norm": 1.343416452407837, "learning_rate": 7.8352e-06, "loss": 0.0137, "step": 10825 }, { "epoch": 0.5771711894226156, "grad_norm": 0.8950802087783813, "learning_rate": 7.835e-06, "loss": 0.0171, "step": 10826 }, { "epoch": 0.5772245028522685, "grad_norm": 1.0702511072158813, "learning_rate": 7.8348e-06, "loss": 0.0178, "step": 10827 }, { "epoch": 0.5772778162819214, "grad_norm": 2.919792890548706, "learning_rate": 7.834600000000001e-06, "loss": 0.0474, "step": 10828 }, { "epoch": 0.5773311297115743, "grad_norm": 3.807854652404785, "learning_rate": 7.8344e-06, "loss": 0.0626, "step": 10829 }, { "epoch": 0.5773844431412273, "grad_norm": 2.2276012897491455, "learning_rate": 7.8342e-06, "loss": 0.0359, "step": 10830 }, { "epoch": 0.5774377565708803, "grad_norm": 3.1493828296661377, "learning_rate": 7.834e-06, "loss": 0.0518, "step": 10831 }, { "epoch": 0.5774910700005331, "grad_norm": 2.597752332687378, "learning_rate": 7.833800000000001e-06, "loss": 0.0504, "step": 10832 }, { "epoch": 0.5775443834301861, "grad_norm": 2.1370906829833984, "learning_rate": 7.8336e-06, "loss": 0.0247, "step": 10833 }, { "epoch": 0.577597696859839, "grad_norm": 1.977414608001709, "learning_rate": 7.8334e-06, "loss": 0.0264, "step": 10834 }, { "epoch": 0.577651010289492, "grad_norm": 4.1474080085754395, "learning_rate": 7.833200000000002e-06, "loss": 0.0315, "step": 10835 }, { "epoch": 0.5777043237191448, "grad_norm": 2.6740827560424805, "learning_rate": 7.833000000000001e-06, "loss": 0.0197, "step": 10836 }, { "epoch": 0.5777576371487978, "grad_norm": 2.769092321395874, "learning_rate": 7.8328e-06, "loss": 0.0279, "step": 10837 }, { "epoch": 0.5778109505784507, "grad_norm": 3.397984743118286, "learning_rate": 7.8326e-06, "loss": 0.0212, "step": 10838 }, { "epoch": 0.5778642640081036, "grad_norm": 1.1596368551254272, "learning_rate": 7.832400000000001e-06, "loss": 0.0157, "step": 10839 }, { "epoch": 0.5779175774377566, "grad_norm": 4.154852867126465, "learning_rate": 7.832200000000001e-06, "loss": 0.0485, "step": 10840 }, { "epoch": 0.5779708908674095, "grad_norm": 3.521688461303711, "learning_rate": 7.832e-06, "loss": 0.052, "step": 10841 }, { "epoch": 0.5780242042970625, "grad_norm": 3.603126287460327, "learning_rate": 7.8318e-06, "loss": 0.0495, "step": 10842 }, { "epoch": 0.5780775177267153, "grad_norm": 3.1385979652404785, "learning_rate": 7.8316e-06, "loss": 0.0581, "step": 10843 }, { "epoch": 0.5781308311563683, "grad_norm": 3.9292144775390625, "learning_rate": 7.831400000000001e-06, "loss": 0.0514, "step": 10844 }, { "epoch": 0.5781841445860212, "grad_norm": 2.001389265060425, "learning_rate": 7.8312e-06, "loss": 0.0253, "step": 10845 }, { "epoch": 0.5782374580156742, "grad_norm": 2.6455819606781006, "learning_rate": 7.831e-06, "loss": 0.0239, "step": 10846 }, { "epoch": 0.578290771445327, "grad_norm": 3.655850887298584, "learning_rate": 7.830800000000001e-06, "loss": 0.0251, "step": 10847 }, { "epoch": 0.57834408487498, "grad_norm": 1.6974117755889893, "learning_rate": 7.830600000000001e-06, "loss": 0.0171, "step": 10848 }, { "epoch": 0.578397398304633, "grad_norm": 3.10837459564209, "learning_rate": 7.8304e-06, "loss": 0.0259, "step": 10849 }, { "epoch": 0.5784507117342859, "grad_norm": 3.193864107131958, "learning_rate": 7.830200000000002e-06, "loss": 0.0521, "step": 10850 }, { "epoch": 0.5785040251639388, "grad_norm": 4.549874305725098, "learning_rate": 7.830000000000001e-06, "loss": 0.036, "step": 10851 }, { "epoch": 0.5785573385935917, "grad_norm": 2.5644278526306152, "learning_rate": 7.8298e-06, "loss": 0.0303, "step": 10852 }, { "epoch": 0.5786106520232447, "grad_norm": 1.5074992179870605, "learning_rate": 7.8296e-06, "loss": 0.0245, "step": 10853 }, { "epoch": 0.5786639654528976, "grad_norm": 3.980459451675415, "learning_rate": 7.8294e-06, "loss": 0.0719, "step": 10854 }, { "epoch": 0.5787172788825505, "grad_norm": 0.5581337809562683, "learning_rate": 7.8292e-06, "loss": 0.0053, "step": 10855 }, { "epoch": 0.5787705923122034, "grad_norm": 3.2706453800201416, "learning_rate": 7.829e-06, "loss": 0.1159, "step": 10856 }, { "epoch": 0.5788239057418564, "grad_norm": 1.5492117404937744, "learning_rate": 7.8288e-06, "loss": 0.0144, "step": 10857 }, { "epoch": 0.5788772191715092, "grad_norm": 1.7975064516067505, "learning_rate": 7.8286e-06, "loss": 0.0119, "step": 10858 }, { "epoch": 0.5789305326011622, "grad_norm": 1.9903560876846313, "learning_rate": 7.828400000000001e-06, "loss": 0.0268, "step": 10859 }, { "epoch": 0.5789838460308152, "grad_norm": 1.8929072618484497, "learning_rate": 7.8282e-06, "loss": 0.037, "step": 10860 }, { "epoch": 0.5790371594604681, "grad_norm": 2.2104811668395996, "learning_rate": 7.828000000000002e-06, "loss": 0.0515, "step": 10861 }, { "epoch": 0.579090472890121, "grad_norm": 2.684210777282715, "learning_rate": 7.827800000000002e-06, "loss": 0.0333, "step": 10862 }, { "epoch": 0.5791437863197739, "grad_norm": 2.0712759494781494, "learning_rate": 7.827600000000001e-06, "loss": 0.0126, "step": 10863 }, { "epoch": 0.5791970997494269, "grad_norm": 2.822322130203247, "learning_rate": 7.8274e-06, "loss": 0.0388, "step": 10864 }, { "epoch": 0.5792504131790798, "grad_norm": 1.3053233623504639, "learning_rate": 7.8272e-06, "loss": 0.0185, "step": 10865 }, { "epoch": 0.5793037266087327, "grad_norm": 3.701974868774414, "learning_rate": 7.827e-06, "loss": 0.0167, "step": 10866 }, { "epoch": 0.5793570400383856, "grad_norm": 2.062135934829712, "learning_rate": 7.826800000000001e-06, "loss": 0.016, "step": 10867 }, { "epoch": 0.5794103534680386, "grad_norm": 1.972703218460083, "learning_rate": 7.8266e-06, "loss": 0.037, "step": 10868 }, { "epoch": 0.5794636668976916, "grad_norm": 1.2023614645004272, "learning_rate": 7.8264e-06, "loss": 0.0358, "step": 10869 }, { "epoch": 0.5795169803273444, "grad_norm": 1.538971185684204, "learning_rate": 7.8262e-06, "loss": 0.0443, "step": 10870 }, { "epoch": 0.5795702937569974, "grad_norm": 2.3213319778442383, "learning_rate": 7.826000000000001e-06, "loss": 0.0364, "step": 10871 }, { "epoch": 0.5796236071866503, "grad_norm": 3.133326768875122, "learning_rate": 7.8258e-06, "loss": 0.0573, "step": 10872 }, { "epoch": 0.5796769206163033, "grad_norm": 2.281935453414917, "learning_rate": 7.825600000000002e-06, "loss": 0.0418, "step": 10873 }, { "epoch": 0.5797302340459561, "grad_norm": 2.368713140487671, "learning_rate": 7.825400000000001e-06, "loss": 0.0343, "step": 10874 }, { "epoch": 0.5797835474756091, "grad_norm": 0.4281122386455536, "learning_rate": 7.825200000000001e-06, "loss": 0.0038, "step": 10875 }, { "epoch": 0.579836860905262, "grad_norm": 1.5400285720825195, "learning_rate": 7.825e-06, "loss": 0.0266, "step": 10876 }, { "epoch": 0.579890174334915, "grad_norm": 2.261314630508423, "learning_rate": 7.8248e-06, "loss": 0.0448, "step": 10877 }, { "epoch": 0.5799434877645679, "grad_norm": 1.6240622997283936, "learning_rate": 7.824600000000001e-06, "loss": 0.0406, "step": 10878 }, { "epoch": 0.5799968011942208, "grad_norm": 1.9618192911148071, "learning_rate": 7.8244e-06, "loss": 0.0213, "step": 10879 }, { "epoch": 0.5800501146238738, "grad_norm": 1.8788355588912964, "learning_rate": 7.8242e-06, "loss": 0.0174, "step": 10880 }, { "epoch": 0.5801034280535267, "grad_norm": 1.26311457157135, "learning_rate": 7.824e-06, "loss": 0.0198, "step": 10881 }, { "epoch": 0.5801567414831796, "grad_norm": 4.392272472381592, "learning_rate": 7.8238e-06, "loss": 0.0633, "step": 10882 }, { "epoch": 0.5802100549128325, "grad_norm": 2.7622408866882324, "learning_rate": 7.8236e-06, "loss": 0.0261, "step": 10883 }, { "epoch": 0.5802633683424855, "grad_norm": 2.1578292846679688, "learning_rate": 7.8234e-06, "loss": 0.0442, "step": 10884 }, { "epoch": 0.5803166817721384, "grad_norm": 1.499375581741333, "learning_rate": 7.823200000000002e-06, "loss": 0.0205, "step": 10885 }, { "epoch": 0.5803699952017913, "grad_norm": 1.836482286453247, "learning_rate": 7.823000000000001e-06, "loss": 0.0192, "step": 10886 }, { "epoch": 0.5804233086314443, "grad_norm": 2.142658233642578, "learning_rate": 7.8228e-06, "loss": 0.0317, "step": 10887 }, { "epoch": 0.5804766220610972, "grad_norm": 1.2634835243225098, "learning_rate": 7.8226e-06, "loss": 0.0179, "step": 10888 }, { "epoch": 0.5805299354907502, "grad_norm": 2.090670585632324, "learning_rate": 7.822400000000002e-06, "loss": 0.0417, "step": 10889 }, { "epoch": 0.580583248920403, "grad_norm": 2.632377862930298, "learning_rate": 7.822200000000001e-06, "loss": 0.0191, "step": 10890 }, { "epoch": 0.580636562350056, "grad_norm": 2.591292381286621, "learning_rate": 7.822e-06, "loss": 0.0189, "step": 10891 }, { "epoch": 0.5806898757797089, "grad_norm": 3.266944169998169, "learning_rate": 7.8218e-06, "loss": 0.0306, "step": 10892 }, { "epoch": 0.5807431892093619, "grad_norm": 3.659484386444092, "learning_rate": 7.8216e-06, "loss": 0.0202, "step": 10893 }, { "epoch": 0.5807965026390147, "grad_norm": 4.935224533081055, "learning_rate": 7.8214e-06, "loss": 0.0636, "step": 10894 }, { "epoch": 0.5808498160686677, "grad_norm": 4.046386241912842, "learning_rate": 7.8212e-06, "loss": 0.0324, "step": 10895 }, { "epoch": 0.5809031294983207, "grad_norm": 1.844576120376587, "learning_rate": 7.821e-06, "loss": 0.0195, "step": 10896 }, { "epoch": 0.5809564429279735, "grad_norm": 6.693434238433838, "learning_rate": 7.820800000000001e-06, "loss": 0.0533, "step": 10897 }, { "epoch": 0.5810097563576265, "grad_norm": 1.4652788639068604, "learning_rate": 7.820600000000001e-06, "loss": 0.0127, "step": 10898 }, { "epoch": 0.5810630697872794, "grad_norm": 2.8130335807800293, "learning_rate": 7.8204e-06, "loss": 0.031, "step": 10899 }, { "epoch": 0.5811163832169324, "grad_norm": 2.9133269786834717, "learning_rate": 7.820200000000002e-06, "loss": 0.045, "step": 10900 }, { "epoch": 0.5811696966465852, "grad_norm": 2.6348838806152344, "learning_rate": 7.820000000000001e-06, "loss": 0.0447, "step": 10901 }, { "epoch": 0.5812230100762382, "grad_norm": 1.5647627115249634, "learning_rate": 7.819800000000001e-06, "loss": 0.0232, "step": 10902 }, { "epoch": 0.5812763235058911, "grad_norm": 1.5987221002578735, "learning_rate": 7.8196e-06, "loss": 0.0204, "step": 10903 }, { "epoch": 0.5813296369355441, "grad_norm": 3.6199872493743896, "learning_rate": 7.8194e-06, "loss": 0.0339, "step": 10904 }, { "epoch": 0.5813829503651969, "grad_norm": 1.3649632930755615, "learning_rate": 7.8192e-06, "loss": 0.0134, "step": 10905 }, { "epoch": 0.5814362637948499, "grad_norm": 2.865015745162964, "learning_rate": 7.819000000000001e-06, "loss": 0.0313, "step": 10906 }, { "epoch": 0.5814895772245029, "grad_norm": 2.9963715076446533, "learning_rate": 7.8188e-06, "loss": 0.0443, "step": 10907 }, { "epoch": 0.5815428906541558, "grad_norm": 5.757296562194824, "learning_rate": 7.8186e-06, "loss": 0.0375, "step": 10908 }, { "epoch": 0.5815962040838087, "grad_norm": 2.3634893894195557, "learning_rate": 7.818400000000001e-06, "loss": 0.0324, "step": 10909 }, { "epoch": 0.5816495175134616, "grad_norm": 3.567152261734009, "learning_rate": 7.8182e-06, "loss": 0.0468, "step": 10910 }, { "epoch": 0.5817028309431146, "grad_norm": 3.3123295307159424, "learning_rate": 7.818e-06, "loss": 0.0408, "step": 10911 }, { "epoch": 0.5817561443727675, "grad_norm": 1.9077088832855225, "learning_rate": 7.817800000000002e-06, "loss": 0.0332, "step": 10912 }, { "epoch": 0.5818094578024204, "grad_norm": 3.676853656768799, "learning_rate": 7.817600000000001e-06, "loss": 0.0459, "step": 10913 }, { "epoch": 0.5818627712320733, "grad_norm": 3.405975818634033, "learning_rate": 7.8174e-06, "loss": 0.0253, "step": 10914 }, { "epoch": 0.5819160846617263, "grad_norm": 2.380563497543335, "learning_rate": 7.8172e-06, "loss": 0.0339, "step": 10915 }, { "epoch": 0.5819693980913793, "grad_norm": 3.640096664428711, "learning_rate": 7.817e-06, "loss": 0.0347, "step": 10916 }, { "epoch": 0.5820227115210321, "grad_norm": 2.0182621479034424, "learning_rate": 7.816800000000001e-06, "loss": 0.0348, "step": 10917 }, { "epoch": 0.5820760249506851, "grad_norm": 3.835634708404541, "learning_rate": 7.8166e-06, "loss": 0.0449, "step": 10918 }, { "epoch": 0.582129338380338, "grad_norm": 2.8296594619750977, "learning_rate": 7.8164e-06, "loss": 0.0457, "step": 10919 }, { "epoch": 0.582182651809991, "grad_norm": 1.493896245956421, "learning_rate": 7.8162e-06, "loss": 0.0247, "step": 10920 }, { "epoch": 0.5822359652396438, "grad_norm": 2.9487688541412354, "learning_rate": 7.816000000000001e-06, "loss": 0.0349, "step": 10921 }, { "epoch": 0.5822892786692968, "grad_norm": 3.1709065437316895, "learning_rate": 7.8158e-06, "loss": 0.022, "step": 10922 }, { "epoch": 0.5823425920989497, "grad_norm": 2.7790162563323975, "learning_rate": 7.8156e-06, "loss": 0.0349, "step": 10923 }, { "epoch": 0.5823959055286027, "grad_norm": 3.1535775661468506, "learning_rate": 7.815400000000001e-06, "loss": 0.0476, "step": 10924 }, { "epoch": 0.5824492189582556, "grad_norm": 2.6782848834991455, "learning_rate": 7.815200000000001e-06, "loss": 0.0376, "step": 10925 }, { "epoch": 0.5825025323879085, "grad_norm": 2.1691064834594727, "learning_rate": 7.815e-06, "loss": 0.0199, "step": 10926 }, { "epoch": 0.5825558458175615, "grad_norm": 2.578726291656494, "learning_rate": 7.8148e-06, "loss": 0.0558, "step": 10927 }, { "epoch": 0.5826091592472143, "grad_norm": 1.6068816184997559, "learning_rate": 7.814600000000001e-06, "loss": 0.0192, "step": 10928 }, { "epoch": 0.5826624726768673, "grad_norm": 2.4091219902038574, "learning_rate": 7.814400000000001e-06, "loss": 0.0604, "step": 10929 }, { "epoch": 0.5827157861065202, "grad_norm": 2.7945733070373535, "learning_rate": 7.8142e-06, "loss": 0.0351, "step": 10930 }, { "epoch": 0.5827690995361732, "grad_norm": 2.4844093322753906, "learning_rate": 7.814e-06, "loss": 0.0395, "step": 10931 }, { "epoch": 0.582822412965826, "grad_norm": 3.220606803894043, "learning_rate": 7.8138e-06, "loss": 0.0403, "step": 10932 }, { "epoch": 0.582875726395479, "grad_norm": 2.212139368057251, "learning_rate": 7.813600000000001e-06, "loss": 0.0491, "step": 10933 }, { "epoch": 0.582929039825132, "grad_norm": 1.289168119430542, "learning_rate": 7.8134e-06, "loss": 0.0087, "step": 10934 }, { "epoch": 0.5829823532547849, "grad_norm": 1.5057142972946167, "learning_rate": 7.8132e-06, "loss": 0.0228, "step": 10935 }, { "epoch": 0.5830356666844378, "grad_norm": 1.466015100479126, "learning_rate": 7.813000000000001e-06, "loss": 0.0142, "step": 10936 }, { "epoch": 0.5830889801140907, "grad_norm": 1.6796587705612183, "learning_rate": 7.8128e-06, "loss": 0.0207, "step": 10937 }, { "epoch": 0.5831422935437437, "grad_norm": 4.369513511657715, "learning_rate": 7.8126e-06, "loss": 0.0568, "step": 10938 }, { "epoch": 0.5831956069733966, "grad_norm": 4.268925666809082, "learning_rate": 7.812400000000002e-06, "loss": 0.0395, "step": 10939 }, { "epoch": 0.5832489204030495, "grad_norm": 2.2478902339935303, "learning_rate": 7.812200000000001e-06, "loss": 0.0321, "step": 10940 }, { "epoch": 0.5833022338327024, "grad_norm": 2.7234091758728027, "learning_rate": 7.812e-06, "loss": 0.0502, "step": 10941 }, { "epoch": 0.5833555472623554, "grad_norm": 2.4065444469451904, "learning_rate": 7.8118e-06, "loss": 0.0503, "step": 10942 }, { "epoch": 0.5834088606920084, "grad_norm": 4.264060974121094, "learning_rate": 7.8116e-06, "loss": 0.0347, "step": 10943 }, { "epoch": 0.5834621741216612, "grad_norm": 2.858322858810425, "learning_rate": 7.8114e-06, "loss": 0.0423, "step": 10944 }, { "epoch": 0.5835154875513142, "grad_norm": 2.4443039894104004, "learning_rate": 7.8112e-06, "loss": 0.0244, "step": 10945 }, { "epoch": 0.5835688009809671, "grad_norm": 2.422719717025757, "learning_rate": 7.811e-06, "loss": 0.034, "step": 10946 }, { "epoch": 0.5836221144106201, "grad_norm": 2.590144395828247, "learning_rate": 7.8108e-06, "loss": 0.0218, "step": 10947 }, { "epoch": 0.5836754278402729, "grad_norm": 2.031860113143921, "learning_rate": 7.810600000000001e-06, "loss": 0.0292, "step": 10948 }, { "epoch": 0.5837287412699259, "grad_norm": 3.0785014629364014, "learning_rate": 7.8104e-06, "loss": 0.0211, "step": 10949 }, { "epoch": 0.5837820546995788, "grad_norm": 1.6023694276809692, "learning_rate": 7.810200000000002e-06, "loss": 0.0232, "step": 10950 }, { "epoch": 0.5838353681292318, "grad_norm": 2.052865982055664, "learning_rate": 7.810000000000001e-06, "loss": 0.0249, "step": 10951 }, { "epoch": 0.5838886815588847, "grad_norm": 1.840116024017334, "learning_rate": 7.809800000000001e-06, "loss": 0.0428, "step": 10952 }, { "epoch": 0.5839419949885376, "grad_norm": 2.318467855453491, "learning_rate": 7.8096e-06, "loss": 0.046, "step": 10953 }, { "epoch": 0.5839953084181906, "grad_norm": 0.9648482799530029, "learning_rate": 7.8094e-06, "loss": 0.0117, "step": 10954 }, { "epoch": 0.5840486218478435, "grad_norm": 2.886953353881836, "learning_rate": 7.8092e-06, "loss": 0.0453, "step": 10955 }, { "epoch": 0.5841019352774964, "grad_norm": 1.8118726015090942, "learning_rate": 7.809000000000001e-06, "loss": 0.0312, "step": 10956 }, { "epoch": 0.5841552487071493, "grad_norm": 1.634134292602539, "learning_rate": 7.8088e-06, "loss": 0.0144, "step": 10957 }, { "epoch": 0.5842085621368023, "grad_norm": 1.6357210874557495, "learning_rate": 7.8086e-06, "loss": 0.0199, "step": 10958 }, { "epoch": 0.5842618755664551, "grad_norm": 1.7298067808151245, "learning_rate": 7.8084e-06, "loss": 0.0118, "step": 10959 }, { "epoch": 0.5843151889961081, "grad_norm": 1.8761757612228394, "learning_rate": 7.808200000000001e-06, "loss": 0.0223, "step": 10960 }, { "epoch": 0.584368502425761, "grad_norm": 2.1088056564331055, "learning_rate": 7.808e-06, "loss": 0.0427, "step": 10961 }, { "epoch": 0.584421815855414, "grad_norm": 1.9863450527191162, "learning_rate": 7.807800000000002e-06, "loss": 0.027, "step": 10962 }, { "epoch": 0.584475129285067, "grad_norm": 1.8431676626205444, "learning_rate": 7.807600000000001e-06, "loss": 0.0305, "step": 10963 }, { "epoch": 0.5845284427147198, "grad_norm": 1.5918447971343994, "learning_rate": 7.8074e-06, "loss": 0.0114, "step": 10964 }, { "epoch": 0.5845817561443728, "grad_norm": 4.6851983070373535, "learning_rate": 7.8072e-06, "loss": 0.0583, "step": 10965 }, { "epoch": 0.5846350695740257, "grad_norm": 3.1881580352783203, "learning_rate": 7.807e-06, "loss": 0.0314, "step": 10966 }, { "epoch": 0.5846883830036786, "grad_norm": 4.914795398712158, "learning_rate": 7.806800000000001e-06, "loss": 0.0201, "step": 10967 }, { "epoch": 0.5847416964333315, "grad_norm": 2.873897075653076, "learning_rate": 7.8066e-06, "loss": 0.058, "step": 10968 }, { "epoch": 0.5847950098629845, "grad_norm": 2.2600505352020264, "learning_rate": 7.8064e-06, "loss": 0.0318, "step": 10969 }, { "epoch": 0.5848483232926374, "grad_norm": 3.683837890625, "learning_rate": 7.8062e-06, "loss": 0.056, "step": 10970 }, { "epoch": 0.5849016367222903, "grad_norm": 1.8971853256225586, "learning_rate": 7.806e-06, "loss": 0.0394, "step": 10971 }, { "epoch": 0.5849549501519433, "grad_norm": 3.8459205627441406, "learning_rate": 7.8058e-06, "loss": 0.0882, "step": 10972 }, { "epoch": 0.5850082635815962, "grad_norm": 1.5738098621368408, "learning_rate": 7.8056e-06, "loss": 0.0093, "step": 10973 }, { "epoch": 0.5850615770112492, "grad_norm": 2.2995917797088623, "learning_rate": 7.805400000000002e-06, "loss": 0.0503, "step": 10974 }, { "epoch": 0.585114890440902, "grad_norm": 2.727109909057617, "learning_rate": 7.805200000000001e-06, "loss": 0.0229, "step": 10975 }, { "epoch": 0.585168203870555, "grad_norm": 2.610349178314209, "learning_rate": 7.805e-06, "loss": 0.0371, "step": 10976 }, { "epoch": 0.5852215173002079, "grad_norm": 1.8702229261398315, "learning_rate": 7.8048e-06, "loss": 0.0148, "step": 10977 }, { "epoch": 0.5852748307298609, "grad_norm": 3.481825351715088, "learning_rate": 7.804600000000001e-06, "loss": 0.029, "step": 10978 }, { "epoch": 0.5853281441595137, "grad_norm": 2.5307910442352295, "learning_rate": 7.804400000000001e-06, "loss": 0.0357, "step": 10979 }, { "epoch": 0.5853814575891667, "grad_norm": 3.273693084716797, "learning_rate": 7.8042e-06, "loss": 0.0508, "step": 10980 }, { "epoch": 0.5854347710188197, "grad_norm": 4.587156772613525, "learning_rate": 7.804e-06, "loss": 0.0515, "step": 10981 }, { "epoch": 0.5854880844484726, "grad_norm": 2.9932007789611816, "learning_rate": 7.8038e-06, "loss": 0.043, "step": 10982 }, { "epoch": 0.5855413978781255, "grad_norm": 4.6083598136901855, "learning_rate": 7.803600000000001e-06, "loss": 0.0292, "step": 10983 }, { "epoch": 0.5855947113077784, "grad_norm": 1.4259976148605347, "learning_rate": 7.8034e-06, "loss": 0.0217, "step": 10984 }, { "epoch": 0.5856480247374314, "grad_norm": 1.5140491724014282, "learning_rate": 7.8032e-06, "loss": 0.0184, "step": 10985 }, { "epoch": 0.5857013381670843, "grad_norm": 3.8812339305877686, "learning_rate": 7.803000000000001e-06, "loss": 0.0455, "step": 10986 }, { "epoch": 0.5857546515967372, "grad_norm": 1.2749558687210083, "learning_rate": 7.802800000000001e-06, "loss": 0.0172, "step": 10987 }, { "epoch": 0.5858079650263901, "grad_norm": 2.216369152069092, "learning_rate": 7.8026e-06, "loss": 0.023, "step": 10988 }, { "epoch": 0.5858612784560431, "grad_norm": 1.1654400825500488, "learning_rate": 7.802400000000002e-06, "loss": 0.0124, "step": 10989 }, { "epoch": 0.5859145918856961, "grad_norm": 1.7394349575042725, "learning_rate": 7.802200000000001e-06, "loss": 0.0355, "step": 10990 }, { "epoch": 0.5859679053153489, "grad_norm": 1.39827561378479, "learning_rate": 7.802000000000001e-06, "loss": 0.0395, "step": 10991 }, { "epoch": 0.5860212187450019, "grad_norm": 2.2010107040405273, "learning_rate": 7.8018e-06, "loss": 0.0437, "step": 10992 }, { "epoch": 0.5860745321746548, "grad_norm": 1.4329673051834106, "learning_rate": 7.8016e-06, "loss": 0.0201, "step": 10993 }, { "epoch": 0.5861278456043078, "grad_norm": 2.42954158782959, "learning_rate": 7.8014e-06, "loss": 0.0466, "step": 10994 }, { "epoch": 0.5861811590339606, "grad_norm": 2.9808554649353027, "learning_rate": 7.8012e-06, "loss": 0.0473, "step": 10995 }, { "epoch": 0.5862344724636136, "grad_norm": 1.6284749507904053, "learning_rate": 7.801e-06, "loss": 0.0224, "step": 10996 }, { "epoch": 0.5862877858932665, "grad_norm": 0.996311366558075, "learning_rate": 7.8008e-06, "loss": 0.0073, "step": 10997 }, { "epoch": 0.5863410993229194, "grad_norm": 3.2057552337646484, "learning_rate": 7.800600000000001e-06, "loss": 0.0509, "step": 10998 }, { "epoch": 0.5863944127525724, "grad_norm": 5.930686950683594, "learning_rate": 7.8004e-06, "loss": 0.0315, "step": 10999 }, { "epoch": 0.5864477261822253, "grad_norm": 3.035353183746338, "learning_rate": 7.8002e-06, "loss": 0.0481, "step": 11000 }, { "epoch": 0.5865010396118783, "grad_norm": 2.781621217727661, "learning_rate": 7.800000000000002e-06, "loss": 0.0258, "step": 11001 }, { "epoch": 0.5865543530415311, "grad_norm": 3.011098861694336, "learning_rate": 7.799800000000001e-06, "loss": 0.0541, "step": 11002 }, { "epoch": 0.5866076664711841, "grad_norm": 1.4982377290725708, "learning_rate": 7.7996e-06, "loss": 0.0278, "step": 11003 }, { "epoch": 0.586660979900837, "grad_norm": 1.5115591287612915, "learning_rate": 7.7994e-06, "loss": 0.0338, "step": 11004 }, { "epoch": 0.58671429333049, "grad_norm": 2.4103469848632812, "learning_rate": 7.7992e-06, "loss": 0.0302, "step": 11005 }, { "epoch": 0.5867676067601428, "grad_norm": 3.3234994411468506, "learning_rate": 7.799000000000001e-06, "loss": 0.0455, "step": 11006 }, { "epoch": 0.5868209201897958, "grad_norm": 2.2018613815307617, "learning_rate": 7.7988e-06, "loss": 0.0295, "step": 11007 }, { "epoch": 0.5868742336194488, "grad_norm": 1.5199342966079712, "learning_rate": 7.7986e-06, "loss": 0.0266, "step": 11008 }, { "epoch": 0.5869275470491017, "grad_norm": 2.721132278442383, "learning_rate": 7.7984e-06, "loss": 0.0289, "step": 11009 }, { "epoch": 0.5869808604787546, "grad_norm": 2.0873003005981445, "learning_rate": 7.798200000000001e-06, "loss": 0.0514, "step": 11010 }, { "epoch": 0.5870341739084075, "grad_norm": 2.185790777206421, "learning_rate": 7.798e-06, "loss": 0.029, "step": 11011 }, { "epoch": 0.5870874873380605, "grad_norm": 2.3214852809906006, "learning_rate": 7.7978e-06, "loss": 0.0232, "step": 11012 }, { "epoch": 0.5871408007677134, "grad_norm": 1.188858151435852, "learning_rate": 7.797600000000001e-06, "loss": 0.0108, "step": 11013 }, { "epoch": 0.5871941141973663, "grad_norm": 2.602113962173462, "learning_rate": 7.797400000000001e-06, "loss": 0.0486, "step": 11014 }, { "epoch": 0.5872474276270192, "grad_norm": 1.4667376279830933, "learning_rate": 7.7972e-06, "loss": 0.0298, "step": 11015 }, { "epoch": 0.5873007410566722, "grad_norm": 2.5965425968170166, "learning_rate": 7.797e-06, "loss": 0.0317, "step": 11016 }, { "epoch": 0.587354054486325, "grad_norm": 3.763878345489502, "learning_rate": 7.796800000000001e-06, "loss": 0.0859, "step": 11017 }, { "epoch": 0.587407367915978, "grad_norm": 2.571305274963379, "learning_rate": 7.796600000000001e-06, "loss": 0.0258, "step": 11018 }, { "epoch": 0.587460681345631, "grad_norm": 1.945755958557129, "learning_rate": 7.7964e-06, "loss": 0.0343, "step": 11019 }, { "epoch": 0.5875139947752839, "grad_norm": 2.653587818145752, "learning_rate": 7.7962e-06, "loss": 0.0216, "step": 11020 }, { "epoch": 0.5875673082049369, "grad_norm": 4.640862941741943, "learning_rate": 7.796e-06, "loss": 0.0337, "step": 11021 }, { "epoch": 0.5876206216345897, "grad_norm": 2.411644458770752, "learning_rate": 7.7958e-06, "loss": 0.0452, "step": 11022 }, { "epoch": 0.5876739350642427, "grad_norm": 2.0019588470458984, "learning_rate": 7.7956e-06, "loss": 0.0277, "step": 11023 }, { "epoch": 0.5877272484938956, "grad_norm": 2.4569692611694336, "learning_rate": 7.7954e-06, "loss": 0.0505, "step": 11024 }, { "epoch": 0.5877805619235486, "grad_norm": 2.8425655364990234, "learning_rate": 7.795200000000001e-06, "loss": 0.0667, "step": 11025 }, { "epoch": 0.5878338753532014, "grad_norm": 2.6727402210235596, "learning_rate": 7.795e-06, "loss": 0.0227, "step": 11026 }, { "epoch": 0.5878871887828544, "grad_norm": 2.5178332328796387, "learning_rate": 7.7948e-06, "loss": 0.031, "step": 11027 }, { "epoch": 0.5879405022125074, "grad_norm": 2.5959391593933105, "learning_rate": 7.794600000000002e-06, "loss": 0.048, "step": 11028 }, { "epoch": 0.5879938156421602, "grad_norm": 1.7178826332092285, "learning_rate": 7.794400000000001e-06, "loss": 0.025, "step": 11029 }, { "epoch": 0.5880471290718132, "grad_norm": 2.3857078552246094, "learning_rate": 7.7942e-06, "loss": 0.0454, "step": 11030 }, { "epoch": 0.5881004425014661, "grad_norm": 1.1994203329086304, "learning_rate": 7.794e-06, "loss": 0.0286, "step": 11031 }, { "epoch": 0.5881537559311191, "grad_norm": 1.9907509088516235, "learning_rate": 7.7938e-06, "loss": 0.0188, "step": 11032 }, { "epoch": 0.5882070693607719, "grad_norm": 5.24990177154541, "learning_rate": 7.793600000000001e-06, "loss": 0.0554, "step": 11033 }, { "epoch": 0.5882603827904249, "grad_norm": 1.5747673511505127, "learning_rate": 7.7934e-06, "loss": 0.0263, "step": 11034 }, { "epoch": 0.5883136962200778, "grad_norm": 1.798383116722107, "learning_rate": 7.7932e-06, "loss": 0.0308, "step": 11035 }, { "epoch": 0.5883670096497308, "grad_norm": 1.7512449026107788, "learning_rate": 7.793e-06, "loss": 0.0213, "step": 11036 }, { "epoch": 0.5884203230793837, "grad_norm": 2.5880963802337646, "learning_rate": 7.792800000000001e-06, "loss": 0.0472, "step": 11037 }, { "epoch": 0.5884736365090366, "grad_norm": 3.4775779247283936, "learning_rate": 7.7926e-06, "loss": 0.0579, "step": 11038 }, { "epoch": 0.5885269499386896, "grad_norm": 1.9244400262832642, "learning_rate": 7.792400000000002e-06, "loss": 0.0297, "step": 11039 }, { "epoch": 0.5885802633683425, "grad_norm": 2.556535243988037, "learning_rate": 7.792200000000001e-06, "loss": 0.0635, "step": 11040 }, { "epoch": 0.5886335767979954, "grad_norm": 1.4800177812576294, "learning_rate": 7.792000000000001e-06, "loss": 0.0199, "step": 11041 }, { "epoch": 0.5886868902276483, "grad_norm": 1.2577273845672607, "learning_rate": 7.7918e-06, "loss": 0.0267, "step": 11042 }, { "epoch": 0.5887402036573013, "grad_norm": 1.691935658454895, "learning_rate": 7.7916e-06, "loss": 0.0348, "step": 11043 }, { "epoch": 0.5887935170869542, "grad_norm": 2.597273826599121, "learning_rate": 7.791400000000001e-06, "loss": 0.0503, "step": 11044 }, { "epoch": 0.5888468305166071, "grad_norm": 1.5670381784439087, "learning_rate": 7.791200000000001e-06, "loss": 0.0232, "step": 11045 }, { "epoch": 0.5889001439462601, "grad_norm": 2.2046730518341064, "learning_rate": 7.791e-06, "loss": 0.0357, "step": 11046 }, { "epoch": 0.588953457375913, "grad_norm": 4.246927738189697, "learning_rate": 7.7908e-06, "loss": 0.0487, "step": 11047 }, { "epoch": 0.589006770805566, "grad_norm": 2.6292693614959717, "learning_rate": 7.7906e-06, "loss": 0.0595, "step": 11048 }, { "epoch": 0.5890600842352188, "grad_norm": 2.6713898181915283, "learning_rate": 7.7904e-06, "loss": 0.044, "step": 11049 }, { "epoch": 0.5891133976648718, "grad_norm": 2.6093106269836426, "learning_rate": 7.7902e-06, "loss": 0.0455, "step": 11050 }, { "epoch": 0.5891667110945247, "grad_norm": 2.4986329078674316, "learning_rate": 7.790000000000002e-06, "loss": 0.0619, "step": 11051 }, { "epoch": 0.5892200245241777, "grad_norm": 2.3182926177978516, "learning_rate": 7.789800000000001e-06, "loss": 0.0371, "step": 11052 }, { "epoch": 0.5892733379538305, "grad_norm": 1.715370535850525, "learning_rate": 7.7896e-06, "loss": 0.0219, "step": 11053 }, { "epoch": 0.5893266513834835, "grad_norm": 1.6170291900634766, "learning_rate": 7.7894e-06, "loss": 0.0338, "step": 11054 }, { "epoch": 0.5893799648131365, "grad_norm": 1.8412697315216064, "learning_rate": 7.7892e-06, "loss": 0.0302, "step": 11055 }, { "epoch": 0.5894332782427893, "grad_norm": 1.921454668045044, "learning_rate": 7.789000000000001e-06, "loss": 0.0446, "step": 11056 }, { "epoch": 0.5894865916724423, "grad_norm": 1.3834683895111084, "learning_rate": 7.7888e-06, "loss": 0.0202, "step": 11057 }, { "epoch": 0.5895399051020952, "grad_norm": 1.3880902528762817, "learning_rate": 7.7886e-06, "loss": 0.0236, "step": 11058 }, { "epoch": 0.5895932185317482, "grad_norm": 1.8500958681106567, "learning_rate": 7.7884e-06, "loss": 0.0201, "step": 11059 }, { "epoch": 0.589646531961401, "grad_norm": 1.2883588075637817, "learning_rate": 7.7882e-06, "loss": 0.0117, "step": 11060 }, { "epoch": 0.589699845391054, "grad_norm": 3.2084617614746094, "learning_rate": 7.788e-06, "loss": 0.081, "step": 11061 }, { "epoch": 0.5897531588207069, "grad_norm": 1.4211962223052979, "learning_rate": 7.7878e-06, "loss": 0.0107, "step": 11062 }, { "epoch": 0.5898064722503599, "grad_norm": 4.727575302124023, "learning_rate": 7.787600000000001e-06, "loss": 0.0721, "step": 11063 }, { "epoch": 0.5898597856800128, "grad_norm": 1.5203489065170288, "learning_rate": 7.787400000000001e-06, "loss": 0.0146, "step": 11064 }, { "epoch": 0.5899130991096657, "grad_norm": 1.9922982454299927, "learning_rate": 7.7872e-06, "loss": 0.0313, "step": 11065 }, { "epoch": 0.5899664125393187, "grad_norm": 3.1799356937408447, "learning_rate": 7.787e-06, "loss": 0.0346, "step": 11066 }, { "epoch": 0.5900197259689716, "grad_norm": 3.5282795429229736, "learning_rate": 7.786800000000001e-06, "loss": 0.0793, "step": 11067 }, { "epoch": 0.5900730393986245, "grad_norm": 1.789920687675476, "learning_rate": 7.786600000000001e-06, "loss": 0.0203, "step": 11068 }, { "epoch": 0.5901263528282774, "grad_norm": 3.932810068130493, "learning_rate": 7.7864e-06, "loss": 0.0267, "step": 11069 }, { "epoch": 0.5901796662579304, "grad_norm": 3.0920894145965576, "learning_rate": 7.7862e-06, "loss": 0.0242, "step": 11070 }, { "epoch": 0.5902329796875833, "grad_norm": 4.157209873199463, "learning_rate": 7.786e-06, "loss": 0.0462, "step": 11071 }, { "epoch": 0.5902862931172362, "grad_norm": 3.6129040718078613, "learning_rate": 7.785800000000001e-06, "loss": 0.0467, "step": 11072 }, { "epoch": 0.5903396065468891, "grad_norm": 2.7965798377990723, "learning_rate": 7.7856e-06, "loss": 0.0217, "step": 11073 }, { "epoch": 0.5903929199765421, "grad_norm": 1.986951470375061, "learning_rate": 7.7854e-06, "loss": 0.0244, "step": 11074 }, { "epoch": 0.5904462334061951, "grad_norm": 2.0590131282806396, "learning_rate": 7.785200000000001e-06, "loss": 0.0405, "step": 11075 }, { "epoch": 0.5904995468358479, "grad_norm": 2.274486780166626, "learning_rate": 7.785000000000001e-06, "loss": 0.0483, "step": 11076 }, { "epoch": 0.5905528602655009, "grad_norm": 2.411050319671631, "learning_rate": 7.7848e-06, "loss": 0.0113, "step": 11077 }, { "epoch": 0.5906061736951538, "grad_norm": 1.1229281425476074, "learning_rate": 7.784600000000002e-06, "loss": 0.0154, "step": 11078 }, { "epoch": 0.5906594871248068, "grad_norm": 1.4823392629623413, "learning_rate": 7.784400000000001e-06, "loss": 0.0236, "step": 11079 }, { "epoch": 0.5907128005544596, "grad_norm": 2.036569356918335, "learning_rate": 7.7842e-06, "loss": 0.0277, "step": 11080 }, { "epoch": 0.5907661139841126, "grad_norm": 1.7717844247817993, "learning_rate": 7.784e-06, "loss": 0.0345, "step": 11081 }, { "epoch": 0.5908194274137655, "grad_norm": 2.051523447036743, "learning_rate": 7.7838e-06, "loss": 0.0374, "step": 11082 }, { "epoch": 0.5908727408434185, "grad_norm": 2.1350297927856445, "learning_rate": 7.783600000000001e-06, "loss": 0.0223, "step": 11083 }, { "epoch": 0.5909260542730714, "grad_norm": 2.195209264755249, "learning_rate": 7.7834e-06, "loss": 0.0484, "step": 11084 }, { "epoch": 0.5909793677027243, "grad_norm": 1.5584837198257446, "learning_rate": 7.7832e-06, "loss": 0.0292, "step": 11085 }, { "epoch": 0.5910326811323773, "grad_norm": 2.3852696418762207, "learning_rate": 7.783e-06, "loss": 0.031, "step": 11086 }, { "epoch": 0.5910859945620301, "grad_norm": 2.713630437850952, "learning_rate": 7.782800000000001e-06, "loss": 0.0327, "step": 11087 }, { "epoch": 0.5911393079916831, "grad_norm": 3.6299850940704346, "learning_rate": 7.7826e-06, "loss": 0.0569, "step": 11088 }, { "epoch": 0.591192621421336, "grad_norm": 3.278634786605835, "learning_rate": 7.7824e-06, "loss": 0.0331, "step": 11089 }, { "epoch": 0.591245934850989, "grad_norm": 4.924698352813721, "learning_rate": 7.782200000000001e-06, "loss": 0.0649, "step": 11090 }, { "epoch": 0.5912992482806418, "grad_norm": 3.728135824203491, "learning_rate": 7.782000000000001e-06, "loss": 0.0577, "step": 11091 }, { "epoch": 0.5913525617102948, "grad_norm": 2.107753038406372, "learning_rate": 7.7818e-06, "loss": 0.0343, "step": 11092 }, { "epoch": 0.5914058751399478, "grad_norm": 2.4246907234191895, "learning_rate": 7.7816e-06, "loss": 0.0361, "step": 11093 }, { "epoch": 0.5914591885696007, "grad_norm": 3.6871578693389893, "learning_rate": 7.781400000000001e-06, "loss": 0.0738, "step": 11094 }, { "epoch": 0.5915125019992536, "grad_norm": 4.485618591308594, "learning_rate": 7.781200000000001e-06, "loss": 0.0417, "step": 11095 }, { "epoch": 0.5915658154289065, "grad_norm": 1.4628969430923462, "learning_rate": 7.781e-06, "loss": 0.0231, "step": 11096 }, { "epoch": 0.5916191288585595, "grad_norm": 2.6897995471954346, "learning_rate": 7.7808e-06, "loss": 0.0326, "step": 11097 }, { "epoch": 0.5916724422882124, "grad_norm": 2.4298503398895264, "learning_rate": 7.7806e-06, "loss": 0.0516, "step": 11098 }, { "epoch": 0.5917257557178653, "grad_norm": 1.5345574617385864, "learning_rate": 7.780400000000001e-06, "loss": 0.0204, "step": 11099 }, { "epoch": 0.5917790691475182, "grad_norm": 1.4704906940460205, "learning_rate": 7.7802e-06, "loss": 0.0205, "step": 11100 }, { "epoch": 0.5918323825771712, "grad_norm": 1.0881012678146362, "learning_rate": 7.78e-06, "loss": 0.0157, "step": 11101 }, { "epoch": 0.5918856960068242, "grad_norm": 1.6628408432006836, "learning_rate": 7.779800000000001e-06, "loss": 0.0131, "step": 11102 }, { "epoch": 0.591939009436477, "grad_norm": 1.9689521789550781, "learning_rate": 7.779600000000001e-06, "loss": 0.0379, "step": 11103 }, { "epoch": 0.59199232286613, "grad_norm": 1.7466541528701782, "learning_rate": 7.7794e-06, "loss": 0.0243, "step": 11104 }, { "epoch": 0.5920456362957829, "grad_norm": 2.5842812061309814, "learning_rate": 7.7792e-06, "loss": 0.049, "step": 11105 }, { "epoch": 0.5920989497254359, "grad_norm": 2.26289963722229, "learning_rate": 7.779000000000001e-06, "loss": 0.0423, "step": 11106 }, { "epoch": 0.5921522631550887, "grad_norm": 1.3591457605361938, "learning_rate": 7.7788e-06, "loss": 0.0255, "step": 11107 }, { "epoch": 0.5922055765847417, "grad_norm": 5.507277011871338, "learning_rate": 7.7786e-06, "loss": 0.0523, "step": 11108 }, { "epoch": 0.5922588900143946, "grad_norm": 1.3702592849731445, "learning_rate": 7.7784e-06, "loss": 0.0204, "step": 11109 }, { "epoch": 0.5923122034440476, "grad_norm": 1.7364720106124878, "learning_rate": 7.7782e-06, "loss": 0.0261, "step": 11110 }, { "epoch": 0.5923655168737005, "grad_norm": 2.6747255325317383, "learning_rate": 7.778e-06, "loss": 0.0231, "step": 11111 }, { "epoch": 0.5924188303033534, "grad_norm": 1.0809268951416016, "learning_rate": 7.7778e-06, "loss": 0.0102, "step": 11112 }, { "epoch": 0.5924721437330064, "grad_norm": 2.557812452316284, "learning_rate": 7.7776e-06, "loss": 0.0449, "step": 11113 }, { "epoch": 0.5925254571626593, "grad_norm": 1.8369437456130981, "learning_rate": 7.777400000000001e-06, "loss": 0.0302, "step": 11114 }, { "epoch": 0.5925787705923122, "grad_norm": 2.4342563152313232, "learning_rate": 7.7772e-06, "loss": 0.0275, "step": 11115 }, { "epoch": 0.5926320840219651, "grad_norm": 2.1480629444122314, "learning_rate": 7.777e-06, "loss": 0.0237, "step": 11116 }, { "epoch": 0.5926853974516181, "grad_norm": 1.5924835205078125, "learning_rate": 7.776800000000002e-06, "loss": 0.0205, "step": 11117 }, { "epoch": 0.592738710881271, "grad_norm": 6.537583351135254, "learning_rate": 7.776600000000001e-06, "loss": 0.1068, "step": 11118 }, { "epoch": 0.5927920243109239, "grad_norm": 5.325036525726318, "learning_rate": 7.7764e-06, "loss": 0.0511, "step": 11119 }, { "epoch": 0.5928453377405769, "grad_norm": 3.477313280105591, "learning_rate": 7.7762e-06, "loss": 0.0476, "step": 11120 }, { "epoch": 0.5928986511702298, "grad_norm": 1.3564399480819702, "learning_rate": 7.776e-06, "loss": 0.0196, "step": 11121 }, { "epoch": 0.5929519645998828, "grad_norm": 1.273206114768982, "learning_rate": 7.775800000000001e-06, "loss": 0.0242, "step": 11122 }, { "epoch": 0.5930052780295356, "grad_norm": 2.989300489425659, "learning_rate": 7.7756e-06, "loss": 0.039, "step": 11123 }, { "epoch": 0.5930585914591886, "grad_norm": 2.6600069999694824, "learning_rate": 7.7754e-06, "loss": 0.0514, "step": 11124 }, { "epoch": 0.5931119048888415, "grad_norm": 1.756917119026184, "learning_rate": 7.7752e-06, "loss": 0.0162, "step": 11125 }, { "epoch": 0.5931652183184944, "grad_norm": 2.435211658477783, "learning_rate": 7.775000000000001e-06, "loss": 0.0128, "step": 11126 }, { "epoch": 0.5932185317481473, "grad_norm": 2.0948617458343506, "learning_rate": 7.7748e-06, "loss": 0.0178, "step": 11127 }, { "epoch": 0.5932718451778003, "grad_norm": 2.1624679565429688, "learning_rate": 7.774600000000002e-06, "loss": 0.0424, "step": 11128 }, { "epoch": 0.5933251586074532, "grad_norm": 2.455575704574585, "learning_rate": 7.774400000000001e-06, "loss": 0.0348, "step": 11129 }, { "epoch": 0.5933784720371061, "grad_norm": 3.191891670227051, "learning_rate": 7.774200000000001e-06, "loss": 0.0471, "step": 11130 }, { "epoch": 0.5934317854667591, "grad_norm": 1.0587844848632812, "learning_rate": 7.774e-06, "loss": 0.0107, "step": 11131 }, { "epoch": 0.593485098896412, "grad_norm": 3.6025915145874023, "learning_rate": 7.7738e-06, "loss": 0.0664, "step": 11132 }, { "epoch": 0.593538412326065, "grad_norm": 3.0673861503601074, "learning_rate": 7.773600000000001e-06, "loss": 0.0256, "step": 11133 }, { "epoch": 0.5935917257557178, "grad_norm": 4.71596097946167, "learning_rate": 7.7734e-06, "loss": 0.0357, "step": 11134 }, { "epoch": 0.5936450391853708, "grad_norm": 3.1650540828704834, "learning_rate": 7.7732e-06, "loss": 0.0144, "step": 11135 }, { "epoch": 0.5936983526150237, "grad_norm": 1.1715962886810303, "learning_rate": 7.773e-06, "loss": 0.0221, "step": 11136 }, { "epoch": 0.5937516660446767, "grad_norm": 2.84133243560791, "learning_rate": 7.7728e-06, "loss": 0.062, "step": 11137 }, { "epoch": 0.5938049794743295, "grad_norm": 1.8684238195419312, "learning_rate": 7.7726e-06, "loss": 0.0186, "step": 11138 }, { "epoch": 0.5938582929039825, "grad_norm": 1.2602401971817017, "learning_rate": 7.7724e-06, "loss": 0.0111, "step": 11139 }, { "epoch": 0.5939116063336355, "grad_norm": 2.377101182937622, "learning_rate": 7.772200000000002e-06, "loss": 0.0361, "step": 11140 }, { "epoch": 0.5939649197632884, "grad_norm": 4.162419319152832, "learning_rate": 7.772000000000001e-06, "loss": 0.0719, "step": 11141 }, { "epoch": 0.5940182331929413, "grad_norm": 1.9414602518081665, "learning_rate": 7.7718e-06, "loss": 0.0205, "step": 11142 }, { "epoch": 0.5940715466225942, "grad_norm": 3.5040416717529297, "learning_rate": 7.7716e-06, "loss": 0.0507, "step": 11143 }, { "epoch": 0.5941248600522472, "grad_norm": 1.4298906326293945, "learning_rate": 7.771400000000002e-06, "loss": 0.0193, "step": 11144 }, { "epoch": 0.5941781734819, "grad_norm": 4.437478542327881, "learning_rate": 7.771200000000001e-06, "loss": 0.0537, "step": 11145 }, { "epoch": 0.594231486911553, "grad_norm": 1.7753589153289795, "learning_rate": 7.771e-06, "loss": 0.0311, "step": 11146 }, { "epoch": 0.5942848003412059, "grad_norm": 1.8283923864364624, "learning_rate": 7.7708e-06, "loss": 0.0275, "step": 11147 }, { "epoch": 0.5943381137708589, "grad_norm": 2.677354097366333, "learning_rate": 7.7706e-06, "loss": 0.0371, "step": 11148 }, { "epoch": 0.5943914272005119, "grad_norm": 4.668371200561523, "learning_rate": 7.7704e-06, "loss": 0.04, "step": 11149 }, { "epoch": 0.5944447406301647, "grad_norm": 2.108562469482422, "learning_rate": 7.7702e-06, "loss": 0.0297, "step": 11150 }, { "epoch": 0.5944980540598177, "grad_norm": 1.981516718864441, "learning_rate": 7.77e-06, "loss": 0.0261, "step": 11151 }, { "epoch": 0.5945513674894706, "grad_norm": 4.009203910827637, "learning_rate": 7.769800000000001e-06, "loss": 0.0547, "step": 11152 }, { "epoch": 0.5946046809191236, "grad_norm": 1.704380989074707, "learning_rate": 7.769600000000001e-06, "loss": 0.0241, "step": 11153 }, { "epoch": 0.5946579943487764, "grad_norm": 1.6828632354736328, "learning_rate": 7.7694e-06, "loss": 0.0363, "step": 11154 }, { "epoch": 0.5947113077784294, "grad_norm": 2.0137839317321777, "learning_rate": 7.769200000000002e-06, "loss": 0.0338, "step": 11155 }, { "epoch": 0.5947646212080823, "grad_norm": 2.6282854080200195, "learning_rate": 7.769000000000001e-06, "loss": 0.0301, "step": 11156 }, { "epoch": 0.5948179346377352, "grad_norm": 1.6348779201507568, "learning_rate": 7.768800000000001e-06, "loss": 0.033, "step": 11157 }, { "epoch": 0.5948712480673882, "grad_norm": 2.440706968307495, "learning_rate": 7.7686e-06, "loss": 0.0382, "step": 11158 }, { "epoch": 0.5949245614970411, "grad_norm": 2.2253658771514893, "learning_rate": 7.7684e-06, "loss": 0.0123, "step": 11159 }, { "epoch": 0.5949778749266941, "grad_norm": 2.6699628829956055, "learning_rate": 7.7682e-06, "loss": 0.048, "step": 11160 }, { "epoch": 0.5950311883563469, "grad_norm": 3.0647170543670654, "learning_rate": 7.768e-06, "loss": 0.0562, "step": 11161 }, { "epoch": 0.5950845017859999, "grad_norm": 2.622269630432129, "learning_rate": 7.7678e-06, "loss": 0.0413, "step": 11162 }, { "epoch": 0.5951378152156528, "grad_norm": 2.9678523540496826, "learning_rate": 7.7676e-06, "loss": 0.053, "step": 11163 }, { "epoch": 0.5951911286453058, "grad_norm": 2.65622878074646, "learning_rate": 7.767400000000001e-06, "loss": 0.0501, "step": 11164 }, { "epoch": 0.5952444420749586, "grad_norm": 1.4448851346969604, "learning_rate": 7.7672e-06, "loss": 0.0214, "step": 11165 }, { "epoch": 0.5952977555046116, "grad_norm": 2.1248273849487305, "learning_rate": 7.767e-06, "loss": 0.0371, "step": 11166 }, { "epoch": 0.5953510689342646, "grad_norm": 2.088352918624878, "learning_rate": 7.766800000000002e-06, "loss": 0.0328, "step": 11167 }, { "epoch": 0.5954043823639175, "grad_norm": 4.2880330085754395, "learning_rate": 7.766600000000001e-06, "loss": 0.0432, "step": 11168 }, { "epoch": 0.5954576957935704, "grad_norm": 4.056400775909424, "learning_rate": 7.7664e-06, "loss": 0.0434, "step": 11169 }, { "epoch": 0.5955110092232233, "grad_norm": 2.4266817569732666, "learning_rate": 7.7662e-06, "loss": 0.0368, "step": 11170 }, { "epoch": 0.5955643226528763, "grad_norm": 2.9794046878814697, "learning_rate": 7.766e-06, "loss": 0.0346, "step": 11171 }, { "epoch": 0.5956176360825292, "grad_norm": 1.6534817218780518, "learning_rate": 7.765800000000001e-06, "loss": 0.0218, "step": 11172 }, { "epoch": 0.5956709495121821, "grad_norm": 3.1643314361572266, "learning_rate": 7.7656e-06, "loss": 0.0578, "step": 11173 }, { "epoch": 0.595724262941835, "grad_norm": 2.4358787536621094, "learning_rate": 7.7654e-06, "loss": 0.0403, "step": 11174 }, { "epoch": 0.595777576371488, "grad_norm": 2.549619436264038, "learning_rate": 7.7652e-06, "loss": 0.0557, "step": 11175 }, { "epoch": 0.595830889801141, "grad_norm": 5.015720367431641, "learning_rate": 7.765000000000001e-06, "loss": 0.0502, "step": 11176 }, { "epoch": 0.5958842032307938, "grad_norm": 2.1873764991760254, "learning_rate": 7.7648e-06, "loss": 0.0298, "step": 11177 }, { "epoch": 0.5959375166604468, "grad_norm": 3.4249985218048096, "learning_rate": 7.764600000000002e-06, "loss": 0.069, "step": 11178 }, { "epoch": 0.5959908300900997, "grad_norm": 3.5164806842803955, "learning_rate": 7.764400000000001e-06, "loss": 0.0535, "step": 11179 }, { "epoch": 0.5960441435197527, "grad_norm": 2.4536170959472656, "learning_rate": 7.764200000000001e-06, "loss": 0.0156, "step": 11180 }, { "epoch": 0.5960974569494055, "grad_norm": 1.6375470161437988, "learning_rate": 7.764e-06, "loss": 0.0387, "step": 11181 }, { "epoch": 0.5961507703790585, "grad_norm": 2.7244858741760254, "learning_rate": 7.7638e-06, "loss": 0.0618, "step": 11182 }, { "epoch": 0.5962040838087114, "grad_norm": 1.7779182195663452, "learning_rate": 7.763600000000001e-06, "loss": 0.041, "step": 11183 }, { "epoch": 0.5962573972383644, "grad_norm": 3.505544662475586, "learning_rate": 7.763400000000001e-06, "loss": 0.0538, "step": 11184 }, { "epoch": 0.5963107106680172, "grad_norm": 1.392143964767456, "learning_rate": 7.7632e-06, "loss": 0.0187, "step": 11185 }, { "epoch": 0.5963640240976702, "grad_norm": 2.4885642528533936, "learning_rate": 7.763e-06, "loss": 0.0478, "step": 11186 }, { "epoch": 0.5964173375273232, "grad_norm": 3.1327524185180664, "learning_rate": 7.7628e-06, "loss": 0.0605, "step": 11187 }, { "epoch": 0.596470650956976, "grad_norm": 1.4504183530807495, "learning_rate": 7.762600000000001e-06, "loss": 0.0233, "step": 11188 }, { "epoch": 0.596523964386629, "grad_norm": 3.379032611846924, "learning_rate": 7.7624e-06, "loss": 0.0301, "step": 11189 }, { "epoch": 0.5965772778162819, "grad_norm": 1.596935510635376, "learning_rate": 7.762200000000002e-06, "loss": 0.0202, "step": 11190 }, { "epoch": 0.5966305912459349, "grad_norm": 1.8497633934020996, "learning_rate": 7.762000000000001e-06, "loss": 0.0273, "step": 11191 }, { "epoch": 0.5966839046755877, "grad_norm": 3.331307888031006, "learning_rate": 7.7618e-06, "loss": 0.0529, "step": 11192 }, { "epoch": 0.5967372181052407, "grad_norm": 2.0706534385681152, "learning_rate": 7.7616e-06, "loss": 0.0351, "step": 11193 }, { "epoch": 0.5967905315348936, "grad_norm": 2.1825780868530273, "learning_rate": 7.761400000000002e-06, "loss": 0.033, "step": 11194 }, { "epoch": 0.5968438449645466, "grad_norm": 4.681847095489502, "learning_rate": 7.761200000000001e-06, "loss": 0.045, "step": 11195 }, { "epoch": 0.5968971583941995, "grad_norm": 1.6189221143722534, "learning_rate": 7.761e-06, "loss": 0.036, "step": 11196 }, { "epoch": 0.5969504718238524, "grad_norm": 2.8675448894500732, "learning_rate": 7.7608e-06, "loss": 0.0429, "step": 11197 }, { "epoch": 0.5970037852535054, "grad_norm": 1.1692173480987549, "learning_rate": 7.7606e-06, "loss": 0.0173, "step": 11198 }, { "epoch": 0.5970570986831583, "grad_norm": 2.953425645828247, "learning_rate": 7.7604e-06, "loss": 0.0359, "step": 11199 }, { "epoch": 0.5971104121128112, "grad_norm": 1.1335281133651733, "learning_rate": 7.7602e-06, "loss": 0.0135, "step": 11200 }, { "epoch": 0.5971637255424641, "grad_norm": 2.6740920543670654, "learning_rate": 7.76e-06, "loss": 0.062, "step": 11201 }, { "epoch": 0.5972170389721171, "grad_norm": 4.666581630706787, "learning_rate": 7.759800000000001e-06, "loss": 0.0441, "step": 11202 }, { "epoch": 0.59727035240177, "grad_norm": 3.2314257621765137, "learning_rate": 7.759600000000001e-06, "loss": 0.0495, "step": 11203 }, { "epoch": 0.5973236658314229, "grad_norm": 2.532327890396118, "learning_rate": 7.7594e-06, "loss": 0.0512, "step": 11204 }, { "epoch": 0.5973769792610759, "grad_norm": 2.653994560241699, "learning_rate": 7.759200000000002e-06, "loss": 0.0403, "step": 11205 }, { "epoch": 0.5974302926907288, "grad_norm": 2.5506155490875244, "learning_rate": 7.759000000000001e-06, "loss": 0.0312, "step": 11206 }, { "epoch": 0.5974836061203818, "grad_norm": 1.9362918138504028, "learning_rate": 7.758800000000001e-06, "loss": 0.033, "step": 11207 }, { "epoch": 0.5975369195500346, "grad_norm": 2.1457014083862305, "learning_rate": 7.7586e-06, "loss": 0.021, "step": 11208 }, { "epoch": 0.5975902329796876, "grad_norm": 1.6646579504013062, "learning_rate": 7.7584e-06, "loss": 0.0372, "step": 11209 }, { "epoch": 0.5976435464093405, "grad_norm": 2.5281195640563965, "learning_rate": 7.7582e-06, "loss": 0.0381, "step": 11210 }, { "epoch": 0.5976968598389935, "grad_norm": 3.0053391456604004, "learning_rate": 7.758000000000001e-06, "loss": 0.0547, "step": 11211 }, { "epoch": 0.5977501732686463, "grad_norm": 3.426076889038086, "learning_rate": 7.7578e-06, "loss": 0.0545, "step": 11212 }, { "epoch": 0.5978034866982993, "grad_norm": 2.5898244380950928, "learning_rate": 7.7576e-06, "loss": 0.0274, "step": 11213 }, { "epoch": 0.5978568001279523, "grad_norm": 1.2641637325286865, "learning_rate": 7.757400000000001e-06, "loss": 0.0151, "step": 11214 }, { "epoch": 0.5979101135576051, "grad_norm": 1.7796626091003418, "learning_rate": 7.757200000000001e-06, "loss": 0.0329, "step": 11215 }, { "epoch": 0.5979634269872581, "grad_norm": 3.121953248977661, "learning_rate": 7.757e-06, "loss": 0.0212, "step": 11216 }, { "epoch": 0.598016740416911, "grad_norm": 3.8153903484344482, "learning_rate": 7.756800000000002e-06, "loss": 0.0215, "step": 11217 }, { "epoch": 0.598070053846564, "grad_norm": 0.8049883842468262, "learning_rate": 7.756600000000001e-06, "loss": 0.0086, "step": 11218 }, { "epoch": 0.5981233672762168, "grad_norm": 2.2621753215789795, "learning_rate": 7.7564e-06, "loss": 0.0383, "step": 11219 }, { "epoch": 0.5981766807058698, "grad_norm": 2.1245498657226562, "learning_rate": 7.7562e-06, "loss": 0.0168, "step": 11220 }, { "epoch": 0.5982299941355227, "grad_norm": 2.4453201293945312, "learning_rate": 7.756e-06, "loss": 0.0385, "step": 11221 }, { "epoch": 0.5982833075651757, "grad_norm": 1.9982489347457886, "learning_rate": 7.755800000000001e-06, "loss": 0.0092, "step": 11222 }, { "epoch": 0.5983366209948286, "grad_norm": 1.9828327894210815, "learning_rate": 7.7556e-06, "loss": 0.0177, "step": 11223 }, { "epoch": 0.5983899344244815, "grad_norm": 3.071352958679199, "learning_rate": 7.7554e-06, "loss": 0.0487, "step": 11224 }, { "epoch": 0.5984432478541345, "grad_norm": 3.4277241230010986, "learning_rate": 7.7552e-06, "loss": 0.0383, "step": 11225 }, { "epoch": 0.5984965612837874, "grad_norm": 1.732053279876709, "learning_rate": 7.755000000000001e-06, "loss": 0.031, "step": 11226 }, { "epoch": 0.5985498747134403, "grad_norm": 4.391106128692627, "learning_rate": 7.7548e-06, "loss": 0.0501, "step": 11227 }, { "epoch": 0.5986031881430932, "grad_norm": 1.644219994544983, "learning_rate": 7.7546e-06, "loss": 0.013, "step": 11228 }, { "epoch": 0.5986565015727462, "grad_norm": 1.6288387775421143, "learning_rate": 7.754400000000002e-06, "loss": 0.0086, "step": 11229 }, { "epoch": 0.5987098150023991, "grad_norm": 2.3935794830322266, "learning_rate": 7.754200000000001e-06, "loss": 0.025, "step": 11230 }, { "epoch": 0.598763128432052, "grad_norm": 2.217221975326538, "learning_rate": 7.754e-06, "loss": 0.0182, "step": 11231 }, { "epoch": 0.598816441861705, "grad_norm": 4.287597179412842, "learning_rate": 7.7538e-06, "loss": 0.0539, "step": 11232 }, { "epoch": 0.5988697552913579, "grad_norm": 2.387770414352417, "learning_rate": 7.753600000000001e-06, "loss": 0.0229, "step": 11233 }, { "epoch": 0.5989230687210109, "grad_norm": 3.6401524543762207, "learning_rate": 7.753400000000001e-06, "loss": 0.0247, "step": 11234 }, { "epoch": 0.5989763821506637, "grad_norm": 2.6385536193847656, "learning_rate": 7.7532e-06, "loss": 0.0383, "step": 11235 }, { "epoch": 0.5990296955803167, "grad_norm": 2.827205181121826, "learning_rate": 7.753e-06, "loss": 0.04, "step": 11236 }, { "epoch": 0.5990830090099696, "grad_norm": 2.2850146293640137, "learning_rate": 7.7528e-06, "loss": 0.0344, "step": 11237 }, { "epoch": 0.5991363224396226, "grad_norm": 2.8725106716156006, "learning_rate": 7.752600000000001e-06, "loss": 0.0437, "step": 11238 }, { "epoch": 0.5991896358692754, "grad_norm": 1.759906530380249, "learning_rate": 7.7524e-06, "loss": 0.0454, "step": 11239 }, { "epoch": 0.5992429492989284, "grad_norm": 1.171805739402771, "learning_rate": 7.7522e-06, "loss": 0.0127, "step": 11240 }, { "epoch": 0.5992962627285813, "grad_norm": 1.8350361585617065, "learning_rate": 7.752000000000001e-06, "loss": 0.0332, "step": 11241 }, { "epoch": 0.5993495761582343, "grad_norm": 2.9624688625335693, "learning_rate": 7.751800000000001e-06, "loss": 0.0458, "step": 11242 }, { "epoch": 0.5994028895878872, "grad_norm": 3.412794589996338, "learning_rate": 7.7516e-06, "loss": 0.0574, "step": 11243 }, { "epoch": 0.5994562030175401, "grad_norm": 3.4205076694488525, "learning_rate": 7.751400000000002e-06, "loss": 0.0465, "step": 11244 }, { "epoch": 0.5995095164471931, "grad_norm": 1.7204376459121704, "learning_rate": 7.751200000000001e-06, "loss": 0.0516, "step": 11245 }, { "epoch": 0.599562829876846, "grad_norm": 2.5329809188842773, "learning_rate": 7.751e-06, "loss": 0.0235, "step": 11246 }, { "epoch": 0.5996161433064989, "grad_norm": 2.1006200313568115, "learning_rate": 7.7508e-06, "loss": 0.0369, "step": 11247 }, { "epoch": 0.5996694567361518, "grad_norm": 1.6134918928146362, "learning_rate": 7.7506e-06, "loss": 0.0253, "step": 11248 }, { "epoch": 0.5997227701658048, "grad_norm": 2.531076431274414, "learning_rate": 7.7504e-06, "loss": 0.0253, "step": 11249 }, { "epoch": 0.5997760835954576, "grad_norm": 1.344056248664856, "learning_rate": 7.7502e-06, "loss": 0.0294, "step": 11250 }, { "epoch": 0.5998293970251106, "grad_norm": 2.9782490730285645, "learning_rate": 7.75e-06, "loss": 0.0255, "step": 11251 }, { "epoch": 0.5998827104547636, "grad_norm": 1.1211018562316895, "learning_rate": 7.7498e-06, "loss": 0.0181, "step": 11252 }, { "epoch": 0.5999360238844165, "grad_norm": 1.7560943365097046, "learning_rate": 7.749600000000001e-06, "loss": 0.0232, "step": 11253 }, { "epoch": 0.5999893373140694, "grad_norm": 2.336681365966797, "learning_rate": 7.7494e-06, "loss": 0.0207, "step": 11254 }, { "epoch": 0.6000426507437223, "grad_norm": 2.846829891204834, "learning_rate": 7.749200000000002e-06, "loss": 0.0702, "step": 11255 }, { "epoch": 0.6000959641733753, "grad_norm": 2.971951723098755, "learning_rate": 7.749000000000002e-06, "loss": 0.0339, "step": 11256 }, { "epoch": 0.6001492776030282, "grad_norm": 3.1548595428466797, "learning_rate": 7.748800000000001e-06, "loss": 0.0521, "step": 11257 }, { "epoch": 0.6002025910326811, "grad_norm": 2.8062744140625, "learning_rate": 7.7486e-06, "loss": 0.0441, "step": 11258 }, { "epoch": 0.600255904462334, "grad_norm": 2.1954081058502197, "learning_rate": 7.7484e-06, "loss": 0.0394, "step": 11259 }, { "epoch": 0.600309217891987, "grad_norm": 3.854541540145874, "learning_rate": 7.7482e-06, "loss": 0.0289, "step": 11260 }, { "epoch": 0.60036253132164, "grad_norm": 1.6012357473373413, "learning_rate": 7.748000000000001e-06, "loss": 0.028, "step": 11261 }, { "epoch": 0.6004158447512928, "grad_norm": 3.315300941467285, "learning_rate": 7.7478e-06, "loss": 0.0726, "step": 11262 }, { "epoch": 0.6004691581809458, "grad_norm": 1.3716641664505005, "learning_rate": 7.7476e-06, "loss": 0.0118, "step": 11263 }, { "epoch": 0.6005224716105987, "grad_norm": 1.232911229133606, "learning_rate": 7.7474e-06, "loss": 0.0162, "step": 11264 }, { "epoch": 0.6005757850402517, "grad_norm": 1.4559764862060547, "learning_rate": 7.747200000000001e-06, "loss": 0.0125, "step": 11265 }, { "epoch": 0.6006290984699045, "grad_norm": 5.5336103439331055, "learning_rate": 7.747e-06, "loss": 0.0794, "step": 11266 }, { "epoch": 0.6006824118995575, "grad_norm": 2.1220483779907227, "learning_rate": 7.746800000000002e-06, "loss": 0.036, "step": 11267 }, { "epoch": 0.6007357253292104, "grad_norm": 2.0377650260925293, "learning_rate": 7.746600000000001e-06, "loss": 0.0309, "step": 11268 }, { "epoch": 0.6007890387588634, "grad_norm": 1.0526118278503418, "learning_rate": 7.746400000000001e-06, "loss": 0.0196, "step": 11269 }, { "epoch": 0.6008423521885163, "grad_norm": 2.465968608856201, "learning_rate": 7.7462e-06, "loss": 0.0499, "step": 11270 }, { "epoch": 0.6008956656181692, "grad_norm": 1.6091113090515137, "learning_rate": 7.746e-06, "loss": 0.0276, "step": 11271 }, { "epoch": 0.6009489790478222, "grad_norm": 2.1989476680755615, "learning_rate": 7.745800000000001e-06, "loss": 0.0217, "step": 11272 }, { "epoch": 0.601002292477475, "grad_norm": 0.8747010827064514, "learning_rate": 7.745600000000001e-06, "loss": 0.008, "step": 11273 }, { "epoch": 0.601055605907128, "grad_norm": 1.0121040344238281, "learning_rate": 7.7454e-06, "loss": 0.0141, "step": 11274 }, { "epoch": 0.6011089193367809, "grad_norm": 2.0560684204101562, "learning_rate": 7.7452e-06, "loss": 0.0298, "step": 11275 }, { "epoch": 0.6011622327664339, "grad_norm": 3.7873289585113525, "learning_rate": 7.745e-06, "loss": 0.0612, "step": 11276 }, { "epoch": 0.6012155461960867, "grad_norm": 1.6617114543914795, "learning_rate": 7.7448e-06, "loss": 0.0427, "step": 11277 }, { "epoch": 0.6012688596257397, "grad_norm": 2.548093795776367, "learning_rate": 7.7446e-06, "loss": 0.032, "step": 11278 }, { "epoch": 0.6013221730553927, "grad_norm": 3.2623510360717773, "learning_rate": 7.744400000000002e-06, "loss": 0.0339, "step": 11279 }, { "epoch": 0.6013754864850456, "grad_norm": 2.4241600036621094, "learning_rate": 7.744200000000001e-06, "loss": 0.0461, "step": 11280 }, { "epoch": 0.6014287999146986, "grad_norm": 2.8930163383483887, "learning_rate": 7.744e-06, "loss": 0.0614, "step": 11281 }, { "epoch": 0.6014821133443514, "grad_norm": 2.544872999191284, "learning_rate": 7.7438e-06, "loss": 0.0392, "step": 11282 }, { "epoch": 0.6015354267740044, "grad_norm": 3.3399322032928467, "learning_rate": 7.743600000000002e-06, "loss": 0.017, "step": 11283 }, { "epoch": 0.6015887402036573, "grad_norm": 2.5814168453216553, "learning_rate": 7.743400000000001e-06, "loss": 0.0394, "step": 11284 }, { "epoch": 0.6016420536333102, "grad_norm": 1.0825481414794922, "learning_rate": 7.7432e-06, "loss": 0.0073, "step": 11285 }, { "epoch": 0.6016953670629631, "grad_norm": 1.908107042312622, "learning_rate": 7.743e-06, "loss": 0.0184, "step": 11286 }, { "epoch": 0.6017486804926161, "grad_norm": 0.9826886057853699, "learning_rate": 7.7428e-06, "loss": 0.0092, "step": 11287 }, { "epoch": 0.6018019939222691, "grad_norm": 2.0046443939208984, "learning_rate": 7.7426e-06, "loss": 0.0291, "step": 11288 }, { "epoch": 0.6018553073519219, "grad_norm": 3.2022907733917236, "learning_rate": 7.7424e-06, "loss": 0.0508, "step": 11289 }, { "epoch": 0.6019086207815749, "grad_norm": 3.933513641357422, "learning_rate": 7.7422e-06, "loss": 0.0522, "step": 11290 }, { "epoch": 0.6019619342112278, "grad_norm": 2.9288811683654785, "learning_rate": 7.742000000000001e-06, "loss": 0.0331, "step": 11291 }, { "epoch": 0.6020152476408808, "grad_norm": 1.8305199146270752, "learning_rate": 7.741800000000001e-06, "loss": 0.0345, "step": 11292 }, { "epoch": 0.6020685610705336, "grad_norm": 2.8951051235198975, "learning_rate": 7.7416e-06, "loss": 0.0349, "step": 11293 }, { "epoch": 0.6021218745001866, "grad_norm": 4.841465473175049, "learning_rate": 7.741400000000002e-06, "loss": 0.0411, "step": 11294 }, { "epoch": 0.6021751879298395, "grad_norm": 4.070362091064453, "learning_rate": 7.741200000000001e-06, "loss": 0.0447, "step": 11295 }, { "epoch": 0.6022285013594925, "grad_norm": 5.465495586395264, "learning_rate": 7.741000000000001e-06, "loss": 0.0449, "step": 11296 }, { "epoch": 0.6022818147891453, "grad_norm": 2.3297910690307617, "learning_rate": 7.7408e-06, "loss": 0.034, "step": 11297 }, { "epoch": 0.6023351282187983, "grad_norm": 1.9405546188354492, "learning_rate": 7.7406e-06, "loss": 0.0251, "step": 11298 }, { "epoch": 0.6023884416484513, "grad_norm": 2.2053322792053223, "learning_rate": 7.7404e-06, "loss": 0.0331, "step": 11299 }, { "epoch": 0.6024417550781042, "grad_norm": 1.4222943782806396, "learning_rate": 7.740200000000001e-06, "loss": 0.0195, "step": 11300 }, { "epoch": 0.6024950685077571, "grad_norm": 1.2877979278564453, "learning_rate": 7.74e-06, "loss": 0.0117, "step": 11301 }, { "epoch": 0.60254838193741, "grad_norm": 0.9288574457168579, "learning_rate": 7.7398e-06, "loss": 0.0045, "step": 11302 }, { "epoch": 0.602601695367063, "grad_norm": 2.966111660003662, "learning_rate": 7.739600000000001e-06, "loss": 0.0264, "step": 11303 }, { "epoch": 0.6026550087967159, "grad_norm": 3.083322286605835, "learning_rate": 7.7394e-06, "loss": 0.0447, "step": 11304 }, { "epoch": 0.6027083222263688, "grad_norm": 3.232820749282837, "learning_rate": 7.7392e-06, "loss": 0.0827, "step": 11305 }, { "epoch": 0.6027616356560217, "grad_norm": 1.0516846179962158, "learning_rate": 7.739000000000002e-06, "loss": 0.0231, "step": 11306 }, { "epoch": 0.6028149490856747, "grad_norm": 1.4367800951004028, "learning_rate": 7.738800000000001e-06, "loss": 0.0117, "step": 11307 }, { "epoch": 0.6028682625153277, "grad_norm": 2.9115610122680664, "learning_rate": 7.7386e-06, "loss": 0.0377, "step": 11308 }, { "epoch": 0.6029215759449805, "grad_norm": 1.342292308807373, "learning_rate": 7.7384e-06, "loss": 0.0146, "step": 11309 }, { "epoch": 0.6029748893746335, "grad_norm": 5.209212303161621, "learning_rate": 7.7382e-06, "loss": 0.0485, "step": 11310 }, { "epoch": 0.6030282028042864, "grad_norm": 1.5831804275512695, "learning_rate": 7.738000000000001e-06, "loss": 0.0164, "step": 11311 }, { "epoch": 0.6030815162339394, "grad_norm": 5.327182292938232, "learning_rate": 7.7378e-06, "loss": 0.0403, "step": 11312 }, { "epoch": 0.6031348296635922, "grad_norm": 2.7453577518463135, "learning_rate": 7.7376e-06, "loss": 0.0408, "step": 11313 }, { "epoch": 0.6031881430932452, "grad_norm": 2.140488624572754, "learning_rate": 7.7374e-06, "loss": 0.0341, "step": 11314 }, { "epoch": 0.6032414565228981, "grad_norm": 1.521987795829773, "learning_rate": 7.737200000000001e-06, "loss": 0.0277, "step": 11315 }, { "epoch": 0.603294769952551, "grad_norm": 1.8321218490600586, "learning_rate": 7.737e-06, "loss": 0.0314, "step": 11316 }, { "epoch": 0.603348083382204, "grad_norm": 1.6950733661651611, "learning_rate": 7.7368e-06, "loss": 0.0312, "step": 11317 }, { "epoch": 0.6034013968118569, "grad_norm": 3.9074923992156982, "learning_rate": 7.736600000000001e-06, "loss": 0.0393, "step": 11318 }, { "epoch": 0.6034547102415099, "grad_norm": 2.423250675201416, "learning_rate": 7.736400000000001e-06, "loss": 0.0235, "step": 11319 }, { "epoch": 0.6035080236711627, "grad_norm": 2.342289686203003, "learning_rate": 7.7362e-06, "loss": 0.0248, "step": 11320 }, { "epoch": 0.6035613371008157, "grad_norm": 2.057865858078003, "learning_rate": 7.736e-06, "loss": 0.0399, "step": 11321 }, { "epoch": 0.6036146505304686, "grad_norm": 1.8866684436798096, "learning_rate": 7.735800000000001e-06, "loss": 0.0285, "step": 11322 }, { "epoch": 0.6036679639601216, "grad_norm": 1.1594818830490112, "learning_rate": 7.735600000000001e-06, "loss": 0.0094, "step": 11323 }, { "epoch": 0.6037212773897744, "grad_norm": 1.7351529598236084, "learning_rate": 7.7354e-06, "loss": 0.0233, "step": 11324 }, { "epoch": 0.6037745908194274, "grad_norm": 1.594032883644104, "learning_rate": 7.7352e-06, "loss": 0.0292, "step": 11325 }, { "epoch": 0.6038279042490804, "grad_norm": 3.028656244277954, "learning_rate": 7.735e-06, "loss": 0.0498, "step": 11326 }, { "epoch": 0.6038812176787333, "grad_norm": 2.9831037521362305, "learning_rate": 7.734800000000001e-06, "loss": 0.0448, "step": 11327 }, { "epoch": 0.6039345311083862, "grad_norm": 3.3961501121520996, "learning_rate": 7.7346e-06, "loss": 0.0486, "step": 11328 }, { "epoch": 0.6039878445380391, "grad_norm": 2.8118441104888916, "learning_rate": 7.7344e-06, "loss": 0.0426, "step": 11329 }, { "epoch": 0.6040411579676921, "grad_norm": 3.858710527420044, "learning_rate": 7.734200000000001e-06, "loss": 0.0361, "step": 11330 }, { "epoch": 0.604094471397345, "grad_norm": 4.280018329620361, "learning_rate": 7.734e-06, "loss": 0.0656, "step": 11331 }, { "epoch": 0.6041477848269979, "grad_norm": 4.891190528869629, "learning_rate": 7.7338e-06, "loss": 0.0437, "step": 11332 }, { "epoch": 0.6042010982566508, "grad_norm": 3.0427348613739014, "learning_rate": 7.733600000000002e-06, "loss": 0.0318, "step": 11333 }, { "epoch": 0.6042544116863038, "grad_norm": 3.0913026332855225, "learning_rate": 7.733400000000001e-06, "loss": 0.0359, "step": 11334 }, { "epoch": 0.6043077251159568, "grad_norm": 2.955713987350464, "learning_rate": 7.7332e-06, "loss": 0.0579, "step": 11335 }, { "epoch": 0.6043610385456096, "grad_norm": 1.2815488576889038, "learning_rate": 7.733e-06, "loss": 0.0279, "step": 11336 }, { "epoch": 0.6044143519752626, "grad_norm": 1.837311863899231, "learning_rate": 7.7328e-06, "loss": 0.0397, "step": 11337 }, { "epoch": 0.6044676654049155, "grad_norm": 2.340872287750244, "learning_rate": 7.7326e-06, "loss": 0.025, "step": 11338 }, { "epoch": 0.6045209788345685, "grad_norm": 2.3343141078948975, "learning_rate": 7.7324e-06, "loss": 0.0408, "step": 11339 }, { "epoch": 0.6045742922642213, "grad_norm": 2.3482425212860107, "learning_rate": 7.7322e-06, "loss": 0.066, "step": 11340 }, { "epoch": 0.6046276056938743, "grad_norm": 1.6101796627044678, "learning_rate": 7.732e-06, "loss": 0.0177, "step": 11341 }, { "epoch": 0.6046809191235272, "grad_norm": 1.8538038730621338, "learning_rate": 7.731800000000001e-06, "loss": 0.0287, "step": 11342 }, { "epoch": 0.6047342325531802, "grad_norm": 3.657210111618042, "learning_rate": 7.7316e-06, "loss": 0.0529, "step": 11343 }, { "epoch": 0.6047875459828331, "grad_norm": 3.200726270675659, "learning_rate": 7.731400000000002e-06, "loss": 0.0394, "step": 11344 }, { "epoch": 0.604840859412486, "grad_norm": 2.0161659717559814, "learning_rate": 7.731200000000001e-06, "loss": 0.0363, "step": 11345 }, { "epoch": 0.604894172842139, "grad_norm": 2.8801729679107666, "learning_rate": 7.731000000000001e-06, "loss": 0.0409, "step": 11346 }, { "epoch": 0.6049474862717918, "grad_norm": 2.5835461616516113, "learning_rate": 7.7308e-06, "loss": 0.0302, "step": 11347 }, { "epoch": 0.6050007997014448, "grad_norm": 4.155864238739014, "learning_rate": 7.7306e-06, "loss": 0.054, "step": 11348 }, { "epoch": 0.6050541131310977, "grad_norm": 3.8402762413024902, "learning_rate": 7.7304e-06, "loss": 0.0475, "step": 11349 }, { "epoch": 0.6051074265607507, "grad_norm": 3.3007760047912598, "learning_rate": 7.730200000000001e-06, "loss": 0.0553, "step": 11350 }, { "epoch": 0.6051607399904035, "grad_norm": 2.6229002475738525, "learning_rate": 7.73e-06, "loss": 0.0251, "step": 11351 }, { "epoch": 0.6052140534200565, "grad_norm": 2.157604694366455, "learning_rate": 7.7298e-06, "loss": 0.0243, "step": 11352 }, { "epoch": 0.6052673668497094, "grad_norm": 2.2602715492248535, "learning_rate": 7.7296e-06, "loss": 0.0374, "step": 11353 }, { "epoch": 0.6053206802793624, "grad_norm": 2.143702983856201, "learning_rate": 7.729400000000001e-06, "loss": 0.0263, "step": 11354 }, { "epoch": 0.6053739937090153, "grad_norm": 3.269221544265747, "learning_rate": 7.7292e-06, "loss": 0.0253, "step": 11355 }, { "epoch": 0.6054273071386682, "grad_norm": 1.6876990795135498, "learning_rate": 7.729000000000002e-06, "loss": 0.0258, "step": 11356 }, { "epoch": 0.6054806205683212, "grad_norm": 3.7573001384735107, "learning_rate": 7.728800000000001e-06, "loss": 0.0457, "step": 11357 }, { "epoch": 0.6055339339979741, "grad_norm": 2.4270918369293213, "learning_rate": 7.728600000000001e-06, "loss": 0.0382, "step": 11358 }, { "epoch": 0.605587247427627, "grad_norm": 3.3764944076538086, "learning_rate": 7.7284e-06, "loss": 0.0433, "step": 11359 }, { "epoch": 0.6056405608572799, "grad_norm": 1.43256413936615, "learning_rate": 7.7282e-06, "loss": 0.0106, "step": 11360 }, { "epoch": 0.6056938742869329, "grad_norm": 1.6070185899734497, "learning_rate": 7.728000000000001e-06, "loss": 0.028, "step": 11361 }, { "epoch": 0.6057471877165858, "grad_norm": 2.0918123722076416, "learning_rate": 7.7278e-06, "loss": 0.0372, "step": 11362 }, { "epoch": 0.6058005011462387, "grad_norm": 1.1456087827682495, "learning_rate": 7.7276e-06, "loss": 0.0163, "step": 11363 }, { "epoch": 0.6058538145758917, "grad_norm": 3.200204849243164, "learning_rate": 7.7274e-06, "loss": 0.0436, "step": 11364 }, { "epoch": 0.6059071280055446, "grad_norm": 2.2547028064727783, "learning_rate": 7.7272e-06, "loss": 0.0235, "step": 11365 }, { "epoch": 0.6059604414351976, "grad_norm": 1.522652268409729, "learning_rate": 7.727e-06, "loss": 0.0318, "step": 11366 }, { "epoch": 0.6060137548648504, "grad_norm": 6.755294322967529, "learning_rate": 7.7268e-06, "loss": 0.0672, "step": 11367 }, { "epoch": 0.6060670682945034, "grad_norm": 1.3796732425689697, "learning_rate": 7.726600000000002e-06, "loss": 0.0204, "step": 11368 }, { "epoch": 0.6061203817241563, "grad_norm": 1.7549649477005005, "learning_rate": 7.726400000000001e-06, "loss": 0.0329, "step": 11369 }, { "epoch": 0.6061736951538093, "grad_norm": 1.7299699783325195, "learning_rate": 7.7262e-06, "loss": 0.0172, "step": 11370 }, { "epoch": 0.6062270085834621, "grad_norm": 1.7582870721817017, "learning_rate": 7.726e-06, "loss": 0.0311, "step": 11371 }, { "epoch": 0.6062803220131151, "grad_norm": 0.8377981781959534, "learning_rate": 7.725800000000001e-06, "loss": 0.0084, "step": 11372 }, { "epoch": 0.6063336354427681, "grad_norm": 2.541107177734375, "learning_rate": 7.725600000000001e-06, "loss": 0.0286, "step": 11373 }, { "epoch": 0.606386948872421, "grad_norm": 2.924856424331665, "learning_rate": 7.7254e-06, "loss": 0.0318, "step": 11374 }, { "epoch": 0.6064402623020739, "grad_norm": 2.4898176193237305, "learning_rate": 7.7252e-06, "loss": 0.0314, "step": 11375 }, { "epoch": 0.6064935757317268, "grad_norm": 1.876638412475586, "learning_rate": 7.725e-06, "loss": 0.0362, "step": 11376 }, { "epoch": 0.6065468891613798, "grad_norm": 2.0672504901885986, "learning_rate": 7.724800000000001e-06, "loss": 0.051, "step": 11377 }, { "epoch": 0.6066002025910326, "grad_norm": 1.5022255182266235, "learning_rate": 7.7246e-06, "loss": 0.0206, "step": 11378 }, { "epoch": 0.6066535160206856, "grad_norm": 1.6278880834579468, "learning_rate": 7.7244e-06, "loss": 0.0231, "step": 11379 }, { "epoch": 0.6067068294503385, "grad_norm": 2.6400222778320312, "learning_rate": 7.724200000000001e-06, "loss": 0.0526, "step": 11380 }, { "epoch": 0.6067601428799915, "grad_norm": 2.5253241062164307, "learning_rate": 7.724000000000001e-06, "loss": 0.0335, "step": 11381 }, { "epoch": 0.6068134563096445, "grad_norm": 2.1330952644348145, "learning_rate": 7.7238e-06, "loss": 0.0518, "step": 11382 }, { "epoch": 0.6068667697392973, "grad_norm": 3.415179967880249, "learning_rate": 7.723600000000002e-06, "loss": 0.0395, "step": 11383 }, { "epoch": 0.6069200831689503, "grad_norm": 2.130535364151001, "learning_rate": 7.723400000000001e-06, "loss": 0.0302, "step": 11384 }, { "epoch": 0.6069733965986032, "grad_norm": 1.7666542530059814, "learning_rate": 7.723200000000001e-06, "loss": 0.0307, "step": 11385 }, { "epoch": 0.6070267100282561, "grad_norm": 4.530416011810303, "learning_rate": 7.723e-06, "loss": 0.042, "step": 11386 }, { "epoch": 0.607080023457909, "grad_norm": 3.339101791381836, "learning_rate": 7.7228e-06, "loss": 0.0279, "step": 11387 }, { "epoch": 0.607133336887562, "grad_norm": 1.4195525646209717, "learning_rate": 7.7226e-06, "loss": 0.0147, "step": 11388 }, { "epoch": 0.6071866503172149, "grad_norm": 3.056379795074463, "learning_rate": 7.7224e-06, "loss": 0.0554, "step": 11389 }, { "epoch": 0.6072399637468678, "grad_norm": 2.5815842151641846, "learning_rate": 7.7222e-06, "loss": 0.0223, "step": 11390 }, { "epoch": 0.6072932771765208, "grad_norm": 2.4190256595611572, "learning_rate": 7.722e-06, "loss": 0.0409, "step": 11391 }, { "epoch": 0.6073465906061737, "grad_norm": 2.6388401985168457, "learning_rate": 7.721800000000001e-06, "loss": 0.0464, "step": 11392 }, { "epoch": 0.6073999040358267, "grad_norm": 1.099515438079834, "learning_rate": 7.7216e-06, "loss": 0.0194, "step": 11393 }, { "epoch": 0.6074532174654795, "grad_norm": 1.8500279188156128, "learning_rate": 7.7214e-06, "loss": 0.0324, "step": 11394 }, { "epoch": 0.6075065308951325, "grad_norm": 2.1617085933685303, "learning_rate": 7.721200000000002e-06, "loss": 0.0288, "step": 11395 }, { "epoch": 0.6075598443247854, "grad_norm": 3.4105544090270996, "learning_rate": 7.721000000000001e-06, "loss": 0.035, "step": 11396 }, { "epoch": 0.6076131577544384, "grad_norm": 2.2338454723358154, "learning_rate": 7.7208e-06, "loss": 0.0296, "step": 11397 }, { "epoch": 0.6076664711840912, "grad_norm": 2.1751253604888916, "learning_rate": 7.7206e-06, "loss": 0.0377, "step": 11398 }, { "epoch": 0.6077197846137442, "grad_norm": 2.891958713531494, "learning_rate": 7.7204e-06, "loss": 0.0567, "step": 11399 }, { "epoch": 0.6077730980433972, "grad_norm": 2.779451847076416, "learning_rate": 7.720200000000001e-06, "loss": 0.0296, "step": 11400 }, { "epoch": 0.60782641147305, "grad_norm": 1.817243218421936, "learning_rate": 7.72e-06, "loss": 0.0544, "step": 11401 }, { "epoch": 0.607879724902703, "grad_norm": 3.158757209777832, "learning_rate": 7.7198e-06, "loss": 0.0275, "step": 11402 }, { "epoch": 0.6079330383323559, "grad_norm": 1.1569291353225708, "learning_rate": 7.7196e-06, "loss": 0.0295, "step": 11403 }, { "epoch": 0.6079863517620089, "grad_norm": 1.6768189668655396, "learning_rate": 7.719400000000001e-06, "loss": 0.0188, "step": 11404 }, { "epoch": 0.6080396651916617, "grad_norm": 1.0435690879821777, "learning_rate": 7.7192e-06, "loss": 0.0154, "step": 11405 }, { "epoch": 0.6080929786213147, "grad_norm": 2.1105399131774902, "learning_rate": 7.719e-06, "loss": 0.0291, "step": 11406 }, { "epoch": 0.6081462920509676, "grad_norm": 1.8124336004257202, "learning_rate": 7.718800000000001e-06, "loss": 0.0452, "step": 11407 }, { "epoch": 0.6081996054806206, "grad_norm": 2.5795061588287354, "learning_rate": 7.718600000000001e-06, "loss": 0.0242, "step": 11408 }, { "epoch": 0.6082529189102734, "grad_norm": 1.3774895668029785, "learning_rate": 7.7184e-06, "loss": 0.021, "step": 11409 }, { "epoch": 0.6083062323399264, "grad_norm": 3.710946798324585, "learning_rate": 7.7182e-06, "loss": 0.0468, "step": 11410 }, { "epoch": 0.6083595457695794, "grad_norm": 2.435049533843994, "learning_rate": 7.718000000000001e-06, "loss": 0.0339, "step": 11411 }, { "epoch": 0.6084128591992323, "grad_norm": 1.6399534940719604, "learning_rate": 7.717800000000001e-06, "loss": 0.011, "step": 11412 }, { "epoch": 0.6084661726288852, "grad_norm": 1.0425454378128052, "learning_rate": 7.7176e-06, "loss": 0.0169, "step": 11413 }, { "epoch": 0.6085194860585381, "grad_norm": 2.2603719234466553, "learning_rate": 7.7174e-06, "loss": 0.0506, "step": 11414 }, { "epoch": 0.6085727994881911, "grad_norm": 1.7254232168197632, "learning_rate": 7.7172e-06, "loss": 0.0443, "step": 11415 }, { "epoch": 0.608626112917844, "grad_norm": 2.403369188308716, "learning_rate": 7.717e-06, "loss": 0.0199, "step": 11416 }, { "epoch": 0.6086794263474969, "grad_norm": 3.509333610534668, "learning_rate": 7.7168e-06, "loss": 0.0672, "step": 11417 }, { "epoch": 0.6087327397771498, "grad_norm": 1.5497981309890747, "learning_rate": 7.7166e-06, "loss": 0.0165, "step": 11418 }, { "epoch": 0.6087860532068028, "grad_norm": 2.6344268321990967, "learning_rate": 7.716400000000001e-06, "loss": 0.0373, "step": 11419 }, { "epoch": 0.6088393666364558, "grad_norm": 3.415668487548828, "learning_rate": 7.7162e-06, "loss": 0.0372, "step": 11420 }, { "epoch": 0.6088926800661086, "grad_norm": 2.2891266345977783, "learning_rate": 7.716e-06, "loss": 0.0271, "step": 11421 }, { "epoch": 0.6089459934957616, "grad_norm": 0.9132469892501831, "learning_rate": 7.715800000000002e-06, "loss": 0.0102, "step": 11422 }, { "epoch": 0.6089993069254145, "grad_norm": 1.9233453273773193, "learning_rate": 7.715600000000001e-06, "loss": 0.0341, "step": 11423 }, { "epoch": 0.6090526203550675, "grad_norm": 1.8956503868103027, "learning_rate": 7.7154e-06, "loss": 0.0364, "step": 11424 }, { "epoch": 0.6091059337847203, "grad_norm": 1.5542638301849365, "learning_rate": 7.7152e-06, "loss": 0.0115, "step": 11425 }, { "epoch": 0.6091592472143733, "grad_norm": 4.452985763549805, "learning_rate": 7.715e-06, "loss": 0.0549, "step": 11426 }, { "epoch": 0.6092125606440262, "grad_norm": 0.5117191672325134, "learning_rate": 7.714800000000001e-06, "loss": 0.0045, "step": 11427 }, { "epoch": 0.6092658740736792, "grad_norm": 1.4744445085525513, "learning_rate": 7.7146e-06, "loss": 0.0147, "step": 11428 }, { "epoch": 0.6093191875033321, "grad_norm": 1.711258888244629, "learning_rate": 7.7144e-06, "loss": 0.0149, "step": 11429 }, { "epoch": 0.609372500932985, "grad_norm": 2.9795689582824707, "learning_rate": 7.7142e-06, "loss": 0.0296, "step": 11430 }, { "epoch": 0.609425814362638, "grad_norm": 3.43380069732666, "learning_rate": 7.714000000000001e-06, "loss": 0.0392, "step": 11431 }, { "epoch": 0.6094791277922909, "grad_norm": 4.0935139656066895, "learning_rate": 7.7138e-06, "loss": 0.0738, "step": 11432 }, { "epoch": 0.6095324412219438, "grad_norm": 2.496476173400879, "learning_rate": 7.713600000000002e-06, "loss": 0.0216, "step": 11433 }, { "epoch": 0.6095857546515967, "grad_norm": 2.771369695663452, "learning_rate": 7.713400000000001e-06, "loss": 0.0396, "step": 11434 }, { "epoch": 0.6096390680812497, "grad_norm": 1.8735402822494507, "learning_rate": 7.713200000000001e-06, "loss": 0.0522, "step": 11435 }, { "epoch": 0.6096923815109025, "grad_norm": 4.071114540100098, "learning_rate": 7.713e-06, "loss": 0.0243, "step": 11436 }, { "epoch": 0.6097456949405555, "grad_norm": 4.519405364990234, "learning_rate": 7.7128e-06, "loss": 0.0845, "step": 11437 }, { "epoch": 0.6097990083702085, "grad_norm": 5.116545677185059, "learning_rate": 7.7126e-06, "loss": 0.0596, "step": 11438 }, { "epoch": 0.6098523217998614, "grad_norm": 1.5600370168685913, "learning_rate": 7.712400000000001e-06, "loss": 0.035, "step": 11439 }, { "epoch": 0.6099056352295144, "grad_norm": 2.4343435764312744, "learning_rate": 7.7122e-06, "loss": 0.0298, "step": 11440 }, { "epoch": 0.6099589486591672, "grad_norm": 1.291999101638794, "learning_rate": 7.712e-06, "loss": 0.0139, "step": 11441 }, { "epoch": 0.6100122620888202, "grad_norm": 3.7581722736358643, "learning_rate": 7.7118e-06, "loss": 0.0602, "step": 11442 }, { "epoch": 0.6100655755184731, "grad_norm": 1.956186294555664, "learning_rate": 7.7116e-06, "loss": 0.0283, "step": 11443 }, { "epoch": 0.610118888948126, "grad_norm": 3.4712882041931152, "learning_rate": 7.7114e-06, "loss": 0.0698, "step": 11444 }, { "epoch": 0.6101722023777789, "grad_norm": 3.6650609970092773, "learning_rate": 7.711200000000002e-06, "loss": 0.0465, "step": 11445 }, { "epoch": 0.6102255158074319, "grad_norm": 2.2389872074127197, "learning_rate": 7.711000000000001e-06, "loss": 0.0447, "step": 11446 }, { "epoch": 0.6102788292370849, "grad_norm": 1.8309736251831055, "learning_rate": 7.7108e-06, "loss": 0.0367, "step": 11447 }, { "epoch": 0.6103321426667377, "grad_norm": 3.979083776473999, "learning_rate": 7.7106e-06, "loss": 0.0534, "step": 11448 }, { "epoch": 0.6103854560963907, "grad_norm": 2.0965027809143066, "learning_rate": 7.7104e-06, "loss": 0.0493, "step": 11449 }, { "epoch": 0.6104387695260436, "grad_norm": 1.503283143043518, "learning_rate": 7.710200000000001e-06, "loss": 0.0189, "step": 11450 }, { "epoch": 0.6104920829556966, "grad_norm": 1.836956262588501, "learning_rate": 7.71e-06, "loss": 0.033, "step": 11451 }, { "epoch": 0.6105453963853494, "grad_norm": 1.7018585205078125, "learning_rate": 7.7098e-06, "loss": 0.0171, "step": 11452 }, { "epoch": 0.6105987098150024, "grad_norm": 2.1821482181549072, "learning_rate": 7.7096e-06, "loss": 0.0511, "step": 11453 }, { "epoch": 0.6106520232446553, "grad_norm": 1.4596046209335327, "learning_rate": 7.7094e-06, "loss": 0.0347, "step": 11454 }, { "epoch": 0.6107053366743083, "grad_norm": 1.709742784500122, "learning_rate": 7.7092e-06, "loss": 0.0231, "step": 11455 }, { "epoch": 0.6107586501039612, "grad_norm": 2.5148513317108154, "learning_rate": 7.709e-06, "loss": 0.0431, "step": 11456 }, { "epoch": 0.6108119635336141, "grad_norm": 3.1442151069641113, "learning_rate": 7.708800000000001e-06, "loss": 0.0477, "step": 11457 }, { "epoch": 0.6108652769632671, "grad_norm": 2.322643995285034, "learning_rate": 7.708600000000001e-06, "loss": 0.0689, "step": 11458 }, { "epoch": 0.61091859039292, "grad_norm": 1.6468418836593628, "learning_rate": 7.7084e-06, "loss": 0.0298, "step": 11459 }, { "epoch": 0.6109719038225729, "grad_norm": 3.0595946311950684, "learning_rate": 7.7082e-06, "loss": 0.0446, "step": 11460 }, { "epoch": 0.6110252172522258, "grad_norm": 1.2090426683425903, "learning_rate": 7.708000000000001e-06, "loss": 0.0139, "step": 11461 }, { "epoch": 0.6110785306818788, "grad_norm": 1.422121524810791, "learning_rate": 7.707800000000001e-06, "loss": 0.023, "step": 11462 }, { "epoch": 0.6111318441115317, "grad_norm": 1.6668370962142944, "learning_rate": 7.7076e-06, "loss": 0.0316, "step": 11463 }, { "epoch": 0.6111851575411846, "grad_norm": 1.964399814605713, "learning_rate": 7.7074e-06, "loss": 0.0284, "step": 11464 }, { "epoch": 0.6112384709708375, "grad_norm": 2.4589524269104004, "learning_rate": 7.7072e-06, "loss": 0.0316, "step": 11465 }, { "epoch": 0.6112917844004905, "grad_norm": 1.2505977153778076, "learning_rate": 7.707000000000001e-06, "loss": 0.011, "step": 11466 }, { "epoch": 0.6113450978301435, "grad_norm": 1.9066979885101318, "learning_rate": 7.7068e-06, "loss": 0.02, "step": 11467 }, { "epoch": 0.6113984112597963, "grad_norm": 1.840321660041809, "learning_rate": 7.7066e-06, "loss": 0.0366, "step": 11468 }, { "epoch": 0.6114517246894493, "grad_norm": 2.0304837226867676, "learning_rate": 7.706400000000001e-06, "loss": 0.0177, "step": 11469 }, { "epoch": 0.6115050381191022, "grad_norm": 3.0553739070892334, "learning_rate": 7.706200000000001e-06, "loss": 0.0216, "step": 11470 }, { "epoch": 0.6115583515487552, "grad_norm": 1.4586617946624756, "learning_rate": 7.706e-06, "loss": 0.0205, "step": 11471 }, { "epoch": 0.611611664978408, "grad_norm": 4.001524448394775, "learning_rate": 7.705800000000002e-06, "loss": 0.0736, "step": 11472 }, { "epoch": 0.611664978408061, "grad_norm": 2.6547298431396484, "learning_rate": 7.705600000000001e-06, "loss": 0.0273, "step": 11473 }, { "epoch": 0.6117182918377139, "grad_norm": 2.3255484104156494, "learning_rate": 7.7054e-06, "loss": 0.0328, "step": 11474 }, { "epoch": 0.6117716052673668, "grad_norm": 3.2526042461395264, "learning_rate": 7.7052e-06, "loss": 0.0197, "step": 11475 }, { "epoch": 0.6118249186970198, "grad_norm": 3.0436761379241943, "learning_rate": 7.705e-06, "loss": 0.0297, "step": 11476 }, { "epoch": 0.6118782321266727, "grad_norm": 2.497846841812134, "learning_rate": 7.704800000000001e-06, "loss": 0.0346, "step": 11477 }, { "epoch": 0.6119315455563257, "grad_norm": 2.351313591003418, "learning_rate": 7.7046e-06, "loss": 0.0431, "step": 11478 }, { "epoch": 0.6119848589859785, "grad_norm": 1.4572577476501465, "learning_rate": 7.7044e-06, "loss": 0.0244, "step": 11479 }, { "epoch": 0.6120381724156315, "grad_norm": 1.9487491846084595, "learning_rate": 7.7042e-06, "loss": 0.0277, "step": 11480 }, { "epoch": 0.6120914858452844, "grad_norm": 2.2259252071380615, "learning_rate": 7.704000000000001e-06, "loss": 0.0441, "step": 11481 }, { "epoch": 0.6121447992749374, "grad_norm": 0.4143889844417572, "learning_rate": 7.7038e-06, "loss": 0.0039, "step": 11482 }, { "epoch": 0.6121981127045902, "grad_norm": 2.171497106552124, "learning_rate": 7.7036e-06, "loss": 0.0496, "step": 11483 }, { "epoch": 0.6122514261342432, "grad_norm": 1.6223866939544678, "learning_rate": 7.703400000000001e-06, "loss": 0.0214, "step": 11484 }, { "epoch": 0.6123047395638962, "grad_norm": 1.352099061012268, "learning_rate": 7.703200000000001e-06, "loss": 0.0271, "step": 11485 }, { "epoch": 0.6123580529935491, "grad_norm": 4.739259243011475, "learning_rate": 7.703e-06, "loss": 0.0399, "step": 11486 }, { "epoch": 0.612411366423202, "grad_norm": 0.9190965890884399, "learning_rate": 7.7028e-06, "loss": 0.0275, "step": 11487 }, { "epoch": 0.6124646798528549, "grad_norm": 1.2803690433502197, "learning_rate": 7.7026e-06, "loss": 0.0198, "step": 11488 }, { "epoch": 0.6125179932825079, "grad_norm": 2.100220203399658, "learning_rate": 7.702400000000001e-06, "loss": 0.0321, "step": 11489 }, { "epoch": 0.6125713067121608, "grad_norm": 2.4864516258239746, "learning_rate": 7.7022e-06, "loss": 0.0333, "step": 11490 }, { "epoch": 0.6126246201418137, "grad_norm": 2.5450797080993652, "learning_rate": 7.702e-06, "loss": 0.0572, "step": 11491 }, { "epoch": 0.6126779335714666, "grad_norm": 2.4238240718841553, "learning_rate": 7.7018e-06, "loss": 0.0157, "step": 11492 }, { "epoch": 0.6127312470011196, "grad_norm": 1.5199936628341675, "learning_rate": 7.701600000000001e-06, "loss": 0.0219, "step": 11493 }, { "epoch": 0.6127845604307726, "grad_norm": 1.6763256788253784, "learning_rate": 7.7014e-06, "loss": 0.0184, "step": 11494 }, { "epoch": 0.6128378738604254, "grad_norm": 1.5850976705551147, "learning_rate": 7.7012e-06, "loss": 0.0414, "step": 11495 }, { "epoch": 0.6128911872900784, "grad_norm": 3.8159735202789307, "learning_rate": 7.701000000000001e-06, "loss": 0.1006, "step": 11496 }, { "epoch": 0.6129445007197313, "grad_norm": 1.845399260520935, "learning_rate": 7.700800000000001e-06, "loss": 0.018, "step": 11497 }, { "epoch": 0.6129978141493843, "grad_norm": 2.1436448097229004, "learning_rate": 7.7006e-06, "loss": 0.0259, "step": 11498 }, { "epoch": 0.6130511275790371, "grad_norm": 3.6365575790405273, "learning_rate": 7.7004e-06, "loss": 0.0713, "step": 11499 }, { "epoch": 0.6131044410086901, "grad_norm": 1.982319951057434, "learning_rate": 7.700200000000001e-06, "loss": 0.0437, "step": 11500 }, { "epoch": 0.613157754438343, "grad_norm": 1.6398415565490723, "learning_rate": 7.7e-06, "loss": 0.022, "step": 11501 }, { "epoch": 0.613211067867996, "grad_norm": 0.9615722894668579, "learning_rate": 7.6998e-06, "loss": 0.0154, "step": 11502 }, { "epoch": 0.6132643812976489, "grad_norm": 1.7637732028961182, "learning_rate": 7.6996e-06, "loss": 0.0401, "step": 11503 }, { "epoch": 0.6133176947273018, "grad_norm": 2.086470127105713, "learning_rate": 7.6994e-06, "loss": 0.0529, "step": 11504 }, { "epoch": 0.6133710081569548, "grad_norm": 1.9667164087295532, "learning_rate": 7.6992e-06, "loss": 0.0407, "step": 11505 }, { "epoch": 0.6134243215866076, "grad_norm": 1.8378287553787231, "learning_rate": 7.699e-06, "loss": 0.0499, "step": 11506 }, { "epoch": 0.6134776350162606, "grad_norm": 2.429457426071167, "learning_rate": 7.6988e-06, "loss": 0.0437, "step": 11507 }, { "epoch": 0.6135309484459135, "grad_norm": 1.8844236135482788, "learning_rate": 7.698600000000001e-06, "loss": 0.0185, "step": 11508 }, { "epoch": 0.6135842618755665, "grad_norm": 1.5866940021514893, "learning_rate": 7.6984e-06, "loss": 0.0453, "step": 11509 }, { "epoch": 0.6136375753052193, "grad_norm": 1.98674476146698, "learning_rate": 7.6982e-06, "loss": 0.0253, "step": 11510 }, { "epoch": 0.6136908887348723, "grad_norm": 1.2179476022720337, "learning_rate": 7.698000000000002e-06, "loss": 0.0194, "step": 11511 }, { "epoch": 0.6137442021645253, "grad_norm": 2.7218122482299805, "learning_rate": 7.697800000000001e-06, "loss": 0.0446, "step": 11512 }, { "epoch": 0.6137975155941782, "grad_norm": 3.3401169776916504, "learning_rate": 7.6976e-06, "loss": 0.028, "step": 11513 }, { "epoch": 0.6138508290238311, "grad_norm": 1.920256495475769, "learning_rate": 7.6974e-06, "loss": 0.0411, "step": 11514 }, { "epoch": 0.613904142453484, "grad_norm": 1.7191344499588013, "learning_rate": 7.6972e-06, "loss": 0.0489, "step": 11515 }, { "epoch": 0.613957455883137, "grad_norm": 1.6285345554351807, "learning_rate": 7.697000000000001e-06, "loss": 0.0155, "step": 11516 }, { "epoch": 0.6140107693127899, "grad_norm": 1.18731689453125, "learning_rate": 7.6968e-06, "loss": 0.0088, "step": 11517 }, { "epoch": 0.6140640827424428, "grad_norm": 1.365396499633789, "learning_rate": 7.6966e-06, "loss": 0.0162, "step": 11518 }, { "epoch": 0.6141173961720957, "grad_norm": 3.1501271724700928, "learning_rate": 7.6964e-06, "loss": 0.0338, "step": 11519 }, { "epoch": 0.6141707096017487, "grad_norm": 1.2298239469528198, "learning_rate": 7.696200000000001e-06, "loss": 0.0177, "step": 11520 }, { "epoch": 0.6142240230314016, "grad_norm": 2.8544089794158936, "learning_rate": 7.696e-06, "loss": 0.0381, "step": 11521 }, { "epoch": 0.6142773364610545, "grad_norm": 2.9433631896972656, "learning_rate": 7.695800000000002e-06, "loss": 0.0293, "step": 11522 }, { "epoch": 0.6143306498907075, "grad_norm": 1.2445749044418335, "learning_rate": 7.695600000000001e-06, "loss": 0.0251, "step": 11523 }, { "epoch": 0.6143839633203604, "grad_norm": 3.2770683765411377, "learning_rate": 7.695400000000001e-06, "loss": 0.0762, "step": 11524 }, { "epoch": 0.6144372767500134, "grad_norm": 2.869467258453369, "learning_rate": 7.6952e-06, "loss": 0.0229, "step": 11525 }, { "epoch": 0.6144905901796662, "grad_norm": 2.7689809799194336, "learning_rate": 7.695e-06, "loss": 0.0454, "step": 11526 }, { "epoch": 0.6145439036093192, "grad_norm": 1.3316937685012817, "learning_rate": 7.694800000000001e-06, "loss": 0.0185, "step": 11527 }, { "epoch": 0.6145972170389721, "grad_norm": 4.218339920043945, "learning_rate": 7.6946e-06, "loss": 0.0639, "step": 11528 }, { "epoch": 0.6146505304686251, "grad_norm": 1.6956379413604736, "learning_rate": 7.6944e-06, "loss": 0.0393, "step": 11529 }, { "epoch": 0.6147038438982779, "grad_norm": 1.140798568725586, "learning_rate": 7.6942e-06, "loss": 0.0161, "step": 11530 }, { "epoch": 0.6147571573279309, "grad_norm": 1.8588457107543945, "learning_rate": 7.694e-06, "loss": 0.0475, "step": 11531 }, { "epoch": 0.6148104707575839, "grad_norm": 3.3766655921936035, "learning_rate": 7.6938e-06, "loss": 0.0292, "step": 11532 }, { "epoch": 0.6148637841872368, "grad_norm": 2.8094277381896973, "learning_rate": 7.6936e-06, "loss": 0.0524, "step": 11533 }, { "epoch": 0.6149170976168897, "grad_norm": 3.8647801876068115, "learning_rate": 7.693400000000002e-06, "loss": 0.0509, "step": 11534 }, { "epoch": 0.6149704110465426, "grad_norm": 3.1462857723236084, "learning_rate": 7.693200000000001e-06, "loss": 0.034, "step": 11535 }, { "epoch": 0.6150237244761956, "grad_norm": 2.5747594833374023, "learning_rate": 7.693e-06, "loss": 0.0164, "step": 11536 }, { "epoch": 0.6150770379058484, "grad_norm": 2.2598724365234375, "learning_rate": 7.6928e-06, "loss": 0.023, "step": 11537 }, { "epoch": 0.6151303513355014, "grad_norm": 2.0822067260742188, "learning_rate": 7.692600000000002e-06, "loss": 0.0382, "step": 11538 }, { "epoch": 0.6151836647651543, "grad_norm": 3.9176909923553467, "learning_rate": 7.692400000000001e-06, "loss": 0.026, "step": 11539 }, { "epoch": 0.6152369781948073, "grad_norm": 2.578935384750366, "learning_rate": 7.6922e-06, "loss": 0.0326, "step": 11540 }, { "epoch": 0.6152902916244603, "grad_norm": 0.8775137066841125, "learning_rate": 7.692e-06, "loss": 0.0189, "step": 11541 }, { "epoch": 0.6153436050541131, "grad_norm": 2.643465042114258, "learning_rate": 7.6918e-06, "loss": 0.0412, "step": 11542 }, { "epoch": 0.6153969184837661, "grad_norm": 1.8476264476776123, "learning_rate": 7.6916e-06, "loss": 0.0176, "step": 11543 }, { "epoch": 0.615450231913419, "grad_norm": 2.8910298347473145, "learning_rate": 7.6914e-06, "loss": 0.0411, "step": 11544 }, { "epoch": 0.615503545343072, "grad_norm": 2.4697864055633545, "learning_rate": 7.6912e-06, "loss": 0.0449, "step": 11545 }, { "epoch": 0.6155568587727248, "grad_norm": 1.2266658544540405, "learning_rate": 7.691000000000001e-06, "loss": 0.0155, "step": 11546 }, { "epoch": 0.6156101722023778, "grad_norm": 1.632561445236206, "learning_rate": 7.690800000000001e-06, "loss": 0.0397, "step": 11547 }, { "epoch": 0.6156634856320307, "grad_norm": 3.11480450630188, "learning_rate": 7.6906e-06, "loss": 0.0351, "step": 11548 }, { "epoch": 0.6157167990616836, "grad_norm": 2.3170385360717773, "learning_rate": 7.6904e-06, "loss": 0.0307, "step": 11549 }, { "epoch": 0.6157701124913366, "grad_norm": 1.4638932943344116, "learning_rate": 7.690200000000001e-06, "loss": 0.0148, "step": 11550 }, { "epoch": 0.6158234259209895, "grad_norm": 1.4054946899414062, "learning_rate": 7.690000000000001e-06, "loss": 0.0278, "step": 11551 }, { "epoch": 0.6158767393506425, "grad_norm": 2.547739028930664, "learning_rate": 7.6898e-06, "loss": 0.0336, "step": 11552 }, { "epoch": 0.6159300527802953, "grad_norm": 2.365917444229126, "learning_rate": 7.6896e-06, "loss": 0.035, "step": 11553 }, { "epoch": 0.6159833662099483, "grad_norm": 3.138819932937622, "learning_rate": 7.6894e-06, "loss": 0.0861, "step": 11554 }, { "epoch": 0.6160366796396012, "grad_norm": 2.347658157348633, "learning_rate": 7.689200000000001e-06, "loss": 0.0219, "step": 11555 }, { "epoch": 0.6160899930692542, "grad_norm": 0.9013023972511292, "learning_rate": 7.689e-06, "loss": 0.0099, "step": 11556 }, { "epoch": 0.616143306498907, "grad_norm": 1.6869372129440308, "learning_rate": 7.6888e-06, "loss": 0.0121, "step": 11557 }, { "epoch": 0.61619661992856, "grad_norm": 1.9302642345428467, "learning_rate": 7.688600000000001e-06, "loss": 0.0532, "step": 11558 }, { "epoch": 0.616249933358213, "grad_norm": 1.7383456230163574, "learning_rate": 7.6884e-06, "loss": 0.0236, "step": 11559 }, { "epoch": 0.6163032467878659, "grad_norm": 1.0295637845993042, "learning_rate": 7.6882e-06, "loss": 0.0146, "step": 11560 }, { "epoch": 0.6163565602175188, "grad_norm": 2.3029651641845703, "learning_rate": 7.688000000000002e-06, "loss": 0.0229, "step": 11561 }, { "epoch": 0.6164098736471717, "grad_norm": 1.9878989458084106, "learning_rate": 7.687800000000001e-06, "loss": 0.0221, "step": 11562 }, { "epoch": 0.6164631870768247, "grad_norm": 3.5301976203918457, "learning_rate": 7.6876e-06, "loss": 0.032, "step": 11563 }, { "epoch": 0.6165165005064775, "grad_norm": 2.473447799682617, "learning_rate": 7.6874e-06, "loss": 0.042, "step": 11564 }, { "epoch": 0.6165698139361305, "grad_norm": 1.8652533292770386, "learning_rate": 7.6872e-06, "loss": 0.0165, "step": 11565 }, { "epoch": 0.6166231273657834, "grad_norm": 3.1629648208618164, "learning_rate": 7.687000000000001e-06, "loss": 0.0549, "step": 11566 }, { "epoch": 0.6166764407954364, "grad_norm": 3.291015148162842, "learning_rate": 7.6868e-06, "loss": 0.0561, "step": 11567 }, { "epoch": 0.6167297542250894, "grad_norm": 2.1342127323150635, "learning_rate": 7.6866e-06, "loss": 0.0193, "step": 11568 }, { "epoch": 0.6167830676547422, "grad_norm": 2.3659586906433105, "learning_rate": 7.6864e-06, "loss": 0.0192, "step": 11569 }, { "epoch": 0.6168363810843952, "grad_norm": 3.192030191421509, "learning_rate": 7.686200000000001e-06, "loss": 0.026, "step": 11570 }, { "epoch": 0.6168896945140481, "grad_norm": 1.88958740234375, "learning_rate": 7.686e-06, "loss": 0.0172, "step": 11571 }, { "epoch": 0.616943007943701, "grad_norm": 3.244839906692505, "learning_rate": 7.685800000000002e-06, "loss": 0.0398, "step": 11572 }, { "epoch": 0.6169963213733539, "grad_norm": 3.240959405899048, "learning_rate": 7.685600000000001e-06, "loss": 0.0487, "step": 11573 }, { "epoch": 0.6170496348030069, "grad_norm": 2.1795742511749268, "learning_rate": 7.685400000000001e-06, "loss": 0.0433, "step": 11574 }, { "epoch": 0.6171029482326598, "grad_norm": 2.1739442348480225, "learning_rate": 7.6852e-06, "loss": 0.0152, "step": 11575 }, { "epoch": 0.6171562616623127, "grad_norm": 1.3212529420852661, "learning_rate": 7.685e-06, "loss": 0.0141, "step": 11576 }, { "epoch": 0.6172095750919656, "grad_norm": 1.2908661365509033, "learning_rate": 7.684800000000001e-06, "loss": 0.0137, "step": 11577 }, { "epoch": 0.6172628885216186, "grad_norm": 2.1023800373077393, "learning_rate": 7.684600000000001e-06, "loss": 0.0377, "step": 11578 }, { "epoch": 0.6173162019512716, "grad_norm": 2.0307023525238037, "learning_rate": 7.6844e-06, "loss": 0.0197, "step": 11579 }, { "epoch": 0.6173695153809244, "grad_norm": 2.2101171016693115, "learning_rate": 7.6842e-06, "loss": 0.0486, "step": 11580 }, { "epoch": 0.6174228288105774, "grad_norm": 2.145235776901245, "learning_rate": 7.684e-06, "loss": 0.0175, "step": 11581 }, { "epoch": 0.6174761422402303, "grad_norm": 2.573319911956787, "learning_rate": 7.683800000000001e-06, "loss": 0.049, "step": 11582 }, { "epoch": 0.6175294556698833, "grad_norm": 2.9731528759002686, "learning_rate": 7.6836e-06, "loss": 0.0254, "step": 11583 }, { "epoch": 0.6175827690995361, "grad_norm": 3.6498422622680664, "learning_rate": 7.683400000000002e-06, "loss": 0.0344, "step": 11584 }, { "epoch": 0.6176360825291891, "grad_norm": 2.342374324798584, "learning_rate": 7.683200000000001e-06, "loss": 0.042, "step": 11585 }, { "epoch": 0.617689395958842, "grad_norm": 1.6749558448791504, "learning_rate": 7.683e-06, "loss": 0.0384, "step": 11586 }, { "epoch": 0.617742709388495, "grad_norm": 2.3674798011779785, "learning_rate": 7.6828e-06, "loss": 0.0281, "step": 11587 }, { "epoch": 0.6177960228181479, "grad_norm": 0.7999829649925232, "learning_rate": 7.682600000000002e-06, "loss": 0.0083, "step": 11588 }, { "epoch": 0.6178493362478008, "grad_norm": 2.0402109622955322, "learning_rate": 7.682400000000001e-06, "loss": 0.0224, "step": 11589 }, { "epoch": 0.6179026496774538, "grad_norm": 2.7871007919311523, "learning_rate": 7.6822e-06, "loss": 0.0348, "step": 11590 }, { "epoch": 0.6179559631071067, "grad_norm": 3.1302618980407715, "learning_rate": 7.682e-06, "loss": 0.0388, "step": 11591 }, { "epoch": 0.6180092765367596, "grad_norm": 3.8683061599731445, "learning_rate": 7.6818e-06, "loss": 0.0523, "step": 11592 }, { "epoch": 0.6180625899664125, "grad_norm": 2.6677744388580322, "learning_rate": 7.6816e-06, "loss": 0.0396, "step": 11593 }, { "epoch": 0.6181159033960655, "grad_norm": 1.6780290603637695, "learning_rate": 7.6814e-06, "loss": 0.035, "step": 11594 }, { "epoch": 0.6181692168257183, "grad_norm": 3.8689680099487305, "learning_rate": 7.6812e-06, "loss": 0.0293, "step": 11595 }, { "epoch": 0.6182225302553713, "grad_norm": 8.75241756439209, "learning_rate": 7.681000000000002e-06, "loss": 0.074, "step": 11596 }, { "epoch": 0.6182758436850243, "grad_norm": 1.553688645362854, "learning_rate": 7.680800000000001e-06, "loss": 0.0233, "step": 11597 }, { "epoch": 0.6183291571146772, "grad_norm": 2.351151943206787, "learning_rate": 7.6806e-06, "loss": 0.0333, "step": 11598 }, { "epoch": 0.6183824705443302, "grad_norm": 2.0723824501037598, "learning_rate": 7.6804e-06, "loss": 0.0319, "step": 11599 }, { "epoch": 0.618435783973983, "grad_norm": 0.2472674548625946, "learning_rate": 7.680200000000001e-06, "loss": 0.0029, "step": 11600 }, { "epoch": 0.618489097403636, "grad_norm": 1.9408652782440186, "learning_rate": 7.680000000000001e-06, "loss": 0.0682, "step": 11601 }, { "epoch": 0.6185424108332889, "grad_norm": 3.5811026096343994, "learning_rate": 7.6798e-06, "loss": 0.0556, "step": 11602 }, { "epoch": 0.6185957242629418, "grad_norm": 1.6277610063552856, "learning_rate": 7.6796e-06, "loss": 0.034, "step": 11603 }, { "epoch": 0.6186490376925947, "grad_norm": 3.29826283454895, "learning_rate": 7.6794e-06, "loss": 0.0497, "step": 11604 }, { "epoch": 0.6187023511222477, "grad_norm": 2.6813950538635254, "learning_rate": 7.679200000000001e-06, "loss": 0.0449, "step": 11605 }, { "epoch": 0.6187556645519007, "grad_norm": 2.222440004348755, "learning_rate": 7.679e-06, "loss": 0.0376, "step": 11606 }, { "epoch": 0.6188089779815535, "grad_norm": 1.6848015785217285, "learning_rate": 7.6788e-06, "loss": 0.0259, "step": 11607 }, { "epoch": 0.6188622914112065, "grad_norm": 1.148472785949707, "learning_rate": 7.678600000000001e-06, "loss": 0.0305, "step": 11608 }, { "epoch": 0.6189156048408594, "grad_norm": 2.091277599334717, "learning_rate": 7.678400000000001e-06, "loss": 0.0304, "step": 11609 }, { "epoch": 0.6189689182705124, "grad_norm": 2.182701826095581, "learning_rate": 7.6782e-06, "loss": 0.0631, "step": 11610 }, { "epoch": 0.6190222317001652, "grad_norm": 2.2798986434936523, "learning_rate": 7.678000000000002e-06, "loss": 0.0263, "step": 11611 }, { "epoch": 0.6190755451298182, "grad_norm": 3.083366632461548, "learning_rate": 7.677800000000001e-06, "loss": 0.0403, "step": 11612 }, { "epoch": 0.6191288585594711, "grad_norm": 3.059518814086914, "learning_rate": 7.6776e-06, "loss": 0.0369, "step": 11613 }, { "epoch": 0.6191821719891241, "grad_norm": 1.783123254776001, "learning_rate": 7.6774e-06, "loss": 0.0272, "step": 11614 }, { "epoch": 0.619235485418777, "grad_norm": 1.9529294967651367, "learning_rate": 7.6772e-06, "loss": 0.0293, "step": 11615 }, { "epoch": 0.6192887988484299, "grad_norm": 1.8554401397705078, "learning_rate": 7.677000000000001e-06, "loss": 0.0297, "step": 11616 }, { "epoch": 0.6193421122780829, "grad_norm": 2.4925215244293213, "learning_rate": 7.6768e-06, "loss": 0.0723, "step": 11617 }, { "epoch": 0.6193954257077358, "grad_norm": 1.668063998222351, "learning_rate": 7.6766e-06, "loss": 0.0401, "step": 11618 }, { "epoch": 0.6194487391373887, "grad_norm": 2.211056709289551, "learning_rate": 7.6764e-06, "loss": 0.028, "step": 11619 }, { "epoch": 0.6195020525670416, "grad_norm": 2.009901523590088, "learning_rate": 7.676200000000001e-06, "loss": 0.034, "step": 11620 }, { "epoch": 0.6195553659966946, "grad_norm": 1.0877667665481567, "learning_rate": 7.676e-06, "loss": 0.0299, "step": 11621 }, { "epoch": 0.6196086794263475, "grad_norm": 1.6865421533584595, "learning_rate": 7.6758e-06, "loss": 0.035, "step": 11622 }, { "epoch": 0.6196619928560004, "grad_norm": 2.488434076309204, "learning_rate": 7.675600000000002e-06, "loss": 0.0277, "step": 11623 }, { "epoch": 0.6197153062856534, "grad_norm": 1.7425708770751953, "learning_rate": 7.675400000000001e-06, "loss": 0.029, "step": 11624 }, { "epoch": 0.6197686197153063, "grad_norm": 0.7484139800071716, "learning_rate": 7.6752e-06, "loss": 0.0063, "step": 11625 }, { "epoch": 0.6198219331449593, "grad_norm": 1.4864522218704224, "learning_rate": 7.675e-06, "loss": 0.0213, "step": 11626 }, { "epoch": 0.6198752465746121, "grad_norm": 1.920389175415039, "learning_rate": 7.674800000000001e-06, "loss": 0.0307, "step": 11627 }, { "epoch": 0.6199285600042651, "grad_norm": 3.150254249572754, "learning_rate": 7.674600000000001e-06, "loss": 0.027, "step": 11628 }, { "epoch": 0.619981873433918, "grad_norm": 1.4653575420379639, "learning_rate": 7.6744e-06, "loss": 0.0277, "step": 11629 }, { "epoch": 0.620035186863571, "grad_norm": 2.449645519256592, "learning_rate": 7.6742e-06, "loss": 0.0386, "step": 11630 }, { "epoch": 0.6200885002932238, "grad_norm": 2.9900879859924316, "learning_rate": 7.674e-06, "loss": 0.0407, "step": 11631 }, { "epoch": 0.6201418137228768, "grad_norm": 1.6084165573120117, "learning_rate": 7.673800000000001e-06, "loss": 0.025, "step": 11632 }, { "epoch": 0.6201951271525297, "grad_norm": 3.561204195022583, "learning_rate": 7.6736e-06, "loss": 0.025, "step": 11633 }, { "epoch": 0.6202484405821826, "grad_norm": 1.869475245475769, "learning_rate": 7.6734e-06, "loss": 0.0342, "step": 11634 }, { "epoch": 0.6203017540118356, "grad_norm": 1.6290737390518188, "learning_rate": 7.673200000000001e-06, "loss": 0.0175, "step": 11635 }, { "epoch": 0.6203550674414885, "grad_norm": 1.408044457435608, "learning_rate": 7.673000000000001e-06, "loss": 0.0202, "step": 11636 }, { "epoch": 0.6204083808711415, "grad_norm": 2.1506361961364746, "learning_rate": 7.6728e-06, "loss": 0.0406, "step": 11637 }, { "epoch": 0.6204616943007943, "grad_norm": 1.0243009328842163, "learning_rate": 7.672600000000002e-06, "loss": 0.0102, "step": 11638 }, { "epoch": 0.6205150077304473, "grad_norm": 1.757915735244751, "learning_rate": 7.672400000000001e-06, "loss": 0.0453, "step": 11639 }, { "epoch": 0.6205683211601002, "grad_norm": 2.3674185276031494, "learning_rate": 7.6722e-06, "loss": 0.0232, "step": 11640 }, { "epoch": 0.6206216345897532, "grad_norm": 1.9755897521972656, "learning_rate": 7.672e-06, "loss": 0.0423, "step": 11641 }, { "epoch": 0.620674948019406, "grad_norm": 1.9680017232894897, "learning_rate": 7.6718e-06, "loss": 0.0421, "step": 11642 }, { "epoch": 0.620728261449059, "grad_norm": 1.690742015838623, "learning_rate": 7.6716e-06, "loss": 0.0433, "step": 11643 }, { "epoch": 0.620781574878712, "grad_norm": 2.3940372467041016, "learning_rate": 7.6714e-06, "loss": 0.0369, "step": 11644 }, { "epoch": 0.6208348883083649, "grad_norm": 2.5271849632263184, "learning_rate": 7.6712e-06, "loss": 0.0247, "step": 11645 }, { "epoch": 0.6208882017380178, "grad_norm": 3.359069347381592, "learning_rate": 7.671e-06, "loss": 0.0604, "step": 11646 }, { "epoch": 0.6209415151676707, "grad_norm": 3.6542978286743164, "learning_rate": 7.670800000000001e-06, "loss": 0.0963, "step": 11647 }, { "epoch": 0.6209948285973237, "grad_norm": 3.5774364471435547, "learning_rate": 7.6706e-06, "loss": 0.0288, "step": 11648 }, { "epoch": 0.6210481420269766, "grad_norm": 1.9099032878875732, "learning_rate": 7.670400000000002e-06, "loss": 0.0234, "step": 11649 }, { "epoch": 0.6211014554566295, "grad_norm": 1.2250311374664307, "learning_rate": 7.670200000000002e-06, "loss": 0.0269, "step": 11650 }, { "epoch": 0.6211547688862824, "grad_norm": 2.064460277557373, "learning_rate": 7.670000000000001e-06, "loss": 0.0209, "step": 11651 }, { "epoch": 0.6212080823159354, "grad_norm": 2.6228299140930176, "learning_rate": 7.6698e-06, "loss": 0.0477, "step": 11652 }, { "epoch": 0.6212613957455884, "grad_norm": 2.1085243225097656, "learning_rate": 7.6696e-06, "loss": 0.0151, "step": 11653 }, { "epoch": 0.6213147091752412, "grad_norm": 3.574660539627075, "learning_rate": 7.6694e-06, "loss": 0.0409, "step": 11654 }, { "epoch": 0.6213680226048942, "grad_norm": 0.7769577503204346, "learning_rate": 7.669200000000001e-06, "loss": 0.0067, "step": 11655 }, { "epoch": 0.6214213360345471, "grad_norm": 3.648796558380127, "learning_rate": 7.669e-06, "loss": 0.0343, "step": 11656 }, { "epoch": 0.6214746494642001, "grad_norm": 3.303419589996338, "learning_rate": 7.6688e-06, "loss": 0.027, "step": 11657 }, { "epoch": 0.6215279628938529, "grad_norm": 1.269991159439087, "learning_rate": 7.6686e-06, "loss": 0.0176, "step": 11658 }, { "epoch": 0.6215812763235059, "grad_norm": 2.9462902545928955, "learning_rate": 7.668400000000001e-06, "loss": 0.0237, "step": 11659 }, { "epoch": 0.6216345897531588, "grad_norm": 1.6476223468780518, "learning_rate": 7.6682e-06, "loss": 0.0195, "step": 11660 }, { "epoch": 0.6216879031828118, "grad_norm": 1.133911371231079, "learning_rate": 7.668000000000002e-06, "loss": 0.0271, "step": 11661 }, { "epoch": 0.6217412166124647, "grad_norm": 1.8601194620132446, "learning_rate": 7.667800000000001e-06, "loss": 0.0465, "step": 11662 }, { "epoch": 0.6217945300421176, "grad_norm": 2.1558756828308105, "learning_rate": 7.667600000000001e-06, "loss": 0.0267, "step": 11663 }, { "epoch": 0.6218478434717706, "grad_norm": 3.638820171356201, "learning_rate": 7.6674e-06, "loss": 0.0565, "step": 11664 }, { "epoch": 0.6219011569014234, "grad_norm": 2.2365992069244385, "learning_rate": 7.6672e-06, "loss": 0.033, "step": 11665 }, { "epoch": 0.6219544703310764, "grad_norm": 0.8666501641273499, "learning_rate": 7.667000000000001e-06, "loss": 0.0077, "step": 11666 }, { "epoch": 0.6220077837607293, "grad_norm": 4.013430118560791, "learning_rate": 7.666800000000001e-06, "loss": 0.0476, "step": 11667 }, { "epoch": 0.6220610971903823, "grad_norm": 4.990156173706055, "learning_rate": 7.6666e-06, "loss": 0.0236, "step": 11668 }, { "epoch": 0.6221144106200351, "grad_norm": 3.230102300643921, "learning_rate": 7.6664e-06, "loss": 0.0278, "step": 11669 }, { "epoch": 0.6221677240496881, "grad_norm": 1.965803861618042, "learning_rate": 7.6662e-06, "loss": 0.0326, "step": 11670 }, { "epoch": 0.6222210374793411, "grad_norm": 1.8628071546554565, "learning_rate": 7.666e-06, "loss": 0.0427, "step": 11671 }, { "epoch": 0.622274350908994, "grad_norm": 4.079311370849609, "learning_rate": 7.6658e-06, "loss": 0.0672, "step": 11672 }, { "epoch": 0.622327664338647, "grad_norm": 2.2587289810180664, "learning_rate": 7.665600000000002e-06, "loss": 0.0355, "step": 11673 }, { "epoch": 0.6223809777682998, "grad_norm": 1.5086768865585327, "learning_rate": 7.665400000000001e-06, "loss": 0.0335, "step": 11674 }, { "epoch": 0.6224342911979528, "grad_norm": 3.1305508613586426, "learning_rate": 7.6652e-06, "loss": 0.0748, "step": 11675 }, { "epoch": 0.6224876046276057, "grad_norm": 1.2553611993789673, "learning_rate": 7.665e-06, "loss": 0.0152, "step": 11676 }, { "epoch": 0.6225409180572586, "grad_norm": 1.6400463581085205, "learning_rate": 7.664800000000002e-06, "loss": 0.0158, "step": 11677 }, { "epoch": 0.6225942314869115, "grad_norm": 1.8590960502624512, "learning_rate": 7.664600000000001e-06, "loss": 0.031, "step": 11678 }, { "epoch": 0.6226475449165645, "grad_norm": 3.418191909790039, "learning_rate": 7.6644e-06, "loss": 0.058, "step": 11679 }, { "epoch": 0.6227008583462175, "grad_norm": 1.0927189588546753, "learning_rate": 7.6642e-06, "loss": 0.0088, "step": 11680 }, { "epoch": 0.6227541717758703, "grad_norm": 4.2106499671936035, "learning_rate": 7.664e-06, "loss": 0.0285, "step": 11681 }, { "epoch": 0.6228074852055233, "grad_norm": 3.9175984859466553, "learning_rate": 7.6638e-06, "loss": 0.0554, "step": 11682 }, { "epoch": 0.6228607986351762, "grad_norm": 2.889313220977783, "learning_rate": 7.6636e-06, "loss": 0.0314, "step": 11683 }, { "epoch": 0.6229141120648292, "grad_norm": 0.9535328149795532, "learning_rate": 7.6634e-06, "loss": 0.0099, "step": 11684 }, { "epoch": 0.622967425494482, "grad_norm": 1.410181999206543, "learning_rate": 7.663200000000001e-06, "loss": 0.0147, "step": 11685 }, { "epoch": 0.623020738924135, "grad_norm": 3.040832042694092, "learning_rate": 7.663000000000001e-06, "loss": 0.0317, "step": 11686 }, { "epoch": 0.6230740523537879, "grad_norm": 2.1739749908447266, "learning_rate": 7.6628e-06, "loss": 0.0327, "step": 11687 }, { "epoch": 0.6231273657834409, "grad_norm": 1.885414719581604, "learning_rate": 7.662600000000002e-06, "loss": 0.0237, "step": 11688 }, { "epoch": 0.6231806792130937, "grad_norm": 1.8632768392562866, "learning_rate": 7.662400000000001e-06, "loss": 0.0332, "step": 11689 }, { "epoch": 0.6232339926427467, "grad_norm": 1.1975027322769165, "learning_rate": 7.662200000000001e-06, "loss": 0.0268, "step": 11690 }, { "epoch": 0.6232873060723997, "grad_norm": 4.70945930480957, "learning_rate": 7.662e-06, "loss": 0.0589, "step": 11691 }, { "epoch": 0.6233406195020526, "grad_norm": 3.7443575859069824, "learning_rate": 7.6618e-06, "loss": 0.032, "step": 11692 }, { "epoch": 0.6233939329317055, "grad_norm": 3.587491989135742, "learning_rate": 7.6616e-06, "loss": 0.0435, "step": 11693 }, { "epoch": 0.6234472463613584, "grad_norm": 1.9355313777923584, "learning_rate": 7.661400000000001e-06, "loss": 0.0197, "step": 11694 }, { "epoch": 0.6235005597910114, "grad_norm": 2.534745216369629, "learning_rate": 7.6612e-06, "loss": 0.0498, "step": 11695 }, { "epoch": 0.6235538732206642, "grad_norm": 1.5668261051177979, "learning_rate": 7.661e-06, "loss": 0.0168, "step": 11696 }, { "epoch": 0.6236071866503172, "grad_norm": 2.3679821491241455, "learning_rate": 7.660800000000001e-06, "loss": 0.0365, "step": 11697 }, { "epoch": 0.6236605000799701, "grad_norm": 1.786478042602539, "learning_rate": 7.6606e-06, "loss": 0.032, "step": 11698 }, { "epoch": 0.6237138135096231, "grad_norm": 1.9633327722549438, "learning_rate": 7.6604e-06, "loss": 0.01, "step": 11699 }, { "epoch": 0.623767126939276, "grad_norm": 4.072812080383301, "learning_rate": 7.660200000000002e-06, "loss": 0.0342, "step": 11700 }, { "epoch": 0.6238204403689289, "grad_norm": 2.4189748764038086, "learning_rate": 7.660000000000001e-06, "loss": 0.0102, "step": 11701 }, { "epoch": 0.6238737537985819, "grad_norm": 1.7514524459838867, "learning_rate": 7.6598e-06, "loss": 0.0398, "step": 11702 }, { "epoch": 0.6239270672282348, "grad_norm": 2.65234112739563, "learning_rate": 7.6596e-06, "loss": 0.0708, "step": 11703 }, { "epoch": 0.6239803806578877, "grad_norm": 4.142723560333252, "learning_rate": 7.6594e-06, "loss": 0.0936, "step": 11704 }, { "epoch": 0.6240336940875406, "grad_norm": 2.699540138244629, "learning_rate": 7.659200000000001e-06, "loss": 0.0637, "step": 11705 }, { "epoch": 0.6240870075171936, "grad_norm": 2.7154390811920166, "learning_rate": 7.659e-06, "loss": 0.0399, "step": 11706 }, { "epoch": 0.6241403209468465, "grad_norm": 2.2192516326904297, "learning_rate": 7.6588e-06, "loss": 0.0362, "step": 11707 }, { "epoch": 0.6241936343764994, "grad_norm": 2.156144380569458, "learning_rate": 7.6586e-06, "loss": 0.0401, "step": 11708 }, { "epoch": 0.6242469478061524, "grad_norm": 3.003399610519409, "learning_rate": 7.658400000000001e-06, "loss": 0.0518, "step": 11709 }, { "epoch": 0.6243002612358053, "grad_norm": 1.4006625413894653, "learning_rate": 7.6582e-06, "loss": 0.0233, "step": 11710 }, { "epoch": 0.6243535746654583, "grad_norm": 2.7731435298919678, "learning_rate": 7.658e-06, "loss": 0.0292, "step": 11711 }, { "epoch": 0.6244068880951111, "grad_norm": 3.9732635021209717, "learning_rate": 7.657800000000001e-06, "loss": 0.0484, "step": 11712 }, { "epoch": 0.6244602015247641, "grad_norm": 2.5857934951782227, "learning_rate": 7.657600000000001e-06, "loss": 0.0364, "step": 11713 }, { "epoch": 0.624513514954417, "grad_norm": 4.337911128997803, "learning_rate": 7.6574e-06, "loss": 0.068, "step": 11714 }, { "epoch": 0.62456682838407, "grad_norm": 2.9085500240325928, "learning_rate": 7.6572e-06, "loss": 0.059, "step": 11715 }, { "epoch": 0.6246201418137228, "grad_norm": 2.8229219913482666, "learning_rate": 7.657000000000001e-06, "loss": 0.0822, "step": 11716 }, { "epoch": 0.6246734552433758, "grad_norm": 1.5039606094360352, "learning_rate": 7.656800000000001e-06, "loss": 0.0209, "step": 11717 }, { "epoch": 0.6247267686730288, "grad_norm": 1.0629920959472656, "learning_rate": 7.6566e-06, "loss": 0.0114, "step": 11718 }, { "epoch": 0.6247800821026817, "grad_norm": 2.045078992843628, "learning_rate": 7.6564e-06, "loss": 0.0335, "step": 11719 }, { "epoch": 0.6248333955323346, "grad_norm": 2.319096326828003, "learning_rate": 7.6562e-06, "loss": 0.0333, "step": 11720 }, { "epoch": 0.6248867089619875, "grad_norm": 1.7076644897460938, "learning_rate": 7.656000000000001e-06, "loss": 0.0209, "step": 11721 }, { "epoch": 0.6249400223916405, "grad_norm": 1.5115306377410889, "learning_rate": 7.6558e-06, "loss": 0.0174, "step": 11722 }, { "epoch": 0.6249933358212933, "grad_norm": 5.42600154876709, "learning_rate": 7.6556e-06, "loss": 0.0364, "step": 11723 }, { "epoch": 0.6250466492509463, "grad_norm": 2.0903115272521973, "learning_rate": 7.655400000000001e-06, "loss": 0.0234, "step": 11724 }, { "epoch": 0.6250999626805992, "grad_norm": 1.2557750940322876, "learning_rate": 7.6552e-06, "loss": 0.0248, "step": 11725 }, { "epoch": 0.6251532761102522, "grad_norm": 2.6653459072113037, "learning_rate": 7.655e-06, "loss": 0.0476, "step": 11726 }, { "epoch": 0.6252065895399052, "grad_norm": 0.5634922385215759, "learning_rate": 7.654800000000002e-06, "loss": 0.0085, "step": 11727 }, { "epoch": 0.625259902969558, "grad_norm": 2.1970324516296387, "learning_rate": 7.654600000000001e-06, "loss": 0.0446, "step": 11728 }, { "epoch": 0.625313216399211, "grad_norm": 2.597872495651245, "learning_rate": 7.6544e-06, "loss": 0.0636, "step": 11729 }, { "epoch": 0.6253665298288639, "grad_norm": 3.1553499698638916, "learning_rate": 7.6542e-06, "loss": 0.0589, "step": 11730 }, { "epoch": 0.6254198432585168, "grad_norm": 5.8595428466796875, "learning_rate": 7.654e-06, "loss": 0.0215, "step": 11731 }, { "epoch": 0.6254731566881697, "grad_norm": 2.659942626953125, "learning_rate": 7.6538e-06, "loss": 0.0564, "step": 11732 }, { "epoch": 0.6255264701178227, "grad_norm": 3.027362108230591, "learning_rate": 7.6536e-06, "loss": 0.0491, "step": 11733 }, { "epoch": 0.6255797835474756, "grad_norm": 3.664574384689331, "learning_rate": 7.6534e-06, "loss": 0.0491, "step": 11734 }, { "epoch": 0.6256330969771285, "grad_norm": 3.7716617584228516, "learning_rate": 7.6532e-06, "loss": 0.0516, "step": 11735 }, { "epoch": 0.6256864104067815, "grad_norm": 2.320998430252075, "learning_rate": 7.653000000000001e-06, "loss": 0.0371, "step": 11736 }, { "epoch": 0.6257397238364344, "grad_norm": 1.5336072444915771, "learning_rate": 7.6528e-06, "loss": 0.038, "step": 11737 }, { "epoch": 0.6257930372660874, "grad_norm": 2.432337760925293, "learning_rate": 7.652600000000002e-06, "loss": 0.0407, "step": 11738 }, { "epoch": 0.6258463506957402, "grad_norm": 1.7689037322998047, "learning_rate": 7.652400000000001e-06, "loss": 0.022, "step": 11739 }, { "epoch": 0.6258996641253932, "grad_norm": 1.6585044860839844, "learning_rate": 7.652200000000001e-06, "loss": 0.0276, "step": 11740 }, { "epoch": 0.6259529775550461, "grad_norm": 1.6035761833190918, "learning_rate": 7.652e-06, "loss": 0.0276, "step": 11741 }, { "epoch": 0.6260062909846991, "grad_norm": 3.0607190132141113, "learning_rate": 7.6518e-06, "loss": 0.0399, "step": 11742 }, { "epoch": 0.6260596044143519, "grad_norm": 3.3975489139556885, "learning_rate": 7.6516e-06, "loss": 0.0312, "step": 11743 }, { "epoch": 0.6261129178440049, "grad_norm": 3.309128999710083, "learning_rate": 7.651400000000001e-06, "loss": 0.0439, "step": 11744 }, { "epoch": 0.6261662312736578, "grad_norm": 2.3134191036224365, "learning_rate": 7.6512e-06, "loss": 0.0216, "step": 11745 }, { "epoch": 0.6262195447033108, "grad_norm": 3.1786386966705322, "learning_rate": 7.651e-06, "loss": 0.0421, "step": 11746 }, { "epoch": 0.6262728581329637, "grad_norm": 2.794952154159546, "learning_rate": 7.6508e-06, "loss": 0.0533, "step": 11747 }, { "epoch": 0.6263261715626166, "grad_norm": 2.8648934364318848, "learning_rate": 7.650600000000001e-06, "loss": 0.0519, "step": 11748 }, { "epoch": 0.6263794849922696, "grad_norm": 2.4331963062286377, "learning_rate": 7.6504e-06, "loss": 0.0455, "step": 11749 }, { "epoch": 0.6264327984219225, "grad_norm": 2.2705607414245605, "learning_rate": 7.650200000000002e-06, "loss": 0.0474, "step": 11750 }, { "epoch": 0.6264861118515754, "grad_norm": 1.534363865852356, "learning_rate": 7.650000000000001e-06, "loss": 0.0537, "step": 11751 }, { "epoch": 0.6265394252812283, "grad_norm": 2.2845215797424316, "learning_rate": 7.649800000000001e-06, "loss": 0.0391, "step": 11752 }, { "epoch": 0.6265927387108813, "grad_norm": 1.8175287246704102, "learning_rate": 7.6496e-06, "loss": 0.0202, "step": 11753 }, { "epoch": 0.6266460521405341, "grad_norm": 2.4108219146728516, "learning_rate": 7.6494e-06, "loss": 0.0288, "step": 11754 }, { "epoch": 0.6266993655701871, "grad_norm": 1.3551514148712158, "learning_rate": 7.649200000000001e-06, "loss": 0.0199, "step": 11755 }, { "epoch": 0.6267526789998401, "grad_norm": 2.557690382003784, "learning_rate": 7.649e-06, "loss": 0.032, "step": 11756 }, { "epoch": 0.626805992429493, "grad_norm": 1.4198051691055298, "learning_rate": 7.6488e-06, "loss": 0.0224, "step": 11757 }, { "epoch": 0.626859305859146, "grad_norm": 3.3594911098480225, "learning_rate": 7.6486e-06, "loss": 0.0696, "step": 11758 }, { "epoch": 0.6269126192887988, "grad_norm": 2.657336950302124, "learning_rate": 7.6484e-06, "loss": 0.0266, "step": 11759 }, { "epoch": 0.6269659327184518, "grad_norm": 1.7947900295257568, "learning_rate": 7.6482e-06, "loss": 0.0374, "step": 11760 }, { "epoch": 0.6270192461481047, "grad_norm": 3.1246728897094727, "learning_rate": 7.648e-06, "loss": 0.0304, "step": 11761 }, { "epoch": 0.6270725595777576, "grad_norm": 1.8618048429489136, "learning_rate": 7.647800000000002e-06, "loss": 0.0358, "step": 11762 }, { "epoch": 0.6271258730074105, "grad_norm": 2.0806031227111816, "learning_rate": 7.647600000000001e-06, "loss": 0.031, "step": 11763 }, { "epoch": 0.6271791864370635, "grad_norm": 1.6790046691894531, "learning_rate": 7.6474e-06, "loss": 0.0248, "step": 11764 }, { "epoch": 0.6272324998667165, "grad_norm": 1.4404408931732178, "learning_rate": 7.6472e-06, "loss": 0.0242, "step": 11765 }, { "epoch": 0.6272858132963693, "grad_norm": 0.8799008727073669, "learning_rate": 7.647000000000001e-06, "loss": 0.0135, "step": 11766 }, { "epoch": 0.6273391267260223, "grad_norm": 2.250110626220703, "learning_rate": 7.646800000000001e-06, "loss": 0.0414, "step": 11767 }, { "epoch": 0.6273924401556752, "grad_norm": 0.9231309294700623, "learning_rate": 7.6466e-06, "loss": 0.0162, "step": 11768 }, { "epoch": 0.6274457535853282, "grad_norm": 5.557306289672852, "learning_rate": 7.6464e-06, "loss": 0.0937, "step": 11769 }, { "epoch": 0.627499067014981, "grad_norm": 2.537325859069824, "learning_rate": 7.6462e-06, "loss": 0.0533, "step": 11770 }, { "epoch": 0.627552380444634, "grad_norm": 1.4690417051315308, "learning_rate": 7.646e-06, "loss": 0.038, "step": 11771 }, { "epoch": 0.6276056938742869, "grad_norm": 1.3344701528549194, "learning_rate": 7.6458e-06, "loss": 0.0218, "step": 11772 }, { "epoch": 0.6276590073039399, "grad_norm": 2.8093435764312744, "learning_rate": 7.6456e-06, "loss": 0.0537, "step": 11773 }, { "epoch": 0.6277123207335928, "grad_norm": 1.7030516862869263, "learning_rate": 7.645400000000001e-06, "loss": 0.0382, "step": 11774 }, { "epoch": 0.6277656341632457, "grad_norm": 2.1896796226501465, "learning_rate": 7.645200000000001e-06, "loss": 0.0311, "step": 11775 }, { "epoch": 0.6278189475928987, "grad_norm": 1.6371504068374634, "learning_rate": 7.645e-06, "loss": 0.0162, "step": 11776 }, { "epoch": 0.6278722610225516, "grad_norm": 2.108708143234253, "learning_rate": 7.644800000000002e-06, "loss": 0.0295, "step": 11777 }, { "epoch": 0.6279255744522045, "grad_norm": 2.4749414920806885, "learning_rate": 7.644600000000001e-06, "loss": 0.0711, "step": 11778 }, { "epoch": 0.6279788878818574, "grad_norm": 2.6402363777160645, "learning_rate": 7.644400000000001e-06, "loss": 0.0434, "step": 11779 }, { "epoch": 0.6280322013115104, "grad_norm": 1.334291696548462, "learning_rate": 7.6442e-06, "loss": 0.0264, "step": 11780 }, { "epoch": 0.6280855147411633, "grad_norm": 2.069803476333618, "learning_rate": 7.644e-06, "loss": 0.0486, "step": 11781 }, { "epoch": 0.6281388281708162, "grad_norm": 2.5671260356903076, "learning_rate": 7.6438e-06, "loss": 0.0435, "step": 11782 }, { "epoch": 0.6281921416004692, "grad_norm": 1.4001715183258057, "learning_rate": 7.6436e-06, "loss": 0.0243, "step": 11783 }, { "epoch": 0.6282454550301221, "grad_norm": 1.56633460521698, "learning_rate": 7.6434e-06, "loss": 0.0228, "step": 11784 }, { "epoch": 0.6282987684597751, "grad_norm": 1.7777774333953857, "learning_rate": 7.6432e-06, "loss": 0.0282, "step": 11785 }, { "epoch": 0.6283520818894279, "grad_norm": 1.5075267553329468, "learning_rate": 7.643000000000001e-06, "loss": 0.0421, "step": 11786 }, { "epoch": 0.6284053953190809, "grad_norm": 1.206824779510498, "learning_rate": 7.6428e-06, "loss": 0.0099, "step": 11787 }, { "epoch": 0.6284587087487338, "grad_norm": 2.732839345932007, "learning_rate": 7.6426e-06, "loss": 0.0313, "step": 11788 }, { "epoch": 0.6285120221783868, "grad_norm": 2.1531260013580322, "learning_rate": 7.642400000000002e-06, "loss": 0.0457, "step": 11789 }, { "epoch": 0.6285653356080396, "grad_norm": 1.974014163017273, "learning_rate": 7.642200000000001e-06, "loss": 0.0436, "step": 11790 }, { "epoch": 0.6286186490376926, "grad_norm": 2.0277302265167236, "learning_rate": 7.642e-06, "loss": 0.0229, "step": 11791 }, { "epoch": 0.6286719624673456, "grad_norm": 1.8434617519378662, "learning_rate": 7.6418e-06, "loss": 0.0343, "step": 11792 }, { "epoch": 0.6287252758969984, "grad_norm": 2.201383113861084, "learning_rate": 7.6416e-06, "loss": 0.0258, "step": 11793 }, { "epoch": 0.6287785893266514, "grad_norm": 3.062098979949951, "learning_rate": 7.641400000000001e-06, "loss": 0.0302, "step": 11794 }, { "epoch": 0.6288319027563043, "grad_norm": 1.3180971145629883, "learning_rate": 7.6412e-06, "loss": 0.0217, "step": 11795 }, { "epoch": 0.6288852161859573, "grad_norm": 2.9769394397735596, "learning_rate": 7.641e-06, "loss": 0.0664, "step": 11796 }, { "epoch": 0.6289385296156101, "grad_norm": 2.6409215927124023, "learning_rate": 7.6408e-06, "loss": 0.039, "step": 11797 }, { "epoch": 0.6289918430452631, "grad_norm": 2.346365451812744, "learning_rate": 7.640600000000001e-06, "loss": 0.0367, "step": 11798 }, { "epoch": 0.629045156474916, "grad_norm": 1.7423837184906006, "learning_rate": 7.6404e-06, "loss": 0.0255, "step": 11799 }, { "epoch": 0.629098469904569, "grad_norm": 1.0139497518539429, "learning_rate": 7.6402e-06, "loss": 0.0224, "step": 11800 }, { "epoch": 0.6291517833342218, "grad_norm": 1.6939706802368164, "learning_rate": 7.640000000000001e-06, "loss": 0.0202, "step": 11801 }, { "epoch": 0.6292050967638748, "grad_norm": 1.5353672504425049, "learning_rate": 7.639800000000001e-06, "loss": 0.02, "step": 11802 }, { "epoch": 0.6292584101935278, "grad_norm": 3.2764241695404053, "learning_rate": 7.6396e-06, "loss": 0.0425, "step": 11803 }, { "epoch": 0.6293117236231807, "grad_norm": 1.2119736671447754, "learning_rate": 7.6394e-06, "loss": 0.0163, "step": 11804 }, { "epoch": 0.6293650370528336, "grad_norm": 2.2818100452423096, "learning_rate": 7.639200000000001e-06, "loss": 0.0476, "step": 11805 }, { "epoch": 0.6294183504824865, "grad_norm": 1.1809111833572388, "learning_rate": 7.639000000000001e-06, "loss": 0.0081, "step": 11806 }, { "epoch": 0.6294716639121395, "grad_norm": 1.8750277757644653, "learning_rate": 7.6388e-06, "loss": 0.0331, "step": 11807 }, { "epoch": 0.6295249773417924, "grad_norm": 1.8691059350967407, "learning_rate": 7.6386e-06, "loss": 0.0258, "step": 11808 }, { "epoch": 0.6295782907714453, "grad_norm": 2.0348610877990723, "learning_rate": 7.6384e-06, "loss": 0.0267, "step": 11809 }, { "epoch": 0.6296316042010982, "grad_norm": 1.1491219997406006, "learning_rate": 7.6382e-06, "loss": 0.0417, "step": 11810 }, { "epoch": 0.6296849176307512, "grad_norm": 3.0071136951446533, "learning_rate": 7.638e-06, "loss": 0.0768, "step": 11811 }, { "epoch": 0.6297382310604042, "grad_norm": 1.5727547407150269, "learning_rate": 7.6378e-06, "loss": 0.0228, "step": 11812 }, { "epoch": 0.629791544490057, "grad_norm": 1.9474916458129883, "learning_rate": 7.637600000000001e-06, "loss": 0.0318, "step": 11813 }, { "epoch": 0.62984485791971, "grad_norm": 2.0592174530029297, "learning_rate": 7.6374e-06, "loss": 0.0245, "step": 11814 }, { "epoch": 0.6298981713493629, "grad_norm": 2.7674367427825928, "learning_rate": 7.6372e-06, "loss": 0.0238, "step": 11815 }, { "epoch": 0.6299514847790159, "grad_norm": 2.934417963027954, "learning_rate": 7.637000000000002e-06, "loss": 0.0476, "step": 11816 }, { "epoch": 0.6300047982086687, "grad_norm": 2.1311123371124268, "learning_rate": 7.636800000000001e-06, "loss": 0.0389, "step": 11817 }, { "epoch": 0.6300581116383217, "grad_norm": 4.216391563415527, "learning_rate": 7.6366e-06, "loss": 0.0348, "step": 11818 }, { "epoch": 0.6301114250679746, "grad_norm": 4.125457763671875, "learning_rate": 7.6364e-06, "loss": 0.045, "step": 11819 }, { "epoch": 0.6301647384976276, "grad_norm": 2.2458879947662354, "learning_rate": 7.6362e-06, "loss": 0.0378, "step": 11820 }, { "epoch": 0.6302180519272805, "grad_norm": 1.4399099349975586, "learning_rate": 7.636e-06, "loss": 0.0272, "step": 11821 }, { "epoch": 0.6302713653569334, "grad_norm": 2.790372133255005, "learning_rate": 7.6358e-06, "loss": 0.0351, "step": 11822 }, { "epoch": 0.6303246787865864, "grad_norm": 1.0779669284820557, "learning_rate": 7.6356e-06, "loss": 0.0207, "step": 11823 }, { "epoch": 0.6303779922162392, "grad_norm": 1.860949993133545, "learning_rate": 7.6354e-06, "loss": 0.0506, "step": 11824 }, { "epoch": 0.6304313056458922, "grad_norm": 1.4329562187194824, "learning_rate": 7.635200000000001e-06, "loss": 0.0216, "step": 11825 }, { "epoch": 0.6304846190755451, "grad_norm": 2.365072011947632, "learning_rate": 7.635e-06, "loss": 0.0577, "step": 11826 }, { "epoch": 0.6305379325051981, "grad_norm": 1.8489271402359009, "learning_rate": 7.634800000000002e-06, "loss": 0.0402, "step": 11827 }, { "epoch": 0.6305912459348509, "grad_norm": 1.7437409162521362, "learning_rate": 7.634600000000001e-06, "loss": 0.0471, "step": 11828 }, { "epoch": 0.6306445593645039, "grad_norm": 0.9203386306762695, "learning_rate": 7.634400000000001e-06, "loss": 0.0083, "step": 11829 }, { "epoch": 0.6306978727941569, "grad_norm": 4.326419353485107, "learning_rate": 7.6342e-06, "loss": 0.044, "step": 11830 }, { "epoch": 0.6307511862238098, "grad_norm": 2.1087965965270996, "learning_rate": 7.634e-06, "loss": 0.0323, "step": 11831 }, { "epoch": 0.6308044996534627, "grad_norm": 1.1506644487380981, "learning_rate": 7.6338e-06, "loss": 0.0106, "step": 11832 }, { "epoch": 0.6308578130831156, "grad_norm": 2.902942657470703, "learning_rate": 7.633600000000001e-06, "loss": 0.0317, "step": 11833 }, { "epoch": 0.6309111265127686, "grad_norm": 1.2382441759109497, "learning_rate": 7.6334e-06, "loss": 0.0255, "step": 11834 }, { "epoch": 0.6309644399424215, "grad_norm": 3.3309338092803955, "learning_rate": 7.6332e-06, "loss": 0.0397, "step": 11835 }, { "epoch": 0.6310177533720744, "grad_norm": 2.8732645511627197, "learning_rate": 7.633e-06, "loss": 0.0422, "step": 11836 }, { "epoch": 0.6310710668017273, "grad_norm": 3.274203300476074, "learning_rate": 7.632800000000001e-06, "loss": 0.0444, "step": 11837 }, { "epoch": 0.6311243802313803, "grad_norm": 1.8377989530563354, "learning_rate": 7.6326e-06, "loss": 0.0315, "step": 11838 }, { "epoch": 0.6311776936610333, "grad_norm": 3.502439022064209, "learning_rate": 7.632400000000002e-06, "loss": 0.0364, "step": 11839 }, { "epoch": 0.6312310070906861, "grad_norm": 1.0870240926742554, "learning_rate": 7.632200000000001e-06, "loss": 0.0161, "step": 11840 }, { "epoch": 0.6312843205203391, "grad_norm": 1.7252979278564453, "learning_rate": 7.632e-06, "loss": 0.0255, "step": 11841 }, { "epoch": 0.631337633949992, "grad_norm": 2.198129653930664, "learning_rate": 7.6318e-06, "loss": 0.0265, "step": 11842 }, { "epoch": 0.631390947379645, "grad_norm": 2.0781545639038086, "learning_rate": 7.6316e-06, "loss": 0.0248, "step": 11843 }, { "epoch": 0.6314442608092978, "grad_norm": 1.0731559991836548, "learning_rate": 7.631400000000001e-06, "loss": 0.0101, "step": 11844 }, { "epoch": 0.6314975742389508, "grad_norm": 4.867969989776611, "learning_rate": 7.6312e-06, "loss": 0.0251, "step": 11845 }, { "epoch": 0.6315508876686037, "grad_norm": 2.226848602294922, "learning_rate": 7.631e-06, "loss": 0.0224, "step": 11846 }, { "epoch": 0.6316042010982567, "grad_norm": 3.326284170150757, "learning_rate": 7.6308e-06, "loss": 0.0284, "step": 11847 }, { "epoch": 0.6316575145279095, "grad_norm": 1.8124172687530518, "learning_rate": 7.6306e-06, "loss": 0.0168, "step": 11848 }, { "epoch": 0.6317108279575625, "grad_norm": 1.703385353088379, "learning_rate": 7.6304e-06, "loss": 0.0257, "step": 11849 }, { "epoch": 0.6317641413872155, "grad_norm": 3.5110511779785156, "learning_rate": 7.6302e-06, "loss": 0.0158, "step": 11850 }, { "epoch": 0.6318174548168684, "grad_norm": 4.9105753898620605, "learning_rate": 7.630000000000001e-06, "loss": 0.0299, "step": 11851 }, { "epoch": 0.6318707682465213, "grad_norm": 1.5557016134262085, "learning_rate": 7.629800000000001e-06, "loss": 0.0308, "step": 11852 }, { "epoch": 0.6319240816761742, "grad_norm": 2.1186037063598633, "learning_rate": 7.6296e-06, "loss": 0.0365, "step": 11853 }, { "epoch": 0.6319773951058272, "grad_norm": 1.9561938047409058, "learning_rate": 7.6294e-06, "loss": 0.0568, "step": 11854 }, { "epoch": 0.63203070853548, "grad_norm": 1.288385272026062, "learning_rate": 7.6292000000000014e-06, "loss": 0.0147, "step": 11855 }, { "epoch": 0.632084021965133, "grad_norm": 1.7360860109329224, "learning_rate": 7.629000000000001e-06, "loss": 0.0225, "step": 11856 }, { "epoch": 0.6321373353947859, "grad_norm": 2.939706563949585, "learning_rate": 7.6288000000000005e-06, "loss": 0.0389, "step": 11857 }, { "epoch": 0.6321906488244389, "grad_norm": 3.256883144378662, "learning_rate": 7.6286e-06, "loss": 0.0274, "step": 11858 }, { "epoch": 0.6322439622540919, "grad_norm": 2.545228958129883, "learning_rate": 7.6284000000000005e-06, "loss": 0.0377, "step": 11859 }, { "epoch": 0.6322972756837447, "grad_norm": 1.5995694398880005, "learning_rate": 7.628200000000001e-06, "loss": 0.0118, "step": 11860 }, { "epoch": 0.6323505891133977, "grad_norm": 2.8951032161712646, "learning_rate": 7.628000000000001e-06, "loss": 0.0282, "step": 11861 }, { "epoch": 0.6324039025430506, "grad_norm": 2.42319393157959, "learning_rate": 7.627800000000001e-06, "loss": 0.0225, "step": 11862 }, { "epoch": 0.6324572159727035, "grad_norm": 4.4679694175720215, "learning_rate": 7.6276000000000005e-06, "loss": 0.0465, "step": 11863 }, { "epoch": 0.6325105294023564, "grad_norm": 3.087991714477539, "learning_rate": 7.6274e-06, "loss": 0.0292, "step": 11864 }, { "epoch": 0.6325638428320094, "grad_norm": 2.771479845046997, "learning_rate": 7.6272000000000004e-06, "loss": 0.0648, "step": 11865 }, { "epoch": 0.6326171562616623, "grad_norm": 1.6793872117996216, "learning_rate": 7.627000000000001e-06, "loss": 0.0296, "step": 11866 }, { "epoch": 0.6326704696913152, "grad_norm": 2.1072986125946045, "learning_rate": 7.626800000000001e-06, "loss": 0.0344, "step": 11867 }, { "epoch": 0.6327237831209682, "grad_norm": 2.8659112453460693, "learning_rate": 7.626600000000001e-06, "loss": 0.0184, "step": 11868 }, { "epoch": 0.6327770965506211, "grad_norm": 1.693610429763794, "learning_rate": 7.6264e-06, "loss": 0.0298, "step": 11869 }, { "epoch": 0.6328304099802741, "grad_norm": 2.617673397064209, "learning_rate": 7.6262e-06, "loss": 0.038, "step": 11870 }, { "epoch": 0.6328837234099269, "grad_norm": 4.167818546295166, "learning_rate": 7.626e-06, "loss": 0.0856, "step": 11871 }, { "epoch": 0.6329370368395799, "grad_norm": 1.9953501224517822, "learning_rate": 7.625800000000001e-06, "loss": 0.0266, "step": 11872 }, { "epoch": 0.6329903502692328, "grad_norm": 1.7402592897415161, "learning_rate": 7.625600000000001e-06, "loss": 0.0309, "step": 11873 }, { "epoch": 0.6330436636988858, "grad_norm": 1.9637556076049805, "learning_rate": 7.625400000000001e-06, "loss": 0.0343, "step": 11874 }, { "epoch": 0.6330969771285386, "grad_norm": 4.175352096557617, "learning_rate": 7.6252e-06, "loss": 0.0581, "step": 11875 }, { "epoch": 0.6331502905581916, "grad_norm": 2.7809531688690186, "learning_rate": 7.625e-06, "loss": 0.0258, "step": 11876 }, { "epoch": 0.6332036039878446, "grad_norm": 2.345855951309204, "learning_rate": 7.624800000000001e-06, "loss": 0.0169, "step": 11877 }, { "epoch": 0.6332569174174975, "grad_norm": 3.511776924133301, "learning_rate": 7.624600000000001e-06, "loss": 0.0301, "step": 11878 }, { "epoch": 0.6333102308471504, "grad_norm": 3.133265495300293, "learning_rate": 7.624400000000001e-06, "loss": 0.0291, "step": 11879 }, { "epoch": 0.6333635442768033, "grad_norm": 0.700038731098175, "learning_rate": 7.624200000000001e-06, "loss": 0.0107, "step": 11880 }, { "epoch": 0.6334168577064563, "grad_norm": 2.235293388366699, "learning_rate": 7.624e-06, "loss": 0.033, "step": 11881 }, { "epoch": 0.6334701711361092, "grad_norm": 4.469478607177734, "learning_rate": 7.6238e-06, "loss": 0.059, "step": 11882 }, { "epoch": 0.6335234845657621, "grad_norm": 2.1267189979553223, "learning_rate": 7.623600000000001e-06, "loss": 0.0272, "step": 11883 }, { "epoch": 0.633576797995415, "grad_norm": 1.716987133026123, "learning_rate": 7.6234000000000006e-06, "loss": 0.0255, "step": 11884 }, { "epoch": 0.633630111425068, "grad_norm": 3.6054604053497314, "learning_rate": 7.623200000000001e-06, "loss": 0.0367, "step": 11885 }, { "epoch": 0.633683424854721, "grad_norm": 1.5902132987976074, "learning_rate": 7.6230000000000005e-06, "loss": 0.0361, "step": 11886 }, { "epoch": 0.6337367382843738, "grad_norm": 1.0549627542495728, "learning_rate": 7.6228e-06, "loss": 0.0122, "step": 11887 }, { "epoch": 0.6337900517140268, "grad_norm": 3.4229493141174316, "learning_rate": 7.622600000000001e-06, "loss": 0.0778, "step": 11888 }, { "epoch": 0.6338433651436797, "grad_norm": 2.3435451984405518, "learning_rate": 7.622400000000001e-06, "loss": 0.0317, "step": 11889 }, { "epoch": 0.6338966785733327, "grad_norm": 4.2383575439453125, "learning_rate": 7.6222000000000005e-06, "loss": 0.0658, "step": 11890 }, { "epoch": 0.6339499920029855, "grad_norm": 2.3224363327026367, "learning_rate": 7.622000000000001e-06, "loss": 0.0542, "step": 11891 }, { "epoch": 0.6340033054326385, "grad_norm": 2.6423795223236084, "learning_rate": 7.6218000000000004e-06, "loss": 0.042, "step": 11892 }, { "epoch": 0.6340566188622914, "grad_norm": 2.622314453125, "learning_rate": 7.6216e-06, "loss": 0.05, "step": 11893 }, { "epoch": 0.6341099322919443, "grad_norm": 4.261847496032715, "learning_rate": 7.621400000000001e-06, "loss": 0.0471, "step": 11894 }, { "epoch": 0.6341632457215973, "grad_norm": 4.2038254737854, "learning_rate": 7.621200000000001e-06, "loss": 0.0886, "step": 11895 }, { "epoch": 0.6342165591512502, "grad_norm": 1.5876802206039429, "learning_rate": 7.621e-06, "loss": 0.0274, "step": 11896 }, { "epoch": 0.6342698725809032, "grad_norm": 4.570821285247803, "learning_rate": 7.620800000000001e-06, "loss": 0.0395, "step": 11897 }, { "epoch": 0.634323186010556, "grad_norm": 2.018371820449829, "learning_rate": 7.6206e-06, "loss": 0.0274, "step": 11898 }, { "epoch": 0.634376499440209, "grad_norm": 1.488189697265625, "learning_rate": 7.620400000000001e-06, "loss": 0.0243, "step": 11899 }, { "epoch": 0.6344298128698619, "grad_norm": 3.6372528076171875, "learning_rate": 7.620200000000001e-06, "loss": 0.0363, "step": 11900 }, { "epoch": 0.6344831262995149, "grad_norm": 1.0037883520126343, "learning_rate": 7.620000000000001e-06, "loss": 0.0306, "step": 11901 }, { "epoch": 0.6345364397291677, "grad_norm": 2.2183175086975098, "learning_rate": 7.6198e-06, "loss": 0.0368, "step": 11902 }, { "epoch": 0.6345897531588207, "grad_norm": 3.1891493797302246, "learning_rate": 7.619600000000001e-06, "loss": 0.0591, "step": 11903 }, { "epoch": 0.6346430665884736, "grad_norm": 1.9870047569274902, "learning_rate": 7.6194e-06, "loss": 0.0507, "step": 11904 }, { "epoch": 0.6346963800181266, "grad_norm": 1.4984283447265625, "learning_rate": 7.619200000000001e-06, "loss": 0.0143, "step": 11905 }, { "epoch": 0.6347496934477795, "grad_norm": 1.789409875869751, "learning_rate": 7.619000000000001e-06, "loss": 0.0412, "step": 11906 }, { "epoch": 0.6348030068774324, "grad_norm": 1.334668755531311, "learning_rate": 7.618800000000001e-06, "loss": 0.0351, "step": 11907 }, { "epoch": 0.6348563203070854, "grad_norm": 1.804648756980896, "learning_rate": 7.6186e-06, "loss": 0.016, "step": 11908 }, { "epoch": 0.6349096337367383, "grad_norm": 3.7889328002929688, "learning_rate": 7.618400000000001e-06, "loss": 0.044, "step": 11909 }, { "epoch": 0.6349629471663912, "grad_norm": 1.3118380308151245, "learning_rate": 7.618200000000001e-06, "loss": 0.0431, "step": 11910 }, { "epoch": 0.6350162605960441, "grad_norm": 1.3914369344711304, "learning_rate": 7.618000000000001e-06, "loss": 0.019, "step": 11911 }, { "epoch": 0.6350695740256971, "grad_norm": 1.4797433614730835, "learning_rate": 7.617800000000001e-06, "loss": 0.0202, "step": 11912 }, { "epoch": 0.63512288745535, "grad_norm": 2.6627604961395264, "learning_rate": 7.6176000000000006e-06, "loss": 0.0504, "step": 11913 }, { "epoch": 0.6351762008850029, "grad_norm": 2.583483934402466, "learning_rate": 7.6174e-06, "loss": 0.0371, "step": 11914 }, { "epoch": 0.6352295143146559, "grad_norm": 1.9718616008758545, "learning_rate": 7.6172000000000005e-06, "loss": 0.0416, "step": 11915 }, { "epoch": 0.6352828277443088, "grad_norm": 2.6209325790405273, "learning_rate": 7.617000000000001e-06, "loss": 0.0164, "step": 11916 }, { "epoch": 0.6353361411739618, "grad_norm": 2.3445680141448975, "learning_rate": 7.6168000000000005e-06, "loss": 0.0246, "step": 11917 }, { "epoch": 0.6353894546036146, "grad_norm": 2.916017532348633, "learning_rate": 7.616600000000001e-06, "loss": 0.0514, "step": 11918 }, { "epoch": 0.6354427680332676, "grad_norm": 1.961360216140747, "learning_rate": 7.6164000000000005e-06, "loss": 0.0437, "step": 11919 }, { "epoch": 0.6354960814629205, "grad_norm": 1.3828327655792236, "learning_rate": 7.6162e-06, "loss": 0.0133, "step": 11920 }, { "epoch": 0.6355493948925734, "grad_norm": 1.8025832176208496, "learning_rate": 7.616000000000001e-06, "loss": 0.0149, "step": 11921 }, { "epoch": 0.6356027083222263, "grad_norm": 1.9053945541381836, "learning_rate": 7.615800000000001e-06, "loss": 0.0318, "step": 11922 }, { "epoch": 0.6356560217518793, "grad_norm": 3.1842844486236572, "learning_rate": 7.6156e-06, "loss": 0.0322, "step": 11923 }, { "epoch": 0.6357093351815323, "grad_norm": 2.468571901321411, "learning_rate": 7.615400000000001e-06, "loss": 0.0449, "step": 11924 }, { "epoch": 0.6357626486111851, "grad_norm": 3.214344024658203, "learning_rate": 7.6152e-06, "loss": 0.0601, "step": 11925 }, { "epoch": 0.6358159620408381, "grad_norm": 3.8587753772735596, "learning_rate": 7.615e-06, "loss": 0.0441, "step": 11926 }, { "epoch": 0.635869275470491, "grad_norm": 3.9091956615448, "learning_rate": 7.614800000000001e-06, "loss": 0.0778, "step": 11927 }, { "epoch": 0.635922588900144, "grad_norm": 2.07041335105896, "learning_rate": 7.614600000000001e-06, "loss": 0.0207, "step": 11928 }, { "epoch": 0.6359759023297968, "grad_norm": 2.694422721862793, "learning_rate": 7.6144e-06, "loss": 0.0511, "step": 11929 }, { "epoch": 0.6360292157594498, "grad_norm": 2.510368824005127, "learning_rate": 7.614200000000001e-06, "loss": 0.0517, "step": 11930 }, { "epoch": 0.6360825291891027, "grad_norm": 2.937023878097534, "learning_rate": 7.614e-06, "loss": 0.0326, "step": 11931 }, { "epoch": 0.6361358426187557, "grad_norm": 3.035989999771118, "learning_rate": 7.6138e-06, "loss": 0.0571, "step": 11932 }, { "epoch": 0.6361891560484086, "grad_norm": 1.710360050201416, "learning_rate": 7.613600000000001e-06, "loss": 0.0366, "step": 11933 }, { "epoch": 0.6362424694780615, "grad_norm": 2.0908524990081787, "learning_rate": 7.613400000000001e-06, "loss": 0.0326, "step": 11934 }, { "epoch": 0.6362957829077145, "grad_norm": 4.743314266204834, "learning_rate": 7.6132e-06, "loss": 0.0512, "step": 11935 }, { "epoch": 0.6363490963373674, "grad_norm": 3.25576114654541, "learning_rate": 7.613000000000001e-06, "loss": 0.0327, "step": 11936 }, { "epoch": 0.6364024097670203, "grad_norm": 2.3746657371520996, "learning_rate": 7.6128e-06, "loss": 0.021, "step": 11937 }, { "epoch": 0.6364557231966732, "grad_norm": 3.4322123527526855, "learning_rate": 7.6126000000000014e-06, "loss": 0.0321, "step": 11938 }, { "epoch": 0.6365090366263262, "grad_norm": 1.8902307748794556, "learning_rate": 7.612400000000001e-06, "loss": 0.0256, "step": 11939 }, { "epoch": 0.636562350055979, "grad_norm": 1.470757007598877, "learning_rate": 7.6122000000000006e-06, "loss": 0.0172, "step": 11940 }, { "epoch": 0.636615663485632, "grad_norm": 1.420048713684082, "learning_rate": 7.612e-06, "loss": 0.03, "step": 11941 }, { "epoch": 0.636668976915285, "grad_norm": 2.831373691558838, "learning_rate": 7.6118000000000005e-06, "loss": 0.0384, "step": 11942 }, { "epoch": 0.6367222903449379, "grad_norm": 1.5260956287384033, "learning_rate": 7.6116e-06, "loss": 0.0326, "step": 11943 }, { "epoch": 0.6367756037745909, "grad_norm": 2.2606499195098877, "learning_rate": 7.611400000000001e-06, "loss": 0.0279, "step": 11944 }, { "epoch": 0.6368289172042437, "grad_norm": 1.4830511808395386, "learning_rate": 7.611200000000001e-06, "loss": 0.0254, "step": 11945 }, { "epoch": 0.6368822306338967, "grad_norm": 2.952850103378296, "learning_rate": 7.6110000000000005e-06, "loss": 0.0518, "step": 11946 }, { "epoch": 0.6369355440635496, "grad_norm": 3.0996153354644775, "learning_rate": 7.6108e-06, "loss": 0.0435, "step": 11947 }, { "epoch": 0.6369888574932026, "grad_norm": 2.427896022796631, "learning_rate": 7.6106000000000005e-06, "loss": 0.0402, "step": 11948 }, { "epoch": 0.6370421709228554, "grad_norm": 2.0083444118499756, "learning_rate": 7.610400000000001e-06, "loss": 0.026, "step": 11949 }, { "epoch": 0.6370954843525084, "grad_norm": 1.5936075448989868, "learning_rate": 7.610200000000001e-06, "loss": 0.0236, "step": 11950 }, { "epoch": 0.6371487977821614, "grad_norm": 1.8873134851455688, "learning_rate": 7.610000000000001e-06, "loss": 0.0346, "step": 11951 }, { "epoch": 0.6372021112118142, "grad_norm": 1.1027108430862427, "learning_rate": 7.6098e-06, "loss": 0.0121, "step": 11952 }, { "epoch": 0.6372554246414672, "grad_norm": 2.8288190364837646, "learning_rate": 7.6096e-06, "loss": 0.0602, "step": 11953 }, { "epoch": 0.6373087380711201, "grad_norm": 3.320058822631836, "learning_rate": 7.6094e-06, "loss": 0.0301, "step": 11954 }, { "epoch": 0.6373620515007731, "grad_norm": 1.8917678594589233, "learning_rate": 7.609200000000001e-06, "loss": 0.03, "step": 11955 }, { "epoch": 0.6374153649304259, "grad_norm": 2.209031343460083, "learning_rate": 7.609000000000001e-06, "loss": 0.0291, "step": 11956 }, { "epoch": 0.6374686783600789, "grad_norm": 1.2429968118667603, "learning_rate": 7.608800000000001e-06, "loss": 0.01, "step": 11957 }, { "epoch": 0.6375219917897318, "grad_norm": 2.6318602561950684, "learning_rate": 7.6086e-06, "loss": 0.0334, "step": 11958 }, { "epoch": 0.6375753052193848, "grad_norm": 2.3989505767822266, "learning_rate": 7.6084e-06, "loss": 0.046, "step": 11959 }, { "epoch": 0.6376286186490376, "grad_norm": 3.252295970916748, "learning_rate": 7.608200000000001e-06, "loss": 0.0387, "step": 11960 }, { "epoch": 0.6376819320786906, "grad_norm": 5.1319499015808105, "learning_rate": 7.608000000000001e-06, "loss": 0.04, "step": 11961 }, { "epoch": 0.6377352455083436, "grad_norm": 2.349337100982666, "learning_rate": 7.607800000000001e-06, "loss": 0.0293, "step": 11962 }, { "epoch": 0.6377885589379965, "grad_norm": 3.1240696907043457, "learning_rate": 7.607600000000001e-06, "loss": 0.0518, "step": 11963 }, { "epoch": 0.6378418723676494, "grad_norm": 2.275747776031494, "learning_rate": 7.6074e-06, "loss": 0.0412, "step": 11964 }, { "epoch": 0.6378951857973023, "grad_norm": 2.3119471073150635, "learning_rate": 7.6072e-06, "loss": 0.0365, "step": 11965 }, { "epoch": 0.6379484992269553, "grad_norm": 3.6053411960601807, "learning_rate": 7.607000000000001e-06, "loss": 0.0441, "step": 11966 }, { "epoch": 0.6380018126566082, "grad_norm": 1.9406601190567017, "learning_rate": 7.606800000000001e-06, "loss": 0.0344, "step": 11967 }, { "epoch": 0.6380551260862611, "grad_norm": 2.7044694423675537, "learning_rate": 7.606600000000001e-06, "loss": 0.0593, "step": 11968 }, { "epoch": 0.638108439515914, "grad_norm": 2.496394157409668, "learning_rate": 7.6064000000000006e-06, "loss": 0.0511, "step": 11969 }, { "epoch": 0.638161752945567, "grad_norm": 5.9773077964782715, "learning_rate": 7.6062e-06, "loss": 0.0474, "step": 11970 }, { "epoch": 0.63821506637522, "grad_norm": 2.484865427017212, "learning_rate": 7.606000000000001e-06, "loss": 0.0545, "step": 11971 }, { "epoch": 0.6382683798048728, "grad_norm": 0.9620271921157837, "learning_rate": 7.605800000000001e-06, "loss": 0.0149, "step": 11972 }, { "epoch": 0.6383216932345258, "grad_norm": 0.850452721118927, "learning_rate": 7.6056000000000005e-06, "loss": 0.0115, "step": 11973 }, { "epoch": 0.6383750066641787, "grad_norm": 2.9918200969696045, "learning_rate": 7.605400000000001e-06, "loss": 0.0731, "step": 11974 }, { "epoch": 0.6384283200938317, "grad_norm": 1.5603805780410767, "learning_rate": 7.6052000000000005e-06, "loss": 0.0256, "step": 11975 }, { "epoch": 0.6384816335234845, "grad_norm": 2.582578182220459, "learning_rate": 7.605e-06, "loss": 0.0213, "step": 11976 }, { "epoch": 0.6385349469531375, "grad_norm": 1.8083090782165527, "learning_rate": 7.604800000000001e-06, "loss": 0.0568, "step": 11977 }, { "epoch": 0.6385882603827904, "grad_norm": 2.659850835800171, "learning_rate": 7.604600000000001e-06, "loss": 0.0366, "step": 11978 }, { "epoch": 0.6386415738124434, "grad_norm": 1.3539022207260132, "learning_rate": 7.6044e-06, "loss": 0.0298, "step": 11979 }, { "epoch": 0.6386948872420963, "grad_norm": 3.011385679244995, "learning_rate": 7.604200000000001e-06, "loss": 0.0889, "step": 11980 }, { "epoch": 0.6387482006717492, "grad_norm": 1.7299270629882812, "learning_rate": 7.604e-06, "loss": 0.0342, "step": 11981 }, { "epoch": 0.6388015141014022, "grad_norm": 2.954820156097412, "learning_rate": 7.6038e-06, "loss": 0.0462, "step": 11982 }, { "epoch": 0.638854827531055, "grad_norm": 3.1133623123168945, "learning_rate": 7.603600000000001e-06, "loss": 0.033, "step": 11983 }, { "epoch": 0.638908140960708, "grad_norm": 2.6379003524780273, "learning_rate": 7.603400000000001e-06, "loss": 0.0261, "step": 11984 }, { "epoch": 0.6389614543903609, "grad_norm": 1.679430603981018, "learning_rate": 7.6032e-06, "loss": 0.037, "step": 11985 }, { "epoch": 0.6390147678200139, "grad_norm": 1.0853654146194458, "learning_rate": 7.603000000000001e-06, "loss": 0.0159, "step": 11986 }, { "epoch": 0.6390680812496667, "grad_norm": 1.9673364162445068, "learning_rate": 7.6028e-06, "loss": 0.0491, "step": 11987 }, { "epoch": 0.6391213946793197, "grad_norm": 1.2959648370742798, "learning_rate": 7.602600000000001e-06, "loss": 0.017, "step": 11988 }, { "epoch": 0.6391747081089727, "grad_norm": 2.6506593227386475, "learning_rate": 7.602400000000001e-06, "loss": 0.0666, "step": 11989 }, { "epoch": 0.6392280215386256, "grad_norm": 2.4975521564483643, "learning_rate": 7.602200000000001e-06, "loss": 0.0514, "step": 11990 }, { "epoch": 0.6392813349682785, "grad_norm": 1.6739214658737183, "learning_rate": 7.602e-06, "loss": 0.0358, "step": 11991 }, { "epoch": 0.6393346483979314, "grad_norm": 1.1957728862762451, "learning_rate": 7.601800000000001e-06, "loss": 0.0197, "step": 11992 }, { "epoch": 0.6393879618275844, "grad_norm": 1.9131494760513306, "learning_rate": 7.6016e-06, "loss": 0.057, "step": 11993 }, { "epoch": 0.6394412752572373, "grad_norm": 1.0883455276489258, "learning_rate": 7.601400000000001e-06, "loss": 0.0248, "step": 11994 }, { "epoch": 0.6394945886868902, "grad_norm": 2.6079890727996826, "learning_rate": 7.601200000000001e-06, "loss": 0.0295, "step": 11995 }, { "epoch": 0.6395479021165431, "grad_norm": 3.698099136352539, "learning_rate": 7.601000000000001e-06, "loss": 0.0647, "step": 11996 }, { "epoch": 0.6396012155461961, "grad_norm": 1.456695318222046, "learning_rate": 7.6008e-06, "loss": 0.0165, "step": 11997 }, { "epoch": 0.6396545289758491, "grad_norm": 1.5214000940322876, "learning_rate": 7.6006000000000005e-06, "loss": 0.032, "step": 11998 }, { "epoch": 0.6397078424055019, "grad_norm": 1.0009510517120361, "learning_rate": 7.600400000000001e-06, "loss": 0.0115, "step": 11999 }, { "epoch": 0.6397611558351549, "grad_norm": 1.0163925886154175, "learning_rate": 7.6002000000000005e-06, "loss": 0.0257, "step": 12000 }, { "epoch": 0.6398144692648078, "grad_norm": 1.6141093969345093, "learning_rate": 7.600000000000001e-06, "loss": 0.018, "step": 12001 }, { "epoch": 0.6398677826944608, "grad_norm": 1.378556728363037, "learning_rate": 7.5998000000000005e-06, "loss": 0.0113, "step": 12002 }, { "epoch": 0.6399210961241136, "grad_norm": 2.5105457305908203, "learning_rate": 7.5996e-06, "loss": 0.0479, "step": 12003 }, { "epoch": 0.6399744095537666, "grad_norm": 1.5898573398590088, "learning_rate": 7.5994000000000005e-06, "loss": 0.0188, "step": 12004 }, { "epoch": 0.6400277229834195, "grad_norm": 2.4604601860046387, "learning_rate": 7.599200000000001e-06, "loss": 0.0291, "step": 12005 }, { "epoch": 0.6400810364130725, "grad_norm": 1.4369829893112183, "learning_rate": 7.5990000000000004e-06, "loss": 0.0223, "step": 12006 }, { "epoch": 0.6401343498427254, "grad_norm": 2.960115909576416, "learning_rate": 7.598800000000001e-06, "loss": 0.0433, "step": 12007 }, { "epoch": 0.6401876632723783, "grad_norm": 1.9087321758270264, "learning_rate": 7.5986e-06, "loss": 0.0266, "step": 12008 }, { "epoch": 0.6402409767020313, "grad_norm": 3.8179707527160645, "learning_rate": 7.5984e-06, "loss": 0.0566, "step": 12009 }, { "epoch": 0.6402942901316842, "grad_norm": 1.7594053745269775, "learning_rate": 7.598200000000001e-06, "loss": 0.0141, "step": 12010 }, { "epoch": 0.6403476035613371, "grad_norm": 3.154690742492676, "learning_rate": 7.598000000000001e-06, "loss": 0.0297, "step": 12011 }, { "epoch": 0.64040091699099, "grad_norm": 2.4143576622009277, "learning_rate": 7.5978e-06, "loss": 0.0428, "step": 12012 }, { "epoch": 0.640454230420643, "grad_norm": 1.7440450191497803, "learning_rate": 7.597600000000001e-06, "loss": 0.0204, "step": 12013 }, { "epoch": 0.6405075438502958, "grad_norm": 1.2488840818405151, "learning_rate": 7.5974e-06, "loss": 0.0206, "step": 12014 }, { "epoch": 0.6405608572799488, "grad_norm": 1.962119221687317, "learning_rate": 7.5972e-06, "loss": 0.0307, "step": 12015 }, { "epoch": 0.6406141707096017, "grad_norm": 2.4644854068756104, "learning_rate": 7.597000000000001e-06, "loss": 0.0295, "step": 12016 }, { "epoch": 0.6406674841392547, "grad_norm": 2.8491525650024414, "learning_rate": 7.596800000000001e-06, "loss": 0.0404, "step": 12017 }, { "epoch": 0.6407207975689077, "grad_norm": 2.878948450088501, "learning_rate": 7.5966e-06, "loss": 0.023, "step": 12018 }, { "epoch": 0.6407741109985605, "grad_norm": 2.5254836082458496, "learning_rate": 7.596400000000001e-06, "loss": 0.0298, "step": 12019 }, { "epoch": 0.6408274244282135, "grad_norm": 2.9209787845611572, "learning_rate": 7.5962e-06, "loss": 0.0609, "step": 12020 }, { "epoch": 0.6408807378578664, "grad_norm": 1.2120096683502197, "learning_rate": 7.5960000000000015e-06, "loss": 0.0286, "step": 12021 }, { "epoch": 0.6409340512875193, "grad_norm": 1.854695439338684, "learning_rate": 7.595800000000001e-06, "loss": 0.0297, "step": 12022 }, { "epoch": 0.6409873647171722, "grad_norm": 2.047586441040039, "learning_rate": 7.595600000000001e-06, "loss": 0.0189, "step": 12023 }, { "epoch": 0.6410406781468252, "grad_norm": 0.9574650526046753, "learning_rate": 7.5954e-06, "loss": 0.0094, "step": 12024 }, { "epoch": 0.6410939915764781, "grad_norm": 2.2988085746765137, "learning_rate": 7.5952000000000006e-06, "loss": 0.047, "step": 12025 }, { "epoch": 0.641147305006131, "grad_norm": 2.146151065826416, "learning_rate": 7.595e-06, "loss": 0.0385, "step": 12026 }, { "epoch": 0.641200618435784, "grad_norm": 3.2137231826782227, "learning_rate": 7.594800000000001e-06, "loss": 0.0579, "step": 12027 }, { "epoch": 0.6412539318654369, "grad_norm": 2.491318702697754, "learning_rate": 7.594600000000001e-06, "loss": 0.0393, "step": 12028 }, { "epoch": 0.6413072452950899, "grad_norm": 2.262988805770874, "learning_rate": 7.5944000000000005e-06, "loss": 0.0444, "step": 12029 }, { "epoch": 0.6413605587247427, "grad_norm": 2.3862342834472656, "learning_rate": 7.5942e-06, "loss": 0.0362, "step": 12030 }, { "epoch": 0.6414138721543957, "grad_norm": 2.6160168647766113, "learning_rate": 7.5940000000000005e-06, "loss": 0.0187, "step": 12031 }, { "epoch": 0.6414671855840486, "grad_norm": 2.5648982524871826, "learning_rate": 7.593800000000001e-06, "loss": 0.023, "step": 12032 }, { "epoch": 0.6415204990137016, "grad_norm": 6.168550491333008, "learning_rate": 7.593600000000001e-06, "loss": 0.064, "step": 12033 }, { "epoch": 0.6415738124433544, "grad_norm": 2.0979843139648438, "learning_rate": 7.593400000000001e-06, "loss": 0.045, "step": 12034 }, { "epoch": 0.6416271258730074, "grad_norm": 1.8872191905975342, "learning_rate": 7.5932e-06, "loss": 0.0307, "step": 12035 }, { "epoch": 0.6416804393026604, "grad_norm": 2.3295321464538574, "learning_rate": 7.593e-06, "loss": 0.0197, "step": 12036 }, { "epoch": 0.6417337527323133, "grad_norm": 1.475940465927124, "learning_rate": 7.5928e-06, "loss": 0.0245, "step": 12037 }, { "epoch": 0.6417870661619662, "grad_norm": 1.8698478937149048, "learning_rate": 7.592600000000001e-06, "loss": 0.0509, "step": 12038 }, { "epoch": 0.6418403795916191, "grad_norm": 2.268731117248535, "learning_rate": 7.592400000000001e-06, "loss": 0.0387, "step": 12039 }, { "epoch": 0.6418936930212721, "grad_norm": 1.9444630146026611, "learning_rate": 7.592200000000001e-06, "loss": 0.02, "step": 12040 }, { "epoch": 0.641947006450925, "grad_norm": 1.9540010690689087, "learning_rate": 7.592e-06, "loss": 0.0425, "step": 12041 }, { "epoch": 0.6420003198805779, "grad_norm": 1.9780035018920898, "learning_rate": 7.5918e-06, "loss": 0.037, "step": 12042 }, { "epoch": 0.6420536333102308, "grad_norm": 3.365431547164917, "learning_rate": 7.5916e-06, "loss": 0.0413, "step": 12043 }, { "epoch": 0.6421069467398838, "grad_norm": 2.6273014545440674, "learning_rate": 7.591400000000001e-06, "loss": 0.0446, "step": 12044 }, { "epoch": 0.6421602601695368, "grad_norm": 0.693260908126831, "learning_rate": 7.591200000000001e-06, "loss": 0.0073, "step": 12045 }, { "epoch": 0.6422135735991896, "grad_norm": 2.2821109294891357, "learning_rate": 7.591000000000001e-06, "loss": 0.0274, "step": 12046 }, { "epoch": 0.6422668870288426, "grad_norm": 1.7950180768966675, "learning_rate": 7.5908e-06, "loss": 0.0418, "step": 12047 }, { "epoch": 0.6423202004584955, "grad_norm": 2.3238651752471924, "learning_rate": 7.5906e-06, "loss": 0.0262, "step": 12048 }, { "epoch": 0.6423735138881485, "grad_norm": 1.7996931076049805, "learning_rate": 7.590400000000001e-06, "loss": 0.0249, "step": 12049 }, { "epoch": 0.6424268273178013, "grad_norm": 2.2912392616271973, "learning_rate": 7.590200000000001e-06, "loss": 0.0499, "step": 12050 }, { "epoch": 0.6424801407474543, "grad_norm": 3.805650472640991, "learning_rate": 7.590000000000001e-06, "loss": 0.0348, "step": 12051 }, { "epoch": 0.6425334541771072, "grad_norm": 2.0435879230499268, "learning_rate": 7.589800000000001e-06, "loss": 0.0164, "step": 12052 }, { "epoch": 0.6425867676067601, "grad_norm": 0.7360038757324219, "learning_rate": 7.5896e-06, "loss": 0.0072, "step": 12053 }, { "epoch": 0.6426400810364131, "grad_norm": 3.0912981033325195, "learning_rate": 7.5894e-06, "loss": 0.0281, "step": 12054 }, { "epoch": 0.642693394466066, "grad_norm": 3.5819473266601562, "learning_rate": 7.589200000000001e-06, "loss": 0.0703, "step": 12055 }, { "epoch": 0.642746707895719, "grad_norm": 1.0666769742965698, "learning_rate": 7.5890000000000005e-06, "loss": 0.0076, "step": 12056 }, { "epoch": 0.6428000213253718, "grad_norm": 0.9087952375411987, "learning_rate": 7.588800000000001e-06, "loss": 0.0107, "step": 12057 }, { "epoch": 0.6428533347550248, "grad_norm": 1.8369152545928955, "learning_rate": 7.5886000000000005e-06, "loss": 0.022, "step": 12058 }, { "epoch": 0.6429066481846777, "grad_norm": 3.4681947231292725, "learning_rate": 7.5884e-06, "loss": 0.0878, "step": 12059 }, { "epoch": 0.6429599616143307, "grad_norm": 2.214663505554199, "learning_rate": 7.588200000000001e-06, "loss": 0.0452, "step": 12060 }, { "epoch": 0.6430132750439835, "grad_norm": 2.319483518600464, "learning_rate": 7.588000000000001e-06, "loss": 0.04, "step": 12061 }, { "epoch": 0.6430665884736365, "grad_norm": 2.0258281230926514, "learning_rate": 7.5878000000000004e-06, "loss": 0.0482, "step": 12062 }, { "epoch": 0.6431199019032895, "grad_norm": 10.980022430419922, "learning_rate": 7.587600000000001e-06, "loss": 0.0259, "step": 12063 }, { "epoch": 0.6431732153329424, "grad_norm": 2.0210537910461426, "learning_rate": 7.5874e-06, "loss": 0.0277, "step": 12064 }, { "epoch": 0.6432265287625953, "grad_norm": 2.836416244506836, "learning_rate": 7.5872e-06, "loss": 0.0307, "step": 12065 }, { "epoch": 0.6432798421922482, "grad_norm": 1.8422778844833374, "learning_rate": 7.587000000000001e-06, "loss": 0.0471, "step": 12066 }, { "epoch": 0.6433331556219012, "grad_norm": 10.372753143310547, "learning_rate": 7.586800000000001e-06, "loss": 0.0479, "step": 12067 }, { "epoch": 0.643386469051554, "grad_norm": 2.8908803462982178, "learning_rate": 7.5866e-06, "loss": 0.0245, "step": 12068 }, { "epoch": 0.643439782481207, "grad_norm": 2.663986921310425, "learning_rate": 7.586400000000001e-06, "loss": 0.034, "step": 12069 }, { "epoch": 0.6434930959108599, "grad_norm": 1.3203685283660889, "learning_rate": 7.5862e-06, "loss": 0.0196, "step": 12070 }, { "epoch": 0.6435464093405129, "grad_norm": 4.55471658706665, "learning_rate": 7.586000000000001e-06, "loss": 0.023, "step": 12071 }, { "epoch": 0.6435997227701657, "grad_norm": 3.513582706451416, "learning_rate": 7.585800000000001e-06, "loss": 0.0577, "step": 12072 }, { "epoch": 0.6436530361998187, "grad_norm": 2.2735981941223145, "learning_rate": 7.585600000000001e-06, "loss": 0.035, "step": 12073 }, { "epoch": 0.6437063496294717, "grad_norm": 3.579679012298584, "learning_rate": 7.5854e-06, "loss": 0.0333, "step": 12074 }, { "epoch": 0.6437596630591246, "grad_norm": 2.1312477588653564, "learning_rate": 7.585200000000001e-06, "loss": 0.0314, "step": 12075 }, { "epoch": 0.6438129764887776, "grad_norm": 1.684893012046814, "learning_rate": 7.585e-06, "loss": 0.0188, "step": 12076 }, { "epoch": 0.6438662899184304, "grad_norm": 2.909447193145752, "learning_rate": 7.584800000000001e-06, "loss": 0.0498, "step": 12077 }, { "epoch": 0.6439196033480834, "grad_norm": 2.6982414722442627, "learning_rate": 7.584600000000001e-06, "loss": 0.0592, "step": 12078 }, { "epoch": 0.6439729167777363, "grad_norm": 2.416572332382202, "learning_rate": 7.584400000000001e-06, "loss": 0.0296, "step": 12079 }, { "epoch": 0.6440262302073892, "grad_norm": 2.2979233264923096, "learning_rate": 7.5842e-06, "loss": 0.0537, "step": 12080 }, { "epoch": 0.6440795436370421, "grad_norm": 1.6428582668304443, "learning_rate": 7.5840000000000006e-06, "loss": 0.0274, "step": 12081 }, { "epoch": 0.6441328570666951, "grad_norm": 1.3688346147537231, "learning_rate": 7.583800000000001e-06, "loss": 0.0194, "step": 12082 }, { "epoch": 0.6441861704963481, "grad_norm": 1.8990463018417358, "learning_rate": 7.5836000000000005e-06, "loss": 0.0269, "step": 12083 }, { "epoch": 0.6442394839260009, "grad_norm": 3.07470440864563, "learning_rate": 7.583400000000001e-06, "loss": 0.0441, "step": 12084 }, { "epoch": 0.6442927973556539, "grad_norm": 1.4370334148406982, "learning_rate": 7.5832000000000005e-06, "loss": 0.0293, "step": 12085 }, { "epoch": 0.6443461107853068, "grad_norm": 5.627572059631348, "learning_rate": 7.583e-06, "loss": 0.087, "step": 12086 }, { "epoch": 0.6443994242149598, "grad_norm": 1.7877393960952759, "learning_rate": 7.5828000000000005e-06, "loss": 0.0256, "step": 12087 }, { "epoch": 0.6444527376446126, "grad_norm": 1.5924514532089233, "learning_rate": 7.582600000000001e-06, "loss": 0.0258, "step": 12088 }, { "epoch": 0.6445060510742656, "grad_norm": 3.2348289489746094, "learning_rate": 7.5824000000000005e-06, "loss": 0.0513, "step": 12089 }, { "epoch": 0.6445593645039185, "grad_norm": 3.4000675678253174, "learning_rate": 7.582200000000001e-06, "loss": 0.0399, "step": 12090 }, { "epoch": 0.6446126779335715, "grad_norm": 2.1306028366088867, "learning_rate": 7.582e-06, "loss": 0.0272, "step": 12091 }, { "epoch": 0.6446659913632244, "grad_norm": 1.7001333236694336, "learning_rate": 7.5818e-06, "loss": 0.0335, "step": 12092 }, { "epoch": 0.6447193047928773, "grad_norm": 1.2244603633880615, "learning_rate": 7.5816e-06, "loss": 0.0197, "step": 12093 }, { "epoch": 0.6447726182225303, "grad_norm": 1.7348692417144775, "learning_rate": 7.581400000000001e-06, "loss": 0.0171, "step": 12094 }, { "epoch": 0.6448259316521832, "grad_norm": 2.4750266075134277, "learning_rate": 7.5812e-06, "loss": 0.0586, "step": 12095 }, { "epoch": 0.6448792450818361, "grad_norm": 5.474537372589111, "learning_rate": 7.581000000000001e-06, "loss": 0.0379, "step": 12096 }, { "epoch": 0.644932558511489, "grad_norm": 2.477369785308838, "learning_rate": 7.5808e-06, "loss": 0.0342, "step": 12097 }, { "epoch": 0.644985871941142, "grad_norm": 1.982712984085083, "learning_rate": 7.5806e-06, "loss": 0.0431, "step": 12098 }, { "epoch": 0.6450391853707949, "grad_norm": 2.1285881996154785, "learning_rate": 7.580400000000001e-06, "loss": 0.0372, "step": 12099 }, { "epoch": 0.6450924988004478, "grad_norm": 2.8118295669555664, "learning_rate": 7.580200000000001e-06, "loss": 0.0132, "step": 12100 }, { "epoch": 0.6451458122301008, "grad_norm": 1.6140596866607666, "learning_rate": 7.58e-06, "loss": 0.0249, "step": 12101 }, { "epoch": 0.6451991256597537, "grad_norm": 3.8077991008758545, "learning_rate": 7.579800000000001e-06, "loss": 0.0313, "step": 12102 }, { "epoch": 0.6452524390894067, "grad_norm": 1.7818121910095215, "learning_rate": 7.5796e-06, "loss": 0.0288, "step": 12103 }, { "epoch": 0.6453057525190595, "grad_norm": 1.5208076238632202, "learning_rate": 7.5794e-06, "loss": 0.0172, "step": 12104 }, { "epoch": 0.6453590659487125, "grad_norm": 2.882786750793457, "learning_rate": 7.579200000000001e-06, "loss": 0.029, "step": 12105 }, { "epoch": 0.6454123793783654, "grad_norm": 0.9868806004524231, "learning_rate": 7.579000000000001e-06, "loss": 0.011, "step": 12106 }, { "epoch": 0.6454656928080184, "grad_norm": 1.4881815910339355, "learning_rate": 7.5788e-06, "loss": 0.0216, "step": 12107 }, { "epoch": 0.6455190062376712, "grad_norm": 3.11846661567688, "learning_rate": 7.578600000000001e-06, "loss": 0.0275, "step": 12108 }, { "epoch": 0.6455723196673242, "grad_norm": 1.1947113275527954, "learning_rate": 7.5784e-06, "loss": 0.0158, "step": 12109 }, { "epoch": 0.6456256330969772, "grad_norm": 1.060109257698059, "learning_rate": 7.578200000000001e-06, "loss": 0.009, "step": 12110 }, { "epoch": 0.64567894652663, "grad_norm": 3.0940332412719727, "learning_rate": 7.578000000000001e-06, "loss": 0.0748, "step": 12111 }, { "epoch": 0.645732259956283, "grad_norm": 1.189038872718811, "learning_rate": 7.5778000000000005e-06, "loss": 0.0267, "step": 12112 }, { "epoch": 0.6457855733859359, "grad_norm": 2.291111946105957, "learning_rate": 7.5776e-06, "loss": 0.0363, "step": 12113 }, { "epoch": 0.6458388868155889, "grad_norm": 2.0855183601379395, "learning_rate": 7.5774000000000005e-06, "loss": 0.0111, "step": 12114 }, { "epoch": 0.6458922002452417, "grad_norm": 1.9423900842666626, "learning_rate": 7.5772e-06, "loss": 0.0352, "step": 12115 }, { "epoch": 0.6459455136748947, "grad_norm": 1.0472406148910522, "learning_rate": 7.577000000000001e-06, "loss": 0.0233, "step": 12116 }, { "epoch": 0.6459988271045476, "grad_norm": 1.173086404800415, "learning_rate": 7.576800000000001e-06, "loss": 0.0082, "step": 12117 }, { "epoch": 0.6460521405342006, "grad_norm": 1.6660510301589966, "learning_rate": 7.5766000000000004e-06, "loss": 0.0218, "step": 12118 }, { "epoch": 0.6461054539638535, "grad_norm": 3.8167734146118164, "learning_rate": 7.5764e-06, "loss": 0.0716, "step": 12119 }, { "epoch": 0.6461587673935064, "grad_norm": 6.082237720489502, "learning_rate": 7.5762e-06, "loss": 0.0572, "step": 12120 }, { "epoch": 0.6462120808231594, "grad_norm": 2.140031576156616, "learning_rate": 7.576000000000001e-06, "loss": 0.0187, "step": 12121 }, { "epoch": 0.6462653942528123, "grad_norm": 1.8083513975143433, "learning_rate": 7.575800000000001e-06, "loss": 0.0299, "step": 12122 }, { "epoch": 0.6463187076824652, "grad_norm": 3.290821075439453, "learning_rate": 7.575600000000001e-06, "loss": 0.0354, "step": 12123 }, { "epoch": 0.6463720211121181, "grad_norm": 1.5132253170013428, "learning_rate": 7.5754e-06, "loss": 0.0229, "step": 12124 }, { "epoch": 0.6464253345417711, "grad_norm": 1.877509593963623, "learning_rate": 7.5752e-06, "loss": 0.0258, "step": 12125 }, { "epoch": 0.646478647971424, "grad_norm": 2.086045265197754, "learning_rate": 7.575e-06, "loss": 0.0253, "step": 12126 }, { "epoch": 0.6465319614010769, "grad_norm": 2.436081886291504, "learning_rate": 7.574800000000001e-06, "loss": 0.0225, "step": 12127 }, { "epoch": 0.6465852748307298, "grad_norm": 1.946751356124878, "learning_rate": 7.574600000000001e-06, "loss": 0.0218, "step": 12128 }, { "epoch": 0.6466385882603828, "grad_norm": 2.9670636653900146, "learning_rate": 7.574400000000001e-06, "loss": 0.0717, "step": 12129 }, { "epoch": 0.6466919016900358, "grad_norm": 3.9078032970428467, "learning_rate": 7.5742e-06, "loss": 0.0409, "step": 12130 }, { "epoch": 0.6467452151196886, "grad_norm": 4.0785813331604, "learning_rate": 7.574e-06, "loss": 0.0414, "step": 12131 }, { "epoch": 0.6467985285493416, "grad_norm": 2.322753429412842, "learning_rate": 7.573800000000001e-06, "loss": 0.0748, "step": 12132 }, { "epoch": 0.6468518419789945, "grad_norm": 3.0609028339385986, "learning_rate": 7.573600000000001e-06, "loss": 0.0868, "step": 12133 }, { "epoch": 0.6469051554086475, "grad_norm": 1.9053194522857666, "learning_rate": 7.573400000000001e-06, "loss": 0.0144, "step": 12134 }, { "epoch": 0.6469584688383003, "grad_norm": 2.7738118171691895, "learning_rate": 7.573200000000001e-06, "loss": 0.0365, "step": 12135 }, { "epoch": 0.6470117822679533, "grad_norm": 2.197260856628418, "learning_rate": 7.573e-06, "loss": 0.0379, "step": 12136 }, { "epoch": 0.6470650956976062, "grad_norm": 2.6614797115325928, "learning_rate": 7.5728e-06, "loss": 0.0338, "step": 12137 }, { "epoch": 0.6471184091272592, "grad_norm": 3.448942184448242, "learning_rate": 7.572600000000001e-06, "loss": 0.045, "step": 12138 }, { "epoch": 0.6471717225569121, "grad_norm": 5.828375816345215, "learning_rate": 7.5724000000000005e-06, "loss": 0.0322, "step": 12139 }, { "epoch": 0.647225035986565, "grad_norm": 1.222011685371399, "learning_rate": 7.572200000000001e-06, "loss": 0.0238, "step": 12140 }, { "epoch": 0.647278349416218, "grad_norm": 3.7888381481170654, "learning_rate": 7.5720000000000005e-06, "loss": 0.0476, "step": 12141 }, { "epoch": 0.6473316628458708, "grad_norm": 1.3584098815917969, "learning_rate": 7.5718e-06, "loss": 0.0149, "step": 12142 }, { "epoch": 0.6473849762755238, "grad_norm": 1.879136323928833, "learning_rate": 7.571600000000001e-06, "loss": 0.0329, "step": 12143 }, { "epoch": 0.6474382897051767, "grad_norm": 2.0341382026672363, "learning_rate": 7.571400000000001e-06, "loss": 0.0184, "step": 12144 }, { "epoch": 0.6474916031348297, "grad_norm": 0.8387947082519531, "learning_rate": 7.5712000000000005e-06, "loss": 0.0055, "step": 12145 }, { "epoch": 0.6475449165644825, "grad_norm": 3.6687326431274414, "learning_rate": 7.571000000000001e-06, "loss": 0.029, "step": 12146 }, { "epoch": 0.6475982299941355, "grad_norm": 2.311319589614868, "learning_rate": 7.5708000000000004e-06, "loss": 0.0387, "step": 12147 }, { "epoch": 0.6476515434237885, "grad_norm": 3.1527631282806396, "learning_rate": 7.5706e-06, "loss": 0.0498, "step": 12148 }, { "epoch": 0.6477048568534414, "grad_norm": 1.6957037448883057, "learning_rate": 7.570400000000001e-06, "loss": 0.042, "step": 12149 }, { "epoch": 0.6477581702830943, "grad_norm": 1.8127915859222412, "learning_rate": 7.570200000000001e-06, "loss": 0.0236, "step": 12150 }, { "epoch": 0.6478114837127472, "grad_norm": 2.9816908836364746, "learning_rate": 7.57e-06, "loss": 0.0289, "step": 12151 }, { "epoch": 0.6478647971424002, "grad_norm": 2.042534828186035, "learning_rate": 7.569800000000001e-06, "loss": 0.0346, "step": 12152 }, { "epoch": 0.6479181105720531, "grad_norm": 2.82262921333313, "learning_rate": 7.5696e-06, "loss": 0.0428, "step": 12153 }, { "epoch": 0.647971424001706, "grad_norm": 3.7470128536224365, "learning_rate": 7.5694e-06, "loss": 0.0497, "step": 12154 }, { "epoch": 0.6480247374313589, "grad_norm": 3.426060914993286, "learning_rate": 7.569200000000001e-06, "loss": 0.0433, "step": 12155 }, { "epoch": 0.6480780508610119, "grad_norm": 2.998220205307007, "learning_rate": 7.569000000000001e-06, "loss": 0.0401, "step": 12156 }, { "epoch": 0.6481313642906649, "grad_norm": 3.487541675567627, "learning_rate": 7.5688e-06, "loss": 0.0157, "step": 12157 }, { "epoch": 0.6481846777203177, "grad_norm": 3.430180311203003, "learning_rate": 7.568600000000001e-06, "loss": 0.0282, "step": 12158 }, { "epoch": 0.6482379911499707, "grad_norm": 2.8657710552215576, "learning_rate": 7.5684e-06, "loss": 0.032, "step": 12159 }, { "epoch": 0.6482913045796236, "grad_norm": 3.0654070377349854, "learning_rate": 7.5682000000000015e-06, "loss": 0.0325, "step": 12160 }, { "epoch": 0.6483446180092766, "grad_norm": 2.972647190093994, "learning_rate": 7.568000000000001e-06, "loss": 0.0211, "step": 12161 }, { "epoch": 0.6483979314389294, "grad_norm": 2.4089109897613525, "learning_rate": 7.567800000000001e-06, "loss": 0.0314, "step": 12162 }, { "epoch": 0.6484512448685824, "grad_norm": 8.586767196655273, "learning_rate": 7.5676e-06, "loss": 0.0742, "step": 12163 }, { "epoch": 0.6485045582982353, "grad_norm": 2.0801262855529785, "learning_rate": 7.567400000000001e-06, "loss": 0.0611, "step": 12164 }, { "epoch": 0.6485578717278883, "grad_norm": 1.7439253330230713, "learning_rate": 7.5672e-06, "loss": 0.0188, "step": 12165 }, { "epoch": 0.6486111851575412, "grad_norm": 1.767079472541809, "learning_rate": 7.567000000000001e-06, "loss": 0.029, "step": 12166 }, { "epoch": 0.6486644985871941, "grad_norm": 1.020101547241211, "learning_rate": 7.566800000000001e-06, "loss": 0.0075, "step": 12167 }, { "epoch": 0.6487178120168471, "grad_norm": 3.004195213317871, "learning_rate": 7.5666000000000005e-06, "loss": 0.0422, "step": 12168 }, { "epoch": 0.6487711254465, "grad_norm": 2.126558542251587, "learning_rate": 7.5664e-06, "loss": 0.0418, "step": 12169 }, { "epoch": 0.6488244388761529, "grad_norm": 2.2148358821868896, "learning_rate": 7.5662000000000005e-06, "loss": 0.0297, "step": 12170 }, { "epoch": 0.6488777523058058, "grad_norm": 4.1791911125183105, "learning_rate": 7.566000000000001e-06, "loss": 0.0487, "step": 12171 }, { "epoch": 0.6489310657354588, "grad_norm": 1.9778810739517212, "learning_rate": 7.565800000000001e-06, "loss": 0.0433, "step": 12172 }, { "epoch": 0.6489843791651116, "grad_norm": 5.371538162231445, "learning_rate": 7.565600000000001e-06, "loss": 0.0372, "step": 12173 }, { "epoch": 0.6490376925947646, "grad_norm": 2.236361265182495, "learning_rate": 7.5654000000000004e-06, "loss": 0.0502, "step": 12174 }, { "epoch": 0.6490910060244176, "grad_norm": 3.241828441619873, "learning_rate": 7.5652e-06, "loss": 0.0259, "step": 12175 }, { "epoch": 0.6491443194540705, "grad_norm": 2.3010945320129395, "learning_rate": 7.565e-06, "loss": 0.0334, "step": 12176 }, { "epoch": 0.6491976328837235, "grad_norm": 1.3487151861190796, "learning_rate": 7.564800000000001e-06, "loss": 0.017, "step": 12177 }, { "epoch": 0.6492509463133763, "grad_norm": 2.2621102333068848, "learning_rate": 7.564600000000001e-06, "loss": 0.0351, "step": 12178 }, { "epoch": 0.6493042597430293, "grad_norm": 2.715776205062866, "learning_rate": 7.564400000000001e-06, "loss": 0.0514, "step": 12179 }, { "epoch": 0.6493575731726822, "grad_norm": 2.2614805698394775, "learning_rate": 7.5642e-06, "loss": 0.0351, "step": 12180 }, { "epoch": 0.6494108866023351, "grad_norm": 2.6254749298095703, "learning_rate": 7.564e-06, "loss": 0.0696, "step": 12181 }, { "epoch": 0.649464200031988, "grad_norm": 2.0740251541137695, "learning_rate": 7.563800000000001e-06, "loss": 0.0381, "step": 12182 }, { "epoch": 0.649517513461641, "grad_norm": 2.408918857574463, "learning_rate": 7.563600000000001e-06, "loss": 0.0624, "step": 12183 }, { "epoch": 0.6495708268912939, "grad_norm": 2.92148494720459, "learning_rate": 7.563400000000001e-06, "loss": 0.0197, "step": 12184 }, { "epoch": 0.6496241403209468, "grad_norm": 4.989606857299805, "learning_rate": 7.563200000000001e-06, "loss": 0.0503, "step": 12185 }, { "epoch": 0.6496774537505998, "grad_norm": 2.7438130378723145, "learning_rate": 7.563e-06, "loss": 0.0405, "step": 12186 }, { "epoch": 0.6497307671802527, "grad_norm": 1.386389136314392, "learning_rate": 7.5628e-06, "loss": 0.0201, "step": 12187 }, { "epoch": 0.6497840806099057, "grad_norm": 2.735206365585327, "learning_rate": 7.562600000000001e-06, "loss": 0.0378, "step": 12188 }, { "epoch": 0.6498373940395585, "grad_norm": 2.044940233230591, "learning_rate": 7.562400000000001e-06, "loss": 0.0316, "step": 12189 }, { "epoch": 0.6498907074692115, "grad_norm": 1.090124249458313, "learning_rate": 7.562200000000001e-06, "loss": 0.0119, "step": 12190 }, { "epoch": 0.6499440208988644, "grad_norm": 3.4146389961242676, "learning_rate": 7.562000000000001e-06, "loss": 0.0634, "step": 12191 }, { "epoch": 0.6499973343285174, "grad_norm": 2.494805097579956, "learning_rate": 7.5618e-06, "loss": 0.0536, "step": 12192 }, { "epoch": 0.6500506477581702, "grad_norm": 1.8055332899093628, "learning_rate": 7.5616000000000014e-06, "loss": 0.0273, "step": 12193 }, { "epoch": 0.6501039611878232, "grad_norm": 3.1915552616119385, "learning_rate": 7.561400000000001e-06, "loss": 0.0475, "step": 12194 }, { "epoch": 0.6501572746174762, "grad_norm": 3.06219220161438, "learning_rate": 7.5612000000000006e-06, "loss": 0.0515, "step": 12195 }, { "epoch": 0.6502105880471291, "grad_norm": 2.5038983821868896, "learning_rate": 7.561000000000001e-06, "loss": 0.0232, "step": 12196 }, { "epoch": 0.650263901476782, "grad_norm": 2.2679781913757324, "learning_rate": 7.5608000000000005e-06, "loss": 0.0484, "step": 12197 }, { "epoch": 0.6503172149064349, "grad_norm": 1.6770169734954834, "learning_rate": 7.5606e-06, "loss": 0.0211, "step": 12198 }, { "epoch": 0.6503705283360879, "grad_norm": 3.7307143211364746, "learning_rate": 7.560400000000001e-06, "loss": 0.0294, "step": 12199 }, { "epoch": 0.6504238417657408, "grad_norm": 3.252034902572632, "learning_rate": 7.560200000000001e-06, "loss": 0.0301, "step": 12200 }, { "epoch": 0.6504771551953937, "grad_norm": 1.7630703449249268, "learning_rate": 7.5600000000000005e-06, "loss": 0.0211, "step": 12201 }, { "epoch": 0.6505304686250466, "grad_norm": 1.7397366762161255, "learning_rate": 7.559800000000001e-06, "loss": 0.0125, "step": 12202 }, { "epoch": 0.6505837820546996, "grad_norm": 2.0546905994415283, "learning_rate": 7.5596000000000004e-06, "loss": 0.0216, "step": 12203 }, { "epoch": 0.6506370954843526, "grad_norm": 3.5772881507873535, "learning_rate": 7.5594e-06, "loss": 0.0506, "step": 12204 }, { "epoch": 0.6506904089140054, "grad_norm": 5.06021785736084, "learning_rate": 7.559200000000001e-06, "loss": 0.0531, "step": 12205 }, { "epoch": 0.6507437223436584, "grad_norm": 1.798447847366333, "learning_rate": 7.559000000000001e-06, "loss": 0.0127, "step": 12206 }, { "epoch": 0.6507970357733113, "grad_norm": 3.3469436168670654, "learning_rate": 7.5588e-06, "loss": 0.0865, "step": 12207 }, { "epoch": 0.6508503492029643, "grad_norm": 2.421121835708618, "learning_rate": 7.558600000000001e-06, "loss": 0.0168, "step": 12208 }, { "epoch": 0.6509036626326171, "grad_norm": 1.3614777326583862, "learning_rate": 7.5584e-06, "loss": 0.0313, "step": 12209 }, { "epoch": 0.6509569760622701, "grad_norm": 2.661672353744507, "learning_rate": 7.558200000000001e-06, "loss": 0.0549, "step": 12210 }, { "epoch": 0.651010289491923, "grad_norm": 1.8951619863510132, "learning_rate": 7.558000000000001e-06, "loss": 0.0175, "step": 12211 }, { "epoch": 0.651063602921576, "grad_norm": 2.455842971801758, "learning_rate": 7.557800000000001e-06, "loss": 0.0233, "step": 12212 }, { "epoch": 0.6511169163512289, "grad_norm": 1.5013501644134521, "learning_rate": 7.5576e-06, "loss": 0.0199, "step": 12213 }, { "epoch": 0.6511702297808818, "grad_norm": 2.211637258529663, "learning_rate": 7.557400000000001e-06, "loss": 0.053, "step": 12214 }, { "epoch": 0.6512235432105348, "grad_norm": 2.1463799476623535, "learning_rate": 7.5572e-06, "loss": 0.0475, "step": 12215 }, { "epoch": 0.6512768566401876, "grad_norm": 1.7936551570892334, "learning_rate": 7.557000000000001e-06, "loss": 0.0318, "step": 12216 }, { "epoch": 0.6513301700698406, "grad_norm": 1.4424759149551392, "learning_rate": 7.556800000000001e-06, "loss": 0.0393, "step": 12217 }, { "epoch": 0.6513834834994935, "grad_norm": 1.9459742307662964, "learning_rate": 7.556600000000001e-06, "loss": 0.0158, "step": 12218 }, { "epoch": 0.6514367969291465, "grad_norm": 1.6712192296981812, "learning_rate": 7.5564e-06, "loss": 0.0136, "step": 12219 }, { "epoch": 0.6514901103587993, "grad_norm": 1.846082091331482, "learning_rate": 7.556200000000001e-06, "loss": 0.0222, "step": 12220 }, { "epoch": 0.6515434237884523, "grad_norm": 1.9429898262023926, "learning_rate": 7.556000000000001e-06, "loss": 0.0392, "step": 12221 }, { "epoch": 0.6515967372181053, "grad_norm": 2.269338369369507, "learning_rate": 7.5558000000000006e-06, "loss": 0.0311, "step": 12222 }, { "epoch": 0.6516500506477582, "grad_norm": 3.4727163314819336, "learning_rate": 7.555600000000001e-06, "loss": 0.0513, "step": 12223 }, { "epoch": 0.6517033640774111, "grad_norm": 4.584686279296875, "learning_rate": 7.5554000000000005e-06, "loss": 0.0529, "step": 12224 }, { "epoch": 0.651756677507064, "grad_norm": 2.4885613918304443, "learning_rate": 7.5552e-06, "loss": 0.0319, "step": 12225 }, { "epoch": 0.651809990936717, "grad_norm": 1.8426423072814941, "learning_rate": 7.5550000000000005e-06, "loss": 0.0416, "step": 12226 }, { "epoch": 0.6518633043663699, "grad_norm": 1.428707242012024, "learning_rate": 7.554800000000001e-06, "loss": 0.0104, "step": 12227 }, { "epoch": 0.6519166177960228, "grad_norm": 1.1139527559280396, "learning_rate": 7.5546000000000005e-06, "loss": 0.0233, "step": 12228 }, { "epoch": 0.6519699312256757, "grad_norm": 2.143460273742676, "learning_rate": 7.554400000000001e-06, "loss": 0.0351, "step": 12229 }, { "epoch": 0.6520232446553287, "grad_norm": 1.0108444690704346, "learning_rate": 7.5542000000000005e-06, "loss": 0.0162, "step": 12230 }, { "epoch": 0.6520765580849817, "grad_norm": 2.0877625942230225, "learning_rate": 7.554e-06, "loss": 0.0288, "step": 12231 }, { "epoch": 0.6521298715146345, "grad_norm": 1.809683918952942, "learning_rate": 7.553800000000001e-06, "loss": 0.0328, "step": 12232 }, { "epoch": 0.6521831849442875, "grad_norm": 1.4056017398834229, "learning_rate": 7.553600000000001e-06, "loss": 0.0315, "step": 12233 }, { "epoch": 0.6522364983739404, "grad_norm": 1.6497244834899902, "learning_rate": 7.5534e-06, "loss": 0.0224, "step": 12234 }, { "epoch": 0.6522898118035934, "grad_norm": 4.145325660705566, "learning_rate": 7.553200000000001e-06, "loss": 0.0811, "step": 12235 }, { "epoch": 0.6523431252332462, "grad_norm": 0.3186599910259247, "learning_rate": 7.553e-06, "loss": 0.0026, "step": 12236 }, { "epoch": 0.6523964386628992, "grad_norm": 2.152911901473999, "learning_rate": 7.5528e-06, "loss": 0.04, "step": 12237 }, { "epoch": 0.6524497520925521, "grad_norm": 1.3936926126480103, "learning_rate": 7.552600000000001e-06, "loss": 0.0131, "step": 12238 }, { "epoch": 0.652503065522205, "grad_norm": 2.526437997817993, "learning_rate": 7.552400000000001e-06, "loss": 0.0181, "step": 12239 }, { "epoch": 0.6525563789518579, "grad_norm": 3.8523759841918945, "learning_rate": 7.5522e-06, "loss": 0.079, "step": 12240 }, { "epoch": 0.6526096923815109, "grad_norm": 3.1116557121276855, "learning_rate": 7.552000000000001e-06, "loss": 0.037, "step": 12241 }, { "epoch": 0.6526630058111639, "grad_norm": 1.7027393579483032, "learning_rate": 7.5518e-06, "loss": 0.0283, "step": 12242 }, { "epoch": 0.6527163192408167, "grad_norm": 3.2104551792144775, "learning_rate": 7.5516000000000015e-06, "loss": 0.0401, "step": 12243 }, { "epoch": 0.6527696326704697, "grad_norm": 2.1040656566619873, "learning_rate": 7.551400000000001e-06, "loss": 0.0316, "step": 12244 }, { "epoch": 0.6528229461001226, "grad_norm": 1.913169503211975, "learning_rate": 7.551200000000001e-06, "loss": 0.0379, "step": 12245 }, { "epoch": 0.6528762595297756, "grad_norm": 1.4520304203033447, "learning_rate": 7.551e-06, "loss": 0.037, "step": 12246 }, { "epoch": 0.6529295729594284, "grad_norm": 1.156325101852417, "learning_rate": 7.550800000000001e-06, "loss": 0.0268, "step": 12247 }, { "epoch": 0.6529828863890814, "grad_norm": 1.9168990850448608, "learning_rate": 7.5506e-06, "loss": 0.0234, "step": 12248 }, { "epoch": 0.6530361998187343, "grad_norm": 2.238326072692871, "learning_rate": 7.5504000000000014e-06, "loss": 0.0448, "step": 12249 }, { "epoch": 0.6530895132483873, "grad_norm": 2.1526899337768555, "learning_rate": 7.550200000000001e-06, "loss": 0.0562, "step": 12250 }, { "epoch": 0.6531428266780402, "grad_norm": 8.09070873260498, "learning_rate": 7.5500000000000006e-06, "loss": 0.0588, "step": 12251 }, { "epoch": 0.6531961401076931, "grad_norm": 3.0519285202026367, "learning_rate": 7.5498e-06, "loss": 0.0748, "step": 12252 }, { "epoch": 0.6532494535373461, "grad_norm": 2.0135438442230225, "learning_rate": 7.5496000000000005e-06, "loss": 0.0313, "step": 12253 }, { "epoch": 0.653302766966999, "grad_norm": 1.5887792110443115, "learning_rate": 7.549400000000001e-06, "loss": 0.0266, "step": 12254 }, { "epoch": 0.6533560803966519, "grad_norm": 1.8661874532699585, "learning_rate": 7.549200000000001e-06, "loss": 0.0249, "step": 12255 }, { "epoch": 0.6534093938263048, "grad_norm": 2.2171549797058105, "learning_rate": 7.549000000000001e-06, "loss": 0.0474, "step": 12256 }, { "epoch": 0.6534627072559578, "grad_norm": 3.336008310317993, "learning_rate": 7.5488000000000005e-06, "loss": 0.0367, "step": 12257 }, { "epoch": 0.6535160206856107, "grad_norm": 6.094459056854248, "learning_rate": 7.5486e-06, "loss": 0.0669, "step": 12258 }, { "epoch": 0.6535693341152636, "grad_norm": 4.638010025024414, "learning_rate": 7.5484000000000004e-06, "loss": 0.0486, "step": 12259 }, { "epoch": 0.6536226475449166, "grad_norm": 1.4792044162750244, "learning_rate": 7.548200000000001e-06, "loss": 0.0198, "step": 12260 }, { "epoch": 0.6536759609745695, "grad_norm": 5.898393630981445, "learning_rate": 7.548000000000001e-06, "loss": 0.0391, "step": 12261 }, { "epoch": 0.6537292744042225, "grad_norm": 1.786806344985962, "learning_rate": 7.547800000000001e-06, "loss": 0.0317, "step": 12262 }, { "epoch": 0.6537825878338753, "grad_norm": 1.0054373741149902, "learning_rate": 7.5476e-06, "loss": 0.0168, "step": 12263 }, { "epoch": 0.6538359012635283, "grad_norm": 3.497811794281006, "learning_rate": 7.5474e-06, "loss": 0.0521, "step": 12264 }, { "epoch": 0.6538892146931812, "grad_norm": 1.6900057792663574, "learning_rate": 7.5472e-06, "loss": 0.0276, "step": 12265 }, { "epoch": 0.6539425281228342, "grad_norm": 2.9018454551696777, "learning_rate": 7.547000000000001e-06, "loss": 0.0614, "step": 12266 }, { "epoch": 0.653995841552487, "grad_norm": 3.9990994930267334, "learning_rate": 7.546800000000001e-06, "loss": 0.0363, "step": 12267 }, { "epoch": 0.65404915498214, "grad_norm": 1.945647954940796, "learning_rate": 7.546600000000001e-06, "loss": 0.0122, "step": 12268 }, { "epoch": 0.654102468411793, "grad_norm": 2.5159950256347656, "learning_rate": 7.5464e-06, "loss": 0.0533, "step": 12269 }, { "epoch": 0.6541557818414458, "grad_norm": 1.8837965726852417, "learning_rate": 7.5462e-06, "loss": 0.0488, "step": 12270 }, { "epoch": 0.6542090952710988, "grad_norm": 2.3085241317749023, "learning_rate": 7.546000000000001e-06, "loss": 0.0332, "step": 12271 }, { "epoch": 0.6542624087007517, "grad_norm": 1.1854422092437744, "learning_rate": 7.545800000000001e-06, "loss": 0.0177, "step": 12272 }, { "epoch": 0.6543157221304047, "grad_norm": 1.980509638786316, "learning_rate": 7.545600000000001e-06, "loss": 0.0405, "step": 12273 }, { "epoch": 0.6543690355600575, "grad_norm": 5.813764572143555, "learning_rate": 7.545400000000001e-06, "loss": 0.0565, "step": 12274 }, { "epoch": 0.6544223489897105, "grad_norm": 2.5564169883728027, "learning_rate": 7.5452e-06, "loss": 0.0283, "step": 12275 }, { "epoch": 0.6544756624193634, "grad_norm": 2.020303726196289, "learning_rate": 7.545e-06, "loss": 0.0287, "step": 12276 }, { "epoch": 0.6545289758490164, "grad_norm": 2.5827572345733643, "learning_rate": 7.544800000000001e-06, "loss": 0.0149, "step": 12277 }, { "epoch": 0.6545822892786693, "grad_norm": 3.067247152328491, "learning_rate": 7.5446000000000006e-06, "loss": 0.0646, "step": 12278 }, { "epoch": 0.6546356027083222, "grad_norm": 3.713444948196411, "learning_rate": 7.544400000000001e-06, "loss": 0.056, "step": 12279 }, { "epoch": 0.6546889161379752, "grad_norm": 2.0104994773864746, "learning_rate": 7.5442000000000005e-06, "loss": 0.0282, "step": 12280 }, { "epoch": 0.6547422295676281, "grad_norm": 0.9649965167045593, "learning_rate": 7.544e-06, "loss": 0.015, "step": 12281 }, { "epoch": 0.654795542997281, "grad_norm": 1.9017695188522339, "learning_rate": 7.543800000000001e-06, "loss": 0.0539, "step": 12282 }, { "epoch": 0.6548488564269339, "grad_norm": 5.265008926391602, "learning_rate": 7.543600000000001e-06, "loss": 0.0551, "step": 12283 }, { "epoch": 0.6549021698565869, "grad_norm": 2.065854072570801, "learning_rate": 7.5434000000000005e-06, "loss": 0.0269, "step": 12284 }, { "epoch": 0.6549554832862398, "grad_norm": 2.310227155685425, "learning_rate": 7.543200000000001e-06, "loss": 0.0368, "step": 12285 }, { "epoch": 0.6550087967158927, "grad_norm": 1.6777597665786743, "learning_rate": 7.5430000000000005e-06, "loss": 0.0257, "step": 12286 }, { "epoch": 0.6550621101455457, "grad_norm": 2.135183334350586, "learning_rate": 7.5428e-06, "loss": 0.0352, "step": 12287 }, { "epoch": 0.6551154235751986, "grad_norm": 1.6250230073928833, "learning_rate": 7.542600000000001e-06, "loss": 0.0251, "step": 12288 }, { "epoch": 0.6551687370048516, "grad_norm": 2.171973943710327, "learning_rate": 7.542400000000001e-06, "loss": 0.0646, "step": 12289 }, { "epoch": 0.6552220504345044, "grad_norm": 2.7180309295654297, "learning_rate": 7.5422e-06, "loss": 0.0363, "step": 12290 }, { "epoch": 0.6552753638641574, "grad_norm": 1.6021219491958618, "learning_rate": 7.542000000000001e-06, "loss": 0.0213, "step": 12291 }, { "epoch": 0.6553286772938103, "grad_norm": 1.47651207447052, "learning_rate": 7.5418e-06, "loss": 0.0237, "step": 12292 }, { "epoch": 0.6553819907234633, "grad_norm": 1.4822828769683838, "learning_rate": 7.541600000000001e-06, "loss": 0.02, "step": 12293 }, { "epoch": 0.6554353041531161, "grad_norm": 1.9971001148223877, "learning_rate": 7.541400000000001e-06, "loss": 0.0172, "step": 12294 }, { "epoch": 0.6554886175827691, "grad_norm": 1.6998696327209473, "learning_rate": 7.541200000000001e-06, "loss": 0.0188, "step": 12295 }, { "epoch": 0.655541931012422, "grad_norm": 3.508162260055542, "learning_rate": 7.541e-06, "loss": 0.0305, "step": 12296 }, { "epoch": 0.655595244442075, "grad_norm": 1.6428261995315552, "learning_rate": 7.540800000000001e-06, "loss": 0.0206, "step": 12297 }, { "epoch": 0.6556485578717279, "grad_norm": 4.635879039764404, "learning_rate": 7.5406e-06, "loss": 0.0547, "step": 12298 }, { "epoch": 0.6557018713013808, "grad_norm": 2.863668918609619, "learning_rate": 7.540400000000001e-06, "loss": 0.0475, "step": 12299 }, { "epoch": 0.6557551847310338, "grad_norm": 1.7432807683944702, "learning_rate": 7.540200000000001e-06, "loss": 0.0201, "step": 12300 }, { "epoch": 0.6558084981606866, "grad_norm": 2.0049002170562744, "learning_rate": 7.540000000000001e-06, "loss": 0.0363, "step": 12301 }, { "epoch": 0.6558618115903396, "grad_norm": 1.6634960174560547, "learning_rate": 7.5398e-06, "loss": 0.0187, "step": 12302 }, { "epoch": 0.6559151250199925, "grad_norm": 2.0274438858032227, "learning_rate": 7.539600000000001e-06, "loss": 0.0252, "step": 12303 }, { "epoch": 0.6559684384496455, "grad_norm": 0.846560001373291, "learning_rate": 7.539400000000001e-06, "loss": 0.008, "step": 12304 }, { "epoch": 0.6560217518792983, "grad_norm": 2.3950743675231934, "learning_rate": 7.539200000000001e-06, "loss": 0.0512, "step": 12305 }, { "epoch": 0.6560750653089513, "grad_norm": 0.7469476461410522, "learning_rate": 7.539000000000001e-06, "loss": 0.0074, "step": 12306 }, { "epoch": 0.6561283787386043, "grad_norm": 3.266497850418091, "learning_rate": 7.5388000000000006e-06, "loss": 0.0383, "step": 12307 }, { "epoch": 0.6561816921682572, "grad_norm": 2.069727659225464, "learning_rate": 7.5386e-06, "loss": 0.0255, "step": 12308 }, { "epoch": 0.6562350055979101, "grad_norm": 3.407914876937866, "learning_rate": 7.5384000000000005e-06, "loss": 0.0414, "step": 12309 }, { "epoch": 0.656288319027563, "grad_norm": 4.1463422775268555, "learning_rate": 7.538200000000001e-06, "loss": 0.044, "step": 12310 }, { "epoch": 0.656341632457216, "grad_norm": 2.6853444576263428, "learning_rate": 7.5380000000000005e-06, "loss": 0.0514, "step": 12311 }, { "epoch": 0.6563949458868689, "grad_norm": 3.035378932952881, "learning_rate": 7.537800000000001e-06, "loss": 0.021, "step": 12312 }, { "epoch": 0.6564482593165218, "grad_norm": 2.061732769012451, "learning_rate": 7.5376000000000005e-06, "loss": 0.0455, "step": 12313 }, { "epoch": 0.6565015727461747, "grad_norm": 1.4196934700012207, "learning_rate": 7.5374e-06, "loss": 0.0205, "step": 12314 }, { "epoch": 0.6565548861758277, "grad_norm": 4.35624361038208, "learning_rate": 7.5372000000000004e-06, "loss": 0.0447, "step": 12315 }, { "epoch": 0.6566081996054807, "grad_norm": 3.12198543548584, "learning_rate": 7.537000000000001e-06, "loss": 0.0264, "step": 12316 }, { "epoch": 0.6566615130351335, "grad_norm": 3.5771610736846924, "learning_rate": 7.5368e-06, "loss": 0.0346, "step": 12317 }, { "epoch": 0.6567148264647865, "grad_norm": 3.780540943145752, "learning_rate": 7.536600000000001e-06, "loss": 0.0377, "step": 12318 }, { "epoch": 0.6567681398944394, "grad_norm": 7.270931720733643, "learning_rate": 7.5364e-06, "loss": 0.0444, "step": 12319 }, { "epoch": 0.6568214533240924, "grad_norm": 1.9816657304763794, "learning_rate": 7.5362e-06, "loss": 0.0351, "step": 12320 }, { "epoch": 0.6568747667537452, "grad_norm": 1.368264079093933, "learning_rate": 7.536000000000001e-06, "loss": 0.0112, "step": 12321 }, { "epoch": 0.6569280801833982, "grad_norm": 2.329632520675659, "learning_rate": 7.535800000000001e-06, "loss": 0.0204, "step": 12322 }, { "epoch": 0.6569813936130511, "grad_norm": 2.0803511142730713, "learning_rate": 7.5356e-06, "loss": 0.034, "step": 12323 }, { "epoch": 0.6570347070427041, "grad_norm": 1.3654677867889404, "learning_rate": 7.535400000000001e-06, "loss": 0.0102, "step": 12324 }, { "epoch": 0.657088020472357, "grad_norm": 2.0755672454833984, "learning_rate": 7.5352e-06, "loss": 0.0304, "step": 12325 }, { "epoch": 0.6571413339020099, "grad_norm": 4.876569747924805, "learning_rate": 7.535e-06, "loss": 0.0455, "step": 12326 }, { "epoch": 0.6571946473316629, "grad_norm": 4.119546890258789, "learning_rate": 7.534800000000001e-06, "loss": 0.089, "step": 12327 }, { "epoch": 0.6572479607613158, "grad_norm": 2.858405828475952, "learning_rate": 7.534600000000001e-06, "loss": 0.0408, "step": 12328 }, { "epoch": 0.6573012741909687, "grad_norm": 2.3791465759277344, "learning_rate": 7.5344e-06, "loss": 0.0423, "step": 12329 }, { "epoch": 0.6573545876206216, "grad_norm": 3.2824289798736572, "learning_rate": 7.534200000000001e-06, "loss": 0.0384, "step": 12330 }, { "epoch": 0.6574079010502746, "grad_norm": 2.8255488872528076, "learning_rate": 7.534e-06, "loss": 0.0227, "step": 12331 }, { "epoch": 0.6574612144799274, "grad_norm": 3.6707451343536377, "learning_rate": 7.5338000000000015e-06, "loss": 0.0371, "step": 12332 }, { "epoch": 0.6575145279095804, "grad_norm": 3.8930776119232178, "learning_rate": 7.533600000000001e-06, "loss": 0.044, "step": 12333 }, { "epoch": 0.6575678413392334, "grad_norm": 4.116536617279053, "learning_rate": 7.533400000000001e-06, "loss": 0.0516, "step": 12334 }, { "epoch": 0.6576211547688863, "grad_norm": 3.13257098197937, "learning_rate": 7.5332e-06, "loss": 0.0405, "step": 12335 }, { "epoch": 0.6576744681985393, "grad_norm": 1.6161456108093262, "learning_rate": 7.5330000000000006e-06, "loss": 0.0314, "step": 12336 }, { "epoch": 0.6577277816281921, "grad_norm": 1.0610754489898682, "learning_rate": 7.5328e-06, "loss": 0.0111, "step": 12337 }, { "epoch": 0.6577810950578451, "grad_norm": 1.9441890716552734, "learning_rate": 7.532600000000001e-06, "loss": 0.0329, "step": 12338 }, { "epoch": 0.657834408487498, "grad_norm": 2.166381359100342, "learning_rate": 7.532400000000001e-06, "loss": 0.0341, "step": 12339 }, { "epoch": 0.657887721917151, "grad_norm": 1.4361293315887451, "learning_rate": 7.5322000000000005e-06, "loss": 0.012, "step": 12340 }, { "epoch": 0.6579410353468038, "grad_norm": 5.617468357086182, "learning_rate": 7.532e-06, "loss": 0.0337, "step": 12341 }, { "epoch": 0.6579943487764568, "grad_norm": 2.158137559890747, "learning_rate": 7.5318000000000005e-06, "loss": 0.0461, "step": 12342 }, { "epoch": 0.6580476622061098, "grad_norm": 1.9345439672470093, "learning_rate": 7.531600000000001e-06, "loss": 0.0177, "step": 12343 }, { "epoch": 0.6581009756357626, "grad_norm": 1.5679839849472046, "learning_rate": 7.531400000000001e-06, "loss": 0.0258, "step": 12344 }, { "epoch": 0.6581542890654156, "grad_norm": 1.5408978462219238, "learning_rate": 7.531200000000001e-06, "loss": 0.0256, "step": 12345 }, { "epoch": 0.6582076024950685, "grad_norm": 3.0237386226654053, "learning_rate": 7.531e-06, "loss": 0.035, "step": 12346 }, { "epoch": 0.6582609159247215, "grad_norm": 3.602625846862793, "learning_rate": 7.5308e-06, "loss": 0.0903, "step": 12347 }, { "epoch": 0.6583142293543743, "grad_norm": 2.6304051876068115, "learning_rate": 7.5306e-06, "loss": 0.0444, "step": 12348 }, { "epoch": 0.6583675427840273, "grad_norm": 2.470900535583496, "learning_rate": 7.530400000000001e-06, "loss": 0.0183, "step": 12349 }, { "epoch": 0.6584208562136802, "grad_norm": 2.2154836654663086, "learning_rate": 7.530200000000001e-06, "loss": 0.0301, "step": 12350 }, { "epoch": 0.6584741696433332, "grad_norm": 1.3687818050384521, "learning_rate": 7.530000000000001e-06, "loss": 0.0241, "step": 12351 }, { "epoch": 0.658527483072986, "grad_norm": 1.6202956438064575, "learning_rate": 7.5298e-06, "loss": 0.0194, "step": 12352 }, { "epoch": 0.658580796502639, "grad_norm": 2.4306325912475586, "learning_rate": 7.5296e-06, "loss": 0.0678, "step": 12353 }, { "epoch": 0.658634109932292, "grad_norm": 7.120530605316162, "learning_rate": 7.529400000000001e-06, "loss": 0.0958, "step": 12354 }, { "epoch": 0.6586874233619449, "grad_norm": 5.141223430633545, "learning_rate": 7.529200000000001e-06, "loss": 0.046, "step": 12355 }, { "epoch": 0.6587407367915978, "grad_norm": 3.9980900287628174, "learning_rate": 7.529000000000001e-06, "loss": 0.0393, "step": 12356 }, { "epoch": 0.6587940502212507, "grad_norm": 3.341815233230591, "learning_rate": 7.528800000000001e-06, "loss": 0.0334, "step": 12357 }, { "epoch": 0.6588473636509037, "grad_norm": 3.963104009628296, "learning_rate": 7.5286e-06, "loss": 0.0478, "step": 12358 }, { "epoch": 0.6589006770805566, "grad_norm": 1.7813867330551147, "learning_rate": 7.5284e-06, "loss": 0.019, "step": 12359 }, { "epoch": 0.6589539905102095, "grad_norm": 2.335299253463745, "learning_rate": 7.528200000000001e-06, "loss": 0.0579, "step": 12360 }, { "epoch": 0.6590073039398624, "grad_norm": 3.4333839416503906, "learning_rate": 7.528000000000001e-06, "loss": 0.0657, "step": 12361 }, { "epoch": 0.6590606173695154, "grad_norm": 2.916121244430542, "learning_rate": 7.527800000000001e-06, "loss": 0.0257, "step": 12362 }, { "epoch": 0.6591139307991684, "grad_norm": 3.999957799911499, "learning_rate": 7.5276000000000006e-06, "loss": 0.0345, "step": 12363 }, { "epoch": 0.6591672442288212, "grad_norm": 2.7604024410247803, "learning_rate": 7.5274e-06, "loss": 0.0707, "step": 12364 }, { "epoch": 0.6592205576584742, "grad_norm": 2.401090383529663, "learning_rate": 7.5272e-06, "loss": 0.0383, "step": 12365 }, { "epoch": 0.6592738710881271, "grad_norm": 1.9463146924972534, "learning_rate": 7.527000000000001e-06, "loss": 0.0294, "step": 12366 }, { "epoch": 0.65932718451778, "grad_norm": 1.2504245042800903, "learning_rate": 7.5268000000000005e-06, "loss": 0.0157, "step": 12367 }, { "epoch": 0.6593804979474329, "grad_norm": 1.6907967329025269, "learning_rate": 7.526600000000001e-06, "loss": 0.0121, "step": 12368 }, { "epoch": 0.6594338113770859, "grad_norm": 1.652334451675415, "learning_rate": 7.5264000000000005e-06, "loss": 0.0341, "step": 12369 }, { "epoch": 0.6594871248067388, "grad_norm": 2.399242877960205, "learning_rate": 7.5262e-06, "loss": 0.0313, "step": 12370 }, { "epoch": 0.6595404382363917, "grad_norm": 1.6786425113677979, "learning_rate": 7.526000000000001e-06, "loss": 0.0372, "step": 12371 }, { "epoch": 0.6595937516660447, "grad_norm": 1.408004879951477, "learning_rate": 7.525800000000001e-06, "loss": 0.0406, "step": 12372 }, { "epoch": 0.6596470650956976, "grad_norm": 2.9176089763641357, "learning_rate": 7.5256e-06, "loss": 0.0712, "step": 12373 }, { "epoch": 0.6597003785253506, "grad_norm": 1.2694405317306519, "learning_rate": 7.525400000000001e-06, "loss": 0.0123, "step": 12374 }, { "epoch": 0.6597536919550034, "grad_norm": 2.418490171432495, "learning_rate": 7.5252e-06, "loss": 0.036, "step": 12375 }, { "epoch": 0.6598070053846564, "grad_norm": 3.3707773685455322, "learning_rate": 7.525e-06, "loss": 0.0688, "step": 12376 }, { "epoch": 0.6598603188143093, "grad_norm": 1.4218705892562866, "learning_rate": 7.524800000000001e-06, "loss": 0.0337, "step": 12377 }, { "epoch": 0.6599136322439623, "grad_norm": 1.0163562297821045, "learning_rate": 7.524600000000001e-06, "loss": 0.0225, "step": 12378 }, { "epoch": 0.6599669456736151, "grad_norm": 2.5090384483337402, "learning_rate": 7.5244e-06, "loss": 0.0452, "step": 12379 }, { "epoch": 0.6600202591032681, "grad_norm": 2.3411412239074707, "learning_rate": 7.524200000000001e-06, "loss": 0.0432, "step": 12380 }, { "epoch": 0.6600735725329211, "grad_norm": 1.2151262760162354, "learning_rate": 7.524e-06, "loss": 0.0143, "step": 12381 }, { "epoch": 0.660126885962574, "grad_norm": 2.896616220474243, "learning_rate": 7.523800000000001e-06, "loss": 0.0281, "step": 12382 }, { "epoch": 0.6601801993922269, "grad_norm": 1.681172490119934, "learning_rate": 7.523600000000001e-06, "loss": 0.0435, "step": 12383 }, { "epoch": 0.6602335128218798, "grad_norm": 4.851950645446777, "learning_rate": 7.523400000000001e-06, "loss": 0.0351, "step": 12384 }, { "epoch": 0.6602868262515328, "grad_norm": 1.6054354906082153, "learning_rate": 7.5232e-06, "loss": 0.0399, "step": 12385 }, { "epoch": 0.6603401396811857, "grad_norm": 1.4998557567596436, "learning_rate": 7.523000000000001e-06, "loss": 0.0154, "step": 12386 }, { "epoch": 0.6603934531108386, "grad_norm": 2.4709532260894775, "learning_rate": 7.5228e-06, "loss": 0.0364, "step": 12387 }, { "epoch": 0.6604467665404915, "grad_norm": 3.036813974380493, "learning_rate": 7.522600000000001e-06, "loss": 0.0777, "step": 12388 }, { "epoch": 0.6605000799701445, "grad_norm": 3.73875093460083, "learning_rate": 7.522400000000001e-06, "loss": 0.072, "step": 12389 }, { "epoch": 0.6605533933997975, "grad_norm": 3.0482900142669678, "learning_rate": 7.522200000000001e-06, "loss": 0.0274, "step": 12390 }, { "epoch": 0.6606067068294503, "grad_norm": 1.9521549940109253, "learning_rate": 7.522e-06, "loss": 0.031, "step": 12391 }, { "epoch": 0.6606600202591033, "grad_norm": 2.119457483291626, "learning_rate": 7.5218000000000006e-06, "loss": 0.0436, "step": 12392 }, { "epoch": 0.6607133336887562, "grad_norm": 1.1389235258102417, "learning_rate": 7.521600000000001e-06, "loss": 0.0224, "step": 12393 }, { "epoch": 0.6607666471184092, "grad_norm": 1.9572125673294067, "learning_rate": 7.5214000000000005e-06, "loss": 0.0331, "step": 12394 }, { "epoch": 0.660819960548062, "grad_norm": 1.546589732170105, "learning_rate": 7.521200000000001e-06, "loss": 0.0121, "step": 12395 }, { "epoch": 0.660873273977715, "grad_norm": 4.408631324768066, "learning_rate": 7.5210000000000005e-06, "loss": 0.0699, "step": 12396 }, { "epoch": 0.6609265874073679, "grad_norm": 2.385753631591797, "learning_rate": 7.5208e-06, "loss": 0.0402, "step": 12397 }, { "epoch": 0.6609799008370209, "grad_norm": 1.1450746059417725, "learning_rate": 7.5206000000000005e-06, "loss": 0.0141, "step": 12398 }, { "epoch": 0.6610332142666738, "grad_norm": 1.8315356969833374, "learning_rate": 7.520400000000001e-06, "loss": 0.0303, "step": 12399 }, { "epoch": 0.6610865276963267, "grad_norm": 1.5503166913986206, "learning_rate": 7.5202000000000004e-06, "loss": 0.0146, "step": 12400 }, { "epoch": 0.6611398411259797, "grad_norm": 2.234018087387085, "learning_rate": 7.520000000000001e-06, "loss": 0.029, "step": 12401 }, { "epoch": 0.6611931545556325, "grad_norm": 1.9396946430206299, "learning_rate": 7.5198e-06, "loss": 0.0375, "step": 12402 }, { "epoch": 0.6612464679852855, "grad_norm": 2.018090009689331, "learning_rate": 7.5196e-06, "loss": 0.0359, "step": 12403 }, { "epoch": 0.6612997814149384, "grad_norm": 2.0361440181732178, "learning_rate": 7.519400000000001e-06, "loss": 0.0335, "step": 12404 }, { "epoch": 0.6613530948445914, "grad_norm": 1.864303708076477, "learning_rate": 7.519200000000001e-06, "loss": 0.034, "step": 12405 }, { "epoch": 0.6614064082742442, "grad_norm": 0.9143535494804382, "learning_rate": 7.519e-06, "loss": 0.0185, "step": 12406 }, { "epoch": 0.6614597217038972, "grad_norm": 1.8425050973892212, "learning_rate": 7.518800000000001e-06, "loss": 0.0296, "step": 12407 }, { "epoch": 0.6615130351335501, "grad_norm": 1.5950483083724976, "learning_rate": 7.5186e-06, "loss": 0.0096, "step": 12408 }, { "epoch": 0.6615663485632031, "grad_norm": 2.3004822731018066, "learning_rate": 7.5184e-06, "loss": 0.023, "step": 12409 }, { "epoch": 0.661619661992856, "grad_norm": 2.6720051765441895, "learning_rate": 7.518200000000001e-06, "loss": 0.0364, "step": 12410 }, { "epoch": 0.6616729754225089, "grad_norm": 3.826036214828491, "learning_rate": 7.518000000000001e-06, "loss": 0.0596, "step": 12411 }, { "epoch": 0.6617262888521619, "grad_norm": 4.017069339752197, "learning_rate": 7.5178e-06, "loss": 0.0384, "step": 12412 }, { "epoch": 0.6617796022818148, "grad_norm": 2.5951459407806396, "learning_rate": 7.517600000000001e-06, "loss": 0.0312, "step": 12413 }, { "epoch": 0.6618329157114677, "grad_norm": 3.2692346572875977, "learning_rate": 7.5174e-06, "loss": 0.0441, "step": 12414 }, { "epoch": 0.6618862291411206, "grad_norm": 1.539059042930603, "learning_rate": 7.5172000000000015e-06, "loss": 0.0198, "step": 12415 }, { "epoch": 0.6619395425707736, "grad_norm": 1.9929392337799072, "learning_rate": 7.517000000000001e-06, "loss": 0.0362, "step": 12416 }, { "epoch": 0.6619928560004265, "grad_norm": 1.2566226720809937, "learning_rate": 7.516800000000001e-06, "loss": 0.0191, "step": 12417 }, { "epoch": 0.6620461694300794, "grad_norm": 3.141127109527588, "learning_rate": 7.5166e-06, "loss": 0.0593, "step": 12418 }, { "epoch": 0.6620994828597324, "grad_norm": 1.6753596067428589, "learning_rate": 7.5164000000000006e-06, "loss": 0.0327, "step": 12419 }, { "epoch": 0.6621527962893853, "grad_norm": 3.9407427310943604, "learning_rate": 7.5162e-06, "loss": 0.0484, "step": 12420 }, { "epoch": 0.6622061097190383, "grad_norm": 2.4489831924438477, "learning_rate": 7.516000000000001e-06, "loss": 0.0294, "step": 12421 }, { "epoch": 0.6622594231486911, "grad_norm": 0.9221163988113403, "learning_rate": 7.515800000000001e-06, "loss": 0.024, "step": 12422 }, { "epoch": 0.6623127365783441, "grad_norm": 3.824007034301758, "learning_rate": 7.5156000000000005e-06, "loss": 0.0342, "step": 12423 }, { "epoch": 0.662366050007997, "grad_norm": 1.0372918844223022, "learning_rate": 7.5154e-06, "loss": 0.0166, "step": 12424 }, { "epoch": 0.66241936343765, "grad_norm": 2.1759402751922607, "learning_rate": 7.5152000000000005e-06, "loss": 0.0437, "step": 12425 }, { "epoch": 0.6624726768673028, "grad_norm": 1.5801782608032227, "learning_rate": 7.515e-06, "loss": 0.03, "step": 12426 }, { "epoch": 0.6625259902969558, "grad_norm": 2.2596540451049805, "learning_rate": 7.514800000000001e-06, "loss": 0.046, "step": 12427 }, { "epoch": 0.6625793037266088, "grad_norm": 2.5420334339141846, "learning_rate": 7.514600000000001e-06, "loss": 0.034, "step": 12428 }, { "epoch": 0.6626326171562616, "grad_norm": 3.320775032043457, "learning_rate": 7.5144000000000004e-06, "loss": 0.0469, "step": 12429 }, { "epoch": 0.6626859305859146, "grad_norm": 3.710442304611206, "learning_rate": 7.5142e-06, "loss": 0.0533, "step": 12430 }, { "epoch": 0.6627392440155675, "grad_norm": 1.5877456665039062, "learning_rate": 7.514e-06, "loss": 0.0341, "step": 12431 }, { "epoch": 0.6627925574452205, "grad_norm": 1.47725248336792, "learning_rate": 7.513800000000001e-06, "loss": 0.0225, "step": 12432 }, { "epoch": 0.6628458708748733, "grad_norm": 1.8785831928253174, "learning_rate": 7.513600000000001e-06, "loss": 0.0392, "step": 12433 }, { "epoch": 0.6628991843045263, "grad_norm": 2.3519139289855957, "learning_rate": 7.513400000000001e-06, "loss": 0.046, "step": 12434 }, { "epoch": 0.6629524977341792, "grad_norm": 3.122736930847168, "learning_rate": 7.5132e-06, "loss": 0.0459, "step": 12435 }, { "epoch": 0.6630058111638322, "grad_norm": 3.2366526126861572, "learning_rate": 7.513e-06, "loss": 0.051, "step": 12436 }, { "epoch": 0.6630591245934851, "grad_norm": 1.8918054103851318, "learning_rate": 7.5128e-06, "loss": 0.0187, "step": 12437 }, { "epoch": 0.663112438023138, "grad_norm": 1.956043004989624, "learning_rate": 7.512600000000001e-06, "loss": 0.0306, "step": 12438 }, { "epoch": 0.663165751452791, "grad_norm": 1.8931714296340942, "learning_rate": 7.512400000000001e-06, "loss": 0.037, "step": 12439 }, { "epoch": 0.6632190648824439, "grad_norm": 1.4546915292739868, "learning_rate": 7.512200000000001e-06, "loss": 0.02, "step": 12440 }, { "epoch": 0.6632723783120968, "grad_norm": 2.253232002258301, "learning_rate": 7.512e-06, "loss": 0.0443, "step": 12441 }, { "epoch": 0.6633256917417497, "grad_norm": 3.1418957710266113, "learning_rate": 7.5118e-06, "loss": 0.0318, "step": 12442 }, { "epoch": 0.6633790051714027, "grad_norm": 2.405611991882324, "learning_rate": 7.511600000000001e-06, "loss": 0.0226, "step": 12443 }, { "epoch": 0.6634323186010556, "grad_norm": 1.6393635272979736, "learning_rate": 7.511400000000001e-06, "loss": 0.0313, "step": 12444 }, { "epoch": 0.6634856320307085, "grad_norm": 2.2118966579437256, "learning_rate": 7.511200000000001e-06, "loss": 0.0191, "step": 12445 }, { "epoch": 0.6635389454603615, "grad_norm": 3.2986152172088623, "learning_rate": 7.511000000000001e-06, "loss": 0.0208, "step": 12446 }, { "epoch": 0.6635922588900144, "grad_norm": 2.024399995803833, "learning_rate": 7.5108e-06, "loss": 0.0274, "step": 12447 }, { "epoch": 0.6636455723196674, "grad_norm": 2.7286274433135986, "learning_rate": 7.5106e-06, "loss": 0.0394, "step": 12448 }, { "epoch": 0.6636988857493202, "grad_norm": 1.5457074642181396, "learning_rate": 7.510400000000001e-06, "loss": 0.0381, "step": 12449 }, { "epoch": 0.6637521991789732, "grad_norm": 1.8862793445587158, "learning_rate": 7.5102000000000005e-06, "loss": 0.0331, "step": 12450 }, { "epoch": 0.6638055126086261, "grad_norm": 1.4509902000427246, "learning_rate": 7.510000000000001e-06, "loss": 0.0293, "step": 12451 }, { "epoch": 0.6638588260382791, "grad_norm": 1.583495855331421, "learning_rate": 7.5098000000000005e-06, "loss": 0.019, "step": 12452 }, { "epoch": 0.6639121394679319, "grad_norm": 3.2033448219299316, "learning_rate": 7.5096e-06, "loss": 0.0336, "step": 12453 }, { "epoch": 0.6639654528975849, "grad_norm": 1.220863699913025, "learning_rate": 7.509400000000001e-06, "loss": 0.0187, "step": 12454 }, { "epoch": 0.6640187663272379, "grad_norm": 1.836417317390442, "learning_rate": 7.509200000000001e-06, "loss": 0.034, "step": 12455 }, { "epoch": 0.6640720797568908, "grad_norm": 1.654162883758545, "learning_rate": 7.5090000000000004e-06, "loss": 0.0165, "step": 12456 }, { "epoch": 0.6641253931865437, "grad_norm": 3.1917715072631836, "learning_rate": 7.508800000000001e-06, "loss": 0.0307, "step": 12457 }, { "epoch": 0.6641787066161966, "grad_norm": 2.4114983081817627, "learning_rate": 7.5086e-06, "loss": 0.0368, "step": 12458 }, { "epoch": 0.6642320200458496, "grad_norm": 1.5689008235931396, "learning_rate": 7.5084e-06, "loss": 0.0212, "step": 12459 }, { "epoch": 0.6642853334755024, "grad_norm": 1.988167405128479, "learning_rate": 7.508200000000001e-06, "loss": 0.0298, "step": 12460 }, { "epoch": 0.6643386469051554, "grad_norm": 1.9537813663482666, "learning_rate": 7.508000000000001e-06, "loss": 0.0318, "step": 12461 }, { "epoch": 0.6643919603348083, "grad_norm": 1.9717910289764404, "learning_rate": 7.5078e-06, "loss": 0.0236, "step": 12462 }, { "epoch": 0.6644452737644613, "grad_norm": 2.519841432571411, "learning_rate": 7.507600000000001e-06, "loss": 0.0485, "step": 12463 }, { "epoch": 0.6644985871941141, "grad_norm": 1.7175617218017578, "learning_rate": 7.5074e-06, "loss": 0.0291, "step": 12464 }, { "epoch": 0.6645519006237671, "grad_norm": 1.6210691928863525, "learning_rate": 7.507200000000001e-06, "loss": 0.0167, "step": 12465 }, { "epoch": 0.6646052140534201, "grad_norm": 2.14670467376709, "learning_rate": 7.507000000000001e-06, "loss": 0.0407, "step": 12466 }, { "epoch": 0.664658527483073, "grad_norm": 2.0674192905426025, "learning_rate": 7.506800000000001e-06, "loss": 0.0263, "step": 12467 }, { "epoch": 0.664711840912726, "grad_norm": 2.199791669845581, "learning_rate": 7.5066e-06, "loss": 0.0377, "step": 12468 }, { "epoch": 0.6647651543423788, "grad_norm": 3.3038840293884277, "learning_rate": 7.506400000000001e-06, "loss": 0.0342, "step": 12469 }, { "epoch": 0.6648184677720318, "grad_norm": 2.0500171184539795, "learning_rate": 7.5062e-06, "loss": 0.0232, "step": 12470 }, { "epoch": 0.6648717812016847, "grad_norm": 2.6852126121520996, "learning_rate": 7.506000000000001e-06, "loss": 0.0372, "step": 12471 }, { "epoch": 0.6649250946313376, "grad_norm": 3.8948705196380615, "learning_rate": 7.505800000000001e-06, "loss": 0.0614, "step": 12472 }, { "epoch": 0.6649784080609905, "grad_norm": 1.017340064048767, "learning_rate": 7.505600000000001e-06, "loss": 0.0165, "step": 12473 }, { "epoch": 0.6650317214906435, "grad_norm": 1.572624683380127, "learning_rate": 7.5054e-06, "loss": 0.0134, "step": 12474 }, { "epoch": 0.6650850349202965, "grad_norm": 1.8298777341842651, "learning_rate": 7.505200000000001e-06, "loss": 0.0308, "step": 12475 }, { "epoch": 0.6651383483499493, "grad_norm": 2.3194358348846436, "learning_rate": 7.505e-06, "loss": 0.0508, "step": 12476 }, { "epoch": 0.6651916617796023, "grad_norm": 2.8715531826019287, "learning_rate": 7.5048000000000006e-06, "loss": 0.0395, "step": 12477 }, { "epoch": 0.6652449752092552, "grad_norm": 4.7933831214904785, "learning_rate": 7.504600000000001e-06, "loss": 0.0433, "step": 12478 }, { "epoch": 0.6652982886389082, "grad_norm": 1.8717585802078247, "learning_rate": 7.5044000000000005e-06, "loss": 0.0233, "step": 12479 }, { "epoch": 0.665351602068561, "grad_norm": 1.6917827129364014, "learning_rate": 7.5042e-06, "loss": 0.0242, "step": 12480 }, { "epoch": 0.665404915498214, "grad_norm": 1.479962944984436, "learning_rate": 7.5040000000000005e-06, "loss": 0.0257, "step": 12481 }, { "epoch": 0.6654582289278669, "grad_norm": 4.6211957931518555, "learning_rate": 7.503800000000001e-06, "loss": 0.1023, "step": 12482 }, { "epoch": 0.6655115423575199, "grad_norm": 2.2857213020324707, "learning_rate": 7.5036000000000005e-06, "loss": 0.0317, "step": 12483 }, { "epoch": 0.6655648557871728, "grad_norm": 2.914541244506836, "learning_rate": 7.503400000000001e-06, "loss": 0.0437, "step": 12484 }, { "epoch": 0.6656181692168257, "grad_norm": 6.089983940124512, "learning_rate": 7.5032000000000004e-06, "loss": 0.0385, "step": 12485 }, { "epoch": 0.6656714826464787, "grad_norm": 1.5016411542892456, "learning_rate": 7.503e-06, "loss": 0.0157, "step": 12486 }, { "epoch": 0.6657247960761316, "grad_norm": 2.2427523136138916, "learning_rate": 7.5028e-06, "loss": 0.0237, "step": 12487 }, { "epoch": 0.6657781095057845, "grad_norm": 3.31900691986084, "learning_rate": 7.502600000000001e-06, "loss": 0.0439, "step": 12488 }, { "epoch": 0.6658314229354374, "grad_norm": 1.8313543796539307, "learning_rate": 7.5024e-06, "loss": 0.0289, "step": 12489 }, { "epoch": 0.6658847363650904, "grad_norm": 1.79450523853302, "learning_rate": 7.502200000000001e-06, "loss": 0.0302, "step": 12490 }, { "epoch": 0.6659380497947432, "grad_norm": 2.6138710975646973, "learning_rate": 7.502e-06, "loss": 0.0484, "step": 12491 }, { "epoch": 0.6659913632243962, "grad_norm": 1.7455729246139526, "learning_rate": 7.5018e-06, "loss": 0.0229, "step": 12492 }, { "epoch": 0.6660446766540492, "grad_norm": 1.8391776084899902, "learning_rate": 7.501600000000001e-06, "loss": 0.0324, "step": 12493 }, { "epoch": 0.6660979900837021, "grad_norm": 4.421199798583984, "learning_rate": 7.501400000000001e-06, "loss": 0.0395, "step": 12494 }, { "epoch": 0.666151303513355, "grad_norm": 2.974133014678955, "learning_rate": 7.5012e-06, "loss": 0.0368, "step": 12495 }, { "epoch": 0.6662046169430079, "grad_norm": 2.2692320346832275, "learning_rate": 7.501000000000001e-06, "loss": 0.052, "step": 12496 }, { "epoch": 0.6662579303726609, "grad_norm": 2.2539782524108887, "learning_rate": 7.5008e-06, "loss": 0.0456, "step": 12497 }, { "epoch": 0.6663112438023138, "grad_norm": 1.6278668642044067, "learning_rate": 7.5006e-06, "loss": 0.0336, "step": 12498 }, { "epoch": 0.6663645572319667, "grad_norm": 2.105276346206665, "learning_rate": 7.500400000000001e-06, "loss": 0.0351, "step": 12499 }, { "epoch": 0.6664178706616196, "grad_norm": 2.2916014194488525, "learning_rate": 7.500200000000001e-06, "loss": 0.0321, "step": 12500 }, { "epoch": 0.6664711840912726, "grad_norm": 5.452954292297363, "learning_rate": 7.500000000000001e-06, "loss": 0.0407, "step": 12501 }, { "epoch": 0.6665244975209256, "grad_norm": 6.7343950271606445, "learning_rate": 7.499800000000001e-06, "loss": 0.0435, "step": 12502 }, { "epoch": 0.6665778109505784, "grad_norm": 4.054458141326904, "learning_rate": 7.4996e-06, "loss": 0.0324, "step": 12503 }, { "epoch": 0.6666311243802314, "grad_norm": 2.990856170654297, "learning_rate": 7.499400000000001e-06, "loss": 0.0594, "step": 12504 }, { "epoch": 0.6666844378098843, "grad_norm": 3.037289619445801, "learning_rate": 7.499200000000001e-06, "loss": 0.0379, "step": 12505 }, { "epoch": 0.6667377512395373, "grad_norm": 2.233855724334717, "learning_rate": 7.4990000000000005e-06, "loss": 0.0356, "step": 12506 }, { "epoch": 0.6667910646691901, "grad_norm": 3.383272409439087, "learning_rate": 7.498800000000001e-06, "loss": 0.0403, "step": 12507 }, { "epoch": 0.6668443780988431, "grad_norm": 7.510941028594971, "learning_rate": 7.4986000000000005e-06, "loss": 0.0519, "step": 12508 }, { "epoch": 0.666897691528496, "grad_norm": 2.751471757888794, "learning_rate": 7.4984e-06, "loss": 0.0314, "step": 12509 }, { "epoch": 0.666951004958149, "grad_norm": 3.8213860988616943, "learning_rate": 7.498200000000001e-06, "loss": 0.0344, "step": 12510 }, { "epoch": 0.6670043183878019, "grad_norm": 2.7935187816619873, "learning_rate": 7.498000000000001e-06, "loss": 0.0504, "step": 12511 }, { "epoch": 0.6670576318174548, "grad_norm": 2.4362714290618896, "learning_rate": 7.4978000000000004e-06, "loss": 0.0499, "step": 12512 }, { "epoch": 0.6671109452471078, "grad_norm": 2.3384501934051514, "learning_rate": 7.497600000000001e-06, "loss": 0.0442, "step": 12513 }, { "epoch": 0.6671642586767607, "grad_norm": 1.7544457912445068, "learning_rate": 7.4974e-06, "loss": 0.0354, "step": 12514 }, { "epoch": 0.6672175721064136, "grad_norm": 2.1555118560791016, "learning_rate": 7.497200000000001e-06, "loss": 0.0338, "step": 12515 }, { "epoch": 0.6672708855360665, "grad_norm": 1.4247177839279175, "learning_rate": 7.497000000000001e-06, "loss": 0.0201, "step": 12516 }, { "epoch": 0.6673241989657195, "grad_norm": 1.7961931228637695, "learning_rate": 7.496800000000001e-06, "loss": 0.0277, "step": 12517 }, { "epoch": 0.6673775123953724, "grad_norm": 2.6056413650512695, "learning_rate": 7.4966e-06, "loss": 0.0224, "step": 12518 }, { "epoch": 0.6674308258250253, "grad_norm": 1.5090903043746948, "learning_rate": 7.496400000000001e-06, "loss": 0.0105, "step": 12519 }, { "epoch": 0.6674841392546782, "grad_norm": 0.784092903137207, "learning_rate": 7.4962e-06, "loss": 0.0089, "step": 12520 }, { "epoch": 0.6675374526843312, "grad_norm": 3.2293484210968018, "learning_rate": 7.496000000000001e-06, "loss": 0.0618, "step": 12521 }, { "epoch": 0.6675907661139842, "grad_norm": 3.424619197845459, "learning_rate": 7.495800000000001e-06, "loss": 0.0648, "step": 12522 }, { "epoch": 0.667644079543637, "grad_norm": 1.5186313390731812, "learning_rate": 7.495600000000001e-06, "loss": 0.0341, "step": 12523 }, { "epoch": 0.66769739297329, "grad_norm": 1.6216658353805542, "learning_rate": 7.4954e-06, "loss": 0.0347, "step": 12524 }, { "epoch": 0.6677507064029429, "grad_norm": 1.9084721803665161, "learning_rate": 7.495200000000001e-06, "loss": 0.0231, "step": 12525 }, { "epoch": 0.6678040198325959, "grad_norm": 2.381589412689209, "learning_rate": 7.495000000000001e-06, "loss": 0.0197, "step": 12526 }, { "epoch": 0.6678573332622487, "grad_norm": 2.3272035121917725, "learning_rate": 7.494800000000001e-06, "loss": 0.025, "step": 12527 }, { "epoch": 0.6679106466919017, "grad_norm": 2.2075610160827637, "learning_rate": 7.494600000000001e-06, "loss": 0.0342, "step": 12528 }, { "epoch": 0.6679639601215546, "grad_norm": 2.358760356903076, "learning_rate": 7.494400000000001e-06, "loss": 0.0324, "step": 12529 }, { "epoch": 0.6680172735512075, "grad_norm": 2.2215752601623535, "learning_rate": 7.4942e-06, "loss": 0.0312, "step": 12530 }, { "epoch": 0.6680705869808605, "grad_norm": 1.5373364686965942, "learning_rate": 7.494000000000001e-06, "loss": 0.0296, "step": 12531 }, { "epoch": 0.6681239004105134, "grad_norm": 3.1634058952331543, "learning_rate": 7.493800000000001e-06, "loss": 0.0516, "step": 12532 }, { "epoch": 0.6681772138401664, "grad_norm": 2.7789642810821533, "learning_rate": 7.4936000000000006e-06, "loss": 0.0251, "step": 12533 }, { "epoch": 0.6682305272698192, "grad_norm": 2.2297730445861816, "learning_rate": 7.493400000000001e-06, "loss": 0.0239, "step": 12534 }, { "epoch": 0.6682838406994722, "grad_norm": 2.1290061473846436, "learning_rate": 7.4932000000000005e-06, "loss": 0.0292, "step": 12535 }, { "epoch": 0.6683371541291251, "grad_norm": 2.828566789627075, "learning_rate": 7.493e-06, "loss": 0.0605, "step": 12536 }, { "epoch": 0.6683904675587781, "grad_norm": 1.9429218769073486, "learning_rate": 7.4928000000000005e-06, "loss": 0.0523, "step": 12537 }, { "epoch": 0.6684437809884309, "grad_norm": 0.9803493022918701, "learning_rate": 7.492600000000001e-06, "loss": 0.0091, "step": 12538 }, { "epoch": 0.6684970944180839, "grad_norm": 2.3134284019470215, "learning_rate": 7.4924000000000005e-06, "loss": 0.032, "step": 12539 }, { "epoch": 0.6685504078477369, "grad_norm": 1.9792416095733643, "learning_rate": 7.492200000000001e-06, "loss": 0.021, "step": 12540 }, { "epoch": 0.6686037212773898, "grad_norm": 2.4624457359313965, "learning_rate": 7.4920000000000004e-06, "loss": 0.038, "step": 12541 }, { "epoch": 0.6686570347070427, "grad_norm": 2.162390947341919, "learning_rate": 7.4918e-06, "loss": 0.0406, "step": 12542 }, { "epoch": 0.6687103481366956, "grad_norm": 2.15739369392395, "learning_rate": 7.491600000000001e-06, "loss": 0.0242, "step": 12543 }, { "epoch": 0.6687636615663486, "grad_norm": 1.765995740890503, "learning_rate": 7.491400000000001e-06, "loss": 0.0189, "step": 12544 }, { "epoch": 0.6688169749960015, "grad_norm": 1.8321807384490967, "learning_rate": 7.4912e-06, "loss": 0.0235, "step": 12545 }, { "epoch": 0.6688702884256544, "grad_norm": 5.123769760131836, "learning_rate": 7.491000000000001e-06, "loss": 0.0896, "step": 12546 }, { "epoch": 0.6689236018553073, "grad_norm": 2.3356070518493652, "learning_rate": 7.4908e-06, "loss": 0.0299, "step": 12547 }, { "epoch": 0.6689769152849603, "grad_norm": 2.2164008617401123, "learning_rate": 7.4906e-06, "loss": 0.0327, "step": 12548 }, { "epoch": 0.6690302287146133, "grad_norm": 2.2641656398773193, "learning_rate": 7.490400000000001e-06, "loss": 0.0147, "step": 12549 }, { "epoch": 0.6690835421442661, "grad_norm": 5.732987880706787, "learning_rate": 7.490200000000001e-06, "loss": 0.0773, "step": 12550 }, { "epoch": 0.6691368555739191, "grad_norm": 2.873696804046631, "learning_rate": 7.49e-06, "loss": 0.0271, "step": 12551 }, { "epoch": 0.669190169003572, "grad_norm": 2.9724981784820557, "learning_rate": 7.489800000000001e-06, "loss": 0.0337, "step": 12552 }, { "epoch": 0.669243482433225, "grad_norm": 2.496633291244507, "learning_rate": 7.4896e-06, "loss": 0.0344, "step": 12553 }, { "epoch": 0.6692967958628778, "grad_norm": 1.3610455989837646, "learning_rate": 7.4894000000000015e-06, "loss": 0.0141, "step": 12554 }, { "epoch": 0.6693501092925308, "grad_norm": 2.2981956005096436, "learning_rate": 7.489200000000001e-06, "loss": 0.0246, "step": 12555 }, { "epoch": 0.6694034227221837, "grad_norm": 3.542785882949829, "learning_rate": 7.489000000000001e-06, "loss": 0.0648, "step": 12556 }, { "epoch": 0.6694567361518367, "grad_norm": 2.9163403511047363, "learning_rate": 7.4888e-06, "loss": 0.0274, "step": 12557 }, { "epoch": 0.6695100495814896, "grad_norm": 2.4402148723602295, "learning_rate": 7.488600000000001e-06, "loss": 0.0189, "step": 12558 }, { "epoch": 0.6695633630111425, "grad_norm": 4.493215084075928, "learning_rate": 7.4884e-06, "loss": 0.0351, "step": 12559 }, { "epoch": 0.6696166764407955, "grad_norm": 1.257333755493164, "learning_rate": 7.488200000000001e-06, "loss": 0.011, "step": 12560 }, { "epoch": 0.6696699898704483, "grad_norm": 1.5141708850860596, "learning_rate": 7.488000000000001e-06, "loss": 0.032, "step": 12561 }, { "epoch": 0.6697233033001013, "grad_norm": 1.1453126668930054, "learning_rate": 7.4878000000000005e-06, "loss": 0.0198, "step": 12562 }, { "epoch": 0.6697766167297542, "grad_norm": 2.421767234802246, "learning_rate": 7.4876e-06, "loss": 0.0384, "step": 12563 }, { "epoch": 0.6698299301594072, "grad_norm": 1.3729948997497559, "learning_rate": 7.4874000000000005e-06, "loss": 0.0138, "step": 12564 }, { "epoch": 0.66988324358906, "grad_norm": 1.0481928586959839, "learning_rate": 7.487200000000001e-06, "loss": 0.0205, "step": 12565 }, { "epoch": 0.669936557018713, "grad_norm": 1.1755708456039429, "learning_rate": 7.487000000000001e-06, "loss": 0.0098, "step": 12566 }, { "epoch": 0.669989870448366, "grad_norm": 1.198625087738037, "learning_rate": 7.486800000000001e-06, "loss": 0.029, "step": 12567 }, { "epoch": 0.6700431838780189, "grad_norm": 2.547940254211426, "learning_rate": 7.4866000000000005e-06, "loss": 0.029, "step": 12568 }, { "epoch": 0.6700964973076718, "grad_norm": 1.4558659791946411, "learning_rate": 7.4864e-06, "loss": 0.0166, "step": 12569 }, { "epoch": 0.6701498107373247, "grad_norm": 2.6035051345825195, "learning_rate": 7.4862000000000004e-06, "loss": 0.0406, "step": 12570 }, { "epoch": 0.6702031241669777, "grad_norm": 2.1711294651031494, "learning_rate": 7.486000000000001e-06, "loss": 0.0461, "step": 12571 }, { "epoch": 0.6702564375966306, "grad_norm": 1.9071720838546753, "learning_rate": 7.485800000000001e-06, "loss": 0.0265, "step": 12572 }, { "epoch": 0.6703097510262835, "grad_norm": 1.2421882152557373, "learning_rate": 7.485600000000001e-06, "loss": 0.019, "step": 12573 }, { "epoch": 0.6703630644559364, "grad_norm": 3.2616937160491943, "learning_rate": 7.4854e-06, "loss": 0.0429, "step": 12574 }, { "epoch": 0.6704163778855894, "grad_norm": 1.3337897062301636, "learning_rate": 7.4852e-06, "loss": 0.0085, "step": 12575 }, { "epoch": 0.6704696913152423, "grad_norm": 2.416252851486206, "learning_rate": 7.485000000000001e-06, "loss": 0.0198, "step": 12576 }, { "epoch": 0.6705230047448952, "grad_norm": 1.5890741348266602, "learning_rate": 7.484800000000001e-06, "loss": 0.0244, "step": 12577 }, { "epoch": 0.6705763181745482, "grad_norm": 1.997503638267517, "learning_rate": 7.484600000000001e-06, "loss": 0.0265, "step": 12578 }, { "epoch": 0.6706296316042011, "grad_norm": 3.2024450302124023, "learning_rate": 7.484400000000001e-06, "loss": 0.0396, "step": 12579 }, { "epoch": 0.6706829450338541, "grad_norm": 1.6483279466629028, "learning_rate": 7.4842e-06, "loss": 0.0115, "step": 12580 }, { "epoch": 0.6707362584635069, "grad_norm": 2.46956467628479, "learning_rate": 7.484e-06, "loss": 0.0202, "step": 12581 }, { "epoch": 0.6707895718931599, "grad_norm": 1.9172961711883545, "learning_rate": 7.483800000000001e-06, "loss": 0.012, "step": 12582 }, { "epoch": 0.6708428853228128, "grad_norm": 2.797549247741699, "learning_rate": 7.483600000000001e-06, "loss": 0.0475, "step": 12583 }, { "epoch": 0.6708961987524658, "grad_norm": 3.411830425262451, "learning_rate": 7.483400000000001e-06, "loss": 0.0369, "step": 12584 }, { "epoch": 0.6709495121821186, "grad_norm": 3.415187120437622, "learning_rate": 7.483200000000001e-06, "loss": 0.033, "step": 12585 }, { "epoch": 0.6710028256117716, "grad_norm": 5.028006553649902, "learning_rate": 7.483e-06, "loss": 0.0552, "step": 12586 }, { "epoch": 0.6710561390414246, "grad_norm": 3.3375535011291504, "learning_rate": 7.4828e-06, "loss": 0.03, "step": 12587 }, { "epoch": 0.6711094524710775, "grad_norm": 1.0194491147994995, "learning_rate": 7.482600000000001e-06, "loss": 0.011, "step": 12588 }, { "epoch": 0.6711627659007304, "grad_norm": 1.38187575340271, "learning_rate": 7.4824000000000006e-06, "loss": 0.0217, "step": 12589 }, { "epoch": 0.6712160793303833, "grad_norm": 2.7848989963531494, "learning_rate": 7.482200000000001e-06, "loss": 0.0385, "step": 12590 }, { "epoch": 0.6712693927600363, "grad_norm": 1.9415888786315918, "learning_rate": 7.4820000000000005e-06, "loss": 0.0152, "step": 12591 }, { "epoch": 0.6713227061896891, "grad_norm": 1.9293344020843506, "learning_rate": 7.4818e-06, "loss": 0.0335, "step": 12592 }, { "epoch": 0.6713760196193421, "grad_norm": 1.5393823385238647, "learning_rate": 7.481600000000001e-06, "loss": 0.04, "step": 12593 }, { "epoch": 0.671429333048995, "grad_norm": 2.7683072090148926, "learning_rate": 7.481400000000001e-06, "loss": 0.0293, "step": 12594 }, { "epoch": 0.671482646478648, "grad_norm": 2.4632561206817627, "learning_rate": 7.4812000000000005e-06, "loss": 0.0172, "step": 12595 }, { "epoch": 0.671535959908301, "grad_norm": 1.411781668663025, "learning_rate": 7.481000000000001e-06, "loss": 0.0213, "step": 12596 }, { "epoch": 0.6715892733379538, "grad_norm": 0.4170069396495819, "learning_rate": 7.4808000000000004e-06, "loss": 0.0042, "step": 12597 }, { "epoch": 0.6716425867676068, "grad_norm": 1.7426567077636719, "learning_rate": 7.4806e-06, "loss": 0.0202, "step": 12598 }, { "epoch": 0.6716959001972597, "grad_norm": 1.2332154512405396, "learning_rate": 7.480400000000001e-06, "loss": 0.0057, "step": 12599 }, { "epoch": 0.6717492136269126, "grad_norm": 4.505399227142334, "learning_rate": 7.480200000000001e-06, "loss": 0.0467, "step": 12600 }, { "epoch": 0.6718025270565655, "grad_norm": 4.213988780975342, "learning_rate": 7.48e-06, "loss": 0.0668, "step": 12601 }, { "epoch": 0.6718558404862185, "grad_norm": 1.1293296813964844, "learning_rate": 7.479800000000001e-06, "loss": 0.0091, "step": 12602 }, { "epoch": 0.6719091539158714, "grad_norm": 0.7213629484176636, "learning_rate": 7.4796e-06, "loss": 0.0047, "step": 12603 }, { "epoch": 0.6719624673455243, "grad_norm": 3.419731616973877, "learning_rate": 7.479400000000001e-06, "loss": 0.0501, "step": 12604 }, { "epoch": 0.6720157807751773, "grad_norm": 1.4454847574234009, "learning_rate": 7.479200000000001e-06, "loss": 0.0152, "step": 12605 }, { "epoch": 0.6720690942048302, "grad_norm": 1.8250409364700317, "learning_rate": 7.479000000000001e-06, "loss": 0.0212, "step": 12606 }, { "epoch": 0.6721224076344832, "grad_norm": 0.6717932224273682, "learning_rate": 7.4788e-06, "loss": 0.004, "step": 12607 }, { "epoch": 0.672175721064136, "grad_norm": 2.774942398071289, "learning_rate": 7.478600000000001e-06, "loss": 0.065, "step": 12608 }, { "epoch": 0.672229034493789, "grad_norm": 4.251163959503174, "learning_rate": 7.4784e-06, "loss": 0.0636, "step": 12609 }, { "epoch": 0.6722823479234419, "grad_norm": 2.940279960632324, "learning_rate": 7.478200000000001e-06, "loss": 0.0554, "step": 12610 }, { "epoch": 0.6723356613530949, "grad_norm": 4.23732328414917, "learning_rate": 7.478000000000001e-06, "loss": 0.0447, "step": 12611 }, { "epoch": 0.6723889747827477, "grad_norm": 1.896329641342163, "learning_rate": 7.477800000000001e-06, "loss": 0.0228, "step": 12612 }, { "epoch": 0.6724422882124007, "grad_norm": 2.8250632286071777, "learning_rate": 7.4776e-06, "loss": 0.0227, "step": 12613 }, { "epoch": 0.6724956016420537, "grad_norm": 2.200172185897827, "learning_rate": 7.477400000000001e-06, "loss": 0.0266, "step": 12614 }, { "epoch": 0.6725489150717066, "grad_norm": 1.2830102443695068, "learning_rate": 7.477200000000001e-06, "loss": 0.019, "step": 12615 }, { "epoch": 0.6726022285013595, "grad_norm": 1.0076043605804443, "learning_rate": 7.477000000000001e-06, "loss": 0.0092, "step": 12616 }, { "epoch": 0.6726555419310124, "grad_norm": 1.1974315643310547, "learning_rate": 7.476800000000001e-06, "loss": 0.0187, "step": 12617 }, { "epoch": 0.6727088553606654, "grad_norm": 2.3951456546783447, "learning_rate": 7.4766000000000005e-06, "loss": 0.0405, "step": 12618 }, { "epoch": 0.6727621687903182, "grad_norm": 2.22586989402771, "learning_rate": 7.4764e-06, "loss": 0.0236, "step": 12619 }, { "epoch": 0.6728154822199712, "grad_norm": 3.858330011367798, "learning_rate": 7.4762000000000005e-06, "loss": 0.0949, "step": 12620 }, { "epoch": 0.6728687956496241, "grad_norm": 1.5861843824386597, "learning_rate": 7.476000000000001e-06, "loss": 0.0221, "step": 12621 }, { "epoch": 0.6729221090792771, "grad_norm": 1.1183092594146729, "learning_rate": 7.4758000000000005e-06, "loss": 0.022, "step": 12622 }, { "epoch": 0.67297542250893, "grad_norm": 3.5358967781066895, "learning_rate": 7.475600000000001e-06, "loss": 0.0403, "step": 12623 }, { "epoch": 0.6730287359385829, "grad_norm": 2.2448699474334717, "learning_rate": 7.4754000000000005e-06, "loss": 0.0393, "step": 12624 }, { "epoch": 0.6730820493682359, "grad_norm": 2.5130269527435303, "learning_rate": 7.4752e-06, "loss": 0.0282, "step": 12625 }, { "epoch": 0.6731353627978888, "grad_norm": 2.133624792098999, "learning_rate": 7.475000000000001e-06, "loss": 0.0425, "step": 12626 }, { "epoch": 0.6731886762275417, "grad_norm": 4.1825737953186035, "learning_rate": 7.474800000000001e-06, "loss": 0.0186, "step": 12627 }, { "epoch": 0.6732419896571946, "grad_norm": 3.657506227493286, "learning_rate": 7.4746e-06, "loss": 0.0272, "step": 12628 }, { "epoch": 0.6732953030868476, "grad_norm": 4.89132022857666, "learning_rate": 7.474400000000001e-06, "loss": 0.0814, "step": 12629 }, { "epoch": 0.6733486165165005, "grad_norm": 5.290380477905273, "learning_rate": 7.4742e-06, "loss": 0.0559, "step": 12630 }, { "epoch": 0.6734019299461534, "grad_norm": 1.4389499425888062, "learning_rate": 7.474e-06, "loss": 0.0311, "step": 12631 }, { "epoch": 0.6734552433758063, "grad_norm": 1.2458759546279907, "learning_rate": 7.473800000000001e-06, "loss": 0.0259, "step": 12632 }, { "epoch": 0.6735085568054593, "grad_norm": 1.162606954574585, "learning_rate": 7.473600000000001e-06, "loss": 0.0156, "step": 12633 }, { "epoch": 0.6735618702351123, "grad_norm": 3.735523223876953, "learning_rate": 7.4734e-06, "loss": 0.0531, "step": 12634 }, { "epoch": 0.6736151836647651, "grad_norm": 0.5887942314147949, "learning_rate": 7.473200000000001e-06, "loss": 0.0058, "step": 12635 }, { "epoch": 0.6736684970944181, "grad_norm": 2.027299404144287, "learning_rate": 7.473e-06, "loss": 0.0421, "step": 12636 }, { "epoch": 0.673721810524071, "grad_norm": 5.521389484405518, "learning_rate": 7.4728000000000015e-06, "loss": 0.0654, "step": 12637 }, { "epoch": 0.673775123953724, "grad_norm": 2.480292558670044, "learning_rate": 7.472600000000001e-06, "loss": 0.0493, "step": 12638 }, { "epoch": 0.6738284373833768, "grad_norm": 4.956600666046143, "learning_rate": 7.472400000000001e-06, "loss": 0.0274, "step": 12639 }, { "epoch": 0.6738817508130298, "grad_norm": 1.9537391662597656, "learning_rate": 7.4722e-06, "loss": 0.0276, "step": 12640 }, { "epoch": 0.6739350642426827, "grad_norm": 3.5189919471740723, "learning_rate": 7.472000000000001e-06, "loss": 0.0354, "step": 12641 }, { "epoch": 0.6739883776723357, "grad_norm": 2.1941263675689697, "learning_rate": 7.4718e-06, "loss": 0.0378, "step": 12642 }, { "epoch": 0.6740416911019886, "grad_norm": 1.4066163301467896, "learning_rate": 7.4716000000000014e-06, "loss": 0.0169, "step": 12643 }, { "epoch": 0.6740950045316415, "grad_norm": 5.455418586730957, "learning_rate": 7.471400000000001e-06, "loss": 0.0343, "step": 12644 }, { "epoch": 0.6741483179612945, "grad_norm": 1.5019253492355347, "learning_rate": 7.4712000000000006e-06, "loss": 0.0323, "step": 12645 }, { "epoch": 0.6742016313909474, "grad_norm": 3.47790265083313, "learning_rate": 7.471e-06, "loss": 0.0413, "step": 12646 }, { "epoch": 0.6742549448206003, "grad_norm": 1.826403260231018, "learning_rate": 7.4708000000000005e-06, "loss": 0.0244, "step": 12647 }, { "epoch": 0.6743082582502532, "grad_norm": 3.964049816131592, "learning_rate": 7.4706e-06, "loss": 0.0403, "step": 12648 }, { "epoch": 0.6743615716799062, "grad_norm": 1.6579430103302002, "learning_rate": 7.470400000000001e-06, "loss": 0.0329, "step": 12649 }, { "epoch": 0.674414885109559, "grad_norm": 2.048363447189331, "learning_rate": 7.470200000000001e-06, "loss": 0.0345, "step": 12650 }, { "epoch": 0.674468198539212, "grad_norm": 1.606677770614624, "learning_rate": 7.4700000000000005e-06, "loss": 0.0269, "step": 12651 }, { "epoch": 0.674521511968865, "grad_norm": 1.6343494653701782, "learning_rate": 7.4698e-06, "loss": 0.0251, "step": 12652 }, { "epoch": 0.6745748253985179, "grad_norm": 2.010556221008301, "learning_rate": 7.4696000000000004e-06, "loss": 0.0485, "step": 12653 }, { "epoch": 0.6746281388281709, "grad_norm": 3.1540024280548096, "learning_rate": 7.469400000000001e-06, "loss": 0.0548, "step": 12654 }, { "epoch": 0.6746814522578237, "grad_norm": 4.117142200469971, "learning_rate": 7.469200000000001e-06, "loss": 0.0562, "step": 12655 }, { "epoch": 0.6747347656874767, "grad_norm": 1.7367652654647827, "learning_rate": 7.469000000000001e-06, "loss": 0.034, "step": 12656 }, { "epoch": 0.6747880791171296, "grad_norm": 2.61073899269104, "learning_rate": 7.4688e-06, "loss": 0.047, "step": 12657 }, { "epoch": 0.6748413925467825, "grad_norm": 1.6002352237701416, "learning_rate": 7.4686e-06, "loss": 0.0168, "step": 12658 }, { "epoch": 0.6748947059764354, "grad_norm": 1.5380243062973022, "learning_rate": 7.4684e-06, "loss": 0.0343, "step": 12659 }, { "epoch": 0.6749480194060884, "grad_norm": 2.0298948287963867, "learning_rate": 7.468200000000001e-06, "loss": 0.0198, "step": 12660 }, { "epoch": 0.6750013328357414, "grad_norm": 2.4946703910827637, "learning_rate": 7.468000000000001e-06, "loss": 0.0559, "step": 12661 }, { "epoch": 0.6750546462653942, "grad_norm": 0.9232527017593384, "learning_rate": 7.467800000000001e-06, "loss": 0.0142, "step": 12662 }, { "epoch": 0.6751079596950472, "grad_norm": 2.051785469055176, "learning_rate": 7.4676e-06, "loss": 0.0448, "step": 12663 }, { "epoch": 0.6751612731247001, "grad_norm": 2.496553421020508, "learning_rate": 7.4674e-06, "loss": 0.0277, "step": 12664 }, { "epoch": 0.6752145865543531, "grad_norm": 0.9423184990882874, "learning_rate": 7.467200000000001e-06, "loss": 0.0108, "step": 12665 }, { "epoch": 0.6752678999840059, "grad_norm": 0.9873550534248352, "learning_rate": 7.467000000000001e-06, "loss": 0.0139, "step": 12666 }, { "epoch": 0.6753212134136589, "grad_norm": 2.224975109100342, "learning_rate": 7.466800000000001e-06, "loss": 0.0142, "step": 12667 }, { "epoch": 0.6753745268433118, "grad_norm": 1.478055477142334, "learning_rate": 7.466600000000001e-06, "loss": 0.0186, "step": 12668 }, { "epoch": 0.6754278402729648, "grad_norm": 2.6361918449401855, "learning_rate": 7.4664e-06, "loss": 0.0355, "step": 12669 }, { "epoch": 0.6754811537026177, "grad_norm": 2.761415481567383, "learning_rate": 7.4662e-06, "loss": 0.0505, "step": 12670 }, { "epoch": 0.6755344671322706, "grad_norm": 2.9812169075012207, "learning_rate": 7.466000000000001e-06, "loss": 0.0811, "step": 12671 }, { "epoch": 0.6755877805619236, "grad_norm": 2.145294666290283, "learning_rate": 7.465800000000001e-06, "loss": 0.0319, "step": 12672 }, { "epoch": 0.6756410939915765, "grad_norm": 2.6535439491271973, "learning_rate": 7.465600000000001e-06, "loss": 0.0215, "step": 12673 }, { "epoch": 0.6756944074212294, "grad_norm": 1.5908559560775757, "learning_rate": 7.4654000000000006e-06, "loss": 0.0166, "step": 12674 }, { "epoch": 0.6757477208508823, "grad_norm": 3.1987102031707764, "learning_rate": 7.4652e-06, "loss": 0.0381, "step": 12675 }, { "epoch": 0.6758010342805353, "grad_norm": 2.609912872314453, "learning_rate": 7.465000000000001e-06, "loss": 0.0215, "step": 12676 }, { "epoch": 0.6758543477101882, "grad_norm": 2.2390758991241455, "learning_rate": 7.464800000000001e-06, "loss": 0.0315, "step": 12677 }, { "epoch": 0.6759076611398411, "grad_norm": 2.918942451477051, "learning_rate": 7.4646000000000005e-06, "loss": 0.0386, "step": 12678 }, { "epoch": 0.6759609745694941, "grad_norm": 4.0294647216796875, "learning_rate": 7.464400000000001e-06, "loss": 0.064, "step": 12679 }, { "epoch": 0.676014287999147, "grad_norm": 1.8985049724578857, "learning_rate": 7.4642000000000005e-06, "loss": 0.0504, "step": 12680 }, { "epoch": 0.6760676014288, "grad_norm": 1.8848038911819458, "learning_rate": 7.464e-06, "loss": 0.0314, "step": 12681 }, { "epoch": 0.6761209148584528, "grad_norm": 1.6325842142105103, "learning_rate": 7.463800000000001e-06, "loss": 0.0239, "step": 12682 }, { "epoch": 0.6761742282881058, "grad_norm": 1.1141788959503174, "learning_rate": 7.463600000000001e-06, "loss": 0.0232, "step": 12683 }, { "epoch": 0.6762275417177587, "grad_norm": 1.7047680616378784, "learning_rate": 7.4634e-06, "loss": 0.0206, "step": 12684 }, { "epoch": 0.6762808551474117, "grad_norm": 1.884830355644226, "learning_rate": 7.463200000000001e-06, "loss": 0.0126, "step": 12685 }, { "epoch": 0.6763341685770645, "grad_norm": 2.135545015335083, "learning_rate": 7.463e-06, "loss": 0.026, "step": 12686 }, { "epoch": 0.6763874820067175, "grad_norm": 2.2829341888427734, "learning_rate": 7.462800000000001e-06, "loss": 0.0289, "step": 12687 }, { "epoch": 0.6764407954363704, "grad_norm": 1.8561317920684814, "learning_rate": 7.462600000000001e-06, "loss": 0.0331, "step": 12688 }, { "epoch": 0.6764941088660233, "grad_norm": 1.9145673513412476, "learning_rate": 7.462400000000001e-06, "loss": 0.0219, "step": 12689 }, { "epoch": 0.6765474222956763, "grad_norm": 1.9701166152954102, "learning_rate": 7.4622e-06, "loss": 0.0519, "step": 12690 }, { "epoch": 0.6766007357253292, "grad_norm": 2.244405508041382, "learning_rate": 7.462000000000001e-06, "loss": 0.0385, "step": 12691 }, { "epoch": 0.6766540491549822, "grad_norm": 2.8886189460754395, "learning_rate": 7.4618e-06, "loss": 0.0567, "step": 12692 }, { "epoch": 0.676707362584635, "grad_norm": 3.0027952194213867, "learning_rate": 7.461600000000001e-06, "loss": 0.0259, "step": 12693 }, { "epoch": 0.676760676014288, "grad_norm": 3.718916416168213, "learning_rate": 7.461400000000001e-06, "loss": 0.0233, "step": 12694 }, { "epoch": 0.6768139894439409, "grad_norm": 3.641996145248413, "learning_rate": 7.461200000000001e-06, "loss": 0.0235, "step": 12695 }, { "epoch": 0.6768673028735939, "grad_norm": 1.735931158065796, "learning_rate": 7.461e-06, "loss": 0.0372, "step": 12696 }, { "epoch": 0.6769206163032467, "grad_norm": 1.954567313194275, "learning_rate": 7.460800000000001e-06, "loss": 0.0528, "step": 12697 }, { "epoch": 0.6769739297328997, "grad_norm": 1.1958410739898682, "learning_rate": 7.4606e-06, "loss": 0.0261, "step": 12698 }, { "epoch": 0.6770272431625527, "grad_norm": 7.126216888427734, "learning_rate": 7.460400000000001e-06, "loss": 0.064, "step": 12699 }, { "epoch": 0.6770805565922056, "grad_norm": 3.3033928871154785, "learning_rate": 7.460200000000001e-06, "loss": 0.0472, "step": 12700 }, { "epoch": 0.6771338700218585, "grad_norm": 2.686652421951294, "learning_rate": 7.4600000000000006e-06, "loss": 0.0476, "step": 12701 }, { "epoch": 0.6771871834515114, "grad_norm": 1.8291821479797363, "learning_rate": 7.4598e-06, "loss": 0.043, "step": 12702 }, { "epoch": 0.6772404968811644, "grad_norm": 1.780605673789978, "learning_rate": 7.4596000000000005e-06, "loss": 0.0243, "step": 12703 }, { "epoch": 0.6772938103108173, "grad_norm": 2.5934958457946777, "learning_rate": 7.459400000000001e-06, "loss": 0.0459, "step": 12704 }, { "epoch": 0.6773471237404702, "grad_norm": 1.38506281375885, "learning_rate": 7.4592000000000005e-06, "loss": 0.0156, "step": 12705 }, { "epoch": 0.6774004371701231, "grad_norm": 2.0904345512390137, "learning_rate": 7.459000000000001e-06, "loss": 0.0304, "step": 12706 }, { "epoch": 0.6774537505997761, "grad_norm": 3.5453364849090576, "learning_rate": 7.4588000000000005e-06, "loss": 0.0271, "step": 12707 }, { "epoch": 0.6775070640294291, "grad_norm": 2.0540850162506104, "learning_rate": 7.4586e-06, "loss": 0.0411, "step": 12708 }, { "epoch": 0.6775603774590819, "grad_norm": 1.9194742441177368, "learning_rate": 7.4584000000000005e-06, "loss": 0.049, "step": 12709 }, { "epoch": 0.6776136908887349, "grad_norm": 1.5891551971435547, "learning_rate": 7.458200000000001e-06, "loss": 0.0314, "step": 12710 }, { "epoch": 0.6776670043183878, "grad_norm": 2.434042453765869, "learning_rate": 7.458e-06, "loss": 0.0192, "step": 12711 }, { "epoch": 0.6777203177480408, "grad_norm": 2.054403305053711, "learning_rate": 7.457800000000001e-06, "loss": 0.0232, "step": 12712 }, { "epoch": 0.6777736311776936, "grad_norm": 2.1560213565826416, "learning_rate": 7.4576e-06, "loss": 0.0378, "step": 12713 }, { "epoch": 0.6778269446073466, "grad_norm": 2.0120489597320557, "learning_rate": 7.4574e-06, "loss": 0.025, "step": 12714 }, { "epoch": 0.6778802580369995, "grad_norm": 5.040493965148926, "learning_rate": 7.457200000000001e-06, "loss": 0.0644, "step": 12715 }, { "epoch": 0.6779335714666525, "grad_norm": 2.885488986968994, "learning_rate": 7.457000000000001e-06, "loss": 0.0307, "step": 12716 }, { "epoch": 0.6779868848963054, "grad_norm": 1.8343369960784912, "learning_rate": 7.4568e-06, "loss": 0.0212, "step": 12717 }, { "epoch": 0.6780401983259583, "grad_norm": 4.925356864929199, "learning_rate": 7.456600000000001e-06, "loss": 0.059, "step": 12718 }, { "epoch": 0.6780935117556113, "grad_norm": 1.5334311723709106, "learning_rate": 7.4564e-06, "loss": 0.0204, "step": 12719 }, { "epoch": 0.6781468251852641, "grad_norm": 2.8573720455169678, "learning_rate": 7.4562e-06, "loss": 0.0548, "step": 12720 }, { "epoch": 0.6782001386149171, "grad_norm": 2.662776231765747, "learning_rate": 7.456000000000001e-06, "loss": 0.0192, "step": 12721 }, { "epoch": 0.67825345204457, "grad_norm": 2.855269432067871, "learning_rate": 7.455800000000001e-06, "loss": 0.0385, "step": 12722 }, { "epoch": 0.678306765474223, "grad_norm": 1.3235708475112915, "learning_rate": 7.4556e-06, "loss": 0.0113, "step": 12723 }, { "epoch": 0.6783600789038758, "grad_norm": 2.3106327056884766, "learning_rate": 7.455400000000001e-06, "loss": 0.0148, "step": 12724 }, { "epoch": 0.6784133923335288, "grad_norm": 1.581385612487793, "learning_rate": 7.4552e-06, "loss": 0.0175, "step": 12725 }, { "epoch": 0.6784667057631818, "grad_norm": 3.7643375396728516, "learning_rate": 7.4550000000000015e-06, "loss": 0.0419, "step": 12726 }, { "epoch": 0.6785200191928347, "grad_norm": 0.9260759949684143, "learning_rate": 7.454800000000001e-06, "loss": 0.0067, "step": 12727 }, { "epoch": 0.6785733326224876, "grad_norm": 2.2865149974823, "learning_rate": 7.454600000000001e-06, "loss": 0.0166, "step": 12728 }, { "epoch": 0.6786266460521405, "grad_norm": 1.6746180057525635, "learning_rate": 7.4544e-06, "loss": 0.0266, "step": 12729 }, { "epoch": 0.6786799594817935, "grad_norm": 2.8194117546081543, "learning_rate": 7.4542000000000006e-06, "loss": 0.0471, "step": 12730 }, { "epoch": 0.6787332729114464, "grad_norm": 3.1202948093414307, "learning_rate": 7.454e-06, "loss": 0.0396, "step": 12731 }, { "epoch": 0.6787865863410993, "grad_norm": 1.8567677736282349, "learning_rate": 7.453800000000001e-06, "loss": 0.0156, "step": 12732 }, { "epoch": 0.6788398997707522, "grad_norm": 0.8298006057739258, "learning_rate": 7.453600000000001e-06, "loss": 0.0077, "step": 12733 }, { "epoch": 0.6788932132004052, "grad_norm": 0.7484642267227173, "learning_rate": 7.4534000000000005e-06, "loss": 0.0177, "step": 12734 }, { "epoch": 0.6789465266300582, "grad_norm": 2.2882509231567383, "learning_rate": 7.4532e-06, "loss": 0.0385, "step": 12735 }, { "epoch": 0.678999840059711, "grad_norm": 2.0761687755584717, "learning_rate": 7.4530000000000005e-06, "loss": 0.0329, "step": 12736 }, { "epoch": 0.679053153489364, "grad_norm": 1.4644798040390015, "learning_rate": 7.452800000000001e-06, "loss": 0.028, "step": 12737 }, { "epoch": 0.6791064669190169, "grad_norm": 2.202030897140503, "learning_rate": 7.452600000000001e-06, "loss": 0.0383, "step": 12738 }, { "epoch": 0.6791597803486699, "grad_norm": 3.523125410079956, "learning_rate": 7.452400000000001e-06, "loss": 0.0627, "step": 12739 }, { "epoch": 0.6792130937783227, "grad_norm": 2.430039644241333, "learning_rate": 7.4522e-06, "loss": 0.0362, "step": 12740 }, { "epoch": 0.6792664072079757, "grad_norm": 4.7751383781433105, "learning_rate": 7.452e-06, "loss": 0.0369, "step": 12741 }, { "epoch": 0.6793197206376286, "grad_norm": 3.6650402545928955, "learning_rate": 7.4518e-06, "loss": 0.072, "step": 12742 }, { "epoch": 0.6793730340672816, "grad_norm": 1.0623334646224976, "learning_rate": 7.451600000000001e-06, "loss": 0.017, "step": 12743 }, { "epoch": 0.6794263474969344, "grad_norm": 1.7485649585723877, "learning_rate": 7.451400000000001e-06, "loss": 0.0121, "step": 12744 }, { "epoch": 0.6794796609265874, "grad_norm": 4.454556941986084, "learning_rate": 7.451200000000001e-06, "loss": 0.0345, "step": 12745 }, { "epoch": 0.6795329743562404, "grad_norm": 1.6306906938552856, "learning_rate": 7.451e-06, "loss": 0.0184, "step": 12746 }, { "epoch": 0.6795862877858933, "grad_norm": 2.909133195877075, "learning_rate": 7.4508e-06, "loss": 0.028, "step": 12747 }, { "epoch": 0.6796396012155462, "grad_norm": 2.2588343620300293, "learning_rate": 7.4506e-06, "loss": 0.0297, "step": 12748 }, { "epoch": 0.6796929146451991, "grad_norm": 2.1532135009765625, "learning_rate": 7.450400000000001e-06, "loss": 0.024, "step": 12749 }, { "epoch": 0.6797462280748521, "grad_norm": 1.5655300617218018, "learning_rate": 7.450200000000001e-06, "loss": 0.0097, "step": 12750 }, { "epoch": 0.679799541504505, "grad_norm": 3.1863608360290527, "learning_rate": 7.450000000000001e-06, "loss": 0.071, "step": 12751 }, { "epoch": 0.6798528549341579, "grad_norm": 2.2601959705352783, "learning_rate": 7.4498e-06, "loss": 0.0397, "step": 12752 }, { "epoch": 0.6799061683638108, "grad_norm": 1.916415810585022, "learning_rate": 7.4496e-06, "loss": 0.0163, "step": 12753 }, { "epoch": 0.6799594817934638, "grad_norm": 1.5781104564666748, "learning_rate": 7.449400000000001e-06, "loss": 0.0264, "step": 12754 }, { "epoch": 0.6800127952231168, "grad_norm": 1.475677490234375, "learning_rate": 7.449200000000001e-06, "loss": 0.0141, "step": 12755 }, { "epoch": 0.6800661086527696, "grad_norm": 2.6372604370117188, "learning_rate": 7.449000000000001e-06, "loss": 0.0581, "step": 12756 }, { "epoch": 0.6801194220824226, "grad_norm": 3.874785900115967, "learning_rate": 7.448800000000001e-06, "loss": 0.0386, "step": 12757 }, { "epoch": 0.6801727355120755, "grad_norm": 1.272276520729065, "learning_rate": 7.4486e-06, "loss": 0.0106, "step": 12758 }, { "epoch": 0.6802260489417284, "grad_norm": 1.0903311967849731, "learning_rate": 7.4484e-06, "loss": 0.0183, "step": 12759 }, { "epoch": 0.6802793623713813, "grad_norm": 2.1649162769317627, "learning_rate": 7.448200000000001e-06, "loss": 0.042, "step": 12760 }, { "epoch": 0.6803326758010343, "grad_norm": 2.959218740463257, "learning_rate": 7.4480000000000005e-06, "loss": 0.0528, "step": 12761 }, { "epoch": 0.6803859892306872, "grad_norm": 3.5506045818328857, "learning_rate": 7.447800000000001e-06, "loss": 0.0429, "step": 12762 }, { "epoch": 0.6804393026603401, "grad_norm": 2.9972169399261475, "learning_rate": 7.4476000000000005e-06, "loss": 0.0303, "step": 12763 }, { "epoch": 0.6804926160899931, "grad_norm": 4.034194469451904, "learning_rate": 7.4474e-06, "loss": 0.106, "step": 12764 }, { "epoch": 0.680545929519646, "grad_norm": 1.4481710195541382, "learning_rate": 7.447200000000001e-06, "loss": 0.0178, "step": 12765 }, { "epoch": 0.680599242949299, "grad_norm": 5.60412073135376, "learning_rate": 7.447000000000001e-06, "loss": 0.0408, "step": 12766 }, { "epoch": 0.6806525563789518, "grad_norm": 1.5354986190795898, "learning_rate": 7.4468000000000004e-06, "loss": 0.0355, "step": 12767 }, { "epoch": 0.6807058698086048, "grad_norm": 5.981777667999268, "learning_rate": 7.446600000000001e-06, "loss": 0.0656, "step": 12768 }, { "epoch": 0.6807591832382577, "grad_norm": 3.9658491611480713, "learning_rate": 7.4464e-06, "loss": 0.0428, "step": 12769 }, { "epoch": 0.6808124966679107, "grad_norm": 1.5067038536071777, "learning_rate": 7.4462e-06, "loss": 0.0375, "step": 12770 }, { "epoch": 0.6808658100975635, "grad_norm": 2.6848816871643066, "learning_rate": 7.446000000000001e-06, "loss": 0.046, "step": 12771 }, { "epoch": 0.6809191235272165, "grad_norm": 1.7309199571609497, "learning_rate": 7.445800000000001e-06, "loss": 0.019, "step": 12772 }, { "epoch": 0.6809724369568695, "grad_norm": 1.168480396270752, "learning_rate": 7.4456e-06, "loss": 0.013, "step": 12773 }, { "epoch": 0.6810257503865224, "grad_norm": 1.6136811971664429, "learning_rate": 7.445400000000001e-06, "loss": 0.0247, "step": 12774 }, { "epoch": 0.6810790638161753, "grad_norm": 2.0055618286132812, "learning_rate": 7.4452e-06, "loss": 0.0484, "step": 12775 }, { "epoch": 0.6811323772458282, "grad_norm": 1.6046051979064941, "learning_rate": 7.445000000000001e-06, "loss": 0.0249, "step": 12776 }, { "epoch": 0.6811856906754812, "grad_norm": 2.1753036975860596, "learning_rate": 7.444800000000001e-06, "loss": 0.0182, "step": 12777 }, { "epoch": 0.681239004105134, "grad_norm": 1.4095804691314697, "learning_rate": 7.444600000000001e-06, "loss": 0.0258, "step": 12778 }, { "epoch": 0.681292317534787, "grad_norm": 1.5020196437835693, "learning_rate": 7.4444e-06, "loss": 0.0408, "step": 12779 }, { "epoch": 0.6813456309644399, "grad_norm": 2.5574188232421875, "learning_rate": 7.444200000000001e-06, "loss": 0.0377, "step": 12780 }, { "epoch": 0.6813989443940929, "grad_norm": 1.977178931236267, "learning_rate": 7.444e-06, "loss": 0.0308, "step": 12781 }, { "epoch": 0.6814522578237459, "grad_norm": 3.9262397289276123, "learning_rate": 7.443800000000001e-06, "loss": 0.0483, "step": 12782 }, { "epoch": 0.6815055712533987, "grad_norm": 2.6284005641937256, "learning_rate": 7.443600000000001e-06, "loss": 0.0368, "step": 12783 }, { "epoch": 0.6815588846830517, "grad_norm": 2.3830792903900146, "learning_rate": 7.443400000000001e-06, "loss": 0.0346, "step": 12784 }, { "epoch": 0.6816121981127046, "grad_norm": 1.265230655670166, "learning_rate": 7.4432e-06, "loss": 0.021, "step": 12785 }, { "epoch": 0.6816655115423575, "grad_norm": 3.6295504570007324, "learning_rate": 7.4430000000000006e-06, "loss": 0.074, "step": 12786 }, { "epoch": 0.6817188249720104, "grad_norm": 3.265815496444702, "learning_rate": 7.442800000000001e-06, "loss": 0.0312, "step": 12787 }, { "epoch": 0.6817721384016634, "grad_norm": 1.8423353433609009, "learning_rate": 7.4426000000000005e-06, "loss": 0.029, "step": 12788 }, { "epoch": 0.6818254518313163, "grad_norm": 1.7460726499557495, "learning_rate": 7.442400000000001e-06, "loss": 0.0304, "step": 12789 }, { "epoch": 0.6818787652609692, "grad_norm": 3.093858242034912, "learning_rate": 7.4422000000000005e-06, "loss": 0.0536, "step": 12790 }, { "epoch": 0.6819320786906222, "grad_norm": 3.422628164291382, "learning_rate": 7.442e-06, "loss": 0.0418, "step": 12791 }, { "epoch": 0.6819853921202751, "grad_norm": 1.2205427885055542, "learning_rate": 7.4418000000000005e-06, "loss": 0.0137, "step": 12792 }, { "epoch": 0.6820387055499281, "grad_norm": 2.279637575149536, "learning_rate": 7.441600000000001e-06, "loss": 0.0302, "step": 12793 }, { "epoch": 0.6820920189795809, "grad_norm": 0.8614098429679871, "learning_rate": 7.4414000000000004e-06, "loss": 0.0061, "step": 12794 }, { "epoch": 0.6821453324092339, "grad_norm": 2.0004873275756836, "learning_rate": 7.441200000000001e-06, "loss": 0.028, "step": 12795 }, { "epoch": 0.6821986458388868, "grad_norm": 3.177168846130371, "learning_rate": 7.441e-06, "loss": 0.0445, "step": 12796 }, { "epoch": 0.6822519592685398, "grad_norm": 1.070841670036316, "learning_rate": 7.4408e-06, "loss": 0.0132, "step": 12797 }, { "epoch": 0.6823052726981926, "grad_norm": 1.4058040380477905, "learning_rate": 7.440600000000001e-06, "loss": 0.0156, "step": 12798 }, { "epoch": 0.6823585861278456, "grad_norm": 2.17988657951355, "learning_rate": 7.440400000000001e-06, "loss": 0.022, "step": 12799 }, { "epoch": 0.6824118995574985, "grad_norm": 3.2696287631988525, "learning_rate": 7.4402e-06, "loss": 0.0508, "step": 12800 }, { "epoch": 0.6824652129871515, "grad_norm": 2.5092639923095703, "learning_rate": 7.440000000000001e-06, "loss": 0.0502, "step": 12801 }, { "epoch": 0.6825185264168044, "grad_norm": 1.5582386255264282, "learning_rate": 7.4398e-06, "loss": 0.0408, "step": 12802 }, { "epoch": 0.6825718398464573, "grad_norm": 2.150330066680908, "learning_rate": 7.4396e-06, "loss": 0.0469, "step": 12803 }, { "epoch": 0.6826251532761103, "grad_norm": 2.7749462127685547, "learning_rate": 7.439400000000001e-06, "loss": 0.0302, "step": 12804 }, { "epoch": 0.6826784667057632, "grad_norm": 1.4477503299713135, "learning_rate": 7.439200000000001e-06, "loss": 0.0102, "step": 12805 }, { "epoch": 0.6827317801354161, "grad_norm": 3.379817008972168, "learning_rate": 7.439e-06, "loss": 0.0474, "step": 12806 }, { "epoch": 0.682785093565069, "grad_norm": 4.032907009124756, "learning_rate": 7.438800000000001e-06, "loss": 0.0602, "step": 12807 }, { "epoch": 0.682838406994722, "grad_norm": 2.1679351329803467, "learning_rate": 7.4386e-06, "loss": 0.0212, "step": 12808 }, { "epoch": 0.6828917204243748, "grad_norm": 1.610105037689209, "learning_rate": 7.4384e-06, "loss": 0.0189, "step": 12809 }, { "epoch": 0.6829450338540278, "grad_norm": 4.063612461090088, "learning_rate": 7.438200000000001e-06, "loss": 0.0198, "step": 12810 }, { "epoch": 0.6829983472836808, "grad_norm": 1.9420573711395264, "learning_rate": 7.438000000000001e-06, "loss": 0.0162, "step": 12811 }, { "epoch": 0.6830516607133337, "grad_norm": 1.428196668624878, "learning_rate": 7.4378e-06, "loss": 0.0331, "step": 12812 }, { "epoch": 0.6831049741429867, "grad_norm": 1.4634901285171509, "learning_rate": 7.437600000000001e-06, "loss": 0.0324, "step": 12813 }, { "epoch": 0.6831582875726395, "grad_norm": 2.7768335342407227, "learning_rate": 7.4374e-06, "loss": 0.0558, "step": 12814 }, { "epoch": 0.6832116010022925, "grad_norm": 2.2720229625701904, "learning_rate": 7.437200000000001e-06, "loss": 0.0352, "step": 12815 }, { "epoch": 0.6832649144319454, "grad_norm": 2.261195659637451, "learning_rate": 7.437000000000001e-06, "loss": 0.0406, "step": 12816 }, { "epoch": 0.6833182278615983, "grad_norm": 2.199939489364624, "learning_rate": 7.4368000000000005e-06, "loss": 0.0276, "step": 12817 }, { "epoch": 0.6833715412912512, "grad_norm": 2.033097982406616, "learning_rate": 7.4366e-06, "loss": 0.0353, "step": 12818 }, { "epoch": 0.6834248547209042, "grad_norm": 1.2841535806655884, "learning_rate": 7.4364000000000005e-06, "loss": 0.017, "step": 12819 }, { "epoch": 0.6834781681505572, "grad_norm": 1.3630726337432861, "learning_rate": 7.4362e-06, "loss": 0.0316, "step": 12820 }, { "epoch": 0.68353148158021, "grad_norm": 2.8319473266601562, "learning_rate": 7.436000000000001e-06, "loss": 0.0598, "step": 12821 }, { "epoch": 0.683584795009863, "grad_norm": 4.100600242614746, "learning_rate": 7.435800000000001e-06, "loss": 0.07, "step": 12822 }, { "epoch": 0.6836381084395159, "grad_norm": 2.661574125289917, "learning_rate": 7.4356000000000004e-06, "loss": 0.0483, "step": 12823 }, { "epoch": 0.6836914218691689, "grad_norm": 2.0897772312164307, "learning_rate": 7.4354e-06, "loss": 0.0621, "step": 12824 }, { "epoch": 0.6837447352988217, "grad_norm": 2.7788937091827393, "learning_rate": 7.4352e-06, "loss": 0.0285, "step": 12825 }, { "epoch": 0.6837980487284747, "grad_norm": 3.1801140308380127, "learning_rate": 7.435000000000001e-06, "loss": 0.041, "step": 12826 }, { "epoch": 0.6838513621581276, "grad_norm": 1.925727367401123, "learning_rate": 7.434800000000001e-06, "loss": 0.062, "step": 12827 }, { "epoch": 0.6839046755877806, "grad_norm": 4.026963710784912, "learning_rate": 7.434600000000001e-06, "loss": 0.037, "step": 12828 }, { "epoch": 0.6839579890174335, "grad_norm": 1.8955233097076416, "learning_rate": 7.4344e-06, "loss": 0.0231, "step": 12829 }, { "epoch": 0.6840113024470864, "grad_norm": 2.059875726699829, "learning_rate": 7.4342e-06, "loss": 0.0306, "step": 12830 }, { "epoch": 0.6840646158767394, "grad_norm": 1.5124207735061646, "learning_rate": 7.434e-06, "loss": 0.0171, "step": 12831 }, { "epoch": 0.6841179293063923, "grad_norm": 1.2382057905197144, "learning_rate": 7.433800000000001e-06, "loss": 0.0244, "step": 12832 }, { "epoch": 0.6841712427360452, "grad_norm": 1.3867578506469727, "learning_rate": 7.433600000000001e-06, "loss": 0.0238, "step": 12833 }, { "epoch": 0.6842245561656981, "grad_norm": 2.8815765380859375, "learning_rate": 7.433400000000001e-06, "loss": 0.0562, "step": 12834 }, { "epoch": 0.6842778695953511, "grad_norm": 2.1448795795440674, "learning_rate": 7.4332e-06, "loss": 0.0398, "step": 12835 }, { "epoch": 0.684331183025004, "grad_norm": 3.0214715003967285, "learning_rate": 7.433e-06, "loss": 0.0585, "step": 12836 }, { "epoch": 0.6843844964546569, "grad_norm": 1.1916178464889526, "learning_rate": 7.432800000000001e-06, "loss": 0.0109, "step": 12837 }, { "epoch": 0.6844378098843099, "grad_norm": 1.1631689071655273, "learning_rate": 7.432600000000001e-06, "loss": 0.0147, "step": 12838 }, { "epoch": 0.6844911233139628, "grad_norm": 1.785457968711853, "learning_rate": 7.432400000000001e-06, "loss": 0.0346, "step": 12839 }, { "epoch": 0.6845444367436158, "grad_norm": 2.149054527282715, "learning_rate": 7.432200000000001e-06, "loss": 0.0475, "step": 12840 }, { "epoch": 0.6845977501732686, "grad_norm": 1.0820940732955933, "learning_rate": 7.432e-06, "loss": 0.01, "step": 12841 }, { "epoch": 0.6846510636029216, "grad_norm": 1.38462495803833, "learning_rate": 7.4318e-06, "loss": 0.0207, "step": 12842 }, { "epoch": 0.6847043770325745, "grad_norm": 3.2704529762268066, "learning_rate": 7.431600000000001e-06, "loss": 0.0561, "step": 12843 }, { "epoch": 0.6847576904622275, "grad_norm": 1.3945318460464478, "learning_rate": 7.4314000000000005e-06, "loss": 0.0251, "step": 12844 }, { "epoch": 0.6848110038918803, "grad_norm": 2.0603718757629395, "learning_rate": 7.431200000000001e-06, "loss": 0.0285, "step": 12845 }, { "epoch": 0.6848643173215333, "grad_norm": 1.2503751516342163, "learning_rate": 7.4310000000000005e-06, "loss": 0.0215, "step": 12846 }, { "epoch": 0.6849176307511862, "grad_norm": 3.5993082523345947, "learning_rate": 7.4308e-06, "loss": 0.0559, "step": 12847 }, { "epoch": 0.6849709441808391, "grad_norm": 1.8740638494491577, "learning_rate": 7.430600000000001e-06, "loss": 0.0416, "step": 12848 }, { "epoch": 0.6850242576104921, "grad_norm": 3.3300552368164062, "learning_rate": 7.430400000000001e-06, "loss": 0.0268, "step": 12849 }, { "epoch": 0.685077571040145, "grad_norm": 3.223299264907837, "learning_rate": 7.4302000000000005e-06, "loss": 0.0351, "step": 12850 }, { "epoch": 0.685130884469798, "grad_norm": 1.7004563808441162, "learning_rate": 7.430000000000001e-06, "loss": 0.0131, "step": 12851 }, { "epoch": 0.6851841978994508, "grad_norm": 1.866666316986084, "learning_rate": 7.4298e-06, "loss": 0.0302, "step": 12852 }, { "epoch": 0.6852375113291038, "grad_norm": 2.7407801151275635, "learning_rate": 7.4296e-06, "loss": 0.0247, "step": 12853 }, { "epoch": 0.6852908247587567, "grad_norm": 1.8849527835845947, "learning_rate": 7.429400000000001e-06, "loss": 0.0427, "step": 12854 }, { "epoch": 0.6853441381884097, "grad_norm": 2.5712974071502686, "learning_rate": 7.429200000000001e-06, "loss": 0.0178, "step": 12855 }, { "epoch": 0.6853974516180625, "grad_norm": 1.4918382167816162, "learning_rate": 7.429e-06, "loss": 0.0406, "step": 12856 }, { "epoch": 0.6854507650477155, "grad_norm": 2.6646320819854736, "learning_rate": 7.428800000000001e-06, "loss": 0.033, "step": 12857 }, { "epoch": 0.6855040784773685, "grad_norm": 2.655778646469116, "learning_rate": 7.4286e-06, "loss": 0.0484, "step": 12858 }, { "epoch": 0.6855573919070214, "grad_norm": 4.179896831512451, "learning_rate": 7.4284e-06, "loss": 0.0637, "step": 12859 }, { "epoch": 0.6856107053366743, "grad_norm": 1.7382984161376953, "learning_rate": 7.428200000000001e-06, "loss": 0.0454, "step": 12860 }, { "epoch": 0.6856640187663272, "grad_norm": 1.666979193687439, "learning_rate": 7.428000000000001e-06, "loss": 0.0287, "step": 12861 }, { "epoch": 0.6857173321959802, "grad_norm": 2.981205940246582, "learning_rate": 7.4278e-06, "loss": 0.0538, "step": 12862 }, { "epoch": 0.6857706456256331, "grad_norm": 1.6757501363754272, "learning_rate": 7.427600000000001e-06, "loss": 0.0233, "step": 12863 }, { "epoch": 0.685823959055286, "grad_norm": 3.0038204193115234, "learning_rate": 7.4274e-06, "loss": 0.0347, "step": 12864 }, { "epoch": 0.6858772724849389, "grad_norm": 2.8965914249420166, "learning_rate": 7.4272000000000015e-06, "loss": 0.0516, "step": 12865 }, { "epoch": 0.6859305859145919, "grad_norm": 2.759342670440674, "learning_rate": 7.427000000000001e-06, "loss": 0.0283, "step": 12866 }, { "epoch": 0.6859838993442449, "grad_norm": 1.55155348777771, "learning_rate": 7.426800000000001e-06, "loss": 0.0313, "step": 12867 }, { "epoch": 0.6860372127738977, "grad_norm": 1.450681447982788, "learning_rate": 7.4266e-06, "loss": 0.0224, "step": 12868 }, { "epoch": 0.6860905262035507, "grad_norm": 1.7206158638000488, "learning_rate": 7.426400000000001e-06, "loss": 0.028, "step": 12869 }, { "epoch": 0.6861438396332036, "grad_norm": 1.7866648435592651, "learning_rate": 7.4262e-06, "loss": 0.0283, "step": 12870 }, { "epoch": 0.6861971530628566, "grad_norm": 2.4598283767700195, "learning_rate": 7.426000000000001e-06, "loss": 0.0484, "step": 12871 }, { "epoch": 0.6862504664925094, "grad_norm": 1.135756254196167, "learning_rate": 7.425800000000001e-06, "loss": 0.0162, "step": 12872 }, { "epoch": 0.6863037799221624, "grad_norm": 1.327002763748169, "learning_rate": 7.4256000000000005e-06, "loss": 0.0144, "step": 12873 }, { "epoch": 0.6863570933518153, "grad_norm": 2.74086332321167, "learning_rate": 7.4254e-06, "loss": 0.042, "step": 12874 }, { "epoch": 0.6864104067814683, "grad_norm": 2.253472328186035, "learning_rate": 7.4252000000000005e-06, "loss": 0.0269, "step": 12875 }, { "epoch": 0.6864637202111212, "grad_norm": 0.6816102862358093, "learning_rate": 7.425000000000001e-06, "loss": 0.0088, "step": 12876 }, { "epoch": 0.6865170336407741, "grad_norm": 1.3682920932769775, "learning_rate": 7.424800000000001e-06, "loss": 0.0145, "step": 12877 }, { "epoch": 0.6865703470704271, "grad_norm": 2.1332361698150635, "learning_rate": 7.424600000000001e-06, "loss": 0.0363, "step": 12878 }, { "epoch": 0.68662366050008, "grad_norm": 2.5860378742218018, "learning_rate": 7.4244000000000004e-06, "loss": 0.0426, "step": 12879 }, { "epoch": 0.6866769739297329, "grad_norm": 1.4054293632507324, "learning_rate": 7.4242e-06, "loss": 0.0211, "step": 12880 }, { "epoch": 0.6867302873593858, "grad_norm": 2.0532336235046387, "learning_rate": 7.424e-06, "loss": 0.0414, "step": 12881 }, { "epoch": 0.6867836007890388, "grad_norm": 2.872842311859131, "learning_rate": 7.423800000000001e-06, "loss": 0.0531, "step": 12882 }, { "epoch": 0.6868369142186916, "grad_norm": 2.6762149333953857, "learning_rate": 7.423600000000001e-06, "loss": 0.045, "step": 12883 }, { "epoch": 0.6868902276483446, "grad_norm": 3.038710594177246, "learning_rate": 7.423400000000001e-06, "loss": 0.0448, "step": 12884 }, { "epoch": 0.6869435410779976, "grad_norm": 1.336155891418457, "learning_rate": 7.4232e-06, "loss": 0.0243, "step": 12885 }, { "epoch": 0.6869968545076505, "grad_norm": 0.6626954078674316, "learning_rate": 7.423e-06, "loss": 0.0052, "step": 12886 }, { "epoch": 0.6870501679373034, "grad_norm": 1.8428651094436646, "learning_rate": 7.422800000000001e-06, "loss": 0.0217, "step": 12887 }, { "epoch": 0.6871034813669563, "grad_norm": 1.778867483139038, "learning_rate": 7.422600000000001e-06, "loss": 0.012, "step": 12888 }, { "epoch": 0.6871567947966093, "grad_norm": 4.881721019744873, "learning_rate": 7.422400000000001e-06, "loss": 0.0208, "step": 12889 }, { "epoch": 0.6872101082262622, "grad_norm": 1.7477110624313354, "learning_rate": 7.422200000000001e-06, "loss": 0.0136, "step": 12890 }, { "epoch": 0.6872634216559151, "grad_norm": 3.144522190093994, "learning_rate": 7.422e-06, "loss": 0.0469, "step": 12891 }, { "epoch": 0.687316735085568, "grad_norm": 1.6287498474121094, "learning_rate": 7.4218e-06, "loss": 0.0175, "step": 12892 }, { "epoch": 0.687370048515221, "grad_norm": 2.083606719970703, "learning_rate": 7.421600000000001e-06, "loss": 0.0224, "step": 12893 }, { "epoch": 0.687423361944874, "grad_norm": 2.620530128479004, "learning_rate": 7.421400000000001e-06, "loss": 0.0296, "step": 12894 }, { "epoch": 0.6874766753745268, "grad_norm": 4.066081523895264, "learning_rate": 7.421200000000001e-06, "loss": 0.035, "step": 12895 }, { "epoch": 0.6875299888041798, "grad_norm": 1.8332631587982178, "learning_rate": 7.421000000000001e-06, "loss": 0.0337, "step": 12896 }, { "epoch": 0.6875833022338327, "grad_norm": 1.4750622510910034, "learning_rate": 7.4208e-06, "loss": 0.0105, "step": 12897 }, { "epoch": 0.6876366156634857, "grad_norm": 2.715113639831543, "learning_rate": 7.4206000000000014e-06, "loss": 0.0257, "step": 12898 }, { "epoch": 0.6876899290931385, "grad_norm": 2.6189911365509033, "learning_rate": 7.420400000000001e-06, "loss": 0.0577, "step": 12899 }, { "epoch": 0.6877432425227915, "grad_norm": 1.4045681953430176, "learning_rate": 7.4202000000000005e-06, "loss": 0.015, "step": 12900 }, { "epoch": 0.6877965559524444, "grad_norm": 2.7249410152435303, "learning_rate": 7.420000000000001e-06, "loss": 0.0315, "step": 12901 }, { "epoch": 0.6878498693820974, "grad_norm": 4.2035932540893555, "learning_rate": 7.4198000000000005e-06, "loss": 0.0527, "step": 12902 }, { "epoch": 0.6879031828117502, "grad_norm": 3.239980697631836, "learning_rate": 7.4196e-06, "loss": 0.0303, "step": 12903 }, { "epoch": 0.6879564962414032, "grad_norm": 2.0472700595855713, "learning_rate": 7.419400000000001e-06, "loss": 0.0224, "step": 12904 }, { "epoch": 0.6880098096710562, "grad_norm": 6.582554340362549, "learning_rate": 7.419200000000001e-06, "loss": 0.054, "step": 12905 }, { "epoch": 0.688063123100709, "grad_norm": 1.0592904090881348, "learning_rate": 7.4190000000000005e-06, "loss": 0.0158, "step": 12906 }, { "epoch": 0.688116436530362, "grad_norm": 2.205453395843506, "learning_rate": 7.418800000000001e-06, "loss": 0.0445, "step": 12907 }, { "epoch": 0.6881697499600149, "grad_norm": 2.006564140319824, "learning_rate": 7.4186000000000004e-06, "loss": 0.0196, "step": 12908 }, { "epoch": 0.6882230633896679, "grad_norm": 1.5798228979110718, "learning_rate": 7.418400000000001e-06, "loss": 0.0146, "step": 12909 }, { "epoch": 0.6882763768193207, "grad_norm": 3.0948808193206787, "learning_rate": 7.418200000000001e-06, "loss": 0.0442, "step": 12910 }, { "epoch": 0.6883296902489737, "grad_norm": 2.183537244796753, "learning_rate": 7.418000000000001e-06, "loss": 0.0401, "step": 12911 }, { "epoch": 0.6883830036786266, "grad_norm": 2.964984893798828, "learning_rate": 7.4178e-06, "loss": 0.0604, "step": 12912 }, { "epoch": 0.6884363171082796, "grad_norm": 1.5008964538574219, "learning_rate": 7.417600000000001e-06, "loss": 0.0559, "step": 12913 }, { "epoch": 0.6884896305379326, "grad_norm": 3.178771734237671, "learning_rate": 7.4174e-06, "loss": 0.0333, "step": 12914 }, { "epoch": 0.6885429439675854, "grad_norm": 1.7921059131622314, "learning_rate": 7.417200000000001e-06, "loss": 0.0257, "step": 12915 }, { "epoch": 0.6885962573972384, "grad_norm": 1.6790796518325806, "learning_rate": 7.417000000000001e-06, "loss": 0.0386, "step": 12916 }, { "epoch": 0.6886495708268913, "grad_norm": 1.1379131078720093, "learning_rate": 7.416800000000001e-06, "loss": 0.014, "step": 12917 }, { "epoch": 0.6887028842565442, "grad_norm": 2.9262452125549316, "learning_rate": 7.4166e-06, "loss": 0.0229, "step": 12918 }, { "epoch": 0.6887561976861971, "grad_norm": 1.8043034076690674, "learning_rate": 7.416400000000001e-06, "loss": 0.0199, "step": 12919 }, { "epoch": 0.6888095111158501, "grad_norm": 2.0305261611938477, "learning_rate": 7.4162e-06, "loss": 0.0289, "step": 12920 }, { "epoch": 0.688862824545503, "grad_norm": 2.1255507469177246, "learning_rate": 7.416000000000001e-06, "loss": 0.0308, "step": 12921 }, { "epoch": 0.6889161379751559, "grad_norm": 1.2904934883117676, "learning_rate": 7.415800000000001e-06, "loss": 0.0145, "step": 12922 }, { "epoch": 0.6889694514048089, "grad_norm": 2.2643628120422363, "learning_rate": 7.415600000000001e-06, "loss": 0.0302, "step": 12923 }, { "epoch": 0.6890227648344618, "grad_norm": 2.2812681198120117, "learning_rate": 7.4154e-06, "loss": 0.0429, "step": 12924 }, { "epoch": 0.6890760782641148, "grad_norm": 2.407604455947876, "learning_rate": 7.415200000000001e-06, "loss": 0.0555, "step": 12925 }, { "epoch": 0.6891293916937676, "grad_norm": 1.7706325054168701, "learning_rate": 7.415000000000001e-06, "loss": 0.0597, "step": 12926 }, { "epoch": 0.6891827051234206, "grad_norm": 1.412310004234314, "learning_rate": 7.4148000000000006e-06, "loss": 0.0253, "step": 12927 }, { "epoch": 0.6892360185530735, "grad_norm": 1.2669179439544678, "learning_rate": 7.414600000000001e-06, "loss": 0.0134, "step": 12928 }, { "epoch": 0.6892893319827265, "grad_norm": 2.1955337524414062, "learning_rate": 7.4144000000000005e-06, "loss": 0.0494, "step": 12929 }, { "epoch": 0.6893426454123793, "grad_norm": 3.1307687759399414, "learning_rate": 7.4142e-06, "loss": 0.0586, "step": 12930 }, { "epoch": 0.6893959588420323, "grad_norm": 2.118408679962158, "learning_rate": 7.4140000000000005e-06, "loss": 0.0364, "step": 12931 }, { "epoch": 0.6894492722716853, "grad_norm": 1.9868499040603638, "learning_rate": 7.413800000000001e-06, "loss": 0.0284, "step": 12932 }, { "epoch": 0.6895025857013382, "grad_norm": 1.8943759202957153, "learning_rate": 7.4136000000000005e-06, "loss": 0.0365, "step": 12933 }, { "epoch": 0.6895558991309911, "grad_norm": 2.2097060680389404, "learning_rate": 7.413400000000001e-06, "loss": 0.0235, "step": 12934 }, { "epoch": 0.689609212560644, "grad_norm": 1.3792226314544678, "learning_rate": 7.4132000000000004e-06, "loss": 0.0215, "step": 12935 }, { "epoch": 0.689662525990297, "grad_norm": 2.204404592514038, "learning_rate": 7.413e-06, "loss": 0.025, "step": 12936 }, { "epoch": 0.6897158394199498, "grad_norm": 1.0770803689956665, "learning_rate": 7.412800000000001e-06, "loss": 0.0115, "step": 12937 }, { "epoch": 0.6897691528496028, "grad_norm": 2.2884576320648193, "learning_rate": 7.412600000000001e-06, "loss": 0.0277, "step": 12938 }, { "epoch": 0.6898224662792557, "grad_norm": 1.6374644041061401, "learning_rate": 7.4124e-06, "loss": 0.0239, "step": 12939 }, { "epoch": 0.6898757797089087, "grad_norm": 1.800581455230713, "learning_rate": 7.412200000000001e-06, "loss": 0.0233, "step": 12940 }, { "epoch": 0.6899290931385617, "grad_norm": 1.8153523206710815, "learning_rate": 7.412e-06, "loss": 0.0214, "step": 12941 }, { "epoch": 0.6899824065682145, "grad_norm": 1.501691222190857, "learning_rate": 7.4118e-06, "loss": 0.0329, "step": 12942 }, { "epoch": 0.6900357199978675, "grad_norm": 1.7837353944778442, "learning_rate": 7.411600000000001e-06, "loss": 0.0171, "step": 12943 }, { "epoch": 0.6900890334275204, "grad_norm": 1.117881178855896, "learning_rate": 7.411400000000001e-06, "loss": 0.0094, "step": 12944 }, { "epoch": 0.6901423468571733, "grad_norm": 3.1197543144226074, "learning_rate": 7.4112e-06, "loss": 0.0585, "step": 12945 }, { "epoch": 0.6901956602868262, "grad_norm": 2.721693754196167, "learning_rate": 7.411000000000001e-06, "loss": 0.0539, "step": 12946 }, { "epoch": 0.6902489737164792, "grad_norm": 1.5704522132873535, "learning_rate": 7.4108e-06, "loss": 0.0201, "step": 12947 }, { "epoch": 0.6903022871461321, "grad_norm": 2.330552577972412, "learning_rate": 7.4106000000000015e-06, "loss": 0.0225, "step": 12948 }, { "epoch": 0.690355600575785, "grad_norm": 1.125058889389038, "learning_rate": 7.410400000000001e-06, "loss": 0.0207, "step": 12949 }, { "epoch": 0.690408914005438, "grad_norm": 1.692838191986084, "learning_rate": 7.410200000000001e-06, "loss": 0.0196, "step": 12950 }, { "epoch": 0.6904622274350909, "grad_norm": 3.1742615699768066, "learning_rate": 7.41e-06, "loss": 0.0435, "step": 12951 }, { "epoch": 0.6905155408647439, "grad_norm": 3.6164660453796387, "learning_rate": 7.409800000000001e-06, "loss": 0.0303, "step": 12952 }, { "epoch": 0.6905688542943967, "grad_norm": 2.1015188694000244, "learning_rate": 7.4096e-06, "loss": 0.0167, "step": 12953 }, { "epoch": 0.6906221677240497, "grad_norm": 3.0496859550476074, "learning_rate": 7.4094000000000014e-06, "loss": 0.102, "step": 12954 }, { "epoch": 0.6906754811537026, "grad_norm": 1.9089915752410889, "learning_rate": 7.409200000000001e-06, "loss": 0.0203, "step": 12955 }, { "epoch": 0.6907287945833556, "grad_norm": 2.7335503101348877, "learning_rate": 7.4090000000000006e-06, "loss": 0.0405, "step": 12956 }, { "epoch": 0.6907821080130084, "grad_norm": 3.1613659858703613, "learning_rate": 7.4088e-06, "loss": 0.1082, "step": 12957 }, { "epoch": 0.6908354214426614, "grad_norm": 0.6524617075920105, "learning_rate": 7.4086000000000005e-06, "loss": 0.0045, "step": 12958 }, { "epoch": 0.6908887348723143, "grad_norm": 3.1930994987487793, "learning_rate": 7.408400000000001e-06, "loss": 0.0391, "step": 12959 }, { "epoch": 0.6909420483019673, "grad_norm": 2.8578836917877197, "learning_rate": 7.408200000000001e-06, "loss": 0.043, "step": 12960 }, { "epoch": 0.6909953617316202, "grad_norm": 3.885843515396118, "learning_rate": 7.408000000000001e-06, "loss": 0.0322, "step": 12961 }, { "epoch": 0.6910486751612731, "grad_norm": 4.254846096038818, "learning_rate": 7.4078000000000005e-06, "loss": 0.0364, "step": 12962 }, { "epoch": 0.6911019885909261, "grad_norm": 2.1801857948303223, "learning_rate": 7.4076e-06, "loss": 0.0459, "step": 12963 }, { "epoch": 0.691155302020579, "grad_norm": 3.428298234939575, "learning_rate": 7.4074000000000004e-06, "loss": 0.0245, "step": 12964 }, { "epoch": 0.6912086154502319, "grad_norm": 2.4579875469207764, "learning_rate": 7.407200000000001e-06, "loss": 0.0234, "step": 12965 }, { "epoch": 0.6912619288798848, "grad_norm": 3.899834632873535, "learning_rate": 7.407000000000001e-06, "loss": 0.0338, "step": 12966 }, { "epoch": 0.6913152423095378, "grad_norm": 2.8501923084259033, "learning_rate": 7.406800000000001e-06, "loss": 0.0337, "step": 12967 }, { "epoch": 0.6913685557391906, "grad_norm": 2.619563341140747, "learning_rate": 7.4066e-06, "loss": 0.0304, "step": 12968 }, { "epoch": 0.6914218691688436, "grad_norm": 2.933830976486206, "learning_rate": 7.4064e-06, "loss": 0.0282, "step": 12969 }, { "epoch": 0.6914751825984966, "grad_norm": 2.62296724319458, "learning_rate": 7.4062e-06, "loss": 0.0325, "step": 12970 }, { "epoch": 0.6915284960281495, "grad_norm": 1.5862200260162354, "learning_rate": 7.406000000000001e-06, "loss": 0.0188, "step": 12971 }, { "epoch": 0.6915818094578025, "grad_norm": 1.0689544677734375, "learning_rate": 7.405800000000001e-06, "loss": 0.0138, "step": 12972 }, { "epoch": 0.6916351228874553, "grad_norm": 2.486304521560669, "learning_rate": 7.405600000000001e-06, "loss": 0.0357, "step": 12973 }, { "epoch": 0.6916884363171083, "grad_norm": 2.9188902378082275, "learning_rate": 7.4054e-06, "loss": 0.0701, "step": 12974 }, { "epoch": 0.6917417497467612, "grad_norm": 2.033914089202881, "learning_rate": 7.4052e-06, "loss": 0.0473, "step": 12975 }, { "epoch": 0.6917950631764141, "grad_norm": 1.338708758354187, "learning_rate": 7.405000000000001e-06, "loss": 0.0285, "step": 12976 }, { "epoch": 0.691848376606067, "grad_norm": 1.7137998342514038, "learning_rate": 7.404800000000001e-06, "loss": 0.0467, "step": 12977 }, { "epoch": 0.69190169003572, "grad_norm": 1.5424751043319702, "learning_rate": 7.404600000000001e-06, "loss": 0.0087, "step": 12978 }, { "epoch": 0.691955003465373, "grad_norm": 1.7233219146728516, "learning_rate": 7.404400000000001e-06, "loss": 0.0236, "step": 12979 }, { "epoch": 0.6920083168950258, "grad_norm": 0.9474153518676758, "learning_rate": 7.4042e-06, "loss": 0.0103, "step": 12980 }, { "epoch": 0.6920616303246788, "grad_norm": 0.9902778267860413, "learning_rate": 7.404e-06, "loss": 0.0211, "step": 12981 }, { "epoch": 0.6921149437543317, "grad_norm": 1.168084740638733, "learning_rate": 7.403800000000001e-06, "loss": 0.0195, "step": 12982 }, { "epoch": 0.6921682571839847, "grad_norm": 1.3958637714385986, "learning_rate": 7.4036000000000006e-06, "loss": 0.0212, "step": 12983 }, { "epoch": 0.6922215706136375, "grad_norm": 2.019935369491577, "learning_rate": 7.403400000000001e-06, "loss": 0.0509, "step": 12984 }, { "epoch": 0.6922748840432905, "grad_norm": 1.3669556379318237, "learning_rate": 7.4032000000000005e-06, "loss": 0.0284, "step": 12985 }, { "epoch": 0.6923281974729434, "grad_norm": 1.341140866279602, "learning_rate": 7.403e-06, "loss": 0.0253, "step": 12986 }, { "epoch": 0.6923815109025964, "grad_norm": 1.5516209602355957, "learning_rate": 7.402800000000001e-06, "loss": 0.0337, "step": 12987 }, { "epoch": 0.6924348243322493, "grad_norm": 1.5323760509490967, "learning_rate": 7.402600000000001e-06, "loss": 0.0186, "step": 12988 }, { "epoch": 0.6924881377619022, "grad_norm": 0.9204791188240051, "learning_rate": 7.4024000000000005e-06, "loss": 0.0102, "step": 12989 }, { "epoch": 0.6925414511915552, "grad_norm": 1.14034903049469, "learning_rate": 7.402200000000001e-06, "loss": 0.0202, "step": 12990 }, { "epoch": 0.6925947646212081, "grad_norm": 2.6317920684814453, "learning_rate": 7.4020000000000005e-06, "loss": 0.0313, "step": 12991 }, { "epoch": 0.692648078050861, "grad_norm": 2.4192309379577637, "learning_rate": 7.4018e-06, "loss": 0.0437, "step": 12992 }, { "epoch": 0.6927013914805139, "grad_norm": 3.801753520965576, "learning_rate": 7.401600000000001e-06, "loss": 0.0617, "step": 12993 }, { "epoch": 0.6927547049101669, "grad_norm": 1.8364521265029907, "learning_rate": 7.401400000000001e-06, "loss": 0.0372, "step": 12994 }, { "epoch": 0.6928080183398198, "grad_norm": 2.169844150543213, "learning_rate": 7.4012e-06, "loss": 0.0177, "step": 12995 }, { "epoch": 0.6928613317694727, "grad_norm": 2.6051230430603027, "learning_rate": 7.401000000000001e-06, "loss": 0.0366, "step": 12996 }, { "epoch": 0.6929146451991257, "grad_norm": 1.8487701416015625, "learning_rate": 7.4008e-06, "loss": 0.0332, "step": 12997 }, { "epoch": 0.6929679586287786, "grad_norm": 2.337294578552246, "learning_rate": 7.400600000000001e-06, "loss": 0.0215, "step": 12998 }, { "epoch": 0.6930212720584316, "grad_norm": 2.1296567916870117, "learning_rate": 7.400400000000001e-06, "loss": 0.0248, "step": 12999 }, { "epoch": 0.6930745854880844, "grad_norm": 1.0572481155395508, "learning_rate": 7.400200000000001e-06, "loss": 0.0154, "step": 13000 }, { "epoch": 0.6931278989177374, "grad_norm": 1.8737269639968872, "learning_rate": 7.4e-06, "loss": 0.0186, "step": 13001 }, { "epoch": 0.6931812123473903, "grad_norm": 3.292795419692993, "learning_rate": 7.399800000000001e-06, "loss": 0.0296, "step": 13002 }, { "epoch": 0.6932345257770433, "grad_norm": 2.0192360877990723, "learning_rate": 7.3996e-06, "loss": 0.0403, "step": 13003 }, { "epoch": 0.6932878392066961, "grad_norm": 2.7846932411193848, "learning_rate": 7.399400000000001e-06, "loss": 0.0304, "step": 13004 }, { "epoch": 0.6933411526363491, "grad_norm": 1.4202868938446045, "learning_rate": 7.399200000000001e-06, "loss": 0.0259, "step": 13005 }, { "epoch": 0.6933944660660021, "grad_norm": 1.2437536716461182, "learning_rate": 7.399000000000001e-06, "loss": 0.0314, "step": 13006 }, { "epoch": 0.693447779495655, "grad_norm": 1.7378112077713013, "learning_rate": 7.3988e-06, "loss": 0.0186, "step": 13007 }, { "epoch": 0.6935010929253079, "grad_norm": 3.173229217529297, "learning_rate": 7.398600000000001e-06, "loss": 0.0458, "step": 13008 }, { "epoch": 0.6935544063549608, "grad_norm": 1.6457639932632446, "learning_rate": 7.398400000000001e-06, "loss": 0.0242, "step": 13009 }, { "epoch": 0.6936077197846138, "grad_norm": 2.4491004943847656, "learning_rate": 7.398200000000001e-06, "loss": 0.0297, "step": 13010 }, { "epoch": 0.6936610332142666, "grad_norm": 1.052565097808838, "learning_rate": 7.398000000000001e-06, "loss": 0.0157, "step": 13011 }, { "epoch": 0.6937143466439196, "grad_norm": 0.8692950010299683, "learning_rate": 7.3978000000000006e-06, "loss": 0.0092, "step": 13012 }, { "epoch": 0.6937676600735725, "grad_norm": 1.735350489616394, "learning_rate": 7.3976e-06, "loss": 0.0351, "step": 13013 }, { "epoch": 0.6938209735032255, "grad_norm": 2.6679224967956543, "learning_rate": 7.3974000000000005e-06, "loss": 0.0429, "step": 13014 }, { "epoch": 0.6938742869328783, "grad_norm": 2.5041656494140625, "learning_rate": 7.397200000000001e-06, "loss": 0.0562, "step": 13015 }, { "epoch": 0.6939276003625313, "grad_norm": 3.670742988586426, "learning_rate": 7.3970000000000005e-06, "loss": 0.0295, "step": 13016 }, { "epoch": 0.6939809137921843, "grad_norm": 2.3518857955932617, "learning_rate": 7.396800000000001e-06, "loss": 0.0188, "step": 13017 }, { "epoch": 0.6940342272218372, "grad_norm": 2.6221985816955566, "learning_rate": 7.3966000000000005e-06, "loss": 0.0309, "step": 13018 }, { "epoch": 0.6940875406514901, "grad_norm": 2.6887385845184326, "learning_rate": 7.3964e-06, "loss": 0.0415, "step": 13019 }, { "epoch": 0.694140854081143, "grad_norm": 2.2757885456085205, "learning_rate": 7.396200000000001e-06, "loss": 0.038, "step": 13020 }, { "epoch": 0.694194167510796, "grad_norm": 2.115222215652466, "learning_rate": 7.396000000000001e-06, "loss": 0.0404, "step": 13021 }, { "epoch": 0.6942474809404489, "grad_norm": 2.8772177696228027, "learning_rate": 7.3958e-06, "loss": 0.0409, "step": 13022 }, { "epoch": 0.6943007943701018, "grad_norm": 1.5115450620651245, "learning_rate": 7.395600000000001e-06, "loss": 0.0228, "step": 13023 }, { "epoch": 0.6943541077997547, "grad_norm": 2.3126766681671143, "learning_rate": 7.3954e-06, "loss": 0.057, "step": 13024 }, { "epoch": 0.6944074212294077, "grad_norm": 1.5311145782470703, "learning_rate": 7.3952e-06, "loss": 0.0222, "step": 13025 }, { "epoch": 0.6944607346590607, "grad_norm": 2.687161922454834, "learning_rate": 7.395000000000001e-06, "loss": 0.06, "step": 13026 }, { "epoch": 0.6945140480887135, "grad_norm": 2.3432424068450928, "learning_rate": 7.394800000000001e-06, "loss": 0.0333, "step": 13027 }, { "epoch": 0.6945673615183665, "grad_norm": 1.8383874893188477, "learning_rate": 7.3946e-06, "loss": 0.0351, "step": 13028 }, { "epoch": 0.6946206749480194, "grad_norm": 1.1727299690246582, "learning_rate": 7.394400000000001e-06, "loss": 0.0196, "step": 13029 }, { "epoch": 0.6946739883776724, "grad_norm": 2.029370069503784, "learning_rate": 7.3942e-06, "loss": 0.018, "step": 13030 }, { "epoch": 0.6947273018073252, "grad_norm": 1.431966781616211, "learning_rate": 7.394e-06, "loss": 0.0277, "step": 13031 }, { "epoch": 0.6947806152369782, "grad_norm": 5.316248416900635, "learning_rate": 7.393800000000001e-06, "loss": 0.0483, "step": 13032 }, { "epoch": 0.6948339286666311, "grad_norm": 2.844677448272705, "learning_rate": 7.393600000000001e-06, "loss": 0.0244, "step": 13033 }, { "epoch": 0.694887242096284, "grad_norm": 1.8225479125976562, "learning_rate": 7.3934e-06, "loss": 0.0313, "step": 13034 }, { "epoch": 0.694940555525937, "grad_norm": 1.867526888847351, "learning_rate": 7.393200000000001e-06, "loss": 0.0122, "step": 13035 }, { "epoch": 0.6949938689555899, "grad_norm": 1.1300151348114014, "learning_rate": 7.393e-06, "loss": 0.0135, "step": 13036 }, { "epoch": 0.6950471823852429, "grad_norm": 1.6060785055160522, "learning_rate": 7.3928000000000015e-06, "loss": 0.0318, "step": 13037 }, { "epoch": 0.6951004958148957, "grad_norm": 2.066823959350586, "learning_rate": 7.392600000000001e-06, "loss": 0.0222, "step": 13038 }, { "epoch": 0.6951538092445487, "grad_norm": 1.5781209468841553, "learning_rate": 7.3924000000000006e-06, "loss": 0.0222, "step": 13039 }, { "epoch": 0.6952071226742016, "grad_norm": 2.319805860519409, "learning_rate": 7.3922e-06, "loss": 0.0431, "step": 13040 }, { "epoch": 0.6952604361038546, "grad_norm": 1.5308866500854492, "learning_rate": 7.3920000000000005e-06, "loss": 0.0178, "step": 13041 }, { "epoch": 0.6953137495335074, "grad_norm": 1.7666897773742676, "learning_rate": 7.3918e-06, "loss": 0.0138, "step": 13042 }, { "epoch": 0.6953670629631604, "grad_norm": 2.090211868286133, "learning_rate": 7.391600000000001e-06, "loss": 0.0459, "step": 13043 }, { "epoch": 0.6954203763928134, "grad_norm": 1.513023853302002, "learning_rate": 7.391400000000001e-06, "loss": 0.0108, "step": 13044 }, { "epoch": 0.6954736898224663, "grad_norm": 2.6929590702056885, "learning_rate": 7.3912000000000005e-06, "loss": 0.0353, "step": 13045 }, { "epoch": 0.6955270032521192, "grad_norm": 3.771557569503784, "learning_rate": 7.391e-06, "loss": 0.0332, "step": 13046 }, { "epoch": 0.6955803166817721, "grad_norm": 3.5491108894348145, "learning_rate": 7.3908000000000005e-06, "loss": 0.0523, "step": 13047 }, { "epoch": 0.6956336301114251, "grad_norm": 2.1507647037506104, "learning_rate": 7.390600000000001e-06, "loss": 0.0301, "step": 13048 }, { "epoch": 0.695686943541078, "grad_norm": 3.1738085746765137, "learning_rate": 7.390400000000001e-06, "loss": 0.0344, "step": 13049 }, { "epoch": 0.6957402569707309, "grad_norm": 0.977975606918335, "learning_rate": 7.390200000000001e-06, "loss": 0.0081, "step": 13050 }, { "epoch": 0.6957935704003838, "grad_norm": 2.895402669906616, "learning_rate": 7.39e-06, "loss": 0.0516, "step": 13051 }, { "epoch": 0.6958468838300368, "grad_norm": 3.5220704078674316, "learning_rate": 7.3898e-06, "loss": 0.0322, "step": 13052 }, { "epoch": 0.6959001972596898, "grad_norm": 4.402233600616455, "learning_rate": 7.3896e-06, "loss": 0.0596, "step": 13053 }, { "epoch": 0.6959535106893426, "grad_norm": 4.467163562774658, "learning_rate": 7.389400000000001e-06, "loss": 0.0261, "step": 13054 }, { "epoch": 0.6960068241189956, "grad_norm": 2.9128265380859375, "learning_rate": 7.389200000000001e-06, "loss": 0.0394, "step": 13055 }, { "epoch": 0.6960601375486485, "grad_norm": 2.5384466648101807, "learning_rate": 7.389000000000001e-06, "loss": 0.0316, "step": 13056 }, { "epoch": 0.6961134509783015, "grad_norm": 2.6101179122924805, "learning_rate": 7.3888e-06, "loss": 0.0236, "step": 13057 }, { "epoch": 0.6961667644079543, "grad_norm": 2.1604957580566406, "learning_rate": 7.3886e-06, "loss": 0.0286, "step": 13058 }, { "epoch": 0.6962200778376073, "grad_norm": 1.7992957830429077, "learning_rate": 7.388400000000001e-06, "loss": 0.0228, "step": 13059 }, { "epoch": 0.6962733912672602, "grad_norm": 2.1908342838287354, "learning_rate": 7.388200000000001e-06, "loss": 0.0247, "step": 13060 }, { "epoch": 0.6963267046969132, "grad_norm": 2.096442937850952, "learning_rate": 7.388000000000001e-06, "loss": 0.0491, "step": 13061 }, { "epoch": 0.6963800181265661, "grad_norm": 1.7213010787963867, "learning_rate": 7.387800000000001e-06, "loss": 0.0279, "step": 13062 }, { "epoch": 0.696433331556219, "grad_norm": 1.1213384866714478, "learning_rate": 7.3876e-06, "loss": 0.0099, "step": 13063 }, { "epoch": 0.696486644985872, "grad_norm": 1.7918022871017456, "learning_rate": 7.3874e-06, "loss": 0.0371, "step": 13064 }, { "epoch": 0.6965399584155249, "grad_norm": 2.8212718963623047, "learning_rate": 7.387200000000001e-06, "loss": 0.0332, "step": 13065 }, { "epoch": 0.6965932718451778, "grad_norm": 1.5645496845245361, "learning_rate": 7.387000000000001e-06, "loss": 0.024, "step": 13066 }, { "epoch": 0.6966465852748307, "grad_norm": 3.364637851715088, "learning_rate": 7.386800000000001e-06, "loss": 0.0416, "step": 13067 }, { "epoch": 0.6966998987044837, "grad_norm": 1.704366683959961, "learning_rate": 7.3866000000000006e-06, "loss": 0.0244, "step": 13068 }, { "epoch": 0.6967532121341365, "grad_norm": 2.609172821044922, "learning_rate": 7.3864e-06, "loss": 0.0432, "step": 13069 }, { "epoch": 0.6968065255637895, "grad_norm": 3.990872621536255, "learning_rate": 7.386200000000001e-06, "loss": 0.0526, "step": 13070 }, { "epoch": 0.6968598389934424, "grad_norm": 4.8496222496032715, "learning_rate": 7.386000000000001e-06, "loss": 0.0615, "step": 13071 }, { "epoch": 0.6969131524230954, "grad_norm": 1.5790430307388306, "learning_rate": 7.3858000000000005e-06, "loss": 0.0238, "step": 13072 }, { "epoch": 0.6969664658527484, "grad_norm": 4.261083126068115, "learning_rate": 7.385600000000001e-06, "loss": 0.0616, "step": 13073 }, { "epoch": 0.6970197792824012, "grad_norm": 5.269283294677734, "learning_rate": 7.3854000000000005e-06, "loss": 0.0454, "step": 13074 }, { "epoch": 0.6970730927120542, "grad_norm": 2.250046968460083, "learning_rate": 7.3852e-06, "loss": 0.0403, "step": 13075 }, { "epoch": 0.6971264061417071, "grad_norm": 3.2694170475006104, "learning_rate": 7.385000000000001e-06, "loss": 0.0729, "step": 13076 }, { "epoch": 0.69717971957136, "grad_norm": 2.407599449157715, "learning_rate": 7.384800000000001e-06, "loss": 0.0551, "step": 13077 }, { "epoch": 0.6972330330010129, "grad_norm": 1.3219573497772217, "learning_rate": 7.3846e-06, "loss": 0.0224, "step": 13078 }, { "epoch": 0.6972863464306659, "grad_norm": 2.683711290359497, "learning_rate": 7.384400000000001e-06, "loss": 0.0432, "step": 13079 }, { "epoch": 0.6973396598603188, "grad_norm": 1.8119056224822998, "learning_rate": 7.3842e-06, "loss": 0.0433, "step": 13080 }, { "epoch": 0.6973929732899717, "grad_norm": 1.9397567510604858, "learning_rate": 7.384e-06, "loss": 0.0174, "step": 13081 }, { "epoch": 0.6974462867196247, "grad_norm": 1.2722585201263428, "learning_rate": 7.383800000000001e-06, "loss": 0.0155, "step": 13082 }, { "epoch": 0.6974996001492776, "grad_norm": 1.6575210094451904, "learning_rate": 7.383600000000001e-06, "loss": 0.0396, "step": 13083 }, { "epoch": 0.6975529135789306, "grad_norm": 1.1560851335525513, "learning_rate": 7.3834e-06, "loss": 0.0114, "step": 13084 }, { "epoch": 0.6976062270085834, "grad_norm": 1.785215139389038, "learning_rate": 7.383200000000001e-06, "loss": 0.014, "step": 13085 }, { "epoch": 0.6976595404382364, "grad_norm": 2.1761693954467773, "learning_rate": 7.383e-06, "loss": 0.0357, "step": 13086 }, { "epoch": 0.6977128538678893, "grad_norm": 2.0785186290740967, "learning_rate": 7.382800000000001e-06, "loss": 0.0195, "step": 13087 }, { "epoch": 0.6977661672975423, "grad_norm": 3.2321298122406006, "learning_rate": 7.382600000000001e-06, "loss": 0.0573, "step": 13088 }, { "epoch": 0.6978194807271951, "grad_norm": 4.343263626098633, "learning_rate": 7.382400000000001e-06, "loss": 0.0593, "step": 13089 }, { "epoch": 0.6978727941568481, "grad_norm": 2.4293034076690674, "learning_rate": 7.3822e-06, "loss": 0.0296, "step": 13090 }, { "epoch": 0.6979261075865011, "grad_norm": 1.7523273229599, "learning_rate": 7.382000000000001e-06, "loss": 0.0294, "step": 13091 }, { "epoch": 0.697979421016154, "grad_norm": 1.3512080907821655, "learning_rate": 7.3818e-06, "loss": 0.0334, "step": 13092 }, { "epoch": 0.6980327344458069, "grad_norm": 2.270120143890381, "learning_rate": 7.381600000000001e-06, "loss": 0.0452, "step": 13093 }, { "epoch": 0.6980860478754598, "grad_norm": 2.431572437286377, "learning_rate": 7.381400000000001e-06, "loss": 0.0568, "step": 13094 }, { "epoch": 0.6981393613051128, "grad_norm": 3.515232801437378, "learning_rate": 7.381200000000001e-06, "loss": 0.078, "step": 13095 }, { "epoch": 0.6981926747347657, "grad_norm": 1.1338019371032715, "learning_rate": 7.381e-06, "loss": 0.02, "step": 13096 }, { "epoch": 0.6982459881644186, "grad_norm": 2.599730968475342, "learning_rate": 7.3808000000000006e-06, "loss": 0.0339, "step": 13097 }, { "epoch": 0.6982993015940715, "grad_norm": 1.6538207530975342, "learning_rate": 7.380600000000001e-06, "loss": 0.028, "step": 13098 }, { "epoch": 0.6983526150237245, "grad_norm": 2.102696418762207, "learning_rate": 7.3804000000000005e-06, "loss": 0.0285, "step": 13099 }, { "epoch": 0.6984059284533775, "grad_norm": 1.7341042757034302, "learning_rate": 7.380200000000001e-06, "loss": 0.0237, "step": 13100 }, { "epoch": 0.6984592418830303, "grad_norm": 2.194639205932617, "learning_rate": 7.3800000000000005e-06, "loss": 0.0413, "step": 13101 }, { "epoch": 0.6985125553126833, "grad_norm": 1.6861605644226074, "learning_rate": 7.3798e-06, "loss": 0.023, "step": 13102 }, { "epoch": 0.6985658687423362, "grad_norm": 1.3336671590805054, "learning_rate": 7.3796000000000005e-06, "loss": 0.0245, "step": 13103 }, { "epoch": 0.6986191821719892, "grad_norm": 1.3333141803741455, "learning_rate": 7.379400000000001e-06, "loss": 0.0163, "step": 13104 }, { "epoch": 0.698672495601642, "grad_norm": 1.9144288301467896, "learning_rate": 7.3792000000000004e-06, "loss": 0.0305, "step": 13105 }, { "epoch": 0.698725809031295, "grad_norm": 1.6388720273971558, "learning_rate": 7.379000000000001e-06, "loss": 0.0249, "step": 13106 }, { "epoch": 0.6987791224609479, "grad_norm": 3.5235278606414795, "learning_rate": 7.3788e-06, "loss": 0.0394, "step": 13107 }, { "epoch": 0.6988324358906008, "grad_norm": 2.8650434017181396, "learning_rate": 7.3786e-06, "loss": 0.0408, "step": 13108 }, { "epoch": 0.6988857493202538, "grad_norm": 3.2062087059020996, "learning_rate": 7.378400000000001e-06, "loss": 0.0578, "step": 13109 }, { "epoch": 0.6989390627499067, "grad_norm": 1.6747359037399292, "learning_rate": 7.378200000000001e-06, "loss": 0.034, "step": 13110 }, { "epoch": 0.6989923761795597, "grad_norm": 1.8887346982955933, "learning_rate": 7.378e-06, "loss": 0.0204, "step": 13111 }, { "epoch": 0.6990456896092125, "grad_norm": 1.0714882612228394, "learning_rate": 7.377800000000001e-06, "loss": 0.0159, "step": 13112 }, { "epoch": 0.6990990030388655, "grad_norm": 1.5912880897521973, "learning_rate": 7.3776e-06, "loss": 0.0518, "step": 13113 }, { "epoch": 0.6991523164685184, "grad_norm": 0.8047403693199158, "learning_rate": 7.3774e-06, "loss": 0.0054, "step": 13114 }, { "epoch": 0.6992056298981714, "grad_norm": 2.0231730937957764, "learning_rate": 7.377200000000001e-06, "loss": 0.0344, "step": 13115 }, { "epoch": 0.6992589433278242, "grad_norm": 1.8561075925827026, "learning_rate": 7.377000000000001e-06, "loss": 0.0216, "step": 13116 }, { "epoch": 0.6993122567574772, "grad_norm": 4.322659015655518, "learning_rate": 7.3768e-06, "loss": 0.0365, "step": 13117 }, { "epoch": 0.6993655701871302, "grad_norm": 2.1782455444335938, "learning_rate": 7.376600000000001e-06, "loss": 0.04, "step": 13118 }, { "epoch": 0.6994188836167831, "grad_norm": 2.2700388431549072, "learning_rate": 7.3764e-06, "loss": 0.0405, "step": 13119 }, { "epoch": 0.699472197046436, "grad_norm": 1.280683159828186, "learning_rate": 7.3762000000000015e-06, "loss": 0.0131, "step": 13120 }, { "epoch": 0.6995255104760889, "grad_norm": 1.84596848487854, "learning_rate": 7.376000000000001e-06, "loss": 0.0297, "step": 13121 }, { "epoch": 0.6995788239057419, "grad_norm": 2.2854883670806885, "learning_rate": 7.375800000000001e-06, "loss": 0.037, "step": 13122 }, { "epoch": 0.6996321373353948, "grad_norm": 3.7504847049713135, "learning_rate": 7.3756e-06, "loss": 0.0426, "step": 13123 }, { "epoch": 0.6996854507650477, "grad_norm": 2.433454990386963, "learning_rate": 7.3754000000000006e-06, "loss": 0.0274, "step": 13124 }, { "epoch": 0.6997387641947006, "grad_norm": 2.2598936557769775, "learning_rate": 7.3752e-06, "loss": 0.0207, "step": 13125 }, { "epoch": 0.6997920776243536, "grad_norm": 3.7457711696624756, "learning_rate": 7.375000000000001e-06, "loss": 0.0426, "step": 13126 }, { "epoch": 0.6998453910540064, "grad_norm": 2.5641050338745117, "learning_rate": 7.374800000000001e-06, "loss": 0.0263, "step": 13127 }, { "epoch": 0.6998987044836594, "grad_norm": 2.0567164421081543, "learning_rate": 7.3746000000000005e-06, "loss": 0.0388, "step": 13128 }, { "epoch": 0.6999520179133124, "grad_norm": 4.046850681304932, "learning_rate": 7.3744e-06, "loss": 0.0444, "step": 13129 }, { "epoch": 0.7000053313429653, "grad_norm": 0.8221419453620911, "learning_rate": 7.3742000000000005e-06, "loss": 0.0065, "step": 13130 }, { "epoch": 0.7000586447726183, "grad_norm": 3.3132901191711426, "learning_rate": 7.374000000000001e-06, "loss": 0.0393, "step": 13131 }, { "epoch": 0.7001119582022711, "grad_norm": 1.285150170326233, "learning_rate": 7.373800000000001e-06, "loss": 0.0241, "step": 13132 }, { "epoch": 0.7001652716319241, "grad_norm": 0.8706592321395874, "learning_rate": 7.373600000000001e-06, "loss": 0.0068, "step": 13133 }, { "epoch": 0.700218585061577, "grad_norm": 1.3756767511367798, "learning_rate": 7.3734e-06, "loss": 0.0202, "step": 13134 }, { "epoch": 0.70027189849123, "grad_norm": 2.299377202987671, "learning_rate": 7.3732e-06, "loss": 0.0224, "step": 13135 }, { "epoch": 0.7003252119208828, "grad_norm": 2.1361117362976074, "learning_rate": 7.373e-06, "loss": 0.0405, "step": 13136 }, { "epoch": 0.7003785253505358, "grad_norm": 2.4433600902557373, "learning_rate": 7.372800000000001e-06, "loss": 0.0366, "step": 13137 }, { "epoch": 0.7004318387801888, "grad_norm": 1.374129056930542, "learning_rate": 7.372600000000001e-06, "loss": 0.0156, "step": 13138 }, { "epoch": 0.7004851522098416, "grad_norm": 1.627170443534851, "learning_rate": 7.372400000000001e-06, "loss": 0.0273, "step": 13139 }, { "epoch": 0.7005384656394946, "grad_norm": 2.529015302658081, "learning_rate": 7.3722e-06, "loss": 0.0277, "step": 13140 }, { "epoch": 0.7005917790691475, "grad_norm": 3.3947277069091797, "learning_rate": 7.372e-06, "loss": 0.0743, "step": 13141 }, { "epoch": 0.7006450924988005, "grad_norm": 2.1447842121124268, "learning_rate": 7.3718e-06, "loss": 0.0243, "step": 13142 }, { "epoch": 0.7006984059284533, "grad_norm": 3.1765847206115723, "learning_rate": 7.371600000000001e-06, "loss": 0.0331, "step": 13143 }, { "epoch": 0.7007517193581063, "grad_norm": 1.2681039571762085, "learning_rate": 7.371400000000001e-06, "loss": 0.014, "step": 13144 }, { "epoch": 0.7008050327877592, "grad_norm": 1.56134831905365, "learning_rate": 7.371200000000001e-06, "loss": 0.0293, "step": 13145 }, { "epoch": 0.7008583462174122, "grad_norm": 1.9346262216567993, "learning_rate": 7.371e-06, "loss": 0.0333, "step": 13146 }, { "epoch": 0.7009116596470651, "grad_norm": 2.2208197116851807, "learning_rate": 7.3708e-06, "loss": 0.0413, "step": 13147 }, { "epoch": 0.700964973076718, "grad_norm": 1.3715362548828125, "learning_rate": 7.370600000000001e-06, "loss": 0.0106, "step": 13148 }, { "epoch": 0.701018286506371, "grad_norm": 2.485806941986084, "learning_rate": 7.370400000000001e-06, "loss": 0.0274, "step": 13149 }, { "epoch": 0.7010715999360239, "grad_norm": 2.2432847023010254, "learning_rate": 7.370200000000001e-06, "loss": 0.0332, "step": 13150 }, { "epoch": 0.7011249133656768, "grad_norm": 1.7315974235534668, "learning_rate": 7.370000000000001e-06, "loss": 0.0254, "step": 13151 }, { "epoch": 0.7011782267953297, "grad_norm": 2.13156795501709, "learning_rate": 7.3698e-06, "loss": 0.037, "step": 13152 }, { "epoch": 0.7012315402249827, "grad_norm": 3.062340259552002, "learning_rate": 7.3696e-06, "loss": 0.0552, "step": 13153 }, { "epoch": 0.7012848536546356, "grad_norm": 1.9697163105010986, "learning_rate": 7.369400000000001e-06, "loss": 0.0293, "step": 13154 }, { "epoch": 0.7013381670842885, "grad_norm": 1.7361958026885986, "learning_rate": 7.3692000000000005e-06, "loss": 0.0237, "step": 13155 }, { "epoch": 0.7013914805139415, "grad_norm": 1.5911976099014282, "learning_rate": 7.369000000000001e-06, "loss": 0.0156, "step": 13156 }, { "epoch": 0.7014447939435944, "grad_norm": 1.5050549507141113, "learning_rate": 7.3688000000000005e-06, "loss": 0.0174, "step": 13157 }, { "epoch": 0.7014981073732474, "grad_norm": 1.8833286762237549, "learning_rate": 7.3686e-06, "loss": 0.0332, "step": 13158 }, { "epoch": 0.7015514208029002, "grad_norm": 2.680837631225586, "learning_rate": 7.368400000000001e-06, "loss": 0.0407, "step": 13159 }, { "epoch": 0.7016047342325532, "grad_norm": 3.748220205307007, "learning_rate": 7.368200000000001e-06, "loss": 0.0465, "step": 13160 }, { "epoch": 0.7016580476622061, "grad_norm": 5.034882545471191, "learning_rate": 7.3680000000000004e-06, "loss": 0.0698, "step": 13161 }, { "epoch": 0.701711361091859, "grad_norm": 2.1319072246551514, "learning_rate": 7.367800000000001e-06, "loss": 0.0252, "step": 13162 }, { "epoch": 0.7017646745215119, "grad_norm": 3.9151859283447266, "learning_rate": 7.3676e-06, "loss": 0.065, "step": 13163 }, { "epoch": 0.7018179879511649, "grad_norm": 1.7644716501235962, "learning_rate": 7.3674e-06, "loss": 0.0248, "step": 13164 }, { "epoch": 0.7018713013808179, "grad_norm": 2.1138296127319336, "learning_rate": 7.367200000000001e-06, "loss": 0.0327, "step": 13165 }, { "epoch": 0.7019246148104707, "grad_norm": 2.0902960300445557, "learning_rate": 7.367000000000001e-06, "loss": 0.0212, "step": 13166 }, { "epoch": 0.7019779282401237, "grad_norm": 2.0146424770355225, "learning_rate": 7.3668e-06, "loss": 0.0249, "step": 13167 }, { "epoch": 0.7020312416697766, "grad_norm": 2.2609925270080566, "learning_rate": 7.366600000000001e-06, "loss": 0.0571, "step": 13168 }, { "epoch": 0.7020845550994296, "grad_norm": 1.7089343070983887, "learning_rate": 7.3664e-06, "loss": 0.0235, "step": 13169 }, { "epoch": 0.7021378685290824, "grad_norm": 1.7599141597747803, "learning_rate": 7.366200000000001e-06, "loss": 0.0627, "step": 13170 }, { "epoch": 0.7021911819587354, "grad_norm": 2.494140148162842, "learning_rate": 7.366000000000001e-06, "loss": 0.0469, "step": 13171 }, { "epoch": 0.7022444953883883, "grad_norm": 1.7113709449768066, "learning_rate": 7.365800000000001e-06, "loss": 0.0295, "step": 13172 }, { "epoch": 0.7022978088180413, "grad_norm": 1.8609873056411743, "learning_rate": 7.3656e-06, "loss": 0.0521, "step": 13173 }, { "epoch": 0.7023511222476942, "grad_norm": 1.7397706508636475, "learning_rate": 7.365400000000001e-06, "loss": 0.017, "step": 13174 }, { "epoch": 0.7024044356773471, "grad_norm": 0.6784273386001587, "learning_rate": 7.3652e-06, "loss": 0.0079, "step": 13175 }, { "epoch": 0.7024577491070001, "grad_norm": 2.193767786026001, "learning_rate": 7.365000000000001e-06, "loss": 0.0378, "step": 13176 }, { "epoch": 0.702511062536653, "grad_norm": 4.168502330780029, "learning_rate": 7.364800000000001e-06, "loss": 0.0205, "step": 13177 }, { "epoch": 0.7025643759663059, "grad_norm": 2.319181203842163, "learning_rate": 7.364600000000001e-06, "loss": 0.0284, "step": 13178 }, { "epoch": 0.7026176893959588, "grad_norm": 1.6992050409317017, "learning_rate": 7.3644e-06, "loss": 0.036, "step": 13179 }, { "epoch": 0.7026710028256118, "grad_norm": 3.035599946975708, "learning_rate": 7.3642000000000006e-06, "loss": 0.0653, "step": 13180 }, { "epoch": 0.7027243162552647, "grad_norm": 1.8498787879943848, "learning_rate": 7.364000000000001e-06, "loss": 0.0139, "step": 13181 }, { "epoch": 0.7027776296849176, "grad_norm": 1.3179954290390015, "learning_rate": 7.3638000000000005e-06, "loss": 0.0184, "step": 13182 }, { "epoch": 0.7028309431145705, "grad_norm": 4.542709827423096, "learning_rate": 7.363600000000001e-06, "loss": 0.0264, "step": 13183 }, { "epoch": 0.7028842565442235, "grad_norm": 0.6794881224632263, "learning_rate": 7.3634000000000005e-06, "loss": 0.0064, "step": 13184 }, { "epoch": 0.7029375699738765, "grad_norm": 2.858074903488159, "learning_rate": 7.3632e-06, "loss": 0.0486, "step": 13185 }, { "epoch": 0.7029908834035293, "grad_norm": 0.8378704786300659, "learning_rate": 7.3630000000000005e-06, "loss": 0.0082, "step": 13186 }, { "epoch": 0.7030441968331823, "grad_norm": 2.105952739715576, "learning_rate": 7.362800000000001e-06, "loss": 0.0228, "step": 13187 }, { "epoch": 0.7030975102628352, "grad_norm": 1.8176155090332031, "learning_rate": 7.3626000000000005e-06, "loss": 0.0441, "step": 13188 }, { "epoch": 0.7031508236924882, "grad_norm": 1.467065453529358, "learning_rate": 7.362400000000001e-06, "loss": 0.0137, "step": 13189 }, { "epoch": 0.703204137122141, "grad_norm": 2.41741681098938, "learning_rate": 7.3622000000000004e-06, "loss": 0.0384, "step": 13190 }, { "epoch": 0.703257450551794, "grad_norm": 1.8476015329360962, "learning_rate": 7.362e-06, "loss": 0.012, "step": 13191 }, { "epoch": 0.7033107639814469, "grad_norm": 2.408500909805298, "learning_rate": 7.3618e-06, "loss": 0.0588, "step": 13192 }, { "epoch": 0.7033640774110999, "grad_norm": 1.622986078262329, "learning_rate": 7.361600000000001e-06, "loss": 0.0283, "step": 13193 }, { "epoch": 0.7034173908407528, "grad_norm": 1.5850040912628174, "learning_rate": 7.3614e-06, "loss": 0.0182, "step": 13194 }, { "epoch": 0.7034707042704057, "grad_norm": 2.898740530014038, "learning_rate": 7.361200000000001e-06, "loss": 0.0776, "step": 13195 }, { "epoch": 0.7035240177000587, "grad_norm": 2.838683843612671, "learning_rate": 7.361e-06, "loss": 0.0317, "step": 13196 }, { "epoch": 0.7035773311297115, "grad_norm": 1.44868004322052, "learning_rate": 7.3608e-06, "loss": 0.0497, "step": 13197 }, { "epoch": 0.7036306445593645, "grad_norm": 3.777070999145508, "learning_rate": 7.360600000000001e-06, "loss": 0.0332, "step": 13198 }, { "epoch": 0.7036839579890174, "grad_norm": 2.3413779735565186, "learning_rate": 7.360400000000001e-06, "loss": 0.0495, "step": 13199 }, { "epoch": 0.7037372714186704, "grad_norm": 3.9923863410949707, "learning_rate": 7.3602e-06, "loss": 0.0437, "step": 13200 }, { "epoch": 0.7037905848483232, "grad_norm": 2.434220552444458, "learning_rate": 7.360000000000001e-06, "loss": 0.0338, "step": 13201 }, { "epoch": 0.7038438982779762, "grad_norm": 1.1921604871749878, "learning_rate": 7.3598e-06, "loss": 0.0128, "step": 13202 }, { "epoch": 0.7038972117076292, "grad_norm": 1.3911229372024536, "learning_rate": 7.3596e-06, "loss": 0.037, "step": 13203 }, { "epoch": 0.7039505251372821, "grad_norm": 1.114815592765808, "learning_rate": 7.359400000000001e-06, "loss": 0.0089, "step": 13204 }, { "epoch": 0.704003838566935, "grad_norm": 2.7328052520751953, "learning_rate": 7.359200000000001e-06, "loss": 0.0484, "step": 13205 }, { "epoch": 0.7040571519965879, "grad_norm": 2.041942596435547, "learning_rate": 7.359e-06, "loss": 0.031, "step": 13206 }, { "epoch": 0.7041104654262409, "grad_norm": 2.8871002197265625, "learning_rate": 7.358800000000001e-06, "loss": 0.0766, "step": 13207 }, { "epoch": 0.7041637788558938, "grad_norm": 1.8879187107086182, "learning_rate": 7.3586e-06, "loss": 0.0521, "step": 13208 }, { "epoch": 0.7042170922855467, "grad_norm": 1.391268253326416, "learning_rate": 7.358400000000001e-06, "loss": 0.0233, "step": 13209 }, { "epoch": 0.7042704057151996, "grad_norm": 0.983391523361206, "learning_rate": 7.358200000000001e-06, "loss": 0.0297, "step": 13210 }, { "epoch": 0.7043237191448526, "grad_norm": 2.5856709480285645, "learning_rate": 7.3580000000000005e-06, "loss": 0.032, "step": 13211 }, { "epoch": 0.7043770325745056, "grad_norm": 1.4832383394241333, "learning_rate": 7.3578e-06, "loss": 0.0326, "step": 13212 }, { "epoch": 0.7044303460041584, "grad_norm": 2.4509811401367188, "learning_rate": 7.3576000000000005e-06, "loss": 0.0687, "step": 13213 }, { "epoch": 0.7044836594338114, "grad_norm": 1.8980005979537964, "learning_rate": 7.3574e-06, "loss": 0.0297, "step": 13214 }, { "epoch": 0.7045369728634643, "grad_norm": 1.7667473554611206, "learning_rate": 7.357200000000001e-06, "loss": 0.0171, "step": 13215 }, { "epoch": 0.7045902862931173, "grad_norm": 2.2587387561798096, "learning_rate": 7.357000000000001e-06, "loss": 0.042, "step": 13216 }, { "epoch": 0.7046435997227701, "grad_norm": 1.3595099449157715, "learning_rate": 7.3568000000000004e-06, "loss": 0.0242, "step": 13217 }, { "epoch": 0.7046969131524231, "grad_norm": 3.344395637512207, "learning_rate": 7.3566e-06, "loss": 0.0571, "step": 13218 }, { "epoch": 0.704750226582076, "grad_norm": 1.798271894454956, "learning_rate": 7.3564e-06, "loss": 0.014, "step": 13219 }, { "epoch": 0.704803540011729, "grad_norm": 1.708171010017395, "learning_rate": 7.356200000000001e-06, "loss": 0.0318, "step": 13220 }, { "epoch": 0.7048568534413819, "grad_norm": 1.680849552154541, "learning_rate": 7.356000000000001e-06, "loss": 0.0257, "step": 13221 }, { "epoch": 0.7049101668710348, "grad_norm": 1.1392766237258911, "learning_rate": 7.355800000000001e-06, "loss": 0.0244, "step": 13222 }, { "epoch": 0.7049634803006878, "grad_norm": 1.1189974546432495, "learning_rate": 7.3556e-06, "loss": 0.0158, "step": 13223 }, { "epoch": 0.7050167937303407, "grad_norm": 1.9694173336029053, "learning_rate": 7.3554e-06, "loss": 0.0377, "step": 13224 }, { "epoch": 0.7050701071599936, "grad_norm": 1.3389289379119873, "learning_rate": 7.3552e-06, "loss": 0.0136, "step": 13225 }, { "epoch": 0.7051234205896465, "grad_norm": 2.254481315612793, "learning_rate": 7.355000000000001e-06, "loss": 0.0409, "step": 13226 }, { "epoch": 0.7051767340192995, "grad_norm": 0.8928671479225159, "learning_rate": 7.354800000000001e-06, "loss": 0.0194, "step": 13227 }, { "epoch": 0.7052300474489523, "grad_norm": 1.7844620943069458, "learning_rate": 7.354600000000001e-06, "loss": 0.0346, "step": 13228 }, { "epoch": 0.7052833608786053, "grad_norm": 2.1939568519592285, "learning_rate": 7.3544e-06, "loss": 0.0327, "step": 13229 }, { "epoch": 0.7053366743082583, "grad_norm": 3.1618688106536865, "learning_rate": 7.3542e-06, "loss": 0.0432, "step": 13230 }, { "epoch": 0.7053899877379112, "grad_norm": 1.5381741523742676, "learning_rate": 7.354000000000001e-06, "loss": 0.0375, "step": 13231 }, { "epoch": 0.7054433011675642, "grad_norm": 2.1238009929656982, "learning_rate": 7.353800000000001e-06, "loss": 0.0505, "step": 13232 }, { "epoch": 0.705496614597217, "grad_norm": 1.9501103162765503, "learning_rate": 7.353600000000001e-06, "loss": 0.0289, "step": 13233 }, { "epoch": 0.70554992802687, "grad_norm": 1.71009361743927, "learning_rate": 7.353400000000001e-06, "loss": 0.022, "step": 13234 }, { "epoch": 0.7056032414565229, "grad_norm": 3.8883090019226074, "learning_rate": 7.3532e-06, "loss": 0.0381, "step": 13235 }, { "epoch": 0.7056565548861758, "grad_norm": 2.0688493251800537, "learning_rate": 7.353e-06, "loss": 0.0423, "step": 13236 }, { "epoch": 0.7057098683158287, "grad_norm": 1.4040783643722534, "learning_rate": 7.352800000000001e-06, "loss": 0.0139, "step": 13237 }, { "epoch": 0.7057631817454817, "grad_norm": 2.5337204933166504, "learning_rate": 7.3526000000000006e-06, "loss": 0.0484, "step": 13238 }, { "epoch": 0.7058164951751346, "grad_norm": 2.4959278106689453, "learning_rate": 7.352400000000001e-06, "loss": 0.0432, "step": 13239 }, { "epoch": 0.7058698086047875, "grad_norm": 2.2484993934631348, "learning_rate": 7.3522000000000005e-06, "loss": 0.0488, "step": 13240 }, { "epoch": 0.7059231220344405, "grad_norm": 1.8615381717681885, "learning_rate": 7.352e-06, "loss": 0.0421, "step": 13241 }, { "epoch": 0.7059764354640934, "grad_norm": 2.6680691242218018, "learning_rate": 7.3518e-06, "loss": 0.0476, "step": 13242 }, { "epoch": 0.7060297488937464, "grad_norm": 1.2515778541564941, "learning_rate": 7.351600000000001e-06, "loss": 0.0081, "step": 13243 }, { "epoch": 0.7060830623233992, "grad_norm": 2.3939592838287354, "learning_rate": 7.3514000000000005e-06, "loss": 0.0423, "step": 13244 }, { "epoch": 0.7061363757530522, "grad_norm": 1.5850142240524292, "learning_rate": 7.351200000000001e-06, "loss": 0.023, "step": 13245 }, { "epoch": 0.7061896891827051, "grad_norm": 0.4779322147369385, "learning_rate": 7.3510000000000004e-06, "loss": 0.0061, "step": 13246 }, { "epoch": 0.7062430026123581, "grad_norm": 1.6779102087020874, "learning_rate": 7.3508e-06, "loss": 0.0334, "step": 13247 }, { "epoch": 0.7062963160420109, "grad_norm": 1.1994985342025757, "learning_rate": 7.350600000000001e-06, "loss": 0.0126, "step": 13248 }, { "epoch": 0.7063496294716639, "grad_norm": 3.155601978302002, "learning_rate": 7.350400000000001e-06, "loss": 0.0671, "step": 13249 }, { "epoch": 0.7064029429013169, "grad_norm": 2.037327527999878, "learning_rate": 7.3502e-06, "loss": 0.029, "step": 13250 }, { "epoch": 0.7064562563309698, "grad_norm": 1.9327179193496704, "learning_rate": 7.350000000000001e-06, "loss": 0.0428, "step": 13251 }, { "epoch": 0.7065095697606227, "grad_norm": 1.6781609058380127, "learning_rate": 7.3498e-06, "loss": 0.0166, "step": 13252 }, { "epoch": 0.7065628831902756, "grad_norm": 1.9995821714401245, "learning_rate": 7.3496e-06, "loss": 0.0282, "step": 13253 }, { "epoch": 0.7066161966199286, "grad_norm": 2.264920711517334, "learning_rate": 7.349400000000001e-06, "loss": 0.0474, "step": 13254 }, { "epoch": 0.7066695100495815, "grad_norm": 2.207646369934082, "learning_rate": 7.349200000000001e-06, "loss": 0.0669, "step": 13255 }, { "epoch": 0.7067228234792344, "grad_norm": 0.8413581252098083, "learning_rate": 7.349e-06, "loss": 0.0237, "step": 13256 }, { "epoch": 0.7067761369088873, "grad_norm": 1.2318731546401978, "learning_rate": 7.348800000000001e-06, "loss": 0.0096, "step": 13257 }, { "epoch": 0.7068294503385403, "grad_norm": 1.3809560537338257, "learning_rate": 7.3486e-06, "loss": 0.0157, "step": 13258 }, { "epoch": 0.7068827637681933, "grad_norm": 1.7313618659973145, "learning_rate": 7.3484000000000015e-06, "loss": 0.0087, "step": 13259 }, { "epoch": 0.7069360771978461, "grad_norm": 1.8004406690597534, "learning_rate": 7.348200000000001e-06, "loss": 0.0134, "step": 13260 }, { "epoch": 0.7069893906274991, "grad_norm": 3.120271921157837, "learning_rate": 7.348000000000001e-06, "loss": 0.0376, "step": 13261 }, { "epoch": 0.707042704057152, "grad_norm": 2.66064715385437, "learning_rate": 7.3478e-06, "loss": 0.0365, "step": 13262 }, { "epoch": 0.707096017486805, "grad_norm": 2.5204885005950928, "learning_rate": 7.347600000000001e-06, "loss": 0.0588, "step": 13263 }, { "epoch": 0.7071493309164578, "grad_norm": 1.6453653573989868, "learning_rate": 7.3474e-06, "loss": 0.0333, "step": 13264 }, { "epoch": 0.7072026443461108, "grad_norm": 2.3543965816497803, "learning_rate": 7.347200000000001e-06, "loss": 0.0531, "step": 13265 }, { "epoch": 0.7072559577757637, "grad_norm": 2.227039098739624, "learning_rate": 7.347000000000001e-06, "loss": 0.0395, "step": 13266 }, { "epoch": 0.7073092712054166, "grad_norm": 3.4207334518432617, "learning_rate": 7.3468000000000005e-06, "loss": 0.0531, "step": 13267 }, { "epoch": 0.7073625846350696, "grad_norm": 2.781804323196411, "learning_rate": 7.3466e-06, "loss": 0.0183, "step": 13268 }, { "epoch": 0.7074158980647225, "grad_norm": 3.6014959812164307, "learning_rate": 7.3464000000000005e-06, "loss": 0.0421, "step": 13269 }, { "epoch": 0.7074692114943755, "grad_norm": 1.6569068431854248, "learning_rate": 7.346200000000001e-06, "loss": 0.0334, "step": 13270 }, { "epoch": 0.7075225249240283, "grad_norm": 2.218902349472046, "learning_rate": 7.346000000000001e-06, "loss": 0.0222, "step": 13271 }, { "epoch": 0.7075758383536813, "grad_norm": 1.8823890686035156, "learning_rate": 7.345800000000001e-06, "loss": 0.0236, "step": 13272 }, { "epoch": 0.7076291517833342, "grad_norm": 1.2395267486572266, "learning_rate": 7.3456000000000004e-06, "loss": 0.0149, "step": 13273 }, { "epoch": 0.7076824652129872, "grad_norm": 4.273445129394531, "learning_rate": 7.3454e-06, "loss": 0.0802, "step": 13274 }, { "epoch": 0.70773577864264, "grad_norm": 1.8503779172897339, "learning_rate": 7.3452e-06, "loss": 0.0234, "step": 13275 }, { "epoch": 0.707789092072293, "grad_norm": 2.4023053646087646, "learning_rate": 7.345000000000001e-06, "loss": 0.0519, "step": 13276 }, { "epoch": 0.707842405501946, "grad_norm": 2.003077983856201, "learning_rate": 7.344800000000001e-06, "loss": 0.014, "step": 13277 }, { "epoch": 0.7078957189315989, "grad_norm": 1.402387261390686, "learning_rate": 7.344600000000001e-06, "loss": 0.0236, "step": 13278 }, { "epoch": 0.7079490323612518, "grad_norm": 1.8449680805206299, "learning_rate": 7.3444e-06, "loss": 0.0252, "step": 13279 }, { "epoch": 0.7080023457909047, "grad_norm": 2.416627883911133, "learning_rate": 7.3442e-06, "loss": 0.0457, "step": 13280 }, { "epoch": 0.7080556592205577, "grad_norm": 2.0464909076690674, "learning_rate": 7.344000000000001e-06, "loss": 0.0342, "step": 13281 }, { "epoch": 0.7081089726502106, "grad_norm": 1.7430613040924072, "learning_rate": 7.343800000000001e-06, "loss": 0.0237, "step": 13282 }, { "epoch": 0.7081622860798635, "grad_norm": 3.8917887210845947, "learning_rate": 7.343600000000001e-06, "loss": 0.0239, "step": 13283 }, { "epoch": 0.7082155995095164, "grad_norm": 3.244541883468628, "learning_rate": 7.343400000000001e-06, "loss": 0.0332, "step": 13284 }, { "epoch": 0.7082689129391694, "grad_norm": 1.9043108224868774, "learning_rate": 7.3432e-06, "loss": 0.035, "step": 13285 }, { "epoch": 0.7083222263688224, "grad_norm": 1.2371755838394165, "learning_rate": 7.343e-06, "loss": 0.0231, "step": 13286 }, { "epoch": 0.7083755397984752, "grad_norm": 2.3718783855438232, "learning_rate": 7.342800000000001e-06, "loss": 0.059, "step": 13287 }, { "epoch": 0.7084288532281282, "grad_norm": 2.127321720123291, "learning_rate": 7.342600000000001e-06, "loss": 0.0408, "step": 13288 }, { "epoch": 0.7084821666577811, "grad_norm": 0.9005014300346375, "learning_rate": 7.342400000000001e-06, "loss": 0.0107, "step": 13289 }, { "epoch": 0.708535480087434, "grad_norm": 1.5064146518707275, "learning_rate": 7.342200000000001e-06, "loss": 0.0436, "step": 13290 }, { "epoch": 0.7085887935170869, "grad_norm": 1.3504458665847778, "learning_rate": 7.342e-06, "loss": 0.0277, "step": 13291 }, { "epoch": 0.7086421069467399, "grad_norm": 2.0539638996124268, "learning_rate": 7.3418000000000014e-06, "loss": 0.0282, "step": 13292 }, { "epoch": 0.7086954203763928, "grad_norm": 2.2248616218566895, "learning_rate": 7.341600000000001e-06, "loss": 0.0547, "step": 13293 }, { "epoch": 0.7087487338060457, "grad_norm": 2.080209970474243, "learning_rate": 7.3414000000000006e-06, "loss": 0.0416, "step": 13294 }, { "epoch": 0.7088020472356986, "grad_norm": 1.6761218309402466, "learning_rate": 7.341200000000001e-06, "loss": 0.026, "step": 13295 }, { "epoch": 0.7088553606653516, "grad_norm": 2.306821823120117, "learning_rate": 7.3410000000000005e-06, "loss": 0.0172, "step": 13296 }, { "epoch": 0.7089086740950046, "grad_norm": 1.7370301485061646, "learning_rate": 7.3408e-06, "loss": 0.0266, "step": 13297 }, { "epoch": 0.7089619875246574, "grad_norm": 2.8323540687561035, "learning_rate": 7.340600000000001e-06, "loss": 0.028, "step": 13298 }, { "epoch": 0.7090153009543104, "grad_norm": 1.879800796508789, "learning_rate": 7.340400000000001e-06, "loss": 0.0317, "step": 13299 }, { "epoch": 0.7090686143839633, "grad_norm": 1.9485639333724976, "learning_rate": 7.3402000000000005e-06, "loss": 0.0159, "step": 13300 }, { "epoch": 0.7091219278136163, "grad_norm": 3.6945548057556152, "learning_rate": 7.340000000000001e-06, "loss": 0.0603, "step": 13301 }, { "epoch": 0.7091752412432691, "grad_norm": 1.046330213546753, "learning_rate": 7.3398000000000004e-06, "loss": 0.0234, "step": 13302 }, { "epoch": 0.7092285546729221, "grad_norm": 1.4933794736862183, "learning_rate": 7.3396e-06, "loss": 0.0126, "step": 13303 }, { "epoch": 0.709281868102575, "grad_norm": 4.111859321594238, "learning_rate": 7.339400000000001e-06, "loss": 0.0746, "step": 13304 }, { "epoch": 0.709335181532228, "grad_norm": 3.795544147491455, "learning_rate": 7.339200000000001e-06, "loss": 0.0323, "step": 13305 }, { "epoch": 0.7093884949618809, "grad_norm": 1.989697813987732, "learning_rate": 7.339e-06, "loss": 0.035, "step": 13306 }, { "epoch": 0.7094418083915338, "grad_norm": 3.3073654174804688, "learning_rate": 7.338800000000001e-06, "loss": 0.0457, "step": 13307 }, { "epoch": 0.7094951218211868, "grad_norm": 1.8677613735198975, "learning_rate": 7.3386e-06, "loss": 0.031, "step": 13308 }, { "epoch": 0.7095484352508397, "grad_norm": 1.2854341268539429, "learning_rate": 7.338400000000001e-06, "loss": 0.0237, "step": 13309 }, { "epoch": 0.7096017486804926, "grad_norm": 1.624847412109375, "learning_rate": 7.338200000000001e-06, "loss": 0.0323, "step": 13310 }, { "epoch": 0.7096550621101455, "grad_norm": 3.068089246749878, "learning_rate": 7.338000000000001e-06, "loss": 0.0365, "step": 13311 }, { "epoch": 0.7097083755397985, "grad_norm": 3.413714647293091, "learning_rate": 7.3378e-06, "loss": 0.029, "step": 13312 }, { "epoch": 0.7097616889694514, "grad_norm": 1.27633798122406, "learning_rate": 7.337600000000001e-06, "loss": 0.012, "step": 13313 }, { "epoch": 0.7098150023991043, "grad_norm": 3.455374240875244, "learning_rate": 7.3374e-06, "loss": 0.0467, "step": 13314 }, { "epoch": 0.7098683158287573, "grad_norm": 3.02998948097229, "learning_rate": 7.337200000000001e-06, "loss": 0.0507, "step": 13315 }, { "epoch": 0.7099216292584102, "grad_norm": 1.8077406883239746, "learning_rate": 7.337000000000001e-06, "loss": 0.0324, "step": 13316 }, { "epoch": 0.7099749426880632, "grad_norm": 3.253713607788086, "learning_rate": 7.336800000000001e-06, "loss": 0.0526, "step": 13317 }, { "epoch": 0.710028256117716, "grad_norm": 3.1184189319610596, "learning_rate": 7.3366e-06, "loss": 0.0381, "step": 13318 }, { "epoch": 0.710081569547369, "grad_norm": 1.3248133659362793, "learning_rate": 7.336400000000001e-06, "loss": 0.0385, "step": 13319 }, { "epoch": 0.7101348829770219, "grad_norm": 3.8502330780029297, "learning_rate": 7.336200000000001e-06, "loss": 0.031, "step": 13320 }, { "epoch": 0.7101881964066749, "grad_norm": 2.15358567237854, "learning_rate": 7.3360000000000006e-06, "loss": 0.0246, "step": 13321 }, { "epoch": 0.7102415098363277, "grad_norm": 3.3981552124023438, "learning_rate": 7.335800000000001e-06, "loss": 0.038, "step": 13322 }, { "epoch": 0.7102948232659807, "grad_norm": 3.029726028442383, "learning_rate": 7.3356000000000005e-06, "loss": 0.0507, "step": 13323 }, { "epoch": 0.7103481366956337, "grad_norm": 3.9453718662261963, "learning_rate": 7.3354e-06, "loss": 0.0388, "step": 13324 }, { "epoch": 0.7104014501252865, "grad_norm": 1.5094354152679443, "learning_rate": 7.3352000000000005e-06, "loss": 0.0418, "step": 13325 }, { "epoch": 0.7104547635549395, "grad_norm": 1.4751373529434204, "learning_rate": 7.335000000000001e-06, "loss": 0.0316, "step": 13326 }, { "epoch": 0.7105080769845924, "grad_norm": 3.3889966011047363, "learning_rate": 7.3348000000000005e-06, "loss": 0.0412, "step": 13327 }, { "epoch": 0.7105613904142454, "grad_norm": 1.719254493713379, "learning_rate": 7.334600000000001e-06, "loss": 0.0316, "step": 13328 }, { "epoch": 0.7106147038438982, "grad_norm": 3.3339414596557617, "learning_rate": 7.3344000000000005e-06, "loss": 0.0681, "step": 13329 }, { "epoch": 0.7106680172735512, "grad_norm": 2.080498218536377, "learning_rate": 7.3342e-06, "loss": 0.0347, "step": 13330 }, { "epoch": 0.7107213307032041, "grad_norm": 1.824886679649353, "learning_rate": 7.334000000000001e-06, "loss": 0.0398, "step": 13331 }, { "epoch": 0.7107746441328571, "grad_norm": 2.475252866744995, "learning_rate": 7.333800000000001e-06, "loss": 0.034, "step": 13332 }, { "epoch": 0.71082795756251, "grad_norm": 3.1101245880126953, "learning_rate": 7.3336e-06, "loss": 0.0241, "step": 13333 }, { "epoch": 0.7108812709921629, "grad_norm": 2.877822160720825, "learning_rate": 7.333400000000001e-06, "loss": 0.036, "step": 13334 }, { "epoch": 0.7109345844218159, "grad_norm": 4.10903787612915, "learning_rate": 7.3332e-06, "loss": 0.0612, "step": 13335 }, { "epoch": 0.7109878978514688, "grad_norm": 1.7868971824645996, "learning_rate": 7.333e-06, "loss": 0.0142, "step": 13336 }, { "epoch": 0.7110412112811217, "grad_norm": 3.4888460636138916, "learning_rate": 7.332800000000001e-06, "loss": 0.025, "step": 13337 }, { "epoch": 0.7110945247107746, "grad_norm": 1.1735131740570068, "learning_rate": 7.332600000000001e-06, "loss": 0.0146, "step": 13338 }, { "epoch": 0.7111478381404276, "grad_norm": 1.9565216302871704, "learning_rate": 7.3324e-06, "loss": 0.0387, "step": 13339 }, { "epoch": 0.7112011515700805, "grad_norm": 2.0488054752349854, "learning_rate": 7.332200000000001e-06, "loss": 0.0293, "step": 13340 }, { "epoch": 0.7112544649997334, "grad_norm": 1.5135316848754883, "learning_rate": 7.332e-06, "loss": 0.0122, "step": 13341 }, { "epoch": 0.7113077784293864, "grad_norm": 4.128178596496582, "learning_rate": 7.3318000000000015e-06, "loss": 0.0366, "step": 13342 }, { "epoch": 0.7113610918590393, "grad_norm": 1.8923077583312988, "learning_rate": 7.331600000000001e-06, "loss": 0.031, "step": 13343 }, { "epoch": 0.7114144052886923, "grad_norm": 2.0397677421569824, "learning_rate": 7.331400000000001e-06, "loss": 0.0415, "step": 13344 }, { "epoch": 0.7114677187183451, "grad_norm": 2.3548967838287354, "learning_rate": 7.3312e-06, "loss": 0.0419, "step": 13345 }, { "epoch": 0.7115210321479981, "grad_norm": 2.1229453086853027, "learning_rate": 7.331000000000001e-06, "loss": 0.0438, "step": 13346 }, { "epoch": 0.711574345577651, "grad_norm": 2.474205255508423, "learning_rate": 7.3308e-06, "loss": 0.066, "step": 13347 }, { "epoch": 0.711627659007304, "grad_norm": 1.6392741203308105, "learning_rate": 7.3306000000000014e-06, "loss": 0.0338, "step": 13348 }, { "epoch": 0.7116809724369568, "grad_norm": 1.8467448949813843, "learning_rate": 7.330400000000001e-06, "loss": 0.0232, "step": 13349 }, { "epoch": 0.7117342858666098, "grad_norm": 1.5266426801681519, "learning_rate": 7.3302000000000006e-06, "loss": 0.0334, "step": 13350 }, { "epoch": 0.7117875992962627, "grad_norm": 0.9699523448944092, "learning_rate": 7.33e-06, "loss": 0.0146, "step": 13351 }, { "epoch": 0.7118409127259157, "grad_norm": 1.391359567642212, "learning_rate": 7.3298000000000005e-06, "loss": 0.0173, "step": 13352 }, { "epoch": 0.7118942261555686, "grad_norm": 1.5296823978424072, "learning_rate": 7.3296e-06, "loss": 0.0265, "step": 13353 }, { "epoch": 0.7119475395852215, "grad_norm": 1.2257580757141113, "learning_rate": 7.329400000000001e-06, "loss": 0.0119, "step": 13354 }, { "epoch": 0.7120008530148745, "grad_norm": 1.9615179300308228, "learning_rate": 7.329200000000001e-06, "loss": 0.0245, "step": 13355 }, { "epoch": 0.7120541664445273, "grad_norm": 0.9968147873878479, "learning_rate": 7.3290000000000005e-06, "loss": 0.0084, "step": 13356 }, { "epoch": 0.7121074798741803, "grad_norm": 1.559234619140625, "learning_rate": 7.3288e-06, "loss": 0.0214, "step": 13357 }, { "epoch": 0.7121607933038332, "grad_norm": 0.9999892115592957, "learning_rate": 7.3286000000000004e-06, "loss": 0.0193, "step": 13358 }, { "epoch": 0.7122141067334862, "grad_norm": 1.913469910621643, "learning_rate": 7.328400000000001e-06, "loss": 0.0442, "step": 13359 }, { "epoch": 0.712267420163139, "grad_norm": 2.2945773601531982, "learning_rate": 7.328200000000001e-06, "loss": 0.0395, "step": 13360 }, { "epoch": 0.712320733592792, "grad_norm": 0.9848688840866089, "learning_rate": 7.328000000000001e-06, "loss": 0.0091, "step": 13361 }, { "epoch": 0.712374047022445, "grad_norm": 2.4467647075653076, "learning_rate": 7.3278e-06, "loss": 0.0496, "step": 13362 }, { "epoch": 0.7124273604520979, "grad_norm": 1.4734505414962769, "learning_rate": 7.3276e-06, "loss": 0.0163, "step": 13363 }, { "epoch": 0.7124806738817508, "grad_norm": 0.5975804924964905, "learning_rate": 7.3274e-06, "loss": 0.0057, "step": 13364 }, { "epoch": 0.7125339873114037, "grad_norm": 2.0075061321258545, "learning_rate": 7.327200000000001e-06, "loss": 0.0313, "step": 13365 }, { "epoch": 0.7125873007410567, "grad_norm": 2.753847360610962, "learning_rate": 7.327000000000001e-06, "loss": 0.032, "step": 13366 }, { "epoch": 0.7126406141707096, "grad_norm": 2.469597101211548, "learning_rate": 7.326800000000001e-06, "loss": 0.0243, "step": 13367 }, { "epoch": 0.7126939276003625, "grad_norm": 2.155796527862549, "learning_rate": 7.3266e-06, "loss": 0.0347, "step": 13368 }, { "epoch": 0.7127472410300154, "grad_norm": 3.2329046726226807, "learning_rate": 7.3264e-06, "loss": 0.0296, "step": 13369 }, { "epoch": 0.7128005544596684, "grad_norm": 3.968656539916992, "learning_rate": 7.326200000000001e-06, "loss": 0.0401, "step": 13370 }, { "epoch": 0.7128538678893214, "grad_norm": 4.35931396484375, "learning_rate": 7.326000000000001e-06, "loss": 0.0381, "step": 13371 }, { "epoch": 0.7129071813189742, "grad_norm": 1.8532655239105225, "learning_rate": 7.325800000000001e-06, "loss": 0.0169, "step": 13372 }, { "epoch": 0.7129604947486272, "grad_norm": 4.224330425262451, "learning_rate": 7.325600000000001e-06, "loss": 0.0377, "step": 13373 }, { "epoch": 0.7130138081782801, "grad_norm": 3.5907135009765625, "learning_rate": 7.3254e-06, "loss": 0.0411, "step": 13374 }, { "epoch": 0.7130671216079331, "grad_norm": 3.064113140106201, "learning_rate": 7.3252e-06, "loss": 0.0672, "step": 13375 }, { "epoch": 0.7131204350375859, "grad_norm": 1.3897141218185425, "learning_rate": 7.325000000000001e-06, "loss": 0.0189, "step": 13376 }, { "epoch": 0.7131737484672389, "grad_norm": 2.118272066116333, "learning_rate": 7.324800000000001e-06, "loss": 0.0419, "step": 13377 }, { "epoch": 0.7132270618968918, "grad_norm": 0.7923309803009033, "learning_rate": 7.324600000000001e-06, "loss": 0.0067, "step": 13378 }, { "epoch": 0.7132803753265448, "grad_norm": 1.6928538084030151, "learning_rate": 7.3244000000000005e-06, "loss": 0.0149, "step": 13379 }, { "epoch": 0.7133336887561977, "grad_norm": 2.0826187133789062, "learning_rate": 7.3242e-06, "loss": 0.0351, "step": 13380 }, { "epoch": 0.7133870021858506, "grad_norm": 1.9724730253219604, "learning_rate": 7.324000000000001e-06, "loss": 0.0293, "step": 13381 }, { "epoch": 0.7134403156155036, "grad_norm": 3.0267632007598877, "learning_rate": 7.323800000000001e-06, "loss": 0.0483, "step": 13382 }, { "epoch": 0.7134936290451565, "grad_norm": 1.8776012659072876, "learning_rate": 7.3236000000000005e-06, "loss": 0.0288, "step": 13383 }, { "epoch": 0.7135469424748094, "grad_norm": 3.3679556846618652, "learning_rate": 7.323400000000001e-06, "loss": 0.0539, "step": 13384 }, { "epoch": 0.7136002559044623, "grad_norm": 1.8746472597122192, "learning_rate": 7.3232000000000005e-06, "loss": 0.0314, "step": 13385 }, { "epoch": 0.7136535693341153, "grad_norm": 2.0165069103240967, "learning_rate": 7.323e-06, "loss": 0.023, "step": 13386 }, { "epoch": 0.7137068827637681, "grad_norm": 1.2889022827148438, "learning_rate": 7.322800000000001e-06, "loss": 0.0265, "step": 13387 }, { "epoch": 0.7137601961934211, "grad_norm": 2.4499051570892334, "learning_rate": 7.322600000000001e-06, "loss": 0.0488, "step": 13388 }, { "epoch": 0.7138135096230741, "grad_norm": 3.393590211868286, "learning_rate": 7.3224e-06, "loss": 0.0314, "step": 13389 }, { "epoch": 0.713866823052727, "grad_norm": 2.5061938762664795, "learning_rate": 7.322200000000001e-06, "loss": 0.0119, "step": 13390 }, { "epoch": 0.71392013648238, "grad_norm": 2.62687611579895, "learning_rate": 7.322e-06, "loss": 0.0365, "step": 13391 }, { "epoch": 0.7139734499120328, "grad_norm": 1.900326132774353, "learning_rate": 7.321800000000001e-06, "loss": 0.0174, "step": 13392 }, { "epoch": 0.7140267633416858, "grad_norm": 1.2272578477859497, "learning_rate": 7.321600000000001e-06, "loss": 0.0175, "step": 13393 }, { "epoch": 0.7140800767713387, "grad_norm": 2.6501731872558594, "learning_rate": 7.321400000000001e-06, "loss": 0.0429, "step": 13394 }, { "epoch": 0.7141333902009916, "grad_norm": 1.3926002979278564, "learning_rate": 7.3212e-06, "loss": 0.0347, "step": 13395 }, { "epoch": 0.7141867036306445, "grad_norm": 1.4541388750076294, "learning_rate": 7.321000000000001e-06, "loss": 0.0219, "step": 13396 }, { "epoch": 0.7142400170602975, "grad_norm": 1.4891527891159058, "learning_rate": 7.3208e-06, "loss": 0.0392, "step": 13397 }, { "epoch": 0.7142933304899505, "grad_norm": 1.5794994831085205, "learning_rate": 7.320600000000001e-06, "loss": 0.0214, "step": 13398 }, { "epoch": 0.7143466439196033, "grad_norm": 1.284602165222168, "learning_rate": 7.320400000000001e-06, "loss": 0.0404, "step": 13399 }, { "epoch": 0.7143999573492563, "grad_norm": 3.377680778503418, "learning_rate": 7.320200000000001e-06, "loss": 0.0558, "step": 13400 }, { "epoch": 0.7144532707789092, "grad_norm": 1.9782555103302002, "learning_rate": 7.32e-06, "loss": 0.0255, "step": 13401 }, { "epoch": 0.7145065842085622, "grad_norm": 3.227023124694824, "learning_rate": 7.319800000000001e-06, "loss": 0.0276, "step": 13402 }, { "epoch": 0.714559897638215, "grad_norm": 1.7656155824661255, "learning_rate": 7.319600000000001e-06, "loss": 0.0196, "step": 13403 }, { "epoch": 0.714613211067868, "grad_norm": 1.5856703519821167, "learning_rate": 7.319400000000001e-06, "loss": 0.0252, "step": 13404 }, { "epoch": 0.7146665244975209, "grad_norm": 2.358900547027588, "learning_rate": 7.319200000000001e-06, "loss": 0.0432, "step": 13405 }, { "epoch": 0.7147198379271739, "grad_norm": 3.4672317504882812, "learning_rate": 7.3190000000000006e-06, "loss": 0.0684, "step": 13406 }, { "epoch": 0.7147731513568267, "grad_norm": 1.3182975053787231, "learning_rate": 7.3188e-06, "loss": 0.0182, "step": 13407 }, { "epoch": 0.7148264647864797, "grad_norm": 2.3679561614990234, "learning_rate": 7.3186000000000005e-06, "loss": 0.0414, "step": 13408 }, { "epoch": 0.7148797782161327, "grad_norm": 0.9859485626220703, "learning_rate": 7.318400000000001e-06, "loss": 0.0132, "step": 13409 }, { "epoch": 0.7149330916457856, "grad_norm": 1.3808542490005493, "learning_rate": 7.3182000000000005e-06, "loss": 0.0283, "step": 13410 }, { "epoch": 0.7149864050754385, "grad_norm": 0.8381564021110535, "learning_rate": 7.318000000000001e-06, "loss": 0.0141, "step": 13411 }, { "epoch": 0.7150397185050914, "grad_norm": 0.9754499793052673, "learning_rate": 7.3178000000000005e-06, "loss": 0.0204, "step": 13412 }, { "epoch": 0.7150930319347444, "grad_norm": 2.6903269290924072, "learning_rate": 7.3176e-06, "loss": 0.0254, "step": 13413 }, { "epoch": 0.7151463453643973, "grad_norm": 2.861727237701416, "learning_rate": 7.3174000000000004e-06, "loss": 0.0465, "step": 13414 }, { "epoch": 0.7151996587940502, "grad_norm": 1.9633009433746338, "learning_rate": 7.317200000000001e-06, "loss": 0.033, "step": 13415 }, { "epoch": 0.7152529722237031, "grad_norm": 1.6705355644226074, "learning_rate": 7.317e-06, "loss": 0.0361, "step": 13416 }, { "epoch": 0.7153062856533561, "grad_norm": 2.7738263607025146, "learning_rate": 7.316800000000001e-06, "loss": 0.034, "step": 13417 }, { "epoch": 0.7153595990830091, "grad_norm": 2.51184344291687, "learning_rate": 7.3166e-06, "loss": 0.02, "step": 13418 }, { "epoch": 0.7154129125126619, "grad_norm": 2.241539239883423, "learning_rate": 7.3164e-06, "loss": 0.0625, "step": 13419 }, { "epoch": 0.7154662259423149, "grad_norm": 2.3907623291015625, "learning_rate": 7.316200000000001e-06, "loss": 0.0524, "step": 13420 }, { "epoch": 0.7155195393719678, "grad_norm": 1.6981335878372192, "learning_rate": 7.316000000000001e-06, "loss": 0.0313, "step": 13421 }, { "epoch": 0.7155728528016208, "grad_norm": 1.0992392301559448, "learning_rate": 7.3158e-06, "loss": 0.0158, "step": 13422 }, { "epoch": 0.7156261662312736, "grad_norm": 1.8790709972381592, "learning_rate": 7.315600000000001e-06, "loss": 0.0314, "step": 13423 }, { "epoch": 0.7156794796609266, "grad_norm": 2.0061142444610596, "learning_rate": 7.3154e-06, "loss": 0.0315, "step": 13424 }, { "epoch": 0.7157327930905795, "grad_norm": 1.6859416961669922, "learning_rate": 7.3152e-06, "loss": 0.0123, "step": 13425 }, { "epoch": 0.7157861065202324, "grad_norm": 1.6962242126464844, "learning_rate": 7.315000000000001e-06, "loss": 0.0509, "step": 13426 }, { "epoch": 0.7158394199498854, "grad_norm": 0.7377659678459167, "learning_rate": 7.314800000000001e-06, "loss": 0.0071, "step": 13427 }, { "epoch": 0.7158927333795383, "grad_norm": 1.2419344186782837, "learning_rate": 7.3146e-06, "loss": 0.028, "step": 13428 }, { "epoch": 0.7159460468091913, "grad_norm": 1.6420291662216187, "learning_rate": 7.314400000000001e-06, "loss": 0.0156, "step": 13429 }, { "epoch": 0.7159993602388441, "grad_norm": 1.2328001260757446, "learning_rate": 7.3142e-06, "loss": 0.0228, "step": 13430 }, { "epoch": 0.7160526736684971, "grad_norm": 1.728287696838379, "learning_rate": 7.3140000000000015e-06, "loss": 0.019, "step": 13431 }, { "epoch": 0.71610598709815, "grad_norm": 1.9616672992706299, "learning_rate": 7.313800000000001e-06, "loss": 0.0156, "step": 13432 }, { "epoch": 0.716159300527803, "grad_norm": 4.037044525146484, "learning_rate": 7.313600000000001e-06, "loss": 0.0587, "step": 13433 }, { "epoch": 0.7162126139574558, "grad_norm": 2.883183717727661, "learning_rate": 7.3134e-06, "loss": 0.0361, "step": 13434 }, { "epoch": 0.7162659273871088, "grad_norm": 2.204852342605591, "learning_rate": 7.3132000000000006e-06, "loss": 0.0409, "step": 13435 }, { "epoch": 0.7163192408167618, "grad_norm": 2.037766933441162, "learning_rate": 7.313e-06, "loss": 0.0348, "step": 13436 }, { "epoch": 0.7163725542464147, "grad_norm": 2.372351884841919, "learning_rate": 7.312800000000001e-06, "loss": 0.0427, "step": 13437 }, { "epoch": 0.7164258676760676, "grad_norm": 0.6716464757919312, "learning_rate": 7.312600000000001e-06, "loss": 0.0082, "step": 13438 }, { "epoch": 0.7164791811057205, "grad_norm": 1.7478471994400024, "learning_rate": 7.3124000000000005e-06, "loss": 0.018, "step": 13439 }, { "epoch": 0.7165324945353735, "grad_norm": 1.3859974145889282, "learning_rate": 7.3122e-06, "loss": 0.0152, "step": 13440 }, { "epoch": 0.7165858079650264, "grad_norm": 1.8301740884780884, "learning_rate": 7.3120000000000005e-06, "loss": 0.0375, "step": 13441 }, { "epoch": 0.7166391213946793, "grad_norm": 2.016864061355591, "learning_rate": 7.311800000000001e-06, "loss": 0.0302, "step": 13442 }, { "epoch": 0.7166924348243322, "grad_norm": 2.337867498397827, "learning_rate": 7.311600000000001e-06, "loss": 0.0308, "step": 13443 }, { "epoch": 0.7167457482539852, "grad_norm": 2.510650157928467, "learning_rate": 7.311400000000001e-06, "loss": 0.0178, "step": 13444 }, { "epoch": 0.7167990616836382, "grad_norm": 2.809143543243408, "learning_rate": 7.3112e-06, "loss": 0.0689, "step": 13445 }, { "epoch": 0.716852375113291, "grad_norm": 1.295927882194519, "learning_rate": 7.311e-06, "loss": 0.0297, "step": 13446 }, { "epoch": 0.716905688542944, "grad_norm": 1.8610483407974243, "learning_rate": 7.3108e-06, "loss": 0.0208, "step": 13447 }, { "epoch": 0.7169590019725969, "grad_norm": 4.701278209686279, "learning_rate": 7.310600000000001e-06, "loss": 0.0344, "step": 13448 }, { "epoch": 0.7170123154022499, "grad_norm": 1.7342398166656494, "learning_rate": 7.310400000000001e-06, "loss": 0.0357, "step": 13449 }, { "epoch": 0.7170656288319027, "grad_norm": 1.6835492849349976, "learning_rate": 7.310200000000001e-06, "loss": 0.0256, "step": 13450 }, { "epoch": 0.7171189422615557, "grad_norm": 1.9246301651000977, "learning_rate": 7.31e-06, "loss": 0.0175, "step": 13451 }, { "epoch": 0.7171722556912086, "grad_norm": 3.2830312252044678, "learning_rate": 7.3098e-06, "loss": 0.0152, "step": 13452 }, { "epoch": 0.7172255691208616, "grad_norm": 1.7822598218917847, "learning_rate": 7.309600000000001e-06, "loss": 0.0178, "step": 13453 }, { "epoch": 0.7172788825505145, "grad_norm": 3.0674264430999756, "learning_rate": 7.309400000000001e-06, "loss": 0.049, "step": 13454 }, { "epoch": 0.7173321959801674, "grad_norm": 1.1368401050567627, "learning_rate": 7.309200000000001e-06, "loss": 0.0198, "step": 13455 }, { "epoch": 0.7173855094098204, "grad_norm": 0.8048896193504333, "learning_rate": 7.309000000000001e-06, "loss": 0.011, "step": 13456 }, { "epoch": 0.7174388228394732, "grad_norm": 2.3553924560546875, "learning_rate": 7.3088e-06, "loss": 0.0184, "step": 13457 }, { "epoch": 0.7174921362691262, "grad_norm": 3.192910671234131, "learning_rate": 7.3086e-06, "loss": 0.032, "step": 13458 }, { "epoch": 0.7175454496987791, "grad_norm": 2.5929665565490723, "learning_rate": 7.308400000000001e-06, "loss": 0.0118, "step": 13459 }, { "epoch": 0.7175987631284321, "grad_norm": 2.08721923828125, "learning_rate": 7.308200000000001e-06, "loss": 0.0215, "step": 13460 }, { "epoch": 0.7176520765580849, "grad_norm": 2.7015812397003174, "learning_rate": 7.308000000000001e-06, "loss": 0.0788, "step": 13461 }, { "epoch": 0.7177053899877379, "grad_norm": 1.1466188430786133, "learning_rate": 7.3078000000000006e-06, "loss": 0.0337, "step": 13462 }, { "epoch": 0.7177587034173908, "grad_norm": 2.314685821533203, "learning_rate": 7.3076e-06, "loss": 0.0346, "step": 13463 }, { "epoch": 0.7178120168470438, "grad_norm": 3.059262990951538, "learning_rate": 7.3074e-06, "loss": 0.0141, "step": 13464 }, { "epoch": 0.7178653302766967, "grad_norm": 2.4721858501434326, "learning_rate": 7.307200000000001e-06, "loss": 0.0511, "step": 13465 }, { "epoch": 0.7179186437063496, "grad_norm": 1.219375729560852, "learning_rate": 7.3070000000000005e-06, "loss": 0.0147, "step": 13466 }, { "epoch": 0.7179719571360026, "grad_norm": 1.3498138189315796, "learning_rate": 7.306800000000001e-06, "loss": 0.0244, "step": 13467 }, { "epoch": 0.7180252705656555, "grad_norm": 3.842803716659546, "learning_rate": 7.3066000000000005e-06, "loss": 0.0397, "step": 13468 }, { "epoch": 0.7180785839953084, "grad_norm": 2.230470895767212, "learning_rate": 7.3064e-06, "loss": 0.031, "step": 13469 }, { "epoch": 0.7181318974249613, "grad_norm": 2.3839895725250244, "learning_rate": 7.306200000000001e-06, "loss": 0.0329, "step": 13470 }, { "epoch": 0.7181852108546143, "grad_norm": 0.843108057975769, "learning_rate": 7.306000000000001e-06, "loss": 0.0169, "step": 13471 }, { "epoch": 0.7182385242842672, "grad_norm": 0.9147575497627258, "learning_rate": 7.3058e-06, "loss": 0.0122, "step": 13472 }, { "epoch": 0.7182918377139201, "grad_norm": 1.571914553642273, "learning_rate": 7.305600000000001e-06, "loss": 0.0496, "step": 13473 }, { "epoch": 0.7183451511435731, "grad_norm": 3.385859251022339, "learning_rate": 7.3054e-06, "loss": 0.0407, "step": 13474 }, { "epoch": 0.718398464573226, "grad_norm": 1.06107759475708, "learning_rate": 7.3052e-06, "loss": 0.0129, "step": 13475 }, { "epoch": 0.718451778002879, "grad_norm": 1.5438246726989746, "learning_rate": 7.305000000000001e-06, "loss": 0.0236, "step": 13476 }, { "epoch": 0.7185050914325318, "grad_norm": 1.591247797012329, "learning_rate": 7.304800000000001e-06, "loss": 0.0165, "step": 13477 }, { "epoch": 0.7185584048621848, "grad_norm": 2.5048303604125977, "learning_rate": 7.3046e-06, "loss": 0.0577, "step": 13478 }, { "epoch": 0.7186117182918377, "grad_norm": 1.435649037361145, "learning_rate": 7.304400000000001e-06, "loss": 0.028, "step": 13479 }, { "epoch": 0.7186650317214907, "grad_norm": 2.006375551223755, "learning_rate": 7.3042e-06, "loss": 0.0228, "step": 13480 }, { "epoch": 0.7187183451511435, "grad_norm": 2.718356132507324, "learning_rate": 7.304000000000001e-06, "loss": 0.0567, "step": 13481 }, { "epoch": 0.7187716585807965, "grad_norm": 2.9008138179779053, "learning_rate": 7.303800000000001e-06, "loss": 0.053, "step": 13482 }, { "epoch": 0.7188249720104495, "grad_norm": 5.274132251739502, "learning_rate": 7.303600000000001e-06, "loss": 0.037, "step": 13483 }, { "epoch": 0.7188782854401023, "grad_norm": 2.6046462059020996, "learning_rate": 7.3034e-06, "loss": 0.044, "step": 13484 }, { "epoch": 0.7189315988697553, "grad_norm": 2.319035768508911, "learning_rate": 7.303200000000001e-06, "loss": 0.0331, "step": 13485 }, { "epoch": 0.7189849122994082, "grad_norm": 1.2305516004562378, "learning_rate": 7.303e-06, "loss": 0.0087, "step": 13486 }, { "epoch": 0.7190382257290612, "grad_norm": 2.2421841621398926, "learning_rate": 7.302800000000001e-06, "loss": 0.0339, "step": 13487 }, { "epoch": 0.719091539158714, "grad_norm": 1.5056506395339966, "learning_rate": 7.302600000000001e-06, "loss": 0.0273, "step": 13488 }, { "epoch": 0.719144852588367, "grad_norm": 2.7584545612335205, "learning_rate": 7.302400000000001e-06, "loss": 0.0486, "step": 13489 }, { "epoch": 0.7191981660180199, "grad_norm": 1.3367708921432495, "learning_rate": 7.3022e-06, "loss": 0.0208, "step": 13490 }, { "epoch": 0.7192514794476729, "grad_norm": 3.012552261352539, "learning_rate": 7.3020000000000006e-06, "loss": 0.0373, "step": 13491 }, { "epoch": 0.7193047928773258, "grad_norm": 2.1682071685791016, "learning_rate": 7.301800000000001e-06, "loss": 0.0387, "step": 13492 }, { "epoch": 0.7193581063069787, "grad_norm": 4.562480926513672, "learning_rate": 7.3016000000000005e-06, "loss": 0.089, "step": 13493 }, { "epoch": 0.7194114197366317, "grad_norm": 1.7903858423233032, "learning_rate": 7.301400000000001e-06, "loss": 0.0318, "step": 13494 }, { "epoch": 0.7194647331662846, "grad_norm": 3.180171012878418, "learning_rate": 7.3012000000000005e-06, "loss": 0.0344, "step": 13495 }, { "epoch": 0.7195180465959375, "grad_norm": 2.029477119445801, "learning_rate": 7.301e-06, "loss": 0.0886, "step": 13496 }, { "epoch": 0.7195713600255904, "grad_norm": 1.4047925472259521, "learning_rate": 7.3008000000000005e-06, "loss": 0.0148, "step": 13497 }, { "epoch": 0.7196246734552434, "grad_norm": 3.3190805912017822, "learning_rate": 7.300600000000001e-06, "loss": 0.0459, "step": 13498 }, { "epoch": 0.7196779868848963, "grad_norm": 2.1211812496185303, "learning_rate": 7.3004000000000004e-06, "loss": 0.0403, "step": 13499 }, { "epoch": 0.7197313003145492, "grad_norm": 2.320909023284912, "learning_rate": 7.300200000000001e-06, "loss": 0.0425, "step": 13500 }, { "epoch": 0.7197846137442022, "grad_norm": 1.3885974884033203, "learning_rate": 7.3e-06, "loss": 0.0165, "step": 13501 }, { "epoch": 0.7198379271738551, "grad_norm": 1.1443389654159546, "learning_rate": 7.2998e-06, "loss": 0.0131, "step": 13502 }, { "epoch": 0.7198912406035081, "grad_norm": 1.835001826286316, "learning_rate": 7.299600000000001e-06, "loss": 0.0364, "step": 13503 }, { "epoch": 0.7199445540331609, "grad_norm": 2.8753793239593506, "learning_rate": 7.299400000000001e-06, "loss": 0.0593, "step": 13504 }, { "epoch": 0.7199978674628139, "grad_norm": 1.9935158491134644, "learning_rate": 7.2992e-06, "loss": 0.0199, "step": 13505 }, { "epoch": 0.7200511808924668, "grad_norm": 0.9100939631462097, "learning_rate": 7.299000000000001e-06, "loss": 0.011, "step": 13506 }, { "epoch": 0.7201044943221198, "grad_norm": 1.6468498706817627, "learning_rate": 7.2988e-06, "loss": 0.0365, "step": 13507 }, { "epoch": 0.7201578077517726, "grad_norm": 2.3986504077911377, "learning_rate": 7.2986e-06, "loss": 0.0535, "step": 13508 }, { "epoch": 0.7202111211814256, "grad_norm": 1.2724207639694214, "learning_rate": 7.298400000000001e-06, "loss": 0.0164, "step": 13509 }, { "epoch": 0.7202644346110786, "grad_norm": 2.2130963802337646, "learning_rate": 7.298200000000001e-06, "loss": 0.0325, "step": 13510 }, { "epoch": 0.7203177480407315, "grad_norm": 2.3765807151794434, "learning_rate": 7.298e-06, "loss": 0.0512, "step": 13511 }, { "epoch": 0.7203710614703844, "grad_norm": 1.5558140277862549, "learning_rate": 7.297800000000001e-06, "loss": 0.018, "step": 13512 }, { "epoch": 0.7204243749000373, "grad_norm": 1.3508362770080566, "learning_rate": 7.2976e-06, "loss": 0.0171, "step": 13513 }, { "epoch": 0.7204776883296903, "grad_norm": 0.6077527403831482, "learning_rate": 7.2974000000000015e-06, "loss": 0.0064, "step": 13514 }, { "epoch": 0.7205310017593431, "grad_norm": 1.9231659173965454, "learning_rate": 7.297200000000001e-06, "loss": 0.0298, "step": 13515 }, { "epoch": 0.7205843151889961, "grad_norm": 2.0583150386810303, "learning_rate": 7.297000000000001e-06, "loss": 0.0447, "step": 13516 }, { "epoch": 0.720637628618649, "grad_norm": 1.4593920707702637, "learning_rate": 7.2968e-06, "loss": 0.034, "step": 13517 }, { "epoch": 0.720690942048302, "grad_norm": 1.9160394668579102, "learning_rate": 7.296600000000001e-06, "loss": 0.0491, "step": 13518 }, { "epoch": 0.7207442554779548, "grad_norm": 3.4526853561401367, "learning_rate": 7.2964e-06, "loss": 0.0332, "step": 13519 }, { "epoch": 0.7207975689076078, "grad_norm": 2.8544576168060303, "learning_rate": 7.296200000000001e-06, "loss": 0.0547, "step": 13520 }, { "epoch": 0.7208508823372608, "grad_norm": 3.7821402549743652, "learning_rate": 7.296000000000001e-06, "loss": 0.0289, "step": 13521 }, { "epoch": 0.7209041957669137, "grad_norm": 2.2630693912506104, "learning_rate": 7.2958000000000005e-06, "loss": 0.0519, "step": 13522 }, { "epoch": 0.7209575091965666, "grad_norm": 1.9490669965744019, "learning_rate": 7.2956e-06, "loss": 0.0372, "step": 13523 }, { "epoch": 0.7210108226262195, "grad_norm": 1.9030393362045288, "learning_rate": 7.2954000000000005e-06, "loss": 0.0399, "step": 13524 }, { "epoch": 0.7210641360558725, "grad_norm": 2.126878023147583, "learning_rate": 7.2952e-06, "loss": 0.0307, "step": 13525 }, { "epoch": 0.7211174494855254, "grad_norm": 1.63606858253479, "learning_rate": 7.295000000000001e-06, "loss": 0.0265, "step": 13526 }, { "epoch": 0.7211707629151783, "grad_norm": 1.098868727684021, "learning_rate": 7.294800000000001e-06, "loss": 0.0329, "step": 13527 }, { "epoch": 0.7212240763448312, "grad_norm": 2.0178463459014893, "learning_rate": 7.2946000000000004e-06, "loss": 0.0363, "step": 13528 }, { "epoch": 0.7212773897744842, "grad_norm": 0.9304801821708679, "learning_rate": 7.2944e-06, "loss": 0.0111, "step": 13529 }, { "epoch": 0.7213307032041372, "grad_norm": 3.367741584777832, "learning_rate": 7.2942e-06, "loss": 0.053, "step": 13530 }, { "epoch": 0.72138401663379, "grad_norm": 0.8183981776237488, "learning_rate": 7.294000000000001e-06, "loss": 0.0114, "step": 13531 }, { "epoch": 0.721437330063443, "grad_norm": 1.4588000774383545, "learning_rate": 7.293800000000001e-06, "loss": 0.0268, "step": 13532 }, { "epoch": 0.7214906434930959, "grad_norm": 1.8672723770141602, "learning_rate": 7.293600000000001e-06, "loss": 0.0415, "step": 13533 }, { "epoch": 0.7215439569227489, "grad_norm": 1.9319283962249756, "learning_rate": 7.2934e-06, "loss": 0.03, "step": 13534 }, { "epoch": 0.7215972703524017, "grad_norm": 1.1126898527145386, "learning_rate": 7.2932e-06, "loss": 0.0189, "step": 13535 }, { "epoch": 0.7216505837820547, "grad_norm": 1.880589246749878, "learning_rate": 7.293e-06, "loss": 0.0158, "step": 13536 }, { "epoch": 0.7217038972117076, "grad_norm": 1.8234950304031372, "learning_rate": 7.292800000000001e-06, "loss": 0.0269, "step": 13537 }, { "epoch": 0.7217572106413606, "grad_norm": 1.7779676914215088, "learning_rate": 7.292600000000001e-06, "loss": 0.0532, "step": 13538 }, { "epoch": 0.7218105240710135, "grad_norm": 3.126830577850342, "learning_rate": 7.292400000000001e-06, "loss": 0.0316, "step": 13539 }, { "epoch": 0.7218638375006664, "grad_norm": 3.897966146469116, "learning_rate": 7.2922e-06, "loss": 0.0482, "step": 13540 }, { "epoch": 0.7219171509303194, "grad_norm": 1.9476746320724487, "learning_rate": 7.292e-06, "loss": 0.0166, "step": 13541 }, { "epoch": 0.7219704643599723, "grad_norm": 2.0003886222839355, "learning_rate": 7.291800000000001e-06, "loss": 0.0287, "step": 13542 }, { "epoch": 0.7220237777896252, "grad_norm": 1.522281527519226, "learning_rate": 7.291600000000001e-06, "loss": 0.0307, "step": 13543 }, { "epoch": 0.7220770912192781, "grad_norm": 1.3124326467514038, "learning_rate": 7.291400000000001e-06, "loss": 0.0271, "step": 13544 }, { "epoch": 0.7221304046489311, "grad_norm": 2.849102258682251, "learning_rate": 7.291200000000001e-06, "loss": 0.0298, "step": 13545 }, { "epoch": 0.722183718078584, "grad_norm": 1.970942497253418, "learning_rate": 7.291e-06, "loss": 0.0327, "step": 13546 }, { "epoch": 0.7222370315082369, "grad_norm": 2.268085479736328, "learning_rate": 7.2908e-06, "loss": 0.0425, "step": 13547 }, { "epoch": 0.7222903449378899, "grad_norm": 0.9688994884490967, "learning_rate": 7.290600000000001e-06, "loss": 0.0135, "step": 13548 }, { "epoch": 0.7223436583675428, "grad_norm": 2.2645890712738037, "learning_rate": 7.2904000000000005e-06, "loss": 0.0227, "step": 13549 }, { "epoch": 0.7223969717971958, "grad_norm": 1.6786831617355347, "learning_rate": 7.290200000000001e-06, "loss": 0.0226, "step": 13550 }, { "epoch": 0.7224502852268486, "grad_norm": 1.9391118288040161, "learning_rate": 7.2900000000000005e-06, "loss": 0.0173, "step": 13551 }, { "epoch": 0.7225035986565016, "grad_norm": 2.3803658485412598, "learning_rate": 7.2898e-06, "loss": 0.0348, "step": 13552 }, { "epoch": 0.7225569120861545, "grad_norm": 2.4248974323272705, "learning_rate": 7.289600000000001e-06, "loss": 0.0437, "step": 13553 }, { "epoch": 0.7226102255158074, "grad_norm": 1.9764392375946045, "learning_rate": 7.289400000000001e-06, "loss": 0.0159, "step": 13554 }, { "epoch": 0.7226635389454603, "grad_norm": 0.8708729147911072, "learning_rate": 7.2892000000000004e-06, "loss": 0.0198, "step": 13555 }, { "epoch": 0.7227168523751133, "grad_norm": 2.212536096572876, "learning_rate": 7.289000000000001e-06, "loss": 0.0319, "step": 13556 }, { "epoch": 0.7227701658047663, "grad_norm": 1.65574312210083, "learning_rate": 7.2888e-06, "loss": 0.021, "step": 13557 }, { "epoch": 0.7228234792344191, "grad_norm": 3.007120370864868, "learning_rate": 7.2886e-06, "loss": 0.0598, "step": 13558 }, { "epoch": 0.7228767926640721, "grad_norm": 2.0490360260009766, "learning_rate": 7.288400000000001e-06, "loss": 0.0199, "step": 13559 }, { "epoch": 0.722930106093725, "grad_norm": 4.872061729431152, "learning_rate": 7.288200000000001e-06, "loss": 0.0321, "step": 13560 }, { "epoch": 0.722983419523378, "grad_norm": 3.7217047214508057, "learning_rate": 7.288e-06, "loss": 0.0675, "step": 13561 }, { "epoch": 0.7230367329530308, "grad_norm": 3.303480386734009, "learning_rate": 7.287800000000001e-06, "loss": 0.036, "step": 13562 }, { "epoch": 0.7230900463826838, "grad_norm": 1.3369148969650269, "learning_rate": 7.2876e-06, "loss": 0.0193, "step": 13563 }, { "epoch": 0.7231433598123367, "grad_norm": 1.853007197380066, "learning_rate": 7.287400000000001e-06, "loss": 0.0267, "step": 13564 }, { "epoch": 0.7231966732419897, "grad_norm": 3.0268568992614746, "learning_rate": 7.287200000000001e-06, "loss": 0.0343, "step": 13565 }, { "epoch": 0.7232499866716426, "grad_norm": 2.5449142456054688, "learning_rate": 7.287000000000001e-06, "loss": 0.067, "step": 13566 }, { "epoch": 0.7233033001012955, "grad_norm": 2.8458945751190186, "learning_rate": 7.2868e-06, "loss": 0.0291, "step": 13567 }, { "epoch": 0.7233566135309485, "grad_norm": 1.9537487030029297, "learning_rate": 7.286600000000001e-06, "loss": 0.0164, "step": 13568 }, { "epoch": 0.7234099269606014, "grad_norm": 2.339325189590454, "learning_rate": 7.2864e-06, "loss": 0.0235, "step": 13569 }, { "epoch": 0.7234632403902543, "grad_norm": 4.949108600616455, "learning_rate": 7.286200000000001e-06, "loss": 0.0848, "step": 13570 }, { "epoch": 0.7235165538199072, "grad_norm": 5.123643398284912, "learning_rate": 7.286000000000001e-06, "loss": 0.0367, "step": 13571 }, { "epoch": 0.7235698672495602, "grad_norm": 3.544527530670166, "learning_rate": 7.285800000000001e-06, "loss": 0.0543, "step": 13572 }, { "epoch": 0.723623180679213, "grad_norm": 1.5617600679397583, "learning_rate": 7.2856e-06, "loss": 0.0216, "step": 13573 }, { "epoch": 0.723676494108866, "grad_norm": 2.7486417293548584, "learning_rate": 7.285400000000001e-06, "loss": 0.047, "step": 13574 }, { "epoch": 0.7237298075385189, "grad_norm": 1.391190767288208, "learning_rate": 7.2852e-06, "loss": 0.0145, "step": 13575 }, { "epoch": 0.7237831209681719, "grad_norm": 3.700080156326294, "learning_rate": 7.2850000000000006e-06, "loss": 0.0657, "step": 13576 }, { "epoch": 0.7238364343978249, "grad_norm": 2.6981253623962402, "learning_rate": 7.284800000000001e-06, "loss": 0.05, "step": 13577 }, { "epoch": 0.7238897478274777, "grad_norm": 1.8492004871368408, "learning_rate": 7.2846000000000005e-06, "loss": 0.0229, "step": 13578 }, { "epoch": 0.7239430612571307, "grad_norm": 2.9491069316864014, "learning_rate": 7.2844e-06, "loss": 0.0406, "step": 13579 }, { "epoch": 0.7239963746867836, "grad_norm": 3.5201117992401123, "learning_rate": 7.2842000000000005e-06, "loss": 0.0339, "step": 13580 }, { "epoch": 0.7240496881164366, "grad_norm": 3.04050874710083, "learning_rate": 7.284000000000001e-06, "loss": 0.0555, "step": 13581 }, { "epoch": 0.7241030015460894, "grad_norm": 2.187873125076294, "learning_rate": 7.2838000000000005e-06, "loss": 0.0405, "step": 13582 }, { "epoch": 0.7241563149757424, "grad_norm": 2.898230791091919, "learning_rate": 7.283600000000001e-06, "loss": 0.0309, "step": 13583 }, { "epoch": 0.7242096284053953, "grad_norm": 3.655813217163086, "learning_rate": 7.2834000000000004e-06, "loss": 0.0491, "step": 13584 }, { "epoch": 0.7242629418350482, "grad_norm": 1.8600661754608154, "learning_rate": 7.2832e-06, "loss": 0.0418, "step": 13585 }, { "epoch": 0.7243162552647012, "grad_norm": 2.3913891315460205, "learning_rate": 7.283e-06, "loss": 0.0382, "step": 13586 }, { "epoch": 0.7243695686943541, "grad_norm": 0.43646883964538574, "learning_rate": 7.282800000000001e-06, "loss": 0.0049, "step": 13587 }, { "epoch": 0.7244228821240071, "grad_norm": 2.1038272380828857, "learning_rate": 7.2826e-06, "loss": 0.0443, "step": 13588 }, { "epoch": 0.7244761955536599, "grad_norm": 1.8500046730041504, "learning_rate": 7.282400000000001e-06, "loss": 0.0275, "step": 13589 }, { "epoch": 0.7245295089833129, "grad_norm": 2.586970567703247, "learning_rate": 7.2822e-06, "loss": 0.0238, "step": 13590 }, { "epoch": 0.7245828224129658, "grad_norm": 2.3858871459960938, "learning_rate": 7.282e-06, "loss": 0.0387, "step": 13591 }, { "epoch": 0.7246361358426188, "grad_norm": 2.2214465141296387, "learning_rate": 7.281800000000001e-06, "loss": 0.0453, "step": 13592 }, { "epoch": 0.7246894492722716, "grad_norm": 0.9635970592498779, "learning_rate": 7.281600000000001e-06, "loss": 0.0151, "step": 13593 }, { "epoch": 0.7247427627019246, "grad_norm": 2.7527658939361572, "learning_rate": 7.2814e-06, "loss": 0.0677, "step": 13594 }, { "epoch": 0.7247960761315776, "grad_norm": 1.836620807647705, "learning_rate": 7.281200000000001e-06, "loss": 0.0508, "step": 13595 }, { "epoch": 0.7248493895612305, "grad_norm": 1.2878140211105347, "learning_rate": 7.281e-06, "loss": 0.0409, "step": 13596 }, { "epoch": 0.7249027029908834, "grad_norm": 1.9021663665771484, "learning_rate": 7.2808e-06, "loss": 0.033, "step": 13597 }, { "epoch": 0.7249560164205363, "grad_norm": 1.199197769165039, "learning_rate": 7.280600000000001e-06, "loss": 0.0296, "step": 13598 }, { "epoch": 0.7250093298501893, "grad_norm": 1.998292088508606, "learning_rate": 7.280400000000001e-06, "loss": 0.0265, "step": 13599 }, { "epoch": 0.7250626432798422, "grad_norm": 3.078927755355835, "learning_rate": 7.2802e-06, "loss": 0.0313, "step": 13600 }, { "epoch": 0.7251159567094951, "grad_norm": 2.6317310333251953, "learning_rate": 7.280000000000001e-06, "loss": 0.0333, "step": 13601 }, { "epoch": 0.725169270139148, "grad_norm": 3.3684568405151367, "learning_rate": 7.2798e-06, "loss": 0.0487, "step": 13602 }, { "epoch": 0.725222583568801, "grad_norm": 2.0505318641662598, "learning_rate": 7.279600000000001e-06, "loss": 0.0351, "step": 13603 }, { "epoch": 0.725275896998454, "grad_norm": 1.7310662269592285, "learning_rate": 7.279400000000001e-06, "loss": 0.0369, "step": 13604 }, { "epoch": 0.7253292104281068, "grad_norm": 1.7475903034210205, "learning_rate": 7.2792000000000005e-06, "loss": 0.0313, "step": 13605 }, { "epoch": 0.7253825238577598, "grad_norm": 2.404381513595581, "learning_rate": 7.279e-06, "loss": 0.0278, "step": 13606 }, { "epoch": 0.7254358372874127, "grad_norm": 1.6748918294906616, "learning_rate": 7.2788000000000005e-06, "loss": 0.0214, "step": 13607 }, { "epoch": 0.7254891507170657, "grad_norm": 0.8757678866386414, "learning_rate": 7.2786e-06, "loss": 0.0177, "step": 13608 }, { "epoch": 0.7255424641467185, "grad_norm": 2.943263053894043, "learning_rate": 7.278400000000001e-06, "loss": 0.0298, "step": 13609 }, { "epoch": 0.7255957775763715, "grad_norm": 1.6943440437316895, "learning_rate": 7.278200000000001e-06, "loss": 0.0354, "step": 13610 }, { "epoch": 0.7256490910060244, "grad_norm": 2.3510091304779053, "learning_rate": 7.2780000000000005e-06, "loss": 0.0291, "step": 13611 }, { "epoch": 0.7257024044356774, "grad_norm": 1.9197814464569092, "learning_rate": 7.2778e-06, "loss": 0.0265, "step": 13612 }, { "epoch": 0.7257557178653303, "grad_norm": 1.7525267601013184, "learning_rate": 7.2776e-06, "loss": 0.0341, "step": 13613 }, { "epoch": 0.7258090312949832, "grad_norm": 1.9121575355529785, "learning_rate": 7.277400000000001e-06, "loss": 0.0444, "step": 13614 }, { "epoch": 0.7258623447246362, "grad_norm": 1.9616447687149048, "learning_rate": 7.277200000000001e-06, "loss": 0.0419, "step": 13615 }, { "epoch": 0.725915658154289, "grad_norm": 3.4162228107452393, "learning_rate": 7.277000000000001e-06, "loss": 0.059, "step": 13616 }, { "epoch": 0.725968971583942, "grad_norm": 1.8381192684173584, "learning_rate": 7.2768e-06, "loss": 0.0338, "step": 13617 }, { "epoch": 0.7260222850135949, "grad_norm": 1.3829338550567627, "learning_rate": 7.2766e-06, "loss": 0.0162, "step": 13618 }, { "epoch": 0.7260755984432479, "grad_norm": 1.7347146272659302, "learning_rate": 7.2764e-06, "loss": 0.0509, "step": 13619 }, { "epoch": 0.7261289118729007, "grad_norm": 1.520455002784729, "learning_rate": 7.276200000000001e-06, "loss": 0.023, "step": 13620 }, { "epoch": 0.7261822253025537, "grad_norm": 1.8397616147994995, "learning_rate": 7.276000000000001e-06, "loss": 0.0256, "step": 13621 }, { "epoch": 0.7262355387322067, "grad_norm": 1.8603698015213013, "learning_rate": 7.275800000000001e-06, "loss": 0.0169, "step": 13622 }, { "epoch": 0.7262888521618596, "grad_norm": 1.8774982690811157, "learning_rate": 7.2756e-06, "loss": 0.0365, "step": 13623 }, { "epoch": 0.7263421655915125, "grad_norm": 1.4160178899765015, "learning_rate": 7.2754e-06, "loss": 0.0178, "step": 13624 }, { "epoch": 0.7263954790211654, "grad_norm": 2.0805745124816895, "learning_rate": 7.2752e-06, "loss": 0.0296, "step": 13625 }, { "epoch": 0.7264487924508184, "grad_norm": 3.051030158996582, "learning_rate": 7.275000000000001e-06, "loss": 0.035, "step": 13626 }, { "epoch": 0.7265021058804713, "grad_norm": 1.1933766603469849, "learning_rate": 7.274800000000001e-06, "loss": 0.0152, "step": 13627 }, { "epoch": 0.7265554193101242, "grad_norm": 1.6031684875488281, "learning_rate": 7.274600000000001e-06, "loss": 0.0253, "step": 13628 }, { "epoch": 0.7266087327397771, "grad_norm": 1.6429802179336548, "learning_rate": 7.2744e-06, "loss": 0.0264, "step": 13629 }, { "epoch": 0.7266620461694301, "grad_norm": 2.6012086868286133, "learning_rate": 7.2742e-06, "loss": 0.036, "step": 13630 }, { "epoch": 0.726715359599083, "grad_norm": 0.747317910194397, "learning_rate": 7.274000000000001e-06, "loss": 0.0047, "step": 13631 }, { "epoch": 0.7267686730287359, "grad_norm": 3.7255208492279053, "learning_rate": 7.2738000000000006e-06, "loss": 0.0565, "step": 13632 }, { "epoch": 0.7268219864583889, "grad_norm": 4.19573450088501, "learning_rate": 7.273600000000001e-06, "loss": 0.0442, "step": 13633 }, { "epoch": 0.7268752998880418, "grad_norm": 1.607796549797058, "learning_rate": 7.2734000000000005e-06, "loss": 0.021, "step": 13634 }, { "epoch": 0.7269286133176948, "grad_norm": 2.7870020866394043, "learning_rate": 7.2732e-06, "loss": 0.0208, "step": 13635 }, { "epoch": 0.7269819267473476, "grad_norm": 1.6279942989349365, "learning_rate": 7.273e-06, "loss": 0.0176, "step": 13636 }, { "epoch": 0.7270352401770006, "grad_norm": 2.010735273361206, "learning_rate": 7.272800000000001e-06, "loss": 0.0297, "step": 13637 }, { "epoch": 0.7270885536066535, "grad_norm": 1.5579167604446411, "learning_rate": 7.2726000000000005e-06, "loss": 0.0209, "step": 13638 }, { "epoch": 0.7271418670363065, "grad_norm": 1.236411452293396, "learning_rate": 7.272400000000001e-06, "loss": 0.0174, "step": 13639 }, { "epoch": 0.7271951804659593, "grad_norm": 2.0866589546203613, "learning_rate": 7.2722000000000004e-06, "loss": 0.0317, "step": 13640 }, { "epoch": 0.7272484938956123, "grad_norm": 1.6818537712097168, "learning_rate": 7.272e-06, "loss": 0.0131, "step": 13641 }, { "epoch": 0.7273018073252653, "grad_norm": 2.0694968700408936, "learning_rate": 7.271800000000001e-06, "loss": 0.0221, "step": 13642 }, { "epoch": 0.7273551207549181, "grad_norm": 2.152385711669922, "learning_rate": 7.271600000000001e-06, "loss": 0.0319, "step": 13643 }, { "epoch": 0.7274084341845711, "grad_norm": 1.2198830842971802, "learning_rate": 7.2714e-06, "loss": 0.0117, "step": 13644 }, { "epoch": 0.727461747614224, "grad_norm": 1.7718288898468018, "learning_rate": 7.271200000000001e-06, "loss": 0.0451, "step": 13645 }, { "epoch": 0.727515061043877, "grad_norm": 1.8029866218566895, "learning_rate": 7.271e-06, "loss": 0.03, "step": 13646 }, { "epoch": 0.7275683744735298, "grad_norm": 2.1907944679260254, "learning_rate": 7.2708e-06, "loss": 0.0524, "step": 13647 }, { "epoch": 0.7276216879031828, "grad_norm": 2.096982479095459, "learning_rate": 7.270600000000001e-06, "loss": 0.0457, "step": 13648 }, { "epoch": 0.7276750013328357, "grad_norm": 1.5057257413864136, "learning_rate": 7.270400000000001e-06, "loss": 0.0153, "step": 13649 }, { "epoch": 0.7277283147624887, "grad_norm": 2.2658770084381104, "learning_rate": 7.2702e-06, "loss": 0.0569, "step": 13650 }, { "epoch": 0.7277816281921416, "grad_norm": 1.9007798433303833, "learning_rate": 7.270000000000001e-06, "loss": 0.0159, "step": 13651 }, { "epoch": 0.7278349416217945, "grad_norm": 6.016185283660889, "learning_rate": 7.2698e-06, "loss": 0.0536, "step": 13652 }, { "epoch": 0.7278882550514475, "grad_norm": 1.9266612529754639, "learning_rate": 7.2696000000000015e-06, "loss": 0.0275, "step": 13653 }, { "epoch": 0.7279415684811004, "grad_norm": 1.739104986190796, "learning_rate": 7.269400000000001e-06, "loss": 0.0232, "step": 13654 }, { "epoch": 0.7279948819107533, "grad_norm": 1.4984389543533325, "learning_rate": 7.269200000000001e-06, "loss": 0.0387, "step": 13655 }, { "epoch": 0.7280481953404062, "grad_norm": 2.2832424640655518, "learning_rate": 7.269e-06, "loss": 0.0302, "step": 13656 }, { "epoch": 0.7281015087700592, "grad_norm": 2.4435782432556152, "learning_rate": 7.268800000000001e-06, "loss": 0.0778, "step": 13657 }, { "epoch": 0.7281548221997121, "grad_norm": 2.533317804336548, "learning_rate": 7.2686e-06, "loss": 0.0273, "step": 13658 }, { "epoch": 0.728208135629365, "grad_norm": 5.003353595733643, "learning_rate": 7.2684000000000014e-06, "loss": 0.0484, "step": 13659 }, { "epoch": 0.728261449059018, "grad_norm": 1.8377571105957031, "learning_rate": 7.268200000000001e-06, "loss": 0.0312, "step": 13660 }, { "epoch": 0.7283147624886709, "grad_norm": 1.546384572982788, "learning_rate": 7.2680000000000005e-06, "loss": 0.0264, "step": 13661 }, { "epoch": 0.7283680759183239, "grad_norm": 0.9729154109954834, "learning_rate": 7.2678e-06, "loss": 0.0223, "step": 13662 }, { "epoch": 0.7284213893479767, "grad_norm": 2.3213627338409424, "learning_rate": 7.2676000000000005e-06, "loss": 0.0392, "step": 13663 }, { "epoch": 0.7284747027776297, "grad_norm": 2.1816537380218506, "learning_rate": 7.267400000000001e-06, "loss": 0.0437, "step": 13664 }, { "epoch": 0.7285280162072826, "grad_norm": 2.1912882328033447, "learning_rate": 7.267200000000001e-06, "loss": 0.0379, "step": 13665 }, { "epoch": 0.7285813296369356, "grad_norm": 3.329160213470459, "learning_rate": 7.267000000000001e-06, "loss": 0.0601, "step": 13666 }, { "epoch": 0.7286346430665884, "grad_norm": 1.2557004690170288, "learning_rate": 7.2668000000000005e-06, "loss": 0.0228, "step": 13667 }, { "epoch": 0.7286879564962414, "grad_norm": 3.93914532661438, "learning_rate": 7.2666e-06, "loss": 0.0445, "step": 13668 }, { "epoch": 0.7287412699258944, "grad_norm": 5.791038513183594, "learning_rate": 7.2664000000000004e-06, "loss": 0.0445, "step": 13669 }, { "epoch": 0.7287945833555473, "grad_norm": 3.0397865772247314, "learning_rate": 7.266200000000001e-06, "loss": 0.0398, "step": 13670 }, { "epoch": 0.7288478967852002, "grad_norm": 1.6238012313842773, "learning_rate": 7.266000000000001e-06, "loss": 0.0378, "step": 13671 }, { "epoch": 0.7289012102148531, "grad_norm": 4.1419677734375, "learning_rate": 7.265800000000001e-06, "loss": 0.0676, "step": 13672 }, { "epoch": 0.7289545236445061, "grad_norm": 1.6884640455245972, "learning_rate": 7.2656e-06, "loss": 0.0381, "step": 13673 }, { "epoch": 0.729007837074159, "grad_norm": 1.5623105764389038, "learning_rate": 7.2654e-06, "loss": 0.0334, "step": 13674 }, { "epoch": 0.7290611505038119, "grad_norm": 1.4578592777252197, "learning_rate": 7.265200000000001e-06, "loss": 0.0329, "step": 13675 }, { "epoch": 0.7291144639334648, "grad_norm": 1.6168535947799683, "learning_rate": 7.265000000000001e-06, "loss": 0.0329, "step": 13676 }, { "epoch": 0.7291677773631178, "grad_norm": 2.029850482940674, "learning_rate": 7.264800000000001e-06, "loss": 0.0193, "step": 13677 }, { "epoch": 0.7292210907927708, "grad_norm": 1.3136955499649048, "learning_rate": 7.264600000000001e-06, "loss": 0.0134, "step": 13678 }, { "epoch": 0.7292744042224236, "grad_norm": 1.5487500429153442, "learning_rate": 7.2644e-06, "loss": 0.0317, "step": 13679 }, { "epoch": 0.7293277176520766, "grad_norm": 1.885077714920044, "learning_rate": 7.2642e-06, "loss": 0.018, "step": 13680 }, { "epoch": 0.7293810310817295, "grad_norm": 1.875193476676941, "learning_rate": 7.264000000000001e-06, "loss": 0.0184, "step": 13681 }, { "epoch": 0.7294343445113824, "grad_norm": 2.343066930770874, "learning_rate": 7.263800000000001e-06, "loss": 0.0261, "step": 13682 }, { "epoch": 0.7294876579410353, "grad_norm": 3.5617120265960693, "learning_rate": 7.263600000000001e-06, "loss": 0.0756, "step": 13683 }, { "epoch": 0.7295409713706883, "grad_norm": 2.3421921730041504, "learning_rate": 7.263400000000001e-06, "loss": 0.0325, "step": 13684 }, { "epoch": 0.7295942848003412, "grad_norm": 3.442898988723755, "learning_rate": 7.2632e-06, "loss": 0.0386, "step": 13685 }, { "epoch": 0.7296475982299941, "grad_norm": 1.0972627401351929, "learning_rate": 7.263e-06, "loss": 0.0199, "step": 13686 }, { "epoch": 0.729700911659647, "grad_norm": 1.4608339071273804, "learning_rate": 7.262800000000001e-06, "loss": 0.017, "step": 13687 }, { "epoch": 0.7297542250893, "grad_norm": 2.2051355838775635, "learning_rate": 7.2626000000000006e-06, "loss": 0.0298, "step": 13688 }, { "epoch": 0.729807538518953, "grad_norm": 1.8232966661453247, "learning_rate": 7.262400000000001e-06, "loss": 0.0141, "step": 13689 }, { "epoch": 0.7298608519486058, "grad_norm": 1.3193506002426147, "learning_rate": 7.2622000000000005e-06, "loss": 0.0212, "step": 13690 }, { "epoch": 0.7299141653782588, "grad_norm": 1.903825044631958, "learning_rate": 7.262e-06, "loss": 0.0409, "step": 13691 }, { "epoch": 0.7299674788079117, "grad_norm": 3.0743372440338135, "learning_rate": 7.261800000000001e-06, "loss": 0.0579, "step": 13692 }, { "epoch": 0.7300207922375647, "grad_norm": 3.6463284492492676, "learning_rate": 7.261600000000001e-06, "loss": 0.0708, "step": 13693 }, { "epoch": 0.7300741056672175, "grad_norm": 2.1278035640716553, "learning_rate": 7.2614000000000005e-06, "loss": 0.0312, "step": 13694 }, { "epoch": 0.7301274190968705, "grad_norm": 2.1179659366607666, "learning_rate": 7.261200000000001e-06, "loss": 0.0143, "step": 13695 }, { "epoch": 0.7301807325265234, "grad_norm": 2.1461007595062256, "learning_rate": 7.2610000000000004e-06, "loss": 0.0326, "step": 13696 }, { "epoch": 0.7302340459561764, "grad_norm": 2.682752847671509, "learning_rate": 7.2608e-06, "loss": 0.0344, "step": 13697 }, { "epoch": 0.7302873593858293, "grad_norm": 2.5669515132904053, "learning_rate": 7.260600000000001e-06, "loss": 0.027, "step": 13698 }, { "epoch": 0.7303406728154822, "grad_norm": 1.6711715459823608, "learning_rate": 7.260400000000001e-06, "loss": 0.0281, "step": 13699 }, { "epoch": 0.7303939862451352, "grad_norm": 2.724332332611084, "learning_rate": 7.2602e-06, "loss": 0.0467, "step": 13700 }, { "epoch": 0.730447299674788, "grad_norm": 2.146016836166382, "learning_rate": 7.260000000000001e-06, "loss": 0.0401, "step": 13701 }, { "epoch": 0.730500613104441, "grad_norm": 1.9172624349594116, "learning_rate": 7.2598e-06, "loss": 0.046, "step": 13702 }, { "epoch": 0.7305539265340939, "grad_norm": 2.8191592693328857, "learning_rate": 7.259600000000001e-06, "loss": 0.036, "step": 13703 }, { "epoch": 0.7306072399637469, "grad_norm": 3.3416566848754883, "learning_rate": 7.259400000000001e-06, "loss": 0.0338, "step": 13704 }, { "epoch": 0.7306605533933997, "grad_norm": 1.6139004230499268, "learning_rate": 7.259200000000001e-06, "loss": 0.048, "step": 13705 }, { "epoch": 0.7307138668230527, "grad_norm": 1.2359713315963745, "learning_rate": 7.259e-06, "loss": 0.0164, "step": 13706 }, { "epoch": 0.7307671802527057, "grad_norm": 2.7373600006103516, "learning_rate": 7.258800000000001e-06, "loss": 0.0208, "step": 13707 }, { "epoch": 0.7308204936823586, "grad_norm": 2.5948777198791504, "learning_rate": 7.2586e-06, "loss": 0.0536, "step": 13708 }, { "epoch": 0.7308738071120116, "grad_norm": 1.3244950771331787, "learning_rate": 7.258400000000001e-06, "loss": 0.0191, "step": 13709 }, { "epoch": 0.7309271205416644, "grad_norm": 1.7239973545074463, "learning_rate": 7.258200000000001e-06, "loss": 0.0239, "step": 13710 }, { "epoch": 0.7309804339713174, "grad_norm": 2.2744505405426025, "learning_rate": 7.258000000000001e-06, "loss": 0.0386, "step": 13711 }, { "epoch": 0.7310337474009703, "grad_norm": 2.5155014991760254, "learning_rate": 7.2578e-06, "loss": 0.0348, "step": 13712 }, { "epoch": 0.7310870608306232, "grad_norm": 1.8337758779525757, "learning_rate": 7.257600000000001e-06, "loss": 0.0184, "step": 13713 }, { "epoch": 0.7311403742602761, "grad_norm": 1.3058174848556519, "learning_rate": 7.257400000000001e-06, "loss": 0.0311, "step": 13714 }, { "epoch": 0.7311936876899291, "grad_norm": 1.924530029296875, "learning_rate": 7.257200000000001e-06, "loss": 0.0518, "step": 13715 }, { "epoch": 0.7312470011195821, "grad_norm": 1.4380954504013062, "learning_rate": 7.257000000000001e-06, "loss": 0.0144, "step": 13716 }, { "epoch": 0.7313003145492349, "grad_norm": 2.6395626068115234, "learning_rate": 7.2568000000000006e-06, "loss": 0.0514, "step": 13717 }, { "epoch": 0.7313536279788879, "grad_norm": 2.9007742404937744, "learning_rate": 7.2566e-06, "loss": 0.0295, "step": 13718 }, { "epoch": 0.7314069414085408, "grad_norm": 1.6978474855422974, "learning_rate": 7.2564000000000005e-06, "loss": 0.0241, "step": 13719 }, { "epoch": 0.7314602548381938, "grad_norm": 3.0632474422454834, "learning_rate": 7.256200000000001e-06, "loss": 0.0513, "step": 13720 }, { "epoch": 0.7315135682678466, "grad_norm": 1.02641761302948, "learning_rate": 7.2560000000000005e-06, "loss": 0.0126, "step": 13721 }, { "epoch": 0.7315668816974996, "grad_norm": 2.1086127758026123, "learning_rate": 7.255800000000001e-06, "loss": 0.0215, "step": 13722 }, { "epoch": 0.7316201951271525, "grad_norm": 2.3325915336608887, "learning_rate": 7.2556000000000005e-06, "loss": 0.0557, "step": 13723 }, { "epoch": 0.7316735085568055, "grad_norm": 2.231262445449829, "learning_rate": 7.2554e-06, "loss": 0.0155, "step": 13724 }, { "epoch": 0.7317268219864584, "grad_norm": 4.094273567199707, "learning_rate": 7.255200000000001e-06, "loss": 0.0387, "step": 13725 }, { "epoch": 0.7317801354161113, "grad_norm": 1.1825352907180786, "learning_rate": 7.255000000000001e-06, "loss": 0.0122, "step": 13726 }, { "epoch": 0.7318334488457643, "grad_norm": 2.5357754230499268, "learning_rate": 7.2548e-06, "loss": 0.0209, "step": 13727 }, { "epoch": 0.7318867622754172, "grad_norm": 1.746414065361023, "learning_rate": 7.254600000000001e-06, "loss": 0.0364, "step": 13728 }, { "epoch": 0.7319400757050701, "grad_norm": 3.382941246032715, "learning_rate": 7.2544e-06, "loss": 0.0484, "step": 13729 }, { "epoch": 0.731993389134723, "grad_norm": 0.7456936240196228, "learning_rate": 7.2542e-06, "loss": 0.007, "step": 13730 }, { "epoch": 0.732046702564376, "grad_norm": 2.885298252105713, "learning_rate": 7.254000000000001e-06, "loss": 0.0564, "step": 13731 }, { "epoch": 0.7321000159940289, "grad_norm": 2.6724884510040283, "learning_rate": 7.253800000000001e-06, "loss": 0.0559, "step": 13732 }, { "epoch": 0.7321533294236818, "grad_norm": 1.9911383390426636, "learning_rate": 7.2536e-06, "loss": 0.0415, "step": 13733 }, { "epoch": 0.7322066428533348, "grad_norm": 1.7251741886138916, "learning_rate": 7.253400000000001e-06, "loss": 0.027, "step": 13734 }, { "epoch": 0.7322599562829877, "grad_norm": 3.851780891418457, "learning_rate": 7.2532e-06, "loss": 0.0452, "step": 13735 }, { "epoch": 0.7323132697126407, "grad_norm": 2.506401538848877, "learning_rate": 7.253e-06, "loss": 0.0407, "step": 13736 }, { "epoch": 0.7323665831422935, "grad_norm": 2.0410447120666504, "learning_rate": 7.252800000000001e-06, "loss": 0.0358, "step": 13737 }, { "epoch": 0.7324198965719465, "grad_norm": 2.220400810241699, "learning_rate": 7.252600000000001e-06, "loss": 0.029, "step": 13738 }, { "epoch": 0.7324732100015994, "grad_norm": 2.453326463699341, "learning_rate": 7.2524e-06, "loss": 0.0387, "step": 13739 }, { "epoch": 0.7325265234312524, "grad_norm": 1.9743223190307617, "learning_rate": 7.252200000000001e-06, "loss": 0.0255, "step": 13740 }, { "epoch": 0.7325798368609052, "grad_norm": 4.263114929199219, "learning_rate": 7.252e-06, "loss": 0.0309, "step": 13741 }, { "epoch": 0.7326331502905582, "grad_norm": 2.318697929382324, "learning_rate": 7.2518000000000014e-06, "loss": 0.0354, "step": 13742 }, { "epoch": 0.7326864637202111, "grad_norm": 3.7802960872650146, "learning_rate": 7.251600000000001e-06, "loss": 0.0779, "step": 13743 }, { "epoch": 0.732739777149864, "grad_norm": 1.790968418121338, "learning_rate": 7.2514000000000006e-06, "loss": 0.024, "step": 13744 }, { "epoch": 0.732793090579517, "grad_norm": 2.3229312896728516, "learning_rate": 7.2512e-06, "loss": 0.0624, "step": 13745 }, { "epoch": 0.7328464040091699, "grad_norm": 2.741786241531372, "learning_rate": 7.2510000000000005e-06, "loss": 0.0439, "step": 13746 }, { "epoch": 0.7328997174388229, "grad_norm": 2.099189519882202, "learning_rate": 7.2508e-06, "loss": 0.028, "step": 13747 }, { "epoch": 0.7329530308684757, "grad_norm": 1.770180106163025, "learning_rate": 7.250600000000001e-06, "loss": 0.0386, "step": 13748 }, { "epoch": 0.7330063442981287, "grad_norm": 2.177129030227661, "learning_rate": 7.250400000000001e-06, "loss": 0.0288, "step": 13749 }, { "epoch": 0.7330596577277816, "grad_norm": 1.4711989164352417, "learning_rate": 7.2502000000000005e-06, "loss": 0.0163, "step": 13750 }, { "epoch": 0.7331129711574346, "grad_norm": 1.493133306503296, "learning_rate": 7.25e-06, "loss": 0.0327, "step": 13751 }, { "epoch": 0.7331662845870874, "grad_norm": 1.4504915475845337, "learning_rate": 7.2498000000000005e-06, "loss": 0.0127, "step": 13752 }, { "epoch": 0.7332195980167404, "grad_norm": 2.182878017425537, "learning_rate": 7.249600000000001e-06, "loss": 0.0421, "step": 13753 }, { "epoch": 0.7332729114463934, "grad_norm": 2.162327527999878, "learning_rate": 7.249400000000001e-06, "loss": 0.0572, "step": 13754 }, { "epoch": 0.7333262248760463, "grad_norm": 2.127108573913574, "learning_rate": 7.249200000000001e-06, "loss": 0.0421, "step": 13755 }, { "epoch": 0.7333795383056992, "grad_norm": 2.1546459197998047, "learning_rate": 7.249e-06, "loss": 0.0398, "step": 13756 }, { "epoch": 0.7334328517353521, "grad_norm": 1.8537349700927734, "learning_rate": 7.2488e-06, "loss": 0.0432, "step": 13757 }, { "epoch": 0.7334861651650051, "grad_norm": 1.3065530061721802, "learning_rate": 7.2486e-06, "loss": 0.0098, "step": 13758 }, { "epoch": 0.733539478594658, "grad_norm": 1.4699084758758545, "learning_rate": 7.248400000000001e-06, "loss": 0.0294, "step": 13759 }, { "epoch": 0.7335927920243109, "grad_norm": 2.398627758026123, "learning_rate": 7.248200000000001e-06, "loss": 0.046, "step": 13760 }, { "epoch": 0.7336461054539638, "grad_norm": 1.3599847555160522, "learning_rate": 7.248000000000001e-06, "loss": 0.0145, "step": 13761 }, { "epoch": 0.7336994188836168, "grad_norm": 2.58634877204895, "learning_rate": 7.2478e-06, "loss": 0.0479, "step": 13762 }, { "epoch": 0.7337527323132698, "grad_norm": 1.9876651763916016, "learning_rate": 7.2476e-06, "loss": 0.0237, "step": 13763 }, { "epoch": 0.7338060457429226, "grad_norm": 2.6308658123016357, "learning_rate": 7.247400000000001e-06, "loss": 0.043, "step": 13764 }, { "epoch": 0.7338593591725756, "grad_norm": 1.5677634477615356, "learning_rate": 7.247200000000001e-06, "loss": 0.028, "step": 13765 }, { "epoch": 0.7339126726022285, "grad_norm": 1.4152485132217407, "learning_rate": 7.247000000000001e-06, "loss": 0.0183, "step": 13766 }, { "epoch": 0.7339659860318815, "grad_norm": 1.5936660766601562, "learning_rate": 7.246800000000001e-06, "loss": 0.0324, "step": 13767 }, { "epoch": 0.7340192994615343, "grad_norm": 3.781102418899536, "learning_rate": 7.2466e-06, "loss": 0.0424, "step": 13768 }, { "epoch": 0.7340726128911873, "grad_norm": 1.7908951044082642, "learning_rate": 7.2464e-06, "loss": 0.0344, "step": 13769 }, { "epoch": 0.7341259263208402, "grad_norm": 2.9449920654296875, "learning_rate": 7.246200000000001e-06, "loss": 0.0396, "step": 13770 }, { "epoch": 0.7341792397504932, "grad_norm": 1.2956055402755737, "learning_rate": 7.246000000000001e-06, "loss": 0.0159, "step": 13771 }, { "epoch": 0.7342325531801461, "grad_norm": 1.7960282564163208, "learning_rate": 7.245800000000001e-06, "loss": 0.0193, "step": 13772 }, { "epoch": 0.734285866609799, "grad_norm": 1.6446359157562256, "learning_rate": 7.2456000000000006e-06, "loss": 0.0278, "step": 13773 }, { "epoch": 0.734339180039452, "grad_norm": 1.4739421606063843, "learning_rate": 7.2454e-06, "loss": 0.0415, "step": 13774 }, { "epoch": 0.7343924934691048, "grad_norm": 3.055223226547241, "learning_rate": 7.245200000000001e-06, "loss": 0.0764, "step": 13775 }, { "epoch": 0.7344458068987578, "grad_norm": 1.7916598320007324, "learning_rate": 7.245000000000001e-06, "loss": 0.0198, "step": 13776 }, { "epoch": 0.7344991203284107, "grad_norm": 1.6618976593017578, "learning_rate": 7.2448000000000005e-06, "loss": 0.0164, "step": 13777 }, { "epoch": 0.7345524337580637, "grad_norm": 1.152793049812317, "learning_rate": 7.244600000000001e-06, "loss": 0.0223, "step": 13778 }, { "epoch": 0.7346057471877165, "grad_norm": 1.2713755369186401, "learning_rate": 7.2444000000000005e-06, "loss": 0.0103, "step": 13779 }, { "epoch": 0.7346590606173695, "grad_norm": 1.09878408908844, "learning_rate": 7.2442e-06, "loss": 0.023, "step": 13780 }, { "epoch": 0.7347123740470225, "grad_norm": 1.7623553276062012, "learning_rate": 7.244000000000001e-06, "loss": 0.024, "step": 13781 }, { "epoch": 0.7347656874766754, "grad_norm": 3.5268378257751465, "learning_rate": 7.243800000000001e-06, "loss": 0.0482, "step": 13782 }, { "epoch": 0.7348190009063283, "grad_norm": 2.371640205383301, "learning_rate": 7.2436e-06, "loss": 0.0355, "step": 13783 }, { "epoch": 0.7348723143359812, "grad_norm": 2.399840831756592, "learning_rate": 7.243400000000001e-06, "loss": 0.0442, "step": 13784 }, { "epoch": 0.7349256277656342, "grad_norm": 1.9327317476272583, "learning_rate": 7.2432e-06, "loss": 0.0271, "step": 13785 }, { "epoch": 0.7349789411952871, "grad_norm": 4.914543628692627, "learning_rate": 7.243000000000001e-06, "loss": 0.0589, "step": 13786 }, { "epoch": 0.73503225462494, "grad_norm": 5.333578109741211, "learning_rate": 7.242800000000001e-06, "loss": 0.0628, "step": 13787 }, { "epoch": 0.7350855680545929, "grad_norm": 2.086071014404297, "learning_rate": 7.242600000000001e-06, "loss": 0.0402, "step": 13788 }, { "epoch": 0.7351388814842459, "grad_norm": 2.6328320503234863, "learning_rate": 7.2424e-06, "loss": 0.0518, "step": 13789 }, { "epoch": 0.7351921949138989, "grad_norm": 3.451268434524536, "learning_rate": 7.242200000000001e-06, "loss": 0.0185, "step": 13790 }, { "epoch": 0.7352455083435517, "grad_norm": 1.350250005722046, "learning_rate": 7.242e-06, "loss": 0.0182, "step": 13791 }, { "epoch": 0.7352988217732047, "grad_norm": 2.1657605171203613, "learning_rate": 7.241800000000001e-06, "loss": 0.0339, "step": 13792 }, { "epoch": 0.7353521352028576, "grad_norm": 1.7178970575332642, "learning_rate": 7.241600000000001e-06, "loss": 0.0226, "step": 13793 }, { "epoch": 0.7354054486325106, "grad_norm": 2.5231759548187256, "learning_rate": 7.241400000000001e-06, "loss": 0.0394, "step": 13794 }, { "epoch": 0.7354587620621634, "grad_norm": 2.2099552154541016, "learning_rate": 7.2412e-06, "loss": 0.0159, "step": 13795 }, { "epoch": 0.7355120754918164, "grad_norm": 1.4952361583709717, "learning_rate": 7.241000000000001e-06, "loss": 0.0411, "step": 13796 }, { "epoch": 0.7355653889214693, "grad_norm": 0.85079026222229, "learning_rate": 7.2408e-06, "loss": 0.0082, "step": 13797 }, { "epoch": 0.7356187023511223, "grad_norm": 2.895308256149292, "learning_rate": 7.240600000000001e-06, "loss": 0.0458, "step": 13798 }, { "epoch": 0.7356720157807751, "grad_norm": 2.8360118865966797, "learning_rate": 7.240400000000001e-06, "loss": 0.0582, "step": 13799 }, { "epoch": 0.7357253292104281, "grad_norm": 2.067366361618042, "learning_rate": 7.240200000000001e-06, "loss": 0.0202, "step": 13800 }, { "epoch": 0.7357786426400811, "grad_norm": 2.122987747192383, "learning_rate": 7.24e-06, "loss": 0.0584, "step": 13801 }, { "epoch": 0.735831956069734, "grad_norm": 2.06754994392395, "learning_rate": 7.2398000000000005e-06, "loss": 0.0219, "step": 13802 }, { "epoch": 0.7358852694993869, "grad_norm": 4.98735237121582, "learning_rate": 7.239600000000001e-06, "loss": 0.0449, "step": 13803 }, { "epoch": 0.7359385829290398, "grad_norm": 2.0741331577301025, "learning_rate": 7.2394000000000005e-06, "loss": 0.0377, "step": 13804 }, { "epoch": 0.7359918963586928, "grad_norm": 1.9485995769500732, "learning_rate": 7.239200000000001e-06, "loss": 0.043, "step": 13805 }, { "epoch": 0.7360452097883456, "grad_norm": 4.7902936935424805, "learning_rate": 7.2390000000000005e-06, "loss": 0.051, "step": 13806 }, { "epoch": 0.7360985232179986, "grad_norm": 2.4543235301971436, "learning_rate": 7.2388e-06, "loss": 0.0193, "step": 13807 }, { "epoch": 0.7361518366476515, "grad_norm": 1.622754693031311, "learning_rate": 7.2386000000000005e-06, "loss": 0.0299, "step": 13808 }, { "epoch": 0.7362051500773045, "grad_norm": 1.918891429901123, "learning_rate": 7.238400000000001e-06, "loss": 0.0471, "step": 13809 }, { "epoch": 0.7362584635069575, "grad_norm": 1.8367663621902466, "learning_rate": 7.2382000000000004e-06, "loss": 0.0353, "step": 13810 }, { "epoch": 0.7363117769366103, "grad_norm": 1.88131582736969, "learning_rate": 7.238000000000001e-06, "loss": 0.0303, "step": 13811 }, { "epoch": 0.7363650903662633, "grad_norm": 2.3489482402801514, "learning_rate": 7.2378e-06, "loss": 0.0352, "step": 13812 }, { "epoch": 0.7364184037959162, "grad_norm": 2.581070899963379, "learning_rate": 7.2376e-06, "loss": 0.0372, "step": 13813 }, { "epoch": 0.7364717172255691, "grad_norm": 1.7102952003479004, "learning_rate": 7.237400000000001e-06, "loss": 0.0267, "step": 13814 }, { "epoch": 0.736525030655222, "grad_norm": 1.120276689529419, "learning_rate": 7.237200000000001e-06, "loss": 0.0131, "step": 13815 }, { "epoch": 0.736578344084875, "grad_norm": 1.535584807395935, "learning_rate": 7.237e-06, "loss": 0.0257, "step": 13816 }, { "epoch": 0.7366316575145279, "grad_norm": 1.31191086769104, "learning_rate": 7.236800000000001e-06, "loss": 0.0168, "step": 13817 }, { "epoch": 0.7366849709441808, "grad_norm": 1.2538329362869263, "learning_rate": 7.2366e-06, "loss": 0.0139, "step": 13818 }, { "epoch": 0.7367382843738338, "grad_norm": 1.4460721015930176, "learning_rate": 7.2364e-06, "loss": 0.0276, "step": 13819 }, { "epoch": 0.7367915978034867, "grad_norm": 2.1176633834838867, "learning_rate": 7.236200000000001e-06, "loss": 0.0321, "step": 13820 }, { "epoch": 0.7368449112331397, "grad_norm": 2.294797658920288, "learning_rate": 7.236000000000001e-06, "loss": 0.0377, "step": 13821 }, { "epoch": 0.7368982246627925, "grad_norm": 2.329883575439453, "learning_rate": 7.2358e-06, "loss": 0.0135, "step": 13822 }, { "epoch": 0.7369515380924455, "grad_norm": 2.327662229537964, "learning_rate": 7.235600000000001e-06, "loss": 0.043, "step": 13823 }, { "epoch": 0.7370048515220984, "grad_norm": 3.32327938079834, "learning_rate": 7.2354e-06, "loss": 0.0588, "step": 13824 }, { "epoch": 0.7370581649517514, "grad_norm": 1.8708592653274536, "learning_rate": 7.2352000000000015e-06, "loss": 0.0339, "step": 13825 }, { "epoch": 0.7371114783814042, "grad_norm": 2.7545387744903564, "learning_rate": 7.235000000000001e-06, "loss": 0.0587, "step": 13826 }, { "epoch": 0.7371647918110572, "grad_norm": 3.848644256591797, "learning_rate": 7.234800000000001e-06, "loss": 0.0297, "step": 13827 }, { "epoch": 0.7372181052407102, "grad_norm": 3.2653448581695557, "learning_rate": 7.2346e-06, "loss": 0.0643, "step": 13828 }, { "epoch": 0.737271418670363, "grad_norm": 1.6988166570663452, "learning_rate": 7.2344000000000006e-06, "loss": 0.0422, "step": 13829 }, { "epoch": 0.737324732100016, "grad_norm": 1.6802281141281128, "learning_rate": 7.2342e-06, "loss": 0.0287, "step": 13830 }, { "epoch": 0.7373780455296689, "grad_norm": 3.069047689437866, "learning_rate": 7.234000000000001e-06, "loss": 0.0354, "step": 13831 }, { "epoch": 0.7374313589593219, "grad_norm": 1.4167463779449463, "learning_rate": 7.233800000000001e-06, "loss": 0.0166, "step": 13832 }, { "epoch": 0.7374846723889747, "grad_norm": 1.0081855058670044, "learning_rate": 7.2336000000000005e-06, "loss": 0.0116, "step": 13833 }, { "epoch": 0.7375379858186277, "grad_norm": 1.5001108646392822, "learning_rate": 7.2334e-06, "loss": 0.024, "step": 13834 }, { "epoch": 0.7375912992482806, "grad_norm": 2.244288444519043, "learning_rate": 7.2332000000000005e-06, "loss": 0.0394, "step": 13835 }, { "epoch": 0.7376446126779336, "grad_norm": 1.0897048711776733, "learning_rate": 7.233000000000001e-06, "loss": 0.0143, "step": 13836 }, { "epoch": 0.7376979261075866, "grad_norm": 3.5371758937835693, "learning_rate": 7.232800000000001e-06, "loss": 0.0577, "step": 13837 }, { "epoch": 0.7377512395372394, "grad_norm": 3.6967103481292725, "learning_rate": 7.232600000000001e-06, "loss": 0.0283, "step": 13838 }, { "epoch": 0.7378045529668924, "grad_norm": 2.903904676437378, "learning_rate": 7.2324e-06, "loss": 0.0438, "step": 13839 }, { "epoch": 0.7378578663965453, "grad_norm": 1.7855746746063232, "learning_rate": 7.2322e-06, "loss": 0.0317, "step": 13840 }, { "epoch": 0.7379111798261982, "grad_norm": 1.4104713201522827, "learning_rate": 7.232e-06, "loss": 0.0144, "step": 13841 }, { "epoch": 0.7379644932558511, "grad_norm": 3.60675048828125, "learning_rate": 7.231800000000001e-06, "loss": 0.0408, "step": 13842 }, { "epoch": 0.7380178066855041, "grad_norm": 1.572642207145691, "learning_rate": 7.231600000000001e-06, "loss": 0.0205, "step": 13843 }, { "epoch": 0.738071120115157, "grad_norm": 1.0304170846939087, "learning_rate": 7.231400000000001e-06, "loss": 0.0089, "step": 13844 }, { "epoch": 0.7381244335448099, "grad_norm": 2.240360975265503, "learning_rate": 7.2312e-06, "loss": 0.0487, "step": 13845 }, { "epoch": 0.7381777469744628, "grad_norm": 3.094099283218384, "learning_rate": 7.231e-06, "loss": 0.0407, "step": 13846 }, { "epoch": 0.7382310604041158, "grad_norm": 2.418274402618408, "learning_rate": 7.2308e-06, "loss": 0.0264, "step": 13847 }, { "epoch": 0.7382843738337688, "grad_norm": 2.3312315940856934, "learning_rate": 7.230600000000001e-06, "loss": 0.0173, "step": 13848 }, { "epoch": 0.7383376872634216, "grad_norm": 1.9330782890319824, "learning_rate": 7.230400000000001e-06, "loss": 0.0211, "step": 13849 }, { "epoch": 0.7383910006930746, "grad_norm": 3.424224615097046, "learning_rate": 7.230200000000001e-06, "loss": 0.0415, "step": 13850 }, { "epoch": 0.7384443141227275, "grad_norm": 3.13919997215271, "learning_rate": 7.23e-06, "loss": 0.0608, "step": 13851 }, { "epoch": 0.7384976275523805, "grad_norm": 5.075433254241943, "learning_rate": 7.2298e-06, "loss": 0.0268, "step": 13852 }, { "epoch": 0.7385509409820333, "grad_norm": 1.0855777263641357, "learning_rate": 7.229600000000001e-06, "loss": 0.0197, "step": 13853 }, { "epoch": 0.7386042544116863, "grad_norm": 2.0624029636383057, "learning_rate": 7.229400000000001e-06, "loss": 0.0443, "step": 13854 }, { "epoch": 0.7386575678413392, "grad_norm": 4.174713134765625, "learning_rate": 7.229200000000001e-06, "loss": 0.021, "step": 13855 }, { "epoch": 0.7387108812709922, "grad_norm": 2.1975107192993164, "learning_rate": 7.229000000000001e-06, "loss": 0.0212, "step": 13856 }, { "epoch": 0.7387641947006451, "grad_norm": 1.5168030261993408, "learning_rate": 7.2288e-06, "loss": 0.0313, "step": 13857 }, { "epoch": 0.738817508130298, "grad_norm": 2.169175148010254, "learning_rate": 7.2286e-06, "loss": 0.0303, "step": 13858 }, { "epoch": 0.738870821559951, "grad_norm": 2.593193531036377, "learning_rate": 7.228400000000001e-06, "loss": 0.0595, "step": 13859 }, { "epoch": 0.7389241349896039, "grad_norm": 1.8568681478500366, "learning_rate": 7.2282000000000005e-06, "loss": 0.0652, "step": 13860 }, { "epoch": 0.7389774484192568, "grad_norm": 3.449186325073242, "learning_rate": 7.228000000000001e-06, "loss": 0.0271, "step": 13861 }, { "epoch": 0.7390307618489097, "grad_norm": 1.891920566558838, "learning_rate": 7.2278000000000005e-06, "loss": 0.0201, "step": 13862 }, { "epoch": 0.7390840752785627, "grad_norm": 2.968932628631592, "learning_rate": 7.2276e-06, "loss": 0.0234, "step": 13863 }, { "epoch": 0.7391373887082155, "grad_norm": 3.0412211418151855, "learning_rate": 7.227400000000001e-06, "loss": 0.0348, "step": 13864 }, { "epoch": 0.7391907021378685, "grad_norm": 2.4244837760925293, "learning_rate": 7.227200000000001e-06, "loss": 0.0588, "step": 13865 }, { "epoch": 0.7392440155675215, "grad_norm": 1.473839521408081, "learning_rate": 7.2270000000000004e-06, "loss": 0.0143, "step": 13866 }, { "epoch": 0.7392973289971744, "grad_norm": 1.6453417539596558, "learning_rate": 7.226800000000001e-06, "loss": 0.0297, "step": 13867 }, { "epoch": 0.7393506424268274, "grad_norm": 2.8965790271759033, "learning_rate": 7.2266e-06, "loss": 0.0604, "step": 13868 }, { "epoch": 0.7394039558564802, "grad_norm": 2.8752520084381104, "learning_rate": 7.2264e-06, "loss": 0.033, "step": 13869 }, { "epoch": 0.7394572692861332, "grad_norm": 2.097154378890991, "learning_rate": 7.226200000000001e-06, "loss": 0.0299, "step": 13870 }, { "epoch": 0.7395105827157861, "grad_norm": 3.756423234939575, "learning_rate": 7.226000000000001e-06, "loss": 0.0425, "step": 13871 }, { "epoch": 0.739563896145439, "grad_norm": 1.4891173839569092, "learning_rate": 7.2258e-06, "loss": 0.0088, "step": 13872 }, { "epoch": 0.7396172095750919, "grad_norm": 1.2624813318252563, "learning_rate": 7.225600000000001e-06, "loss": 0.0141, "step": 13873 }, { "epoch": 0.7396705230047449, "grad_norm": 1.2697147130966187, "learning_rate": 7.2254e-06, "loss": 0.0135, "step": 13874 }, { "epoch": 0.7397238364343979, "grad_norm": 2.2743823528289795, "learning_rate": 7.225200000000001e-06, "loss": 0.0238, "step": 13875 }, { "epoch": 0.7397771498640507, "grad_norm": 0.790819525718689, "learning_rate": 7.225000000000001e-06, "loss": 0.0079, "step": 13876 }, { "epoch": 0.7398304632937037, "grad_norm": 1.228713035583496, "learning_rate": 7.224800000000001e-06, "loss": 0.012, "step": 13877 }, { "epoch": 0.7398837767233566, "grad_norm": 1.7978588342666626, "learning_rate": 7.2246e-06, "loss": 0.0213, "step": 13878 }, { "epoch": 0.7399370901530096, "grad_norm": 2.0498862266540527, "learning_rate": 7.224400000000001e-06, "loss": 0.0156, "step": 13879 }, { "epoch": 0.7399904035826624, "grad_norm": 1.9796572923660278, "learning_rate": 7.2242e-06, "loss": 0.029, "step": 13880 }, { "epoch": 0.7400437170123154, "grad_norm": 4.261348247528076, "learning_rate": 7.224000000000001e-06, "loss": 0.0369, "step": 13881 }, { "epoch": 0.7400970304419683, "grad_norm": 0.4579518139362335, "learning_rate": 7.223800000000001e-06, "loss": 0.0033, "step": 13882 }, { "epoch": 0.7401503438716213, "grad_norm": 1.7353131771087646, "learning_rate": 7.223600000000001e-06, "loss": 0.0305, "step": 13883 }, { "epoch": 0.7402036573012742, "grad_norm": 2.5082626342773438, "learning_rate": 7.2234e-06, "loss": 0.0387, "step": 13884 }, { "epoch": 0.7402569707309271, "grad_norm": 2.440997362136841, "learning_rate": 7.2232000000000006e-06, "loss": 0.0125, "step": 13885 }, { "epoch": 0.7403102841605801, "grad_norm": 2.545785665512085, "learning_rate": 7.223000000000001e-06, "loss": 0.0458, "step": 13886 }, { "epoch": 0.740363597590233, "grad_norm": 2.5428552627563477, "learning_rate": 7.2228000000000005e-06, "loss": 0.0409, "step": 13887 }, { "epoch": 0.7404169110198859, "grad_norm": 0.9045555591583252, "learning_rate": 7.222600000000001e-06, "loss": 0.0102, "step": 13888 }, { "epoch": 0.7404702244495388, "grad_norm": 2.048178195953369, "learning_rate": 7.2224000000000005e-06, "loss": 0.0351, "step": 13889 }, { "epoch": 0.7405235378791918, "grad_norm": 1.0072779655456543, "learning_rate": 7.2222e-06, "loss": 0.0094, "step": 13890 }, { "epoch": 0.7405768513088447, "grad_norm": 1.6896286010742188, "learning_rate": 7.2220000000000005e-06, "loss": 0.026, "step": 13891 }, { "epoch": 0.7406301647384976, "grad_norm": 2.8522350788116455, "learning_rate": 7.221800000000001e-06, "loss": 0.0787, "step": 13892 }, { "epoch": 0.7406834781681506, "grad_norm": 1.177111268043518, "learning_rate": 7.2216000000000004e-06, "loss": 0.0069, "step": 13893 }, { "epoch": 0.7407367915978035, "grad_norm": 1.3066997528076172, "learning_rate": 7.221400000000001e-06, "loss": 0.0128, "step": 13894 }, { "epoch": 0.7407901050274565, "grad_norm": 1.3359602689743042, "learning_rate": 7.2212e-06, "loss": 0.0241, "step": 13895 }, { "epoch": 0.7408434184571093, "grad_norm": 2.3232173919677734, "learning_rate": 7.221e-06, "loss": 0.0237, "step": 13896 }, { "epoch": 0.7408967318867623, "grad_norm": 1.3238718509674072, "learning_rate": 7.220800000000001e-06, "loss": 0.0198, "step": 13897 }, { "epoch": 0.7409500453164152, "grad_norm": 2.017749309539795, "learning_rate": 7.220600000000001e-06, "loss": 0.0148, "step": 13898 }, { "epoch": 0.7410033587460682, "grad_norm": 1.4402292966842651, "learning_rate": 7.2204e-06, "loss": 0.0217, "step": 13899 }, { "epoch": 0.741056672175721, "grad_norm": 3.56986403465271, "learning_rate": 7.220200000000001e-06, "loss": 0.0386, "step": 13900 }, { "epoch": 0.741109985605374, "grad_norm": 2.309514045715332, "learning_rate": 7.22e-06, "loss": 0.0335, "step": 13901 }, { "epoch": 0.7411632990350269, "grad_norm": 2.690234899520874, "learning_rate": 7.2198e-06, "loss": 0.038, "step": 13902 }, { "epoch": 0.7412166124646798, "grad_norm": 2.271268606185913, "learning_rate": 7.219600000000001e-06, "loss": 0.0202, "step": 13903 }, { "epoch": 0.7412699258943328, "grad_norm": 2.041053533554077, "learning_rate": 7.219400000000001e-06, "loss": 0.0195, "step": 13904 }, { "epoch": 0.7413232393239857, "grad_norm": 2.3338959217071533, "learning_rate": 7.2192e-06, "loss": 0.0585, "step": 13905 }, { "epoch": 0.7413765527536387, "grad_norm": 1.6536767482757568, "learning_rate": 7.219000000000001e-06, "loss": 0.0248, "step": 13906 }, { "epoch": 0.7414298661832915, "grad_norm": 1.2757716178894043, "learning_rate": 7.2188e-06, "loss": 0.0129, "step": 13907 }, { "epoch": 0.7414831796129445, "grad_norm": 3.20806884765625, "learning_rate": 7.2186e-06, "loss": 0.0348, "step": 13908 }, { "epoch": 0.7415364930425974, "grad_norm": 0.5581604242324829, "learning_rate": 7.218400000000001e-06, "loss": 0.005, "step": 13909 }, { "epoch": 0.7415898064722504, "grad_norm": 2.6188204288482666, "learning_rate": 7.218200000000001e-06, "loss": 0.036, "step": 13910 }, { "epoch": 0.7416431199019032, "grad_norm": 1.9361023902893066, "learning_rate": 7.218e-06, "loss": 0.0504, "step": 13911 }, { "epoch": 0.7416964333315562, "grad_norm": 2.6480283737182617, "learning_rate": 7.217800000000001e-06, "loss": 0.046, "step": 13912 }, { "epoch": 0.7417497467612092, "grad_norm": 2.3672068119049072, "learning_rate": 7.2176e-06, "loss": 0.0407, "step": 13913 }, { "epoch": 0.7418030601908621, "grad_norm": 1.9817461967468262, "learning_rate": 7.217400000000001e-06, "loss": 0.0158, "step": 13914 }, { "epoch": 0.741856373620515, "grad_norm": 1.0415970087051392, "learning_rate": 7.217200000000001e-06, "loss": 0.0064, "step": 13915 }, { "epoch": 0.7419096870501679, "grad_norm": 1.3099052906036377, "learning_rate": 7.2170000000000005e-06, "loss": 0.0155, "step": 13916 }, { "epoch": 0.7419630004798209, "grad_norm": 1.8748059272766113, "learning_rate": 7.2168e-06, "loss": 0.0173, "step": 13917 }, { "epoch": 0.7420163139094738, "grad_norm": 2.2454476356506348, "learning_rate": 7.2166000000000005e-06, "loss": 0.0217, "step": 13918 }, { "epoch": 0.7420696273391267, "grad_norm": 1.3751106262207031, "learning_rate": 7.2164e-06, "loss": 0.0236, "step": 13919 }, { "epoch": 0.7421229407687796, "grad_norm": 2.0383191108703613, "learning_rate": 7.216200000000001e-06, "loss": 0.0202, "step": 13920 }, { "epoch": 0.7421762541984326, "grad_norm": 1.0345244407653809, "learning_rate": 7.216000000000001e-06, "loss": 0.0165, "step": 13921 }, { "epoch": 0.7422295676280856, "grad_norm": 3.02813458442688, "learning_rate": 7.2158000000000004e-06, "loss": 0.0405, "step": 13922 }, { "epoch": 0.7422828810577384, "grad_norm": 2.6939566135406494, "learning_rate": 7.2156e-06, "loss": 0.0325, "step": 13923 }, { "epoch": 0.7423361944873914, "grad_norm": 3.317256212234497, "learning_rate": 7.2154e-06, "loss": 0.0519, "step": 13924 }, { "epoch": 0.7423895079170443, "grad_norm": 1.7844730615615845, "learning_rate": 7.215200000000001e-06, "loss": 0.0195, "step": 13925 }, { "epoch": 0.7424428213466973, "grad_norm": 2.44594144821167, "learning_rate": 7.215000000000001e-06, "loss": 0.0227, "step": 13926 }, { "epoch": 0.7424961347763501, "grad_norm": 2.070741891860962, "learning_rate": 7.214800000000001e-06, "loss": 0.0316, "step": 13927 }, { "epoch": 0.7425494482060031, "grad_norm": 2.128067970275879, "learning_rate": 7.2146e-06, "loss": 0.0203, "step": 13928 }, { "epoch": 0.742602761635656, "grad_norm": 4.249642848968506, "learning_rate": 7.2144e-06, "loss": 0.0181, "step": 13929 }, { "epoch": 0.742656075065309, "grad_norm": 3.3014166355133057, "learning_rate": 7.2142e-06, "loss": 0.0705, "step": 13930 }, { "epoch": 0.7427093884949619, "grad_norm": 1.6390130519866943, "learning_rate": 7.214000000000001e-06, "loss": 0.0177, "step": 13931 }, { "epoch": 0.7427627019246148, "grad_norm": 2.3600046634674072, "learning_rate": 7.213800000000001e-06, "loss": 0.0289, "step": 13932 }, { "epoch": 0.7428160153542678, "grad_norm": 2.8955960273742676, "learning_rate": 7.213600000000001e-06, "loss": 0.0236, "step": 13933 }, { "epoch": 0.7428693287839206, "grad_norm": 4.048666000366211, "learning_rate": 7.2134e-06, "loss": 0.0486, "step": 13934 }, { "epoch": 0.7429226422135736, "grad_norm": 3.588679075241089, "learning_rate": 7.2132e-06, "loss": 0.0764, "step": 13935 }, { "epoch": 0.7429759556432265, "grad_norm": 3.3157546520233154, "learning_rate": 7.213000000000001e-06, "loss": 0.0364, "step": 13936 }, { "epoch": 0.7430292690728795, "grad_norm": 1.8131701946258545, "learning_rate": 7.212800000000001e-06, "loss": 0.0168, "step": 13937 }, { "epoch": 0.7430825825025323, "grad_norm": 3.511089563369751, "learning_rate": 7.212600000000001e-06, "loss": 0.029, "step": 13938 }, { "epoch": 0.7431358959321853, "grad_norm": 2.0778539180755615, "learning_rate": 7.212400000000001e-06, "loss": 0.0518, "step": 13939 }, { "epoch": 0.7431892093618383, "grad_norm": 1.862931728363037, "learning_rate": 7.2122e-06, "loss": 0.0166, "step": 13940 }, { "epoch": 0.7432425227914912, "grad_norm": 2.149439573287964, "learning_rate": 7.212e-06, "loss": 0.0435, "step": 13941 }, { "epoch": 0.7432958362211441, "grad_norm": 1.578444242477417, "learning_rate": 7.211800000000001e-06, "loss": 0.0384, "step": 13942 }, { "epoch": 0.743349149650797, "grad_norm": 1.1608506441116333, "learning_rate": 7.2116000000000005e-06, "loss": 0.0129, "step": 13943 }, { "epoch": 0.74340246308045, "grad_norm": 1.8657275438308716, "learning_rate": 7.211400000000001e-06, "loss": 0.0371, "step": 13944 }, { "epoch": 0.7434557765101029, "grad_norm": 1.7540642023086548, "learning_rate": 7.2112000000000005e-06, "loss": 0.0247, "step": 13945 }, { "epoch": 0.7435090899397558, "grad_norm": 2.103634834289551, "learning_rate": 7.211e-06, "loss": 0.0259, "step": 13946 }, { "epoch": 0.7435624033694087, "grad_norm": 3.6533772945404053, "learning_rate": 7.210800000000001e-06, "loss": 0.0291, "step": 13947 }, { "epoch": 0.7436157167990617, "grad_norm": 3.013265609741211, "learning_rate": 7.210600000000001e-06, "loss": 0.0413, "step": 13948 }, { "epoch": 0.7436690302287147, "grad_norm": 3.124753713607788, "learning_rate": 7.2104000000000005e-06, "loss": 0.0485, "step": 13949 }, { "epoch": 0.7437223436583675, "grad_norm": 2.0243401527404785, "learning_rate": 7.210200000000001e-06, "loss": 0.0377, "step": 13950 }, { "epoch": 0.7437756570880205, "grad_norm": 6.519991874694824, "learning_rate": 7.2100000000000004e-06, "loss": 0.0332, "step": 13951 }, { "epoch": 0.7438289705176734, "grad_norm": 1.7578134536743164, "learning_rate": 7.2098e-06, "loss": 0.0188, "step": 13952 }, { "epoch": 0.7438822839473264, "grad_norm": 1.457613229751587, "learning_rate": 7.209600000000001e-06, "loss": 0.0136, "step": 13953 }, { "epoch": 0.7439355973769792, "grad_norm": 5.138843536376953, "learning_rate": 7.209400000000001e-06, "loss": 0.0423, "step": 13954 }, { "epoch": 0.7439889108066322, "grad_norm": 1.4366974830627441, "learning_rate": 7.2092e-06, "loss": 0.0242, "step": 13955 }, { "epoch": 0.7440422242362851, "grad_norm": 2.530575752258301, "learning_rate": 7.209000000000001e-06, "loss": 0.0237, "step": 13956 }, { "epoch": 0.7440955376659381, "grad_norm": 2.3432443141937256, "learning_rate": 7.2088e-06, "loss": 0.0422, "step": 13957 }, { "epoch": 0.7441488510955909, "grad_norm": 2.305814027786255, "learning_rate": 7.2086e-06, "loss": 0.0339, "step": 13958 }, { "epoch": 0.7442021645252439, "grad_norm": 2.768177032470703, "learning_rate": 7.208400000000001e-06, "loss": 0.0528, "step": 13959 }, { "epoch": 0.7442554779548969, "grad_norm": 1.882768988609314, "learning_rate": 7.208200000000001e-06, "loss": 0.0245, "step": 13960 }, { "epoch": 0.7443087913845498, "grad_norm": 3.0028631687164307, "learning_rate": 7.208e-06, "loss": 0.0407, "step": 13961 }, { "epoch": 0.7443621048142027, "grad_norm": 2.387443780899048, "learning_rate": 7.207800000000001e-06, "loss": 0.0257, "step": 13962 }, { "epoch": 0.7444154182438556, "grad_norm": 1.695137858390808, "learning_rate": 7.2076e-06, "loss": 0.0183, "step": 13963 }, { "epoch": 0.7444687316735086, "grad_norm": 3.6726763248443604, "learning_rate": 7.207400000000001e-06, "loss": 0.033, "step": 13964 }, { "epoch": 0.7445220451031614, "grad_norm": 2.2573587894439697, "learning_rate": 7.207200000000001e-06, "loss": 0.0582, "step": 13965 }, { "epoch": 0.7445753585328144, "grad_norm": 2.1016845703125, "learning_rate": 7.207000000000001e-06, "loss": 0.0331, "step": 13966 }, { "epoch": 0.7446286719624673, "grad_norm": 3.7783710956573486, "learning_rate": 7.2068e-06, "loss": 0.0684, "step": 13967 }, { "epoch": 0.7446819853921203, "grad_norm": 1.428181767463684, "learning_rate": 7.206600000000001e-06, "loss": 0.0187, "step": 13968 }, { "epoch": 0.7447352988217733, "grad_norm": 3.9161150455474854, "learning_rate": 7.2064e-06, "loss": 0.0501, "step": 13969 }, { "epoch": 0.7447886122514261, "grad_norm": 2.7855224609375, "learning_rate": 7.2062000000000006e-06, "loss": 0.0232, "step": 13970 }, { "epoch": 0.7448419256810791, "grad_norm": 1.731054663658142, "learning_rate": 7.206000000000001e-06, "loss": 0.029, "step": 13971 }, { "epoch": 0.744895239110732, "grad_norm": 1.422160267829895, "learning_rate": 7.2058000000000005e-06, "loss": 0.0423, "step": 13972 }, { "epoch": 0.744948552540385, "grad_norm": 0.44922158122062683, "learning_rate": 7.2056e-06, "loss": 0.0034, "step": 13973 }, { "epoch": 0.7450018659700378, "grad_norm": 2.0425515174865723, "learning_rate": 7.2054000000000005e-06, "loss": 0.0612, "step": 13974 }, { "epoch": 0.7450551793996908, "grad_norm": 2.2535781860351562, "learning_rate": 7.205200000000001e-06, "loss": 0.0233, "step": 13975 }, { "epoch": 0.7451084928293437, "grad_norm": 1.3518224954605103, "learning_rate": 7.2050000000000005e-06, "loss": 0.0106, "step": 13976 }, { "epoch": 0.7451618062589966, "grad_norm": 2.1145195960998535, "learning_rate": 7.204800000000001e-06, "loss": 0.0504, "step": 13977 }, { "epoch": 0.7452151196886496, "grad_norm": 2.3117547035217285, "learning_rate": 7.2046000000000004e-06, "loss": 0.0342, "step": 13978 }, { "epoch": 0.7452684331183025, "grad_norm": 1.1818841695785522, "learning_rate": 7.2044e-06, "loss": 0.0242, "step": 13979 }, { "epoch": 0.7453217465479555, "grad_norm": 1.8566584587097168, "learning_rate": 7.2042e-06, "loss": 0.0203, "step": 13980 }, { "epoch": 0.7453750599776083, "grad_norm": 1.1432533264160156, "learning_rate": 7.204000000000001e-06, "loss": 0.0135, "step": 13981 }, { "epoch": 0.7454283734072613, "grad_norm": 2.1615793704986572, "learning_rate": 7.2038e-06, "loss": 0.0522, "step": 13982 }, { "epoch": 0.7454816868369142, "grad_norm": 1.4859508275985718, "learning_rate": 7.203600000000001e-06, "loss": 0.0127, "step": 13983 }, { "epoch": 0.7455350002665672, "grad_norm": 1.69248366355896, "learning_rate": 7.2034e-06, "loss": 0.0182, "step": 13984 }, { "epoch": 0.74558831369622, "grad_norm": 2.324944257736206, "learning_rate": 7.2032e-06, "loss": 0.0195, "step": 13985 }, { "epoch": 0.745641627125873, "grad_norm": 1.0602467060089111, "learning_rate": 7.203000000000001e-06, "loss": 0.0087, "step": 13986 }, { "epoch": 0.745694940555526, "grad_norm": 2.429318904876709, "learning_rate": 7.202800000000001e-06, "loss": 0.0368, "step": 13987 }, { "epoch": 0.7457482539851789, "grad_norm": 0.6294770240783691, "learning_rate": 7.2026e-06, "loss": 0.0068, "step": 13988 }, { "epoch": 0.7458015674148318, "grad_norm": 1.0604029893875122, "learning_rate": 7.202400000000001e-06, "loss": 0.0111, "step": 13989 }, { "epoch": 0.7458548808444847, "grad_norm": 1.5452146530151367, "learning_rate": 7.2022e-06, "loss": 0.0154, "step": 13990 }, { "epoch": 0.7459081942741377, "grad_norm": 3.806213140487671, "learning_rate": 7.202e-06, "loss": 0.0685, "step": 13991 }, { "epoch": 0.7459615077037905, "grad_norm": 2.7546849250793457, "learning_rate": 7.201800000000001e-06, "loss": 0.0243, "step": 13992 }, { "epoch": 0.7460148211334435, "grad_norm": 3.24830961227417, "learning_rate": 7.201600000000001e-06, "loss": 0.0463, "step": 13993 }, { "epoch": 0.7460681345630964, "grad_norm": 1.0089614391326904, "learning_rate": 7.2014e-06, "loss": 0.0155, "step": 13994 }, { "epoch": 0.7461214479927494, "grad_norm": 3.0728957653045654, "learning_rate": 7.201200000000001e-06, "loss": 0.0316, "step": 13995 }, { "epoch": 0.7461747614224024, "grad_norm": 3.414435863494873, "learning_rate": 7.201e-06, "loss": 0.0404, "step": 13996 }, { "epoch": 0.7462280748520552, "grad_norm": 1.7624504566192627, "learning_rate": 7.2008000000000014e-06, "loss": 0.0322, "step": 13997 }, { "epoch": 0.7462813882817082, "grad_norm": 3.0038414001464844, "learning_rate": 7.200600000000001e-06, "loss": 0.0525, "step": 13998 }, { "epoch": 0.7463347017113611, "grad_norm": 2.098994731903076, "learning_rate": 7.2004000000000006e-06, "loss": 0.0262, "step": 13999 }, { "epoch": 0.746388015141014, "grad_norm": 2.7071409225463867, "learning_rate": 7.2002e-06, "loss": 0.0546, "step": 14000 }, { "epoch": 0.7464413285706669, "grad_norm": 2.084761381149292, "learning_rate": 7.2000000000000005e-06, "loss": 0.0207, "step": 14001 }, { "epoch": 0.7464946420003199, "grad_norm": 1.5325994491577148, "learning_rate": 7.1998e-06, "loss": 0.0132, "step": 14002 }, { "epoch": 0.7465479554299728, "grad_norm": 2.955993413925171, "learning_rate": 7.199600000000001e-06, "loss": 0.042, "step": 14003 }, { "epoch": 0.7466012688596257, "grad_norm": 4.568179607391357, "learning_rate": 7.199400000000001e-06, "loss": 0.037, "step": 14004 }, { "epoch": 0.7466545822892787, "grad_norm": 0.9640077352523804, "learning_rate": 7.1992000000000005e-06, "loss": 0.0068, "step": 14005 }, { "epoch": 0.7467078957189316, "grad_norm": 2.2664506435394287, "learning_rate": 7.199e-06, "loss": 0.03, "step": 14006 }, { "epoch": 0.7467612091485846, "grad_norm": 1.7385274171829224, "learning_rate": 7.1988000000000004e-06, "loss": 0.0457, "step": 14007 }, { "epoch": 0.7468145225782374, "grad_norm": 1.032836675643921, "learning_rate": 7.198600000000001e-06, "loss": 0.0163, "step": 14008 }, { "epoch": 0.7468678360078904, "grad_norm": 1.3053569793701172, "learning_rate": 7.198400000000001e-06, "loss": 0.0114, "step": 14009 }, { "epoch": 0.7469211494375433, "grad_norm": 1.8620989322662354, "learning_rate": 7.198200000000001e-06, "loss": 0.0468, "step": 14010 }, { "epoch": 0.7469744628671963, "grad_norm": 1.903643250465393, "learning_rate": 7.198e-06, "loss": 0.0402, "step": 14011 }, { "epoch": 0.7470277762968491, "grad_norm": 2.7680482864379883, "learning_rate": 7.1978e-06, "loss": 0.035, "step": 14012 }, { "epoch": 0.7470810897265021, "grad_norm": 2.025313377380371, "learning_rate": 7.1976e-06, "loss": 0.0365, "step": 14013 }, { "epoch": 0.747134403156155, "grad_norm": 1.6408491134643555, "learning_rate": 7.197400000000001e-06, "loss": 0.0411, "step": 14014 }, { "epoch": 0.747187716585808, "grad_norm": 1.877159595489502, "learning_rate": 7.197200000000001e-06, "loss": 0.0156, "step": 14015 }, { "epoch": 0.7472410300154609, "grad_norm": 3.226996421813965, "learning_rate": 7.197000000000001e-06, "loss": 0.0232, "step": 14016 }, { "epoch": 0.7472943434451138, "grad_norm": 2.967423439025879, "learning_rate": 7.1968e-06, "loss": 0.0269, "step": 14017 }, { "epoch": 0.7473476568747668, "grad_norm": 1.7055455446243286, "learning_rate": 7.1966e-06, "loss": 0.0131, "step": 14018 }, { "epoch": 0.7474009703044197, "grad_norm": 2.1516880989074707, "learning_rate": 7.1964e-06, "loss": 0.0351, "step": 14019 }, { "epoch": 0.7474542837340726, "grad_norm": 1.44758141040802, "learning_rate": 7.196200000000001e-06, "loss": 0.0213, "step": 14020 }, { "epoch": 0.7475075971637255, "grad_norm": 2.7143771648406982, "learning_rate": 7.196000000000001e-06, "loss": 0.0781, "step": 14021 }, { "epoch": 0.7475609105933785, "grad_norm": 2.5747601985931396, "learning_rate": 7.195800000000001e-06, "loss": 0.0307, "step": 14022 }, { "epoch": 0.7476142240230313, "grad_norm": 2.9644508361816406, "learning_rate": 7.1956e-06, "loss": 0.0425, "step": 14023 }, { "epoch": 0.7476675374526843, "grad_norm": 2.164503335952759, "learning_rate": 7.1954e-06, "loss": 0.0239, "step": 14024 }, { "epoch": 0.7477208508823373, "grad_norm": 2.045621633529663, "learning_rate": 7.195200000000001e-06, "loss": 0.0476, "step": 14025 }, { "epoch": 0.7477741643119902, "grad_norm": 1.7195751667022705, "learning_rate": 7.1950000000000006e-06, "loss": 0.0322, "step": 14026 }, { "epoch": 0.7478274777416432, "grad_norm": 1.874049425125122, "learning_rate": 7.194800000000001e-06, "loss": 0.0247, "step": 14027 }, { "epoch": 0.747880791171296, "grad_norm": 2.849156618118286, "learning_rate": 7.1946000000000005e-06, "loss": 0.0341, "step": 14028 }, { "epoch": 0.747934104600949, "grad_norm": 2.631192207336426, "learning_rate": 7.1944e-06, "loss": 0.0234, "step": 14029 }, { "epoch": 0.7479874180306019, "grad_norm": 1.6447336673736572, "learning_rate": 7.1942e-06, "loss": 0.0314, "step": 14030 }, { "epoch": 0.7480407314602548, "grad_norm": 2.3165130615234375, "learning_rate": 7.194000000000001e-06, "loss": 0.0164, "step": 14031 }, { "epoch": 0.7480940448899077, "grad_norm": 2.1057286262512207, "learning_rate": 7.1938000000000005e-06, "loss": 0.0343, "step": 14032 }, { "epoch": 0.7481473583195607, "grad_norm": 2.5491576194763184, "learning_rate": 7.193600000000001e-06, "loss": 0.0462, "step": 14033 }, { "epoch": 0.7482006717492137, "grad_norm": 1.1773278713226318, "learning_rate": 7.1934000000000004e-06, "loss": 0.0255, "step": 14034 }, { "epoch": 0.7482539851788665, "grad_norm": 3.5865657329559326, "learning_rate": 7.1932e-06, "loss": 0.0807, "step": 14035 }, { "epoch": 0.7483072986085195, "grad_norm": 1.4369627237319946, "learning_rate": 7.193000000000001e-06, "loss": 0.0219, "step": 14036 }, { "epoch": 0.7483606120381724, "grad_norm": 1.370061993598938, "learning_rate": 7.192800000000001e-06, "loss": 0.0208, "step": 14037 }, { "epoch": 0.7484139254678254, "grad_norm": 3.985387086868286, "learning_rate": 7.1926e-06, "loss": 0.0613, "step": 14038 }, { "epoch": 0.7484672388974782, "grad_norm": 1.1887574195861816, "learning_rate": 7.192400000000001e-06, "loss": 0.0195, "step": 14039 }, { "epoch": 0.7485205523271312, "grad_norm": 3.0095410346984863, "learning_rate": 7.1922e-06, "loss": 0.0195, "step": 14040 }, { "epoch": 0.7485738657567841, "grad_norm": 2.080467939376831, "learning_rate": 7.192e-06, "loss": 0.0327, "step": 14041 }, { "epoch": 0.7486271791864371, "grad_norm": 2.1717519760131836, "learning_rate": 7.191800000000001e-06, "loss": 0.0199, "step": 14042 }, { "epoch": 0.74868049261609, "grad_norm": 4.497096061706543, "learning_rate": 7.191600000000001e-06, "loss": 0.0848, "step": 14043 }, { "epoch": 0.7487338060457429, "grad_norm": 1.440838098526001, "learning_rate": 7.1914e-06, "loss": 0.0248, "step": 14044 }, { "epoch": 0.7487871194753959, "grad_norm": 2.48970103263855, "learning_rate": 7.191200000000001e-06, "loss": 0.0594, "step": 14045 }, { "epoch": 0.7488404329050488, "grad_norm": 2.2229318618774414, "learning_rate": 7.191e-06, "loss": 0.0398, "step": 14046 }, { "epoch": 0.7488937463347017, "grad_norm": 2.4567832946777344, "learning_rate": 7.1908000000000015e-06, "loss": 0.0349, "step": 14047 }, { "epoch": 0.7489470597643546, "grad_norm": 2.007215976715088, "learning_rate": 7.190600000000001e-06, "loss": 0.0204, "step": 14048 }, { "epoch": 0.7490003731940076, "grad_norm": 1.6651712656021118, "learning_rate": 7.190400000000001e-06, "loss": 0.0279, "step": 14049 }, { "epoch": 0.7490536866236605, "grad_norm": 1.6547735929489136, "learning_rate": 7.1902e-06, "loss": 0.0303, "step": 14050 }, { "epoch": 0.7491070000533134, "grad_norm": 3.6243319511413574, "learning_rate": 7.190000000000001e-06, "loss": 0.0394, "step": 14051 }, { "epoch": 0.7491603134829664, "grad_norm": 0.8386621475219727, "learning_rate": 7.1898e-06, "loss": 0.0112, "step": 14052 }, { "epoch": 0.7492136269126193, "grad_norm": 1.0444000959396362, "learning_rate": 7.1896000000000014e-06, "loss": 0.0231, "step": 14053 }, { "epoch": 0.7492669403422723, "grad_norm": 2.062749147415161, "learning_rate": 7.189400000000001e-06, "loss": 0.0458, "step": 14054 }, { "epoch": 0.7493202537719251, "grad_norm": 1.8000432252883911, "learning_rate": 7.1892000000000006e-06, "loss": 0.0469, "step": 14055 }, { "epoch": 0.7493735672015781, "grad_norm": 2.5671725273132324, "learning_rate": 7.189e-06, "loss": 0.0248, "step": 14056 }, { "epoch": 0.749426880631231, "grad_norm": 3.8749301433563232, "learning_rate": 7.1888000000000005e-06, "loss": 0.0437, "step": 14057 }, { "epoch": 0.749480194060884, "grad_norm": 2.2076096534729004, "learning_rate": 7.188600000000001e-06, "loss": 0.0244, "step": 14058 }, { "epoch": 0.7495335074905368, "grad_norm": 1.2646379470825195, "learning_rate": 7.188400000000001e-06, "loss": 0.028, "step": 14059 }, { "epoch": 0.7495868209201898, "grad_norm": 1.1880152225494385, "learning_rate": 7.188200000000001e-06, "loss": 0.0205, "step": 14060 }, { "epoch": 0.7496401343498428, "grad_norm": 1.6707944869995117, "learning_rate": 7.1880000000000005e-06, "loss": 0.026, "step": 14061 }, { "epoch": 0.7496934477794956, "grad_norm": 1.73844575881958, "learning_rate": 7.1878e-06, "loss": 0.0314, "step": 14062 }, { "epoch": 0.7497467612091486, "grad_norm": 2.7703194618225098, "learning_rate": 7.1876000000000004e-06, "loss": 0.0688, "step": 14063 }, { "epoch": 0.7498000746388015, "grad_norm": 1.9002288579940796, "learning_rate": 7.187400000000001e-06, "loss": 0.054, "step": 14064 }, { "epoch": 0.7498533880684545, "grad_norm": 1.1300380229949951, "learning_rate": 7.187200000000001e-06, "loss": 0.01, "step": 14065 }, { "epoch": 0.7499067014981073, "grad_norm": 2.5161385536193848, "learning_rate": 7.187000000000001e-06, "loss": 0.0258, "step": 14066 }, { "epoch": 0.7499600149277603, "grad_norm": 1.632783055305481, "learning_rate": 7.1868e-06, "loss": 0.0313, "step": 14067 }, { "epoch": 0.7500133283574132, "grad_norm": 1.9610543251037598, "learning_rate": 7.1866e-06, "loss": 0.0281, "step": 14068 }, { "epoch": 0.7500666417870662, "grad_norm": 1.7999155521392822, "learning_rate": 7.1864e-06, "loss": 0.0221, "step": 14069 }, { "epoch": 0.750119955216719, "grad_norm": 2.7556240558624268, "learning_rate": 7.186200000000001e-06, "loss": 0.056, "step": 14070 }, { "epoch": 0.750173268646372, "grad_norm": 1.51035475730896, "learning_rate": 7.186000000000001e-06, "loss": 0.013, "step": 14071 }, { "epoch": 0.750226582076025, "grad_norm": 2.1044609546661377, "learning_rate": 7.185800000000001e-06, "loss": 0.0533, "step": 14072 }, { "epoch": 0.7502798955056779, "grad_norm": 2.52769136428833, "learning_rate": 7.1856e-06, "loss": 0.0238, "step": 14073 }, { "epoch": 0.7503332089353308, "grad_norm": 3.986126184463501, "learning_rate": 7.1854e-06, "loss": 0.0292, "step": 14074 }, { "epoch": 0.7503865223649837, "grad_norm": 4.8758392333984375, "learning_rate": 7.185200000000001e-06, "loss": 0.0387, "step": 14075 }, { "epoch": 0.7504398357946367, "grad_norm": 1.7895532846450806, "learning_rate": 7.185000000000001e-06, "loss": 0.0318, "step": 14076 }, { "epoch": 0.7504931492242896, "grad_norm": 1.3310046195983887, "learning_rate": 7.184800000000001e-06, "loss": 0.0195, "step": 14077 }, { "epoch": 0.7505464626539425, "grad_norm": 1.29897940158844, "learning_rate": 7.184600000000001e-06, "loss": 0.0096, "step": 14078 }, { "epoch": 0.7505997760835954, "grad_norm": 3.526052713394165, "learning_rate": 7.1844e-06, "loss": 0.0342, "step": 14079 }, { "epoch": 0.7506530895132484, "grad_norm": 2.335416555404663, "learning_rate": 7.1842e-06, "loss": 0.0295, "step": 14080 }, { "epoch": 0.7507064029429014, "grad_norm": 1.3856589794158936, "learning_rate": 7.184000000000001e-06, "loss": 0.0267, "step": 14081 }, { "epoch": 0.7507597163725542, "grad_norm": 1.9730427265167236, "learning_rate": 7.1838000000000006e-06, "loss": 0.0329, "step": 14082 }, { "epoch": 0.7508130298022072, "grad_norm": 2.0863137245178223, "learning_rate": 7.183600000000001e-06, "loss": 0.0172, "step": 14083 }, { "epoch": 0.7508663432318601, "grad_norm": 2.8797309398651123, "learning_rate": 7.1834000000000005e-06, "loss": 0.0327, "step": 14084 }, { "epoch": 0.7509196566615131, "grad_norm": 1.056731104850769, "learning_rate": 7.1832e-06, "loss": 0.0068, "step": 14085 }, { "epoch": 0.7509729700911659, "grad_norm": 1.5631086826324463, "learning_rate": 7.183000000000001e-06, "loss": 0.0152, "step": 14086 }, { "epoch": 0.7510262835208189, "grad_norm": 1.847901701927185, "learning_rate": 7.182800000000001e-06, "loss": 0.0184, "step": 14087 }, { "epoch": 0.7510795969504718, "grad_norm": 2.7670838832855225, "learning_rate": 7.1826000000000005e-06, "loss": 0.023, "step": 14088 }, { "epoch": 0.7511329103801248, "grad_norm": 1.596113681793213, "learning_rate": 7.182400000000001e-06, "loss": 0.0159, "step": 14089 }, { "epoch": 0.7511862238097777, "grad_norm": 4.2632341384887695, "learning_rate": 7.1822000000000005e-06, "loss": 0.0552, "step": 14090 }, { "epoch": 0.7512395372394306, "grad_norm": 2.0670485496520996, "learning_rate": 7.182e-06, "loss": 0.0192, "step": 14091 }, { "epoch": 0.7512928506690836, "grad_norm": 4.411128520965576, "learning_rate": 7.181800000000001e-06, "loss": 0.0483, "step": 14092 }, { "epoch": 0.7513461640987364, "grad_norm": 1.907562017440796, "learning_rate": 7.181600000000001e-06, "loss": 0.0167, "step": 14093 }, { "epoch": 0.7513994775283894, "grad_norm": 1.8831301927566528, "learning_rate": 7.1814e-06, "loss": 0.043, "step": 14094 }, { "epoch": 0.7514527909580423, "grad_norm": 1.6715683937072754, "learning_rate": 7.181200000000001e-06, "loss": 0.0124, "step": 14095 }, { "epoch": 0.7515061043876953, "grad_norm": 4.5152974128723145, "learning_rate": 7.181e-06, "loss": 0.049, "step": 14096 }, { "epoch": 0.7515594178173481, "grad_norm": 2.973980665206909, "learning_rate": 7.180800000000001e-06, "loss": 0.0682, "step": 14097 }, { "epoch": 0.7516127312470011, "grad_norm": 1.9046119451522827, "learning_rate": 7.180600000000001e-06, "loss": 0.0151, "step": 14098 }, { "epoch": 0.7516660446766541, "grad_norm": 2.1805808544158936, "learning_rate": 7.180400000000001e-06, "loss": 0.0291, "step": 14099 }, { "epoch": 0.751719358106307, "grad_norm": 3.5689775943756104, "learning_rate": 7.1802e-06, "loss": 0.0305, "step": 14100 }, { "epoch": 0.75177267153596, "grad_norm": 2.396470308303833, "learning_rate": 7.180000000000001e-06, "loss": 0.0308, "step": 14101 }, { "epoch": 0.7518259849656128, "grad_norm": 1.297890067100525, "learning_rate": 7.1798e-06, "loss": 0.0148, "step": 14102 }, { "epoch": 0.7518792983952658, "grad_norm": 1.5228701829910278, "learning_rate": 7.179600000000001e-06, "loss": 0.0164, "step": 14103 }, { "epoch": 0.7519326118249187, "grad_norm": 1.21344792842865, "learning_rate": 7.179400000000001e-06, "loss": 0.0113, "step": 14104 }, { "epoch": 0.7519859252545716, "grad_norm": 2.7172772884368896, "learning_rate": 7.179200000000001e-06, "loss": 0.037, "step": 14105 }, { "epoch": 0.7520392386842245, "grad_norm": 2.7313966751098633, "learning_rate": 7.179e-06, "loss": 0.0644, "step": 14106 }, { "epoch": 0.7520925521138775, "grad_norm": 2.235954761505127, "learning_rate": 7.178800000000001e-06, "loss": 0.0191, "step": 14107 }, { "epoch": 0.7521458655435305, "grad_norm": 4.103490352630615, "learning_rate": 7.178600000000001e-06, "loss": 0.0317, "step": 14108 }, { "epoch": 0.7521991789731833, "grad_norm": 2.9450294971466064, "learning_rate": 7.178400000000001e-06, "loss": 0.0669, "step": 14109 }, { "epoch": 0.7522524924028363, "grad_norm": 1.5715842247009277, "learning_rate": 7.178200000000001e-06, "loss": 0.0242, "step": 14110 }, { "epoch": 0.7523058058324892, "grad_norm": 1.5557328462600708, "learning_rate": 7.1780000000000006e-06, "loss": 0.0194, "step": 14111 }, { "epoch": 0.7523591192621422, "grad_norm": 0.9102042317390442, "learning_rate": 7.1778e-06, "loss": 0.0217, "step": 14112 }, { "epoch": 0.752412432691795, "grad_norm": 1.7447021007537842, "learning_rate": 7.1776000000000005e-06, "loss": 0.024, "step": 14113 }, { "epoch": 0.752465746121448, "grad_norm": 1.7546813488006592, "learning_rate": 7.177400000000001e-06, "loss": 0.0223, "step": 14114 }, { "epoch": 0.7525190595511009, "grad_norm": 2.0279219150543213, "learning_rate": 7.1772000000000005e-06, "loss": 0.0236, "step": 14115 }, { "epoch": 0.7525723729807539, "grad_norm": 1.6385964155197144, "learning_rate": 7.177000000000001e-06, "loss": 0.0096, "step": 14116 }, { "epoch": 0.7526256864104068, "grad_norm": 4.12222146987915, "learning_rate": 7.1768000000000005e-06, "loss": 0.0482, "step": 14117 }, { "epoch": 0.7526789998400597, "grad_norm": 1.4276413917541504, "learning_rate": 7.1766e-06, "loss": 0.0211, "step": 14118 }, { "epoch": 0.7527323132697127, "grad_norm": 2.4168925285339355, "learning_rate": 7.1764000000000004e-06, "loss": 0.0805, "step": 14119 }, { "epoch": 0.7527856266993656, "grad_norm": 5.291404724121094, "learning_rate": 7.176200000000001e-06, "loss": 0.0325, "step": 14120 }, { "epoch": 0.7528389401290185, "grad_norm": 2.283418893814087, "learning_rate": 7.176e-06, "loss": 0.0337, "step": 14121 }, { "epoch": 0.7528922535586714, "grad_norm": 1.9872691631317139, "learning_rate": 7.175800000000001e-06, "loss": 0.0292, "step": 14122 }, { "epoch": 0.7529455669883244, "grad_norm": 2.3805043697357178, "learning_rate": 7.1756e-06, "loss": 0.0178, "step": 14123 }, { "epoch": 0.7529988804179772, "grad_norm": 1.0508100986480713, "learning_rate": 7.1754e-06, "loss": 0.02, "step": 14124 }, { "epoch": 0.7530521938476302, "grad_norm": 1.570131778717041, "learning_rate": 7.175200000000001e-06, "loss": 0.0344, "step": 14125 }, { "epoch": 0.7531055072772831, "grad_norm": 1.5698440074920654, "learning_rate": 7.175000000000001e-06, "loss": 0.0256, "step": 14126 }, { "epoch": 0.7531588207069361, "grad_norm": 2.6939313411712646, "learning_rate": 7.1748e-06, "loss": 0.0546, "step": 14127 }, { "epoch": 0.753212134136589, "grad_norm": 1.3377983570098877, "learning_rate": 7.174600000000001e-06, "loss": 0.0126, "step": 14128 }, { "epoch": 0.7532654475662419, "grad_norm": 1.0367902517318726, "learning_rate": 7.1744e-06, "loss": 0.0114, "step": 14129 }, { "epoch": 0.7533187609958949, "grad_norm": 1.630962610244751, "learning_rate": 7.1742e-06, "loss": 0.0152, "step": 14130 }, { "epoch": 0.7533720744255478, "grad_norm": 1.5691412687301636, "learning_rate": 7.174000000000001e-06, "loss": 0.0339, "step": 14131 }, { "epoch": 0.7534253878552007, "grad_norm": 2.77229905128479, "learning_rate": 7.173800000000001e-06, "loss": 0.0697, "step": 14132 }, { "epoch": 0.7534787012848536, "grad_norm": 3.471944808959961, "learning_rate": 7.1736e-06, "loss": 0.0562, "step": 14133 }, { "epoch": 0.7535320147145066, "grad_norm": 1.3566826581954956, "learning_rate": 7.173400000000001e-06, "loss": 0.0172, "step": 14134 }, { "epoch": 0.7535853281441595, "grad_norm": 1.7330166101455688, "learning_rate": 7.1732e-06, "loss": 0.0454, "step": 14135 }, { "epoch": 0.7536386415738124, "grad_norm": 0.895645022392273, "learning_rate": 7.1730000000000015e-06, "loss": 0.0189, "step": 14136 }, { "epoch": 0.7536919550034654, "grad_norm": 1.9207435846328735, "learning_rate": 7.172800000000001e-06, "loss": 0.032, "step": 14137 }, { "epoch": 0.7537452684331183, "grad_norm": 1.9324928522109985, "learning_rate": 7.172600000000001e-06, "loss": 0.0308, "step": 14138 }, { "epoch": 0.7537985818627713, "grad_norm": 2.8422460556030273, "learning_rate": 7.1724e-06, "loss": 0.0244, "step": 14139 }, { "epoch": 0.7538518952924241, "grad_norm": 3.1403183937072754, "learning_rate": 7.1722000000000006e-06, "loss": 0.0371, "step": 14140 }, { "epoch": 0.7539052087220771, "grad_norm": 2.8653526306152344, "learning_rate": 7.172e-06, "loss": 0.0202, "step": 14141 }, { "epoch": 0.75395852215173, "grad_norm": 2.840360403060913, "learning_rate": 7.171800000000001e-06, "loss": 0.0451, "step": 14142 }, { "epoch": 0.754011835581383, "grad_norm": 1.6520495414733887, "learning_rate": 7.171600000000001e-06, "loss": 0.0227, "step": 14143 }, { "epoch": 0.7540651490110358, "grad_norm": 1.9598655700683594, "learning_rate": 7.1714000000000005e-06, "loss": 0.0377, "step": 14144 }, { "epoch": 0.7541184624406888, "grad_norm": 3.055922269821167, "learning_rate": 7.1712e-06, "loss": 0.0491, "step": 14145 }, { "epoch": 0.7541717758703418, "grad_norm": 1.1158747673034668, "learning_rate": 7.1710000000000005e-06, "loss": 0.017, "step": 14146 }, { "epoch": 0.7542250892999947, "grad_norm": 2.043107509613037, "learning_rate": 7.170800000000001e-06, "loss": 0.032, "step": 14147 }, { "epoch": 0.7542784027296476, "grad_norm": 2.682753801345825, "learning_rate": 7.170600000000001e-06, "loss": 0.038, "step": 14148 }, { "epoch": 0.7543317161593005, "grad_norm": 3.1265947818756104, "learning_rate": 7.170400000000001e-06, "loss": 0.0533, "step": 14149 }, { "epoch": 0.7543850295889535, "grad_norm": 1.2045693397521973, "learning_rate": 7.1702e-06, "loss": 0.0151, "step": 14150 }, { "epoch": 0.7544383430186063, "grad_norm": 2.0228500366210938, "learning_rate": 7.17e-06, "loss": 0.0233, "step": 14151 }, { "epoch": 0.7544916564482593, "grad_norm": 2.8526811599731445, "learning_rate": 7.1698e-06, "loss": 0.0349, "step": 14152 }, { "epoch": 0.7545449698779122, "grad_norm": 1.9731305837631226, "learning_rate": 7.169600000000001e-06, "loss": 0.0187, "step": 14153 }, { "epoch": 0.7545982833075652, "grad_norm": 1.4477967023849487, "learning_rate": 7.169400000000001e-06, "loss": 0.0129, "step": 14154 }, { "epoch": 0.7546515967372182, "grad_norm": 3.0471866130828857, "learning_rate": 7.169200000000001e-06, "loss": 0.0354, "step": 14155 }, { "epoch": 0.754704910166871, "grad_norm": 1.2728887796401978, "learning_rate": 7.169e-06, "loss": 0.0144, "step": 14156 }, { "epoch": 0.754758223596524, "grad_norm": 1.637034296989441, "learning_rate": 7.1688e-06, "loss": 0.0154, "step": 14157 }, { "epoch": 0.7548115370261769, "grad_norm": 1.270167589187622, "learning_rate": 7.168600000000001e-06, "loss": 0.019, "step": 14158 }, { "epoch": 0.7548648504558298, "grad_norm": 2.1027259826660156, "learning_rate": 7.168400000000001e-06, "loss": 0.0271, "step": 14159 }, { "epoch": 0.7549181638854827, "grad_norm": 2.463972806930542, "learning_rate": 7.168200000000001e-06, "loss": 0.0359, "step": 14160 }, { "epoch": 0.7549714773151357, "grad_norm": 3.015092611312866, "learning_rate": 7.168000000000001e-06, "loss": 0.0282, "step": 14161 }, { "epoch": 0.7550247907447886, "grad_norm": 2.461179256439209, "learning_rate": 7.1678e-06, "loss": 0.0369, "step": 14162 }, { "epoch": 0.7550781041744415, "grad_norm": 1.2711517810821533, "learning_rate": 7.1676e-06, "loss": 0.0179, "step": 14163 }, { "epoch": 0.7551314176040945, "grad_norm": 3.2854080200195312, "learning_rate": 7.167400000000001e-06, "loss": 0.0538, "step": 14164 }, { "epoch": 0.7551847310337474, "grad_norm": 2.214191436767578, "learning_rate": 7.167200000000001e-06, "loss": 0.0507, "step": 14165 }, { "epoch": 0.7552380444634004, "grad_norm": 1.9101481437683105, "learning_rate": 7.167000000000001e-06, "loss": 0.0275, "step": 14166 }, { "epoch": 0.7552913578930532, "grad_norm": 2.6586859226226807, "learning_rate": 7.1668000000000006e-06, "loss": 0.0604, "step": 14167 }, { "epoch": 0.7553446713227062, "grad_norm": 4.240939617156982, "learning_rate": 7.1666e-06, "loss": 0.0347, "step": 14168 }, { "epoch": 0.7553979847523591, "grad_norm": 2.2706940174102783, "learning_rate": 7.166400000000001e-06, "loss": 0.0339, "step": 14169 }, { "epoch": 0.7554512981820121, "grad_norm": 1.5128929615020752, "learning_rate": 7.166200000000001e-06, "loss": 0.0225, "step": 14170 }, { "epoch": 0.7555046116116649, "grad_norm": 2.2248871326446533, "learning_rate": 7.1660000000000005e-06, "loss": 0.0345, "step": 14171 }, { "epoch": 0.7555579250413179, "grad_norm": 3.012272596359253, "learning_rate": 7.165800000000001e-06, "loss": 0.0434, "step": 14172 }, { "epoch": 0.7556112384709709, "grad_norm": 1.9581992626190186, "learning_rate": 7.1656000000000005e-06, "loss": 0.0202, "step": 14173 }, { "epoch": 0.7556645519006238, "grad_norm": 1.5974137783050537, "learning_rate": 7.1654e-06, "loss": 0.014, "step": 14174 }, { "epoch": 0.7557178653302767, "grad_norm": 1.7819029092788696, "learning_rate": 7.165200000000001e-06, "loss": 0.0229, "step": 14175 }, { "epoch": 0.7557711787599296, "grad_norm": 1.7317136526107788, "learning_rate": 7.165000000000001e-06, "loss": 0.035, "step": 14176 }, { "epoch": 0.7558244921895826, "grad_norm": 3.918424129486084, "learning_rate": 7.1648e-06, "loss": 0.0408, "step": 14177 }, { "epoch": 0.7558778056192355, "grad_norm": 1.7279471158981323, "learning_rate": 7.164600000000001e-06, "loss": 0.0253, "step": 14178 }, { "epoch": 0.7559311190488884, "grad_norm": 1.4121665954589844, "learning_rate": 7.1644e-06, "loss": 0.0284, "step": 14179 }, { "epoch": 0.7559844324785413, "grad_norm": 2.3205044269561768, "learning_rate": 7.1642e-06, "loss": 0.0353, "step": 14180 }, { "epoch": 0.7560377459081943, "grad_norm": 1.71712064743042, "learning_rate": 7.164000000000001e-06, "loss": 0.0166, "step": 14181 }, { "epoch": 0.7560910593378471, "grad_norm": 2.67117977142334, "learning_rate": 7.163800000000001e-06, "loss": 0.036, "step": 14182 }, { "epoch": 0.7561443727675001, "grad_norm": 0.8407235145568848, "learning_rate": 7.1636e-06, "loss": 0.0048, "step": 14183 }, { "epoch": 0.7561976861971531, "grad_norm": 2.752354145050049, "learning_rate": 7.163400000000001e-06, "loss": 0.063, "step": 14184 }, { "epoch": 0.756250999626806, "grad_norm": 2.933227300643921, "learning_rate": 7.1632e-06, "loss": 0.0476, "step": 14185 }, { "epoch": 0.756304313056459, "grad_norm": 2.2156436443328857, "learning_rate": 7.163000000000001e-06, "loss": 0.0166, "step": 14186 }, { "epoch": 0.7563576264861118, "grad_norm": 1.0569398403167725, "learning_rate": 7.162800000000001e-06, "loss": 0.0248, "step": 14187 }, { "epoch": 0.7564109399157648, "grad_norm": 1.5696057081222534, "learning_rate": 7.162600000000001e-06, "loss": 0.0169, "step": 14188 }, { "epoch": 0.7564642533454177, "grad_norm": 1.8338398933410645, "learning_rate": 7.1624e-06, "loss": 0.0253, "step": 14189 }, { "epoch": 0.7565175667750706, "grad_norm": 2.2831180095672607, "learning_rate": 7.162200000000001e-06, "loss": 0.0376, "step": 14190 }, { "epoch": 0.7565708802047235, "grad_norm": 2.5996949672698975, "learning_rate": 7.162e-06, "loss": 0.0162, "step": 14191 }, { "epoch": 0.7566241936343765, "grad_norm": 3.3167948722839355, "learning_rate": 7.161800000000001e-06, "loss": 0.0231, "step": 14192 }, { "epoch": 0.7566775070640295, "grad_norm": 3.2683117389678955, "learning_rate": 7.161600000000001e-06, "loss": 0.0238, "step": 14193 }, { "epoch": 0.7567308204936823, "grad_norm": 4.134608745574951, "learning_rate": 7.161400000000001e-06, "loss": 0.0224, "step": 14194 }, { "epoch": 0.7567841339233353, "grad_norm": 2.2707715034484863, "learning_rate": 7.1612e-06, "loss": 0.0188, "step": 14195 }, { "epoch": 0.7568374473529882, "grad_norm": 4.129819869995117, "learning_rate": 7.1610000000000006e-06, "loss": 0.0567, "step": 14196 }, { "epoch": 0.7568907607826412, "grad_norm": 1.7810393571853638, "learning_rate": 7.160800000000001e-06, "loss": 0.0261, "step": 14197 }, { "epoch": 0.756944074212294, "grad_norm": 2.4106197357177734, "learning_rate": 7.1606000000000005e-06, "loss": 0.0338, "step": 14198 }, { "epoch": 0.756997387641947, "grad_norm": 1.5428624153137207, "learning_rate": 7.160400000000001e-06, "loss": 0.013, "step": 14199 }, { "epoch": 0.7570507010715999, "grad_norm": 1.227486491203308, "learning_rate": 7.1602000000000005e-06, "loss": 0.0139, "step": 14200 }, { "epoch": 0.7571040145012529, "grad_norm": 2.4063148498535156, "learning_rate": 7.16e-06, "loss": 0.0235, "step": 14201 }, { "epoch": 0.7571573279309058, "grad_norm": 2.412513256072998, "learning_rate": 7.1598000000000005e-06, "loss": 0.0608, "step": 14202 }, { "epoch": 0.7572106413605587, "grad_norm": 2.725752830505371, "learning_rate": 7.159600000000001e-06, "loss": 0.0625, "step": 14203 }, { "epoch": 0.7572639547902117, "grad_norm": 2.577589988708496, "learning_rate": 7.1594000000000004e-06, "loss": 0.0351, "step": 14204 }, { "epoch": 0.7573172682198646, "grad_norm": 1.6569913625717163, "learning_rate": 7.159200000000001e-06, "loss": 0.0263, "step": 14205 }, { "epoch": 0.7573705816495175, "grad_norm": 0.8665189743041992, "learning_rate": 7.159e-06, "loss": 0.0105, "step": 14206 }, { "epoch": 0.7574238950791704, "grad_norm": 2.8917155265808105, "learning_rate": 7.1588e-06, "loss": 0.0309, "step": 14207 }, { "epoch": 0.7574772085088234, "grad_norm": 1.9113787412643433, "learning_rate": 7.158600000000001e-06, "loss": 0.0475, "step": 14208 }, { "epoch": 0.7575305219384763, "grad_norm": 3.052806854248047, "learning_rate": 7.158400000000001e-06, "loss": 0.0472, "step": 14209 }, { "epoch": 0.7575838353681292, "grad_norm": 5.075059413909912, "learning_rate": 7.1582e-06, "loss": 0.0267, "step": 14210 }, { "epoch": 0.7576371487977822, "grad_norm": 1.2214804887771606, "learning_rate": 7.158000000000001e-06, "loss": 0.0124, "step": 14211 }, { "epoch": 0.7576904622274351, "grad_norm": 2.0559747219085693, "learning_rate": 7.1578e-06, "loss": 0.0293, "step": 14212 }, { "epoch": 0.7577437756570881, "grad_norm": 3.713423013687134, "learning_rate": 7.1576e-06, "loss": 0.0261, "step": 14213 }, { "epoch": 0.7577970890867409, "grad_norm": 3.047175168991089, "learning_rate": 7.157400000000001e-06, "loss": 0.0582, "step": 14214 }, { "epoch": 0.7578504025163939, "grad_norm": 1.5315228700637817, "learning_rate": 7.157200000000001e-06, "loss": 0.0266, "step": 14215 }, { "epoch": 0.7579037159460468, "grad_norm": 1.9700801372528076, "learning_rate": 7.157e-06, "loss": 0.0365, "step": 14216 }, { "epoch": 0.7579570293756998, "grad_norm": 1.7873975038528442, "learning_rate": 7.156800000000001e-06, "loss": 0.0261, "step": 14217 }, { "epoch": 0.7580103428053526, "grad_norm": 2.5456430912017822, "learning_rate": 7.1566e-06, "loss": 0.0363, "step": 14218 }, { "epoch": 0.7580636562350056, "grad_norm": 1.6776617765426636, "learning_rate": 7.1564000000000015e-06, "loss": 0.0372, "step": 14219 }, { "epoch": 0.7581169696646586, "grad_norm": 1.5220866203308105, "learning_rate": 7.156200000000001e-06, "loss": 0.0162, "step": 14220 }, { "epoch": 0.7581702830943114, "grad_norm": 1.6427956819534302, "learning_rate": 7.156000000000001e-06, "loss": 0.0216, "step": 14221 }, { "epoch": 0.7582235965239644, "grad_norm": 2.8336918354034424, "learning_rate": 7.1558e-06, "loss": 0.0511, "step": 14222 }, { "epoch": 0.7582769099536173, "grad_norm": 1.7657290697097778, "learning_rate": 7.1556000000000006e-06, "loss": 0.0373, "step": 14223 }, { "epoch": 0.7583302233832703, "grad_norm": 1.8545353412628174, "learning_rate": 7.1554e-06, "loss": 0.0275, "step": 14224 }, { "epoch": 0.7583835368129231, "grad_norm": 2.2042617797851562, "learning_rate": 7.155200000000001e-06, "loss": 0.0377, "step": 14225 }, { "epoch": 0.7584368502425761, "grad_norm": 1.21506667137146, "learning_rate": 7.155000000000001e-06, "loss": 0.018, "step": 14226 }, { "epoch": 0.758490163672229, "grad_norm": 3.0987448692321777, "learning_rate": 7.1548000000000005e-06, "loss": 0.0451, "step": 14227 }, { "epoch": 0.758543477101882, "grad_norm": 1.7364431619644165, "learning_rate": 7.1546e-06, "loss": 0.0215, "step": 14228 }, { "epoch": 0.758596790531535, "grad_norm": 3.041348457336426, "learning_rate": 7.1544000000000005e-06, "loss": 0.0427, "step": 14229 }, { "epoch": 0.7586501039611878, "grad_norm": 2.4865009784698486, "learning_rate": 7.1542e-06, "loss": 0.0202, "step": 14230 }, { "epoch": 0.7587034173908408, "grad_norm": 2.627026319503784, "learning_rate": 7.154000000000001e-06, "loss": 0.0419, "step": 14231 }, { "epoch": 0.7587567308204937, "grad_norm": 2.7688448429107666, "learning_rate": 7.153800000000001e-06, "loss": 0.0238, "step": 14232 }, { "epoch": 0.7588100442501466, "grad_norm": 2.2548134326934814, "learning_rate": 7.1536e-06, "loss": 0.0178, "step": 14233 }, { "epoch": 0.7588633576797995, "grad_norm": 2.7022204399108887, "learning_rate": 7.1534e-06, "loss": 0.053, "step": 14234 }, { "epoch": 0.7589166711094525, "grad_norm": 1.3080432415008545, "learning_rate": 7.1532e-06, "loss": 0.018, "step": 14235 }, { "epoch": 0.7589699845391054, "grad_norm": 1.6282799243927002, "learning_rate": 7.153000000000001e-06, "loss": 0.033, "step": 14236 }, { "epoch": 0.7590232979687583, "grad_norm": 1.456465244293213, "learning_rate": 7.152800000000001e-06, "loss": 0.015, "step": 14237 }, { "epoch": 0.7590766113984112, "grad_norm": 1.5945602655410767, "learning_rate": 7.152600000000001e-06, "loss": 0.0285, "step": 14238 }, { "epoch": 0.7591299248280642, "grad_norm": 1.0264339447021484, "learning_rate": 7.1524e-06, "loss": 0.0244, "step": 14239 }, { "epoch": 0.7591832382577172, "grad_norm": 1.3478806018829346, "learning_rate": 7.1522e-06, "loss": 0.0174, "step": 14240 }, { "epoch": 0.75923655168737, "grad_norm": 1.5125519037246704, "learning_rate": 7.152e-06, "loss": 0.0207, "step": 14241 }, { "epoch": 0.759289865117023, "grad_norm": 3.116638660430908, "learning_rate": 7.151800000000001e-06, "loss": 0.0345, "step": 14242 }, { "epoch": 0.7593431785466759, "grad_norm": 2.071532726287842, "learning_rate": 7.151600000000001e-06, "loss": 0.0197, "step": 14243 }, { "epoch": 0.7593964919763289, "grad_norm": 1.689172625541687, "learning_rate": 7.151400000000001e-06, "loss": 0.0221, "step": 14244 }, { "epoch": 0.7594498054059817, "grad_norm": 3.998382329940796, "learning_rate": 7.1512e-06, "loss": 0.0347, "step": 14245 }, { "epoch": 0.7595031188356347, "grad_norm": 3.203073740005493, "learning_rate": 7.151e-06, "loss": 0.0285, "step": 14246 }, { "epoch": 0.7595564322652876, "grad_norm": 1.9453192949295044, "learning_rate": 7.150800000000001e-06, "loss": 0.0133, "step": 14247 }, { "epoch": 0.7596097456949406, "grad_norm": 3.5299184322357178, "learning_rate": 7.150600000000001e-06, "loss": 0.0194, "step": 14248 }, { "epoch": 0.7596630591245935, "grad_norm": 2.2694904804229736, "learning_rate": 7.150400000000001e-06, "loss": 0.0178, "step": 14249 }, { "epoch": 0.7597163725542464, "grad_norm": 1.6721004247665405, "learning_rate": 7.150200000000001e-06, "loss": 0.0106, "step": 14250 }, { "epoch": 0.7597696859838994, "grad_norm": 1.5404325723648071, "learning_rate": 7.15e-06, "loss": 0.0176, "step": 14251 }, { "epoch": 0.7598229994135522, "grad_norm": 1.3978458642959595, "learning_rate": 7.1498e-06, "loss": 0.0346, "step": 14252 }, { "epoch": 0.7598763128432052, "grad_norm": 2.6144461631774902, "learning_rate": 7.149600000000001e-06, "loss": 0.0426, "step": 14253 }, { "epoch": 0.7599296262728581, "grad_norm": 1.6098670959472656, "learning_rate": 7.1494000000000005e-06, "loss": 0.0103, "step": 14254 }, { "epoch": 0.7599829397025111, "grad_norm": 1.675958514213562, "learning_rate": 7.149200000000001e-06, "loss": 0.0156, "step": 14255 }, { "epoch": 0.7600362531321639, "grad_norm": 3.1797702312469482, "learning_rate": 7.1490000000000005e-06, "loss": 0.0382, "step": 14256 }, { "epoch": 0.7600895665618169, "grad_norm": 2.1272242069244385, "learning_rate": 7.1488e-06, "loss": 0.0139, "step": 14257 }, { "epoch": 0.7601428799914699, "grad_norm": 3.7927961349487305, "learning_rate": 7.148600000000001e-06, "loss": 0.0237, "step": 14258 }, { "epoch": 0.7601961934211228, "grad_norm": 1.8431487083435059, "learning_rate": 7.148400000000001e-06, "loss": 0.0246, "step": 14259 }, { "epoch": 0.7602495068507757, "grad_norm": 2.8661651611328125, "learning_rate": 7.1482000000000004e-06, "loss": 0.0358, "step": 14260 }, { "epoch": 0.7603028202804286, "grad_norm": 3.0654356479644775, "learning_rate": 7.148000000000001e-06, "loss": 0.0391, "step": 14261 }, { "epoch": 0.7603561337100816, "grad_norm": 4.504868984222412, "learning_rate": 7.1478e-06, "loss": 0.0601, "step": 14262 }, { "epoch": 0.7604094471397345, "grad_norm": 0.6131031513214111, "learning_rate": 7.1476e-06, "loss": 0.0068, "step": 14263 }, { "epoch": 0.7604627605693874, "grad_norm": 3.642932653427124, "learning_rate": 7.147400000000001e-06, "loss": 0.0442, "step": 14264 }, { "epoch": 0.7605160739990403, "grad_norm": 6.370126724243164, "learning_rate": 7.147200000000001e-06, "loss": 0.0182, "step": 14265 }, { "epoch": 0.7605693874286933, "grad_norm": 1.5926350355148315, "learning_rate": 7.147e-06, "loss": 0.0174, "step": 14266 }, { "epoch": 0.7606227008583463, "grad_norm": 2.9460718631744385, "learning_rate": 7.146800000000001e-06, "loss": 0.0593, "step": 14267 }, { "epoch": 0.7606760142879991, "grad_norm": 1.4190757274627686, "learning_rate": 7.1466e-06, "loss": 0.0199, "step": 14268 }, { "epoch": 0.7607293277176521, "grad_norm": 3.7052035331726074, "learning_rate": 7.146400000000001e-06, "loss": 0.0429, "step": 14269 }, { "epoch": 0.760782641147305, "grad_norm": 5.534246921539307, "learning_rate": 7.146200000000001e-06, "loss": 0.1031, "step": 14270 }, { "epoch": 0.760835954576958, "grad_norm": 1.0907613039016724, "learning_rate": 7.146000000000001e-06, "loss": 0.0219, "step": 14271 }, { "epoch": 0.7608892680066108, "grad_norm": 2.109823226928711, "learning_rate": 7.1458e-06, "loss": 0.0197, "step": 14272 }, { "epoch": 0.7609425814362638, "grad_norm": 1.8836700916290283, "learning_rate": 7.145600000000001e-06, "loss": 0.0109, "step": 14273 }, { "epoch": 0.7609958948659167, "grad_norm": 2.164452314376831, "learning_rate": 7.1454e-06, "loss": 0.0237, "step": 14274 }, { "epoch": 0.7610492082955697, "grad_norm": 2.9398293495178223, "learning_rate": 7.145200000000001e-06, "loss": 0.0357, "step": 14275 }, { "epoch": 0.7611025217252226, "grad_norm": 0.7246730923652649, "learning_rate": 7.145000000000001e-06, "loss": 0.0089, "step": 14276 }, { "epoch": 0.7611558351548755, "grad_norm": 2.4821088314056396, "learning_rate": 7.144800000000001e-06, "loss": 0.047, "step": 14277 }, { "epoch": 0.7612091485845285, "grad_norm": 1.6332151889801025, "learning_rate": 7.1446e-06, "loss": 0.0285, "step": 14278 }, { "epoch": 0.7612624620141814, "grad_norm": 2.35986065864563, "learning_rate": 7.144400000000001e-06, "loss": 0.0497, "step": 14279 }, { "epoch": 0.7613157754438343, "grad_norm": 1.2400450706481934, "learning_rate": 7.144200000000001e-06, "loss": 0.0146, "step": 14280 }, { "epoch": 0.7613690888734872, "grad_norm": 1.9394665956497192, "learning_rate": 7.1440000000000005e-06, "loss": 0.0174, "step": 14281 }, { "epoch": 0.7614224023031402, "grad_norm": 1.5154845714569092, "learning_rate": 7.143800000000001e-06, "loss": 0.0299, "step": 14282 }, { "epoch": 0.761475715732793, "grad_norm": 2.230402708053589, "learning_rate": 7.1436000000000005e-06, "loss": 0.0441, "step": 14283 }, { "epoch": 0.761529029162446, "grad_norm": 2.8435089588165283, "learning_rate": 7.1434e-06, "loss": 0.0215, "step": 14284 }, { "epoch": 0.761582342592099, "grad_norm": 2.2558844089508057, "learning_rate": 7.1432000000000005e-06, "loss": 0.033, "step": 14285 }, { "epoch": 0.7616356560217519, "grad_norm": 1.3836541175842285, "learning_rate": 7.143000000000001e-06, "loss": 0.0107, "step": 14286 }, { "epoch": 0.7616889694514049, "grad_norm": 1.8683273792266846, "learning_rate": 7.1428000000000005e-06, "loss": 0.0211, "step": 14287 }, { "epoch": 0.7617422828810577, "grad_norm": 1.0932198762893677, "learning_rate": 7.142600000000001e-06, "loss": 0.0231, "step": 14288 }, { "epoch": 0.7617955963107107, "grad_norm": 2.783933639526367, "learning_rate": 7.1424000000000004e-06, "loss": 0.0622, "step": 14289 }, { "epoch": 0.7618489097403636, "grad_norm": 0.9908202886581421, "learning_rate": 7.1422e-06, "loss": 0.0093, "step": 14290 }, { "epoch": 0.7619022231700165, "grad_norm": 0.9173946380615234, "learning_rate": 7.142e-06, "loss": 0.0108, "step": 14291 }, { "epoch": 0.7619555365996694, "grad_norm": 2.4955499172210693, "learning_rate": 7.141800000000001e-06, "loss": 0.0721, "step": 14292 }, { "epoch": 0.7620088500293224, "grad_norm": 3.529104709625244, "learning_rate": 7.1416e-06, "loss": 0.0618, "step": 14293 }, { "epoch": 0.7620621634589753, "grad_norm": 3.5655386447906494, "learning_rate": 7.141400000000001e-06, "loss": 0.0298, "step": 14294 }, { "epoch": 0.7621154768886282, "grad_norm": 2.65718150138855, "learning_rate": 7.1412e-06, "loss": 0.0554, "step": 14295 }, { "epoch": 0.7621687903182812, "grad_norm": 3.7363767623901367, "learning_rate": 7.141e-06, "loss": 0.067, "step": 14296 }, { "epoch": 0.7622221037479341, "grad_norm": 2.0219566822052, "learning_rate": 7.140800000000001e-06, "loss": 0.0205, "step": 14297 }, { "epoch": 0.7622754171775871, "grad_norm": 3.294607400894165, "learning_rate": 7.140600000000001e-06, "loss": 0.0451, "step": 14298 }, { "epoch": 0.7623287306072399, "grad_norm": 1.333315134048462, "learning_rate": 7.1404e-06, "loss": 0.033, "step": 14299 }, { "epoch": 0.7623820440368929, "grad_norm": 2.274178981781006, "learning_rate": 7.140200000000001e-06, "loss": 0.051, "step": 14300 }, { "epoch": 0.7624353574665458, "grad_norm": 1.4418058395385742, "learning_rate": 7.14e-06, "loss": 0.0158, "step": 14301 }, { "epoch": 0.7624886708961988, "grad_norm": 1.8973499536514282, "learning_rate": 7.1398e-06, "loss": 0.0271, "step": 14302 }, { "epoch": 0.7625419843258516, "grad_norm": 2.218096971511841, "learning_rate": 7.139600000000001e-06, "loss": 0.022, "step": 14303 }, { "epoch": 0.7625952977555046, "grad_norm": 2.8570666313171387, "learning_rate": 7.139400000000001e-06, "loss": 0.0665, "step": 14304 }, { "epoch": 0.7626486111851576, "grad_norm": 2.0428550243377686, "learning_rate": 7.1392e-06, "loss": 0.0426, "step": 14305 }, { "epoch": 0.7627019246148105, "grad_norm": 2.313668966293335, "learning_rate": 7.139000000000001e-06, "loss": 0.0487, "step": 14306 }, { "epoch": 0.7627552380444634, "grad_norm": 1.9902812242507935, "learning_rate": 7.1388e-06, "loss": 0.0299, "step": 14307 }, { "epoch": 0.7628085514741163, "grad_norm": 1.6862674951553345, "learning_rate": 7.138600000000001e-06, "loss": 0.0407, "step": 14308 }, { "epoch": 0.7628618649037693, "grad_norm": 2.4148993492126465, "learning_rate": 7.138400000000001e-06, "loss": 0.04, "step": 14309 }, { "epoch": 0.7629151783334222, "grad_norm": 2.685471773147583, "learning_rate": 7.1382000000000005e-06, "loss": 0.0459, "step": 14310 }, { "epoch": 0.7629684917630751, "grad_norm": 1.5902669429779053, "learning_rate": 7.138e-06, "loss": 0.0427, "step": 14311 }, { "epoch": 0.763021805192728, "grad_norm": 0.9515910148620605, "learning_rate": 7.1378000000000005e-06, "loss": 0.0117, "step": 14312 }, { "epoch": 0.763075118622381, "grad_norm": 1.8331024646759033, "learning_rate": 7.1376e-06, "loss": 0.0181, "step": 14313 }, { "epoch": 0.763128432052034, "grad_norm": 0.6327447295188904, "learning_rate": 7.137400000000001e-06, "loss": 0.0075, "step": 14314 }, { "epoch": 0.7631817454816868, "grad_norm": 1.825134515762329, "learning_rate": 7.137200000000001e-06, "loss": 0.0196, "step": 14315 }, { "epoch": 0.7632350589113398, "grad_norm": 1.6583714485168457, "learning_rate": 7.1370000000000004e-06, "loss": 0.0241, "step": 14316 }, { "epoch": 0.7632883723409927, "grad_norm": 1.7794129848480225, "learning_rate": 7.1368e-06, "loss": 0.0251, "step": 14317 }, { "epoch": 0.7633416857706457, "grad_norm": 2.6610171794891357, "learning_rate": 7.1366e-06, "loss": 0.0412, "step": 14318 }, { "epoch": 0.7633949992002985, "grad_norm": 2.1150639057159424, "learning_rate": 7.136400000000001e-06, "loss": 0.0345, "step": 14319 }, { "epoch": 0.7634483126299515, "grad_norm": 1.777259111404419, "learning_rate": 7.136200000000001e-06, "loss": 0.0282, "step": 14320 }, { "epoch": 0.7635016260596044, "grad_norm": 1.9276362657546997, "learning_rate": 7.136000000000001e-06, "loss": 0.039, "step": 14321 }, { "epoch": 0.7635549394892573, "grad_norm": 1.2069810628890991, "learning_rate": 7.1358e-06, "loss": 0.0107, "step": 14322 }, { "epoch": 0.7636082529189103, "grad_norm": 3.4770941734313965, "learning_rate": 7.1356e-06, "loss": 0.0131, "step": 14323 }, { "epoch": 0.7636615663485632, "grad_norm": 2.3063979148864746, "learning_rate": 7.1354e-06, "loss": 0.051, "step": 14324 }, { "epoch": 0.7637148797782162, "grad_norm": 2.3778574466705322, "learning_rate": 7.135200000000001e-06, "loss": 0.0319, "step": 14325 }, { "epoch": 0.763768193207869, "grad_norm": 1.9227955341339111, "learning_rate": 7.135000000000001e-06, "loss": 0.0218, "step": 14326 }, { "epoch": 0.763821506637522, "grad_norm": 1.1546235084533691, "learning_rate": 7.134800000000001e-06, "loss": 0.0186, "step": 14327 }, { "epoch": 0.7638748200671749, "grad_norm": 2.553189992904663, "learning_rate": 7.1346e-06, "loss": 0.0479, "step": 14328 }, { "epoch": 0.7639281334968279, "grad_norm": 1.3658450841903687, "learning_rate": 7.1344e-06, "loss": 0.0166, "step": 14329 }, { "epoch": 0.7639814469264807, "grad_norm": 2.515286445617676, "learning_rate": 7.134200000000001e-06, "loss": 0.0176, "step": 14330 }, { "epoch": 0.7640347603561337, "grad_norm": 1.7333835363388062, "learning_rate": 7.134000000000001e-06, "loss": 0.0164, "step": 14331 }, { "epoch": 0.7640880737857867, "grad_norm": 3.708369493484497, "learning_rate": 7.133800000000001e-06, "loss": 0.0264, "step": 14332 }, { "epoch": 0.7641413872154396, "grad_norm": 1.817118525505066, "learning_rate": 7.133600000000001e-06, "loss": 0.018, "step": 14333 }, { "epoch": 0.7641947006450925, "grad_norm": 2.810476541519165, "learning_rate": 7.1334e-06, "loss": 0.039, "step": 14334 }, { "epoch": 0.7642480140747454, "grad_norm": 6.307971954345703, "learning_rate": 7.1332e-06, "loss": 0.0467, "step": 14335 }, { "epoch": 0.7643013275043984, "grad_norm": 1.0846368074417114, "learning_rate": 7.133000000000001e-06, "loss": 0.0094, "step": 14336 }, { "epoch": 0.7643546409340513, "grad_norm": 1.9270602464675903, "learning_rate": 7.1328000000000006e-06, "loss": 0.025, "step": 14337 }, { "epoch": 0.7644079543637042, "grad_norm": 2.469818353652954, "learning_rate": 7.132600000000001e-06, "loss": 0.048, "step": 14338 }, { "epoch": 0.7644612677933571, "grad_norm": 3.6053760051727295, "learning_rate": 7.1324000000000005e-06, "loss": 0.0445, "step": 14339 }, { "epoch": 0.7645145812230101, "grad_norm": 3.6084513664245605, "learning_rate": 7.1322e-06, "loss": 0.0237, "step": 14340 }, { "epoch": 0.7645678946526631, "grad_norm": 1.309333324432373, "learning_rate": 7.132e-06, "loss": 0.012, "step": 14341 }, { "epoch": 0.7646212080823159, "grad_norm": 3.121701240539551, "learning_rate": 7.131800000000001e-06, "loss": 0.0112, "step": 14342 }, { "epoch": 0.7646745215119689, "grad_norm": 1.515193223953247, "learning_rate": 7.1316000000000005e-06, "loss": 0.0237, "step": 14343 }, { "epoch": 0.7647278349416218, "grad_norm": 2.469487190246582, "learning_rate": 7.131400000000001e-06, "loss": 0.019, "step": 14344 }, { "epoch": 0.7647811483712748, "grad_norm": 2.0642147064208984, "learning_rate": 7.1312000000000004e-06, "loss": 0.0336, "step": 14345 }, { "epoch": 0.7648344618009276, "grad_norm": 2.6865100860595703, "learning_rate": 7.131e-06, "loss": 0.0546, "step": 14346 }, { "epoch": 0.7648877752305806, "grad_norm": 1.462775707244873, "learning_rate": 7.130800000000001e-06, "loss": 0.0217, "step": 14347 }, { "epoch": 0.7649410886602335, "grad_norm": 1.7018767595291138, "learning_rate": 7.130600000000001e-06, "loss": 0.0269, "step": 14348 }, { "epoch": 0.7649944020898864, "grad_norm": 0.7979613542556763, "learning_rate": 7.1304e-06, "loss": 0.0066, "step": 14349 }, { "epoch": 0.7650477155195393, "grad_norm": 1.0659432411193848, "learning_rate": 7.130200000000001e-06, "loss": 0.008, "step": 14350 }, { "epoch": 0.7651010289491923, "grad_norm": 2.9172163009643555, "learning_rate": 7.13e-06, "loss": 0.0183, "step": 14351 }, { "epoch": 0.7651543423788453, "grad_norm": 1.683320164680481, "learning_rate": 7.1298e-06, "loss": 0.0249, "step": 14352 }, { "epoch": 0.7652076558084981, "grad_norm": 2.5376672744750977, "learning_rate": 7.129600000000001e-06, "loss": 0.0276, "step": 14353 }, { "epoch": 0.7652609692381511, "grad_norm": 2.2576818466186523, "learning_rate": 7.129400000000001e-06, "loss": 0.0428, "step": 14354 }, { "epoch": 0.765314282667804, "grad_norm": 1.2810560464859009, "learning_rate": 7.1292e-06, "loss": 0.013, "step": 14355 }, { "epoch": 0.765367596097457, "grad_norm": 2.3942317962646484, "learning_rate": 7.129000000000001e-06, "loss": 0.0273, "step": 14356 }, { "epoch": 0.7654209095271098, "grad_norm": 1.9002060890197754, "learning_rate": 7.1288e-06, "loss": 0.038, "step": 14357 }, { "epoch": 0.7654742229567628, "grad_norm": 1.6735934019088745, "learning_rate": 7.128600000000001e-06, "loss": 0.0179, "step": 14358 }, { "epoch": 0.7655275363864157, "grad_norm": 3.441795825958252, "learning_rate": 7.128400000000001e-06, "loss": 0.031, "step": 14359 }, { "epoch": 0.7655808498160687, "grad_norm": 3.130842447280884, "learning_rate": 7.128200000000001e-06, "loss": 0.0514, "step": 14360 }, { "epoch": 0.7656341632457216, "grad_norm": 1.10149085521698, "learning_rate": 7.128e-06, "loss": 0.0188, "step": 14361 }, { "epoch": 0.7656874766753745, "grad_norm": 1.947178840637207, "learning_rate": 7.127800000000001e-06, "loss": 0.0285, "step": 14362 }, { "epoch": 0.7657407901050275, "grad_norm": 1.582580804824829, "learning_rate": 7.1276e-06, "loss": 0.0205, "step": 14363 }, { "epoch": 0.7657941035346804, "grad_norm": 2.4129557609558105, "learning_rate": 7.1274000000000006e-06, "loss": 0.0523, "step": 14364 }, { "epoch": 0.7658474169643333, "grad_norm": 0.9447708129882812, "learning_rate": 7.127200000000001e-06, "loss": 0.0102, "step": 14365 }, { "epoch": 0.7659007303939862, "grad_norm": 3.199582815170288, "learning_rate": 7.1270000000000005e-06, "loss": 0.038, "step": 14366 }, { "epoch": 0.7659540438236392, "grad_norm": 1.013060212135315, "learning_rate": 7.1268e-06, "loss": 0.0146, "step": 14367 }, { "epoch": 0.766007357253292, "grad_norm": 2.543738842010498, "learning_rate": 7.1266000000000005e-06, "loss": 0.0158, "step": 14368 }, { "epoch": 0.766060670682945, "grad_norm": 3.5282704830169678, "learning_rate": 7.126400000000001e-06, "loss": 0.0293, "step": 14369 }, { "epoch": 0.766113984112598, "grad_norm": 1.330938696861267, "learning_rate": 7.1262000000000005e-06, "loss": 0.0227, "step": 14370 }, { "epoch": 0.7661672975422509, "grad_norm": 1.2642008066177368, "learning_rate": 7.126000000000001e-06, "loss": 0.0158, "step": 14371 }, { "epoch": 0.7662206109719039, "grad_norm": 1.5301593542099, "learning_rate": 7.1258000000000005e-06, "loss": 0.027, "step": 14372 }, { "epoch": 0.7662739244015567, "grad_norm": 2.2631049156188965, "learning_rate": 7.1256e-06, "loss": 0.0303, "step": 14373 }, { "epoch": 0.7663272378312097, "grad_norm": 1.836638331413269, "learning_rate": 7.1254e-06, "loss": 0.0072, "step": 14374 }, { "epoch": 0.7663805512608626, "grad_norm": 4.164103031158447, "learning_rate": 7.125200000000001e-06, "loss": 0.0536, "step": 14375 }, { "epoch": 0.7664338646905156, "grad_norm": 1.4735995531082153, "learning_rate": 7.125e-06, "loss": 0.0198, "step": 14376 }, { "epoch": 0.7664871781201684, "grad_norm": 2.475743532180786, "learning_rate": 7.124800000000001e-06, "loss": 0.0337, "step": 14377 }, { "epoch": 0.7665404915498214, "grad_norm": 2.777400493621826, "learning_rate": 7.1246e-06, "loss": 0.0256, "step": 14378 }, { "epoch": 0.7665938049794744, "grad_norm": 1.8235642910003662, "learning_rate": 7.1244e-06, "loss": 0.0322, "step": 14379 }, { "epoch": 0.7666471184091272, "grad_norm": 1.432307243347168, "learning_rate": 7.124200000000001e-06, "loss": 0.0062, "step": 14380 }, { "epoch": 0.7667004318387802, "grad_norm": 3.9262375831604004, "learning_rate": 7.124000000000001e-06, "loss": 0.1172, "step": 14381 }, { "epoch": 0.7667537452684331, "grad_norm": 2.578199863433838, "learning_rate": 7.1238e-06, "loss": 0.0229, "step": 14382 }, { "epoch": 0.7668070586980861, "grad_norm": 3.2287144660949707, "learning_rate": 7.123600000000001e-06, "loss": 0.0308, "step": 14383 }, { "epoch": 0.7668603721277389, "grad_norm": 3.261045217514038, "learning_rate": 7.1234e-06, "loss": 0.0403, "step": 14384 }, { "epoch": 0.7669136855573919, "grad_norm": 2.6067557334899902, "learning_rate": 7.1232e-06, "loss": 0.0286, "step": 14385 }, { "epoch": 0.7669669989870448, "grad_norm": 2.3017466068267822, "learning_rate": 7.123000000000001e-06, "loss": 0.0146, "step": 14386 }, { "epoch": 0.7670203124166978, "grad_norm": 1.7280746698379517, "learning_rate": 7.122800000000001e-06, "loss": 0.0198, "step": 14387 }, { "epoch": 0.7670736258463507, "grad_norm": 0.7787212133407593, "learning_rate": 7.1226e-06, "loss": 0.0224, "step": 14388 }, { "epoch": 0.7671269392760036, "grad_norm": 1.1046335697174072, "learning_rate": 7.122400000000001e-06, "loss": 0.021, "step": 14389 }, { "epoch": 0.7671802527056566, "grad_norm": 3.2252840995788574, "learning_rate": 7.1222e-06, "loss": 0.0605, "step": 14390 }, { "epoch": 0.7672335661353095, "grad_norm": 2.491992712020874, "learning_rate": 7.1220000000000014e-06, "loss": 0.0409, "step": 14391 }, { "epoch": 0.7672868795649624, "grad_norm": 1.549044132232666, "learning_rate": 7.121800000000001e-06, "loss": 0.0339, "step": 14392 }, { "epoch": 0.7673401929946153, "grad_norm": 1.9691554307937622, "learning_rate": 7.1216000000000006e-06, "loss": 0.0491, "step": 14393 }, { "epoch": 0.7673935064242683, "grad_norm": 1.9608755111694336, "learning_rate": 7.1214e-06, "loss": 0.0212, "step": 14394 }, { "epoch": 0.7674468198539212, "grad_norm": 3.1770853996276855, "learning_rate": 7.1212000000000005e-06, "loss": 0.0221, "step": 14395 }, { "epoch": 0.7675001332835741, "grad_norm": 3.6329617500305176, "learning_rate": 7.121e-06, "loss": 0.0492, "step": 14396 }, { "epoch": 0.7675534467132271, "grad_norm": 2.2817113399505615, "learning_rate": 7.120800000000001e-06, "loss": 0.0348, "step": 14397 }, { "epoch": 0.76760676014288, "grad_norm": 2.875112295150757, "learning_rate": 7.120600000000001e-06, "loss": 0.0401, "step": 14398 }, { "epoch": 0.767660073572533, "grad_norm": 1.8167213201522827, "learning_rate": 7.1204000000000005e-06, "loss": 0.0144, "step": 14399 }, { "epoch": 0.7677133870021858, "grad_norm": 2.6105382442474365, "learning_rate": 7.1202e-06, "loss": 0.0451, "step": 14400 }, { "epoch": 0.7677667004318388, "grad_norm": 1.886824369430542, "learning_rate": 7.1200000000000004e-06, "loss": 0.0445, "step": 14401 }, { "epoch": 0.7678200138614917, "grad_norm": 3.856613874435425, "learning_rate": 7.1198e-06, "loss": 0.044, "step": 14402 }, { "epoch": 0.7678733272911447, "grad_norm": 1.6183176040649414, "learning_rate": 7.119600000000001e-06, "loss": 0.0179, "step": 14403 }, { "epoch": 0.7679266407207975, "grad_norm": 3.843363046646118, "learning_rate": 7.119400000000001e-06, "loss": 0.033, "step": 14404 }, { "epoch": 0.7679799541504505, "grad_norm": 1.9372111558914185, "learning_rate": 7.1192e-06, "loss": 0.0434, "step": 14405 }, { "epoch": 0.7680332675801034, "grad_norm": 2.863226890563965, "learning_rate": 7.119e-06, "loss": 0.0417, "step": 14406 }, { "epoch": 0.7680865810097564, "grad_norm": 1.7619224786758423, "learning_rate": 7.1188e-06, "loss": 0.027, "step": 14407 }, { "epoch": 0.7681398944394093, "grad_norm": 1.3765536546707153, "learning_rate": 7.118600000000001e-06, "loss": 0.0221, "step": 14408 }, { "epoch": 0.7681932078690622, "grad_norm": 2.2316713333129883, "learning_rate": 7.118400000000001e-06, "loss": 0.0273, "step": 14409 }, { "epoch": 0.7682465212987152, "grad_norm": 5.826817512512207, "learning_rate": 7.118200000000001e-06, "loss": 0.0498, "step": 14410 }, { "epoch": 0.768299834728368, "grad_norm": 2.4324092864990234, "learning_rate": 7.118e-06, "loss": 0.0316, "step": 14411 }, { "epoch": 0.768353148158021, "grad_norm": 4.050057411193848, "learning_rate": 7.1178e-06, "loss": 0.0389, "step": 14412 }, { "epoch": 0.7684064615876739, "grad_norm": 2.773279905319214, "learning_rate": 7.1176e-06, "loss": 0.0446, "step": 14413 }, { "epoch": 0.7684597750173269, "grad_norm": 3.416818618774414, "learning_rate": 7.117400000000001e-06, "loss": 0.0303, "step": 14414 }, { "epoch": 0.7685130884469797, "grad_norm": 2.9824352264404297, "learning_rate": 7.117200000000001e-06, "loss": 0.0187, "step": 14415 }, { "epoch": 0.7685664018766327, "grad_norm": 3.330315589904785, "learning_rate": 7.117000000000001e-06, "loss": 0.0521, "step": 14416 }, { "epoch": 0.7686197153062857, "grad_norm": 1.8734781742095947, "learning_rate": 7.1168e-06, "loss": 0.0391, "step": 14417 }, { "epoch": 0.7686730287359386, "grad_norm": 2.4393484592437744, "learning_rate": 7.1166e-06, "loss": 0.0219, "step": 14418 }, { "epoch": 0.7687263421655915, "grad_norm": 1.7455189228057861, "learning_rate": 7.116400000000001e-06, "loss": 0.0208, "step": 14419 }, { "epoch": 0.7687796555952444, "grad_norm": 2.448253870010376, "learning_rate": 7.116200000000001e-06, "loss": 0.0484, "step": 14420 }, { "epoch": 0.7688329690248974, "grad_norm": 3.1303884983062744, "learning_rate": 7.116000000000001e-06, "loss": 0.0437, "step": 14421 }, { "epoch": 0.7688862824545503, "grad_norm": 1.1227532625198364, "learning_rate": 7.1158000000000005e-06, "loss": 0.0181, "step": 14422 }, { "epoch": 0.7689395958842032, "grad_norm": 2.750603675842285, "learning_rate": 7.1156e-06, "loss": 0.0627, "step": 14423 }, { "epoch": 0.7689929093138561, "grad_norm": 2.043259620666504, "learning_rate": 7.1154e-06, "loss": 0.0347, "step": 14424 }, { "epoch": 0.7690462227435091, "grad_norm": 3.262265205383301, "learning_rate": 7.115200000000001e-06, "loss": 0.0603, "step": 14425 }, { "epoch": 0.7690995361731621, "grad_norm": 2.4511759281158447, "learning_rate": 7.1150000000000005e-06, "loss": 0.0548, "step": 14426 }, { "epoch": 0.7691528496028149, "grad_norm": 2.0236101150512695, "learning_rate": 7.114800000000001e-06, "loss": 0.0136, "step": 14427 }, { "epoch": 0.7692061630324679, "grad_norm": 1.1729423999786377, "learning_rate": 7.1146000000000005e-06, "loss": 0.0132, "step": 14428 }, { "epoch": 0.7692594764621208, "grad_norm": 2.947242259979248, "learning_rate": 7.1144e-06, "loss": 0.0537, "step": 14429 }, { "epoch": 0.7693127898917738, "grad_norm": 1.78037691116333, "learning_rate": 7.114200000000001e-06, "loss": 0.0264, "step": 14430 }, { "epoch": 0.7693661033214266, "grad_norm": 2.259389877319336, "learning_rate": 7.114000000000001e-06, "loss": 0.0514, "step": 14431 }, { "epoch": 0.7694194167510796, "grad_norm": 1.2229617834091187, "learning_rate": 7.1138e-06, "loss": 0.0289, "step": 14432 }, { "epoch": 0.7694727301807325, "grad_norm": 2.0318615436553955, "learning_rate": 7.113600000000001e-06, "loss": 0.0296, "step": 14433 }, { "epoch": 0.7695260436103855, "grad_norm": 1.3272490501403809, "learning_rate": 7.1134e-06, "loss": 0.0253, "step": 14434 }, { "epoch": 0.7695793570400384, "grad_norm": 2.4392285346984863, "learning_rate": 7.1132e-06, "loss": 0.0334, "step": 14435 }, { "epoch": 0.7696326704696913, "grad_norm": 1.7374093532562256, "learning_rate": 7.113000000000001e-06, "loss": 0.0313, "step": 14436 }, { "epoch": 0.7696859838993443, "grad_norm": 3.7791688442230225, "learning_rate": 7.112800000000001e-06, "loss": 0.0392, "step": 14437 }, { "epoch": 0.7697392973289972, "grad_norm": 1.271144986152649, "learning_rate": 7.1126e-06, "loss": 0.0284, "step": 14438 }, { "epoch": 0.7697926107586501, "grad_norm": 1.8949687480926514, "learning_rate": 7.112400000000001e-06, "loss": 0.022, "step": 14439 }, { "epoch": 0.769845924188303, "grad_norm": 2.837557554244995, "learning_rate": 7.1122e-06, "loss": 0.0549, "step": 14440 }, { "epoch": 0.769899237617956, "grad_norm": 3.071502447128296, "learning_rate": 7.1120000000000015e-06, "loss": 0.0436, "step": 14441 }, { "epoch": 0.7699525510476088, "grad_norm": 2.443561553955078, "learning_rate": 7.111800000000001e-06, "loss": 0.049, "step": 14442 }, { "epoch": 0.7700058644772618, "grad_norm": 2.207190752029419, "learning_rate": 7.111600000000001e-06, "loss": 0.0395, "step": 14443 }, { "epoch": 0.7700591779069148, "grad_norm": 1.2530995607376099, "learning_rate": 7.1114e-06, "loss": 0.0121, "step": 14444 }, { "epoch": 0.7701124913365677, "grad_norm": 1.984108805656433, "learning_rate": 7.111200000000001e-06, "loss": 0.0383, "step": 14445 }, { "epoch": 0.7701658047662207, "grad_norm": 2.220896005630493, "learning_rate": 7.111e-06, "loss": 0.0196, "step": 14446 }, { "epoch": 0.7702191181958735, "grad_norm": 2.990288257598877, "learning_rate": 7.1108000000000014e-06, "loss": 0.0544, "step": 14447 }, { "epoch": 0.7702724316255265, "grad_norm": 3.093278408050537, "learning_rate": 7.110600000000001e-06, "loss": 0.0566, "step": 14448 }, { "epoch": 0.7703257450551794, "grad_norm": 3.8345954418182373, "learning_rate": 7.1104000000000006e-06, "loss": 0.0592, "step": 14449 }, { "epoch": 0.7703790584848323, "grad_norm": 1.519704818725586, "learning_rate": 7.1102e-06, "loss": 0.0216, "step": 14450 }, { "epoch": 0.7704323719144852, "grad_norm": 4.013216972351074, "learning_rate": 7.1100000000000005e-06, "loss": 0.0361, "step": 14451 }, { "epoch": 0.7704856853441382, "grad_norm": 1.839638352394104, "learning_rate": 7.1098e-06, "loss": 0.0317, "step": 14452 }, { "epoch": 0.7705389987737912, "grad_norm": 1.7554922103881836, "learning_rate": 7.109600000000001e-06, "loss": 0.0412, "step": 14453 }, { "epoch": 0.770592312203444, "grad_norm": 1.1971153020858765, "learning_rate": 7.109400000000001e-06, "loss": 0.0364, "step": 14454 }, { "epoch": 0.770645625633097, "grad_norm": 3.9435973167419434, "learning_rate": 7.1092000000000005e-06, "loss": 0.0529, "step": 14455 }, { "epoch": 0.7706989390627499, "grad_norm": 2.128180742263794, "learning_rate": 7.109e-06, "loss": 0.0415, "step": 14456 }, { "epoch": 0.7707522524924029, "grad_norm": 2.215935230255127, "learning_rate": 7.1088000000000004e-06, "loss": 0.0301, "step": 14457 }, { "epoch": 0.7708055659220557, "grad_norm": 2.548487901687622, "learning_rate": 7.108600000000001e-06, "loss": 0.0459, "step": 14458 }, { "epoch": 0.7708588793517087, "grad_norm": 2.5655128955841064, "learning_rate": 7.108400000000001e-06, "loss": 0.0253, "step": 14459 }, { "epoch": 0.7709121927813616, "grad_norm": 1.6669645309448242, "learning_rate": 7.108200000000001e-06, "loss": 0.026, "step": 14460 }, { "epoch": 0.7709655062110146, "grad_norm": 1.1704225540161133, "learning_rate": 7.108e-06, "loss": 0.0166, "step": 14461 }, { "epoch": 0.7710188196406674, "grad_norm": 1.6898101568222046, "learning_rate": 7.1078e-06, "loss": 0.0313, "step": 14462 }, { "epoch": 0.7710721330703204, "grad_norm": 1.194304347038269, "learning_rate": 7.1076e-06, "loss": 0.0131, "step": 14463 }, { "epoch": 0.7711254464999734, "grad_norm": 1.5343332290649414, "learning_rate": 7.107400000000001e-06, "loss": 0.0329, "step": 14464 }, { "epoch": 0.7711787599296263, "grad_norm": 3.5135936737060547, "learning_rate": 7.107200000000001e-06, "loss": 0.052, "step": 14465 }, { "epoch": 0.7712320733592792, "grad_norm": 2.97147274017334, "learning_rate": 7.107000000000001e-06, "loss": 0.0411, "step": 14466 }, { "epoch": 0.7712853867889321, "grad_norm": 1.1212339401245117, "learning_rate": 7.1068e-06, "loss": 0.0217, "step": 14467 }, { "epoch": 0.7713387002185851, "grad_norm": 1.9003252983093262, "learning_rate": 7.1066e-06, "loss": 0.0196, "step": 14468 }, { "epoch": 0.771392013648238, "grad_norm": 1.6915396451950073, "learning_rate": 7.106400000000001e-06, "loss": 0.0286, "step": 14469 }, { "epoch": 0.7714453270778909, "grad_norm": 1.934356451034546, "learning_rate": 7.106200000000001e-06, "loss": 0.0253, "step": 14470 }, { "epoch": 0.7714986405075438, "grad_norm": 1.7334858179092407, "learning_rate": 7.106000000000001e-06, "loss": 0.0343, "step": 14471 }, { "epoch": 0.7715519539371968, "grad_norm": 2.154125928878784, "learning_rate": 7.105800000000001e-06, "loss": 0.0648, "step": 14472 }, { "epoch": 0.7716052673668498, "grad_norm": 3.797774314880371, "learning_rate": 7.1056e-06, "loss": 0.0644, "step": 14473 }, { "epoch": 0.7716585807965026, "grad_norm": 1.9999537467956543, "learning_rate": 7.1054e-06, "loss": 0.0477, "step": 14474 }, { "epoch": 0.7717118942261556, "grad_norm": 1.9447109699249268, "learning_rate": 7.105200000000001e-06, "loss": 0.0317, "step": 14475 }, { "epoch": 0.7717652076558085, "grad_norm": 1.804636001586914, "learning_rate": 7.105000000000001e-06, "loss": 0.0234, "step": 14476 }, { "epoch": 0.7718185210854615, "grad_norm": 2.5562026500701904, "learning_rate": 7.104800000000001e-06, "loss": 0.0501, "step": 14477 }, { "epoch": 0.7718718345151143, "grad_norm": 1.8401267528533936, "learning_rate": 7.1046000000000006e-06, "loss": 0.0311, "step": 14478 }, { "epoch": 0.7719251479447673, "grad_norm": 0.8821523189544678, "learning_rate": 7.1044e-06, "loss": 0.0162, "step": 14479 }, { "epoch": 0.7719784613744202, "grad_norm": 2.059499740600586, "learning_rate": 7.104200000000001e-06, "loss": 0.0224, "step": 14480 }, { "epoch": 0.7720317748040731, "grad_norm": 2.2596077919006348, "learning_rate": 7.104000000000001e-06, "loss": 0.0288, "step": 14481 }, { "epoch": 0.7720850882337261, "grad_norm": 1.8048428297042847, "learning_rate": 7.1038000000000005e-06, "loss": 0.0199, "step": 14482 }, { "epoch": 0.772138401663379, "grad_norm": 2.6799254417419434, "learning_rate": 7.103600000000001e-06, "loss": 0.0327, "step": 14483 }, { "epoch": 0.772191715093032, "grad_norm": 2.025667905807495, "learning_rate": 7.1034000000000005e-06, "loss": 0.0323, "step": 14484 }, { "epoch": 0.7722450285226848, "grad_norm": 3.169656753540039, "learning_rate": 7.1032e-06, "loss": 0.049, "step": 14485 }, { "epoch": 0.7722983419523378, "grad_norm": 2.0450997352600098, "learning_rate": 7.103000000000001e-06, "loss": 0.0214, "step": 14486 }, { "epoch": 0.7723516553819907, "grad_norm": 2.965031862258911, "learning_rate": 7.102800000000001e-06, "loss": 0.0496, "step": 14487 }, { "epoch": 0.7724049688116437, "grad_norm": 0.9466536641120911, "learning_rate": 7.1026e-06, "loss": 0.0094, "step": 14488 }, { "epoch": 0.7724582822412965, "grad_norm": 1.8885352611541748, "learning_rate": 7.102400000000001e-06, "loss": 0.0352, "step": 14489 }, { "epoch": 0.7725115956709495, "grad_norm": 2.841259479522705, "learning_rate": 7.1022e-06, "loss": 0.0441, "step": 14490 }, { "epoch": 0.7725649091006025, "grad_norm": 0.770152747631073, "learning_rate": 7.102000000000001e-06, "loss": 0.0097, "step": 14491 }, { "epoch": 0.7726182225302554, "grad_norm": 3.117825508117676, "learning_rate": 7.101800000000001e-06, "loss": 0.0368, "step": 14492 }, { "epoch": 0.7726715359599083, "grad_norm": 4.243731498718262, "learning_rate": 7.101600000000001e-06, "loss": 0.0375, "step": 14493 }, { "epoch": 0.7727248493895612, "grad_norm": 1.5643926858901978, "learning_rate": 7.1014e-06, "loss": 0.0198, "step": 14494 }, { "epoch": 0.7727781628192142, "grad_norm": 1.861546516418457, "learning_rate": 7.101200000000001e-06, "loss": 0.0335, "step": 14495 }, { "epoch": 0.772831476248867, "grad_norm": 1.8397775888442993, "learning_rate": 7.101e-06, "loss": 0.0282, "step": 14496 }, { "epoch": 0.77288478967852, "grad_norm": 1.9741226434707642, "learning_rate": 7.100800000000001e-06, "loss": 0.0298, "step": 14497 }, { "epoch": 0.7729381031081729, "grad_norm": 0.9679743051528931, "learning_rate": 7.100600000000001e-06, "loss": 0.0089, "step": 14498 }, { "epoch": 0.7729914165378259, "grad_norm": 1.4819295406341553, "learning_rate": 7.100400000000001e-06, "loss": 0.015, "step": 14499 }, { "epoch": 0.7730447299674789, "grad_norm": 1.5928370952606201, "learning_rate": 7.1002e-06, "loss": 0.0335, "step": 14500 }, { "epoch": 0.7730980433971317, "grad_norm": 1.0949463844299316, "learning_rate": 7.100000000000001e-06, "loss": 0.0121, "step": 14501 }, { "epoch": 0.7731513568267847, "grad_norm": 4.6429524421691895, "learning_rate": 7.099800000000001e-06, "loss": 0.0435, "step": 14502 }, { "epoch": 0.7732046702564376, "grad_norm": 1.4150253534317017, "learning_rate": 7.099600000000001e-06, "loss": 0.013, "step": 14503 }, { "epoch": 0.7732579836860906, "grad_norm": 1.1193952560424805, "learning_rate": 7.099400000000001e-06, "loss": 0.0096, "step": 14504 }, { "epoch": 0.7733112971157434, "grad_norm": 2.261059284210205, "learning_rate": 7.0992000000000006e-06, "loss": 0.0253, "step": 14505 }, { "epoch": 0.7733646105453964, "grad_norm": 1.3234405517578125, "learning_rate": 7.099e-06, "loss": 0.0141, "step": 14506 }, { "epoch": 0.7734179239750493, "grad_norm": 3.3998711109161377, "learning_rate": 7.0988000000000005e-06, "loss": 0.0303, "step": 14507 }, { "epoch": 0.7734712374047022, "grad_norm": 1.1581988334655762, "learning_rate": 7.098600000000001e-06, "loss": 0.0095, "step": 14508 }, { "epoch": 0.7735245508343552, "grad_norm": 0.8027582764625549, "learning_rate": 7.0984000000000005e-06, "loss": 0.0056, "step": 14509 }, { "epoch": 0.7735778642640081, "grad_norm": 0.9614761471748352, "learning_rate": 7.098200000000001e-06, "loss": 0.0079, "step": 14510 }, { "epoch": 0.7736311776936611, "grad_norm": 3.1929569244384766, "learning_rate": 7.0980000000000005e-06, "loss": 0.0474, "step": 14511 }, { "epoch": 0.7736844911233139, "grad_norm": 3.7900784015655518, "learning_rate": 7.0978e-06, "loss": 0.0406, "step": 14512 }, { "epoch": 0.7737378045529669, "grad_norm": 1.8437258005142212, "learning_rate": 7.0976000000000005e-06, "loss": 0.0355, "step": 14513 }, { "epoch": 0.7737911179826198, "grad_norm": 3.207148790359497, "learning_rate": 7.097400000000001e-06, "loss": 0.0281, "step": 14514 }, { "epoch": 0.7738444314122728, "grad_norm": 2.384369373321533, "learning_rate": 7.0972e-06, "loss": 0.0369, "step": 14515 }, { "epoch": 0.7738977448419256, "grad_norm": 1.7737054824829102, "learning_rate": 7.097000000000001e-06, "loss": 0.0081, "step": 14516 }, { "epoch": 0.7739510582715786, "grad_norm": 3.6744213104248047, "learning_rate": 7.0968e-06, "loss": 0.0954, "step": 14517 }, { "epoch": 0.7740043717012315, "grad_norm": 2.076580286026001, "learning_rate": 7.0966e-06, "loss": 0.0371, "step": 14518 }, { "epoch": 0.7740576851308845, "grad_norm": 2.1074841022491455, "learning_rate": 7.096400000000001e-06, "loss": 0.0302, "step": 14519 }, { "epoch": 0.7741109985605374, "grad_norm": 3.717320203781128, "learning_rate": 7.096200000000001e-06, "loss": 0.047, "step": 14520 }, { "epoch": 0.7741643119901903, "grad_norm": 2.2811572551727295, "learning_rate": 7.096e-06, "loss": 0.0338, "step": 14521 }, { "epoch": 0.7742176254198433, "grad_norm": 2.3455443382263184, "learning_rate": 7.095800000000001e-06, "loss": 0.0238, "step": 14522 }, { "epoch": 0.7742709388494962, "grad_norm": 1.1064040660858154, "learning_rate": 7.0956e-06, "loss": 0.0092, "step": 14523 }, { "epoch": 0.7743242522791491, "grad_norm": 2.3812358379364014, "learning_rate": 7.0954e-06, "loss": 0.049, "step": 14524 }, { "epoch": 0.774377565708802, "grad_norm": 1.8111191987991333, "learning_rate": 7.095200000000001e-06, "loss": 0.0266, "step": 14525 }, { "epoch": 0.774430879138455, "grad_norm": 0.8785043358802795, "learning_rate": 7.095000000000001e-06, "loss": 0.0074, "step": 14526 }, { "epoch": 0.7744841925681079, "grad_norm": 2.8058886528015137, "learning_rate": 7.0948e-06, "loss": 0.0255, "step": 14527 }, { "epoch": 0.7745375059977608, "grad_norm": 1.9410020112991333, "learning_rate": 7.094600000000001e-06, "loss": 0.021, "step": 14528 }, { "epoch": 0.7745908194274138, "grad_norm": 2.160130500793457, "learning_rate": 7.0944e-06, "loss": 0.0315, "step": 14529 }, { "epoch": 0.7746441328570667, "grad_norm": 2.79240083694458, "learning_rate": 7.0942000000000015e-06, "loss": 0.0366, "step": 14530 }, { "epoch": 0.7746974462867197, "grad_norm": 1.402408242225647, "learning_rate": 7.094000000000001e-06, "loss": 0.0162, "step": 14531 }, { "epoch": 0.7747507597163725, "grad_norm": 2.0527212619781494, "learning_rate": 7.093800000000001e-06, "loss": 0.0485, "step": 14532 }, { "epoch": 0.7748040731460255, "grad_norm": 2.5059804916381836, "learning_rate": 7.0936e-06, "loss": 0.0485, "step": 14533 }, { "epoch": 0.7748573865756784, "grad_norm": 2.592329740524292, "learning_rate": 7.0934000000000006e-06, "loss": 0.0427, "step": 14534 }, { "epoch": 0.7749107000053314, "grad_norm": 1.54013192653656, "learning_rate": 7.0932e-06, "loss": 0.0333, "step": 14535 }, { "epoch": 0.7749640134349842, "grad_norm": 2.2465906143188477, "learning_rate": 7.093000000000001e-06, "loss": 0.0216, "step": 14536 }, { "epoch": 0.7750173268646372, "grad_norm": 2.110724925994873, "learning_rate": 7.092800000000001e-06, "loss": 0.043, "step": 14537 }, { "epoch": 0.7750706402942902, "grad_norm": 0.9575936794281006, "learning_rate": 7.0926000000000005e-06, "loss": 0.0126, "step": 14538 }, { "epoch": 0.775123953723943, "grad_norm": 2.8888351917266846, "learning_rate": 7.0924e-06, "loss": 0.0458, "step": 14539 }, { "epoch": 0.775177267153596, "grad_norm": 1.6659457683563232, "learning_rate": 7.0922000000000005e-06, "loss": 0.0243, "step": 14540 }, { "epoch": 0.7752305805832489, "grad_norm": 1.3753608465194702, "learning_rate": 7.092000000000001e-06, "loss": 0.0214, "step": 14541 }, { "epoch": 0.7752838940129019, "grad_norm": 1.7591311931610107, "learning_rate": 7.091800000000001e-06, "loss": 0.0286, "step": 14542 }, { "epoch": 0.7753372074425547, "grad_norm": 3.879146099090576, "learning_rate": 7.091600000000001e-06, "loss": 0.0516, "step": 14543 }, { "epoch": 0.7753905208722077, "grad_norm": 0.7728033065795898, "learning_rate": 7.0914e-06, "loss": 0.013, "step": 14544 }, { "epoch": 0.7754438343018606, "grad_norm": 2.198856830596924, "learning_rate": 7.0912e-06, "loss": 0.0283, "step": 14545 }, { "epoch": 0.7754971477315136, "grad_norm": 2.562021255493164, "learning_rate": 7.091e-06, "loss": 0.023, "step": 14546 }, { "epoch": 0.7755504611611665, "grad_norm": 2.3639185428619385, "learning_rate": 7.090800000000001e-06, "loss": 0.0283, "step": 14547 }, { "epoch": 0.7756037745908194, "grad_norm": 1.5356428623199463, "learning_rate": 7.090600000000001e-06, "loss": 0.0211, "step": 14548 }, { "epoch": 0.7756570880204724, "grad_norm": 1.9894901514053345, "learning_rate": 7.090400000000001e-06, "loss": 0.0432, "step": 14549 }, { "epoch": 0.7757104014501253, "grad_norm": 2.3287546634674072, "learning_rate": 7.0902e-06, "loss": 0.048, "step": 14550 }, { "epoch": 0.7757637148797782, "grad_norm": 2.493305206298828, "learning_rate": 7.09e-06, "loss": 0.0607, "step": 14551 }, { "epoch": 0.7758170283094311, "grad_norm": 1.208970546722412, "learning_rate": 7.089800000000001e-06, "loss": 0.0112, "step": 14552 }, { "epoch": 0.7758703417390841, "grad_norm": 1.8443188667297363, "learning_rate": 7.089600000000001e-06, "loss": 0.0271, "step": 14553 }, { "epoch": 0.775923655168737, "grad_norm": 2.9656026363372803, "learning_rate": 7.089400000000001e-06, "loss": 0.0201, "step": 14554 }, { "epoch": 0.7759769685983899, "grad_norm": 1.1248716115951538, "learning_rate": 7.089200000000001e-06, "loss": 0.0142, "step": 14555 }, { "epoch": 0.7760302820280429, "grad_norm": 1.688807487487793, "learning_rate": 7.089e-06, "loss": 0.0351, "step": 14556 }, { "epoch": 0.7760835954576958, "grad_norm": 2.2755587100982666, "learning_rate": 7.0888e-06, "loss": 0.0484, "step": 14557 }, { "epoch": 0.7761369088873488, "grad_norm": 3.0803987979888916, "learning_rate": 7.088600000000001e-06, "loss": 0.0413, "step": 14558 }, { "epoch": 0.7761902223170016, "grad_norm": 2.365323066711426, "learning_rate": 7.088400000000001e-06, "loss": 0.0229, "step": 14559 }, { "epoch": 0.7762435357466546, "grad_norm": 1.464083194732666, "learning_rate": 7.088200000000001e-06, "loss": 0.0202, "step": 14560 }, { "epoch": 0.7762968491763075, "grad_norm": 2.199066400527954, "learning_rate": 7.088000000000001e-06, "loss": 0.037, "step": 14561 }, { "epoch": 0.7763501626059605, "grad_norm": 1.647230863571167, "learning_rate": 7.0878e-06, "loss": 0.0139, "step": 14562 }, { "epoch": 0.7764034760356133, "grad_norm": 2.4388391971588135, "learning_rate": 7.0876e-06, "loss": 0.0291, "step": 14563 }, { "epoch": 0.7764567894652663, "grad_norm": 2.0736491680145264, "learning_rate": 7.087400000000001e-06, "loss": 0.0184, "step": 14564 }, { "epoch": 0.7765101028949193, "grad_norm": 4.810314178466797, "learning_rate": 7.0872000000000005e-06, "loss": 0.0651, "step": 14565 }, { "epoch": 0.7765634163245722, "grad_norm": 4.008056163787842, "learning_rate": 7.087000000000001e-06, "loss": 0.0294, "step": 14566 }, { "epoch": 0.7766167297542251, "grad_norm": 0.6230133175849915, "learning_rate": 7.0868000000000005e-06, "loss": 0.0064, "step": 14567 }, { "epoch": 0.776670043183878, "grad_norm": 4.736561298370361, "learning_rate": 7.0866e-06, "loss": 0.0489, "step": 14568 }, { "epoch": 0.776723356613531, "grad_norm": 3.1297531127929688, "learning_rate": 7.086400000000001e-06, "loss": 0.0466, "step": 14569 }, { "epoch": 0.7767766700431838, "grad_norm": 1.587092399597168, "learning_rate": 7.086200000000001e-06, "loss": 0.0158, "step": 14570 }, { "epoch": 0.7768299834728368, "grad_norm": 2.3153021335601807, "learning_rate": 7.0860000000000004e-06, "loss": 0.0299, "step": 14571 }, { "epoch": 0.7768832969024897, "grad_norm": 3.366119384765625, "learning_rate": 7.085800000000001e-06, "loss": 0.0414, "step": 14572 }, { "epoch": 0.7769366103321427, "grad_norm": 2.2164225578308105, "learning_rate": 7.0856e-06, "loss": 0.0418, "step": 14573 }, { "epoch": 0.7769899237617955, "grad_norm": 1.5731544494628906, "learning_rate": 7.0854e-06, "loss": 0.0141, "step": 14574 }, { "epoch": 0.7770432371914485, "grad_norm": 1.5844529867172241, "learning_rate": 7.085200000000001e-06, "loss": 0.0095, "step": 14575 }, { "epoch": 0.7770965506211015, "grad_norm": 2.95766019821167, "learning_rate": 7.085000000000001e-06, "loss": 0.0514, "step": 14576 }, { "epoch": 0.7771498640507544, "grad_norm": 2.165688991546631, "learning_rate": 7.0848e-06, "loss": 0.0434, "step": 14577 }, { "epoch": 0.7772031774804073, "grad_norm": 4.07206392288208, "learning_rate": 7.084600000000001e-06, "loss": 0.0343, "step": 14578 }, { "epoch": 0.7772564909100602, "grad_norm": 2.6201183795928955, "learning_rate": 7.0844e-06, "loss": 0.0401, "step": 14579 }, { "epoch": 0.7773098043397132, "grad_norm": 3.0791544914245605, "learning_rate": 7.084200000000001e-06, "loss": 0.0213, "step": 14580 }, { "epoch": 0.7773631177693661, "grad_norm": 2.1891584396362305, "learning_rate": 7.084000000000001e-06, "loss": 0.0473, "step": 14581 }, { "epoch": 0.777416431199019, "grad_norm": 2.1420562267303467, "learning_rate": 7.083800000000001e-06, "loss": 0.0288, "step": 14582 }, { "epoch": 0.7774697446286719, "grad_norm": 1.9793133735656738, "learning_rate": 7.0836e-06, "loss": 0.027, "step": 14583 }, { "epoch": 0.7775230580583249, "grad_norm": 1.428210973739624, "learning_rate": 7.083400000000001e-06, "loss": 0.0127, "step": 14584 }, { "epoch": 0.7775763714879779, "grad_norm": 1.8013652563095093, "learning_rate": 7.0832e-06, "loss": 0.0348, "step": 14585 }, { "epoch": 0.7776296849176307, "grad_norm": 3.7792913913726807, "learning_rate": 7.083000000000001e-06, "loss": 0.0379, "step": 14586 }, { "epoch": 0.7776829983472837, "grad_norm": 1.2510757446289062, "learning_rate": 7.082800000000001e-06, "loss": 0.0131, "step": 14587 }, { "epoch": 0.7777363117769366, "grad_norm": 3.609882354736328, "learning_rate": 7.082600000000001e-06, "loss": 0.014, "step": 14588 }, { "epoch": 0.7777896252065896, "grad_norm": 1.4948337078094482, "learning_rate": 7.0824e-06, "loss": 0.0321, "step": 14589 }, { "epoch": 0.7778429386362424, "grad_norm": 1.9962303638458252, "learning_rate": 7.0822000000000006e-06, "loss": 0.0162, "step": 14590 }, { "epoch": 0.7778962520658954, "grad_norm": 2.2314767837524414, "learning_rate": 7.082000000000001e-06, "loss": 0.0159, "step": 14591 }, { "epoch": 0.7779495654955483, "grad_norm": 1.046561360359192, "learning_rate": 7.0818000000000005e-06, "loss": 0.0114, "step": 14592 }, { "epoch": 0.7780028789252013, "grad_norm": 3.3849828243255615, "learning_rate": 7.081600000000001e-06, "loss": 0.0643, "step": 14593 }, { "epoch": 0.7780561923548542, "grad_norm": 2.5296242237091064, "learning_rate": 7.0814000000000005e-06, "loss": 0.0277, "step": 14594 }, { "epoch": 0.7781095057845071, "grad_norm": 1.783931851387024, "learning_rate": 7.0812e-06, "loss": 0.0221, "step": 14595 }, { "epoch": 0.7781628192141601, "grad_norm": 1.0711426734924316, "learning_rate": 7.0810000000000005e-06, "loss": 0.0084, "step": 14596 }, { "epoch": 0.778216132643813, "grad_norm": 1.501711368560791, "learning_rate": 7.080800000000001e-06, "loss": 0.0138, "step": 14597 }, { "epoch": 0.7782694460734659, "grad_norm": 2.6247026920318604, "learning_rate": 7.0806000000000004e-06, "loss": 0.0434, "step": 14598 }, { "epoch": 0.7783227595031188, "grad_norm": 1.8883914947509766, "learning_rate": 7.080400000000001e-06, "loss": 0.0621, "step": 14599 }, { "epoch": 0.7783760729327718, "grad_norm": 1.9181123971939087, "learning_rate": 7.0802e-06, "loss": 0.032, "step": 14600 }, { "epoch": 0.7784293863624246, "grad_norm": 2.7463202476501465, "learning_rate": 7.08e-06, "loss": 0.0628, "step": 14601 }, { "epoch": 0.7784826997920776, "grad_norm": 1.9910001754760742, "learning_rate": 7.079800000000001e-06, "loss": 0.0503, "step": 14602 }, { "epoch": 0.7785360132217306, "grad_norm": 2.661142587661743, "learning_rate": 7.079600000000001e-06, "loss": 0.0307, "step": 14603 }, { "epoch": 0.7785893266513835, "grad_norm": 1.2461040019989014, "learning_rate": 7.0794e-06, "loss": 0.0396, "step": 14604 }, { "epoch": 0.7786426400810365, "grad_norm": 2.985379219055176, "learning_rate": 7.079200000000001e-06, "loss": 0.029, "step": 14605 }, { "epoch": 0.7786959535106893, "grad_norm": 3.4865994453430176, "learning_rate": 7.079e-06, "loss": 0.0416, "step": 14606 }, { "epoch": 0.7787492669403423, "grad_norm": 2.8827133178710938, "learning_rate": 7.0788e-06, "loss": 0.0356, "step": 14607 }, { "epoch": 0.7788025803699952, "grad_norm": 2.4395358562469482, "learning_rate": 7.078600000000001e-06, "loss": 0.0377, "step": 14608 }, { "epoch": 0.7788558937996481, "grad_norm": 2.321993827819824, "learning_rate": 7.078400000000001e-06, "loss": 0.0203, "step": 14609 }, { "epoch": 0.778909207229301, "grad_norm": 2.2489333152770996, "learning_rate": 7.0782e-06, "loss": 0.0489, "step": 14610 }, { "epoch": 0.778962520658954, "grad_norm": 3.677445411682129, "learning_rate": 7.078000000000001e-06, "loss": 0.0217, "step": 14611 }, { "epoch": 0.779015834088607, "grad_norm": 2.08585786819458, "learning_rate": 7.0778e-06, "loss": 0.0422, "step": 14612 }, { "epoch": 0.7790691475182598, "grad_norm": 2.2546722888946533, "learning_rate": 7.0776e-06, "loss": 0.0392, "step": 14613 }, { "epoch": 0.7791224609479128, "grad_norm": 1.9482334852218628, "learning_rate": 7.077400000000001e-06, "loss": 0.0317, "step": 14614 }, { "epoch": 0.7791757743775657, "grad_norm": 1.3647574186325073, "learning_rate": 7.077200000000001e-06, "loss": 0.0299, "step": 14615 }, { "epoch": 0.7792290878072187, "grad_norm": 1.7003597021102905, "learning_rate": 7.077e-06, "loss": 0.0179, "step": 14616 }, { "epoch": 0.7792824012368715, "grad_norm": 2.619621515274048, "learning_rate": 7.076800000000001e-06, "loss": 0.0416, "step": 14617 }, { "epoch": 0.7793357146665245, "grad_norm": 2.156148910522461, "learning_rate": 7.0766e-06, "loss": 0.0239, "step": 14618 }, { "epoch": 0.7793890280961774, "grad_norm": 2.1914472579956055, "learning_rate": 7.076400000000001e-06, "loss": 0.0163, "step": 14619 }, { "epoch": 0.7794423415258304, "grad_norm": 2.4447529315948486, "learning_rate": 7.076200000000001e-06, "loss": 0.0251, "step": 14620 }, { "epoch": 0.7794956549554833, "grad_norm": 2.580016851425171, "learning_rate": 7.0760000000000005e-06, "loss": 0.035, "step": 14621 }, { "epoch": 0.7795489683851362, "grad_norm": 1.147798776626587, "learning_rate": 7.0758e-06, "loss": 0.0142, "step": 14622 }, { "epoch": 0.7796022818147892, "grad_norm": 1.8916813135147095, "learning_rate": 7.0756000000000005e-06, "loss": 0.018, "step": 14623 }, { "epoch": 0.7796555952444421, "grad_norm": 2.936086654663086, "learning_rate": 7.0754e-06, "loss": 0.0577, "step": 14624 }, { "epoch": 0.779708908674095, "grad_norm": 1.0547995567321777, "learning_rate": 7.075200000000001e-06, "loss": 0.0281, "step": 14625 }, { "epoch": 0.7797622221037479, "grad_norm": 0.8527222275733948, "learning_rate": 7.075000000000001e-06, "loss": 0.0101, "step": 14626 }, { "epoch": 0.7798155355334009, "grad_norm": 2.337916374206543, "learning_rate": 7.0748000000000004e-06, "loss": 0.0418, "step": 14627 }, { "epoch": 0.7798688489630538, "grad_norm": 1.6816768646240234, "learning_rate": 7.0746e-06, "loss": 0.028, "step": 14628 }, { "epoch": 0.7799221623927067, "grad_norm": 1.1275584697723389, "learning_rate": 7.0744e-06, "loss": 0.0097, "step": 14629 }, { "epoch": 0.7799754758223596, "grad_norm": 1.9093234539031982, "learning_rate": 7.074200000000001e-06, "loss": 0.0383, "step": 14630 }, { "epoch": 0.7800287892520126, "grad_norm": 1.5854452848434448, "learning_rate": 7.074000000000001e-06, "loss": 0.0254, "step": 14631 }, { "epoch": 0.7800821026816656, "grad_norm": 3.0353453159332275, "learning_rate": 7.073800000000001e-06, "loss": 0.017, "step": 14632 }, { "epoch": 0.7801354161113184, "grad_norm": 1.3566768169403076, "learning_rate": 7.0736e-06, "loss": 0.0105, "step": 14633 }, { "epoch": 0.7801887295409714, "grad_norm": 2.3829619884490967, "learning_rate": 7.0734e-06, "loss": 0.0647, "step": 14634 }, { "epoch": 0.7802420429706243, "grad_norm": 1.2270643711090088, "learning_rate": 7.0732e-06, "loss": 0.0069, "step": 14635 }, { "epoch": 0.7802953564002773, "grad_norm": 2.5551819801330566, "learning_rate": 7.073000000000001e-06, "loss": 0.0353, "step": 14636 }, { "epoch": 0.7803486698299301, "grad_norm": 1.9068444967269897, "learning_rate": 7.072800000000001e-06, "loss": 0.0314, "step": 14637 }, { "epoch": 0.7804019832595831, "grad_norm": 3.5339648723602295, "learning_rate": 7.072600000000001e-06, "loss": 0.0575, "step": 14638 }, { "epoch": 0.780455296689236, "grad_norm": 4.474453449249268, "learning_rate": 7.0724e-06, "loss": 0.0489, "step": 14639 }, { "epoch": 0.780508610118889, "grad_norm": 1.4022480249404907, "learning_rate": 7.0722e-06, "loss": 0.042, "step": 14640 }, { "epoch": 0.7805619235485419, "grad_norm": 1.7647937536239624, "learning_rate": 7.072000000000001e-06, "loss": 0.0351, "step": 14641 }, { "epoch": 0.7806152369781948, "grad_norm": 2.7822253704071045, "learning_rate": 7.071800000000001e-06, "loss": 0.0354, "step": 14642 }, { "epoch": 0.7806685504078478, "grad_norm": 1.9653295278549194, "learning_rate": 7.071600000000001e-06, "loss": 0.0235, "step": 14643 }, { "epoch": 0.7807218638375006, "grad_norm": 2.3789961338043213, "learning_rate": 7.071400000000001e-06, "loss": 0.0242, "step": 14644 }, { "epoch": 0.7807751772671536, "grad_norm": 1.3598718643188477, "learning_rate": 7.0712e-06, "loss": 0.0163, "step": 14645 }, { "epoch": 0.7808284906968065, "grad_norm": 0.9929391741752625, "learning_rate": 7.071e-06, "loss": 0.0131, "step": 14646 }, { "epoch": 0.7808818041264595, "grad_norm": 2.671064853668213, "learning_rate": 7.070800000000001e-06, "loss": 0.0363, "step": 14647 }, { "epoch": 0.7809351175561123, "grad_norm": 4.171601295471191, "learning_rate": 7.0706000000000005e-06, "loss": 0.0784, "step": 14648 }, { "epoch": 0.7809884309857653, "grad_norm": 1.7490851879119873, "learning_rate": 7.070400000000001e-06, "loss": 0.0297, "step": 14649 }, { "epoch": 0.7810417444154183, "grad_norm": 1.1865042448043823, "learning_rate": 7.0702000000000005e-06, "loss": 0.013, "step": 14650 }, { "epoch": 0.7810950578450712, "grad_norm": 4.169662952423096, "learning_rate": 7.07e-06, "loss": 0.0315, "step": 14651 }, { "epoch": 0.7811483712747241, "grad_norm": 2.3620381355285645, "learning_rate": 7.069800000000001e-06, "loss": 0.0193, "step": 14652 }, { "epoch": 0.781201684704377, "grad_norm": 1.92606782913208, "learning_rate": 7.069600000000001e-06, "loss": 0.0563, "step": 14653 }, { "epoch": 0.78125499813403, "grad_norm": 1.8059699535369873, "learning_rate": 7.0694000000000005e-06, "loss": 0.0154, "step": 14654 }, { "epoch": 0.7813083115636829, "grad_norm": 2.444131851196289, "learning_rate": 7.069200000000001e-06, "loss": 0.0323, "step": 14655 }, { "epoch": 0.7813616249933358, "grad_norm": 1.1004160642623901, "learning_rate": 7.069e-06, "loss": 0.0106, "step": 14656 }, { "epoch": 0.7814149384229887, "grad_norm": 1.1568629741668701, "learning_rate": 7.0688e-06, "loss": 0.0312, "step": 14657 }, { "epoch": 0.7814682518526417, "grad_norm": 2.718212842941284, "learning_rate": 7.068600000000001e-06, "loss": 0.0603, "step": 14658 }, { "epoch": 0.7815215652822947, "grad_norm": 2.699848175048828, "learning_rate": 7.068400000000001e-06, "loss": 0.0531, "step": 14659 }, { "epoch": 0.7815748787119475, "grad_norm": 2.0594255924224854, "learning_rate": 7.0682e-06, "loss": 0.0332, "step": 14660 }, { "epoch": 0.7816281921416005, "grad_norm": 0.7725837826728821, "learning_rate": 7.068000000000001e-06, "loss": 0.0144, "step": 14661 }, { "epoch": 0.7816815055712534, "grad_norm": 1.9607453346252441, "learning_rate": 7.0678e-06, "loss": 0.0214, "step": 14662 }, { "epoch": 0.7817348190009064, "grad_norm": 1.5313551425933838, "learning_rate": 7.067600000000001e-06, "loss": 0.0295, "step": 14663 }, { "epoch": 0.7817881324305592, "grad_norm": 2.987656354904175, "learning_rate": 7.067400000000001e-06, "loss": 0.0483, "step": 14664 }, { "epoch": 0.7818414458602122, "grad_norm": 1.6181707382202148, "learning_rate": 7.067200000000001e-06, "loss": 0.025, "step": 14665 }, { "epoch": 0.7818947592898651, "grad_norm": 1.6633650064468384, "learning_rate": 7.067e-06, "loss": 0.0146, "step": 14666 }, { "epoch": 0.781948072719518, "grad_norm": 2.001535654067993, "learning_rate": 7.066800000000001e-06, "loss": 0.0376, "step": 14667 }, { "epoch": 0.782001386149171, "grad_norm": 1.567720890045166, "learning_rate": 7.0666e-06, "loss": 0.0244, "step": 14668 }, { "epoch": 0.7820546995788239, "grad_norm": 1.034568428993225, "learning_rate": 7.066400000000001e-06, "loss": 0.0202, "step": 14669 }, { "epoch": 0.7821080130084769, "grad_norm": 1.1880216598510742, "learning_rate": 7.066200000000001e-06, "loss": 0.0099, "step": 14670 }, { "epoch": 0.7821613264381297, "grad_norm": 2.301783323287964, "learning_rate": 7.066000000000001e-06, "loss": 0.0221, "step": 14671 }, { "epoch": 0.7822146398677827, "grad_norm": 1.6001665592193604, "learning_rate": 7.0658e-06, "loss": 0.0151, "step": 14672 }, { "epoch": 0.7822679532974356, "grad_norm": 1.834517240524292, "learning_rate": 7.065600000000001e-06, "loss": 0.015, "step": 14673 }, { "epoch": 0.7823212667270886, "grad_norm": 3.6621553897857666, "learning_rate": 7.0654e-06, "loss": 0.0302, "step": 14674 }, { "epoch": 0.7823745801567414, "grad_norm": 3.1617326736450195, "learning_rate": 7.0652000000000006e-06, "loss": 0.0654, "step": 14675 }, { "epoch": 0.7824278935863944, "grad_norm": 2.406916856765747, "learning_rate": 7.065000000000001e-06, "loss": 0.0459, "step": 14676 }, { "epoch": 0.7824812070160474, "grad_norm": 1.9102882146835327, "learning_rate": 7.0648000000000005e-06, "loss": 0.0114, "step": 14677 }, { "epoch": 0.7825345204457003, "grad_norm": 2.7140674591064453, "learning_rate": 7.0646e-06, "loss": 0.0464, "step": 14678 }, { "epoch": 0.7825878338753532, "grad_norm": 1.0173976421356201, "learning_rate": 7.0644000000000005e-06, "loss": 0.0079, "step": 14679 }, { "epoch": 0.7826411473050061, "grad_norm": 2.98964524269104, "learning_rate": 7.064200000000001e-06, "loss": 0.0549, "step": 14680 }, { "epoch": 0.7826944607346591, "grad_norm": 2.6223955154418945, "learning_rate": 7.0640000000000005e-06, "loss": 0.0194, "step": 14681 }, { "epoch": 0.782747774164312, "grad_norm": 2.5382680892944336, "learning_rate": 7.063800000000001e-06, "loss": 0.0305, "step": 14682 }, { "epoch": 0.7828010875939649, "grad_norm": 1.5785651206970215, "learning_rate": 7.0636000000000004e-06, "loss": 0.0238, "step": 14683 }, { "epoch": 0.7828544010236178, "grad_norm": 2.6821253299713135, "learning_rate": 7.0634e-06, "loss": 0.021, "step": 14684 }, { "epoch": 0.7829077144532708, "grad_norm": 3.5793657302856445, "learning_rate": 7.0632e-06, "loss": 0.0421, "step": 14685 }, { "epoch": 0.7829610278829237, "grad_norm": 2.4735124111175537, "learning_rate": 7.063000000000001e-06, "loss": 0.0191, "step": 14686 }, { "epoch": 0.7830143413125766, "grad_norm": 3.064396858215332, "learning_rate": 7.0628e-06, "loss": 0.0547, "step": 14687 }, { "epoch": 0.7830676547422296, "grad_norm": 1.6405707597732544, "learning_rate": 7.062600000000001e-06, "loss": 0.0243, "step": 14688 }, { "epoch": 0.7831209681718825, "grad_norm": 3.3324568271636963, "learning_rate": 7.0624e-06, "loss": 0.0685, "step": 14689 }, { "epoch": 0.7831742816015355, "grad_norm": 2.721315622329712, "learning_rate": 7.0622e-06, "loss": 0.048, "step": 14690 }, { "epoch": 0.7832275950311883, "grad_norm": 1.4985257387161255, "learning_rate": 7.062000000000001e-06, "loss": 0.0448, "step": 14691 }, { "epoch": 0.7832809084608413, "grad_norm": 1.6798739433288574, "learning_rate": 7.061800000000001e-06, "loss": 0.0218, "step": 14692 }, { "epoch": 0.7833342218904942, "grad_norm": 3.4172375202178955, "learning_rate": 7.0616e-06, "loss": 0.03, "step": 14693 }, { "epoch": 0.7833875353201472, "grad_norm": 3.0338823795318604, "learning_rate": 7.061400000000001e-06, "loss": 0.0214, "step": 14694 }, { "epoch": 0.7834408487498, "grad_norm": 2.3501229286193848, "learning_rate": 7.0612e-06, "loss": 0.0369, "step": 14695 }, { "epoch": 0.783494162179453, "grad_norm": 0.934238851070404, "learning_rate": 7.061e-06, "loss": 0.0133, "step": 14696 }, { "epoch": 0.783547475609106, "grad_norm": 2.0993497371673584, "learning_rate": 7.060800000000001e-06, "loss": 0.0323, "step": 14697 }, { "epoch": 0.7836007890387588, "grad_norm": 2.626600980758667, "learning_rate": 7.060600000000001e-06, "loss": 0.0379, "step": 14698 }, { "epoch": 0.7836541024684118, "grad_norm": 1.5519546270370483, "learning_rate": 7.0604e-06, "loss": 0.0325, "step": 14699 }, { "epoch": 0.7837074158980647, "grad_norm": 1.6750565767288208, "learning_rate": 7.060200000000001e-06, "loss": 0.0293, "step": 14700 }, { "epoch": 0.7837607293277177, "grad_norm": 2.211341381072998, "learning_rate": 7.06e-06, "loss": 0.0112, "step": 14701 }, { "epoch": 0.7838140427573705, "grad_norm": 2.113147258758545, "learning_rate": 7.059800000000001e-06, "loss": 0.0362, "step": 14702 }, { "epoch": 0.7838673561870235, "grad_norm": 1.746168851852417, "learning_rate": 7.059600000000001e-06, "loss": 0.0364, "step": 14703 }, { "epoch": 0.7839206696166764, "grad_norm": 0.9186519980430603, "learning_rate": 7.0594000000000005e-06, "loss": 0.0158, "step": 14704 }, { "epoch": 0.7839739830463294, "grad_norm": 1.3847930431365967, "learning_rate": 7.0592e-06, "loss": 0.0334, "step": 14705 }, { "epoch": 0.7840272964759823, "grad_norm": 1.5844712257385254, "learning_rate": 7.0590000000000005e-06, "loss": 0.0204, "step": 14706 }, { "epoch": 0.7840806099056352, "grad_norm": 1.5504934787750244, "learning_rate": 7.0588e-06, "loss": 0.0189, "step": 14707 }, { "epoch": 0.7841339233352882, "grad_norm": 1.6618924140930176, "learning_rate": 7.058600000000001e-06, "loss": 0.0412, "step": 14708 }, { "epoch": 0.7841872367649411, "grad_norm": 1.5462652444839478, "learning_rate": 7.058400000000001e-06, "loss": 0.0187, "step": 14709 }, { "epoch": 0.784240550194594, "grad_norm": 2.283082962036133, "learning_rate": 7.0582000000000005e-06, "loss": 0.0329, "step": 14710 }, { "epoch": 0.7842938636242469, "grad_norm": 0.522487461566925, "learning_rate": 7.058e-06, "loss": 0.0072, "step": 14711 }, { "epoch": 0.7843471770538999, "grad_norm": 1.4029196500778198, "learning_rate": 7.0578000000000004e-06, "loss": 0.0145, "step": 14712 }, { "epoch": 0.7844004904835528, "grad_norm": 1.2057318687438965, "learning_rate": 7.057600000000001e-06, "loss": 0.0168, "step": 14713 }, { "epoch": 0.7844538039132057, "grad_norm": 2.1255760192871094, "learning_rate": 7.057400000000001e-06, "loss": 0.0238, "step": 14714 }, { "epoch": 0.7845071173428587, "grad_norm": 1.5036423206329346, "learning_rate": 7.057200000000001e-06, "loss": 0.0138, "step": 14715 }, { "epoch": 0.7845604307725116, "grad_norm": 1.1218088865280151, "learning_rate": 7.057e-06, "loss": 0.0232, "step": 14716 }, { "epoch": 0.7846137442021646, "grad_norm": 1.1020678281784058, "learning_rate": 7.0568e-06, "loss": 0.0211, "step": 14717 }, { "epoch": 0.7846670576318174, "grad_norm": 2.351839303970337, "learning_rate": 7.0566e-06, "loss": 0.0225, "step": 14718 }, { "epoch": 0.7847203710614704, "grad_norm": 1.8835922479629517, "learning_rate": 7.056400000000001e-06, "loss": 0.0192, "step": 14719 }, { "epoch": 0.7847736844911233, "grad_norm": 1.9313210248947144, "learning_rate": 7.056200000000001e-06, "loss": 0.0457, "step": 14720 }, { "epoch": 0.7848269979207763, "grad_norm": 3.5120506286621094, "learning_rate": 7.056000000000001e-06, "loss": 0.063, "step": 14721 }, { "epoch": 0.7848803113504291, "grad_norm": 1.3305026292800903, "learning_rate": 7.0558e-06, "loss": 0.0183, "step": 14722 }, { "epoch": 0.7849336247800821, "grad_norm": 2.243116617202759, "learning_rate": 7.0556e-06, "loss": 0.0266, "step": 14723 }, { "epoch": 0.7849869382097351, "grad_norm": 1.154471755027771, "learning_rate": 7.0554e-06, "loss": 0.007, "step": 14724 }, { "epoch": 0.785040251639388, "grad_norm": 2.4982612133026123, "learning_rate": 7.055200000000001e-06, "loss": 0.0399, "step": 14725 }, { "epoch": 0.7850935650690409, "grad_norm": 3.9658567905426025, "learning_rate": 7.055000000000001e-06, "loss": 0.0524, "step": 14726 }, { "epoch": 0.7851468784986938, "grad_norm": 2.482391595840454, "learning_rate": 7.054800000000001e-06, "loss": 0.0521, "step": 14727 }, { "epoch": 0.7852001919283468, "grad_norm": 1.6517037153244019, "learning_rate": 7.0546e-06, "loss": 0.0218, "step": 14728 }, { "epoch": 0.7852535053579996, "grad_norm": 4.017817497253418, "learning_rate": 7.0544e-06, "loss": 0.0485, "step": 14729 }, { "epoch": 0.7853068187876526, "grad_norm": 5.34055757522583, "learning_rate": 7.054200000000001e-06, "loss": 0.0345, "step": 14730 }, { "epoch": 0.7853601322173055, "grad_norm": 2.0925135612487793, "learning_rate": 7.0540000000000006e-06, "loss": 0.023, "step": 14731 }, { "epoch": 0.7854134456469585, "grad_norm": 2.3106775283813477, "learning_rate": 7.053800000000001e-06, "loss": 0.0256, "step": 14732 }, { "epoch": 0.7854667590766115, "grad_norm": 2.2767534255981445, "learning_rate": 7.0536000000000005e-06, "loss": 0.0196, "step": 14733 }, { "epoch": 0.7855200725062643, "grad_norm": 1.5301228761672974, "learning_rate": 7.0534e-06, "loss": 0.011, "step": 14734 }, { "epoch": 0.7855733859359173, "grad_norm": 1.2181086540222168, "learning_rate": 7.0532e-06, "loss": 0.0221, "step": 14735 }, { "epoch": 0.7856266993655702, "grad_norm": 1.082703709602356, "learning_rate": 7.053000000000001e-06, "loss": 0.0146, "step": 14736 }, { "epoch": 0.7856800127952231, "grad_norm": 2.9178950786590576, "learning_rate": 7.0528000000000005e-06, "loss": 0.0772, "step": 14737 }, { "epoch": 0.785733326224876, "grad_norm": 2.1555681228637695, "learning_rate": 7.052600000000001e-06, "loss": 0.0177, "step": 14738 }, { "epoch": 0.785786639654529, "grad_norm": 3.3171277046203613, "learning_rate": 7.0524000000000004e-06, "loss": 0.0529, "step": 14739 }, { "epoch": 0.7858399530841819, "grad_norm": 2.6292624473571777, "learning_rate": 7.0522e-06, "loss": 0.039, "step": 14740 }, { "epoch": 0.7858932665138348, "grad_norm": 1.5207045078277588, "learning_rate": 7.052000000000001e-06, "loss": 0.0173, "step": 14741 }, { "epoch": 0.7859465799434877, "grad_norm": 1.109461784362793, "learning_rate": 7.051800000000001e-06, "loss": 0.0135, "step": 14742 }, { "epoch": 0.7859998933731407, "grad_norm": 0.688230574131012, "learning_rate": 7.0516e-06, "loss": 0.0053, "step": 14743 }, { "epoch": 0.7860532068027937, "grad_norm": 1.452728271484375, "learning_rate": 7.051400000000001e-06, "loss": 0.0158, "step": 14744 }, { "epoch": 0.7861065202324465, "grad_norm": 3.3677890300750732, "learning_rate": 7.0512e-06, "loss": 0.0243, "step": 14745 }, { "epoch": 0.7861598336620995, "grad_norm": 1.199021339416504, "learning_rate": 7.051e-06, "loss": 0.011, "step": 14746 }, { "epoch": 0.7862131470917524, "grad_norm": 1.554788589477539, "learning_rate": 7.050800000000001e-06, "loss": 0.0345, "step": 14747 }, { "epoch": 0.7862664605214054, "grad_norm": 2.7105355262756348, "learning_rate": 7.050600000000001e-06, "loss": 0.045, "step": 14748 }, { "epoch": 0.7863197739510582, "grad_norm": 1.166670799255371, "learning_rate": 7.0504e-06, "loss": 0.0192, "step": 14749 }, { "epoch": 0.7863730873807112, "grad_norm": 1.795515537261963, "learning_rate": 7.050200000000001e-06, "loss": 0.0302, "step": 14750 }, { "epoch": 0.7864264008103641, "grad_norm": 2.537893056869507, "learning_rate": 7.05e-06, "loss": 0.0252, "step": 14751 }, { "epoch": 0.7864797142400171, "grad_norm": 1.4444293975830078, "learning_rate": 7.049800000000001e-06, "loss": 0.0107, "step": 14752 }, { "epoch": 0.78653302766967, "grad_norm": 1.5031285285949707, "learning_rate": 7.049600000000001e-06, "loss": 0.0359, "step": 14753 }, { "epoch": 0.7865863410993229, "grad_norm": 1.638354778289795, "learning_rate": 7.049400000000001e-06, "loss": 0.0203, "step": 14754 }, { "epoch": 0.7866396545289759, "grad_norm": 3.108515977859497, "learning_rate": 7.0492e-06, "loss": 0.0247, "step": 14755 }, { "epoch": 0.7866929679586288, "grad_norm": 2.211738109588623, "learning_rate": 7.049000000000001e-06, "loss": 0.0316, "step": 14756 }, { "epoch": 0.7867462813882817, "grad_norm": 2.30908465385437, "learning_rate": 7.0488e-06, "loss": 0.0307, "step": 14757 }, { "epoch": 0.7867995948179346, "grad_norm": 2.5673911571502686, "learning_rate": 7.048600000000001e-06, "loss": 0.045, "step": 14758 }, { "epoch": 0.7868529082475876, "grad_norm": 5.688793182373047, "learning_rate": 7.048400000000001e-06, "loss": 0.0449, "step": 14759 }, { "epoch": 0.7869062216772404, "grad_norm": 1.7181965112686157, "learning_rate": 7.0482000000000006e-06, "loss": 0.0222, "step": 14760 }, { "epoch": 0.7869595351068934, "grad_norm": 1.8388715982437134, "learning_rate": 7.048e-06, "loss": 0.0401, "step": 14761 }, { "epoch": 0.7870128485365464, "grad_norm": 0.6907504200935364, "learning_rate": 7.0478000000000005e-06, "loss": 0.0063, "step": 14762 }, { "epoch": 0.7870661619661993, "grad_norm": 1.0830484628677368, "learning_rate": 7.047600000000001e-06, "loss": 0.0277, "step": 14763 }, { "epoch": 0.7871194753958523, "grad_norm": 0.9587051868438721, "learning_rate": 7.0474000000000005e-06, "loss": 0.0084, "step": 14764 }, { "epoch": 0.7871727888255051, "grad_norm": 1.970828652381897, "learning_rate": 7.047200000000001e-06, "loss": 0.031, "step": 14765 }, { "epoch": 0.7872261022551581, "grad_norm": 2.298868417739868, "learning_rate": 7.0470000000000005e-06, "loss": 0.0503, "step": 14766 }, { "epoch": 0.787279415684811, "grad_norm": 2.6452956199645996, "learning_rate": 7.0468e-06, "loss": 0.0205, "step": 14767 }, { "epoch": 0.787332729114464, "grad_norm": 3.83647084236145, "learning_rate": 7.0466000000000004e-06, "loss": 0.0295, "step": 14768 }, { "epoch": 0.7873860425441168, "grad_norm": 1.723413109779358, "learning_rate": 7.046400000000001e-06, "loss": 0.0249, "step": 14769 }, { "epoch": 0.7874393559737698, "grad_norm": 2.3377254009246826, "learning_rate": 7.0462e-06, "loss": 0.0458, "step": 14770 }, { "epoch": 0.7874926694034228, "grad_norm": 4.154457092285156, "learning_rate": 7.046000000000001e-06, "loss": 0.0337, "step": 14771 }, { "epoch": 0.7875459828330756, "grad_norm": 3.862563133239746, "learning_rate": 7.0458e-06, "loss": 0.0472, "step": 14772 }, { "epoch": 0.7875992962627286, "grad_norm": 3.402231454849243, "learning_rate": 7.0456e-06, "loss": 0.0242, "step": 14773 }, { "epoch": 0.7876526096923815, "grad_norm": 1.124842643737793, "learning_rate": 7.045400000000001e-06, "loss": 0.0233, "step": 14774 }, { "epoch": 0.7877059231220345, "grad_norm": 2.7884743213653564, "learning_rate": 7.045200000000001e-06, "loss": 0.041, "step": 14775 }, { "epoch": 0.7877592365516873, "grad_norm": 2.913438558578491, "learning_rate": 7.045e-06, "loss": 0.0509, "step": 14776 }, { "epoch": 0.7878125499813403, "grad_norm": 0.9553922414779663, "learning_rate": 7.044800000000001e-06, "loss": 0.0084, "step": 14777 }, { "epoch": 0.7878658634109932, "grad_norm": 0.9640835523605347, "learning_rate": 7.0446e-06, "loss": 0.0218, "step": 14778 }, { "epoch": 0.7879191768406462, "grad_norm": 1.4081966876983643, "learning_rate": 7.0444e-06, "loss": 0.0143, "step": 14779 }, { "epoch": 0.7879724902702991, "grad_norm": 2.0341973304748535, "learning_rate": 7.044200000000001e-06, "loss": 0.0238, "step": 14780 }, { "epoch": 0.788025803699952, "grad_norm": 3.139540195465088, "learning_rate": 7.044000000000001e-06, "loss": 0.0442, "step": 14781 }, { "epoch": 0.788079117129605, "grad_norm": 1.2113661766052246, "learning_rate": 7.0438e-06, "loss": 0.0209, "step": 14782 }, { "epoch": 0.7881324305592579, "grad_norm": 3.2154741287231445, "learning_rate": 7.043600000000001e-06, "loss": 0.0398, "step": 14783 }, { "epoch": 0.7881857439889108, "grad_norm": 1.1833701133728027, "learning_rate": 7.0434e-06, "loss": 0.0202, "step": 14784 }, { "epoch": 0.7882390574185637, "grad_norm": 1.2103686332702637, "learning_rate": 7.0432e-06, "loss": 0.0082, "step": 14785 }, { "epoch": 0.7882923708482167, "grad_norm": 2.872654438018799, "learning_rate": 7.043000000000001e-06, "loss": 0.0351, "step": 14786 }, { "epoch": 0.7883456842778696, "grad_norm": 1.5286228656768799, "learning_rate": 7.0428000000000006e-06, "loss": 0.012, "step": 14787 }, { "epoch": 0.7883989977075225, "grad_norm": 4.385089874267578, "learning_rate": 7.0426e-06, "loss": 0.0856, "step": 14788 }, { "epoch": 0.7884523111371755, "grad_norm": 2.32234787940979, "learning_rate": 7.0424000000000005e-06, "loss": 0.0321, "step": 14789 }, { "epoch": 0.7885056245668284, "grad_norm": 2.758511543273926, "learning_rate": 7.0422e-06, "loss": 0.025, "step": 14790 }, { "epoch": 0.7885589379964814, "grad_norm": 1.5546282529830933, "learning_rate": 7.042000000000001e-06, "loss": 0.0213, "step": 14791 }, { "epoch": 0.7886122514261342, "grad_norm": 1.9414678812026978, "learning_rate": 7.041800000000001e-06, "loss": 0.0488, "step": 14792 }, { "epoch": 0.7886655648557872, "grad_norm": 2.2865891456604004, "learning_rate": 7.0416000000000005e-06, "loss": 0.0374, "step": 14793 }, { "epoch": 0.7887188782854401, "grad_norm": 1.0240875482559204, "learning_rate": 7.0414e-06, "loss": 0.0139, "step": 14794 }, { "epoch": 0.788772191715093, "grad_norm": 1.3204203844070435, "learning_rate": 7.0412000000000004e-06, "loss": 0.011, "step": 14795 }, { "epoch": 0.7888255051447459, "grad_norm": 0.7919554114341736, "learning_rate": 7.041e-06, "loss": 0.025, "step": 14796 }, { "epoch": 0.7888788185743989, "grad_norm": 0.8586834073066711, "learning_rate": 7.040800000000001e-06, "loss": 0.0056, "step": 14797 }, { "epoch": 0.7889321320040518, "grad_norm": 1.9028996229171753, "learning_rate": 7.040600000000001e-06, "loss": 0.0424, "step": 14798 }, { "epoch": 0.7889854454337047, "grad_norm": 2.375633955001831, "learning_rate": 7.0404e-06, "loss": 0.0399, "step": 14799 }, { "epoch": 0.7890387588633577, "grad_norm": 1.4439769983291626, "learning_rate": 7.0402e-06, "loss": 0.0112, "step": 14800 }, { "epoch": 0.7890920722930106, "grad_norm": 4.334094047546387, "learning_rate": 7.04e-06, "loss": 0.0302, "step": 14801 }, { "epoch": 0.7891453857226636, "grad_norm": 1.7534425258636475, "learning_rate": 7.039800000000001e-06, "loss": 0.0163, "step": 14802 }, { "epoch": 0.7891986991523164, "grad_norm": 0.8815467953681946, "learning_rate": 7.039600000000001e-06, "loss": 0.0086, "step": 14803 }, { "epoch": 0.7892520125819694, "grad_norm": 1.8231779336929321, "learning_rate": 7.039400000000001e-06, "loss": 0.0219, "step": 14804 }, { "epoch": 0.7893053260116223, "grad_norm": 1.6619881391525269, "learning_rate": 7.0392e-06, "loss": 0.0238, "step": 14805 }, { "epoch": 0.7893586394412753, "grad_norm": 3.161076545715332, "learning_rate": 7.039e-06, "loss": 0.0235, "step": 14806 }, { "epoch": 0.7894119528709281, "grad_norm": 1.5245932340621948, "learning_rate": 7.0388e-06, "loss": 0.0202, "step": 14807 }, { "epoch": 0.7894652663005811, "grad_norm": 3.78922963142395, "learning_rate": 7.038600000000001e-06, "loss": 0.0597, "step": 14808 }, { "epoch": 0.7895185797302341, "grad_norm": 2.316255569458008, "learning_rate": 7.038400000000001e-06, "loss": 0.0202, "step": 14809 }, { "epoch": 0.789571893159887, "grad_norm": 1.5580623149871826, "learning_rate": 7.038200000000001e-06, "loss": 0.016, "step": 14810 }, { "epoch": 0.7896252065895399, "grad_norm": 5.334158420562744, "learning_rate": 7.038e-06, "loss": 0.0678, "step": 14811 }, { "epoch": 0.7896785200191928, "grad_norm": 1.6840015649795532, "learning_rate": 7.0378e-06, "loss": 0.0167, "step": 14812 }, { "epoch": 0.7897318334488458, "grad_norm": 4.433502197265625, "learning_rate": 7.037600000000001e-06, "loss": 0.0464, "step": 14813 }, { "epoch": 0.7897851468784987, "grad_norm": 2.548262357711792, "learning_rate": 7.037400000000001e-06, "loss": 0.0247, "step": 14814 }, { "epoch": 0.7898384603081516, "grad_norm": 2.0465385913848877, "learning_rate": 7.037200000000001e-06, "loss": 0.0214, "step": 14815 }, { "epoch": 0.7898917737378045, "grad_norm": 2.6988525390625, "learning_rate": 7.0370000000000006e-06, "loss": 0.0286, "step": 14816 }, { "epoch": 0.7899450871674575, "grad_norm": 0.740817129611969, "learning_rate": 7.0368e-06, "loss": 0.0063, "step": 14817 }, { "epoch": 0.7899984005971105, "grad_norm": 2.9191081523895264, "learning_rate": 7.0366e-06, "loss": 0.0589, "step": 14818 }, { "epoch": 0.7900517140267633, "grad_norm": 4.626425266265869, "learning_rate": 7.036400000000001e-06, "loss": 0.072, "step": 14819 }, { "epoch": 0.7901050274564163, "grad_norm": 1.2720485925674438, "learning_rate": 7.0362000000000005e-06, "loss": 0.0294, "step": 14820 }, { "epoch": 0.7901583408860692, "grad_norm": 1.8580516576766968, "learning_rate": 7.036000000000001e-06, "loss": 0.0454, "step": 14821 }, { "epoch": 0.7902116543157222, "grad_norm": 0.8114223480224609, "learning_rate": 7.0358000000000005e-06, "loss": 0.0058, "step": 14822 }, { "epoch": 0.790264967745375, "grad_norm": 1.5811678171157837, "learning_rate": 7.0356e-06, "loss": 0.0281, "step": 14823 }, { "epoch": 0.790318281175028, "grad_norm": 1.2400110960006714, "learning_rate": 7.035400000000001e-06, "loss": 0.034, "step": 14824 }, { "epoch": 0.7903715946046809, "grad_norm": 2.1118524074554443, "learning_rate": 7.035200000000001e-06, "loss": 0.0249, "step": 14825 }, { "epoch": 0.7904249080343339, "grad_norm": 2.611884355545044, "learning_rate": 7.035e-06, "loss": 0.0349, "step": 14826 }, { "epoch": 0.7904782214639868, "grad_norm": 5.9196858406066895, "learning_rate": 7.034800000000001e-06, "loss": 0.0471, "step": 14827 }, { "epoch": 0.7905315348936397, "grad_norm": 2.7010281085968018, "learning_rate": 7.0346e-06, "loss": 0.0314, "step": 14828 }, { "epoch": 0.7905848483232927, "grad_norm": 3.300067186355591, "learning_rate": 7.0344e-06, "loss": 0.0534, "step": 14829 }, { "epoch": 0.7906381617529455, "grad_norm": 3.0402109622955322, "learning_rate": 7.034200000000001e-06, "loss": 0.0284, "step": 14830 }, { "epoch": 0.7906914751825985, "grad_norm": 2.808166742324829, "learning_rate": 7.034000000000001e-06, "loss": 0.0276, "step": 14831 }, { "epoch": 0.7907447886122514, "grad_norm": 1.7554365396499634, "learning_rate": 7.0338e-06, "loss": 0.0435, "step": 14832 }, { "epoch": 0.7907981020419044, "grad_norm": 2.239769458770752, "learning_rate": 7.033600000000001e-06, "loss": 0.0244, "step": 14833 }, { "epoch": 0.7908514154715572, "grad_norm": 1.5208990573883057, "learning_rate": 7.0334e-06, "loss": 0.0155, "step": 14834 }, { "epoch": 0.7909047289012102, "grad_norm": 4.592194557189941, "learning_rate": 7.0332e-06, "loss": 0.027, "step": 14835 }, { "epoch": 0.7909580423308632, "grad_norm": 0.9310489296913147, "learning_rate": 7.033000000000001e-06, "loss": 0.0278, "step": 14836 }, { "epoch": 0.7910113557605161, "grad_norm": 2.0505778789520264, "learning_rate": 7.032800000000001e-06, "loss": 0.0237, "step": 14837 }, { "epoch": 0.791064669190169, "grad_norm": 0.9440212249755859, "learning_rate": 7.0326e-06, "loss": 0.008, "step": 14838 }, { "epoch": 0.7911179826198219, "grad_norm": 1.35736083984375, "learning_rate": 7.032400000000001e-06, "loss": 0.0125, "step": 14839 }, { "epoch": 0.7911712960494749, "grad_norm": 1.4536525011062622, "learning_rate": 7.0322e-06, "loss": 0.0149, "step": 14840 }, { "epoch": 0.7912246094791278, "grad_norm": 0.7315523624420166, "learning_rate": 7.0320000000000015e-06, "loss": 0.0078, "step": 14841 }, { "epoch": 0.7912779229087807, "grad_norm": 1.824000358581543, "learning_rate": 7.031800000000001e-06, "loss": 0.0324, "step": 14842 }, { "epoch": 0.7913312363384336, "grad_norm": 3.2179982662200928, "learning_rate": 7.0316000000000006e-06, "loss": 0.0361, "step": 14843 }, { "epoch": 0.7913845497680866, "grad_norm": 1.9782427549362183, "learning_rate": 7.0314e-06, "loss": 0.0256, "step": 14844 }, { "epoch": 0.7914378631977395, "grad_norm": 2.5107367038726807, "learning_rate": 7.0312000000000005e-06, "loss": 0.0355, "step": 14845 }, { "epoch": 0.7914911766273924, "grad_norm": 2.1016387939453125, "learning_rate": 7.031e-06, "loss": 0.0118, "step": 14846 }, { "epoch": 0.7915444900570454, "grad_norm": 1.681837797164917, "learning_rate": 7.030800000000001e-06, "loss": 0.0163, "step": 14847 }, { "epoch": 0.7915978034866983, "grad_norm": 1.4223860502243042, "learning_rate": 7.030600000000001e-06, "loss": 0.0118, "step": 14848 }, { "epoch": 0.7916511169163513, "grad_norm": 2.361851215362549, "learning_rate": 7.0304000000000005e-06, "loss": 0.0518, "step": 14849 }, { "epoch": 0.7917044303460041, "grad_norm": 4.1191558837890625, "learning_rate": 7.0302e-06, "loss": 0.0313, "step": 14850 }, { "epoch": 0.7917577437756571, "grad_norm": 4.0293450355529785, "learning_rate": 7.0300000000000005e-06, "loss": 0.0357, "step": 14851 }, { "epoch": 0.79181105720531, "grad_norm": 0.8031758069992065, "learning_rate": 7.029800000000001e-06, "loss": 0.0058, "step": 14852 }, { "epoch": 0.791864370634963, "grad_norm": 1.5703867673873901, "learning_rate": 7.029600000000001e-06, "loss": 0.0293, "step": 14853 }, { "epoch": 0.7919176840646158, "grad_norm": 2.2659127712249756, "learning_rate": 7.029400000000001e-06, "loss": 0.0389, "step": 14854 }, { "epoch": 0.7919709974942688, "grad_norm": 2.1174774169921875, "learning_rate": 7.0292e-06, "loss": 0.0151, "step": 14855 }, { "epoch": 0.7920243109239218, "grad_norm": 4.8437628746032715, "learning_rate": 7.029e-06, "loss": 0.0237, "step": 14856 }, { "epoch": 0.7920776243535746, "grad_norm": 2.0064852237701416, "learning_rate": 7.0288e-06, "loss": 0.0227, "step": 14857 }, { "epoch": 0.7921309377832276, "grad_norm": 1.8195456266403198, "learning_rate": 7.028600000000001e-06, "loss": 0.0265, "step": 14858 }, { "epoch": 0.7921842512128805, "grad_norm": 1.4897783994674683, "learning_rate": 7.028400000000001e-06, "loss": 0.0199, "step": 14859 }, { "epoch": 0.7922375646425335, "grad_norm": 1.469185709953308, "learning_rate": 7.028200000000001e-06, "loss": 0.0208, "step": 14860 }, { "epoch": 0.7922908780721863, "grad_norm": 2.00264573097229, "learning_rate": 7.028e-06, "loss": 0.0263, "step": 14861 }, { "epoch": 0.7923441915018393, "grad_norm": 2.0689947605133057, "learning_rate": 7.0278e-06, "loss": 0.0159, "step": 14862 }, { "epoch": 0.7923975049314922, "grad_norm": 3.2262916564941406, "learning_rate": 7.027600000000001e-06, "loss": 0.0536, "step": 14863 }, { "epoch": 0.7924508183611452, "grad_norm": 2.78729248046875, "learning_rate": 7.027400000000001e-06, "loss": 0.0209, "step": 14864 }, { "epoch": 0.7925041317907981, "grad_norm": 2.5884079933166504, "learning_rate": 7.027200000000001e-06, "loss": 0.0462, "step": 14865 }, { "epoch": 0.792557445220451, "grad_norm": 4.954736709594727, "learning_rate": 7.027000000000001e-06, "loss": 0.0484, "step": 14866 }, { "epoch": 0.792610758650104, "grad_norm": 2.0043680667877197, "learning_rate": 7.0268e-06, "loss": 0.0317, "step": 14867 }, { "epoch": 0.7926640720797569, "grad_norm": 1.8422831296920776, "learning_rate": 7.0266e-06, "loss": 0.0143, "step": 14868 }, { "epoch": 0.7927173855094098, "grad_norm": 3.5079305171966553, "learning_rate": 7.026400000000001e-06, "loss": 0.0267, "step": 14869 }, { "epoch": 0.7927706989390627, "grad_norm": 3.4715888500213623, "learning_rate": 7.026200000000001e-06, "loss": 0.0241, "step": 14870 }, { "epoch": 0.7928240123687157, "grad_norm": 1.5278503894805908, "learning_rate": 7.026000000000001e-06, "loss": 0.0173, "step": 14871 }, { "epoch": 0.7928773257983686, "grad_norm": 2.678168296813965, "learning_rate": 7.0258000000000006e-06, "loss": 0.0124, "step": 14872 }, { "epoch": 0.7929306392280215, "grad_norm": 4.133834362030029, "learning_rate": 7.0256e-06, "loss": 0.0647, "step": 14873 }, { "epoch": 0.7929839526576745, "grad_norm": 2.0772171020507812, "learning_rate": 7.025400000000001e-06, "loss": 0.0301, "step": 14874 }, { "epoch": 0.7930372660873274, "grad_norm": 5.042524337768555, "learning_rate": 7.025200000000001e-06, "loss": 0.0697, "step": 14875 }, { "epoch": 0.7930905795169804, "grad_norm": 3.296802043914795, "learning_rate": 7.0250000000000005e-06, "loss": 0.0384, "step": 14876 }, { "epoch": 0.7931438929466332, "grad_norm": 2.211520195007324, "learning_rate": 7.024800000000001e-06, "loss": 0.0367, "step": 14877 }, { "epoch": 0.7931972063762862, "grad_norm": 3.6223113536834717, "learning_rate": 7.0246000000000005e-06, "loss": 0.0329, "step": 14878 }, { "epoch": 0.7932505198059391, "grad_norm": 2.544233798980713, "learning_rate": 7.0244e-06, "loss": 0.0378, "step": 14879 }, { "epoch": 0.7933038332355921, "grad_norm": 4.853402614593506, "learning_rate": 7.024200000000001e-06, "loss": 0.0434, "step": 14880 }, { "epoch": 0.7933571466652449, "grad_norm": 2.3090386390686035, "learning_rate": 7.024000000000001e-06, "loss": 0.0428, "step": 14881 }, { "epoch": 0.7934104600948979, "grad_norm": 4.342641353607178, "learning_rate": 7.0238e-06, "loss": 0.024, "step": 14882 }, { "epoch": 0.7934637735245509, "grad_norm": 1.780788779258728, "learning_rate": 7.023600000000001e-06, "loss": 0.0291, "step": 14883 }, { "epoch": 0.7935170869542038, "grad_norm": 1.5598064661026, "learning_rate": 7.0234e-06, "loss": 0.0348, "step": 14884 }, { "epoch": 0.7935704003838567, "grad_norm": 2.7612204551696777, "learning_rate": 7.023200000000001e-06, "loss": 0.0193, "step": 14885 }, { "epoch": 0.7936237138135096, "grad_norm": 2.9021923542022705, "learning_rate": 7.023000000000001e-06, "loss": 0.0441, "step": 14886 }, { "epoch": 0.7936770272431626, "grad_norm": 1.5703870058059692, "learning_rate": 7.022800000000001e-06, "loss": 0.0316, "step": 14887 }, { "epoch": 0.7937303406728154, "grad_norm": 1.5707532167434692, "learning_rate": 7.0226e-06, "loss": 0.0183, "step": 14888 }, { "epoch": 0.7937836541024684, "grad_norm": 1.3083131313323975, "learning_rate": 7.022400000000001e-06, "loss": 0.0241, "step": 14889 }, { "epoch": 0.7938369675321213, "grad_norm": 1.415328860282898, "learning_rate": 7.0222e-06, "loss": 0.0146, "step": 14890 }, { "epoch": 0.7938902809617743, "grad_norm": 0.9379932880401611, "learning_rate": 7.022000000000001e-06, "loss": 0.0147, "step": 14891 }, { "epoch": 0.7939435943914273, "grad_norm": 2.1189634799957275, "learning_rate": 7.021800000000001e-06, "loss": 0.0278, "step": 14892 }, { "epoch": 0.7939969078210801, "grad_norm": 1.899145483970642, "learning_rate": 7.021600000000001e-06, "loss": 0.032, "step": 14893 }, { "epoch": 0.7940502212507331, "grad_norm": 3.057710886001587, "learning_rate": 7.0214e-06, "loss": 0.0504, "step": 14894 }, { "epoch": 0.794103534680386, "grad_norm": 1.729310154914856, "learning_rate": 7.021200000000001e-06, "loss": 0.0332, "step": 14895 }, { "epoch": 0.794156848110039, "grad_norm": 2.2213642597198486, "learning_rate": 7.021e-06, "loss": 0.0367, "step": 14896 }, { "epoch": 0.7942101615396918, "grad_norm": 1.4680925607681274, "learning_rate": 7.020800000000001e-06, "loss": 0.0203, "step": 14897 }, { "epoch": 0.7942634749693448, "grad_norm": 0.8021542429924011, "learning_rate": 7.020600000000001e-06, "loss": 0.0096, "step": 14898 }, { "epoch": 0.7943167883989977, "grad_norm": 2.6599178314208984, "learning_rate": 7.020400000000001e-06, "loss": 0.0226, "step": 14899 }, { "epoch": 0.7943701018286506, "grad_norm": 2.411890745162964, "learning_rate": 7.0202e-06, "loss": 0.0205, "step": 14900 }, { "epoch": 0.7944234152583035, "grad_norm": 2.2468504905700684, "learning_rate": 7.0200000000000006e-06, "loss": 0.0247, "step": 14901 }, { "epoch": 0.7944767286879565, "grad_norm": 1.2377527952194214, "learning_rate": 7.019800000000001e-06, "loss": 0.0228, "step": 14902 }, { "epoch": 0.7945300421176095, "grad_norm": 2.325338363647461, "learning_rate": 7.0196000000000005e-06, "loss": 0.0362, "step": 14903 }, { "epoch": 0.7945833555472623, "grad_norm": 3.525879383087158, "learning_rate": 7.019400000000001e-06, "loss": 0.0496, "step": 14904 }, { "epoch": 0.7946366689769153, "grad_norm": 1.1275907754898071, "learning_rate": 7.0192000000000005e-06, "loss": 0.0279, "step": 14905 }, { "epoch": 0.7946899824065682, "grad_norm": 0.8187944889068604, "learning_rate": 7.019e-06, "loss": 0.0149, "step": 14906 }, { "epoch": 0.7947432958362212, "grad_norm": 3.5302252769470215, "learning_rate": 7.0188000000000005e-06, "loss": 0.0224, "step": 14907 }, { "epoch": 0.794796609265874, "grad_norm": 2.3612029552459717, "learning_rate": 7.018600000000001e-06, "loss": 0.0346, "step": 14908 }, { "epoch": 0.794849922695527, "grad_norm": 3.1680784225463867, "learning_rate": 7.0184000000000004e-06, "loss": 0.0426, "step": 14909 }, { "epoch": 0.7949032361251799, "grad_norm": 1.2503584623336792, "learning_rate": 7.018200000000001e-06, "loss": 0.03, "step": 14910 }, { "epoch": 0.7949565495548329, "grad_norm": 1.952958345413208, "learning_rate": 7.018e-06, "loss": 0.0493, "step": 14911 }, { "epoch": 0.7950098629844858, "grad_norm": 2.1338624954223633, "learning_rate": 7.0178e-06, "loss": 0.0213, "step": 14912 }, { "epoch": 0.7950631764141387, "grad_norm": 2.0490705966949463, "learning_rate": 7.017600000000001e-06, "loss": 0.0155, "step": 14913 }, { "epoch": 0.7951164898437917, "grad_norm": 1.6090285778045654, "learning_rate": 7.017400000000001e-06, "loss": 0.0219, "step": 14914 }, { "epoch": 0.7951698032734446, "grad_norm": 2.154884099960327, "learning_rate": 7.0172e-06, "loss": 0.0444, "step": 14915 }, { "epoch": 0.7952231167030975, "grad_norm": 2.5934677124023438, "learning_rate": 7.017000000000001e-06, "loss": 0.0258, "step": 14916 }, { "epoch": 0.7952764301327504, "grad_norm": 2.0625834465026855, "learning_rate": 7.0168e-06, "loss": 0.0199, "step": 14917 }, { "epoch": 0.7953297435624034, "grad_norm": 1.4090352058410645, "learning_rate": 7.0166e-06, "loss": 0.0398, "step": 14918 }, { "epoch": 0.7953830569920562, "grad_norm": 2.6950976848602295, "learning_rate": 7.016400000000001e-06, "loss": 0.0256, "step": 14919 }, { "epoch": 0.7954363704217092, "grad_norm": 1.7603639364242554, "learning_rate": 7.016200000000001e-06, "loss": 0.0134, "step": 14920 }, { "epoch": 0.7954896838513622, "grad_norm": 3.4239296913146973, "learning_rate": 7.016e-06, "loss": 0.0291, "step": 14921 }, { "epoch": 0.7955429972810151, "grad_norm": 1.6684837341308594, "learning_rate": 7.015800000000001e-06, "loss": 0.0389, "step": 14922 }, { "epoch": 0.795596310710668, "grad_norm": 1.4189136028289795, "learning_rate": 7.0156e-06, "loss": 0.0178, "step": 14923 }, { "epoch": 0.7956496241403209, "grad_norm": 2.5766584873199463, "learning_rate": 7.0154000000000015e-06, "loss": 0.0729, "step": 14924 }, { "epoch": 0.7957029375699739, "grad_norm": 3.578449249267578, "learning_rate": 7.015200000000001e-06, "loss": 0.0452, "step": 14925 }, { "epoch": 0.7957562509996268, "grad_norm": 2.4481043815612793, "learning_rate": 7.015000000000001e-06, "loss": 0.0381, "step": 14926 }, { "epoch": 0.7958095644292797, "grad_norm": 1.2791661024093628, "learning_rate": 7.0148e-06, "loss": 0.021, "step": 14927 }, { "epoch": 0.7958628778589326, "grad_norm": 2.3259387016296387, "learning_rate": 7.0146000000000006e-06, "loss": 0.0345, "step": 14928 }, { "epoch": 0.7959161912885856, "grad_norm": 2.879683494567871, "learning_rate": 7.0144e-06, "loss": 0.0248, "step": 14929 }, { "epoch": 0.7959695047182386, "grad_norm": 0.8497312664985657, "learning_rate": 7.014200000000001e-06, "loss": 0.0078, "step": 14930 }, { "epoch": 0.7960228181478914, "grad_norm": 2.5313427448272705, "learning_rate": 7.014000000000001e-06, "loss": 0.052, "step": 14931 }, { "epoch": 0.7960761315775444, "grad_norm": 1.3429269790649414, "learning_rate": 7.0138000000000005e-06, "loss": 0.0187, "step": 14932 }, { "epoch": 0.7961294450071973, "grad_norm": 2.0420260429382324, "learning_rate": 7.0136e-06, "loss": 0.0252, "step": 14933 }, { "epoch": 0.7961827584368503, "grad_norm": 0.8961929678916931, "learning_rate": 7.0134000000000005e-06, "loss": 0.0121, "step": 14934 }, { "epoch": 0.7962360718665031, "grad_norm": 2.2188644409179688, "learning_rate": 7.013200000000001e-06, "loss": 0.035, "step": 14935 }, { "epoch": 0.7962893852961561, "grad_norm": 1.6036748886108398, "learning_rate": 7.013000000000001e-06, "loss": 0.0189, "step": 14936 }, { "epoch": 0.796342698725809, "grad_norm": 2.8553848266601562, "learning_rate": 7.012800000000001e-06, "loss": 0.0543, "step": 14937 }, { "epoch": 0.796396012155462, "grad_norm": 2.607649564743042, "learning_rate": 7.0126e-06, "loss": 0.0553, "step": 14938 }, { "epoch": 0.7964493255851149, "grad_norm": 3.3717479705810547, "learning_rate": 7.0124e-06, "loss": 0.0561, "step": 14939 }, { "epoch": 0.7965026390147678, "grad_norm": 1.889484167098999, "learning_rate": 7.0122e-06, "loss": 0.0115, "step": 14940 }, { "epoch": 0.7965559524444208, "grad_norm": 1.8755557537078857, "learning_rate": 7.012000000000001e-06, "loss": 0.0304, "step": 14941 }, { "epoch": 0.7966092658740737, "grad_norm": 1.7113717794418335, "learning_rate": 7.011800000000001e-06, "loss": 0.0205, "step": 14942 }, { "epoch": 0.7966625793037266, "grad_norm": 2.7252492904663086, "learning_rate": 7.011600000000001e-06, "loss": 0.0165, "step": 14943 }, { "epoch": 0.7967158927333795, "grad_norm": 5.87089729309082, "learning_rate": 7.0114e-06, "loss": 0.0681, "step": 14944 }, { "epoch": 0.7967692061630325, "grad_norm": 4.805250644683838, "learning_rate": 7.0112e-06, "loss": 0.0355, "step": 14945 }, { "epoch": 0.7968225195926854, "grad_norm": 2.1885318756103516, "learning_rate": 7.011e-06, "loss": 0.0266, "step": 14946 }, { "epoch": 0.7968758330223383, "grad_norm": 3.8984806537628174, "learning_rate": 7.010800000000001e-06, "loss": 0.0282, "step": 14947 }, { "epoch": 0.7969291464519913, "grad_norm": 1.5862925052642822, "learning_rate": 7.010600000000001e-06, "loss": 0.0217, "step": 14948 }, { "epoch": 0.7969824598816442, "grad_norm": 1.7062063217163086, "learning_rate": 7.010400000000001e-06, "loss": 0.0361, "step": 14949 }, { "epoch": 0.7970357733112972, "grad_norm": 4.951470851898193, "learning_rate": 7.0102e-06, "loss": 0.0326, "step": 14950 }, { "epoch": 0.79708908674095, "grad_norm": 5.313680171966553, "learning_rate": 7.01e-06, "loss": 0.0349, "step": 14951 }, { "epoch": 0.797142400170603, "grad_norm": 4.506253242492676, "learning_rate": 7.009800000000001e-06, "loss": 0.0452, "step": 14952 }, { "epoch": 0.7971957136002559, "grad_norm": 2.2557456493377686, "learning_rate": 7.009600000000001e-06, "loss": 0.0398, "step": 14953 }, { "epoch": 0.7972490270299089, "grad_norm": 2.1381895542144775, "learning_rate": 7.009400000000001e-06, "loss": 0.0346, "step": 14954 }, { "epoch": 0.7973023404595617, "grad_norm": 1.8930542469024658, "learning_rate": 7.009200000000001e-06, "loss": 0.0386, "step": 14955 }, { "epoch": 0.7973556538892147, "grad_norm": 3.8365063667297363, "learning_rate": 7.009e-06, "loss": 0.0526, "step": 14956 }, { "epoch": 0.7974089673188676, "grad_norm": 2.336737871170044, "learning_rate": 7.0088e-06, "loss": 0.0349, "step": 14957 }, { "epoch": 0.7974622807485205, "grad_norm": 3.4916646480560303, "learning_rate": 7.008600000000001e-06, "loss": 0.033, "step": 14958 }, { "epoch": 0.7975155941781735, "grad_norm": 8.139070510864258, "learning_rate": 7.0084000000000005e-06, "loss": 0.0328, "step": 14959 }, { "epoch": 0.7975689076078264, "grad_norm": 3.598740339279175, "learning_rate": 7.008200000000001e-06, "loss": 0.0718, "step": 14960 }, { "epoch": 0.7976222210374794, "grad_norm": 1.8277397155761719, "learning_rate": 7.0080000000000005e-06, "loss": 0.0138, "step": 14961 }, { "epoch": 0.7976755344671322, "grad_norm": 1.9192200899124146, "learning_rate": 7.0078e-06, "loss": 0.037, "step": 14962 }, { "epoch": 0.7977288478967852, "grad_norm": 2.9593982696533203, "learning_rate": 7.007600000000001e-06, "loss": 0.0428, "step": 14963 }, { "epoch": 0.7977821613264381, "grad_norm": 1.4817078113555908, "learning_rate": 7.007400000000001e-06, "loss": 0.0192, "step": 14964 }, { "epoch": 0.7978354747560911, "grad_norm": 1.8251866102218628, "learning_rate": 7.0072000000000004e-06, "loss": 0.0338, "step": 14965 }, { "epoch": 0.7978887881857439, "grad_norm": 2.972862958908081, "learning_rate": 7.007000000000001e-06, "loss": 0.0475, "step": 14966 }, { "epoch": 0.7979421016153969, "grad_norm": 0.5335277318954468, "learning_rate": 7.0068e-06, "loss": 0.0061, "step": 14967 }, { "epoch": 0.7979954150450499, "grad_norm": 2.019853353500366, "learning_rate": 7.0066e-06, "loss": 0.0553, "step": 14968 }, { "epoch": 0.7980487284747028, "grad_norm": 1.687429666519165, "learning_rate": 7.006400000000001e-06, "loss": 0.0233, "step": 14969 }, { "epoch": 0.7981020419043557, "grad_norm": 3.2585372924804688, "learning_rate": 7.006200000000001e-06, "loss": 0.034, "step": 14970 }, { "epoch": 0.7981553553340086, "grad_norm": 0.7766965627670288, "learning_rate": 7.006e-06, "loss": 0.0092, "step": 14971 }, { "epoch": 0.7982086687636616, "grad_norm": 1.8289765119552612, "learning_rate": 7.005800000000001e-06, "loss": 0.0233, "step": 14972 }, { "epoch": 0.7982619821933145, "grad_norm": 1.3359031677246094, "learning_rate": 7.0056e-06, "loss": 0.0164, "step": 14973 }, { "epoch": 0.7983152956229674, "grad_norm": 3.5668392181396484, "learning_rate": 7.005400000000001e-06, "loss": 0.0248, "step": 14974 }, { "epoch": 0.7983686090526203, "grad_norm": 1.6579452753067017, "learning_rate": 7.005200000000001e-06, "loss": 0.0406, "step": 14975 }, { "epoch": 0.7984219224822733, "grad_norm": 1.6303261518478394, "learning_rate": 7.005000000000001e-06, "loss": 0.023, "step": 14976 }, { "epoch": 0.7984752359119263, "grad_norm": 1.4737430810928345, "learning_rate": 7.0048e-06, "loss": 0.0338, "step": 14977 }, { "epoch": 0.7985285493415791, "grad_norm": 1.8792014122009277, "learning_rate": 7.004600000000001e-06, "loss": 0.0274, "step": 14978 }, { "epoch": 0.7985818627712321, "grad_norm": 1.6711450815200806, "learning_rate": 7.0044e-06, "loss": 0.0225, "step": 14979 }, { "epoch": 0.798635176200885, "grad_norm": 1.2240846157073975, "learning_rate": 7.004200000000001e-06, "loss": 0.0161, "step": 14980 }, { "epoch": 0.798688489630538, "grad_norm": 2.5811285972595215, "learning_rate": 7.004000000000001e-06, "loss": 0.0337, "step": 14981 }, { "epoch": 0.7987418030601908, "grad_norm": 2.4002788066864014, "learning_rate": 7.003800000000001e-06, "loss": 0.0501, "step": 14982 }, { "epoch": 0.7987951164898438, "grad_norm": 2.3926961421966553, "learning_rate": 7.0036e-06, "loss": 0.018, "step": 14983 }, { "epoch": 0.7988484299194967, "grad_norm": 1.6077280044555664, "learning_rate": 7.0034000000000006e-06, "loss": 0.0308, "step": 14984 }, { "epoch": 0.7989017433491497, "grad_norm": 2.0788393020629883, "learning_rate": 7.003200000000001e-06, "loss": 0.0418, "step": 14985 }, { "epoch": 0.7989550567788026, "grad_norm": 1.8003374338150024, "learning_rate": 7.0030000000000005e-06, "loss": 0.0295, "step": 14986 }, { "epoch": 0.7990083702084555, "grad_norm": 1.7194541692733765, "learning_rate": 7.002800000000001e-06, "loss": 0.036, "step": 14987 }, { "epoch": 0.7990616836381085, "grad_norm": 1.7842680215835571, "learning_rate": 7.0026000000000005e-06, "loss": 0.0402, "step": 14988 }, { "epoch": 0.7991149970677613, "grad_norm": 2.44793438911438, "learning_rate": 7.0024e-06, "loss": 0.0576, "step": 14989 }, { "epoch": 0.7991683104974143, "grad_norm": 4.731324672698975, "learning_rate": 7.0022000000000005e-06, "loss": 0.0372, "step": 14990 }, { "epoch": 0.7992216239270672, "grad_norm": 1.3432817459106445, "learning_rate": 7.002000000000001e-06, "loss": 0.0356, "step": 14991 }, { "epoch": 0.7992749373567202, "grad_norm": 1.6437625885009766, "learning_rate": 7.0018000000000005e-06, "loss": 0.0182, "step": 14992 }, { "epoch": 0.799328250786373, "grad_norm": 2.6768295764923096, "learning_rate": 7.001600000000001e-06, "loss": 0.0357, "step": 14993 }, { "epoch": 0.799381564216026, "grad_norm": 2.1574201583862305, "learning_rate": 7.0014000000000004e-06, "loss": 0.0345, "step": 14994 }, { "epoch": 0.799434877645679, "grad_norm": 2.285996675491333, "learning_rate": 7.0012e-06, "loss": 0.0445, "step": 14995 }, { "epoch": 0.7994881910753319, "grad_norm": 2.561509370803833, "learning_rate": 7.001e-06, "loss": 0.0238, "step": 14996 }, { "epoch": 0.7995415045049848, "grad_norm": 1.9348970651626587, "learning_rate": 7.000800000000001e-06, "loss": 0.0157, "step": 14997 }, { "epoch": 0.7995948179346377, "grad_norm": 2.28583025932312, "learning_rate": 7.0006e-06, "loss": 0.0412, "step": 14998 }, { "epoch": 0.7996481313642907, "grad_norm": 2.2342655658721924, "learning_rate": 7.000400000000001e-06, "loss": 0.0315, "step": 14999 }, { "epoch": 0.7997014447939436, "grad_norm": 1.6740798950195312, "learning_rate": 7.0002e-06, "loss": 0.0237, "step": 15000 }, { "epoch": 0.7997547582235965, "grad_norm": 3.50466251373291, "learning_rate": 7e-06, "loss": 0.0513, "step": 15001 }, { "epoch": 0.7998080716532494, "grad_norm": 2.2825865745544434, "learning_rate": 6.999800000000001e-06, "loss": 0.0309, "step": 15002 }, { "epoch": 0.7998613850829024, "grad_norm": 2.2890307903289795, "learning_rate": 6.999600000000001e-06, "loss": 0.0273, "step": 15003 }, { "epoch": 0.7999146985125554, "grad_norm": 3.2593092918395996, "learning_rate": 6.9994e-06, "loss": 0.0426, "step": 15004 }, { "epoch": 0.7999680119422082, "grad_norm": 0.6954740285873413, "learning_rate": 6.999200000000001e-06, "loss": 0.006, "step": 15005 }, { "epoch": 0.8000213253718612, "grad_norm": 2.7771542072296143, "learning_rate": 6.999e-06, "loss": 0.0489, "step": 15006 }, { "epoch": 0.8000746388015141, "grad_norm": 1.2529164552688599, "learning_rate": 6.9988e-06, "loss": 0.0264, "step": 15007 }, { "epoch": 0.8001279522311671, "grad_norm": 1.7123382091522217, "learning_rate": 6.998600000000001e-06, "loss": 0.0286, "step": 15008 }, { "epoch": 0.8001812656608199, "grad_norm": 2.2886900901794434, "learning_rate": 6.998400000000001e-06, "loss": 0.0399, "step": 15009 }, { "epoch": 0.8002345790904729, "grad_norm": 1.4457736015319824, "learning_rate": 6.9982e-06, "loss": 0.0295, "step": 15010 }, { "epoch": 0.8002878925201258, "grad_norm": 2.7459003925323486, "learning_rate": 6.998000000000001e-06, "loss": 0.0333, "step": 15011 }, { "epoch": 0.8003412059497788, "grad_norm": 0.8124279975891113, "learning_rate": 6.9978e-06, "loss": 0.0196, "step": 15012 }, { "epoch": 0.8003945193794316, "grad_norm": 2.312321424484253, "learning_rate": 6.997600000000001e-06, "loss": 0.0277, "step": 15013 }, { "epoch": 0.8004478328090846, "grad_norm": 2.9513497352600098, "learning_rate": 6.997400000000001e-06, "loss": 0.0166, "step": 15014 }, { "epoch": 0.8005011462387376, "grad_norm": 1.9098914861679077, "learning_rate": 6.9972000000000005e-06, "loss": 0.0345, "step": 15015 }, { "epoch": 0.8005544596683905, "grad_norm": 1.1179980039596558, "learning_rate": 6.997e-06, "loss": 0.0116, "step": 15016 }, { "epoch": 0.8006077730980434, "grad_norm": 1.287001371383667, "learning_rate": 6.9968000000000005e-06, "loss": 0.0371, "step": 15017 }, { "epoch": 0.8006610865276963, "grad_norm": 1.6570043563842773, "learning_rate": 6.9966e-06, "loss": 0.0271, "step": 15018 }, { "epoch": 0.8007143999573493, "grad_norm": 1.671449065208435, "learning_rate": 6.996400000000001e-06, "loss": 0.0092, "step": 15019 }, { "epoch": 0.8007677133870021, "grad_norm": 1.9917442798614502, "learning_rate": 6.996200000000001e-06, "loss": 0.0453, "step": 15020 }, { "epoch": 0.8008210268166551, "grad_norm": 1.8576985597610474, "learning_rate": 6.9960000000000004e-06, "loss": 0.0184, "step": 15021 }, { "epoch": 0.800874340246308, "grad_norm": 2.0854694843292236, "learning_rate": 6.9958e-06, "loss": 0.0246, "step": 15022 }, { "epoch": 0.800927653675961, "grad_norm": 1.3825112581253052, "learning_rate": 6.9956e-06, "loss": 0.0177, "step": 15023 }, { "epoch": 0.800980967105614, "grad_norm": 1.6523040533065796, "learning_rate": 6.995400000000001e-06, "loss": 0.0109, "step": 15024 }, { "epoch": 0.8010342805352668, "grad_norm": 2.279566526412964, "learning_rate": 6.995200000000001e-06, "loss": 0.0274, "step": 15025 }, { "epoch": 0.8010875939649198, "grad_norm": 2.1864845752716064, "learning_rate": 6.995000000000001e-06, "loss": 0.0438, "step": 15026 }, { "epoch": 0.8011409073945727, "grad_norm": 1.5951815843582153, "learning_rate": 6.9948e-06, "loss": 0.0288, "step": 15027 }, { "epoch": 0.8011942208242256, "grad_norm": 3.3288028240203857, "learning_rate": 6.9946e-06, "loss": 0.0356, "step": 15028 }, { "epoch": 0.8012475342538785, "grad_norm": 1.4036732912063599, "learning_rate": 6.9944e-06, "loss": 0.0148, "step": 15029 }, { "epoch": 0.8013008476835315, "grad_norm": 1.6576488018035889, "learning_rate": 6.994200000000001e-06, "loss": 0.0224, "step": 15030 }, { "epoch": 0.8013541611131844, "grad_norm": 2.099916696548462, "learning_rate": 6.994000000000001e-06, "loss": 0.0318, "step": 15031 }, { "epoch": 0.8014074745428373, "grad_norm": 2.012875556945801, "learning_rate": 6.993800000000001e-06, "loss": 0.0464, "step": 15032 }, { "epoch": 0.8014607879724903, "grad_norm": 1.8015172481536865, "learning_rate": 6.9936e-06, "loss": 0.0172, "step": 15033 }, { "epoch": 0.8015141014021432, "grad_norm": 2.199198007583618, "learning_rate": 6.9934e-06, "loss": 0.042, "step": 15034 }, { "epoch": 0.8015674148317962, "grad_norm": 1.1914509534835815, "learning_rate": 6.993200000000001e-06, "loss": 0.0215, "step": 15035 }, { "epoch": 0.801620728261449, "grad_norm": 1.9282686710357666, "learning_rate": 6.993000000000001e-06, "loss": 0.0171, "step": 15036 }, { "epoch": 0.801674041691102, "grad_norm": 1.87396240234375, "learning_rate": 6.992800000000001e-06, "loss": 0.0176, "step": 15037 }, { "epoch": 0.8017273551207549, "grad_norm": 1.6344281435012817, "learning_rate": 6.992600000000001e-06, "loss": 0.038, "step": 15038 }, { "epoch": 0.8017806685504079, "grad_norm": 2.501742124557495, "learning_rate": 6.9924e-06, "loss": 0.0396, "step": 15039 }, { "epoch": 0.8018339819800607, "grad_norm": 1.9401930570602417, "learning_rate": 6.9922e-06, "loss": 0.0253, "step": 15040 }, { "epoch": 0.8018872954097137, "grad_norm": 2.817638397216797, "learning_rate": 6.992000000000001e-06, "loss": 0.0276, "step": 15041 }, { "epoch": 0.8019406088393667, "grad_norm": 2.5071170330047607, "learning_rate": 6.9918000000000005e-06, "loss": 0.0187, "step": 15042 }, { "epoch": 0.8019939222690196, "grad_norm": 0.7871004939079285, "learning_rate": 6.991600000000001e-06, "loss": 0.0092, "step": 15043 }, { "epoch": 0.8020472356986725, "grad_norm": 1.5144975185394287, "learning_rate": 6.9914000000000005e-06, "loss": 0.0223, "step": 15044 }, { "epoch": 0.8021005491283254, "grad_norm": 2.0284526348114014, "learning_rate": 6.9912e-06, "loss": 0.0354, "step": 15045 }, { "epoch": 0.8021538625579784, "grad_norm": 2.056415319442749, "learning_rate": 6.991000000000001e-06, "loss": 0.0475, "step": 15046 }, { "epoch": 0.8022071759876312, "grad_norm": 1.738356113433838, "learning_rate": 6.990800000000001e-06, "loss": 0.0307, "step": 15047 }, { "epoch": 0.8022604894172842, "grad_norm": 1.1302590370178223, "learning_rate": 6.9906000000000005e-06, "loss": 0.0278, "step": 15048 }, { "epoch": 0.8023138028469371, "grad_norm": 1.6523419618606567, "learning_rate": 6.990400000000001e-06, "loss": 0.022, "step": 15049 }, { "epoch": 0.8023671162765901, "grad_norm": 3.3351528644561768, "learning_rate": 6.9902000000000004e-06, "loss": 0.0245, "step": 15050 }, { "epoch": 0.802420429706243, "grad_norm": 4.530572414398193, "learning_rate": 6.99e-06, "loss": 0.0421, "step": 15051 }, { "epoch": 0.8024737431358959, "grad_norm": 1.406535029411316, "learning_rate": 6.989800000000001e-06, "loss": 0.0126, "step": 15052 }, { "epoch": 0.8025270565655489, "grad_norm": 1.7247196435928345, "learning_rate": 6.989600000000001e-06, "loss": 0.0231, "step": 15053 }, { "epoch": 0.8025803699952018, "grad_norm": 1.2764885425567627, "learning_rate": 6.9894e-06, "loss": 0.0096, "step": 15054 }, { "epoch": 0.8026336834248547, "grad_norm": 2.6171786785125732, "learning_rate": 6.989200000000001e-06, "loss": 0.0241, "step": 15055 }, { "epoch": 0.8026869968545076, "grad_norm": 2.86734676361084, "learning_rate": 6.989e-06, "loss": 0.0462, "step": 15056 }, { "epoch": 0.8027403102841606, "grad_norm": 3.5848710536956787, "learning_rate": 6.9888e-06, "loss": 0.042, "step": 15057 }, { "epoch": 0.8027936237138135, "grad_norm": 1.2309361696243286, "learning_rate": 6.988600000000001e-06, "loss": 0.0245, "step": 15058 }, { "epoch": 0.8028469371434664, "grad_norm": 3.594247817993164, "learning_rate": 6.988400000000001e-06, "loss": 0.0755, "step": 15059 }, { "epoch": 0.8029002505731194, "grad_norm": 3.0558457374572754, "learning_rate": 6.9882e-06, "loss": 0.0403, "step": 15060 }, { "epoch": 0.8029535640027723, "grad_norm": 1.4083391427993774, "learning_rate": 6.988000000000001e-06, "loss": 0.0123, "step": 15061 }, { "epoch": 0.8030068774324253, "grad_norm": 2.5047085285186768, "learning_rate": 6.9878e-06, "loss": 0.0356, "step": 15062 }, { "epoch": 0.8030601908620781, "grad_norm": 3.5067906379699707, "learning_rate": 6.987600000000001e-06, "loss": 0.0197, "step": 15063 }, { "epoch": 0.8031135042917311, "grad_norm": 2.317800760269165, "learning_rate": 6.987400000000001e-06, "loss": 0.0427, "step": 15064 }, { "epoch": 0.803166817721384, "grad_norm": 1.9449177980422974, "learning_rate": 6.987200000000001e-06, "loss": 0.0214, "step": 15065 }, { "epoch": 0.803220131151037, "grad_norm": 2.5948877334594727, "learning_rate": 6.987e-06, "loss": 0.0542, "step": 15066 }, { "epoch": 0.8032734445806898, "grad_norm": 0.7650384306907654, "learning_rate": 6.986800000000001e-06, "loss": 0.0156, "step": 15067 }, { "epoch": 0.8033267580103428, "grad_norm": 1.564200758934021, "learning_rate": 6.9866e-06, "loss": 0.027, "step": 15068 }, { "epoch": 0.8033800714399957, "grad_norm": 2.320899724960327, "learning_rate": 6.9864000000000006e-06, "loss": 0.0264, "step": 15069 }, { "epoch": 0.8034333848696487, "grad_norm": 1.801896095275879, "learning_rate": 6.986200000000001e-06, "loss": 0.032, "step": 15070 }, { "epoch": 0.8034866982993016, "grad_norm": 0.7444048523902893, "learning_rate": 6.9860000000000005e-06, "loss": 0.0199, "step": 15071 }, { "epoch": 0.8035400117289545, "grad_norm": 6.751319885253906, "learning_rate": 6.9858e-06, "loss": 0.0715, "step": 15072 }, { "epoch": 0.8035933251586075, "grad_norm": 3.076465606689453, "learning_rate": 6.9856000000000005e-06, "loss": 0.0679, "step": 15073 }, { "epoch": 0.8036466385882604, "grad_norm": 1.957560658454895, "learning_rate": 6.985400000000001e-06, "loss": 0.0336, "step": 15074 }, { "epoch": 0.8036999520179133, "grad_norm": 4.276285648345947, "learning_rate": 6.9852000000000005e-06, "loss": 0.0703, "step": 15075 }, { "epoch": 0.8037532654475662, "grad_norm": 2.6383519172668457, "learning_rate": 6.985000000000001e-06, "loss": 0.0535, "step": 15076 }, { "epoch": 0.8038065788772192, "grad_norm": 1.3625156879425049, "learning_rate": 6.9848000000000004e-06, "loss": 0.0245, "step": 15077 }, { "epoch": 0.803859892306872, "grad_norm": 2.055722951889038, "learning_rate": 6.9846e-06, "loss": 0.0235, "step": 15078 }, { "epoch": 0.803913205736525, "grad_norm": 3.787938117980957, "learning_rate": 6.9844e-06, "loss": 0.0277, "step": 15079 }, { "epoch": 0.803966519166178, "grad_norm": 1.8382749557495117, "learning_rate": 6.984200000000001e-06, "loss": 0.0513, "step": 15080 }, { "epoch": 0.8040198325958309, "grad_norm": 1.4404630661010742, "learning_rate": 6.984e-06, "loss": 0.0219, "step": 15081 }, { "epoch": 0.8040731460254839, "grad_norm": 2.1123578548431396, "learning_rate": 6.983800000000001e-06, "loss": 0.0179, "step": 15082 }, { "epoch": 0.8041264594551367, "grad_norm": 2.1764180660247803, "learning_rate": 6.9836e-06, "loss": 0.0445, "step": 15083 }, { "epoch": 0.8041797728847897, "grad_norm": 1.3587946891784668, "learning_rate": 6.9834e-06, "loss": 0.0206, "step": 15084 }, { "epoch": 0.8042330863144426, "grad_norm": 2.792315721511841, "learning_rate": 6.983200000000001e-06, "loss": 0.0412, "step": 15085 }, { "epoch": 0.8042863997440955, "grad_norm": 6.083206653594971, "learning_rate": 6.983000000000001e-06, "loss": 0.0393, "step": 15086 }, { "epoch": 0.8043397131737484, "grad_norm": 1.2007337808609009, "learning_rate": 6.9828e-06, "loss": 0.0182, "step": 15087 }, { "epoch": 0.8043930266034014, "grad_norm": 2.2130565643310547, "learning_rate": 6.982600000000001e-06, "loss": 0.042, "step": 15088 }, { "epoch": 0.8044463400330544, "grad_norm": 1.7225687503814697, "learning_rate": 6.9824e-06, "loss": 0.0157, "step": 15089 }, { "epoch": 0.8044996534627072, "grad_norm": 1.551015019416809, "learning_rate": 6.9822e-06, "loss": 0.041, "step": 15090 }, { "epoch": 0.8045529668923602, "grad_norm": 1.6209949254989624, "learning_rate": 6.982000000000001e-06, "loss": 0.0277, "step": 15091 }, { "epoch": 0.8046062803220131, "grad_norm": 1.6512552499771118, "learning_rate": 6.981800000000001e-06, "loss": 0.0327, "step": 15092 }, { "epoch": 0.8046595937516661, "grad_norm": 1.5358562469482422, "learning_rate": 6.9816e-06, "loss": 0.0175, "step": 15093 }, { "epoch": 0.8047129071813189, "grad_norm": 2.973837375640869, "learning_rate": 6.981400000000001e-06, "loss": 0.0242, "step": 15094 }, { "epoch": 0.8047662206109719, "grad_norm": 1.5967687368392944, "learning_rate": 6.9812e-06, "loss": 0.0225, "step": 15095 }, { "epoch": 0.8048195340406248, "grad_norm": 1.30286705493927, "learning_rate": 6.9810000000000014e-06, "loss": 0.0176, "step": 15096 }, { "epoch": 0.8048728474702778, "grad_norm": 2.804324150085449, "learning_rate": 6.980800000000001e-06, "loss": 0.0434, "step": 15097 }, { "epoch": 0.8049261608999307, "grad_norm": 1.8328312635421753, "learning_rate": 6.9806000000000006e-06, "loss": 0.0317, "step": 15098 }, { "epoch": 0.8049794743295836, "grad_norm": 2.03196120262146, "learning_rate": 6.9804e-06, "loss": 0.0161, "step": 15099 }, { "epoch": 0.8050327877592366, "grad_norm": 1.4692374467849731, "learning_rate": 6.9802000000000005e-06, "loss": 0.0223, "step": 15100 }, { "epoch": 0.8050861011888895, "grad_norm": 2.1037378311157227, "learning_rate": 6.98e-06, "loss": 0.056, "step": 15101 }, { "epoch": 0.8051394146185424, "grad_norm": 3.49324107170105, "learning_rate": 6.979800000000001e-06, "loss": 0.0364, "step": 15102 }, { "epoch": 0.8051927280481953, "grad_norm": 2.563385486602783, "learning_rate": 6.979600000000001e-06, "loss": 0.0183, "step": 15103 }, { "epoch": 0.8052460414778483, "grad_norm": 5.083734512329102, "learning_rate": 6.9794000000000005e-06, "loss": 0.044, "step": 15104 }, { "epoch": 0.8052993549075012, "grad_norm": 0.9538447856903076, "learning_rate": 6.9792e-06, "loss": 0.0111, "step": 15105 }, { "epoch": 0.8053526683371541, "grad_norm": 2.3920512199401855, "learning_rate": 6.9790000000000004e-06, "loss": 0.0229, "step": 15106 }, { "epoch": 0.8054059817668071, "grad_norm": 1.0010024309158325, "learning_rate": 6.9788e-06, "loss": 0.0085, "step": 15107 }, { "epoch": 0.80545929519646, "grad_norm": 2.2368228435516357, "learning_rate": 6.978600000000001e-06, "loss": 0.0428, "step": 15108 }, { "epoch": 0.805512608626113, "grad_norm": 2.910496473312378, "learning_rate": 6.978400000000001e-06, "loss": 0.0533, "step": 15109 }, { "epoch": 0.8055659220557658, "grad_norm": 2.7523441314697266, "learning_rate": 6.9782e-06, "loss": 0.0176, "step": 15110 }, { "epoch": 0.8056192354854188, "grad_norm": 1.0613316297531128, "learning_rate": 6.978e-06, "loss": 0.0215, "step": 15111 }, { "epoch": 0.8056725489150717, "grad_norm": 1.522189974784851, "learning_rate": 6.9778e-06, "loss": 0.0384, "step": 15112 }, { "epoch": 0.8057258623447247, "grad_norm": 2.912724733352661, "learning_rate": 6.977600000000001e-06, "loss": 0.0458, "step": 15113 }, { "epoch": 0.8057791757743775, "grad_norm": 2.408633232116699, "learning_rate": 6.977400000000001e-06, "loss": 0.0311, "step": 15114 }, { "epoch": 0.8058324892040305, "grad_norm": 3.2396583557128906, "learning_rate": 6.977200000000001e-06, "loss": 0.0464, "step": 15115 }, { "epoch": 0.8058858026336835, "grad_norm": 1.2538045644760132, "learning_rate": 6.977e-06, "loss": 0.0144, "step": 15116 }, { "epoch": 0.8059391160633363, "grad_norm": 1.540383219718933, "learning_rate": 6.9768e-06, "loss": 0.0605, "step": 15117 }, { "epoch": 0.8059924294929893, "grad_norm": 2.21098256111145, "learning_rate": 6.9766e-06, "loss": 0.0277, "step": 15118 }, { "epoch": 0.8060457429226422, "grad_norm": 1.633297085762024, "learning_rate": 6.976400000000001e-06, "loss": 0.0286, "step": 15119 }, { "epoch": 0.8060990563522952, "grad_norm": 2.8652265071868896, "learning_rate": 6.976200000000001e-06, "loss": 0.0472, "step": 15120 }, { "epoch": 0.806152369781948, "grad_norm": 1.997700572013855, "learning_rate": 6.976000000000001e-06, "loss": 0.0297, "step": 15121 }, { "epoch": 0.806205683211601, "grad_norm": 1.1524513959884644, "learning_rate": 6.9758e-06, "loss": 0.0164, "step": 15122 }, { "epoch": 0.8062589966412539, "grad_norm": 2.498542308807373, "learning_rate": 6.9756e-06, "loss": 0.0239, "step": 15123 }, { "epoch": 0.8063123100709069, "grad_norm": 1.5243124961853027, "learning_rate": 6.975400000000001e-06, "loss": 0.022, "step": 15124 }, { "epoch": 0.8063656235005597, "grad_norm": 3.295593500137329, "learning_rate": 6.9752000000000006e-06, "loss": 0.0287, "step": 15125 }, { "epoch": 0.8064189369302127, "grad_norm": 1.4868234395980835, "learning_rate": 6.975000000000001e-06, "loss": 0.0145, "step": 15126 }, { "epoch": 0.8064722503598657, "grad_norm": 3.733069896697998, "learning_rate": 6.9748000000000005e-06, "loss": 0.0281, "step": 15127 }, { "epoch": 0.8065255637895186, "grad_norm": 2.437746286392212, "learning_rate": 6.9746e-06, "loss": 0.0267, "step": 15128 }, { "epoch": 0.8065788772191715, "grad_norm": 2.2760612964630127, "learning_rate": 6.9744e-06, "loss": 0.0408, "step": 15129 }, { "epoch": 0.8066321906488244, "grad_norm": 1.6006155014038086, "learning_rate": 6.974200000000001e-06, "loss": 0.0114, "step": 15130 }, { "epoch": 0.8066855040784774, "grad_norm": 1.8008898496627808, "learning_rate": 6.9740000000000005e-06, "loss": 0.0426, "step": 15131 }, { "epoch": 0.8067388175081303, "grad_norm": 2.523231029510498, "learning_rate": 6.973800000000001e-06, "loss": 0.0539, "step": 15132 }, { "epoch": 0.8067921309377832, "grad_norm": 0.6980744004249573, "learning_rate": 6.9736000000000005e-06, "loss": 0.0078, "step": 15133 }, { "epoch": 0.8068454443674361, "grad_norm": 0.8976067900657654, "learning_rate": 6.9734e-06, "loss": 0.0104, "step": 15134 }, { "epoch": 0.8068987577970891, "grad_norm": 1.7234787940979004, "learning_rate": 6.973200000000001e-06, "loss": 0.029, "step": 15135 }, { "epoch": 0.8069520712267421, "grad_norm": 1.8359192609786987, "learning_rate": 6.973000000000001e-06, "loss": 0.0348, "step": 15136 }, { "epoch": 0.8070053846563949, "grad_norm": 2.8996217250823975, "learning_rate": 6.9728e-06, "loss": 0.0331, "step": 15137 }, { "epoch": 0.8070586980860479, "grad_norm": 4.003110408782959, "learning_rate": 6.972600000000001e-06, "loss": 0.045, "step": 15138 }, { "epoch": 0.8071120115157008, "grad_norm": 2.5698721408843994, "learning_rate": 6.9724e-06, "loss": 0.0593, "step": 15139 }, { "epoch": 0.8071653249453538, "grad_norm": 4.3852386474609375, "learning_rate": 6.9722e-06, "loss": 0.0507, "step": 15140 }, { "epoch": 0.8072186383750066, "grad_norm": 1.426523208618164, "learning_rate": 6.972000000000001e-06, "loss": 0.0337, "step": 15141 }, { "epoch": 0.8072719518046596, "grad_norm": 0.9742857813835144, "learning_rate": 6.971800000000001e-06, "loss": 0.0092, "step": 15142 }, { "epoch": 0.8073252652343125, "grad_norm": 1.5053976774215698, "learning_rate": 6.9716e-06, "loss": 0.0261, "step": 15143 }, { "epoch": 0.8073785786639655, "grad_norm": 2.734344720840454, "learning_rate": 6.971400000000001e-06, "loss": 0.017, "step": 15144 }, { "epoch": 0.8074318920936184, "grad_norm": 3.2671334743499756, "learning_rate": 6.9712e-06, "loss": 0.046, "step": 15145 }, { "epoch": 0.8074852055232713, "grad_norm": 1.6938748359680176, "learning_rate": 6.971000000000001e-06, "loss": 0.0237, "step": 15146 }, { "epoch": 0.8075385189529243, "grad_norm": 1.1223527193069458, "learning_rate": 6.970800000000001e-06, "loss": 0.0193, "step": 15147 }, { "epoch": 0.8075918323825771, "grad_norm": 1.659087896347046, "learning_rate": 6.970600000000001e-06, "loss": 0.0251, "step": 15148 }, { "epoch": 0.8076451458122301, "grad_norm": 3.1790974140167236, "learning_rate": 6.9704e-06, "loss": 0.0679, "step": 15149 }, { "epoch": 0.807698459241883, "grad_norm": 2.5846762657165527, "learning_rate": 6.970200000000001e-06, "loss": 0.0269, "step": 15150 }, { "epoch": 0.807751772671536, "grad_norm": 2.058803081512451, "learning_rate": 6.97e-06, "loss": 0.0384, "step": 15151 }, { "epoch": 0.8078050861011888, "grad_norm": 1.9952834844589233, "learning_rate": 6.969800000000001e-06, "loss": 0.0284, "step": 15152 }, { "epoch": 0.8078583995308418, "grad_norm": 0.7764873504638672, "learning_rate": 6.969600000000001e-06, "loss": 0.0087, "step": 15153 }, { "epoch": 0.8079117129604948, "grad_norm": 1.437130331993103, "learning_rate": 6.9694000000000006e-06, "loss": 0.0289, "step": 15154 }, { "epoch": 0.8079650263901477, "grad_norm": 1.927802562713623, "learning_rate": 6.9692e-06, "loss": 0.0321, "step": 15155 }, { "epoch": 0.8080183398198006, "grad_norm": 2.6186740398406982, "learning_rate": 6.9690000000000005e-06, "loss": 0.0512, "step": 15156 }, { "epoch": 0.8080716532494535, "grad_norm": 2.5385239124298096, "learning_rate": 6.968800000000001e-06, "loss": 0.046, "step": 15157 }, { "epoch": 0.8081249666791065, "grad_norm": 2.662039041519165, "learning_rate": 6.9686000000000005e-06, "loss": 0.0284, "step": 15158 }, { "epoch": 0.8081782801087594, "grad_norm": 1.821730136871338, "learning_rate": 6.968400000000001e-06, "loss": 0.0177, "step": 15159 }, { "epoch": 0.8082315935384123, "grad_norm": 2.302212715148926, "learning_rate": 6.9682000000000005e-06, "loss": 0.0377, "step": 15160 }, { "epoch": 0.8082849069680652, "grad_norm": 1.548854112625122, "learning_rate": 6.968e-06, "loss": 0.0388, "step": 15161 }, { "epoch": 0.8083382203977182, "grad_norm": 2.1890785694122314, "learning_rate": 6.9678000000000004e-06, "loss": 0.0269, "step": 15162 }, { "epoch": 0.8083915338273712, "grad_norm": 1.5232768058776855, "learning_rate": 6.967600000000001e-06, "loss": 0.0256, "step": 15163 }, { "epoch": 0.808444847257024, "grad_norm": 5.0110578536987305, "learning_rate": 6.9674e-06, "loss": 0.0378, "step": 15164 }, { "epoch": 0.808498160686677, "grad_norm": 2.2072601318359375, "learning_rate": 6.967200000000001e-06, "loss": 0.018, "step": 15165 }, { "epoch": 0.8085514741163299, "grad_norm": 2.9317610263824463, "learning_rate": 6.967e-06, "loss": 0.0263, "step": 15166 }, { "epoch": 0.8086047875459829, "grad_norm": 1.5353264808654785, "learning_rate": 6.9668e-06, "loss": 0.0135, "step": 15167 }, { "epoch": 0.8086581009756357, "grad_norm": 1.7363166809082031, "learning_rate": 6.9666e-06, "loss": 0.0275, "step": 15168 }, { "epoch": 0.8087114144052887, "grad_norm": 1.8874748945236206, "learning_rate": 6.966400000000001e-06, "loss": 0.032, "step": 15169 }, { "epoch": 0.8087647278349416, "grad_norm": 1.1541376113891602, "learning_rate": 6.9662e-06, "loss": 0.0187, "step": 15170 }, { "epoch": 0.8088180412645946, "grad_norm": 3.579374313354492, "learning_rate": 6.966000000000001e-06, "loss": 0.0329, "step": 15171 }, { "epoch": 0.8088713546942475, "grad_norm": 1.8358981609344482, "learning_rate": 6.9658e-06, "loss": 0.0248, "step": 15172 }, { "epoch": 0.8089246681239004, "grad_norm": 1.6770434379577637, "learning_rate": 6.9656e-06, "loss": 0.0165, "step": 15173 }, { "epoch": 0.8089779815535534, "grad_norm": 1.4809600114822388, "learning_rate": 6.965400000000001e-06, "loss": 0.0145, "step": 15174 }, { "epoch": 0.8090312949832063, "grad_norm": 2.0373356342315674, "learning_rate": 6.965200000000001e-06, "loss": 0.0296, "step": 15175 }, { "epoch": 0.8090846084128592, "grad_norm": 1.1115926504135132, "learning_rate": 6.965e-06, "loss": 0.0221, "step": 15176 }, { "epoch": 0.8091379218425121, "grad_norm": 1.730688214302063, "learning_rate": 6.964800000000001e-06, "loss": 0.021, "step": 15177 }, { "epoch": 0.8091912352721651, "grad_norm": 0.9843679666519165, "learning_rate": 6.9646e-06, "loss": 0.007, "step": 15178 }, { "epoch": 0.809244548701818, "grad_norm": 2.7091891765594482, "learning_rate": 6.9644e-06, "loss": 0.0292, "step": 15179 }, { "epoch": 0.8092978621314709, "grad_norm": 1.7160483598709106, "learning_rate": 6.964200000000001e-06, "loss": 0.0125, "step": 15180 }, { "epoch": 0.8093511755611238, "grad_norm": 2.5975732803344727, "learning_rate": 6.964000000000001e-06, "loss": 0.0411, "step": 15181 }, { "epoch": 0.8094044889907768, "grad_norm": 2.5691606998443604, "learning_rate": 6.9638e-06, "loss": 0.0363, "step": 15182 }, { "epoch": 0.8094578024204298, "grad_norm": 0.876769483089447, "learning_rate": 6.9636000000000005e-06, "loss": 0.0086, "step": 15183 }, { "epoch": 0.8095111158500826, "grad_norm": 1.0588772296905518, "learning_rate": 6.9634e-06, "loss": 0.008, "step": 15184 }, { "epoch": 0.8095644292797356, "grad_norm": 2.768968343734741, "learning_rate": 6.963200000000001e-06, "loss": 0.0453, "step": 15185 }, { "epoch": 0.8096177427093885, "grad_norm": 2.451608180999756, "learning_rate": 6.963000000000001e-06, "loss": 0.0253, "step": 15186 }, { "epoch": 0.8096710561390414, "grad_norm": 5.898502349853516, "learning_rate": 6.9628000000000005e-06, "loss": 0.0355, "step": 15187 }, { "epoch": 0.8097243695686943, "grad_norm": 3.6324572563171387, "learning_rate": 6.9626e-06, "loss": 0.0414, "step": 15188 }, { "epoch": 0.8097776829983473, "grad_norm": 1.814218282699585, "learning_rate": 6.9624000000000005e-06, "loss": 0.0133, "step": 15189 }, { "epoch": 0.8098309964280002, "grad_norm": 2.3226585388183594, "learning_rate": 6.9622e-06, "loss": 0.0154, "step": 15190 }, { "epoch": 0.8098843098576531, "grad_norm": 0.9880154132843018, "learning_rate": 6.962000000000001e-06, "loss": 0.0112, "step": 15191 }, { "epoch": 0.8099376232873061, "grad_norm": 1.7976311445236206, "learning_rate": 6.961800000000001e-06, "loss": 0.0148, "step": 15192 }, { "epoch": 0.809990936716959, "grad_norm": 2.1173477172851562, "learning_rate": 6.9616e-06, "loss": 0.0202, "step": 15193 }, { "epoch": 0.810044250146612, "grad_norm": 2.238304615020752, "learning_rate": 6.9614e-06, "loss": 0.0327, "step": 15194 }, { "epoch": 0.8100975635762648, "grad_norm": 2.864166259765625, "learning_rate": 6.9612e-06, "loss": 0.057, "step": 15195 }, { "epoch": 0.8101508770059178, "grad_norm": 4.014263153076172, "learning_rate": 6.961000000000001e-06, "loss": 0.0211, "step": 15196 }, { "epoch": 0.8102041904355707, "grad_norm": 3.4257919788360596, "learning_rate": 6.960800000000001e-06, "loss": 0.0473, "step": 15197 }, { "epoch": 0.8102575038652237, "grad_norm": 2.586775779724121, "learning_rate": 6.960600000000001e-06, "loss": 0.0415, "step": 15198 }, { "epoch": 0.8103108172948765, "grad_norm": 2.3635458946228027, "learning_rate": 6.9604e-06, "loss": 0.0446, "step": 15199 }, { "epoch": 0.8103641307245295, "grad_norm": 1.2096222639083862, "learning_rate": 6.9602e-06, "loss": 0.0084, "step": 15200 }, { "epoch": 0.8104174441541825, "grad_norm": 0.8681208491325378, "learning_rate": 6.96e-06, "loss": 0.0055, "step": 15201 }, { "epoch": 0.8104707575838354, "grad_norm": 2.5492262840270996, "learning_rate": 6.959800000000001e-06, "loss": 0.0235, "step": 15202 }, { "epoch": 0.8105240710134883, "grad_norm": 3.314077377319336, "learning_rate": 6.959600000000001e-06, "loss": 0.0323, "step": 15203 }, { "epoch": 0.8105773844431412, "grad_norm": 2.9214119911193848, "learning_rate": 6.959400000000001e-06, "loss": 0.0569, "step": 15204 }, { "epoch": 0.8106306978727942, "grad_norm": 3.3717033863067627, "learning_rate": 6.9592e-06, "loss": 0.022, "step": 15205 }, { "epoch": 0.810684011302447, "grad_norm": 4.775719165802002, "learning_rate": 6.959e-06, "loss": 0.0289, "step": 15206 }, { "epoch": 0.8107373247321, "grad_norm": 1.08187997341156, "learning_rate": 6.958800000000001e-06, "loss": 0.0095, "step": 15207 }, { "epoch": 0.8107906381617529, "grad_norm": 1.4485799074172974, "learning_rate": 6.958600000000001e-06, "loss": 0.01, "step": 15208 }, { "epoch": 0.8108439515914059, "grad_norm": 1.9307540655136108, "learning_rate": 6.958400000000001e-06, "loss": 0.0226, "step": 15209 }, { "epoch": 0.8108972650210589, "grad_norm": 3.2102596759796143, "learning_rate": 6.9582000000000006e-06, "loss": 0.019, "step": 15210 }, { "epoch": 0.8109505784507117, "grad_norm": 4.430699825286865, "learning_rate": 6.958e-06, "loss": 0.0322, "step": 15211 }, { "epoch": 0.8110038918803647, "grad_norm": 2.832846164703369, "learning_rate": 6.9578e-06, "loss": 0.045, "step": 15212 }, { "epoch": 0.8110572053100176, "grad_norm": 1.1216707229614258, "learning_rate": 6.957600000000001e-06, "loss": 0.0082, "step": 15213 }, { "epoch": 0.8111105187396705, "grad_norm": 3.1323866844177246, "learning_rate": 6.9574000000000005e-06, "loss": 0.0676, "step": 15214 }, { "epoch": 0.8111638321693234, "grad_norm": 1.8945130109786987, "learning_rate": 6.957200000000001e-06, "loss": 0.021, "step": 15215 }, { "epoch": 0.8112171455989764, "grad_norm": 2.825511932373047, "learning_rate": 6.9570000000000005e-06, "loss": 0.0559, "step": 15216 }, { "epoch": 0.8112704590286293, "grad_norm": 2.973440170288086, "learning_rate": 6.9568e-06, "loss": 0.0363, "step": 15217 }, { "epoch": 0.8113237724582822, "grad_norm": 2.7766969203948975, "learning_rate": 6.9566e-06, "loss": 0.0247, "step": 15218 }, { "epoch": 0.8113770858879352, "grad_norm": 2.0661518573760986, "learning_rate": 6.956400000000001e-06, "loss": 0.0199, "step": 15219 }, { "epoch": 0.8114303993175881, "grad_norm": 3.014244318008423, "learning_rate": 6.9562e-06, "loss": 0.0341, "step": 15220 }, { "epoch": 0.8114837127472411, "grad_norm": 3.0294783115386963, "learning_rate": 6.956000000000001e-06, "loss": 0.0413, "step": 15221 }, { "epoch": 0.8115370261768939, "grad_norm": 2.336941957473755, "learning_rate": 6.9558e-06, "loss": 0.0243, "step": 15222 }, { "epoch": 0.8115903396065469, "grad_norm": 1.990023136138916, "learning_rate": 6.9556e-06, "loss": 0.0342, "step": 15223 }, { "epoch": 0.8116436530361998, "grad_norm": 2.1336984634399414, "learning_rate": 6.955400000000001e-06, "loss": 0.0484, "step": 15224 }, { "epoch": 0.8116969664658528, "grad_norm": 4.457493782043457, "learning_rate": 6.955200000000001e-06, "loss": 0.0588, "step": 15225 }, { "epoch": 0.8117502798955056, "grad_norm": 5.022326946258545, "learning_rate": 6.955e-06, "loss": 0.0256, "step": 15226 }, { "epoch": 0.8118035933251586, "grad_norm": 2.300995349884033, "learning_rate": 6.954800000000001e-06, "loss": 0.0232, "step": 15227 }, { "epoch": 0.8118569067548116, "grad_norm": 3.312223196029663, "learning_rate": 6.9546e-06, "loss": 0.0232, "step": 15228 }, { "epoch": 0.8119102201844645, "grad_norm": 1.2921051979064941, "learning_rate": 6.9544e-06, "loss": 0.0159, "step": 15229 }, { "epoch": 0.8119635336141174, "grad_norm": 0.8754603862762451, "learning_rate": 6.954200000000001e-06, "loss": 0.0078, "step": 15230 }, { "epoch": 0.8120168470437703, "grad_norm": 3.939842939376831, "learning_rate": 6.954000000000001e-06, "loss": 0.0599, "step": 15231 }, { "epoch": 0.8120701604734233, "grad_norm": 2.3815042972564697, "learning_rate": 6.9538e-06, "loss": 0.0158, "step": 15232 }, { "epoch": 0.8121234739030762, "grad_norm": 3.2010700702667236, "learning_rate": 6.953600000000001e-06, "loss": 0.0437, "step": 15233 }, { "epoch": 0.8121767873327291, "grad_norm": 1.5508582592010498, "learning_rate": 6.9534e-06, "loss": 0.03, "step": 15234 }, { "epoch": 0.812230100762382, "grad_norm": 1.8231232166290283, "learning_rate": 6.9532000000000015e-06, "loss": 0.0181, "step": 15235 }, { "epoch": 0.812283414192035, "grad_norm": 2.8293802738189697, "learning_rate": 6.953000000000001e-06, "loss": 0.0417, "step": 15236 }, { "epoch": 0.8123367276216878, "grad_norm": 1.6469515562057495, "learning_rate": 6.952800000000001e-06, "loss": 0.0346, "step": 15237 }, { "epoch": 0.8123900410513408, "grad_norm": 3.344672441482544, "learning_rate": 6.9526e-06, "loss": 0.0376, "step": 15238 }, { "epoch": 0.8124433544809938, "grad_norm": 1.9961062669754028, "learning_rate": 6.9524000000000006e-06, "loss": 0.029, "step": 15239 }, { "epoch": 0.8124966679106467, "grad_norm": 1.9173057079315186, "learning_rate": 6.9522e-06, "loss": 0.0238, "step": 15240 }, { "epoch": 0.8125499813402997, "grad_norm": 1.3408564329147339, "learning_rate": 6.952000000000001e-06, "loss": 0.029, "step": 15241 }, { "epoch": 0.8126032947699525, "grad_norm": 2.9430911540985107, "learning_rate": 6.951800000000001e-06, "loss": 0.0467, "step": 15242 }, { "epoch": 0.8126566081996055, "grad_norm": 4.225436687469482, "learning_rate": 6.9516000000000005e-06, "loss": 0.0619, "step": 15243 }, { "epoch": 0.8127099216292584, "grad_norm": 2.297661066055298, "learning_rate": 6.9514e-06, "loss": 0.0504, "step": 15244 }, { "epoch": 0.8127632350589113, "grad_norm": 1.5160484313964844, "learning_rate": 6.9512000000000005e-06, "loss": 0.0192, "step": 15245 }, { "epoch": 0.8128165484885642, "grad_norm": 1.2622519731521606, "learning_rate": 6.951000000000001e-06, "loss": 0.026, "step": 15246 }, { "epoch": 0.8128698619182172, "grad_norm": 2.3193106651306152, "learning_rate": 6.950800000000001e-06, "loss": 0.0375, "step": 15247 }, { "epoch": 0.8129231753478702, "grad_norm": 1.2386982440948486, "learning_rate": 6.950600000000001e-06, "loss": 0.014, "step": 15248 }, { "epoch": 0.812976488777523, "grad_norm": 0.5018779039382935, "learning_rate": 6.9504e-06, "loss": 0.0057, "step": 15249 }, { "epoch": 0.813029802207176, "grad_norm": 1.4388856887817383, "learning_rate": 6.9502e-06, "loss": 0.0173, "step": 15250 }, { "epoch": 0.8130831156368289, "grad_norm": 2.9375815391540527, "learning_rate": 6.95e-06, "loss": 0.0864, "step": 15251 }, { "epoch": 0.8131364290664819, "grad_norm": 0.9840106964111328, "learning_rate": 6.949800000000001e-06, "loss": 0.0079, "step": 15252 }, { "epoch": 0.8131897424961347, "grad_norm": 3.0423853397369385, "learning_rate": 6.949600000000001e-06, "loss": 0.0314, "step": 15253 }, { "epoch": 0.8132430559257877, "grad_norm": 2.2940993309020996, "learning_rate": 6.949400000000001e-06, "loss": 0.0425, "step": 15254 }, { "epoch": 0.8132963693554406, "grad_norm": 1.0646824836730957, "learning_rate": 6.9492e-06, "loss": 0.0217, "step": 15255 }, { "epoch": 0.8133496827850936, "grad_norm": 2.158695936203003, "learning_rate": 6.949e-06, "loss": 0.0329, "step": 15256 }, { "epoch": 0.8134029962147465, "grad_norm": 1.2202656269073486, "learning_rate": 6.948800000000001e-06, "loss": 0.0165, "step": 15257 }, { "epoch": 0.8134563096443994, "grad_norm": 1.2701637744903564, "learning_rate": 6.948600000000001e-06, "loss": 0.0187, "step": 15258 }, { "epoch": 0.8135096230740524, "grad_norm": 1.8612360954284668, "learning_rate": 6.948400000000001e-06, "loss": 0.0383, "step": 15259 }, { "epoch": 0.8135629365037053, "grad_norm": 2.2066385746002197, "learning_rate": 6.948200000000001e-06, "loss": 0.0489, "step": 15260 }, { "epoch": 0.8136162499333582, "grad_norm": 4.750401973724365, "learning_rate": 6.948e-06, "loss": 0.0435, "step": 15261 }, { "epoch": 0.8136695633630111, "grad_norm": 2.0892882347106934, "learning_rate": 6.9478e-06, "loss": 0.053, "step": 15262 }, { "epoch": 0.8137228767926641, "grad_norm": 2.787179946899414, "learning_rate": 6.947600000000001e-06, "loss": 0.0396, "step": 15263 }, { "epoch": 0.813776190222317, "grad_norm": 2.140474557876587, "learning_rate": 6.947400000000001e-06, "loss": 0.0357, "step": 15264 }, { "epoch": 0.8138295036519699, "grad_norm": 1.497261881828308, "learning_rate": 6.947200000000001e-06, "loss": 0.0418, "step": 15265 }, { "epoch": 0.8138828170816229, "grad_norm": 3.155200481414795, "learning_rate": 6.9470000000000006e-06, "loss": 0.051, "step": 15266 }, { "epoch": 0.8139361305112758, "grad_norm": 1.428971290588379, "learning_rate": 6.9468e-06, "loss": 0.0122, "step": 15267 }, { "epoch": 0.8139894439409288, "grad_norm": 1.7195855379104614, "learning_rate": 6.946600000000001e-06, "loss": 0.0619, "step": 15268 }, { "epoch": 0.8140427573705816, "grad_norm": 1.5989339351654053, "learning_rate": 6.946400000000001e-06, "loss": 0.0234, "step": 15269 }, { "epoch": 0.8140960708002346, "grad_norm": 1.8230482339859009, "learning_rate": 6.9462000000000005e-06, "loss": 0.0471, "step": 15270 }, { "epoch": 0.8141493842298875, "grad_norm": 1.1920926570892334, "learning_rate": 6.946000000000001e-06, "loss": 0.0273, "step": 15271 }, { "epoch": 0.8142026976595405, "grad_norm": 1.7915700674057007, "learning_rate": 6.9458000000000005e-06, "loss": 0.0261, "step": 15272 }, { "epoch": 0.8142560110891933, "grad_norm": 3.2283236980438232, "learning_rate": 6.9456e-06, "loss": 0.0559, "step": 15273 }, { "epoch": 0.8143093245188463, "grad_norm": 2.2256975173950195, "learning_rate": 6.945400000000001e-06, "loss": 0.0212, "step": 15274 }, { "epoch": 0.8143626379484993, "grad_norm": 2.1992008686065674, "learning_rate": 6.945200000000001e-06, "loss": 0.0359, "step": 15275 }, { "epoch": 0.8144159513781521, "grad_norm": 1.7434091567993164, "learning_rate": 6.945e-06, "loss": 0.0273, "step": 15276 }, { "epoch": 0.8144692648078051, "grad_norm": 1.3551441431045532, "learning_rate": 6.944800000000001e-06, "loss": 0.0144, "step": 15277 }, { "epoch": 0.814522578237458, "grad_norm": 1.429573893547058, "learning_rate": 6.9446e-06, "loss": 0.0219, "step": 15278 }, { "epoch": 0.814575891667111, "grad_norm": 3.1761820316314697, "learning_rate": 6.9444e-06, "loss": 0.0494, "step": 15279 }, { "epoch": 0.8146292050967638, "grad_norm": 1.4293088912963867, "learning_rate": 6.944200000000001e-06, "loss": 0.0474, "step": 15280 }, { "epoch": 0.8146825185264168, "grad_norm": 1.6692308187484741, "learning_rate": 6.944000000000001e-06, "loss": 0.0434, "step": 15281 }, { "epoch": 0.8147358319560697, "grad_norm": 1.9040610790252686, "learning_rate": 6.9438e-06, "loss": 0.0322, "step": 15282 }, { "epoch": 0.8147891453857227, "grad_norm": 0.7984123826026917, "learning_rate": 6.943600000000001e-06, "loss": 0.0119, "step": 15283 }, { "epoch": 0.8148424588153756, "grad_norm": 1.5643755197525024, "learning_rate": 6.9434e-06, "loss": 0.0168, "step": 15284 }, { "epoch": 0.8148957722450285, "grad_norm": 1.4941821098327637, "learning_rate": 6.943200000000001e-06, "loss": 0.0216, "step": 15285 }, { "epoch": 0.8149490856746815, "grad_norm": 1.1525640487670898, "learning_rate": 6.943000000000001e-06, "loss": 0.0104, "step": 15286 }, { "epoch": 0.8150023991043344, "grad_norm": 2.516334295272827, "learning_rate": 6.942800000000001e-06, "loss": 0.0422, "step": 15287 }, { "epoch": 0.8150557125339873, "grad_norm": 2.6653695106506348, "learning_rate": 6.9426e-06, "loss": 0.0519, "step": 15288 }, { "epoch": 0.8151090259636402, "grad_norm": 1.3727401494979858, "learning_rate": 6.942400000000001e-06, "loss": 0.0123, "step": 15289 }, { "epoch": 0.8151623393932932, "grad_norm": 1.2620067596435547, "learning_rate": 6.9422e-06, "loss": 0.0161, "step": 15290 }, { "epoch": 0.8152156528229461, "grad_norm": 2.5293378829956055, "learning_rate": 6.942000000000001e-06, "loss": 0.0211, "step": 15291 }, { "epoch": 0.815268966252599, "grad_norm": 1.3714312314987183, "learning_rate": 6.941800000000001e-06, "loss": 0.0239, "step": 15292 }, { "epoch": 0.8153222796822519, "grad_norm": 1.7848600149154663, "learning_rate": 6.941600000000001e-06, "loss": 0.0319, "step": 15293 }, { "epoch": 0.8153755931119049, "grad_norm": 1.1479549407958984, "learning_rate": 6.9414e-06, "loss": 0.0163, "step": 15294 }, { "epoch": 0.8154289065415579, "grad_norm": 1.927011251449585, "learning_rate": 6.9412000000000006e-06, "loss": 0.0173, "step": 15295 }, { "epoch": 0.8154822199712107, "grad_norm": 1.3928848505020142, "learning_rate": 6.941000000000001e-06, "loss": 0.0101, "step": 15296 }, { "epoch": 0.8155355334008637, "grad_norm": 2.0229873657226562, "learning_rate": 6.9408000000000005e-06, "loss": 0.0365, "step": 15297 }, { "epoch": 0.8155888468305166, "grad_norm": 2.347010374069214, "learning_rate": 6.940600000000001e-06, "loss": 0.037, "step": 15298 }, { "epoch": 0.8156421602601696, "grad_norm": 1.9088258743286133, "learning_rate": 6.9404000000000005e-06, "loss": 0.0264, "step": 15299 }, { "epoch": 0.8156954736898224, "grad_norm": 2.14619779586792, "learning_rate": 6.9402e-06, "loss": 0.0311, "step": 15300 }, { "epoch": 0.8157487871194754, "grad_norm": 2.117421865463257, "learning_rate": 6.9400000000000005e-06, "loss": 0.0248, "step": 15301 }, { "epoch": 0.8158021005491283, "grad_norm": 3.217233657836914, "learning_rate": 6.939800000000001e-06, "loss": 0.036, "step": 15302 }, { "epoch": 0.8158554139787813, "grad_norm": 2.3531646728515625, "learning_rate": 6.9396000000000004e-06, "loss": 0.0245, "step": 15303 }, { "epoch": 0.8159087274084342, "grad_norm": 2.7866692543029785, "learning_rate": 6.939400000000001e-06, "loss": 0.0389, "step": 15304 }, { "epoch": 0.8159620408380871, "grad_norm": 3.3623485565185547, "learning_rate": 6.9392e-06, "loss": 0.0559, "step": 15305 }, { "epoch": 0.8160153542677401, "grad_norm": 0.6424383521080017, "learning_rate": 6.939e-06, "loss": 0.0073, "step": 15306 }, { "epoch": 0.816068667697393, "grad_norm": 2.019008159637451, "learning_rate": 6.938800000000001e-06, "loss": 0.0341, "step": 15307 }, { "epoch": 0.8161219811270459, "grad_norm": 2.7105259895324707, "learning_rate": 6.938600000000001e-06, "loss": 0.0556, "step": 15308 }, { "epoch": 0.8161752945566988, "grad_norm": 4.79880952835083, "learning_rate": 6.9384e-06, "loss": 0.0668, "step": 15309 }, { "epoch": 0.8162286079863518, "grad_norm": 3.33949875831604, "learning_rate": 6.938200000000001e-06, "loss": 0.0329, "step": 15310 }, { "epoch": 0.8162819214160046, "grad_norm": 2.6200430393218994, "learning_rate": 6.938e-06, "loss": 0.0229, "step": 15311 }, { "epoch": 0.8163352348456576, "grad_norm": 1.4162417650222778, "learning_rate": 6.9378e-06, "loss": 0.022, "step": 15312 }, { "epoch": 0.8163885482753106, "grad_norm": 2.015528917312622, "learning_rate": 6.937600000000001e-06, "loss": 0.0359, "step": 15313 }, { "epoch": 0.8164418617049635, "grad_norm": 2.422719955444336, "learning_rate": 6.937400000000001e-06, "loss": 0.0243, "step": 15314 }, { "epoch": 0.8164951751346164, "grad_norm": 1.7861782312393188, "learning_rate": 6.9372e-06, "loss": 0.0231, "step": 15315 }, { "epoch": 0.8165484885642693, "grad_norm": 3.6712405681610107, "learning_rate": 6.937000000000001e-06, "loss": 0.0345, "step": 15316 }, { "epoch": 0.8166018019939223, "grad_norm": 1.8865171670913696, "learning_rate": 6.9368e-06, "loss": 0.0174, "step": 15317 }, { "epoch": 0.8166551154235752, "grad_norm": 2.138141632080078, "learning_rate": 6.9366000000000015e-06, "loss": 0.0332, "step": 15318 }, { "epoch": 0.8167084288532281, "grad_norm": 1.7220505475997925, "learning_rate": 6.936400000000001e-06, "loss": 0.0202, "step": 15319 }, { "epoch": 0.816761742282881, "grad_norm": 2.360466718673706, "learning_rate": 6.936200000000001e-06, "loss": 0.0496, "step": 15320 }, { "epoch": 0.816815055712534, "grad_norm": 1.6588232517242432, "learning_rate": 6.936e-06, "loss": 0.0151, "step": 15321 }, { "epoch": 0.816868369142187, "grad_norm": 1.9196850061416626, "learning_rate": 6.935800000000001e-06, "loss": 0.0262, "step": 15322 }, { "epoch": 0.8169216825718398, "grad_norm": 3.079320192337036, "learning_rate": 6.9356e-06, "loss": 0.0259, "step": 15323 }, { "epoch": 0.8169749960014928, "grad_norm": 1.738587498664856, "learning_rate": 6.935400000000001e-06, "loss": 0.0118, "step": 15324 }, { "epoch": 0.8170283094311457, "grad_norm": 1.6988333463668823, "learning_rate": 6.935200000000001e-06, "loss": 0.0317, "step": 15325 }, { "epoch": 0.8170816228607987, "grad_norm": 1.3964899778366089, "learning_rate": 6.9350000000000005e-06, "loss": 0.0123, "step": 15326 }, { "epoch": 0.8171349362904515, "grad_norm": 2.4494128227233887, "learning_rate": 6.9348e-06, "loss": 0.0328, "step": 15327 }, { "epoch": 0.8171882497201045, "grad_norm": 0.8097093105316162, "learning_rate": 6.9346000000000005e-06, "loss": 0.006, "step": 15328 }, { "epoch": 0.8172415631497574, "grad_norm": 1.7157714366912842, "learning_rate": 6.9344e-06, "loss": 0.0267, "step": 15329 }, { "epoch": 0.8172948765794104, "grad_norm": 3.2004854679107666, "learning_rate": 6.934200000000001e-06, "loss": 0.0545, "step": 15330 }, { "epoch": 0.8173481900090633, "grad_norm": 1.6980994939804077, "learning_rate": 6.934000000000001e-06, "loss": 0.0278, "step": 15331 }, { "epoch": 0.8174015034387162, "grad_norm": 2.4212253093719482, "learning_rate": 6.9338000000000004e-06, "loss": 0.0411, "step": 15332 }, { "epoch": 0.8174548168683692, "grad_norm": 4.808655738830566, "learning_rate": 6.9336e-06, "loss": 0.0488, "step": 15333 }, { "epoch": 0.817508130298022, "grad_norm": 1.9109526872634888, "learning_rate": 6.9334e-06, "loss": 0.0265, "step": 15334 }, { "epoch": 0.817561443727675, "grad_norm": 0.8423230648040771, "learning_rate": 6.933200000000001e-06, "loss": 0.0064, "step": 15335 }, { "epoch": 0.8176147571573279, "grad_norm": 1.7053388357162476, "learning_rate": 6.933000000000001e-06, "loss": 0.0294, "step": 15336 }, { "epoch": 0.8176680705869809, "grad_norm": 2.758671283721924, "learning_rate": 6.932800000000001e-06, "loss": 0.0428, "step": 15337 }, { "epoch": 0.8177213840166337, "grad_norm": 3.7803969383239746, "learning_rate": 6.9326e-06, "loss": 0.0284, "step": 15338 }, { "epoch": 0.8177746974462867, "grad_norm": 2.5188345909118652, "learning_rate": 6.9324e-06, "loss": 0.025, "step": 15339 }, { "epoch": 0.8178280108759397, "grad_norm": 2.226041793823242, "learning_rate": 6.9322e-06, "loss": 0.0435, "step": 15340 }, { "epoch": 0.8178813243055926, "grad_norm": 3.289062261581421, "learning_rate": 6.932000000000001e-06, "loss": 0.0475, "step": 15341 }, { "epoch": 0.8179346377352456, "grad_norm": 1.407259225845337, "learning_rate": 6.931800000000001e-06, "loss": 0.0331, "step": 15342 }, { "epoch": 0.8179879511648984, "grad_norm": 3.1344242095947266, "learning_rate": 6.931600000000001e-06, "loss": 0.052, "step": 15343 }, { "epoch": 0.8180412645945514, "grad_norm": 1.6776643991470337, "learning_rate": 6.9314e-06, "loss": 0.0212, "step": 15344 }, { "epoch": 0.8180945780242043, "grad_norm": 2.873169183731079, "learning_rate": 6.9312e-06, "loss": 0.0503, "step": 15345 }, { "epoch": 0.8181478914538572, "grad_norm": 1.219706654548645, "learning_rate": 6.931000000000001e-06, "loss": 0.0064, "step": 15346 }, { "epoch": 0.8182012048835101, "grad_norm": 2.0997543334960938, "learning_rate": 6.930800000000001e-06, "loss": 0.03, "step": 15347 }, { "epoch": 0.8182545183131631, "grad_norm": 2.238780975341797, "learning_rate": 6.930600000000001e-06, "loss": 0.0315, "step": 15348 }, { "epoch": 0.818307831742816, "grad_norm": 2.0797159671783447, "learning_rate": 6.930400000000001e-06, "loss": 0.0429, "step": 15349 }, { "epoch": 0.8183611451724689, "grad_norm": 3.173774242401123, "learning_rate": 6.9302e-06, "loss": 0.0673, "step": 15350 }, { "epoch": 0.8184144586021219, "grad_norm": 2.0879480838775635, "learning_rate": 6.93e-06, "loss": 0.0484, "step": 15351 }, { "epoch": 0.8184677720317748, "grad_norm": 1.5032174587249756, "learning_rate": 6.929800000000001e-06, "loss": 0.0165, "step": 15352 }, { "epoch": 0.8185210854614278, "grad_norm": 1.8891271352767944, "learning_rate": 6.9296000000000005e-06, "loss": 0.0405, "step": 15353 }, { "epoch": 0.8185743988910806, "grad_norm": 1.4715203046798706, "learning_rate": 6.929400000000001e-06, "loss": 0.0106, "step": 15354 }, { "epoch": 0.8186277123207336, "grad_norm": 2.225750207901001, "learning_rate": 6.9292000000000005e-06, "loss": 0.0394, "step": 15355 }, { "epoch": 0.8186810257503865, "grad_norm": 3.6712067127227783, "learning_rate": 6.929e-06, "loss": 0.0363, "step": 15356 }, { "epoch": 0.8187343391800395, "grad_norm": 1.5500439405441284, "learning_rate": 6.928800000000001e-06, "loss": 0.0353, "step": 15357 }, { "epoch": 0.8187876526096923, "grad_norm": 2.5872926712036133, "learning_rate": 6.928600000000001e-06, "loss": 0.0292, "step": 15358 }, { "epoch": 0.8188409660393453, "grad_norm": 1.038880467414856, "learning_rate": 6.9284000000000004e-06, "loss": 0.0129, "step": 15359 }, { "epoch": 0.8188942794689983, "grad_norm": 1.8400533199310303, "learning_rate": 6.928200000000001e-06, "loss": 0.0283, "step": 15360 }, { "epoch": 0.8189475928986512, "grad_norm": 2.377143144607544, "learning_rate": 6.928e-06, "loss": 0.0367, "step": 15361 }, { "epoch": 0.8190009063283041, "grad_norm": 1.9842995405197144, "learning_rate": 6.9278e-06, "loss": 0.0633, "step": 15362 }, { "epoch": 0.819054219757957, "grad_norm": 3.616950750350952, "learning_rate": 6.927600000000001e-06, "loss": 0.0317, "step": 15363 }, { "epoch": 0.81910753318761, "grad_norm": 2.3399369716644287, "learning_rate": 6.927400000000001e-06, "loss": 0.0307, "step": 15364 }, { "epoch": 0.8191608466172628, "grad_norm": 2.6672701835632324, "learning_rate": 6.9272e-06, "loss": 0.0659, "step": 15365 }, { "epoch": 0.8192141600469158, "grad_norm": 1.6233075857162476, "learning_rate": 6.927000000000001e-06, "loss": 0.0123, "step": 15366 }, { "epoch": 0.8192674734765687, "grad_norm": 1.8558943271636963, "learning_rate": 6.9268e-06, "loss": 0.0415, "step": 15367 }, { "epoch": 0.8193207869062217, "grad_norm": 3.639578104019165, "learning_rate": 6.926600000000001e-06, "loss": 0.0781, "step": 15368 }, { "epoch": 0.8193741003358747, "grad_norm": 2.0786755084991455, "learning_rate": 6.926400000000001e-06, "loss": 0.0184, "step": 15369 }, { "epoch": 0.8194274137655275, "grad_norm": 3.0767364501953125, "learning_rate": 6.926200000000001e-06, "loss": 0.0457, "step": 15370 }, { "epoch": 0.8194807271951805, "grad_norm": 2.3299028873443604, "learning_rate": 6.926e-06, "loss": 0.0208, "step": 15371 }, { "epoch": 0.8195340406248334, "grad_norm": 1.0943224430084229, "learning_rate": 6.925800000000001e-06, "loss": 0.0186, "step": 15372 }, { "epoch": 0.8195873540544864, "grad_norm": 0.6641717553138733, "learning_rate": 6.9256e-06, "loss": 0.0063, "step": 15373 }, { "epoch": 0.8196406674841392, "grad_norm": 2.2373886108398438, "learning_rate": 6.925400000000001e-06, "loss": 0.0284, "step": 15374 }, { "epoch": 0.8196939809137922, "grad_norm": 1.7625247240066528, "learning_rate": 6.925200000000001e-06, "loss": 0.0349, "step": 15375 }, { "epoch": 0.8197472943434451, "grad_norm": 2.1410129070281982, "learning_rate": 6.925000000000001e-06, "loss": 0.0379, "step": 15376 }, { "epoch": 0.819800607773098, "grad_norm": 5.923521041870117, "learning_rate": 6.9248e-06, "loss": 0.0645, "step": 15377 }, { "epoch": 0.819853921202751, "grad_norm": 2.599378824234009, "learning_rate": 6.924600000000001e-06, "loss": 0.0522, "step": 15378 }, { "epoch": 0.8199072346324039, "grad_norm": 4.564156532287598, "learning_rate": 6.924400000000001e-06, "loss": 0.0956, "step": 15379 }, { "epoch": 0.8199605480620569, "grad_norm": 3.0664944648742676, "learning_rate": 6.9242000000000006e-06, "loss": 0.0333, "step": 15380 }, { "epoch": 0.8200138614917097, "grad_norm": 2.5110208988189697, "learning_rate": 6.924000000000001e-06, "loss": 0.0472, "step": 15381 }, { "epoch": 0.8200671749213627, "grad_norm": 1.7914751768112183, "learning_rate": 6.9238000000000005e-06, "loss": 0.0205, "step": 15382 }, { "epoch": 0.8201204883510156, "grad_norm": 1.771586537361145, "learning_rate": 6.9236e-06, "loss": 0.0261, "step": 15383 }, { "epoch": 0.8201738017806686, "grad_norm": 1.4027683734893799, "learning_rate": 6.9234000000000005e-06, "loss": 0.0229, "step": 15384 }, { "epoch": 0.8202271152103214, "grad_norm": 1.6102535724639893, "learning_rate": 6.923200000000001e-06, "loss": 0.0139, "step": 15385 }, { "epoch": 0.8202804286399744, "grad_norm": 1.2818247079849243, "learning_rate": 6.9230000000000005e-06, "loss": 0.0241, "step": 15386 }, { "epoch": 0.8203337420696274, "grad_norm": 3.942539930343628, "learning_rate": 6.922800000000001e-06, "loss": 0.0605, "step": 15387 }, { "epoch": 0.8203870554992803, "grad_norm": 1.7047662734985352, "learning_rate": 6.9226000000000004e-06, "loss": 0.0274, "step": 15388 }, { "epoch": 0.8204403689289332, "grad_norm": 0.857399046421051, "learning_rate": 6.9224e-06, "loss": 0.0186, "step": 15389 }, { "epoch": 0.8204936823585861, "grad_norm": 1.9602991342544556, "learning_rate": 6.9222e-06, "loss": 0.0146, "step": 15390 }, { "epoch": 0.8205469957882391, "grad_norm": 1.7668043375015259, "learning_rate": 6.922000000000001e-06, "loss": 0.0352, "step": 15391 }, { "epoch": 0.820600309217892, "grad_norm": 1.9877865314483643, "learning_rate": 6.9218e-06, "loss": 0.0219, "step": 15392 }, { "epoch": 0.8206536226475449, "grad_norm": 1.9400722980499268, "learning_rate": 6.921600000000001e-06, "loss": 0.0447, "step": 15393 }, { "epoch": 0.8207069360771978, "grad_norm": 1.6908318996429443, "learning_rate": 6.9214e-06, "loss": 0.0289, "step": 15394 }, { "epoch": 0.8207602495068508, "grad_norm": 0.9347314834594727, "learning_rate": 6.9212e-06, "loss": 0.0137, "step": 15395 }, { "epoch": 0.8208135629365038, "grad_norm": 3.4370877742767334, "learning_rate": 6.921000000000001e-06, "loss": 0.0505, "step": 15396 }, { "epoch": 0.8208668763661566, "grad_norm": 1.3081544637680054, "learning_rate": 6.920800000000001e-06, "loss": 0.0219, "step": 15397 }, { "epoch": 0.8209201897958096, "grad_norm": 1.384605050086975, "learning_rate": 6.9206e-06, "loss": 0.0101, "step": 15398 }, { "epoch": 0.8209735032254625, "grad_norm": 3.236250877380371, "learning_rate": 6.920400000000001e-06, "loss": 0.0587, "step": 15399 }, { "epoch": 0.8210268166551155, "grad_norm": 1.9028085470199585, "learning_rate": 6.9202e-06, "loss": 0.0253, "step": 15400 }, { "epoch": 0.8210801300847683, "grad_norm": 1.5596250295639038, "learning_rate": 6.92e-06, "loss": 0.0165, "step": 15401 }, { "epoch": 0.8211334435144213, "grad_norm": 1.5518162250518799, "learning_rate": 6.919800000000001e-06, "loss": 0.0244, "step": 15402 }, { "epoch": 0.8211867569440742, "grad_norm": 1.7984837293624878, "learning_rate": 6.919600000000001e-06, "loss": 0.0347, "step": 15403 }, { "epoch": 0.8212400703737271, "grad_norm": 1.9214588403701782, "learning_rate": 6.9194e-06, "loss": 0.0292, "step": 15404 }, { "epoch": 0.82129338380338, "grad_norm": 2.4302570819854736, "learning_rate": 6.919200000000001e-06, "loss": 0.0325, "step": 15405 }, { "epoch": 0.821346697233033, "grad_norm": 2.594913959503174, "learning_rate": 6.919e-06, "loss": 0.0264, "step": 15406 }, { "epoch": 0.821400010662686, "grad_norm": 1.8203257322311401, "learning_rate": 6.918800000000001e-06, "loss": 0.024, "step": 15407 }, { "epoch": 0.8214533240923388, "grad_norm": 1.0854158401489258, "learning_rate": 6.918600000000001e-06, "loss": 0.0184, "step": 15408 }, { "epoch": 0.8215066375219918, "grad_norm": 1.5034055709838867, "learning_rate": 6.9184000000000005e-06, "loss": 0.0273, "step": 15409 }, { "epoch": 0.8215599509516447, "grad_norm": 2.9540679454803467, "learning_rate": 6.9182e-06, "loss": 0.0266, "step": 15410 }, { "epoch": 0.8216132643812977, "grad_norm": 1.5520027875900269, "learning_rate": 6.9180000000000005e-06, "loss": 0.0229, "step": 15411 }, { "epoch": 0.8216665778109505, "grad_norm": 1.7629228830337524, "learning_rate": 6.9178e-06, "loss": 0.0142, "step": 15412 }, { "epoch": 0.8217198912406035, "grad_norm": 0.7937225103378296, "learning_rate": 6.917600000000001e-06, "loss": 0.0085, "step": 15413 }, { "epoch": 0.8217732046702564, "grad_norm": 1.479934573173523, "learning_rate": 6.917400000000001e-06, "loss": 0.0513, "step": 15414 }, { "epoch": 0.8218265180999094, "grad_norm": 0.4815960228443146, "learning_rate": 6.9172000000000005e-06, "loss": 0.0046, "step": 15415 }, { "epoch": 0.8218798315295623, "grad_norm": 1.6047924757003784, "learning_rate": 6.917e-06, "loss": 0.0196, "step": 15416 }, { "epoch": 0.8219331449592152, "grad_norm": 3.0014777183532715, "learning_rate": 6.9168e-06, "loss": 0.0501, "step": 15417 }, { "epoch": 0.8219864583888682, "grad_norm": 0.9394479990005493, "learning_rate": 6.916600000000001e-06, "loss": 0.0072, "step": 15418 }, { "epoch": 0.8220397718185211, "grad_norm": 2.298740863800049, "learning_rate": 6.916400000000001e-06, "loss": 0.0207, "step": 15419 }, { "epoch": 0.822093085248174, "grad_norm": 2.18410587310791, "learning_rate": 6.916200000000001e-06, "loss": 0.0286, "step": 15420 }, { "epoch": 0.8221463986778269, "grad_norm": 0.9468145370483398, "learning_rate": 6.916e-06, "loss": 0.0158, "step": 15421 }, { "epoch": 0.8221997121074799, "grad_norm": 2.1661972999572754, "learning_rate": 6.9158e-06, "loss": 0.0325, "step": 15422 }, { "epoch": 0.8222530255371328, "grad_norm": 5.330305576324463, "learning_rate": 6.9156e-06, "loss": 0.0575, "step": 15423 }, { "epoch": 0.8223063389667857, "grad_norm": 2.386632204055786, "learning_rate": 6.915400000000001e-06, "loss": 0.046, "step": 15424 }, { "epoch": 0.8223596523964387, "grad_norm": 1.8321163654327393, "learning_rate": 6.915200000000001e-06, "loss": 0.0217, "step": 15425 }, { "epoch": 0.8224129658260916, "grad_norm": 1.1503629684448242, "learning_rate": 6.915000000000001e-06, "loss": 0.0238, "step": 15426 }, { "epoch": 0.8224662792557446, "grad_norm": 1.5726510286331177, "learning_rate": 6.9148e-06, "loss": 0.022, "step": 15427 }, { "epoch": 0.8225195926853974, "grad_norm": 2.2609663009643555, "learning_rate": 6.9146e-06, "loss": 0.055, "step": 15428 }, { "epoch": 0.8225729061150504, "grad_norm": 2.124680757522583, "learning_rate": 6.914400000000001e-06, "loss": 0.0279, "step": 15429 }, { "epoch": 0.8226262195447033, "grad_norm": 2.3579914569854736, "learning_rate": 6.914200000000001e-06, "loss": 0.0366, "step": 15430 }, { "epoch": 0.8226795329743563, "grad_norm": 1.8128433227539062, "learning_rate": 6.914000000000001e-06, "loss": 0.0177, "step": 15431 }, { "epoch": 0.8227328464040091, "grad_norm": 2.0449881553649902, "learning_rate": 6.913800000000001e-06, "loss": 0.0408, "step": 15432 }, { "epoch": 0.8227861598336621, "grad_norm": 2.770657539367676, "learning_rate": 6.9136e-06, "loss": 0.044, "step": 15433 }, { "epoch": 0.8228394732633151, "grad_norm": 2.5242726802825928, "learning_rate": 6.9134e-06, "loss": 0.0556, "step": 15434 }, { "epoch": 0.822892786692968, "grad_norm": 1.5436571836471558, "learning_rate": 6.913200000000001e-06, "loss": 0.0194, "step": 15435 }, { "epoch": 0.8229461001226209, "grad_norm": 1.6731469631195068, "learning_rate": 6.9130000000000006e-06, "loss": 0.0338, "step": 15436 }, { "epoch": 0.8229994135522738, "grad_norm": 1.9634798765182495, "learning_rate": 6.912800000000001e-06, "loss": 0.0543, "step": 15437 }, { "epoch": 0.8230527269819268, "grad_norm": 2.7872698307037354, "learning_rate": 6.9126000000000005e-06, "loss": 0.0305, "step": 15438 }, { "epoch": 0.8231060404115796, "grad_norm": 1.470619559288025, "learning_rate": 6.9124e-06, "loss": 0.0141, "step": 15439 }, { "epoch": 0.8231593538412326, "grad_norm": 2.077026605606079, "learning_rate": 6.9122e-06, "loss": 0.0197, "step": 15440 }, { "epoch": 0.8232126672708855, "grad_norm": 3.4840517044067383, "learning_rate": 6.912000000000001e-06, "loss": 0.0289, "step": 15441 }, { "epoch": 0.8232659807005385, "grad_norm": 3.958258628845215, "learning_rate": 6.9118000000000005e-06, "loss": 0.0414, "step": 15442 }, { "epoch": 0.8233192941301914, "grad_norm": 1.57636296749115, "learning_rate": 6.911600000000001e-06, "loss": 0.0233, "step": 15443 }, { "epoch": 0.8233726075598443, "grad_norm": 1.7998651266098022, "learning_rate": 6.9114000000000004e-06, "loss": 0.022, "step": 15444 }, { "epoch": 0.8234259209894973, "grad_norm": 1.122010588645935, "learning_rate": 6.9112e-06, "loss": 0.0232, "step": 15445 }, { "epoch": 0.8234792344191502, "grad_norm": 1.8481050729751587, "learning_rate": 6.911000000000001e-06, "loss": 0.0144, "step": 15446 }, { "epoch": 0.8235325478488031, "grad_norm": 1.391683578491211, "learning_rate": 6.910800000000001e-06, "loss": 0.0213, "step": 15447 }, { "epoch": 0.823585861278456, "grad_norm": 2.257312297821045, "learning_rate": 6.9106e-06, "loss": 0.0379, "step": 15448 }, { "epoch": 0.823639174708109, "grad_norm": 3.6640119552612305, "learning_rate": 6.910400000000001e-06, "loss": 0.0319, "step": 15449 }, { "epoch": 0.8236924881377619, "grad_norm": 1.551577091217041, "learning_rate": 6.9102e-06, "loss": 0.0342, "step": 15450 }, { "epoch": 0.8237458015674148, "grad_norm": 2.362093210220337, "learning_rate": 6.91e-06, "loss": 0.0619, "step": 15451 }, { "epoch": 0.8237991149970678, "grad_norm": 2.8328139781951904, "learning_rate": 6.909800000000001e-06, "loss": 0.0318, "step": 15452 }, { "epoch": 0.8238524284267207, "grad_norm": 4.339869499206543, "learning_rate": 6.909600000000001e-06, "loss": 0.0472, "step": 15453 }, { "epoch": 0.8239057418563737, "grad_norm": 1.585952639579773, "learning_rate": 6.9094e-06, "loss": 0.0151, "step": 15454 }, { "epoch": 0.8239590552860265, "grad_norm": 1.8404395580291748, "learning_rate": 6.909200000000001e-06, "loss": 0.0446, "step": 15455 }, { "epoch": 0.8240123687156795, "grad_norm": 0.7216576933860779, "learning_rate": 6.909e-06, "loss": 0.0057, "step": 15456 }, { "epoch": 0.8240656821453324, "grad_norm": 1.4876667261123657, "learning_rate": 6.908800000000001e-06, "loss": 0.0152, "step": 15457 }, { "epoch": 0.8241189955749854, "grad_norm": 1.559288501739502, "learning_rate": 6.908600000000001e-06, "loss": 0.0396, "step": 15458 }, { "epoch": 0.8241723090046382, "grad_norm": 1.408270239830017, "learning_rate": 6.908400000000001e-06, "loss": 0.0304, "step": 15459 }, { "epoch": 0.8242256224342912, "grad_norm": 0.6698840260505676, "learning_rate": 6.9082e-06, "loss": 0.0186, "step": 15460 }, { "epoch": 0.8242789358639441, "grad_norm": 0.964148998260498, "learning_rate": 6.908000000000001e-06, "loss": 0.0149, "step": 15461 }, { "epoch": 0.824332249293597, "grad_norm": 0.7827339768409729, "learning_rate": 6.9078e-06, "loss": 0.0228, "step": 15462 }, { "epoch": 0.82438556272325, "grad_norm": 0.9062256813049316, "learning_rate": 6.9076000000000006e-06, "loss": 0.0142, "step": 15463 }, { "epoch": 0.8244388761529029, "grad_norm": 2.2397396564483643, "learning_rate": 6.907400000000001e-06, "loss": 0.0247, "step": 15464 }, { "epoch": 0.8244921895825559, "grad_norm": 1.5683454275131226, "learning_rate": 6.9072000000000005e-06, "loss": 0.0243, "step": 15465 }, { "epoch": 0.8245455030122087, "grad_norm": 1.7341618537902832, "learning_rate": 6.907e-06, "loss": 0.0418, "step": 15466 }, { "epoch": 0.8245988164418617, "grad_norm": 2.855721950531006, "learning_rate": 6.9068000000000005e-06, "loss": 0.0125, "step": 15467 }, { "epoch": 0.8246521298715146, "grad_norm": 1.623342752456665, "learning_rate": 6.906600000000001e-06, "loss": 0.0132, "step": 15468 }, { "epoch": 0.8247054433011676, "grad_norm": 2.569597005844116, "learning_rate": 6.9064000000000005e-06, "loss": 0.0301, "step": 15469 }, { "epoch": 0.8247587567308204, "grad_norm": 1.6231058835983276, "learning_rate": 6.906200000000001e-06, "loss": 0.0375, "step": 15470 }, { "epoch": 0.8248120701604734, "grad_norm": 1.499720811843872, "learning_rate": 6.9060000000000005e-06, "loss": 0.0329, "step": 15471 }, { "epoch": 0.8248653835901264, "grad_norm": 2.2477405071258545, "learning_rate": 6.9058e-06, "loss": 0.0313, "step": 15472 }, { "epoch": 0.8249186970197793, "grad_norm": 2.9230709075927734, "learning_rate": 6.9056000000000004e-06, "loss": 0.046, "step": 15473 }, { "epoch": 0.8249720104494322, "grad_norm": 1.7064380645751953, "learning_rate": 6.905400000000001e-06, "loss": 0.0477, "step": 15474 }, { "epoch": 0.8250253238790851, "grad_norm": 3.195762872695923, "learning_rate": 6.9052e-06, "loss": 0.0196, "step": 15475 }, { "epoch": 0.8250786373087381, "grad_norm": 1.2226506471633911, "learning_rate": 6.905000000000001e-06, "loss": 0.011, "step": 15476 }, { "epoch": 0.825131950738391, "grad_norm": 5.945502281188965, "learning_rate": 6.9048e-06, "loss": 0.0705, "step": 15477 }, { "epoch": 0.8251852641680439, "grad_norm": 2.027329206466675, "learning_rate": 6.9046e-06, "loss": 0.0144, "step": 15478 }, { "epoch": 0.8252385775976968, "grad_norm": 1.5760210752487183, "learning_rate": 6.904400000000001e-06, "loss": 0.0151, "step": 15479 }, { "epoch": 0.8252918910273498, "grad_norm": 3.390429973602295, "learning_rate": 6.904200000000001e-06, "loss": 0.0389, "step": 15480 }, { "epoch": 0.8253452044570028, "grad_norm": 2.036226511001587, "learning_rate": 6.904e-06, "loss": 0.0154, "step": 15481 }, { "epoch": 0.8253985178866556, "grad_norm": 4.365905284881592, "learning_rate": 6.903800000000001e-06, "loss": 0.0167, "step": 15482 }, { "epoch": 0.8254518313163086, "grad_norm": 1.207899808883667, "learning_rate": 6.9036e-06, "loss": 0.0111, "step": 15483 }, { "epoch": 0.8255051447459615, "grad_norm": 1.8424721956253052, "learning_rate": 6.9034e-06, "loss": 0.0303, "step": 15484 }, { "epoch": 0.8255584581756145, "grad_norm": 1.7441837787628174, "learning_rate": 6.903200000000001e-06, "loss": 0.0185, "step": 15485 }, { "epoch": 0.8256117716052673, "grad_norm": 3.0443339347839355, "learning_rate": 6.903000000000001e-06, "loss": 0.0243, "step": 15486 }, { "epoch": 0.8256650850349203, "grad_norm": 3.9495866298675537, "learning_rate": 6.9028e-06, "loss": 0.077, "step": 15487 }, { "epoch": 0.8257183984645732, "grad_norm": 1.790981650352478, "learning_rate": 6.902600000000001e-06, "loss": 0.0205, "step": 15488 }, { "epoch": 0.8257717118942262, "grad_norm": 1.0782976150512695, "learning_rate": 6.9024e-06, "loss": 0.0143, "step": 15489 }, { "epoch": 0.8258250253238791, "grad_norm": 2.6506195068359375, "learning_rate": 6.9022e-06, "loss": 0.0448, "step": 15490 }, { "epoch": 0.825878338753532, "grad_norm": 2.6471221446990967, "learning_rate": 6.902000000000001e-06, "loss": 0.0263, "step": 15491 }, { "epoch": 0.825931652183185, "grad_norm": 1.2164194583892822, "learning_rate": 6.9018000000000006e-06, "loss": 0.0099, "step": 15492 }, { "epoch": 0.8259849656128379, "grad_norm": 4.479999542236328, "learning_rate": 6.9016e-06, "loss": 0.0389, "step": 15493 }, { "epoch": 0.8260382790424908, "grad_norm": 3.208738088607788, "learning_rate": 6.9014000000000005e-06, "loss": 0.027, "step": 15494 }, { "epoch": 0.8260915924721437, "grad_norm": 2.94525408744812, "learning_rate": 6.9012e-06, "loss": 0.0324, "step": 15495 }, { "epoch": 0.8261449059017967, "grad_norm": 2.0489938259124756, "learning_rate": 6.901000000000001e-06, "loss": 0.0286, "step": 15496 }, { "epoch": 0.8261982193314495, "grad_norm": 3.771190881729126, "learning_rate": 6.900800000000001e-06, "loss": 0.0278, "step": 15497 }, { "epoch": 0.8262515327611025, "grad_norm": 2.1668121814727783, "learning_rate": 6.9006000000000005e-06, "loss": 0.0266, "step": 15498 }, { "epoch": 0.8263048461907555, "grad_norm": 6.429938316345215, "learning_rate": 6.9004e-06, "loss": 0.0423, "step": 15499 }, { "epoch": 0.8263581596204084, "grad_norm": 5.222543716430664, "learning_rate": 6.9002000000000004e-06, "loss": 0.048, "step": 15500 }, { "epoch": 0.8264114730500614, "grad_norm": 4.004080772399902, "learning_rate": 6.9e-06, "loss": 0.0364, "step": 15501 }, { "epoch": 0.8264647864797142, "grad_norm": 1.4258500337600708, "learning_rate": 6.899800000000001e-06, "loss": 0.0272, "step": 15502 }, { "epoch": 0.8265180999093672, "grad_norm": 2.9805803298950195, "learning_rate": 6.899600000000001e-06, "loss": 0.0549, "step": 15503 }, { "epoch": 0.8265714133390201, "grad_norm": 2.575000286102295, "learning_rate": 6.8994e-06, "loss": 0.0445, "step": 15504 }, { "epoch": 0.826624726768673, "grad_norm": 3.234358787536621, "learning_rate": 6.8992e-06, "loss": 0.0222, "step": 15505 }, { "epoch": 0.8266780401983259, "grad_norm": 1.701005458831787, "learning_rate": 6.899e-06, "loss": 0.0136, "step": 15506 }, { "epoch": 0.8267313536279789, "grad_norm": 3.3571317195892334, "learning_rate": 6.898800000000001e-06, "loss": 0.0736, "step": 15507 }, { "epoch": 0.8267846670576319, "grad_norm": 2.7250335216522217, "learning_rate": 6.898600000000001e-06, "loss": 0.0501, "step": 15508 }, { "epoch": 0.8268379804872847, "grad_norm": 4.056675434112549, "learning_rate": 6.898400000000001e-06, "loss": 0.0556, "step": 15509 }, { "epoch": 0.8268912939169377, "grad_norm": 6.371401786804199, "learning_rate": 6.8982e-06, "loss": 0.0221, "step": 15510 }, { "epoch": 0.8269446073465906, "grad_norm": 4.624333381652832, "learning_rate": 6.898e-06, "loss": 0.064, "step": 15511 }, { "epoch": 0.8269979207762436, "grad_norm": 3.658522844314575, "learning_rate": 6.8978e-06, "loss": 0.0536, "step": 15512 }, { "epoch": 0.8270512342058964, "grad_norm": 4.995629787445068, "learning_rate": 6.897600000000001e-06, "loss": 0.0526, "step": 15513 }, { "epoch": 0.8271045476355494, "grad_norm": 4.361952781677246, "learning_rate": 6.897400000000001e-06, "loss": 0.0452, "step": 15514 }, { "epoch": 0.8271578610652023, "grad_norm": 2.0628111362457275, "learning_rate": 6.897200000000001e-06, "loss": 0.0209, "step": 15515 }, { "epoch": 0.8272111744948553, "grad_norm": 2.3741297721862793, "learning_rate": 6.897e-06, "loss": 0.0381, "step": 15516 }, { "epoch": 0.8272644879245081, "grad_norm": 0.7977494597434998, "learning_rate": 6.8968e-06, "loss": 0.0077, "step": 15517 }, { "epoch": 0.8273178013541611, "grad_norm": 2.0922629833221436, "learning_rate": 6.896600000000001e-06, "loss": 0.0225, "step": 15518 }, { "epoch": 0.8273711147838141, "grad_norm": 3.9593238830566406, "learning_rate": 6.896400000000001e-06, "loss": 0.0217, "step": 15519 }, { "epoch": 0.827424428213467, "grad_norm": 2.0946786403656006, "learning_rate": 6.896200000000001e-06, "loss": 0.0269, "step": 15520 }, { "epoch": 0.8274777416431199, "grad_norm": 4.247014045715332, "learning_rate": 6.8960000000000006e-06, "loss": 0.0479, "step": 15521 }, { "epoch": 0.8275310550727728, "grad_norm": 2.3830714225769043, "learning_rate": 6.8958e-06, "loss": 0.0252, "step": 15522 }, { "epoch": 0.8275843685024258, "grad_norm": 1.7498916387557983, "learning_rate": 6.8956e-06, "loss": 0.0086, "step": 15523 }, { "epoch": 0.8276376819320787, "grad_norm": 3.685964584350586, "learning_rate": 6.895400000000001e-06, "loss": 0.0447, "step": 15524 }, { "epoch": 0.8276909953617316, "grad_norm": 1.6562299728393555, "learning_rate": 6.8952000000000005e-06, "loss": 0.0386, "step": 15525 }, { "epoch": 0.8277443087913845, "grad_norm": 2.8568387031555176, "learning_rate": 6.895000000000001e-06, "loss": 0.039, "step": 15526 }, { "epoch": 0.8277976222210375, "grad_norm": 1.5269067287445068, "learning_rate": 6.8948000000000005e-06, "loss": 0.0209, "step": 15527 }, { "epoch": 0.8278509356506905, "grad_norm": 2.1100499629974365, "learning_rate": 6.8946e-06, "loss": 0.0238, "step": 15528 }, { "epoch": 0.8279042490803433, "grad_norm": 2.1634840965270996, "learning_rate": 6.894400000000001e-06, "loss": 0.0334, "step": 15529 }, { "epoch": 0.8279575625099963, "grad_norm": 2.3162612915039062, "learning_rate": 6.894200000000001e-06, "loss": 0.0422, "step": 15530 }, { "epoch": 0.8280108759396492, "grad_norm": 3.546125888824463, "learning_rate": 6.894e-06, "loss": 0.0373, "step": 15531 }, { "epoch": 0.8280641893693022, "grad_norm": 4.913278579711914, "learning_rate": 6.893800000000001e-06, "loss": 0.0394, "step": 15532 }, { "epoch": 0.828117502798955, "grad_norm": 2.416562557220459, "learning_rate": 6.8936e-06, "loss": 0.0206, "step": 15533 }, { "epoch": 0.828170816228608, "grad_norm": 5.8706560134887695, "learning_rate": 6.8934e-06, "loss": 0.0248, "step": 15534 }, { "epoch": 0.8282241296582609, "grad_norm": 1.796510934829712, "learning_rate": 6.893200000000001e-06, "loss": 0.0158, "step": 15535 }, { "epoch": 0.8282774430879138, "grad_norm": 2.295189380645752, "learning_rate": 6.893000000000001e-06, "loss": 0.0234, "step": 15536 }, { "epoch": 0.8283307565175668, "grad_norm": 2.2650742530822754, "learning_rate": 6.8928e-06, "loss": 0.0264, "step": 15537 }, { "epoch": 0.8283840699472197, "grad_norm": 1.9907336235046387, "learning_rate": 6.892600000000001e-06, "loss": 0.0276, "step": 15538 }, { "epoch": 0.8284373833768727, "grad_norm": 3.3451988697052, "learning_rate": 6.8924e-06, "loss": 0.0285, "step": 15539 }, { "epoch": 0.8284906968065255, "grad_norm": 1.4676507711410522, "learning_rate": 6.892200000000001e-06, "loss": 0.0214, "step": 15540 }, { "epoch": 0.8285440102361785, "grad_norm": 1.5003199577331543, "learning_rate": 6.892000000000001e-06, "loss": 0.0238, "step": 15541 }, { "epoch": 0.8285973236658314, "grad_norm": 1.1249582767486572, "learning_rate": 6.891800000000001e-06, "loss": 0.0169, "step": 15542 }, { "epoch": 0.8286506370954844, "grad_norm": 1.7946006059646606, "learning_rate": 6.8916e-06, "loss": 0.0185, "step": 15543 }, { "epoch": 0.8287039505251372, "grad_norm": 1.2620517015457153, "learning_rate": 6.891400000000001e-06, "loss": 0.0191, "step": 15544 }, { "epoch": 0.8287572639547902, "grad_norm": 2.023906946182251, "learning_rate": 6.8912e-06, "loss": 0.0301, "step": 15545 }, { "epoch": 0.8288105773844432, "grad_norm": 2.279864549636841, "learning_rate": 6.891000000000001e-06, "loss": 0.0402, "step": 15546 }, { "epoch": 0.8288638908140961, "grad_norm": 3.294257640838623, "learning_rate": 6.890800000000001e-06, "loss": 0.0359, "step": 15547 }, { "epoch": 0.828917204243749, "grad_norm": 1.318244457244873, "learning_rate": 6.8906000000000006e-06, "loss": 0.0097, "step": 15548 }, { "epoch": 0.8289705176734019, "grad_norm": 1.1334995031356812, "learning_rate": 6.8904e-06, "loss": 0.0157, "step": 15549 }, { "epoch": 0.8290238311030549, "grad_norm": 2.514852285385132, "learning_rate": 6.8902000000000005e-06, "loss": 0.0327, "step": 15550 }, { "epoch": 0.8290771445327078, "grad_norm": 2.7040131092071533, "learning_rate": 6.89e-06, "loss": 0.0257, "step": 15551 }, { "epoch": 0.8291304579623607, "grad_norm": 3.297136068344116, "learning_rate": 6.8898000000000005e-06, "loss": 0.0244, "step": 15552 }, { "epoch": 0.8291837713920136, "grad_norm": 1.2793724536895752, "learning_rate": 6.889600000000001e-06, "loss": 0.0453, "step": 15553 }, { "epoch": 0.8292370848216666, "grad_norm": 4.866113185882568, "learning_rate": 6.8894000000000005e-06, "loss": 0.0438, "step": 15554 }, { "epoch": 0.8292903982513196, "grad_norm": 2.6911256313323975, "learning_rate": 6.8892e-06, "loss": 0.0239, "step": 15555 }, { "epoch": 0.8293437116809724, "grad_norm": 1.458810806274414, "learning_rate": 6.8890000000000004e-06, "loss": 0.0141, "step": 15556 }, { "epoch": 0.8293970251106254, "grad_norm": 1.2453243732452393, "learning_rate": 6.888800000000001e-06, "loss": 0.0118, "step": 15557 }, { "epoch": 0.8294503385402783, "grad_norm": 2.094625949859619, "learning_rate": 6.8886e-06, "loss": 0.0513, "step": 15558 }, { "epoch": 0.8295036519699313, "grad_norm": 1.965239405632019, "learning_rate": 6.888400000000001e-06, "loss": 0.0444, "step": 15559 }, { "epoch": 0.8295569653995841, "grad_norm": 1.5355552434921265, "learning_rate": 6.8882e-06, "loss": 0.0119, "step": 15560 }, { "epoch": 0.8296102788292371, "grad_norm": 3.2869231700897217, "learning_rate": 6.888e-06, "loss": 0.0717, "step": 15561 }, { "epoch": 0.82966359225889, "grad_norm": 3.1321628093719482, "learning_rate": 6.8878e-06, "loss": 0.0813, "step": 15562 }, { "epoch": 0.829716905688543, "grad_norm": 2.9463682174682617, "learning_rate": 6.887600000000001e-06, "loss": 0.0469, "step": 15563 }, { "epoch": 0.8297702191181959, "grad_norm": 1.819205403327942, "learning_rate": 6.8874e-06, "loss": 0.0415, "step": 15564 }, { "epoch": 0.8298235325478488, "grad_norm": 1.3934625387191772, "learning_rate": 6.887200000000001e-06, "loss": 0.0266, "step": 15565 }, { "epoch": 0.8298768459775018, "grad_norm": 2.900614023208618, "learning_rate": 6.887e-06, "loss": 0.0286, "step": 15566 }, { "epoch": 0.8299301594071546, "grad_norm": 3.4117116928100586, "learning_rate": 6.8868e-06, "loss": 0.0538, "step": 15567 }, { "epoch": 0.8299834728368076, "grad_norm": 1.5170058012008667, "learning_rate": 6.886600000000001e-06, "loss": 0.0274, "step": 15568 }, { "epoch": 0.8300367862664605, "grad_norm": 5.500540733337402, "learning_rate": 6.886400000000001e-06, "loss": 0.042, "step": 15569 }, { "epoch": 0.8300900996961135, "grad_norm": 2.5643420219421387, "learning_rate": 6.8862e-06, "loss": 0.0304, "step": 15570 }, { "epoch": 0.8301434131257663, "grad_norm": 1.4411413669586182, "learning_rate": 6.886000000000001e-06, "loss": 0.0119, "step": 15571 }, { "epoch": 0.8301967265554193, "grad_norm": 2.2094216346740723, "learning_rate": 6.8858e-06, "loss": 0.0224, "step": 15572 }, { "epoch": 0.8302500399850722, "grad_norm": 0.6475786566734314, "learning_rate": 6.8856e-06, "loss": 0.0072, "step": 15573 }, { "epoch": 0.8303033534147252, "grad_norm": 1.3952794075012207, "learning_rate": 6.885400000000001e-06, "loss": 0.0255, "step": 15574 }, { "epoch": 0.8303566668443781, "grad_norm": 1.5572452545166016, "learning_rate": 6.885200000000001e-06, "loss": 0.0216, "step": 15575 }, { "epoch": 0.830409980274031, "grad_norm": 1.9412870407104492, "learning_rate": 6.885e-06, "loss": 0.0327, "step": 15576 }, { "epoch": 0.830463293703684, "grad_norm": 2.2589476108551025, "learning_rate": 6.8848000000000006e-06, "loss": 0.0439, "step": 15577 }, { "epoch": 0.8305166071333369, "grad_norm": 1.9026834964752197, "learning_rate": 6.8846e-06, "loss": 0.0186, "step": 15578 }, { "epoch": 0.8305699205629898, "grad_norm": 1.737204670906067, "learning_rate": 6.884400000000001e-06, "loss": 0.0332, "step": 15579 }, { "epoch": 0.8306232339926427, "grad_norm": 1.6898608207702637, "learning_rate": 6.884200000000001e-06, "loss": 0.0379, "step": 15580 }, { "epoch": 0.8306765474222957, "grad_norm": 1.3291152715682983, "learning_rate": 6.8840000000000005e-06, "loss": 0.0394, "step": 15581 }, { "epoch": 0.8307298608519486, "grad_norm": 1.4866790771484375, "learning_rate": 6.8838e-06, "loss": 0.0271, "step": 15582 }, { "epoch": 0.8307831742816015, "grad_norm": 1.4751720428466797, "learning_rate": 6.8836000000000005e-06, "loss": 0.0299, "step": 15583 }, { "epoch": 0.8308364877112545, "grad_norm": 1.7296727895736694, "learning_rate": 6.8834e-06, "loss": 0.033, "step": 15584 }, { "epoch": 0.8308898011409074, "grad_norm": 2.9115822315216064, "learning_rate": 6.883200000000001e-06, "loss": 0.0264, "step": 15585 }, { "epoch": 0.8309431145705604, "grad_norm": 1.4796375036239624, "learning_rate": 6.883000000000001e-06, "loss": 0.0125, "step": 15586 }, { "epoch": 0.8309964280002132, "grad_norm": 1.5351043939590454, "learning_rate": 6.8828e-06, "loss": 0.0194, "step": 15587 }, { "epoch": 0.8310497414298662, "grad_norm": 1.1965560913085938, "learning_rate": 6.8826e-06, "loss": 0.0134, "step": 15588 }, { "epoch": 0.8311030548595191, "grad_norm": 3.2845723628997803, "learning_rate": 6.8824e-06, "loss": 0.0687, "step": 15589 }, { "epoch": 0.831156368289172, "grad_norm": 1.5732593536376953, "learning_rate": 6.882200000000001e-06, "loss": 0.0221, "step": 15590 }, { "epoch": 0.8312096817188249, "grad_norm": 3.3634097576141357, "learning_rate": 6.882000000000001e-06, "loss": 0.0666, "step": 15591 }, { "epoch": 0.8312629951484779, "grad_norm": 3.589613437652588, "learning_rate": 6.881800000000001e-06, "loss": 0.0442, "step": 15592 }, { "epoch": 0.8313163085781309, "grad_norm": 2.210538387298584, "learning_rate": 6.8816e-06, "loss": 0.0253, "step": 15593 }, { "epoch": 0.8313696220077837, "grad_norm": 2.451934337615967, "learning_rate": 6.881400000000001e-06, "loss": 0.0414, "step": 15594 }, { "epoch": 0.8314229354374367, "grad_norm": 2.0787100791931152, "learning_rate": 6.8812e-06, "loss": 0.017, "step": 15595 }, { "epoch": 0.8314762488670896, "grad_norm": 1.2768245935440063, "learning_rate": 6.881000000000001e-06, "loss": 0.0291, "step": 15596 }, { "epoch": 0.8315295622967426, "grad_norm": 1.6231329441070557, "learning_rate": 6.880800000000001e-06, "loss": 0.0305, "step": 15597 }, { "epoch": 0.8315828757263954, "grad_norm": 1.62626051902771, "learning_rate": 6.880600000000001e-06, "loss": 0.0145, "step": 15598 }, { "epoch": 0.8316361891560484, "grad_norm": 1.3891425132751465, "learning_rate": 6.8804e-06, "loss": 0.0283, "step": 15599 }, { "epoch": 0.8316895025857013, "grad_norm": 1.2396076917648315, "learning_rate": 6.880200000000001e-06, "loss": 0.0117, "step": 15600 }, { "epoch": 0.8317428160153543, "grad_norm": 2.6705613136291504, "learning_rate": 6.88e-06, "loss": 0.0577, "step": 15601 }, { "epoch": 0.8317961294450072, "grad_norm": 1.3567732572555542, "learning_rate": 6.879800000000001e-06, "loss": 0.0182, "step": 15602 }, { "epoch": 0.8318494428746601, "grad_norm": 1.655873417854309, "learning_rate": 6.879600000000001e-06, "loss": 0.0106, "step": 15603 }, { "epoch": 0.8319027563043131, "grad_norm": 0.7388539910316467, "learning_rate": 6.8794000000000006e-06, "loss": 0.0079, "step": 15604 }, { "epoch": 0.831956069733966, "grad_norm": 2.6868138313293457, "learning_rate": 6.8792e-06, "loss": 0.0407, "step": 15605 }, { "epoch": 0.8320093831636189, "grad_norm": 1.6067028045654297, "learning_rate": 6.8790000000000005e-06, "loss": 0.0178, "step": 15606 }, { "epoch": 0.8320626965932718, "grad_norm": 1.8200782537460327, "learning_rate": 6.878800000000001e-06, "loss": 0.0173, "step": 15607 }, { "epoch": 0.8321160100229248, "grad_norm": 2.206217050552368, "learning_rate": 6.8786000000000005e-06, "loss": 0.0349, "step": 15608 }, { "epoch": 0.8321693234525777, "grad_norm": 2.4199535846710205, "learning_rate": 6.878400000000001e-06, "loss": 0.0527, "step": 15609 }, { "epoch": 0.8322226368822306, "grad_norm": 0.9251025915145874, "learning_rate": 6.8782000000000005e-06, "loss": 0.0082, "step": 15610 }, { "epoch": 0.8322759503118836, "grad_norm": 2.442492723464966, "learning_rate": 6.878e-06, "loss": 0.0205, "step": 15611 }, { "epoch": 0.8323292637415365, "grad_norm": 3.9984676837921143, "learning_rate": 6.8778000000000005e-06, "loss": 0.0425, "step": 15612 }, { "epoch": 0.8323825771711895, "grad_norm": 3.242192506790161, "learning_rate": 6.877600000000001e-06, "loss": 0.0598, "step": 15613 }, { "epoch": 0.8324358906008423, "grad_norm": 1.9227181673049927, "learning_rate": 6.8774000000000004e-06, "loss": 0.0192, "step": 15614 }, { "epoch": 0.8324892040304953, "grad_norm": 2.338008165359497, "learning_rate": 6.877200000000001e-06, "loss": 0.0271, "step": 15615 }, { "epoch": 0.8325425174601482, "grad_norm": 2.40483021736145, "learning_rate": 6.877e-06, "loss": 0.0425, "step": 15616 }, { "epoch": 0.8325958308898012, "grad_norm": 1.8534611463546753, "learning_rate": 6.8768e-06, "loss": 0.0466, "step": 15617 }, { "epoch": 0.832649144319454, "grad_norm": 2.966073989868164, "learning_rate": 6.876600000000001e-06, "loss": 0.0231, "step": 15618 }, { "epoch": 0.832702457749107, "grad_norm": 4.008759021759033, "learning_rate": 6.876400000000001e-06, "loss": 0.0491, "step": 15619 }, { "epoch": 0.83275577117876, "grad_norm": 1.1855850219726562, "learning_rate": 6.8762e-06, "loss": 0.0187, "step": 15620 }, { "epoch": 0.8328090846084129, "grad_norm": 1.9301701784133911, "learning_rate": 6.876000000000001e-06, "loss": 0.0216, "step": 15621 }, { "epoch": 0.8328623980380658, "grad_norm": 7.536624431610107, "learning_rate": 6.8758e-06, "loss": 0.0265, "step": 15622 }, { "epoch": 0.8329157114677187, "grad_norm": 1.052722692489624, "learning_rate": 6.8756e-06, "loss": 0.0209, "step": 15623 }, { "epoch": 0.8329690248973717, "grad_norm": 2.583840847015381, "learning_rate": 6.875400000000001e-06, "loss": 0.0421, "step": 15624 }, { "epoch": 0.8330223383270245, "grad_norm": 1.411966323852539, "learning_rate": 6.875200000000001e-06, "loss": 0.0251, "step": 15625 }, { "epoch": 0.8330756517566775, "grad_norm": 2.00394868850708, "learning_rate": 6.875e-06, "loss": 0.0144, "step": 15626 }, { "epoch": 0.8331289651863304, "grad_norm": 1.6288028955459595, "learning_rate": 6.874800000000001e-06, "loss": 0.0232, "step": 15627 }, { "epoch": 0.8331822786159834, "grad_norm": 3.6470587253570557, "learning_rate": 6.8746e-06, "loss": 0.0314, "step": 15628 }, { "epoch": 0.8332355920456362, "grad_norm": 3.225409984588623, "learning_rate": 6.8744000000000015e-06, "loss": 0.0573, "step": 15629 }, { "epoch": 0.8332889054752892, "grad_norm": 2.51363205909729, "learning_rate": 6.874200000000001e-06, "loss": 0.0302, "step": 15630 }, { "epoch": 0.8333422189049422, "grad_norm": 1.749254822731018, "learning_rate": 6.874000000000001e-06, "loss": 0.0088, "step": 15631 }, { "epoch": 0.8333955323345951, "grad_norm": 5.151574611663818, "learning_rate": 6.8738e-06, "loss": 0.0412, "step": 15632 }, { "epoch": 0.833448845764248, "grad_norm": 1.6799243688583374, "learning_rate": 6.8736000000000006e-06, "loss": 0.0118, "step": 15633 }, { "epoch": 0.8335021591939009, "grad_norm": 3.0232057571411133, "learning_rate": 6.8734e-06, "loss": 0.037, "step": 15634 }, { "epoch": 0.8335554726235539, "grad_norm": 4.648103713989258, "learning_rate": 6.873200000000001e-06, "loss": 0.0294, "step": 15635 }, { "epoch": 0.8336087860532068, "grad_norm": 4.725656986236572, "learning_rate": 6.873000000000001e-06, "loss": 0.05, "step": 15636 }, { "epoch": 0.8336620994828597, "grad_norm": 2.955113410949707, "learning_rate": 6.8728000000000005e-06, "loss": 0.0396, "step": 15637 }, { "epoch": 0.8337154129125126, "grad_norm": 1.2729356288909912, "learning_rate": 6.8726e-06, "loss": 0.0186, "step": 15638 }, { "epoch": 0.8337687263421656, "grad_norm": 2.2633190155029297, "learning_rate": 6.8724000000000005e-06, "loss": 0.0227, "step": 15639 }, { "epoch": 0.8338220397718186, "grad_norm": 3.0409750938415527, "learning_rate": 6.872200000000001e-06, "loss": 0.0396, "step": 15640 }, { "epoch": 0.8338753532014714, "grad_norm": 2.280482053756714, "learning_rate": 6.872000000000001e-06, "loss": 0.0485, "step": 15641 }, { "epoch": 0.8339286666311244, "grad_norm": 1.6813478469848633, "learning_rate": 6.871800000000001e-06, "loss": 0.0369, "step": 15642 }, { "epoch": 0.8339819800607773, "grad_norm": 1.4209868907928467, "learning_rate": 6.8716e-06, "loss": 0.0274, "step": 15643 }, { "epoch": 0.8340352934904303, "grad_norm": 1.5252481698989868, "learning_rate": 6.8714e-06, "loss": 0.0257, "step": 15644 }, { "epoch": 0.8340886069200831, "grad_norm": 2.8863754272460938, "learning_rate": 6.8712e-06, "loss": 0.0376, "step": 15645 }, { "epoch": 0.8341419203497361, "grad_norm": 1.5880247354507446, "learning_rate": 6.871000000000001e-06, "loss": 0.0144, "step": 15646 }, { "epoch": 0.834195233779389, "grad_norm": 2.0838425159454346, "learning_rate": 6.870800000000001e-06, "loss": 0.0126, "step": 15647 }, { "epoch": 0.834248547209042, "grad_norm": 2.449716806411743, "learning_rate": 6.870600000000001e-06, "loss": 0.0561, "step": 15648 }, { "epoch": 0.8343018606386949, "grad_norm": 1.8235118389129639, "learning_rate": 6.8704e-06, "loss": 0.0334, "step": 15649 }, { "epoch": 0.8343551740683478, "grad_norm": 2.239790916442871, "learning_rate": 6.8702e-06, "loss": 0.0312, "step": 15650 }, { "epoch": 0.8344084874980008, "grad_norm": 1.8728737831115723, "learning_rate": 6.870000000000001e-06, "loss": 0.0302, "step": 15651 }, { "epoch": 0.8344618009276537, "grad_norm": 2.4933056831359863, "learning_rate": 6.869800000000001e-06, "loss": 0.0245, "step": 15652 }, { "epoch": 0.8345151143573066, "grad_norm": 1.6684762239456177, "learning_rate": 6.869600000000001e-06, "loss": 0.0213, "step": 15653 }, { "epoch": 0.8345684277869595, "grad_norm": 1.3861812353134155, "learning_rate": 6.869400000000001e-06, "loss": 0.0136, "step": 15654 }, { "epoch": 0.8346217412166125, "grad_norm": 1.1595653295516968, "learning_rate": 6.8692e-06, "loss": 0.0085, "step": 15655 }, { "epoch": 0.8346750546462653, "grad_norm": 2.9465503692626953, "learning_rate": 6.869e-06, "loss": 0.0416, "step": 15656 }, { "epoch": 0.8347283680759183, "grad_norm": 2.9751436710357666, "learning_rate": 6.868800000000001e-06, "loss": 0.0221, "step": 15657 }, { "epoch": 0.8347816815055713, "grad_norm": 2.9884204864501953, "learning_rate": 6.868600000000001e-06, "loss": 0.0246, "step": 15658 }, { "epoch": 0.8348349949352242, "grad_norm": 1.2647770643234253, "learning_rate": 6.868400000000001e-06, "loss": 0.0184, "step": 15659 }, { "epoch": 0.8348883083648772, "grad_norm": 3.028242588043213, "learning_rate": 6.868200000000001e-06, "loss": 0.0234, "step": 15660 }, { "epoch": 0.83494162179453, "grad_norm": 7.652839183807373, "learning_rate": 6.868e-06, "loss": 0.0426, "step": 15661 }, { "epoch": 0.834994935224183, "grad_norm": 1.936186671257019, "learning_rate": 6.8678e-06, "loss": 0.014, "step": 15662 }, { "epoch": 0.8350482486538359, "grad_norm": 1.9616954326629639, "learning_rate": 6.867600000000001e-06, "loss": 0.0419, "step": 15663 }, { "epoch": 0.8351015620834888, "grad_norm": 4.329916000366211, "learning_rate": 6.8674000000000005e-06, "loss": 0.0337, "step": 15664 }, { "epoch": 0.8351548755131417, "grad_norm": 0.5021361708641052, "learning_rate": 6.867200000000001e-06, "loss": 0.0054, "step": 15665 }, { "epoch": 0.8352081889427947, "grad_norm": 3.1456143856048584, "learning_rate": 6.8670000000000005e-06, "loss": 0.0364, "step": 15666 }, { "epoch": 0.8352615023724477, "grad_norm": 1.5782129764556885, "learning_rate": 6.8668e-06, "loss": 0.0125, "step": 15667 }, { "epoch": 0.8353148158021005, "grad_norm": 1.702333927154541, "learning_rate": 6.866600000000001e-06, "loss": 0.0117, "step": 15668 }, { "epoch": 0.8353681292317535, "grad_norm": 1.4524497985839844, "learning_rate": 6.866400000000001e-06, "loss": 0.0213, "step": 15669 }, { "epoch": 0.8354214426614064, "grad_norm": 2.1600544452667236, "learning_rate": 6.8662000000000004e-06, "loss": 0.0182, "step": 15670 }, { "epoch": 0.8354747560910594, "grad_norm": 3.2250187397003174, "learning_rate": 6.866000000000001e-06, "loss": 0.0577, "step": 15671 }, { "epoch": 0.8355280695207122, "grad_norm": 2.9053850173950195, "learning_rate": 6.8658e-06, "loss": 0.0491, "step": 15672 }, { "epoch": 0.8355813829503652, "grad_norm": 3.8183443546295166, "learning_rate": 6.8656e-06, "loss": 0.035, "step": 15673 }, { "epoch": 0.8356346963800181, "grad_norm": 3.458970785140991, "learning_rate": 6.865400000000001e-06, "loss": 0.0472, "step": 15674 }, { "epoch": 0.8356880098096711, "grad_norm": 2.6033077239990234, "learning_rate": 6.865200000000001e-06, "loss": 0.0314, "step": 15675 }, { "epoch": 0.835741323239324, "grad_norm": 1.6626222133636475, "learning_rate": 6.865e-06, "loss": 0.0137, "step": 15676 }, { "epoch": 0.8357946366689769, "grad_norm": 2.4338340759277344, "learning_rate": 6.864800000000001e-06, "loss": 0.0496, "step": 15677 }, { "epoch": 0.8358479500986299, "grad_norm": 5.263943672180176, "learning_rate": 6.8646e-06, "loss": 0.0362, "step": 15678 }, { "epoch": 0.8359012635282828, "grad_norm": 1.9264028072357178, "learning_rate": 6.864400000000001e-06, "loss": 0.0313, "step": 15679 }, { "epoch": 0.8359545769579357, "grad_norm": 1.4299468994140625, "learning_rate": 6.864200000000001e-06, "loss": 0.0268, "step": 15680 }, { "epoch": 0.8360078903875886, "grad_norm": 7.603318691253662, "learning_rate": 6.864000000000001e-06, "loss": 0.0509, "step": 15681 }, { "epoch": 0.8360612038172416, "grad_norm": 1.593670129776001, "learning_rate": 6.8638e-06, "loss": 0.0337, "step": 15682 }, { "epoch": 0.8361145172468945, "grad_norm": 2.0689475536346436, "learning_rate": 6.863600000000001e-06, "loss": 0.0283, "step": 15683 }, { "epoch": 0.8361678306765474, "grad_norm": 1.4599356651306152, "learning_rate": 6.8634e-06, "loss": 0.0142, "step": 15684 }, { "epoch": 0.8362211441062003, "grad_norm": 1.9516947269439697, "learning_rate": 6.863200000000001e-06, "loss": 0.0207, "step": 15685 }, { "epoch": 0.8362744575358533, "grad_norm": 1.732628583908081, "learning_rate": 6.863000000000001e-06, "loss": 0.023, "step": 15686 }, { "epoch": 0.8363277709655063, "grad_norm": 1.4573742151260376, "learning_rate": 6.862800000000001e-06, "loss": 0.0268, "step": 15687 }, { "epoch": 0.8363810843951591, "grad_norm": 3.215189218521118, "learning_rate": 6.8626e-06, "loss": 0.0384, "step": 15688 }, { "epoch": 0.8364343978248121, "grad_norm": 2.1546878814697266, "learning_rate": 6.8624000000000006e-06, "loss": 0.0285, "step": 15689 }, { "epoch": 0.836487711254465, "grad_norm": 4.07520866394043, "learning_rate": 6.862200000000001e-06, "loss": 0.0737, "step": 15690 }, { "epoch": 0.836541024684118, "grad_norm": 1.7129184007644653, "learning_rate": 6.8620000000000005e-06, "loss": 0.0251, "step": 15691 }, { "epoch": 0.8365943381137708, "grad_norm": 1.3764100074768066, "learning_rate": 6.861800000000001e-06, "loss": 0.0188, "step": 15692 }, { "epoch": 0.8366476515434238, "grad_norm": 2.1556081771850586, "learning_rate": 6.8616000000000005e-06, "loss": 0.0449, "step": 15693 }, { "epoch": 0.8367009649730767, "grad_norm": 1.8138272762298584, "learning_rate": 6.8614e-06, "loss": 0.0177, "step": 15694 }, { "epoch": 0.8367542784027296, "grad_norm": 1.4329509735107422, "learning_rate": 6.8612000000000005e-06, "loss": 0.0358, "step": 15695 }, { "epoch": 0.8368075918323826, "grad_norm": 1.6306657791137695, "learning_rate": 6.861000000000001e-06, "loss": 0.041, "step": 15696 }, { "epoch": 0.8368609052620355, "grad_norm": 1.579222559928894, "learning_rate": 6.8608000000000004e-06, "loss": 0.018, "step": 15697 }, { "epoch": 0.8369142186916885, "grad_norm": 2.105276107788086, "learning_rate": 6.860600000000001e-06, "loss": 0.0167, "step": 15698 }, { "epoch": 0.8369675321213413, "grad_norm": 1.7133711576461792, "learning_rate": 6.8604e-06, "loss": 0.0229, "step": 15699 }, { "epoch": 0.8370208455509943, "grad_norm": 4.242223739624023, "learning_rate": 6.8602e-06, "loss": 0.0333, "step": 15700 }, { "epoch": 0.8370741589806472, "grad_norm": 4.748162746429443, "learning_rate": 6.860000000000001e-06, "loss": 0.0572, "step": 15701 }, { "epoch": 0.8371274724103002, "grad_norm": 1.259566307067871, "learning_rate": 6.859800000000001e-06, "loss": 0.0165, "step": 15702 }, { "epoch": 0.837180785839953, "grad_norm": 3.0444562435150146, "learning_rate": 6.8596e-06, "loss": 0.0447, "step": 15703 }, { "epoch": 0.837234099269606, "grad_norm": 1.377509593963623, "learning_rate": 6.859400000000001e-06, "loss": 0.0131, "step": 15704 }, { "epoch": 0.837287412699259, "grad_norm": 2.2600252628326416, "learning_rate": 6.8592e-06, "loss": 0.0374, "step": 15705 }, { "epoch": 0.8373407261289119, "grad_norm": 2.7722055912017822, "learning_rate": 6.859e-06, "loss": 0.0292, "step": 15706 }, { "epoch": 0.8373940395585648, "grad_norm": 1.3144394159317017, "learning_rate": 6.858800000000001e-06, "loss": 0.0202, "step": 15707 }, { "epoch": 0.8374473529882177, "grad_norm": 2.105832815170288, "learning_rate": 6.858600000000001e-06, "loss": 0.0152, "step": 15708 }, { "epoch": 0.8375006664178707, "grad_norm": 2.6003358364105225, "learning_rate": 6.8584e-06, "loss": 0.0413, "step": 15709 }, { "epoch": 0.8375539798475236, "grad_norm": 2.0165634155273438, "learning_rate": 6.858200000000001e-06, "loss": 0.035, "step": 15710 }, { "epoch": 0.8376072932771765, "grad_norm": 1.9725828170776367, "learning_rate": 6.858e-06, "loss": 0.025, "step": 15711 }, { "epoch": 0.8376606067068294, "grad_norm": 0.9866325259208679, "learning_rate": 6.8578e-06, "loss": 0.016, "step": 15712 }, { "epoch": 0.8377139201364824, "grad_norm": 1.6119381189346313, "learning_rate": 6.857600000000001e-06, "loss": 0.046, "step": 15713 }, { "epoch": 0.8377672335661354, "grad_norm": 1.382752776145935, "learning_rate": 6.857400000000001e-06, "loss": 0.0281, "step": 15714 }, { "epoch": 0.8378205469957882, "grad_norm": 1.8029818534851074, "learning_rate": 6.8572e-06, "loss": 0.0433, "step": 15715 }, { "epoch": 0.8378738604254412, "grad_norm": 2.021327495574951, "learning_rate": 6.857000000000001e-06, "loss": 0.0425, "step": 15716 }, { "epoch": 0.8379271738550941, "grad_norm": 1.860522747039795, "learning_rate": 6.8568e-06, "loss": 0.0298, "step": 15717 }, { "epoch": 0.837980487284747, "grad_norm": 2.3210601806640625, "learning_rate": 6.856600000000001e-06, "loss": 0.0319, "step": 15718 }, { "epoch": 0.8380338007143999, "grad_norm": 1.7303075790405273, "learning_rate": 6.856400000000001e-06, "loss": 0.0113, "step": 15719 }, { "epoch": 0.8380871141440529, "grad_norm": 1.7844622135162354, "learning_rate": 6.8562000000000005e-06, "loss": 0.0125, "step": 15720 }, { "epoch": 0.8381404275737058, "grad_norm": 2.665299654006958, "learning_rate": 6.856e-06, "loss": 0.0495, "step": 15721 }, { "epoch": 0.8381937410033587, "grad_norm": 1.9666739702224731, "learning_rate": 6.8558000000000005e-06, "loss": 0.0389, "step": 15722 }, { "epoch": 0.8382470544330117, "grad_norm": 1.7468940019607544, "learning_rate": 6.8556e-06, "loss": 0.0426, "step": 15723 }, { "epoch": 0.8383003678626646, "grad_norm": 0.9101213812828064, "learning_rate": 6.855400000000001e-06, "loss": 0.0109, "step": 15724 }, { "epoch": 0.8383536812923176, "grad_norm": 2.0846633911132812, "learning_rate": 6.855200000000001e-06, "loss": 0.0206, "step": 15725 }, { "epoch": 0.8384069947219704, "grad_norm": 2.7496962547302246, "learning_rate": 6.8550000000000004e-06, "loss": 0.0347, "step": 15726 }, { "epoch": 0.8384603081516234, "grad_norm": 0.9909922480583191, "learning_rate": 6.8548e-06, "loss": 0.0236, "step": 15727 }, { "epoch": 0.8385136215812763, "grad_norm": 2.2129557132720947, "learning_rate": 6.8546e-06, "loss": 0.0283, "step": 15728 }, { "epoch": 0.8385669350109293, "grad_norm": 2.4628920555114746, "learning_rate": 6.854400000000001e-06, "loss": 0.0408, "step": 15729 }, { "epoch": 0.8386202484405821, "grad_norm": 2.813471555709839, "learning_rate": 6.854200000000001e-06, "loss": 0.0489, "step": 15730 }, { "epoch": 0.8386735618702351, "grad_norm": 3.7462949752807617, "learning_rate": 6.854000000000001e-06, "loss": 0.0207, "step": 15731 }, { "epoch": 0.8387268752998881, "grad_norm": 2.5436129570007324, "learning_rate": 6.8538e-06, "loss": 0.021, "step": 15732 }, { "epoch": 0.838780188729541, "grad_norm": 1.6103683710098267, "learning_rate": 6.8536e-06, "loss": 0.0274, "step": 15733 }, { "epoch": 0.8388335021591939, "grad_norm": 2.2361090183258057, "learning_rate": 6.8534e-06, "loss": 0.0573, "step": 15734 }, { "epoch": 0.8388868155888468, "grad_norm": 1.14011549949646, "learning_rate": 6.853200000000001e-06, "loss": 0.0273, "step": 15735 }, { "epoch": 0.8389401290184998, "grad_norm": 1.5391013622283936, "learning_rate": 6.853000000000001e-06, "loss": 0.0169, "step": 15736 }, { "epoch": 0.8389934424481527, "grad_norm": 3.863741874694824, "learning_rate": 6.852800000000001e-06, "loss": 0.0385, "step": 15737 }, { "epoch": 0.8390467558778056, "grad_norm": 3.4126057624816895, "learning_rate": 6.8526e-06, "loss": 0.0402, "step": 15738 }, { "epoch": 0.8391000693074585, "grad_norm": 1.3672702312469482, "learning_rate": 6.8524e-06, "loss": 0.0149, "step": 15739 }, { "epoch": 0.8391533827371115, "grad_norm": 1.6179691553115845, "learning_rate": 6.852200000000001e-06, "loss": 0.018, "step": 15740 }, { "epoch": 0.8392066961667644, "grad_norm": 1.0222288370132446, "learning_rate": 6.852000000000001e-06, "loss": 0.0126, "step": 15741 }, { "epoch": 0.8392600095964173, "grad_norm": 0.34794068336486816, "learning_rate": 6.851800000000001e-06, "loss": 0.0033, "step": 15742 }, { "epoch": 0.8393133230260703, "grad_norm": 1.0859241485595703, "learning_rate": 6.851600000000001e-06, "loss": 0.0087, "step": 15743 }, { "epoch": 0.8393666364557232, "grad_norm": 1.4333359003067017, "learning_rate": 6.8514e-06, "loss": 0.0207, "step": 15744 }, { "epoch": 0.8394199498853762, "grad_norm": 2.7868778705596924, "learning_rate": 6.8512e-06, "loss": 0.037, "step": 15745 }, { "epoch": 0.839473263315029, "grad_norm": 2.0560150146484375, "learning_rate": 6.851000000000001e-06, "loss": 0.0403, "step": 15746 }, { "epoch": 0.839526576744682, "grad_norm": 1.4280898571014404, "learning_rate": 6.8508000000000005e-06, "loss": 0.0157, "step": 15747 }, { "epoch": 0.8395798901743349, "grad_norm": 3.354135513305664, "learning_rate": 6.850600000000001e-06, "loss": 0.0313, "step": 15748 }, { "epoch": 0.8396332036039879, "grad_norm": 1.8505946397781372, "learning_rate": 6.8504000000000005e-06, "loss": 0.0279, "step": 15749 }, { "epoch": 0.8396865170336407, "grad_norm": 2.646968126296997, "learning_rate": 6.8502e-06, "loss": 0.0262, "step": 15750 }, { "epoch": 0.8397398304632937, "grad_norm": 1.329837441444397, "learning_rate": 6.850000000000001e-06, "loss": 0.0173, "step": 15751 }, { "epoch": 0.8397931438929467, "grad_norm": 1.052917718887329, "learning_rate": 6.849800000000001e-06, "loss": 0.0103, "step": 15752 }, { "epoch": 0.8398464573225995, "grad_norm": 2.093410015106201, "learning_rate": 6.8496000000000005e-06, "loss": 0.0311, "step": 15753 }, { "epoch": 0.8398997707522525, "grad_norm": 2.400517463684082, "learning_rate": 6.849400000000001e-06, "loss": 0.0513, "step": 15754 }, { "epoch": 0.8399530841819054, "grad_norm": 1.8078111410140991, "learning_rate": 6.8492000000000004e-06, "loss": 0.0379, "step": 15755 }, { "epoch": 0.8400063976115584, "grad_norm": 5.157608509063721, "learning_rate": 6.849e-06, "loss": 0.0618, "step": 15756 }, { "epoch": 0.8400597110412112, "grad_norm": 2.5888888835906982, "learning_rate": 6.848800000000001e-06, "loss": 0.014, "step": 15757 }, { "epoch": 0.8401130244708642, "grad_norm": 3.254236936569214, "learning_rate": 6.848600000000001e-06, "loss": 0.0307, "step": 15758 }, { "epoch": 0.8401663379005171, "grad_norm": 1.7332454919815063, "learning_rate": 6.8484e-06, "loss": 0.0245, "step": 15759 }, { "epoch": 0.8402196513301701, "grad_norm": 3.068915367126465, "learning_rate": 6.848200000000001e-06, "loss": 0.0329, "step": 15760 }, { "epoch": 0.840272964759823, "grad_norm": 1.7637728452682495, "learning_rate": 6.848e-06, "loss": 0.0272, "step": 15761 }, { "epoch": 0.8403262781894759, "grad_norm": 0.9138585329055786, "learning_rate": 6.847800000000001e-06, "loss": 0.021, "step": 15762 }, { "epoch": 0.8403795916191289, "grad_norm": 1.2217469215393066, "learning_rate": 6.847600000000001e-06, "loss": 0.0142, "step": 15763 }, { "epoch": 0.8404329050487818, "grad_norm": 5.5184197425842285, "learning_rate": 6.847400000000001e-06, "loss": 0.0458, "step": 15764 }, { "epoch": 0.8404862184784347, "grad_norm": 2.2527382373809814, "learning_rate": 6.8472e-06, "loss": 0.0253, "step": 15765 }, { "epoch": 0.8405395319080876, "grad_norm": 2.4927546977996826, "learning_rate": 6.847000000000001e-06, "loss": 0.0509, "step": 15766 }, { "epoch": 0.8405928453377406, "grad_norm": 2.6127660274505615, "learning_rate": 6.8468e-06, "loss": 0.0422, "step": 15767 }, { "epoch": 0.8406461587673935, "grad_norm": 1.819867491722107, "learning_rate": 6.846600000000001e-06, "loss": 0.0298, "step": 15768 }, { "epoch": 0.8406994721970464, "grad_norm": 1.0474807024002075, "learning_rate": 6.846400000000001e-06, "loss": 0.0219, "step": 15769 }, { "epoch": 0.8407527856266994, "grad_norm": 1.6416990756988525, "learning_rate": 6.846200000000001e-06, "loss": 0.0184, "step": 15770 }, { "epoch": 0.8408060990563523, "grad_norm": 1.6664602756500244, "learning_rate": 6.846e-06, "loss": 0.0393, "step": 15771 }, { "epoch": 0.8408594124860053, "grad_norm": 1.9355114698410034, "learning_rate": 6.845800000000001e-06, "loss": 0.0365, "step": 15772 }, { "epoch": 0.8409127259156581, "grad_norm": 1.5741796493530273, "learning_rate": 6.8456e-06, "loss": 0.034, "step": 15773 }, { "epoch": 0.8409660393453111, "grad_norm": 2.5223355293273926, "learning_rate": 6.8454000000000006e-06, "loss": 0.0428, "step": 15774 }, { "epoch": 0.841019352774964, "grad_norm": 2.5274083614349365, "learning_rate": 6.845200000000001e-06, "loss": 0.0784, "step": 15775 }, { "epoch": 0.841072666204617, "grad_norm": 1.1328065395355225, "learning_rate": 6.8450000000000005e-06, "loss": 0.0138, "step": 15776 }, { "epoch": 0.8411259796342698, "grad_norm": 2.1384570598602295, "learning_rate": 6.8448e-06, "loss": 0.0162, "step": 15777 }, { "epoch": 0.8411792930639228, "grad_norm": 3.0611448287963867, "learning_rate": 6.8446000000000005e-06, "loss": 0.0523, "step": 15778 }, { "epoch": 0.8412326064935758, "grad_norm": 1.9967957735061646, "learning_rate": 6.844400000000001e-06, "loss": 0.0211, "step": 15779 }, { "epoch": 0.8412859199232287, "grad_norm": 2.9763965606689453, "learning_rate": 6.8442000000000005e-06, "loss": 0.0543, "step": 15780 }, { "epoch": 0.8413392333528816, "grad_norm": 3.442204713821411, "learning_rate": 6.844000000000001e-06, "loss": 0.0285, "step": 15781 }, { "epoch": 0.8413925467825345, "grad_norm": 3.577068567276001, "learning_rate": 6.8438000000000004e-06, "loss": 0.0585, "step": 15782 }, { "epoch": 0.8414458602121875, "grad_norm": 3.478464126586914, "learning_rate": 6.8436e-06, "loss": 0.0262, "step": 15783 }, { "epoch": 0.8414991736418403, "grad_norm": 2.334726572036743, "learning_rate": 6.8434e-06, "loss": 0.02, "step": 15784 }, { "epoch": 0.8415524870714933, "grad_norm": 2.1964099407196045, "learning_rate": 6.843200000000001e-06, "loss": 0.0197, "step": 15785 }, { "epoch": 0.8416058005011462, "grad_norm": 2.213304042816162, "learning_rate": 6.843e-06, "loss": 0.0291, "step": 15786 }, { "epoch": 0.8416591139307992, "grad_norm": 2.2783751487731934, "learning_rate": 6.842800000000001e-06, "loss": 0.0512, "step": 15787 }, { "epoch": 0.8417124273604522, "grad_norm": 1.2556673288345337, "learning_rate": 6.8426e-06, "loss": 0.0192, "step": 15788 }, { "epoch": 0.841765740790105, "grad_norm": 2.7296879291534424, "learning_rate": 6.8424e-06, "loss": 0.0272, "step": 15789 }, { "epoch": 0.841819054219758, "grad_norm": 3.458662509918213, "learning_rate": 6.842200000000001e-06, "loss": 0.0581, "step": 15790 }, { "epoch": 0.8418723676494109, "grad_norm": 1.939943790435791, "learning_rate": 6.842000000000001e-06, "loss": 0.0342, "step": 15791 }, { "epoch": 0.8419256810790638, "grad_norm": 1.3121882677078247, "learning_rate": 6.8418e-06, "loss": 0.0272, "step": 15792 }, { "epoch": 0.8419789945087167, "grad_norm": 1.5331884622573853, "learning_rate": 6.841600000000001e-06, "loss": 0.0267, "step": 15793 }, { "epoch": 0.8420323079383697, "grad_norm": 2.1802520751953125, "learning_rate": 6.8414e-06, "loss": 0.0486, "step": 15794 }, { "epoch": 0.8420856213680226, "grad_norm": 1.6986762285232544, "learning_rate": 6.8412e-06, "loss": 0.0299, "step": 15795 }, { "epoch": 0.8421389347976755, "grad_norm": 2.831934690475464, "learning_rate": 6.841000000000001e-06, "loss": 0.0224, "step": 15796 }, { "epoch": 0.8421922482273284, "grad_norm": 2.074711561203003, "learning_rate": 6.840800000000001e-06, "loss": 0.0423, "step": 15797 }, { "epoch": 0.8422455616569814, "grad_norm": 4.13038444519043, "learning_rate": 6.8406e-06, "loss": 0.0409, "step": 15798 }, { "epoch": 0.8422988750866344, "grad_norm": 3.5861027240753174, "learning_rate": 6.840400000000001e-06, "loss": 0.0337, "step": 15799 }, { "epoch": 0.8423521885162872, "grad_norm": 1.4421093463897705, "learning_rate": 6.8402e-06, "loss": 0.0121, "step": 15800 }, { "epoch": 0.8424055019459402, "grad_norm": 1.8250725269317627, "learning_rate": 6.8400000000000014e-06, "loss": 0.0226, "step": 15801 }, { "epoch": 0.8424588153755931, "grad_norm": 1.987243413925171, "learning_rate": 6.839800000000001e-06, "loss": 0.0222, "step": 15802 }, { "epoch": 0.8425121288052461, "grad_norm": 2.3755900859832764, "learning_rate": 6.8396000000000005e-06, "loss": 0.0302, "step": 15803 }, { "epoch": 0.8425654422348989, "grad_norm": 1.9232038259506226, "learning_rate": 6.8394e-06, "loss": 0.0341, "step": 15804 }, { "epoch": 0.8426187556645519, "grad_norm": 3.680973768234253, "learning_rate": 6.8392000000000005e-06, "loss": 0.0775, "step": 15805 }, { "epoch": 0.8426720690942048, "grad_norm": 2.219320058822632, "learning_rate": 6.839e-06, "loss": 0.0434, "step": 15806 }, { "epoch": 0.8427253825238578, "grad_norm": 1.9460299015045166, "learning_rate": 6.838800000000001e-06, "loss": 0.0289, "step": 15807 }, { "epoch": 0.8427786959535107, "grad_norm": 1.5400747060775757, "learning_rate": 6.838600000000001e-06, "loss": 0.0207, "step": 15808 }, { "epoch": 0.8428320093831636, "grad_norm": 2.587944269180298, "learning_rate": 6.8384000000000005e-06, "loss": 0.0275, "step": 15809 }, { "epoch": 0.8428853228128166, "grad_norm": 1.200554609298706, "learning_rate": 6.8382e-06, "loss": 0.0153, "step": 15810 }, { "epoch": 0.8429386362424695, "grad_norm": 1.7482032775878906, "learning_rate": 6.8380000000000004e-06, "loss": 0.0277, "step": 15811 }, { "epoch": 0.8429919496721224, "grad_norm": 3.9541335105895996, "learning_rate": 6.837800000000001e-06, "loss": 0.0325, "step": 15812 }, { "epoch": 0.8430452631017753, "grad_norm": 1.9145379066467285, "learning_rate": 6.837600000000001e-06, "loss": 0.0308, "step": 15813 }, { "epoch": 0.8430985765314283, "grad_norm": 3.3070075511932373, "learning_rate": 6.837400000000001e-06, "loss": 0.0299, "step": 15814 }, { "epoch": 0.8431518899610811, "grad_norm": 4.630049228668213, "learning_rate": 6.8372e-06, "loss": 0.0417, "step": 15815 }, { "epoch": 0.8432052033907341, "grad_norm": 0.27882739901542664, "learning_rate": 6.837e-06, "loss": 0.003, "step": 15816 }, { "epoch": 0.8432585168203871, "grad_norm": 1.837160348892212, "learning_rate": 6.8368e-06, "loss": 0.0224, "step": 15817 }, { "epoch": 0.84331183025004, "grad_norm": 1.2927533388137817, "learning_rate": 6.836600000000001e-06, "loss": 0.0381, "step": 15818 }, { "epoch": 0.843365143679693, "grad_norm": 2.6508336067199707, "learning_rate": 6.836400000000001e-06, "loss": 0.0428, "step": 15819 }, { "epoch": 0.8434184571093458, "grad_norm": 0.8324928879737854, "learning_rate": 6.836200000000001e-06, "loss": 0.0092, "step": 15820 }, { "epoch": 0.8434717705389988, "grad_norm": 1.6838107109069824, "learning_rate": 6.836e-06, "loss": 0.018, "step": 15821 }, { "epoch": 0.8435250839686517, "grad_norm": 1.1508018970489502, "learning_rate": 6.8358e-06, "loss": 0.0286, "step": 15822 }, { "epoch": 0.8435783973983046, "grad_norm": 1.96743643283844, "learning_rate": 6.8356e-06, "loss": 0.0352, "step": 15823 }, { "epoch": 0.8436317108279575, "grad_norm": 1.5947247743606567, "learning_rate": 6.835400000000001e-06, "loss": 0.0366, "step": 15824 }, { "epoch": 0.8436850242576105, "grad_norm": 1.8994728326797485, "learning_rate": 6.835200000000001e-06, "loss": 0.0289, "step": 15825 }, { "epoch": 0.8437383376872635, "grad_norm": 2.1540634632110596, "learning_rate": 6.835000000000001e-06, "loss": 0.0536, "step": 15826 }, { "epoch": 0.8437916511169163, "grad_norm": 2.4327967166900635, "learning_rate": 6.8348e-06, "loss": 0.0592, "step": 15827 }, { "epoch": 0.8438449645465693, "grad_norm": 2.7360599040985107, "learning_rate": 6.8346e-06, "loss": 0.0264, "step": 15828 }, { "epoch": 0.8438982779762222, "grad_norm": 1.7875815629959106, "learning_rate": 6.834400000000001e-06, "loss": 0.0175, "step": 15829 }, { "epoch": 0.8439515914058752, "grad_norm": 1.8819228410720825, "learning_rate": 6.8342000000000006e-06, "loss": 0.0252, "step": 15830 }, { "epoch": 0.844004904835528, "grad_norm": 1.2505881786346436, "learning_rate": 6.834000000000001e-06, "loss": 0.0184, "step": 15831 }, { "epoch": 0.844058218265181, "grad_norm": 1.561367392539978, "learning_rate": 6.8338000000000005e-06, "loss": 0.0194, "step": 15832 }, { "epoch": 0.8441115316948339, "grad_norm": 2.9116077423095703, "learning_rate": 6.8336e-06, "loss": 0.0288, "step": 15833 }, { "epoch": 0.8441648451244869, "grad_norm": 1.410327672958374, "learning_rate": 6.8334e-06, "loss": 0.012, "step": 15834 }, { "epoch": 0.8442181585541398, "grad_norm": 1.01942777633667, "learning_rate": 6.833200000000001e-06, "loss": 0.0251, "step": 15835 }, { "epoch": 0.8442714719837927, "grad_norm": 1.4944000244140625, "learning_rate": 6.8330000000000005e-06, "loss": 0.0332, "step": 15836 }, { "epoch": 0.8443247854134457, "grad_norm": 0.87083500623703, "learning_rate": 6.832800000000001e-06, "loss": 0.0093, "step": 15837 }, { "epoch": 0.8443780988430986, "grad_norm": 1.373732328414917, "learning_rate": 6.8326000000000004e-06, "loss": 0.0203, "step": 15838 }, { "epoch": 0.8444314122727515, "grad_norm": 0.950925350189209, "learning_rate": 6.8324e-06, "loss": 0.0118, "step": 15839 }, { "epoch": 0.8444847257024044, "grad_norm": 0.6190917491912842, "learning_rate": 6.832200000000001e-06, "loss": 0.0067, "step": 15840 }, { "epoch": 0.8445380391320574, "grad_norm": 1.490370512008667, "learning_rate": 6.832000000000001e-06, "loss": 0.0315, "step": 15841 }, { "epoch": 0.8445913525617103, "grad_norm": 3.3831520080566406, "learning_rate": 6.8318e-06, "loss": 0.0363, "step": 15842 }, { "epoch": 0.8446446659913632, "grad_norm": 2.3545517921447754, "learning_rate": 6.831600000000001e-06, "loss": 0.0539, "step": 15843 }, { "epoch": 0.8446979794210161, "grad_norm": 2.3120691776275635, "learning_rate": 6.8314e-06, "loss": 0.0448, "step": 15844 }, { "epoch": 0.8447512928506691, "grad_norm": 1.2003196477890015, "learning_rate": 6.8312e-06, "loss": 0.011, "step": 15845 }, { "epoch": 0.8448046062803221, "grad_norm": 2.5215208530426025, "learning_rate": 6.831000000000001e-06, "loss": 0.0247, "step": 15846 }, { "epoch": 0.8448579197099749, "grad_norm": 0.34141045808792114, "learning_rate": 6.830800000000001e-06, "loss": 0.0036, "step": 15847 }, { "epoch": 0.8449112331396279, "grad_norm": 2.3317971229553223, "learning_rate": 6.8306e-06, "loss": 0.027, "step": 15848 }, { "epoch": 0.8449645465692808, "grad_norm": 1.1309090852737427, "learning_rate": 6.830400000000001e-06, "loss": 0.0118, "step": 15849 }, { "epoch": 0.8450178599989338, "grad_norm": 6.4969563484191895, "learning_rate": 6.8302e-06, "loss": 0.0305, "step": 15850 }, { "epoch": 0.8450711734285866, "grad_norm": 2.8128044605255127, "learning_rate": 6.830000000000001e-06, "loss": 0.0327, "step": 15851 }, { "epoch": 0.8451244868582396, "grad_norm": 2.6625945568084717, "learning_rate": 6.829800000000001e-06, "loss": 0.0549, "step": 15852 }, { "epoch": 0.8451778002878925, "grad_norm": 0.813269853591919, "learning_rate": 6.829600000000001e-06, "loss": 0.0074, "step": 15853 }, { "epoch": 0.8452311137175454, "grad_norm": 0.8876663446426392, "learning_rate": 6.8294e-06, "loss": 0.0088, "step": 15854 }, { "epoch": 0.8452844271471984, "grad_norm": 2.6616358757019043, "learning_rate": 6.829200000000001e-06, "loss": 0.0181, "step": 15855 }, { "epoch": 0.8453377405768513, "grad_norm": 2.265313148498535, "learning_rate": 6.829e-06, "loss": 0.0464, "step": 15856 }, { "epoch": 0.8453910540065043, "grad_norm": 1.3819605112075806, "learning_rate": 6.828800000000001e-06, "loss": 0.0173, "step": 15857 }, { "epoch": 0.8454443674361571, "grad_norm": 2.054537534713745, "learning_rate": 6.828600000000001e-06, "loss": 0.0233, "step": 15858 }, { "epoch": 0.8454976808658101, "grad_norm": 3.2405829429626465, "learning_rate": 6.8284000000000006e-06, "loss": 0.0379, "step": 15859 }, { "epoch": 0.845550994295463, "grad_norm": 1.8628931045532227, "learning_rate": 6.8282e-06, "loss": 0.0127, "step": 15860 }, { "epoch": 0.845604307725116, "grad_norm": 1.7974390983581543, "learning_rate": 6.8280000000000005e-06, "loss": 0.016, "step": 15861 }, { "epoch": 0.8456576211547688, "grad_norm": 1.9114201068878174, "learning_rate": 6.827800000000001e-06, "loss": 0.0156, "step": 15862 }, { "epoch": 0.8457109345844218, "grad_norm": 2.257723331451416, "learning_rate": 6.8276000000000005e-06, "loss": 0.0223, "step": 15863 }, { "epoch": 0.8457642480140748, "grad_norm": 2.319601058959961, "learning_rate": 6.827400000000001e-06, "loss": 0.026, "step": 15864 }, { "epoch": 0.8458175614437277, "grad_norm": 2.853656768798828, "learning_rate": 6.8272000000000005e-06, "loss": 0.0346, "step": 15865 }, { "epoch": 0.8458708748733806, "grad_norm": 1.3250598907470703, "learning_rate": 6.827e-06, "loss": 0.0288, "step": 15866 }, { "epoch": 0.8459241883030335, "grad_norm": 2.417268753051758, "learning_rate": 6.8268000000000004e-06, "loss": 0.0281, "step": 15867 }, { "epoch": 0.8459775017326865, "grad_norm": 2.8479979038238525, "learning_rate": 6.826600000000001e-06, "loss": 0.0168, "step": 15868 }, { "epoch": 0.8460308151623394, "grad_norm": 2.616269588470459, "learning_rate": 6.8264e-06, "loss": 0.0605, "step": 15869 }, { "epoch": 0.8460841285919923, "grad_norm": 2.1007001399993896, "learning_rate": 6.826200000000001e-06, "loss": 0.0178, "step": 15870 }, { "epoch": 0.8461374420216452, "grad_norm": 4.299061298370361, "learning_rate": 6.826e-06, "loss": 0.0262, "step": 15871 }, { "epoch": 0.8461907554512982, "grad_norm": 2.012566328048706, "learning_rate": 6.8258e-06, "loss": 0.0129, "step": 15872 }, { "epoch": 0.8462440688809512, "grad_norm": 2.342409610748291, "learning_rate": 6.8256e-06, "loss": 0.0363, "step": 15873 }, { "epoch": 0.846297382310604, "grad_norm": 2.6803951263427734, "learning_rate": 6.825400000000001e-06, "loss": 0.069, "step": 15874 }, { "epoch": 0.846350695740257, "grad_norm": 1.3926833868026733, "learning_rate": 6.8252e-06, "loss": 0.0279, "step": 15875 }, { "epoch": 0.8464040091699099, "grad_norm": 2.3628945350646973, "learning_rate": 6.825000000000001e-06, "loss": 0.0238, "step": 15876 }, { "epoch": 0.8464573225995629, "grad_norm": 3.3730759620666504, "learning_rate": 6.8248e-06, "loss": 0.0297, "step": 15877 }, { "epoch": 0.8465106360292157, "grad_norm": 1.621455192565918, "learning_rate": 6.8246e-06, "loss": 0.0174, "step": 15878 }, { "epoch": 0.8465639494588687, "grad_norm": 1.5201750993728638, "learning_rate": 6.824400000000001e-06, "loss": 0.0253, "step": 15879 }, { "epoch": 0.8466172628885216, "grad_norm": 1.5493743419647217, "learning_rate": 6.824200000000001e-06, "loss": 0.0212, "step": 15880 }, { "epoch": 0.8466705763181746, "grad_norm": 3.707501173019409, "learning_rate": 6.824e-06, "loss": 0.033, "step": 15881 }, { "epoch": 0.8467238897478275, "grad_norm": 2.3733389377593994, "learning_rate": 6.823800000000001e-06, "loss": 0.037, "step": 15882 }, { "epoch": 0.8467772031774804, "grad_norm": 1.9996737241744995, "learning_rate": 6.8236e-06, "loss": 0.0193, "step": 15883 }, { "epoch": 0.8468305166071334, "grad_norm": 2.0101091861724854, "learning_rate": 6.8234e-06, "loss": 0.0282, "step": 15884 }, { "epoch": 0.8468838300367862, "grad_norm": 1.5461087226867676, "learning_rate": 6.823200000000001e-06, "loss": 0.0212, "step": 15885 }, { "epoch": 0.8469371434664392, "grad_norm": 1.3107378482818604, "learning_rate": 6.8230000000000006e-06, "loss": 0.025, "step": 15886 }, { "epoch": 0.8469904568960921, "grad_norm": 2.5837149620056152, "learning_rate": 6.8228e-06, "loss": 0.0521, "step": 15887 }, { "epoch": 0.8470437703257451, "grad_norm": 1.1884390115737915, "learning_rate": 6.8226000000000005e-06, "loss": 0.018, "step": 15888 }, { "epoch": 0.8470970837553979, "grad_norm": 3.5143537521362305, "learning_rate": 6.8224e-06, "loss": 0.0552, "step": 15889 }, { "epoch": 0.8471503971850509, "grad_norm": 2.005898952484131, "learning_rate": 6.822200000000001e-06, "loss": 0.0403, "step": 15890 }, { "epoch": 0.8472037106147039, "grad_norm": 1.8402721881866455, "learning_rate": 6.822000000000001e-06, "loss": 0.0198, "step": 15891 }, { "epoch": 0.8472570240443568, "grad_norm": 1.8151687383651733, "learning_rate": 6.8218000000000005e-06, "loss": 0.0177, "step": 15892 }, { "epoch": 0.8473103374740097, "grad_norm": 1.7287929058074951, "learning_rate": 6.8216e-06, "loss": 0.0256, "step": 15893 }, { "epoch": 0.8473636509036626, "grad_norm": 3.8992156982421875, "learning_rate": 6.8214000000000005e-06, "loss": 0.043, "step": 15894 }, { "epoch": 0.8474169643333156, "grad_norm": 2.1161367893218994, "learning_rate": 6.8212e-06, "loss": 0.0333, "step": 15895 }, { "epoch": 0.8474702777629685, "grad_norm": 2.4897122383117676, "learning_rate": 6.821000000000001e-06, "loss": 0.0303, "step": 15896 }, { "epoch": 0.8475235911926214, "grad_norm": 1.6189478635787964, "learning_rate": 6.820800000000001e-06, "loss": 0.0239, "step": 15897 }, { "epoch": 0.8475769046222743, "grad_norm": 2.5196235179901123, "learning_rate": 6.8206e-06, "loss": 0.0444, "step": 15898 }, { "epoch": 0.8476302180519273, "grad_norm": 1.2657865285873413, "learning_rate": 6.8204e-06, "loss": 0.0258, "step": 15899 }, { "epoch": 0.8476835314815802, "grad_norm": 2.4897360801696777, "learning_rate": 6.8202e-06, "loss": 0.0217, "step": 15900 }, { "epoch": 0.8477368449112331, "grad_norm": 1.2811073064804077, "learning_rate": 6.820000000000001e-06, "loss": 0.0162, "step": 15901 }, { "epoch": 0.8477901583408861, "grad_norm": 0.9887682795524597, "learning_rate": 6.819800000000001e-06, "loss": 0.015, "step": 15902 }, { "epoch": 0.847843471770539, "grad_norm": 1.3854840993881226, "learning_rate": 6.819600000000001e-06, "loss": 0.035, "step": 15903 }, { "epoch": 0.847896785200192, "grad_norm": 1.6077890396118164, "learning_rate": 6.8194e-06, "loss": 0.0284, "step": 15904 }, { "epoch": 0.8479500986298448, "grad_norm": 0.8817199468612671, "learning_rate": 6.8192e-06, "loss": 0.0095, "step": 15905 }, { "epoch": 0.8480034120594978, "grad_norm": 0.9758272767066956, "learning_rate": 6.819e-06, "loss": 0.0218, "step": 15906 }, { "epoch": 0.8480567254891507, "grad_norm": 2.386276960372925, "learning_rate": 6.818800000000001e-06, "loss": 0.0406, "step": 15907 }, { "epoch": 0.8481100389188037, "grad_norm": 2.543545961380005, "learning_rate": 6.818600000000001e-06, "loss": 0.0539, "step": 15908 }, { "epoch": 0.8481633523484565, "grad_norm": 2.1071300506591797, "learning_rate": 6.818400000000001e-06, "loss": 0.025, "step": 15909 }, { "epoch": 0.8482166657781095, "grad_norm": 1.6718218326568604, "learning_rate": 6.8182e-06, "loss": 0.0326, "step": 15910 }, { "epoch": 0.8482699792077625, "grad_norm": 1.2130101919174194, "learning_rate": 6.818e-06, "loss": 0.0145, "step": 15911 }, { "epoch": 0.8483232926374153, "grad_norm": 2.650615930557251, "learning_rate": 6.817800000000001e-06, "loss": 0.0236, "step": 15912 }, { "epoch": 0.8483766060670683, "grad_norm": 1.9009777307510376, "learning_rate": 6.817600000000001e-06, "loss": 0.0395, "step": 15913 }, { "epoch": 0.8484299194967212, "grad_norm": 1.2187249660491943, "learning_rate": 6.817400000000001e-06, "loss": 0.0216, "step": 15914 }, { "epoch": 0.8484832329263742, "grad_norm": 1.5826365947723389, "learning_rate": 6.8172000000000006e-06, "loss": 0.0355, "step": 15915 }, { "epoch": 0.848536546356027, "grad_norm": 2.7103893756866455, "learning_rate": 6.817e-06, "loss": 0.0566, "step": 15916 }, { "epoch": 0.84858985978568, "grad_norm": 2.088327169418335, "learning_rate": 6.8168e-06, "loss": 0.029, "step": 15917 }, { "epoch": 0.8486431732153329, "grad_norm": 2.4496734142303467, "learning_rate": 6.816600000000001e-06, "loss": 0.0489, "step": 15918 }, { "epoch": 0.8486964866449859, "grad_norm": 2.312372922897339, "learning_rate": 6.8164000000000005e-06, "loss": 0.0339, "step": 15919 }, { "epoch": 0.8487498000746388, "grad_norm": 1.5423694849014282, "learning_rate": 6.816200000000001e-06, "loss": 0.0253, "step": 15920 }, { "epoch": 0.8488031135042917, "grad_norm": 2.966914415359497, "learning_rate": 6.8160000000000005e-06, "loss": 0.0243, "step": 15921 }, { "epoch": 0.8488564269339447, "grad_norm": 0.9479120373725891, "learning_rate": 6.8158e-06, "loss": 0.0138, "step": 15922 }, { "epoch": 0.8489097403635976, "grad_norm": 2.6891133785247803, "learning_rate": 6.815600000000001e-06, "loss": 0.0479, "step": 15923 }, { "epoch": 0.8489630537932505, "grad_norm": 2.515418529510498, "learning_rate": 6.815400000000001e-06, "loss": 0.0538, "step": 15924 }, { "epoch": 0.8490163672229034, "grad_norm": 1.310443639755249, "learning_rate": 6.8152e-06, "loss": 0.0285, "step": 15925 }, { "epoch": 0.8490696806525564, "grad_norm": 2.3956785202026367, "learning_rate": 6.815000000000001e-06, "loss": 0.0419, "step": 15926 }, { "epoch": 0.8491229940822093, "grad_norm": 1.528356671333313, "learning_rate": 6.8148e-06, "loss": 0.0367, "step": 15927 }, { "epoch": 0.8491763075118622, "grad_norm": 2.357567310333252, "learning_rate": 6.8146e-06, "loss": 0.0573, "step": 15928 }, { "epoch": 0.8492296209415152, "grad_norm": 2.260560989379883, "learning_rate": 6.814400000000001e-06, "loss": 0.0462, "step": 15929 }, { "epoch": 0.8492829343711681, "grad_norm": 2.7610650062561035, "learning_rate": 6.814200000000001e-06, "loss": 0.0647, "step": 15930 }, { "epoch": 0.8493362478008211, "grad_norm": 1.6507290601730347, "learning_rate": 6.814e-06, "loss": 0.0314, "step": 15931 }, { "epoch": 0.8493895612304739, "grad_norm": 2.6239492893218994, "learning_rate": 6.813800000000001e-06, "loss": 0.0225, "step": 15932 }, { "epoch": 0.8494428746601269, "grad_norm": 3.019908905029297, "learning_rate": 6.8136e-06, "loss": 0.0276, "step": 15933 }, { "epoch": 0.8494961880897798, "grad_norm": 1.84519362449646, "learning_rate": 6.8134e-06, "loss": 0.0452, "step": 15934 }, { "epoch": 0.8495495015194328, "grad_norm": 2.2443747520446777, "learning_rate": 6.813200000000001e-06, "loss": 0.0373, "step": 15935 }, { "epoch": 0.8496028149490856, "grad_norm": 1.5004141330718994, "learning_rate": 6.813000000000001e-06, "loss": 0.0348, "step": 15936 }, { "epoch": 0.8496561283787386, "grad_norm": 2.0837433338165283, "learning_rate": 6.8128e-06, "loss": 0.0399, "step": 15937 }, { "epoch": 0.8497094418083916, "grad_norm": 4.110506534576416, "learning_rate": 6.812600000000001e-06, "loss": 0.0677, "step": 15938 }, { "epoch": 0.8497627552380445, "grad_norm": 1.5955933332443237, "learning_rate": 6.8124e-06, "loss": 0.0234, "step": 15939 }, { "epoch": 0.8498160686676974, "grad_norm": 2.1377041339874268, "learning_rate": 6.8122000000000015e-06, "loss": 0.0437, "step": 15940 }, { "epoch": 0.8498693820973503, "grad_norm": 1.1496392488479614, "learning_rate": 6.812000000000001e-06, "loss": 0.0368, "step": 15941 }, { "epoch": 0.8499226955270033, "grad_norm": 1.390843391418457, "learning_rate": 6.811800000000001e-06, "loss": 0.0203, "step": 15942 }, { "epoch": 0.8499760089566561, "grad_norm": 2.1495285034179688, "learning_rate": 6.8116e-06, "loss": 0.0445, "step": 15943 }, { "epoch": 0.8500293223863091, "grad_norm": 1.389811635017395, "learning_rate": 6.8114000000000005e-06, "loss": 0.0318, "step": 15944 }, { "epoch": 0.850082635815962, "grad_norm": 1.922883152961731, "learning_rate": 6.8112e-06, "loss": 0.0482, "step": 15945 }, { "epoch": 0.850135949245615, "grad_norm": 1.043891191482544, "learning_rate": 6.811000000000001e-06, "loss": 0.0341, "step": 15946 }, { "epoch": 0.850189262675268, "grad_norm": 1.3801465034484863, "learning_rate": 6.810800000000001e-06, "loss": 0.0255, "step": 15947 }, { "epoch": 0.8502425761049208, "grad_norm": 1.257172703742981, "learning_rate": 6.8106000000000005e-06, "loss": 0.036, "step": 15948 }, { "epoch": 0.8502958895345738, "grad_norm": 2.504009246826172, "learning_rate": 6.8104e-06, "loss": 0.0515, "step": 15949 }, { "epoch": 0.8503492029642267, "grad_norm": 2.667752504348755, "learning_rate": 6.8102000000000005e-06, "loss": 0.0339, "step": 15950 }, { "epoch": 0.8504025163938796, "grad_norm": 2.778172016143799, "learning_rate": 6.810000000000001e-06, "loss": 0.0178, "step": 15951 }, { "epoch": 0.8504558298235325, "grad_norm": 1.121280550956726, "learning_rate": 6.809800000000001e-06, "loss": 0.0132, "step": 15952 }, { "epoch": 0.8505091432531855, "grad_norm": 2.319805145263672, "learning_rate": 6.809600000000001e-06, "loss": 0.0463, "step": 15953 }, { "epoch": 0.8505624566828384, "grad_norm": 2.9909870624542236, "learning_rate": 6.8094e-06, "loss": 0.0317, "step": 15954 }, { "epoch": 0.8506157701124913, "grad_norm": 1.4130412340164185, "learning_rate": 6.8092e-06, "loss": 0.0132, "step": 15955 }, { "epoch": 0.8506690835421442, "grad_norm": 2.290099859237671, "learning_rate": 6.809e-06, "loss": 0.0399, "step": 15956 }, { "epoch": 0.8507223969717972, "grad_norm": 1.7558718919754028, "learning_rate": 6.808800000000001e-06, "loss": 0.0345, "step": 15957 }, { "epoch": 0.8507757104014502, "grad_norm": 1.321869134902954, "learning_rate": 6.808600000000001e-06, "loss": 0.0185, "step": 15958 }, { "epoch": 0.850829023831103, "grad_norm": 2.0631494522094727, "learning_rate": 6.808400000000001e-06, "loss": 0.0357, "step": 15959 }, { "epoch": 0.850882337260756, "grad_norm": 0.9530139565467834, "learning_rate": 6.8082e-06, "loss": 0.0276, "step": 15960 }, { "epoch": 0.8509356506904089, "grad_norm": 1.2625155448913574, "learning_rate": 6.808e-06, "loss": 0.0219, "step": 15961 }, { "epoch": 0.8509889641200619, "grad_norm": 1.836754560470581, "learning_rate": 6.807800000000001e-06, "loss": 0.0159, "step": 15962 }, { "epoch": 0.8510422775497147, "grad_norm": 2.039233684539795, "learning_rate": 6.807600000000001e-06, "loss": 0.0482, "step": 15963 }, { "epoch": 0.8510955909793677, "grad_norm": 1.9663496017456055, "learning_rate": 6.807400000000001e-06, "loss": 0.0166, "step": 15964 }, { "epoch": 0.8511489044090206, "grad_norm": 1.7266449928283691, "learning_rate": 6.807200000000001e-06, "loss": 0.0225, "step": 15965 }, { "epoch": 0.8512022178386736, "grad_norm": 2.5890419483184814, "learning_rate": 6.807e-06, "loss": 0.0465, "step": 15966 }, { "epoch": 0.8512555312683265, "grad_norm": 1.0696213245391846, "learning_rate": 6.8068e-06, "loss": 0.0162, "step": 15967 }, { "epoch": 0.8513088446979794, "grad_norm": 1.5409685373306274, "learning_rate": 6.806600000000001e-06, "loss": 0.044, "step": 15968 }, { "epoch": 0.8513621581276324, "grad_norm": 3.2418506145477295, "learning_rate": 6.806400000000001e-06, "loss": 0.0487, "step": 15969 }, { "epoch": 0.8514154715572853, "grad_norm": 1.777477741241455, "learning_rate": 6.806200000000001e-06, "loss": 0.0213, "step": 15970 }, { "epoch": 0.8514687849869382, "grad_norm": 3.0109426975250244, "learning_rate": 6.8060000000000006e-06, "loss": 0.0298, "step": 15971 }, { "epoch": 0.8515220984165911, "grad_norm": 1.6017111539840698, "learning_rate": 6.8058e-06, "loss": 0.0185, "step": 15972 }, { "epoch": 0.8515754118462441, "grad_norm": 1.8723759651184082, "learning_rate": 6.805600000000001e-06, "loss": 0.0461, "step": 15973 }, { "epoch": 0.851628725275897, "grad_norm": 1.6594054698944092, "learning_rate": 6.805400000000001e-06, "loss": 0.0325, "step": 15974 }, { "epoch": 0.8516820387055499, "grad_norm": 2.015057325363159, "learning_rate": 6.8052000000000005e-06, "loss": 0.0478, "step": 15975 }, { "epoch": 0.8517353521352029, "grad_norm": 1.0806289911270142, "learning_rate": 6.805000000000001e-06, "loss": 0.0123, "step": 15976 }, { "epoch": 0.8517886655648558, "grad_norm": 1.6089835166931152, "learning_rate": 6.8048000000000005e-06, "loss": 0.028, "step": 15977 }, { "epoch": 0.8518419789945088, "grad_norm": 1.4869534969329834, "learning_rate": 6.8046e-06, "loss": 0.0099, "step": 15978 }, { "epoch": 0.8518952924241616, "grad_norm": 4.804553508758545, "learning_rate": 6.804400000000001e-06, "loss": 0.0206, "step": 15979 }, { "epoch": 0.8519486058538146, "grad_norm": 1.3703815937042236, "learning_rate": 6.804200000000001e-06, "loss": 0.0175, "step": 15980 }, { "epoch": 0.8520019192834675, "grad_norm": 1.2872941493988037, "learning_rate": 6.804e-06, "loss": 0.0198, "step": 15981 }, { "epoch": 0.8520552327131204, "grad_norm": 2.940598249435425, "learning_rate": 6.803800000000001e-06, "loss": 0.0348, "step": 15982 }, { "epoch": 0.8521085461427733, "grad_norm": 1.411353588104248, "learning_rate": 6.8036e-06, "loss": 0.0229, "step": 15983 }, { "epoch": 0.8521618595724263, "grad_norm": 1.1595474481582642, "learning_rate": 6.8034e-06, "loss": 0.0245, "step": 15984 }, { "epoch": 0.8522151730020793, "grad_norm": 3.574775457382202, "learning_rate": 6.803200000000001e-06, "loss": 0.0402, "step": 15985 }, { "epoch": 0.8522684864317321, "grad_norm": 1.5109024047851562, "learning_rate": 6.803000000000001e-06, "loss": 0.0304, "step": 15986 }, { "epoch": 0.8523217998613851, "grad_norm": 3.491509437561035, "learning_rate": 6.8028e-06, "loss": 0.0533, "step": 15987 }, { "epoch": 0.852375113291038, "grad_norm": 4.386216163635254, "learning_rate": 6.802600000000001e-06, "loss": 0.1033, "step": 15988 }, { "epoch": 0.852428426720691, "grad_norm": 2.4779975414276123, "learning_rate": 6.8024e-06, "loss": 0.0335, "step": 15989 }, { "epoch": 0.8524817401503438, "grad_norm": 2.378965377807617, "learning_rate": 6.802200000000001e-06, "loss": 0.0214, "step": 15990 }, { "epoch": 0.8525350535799968, "grad_norm": 2.6018385887145996, "learning_rate": 6.802000000000001e-06, "loss": 0.0286, "step": 15991 }, { "epoch": 0.8525883670096497, "grad_norm": 2.268968343734741, "learning_rate": 6.801800000000001e-06, "loss": 0.0344, "step": 15992 }, { "epoch": 0.8526416804393027, "grad_norm": 1.8305801153182983, "learning_rate": 6.8016e-06, "loss": 0.0279, "step": 15993 }, { "epoch": 0.8526949938689556, "grad_norm": 2.098639965057373, "learning_rate": 6.801400000000001e-06, "loss": 0.0221, "step": 15994 }, { "epoch": 0.8527483072986085, "grad_norm": 1.4576199054718018, "learning_rate": 6.8012e-06, "loss": 0.0245, "step": 15995 }, { "epoch": 0.8528016207282615, "grad_norm": 1.2673835754394531, "learning_rate": 6.801000000000001e-06, "loss": 0.0184, "step": 15996 }, { "epoch": 0.8528549341579144, "grad_norm": 2.6978132724761963, "learning_rate": 6.800800000000001e-06, "loss": 0.0599, "step": 15997 }, { "epoch": 0.8529082475875673, "grad_norm": 2.0501556396484375, "learning_rate": 6.800600000000001e-06, "loss": 0.0333, "step": 15998 }, { "epoch": 0.8529615610172202, "grad_norm": 2.6854891777038574, "learning_rate": 6.8004e-06, "loss": 0.0582, "step": 15999 }, { "epoch": 0.8530148744468732, "grad_norm": 1.3781272172927856, "learning_rate": 6.8002000000000006e-06, "loss": 0.0254, "step": 16000 }, { "epoch": 0.853068187876526, "grad_norm": 1.4835882186889648, "learning_rate": 6.800000000000001e-06, "loss": 0.0115, "step": 16001 }, { "epoch": 0.853121501306179, "grad_norm": 2.2657294273376465, "learning_rate": 6.7998000000000005e-06, "loss": 0.0284, "step": 16002 }, { "epoch": 0.853174814735832, "grad_norm": 2.0171258449554443, "learning_rate": 6.799600000000001e-06, "loss": 0.0179, "step": 16003 }, { "epoch": 0.8532281281654849, "grad_norm": 1.7970280647277832, "learning_rate": 6.7994000000000005e-06, "loss": 0.0335, "step": 16004 }, { "epoch": 0.8532814415951379, "grad_norm": 1.8489713668823242, "learning_rate": 6.7992e-06, "loss": 0.0259, "step": 16005 }, { "epoch": 0.8533347550247907, "grad_norm": 0.6566013693809509, "learning_rate": 6.7990000000000005e-06, "loss": 0.0073, "step": 16006 }, { "epoch": 0.8533880684544437, "grad_norm": 2.3109827041625977, "learning_rate": 6.798800000000001e-06, "loss": 0.036, "step": 16007 }, { "epoch": 0.8534413818840966, "grad_norm": 1.794822096824646, "learning_rate": 6.7986000000000004e-06, "loss": 0.0476, "step": 16008 }, { "epoch": 0.8534946953137496, "grad_norm": 1.3045282363891602, "learning_rate": 6.798400000000001e-06, "loss": 0.0181, "step": 16009 }, { "epoch": 0.8535480087434024, "grad_norm": 1.574151873588562, "learning_rate": 6.7982e-06, "loss": 0.0335, "step": 16010 }, { "epoch": 0.8536013221730554, "grad_norm": 1.804013729095459, "learning_rate": 6.798e-06, "loss": 0.0206, "step": 16011 }, { "epoch": 0.8536546356027083, "grad_norm": 1.5508077144622803, "learning_rate": 6.797800000000001e-06, "loss": 0.0128, "step": 16012 }, { "epoch": 0.8537079490323612, "grad_norm": 0.944789707660675, "learning_rate": 6.797600000000001e-06, "loss": 0.017, "step": 16013 }, { "epoch": 0.8537612624620142, "grad_norm": 1.6006360054016113, "learning_rate": 6.7974e-06, "loss": 0.0135, "step": 16014 }, { "epoch": 0.8538145758916671, "grad_norm": 1.288464069366455, "learning_rate": 6.797200000000001e-06, "loss": 0.0126, "step": 16015 }, { "epoch": 0.8538678893213201, "grad_norm": 1.7594783306121826, "learning_rate": 6.797e-06, "loss": 0.018, "step": 16016 }, { "epoch": 0.8539212027509729, "grad_norm": 2.407331705093384, "learning_rate": 6.7968e-06, "loss": 0.0343, "step": 16017 }, { "epoch": 0.8539745161806259, "grad_norm": 2.007478713989258, "learning_rate": 6.796600000000001e-06, "loss": 0.0438, "step": 16018 }, { "epoch": 0.8540278296102788, "grad_norm": 0.9454593658447266, "learning_rate": 6.796400000000001e-06, "loss": 0.0102, "step": 16019 }, { "epoch": 0.8540811430399318, "grad_norm": 2.3922226428985596, "learning_rate": 6.7962e-06, "loss": 0.0228, "step": 16020 }, { "epoch": 0.8541344564695846, "grad_norm": 1.795520544052124, "learning_rate": 6.796000000000001e-06, "loss": 0.0184, "step": 16021 }, { "epoch": 0.8541877698992376, "grad_norm": 2.094972848892212, "learning_rate": 6.7958e-06, "loss": 0.052, "step": 16022 }, { "epoch": 0.8542410833288906, "grad_norm": 2.9582414627075195, "learning_rate": 6.7956000000000015e-06, "loss": 0.0225, "step": 16023 }, { "epoch": 0.8542943967585435, "grad_norm": 2.810847759246826, "learning_rate": 6.795400000000001e-06, "loss": 0.0622, "step": 16024 }, { "epoch": 0.8543477101881964, "grad_norm": 2.395197629928589, "learning_rate": 6.795200000000001e-06, "loss": 0.0207, "step": 16025 }, { "epoch": 0.8544010236178493, "grad_norm": 2.535957098007202, "learning_rate": 6.795e-06, "loss": 0.0366, "step": 16026 }, { "epoch": 0.8544543370475023, "grad_norm": 2.3175339698791504, "learning_rate": 6.7948000000000006e-06, "loss": 0.0205, "step": 16027 }, { "epoch": 0.8545076504771552, "grad_norm": 2.4715049266815186, "learning_rate": 6.7946e-06, "loss": 0.0259, "step": 16028 }, { "epoch": 0.8545609639068081, "grad_norm": 3.370694398880005, "learning_rate": 6.794400000000001e-06, "loss": 0.0206, "step": 16029 }, { "epoch": 0.854614277336461, "grad_norm": 1.710121989250183, "learning_rate": 6.794200000000001e-06, "loss": 0.029, "step": 16030 }, { "epoch": 0.854667590766114, "grad_norm": 4.151062488555908, "learning_rate": 6.7940000000000005e-06, "loss": 0.047, "step": 16031 }, { "epoch": 0.854720904195767, "grad_norm": 2.4723713397979736, "learning_rate": 6.7938e-06, "loss": 0.0429, "step": 16032 }, { "epoch": 0.8547742176254198, "grad_norm": 4.901519775390625, "learning_rate": 6.7936000000000005e-06, "loss": 0.0602, "step": 16033 }, { "epoch": 0.8548275310550728, "grad_norm": 1.4140247106552124, "learning_rate": 6.793400000000001e-06, "loss": 0.0123, "step": 16034 }, { "epoch": 0.8548808444847257, "grad_norm": 1.435311198234558, "learning_rate": 6.793200000000001e-06, "loss": 0.0129, "step": 16035 }, { "epoch": 0.8549341579143787, "grad_norm": 1.8175185918807983, "learning_rate": 6.793000000000001e-06, "loss": 0.0479, "step": 16036 }, { "epoch": 0.8549874713440315, "grad_norm": 1.6871956586837769, "learning_rate": 6.7928e-06, "loss": 0.0156, "step": 16037 }, { "epoch": 0.8550407847736845, "grad_norm": 2.257148027420044, "learning_rate": 6.7926e-06, "loss": 0.028, "step": 16038 }, { "epoch": 0.8550940982033374, "grad_norm": 2.29954195022583, "learning_rate": 6.7924e-06, "loss": 0.0371, "step": 16039 }, { "epoch": 0.8551474116329904, "grad_norm": 2.343181848526001, "learning_rate": 6.792200000000001e-06, "loss": 0.0473, "step": 16040 }, { "epoch": 0.8552007250626433, "grad_norm": 1.9067723751068115, "learning_rate": 6.792000000000001e-06, "loss": 0.0159, "step": 16041 }, { "epoch": 0.8552540384922962, "grad_norm": 1.738572120666504, "learning_rate": 6.791800000000001e-06, "loss": 0.021, "step": 16042 }, { "epoch": 0.8553073519219492, "grad_norm": 2.888669490814209, "learning_rate": 6.7916e-06, "loss": 0.0533, "step": 16043 }, { "epoch": 0.855360665351602, "grad_norm": 1.8335496187210083, "learning_rate": 6.7914e-06, "loss": 0.0416, "step": 16044 }, { "epoch": 0.855413978781255, "grad_norm": 1.2020514011383057, "learning_rate": 6.7912e-06, "loss": 0.0156, "step": 16045 }, { "epoch": 0.8554672922109079, "grad_norm": 1.0035731792449951, "learning_rate": 6.791000000000001e-06, "loss": 0.0096, "step": 16046 }, { "epoch": 0.8555206056405609, "grad_norm": 2.8623173236846924, "learning_rate": 6.790800000000001e-06, "loss": 0.0436, "step": 16047 }, { "epoch": 0.8555739190702137, "grad_norm": 1.3064614534378052, "learning_rate": 6.790600000000001e-06, "loss": 0.0301, "step": 16048 }, { "epoch": 0.8556272324998667, "grad_norm": 2.648125648498535, "learning_rate": 6.7904e-06, "loss": 0.0277, "step": 16049 }, { "epoch": 0.8556805459295197, "grad_norm": 0.9213162660598755, "learning_rate": 6.7902e-06, "loss": 0.018, "step": 16050 }, { "epoch": 0.8557338593591726, "grad_norm": 2.9875195026397705, "learning_rate": 6.790000000000001e-06, "loss": 0.0417, "step": 16051 }, { "epoch": 0.8557871727888255, "grad_norm": 2.409144401550293, "learning_rate": 6.789800000000001e-06, "loss": 0.0255, "step": 16052 }, { "epoch": 0.8558404862184784, "grad_norm": 1.9923218488693237, "learning_rate": 6.789600000000001e-06, "loss": 0.0232, "step": 16053 }, { "epoch": 0.8558937996481314, "grad_norm": 2.7511284351348877, "learning_rate": 6.789400000000001e-06, "loss": 0.0327, "step": 16054 }, { "epoch": 0.8559471130777843, "grad_norm": 1.158780574798584, "learning_rate": 6.7892e-06, "loss": 0.0126, "step": 16055 }, { "epoch": 0.8560004265074372, "grad_norm": 2.298861265182495, "learning_rate": 6.789e-06, "loss": 0.0303, "step": 16056 }, { "epoch": 0.8560537399370901, "grad_norm": 1.8904167413711548, "learning_rate": 6.788800000000001e-06, "loss": 0.0279, "step": 16057 }, { "epoch": 0.8561070533667431, "grad_norm": 1.2205220460891724, "learning_rate": 6.7886000000000005e-06, "loss": 0.0236, "step": 16058 }, { "epoch": 0.8561603667963961, "grad_norm": 0.8490707278251648, "learning_rate": 6.788400000000001e-06, "loss": 0.0176, "step": 16059 }, { "epoch": 0.8562136802260489, "grad_norm": 2.1465976238250732, "learning_rate": 6.7882000000000005e-06, "loss": 0.0228, "step": 16060 }, { "epoch": 0.8562669936557019, "grad_norm": 2.7855606079101562, "learning_rate": 6.788e-06, "loss": 0.0331, "step": 16061 }, { "epoch": 0.8563203070853548, "grad_norm": 2.6572673320770264, "learning_rate": 6.787800000000001e-06, "loss": 0.0479, "step": 16062 }, { "epoch": 0.8563736205150078, "grad_norm": 1.4348585605621338, "learning_rate": 6.787600000000001e-06, "loss": 0.0108, "step": 16063 }, { "epoch": 0.8564269339446606, "grad_norm": 2.1068756580352783, "learning_rate": 6.7874000000000004e-06, "loss": 0.0414, "step": 16064 }, { "epoch": 0.8564802473743136, "grad_norm": 3.116870164871216, "learning_rate": 6.787200000000001e-06, "loss": 0.0282, "step": 16065 }, { "epoch": 0.8565335608039665, "grad_norm": 2.0332541465759277, "learning_rate": 6.787e-06, "loss": 0.0403, "step": 16066 }, { "epoch": 0.8565868742336195, "grad_norm": 3.1718966960906982, "learning_rate": 6.7868e-06, "loss": 0.0575, "step": 16067 }, { "epoch": 0.8566401876632723, "grad_norm": 2.0865249633789062, "learning_rate": 6.786600000000001e-06, "loss": 0.0345, "step": 16068 }, { "epoch": 0.8566935010929253, "grad_norm": 1.534908413887024, "learning_rate": 6.786400000000001e-06, "loss": 0.013, "step": 16069 }, { "epoch": 0.8567468145225783, "grad_norm": 1.3605711460113525, "learning_rate": 6.7862e-06, "loss": 0.0188, "step": 16070 }, { "epoch": 0.8568001279522311, "grad_norm": 1.5623406171798706, "learning_rate": 6.786000000000001e-06, "loss": 0.0308, "step": 16071 }, { "epoch": 0.8568534413818841, "grad_norm": 1.9543167352676392, "learning_rate": 6.7858e-06, "loss": 0.0378, "step": 16072 }, { "epoch": 0.856906754811537, "grad_norm": 2.5392353534698486, "learning_rate": 6.785600000000001e-06, "loss": 0.0279, "step": 16073 }, { "epoch": 0.85696006824119, "grad_norm": 2.1373507976531982, "learning_rate": 6.785400000000001e-06, "loss": 0.0166, "step": 16074 }, { "epoch": 0.8570133816708428, "grad_norm": 1.500427484512329, "learning_rate": 6.785200000000001e-06, "loss": 0.0115, "step": 16075 }, { "epoch": 0.8570666951004958, "grad_norm": 2.08414626121521, "learning_rate": 6.785e-06, "loss": 0.0491, "step": 16076 }, { "epoch": 0.8571200085301487, "grad_norm": 2.169370174407959, "learning_rate": 6.784800000000001e-06, "loss": 0.0449, "step": 16077 }, { "epoch": 0.8571733219598017, "grad_norm": 0.9018104672431946, "learning_rate": 6.7846e-06, "loss": 0.018, "step": 16078 }, { "epoch": 0.8572266353894546, "grad_norm": 2.322392702102661, "learning_rate": 6.784400000000001e-06, "loss": 0.0441, "step": 16079 }, { "epoch": 0.8572799488191075, "grad_norm": 1.6295363903045654, "learning_rate": 6.784200000000001e-06, "loss": 0.039, "step": 16080 }, { "epoch": 0.8573332622487605, "grad_norm": 1.4824752807617188, "learning_rate": 6.784000000000001e-06, "loss": 0.0248, "step": 16081 }, { "epoch": 0.8573865756784134, "grad_norm": 1.6831004619598389, "learning_rate": 6.7838e-06, "loss": 0.0297, "step": 16082 }, { "epoch": 0.8574398891080663, "grad_norm": 1.2397098541259766, "learning_rate": 6.783600000000001e-06, "loss": 0.0188, "step": 16083 }, { "epoch": 0.8574932025377192, "grad_norm": 2.1913015842437744, "learning_rate": 6.783400000000001e-06, "loss": 0.0279, "step": 16084 }, { "epoch": 0.8575465159673722, "grad_norm": 3.737872838973999, "learning_rate": 6.7832000000000005e-06, "loss": 0.0343, "step": 16085 }, { "epoch": 0.8575998293970251, "grad_norm": 3.4264116287231445, "learning_rate": 6.783000000000001e-06, "loss": 0.062, "step": 16086 }, { "epoch": 0.857653142826678, "grad_norm": 2.659337043762207, "learning_rate": 6.7828000000000005e-06, "loss": 0.0306, "step": 16087 }, { "epoch": 0.857706456256331, "grad_norm": 1.9266458749771118, "learning_rate": 6.7826e-06, "loss": 0.0208, "step": 16088 }, { "epoch": 0.8577597696859839, "grad_norm": 2.390172004699707, "learning_rate": 6.7824000000000005e-06, "loss": 0.0306, "step": 16089 }, { "epoch": 0.8578130831156369, "grad_norm": 2.6669979095458984, "learning_rate": 6.782200000000001e-06, "loss": 0.0417, "step": 16090 }, { "epoch": 0.8578663965452897, "grad_norm": 1.7067084312438965, "learning_rate": 6.7820000000000005e-06, "loss": 0.0229, "step": 16091 }, { "epoch": 0.8579197099749427, "grad_norm": 2.496936559677124, "learning_rate": 6.781800000000001e-06, "loss": 0.0436, "step": 16092 }, { "epoch": 0.8579730234045956, "grad_norm": 2.3078784942626953, "learning_rate": 6.7816000000000004e-06, "loss": 0.0351, "step": 16093 }, { "epoch": 0.8580263368342486, "grad_norm": 0.7534682154655457, "learning_rate": 6.7814e-06, "loss": 0.0107, "step": 16094 }, { "epoch": 0.8580796502639014, "grad_norm": 2.0735056400299072, "learning_rate": 6.7812e-06, "loss": 0.022, "step": 16095 }, { "epoch": 0.8581329636935544, "grad_norm": 2.107840061187744, "learning_rate": 6.781000000000001e-06, "loss": 0.0261, "step": 16096 }, { "epoch": 0.8581862771232074, "grad_norm": 1.7940009832382202, "learning_rate": 6.7808e-06, "loss": 0.0322, "step": 16097 }, { "epoch": 0.8582395905528603, "grad_norm": 1.5543103218078613, "learning_rate": 6.780600000000001e-06, "loss": 0.0066, "step": 16098 }, { "epoch": 0.8582929039825132, "grad_norm": 1.846245527267456, "learning_rate": 6.7804e-06, "loss": 0.0498, "step": 16099 }, { "epoch": 0.8583462174121661, "grad_norm": 2.0311498641967773, "learning_rate": 6.7802e-06, "loss": 0.0313, "step": 16100 }, { "epoch": 0.8583995308418191, "grad_norm": 0.7244923114776611, "learning_rate": 6.780000000000001e-06, "loss": 0.0069, "step": 16101 }, { "epoch": 0.858452844271472, "grad_norm": 2.7407116889953613, "learning_rate": 6.779800000000001e-06, "loss": 0.0714, "step": 16102 }, { "epoch": 0.8585061577011249, "grad_norm": 1.8458834886550903, "learning_rate": 6.7796e-06, "loss": 0.0393, "step": 16103 }, { "epoch": 0.8585594711307778, "grad_norm": 2.1162779331207275, "learning_rate": 6.779400000000001e-06, "loss": 0.032, "step": 16104 }, { "epoch": 0.8586127845604308, "grad_norm": 1.6245458126068115, "learning_rate": 6.7792e-06, "loss": 0.0314, "step": 16105 }, { "epoch": 0.8586660979900838, "grad_norm": 1.1082048416137695, "learning_rate": 6.779e-06, "loss": 0.0155, "step": 16106 }, { "epoch": 0.8587194114197366, "grad_norm": 3.5186328887939453, "learning_rate": 6.778800000000001e-06, "loss": 0.0673, "step": 16107 }, { "epoch": 0.8587727248493896, "grad_norm": 2.6002860069274902, "learning_rate": 6.778600000000001e-06, "loss": 0.0648, "step": 16108 }, { "epoch": 0.8588260382790425, "grad_norm": 1.2340137958526611, "learning_rate": 6.7784e-06, "loss": 0.0111, "step": 16109 }, { "epoch": 0.8588793517086954, "grad_norm": 5.079522132873535, "learning_rate": 6.778200000000001e-06, "loss": 0.0233, "step": 16110 }, { "epoch": 0.8589326651383483, "grad_norm": 1.0009111166000366, "learning_rate": 6.778e-06, "loss": 0.0235, "step": 16111 }, { "epoch": 0.8589859785680013, "grad_norm": 1.4924116134643555, "learning_rate": 6.777800000000001e-06, "loss": 0.0246, "step": 16112 }, { "epoch": 0.8590392919976542, "grad_norm": 1.6859122514724731, "learning_rate": 6.777600000000001e-06, "loss": 0.0224, "step": 16113 }, { "epoch": 0.8590926054273071, "grad_norm": 1.4150360822677612, "learning_rate": 6.7774000000000005e-06, "loss": 0.0363, "step": 16114 }, { "epoch": 0.8591459188569601, "grad_norm": 1.9063200950622559, "learning_rate": 6.7772e-06, "loss": 0.0679, "step": 16115 }, { "epoch": 0.859199232286613, "grad_norm": 1.2151285409927368, "learning_rate": 6.7770000000000005e-06, "loss": 0.0135, "step": 16116 }, { "epoch": 0.859252545716266, "grad_norm": 1.6138006448745728, "learning_rate": 6.7768e-06, "loss": 0.0274, "step": 16117 }, { "epoch": 0.8593058591459188, "grad_norm": 2.7155487537384033, "learning_rate": 6.776600000000001e-06, "loss": 0.0726, "step": 16118 }, { "epoch": 0.8593591725755718, "grad_norm": 1.0986828804016113, "learning_rate": 6.776400000000001e-06, "loss": 0.0115, "step": 16119 }, { "epoch": 0.8594124860052247, "grad_norm": 1.722300410270691, "learning_rate": 6.7762000000000004e-06, "loss": 0.0263, "step": 16120 }, { "epoch": 0.8594657994348777, "grad_norm": 1.645302176475525, "learning_rate": 6.776e-06, "loss": 0.0573, "step": 16121 }, { "epoch": 0.8595191128645305, "grad_norm": 1.4875462055206299, "learning_rate": 6.7758e-06, "loss": 0.0287, "step": 16122 }, { "epoch": 0.8595724262941835, "grad_norm": 2.231370449066162, "learning_rate": 6.775600000000001e-06, "loss": 0.0411, "step": 16123 }, { "epoch": 0.8596257397238364, "grad_norm": 1.0103626251220703, "learning_rate": 6.775400000000001e-06, "loss": 0.0097, "step": 16124 }, { "epoch": 0.8596790531534894, "grad_norm": 0.9427108764648438, "learning_rate": 6.775200000000001e-06, "loss": 0.0148, "step": 16125 }, { "epoch": 0.8597323665831423, "grad_norm": 2.321051597595215, "learning_rate": 6.775e-06, "loss": 0.0496, "step": 16126 }, { "epoch": 0.8597856800127952, "grad_norm": 1.512304663658142, "learning_rate": 6.7748e-06, "loss": 0.0194, "step": 16127 }, { "epoch": 0.8598389934424482, "grad_norm": 1.8886581659317017, "learning_rate": 6.7746e-06, "loss": 0.0296, "step": 16128 }, { "epoch": 0.859892306872101, "grad_norm": 3.666498899459839, "learning_rate": 6.774400000000001e-06, "loss": 0.0606, "step": 16129 }, { "epoch": 0.859945620301754, "grad_norm": 3.0885462760925293, "learning_rate": 6.774200000000001e-06, "loss": 0.0381, "step": 16130 }, { "epoch": 0.8599989337314069, "grad_norm": 1.1408731937408447, "learning_rate": 6.774000000000001e-06, "loss": 0.0251, "step": 16131 }, { "epoch": 0.8600522471610599, "grad_norm": 1.5565587282180786, "learning_rate": 6.7738e-06, "loss": 0.0297, "step": 16132 }, { "epoch": 0.8601055605907127, "grad_norm": 0.8503583669662476, "learning_rate": 6.7736e-06, "loss": 0.0104, "step": 16133 }, { "epoch": 0.8601588740203657, "grad_norm": 1.257441759109497, "learning_rate": 6.773400000000001e-06, "loss": 0.0209, "step": 16134 }, { "epoch": 0.8602121874500187, "grad_norm": 2.3650665283203125, "learning_rate": 6.773200000000001e-06, "loss": 0.0395, "step": 16135 }, { "epoch": 0.8602655008796716, "grad_norm": 1.7976819276809692, "learning_rate": 6.773000000000001e-06, "loss": 0.0379, "step": 16136 }, { "epoch": 0.8603188143093246, "grad_norm": 1.2255135774612427, "learning_rate": 6.772800000000001e-06, "loss": 0.0237, "step": 16137 }, { "epoch": 0.8603721277389774, "grad_norm": 1.8381178379058838, "learning_rate": 6.7726e-06, "loss": 0.0223, "step": 16138 }, { "epoch": 0.8604254411686304, "grad_norm": 2.8306260108947754, "learning_rate": 6.7724e-06, "loss": 0.0293, "step": 16139 }, { "epoch": 0.8604787545982833, "grad_norm": 2.570096015930176, "learning_rate": 6.772200000000001e-06, "loss": 0.0287, "step": 16140 }, { "epoch": 0.8605320680279362, "grad_norm": 2.6935131549835205, "learning_rate": 6.7720000000000006e-06, "loss": 0.0458, "step": 16141 }, { "epoch": 0.8605853814575891, "grad_norm": 1.2569732666015625, "learning_rate": 6.771800000000001e-06, "loss": 0.0335, "step": 16142 }, { "epoch": 0.8606386948872421, "grad_norm": 2.8836803436279297, "learning_rate": 6.7716000000000005e-06, "loss": 0.0506, "step": 16143 }, { "epoch": 0.8606920083168951, "grad_norm": 3.409990072250366, "learning_rate": 6.7714e-06, "loss": 0.0446, "step": 16144 }, { "epoch": 0.8607453217465479, "grad_norm": 3.022529363632202, "learning_rate": 6.771200000000001e-06, "loss": 0.0449, "step": 16145 }, { "epoch": 0.8607986351762009, "grad_norm": 1.7184016704559326, "learning_rate": 6.771000000000001e-06, "loss": 0.0166, "step": 16146 }, { "epoch": 0.8608519486058538, "grad_norm": 1.2045745849609375, "learning_rate": 6.7708000000000005e-06, "loss": 0.0175, "step": 16147 }, { "epoch": 0.8609052620355068, "grad_norm": 3.9105677604675293, "learning_rate": 6.770600000000001e-06, "loss": 0.029, "step": 16148 }, { "epoch": 0.8609585754651596, "grad_norm": 1.4508037567138672, "learning_rate": 6.7704000000000004e-06, "loss": 0.0247, "step": 16149 }, { "epoch": 0.8610118888948126, "grad_norm": 1.2687066793441772, "learning_rate": 6.7702e-06, "loss": 0.0134, "step": 16150 }, { "epoch": 0.8610652023244655, "grad_norm": 1.723126769065857, "learning_rate": 6.770000000000001e-06, "loss": 0.036, "step": 16151 }, { "epoch": 0.8611185157541185, "grad_norm": 2.150744676589966, "learning_rate": 6.769800000000001e-06, "loss": 0.049, "step": 16152 }, { "epoch": 0.8611718291837714, "grad_norm": 1.6890637874603271, "learning_rate": 6.7696e-06, "loss": 0.0301, "step": 16153 }, { "epoch": 0.8612251426134243, "grad_norm": 1.090911626815796, "learning_rate": 6.769400000000001e-06, "loss": 0.0165, "step": 16154 }, { "epoch": 0.8612784560430773, "grad_norm": 1.6771503686904907, "learning_rate": 6.7692e-06, "loss": 0.0426, "step": 16155 }, { "epoch": 0.8613317694727302, "grad_norm": 1.6244016885757446, "learning_rate": 6.769e-06, "loss": 0.0194, "step": 16156 }, { "epoch": 0.8613850829023831, "grad_norm": 1.2079087495803833, "learning_rate": 6.768800000000001e-06, "loss": 0.0217, "step": 16157 }, { "epoch": 0.861438396332036, "grad_norm": 1.3615092039108276, "learning_rate": 6.768600000000001e-06, "loss": 0.0092, "step": 16158 }, { "epoch": 0.861491709761689, "grad_norm": 1.0711132287979126, "learning_rate": 6.7684e-06, "loss": 0.0249, "step": 16159 }, { "epoch": 0.8615450231913419, "grad_norm": 2.3767058849334717, "learning_rate": 6.768200000000001e-06, "loss": 0.021, "step": 16160 }, { "epoch": 0.8615983366209948, "grad_norm": 1.0608983039855957, "learning_rate": 6.768e-06, "loss": 0.0125, "step": 16161 }, { "epoch": 0.8616516500506478, "grad_norm": 3.86519455909729, "learning_rate": 6.767800000000001e-06, "loss": 0.0376, "step": 16162 }, { "epoch": 0.8617049634803007, "grad_norm": 1.0113004446029663, "learning_rate": 6.767600000000001e-06, "loss": 0.007, "step": 16163 }, { "epoch": 0.8617582769099537, "grad_norm": 1.6974596977233887, "learning_rate": 6.767400000000001e-06, "loss": 0.0137, "step": 16164 }, { "epoch": 0.8618115903396065, "grad_norm": 1.0317437648773193, "learning_rate": 6.7672e-06, "loss": 0.0102, "step": 16165 }, { "epoch": 0.8618649037692595, "grad_norm": 2.4039883613586426, "learning_rate": 6.767000000000001e-06, "loss": 0.0448, "step": 16166 }, { "epoch": 0.8619182171989124, "grad_norm": 2.090726137161255, "learning_rate": 6.7668e-06, "loss": 0.0452, "step": 16167 }, { "epoch": 0.8619715306285654, "grad_norm": 1.4422309398651123, "learning_rate": 6.7666000000000006e-06, "loss": 0.0242, "step": 16168 }, { "epoch": 0.8620248440582182, "grad_norm": 1.8513754606246948, "learning_rate": 6.766400000000001e-06, "loss": 0.04, "step": 16169 }, { "epoch": 0.8620781574878712, "grad_norm": 1.3808915615081787, "learning_rate": 6.7662000000000005e-06, "loss": 0.0243, "step": 16170 }, { "epoch": 0.8621314709175242, "grad_norm": 3.0665557384490967, "learning_rate": 6.766e-06, "loss": 0.0761, "step": 16171 }, { "epoch": 0.862184784347177, "grad_norm": 0.8649063110351562, "learning_rate": 6.7658000000000005e-06, "loss": 0.0066, "step": 16172 }, { "epoch": 0.86223809777683, "grad_norm": 3.4687137603759766, "learning_rate": 6.765600000000001e-06, "loss": 0.0428, "step": 16173 }, { "epoch": 0.8622914112064829, "grad_norm": 1.8876299858093262, "learning_rate": 6.7654000000000005e-06, "loss": 0.0227, "step": 16174 }, { "epoch": 0.8623447246361359, "grad_norm": 0.8622595071792603, "learning_rate": 6.765200000000001e-06, "loss": 0.0239, "step": 16175 }, { "epoch": 0.8623980380657887, "grad_norm": 2.021758556365967, "learning_rate": 6.7650000000000005e-06, "loss": 0.0323, "step": 16176 }, { "epoch": 0.8624513514954417, "grad_norm": 1.634019374847412, "learning_rate": 6.7648e-06, "loss": 0.0207, "step": 16177 }, { "epoch": 0.8625046649250946, "grad_norm": 1.831934928894043, "learning_rate": 6.7646e-06, "loss": 0.0284, "step": 16178 }, { "epoch": 0.8625579783547476, "grad_norm": 2.4295637607574463, "learning_rate": 6.764400000000001e-06, "loss": 0.0468, "step": 16179 }, { "epoch": 0.8626112917844004, "grad_norm": 1.4954051971435547, "learning_rate": 6.7642e-06, "loss": 0.0207, "step": 16180 }, { "epoch": 0.8626646052140534, "grad_norm": 2.1727631092071533, "learning_rate": 6.764000000000001e-06, "loss": 0.0343, "step": 16181 }, { "epoch": 0.8627179186437064, "grad_norm": 0.9618098139762878, "learning_rate": 6.7638e-06, "loss": 0.0194, "step": 16182 }, { "epoch": 0.8627712320733593, "grad_norm": 2.4476888179779053, "learning_rate": 6.7636e-06, "loss": 0.0662, "step": 16183 }, { "epoch": 0.8628245455030122, "grad_norm": 2.3254706859588623, "learning_rate": 6.763400000000001e-06, "loss": 0.0432, "step": 16184 }, { "epoch": 0.8628778589326651, "grad_norm": 1.6728113889694214, "learning_rate": 6.763200000000001e-06, "loss": 0.0206, "step": 16185 }, { "epoch": 0.8629311723623181, "grad_norm": 1.8269987106323242, "learning_rate": 6.763e-06, "loss": 0.0198, "step": 16186 }, { "epoch": 0.862984485791971, "grad_norm": 2.026677370071411, "learning_rate": 6.762800000000001e-06, "loss": 0.0316, "step": 16187 }, { "epoch": 0.8630377992216239, "grad_norm": 1.0076653957366943, "learning_rate": 6.7626e-06, "loss": 0.0192, "step": 16188 }, { "epoch": 0.8630911126512768, "grad_norm": 1.1926510334014893, "learning_rate": 6.7624e-06, "loss": 0.0268, "step": 16189 }, { "epoch": 0.8631444260809298, "grad_norm": 1.8890362977981567, "learning_rate": 6.762200000000001e-06, "loss": 0.043, "step": 16190 }, { "epoch": 0.8631977395105828, "grad_norm": 4.444675922393799, "learning_rate": 6.762000000000001e-06, "loss": 0.0513, "step": 16191 }, { "epoch": 0.8632510529402356, "grad_norm": 0.9823793172836304, "learning_rate": 6.7618e-06, "loss": 0.0244, "step": 16192 }, { "epoch": 0.8633043663698886, "grad_norm": 1.246924877166748, "learning_rate": 6.761600000000001e-06, "loss": 0.0192, "step": 16193 }, { "epoch": 0.8633576797995415, "grad_norm": 1.2660540342330933, "learning_rate": 6.7614e-06, "loss": 0.0118, "step": 16194 }, { "epoch": 0.8634109932291945, "grad_norm": 2.1461522579193115, "learning_rate": 6.7612000000000014e-06, "loss": 0.048, "step": 16195 }, { "epoch": 0.8634643066588473, "grad_norm": 1.5310828685760498, "learning_rate": 6.761000000000001e-06, "loss": 0.0344, "step": 16196 }, { "epoch": 0.8635176200885003, "grad_norm": 1.3652825355529785, "learning_rate": 6.7608000000000006e-06, "loss": 0.0224, "step": 16197 }, { "epoch": 0.8635709335181532, "grad_norm": 1.1757053136825562, "learning_rate": 6.7606e-06, "loss": 0.0119, "step": 16198 }, { "epoch": 0.8636242469478062, "grad_norm": 1.5152802467346191, "learning_rate": 6.7604000000000005e-06, "loss": 0.0236, "step": 16199 }, { "epoch": 0.8636775603774591, "grad_norm": 2.213927984237671, "learning_rate": 6.7602e-06, "loss": 0.022, "step": 16200 }, { "epoch": 0.863730873807112, "grad_norm": 1.6082783937454224, "learning_rate": 6.760000000000001e-06, "loss": 0.0398, "step": 16201 }, { "epoch": 0.863784187236765, "grad_norm": 1.6471402645111084, "learning_rate": 6.759800000000001e-06, "loss": 0.0293, "step": 16202 }, { "epoch": 0.8638375006664178, "grad_norm": 1.2135757207870483, "learning_rate": 6.7596000000000005e-06, "loss": 0.0148, "step": 16203 }, { "epoch": 0.8638908140960708, "grad_norm": 1.20247483253479, "learning_rate": 6.7594e-06, "loss": 0.0227, "step": 16204 }, { "epoch": 0.8639441275257237, "grad_norm": 1.730826497077942, "learning_rate": 6.7592000000000004e-06, "loss": 0.0242, "step": 16205 }, { "epoch": 0.8639974409553767, "grad_norm": 1.8462597131729126, "learning_rate": 6.759e-06, "loss": 0.0249, "step": 16206 }, { "epoch": 0.8640507543850295, "grad_norm": 2.220913887023926, "learning_rate": 6.758800000000001e-06, "loss": 0.0514, "step": 16207 }, { "epoch": 0.8641040678146825, "grad_norm": 1.647918939590454, "learning_rate": 6.758600000000001e-06, "loss": 0.0473, "step": 16208 }, { "epoch": 0.8641573812443355, "grad_norm": 1.6696003675460815, "learning_rate": 6.7584e-06, "loss": 0.0449, "step": 16209 }, { "epoch": 0.8642106946739884, "grad_norm": 6.160879135131836, "learning_rate": 6.7582e-06, "loss": 0.0513, "step": 16210 }, { "epoch": 0.8642640081036413, "grad_norm": 2.2657840251922607, "learning_rate": 6.758e-06, "loss": 0.0298, "step": 16211 }, { "epoch": 0.8643173215332942, "grad_norm": 2.7129898071289062, "learning_rate": 6.757800000000001e-06, "loss": 0.0573, "step": 16212 }, { "epoch": 0.8643706349629472, "grad_norm": 2.5323596000671387, "learning_rate": 6.757600000000001e-06, "loss": 0.0208, "step": 16213 }, { "epoch": 0.8644239483926001, "grad_norm": 1.9569532871246338, "learning_rate": 6.757400000000001e-06, "loss": 0.0304, "step": 16214 }, { "epoch": 0.864477261822253, "grad_norm": 1.965421438217163, "learning_rate": 6.7572e-06, "loss": 0.0366, "step": 16215 }, { "epoch": 0.8645305752519059, "grad_norm": 2.1352179050445557, "learning_rate": 6.757e-06, "loss": 0.0205, "step": 16216 }, { "epoch": 0.8645838886815589, "grad_norm": 0.9859024882316589, "learning_rate": 6.7568e-06, "loss": 0.0192, "step": 16217 }, { "epoch": 0.8646372021112119, "grad_norm": 2.4622597694396973, "learning_rate": 6.756600000000001e-06, "loss": 0.0326, "step": 16218 }, { "epoch": 0.8646905155408647, "grad_norm": 2.1094276905059814, "learning_rate": 6.756400000000001e-06, "loss": 0.0388, "step": 16219 }, { "epoch": 0.8647438289705177, "grad_norm": 1.7387945652008057, "learning_rate": 6.756200000000001e-06, "loss": 0.0498, "step": 16220 }, { "epoch": 0.8647971424001706, "grad_norm": 2.813079595565796, "learning_rate": 6.756e-06, "loss": 0.0411, "step": 16221 }, { "epoch": 0.8648504558298236, "grad_norm": 0.9974082112312317, "learning_rate": 6.7558e-06, "loss": 0.0204, "step": 16222 }, { "epoch": 0.8649037692594764, "grad_norm": 1.5015685558319092, "learning_rate": 6.755600000000001e-06, "loss": 0.0186, "step": 16223 }, { "epoch": 0.8649570826891294, "grad_norm": 2.2895424365997314, "learning_rate": 6.755400000000001e-06, "loss": 0.0412, "step": 16224 }, { "epoch": 0.8650103961187823, "grad_norm": 1.0323598384857178, "learning_rate": 6.755200000000001e-06, "loss": 0.015, "step": 16225 }, { "epoch": 0.8650637095484353, "grad_norm": 2.056546211242676, "learning_rate": 6.7550000000000005e-06, "loss": 0.0165, "step": 16226 }, { "epoch": 0.8651170229780882, "grad_norm": 1.9371370077133179, "learning_rate": 6.7548e-06, "loss": 0.0428, "step": 16227 }, { "epoch": 0.8651703364077411, "grad_norm": 1.7435985803604126, "learning_rate": 6.7546e-06, "loss": 0.0142, "step": 16228 }, { "epoch": 0.8652236498373941, "grad_norm": 1.7281460762023926, "learning_rate": 6.754400000000001e-06, "loss": 0.014, "step": 16229 }, { "epoch": 0.865276963267047, "grad_norm": 2.0250015258789062, "learning_rate": 6.7542000000000005e-06, "loss": 0.032, "step": 16230 }, { "epoch": 0.8653302766966999, "grad_norm": 1.5295993089675903, "learning_rate": 6.754000000000001e-06, "loss": 0.0162, "step": 16231 }, { "epoch": 0.8653835901263528, "grad_norm": 1.2523051500320435, "learning_rate": 6.7538000000000005e-06, "loss": 0.0272, "step": 16232 }, { "epoch": 0.8654369035560058, "grad_norm": 1.5210727453231812, "learning_rate": 6.7536e-06, "loss": 0.0165, "step": 16233 }, { "epoch": 0.8654902169856586, "grad_norm": 1.0501587390899658, "learning_rate": 6.753400000000001e-06, "loss": 0.0118, "step": 16234 }, { "epoch": 0.8655435304153116, "grad_norm": 2.2073609828948975, "learning_rate": 6.753200000000001e-06, "loss": 0.0266, "step": 16235 }, { "epoch": 0.8655968438449645, "grad_norm": 2.0020036697387695, "learning_rate": 6.753e-06, "loss": 0.0242, "step": 16236 }, { "epoch": 0.8656501572746175, "grad_norm": 1.862304449081421, "learning_rate": 6.752800000000001e-06, "loss": 0.0136, "step": 16237 }, { "epoch": 0.8657034707042705, "grad_norm": 1.979989767074585, "learning_rate": 6.7526e-06, "loss": 0.0295, "step": 16238 }, { "epoch": 0.8657567841339233, "grad_norm": 1.5138624906539917, "learning_rate": 6.7524e-06, "loss": 0.0089, "step": 16239 }, { "epoch": 0.8658100975635763, "grad_norm": 2.2768373489379883, "learning_rate": 6.752200000000001e-06, "loss": 0.0244, "step": 16240 }, { "epoch": 0.8658634109932292, "grad_norm": 2.1191537380218506, "learning_rate": 6.752000000000001e-06, "loss": 0.0457, "step": 16241 }, { "epoch": 0.8659167244228821, "grad_norm": 1.54502272605896, "learning_rate": 6.7518e-06, "loss": 0.027, "step": 16242 }, { "epoch": 0.865970037852535, "grad_norm": 3.7535741329193115, "learning_rate": 6.751600000000001e-06, "loss": 0.035, "step": 16243 }, { "epoch": 0.866023351282188, "grad_norm": 2.839319944381714, "learning_rate": 6.7514e-06, "loss": 0.0335, "step": 16244 }, { "epoch": 0.8660766647118409, "grad_norm": 1.8503820896148682, "learning_rate": 6.751200000000001e-06, "loss": 0.0108, "step": 16245 }, { "epoch": 0.8661299781414938, "grad_norm": 1.5767778158187866, "learning_rate": 6.751000000000001e-06, "loss": 0.0166, "step": 16246 }, { "epoch": 0.8661832915711468, "grad_norm": 1.8336036205291748, "learning_rate": 6.750800000000001e-06, "loss": 0.0238, "step": 16247 }, { "epoch": 0.8662366050007997, "grad_norm": 2.6113121509552, "learning_rate": 6.7506e-06, "loss": 0.0541, "step": 16248 }, { "epoch": 0.8662899184304527, "grad_norm": 0.3540407121181488, "learning_rate": 6.750400000000001e-06, "loss": 0.0035, "step": 16249 }, { "epoch": 0.8663432318601055, "grad_norm": 1.141409993171692, "learning_rate": 6.7502e-06, "loss": 0.0253, "step": 16250 }, { "epoch": 0.8663965452897585, "grad_norm": 1.5832924842834473, "learning_rate": 6.750000000000001e-06, "loss": 0.0131, "step": 16251 }, { "epoch": 0.8664498587194114, "grad_norm": 2.011396646499634, "learning_rate": 6.749800000000001e-06, "loss": 0.0235, "step": 16252 }, { "epoch": 0.8665031721490644, "grad_norm": 3.13057541847229, "learning_rate": 6.7496000000000006e-06, "loss": 0.0719, "step": 16253 }, { "epoch": 0.8665564855787172, "grad_norm": 3.4987704753875732, "learning_rate": 6.7494e-06, "loss": 0.0708, "step": 16254 }, { "epoch": 0.8666097990083702, "grad_norm": 4.226099491119385, "learning_rate": 6.7492000000000005e-06, "loss": 0.0179, "step": 16255 }, { "epoch": 0.8666631124380232, "grad_norm": 3.0938329696655273, "learning_rate": 6.749000000000001e-06, "loss": 0.0508, "step": 16256 }, { "epoch": 0.866716425867676, "grad_norm": 2.6814682483673096, "learning_rate": 6.7488000000000005e-06, "loss": 0.0447, "step": 16257 }, { "epoch": 0.866769739297329, "grad_norm": 3.8789892196655273, "learning_rate": 6.748600000000001e-06, "loss": 0.0311, "step": 16258 }, { "epoch": 0.8668230527269819, "grad_norm": 3.5796144008636475, "learning_rate": 6.7484000000000005e-06, "loss": 0.0211, "step": 16259 }, { "epoch": 0.8668763661566349, "grad_norm": 2.312173366546631, "learning_rate": 6.7482e-06, "loss": 0.0139, "step": 16260 }, { "epoch": 0.8669296795862877, "grad_norm": 2.5445821285247803, "learning_rate": 6.7480000000000004e-06, "loss": 0.0187, "step": 16261 }, { "epoch": 0.8669829930159407, "grad_norm": 1.7968610525131226, "learning_rate": 6.747800000000001e-06, "loss": 0.0361, "step": 16262 }, { "epoch": 0.8670363064455936, "grad_norm": 0.8333032131195068, "learning_rate": 6.7476e-06, "loss": 0.0063, "step": 16263 }, { "epoch": 0.8670896198752466, "grad_norm": 2.058145761489868, "learning_rate": 6.747400000000001e-06, "loss": 0.0396, "step": 16264 }, { "epoch": 0.8671429333048996, "grad_norm": 2.066392660140991, "learning_rate": 6.7472e-06, "loss": 0.0345, "step": 16265 }, { "epoch": 0.8671962467345524, "grad_norm": 1.4050360918045044, "learning_rate": 6.747e-06, "loss": 0.0316, "step": 16266 }, { "epoch": 0.8672495601642054, "grad_norm": 1.5487903356552124, "learning_rate": 6.7468e-06, "loss": 0.0397, "step": 16267 }, { "epoch": 0.8673028735938583, "grad_norm": 1.7920825481414795, "learning_rate": 6.746600000000001e-06, "loss": 0.0229, "step": 16268 }, { "epoch": 0.8673561870235112, "grad_norm": 3.066094160079956, "learning_rate": 6.7464e-06, "loss": 0.0562, "step": 16269 }, { "epoch": 0.8674095004531641, "grad_norm": 0.8496443033218384, "learning_rate": 6.746200000000001e-06, "loss": 0.0094, "step": 16270 }, { "epoch": 0.8674628138828171, "grad_norm": 2.0590906143188477, "learning_rate": 6.746e-06, "loss": 0.0309, "step": 16271 }, { "epoch": 0.86751612731247, "grad_norm": 2.493217706680298, "learning_rate": 6.7458e-06, "loss": 0.0566, "step": 16272 }, { "epoch": 0.8675694407421229, "grad_norm": 2.1623404026031494, "learning_rate": 6.745600000000001e-06, "loss": 0.0526, "step": 16273 }, { "epoch": 0.8676227541717759, "grad_norm": 2.6145989894866943, "learning_rate": 6.745400000000001e-06, "loss": 0.0437, "step": 16274 }, { "epoch": 0.8676760676014288, "grad_norm": 1.4891633987426758, "learning_rate": 6.7452e-06, "loss": 0.0262, "step": 16275 }, { "epoch": 0.8677293810310818, "grad_norm": 5.033188343048096, "learning_rate": 6.745000000000001e-06, "loss": 0.0476, "step": 16276 }, { "epoch": 0.8677826944607346, "grad_norm": 3.3733251094818115, "learning_rate": 6.7448e-06, "loss": 0.0345, "step": 16277 }, { "epoch": 0.8678360078903876, "grad_norm": 1.169905424118042, "learning_rate": 6.7446e-06, "loss": 0.0153, "step": 16278 }, { "epoch": 0.8678893213200405, "grad_norm": 1.708045244216919, "learning_rate": 6.744400000000001e-06, "loss": 0.0255, "step": 16279 }, { "epoch": 0.8679426347496935, "grad_norm": 3.7658305168151855, "learning_rate": 6.744200000000001e-06, "loss": 0.0511, "step": 16280 }, { "epoch": 0.8679959481793463, "grad_norm": 0.9932352900505066, "learning_rate": 6.744e-06, "loss": 0.0237, "step": 16281 }, { "epoch": 0.8680492616089993, "grad_norm": 1.3603460788726807, "learning_rate": 6.7438000000000006e-06, "loss": 0.016, "step": 16282 }, { "epoch": 0.8681025750386523, "grad_norm": 0.9445914626121521, "learning_rate": 6.7436e-06, "loss": 0.0072, "step": 16283 }, { "epoch": 0.8681558884683052, "grad_norm": 2.022397041320801, "learning_rate": 6.743400000000001e-06, "loss": 0.0204, "step": 16284 }, { "epoch": 0.8682092018979581, "grad_norm": 1.5980850458145142, "learning_rate": 6.743200000000001e-06, "loss": 0.0131, "step": 16285 }, { "epoch": 0.868262515327611, "grad_norm": 1.6597745418548584, "learning_rate": 6.7430000000000005e-06, "loss": 0.0245, "step": 16286 }, { "epoch": 0.868315828757264, "grad_norm": 2.314131498336792, "learning_rate": 6.7428e-06, "loss": 0.0389, "step": 16287 }, { "epoch": 0.8683691421869169, "grad_norm": 1.981879472732544, "learning_rate": 6.7426000000000005e-06, "loss": 0.0449, "step": 16288 }, { "epoch": 0.8684224556165698, "grad_norm": 1.4797255992889404, "learning_rate": 6.7424e-06, "loss": 0.0163, "step": 16289 }, { "epoch": 0.8684757690462227, "grad_norm": 2.25801157951355, "learning_rate": 6.742200000000001e-06, "loss": 0.0408, "step": 16290 }, { "epoch": 0.8685290824758757, "grad_norm": 2.9082634449005127, "learning_rate": 6.742000000000001e-06, "loss": 0.0561, "step": 16291 }, { "epoch": 0.8685823959055285, "grad_norm": 2.8799283504486084, "learning_rate": 6.7418e-06, "loss": 0.0524, "step": 16292 }, { "epoch": 0.8686357093351815, "grad_norm": 2.857814311981201, "learning_rate": 6.7416e-06, "loss": 0.0692, "step": 16293 }, { "epoch": 0.8686890227648345, "grad_norm": 3.7827584743499756, "learning_rate": 6.7414e-06, "loss": 0.0266, "step": 16294 }, { "epoch": 0.8687423361944874, "grad_norm": 1.779787540435791, "learning_rate": 6.741200000000001e-06, "loss": 0.0401, "step": 16295 }, { "epoch": 0.8687956496241404, "grad_norm": 1.8297475576400757, "learning_rate": 6.741000000000001e-06, "loss": 0.0271, "step": 16296 }, { "epoch": 0.8688489630537932, "grad_norm": 1.702268362045288, "learning_rate": 6.740800000000001e-06, "loss": 0.053, "step": 16297 }, { "epoch": 0.8689022764834462, "grad_norm": 1.2880632877349854, "learning_rate": 6.7406e-06, "loss": 0.0157, "step": 16298 }, { "epoch": 0.8689555899130991, "grad_norm": 2.5317087173461914, "learning_rate": 6.7404e-06, "loss": 0.0231, "step": 16299 }, { "epoch": 0.869008903342752, "grad_norm": 1.6897131204605103, "learning_rate": 6.7402e-06, "loss": 0.0283, "step": 16300 }, { "epoch": 0.8690622167724049, "grad_norm": 1.5874303579330444, "learning_rate": 6.740000000000001e-06, "loss": 0.0259, "step": 16301 }, { "epoch": 0.8691155302020579, "grad_norm": 2.3499529361724854, "learning_rate": 6.739800000000001e-06, "loss": 0.0314, "step": 16302 }, { "epoch": 0.8691688436317109, "grad_norm": 1.6630499362945557, "learning_rate": 6.739600000000001e-06, "loss": 0.0136, "step": 16303 }, { "epoch": 0.8692221570613637, "grad_norm": 1.7484601736068726, "learning_rate": 6.7394e-06, "loss": 0.0418, "step": 16304 }, { "epoch": 0.8692754704910167, "grad_norm": 0.8716051578521729, "learning_rate": 6.7392e-06, "loss": 0.0134, "step": 16305 }, { "epoch": 0.8693287839206696, "grad_norm": 2.215423345565796, "learning_rate": 6.739000000000001e-06, "loss": 0.0428, "step": 16306 }, { "epoch": 0.8693820973503226, "grad_norm": 1.4762110710144043, "learning_rate": 6.738800000000001e-06, "loss": 0.024, "step": 16307 }, { "epoch": 0.8694354107799754, "grad_norm": 2.268406391143799, "learning_rate": 6.738600000000001e-06, "loss": 0.0308, "step": 16308 }, { "epoch": 0.8694887242096284, "grad_norm": 2.5992064476013184, "learning_rate": 6.7384000000000006e-06, "loss": 0.0704, "step": 16309 }, { "epoch": 0.8695420376392813, "grad_norm": 2.368720054626465, "learning_rate": 6.7382e-06, "loss": 0.021, "step": 16310 }, { "epoch": 0.8695953510689343, "grad_norm": 1.087028980255127, "learning_rate": 6.738e-06, "loss": 0.0194, "step": 16311 }, { "epoch": 0.8696486644985872, "grad_norm": 2.5833797454833984, "learning_rate": 6.737800000000001e-06, "loss": 0.0257, "step": 16312 }, { "epoch": 0.8697019779282401, "grad_norm": 1.6142208576202393, "learning_rate": 6.7376000000000005e-06, "loss": 0.0208, "step": 16313 }, { "epoch": 0.8697552913578931, "grad_norm": 1.9403754472732544, "learning_rate": 6.737400000000001e-06, "loss": 0.0375, "step": 16314 }, { "epoch": 0.869808604787546, "grad_norm": 0.7838788032531738, "learning_rate": 6.7372000000000005e-06, "loss": 0.0094, "step": 16315 }, { "epoch": 0.8698619182171989, "grad_norm": 1.8470600843429565, "learning_rate": 6.737e-06, "loss": 0.0156, "step": 16316 }, { "epoch": 0.8699152316468518, "grad_norm": 3.483088731765747, "learning_rate": 6.7368e-06, "loss": 0.0373, "step": 16317 }, { "epoch": 0.8699685450765048, "grad_norm": 1.261782169342041, "learning_rate": 6.736600000000001e-06, "loss": 0.0197, "step": 16318 }, { "epoch": 0.8700218585061577, "grad_norm": 1.5730342864990234, "learning_rate": 6.7364e-06, "loss": 0.0354, "step": 16319 }, { "epoch": 0.8700751719358106, "grad_norm": 1.4673473834991455, "learning_rate": 6.736200000000001e-06, "loss": 0.0228, "step": 16320 }, { "epoch": 0.8701284853654636, "grad_norm": 1.941739559173584, "learning_rate": 6.736e-06, "loss": 0.0294, "step": 16321 }, { "epoch": 0.8701817987951165, "grad_norm": 2.3511719703674316, "learning_rate": 6.7358e-06, "loss": 0.0403, "step": 16322 }, { "epoch": 0.8702351122247695, "grad_norm": 2.9089388847351074, "learning_rate": 6.735600000000001e-06, "loss": 0.0341, "step": 16323 }, { "epoch": 0.8702884256544223, "grad_norm": 1.5439224243164062, "learning_rate": 6.735400000000001e-06, "loss": 0.0272, "step": 16324 }, { "epoch": 0.8703417390840753, "grad_norm": 1.881508231163025, "learning_rate": 6.7352e-06, "loss": 0.0287, "step": 16325 }, { "epoch": 0.8703950525137282, "grad_norm": 1.6594135761260986, "learning_rate": 6.735000000000001e-06, "loss": 0.0121, "step": 16326 }, { "epoch": 0.8704483659433812, "grad_norm": 2.9069015979766846, "learning_rate": 6.7348e-06, "loss": 0.0339, "step": 16327 }, { "epoch": 0.870501679373034, "grad_norm": 2.8349051475524902, "learning_rate": 6.7346e-06, "loss": 0.0206, "step": 16328 }, { "epoch": 0.870554992802687, "grad_norm": 3.7896106243133545, "learning_rate": 6.734400000000001e-06, "loss": 0.0416, "step": 16329 }, { "epoch": 0.87060830623234, "grad_norm": 1.5308074951171875, "learning_rate": 6.734200000000001e-06, "loss": 0.0368, "step": 16330 }, { "epoch": 0.8706616196619928, "grad_norm": 2.446155548095703, "learning_rate": 6.734e-06, "loss": 0.0193, "step": 16331 }, { "epoch": 0.8707149330916458, "grad_norm": 5.8424248695373535, "learning_rate": 6.733800000000001e-06, "loss": 0.0436, "step": 16332 }, { "epoch": 0.8707682465212987, "grad_norm": 2.4741344451904297, "learning_rate": 6.7336e-06, "loss": 0.0315, "step": 16333 }, { "epoch": 0.8708215599509517, "grad_norm": 3.388418674468994, "learning_rate": 6.7334000000000015e-06, "loss": 0.0486, "step": 16334 }, { "epoch": 0.8708748733806045, "grad_norm": 2.1543149948120117, "learning_rate": 6.733200000000001e-06, "loss": 0.0282, "step": 16335 }, { "epoch": 0.8709281868102575, "grad_norm": 2.826677083969116, "learning_rate": 6.733000000000001e-06, "loss": 0.0599, "step": 16336 }, { "epoch": 0.8709815002399104, "grad_norm": 1.1430487632751465, "learning_rate": 6.7328e-06, "loss": 0.0109, "step": 16337 }, { "epoch": 0.8710348136695634, "grad_norm": 1.061354398727417, "learning_rate": 6.7326000000000006e-06, "loss": 0.012, "step": 16338 }, { "epoch": 0.8710881270992163, "grad_norm": 1.1568869352340698, "learning_rate": 6.7324e-06, "loss": 0.0197, "step": 16339 }, { "epoch": 0.8711414405288692, "grad_norm": 1.3131177425384521, "learning_rate": 6.732200000000001e-06, "loss": 0.0209, "step": 16340 }, { "epoch": 0.8711947539585222, "grad_norm": 1.692352294921875, "learning_rate": 6.732000000000001e-06, "loss": 0.0195, "step": 16341 }, { "epoch": 0.8712480673881751, "grad_norm": 1.9310734272003174, "learning_rate": 6.7318000000000005e-06, "loss": 0.0308, "step": 16342 }, { "epoch": 0.871301380817828, "grad_norm": 1.3885935544967651, "learning_rate": 6.7316e-06, "loss": 0.014, "step": 16343 }, { "epoch": 0.8713546942474809, "grad_norm": 0.8737527132034302, "learning_rate": 6.7314000000000005e-06, "loss": 0.0193, "step": 16344 }, { "epoch": 0.8714080076771339, "grad_norm": 1.5663752555847168, "learning_rate": 6.731200000000001e-06, "loss": 0.0473, "step": 16345 }, { "epoch": 0.8714613211067868, "grad_norm": 1.7776423692703247, "learning_rate": 6.731000000000001e-06, "loss": 0.0239, "step": 16346 }, { "epoch": 0.8715146345364397, "grad_norm": 0.8658280372619629, "learning_rate": 6.730800000000001e-06, "loss": 0.0063, "step": 16347 }, { "epoch": 0.8715679479660926, "grad_norm": 1.7789911031723022, "learning_rate": 6.7306e-06, "loss": 0.0206, "step": 16348 }, { "epoch": 0.8716212613957456, "grad_norm": 1.733079195022583, "learning_rate": 6.7304e-06, "loss": 0.047, "step": 16349 }, { "epoch": 0.8716745748253986, "grad_norm": 2.1032493114471436, "learning_rate": 6.7302e-06, "loss": 0.02, "step": 16350 }, { "epoch": 0.8717278882550514, "grad_norm": 1.2462414503097534, "learning_rate": 6.730000000000001e-06, "loss": 0.0076, "step": 16351 }, { "epoch": 0.8717812016847044, "grad_norm": 1.2273081541061401, "learning_rate": 6.729800000000001e-06, "loss": 0.0283, "step": 16352 }, { "epoch": 0.8718345151143573, "grad_norm": 2.7059836387634277, "learning_rate": 6.729600000000001e-06, "loss": 0.0253, "step": 16353 }, { "epoch": 0.8718878285440103, "grad_norm": 2.7403483390808105, "learning_rate": 6.7294e-06, "loss": 0.0295, "step": 16354 }, { "epoch": 0.8719411419736631, "grad_norm": 1.536324143409729, "learning_rate": 6.7292e-06, "loss": 0.0292, "step": 16355 }, { "epoch": 0.8719944554033161, "grad_norm": 1.4270422458648682, "learning_rate": 6.729000000000001e-06, "loss": 0.0258, "step": 16356 }, { "epoch": 0.872047768832969, "grad_norm": 2.773930311203003, "learning_rate": 6.728800000000001e-06, "loss": 0.0699, "step": 16357 }, { "epoch": 0.872101082262622, "grad_norm": 2.018145799636841, "learning_rate": 6.728600000000001e-06, "loss": 0.048, "step": 16358 }, { "epoch": 0.8721543956922749, "grad_norm": 1.4796689748764038, "learning_rate": 6.728400000000001e-06, "loss": 0.0228, "step": 16359 }, { "epoch": 0.8722077091219278, "grad_norm": 1.9244732856750488, "learning_rate": 6.7282e-06, "loss": 0.0306, "step": 16360 }, { "epoch": 0.8722610225515808, "grad_norm": 1.6953866481781006, "learning_rate": 6.728e-06, "loss": 0.029, "step": 16361 }, { "epoch": 0.8723143359812336, "grad_norm": 1.4731898307800293, "learning_rate": 6.727800000000001e-06, "loss": 0.0284, "step": 16362 }, { "epoch": 0.8723676494108866, "grad_norm": 1.7655267715454102, "learning_rate": 6.727600000000001e-06, "loss": 0.0318, "step": 16363 }, { "epoch": 0.8724209628405395, "grad_norm": 1.6518782377243042, "learning_rate": 6.727400000000001e-06, "loss": 0.0329, "step": 16364 }, { "epoch": 0.8724742762701925, "grad_norm": 1.7680004835128784, "learning_rate": 6.7272000000000006e-06, "loss": 0.0322, "step": 16365 }, { "epoch": 0.8725275896998453, "grad_norm": 1.6270064115524292, "learning_rate": 6.727e-06, "loss": 0.0241, "step": 16366 }, { "epoch": 0.8725809031294983, "grad_norm": 1.3487074375152588, "learning_rate": 6.7268e-06, "loss": 0.0134, "step": 16367 }, { "epoch": 0.8726342165591513, "grad_norm": 1.856062889099121, "learning_rate": 6.726600000000001e-06, "loss": 0.0234, "step": 16368 }, { "epoch": 0.8726875299888042, "grad_norm": 1.9582651853561401, "learning_rate": 6.7264000000000005e-06, "loss": 0.032, "step": 16369 }, { "epoch": 0.8727408434184571, "grad_norm": 1.6438535451889038, "learning_rate": 6.726200000000001e-06, "loss": 0.041, "step": 16370 }, { "epoch": 0.87279415684811, "grad_norm": 2.0192956924438477, "learning_rate": 6.7260000000000005e-06, "loss": 0.0297, "step": 16371 }, { "epoch": 0.872847470277763, "grad_norm": 2.5745832920074463, "learning_rate": 6.7258e-06, "loss": 0.024, "step": 16372 }, { "epoch": 0.8729007837074159, "grad_norm": 2.9668097496032715, "learning_rate": 6.725600000000001e-06, "loss": 0.0222, "step": 16373 }, { "epoch": 0.8729540971370688, "grad_norm": 2.584611654281616, "learning_rate": 6.725400000000001e-06, "loss": 0.0229, "step": 16374 }, { "epoch": 0.8730074105667217, "grad_norm": 1.6270521879196167, "learning_rate": 6.7252000000000004e-06, "loss": 0.0202, "step": 16375 }, { "epoch": 0.8730607239963747, "grad_norm": 2.926982879638672, "learning_rate": 6.725000000000001e-06, "loss": 0.0471, "step": 16376 }, { "epoch": 0.8731140374260277, "grad_norm": 1.7098637819290161, "learning_rate": 6.7248e-06, "loss": 0.0385, "step": 16377 }, { "epoch": 0.8731673508556805, "grad_norm": 1.6667886972427368, "learning_rate": 6.7246e-06, "loss": 0.0438, "step": 16378 }, { "epoch": 0.8732206642853335, "grad_norm": 1.6993881464004517, "learning_rate": 6.724400000000001e-06, "loss": 0.0276, "step": 16379 }, { "epoch": 0.8732739777149864, "grad_norm": 0.8986055850982666, "learning_rate": 6.724200000000001e-06, "loss": 0.0218, "step": 16380 }, { "epoch": 0.8733272911446394, "grad_norm": 2.0312812328338623, "learning_rate": 6.724e-06, "loss": 0.012, "step": 16381 }, { "epoch": 0.8733806045742922, "grad_norm": 1.3876407146453857, "learning_rate": 6.723800000000001e-06, "loss": 0.0326, "step": 16382 }, { "epoch": 0.8734339180039452, "grad_norm": 1.6249748468399048, "learning_rate": 6.7236e-06, "loss": 0.0097, "step": 16383 }, { "epoch": 0.8734872314335981, "grad_norm": 1.1427096128463745, "learning_rate": 6.723400000000001e-06, "loss": 0.0103, "step": 16384 }, { "epoch": 0.8735405448632511, "grad_norm": 2.4959123134613037, "learning_rate": 6.723200000000001e-06, "loss": 0.0298, "step": 16385 }, { "epoch": 0.873593858292904, "grad_norm": 0.7449691295623779, "learning_rate": 6.723000000000001e-06, "loss": 0.0056, "step": 16386 }, { "epoch": 0.8736471717225569, "grad_norm": 3.1148767471313477, "learning_rate": 6.7228e-06, "loss": 0.0508, "step": 16387 }, { "epoch": 0.8737004851522099, "grad_norm": 1.8645834922790527, "learning_rate": 6.722600000000001e-06, "loss": 0.0366, "step": 16388 }, { "epoch": 0.8737537985818628, "grad_norm": 2.090050220489502, "learning_rate": 6.7224e-06, "loss": 0.0245, "step": 16389 }, { "epoch": 0.8738071120115157, "grad_norm": 2.0209500789642334, "learning_rate": 6.722200000000001e-06, "loss": 0.0416, "step": 16390 }, { "epoch": 0.8738604254411686, "grad_norm": 1.5045868158340454, "learning_rate": 6.722000000000001e-06, "loss": 0.0222, "step": 16391 }, { "epoch": 0.8739137388708216, "grad_norm": 2.1571412086486816, "learning_rate": 6.721800000000001e-06, "loss": 0.0345, "step": 16392 }, { "epoch": 0.8739670523004744, "grad_norm": 3.680112838745117, "learning_rate": 6.7216e-06, "loss": 0.0306, "step": 16393 }, { "epoch": 0.8740203657301274, "grad_norm": 1.9618566036224365, "learning_rate": 6.7214000000000006e-06, "loss": 0.0216, "step": 16394 }, { "epoch": 0.8740736791597804, "grad_norm": 2.385434150695801, "learning_rate": 6.721200000000001e-06, "loss": 0.0653, "step": 16395 }, { "epoch": 0.8741269925894333, "grad_norm": 1.4815706014633179, "learning_rate": 6.7210000000000005e-06, "loss": 0.0328, "step": 16396 }, { "epoch": 0.8741803060190863, "grad_norm": 2.2572133541107178, "learning_rate": 6.720800000000001e-06, "loss": 0.0313, "step": 16397 }, { "epoch": 0.8742336194487391, "grad_norm": 2.3310065269470215, "learning_rate": 6.7206000000000005e-06, "loss": 0.0242, "step": 16398 }, { "epoch": 0.8742869328783921, "grad_norm": 1.4603947401046753, "learning_rate": 6.7204e-06, "loss": 0.0098, "step": 16399 }, { "epoch": 0.874340246308045, "grad_norm": 1.7992223501205444, "learning_rate": 6.7202000000000005e-06, "loss": 0.0181, "step": 16400 }, { "epoch": 0.874393559737698, "grad_norm": 1.9335911273956299, "learning_rate": 6.720000000000001e-06, "loss": 0.0147, "step": 16401 }, { "epoch": 0.8744468731673508, "grad_norm": 1.7451750040054321, "learning_rate": 6.7198000000000004e-06, "loss": 0.0249, "step": 16402 }, { "epoch": 0.8745001865970038, "grad_norm": 1.103570818901062, "learning_rate": 6.719600000000001e-06, "loss": 0.0078, "step": 16403 }, { "epoch": 0.8745535000266567, "grad_norm": 4.136918544769287, "learning_rate": 6.7194e-06, "loss": 0.0614, "step": 16404 }, { "epoch": 0.8746068134563096, "grad_norm": 2.607583999633789, "learning_rate": 6.7192e-06, "loss": 0.0256, "step": 16405 }, { "epoch": 0.8746601268859626, "grad_norm": 1.4277560710906982, "learning_rate": 6.719000000000001e-06, "loss": 0.0127, "step": 16406 }, { "epoch": 0.8747134403156155, "grad_norm": 1.52776038646698, "learning_rate": 6.718800000000001e-06, "loss": 0.0297, "step": 16407 }, { "epoch": 0.8747667537452685, "grad_norm": 2.8281733989715576, "learning_rate": 6.7186e-06, "loss": 0.0494, "step": 16408 }, { "epoch": 0.8748200671749213, "grad_norm": 0.970970094203949, "learning_rate": 6.718400000000001e-06, "loss": 0.0235, "step": 16409 }, { "epoch": 0.8748733806045743, "grad_norm": 1.3300176858901978, "learning_rate": 6.7182e-06, "loss": 0.014, "step": 16410 }, { "epoch": 0.8749266940342272, "grad_norm": 1.5627102851867676, "learning_rate": 6.718e-06, "loss": 0.0237, "step": 16411 }, { "epoch": 0.8749800074638802, "grad_norm": 1.7266887426376343, "learning_rate": 6.717800000000001e-06, "loss": 0.0339, "step": 16412 }, { "epoch": 0.875033320893533, "grad_norm": 2.786501407623291, "learning_rate": 6.717600000000001e-06, "loss": 0.0424, "step": 16413 }, { "epoch": 0.875086634323186, "grad_norm": 2.1763734817504883, "learning_rate": 6.7174e-06, "loss": 0.0375, "step": 16414 }, { "epoch": 0.875139947752839, "grad_norm": 2.0264499187469482, "learning_rate": 6.717200000000001e-06, "loss": 0.0229, "step": 16415 }, { "epoch": 0.8751932611824919, "grad_norm": 1.7749733924865723, "learning_rate": 6.717e-06, "loss": 0.0271, "step": 16416 }, { "epoch": 0.8752465746121448, "grad_norm": 2.8445346355438232, "learning_rate": 6.7168000000000015e-06, "loss": 0.0329, "step": 16417 }, { "epoch": 0.8752998880417977, "grad_norm": 4.587591648101807, "learning_rate": 6.716600000000001e-06, "loss": 0.0237, "step": 16418 }, { "epoch": 0.8753532014714507, "grad_norm": 2.3605377674102783, "learning_rate": 6.716400000000001e-06, "loss": 0.0407, "step": 16419 }, { "epoch": 0.8754065149011035, "grad_norm": 1.6438331604003906, "learning_rate": 6.7162e-06, "loss": 0.0274, "step": 16420 }, { "epoch": 0.8754598283307565, "grad_norm": 2.600050210952759, "learning_rate": 6.716000000000001e-06, "loss": 0.0552, "step": 16421 }, { "epoch": 0.8755131417604094, "grad_norm": 1.5698636770248413, "learning_rate": 6.7158e-06, "loss": 0.0115, "step": 16422 }, { "epoch": 0.8755664551900624, "grad_norm": 1.5117452144622803, "learning_rate": 6.715600000000001e-06, "loss": 0.0142, "step": 16423 }, { "epoch": 0.8756197686197154, "grad_norm": 1.9733678102493286, "learning_rate": 6.715400000000001e-06, "loss": 0.0304, "step": 16424 }, { "epoch": 0.8756730820493682, "grad_norm": 1.5843302011489868, "learning_rate": 6.7152000000000005e-06, "loss": 0.0269, "step": 16425 }, { "epoch": 0.8757263954790212, "grad_norm": 0.7312785983085632, "learning_rate": 6.715e-06, "loss": 0.0118, "step": 16426 }, { "epoch": 0.8757797089086741, "grad_norm": 1.1369248628616333, "learning_rate": 6.7148000000000005e-06, "loss": 0.0151, "step": 16427 }, { "epoch": 0.875833022338327, "grad_norm": 2.116128444671631, "learning_rate": 6.7146e-06, "loss": 0.0398, "step": 16428 }, { "epoch": 0.8758863357679799, "grad_norm": 1.7729175090789795, "learning_rate": 6.714400000000001e-06, "loss": 0.0267, "step": 16429 }, { "epoch": 0.8759396491976329, "grad_norm": 1.590423822402954, "learning_rate": 6.714200000000001e-06, "loss": 0.0272, "step": 16430 }, { "epoch": 0.8759929626272858, "grad_norm": 1.973930835723877, "learning_rate": 6.7140000000000004e-06, "loss": 0.0569, "step": 16431 }, { "epoch": 0.8760462760569387, "grad_norm": 2.815281391143799, "learning_rate": 6.7138e-06, "loss": 0.0603, "step": 16432 }, { "epoch": 0.8760995894865917, "grad_norm": 1.8522433042526245, "learning_rate": 6.7136e-06, "loss": 0.0372, "step": 16433 }, { "epoch": 0.8761529029162446, "grad_norm": 2.7185044288635254, "learning_rate": 6.713400000000001e-06, "loss": 0.0511, "step": 16434 }, { "epoch": 0.8762062163458976, "grad_norm": 1.316131353378296, "learning_rate": 6.713200000000001e-06, "loss": 0.0168, "step": 16435 }, { "epoch": 0.8762595297755504, "grad_norm": 3.0143957138061523, "learning_rate": 6.713000000000001e-06, "loss": 0.0368, "step": 16436 }, { "epoch": 0.8763128432052034, "grad_norm": 2.8187177181243896, "learning_rate": 6.7128e-06, "loss": 0.049, "step": 16437 }, { "epoch": 0.8763661566348563, "grad_norm": 1.780389428138733, "learning_rate": 6.7126e-06, "loss": 0.0369, "step": 16438 }, { "epoch": 0.8764194700645093, "grad_norm": 2.559009313583374, "learning_rate": 6.7124e-06, "loss": 0.0531, "step": 16439 }, { "epoch": 0.8764727834941621, "grad_norm": 1.038082480430603, "learning_rate": 6.712200000000001e-06, "loss": 0.0131, "step": 16440 }, { "epoch": 0.8765260969238151, "grad_norm": 0.9630255103111267, "learning_rate": 6.712000000000001e-06, "loss": 0.0142, "step": 16441 }, { "epoch": 0.8765794103534681, "grad_norm": 1.2522311210632324, "learning_rate": 6.711800000000001e-06, "loss": 0.012, "step": 16442 }, { "epoch": 0.876632723783121, "grad_norm": 1.0255845785140991, "learning_rate": 6.7116e-06, "loss": 0.0101, "step": 16443 }, { "epoch": 0.8766860372127739, "grad_norm": 1.8517539501190186, "learning_rate": 6.7114e-06, "loss": 0.0253, "step": 16444 }, { "epoch": 0.8767393506424268, "grad_norm": 1.0505759716033936, "learning_rate": 6.711200000000001e-06, "loss": 0.0226, "step": 16445 }, { "epoch": 0.8767926640720798, "grad_norm": 2.8695693016052246, "learning_rate": 6.711000000000001e-06, "loss": 0.0173, "step": 16446 }, { "epoch": 0.8768459775017327, "grad_norm": 1.8788281679153442, "learning_rate": 6.710800000000001e-06, "loss": 0.0283, "step": 16447 }, { "epoch": 0.8768992909313856, "grad_norm": 0.941682755947113, "learning_rate": 6.710600000000001e-06, "loss": 0.0082, "step": 16448 }, { "epoch": 0.8769526043610385, "grad_norm": 0.5095216035842896, "learning_rate": 6.7104e-06, "loss": 0.0056, "step": 16449 }, { "epoch": 0.8770059177906915, "grad_norm": 3.186255931854248, "learning_rate": 6.7102e-06, "loss": 0.0737, "step": 16450 }, { "epoch": 0.8770592312203445, "grad_norm": 1.6765941381454468, "learning_rate": 6.710000000000001e-06, "loss": 0.0133, "step": 16451 }, { "epoch": 0.8771125446499973, "grad_norm": 2.385298728942871, "learning_rate": 6.7098000000000005e-06, "loss": 0.0195, "step": 16452 }, { "epoch": 0.8771658580796503, "grad_norm": 0.986689567565918, "learning_rate": 6.709600000000001e-06, "loss": 0.0087, "step": 16453 }, { "epoch": 0.8772191715093032, "grad_norm": 0.7150309085845947, "learning_rate": 6.7094000000000005e-06, "loss": 0.0178, "step": 16454 }, { "epoch": 0.8772724849389562, "grad_norm": 2.006812810897827, "learning_rate": 6.7092e-06, "loss": 0.0394, "step": 16455 }, { "epoch": 0.877325798368609, "grad_norm": 2.7976160049438477, "learning_rate": 6.709000000000001e-06, "loss": 0.0421, "step": 16456 }, { "epoch": 0.877379111798262, "grad_norm": 1.2090904712677002, "learning_rate": 6.708800000000001e-06, "loss": 0.0087, "step": 16457 }, { "epoch": 0.8774324252279149, "grad_norm": 3.390918493270874, "learning_rate": 6.7086000000000004e-06, "loss": 0.0439, "step": 16458 }, { "epoch": 0.8774857386575678, "grad_norm": 2.8834898471832275, "learning_rate": 6.708400000000001e-06, "loss": 0.0529, "step": 16459 }, { "epoch": 0.8775390520872207, "grad_norm": 1.5078516006469727, "learning_rate": 6.7082e-06, "loss": 0.0198, "step": 16460 }, { "epoch": 0.8775923655168737, "grad_norm": 6.9449663162231445, "learning_rate": 6.708e-06, "loss": 0.0517, "step": 16461 }, { "epoch": 0.8776456789465267, "grad_norm": 2.7418601512908936, "learning_rate": 6.707800000000001e-06, "loss": 0.0241, "step": 16462 }, { "epoch": 0.8776989923761795, "grad_norm": 2.0151047706604004, "learning_rate": 6.707600000000001e-06, "loss": 0.0154, "step": 16463 }, { "epoch": 0.8777523058058325, "grad_norm": 2.5114643573760986, "learning_rate": 6.7074e-06, "loss": 0.0611, "step": 16464 }, { "epoch": 0.8778056192354854, "grad_norm": 3.565851926803589, "learning_rate": 6.707200000000001e-06, "loss": 0.0631, "step": 16465 }, { "epoch": 0.8778589326651384, "grad_norm": 0.7243911623954773, "learning_rate": 6.707e-06, "loss": 0.0046, "step": 16466 }, { "epoch": 0.8779122460947912, "grad_norm": 2.9432730674743652, "learning_rate": 6.706800000000001e-06, "loss": 0.0663, "step": 16467 }, { "epoch": 0.8779655595244442, "grad_norm": 2.082231044769287, "learning_rate": 6.706600000000001e-06, "loss": 0.0475, "step": 16468 }, { "epoch": 0.8780188729540971, "grad_norm": 1.6146323680877686, "learning_rate": 6.706400000000001e-06, "loss": 0.0104, "step": 16469 }, { "epoch": 0.8780721863837501, "grad_norm": 2.6136574745178223, "learning_rate": 6.7062e-06, "loss": 0.0501, "step": 16470 }, { "epoch": 0.878125499813403, "grad_norm": 1.1908555030822754, "learning_rate": 6.706000000000001e-06, "loss": 0.0321, "step": 16471 }, { "epoch": 0.8781788132430559, "grad_norm": 0.730876088142395, "learning_rate": 6.7058e-06, "loss": 0.0248, "step": 16472 }, { "epoch": 0.8782321266727089, "grad_norm": 3.5058062076568604, "learning_rate": 6.705600000000001e-06, "loss": 0.0666, "step": 16473 }, { "epoch": 0.8782854401023618, "grad_norm": 1.385783314704895, "learning_rate": 6.705400000000001e-06, "loss": 0.0111, "step": 16474 }, { "epoch": 0.8783387535320147, "grad_norm": 2.7634246349334717, "learning_rate": 6.705200000000001e-06, "loss": 0.0449, "step": 16475 }, { "epoch": 0.8783920669616676, "grad_norm": 1.6412303447723389, "learning_rate": 6.705e-06, "loss": 0.0154, "step": 16476 }, { "epoch": 0.8784453803913206, "grad_norm": 2.4686872959136963, "learning_rate": 6.704800000000001e-06, "loss": 0.0399, "step": 16477 }, { "epoch": 0.8784986938209735, "grad_norm": 3.0619139671325684, "learning_rate": 6.7046e-06, "loss": 0.0432, "step": 16478 }, { "epoch": 0.8785520072506264, "grad_norm": 1.7446002960205078, "learning_rate": 6.7044000000000006e-06, "loss": 0.028, "step": 16479 }, { "epoch": 0.8786053206802794, "grad_norm": 6.611715316772461, "learning_rate": 6.704200000000001e-06, "loss": 0.0519, "step": 16480 }, { "epoch": 0.8786586341099323, "grad_norm": 1.9906367063522339, "learning_rate": 6.7040000000000005e-06, "loss": 0.0347, "step": 16481 }, { "epoch": 0.8787119475395853, "grad_norm": 1.632246732711792, "learning_rate": 6.7038e-06, "loss": 0.0353, "step": 16482 }, { "epoch": 0.8787652609692381, "grad_norm": 2.6555304527282715, "learning_rate": 6.7036000000000005e-06, "loss": 0.0541, "step": 16483 }, { "epoch": 0.8788185743988911, "grad_norm": 0.8905225396156311, "learning_rate": 6.703400000000001e-06, "loss": 0.0117, "step": 16484 }, { "epoch": 0.878871887828544, "grad_norm": 1.574980616569519, "learning_rate": 6.7032000000000005e-06, "loss": 0.0267, "step": 16485 }, { "epoch": 0.878925201258197, "grad_norm": 1.8793811798095703, "learning_rate": 6.703000000000001e-06, "loss": 0.0428, "step": 16486 }, { "epoch": 0.8789785146878498, "grad_norm": 1.6149768829345703, "learning_rate": 6.7028000000000004e-06, "loss": 0.0409, "step": 16487 }, { "epoch": 0.8790318281175028, "grad_norm": 1.2256016731262207, "learning_rate": 6.7026e-06, "loss": 0.0191, "step": 16488 }, { "epoch": 0.8790851415471558, "grad_norm": 2.158371925354004, "learning_rate": 6.7024e-06, "loss": 0.0556, "step": 16489 }, { "epoch": 0.8791384549768086, "grad_norm": 3.188045024871826, "learning_rate": 6.702200000000001e-06, "loss": 0.0623, "step": 16490 }, { "epoch": 0.8791917684064616, "grad_norm": 0.25390928983688354, "learning_rate": 6.702e-06, "loss": 0.0033, "step": 16491 }, { "epoch": 0.8792450818361145, "grad_norm": 4.080973148345947, "learning_rate": 6.701800000000001e-06, "loss": 0.0252, "step": 16492 }, { "epoch": 0.8792983952657675, "grad_norm": 0.7677863836288452, "learning_rate": 6.7016e-06, "loss": 0.0217, "step": 16493 }, { "epoch": 0.8793517086954203, "grad_norm": 0.8351125717163086, "learning_rate": 6.7014e-06, "loss": 0.01, "step": 16494 }, { "epoch": 0.8794050221250733, "grad_norm": 3.051520824432373, "learning_rate": 6.701200000000001e-06, "loss": 0.0521, "step": 16495 }, { "epoch": 0.8794583355547262, "grad_norm": 0.925545871257782, "learning_rate": 6.701000000000001e-06, "loss": 0.0128, "step": 16496 }, { "epoch": 0.8795116489843792, "grad_norm": 1.8230431079864502, "learning_rate": 6.7008e-06, "loss": 0.0251, "step": 16497 }, { "epoch": 0.8795649624140321, "grad_norm": 1.0571637153625488, "learning_rate": 6.700600000000001e-06, "loss": 0.014, "step": 16498 }, { "epoch": 0.879618275843685, "grad_norm": 2.0840883255004883, "learning_rate": 6.7004e-06, "loss": 0.048, "step": 16499 }, { "epoch": 0.879671589273338, "grad_norm": 3.0310165882110596, "learning_rate": 6.7002e-06, "loss": 0.0306, "step": 16500 }, { "epoch": 0.8797249027029909, "grad_norm": 2.969510078430176, "learning_rate": 6.700000000000001e-06, "loss": 0.0551, "step": 16501 }, { "epoch": 0.8797782161326438, "grad_norm": 1.7840646505355835, "learning_rate": 6.699800000000001e-06, "loss": 0.0587, "step": 16502 }, { "epoch": 0.8798315295622967, "grad_norm": 1.7825791835784912, "learning_rate": 6.6996e-06, "loss": 0.024, "step": 16503 }, { "epoch": 0.8798848429919497, "grad_norm": 1.897813081741333, "learning_rate": 6.699400000000001e-06, "loss": 0.036, "step": 16504 }, { "epoch": 0.8799381564216026, "grad_norm": 2.8912947177886963, "learning_rate": 6.6992e-06, "loss": 0.0499, "step": 16505 }, { "epoch": 0.8799914698512555, "grad_norm": 1.301110029220581, "learning_rate": 6.699000000000001e-06, "loss": 0.0152, "step": 16506 }, { "epoch": 0.8800447832809085, "grad_norm": 1.4540908336639404, "learning_rate": 6.698800000000001e-06, "loss": 0.0234, "step": 16507 }, { "epoch": 0.8800980967105614, "grad_norm": 1.6266270875930786, "learning_rate": 6.6986000000000005e-06, "loss": 0.0301, "step": 16508 }, { "epoch": 0.8801514101402144, "grad_norm": 1.895613670349121, "learning_rate": 6.6984e-06, "loss": 0.0356, "step": 16509 }, { "epoch": 0.8802047235698672, "grad_norm": 1.8898327350616455, "learning_rate": 6.6982000000000005e-06, "loss": 0.039, "step": 16510 }, { "epoch": 0.8802580369995202, "grad_norm": 1.6376296281814575, "learning_rate": 6.698e-06, "loss": 0.0243, "step": 16511 }, { "epoch": 0.8803113504291731, "grad_norm": 1.4375927448272705, "learning_rate": 6.697800000000001e-06, "loss": 0.0275, "step": 16512 }, { "epoch": 0.8803646638588261, "grad_norm": 1.4090110063552856, "learning_rate": 6.697600000000001e-06, "loss": 0.0251, "step": 16513 }, { "epoch": 0.8804179772884789, "grad_norm": 1.6289411783218384, "learning_rate": 6.6974000000000005e-06, "loss": 0.0158, "step": 16514 }, { "epoch": 0.8804712907181319, "grad_norm": 1.5986461639404297, "learning_rate": 6.6972e-06, "loss": 0.0269, "step": 16515 }, { "epoch": 0.8805246041477848, "grad_norm": 2.8148205280303955, "learning_rate": 6.6970000000000004e-06, "loss": 0.0285, "step": 16516 }, { "epoch": 0.8805779175774378, "grad_norm": 0.7522982358932495, "learning_rate": 6.696800000000001e-06, "loss": 0.0079, "step": 16517 }, { "epoch": 0.8806312310070907, "grad_norm": 2.114318370819092, "learning_rate": 6.696600000000001e-06, "loss": 0.0279, "step": 16518 }, { "epoch": 0.8806845444367436, "grad_norm": 1.859276533126831, "learning_rate": 6.696400000000001e-06, "loss": 0.0225, "step": 16519 }, { "epoch": 0.8807378578663966, "grad_norm": 2.1326537132263184, "learning_rate": 6.6962e-06, "loss": 0.06, "step": 16520 }, { "epoch": 0.8807911712960494, "grad_norm": 0.9946470856666565, "learning_rate": 6.696e-06, "loss": 0.0119, "step": 16521 }, { "epoch": 0.8808444847257024, "grad_norm": 2.0328187942504883, "learning_rate": 6.6958e-06, "loss": 0.0289, "step": 16522 }, { "epoch": 0.8808977981553553, "grad_norm": 1.8660321235656738, "learning_rate": 6.695600000000001e-06, "loss": 0.0226, "step": 16523 }, { "epoch": 0.8809511115850083, "grad_norm": 2.036222457885742, "learning_rate": 6.695400000000001e-06, "loss": 0.0142, "step": 16524 }, { "epoch": 0.8810044250146611, "grad_norm": 1.3036290407180786, "learning_rate": 6.695200000000001e-06, "loss": 0.0213, "step": 16525 }, { "epoch": 0.8810577384443141, "grad_norm": 2.2556638717651367, "learning_rate": 6.695e-06, "loss": 0.024, "step": 16526 }, { "epoch": 0.8811110518739671, "grad_norm": 2.5414960384368896, "learning_rate": 6.6948e-06, "loss": 0.0605, "step": 16527 }, { "epoch": 0.88116436530362, "grad_norm": 3.626359224319458, "learning_rate": 6.694600000000001e-06, "loss": 0.0745, "step": 16528 }, { "epoch": 0.881217678733273, "grad_norm": 3.214301347732544, "learning_rate": 6.694400000000001e-06, "loss": 0.0559, "step": 16529 }, { "epoch": 0.8812709921629258, "grad_norm": 2.7365334033966064, "learning_rate": 6.694200000000001e-06, "loss": 0.0237, "step": 16530 }, { "epoch": 0.8813243055925788, "grad_norm": 2.0796971321105957, "learning_rate": 6.694000000000001e-06, "loss": 0.034, "step": 16531 }, { "epoch": 0.8813776190222317, "grad_norm": 4.925380706787109, "learning_rate": 6.6938e-06, "loss": 0.0372, "step": 16532 }, { "epoch": 0.8814309324518846, "grad_norm": 1.2138583660125732, "learning_rate": 6.6936e-06, "loss": 0.0173, "step": 16533 }, { "epoch": 0.8814842458815375, "grad_norm": 1.444828748703003, "learning_rate": 6.693400000000001e-06, "loss": 0.0172, "step": 16534 }, { "epoch": 0.8815375593111905, "grad_norm": 1.1948909759521484, "learning_rate": 6.6932000000000006e-06, "loss": 0.0272, "step": 16535 }, { "epoch": 0.8815908727408435, "grad_norm": 2.744917869567871, "learning_rate": 6.693000000000001e-06, "loss": 0.0235, "step": 16536 }, { "epoch": 0.8816441861704963, "grad_norm": 2.2206637859344482, "learning_rate": 6.6928000000000005e-06, "loss": 0.0302, "step": 16537 }, { "epoch": 0.8816974996001493, "grad_norm": 1.6628612279891968, "learning_rate": 6.6926e-06, "loss": 0.0326, "step": 16538 }, { "epoch": 0.8817508130298022, "grad_norm": 1.6262004375457764, "learning_rate": 6.6924e-06, "loss": 0.0286, "step": 16539 }, { "epoch": 0.8818041264594552, "grad_norm": 1.781455636024475, "learning_rate": 6.692200000000001e-06, "loss": 0.0287, "step": 16540 }, { "epoch": 0.881857439889108, "grad_norm": 2.478987693786621, "learning_rate": 6.6920000000000005e-06, "loss": 0.0666, "step": 16541 }, { "epoch": 0.881910753318761, "grad_norm": 0.7611057162284851, "learning_rate": 6.691800000000001e-06, "loss": 0.0081, "step": 16542 }, { "epoch": 0.8819640667484139, "grad_norm": 2.548463821411133, "learning_rate": 6.6916000000000004e-06, "loss": 0.0398, "step": 16543 }, { "epoch": 0.8820173801780669, "grad_norm": 1.2252953052520752, "learning_rate": 6.6914e-06, "loss": 0.0171, "step": 16544 }, { "epoch": 0.8820706936077198, "grad_norm": 1.0772435665130615, "learning_rate": 6.691200000000001e-06, "loss": 0.0106, "step": 16545 }, { "epoch": 0.8821240070373727, "grad_norm": 2.202239751815796, "learning_rate": 6.691000000000001e-06, "loss": 0.0523, "step": 16546 }, { "epoch": 0.8821773204670257, "grad_norm": 2.39371657371521, "learning_rate": 6.6908e-06, "loss": 0.0473, "step": 16547 }, { "epoch": 0.8822306338966786, "grad_norm": 1.6085193157196045, "learning_rate": 6.690600000000001e-06, "loss": 0.0133, "step": 16548 }, { "epoch": 0.8822839473263315, "grad_norm": 1.1843385696411133, "learning_rate": 6.6904e-06, "loss": 0.011, "step": 16549 }, { "epoch": 0.8823372607559844, "grad_norm": 1.6210289001464844, "learning_rate": 6.6902e-06, "loss": 0.0388, "step": 16550 }, { "epoch": 0.8823905741856374, "grad_norm": 2.162944793701172, "learning_rate": 6.690000000000001e-06, "loss": 0.0355, "step": 16551 }, { "epoch": 0.8824438876152902, "grad_norm": 1.9868998527526855, "learning_rate": 6.689800000000001e-06, "loss": 0.0164, "step": 16552 }, { "epoch": 0.8824972010449432, "grad_norm": 0.6194581389427185, "learning_rate": 6.6896e-06, "loss": 0.0065, "step": 16553 }, { "epoch": 0.8825505144745962, "grad_norm": 2.277348518371582, "learning_rate": 6.689400000000001e-06, "loss": 0.0481, "step": 16554 }, { "epoch": 0.8826038279042491, "grad_norm": 1.4483691453933716, "learning_rate": 6.6892e-06, "loss": 0.0255, "step": 16555 }, { "epoch": 0.882657141333902, "grad_norm": 0.7135029435157776, "learning_rate": 6.689000000000001e-06, "loss": 0.0117, "step": 16556 }, { "epoch": 0.8827104547635549, "grad_norm": 2.185593605041504, "learning_rate": 6.688800000000001e-06, "loss": 0.0264, "step": 16557 }, { "epoch": 0.8827637681932079, "grad_norm": 1.4688520431518555, "learning_rate": 6.688600000000001e-06, "loss": 0.0149, "step": 16558 }, { "epoch": 0.8828170816228608, "grad_norm": 1.48749577999115, "learning_rate": 6.6884e-06, "loss": 0.0203, "step": 16559 }, { "epoch": 0.8828703950525137, "grad_norm": 2.1891956329345703, "learning_rate": 6.688200000000001e-06, "loss": 0.0374, "step": 16560 }, { "epoch": 0.8829237084821666, "grad_norm": 1.8826314210891724, "learning_rate": 6.688e-06, "loss": 0.0378, "step": 16561 }, { "epoch": 0.8829770219118196, "grad_norm": 1.4549669027328491, "learning_rate": 6.687800000000001e-06, "loss": 0.0308, "step": 16562 }, { "epoch": 0.8830303353414726, "grad_norm": 0.849706768989563, "learning_rate": 6.687600000000001e-06, "loss": 0.0158, "step": 16563 }, { "epoch": 0.8830836487711254, "grad_norm": 1.0007846355438232, "learning_rate": 6.6874000000000006e-06, "loss": 0.0091, "step": 16564 }, { "epoch": 0.8831369622007784, "grad_norm": 1.3866759538650513, "learning_rate": 6.6872e-06, "loss": 0.0344, "step": 16565 }, { "epoch": 0.8831902756304313, "grad_norm": 1.3721596002578735, "learning_rate": 6.6870000000000005e-06, "loss": 0.0245, "step": 16566 }, { "epoch": 0.8832435890600843, "grad_norm": 2.142906427383423, "learning_rate": 6.686800000000001e-06, "loss": 0.0546, "step": 16567 }, { "epoch": 0.8832969024897371, "grad_norm": 3.666287422180176, "learning_rate": 6.6866000000000005e-06, "loss": 0.0279, "step": 16568 }, { "epoch": 0.8833502159193901, "grad_norm": 2.082660675048828, "learning_rate": 6.686400000000001e-06, "loss": 0.0331, "step": 16569 }, { "epoch": 0.883403529349043, "grad_norm": 1.8637700080871582, "learning_rate": 6.6862000000000005e-06, "loss": 0.0241, "step": 16570 }, { "epoch": 0.883456842778696, "grad_norm": 2.4978601932525635, "learning_rate": 6.686e-06, "loss": 0.0221, "step": 16571 }, { "epoch": 0.8835101562083488, "grad_norm": 1.5399471521377563, "learning_rate": 6.6858000000000004e-06, "loss": 0.0214, "step": 16572 }, { "epoch": 0.8835634696380018, "grad_norm": 1.0737354755401611, "learning_rate": 6.685600000000001e-06, "loss": 0.0203, "step": 16573 }, { "epoch": 0.8836167830676548, "grad_norm": 1.432127833366394, "learning_rate": 6.6854e-06, "loss": 0.0216, "step": 16574 }, { "epoch": 0.8836700964973077, "grad_norm": 2.1470069885253906, "learning_rate": 6.685200000000001e-06, "loss": 0.0235, "step": 16575 }, { "epoch": 0.8837234099269606, "grad_norm": 1.3212441205978394, "learning_rate": 6.685e-06, "loss": 0.0171, "step": 16576 }, { "epoch": 0.8837767233566135, "grad_norm": 1.902442455291748, "learning_rate": 6.6848e-06, "loss": 0.0491, "step": 16577 }, { "epoch": 0.8838300367862665, "grad_norm": 1.1091972589492798, "learning_rate": 6.684600000000001e-06, "loss": 0.0291, "step": 16578 }, { "epoch": 0.8838833502159194, "grad_norm": 2.677564859390259, "learning_rate": 6.684400000000001e-06, "loss": 0.028, "step": 16579 }, { "epoch": 0.8839366636455723, "grad_norm": 3.8320581912994385, "learning_rate": 6.6842e-06, "loss": 0.0395, "step": 16580 }, { "epoch": 0.8839899770752252, "grad_norm": 0.9616488814353943, "learning_rate": 6.684000000000001e-06, "loss": 0.0228, "step": 16581 }, { "epoch": 0.8840432905048782, "grad_norm": 1.2786182165145874, "learning_rate": 6.6838e-06, "loss": 0.0276, "step": 16582 }, { "epoch": 0.8840966039345312, "grad_norm": 1.4088081121444702, "learning_rate": 6.6836e-06, "loss": 0.0339, "step": 16583 }, { "epoch": 0.884149917364184, "grad_norm": 3.4087226390838623, "learning_rate": 6.683400000000001e-06, "loss": 0.0503, "step": 16584 }, { "epoch": 0.884203230793837, "grad_norm": 2.0902445316314697, "learning_rate": 6.683200000000001e-06, "loss": 0.0181, "step": 16585 }, { "epoch": 0.8842565442234899, "grad_norm": 2.2822210788726807, "learning_rate": 6.683e-06, "loss": 0.0434, "step": 16586 }, { "epoch": 0.8843098576531429, "grad_norm": 2.8411362171173096, "learning_rate": 6.682800000000001e-06, "loss": 0.0787, "step": 16587 }, { "epoch": 0.8843631710827957, "grad_norm": 2.1203415393829346, "learning_rate": 6.6826e-06, "loss": 0.0216, "step": 16588 }, { "epoch": 0.8844164845124487, "grad_norm": 2.470301389694214, "learning_rate": 6.6824e-06, "loss": 0.0501, "step": 16589 }, { "epoch": 0.8844697979421016, "grad_norm": 1.4298826456069946, "learning_rate": 6.682200000000001e-06, "loss": 0.0141, "step": 16590 }, { "epoch": 0.8845231113717545, "grad_norm": 1.8797913789749146, "learning_rate": 6.6820000000000006e-06, "loss": 0.0183, "step": 16591 }, { "epoch": 0.8845764248014075, "grad_norm": 2.1148102283477783, "learning_rate": 6.6818e-06, "loss": 0.0114, "step": 16592 }, { "epoch": 0.8846297382310604, "grad_norm": 0.675781786441803, "learning_rate": 6.6816000000000005e-06, "loss": 0.0085, "step": 16593 }, { "epoch": 0.8846830516607134, "grad_norm": 1.88828444480896, "learning_rate": 6.6814e-06, "loss": 0.0319, "step": 16594 }, { "epoch": 0.8847363650903662, "grad_norm": 2.126716375350952, "learning_rate": 6.681200000000001e-06, "loss": 0.0267, "step": 16595 }, { "epoch": 0.8847896785200192, "grad_norm": 0.8546549677848816, "learning_rate": 6.681000000000001e-06, "loss": 0.0105, "step": 16596 }, { "epoch": 0.8848429919496721, "grad_norm": 2.297245502471924, "learning_rate": 6.6808000000000005e-06, "loss": 0.0162, "step": 16597 }, { "epoch": 0.8848963053793251, "grad_norm": 2.568887710571289, "learning_rate": 6.6806e-06, "loss": 0.0498, "step": 16598 }, { "epoch": 0.8849496188089779, "grad_norm": 1.3447941541671753, "learning_rate": 6.6804000000000004e-06, "loss": 0.027, "step": 16599 }, { "epoch": 0.8850029322386309, "grad_norm": 1.5020891427993774, "learning_rate": 6.6802e-06, "loss": 0.024, "step": 16600 }, { "epoch": 0.8850562456682839, "grad_norm": 2.4217498302459717, "learning_rate": 6.680000000000001e-06, "loss": 0.0373, "step": 16601 }, { "epoch": 0.8851095590979368, "grad_norm": 2.6250081062316895, "learning_rate": 6.679800000000001e-06, "loss": 0.0396, "step": 16602 }, { "epoch": 0.8851628725275897, "grad_norm": 2.6031699180603027, "learning_rate": 6.6796e-06, "loss": 0.0512, "step": 16603 }, { "epoch": 0.8852161859572426, "grad_norm": 3.402088165283203, "learning_rate": 6.6794e-06, "loss": 0.065, "step": 16604 }, { "epoch": 0.8852694993868956, "grad_norm": 2.314364433288574, "learning_rate": 6.6792e-06, "loss": 0.0351, "step": 16605 }, { "epoch": 0.8853228128165485, "grad_norm": 1.677546739578247, "learning_rate": 6.679000000000001e-06, "loss": 0.0262, "step": 16606 }, { "epoch": 0.8853761262462014, "grad_norm": 1.2053462266921997, "learning_rate": 6.678800000000001e-06, "loss": 0.0319, "step": 16607 }, { "epoch": 0.8854294396758543, "grad_norm": 1.6523627042770386, "learning_rate": 6.678600000000001e-06, "loss": 0.0227, "step": 16608 }, { "epoch": 0.8854827531055073, "grad_norm": 1.69058358669281, "learning_rate": 6.6784e-06, "loss": 0.0209, "step": 16609 }, { "epoch": 0.8855360665351603, "grad_norm": 2.3495001792907715, "learning_rate": 6.6782e-06, "loss": 0.0162, "step": 16610 }, { "epoch": 0.8855893799648131, "grad_norm": 2.0700814723968506, "learning_rate": 6.678e-06, "loss": 0.0464, "step": 16611 }, { "epoch": 0.8856426933944661, "grad_norm": 1.3014914989471436, "learning_rate": 6.677800000000001e-06, "loss": 0.0254, "step": 16612 }, { "epoch": 0.885696006824119, "grad_norm": 1.7235908508300781, "learning_rate": 6.677600000000001e-06, "loss": 0.0143, "step": 16613 }, { "epoch": 0.885749320253772, "grad_norm": 1.3370603322982788, "learning_rate": 6.677400000000001e-06, "loss": 0.0174, "step": 16614 }, { "epoch": 0.8858026336834248, "grad_norm": 2.088545799255371, "learning_rate": 6.6772e-06, "loss": 0.0324, "step": 16615 }, { "epoch": 0.8858559471130778, "grad_norm": 0.9258409142494202, "learning_rate": 6.677e-06, "loss": 0.0091, "step": 16616 }, { "epoch": 0.8859092605427307, "grad_norm": 1.1397820711135864, "learning_rate": 6.676800000000001e-06, "loss": 0.0095, "step": 16617 }, { "epoch": 0.8859625739723836, "grad_norm": 2.2040457725524902, "learning_rate": 6.676600000000001e-06, "loss": 0.0183, "step": 16618 }, { "epoch": 0.8860158874020366, "grad_norm": 0.7817860841751099, "learning_rate": 6.676400000000001e-06, "loss": 0.0069, "step": 16619 }, { "epoch": 0.8860692008316895, "grad_norm": 2.659487724304199, "learning_rate": 6.6762000000000006e-06, "loss": 0.0299, "step": 16620 }, { "epoch": 0.8861225142613425, "grad_norm": 2.234282970428467, "learning_rate": 6.676e-06, "loss": 0.023, "step": 16621 }, { "epoch": 0.8861758276909953, "grad_norm": 2.4325592517852783, "learning_rate": 6.6758e-06, "loss": 0.0717, "step": 16622 }, { "epoch": 0.8862291411206483, "grad_norm": 1.4207943677902222, "learning_rate": 6.675600000000001e-06, "loss": 0.0214, "step": 16623 }, { "epoch": 0.8862824545503012, "grad_norm": 1.5843756198883057, "learning_rate": 6.6754000000000005e-06, "loss": 0.0118, "step": 16624 }, { "epoch": 0.8863357679799542, "grad_norm": 1.6073691844940186, "learning_rate": 6.675200000000001e-06, "loss": 0.0329, "step": 16625 }, { "epoch": 0.886389081409607, "grad_norm": 1.7874889373779297, "learning_rate": 6.6750000000000005e-06, "loss": 0.0319, "step": 16626 }, { "epoch": 0.88644239483926, "grad_norm": 1.6196099519729614, "learning_rate": 6.6748e-06, "loss": 0.0367, "step": 16627 }, { "epoch": 0.8864957082689129, "grad_norm": 1.5791577100753784, "learning_rate": 6.674600000000001e-06, "loss": 0.0328, "step": 16628 }, { "epoch": 0.8865490216985659, "grad_norm": 2.661461591720581, "learning_rate": 6.674400000000001e-06, "loss": 0.0444, "step": 16629 }, { "epoch": 0.8866023351282188, "grad_norm": 1.651021122932434, "learning_rate": 6.6742e-06, "loss": 0.0261, "step": 16630 }, { "epoch": 0.8866556485578717, "grad_norm": 3.631967306137085, "learning_rate": 6.674000000000001e-06, "loss": 0.0323, "step": 16631 }, { "epoch": 0.8867089619875247, "grad_norm": 1.587178111076355, "learning_rate": 6.6738e-06, "loss": 0.0348, "step": 16632 }, { "epoch": 0.8867622754171776, "grad_norm": 3.644411325454712, "learning_rate": 6.6736e-06, "loss": 0.0297, "step": 16633 }, { "epoch": 0.8868155888468305, "grad_norm": 1.7609543800354004, "learning_rate": 6.673400000000001e-06, "loss": 0.017, "step": 16634 }, { "epoch": 0.8868689022764834, "grad_norm": 1.6178034543991089, "learning_rate": 6.673200000000001e-06, "loss": 0.0168, "step": 16635 }, { "epoch": 0.8869222157061364, "grad_norm": 2.524630308151245, "learning_rate": 6.673e-06, "loss": 0.0316, "step": 16636 }, { "epoch": 0.8869755291357893, "grad_norm": 1.6832432746887207, "learning_rate": 6.672800000000001e-06, "loss": 0.0185, "step": 16637 }, { "epoch": 0.8870288425654422, "grad_norm": 2.6647186279296875, "learning_rate": 6.6726e-06, "loss": 0.0469, "step": 16638 }, { "epoch": 0.8870821559950952, "grad_norm": 2.6747448444366455, "learning_rate": 6.672400000000001e-06, "loss": 0.0287, "step": 16639 }, { "epoch": 0.8871354694247481, "grad_norm": 2.324873208999634, "learning_rate": 6.672200000000001e-06, "loss": 0.0202, "step": 16640 }, { "epoch": 0.8871887828544011, "grad_norm": 2.945253610610962, "learning_rate": 6.672000000000001e-06, "loss": 0.0341, "step": 16641 }, { "epoch": 0.8872420962840539, "grad_norm": 1.7803864479064941, "learning_rate": 6.6718e-06, "loss": 0.0328, "step": 16642 }, { "epoch": 0.8872954097137069, "grad_norm": 1.067244052886963, "learning_rate": 6.671600000000001e-06, "loss": 0.0182, "step": 16643 }, { "epoch": 0.8873487231433598, "grad_norm": 2.7667715549468994, "learning_rate": 6.6714e-06, "loss": 0.0585, "step": 16644 }, { "epoch": 0.8874020365730128, "grad_norm": 2.3022680282592773, "learning_rate": 6.671200000000001e-06, "loss": 0.0438, "step": 16645 }, { "epoch": 0.8874553500026656, "grad_norm": 2.335963487625122, "learning_rate": 6.671000000000001e-06, "loss": 0.0169, "step": 16646 }, { "epoch": 0.8875086634323186, "grad_norm": 2.351318597793579, "learning_rate": 6.6708000000000006e-06, "loss": 0.0255, "step": 16647 }, { "epoch": 0.8875619768619716, "grad_norm": 2.2806568145751953, "learning_rate": 6.6706e-06, "loss": 0.0392, "step": 16648 }, { "epoch": 0.8876152902916244, "grad_norm": 1.7586697340011597, "learning_rate": 6.6704000000000005e-06, "loss": 0.0447, "step": 16649 }, { "epoch": 0.8876686037212774, "grad_norm": 1.0619838237762451, "learning_rate": 6.6702e-06, "loss": 0.0121, "step": 16650 }, { "epoch": 0.8877219171509303, "grad_norm": 3.8871843814849854, "learning_rate": 6.6700000000000005e-06, "loss": 0.0487, "step": 16651 }, { "epoch": 0.8877752305805833, "grad_norm": 2.6937079429626465, "learning_rate": 6.669800000000001e-06, "loss": 0.0529, "step": 16652 }, { "epoch": 0.8878285440102361, "grad_norm": 2.6465985774993896, "learning_rate": 6.6696000000000005e-06, "loss": 0.0277, "step": 16653 }, { "epoch": 0.8878818574398891, "grad_norm": 1.5171301364898682, "learning_rate": 6.6694e-06, "loss": 0.0206, "step": 16654 }, { "epoch": 0.887935170869542, "grad_norm": 3.1407182216644287, "learning_rate": 6.6692000000000005e-06, "loss": 0.0389, "step": 16655 }, { "epoch": 0.887988484299195, "grad_norm": 3.598439931869507, "learning_rate": 6.669000000000001e-06, "loss": 0.0721, "step": 16656 }, { "epoch": 0.888041797728848, "grad_norm": 2.336311101913452, "learning_rate": 6.6688e-06, "loss": 0.0399, "step": 16657 }, { "epoch": 0.8880951111585008, "grad_norm": 1.2399545907974243, "learning_rate": 6.668600000000001e-06, "loss": 0.0158, "step": 16658 }, { "epoch": 0.8881484245881538, "grad_norm": 4.091381072998047, "learning_rate": 6.6684e-06, "loss": 0.0487, "step": 16659 }, { "epoch": 0.8882017380178067, "grad_norm": 2.7686636447906494, "learning_rate": 6.6682e-06, "loss": 0.0427, "step": 16660 }, { "epoch": 0.8882550514474596, "grad_norm": 1.7255158424377441, "learning_rate": 6.668e-06, "loss": 0.0443, "step": 16661 }, { "epoch": 0.8883083648771125, "grad_norm": 1.552865743637085, "learning_rate": 6.667800000000001e-06, "loss": 0.042, "step": 16662 }, { "epoch": 0.8883616783067655, "grad_norm": 1.9498611688613892, "learning_rate": 6.6676e-06, "loss": 0.0245, "step": 16663 }, { "epoch": 0.8884149917364184, "grad_norm": 1.5433745384216309, "learning_rate": 6.667400000000001e-06, "loss": 0.0235, "step": 16664 }, { "epoch": 0.8884683051660713, "grad_norm": 1.043196678161621, "learning_rate": 6.6672e-06, "loss": 0.0214, "step": 16665 }, { "epoch": 0.8885216185957243, "grad_norm": 1.1334865093231201, "learning_rate": 6.667e-06, "loss": 0.0193, "step": 16666 }, { "epoch": 0.8885749320253772, "grad_norm": 2.272163152694702, "learning_rate": 6.666800000000001e-06, "loss": 0.0549, "step": 16667 }, { "epoch": 0.8886282454550302, "grad_norm": 2.065296173095703, "learning_rate": 6.666600000000001e-06, "loss": 0.022, "step": 16668 }, { "epoch": 0.888681558884683, "grad_norm": 1.0203765630722046, "learning_rate": 6.6664e-06, "loss": 0.0129, "step": 16669 }, { "epoch": 0.888734872314336, "grad_norm": 1.4077434539794922, "learning_rate": 6.666200000000001e-06, "loss": 0.0307, "step": 16670 }, { "epoch": 0.8887881857439889, "grad_norm": 1.7070516347885132, "learning_rate": 6.666e-06, "loss": 0.0322, "step": 16671 }, { "epoch": 0.8888414991736419, "grad_norm": 1.8115055561065674, "learning_rate": 6.6658e-06, "loss": 0.0342, "step": 16672 }, { "epoch": 0.8888948126032947, "grad_norm": 1.967663288116455, "learning_rate": 6.665600000000001e-06, "loss": 0.0413, "step": 16673 }, { "epoch": 0.8889481260329477, "grad_norm": 1.706088662147522, "learning_rate": 6.665400000000001e-06, "loss": 0.0241, "step": 16674 }, { "epoch": 0.8890014394626007, "grad_norm": 1.9067535400390625, "learning_rate": 6.6652e-06, "loss": 0.033, "step": 16675 }, { "epoch": 0.8890547528922536, "grad_norm": 2.0595362186431885, "learning_rate": 6.6650000000000006e-06, "loss": 0.0457, "step": 16676 }, { "epoch": 0.8891080663219065, "grad_norm": 1.8820489645004272, "learning_rate": 6.6648e-06, "loss": 0.0241, "step": 16677 }, { "epoch": 0.8891613797515594, "grad_norm": 3.3156063556671143, "learning_rate": 6.664600000000001e-06, "loss": 0.0301, "step": 16678 }, { "epoch": 0.8892146931812124, "grad_norm": 1.0032355785369873, "learning_rate": 6.664400000000001e-06, "loss": 0.0089, "step": 16679 }, { "epoch": 0.8892680066108652, "grad_norm": 2.3740503787994385, "learning_rate": 6.6642000000000005e-06, "loss": 0.0538, "step": 16680 }, { "epoch": 0.8893213200405182, "grad_norm": 2.315120220184326, "learning_rate": 6.664e-06, "loss": 0.0332, "step": 16681 }, { "epoch": 0.8893746334701711, "grad_norm": 1.7607988119125366, "learning_rate": 6.6638000000000005e-06, "loss": 0.0231, "step": 16682 }, { "epoch": 0.8894279468998241, "grad_norm": 1.5755618810653687, "learning_rate": 6.6636e-06, "loss": 0.0309, "step": 16683 }, { "epoch": 0.8894812603294769, "grad_norm": 2.5252389907836914, "learning_rate": 6.663400000000001e-06, "loss": 0.0226, "step": 16684 }, { "epoch": 0.8895345737591299, "grad_norm": 1.4970165491104126, "learning_rate": 6.663200000000001e-06, "loss": 0.0227, "step": 16685 }, { "epoch": 0.8895878871887829, "grad_norm": 1.5115858316421509, "learning_rate": 6.663e-06, "loss": 0.0185, "step": 16686 }, { "epoch": 0.8896412006184358, "grad_norm": 1.8876947164535522, "learning_rate": 6.6628e-06, "loss": 0.011, "step": 16687 }, { "epoch": 0.8896945140480887, "grad_norm": 1.1919410228729248, "learning_rate": 6.6626e-06, "loss": 0.0205, "step": 16688 }, { "epoch": 0.8897478274777416, "grad_norm": 3.621884822845459, "learning_rate": 6.662400000000001e-06, "loss": 0.0498, "step": 16689 }, { "epoch": 0.8898011409073946, "grad_norm": 3.6701653003692627, "learning_rate": 6.662200000000001e-06, "loss": 0.044, "step": 16690 }, { "epoch": 0.8898544543370475, "grad_norm": 2.2723536491394043, "learning_rate": 6.662000000000001e-06, "loss": 0.0381, "step": 16691 }, { "epoch": 0.8899077677667004, "grad_norm": 3.605027675628662, "learning_rate": 6.6618e-06, "loss": 0.0351, "step": 16692 }, { "epoch": 0.8899610811963533, "grad_norm": 1.5426440238952637, "learning_rate": 6.6616e-06, "loss": 0.0376, "step": 16693 }, { "epoch": 0.8900143946260063, "grad_norm": 2.6258695125579834, "learning_rate": 6.6614e-06, "loss": 0.0309, "step": 16694 }, { "epoch": 0.8900677080556593, "grad_norm": 2.654399871826172, "learning_rate": 6.661200000000001e-06, "loss": 0.0519, "step": 16695 }, { "epoch": 0.8901210214853121, "grad_norm": 2.3259263038635254, "learning_rate": 6.661000000000001e-06, "loss": 0.0585, "step": 16696 }, { "epoch": 0.8901743349149651, "grad_norm": 2.4712047576904297, "learning_rate": 6.660800000000001e-06, "loss": 0.0399, "step": 16697 }, { "epoch": 0.890227648344618, "grad_norm": 2.9384336471557617, "learning_rate": 6.6606e-06, "loss": 0.0437, "step": 16698 }, { "epoch": 0.890280961774271, "grad_norm": 0.9493128657341003, "learning_rate": 6.6604e-06, "loss": 0.0102, "step": 16699 }, { "epoch": 0.8903342752039238, "grad_norm": 2.164666175842285, "learning_rate": 6.6602e-06, "loss": 0.0413, "step": 16700 }, { "epoch": 0.8903875886335768, "grad_norm": 1.6939468383789062, "learning_rate": 6.660000000000001e-06, "loss": 0.0254, "step": 16701 }, { "epoch": 0.8904409020632297, "grad_norm": 1.7838950157165527, "learning_rate": 6.659800000000001e-06, "loss": 0.0252, "step": 16702 }, { "epoch": 0.8904942154928827, "grad_norm": 1.2100098133087158, "learning_rate": 6.659600000000001e-06, "loss": 0.0122, "step": 16703 }, { "epoch": 0.8905475289225356, "grad_norm": 1.626658320426941, "learning_rate": 6.6594e-06, "loss": 0.0213, "step": 16704 }, { "epoch": 0.8906008423521885, "grad_norm": 1.5914146900177002, "learning_rate": 6.6592e-06, "loss": 0.023, "step": 16705 }, { "epoch": 0.8906541557818415, "grad_norm": 2.481717586517334, "learning_rate": 6.659000000000001e-06, "loss": 0.0335, "step": 16706 }, { "epoch": 0.8907074692114944, "grad_norm": 2.0093343257904053, "learning_rate": 6.6588000000000005e-06, "loss": 0.016, "step": 16707 }, { "epoch": 0.8907607826411473, "grad_norm": 2.232853889465332, "learning_rate": 6.658600000000001e-06, "loss": 0.0323, "step": 16708 }, { "epoch": 0.8908140960708002, "grad_norm": 1.5594133138656616, "learning_rate": 6.6584000000000005e-06, "loss": 0.0158, "step": 16709 }, { "epoch": 0.8908674095004532, "grad_norm": 1.6456022262573242, "learning_rate": 6.6582e-06, "loss": 0.0409, "step": 16710 }, { "epoch": 0.890920722930106, "grad_norm": 1.8290804624557495, "learning_rate": 6.658e-06, "loss": 0.0177, "step": 16711 }, { "epoch": 0.890974036359759, "grad_norm": 2.159646987915039, "learning_rate": 6.657800000000001e-06, "loss": 0.0292, "step": 16712 }, { "epoch": 0.891027349789412, "grad_norm": 1.875350832939148, "learning_rate": 6.6576000000000004e-06, "loss": 0.0338, "step": 16713 }, { "epoch": 0.8910806632190649, "grad_norm": 0.701239824295044, "learning_rate": 6.657400000000001e-06, "loss": 0.0077, "step": 16714 }, { "epoch": 0.8911339766487179, "grad_norm": 1.316726803779602, "learning_rate": 6.6572e-06, "loss": 0.0157, "step": 16715 }, { "epoch": 0.8911872900783707, "grad_norm": 2.3289027214050293, "learning_rate": 6.657e-06, "loss": 0.0516, "step": 16716 }, { "epoch": 0.8912406035080237, "grad_norm": 0.7089751362800598, "learning_rate": 6.656800000000001e-06, "loss": 0.0066, "step": 16717 }, { "epoch": 0.8912939169376766, "grad_norm": 1.5455485582351685, "learning_rate": 6.656600000000001e-06, "loss": 0.0204, "step": 16718 }, { "epoch": 0.8913472303673295, "grad_norm": 0.7773829102516174, "learning_rate": 6.6564e-06, "loss": 0.0105, "step": 16719 }, { "epoch": 0.8914005437969824, "grad_norm": 1.4732892513275146, "learning_rate": 6.656200000000001e-06, "loss": 0.0198, "step": 16720 }, { "epoch": 0.8914538572266354, "grad_norm": 1.9325155019760132, "learning_rate": 6.656e-06, "loss": 0.0294, "step": 16721 }, { "epoch": 0.8915071706562884, "grad_norm": 5.507991313934326, "learning_rate": 6.6558e-06, "loss": 0.0706, "step": 16722 }, { "epoch": 0.8915604840859412, "grad_norm": 2.161616325378418, "learning_rate": 6.655600000000001e-06, "loss": 0.0354, "step": 16723 }, { "epoch": 0.8916137975155942, "grad_norm": 1.1162760257720947, "learning_rate": 6.655400000000001e-06, "loss": 0.0129, "step": 16724 }, { "epoch": 0.8916671109452471, "grad_norm": 0.659750759601593, "learning_rate": 6.6552e-06, "loss": 0.0089, "step": 16725 }, { "epoch": 0.8917204243749001, "grad_norm": 2.7170732021331787, "learning_rate": 6.655000000000001e-06, "loss": 0.0382, "step": 16726 }, { "epoch": 0.8917737378045529, "grad_norm": 1.9343483448028564, "learning_rate": 6.6548e-06, "loss": 0.0263, "step": 16727 }, { "epoch": 0.8918270512342059, "grad_norm": 2.2573838233947754, "learning_rate": 6.6546000000000015e-06, "loss": 0.0368, "step": 16728 }, { "epoch": 0.8918803646638588, "grad_norm": 2.812100648880005, "learning_rate": 6.654400000000001e-06, "loss": 0.041, "step": 16729 }, { "epoch": 0.8919336780935118, "grad_norm": 1.1439977884292603, "learning_rate": 6.654200000000001e-06, "loss": 0.0372, "step": 16730 }, { "epoch": 0.8919869915231647, "grad_norm": 1.9671597480773926, "learning_rate": 6.654e-06, "loss": 0.0437, "step": 16731 }, { "epoch": 0.8920403049528176, "grad_norm": 1.6542773246765137, "learning_rate": 6.6538000000000006e-06, "loss": 0.0357, "step": 16732 }, { "epoch": 0.8920936183824706, "grad_norm": 2.0537922382354736, "learning_rate": 6.6536e-06, "loss": 0.0439, "step": 16733 }, { "epoch": 0.8921469318121235, "grad_norm": 1.879190444946289, "learning_rate": 6.653400000000001e-06, "loss": 0.0345, "step": 16734 }, { "epoch": 0.8922002452417764, "grad_norm": 1.6074074506759644, "learning_rate": 6.653200000000001e-06, "loss": 0.0319, "step": 16735 }, { "epoch": 0.8922535586714293, "grad_norm": 2.590278387069702, "learning_rate": 6.6530000000000005e-06, "loss": 0.0165, "step": 16736 }, { "epoch": 0.8923068721010823, "grad_norm": 4.937159061431885, "learning_rate": 6.6528e-06, "loss": 0.0221, "step": 16737 }, { "epoch": 0.8923601855307352, "grad_norm": 2.3267905712127686, "learning_rate": 6.6526000000000005e-06, "loss": 0.0264, "step": 16738 }, { "epoch": 0.8924134989603881, "grad_norm": 1.918731689453125, "learning_rate": 6.652400000000001e-06, "loss": 0.0457, "step": 16739 }, { "epoch": 0.892466812390041, "grad_norm": 2.362736701965332, "learning_rate": 6.652200000000001e-06, "loss": 0.0465, "step": 16740 }, { "epoch": 0.892520125819694, "grad_norm": 2.0210561752319336, "learning_rate": 6.652000000000001e-06, "loss": 0.0294, "step": 16741 }, { "epoch": 0.892573439249347, "grad_norm": 3.378107786178589, "learning_rate": 6.6518e-06, "loss": 0.0308, "step": 16742 }, { "epoch": 0.8926267526789998, "grad_norm": 2.4187004566192627, "learning_rate": 6.6516e-06, "loss": 0.0189, "step": 16743 }, { "epoch": 0.8926800661086528, "grad_norm": 1.528952717781067, "learning_rate": 6.6514e-06, "loss": 0.0273, "step": 16744 }, { "epoch": 0.8927333795383057, "grad_norm": 2.8943300247192383, "learning_rate": 6.651200000000001e-06, "loss": 0.0331, "step": 16745 }, { "epoch": 0.8927866929679587, "grad_norm": 2.445528507232666, "learning_rate": 6.651000000000001e-06, "loss": 0.0236, "step": 16746 }, { "epoch": 0.8928400063976115, "grad_norm": 2.442754030227661, "learning_rate": 6.650800000000001e-06, "loss": 0.0287, "step": 16747 }, { "epoch": 0.8928933198272645, "grad_norm": 2.42722749710083, "learning_rate": 6.6506e-06, "loss": 0.0318, "step": 16748 }, { "epoch": 0.8929466332569174, "grad_norm": 1.1523348093032837, "learning_rate": 6.6504e-06, "loss": 0.0223, "step": 16749 }, { "epoch": 0.8929999466865703, "grad_norm": 2.752495050430298, "learning_rate": 6.6502e-06, "loss": 0.0246, "step": 16750 }, { "epoch": 0.8930532601162233, "grad_norm": 1.0758411884307861, "learning_rate": 6.650000000000001e-06, "loss": 0.017, "step": 16751 }, { "epoch": 0.8931065735458762, "grad_norm": 1.6282678842544556, "learning_rate": 6.649800000000001e-06, "loss": 0.0205, "step": 16752 }, { "epoch": 0.8931598869755292, "grad_norm": 3.0393190383911133, "learning_rate": 6.649600000000001e-06, "loss": 0.0448, "step": 16753 }, { "epoch": 0.893213200405182, "grad_norm": 1.6504225730895996, "learning_rate": 6.6494e-06, "loss": 0.0161, "step": 16754 }, { "epoch": 0.893266513834835, "grad_norm": 1.312987208366394, "learning_rate": 6.6492e-06, "loss": 0.0118, "step": 16755 }, { "epoch": 0.8933198272644879, "grad_norm": 1.508285403251648, "learning_rate": 6.649000000000001e-06, "loss": 0.0188, "step": 16756 }, { "epoch": 0.8933731406941409, "grad_norm": 2.8560962677001953, "learning_rate": 6.648800000000001e-06, "loss": 0.0231, "step": 16757 }, { "epoch": 0.8934264541237937, "grad_norm": 0.5254455804824829, "learning_rate": 6.648600000000001e-06, "loss": 0.0053, "step": 16758 }, { "epoch": 0.8934797675534467, "grad_norm": 0.520582377910614, "learning_rate": 6.648400000000001e-06, "loss": 0.0064, "step": 16759 }, { "epoch": 0.8935330809830997, "grad_norm": 2.814563274383545, "learning_rate": 6.6482e-06, "loss": 0.053, "step": 16760 }, { "epoch": 0.8935863944127526, "grad_norm": 2.982632875442505, "learning_rate": 6.648e-06, "loss": 0.0346, "step": 16761 }, { "epoch": 0.8936397078424055, "grad_norm": 1.3205281496047974, "learning_rate": 6.647800000000001e-06, "loss": 0.014, "step": 16762 }, { "epoch": 0.8936930212720584, "grad_norm": 1.6308974027633667, "learning_rate": 6.6476000000000005e-06, "loss": 0.0238, "step": 16763 }, { "epoch": 0.8937463347017114, "grad_norm": 1.7895971536636353, "learning_rate": 6.647400000000001e-06, "loss": 0.0262, "step": 16764 }, { "epoch": 0.8937996481313643, "grad_norm": 1.7576388120651245, "learning_rate": 6.6472000000000005e-06, "loss": 0.0181, "step": 16765 }, { "epoch": 0.8938529615610172, "grad_norm": 1.7024701833724976, "learning_rate": 6.647e-06, "loss": 0.0227, "step": 16766 }, { "epoch": 0.8939062749906701, "grad_norm": 1.9443330764770508, "learning_rate": 6.646800000000001e-06, "loss": 0.035, "step": 16767 }, { "epoch": 0.8939595884203231, "grad_norm": 2.1635591983795166, "learning_rate": 6.646600000000001e-06, "loss": 0.0299, "step": 16768 }, { "epoch": 0.8940129018499761, "grad_norm": 1.7110506296157837, "learning_rate": 6.6464000000000004e-06, "loss": 0.0109, "step": 16769 }, { "epoch": 0.8940662152796289, "grad_norm": 2.4768171310424805, "learning_rate": 6.646200000000001e-06, "loss": 0.0304, "step": 16770 }, { "epoch": 0.8941195287092819, "grad_norm": 2.559305191040039, "learning_rate": 6.646e-06, "loss": 0.0444, "step": 16771 }, { "epoch": 0.8941728421389348, "grad_norm": 2.487525701522827, "learning_rate": 6.6458e-06, "loss": 0.0453, "step": 16772 }, { "epoch": 0.8942261555685878, "grad_norm": 2.5988426208496094, "learning_rate": 6.645600000000001e-06, "loss": 0.0493, "step": 16773 }, { "epoch": 0.8942794689982406, "grad_norm": 8.335728645324707, "learning_rate": 6.645400000000001e-06, "loss": 0.0871, "step": 16774 }, { "epoch": 0.8943327824278936, "grad_norm": 2.1545002460479736, "learning_rate": 6.6452e-06, "loss": 0.023, "step": 16775 }, { "epoch": 0.8943860958575465, "grad_norm": 2.8263046741485596, "learning_rate": 6.645000000000001e-06, "loss": 0.0502, "step": 16776 }, { "epoch": 0.8944394092871994, "grad_norm": 3.134303569793701, "learning_rate": 6.6448e-06, "loss": 0.0542, "step": 16777 }, { "epoch": 0.8944927227168524, "grad_norm": 1.2399886846542358, "learning_rate": 6.644600000000001e-06, "loss": 0.0344, "step": 16778 }, { "epoch": 0.8945460361465053, "grad_norm": 1.988906979560852, "learning_rate": 6.644400000000001e-06, "loss": 0.0429, "step": 16779 }, { "epoch": 0.8945993495761583, "grad_norm": 1.5038135051727295, "learning_rate": 6.644200000000001e-06, "loss": 0.0403, "step": 16780 }, { "epoch": 0.8946526630058111, "grad_norm": 0.933944582939148, "learning_rate": 6.644e-06, "loss": 0.0193, "step": 16781 }, { "epoch": 0.8947059764354641, "grad_norm": 1.5080903768539429, "learning_rate": 6.643800000000001e-06, "loss": 0.0472, "step": 16782 }, { "epoch": 0.894759289865117, "grad_norm": 1.272027611732483, "learning_rate": 6.6436e-06, "loss": 0.011, "step": 16783 }, { "epoch": 0.89481260329477, "grad_norm": 2.749333620071411, "learning_rate": 6.643400000000001e-06, "loss": 0.0392, "step": 16784 }, { "epoch": 0.8948659167244228, "grad_norm": 1.7975318431854248, "learning_rate": 6.643200000000001e-06, "loss": 0.0124, "step": 16785 }, { "epoch": 0.8949192301540758, "grad_norm": 1.5920110940933228, "learning_rate": 6.643000000000001e-06, "loss": 0.0471, "step": 16786 }, { "epoch": 0.8949725435837287, "grad_norm": 1.531005859375, "learning_rate": 6.6428e-06, "loss": 0.0188, "step": 16787 }, { "epoch": 0.8950258570133817, "grad_norm": 1.2522720098495483, "learning_rate": 6.6426000000000006e-06, "loss": 0.0167, "step": 16788 }, { "epoch": 0.8950791704430346, "grad_norm": 0.8562897443771362, "learning_rate": 6.642400000000001e-06, "loss": 0.012, "step": 16789 }, { "epoch": 0.8951324838726875, "grad_norm": 2.0147972106933594, "learning_rate": 6.6422000000000005e-06, "loss": 0.063, "step": 16790 }, { "epoch": 0.8951857973023405, "grad_norm": 1.1113531589508057, "learning_rate": 6.642000000000001e-06, "loss": 0.0146, "step": 16791 }, { "epoch": 0.8952391107319934, "grad_norm": 1.669382095336914, "learning_rate": 6.6418000000000005e-06, "loss": 0.0375, "step": 16792 }, { "epoch": 0.8952924241616463, "grad_norm": 1.55794095993042, "learning_rate": 6.6416e-06, "loss": 0.0344, "step": 16793 }, { "epoch": 0.8953457375912992, "grad_norm": 1.585935115814209, "learning_rate": 6.6414000000000005e-06, "loss": 0.0304, "step": 16794 }, { "epoch": 0.8953990510209522, "grad_norm": 2.2534263134002686, "learning_rate": 6.641200000000001e-06, "loss": 0.0329, "step": 16795 }, { "epoch": 0.895452364450605, "grad_norm": 3.6841471195220947, "learning_rate": 6.6410000000000005e-06, "loss": 0.0391, "step": 16796 }, { "epoch": 0.895505677880258, "grad_norm": 0.9712167382240295, "learning_rate": 6.640800000000001e-06, "loss": 0.0176, "step": 16797 }, { "epoch": 0.895558991309911, "grad_norm": 1.7534550428390503, "learning_rate": 6.6406e-06, "loss": 0.0313, "step": 16798 }, { "epoch": 0.8956123047395639, "grad_norm": 2.447659730911255, "learning_rate": 6.6404e-06, "loss": 0.0447, "step": 16799 }, { "epoch": 0.8956656181692169, "grad_norm": 2.9615285396575928, "learning_rate": 6.640200000000001e-06, "loss": 0.0425, "step": 16800 }, { "epoch": 0.8957189315988697, "grad_norm": 2.167241096496582, "learning_rate": 6.640000000000001e-06, "loss": 0.0214, "step": 16801 }, { "epoch": 0.8957722450285227, "grad_norm": 1.9435553550720215, "learning_rate": 6.6398e-06, "loss": 0.0329, "step": 16802 }, { "epoch": 0.8958255584581756, "grad_norm": 1.1592851877212524, "learning_rate": 6.639600000000001e-06, "loss": 0.0151, "step": 16803 }, { "epoch": 0.8958788718878286, "grad_norm": 4.623887538909912, "learning_rate": 6.6394e-06, "loss": 0.039, "step": 16804 }, { "epoch": 0.8959321853174814, "grad_norm": 1.3673230409622192, "learning_rate": 6.6392e-06, "loss": 0.0375, "step": 16805 }, { "epoch": 0.8959854987471344, "grad_norm": 1.8922040462493896, "learning_rate": 6.639000000000001e-06, "loss": 0.018, "step": 16806 }, { "epoch": 0.8960388121767874, "grad_norm": 2.3052315711975098, "learning_rate": 6.638800000000001e-06, "loss": 0.0379, "step": 16807 }, { "epoch": 0.8960921256064402, "grad_norm": 4.350467681884766, "learning_rate": 6.6386e-06, "loss": 0.0335, "step": 16808 }, { "epoch": 0.8961454390360932, "grad_norm": 1.4169988632202148, "learning_rate": 6.638400000000001e-06, "loss": 0.0182, "step": 16809 }, { "epoch": 0.8961987524657461, "grad_norm": 0.8085911870002747, "learning_rate": 6.6382e-06, "loss": 0.0206, "step": 16810 }, { "epoch": 0.8962520658953991, "grad_norm": 1.4183341264724731, "learning_rate": 6.638e-06, "loss": 0.0137, "step": 16811 }, { "epoch": 0.8963053793250519, "grad_norm": 2.3195602893829346, "learning_rate": 6.637800000000001e-06, "loss": 0.0419, "step": 16812 }, { "epoch": 0.8963586927547049, "grad_norm": 1.2402132749557495, "learning_rate": 6.637600000000001e-06, "loss": 0.0199, "step": 16813 }, { "epoch": 0.8964120061843578, "grad_norm": 2.564432382583618, "learning_rate": 6.6374e-06, "loss": 0.035, "step": 16814 }, { "epoch": 0.8964653196140108, "grad_norm": 4.785248756408691, "learning_rate": 6.637200000000001e-06, "loss": 0.0311, "step": 16815 }, { "epoch": 0.8965186330436637, "grad_norm": 2.7012691497802734, "learning_rate": 6.637e-06, "loss": 0.0699, "step": 16816 }, { "epoch": 0.8965719464733166, "grad_norm": 1.6809663772583008, "learning_rate": 6.636800000000001e-06, "loss": 0.022, "step": 16817 }, { "epoch": 0.8966252599029696, "grad_norm": 2.8623125553131104, "learning_rate": 6.636600000000001e-06, "loss": 0.0342, "step": 16818 }, { "epoch": 0.8966785733326225, "grad_norm": 1.4597800970077515, "learning_rate": 6.6364000000000005e-06, "loss": 0.0159, "step": 16819 }, { "epoch": 0.8967318867622754, "grad_norm": 5.215802192687988, "learning_rate": 6.6362e-06, "loss": 0.0685, "step": 16820 }, { "epoch": 0.8967852001919283, "grad_norm": 2.2050840854644775, "learning_rate": 6.6360000000000005e-06, "loss": 0.0337, "step": 16821 }, { "epoch": 0.8968385136215813, "grad_norm": 2.124126434326172, "learning_rate": 6.6358e-06, "loss": 0.0368, "step": 16822 }, { "epoch": 0.8968918270512342, "grad_norm": 2.2982490062713623, "learning_rate": 6.635600000000001e-06, "loss": 0.0419, "step": 16823 }, { "epoch": 0.8969451404808871, "grad_norm": 0.936028003692627, "learning_rate": 6.635400000000001e-06, "loss": 0.0106, "step": 16824 }, { "epoch": 0.8969984539105401, "grad_norm": 1.1748417615890503, "learning_rate": 6.6352000000000004e-06, "loss": 0.0289, "step": 16825 }, { "epoch": 0.897051767340193, "grad_norm": 1.2291030883789062, "learning_rate": 6.635e-06, "loss": 0.021, "step": 16826 }, { "epoch": 0.897105080769846, "grad_norm": 2.030489206314087, "learning_rate": 6.6348e-06, "loss": 0.0223, "step": 16827 }, { "epoch": 0.8971583941994988, "grad_norm": 2.6029369831085205, "learning_rate": 6.634600000000001e-06, "loss": 0.0212, "step": 16828 }, { "epoch": 0.8972117076291518, "grad_norm": 1.0257601737976074, "learning_rate": 6.634400000000001e-06, "loss": 0.0082, "step": 16829 }, { "epoch": 0.8972650210588047, "grad_norm": 1.1534960269927979, "learning_rate": 6.634200000000001e-06, "loss": 0.0191, "step": 16830 }, { "epoch": 0.8973183344884577, "grad_norm": 0.9685550928115845, "learning_rate": 6.634e-06, "loss": 0.0091, "step": 16831 }, { "epoch": 0.8973716479181105, "grad_norm": 1.8800275325775146, "learning_rate": 6.6338e-06, "loss": 0.0416, "step": 16832 }, { "epoch": 0.8974249613477635, "grad_norm": 3.9864463806152344, "learning_rate": 6.6336e-06, "loss": 0.0694, "step": 16833 }, { "epoch": 0.8974782747774165, "grad_norm": 2.2173221111297607, "learning_rate": 6.633400000000001e-06, "loss": 0.0175, "step": 16834 }, { "epoch": 0.8975315882070694, "grad_norm": 2.18510365486145, "learning_rate": 6.633200000000001e-06, "loss": 0.0331, "step": 16835 }, { "epoch": 0.8975849016367223, "grad_norm": 0.8692185282707214, "learning_rate": 6.633000000000001e-06, "loss": 0.0049, "step": 16836 }, { "epoch": 0.8976382150663752, "grad_norm": 1.1139920949935913, "learning_rate": 6.6328e-06, "loss": 0.0249, "step": 16837 }, { "epoch": 0.8976915284960282, "grad_norm": 1.8324683904647827, "learning_rate": 6.6326e-06, "loss": 0.0443, "step": 16838 }, { "epoch": 0.897744841925681, "grad_norm": 2.464539051055908, "learning_rate": 6.632400000000001e-06, "loss": 0.0428, "step": 16839 }, { "epoch": 0.897798155355334, "grad_norm": 2.582611322402954, "learning_rate": 6.632200000000001e-06, "loss": 0.0377, "step": 16840 }, { "epoch": 0.8978514687849869, "grad_norm": 2.403656005859375, "learning_rate": 6.632000000000001e-06, "loss": 0.0324, "step": 16841 }, { "epoch": 0.8979047822146399, "grad_norm": 0.8893855810165405, "learning_rate": 6.631800000000001e-06, "loss": 0.0062, "step": 16842 }, { "epoch": 0.8979580956442927, "grad_norm": 1.414993166923523, "learning_rate": 6.6316e-06, "loss": 0.0085, "step": 16843 }, { "epoch": 0.8980114090739457, "grad_norm": 2.410316228866577, "learning_rate": 6.6314e-06, "loss": 0.0337, "step": 16844 }, { "epoch": 0.8980647225035987, "grad_norm": 2.29282283782959, "learning_rate": 6.631200000000001e-06, "loss": 0.0468, "step": 16845 }, { "epoch": 0.8981180359332516, "grad_norm": 1.7921924591064453, "learning_rate": 6.6310000000000005e-06, "loss": 0.0235, "step": 16846 }, { "epoch": 0.8981713493629045, "grad_norm": 2.645368814468384, "learning_rate": 6.630800000000001e-06, "loss": 0.0384, "step": 16847 }, { "epoch": 0.8982246627925574, "grad_norm": 3.046234369277954, "learning_rate": 6.6306000000000005e-06, "loss": 0.0358, "step": 16848 }, { "epoch": 0.8982779762222104, "grad_norm": 1.2541099786758423, "learning_rate": 6.6304e-06, "loss": 0.0183, "step": 16849 }, { "epoch": 0.8983312896518633, "grad_norm": 2.259072780609131, "learning_rate": 6.630200000000001e-06, "loss": 0.0376, "step": 16850 }, { "epoch": 0.8983846030815162, "grad_norm": 1.5198591947555542, "learning_rate": 6.630000000000001e-06, "loss": 0.0201, "step": 16851 }, { "epoch": 0.8984379165111691, "grad_norm": 1.07382333278656, "learning_rate": 6.6298000000000005e-06, "loss": 0.0169, "step": 16852 }, { "epoch": 0.8984912299408221, "grad_norm": 1.0079703330993652, "learning_rate": 6.629600000000001e-06, "loss": 0.0098, "step": 16853 }, { "epoch": 0.8985445433704751, "grad_norm": 1.4072962999343872, "learning_rate": 6.6294000000000004e-06, "loss": 0.0172, "step": 16854 }, { "epoch": 0.8985978568001279, "grad_norm": 2.1910908222198486, "learning_rate": 6.6292e-06, "loss": 0.0251, "step": 16855 }, { "epoch": 0.8986511702297809, "grad_norm": 1.1922221183776855, "learning_rate": 6.629000000000001e-06, "loss": 0.0127, "step": 16856 }, { "epoch": 0.8987044836594338, "grad_norm": 0.9127889275550842, "learning_rate": 6.628800000000001e-06, "loss": 0.0079, "step": 16857 }, { "epoch": 0.8987577970890868, "grad_norm": 1.16053307056427, "learning_rate": 6.6286e-06, "loss": 0.0125, "step": 16858 }, { "epoch": 0.8988111105187396, "grad_norm": 3.168222427368164, "learning_rate": 6.628400000000001e-06, "loss": 0.0701, "step": 16859 }, { "epoch": 0.8988644239483926, "grad_norm": 2.9625368118286133, "learning_rate": 6.6282e-06, "loss": 0.0501, "step": 16860 }, { "epoch": 0.8989177373780455, "grad_norm": 1.3016502857208252, "learning_rate": 6.628e-06, "loss": 0.0186, "step": 16861 }, { "epoch": 0.8989710508076985, "grad_norm": 2.4689531326293945, "learning_rate": 6.627800000000001e-06, "loss": 0.0316, "step": 16862 }, { "epoch": 0.8990243642373514, "grad_norm": 1.1361504793167114, "learning_rate": 6.627600000000001e-06, "loss": 0.0099, "step": 16863 }, { "epoch": 0.8990776776670043, "grad_norm": 1.8761590719223022, "learning_rate": 6.6274e-06, "loss": 0.0408, "step": 16864 }, { "epoch": 0.8991309910966573, "grad_norm": 2.6576125621795654, "learning_rate": 6.627200000000001e-06, "loss": 0.0477, "step": 16865 }, { "epoch": 0.8991843045263102, "grad_norm": 2.228032350540161, "learning_rate": 6.627e-06, "loss": 0.0211, "step": 16866 }, { "epoch": 0.8992376179559631, "grad_norm": 2.1768076419830322, "learning_rate": 6.626800000000001e-06, "loss": 0.0155, "step": 16867 }, { "epoch": 0.899290931385616, "grad_norm": 2.0980441570281982, "learning_rate": 6.626600000000001e-06, "loss": 0.0241, "step": 16868 }, { "epoch": 0.899344244815269, "grad_norm": 1.355492353439331, "learning_rate": 6.626400000000001e-06, "loss": 0.0311, "step": 16869 }, { "epoch": 0.8993975582449218, "grad_norm": 1.2864272594451904, "learning_rate": 6.6262e-06, "loss": 0.0174, "step": 16870 }, { "epoch": 0.8994508716745748, "grad_norm": 1.8729054927825928, "learning_rate": 6.626000000000001e-06, "loss": 0.0344, "step": 16871 }, { "epoch": 0.8995041851042278, "grad_norm": 1.2217241525650024, "learning_rate": 6.6258e-06, "loss": 0.0131, "step": 16872 }, { "epoch": 0.8995574985338807, "grad_norm": 3.063096046447754, "learning_rate": 6.6256000000000006e-06, "loss": 0.0277, "step": 16873 }, { "epoch": 0.8996108119635337, "grad_norm": 0.3635268211364746, "learning_rate": 6.625400000000001e-06, "loss": 0.0032, "step": 16874 }, { "epoch": 0.8996641253931865, "grad_norm": 2.344632625579834, "learning_rate": 6.6252000000000005e-06, "loss": 0.0343, "step": 16875 }, { "epoch": 0.8997174388228395, "grad_norm": 3.9713375568389893, "learning_rate": 6.625e-06, "loss": 0.0687, "step": 16876 }, { "epoch": 0.8997707522524924, "grad_norm": 2.0618691444396973, "learning_rate": 6.6248000000000005e-06, "loss": 0.0361, "step": 16877 }, { "epoch": 0.8998240656821453, "grad_norm": 2.8098292350769043, "learning_rate": 6.624600000000001e-06, "loss": 0.0423, "step": 16878 }, { "epoch": 0.8998773791117982, "grad_norm": 3.106201410293579, "learning_rate": 6.6244000000000005e-06, "loss": 0.0797, "step": 16879 }, { "epoch": 0.8999306925414512, "grad_norm": 0.34798458218574524, "learning_rate": 6.624200000000001e-06, "loss": 0.004, "step": 16880 }, { "epoch": 0.8999840059711042, "grad_norm": 1.6334108114242554, "learning_rate": 6.6240000000000004e-06, "loss": 0.0237, "step": 16881 }, { "epoch": 0.900037319400757, "grad_norm": 2.1232259273529053, "learning_rate": 6.6238e-06, "loss": 0.0334, "step": 16882 }, { "epoch": 0.90009063283041, "grad_norm": 1.555328369140625, "learning_rate": 6.6236e-06, "loss": 0.0143, "step": 16883 }, { "epoch": 0.9001439462600629, "grad_norm": 1.6251438856124878, "learning_rate": 6.623400000000001e-06, "loss": 0.0248, "step": 16884 }, { "epoch": 0.9001972596897159, "grad_norm": 3.66756272315979, "learning_rate": 6.6232e-06, "loss": 0.0321, "step": 16885 }, { "epoch": 0.9002505731193687, "grad_norm": 1.5236001014709473, "learning_rate": 6.623000000000001e-06, "loss": 0.0245, "step": 16886 }, { "epoch": 0.9003038865490217, "grad_norm": 1.7228235006332397, "learning_rate": 6.6228e-06, "loss": 0.0411, "step": 16887 }, { "epoch": 0.9003571999786746, "grad_norm": 2.941739797592163, "learning_rate": 6.6226e-06, "loss": 0.0361, "step": 16888 }, { "epoch": 0.9004105134083276, "grad_norm": 1.2799400091171265, "learning_rate": 6.622400000000001e-06, "loss": 0.0138, "step": 16889 }, { "epoch": 0.9004638268379805, "grad_norm": 2.026576519012451, "learning_rate": 6.622200000000001e-06, "loss": 0.027, "step": 16890 }, { "epoch": 0.9005171402676334, "grad_norm": 1.7859753370285034, "learning_rate": 6.622e-06, "loss": 0.0372, "step": 16891 }, { "epoch": 0.9005704536972864, "grad_norm": 1.2550785541534424, "learning_rate": 6.621800000000001e-06, "loss": 0.0192, "step": 16892 }, { "epoch": 0.9006237671269393, "grad_norm": 2.549042224884033, "learning_rate": 6.6216e-06, "loss": 0.0399, "step": 16893 }, { "epoch": 0.9006770805565922, "grad_norm": 1.6445969343185425, "learning_rate": 6.6214e-06, "loss": 0.014, "step": 16894 }, { "epoch": 0.9007303939862451, "grad_norm": 3.1997432708740234, "learning_rate": 6.621200000000001e-06, "loss": 0.035, "step": 16895 }, { "epoch": 0.9007837074158981, "grad_norm": 0.6928825974464417, "learning_rate": 6.621000000000001e-06, "loss": 0.0077, "step": 16896 }, { "epoch": 0.900837020845551, "grad_norm": 1.1715704202651978, "learning_rate": 6.6208e-06, "loss": 0.0144, "step": 16897 }, { "epoch": 0.9008903342752039, "grad_norm": 2.4533591270446777, "learning_rate": 6.620600000000001e-06, "loss": 0.033, "step": 16898 }, { "epoch": 0.9009436477048568, "grad_norm": 0.9510758519172668, "learning_rate": 6.6204e-06, "loss": 0.0069, "step": 16899 }, { "epoch": 0.9009969611345098, "grad_norm": 3.0931546688079834, "learning_rate": 6.6202000000000014e-06, "loss": 0.0324, "step": 16900 }, { "epoch": 0.9010502745641628, "grad_norm": 1.8609501123428345, "learning_rate": 6.620000000000001e-06, "loss": 0.015, "step": 16901 }, { "epoch": 0.9011035879938156, "grad_norm": 1.8095004558563232, "learning_rate": 6.6198000000000006e-06, "loss": 0.0172, "step": 16902 }, { "epoch": 0.9011569014234686, "grad_norm": 1.2396388053894043, "learning_rate": 6.6196e-06, "loss": 0.0281, "step": 16903 }, { "epoch": 0.9012102148531215, "grad_norm": 2.024854898452759, "learning_rate": 6.6194000000000005e-06, "loss": 0.0266, "step": 16904 }, { "epoch": 0.9012635282827745, "grad_norm": 2.1553540229797363, "learning_rate": 6.6192e-06, "loss": 0.0234, "step": 16905 }, { "epoch": 0.9013168417124273, "grad_norm": 2.57800030708313, "learning_rate": 6.619000000000001e-06, "loss": 0.0309, "step": 16906 }, { "epoch": 0.9013701551420803, "grad_norm": 2.347593069076538, "learning_rate": 6.618800000000001e-06, "loss": 0.0144, "step": 16907 }, { "epoch": 0.9014234685717332, "grad_norm": 1.1342030763626099, "learning_rate": 6.6186000000000005e-06, "loss": 0.0103, "step": 16908 }, { "epoch": 0.9014767820013861, "grad_norm": 1.1237930059432983, "learning_rate": 6.6184e-06, "loss": 0.0094, "step": 16909 }, { "epoch": 0.9015300954310391, "grad_norm": 1.612253189086914, "learning_rate": 6.6182000000000004e-06, "loss": 0.0182, "step": 16910 }, { "epoch": 0.901583408860692, "grad_norm": 2.3871474266052246, "learning_rate": 6.618000000000001e-06, "loss": 0.0424, "step": 16911 }, { "epoch": 0.901636722290345, "grad_norm": 1.9596792459487915, "learning_rate": 6.617800000000001e-06, "loss": 0.0389, "step": 16912 }, { "epoch": 0.9016900357199978, "grad_norm": 1.4176630973815918, "learning_rate": 6.617600000000001e-06, "loss": 0.0135, "step": 16913 }, { "epoch": 0.9017433491496508, "grad_norm": 1.8435561656951904, "learning_rate": 6.6174e-06, "loss": 0.0313, "step": 16914 }, { "epoch": 0.9017966625793037, "grad_norm": 2.106264591217041, "learning_rate": 6.6172e-06, "loss": 0.0211, "step": 16915 }, { "epoch": 0.9018499760089567, "grad_norm": 1.2619283199310303, "learning_rate": 6.617e-06, "loss": 0.007, "step": 16916 }, { "epoch": 0.9019032894386095, "grad_norm": 1.4885139465332031, "learning_rate": 6.616800000000001e-06, "loss": 0.0078, "step": 16917 }, { "epoch": 0.9019566028682625, "grad_norm": 2.1487882137298584, "learning_rate": 6.616600000000001e-06, "loss": 0.0499, "step": 16918 }, { "epoch": 0.9020099162979155, "grad_norm": 1.0172991752624512, "learning_rate": 6.616400000000001e-06, "loss": 0.0175, "step": 16919 }, { "epoch": 0.9020632297275684, "grad_norm": 3.254607915878296, "learning_rate": 6.6162e-06, "loss": 0.0428, "step": 16920 }, { "epoch": 0.9021165431572213, "grad_norm": 1.5794638395309448, "learning_rate": 6.616e-06, "loss": 0.0256, "step": 16921 }, { "epoch": 0.9021698565868742, "grad_norm": 2.5706470012664795, "learning_rate": 6.6158e-06, "loss": 0.0684, "step": 16922 }, { "epoch": 0.9022231700165272, "grad_norm": 1.2358505725860596, "learning_rate": 6.615600000000001e-06, "loss": 0.0148, "step": 16923 }, { "epoch": 0.90227648344618, "grad_norm": 2.0318634510040283, "learning_rate": 6.615400000000001e-06, "loss": 0.0267, "step": 16924 }, { "epoch": 0.902329796875833, "grad_norm": 5.347184658050537, "learning_rate": 6.615200000000001e-06, "loss": 0.062, "step": 16925 }, { "epoch": 0.9023831103054859, "grad_norm": 2.8553247451782227, "learning_rate": 6.615e-06, "loss": 0.0214, "step": 16926 }, { "epoch": 0.9024364237351389, "grad_norm": 1.4622691869735718, "learning_rate": 6.6148e-06, "loss": 0.0291, "step": 16927 }, { "epoch": 0.9024897371647919, "grad_norm": 1.6334388256072998, "learning_rate": 6.614600000000001e-06, "loss": 0.0147, "step": 16928 }, { "epoch": 0.9025430505944447, "grad_norm": 2.1550099849700928, "learning_rate": 6.6144000000000006e-06, "loss": 0.0378, "step": 16929 }, { "epoch": 0.9025963640240977, "grad_norm": 1.1870344877243042, "learning_rate": 6.614200000000001e-06, "loss": 0.0122, "step": 16930 }, { "epoch": 0.9026496774537506, "grad_norm": 3.992804527282715, "learning_rate": 6.6140000000000005e-06, "loss": 0.0465, "step": 16931 }, { "epoch": 0.9027029908834036, "grad_norm": 2.450815200805664, "learning_rate": 6.6138e-06, "loss": 0.0541, "step": 16932 }, { "epoch": 0.9027563043130564, "grad_norm": 4.909051418304443, "learning_rate": 6.6136e-06, "loss": 0.0376, "step": 16933 }, { "epoch": 0.9028096177427094, "grad_norm": 1.8030163049697876, "learning_rate": 6.613400000000001e-06, "loss": 0.013, "step": 16934 }, { "epoch": 0.9028629311723623, "grad_norm": 1.4827369451522827, "learning_rate": 6.6132000000000005e-06, "loss": 0.0124, "step": 16935 }, { "epoch": 0.9029162446020152, "grad_norm": 1.6788805723190308, "learning_rate": 6.613000000000001e-06, "loss": 0.0198, "step": 16936 }, { "epoch": 0.9029695580316682, "grad_norm": 1.6012531518936157, "learning_rate": 6.6128000000000005e-06, "loss": 0.0189, "step": 16937 }, { "epoch": 0.9030228714613211, "grad_norm": 2.3373425006866455, "learning_rate": 6.6126e-06, "loss": 0.0481, "step": 16938 }, { "epoch": 0.9030761848909741, "grad_norm": 1.1386562585830688, "learning_rate": 6.612400000000001e-06, "loss": 0.0255, "step": 16939 }, { "epoch": 0.9031294983206269, "grad_norm": 2.5420010089874268, "learning_rate": 6.612200000000001e-06, "loss": 0.0417, "step": 16940 }, { "epoch": 0.9031828117502799, "grad_norm": 4.5150980949401855, "learning_rate": 6.612e-06, "loss": 0.053, "step": 16941 }, { "epoch": 0.9032361251799328, "grad_norm": 1.7619292736053467, "learning_rate": 6.611800000000001e-06, "loss": 0.0248, "step": 16942 }, { "epoch": 0.9032894386095858, "grad_norm": 1.2848179340362549, "learning_rate": 6.6116e-06, "loss": 0.0198, "step": 16943 }, { "epoch": 0.9033427520392386, "grad_norm": 0.8541138768196106, "learning_rate": 6.6114e-06, "loss": 0.019, "step": 16944 }, { "epoch": 0.9033960654688916, "grad_norm": 1.5352336168289185, "learning_rate": 6.611200000000001e-06, "loss": 0.022, "step": 16945 }, { "epoch": 0.9034493788985446, "grad_norm": 1.5050610303878784, "learning_rate": 6.611000000000001e-06, "loss": 0.0157, "step": 16946 }, { "epoch": 0.9035026923281975, "grad_norm": 3.170717477798462, "learning_rate": 6.6108e-06, "loss": 0.0288, "step": 16947 }, { "epoch": 0.9035560057578504, "grad_norm": 2.9388515949249268, "learning_rate": 6.610600000000001e-06, "loss": 0.0186, "step": 16948 }, { "epoch": 0.9036093191875033, "grad_norm": 1.226944088935852, "learning_rate": 6.6104e-06, "loss": 0.0192, "step": 16949 }, { "epoch": 0.9036626326171563, "grad_norm": 1.6079436540603638, "learning_rate": 6.610200000000001e-06, "loss": 0.0191, "step": 16950 }, { "epoch": 0.9037159460468092, "grad_norm": 1.6053788661956787, "learning_rate": 6.610000000000001e-06, "loss": 0.0155, "step": 16951 }, { "epoch": 0.9037692594764621, "grad_norm": 1.856595754623413, "learning_rate": 6.609800000000001e-06, "loss": 0.0257, "step": 16952 }, { "epoch": 0.903822572906115, "grad_norm": 1.950964331626892, "learning_rate": 6.6096e-06, "loss": 0.0444, "step": 16953 }, { "epoch": 0.903875886335768, "grad_norm": 1.5594786405563354, "learning_rate": 6.609400000000001e-06, "loss": 0.0374, "step": 16954 }, { "epoch": 0.9039291997654209, "grad_norm": 1.0661901235580444, "learning_rate": 6.6092e-06, "loss": 0.0172, "step": 16955 }, { "epoch": 0.9039825131950738, "grad_norm": 3.21991229057312, "learning_rate": 6.609000000000001e-06, "loss": 0.0213, "step": 16956 }, { "epoch": 0.9040358266247268, "grad_norm": 5.390723705291748, "learning_rate": 6.608800000000001e-06, "loss": 0.0569, "step": 16957 }, { "epoch": 0.9040891400543797, "grad_norm": 1.7790240049362183, "learning_rate": 6.6086000000000006e-06, "loss": 0.0134, "step": 16958 }, { "epoch": 0.9041424534840327, "grad_norm": 2.134019374847412, "learning_rate": 6.6084e-06, "loss": 0.0178, "step": 16959 }, { "epoch": 0.9041957669136855, "grad_norm": 2.1239936351776123, "learning_rate": 6.6082000000000005e-06, "loss": 0.0514, "step": 16960 }, { "epoch": 0.9042490803433385, "grad_norm": 2.4698867797851562, "learning_rate": 6.608000000000001e-06, "loss": 0.0721, "step": 16961 }, { "epoch": 0.9043023937729914, "grad_norm": 1.3986544609069824, "learning_rate": 6.6078000000000005e-06, "loss": 0.0349, "step": 16962 }, { "epoch": 0.9043557072026444, "grad_norm": 1.3448903560638428, "learning_rate": 6.607600000000001e-06, "loss": 0.0346, "step": 16963 }, { "epoch": 0.9044090206322972, "grad_norm": 0.915630042552948, "learning_rate": 6.6074000000000005e-06, "loss": 0.0218, "step": 16964 }, { "epoch": 0.9044623340619502, "grad_norm": 2.237766742706299, "learning_rate": 6.6072e-06, "loss": 0.0481, "step": 16965 }, { "epoch": 0.9045156474916032, "grad_norm": 1.6435481309890747, "learning_rate": 6.6070000000000004e-06, "loss": 0.0198, "step": 16966 }, { "epoch": 0.904568960921256, "grad_norm": 1.7978565692901611, "learning_rate": 6.606800000000001e-06, "loss": 0.0405, "step": 16967 }, { "epoch": 0.904622274350909, "grad_norm": 1.5072447061538696, "learning_rate": 6.6066e-06, "loss": 0.019, "step": 16968 }, { "epoch": 0.9046755877805619, "grad_norm": 2.10329008102417, "learning_rate": 6.606400000000001e-06, "loss": 0.0373, "step": 16969 }, { "epoch": 0.9047289012102149, "grad_norm": 2.0064613819122314, "learning_rate": 6.6062e-06, "loss": 0.044, "step": 16970 }, { "epoch": 0.9047822146398677, "grad_norm": 2.8137736320495605, "learning_rate": 6.606e-06, "loss": 0.0444, "step": 16971 }, { "epoch": 0.9048355280695207, "grad_norm": 3.0904579162597656, "learning_rate": 6.6058e-06, "loss": 0.035, "step": 16972 }, { "epoch": 0.9048888414991736, "grad_norm": 3.5630295276641846, "learning_rate": 6.605600000000001e-06, "loss": 0.0256, "step": 16973 }, { "epoch": 0.9049421549288266, "grad_norm": 1.700868844985962, "learning_rate": 6.6054e-06, "loss": 0.0395, "step": 16974 }, { "epoch": 0.9049954683584795, "grad_norm": 1.6749259233474731, "learning_rate": 6.605200000000001e-06, "loss": 0.0309, "step": 16975 }, { "epoch": 0.9050487817881324, "grad_norm": 1.4373465776443481, "learning_rate": 6.605e-06, "loss": 0.016, "step": 16976 }, { "epoch": 0.9051020952177854, "grad_norm": 1.7022199630737305, "learning_rate": 6.6048e-06, "loss": 0.0319, "step": 16977 }, { "epoch": 0.9051554086474383, "grad_norm": 1.312915325164795, "learning_rate": 6.604600000000001e-06, "loss": 0.0258, "step": 16978 }, { "epoch": 0.9052087220770912, "grad_norm": 2.4908339977264404, "learning_rate": 6.604400000000001e-06, "loss": 0.0447, "step": 16979 }, { "epoch": 0.9052620355067441, "grad_norm": 3.9050815105438232, "learning_rate": 6.6042e-06, "loss": 0.0355, "step": 16980 }, { "epoch": 0.9053153489363971, "grad_norm": 1.85881507396698, "learning_rate": 6.604000000000001e-06, "loss": 0.0299, "step": 16981 }, { "epoch": 0.90536866236605, "grad_norm": 0.9987999796867371, "learning_rate": 6.6038e-06, "loss": 0.0086, "step": 16982 }, { "epoch": 0.9054219757957029, "grad_norm": 1.940908432006836, "learning_rate": 6.6036e-06, "loss": 0.0328, "step": 16983 }, { "epoch": 0.9054752892253559, "grad_norm": 2.2195050716400146, "learning_rate": 6.603400000000001e-06, "loss": 0.048, "step": 16984 }, { "epoch": 0.9055286026550088, "grad_norm": 1.09499990940094, "learning_rate": 6.603200000000001e-06, "loss": 0.0176, "step": 16985 }, { "epoch": 0.9055819160846618, "grad_norm": 1.7659872770309448, "learning_rate": 6.603e-06, "loss": 0.0149, "step": 16986 }, { "epoch": 0.9056352295143146, "grad_norm": 1.789721131324768, "learning_rate": 6.6028000000000005e-06, "loss": 0.0187, "step": 16987 }, { "epoch": 0.9056885429439676, "grad_norm": 0.8885536789894104, "learning_rate": 6.6026e-06, "loss": 0.0107, "step": 16988 }, { "epoch": 0.9057418563736205, "grad_norm": 1.79857337474823, "learning_rate": 6.602400000000001e-06, "loss": 0.0275, "step": 16989 }, { "epoch": 0.9057951698032735, "grad_norm": 4.977673053741455, "learning_rate": 6.602200000000001e-06, "loss": 0.0439, "step": 16990 }, { "epoch": 0.9058484832329263, "grad_norm": 1.611710786819458, "learning_rate": 6.6020000000000005e-06, "loss": 0.0316, "step": 16991 }, { "epoch": 0.9059017966625793, "grad_norm": 1.8149744272232056, "learning_rate": 6.6018e-06, "loss": 0.0377, "step": 16992 }, { "epoch": 0.9059551100922323, "grad_norm": 2.2983813285827637, "learning_rate": 6.6016000000000005e-06, "loss": 0.0233, "step": 16993 }, { "epoch": 0.9060084235218852, "grad_norm": 2.0752551555633545, "learning_rate": 6.6014e-06, "loss": 0.0369, "step": 16994 }, { "epoch": 0.9060617369515381, "grad_norm": 2.562617301940918, "learning_rate": 6.601200000000001e-06, "loss": 0.0263, "step": 16995 }, { "epoch": 0.906115050381191, "grad_norm": 1.2026433944702148, "learning_rate": 6.601000000000001e-06, "loss": 0.0263, "step": 16996 }, { "epoch": 0.906168363810844, "grad_norm": 1.302926778793335, "learning_rate": 6.6008e-06, "loss": 0.0285, "step": 16997 }, { "epoch": 0.9062216772404968, "grad_norm": 1.8287806510925293, "learning_rate": 6.6006e-06, "loss": 0.036, "step": 16998 }, { "epoch": 0.9062749906701498, "grad_norm": 1.5809258222579956, "learning_rate": 6.6004e-06, "loss": 0.0158, "step": 16999 }, { "epoch": 0.9063283040998027, "grad_norm": 1.0712884664535522, "learning_rate": 6.600200000000001e-06, "loss": 0.023, "step": 17000 }, { "epoch": 0.9063816175294557, "grad_norm": 1.4235860109329224, "learning_rate": 6.600000000000001e-06, "loss": 0.0247, "step": 17001 }, { "epoch": 0.9064349309591087, "grad_norm": 1.136860966682434, "learning_rate": 6.599800000000001e-06, "loss": 0.0205, "step": 17002 }, { "epoch": 0.9064882443887615, "grad_norm": 3.1021909713745117, "learning_rate": 6.5996e-06, "loss": 0.0287, "step": 17003 }, { "epoch": 0.9065415578184145, "grad_norm": 1.9564728736877441, "learning_rate": 6.5994e-06, "loss": 0.0229, "step": 17004 }, { "epoch": 0.9065948712480674, "grad_norm": 1.8158366680145264, "learning_rate": 6.5992e-06, "loss": 0.0366, "step": 17005 }, { "epoch": 0.9066481846777203, "grad_norm": 6.54611873626709, "learning_rate": 6.599000000000001e-06, "loss": 0.0778, "step": 17006 }, { "epoch": 0.9067014981073732, "grad_norm": 4.097296237945557, "learning_rate": 6.598800000000001e-06, "loss": 0.0419, "step": 17007 }, { "epoch": 0.9067548115370262, "grad_norm": 2.073873996734619, "learning_rate": 6.598600000000001e-06, "loss": 0.0354, "step": 17008 }, { "epoch": 0.9068081249666791, "grad_norm": 1.690294623374939, "learning_rate": 6.5984e-06, "loss": 0.0294, "step": 17009 }, { "epoch": 0.906861438396332, "grad_norm": 1.5757325887680054, "learning_rate": 6.5982e-06, "loss": 0.0161, "step": 17010 }, { "epoch": 0.9069147518259849, "grad_norm": 3.0474562644958496, "learning_rate": 6.598000000000001e-06, "loss": 0.0166, "step": 17011 }, { "epoch": 0.9069680652556379, "grad_norm": 1.352616310119629, "learning_rate": 6.597800000000001e-06, "loss": 0.0124, "step": 17012 }, { "epoch": 0.9070213786852909, "grad_norm": 2.4750354290008545, "learning_rate": 6.597600000000001e-06, "loss": 0.0133, "step": 17013 }, { "epoch": 0.9070746921149437, "grad_norm": 0.9804781079292297, "learning_rate": 6.5974000000000006e-06, "loss": 0.0119, "step": 17014 }, { "epoch": 0.9071280055445967, "grad_norm": 2.1389670372009277, "learning_rate": 6.5972e-06, "loss": 0.0214, "step": 17015 }, { "epoch": 0.9071813189742496, "grad_norm": 1.517561912536621, "learning_rate": 6.597e-06, "loss": 0.0226, "step": 17016 }, { "epoch": 0.9072346324039026, "grad_norm": 3.930285930633545, "learning_rate": 6.596800000000001e-06, "loss": 0.0495, "step": 17017 }, { "epoch": 0.9072879458335554, "grad_norm": 2.8428869247436523, "learning_rate": 6.5966000000000005e-06, "loss": 0.037, "step": 17018 }, { "epoch": 0.9073412592632084, "grad_norm": 1.4494637250900269, "learning_rate": 6.596400000000001e-06, "loss": 0.032, "step": 17019 }, { "epoch": 0.9073945726928613, "grad_norm": 2.0638222694396973, "learning_rate": 6.5962000000000005e-06, "loss": 0.039, "step": 17020 }, { "epoch": 0.9074478861225143, "grad_norm": 2.3893110752105713, "learning_rate": 6.596e-06, "loss": 0.0248, "step": 17021 }, { "epoch": 0.9075011995521672, "grad_norm": 1.6618374586105347, "learning_rate": 6.595800000000001e-06, "loss": 0.0171, "step": 17022 }, { "epoch": 0.9075545129818201, "grad_norm": 2.842506170272827, "learning_rate": 6.595600000000001e-06, "loss": 0.0316, "step": 17023 }, { "epoch": 0.9076078264114731, "grad_norm": 5.401190757751465, "learning_rate": 6.5954e-06, "loss": 0.043, "step": 17024 }, { "epoch": 0.907661139841126, "grad_norm": 2.5544955730438232, "learning_rate": 6.595200000000001e-06, "loss": 0.022, "step": 17025 }, { "epoch": 0.9077144532707789, "grad_norm": 2.8326199054718018, "learning_rate": 6.595e-06, "loss": 0.0261, "step": 17026 }, { "epoch": 0.9077677667004318, "grad_norm": 2.683950424194336, "learning_rate": 6.5948e-06, "loss": 0.0269, "step": 17027 }, { "epoch": 0.9078210801300848, "grad_norm": 1.617960810661316, "learning_rate": 6.594600000000001e-06, "loss": 0.02, "step": 17028 }, { "epoch": 0.9078743935597376, "grad_norm": 1.2579840421676636, "learning_rate": 6.594400000000001e-06, "loss": 0.0195, "step": 17029 }, { "epoch": 0.9079277069893906, "grad_norm": 3.633439302444458, "learning_rate": 6.5942e-06, "loss": 0.0266, "step": 17030 }, { "epoch": 0.9079810204190436, "grad_norm": 2.510305404663086, "learning_rate": 6.594000000000001e-06, "loss": 0.047, "step": 17031 }, { "epoch": 0.9080343338486965, "grad_norm": 2.5029349327087402, "learning_rate": 6.5938e-06, "loss": 0.0655, "step": 17032 }, { "epoch": 0.9080876472783495, "grad_norm": 1.2447549104690552, "learning_rate": 6.5936e-06, "loss": 0.0251, "step": 17033 }, { "epoch": 0.9081409607080023, "grad_norm": 3.0468132495880127, "learning_rate": 6.593400000000001e-06, "loss": 0.0209, "step": 17034 }, { "epoch": 0.9081942741376553, "grad_norm": 2.267202615737915, "learning_rate": 6.593200000000001e-06, "loss": 0.065, "step": 17035 }, { "epoch": 0.9082475875673082, "grad_norm": 3.643319606781006, "learning_rate": 6.593e-06, "loss": 0.0502, "step": 17036 }, { "epoch": 0.9083009009969611, "grad_norm": 1.3750113248825073, "learning_rate": 6.592800000000001e-06, "loss": 0.013, "step": 17037 }, { "epoch": 0.908354214426614, "grad_norm": 4.0374321937561035, "learning_rate": 6.5926e-06, "loss": 0.0213, "step": 17038 }, { "epoch": 0.908407527856267, "grad_norm": 3.571331024169922, "learning_rate": 6.592400000000001e-06, "loss": 0.0269, "step": 17039 }, { "epoch": 0.90846084128592, "grad_norm": 0.5510886311531067, "learning_rate": 6.592200000000001e-06, "loss": 0.0062, "step": 17040 }, { "epoch": 0.9085141547155728, "grad_norm": 0.809969425201416, "learning_rate": 6.592000000000001e-06, "loss": 0.0156, "step": 17041 }, { "epoch": 0.9085674681452258, "grad_norm": 0.9326559901237488, "learning_rate": 6.5918e-06, "loss": 0.0119, "step": 17042 }, { "epoch": 0.9086207815748787, "grad_norm": 1.6445777416229248, "learning_rate": 6.5916000000000006e-06, "loss": 0.0205, "step": 17043 }, { "epoch": 0.9086740950045317, "grad_norm": 3.1154136657714844, "learning_rate": 6.5914e-06, "loss": 0.0236, "step": 17044 }, { "epoch": 0.9087274084341845, "grad_norm": 2.1700098514556885, "learning_rate": 6.5912000000000005e-06, "loss": 0.0444, "step": 17045 }, { "epoch": 0.9087807218638375, "grad_norm": 0.6486749053001404, "learning_rate": 6.591000000000001e-06, "loss": 0.0067, "step": 17046 }, { "epoch": 0.9088340352934904, "grad_norm": 0.7964904308319092, "learning_rate": 6.5908000000000005e-06, "loss": 0.0209, "step": 17047 }, { "epoch": 0.9088873487231434, "grad_norm": 1.3305743932724, "learning_rate": 6.5906e-06, "loss": 0.0383, "step": 17048 }, { "epoch": 0.9089406621527963, "grad_norm": 2.886259078979492, "learning_rate": 6.5904000000000005e-06, "loss": 0.04, "step": 17049 }, { "epoch": 0.9089939755824492, "grad_norm": 0.9476210474967957, "learning_rate": 6.590200000000001e-06, "loss": 0.0078, "step": 17050 }, { "epoch": 0.9090472890121022, "grad_norm": 2.1111068725585938, "learning_rate": 6.5900000000000004e-06, "loss": 0.0452, "step": 17051 }, { "epoch": 0.9091006024417551, "grad_norm": 2.369987964630127, "learning_rate": 6.589800000000001e-06, "loss": 0.0355, "step": 17052 }, { "epoch": 0.909153915871408, "grad_norm": 3.481544017791748, "learning_rate": 6.5896e-06, "loss": 0.0531, "step": 17053 }, { "epoch": 0.9092072293010609, "grad_norm": 1.1608022451400757, "learning_rate": 6.5894e-06, "loss": 0.0126, "step": 17054 }, { "epoch": 0.9092605427307139, "grad_norm": 1.578035593032837, "learning_rate": 6.5892e-06, "loss": 0.0218, "step": 17055 }, { "epoch": 0.9093138561603668, "grad_norm": 2.2941625118255615, "learning_rate": 6.589000000000001e-06, "loss": 0.073, "step": 17056 }, { "epoch": 0.9093671695900197, "grad_norm": 2.564772367477417, "learning_rate": 6.5888e-06, "loss": 0.0263, "step": 17057 }, { "epoch": 0.9094204830196727, "grad_norm": 0.9755510091781616, "learning_rate": 6.588600000000001e-06, "loss": 0.0235, "step": 17058 }, { "epoch": 0.9094737964493256, "grad_norm": 1.5125305652618408, "learning_rate": 6.5884e-06, "loss": 0.0303, "step": 17059 }, { "epoch": 0.9095271098789786, "grad_norm": 0.8898825645446777, "learning_rate": 6.5882e-06, "loss": 0.01, "step": 17060 }, { "epoch": 0.9095804233086314, "grad_norm": 0.9829683303833008, "learning_rate": 6.588000000000001e-06, "loss": 0.0107, "step": 17061 }, { "epoch": 0.9096337367382844, "grad_norm": 1.9237090349197388, "learning_rate": 6.587800000000001e-06, "loss": 0.0325, "step": 17062 }, { "epoch": 0.9096870501679373, "grad_norm": 1.9073104858398438, "learning_rate": 6.5876e-06, "loss": 0.0217, "step": 17063 }, { "epoch": 0.9097403635975903, "grad_norm": 2.0634548664093018, "learning_rate": 6.587400000000001e-06, "loss": 0.0173, "step": 17064 }, { "epoch": 0.9097936770272431, "grad_norm": 1.458387017250061, "learning_rate": 6.5872e-06, "loss": 0.0413, "step": 17065 }, { "epoch": 0.9098469904568961, "grad_norm": 1.3960676193237305, "learning_rate": 6.587e-06, "loss": 0.0215, "step": 17066 }, { "epoch": 0.909900303886549, "grad_norm": 3.2416553497314453, "learning_rate": 6.586800000000001e-06, "loss": 0.0232, "step": 17067 }, { "epoch": 0.909953617316202, "grad_norm": 0.9440760016441345, "learning_rate": 6.586600000000001e-06, "loss": 0.0204, "step": 17068 }, { "epoch": 0.9100069307458549, "grad_norm": 1.2912266254425049, "learning_rate": 6.5864e-06, "loss": 0.01, "step": 17069 }, { "epoch": 0.9100602441755078, "grad_norm": 2.1334118843078613, "learning_rate": 6.5862000000000006e-06, "loss": 0.0527, "step": 17070 }, { "epoch": 0.9101135576051608, "grad_norm": 1.0084469318389893, "learning_rate": 6.586e-06, "loss": 0.0098, "step": 17071 }, { "epoch": 0.9101668710348136, "grad_norm": 1.523789882659912, "learning_rate": 6.585800000000001e-06, "loss": 0.0145, "step": 17072 }, { "epoch": 0.9102201844644666, "grad_norm": 3.0004026889801025, "learning_rate": 6.585600000000001e-06, "loss": 0.026, "step": 17073 }, { "epoch": 0.9102734978941195, "grad_norm": 1.8613133430480957, "learning_rate": 6.5854000000000005e-06, "loss": 0.0465, "step": 17074 }, { "epoch": 0.9103268113237725, "grad_norm": 1.87627375125885, "learning_rate": 6.5852e-06, "loss": 0.0279, "step": 17075 }, { "epoch": 0.9103801247534253, "grad_norm": 1.779831051826477, "learning_rate": 6.5850000000000005e-06, "loss": 0.018, "step": 17076 }, { "epoch": 0.9104334381830783, "grad_norm": 1.5243550539016724, "learning_rate": 6.5848e-06, "loss": 0.0529, "step": 17077 }, { "epoch": 0.9104867516127313, "grad_norm": 0.3322513997554779, "learning_rate": 6.584600000000001e-06, "loss": 0.0033, "step": 17078 }, { "epoch": 0.9105400650423842, "grad_norm": 1.531108021736145, "learning_rate": 6.584400000000001e-06, "loss": 0.0311, "step": 17079 }, { "epoch": 0.9105933784720371, "grad_norm": 1.1209943294525146, "learning_rate": 6.5842e-06, "loss": 0.0115, "step": 17080 }, { "epoch": 0.91064669190169, "grad_norm": 2.303992509841919, "learning_rate": 6.584e-06, "loss": 0.0197, "step": 17081 }, { "epoch": 0.910700005331343, "grad_norm": 1.5733381509780884, "learning_rate": 6.5838e-06, "loss": 0.0093, "step": 17082 }, { "epoch": 0.9107533187609959, "grad_norm": 1.0434902906417847, "learning_rate": 6.5836e-06, "loss": 0.0227, "step": 17083 }, { "epoch": 0.9108066321906488, "grad_norm": 2.387159585952759, "learning_rate": 6.583400000000001e-06, "loss": 0.0408, "step": 17084 }, { "epoch": 0.9108599456203017, "grad_norm": 2.373762607574463, "learning_rate": 6.583200000000001e-06, "loss": 0.0198, "step": 17085 }, { "epoch": 0.9109132590499547, "grad_norm": 1.0136659145355225, "learning_rate": 6.583e-06, "loss": 0.0071, "step": 17086 }, { "epoch": 0.9109665724796077, "grad_norm": 3.4267265796661377, "learning_rate": 6.5828e-06, "loss": 0.0492, "step": 17087 }, { "epoch": 0.9110198859092605, "grad_norm": 3.7891557216644287, "learning_rate": 6.5826e-06, "loss": 0.0451, "step": 17088 }, { "epoch": 0.9110731993389135, "grad_norm": 2.918212652206421, "learning_rate": 6.582400000000001e-06, "loss": 0.0269, "step": 17089 }, { "epoch": 0.9111265127685664, "grad_norm": 1.8486050367355347, "learning_rate": 6.582200000000001e-06, "loss": 0.0312, "step": 17090 }, { "epoch": 0.9111798261982194, "grad_norm": 2.504683256149292, "learning_rate": 6.582000000000001e-06, "loss": 0.0458, "step": 17091 }, { "epoch": 0.9112331396278722, "grad_norm": 1.7058590650558472, "learning_rate": 6.5818e-06, "loss": 0.0448, "step": 17092 }, { "epoch": 0.9112864530575252, "grad_norm": 2.492393732070923, "learning_rate": 6.5816e-06, "loss": 0.0449, "step": 17093 }, { "epoch": 0.9113397664871781, "grad_norm": 1.0029107332229614, "learning_rate": 6.5814e-06, "loss": 0.0203, "step": 17094 }, { "epoch": 0.911393079916831, "grad_norm": 1.5168282985687256, "learning_rate": 6.581200000000001e-06, "loss": 0.0157, "step": 17095 }, { "epoch": 0.911446393346484, "grad_norm": 2.685666799545288, "learning_rate": 6.581000000000001e-06, "loss": 0.0452, "step": 17096 }, { "epoch": 0.9114997067761369, "grad_norm": 2.4430630207061768, "learning_rate": 6.580800000000001e-06, "loss": 0.0246, "step": 17097 }, { "epoch": 0.9115530202057899, "grad_norm": 1.6356468200683594, "learning_rate": 6.5806e-06, "loss": 0.0434, "step": 17098 }, { "epoch": 0.9116063336354427, "grad_norm": 1.917304277420044, "learning_rate": 6.5804e-06, "loss": 0.0313, "step": 17099 }, { "epoch": 0.9116596470650957, "grad_norm": 1.4093616008758545, "learning_rate": 6.580200000000001e-06, "loss": 0.0261, "step": 17100 }, { "epoch": 0.9117129604947486, "grad_norm": 1.6848764419555664, "learning_rate": 6.5800000000000005e-06, "loss": 0.0161, "step": 17101 }, { "epoch": 0.9117662739244016, "grad_norm": 1.5617817640304565, "learning_rate": 6.579800000000001e-06, "loss": 0.0134, "step": 17102 }, { "epoch": 0.9118195873540544, "grad_norm": 2.111008405685425, "learning_rate": 6.5796000000000005e-06, "loss": 0.0294, "step": 17103 }, { "epoch": 0.9118729007837074, "grad_norm": 0.7931459546089172, "learning_rate": 6.5794e-06, "loss": 0.0207, "step": 17104 }, { "epoch": 0.9119262142133604, "grad_norm": 1.2162456512451172, "learning_rate": 6.5792e-06, "loss": 0.0306, "step": 17105 }, { "epoch": 0.9119795276430133, "grad_norm": 2.5537731647491455, "learning_rate": 6.579000000000001e-06, "loss": 0.0228, "step": 17106 }, { "epoch": 0.9120328410726662, "grad_norm": 1.4701265096664429, "learning_rate": 6.5788000000000004e-06, "loss": 0.0183, "step": 17107 }, { "epoch": 0.9120861545023191, "grad_norm": 1.8575695753097534, "learning_rate": 6.578600000000001e-06, "loss": 0.0213, "step": 17108 }, { "epoch": 0.9121394679319721, "grad_norm": 4.013955116271973, "learning_rate": 6.5784e-06, "loss": 0.0874, "step": 17109 }, { "epoch": 0.912192781361625, "grad_norm": 1.3881816864013672, "learning_rate": 6.5782e-06, "loss": 0.0195, "step": 17110 }, { "epoch": 0.9122460947912779, "grad_norm": 2.1745026111602783, "learning_rate": 6.578000000000001e-06, "loss": 0.0206, "step": 17111 }, { "epoch": 0.9122994082209308, "grad_norm": 2.5047104358673096, "learning_rate": 6.577800000000001e-06, "loss": 0.0493, "step": 17112 }, { "epoch": 0.9123527216505838, "grad_norm": 2.0699212551116943, "learning_rate": 6.5776e-06, "loss": 0.0193, "step": 17113 }, { "epoch": 0.9124060350802368, "grad_norm": 2.936013698577881, "learning_rate": 6.577400000000001e-06, "loss": 0.0571, "step": 17114 }, { "epoch": 0.9124593485098896, "grad_norm": 2.079822540283203, "learning_rate": 6.5772e-06, "loss": 0.0249, "step": 17115 }, { "epoch": 0.9125126619395426, "grad_norm": 1.9112635850906372, "learning_rate": 6.577e-06, "loss": 0.0111, "step": 17116 }, { "epoch": 0.9125659753691955, "grad_norm": 1.6093517541885376, "learning_rate": 6.576800000000001e-06, "loss": 0.0339, "step": 17117 }, { "epoch": 0.9126192887988485, "grad_norm": 1.4939748048782349, "learning_rate": 6.576600000000001e-06, "loss": 0.0171, "step": 17118 }, { "epoch": 0.9126726022285013, "grad_norm": 1.0078692436218262, "learning_rate": 6.5764e-06, "loss": 0.018, "step": 17119 }, { "epoch": 0.9127259156581543, "grad_norm": 2.016068696975708, "learning_rate": 6.576200000000001e-06, "loss": 0.0269, "step": 17120 }, { "epoch": 0.9127792290878072, "grad_norm": 1.127335548400879, "learning_rate": 6.576e-06, "loss": 0.0256, "step": 17121 }, { "epoch": 0.9128325425174602, "grad_norm": 1.6132010221481323, "learning_rate": 6.5758000000000015e-06, "loss": 0.0283, "step": 17122 }, { "epoch": 0.912885855947113, "grad_norm": 1.7038066387176514, "learning_rate": 6.575600000000001e-06, "loss": 0.0151, "step": 17123 }, { "epoch": 0.912939169376766, "grad_norm": 3.74019718170166, "learning_rate": 6.575400000000001e-06, "loss": 0.0428, "step": 17124 }, { "epoch": 0.912992482806419, "grad_norm": 0.826839029788971, "learning_rate": 6.5752e-06, "loss": 0.0089, "step": 17125 }, { "epoch": 0.9130457962360718, "grad_norm": 2.6357076168060303, "learning_rate": 6.5750000000000006e-06, "loss": 0.0327, "step": 17126 }, { "epoch": 0.9130991096657248, "grad_norm": 2.55285382270813, "learning_rate": 6.5748e-06, "loss": 0.0383, "step": 17127 }, { "epoch": 0.9131524230953777, "grad_norm": 1.925911784172058, "learning_rate": 6.574600000000001e-06, "loss": 0.0239, "step": 17128 }, { "epoch": 0.9132057365250307, "grad_norm": 1.561825156211853, "learning_rate": 6.574400000000001e-06, "loss": 0.0192, "step": 17129 }, { "epoch": 0.9132590499546835, "grad_norm": 0.9781075716018677, "learning_rate": 6.5742000000000005e-06, "loss": 0.0162, "step": 17130 }, { "epoch": 0.9133123633843365, "grad_norm": 0.959247350692749, "learning_rate": 6.574e-06, "loss": 0.0131, "step": 17131 }, { "epoch": 0.9133656768139894, "grad_norm": 1.5003970861434937, "learning_rate": 6.5738000000000005e-06, "loss": 0.0122, "step": 17132 }, { "epoch": 0.9134189902436424, "grad_norm": 1.5117669105529785, "learning_rate": 6.573600000000001e-06, "loss": 0.0347, "step": 17133 }, { "epoch": 0.9134723036732953, "grad_norm": 2.8402607440948486, "learning_rate": 6.573400000000001e-06, "loss": 0.0579, "step": 17134 }, { "epoch": 0.9135256171029482, "grad_norm": 2.1045966148376465, "learning_rate": 6.573200000000001e-06, "loss": 0.0284, "step": 17135 }, { "epoch": 0.9135789305326012, "grad_norm": 2.803093671798706, "learning_rate": 6.5730000000000004e-06, "loss": 0.0312, "step": 17136 }, { "epoch": 0.9136322439622541, "grad_norm": 2.5662014484405518, "learning_rate": 6.5728e-06, "loss": 0.0546, "step": 17137 }, { "epoch": 0.913685557391907, "grad_norm": 0.7692999839782715, "learning_rate": 6.5726e-06, "loss": 0.0066, "step": 17138 }, { "epoch": 0.9137388708215599, "grad_norm": 2.705212354660034, "learning_rate": 6.572400000000001e-06, "loss": 0.0358, "step": 17139 }, { "epoch": 0.9137921842512129, "grad_norm": 1.4338734149932861, "learning_rate": 6.572200000000001e-06, "loss": 0.0114, "step": 17140 }, { "epoch": 0.9138454976808658, "grad_norm": 2.8161988258361816, "learning_rate": 6.572000000000001e-06, "loss": 0.0265, "step": 17141 }, { "epoch": 0.9138988111105187, "grad_norm": 1.8915563821792603, "learning_rate": 6.5718e-06, "loss": 0.0288, "step": 17142 }, { "epoch": 0.9139521245401717, "grad_norm": 1.2767032384872437, "learning_rate": 6.5716e-06, "loss": 0.0098, "step": 17143 }, { "epoch": 0.9140054379698246, "grad_norm": 2.1901803016662598, "learning_rate": 6.5714e-06, "loss": 0.0232, "step": 17144 }, { "epoch": 0.9140587513994776, "grad_norm": 4.203003883361816, "learning_rate": 6.571200000000001e-06, "loss": 0.0378, "step": 17145 }, { "epoch": 0.9141120648291304, "grad_norm": 3.0002870559692383, "learning_rate": 6.571000000000001e-06, "loss": 0.0419, "step": 17146 }, { "epoch": 0.9141653782587834, "grad_norm": 2.388975143432617, "learning_rate": 6.570800000000001e-06, "loss": 0.0197, "step": 17147 }, { "epoch": 0.9142186916884363, "grad_norm": 1.4046905040740967, "learning_rate": 6.5706e-06, "loss": 0.0186, "step": 17148 }, { "epoch": 0.9142720051180893, "grad_norm": 1.8172399997711182, "learning_rate": 6.5704e-06, "loss": 0.0172, "step": 17149 }, { "epoch": 0.9143253185477421, "grad_norm": 1.2952426671981812, "learning_rate": 6.570200000000001e-06, "loss": 0.0122, "step": 17150 }, { "epoch": 0.9143786319773951, "grad_norm": 2.6025121212005615, "learning_rate": 6.570000000000001e-06, "loss": 0.0257, "step": 17151 }, { "epoch": 0.9144319454070481, "grad_norm": 0.9510686993598938, "learning_rate": 6.569800000000001e-06, "loss": 0.0085, "step": 17152 }, { "epoch": 0.914485258836701, "grad_norm": 2.070234775543213, "learning_rate": 6.569600000000001e-06, "loss": 0.0239, "step": 17153 }, { "epoch": 0.9145385722663539, "grad_norm": 2.1927268505096436, "learning_rate": 6.5694e-06, "loss": 0.0324, "step": 17154 }, { "epoch": 0.9145918856960068, "grad_norm": 1.0980714559555054, "learning_rate": 6.5692e-06, "loss": 0.0135, "step": 17155 }, { "epoch": 0.9146451991256598, "grad_norm": 1.5852762460708618, "learning_rate": 6.569000000000001e-06, "loss": 0.0296, "step": 17156 }, { "epoch": 0.9146985125553126, "grad_norm": 1.4986464977264404, "learning_rate": 6.5688000000000005e-06, "loss": 0.0199, "step": 17157 }, { "epoch": 0.9147518259849656, "grad_norm": 3.1877713203430176, "learning_rate": 6.568600000000001e-06, "loss": 0.02, "step": 17158 }, { "epoch": 0.9148051394146185, "grad_norm": 2.2693932056427, "learning_rate": 6.5684000000000005e-06, "loss": 0.0431, "step": 17159 }, { "epoch": 0.9148584528442715, "grad_norm": 2.970975160598755, "learning_rate": 6.5682e-06, "loss": 0.0539, "step": 17160 }, { "epoch": 0.9149117662739245, "grad_norm": 2.3365461826324463, "learning_rate": 6.568000000000001e-06, "loss": 0.037, "step": 17161 }, { "epoch": 0.9149650797035773, "grad_norm": 1.8454617261886597, "learning_rate": 6.567800000000001e-06, "loss": 0.034, "step": 17162 }, { "epoch": 0.9150183931332303, "grad_norm": 1.8292009830474854, "learning_rate": 6.5676000000000004e-06, "loss": 0.0231, "step": 17163 }, { "epoch": 0.9150717065628832, "grad_norm": 0.9666395783424377, "learning_rate": 6.567400000000001e-06, "loss": 0.0174, "step": 17164 }, { "epoch": 0.9151250199925361, "grad_norm": 1.4387671947479248, "learning_rate": 6.5672e-06, "loss": 0.0207, "step": 17165 }, { "epoch": 0.915178333422189, "grad_norm": 1.7399319410324097, "learning_rate": 6.567e-06, "loss": 0.0404, "step": 17166 }, { "epoch": 0.915231646851842, "grad_norm": 2.678116798400879, "learning_rate": 6.566800000000001e-06, "loss": 0.0253, "step": 17167 }, { "epoch": 0.9152849602814949, "grad_norm": 2.460988759994507, "learning_rate": 6.566600000000001e-06, "loss": 0.037, "step": 17168 }, { "epoch": 0.9153382737111478, "grad_norm": 2.710495710372925, "learning_rate": 6.5664e-06, "loss": 0.0203, "step": 17169 }, { "epoch": 0.9153915871408008, "grad_norm": 1.7400212287902832, "learning_rate": 6.566200000000001e-06, "loss": 0.0162, "step": 17170 }, { "epoch": 0.9154449005704537, "grad_norm": 1.7591774463653564, "learning_rate": 6.566e-06, "loss": 0.0113, "step": 17171 }, { "epoch": 0.9154982140001067, "grad_norm": 2.0048418045043945, "learning_rate": 6.565800000000001e-06, "loss": 0.0225, "step": 17172 }, { "epoch": 0.9155515274297595, "grad_norm": 2.5231235027313232, "learning_rate": 6.565600000000001e-06, "loss": 0.0271, "step": 17173 }, { "epoch": 0.9156048408594125, "grad_norm": 2.2045135498046875, "learning_rate": 6.565400000000001e-06, "loss": 0.0252, "step": 17174 }, { "epoch": 0.9156581542890654, "grad_norm": 1.223855972290039, "learning_rate": 6.5652e-06, "loss": 0.0149, "step": 17175 }, { "epoch": 0.9157114677187184, "grad_norm": 1.627854347229004, "learning_rate": 6.565000000000001e-06, "loss": 0.046, "step": 17176 }, { "epoch": 0.9157647811483712, "grad_norm": 3.3199844360351562, "learning_rate": 6.5648e-06, "loss": 0.0145, "step": 17177 }, { "epoch": 0.9158180945780242, "grad_norm": 1.230470895767212, "learning_rate": 6.564600000000001e-06, "loss": 0.0066, "step": 17178 }, { "epoch": 0.9158714080076771, "grad_norm": 1.0789376497268677, "learning_rate": 6.564400000000001e-06, "loss": 0.0206, "step": 17179 }, { "epoch": 0.9159247214373301, "grad_norm": 1.3331409692764282, "learning_rate": 6.564200000000001e-06, "loss": 0.0285, "step": 17180 }, { "epoch": 0.915978034866983, "grad_norm": 1.9157084226608276, "learning_rate": 6.564e-06, "loss": 0.0412, "step": 17181 }, { "epoch": 0.9160313482966359, "grad_norm": 1.4513909816741943, "learning_rate": 6.563800000000001e-06, "loss": 0.0273, "step": 17182 }, { "epoch": 0.9160846617262889, "grad_norm": 2.285029649734497, "learning_rate": 6.563600000000001e-06, "loss": 0.0465, "step": 17183 }, { "epoch": 0.9161379751559418, "grad_norm": 2.737290382385254, "learning_rate": 6.5634000000000006e-06, "loss": 0.0198, "step": 17184 }, { "epoch": 0.9161912885855947, "grad_norm": 2.4858765602111816, "learning_rate": 6.563200000000001e-06, "loss": 0.0312, "step": 17185 }, { "epoch": 0.9162446020152476, "grad_norm": 1.7497851848602295, "learning_rate": 6.5630000000000005e-06, "loss": 0.0308, "step": 17186 }, { "epoch": 0.9162979154449006, "grad_norm": 0.5392107963562012, "learning_rate": 6.5628e-06, "loss": 0.0058, "step": 17187 }, { "epoch": 0.9163512288745534, "grad_norm": 2.0768373012542725, "learning_rate": 6.5626000000000005e-06, "loss": 0.0171, "step": 17188 }, { "epoch": 0.9164045423042064, "grad_norm": 2.1771340370178223, "learning_rate": 6.562400000000001e-06, "loss": 0.0296, "step": 17189 }, { "epoch": 0.9164578557338594, "grad_norm": 2.2208056449890137, "learning_rate": 6.5622000000000005e-06, "loss": 0.0237, "step": 17190 }, { "epoch": 0.9165111691635123, "grad_norm": 1.4332151412963867, "learning_rate": 6.562000000000001e-06, "loss": 0.0224, "step": 17191 }, { "epoch": 0.9165644825931653, "grad_norm": 2.3133485317230225, "learning_rate": 6.5618000000000004e-06, "loss": 0.0153, "step": 17192 }, { "epoch": 0.9166177960228181, "grad_norm": 4.910070896148682, "learning_rate": 6.5616e-06, "loss": 0.0222, "step": 17193 }, { "epoch": 0.9166711094524711, "grad_norm": 2.0423519611358643, "learning_rate": 6.5614e-06, "loss": 0.0187, "step": 17194 }, { "epoch": 0.916724422882124, "grad_norm": 3.0492238998413086, "learning_rate": 6.561200000000001e-06, "loss": 0.0588, "step": 17195 }, { "epoch": 0.916777736311777, "grad_norm": 0.6688684821128845, "learning_rate": 6.561e-06, "loss": 0.0058, "step": 17196 }, { "epoch": 0.9168310497414298, "grad_norm": 1.8107308149337769, "learning_rate": 6.560800000000001e-06, "loss": 0.0246, "step": 17197 }, { "epoch": 0.9168843631710828, "grad_norm": 2.2754621505737305, "learning_rate": 6.5606e-06, "loss": 0.0277, "step": 17198 }, { "epoch": 0.9169376766007358, "grad_norm": 1.5373663902282715, "learning_rate": 6.5604e-06, "loss": 0.0111, "step": 17199 }, { "epoch": 0.9169909900303886, "grad_norm": 1.8837708234786987, "learning_rate": 6.560200000000001e-06, "loss": 0.0346, "step": 17200 }, { "epoch": 0.9170443034600416, "grad_norm": 2.26684308052063, "learning_rate": 6.560000000000001e-06, "loss": 0.0436, "step": 17201 }, { "epoch": 0.9170976168896945, "grad_norm": 2.828589916229248, "learning_rate": 6.5598e-06, "loss": 0.0213, "step": 17202 }, { "epoch": 0.9171509303193475, "grad_norm": 7.005680084228516, "learning_rate": 6.559600000000001e-06, "loss": 0.0664, "step": 17203 }, { "epoch": 0.9172042437490003, "grad_norm": 1.4635496139526367, "learning_rate": 6.5594e-06, "loss": 0.0093, "step": 17204 }, { "epoch": 0.9172575571786533, "grad_norm": 2.36633038520813, "learning_rate": 6.5592e-06, "loss": 0.0416, "step": 17205 }, { "epoch": 0.9173108706083062, "grad_norm": 0.8624116778373718, "learning_rate": 6.559000000000001e-06, "loss": 0.0118, "step": 17206 }, { "epoch": 0.9173641840379592, "grad_norm": 1.6679060459136963, "learning_rate": 6.558800000000001e-06, "loss": 0.0264, "step": 17207 }, { "epoch": 0.9174174974676121, "grad_norm": 1.6173561811447144, "learning_rate": 6.5586e-06, "loss": 0.0306, "step": 17208 }, { "epoch": 0.917470810897265, "grad_norm": 2.0015604496002197, "learning_rate": 6.558400000000001e-06, "loss": 0.0251, "step": 17209 }, { "epoch": 0.917524124326918, "grad_norm": 2.379253387451172, "learning_rate": 6.5582e-06, "loss": 0.037, "step": 17210 }, { "epoch": 0.9175774377565709, "grad_norm": 2.33516526222229, "learning_rate": 6.558000000000001e-06, "loss": 0.0693, "step": 17211 }, { "epoch": 0.9176307511862238, "grad_norm": 1.810300588607788, "learning_rate": 6.557800000000001e-06, "loss": 0.0273, "step": 17212 }, { "epoch": 0.9176840646158767, "grad_norm": 1.2222156524658203, "learning_rate": 6.5576000000000005e-06, "loss": 0.0123, "step": 17213 }, { "epoch": 0.9177373780455297, "grad_norm": 1.255149483680725, "learning_rate": 6.5574e-06, "loss": 0.0131, "step": 17214 }, { "epoch": 0.9177906914751826, "grad_norm": 2.463756561279297, "learning_rate": 6.5572000000000005e-06, "loss": 0.0262, "step": 17215 }, { "epoch": 0.9178440049048355, "grad_norm": 1.3393809795379639, "learning_rate": 6.557e-06, "loss": 0.0202, "step": 17216 }, { "epoch": 0.9178973183344885, "grad_norm": 1.9075583219528198, "learning_rate": 6.556800000000001e-06, "loss": 0.0235, "step": 17217 }, { "epoch": 0.9179506317641414, "grad_norm": 3.6503148078918457, "learning_rate": 6.556600000000001e-06, "loss": 0.0188, "step": 17218 }, { "epoch": 0.9180039451937944, "grad_norm": 0.9693248271942139, "learning_rate": 6.5564000000000004e-06, "loss": 0.0103, "step": 17219 }, { "epoch": 0.9180572586234472, "grad_norm": 2.514125347137451, "learning_rate": 6.5562e-06, "loss": 0.0402, "step": 17220 }, { "epoch": 0.9181105720531002, "grad_norm": 2.6331803798675537, "learning_rate": 6.556e-06, "loss": 0.0436, "step": 17221 }, { "epoch": 0.9181638854827531, "grad_norm": 1.5397249460220337, "learning_rate": 6.555800000000001e-06, "loss": 0.0123, "step": 17222 }, { "epoch": 0.918217198912406, "grad_norm": 1.8178231716156006, "learning_rate": 6.555600000000001e-06, "loss": 0.0308, "step": 17223 }, { "epoch": 0.9182705123420589, "grad_norm": 1.9195396900177002, "learning_rate": 6.555400000000001e-06, "loss": 0.027, "step": 17224 }, { "epoch": 0.9183238257717119, "grad_norm": 2.397073984146118, "learning_rate": 6.5552e-06, "loss": 0.0311, "step": 17225 }, { "epoch": 0.9183771392013649, "grad_norm": 1.4464939832687378, "learning_rate": 6.555e-06, "loss": 0.0473, "step": 17226 }, { "epoch": 0.9184304526310177, "grad_norm": 2.127274990081787, "learning_rate": 6.5548e-06, "loss": 0.0299, "step": 17227 }, { "epoch": 0.9184837660606707, "grad_norm": 1.3400965929031372, "learning_rate": 6.554600000000001e-06, "loss": 0.0186, "step": 17228 }, { "epoch": 0.9185370794903236, "grad_norm": 0.7031359672546387, "learning_rate": 6.554400000000001e-06, "loss": 0.0067, "step": 17229 }, { "epoch": 0.9185903929199766, "grad_norm": 2.0717952251434326, "learning_rate": 6.554200000000001e-06, "loss": 0.0496, "step": 17230 }, { "epoch": 0.9186437063496294, "grad_norm": 4.17569637298584, "learning_rate": 6.554e-06, "loss": 0.0158, "step": 17231 }, { "epoch": 0.9186970197792824, "grad_norm": 1.72865891456604, "learning_rate": 6.5538e-06, "loss": 0.0233, "step": 17232 }, { "epoch": 0.9187503332089353, "grad_norm": 2.7829396724700928, "learning_rate": 6.553600000000001e-06, "loss": 0.0362, "step": 17233 }, { "epoch": 0.9188036466385883, "grad_norm": 1.51378333568573, "learning_rate": 6.553400000000001e-06, "loss": 0.011, "step": 17234 }, { "epoch": 0.9188569600682411, "grad_norm": 2.0835812091827393, "learning_rate": 6.553200000000001e-06, "loss": 0.0324, "step": 17235 }, { "epoch": 0.9189102734978941, "grad_norm": 2.532015562057495, "learning_rate": 6.553000000000001e-06, "loss": 0.0252, "step": 17236 }, { "epoch": 0.9189635869275471, "grad_norm": 1.6284904479980469, "learning_rate": 6.5528e-06, "loss": 0.0334, "step": 17237 }, { "epoch": 0.9190169003572, "grad_norm": 1.0913307666778564, "learning_rate": 6.5526e-06, "loss": 0.012, "step": 17238 }, { "epoch": 0.9190702137868529, "grad_norm": 0.9082802534103394, "learning_rate": 6.552400000000001e-06, "loss": 0.0059, "step": 17239 }, { "epoch": 0.9191235272165058, "grad_norm": 1.5768827199935913, "learning_rate": 6.5522000000000006e-06, "loss": 0.0302, "step": 17240 }, { "epoch": 0.9191768406461588, "grad_norm": 2.4764668941497803, "learning_rate": 6.552000000000001e-06, "loss": 0.0261, "step": 17241 }, { "epoch": 0.9192301540758117, "grad_norm": 1.5905100107192993, "learning_rate": 6.5518000000000005e-06, "loss": 0.0289, "step": 17242 }, { "epoch": 0.9192834675054646, "grad_norm": 2.9792354106903076, "learning_rate": 6.5516e-06, "loss": 0.0595, "step": 17243 }, { "epoch": 0.9193367809351175, "grad_norm": 2.4826815128326416, "learning_rate": 6.5514e-06, "loss": 0.0289, "step": 17244 }, { "epoch": 0.9193900943647705, "grad_norm": 4.859938144683838, "learning_rate": 6.551200000000001e-06, "loss": 0.0286, "step": 17245 }, { "epoch": 0.9194434077944235, "grad_norm": 2.168203115463257, "learning_rate": 6.5510000000000005e-06, "loss": 0.0617, "step": 17246 }, { "epoch": 0.9194967212240763, "grad_norm": 3.4857897758483887, "learning_rate": 6.550800000000001e-06, "loss": 0.0794, "step": 17247 }, { "epoch": 0.9195500346537293, "grad_norm": 1.3285927772521973, "learning_rate": 6.5506000000000004e-06, "loss": 0.0167, "step": 17248 }, { "epoch": 0.9196033480833822, "grad_norm": 3.14686918258667, "learning_rate": 6.5504e-06, "loss": 0.0385, "step": 17249 }, { "epoch": 0.9196566615130352, "grad_norm": 2.0846469402313232, "learning_rate": 6.550200000000001e-06, "loss": 0.0554, "step": 17250 }, { "epoch": 0.919709974942688, "grad_norm": 1.6921449899673462, "learning_rate": 6.550000000000001e-06, "loss": 0.0185, "step": 17251 }, { "epoch": 0.919763288372341, "grad_norm": 1.400668978691101, "learning_rate": 6.5498e-06, "loss": 0.011, "step": 17252 }, { "epoch": 0.9198166018019939, "grad_norm": 0.21666282415390015, "learning_rate": 6.549600000000001e-06, "loss": 0.0024, "step": 17253 }, { "epoch": 0.9198699152316469, "grad_norm": 2.0611822605133057, "learning_rate": 6.5494e-06, "loss": 0.0273, "step": 17254 }, { "epoch": 0.9199232286612998, "grad_norm": 2.0374257564544678, "learning_rate": 6.5492e-06, "loss": 0.0426, "step": 17255 }, { "epoch": 0.9199765420909527, "grad_norm": 1.7685235738754272, "learning_rate": 6.549000000000001e-06, "loss": 0.0247, "step": 17256 }, { "epoch": 0.9200298555206057, "grad_norm": 2.967355251312256, "learning_rate": 6.548800000000001e-06, "loss": 0.0397, "step": 17257 }, { "epoch": 0.9200831689502585, "grad_norm": 1.5187277793884277, "learning_rate": 6.5486e-06, "loss": 0.0195, "step": 17258 }, { "epoch": 0.9201364823799115, "grad_norm": 1.661956548690796, "learning_rate": 6.548400000000001e-06, "loss": 0.0247, "step": 17259 }, { "epoch": 0.9201897958095644, "grad_norm": 3.069122552871704, "learning_rate": 6.5482e-06, "loss": 0.0538, "step": 17260 }, { "epoch": 0.9202431092392174, "grad_norm": 2.996387481689453, "learning_rate": 6.548000000000001e-06, "loss": 0.0467, "step": 17261 }, { "epoch": 0.9202964226688702, "grad_norm": 3.938375234603882, "learning_rate": 6.547800000000001e-06, "loss": 0.0399, "step": 17262 }, { "epoch": 0.9203497360985232, "grad_norm": 2.349461317062378, "learning_rate": 6.547600000000001e-06, "loss": 0.0429, "step": 17263 }, { "epoch": 0.9204030495281762, "grad_norm": 1.8530391454696655, "learning_rate": 6.5474e-06, "loss": 0.0227, "step": 17264 }, { "epoch": 0.9204563629578291, "grad_norm": 1.4054999351501465, "learning_rate": 6.547200000000001e-06, "loss": 0.0113, "step": 17265 }, { "epoch": 0.920509676387482, "grad_norm": 2.1235389709472656, "learning_rate": 6.547e-06, "loss": 0.0385, "step": 17266 }, { "epoch": 0.9205629898171349, "grad_norm": 2.5569608211517334, "learning_rate": 6.5468000000000006e-06, "loss": 0.0293, "step": 17267 }, { "epoch": 0.9206163032467879, "grad_norm": 5.045124530792236, "learning_rate": 6.546600000000001e-06, "loss": 0.0548, "step": 17268 }, { "epoch": 0.9206696166764408, "grad_norm": 2.2743101119995117, "learning_rate": 6.5464000000000005e-06, "loss": 0.0494, "step": 17269 }, { "epoch": 0.9207229301060937, "grad_norm": 1.9191796779632568, "learning_rate": 6.5462e-06, "loss": 0.04, "step": 17270 }, { "epoch": 0.9207762435357466, "grad_norm": 0.7385162115097046, "learning_rate": 6.5460000000000005e-06, "loss": 0.0053, "step": 17271 }, { "epoch": 0.9208295569653996, "grad_norm": 0.9186854958534241, "learning_rate": 6.545800000000001e-06, "loss": 0.0133, "step": 17272 }, { "epoch": 0.9208828703950526, "grad_norm": 0.9314083456993103, "learning_rate": 6.5456000000000005e-06, "loss": 0.0069, "step": 17273 }, { "epoch": 0.9209361838247054, "grad_norm": 2.3125040531158447, "learning_rate": 6.545400000000001e-06, "loss": 0.0318, "step": 17274 }, { "epoch": 0.9209894972543584, "grad_norm": 1.9205176830291748, "learning_rate": 6.5452000000000005e-06, "loss": 0.0217, "step": 17275 }, { "epoch": 0.9210428106840113, "grad_norm": 2.290761709213257, "learning_rate": 6.545e-06, "loss": 0.0149, "step": 17276 }, { "epoch": 0.9210961241136643, "grad_norm": 1.7448034286499023, "learning_rate": 6.5448000000000004e-06, "loss": 0.0206, "step": 17277 }, { "epoch": 0.9211494375433171, "grad_norm": 0.9561669826507568, "learning_rate": 6.544600000000001e-06, "loss": 0.0072, "step": 17278 }, { "epoch": 0.9212027509729701, "grad_norm": 3.5477190017700195, "learning_rate": 6.5444e-06, "loss": 0.0585, "step": 17279 }, { "epoch": 0.921256064402623, "grad_norm": 2.7341501712799072, "learning_rate": 6.544200000000001e-06, "loss": 0.0328, "step": 17280 }, { "epoch": 0.921309377832276, "grad_norm": 1.6998777389526367, "learning_rate": 6.544e-06, "loss": 0.0225, "step": 17281 }, { "epoch": 0.9213626912619289, "grad_norm": 2.693425416946411, "learning_rate": 6.5438e-06, "loss": 0.0393, "step": 17282 }, { "epoch": 0.9214160046915818, "grad_norm": 1.630143642425537, "learning_rate": 6.543600000000001e-06, "loss": 0.0224, "step": 17283 }, { "epoch": 0.9214693181212348, "grad_norm": 2.9970576763153076, "learning_rate": 6.543400000000001e-06, "loss": 0.0507, "step": 17284 }, { "epoch": 0.9215226315508876, "grad_norm": 1.2168463468551636, "learning_rate": 6.5432e-06, "loss": 0.02, "step": 17285 }, { "epoch": 0.9215759449805406, "grad_norm": 4.0231852531433105, "learning_rate": 6.543000000000001e-06, "loss": 0.0467, "step": 17286 }, { "epoch": 0.9216292584101935, "grad_norm": 1.4304195642471313, "learning_rate": 6.5428e-06, "loss": 0.0269, "step": 17287 }, { "epoch": 0.9216825718398465, "grad_norm": 2.9604544639587402, "learning_rate": 6.5426e-06, "loss": 0.0361, "step": 17288 }, { "epoch": 0.9217358852694993, "grad_norm": 1.5065910816192627, "learning_rate": 6.542400000000001e-06, "loss": 0.0213, "step": 17289 }, { "epoch": 0.9217891986991523, "grad_norm": 0.9997108578681946, "learning_rate": 6.542200000000001e-06, "loss": 0.0283, "step": 17290 }, { "epoch": 0.9218425121288052, "grad_norm": 1.663887858390808, "learning_rate": 6.542e-06, "loss": 0.0211, "step": 17291 }, { "epoch": 0.9218958255584582, "grad_norm": 1.7999945878982544, "learning_rate": 6.541800000000001e-06, "loss": 0.0168, "step": 17292 }, { "epoch": 0.9219491389881111, "grad_norm": 1.8507258892059326, "learning_rate": 6.5416e-06, "loss": 0.0219, "step": 17293 }, { "epoch": 0.922002452417764, "grad_norm": 1.4152313470840454, "learning_rate": 6.5414000000000014e-06, "loss": 0.0271, "step": 17294 }, { "epoch": 0.922055765847417, "grad_norm": 2.3395864963531494, "learning_rate": 6.541200000000001e-06, "loss": 0.022, "step": 17295 }, { "epoch": 0.9221090792770699, "grad_norm": 1.69198477268219, "learning_rate": 6.5410000000000006e-06, "loss": 0.0153, "step": 17296 }, { "epoch": 0.9221623927067228, "grad_norm": 0.8975746035575867, "learning_rate": 6.5408e-06, "loss": 0.0074, "step": 17297 }, { "epoch": 0.9222157061363757, "grad_norm": 0.6898269653320312, "learning_rate": 6.5406000000000005e-06, "loss": 0.0204, "step": 17298 }, { "epoch": 0.9222690195660287, "grad_norm": 1.833630919456482, "learning_rate": 6.5404e-06, "loss": 0.0436, "step": 17299 }, { "epoch": 0.9223223329956816, "grad_norm": 1.6122841835021973, "learning_rate": 6.540200000000001e-06, "loss": 0.0188, "step": 17300 }, { "epoch": 0.9223756464253345, "grad_norm": 0.6182820200920105, "learning_rate": 6.540000000000001e-06, "loss": 0.0097, "step": 17301 }, { "epoch": 0.9224289598549875, "grad_norm": 2.741241931915283, "learning_rate": 6.5398000000000005e-06, "loss": 0.0383, "step": 17302 }, { "epoch": 0.9224822732846404, "grad_norm": 1.4100415706634521, "learning_rate": 6.5396e-06, "loss": 0.0383, "step": 17303 }, { "epoch": 0.9225355867142934, "grad_norm": 1.8678675889968872, "learning_rate": 6.5394000000000004e-06, "loss": 0.0304, "step": 17304 }, { "epoch": 0.9225889001439462, "grad_norm": 0.9341151714324951, "learning_rate": 6.5392e-06, "loss": 0.0083, "step": 17305 }, { "epoch": 0.9226422135735992, "grad_norm": 2.069718599319458, "learning_rate": 6.539000000000001e-06, "loss": 0.0211, "step": 17306 }, { "epoch": 0.9226955270032521, "grad_norm": 0.7771971225738525, "learning_rate": 6.538800000000001e-06, "loss": 0.0063, "step": 17307 }, { "epoch": 0.9227488404329051, "grad_norm": 1.1783843040466309, "learning_rate": 6.5386e-06, "loss": 0.0085, "step": 17308 }, { "epoch": 0.9228021538625579, "grad_norm": 2.1291890144348145, "learning_rate": 6.5384e-06, "loss": 0.0115, "step": 17309 }, { "epoch": 0.9228554672922109, "grad_norm": 0.7468441724777222, "learning_rate": 6.5382e-06, "loss": 0.0069, "step": 17310 }, { "epoch": 0.9229087807218639, "grad_norm": 2.9618284702301025, "learning_rate": 6.538000000000001e-06, "loss": 0.0464, "step": 17311 }, { "epoch": 0.9229620941515168, "grad_norm": 0.9381659626960754, "learning_rate": 6.537800000000001e-06, "loss": 0.0175, "step": 17312 }, { "epoch": 0.9230154075811697, "grad_norm": 1.7944797277450562, "learning_rate": 6.537600000000001e-06, "loss": 0.0205, "step": 17313 }, { "epoch": 0.9230687210108226, "grad_norm": 2.5632855892181396, "learning_rate": 6.5374e-06, "loss": 0.0263, "step": 17314 }, { "epoch": 0.9231220344404756, "grad_norm": 1.1167187690734863, "learning_rate": 6.5372e-06, "loss": 0.0139, "step": 17315 }, { "epoch": 0.9231753478701284, "grad_norm": 1.2084826231002808, "learning_rate": 6.537e-06, "loss": 0.0276, "step": 17316 }, { "epoch": 0.9232286612997814, "grad_norm": 3.7518882751464844, "learning_rate": 6.536800000000001e-06, "loss": 0.0389, "step": 17317 }, { "epoch": 0.9232819747294343, "grad_norm": 2.2568347454071045, "learning_rate": 6.536600000000001e-06, "loss": 0.0505, "step": 17318 }, { "epoch": 0.9233352881590873, "grad_norm": 1.8405884504318237, "learning_rate": 6.536400000000001e-06, "loss": 0.0225, "step": 17319 }, { "epoch": 0.9233886015887403, "grad_norm": 1.76751708984375, "learning_rate": 6.5362e-06, "loss": 0.0239, "step": 17320 }, { "epoch": 0.9234419150183931, "grad_norm": 1.956631064414978, "learning_rate": 6.536e-06, "loss": 0.0274, "step": 17321 }, { "epoch": 0.9234952284480461, "grad_norm": 2.3660714626312256, "learning_rate": 6.535800000000001e-06, "loss": 0.0398, "step": 17322 }, { "epoch": 0.923548541877699, "grad_norm": 1.6983458995819092, "learning_rate": 6.535600000000001e-06, "loss": 0.0477, "step": 17323 }, { "epoch": 0.923601855307352, "grad_norm": 1.6819041967391968, "learning_rate": 6.535400000000001e-06, "loss": 0.0359, "step": 17324 }, { "epoch": 0.9236551687370048, "grad_norm": 3.0714590549468994, "learning_rate": 6.5352000000000006e-06, "loss": 0.0365, "step": 17325 }, { "epoch": 0.9237084821666578, "grad_norm": 1.6441267728805542, "learning_rate": 6.535e-06, "loss": 0.021, "step": 17326 }, { "epoch": 0.9237617955963107, "grad_norm": 1.207948923110962, "learning_rate": 6.5348e-06, "loss": 0.0294, "step": 17327 }, { "epoch": 0.9238151090259636, "grad_norm": 0.9893777370452881, "learning_rate": 6.534600000000001e-06, "loss": 0.0124, "step": 17328 }, { "epoch": 0.9238684224556166, "grad_norm": 2.0805723667144775, "learning_rate": 6.5344000000000005e-06, "loss": 0.022, "step": 17329 }, { "epoch": 0.9239217358852695, "grad_norm": 1.8260542154312134, "learning_rate": 6.534200000000001e-06, "loss": 0.0374, "step": 17330 }, { "epoch": 0.9239750493149225, "grad_norm": 1.1458492279052734, "learning_rate": 6.5340000000000005e-06, "loss": 0.0238, "step": 17331 }, { "epoch": 0.9240283627445753, "grad_norm": 1.6443463563919067, "learning_rate": 6.5338e-06, "loss": 0.0306, "step": 17332 }, { "epoch": 0.9240816761742283, "grad_norm": 2.474059581756592, "learning_rate": 6.533600000000001e-06, "loss": 0.0304, "step": 17333 }, { "epoch": 0.9241349896038812, "grad_norm": 2.386807441711426, "learning_rate": 6.533400000000001e-06, "loss": 0.0624, "step": 17334 }, { "epoch": 0.9241883030335342, "grad_norm": 1.8558605909347534, "learning_rate": 6.5332e-06, "loss": 0.0176, "step": 17335 }, { "epoch": 0.924241616463187, "grad_norm": 2.318657398223877, "learning_rate": 6.533000000000001e-06, "loss": 0.0388, "step": 17336 }, { "epoch": 0.92429492989284, "grad_norm": 2.0622286796569824, "learning_rate": 6.5328e-06, "loss": 0.0274, "step": 17337 }, { "epoch": 0.924348243322493, "grad_norm": 1.1962956190109253, "learning_rate": 6.5326e-06, "loss": 0.029, "step": 17338 }, { "epoch": 0.9244015567521459, "grad_norm": 1.6133661270141602, "learning_rate": 6.532400000000001e-06, "loss": 0.0333, "step": 17339 }, { "epoch": 0.9244548701817988, "grad_norm": 1.3740143775939941, "learning_rate": 6.532200000000001e-06, "loss": 0.0342, "step": 17340 }, { "epoch": 0.9245081836114517, "grad_norm": 1.5968743562698364, "learning_rate": 6.532e-06, "loss": 0.0201, "step": 17341 }, { "epoch": 0.9245614970411047, "grad_norm": 1.7602235078811646, "learning_rate": 6.531800000000001e-06, "loss": 0.0142, "step": 17342 }, { "epoch": 0.9246148104707576, "grad_norm": 3.39058518409729, "learning_rate": 6.5316e-06, "loss": 0.0329, "step": 17343 }, { "epoch": 0.9246681239004105, "grad_norm": 1.134793996810913, "learning_rate": 6.531400000000001e-06, "loss": 0.0141, "step": 17344 }, { "epoch": 0.9247214373300634, "grad_norm": 1.0610013008117676, "learning_rate": 6.531200000000001e-06, "loss": 0.0097, "step": 17345 }, { "epoch": 0.9247747507597164, "grad_norm": 1.4600132703781128, "learning_rate": 6.531000000000001e-06, "loss": 0.0218, "step": 17346 }, { "epoch": 0.9248280641893692, "grad_norm": 3.282440185546875, "learning_rate": 6.5308e-06, "loss": 0.0368, "step": 17347 }, { "epoch": 0.9248813776190222, "grad_norm": 3.1953794956207275, "learning_rate": 6.530600000000001e-06, "loss": 0.0654, "step": 17348 }, { "epoch": 0.9249346910486752, "grad_norm": 1.5486924648284912, "learning_rate": 6.5304e-06, "loss": 0.0352, "step": 17349 }, { "epoch": 0.9249880044783281, "grad_norm": 1.610761284828186, "learning_rate": 6.530200000000001e-06, "loss": 0.0365, "step": 17350 }, { "epoch": 0.925041317907981, "grad_norm": 1.1207278966903687, "learning_rate": 6.530000000000001e-06, "loss": 0.0062, "step": 17351 }, { "epoch": 0.9250946313376339, "grad_norm": 1.7441500425338745, "learning_rate": 6.5298000000000006e-06, "loss": 0.0195, "step": 17352 }, { "epoch": 0.9251479447672869, "grad_norm": 1.1952135562896729, "learning_rate": 6.5296e-06, "loss": 0.0153, "step": 17353 }, { "epoch": 0.9252012581969398, "grad_norm": 4.21809196472168, "learning_rate": 6.5294000000000005e-06, "loss": 0.0324, "step": 17354 }, { "epoch": 0.9252545716265927, "grad_norm": 3.046954393386841, "learning_rate": 6.5292e-06, "loss": 0.0249, "step": 17355 }, { "epoch": 0.9253078850562456, "grad_norm": 1.0190908908843994, "learning_rate": 6.5290000000000005e-06, "loss": 0.0179, "step": 17356 }, { "epoch": 0.9253611984858986, "grad_norm": 3.226897716522217, "learning_rate": 6.528800000000001e-06, "loss": 0.0307, "step": 17357 }, { "epoch": 0.9254145119155516, "grad_norm": 1.8612136840820312, "learning_rate": 6.5286000000000005e-06, "loss": 0.0372, "step": 17358 }, { "epoch": 0.9254678253452044, "grad_norm": 2.07318115234375, "learning_rate": 6.5284e-06, "loss": 0.0261, "step": 17359 }, { "epoch": 0.9255211387748574, "grad_norm": 1.6165997982025146, "learning_rate": 6.5282000000000004e-06, "loss": 0.0204, "step": 17360 }, { "epoch": 0.9255744522045103, "grad_norm": 2.0094189643859863, "learning_rate": 6.528000000000001e-06, "loss": 0.0352, "step": 17361 }, { "epoch": 0.9256277656341633, "grad_norm": 3.8903348445892334, "learning_rate": 6.5278e-06, "loss": 0.0493, "step": 17362 }, { "epoch": 0.9256810790638161, "grad_norm": 2.1015830039978027, "learning_rate": 6.527600000000001e-06, "loss": 0.0321, "step": 17363 }, { "epoch": 0.9257343924934691, "grad_norm": 1.373698115348816, "learning_rate": 6.5274e-06, "loss": 0.0203, "step": 17364 }, { "epoch": 0.925787705923122, "grad_norm": 1.1118181943893433, "learning_rate": 6.5272e-06, "loss": 0.0157, "step": 17365 }, { "epoch": 0.925841019352775, "grad_norm": 2.564530849456787, "learning_rate": 6.527e-06, "loss": 0.0189, "step": 17366 }, { "epoch": 0.9258943327824279, "grad_norm": 2.268474817276001, "learning_rate": 6.526800000000001e-06, "loss": 0.0391, "step": 17367 }, { "epoch": 0.9259476462120808, "grad_norm": 1.3158024549484253, "learning_rate": 6.5266e-06, "loss": 0.0199, "step": 17368 }, { "epoch": 0.9260009596417338, "grad_norm": 1.4811146259307861, "learning_rate": 6.526400000000001e-06, "loss": 0.0219, "step": 17369 }, { "epoch": 0.9260542730713867, "grad_norm": 1.6516926288604736, "learning_rate": 6.5262e-06, "loss": 0.0331, "step": 17370 }, { "epoch": 0.9261075865010396, "grad_norm": 1.6205590963363647, "learning_rate": 6.526e-06, "loss": 0.0413, "step": 17371 }, { "epoch": 0.9261608999306925, "grad_norm": 2.0326907634735107, "learning_rate": 6.525800000000001e-06, "loss": 0.0172, "step": 17372 }, { "epoch": 0.9262142133603455, "grad_norm": 1.6619343757629395, "learning_rate": 6.525600000000001e-06, "loss": 0.0254, "step": 17373 }, { "epoch": 0.9262675267899984, "grad_norm": 1.968498706817627, "learning_rate": 6.5254e-06, "loss": 0.0308, "step": 17374 }, { "epoch": 0.9263208402196513, "grad_norm": 1.1724598407745361, "learning_rate": 6.525200000000001e-06, "loss": 0.0117, "step": 17375 }, { "epoch": 0.9263741536493043, "grad_norm": 1.4987884759902954, "learning_rate": 6.525e-06, "loss": 0.0121, "step": 17376 }, { "epoch": 0.9264274670789572, "grad_norm": 2.0582315921783447, "learning_rate": 6.5248e-06, "loss": 0.0178, "step": 17377 }, { "epoch": 0.9264807805086102, "grad_norm": 1.0681767463684082, "learning_rate": 6.524600000000001e-06, "loss": 0.0084, "step": 17378 }, { "epoch": 0.926534093938263, "grad_norm": 2.2125043869018555, "learning_rate": 6.524400000000001e-06, "loss": 0.0195, "step": 17379 }, { "epoch": 0.926587407367916, "grad_norm": 2.157320022583008, "learning_rate": 6.5242e-06, "loss": 0.0234, "step": 17380 }, { "epoch": 0.9266407207975689, "grad_norm": 1.1878851652145386, "learning_rate": 6.5240000000000006e-06, "loss": 0.0172, "step": 17381 }, { "epoch": 0.9266940342272219, "grad_norm": 0.8252378702163696, "learning_rate": 6.5238e-06, "loss": 0.0066, "step": 17382 }, { "epoch": 0.9267473476568747, "grad_norm": 1.7787203788757324, "learning_rate": 6.523600000000001e-06, "loss": 0.0204, "step": 17383 }, { "epoch": 0.9268006610865277, "grad_norm": 1.8332635164260864, "learning_rate": 6.523400000000001e-06, "loss": 0.0508, "step": 17384 }, { "epoch": 0.9268539745161807, "grad_norm": 1.347293496131897, "learning_rate": 6.5232000000000005e-06, "loss": 0.0154, "step": 17385 }, { "epoch": 0.9269072879458335, "grad_norm": 2.2989439964294434, "learning_rate": 6.523e-06, "loss": 0.0342, "step": 17386 }, { "epoch": 0.9269606013754865, "grad_norm": 1.4295624494552612, "learning_rate": 6.5228000000000005e-06, "loss": 0.0205, "step": 17387 }, { "epoch": 0.9270139148051394, "grad_norm": 2.0808300971984863, "learning_rate": 6.5226e-06, "loss": 0.0241, "step": 17388 }, { "epoch": 0.9270672282347924, "grad_norm": 3.2803354263305664, "learning_rate": 6.522400000000001e-06, "loss": 0.05, "step": 17389 }, { "epoch": 0.9271205416644452, "grad_norm": 0.7460275292396545, "learning_rate": 6.522200000000001e-06, "loss": 0.0058, "step": 17390 }, { "epoch": 0.9271738550940982, "grad_norm": 2.9925482273101807, "learning_rate": 6.522e-06, "loss": 0.0222, "step": 17391 }, { "epoch": 0.9272271685237511, "grad_norm": 3.640413761138916, "learning_rate": 6.5218e-06, "loss": 0.0203, "step": 17392 }, { "epoch": 0.9272804819534041, "grad_norm": 2.025733470916748, "learning_rate": 6.5216e-06, "loss": 0.0247, "step": 17393 }, { "epoch": 0.927333795383057, "grad_norm": 2.2280211448669434, "learning_rate": 6.521400000000001e-06, "loss": 0.0155, "step": 17394 }, { "epoch": 0.9273871088127099, "grad_norm": 2.0018045902252197, "learning_rate": 6.521200000000001e-06, "loss": 0.0223, "step": 17395 }, { "epoch": 0.9274404222423629, "grad_norm": 2.517080783843994, "learning_rate": 6.521000000000001e-06, "loss": 0.0125, "step": 17396 }, { "epoch": 0.9274937356720158, "grad_norm": 1.4287173748016357, "learning_rate": 6.5208e-06, "loss": 0.0267, "step": 17397 }, { "epoch": 0.9275470491016687, "grad_norm": 1.0820399522781372, "learning_rate": 6.5206e-06, "loss": 0.0115, "step": 17398 }, { "epoch": 0.9276003625313216, "grad_norm": 2.4979491233825684, "learning_rate": 6.5204e-06, "loss": 0.0252, "step": 17399 }, { "epoch": 0.9276536759609746, "grad_norm": 1.0517383813858032, "learning_rate": 6.520200000000001e-06, "loss": 0.0119, "step": 17400 }, { "epoch": 0.9277069893906275, "grad_norm": 1.731819987297058, "learning_rate": 6.520000000000001e-06, "loss": 0.0357, "step": 17401 }, { "epoch": 0.9277603028202804, "grad_norm": 4.586219310760498, "learning_rate": 6.519800000000001e-06, "loss": 0.0401, "step": 17402 }, { "epoch": 0.9278136162499333, "grad_norm": 1.683852195739746, "learning_rate": 6.5196e-06, "loss": 0.0362, "step": 17403 }, { "epoch": 0.9278669296795863, "grad_norm": 0.5679886937141418, "learning_rate": 6.5194e-06, "loss": 0.0044, "step": 17404 }, { "epoch": 0.9279202431092393, "grad_norm": 0.9557117819786072, "learning_rate": 6.519200000000001e-06, "loss": 0.021, "step": 17405 }, { "epoch": 0.9279735565388921, "grad_norm": 0.5680315494537354, "learning_rate": 6.519000000000001e-06, "loss": 0.0074, "step": 17406 }, { "epoch": 0.9280268699685451, "grad_norm": 0.9820294380187988, "learning_rate": 6.518800000000001e-06, "loss": 0.0164, "step": 17407 }, { "epoch": 0.928080183398198, "grad_norm": 2.3192951679229736, "learning_rate": 6.5186000000000006e-06, "loss": 0.0159, "step": 17408 }, { "epoch": 0.928133496827851, "grad_norm": 1.6506811380386353, "learning_rate": 6.5184e-06, "loss": 0.0119, "step": 17409 }, { "epoch": 0.9281868102575038, "grad_norm": 2.5772690773010254, "learning_rate": 6.5182e-06, "loss": 0.0268, "step": 17410 }, { "epoch": 0.9282401236871568, "grad_norm": 1.8491709232330322, "learning_rate": 6.518000000000001e-06, "loss": 0.0169, "step": 17411 }, { "epoch": 0.9282934371168097, "grad_norm": 1.0129908323287964, "learning_rate": 6.5178000000000005e-06, "loss": 0.0126, "step": 17412 }, { "epoch": 0.9283467505464627, "grad_norm": 0.9753547310829163, "learning_rate": 6.517600000000001e-06, "loss": 0.0141, "step": 17413 }, { "epoch": 0.9284000639761156, "grad_norm": 2.229701280593872, "learning_rate": 6.5174000000000005e-06, "loss": 0.0322, "step": 17414 }, { "epoch": 0.9284533774057685, "grad_norm": 1.6119860410690308, "learning_rate": 6.5172e-06, "loss": 0.0188, "step": 17415 }, { "epoch": 0.9285066908354215, "grad_norm": 1.059906005859375, "learning_rate": 6.517e-06, "loss": 0.007, "step": 17416 }, { "epoch": 0.9285600042650743, "grad_norm": 1.4178639650344849, "learning_rate": 6.516800000000001e-06, "loss": 0.0289, "step": 17417 }, { "epoch": 0.9286133176947273, "grad_norm": 3.1866748332977295, "learning_rate": 6.5166000000000004e-06, "loss": 0.0369, "step": 17418 }, { "epoch": 0.9286666311243802, "grad_norm": 1.9432629346847534, "learning_rate": 6.516400000000001e-06, "loss": 0.0222, "step": 17419 }, { "epoch": 0.9287199445540332, "grad_norm": 4.50145149230957, "learning_rate": 6.5162e-06, "loss": 0.0458, "step": 17420 }, { "epoch": 0.928773257983686, "grad_norm": 3.197446823120117, "learning_rate": 6.516e-06, "loss": 0.0335, "step": 17421 }, { "epoch": 0.928826571413339, "grad_norm": 2.5916898250579834, "learning_rate": 6.515800000000001e-06, "loss": 0.0376, "step": 17422 }, { "epoch": 0.928879884842992, "grad_norm": 1.3739686012268066, "learning_rate": 6.515600000000001e-06, "loss": 0.0179, "step": 17423 }, { "epoch": 0.9289331982726449, "grad_norm": 1.282509207725525, "learning_rate": 6.5154e-06, "loss": 0.0155, "step": 17424 }, { "epoch": 0.9289865117022978, "grad_norm": 1.9211570024490356, "learning_rate": 6.515200000000001e-06, "loss": 0.0274, "step": 17425 }, { "epoch": 0.9290398251319507, "grad_norm": 1.7118700742721558, "learning_rate": 6.515e-06, "loss": 0.0264, "step": 17426 }, { "epoch": 0.9290931385616037, "grad_norm": 2.848649501800537, "learning_rate": 6.5148e-06, "loss": 0.0401, "step": 17427 }, { "epoch": 0.9291464519912566, "grad_norm": 2.3486218452453613, "learning_rate": 6.514600000000001e-06, "loss": 0.0341, "step": 17428 }, { "epoch": 0.9291997654209095, "grad_norm": 0.6274452805519104, "learning_rate": 6.514400000000001e-06, "loss": 0.0059, "step": 17429 }, { "epoch": 0.9292530788505624, "grad_norm": 3.2212233543395996, "learning_rate": 6.5142e-06, "loss": 0.0449, "step": 17430 }, { "epoch": 0.9293063922802154, "grad_norm": 1.1609725952148438, "learning_rate": 6.514000000000001e-06, "loss": 0.0136, "step": 17431 }, { "epoch": 0.9293597057098684, "grad_norm": 1.7186501026153564, "learning_rate": 6.5138e-06, "loss": 0.0225, "step": 17432 }, { "epoch": 0.9294130191395212, "grad_norm": 0.8710095286369324, "learning_rate": 6.513600000000001e-06, "loss": 0.0077, "step": 17433 }, { "epoch": 0.9294663325691742, "grad_norm": 0.7242566347122192, "learning_rate": 6.513400000000001e-06, "loss": 0.0075, "step": 17434 }, { "epoch": 0.9295196459988271, "grad_norm": 2.438155174255371, "learning_rate": 6.513200000000001e-06, "loss": 0.0208, "step": 17435 }, { "epoch": 0.9295729594284801, "grad_norm": 1.8237429857254028, "learning_rate": 6.513e-06, "loss": 0.0277, "step": 17436 }, { "epoch": 0.9296262728581329, "grad_norm": 1.4904788732528687, "learning_rate": 6.5128000000000006e-06, "loss": 0.0329, "step": 17437 }, { "epoch": 0.9296795862877859, "grad_norm": 2.3489954471588135, "learning_rate": 6.5126e-06, "loss": 0.0216, "step": 17438 }, { "epoch": 0.9297328997174388, "grad_norm": 1.2115576267242432, "learning_rate": 6.5124000000000005e-06, "loss": 0.0076, "step": 17439 }, { "epoch": 0.9297862131470918, "grad_norm": 2.6750659942626953, "learning_rate": 6.512200000000001e-06, "loss": 0.0292, "step": 17440 }, { "epoch": 0.9298395265767447, "grad_norm": 1.7008187770843506, "learning_rate": 6.5120000000000005e-06, "loss": 0.0194, "step": 17441 }, { "epoch": 0.9298928400063976, "grad_norm": 1.5697544813156128, "learning_rate": 6.5118e-06, "loss": 0.0179, "step": 17442 }, { "epoch": 0.9299461534360506, "grad_norm": 1.8204667568206787, "learning_rate": 6.5116000000000005e-06, "loss": 0.0325, "step": 17443 }, { "epoch": 0.9299994668657035, "grad_norm": 0.7749442458152771, "learning_rate": 6.511400000000001e-06, "loss": 0.0181, "step": 17444 }, { "epoch": 0.9300527802953564, "grad_norm": 1.432742714881897, "learning_rate": 6.5112000000000004e-06, "loss": 0.0083, "step": 17445 }, { "epoch": 0.9301060937250093, "grad_norm": 1.8226120471954346, "learning_rate": 6.511000000000001e-06, "loss": 0.0175, "step": 17446 }, { "epoch": 0.9301594071546623, "grad_norm": 4.311884880065918, "learning_rate": 6.5108e-06, "loss": 0.0507, "step": 17447 }, { "epoch": 0.9302127205843151, "grad_norm": 2.225792646408081, "learning_rate": 6.5106e-06, "loss": 0.0173, "step": 17448 }, { "epoch": 0.9302660340139681, "grad_norm": 2.245943784713745, "learning_rate": 6.5104e-06, "loss": 0.0424, "step": 17449 }, { "epoch": 0.9303193474436211, "grad_norm": 2.032785654067993, "learning_rate": 6.510200000000001e-06, "loss": 0.0365, "step": 17450 }, { "epoch": 0.930372660873274, "grad_norm": 2.8689239025115967, "learning_rate": 6.51e-06, "loss": 0.0247, "step": 17451 }, { "epoch": 0.930425974302927, "grad_norm": 5.646244525909424, "learning_rate": 6.509800000000001e-06, "loss": 0.0361, "step": 17452 }, { "epoch": 0.9304792877325798, "grad_norm": 3.3787331581115723, "learning_rate": 6.5096e-06, "loss": 0.0167, "step": 17453 }, { "epoch": 0.9305326011622328, "grad_norm": 2.273725986480713, "learning_rate": 6.5094e-06, "loss": 0.0312, "step": 17454 }, { "epoch": 0.9305859145918857, "grad_norm": 0.784951388835907, "learning_rate": 6.509200000000001e-06, "loss": 0.0071, "step": 17455 }, { "epoch": 0.9306392280215386, "grad_norm": 1.5160752534866333, "learning_rate": 6.509000000000001e-06, "loss": 0.0132, "step": 17456 }, { "epoch": 0.9306925414511915, "grad_norm": 2.305170774459839, "learning_rate": 6.5088e-06, "loss": 0.0358, "step": 17457 }, { "epoch": 0.9307458548808445, "grad_norm": 2.3585331439971924, "learning_rate": 6.508600000000001e-06, "loss": 0.0423, "step": 17458 }, { "epoch": 0.9307991683104974, "grad_norm": 2.7038424015045166, "learning_rate": 6.5084e-06, "loss": 0.0149, "step": 17459 }, { "epoch": 0.9308524817401503, "grad_norm": 2.0010054111480713, "learning_rate": 6.5082e-06, "loss": 0.0269, "step": 17460 }, { "epoch": 0.9309057951698033, "grad_norm": 4.502321720123291, "learning_rate": 6.508000000000001e-06, "loss": 0.0358, "step": 17461 }, { "epoch": 0.9309591085994562, "grad_norm": 1.582044005393982, "learning_rate": 6.507800000000001e-06, "loss": 0.0167, "step": 17462 }, { "epoch": 0.9310124220291092, "grad_norm": 2.0913422107696533, "learning_rate": 6.5076e-06, "loss": 0.0554, "step": 17463 }, { "epoch": 0.931065735458762, "grad_norm": 3.0578887462615967, "learning_rate": 6.507400000000001e-06, "loss": 0.02, "step": 17464 }, { "epoch": 0.931119048888415, "grad_norm": 1.1079833507537842, "learning_rate": 6.5072e-06, "loss": 0.0136, "step": 17465 }, { "epoch": 0.9311723623180679, "grad_norm": 1.7310913801193237, "learning_rate": 6.507e-06, "loss": 0.021, "step": 17466 }, { "epoch": 0.9312256757477209, "grad_norm": 2.255260944366455, "learning_rate": 6.506800000000001e-06, "loss": 0.0482, "step": 17467 }, { "epoch": 0.9312789891773737, "grad_norm": 1.147841453552246, "learning_rate": 6.5066000000000005e-06, "loss": 0.0147, "step": 17468 }, { "epoch": 0.9313323026070267, "grad_norm": 1.6500755548477173, "learning_rate": 6.5064e-06, "loss": 0.027, "step": 17469 }, { "epoch": 0.9313856160366797, "grad_norm": 2.318052291870117, "learning_rate": 6.5062000000000005e-06, "loss": 0.0155, "step": 17470 }, { "epoch": 0.9314389294663326, "grad_norm": 2.9306020736694336, "learning_rate": 6.506e-06, "loss": 0.0429, "step": 17471 }, { "epoch": 0.9314922428959855, "grad_norm": 1.9760055541992188, "learning_rate": 6.505800000000001e-06, "loss": 0.0184, "step": 17472 }, { "epoch": 0.9315455563256384, "grad_norm": 1.9256587028503418, "learning_rate": 6.505600000000001e-06, "loss": 0.0169, "step": 17473 }, { "epoch": 0.9315988697552914, "grad_norm": 0.8050686120986938, "learning_rate": 6.5054000000000004e-06, "loss": 0.0179, "step": 17474 }, { "epoch": 0.9316521831849442, "grad_norm": 0.9367348551750183, "learning_rate": 6.5052e-06, "loss": 0.0174, "step": 17475 }, { "epoch": 0.9317054966145972, "grad_norm": 3.211951732635498, "learning_rate": 6.505e-06, "loss": 0.0352, "step": 17476 }, { "epoch": 0.9317588100442501, "grad_norm": 0.9950985908508301, "learning_rate": 6.5048e-06, "loss": 0.0225, "step": 17477 }, { "epoch": 0.9318121234739031, "grad_norm": 2.1477420330047607, "learning_rate": 6.504600000000001e-06, "loss": 0.0315, "step": 17478 }, { "epoch": 0.931865436903556, "grad_norm": 1.583662748336792, "learning_rate": 6.504400000000001e-06, "loss": 0.0309, "step": 17479 }, { "epoch": 0.9319187503332089, "grad_norm": 2.2933437824249268, "learning_rate": 6.5042e-06, "loss": 0.054, "step": 17480 }, { "epoch": 0.9319720637628619, "grad_norm": 1.5994666814804077, "learning_rate": 6.504e-06, "loss": 0.0247, "step": 17481 }, { "epoch": 0.9320253771925148, "grad_norm": 3.4062161445617676, "learning_rate": 6.5038e-06, "loss": 0.0431, "step": 17482 }, { "epoch": 0.9320786906221677, "grad_norm": 1.9507695436477661, "learning_rate": 6.503600000000001e-06, "loss": 0.0357, "step": 17483 }, { "epoch": 0.9321320040518206, "grad_norm": 1.7506121397018433, "learning_rate": 6.503400000000001e-06, "loss": 0.0252, "step": 17484 }, { "epoch": 0.9321853174814736, "grad_norm": 2.0477519035339355, "learning_rate": 6.503200000000001e-06, "loss": 0.0218, "step": 17485 }, { "epoch": 0.9322386309111265, "grad_norm": 0.8554439544677734, "learning_rate": 6.503e-06, "loss": 0.0082, "step": 17486 }, { "epoch": 0.9322919443407794, "grad_norm": 2.175265312194824, "learning_rate": 6.5028e-06, "loss": 0.0219, "step": 17487 }, { "epoch": 0.9323452577704324, "grad_norm": 1.9336378574371338, "learning_rate": 6.5026e-06, "loss": 0.0323, "step": 17488 }, { "epoch": 0.9323985712000853, "grad_norm": 1.612207055091858, "learning_rate": 6.502400000000001e-06, "loss": 0.0205, "step": 17489 }, { "epoch": 0.9324518846297383, "grad_norm": 1.6608920097351074, "learning_rate": 6.502200000000001e-06, "loss": 0.0307, "step": 17490 }, { "epoch": 0.9325051980593911, "grad_norm": 2.635012626647949, "learning_rate": 6.502000000000001e-06, "loss": 0.0427, "step": 17491 }, { "epoch": 0.9325585114890441, "grad_norm": 1.8086246252059937, "learning_rate": 6.5018e-06, "loss": 0.0326, "step": 17492 }, { "epoch": 0.932611824918697, "grad_norm": 5.768473148345947, "learning_rate": 6.5016e-06, "loss": 0.0545, "step": 17493 }, { "epoch": 0.93266513834835, "grad_norm": 1.3903201818466187, "learning_rate": 6.501400000000001e-06, "loss": 0.0098, "step": 17494 }, { "epoch": 0.9327184517780028, "grad_norm": 1.2395169734954834, "learning_rate": 6.5012000000000005e-06, "loss": 0.0136, "step": 17495 }, { "epoch": 0.9327717652076558, "grad_norm": 2.1243603229522705, "learning_rate": 6.501000000000001e-06, "loss": 0.0288, "step": 17496 }, { "epoch": 0.9328250786373088, "grad_norm": 1.3813567161560059, "learning_rate": 6.5008000000000005e-06, "loss": 0.0168, "step": 17497 }, { "epoch": 0.9328783920669617, "grad_norm": 1.0807400941848755, "learning_rate": 6.5006e-06, "loss": 0.0153, "step": 17498 }, { "epoch": 0.9329317054966146, "grad_norm": 3.028629779815674, "learning_rate": 6.5004e-06, "loss": 0.0663, "step": 17499 }, { "epoch": 0.9329850189262675, "grad_norm": 0.7258392572402954, "learning_rate": 6.500200000000001e-06, "loss": 0.0057, "step": 17500 }, { "epoch": 0.9330383323559205, "grad_norm": 2.5015952587127686, "learning_rate": 6.5000000000000004e-06, "loss": 0.0207, "step": 17501 }, { "epoch": 0.9330916457855734, "grad_norm": 2.00635027885437, "learning_rate": 6.499800000000001e-06, "loss": 0.0342, "step": 17502 }, { "epoch": 0.9331449592152263, "grad_norm": 0.5647954344749451, "learning_rate": 6.4996e-06, "loss": 0.0037, "step": 17503 }, { "epoch": 0.9331982726448792, "grad_norm": 1.2541605234146118, "learning_rate": 6.4994e-06, "loss": 0.0115, "step": 17504 }, { "epoch": 0.9332515860745322, "grad_norm": 1.6119046211242676, "learning_rate": 6.499200000000001e-06, "loss": 0.0182, "step": 17505 }, { "epoch": 0.9333048995041852, "grad_norm": 1.7619175910949707, "learning_rate": 6.499000000000001e-06, "loss": 0.0439, "step": 17506 }, { "epoch": 0.933358212933838, "grad_norm": 1.4863255023956299, "learning_rate": 6.4988e-06, "loss": 0.0378, "step": 17507 }, { "epoch": 0.933411526363491, "grad_norm": 2.1602602005004883, "learning_rate": 6.498600000000001e-06, "loss": 0.0283, "step": 17508 }, { "epoch": 0.9334648397931439, "grad_norm": 0.8874292373657227, "learning_rate": 6.4984e-06, "loss": 0.0077, "step": 17509 }, { "epoch": 0.9335181532227969, "grad_norm": 3.0884859561920166, "learning_rate": 6.4982e-06, "loss": 0.0228, "step": 17510 }, { "epoch": 0.9335714666524497, "grad_norm": 3.315959930419922, "learning_rate": 6.498000000000001e-06, "loss": 0.0499, "step": 17511 }, { "epoch": 0.9336247800821027, "grad_norm": 2.724735975265503, "learning_rate": 6.497800000000001e-06, "loss": 0.0563, "step": 17512 }, { "epoch": 0.9336780935117556, "grad_norm": 1.5559537410736084, "learning_rate": 6.4976e-06, "loss": 0.0322, "step": 17513 }, { "epoch": 0.9337314069414085, "grad_norm": 1.8984266519546509, "learning_rate": 6.497400000000001e-06, "loss": 0.0277, "step": 17514 }, { "epoch": 0.9337847203710614, "grad_norm": 1.9687496423721313, "learning_rate": 6.4972e-06, "loss": 0.0199, "step": 17515 }, { "epoch": 0.9338380338007144, "grad_norm": 2.3471126556396484, "learning_rate": 6.4970000000000015e-06, "loss": 0.0514, "step": 17516 }, { "epoch": 0.9338913472303674, "grad_norm": 2.0076236724853516, "learning_rate": 6.496800000000001e-06, "loss": 0.0498, "step": 17517 }, { "epoch": 0.9339446606600202, "grad_norm": 1.5597580671310425, "learning_rate": 6.496600000000001e-06, "loss": 0.0121, "step": 17518 }, { "epoch": 0.9339979740896732, "grad_norm": 1.5708364248275757, "learning_rate": 6.4964e-06, "loss": 0.0411, "step": 17519 }, { "epoch": 0.9340512875193261, "grad_norm": 1.8967629671096802, "learning_rate": 6.496200000000001e-06, "loss": 0.029, "step": 17520 }, { "epoch": 0.9341046009489791, "grad_norm": 1.4561073780059814, "learning_rate": 6.496e-06, "loss": 0.0271, "step": 17521 }, { "epoch": 0.9341579143786319, "grad_norm": 2.0701260566711426, "learning_rate": 6.495800000000001e-06, "loss": 0.043, "step": 17522 }, { "epoch": 0.9342112278082849, "grad_norm": 1.1837973594665527, "learning_rate": 6.495600000000001e-06, "loss": 0.0151, "step": 17523 }, { "epoch": 0.9342645412379378, "grad_norm": 1.1087961196899414, "learning_rate": 6.4954000000000005e-06, "loss": 0.0156, "step": 17524 }, { "epoch": 0.9343178546675908, "grad_norm": 1.3769476413726807, "learning_rate": 6.4952e-06, "loss": 0.0225, "step": 17525 }, { "epoch": 0.9343711680972437, "grad_norm": 1.4889299869537354, "learning_rate": 6.4950000000000005e-06, "loss": 0.0323, "step": 17526 }, { "epoch": 0.9344244815268966, "grad_norm": 1.0398480892181396, "learning_rate": 6.4948e-06, "loss": 0.0096, "step": 17527 }, { "epoch": 0.9344777949565496, "grad_norm": 1.356452465057373, "learning_rate": 6.494600000000001e-06, "loss": 0.0161, "step": 17528 }, { "epoch": 0.9345311083862025, "grad_norm": 3.1634607315063477, "learning_rate": 6.494400000000001e-06, "loss": 0.0279, "step": 17529 }, { "epoch": 0.9345844218158554, "grad_norm": 1.1507271528244019, "learning_rate": 6.4942000000000004e-06, "loss": 0.0386, "step": 17530 }, { "epoch": 0.9346377352455083, "grad_norm": 1.759579062461853, "learning_rate": 6.494e-06, "loss": 0.0202, "step": 17531 }, { "epoch": 0.9346910486751613, "grad_norm": 1.1872807741165161, "learning_rate": 6.4938e-06, "loss": 0.0199, "step": 17532 }, { "epoch": 0.9347443621048142, "grad_norm": 0.7171285152435303, "learning_rate": 6.493600000000001e-06, "loss": 0.0088, "step": 17533 }, { "epoch": 0.9347976755344671, "grad_norm": 1.1780705451965332, "learning_rate": 6.493400000000001e-06, "loss": 0.0181, "step": 17534 }, { "epoch": 0.9348509889641201, "grad_norm": 0.8238565921783447, "learning_rate": 6.493200000000001e-06, "loss": 0.0074, "step": 17535 }, { "epoch": 0.934904302393773, "grad_norm": 2.735534191131592, "learning_rate": 6.493e-06, "loss": 0.0295, "step": 17536 }, { "epoch": 0.934957615823426, "grad_norm": 2.766183853149414, "learning_rate": 6.4928e-06, "loss": 0.0231, "step": 17537 }, { "epoch": 0.9350109292530788, "grad_norm": 1.2428805828094482, "learning_rate": 6.4926e-06, "loss": 0.0334, "step": 17538 }, { "epoch": 0.9350642426827318, "grad_norm": 0.9583643674850464, "learning_rate": 6.492400000000001e-06, "loss": 0.0117, "step": 17539 }, { "epoch": 0.9351175561123847, "grad_norm": 2.4381582736968994, "learning_rate": 6.492200000000001e-06, "loss": 0.0319, "step": 17540 }, { "epoch": 0.9351708695420377, "grad_norm": 1.5167455673217773, "learning_rate": 6.492000000000001e-06, "loss": 0.0211, "step": 17541 }, { "epoch": 0.9352241829716905, "grad_norm": 2.639939308166504, "learning_rate": 6.4918e-06, "loss": 0.0169, "step": 17542 }, { "epoch": 0.9352774964013435, "grad_norm": 2.4651777744293213, "learning_rate": 6.4916e-06, "loss": 0.052, "step": 17543 }, { "epoch": 0.9353308098309965, "grad_norm": 1.8208949565887451, "learning_rate": 6.491400000000001e-06, "loss": 0.0307, "step": 17544 }, { "epoch": 0.9353841232606493, "grad_norm": 1.3985971212387085, "learning_rate": 6.491200000000001e-06, "loss": 0.0388, "step": 17545 }, { "epoch": 0.9354374366903023, "grad_norm": 1.9271965026855469, "learning_rate": 6.491000000000001e-06, "loss": 0.0455, "step": 17546 }, { "epoch": 0.9354907501199552, "grad_norm": 1.3226689100265503, "learning_rate": 6.490800000000001e-06, "loss": 0.0254, "step": 17547 }, { "epoch": 0.9355440635496082, "grad_norm": 1.9415415525436401, "learning_rate": 6.4906e-06, "loss": 0.0352, "step": 17548 }, { "epoch": 0.935597376979261, "grad_norm": 1.7388430833816528, "learning_rate": 6.4904e-06, "loss": 0.0265, "step": 17549 }, { "epoch": 0.935650690408914, "grad_norm": 2.6780178546905518, "learning_rate": 6.490200000000001e-06, "loss": 0.0498, "step": 17550 }, { "epoch": 0.9357040038385669, "grad_norm": 1.6646302938461304, "learning_rate": 6.4900000000000005e-06, "loss": 0.0207, "step": 17551 }, { "epoch": 0.9357573172682199, "grad_norm": 1.8298518657684326, "learning_rate": 6.489800000000001e-06, "loss": 0.0177, "step": 17552 }, { "epoch": 0.9358106306978728, "grad_norm": 2.3224713802337646, "learning_rate": 6.4896000000000005e-06, "loss": 0.0308, "step": 17553 }, { "epoch": 0.9358639441275257, "grad_norm": 1.9916564226150513, "learning_rate": 6.4894e-06, "loss": 0.0302, "step": 17554 }, { "epoch": 0.9359172575571787, "grad_norm": 2.1958913803100586, "learning_rate": 6.489200000000001e-06, "loss": 0.0236, "step": 17555 }, { "epoch": 0.9359705709868316, "grad_norm": 1.9986622333526611, "learning_rate": 6.489000000000001e-06, "loss": 0.04, "step": 17556 }, { "epoch": 0.9360238844164845, "grad_norm": 3.7294085025787354, "learning_rate": 6.4888000000000005e-06, "loss": 0.0257, "step": 17557 }, { "epoch": 0.9360771978461374, "grad_norm": 2.6231114864349365, "learning_rate": 6.488600000000001e-06, "loss": 0.0501, "step": 17558 }, { "epoch": 0.9361305112757904, "grad_norm": 2.4751222133636475, "learning_rate": 6.4884e-06, "loss": 0.0263, "step": 17559 }, { "epoch": 0.9361838247054433, "grad_norm": 2.1511313915252686, "learning_rate": 6.4882e-06, "loss": 0.0183, "step": 17560 }, { "epoch": 0.9362371381350962, "grad_norm": 5.723803997039795, "learning_rate": 6.488000000000001e-06, "loss": 0.0624, "step": 17561 }, { "epoch": 0.9362904515647492, "grad_norm": 1.2627040147781372, "learning_rate": 6.487800000000001e-06, "loss": 0.0188, "step": 17562 }, { "epoch": 0.9363437649944021, "grad_norm": 0.8935441970825195, "learning_rate": 6.4876e-06, "loss": 0.0231, "step": 17563 }, { "epoch": 0.9363970784240551, "grad_norm": 2.8685085773468018, "learning_rate": 6.487400000000001e-06, "loss": 0.0364, "step": 17564 }, { "epoch": 0.9364503918537079, "grad_norm": 2.7385647296905518, "learning_rate": 6.4872e-06, "loss": 0.0318, "step": 17565 }, { "epoch": 0.9365037052833609, "grad_norm": 1.7739955186843872, "learning_rate": 6.487000000000001e-06, "loss": 0.0123, "step": 17566 }, { "epoch": 0.9365570187130138, "grad_norm": 2.033163070678711, "learning_rate": 6.486800000000001e-06, "loss": 0.0263, "step": 17567 }, { "epoch": 0.9366103321426668, "grad_norm": 3.8660757541656494, "learning_rate": 6.486600000000001e-06, "loss": 0.0341, "step": 17568 }, { "epoch": 0.9366636455723196, "grad_norm": 2.612124443054199, "learning_rate": 6.4864e-06, "loss": 0.02, "step": 17569 }, { "epoch": 0.9367169590019726, "grad_norm": 1.300088882446289, "learning_rate": 6.486200000000001e-06, "loss": 0.0131, "step": 17570 }, { "epoch": 0.9367702724316255, "grad_norm": 1.4718412160873413, "learning_rate": 6.486e-06, "loss": 0.0152, "step": 17571 }, { "epoch": 0.9368235858612785, "grad_norm": 1.956510305404663, "learning_rate": 6.485800000000001e-06, "loss": 0.0194, "step": 17572 }, { "epoch": 0.9368768992909314, "grad_norm": 3.238528251647949, "learning_rate": 6.485600000000001e-06, "loss": 0.0342, "step": 17573 }, { "epoch": 0.9369302127205843, "grad_norm": 4.392801761627197, "learning_rate": 6.485400000000001e-06, "loss": 0.0463, "step": 17574 }, { "epoch": 0.9369835261502373, "grad_norm": 2.845200777053833, "learning_rate": 6.4852e-06, "loss": 0.0609, "step": 17575 }, { "epoch": 0.9370368395798901, "grad_norm": 3.5628654956817627, "learning_rate": 6.485000000000001e-06, "loss": 0.0449, "step": 17576 }, { "epoch": 0.9370901530095431, "grad_norm": 2.204606294631958, "learning_rate": 6.4848e-06, "loss": 0.0363, "step": 17577 }, { "epoch": 0.937143466439196, "grad_norm": 0.7661311030387878, "learning_rate": 6.4846000000000006e-06, "loss": 0.0092, "step": 17578 }, { "epoch": 0.937196779868849, "grad_norm": 3.4809987545013428, "learning_rate": 6.484400000000001e-06, "loss": 0.0455, "step": 17579 }, { "epoch": 0.9372500932985018, "grad_norm": 4.337690830230713, "learning_rate": 6.4842000000000005e-06, "loss": 0.0249, "step": 17580 }, { "epoch": 0.9373034067281548, "grad_norm": 1.3690494298934937, "learning_rate": 6.484e-06, "loss": 0.0292, "step": 17581 }, { "epoch": 0.9373567201578078, "grad_norm": 2.282397747039795, "learning_rate": 6.4838000000000005e-06, "loss": 0.0274, "step": 17582 }, { "epoch": 0.9374100335874607, "grad_norm": 2.4497203826904297, "learning_rate": 6.483600000000001e-06, "loss": 0.0168, "step": 17583 }, { "epoch": 0.9374633470171136, "grad_norm": 4.895178318023682, "learning_rate": 6.4834000000000005e-06, "loss": 0.0377, "step": 17584 }, { "epoch": 0.9375166604467665, "grad_norm": 2.7357370853424072, "learning_rate": 6.483200000000001e-06, "loss": 0.0481, "step": 17585 }, { "epoch": 0.9375699738764195, "grad_norm": 1.3801937103271484, "learning_rate": 6.4830000000000004e-06, "loss": 0.0165, "step": 17586 }, { "epoch": 0.9376232873060724, "grad_norm": 1.6415609121322632, "learning_rate": 6.4828e-06, "loss": 0.0283, "step": 17587 }, { "epoch": 0.9376766007357253, "grad_norm": 2.269177198410034, "learning_rate": 6.4826e-06, "loss": 0.0376, "step": 17588 }, { "epoch": 0.9377299141653782, "grad_norm": 1.550204873085022, "learning_rate": 6.482400000000001e-06, "loss": 0.026, "step": 17589 }, { "epoch": 0.9377832275950312, "grad_norm": 2.5116426944732666, "learning_rate": 6.4822e-06, "loss": 0.0359, "step": 17590 }, { "epoch": 0.9378365410246842, "grad_norm": 2.210425853729248, "learning_rate": 6.482000000000001e-06, "loss": 0.0316, "step": 17591 }, { "epoch": 0.937889854454337, "grad_norm": 3.5605154037475586, "learning_rate": 6.4818e-06, "loss": 0.0521, "step": 17592 }, { "epoch": 0.93794316788399, "grad_norm": 1.9285823106765747, "learning_rate": 6.4816e-06, "loss": 0.0274, "step": 17593 }, { "epoch": 0.9379964813136429, "grad_norm": 1.5147134065628052, "learning_rate": 6.481400000000001e-06, "loss": 0.0168, "step": 17594 }, { "epoch": 0.9380497947432959, "grad_norm": 1.5815083980560303, "learning_rate": 6.481200000000001e-06, "loss": 0.0234, "step": 17595 }, { "epoch": 0.9381031081729487, "grad_norm": 1.7867026329040527, "learning_rate": 6.481e-06, "loss": 0.0378, "step": 17596 }, { "epoch": 0.9381564216026017, "grad_norm": 1.2709637880325317, "learning_rate": 6.480800000000001e-06, "loss": 0.022, "step": 17597 }, { "epoch": 0.9382097350322546, "grad_norm": 1.737504243850708, "learning_rate": 6.4806e-06, "loss": 0.0216, "step": 17598 }, { "epoch": 0.9382630484619076, "grad_norm": 3.1348421573638916, "learning_rate": 6.4804e-06, "loss": 0.0466, "step": 17599 }, { "epoch": 0.9383163618915605, "grad_norm": 3.110015392303467, "learning_rate": 6.480200000000001e-06, "loss": 0.029, "step": 17600 }, { "epoch": 0.9383696753212134, "grad_norm": 1.187333345413208, "learning_rate": 6.480000000000001e-06, "loss": 0.024, "step": 17601 }, { "epoch": 0.9384229887508664, "grad_norm": 2.9170968532562256, "learning_rate": 6.4798e-06, "loss": 0.0464, "step": 17602 }, { "epoch": 0.9384763021805193, "grad_norm": 2.0364179611206055, "learning_rate": 6.479600000000001e-06, "loss": 0.0249, "step": 17603 }, { "epoch": 0.9385296156101722, "grad_norm": 2.343815565109253, "learning_rate": 6.4794e-06, "loss": 0.0466, "step": 17604 }, { "epoch": 0.9385829290398251, "grad_norm": 2.189185857772827, "learning_rate": 6.4792000000000014e-06, "loss": 0.0504, "step": 17605 }, { "epoch": 0.9386362424694781, "grad_norm": 2.5728952884674072, "learning_rate": 6.479000000000001e-06, "loss": 0.0686, "step": 17606 }, { "epoch": 0.938689555899131, "grad_norm": 1.7281581163406372, "learning_rate": 6.4788000000000005e-06, "loss": 0.0264, "step": 17607 }, { "epoch": 0.9387428693287839, "grad_norm": 1.2603603601455688, "learning_rate": 6.4786e-06, "loss": 0.0187, "step": 17608 }, { "epoch": 0.9387961827584369, "grad_norm": 1.3009755611419678, "learning_rate": 6.4784000000000005e-06, "loss": 0.0207, "step": 17609 }, { "epoch": 0.9388494961880898, "grad_norm": 2.4948976039886475, "learning_rate": 6.4782e-06, "loss": 0.0276, "step": 17610 }, { "epoch": 0.9389028096177428, "grad_norm": 2.2482035160064697, "learning_rate": 6.478000000000001e-06, "loss": 0.0271, "step": 17611 }, { "epoch": 0.9389561230473956, "grad_norm": 1.1892287731170654, "learning_rate": 6.477800000000001e-06, "loss": 0.0145, "step": 17612 }, { "epoch": 0.9390094364770486, "grad_norm": 1.5477826595306396, "learning_rate": 6.4776000000000005e-06, "loss": 0.0322, "step": 17613 }, { "epoch": 0.9390627499067015, "grad_norm": 1.3168798685073853, "learning_rate": 6.4774e-06, "loss": 0.0186, "step": 17614 }, { "epoch": 0.9391160633363544, "grad_norm": 1.295936107635498, "learning_rate": 6.4772000000000004e-06, "loss": 0.0153, "step": 17615 }, { "epoch": 0.9391693767660073, "grad_norm": 1.8361971378326416, "learning_rate": 6.477000000000001e-06, "loss": 0.0514, "step": 17616 }, { "epoch": 0.9392226901956603, "grad_norm": 2.1953773498535156, "learning_rate": 6.476800000000001e-06, "loss": 0.0327, "step": 17617 }, { "epoch": 0.9392760036253133, "grad_norm": 1.6651731729507446, "learning_rate": 6.476600000000001e-06, "loss": 0.0207, "step": 17618 }, { "epoch": 0.9393293170549661, "grad_norm": 1.3603122234344482, "learning_rate": 6.4764e-06, "loss": 0.0173, "step": 17619 }, { "epoch": 0.9393826304846191, "grad_norm": 1.8784844875335693, "learning_rate": 6.4762e-06, "loss": 0.0567, "step": 17620 }, { "epoch": 0.939435943914272, "grad_norm": 1.4096705913543701, "learning_rate": 6.476e-06, "loss": 0.016, "step": 17621 }, { "epoch": 0.939489257343925, "grad_norm": 4.148437976837158, "learning_rate": 6.475800000000001e-06, "loss": 0.0224, "step": 17622 }, { "epoch": 0.9395425707735778, "grad_norm": 2.938570261001587, "learning_rate": 6.475600000000001e-06, "loss": 0.0906, "step": 17623 }, { "epoch": 0.9395958842032308, "grad_norm": 1.7254273891448975, "learning_rate": 6.475400000000001e-06, "loss": 0.0129, "step": 17624 }, { "epoch": 0.9396491976328837, "grad_norm": 1.3107422590255737, "learning_rate": 6.4752e-06, "loss": 0.0397, "step": 17625 }, { "epoch": 0.9397025110625367, "grad_norm": 1.7074705362319946, "learning_rate": 6.475e-06, "loss": 0.0271, "step": 17626 }, { "epoch": 0.9397558244921895, "grad_norm": 1.706583023071289, "learning_rate": 6.474800000000001e-06, "loss": 0.0306, "step": 17627 }, { "epoch": 0.9398091379218425, "grad_norm": 2.5550079345703125, "learning_rate": 6.474600000000001e-06, "loss": 0.0349, "step": 17628 }, { "epoch": 0.9398624513514955, "grad_norm": 2.148378372192383, "learning_rate": 6.474400000000001e-06, "loss": 0.0563, "step": 17629 }, { "epoch": 0.9399157647811484, "grad_norm": 4.010581970214844, "learning_rate": 6.474200000000001e-06, "loss": 0.0437, "step": 17630 }, { "epoch": 0.9399690782108013, "grad_norm": 1.91505765914917, "learning_rate": 6.474e-06, "loss": 0.0154, "step": 17631 }, { "epoch": 0.9400223916404542, "grad_norm": 1.2245619297027588, "learning_rate": 6.4738e-06, "loss": 0.0215, "step": 17632 }, { "epoch": 0.9400757050701072, "grad_norm": 1.010054588317871, "learning_rate": 6.473600000000001e-06, "loss": 0.0121, "step": 17633 }, { "epoch": 0.94012901849976, "grad_norm": 3.266298770904541, "learning_rate": 6.4734000000000006e-06, "loss": 0.0821, "step": 17634 }, { "epoch": 0.940182331929413, "grad_norm": 1.1376575231552124, "learning_rate": 6.473200000000001e-06, "loss": 0.0102, "step": 17635 }, { "epoch": 0.9402356453590659, "grad_norm": 3.050759792327881, "learning_rate": 6.4730000000000005e-06, "loss": 0.0307, "step": 17636 }, { "epoch": 0.9402889587887189, "grad_norm": 1.370851993560791, "learning_rate": 6.4728e-06, "loss": 0.0168, "step": 17637 }, { "epoch": 0.9403422722183719, "grad_norm": 1.3848258256912231, "learning_rate": 6.4726e-06, "loss": 0.0142, "step": 17638 }, { "epoch": 0.9403955856480247, "grad_norm": 1.0885246992111206, "learning_rate": 6.472400000000001e-06, "loss": 0.019, "step": 17639 }, { "epoch": 0.9404488990776777, "grad_norm": 2.4280827045440674, "learning_rate": 6.4722000000000005e-06, "loss": 0.0766, "step": 17640 }, { "epoch": 0.9405022125073306, "grad_norm": 1.859505295753479, "learning_rate": 6.472000000000001e-06, "loss": 0.0509, "step": 17641 }, { "epoch": 0.9405555259369835, "grad_norm": 1.783527135848999, "learning_rate": 6.4718000000000004e-06, "loss": 0.0361, "step": 17642 }, { "epoch": 0.9406088393666364, "grad_norm": 2.2215914726257324, "learning_rate": 6.4716e-06, "loss": 0.0303, "step": 17643 }, { "epoch": 0.9406621527962894, "grad_norm": 2.39108943939209, "learning_rate": 6.471400000000001e-06, "loss": 0.0291, "step": 17644 }, { "epoch": 0.9407154662259423, "grad_norm": 2.1967828273773193, "learning_rate": 6.471200000000001e-06, "loss": 0.0478, "step": 17645 }, { "epoch": 0.9407687796555952, "grad_norm": 0.9400662779808044, "learning_rate": 6.471e-06, "loss": 0.018, "step": 17646 }, { "epoch": 0.9408220930852482, "grad_norm": 0.8193909525871277, "learning_rate": 6.470800000000001e-06, "loss": 0.0084, "step": 17647 }, { "epoch": 0.9408754065149011, "grad_norm": 2.4700510501861572, "learning_rate": 6.4706e-06, "loss": 0.035, "step": 17648 }, { "epoch": 0.9409287199445541, "grad_norm": 1.1624679565429688, "learning_rate": 6.4704e-06, "loss": 0.0171, "step": 17649 }, { "epoch": 0.9409820333742069, "grad_norm": 2.402757406234741, "learning_rate": 6.470200000000001e-06, "loss": 0.033, "step": 17650 }, { "epoch": 0.9410353468038599, "grad_norm": 4.202033042907715, "learning_rate": 6.470000000000001e-06, "loss": 0.0466, "step": 17651 }, { "epoch": 0.9410886602335128, "grad_norm": 2.3576724529266357, "learning_rate": 6.4698e-06, "loss": 0.0238, "step": 17652 }, { "epoch": 0.9411419736631658, "grad_norm": 1.2750041484832764, "learning_rate": 6.469600000000001e-06, "loss": 0.015, "step": 17653 }, { "epoch": 0.9411952870928186, "grad_norm": 2.149742364883423, "learning_rate": 6.4694e-06, "loss": 0.0483, "step": 17654 }, { "epoch": 0.9412486005224716, "grad_norm": 2.692945718765259, "learning_rate": 6.469200000000001e-06, "loss": 0.022, "step": 17655 }, { "epoch": 0.9413019139521246, "grad_norm": 0.578517496585846, "learning_rate": 6.469000000000001e-06, "loss": 0.0078, "step": 17656 }, { "epoch": 0.9413552273817775, "grad_norm": 1.6280685663223267, "learning_rate": 6.468800000000001e-06, "loss": 0.0175, "step": 17657 }, { "epoch": 0.9414085408114304, "grad_norm": 1.5153125524520874, "learning_rate": 6.4686e-06, "loss": 0.0165, "step": 17658 }, { "epoch": 0.9414618542410833, "grad_norm": 2.5764644145965576, "learning_rate": 6.468400000000001e-06, "loss": 0.0494, "step": 17659 }, { "epoch": 0.9415151676707363, "grad_norm": 1.1907509565353394, "learning_rate": 6.4682e-06, "loss": 0.0133, "step": 17660 }, { "epoch": 0.9415684811003892, "grad_norm": 1.6659951210021973, "learning_rate": 6.468000000000001e-06, "loss": 0.0143, "step": 17661 }, { "epoch": 0.9416217945300421, "grad_norm": 1.3634533882141113, "learning_rate": 6.467800000000001e-06, "loss": 0.0241, "step": 17662 }, { "epoch": 0.941675107959695, "grad_norm": 2.231588125228882, "learning_rate": 6.4676000000000006e-06, "loss": 0.0416, "step": 17663 }, { "epoch": 0.941728421389348, "grad_norm": 2.228858232498169, "learning_rate": 6.4674e-06, "loss": 0.0155, "step": 17664 }, { "epoch": 0.941781734819001, "grad_norm": 2.2705187797546387, "learning_rate": 6.4672000000000005e-06, "loss": 0.0277, "step": 17665 }, { "epoch": 0.9418350482486538, "grad_norm": 2.8151254653930664, "learning_rate": 6.467000000000001e-06, "loss": 0.0265, "step": 17666 }, { "epoch": 0.9418883616783068, "grad_norm": 1.9207655191421509, "learning_rate": 6.4668000000000005e-06, "loss": 0.0206, "step": 17667 }, { "epoch": 0.9419416751079597, "grad_norm": 2.288914918899536, "learning_rate": 6.466600000000001e-06, "loss": 0.0624, "step": 17668 }, { "epoch": 0.9419949885376127, "grad_norm": 2.1400647163391113, "learning_rate": 6.4664000000000005e-06, "loss": 0.0319, "step": 17669 }, { "epoch": 0.9420483019672655, "grad_norm": 1.6515820026397705, "learning_rate": 6.4662e-06, "loss": 0.0424, "step": 17670 }, { "epoch": 0.9421016153969185, "grad_norm": 3.241605281829834, "learning_rate": 6.4660000000000004e-06, "loss": 0.0462, "step": 17671 }, { "epoch": 0.9421549288265714, "grad_norm": 1.1630650758743286, "learning_rate": 6.465800000000001e-06, "loss": 0.0115, "step": 17672 }, { "epoch": 0.9422082422562243, "grad_norm": 1.8676462173461914, "learning_rate": 6.4656e-06, "loss": 0.0341, "step": 17673 }, { "epoch": 0.9422615556858773, "grad_norm": 1.3771039247512817, "learning_rate": 6.465400000000001e-06, "loss": 0.02, "step": 17674 }, { "epoch": 0.9423148691155302, "grad_norm": 2.0693249702453613, "learning_rate": 6.4652e-06, "loss": 0.0255, "step": 17675 }, { "epoch": 0.9423681825451832, "grad_norm": 1.8304425477981567, "learning_rate": 6.465e-06, "loss": 0.019, "step": 17676 }, { "epoch": 0.942421495974836, "grad_norm": 1.9462724924087524, "learning_rate": 6.464800000000001e-06, "loss": 0.0248, "step": 17677 }, { "epoch": 0.942474809404489, "grad_norm": 4.8201494216918945, "learning_rate": 6.464600000000001e-06, "loss": 0.0819, "step": 17678 }, { "epoch": 0.9425281228341419, "grad_norm": 3.100991725921631, "learning_rate": 6.4644e-06, "loss": 0.0414, "step": 17679 }, { "epoch": 0.9425814362637949, "grad_norm": 3.771123170852661, "learning_rate": 6.464200000000001e-06, "loss": 0.0479, "step": 17680 }, { "epoch": 0.9426347496934477, "grad_norm": 1.5879372358322144, "learning_rate": 6.464e-06, "loss": 0.0398, "step": 17681 }, { "epoch": 0.9426880631231007, "grad_norm": 1.3426268100738525, "learning_rate": 6.4638e-06, "loss": 0.0267, "step": 17682 }, { "epoch": 0.9427413765527536, "grad_norm": 0.7131629586219788, "learning_rate": 6.463600000000001e-06, "loss": 0.0065, "step": 17683 }, { "epoch": 0.9427946899824066, "grad_norm": 3.2805328369140625, "learning_rate": 6.463400000000001e-06, "loss": 0.0304, "step": 17684 }, { "epoch": 0.9428480034120595, "grad_norm": 1.8240692615509033, "learning_rate": 6.4632e-06, "loss": 0.0487, "step": 17685 }, { "epoch": 0.9429013168417124, "grad_norm": 1.5115044116973877, "learning_rate": 6.463000000000001e-06, "loss": 0.0243, "step": 17686 }, { "epoch": 0.9429546302713654, "grad_norm": 1.4287419319152832, "learning_rate": 6.4628e-06, "loss": 0.015, "step": 17687 }, { "epoch": 0.9430079437010183, "grad_norm": 1.7956781387329102, "learning_rate": 6.4626e-06, "loss": 0.0383, "step": 17688 }, { "epoch": 0.9430612571306712, "grad_norm": 0.7166245579719543, "learning_rate": 6.462400000000001e-06, "loss": 0.0089, "step": 17689 }, { "epoch": 0.9431145705603241, "grad_norm": 2.4163155555725098, "learning_rate": 6.4622000000000006e-06, "loss": 0.0424, "step": 17690 }, { "epoch": 0.9431678839899771, "grad_norm": 1.254883050918579, "learning_rate": 6.462e-06, "loss": 0.028, "step": 17691 }, { "epoch": 0.94322119741963, "grad_norm": 3.9339895248413086, "learning_rate": 6.4618000000000005e-06, "loss": 0.0219, "step": 17692 }, { "epoch": 0.9432745108492829, "grad_norm": 1.5133386850357056, "learning_rate": 6.4616e-06, "loss": 0.0129, "step": 17693 }, { "epoch": 0.9433278242789359, "grad_norm": 2.601531982421875, "learning_rate": 6.461400000000001e-06, "loss": 0.0645, "step": 17694 }, { "epoch": 0.9433811377085888, "grad_norm": 2.24497652053833, "learning_rate": 6.461200000000001e-06, "loss": 0.0217, "step": 17695 }, { "epoch": 0.9434344511382418, "grad_norm": 0.9760181903839111, "learning_rate": 6.4610000000000005e-06, "loss": 0.007, "step": 17696 }, { "epoch": 0.9434877645678946, "grad_norm": 2.3420331478118896, "learning_rate": 6.4608e-06, "loss": 0.0428, "step": 17697 }, { "epoch": 0.9435410779975476, "grad_norm": 1.0536980628967285, "learning_rate": 6.4606000000000005e-06, "loss": 0.0131, "step": 17698 }, { "epoch": 0.9435943914272005, "grad_norm": 1.78372061252594, "learning_rate": 6.4604e-06, "loss": 0.0237, "step": 17699 }, { "epoch": 0.9436477048568535, "grad_norm": 1.897887110710144, "learning_rate": 6.460200000000001e-06, "loss": 0.0614, "step": 17700 }, { "epoch": 0.9437010182865063, "grad_norm": 0.7487454414367676, "learning_rate": 6.460000000000001e-06, "loss": 0.0086, "step": 17701 }, { "epoch": 0.9437543317161593, "grad_norm": 1.141902208328247, "learning_rate": 6.4598e-06, "loss": 0.0242, "step": 17702 }, { "epoch": 0.9438076451458123, "grad_norm": 1.7664620876312256, "learning_rate": 6.4596e-06, "loss": 0.0444, "step": 17703 }, { "epoch": 0.9438609585754651, "grad_norm": 1.2032644748687744, "learning_rate": 6.4594e-06, "loss": 0.0302, "step": 17704 }, { "epoch": 0.9439142720051181, "grad_norm": 3.2173619270324707, "learning_rate": 6.459200000000001e-06, "loss": 0.0377, "step": 17705 }, { "epoch": 0.943967585434771, "grad_norm": 1.8569557666778564, "learning_rate": 6.459000000000001e-06, "loss": 0.0141, "step": 17706 }, { "epoch": 0.944020898864424, "grad_norm": 1.5471290349960327, "learning_rate": 6.458800000000001e-06, "loss": 0.0269, "step": 17707 }, { "epoch": 0.9440742122940768, "grad_norm": 1.3136727809906006, "learning_rate": 6.4586e-06, "loss": 0.0111, "step": 17708 }, { "epoch": 0.9441275257237298, "grad_norm": 1.6597256660461426, "learning_rate": 6.4584e-06, "loss": 0.015, "step": 17709 }, { "epoch": 0.9441808391533827, "grad_norm": 1.2000929117202759, "learning_rate": 6.4582e-06, "loss": 0.0153, "step": 17710 }, { "epoch": 0.9442341525830357, "grad_norm": 2.175826072692871, "learning_rate": 6.458000000000001e-06, "loss": 0.0143, "step": 17711 }, { "epoch": 0.9442874660126886, "grad_norm": 1.221277117729187, "learning_rate": 6.457800000000001e-06, "loss": 0.0129, "step": 17712 }, { "epoch": 0.9443407794423415, "grad_norm": 1.5691829919815063, "learning_rate": 6.457600000000001e-06, "loss": 0.0164, "step": 17713 }, { "epoch": 0.9443940928719945, "grad_norm": 3.3368422985076904, "learning_rate": 6.4574e-06, "loss": 0.0254, "step": 17714 }, { "epoch": 0.9444474063016474, "grad_norm": 1.8352159261703491, "learning_rate": 6.4572e-06, "loss": 0.0256, "step": 17715 }, { "epoch": 0.9445007197313003, "grad_norm": 2.243114709854126, "learning_rate": 6.457000000000001e-06, "loss": 0.0331, "step": 17716 }, { "epoch": 0.9445540331609532, "grad_norm": 1.6767431497573853, "learning_rate": 6.456800000000001e-06, "loss": 0.0233, "step": 17717 }, { "epoch": 0.9446073465906062, "grad_norm": 1.4918395280838013, "learning_rate": 6.456600000000001e-06, "loss": 0.0363, "step": 17718 }, { "epoch": 0.9446606600202591, "grad_norm": 2.4466819763183594, "learning_rate": 6.4564000000000006e-06, "loss": 0.0283, "step": 17719 }, { "epoch": 0.944713973449912, "grad_norm": 1.440765142440796, "learning_rate": 6.4562e-06, "loss": 0.0345, "step": 17720 }, { "epoch": 0.944767286879565, "grad_norm": 1.5034289360046387, "learning_rate": 6.456e-06, "loss": 0.0446, "step": 17721 }, { "epoch": 0.9448206003092179, "grad_norm": 1.1595191955566406, "learning_rate": 6.455800000000001e-06, "loss": 0.0087, "step": 17722 }, { "epoch": 0.9448739137388709, "grad_norm": 2.8085806369781494, "learning_rate": 6.4556000000000005e-06, "loss": 0.0377, "step": 17723 }, { "epoch": 0.9449272271685237, "grad_norm": 1.8550972938537598, "learning_rate": 6.455400000000001e-06, "loss": 0.0477, "step": 17724 }, { "epoch": 0.9449805405981767, "grad_norm": 3.0235278606414795, "learning_rate": 6.4552000000000005e-06, "loss": 0.0484, "step": 17725 }, { "epoch": 0.9450338540278296, "grad_norm": 2.5541672706604004, "learning_rate": 6.455e-06, "loss": 0.0275, "step": 17726 }, { "epoch": 0.9450871674574826, "grad_norm": 1.417556881904602, "learning_rate": 6.454800000000001e-06, "loss": 0.0456, "step": 17727 }, { "epoch": 0.9451404808871354, "grad_norm": 1.2056403160095215, "learning_rate": 6.454600000000001e-06, "loss": 0.0171, "step": 17728 }, { "epoch": 0.9451937943167884, "grad_norm": 1.5442866086959839, "learning_rate": 6.4544e-06, "loss": 0.0388, "step": 17729 }, { "epoch": 0.9452471077464414, "grad_norm": 1.5145729780197144, "learning_rate": 6.454200000000001e-06, "loss": 0.0122, "step": 17730 }, { "epoch": 0.9453004211760943, "grad_norm": 1.7545384168624878, "learning_rate": 6.454e-06, "loss": 0.0143, "step": 17731 }, { "epoch": 0.9453537346057472, "grad_norm": 4.119309425354004, "learning_rate": 6.4538e-06, "loss": 0.0386, "step": 17732 }, { "epoch": 0.9454070480354001, "grad_norm": 1.7922977209091187, "learning_rate": 6.453600000000001e-06, "loss": 0.0199, "step": 17733 }, { "epoch": 0.9454603614650531, "grad_norm": 1.9567705392837524, "learning_rate": 6.453400000000001e-06, "loss": 0.0237, "step": 17734 }, { "epoch": 0.945513674894706, "grad_norm": 0.541675865650177, "learning_rate": 6.4532e-06, "loss": 0.0048, "step": 17735 }, { "epoch": 0.9455669883243589, "grad_norm": 1.8761601448059082, "learning_rate": 6.453000000000001e-06, "loss": 0.018, "step": 17736 }, { "epoch": 0.9456203017540118, "grad_norm": 2.213733673095703, "learning_rate": 6.4528e-06, "loss": 0.0412, "step": 17737 }, { "epoch": 0.9456736151836648, "grad_norm": 1.164796233177185, "learning_rate": 6.4526e-06, "loss": 0.0158, "step": 17738 }, { "epoch": 0.9457269286133176, "grad_norm": 2.223714828491211, "learning_rate": 6.452400000000001e-06, "loss": 0.0337, "step": 17739 }, { "epoch": 0.9457802420429706, "grad_norm": 2.027639150619507, "learning_rate": 6.452200000000001e-06, "loss": 0.0088, "step": 17740 }, { "epoch": 0.9458335554726236, "grad_norm": 2.053649425506592, "learning_rate": 6.452e-06, "loss": 0.0473, "step": 17741 }, { "epoch": 0.9458868689022765, "grad_norm": 1.5135366916656494, "learning_rate": 6.451800000000001e-06, "loss": 0.032, "step": 17742 }, { "epoch": 0.9459401823319294, "grad_norm": 3.252758741378784, "learning_rate": 6.4516e-06, "loss": 0.0347, "step": 17743 }, { "epoch": 0.9459934957615823, "grad_norm": 1.9662245512008667, "learning_rate": 6.451400000000001e-06, "loss": 0.0332, "step": 17744 }, { "epoch": 0.9460468091912353, "grad_norm": 2.4517099857330322, "learning_rate": 6.451200000000001e-06, "loss": 0.0313, "step": 17745 }, { "epoch": 0.9461001226208882, "grad_norm": 1.7983438968658447, "learning_rate": 6.451000000000001e-06, "loss": 0.0179, "step": 17746 }, { "epoch": 0.9461534360505411, "grad_norm": 1.6561864614486694, "learning_rate": 6.4508e-06, "loss": 0.0279, "step": 17747 }, { "epoch": 0.946206749480194, "grad_norm": 2.1165974140167236, "learning_rate": 6.4506000000000005e-06, "loss": 0.05, "step": 17748 }, { "epoch": 0.946260062909847, "grad_norm": 2.529460906982422, "learning_rate": 6.4504e-06, "loss": 0.0335, "step": 17749 }, { "epoch": 0.9463133763395, "grad_norm": 4.435429096221924, "learning_rate": 6.4502000000000005e-06, "loss": 0.0498, "step": 17750 }, { "epoch": 0.9463666897691528, "grad_norm": 2.6557769775390625, "learning_rate": 6.450000000000001e-06, "loss": 0.0231, "step": 17751 }, { "epoch": 0.9464200031988058, "grad_norm": 4.280821800231934, "learning_rate": 6.4498000000000005e-06, "loss": 0.0666, "step": 17752 }, { "epoch": 0.9464733166284587, "grad_norm": 3.6367993354797363, "learning_rate": 6.4496e-06, "loss": 0.0334, "step": 17753 }, { "epoch": 0.9465266300581117, "grad_norm": 1.1714165210723877, "learning_rate": 6.4494000000000005e-06, "loss": 0.0168, "step": 17754 }, { "epoch": 0.9465799434877645, "grad_norm": 1.687794804573059, "learning_rate": 6.449200000000001e-06, "loss": 0.0132, "step": 17755 }, { "epoch": 0.9466332569174175, "grad_norm": 1.9491583108901978, "learning_rate": 6.4490000000000004e-06, "loss": 0.0411, "step": 17756 }, { "epoch": 0.9466865703470704, "grad_norm": 2.6754326820373535, "learning_rate": 6.448800000000001e-06, "loss": 0.0492, "step": 17757 }, { "epoch": 0.9467398837767234, "grad_norm": 2.550067901611328, "learning_rate": 6.4486e-06, "loss": 0.0296, "step": 17758 }, { "epoch": 0.9467931972063763, "grad_norm": 1.5178202390670776, "learning_rate": 6.4484e-06, "loss": 0.0229, "step": 17759 }, { "epoch": 0.9468465106360292, "grad_norm": 2.9180049896240234, "learning_rate": 6.4482e-06, "loss": 0.028, "step": 17760 }, { "epoch": 0.9468998240656822, "grad_norm": 2.3199098110198975, "learning_rate": 6.448000000000001e-06, "loss": 0.0517, "step": 17761 }, { "epoch": 0.946953137495335, "grad_norm": 0.9528489112854004, "learning_rate": 6.4478e-06, "loss": 0.0117, "step": 17762 }, { "epoch": 0.947006450924988, "grad_norm": 1.3505680561065674, "learning_rate": 6.447600000000001e-06, "loss": 0.0164, "step": 17763 }, { "epoch": 0.9470597643546409, "grad_norm": 2.0163121223449707, "learning_rate": 6.4474e-06, "loss": 0.0291, "step": 17764 }, { "epoch": 0.9471130777842939, "grad_norm": 1.7457525730133057, "learning_rate": 6.4472e-06, "loss": 0.0326, "step": 17765 }, { "epoch": 0.9471663912139467, "grad_norm": 2.0569303035736084, "learning_rate": 6.447000000000001e-06, "loss": 0.0368, "step": 17766 }, { "epoch": 0.9472197046435997, "grad_norm": 3.1917154788970947, "learning_rate": 6.446800000000001e-06, "loss": 0.0536, "step": 17767 }, { "epoch": 0.9472730180732527, "grad_norm": 2.04532790184021, "learning_rate": 6.4466e-06, "loss": 0.0659, "step": 17768 }, { "epoch": 0.9473263315029056, "grad_norm": 1.42659592628479, "learning_rate": 6.446400000000001e-06, "loss": 0.0264, "step": 17769 }, { "epoch": 0.9473796449325586, "grad_norm": 2.3580634593963623, "learning_rate": 6.4462e-06, "loss": 0.019, "step": 17770 }, { "epoch": 0.9474329583622114, "grad_norm": 1.7830873727798462, "learning_rate": 6.446e-06, "loss": 0.0232, "step": 17771 }, { "epoch": 0.9474862717918644, "grad_norm": 4.096069812774658, "learning_rate": 6.445800000000001e-06, "loss": 0.0279, "step": 17772 }, { "epoch": 0.9475395852215173, "grad_norm": 1.9546064138412476, "learning_rate": 6.445600000000001e-06, "loss": 0.037, "step": 17773 }, { "epoch": 0.9475928986511702, "grad_norm": 1.3167701959609985, "learning_rate": 6.4454e-06, "loss": 0.0254, "step": 17774 }, { "epoch": 0.9476462120808231, "grad_norm": 2.5962491035461426, "learning_rate": 6.4452000000000006e-06, "loss": 0.046, "step": 17775 }, { "epoch": 0.9476995255104761, "grad_norm": 2.2918028831481934, "learning_rate": 6.445e-06, "loss": 0.0551, "step": 17776 }, { "epoch": 0.9477528389401291, "grad_norm": 0.9450276494026184, "learning_rate": 6.444800000000001e-06, "loss": 0.0188, "step": 17777 }, { "epoch": 0.9478061523697819, "grad_norm": 2.2218549251556396, "learning_rate": 6.444600000000001e-06, "loss": 0.0318, "step": 17778 }, { "epoch": 0.9478594657994349, "grad_norm": 2.295121192932129, "learning_rate": 6.4444000000000005e-06, "loss": 0.0274, "step": 17779 }, { "epoch": 0.9479127792290878, "grad_norm": 1.899329423904419, "learning_rate": 6.4442e-06, "loss": 0.0357, "step": 17780 }, { "epoch": 0.9479660926587408, "grad_norm": 2.352961540222168, "learning_rate": 6.4440000000000005e-06, "loss": 0.025, "step": 17781 }, { "epoch": 0.9480194060883936, "grad_norm": 2.0030453205108643, "learning_rate": 6.4438e-06, "loss": 0.0407, "step": 17782 }, { "epoch": 0.9480727195180466, "grad_norm": 1.2300137281417847, "learning_rate": 6.443600000000001e-06, "loss": 0.0306, "step": 17783 }, { "epoch": 0.9481260329476995, "grad_norm": 0.8065837025642395, "learning_rate": 6.443400000000001e-06, "loss": 0.0086, "step": 17784 }, { "epoch": 0.9481793463773525, "grad_norm": 1.6018502712249756, "learning_rate": 6.4432e-06, "loss": 0.0272, "step": 17785 }, { "epoch": 0.9482326598070053, "grad_norm": 3.1599066257476807, "learning_rate": 6.443e-06, "loss": 0.0483, "step": 17786 }, { "epoch": 0.9482859732366583, "grad_norm": 1.8818265199661255, "learning_rate": 6.4428e-06, "loss": 0.0302, "step": 17787 }, { "epoch": 0.9483392866663113, "grad_norm": 1.9536978006362915, "learning_rate": 6.442600000000001e-06, "loss": 0.0225, "step": 17788 }, { "epoch": 0.9483926000959642, "grad_norm": 1.4008651971817017, "learning_rate": 6.442400000000001e-06, "loss": 0.0232, "step": 17789 }, { "epoch": 0.9484459135256171, "grad_norm": 1.4813976287841797, "learning_rate": 6.442200000000001e-06, "loss": 0.0125, "step": 17790 }, { "epoch": 0.94849922695527, "grad_norm": 2.1734039783477783, "learning_rate": 6.442e-06, "loss": 0.0289, "step": 17791 }, { "epoch": 0.948552540384923, "grad_norm": 2.347121238708496, "learning_rate": 6.4418e-06, "loss": 0.0286, "step": 17792 }, { "epoch": 0.9486058538145759, "grad_norm": 2.0108840465545654, "learning_rate": 6.4416e-06, "loss": 0.0088, "step": 17793 }, { "epoch": 0.9486591672442288, "grad_norm": 1.2071532011032104, "learning_rate": 6.441400000000001e-06, "loss": 0.0132, "step": 17794 }, { "epoch": 0.9487124806738817, "grad_norm": 1.535765528678894, "learning_rate": 6.441200000000001e-06, "loss": 0.0235, "step": 17795 }, { "epoch": 0.9487657941035347, "grad_norm": 2.171229362487793, "learning_rate": 6.441000000000001e-06, "loss": 0.0241, "step": 17796 }, { "epoch": 0.9488191075331877, "grad_norm": 2.4825687408447266, "learning_rate": 6.4408e-06, "loss": 0.0436, "step": 17797 }, { "epoch": 0.9488724209628405, "grad_norm": 1.356689453125, "learning_rate": 6.4406e-06, "loss": 0.0188, "step": 17798 }, { "epoch": 0.9489257343924935, "grad_norm": 1.7029526233673096, "learning_rate": 6.4404e-06, "loss": 0.0202, "step": 17799 }, { "epoch": 0.9489790478221464, "grad_norm": 1.6324375867843628, "learning_rate": 6.440200000000001e-06, "loss": 0.0289, "step": 17800 }, { "epoch": 0.9490323612517994, "grad_norm": 3.3669254779815674, "learning_rate": 6.440000000000001e-06, "loss": 0.0424, "step": 17801 }, { "epoch": 0.9490856746814522, "grad_norm": 1.9938815832138062, "learning_rate": 6.439800000000001e-06, "loss": 0.0612, "step": 17802 }, { "epoch": 0.9491389881111052, "grad_norm": 1.18634033203125, "learning_rate": 6.4396e-06, "loss": 0.0215, "step": 17803 }, { "epoch": 0.9491923015407581, "grad_norm": 2.101533889770508, "learning_rate": 6.4394e-06, "loss": 0.0198, "step": 17804 }, { "epoch": 0.949245614970411, "grad_norm": 1.628420352935791, "learning_rate": 6.439200000000001e-06, "loss": 0.0126, "step": 17805 }, { "epoch": 0.949298928400064, "grad_norm": 2.5110127925872803, "learning_rate": 6.4390000000000005e-06, "loss": 0.0264, "step": 17806 }, { "epoch": 0.9493522418297169, "grad_norm": 1.8481338024139404, "learning_rate": 6.438800000000001e-06, "loss": 0.032, "step": 17807 }, { "epoch": 0.9494055552593699, "grad_norm": 2.4418392181396484, "learning_rate": 6.4386000000000005e-06, "loss": 0.0458, "step": 17808 }, { "epoch": 0.9494588686890227, "grad_norm": 1.1559265851974487, "learning_rate": 6.4384e-06, "loss": 0.0103, "step": 17809 }, { "epoch": 0.9495121821186757, "grad_norm": 2.15588641166687, "learning_rate": 6.4382e-06, "loss": 0.0517, "step": 17810 }, { "epoch": 0.9495654955483286, "grad_norm": 2.856407642364502, "learning_rate": 6.438000000000001e-06, "loss": 0.0477, "step": 17811 }, { "epoch": 0.9496188089779816, "grad_norm": 1.4789601564407349, "learning_rate": 6.4378000000000004e-06, "loss": 0.0167, "step": 17812 }, { "epoch": 0.9496721224076344, "grad_norm": 2.5532166957855225, "learning_rate": 6.437600000000001e-06, "loss": 0.0524, "step": 17813 }, { "epoch": 0.9497254358372874, "grad_norm": 1.1048020124435425, "learning_rate": 6.4374e-06, "loss": 0.0119, "step": 17814 }, { "epoch": 0.9497787492669404, "grad_norm": 3.5869603157043457, "learning_rate": 6.4372e-06, "loss": 0.0157, "step": 17815 }, { "epoch": 0.9498320626965933, "grad_norm": 4.072320461273193, "learning_rate": 6.437000000000001e-06, "loss": 0.0262, "step": 17816 }, { "epoch": 0.9498853761262462, "grad_norm": 1.2553918361663818, "learning_rate": 6.436800000000001e-06, "loss": 0.0301, "step": 17817 }, { "epoch": 0.9499386895558991, "grad_norm": 1.7673251628875732, "learning_rate": 6.4366e-06, "loss": 0.0402, "step": 17818 }, { "epoch": 0.9499920029855521, "grad_norm": 1.5757441520690918, "learning_rate": 6.436400000000001e-06, "loss": 0.0215, "step": 17819 }, { "epoch": 0.950045316415205, "grad_norm": 2.736692428588867, "learning_rate": 6.4362e-06, "loss": 0.0257, "step": 17820 }, { "epoch": 0.9500986298448579, "grad_norm": 1.5148378610610962, "learning_rate": 6.436e-06, "loss": 0.0253, "step": 17821 }, { "epoch": 0.9501519432745108, "grad_norm": 1.4797096252441406, "learning_rate": 6.435800000000001e-06, "loss": 0.0406, "step": 17822 }, { "epoch": 0.9502052567041638, "grad_norm": 1.7221572399139404, "learning_rate": 6.435600000000001e-06, "loss": 0.0237, "step": 17823 }, { "epoch": 0.9502585701338168, "grad_norm": 1.1631664037704468, "learning_rate": 6.4354e-06, "loss": 0.027, "step": 17824 }, { "epoch": 0.9503118835634696, "grad_norm": 2.383634090423584, "learning_rate": 6.435200000000001e-06, "loss": 0.0574, "step": 17825 }, { "epoch": 0.9503651969931226, "grad_norm": 0.910172700881958, "learning_rate": 6.435e-06, "loss": 0.0106, "step": 17826 }, { "epoch": 0.9504185104227755, "grad_norm": 1.5473734140396118, "learning_rate": 6.434800000000001e-06, "loss": 0.0228, "step": 17827 }, { "epoch": 0.9504718238524285, "grad_norm": 0.885561466217041, "learning_rate": 6.434600000000001e-06, "loss": 0.0123, "step": 17828 }, { "epoch": 0.9505251372820813, "grad_norm": 1.0077733993530273, "learning_rate": 6.434400000000001e-06, "loss": 0.0247, "step": 17829 }, { "epoch": 0.9505784507117343, "grad_norm": 1.933954119682312, "learning_rate": 6.4342e-06, "loss": 0.0275, "step": 17830 }, { "epoch": 0.9506317641413872, "grad_norm": 1.865342140197754, "learning_rate": 6.4340000000000006e-06, "loss": 0.0343, "step": 17831 }, { "epoch": 0.9506850775710401, "grad_norm": 0.9790434241294861, "learning_rate": 6.4338e-06, "loss": 0.009, "step": 17832 }, { "epoch": 0.9507383910006931, "grad_norm": 1.6885126829147339, "learning_rate": 6.4336000000000005e-06, "loss": 0.0129, "step": 17833 }, { "epoch": 0.950791704430346, "grad_norm": 2.174666404724121, "learning_rate": 6.433400000000001e-06, "loss": 0.031, "step": 17834 }, { "epoch": 0.950845017859999, "grad_norm": 2.4659340381622314, "learning_rate": 6.4332000000000005e-06, "loss": 0.0404, "step": 17835 }, { "epoch": 0.9508983312896518, "grad_norm": 2.305499792098999, "learning_rate": 6.433e-06, "loss": 0.0204, "step": 17836 }, { "epoch": 0.9509516447193048, "grad_norm": 2.107942581176758, "learning_rate": 6.4328000000000005e-06, "loss": 0.0355, "step": 17837 }, { "epoch": 0.9510049581489577, "grad_norm": 0.6541036367416382, "learning_rate": 6.432600000000001e-06, "loss": 0.0065, "step": 17838 }, { "epoch": 0.9510582715786107, "grad_norm": 1.4205238819122314, "learning_rate": 6.4324000000000005e-06, "loss": 0.0342, "step": 17839 }, { "epoch": 0.9511115850082635, "grad_norm": 1.8422508239746094, "learning_rate": 6.432200000000001e-06, "loss": 0.0282, "step": 17840 }, { "epoch": 0.9511648984379165, "grad_norm": 1.4126759767532349, "learning_rate": 6.432e-06, "loss": 0.0237, "step": 17841 }, { "epoch": 0.9512182118675694, "grad_norm": 1.454419493675232, "learning_rate": 6.4318e-06, "loss": 0.0163, "step": 17842 }, { "epoch": 0.9512715252972224, "grad_norm": 1.3724184036254883, "learning_rate": 6.4316e-06, "loss": 0.0305, "step": 17843 }, { "epoch": 0.9513248387268753, "grad_norm": 2.4427874088287354, "learning_rate": 6.431400000000001e-06, "loss": 0.0273, "step": 17844 }, { "epoch": 0.9513781521565282, "grad_norm": 1.4409881830215454, "learning_rate": 6.4312e-06, "loss": 0.01, "step": 17845 }, { "epoch": 0.9514314655861812, "grad_norm": 0.6424375176429749, "learning_rate": 6.431000000000001e-06, "loss": 0.0164, "step": 17846 }, { "epoch": 0.9514847790158341, "grad_norm": 1.655158281326294, "learning_rate": 6.4308e-06, "loss": 0.03, "step": 17847 }, { "epoch": 0.951538092445487, "grad_norm": 0.8549035787582397, "learning_rate": 6.4306e-06, "loss": 0.0134, "step": 17848 }, { "epoch": 0.9515914058751399, "grad_norm": 2.399620294570923, "learning_rate": 6.4304e-06, "loss": 0.0369, "step": 17849 }, { "epoch": 0.9516447193047929, "grad_norm": 1.6593869924545288, "learning_rate": 6.430200000000001e-06, "loss": 0.0172, "step": 17850 }, { "epoch": 0.9516980327344458, "grad_norm": 1.18641996383667, "learning_rate": 6.43e-06, "loss": 0.0213, "step": 17851 }, { "epoch": 0.9517513461640987, "grad_norm": 2.160149335861206, "learning_rate": 6.429800000000001e-06, "loss": 0.0479, "step": 17852 }, { "epoch": 0.9518046595937517, "grad_norm": 3.8406803607940674, "learning_rate": 6.4296e-06, "loss": 0.0299, "step": 17853 }, { "epoch": 0.9518579730234046, "grad_norm": 2.701667070388794, "learning_rate": 6.4294e-06, "loss": 0.0254, "step": 17854 }, { "epoch": 0.9519112864530576, "grad_norm": 1.4218333959579468, "learning_rate": 6.429200000000001e-06, "loss": 0.0149, "step": 17855 }, { "epoch": 0.9519645998827104, "grad_norm": 1.8968825340270996, "learning_rate": 6.429000000000001e-06, "loss": 0.0414, "step": 17856 }, { "epoch": 0.9520179133123634, "grad_norm": 2.8027379512786865, "learning_rate": 6.4288e-06, "loss": 0.0461, "step": 17857 }, { "epoch": 0.9520712267420163, "grad_norm": 3.0970394611358643, "learning_rate": 6.428600000000001e-06, "loss": 0.0203, "step": 17858 }, { "epoch": 0.9521245401716693, "grad_norm": 1.4382373094558716, "learning_rate": 6.4284e-06, "loss": 0.0273, "step": 17859 }, { "epoch": 0.9521778536013221, "grad_norm": 1.8677006959915161, "learning_rate": 6.4282e-06, "loss": 0.0182, "step": 17860 }, { "epoch": 0.9522311670309751, "grad_norm": 1.6693830490112305, "learning_rate": 6.428000000000001e-06, "loss": 0.0167, "step": 17861 }, { "epoch": 0.9522844804606281, "grad_norm": 2.693586587905884, "learning_rate": 6.4278000000000005e-06, "loss": 0.0637, "step": 17862 }, { "epoch": 0.952337793890281, "grad_norm": 2.6464266777038574, "learning_rate": 6.4276e-06, "loss": 0.0617, "step": 17863 }, { "epoch": 0.9523911073199339, "grad_norm": 1.9460524320602417, "learning_rate": 6.4274000000000005e-06, "loss": 0.0465, "step": 17864 }, { "epoch": 0.9524444207495868, "grad_norm": 1.6435412168502808, "learning_rate": 6.4272e-06, "loss": 0.0388, "step": 17865 }, { "epoch": 0.9524977341792398, "grad_norm": 1.4208213090896606, "learning_rate": 6.427000000000001e-06, "loss": 0.0334, "step": 17866 }, { "epoch": 0.9525510476088926, "grad_norm": 2.8657798767089844, "learning_rate": 6.426800000000001e-06, "loss": 0.043, "step": 17867 }, { "epoch": 0.9526043610385456, "grad_norm": 2.518144369125366, "learning_rate": 6.4266000000000004e-06, "loss": 0.0244, "step": 17868 }, { "epoch": 0.9526576744681985, "grad_norm": 2.825941801071167, "learning_rate": 6.4264e-06, "loss": 0.0321, "step": 17869 }, { "epoch": 0.9527109878978515, "grad_norm": 1.4114441871643066, "learning_rate": 6.4262e-06, "loss": 0.0193, "step": 17870 }, { "epoch": 0.9527643013275044, "grad_norm": 1.2096792459487915, "learning_rate": 6.426e-06, "loss": 0.0123, "step": 17871 }, { "epoch": 0.9528176147571573, "grad_norm": 1.56659996509552, "learning_rate": 6.425800000000001e-06, "loss": 0.0427, "step": 17872 }, { "epoch": 0.9528709281868103, "grad_norm": 1.4605317115783691, "learning_rate": 6.425600000000001e-06, "loss": 0.0349, "step": 17873 }, { "epoch": 0.9529242416164632, "grad_norm": 1.057679295539856, "learning_rate": 6.4254e-06, "loss": 0.0149, "step": 17874 }, { "epoch": 0.9529775550461161, "grad_norm": 2.0721137523651123, "learning_rate": 6.4252e-06, "loss": 0.0387, "step": 17875 }, { "epoch": 0.953030868475769, "grad_norm": 1.9289014339447021, "learning_rate": 6.425e-06, "loss": 0.044, "step": 17876 }, { "epoch": 0.953084181905422, "grad_norm": 1.4185928106307983, "learning_rate": 6.424800000000001e-06, "loss": 0.0125, "step": 17877 }, { "epoch": 0.9531374953350749, "grad_norm": 1.4222888946533203, "learning_rate": 6.424600000000001e-06, "loss": 0.0298, "step": 17878 }, { "epoch": 0.9531908087647278, "grad_norm": 2.019508123397827, "learning_rate": 6.424400000000001e-06, "loss": 0.0369, "step": 17879 }, { "epoch": 0.9532441221943808, "grad_norm": 1.2213866710662842, "learning_rate": 6.4242e-06, "loss": 0.0226, "step": 17880 }, { "epoch": 0.9532974356240337, "grad_norm": 1.6579481363296509, "learning_rate": 6.424e-06, "loss": 0.0311, "step": 17881 }, { "epoch": 0.9533507490536867, "grad_norm": 1.705112338066101, "learning_rate": 6.4238e-06, "loss": 0.0281, "step": 17882 }, { "epoch": 0.9534040624833395, "grad_norm": 1.631248950958252, "learning_rate": 6.423600000000001e-06, "loss": 0.0259, "step": 17883 }, { "epoch": 0.9534573759129925, "grad_norm": 0.7143641710281372, "learning_rate": 6.423400000000001e-06, "loss": 0.0084, "step": 17884 }, { "epoch": 0.9535106893426454, "grad_norm": 2.1139798164367676, "learning_rate": 6.423200000000001e-06, "loss": 0.0257, "step": 17885 }, { "epoch": 0.9535640027722984, "grad_norm": 2.6805036067962646, "learning_rate": 6.423e-06, "loss": 0.0366, "step": 17886 }, { "epoch": 0.9536173162019512, "grad_norm": 1.4414044618606567, "learning_rate": 6.4228e-06, "loss": 0.0104, "step": 17887 }, { "epoch": 0.9536706296316042, "grad_norm": 2.4867923259735107, "learning_rate": 6.422600000000001e-06, "loss": 0.037, "step": 17888 }, { "epoch": 0.9537239430612572, "grad_norm": 1.2897359132766724, "learning_rate": 6.4224000000000005e-06, "loss": 0.019, "step": 17889 }, { "epoch": 0.95377725649091, "grad_norm": 0.8803098797798157, "learning_rate": 6.422200000000001e-06, "loss": 0.0191, "step": 17890 }, { "epoch": 0.953830569920563, "grad_norm": 1.6409780979156494, "learning_rate": 6.4220000000000005e-06, "loss": 0.0287, "step": 17891 }, { "epoch": 0.9538838833502159, "grad_norm": 3.064938545227051, "learning_rate": 6.4218e-06, "loss": 0.0264, "step": 17892 }, { "epoch": 0.9539371967798689, "grad_norm": 1.4060417413711548, "learning_rate": 6.4216e-06, "loss": 0.0119, "step": 17893 }, { "epoch": 0.9539905102095217, "grad_norm": 3.0232489109039307, "learning_rate": 6.421400000000001e-06, "loss": 0.0489, "step": 17894 }, { "epoch": 0.9540438236391747, "grad_norm": 1.4875658750534058, "learning_rate": 6.4212000000000005e-06, "loss": 0.0439, "step": 17895 }, { "epoch": 0.9540971370688276, "grad_norm": 2.848698616027832, "learning_rate": 6.421000000000001e-06, "loss": 0.0489, "step": 17896 }, { "epoch": 0.9541504504984806, "grad_norm": 0.4931907653808594, "learning_rate": 6.4208000000000004e-06, "loss": 0.0047, "step": 17897 }, { "epoch": 0.9542037639281334, "grad_norm": 1.0439794063568115, "learning_rate": 6.4206e-06, "loss": 0.0054, "step": 17898 }, { "epoch": 0.9542570773577864, "grad_norm": 1.6520941257476807, "learning_rate": 6.420400000000001e-06, "loss": 0.0262, "step": 17899 }, { "epoch": 0.9543103907874394, "grad_norm": 1.672789454460144, "learning_rate": 6.420200000000001e-06, "loss": 0.0325, "step": 17900 }, { "epoch": 0.9543637042170923, "grad_norm": 1.749715805053711, "learning_rate": 6.42e-06, "loss": 0.0387, "step": 17901 }, { "epoch": 0.9544170176467452, "grad_norm": 2.121506452560425, "learning_rate": 6.419800000000001e-06, "loss": 0.0218, "step": 17902 }, { "epoch": 0.9544703310763981, "grad_norm": 3.027273416519165, "learning_rate": 6.4196e-06, "loss": 0.0336, "step": 17903 }, { "epoch": 0.9545236445060511, "grad_norm": 2.211846351623535, "learning_rate": 6.4194e-06, "loss": 0.0295, "step": 17904 }, { "epoch": 0.954576957935704, "grad_norm": 2.385956287384033, "learning_rate": 6.419200000000001e-06, "loss": 0.0144, "step": 17905 }, { "epoch": 0.9546302713653569, "grad_norm": 2.3507542610168457, "learning_rate": 6.419000000000001e-06, "loss": 0.0183, "step": 17906 }, { "epoch": 0.9546835847950098, "grad_norm": 1.7237948179244995, "learning_rate": 6.4188e-06, "loss": 0.0442, "step": 17907 }, { "epoch": 0.9547368982246628, "grad_norm": 2.117715835571289, "learning_rate": 6.418600000000001e-06, "loss": 0.014, "step": 17908 }, { "epoch": 0.9547902116543158, "grad_norm": 2.5960724353790283, "learning_rate": 6.4184e-06, "loss": 0.0282, "step": 17909 }, { "epoch": 0.9548435250839686, "grad_norm": 3.049879789352417, "learning_rate": 6.4182e-06, "loss": 0.0184, "step": 17910 }, { "epoch": 0.9548968385136216, "grad_norm": 1.4325042963027954, "learning_rate": 6.418000000000001e-06, "loss": 0.0324, "step": 17911 }, { "epoch": 0.9549501519432745, "grad_norm": 1.3921853303909302, "learning_rate": 6.417800000000001e-06, "loss": 0.0164, "step": 17912 }, { "epoch": 0.9550034653729275, "grad_norm": 4.829133033752441, "learning_rate": 6.4176e-06, "loss": 0.054, "step": 17913 }, { "epoch": 0.9550567788025803, "grad_norm": 2.1963794231414795, "learning_rate": 6.417400000000001e-06, "loss": 0.0391, "step": 17914 }, { "epoch": 0.9551100922322333, "grad_norm": 1.5456769466400146, "learning_rate": 6.4172e-06, "loss": 0.0122, "step": 17915 }, { "epoch": 0.9551634056618862, "grad_norm": 1.3491548299789429, "learning_rate": 6.417000000000001e-06, "loss": 0.0377, "step": 17916 }, { "epoch": 0.9552167190915392, "grad_norm": 2.3256940841674805, "learning_rate": 6.416800000000001e-06, "loss": 0.0453, "step": 17917 }, { "epoch": 0.9552700325211921, "grad_norm": 1.854483962059021, "learning_rate": 6.4166000000000005e-06, "loss": 0.0271, "step": 17918 }, { "epoch": 0.955323345950845, "grad_norm": 2.931481122970581, "learning_rate": 6.4164e-06, "loss": 0.0419, "step": 17919 }, { "epoch": 0.955376659380498, "grad_norm": 2.6276984214782715, "learning_rate": 6.4162000000000005e-06, "loss": 0.0281, "step": 17920 }, { "epoch": 0.9554299728101509, "grad_norm": 1.288413166999817, "learning_rate": 6.416e-06, "loss": 0.0271, "step": 17921 }, { "epoch": 0.9554832862398038, "grad_norm": 2.6991941928863525, "learning_rate": 6.415800000000001e-06, "loss": 0.0385, "step": 17922 }, { "epoch": 0.9555365996694567, "grad_norm": 1.96945059299469, "learning_rate": 6.415600000000001e-06, "loss": 0.0334, "step": 17923 }, { "epoch": 0.9555899130991097, "grad_norm": 2.437382459640503, "learning_rate": 6.4154000000000004e-06, "loss": 0.0346, "step": 17924 }, { "epoch": 0.9556432265287625, "grad_norm": 1.528953194618225, "learning_rate": 6.4152e-06, "loss": 0.0293, "step": 17925 }, { "epoch": 0.9556965399584155, "grad_norm": 1.9609193801879883, "learning_rate": 6.415e-06, "loss": 0.0203, "step": 17926 }, { "epoch": 0.9557498533880685, "grad_norm": 1.7696192264556885, "learning_rate": 6.414800000000001e-06, "loss": 0.03, "step": 17927 }, { "epoch": 0.9558031668177214, "grad_norm": 1.5814340114593506, "learning_rate": 6.414600000000001e-06, "loss": 0.0377, "step": 17928 }, { "epoch": 0.9558564802473744, "grad_norm": 2.403496026992798, "learning_rate": 6.414400000000001e-06, "loss": 0.0331, "step": 17929 }, { "epoch": 0.9559097936770272, "grad_norm": 1.0616357326507568, "learning_rate": 6.4142e-06, "loss": 0.0205, "step": 17930 }, { "epoch": 0.9559631071066802, "grad_norm": 2.2395529747009277, "learning_rate": 6.414e-06, "loss": 0.0664, "step": 17931 }, { "epoch": 0.9560164205363331, "grad_norm": 1.598552942276001, "learning_rate": 6.4138e-06, "loss": 0.0382, "step": 17932 }, { "epoch": 0.956069733965986, "grad_norm": 1.5428940057754517, "learning_rate": 6.413600000000001e-06, "loss": 0.0156, "step": 17933 }, { "epoch": 0.9561230473956389, "grad_norm": 1.2556344270706177, "learning_rate": 6.413400000000001e-06, "loss": 0.0257, "step": 17934 }, { "epoch": 0.9561763608252919, "grad_norm": 1.7010058164596558, "learning_rate": 6.413200000000001e-06, "loss": 0.0211, "step": 17935 }, { "epoch": 0.9562296742549449, "grad_norm": 2.663330078125, "learning_rate": 6.413e-06, "loss": 0.0323, "step": 17936 }, { "epoch": 0.9562829876845977, "grad_norm": 2.5443387031555176, "learning_rate": 6.4128e-06, "loss": 0.0608, "step": 17937 }, { "epoch": 0.9563363011142507, "grad_norm": 2.481968402862549, "learning_rate": 6.412600000000001e-06, "loss": 0.0341, "step": 17938 }, { "epoch": 0.9563896145439036, "grad_norm": 1.1325374841690063, "learning_rate": 6.412400000000001e-06, "loss": 0.0247, "step": 17939 }, { "epoch": 0.9564429279735566, "grad_norm": 1.853353500366211, "learning_rate": 6.412200000000001e-06, "loss": 0.047, "step": 17940 }, { "epoch": 0.9564962414032094, "grad_norm": 1.967329502105713, "learning_rate": 6.412000000000001e-06, "loss": 0.0481, "step": 17941 }, { "epoch": 0.9565495548328624, "grad_norm": 3.935154914855957, "learning_rate": 6.4118e-06, "loss": 0.0599, "step": 17942 }, { "epoch": 0.9566028682625153, "grad_norm": 1.3172780275344849, "learning_rate": 6.4116e-06, "loss": 0.0286, "step": 17943 }, { "epoch": 0.9566561816921683, "grad_norm": 1.1023809909820557, "learning_rate": 6.411400000000001e-06, "loss": 0.0237, "step": 17944 }, { "epoch": 0.9567094951218212, "grad_norm": 1.844328761100769, "learning_rate": 6.4112000000000006e-06, "loss": 0.0239, "step": 17945 }, { "epoch": 0.9567628085514741, "grad_norm": 1.7755879163742065, "learning_rate": 6.411000000000001e-06, "loss": 0.0215, "step": 17946 }, { "epoch": 0.9568161219811271, "grad_norm": 2.1047379970550537, "learning_rate": 6.4108000000000005e-06, "loss": 0.027, "step": 17947 }, { "epoch": 0.95686943541078, "grad_norm": 1.4790860414505005, "learning_rate": 6.4106e-06, "loss": 0.0297, "step": 17948 }, { "epoch": 0.9569227488404329, "grad_norm": 1.66400146484375, "learning_rate": 6.410400000000001e-06, "loss": 0.0441, "step": 17949 }, { "epoch": 0.9569760622700858, "grad_norm": 0.7358384132385254, "learning_rate": 6.410200000000001e-06, "loss": 0.0107, "step": 17950 }, { "epoch": 0.9570293756997388, "grad_norm": 1.7553606033325195, "learning_rate": 6.4100000000000005e-06, "loss": 0.0367, "step": 17951 }, { "epoch": 0.9570826891293917, "grad_norm": 1.1071451902389526, "learning_rate": 6.409800000000001e-06, "loss": 0.0106, "step": 17952 }, { "epoch": 0.9571360025590446, "grad_norm": 1.294077754020691, "learning_rate": 6.4096000000000004e-06, "loss": 0.0327, "step": 17953 }, { "epoch": 0.9571893159886975, "grad_norm": 1.727237582206726, "learning_rate": 6.4094e-06, "loss": 0.0379, "step": 17954 }, { "epoch": 0.9572426294183505, "grad_norm": 0.8382684588432312, "learning_rate": 6.409200000000001e-06, "loss": 0.008, "step": 17955 }, { "epoch": 0.9572959428480035, "grad_norm": 1.1299855709075928, "learning_rate": 6.409000000000001e-06, "loss": 0.0293, "step": 17956 }, { "epoch": 0.9573492562776563, "grad_norm": 2.8461191654205322, "learning_rate": 6.4088e-06, "loss": 0.0389, "step": 17957 }, { "epoch": 0.9574025697073093, "grad_norm": 1.174036979675293, "learning_rate": 6.408600000000001e-06, "loss": 0.0209, "step": 17958 }, { "epoch": 0.9574558831369622, "grad_norm": 1.1123347282409668, "learning_rate": 6.4084e-06, "loss": 0.0097, "step": 17959 }, { "epoch": 0.9575091965666152, "grad_norm": 2.073413610458374, "learning_rate": 6.4082e-06, "loss": 0.0383, "step": 17960 }, { "epoch": 0.957562509996268, "grad_norm": 1.2094581127166748, "learning_rate": 6.408000000000001e-06, "loss": 0.0135, "step": 17961 }, { "epoch": 0.957615823425921, "grad_norm": 2.8082971572875977, "learning_rate": 6.407800000000001e-06, "loss": 0.0273, "step": 17962 }, { "epoch": 0.9576691368555739, "grad_norm": 1.8960597515106201, "learning_rate": 6.4076e-06, "loss": 0.0253, "step": 17963 }, { "epoch": 0.9577224502852268, "grad_norm": 1.8842251300811768, "learning_rate": 6.407400000000001e-06, "loss": 0.0296, "step": 17964 }, { "epoch": 0.9577757637148798, "grad_norm": 1.858809232711792, "learning_rate": 6.4072e-06, "loss": 0.0424, "step": 17965 }, { "epoch": 0.9578290771445327, "grad_norm": 1.9223252534866333, "learning_rate": 6.407000000000001e-06, "loss": 0.0473, "step": 17966 }, { "epoch": 0.9578823905741857, "grad_norm": 2.984654664993286, "learning_rate": 6.406800000000001e-06, "loss": 0.0553, "step": 17967 }, { "epoch": 0.9579357040038385, "grad_norm": 1.6915552616119385, "learning_rate": 6.406600000000001e-06, "loss": 0.0436, "step": 17968 }, { "epoch": 0.9579890174334915, "grad_norm": 1.8385502099990845, "learning_rate": 6.4064e-06, "loss": 0.0331, "step": 17969 }, { "epoch": 0.9580423308631444, "grad_norm": 1.959930181503296, "learning_rate": 6.406200000000001e-06, "loss": 0.0153, "step": 17970 }, { "epoch": 0.9580956442927974, "grad_norm": 2.1192309856414795, "learning_rate": 6.406e-06, "loss": 0.0232, "step": 17971 }, { "epoch": 0.9581489577224502, "grad_norm": 1.9791325330734253, "learning_rate": 6.4058000000000006e-06, "loss": 0.0344, "step": 17972 }, { "epoch": 0.9582022711521032, "grad_norm": 1.9659628868103027, "learning_rate": 6.405600000000001e-06, "loss": 0.0351, "step": 17973 }, { "epoch": 0.9582555845817562, "grad_norm": 1.4400299787521362, "learning_rate": 6.4054000000000005e-06, "loss": 0.0182, "step": 17974 }, { "epoch": 0.9583088980114091, "grad_norm": 3.44492769241333, "learning_rate": 6.4052e-06, "loss": 0.0367, "step": 17975 }, { "epoch": 0.958362211441062, "grad_norm": 2.344162940979004, "learning_rate": 6.4050000000000005e-06, "loss": 0.0266, "step": 17976 }, { "epoch": 0.9584155248707149, "grad_norm": 1.4784226417541504, "learning_rate": 6.404800000000001e-06, "loss": 0.024, "step": 17977 }, { "epoch": 0.9584688383003679, "grad_norm": 2.781087875366211, "learning_rate": 6.4046000000000005e-06, "loss": 0.0574, "step": 17978 }, { "epoch": 0.9585221517300208, "grad_norm": 0.959977388381958, "learning_rate": 6.404400000000001e-06, "loss": 0.0178, "step": 17979 }, { "epoch": 0.9585754651596737, "grad_norm": 0.9829025864601135, "learning_rate": 6.4042000000000004e-06, "loss": 0.029, "step": 17980 }, { "epoch": 0.9586287785893266, "grad_norm": 0.7823353409767151, "learning_rate": 6.404e-06, "loss": 0.0171, "step": 17981 }, { "epoch": 0.9586820920189796, "grad_norm": 1.162594199180603, "learning_rate": 6.4038e-06, "loss": 0.0132, "step": 17982 }, { "epoch": 0.9587354054486326, "grad_norm": 1.5134629011154175, "learning_rate": 6.403600000000001e-06, "loss": 0.05, "step": 17983 }, { "epoch": 0.9587887188782854, "grad_norm": 1.5441125631332397, "learning_rate": 6.4034e-06, "loss": 0.007, "step": 17984 }, { "epoch": 0.9588420323079384, "grad_norm": 2.583791971206665, "learning_rate": 6.403200000000001e-06, "loss": 0.0474, "step": 17985 }, { "epoch": 0.9588953457375913, "grad_norm": 3.5075485706329346, "learning_rate": 6.403e-06, "loss": 0.0295, "step": 17986 }, { "epoch": 0.9589486591672443, "grad_norm": 1.4381814002990723, "learning_rate": 6.4028e-06, "loss": 0.0353, "step": 17987 }, { "epoch": 0.9590019725968971, "grad_norm": 2.9440484046936035, "learning_rate": 6.402600000000001e-06, "loss": 0.0349, "step": 17988 }, { "epoch": 0.9590552860265501, "grad_norm": 1.6535943746566772, "learning_rate": 6.402400000000001e-06, "loss": 0.0447, "step": 17989 }, { "epoch": 0.959108599456203, "grad_norm": 2.5359208583831787, "learning_rate": 6.4022e-06, "loss": 0.0353, "step": 17990 }, { "epoch": 0.959161912885856, "grad_norm": 1.4684983491897583, "learning_rate": 6.402000000000001e-06, "loss": 0.0182, "step": 17991 }, { "epoch": 0.9592152263155089, "grad_norm": 2.2797484397888184, "learning_rate": 6.4018e-06, "loss": 0.0217, "step": 17992 }, { "epoch": 0.9592685397451618, "grad_norm": 2.6299777030944824, "learning_rate": 6.4016e-06, "loss": 0.0289, "step": 17993 }, { "epoch": 0.9593218531748148, "grad_norm": 1.6951500177383423, "learning_rate": 6.401400000000001e-06, "loss": 0.0181, "step": 17994 }, { "epoch": 0.9593751666044676, "grad_norm": 1.7099679708480835, "learning_rate": 6.401200000000001e-06, "loss": 0.042, "step": 17995 }, { "epoch": 0.9594284800341206, "grad_norm": 2.233882188796997, "learning_rate": 6.401e-06, "loss": 0.0186, "step": 17996 }, { "epoch": 0.9594817934637735, "grad_norm": 2.6191959381103516, "learning_rate": 6.400800000000001e-06, "loss": 0.0781, "step": 17997 }, { "epoch": 0.9595351068934265, "grad_norm": 1.6106481552124023, "learning_rate": 6.4006e-06, "loss": 0.025, "step": 17998 }, { "epoch": 0.9595884203230793, "grad_norm": 1.00117826461792, "learning_rate": 6.4004000000000014e-06, "loss": 0.0143, "step": 17999 }, { "epoch": 0.9596417337527323, "grad_norm": 1.7443300485610962, "learning_rate": 6.400200000000001e-06, "loss": 0.0217, "step": 18000 }, { "epoch": 0.9596950471823853, "grad_norm": 1.0598490238189697, "learning_rate": 6.4000000000000006e-06, "loss": 0.0176, "step": 18001 }, { "epoch": 0.9597483606120382, "grad_norm": 2.9895057678222656, "learning_rate": 6.3998e-06, "loss": 0.0301, "step": 18002 }, { "epoch": 0.9598016740416911, "grad_norm": 1.466334581375122, "learning_rate": 6.3996000000000005e-06, "loss": 0.0381, "step": 18003 }, { "epoch": 0.959854987471344, "grad_norm": 1.1586097478866577, "learning_rate": 6.3994e-06, "loss": 0.0336, "step": 18004 }, { "epoch": 0.959908300900997, "grad_norm": 2.920757532119751, "learning_rate": 6.399200000000001e-06, "loss": 0.031, "step": 18005 }, { "epoch": 0.9599616143306499, "grad_norm": 1.0977834463119507, "learning_rate": 6.399000000000001e-06, "loss": 0.0117, "step": 18006 }, { "epoch": 0.9600149277603028, "grad_norm": 1.7065595388412476, "learning_rate": 6.3988000000000005e-06, "loss": 0.0268, "step": 18007 }, { "epoch": 0.9600682411899557, "grad_norm": 3.420677661895752, "learning_rate": 6.3986e-06, "loss": 0.055, "step": 18008 }, { "epoch": 0.9601215546196087, "grad_norm": 2.164078950881958, "learning_rate": 6.3984000000000004e-06, "loss": 0.0156, "step": 18009 }, { "epoch": 0.9601748680492616, "grad_norm": 3.2160284519195557, "learning_rate": 6.398200000000001e-06, "loss": 0.0369, "step": 18010 }, { "epoch": 0.9602281814789145, "grad_norm": 2.2996490001678467, "learning_rate": 6.398000000000001e-06, "loss": 0.0362, "step": 18011 }, { "epoch": 0.9602814949085675, "grad_norm": 2.449871063232422, "learning_rate": 6.397800000000001e-06, "loss": 0.0261, "step": 18012 }, { "epoch": 0.9603348083382204, "grad_norm": 2.6147572994232178, "learning_rate": 6.3976e-06, "loss": 0.0273, "step": 18013 }, { "epoch": 0.9603881217678734, "grad_norm": 1.9704041481018066, "learning_rate": 6.3974e-06, "loss": 0.0303, "step": 18014 }, { "epoch": 0.9604414351975262, "grad_norm": 1.740325689315796, "learning_rate": 6.3972e-06, "loss": 0.0265, "step": 18015 }, { "epoch": 0.9604947486271792, "grad_norm": 1.1018507480621338, "learning_rate": 6.397000000000001e-06, "loss": 0.0125, "step": 18016 }, { "epoch": 0.9605480620568321, "grad_norm": 1.7661869525909424, "learning_rate": 6.396800000000001e-06, "loss": 0.0244, "step": 18017 }, { "epoch": 0.960601375486485, "grad_norm": 1.8951352834701538, "learning_rate": 6.396600000000001e-06, "loss": 0.0295, "step": 18018 }, { "epoch": 0.9606546889161379, "grad_norm": 1.4637852907180786, "learning_rate": 6.3964e-06, "loss": 0.0265, "step": 18019 }, { "epoch": 0.9607080023457909, "grad_norm": 2.6035709381103516, "learning_rate": 6.3962e-06, "loss": 0.0376, "step": 18020 }, { "epoch": 0.9607613157754439, "grad_norm": 2.00978684425354, "learning_rate": 6.396e-06, "loss": 0.0287, "step": 18021 }, { "epoch": 0.9608146292050967, "grad_norm": 1.3307403326034546, "learning_rate": 6.395800000000001e-06, "loss": 0.0139, "step": 18022 }, { "epoch": 0.9608679426347497, "grad_norm": 1.1139898300170898, "learning_rate": 6.395600000000001e-06, "loss": 0.014, "step": 18023 }, { "epoch": 0.9609212560644026, "grad_norm": 1.6130597591400146, "learning_rate": 6.395400000000001e-06, "loss": 0.0303, "step": 18024 }, { "epoch": 0.9609745694940556, "grad_norm": 1.4557113647460938, "learning_rate": 6.3952e-06, "loss": 0.032, "step": 18025 }, { "epoch": 0.9610278829237084, "grad_norm": 2.197253942489624, "learning_rate": 6.395e-06, "loss": 0.0185, "step": 18026 }, { "epoch": 0.9610811963533614, "grad_norm": 2.3798530101776123, "learning_rate": 6.394800000000001e-06, "loss": 0.024, "step": 18027 }, { "epoch": 0.9611345097830143, "grad_norm": 2.2043240070343018, "learning_rate": 6.3946000000000006e-06, "loss": 0.0409, "step": 18028 }, { "epoch": 0.9611878232126673, "grad_norm": 1.7688957452774048, "learning_rate": 6.394400000000001e-06, "loss": 0.0196, "step": 18029 }, { "epoch": 0.9612411366423202, "grad_norm": 1.6453771591186523, "learning_rate": 6.3942000000000005e-06, "loss": 0.0168, "step": 18030 }, { "epoch": 0.9612944500719731, "grad_norm": 2.2767932415008545, "learning_rate": 6.394e-06, "loss": 0.0445, "step": 18031 }, { "epoch": 0.9613477635016261, "grad_norm": 1.1633626222610474, "learning_rate": 6.3938e-06, "loss": 0.0196, "step": 18032 }, { "epoch": 0.961401076931279, "grad_norm": 1.6909189224243164, "learning_rate": 6.393600000000001e-06, "loss": 0.0203, "step": 18033 }, { "epoch": 0.9614543903609319, "grad_norm": 1.3459150791168213, "learning_rate": 6.3934000000000005e-06, "loss": 0.0166, "step": 18034 }, { "epoch": 0.9615077037905848, "grad_norm": 2.196981191635132, "learning_rate": 6.393200000000001e-06, "loss": 0.0234, "step": 18035 }, { "epoch": 0.9615610172202378, "grad_norm": 2.539330244064331, "learning_rate": 6.3930000000000005e-06, "loss": 0.0221, "step": 18036 }, { "epoch": 0.9616143306498907, "grad_norm": 1.7829501628875732, "learning_rate": 6.3928e-06, "loss": 0.0166, "step": 18037 }, { "epoch": 0.9616676440795436, "grad_norm": 2.360612154006958, "learning_rate": 6.392600000000001e-06, "loss": 0.0296, "step": 18038 }, { "epoch": 0.9617209575091966, "grad_norm": 1.3100038766860962, "learning_rate": 6.392400000000001e-06, "loss": 0.026, "step": 18039 }, { "epoch": 0.9617742709388495, "grad_norm": 4.872581958770752, "learning_rate": 6.3922e-06, "loss": 0.0191, "step": 18040 }, { "epoch": 0.9618275843685025, "grad_norm": 0.5756206512451172, "learning_rate": 6.392000000000001e-06, "loss": 0.0052, "step": 18041 }, { "epoch": 0.9618808977981553, "grad_norm": 0.447441965341568, "learning_rate": 6.3918e-06, "loss": 0.0045, "step": 18042 }, { "epoch": 0.9619342112278083, "grad_norm": 0.8285669088363647, "learning_rate": 6.3916e-06, "loss": 0.0057, "step": 18043 }, { "epoch": 0.9619875246574612, "grad_norm": 3.9011752605438232, "learning_rate": 6.391400000000001e-06, "loss": 0.0553, "step": 18044 }, { "epoch": 0.9620408380871142, "grad_norm": 1.601794958114624, "learning_rate": 6.391200000000001e-06, "loss": 0.0277, "step": 18045 }, { "epoch": 0.962094151516767, "grad_norm": 1.5460516214370728, "learning_rate": 6.391e-06, "loss": 0.0152, "step": 18046 }, { "epoch": 0.96214746494642, "grad_norm": 2.1277220249176025, "learning_rate": 6.390800000000001e-06, "loss": 0.0204, "step": 18047 }, { "epoch": 0.962200778376073, "grad_norm": 1.9837490320205688, "learning_rate": 6.3906e-06, "loss": 0.0866, "step": 18048 }, { "epoch": 0.9622540918057259, "grad_norm": 1.5898030996322632, "learning_rate": 6.390400000000001e-06, "loss": 0.0154, "step": 18049 }, { "epoch": 0.9623074052353788, "grad_norm": 2.394946575164795, "learning_rate": 6.390200000000001e-06, "loss": 0.0288, "step": 18050 }, { "epoch": 0.9623607186650317, "grad_norm": 1.4664472341537476, "learning_rate": 6.390000000000001e-06, "loss": 0.0207, "step": 18051 }, { "epoch": 0.9624140320946847, "grad_norm": 2.5592780113220215, "learning_rate": 6.3898e-06, "loss": 0.0278, "step": 18052 }, { "epoch": 0.9624673455243375, "grad_norm": 1.6752837896347046, "learning_rate": 6.389600000000001e-06, "loss": 0.018, "step": 18053 }, { "epoch": 0.9625206589539905, "grad_norm": 2.7182164192199707, "learning_rate": 6.3894e-06, "loss": 0.0414, "step": 18054 }, { "epoch": 0.9625739723836434, "grad_norm": 0.5472725629806519, "learning_rate": 6.389200000000001e-06, "loss": 0.0037, "step": 18055 }, { "epoch": 0.9626272858132964, "grad_norm": 0.629649817943573, "learning_rate": 6.389000000000001e-06, "loss": 0.0042, "step": 18056 }, { "epoch": 0.9626805992429494, "grad_norm": 4.061097145080566, "learning_rate": 6.3888000000000006e-06, "loss": 0.026, "step": 18057 }, { "epoch": 0.9627339126726022, "grad_norm": 0.8546110987663269, "learning_rate": 6.3886e-06, "loss": 0.0112, "step": 18058 }, { "epoch": 0.9627872261022552, "grad_norm": 2.730409622192383, "learning_rate": 6.3884000000000005e-06, "loss": 0.0457, "step": 18059 }, { "epoch": 0.9628405395319081, "grad_norm": 1.2883950471878052, "learning_rate": 6.388200000000001e-06, "loss": 0.0107, "step": 18060 }, { "epoch": 0.962893852961561, "grad_norm": 2.4141650199890137, "learning_rate": 6.3880000000000005e-06, "loss": 0.0293, "step": 18061 }, { "epoch": 0.9629471663912139, "grad_norm": 1.708266019821167, "learning_rate": 6.387800000000001e-06, "loss": 0.0307, "step": 18062 }, { "epoch": 0.9630004798208669, "grad_norm": 2.7841298580169678, "learning_rate": 6.3876000000000005e-06, "loss": 0.0301, "step": 18063 }, { "epoch": 0.9630537932505198, "grad_norm": 2.1103756427764893, "learning_rate": 6.3874e-06, "loss": 0.0343, "step": 18064 }, { "epoch": 0.9631071066801727, "grad_norm": 2.7735636234283447, "learning_rate": 6.3872000000000004e-06, "loss": 0.0191, "step": 18065 }, { "epoch": 0.9631604201098256, "grad_norm": 0.7241188883781433, "learning_rate": 6.387000000000001e-06, "loss": 0.0062, "step": 18066 }, { "epoch": 0.9632137335394786, "grad_norm": 0.8917720317840576, "learning_rate": 6.3868e-06, "loss": 0.0068, "step": 18067 }, { "epoch": 0.9632670469691316, "grad_norm": 3.2886807918548584, "learning_rate": 6.386600000000001e-06, "loss": 0.0254, "step": 18068 }, { "epoch": 0.9633203603987844, "grad_norm": 1.2390472888946533, "learning_rate": 6.3864e-06, "loss": 0.0088, "step": 18069 }, { "epoch": 0.9633736738284374, "grad_norm": 2.158247709274292, "learning_rate": 6.3862e-06, "loss": 0.0399, "step": 18070 }, { "epoch": 0.9634269872580903, "grad_norm": 1.4583133459091187, "learning_rate": 6.386e-06, "loss": 0.0432, "step": 18071 }, { "epoch": 0.9634803006877433, "grad_norm": 2.383225917816162, "learning_rate": 6.385800000000001e-06, "loss": 0.0269, "step": 18072 }, { "epoch": 0.9635336141173961, "grad_norm": 2.983832359313965, "learning_rate": 6.3856e-06, "loss": 0.0389, "step": 18073 }, { "epoch": 0.9635869275470491, "grad_norm": 1.3790253400802612, "learning_rate": 6.385400000000001e-06, "loss": 0.0153, "step": 18074 }, { "epoch": 0.963640240976702, "grad_norm": 2.582911968231201, "learning_rate": 6.3852e-06, "loss": 0.0176, "step": 18075 }, { "epoch": 0.963693554406355, "grad_norm": 2.0964157581329346, "learning_rate": 6.385e-06, "loss": 0.0286, "step": 18076 }, { "epoch": 0.9637468678360079, "grad_norm": 1.9862858057022095, "learning_rate": 6.384800000000001e-06, "loss": 0.0172, "step": 18077 }, { "epoch": 0.9638001812656608, "grad_norm": 1.7214916944503784, "learning_rate": 6.384600000000001e-06, "loss": 0.0157, "step": 18078 }, { "epoch": 0.9638534946953138, "grad_norm": 1.3967375755310059, "learning_rate": 6.3844e-06, "loss": 0.0128, "step": 18079 }, { "epoch": 0.9639068081249667, "grad_norm": 1.6179510354995728, "learning_rate": 6.384200000000001e-06, "loss": 0.015, "step": 18080 }, { "epoch": 0.9639601215546196, "grad_norm": 1.8230336904525757, "learning_rate": 6.384e-06, "loss": 0.0511, "step": 18081 }, { "epoch": 0.9640134349842725, "grad_norm": 1.3680768013000488, "learning_rate": 6.3838e-06, "loss": 0.0098, "step": 18082 }, { "epoch": 0.9640667484139255, "grad_norm": 2.056411027908325, "learning_rate": 6.383600000000001e-06, "loss": 0.0379, "step": 18083 }, { "epoch": 0.9641200618435783, "grad_norm": 2.4264495372772217, "learning_rate": 6.383400000000001e-06, "loss": 0.0389, "step": 18084 }, { "epoch": 0.9641733752732313, "grad_norm": 5.51072883605957, "learning_rate": 6.3832e-06, "loss": 0.0236, "step": 18085 }, { "epoch": 0.9642266887028843, "grad_norm": 1.6245533227920532, "learning_rate": 6.3830000000000006e-06, "loss": 0.0095, "step": 18086 }, { "epoch": 0.9642800021325372, "grad_norm": 4.368186950683594, "learning_rate": 6.3828e-06, "loss": 0.0344, "step": 18087 }, { "epoch": 0.9643333155621902, "grad_norm": 1.7389276027679443, "learning_rate": 6.382600000000001e-06, "loss": 0.0178, "step": 18088 }, { "epoch": 0.964386628991843, "grad_norm": 1.4297199249267578, "learning_rate": 6.382400000000001e-06, "loss": 0.0076, "step": 18089 }, { "epoch": 0.964439942421496, "grad_norm": 1.9397023916244507, "learning_rate": 6.3822000000000005e-06, "loss": 0.0281, "step": 18090 }, { "epoch": 0.9644932558511489, "grad_norm": 1.9965161085128784, "learning_rate": 6.382e-06, "loss": 0.0314, "step": 18091 }, { "epoch": 0.9645465692808018, "grad_norm": 2.2068700790405273, "learning_rate": 6.3818000000000005e-06, "loss": 0.021, "step": 18092 }, { "epoch": 0.9645998827104547, "grad_norm": 3.853365182876587, "learning_rate": 6.3816e-06, "loss": 0.0786, "step": 18093 }, { "epoch": 0.9646531961401077, "grad_norm": 2.584906578063965, "learning_rate": 6.381400000000001e-06, "loss": 0.02, "step": 18094 }, { "epoch": 0.9647065095697607, "grad_norm": 1.4977636337280273, "learning_rate": 6.381200000000001e-06, "loss": 0.0322, "step": 18095 }, { "epoch": 0.9647598229994135, "grad_norm": 2.98335599899292, "learning_rate": 6.381e-06, "loss": 0.0548, "step": 18096 }, { "epoch": 0.9648131364290665, "grad_norm": 2.179218053817749, "learning_rate": 6.3808e-06, "loss": 0.0212, "step": 18097 }, { "epoch": 0.9648664498587194, "grad_norm": 1.8082112073898315, "learning_rate": 6.3806e-06, "loss": 0.0156, "step": 18098 }, { "epoch": 0.9649197632883724, "grad_norm": 2.575993061065674, "learning_rate": 6.380400000000001e-06, "loss": 0.0301, "step": 18099 }, { "epoch": 0.9649730767180252, "grad_norm": 2.77944278717041, "learning_rate": 6.380200000000001e-06, "loss": 0.0538, "step": 18100 }, { "epoch": 0.9650263901476782, "grad_norm": 3.1674814224243164, "learning_rate": 6.380000000000001e-06, "loss": 0.0241, "step": 18101 }, { "epoch": 0.9650797035773311, "grad_norm": 3.099398374557495, "learning_rate": 6.3798e-06, "loss": 0.0413, "step": 18102 }, { "epoch": 0.9651330170069841, "grad_norm": 2.4071335792541504, "learning_rate": 6.3796e-06, "loss": 0.0309, "step": 18103 }, { "epoch": 0.965186330436637, "grad_norm": 3.5712966918945312, "learning_rate": 6.3794e-06, "loss": 0.0431, "step": 18104 }, { "epoch": 0.9652396438662899, "grad_norm": 2.467381000518799, "learning_rate": 6.379200000000001e-06, "loss": 0.0411, "step": 18105 }, { "epoch": 0.9652929572959429, "grad_norm": 2.209968090057373, "learning_rate": 6.379000000000001e-06, "loss": 0.0444, "step": 18106 }, { "epoch": 0.9653462707255958, "grad_norm": 2.306553840637207, "learning_rate": 6.378800000000001e-06, "loss": 0.0356, "step": 18107 }, { "epoch": 0.9653995841552487, "grad_norm": 3.2889132499694824, "learning_rate": 6.3786e-06, "loss": 0.0298, "step": 18108 }, { "epoch": 0.9654528975849016, "grad_norm": 3.2690367698669434, "learning_rate": 6.3784e-06, "loss": 0.0277, "step": 18109 }, { "epoch": 0.9655062110145546, "grad_norm": 2.2729930877685547, "learning_rate": 6.378200000000001e-06, "loss": 0.0379, "step": 18110 }, { "epoch": 0.9655595244442075, "grad_norm": 1.7085505723953247, "learning_rate": 6.378000000000001e-06, "loss": 0.0332, "step": 18111 }, { "epoch": 0.9656128378738604, "grad_norm": 1.6490153074264526, "learning_rate": 6.377800000000001e-06, "loss": 0.0292, "step": 18112 }, { "epoch": 0.9656661513035134, "grad_norm": 1.8361610174179077, "learning_rate": 6.3776000000000006e-06, "loss": 0.0324, "step": 18113 }, { "epoch": 0.9657194647331663, "grad_norm": 1.6561129093170166, "learning_rate": 6.3774e-06, "loss": 0.0144, "step": 18114 }, { "epoch": 0.9657727781628193, "grad_norm": 2.7877302169799805, "learning_rate": 6.3772e-06, "loss": 0.0348, "step": 18115 }, { "epoch": 0.9658260915924721, "grad_norm": 1.2197481393814087, "learning_rate": 6.377000000000001e-06, "loss": 0.016, "step": 18116 }, { "epoch": 0.9658794050221251, "grad_norm": 1.9784413576126099, "learning_rate": 6.3768000000000005e-06, "loss": 0.031, "step": 18117 }, { "epoch": 0.965932718451778, "grad_norm": 2.211136817932129, "learning_rate": 6.376600000000001e-06, "loss": 0.0312, "step": 18118 }, { "epoch": 0.965986031881431, "grad_norm": 1.218124508857727, "learning_rate": 6.3764000000000005e-06, "loss": 0.0207, "step": 18119 }, { "epoch": 0.9660393453110838, "grad_norm": 6.47661018371582, "learning_rate": 6.3762e-06, "loss": 0.0404, "step": 18120 }, { "epoch": 0.9660926587407368, "grad_norm": 1.40505051612854, "learning_rate": 6.376e-06, "loss": 0.0298, "step": 18121 }, { "epoch": 0.9661459721703897, "grad_norm": 4.869135856628418, "learning_rate": 6.375800000000001e-06, "loss": 0.0255, "step": 18122 }, { "epoch": 0.9661992856000426, "grad_norm": 1.2651640176773071, "learning_rate": 6.3756e-06, "loss": 0.0325, "step": 18123 }, { "epoch": 0.9662525990296956, "grad_norm": 2.074331045150757, "learning_rate": 6.375400000000001e-06, "loss": 0.0198, "step": 18124 }, { "epoch": 0.9663059124593485, "grad_norm": 1.4115079641342163, "learning_rate": 6.3752e-06, "loss": 0.024, "step": 18125 }, { "epoch": 0.9663592258890015, "grad_norm": 1.6820224523544312, "learning_rate": 6.375e-06, "loss": 0.0495, "step": 18126 }, { "epoch": 0.9664125393186543, "grad_norm": 1.3480693101882935, "learning_rate": 6.374800000000001e-06, "loss": 0.026, "step": 18127 }, { "epoch": 0.9664658527483073, "grad_norm": 1.9868324995040894, "learning_rate": 6.374600000000001e-06, "loss": 0.0352, "step": 18128 }, { "epoch": 0.9665191661779602, "grad_norm": 2.4360036849975586, "learning_rate": 6.3744e-06, "loss": 0.036, "step": 18129 }, { "epoch": 0.9665724796076132, "grad_norm": 2.2811264991760254, "learning_rate": 6.374200000000001e-06, "loss": 0.027, "step": 18130 }, { "epoch": 0.966625793037266, "grad_norm": 1.3144370317459106, "learning_rate": 6.374e-06, "loss": 0.0109, "step": 18131 }, { "epoch": 0.966679106466919, "grad_norm": 1.4393606185913086, "learning_rate": 6.3738e-06, "loss": 0.0186, "step": 18132 }, { "epoch": 0.966732419896572, "grad_norm": 2.050992250442505, "learning_rate": 6.373600000000001e-06, "loss": 0.036, "step": 18133 }, { "epoch": 0.9667857333262249, "grad_norm": 3.2676219940185547, "learning_rate": 6.373400000000001e-06, "loss": 0.0483, "step": 18134 }, { "epoch": 0.9668390467558778, "grad_norm": 1.635661244392395, "learning_rate": 6.3732e-06, "loss": 0.034, "step": 18135 }, { "epoch": 0.9668923601855307, "grad_norm": 1.8338940143585205, "learning_rate": 6.373000000000001e-06, "loss": 0.0372, "step": 18136 }, { "epoch": 0.9669456736151837, "grad_norm": 1.345046877861023, "learning_rate": 6.3728e-06, "loss": 0.0323, "step": 18137 }, { "epoch": 0.9669989870448366, "grad_norm": 1.2959978580474854, "learning_rate": 6.372600000000001e-06, "loss": 0.0169, "step": 18138 }, { "epoch": 0.9670523004744895, "grad_norm": 2.4628210067749023, "learning_rate": 6.372400000000001e-06, "loss": 0.0207, "step": 18139 }, { "epoch": 0.9671056139041424, "grad_norm": 1.1193591356277466, "learning_rate": 6.372200000000001e-06, "loss": 0.0136, "step": 18140 }, { "epoch": 0.9671589273337954, "grad_norm": 3.998770236968994, "learning_rate": 6.372e-06, "loss": 0.0375, "step": 18141 }, { "epoch": 0.9672122407634484, "grad_norm": 2.7234692573547363, "learning_rate": 6.3718000000000006e-06, "loss": 0.0451, "step": 18142 }, { "epoch": 0.9672655541931012, "grad_norm": 1.36408269405365, "learning_rate": 6.3716e-06, "loss": 0.0321, "step": 18143 }, { "epoch": 0.9673188676227542, "grad_norm": 1.841990351676941, "learning_rate": 6.3714000000000005e-06, "loss": 0.0311, "step": 18144 }, { "epoch": 0.9673721810524071, "grad_norm": 0.6587584614753723, "learning_rate": 6.371200000000001e-06, "loss": 0.0068, "step": 18145 }, { "epoch": 0.96742549448206, "grad_norm": 1.2316064834594727, "learning_rate": 6.3710000000000005e-06, "loss": 0.0143, "step": 18146 }, { "epoch": 0.9674788079117129, "grad_norm": 1.8664076328277588, "learning_rate": 6.3708e-06, "loss": 0.0245, "step": 18147 }, { "epoch": 0.9675321213413659, "grad_norm": 2.6547820568084717, "learning_rate": 6.3706000000000005e-06, "loss": 0.0405, "step": 18148 }, { "epoch": 0.9675854347710188, "grad_norm": 1.6263145208358765, "learning_rate": 6.370400000000001e-06, "loss": 0.0357, "step": 18149 }, { "epoch": 0.9676387482006717, "grad_norm": 0.5614028573036194, "learning_rate": 6.3702000000000004e-06, "loss": 0.0053, "step": 18150 }, { "epoch": 0.9676920616303247, "grad_norm": 2.9104855060577393, "learning_rate": 6.370000000000001e-06, "loss": 0.0319, "step": 18151 }, { "epoch": 0.9677453750599776, "grad_norm": 1.6393319368362427, "learning_rate": 6.3698e-06, "loss": 0.0208, "step": 18152 }, { "epoch": 0.9677986884896306, "grad_norm": 2.1569533348083496, "learning_rate": 6.3696e-06, "loss": 0.0348, "step": 18153 }, { "epoch": 0.9678520019192834, "grad_norm": 2.920337438583374, "learning_rate": 6.3694e-06, "loss": 0.0459, "step": 18154 }, { "epoch": 0.9679053153489364, "grad_norm": 1.4979479312896729, "learning_rate": 6.369200000000001e-06, "loss": 0.0223, "step": 18155 }, { "epoch": 0.9679586287785893, "grad_norm": 4.6518449783325195, "learning_rate": 6.369e-06, "loss": 0.0471, "step": 18156 }, { "epoch": 0.9680119422082423, "grad_norm": 2.257455587387085, "learning_rate": 6.368800000000001e-06, "loss": 0.0147, "step": 18157 }, { "epoch": 0.9680652556378951, "grad_norm": 1.8728951215744019, "learning_rate": 6.3686e-06, "loss": 0.047, "step": 18158 }, { "epoch": 0.9681185690675481, "grad_norm": 2.482869863510132, "learning_rate": 6.3684e-06, "loss": 0.0559, "step": 18159 }, { "epoch": 0.9681718824972011, "grad_norm": 2.319591999053955, "learning_rate": 6.368200000000001e-06, "loss": 0.0299, "step": 18160 }, { "epoch": 0.968225195926854, "grad_norm": 3.2429773807525635, "learning_rate": 6.368000000000001e-06, "loss": 0.0603, "step": 18161 }, { "epoch": 0.968278509356507, "grad_norm": 2.6606192588806152, "learning_rate": 6.3678e-06, "loss": 0.0266, "step": 18162 }, { "epoch": 0.9683318227861598, "grad_norm": 3.0937705039978027, "learning_rate": 6.367600000000001e-06, "loss": 0.0296, "step": 18163 }, { "epoch": 0.9683851362158128, "grad_norm": 3.312838315963745, "learning_rate": 6.3674e-06, "loss": 0.0176, "step": 18164 }, { "epoch": 0.9684384496454657, "grad_norm": 1.0863300561904907, "learning_rate": 6.3672e-06, "loss": 0.0115, "step": 18165 }, { "epoch": 0.9684917630751186, "grad_norm": 2.9947824478149414, "learning_rate": 6.367000000000001e-06, "loss": 0.0449, "step": 18166 }, { "epoch": 0.9685450765047715, "grad_norm": 1.361602544784546, "learning_rate": 6.366800000000001e-06, "loss": 0.024, "step": 18167 }, { "epoch": 0.9685983899344245, "grad_norm": 3.27055287361145, "learning_rate": 6.3666e-06, "loss": 0.0507, "step": 18168 }, { "epoch": 0.9686517033640775, "grad_norm": 2.0796661376953125, "learning_rate": 6.3664000000000006e-06, "loss": 0.0354, "step": 18169 }, { "epoch": 0.9687050167937303, "grad_norm": 1.5175520181655884, "learning_rate": 6.3662e-06, "loss": 0.0106, "step": 18170 }, { "epoch": 0.9687583302233833, "grad_norm": 1.9447931051254272, "learning_rate": 6.366000000000001e-06, "loss": 0.0383, "step": 18171 }, { "epoch": 0.9688116436530362, "grad_norm": 2.3513617515563965, "learning_rate": 6.365800000000001e-06, "loss": 0.0241, "step": 18172 }, { "epoch": 0.9688649570826892, "grad_norm": 1.9208239316940308, "learning_rate": 6.3656000000000005e-06, "loss": 0.0673, "step": 18173 }, { "epoch": 0.968918270512342, "grad_norm": 2.589060068130493, "learning_rate": 6.3654e-06, "loss": 0.0185, "step": 18174 }, { "epoch": 0.968971583941995, "grad_norm": 2.857029914855957, "learning_rate": 6.3652000000000005e-06, "loss": 0.0493, "step": 18175 }, { "epoch": 0.9690248973716479, "grad_norm": 1.2573161125183105, "learning_rate": 6.365e-06, "loss": 0.0169, "step": 18176 }, { "epoch": 0.9690782108013009, "grad_norm": 4.285191535949707, "learning_rate": 6.364800000000001e-06, "loss": 0.042, "step": 18177 }, { "epoch": 0.9691315242309537, "grad_norm": 1.5730903148651123, "learning_rate": 6.364600000000001e-06, "loss": 0.0284, "step": 18178 }, { "epoch": 0.9691848376606067, "grad_norm": 1.0626198053359985, "learning_rate": 6.3644000000000004e-06, "loss": 0.0215, "step": 18179 }, { "epoch": 0.9692381510902597, "grad_norm": 1.2247878313064575, "learning_rate": 6.3642e-06, "loss": 0.0201, "step": 18180 }, { "epoch": 0.9692914645199125, "grad_norm": 0.655113935470581, "learning_rate": 6.364e-06, "loss": 0.007, "step": 18181 }, { "epoch": 0.9693447779495655, "grad_norm": 3.648385763168335, "learning_rate": 6.3638e-06, "loss": 0.027, "step": 18182 }, { "epoch": 0.9693980913792184, "grad_norm": 1.8961259126663208, "learning_rate": 6.363600000000001e-06, "loss": 0.0265, "step": 18183 }, { "epoch": 0.9694514048088714, "grad_norm": 2.547952175140381, "learning_rate": 6.363400000000001e-06, "loss": 0.0516, "step": 18184 }, { "epoch": 0.9695047182385242, "grad_norm": 1.8777445554733276, "learning_rate": 6.3632e-06, "loss": 0.0219, "step": 18185 }, { "epoch": 0.9695580316681772, "grad_norm": 3.0044288635253906, "learning_rate": 6.363e-06, "loss": 0.0228, "step": 18186 }, { "epoch": 0.9696113450978301, "grad_norm": 1.8071081638336182, "learning_rate": 6.3628e-06, "loss": 0.0364, "step": 18187 }, { "epoch": 0.9696646585274831, "grad_norm": 1.424665093421936, "learning_rate": 6.362600000000001e-06, "loss": 0.0208, "step": 18188 }, { "epoch": 0.969717971957136, "grad_norm": 1.433226466178894, "learning_rate": 6.362400000000001e-06, "loss": 0.0269, "step": 18189 }, { "epoch": 0.9697712853867889, "grad_norm": 1.4034770727157593, "learning_rate": 6.362200000000001e-06, "loss": 0.0122, "step": 18190 }, { "epoch": 0.9698245988164419, "grad_norm": 2.337251901626587, "learning_rate": 6.362e-06, "loss": 0.0266, "step": 18191 }, { "epoch": 0.9698779122460948, "grad_norm": 1.9880194664001465, "learning_rate": 6.3618e-06, "loss": 0.0279, "step": 18192 }, { "epoch": 0.9699312256757477, "grad_norm": 1.7335364818572998, "learning_rate": 6.3616e-06, "loss": 0.0346, "step": 18193 }, { "epoch": 0.9699845391054006, "grad_norm": 1.5796936750411987, "learning_rate": 6.361400000000001e-06, "loss": 0.0182, "step": 18194 }, { "epoch": 0.9700378525350536, "grad_norm": 1.2746788263320923, "learning_rate": 6.361200000000001e-06, "loss": 0.0104, "step": 18195 }, { "epoch": 0.9700911659647065, "grad_norm": 0.6938812136650085, "learning_rate": 6.361000000000001e-06, "loss": 0.0211, "step": 18196 }, { "epoch": 0.9701444793943594, "grad_norm": 4.449671745300293, "learning_rate": 6.3608e-06, "loss": 0.0222, "step": 18197 }, { "epoch": 0.9701977928240124, "grad_norm": 1.4557017087936401, "learning_rate": 6.3606e-06, "loss": 0.0215, "step": 18198 }, { "epoch": 0.9702511062536653, "grad_norm": 2.5097126960754395, "learning_rate": 6.360400000000001e-06, "loss": 0.0126, "step": 18199 }, { "epoch": 0.9703044196833183, "grad_norm": 1.8966500759124756, "learning_rate": 6.3602000000000005e-06, "loss": 0.0116, "step": 18200 }, { "epoch": 0.9703577331129711, "grad_norm": 2.389934778213501, "learning_rate": 6.360000000000001e-06, "loss": 0.0182, "step": 18201 }, { "epoch": 0.9704110465426241, "grad_norm": 2.860133171081543, "learning_rate": 6.3598000000000005e-06, "loss": 0.0693, "step": 18202 }, { "epoch": 0.970464359972277, "grad_norm": 1.8445860147476196, "learning_rate": 6.3596e-06, "loss": 0.024, "step": 18203 }, { "epoch": 0.97051767340193, "grad_norm": 2.488384485244751, "learning_rate": 6.3594e-06, "loss": 0.0332, "step": 18204 }, { "epoch": 0.9705709868315828, "grad_norm": 3.28883957862854, "learning_rate": 6.359200000000001e-06, "loss": 0.0327, "step": 18205 }, { "epoch": 0.9706243002612358, "grad_norm": 1.8194496631622314, "learning_rate": 6.3590000000000004e-06, "loss": 0.0468, "step": 18206 }, { "epoch": 0.9706776136908888, "grad_norm": 2.902679443359375, "learning_rate": 6.358800000000001e-06, "loss": 0.0299, "step": 18207 }, { "epoch": 0.9707309271205417, "grad_norm": 1.5483201742172241, "learning_rate": 6.3586e-06, "loss": 0.0072, "step": 18208 }, { "epoch": 0.9707842405501946, "grad_norm": 1.7499488592147827, "learning_rate": 6.3584e-06, "loss": 0.0246, "step": 18209 }, { "epoch": 0.9708375539798475, "grad_norm": 1.2129918336868286, "learning_rate": 6.358200000000001e-06, "loss": 0.0146, "step": 18210 }, { "epoch": 0.9708908674095005, "grad_norm": 1.589189052581787, "learning_rate": 6.358000000000001e-06, "loss": 0.0168, "step": 18211 }, { "epoch": 0.9709441808391533, "grad_norm": 2.570152521133423, "learning_rate": 6.3578e-06, "loss": 0.0322, "step": 18212 }, { "epoch": 0.9709974942688063, "grad_norm": 1.4136695861816406, "learning_rate": 6.357600000000001e-06, "loss": 0.0337, "step": 18213 }, { "epoch": 0.9710508076984592, "grad_norm": 1.7381452322006226, "learning_rate": 6.3574e-06, "loss": 0.0229, "step": 18214 }, { "epoch": 0.9711041211281122, "grad_norm": 2.207641124725342, "learning_rate": 6.3572e-06, "loss": 0.0227, "step": 18215 }, { "epoch": 0.9711574345577652, "grad_norm": 0.9975913763046265, "learning_rate": 6.357000000000001e-06, "loss": 0.0089, "step": 18216 }, { "epoch": 0.971210747987418, "grad_norm": 1.4927780628204346, "learning_rate": 6.356800000000001e-06, "loss": 0.012, "step": 18217 }, { "epoch": 0.971264061417071, "grad_norm": 1.5712114572525024, "learning_rate": 6.3566e-06, "loss": 0.0137, "step": 18218 }, { "epoch": 0.9713173748467239, "grad_norm": 4.558958053588867, "learning_rate": 6.356400000000001e-06, "loss": 0.0659, "step": 18219 }, { "epoch": 0.9713706882763768, "grad_norm": 5.594197750091553, "learning_rate": 6.3562e-06, "loss": 0.0274, "step": 18220 }, { "epoch": 0.9714240017060297, "grad_norm": 3.9651615619659424, "learning_rate": 6.356000000000001e-06, "loss": 0.0601, "step": 18221 }, { "epoch": 0.9714773151356827, "grad_norm": 1.565360426902771, "learning_rate": 6.355800000000001e-06, "loss": 0.0288, "step": 18222 }, { "epoch": 0.9715306285653356, "grad_norm": 1.7315163612365723, "learning_rate": 6.355600000000001e-06, "loss": 0.039, "step": 18223 }, { "epoch": 0.9715839419949885, "grad_norm": 1.7079895734786987, "learning_rate": 6.3554e-06, "loss": 0.0397, "step": 18224 }, { "epoch": 0.9716372554246415, "grad_norm": 2.232693910598755, "learning_rate": 6.355200000000001e-06, "loss": 0.0386, "step": 18225 }, { "epoch": 0.9716905688542944, "grad_norm": 0.5777382254600525, "learning_rate": 6.355e-06, "loss": 0.0052, "step": 18226 }, { "epoch": 0.9717438822839474, "grad_norm": 3.3486735820770264, "learning_rate": 6.3548000000000005e-06, "loss": 0.0399, "step": 18227 }, { "epoch": 0.9717971957136002, "grad_norm": 3.583970785140991, "learning_rate": 6.354600000000001e-06, "loss": 0.0492, "step": 18228 }, { "epoch": 0.9718505091432532, "grad_norm": 3.2412478923797607, "learning_rate": 6.3544000000000005e-06, "loss": 0.0463, "step": 18229 }, { "epoch": 0.9719038225729061, "grad_norm": 1.706135869026184, "learning_rate": 6.3542e-06, "loss": 0.0435, "step": 18230 }, { "epoch": 0.9719571360025591, "grad_norm": 1.366950273513794, "learning_rate": 6.3540000000000005e-06, "loss": 0.0113, "step": 18231 }, { "epoch": 0.9720104494322119, "grad_norm": 1.733148455619812, "learning_rate": 6.3538e-06, "loss": 0.0203, "step": 18232 }, { "epoch": 0.9720637628618649, "grad_norm": 4.504438400268555, "learning_rate": 6.3536000000000005e-06, "loss": 0.036, "step": 18233 }, { "epoch": 0.9721170762915178, "grad_norm": 1.1157944202423096, "learning_rate": 6.353400000000001e-06, "loss": 0.0145, "step": 18234 }, { "epoch": 0.9721703897211708, "grad_norm": 1.4613417387008667, "learning_rate": 6.3532000000000004e-06, "loss": 0.0314, "step": 18235 }, { "epoch": 0.9722237031508237, "grad_norm": 1.818856120109558, "learning_rate": 6.353e-06, "loss": 0.0309, "step": 18236 }, { "epoch": 0.9722770165804766, "grad_norm": 1.849321722984314, "learning_rate": 6.3528e-06, "loss": 0.017, "step": 18237 }, { "epoch": 0.9723303300101296, "grad_norm": 1.3045413494110107, "learning_rate": 6.352600000000001e-06, "loss": 0.0167, "step": 18238 }, { "epoch": 0.9723836434397825, "grad_norm": 1.6887595653533936, "learning_rate": 6.3524e-06, "loss": 0.0195, "step": 18239 }, { "epoch": 0.9724369568694354, "grad_norm": 1.8499045372009277, "learning_rate": 6.352200000000001e-06, "loss": 0.0164, "step": 18240 }, { "epoch": 0.9724902702990883, "grad_norm": 2.146249294281006, "learning_rate": 6.352e-06, "loss": 0.0149, "step": 18241 }, { "epoch": 0.9725435837287413, "grad_norm": 1.2910808324813843, "learning_rate": 6.3518e-06, "loss": 0.0116, "step": 18242 }, { "epoch": 0.9725968971583941, "grad_norm": 0.8066089153289795, "learning_rate": 6.3516e-06, "loss": 0.0219, "step": 18243 }, { "epoch": 0.9726502105880471, "grad_norm": 0.9461848735809326, "learning_rate": 6.351400000000001e-06, "loss": 0.0153, "step": 18244 }, { "epoch": 0.9727035240177001, "grad_norm": 1.598774790763855, "learning_rate": 6.3512e-06, "loss": 0.0243, "step": 18245 }, { "epoch": 0.972756837447353, "grad_norm": 3.186863660812378, "learning_rate": 6.351000000000001e-06, "loss": 0.0192, "step": 18246 }, { "epoch": 0.972810150877006, "grad_norm": 1.5565437078475952, "learning_rate": 6.3508e-06, "loss": 0.0162, "step": 18247 }, { "epoch": 0.9728634643066588, "grad_norm": 2.7929813861846924, "learning_rate": 6.3506e-06, "loss": 0.0744, "step": 18248 }, { "epoch": 0.9729167777363118, "grad_norm": 2.3612008094787598, "learning_rate": 6.350400000000001e-06, "loss": 0.0162, "step": 18249 }, { "epoch": 0.9729700911659647, "grad_norm": 1.188382863998413, "learning_rate": 6.350200000000001e-06, "loss": 0.0166, "step": 18250 }, { "epoch": 0.9730234045956176, "grad_norm": 1.1928690671920776, "learning_rate": 6.35e-06, "loss": 0.0112, "step": 18251 }, { "epoch": 0.9730767180252705, "grad_norm": 2.3249878883361816, "learning_rate": 6.349800000000001e-06, "loss": 0.0141, "step": 18252 }, { "epoch": 0.9731300314549235, "grad_norm": 1.5061811208724976, "learning_rate": 6.3496e-06, "loss": 0.0198, "step": 18253 }, { "epoch": 0.9731833448845765, "grad_norm": 1.692017674446106, "learning_rate": 6.3494e-06, "loss": 0.0184, "step": 18254 }, { "epoch": 0.9732366583142293, "grad_norm": 2.7893550395965576, "learning_rate": 6.349200000000001e-06, "loss": 0.0513, "step": 18255 }, { "epoch": 0.9732899717438823, "grad_norm": 4.132758617401123, "learning_rate": 6.3490000000000005e-06, "loss": 0.0314, "step": 18256 }, { "epoch": 0.9733432851735352, "grad_norm": 2.365408420562744, "learning_rate": 6.3488e-06, "loss": 0.0286, "step": 18257 }, { "epoch": 0.9733965986031882, "grad_norm": 0.6358179450035095, "learning_rate": 6.3486000000000005e-06, "loss": 0.0063, "step": 18258 }, { "epoch": 0.973449912032841, "grad_norm": 2.533038377761841, "learning_rate": 6.3484e-06, "loss": 0.0182, "step": 18259 }, { "epoch": 0.973503225462494, "grad_norm": 1.532423734664917, "learning_rate": 6.348200000000001e-06, "loss": 0.0296, "step": 18260 }, { "epoch": 0.9735565388921469, "grad_norm": 1.533967137336731, "learning_rate": 6.348000000000001e-06, "loss": 0.0132, "step": 18261 }, { "epoch": 0.9736098523217999, "grad_norm": 2.097092390060425, "learning_rate": 6.3478000000000004e-06, "loss": 0.0515, "step": 18262 }, { "epoch": 0.9736631657514528, "grad_norm": 0.6641369462013245, "learning_rate": 6.3476e-06, "loss": 0.0058, "step": 18263 }, { "epoch": 0.9737164791811057, "grad_norm": 1.9161601066589355, "learning_rate": 6.3474e-06, "loss": 0.0388, "step": 18264 }, { "epoch": 0.9737697926107587, "grad_norm": 2.3795228004455566, "learning_rate": 6.3472e-06, "loss": 0.0165, "step": 18265 }, { "epoch": 0.9738231060404116, "grad_norm": 2.2770731449127197, "learning_rate": 6.347000000000001e-06, "loss": 0.0715, "step": 18266 }, { "epoch": 0.9738764194700645, "grad_norm": 1.5802679061889648, "learning_rate": 6.346800000000001e-06, "loss": 0.0323, "step": 18267 }, { "epoch": 0.9739297328997174, "grad_norm": 3.654649496078491, "learning_rate": 6.3466e-06, "loss": 0.0333, "step": 18268 }, { "epoch": 0.9739830463293704, "grad_norm": 3.1953811645507812, "learning_rate": 6.3464e-06, "loss": 0.0297, "step": 18269 }, { "epoch": 0.9740363597590233, "grad_norm": 3.059013605117798, "learning_rate": 6.3462e-06, "loss": 0.0432, "step": 18270 }, { "epoch": 0.9740896731886762, "grad_norm": 2.23909068107605, "learning_rate": 6.346000000000001e-06, "loss": 0.0262, "step": 18271 }, { "epoch": 0.9741429866183292, "grad_norm": 1.6591538190841675, "learning_rate": 6.345800000000001e-06, "loss": 0.0223, "step": 18272 }, { "epoch": 0.9741963000479821, "grad_norm": 1.9560213088989258, "learning_rate": 6.345600000000001e-06, "loss": 0.0548, "step": 18273 }, { "epoch": 0.9742496134776351, "grad_norm": 2.6077938079833984, "learning_rate": 6.3454e-06, "loss": 0.0396, "step": 18274 }, { "epoch": 0.9743029269072879, "grad_norm": 3.1471517086029053, "learning_rate": 6.3452e-06, "loss": 0.047, "step": 18275 }, { "epoch": 0.9743562403369409, "grad_norm": 1.6747597455978394, "learning_rate": 6.345e-06, "loss": 0.0212, "step": 18276 }, { "epoch": 0.9744095537665938, "grad_norm": 3.0375876426696777, "learning_rate": 6.344800000000001e-06, "loss": 0.0389, "step": 18277 }, { "epoch": 0.9744628671962468, "grad_norm": 3.9492132663726807, "learning_rate": 6.344600000000001e-06, "loss": 0.0486, "step": 18278 }, { "epoch": 0.9745161806258996, "grad_norm": 3.220369577407837, "learning_rate": 6.344400000000001e-06, "loss": 0.0229, "step": 18279 }, { "epoch": 0.9745694940555526, "grad_norm": 1.7533034086227417, "learning_rate": 6.3442e-06, "loss": 0.015, "step": 18280 }, { "epoch": 0.9746228074852056, "grad_norm": 2.256352663040161, "learning_rate": 6.344e-06, "loss": 0.022, "step": 18281 }, { "epoch": 0.9746761209148584, "grad_norm": 1.62859046459198, "learning_rate": 6.343800000000001e-06, "loss": 0.0297, "step": 18282 }, { "epoch": 0.9747294343445114, "grad_norm": 2.1331968307495117, "learning_rate": 6.3436000000000006e-06, "loss": 0.0338, "step": 18283 }, { "epoch": 0.9747827477741643, "grad_norm": 1.6963274478912354, "learning_rate": 6.343400000000001e-06, "loss": 0.0197, "step": 18284 }, { "epoch": 0.9748360612038173, "grad_norm": 2.3967316150665283, "learning_rate": 6.3432000000000005e-06, "loss": 0.0352, "step": 18285 }, { "epoch": 0.9748893746334701, "grad_norm": 1.8790301084518433, "learning_rate": 6.343e-06, "loss": 0.0225, "step": 18286 }, { "epoch": 0.9749426880631231, "grad_norm": 1.0260812044143677, "learning_rate": 6.3428e-06, "loss": 0.0274, "step": 18287 }, { "epoch": 0.974996001492776, "grad_norm": 1.2048109769821167, "learning_rate": 6.342600000000001e-06, "loss": 0.0233, "step": 18288 }, { "epoch": 0.975049314922429, "grad_norm": 1.387900471687317, "learning_rate": 6.3424000000000005e-06, "loss": 0.0287, "step": 18289 }, { "epoch": 0.9751026283520818, "grad_norm": 1.3250149488449097, "learning_rate": 6.342200000000001e-06, "loss": 0.0161, "step": 18290 }, { "epoch": 0.9751559417817348, "grad_norm": 1.981919527053833, "learning_rate": 6.3420000000000004e-06, "loss": 0.0347, "step": 18291 }, { "epoch": 0.9752092552113878, "grad_norm": 1.8568722009658813, "learning_rate": 6.3418e-06, "loss": 0.0181, "step": 18292 }, { "epoch": 0.9752625686410407, "grad_norm": 1.8692514896392822, "learning_rate": 6.3415999999999996e-06, "loss": 0.0401, "step": 18293 }, { "epoch": 0.9753158820706936, "grad_norm": 1.8930109739303589, "learning_rate": 6.341400000000001e-06, "loss": 0.0321, "step": 18294 }, { "epoch": 0.9753691955003465, "grad_norm": 1.7398152351379395, "learning_rate": 6.3412e-06, "loss": 0.0256, "step": 18295 }, { "epoch": 0.9754225089299995, "grad_norm": 1.4290870428085327, "learning_rate": 6.341000000000001e-06, "loss": 0.0283, "step": 18296 }, { "epoch": 0.9754758223596524, "grad_norm": 1.8057043552398682, "learning_rate": 6.3408e-06, "loss": 0.0396, "step": 18297 }, { "epoch": 0.9755291357893053, "grad_norm": 1.4330992698669434, "learning_rate": 6.3406e-06, "loss": 0.0191, "step": 18298 }, { "epoch": 0.9755824492189582, "grad_norm": 3.066577672958374, "learning_rate": 6.340400000000001e-06, "loss": 0.0237, "step": 18299 }, { "epoch": 0.9756357626486112, "grad_norm": 1.2627137899398804, "learning_rate": 6.340200000000001e-06, "loss": 0.0361, "step": 18300 }, { "epoch": 0.9756890760782642, "grad_norm": 1.8352285623550415, "learning_rate": 6.34e-06, "loss": 0.0237, "step": 18301 }, { "epoch": 0.975742389507917, "grad_norm": 2.1740455627441406, "learning_rate": 6.339800000000001e-06, "loss": 0.0463, "step": 18302 }, { "epoch": 0.97579570293757, "grad_norm": 1.5482902526855469, "learning_rate": 6.3396e-06, "loss": 0.015, "step": 18303 }, { "epoch": 0.9758490163672229, "grad_norm": 1.6239782571792603, "learning_rate": 6.3394e-06, "loss": 0.0347, "step": 18304 }, { "epoch": 0.9759023297968759, "grad_norm": 1.105588436126709, "learning_rate": 6.339200000000001e-06, "loss": 0.0231, "step": 18305 }, { "epoch": 0.9759556432265287, "grad_norm": 1.4066908359527588, "learning_rate": 6.339000000000001e-06, "loss": 0.0496, "step": 18306 }, { "epoch": 0.9760089566561817, "grad_norm": 1.4027074575424194, "learning_rate": 6.3388e-06, "loss": 0.0261, "step": 18307 }, { "epoch": 0.9760622700858346, "grad_norm": 2.6559250354766846, "learning_rate": 6.338600000000001e-06, "loss": 0.024, "step": 18308 }, { "epoch": 0.9761155835154876, "grad_norm": 2.2804503440856934, "learning_rate": 6.3384e-06, "loss": 0.0178, "step": 18309 }, { "epoch": 0.9761688969451405, "grad_norm": 0.8875671029090881, "learning_rate": 6.338200000000001e-06, "loss": 0.0065, "step": 18310 }, { "epoch": 0.9762222103747934, "grad_norm": 1.8984917402267456, "learning_rate": 6.338000000000001e-06, "loss": 0.0253, "step": 18311 }, { "epoch": 0.9762755238044464, "grad_norm": 0.7106106877326965, "learning_rate": 6.3378000000000005e-06, "loss": 0.0078, "step": 18312 }, { "epoch": 0.9763288372340992, "grad_norm": 2.2744529247283936, "learning_rate": 6.3376e-06, "loss": 0.0429, "step": 18313 }, { "epoch": 0.9763821506637522, "grad_norm": 2.067629814147949, "learning_rate": 6.3374000000000005e-06, "loss": 0.0382, "step": 18314 }, { "epoch": 0.9764354640934051, "grad_norm": 1.197791576385498, "learning_rate": 6.3372e-06, "loss": 0.0094, "step": 18315 }, { "epoch": 0.9764887775230581, "grad_norm": 1.3471410274505615, "learning_rate": 6.337000000000001e-06, "loss": 0.0122, "step": 18316 }, { "epoch": 0.9765420909527109, "grad_norm": 1.5667551755905151, "learning_rate": 6.336800000000001e-06, "loss": 0.0249, "step": 18317 }, { "epoch": 0.9765954043823639, "grad_norm": 0.9510539174079895, "learning_rate": 6.3366000000000005e-06, "loss": 0.0135, "step": 18318 }, { "epoch": 0.9766487178120169, "grad_norm": 1.760117769241333, "learning_rate": 6.3364e-06, "loss": 0.0209, "step": 18319 }, { "epoch": 0.9767020312416698, "grad_norm": 1.008241057395935, "learning_rate": 6.3362e-06, "loss": 0.0089, "step": 18320 }, { "epoch": 0.9767553446713227, "grad_norm": 1.2746002674102783, "learning_rate": 6.336000000000001e-06, "loss": 0.0216, "step": 18321 }, { "epoch": 0.9768086581009756, "grad_norm": 1.4518074989318848, "learning_rate": 6.335800000000001e-06, "loss": 0.0303, "step": 18322 }, { "epoch": 0.9768619715306286, "grad_norm": 1.230047345161438, "learning_rate": 6.335600000000001e-06, "loss": 0.0416, "step": 18323 }, { "epoch": 0.9769152849602815, "grad_norm": 0.8850000500679016, "learning_rate": 6.3354e-06, "loss": 0.0076, "step": 18324 }, { "epoch": 0.9769685983899344, "grad_norm": 0.9564645290374756, "learning_rate": 6.3352e-06, "loss": 0.0185, "step": 18325 }, { "epoch": 0.9770219118195873, "grad_norm": 2.923800468444824, "learning_rate": 6.335e-06, "loss": 0.045, "step": 18326 }, { "epoch": 0.9770752252492403, "grad_norm": 1.5218573808670044, "learning_rate": 6.334800000000001e-06, "loss": 0.0255, "step": 18327 }, { "epoch": 0.9771285386788933, "grad_norm": 1.994010090827942, "learning_rate": 6.334600000000001e-06, "loss": 0.0164, "step": 18328 }, { "epoch": 0.9771818521085461, "grad_norm": 1.884515404701233, "learning_rate": 6.334400000000001e-06, "loss": 0.0282, "step": 18329 }, { "epoch": 0.9772351655381991, "grad_norm": 1.8743164539337158, "learning_rate": 6.3342e-06, "loss": 0.0189, "step": 18330 }, { "epoch": 0.977288478967852, "grad_norm": 0.7647665739059448, "learning_rate": 6.334e-06, "loss": 0.0054, "step": 18331 }, { "epoch": 0.977341792397505, "grad_norm": 2.704322099685669, "learning_rate": 6.333800000000001e-06, "loss": 0.027, "step": 18332 }, { "epoch": 0.9773951058271578, "grad_norm": 2.789788246154785, "learning_rate": 6.333600000000001e-06, "loss": 0.0497, "step": 18333 }, { "epoch": 0.9774484192568108, "grad_norm": 1.8390939235687256, "learning_rate": 6.333400000000001e-06, "loss": 0.0132, "step": 18334 }, { "epoch": 0.9775017326864637, "grad_norm": 0.23286963999271393, "learning_rate": 6.333200000000001e-06, "loss": 0.0021, "step": 18335 }, { "epoch": 0.9775550461161167, "grad_norm": 1.7465784549713135, "learning_rate": 6.333e-06, "loss": 0.0168, "step": 18336 }, { "epoch": 0.9776083595457696, "grad_norm": 2.4577558040618896, "learning_rate": 6.3328e-06, "loss": 0.0362, "step": 18337 }, { "epoch": 0.9776616729754225, "grad_norm": 1.8993537425994873, "learning_rate": 6.332600000000001e-06, "loss": 0.0183, "step": 18338 }, { "epoch": 0.9777149864050755, "grad_norm": 1.5311278104782104, "learning_rate": 6.3324000000000006e-06, "loss": 0.0336, "step": 18339 }, { "epoch": 0.9777682998347283, "grad_norm": 1.998828649520874, "learning_rate": 6.332200000000001e-06, "loss": 0.0321, "step": 18340 }, { "epoch": 0.9778216132643813, "grad_norm": 1.5084524154663086, "learning_rate": 6.3320000000000005e-06, "loss": 0.0186, "step": 18341 }, { "epoch": 0.9778749266940342, "grad_norm": 1.350991129875183, "learning_rate": 6.3318e-06, "loss": 0.0126, "step": 18342 }, { "epoch": 0.9779282401236872, "grad_norm": 1.8838425874710083, "learning_rate": 6.3316e-06, "loss": 0.0442, "step": 18343 }, { "epoch": 0.97798155355334, "grad_norm": 2.812234401702881, "learning_rate": 6.331400000000001e-06, "loss": 0.0412, "step": 18344 }, { "epoch": 0.978034866982993, "grad_norm": 0.6233837008476257, "learning_rate": 6.3312000000000005e-06, "loss": 0.0068, "step": 18345 }, { "epoch": 0.9780881804126459, "grad_norm": 1.186021089553833, "learning_rate": 6.331000000000001e-06, "loss": 0.0176, "step": 18346 }, { "epoch": 0.9781414938422989, "grad_norm": 2.5500690937042236, "learning_rate": 6.3308000000000004e-06, "loss": 0.0402, "step": 18347 }, { "epoch": 0.9781948072719518, "grad_norm": 1.5815294981002808, "learning_rate": 6.3306e-06, "loss": 0.0257, "step": 18348 }, { "epoch": 0.9782481207016047, "grad_norm": 5.31862735748291, "learning_rate": 6.330400000000001e-06, "loss": 0.0699, "step": 18349 }, { "epoch": 0.9783014341312577, "grad_norm": 2.0177195072174072, "learning_rate": 6.330200000000001e-06, "loss": 0.0164, "step": 18350 }, { "epoch": 0.9783547475609106, "grad_norm": 3.744666576385498, "learning_rate": 6.33e-06, "loss": 0.0467, "step": 18351 }, { "epoch": 0.9784080609905635, "grad_norm": 1.7721664905548096, "learning_rate": 6.329800000000001e-06, "loss": 0.0126, "step": 18352 }, { "epoch": 0.9784613744202164, "grad_norm": 4.073359489440918, "learning_rate": 6.3296e-06, "loss": 0.0408, "step": 18353 }, { "epoch": 0.9785146878498694, "grad_norm": 0.9140855669975281, "learning_rate": 6.3294e-06, "loss": 0.0125, "step": 18354 }, { "epoch": 0.9785680012795223, "grad_norm": 2.922006368637085, "learning_rate": 6.329200000000001e-06, "loss": 0.0245, "step": 18355 }, { "epoch": 0.9786213147091752, "grad_norm": 2.5998036861419678, "learning_rate": 6.329000000000001e-06, "loss": 0.0375, "step": 18356 }, { "epoch": 0.9786746281388282, "grad_norm": 1.6017948389053345, "learning_rate": 6.3288e-06, "loss": 0.022, "step": 18357 }, { "epoch": 0.9787279415684811, "grad_norm": 2.222118616104126, "learning_rate": 6.328600000000001e-06, "loss": 0.0559, "step": 18358 }, { "epoch": 0.9787812549981341, "grad_norm": 2.136014938354492, "learning_rate": 6.3284e-06, "loss": 0.0427, "step": 18359 }, { "epoch": 0.9788345684277869, "grad_norm": 4.365929126739502, "learning_rate": 6.328200000000001e-06, "loss": 0.0546, "step": 18360 }, { "epoch": 0.9788878818574399, "grad_norm": 0.7266733050346375, "learning_rate": 6.328000000000001e-06, "loss": 0.0074, "step": 18361 }, { "epoch": 0.9789411952870928, "grad_norm": 1.5178524255752563, "learning_rate": 6.327800000000001e-06, "loss": 0.0466, "step": 18362 }, { "epoch": 0.9789945087167458, "grad_norm": 2.0880889892578125, "learning_rate": 6.3276e-06, "loss": 0.0463, "step": 18363 }, { "epoch": 0.9790478221463986, "grad_norm": 2.6714401245117188, "learning_rate": 6.327400000000001e-06, "loss": 0.0412, "step": 18364 }, { "epoch": 0.9791011355760516, "grad_norm": 1.0688016414642334, "learning_rate": 6.3272e-06, "loss": 0.0275, "step": 18365 }, { "epoch": 0.9791544490057046, "grad_norm": 1.6058510541915894, "learning_rate": 6.327000000000001e-06, "loss": 0.0241, "step": 18366 }, { "epoch": 0.9792077624353575, "grad_norm": 1.705748438835144, "learning_rate": 6.326800000000001e-06, "loss": 0.0184, "step": 18367 }, { "epoch": 0.9792610758650104, "grad_norm": 1.6032987833023071, "learning_rate": 6.3266000000000005e-06, "loss": 0.0151, "step": 18368 }, { "epoch": 0.9793143892946633, "grad_norm": 1.8808609247207642, "learning_rate": 6.3264e-06, "loss": 0.0087, "step": 18369 }, { "epoch": 0.9793677027243163, "grad_norm": 1.5541685819625854, "learning_rate": 6.3262000000000005e-06, "loss": 0.0195, "step": 18370 }, { "epoch": 0.9794210161539691, "grad_norm": 1.1454739570617676, "learning_rate": 6.326000000000001e-06, "loss": 0.0128, "step": 18371 }, { "epoch": 0.9794743295836221, "grad_norm": 1.9755016565322876, "learning_rate": 6.3258000000000005e-06, "loss": 0.0153, "step": 18372 }, { "epoch": 0.979527643013275, "grad_norm": 1.581065058708191, "learning_rate": 6.325600000000001e-06, "loss": 0.0232, "step": 18373 }, { "epoch": 0.979580956442928, "grad_norm": 2.1318209171295166, "learning_rate": 6.3254000000000005e-06, "loss": 0.0437, "step": 18374 }, { "epoch": 0.979634269872581, "grad_norm": 1.2767179012298584, "learning_rate": 6.3252e-06, "loss": 0.0174, "step": 18375 }, { "epoch": 0.9796875833022338, "grad_norm": 3.5333991050720215, "learning_rate": 6.3250000000000004e-06, "loss": 0.0572, "step": 18376 }, { "epoch": 0.9797408967318868, "grad_norm": 0.9328632354736328, "learning_rate": 6.324800000000001e-06, "loss": 0.0188, "step": 18377 }, { "epoch": 0.9797942101615397, "grad_norm": 1.4918869733810425, "learning_rate": 6.3246e-06, "loss": 0.0205, "step": 18378 }, { "epoch": 0.9798475235911926, "grad_norm": 2.8184421062469482, "learning_rate": 6.324400000000001e-06, "loss": 0.0282, "step": 18379 }, { "epoch": 0.9799008370208455, "grad_norm": 2.2879626750946045, "learning_rate": 6.3242e-06, "loss": 0.0526, "step": 18380 }, { "epoch": 0.9799541504504985, "grad_norm": 1.3701227903366089, "learning_rate": 6.324e-06, "loss": 0.0275, "step": 18381 }, { "epoch": 0.9800074638801514, "grad_norm": 1.391033411026001, "learning_rate": 6.323800000000001e-06, "loss": 0.0139, "step": 18382 }, { "epoch": 0.9800607773098043, "grad_norm": 1.3534258604049683, "learning_rate": 6.323600000000001e-06, "loss": 0.0269, "step": 18383 }, { "epoch": 0.9801140907394573, "grad_norm": 1.4529976844787598, "learning_rate": 6.3234e-06, "loss": 0.0203, "step": 18384 }, { "epoch": 0.9801674041691102, "grad_norm": 2.274430751800537, "learning_rate": 6.323200000000001e-06, "loss": 0.0306, "step": 18385 }, { "epoch": 0.9802207175987632, "grad_norm": 1.451961636543274, "learning_rate": 6.323e-06, "loss": 0.0219, "step": 18386 }, { "epoch": 0.980274031028416, "grad_norm": 1.1951591968536377, "learning_rate": 6.3228e-06, "loss": 0.0131, "step": 18387 }, { "epoch": 0.980327344458069, "grad_norm": 1.3524184226989746, "learning_rate": 6.322600000000001e-06, "loss": 0.0234, "step": 18388 }, { "epoch": 0.9803806578877219, "grad_norm": 1.9348543882369995, "learning_rate": 6.322400000000001e-06, "loss": 0.0215, "step": 18389 }, { "epoch": 0.9804339713173749, "grad_norm": 3.163861036300659, "learning_rate": 6.3222e-06, "loss": 0.0276, "step": 18390 }, { "epoch": 0.9804872847470277, "grad_norm": 4.805883884429932, "learning_rate": 6.322000000000001e-06, "loss": 0.038, "step": 18391 }, { "epoch": 0.9805405981766807, "grad_norm": 0.8125647306442261, "learning_rate": 6.3218e-06, "loss": 0.0088, "step": 18392 }, { "epoch": 0.9805939116063337, "grad_norm": 1.0392454862594604, "learning_rate": 6.3216000000000014e-06, "loss": 0.0239, "step": 18393 }, { "epoch": 0.9806472250359866, "grad_norm": 2.08652925491333, "learning_rate": 6.321400000000001e-06, "loss": 0.0299, "step": 18394 }, { "epoch": 0.9807005384656395, "grad_norm": 1.9632383584976196, "learning_rate": 6.3212000000000006e-06, "loss": 0.0257, "step": 18395 }, { "epoch": 0.9807538518952924, "grad_norm": 1.4021600484848022, "learning_rate": 6.321e-06, "loss": 0.0193, "step": 18396 }, { "epoch": 0.9808071653249454, "grad_norm": 1.6395848989486694, "learning_rate": 6.3208000000000005e-06, "loss": 0.013, "step": 18397 }, { "epoch": 0.9808604787545983, "grad_norm": 0.9212459325790405, "learning_rate": 6.3206e-06, "loss": 0.0136, "step": 18398 }, { "epoch": 0.9809137921842512, "grad_norm": 2.7122199535369873, "learning_rate": 6.320400000000001e-06, "loss": 0.053, "step": 18399 }, { "epoch": 0.9809671056139041, "grad_norm": 2.0208775997161865, "learning_rate": 6.320200000000001e-06, "loss": 0.0273, "step": 18400 }, { "epoch": 0.9810204190435571, "grad_norm": 1.7934317588806152, "learning_rate": 6.3200000000000005e-06, "loss": 0.038, "step": 18401 }, { "epoch": 0.98107373247321, "grad_norm": 1.7484129667282104, "learning_rate": 6.3198e-06, "loss": 0.0125, "step": 18402 }, { "epoch": 0.9811270459028629, "grad_norm": 3.321281909942627, "learning_rate": 6.3196000000000004e-06, "loss": 0.052, "step": 18403 }, { "epoch": 0.9811803593325159, "grad_norm": 2.303971767425537, "learning_rate": 6.3194e-06, "loss": 0.0301, "step": 18404 }, { "epoch": 0.9812336727621688, "grad_norm": 1.6033648252487183, "learning_rate": 6.319200000000001e-06, "loss": 0.0255, "step": 18405 }, { "epoch": 0.9812869861918218, "grad_norm": 1.8900548219680786, "learning_rate": 6.319000000000001e-06, "loss": 0.0291, "step": 18406 }, { "epoch": 0.9813402996214746, "grad_norm": 1.6336268186569214, "learning_rate": 6.3188e-06, "loss": 0.0148, "step": 18407 }, { "epoch": 0.9813936130511276, "grad_norm": 6.358250141143799, "learning_rate": 6.3186e-06, "loss": 0.0396, "step": 18408 }, { "epoch": 0.9814469264807805, "grad_norm": 2.1316134929656982, "learning_rate": 6.3184e-06, "loss": 0.0367, "step": 18409 }, { "epoch": 0.9815002399104334, "grad_norm": 2.137690305709839, "learning_rate": 6.318200000000001e-06, "loss": 0.012, "step": 18410 }, { "epoch": 0.9815535533400863, "grad_norm": 1.155549168586731, "learning_rate": 6.318000000000001e-06, "loss": 0.0205, "step": 18411 }, { "epoch": 0.9816068667697393, "grad_norm": 1.287278175354004, "learning_rate": 6.317800000000001e-06, "loss": 0.0349, "step": 18412 }, { "epoch": 0.9816601801993923, "grad_norm": 4.110031604766846, "learning_rate": 6.3176e-06, "loss": 0.0289, "step": 18413 }, { "epoch": 0.9817134936290451, "grad_norm": 2.484492540359497, "learning_rate": 6.3174e-06, "loss": 0.0263, "step": 18414 }, { "epoch": 0.9817668070586981, "grad_norm": 4.0200300216674805, "learning_rate": 6.3172e-06, "loss": 0.0208, "step": 18415 }, { "epoch": 0.981820120488351, "grad_norm": 1.8886427879333496, "learning_rate": 6.317000000000001e-06, "loss": 0.0273, "step": 18416 }, { "epoch": 0.981873433918004, "grad_norm": 3.1372079849243164, "learning_rate": 6.316800000000001e-06, "loss": 0.0366, "step": 18417 }, { "epoch": 0.9819267473476568, "grad_norm": 1.9584516286849976, "learning_rate": 6.316600000000001e-06, "loss": 0.0233, "step": 18418 }, { "epoch": 0.9819800607773098, "grad_norm": 3.328996419906616, "learning_rate": 6.3164e-06, "loss": 0.0636, "step": 18419 }, { "epoch": 0.9820333742069627, "grad_norm": 1.8983489274978638, "learning_rate": 6.3162e-06, "loss": 0.0501, "step": 18420 }, { "epoch": 0.9820866876366157, "grad_norm": 3.3490400314331055, "learning_rate": 6.316000000000001e-06, "loss": 0.0409, "step": 18421 }, { "epoch": 0.9821400010662686, "grad_norm": 3.524850845336914, "learning_rate": 6.315800000000001e-06, "loss": 0.0245, "step": 18422 }, { "epoch": 0.9821933144959215, "grad_norm": 1.1493194103240967, "learning_rate": 6.315600000000001e-06, "loss": 0.0135, "step": 18423 }, { "epoch": 0.9822466279255745, "grad_norm": 2.465362787246704, "learning_rate": 6.3154000000000006e-06, "loss": 0.0356, "step": 18424 }, { "epoch": 0.9822999413552274, "grad_norm": 2.4411866664886475, "learning_rate": 6.3152e-06, "loss": 0.0284, "step": 18425 }, { "epoch": 0.9823532547848803, "grad_norm": 1.470012903213501, "learning_rate": 6.315e-06, "loss": 0.0177, "step": 18426 }, { "epoch": 0.9824065682145332, "grad_norm": 1.2542248964309692, "learning_rate": 6.314800000000001e-06, "loss": 0.0207, "step": 18427 }, { "epoch": 0.9824598816441862, "grad_norm": 2.961045026779175, "learning_rate": 6.3146000000000005e-06, "loss": 0.0393, "step": 18428 }, { "epoch": 0.982513195073839, "grad_norm": 1.7774063348770142, "learning_rate": 6.314400000000001e-06, "loss": 0.0315, "step": 18429 }, { "epoch": 0.982566508503492, "grad_norm": 1.114504098892212, "learning_rate": 6.3142000000000005e-06, "loss": 0.0265, "step": 18430 }, { "epoch": 0.982619821933145, "grad_norm": 1.8483161926269531, "learning_rate": 6.314e-06, "loss": 0.012, "step": 18431 }, { "epoch": 0.9826731353627979, "grad_norm": 2.264092445373535, "learning_rate": 6.313800000000001e-06, "loss": 0.0276, "step": 18432 }, { "epoch": 0.9827264487924509, "grad_norm": 2.9398863315582275, "learning_rate": 6.313600000000001e-06, "loss": 0.0118, "step": 18433 }, { "epoch": 0.9827797622221037, "grad_norm": 1.7454646825790405, "learning_rate": 6.3134e-06, "loss": 0.027, "step": 18434 }, { "epoch": 0.9828330756517567, "grad_norm": 0.8674755692481995, "learning_rate": 6.313200000000001e-06, "loss": 0.0082, "step": 18435 }, { "epoch": 0.9828863890814096, "grad_norm": 1.8444230556488037, "learning_rate": 6.313e-06, "loss": 0.0352, "step": 18436 }, { "epoch": 0.9829397025110626, "grad_norm": 1.6884660720825195, "learning_rate": 6.3128e-06, "loss": 0.0421, "step": 18437 }, { "epoch": 0.9829930159407154, "grad_norm": 0.5824263691902161, "learning_rate": 6.312600000000001e-06, "loss": 0.004, "step": 18438 }, { "epoch": 0.9830463293703684, "grad_norm": 1.5804505348205566, "learning_rate": 6.312400000000001e-06, "loss": 0.0344, "step": 18439 }, { "epoch": 0.9830996428000214, "grad_norm": 2.0159220695495605, "learning_rate": 6.3122e-06, "loss": 0.0258, "step": 18440 }, { "epoch": 0.9831529562296742, "grad_norm": 2.34641695022583, "learning_rate": 6.312000000000001e-06, "loss": 0.04, "step": 18441 }, { "epoch": 0.9832062696593272, "grad_norm": 2.164768695831299, "learning_rate": 6.3118e-06, "loss": 0.0287, "step": 18442 }, { "epoch": 0.9832595830889801, "grad_norm": 1.6660958528518677, "learning_rate": 6.311600000000001e-06, "loss": 0.0418, "step": 18443 }, { "epoch": 0.9833128965186331, "grad_norm": 2.171891212463379, "learning_rate": 6.311400000000001e-06, "loss": 0.0307, "step": 18444 }, { "epoch": 0.9833662099482859, "grad_norm": 0.9072350859642029, "learning_rate": 6.311200000000001e-06, "loss": 0.0095, "step": 18445 }, { "epoch": 0.9834195233779389, "grad_norm": 1.7630233764648438, "learning_rate": 6.311e-06, "loss": 0.0155, "step": 18446 }, { "epoch": 0.9834728368075918, "grad_norm": 1.414058804512024, "learning_rate": 6.310800000000001e-06, "loss": 0.0151, "step": 18447 }, { "epoch": 0.9835261502372448, "grad_norm": 2.8532559871673584, "learning_rate": 6.3106e-06, "loss": 0.0275, "step": 18448 }, { "epoch": 0.9835794636668977, "grad_norm": 0.6912444233894348, "learning_rate": 6.310400000000001e-06, "loss": 0.0065, "step": 18449 }, { "epoch": 0.9836327770965506, "grad_norm": 2.250999689102173, "learning_rate": 6.310200000000001e-06, "loss": 0.0247, "step": 18450 }, { "epoch": 0.9836860905262036, "grad_norm": 0.7044566869735718, "learning_rate": 6.3100000000000006e-06, "loss": 0.0059, "step": 18451 }, { "epoch": 0.9837394039558565, "grad_norm": 1.3420621156692505, "learning_rate": 6.3098e-06, "loss": 0.0133, "step": 18452 }, { "epoch": 0.9837927173855094, "grad_norm": 1.1931169033050537, "learning_rate": 6.3096000000000005e-06, "loss": 0.0142, "step": 18453 }, { "epoch": 0.9838460308151623, "grad_norm": 1.2181179523468018, "learning_rate": 6.3094e-06, "loss": 0.0226, "step": 18454 }, { "epoch": 0.9838993442448153, "grad_norm": 1.2812689542770386, "learning_rate": 6.3092000000000005e-06, "loss": 0.0095, "step": 18455 }, { "epoch": 0.9839526576744682, "grad_norm": 1.848280668258667, "learning_rate": 6.309000000000001e-06, "loss": 0.0186, "step": 18456 }, { "epoch": 0.9840059711041211, "grad_norm": 0.7093490362167358, "learning_rate": 6.3088000000000005e-06, "loss": 0.0095, "step": 18457 }, { "epoch": 0.984059284533774, "grad_norm": 2.0629966259002686, "learning_rate": 6.3086e-06, "loss": 0.0175, "step": 18458 }, { "epoch": 0.984112597963427, "grad_norm": 1.0711965560913086, "learning_rate": 6.3084000000000005e-06, "loss": 0.0154, "step": 18459 }, { "epoch": 0.98416591139308, "grad_norm": 0.912677526473999, "learning_rate": 6.308200000000001e-06, "loss": 0.0157, "step": 18460 }, { "epoch": 0.9842192248227328, "grad_norm": 1.6289726495742798, "learning_rate": 6.308e-06, "loss": 0.0142, "step": 18461 }, { "epoch": 0.9842725382523858, "grad_norm": 1.8998762369155884, "learning_rate": 6.307800000000001e-06, "loss": 0.0342, "step": 18462 }, { "epoch": 0.9843258516820387, "grad_norm": 2.6841447353363037, "learning_rate": 6.3076e-06, "loss": 0.0395, "step": 18463 }, { "epoch": 0.9843791651116917, "grad_norm": 1.2380939722061157, "learning_rate": 6.3074e-06, "loss": 0.008, "step": 18464 }, { "epoch": 0.9844324785413445, "grad_norm": 0.9042811989784241, "learning_rate": 6.3072e-06, "loss": 0.0054, "step": 18465 }, { "epoch": 0.9844857919709975, "grad_norm": 1.5731135606765747, "learning_rate": 6.307000000000001e-06, "loss": 0.0132, "step": 18466 }, { "epoch": 0.9845391054006504, "grad_norm": 0.7908489108085632, "learning_rate": 6.3068e-06, "loss": 0.0046, "step": 18467 }, { "epoch": 0.9845924188303034, "grad_norm": 2.6796762943267822, "learning_rate": 6.306600000000001e-06, "loss": 0.0303, "step": 18468 }, { "epoch": 0.9846457322599563, "grad_norm": 2.768921375274658, "learning_rate": 6.3064e-06, "loss": 0.0296, "step": 18469 }, { "epoch": 0.9846990456896092, "grad_norm": 3.812757730484009, "learning_rate": 6.3062e-06, "loss": 0.0451, "step": 18470 }, { "epoch": 0.9847523591192622, "grad_norm": 1.6835449934005737, "learning_rate": 6.306000000000001e-06, "loss": 0.0146, "step": 18471 }, { "epoch": 0.984805672548915, "grad_norm": 3.247070074081421, "learning_rate": 6.305800000000001e-06, "loss": 0.0437, "step": 18472 }, { "epoch": 0.984858985978568, "grad_norm": 3.917391538619995, "learning_rate": 6.3056e-06, "loss": 0.0276, "step": 18473 }, { "epoch": 0.9849122994082209, "grad_norm": 2.1866488456726074, "learning_rate": 6.305400000000001e-06, "loss": 0.0385, "step": 18474 }, { "epoch": 0.9849656128378739, "grad_norm": 1.6344497203826904, "learning_rate": 6.3052e-06, "loss": 0.0236, "step": 18475 }, { "epoch": 0.9850189262675267, "grad_norm": 1.3900713920593262, "learning_rate": 6.305e-06, "loss": 0.0196, "step": 18476 }, { "epoch": 0.9850722396971797, "grad_norm": 2.360414981842041, "learning_rate": 6.304800000000001e-06, "loss": 0.045, "step": 18477 }, { "epoch": 0.9851255531268327, "grad_norm": 2.8756561279296875, "learning_rate": 6.304600000000001e-06, "loss": 0.0188, "step": 18478 }, { "epoch": 0.9851788665564856, "grad_norm": 2.5248968601226807, "learning_rate": 6.3044e-06, "loss": 0.0348, "step": 18479 }, { "epoch": 0.9852321799861385, "grad_norm": 3.0732858180999756, "learning_rate": 6.3042000000000006e-06, "loss": 0.0778, "step": 18480 }, { "epoch": 0.9852854934157914, "grad_norm": 3.6970632076263428, "learning_rate": 6.304e-06, "loss": 0.0914, "step": 18481 }, { "epoch": 0.9853388068454444, "grad_norm": 1.9438621997833252, "learning_rate": 6.303800000000001e-06, "loss": 0.024, "step": 18482 }, { "epoch": 0.9853921202750973, "grad_norm": 3.1783862113952637, "learning_rate": 6.303600000000001e-06, "loss": 0.0237, "step": 18483 }, { "epoch": 0.9854454337047502, "grad_norm": 4.806673049926758, "learning_rate": 6.3034000000000005e-06, "loss": 0.0606, "step": 18484 }, { "epoch": 0.9854987471344031, "grad_norm": 1.1432937383651733, "learning_rate": 6.3032e-06, "loss": 0.0111, "step": 18485 }, { "epoch": 0.9855520605640561, "grad_norm": 1.6957979202270508, "learning_rate": 6.3030000000000005e-06, "loss": 0.0298, "step": 18486 }, { "epoch": 0.9856053739937091, "grad_norm": 2.1302218437194824, "learning_rate": 6.3028e-06, "loss": 0.0285, "step": 18487 }, { "epoch": 0.9856586874233619, "grad_norm": 2.256345510482788, "learning_rate": 6.302600000000001e-06, "loss": 0.0316, "step": 18488 }, { "epoch": 0.9857120008530149, "grad_norm": 1.9959369897842407, "learning_rate": 6.302400000000001e-06, "loss": 0.0541, "step": 18489 }, { "epoch": 0.9857653142826678, "grad_norm": 3.0185725688934326, "learning_rate": 6.3022e-06, "loss": 0.0546, "step": 18490 }, { "epoch": 0.9858186277123208, "grad_norm": 2.7141339778900146, "learning_rate": 6.302e-06, "loss": 0.0448, "step": 18491 }, { "epoch": 0.9858719411419736, "grad_norm": 2.1194796562194824, "learning_rate": 6.3018e-06, "loss": 0.0312, "step": 18492 }, { "epoch": 0.9859252545716266, "grad_norm": 1.0048764944076538, "learning_rate": 6.301600000000001e-06, "loss": 0.0141, "step": 18493 }, { "epoch": 0.9859785680012795, "grad_norm": 2.3161098957061768, "learning_rate": 6.301400000000001e-06, "loss": 0.0387, "step": 18494 }, { "epoch": 0.9860318814309325, "grad_norm": 3.389335870742798, "learning_rate": 6.301200000000001e-06, "loss": 0.0239, "step": 18495 }, { "epoch": 0.9860851948605854, "grad_norm": 3.4482805728912354, "learning_rate": 6.301e-06, "loss": 0.0432, "step": 18496 }, { "epoch": 0.9861385082902383, "grad_norm": 2.1709017753601074, "learning_rate": 6.3008e-06, "loss": 0.0226, "step": 18497 }, { "epoch": 0.9861918217198913, "grad_norm": 1.0053621530532837, "learning_rate": 6.3006e-06, "loss": 0.0097, "step": 18498 }, { "epoch": 0.9862451351495441, "grad_norm": 2.690999984741211, "learning_rate": 6.300400000000001e-06, "loss": 0.0579, "step": 18499 }, { "epoch": 0.9862984485791971, "grad_norm": 1.4508236646652222, "learning_rate": 6.300200000000001e-06, "loss": 0.0107, "step": 18500 }, { "epoch": 0.98635176200885, "grad_norm": 4.116832733154297, "learning_rate": 6.300000000000001e-06, "loss": 0.0678, "step": 18501 }, { "epoch": 0.986405075438503, "grad_norm": 1.950127363204956, "learning_rate": 6.2998e-06, "loss": 0.0381, "step": 18502 }, { "epoch": 0.9864583888681558, "grad_norm": 1.7988708019256592, "learning_rate": 6.2996e-06, "loss": 0.0245, "step": 18503 }, { "epoch": 0.9865117022978088, "grad_norm": 1.2762166261672974, "learning_rate": 6.299400000000001e-06, "loss": 0.0125, "step": 18504 }, { "epoch": 0.9865650157274618, "grad_norm": 1.2022569179534912, "learning_rate": 6.299200000000001e-06, "loss": 0.0216, "step": 18505 }, { "epoch": 0.9866183291571147, "grad_norm": 2.698084592819214, "learning_rate": 6.299000000000001e-06, "loss": 0.0314, "step": 18506 }, { "epoch": 0.9866716425867676, "grad_norm": 1.6612476110458374, "learning_rate": 6.298800000000001e-06, "loss": 0.0199, "step": 18507 }, { "epoch": 0.9867249560164205, "grad_norm": 0.4953737258911133, "learning_rate": 6.2986e-06, "loss": 0.0064, "step": 18508 }, { "epoch": 0.9867782694460735, "grad_norm": 1.228782057762146, "learning_rate": 6.2984e-06, "loss": 0.0319, "step": 18509 }, { "epoch": 0.9868315828757264, "grad_norm": 0.4001123309135437, "learning_rate": 6.298200000000001e-06, "loss": 0.0048, "step": 18510 }, { "epoch": 0.9868848963053793, "grad_norm": 1.0130659341812134, "learning_rate": 6.2980000000000005e-06, "loss": 0.0162, "step": 18511 }, { "epoch": 0.9869382097350322, "grad_norm": 1.6258792877197266, "learning_rate": 6.297800000000001e-06, "loss": 0.0359, "step": 18512 }, { "epoch": 0.9869915231646852, "grad_norm": 3.197201728820801, "learning_rate": 6.2976000000000005e-06, "loss": 0.0284, "step": 18513 }, { "epoch": 0.9870448365943381, "grad_norm": 2.178727388381958, "learning_rate": 6.2974e-06, "loss": 0.0247, "step": 18514 }, { "epoch": 0.987098150023991, "grad_norm": 0.333688884973526, "learning_rate": 6.2972e-06, "loss": 0.0036, "step": 18515 }, { "epoch": 0.987151463453644, "grad_norm": 1.8296748399734497, "learning_rate": 6.297000000000001e-06, "loss": 0.0294, "step": 18516 }, { "epoch": 0.9872047768832969, "grad_norm": 5.524407386779785, "learning_rate": 6.2968000000000004e-06, "loss": 0.0397, "step": 18517 }, { "epoch": 0.9872580903129499, "grad_norm": 1.688086986541748, "learning_rate": 6.296600000000001e-06, "loss": 0.0234, "step": 18518 }, { "epoch": 0.9873114037426027, "grad_norm": 1.1756927967071533, "learning_rate": 6.2964e-06, "loss": 0.0223, "step": 18519 }, { "epoch": 0.9873647171722557, "grad_norm": 2.2356202602386475, "learning_rate": 6.2962e-06, "loss": 0.0423, "step": 18520 }, { "epoch": 0.9874180306019086, "grad_norm": 2.166754961013794, "learning_rate": 6.296000000000001e-06, "loss": 0.0326, "step": 18521 }, { "epoch": 0.9874713440315616, "grad_norm": 2.5148561000823975, "learning_rate": 6.295800000000001e-06, "loss": 0.0503, "step": 18522 }, { "epoch": 0.9875246574612144, "grad_norm": 0.6036471128463745, "learning_rate": 6.2956e-06, "loss": 0.0082, "step": 18523 }, { "epoch": 0.9875779708908674, "grad_norm": 3.2408950328826904, "learning_rate": 6.295400000000001e-06, "loss": 0.049, "step": 18524 }, { "epoch": 0.9876312843205204, "grad_norm": 1.1977688074111938, "learning_rate": 6.2952e-06, "loss": 0.0251, "step": 18525 }, { "epoch": 0.9876845977501733, "grad_norm": 1.5791527032852173, "learning_rate": 6.295e-06, "loss": 0.0337, "step": 18526 }, { "epoch": 0.9877379111798262, "grad_norm": 2.650391101837158, "learning_rate": 6.294800000000001e-06, "loss": 0.0634, "step": 18527 }, { "epoch": 0.9877912246094791, "grad_norm": 2.044764280319214, "learning_rate": 6.294600000000001e-06, "loss": 0.0333, "step": 18528 }, { "epoch": 0.9878445380391321, "grad_norm": 1.2598670721054077, "learning_rate": 6.2944e-06, "loss": 0.0274, "step": 18529 }, { "epoch": 0.987897851468785, "grad_norm": 2.5295019149780273, "learning_rate": 6.294200000000001e-06, "loss": 0.0366, "step": 18530 }, { "epoch": 0.9879511648984379, "grad_norm": 2.79718279838562, "learning_rate": 6.294e-06, "loss": 0.0263, "step": 18531 }, { "epoch": 0.9880044783280908, "grad_norm": 4.307207107543945, "learning_rate": 6.293800000000001e-06, "loss": 0.0492, "step": 18532 }, { "epoch": 0.9880577917577438, "grad_norm": 1.5013073682785034, "learning_rate": 6.293600000000001e-06, "loss": 0.0286, "step": 18533 }, { "epoch": 0.9881111051873968, "grad_norm": 4.22424840927124, "learning_rate": 6.293400000000001e-06, "loss": 0.0278, "step": 18534 }, { "epoch": 0.9881644186170496, "grad_norm": 1.7766740322113037, "learning_rate": 6.2932e-06, "loss": 0.0346, "step": 18535 }, { "epoch": 0.9882177320467026, "grad_norm": 0.5076455473899841, "learning_rate": 6.2930000000000006e-06, "loss": 0.0059, "step": 18536 }, { "epoch": 0.9882710454763555, "grad_norm": 1.402411937713623, "learning_rate": 6.2928e-06, "loss": 0.0135, "step": 18537 }, { "epoch": 0.9883243589060084, "grad_norm": 1.4473862648010254, "learning_rate": 6.2926000000000005e-06, "loss": 0.0193, "step": 18538 }, { "epoch": 0.9883776723356613, "grad_norm": 0.9791440963745117, "learning_rate": 6.292400000000001e-06, "loss": 0.0168, "step": 18539 }, { "epoch": 0.9884309857653143, "grad_norm": 1.7805570363998413, "learning_rate": 6.2922000000000005e-06, "loss": 0.034, "step": 18540 }, { "epoch": 0.9884842991949672, "grad_norm": 1.0932185649871826, "learning_rate": 6.292e-06, "loss": 0.0096, "step": 18541 }, { "epoch": 0.9885376126246201, "grad_norm": 1.4865431785583496, "learning_rate": 6.2918000000000005e-06, "loss": 0.0161, "step": 18542 }, { "epoch": 0.9885909260542731, "grad_norm": 1.5426685810089111, "learning_rate": 6.291600000000001e-06, "loss": 0.0243, "step": 18543 }, { "epoch": 0.988644239483926, "grad_norm": 1.9482537508010864, "learning_rate": 6.2914000000000004e-06, "loss": 0.0437, "step": 18544 }, { "epoch": 0.988697552913579, "grad_norm": 0.8187101483345032, "learning_rate": 6.291200000000001e-06, "loss": 0.0208, "step": 18545 }, { "epoch": 0.9887508663432318, "grad_norm": 1.6237019300460815, "learning_rate": 6.291e-06, "loss": 0.0091, "step": 18546 }, { "epoch": 0.9888041797728848, "grad_norm": 0.4907078742980957, "learning_rate": 6.2908e-06, "loss": 0.0056, "step": 18547 }, { "epoch": 0.9888574932025377, "grad_norm": 1.3688989877700806, "learning_rate": 6.2906e-06, "loss": 0.028, "step": 18548 }, { "epoch": 0.9889108066321907, "grad_norm": 1.9479327201843262, "learning_rate": 6.290400000000001e-06, "loss": 0.0336, "step": 18549 }, { "epoch": 0.9889641200618435, "grad_norm": 1.9594136476516724, "learning_rate": 6.2902e-06, "loss": 0.0187, "step": 18550 }, { "epoch": 0.9890174334914965, "grad_norm": 2.077103853225708, "learning_rate": 6.290000000000001e-06, "loss": 0.0226, "step": 18551 }, { "epoch": 0.9890707469211495, "grad_norm": 2.2208900451660156, "learning_rate": 6.2898e-06, "loss": 0.0203, "step": 18552 }, { "epoch": 0.9891240603508024, "grad_norm": 3.966876268386841, "learning_rate": 6.2896e-06, "loss": 0.0303, "step": 18553 }, { "epoch": 0.9891773737804553, "grad_norm": 2.9205455780029297, "learning_rate": 6.289400000000001e-06, "loss": 0.0513, "step": 18554 }, { "epoch": 0.9892306872101082, "grad_norm": 4.020889759063721, "learning_rate": 6.289200000000001e-06, "loss": 0.0668, "step": 18555 }, { "epoch": 0.9892840006397612, "grad_norm": 1.7019200325012207, "learning_rate": 6.289e-06, "loss": 0.0159, "step": 18556 }, { "epoch": 0.989337314069414, "grad_norm": 2.6458323001861572, "learning_rate": 6.288800000000001e-06, "loss": 0.073, "step": 18557 }, { "epoch": 0.989390627499067, "grad_norm": 1.4652076959609985, "learning_rate": 6.2886e-06, "loss": 0.0147, "step": 18558 }, { "epoch": 0.9894439409287199, "grad_norm": 0.9373691082000732, "learning_rate": 6.2884e-06, "loss": 0.0102, "step": 18559 }, { "epoch": 0.9894972543583729, "grad_norm": 1.3293721675872803, "learning_rate": 6.288200000000001e-06, "loss": 0.0127, "step": 18560 }, { "epoch": 0.9895505677880259, "grad_norm": 2.4981250762939453, "learning_rate": 6.288000000000001e-06, "loss": 0.0665, "step": 18561 }, { "epoch": 0.9896038812176787, "grad_norm": 2.6622703075408936, "learning_rate": 6.2878e-06, "loss": 0.0285, "step": 18562 }, { "epoch": 0.9896571946473317, "grad_norm": 1.3558125495910645, "learning_rate": 6.287600000000001e-06, "loss": 0.0153, "step": 18563 }, { "epoch": 0.9897105080769846, "grad_norm": 1.986745834350586, "learning_rate": 6.2874e-06, "loss": 0.0415, "step": 18564 }, { "epoch": 0.9897638215066376, "grad_norm": 1.2189918756484985, "learning_rate": 6.2872e-06, "loss": 0.0111, "step": 18565 }, { "epoch": 0.9898171349362904, "grad_norm": 1.2252533435821533, "learning_rate": 6.287000000000001e-06, "loss": 0.0113, "step": 18566 }, { "epoch": 0.9898704483659434, "grad_norm": 1.125945806503296, "learning_rate": 6.2868000000000005e-06, "loss": 0.0085, "step": 18567 }, { "epoch": 0.9899237617955963, "grad_norm": 0.9633812308311462, "learning_rate": 6.2866e-06, "loss": 0.0077, "step": 18568 }, { "epoch": 0.9899770752252492, "grad_norm": 1.1889753341674805, "learning_rate": 6.2864000000000005e-06, "loss": 0.0221, "step": 18569 }, { "epoch": 0.9900303886549021, "grad_norm": 1.4775925874710083, "learning_rate": 6.2862e-06, "loss": 0.007, "step": 18570 }, { "epoch": 0.9900837020845551, "grad_norm": 2.3081607818603516, "learning_rate": 6.286000000000001e-06, "loss": 0.0443, "step": 18571 }, { "epoch": 0.9901370155142081, "grad_norm": 2.449852228164673, "learning_rate": 6.285800000000001e-06, "loss": 0.0179, "step": 18572 }, { "epoch": 0.9901903289438609, "grad_norm": 2.6874308586120605, "learning_rate": 6.2856000000000004e-06, "loss": 0.0555, "step": 18573 }, { "epoch": 0.9902436423735139, "grad_norm": 1.3035483360290527, "learning_rate": 6.2854e-06, "loss": 0.0097, "step": 18574 }, { "epoch": 0.9902969558031668, "grad_norm": 0.5447563529014587, "learning_rate": 6.2852e-06, "loss": 0.0043, "step": 18575 }, { "epoch": 0.9903502692328198, "grad_norm": 1.2425355911254883, "learning_rate": 6.285e-06, "loss": 0.008, "step": 18576 }, { "epoch": 0.9904035826624726, "grad_norm": 3.759894371032715, "learning_rate": 6.284800000000001e-06, "loss": 0.0359, "step": 18577 }, { "epoch": 0.9904568960921256, "grad_norm": 1.738621473312378, "learning_rate": 6.284600000000001e-06, "loss": 0.03, "step": 18578 }, { "epoch": 0.9905102095217785, "grad_norm": 1.9617964029312134, "learning_rate": 6.2844e-06, "loss": 0.0168, "step": 18579 }, { "epoch": 0.9905635229514315, "grad_norm": 2.13957142829895, "learning_rate": 6.2842e-06, "loss": 0.0363, "step": 18580 }, { "epoch": 0.9906168363810844, "grad_norm": 4.37744140625, "learning_rate": 6.284e-06, "loss": 0.0582, "step": 18581 }, { "epoch": 0.9906701498107373, "grad_norm": 1.699547290802002, "learning_rate": 6.283800000000001e-06, "loss": 0.0266, "step": 18582 }, { "epoch": 0.9907234632403903, "grad_norm": 3.7200629711151123, "learning_rate": 6.283600000000001e-06, "loss": 0.0478, "step": 18583 }, { "epoch": 0.9907767766700432, "grad_norm": 0.8231596350669861, "learning_rate": 6.283400000000001e-06, "loss": 0.0181, "step": 18584 }, { "epoch": 0.9908300900996961, "grad_norm": 1.3952277898788452, "learning_rate": 6.2832e-06, "loss": 0.0219, "step": 18585 }, { "epoch": 0.990883403529349, "grad_norm": 1.1856778860092163, "learning_rate": 6.283e-06, "loss": 0.0305, "step": 18586 }, { "epoch": 0.990936716959002, "grad_norm": 1.8630880117416382, "learning_rate": 6.2828e-06, "loss": 0.0273, "step": 18587 }, { "epoch": 0.9909900303886549, "grad_norm": 0.4117542505264282, "learning_rate": 6.282600000000001e-06, "loss": 0.0051, "step": 18588 }, { "epoch": 0.9910433438183078, "grad_norm": 2.2113559246063232, "learning_rate": 6.282400000000001e-06, "loss": 0.0472, "step": 18589 }, { "epoch": 0.9910966572479608, "grad_norm": 2.2135815620422363, "learning_rate": 6.282200000000001e-06, "loss": 0.027, "step": 18590 }, { "epoch": 0.9911499706776137, "grad_norm": 1.0245124101638794, "learning_rate": 6.282e-06, "loss": 0.0131, "step": 18591 }, { "epoch": 0.9912032841072667, "grad_norm": 2.2694015502929688, "learning_rate": 6.2818e-06, "loss": 0.0236, "step": 18592 }, { "epoch": 0.9912565975369195, "grad_norm": 2.024547576904297, "learning_rate": 6.281600000000001e-06, "loss": 0.0352, "step": 18593 }, { "epoch": 0.9913099109665725, "grad_norm": 3.1413486003875732, "learning_rate": 6.2814000000000005e-06, "loss": 0.0376, "step": 18594 }, { "epoch": 0.9913632243962254, "grad_norm": 3.2342560291290283, "learning_rate": 6.281200000000001e-06, "loss": 0.0545, "step": 18595 }, { "epoch": 0.9914165378258784, "grad_norm": 1.657741665840149, "learning_rate": 6.2810000000000005e-06, "loss": 0.0269, "step": 18596 }, { "epoch": 0.9914698512555312, "grad_norm": 3.0906949043273926, "learning_rate": 6.2808e-06, "loss": 0.0236, "step": 18597 }, { "epoch": 0.9915231646851842, "grad_norm": 2.1919941902160645, "learning_rate": 6.2806e-06, "loss": 0.0231, "step": 18598 }, { "epoch": 0.9915764781148372, "grad_norm": 7.089130401611328, "learning_rate": 6.280400000000001e-06, "loss": 0.0788, "step": 18599 }, { "epoch": 0.99162979154449, "grad_norm": 3.3347976207733154, "learning_rate": 6.2802000000000005e-06, "loss": 0.0197, "step": 18600 }, { "epoch": 0.991683104974143, "grad_norm": 1.2510948181152344, "learning_rate": 6.280000000000001e-06, "loss": 0.0172, "step": 18601 }, { "epoch": 0.9917364184037959, "grad_norm": 1.457574486732483, "learning_rate": 6.2798e-06, "loss": 0.0159, "step": 18602 }, { "epoch": 0.9917897318334489, "grad_norm": 1.5709091424942017, "learning_rate": 6.2796e-06, "loss": 0.0283, "step": 18603 }, { "epoch": 0.9918430452631017, "grad_norm": 4.354374408721924, "learning_rate": 6.279400000000001e-06, "loss": 0.0229, "step": 18604 }, { "epoch": 0.9918963586927547, "grad_norm": 2.800088882446289, "learning_rate": 6.279200000000001e-06, "loss": 0.0468, "step": 18605 }, { "epoch": 0.9919496721224076, "grad_norm": 2.032890796661377, "learning_rate": 6.279e-06, "loss": 0.0508, "step": 18606 }, { "epoch": 0.9920029855520606, "grad_norm": 1.0321940183639526, "learning_rate": 6.278800000000001e-06, "loss": 0.0288, "step": 18607 }, { "epoch": 0.9920562989817135, "grad_norm": 1.9853851795196533, "learning_rate": 6.2786e-06, "loss": 0.0366, "step": 18608 }, { "epoch": 0.9921096124113664, "grad_norm": 1.9551113843917847, "learning_rate": 6.2784e-06, "loss": 0.0296, "step": 18609 }, { "epoch": 0.9921629258410194, "grad_norm": 1.3647881746292114, "learning_rate": 6.278200000000001e-06, "loss": 0.0144, "step": 18610 }, { "epoch": 0.9922162392706723, "grad_norm": 1.6196120977401733, "learning_rate": 6.278000000000001e-06, "loss": 0.0264, "step": 18611 }, { "epoch": 0.9922695527003252, "grad_norm": 2.2427587509155273, "learning_rate": 6.2778e-06, "loss": 0.0213, "step": 18612 }, { "epoch": 0.9923228661299781, "grad_norm": 1.926620364189148, "learning_rate": 6.277600000000001e-06, "loss": 0.0168, "step": 18613 }, { "epoch": 0.9923761795596311, "grad_norm": 1.417298674583435, "learning_rate": 6.2774e-06, "loss": 0.0244, "step": 18614 }, { "epoch": 0.992429492989284, "grad_norm": 4.4636054039001465, "learning_rate": 6.2772e-06, "loss": 0.0528, "step": 18615 }, { "epoch": 0.9924828064189369, "grad_norm": 2.4800095558166504, "learning_rate": 6.277000000000001e-06, "loss": 0.0334, "step": 18616 }, { "epoch": 0.9925361198485899, "grad_norm": 1.3689237833023071, "learning_rate": 6.276800000000001e-06, "loss": 0.0231, "step": 18617 }, { "epoch": 0.9925894332782428, "grad_norm": 1.7810832262039185, "learning_rate": 6.2766e-06, "loss": 0.0456, "step": 18618 }, { "epoch": 0.9926427467078958, "grad_norm": 1.7536464929580688, "learning_rate": 6.276400000000001e-06, "loss": 0.0401, "step": 18619 }, { "epoch": 0.9926960601375486, "grad_norm": 1.3007603883743286, "learning_rate": 6.2762e-06, "loss": 0.0238, "step": 18620 }, { "epoch": 0.9927493735672016, "grad_norm": 1.172783613204956, "learning_rate": 6.2760000000000006e-06, "loss": 0.0138, "step": 18621 }, { "epoch": 0.9928026869968545, "grad_norm": 2.7065765857696533, "learning_rate": 6.275800000000001e-06, "loss": 0.0374, "step": 18622 }, { "epoch": 0.9928560004265075, "grad_norm": 2.018475294113159, "learning_rate": 6.2756000000000005e-06, "loss": 0.0411, "step": 18623 }, { "epoch": 0.9929093138561603, "grad_norm": 1.5538015365600586, "learning_rate": 6.2754e-06, "loss": 0.0234, "step": 18624 }, { "epoch": 0.9929626272858133, "grad_norm": 0.7998035550117493, "learning_rate": 6.2752000000000005e-06, "loss": 0.012, "step": 18625 }, { "epoch": 0.9930159407154662, "grad_norm": 1.2502387762069702, "learning_rate": 6.275e-06, "loss": 0.0193, "step": 18626 }, { "epoch": 0.9930692541451192, "grad_norm": 1.6859840154647827, "learning_rate": 6.2748000000000005e-06, "loss": 0.0356, "step": 18627 }, { "epoch": 0.9931225675747721, "grad_norm": 3.7908077239990234, "learning_rate": 6.274600000000001e-06, "loss": 0.0672, "step": 18628 }, { "epoch": 0.993175881004425, "grad_norm": 5.641821384429932, "learning_rate": 6.2744000000000004e-06, "loss": 0.0558, "step": 18629 }, { "epoch": 0.993229194434078, "grad_norm": 0.6834263205528259, "learning_rate": 6.2742e-06, "loss": 0.0078, "step": 18630 }, { "epoch": 0.9932825078637308, "grad_norm": 1.840027928352356, "learning_rate": 6.274e-06, "loss": 0.0315, "step": 18631 }, { "epoch": 0.9933358212933838, "grad_norm": 4.052925109863281, "learning_rate": 6.273800000000001e-06, "loss": 0.0688, "step": 18632 }, { "epoch": 0.9933891347230367, "grad_norm": 2.3737823963165283, "learning_rate": 6.2736e-06, "loss": 0.0294, "step": 18633 }, { "epoch": 0.9934424481526897, "grad_norm": 3.087763547897339, "learning_rate": 6.273400000000001e-06, "loss": 0.043, "step": 18634 }, { "epoch": 0.9934957615823425, "grad_norm": 1.9657686948776245, "learning_rate": 6.2732e-06, "loss": 0.0496, "step": 18635 }, { "epoch": 0.9935490750119955, "grad_norm": 1.8450298309326172, "learning_rate": 6.273e-06, "loss": 0.0215, "step": 18636 }, { "epoch": 0.9936023884416485, "grad_norm": 1.4899557828903198, "learning_rate": 6.2728e-06, "loss": 0.0301, "step": 18637 }, { "epoch": 0.9936557018713014, "grad_norm": 1.2640827894210815, "learning_rate": 6.272600000000001e-06, "loss": 0.0142, "step": 18638 }, { "epoch": 0.9937090153009543, "grad_norm": 1.4499151706695557, "learning_rate": 6.2724e-06, "loss": 0.0184, "step": 18639 }, { "epoch": 0.9937623287306072, "grad_norm": 1.8873844146728516, "learning_rate": 6.272200000000001e-06, "loss": 0.0151, "step": 18640 }, { "epoch": 0.9938156421602602, "grad_norm": 2.339564561843872, "learning_rate": 6.272e-06, "loss": 0.0535, "step": 18641 }, { "epoch": 0.9938689555899131, "grad_norm": 1.0555340051651, "learning_rate": 6.2718e-06, "loss": 0.0233, "step": 18642 }, { "epoch": 0.993922269019566, "grad_norm": 1.1779136657714844, "learning_rate": 6.271600000000001e-06, "loss": 0.0219, "step": 18643 }, { "epoch": 0.9939755824492189, "grad_norm": 1.0835118293762207, "learning_rate": 6.271400000000001e-06, "loss": 0.0152, "step": 18644 }, { "epoch": 0.9940288958788719, "grad_norm": 2.968698501586914, "learning_rate": 6.2712e-06, "loss": 0.067, "step": 18645 }, { "epoch": 0.9940822093085249, "grad_norm": 1.2235097885131836, "learning_rate": 6.271000000000001e-06, "loss": 0.0259, "step": 18646 }, { "epoch": 0.9941355227381777, "grad_norm": 1.4155774116516113, "learning_rate": 6.2708e-06, "loss": 0.0246, "step": 18647 }, { "epoch": 0.9941888361678307, "grad_norm": 2.9040744304656982, "learning_rate": 6.2706e-06, "loss": 0.0477, "step": 18648 }, { "epoch": 0.9942421495974836, "grad_norm": 1.8444035053253174, "learning_rate": 6.270400000000001e-06, "loss": 0.0206, "step": 18649 }, { "epoch": 0.9942954630271366, "grad_norm": 1.4105043411254883, "learning_rate": 6.2702000000000005e-06, "loss": 0.0162, "step": 18650 }, { "epoch": 0.9943487764567894, "grad_norm": 1.7727643251419067, "learning_rate": 6.27e-06, "loss": 0.0248, "step": 18651 }, { "epoch": 0.9944020898864424, "grad_norm": 1.4298286437988281, "learning_rate": 6.2698000000000005e-06, "loss": 0.0164, "step": 18652 }, { "epoch": 0.9944554033160953, "grad_norm": 4.137618064880371, "learning_rate": 6.2696e-06, "loss": 0.0492, "step": 18653 }, { "epoch": 0.9945087167457483, "grad_norm": 1.1783778667449951, "learning_rate": 6.269400000000001e-06, "loss": 0.0258, "step": 18654 }, { "epoch": 0.9945620301754012, "grad_norm": 2.956110715866089, "learning_rate": 6.269200000000001e-06, "loss": 0.0337, "step": 18655 }, { "epoch": 0.9946153436050541, "grad_norm": 1.5934447050094604, "learning_rate": 6.2690000000000005e-06, "loss": 0.0154, "step": 18656 }, { "epoch": 0.9946686570347071, "grad_norm": 1.1576411724090576, "learning_rate": 6.2688e-06, "loss": 0.0295, "step": 18657 }, { "epoch": 0.99472197046436, "grad_norm": 0.8677541017532349, "learning_rate": 6.2686000000000004e-06, "loss": 0.0121, "step": 18658 }, { "epoch": 0.9947752838940129, "grad_norm": 1.7979702949523926, "learning_rate": 6.2684e-06, "loss": 0.0514, "step": 18659 }, { "epoch": 0.9948285973236658, "grad_norm": 1.105747938156128, "learning_rate": 6.268200000000001e-06, "loss": 0.0101, "step": 18660 }, { "epoch": 0.9948819107533188, "grad_norm": 1.8532624244689941, "learning_rate": 6.268000000000001e-06, "loss": 0.0243, "step": 18661 }, { "epoch": 0.9949352241829716, "grad_norm": 1.5975918769836426, "learning_rate": 6.2678e-06, "loss": 0.0292, "step": 18662 }, { "epoch": 0.9949885376126246, "grad_norm": 2.4556281566619873, "learning_rate": 6.2676e-06, "loss": 0.0373, "step": 18663 }, { "epoch": 0.9950418510422776, "grad_norm": 2.5442628860473633, "learning_rate": 6.2674e-06, "loss": 0.0407, "step": 18664 }, { "epoch": 0.9950951644719305, "grad_norm": 0.9133808612823486, "learning_rate": 6.267200000000001e-06, "loss": 0.0203, "step": 18665 }, { "epoch": 0.9951484779015835, "grad_norm": 1.5824012756347656, "learning_rate": 6.267000000000001e-06, "loss": 0.0319, "step": 18666 }, { "epoch": 0.9952017913312363, "grad_norm": 1.1092959642410278, "learning_rate": 6.266800000000001e-06, "loss": 0.0086, "step": 18667 }, { "epoch": 0.9952551047608893, "grad_norm": 1.7177702188491821, "learning_rate": 6.2666e-06, "loss": 0.0506, "step": 18668 }, { "epoch": 0.9953084181905422, "grad_norm": 1.6381735801696777, "learning_rate": 6.2664e-06, "loss": 0.0144, "step": 18669 }, { "epoch": 0.9953617316201951, "grad_norm": 1.6302744150161743, "learning_rate": 6.2662e-06, "loss": 0.0358, "step": 18670 }, { "epoch": 0.995415045049848, "grad_norm": 1.759490966796875, "learning_rate": 6.266000000000001e-06, "loss": 0.0413, "step": 18671 }, { "epoch": 0.995468358479501, "grad_norm": 0.7823902368545532, "learning_rate": 6.265800000000001e-06, "loss": 0.0083, "step": 18672 }, { "epoch": 0.995521671909154, "grad_norm": 2.9485654830932617, "learning_rate": 6.265600000000001e-06, "loss": 0.0271, "step": 18673 }, { "epoch": 0.9955749853388068, "grad_norm": 1.496595859527588, "learning_rate": 6.2654e-06, "loss": 0.0141, "step": 18674 }, { "epoch": 0.9956282987684598, "grad_norm": 1.4946231842041016, "learning_rate": 6.2652e-06, "loss": 0.0291, "step": 18675 }, { "epoch": 0.9956816121981127, "grad_norm": 1.8179885149002075, "learning_rate": 6.265e-06, "loss": 0.0482, "step": 18676 }, { "epoch": 0.9957349256277657, "grad_norm": 1.614577054977417, "learning_rate": 6.2648000000000006e-06, "loss": 0.0362, "step": 18677 }, { "epoch": 0.9957882390574185, "grad_norm": 2.720503568649292, "learning_rate": 6.264600000000001e-06, "loss": 0.0806, "step": 18678 }, { "epoch": 0.9958415524870715, "grad_norm": 1.646298885345459, "learning_rate": 6.2644000000000005e-06, "loss": 0.0208, "step": 18679 }, { "epoch": 0.9958948659167244, "grad_norm": 1.369977355003357, "learning_rate": 6.2642e-06, "loss": 0.0145, "step": 18680 }, { "epoch": 0.9959481793463774, "grad_norm": 2.03682279586792, "learning_rate": 6.264e-06, "loss": 0.0306, "step": 18681 }, { "epoch": 0.9960014927760302, "grad_norm": 3.119581699371338, "learning_rate": 6.263800000000001e-06, "loss": 0.0581, "step": 18682 }, { "epoch": 0.9960548062056832, "grad_norm": 2.607523202896118, "learning_rate": 6.2636000000000005e-06, "loss": 0.0349, "step": 18683 }, { "epoch": 0.9961081196353362, "grad_norm": 2.356980562210083, "learning_rate": 6.263400000000001e-06, "loss": 0.0525, "step": 18684 }, { "epoch": 0.996161433064989, "grad_norm": 1.1897951364517212, "learning_rate": 6.2632000000000004e-06, "loss": 0.0185, "step": 18685 }, { "epoch": 0.996214746494642, "grad_norm": 1.992852807044983, "learning_rate": 6.263e-06, "loss": 0.0397, "step": 18686 }, { "epoch": 0.9962680599242949, "grad_norm": 1.8276987075805664, "learning_rate": 6.2628e-06, "loss": 0.0147, "step": 18687 }, { "epoch": 0.9963213733539479, "grad_norm": 2.2871227264404297, "learning_rate": 6.262600000000001e-06, "loss": 0.0516, "step": 18688 }, { "epoch": 0.9963746867836007, "grad_norm": 1.6024023294448853, "learning_rate": 6.2624e-06, "loss": 0.0235, "step": 18689 }, { "epoch": 0.9964280002132537, "grad_norm": 1.790853500366211, "learning_rate": 6.262200000000001e-06, "loss": 0.0436, "step": 18690 }, { "epoch": 0.9964813136429066, "grad_norm": 0.853715181350708, "learning_rate": 6.262e-06, "loss": 0.0082, "step": 18691 }, { "epoch": 0.9965346270725596, "grad_norm": 1.7823452949523926, "learning_rate": 6.2618e-06, "loss": 0.0216, "step": 18692 }, { "epoch": 0.9965879405022126, "grad_norm": 1.1696486473083496, "learning_rate": 6.261600000000001e-06, "loss": 0.034, "step": 18693 }, { "epoch": 0.9966412539318654, "grad_norm": 2.782640218734741, "learning_rate": 6.261400000000001e-06, "loss": 0.0518, "step": 18694 }, { "epoch": 0.9966945673615184, "grad_norm": 2.25966477394104, "learning_rate": 6.2612e-06, "loss": 0.0487, "step": 18695 }, { "epoch": 0.9967478807911713, "grad_norm": 2.247262716293335, "learning_rate": 6.261000000000001e-06, "loss": 0.0704, "step": 18696 }, { "epoch": 0.9968011942208242, "grad_norm": 1.3621129989624023, "learning_rate": 6.2608e-06, "loss": 0.0213, "step": 18697 }, { "epoch": 0.9968545076504771, "grad_norm": 1.160162091255188, "learning_rate": 6.2606e-06, "loss": 0.0166, "step": 18698 }, { "epoch": 0.9969078210801301, "grad_norm": 1.2185341119766235, "learning_rate": 6.260400000000001e-06, "loss": 0.0204, "step": 18699 }, { "epoch": 0.996961134509783, "grad_norm": 1.2316155433654785, "learning_rate": 6.260200000000001e-06, "loss": 0.0184, "step": 18700 }, { "epoch": 0.9970144479394359, "grad_norm": 0.7908226847648621, "learning_rate": 6.26e-06, "loss": 0.0098, "step": 18701 }, { "epoch": 0.9970677613690889, "grad_norm": 2.332529067993164, "learning_rate": 6.259800000000001e-06, "loss": 0.0394, "step": 18702 }, { "epoch": 0.9971210747987418, "grad_norm": 1.0339926481246948, "learning_rate": 6.2596e-06, "loss": 0.0202, "step": 18703 }, { "epoch": 0.9971743882283948, "grad_norm": 1.0326143503189087, "learning_rate": 6.2594000000000014e-06, "loss": 0.0114, "step": 18704 }, { "epoch": 0.9972277016580476, "grad_norm": 1.0944128036499023, "learning_rate": 6.259200000000001e-06, "loss": 0.0243, "step": 18705 }, { "epoch": 0.9972810150877006, "grad_norm": 1.8904494047164917, "learning_rate": 6.2590000000000006e-06, "loss": 0.0306, "step": 18706 }, { "epoch": 0.9973343285173535, "grad_norm": 1.3609235286712646, "learning_rate": 6.2588e-06, "loss": 0.042, "step": 18707 }, { "epoch": 0.9973876419470065, "grad_norm": 1.020402431488037, "learning_rate": 6.2586000000000005e-06, "loss": 0.0115, "step": 18708 }, { "epoch": 0.9974409553766593, "grad_norm": 1.743614673614502, "learning_rate": 6.2584e-06, "loss": 0.0402, "step": 18709 }, { "epoch": 0.9974942688063123, "grad_norm": 0.6533253192901611, "learning_rate": 6.258200000000001e-06, "loss": 0.0087, "step": 18710 }, { "epoch": 0.9975475822359653, "grad_norm": 2.39963436126709, "learning_rate": 6.258000000000001e-06, "loss": 0.0283, "step": 18711 }, { "epoch": 0.9976008956656182, "grad_norm": 1.3904443979263306, "learning_rate": 6.2578000000000005e-06, "loss": 0.0224, "step": 18712 }, { "epoch": 0.9976542090952711, "grad_norm": 1.765369176864624, "learning_rate": 6.2576e-06, "loss": 0.015, "step": 18713 }, { "epoch": 0.997707522524924, "grad_norm": 2.6020987033843994, "learning_rate": 6.2574000000000004e-06, "loss": 0.0426, "step": 18714 }, { "epoch": 0.997760835954577, "grad_norm": 0.7412105202674866, "learning_rate": 6.257200000000001e-06, "loss": 0.0155, "step": 18715 }, { "epoch": 0.9978141493842299, "grad_norm": 1.6097966432571411, "learning_rate": 6.257000000000001e-06, "loss": 0.035, "step": 18716 }, { "epoch": 0.9978674628138828, "grad_norm": 2.6530601978302, "learning_rate": 6.256800000000001e-06, "loss": 0.0464, "step": 18717 }, { "epoch": 0.9979207762435357, "grad_norm": 0.9702723026275635, "learning_rate": 6.2566e-06, "loss": 0.0117, "step": 18718 }, { "epoch": 0.9979740896731887, "grad_norm": 2.880650520324707, "learning_rate": 6.2564e-06, "loss": 0.028, "step": 18719 }, { "epoch": 0.9980274031028417, "grad_norm": 1.1604644060134888, "learning_rate": 6.2562e-06, "loss": 0.0291, "step": 18720 }, { "epoch": 0.9980807165324945, "grad_norm": 2.6556224822998047, "learning_rate": 6.256000000000001e-06, "loss": 0.0432, "step": 18721 }, { "epoch": 0.9981340299621475, "grad_norm": 4.857733249664307, "learning_rate": 6.255800000000001e-06, "loss": 0.0505, "step": 18722 }, { "epoch": 0.9981873433918004, "grad_norm": 1.2818458080291748, "learning_rate": 6.255600000000001e-06, "loss": 0.0149, "step": 18723 }, { "epoch": 0.9982406568214534, "grad_norm": 1.6639821529388428, "learning_rate": 6.2554e-06, "loss": 0.0208, "step": 18724 }, { "epoch": 0.9982939702511062, "grad_norm": 2.1165237426757812, "learning_rate": 6.2552e-06, "loss": 0.0219, "step": 18725 }, { "epoch": 0.9983472836807592, "grad_norm": 1.7787492275238037, "learning_rate": 6.255e-06, "loss": 0.0189, "step": 18726 }, { "epoch": 0.9984005971104121, "grad_norm": 3.128984212875366, "learning_rate": 6.254800000000001e-06, "loss": 0.0246, "step": 18727 }, { "epoch": 0.998453910540065, "grad_norm": 3.6696834564208984, "learning_rate": 6.254600000000001e-06, "loss": 0.0414, "step": 18728 }, { "epoch": 0.998507223969718, "grad_norm": 1.6964584589004517, "learning_rate": 6.254400000000001e-06, "loss": 0.0234, "step": 18729 }, { "epoch": 0.9985605373993709, "grad_norm": 1.993021011352539, "learning_rate": 6.2542e-06, "loss": 0.0131, "step": 18730 }, { "epoch": 0.9986138508290239, "grad_norm": 2.2092020511627197, "learning_rate": 6.254e-06, "loss": 0.0475, "step": 18731 }, { "epoch": 0.9986671642586767, "grad_norm": 1.017508625984192, "learning_rate": 6.253800000000001e-06, "loss": 0.0089, "step": 18732 }, { "epoch": 0.9987204776883297, "grad_norm": 1.0118249654769897, "learning_rate": 6.2536000000000006e-06, "loss": 0.0108, "step": 18733 }, { "epoch": 0.9987737911179826, "grad_norm": 1.567049264907837, "learning_rate": 6.253400000000001e-06, "loss": 0.0363, "step": 18734 }, { "epoch": 0.9988271045476356, "grad_norm": 2.1351184844970703, "learning_rate": 6.2532000000000005e-06, "loss": 0.0132, "step": 18735 }, { "epoch": 0.9988804179772884, "grad_norm": 1.0909518003463745, "learning_rate": 6.253e-06, "loss": 0.0106, "step": 18736 }, { "epoch": 0.9989337314069414, "grad_norm": 2.126246690750122, "learning_rate": 6.2528e-06, "loss": 0.0288, "step": 18737 }, { "epoch": 0.9989870448365943, "grad_norm": 0.9808631539344788, "learning_rate": 6.252600000000001e-06, "loss": 0.0143, "step": 18738 }, { "epoch": 0.9990403582662473, "grad_norm": 1.9848312139511108, "learning_rate": 6.2524000000000005e-06, "loss": 0.0245, "step": 18739 }, { "epoch": 0.9990936716959002, "grad_norm": 0.2997114062309265, "learning_rate": 6.252200000000001e-06, "loss": 0.0031, "step": 18740 }, { "epoch": 0.9991469851255531, "grad_norm": 0.8381993770599365, "learning_rate": 6.2520000000000004e-06, "loss": 0.0063, "step": 18741 }, { "epoch": 0.9992002985552061, "grad_norm": 1.4937678575515747, "learning_rate": 6.2518e-06, "loss": 0.0076, "step": 18742 }, { "epoch": 0.999253611984859, "grad_norm": 2.147928237915039, "learning_rate": 6.251600000000001e-06, "loss": 0.0391, "step": 18743 }, { "epoch": 0.9993069254145119, "grad_norm": 3.087061643600464, "learning_rate": 6.251400000000001e-06, "loss": 0.019, "step": 18744 }, { "epoch": 0.9993602388441648, "grad_norm": 1.933730959892273, "learning_rate": 6.2512e-06, "loss": 0.0179, "step": 18745 }, { "epoch": 0.9994135522738178, "grad_norm": 0.7606692910194397, "learning_rate": 6.251000000000001e-06, "loss": 0.0088, "step": 18746 }, { "epoch": 0.9994668657034707, "grad_norm": 1.2530871629714966, "learning_rate": 6.2508e-06, "loss": 0.0108, "step": 18747 }, { "epoch": 0.9995201791331236, "grad_norm": 6.488406181335449, "learning_rate": 6.2506e-06, "loss": 0.0285, "step": 18748 }, { "epoch": 0.9995734925627766, "grad_norm": 1.4885259866714478, "learning_rate": 6.250400000000001e-06, "loss": 0.0108, "step": 18749 }, { "epoch": 0.9996268059924295, "grad_norm": 1.8763806819915771, "learning_rate": 6.250200000000001e-06, "loss": 0.0239, "step": 18750 }, { "epoch": 0.9996801194220825, "grad_norm": 1.3375765085220337, "learning_rate": 6.25e-06, "loss": 0.0314, "step": 18751 }, { "epoch": 0.9997334328517353, "grad_norm": 2.931241035461426, "learning_rate": 6.249800000000001e-06, "loss": 0.0143, "step": 18752 }, { "epoch": 0.9997867462813883, "grad_norm": 3.515903949737549, "learning_rate": 6.2496e-06, "loss": 0.0286, "step": 18753 }, { "epoch": 0.9998400597110412, "grad_norm": 1.8258106708526611, "learning_rate": 6.249400000000001e-06, "loss": 0.0169, "step": 18754 }, { "epoch": 0.9998933731406942, "grad_norm": 2.146333932876587, "learning_rate": 6.249200000000001e-06, "loss": 0.0118, "step": 18755 }, { "epoch": 0.999946686570347, "grad_norm": 0.9668365716934204, "learning_rate": 6.249000000000001e-06, "loss": 0.0071, "step": 18756 }, { "epoch": 1.0, "grad_norm": 2.2547175884246826, "learning_rate": 6.2488e-06, "loss": 0.0152, "step": 18757 }, { "epoch": 1.000053313429653, "grad_norm": 1.9094736576080322, "learning_rate": 6.248600000000001e-06, "loss": 0.0246, "step": 18758 }, { "epoch": 1.000106626859306, "grad_norm": 1.6982715129852295, "learning_rate": 6.2484e-06, "loss": 0.0195, "step": 18759 }, { "epoch": 1.0001599402889587, "grad_norm": 2.079744577407837, "learning_rate": 6.248200000000001e-06, "loss": 0.0178, "step": 18760 }, { "epoch": 1.0002132537186117, "grad_norm": 1.8800221681594849, "learning_rate": 6.248000000000001e-06, "loss": 0.0139, "step": 18761 }, { "epoch": 1.0002665671482647, "grad_norm": 4.737714767456055, "learning_rate": 6.2478000000000006e-06, "loss": 0.0504, "step": 18762 }, { "epoch": 1.0003198805779177, "grad_norm": 1.0247523784637451, "learning_rate": 6.2476e-06, "loss": 0.0064, "step": 18763 }, { "epoch": 1.0003731940075704, "grad_norm": 2.4672586917877197, "learning_rate": 6.2474000000000005e-06, "loss": 0.0354, "step": 18764 }, { "epoch": 1.0004265074372234, "grad_norm": 1.2767040729522705, "learning_rate": 6.247200000000001e-06, "loss": 0.0244, "step": 18765 }, { "epoch": 1.0004798208668764, "grad_norm": 2.6254758834838867, "learning_rate": 6.2470000000000005e-06, "loss": 0.023, "step": 18766 }, { "epoch": 1.0005331342965293, "grad_norm": 2.7831594944000244, "learning_rate": 6.246800000000001e-06, "loss": 0.0417, "step": 18767 }, { "epoch": 1.0005864477261823, "grad_norm": 1.7980225086212158, "learning_rate": 6.2466000000000005e-06, "loss": 0.0316, "step": 18768 }, { "epoch": 1.000639761155835, "grad_norm": 0.9943082332611084, "learning_rate": 6.2464e-06, "loss": 0.0139, "step": 18769 }, { "epoch": 1.000693074585488, "grad_norm": 2.8596436977386475, "learning_rate": 6.2462000000000004e-06, "loss": 0.0242, "step": 18770 }, { "epoch": 1.000746388015141, "grad_norm": 1.2213064432144165, "learning_rate": 6.246000000000001e-06, "loss": 0.0265, "step": 18771 }, { "epoch": 1.000799701444794, "grad_norm": 2.0188961029052734, "learning_rate": 6.2458e-06, "loss": 0.019, "step": 18772 }, { "epoch": 1.0008530148744468, "grad_norm": 2.067547082901001, "learning_rate": 6.245600000000001e-06, "loss": 0.0579, "step": 18773 }, { "epoch": 1.0009063283040998, "grad_norm": 6.696518421173096, "learning_rate": 6.2454e-06, "loss": 0.0721, "step": 18774 }, { "epoch": 1.0009596417337527, "grad_norm": 2.3981194496154785, "learning_rate": 6.2452e-06, "loss": 0.0464, "step": 18775 }, { "epoch": 1.0010129551634057, "grad_norm": 2.8238704204559326, "learning_rate": 6.245000000000001e-06, "loss": 0.0113, "step": 18776 }, { "epoch": 1.0010662685930587, "grad_norm": 2.314422845840454, "learning_rate": 6.244800000000001e-06, "loss": 0.0285, "step": 18777 }, { "epoch": 1.0011195820227115, "grad_norm": 2.1582329273223877, "learning_rate": 6.2446e-06, "loss": 0.0608, "step": 18778 }, { "epoch": 1.0011728954523644, "grad_norm": 2.1855499744415283, "learning_rate": 6.244400000000001e-06, "loss": 0.0637, "step": 18779 }, { "epoch": 1.0012262088820174, "grad_norm": 2.759437084197998, "learning_rate": 6.2442e-06, "loss": 0.0345, "step": 18780 }, { "epoch": 1.0012795223116704, "grad_norm": 1.9725291728973389, "learning_rate": 6.244e-06, "loss": 0.0189, "step": 18781 }, { "epoch": 1.0013328357413231, "grad_norm": 0.8861587643623352, "learning_rate": 6.243800000000001e-06, "loss": 0.0119, "step": 18782 }, { "epoch": 1.0013861491709761, "grad_norm": 1.218669056892395, "learning_rate": 6.243600000000001e-06, "loss": 0.0176, "step": 18783 }, { "epoch": 1.001439462600629, "grad_norm": 2.290699005126953, "learning_rate": 6.2434e-06, "loss": 0.0233, "step": 18784 }, { "epoch": 1.001492776030282, "grad_norm": 2.9315104484558105, "learning_rate": 6.243200000000001e-06, "loss": 0.0567, "step": 18785 }, { "epoch": 1.001546089459935, "grad_norm": 2.8246917724609375, "learning_rate": 6.243e-06, "loss": 0.021, "step": 18786 }, { "epoch": 1.0015994028895878, "grad_norm": 1.8849042654037476, "learning_rate": 6.2428e-06, "loss": 0.0198, "step": 18787 }, { "epoch": 1.0016527163192408, "grad_norm": 1.3990541696548462, "learning_rate": 6.242600000000001e-06, "loss": 0.0395, "step": 18788 }, { "epoch": 1.0017060297488938, "grad_norm": 0.7880586385726929, "learning_rate": 6.2424000000000006e-06, "loss": 0.0158, "step": 18789 }, { "epoch": 1.0017593431785468, "grad_norm": 1.5399205684661865, "learning_rate": 6.2422e-06, "loss": 0.034, "step": 18790 }, { "epoch": 1.0018126566081995, "grad_norm": 1.14618980884552, "learning_rate": 6.2420000000000005e-06, "loss": 0.0239, "step": 18791 }, { "epoch": 1.0018659700378525, "grad_norm": 1.2542320489883423, "learning_rate": 6.2418e-06, "loss": 0.0177, "step": 18792 }, { "epoch": 1.0019192834675055, "grad_norm": 1.166957139968872, "learning_rate": 6.241600000000001e-06, "loss": 0.022, "step": 18793 }, { "epoch": 1.0019725968971585, "grad_norm": 2.3389768600463867, "learning_rate": 6.241400000000001e-06, "loss": 0.0189, "step": 18794 }, { "epoch": 1.0020259103268114, "grad_norm": 2.0342214107513428, "learning_rate": 6.2412000000000005e-06, "loss": 0.019, "step": 18795 }, { "epoch": 1.0020792237564642, "grad_norm": 0.7756534814834595, "learning_rate": 6.241e-06, "loss": 0.0163, "step": 18796 }, { "epoch": 1.0021325371861172, "grad_norm": 2.1209840774536133, "learning_rate": 6.2408000000000005e-06, "loss": 0.0418, "step": 18797 }, { "epoch": 1.0021858506157701, "grad_norm": 0.9919593930244446, "learning_rate": 6.2406e-06, "loss": 0.011, "step": 18798 }, { "epoch": 1.0022391640454231, "grad_norm": 1.9178873300552368, "learning_rate": 6.240400000000001e-06, "loss": 0.0168, "step": 18799 }, { "epoch": 1.0022924774750759, "grad_norm": 1.8433496952056885, "learning_rate": 6.240200000000001e-06, "loss": 0.0076, "step": 18800 }, { "epoch": 1.0023457909047289, "grad_norm": 1.6969096660614014, "learning_rate": 6.24e-06, "loss": 0.0352, "step": 18801 }, { "epoch": 1.0023991043343818, "grad_norm": 1.5124926567077637, "learning_rate": 6.2398e-06, "loss": 0.0187, "step": 18802 }, { "epoch": 1.0024524177640348, "grad_norm": 1.598617434501648, "learning_rate": 6.2396e-06, "loss": 0.0138, "step": 18803 }, { "epoch": 1.0025057311936878, "grad_norm": 0.7708116173744202, "learning_rate": 6.239400000000001e-06, "loss": 0.0071, "step": 18804 }, { "epoch": 1.0025590446233406, "grad_norm": 1.6778725385665894, "learning_rate": 6.239200000000001e-06, "loss": 0.0244, "step": 18805 }, { "epoch": 1.0026123580529935, "grad_norm": 3.753483295440674, "learning_rate": 6.239000000000001e-06, "loss": 0.0383, "step": 18806 }, { "epoch": 1.0026656714826465, "grad_norm": 1.7004203796386719, "learning_rate": 6.2388e-06, "loss": 0.0218, "step": 18807 }, { "epoch": 1.0027189849122995, "grad_norm": 1.2023817300796509, "learning_rate": 6.2386e-06, "loss": 0.0258, "step": 18808 }, { "epoch": 1.0027722983419523, "grad_norm": 2.0778768062591553, "learning_rate": 6.2384e-06, "loss": 0.0139, "step": 18809 }, { "epoch": 1.0028256117716052, "grad_norm": 1.5517640113830566, "learning_rate": 6.238200000000001e-06, "loss": 0.036, "step": 18810 }, { "epoch": 1.0028789252012582, "grad_norm": 1.0212581157684326, "learning_rate": 6.238000000000001e-06, "loss": 0.0065, "step": 18811 }, { "epoch": 1.0029322386309112, "grad_norm": 1.4342806339263916, "learning_rate": 6.237800000000001e-06, "loss": 0.0187, "step": 18812 }, { "epoch": 1.0029855520605642, "grad_norm": 1.4496562480926514, "learning_rate": 6.2376e-06, "loss": 0.023, "step": 18813 }, { "epoch": 1.003038865490217, "grad_norm": 3.246711492538452, "learning_rate": 6.2374e-06, "loss": 0.0419, "step": 18814 }, { "epoch": 1.00309217891987, "grad_norm": 1.8849549293518066, "learning_rate": 6.237200000000001e-06, "loss": 0.0444, "step": 18815 }, { "epoch": 1.0031454923495229, "grad_norm": 2.3505969047546387, "learning_rate": 6.237000000000001e-06, "loss": 0.0679, "step": 18816 }, { "epoch": 1.0031988057791759, "grad_norm": 1.1868126392364502, "learning_rate": 6.236800000000001e-06, "loss": 0.0199, "step": 18817 }, { "epoch": 1.0032521192088286, "grad_norm": 1.6608951091766357, "learning_rate": 6.2366000000000006e-06, "loss": 0.0259, "step": 18818 }, { "epoch": 1.0033054326384816, "grad_norm": 1.9220976829528809, "learning_rate": 6.2364e-06, "loss": 0.0299, "step": 18819 }, { "epoch": 1.0033587460681346, "grad_norm": 2.264868974685669, "learning_rate": 6.2362e-06, "loss": 0.0434, "step": 18820 }, { "epoch": 1.0034120594977876, "grad_norm": 1.5236120223999023, "learning_rate": 6.236000000000001e-06, "loss": 0.0179, "step": 18821 }, { "epoch": 1.0034653729274403, "grad_norm": 2.9090614318847656, "learning_rate": 6.2358000000000005e-06, "loss": 0.0257, "step": 18822 }, { "epoch": 1.0035186863570933, "grad_norm": 1.4255281686782837, "learning_rate": 6.235600000000001e-06, "loss": 0.025, "step": 18823 }, { "epoch": 1.0035719997867463, "grad_norm": 2.0957934856414795, "learning_rate": 6.2354000000000005e-06, "loss": 0.0255, "step": 18824 }, { "epoch": 1.0036253132163993, "grad_norm": 2.667201280593872, "learning_rate": 6.2352e-06, "loss": 0.0291, "step": 18825 }, { "epoch": 1.0036786266460522, "grad_norm": 2.341653823852539, "learning_rate": 6.235000000000001e-06, "loss": 0.0195, "step": 18826 }, { "epoch": 1.003731940075705, "grad_norm": 3.3467893600463867, "learning_rate": 6.234800000000001e-06, "loss": 0.0254, "step": 18827 }, { "epoch": 1.003785253505358, "grad_norm": 1.2249995470046997, "learning_rate": 6.2346e-06, "loss": 0.0102, "step": 18828 }, { "epoch": 1.003838566935011, "grad_norm": 1.234688401222229, "learning_rate": 6.234400000000001e-06, "loss": 0.0145, "step": 18829 }, { "epoch": 1.003891880364664, "grad_norm": 2.155494451522827, "learning_rate": 6.2342e-06, "loss": 0.0462, "step": 18830 }, { "epoch": 1.0039451937943167, "grad_norm": 2.6140060424804688, "learning_rate": 6.234e-06, "loss": 0.0276, "step": 18831 }, { "epoch": 1.0039985072239697, "grad_norm": 1.8125131130218506, "learning_rate": 6.233800000000001e-06, "loss": 0.0239, "step": 18832 }, { "epoch": 1.0040518206536226, "grad_norm": 2.0513393878936768, "learning_rate": 6.233600000000001e-06, "loss": 0.0147, "step": 18833 }, { "epoch": 1.0041051340832756, "grad_norm": 1.1689871549606323, "learning_rate": 6.2334e-06, "loss": 0.0061, "step": 18834 }, { "epoch": 1.0041584475129286, "grad_norm": 1.6881003379821777, "learning_rate": 6.233200000000001e-06, "loss": 0.0487, "step": 18835 }, { "epoch": 1.0042117609425814, "grad_norm": 1.8582899570465088, "learning_rate": 6.233e-06, "loss": 0.0458, "step": 18836 }, { "epoch": 1.0042650743722343, "grad_norm": 1.9968764781951904, "learning_rate": 6.2328e-06, "loss": 0.0391, "step": 18837 }, { "epoch": 1.0043183878018873, "grad_norm": 2.470893621444702, "learning_rate": 6.232600000000001e-06, "loss": 0.0415, "step": 18838 }, { "epoch": 1.0043717012315403, "grad_norm": 2.0036768913269043, "learning_rate": 6.232400000000001e-06, "loss": 0.0226, "step": 18839 }, { "epoch": 1.004425014661193, "grad_norm": 1.6934404373168945, "learning_rate": 6.2322e-06, "loss": 0.0228, "step": 18840 }, { "epoch": 1.004478328090846, "grad_norm": 1.5094492435455322, "learning_rate": 6.232000000000001e-06, "loss": 0.0149, "step": 18841 }, { "epoch": 1.004531641520499, "grad_norm": 1.5243921279907227, "learning_rate": 6.2318e-06, "loss": 0.0203, "step": 18842 }, { "epoch": 1.004584954950152, "grad_norm": 2.8970799446105957, "learning_rate": 6.231600000000001e-06, "loss": 0.0662, "step": 18843 }, { "epoch": 1.004638268379805, "grad_norm": 1.139479398727417, "learning_rate": 6.231400000000001e-06, "loss": 0.0152, "step": 18844 }, { "epoch": 1.0046915818094577, "grad_norm": 2.097088098526001, "learning_rate": 6.231200000000001e-06, "loss": 0.0168, "step": 18845 }, { "epoch": 1.0047448952391107, "grad_norm": 1.9041792154312134, "learning_rate": 6.231e-06, "loss": 0.0244, "step": 18846 }, { "epoch": 1.0047982086687637, "grad_norm": 3.185910224914551, "learning_rate": 6.2308000000000006e-06, "loss": 0.0326, "step": 18847 }, { "epoch": 1.0048515220984167, "grad_norm": 4.27756929397583, "learning_rate": 6.2306e-06, "loss": 0.0273, "step": 18848 }, { "epoch": 1.0049048355280694, "grad_norm": 2.3984146118164062, "learning_rate": 6.2304000000000005e-06, "loss": 0.0288, "step": 18849 }, { "epoch": 1.0049581489577224, "grad_norm": 2.6861283779144287, "learning_rate": 6.230200000000001e-06, "loss": 0.0233, "step": 18850 }, { "epoch": 1.0050114623873754, "grad_norm": 4.413354396820068, "learning_rate": 6.2300000000000005e-06, "loss": 0.0415, "step": 18851 }, { "epoch": 1.0050647758170284, "grad_norm": 1.743719458580017, "learning_rate": 6.2298e-06, "loss": 0.0219, "step": 18852 }, { "epoch": 1.0051180892466813, "grad_norm": 1.946146011352539, "learning_rate": 6.2296000000000005e-06, "loss": 0.0172, "step": 18853 }, { "epoch": 1.005171402676334, "grad_norm": 1.4990168809890747, "learning_rate": 6.229400000000001e-06, "loss": 0.0246, "step": 18854 }, { "epoch": 1.005224716105987, "grad_norm": 1.61243736743927, "learning_rate": 6.2292000000000004e-06, "loss": 0.0282, "step": 18855 }, { "epoch": 1.00527802953564, "grad_norm": 2.887117624282837, "learning_rate": 6.229000000000001e-06, "loss": 0.0454, "step": 18856 }, { "epoch": 1.005331342965293, "grad_norm": 0.9890283346176147, "learning_rate": 6.2288e-06, "loss": 0.0082, "step": 18857 }, { "epoch": 1.0053846563949458, "grad_norm": 3.442436695098877, "learning_rate": 6.2286e-06, "loss": 0.0313, "step": 18858 }, { "epoch": 1.0054379698245988, "grad_norm": 2.5890185832977295, "learning_rate": 6.2284e-06, "loss": 0.0203, "step": 18859 }, { "epoch": 1.0054912832542517, "grad_norm": 1.9751336574554443, "learning_rate": 6.228200000000001e-06, "loss": 0.0153, "step": 18860 }, { "epoch": 1.0055445966839047, "grad_norm": 0.6490930914878845, "learning_rate": 6.228e-06, "loss": 0.0081, "step": 18861 }, { "epoch": 1.0055979101135577, "grad_norm": 1.3581652641296387, "learning_rate": 6.227800000000001e-06, "loss": 0.0256, "step": 18862 }, { "epoch": 1.0056512235432105, "grad_norm": 5.21230936050415, "learning_rate": 6.2276e-06, "loss": 0.0497, "step": 18863 }, { "epoch": 1.0057045369728634, "grad_norm": 3.830244779586792, "learning_rate": 6.2274e-06, "loss": 0.0287, "step": 18864 }, { "epoch": 1.0057578504025164, "grad_norm": 2.564417600631714, "learning_rate": 6.227200000000001e-06, "loss": 0.0591, "step": 18865 }, { "epoch": 1.0058111638321694, "grad_norm": 1.5817984342575073, "learning_rate": 6.227000000000001e-06, "loss": 0.0251, "step": 18866 }, { "epoch": 1.0058644772618222, "grad_norm": 1.6730016469955444, "learning_rate": 6.2268e-06, "loss": 0.0216, "step": 18867 }, { "epoch": 1.0059177906914751, "grad_norm": 3.8968193531036377, "learning_rate": 6.226600000000001e-06, "loss": 0.0367, "step": 18868 }, { "epoch": 1.0059711041211281, "grad_norm": 1.9511082172393799, "learning_rate": 6.2264e-06, "loss": 0.0383, "step": 18869 }, { "epoch": 1.006024417550781, "grad_norm": 2.6001570224761963, "learning_rate": 6.2262e-06, "loss": 0.0311, "step": 18870 }, { "epoch": 1.006077730980434, "grad_norm": 1.1055246591567993, "learning_rate": 6.226000000000001e-06, "loss": 0.0161, "step": 18871 }, { "epoch": 1.0061310444100868, "grad_norm": 1.1127662658691406, "learning_rate": 6.225800000000001e-06, "loss": 0.0181, "step": 18872 }, { "epoch": 1.0061843578397398, "grad_norm": 2.2922427654266357, "learning_rate": 6.2256e-06, "loss": 0.0405, "step": 18873 }, { "epoch": 1.0062376712693928, "grad_norm": 1.5722167491912842, "learning_rate": 6.2254000000000006e-06, "loss": 0.0201, "step": 18874 }, { "epoch": 1.0062909846990458, "grad_norm": 1.6433049440383911, "learning_rate": 6.2252e-06, "loss": 0.0316, "step": 18875 }, { "epoch": 1.0063442981286985, "grad_norm": 1.7468830347061157, "learning_rate": 6.225000000000001e-06, "loss": 0.0326, "step": 18876 }, { "epoch": 1.0063976115583515, "grad_norm": 1.7134193181991577, "learning_rate": 6.224800000000001e-06, "loss": 0.0247, "step": 18877 }, { "epoch": 1.0064509249880045, "grad_norm": 0.7424091100692749, "learning_rate": 6.2246000000000005e-06, "loss": 0.0088, "step": 18878 }, { "epoch": 1.0065042384176575, "grad_norm": 1.241214394569397, "learning_rate": 6.2244e-06, "loss": 0.0119, "step": 18879 }, { "epoch": 1.0065575518473104, "grad_norm": 2.0169055461883545, "learning_rate": 6.2242000000000005e-06, "loss": 0.0517, "step": 18880 }, { "epoch": 1.0066108652769632, "grad_norm": 1.8221652507781982, "learning_rate": 6.224e-06, "loss": 0.024, "step": 18881 }, { "epoch": 1.0066641787066162, "grad_norm": 2.0273404121398926, "learning_rate": 6.223800000000001e-06, "loss": 0.0323, "step": 18882 }, { "epoch": 1.0067174921362692, "grad_norm": 1.8265807628631592, "learning_rate": 6.223600000000001e-06, "loss": 0.0239, "step": 18883 }, { "epoch": 1.0067708055659221, "grad_norm": 2.2286863327026367, "learning_rate": 6.2234e-06, "loss": 0.0357, "step": 18884 }, { "epoch": 1.006824118995575, "grad_norm": 0.8650333285331726, "learning_rate": 6.2232e-06, "loss": 0.0225, "step": 18885 }, { "epoch": 1.0068774324252279, "grad_norm": 1.3802618980407715, "learning_rate": 6.223e-06, "loss": 0.0156, "step": 18886 }, { "epoch": 1.0069307458548808, "grad_norm": 1.9830420017242432, "learning_rate": 6.222800000000001e-06, "loss": 0.0333, "step": 18887 }, { "epoch": 1.0069840592845338, "grad_norm": 3.4843714237213135, "learning_rate": 6.222600000000001e-06, "loss": 0.0167, "step": 18888 }, { "epoch": 1.0070373727141868, "grad_norm": 1.3813059329986572, "learning_rate": 6.222400000000001e-06, "loss": 0.0135, "step": 18889 }, { "epoch": 1.0070906861438396, "grad_norm": 1.7802350521087646, "learning_rate": 6.2222e-06, "loss": 0.0387, "step": 18890 }, { "epoch": 1.0071439995734925, "grad_norm": 1.3157198429107666, "learning_rate": 6.222e-06, "loss": 0.0281, "step": 18891 }, { "epoch": 1.0071973130031455, "grad_norm": 1.6324325799942017, "learning_rate": 6.2218e-06, "loss": 0.0405, "step": 18892 }, { "epoch": 1.0072506264327985, "grad_norm": 1.9827966690063477, "learning_rate": 6.221600000000001e-06, "loss": 0.0256, "step": 18893 }, { "epoch": 1.0073039398624513, "grad_norm": 2.2094576358795166, "learning_rate": 6.221400000000001e-06, "loss": 0.0296, "step": 18894 }, { "epoch": 1.0073572532921042, "grad_norm": 3.0982143878936768, "learning_rate": 6.221200000000001e-06, "loss": 0.0434, "step": 18895 }, { "epoch": 1.0074105667217572, "grad_norm": 2.029932737350464, "learning_rate": 6.221e-06, "loss": 0.0218, "step": 18896 }, { "epoch": 1.0074638801514102, "grad_norm": 2.0754966735839844, "learning_rate": 6.2208e-06, "loss": 0.0509, "step": 18897 }, { "epoch": 1.0075171935810632, "grad_norm": 1.2026312351226807, "learning_rate": 6.2206e-06, "loss": 0.012, "step": 18898 }, { "epoch": 1.007570507010716, "grad_norm": 3.054216146469116, "learning_rate": 6.220400000000001e-06, "loss": 0.0456, "step": 18899 }, { "epoch": 1.007623820440369, "grad_norm": 2.921964406967163, "learning_rate": 6.220200000000001e-06, "loss": 0.046, "step": 18900 }, { "epoch": 1.007677133870022, "grad_norm": 2.4869163036346436, "learning_rate": 6.220000000000001e-06, "loss": 0.0534, "step": 18901 }, { "epoch": 1.0077304472996749, "grad_norm": 3.069117784500122, "learning_rate": 6.2198e-06, "loss": 0.0261, "step": 18902 }, { "epoch": 1.0077837607293276, "grad_norm": 0.8975605368614197, "learning_rate": 6.2196e-06, "loss": 0.0084, "step": 18903 }, { "epoch": 1.0078370741589806, "grad_norm": 2.3001129627227783, "learning_rate": 6.219400000000001e-06, "loss": 0.0395, "step": 18904 }, { "epoch": 1.0078903875886336, "grad_norm": 1.876330852508545, "learning_rate": 6.2192000000000005e-06, "loss": 0.0178, "step": 18905 }, { "epoch": 1.0079437010182866, "grad_norm": 1.197977066040039, "learning_rate": 6.219000000000001e-06, "loss": 0.0169, "step": 18906 }, { "epoch": 1.0079970144479395, "grad_norm": 1.6095067262649536, "learning_rate": 6.2188000000000005e-06, "loss": 0.0202, "step": 18907 }, { "epoch": 1.0080503278775923, "grad_norm": 3.1504883766174316, "learning_rate": 6.2186e-06, "loss": 0.0563, "step": 18908 }, { "epoch": 1.0081036413072453, "grad_norm": 1.1213146448135376, "learning_rate": 6.2184e-06, "loss": 0.0187, "step": 18909 }, { "epoch": 1.0081569547368983, "grad_norm": 2.720202922821045, "learning_rate": 6.218200000000001e-06, "loss": 0.0339, "step": 18910 }, { "epoch": 1.0082102681665512, "grad_norm": 2.564831256866455, "learning_rate": 6.2180000000000004e-06, "loss": 0.0151, "step": 18911 }, { "epoch": 1.008263581596204, "grad_norm": 1.4735645055770874, "learning_rate": 6.217800000000001e-06, "loss": 0.0163, "step": 18912 }, { "epoch": 1.008316895025857, "grad_norm": 4.017642498016357, "learning_rate": 6.2176e-06, "loss": 0.0634, "step": 18913 }, { "epoch": 1.00837020845551, "grad_norm": 1.6382989883422852, "learning_rate": 6.2174e-06, "loss": 0.0254, "step": 18914 }, { "epoch": 1.008423521885163, "grad_norm": 1.2587507963180542, "learning_rate": 6.217200000000001e-06, "loss": 0.0312, "step": 18915 }, { "epoch": 1.008476835314816, "grad_norm": 4.793339252471924, "learning_rate": 6.217000000000001e-06, "loss": 0.0316, "step": 18916 }, { "epoch": 1.0085301487444687, "grad_norm": 2.4362456798553467, "learning_rate": 6.2168e-06, "loss": 0.0338, "step": 18917 }, { "epoch": 1.0085834621741216, "grad_norm": 2.684250593185425, "learning_rate": 6.216600000000001e-06, "loss": 0.017, "step": 18918 }, { "epoch": 1.0086367756037746, "grad_norm": 1.7374451160430908, "learning_rate": 6.2164e-06, "loss": 0.0384, "step": 18919 }, { "epoch": 1.0086900890334276, "grad_norm": 1.5841821432113647, "learning_rate": 6.2162e-06, "loss": 0.018, "step": 18920 }, { "epoch": 1.0087434024630804, "grad_norm": 1.0053256750106812, "learning_rate": 6.216000000000001e-06, "loss": 0.014, "step": 18921 }, { "epoch": 1.0087967158927333, "grad_norm": 4.89365816116333, "learning_rate": 6.215800000000001e-06, "loss": 0.0789, "step": 18922 }, { "epoch": 1.0088500293223863, "grad_norm": 2.9116809368133545, "learning_rate": 6.2156e-06, "loss": 0.0411, "step": 18923 }, { "epoch": 1.0089033427520393, "grad_norm": 4.011569976806641, "learning_rate": 6.215400000000001e-06, "loss": 0.0498, "step": 18924 }, { "epoch": 1.008956656181692, "grad_norm": 1.0211663246154785, "learning_rate": 6.2152e-06, "loss": 0.0215, "step": 18925 }, { "epoch": 1.009009969611345, "grad_norm": 2.024768590927124, "learning_rate": 6.215000000000001e-06, "loss": 0.0143, "step": 18926 }, { "epoch": 1.009063283040998, "grad_norm": 1.4560550451278687, "learning_rate": 6.214800000000001e-06, "loss": 0.0227, "step": 18927 }, { "epoch": 1.009116596470651, "grad_norm": 1.9261759519577026, "learning_rate": 6.214600000000001e-06, "loss": 0.031, "step": 18928 }, { "epoch": 1.009169909900304, "grad_norm": 1.5286165475845337, "learning_rate": 6.2144e-06, "loss": 0.0091, "step": 18929 }, { "epoch": 1.0092232233299567, "grad_norm": 3.23222017288208, "learning_rate": 6.2142000000000006e-06, "loss": 0.0473, "step": 18930 }, { "epoch": 1.0092765367596097, "grad_norm": 2.617957830429077, "learning_rate": 6.214e-06, "loss": 0.0422, "step": 18931 }, { "epoch": 1.0093298501892627, "grad_norm": 2.133674383163452, "learning_rate": 6.2138000000000005e-06, "loss": 0.0108, "step": 18932 }, { "epoch": 1.0093831636189157, "grad_norm": 4.275647163391113, "learning_rate": 6.213600000000001e-06, "loss": 0.0416, "step": 18933 }, { "epoch": 1.0094364770485684, "grad_norm": 2.277346134185791, "learning_rate": 6.2134000000000005e-06, "loss": 0.0361, "step": 18934 }, { "epoch": 1.0094897904782214, "grad_norm": 1.5655678510665894, "learning_rate": 6.2132e-06, "loss": 0.0138, "step": 18935 }, { "epoch": 1.0095431039078744, "grad_norm": 2.296273946762085, "learning_rate": 6.2130000000000005e-06, "loss": 0.026, "step": 18936 }, { "epoch": 1.0095964173375274, "grad_norm": 3.409088373184204, "learning_rate": 6.212800000000001e-06, "loss": 0.0259, "step": 18937 }, { "epoch": 1.0096497307671803, "grad_norm": 1.008124828338623, "learning_rate": 6.2126000000000005e-06, "loss": 0.0151, "step": 18938 }, { "epoch": 1.009703044196833, "grad_norm": 3.689399003982544, "learning_rate": 6.212400000000001e-06, "loss": 0.0332, "step": 18939 }, { "epoch": 1.009756357626486, "grad_norm": 1.4829360246658325, "learning_rate": 6.2122000000000004e-06, "loss": 0.0105, "step": 18940 }, { "epoch": 1.009809671056139, "grad_norm": 2.1673102378845215, "learning_rate": 6.212e-06, "loss": 0.0304, "step": 18941 }, { "epoch": 1.009862984485792, "grad_norm": 1.582536220550537, "learning_rate": 6.2118e-06, "loss": 0.0228, "step": 18942 }, { "epoch": 1.0099162979154448, "grad_norm": 1.0681830644607544, "learning_rate": 6.211600000000001e-06, "loss": 0.0165, "step": 18943 }, { "epoch": 1.0099696113450978, "grad_norm": 1.1844391822814941, "learning_rate": 6.2114e-06, "loss": 0.0242, "step": 18944 }, { "epoch": 1.0100229247747508, "grad_norm": 2.098618745803833, "learning_rate": 6.211200000000001e-06, "loss": 0.0386, "step": 18945 }, { "epoch": 1.0100762382044037, "grad_norm": 1.991524577140808, "learning_rate": 6.211e-06, "loss": 0.0319, "step": 18946 }, { "epoch": 1.0101295516340567, "grad_norm": 1.6157255172729492, "learning_rate": 6.2108e-06, "loss": 0.0151, "step": 18947 }, { "epoch": 1.0101828650637095, "grad_norm": 1.4857887029647827, "learning_rate": 6.2106e-06, "loss": 0.0318, "step": 18948 }, { "epoch": 1.0102361784933624, "grad_norm": 1.80869460105896, "learning_rate": 6.210400000000001e-06, "loss": 0.0525, "step": 18949 }, { "epoch": 1.0102894919230154, "grad_norm": 2.1068620681762695, "learning_rate": 6.2102e-06, "loss": 0.0279, "step": 18950 }, { "epoch": 1.0103428053526684, "grad_norm": 1.5828375816345215, "learning_rate": 6.210000000000001e-06, "loss": 0.0242, "step": 18951 }, { "epoch": 1.0103961187823212, "grad_norm": 2.288661003112793, "learning_rate": 6.2098e-06, "loss": 0.0267, "step": 18952 }, { "epoch": 1.0104494322119741, "grad_norm": 2.838794708251953, "learning_rate": 6.2096e-06, "loss": 0.0334, "step": 18953 }, { "epoch": 1.0105027456416271, "grad_norm": 3.1529033184051514, "learning_rate": 6.209400000000001e-06, "loss": 0.0537, "step": 18954 }, { "epoch": 1.01055605907128, "grad_norm": 4.02100133895874, "learning_rate": 6.209200000000001e-06, "loss": 0.0476, "step": 18955 }, { "epoch": 1.010609372500933, "grad_norm": 2.681981086730957, "learning_rate": 6.209e-06, "loss": 0.0342, "step": 18956 }, { "epoch": 1.0106626859305858, "grad_norm": 2.1412644386291504, "learning_rate": 6.208800000000001e-06, "loss": 0.053, "step": 18957 }, { "epoch": 1.0107159993602388, "grad_norm": 1.1137175559997559, "learning_rate": 6.2086e-06, "loss": 0.0131, "step": 18958 }, { "epoch": 1.0107693127898918, "grad_norm": 0.8635106682777405, "learning_rate": 6.2084e-06, "loss": 0.0084, "step": 18959 }, { "epoch": 1.0108226262195448, "grad_norm": 1.6337966918945312, "learning_rate": 6.208200000000001e-06, "loss": 0.0424, "step": 18960 }, { "epoch": 1.0108759396491975, "grad_norm": 1.5319833755493164, "learning_rate": 6.2080000000000005e-06, "loss": 0.0351, "step": 18961 }, { "epoch": 1.0109292530788505, "grad_norm": 2.703003168106079, "learning_rate": 6.2078e-06, "loss": 0.034, "step": 18962 }, { "epoch": 1.0109825665085035, "grad_norm": 1.005561351776123, "learning_rate": 6.2076000000000005e-06, "loss": 0.009, "step": 18963 }, { "epoch": 1.0110358799381565, "grad_norm": 3.35398268699646, "learning_rate": 6.2074e-06, "loss": 0.0582, "step": 18964 }, { "epoch": 1.0110891933678094, "grad_norm": 2.2850425243377686, "learning_rate": 6.207200000000001e-06, "loss": 0.0252, "step": 18965 }, { "epoch": 1.0111425067974622, "grad_norm": 1.9753869771957397, "learning_rate": 6.207000000000001e-06, "loss": 0.0114, "step": 18966 }, { "epoch": 1.0111958202271152, "grad_norm": 1.7395987510681152, "learning_rate": 6.2068000000000004e-06, "loss": 0.0321, "step": 18967 }, { "epoch": 1.0112491336567682, "grad_norm": 1.0659375190734863, "learning_rate": 6.2066e-06, "loss": 0.0299, "step": 18968 }, { "epoch": 1.0113024470864211, "grad_norm": 1.5509155988693237, "learning_rate": 6.2064e-06, "loss": 0.0145, "step": 18969 }, { "epoch": 1.011355760516074, "grad_norm": 2.077152967453003, "learning_rate": 6.2062e-06, "loss": 0.0403, "step": 18970 }, { "epoch": 1.0114090739457269, "grad_norm": 2.414586067199707, "learning_rate": 6.206000000000001e-06, "loss": 0.0191, "step": 18971 }, { "epoch": 1.0114623873753799, "grad_norm": 1.7976527214050293, "learning_rate": 6.205800000000001e-06, "loss": 0.0292, "step": 18972 }, { "epoch": 1.0115157008050328, "grad_norm": 1.979550838470459, "learning_rate": 6.2056e-06, "loss": 0.0289, "step": 18973 }, { "epoch": 1.0115690142346858, "grad_norm": 4.095633029937744, "learning_rate": 6.2054e-06, "loss": 0.0311, "step": 18974 }, { "epoch": 1.0116223276643386, "grad_norm": 2.3495073318481445, "learning_rate": 6.2052e-06, "loss": 0.0182, "step": 18975 }, { "epoch": 1.0116756410939916, "grad_norm": 1.4670214653015137, "learning_rate": 6.205000000000001e-06, "loss": 0.0217, "step": 18976 }, { "epoch": 1.0117289545236445, "grad_norm": 1.4741383790969849, "learning_rate": 6.204800000000001e-06, "loss": 0.0086, "step": 18977 }, { "epoch": 1.0117822679532975, "grad_norm": 2.9469990730285645, "learning_rate": 6.204600000000001e-06, "loss": 0.0702, "step": 18978 }, { "epoch": 1.0118355813829503, "grad_norm": 1.220889687538147, "learning_rate": 6.2044e-06, "loss": 0.0245, "step": 18979 }, { "epoch": 1.0118888948126032, "grad_norm": 3.491581439971924, "learning_rate": 6.2042e-06, "loss": 0.0465, "step": 18980 }, { "epoch": 1.0119422082422562, "grad_norm": 1.0445005893707275, "learning_rate": 6.204e-06, "loss": 0.011, "step": 18981 }, { "epoch": 1.0119955216719092, "grad_norm": 0.9462077617645264, "learning_rate": 6.203800000000001e-06, "loss": 0.013, "step": 18982 }, { "epoch": 1.0120488351015622, "grad_norm": 2.0964407920837402, "learning_rate": 6.203600000000001e-06, "loss": 0.0409, "step": 18983 }, { "epoch": 1.012102148531215, "grad_norm": 0.5469460487365723, "learning_rate": 6.203400000000001e-06, "loss": 0.0079, "step": 18984 }, { "epoch": 1.012155461960868, "grad_norm": 1.406481146812439, "learning_rate": 6.2032e-06, "loss": 0.0177, "step": 18985 }, { "epoch": 1.012208775390521, "grad_norm": 1.5190856456756592, "learning_rate": 6.203e-06, "loss": 0.0204, "step": 18986 }, { "epoch": 1.0122620888201739, "grad_norm": 2.00148606300354, "learning_rate": 6.202800000000001e-06, "loss": 0.0195, "step": 18987 }, { "epoch": 1.0123154022498266, "grad_norm": 2.236440896987915, "learning_rate": 6.2026000000000006e-06, "loss": 0.0274, "step": 18988 }, { "epoch": 1.0123687156794796, "grad_norm": 2.708587169647217, "learning_rate": 6.202400000000001e-06, "loss": 0.0435, "step": 18989 }, { "epoch": 1.0124220291091326, "grad_norm": 2.0545778274536133, "learning_rate": 6.2022000000000005e-06, "loss": 0.0315, "step": 18990 }, { "epoch": 1.0124753425387856, "grad_norm": 2.160975933074951, "learning_rate": 6.202e-06, "loss": 0.0272, "step": 18991 }, { "epoch": 1.0125286559684386, "grad_norm": 0.6283487677574158, "learning_rate": 6.2018e-06, "loss": 0.0064, "step": 18992 }, { "epoch": 1.0125819693980913, "grad_norm": 1.287970781326294, "learning_rate": 6.201600000000001e-06, "loss": 0.0084, "step": 18993 }, { "epoch": 1.0126352828277443, "grad_norm": 1.2844054698944092, "learning_rate": 6.2014000000000005e-06, "loss": 0.0118, "step": 18994 }, { "epoch": 1.0126885962573973, "grad_norm": 1.351245641708374, "learning_rate": 6.201200000000001e-06, "loss": 0.016, "step": 18995 }, { "epoch": 1.0127419096870502, "grad_norm": 2.0582780838012695, "learning_rate": 6.2010000000000004e-06, "loss": 0.0381, "step": 18996 }, { "epoch": 1.012795223116703, "grad_norm": 2.5703494548797607, "learning_rate": 6.2008e-06, "loss": 0.0472, "step": 18997 }, { "epoch": 1.012848536546356, "grad_norm": 1.1609679460525513, "learning_rate": 6.2005999999999996e-06, "loss": 0.0256, "step": 18998 }, { "epoch": 1.012901849976009, "grad_norm": 2.94590163230896, "learning_rate": 6.200400000000001e-06, "loss": 0.0442, "step": 18999 }, { "epoch": 1.012955163405662, "grad_norm": 1.0499770641326904, "learning_rate": 6.2002e-06, "loss": 0.0085, "step": 19000 }, { "epoch": 1.013008476835315, "grad_norm": 2.059447765350342, "learning_rate": 6.200000000000001e-06, "loss": 0.0555, "step": 19001 }, { "epoch": 1.0130617902649677, "grad_norm": 2.066657781600952, "learning_rate": 6.1998e-06, "loss": 0.0138, "step": 19002 }, { "epoch": 1.0131151036946207, "grad_norm": 1.860310673713684, "learning_rate": 6.1996e-06, "loss": 0.0143, "step": 19003 }, { "epoch": 1.0131684171242736, "grad_norm": 1.976100206375122, "learning_rate": 6.199400000000001e-06, "loss": 0.0344, "step": 19004 }, { "epoch": 1.0132217305539266, "grad_norm": 3.696237802505493, "learning_rate": 6.199200000000001e-06, "loss": 0.0447, "step": 19005 }, { "epoch": 1.0132750439835794, "grad_norm": 1.7839901447296143, "learning_rate": 6.199e-06, "loss": 0.0453, "step": 19006 }, { "epoch": 1.0133283574132324, "grad_norm": 2.1644434928894043, "learning_rate": 6.198800000000001e-06, "loss": 0.0181, "step": 19007 }, { "epoch": 1.0133816708428853, "grad_norm": 1.763246774673462, "learning_rate": 6.1986e-06, "loss": 0.0113, "step": 19008 }, { "epoch": 1.0134349842725383, "grad_norm": 2.1378047466278076, "learning_rate": 6.1984e-06, "loss": 0.0387, "step": 19009 }, { "epoch": 1.0134882977021913, "grad_norm": 0.5001675486564636, "learning_rate": 6.198200000000001e-06, "loss": 0.0036, "step": 19010 }, { "epoch": 1.013541611131844, "grad_norm": 0.4559316337108612, "learning_rate": 6.198000000000001e-06, "loss": 0.005, "step": 19011 }, { "epoch": 1.013594924561497, "grad_norm": 1.8544424772262573, "learning_rate": 6.1978e-06, "loss": 0.0346, "step": 19012 }, { "epoch": 1.01364823799115, "grad_norm": 1.0634626150131226, "learning_rate": 6.197600000000001e-06, "loss": 0.0374, "step": 19013 }, { "epoch": 1.013701551420803, "grad_norm": 2.4195401668548584, "learning_rate": 6.1974e-06, "loss": 0.0478, "step": 19014 }, { "epoch": 1.0137548648504557, "grad_norm": 2.8712499141693115, "learning_rate": 6.197200000000001e-06, "loss": 0.0258, "step": 19015 }, { "epoch": 1.0138081782801087, "grad_norm": 2.0343034267425537, "learning_rate": 6.197000000000001e-06, "loss": 0.0328, "step": 19016 }, { "epoch": 1.0138614917097617, "grad_norm": 1.0819287300109863, "learning_rate": 6.1968000000000005e-06, "loss": 0.0192, "step": 19017 }, { "epoch": 1.0139148051394147, "grad_norm": 1.716367244720459, "learning_rate": 6.1966e-06, "loss": 0.0149, "step": 19018 }, { "epoch": 1.0139681185690677, "grad_norm": 1.819306492805481, "learning_rate": 6.1964000000000005e-06, "loss": 0.0369, "step": 19019 }, { "epoch": 1.0140214319987204, "grad_norm": 1.0452972650527954, "learning_rate": 6.1962e-06, "loss": 0.0129, "step": 19020 }, { "epoch": 1.0140747454283734, "grad_norm": 2.352531909942627, "learning_rate": 6.196000000000001e-06, "loss": 0.032, "step": 19021 }, { "epoch": 1.0141280588580264, "grad_norm": 1.874241590499878, "learning_rate": 6.195800000000001e-06, "loss": 0.0206, "step": 19022 }, { "epoch": 1.0141813722876794, "grad_norm": 3.850191354751587, "learning_rate": 6.1956000000000004e-06, "loss": 0.0284, "step": 19023 }, { "epoch": 1.014234685717332, "grad_norm": 0.8882931470870972, "learning_rate": 6.1954e-06, "loss": 0.015, "step": 19024 }, { "epoch": 1.014287999146985, "grad_norm": 1.774454116821289, "learning_rate": 6.1952e-06, "loss": 0.0136, "step": 19025 }, { "epoch": 1.014341312576638, "grad_norm": 1.0337669849395752, "learning_rate": 6.195000000000001e-06, "loss": 0.0136, "step": 19026 }, { "epoch": 1.014394626006291, "grad_norm": 0.929635226726532, "learning_rate": 6.194800000000001e-06, "loss": 0.0124, "step": 19027 }, { "epoch": 1.014447939435944, "grad_norm": 1.8703023195266724, "learning_rate": 6.194600000000001e-06, "loss": 0.0317, "step": 19028 }, { "epoch": 1.0145012528655968, "grad_norm": 0.9012182354927063, "learning_rate": 6.1944e-06, "loss": 0.0087, "step": 19029 }, { "epoch": 1.0145545662952498, "grad_norm": 1.27033269405365, "learning_rate": 6.1942e-06, "loss": 0.0425, "step": 19030 }, { "epoch": 1.0146078797249027, "grad_norm": 1.0683619976043701, "learning_rate": 6.194e-06, "loss": 0.0124, "step": 19031 }, { "epoch": 1.0146611931545557, "grad_norm": 1.137198805809021, "learning_rate": 6.193800000000001e-06, "loss": 0.0135, "step": 19032 }, { "epoch": 1.0147145065842085, "grad_norm": 2.0108516216278076, "learning_rate": 6.193600000000001e-06, "loss": 0.0233, "step": 19033 }, { "epoch": 1.0147678200138615, "grad_norm": 2.722691774368286, "learning_rate": 6.193400000000001e-06, "loss": 0.0489, "step": 19034 }, { "epoch": 1.0148211334435144, "grad_norm": 1.5996164083480835, "learning_rate": 6.1932e-06, "loss": 0.0309, "step": 19035 }, { "epoch": 1.0148744468731674, "grad_norm": 1.6281288862228394, "learning_rate": 6.193e-06, "loss": 0.0294, "step": 19036 }, { "epoch": 1.0149277603028204, "grad_norm": 1.7712100744247437, "learning_rate": 6.192800000000001e-06, "loss": 0.0118, "step": 19037 }, { "epoch": 1.0149810737324731, "grad_norm": 1.1265926361083984, "learning_rate": 6.192600000000001e-06, "loss": 0.0139, "step": 19038 }, { "epoch": 1.0150343871621261, "grad_norm": 2.0363266468048096, "learning_rate": 6.192400000000001e-06, "loss": 0.0396, "step": 19039 }, { "epoch": 1.015087700591779, "grad_norm": 2.0436816215515137, "learning_rate": 6.192200000000001e-06, "loss": 0.0185, "step": 19040 }, { "epoch": 1.015141014021432, "grad_norm": 1.4109630584716797, "learning_rate": 6.192e-06, "loss": 0.0245, "step": 19041 }, { "epoch": 1.0151943274510848, "grad_norm": 1.9177613258361816, "learning_rate": 6.1918e-06, "loss": 0.0137, "step": 19042 }, { "epoch": 1.0152476408807378, "grad_norm": 1.5078338384628296, "learning_rate": 6.191600000000001e-06, "loss": 0.0206, "step": 19043 }, { "epoch": 1.0153009543103908, "grad_norm": 3.759284496307373, "learning_rate": 6.1914000000000006e-06, "loss": 0.0385, "step": 19044 }, { "epoch": 1.0153542677400438, "grad_norm": 4.541367053985596, "learning_rate": 6.191200000000001e-06, "loss": 0.0327, "step": 19045 }, { "epoch": 1.0154075811696965, "grad_norm": 1.32133948802948, "learning_rate": 6.1910000000000005e-06, "loss": 0.0153, "step": 19046 }, { "epoch": 1.0154608945993495, "grad_norm": 0.7921387553215027, "learning_rate": 6.1908e-06, "loss": 0.0173, "step": 19047 }, { "epoch": 1.0155142080290025, "grad_norm": 2.4110817909240723, "learning_rate": 6.190600000000001e-06, "loss": 0.0488, "step": 19048 }, { "epoch": 1.0155675214586555, "grad_norm": 1.4309868812561035, "learning_rate": 6.190400000000001e-06, "loss": 0.0202, "step": 19049 }, { "epoch": 1.0156208348883085, "grad_norm": 1.9716565608978271, "learning_rate": 6.1902000000000005e-06, "loss": 0.0343, "step": 19050 }, { "epoch": 1.0156741483179612, "grad_norm": 0.978908360004425, "learning_rate": 6.190000000000001e-06, "loss": 0.0223, "step": 19051 }, { "epoch": 1.0157274617476142, "grad_norm": 1.469138264656067, "learning_rate": 6.1898000000000004e-06, "loss": 0.0235, "step": 19052 }, { "epoch": 1.0157807751772672, "grad_norm": 1.484931230545044, "learning_rate": 6.1896e-06, "loss": 0.0339, "step": 19053 }, { "epoch": 1.0158340886069201, "grad_norm": 2.960824966430664, "learning_rate": 6.189400000000001e-06, "loss": 0.0157, "step": 19054 }, { "epoch": 1.015887402036573, "grad_norm": 2.1279938220977783, "learning_rate": 6.189200000000001e-06, "loss": 0.0303, "step": 19055 }, { "epoch": 1.0159407154662259, "grad_norm": 0.6013657450675964, "learning_rate": 6.189e-06, "loss": 0.0082, "step": 19056 }, { "epoch": 1.0159940288958789, "grad_norm": 2.0916748046875, "learning_rate": 6.188800000000001e-06, "loss": 0.0512, "step": 19057 }, { "epoch": 1.0160473423255318, "grad_norm": 4.855214595794678, "learning_rate": 6.1886e-06, "loss": 0.037, "step": 19058 }, { "epoch": 1.0161006557551848, "grad_norm": 2.145151376724243, "learning_rate": 6.1884e-06, "loss": 0.0127, "step": 19059 }, { "epoch": 1.0161539691848376, "grad_norm": 0.9803447723388672, "learning_rate": 6.188200000000001e-06, "loss": 0.007, "step": 19060 }, { "epoch": 1.0162072826144906, "grad_norm": 1.5885900259017944, "learning_rate": 6.188000000000001e-06, "loss": 0.0159, "step": 19061 }, { "epoch": 1.0162605960441435, "grad_norm": 1.1403839588165283, "learning_rate": 6.1878e-06, "loss": 0.0185, "step": 19062 }, { "epoch": 1.0163139094737965, "grad_norm": 2.9059441089630127, "learning_rate": 6.187600000000001e-06, "loss": 0.0302, "step": 19063 }, { "epoch": 1.0163672229034493, "grad_norm": 4.024568557739258, "learning_rate": 6.1874e-06, "loss": 0.0434, "step": 19064 }, { "epoch": 1.0164205363331023, "grad_norm": 1.201703429222107, "learning_rate": 6.187200000000001e-06, "loss": 0.0288, "step": 19065 }, { "epoch": 1.0164738497627552, "grad_norm": 4.781032085418701, "learning_rate": 6.187000000000001e-06, "loss": 0.0525, "step": 19066 }, { "epoch": 1.0165271631924082, "grad_norm": 2.8785767555236816, "learning_rate": 6.186800000000001e-06, "loss": 0.0225, "step": 19067 }, { "epoch": 1.0165804766220612, "grad_norm": 0.6163713932037354, "learning_rate": 6.1866e-06, "loss": 0.0059, "step": 19068 }, { "epoch": 1.016633790051714, "grad_norm": 2.3215925693511963, "learning_rate": 6.186400000000001e-06, "loss": 0.0302, "step": 19069 }, { "epoch": 1.016687103481367, "grad_norm": 1.161137580871582, "learning_rate": 6.1862e-06, "loss": 0.0149, "step": 19070 }, { "epoch": 1.01674041691102, "grad_norm": 2.347653388977051, "learning_rate": 6.1860000000000006e-06, "loss": 0.0416, "step": 19071 }, { "epoch": 1.0167937303406729, "grad_norm": 1.2346981763839722, "learning_rate": 6.185800000000001e-06, "loss": 0.0211, "step": 19072 }, { "epoch": 1.0168470437703256, "grad_norm": 2.092837333679199, "learning_rate": 6.1856000000000005e-06, "loss": 0.0357, "step": 19073 }, { "epoch": 1.0169003571999786, "grad_norm": 2.1960275173187256, "learning_rate": 6.1854e-06, "loss": 0.0282, "step": 19074 }, { "epoch": 1.0169536706296316, "grad_norm": 1.9423726797103882, "learning_rate": 6.1852000000000005e-06, "loss": 0.0452, "step": 19075 }, { "epoch": 1.0170069840592846, "grad_norm": 2.659660577774048, "learning_rate": 6.185000000000001e-06, "loss": 0.0644, "step": 19076 }, { "epoch": 1.0170602974889376, "grad_norm": 2.537753105163574, "learning_rate": 6.1848000000000005e-06, "loss": 0.0487, "step": 19077 }, { "epoch": 1.0171136109185903, "grad_norm": 2.487879514694214, "learning_rate": 6.184600000000001e-06, "loss": 0.0122, "step": 19078 }, { "epoch": 1.0171669243482433, "grad_norm": 2.771477699279785, "learning_rate": 6.1844000000000005e-06, "loss": 0.0274, "step": 19079 }, { "epoch": 1.0172202377778963, "grad_norm": 3.574061632156372, "learning_rate": 6.1842e-06, "loss": 0.0283, "step": 19080 }, { "epoch": 1.0172735512075493, "grad_norm": 2.0443215370178223, "learning_rate": 6.184e-06, "loss": 0.0224, "step": 19081 }, { "epoch": 1.017326864637202, "grad_norm": 3.280078887939453, "learning_rate": 6.183800000000001e-06, "loss": 0.0301, "step": 19082 }, { "epoch": 1.017380178066855, "grad_norm": 1.245042085647583, "learning_rate": 6.1836e-06, "loss": 0.0099, "step": 19083 }, { "epoch": 1.017433491496508, "grad_norm": 1.6972038745880127, "learning_rate": 6.183400000000001e-06, "loss": 0.0264, "step": 19084 }, { "epoch": 1.017486804926161, "grad_norm": 3.970999002456665, "learning_rate": 6.1832e-06, "loss": 0.0442, "step": 19085 }, { "epoch": 1.017540118355814, "grad_norm": 2.6975631713867188, "learning_rate": 6.183e-06, "loss": 0.0239, "step": 19086 }, { "epoch": 1.0175934317854667, "grad_norm": 1.9774099588394165, "learning_rate": 6.182800000000001e-06, "loss": 0.0364, "step": 19087 }, { "epoch": 1.0176467452151197, "grad_norm": 1.7196359634399414, "learning_rate": 6.182600000000001e-06, "loss": 0.0214, "step": 19088 }, { "epoch": 1.0177000586447726, "grad_norm": 1.7906873226165771, "learning_rate": 6.1824e-06, "loss": 0.0168, "step": 19089 }, { "epoch": 1.0177533720744256, "grad_norm": 2.061378240585327, "learning_rate": 6.182200000000001e-06, "loss": 0.0186, "step": 19090 }, { "epoch": 1.0178066855040784, "grad_norm": 2.56103515625, "learning_rate": 6.182e-06, "loss": 0.0426, "step": 19091 }, { "epoch": 1.0178599989337314, "grad_norm": 1.6742249727249146, "learning_rate": 6.1818e-06, "loss": 0.0381, "step": 19092 }, { "epoch": 1.0179133123633843, "grad_norm": 1.0459964275360107, "learning_rate": 6.181600000000001e-06, "loss": 0.0153, "step": 19093 }, { "epoch": 1.0179666257930373, "grad_norm": 1.9165778160095215, "learning_rate": 6.181400000000001e-06, "loss": 0.0314, "step": 19094 }, { "epoch": 1.0180199392226903, "grad_norm": 1.8116340637207031, "learning_rate": 6.1812e-06, "loss": 0.0178, "step": 19095 }, { "epoch": 1.018073252652343, "grad_norm": 2.5400376319885254, "learning_rate": 6.181000000000001e-06, "loss": 0.0358, "step": 19096 }, { "epoch": 1.018126566081996, "grad_norm": 1.6648119688034058, "learning_rate": 6.1808e-06, "loss": 0.0129, "step": 19097 }, { "epoch": 1.018179879511649, "grad_norm": 2.4755663871765137, "learning_rate": 6.1806000000000014e-06, "loss": 0.0174, "step": 19098 }, { "epoch": 1.018233192941302, "grad_norm": 2.3231987953186035, "learning_rate": 6.180400000000001e-06, "loss": 0.0063, "step": 19099 }, { "epoch": 1.0182865063709547, "grad_norm": 1.9242892265319824, "learning_rate": 6.1802000000000006e-06, "loss": 0.034, "step": 19100 }, { "epoch": 1.0183398198006077, "grad_norm": 1.5380685329437256, "learning_rate": 6.18e-06, "loss": 0.0099, "step": 19101 }, { "epoch": 1.0183931332302607, "grad_norm": 2.119929313659668, "learning_rate": 6.1798000000000005e-06, "loss": 0.0381, "step": 19102 }, { "epoch": 1.0184464466599137, "grad_norm": 2.1116976737976074, "learning_rate": 6.1796e-06, "loss": 0.0423, "step": 19103 }, { "epoch": 1.0184997600895667, "grad_norm": 1.1471061706542969, "learning_rate": 6.179400000000001e-06, "loss": 0.0132, "step": 19104 }, { "epoch": 1.0185530735192194, "grad_norm": 1.3115185499191284, "learning_rate": 6.179200000000001e-06, "loss": 0.0079, "step": 19105 }, { "epoch": 1.0186063869488724, "grad_norm": 1.576425552368164, "learning_rate": 6.1790000000000005e-06, "loss": 0.016, "step": 19106 }, { "epoch": 1.0186597003785254, "grad_norm": 2.3746566772460938, "learning_rate": 6.1788e-06, "loss": 0.0471, "step": 19107 }, { "epoch": 1.0187130138081784, "grad_norm": 1.3114206790924072, "learning_rate": 6.1786000000000004e-06, "loss": 0.0127, "step": 19108 }, { "epoch": 1.0187663272378311, "grad_norm": 0.8805758953094482, "learning_rate": 6.1784e-06, "loss": 0.0121, "step": 19109 }, { "epoch": 1.018819640667484, "grad_norm": 1.9075154066085815, "learning_rate": 6.178200000000001e-06, "loss": 0.0288, "step": 19110 }, { "epoch": 1.018872954097137, "grad_norm": 2.039342164993286, "learning_rate": 6.178000000000001e-06, "loss": 0.0409, "step": 19111 }, { "epoch": 1.01892626752679, "grad_norm": 1.4460952281951904, "learning_rate": 6.1778e-06, "loss": 0.0146, "step": 19112 }, { "epoch": 1.018979580956443, "grad_norm": 2.0730502605438232, "learning_rate": 6.1776e-06, "loss": 0.0201, "step": 19113 }, { "epoch": 1.0190328943860958, "grad_norm": 2.2349770069122314, "learning_rate": 6.1774e-06, "loss": 0.0296, "step": 19114 }, { "epoch": 1.0190862078157488, "grad_norm": 1.9743361473083496, "learning_rate": 6.177200000000001e-06, "loss": 0.0377, "step": 19115 }, { "epoch": 1.0191395212454017, "grad_norm": 2.101651668548584, "learning_rate": 6.177000000000001e-06, "loss": 0.0254, "step": 19116 }, { "epoch": 1.0191928346750547, "grad_norm": 1.9163321256637573, "learning_rate": 6.176800000000001e-06, "loss": 0.0316, "step": 19117 }, { "epoch": 1.0192461481047075, "grad_norm": 0.9193816781044006, "learning_rate": 6.1766e-06, "loss": 0.0125, "step": 19118 }, { "epoch": 1.0192994615343605, "grad_norm": 3.1430130004882812, "learning_rate": 6.1764e-06, "loss": 0.0232, "step": 19119 }, { "epoch": 1.0193527749640134, "grad_norm": 6.313593864440918, "learning_rate": 6.1762e-06, "loss": 0.0584, "step": 19120 }, { "epoch": 1.0194060883936664, "grad_norm": 1.6719332933425903, "learning_rate": 6.176000000000001e-06, "loss": 0.0217, "step": 19121 }, { "epoch": 1.0194594018233194, "grad_norm": 2.2052507400512695, "learning_rate": 6.175800000000001e-06, "loss": 0.0481, "step": 19122 }, { "epoch": 1.0195127152529722, "grad_norm": 0.9931357502937317, "learning_rate": 6.175600000000001e-06, "loss": 0.0168, "step": 19123 }, { "epoch": 1.0195660286826251, "grad_norm": 0.6758474111557007, "learning_rate": 6.1754e-06, "loss": 0.0049, "step": 19124 }, { "epoch": 1.0196193421122781, "grad_norm": 1.451275110244751, "learning_rate": 6.1752e-06, "loss": 0.0216, "step": 19125 }, { "epoch": 1.019672655541931, "grad_norm": 3.6167685985565186, "learning_rate": 6.175000000000001e-06, "loss": 0.0581, "step": 19126 }, { "epoch": 1.0197259689715839, "grad_norm": 2.4828603267669678, "learning_rate": 6.174800000000001e-06, "loss": 0.0439, "step": 19127 }, { "epoch": 1.0197792824012368, "grad_norm": 4.024209976196289, "learning_rate": 6.174600000000001e-06, "loss": 0.0649, "step": 19128 }, { "epoch": 1.0198325958308898, "grad_norm": 1.016382098197937, "learning_rate": 6.1744000000000005e-06, "loss": 0.0205, "step": 19129 }, { "epoch": 1.0198859092605428, "grad_norm": 1.2614625692367554, "learning_rate": 6.1742e-06, "loss": 0.0109, "step": 19130 }, { "epoch": 1.0199392226901958, "grad_norm": 1.6116002798080444, "learning_rate": 6.174e-06, "loss": 0.0224, "step": 19131 }, { "epoch": 1.0199925361198485, "grad_norm": 1.5359247922897339, "learning_rate": 6.173800000000001e-06, "loss": 0.0165, "step": 19132 }, { "epoch": 1.0200458495495015, "grad_norm": 1.1523929834365845, "learning_rate": 6.1736000000000005e-06, "loss": 0.0129, "step": 19133 }, { "epoch": 1.0200991629791545, "grad_norm": 2.1454837322235107, "learning_rate": 6.173400000000001e-06, "loss": 0.0335, "step": 19134 }, { "epoch": 1.0201524764088075, "grad_norm": 1.8697690963745117, "learning_rate": 6.1732000000000005e-06, "loss": 0.0276, "step": 19135 }, { "epoch": 1.0202057898384602, "grad_norm": 1.758256435394287, "learning_rate": 6.173e-06, "loss": 0.0347, "step": 19136 }, { "epoch": 1.0202591032681132, "grad_norm": 1.7460713386535645, "learning_rate": 6.172800000000001e-06, "loss": 0.0277, "step": 19137 }, { "epoch": 1.0203124166977662, "grad_norm": 2.0811965465545654, "learning_rate": 6.172600000000001e-06, "loss": 0.0275, "step": 19138 }, { "epoch": 1.0203657301274192, "grad_norm": 2.7497646808624268, "learning_rate": 6.1724e-06, "loss": 0.0199, "step": 19139 }, { "epoch": 1.0204190435570721, "grad_norm": 1.2718350887298584, "learning_rate": 6.172200000000001e-06, "loss": 0.0245, "step": 19140 }, { "epoch": 1.020472356986725, "grad_norm": 1.734845519065857, "learning_rate": 6.172e-06, "loss": 0.012, "step": 19141 }, { "epoch": 1.0205256704163779, "grad_norm": 2.436770439147949, "learning_rate": 6.1718e-06, "loss": 0.016, "step": 19142 }, { "epoch": 1.0205789838460309, "grad_norm": 0.8157676458358765, "learning_rate": 6.171600000000001e-06, "loss": 0.0077, "step": 19143 }, { "epoch": 1.0206322972756838, "grad_norm": 1.2394349575042725, "learning_rate": 6.171400000000001e-06, "loss": 0.0414, "step": 19144 }, { "epoch": 1.0206856107053366, "grad_norm": 2.546691656112671, "learning_rate": 6.1712e-06, "loss": 0.0491, "step": 19145 }, { "epoch": 1.0207389241349896, "grad_norm": 2.0726101398468018, "learning_rate": 6.171000000000001e-06, "loss": 0.027, "step": 19146 }, { "epoch": 1.0207922375646425, "grad_norm": 2.2666871547698975, "learning_rate": 6.1708e-06, "loss": 0.0298, "step": 19147 }, { "epoch": 1.0208455509942955, "grad_norm": 1.4598802328109741, "learning_rate": 6.170600000000001e-06, "loss": 0.0299, "step": 19148 }, { "epoch": 1.0208988644239483, "grad_norm": 1.6718374490737915, "learning_rate": 6.170400000000001e-06, "loss": 0.0168, "step": 19149 }, { "epoch": 1.0209521778536013, "grad_norm": 1.1979707479476929, "learning_rate": 6.170200000000001e-06, "loss": 0.0151, "step": 19150 }, { "epoch": 1.0210054912832542, "grad_norm": 2.16706919670105, "learning_rate": 6.17e-06, "loss": 0.0356, "step": 19151 }, { "epoch": 1.0210588047129072, "grad_norm": 1.3901556730270386, "learning_rate": 6.169800000000001e-06, "loss": 0.0263, "step": 19152 }, { "epoch": 1.0211121181425602, "grad_norm": 2.9896609783172607, "learning_rate": 6.1696e-06, "loss": 0.027, "step": 19153 }, { "epoch": 1.021165431572213, "grad_norm": 1.9816522598266602, "learning_rate": 6.169400000000001e-06, "loss": 0.0365, "step": 19154 }, { "epoch": 1.021218745001866, "grad_norm": 2.9084417819976807, "learning_rate": 6.169200000000001e-06, "loss": 0.0523, "step": 19155 }, { "epoch": 1.021272058431519, "grad_norm": 1.5906888246536255, "learning_rate": 6.1690000000000006e-06, "loss": 0.0273, "step": 19156 }, { "epoch": 1.021325371861172, "grad_norm": 2.5956733226776123, "learning_rate": 6.1688e-06, "loss": 0.0203, "step": 19157 }, { "epoch": 1.0213786852908247, "grad_norm": 2.2825779914855957, "learning_rate": 6.1686000000000005e-06, "loss": 0.0547, "step": 19158 }, { "epoch": 1.0214319987204776, "grad_norm": 1.6567838191986084, "learning_rate": 6.168400000000001e-06, "loss": 0.0124, "step": 19159 }, { "epoch": 1.0214853121501306, "grad_norm": 3.5453271865844727, "learning_rate": 6.1682000000000005e-06, "loss": 0.0272, "step": 19160 }, { "epoch": 1.0215386255797836, "grad_norm": 1.8173069953918457, "learning_rate": 6.168000000000001e-06, "loss": 0.025, "step": 19161 }, { "epoch": 1.0215919390094366, "grad_norm": 1.2134393453598022, "learning_rate": 6.1678000000000005e-06, "loss": 0.0239, "step": 19162 }, { "epoch": 1.0216452524390893, "grad_norm": 1.722428321838379, "learning_rate": 6.1676e-06, "loss": 0.022, "step": 19163 }, { "epoch": 1.0216985658687423, "grad_norm": 2.174595355987549, "learning_rate": 6.1674000000000004e-06, "loss": 0.0247, "step": 19164 }, { "epoch": 1.0217518792983953, "grad_norm": 2.401111125946045, "learning_rate": 6.167200000000001e-06, "loss": 0.0344, "step": 19165 }, { "epoch": 1.0218051927280483, "grad_norm": 1.2132517099380493, "learning_rate": 6.167e-06, "loss": 0.0178, "step": 19166 }, { "epoch": 1.021858506157701, "grad_norm": 1.5057569742202759, "learning_rate": 6.166800000000001e-06, "loss": 0.0384, "step": 19167 }, { "epoch": 1.021911819587354, "grad_norm": 0.7306671142578125, "learning_rate": 6.1666e-06, "loss": 0.021, "step": 19168 }, { "epoch": 1.021965133017007, "grad_norm": 1.1060467958450317, "learning_rate": 6.1664e-06, "loss": 0.0337, "step": 19169 }, { "epoch": 1.02201844644666, "grad_norm": 1.2938001155853271, "learning_rate": 6.1662e-06, "loss": 0.0183, "step": 19170 }, { "epoch": 1.022071759876313, "grad_norm": 1.6414453983306885, "learning_rate": 6.166000000000001e-06, "loss": 0.022, "step": 19171 }, { "epoch": 1.0221250733059657, "grad_norm": 3.0141348838806152, "learning_rate": 6.1658e-06, "loss": 0.0528, "step": 19172 }, { "epoch": 1.0221783867356187, "grad_norm": 1.089368224143982, "learning_rate": 6.165600000000001e-06, "loss": 0.0186, "step": 19173 }, { "epoch": 1.0222317001652717, "grad_norm": 2.416956901550293, "learning_rate": 6.1654e-06, "loss": 0.0427, "step": 19174 }, { "epoch": 1.0222850135949246, "grad_norm": 1.561739206314087, "learning_rate": 6.1652e-06, "loss": 0.0361, "step": 19175 }, { "epoch": 1.0223383270245774, "grad_norm": 1.8356480598449707, "learning_rate": 6.165000000000001e-06, "loss": 0.0397, "step": 19176 }, { "epoch": 1.0223916404542304, "grad_norm": 4.799711227416992, "learning_rate": 6.164800000000001e-06, "loss": 0.0142, "step": 19177 }, { "epoch": 1.0224449538838833, "grad_norm": 5.049735069274902, "learning_rate": 6.1646e-06, "loss": 0.0479, "step": 19178 }, { "epoch": 1.0224982673135363, "grad_norm": 1.1871294975280762, "learning_rate": 6.164400000000001e-06, "loss": 0.0268, "step": 19179 }, { "epoch": 1.0225515807431893, "grad_norm": 2.747143507003784, "learning_rate": 6.1642e-06, "loss": 0.0201, "step": 19180 }, { "epoch": 1.022604894172842, "grad_norm": 2.2566826343536377, "learning_rate": 6.164e-06, "loss": 0.0246, "step": 19181 }, { "epoch": 1.022658207602495, "grad_norm": 2.2875235080718994, "learning_rate": 6.163800000000001e-06, "loss": 0.0164, "step": 19182 }, { "epoch": 1.022711521032148, "grad_norm": 3.011476516723633, "learning_rate": 6.163600000000001e-06, "loss": 0.0557, "step": 19183 }, { "epoch": 1.022764834461801, "grad_norm": 1.887486457824707, "learning_rate": 6.1634e-06, "loss": 0.0436, "step": 19184 }, { "epoch": 1.0228181478914538, "grad_norm": 1.5939021110534668, "learning_rate": 6.1632000000000006e-06, "loss": 0.0328, "step": 19185 }, { "epoch": 1.0228714613211067, "grad_norm": 3.22233510017395, "learning_rate": 6.163e-06, "loss": 0.0572, "step": 19186 }, { "epoch": 1.0229247747507597, "grad_norm": 1.6698468923568726, "learning_rate": 6.162800000000001e-06, "loss": 0.0169, "step": 19187 }, { "epoch": 1.0229780881804127, "grad_norm": 4.25574254989624, "learning_rate": 6.162600000000001e-06, "loss": 0.0344, "step": 19188 }, { "epoch": 1.0230314016100657, "grad_norm": 0.9599496722221375, "learning_rate": 6.1624000000000005e-06, "loss": 0.0075, "step": 19189 }, { "epoch": 1.0230847150397184, "grad_norm": 1.3094124794006348, "learning_rate": 6.1622e-06, "loss": 0.0198, "step": 19190 }, { "epoch": 1.0231380284693714, "grad_norm": 1.6374527215957642, "learning_rate": 6.1620000000000005e-06, "loss": 0.0117, "step": 19191 }, { "epoch": 1.0231913418990244, "grad_norm": 0.804792582988739, "learning_rate": 6.1618e-06, "loss": 0.0091, "step": 19192 }, { "epoch": 1.0232446553286774, "grad_norm": 0.8092148303985596, "learning_rate": 6.161600000000001e-06, "loss": 0.011, "step": 19193 }, { "epoch": 1.0232979687583301, "grad_norm": 1.021833062171936, "learning_rate": 6.161400000000001e-06, "loss": 0.0136, "step": 19194 }, { "epoch": 1.023351282187983, "grad_norm": 2.5638697147369385, "learning_rate": 6.1612e-06, "loss": 0.0196, "step": 19195 }, { "epoch": 1.023404595617636, "grad_norm": 3.7812483310699463, "learning_rate": 6.161e-06, "loss": 0.04, "step": 19196 }, { "epoch": 1.023457909047289, "grad_norm": 1.4331005811691284, "learning_rate": 6.1608e-06, "loss": 0.0109, "step": 19197 }, { "epoch": 1.023511222476942, "grad_norm": 1.92478609085083, "learning_rate": 6.160600000000001e-06, "loss": 0.0435, "step": 19198 }, { "epoch": 1.0235645359065948, "grad_norm": 2.2504281997680664, "learning_rate": 6.160400000000001e-06, "loss": 0.0319, "step": 19199 }, { "epoch": 1.0236178493362478, "grad_norm": 1.6270785331726074, "learning_rate": 6.160200000000001e-06, "loss": 0.0244, "step": 19200 }, { "epoch": 1.0236711627659008, "grad_norm": 1.456672191619873, "learning_rate": 6.16e-06, "loss": 0.028, "step": 19201 }, { "epoch": 1.0237244761955537, "grad_norm": 2.4286653995513916, "learning_rate": 6.1598e-06, "loss": 0.0333, "step": 19202 }, { "epoch": 1.0237777896252065, "grad_norm": 2.1560497283935547, "learning_rate": 6.1596e-06, "loss": 0.0834, "step": 19203 }, { "epoch": 1.0238311030548595, "grad_norm": 1.0824394226074219, "learning_rate": 6.159400000000001e-06, "loss": 0.0113, "step": 19204 }, { "epoch": 1.0238844164845124, "grad_norm": 2.0233590602874756, "learning_rate": 6.159200000000001e-06, "loss": 0.0282, "step": 19205 }, { "epoch": 1.0239377299141654, "grad_norm": 2.0230958461761475, "learning_rate": 6.159000000000001e-06, "loss": 0.0205, "step": 19206 }, { "epoch": 1.0239910433438184, "grad_norm": 1.2754089832305908, "learning_rate": 6.1588e-06, "loss": 0.0083, "step": 19207 }, { "epoch": 1.0240443567734712, "grad_norm": 4.477837085723877, "learning_rate": 6.1586e-06, "loss": 0.0317, "step": 19208 }, { "epoch": 1.0240976702031241, "grad_norm": 2.833615779876709, "learning_rate": 6.158400000000001e-06, "loss": 0.0308, "step": 19209 }, { "epoch": 1.0241509836327771, "grad_norm": 3.937678098678589, "learning_rate": 6.158200000000001e-06, "loss": 0.0709, "step": 19210 }, { "epoch": 1.02420429706243, "grad_norm": 1.3702412843704224, "learning_rate": 6.158000000000001e-06, "loss": 0.0149, "step": 19211 }, { "epoch": 1.0242576104920829, "grad_norm": 1.5138046741485596, "learning_rate": 6.1578000000000006e-06, "loss": 0.0222, "step": 19212 }, { "epoch": 1.0243109239217358, "grad_norm": 1.4643042087554932, "learning_rate": 6.1576e-06, "loss": 0.0273, "step": 19213 }, { "epoch": 1.0243642373513888, "grad_norm": 1.47256338596344, "learning_rate": 6.1574e-06, "loss": 0.0228, "step": 19214 }, { "epoch": 1.0244175507810418, "grad_norm": 1.9056806564331055, "learning_rate": 6.157200000000001e-06, "loss": 0.0458, "step": 19215 }, { "epoch": 1.0244708642106948, "grad_norm": 2.1366190910339355, "learning_rate": 6.1570000000000005e-06, "loss": 0.0225, "step": 19216 }, { "epoch": 1.0245241776403475, "grad_norm": 3.0144269466400146, "learning_rate": 6.156800000000001e-06, "loss": 0.0355, "step": 19217 }, { "epoch": 1.0245774910700005, "grad_norm": 7.1524553298950195, "learning_rate": 6.1566000000000005e-06, "loss": 0.0478, "step": 19218 }, { "epoch": 1.0246308044996535, "grad_norm": 1.4706460237503052, "learning_rate": 6.1564e-06, "loss": 0.0189, "step": 19219 }, { "epoch": 1.0246841179293065, "grad_norm": 1.1295746564865112, "learning_rate": 6.1562e-06, "loss": 0.0133, "step": 19220 }, { "epoch": 1.0247374313589592, "grad_norm": 1.608500361442566, "learning_rate": 6.156000000000001e-06, "loss": 0.0169, "step": 19221 }, { "epoch": 1.0247907447886122, "grad_norm": 2.0287094116210938, "learning_rate": 6.1558e-06, "loss": 0.0343, "step": 19222 }, { "epoch": 1.0248440582182652, "grad_norm": 2.7322402000427246, "learning_rate": 6.155600000000001e-06, "loss": 0.071, "step": 19223 }, { "epoch": 1.0248973716479182, "grad_norm": 3.0052902698516846, "learning_rate": 6.1554e-06, "loss": 0.0866, "step": 19224 }, { "epoch": 1.0249506850775711, "grad_norm": 0.8405596613883972, "learning_rate": 6.1552e-06, "loss": 0.0131, "step": 19225 }, { "epoch": 1.025003998507224, "grad_norm": 1.3406901359558105, "learning_rate": 6.155000000000001e-06, "loss": 0.0151, "step": 19226 }, { "epoch": 1.0250573119368769, "grad_norm": 1.4291651248931885, "learning_rate": 6.154800000000001e-06, "loss": 0.0136, "step": 19227 }, { "epoch": 1.0251106253665299, "grad_norm": 3.568843126296997, "learning_rate": 6.1546e-06, "loss": 0.0561, "step": 19228 }, { "epoch": 1.0251639387961828, "grad_norm": 1.5804411172866821, "learning_rate": 6.154400000000001e-06, "loss": 0.0127, "step": 19229 }, { "epoch": 1.0252172522258356, "grad_norm": 1.6683006286621094, "learning_rate": 6.1542e-06, "loss": 0.033, "step": 19230 }, { "epoch": 1.0252705656554886, "grad_norm": 0.8172817230224609, "learning_rate": 6.154e-06, "loss": 0.0157, "step": 19231 }, { "epoch": 1.0253238790851416, "grad_norm": 1.3096030950546265, "learning_rate": 6.153800000000001e-06, "loss": 0.0284, "step": 19232 }, { "epoch": 1.0253771925147945, "grad_norm": 0.8907626271247864, "learning_rate": 6.153600000000001e-06, "loss": 0.0105, "step": 19233 }, { "epoch": 1.0254305059444475, "grad_norm": 0.8241353631019592, "learning_rate": 6.1534e-06, "loss": 0.0082, "step": 19234 }, { "epoch": 1.0254838193741003, "grad_norm": 2.9056551456451416, "learning_rate": 6.153200000000001e-06, "loss": 0.0327, "step": 19235 }, { "epoch": 1.0255371328037532, "grad_norm": 1.15311861038208, "learning_rate": 6.153e-06, "loss": 0.0213, "step": 19236 }, { "epoch": 1.0255904462334062, "grad_norm": 1.4652243852615356, "learning_rate": 6.152800000000001e-06, "loss": 0.0351, "step": 19237 }, { "epoch": 1.0256437596630592, "grad_norm": 1.3590081930160522, "learning_rate": 6.152600000000001e-06, "loss": 0.0155, "step": 19238 }, { "epoch": 1.025697073092712, "grad_norm": 4.448612213134766, "learning_rate": 6.152400000000001e-06, "loss": 0.0526, "step": 19239 }, { "epoch": 1.025750386522365, "grad_norm": 2.1803536415100098, "learning_rate": 6.1522e-06, "loss": 0.0192, "step": 19240 }, { "epoch": 1.025803699952018, "grad_norm": 1.0177828073501587, "learning_rate": 6.1520000000000006e-06, "loss": 0.0111, "step": 19241 }, { "epoch": 1.025857013381671, "grad_norm": 4.146580696105957, "learning_rate": 6.1518e-06, "loss": 0.0389, "step": 19242 }, { "epoch": 1.0259103268113239, "grad_norm": 0.7008031606674194, "learning_rate": 6.1516000000000005e-06, "loss": 0.0066, "step": 19243 }, { "epoch": 1.0259636402409766, "grad_norm": 1.000747561454773, "learning_rate": 6.151400000000001e-06, "loss": 0.0148, "step": 19244 }, { "epoch": 1.0260169536706296, "grad_norm": 1.6200348138809204, "learning_rate": 6.1512000000000005e-06, "loss": 0.0318, "step": 19245 }, { "epoch": 1.0260702671002826, "grad_norm": 1.401077389717102, "learning_rate": 6.151e-06, "loss": 0.0347, "step": 19246 }, { "epoch": 1.0261235805299356, "grad_norm": 2.729684591293335, "learning_rate": 6.1508000000000005e-06, "loss": 0.0267, "step": 19247 }, { "epoch": 1.0261768939595883, "grad_norm": 1.2423040866851807, "learning_rate": 6.150600000000001e-06, "loss": 0.0199, "step": 19248 }, { "epoch": 1.0262302073892413, "grad_norm": 0.4577672779560089, "learning_rate": 6.1504000000000004e-06, "loss": 0.0051, "step": 19249 }, { "epoch": 1.0262835208188943, "grad_norm": 1.0125963687896729, "learning_rate": 6.150200000000001e-06, "loss": 0.0212, "step": 19250 }, { "epoch": 1.0263368342485473, "grad_norm": 1.465464472770691, "learning_rate": 6.15e-06, "loss": 0.0204, "step": 19251 }, { "epoch": 1.0263901476782, "grad_norm": 1.058631420135498, "learning_rate": 6.1498e-06, "loss": 0.0201, "step": 19252 }, { "epoch": 1.026443461107853, "grad_norm": 2.242239475250244, "learning_rate": 6.1496e-06, "loss": 0.023, "step": 19253 }, { "epoch": 1.026496774537506, "grad_norm": 0.885662317276001, "learning_rate": 6.149400000000001e-06, "loss": 0.0194, "step": 19254 }, { "epoch": 1.026550087967159, "grad_norm": 3.697845935821533, "learning_rate": 6.1492e-06, "loss": 0.039, "step": 19255 }, { "epoch": 1.026603401396812, "grad_norm": 2.408395290374756, "learning_rate": 6.149000000000001e-06, "loss": 0.0331, "step": 19256 }, { "epoch": 1.0266567148264647, "grad_norm": 1.4118380546569824, "learning_rate": 6.1488e-06, "loss": 0.0135, "step": 19257 }, { "epoch": 1.0267100282561177, "grad_norm": 2.1160197257995605, "learning_rate": 6.1486e-06, "loss": 0.0705, "step": 19258 }, { "epoch": 1.0267633416857707, "grad_norm": 4.5508880615234375, "learning_rate": 6.148400000000001e-06, "loss": 0.0094, "step": 19259 }, { "epoch": 1.0268166551154236, "grad_norm": 1.5448439121246338, "learning_rate": 6.148200000000001e-06, "loss": 0.0167, "step": 19260 }, { "epoch": 1.0268699685450766, "grad_norm": 2.8840341567993164, "learning_rate": 6.148e-06, "loss": 0.0432, "step": 19261 }, { "epoch": 1.0269232819747294, "grad_norm": 1.352528691291809, "learning_rate": 6.147800000000001e-06, "loss": 0.017, "step": 19262 }, { "epoch": 1.0269765954043824, "grad_norm": 1.4516152143478394, "learning_rate": 6.1476e-06, "loss": 0.0477, "step": 19263 }, { "epoch": 1.0270299088340353, "grad_norm": 2.847910165786743, "learning_rate": 6.1474e-06, "loss": 0.0131, "step": 19264 }, { "epoch": 1.0270832222636883, "grad_norm": 1.5129905939102173, "learning_rate": 6.147200000000001e-06, "loss": 0.0236, "step": 19265 }, { "epoch": 1.027136535693341, "grad_norm": 0.5819046497344971, "learning_rate": 6.147000000000001e-06, "loss": 0.0063, "step": 19266 }, { "epoch": 1.027189849122994, "grad_norm": 1.1162937879562378, "learning_rate": 6.1468e-06, "loss": 0.0214, "step": 19267 }, { "epoch": 1.027243162552647, "grad_norm": 1.5000096559524536, "learning_rate": 6.146600000000001e-06, "loss": 0.0162, "step": 19268 }, { "epoch": 1.0272964759823, "grad_norm": 0.9572567939758301, "learning_rate": 6.1464e-06, "loss": 0.0092, "step": 19269 }, { "epoch": 1.0273497894119528, "grad_norm": 0.7456615567207336, "learning_rate": 6.146200000000001e-06, "loss": 0.0066, "step": 19270 }, { "epoch": 1.0274031028416057, "grad_norm": 2.245026111602783, "learning_rate": 6.146000000000001e-06, "loss": 0.0304, "step": 19271 }, { "epoch": 1.0274564162712587, "grad_norm": 0.9280598759651184, "learning_rate": 6.1458000000000005e-06, "loss": 0.0083, "step": 19272 }, { "epoch": 1.0275097297009117, "grad_norm": 4.692578315734863, "learning_rate": 6.1456e-06, "loss": 0.0309, "step": 19273 }, { "epoch": 1.0275630431305647, "grad_norm": 1.6395413875579834, "learning_rate": 6.1454000000000005e-06, "loss": 0.0088, "step": 19274 }, { "epoch": 1.0276163565602174, "grad_norm": 1.3335810899734497, "learning_rate": 6.1452e-06, "loss": 0.0186, "step": 19275 }, { "epoch": 1.0276696699898704, "grad_norm": 1.5176997184753418, "learning_rate": 6.145000000000001e-06, "loss": 0.0351, "step": 19276 }, { "epoch": 1.0277229834195234, "grad_norm": 0.9252996444702148, "learning_rate": 6.144800000000001e-06, "loss": 0.0084, "step": 19277 }, { "epoch": 1.0277762968491764, "grad_norm": 1.7616862058639526, "learning_rate": 6.1446000000000004e-06, "loss": 0.0133, "step": 19278 }, { "epoch": 1.0278296102788291, "grad_norm": 1.189760446548462, "learning_rate": 6.1444e-06, "loss": 0.0075, "step": 19279 }, { "epoch": 1.027882923708482, "grad_norm": 2.733200788497925, "learning_rate": 6.1442e-06, "loss": 0.0382, "step": 19280 }, { "epoch": 1.027936237138135, "grad_norm": 1.276508092880249, "learning_rate": 6.144e-06, "loss": 0.0166, "step": 19281 }, { "epoch": 1.027989550567788, "grad_norm": 1.6965032815933228, "learning_rate": 6.143800000000001e-06, "loss": 0.019, "step": 19282 }, { "epoch": 1.028042863997441, "grad_norm": 0.7991171479225159, "learning_rate": 6.143600000000001e-06, "loss": 0.0246, "step": 19283 }, { "epoch": 1.0280961774270938, "grad_norm": 1.4952502250671387, "learning_rate": 6.1434e-06, "loss": 0.0252, "step": 19284 }, { "epoch": 1.0281494908567468, "grad_norm": 1.031538724899292, "learning_rate": 6.1432e-06, "loss": 0.0059, "step": 19285 }, { "epoch": 1.0282028042863998, "grad_norm": 1.7944183349609375, "learning_rate": 6.143e-06, "loss": 0.024, "step": 19286 }, { "epoch": 1.0282561177160527, "grad_norm": 2.5016019344329834, "learning_rate": 6.142800000000001e-06, "loss": 0.0271, "step": 19287 }, { "epoch": 1.0283094311457055, "grad_norm": 0.8340848684310913, "learning_rate": 6.142600000000001e-06, "loss": 0.0131, "step": 19288 }, { "epoch": 1.0283627445753585, "grad_norm": 2.8708481788635254, "learning_rate": 6.142400000000001e-06, "loss": 0.02, "step": 19289 }, { "epoch": 1.0284160580050115, "grad_norm": 2.0134799480438232, "learning_rate": 6.1422e-06, "loss": 0.0173, "step": 19290 }, { "epoch": 1.0284693714346644, "grad_norm": 1.813812494277954, "learning_rate": 6.142e-06, "loss": 0.0305, "step": 19291 }, { "epoch": 1.0285226848643174, "grad_norm": 0.8156504034996033, "learning_rate": 6.1418e-06, "loss": 0.0156, "step": 19292 }, { "epoch": 1.0285759982939702, "grad_norm": 1.1611248254776, "learning_rate": 6.141600000000001e-06, "loss": 0.0134, "step": 19293 }, { "epoch": 1.0286293117236232, "grad_norm": 2.6819891929626465, "learning_rate": 6.141400000000001e-06, "loss": 0.0318, "step": 19294 }, { "epoch": 1.0286826251532761, "grad_norm": 2.9221394062042236, "learning_rate": 6.141200000000001e-06, "loss": 0.0694, "step": 19295 }, { "epoch": 1.028735938582929, "grad_norm": 0.9863402247428894, "learning_rate": 6.141e-06, "loss": 0.0071, "step": 19296 }, { "epoch": 1.0287892520125819, "grad_norm": 0.8394542932510376, "learning_rate": 6.1408e-06, "loss": 0.0064, "step": 19297 }, { "epoch": 1.0288425654422348, "grad_norm": 1.028250813484192, "learning_rate": 6.140600000000001e-06, "loss": 0.007, "step": 19298 }, { "epoch": 1.0288958788718878, "grad_norm": 1.7670258283615112, "learning_rate": 6.1404000000000005e-06, "loss": 0.0258, "step": 19299 }, { "epoch": 1.0289491923015408, "grad_norm": 1.9313372373580933, "learning_rate": 6.140200000000001e-06, "loss": 0.0357, "step": 19300 }, { "epoch": 1.0290025057311938, "grad_norm": 0.6221982836723328, "learning_rate": 6.1400000000000005e-06, "loss": 0.0029, "step": 19301 }, { "epoch": 1.0290558191608465, "grad_norm": 2.2567100524902344, "learning_rate": 6.1398e-06, "loss": 0.0248, "step": 19302 }, { "epoch": 1.0291091325904995, "grad_norm": 2.0279762744903564, "learning_rate": 6.1396e-06, "loss": 0.0367, "step": 19303 }, { "epoch": 1.0291624460201525, "grad_norm": 1.0568604469299316, "learning_rate": 6.139400000000001e-06, "loss": 0.0064, "step": 19304 }, { "epoch": 1.0292157594498055, "grad_norm": 3.1133809089660645, "learning_rate": 6.1392000000000004e-06, "loss": 0.0537, "step": 19305 }, { "epoch": 1.0292690728794582, "grad_norm": 1.883199691772461, "learning_rate": 6.139000000000001e-06, "loss": 0.0207, "step": 19306 }, { "epoch": 1.0293223863091112, "grad_norm": 1.9326884746551514, "learning_rate": 6.1388e-06, "loss": 0.0125, "step": 19307 }, { "epoch": 1.0293756997387642, "grad_norm": 2.4947712421417236, "learning_rate": 6.1386e-06, "loss": 0.0542, "step": 19308 }, { "epoch": 1.0294290131684172, "grad_norm": 2.6929755210876465, "learning_rate": 6.138400000000001e-06, "loss": 0.0356, "step": 19309 }, { "epoch": 1.0294823265980702, "grad_norm": 2.18627667427063, "learning_rate": 6.138200000000001e-06, "loss": 0.0343, "step": 19310 }, { "epoch": 1.029535640027723, "grad_norm": 1.617562174797058, "learning_rate": 6.138e-06, "loss": 0.0349, "step": 19311 }, { "epoch": 1.0295889534573759, "grad_norm": 2.3604187965393066, "learning_rate": 6.137800000000001e-06, "loss": 0.0259, "step": 19312 }, { "epoch": 1.0296422668870289, "grad_norm": 1.6165777444839478, "learning_rate": 6.1376e-06, "loss": 0.0255, "step": 19313 }, { "epoch": 1.0296955803166818, "grad_norm": 1.678320288658142, "learning_rate": 6.1374e-06, "loss": 0.0211, "step": 19314 }, { "epoch": 1.0297488937463346, "grad_norm": 1.8193774223327637, "learning_rate": 6.137200000000001e-06, "loss": 0.0491, "step": 19315 }, { "epoch": 1.0298022071759876, "grad_norm": 2.169059991836548, "learning_rate": 6.137000000000001e-06, "loss": 0.0123, "step": 19316 }, { "epoch": 1.0298555206056406, "grad_norm": 1.9890954494476318, "learning_rate": 6.1368e-06, "loss": 0.0267, "step": 19317 }, { "epoch": 1.0299088340352935, "grad_norm": 1.7221159934997559, "learning_rate": 6.136600000000001e-06, "loss": 0.0235, "step": 19318 }, { "epoch": 1.0299621474649465, "grad_norm": 2.722247838973999, "learning_rate": 6.1364e-06, "loss": 0.0322, "step": 19319 }, { "epoch": 1.0300154608945993, "grad_norm": 1.3569862842559814, "learning_rate": 6.136200000000001e-06, "loss": 0.022, "step": 19320 }, { "epoch": 1.0300687743242523, "grad_norm": 2.691112756729126, "learning_rate": 6.136000000000001e-06, "loss": 0.0136, "step": 19321 }, { "epoch": 1.0301220877539052, "grad_norm": 1.2548915147781372, "learning_rate": 6.135800000000001e-06, "loss": 0.0106, "step": 19322 }, { "epoch": 1.0301754011835582, "grad_norm": 2.301234006881714, "learning_rate": 6.1356e-06, "loss": 0.0484, "step": 19323 }, { "epoch": 1.030228714613211, "grad_norm": 1.0421735048294067, "learning_rate": 6.135400000000001e-06, "loss": 0.0219, "step": 19324 }, { "epoch": 1.030282028042864, "grad_norm": 2.035651445388794, "learning_rate": 6.1352e-06, "loss": 0.0292, "step": 19325 }, { "epoch": 1.030335341472517, "grad_norm": 1.226989984512329, "learning_rate": 6.1350000000000006e-06, "loss": 0.0208, "step": 19326 }, { "epoch": 1.03038865490217, "grad_norm": 1.767911434173584, "learning_rate": 6.134800000000001e-06, "loss": 0.0449, "step": 19327 }, { "epoch": 1.0304419683318229, "grad_norm": 1.764725685119629, "learning_rate": 6.1346000000000005e-06, "loss": 0.0139, "step": 19328 }, { "epoch": 1.0304952817614756, "grad_norm": 1.7017455101013184, "learning_rate": 6.1344e-06, "loss": 0.0299, "step": 19329 }, { "epoch": 1.0305485951911286, "grad_norm": 0.7626410126686096, "learning_rate": 6.1342000000000005e-06, "loss": 0.0189, "step": 19330 }, { "epoch": 1.0306019086207816, "grad_norm": 2.533679723739624, "learning_rate": 6.134e-06, "loss": 0.0345, "step": 19331 }, { "epoch": 1.0306552220504346, "grad_norm": 2.249201536178589, "learning_rate": 6.1338000000000005e-06, "loss": 0.0305, "step": 19332 }, { "epoch": 1.0307085354800873, "grad_norm": 0.648797333240509, "learning_rate": 6.133600000000001e-06, "loss": 0.0056, "step": 19333 }, { "epoch": 1.0307618489097403, "grad_norm": 1.3505085706710815, "learning_rate": 6.1334000000000004e-06, "loss": 0.0087, "step": 19334 }, { "epoch": 1.0308151623393933, "grad_norm": 2.7976362705230713, "learning_rate": 6.1332e-06, "loss": 0.0411, "step": 19335 }, { "epoch": 1.0308684757690463, "grad_norm": 1.7106168270111084, "learning_rate": 6.133e-06, "loss": 0.01, "step": 19336 }, { "epoch": 1.0309217891986993, "grad_norm": 3.7932991981506348, "learning_rate": 6.132800000000001e-06, "loss": 0.0532, "step": 19337 }, { "epoch": 1.030975102628352, "grad_norm": 2.074730634689331, "learning_rate": 6.1326e-06, "loss": 0.0648, "step": 19338 }, { "epoch": 1.031028416058005, "grad_norm": 1.5825105905532837, "learning_rate": 6.132400000000001e-06, "loss": 0.0212, "step": 19339 }, { "epoch": 1.031081729487658, "grad_norm": 2.0184152126312256, "learning_rate": 6.1322e-06, "loss": 0.024, "step": 19340 }, { "epoch": 1.031135042917311, "grad_norm": 1.9343208074569702, "learning_rate": 6.132e-06, "loss": 0.026, "step": 19341 }, { "epoch": 1.0311883563469637, "grad_norm": 1.152691125869751, "learning_rate": 6.1318e-06, "loss": 0.0237, "step": 19342 }, { "epoch": 1.0312416697766167, "grad_norm": 1.942132830619812, "learning_rate": 6.131600000000001e-06, "loss": 0.0378, "step": 19343 }, { "epoch": 1.0312949832062697, "grad_norm": 2.3343966007232666, "learning_rate": 6.1314e-06, "loss": 0.0276, "step": 19344 }, { "epoch": 1.0313482966359226, "grad_norm": 3.0280380249023438, "learning_rate": 6.131200000000001e-06, "loss": 0.0788, "step": 19345 }, { "epoch": 1.0314016100655756, "grad_norm": 3.2260162830352783, "learning_rate": 6.131e-06, "loss": 0.0463, "step": 19346 }, { "epoch": 1.0314549234952284, "grad_norm": 1.2407077550888062, "learning_rate": 6.1308e-06, "loss": 0.0142, "step": 19347 }, { "epoch": 1.0315082369248814, "grad_norm": 2.9192047119140625, "learning_rate": 6.130600000000001e-06, "loss": 0.0337, "step": 19348 }, { "epoch": 1.0315615503545343, "grad_norm": 2.515721321105957, "learning_rate": 6.130400000000001e-06, "loss": 0.0478, "step": 19349 }, { "epoch": 1.0316148637841873, "grad_norm": 1.2896084785461426, "learning_rate": 6.1302e-06, "loss": 0.0285, "step": 19350 }, { "epoch": 1.03166817721384, "grad_norm": 1.5157614946365356, "learning_rate": 6.130000000000001e-06, "loss": 0.0276, "step": 19351 }, { "epoch": 1.031721490643493, "grad_norm": 2.2655742168426514, "learning_rate": 6.1298e-06, "loss": 0.0339, "step": 19352 }, { "epoch": 1.031774804073146, "grad_norm": 2.822953462600708, "learning_rate": 6.1296e-06, "loss": 0.0464, "step": 19353 }, { "epoch": 1.031828117502799, "grad_norm": 2.4591915607452393, "learning_rate": 6.129400000000001e-06, "loss": 0.0434, "step": 19354 }, { "epoch": 1.0318814309324518, "grad_norm": 0.8745144009590149, "learning_rate": 6.1292000000000005e-06, "loss": 0.0073, "step": 19355 }, { "epoch": 1.0319347443621047, "grad_norm": 1.008692979812622, "learning_rate": 6.129e-06, "loss": 0.0114, "step": 19356 }, { "epoch": 1.0319880577917577, "grad_norm": 2.1794159412384033, "learning_rate": 6.1288000000000005e-06, "loss": 0.0496, "step": 19357 }, { "epoch": 1.0320413712214107, "grad_norm": 2.305535316467285, "learning_rate": 6.1286e-06, "loss": 0.0301, "step": 19358 }, { "epoch": 1.0320946846510637, "grad_norm": 1.5700335502624512, "learning_rate": 6.128400000000001e-06, "loss": 0.0182, "step": 19359 }, { "epoch": 1.0321479980807164, "grad_norm": 2.7932562828063965, "learning_rate": 6.128200000000001e-06, "loss": 0.0348, "step": 19360 }, { "epoch": 1.0322013115103694, "grad_norm": 1.1375656127929688, "learning_rate": 6.1280000000000005e-06, "loss": 0.0353, "step": 19361 }, { "epoch": 1.0322546249400224, "grad_norm": 0.9510666728019714, "learning_rate": 6.1278e-06, "loss": 0.0089, "step": 19362 }, { "epoch": 1.0323079383696754, "grad_norm": 1.3532090187072754, "learning_rate": 6.1276e-06, "loss": 0.0129, "step": 19363 }, { "epoch": 1.0323612517993284, "grad_norm": 1.225115418434143, "learning_rate": 6.1274e-06, "loss": 0.0266, "step": 19364 }, { "epoch": 1.0324145652289811, "grad_norm": 1.1448628902435303, "learning_rate": 6.127200000000001e-06, "loss": 0.0215, "step": 19365 }, { "epoch": 1.032467878658634, "grad_norm": 4.314942836761475, "learning_rate": 6.127000000000001e-06, "loss": 0.0252, "step": 19366 }, { "epoch": 1.032521192088287, "grad_norm": 2.333078384399414, "learning_rate": 6.1268e-06, "loss": 0.0338, "step": 19367 }, { "epoch": 1.03257450551794, "grad_norm": 1.278946876525879, "learning_rate": 6.1266e-06, "loss": 0.0186, "step": 19368 }, { "epoch": 1.0326278189475928, "grad_norm": 1.6878401041030884, "learning_rate": 6.1264e-06, "loss": 0.044, "step": 19369 }, { "epoch": 1.0326811323772458, "grad_norm": 1.8742437362670898, "learning_rate": 6.126200000000001e-06, "loss": 0.0472, "step": 19370 }, { "epoch": 1.0327344458068988, "grad_norm": 0.7337755560874939, "learning_rate": 6.126000000000001e-06, "loss": 0.0061, "step": 19371 }, { "epoch": 1.0327877592365518, "grad_norm": 1.1333253383636475, "learning_rate": 6.125800000000001e-06, "loss": 0.0111, "step": 19372 }, { "epoch": 1.0328410726662045, "grad_norm": 3.360931396484375, "learning_rate": 6.1256e-06, "loss": 0.0347, "step": 19373 }, { "epoch": 1.0328943860958575, "grad_norm": 1.63062584400177, "learning_rate": 6.1254e-06, "loss": 0.0275, "step": 19374 }, { "epoch": 1.0329476995255105, "grad_norm": 1.0182167291641235, "learning_rate": 6.1252e-06, "loss": 0.0166, "step": 19375 }, { "epoch": 1.0330010129551634, "grad_norm": 0.6159800291061401, "learning_rate": 6.125000000000001e-06, "loss": 0.0055, "step": 19376 }, { "epoch": 1.0330543263848164, "grad_norm": 3.4560647010803223, "learning_rate": 6.124800000000001e-06, "loss": 0.0471, "step": 19377 }, { "epoch": 1.0331076398144692, "grad_norm": 2.623572826385498, "learning_rate": 6.124600000000001e-06, "loss": 0.0384, "step": 19378 }, { "epoch": 1.0331609532441222, "grad_norm": 3.959561347961426, "learning_rate": 6.1244e-06, "loss": 0.0313, "step": 19379 }, { "epoch": 1.0332142666737751, "grad_norm": 1.1896663904190063, "learning_rate": 6.1242e-06, "loss": 0.0084, "step": 19380 }, { "epoch": 1.0332675801034281, "grad_norm": 2.81117582321167, "learning_rate": 6.124000000000001e-06, "loss": 0.0262, "step": 19381 }, { "epoch": 1.033320893533081, "grad_norm": 2.758295774459839, "learning_rate": 6.1238000000000006e-06, "loss": 0.0345, "step": 19382 }, { "epoch": 1.0333742069627339, "grad_norm": 1.04061758518219, "learning_rate": 6.123600000000001e-06, "loss": 0.0103, "step": 19383 }, { "epoch": 1.0334275203923868, "grad_norm": 1.4383450746536255, "learning_rate": 6.1234000000000005e-06, "loss": 0.0303, "step": 19384 }, { "epoch": 1.0334808338220398, "grad_norm": 1.2529569864273071, "learning_rate": 6.1232e-06, "loss": 0.0388, "step": 19385 }, { "epoch": 1.0335341472516928, "grad_norm": 1.8455582857131958, "learning_rate": 6.123e-06, "loss": 0.0389, "step": 19386 }, { "epoch": 1.0335874606813455, "grad_norm": 1.3716685771942139, "learning_rate": 6.122800000000001e-06, "loss": 0.0113, "step": 19387 }, { "epoch": 1.0336407741109985, "grad_norm": 1.8435277938842773, "learning_rate": 6.1226000000000005e-06, "loss": 0.0445, "step": 19388 }, { "epoch": 1.0336940875406515, "grad_norm": 2.824209451675415, "learning_rate": 6.122400000000001e-06, "loss": 0.0864, "step": 19389 }, { "epoch": 1.0337474009703045, "grad_norm": 4.279970169067383, "learning_rate": 6.1222000000000004e-06, "loss": 0.0453, "step": 19390 }, { "epoch": 1.0338007143999572, "grad_norm": 1.493784785270691, "learning_rate": 6.122e-06, "loss": 0.0208, "step": 19391 }, { "epoch": 1.0338540278296102, "grad_norm": 1.5748445987701416, "learning_rate": 6.1217999999999996e-06, "loss": 0.0235, "step": 19392 }, { "epoch": 1.0339073412592632, "grad_norm": 1.2198686599731445, "learning_rate": 6.121600000000001e-06, "loss": 0.016, "step": 19393 }, { "epoch": 1.0339606546889162, "grad_norm": 1.62799870967865, "learning_rate": 6.1214e-06, "loss": 0.0441, "step": 19394 }, { "epoch": 1.0340139681185692, "grad_norm": 0.9309687614440918, "learning_rate": 6.121200000000001e-06, "loss": 0.0108, "step": 19395 }, { "epoch": 1.034067281548222, "grad_norm": 2.0719735622406006, "learning_rate": 6.121e-06, "loss": 0.0486, "step": 19396 }, { "epoch": 1.034120594977875, "grad_norm": 2.0496666431427, "learning_rate": 6.1208e-06, "loss": 0.026, "step": 19397 }, { "epoch": 1.0341739084075279, "grad_norm": 2.15523362159729, "learning_rate": 6.120600000000001e-06, "loss": 0.0463, "step": 19398 }, { "epoch": 1.0342272218371809, "grad_norm": 0.9605041742324829, "learning_rate": 6.120400000000001e-06, "loss": 0.0141, "step": 19399 }, { "epoch": 1.0342805352668336, "grad_norm": 1.9148200750350952, "learning_rate": 6.1202e-06, "loss": 0.0212, "step": 19400 }, { "epoch": 1.0343338486964866, "grad_norm": 2.5368075370788574, "learning_rate": 6.120000000000001e-06, "loss": 0.0307, "step": 19401 }, { "epoch": 1.0343871621261396, "grad_norm": 1.7990864515304565, "learning_rate": 6.1198e-06, "loss": 0.0388, "step": 19402 }, { "epoch": 1.0344404755557925, "grad_norm": 1.075027585029602, "learning_rate": 6.1196e-06, "loss": 0.0128, "step": 19403 }, { "epoch": 1.0344937889854455, "grad_norm": 1.2278053760528564, "learning_rate": 6.119400000000001e-06, "loss": 0.027, "step": 19404 }, { "epoch": 1.0345471024150983, "grad_norm": 2.158003330230713, "learning_rate": 6.119200000000001e-06, "loss": 0.0444, "step": 19405 }, { "epoch": 1.0346004158447513, "grad_norm": 1.4571901559829712, "learning_rate": 6.119e-06, "loss": 0.0293, "step": 19406 }, { "epoch": 1.0346537292744042, "grad_norm": 1.49996018409729, "learning_rate": 6.118800000000001e-06, "loss": 0.0401, "step": 19407 }, { "epoch": 1.0347070427040572, "grad_norm": 2.8209407329559326, "learning_rate": 6.1186e-06, "loss": 0.0456, "step": 19408 }, { "epoch": 1.03476035613371, "grad_norm": 1.7639260292053223, "learning_rate": 6.1184000000000014e-06, "loss": 0.0437, "step": 19409 }, { "epoch": 1.034813669563363, "grad_norm": 2.0377652645111084, "learning_rate": 6.118200000000001e-06, "loss": 0.0409, "step": 19410 }, { "epoch": 1.034866982993016, "grad_norm": 0.996806800365448, "learning_rate": 6.1180000000000005e-06, "loss": 0.0181, "step": 19411 }, { "epoch": 1.034920296422669, "grad_norm": 1.4041606187820435, "learning_rate": 6.1178e-06, "loss": 0.0218, "step": 19412 }, { "epoch": 1.034973609852322, "grad_norm": 1.3828938007354736, "learning_rate": 6.1176000000000005e-06, "loss": 0.0218, "step": 19413 }, { "epoch": 1.0350269232819747, "grad_norm": 1.3992925882339478, "learning_rate": 6.1174e-06, "loss": 0.0199, "step": 19414 }, { "epoch": 1.0350802367116276, "grad_norm": 1.63654625415802, "learning_rate": 6.117200000000001e-06, "loss": 0.058, "step": 19415 }, { "epoch": 1.0351335501412806, "grad_norm": 4.040432929992676, "learning_rate": 6.117000000000001e-06, "loss": 0.0218, "step": 19416 }, { "epoch": 1.0351868635709336, "grad_norm": 2.672161340713501, "learning_rate": 6.1168000000000005e-06, "loss": 0.0273, "step": 19417 }, { "epoch": 1.0352401770005863, "grad_norm": 2.33388090133667, "learning_rate": 6.1166e-06, "loss": 0.0319, "step": 19418 }, { "epoch": 1.0352934904302393, "grad_norm": 2.841217279434204, "learning_rate": 6.1164000000000004e-06, "loss": 0.0538, "step": 19419 }, { "epoch": 1.0353468038598923, "grad_norm": 0.852745771408081, "learning_rate": 6.116200000000001e-06, "loss": 0.0095, "step": 19420 }, { "epoch": 1.0354001172895453, "grad_norm": 1.237565040588379, "learning_rate": 6.116000000000001e-06, "loss": 0.026, "step": 19421 }, { "epoch": 1.0354534307191983, "grad_norm": 0.9569385051727295, "learning_rate": 6.115800000000001e-06, "loss": 0.0153, "step": 19422 }, { "epoch": 1.035506744148851, "grad_norm": 1.2645879983901978, "learning_rate": 6.1156e-06, "loss": 0.0118, "step": 19423 }, { "epoch": 1.035560057578504, "grad_norm": 1.756341576576233, "learning_rate": 6.1154e-06, "loss": 0.039, "step": 19424 }, { "epoch": 1.035613371008157, "grad_norm": 1.3588366508483887, "learning_rate": 6.1152e-06, "loss": 0.0288, "step": 19425 }, { "epoch": 1.03566668443781, "grad_norm": 1.1377829313278198, "learning_rate": 6.115000000000001e-06, "loss": 0.0149, "step": 19426 }, { "epoch": 1.0357199978674627, "grad_norm": 1.6923110485076904, "learning_rate": 6.114800000000001e-06, "loss": 0.0179, "step": 19427 }, { "epoch": 1.0357733112971157, "grad_norm": 1.9675337076187134, "learning_rate": 6.114600000000001e-06, "loss": 0.0231, "step": 19428 }, { "epoch": 1.0358266247267687, "grad_norm": 2.072024345397949, "learning_rate": 6.1144e-06, "loss": 0.0325, "step": 19429 }, { "epoch": 1.0358799381564217, "grad_norm": 1.1386550664901733, "learning_rate": 6.1142e-06, "loss": 0.01, "step": 19430 }, { "epoch": 1.0359332515860746, "grad_norm": 1.2536957263946533, "learning_rate": 6.114000000000001e-06, "loss": 0.0163, "step": 19431 }, { "epoch": 1.0359865650157274, "grad_norm": 0.8997819423675537, "learning_rate": 6.113800000000001e-06, "loss": 0.0203, "step": 19432 }, { "epoch": 1.0360398784453804, "grad_norm": 2.565683603286743, "learning_rate": 6.113600000000001e-06, "loss": 0.0355, "step": 19433 }, { "epoch": 1.0360931918750333, "grad_norm": 1.1805294752120972, "learning_rate": 6.113400000000001e-06, "loss": 0.0125, "step": 19434 }, { "epoch": 1.0361465053046863, "grad_norm": 1.6926672458648682, "learning_rate": 6.1132e-06, "loss": 0.0161, "step": 19435 }, { "epoch": 1.036199818734339, "grad_norm": 2.616353750228882, "learning_rate": 6.113e-06, "loss": 0.0442, "step": 19436 }, { "epoch": 1.036253132163992, "grad_norm": 2.502504587173462, "learning_rate": 6.112800000000001e-06, "loss": 0.0353, "step": 19437 }, { "epoch": 1.036306445593645, "grad_norm": 2.370152473449707, "learning_rate": 6.1126000000000006e-06, "loss": 0.0223, "step": 19438 }, { "epoch": 1.036359759023298, "grad_norm": 1.2698136568069458, "learning_rate": 6.112400000000001e-06, "loss": 0.0155, "step": 19439 }, { "epoch": 1.036413072452951, "grad_norm": 3.840022087097168, "learning_rate": 6.1122000000000005e-06, "loss": 0.0332, "step": 19440 }, { "epoch": 1.0364663858826038, "grad_norm": 1.1781444549560547, "learning_rate": 6.112e-06, "loss": 0.0217, "step": 19441 }, { "epoch": 1.0365196993122567, "grad_norm": 1.0160281658172607, "learning_rate": 6.1118e-06, "loss": 0.013, "step": 19442 }, { "epoch": 1.0365730127419097, "grad_norm": 0.8711267709732056, "learning_rate": 6.111600000000001e-06, "loss": 0.0131, "step": 19443 }, { "epoch": 1.0366263261715627, "grad_norm": 0.8506693243980408, "learning_rate": 6.1114000000000005e-06, "loss": 0.0164, "step": 19444 }, { "epoch": 1.0366796396012155, "grad_norm": 4.686463832855225, "learning_rate": 6.111200000000001e-06, "loss": 0.085, "step": 19445 }, { "epoch": 1.0367329530308684, "grad_norm": 3.2855284214019775, "learning_rate": 6.1110000000000004e-06, "loss": 0.0259, "step": 19446 }, { "epoch": 1.0367862664605214, "grad_norm": 2.4608211517333984, "learning_rate": 6.1108e-06, "loss": 0.0366, "step": 19447 }, { "epoch": 1.0368395798901744, "grad_norm": 1.8073482513427734, "learning_rate": 6.110600000000001e-06, "loss": 0.026, "step": 19448 }, { "epoch": 1.0368928933198274, "grad_norm": 3.4367666244506836, "learning_rate": 6.110400000000001e-06, "loss": 0.0353, "step": 19449 }, { "epoch": 1.0369462067494801, "grad_norm": 1.194846272468567, "learning_rate": 6.1102e-06, "loss": 0.0264, "step": 19450 }, { "epoch": 1.036999520179133, "grad_norm": 1.2878481149673462, "learning_rate": 6.110000000000001e-06, "loss": 0.0063, "step": 19451 }, { "epoch": 1.037052833608786, "grad_norm": 3.0114266872406006, "learning_rate": 6.1098e-06, "loss": 0.0241, "step": 19452 }, { "epoch": 1.037106147038439, "grad_norm": 2.2161314487457275, "learning_rate": 6.1096e-06, "loss": 0.0294, "step": 19453 }, { "epoch": 1.0371594604680918, "grad_norm": 4.554120063781738, "learning_rate": 6.109400000000001e-06, "loss": 0.0303, "step": 19454 }, { "epoch": 1.0372127738977448, "grad_norm": 1.6479908227920532, "learning_rate": 6.109200000000001e-06, "loss": 0.0209, "step": 19455 }, { "epoch": 1.0372660873273978, "grad_norm": 1.6793186664581299, "learning_rate": 6.109e-06, "loss": 0.0153, "step": 19456 }, { "epoch": 1.0373194007570508, "grad_norm": 2.8434693813323975, "learning_rate": 6.108800000000001e-06, "loss": 0.0294, "step": 19457 }, { "epoch": 1.0373727141867037, "grad_norm": 4.383330821990967, "learning_rate": 6.1086e-06, "loss": 0.0088, "step": 19458 }, { "epoch": 1.0374260276163565, "grad_norm": 2.3588790893554688, "learning_rate": 6.108400000000001e-06, "loss": 0.0718, "step": 19459 }, { "epoch": 1.0374793410460095, "grad_norm": 1.5603604316711426, "learning_rate": 6.108200000000001e-06, "loss": 0.0251, "step": 19460 }, { "epoch": 1.0375326544756625, "grad_norm": 1.8998373746871948, "learning_rate": 6.108000000000001e-06, "loss": 0.0217, "step": 19461 }, { "epoch": 1.0375859679053154, "grad_norm": 1.811747670173645, "learning_rate": 6.1078e-06, "loss": 0.0267, "step": 19462 }, { "epoch": 1.0376392813349682, "grad_norm": 1.378045916557312, "learning_rate": 6.107600000000001e-06, "loss": 0.0311, "step": 19463 }, { "epoch": 1.0376925947646212, "grad_norm": 1.05234956741333, "learning_rate": 6.1074e-06, "loss": 0.0069, "step": 19464 }, { "epoch": 1.0377459081942741, "grad_norm": 1.7679435014724731, "learning_rate": 6.107200000000001e-06, "loss": 0.021, "step": 19465 }, { "epoch": 1.0377992216239271, "grad_norm": 2.5137999057769775, "learning_rate": 6.107000000000001e-06, "loss": 0.0392, "step": 19466 }, { "epoch": 1.03785253505358, "grad_norm": 1.8069318532943726, "learning_rate": 6.1068000000000006e-06, "loss": 0.0126, "step": 19467 }, { "epoch": 1.0379058484832329, "grad_norm": 1.7300254106521606, "learning_rate": 6.1066e-06, "loss": 0.014, "step": 19468 }, { "epoch": 1.0379591619128858, "grad_norm": 1.7238901853561401, "learning_rate": 6.1064000000000005e-06, "loss": 0.0137, "step": 19469 }, { "epoch": 1.0380124753425388, "grad_norm": 7.194110870361328, "learning_rate": 6.106200000000001e-06, "loss": 0.0406, "step": 19470 }, { "epoch": 1.0380657887721918, "grad_norm": 4.775673866271973, "learning_rate": 6.1060000000000005e-06, "loss": 0.0281, "step": 19471 }, { "epoch": 1.0381191022018446, "grad_norm": 3.902590036392212, "learning_rate": 6.105800000000001e-06, "loss": 0.0443, "step": 19472 }, { "epoch": 1.0381724156314975, "grad_norm": 2.7931621074676514, "learning_rate": 6.1056000000000005e-06, "loss": 0.069, "step": 19473 }, { "epoch": 1.0382257290611505, "grad_norm": 1.8040882349014282, "learning_rate": 6.1054e-06, "loss": 0.022, "step": 19474 }, { "epoch": 1.0382790424908035, "grad_norm": 1.6493474245071411, "learning_rate": 6.1052000000000004e-06, "loss": 0.0291, "step": 19475 }, { "epoch": 1.0383323559204563, "grad_norm": 3.8827264308929443, "learning_rate": 6.105000000000001e-06, "loss": 0.0382, "step": 19476 }, { "epoch": 1.0383856693501092, "grad_norm": 4.211691856384277, "learning_rate": 6.1048e-06, "loss": 0.0285, "step": 19477 }, { "epoch": 1.0384389827797622, "grad_norm": 3.7160651683807373, "learning_rate": 6.104600000000001e-06, "loss": 0.0318, "step": 19478 }, { "epoch": 1.0384922962094152, "grad_norm": 2.4291744232177734, "learning_rate": 6.1044e-06, "loss": 0.032, "step": 19479 }, { "epoch": 1.0385456096390682, "grad_norm": 2.241407632827759, "learning_rate": 6.1042e-06, "loss": 0.0339, "step": 19480 }, { "epoch": 1.038598923068721, "grad_norm": 1.8625904321670532, "learning_rate": 6.104000000000001e-06, "loss": 0.0214, "step": 19481 }, { "epoch": 1.038652236498374, "grad_norm": 1.1804625988006592, "learning_rate": 6.103800000000001e-06, "loss": 0.0093, "step": 19482 }, { "epoch": 1.0387055499280269, "grad_norm": 1.4677536487579346, "learning_rate": 6.1036e-06, "loss": 0.0103, "step": 19483 }, { "epoch": 1.0387588633576799, "grad_norm": 1.2774518728256226, "learning_rate": 6.103400000000001e-06, "loss": 0.0093, "step": 19484 }, { "epoch": 1.0388121767873328, "grad_norm": 1.99997878074646, "learning_rate": 6.1032e-06, "loss": 0.0334, "step": 19485 }, { "epoch": 1.0388654902169856, "grad_norm": 1.6105401515960693, "learning_rate": 6.103e-06, "loss": 0.0149, "step": 19486 }, { "epoch": 1.0389188036466386, "grad_norm": 5.321464538574219, "learning_rate": 6.102800000000001e-06, "loss": 0.0545, "step": 19487 }, { "epoch": 1.0389721170762916, "grad_norm": 4.795561790466309, "learning_rate": 6.102600000000001e-06, "loss": 0.0315, "step": 19488 }, { "epoch": 1.0390254305059445, "grad_norm": 0.962184488773346, "learning_rate": 6.1024e-06, "loss": 0.0076, "step": 19489 }, { "epoch": 1.0390787439355973, "grad_norm": 4.531588077545166, "learning_rate": 6.102200000000001e-06, "loss": 0.0398, "step": 19490 }, { "epoch": 1.0391320573652503, "grad_norm": 1.598727822303772, "learning_rate": 6.102e-06, "loss": 0.0261, "step": 19491 }, { "epoch": 1.0391853707949033, "grad_norm": 2.3553555011749268, "learning_rate": 6.1018e-06, "loss": 0.0667, "step": 19492 }, { "epoch": 1.0392386842245562, "grad_norm": 0.488800972700119, "learning_rate": 6.101600000000001e-06, "loss": 0.0053, "step": 19493 }, { "epoch": 1.039291997654209, "grad_norm": 2.8398516178131104, "learning_rate": 6.1014000000000006e-06, "loss": 0.0281, "step": 19494 }, { "epoch": 1.039345311083862, "grad_norm": 2.6410868167877197, "learning_rate": 6.1012e-06, "loss": 0.0299, "step": 19495 }, { "epoch": 1.039398624513515, "grad_norm": 2.3057663440704346, "learning_rate": 6.1010000000000005e-06, "loss": 0.0253, "step": 19496 }, { "epoch": 1.039451937943168, "grad_norm": 3.3779237270355225, "learning_rate": 6.1008e-06, "loss": 0.0189, "step": 19497 }, { "epoch": 1.039505251372821, "grad_norm": 3.5745341777801514, "learning_rate": 6.100600000000001e-06, "loss": 0.0579, "step": 19498 }, { "epoch": 1.0395585648024737, "grad_norm": 0.6778956651687622, "learning_rate": 6.100400000000001e-06, "loss": 0.0078, "step": 19499 }, { "epoch": 1.0396118782321266, "grad_norm": 1.3231496810913086, "learning_rate": 6.1002000000000005e-06, "loss": 0.0307, "step": 19500 }, { "epoch": 1.0396651916617796, "grad_norm": 2.175551414489746, "learning_rate": 6.1e-06, "loss": 0.0384, "step": 19501 }, { "epoch": 1.0397185050914326, "grad_norm": 1.6768029928207397, "learning_rate": 6.0998000000000005e-06, "loss": 0.019, "step": 19502 }, { "epoch": 1.0397718185210854, "grad_norm": 2.876020669937134, "learning_rate": 6.0996e-06, "loss": 0.0472, "step": 19503 }, { "epoch": 1.0398251319507383, "grad_norm": 1.7558293342590332, "learning_rate": 6.099400000000001e-06, "loss": 0.037, "step": 19504 }, { "epoch": 1.0398784453803913, "grad_norm": 1.869420051574707, "learning_rate": 6.099200000000001e-06, "loss": 0.0263, "step": 19505 }, { "epoch": 1.0399317588100443, "grad_norm": 1.755495548248291, "learning_rate": 6.099e-06, "loss": 0.0335, "step": 19506 }, { "epoch": 1.0399850722396973, "grad_norm": 2.087019920349121, "learning_rate": 6.0988e-06, "loss": 0.0383, "step": 19507 }, { "epoch": 1.04003838566935, "grad_norm": 0.6011625528335571, "learning_rate": 6.0986e-06, "loss": 0.0053, "step": 19508 }, { "epoch": 1.040091699099003, "grad_norm": 1.670512318611145, "learning_rate": 6.098400000000001e-06, "loss": 0.0287, "step": 19509 }, { "epoch": 1.040145012528656, "grad_norm": 1.7425469160079956, "learning_rate": 6.098200000000001e-06, "loss": 0.0241, "step": 19510 }, { "epoch": 1.040198325958309, "grad_norm": 2.167311906814575, "learning_rate": 6.098000000000001e-06, "loss": 0.0251, "step": 19511 }, { "epoch": 1.0402516393879617, "grad_norm": 1.8894826173782349, "learning_rate": 6.0978e-06, "loss": 0.0473, "step": 19512 }, { "epoch": 1.0403049528176147, "grad_norm": 2.3428242206573486, "learning_rate": 6.0976e-06, "loss": 0.031, "step": 19513 }, { "epoch": 1.0403582662472677, "grad_norm": 4.605149269104004, "learning_rate": 6.0974e-06, "loss": 0.0385, "step": 19514 }, { "epoch": 1.0404115796769207, "grad_norm": 2.397024631500244, "learning_rate": 6.097200000000001e-06, "loss": 0.0415, "step": 19515 }, { "epoch": 1.0404648931065736, "grad_norm": 1.518808364868164, "learning_rate": 6.097000000000001e-06, "loss": 0.0174, "step": 19516 }, { "epoch": 1.0405182065362264, "grad_norm": 1.7643132209777832, "learning_rate": 6.096800000000001e-06, "loss": 0.0255, "step": 19517 }, { "epoch": 1.0405715199658794, "grad_norm": 1.9354666471481323, "learning_rate": 6.0966e-06, "loss": 0.0375, "step": 19518 }, { "epoch": 1.0406248333955324, "grad_norm": 0.9211205840110779, "learning_rate": 6.0964e-06, "loss": 0.0176, "step": 19519 }, { "epoch": 1.0406781468251853, "grad_norm": 2.896043062210083, "learning_rate": 6.096200000000001e-06, "loss": 0.034, "step": 19520 }, { "epoch": 1.040731460254838, "grad_norm": 2.7584915161132812, "learning_rate": 6.096000000000001e-06, "loss": 0.031, "step": 19521 }, { "epoch": 1.040784773684491, "grad_norm": 3.271596670150757, "learning_rate": 6.095800000000001e-06, "loss": 0.0652, "step": 19522 }, { "epoch": 1.040838087114144, "grad_norm": 1.2720694541931152, "learning_rate": 6.0956000000000006e-06, "loss": 0.0237, "step": 19523 }, { "epoch": 1.040891400543797, "grad_norm": 1.925061821937561, "learning_rate": 6.0954e-06, "loss": 0.0433, "step": 19524 }, { "epoch": 1.04094471397345, "grad_norm": 1.1559447050094604, "learning_rate": 6.0952e-06, "loss": 0.0126, "step": 19525 }, { "epoch": 1.0409980274031028, "grad_norm": 3.8574676513671875, "learning_rate": 6.095000000000001e-06, "loss": 0.0571, "step": 19526 }, { "epoch": 1.0410513408327557, "grad_norm": 3.4631717205047607, "learning_rate": 6.0948000000000005e-06, "loss": 0.0554, "step": 19527 }, { "epoch": 1.0411046542624087, "grad_norm": 1.9153530597686768, "learning_rate": 6.094600000000001e-06, "loss": 0.0418, "step": 19528 }, { "epoch": 1.0411579676920617, "grad_norm": 1.2993519306182861, "learning_rate": 6.0944000000000005e-06, "loss": 0.0239, "step": 19529 }, { "epoch": 1.0412112811217145, "grad_norm": 2.901038646697998, "learning_rate": 6.0942e-06, "loss": 0.0299, "step": 19530 }, { "epoch": 1.0412645945513674, "grad_norm": 2.5275988578796387, "learning_rate": 6.094000000000001e-06, "loss": 0.0498, "step": 19531 }, { "epoch": 1.0413179079810204, "grad_norm": 4.43385124206543, "learning_rate": 6.093800000000001e-06, "loss": 0.044, "step": 19532 }, { "epoch": 1.0413712214106734, "grad_norm": 1.6737308502197266, "learning_rate": 6.0936e-06, "loss": 0.0324, "step": 19533 }, { "epoch": 1.0414245348403264, "grad_norm": 2.0381920337677, "learning_rate": 6.093400000000001e-06, "loss": 0.0469, "step": 19534 }, { "epoch": 1.0414778482699791, "grad_norm": 1.6959117650985718, "learning_rate": 6.0932e-06, "loss": 0.0197, "step": 19535 }, { "epoch": 1.0415311616996321, "grad_norm": 1.7050001621246338, "learning_rate": 6.093e-06, "loss": 0.029, "step": 19536 }, { "epoch": 1.041584475129285, "grad_norm": 5.311837196350098, "learning_rate": 6.092800000000001e-06, "loss": 0.0577, "step": 19537 }, { "epoch": 1.041637788558938, "grad_norm": 2.193976879119873, "learning_rate": 6.092600000000001e-06, "loss": 0.0233, "step": 19538 }, { "epoch": 1.0416911019885908, "grad_norm": 2.1846582889556885, "learning_rate": 6.0924e-06, "loss": 0.0487, "step": 19539 }, { "epoch": 1.0417444154182438, "grad_norm": 2.376462697982788, "learning_rate": 6.092200000000001e-06, "loss": 0.0278, "step": 19540 }, { "epoch": 1.0417977288478968, "grad_norm": 1.7284029722213745, "learning_rate": 6.092e-06, "loss": 0.0252, "step": 19541 }, { "epoch": 1.0418510422775498, "grad_norm": 1.7794138193130493, "learning_rate": 6.091800000000001e-06, "loss": 0.0212, "step": 19542 }, { "epoch": 1.0419043557072027, "grad_norm": 2.9012818336486816, "learning_rate": 6.091600000000001e-06, "loss": 0.0358, "step": 19543 }, { "epoch": 1.0419576691368555, "grad_norm": 5.223454475402832, "learning_rate": 6.091400000000001e-06, "loss": 0.0564, "step": 19544 }, { "epoch": 1.0420109825665085, "grad_norm": 1.1037718057632446, "learning_rate": 6.0912e-06, "loss": 0.0101, "step": 19545 }, { "epoch": 1.0420642959961615, "grad_norm": 1.6243611574172974, "learning_rate": 6.091000000000001e-06, "loss": 0.0326, "step": 19546 }, { "epoch": 1.0421176094258144, "grad_norm": 1.2591404914855957, "learning_rate": 6.0908e-06, "loss": 0.0151, "step": 19547 }, { "epoch": 1.0421709228554672, "grad_norm": 1.4682427644729614, "learning_rate": 6.090600000000001e-06, "loss": 0.0255, "step": 19548 }, { "epoch": 1.0422242362851202, "grad_norm": 3.763605833053589, "learning_rate": 6.090400000000001e-06, "loss": 0.0353, "step": 19549 }, { "epoch": 1.0422775497147732, "grad_norm": 1.6582814455032349, "learning_rate": 6.0902000000000006e-06, "loss": 0.0272, "step": 19550 }, { "epoch": 1.0423308631444261, "grad_norm": 1.7206628322601318, "learning_rate": 6.09e-06, "loss": 0.0205, "step": 19551 }, { "epoch": 1.0423841765740791, "grad_norm": 3.107097625732422, "learning_rate": 6.0898000000000005e-06, "loss": 0.0468, "step": 19552 }, { "epoch": 1.0424374900037319, "grad_norm": 3.0189619064331055, "learning_rate": 6.0896e-06, "loss": 0.0263, "step": 19553 }, { "epoch": 1.0424908034333848, "grad_norm": 1.3141796588897705, "learning_rate": 6.0894000000000005e-06, "loss": 0.0176, "step": 19554 }, { "epoch": 1.0425441168630378, "grad_norm": 2.4832675457000732, "learning_rate": 6.089200000000001e-06, "loss": 0.0573, "step": 19555 }, { "epoch": 1.0425974302926908, "grad_norm": 2.2063796520233154, "learning_rate": 6.0890000000000005e-06, "loss": 0.0483, "step": 19556 }, { "epoch": 1.0426507437223436, "grad_norm": 1.1573078632354736, "learning_rate": 6.0888e-06, "loss": 0.0109, "step": 19557 }, { "epoch": 1.0427040571519965, "grad_norm": 1.8354424238204956, "learning_rate": 6.0886000000000005e-06, "loss": 0.0195, "step": 19558 }, { "epoch": 1.0427573705816495, "grad_norm": 2.995116710662842, "learning_rate": 6.088400000000001e-06, "loss": 0.031, "step": 19559 }, { "epoch": 1.0428106840113025, "grad_norm": 2.270441770553589, "learning_rate": 6.0882000000000004e-06, "loss": 0.0158, "step": 19560 }, { "epoch": 1.0428639974409555, "grad_norm": 1.6814351081848145, "learning_rate": 6.088000000000001e-06, "loss": 0.0322, "step": 19561 }, { "epoch": 1.0429173108706082, "grad_norm": 1.8353971242904663, "learning_rate": 6.0878e-06, "loss": 0.0167, "step": 19562 }, { "epoch": 1.0429706243002612, "grad_norm": 1.3929603099822998, "learning_rate": 6.0876e-06, "loss": 0.0314, "step": 19563 }, { "epoch": 1.0430239377299142, "grad_norm": 2.2221105098724365, "learning_rate": 6.0874e-06, "loss": 0.0443, "step": 19564 }, { "epoch": 1.0430772511595672, "grad_norm": 2.2146482467651367, "learning_rate": 6.087200000000001e-06, "loss": 0.0417, "step": 19565 }, { "epoch": 1.04313056458922, "grad_norm": 1.2347594499588013, "learning_rate": 6.087e-06, "loss": 0.009, "step": 19566 }, { "epoch": 1.043183878018873, "grad_norm": 0.33792442083358765, "learning_rate": 6.086800000000001e-06, "loss": 0.0034, "step": 19567 }, { "epoch": 1.043237191448526, "grad_norm": 1.2230796813964844, "learning_rate": 6.0866e-06, "loss": 0.0378, "step": 19568 }, { "epoch": 1.0432905048781789, "grad_norm": 2.9508895874023438, "learning_rate": 6.0864e-06, "loss": 0.0327, "step": 19569 }, { "epoch": 1.0433438183078318, "grad_norm": 1.8440673351287842, "learning_rate": 6.086200000000001e-06, "loss": 0.0146, "step": 19570 }, { "epoch": 1.0433971317374846, "grad_norm": 2.743192434310913, "learning_rate": 6.086000000000001e-06, "loss": 0.0307, "step": 19571 }, { "epoch": 1.0434504451671376, "grad_norm": 1.2207927703857422, "learning_rate": 6.0858e-06, "loss": 0.0139, "step": 19572 }, { "epoch": 1.0435037585967906, "grad_norm": 3.1602225303649902, "learning_rate": 6.085600000000001e-06, "loss": 0.0501, "step": 19573 }, { "epoch": 1.0435570720264435, "grad_norm": 2.6926193237304688, "learning_rate": 6.0854e-06, "loss": 0.0417, "step": 19574 }, { "epoch": 1.0436103854560963, "grad_norm": 2.738938093185425, "learning_rate": 6.0852e-06, "loss": 0.0296, "step": 19575 }, { "epoch": 1.0436636988857493, "grad_norm": 3.149324893951416, "learning_rate": 6.085000000000001e-06, "loss": 0.0297, "step": 19576 }, { "epoch": 1.0437170123154023, "grad_norm": 1.1743147373199463, "learning_rate": 6.084800000000001e-06, "loss": 0.0144, "step": 19577 }, { "epoch": 1.0437703257450552, "grad_norm": 4.076149940490723, "learning_rate": 6.0846e-06, "loss": 0.039, "step": 19578 }, { "epoch": 1.043823639174708, "grad_norm": 1.1300315856933594, "learning_rate": 6.0844000000000006e-06, "loss": 0.0103, "step": 19579 }, { "epoch": 1.043876952604361, "grad_norm": 1.349006175994873, "learning_rate": 6.0842e-06, "loss": 0.0296, "step": 19580 }, { "epoch": 1.043930266034014, "grad_norm": 1.7786177396774292, "learning_rate": 6.084000000000001e-06, "loss": 0.0216, "step": 19581 }, { "epoch": 1.043983579463667, "grad_norm": 3.45015811920166, "learning_rate": 6.083800000000001e-06, "loss": 0.036, "step": 19582 }, { "epoch": 1.04403689289332, "grad_norm": 1.820353388786316, "learning_rate": 6.0836000000000005e-06, "loss": 0.0253, "step": 19583 }, { "epoch": 1.0440902063229727, "grad_norm": 1.2093799114227295, "learning_rate": 6.0834e-06, "loss": 0.0167, "step": 19584 }, { "epoch": 1.0441435197526256, "grad_norm": 3.0554916858673096, "learning_rate": 6.0832000000000005e-06, "loss": 0.0607, "step": 19585 }, { "epoch": 1.0441968331822786, "grad_norm": 1.2996602058410645, "learning_rate": 6.083e-06, "loss": 0.0071, "step": 19586 }, { "epoch": 1.0442501466119316, "grad_norm": 1.2885453701019287, "learning_rate": 6.082800000000001e-06, "loss": 0.0199, "step": 19587 }, { "epoch": 1.0443034600415846, "grad_norm": 2.597316026687622, "learning_rate": 6.082600000000001e-06, "loss": 0.0543, "step": 19588 }, { "epoch": 1.0443567734712373, "grad_norm": 5.075491428375244, "learning_rate": 6.0824e-06, "loss": 0.0409, "step": 19589 }, { "epoch": 1.0444100869008903, "grad_norm": 2.11628794670105, "learning_rate": 6.0822e-06, "loss": 0.0306, "step": 19590 }, { "epoch": 1.0444634003305433, "grad_norm": 1.8217341899871826, "learning_rate": 6.082e-06, "loss": 0.034, "step": 19591 }, { "epoch": 1.0445167137601963, "grad_norm": 1.5021636486053467, "learning_rate": 6.081800000000001e-06, "loss": 0.0195, "step": 19592 }, { "epoch": 1.044570027189849, "grad_norm": 1.212812900543213, "learning_rate": 6.081600000000001e-06, "loss": 0.0154, "step": 19593 }, { "epoch": 1.044623340619502, "grad_norm": 1.543363094329834, "learning_rate": 6.081400000000001e-06, "loss": 0.0141, "step": 19594 }, { "epoch": 1.044676654049155, "grad_norm": 1.7019137144088745, "learning_rate": 6.0812e-06, "loss": 0.0227, "step": 19595 }, { "epoch": 1.044729967478808, "grad_norm": 1.8923001289367676, "learning_rate": 6.081e-06, "loss": 0.0155, "step": 19596 }, { "epoch": 1.0447832809084607, "grad_norm": 1.9704339504241943, "learning_rate": 6.0808e-06, "loss": 0.0259, "step": 19597 }, { "epoch": 1.0448365943381137, "grad_norm": 1.3505879640579224, "learning_rate": 6.080600000000001e-06, "loss": 0.0215, "step": 19598 }, { "epoch": 1.0448899077677667, "grad_norm": 1.9010204076766968, "learning_rate": 6.080400000000001e-06, "loss": 0.0431, "step": 19599 }, { "epoch": 1.0449432211974197, "grad_norm": 2.5672919750213623, "learning_rate": 6.080200000000001e-06, "loss": 0.0412, "step": 19600 }, { "epoch": 1.0449965346270726, "grad_norm": 1.864399790763855, "learning_rate": 6.08e-06, "loss": 0.0319, "step": 19601 }, { "epoch": 1.0450498480567254, "grad_norm": 0.6254717707633972, "learning_rate": 6.0798e-06, "loss": 0.0057, "step": 19602 }, { "epoch": 1.0451031614863784, "grad_norm": 1.2973887920379639, "learning_rate": 6.0796e-06, "loss": 0.0181, "step": 19603 }, { "epoch": 1.0451564749160314, "grad_norm": 1.0006542205810547, "learning_rate": 6.079400000000001e-06, "loss": 0.0164, "step": 19604 }, { "epoch": 1.0452097883456843, "grad_norm": 2.1018338203430176, "learning_rate": 6.079200000000001e-06, "loss": 0.0278, "step": 19605 }, { "epoch": 1.045263101775337, "grad_norm": 4.659823417663574, "learning_rate": 6.079000000000001e-06, "loss": 0.039, "step": 19606 }, { "epoch": 1.04531641520499, "grad_norm": 1.5934984683990479, "learning_rate": 6.0788e-06, "loss": 0.0274, "step": 19607 }, { "epoch": 1.045369728634643, "grad_norm": 2.472075939178467, "learning_rate": 6.0786e-06, "loss": 0.0519, "step": 19608 }, { "epoch": 1.045423042064296, "grad_norm": 0.4199928641319275, "learning_rate": 6.078400000000001e-06, "loss": 0.004, "step": 19609 }, { "epoch": 1.045476355493949, "grad_norm": 2.0661022663116455, "learning_rate": 6.0782000000000005e-06, "loss": 0.045, "step": 19610 }, { "epoch": 1.0455296689236018, "grad_norm": 2.365628242492676, "learning_rate": 6.078000000000001e-06, "loss": 0.0317, "step": 19611 }, { "epoch": 1.0455829823532548, "grad_norm": 1.2738837003707886, "learning_rate": 6.0778000000000005e-06, "loss": 0.0404, "step": 19612 }, { "epoch": 1.0456362957829077, "grad_norm": 3.059727668762207, "learning_rate": 6.0776e-06, "loss": 0.0312, "step": 19613 }, { "epoch": 1.0456896092125607, "grad_norm": 2.351365327835083, "learning_rate": 6.0774e-06, "loss": 0.037, "step": 19614 }, { "epoch": 1.0457429226422135, "grad_norm": 2.0031826496124268, "learning_rate": 6.077200000000001e-06, "loss": 0.0667, "step": 19615 }, { "epoch": 1.0457962360718664, "grad_norm": 1.5094072818756104, "learning_rate": 6.0770000000000004e-06, "loss": 0.0304, "step": 19616 }, { "epoch": 1.0458495495015194, "grad_norm": 1.1395338773727417, "learning_rate": 6.076800000000001e-06, "loss": 0.0127, "step": 19617 }, { "epoch": 1.0459028629311724, "grad_norm": 0.5482043623924255, "learning_rate": 6.0766e-06, "loss": 0.0057, "step": 19618 }, { "epoch": 1.0459561763608254, "grad_norm": 2.1334316730499268, "learning_rate": 6.0764e-06, "loss": 0.0227, "step": 19619 }, { "epoch": 1.0460094897904781, "grad_norm": 1.7820371389389038, "learning_rate": 6.076200000000001e-06, "loss": 0.0321, "step": 19620 }, { "epoch": 1.0460628032201311, "grad_norm": 2.034658193588257, "learning_rate": 6.076000000000001e-06, "loss": 0.0517, "step": 19621 }, { "epoch": 1.046116116649784, "grad_norm": 1.6951699256896973, "learning_rate": 6.0758e-06, "loss": 0.0172, "step": 19622 }, { "epoch": 1.046169430079437, "grad_norm": 0.7015511989593506, "learning_rate": 6.075600000000001e-06, "loss": 0.0115, "step": 19623 }, { "epoch": 1.0462227435090898, "grad_norm": 1.7736155986785889, "learning_rate": 6.0754e-06, "loss": 0.0257, "step": 19624 }, { "epoch": 1.0462760569387428, "grad_norm": 1.107335090637207, "learning_rate": 6.0752e-06, "loss": 0.0119, "step": 19625 }, { "epoch": 1.0463293703683958, "grad_norm": 1.0608091354370117, "learning_rate": 6.075000000000001e-06, "loss": 0.0202, "step": 19626 }, { "epoch": 1.0463826837980488, "grad_norm": 1.7242014408111572, "learning_rate": 6.074800000000001e-06, "loss": 0.0153, "step": 19627 }, { "epoch": 1.0464359972277018, "grad_norm": 2.1852025985717773, "learning_rate": 6.0746e-06, "loss": 0.0281, "step": 19628 }, { "epoch": 1.0464893106573545, "grad_norm": 2.2499516010284424, "learning_rate": 6.074400000000001e-06, "loss": 0.0414, "step": 19629 }, { "epoch": 1.0465426240870075, "grad_norm": 1.2357674837112427, "learning_rate": 6.0742e-06, "loss": 0.0286, "step": 19630 }, { "epoch": 1.0465959375166605, "grad_norm": 0.5556276440620422, "learning_rate": 6.074000000000001e-06, "loss": 0.0111, "step": 19631 }, { "epoch": 1.0466492509463134, "grad_norm": 2.2077136039733887, "learning_rate": 6.073800000000001e-06, "loss": 0.0253, "step": 19632 }, { "epoch": 1.0467025643759662, "grad_norm": 1.5703372955322266, "learning_rate": 6.073600000000001e-06, "loss": 0.0264, "step": 19633 }, { "epoch": 1.0467558778056192, "grad_norm": 1.5410287380218506, "learning_rate": 6.0734e-06, "loss": 0.0192, "step": 19634 }, { "epoch": 1.0468091912352722, "grad_norm": 1.8846389055252075, "learning_rate": 6.0732000000000006e-06, "loss": 0.0291, "step": 19635 }, { "epoch": 1.0468625046649251, "grad_norm": 1.3962229490280151, "learning_rate": 6.073e-06, "loss": 0.0288, "step": 19636 }, { "epoch": 1.0469158180945781, "grad_norm": 2.8810360431671143, "learning_rate": 6.0728000000000005e-06, "loss": 0.0297, "step": 19637 }, { "epoch": 1.0469691315242309, "grad_norm": 2.798344612121582, "learning_rate": 6.072600000000001e-06, "loss": 0.0342, "step": 19638 }, { "epoch": 1.0470224449538839, "grad_norm": 1.5043554306030273, "learning_rate": 6.0724000000000005e-06, "loss": 0.0328, "step": 19639 }, { "epoch": 1.0470757583835368, "grad_norm": 1.4340838193893433, "learning_rate": 6.0722e-06, "loss": 0.0141, "step": 19640 }, { "epoch": 1.0471290718131898, "grad_norm": 1.269355058670044, "learning_rate": 6.0720000000000005e-06, "loss": 0.0198, "step": 19641 }, { "epoch": 1.0471823852428426, "grad_norm": 1.2560019493103027, "learning_rate": 6.071800000000001e-06, "loss": 0.0121, "step": 19642 }, { "epoch": 1.0472356986724956, "grad_norm": 3.192007541656494, "learning_rate": 6.0716000000000004e-06, "loss": 0.0572, "step": 19643 }, { "epoch": 1.0472890121021485, "grad_norm": 1.5848678350448608, "learning_rate": 6.071400000000001e-06, "loss": 0.0353, "step": 19644 }, { "epoch": 1.0473423255318015, "grad_norm": 1.6342296600341797, "learning_rate": 6.0712e-06, "loss": 0.0194, "step": 19645 }, { "epoch": 1.0473956389614545, "grad_norm": 1.6190259456634521, "learning_rate": 6.071e-06, "loss": 0.017, "step": 19646 }, { "epoch": 1.0474489523911072, "grad_norm": 1.3266998529434204, "learning_rate": 6.0708e-06, "loss": 0.017, "step": 19647 }, { "epoch": 1.0475022658207602, "grad_norm": 2.3856606483459473, "learning_rate": 6.070600000000001e-06, "loss": 0.0612, "step": 19648 }, { "epoch": 1.0475555792504132, "grad_norm": 1.8518528938293457, "learning_rate": 6.0704e-06, "loss": 0.0278, "step": 19649 }, { "epoch": 1.0476088926800662, "grad_norm": 1.1258352994918823, "learning_rate": 6.070200000000001e-06, "loss": 0.0107, "step": 19650 }, { "epoch": 1.047662206109719, "grad_norm": 1.5911600589752197, "learning_rate": 6.07e-06, "loss": 0.0211, "step": 19651 }, { "epoch": 1.047715519539372, "grad_norm": 2.3176376819610596, "learning_rate": 6.0698e-06, "loss": 0.0423, "step": 19652 }, { "epoch": 1.047768832969025, "grad_norm": 1.9379483461380005, "learning_rate": 6.069600000000001e-06, "loss": 0.0418, "step": 19653 }, { "epoch": 1.0478221463986779, "grad_norm": 1.5883233547210693, "learning_rate": 6.069400000000001e-06, "loss": 0.0167, "step": 19654 }, { "epoch": 1.0478754598283309, "grad_norm": 2.169647216796875, "learning_rate": 6.0692e-06, "loss": 0.0147, "step": 19655 }, { "epoch": 1.0479287732579836, "grad_norm": 2.745779037475586, "learning_rate": 6.069000000000001e-06, "loss": 0.0405, "step": 19656 }, { "epoch": 1.0479820866876366, "grad_norm": 1.423433542251587, "learning_rate": 6.0688e-06, "loss": 0.0158, "step": 19657 }, { "epoch": 1.0480354001172896, "grad_norm": 1.3926007747650146, "learning_rate": 6.0686e-06, "loss": 0.0204, "step": 19658 }, { "epoch": 1.0480887135469426, "grad_norm": 1.0611416101455688, "learning_rate": 6.068400000000001e-06, "loss": 0.0224, "step": 19659 }, { "epoch": 1.0481420269765953, "grad_norm": 3.6102664470672607, "learning_rate": 6.068200000000001e-06, "loss": 0.0218, "step": 19660 }, { "epoch": 1.0481953404062483, "grad_norm": 1.650413990020752, "learning_rate": 6.068e-06, "loss": 0.0179, "step": 19661 }, { "epoch": 1.0482486538359013, "grad_norm": 1.1930453777313232, "learning_rate": 6.067800000000001e-06, "loss": 0.0101, "step": 19662 }, { "epoch": 1.0483019672655542, "grad_norm": 1.3250372409820557, "learning_rate": 6.0676e-06, "loss": 0.0292, "step": 19663 }, { "epoch": 1.0483552806952072, "grad_norm": 1.016381859779358, "learning_rate": 6.0674e-06, "loss": 0.0209, "step": 19664 }, { "epoch": 1.04840859412486, "grad_norm": 2.6074938774108887, "learning_rate": 6.067200000000001e-06, "loss": 0.043, "step": 19665 }, { "epoch": 1.048461907554513, "grad_norm": 1.0603595972061157, "learning_rate": 6.0670000000000005e-06, "loss": 0.0182, "step": 19666 }, { "epoch": 1.048515220984166, "grad_norm": 1.7839807271957397, "learning_rate": 6.0668e-06, "loss": 0.0107, "step": 19667 }, { "epoch": 1.048568534413819, "grad_norm": 2.733520746231079, "learning_rate": 6.0666000000000005e-06, "loss": 0.0292, "step": 19668 }, { "epoch": 1.0486218478434717, "grad_norm": 2.8820414543151855, "learning_rate": 6.0664e-06, "loss": 0.052, "step": 19669 }, { "epoch": 1.0486751612731247, "grad_norm": 1.2689849138259888, "learning_rate": 6.066200000000001e-06, "loss": 0.024, "step": 19670 }, { "epoch": 1.0487284747027776, "grad_norm": 0.9393661618232727, "learning_rate": 6.066000000000001e-06, "loss": 0.0067, "step": 19671 }, { "epoch": 1.0487817881324306, "grad_norm": 1.9000544548034668, "learning_rate": 6.0658000000000004e-06, "loss": 0.0245, "step": 19672 }, { "epoch": 1.0488351015620836, "grad_norm": 2.1591317653656006, "learning_rate": 6.0656e-06, "loss": 0.0248, "step": 19673 }, { "epoch": 1.0488884149917364, "grad_norm": 1.6396405696868896, "learning_rate": 6.0654e-06, "loss": 0.0124, "step": 19674 }, { "epoch": 1.0489417284213893, "grad_norm": 1.3405791521072388, "learning_rate": 6.0652e-06, "loss": 0.0147, "step": 19675 }, { "epoch": 1.0489950418510423, "grad_norm": 2.810384511947632, "learning_rate": 6.065000000000001e-06, "loss": 0.0225, "step": 19676 }, { "epoch": 1.0490483552806953, "grad_norm": 1.9993010759353638, "learning_rate": 6.064800000000001e-06, "loss": 0.0243, "step": 19677 }, { "epoch": 1.049101668710348, "grad_norm": 2.2036900520324707, "learning_rate": 6.0646e-06, "loss": 0.0176, "step": 19678 }, { "epoch": 1.049154982140001, "grad_norm": 1.482417106628418, "learning_rate": 6.0644e-06, "loss": 0.0276, "step": 19679 }, { "epoch": 1.049208295569654, "grad_norm": 2.496425151824951, "learning_rate": 6.0642e-06, "loss": 0.0492, "step": 19680 }, { "epoch": 1.049261608999307, "grad_norm": 1.871866226196289, "learning_rate": 6.064000000000001e-06, "loss": 0.0278, "step": 19681 }, { "epoch": 1.04931492242896, "grad_norm": 1.403472661972046, "learning_rate": 6.063800000000001e-06, "loss": 0.0228, "step": 19682 }, { "epoch": 1.0493682358586127, "grad_norm": 1.364678978919983, "learning_rate": 6.063600000000001e-06, "loss": 0.0186, "step": 19683 }, { "epoch": 1.0494215492882657, "grad_norm": 1.185851812362671, "learning_rate": 6.0634e-06, "loss": 0.0149, "step": 19684 }, { "epoch": 1.0494748627179187, "grad_norm": 2.3677825927734375, "learning_rate": 6.0632e-06, "loss": 0.0364, "step": 19685 }, { "epoch": 1.0495281761475717, "grad_norm": 1.6523727178573608, "learning_rate": 6.063e-06, "loss": 0.0118, "step": 19686 }, { "epoch": 1.0495814895772244, "grad_norm": 1.8789526224136353, "learning_rate": 6.062800000000001e-06, "loss": 0.0299, "step": 19687 }, { "epoch": 1.0496348030068774, "grad_norm": 1.6598104238510132, "learning_rate": 6.062600000000001e-06, "loss": 0.0183, "step": 19688 }, { "epoch": 1.0496881164365304, "grad_norm": 1.5454530715942383, "learning_rate": 6.062400000000001e-06, "loss": 0.0199, "step": 19689 }, { "epoch": 1.0497414298661834, "grad_norm": 1.2025964260101318, "learning_rate": 6.0622e-06, "loss": 0.034, "step": 19690 }, { "epoch": 1.0497947432958363, "grad_norm": 1.0362107753753662, "learning_rate": 6.062e-06, "loss": 0.0171, "step": 19691 }, { "epoch": 1.049848056725489, "grad_norm": 2.220768928527832, "learning_rate": 6.061800000000001e-06, "loss": 0.0359, "step": 19692 }, { "epoch": 1.049901370155142, "grad_norm": 1.6437702178955078, "learning_rate": 6.0616000000000005e-06, "loss": 0.0362, "step": 19693 }, { "epoch": 1.049954683584795, "grad_norm": 3.008683681488037, "learning_rate": 6.061400000000001e-06, "loss": 0.0439, "step": 19694 }, { "epoch": 1.050007997014448, "grad_norm": 3.451707601547241, "learning_rate": 6.0612000000000005e-06, "loss": 0.053, "step": 19695 }, { "epoch": 1.0500613104441008, "grad_norm": 2.564459800720215, "learning_rate": 6.061e-06, "loss": 0.0383, "step": 19696 }, { "epoch": 1.0501146238737538, "grad_norm": 1.251319169998169, "learning_rate": 6.0608e-06, "loss": 0.0255, "step": 19697 }, { "epoch": 1.0501679373034067, "grad_norm": 1.1784536838531494, "learning_rate": 6.060600000000001e-06, "loss": 0.0222, "step": 19698 }, { "epoch": 1.0502212507330597, "grad_norm": 3.162872791290283, "learning_rate": 6.0604000000000005e-06, "loss": 0.0397, "step": 19699 }, { "epoch": 1.0502745641627125, "grad_norm": 1.185443639755249, "learning_rate": 6.060200000000001e-06, "loss": 0.0191, "step": 19700 }, { "epoch": 1.0503278775923655, "grad_norm": 1.8605120182037354, "learning_rate": 6.0600000000000004e-06, "loss": 0.0237, "step": 19701 }, { "epoch": 1.0503811910220184, "grad_norm": 1.0030099153518677, "learning_rate": 6.0598e-06, "loss": 0.01, "step": 19702 }, { "epoch": 1.0504345044516714, "grad_norm": 1.7735168933868408, "learning_rate": 6.059600000000001e-06, "loss": 0.0512, "step": 19703 }, { "epoch": 1.0504878178813244, "grad_norm": 1.2382800579071045, "learning_rate": 6.059400000000001e-06, "loss": 0.0263, "step": 19704 }, { "epoch": 1.0505411313109771, "grad_norm": 0.5965563654899597, "learning_rate": 6.0592e-06, "loss": 0.0059, "step": 19705 }, { "epoch": 1.0505944447406301, "grad_norm": 2.2704882621765137, "learning_rate": 6.059000000000001e-06, "loss": 0.0296, "step": 19706 }, { "epoch": 1.050647758170283, "grad_norm": 1.500989556312561, "learning_rate": 6.0588e-06, "loss": 0.013, "step": 19707 }, { "epoch": 1.050701071599936, "grad_norm": 0.7615068554878235, "learning_rate": 6.0586e-06, "loss": 0.0123, "step": 19708 }, { "epoch": 1.050754385029589, "grad_norm": 1.2621549367904663, "learning_rate": 6.058400000000001e-06, "loss": 0.0235, "step": 19709 }, { "epoch": 1.0508076984592418, "grad_norm": 2.437222719192505, "learning_rate": 6.058200000000001e-06, "loss": 0.0246, "step": 19710 }, { "epoch": 1.0508610118888948, "grad_norm": 1.2818324565887451, "learning_rate": 6.058e-06, "loss": 0.0156, "step": 19711 }, { "epoch": 1.0509143253185478, "grad_norm": 1.1349238157272339, "learning_rate": 6.057800000000001e-06, "loss": 0.021, "step": 19712 }, { "epoch": 1.0509676387482008, "grad_norm": 1.0170844793319702, "learning_rate": 6.0576e-06, "loss": 0.0111, "step": 19713 }, { "epoch": 1.0510209521778535, "grad_norm": 2.394176483154297, "learning_rate": 6.0574e-06, "loss": 0.0412, "step": 19714 }, { "epoch": 1.0510742656075065, "grad_norm": 1.4899888038635254, "learning_rate": 6.057200000000001e-06, "loss": 0.0155, "step": 19715 }, { "epoch": 1.0511275790371595, "grad_norm": 0.9670782089233398, "learning_rate": 6.057000000000001e-06, "loss": 0.0127, "step": 19716 }, { "epoch": 1.0511808924668125, "grad_norm": 0.6649871468544006, "learning_rate": 6.0568e-06, "loss": 0.0049, "step": 19717 }, { "epoch": 1.0512342058964652, "grad_norm": 2.0696861743927, "learning_rate": 6.056600000000001e-06, "loss": 0.0212, "step": 19718 }, { "epoch": 1.0512875193261182, "grad_norm": 0.8501288890838623, "learning_rate": 6.0564e-06, "loss": 0.0118, "step": 19719 }, { "epoch": 1.0513408327557712, "grad_norm": 1.3224339485168457, "learning_rate": 6.0562000000000006e-06, "loss": 0.0096, "step": 19720 }, { "epoch": 1.0513941461854241, "grad_norm": 0.8652468323707581, "learning_rate": 6.056000000000001e-06, "loss": 0.0069, "step": 19721 }, { "epoch": 1.0514474596150771, "grad_norm": 4.699671745300293, "learning_rate": 6.0558000000000005e-06, "loss": 0.03, "step": 19722 }, { "epoch": 1.0515007730447299, "grad_norm": 1.9163358211517334, "learning_rate": 6.0556e-06, "loss": 0.0312, "step": 19723 }, { "epoch": 1.0515540864743829, "grad_norm": 1.609910488128662, "learning_rate": 6.0554000000000005e-06, "loss": 0.0234, "step": 19724 }, { "epoch": 1.0516073999040358, "grad_norm": 1.839613914489746, "learning_rate": 6.0552e-06, "loss": 0.0288, "step": 19725 }, { "epoch": 1.0516607133336888, "grad_norm": 1.3802905082702637, "learning_rate": 6.0550000000000005e-06, "loss": 0.0238, "step": 19726 }, { "epoch": 1.0517140267633416, "grad_norm": 2.0688560009002686, "learning_rate": 6.054800000000001e-06, "loss": 0.0436, "step": 19727 }, { "epoch": 1.0517673401929946, "grad_norm": 1.097242832183838, "learning_rate": 6.0546000000000004e-06, "loss": 0.0084, "step": 19728 }, { "epoch": 1.0518206536226475, "grad_norm": 1.4617348909378052, "learning_rate": 6.0544e-06, "loss": 0.0196, "step": 19729 }, { "epoch": 1.0518739670523005, "grad_norm": 3.1716675758361816, "learning_rate": 6.0542e-06, "loss": 0.0588, "step": 19730 }, { "epoch": 1.0519272804819535, "grad_norm": 1.0028561353683472, "learning_rate": 6.054000000000001e-06, "loss": 0.0105, "step": 19731 }, { "epoch": 1.0519805939116063, "grad_norm": 2.134816884994507, "learning_rate": 6.0538e-06, "loss": 0.0131, "step": 19732 }, { "epoch": 1.0520339073412592, "grad_norm": 1.1267662048339844, "learning_rate": 6.053600000000001e-06, "loss": 0.0105, "step": 19733 }, { "epoch": 1.0520872207709122, "grad_norm": 3.3631279468536377, "learning_rate": 6.0534e-06, "loss": 0.0456, "step": 19734 }, { "epoch": 1.0521405342005652, "grad_norm": 1.790520429611206, "learning_rate": 6.0532e-06, "loss": 0.0334, "step": 19735 }, { "epoch": 1.052193847630218, "grad_norm": 4.808870792388916, "learning_rate": 6.053e-06, "loss": 0.0526, "step": 19736 }, { "epoch": 1.052247161059871, "grad_norm": 4.20173978805542, "learning_rate": 6.052800000000001e-06, "loss": 0.0474, "step": 19737 }, { "epoch": 1.052300474489524, "grad_norm": 4.990987300872803, "learning_rate": 6.0526e-06, "loss": 0.0454, "step": 19738 }, { "epoch": 1.0523537879191769, "grad_norm": 1.9217063188552856, "learning_rate": 6.052400000000001e-06, "loss": 0.0245, "step": 19739 }, { "epoch": 1.0524071013488299, "grad_norm": 1.3652753829956055, "learning_rate": 6.0522e-06, "loss": 0.0203, "step": 19740 }, { "epoch": 1.0524604147784826, "grad_norm": 1.7041568756103516, "learning_rate": 6.052e-06, "loss": 0.041, "step": 19741 }, { "epoch": 1.0525137282081356, "grad_norm": 2.1363587379455566, "learning_rate": 6.051800000000001e-06, "loss": 0.0409, "step": 19742 }, { "epoch": 1.0525670416377886, "grad_norm": 0.5725778341293335, "learning_rate": 6.051600000000001e-06, "loss": 0.0056, "step": 19743 }, { "epoch": 1.0526203550674416, "grad_norm": 1.044909954071045, "learning_rate": 6.0514e-06, "loss": 0.0156, "step": 19744 }, { "epoch": 1.0526736684970943, "grad_norm": 3.8959436416625977, "learning_rate": 6.051200000000001e-06, "loss": 0.0633, "step": 19745 }, { "epoch": 1.0527269819267473, "grad_norm": 2.0331783294677734, "learning_rate": 6.051e-06, "loss": 0.0261, "step": 19746 }, { "epoch": 1.0527802953564003, "grad_norm": 2.45512318611145, "learning_rate": 6.0508e-06, "loss": 0.0563, "step": 19747 }, { "epoch": 1.0528336087860533, "grad_norm": 1.6251522302627563, "learning_rate": 6.050600000000001e-06, "loss": 0.0308, "step": 19748 }, { "epoch": 1.0528869222157062, "grad_norm": 2.4461469650268555, "learning_rate": 6.0504000000000006e-06, "loss": 0.0286, "step": 19749 }, { "epoch": 1.052940235645359, "grad_norm": 1.7409321069717407, "learning_rate": 6.0502e-06, "loss": 0.0306, "step": 19750 }, { "epoch": 1.052993549075012, "grad_norm": 0.5240349769592285, "learning_rate": 6.0500000000000005e-06, "loss": 0.0063, "step": 19751 }, { "epoch": 1.053046862504665, "grad_norm": 1.3771127462387085, "learning_rate": 6.0498e-06, "loss": 0.0291, "step": 19752 }, { "epoch": 1.053100175934318, "grad_norm": 2.360382080078125, "learning_rate": 6.049600000000001e-06, "loss": 0.0481, "step": 19753 }, { "epoch": 1.0531534893639707, "grad_norm": 1.8240991830825806, "learning_rate": 6.049400000000001e-06, "loss": 0.0287, "step": 19754 }, { "epoch": 1.0532068027936237, "grad_norm": 3.717633008956909, "learning_rate": 6.0492000000000005e-06, "loss": 0.0398, "step": 19755 }, { "epoch": 1.0532601162232766, "grad_norm": 1.8092201948165894, "learning_rate": 6.049e-06, "loss": 0.0175, "step": 19756 }, { "epoch": 1.0533134296529296, "grad_norm": 2.023658275604248, "learning_rate": 6.0488000000000004e-06, "loss": 0.0622, "step": 19757 }, { "epoch": 1.0533667430825826, "grad_norm": 1.5550386905670166, "learning_rate": 6.0486e-06, "loss": 0.0305, "step": 19758 }, { "epoch": 1.0534200565122354, "grad_norm": 2.236891984939575, "learning_rate": 6.048400000000001e-06, "loss": 0.0371, "step": 19759 }, { "epoch": 1.0534733699418883, "grad_norm": 1.2353181838989258, "learning_rate": 6.048200000000001e-06, "loss": 0.0187, "step": 19760 }, { "epoch": 1.0535266833715413, "grad_norm": 1.348266839981079, "learning_rate": 6.048e-06, "loss": 0.0215, "step": 19761 }, { "epoch": 1.0535799968011943, "grad_norm": 1.7228813171386719, "learning_rate": 6.0478e-06, "loss": 0.0337, "step": 19762 }, { "epoch": 1.053633310230847, "grad_norm": 2.4705898761749268, "learning_rate": 6.0476e-06, "loss": 0.0504, "step": 19763 }, { "epoch": 1.0536866236605, "grad_norm": 2.251253604888916, "learning_rate": 6.047400000000001e-06, "loss": 0.0557, "step": 19764 }, { "epoch": 1.053739937090153, "grad_norm": 1.1346988677978516, "learning_rate": 6.047200000000001e-06, "loss": 0.0346, "step": 19765 }, { "epoch": 1.053793250519806, "grad_norm": 2.4902164936065674, "learning_rate": 6.047000000000001e-06, "loss": 0.0504, "step": 19766 }, { "epoch": 1.053846563949459, "grad_norm": 1.8382697105407715, "learning_rate": 6.0468e-06, "loss": 0.0284, "step": 19767 }, { "epoch": 1.0538998773791117, "grad_norm": 1.2690281867980957, "learning_rate": 6.0466e-06, "loss": 0.0188, "step": 19768 }, { "epoch": 1.0539531908087647, "grad_norm": 1.5311821699142456, "learning_rate": 6.0464e-06, "loss": 0.0173, "step": 19769 }, { "epoch": 1.0540065042384177, "grad_norm": 1.4159618616104126, "learning_rate": 6.046200000000001e-06, "loss": 0.0267, "step": 19770 }, { "epoch": 1.0540598176680707, "grad_norm": 1.0529402494430542, "learning_rate": 6.046000000000001e-06, "loss": 0.0192, "step": 19771 }, { "epoch": 1.0541131310977234, "grad_norm": 1.7240444421768188, "learning_rate": 6.045800000000001e-06, "loss": 0.0115, "step": 19772 }, { "epoch": 1.0541664445273764, "grad_norm": 1.5081969499588013, "learning_rate": 6.0456e-06, "loss": 0.0435, "step": 19773 }, { "epoch": 1.0542197579570294, "grad_norm": 1.4399253129959106, "learning_rate": 6.0454e-06, "loss": 0.0401, "step": 19774 }, { "epoch": 1.0542730713866824, "grad_norm": 1.8110878467559814, "learning_rate": 6.0452e-06, "loss": 0.0205, "step": 19775 }, { "epoch": 1.0543263848163353, "grad_norm": 1.9926623106002808, "learning_rate": 6.0450000000000006e-06, "loss": 0.0301, "step": 19776 }, { "epoch": 1.054379698245988, "grad_norm": 1.470987319946289, "learning_rate": 6.044800000000001e-06, "loss": 0.0312, "step": 19777 }, { "epoch": 1.054433011675641, "grad_norm": 2.148322582244873, "learning_rate": 6.0446000000000005e-06, "loss": 0.0435, "step": 19778 }, { "epoch": 1.054486325105294, "grad_norm": 1.0448895692825317, "learning_rate": 6.0444e-06, "loss": 0.0071, "step": 19779 }, { "epoch": 1.054539638534947, "grad_norm": 2.2165377140045166, "learning_rate": 6.0442e-06, "loss": 0.0406, "step": 19780 }, { "epoch": 1.0545929519645998, "grad_norm": 1.6223942041397095, "learning_rate": 6.044000000000001e-06, "loss": 0.031, "step": 19781 }, { "epoch": 1.0546462653942528, "grad_norm": 1.6664918661117554, "learning_rate": 6.0438000000000005e-06, "loss": 0.0429, "step": 19782 }, { "epoch": 1.0546995788239057, "grad_norm": 2.7282683849334717, "learning_rate": 6.043600000000001e-06, "loss": 0.0216, "step": 19783 }, { "epoch": 1.0547528922535587, "grad_norm": 1.6393523216247559, "learning_rate": 6.0434000000000004e-06, "loss": 0.0298, "step": 19784 }, { "epoch": 1.0548062056832117, "grad_norm": 1.2035020589828491, "learning_rate": 6.0432e-06, "loss": 0.0239, "step": 19785 }, { "epoch": 1.0548595191128645, "grad_norm": 1.6784558296203613, "learning_rate": 6.0429999999999996e-06, "loss": 0.0464, "step": 19786 }, { "epoch": 1.0549128325425174, "grad_norm": 1.6981215476989746, "learning_rate": 6.042800000000001e-06, "loss": 0.0382, "step": 19787 }, { "epoch": 1.0549661459721704, "grad_norm": 3.011922597885132, "learning_rate": 6.0426e-06, "loss": 0.0496, "step": 19788 }, { "epoch": 1.0550194594018234, "grad_norm": 1.3403431177139282, "learning_rate": 6.042400000000001e-06, "loss": 0.0146, "step": 19789 }, { "epoch": 1.0550727728314762, "grad_norm": 2.1769747734069824, "learning_rate": 6.0422e-06, "loss": 0.0354, "step": 19790 }, { "epoch": 1.0551260862611291, "grad_norm": 1.2088865041732788, "learning_rate": 6.042e-06, "loss": 0.0216, "step": 19791 }, { "epoch": 1.0551793996907821, "grad_norm": 2.0195724964141846, "learning_rate": 6.041800000000001e-06, "loss": 0.0271, "step": 19792 }, { "epoch": 1.055232713120435, "grad_norm": 1.4427024126052856, "learning_rate": 6.041600000000001e-06, "loss": 0.0225, "step": 19793 }, { "epoch": 1.055286026550088, "grad_norm": 1.7483514547348022, "learning_rate": 6.0414e-06, "loss": 0.0436, "step": 19794 }, { "epoch": 1.0553393399797408, "grad_norm": 5.820172309875488, "learning_rate": 6.041200000000001e-06, "loss": 0.0458, "step": 19795 }, { "epoch": 1.0553926534093938, "grad_norm": 5.073439121246338, "learning_rate": 6.041e-06, "loss": 0.0434, "step": 19796 }, { "epoch": 1.0554459668390468, "grad_norm": 2.1509368419647217, "learning_rate": 6.0408e-06, "loss": 0.0421, "step": 19797 }, { "epoch": 1.0554992802686998, "grad_norm": 0.7630413770675659, "learning_rate": 6.040600000000001e-06, "loss": 0.0331, "step": 19798 }, { "epoch": 1.0555525936983525, "grad_norm": 1.0480458736419678, "learning_rate": 6.040400000000001e-06, "loss": 0.0169, "step": 19799 }, { "epoch": 1.0556059071280055, "grad_norm": 2.0957517623901367, "learning_rate": 6.0402e-06, "loss": 0.0264, "step": 19800 }, { "epoch": 1.0556592205576585, "grad_norm": 1.0903199911117554, "learning_rate": 6.040000000000001e-06, "loss": 0.0225, "step": 19801 }, { "epoch": 1.0557125339873115, "grad_norm": 1.8625813722610474, "learning_rate": 6.0398e-06, "loss": 0.0313, "step": 19802 }, { "epoch": 1.0557658474169642, "grad_norm": 2.8099656105041504, "learning_rate": 6.0396000000000014e-06, "loss": 0.0195, "step": 19803 }, { "epoch": 1.0558191608466172, "grad_norm": 1.1802008152008057, "learning_rate": 6.039400000000001e-06, "loss": 0.0117, "step": 19804 }, { "epoch": 1.0558724742762702, "grad_norm": 1.9652093648910522, "learning_rate": 6.0392000000000006e-06, "loss": 0.0427, "step": 19805 }, { "epoch": 1.0559257877059232, "grad_norm": 2.1611921787261963, "learning_rate": 6.039e-06, "loss": 0.0289, "step": 19806 }, { "epoch": 1.0559791011355761, "grad_norm": 0.9377380609512329, "learning_rate": 6.0388000000000005e-06, "loss": 0.0075, "step": 19807 }, { "epoch": 1.056032414565229, "grad_norm": 1.3630321025848389, "learning_rate": 6.0386e-06, "loss": 0.0281, "step": 19808 }, { "epoch": 1.0560857279948819, "grad_norm": 1.281422734260559, "learning_rate": 6.038400000000001e-06, "loss": 0.0139, "step": 19809 }, { "epoch": 1.0561390414245349, "grad_norm": 2.9181201457977295, "learning_rate": 6.038200000000001e-06, "loss": 0.0679, "step": 19810 }, { "epoch": 1.0561923548541878, "grad_norm": 1.2266098260879517, "learning_rate": 6.0380000000000005e-06, "loss": 0.0255, "step": 19811 }, { "epoch": 1.0562456682838408, "grad_norm": 1.6856321096420288, "learning_rate": 6.0378e-06, "loss": 0.0212, "step": 19812 }, { "epoch": 1.0562989817134936, "grad_norm": 1.8577641248703003, "learning_rate": 6.0376000000000004e-06, "loss": 0.0255, "step": 19813 }, { "epoch": 1.0563522951431465, "grad_norm": 1.4834309816360474, "learning_rate": 6.037400000000001e-06, "loss": 0.0165, "step": 19814 }, { "epoch": 1.0564056085727995, "grad_norm": 1.5692960023880005, "learning_rate": 6.037200000000001e-06, "loss": 0.0309, "step": 19815 }, { "epoch": 1.0564589220024525, "grad_norm": 1.9231352806091309, "learning_rate": 6.037000000000001e-06, "loss": 0.0243, "step": 19816 }, { "epoch": 1.0565122354321053, "grad_norm": 1.0236681699752808, "learning_rate": 6.0368e-06, "loss": 0.0083, "step": 19817 }, { "epoch": 1.0565655488617582, "grad_norm": 2.71622896194458, "learning_rate": 6.0366e-06, "loss": 0.0243, "step": 19818 }, { "epoch": 1.0566188622914112, "grad_norm": 1.7419768571853638, "learning_rate": 6.0364e-06, "loss": 0.0227, "step": 19819 }, { "epoch": 1.0566721757210642, "grad_norm": 1.161037564277649, "learning_rate": 6.036200000000001e-06, "loss": 0.011, "step": 19820 }, { "epoch": 1.056725489150717, "grad_norm": 2.0958800315856934, "learning_rate": 6.036000000000001e-06, "loss": 0.0179, "step": 19821 }, { "epoch": 1.05677880258037, "grad_norm": 1.2076704502105713, "learning_rate": 6.035800000000001e-06, "loss": 0.0106, "step": 19822 }, { "epoch": 1.056832116010023, "grad_norm": 1.653825283050537, "learning_rate": 6.0356e-06, "loss": 0.0192, "step": 19823 }, { "epoch": 1.056885429439676, "grad_norm": 1.0855809450149536, "learning_rate": 6.0354e-06, "loss": 0.0083, "step": 19824 }, { "epoch": 1.0569387428693289, "grad_norm": 1.6654455661773682, "learning_rate": 6.0352e-06, "loss": 0.0224, "step": 19825 }, { "epoch": 1.0569920562989816, "grad_norm": 1.0960313081741333, "learning_rate": 6.035000000000001e-06, "loss": 0.0092, "step": 19826 }, { "epoch": 1.0570453697286346, "grad_norm": 1.5020508766174316, "learning_rate": 6.034800000000001e-06, "loss": 0.0087, "step": 19827 }, { "epoch": 1.0570986831582876, "grad_norm": 1.0350452661514282, "learning_rate": 6.034600000000001e-06, "loss": 0.0075, "step": 19828 }, { "epoch": 1.0571519965879406, "grad_norm": 2.5472850799560547, "learning_rate": 6.0344e-06, "loss": 0.0253, "step": 19829 }, { "epoch": 1.0572053100175933, "grad_norm": 1.8243156671524048, "learning_rate": 6.0342e-06, "loss": 0.0458, "step": 19830 }, { "epoch": 1.0572586234472463, "grad_norm": 2.15120267868042, "learning_rate": 6.034000000000001e-06, "loss": 0.0159, "step": 19831 }, { "epoch": 1.0573119368768993, "grad_norm": 2.40975284576416, "learning_rate": 6.0338000000000006e-06, "loss": 0.0308, "step": 19832 }, { "epoch": 1.0573652503065523, "grad_norm": 1.8661508560180664, "learning_rate": 6.033600000000001e-06, "loss": 0.0164, "step": 19833 }, { "epoch": 1.0574185637362052, "grad_norm": 3.4859507083892822, "learning_rate": 6.0334000000000005e-06, "loss": 0.0438, "step": 19834 }, { "epoch": 1.057471877165858, "grad_norm": 1.2261775732040405, "learning_rate": 6.0332e-06, "loss": 0.0097, "step": 19835 }, { "epoch": 1.057525190595511, "grad_norm": 1.3796555995941162, "learning_rate": 6.033e-06, "loss": 0.0128, "step": 19836 }, { "epoch": 1.057578504025164, "grad_norm": 1.4632558822631836, "learning_rate": 6.032800000000001e-06, "loss": 0.0231, "step": 19837 }, { "epoch": 1.057631817454817, "grad_norm": 1.4956859350204468, "learning_rate": 6.0326000000000005e-06, "loss": 0.0165, "step": 19838 }, { "epoch": 1.0576851308844697, "grad_norm": 1.1435827016830444, "learning_rate": 6.032400000000001e-06, "loss": 0.0115, "step": 19839 }, { "epoch": 1.0577384443141227, "grad_norm": 1.1350226402282715, "learning_rate": 6.0322000000000005e-06, "loss": 0.0108, "step": 19840 }, { "epoch": 1.0577917577437757, "grad_norm": 1.4694719314575195, "learning_rate": 6.032e-06, "loss": 0.0157, "step": 19841 }, { "epoch": 1.0578450711734286, "grad_norm": 5.064457416534424, "learning_rate": 6.031800000000001e-06, "loss": 0.0538, "step": 19842 }, { "epoch": 1.0578983846030816, "grad_norm": 1.2222825288772583, "learning_rate": 6.031600000000001e-06, "loss": 0.0105, "step": 19843 }, { "epoch": 1.0579516980327344, "grad_norm": 1.3717260360717773, "learning_rate": 6.0314e-06, "loss": 0.0124, "step": 19844 }, { "epoch": 1.0580050114623873, "grad_norm": 4.285172939300537, "learning_rate": 6.031200000000001e-06, "loss": 0.021, "step": 19845 }, { "epoch": 1.0580583248920403, "grad_norm": 2.0852484703063965, "learning_rate": 6.031e-06, "loss": 0.0263, "step": 19846 }, { "epoch": 1.0581116383216933, "grad_norm": 1.6480052471160889, "learning_rate": 6.0308e-06, "loss": 0.0217, "step": 19847 }, { "epoch": 1.058164951751346, "grad_norm": 2.9143924713134766, "learning_rate": 6.030600000000001e-06, "loss": 0.0188, "step": 19848 }, { "epoch": 1.058218265180999, "grad_norm": 2.1148476600646973, "learning_rate": 6.030400000000001e-06, "loss": 0.0356, "step": 19849 }, { "epoch": 1.058271578610652, "grad_norm": 2.7652626037597656, "learning_rate": 6.0302e-06, "loss": 0.0193, "step": 19850 }, { "epoch": 1.058324892040305, "grad_norm": 6.059258937835693, "learning_rate": 6.030000000000001e-06, "loss": 0.0455, "step": 19851 }, { "epoch": 1.058378205469958, "grad_norm": 3.6376960277557373, "learning_rate": 6.0298e-06, "loss": 0.0412, "step": 19852 }, { "epoch": 1.0584315188996107, "grad_norm": 1.5604510307312012, "learning_rate": 6.029600000000001e-06, "loss": 0.0107, "step": 19853 }, { "epoch": 1.0584848323292637, "grad_norm": 3.7623469829559326, "learning_rate": 6.029400000000001e-06, "loss": 0.0581, "step": 19854 }, { "epoch": 1.0585381457589167, "grad_norm": 1.037534236907959, "learning_rate": 6.029200000000001e-06, "loss": 0.0165, "step": 19855 }, { "epoch": 1.0585914591885697, "grad_norm": 3.5124359130859375, "learning_rate": 6.029e-06, "loss": 0.0231, "step": 19856 }, { "epoch": 1.0586447726182224, "grad_norm": 1.8244836330413818, "learning_rate": 6.028800000000001e-06, "loss": 0.0123, "step": 19857 }, { "epoch": 1.0586980860478754, "grad_norm": 3.2627451419830322, "learning_rate": 6.0286e-06, "loss": 0.0186, "step": 19858 }, { "epoch": 1.0587513994775284, "grad_norm": 3.226452589035034, "learning_rate": 6.028400000000001e-06, "loss": 0.0515, "step": 19859 }, { "epoch": 1.0588047129071814, "grad_norm": 3.2109110355377197, "learning_rate": 6.028200000000001e-06, "loss": 0.0238, "step": 19860 }, { "epoch": 1.0588580263368343, "grad_norm": 1.9867573976516724, "learning_rate": 6.0280000000000006e-06, "loss": 0.0284, "step": 19861 }, { "epoch": 1.058911339766487, "grad_norm": 1.639411449432373, "learning_rate": 6.0278e-06, "loss": 0.0362, "step": 19862 }, { "epoch": 1.05896465319614, "grad_norm": 2.3546195030212402, "learning_rate": 6.0276000000000005e-06, "loss": 0.0556, "step": 19863 }, { "epoch": 1.059017966625793, "grad_norm": 1.8061226606369019, "learning_rate": 6.027400000000001e-06, "loss": 0.0352, "step": 19864 }, { "epoch": 1.059071280055446, "grad_norm": 0.9322019815444946, "learning_rate": 6.0272000000000005e-06, "loss": 0.0121, "step": 19865 }, { "epoch": 1.0591245934850988, "grad_norm": 2.857387065887451, "learning_rate": 6.027000000000001e-06, "loss": 0.0621, "step": 19866 }, { "epoch": 1.0591779069147518, "grad_norm": 2.2041923999786377, "learning_rate": 6.0268000000000005e-06, "loss": 0.0488, "step": 19867 }, { "epoch": 1.0592312203444048, "grad_norm": 2.3502602577209473, "learning_rate": 6.0266e-06, "loss": 0.037, "step": 19868 }, { "epoch": 1.0592845337740577, "grad_norm": 1.379305124282837, "learning_rate": 6.0264000000000004e-06, "loss": 0.0117, "step": 19869 }, { "epoch": 1.0593378472037107, "grad_norm": 1.9016324281692505, "learning_rate": 6.026200000000001e-06, "loss": 0.0295, "step": 19870 }, { "epoch": 1.0593911606333635, "grad_norm": 6.952711582183838, "learning_rate": 6.026e-06, "loss": 0.0358, "step": 19871 }, { "epoch": 1.0594444740630165, "grad_norm": 2.713608503341675, "learning_rate": 6.025800000000001e-06, "loss": 0.0339, "step": 19872 }, { "epoch": 1.0594977874926694, "grad_norm": 2.115877151489258, "learning_rate": 6.0256e-06, "loss": 0.042, "step": 19873 }, { "epoch": 1.0595511009223224, "grad_norm": 1.8970041275024414, "learning_rate": 6.0254e-06, "loss": 0.039, "step": 19874 }, { "epoch": 1.0596044143519752, "grad_norm": 1.1521484851837158, "learning_rate": 6.0252e-06, "loss": 0.0148, "step": 19875 }, { "epoch": 1.0596577277816281, "grad_norm": 1.3141072988510132, "learning_rate": 6.025000000000001e-06, "loss": 0.0133, "step": 19876 }, { "epoch": 1.0597110412112811, "grad_norm": 1.3918712139129639, "learning_rate": 6.0248e-06, "loss": 0.0242, "step": 19877 }, { "epoch": 1.059764354640934, "grad_norm": 0.5140663981437683, "learning_rate": 6.024600000000001e-06, "loss": 0.004, "step": 19878 }, { "epoch": 1.059817668070587, "grad_norm": 1.3021512031555176, "learning_rate": 6.0244e-06, "loss": 0.0222, "step": 19879 }, { "epoch": 1.0598709815002398, "grad_norm": 2.1141622066497803, "learning_rate": 6.0242e-06, "loss": 0.0154, "step": 19880 }, { "epoch": 1.0599242949298928, "grad_norm": 2.354139804840088, "learning_rate": 6.024000000000001e-06, "loss": 0.0173, "step": 19881 }, { "epoch": 1.0599776083595458, "grad_norm": 2.15281343460083, "learning_rate": 6.023800000000001e-06, "loss": 0.0277, "step": 19882 }, { "epoch": 1.0600309217891988, "grad_norm": 2.455895185470581, "learning_rate": 6.0236e-06, "loss": 0.0067, "step": 19883 }, { "epoch": 1.0600842352188515, "grad_norm": 1.728585124015808, "learning_rate": 6.023400000000001e-06, "loss": 0.0424, "step": 19884 }, { "epoch": 1.0601375486485045, "grad_norm": 2.159937620162964, "learning_rate": 6.0232e-06, "loss": 0.0296, "step": 19885 }, { "epoch": 1.0601908620781575, "grad_norm": 0.2739293575286865, "learning_rate": 6.023e-06, "loss": 0.0027, "step": 19886 }, { "epoch": 1.0602441755078105, "grad_norm": 1.7918848991394043, "learning_rate": 6.022800000000001e-06, "loss": 0.0175, "step": 19887 }, { "epoch": 1.0602974889374635, "grad_norm": 1.1050400733947754, "learning_rate": 6.022600000000001e-06, "loss": 0.0405, "step": 19888 }, { "epoch": 1.0603508023671162, "grad_norm": 0.8001611828804016, "learning_rate": 6.0224e-06, "loss": 0.0076, "step": 19889 }, { "epoch": 1.0604041157967692, "grad_norm": 1.4933276176452637, "learning_rate": 6.0222000000000005e-06, "loss": 0.0197, "step": 19890 }, { "epoch": 1.0604574292264222, "grad_norm": 1.1123392581939697, "learning_rate": 6.022e-06, "loss": 0.0127, "step": 19891 }, { "epoch": 1.0605107426560751, "grad_norm": 1.04823899269104, "learning_rate": 6.021800000000001e-06, "loss": 0.0283, "step": 19892 }, { "epoch": 1.060564056085728, "grad_norm": 1.1919012069702148, "learning_rate": 6.021600000000001e-06, "loss": 0.0138, "step": 19893 }, { "epoch": 1.0606173695153809, "grad_norm": 0.7866219878196716, "learning_rate": 6.0214000000000005e-06, "loss": 0.0085, "step": 19894 }, { "epoch": 1.0606706829450339, "grad_norm": 1.3821885585784912, "learning_rate": 6.0212e-06, "loss": 0.0193, "step": 19895 }, { "epoch": 1.0607239963746868, "grad_norm": 2.9074292182922363, "learning_rate": 6.0210000000000005e-06, "loss": 0.0217, "step": 19896 }, { "epoch": 1.0607773098043398, "grad_norm": 1.7499605417251587, "learning_rate": 6.0208e-06, "loss": 0.0251, "step": 19897 }, { "epoch": 1.0608306232339926, "grad_norm": 1.508727788925171, "learning_rate": 6.020600000000001e-06, "loss": 0.0234, "step": 19898 }, { "epoch": 1.0608839366636456, "grad_norm": 0.751622200012207, "learning_rate": 6.020400000000001e-06, "loss": 0.0052, "step": 19899 }, { "epoch": 1.0609372500932985, "grad_norm": 1.7472485303878784, "learning_rate": 6.0202e-06, "loss": 0.0179, "step": 19900 }, { "epoch": 1.0609905635229515, "grad_norm": 1.0320677757263184, "learning_rate": 6.02e-06, "loss": 0.0139, "step": 19901 }, { "epoch": 1.0610438769526043, "grad_norm": 1.6465044021606445, "learning_rate": 6.0198e-06, "loss": 0.0391, "step": 19902 }, { "epoch": 1.0610971903822572, "grad_norm": 1.3568512201309204, "learning_rate": 6.019600000000001e-06, "loss": 0.024, "step": 19903 }, { "epoch": 1.0611505038119102, "grad_norm": 2.5216379165649414, "learning_rate": 6.019400000000001e-06, "loss": 0.0256, "step": 19904 }, { "epoch": 1.0612038172415632, "grad_norm": 0.6695784330368042, "learning_rate": 6.019200000000001e-06, "loss": 0.0057, "step": 19905 }, { "epoch": 1.0612571306712162, "grad_norm": 2.426858425140381, "learning_rate": 6.019e-06, "loss": 0.0336, "step": 19906 }, { "epoch": 1.061310444100869, "grad_norm": 2.679299831390381, "learning_rate": 6.0188e-06, "loss": 0.0281, "step": 19907 }, { "epoch": 1.061363757530522, "grad_norm": 2.4762089252471924, "learning_rate": 6.0186e-06, "loss": 0.0613, "step": 19908 }, { "epoch": 1.061417070960175, "grad_norm": 1.03423011302948, "learning_rate": 6.018400000000001e-06, "loss": 0.0143, "step": 19909 }, { "epoch": 1.0614703843898279, "grad_norm": 2.6112027168273926, "learning_rate": 6.018200000000001e-06, "loss": 0.0231, "step": 19910 }, { "epoch": 1.0615236978194806, "grad_norm": 1.5834572315216064, "learning_rate": 6.018000000000001e-06, "loss": 0.0066, "step": 19911 }, { "epoch": 1.0615770112491336, "grad_norm": 1.7434598207473755, "learning_rate": 6.0178e-06, "loss": 0.0133, "step": 19912 }, { "epoch": 1.0616303246787866, "grad_norm": 1.0797582864761353, "learning_rate": 6.0176e-06, "loss": 0.0091, "step": 19913 }, { "epoch": 1.0616836381084396, "grad_norm": 2.2750437259674072, "learning_rate": 6.017400000000001e-06, "loss": 0.0384, "step": 19914 }, { "epoch": 1.0617369515380926, "grad_norm": 1.202247142791748, "learning_rate": 6.017200000000001e-06, "loss": 0.0119, "step": 19915 }, { "epoch": 1.0617902649677453, "grad_norm": 1.8528025150299072, "learning_rate": 6.017000000000001e-06, "loss": 0.0256, "step": 19916 }, { "epoch": 1.0618435783973983, "grad_norm": 1.1058886051177979, "learning_rate": 6.0168000000000006e-06, "loss": 0.0101, "step": 19917 }, { "epoch": 1.0618968918270513, "grad_norm": 1.482504963874817, "learning_rate": 6.0166e-06, "loss": 0.0244, "step": 19918 }, { "epoch": 1.0619502052567042, "grad_norm": 1.9057620763778687, "learning_rate": 6.0164e-06, "loss": 0.0563, "step": 19919 }, { "epoch": 1.062003518686357, "grad_norm": 2.486882448196411, "learning_rate": 6.016200000000001e-06, "loss": 0.0144, "step": 19920 }, { "epoch": 1.06205683211601, "grad_norm": 3.1547722816467285, "learning_rate": 6.0160000000000005e-06, "loss": 0.0454, "step": 19921 }, { "epoch": 1.062110145545663, "grad_norm": 1.8256094455718994, "learning_rate": 6.015800000000001e-06, "loss": 0.037, "step": 19922 }, { "epoch": 1.062163458975316, "grad_norm": 1.3720312118530273, "learning_rate": 6.0156000000000005e-06, "loss": 0.019, "step": 19923 }, { "epoch": 1.0622167724049687, "grad_norm": 1.8761365413665771, "learning_rate": 6.0154e-06, "loss": 0.0204, "step": 19924 }, { "epoch": 1.0622700858346217, "grad_norm": 1.4241758584976196, "learning_rate": 6.015200000000001e-06, "loss": 0.024, "step": 19925 }, { "epoch": 1.0623233992642747, "grad_norm": 1.0204273462295532, "learning_rate": 6.015000000000001e-06, "loss": 0.0189, "step": 19926 }, { "epoch": 1.0623767126939276, "grad_norm": 1.7272276878356934, "learning_rate": 6.0148e-06, "loss": 0.0418, "step": 19927 }, { "epoch": 1.0624300261235806, "grad_norm": 0.4500451385974884, "learning_rate": 6.014600000000001e-06, "loss": 0.0037, "step": 19928 }, { "epoch": 1.0624833395532334, "grad_norm": 1.6492830514907837, "learning_rate": 6.0144e-06, "loss": 0.0337, "step": 19929 }, { "epoch": 1.0625366529828864, "grad_norm": 1.4954699277877808, "learning_rate": 6.0142e-06, "loss": 0.0475, "step": 19930 }, { "epoch": 1.0625899664125393, "grad_norm": 1.5925638675689697, "learning_rate": 6.014000000000001e-06, "loss": 0.0123, "step": 19931 }, { "epoch": 1.0626432798421923, "grad_norm": 1.6724843978881836, "learning_rate": 6.013800000000001e-06, "loss": 0.0203, "step": 19932 }, { "epoch": 1.0626965932718453, "grad_norm": 2.091111421585083, "learning_rate": 6.0136e-06, "loss": 0.0241, "step": 19933 }, { "epoch": 1.062749906701498, "grad_norm": 1.7904276847839355, "learning_rate": 6.013400000000001e-06, "loss": 0.0469, "step": 19934 }, { "epoch": 1.062803220131151, "grad_norm": 2.579772710800171, "learning_rate": 6.0132e-06, "loss": 0.0737, "step": 19935 }, { "epoch": 1.062856533560804, "grad_norm": 1.9152551889419556, "learning_rate": 6.013e-06, "loss": 0.0291, "step": 19936 }, { "epoch": 1.062909846990457, "grad_norm": 3.3388004302978516, "learning_rate": 6.012800000000001e-06, "loss": 0.0373, "step": 19937 }, { "epoch": 1.0629631604201097, "grad_norm": 1.7635670900344849, "learning_rate": 6.012600000000001e-06, "loss": 0.0147, "step": 19938 }, { "epoch": 1.0630164738497627, "grad_norm": 1.513817548751831, "learning_rate": 6.0124e-06, "loss": 0.0237, "step": 19939 }, { "epoch": 1.0630697872794157, "grad_norm": 1.786589503288269, "learning_rate": 6.012200000000001e-06, "loss": 0.0368, "step": 19940 }, { "epoch": 1.0631231007090687, "grad_norm": 2.010026693344116, "learning_rate": 6.012e-06, "loss": 0.0143, "step": 19941 }, { "epoch": 1.0631764141387214, "grad_norm": 3.472980499267578, "learning_rate": 6.011800000000001e-06, "loss": 0.0317, "step": 19942 }, { "epoch": 1.0632297275683744, "grad_norm": 1.2089786529541016, "learning_rate": 6.011600000000001e-06, "loss": 0.0204, "step": 19943 }, { "epoch": 1.0632830409980274, "grad_norm": 1.8454381227493286, "learning_rate": 6.011400000000001e-06, "loss": 0.052, "step": 19944 }, { "epoch": 1.0633363544276804, "grad_norm": 2.289837598800659, "learning_rate": 6.0112e-06, "loss": 0.0155, "step": 19945 }, { "epoch": 1.0633896678573334, "grad_norm": 0.8610074520111084, "learning_rate": 6.0110000000000006e-06, "loss": 0.0062, "step": 19946 }, { "epoch": 1.063442981286986, "grad_norm": 2.9828593730926514, "learning_rate": 6.0108e-06, "loss": 0.0227, "step": 19947 }, { "epoch": 1.063496294716639, "grad_norm": 2.033513307571411, "learning_rate": 6.0106000000000005e-06, "loss": 0.0399, "step": 19948 }, { "epoch": 1.063549608146292, "grad_norm": 4.03729772567749, "learning_rate": 6.010400000000001e-06, "loss": 0.0356, "step": 19949 }, { "epoch": 1.063602921575945, "grad_norm": 1.9461846351623535, "learning_rate": 6.0102000000000005e-06, "loss": 0.0302, "step": 19950 }, { "epoch": 1.063656235005598, "grad_norm": 1.2375857830047607, "learning_rate": 6.01e-06, "loss": 0.0221, "step": 19951 }, { "epoch": 1.0637095484352508, "grad_norm": 2.2320449352264404, "learning_rate": 6.0098000000000005e-06, "loss": 0.0426, "step": 19952 }, { "epoch": 1.0637628618649038, "grad_norm": 1.1426454782485962, "learning_rate": 6.009600000000001e-06, "loss": 0.029, "step": 19953 }, { "epoch": 1.0638161752945567, "grad_norm": 1.6240757703781128, "learning_rate": 6.0094000000000004e-06, "loss": 0.0085, "step": 19954 }, { "epoch": 1.0638694887242097, "grad_norm": 2.4632112979888916, "learning_rate": 6.009200000000001e-06, "loss": 0.0421, "step": 19955 }, { "epoch": 1.0639228021538625, "grad_norm": 3.0196526050567627, "learning_rate": 6.009e-06, "loss": 0.0311, "step": 19956 }, { "epoch": 1.0639761155835155, "grad_norm": 2.1479594707489014, "learning_rate": 6.0088e-06, "loss": 0.0178, "step": 19957 }, { "epoch": 1.0640294290131684, "grad_norm": 1.1193698644638062, "learning_rate": 6.0086e-06, "loss": 0.0123, "step": 19958 }, { "epoch": 1.0640827424428214, "grad_norm": 1.5742518901824951, "learning_rate": 6.008400000000001e-06, "loss": 0.0381, "step": 19959 }, { "epoch": 1.0641360558724742, "grad_norm": 2.0581984519958496, "learning_rate": 6.0082e-06, "loss": 0.0323, "step": 19960 }, { "epoch": 1.0641893693021272, "grad_norm": 1.6867668628692627, "learning_rate": 6.008000000000001e-06, "loss": 0.0105, "step": 19961 }, { "epoch": 1.0642426827317801, "grad_norm": 2.1661324501037598, "learning_rate": 6.0078e-06, "loss": 0.0143, "step": 19962 }, { "epoch": 1.0642959961614331, "grad_norm": 1.8151367902755737, "learning_rate": 6.0076e-06, "loss": 0.0241, "step": 19963 }, { "epoch": 1.064349309591086, "grad_norm": 1.4725627899169922, "learning_rate": 6.007400000000001e-06, "loss": 0.0253, "step": 19964 }, { "epoch": 1.0644026230207388, "grad_norm": 1.2439545392990112, "learning_rate": 6.007200000000001e-06, "loss": 0.0165, "step": 19965 }, { "epoch": 1.0644559364503918, "grad_norm": 2.6681928634643555, "learning_rate": 6.007e-06, "loss": 0.0173, "step": 19966 }, { "epoch": 1.0645092498800448, "grad_norm": 2.5356431007385254, "learning_rate": 6.006800000000001e-06, "loss": 0.0211, "step": 19967 }, { "epoch": 1.0645625633096978, "grad_norm": 1.3135448694229126, "learning_rate": 6.0066e-06, "loss": 0.0141, "step": 19968 }, { "epoch": 1.0646158767393505, "grad_norm": 0.5431568622589111, "learning_rate": 6.0064e-06, "loss": 0.0055, "step": 19969 }, { "epoch": 1.0646691901690035, "grad_norm": 2.5761706829071045, "learning_rate": 6.006200000000001e-06, "loss": 0.0219, "step": 19970 }, { "epoch": 1.0647225035986565, "grad_norm": 1.5189383029937744, "learning_rate": 6.006000000000001e-06, "loss": 0.0145, "step": 19971 }, { "epoch": 1.0647758170283095, "grad_norm": 2.0990664958953857, "learning_rate": 6.0058e-06, "loss": 0.0618, "step": 19972 }, { "epoch": 1.0648291304579625, "grad_norm": 0.5389472246170044, "learning_rate": 6.0056000000000006e-06, "loss": 0.0313, "step": 19973 }, { "epoch": 1.0648824438876152, "grad_norm": 2.130241870880127, "learning_rate": 6.0054e-06, "loss": 0.0359, "step": 19974 }, { "epoch": 1.0649357573172682, "grad_norm": 2.5167453289031982, "learning_rate": 6.005200000000001e-06, "loss": 0.0343, "step": 19975 }, { "epoch": 1.0649890707469212, "grad_norm": 0.8216534852981567, "learning_rate": 6.005000000000001e-06, "loss": 0.0053, "step": 19976 }, { "epoch": 1.0650423841765742, "grad_norm": 1.3707748651504517, "learning_rate": 6.0048000000000005e-06, "loss": 0.0412, "step": 19977 }, { "epoch": 1.065095697606227, "grad_norm": 1.8268425464630127, "learning_rate": 6.0046e-06, "loss": 0.0323, "step": 19978 }, { "epoch": 1.06514901103588, "grad_norm": 1.046463131904602, "learning_rate": 6.0044000000000005e-06, "loss": 0.0129, "step": 19979 }, { "epoch": 1.0652023244655329, "grad_norm": 2.095679521560669, "learning_rate": 6.0042e-06, "loss": 0.0395, "step": 19980 }, { "epoch": 1.0652556378951858, "grad_norm": 2.819601058959961, "learning_rate": 6.004000000000001e-06, "loss": 0.0348, "step": 19981 }, { "epoch": 1.0653089513248388, "grad_norm": 1.8844023942947388, "learning_rate": 6.003800000000001e-06, "loss": 0.0363, "step": 19982 }, { "epoch": 1.0653622647544916, "grad_norm": 1.6022520065307617, "learning_rate": 6.0036e-06, "loss": 0.0186, "step": 19983 }, { "epoch": 1.0654155781841446, "grad_norm": 1.3433054685592651, "learning_rate": 6.0034e-06, "loss": 0.0218, "step": 19984 }, { "epoch": 1.0654688916137975, "grad_norm": 2.4955697059631348, "learning_rate": 6.0032e-06, "loss": 0.0369, "step": 19985 }, { "epoch": 1.0655222050434505, "grad_norm": 1.4001591205596924, "learning_rate": 6.003e-06, "loss": 0.0105, "step": 19986 }, { "epoch": 1.0655755184731033, "grad_norm": 3.9410789012908936, "learning_rate": 6.002800000000001e-06, "loss": 0.04, "step": 19987 }, { "epoch": 1.0656288319027563, "grad_norm": 2.5068891048431396, "learning_rate": 6.002600000000001e-06, "loss": 0.0193, "step": 19988 }, { "epoch": 1.0656821453324092, "grad_norm": 1.6087450981140137, "learning_rate": 6.0024e-06, "loss": 0.0395, "step": 19989 }, { "epoch": 1.0657354587620622, "grad_norm": 1.9662113189697266, "learning_rate": 6.0022e-06, "loss": 0.0396, "step": 19990 }, { "epoch": 1.0657887721917152, "grad_norm": 1.5542335510253906, "learning_rate": 6.002e-06, "loss": 0.0238, "step": 19991 }, { "epoch": 1.065842085621368, "grad_norm": 1.4094295501708984, "learning_rate": 6.001800000000001e-06, "loss": 0.0294, "step": 19992 }, { "epoch": 1.065895399051021, "grad_norm": 2.9945266246795654, "learning_rate": 6.001600000000001e-06, "loss": 0.0335, "step": 19993 }, { "epoch": 1.065948712480674, "grad_norm": 1.0898023843765259, "learning_rate": 6.001400000000001e-06, "loss": 0.0131, "step": 19994 }, { "epoch": 1.066002025910327, "grad_norm": 2.506075620651245, "learning_rate": 6.0012e-06, "loss": 0.0162, "step": 19995 }, { "epoch": 1.0660553393399796, "grad_norm": 3.96058988571167, "learning_rate": 6.001e-06, "loss": 0.0339, "step": 19996 }, { "epoch": 1.0661086527696326, "grad_norm": 1.1762546300888062, "learning_rate": 6.0008e-06, "loss": 0.0218, "step": 19997 }, { "epoch": 1.0661619661992856, "grad_norm": 1.3834420442581177, "learning_rate": 6.000600000000001e-06, "loss": 0.0246, "step": 19998 }, { "epoch": 1.0662152796289386, "grad_norm": 1.1796804666519165, "learning_rate": 6.000400000000001e-06, "loss": 0.012, "step": 19999 }, { "epoch": 1.0662685930585916, "grad_norm": 1.4727811813354492, "learning_rate": 6.000200000000001e-06, "loss": 0.0112, "step": 20000 }, { "epoch": 1.0663219064882443, "grad_norm": 0.9052754640579224, "learning_rate": 6e-06, "loss": 0.0069, "step": 20001 }, { "epoch": 1.0663752199178973, "grad_norm": 1.683988332748413, "learning_rate": 5.9998e-06, "loss": 0.0179, "step": 20002 }, { "epoch": 1.0664285333475503, "grad_norm": 2.505653142929077, "learning_rate": 5.999600000000001e-06, "loss": 0.0381, "step": 20003 }, { "epoch": 1.0664818467772033, "grad_norm": 2.3639743328094482, "learning_rate": 5.9994000000000005e-06, "loss": 0.0502, "step": 20004 }, { "epoch": 1.066535160206856, "grad_norm": 1.4390939474105835, "learning_rate": 5.999200000000001e-06, "loss": 0.0072, "step": 20005 }, { "epoch": 1.066588473636509, "grad_norm": 1.2648096084594727, "learning_rate": 5.9990000000000005e-06, "loss": 0.0176, "step": 20006 }, { "epoch": 1.066641787066162, "grad_norm": 1.341996431350708, "learning_rate": 5.9988e-06, "loss": 0.0266, "step": 20007 }, { "epoch": 1.066695100495815, "grad_norm": 2.9395883083343506, "learning_rate": 5.9986e-06, "loss": 0.0561, "step": 20008 }, { "epoch": 1.0667484139254677, "grad_norm": 1.1860016584396362, "learning_rate": 5.998400000000001e-06, "loss": 0.026, "step": 20009 }, { "epoch": 1.0668017273551207, "grad_norm": 2.997321367263794, "learning_rate": 5.9982000000000004e-06, "loss": 0.0508, "step": 20010 }, { "epoch": 1.0668550407847737, "grad_norm": 1.9301798343658447, "learning_rate": 5.998000000000001e-06, "loss": 0.0438, "step": 20011 }, { "epoch": 1.0669083542144266, "grad_norm": 1.1523058414459229, "learning_rate": 5.9978e-06, "loss": 0.0083, "step": 20012 }, { "epoch": 1.0669616676440796, "grad_norm": 1.6663764715194702, "learning_rate": 5.9976e-06, "loss": 0.0309, "step": 20013 }, { "epoch": 1.0670149810737324, "grad_norm": 2.703080177307129, "learning_rate": 5.997400000000001e-06, "loss": 0.028, "step": 20014 }, { "epoch": 1.0670682945033854, "grad_norm": 1.9258882999420166, "learning_rate": 5.997200000000001e-06, "loss": 0.0214, "step": 20015 }, { "epoch": 1.0671216079330383, "grad_norm": 5.001427173614502, "learning_rate": 5.997e-06, "loss": 0.0613, "step": 20016 }, { "epoch": 1.0671749213626913, "grad_norm": 5.6054463386535645, "learning_rate": 5.996800000000001e-06, "loss": 0.0378, "step": 20017 }, { "epoch": 1.0672282347923443, "grad_norm": 3.2718253135681152, "learning_rate": 5.9966e-06, "loss": 0.0669, "step": 20018 }, { "epoch": 1.067281548221997, "grad_norm": 2.2386345863342285, "learning_rate": 5.9964e-06, "loss": 0.0163, "step": 20019 }, { "epoch": 1.06733486165165, "grad_norm": 2.6075475215911865, "learning_rate": 5.996200000000001e-06, "loss": 0.0345, "step": 20020 }, { "epoch": 1.067388175081303, "grad_norm": 0.911710798740387, "learning_rate": 5.996000000000001e-06, "loss": 0.0191, "step": 20021 }, { "epoch": 1.067441488510956, "grad_norm": 3.200404405593872, "learning_rate": 5.9958e-06, "loss": 0.0551, "step": 20022 }, { "epoch": 1.0674948019406088, "grad_norm": 1.6700795888900757, "learning_rate": 5.995600000000001e-06, "loss": 0.0375, "step": 20023 }, { "epoch": 1.0675481153702617, "grad_norm": 1.8154736757278442, "learning_rate": 5.9954e-06, "loss": 0.0286, "step": 20024 }, { "epoch": 1.0676014287999147, "grad_norm": 1.0786051750183105, "learning_rate": 5.995200000000001e-06, "loss": 0.0163, "step": 20025 }, { "epoch": 1.0676547422295677, "grad_norm": 1.8751293420791626, "learning_rate": 5.995000000000001e-06, "loss": 0.0233, "step": 20026 }, { "epoch": 1.0677080556592204, "grad_norm": 1.6485804319381714, "learning_rate": 5.994800000000001e-06, "loss": 0.037, "step": 20027 }, { "epoch": 1.0677613690888734, "grad_norm": 1.1304967403411865, "learning_rate": 5.9946e-06, "loss": 0.0076, "step": 20028 }, { "epoch": 1.0678146825185264, "grad_norm": 2.403545379638672, "learning_rate": 5.994400000000001e-06, "loss": 0.0367, "step": 20029 }, { "epoch": 1.0678679959481794, "grad_norm": 2.793896436691284, "learning_rate": 5.9942e-06, "loss": 0.0142, "step": 20030 }, { "epoch": 1.0679213093778324, "grad_norm": 2.1631929874420166, "learning_rate": 5.9940000000000005e-06, "loss": 0.0218, "step": 20031 }, { "epoch": 1.0679746228074851, "grad_norm": 0.41080042719841003, "learning_rate": 5.993800000000001e-06, "loss": 0.0042, "step": 20032 }, { "epoch": 1.068027936237138, "grad_norm": 2.0989086627960205, "learning_rate": 5.9936000000000005e-06, "loss": 0.0522, "step": 20033 }, { "epoch": 1.068081249666791, "grad_norm": 0.9933579564094543, "learning_rate": 5.9934e-06, "loss": 0.0107, "step": 20034 }, { "epoch": 1.068134563096444, "grad_norm": 3.0436434745788574, "learning_rate": 5.9932000000000005e-06, "loss": 0.0387, "step": 20035 }, { "epoch": 1.068187876526097, "grad_norm": 0.5542550086975098, "learning_rate": 5.993000000000001e-06, "loss": 0.0069, "step": 20036 }, { "epoch": 1.0682411899557498, "grad_norm": 2.1993911266326904, "learning_rate": 5.9928000000000005e-06, "loss": 0.0183, "step": 20037 }, { "epoch": 1.0682945033854028, "grad_norm": 0.7967157959938049, "learning_rate": 5.992600000000001e-06, "loss": 0.0107, "step": 20038 }, { "epoch": 1.0683478168150558, "grad_norm": 1.9487255811691284, "learning_rate": 5.9924000000000004e-06, "loss": 0.0314, "step": 20039 }, { "epoch": 1.0684011302447087, "grad_norm": 1.901229977607727, "learning_rate": 5.9922e-06, "loss": 0.0431, "step": 20040 }, { "epoch": 1.0684544436743615, "grad_norm": 2.119508743286133, "learning_rate": 5.992e-06, "loss": 0.0306, "step": 20041 }, { "epoch": 1.0685077571040145, "grad_norm": 2.80847430229187, "learning_rate": 5.991800000000001e-06, "loss": 0.0298, "step": 20042 }, { "epoch": 1.0685610705336674, "grad_norm": 1.0316345691680908, "learning_rate": 5.9916e-06, "loss": 0.0096, "step": 20043 }, { "epoch": 1.0686143839633204, "grad_norm": 1.49027681350708, "learning_rate": 5.991400000000001e-06, "loss": 0.0197, "step": 20044 }, { "epoch": 1.0686676973929732, "grad_norm": 1.6413744688034058, "learning_rate": 5.9912e-06, "loss": 0.0284, "step": 20045 }, { "epoch": 1.0687210108226262, "grad_norm": 4.499964714050293, "learning_rate": 5.991e-06, "loss": 0.031, "step": 20046 }, { "epoch": 1.0687743242522791, "grad_norm": 1.5202244520187378, "learning_rate": 5.9908e-06, "loss": 0.0465, "step": 20047 }, { "epoch": 1.0688276376819321, "grad_norm": 2.1739418506622314, "learning_rate": 5.990600000000001e-06, "loss": 0.0401, "step": 20048 }, { "epoch": 1.068880951111585, "grad_norm": 6.910149574279785, "learning_rate": 5.9904e-06, "loss": 0.059, "step": 20049 }, { "epoch": 1.0689342645412379, "grad_norm": 2.0935311317443848, "learning_rate": 5.990200000000001e-06, "loss": 0.0168, "step": 20050 }, { "epoch": 1.0689875779708908, "grad_norm": 3.8473918437957764, "learning_rate": 5.99e-06, "loss": 0.0538, "step": 20051 }, { "epoch": 1.0690408914005438, "grad_norm": 2.108525276184082, "learning_rate": 5.9898e-06, "loss": 0.0431, "step": 20052 }, { "epoch": 1.0690942048301968, "grad_norm": 1.79822838306427, "learning_rate": 5.989600000000001e-06, "loss": 0.031, "step": 20053 }, { "epoch": 1.0691475182598498, "grad_norm": 0.7556445598602295, "learning_rate": 5.989400000000001e-06, "loss": 0.0111, "step": 20054 }, { "epoch": 1.0692008316895025, "grad_norm": 1.4260228872299194, "learning_rate": 5.9892e-06, "loss": 0.0289, "step": 20055 }, { "epoch": 1.0692541451191555, "grad_norm": 1.3333340883255005, "learning_rate": 5.989000000000001e-06, "loss": 0.0077, "step": 20056 }, { "epoch": 1.0693074585488085, "grad_norm": 1.3399326801300049, "learning_rate": 5.9888e-06, "loss": 0.0164, "step": 20057 }, { "epoch": 1.0693607719784615, "grad_norm": 1.013027310371399, "learning_rate": 5.9886e-06, "loss": 0.0243, "step": 20058 }, { "epoch": 1.0694140854081142, "grad_norm": 1.8742440938949585, "learning_rate": 5.988400000000001e-06, "loss": 0.0171, "step": 20059 }, { "epoch": 1.0694673988377672, "grad_norm": 1.1339960098266602, "learning_rate": 5.9882000000000005e-06, "loss": 0.0122, "step": 20060 }, { "epoch": 1.0695207122674202, "grad_norm": 2.7368175983428955, "learning_rate": 5.988e-06, "loss": 0.0329, "step": 20061 }, { "epoch": 1.0695740256970732, "grad_norm": 1.1582002639770508, "learning_rate": 5.9878000000000005e-06, "loss": 0.0153, "step": 20062 }, { "epoch": 1.069627339126726, "grad_norm": 4.698726177215576, "learning_rate": 5.9876e-06, "loss": 0.0306, "step": 20063 }, { "epoch": 1.069680652556379, "grad_norm": 3.790475368499756, "learning_rate": 5.987400000000001e-06, "loss": 0.041, "step": 20064 }, { "epoch": 1.0697339659860319, "grad_norm": 2.574265480041504, "learning_rate": 5.987200000000001e-06, "loss": 0.0172, "step": 20065 }, { "epoch": 1.0697872794156849, "grad_norm": 0.981805145740509, "learning_rate": 5.9870000000000004e-06, "loss": 0.0154, "step": 20066 }, { "epoch": 1.0698405928453378, "grad_norm": 1.5087193250656128, "learning_rate": 5.9868e-06, "loss": 0.0226, "step": 20067 }, { "epoch": 1.0698939062749906, "grad_norm": 2.5262508392333984, "learning_rate": 5.9866e-06, "loss": 0.0141, "step": 20068 }, { "epoch": 1.0699472197046436, "grad_norm": 1.6998697519302368, "learning_rate": 5.9864e-06, "loss": 0.0155, "step": 20069 }, { "epoch": 1.0700005331342965, "grad_norm": 1.2910107374191284, "learning_rate": 5.986200000000001e-06, "loss": 0.0181, "step": 20070 }, { "epoch": 1.0700538465639495, "grad_norm": 1.1060974597930908, "learning_rate": 5.986000000000001e-06, "loss": 0.0104, "step": 20071 }, { "epoch": 1.0701071599936023, "grad_norm": 1.656501054763794, "learning_rate": 5.9858e-06, "loss": 0.0415, "step": 20072 }, { "epoch": 1.0701604734232553, "grad_norm": 1.7482508420944214, "learning_rate": 5.9856e-06, "loss": 0.0145, "step": 20073 }, { "epoch": 1.0702137868529082, "grad_norm": 1.6662205457687378, "learning_rate": 5.9854e-06, "loss": 0.0238, "step": 20074 }, { "epoch": 1.0702671002825612, "grad_norm": 1.4951413869857788, "learning_rate": 5.985200000000001e-06, "loss": 0.0257, "step": 20075 }, { "epoch": 1.0703204137122142, "grad_norm": 1.7551542520523071, "learning_rate": 5.985000000000001e-06, "loss": 0.0361, "step": 20076 }, { "epoch": 1.070373727141867, "grad_norm": 2.0261588096618652, "learning_rate": 5.984800000000001e-06, "loss": 0.027, "step": 20077 }, { "epoch": 1.07042704057152, "grad_norm": 1.6556600332260132, "learning_rate": 5.9846e-06, "loss": 0.0206, "step": 20078 }, { "epoch": 1.070480354001173, "grad_norm": 2.288116216659546, "learning_rate": 5.9844e-06, "loss": 0.0327, "step": 20079 }, { "epoch": 1.070533667430826, "grad_norm": 1.8696820735931396, "learning_rate": 5.9842e-06, "loss": 0.0226, "step": 20080 }, { "epoch": 1.0705869808604787, "grad_norm": 2.2544853687286377, "learning_rate": 5.984000000000001e-06, "loss": 0.0238, "step": 20081 }, { "epoch": 1.0706402942901316, "grad_norm": 1.1103628873825073, "learning_rate": 5.983800000000001e-06, "loss": 0.0071, "step": 20082 }, { "epoch": 1.0706936077197846, "grad_norm": 1.6637996435165405, "learning_rate": 5.983600000000001e-06, "loss": 0.0298, "step": 20083 }, { "epoch": 1.0707469211494376, "grad_norm": 2.3560943603515625, "learning_rate": 5.9834e-06, "loss": 0.0403, "step": 20084 }, { "epoch": 1.0708002345790906, "grad_norm": 1.1157143115997314, "learning_rate": 5.9832e-06, "loss": 0.0222, "step": 20085 }, { "epoch": 1.0708535480087433, "grad_norm": 0.8054904937744141, "learning_rate": 5.983000000000001e-06, "loss": 0.0072, "step": 20086 }, { "epoch": 1.0709068614383963, "grad_norm": 1.6375280618667603, "learning_rate": 5.9828000000000006e-06, "loss": 0.0138, "step": 20087 }, { "epoch": 1.0709601748680493, "grad_norm": 3.7028462886810303, "learning_rate": 5.982600000000001e-06, "loss": 0.0283, "step": 20088 }, { "epoch": 1.0710134882977023, "grad_norm": 3.6303136348724365, "learning_rate": 5.9824000000000005e-06, "loss": 0.0344, "step": 20089 }, { "epoch": 1.071066801727355, "grad_norm": 2.525139808654785, "learning_rate": 5.9822e-06, "loss": 0.0463, "step": 20090 }, { "epoch": 1.071120115157008, "grad_norm": 2.141719341278076, "learning_rate": 5.982e-06, "loss": 0.0318, "step": 20091 }, { "epoch": 1.071173428586661, "grad_norm": 2.0710349082946777, "learning_rate": 5.981800000000001e-06, "loss": 0.0159, "step": 20092 }, { "epoch": 1.071226742016314, "grad_norm": 2.205503463745117, "learning_rate": 5.9816000000000005e-06, "loss": 0.0452, "step": 20093 }, { "epoch": 1.071280055445967, "grad_norm": 2.411233425140381, "learning_rate": 5.981400000000001e-06, "loss": 0.0219, "step": 20094 }, { "epoch": 1.0713333688756197, "grad_norm": 1.2732828855514526, "learning_rate": 5.9812000000000004e-06, "loss": 0.0216, "step": 20095 }, { "epoch": 1.0713866823052727, "grad_norm": 6.184657096862793, "learning_rate": 5.981e-06, "loss": 0.0689, "step": 20096 }, { "epoch": 1.0714399957349257, "grad_norm": 1.3709083795547485, "learning_rate": 5.9807999999999996e-06, "loss": 0.0131, "step": 20097 }, { "epoch": 1.0714933091645786, "grad_norm": 1.6104432344436646, "learning_rate": 5.980600000000001e-06, "loss": 0.0371, "step": 20098 }, { "epoch": 1.0715466225942314, "grad_norm": 1.8454904556274414, "learning_rate": 5.9804e-06, "loss": 0.0424, "step": 20099 }, { "epoch": 1.0715999360238844, "grad_norm": 2.6496238708496094, "learning_rate": 5.980200000000001e-06, "loss": 0.0386, "step": 20100 }, { "epoch": 1.0716532494535373, "grad_norm": 1.3266149759292603, "learning_rate": 5.98e-06, "loss": 0.0219, "step": 20101 }, { "epoch": 1.0717065628831903, "grad_norm": 1.3806045055389404, "learning_rate": 5.9798e-06, "loss": 0.0077, "step": 20102 }, { "epoch": 1.0717598763128433, "grad_norm": 1.943280816078186, "learning_rate": 5.979600000000001e-06, "loss": 0.0118, "step": 20103 }, { "epoch": 1.071813189742496, "grad_norm": 4.016180992126465, "learning_rate": 5.979400000000001e-06, "loss": 0.0337, "step": 20104 }, { "epoch": 1.071866503172149, "grad_norm": 2.7079145908355713, "learning_rate": 5.9792e-06, "loss": 0.0421, "step": 20105 }, { "epoch": 1.071919816601802, "grad_norm": 1.7333756685256958, "learning_rate": 5.979000000000001e-06, "loss": 0.0249, "step": 20106 }, { "epoch": 1.071973130031455, "grad_norm": 2.149989366531372, "learning_rate": 5.9788e-06, "loss": 0.0144, "step": 20107 }, { "epoch": 1.0720264434611078, "grad_norm": 2.1614294052124023, "learning_rate": 5.9786e-06, "loss": 0.0666, "step": 20108 }, { "epoch": 1.0720797568907607, "grad_norm": 2.531426429748535, "learning_rate": 5.978400000000001e-06, "loss": 0.0289, "step": 20109 }, { "epoch": 1.0721330703204137, "grad_norm": 2.2562355995178223, "learning_rate": 5.978200000000001e-06, "loss": 0.026, "step": 20110 }, { "epoch": 1.0721863837500667, "grad_norm": 1.912886142730713, "learning_rate": 5.978e-06, "loss": 0.0113, "step": 20111 }, { "epoch": 1.0722396971797197, "grad_norm": 2.688408851623535, "learning_rate": 5.977800000000001e-06, "loss": 0.047, "step": 20112 }, { "epoch": 1.0722930106093724, "grad_norm": 1.9859755039215088, "learning_rate": 5.9776e-06, "loss": 0.0407, "step": 20113 }, { "epoch": 1.0723463240390254, "grad_norm": 3.477050542831421, "learning_rate": 5.9774000000000006e-06, "loss": 0.0274, "step": 20114 }, { "epoch": 1.0723996374686784, "grad_norm": 1.5716660022735596, "learning_rate": 5.977200000000001e-06, "loss": 0.0135, "step": 20115 }, { "epoch": 1.0724529508983314, "grad_norm": 1.8652465343475342, "learning_rate": 5.9770000000000005e-06, "loss": 0.0315, "step": 20116 }, { "epoch": 1.0725062643279841, "grad_norm": 3.3977372646331787, "learning_rate": 5.9768e-06, "loss": 0.0287, "step": 20117 }, { "epoch": 1.072559577757637, "grad_norm": 1.2372324466705322, "learning_rate": 5.9766000000000005e-06, "loss": 0.0098, "step": 20118 }, { "epoch": 1.07261289118729, "grad_norm": 2.264326810836792, "learning_rate": 5.9764e-06, "loss": 0.0284, "step": 20119 }, { "epoch": 1.072666204616943, "grad_norm": 1.8961156606674194, "learning_rate": 5.9762000000000005e-06, "loss": 0.0296, "step": 20120 }, { "epoch": 1.072719518046596, "grad_norm": 0.7594112753868103, "learning_rate": 5.976000000000001e-06, "loss": 0.0231, "step": 20121 }, { "epoch": 1.0727728314762488, "grad_norm": 1.473068118095398, "learning_rate": 5.9758000000000005e-06, "loss": 0.0282, "step": 20122 }, { "epoch": 1.0728261449059018, "grad_norm": 2.3591785430908203, "learning_rate": 5.9756e-06, "loss": 0.0471, "step": 20123 }, { "epoch": 1.0728794583355548, "grad_norm": 1.8138959407806396, "learning_rate": 5.9754e-06, "loss": 0.0198, "step": 20124 }, { "epoch": 1.0729327717652077, "grad_norm": 1.9159977436065674, "learning_rate": 5.975200000000001e-06, "loss": 0.0471, "step": 20125 }, { "epoch": 1.0729860851948605, "grad_norm": 2.2177181243896484, "learning_rate": 5.975e-06, "loss": 0.0402, "step": 20126 }, { "epoch": 1.0730393986245135, "grad_norm": 1.6503853797912598, "learning_rate": 5.974800000000001e-06, "loss": 0.0267, "step": 20127 }, { "epoch": 1.0730927120541665, "grad_norm": 2.15148663520813, "learning_rate": 5.9746e-06, "loss": 0.0447, "step": 20128 }, { "epoch": 1.0731460254838194, "grad_norm": 1.5912400484085083, "learning_rate": 5.9744e-06, "loss": 0.0183, "step": 20129 }, { "epoch": 1.0731993389134722, "grad_norm": 1.4795914888381958, "learning_rate": 5.9742e-06, "loss": 0.016, "step": 20130 }, { "epoch": 1.0732526523431252, "grad_norm": 2.125825881958008, "learning_rate": 5.974000000000001e-06, "loss": 0.0372, "step": 20131 }, { "epoch": 1.0733059657727781, "grad_norm": 2.2687599658966064, "learning_rate": 5.9738e-06, "loss": 0.0121, "step": 20132 }, { "epoch": 1.0733592792024311, "grad_norm": 1.9802199602127075, "learning_rate": 5.973600000000001e-06, "loss": 0.0269, "step": 20133 }, { "epoch": 1.073412592632084, "grad_norm": 3.70912504196167, "learning_rate": 5.9734e-06, "loss": 0.067, "step": 20134 }, { "epoch": 1.0734659060617369, "grad_norm": 1.2558833360671997, "learning_rate": 5.9732e-06, "loss": 0.0214, "step": 20135 }, { "epoch": 1.0735192194913898, "grad_norm": 3.584641456604004, "learning_rate": 5.973000000000001e-06, "loss": 0.0304, "step": 20136 }, { "epoch": 1.0735725329210428, "grad_norm": 1.6659038066864014, "learning_rate": 5.972800000000001e-06, "loss": 0.0347, "step": 20137 }, { "epoch": 1.0736258463506958, "grad_norm": 2.2058095932006836, "learning_rate": 5.9726e-06, "loss": 0.0385, "step": 20138 }, { "epoch": 1.0736791597803488, "grad_norm": 1.0172048807144165, "learning_rate": 5.972400000000001e-06, "loss": 0.0214, "step": 20139 }, { "epoch": 1.0737324732100015, "grad_norm": 2.303745985031128, "learning_rate": 5.9722e-06, "loss": 0.0464, "step": 20140 }, { "epoch": 1.0737857866396545, "grad_norm": 1.6757392883300781, "learning_rate": 5.972e-06, "loss": 0.0471, "step": 20141 }, { "epoch": 1.0738391000693075, "grad_norm": 1.0875024795532227, "learning_rate": 5.971800000000001e-06, "loss": 0.0134, "step": 20142 }, { "epoch": 1.0738924134989605, "grad_norm": 0.9384878277778625, "learning_rate": 5.9716000000000006e-06, "loss": 0.0103, "step": 20143 }, { "epoch": 1.0739457269286132, "grad_norm": 3.235783815383911, "learning_rate": 5.9714e-06, "loss": 0.0311, "step": 20144 }, { "epoch": 1.0739990403582662, "grad_norm": 1.7776620388031006, "learning_rate": 5.9712000000000005e-06, "loss": 0.0454, "step": 20145 }, { "epoch": 1.0740523537879192, "grad_norm": 3.7036402225494385, "learning_rate": 5.971e-06, "loss": 0.0643, "step": 20146 }, { "epoch": 1.0741056672175722, "grad_norm": 2.4253835678100586, "learning_rate": 5.970800000000001e-06, "loss": 0.0244, "step": 20147 }, { "epoch": 1.074158980647225, "grad_norm": 1.0914968252182007, "learning_rate": 5.970600000000001e-06, "loss": 0.0205, "step": 20148 }, { "epoch": 1.074212294076878, "grad_norm": 2.4810373783111572, "learning_rate": 5.9704000000000005e-06, "loss": 0.0218, "step": 20149 }, { "epoch": 1.0742656075065309, "grad_norm": 2.1263394355773926, "learning_rate": 5.9702e-06, "loss": 0.0341, "step": 20150 }, { "epoch": 1.0743189209361839, "grad_norm": 1.3253182172775269, "learning_rate": 5.9700000000000004e-06, "loss": 0.0191, "step": 20151 }, { "epoch": 1.0743722343658368, "grad_norm": 0.899968683719635, "learning_rate": 5.9698e-06, "loss": 0.0071, "step": 20152 }, { "epoch": 1.0744255477954896, "grad_norm": 4.1625447273254395, "learning_rate": 5.969600000000001e-06, "loss": 0.0459, "step": 20153 }, { "epoch": 1.0744788612251426, "grad_norm": 1.2450097799301147, "learning_rate": 5.969400000000001e-06, "loss": 0.0178, "step": 20154 }, { "epoch": 1.0745321746547956, "grad_norm": 2.693201780319214, "learning_rate": 5.9692e-06, "loss": 0.0348, "step": 20155 }, { "epoch": 1.0745854880844485, "grad_norm": 1.0972812175750732, "learning_rate": 5.969e-06, "loss": 0.0234, "step": 20156 }, { "epoch": 1.0746388015141015, "grad_norm": 1.4105154275894165, "learning_rate": 5.9688e-06, "loss": 0.0301, "step": 20157 }, { "epoch": 1.0746921149437543, "grad_norm": 0.8619718551635742, "learning_rate": 5.9686e-06, "loss": 0.0126, "step": 20158 }, { "epoch": 1.0747454283734073, "grad_norm": 0.7499278783798218, "learning_rate": 5.968400000000001e-06, "loss": 0.007, "step": 20159 }, { "epoch": 1.0747987418030602, "grad_norm": 2.0898752212524414, "learning_rate": 5.968200000000001e-06, "loss": 0.0196, "step": 20160 }, { "epoch": 1.0748520552327132, "grad_norm": 3.611912250518799, "learning_rate": 5.968e-06, "loss": 0.0187, "step": 20161 }, { "epoch": 1.074905368662366, "grad_norm": 3.324904441833496, "learning_rate": 5.9678e-06, "loss": 0.0583, "step": 20162 }, { "epoch": 1.074958682092019, "grad_norm": 2.167353868484497, "learning_rate": 5.9676e-06, "loss": 0.0417, "step": 20163 }, { "epoch": 1.075011995521672, "grad_norm": 0.7396671175956726, "learning_rate": 5.967400000000001e-06, "loss": 0.0196, "step": 20164 }, { "epoch": 1.075065308951325, "grad_norm": 2.6149239540100098, "learning_rate": 5.967200000000001e-06, "loss": 0.0531, "step": 20165 }, { "epoch": 1.0751186223809777, "grad_norm": 1.7294659614562988, "learning_rate": 5.967000000000001e-06, "loss": 0.0393, "step": 20166 }, { "epoch": 1.0751719358106306, "grad_norm": 2.159795045852661, "learning_rate": 5.9668e-06, "loss": 0.019, "step": 20167 }, { "epoch": 1.0752252492402836, "grad_norm": 2.2691807746887207, "learning_rate": 5.9666e-06, "loss": 0.0278, "step": 20168 }, { "epoch": 1.0752785626699366, "grad_norm": 1.1564958095550537, "learning_rate": 5.9664e-06, "loss": 0.0362, "step": 20169 }, { "epoch": 1.0753318760995896, "grad_norm": 0.9652665853500366, "learning_rate": 5.966200000000001e-06, "loss": 0.0163, "step": 20170 }, { "epoch": 1.0753851895292423, "grad_norm": 2.9642117023468018, "learning_rate": 5.966000000000001e-06, "loss": 0.0386, "step": 20171 }, { "epoch": 1.0754385029588953, "grad_norm": 3.8243095874786377, "learning_rate": 5.9658000000000005e-06, "loss": 0.0295, "step": 20172 }, { "epoch": 1.0754918163885483, "grad_norm": 1.6091893911361694, "learning_rate": 5.9656e-06, "loss": 0.0287, "step": 20173 }, { "epoch": 1.0755451298182013, "grad_norm": 1.4453023672103882, "learning_rate": 5.9654e-06, "loss": 0.0187, "step": 20174 }, { "epoch": 1.0755984432478543, "grad_norm": 3.279454231262207, "learning_rate": 5.965200000000001e-06, "loss": 0.0307, "step": 20175 }, { "epoch": 1.075651756677507, "grad_norm": 3.3592073917388916, "learning_rate": 5.9650000000000005e-06, "loss": 0.0238, "step": 20176 }, { "epoch": 1.07570507010716, "grad_norm": 1.1164398193359375, "learning_rate": 5.964800000000001e-06, "loss": 0.0172, "step": 20177 }, { "epoch": 1.075758383536813, "grad_norm": 3.9032974243164062, "learning_rate": 5.9646000000000005e-06, "loss": 0.0356, "step": 20178 }, { "epoch": 1.075811696966466, "grad_norm": 1.3878233432769775, "learning_rate": 5.9644e-06, "loss": 0.0388, "step": 20179 }, { "epoch": 1.0758650103961187, "grad_norm": 1.3089604377746582, "learning_rate": 5.9642e-06, "loss": 0.0246, "step": 20180 }, { "epoch": 1.0759183238257717, "grad_norm": 1.14486563205719, "learning_rate": 5.964000000000001e-06, "loss": 0.0328, "step": 20181 }, { "epoch": 1.0759716372554247, "grad_norm": 0.8936038613319397, "learning_rate": 5.9638e-06, "loss": 0.0156, "step": 20182 }, { "epoch": 1.0760249506850776, "grad_norm": 1.9540915489196777, "learning_rate": 5.963600000000001e-06, "loss": 0.0267, "step": 20183 }, { "epoch": 1.0760782641147304, "grad_norm": 0.7302449941635132, "learning_rate": 5.9634e-06, "loss": 0.0079, "step": 20184 }, { "epoch": 1.0761315775443834, "grad_norm": 1.9791650772094727, "learning_rate": 5.9632e-06, "loss": 0.035, "step": 20185 }, { "epoch": 1.0761848909740364, "grad_norm": 2.5471296310424805, "learning_rate": 5.963000000000001e-06, "loss": 0.0626, "step": 20186 }, { "epoch": 1.0762382044036893, "grad_norm": 2.0357272624969482, "learning_rate": 5.962800000000001e-06, "loss": 0.0298, "step": 20187 }, { "epoch": 1.0762915178333423, "grad_norm": 1.8298300504684448, "learning_rate": 5.9626e-06, "loss": 0.0299, "step": 20188 }, { "epoch": 1.076344831262995, "grad_norm": 1.1577610969543457, "learning_rate": 5.962400000000001e-06, "loss": 0.0354, "step": 20189 }, { "epoch": 1.076398144692648, "grad_norm": 0.8480247259140015, "learning_rate": 5.9622e-06, "loss": 0.0096, "step": 20190 }, { "epoch": 1.076451458122301, "grad_norm": 1.1635571718215942, "learning_rate": 5.962e-06, "loss": 0.0138, "step": 20191 }, { "epoch": 1.076504771551954, "grad_norm": 1.8030606508255005, "learning_rate": 5.961800000000001e-06, "loss": 0.0198, "step": 20192 }, { "epoch": 1.0765580849816068, "grad_norm": 3.7873876094818115, "learning_rate": 5.961600000000001e-06, "loss": 0.0464, "step": 20193 }, { "epoch": 1.0766113984112597, "grad_norm": 0.5676878690719604, "learning_rate": 5.9614e-06, "loss": 0.0071, "step": 20194 }, { "epoch": 1.0766647118409127, "grad_norm": 0.9838398694992065, "learning_rate": 5.961200000000001e-06, "loss": 0.0298, "step": 20195 }, { "epoch": 1.0767180252705657, "grad_norm": 1.808713436126709, "learning_rate": 5.961e-06, "loss": 0.0325, "step": 20196 }, { "epoch": 1.0767713387002187, "grad_norm": 2.210451602935791, "learning_rate": 5.9608000000000014e-06, "loss": 0.043, "step": 20197 }, { "epoch": 1.0768246521298714, "grad_norm": 1.4326043128967285, "learning_rate": 5.960600000000001e-06, "loss": 0.0382, "step": 20198 }, { "epoch": 1.0768779655595244, "grad_norm": 1.4579945802688599, "learning_rate": 5.9604000000000006e-06, "loss": 0.019, "step": 20199 }, { "epoch": 1.0769312789891774, "grad_norm": 2.5339884757995605, "learning_rate": 5.9602e-06, "loss": 0.0401, "step": 20200 }, { "epoch": 1.0769845924188304, "grad_norm": 1.8022257089614868, "learning_rate": 5.9600000000000005e-06, "loss": 0.0339, "step": 20201 }, { "epoch": 1.0770379058484831, "grad_norm": 2.965916395187378, "learning_rate": 5.9598e-06, "loss": 0.0443, "step": 20202 }, { "epoch": 1.0770912192781361, "grad_norm": 2.1830687522888184, "learning_rate": 5.959600000000001e-06, "loss": 0.0331, "step": 20203 }, { "epoch": 1.077144532707789, "grad_norm": 1.9524089097976685, "learning_rate": 5.959400000000001e-06, "loss": 0.0284, "step": 20204 }, { "epoch": 1.077197846137442, "grad_norm": 1.3927336931228638, "learning_rate": 5.9592000000000005e-06, "loss": 0.0221, "step": 20205 }, { "epoch": 1.077251159567095, "grad_norm": 2.0816259384155273, "learning_rate": 5.959e-06, "loss": 0.0353, "step": 20206 }, { "epoch": 1.0773044729967478, "grad_norm": 1.001665472984314, "learning_rate": 5.9588000000000004e-06, "loss": 0.013, "step": 20207 }, { "epoch": 1.0773577864264008, "grad_norm": 1.5721244812011719, "learning_rate": 5.9586e-06, "loss": 0.0151, "step": 20208 }, { "epoch": 1.0774110998560538, "grad_norm": 1.511025309562683, "learning_rate": 5.958400000000001e-06, "loss": 0.0413, "step": 20209 }, { "epoch": 1.0774644132857067, "grad_norm": 0.6078256368637085, "learning_rate": 5.958200000000001e-06, "loss": 0.0056, "step": 20210 }, { "epoch": 1.0775177267153595, "grad_norm": 1.2612121105194092, "learning_rate": 5.958e-06, "loss": 0.015, "step": 20211 }, { "epoch": 1.0775710401450125, "grad_norm": 0.8717972636222839, "learning_rate": 5.9578e-06, "loss": 0.018, "step": 20212 }, { "epoch": 1.0776243535746655, "grad_norm": 1.9613995552062988, "learning_rate": 5.9576e-06, "loss": 0.0164, "step": 20213 }, { "epoch": 1.0776776670043184, "grad_norm": 3.776045322418213, "learning_rate": 5.957400000000001e-06, "loss": 0.0387, "step": 20214 }, { "epoch": 1.0777309804339714, "grad_norm": 4.32127571105957, "learning_rate": 5.957200000000001e-06, "loss": 0.0412, "step": 20215 }, { "epoch": 1.0777842938636242, "grad_norm": 0.9696156978607178, "learning_rate": 5.957000000000001e-06, "loss": 0.0127, "step": 20216 }, { "epoch": 1.0778376072932772, "grad_norm": 1.3513163328170776, "learning_rate": 5.9568e-06, "loss": 0.0261, "step": 20217 }, { "epoch": 1.0778909207229301, "grad_norm": 0.9476693272590637, "learning_rate": 5.9566e-06, "loss": 0.0143, "step": 20218 }, { "epoch": 1.0779442341525831, "grad_norm": 2.099177837371826, "learning_rate": 5.9564e-06, "loss": 0.0202, "step": 20219 }, { "epoch": 1.0779975475822359, "grad_norm": 1.2704352140426636, "learning_rate": 5.956200000000001e-06, "loss": 0.0321, "step": 20220 }, { "epoch": 1.0780508610118889, "grad_norm": 1.8142023086547852, "learning_rate": 5.956000000000001e-06, "loss": 0.0352, "step": 20221 }, { "epoch": 1.0781041744415418, "grad_norm": 2.025644302368164, "learning_rate": 5.955800000000001e-06, "loss": 0.0123, "step": 20222 }, { "epoch": 1.0781574878711948, "grad_norm": 1.385050654411316, "learning_rate": 5.9556e-06, "loss": 0.0201, "step": 20223 }, { "epoch": 1.0782108013008478, "grad_norm": 1.795654058456421, "learning_rate": 5.9554e-06, "loss": 0.0209, "step": 20224 }, { "epoch": 1.0782641147305005, "grad_norm": 1.4649848937988281, "learning_rate": 5.955200000000001e-06, "loss": 0.0166, "step": 20225 }, { "epoch": 1.0783174281601535, "grad_norm": 2.44335675239563, "learning_rate": 5.955000000000001e-06, "loss": 0.0152, "step": 20226 }, { "epoch": 1.0783707415898065, "grad_norm": 1.23157799243927, "learning_rate": 5.954800000000001e-06, "loss": 0.0255, "step": 20227 }, { "epoch": 1.0784240550194595, "grad_norm": 3.7144107818603516, "learning_rate": 5.9546000000000006e-06, "loss": 0.0222, "step": 20228 }, { "epoch": 1.0784773684491122, "grad_norm": 1.397385597229004, "learning_rate": 5.9544e-06, "loss": 0.0479, "step": 20229 }, { "epoch": 1.0785306818787652, "grad_norm": 1.3869946002960205, "learning_rate": 5.9542e-06, "loss": 0.0654, "step": 20230 }, { "epoch": 1.0785839953084182, "grad_norm": 1.6905713081359863, "learning_rate": 5.954000000000001e-06, "loss": 0.0319, "step": 20231 }, { "epoch": 1.0786373087380712, "grad_norm": 0.7920662760734558, "learning_rate": 5.9538000000000005e-06, "loss": 0.0079, "step": 20232 }, { "epoch": 1.078690622167724, "grad_norm": 2.259194850921631, "learning_rate": 5.953600000000001e-06, "loss": 0.044, "step": 20233 }, { "epoch": 1.078743935597377, "grad_norm": 1.982828140258789, "learning_rate": 5.9534000000000005e-06, "loss": 0.028, "step": 20234 }, { "epoch": 1.07879724902703, "grad_norm": 3.4072468280792236, "learning_rate": 5.9532e-06, "loss": 0.0432, "step": 20235 }, { "epoch": 1.0788505624566829, "grad_norm": 1.1040610074996948, "learning_rate": 5.953000000000001e-06, "loss": 0.025, "step": 20236 }, { "epoch": 1.0789038758863359, "grad_norm": 2.2426486015319824, "learning_rate": 5.952800000000001e-06, "loss": 0.0487, "step": 20237 }, { "epoch": 1.0789571893159886, "grad_norm": 2.1634469032287598, "learning_rate": 5.9526e-06, "loss": 0.0169, "step": 20238 }, { "epoch": 1.0790105027456416, "grad_norm": 1.6240801811218262, "learning_rate": 5.952400000000001e-06, "loss": 0.0349, "step": 20239 }, { "epoch": 1.0790638161752946, "grad_norm": 1.596386432647705, "learning_rate": 5.9522e-06, "loss": 0.0214, "step": 20240 }, { "epoch": 1.0791171296049475, "grad_norm": 1.1704328060150146, "learning_rate": 5.952e-06, "loss": 0.0098, "step": 20241 }, { "epoch": 1.0791704430346005, "grad_norm": 3.634873151779175, "learning_rate": 5.951800000000001e-06, "loss": 0.0393, "step": 20242 }, { "epoch": 1.0792237564642533, "grad_norm": 3.6513688564300537, "learning_rate": 5.951600000000001e-06, "loss": 0.0337, "step": 20243 }, { "epoch": 1.0792770698939063, "grad_norm": 2.576160192489624, "learning_rate": 5.9514e-06, "loss": 0.0628, "step": 20244 }, { "epoch": 1.0793303833235592, "grad_norm": 3.7204794883728027, "learning_rate": 5.951200000000001e-06, "loss": 0.0127, "step": 20245 }, { "epoch": 1.0793836967532122, "grad_norm": 0.5823072791099548, "learning_rate": 5.951e-06, "loss": 0.0065, "step": 20246 }, { "epoch": 1.079437010182865, "grad_norm": 1.2103253602981567, "learning_rate": 5.950800000000001e-06, "loss": 0.0186, "step": 20247 }, { "epoch": 1.079490323612518, "grad_norm": 1.6511530876159668, "learning_rate": 5.950600000000001e-06, "loss": 0.0365, "step": 20248 }, { "epoch": 1.079543637042171, "grad_norm": 1.8494890928268433, "learning_rate": 5.950400000000001e-06, "loss": 0.0549, "step": 20249 }, { "epoch": 1.079596950471824, "grad_norm": 1.5416573286056519, "learning_rate": 5.9502e-06, "loss": 0.0304, "step": 20250 }, { "epoch": 1.0796502639014767, "grad_norm": 0.7840818762779236, "learning_rate": 5.950000000000001e-06, "loss": 0.007, "step": 20251 }, { "epoch": 1.0797035773311296, "grad_norm": 4.040489196777344, "learning_rate": 5.9498e-06, "loss": 0.0569, "step": 20252 }, { "epoch": 1.0797568907607826, "grad_norm": 1.7632638216018677, "learning_rate": 5.949600000000001e-06, "loss": 0.039, "step": 20253 }, { "epoch": 1.0798102041904356, "grad_norm": 2.234893798828125, "learning_rate": 5.949400000000001e-06, "loss": 0.0348, "step": 20254 }, { "epoch": 1.0798635176200886, "grad_norm": 1.5139515399932861, "learning_rate": 5.9492000000000006e-06, "loss": 0.0118, "step": 20255 }, { "epoch": 1.0799168310497413, "grad_norm": 0.7299802899360657, "learning_rate": 5.949e-06, "loss": 0.0077, "step": 20256 }, { "epoch": 1.0799701444793943, "grad_norm": 1.2508703470230103, "learning_rate": 5.9488000000000005e-06, "loss": 0.0265, "step": 20257 }, { "epoch": 1.0800234579090473, "grad_norm": 1.7945457696914673, "learning_rate": 5.948600000000001e-06, "loss": 0.041, "step": 20258 }, { "epoch": 1.0800767713387003, "grad_norm": 4.231510162353516, "learning_rate": 5.9484000000000005e-06, "loss": 0.0252, "step": 20259 }, { "epoch": 1.0801300847683533, "grad_norm": 1.732539176940918, "learning_rate": 5.948200000000001e-06, "loss": 0.0229, "step": 20260 }, { "epoch": 1.080183398198006, "grad_norm": 0.7815548181533813, "learning_rate": 5.9480000000000005e-06, "loss": 0.0102, "step": 20261 }, { "epoch": 1.080236711627659, "grad_norm": 2.0552756786346436, "learning_rate": 5.9478e-06, "loss": 0.0388, "step": 20262 }, { "epoch": 1.080290025057312, "grad_norm": 1.421011209487915, "learning_rate": 5.9476000000000005e-06, "loss": 0.0114, "step": 20263 }, { "epoch": 1.080343338486965, "grad_norm": 3.009092092514038, "learning_rate": 5.947400000000001e-06, "loss": 0.0322, "step": 20264 }, { "epoch": 1.0803966519166177, "grad_norm": 4.400805950164795, "learning_rate": 5.9472e-06, "loss": 0.0199, "step": 20265 }, { "epoch": 1.0804499653462707, "grad_norm": 1.5907355546951294, "learning_rate": 5.947000000000001e-06, "loss": 0.0175, "step": 20266 }, { "epoch": 1.0805032787759237, "grad_norm": 1.1541638374328613, "learning_rate": 5.9468e-06, "loss": 0.0135, "step": 20267 }, { "epoch": 1.0805565922055766, "grad_norm": 0.6593321561813354, "learning_rate": 5.9466e-06, "loss": 0.0059, "step": 20268 }, { "epoch": 1.0806099056352294, "grad_norm": 1.407309651374817, "learning_rate": 5.9464e-06, "loss": 0.0166, "step": 20269 }, { "epoch": 1.0806632190648824, "grad_norm": 1.893234133720398, "learning_rate": 5.946200000000001e-06, "loss": 0.0375, "step": 20270 }, { "epoch": 1.0807165324945354, "grad_norm": 0.934939980506897, "learning_rate": 5.946e-06, "loss": 0.0159, "step": 20271 }, { "epoch": 1.0807698459241883, "grad_norm": 1.8092987537384033, "learning_rate": 5.945800000000001e-06, "loss": 0.0425, "step": 20272 }, { "epoch": 1.0808231593538413, "grad_norm": 0.9571378827095032, "learning_rate": 5.9456e-06, "loss": 0.0121, "step": 20273 }, { "epoch": 1.080876472783494, "grad_norm": 1.0903325080871582, "learning_rate": 5.9454e-06, "loss": 0.0174, "step": 20274 }, { "epoch": 1.080929786213147, "grad_norm": 1.6528981924057007, "learning_rate": 5.945200000000001e-06, "loss": 0.0183, "step": 20275 }, { "epoch": 1.0809830996428, "grad_norm": 1.7979429960250854, "learning_rate": 5.945000000000001e-06, "loss": 0.0368, "step": 20276 }, { "epoch": 1.081036413072453, "grad_norm": 1.0970691442489624, "learning_rate": 5.9448e-06, "loss": 0.0089, "step": 20277 }, { "epoch": 1.081089726502106, "grad_norm": 1.082161545753479, "learning_rate": 5.944600000000001e-06, "loss": 0.0072, "step": 20278 }, { "epoch": 1.0811430399317588, "grad_norm": 2.0840940475463867, "learning_rate": 5.9444e-06, "loss": 0.0204, "step": 20279 }, { "epoch": 1.0811963533614117, "grad_norm": 2.588289737701416, "learning_rate": 5.9442e-06, "loss": 0.0312, "step": 20280 }, { "epoch": 1.0812496667910647, "grad_norm": 2.0863187313079834, "learning_rate": 5.944000000000001e-06, "loss": 0.0518, "step": 20281 }, { "epoch": 1.0813029802207177, "grad_norm": 1.162802815437317, "learning_rate": 5.943800000000001e-06, "loss": 0.0185, "step": 20282 }, { "epoch": 1.0813562936503704, "grad_norm": 2.4879064559936523, "learning_rate": 5.9436e-06, "loss": 0.01, "step": 20283 }, { "epoch": 1.0814096070800234, "grad_norm": 1.7538495063781738, "learning_rate": 5.9434000000000006e-06, "loss": 0.0183, "step": 20284 }, { "epoch": 1.0814629205096764, "grad_norm": 2.3305704593658447, "learning_rate": 5.9432e-06, "loss": 0.0382, "step": 20285 }, { "epoch": 1.0815162339393294, "grad_norm": 2.223278522491455, "learning_rate": 5.943000000000001e-06, "loss": 0.029, "step": 20286 }, { "epoch": 1.0815695473689821, "grad_norm": 1.7306172847747803, "learning_rate": 5.942800000000001e-06, "loss": 0.0317, "step": 20287 }, { "epoch": 1.0816228607986351, "grad_norm": 2.85233736038208, "learning_rate": 5.9426000000000005e-06, "loss": 0.0191, "step": 20288 }, { "epoch": 1.081676174228288, "grad_norm": 1.598608136177063, "learning_rate": 5.9424e-06, "loss": 0.0178, "step": 20289 }, { "epoch": 1.081729487657941, "grad_norm": 1.2207902669906616, "learning_rate": 5.9422000000000005e-06, "loss": 0.0115, "step": 20290 }, { "epoch": 1.081782801087594, "grad_norm": 1.8311219215393066, "learning_rate": 5.942e-06, "loss": 0.027, "step": 20291 }, { "epoch": 1.0818361145172468, "grad_norm": 1.7337952852249146, "learning_rate": 5.941800000000001e-06, "loss": 0.0212, "step": 20292 }, { "epoch": 1.0818894279468998, "grad_norm": 2.7825827598571777, "learning_rate": 5.941600000000001e-06, "loss": 0.0223, "step": 20293 }, { "epoch": 1.0819427413765528, "grad_norm": 3.8584823608398438, "learning_rate": 5.9414e-06, "loss": 0.0458, "step": 20294 }, { "epoch": 1.0819960548062058, "grad_norm": 2.0499138832092285, "learning_rate": 5.9412e-06, "loss": 0.0352, "step": 20295 }, { "epoch": 1.0820493682358585, "grad_norm": 2.014486789703369, "learning_rate": 5.941e-06, "loss": 0.0359, "step": 20296 }, { "epoch": 1.0821026816655115, "grad_norm": 2.429966449737549, "learning_rate": 5.940800000000001e-06, "loss": 0.0274, "step": 20297 }, { "epoch": 1.0821559950951645, "grad_norm": 4.890471935272217, "learning_rate": 5.940600000000001e-06, "loss": 0.0632, "step": 20298 }, { "epoch": 1.0822093085248174, "grad_norm": 1.961145281791687, "learning_rate": 5.940400000000001e-06, "loss": 0.0237, "step": 20299 }, { "epoch": 1.0822626219544704, "grad_norm": 3.125030994415283, "learning_rate": 5.9402e-06, "loss": 0.0432, "step": 20300 }, { "epoch": 1.0823159353841232, "grad_norm": 6.195494651794434, "learning_rate": 5.94e-06, "loss": 0.0646, "step": 20301 }, { "epoch": 1.0823692488137762, "grad_norm": 2.0600931644439697, "learning_rate": 5.9398e-06, "loss": 0.0206, "step": 20302 }, { "epoch": 1.0824225622434291, "grad_norm": 1.779776930809021, "learning_rate": 5.939600000000001e-06, "loss": 0.0107, "step": 20303 }, { "epoch": 1.0824758756730821, "grad_norm": 0.46421003341674805, "learning_rate": 5.939400000000001e-06, "loss": 0.0045, "step": 20304 }, { "epoch": 1.0825291891027349, "grad_norm": 3.698134422302246, "learning_rate": 5.939200000000001e-06, "loss": 0.0229, "step": 20305 }, { "epoch": 1.0825825025323879, "grad_norm": 3.9227218627929688, "learning_rate": 5.939e-06, "loss": 0.0381, "step": 20306 }, { "epoch": 1.0826358159620408, "grad_norm": 1.5766769647598267, "learning_rate": 5.9388e-06, "loss": 0.0314, "step": 20307 }, { "epoch": 1.0826891293916938, "grad_norm": 1.79776132106781, "learning_rate": 5.938600000000001e-06, "loss": 0.0493, "step": 20308 }, { "epoch": 1.0827424428213468, "grad_norm": 2.304914712905884, "learning_rate": 5.938400000000001e-06, "loss": 0.0516, "step": 20309 }, { "epoch": 1.0827957562509996, "grad_norm": 2.7022366523742676, "learning_rate": 5.938200000000001e-06, "loss": 0.0595, "step": 20310 }, { "epoch": 1.0828490696806525, "grad_norm": 3.2945473194122314, "learning_rate": 5.9380000000000006e-06, "loss": 0.0293, "step": 20311 }, { "epoch": 1.0829023831103055, "grad_norm": 1.0014100074768066, "learning_rate": 5.9378e-06, "loss": 0.013, "step": 20312 }, { "epoch": 1.0829556965399585, "grad_norm": 1.5917787551879883, "learning_rate": 5.9376e-06, "loss": 0.0249, "step": 20313 }, { "epoch": 1.0830090099696112, "grad_norm": 0.9378961324691772, "learning_rate": 5.937400000000001e-06, "loss": 0.0212, "step": 20314 }, { "epoch": 1.0830623233992642, "grad_norm": 2.18416166305542, "learning_rate": 5.9372000000000005e-06, "loss": 0.0476, "step": 20315 }, { "epoch": 1.0831156368289172, "grad_norm": 0.8740324974060059, "learning_rate": 5.937000000000001e-06, "loss": 0.0089, "step": 20316 }, { "epoch": 1.0831689502585702, "grad_norm": 2.767210006713867, "learning_rate": 5.9368000000000005e-06, "loss": 0.0348, "step": 20317 }, { "epoch": 1.0832222636882232, "grad_norm": 1.5833395719528198, "learning_rate": 5.9366e-06, "loss": 0.0208, "step": 20318 }, { "epoch": 1.083275577117876, "grad_norm": 3.033916711807251, "learning_rate": 5.9364e-06, "loss": 0.0317, "step": 20319 }, { "epoch": 1.083328890547529, "grad_norm": 1.8843762874603271, "learning_rate": 5.936200000000001e-06, "loss": 0.0355, "step": 20320 }, { "epoch": 1.0833822039771819, "grad_norm": 1.179945945739746, "learning_rate": 5.9360000000000004e-06, "loss": 0.0271, "step": 20321 }, { "epoch": 1.0834355174068349, "grad_norm": 1.8761165142059326, "learning_rate": 5.935800000000001e-06, "loss": 0.0243, "step": 20322 }, { "epoch": 1.0834888308364876, "grad_norm": 0.9247312545776367, "learning_rate": 5.9356e-06, "loss": 0.0082, "step": 20323 }, { "epoch": 1.0835421442661406, "grad_norm": 3.1860663890838623, "learning_rate": 5.9354e-06, "loss": 0.0313, "step": 20324 }, { "epoch": 1.0835954576957936, "grad_norm": 2.1767430305480957, "learning_rate": 5.935200000000001e-06, "loss": 0.0329, "step": 20325 }, { "epoch": 1.0836487711254466, "grad_norm": 0.9236921668052673, "learning_rate": 5.935000000000001e-06, "loss": 0.0201, "step": 20326 }, { "epoch": 1.0837020845550995, "grad_norm": 2.605720043182373, "learning_rate": 5.9348e-06, "loss": 0.0278, "step": 20327 }, { "epoch": 1.0837553979847523, "grad_norm": 0.6901886463165283, "learning_rate": 5.934600000000001e-06, "loss": 0.0055, "step": 20328 }, { "epoch": 1.0838087114144053, "grad_norm": 1.7868366241455078, "learning_rate": 5.9344e-06, "loss": 0.0157, "step": 20329 }, { "epoch": 1.0838620248440582, "grad_norm": 5.581611156463623, "learning_rate": 5.9342e-06, "loss": 0.0472, "step": 20330 }, { "epoch": 1.0839153382737112, "grad_norm": 2.261838674545288, "learning_rate": 5.934000000000001e-06, "loss": 0.0168, "step": 20331 }, { "epoch": 1.083968651703364, "grad_norm": 2.298872232437134, "learning_rate": 5.933800000000001e-06, "loss": 0.0465, "step": 20332 }, { "epoch": 1.084021965133017, "grad_norm": 1.568200945854187, "learning_rate": 5.9336e-06, "loss": 0.0139, "step": 20333 }, { "epoch": 1.08407527856267, "grad_norm": 1.6320563554763794, "learning_rate": 5.933400000000001e-06, "loss": 0.0211, "step": 20334 }, { "epoch": 1.084128591992323, "grad_norm": 1.593939185142517, "learning_rate": 5.9332e-06, "loss": 0.0229, "step": 20335 }, { "epoch": 1.0841819054219757, "grad_norm": 0.789836585521698, "learning_rate": 5.933000000000001e-06, "loss": 0.0115, "step": 20336 }, { "epoch": 1.0842352188516287, "grad_norm": 1.953696370124817, "learning_rate": 5.932800000000001e-06, "loss": 0.0189, "step": 20337 }, { "epoch": 1.0842885322812816, "grad_norm": 1.9322518110275269, "learning_rate": 5.932600000000001e-06, "loss": 0.0269, "step": 20338 }, { "epoch": 1.0843418457109346, "grad_norm": 1.9152483940124512, "learning_rate": 5.9324e-06, "loss": 0.0144, "step": 20339 }, { "epoch": 1.0843951591405876, "grad_norm": 1.3087573051452637, "learning_rate": 5.9322000000000006e-06, "loss": 0.0225, "step": 20340 }, { "epoch": 1.0844484725702404, "grad_norm": 0.9509873390197754, "learning_rate": 5.932e-06, "loss": 0.0094, "step": 20341 }, { "epoch": 1.0845017859998933, "grad_norm": 1.13055419921875, "learning_rate": 5.9318000000000005e-06, "loss": 0.0136, "step": 20342 }, { "epoch": 1.0845550994295463, "grad_norm": 2.3617258071899414, "learning_rate": 5.931600000000001e-06, "loss": 0.0272, "step": 20343 }, { "epoch": 1.0846084128591993, "grad_norm": 0.3038690388202667, "learning_rate": 5.9314000000000005e-06, "loss": 0.0029, "step": 20344 }, { "epoch": 1.0846617262888523, "grad_norm": 3.2020857334136963, "learning_rate": 5.9312e-06, "loss": 0.0501, "step": 20345 }, { "epoch": 1.084715039718505, "grad_norm": 1.1136934757232666, "learning_rate": 5.9310000000000005e-06, "loss": 0.0247, "step": 20346 }, { "epoch": 1.084768353148158, "grad_norm": 2.0540802478790283, "learning_rate": 5.930800000000001e-06, "loss": 0.0221, "step": 20347 }, { "epoch": 1.084821666577811, "grad_norm": 1.8070777654647827, "learning_rate": 5.9306000000000004e-06, "loss": 0.0172, "step": 20348 }, { "epoch": 1.084874980007464, "grad_norm": 1.7095983028411865, "learning_rate": 5.930400000000001e-06, "loss": 0.0264, "step": 20349 }, { "epoch": 1.0849282934371167, "grad_norm": 1.3035075664520264, "learning_rate": 5.9302e-06, "loss": 0.0289, "step": 20350 }, { "epoch": 1.0849816068667697, "grad_norm": 1.4741075038909912, "learning_rate": 5.93e-06, "loss": 0.0158, "step": 20351 }, { "epoch": 1.0850349202964227, "grad_norm": 1.490999460220337, "learning_rate": 5.9298e-06, "loss": 0.0094, "step": 20352 }, { "epoch": 1.0850882337260757, "grad_norm": 1.4487426280975342, "learning_rate": 5.929600000000001e-06, "loss": 0.0291, "step": 20353 }, { "epoch": 1.0851415471557284, "grad_norm": 1.9388710260391235, "learning_rate": 5.9294e-06, "loss": 0.033, "step": 20354 }, { "epoch": 1.0851948605853814, "grad_norm": 1.0868895053863525, "learning_rate": 5.929200000000001e-06, "loss": 0.0095, "step": 20355 }, { "epoch": 1.0852481740150344, "grad_norm": 1.8575177192687988, "learning_rate": 5.929e-06, "loss": 0.0287, "step": 20356 }, { "epoch": 1.0853014874446874, "grad_norm": 0.9054358005523682, "learning_rate": 5.9288e-06, "loss": 0.0235, "step": 20357 }, { "epoch": 1.0853548008743403, "grad_norm": 1.1604005098342896, "learning_rate": 5.928600000000001e-06, "loss": 0.024, "step": 20358 }, { "epoch": 1.085408114303993, "grad_norm": 2.194108009338379, "learning_rate": 5.928400000000001e-06, "loss": 0.0166, "step": 20359 }, { "epoch": 1.085461427733646, "grad_norm": 2.9210116863250732, "learning_rate": 5.9282e-06, "loss": 0.0293, "step": 20360 }, { "epoch": 1.085514741163299, "grad_norm": 2.5804576873779297, "learning_rate": 5.928000000000001e-06, "loss": 0.0164, "step": 20361 }, { "epoch": 1.085568054592952, "grad_norm": 1.5018868446350098, "learning_rate": 5.9278e-06, "loss": 0.0159, "step": 20362 }, { "epoch": 1.085621368022605, "grad_norm": 0.9294931888580322, "learning_rate": 5.9276e-06, "loss": 0.0071, "step": 20363 }, { "epoch": 1.0856746814522578, "grad_norm": 2.3436524868011475, "learning_rate": 5.927400000000001e-06, "loss": 0.0372, "step": 20364 }, { "epoch": 1.0857279948819107, "grad_norm": 3.395728588104248, "learning_rate": 5.927200000000001e-06, "loss": 0.0264, "step": 20365 }, { "epoch": 1.0857813083115637, "grad_norm": 4.559617042541504, "learning_rate": 5.927e-06, "loss": 0.0337, "step": 20366 }, { "epoch": 1.0858346217412167, "grad_norm": 1.7167587280273438, "learning_rate": 5.926800000000001e-06, "loss": 0.0324, "step": 20367 }, { "epoch": 1.0858879351708695, "grad_norm": 2.0287060737609863, "learning_rate": 5.9266e-06, "loss": 0.0324, "step": 20368 }, { "epoch": 1.0859412486005224, "grad_norm": 1.6082346439361572, "learning_rate": 5.9264e-06, "loss": 0.0261, "step": 20369 }, { "epoch": 1.0859945620301754, "grad_norm": 2.496490478515625, "learning_rate": 5.926200000000001e-06, "loss": 0.0485, "step": 20370 }, { "epoch": 1.0860478754598284, "grad_norm": 3.971813201904297, "learning_rate": 5.9260000000000005e-06, "loss": 0.0344, "step": 20371 }, { "epoch": 1.0861011888894812, "grad_norm": 1.6992145776748657, "learning_rate": 5.9258e-06, "loss": 0.0233, "step": 20372 }, { "epoch": 1.0861545023191341, "grad_norm": 0.5122431516647339, "learning_rate": 5.9256000000000005e-06, "loss": 0.0039, "step": 20373 }, { "epoch": 1.086207815748787, "grad_norm": 1.2498358488082886, "learning_rate": 5.9254e-06, "loss": 0.0164, "step": 20374 }, { "epoch": 1.08626112917844, "grad_norm": 2.6135902404785156, "learning_rate": 5.925200000000001e-06, "loss": 0.0429, "step": 20375 }, { "epoch": 1.086314442608093, "grad_norm": 1.0812278985977173, "learning_rate": 5.925000000000001e-06, "loss": 0.0192, "step": 20376 }, { "epoch": 1.0863677560377458, "grad_norm": 2.056727886199951, "learning_rate": 5.9248000000000004e-06, "loss": 0.0328, "step": 20377 }, { "epoch": 1.0864210694673988, "grad_norm": 0.9639642238616943, "learning_rate": 5.9246e-06, "loss": 0.0062, "step": 20378 }, { "epoch": 1.0864743828970518, "grad_norm": 2.167365074157715, "learning_rate": 5.9244e-06, "loss": 0.0348, "step": 20379 }, { "epoch": 1.0865276963267048, "grad_norm": 1.871297001838684, "learning_rate": 5.9242e-06, "loss": 0.0232, "step": 20380 }, { "epoch": 1.0865810097563577, "grad_norm": 1.7174443006515503, "learning_rate": 5.924000000000001e-06, "loss": 0.026, "step": 20381 }, { "epoch": 1.0866343231860105, "grad_norm": 1.0586434602737427, "learning_rate": 5.923800000000001e-06, "loss": 0.0087, "step": 20382 }, { "epoch": 1.0866876366156635, "grad_norm": 2.354292869567871, "learning_rate": 5.9236e-06, "loss": 0.0178, "step": 20383 }, { "epoch": 1.0867409500453165, "grad_norm": 2.051557779312134, "learning_rate": 5.9234e-06, "loss": 0.0245, "step": 20384 }, { "epoch": 1.0867942634749694, "grad_norm": 6.364082336425781, "learning_rate": 5.9232e-06, "loss": 0.0849, "step": 20385 }, { "epoch": 1.0868475769046222, "grad_norm": 1.8219836950302124, "learning_rate": 5.923000000000001e-06, "loss": 0.0274, "step": 20386 }, { "epoch": 1.0869008903342752, "grad_norm": 2.1270532608032227, "learning_rate": 5.922800000000001e-06, "loss": 0.0387, "step": 20387 }, { "epoch": 1.0869542037639282, "grad_norm": 1.8971610069274902, "learning_rate": 5.922600000000001e-06, "loss": 0.0287, "step": 20388 }, { "epoch": 1.0870075171935811, "grad_norm": 1.7938737869262695, "learning_rate": 5.9224e-06, "loss": 0.021, "step": 20389 }, { "epoch": 1.0870608306232339, "grad_norm": 1.3333659172058105, "learning_rate": 5.9222e-06, "loss": 0.0107, "step": 20390 }, { "epoch": 1.0871141440528869, "grad_norm": 2.5672686100006104, "learning_rate": 5.922e-06, "loss": 0.0309, "step": 20391 }, { "epoch": 1.0871674574825398, "grad_norm": 2.718306064605713, "learning_rate": 5.921800000000001e-06, "loss": 0.0456, "step": 20392 }, { "epoch": 1.0872207709121928, "grad_norm": 3.807405471801758, "learning_rate": 5.921600000000001e-06, "loss": 0.0425, "step": 20393 }, { "epoch": 1.0872740843418458, "grad_norm": 1.2879753112792969, "learning_rate": 5.921400000000001e-06, "loss": 0.0221, "step": 20394 }, { "epoch": 1.0873273977714986, "grad_norm": 1.6696661710739136, "learning_rate": 5.9212e-06, "loss": 0.0211, "step": 20395 }, { "epoch": 1.0873807112011515, "grad_norm": 1.5285848379135132, "learning_rate": 5.921e-06, "loss": 0.021, "step": 20396 }, { "epoch": 1.0874340246308045, "grad_norm": 1.3340147733688354, "learning_rate": 5.920800000000001e-06, "loss": 0.0127, "step": 20397 }, { "epoch": 1.0874873380604575, "grad_norm": 2.014932632446289, "learning_rate": 5.9206000000000005e-06, "loss": 0.0203, "step": 20398 }, { "epoch": 1.0875406514901103, "grad_norm": 1.7478185892105103, "learning_rate": 5.920400000000001e-06, "loss": 0.0324, "step": 20399 }, { "epoch": 1.0875939649197632, "grad_norm": 1.6528300046920776, "learning_rate": 5.9202000000000005e-06, "loss": 0.0159, "step": 20400 }, { "epoch": 1.0876472783494162, "grad_norm": 2.2622735500335693, "learning_rate": 5.92e-06, "loss": 0.0256, "step": 20401 }, { "epoch": 1.0877005917790692, "grad_norm": 1.0705703496932983, "learning_rate": 5.9198e-06, "loss": 0.0084, "step": 20402 }, { "epoch": 1.0877539052087222, "grad_norm": 5.655428886413574, "learning_rate": 5.919600000000001e-06, "loss": 0.031, "step": 20403 }, { "epoch": 1.087807218638375, "grad_norm": 1.6101902723312378, "learning_rate": 5.9194000000000004e-06, "loss": 0.0371, "step": 20404 }, { "epoch": 1.087860532068028, "grad_norm": 1.509993553161621, "learning_rate": 5.919200000000001e-06, "loss": 0.0187, "step": 20405 }, { "epoch": 1.0879138454976809, "grad_norm": 2.20225191116333, "learning_rate": 5.919e-06, "loss": 0.0183, "step": 20406 }, { "epoch": 1.0879671589273339, "grad_norm": 2.2506542205810547, "learning_rate": 5.9188e-06, "loss": 0.039, "step": 20407 }, { "epoch": 1.0880204723569866, "grad_norm": 2.386188507080078, "learning_rate": 5.918600000000001e-06, "loss": 0.0442, "step": 20408 }, { "epoch": 1.0880737857866396, "grad_norm": 3.5603370666503906, "learning_rate": 5.918400000000001e-06, "loss": 0.0311, "step": 20409 }, { "epoch": 1.0881270992162926, "grad_norm": 4.216343402862549, "learning_rate": 5.9182e-06, "loss": 0.0155, "step": 20410 }, { "epoch": 1.0881804126459456, "grad_norm": 1.239581823348999, "learning_rate": 5.918000000000001e-06, "loss": 0.0209, "step": 20411 }, { "epoch": 1.0882337260755985, "grad_norm": 1.7417120933532715, "learning_rate": 5.9178e-06, "loss": 0.0293, "step": 20412 }, { "epoch": 1.0882870395052513, "grad_norm": 2.934037208557129, "learning_rate": 5.9176e-06, "loss": 0.0422, "step": 20413 }, { "epoch": 1.0883403529349043, "grad_norm": 1.1506489515304565, "learning_rate": 5.917400000000001e-06, "loss": 0.0111, "step": 20414 }, { "epoch": 1.0883936663645573, "grad_norm": 0.9693875312805176, "learning_rate": 5.917200000000001e-06, "loss": 0.0105, "step": 20415 }, { "epoch": 1.0884469797942102, "grad_norm": 4.1065993309021, "learning_rate": 5.917e-06, "loss": 0.0328, "step": 20416 }, { "epoch": 1.088500293223863, "grad_norm": 2.139496326446533, "learning_rate": 5.916800000000001e-06, "loss": 0.0126, "step": 20417 }, { "epoch": 1.088553606653516, "grad_norm": 2.2694921493530273, "learning_rate": 5.9166e-06, "loss": 0.034, "step": 20418 }, { "epoch": 1.088606920083169, "grad_norm": 2.707937479019165, "learning_rate": 5.916400000000001e-06, "loss": 0.0802, "step": 20419 }, { "epoch": 1.088660233512822, "grad_norm": 2.285195827484131, "learning_rate": 5.916200000000001e-06, "loss": 0.0245, "step": 20420 }, { "epoch": 1.088713546942475, "grad_norm": 1.150938868522644, "learning_rate": 5.916000000000001e-06, "loss": 0.0214, "step": 20421 }, { "epoch": 1.0887668603721277, "grad_norm": 2.339167833328247, "learning_rate": 5.9158e-06, "loss": 0.0365, "step": 20422 }, { "epoch": 1.0888201738017806, "grad_norm": 1.5235987901687622, "learning_rate": 5.915600000000001e-06, "loss": 0.0208, "step": 20423 }, { "epoch": 1.0888734872314336, "grad_norm": 1.638093113899231, "learning_rate": 5.9154e-06, "loss": 0.0307, "step": 20424 }, { "epoch": 1.0889268006610866, "grad_norm": 1.675330638885498, "learning_rate": 5.9152000000000006e-06, "loss": 0.0266, "step": 20425 }, { "epoch": 1.0889801140907394, "grad_norm": 1.743837594985962, "learning_rate": 5.915000000000001e-06, "loss": 0.0276, "step": 20426 }, { "epoch": 1.0890334275203923, "grad_norm": 0.8198032975196838, "learning_rate": 5.9148000000000005e-06, "loss": 0.0075, "step": 20427 }, { "epoch": 1.0890867409500453, "grad_norm": 1.1700913906097412, "learning_rate": 5.9146e-06, "loss": 0.0239, "step": 20428 }, { "epoch": 1.0891400543796983, "grad_norm": 2.1049540042877197, "learning_rate": 5.9144000000000005e-06, "loss": 0.0177, "step": 20429 }, { "epoch": 1.0891933678093513, "grad_norm": 1.7881611585617065, "learning_rate": 5.9142e-06, "loss": 0.0423, "step": 20430 }, { "epoch": 1.089246681239004, "grad_norm": 1.6749192476272583, "learning_rate": 5.9140000000000005e-06, "loss": 0.0411, "step": 20431 }, { "epoch": 1.089299994668657, "grad_norm": 2.001610279083252, "learning_rate": 5.913800000000001e-06, "loss": 0.0168, "step": 20432 }, { "epoch": 1.08935330809831, "grad_norm": 1.6169723272323608, "learning_rate": 5.9136000000000004e-06, "loss": 0.0492, "step": 20433 }, { "epoch": 1.089406621527963, "grad_norm": 3.048013210296631, "learning_rate": 5.9134e-06, "loss": 0.0352, "step": 20434 }, { "epoch": 1.0894599349576157, "grad_norm": 0.6729569435119629, "learning_rate": 5.9132e-06, "loss": 0.0096, "step": 20435 }, { "epoch": 1.0895132483872687, "grad_norm": 1.6787757873535156, "learning_rate": 5.913000000000001e-06, "loss": 0.0367, "step": 20436 }, { "epoch": 1.0895665618169217, "grad_norm": 1.4411015510559082, "learning_rate": 5.9128e-06, "loss": 0.0351, "step": 20437 }, { "epoch": 1.0896198752465747, "grad_norm": 1.1591370105743408, "learning_rate": 5.912600000000001e-06, "loss": 0.0134, "step": 20438 }, { "epoch": 1.0896731886762276, "grad_norm": 2.200193405151367, "learning_rate": 5.9124e-06, "loss": 0.0291, "step": 20439 }, { "epoch": 1.0897265021058804, "grad_norm": 2.6583054065704346, "learning_rate": 5.9122e-06, "loss": 0.0298, "step": 20440 }, { "epoch": 1.0897798155355334, "grad_norm": 1.3052140474319458, "learning_rate": 5.912e-06, "loss": 0.0117, "step": 20441 }, { "epoch": 1.0898331289651864, "grad_norm": 1.1727828979492188, "learning_rate": 5.911800000000001e-06, "loss": 0.0139, "step": 20442 }, { "epoch": 1.0898864423948393, "grad_norm": 1.330634593963623, "learning_rate": 5.9116e-06, "loss": 0.0193, "step": 20443 }, { "epoch": 1.089939755824492, "grad_norm": 1.335679292678833, "learning_rate": 5.911400000000001e-06, "loss": 0.0187, "step": 20444 }, { "epoch": 1.089993069254145, "grad_norm": 1.1937464475631714, "learning_rate": 5.9112e-06, "loss": 0.0071, "step": 20445 }, { "epoch": 1.090046382683798, "grad_norm": 2.2731850147247314, "learning_rate": 5.911e-06, "loss": 0.0229, "step": 20446 }, { "epoch": 1.090099696113451, "grad_norm": 1.6559014320373535, "learning_rate": 5.910800000000001e-06, "loss": 0.046, "step": 20447 }, { "epoch": 1.090153009543104, "grad_norm": 1.7367826700210571, "learning_rate": 5.910600000000001e-06, "loss": 0.0216, "step": 20448 }, { "epoch": 1.0902063229727568, "grad_norm": 1.869602084159851, "learning_rate": 5.9104e-06, "loss": 0.0447, "step": 20449 }, { "epoch": 1.0902596364024097, "grad_norm": 1.078890323638916, "learning_rate": 5.910200000000001e-06, "loss": 0.0108, "step": 20450 }, { "epoch": 1.0903129498320627, "grad_norm": 2.888558864593506, "learning_rate": 5.91e-06, "loss": 0.019, "step": 20451 }, { "epoch": 1.0903662632617157, "grad_norm": 1.528655767440796, "learning_rate": 5.9098e-06, "loss": 0.0273, "step": 20452 }, { "epoch": 1.0904195766913685, "grad_norm": 1.7801597118377686, "learning_rate": 5.909600000000001e-06, "loss": 0.0446, "step": 20453 }, { "epoch": 1.0904728901210214, "grad_norm": 4.28579568862915, "learning_rate": 5.9094000000000005e-06, "loss": 0.0287, "step": 20454 }, { "epoch": 1.0905262035506744, "grad_norm": 1.6081340312957764, "learning_rate": 5.9092e-06, "loss": 0.0143, "step": 20455 }, { "epoch": 1.0905795169803274, "grad_norm": 2.274115800857544, "learning_rate": 5.9090000000000005e-06, "loss": 0.0503, "step": 20456 }, { "epoch": 1.0906328304099802, "grad_norm": 1.4286847114562988, "learning_rate": 5.9088e-06, "loss": 0.0176, "step": 20457 }, { "epoch": 1.0906861438396331, "grad_norm": 3.3261661529541016, "learning_rate": 5.908600000000001e-06, "loss": 0.0179, "step": 20458 }, { "epoch": 1.0907394572692861, "grad_norm": 2.171685218811035, "learning_rate": 5.908400000000001e-06, "loss": 0.0208, "step": 20459 }, { "epoch": 1.090792770698939, "grad_norm": 2.3367722034454346, "learning_rate": 5.9082000000000005e-06, "loss": 0.0235, "step": 20460 }, { "epoch": 1.090846084128592, "grad_norm": 1.7071911096572876, "learning_rate": 5.908e-06, "loss": 0.0268, "step": 20461 }, { "epoch": 1.0908993975582448, "grad_norm": 1.0844926834106445, "learning_rate": 5.9078000000000004e-06, "loss": 0.0058, "step": 20462 }, { "epoch": 1.0909527109878978, "grad_norm": 1.3780479431152344, "learning_rate": 5.9076e-06, "loss": 0.012, "step": 20463 }, { "epoch": 1.0910060244175508, "grad_norm": 1.7159979343414307, "learning_rate": 5.907400000000001e-06, "loss": 0.0178, "step": 20464 }, { "epoch": 1.0910593378472038, "grad_norm": 1.3035069704055786, "learning_rate": 5.907200000000001e-06, "loss": 0.041, "step": 20465 }, { "epoch": 1.0911126512768567, "grad_norm": 1.791907787322998, "learning_rate": 5.907e-06, "loss": 0.0135, "step": 20466 }, { "epoch": 1.0911659647065095, "grad_norm": 2.173668146133423, "learning_rate": 5.9068e-06, "loss": 0.0077, "step": 20467 }, { "epoch": 1.0912192781361625, "grad_norm": 1.790358543395996, "learning_rate": 5.9066e-06, "loss": 0.0189, "step": 20468 }, { "epoch": 1.0912725915658155, "grad_norm": 0.9370220303535461, "learning_rate": 5.906400000000001e-06, "loss": 0.007, "step": 20469 }, { "epoch": 1.0913259049954684, "grad_norm": 0.7859583497047424, "learning_rate": 5.906200000000001e-06, "loss": 0.0151, "step": 20470 }, { "epoch": 1.0913792184251212, "grad_norm": 0.9720405340194702, "learning_rate": 5.906000000000001e-06, "loss": 0.0207, "step": 20471 }, { "epoch": 1.0914325318547742, "grad_norm": 1.3943781852722168, "learning_rate": 5.9058e-06, "loss": 0.0259, "step": 20472 }, { "epoch": 1.0914858452844272, "grad_norm": 1.8982000350952148, "learning_rate": 5.9056e-06, "loss": 0.0277, "step": 20473 }, { "epoch": 1.0915391587140801, "grad_norm": 1.4265241622924805, "learning_rate": 5.9054e-06, "loss": 0.0321, "step": 20474 }, { "epoch": 1.091592472143733, "grad_norm": 0.6343022584915161, "learning_rate": 5.905200000000001e-06, "loss": 0.0062, "step": 20475 }, { "epoch": 1.0916457855733859, "grad_norm": 1.98098886013031, "learning_rate": 5.905000000000001e-06, "loss": 0.0246, "step": 20476 }, { "epoch": 1.0916990990030389, "grad_norm": 1.344572901725769, "learning_rate": 5.904800000000001e-06, "loss": 0.0177, "step": 20477 }, { "epoch": 1.0917524124326918, "grad_norm": 1.6855409145355225, "learning_rate": 5.9046e-06, "loss": 0.0131, "step": 20478 }, { "epoch": 1.0918057258623448, "grad_norm": 2.2507882118225098, "learning_rate": 5.9044e-06, "loss": 0.0137, "step": 20479 }, { "epoch": 1.0918590392919976, "grad_norm": 5.658585548400879, "learning_rate": 5.9042e-06, "loss": 0.0606, "step": 20480 }, { "epoch": 1.0919123527216505, "grad_norm": 1.8516594171524048, "learning_rate": 5.9040000000000006e-06, "loss": 0.0221, "step": 20481 }, { "epoch": 1.0919656661513035, "grad_norm": 2.6642303466796875, "learning_rate": 5.903800000000001e-06, "loss": 0.0479, "step": 20482 }, { "epoch": 1.0920189795809565, "grad_norm": 2.5810816287994385, "learning_rate": 5.9036000000000005e-06, "loss": 0.068, "step": 20483 }, { "epoch": 1.0920722930106095, "grad_norm": 2.5572140216827393, "learning_rate": 5.9034e-06, "loss": 0.0421, "step": 20484 }, { "epoch": 1.0921256064402622, "grad_norm": 2.974496603012085, "learning_rate": 5.9032e-06, "loss": 0.0792, "step": 20485 }, { "epoch": 1.0921789198699152, "grad_norm": 2.274674415588379, "learning_rate": 5.903000000000001e-06, "loss": 0.0351, "step": 20486 }, { "epoch": 1.0922322332995682, "grad_norm": 2.553321123123169, "learning_rate": 5.9028000000000005e-06, "loss": 0.0419, "step": 20487 }, { "epoch": 1.0922855467292212, "grad_norm": 1.851381540298462, "learning_rate": 5.902600000000001e-06, "loss": 0.0158, "step": 20488 }, { "epoch": 1.092338860158874, "grad_norm": 2.8766252994537354, "learning_rate": 5.9024000000000004e-06, "loss": 0.0283, "step": 20489 }, { "epoch": 1.092392173588527, "grad_norm": 3.035710573196411, "learning_rate": 5.9022e-06, "loss": 0.0213, "step": 20490 }, { "epoch": 1.09244548701818, "grad_norm": 3.1369259357452393, "learning_rate": 5.9019999999999996e-06, "loss": 0.032, "step": 20491 }, { "epoch": 1.0924988004478329, "grad_norm": 2.6389801502227783, "learning_rate": 5.901800000000001e-06, "loss": 0.0259, "step": 20492 }, { "epoch": 1.0925521138774856, "grad_norm": 3.195350408554077, "learning_rate": 5.9016e-06, "loss": 0.0235, "step": 20493 }, { "epoch": 1.0926054273071386, "grad_norm": 1.3931477069854736, "learning_rate": 5.901400000000001e-06, "loss": 0.0263, "step": 20494 }, { "epoch": 1.0926587407367916, "grad_norm": 1.7243388891220093, "learning_rate": 5.9012e-06, "loss": 0.0238, "step": 20495 }, { "epoch": 1.0927120541664446, "grad_norm": 1.9211112260818481, "learning_rate": 5.901e-06, "loss": 0.0358, "step": 20496 }, { "epoch": 1.0927653675960975, "grad_norm": 3.5779972076416016, "learning_rate": 5.900800000000001e-06, "loss": 0.0635, "step": 20497 }, { "epoch": 1.0928186810257503, "grad_norm": 1.5982692241668701, "learning_rate": 5.900600000000001e-06, "loss": 0.0226, "step": 20498 }, { "epoch": 1.0928719944554033, "grad_norm": 2.9872655868530273, "learning_rate": 5.9004e-06, "loss": 0.0181, "step": 20499 }, { "epoch": 1.0929253078850563, "grad_norm": 3.282214879989624, "learning_rate": 5.900200000000001e-06, "loss": 0.0265, "step": 20500 }, { "epoch": 1.0929786213147092, "grad_norm": 1.6472702026367188, "learning_rate": 5.9e-06, "loss": 0.0141, "step": 20501 }, { "epoch": 1.0930319347443622, "grad_norm": 1.973130464553833, "learning_rate": 5.8998e-06, "loss": 0.0261, "step": 20502 }, { "epoch": 1.093085248174015, "grad_norm": 2.5855517387390137, "learning_rate": 5.899600000000001e-06, "loss": 0.0383, "step": 20503 }, { "epoch": 1.093138561603668, "grad_norm": 1.8331100940704346, "learning_rate": 5.899400000000001e-06, "loss": 0.0158, "step": 20504 }, { "epoch": 1.093191875033321, "grad_norm": 2.0998120307922363, "learning_rate": 5.8992e-06, "loss": 0.0169, "step": 20505 }, { "epoch": 1.093245188462974, "grad_norm": 1.9206812381744385, "learning_rate": 5.899000000000001e-06, "loss": 0.0268, "step": 20506 }, { "epoch": 1.0932985018926267, "grad_norm": 1.936711072921753, "learning_rate": 5.8988e-06, "loss": 0.0276, "step": 20507 }, { "epoch": 1.0933518153222797, "grad_norm": 0.7364866733551025, "learning_rate": 5.898600000000001e-06, "loss": 0.0075, "step": 20508 }, { "epoch": 1.0934051287519326, "grad_norm": 2.2762069702148438, "learning_rate": 5.898400000000001e-06, "loss": 0.0408, "step": 20509 }, { "epoch": 1.0934584421815856, "grad_norm": 2.0257277488708496, "learning_rate": 5.8982000000000006e-06, "loss": 0.0429, "step": 20510 }, { "epoch": 1.0935117556112384, "grad_norm": 1.0158871412277222, "learning_rate": 5.898e-06, "loss": 0.0167, "step": 20511 }, { "epoch": 1.0935650690408913, "grad_norm": 1.6761080026626587, "learning_rate": 5.8978000000000005e-06, "loss": 0.0475, "step": 20512 }, { "epoch": 1.0936183824705443, "grad_norm": 0.6745355725288391, "learning_rate": 5.8976e-06, "loss": 0.009, "step": 20513 }, { "epoch": 1.0936716959001973, "grad_norm": 1.7197401523590088, "learning_rate": 5.8974000000000005e-06, "loss": 0.0417, "step": 20514 }, { "epoch": 1.0937250093298503, "grad_norm": 0.6626800298690796, "learning_rate": 5.897200000000001e-06, "loss": 0.0081, "step": 20515 }, { "epoch": 1.093778322759503, "grad_norm": 2.4998481273651123, "learning_rate": 5.8970000000000005e-06, "loss": 0.041, "step": 20516 }, { "epoch": 1.093831636189156, "grad_norm": 1.8317053318023682, "learning_rate": 5.8968e-06, "loss": 0.0142, "step": 20517 }, { "epoch": 1.093884949618809, "grad_norm": 3.069275140762329, "learning_rate": 5.8966000000000004e-06, "loss": 0.016, "step": 20518 }, { "epoch": 1.093938263048462, "grad_norm": 2.7612838745117188, "learning_rate": 5.896400000000001e-06, "loss": 0.0284, "step": 20519 }, { "epoch": 1.0939915764781147, "grad_norm": 2.866488218307495, "learning_rate": 5.8962e-06, "loss": 0.075, "step": 20520 }, { "epoch": 1.0940448899077677, "grad_norm": 0.7724349498748779, "learning_rate": 5.896000000000001e-06, "loss": 0.0088, "step": 20521 }, { "epoch": 1.0940982033374207, "grad_norm": 1.0962883234024048, "learning_rate": 5.8958e-06, "loss": 0.0191, "step": 20522 }, { "epoch": 1.0941515167670737, "grad_norm": 2.08540415763855, "learning_rate": 5.8956e-06, "loss": 0.0081, "step": 20523 }, { "epoch": 1.0942048301967267, "grad_norm": 0.6510523557662964, "learning_rate": 5.8954e-06, "loss": 0.0072, "step": 20524 }, { "epoch": 1.0942581436263794, "grad_norm": 3.3964247703552246, "learning_rate": 5.895200000000001e-06, "loss": 0.0361, "step": 20525 }, { "epoch": 1.0943114570560324, "grad_norm": 0.847763180732727, "learning_rate": 5.895e-06, "loss": 0.017, "step": 20526 }, { "epoch": 1.0943647704856854, "grad_norm": 2.135352849960327, "learning_rate": 5.894800000000001e-06, "loss": 0.0652, "step": 20527 }, { "epoch": 1.0944180839153383, "grad_norm": 2.396085262298584, "learning_rate": 5.8946e-06, "loss": 0.0104, "step": 20528 }, { "epoch": 1.094471397344991, "grad_norm": 1.2366547584533691, "learning_rate": 5.8944e-06, "loss": 0.0147, "step": 20529 }, { "epoch": 1.094524710774644, "grad_norm": 3.309875011444092, "learning_rate": 5.894200000000001e-06, "loss": 0.05, "step": 20530 }, { "epoch": 1.094578024204297, "grad_norm": 1.8307768106460571, "learning_rate": 5.894000000000001e-06, "loss": 0.0473, "step": 20531 }, { "epoch": 1.09463133763395, "grad_norm": 1.0010011196136475, "learning_rate": 5.8938e-06, "loss": 0.0131, "step": 20532 }, { "epoch": 1.094684651063603, "grad_norm": 1.4272838830947876, "learning_rate": 5.893600000000001e-06, "loss": 0.0324, "step": 20533 }, { "epoch": 1.0947379644932558, "grad_norm": 1.037253737449646, "learning_rate": 5.8934e-06, "loss": 0.0106, "step": 20534 }, { "epoch": 1.0947912779229088, "grad_norm": 1.4940706491470337, "learning_rate": 5.8932e-06, "loss": 0.0129, "step": 20535 }, { "epoch": 1.0948445913525617, "grad_norm": 0.6857571005821228, "learning_rate": 5.893000000000001e-06, "loss": 0.0073, "step": 20536 }, { "epoch": 1.0948979047822147, "grad_norm": 1.2378515005111694, "learning_rate": 5.8928000000000006e-06, "loss": 0.0146, "step": 20537 }, { "epoch": 1.0949512182118675, "grad_norm": 1.5143572092056274, "learning_rate": 5.8926e-06, "loss": 0.0195, "step": 20538 }, { "epoch": 1.0950045316415205, "grad_norm": 2.1552586555480957, "learning_rate": 5.8924000000000005e-06, "loss": 0.0355, "step": 20539 }, { "epoch": 1.0950578450711734, "grad_norm": 2.2140324115753174, "learning_rate": 5.8922e-06, "loss": 0.0311, "step": 20540 }, { "epoch": 1.0951111585008264, "grad_norm": 1.7805101871490479, "learning_rate": 5.892e-06, "loss": 0.0289, "step": 20541 }, { "epoch": 1.0951644719304794, "grad_norm": 1.2931069135665894, "learning_rate": 5.891800000000001e-06, "loss": 0.0157, "step": 20542 }, { "epoch": 1.0952177853601321, "grad_norm": 2.475973606109619, "learning_rate": 5.8916000000000005e-06, "loss": 0.0388, "step": 20543 }, { "epoch": 1.0952710987897851, "grad_norm": 2.035694122314453, "learning_rate": 5.8914e-06, "loss": 0.0332, "step": 20544 }, { "epoch": 1.095324412219438, "grad_norm": 2.7414402961730957, "learning_rate": 5.8912000000000004e-06, "loss": 0.0304, "step": 20545 }, { "epoch": 1.095377725649091, "grad_norm": 3.2888448238372803, "learning_rate": 5.891e-06, "loss": 0.05, "step": 20546 }, { "epoch": 1.0954310390787438, "grad_norm": 1.7129443883895874, "learning_rate": 5.890800000000001e-06, "loss": 0.0191, "step": 20547 }, { "epoch": 1.0954843525083968, "grad_norm": 1.9111279249191284, "learning_rate": 5.890600000000001e-06, "loss": 0.0245, "step": 20548 }, { "epoch": 1.0955376659380498, "grad_norm": 8.185490608215332, "learning_rate": 5.8904e-06, "loss": 0.0529, "step": 20549 }, { "epoch": 1.0955909793677028, "grad_norm": 2.006516218185425, "learning_rate": 5.8902e-06, "loss": 0.0305, "step": 20550 }, { "epoch": 1.0956442927973558, "grad_norm": 5.066504955291748, "learning_rate": 5.89e-06, "loss": 0.0257, "step": 20551 }, { "epoch": 1.0956976062270085, "grad_norm": 1.1728291511535645, "learning_rate": 5.8898e-06, "loss": 0.0067, "step": 20552 }, { "epoch": 1.0957509196566615, "grad_norm": 1.3791424036026, "learning_rate": 5.889600000000001e-06, "loss": 0.0175, "step": 20553 }, { "epoch": 1.0958042330863145, "grad_norm": 1.5235912799835205, "learning_rate": 5.889400000000001e-06, "loss": 0.0285, "step": 20554 }, { "epoch": 1.0958575465159675, "grad_norm": 1.3247976303100586, "learning_rate": 5.8892e-06, "loss": 0.0104, "step": 20555 }, { "epoch": 1.0959108599456202, "grad_norm": 2.154770851135254, "learning_rate": 5.889e-06, "loss": 0.0167, "step": 20556 }, { "epoch": 1.0959641733752732, "grad_norm": 1.7045478820800781, "learning_rate": 5.8888e-06, "loss": 0.0155, "step": 20557 }, { "epoch": 1.0960174868049262, "grad_norm": 1.2715704441070557, "learning_rate": 5.888600000000001e-06, "loss": 0.0151, "step": 20558 }, { "epoch": 1.0960708002345791, "grad_norm": 1.8839607238769531, "learning_rate": 5.888400000000001e-06, "loss": 0.0226, "step": 20559 }, { "epoch": 1.096124113664232, "grad_norm": 1.5152223110198975, "learning_rate": 5.888200000000001e-06, "loss": 0.0303, "step": 20560 }, { "epoch": 1.0961774270938849, "grad_norm": 2.254939556121826, "learning_rate": 5.888e-06, "loss": 0.0277, "step": 20561 }, { "epoch": 1.0962307405235379, "grad_norm": 2.5999629497528076, "learning_rate": 5.8878e-06, "loss": 0.0349, "step": 20562 }, { "epoch": 1.0962840539531908, "grad_norm": 1.9801758527755737, "learning_rate": 5.8876e-06, "loss": 0.0275, "step": 20563 }, { "epoch": 1.0963373673828438, "grad_norm": 3.783740282058716, "learning_rate": 5.887400000000001e-06, "loss": 0.0298, "step": 20564 }, { "epoch": 1.0963906808124966, "grad_norm": 1.9750134944915771, "learning_rate": 5.887200000000001e-06, "loss": 0.0326, "step": 20565 }, { "epoch": 1.0964439942421496, "grad_norm": 0.9320794939994812, "learning_rate": 5.8870000000000006e-06, "loss": 0.0187, "step": 20566 }, { "epoch": 1.0964973076718025, "grad_norm": 1.5346778631210327, "learning_rate": 5.8868e-06, "loss": 0.0182, "step": 20567 }, { "epoch": 1.0965506211014555, "grad_norm": 2.583632707595825, "learning_rate": 5.8866e-06, "loss": 0.02, "step": 20568 }, { "epoch": 1.0966039345311085, "grad_norm": 0.9220672249794006, "learning_rate": 5.886400000000001e-06, "loss": 0.0212, "step": 20569 }, { "epoch": 1.0966572479607613, "grad_norm": 0.5682949423789978, "learning_rate": 5.8862000000000005e-06, "loss": 0.0044, "step": 20570 }, { "epoch": 1.0967105613904142, "grad_norm": 2.046783924102783, "learning_rate": 5.886000000000001e-06, "loss": 0.0348, "step": 20571 }, { "epoch": 1.0967638748200672, "grad_norm": 1.9455446004867554, "learning_rate": 5.8858000000000005e-06, "loss": 0.0222, "step": 20572 }, { "epoch": 1.0968171882497202, "grad_norm": 1.8624613285064697, "learning_rate": 5.8856e-06, "loss": 0.0269, "step": 20573 }, { "epoch": 1.096870501679373, "grad_norm": 0.895370364189148, "learning_rate": 5.8854e-06, "loss": 0.0163, "step": 20574 }, { "epoch": 1.096923815109026, "grad_norm": 2.493008852005005, "learning_rate": 5.885200000000001e-06, "loss": 0.036, "step": 20575 }, { "epoch": 1.096977128538679, "grad_norm": 1.633831262588501, "learning_rate": 5.885e-06, "loss": 0.0118, "step": 20576 }, { "epoch": 1.0970304419683319, "grad_norm": 1.7696038484573364, "learning_rate": 5.884800000000001e-06, "loss": 0.0156, "step": 20577 }, { "epoch": 1.0970837553979846, "grad_norm": 0.8479782938957214, "learning_rate": 5.8846e-06, "loss": 0.0072, "step": 20578 }, { "epoch": 1.0971370688276376, "grad_norm": 1.6804319620132446, "learning_rate": 5.8844e-06, "loss": 0.0314, "step": 20579 }, { "epoch": 1.0971903822572906, "grad_norm": 2.4223673343658447, "learning_rate": 5.884200000000001e-06, "loss": 0.0321, "step": 20580 }, { "epoch": 1.0972436956869436, "grad_norm": 1.4446439743041992, "learning_rate": 5.884000000000001e-06, "loss": 0.0098, "step": 20581 }, { "epoch": 1.0972970091165966, "grad_norm": 1.0583566427230835, "learning_rate": 5.8838e-06, "loss": 0.0116, "step": 20582 }, { "epoch": 1.0973503225462493, "grad_norm": 1.7515469789505005, "learning_rate": 5.883600000000001e-06, "loss": 0.0505, "step": 20583 }, { "epoch": 1.0974036359759023, "grad_norm": 0.9622714519500732, "learning_rate": 5.8834e-06, "loss": 0.0063, "step": 20584 }, { "epoch": 1.0974569494055553, "grad_norm": 2.4200010299682617, "learning_rate": 5.8832e-06, "loss": 0.0311, "step": 20585 }, { "epoch": 1.0975102628352083, "grad_norm": 1.539757251739502, "learning_rate": 5.883000000000001e-06, "loss": 0.0398, "step": 20586 }, { "epoch": 1.0975635762648612, "grad_norm": 1.6540778875350952, "learning_rate": 5.882800000000001e-06, "loss": 0.0209, "step": 20587 }, { "epoch": 1.097616889694514, "grad_norm": 3.281991958618164, "learning_rate": 5.8826e-06, "loss": 0.0279, "step": 20588 }, { "epoch": 1.097670203124167, "grad_norm": 1.4044934511184692, "learning_rate": 5.882400000000001e-06, "loss": 0.0085, "step": 20589 }, { "epoch": 1.09772351655382, "grad_norm": 1.1867538690567017, "learning_rate": 5.8822e-06, "loss": 0.0109, "step": 20590 }, { "epoch": 1.097776829983473, "grad_norm": 3.216115713119507, "learning_rate": 5.882e-06, "loss": 0.0327, "step": 20591 }, { "epoch": 1.0978301434131257, "grad_norm": 2.840301513671875, "learning_rate": 5.881800000000001e-06, "loss": 0.0233, "step": 20592 }, { "epoch": 1.0978834568427787, "grad_norm": 1.2625943422317505, "learning_rate": 5.8816000000000006e-06, "loss": 0.0101, "step": 20593 }, { "epoch": 1.0979367702724316, "grad_norm": 1.63076913356781, "learning_rate": 5.8814e-06, "loss": 0.0169, "step": 20594 }, { "epoch": 1.0979900837020846, "grad_norm": 1.7457302808761597, "learning_rate": 5.8812000000000005e-06, "loss": 0.0171, "step": 20595 }, { "epoch": 1.0980433971317374, "grad_norm": 1.4672964811325073, "learning_rate": 5.881e-06, "loss": 0.0205, "step": 20596 }, { "epoch": 1.0980967105613904, "grad_norm": 3.2501769065856934, "learning_rate": 5.880800000000001e-06, "loss": 0.0269, "step": 20597 }, { "epoch": 1.0981500239910433, "grad_norm": 3.7326927185058594, "learning_rate": 5.880600000000001e-06, "loss": 0.0467, "step": 20598 }, { "epoch": 1.0982033374206963, "grad_norm": 1.7252062559127808, "learning_rate": 5.8804000000000005e-06, "loss": 0.016, "step": 20599 }, { "epoch": 1.0982566508503493, "grad_norm": 2.743717670440674, "learning_rate": 5.8802e-06, "loss": 0.0291, "step": 20600 }, { "epoch": 1.098309964280002, "grad_norm": 1.191290259361267, "learning_rate": 5.8800000000000005e-06, "loss": 0.0107, "step": 20601 }, { "epoch": 1.098363277709655, "grad_norm": 4.364087104797363, "learning_rate": 5.8798e-06, "loss": 0.0826, "step": 20602 }, { "epoch": 1.098416591139308, "grad_norm": 2.235435962677002, "learning_rate": 5.879600000000001e-06, "loss": 0.0479, "step": 20603 }, { "epoch": 1.098469904568961, "grad_norm": 2.034139394760132, "learning_rate": 5.879400000000001e-06, "loss": 0.0331, "step": 20604 }, { "epoch": 1.098523217998614, "grad_norm": 2.9922473430633545, "learning_rate": 5.8792e-06, "loss": 0.0324, "step": 20605 }, { "epoch": 1.0985765314282667, "grad_norm": 1.036461353302002, "learning_rate": 5.879e-06, "loss": 0.0077, "step": 20606 }, { "epoch": 1.0986298448579197, "grad_norm": 4.526946544647217, "learning_rate": 5.8788e-06, "loss": 0.0183, "step": 20607 }, { "epoch": 1.0986831582875727, "grad_norm": 1.9051746129989624, "learning_rate": 5.878600000000001e-06, "loss": 0.0201, "step": 20608 }, { "epoch": 1.0987364717172257, "grad_norm": 1.6221321821212769, "learning_rate": 5.878400000000001e-06, "loss": 0.0246, "step": 20609 }, { "epoch": 1.0987897851468784, "grad_norm": 2.5498557090759277, "learning_rate": 5.878200000000001e-06, "loss": 0.0407, "step": 20610 }, { "epoch": 1.0988430985765314, "grad_norm": 2.637057304382324, "learning_rate": 5.878e-06, "loss": 0.0214, "step": 20611 }, { "epoch": 1.0988964120061844, "grad_norm": 2.379394769668579, "learning_rate": 5.8778e-06, "loss": 0.0453, "step": 20612 }, { "epoch": 1.0989497254358374, "grad_norm": 2.705897092819214, "learning_rate": 5.8776e-06, "loss": 0.0305, "step": 20613 }, { "epoch": 1.0990030388654901, "grad_norm": 3.4051766395568848, "learning_rate": 5.877400000000001e-06, "loss": 0.0354, "step": 20614 }, { "epoch": 1.099056352295143, "grad_norm": 1.157343864440918, "learning_rate": 5.877200000000001e-06, "loss": 0.0135, "step": 20615 }, { "epoch": 1.099109665724796, "grad_norm": 2.5023560523986816, "learning_rate": 5.877000000000001e-06, "loss": 0.0306, "step": 20616 }, { "epoch": 1.099162979154449, "grad_norm": 2.3821158409118652, "learning_rate": 5.8768e-06, "loss": 0.0398, "step": 20617 }, { "epoch": 1.099216292584102, "grad_norm": 1.3166025876998901, "learning_rate": 5.8766e-06, "loss": 0.0136, "step": 20618 }, { "epoch": 1.0992696060137548, "grad_norm": 1.0117452144622803, "learning_rate": 5.876400000000001e-06, "loss": 0.0107, "step": 20619 }, { "epoch": 1.0993229194434078, "grad_norm": 2.391441583633423, "learning_rate": 5.876200000000001e-06, "loss": 0.0338, "step": 20620 }, { "epoch": 1.0993762328730607, "grad_norm": 0.6406394839286804, "learning_rate": 5.876000000000001e-06, "loss": 0.0063, "step": 20621 }, { "epoch": 1.0994295463027137, "grad_norm": 2.73085618019104, "learning_rate": 5.8758000000000006e-06, "loss": 0.0474, "step": 20622 }, { "epoch": 1.0994828597323665, "grad_norm": 4.381968975067139, "learning_rate": 5.8756e-06, "loss": 0.0461, "step": 20623 }, { "epoch": 1.0995361731620195, "grad_norm": 0.7537603974342346, "learning_rate": 5.8754e-06, "loss": 0.0079, "step": 20624 }, { "epoch": 1.0995894865916724, "grad_norm": 0.9854620099067688, "learning_rate": 5.875200000000001e-06, "loss": 0.0108, "step": 20625 }, { "epoch": 1.0996428000213254, "grad_norm": 1.5219275951385498, "learning_rate": 5.8750000000000005e-06, "loss": 0.0333, "step": 20626 }, { "epoch": 1.0996961134509784, "grad_norm": 2.5396995544433594, "learning_rate": 5.874800000000001e-06, "loss": 0.0555, "step": 20627 }, { "epoch": 1.0997494268806312, "grad_norm": 1.9921135902404785, "learning_rate": 5.8746000000000005e-06, "loss": 0.0396, "step": 20628 }, { "epoch": 1.0998027403102841, "grad_norm": 1.8436046838760376, "learning_rate": 5.8744e-06, "loss": 0.0404, "step": 20629 }, { "epoch": 1.0998560537399371, "grad_norm": 1.3927322626113892, "learning_rate": 5.874200000000001e-06, "loss": 0.0145, "step": 20630 }, { "epoch": 1.09990936716959, "grad_norm": 2.555968761444092, "learning_rate": 5.874000000000001e-06, "loss": 0.0297, "step": 20631 }, { "epoch": 1.0999626805992428, "grad_norm": 1.8202199935913086, "learning_rate": 5.8738e-06, "loss": 0.0254, "step": 20632 }, { "epoch": 1.1000159940288958, "grad_norm": 2.2964863777160645, "learning_rate": 5.873600000000001e-06, "loss": 0.0356, "step": 20633 }, { "epoch": 1.1000693074585488, "grad_norm": 2.7533135414123535, "learning_rate": 5.8734e-06, "loss": 0.0315, "step": 20634 }, { "epoch": 1.1001226208882018, "grad_norm": 2.396523952484131, "learning_rate": 5.8732e-06, "loss": 0.0577, "step": 20635 }, { "epoch": 1.1001759343178548, "grad_norm": 2.7860395908355713, "learning_rate": 5.873000000000001e-06, "loss": 0.0572, "step": 20636 }, { "epoch": 1.1002292477475075, "grad_norm": 2.0244557857513428, "learning_rate": 5.872800000000001e-06, "loss": 0.0315, "step": 20637 }, { "epoch": 1.1002825611771605, "grad_norm": 1.4145580530166626, "learning_rate": 5.8726e-06, "loss": 0.0177, "step": 20638 }, { "epoch": 1.1003358746068135, "grad_norm": 1.0549061298370361, "learning_rate": 5.872400000000001e-06, "loss": 0.0149, "step": 20639 }, { "epoch": 1.1003891880364665, "grad_norm": 1.7313421964645386, "learning_rate": 5.8722e-06, "loss": 0.0219, "step": 20640 }, { "epoch": 1.1004425014661192, "grad_norm": 2.0296590328216553, "learning_rate": 5.872000000000001e-06, "loss": 0.0553, "step": 20641 }, { "epoch": 1.1004958148957722, "grad_norm": 2.9463820457458496, "learning_rate": 5.871800000000001e-06, "loss": 0.0361, "step": 20642 }, { "epoch": 1.1005491283254252, "grad_norm": 0.22233986854553223, "learning_rate": 5.871600000000001e-06, "loss": 0.0024, "step": 20643 }, { "epoch": 1.1006024417550782, "grad_norm": 2.10044264793396, "learning_rate": 5.8714e-06, "loss": 0.027, "step": 20644 }, { "epoch": 1.1006557551847311, "grad_norm": 1.7991588115692139, "learning_rate": 5.871200000000001e-06, "loss": 0.0164, "step": 20645 }, { "epoch": 1.100709068614384, "grad_norm": 1.6467339992523193, "learning_rate": 5.871e-06, "loss": 0.0396, "step": 20646 }, { "epoch": 1.1007623820440369, "grad_norm": 0.8541643023490906, "learning_rate": 5.870800000000001e-06, "loss": 0.0103, "step": 20647 }, { "epoch": 1.1008156954736898, "grad_norm": 2.854680061340332, "learning_rate": 5.870600000000001e-06, "loss": 0.041, "step": 20648 }, { "epoch": 1.1008690089033428, "grad_norm": 1.8634982109069824, "learning_rate": 5.870400000000001e-06, "loss": 0.0213, "step": 20649 }, { "epoch": 1.1009223223329956, "grad_norm": 1.2570679187774658, "learning_rate": 5.8702e-06, "loss": 0.0152, "step": 20650 }, { "epoch": 1.1009756357626486, "grad_norm": 1.2377617359161377, "learning_rate": 5.8700000000000005e-06, "loss": 0.0143, "step": 20651 }, { "epoch": 1.1010289491923015, "grad_norm": 1.9913711547851562, "learning_rate": 5.8698e-06, "loss": 0.0312, "step": 20652 }, { "epoch": 1.1010822626219545, "grad_norm": 1.5178934335708618, "learning_rate": 5.8696000000000005e-06, "loss": 0.0101, "step": 20653 }, { "epoch": 1.1011355760516075, "grad_norm": 1.575892448425293, "learning_rate": 5.869400000000001e-06, "loss": 0.0093, "step": 20654 }, { "epoch": 1.1011888894812603, "grad_norm": 0.8212512731552124, "learning_rate": 5.8692000000000005e-06, "loss": 0.008, "step": 20655 }, { "epoch": 1.1012422029109132, "grad_norm": 1.6757831573486328, "learning_rate": 5.869e-06, "loss": 0.0141, "step": 20656 }, { "epoch": 1.1012955163405662, "grad_norm": 2.0195505619049072, "learning_rate": 5.8688000000000005e-06, "loss": 0.0464, "step": 20657 }, { "epoch": 1.1013488297702192, "grad_norm": 2.037583112716675, "learning_rate": 5.868600000000001e-06, "loss": 0.0294, "step": 20658 }, { "epoch": 1.101402143199872, "grad_norm": 1.649700403213501, "learning_rate": 5.8684000000000004e-06, "loss": 0.0234, "step": 20659 }, { "epoch": 1.101455456629525, "grad_norm": 1.2519935369491577, "learning_rate": 5.868200000000001e-06, "loss": 0.0197, "step": 20660 }, { "epoch": 1.101508770059178, "grad_norm": 2.673689126968384, "learning_rate": 5.868e-06, "loss": 0.053, "step": 20661 }, { "epoch": 1.101562083488831, "grad_norm": 0.8690509796142578, "learning_rate": 5.8678e-06, "loss": 0.0051, "step": 20662 }, { "epoch": 1.1016153969184839, "grad_norm": 2.631894111633301, "learning_rate": 5.8676e-06, "loss": 0.0275, "step": 20663 }, { "epoch": 1.1016687103481366, "grad_norm": 4.573931694030762, "learning_rate": 5.867400000000001e-06, "loss": 0.0275, "step": 20664 }, { "epoch": 1.1017220237777896, "grad_norm": 1.050281286239624, "learning_rate": 5.8672e-06, "loss": 0.0062, "step": 20665 }, { "epoch": 1.1017753372074426, "grad_norm": 1.30263090133667, "learning_rate": 5.867000000000001e-06, "loss": 0.0105, "step": 20666 }, { "epoch": 1.1018286506370956, "grad_norm": 1.4644426107406616, "learning_rate": 5.8668e-06, "loss": 0.0208, "step": 20667 }, { "epoch": 1.1018819640667483, "grad_norm": 4.79132080078125, "learning_rate": 5.8666e-06, "loss": 0.022, "step": 20668 }, { "epoch": 1.1019352774964013, "grad_norm": 2.0182125568389893, "learning_rate": 5.866400000000001e-06, "loss": 0.0403, "step": 20669 }, { "epoch": 1.1019885909260543, "grad_norm": 2.5904219150543213, "learning_rate": 5.866200000000001e-06, "loss": 0.0414, "step": 20670 }, { "epoch": 1.1020419043557073, "grad_norm": 1.5458815097808838, "learning_rate": 5.866e-06, "loss": 0.0248, "step": 20671 }, { "epoch": 1.1020952177853602, "grad_norm": 1.2289912700653076, "learning_rate": 5.865800000000001e-06, "loss": 0.0261, "step": 20672 }, { "epoch": 1.102148531215013, "grad_norm": 0.9827192425727844, "learning_rate": 5.8656e-06, "loss": 0.0119, "step": 20673 }, { "epoch": 1.102201844644666, "grad_norm": 0.978264570236206, "learning_rate": 5.8654e-06, "loss": 0.019, "step": 20674 }, { "epoch": 1.102255158074319, "grad_norm": 2.518321990966797, "learning_rate": 5.865200000000001e-06, "loss": 0.0284, "step": 20675 }, { "epoch": 1.102308471503972, "grad_norm": 1.5287591218948364, "learning_rate": 5.865000000000001e-06, "loss": 0.0073, "step": 20676 }, { "epoch": 1.1023617849336247, "grad_norm": 2.1265175342559814, "learning_rate": 5.8648e-06, "loss": 0.0199, "step": 20677 }, { "epoch": 1.1024150983632777, "grad_norm": 1.1425117254257202, "learning_rate": 5.8646000000000006e-06, "loss": 0.0286, "step": 20678 }, { "epoch": 1.1024684117929306, "grad_norm": 1.292769432067871, "learning_rate": 5.8644e-06, "loss": 0.0211, "step": 20679 }, { "epoch": 1.1025217252225836, "grad_norm": 2.2845585346221924, "learning_rate": 5.864200000000001e-06, "loss": 0.0432, "step": 20680 }, { "epoch": 1.1025750386522364, "grad_norm": 3.757288932800293, "learning_rate": 5.864000000000001e-06, "loss": 0.0227, "step": 20681 }, { "epoch": 1.1026283520818894, "grad_norm": 1.7579137086868286, "learning_rate": 5.8638000000000005e-06, "loss": 0.0265, "step": 20682 }, { "epoch": 1.1026816655115423, "grad_norm": 4.228395938873291, "learning_rate": 5.8636e-06, "loss": 0.0266, "step": 20683 }, { "epoch": 1.1027349789411953, "grad_norm": 1.636217713356018, "learning_rate": 5.8634000000000005e-06, "loss": 0.0346, "step": 20684 }, { "epoch": 1.1027882923708483, "grad_norm": 2.0565834045410156, "learning_rate": 5.8632e-06, "loss": 0.0178, "step": 20685 }, { "epoch": 1.102841605800501, "grad_norm": 2.363478899002075, "learning_rate": 5.863000000000001e-06, "loss": 0.0162, "step": 20686 }, { "epoch": 1.102894919230154, "grad_norm": 1.6364566087722778, "learning_rate": 5.862800000000001e-06, "loss": 0.0259, "step": 20687 }, { "epoch": 1.102948232659807, "grad_norm": 0.7112984657287598, "learning_rate": 5.8626e-06, "loss": 0.005, "step": 20688 }, { "epoch": 1.10300154608946, "grad_norm": 0.652543842792511, "learning_rate": 5.8624e-06, "loss": 0.004, "step": 20689 }, { "epoch": 1.103054859519113, "grad_norm": 2.429096221923828, "learning_rate": 5.8622e-06, "loss": 0.0262, "step": 20690 }, { "epoch": 1.1031081729487657, "grad_norm": 2.7677688598632812, "learning_rate": 5.862000000000001e-06, "loss": 0.0144, "step": 20691 }, { "epoch": 1.1031614863784187, "grad_norm": 1.1156336069107056, "learning_rate": 5.861800000000001e-06, "loss": 0.0098, "step": 20692 }, { "epoch": 1.1032147998080717, "grad_norm": 2.154210329055786, "learning_rate": 5.861600000000001e-06, "loss": 0.0788, "step": 20693 }, { "epoch": 1.1032681132377247, "grad_norm": 1.5364501476287842, "learning_rate": 5.8614e-06, "loss": 0.014, "step": 20694 }, { "epoch": 1.1033214266673774, "grad_norm": 2.3310391902923584, "learning_rate": 5.8612e-06, "loss": 0.0631, "step": 20695 }, { "epoch": 1.1033747400970304, "grad_norm": 0.8443261981010437, "learning_rate": 5.861e-06, "loss": 0.007, "step": 20696 }, { "epoch": 1.1034280535266834, "grad_norm": 2.4370439052581787, "learning_rate": 5.860800000000001e-06, "loss": 0.0134, "step": 20697 }, { "epoch": 1.1034813669563364, "grad_norm": 2.5649755001068115, "learning_rate": 5.860600000000001e-06, "loss": 0.0786, "step": 20698 }, { "epoch": 1.1035346803859891, "grad_norm": 1.2747697830200195, "learning_rate": 5.860400000000001e-06, "loss": 0.0191, "step": 20699 }, { "epoch": 1.103587993815642, "grad_norm": 2.4182894229888916, "learning_rate": 5.8602e-06, "loss": 0.0325, "step": 20700 }, { "epoch": 1.103641307245295, "grad_norm": 2.217410087585449, "learning_rate": 5.86e-06, "loss": 0.0263, "step": 20701 }, { "epoch": 1.103694620674948, "grad_norm": 2.2958197593688965, "learning_rate": 5.8598e-06, "loss": 0.0203, "step": 20702 }, { "epoch": 1.103747934104601, "grad_norm": 3.1307735443115234, "learning_rate": 5.859600000000001e-06, "loss": 0.0252, "step": 20703 }, { "epoch": 1.1038012475342538, "grad_norm": 2.4933319091796875, "learning_rate": 5.859400000000001e-06, "loss": 0.0228, "step": 20704 }, { "epoch": 1.1038545609639068, "grad_norm": 1.8436253070831299, "learning_rate": 5.859200000000001e-06, "loss": 0.015, "step": 20705 }, { "epoch": 1.1039078743935598, "grad_norm": 2.987240791320801, "learning_rate": 5.859e-06, "loss": 0.0654, "step": 20706 }, { "epoch": 1.1039611878232127, "grad_norm": 1.5043625831604004, "learning_rate": 5.8588e-06, "loss": 0.0516, "step": 20707 }, { "epoch": 1.1040145012528657, "grad_norm": 1.2238417863845825, "learning_rate": 5.858600000000001e-06, "loss": 0.0223, "step": 20708 }, { "epoch": 1.1040678146825185, "grad_norm": 1.9007619619369507, "learning_rate": 5.8584000000000005e-06, "loss": 0.0282, "step": 20709 }, { "epoch": 1.1041211281121714, "grad_norm": 2.390934705734253, "learning_rate": 5.858200000000001e-06, "loss": 0.0317, "step": 20710 }, { "epoch": 1.1041744415418244, "grad_norm": 1.209753394126892, "learning_rate": 5.8580000000000005e-06, "loss": 0.0097, "step": 20711 }, { "epoch": 1.1042277549714774, "grad_norm": 2.0902791023254395, "learning_rate": 5.8578e-06, "loss": 0.0457, "step": 20712 }, { "epoch": 1.1042810684011302, "grad_norm": 1.3534671068191528, "learning_rate": 5.8576e-06, "loss": 0.0092, "step": 20713 }, { "epoch": 1.1043343818307831, "grad_norm": 0.6949659585952759, "learning_rate": 5.857400000000001e-06, "loss": 0.0051, "step": 20714 }, { "epoch": 1.1043876952604361, "grad_norm": 3.4873857498168945, "learning_rate": 5.8572000000000004e-06, "loss": 0.0704, "step": 20715 }, { "epoch": 1.104441008690089, "grad_norm": 1.4173036813735962, "learning_rate": 5.857000000000001e-06, "loss": 0.0164, "step": 20716 }, { "epoch": 1.1044943221197419, "grad_norm": 1.7963365316390991, "learning_rate": 5.8568e-06, "loss": 0.0185, "step": 20717 }, { "epoch": 1.1045476355493948, "grad_norm": 2.6377758979797363, "learning_rate": 5.8566e-06, "loss": 0.0528, "step": 20718 }, { "epoch": 1.1046009489790478, "grad_norm": 2.575463056564331, "learning_rate": 5.856400000000001e-06, "loss": 0.0505, "step": 20719 }, { "epoch": 1.1046542624087008, "grad_norm": 2.178748846054077, "learning_rate": 5.856200000000001e-06, "loss": 0.0318, "step": 20720 }, { "epoch": 1.1047075758383538, "grad_norm": 2.0361905097961426, "learning_rate": 5.856e-06, "loss": 0.0321, "step": 20721 }, { "epoch": 1.1047608892680065, "grad_norm": 1.4378081560134888, "learning_rate": 5.855800000000001e-06, "loss": 0.0186, "step": 20722 }, { "epoch": 1.1048142026976595, "grad_norm": 1.5938764810562134, "learning_rate": 5.8556e-06, "loss": 0.0302, "step": 20723 }, { "epoch": 1.1048675161273125, "grad_norm": 1.912312388420105, "learning_rate": 5.8554e-06, "loss": 0.026, "step": 20724 }, { "epoch": 1.1049208295569655, "grad_norm": 1.7624820470809937, "learning_rate": 5.855200000000001e-06, "loss": 0.0167, "step": 20725 }, { "epoch": 1.1049741429866184, "grad_norm": 1.4858813285827637, "learning_rate": 5.855000000000001e-06, "loss": 0.0248, "step": 20726 }, { "epoch": 1.1050274564162712, "grad_norm": 1.4479475021362305, "learning_rate": 5.8548e-06, "loss": 0.0287, "step": 20727 }, { "epoch": 1.1050807698459242, "grad_norm": 0.8890868425369263, "learning_rate": 5.854600000000001e-06, "loss": 0.0077, "step": 20728 }, { "epoch": 1.1051340832755772, "grad_norm": 1.2860803604125977, "learning_rate": 5.8544e-06, "loss": 0.029, "step": 20729 }, { "epoch": 1.1051873967052301, "grad_norm": 1.558493733406067, "learning_rate": 5.854200000000001e-06, "loss": 0.037, "step": 20730 }, { "epoch": 1.105240710134883, "grad_norm": 1.420894742012024, "learning_rate": 5.854000000000001e-06, "loss": 0.0255, "step": 20731 }, { "epoch": 1.1052940235645359, "grad_norm": 1.5247247219085693, "learning_rate": 5.853800000000001e-06, "loss": 0.0328, "step": 20732 }, { "epoch": 1.1053473369941889, "grad_norm": 1.941884994506836, "learning_rate": 5.8536e-06, "loss": 0.0502, "step": 20733 }, { "epoch": 1.1054006504238418, "grad_norm": 4.062171936035156, "learning_rate": 5.8534000000000006e-06, "loss": 0.0391, "step": 20734 }, { "epoch": 1.1054539638534946, "grad_norm": 1.104468584060669, "learning_rate": 5.8532e-06, "loss": 0.0263, "step": 20735 }, { "epoch": 1.1055072772831476, "grad_norm": 1.3177669048309326, "learning_rate": 5.8530000000000005e-06, "loss": 0.0169, "step": 20736 }, { "epoch": 1.1055605907128006, "grad_norm": 2.2471201419830322, "learning_rate": 5.852800000000001e-06, "loss": 0.0362, "step": 20737 }, { "epoch": 1.1056139041424535, "grad_norm": 0.9426743388175964, "learning_rate": 5.8526000000000005e-06, "loss": 0.0178, "step": 20738 }, { "epoch": 1.1056672175721065, "grad_norm": 4.487947940826416, "learning_rate": 5.8524e-06, "loss": 0.0861, "step": 20739 }, { "epoch": 1.1057205310017593, "grad_norm": 1.329690933227539, "learning_rate": 5.8522000000000005e-06, "loss": 0.0261, "step": 20740 }, { "epoch": 1.1057738444314122, "grad_norm": 1.3586925268173218, "learning_rate": 5.852000000000001e-06, "loss": 0.0441, "step": 20741 }, { "epoch": 1.1058271578610652, "grad_norm": 1.6399309635162354, "learning_rate": 5.8518000000000005e-06, "loss": 0.0185, "step": 20742 }, { "epoch": 1.1058804712907182, "grad_norm": 1.5828756093978882, "learning_rate": 5.851600000000001e-06, "loss": 0.0345, "step": 20743 }, { "epoch": 1.105933784720371, "grad_norm": 1.0819138288497925, "learning_rate": 5.8514e-06, "loss": 0.018, "step": 20744 }, { "epoch": 1.105987098150024, "grad_norm": 1.9623007774353027, "learning_rate": 5.8512e-06, "loss": 0.0402, "step": 20745 }, { "epoch": 1.106040411579677, "grad_norm": 2.441601276397705, "learning_rate": 5.851e-06, "loss": 0.0352, "step": 20746 }, { "epoch": 1.10609372500933, "grad_norm": 1.278355598449707, "learning_rate": 5.850800000000001e-06, "loss": 0.0279, "step": 20747 }, { "epoch": 1.1061470384389829, "grad_norm": 2.289853572845459, "learning_rate": 5.8506e-06, "loss": 0.04, "step": 20748 }, { "epoch": 1.1062003518686356, "grad_norm": 1.0340867042541504, "learning_rate": 5.850400000000001e-06, "loss": 0.0161, "step": 20749 }, { "epoch": 1.1062536652982886, "grad_norm": 0.8617945909500122, "learning_rate": 5.8502e-06, "loss": 0.0161, "step": 20750 }, { "epoch": 1.1063069787279416, "grad_norm": 1.3456141948699951, "learning_rate": 5.85e-06, "loss": 0.0314, "step": 20751 }, { "epoch": 1.1063602921575946, "grad_norm": 1.128796100616455, "learning_rate": 5.849800000000001e-06, "loss": 0.0115, "step": 20752 }, { "epoch": 1.1064136055872473, "grad_norm": 1.631314754486084, "learning_rate": 5.849600000000001e-06, "loss": 0.0125, "step": 20753 }, { "epoch": 1.1064669190169003, "grad_norm": 1.5909844636917114, "learning_rate": 5.8494e-06, "loss": 0.0289, "step": 20754 }, { "epoch": 1.1065202324465533, "grad_norm": 1.788949728012085, "learning_rate": 5.849200000000001e-06, "loss": 0.0214, "step": 20755 }, { "epoch": 1.1065735458762063, "grad_norm": 1.659662127494812, "learning_rate": 5.849e-06, "loss": 0.0333, "step": 20756 }, { "epoch": 1.1066268593058592, "grad_norm": 0.6371352076530457, "learning_rate": 5.8488e-06, "loss": 0.0118, "step": 20757 }, { "epoch": 1.106680172735512, "grad_norm": 1.281529426574707, "learning_rate": 5.848600000000001e-06, "loss": 0.012, "step": 20758 }, { "epoch": 1.106733486165165, "grad_norm": 2.2682266235351562, "learning_rate": 5.848400000000001e-06, "loss": 0.0357, "step": 20759 }, { "epoch": 1.106786799594818, "grad_norm": 2.046780824661255, "learning_rate": 5.8482e-06, "loss": 0.0415, "step": 20760 }, { "epoch": 1.106840113024471, "grad_norm": 2.633854627609253, "learning_rate": 5.848000000000001e-06, "loss": 0.0344, "step": 20761 }, { "epoch": 1.1068934264541237, "grad_norm": 0.7239677906036377, "learning_rate": 5.8478e-06, "loss": 0.0085, "step": 20762 }, { "epoch": 1.1069467398837767, "grad_norm": 1.927560567855835, "learning_rate": 5.8476e-06, "loss": 0.0238, "step": 20763 }, { "epoch": 1.1070000533134297, "grad_norm": 1.5637435913085938, "learning_rate": 5.847400000000001e-06, "loss": 0.0169, "step": 20764 }, { "epoch": 1.1070533667430826, "grad_norm": 1.1662993431091309, "learning_rate": 5.8472000000000005e-06, "loss": 0.0209, "step": 20765 }, { "epoch": 1.1071066801727356, "grad_norm": 2.2131829261779785, "learning_rate": 5.847e-06, "loss": 0.0263, "step": 20766 }, { "epoch": 1.1071599936023884, "grad_norm": 2.1290762424468994, "learning_rate": 5.8468000000000005e-06, "loss": 0.0511, "step": 20767 }, { "epoch": 1.1072133070320413, "grad_norm": 3.1480846405029297, "learning_rate": 5.8466e-06, "loss": 0.0499, "step": 20768 }, { "epoch": 1.1072666204616943, "grad_norm": 2.9481465816497803, "learning_rate": 5.846400000000001e-06, "loss": 0.0454, "step": 20769 }, { "epoch": 1.1073199338913473, "grad_norm": 4.928228855133057, "learning_rate": 5.846200000000001e-06, "loss": 0.0762, "step": 20770 }, { "epoch": 1.107373247321, "grad_norm": 1.9762650728225708, "learning_rate": 5.8460000000000004e-06, "loss": 0.0155, "step": 20771 }, { "epoch": 1.107426560750653, "grad_norm": 1.9679852724075317, "learning_rate": 5.8458e-06, "loss": 0.0312, "step": 20772 }, { "epoch": 1.107479874180306, "grad_norm": 1.036287784576416, "learning_rate": 5.8456e-06, "loss": 0.0291, "step": 20773 }, { "epoch": 1.107533187609959, "grad_norm": 1.6055805683135986, "learning_rate": 5.8454e-06, "loss": 0.0485, "step": 20774 }, { "epoch": 1.107586501039612, "grad_norm": 1.8824186325073242, "learning_rate": 5.845200000000001e-06, "loss": 0.0245, "step": 20775 }, { "epoch": 1.1076398144692647, "grad_norm": 4.007626533508301, "learning_rate": 5.845000000000001e-06, "loss": 0.0507, "step": 20776 }, { "epoch": 1.1076931278989177, "grad_norm": 1.4634177684783936, "learning_rate": 5.8448e-06, "loss": 0.0209, "step": 20777 }, { "epoch": 1.1077464413285707, "grad_norm": 1.5194147825241089, "learning_rate": 5.8446e-06, "loss": 0.0196, "step": 20778 }, { "epoch": 1.1077997547582237, "grad_norm": 1.7748750448226929, "learning_rate": 5.8444e-06, "loss": 0.0135, "step": 20779 }, { "epoch": 1.1078530681878764, "grad_norm": 3.7517995834350586, "learning_rate": 5.844200000000001e-06, "loss": 0.0549, "step": 20780 }, { "epoch": 1.1079063816175294, "grad_norm": 1.8433139324188232, "learning_rate": 5.844000000000001e-06, "loss": 0.0209, "step": 20781 }, { "epoch": 1.1079596950471824, "grad_norm": 1.3236498832702637, "learning_rate": 5.843800000000001e-06, "loss": 0.0314, "step": 20782 }, { "epoch": 1.1080130084768354, "grad_norm": 1.2718324661254883, "learning_rate": 5.8436e-06, "loss": 0.0314, "step": 20783 }, { "epoch": 1.1080663219064881, "grad_norm": 2.7546465396881104, "learning_rate": 5.8434e-06, "loss": 0.0205, "step": 20784 }, { "epoch": 1.108119635336141, "grad_norm": 1.199110984802246, "learning_rate": 5.8432e-06, "loss": 0.0112, "step": 20785 }, { "epoch": 1.108172948765794, "grad_norm": 2.169354200363159, "learning_rate": 5.843000000000001e-06, "loss": 0.0197, "step": 20786 }, { "epoch": 1.108226262195447, "grad_norm": 2.441009044647217, "learning_rate": 5.842800000000001e-06, "loss": 0.0211, "step": 20787 }, { "epoch": 1.1082795756251, "grad_norm": 1.4667903184890747, "learning_rate": 5.842600000000001e-06, "loss": 0.0311, "step": 20788 }, { "epoch": 1.1083328890547528, "grad_norm": 1.2937730550765991, "learning_rate": 5.8424e-06, "loss": 0.0235, "step": 20789 }, { "epoch": 1.1083862024844058, "grad_norm": 0.6049525737762451, "learning_rate": 5.8422e-06, "loss": 0.0073, "step": 20790 }, { "epoch": 1.1084395159140588, "grad_norm": 1.233845829963684, "learning_rate": 5.842000000000001e-06, "loss": 0.0185, "step": 20791 }, { "epoch": 1.1084928293437117, "grad_norm": 1.381272554397583, "learning_rate": 5.8418000000000005e-06, "loss": 0.0203, "step": 20792 }, { "epoch": 1.1085461427733647, "grad_norm": 2.84175443649292, "learning_rate": 5.841600000000001e-06, "loss": 0.0238, "step": 20793 }, { "epoch": 1.1085994562030175, "grad_norm": 2.4362447261810303, "learning_rate": 5.8414000000000005e-06, "loss": 0.0238, "step": 20794 }, { "epoch": 1.1086527696326705, "grad_norm": 1.3805756568908691, "learning_rate": 5.8412e-06, "loss": 0.0193, "step": 20795 }, { "epoch": 1.1087060830623234, "grad_norm": 1.4525530338287354, "learning_rate": 5.841e-06, "loss": 0.0256, "step": 20796 }, { "epoch": 1.1087593964919764, "grad_norm": 1.4356049299240112, "learning_rate": 5.840800000000001e-06, "loss": 0.0274, "step": 20797 }, { "epoch": 1.1088127099216292, "grad_norm": 1.2016366720199585, "learning_rate": 5.8406000000000005e-06, "loss": 0.0202, "step": 20798 }, { "epoch": 1.1088660233512821, "grad_norm": 2.2845468521118164, "learning_rate": 5.840400000000001e-06, "loss": 0.0198, "step": 20799 }, { "epoch": 1.1089193367809351, "grad_norm": 3.2393548488616943, "learning_rate": 5.8402000000000004e-06, "loss": 0.0559, "step": 20800 }, { "epoch": 1.108972650210588, "grad_norm": 2.6173431873321533, "learning_rate": 5.84e-06, "loss": 0.0346, "step": 20801 }, { "epoch": 1.1090259636402409, "grad_norm": 1.0362337827682495, "learning_rate": 5.839800000000001e-06, "loss": 0.0208, "step": 20802 }, { "epoch": 1.1090792770698938, "grad_norm": 1.4167205095291138, "learning_rate": 5.839600000000001e-06, "loss": 0.0416, "step": 20803 }, { "epoch": 1.1091325904995468, "grad_norm": 1.61846923828125, "learning_rate": 5.8394e-06, "loss": 0.0348, "step": 20804 }, { "epoch": 1.1091859039291998, "grad_norm": 1.5317168235778809, "learning_rate": 5.839200000000001e-06, "loss": 0.0212, "step": 20805 }, { "epoch": 1.1092392173588528, "grad_norm": 1.5515906810760498, "learning_rate": 5.839e-06, "loss": 0.0289, "step": 20806 }, { "epoch": 1.1092925307885055, "grad_norm": 2.43180513381958, "learning_rate": 5.8388e-06, "loss": 0.0685, "step": 20807 }, { "epoch": 1.1093458442181585, "grad_norm": 1.5401653051376343, "learning_rate": 5.838600000000001e-06, "loss": 0.0113, "step": 20808 }, { "epoch": 1.1093991576478115, "grad_norm": 2.05981183052063, "learning_rate": 5.838400000000001e-06, "loss": 0.025, "step": 20809 }, { "epoch": 1.1094524710774645, "grad_norm": 3.656984806060791, "learning_rate": 5.8382e-06, "loss": 0.0311, "step": 20810 }, { "epoch": 1.1095057845071175, "grad_norm": 2.2912721633911133, "learning_rate": 5.838000000000001e-06, "loss": 0.0165, "step": 20811 }, { "epoch": 1.1095590979367702, "grad_norm": 1.065625786781311, "learning_rate": 5.8378e-06, "loss": 0.0232, "step": 20812 }, { "epoch": 1.1096124113664232, "grad_norm": 1.724879264831543, "learning_rate": 5.8376e-06, "loss": 0.0279, "step": 20813 }, { "epoch": 1.1096657247960762, "grad_norm": 1.9599010944366455, "learning_rate": 5.837400000000001e-06, "loss": 0.0335, "step": 20814 }, { "epoch": 1.1097190382257291, "grad_norm": 1.9846107959747314, "learning_rate": 5.837200000000001e-06, "loss": 0.0413, "step": 20815 }, { "epoch": 1.109772351655382, "grad_norm": 4.325468063354492, "learning_rate": 5.837e-06, "loss": 0.0273, "step": 20816 }, { "epoch": 1.1098256650850349, "grad_norm": 1.3701328039169312, "learning_rate": 5.836800000000001e-06, "loss": 0.0122, "step": 20817 }, { "epoch": 1.1098789785146879, "grad_norm": 0.7132276892662048, "learning_rate": 5.8366e-06, "loss": 0.0077, "step": 20818 }, { "epoch": 1.1099322919443408, "grad_norm": 1.752879023551941, "learning_rate": 5.8364000000000006e-06, "loss": 0.0201, "step": 20819 }, { "epoch": 1.1099856053739936, "grad_norm": 2.0382442474365234, "learning_rate": 5.836200000000001e-06, "loss": 0.0185, "step": 20820 }, { "epoch": 1.1100389188036466, "grad_norm": 0.9124921560287476, "learning_rate": 5.8360000000000005e-06, "loss": 0.0075, "step": 20821 }, { "epoch": 1.1100922322332996, "grad_norm": 2.0571682453155518, "learning_rate": 5.8358e-06, "loss": 0.029, "step": 20822 }, { "epoch": 1.1101455456629525, "grad_norm": 1.9954317808151245, "learning_rate": 5.8356000000000005e-06, "loss": 0.0378, "step": 20823 }, { "epoch": 1.1101988590926055, "grad_norm": 1.7687214612960815, "learning_rate": 5.8354e-06, "loss": 0.0406, "step": 20824 }, { "epoch": 1.1102521725222583, "grad_norm": 1.1134610176086426, "learning_rate": 5.8352000000000005e-06, "loss": 0.0139, "step": 20825 }, { "epoch": 1.1103054859519113, "grad_norm": 2.493727445602417, "learning_rate": 5.835000000000001e-06, "loss": 0.0432, "step": 20826 }, { "epoch": 1.1103587993815642, "grad_norm": 3.1352322101593018, "learning_rate": 5.8348000000000004e-06, "loss": 0.0455, "step": 20827 }, { "epoch": 1.1104121128112172, "grad_norm": 0.9552186727523804, "learning_rate": 5.8346e-06, "loss": 0.0068, "step": 20828 }, { "epoch": 1.1104654262408702, "grad_norm": 1.4865535497665405, "learning_rate": 5.8344e-06, "loss": 0.0145, "step": 20829 }, { "epoch": 1.110518739670523, "grad_norm": 1.8183622360229492, "learning_rate": 5.834200000000001e-06, "loss": 0.0256, "step": 20830 }, { "epoch": 1.110572053100176, "grad_norm": 3.094238042831421, "learning_rate": 5.834e-06, "loss": 0.0186, "step": 20831 }, { "epoch": 1.110625366529829, "grad_norm": 3.028390645980835, "learning_rate": 5.833800000000001e-06, "loss": 0.0266, "step": 20832 }, { "epoch": 1.1106786799594819, "grad_norm": 0.4771585166454315, "learning_rate": 5.8336e-06, "loss": 0.0044, "step": 20833 }, { "epoch": 1.1107319933891346, "grad_norm": 0.8875305652618408, "learning_rate": 5.8334e-06, "loss": 0.0223, "step": 20834 }, { "epoch": 1.1107853068187876, "grad_norm": 4.084014892578125, "learning_rate": 5.8332e-06, "loss": 0.0402, "step": 20835 }, { "epoch": 1.1108386202484406, "grad_norm": 1.6986775398254395, "learning_rate": 5.833000000000001e-06, "loss": 0.0293, "step": 20836 }, { "epoch": 1.1108919336780936, "grad_norm": 1.458463430404663, "learning_rate": 5.8328e-06, "loss": 0.0218, "step": 20837 }, { "epoch": 1.1109452471077463, "grad_norm": 1.3431758880615234, "learning_rate": 5.832600000000001e-06, "loss": 0.0216, "step": 20838 }, { "epoch": 1.1109985605373993, "grad_norm": 2.5814623832702637, "learning_rate": 5.8324e-06, "loss": 0.0441, "step": 20839 }, { "epoch": 1.1110518739670523, "grad_norm": 2.595388650894165, "learning_rate": 5.8322e-06, "loss": 0.0576, "step": 20840 }, { "epoch": 1.1111051873967053, "grad_norm": 1.4612572193145752, "learning_rate": 5.832000000000001e-06, "loss": 0.019, "step": 20841 }, { "epoch": 1.1111585008263583, "grad_norm": 1.3710945844650269, "learning_rate": 5.831800000000001e-06, "loss": 0.012, "step": 20842 }, { "epoch": 1.111211814256011, "grad_norm": 0.770848274230957, "learning_rate": 5.8316e-06, "loss": 0.01, "step": 20843 }, { "epoch": 1.111265127685664, "grad_norm": 2.087059259414673, "learning_rate": 5.831400000000001e-06, "loss": 0.0284, "step": 20844 }, { "epoch": 1.111318441115317, "grad_norm": 1.5236015319824219, "learning_rate": 5.8312e-06, "loss": 0.0298, "step": 20845 }, { "epoch": 1.11137175454497, "grad_norm": 2.928514003753662, "learning_rate": 5.831e-06, "loss": 0.0313, "step": 20846 }, { "epoch": 1.1114250679746227, "grad_norm": 2.6936075687408447, "learning_rate": 5.830800000000001e-06, "loss": 0.0202, "step": 20847 }, { "epoch": 1.1114783814042757, "grad_norm": 2.21132755279541, "learning_rate": 5.8306000000000006e-06, "loss": 0.0432, "step": 20848 }, { "epoch": 1.1115316948339287, "grad_norm": 2.157221555709839, "learning_rate": 5.8304e-06, "loss": 0.0284, "step": 20849 }, { "epoch": 1.1115850082635816, "grad_norm": 2.5910069942474365, "learning_rate": 5.8302000000000005e-06, "loss": 0.0466, "step": 20850 }, { "epoch": 1.1116383216932346, "grad_norm": 1.9772226810455322, "learning_rate": 5.83e-06, "loss": 0.0389, "step": 20851 }, { "epoch": 1.1116916351228874, "grad_norm": 1.4598380327224731, "learning_rate": 5.829800000000001e-06, "loss": 0.0124, "step": 20852 }, { "epoch": 1.1117449485525404, "grad_norm": 0.7843707799911499, "learning_rate": 5.829600000000001e-06, "loss": 0.0069, "step": 20853 }, { "epoch": 1.1117982619821933, "grad_norm": 2.0050199031829834, "learning_rate": 5.8294000000000005e-06, "loss": 0.0338, "step": 20854 }, { "epoch": 1.1118515754118463, "grad_norm": 1.1357421875, "learning_rate": 5.8292e-06, "loss": 0.0188, "step": 20855 }, { "epoch": 1.111904888841499, "grad_norm": 1.9153190851211548, "learning_rate": 5.8290000000000004e-06, "loss": 0.0244, "step": 20856 }, { "epoch": 1.111958202271152, "grad_norm": 1.9774448871612549, "learning_rate": 5.8288e-06, "loss": 0.0251, "step": 20857 }, { "epoch": 1.112011515700805, "grad_norm": 1.4333012104034424, "learning_rate": 5.828600000000001e-06, "loss": 0.0179, "step": 20858 }, { "epoch": 1.112064829130458, "grad_norm": 0.9503999352455139, "learning_rate": 5.828400000000001e-06, "loss": 0.0092, "step": 20859 }, { "epoch": 1.112118142560111, "grad_norm": 1.5115272998809814, "learning_rate": 5.8282e-06, "loss": 0.0301, "step": 20860 }, { "epoch": 1.1121714559897637, "grad_norm": 1.7995805740356445, "learning_rate": 5.828e-06, "loss": 0.0232, "step": 20861 }, { "epoch": 1.1122247694194167, "grad_norm": 1.8416143655776978, "learning_rate": 5.8278e-06, "loss": 0.0228, "step": 20862 }, { "epoch": 1.1122780828490697, "grad_norm": 3.4232797622680664, "learning_rate": 5.8276e-06, "loss": 0.0236, "step": 20863 }, { "epoch": 1.1123313962787227, "grad_norm": 1.4293491840362549, "learning_rate": 5.827400000000001e-06, "loss": 0.0254, "step": 20864 }, { "epoch": 1.1123847097083754, "grad_norm": 0.7729626893997192, "learning_rate": 5.827200000000001e-06, "loss": 0.0082, "step": 20865 }, { "epoch": 1.1124380231380284, "grad_norm": 0.6321392059326172, "learning_rate": 5.827e-06, "loss": 0.0051, "step": 20866 }, { "epoch": 1.1124913365676814, "grad_norm": 2.586963653564453, "learning_rate": 5.8268e-06, "loss": 0.051, "step": 20867 }, { "epoch": 1.1125446499973344, "grad_norm": 1.2005529403686523, "learning_rate": 5.8266e-06, "loss": 0.0203, "step": 20868 }, { "epoch": 1.1125979634269874, "grad_norm": 1.3010908365249634, "learning_rate": 5.826400000000001e-06, "loss": 0.0157, "step": 20869 }, { "epoch": 1.1126512768566401, "grad_norm": 2.2281081676483154, "learning_rate": 5.826200000000001e-06, "loss": 0.036, "step": 20870 }, { "epoch": 1.112704590286293, "grad_norm": 1.3349019289016724, "learning_rate": 5.826000000000001e-06, "loss": 0.01, "step": 20871 }, { "epoch": 1.112757903715946, "grad_norm": 0.9413119554519653, "learning_rate": 5.8258e-06, "loss": 0.0083, "step": 20872 }, { "epoch": 1.112811217145599, "grad_norm": 1.0578184127807617, "learning_rate": 5.8256e-06, "loss": 0.0148, "step": 20873 }, { "epoch": 1.1128645305752518, "grad_norm": 1.130433201789856, "learning_rate": 5.8254e-06, "loss": 0.0165, "step": 20874 }, { "epoch": 1.1129178440049048, "grad_norm": 1.9879378080368042, "learning_rate": 5.8252000000000006e-06, "loss": 0.0122, "step": 20875 }, { "epoch": 1.1129711574345578, "grad_norm": 2.3251731395721436, "learning_rate": 5.825000000000001e-06, "loss": 0.0316, "step": 20876 }, { "epoch": 1.1130244708642107, "grad_norm": 1.273098349571228, "learning_rate": 5.8248000000000005e-06, "loss": 0.0222, "step": 20877 }, { "epoch": 1.1130777842938637, "grad_norm": 0.6091499328613281, "learning_rate": 5.8246e-06, "loss": 0.0044, "step": 20878 }, { "epoch": 1.1131310977235165, "grad_norm": 1.7349764108657837, "learning_rate": 5.8244e-06, "loss": 0.0269, "step": 20879 }, { "epoch": 1.1131844111531695, "grad_norm": 1.8147066831588745, "learning_rate": 5.824200000000001e-06, "loss": 0.0118, "step": 20880 }, { "epoch": 1.1132377245828224, "grad_norm": 2.626382827758789, "learning_rate": 5.8240000000000005e-06, "loss": 0.0467, "step": 20881 }, { "epoch": 1.1132910380124754, "grad_norm": 1.9748272895812988, "learning_rate": 5.823800000000001e-06, "loss": 0.0172, "step": 20882 }, { "epoch": 1.1133443514421282, "grad_norm": 3.573596715927124, "learning_rate": 5.8236000000000005e-06, "loss": 0.0239, "step": 20883 }, { "epoch": 1.1133976648717812, "grad_norm": 1.5561074018478394, "learning_rate": 5.8234e-06, "loss": 0.0406, "step": 20884 }, { "epoch": 1.1134509783014341, "grad_norm": 1.5477585792541504, "learning_rate": 5.8231999999999996e-06, "loss": 0.0311, "step": 20885 }, { "epoch": 1.1135042917310871, "grad_norm": 0.7942876219749451, "learning_rate": 5.823000000000001e-06, "loss": 0.0188, "step": 20886 }, { "epoch": 1.11355760516074, "grad_norm": 1.6567548513412476, "learning_rate": 5.8228e-06, "loss": 0.0219, "step": 20887 }, { "epoch": 1.1136109185903929, "grad_norm": 2.50427508354187, "learning_rate": 5.822600000000001e-06, "loss": 0.0182, "step": 20888 }, { "epoch": 1.1136642320200458, "grad_norm": 1.8937788009643555, "learning_rate": 5.8224e-06, "loss": 0.0142, "step": 20889 }, { "epoch": 1.1137175454496988, "grad_norm": 1.1788344383239746, "learning_rate": 5.8222e-06, "loss": 0.0168, "step": 20890 }, { "epoch": 1.1137708588793518, "grad_norm": 1.5250340700149536, "learning_rate": 5.822000000000001e-06, "loss": 0.0164, "step": 20891 }, { "epoch": 1.1138241723090045, "grad_norm": 1.926365852355957, "learning_rate": 5.821800000000001e-06, "loss": 0.0254, "step": 20892 }, { "epoch": 1.1138774857386575, "grad_norm": 1.8167234659194946, "learning_rate": 5.8216e-06, "loss": 0.0208, "step": 20893 }, { "epoch": 1.1139307991683105, "grad_norm": 1.4969801902770996, "learning_rate": 5.821400000000001e-06, "loss": 0.0275, "step": 20894 }, { "epoch": 1.1139841125979635, "grad_norm": 1.651738166809082, "learning_rate": 5.8212e-06, "loss": 0.014, "step": 20895 }, { "epoch": 1.1140374260276165, "grad_norm": 2.4558115005493164, "learning_rate": 5.821e-06, "loss": 0.0396, "step": 20896 }, { "epoch": 1.1140907394572692, "grad_norm": 1.5633994340896606, "learning_rate": 5.820800000000001e-06, "loss": 0.0225, "step": 20897 }, { "epoch": 1.1141440528869222, "grad_norm": 1.4833149909973145, "learning_rate": 5.820600000000001e-06, "loss": 0.0093, "step": 20898 }, { "epoch": 1.1141973663165752, "grad_norm": 2.563586950302124, "learning_rate": 5.8204e-06, "loss": 0.0602, "step": 20899 }, { "epoch": 1.1142506797462282, "grad_norm": 1.2389299869537354, "learning_rate": 5.820200000000001e-06, "loss": 0.0196, "step": 20900 }, { "epoch": 1.114303993175881, "grad_norm": 1.8483535051345825, "learning_rate": 5.82e-06, "loss": 0.0253, "step": 20901 }, { "epoch": 1.114357306605534, "grad_norm": 2.485006093978882, "learning_rate": 5.819800000000001e-06, "loss": 0.0269, "step": 20902 }, { "epoch": 1.1144106200351869, "grad_norm": 2.216799020767212, "learning_rate": 5.819600000000001e-06, "loss": 0.0225, "step": 20903 }, { "epoch": 1.1144639334648399, "grad_norm": 1.6835170984268188, "learning_rate": 5.8194000000000006e-06, "loss": 0.0231, "step": 20904 }, { "epoch": 1.1145172468944926, "grad_norm": 2.469348192214966, "learning_rate": 5.8192e-06, "loss": 0.0148, "step": 20905 }, { "epoch": 1.1145705603241456, "grad_norm": 2.5453808307647705, "learning_rate": 5.8190000000000005e-06, "loss": 0.0205, "step": 20906 }, { "epoch": 1.1146238737537986, "grad_norm": 1.3221213817596436, "learning_rate": 5.8188e-06, "loss": 0.022, "step": 20907 }, { "epoch": 1.1146771871834515, "grad_norm": 1.4603666067123413, "learning_rate": 5.8186000000000005e-06, "loss": 0.023, "step": 20908 }, { "epoch": 1.1147305006131045, "grad_norm": 2.0021984577178955, "learning_rate": 5.818400000000001e-06, "loss": 0.0309, "step": 20909 }, { "epoch": 1.1147838140427573, "grad_norm": 1.302758812904358, "learning_rate": 5.8182000000000005e-06, "loss": 0.0141, "step": 20910 }, { "epoch": 1.1148371274724103, "grad_norm": 1.6466238498687744, "learning_rate": 5.818e-06, "loss": 0.0228, "step": 20911 }, { "epoch": 1.1148904409020632, "grad_norm": 2.543161392211914, "learning_rate": 5.8178000000000004e-06, "loss": 0.032, "step": 20912 }, { "epoch": 1.1149437543317162, "grad_norm": 2.200007438659668, "learning_rate": 5.817600000000001e-06, "loss": 0.0363, "step": 20913 }, { "epoch": 1.1149970677613692, "grad_norm": 1.6046409606933594, "learning_rate": 5.8174e-06, "loss": 0.0219, "step": 20914 }, { "epoch": 1.115050381191022, "grad_norm": 2.2836647033691406, "learning_rate": 5.817200000000001e-06, "loss": 0.0326, "step": 20915 }, { "epoch": 1.115103694620675, "grad_norm": 2.1881587505340576, "learning_rate": 5.817e-06, "loss": 0.0364, "step": 20916 }, { "epoch": 1.115157008050328, "grad_norm": 2.3420603275299072, "learning_rate": 5.8168e-06, "loss": 0.0126, "step": 20917 }, { "epoch": 1.115210321479981, "grad_norm": 2.55906343460083, "learning_rate": 5.8166e-06, "loss": 0.0657, "step": 20918 }, { "epoch": 1.1152636349096336, "grad_norm": 2.2643401622772217, "learning_rate": 5.816400000000001e-06, "loss": 0.0185, "step": 20919 }, { "epoch": 1.1153169483392866, "grad_norm": 2.7381350994110107, "learning_rate": 5.8162e-06, "loss": 0.0237, "step": 20920 }, { "epoch": 1.1153702617689396, "grad_norm": 0.6932117342948914, "learning_rate": 5.816000000000001e-06, "loss": 0.0047, "step": 20921 }, { "epoch": 1.1154235751985926, "grad_norm": 0.466770738363266, "learning_rate": 5.8158e-06, "loss": 0.0038, "step": 20922 }, { "epoch": 1.1154768886282453, "grad_norm": 2.119452714920044, "learning_rate": 5.8156e-06, "loss": 0.0256, "step": 20923 }, { "epoch": 1.1155302020578983, "grad_norm": 0.8874190449714661, "learning_rate": 5.8154e-06, "loss": 0.0133, "step": 20924 }, { "epoch": 1.1155835154875513, "grad_norm": 0.8935820460319519, "learning_rate": 5.815200000000001e-06, "loss": 0.0067, "step": 20925 }, { "epoch": 1.1156368289172043, "grad_norm": 1.7499853372573853, "learning_rate": 5.815e-06, "loss": 0.0319, "step": 20926 }, { "epoch": 1.1156901423468573, "grad_norm": 1.7569535970687866, "learning_rate": 5.814800000000001e-06, "loss": 0.0195, "step": 20927 }, { "epoch": 1.11574345577651, "grad_norm": 1.6671440601348877, "learning_rate": 5.8146e-06, "loss": 0.0372, "step": 20928 }, { "epoch": 1.115796769206163, "grad_norm": 0.688697099685669, "learning_rate": 5.8144e-06, "loss": 0.0055, "step": 20929 }, { "epoch": 1.115850082635816, "grad_norm": 1.4738199710845947, "learning_rate": 5.814200000000001e-06, "loss": 0.0247, "step": 20930 }, { "epoch": 1.115903396065469, "grad_norm": 1.429145097732544, "learning_rate": 5.814000000000001e-06, "loss": 0.0085, "step": 20931 }, { "epoch": 1.115956709495122, "grad_norm": 2.193333387374878, "learning_rate": 5.8138e-06, "loss": 0.0143, "step": 20932 }, { "epoch": 1.1160100229247747, "grad_norm": 2.849320888519287, "learning_rate": 5.8136000000000005e-06, "loss": 0.0468, "step": 20933 }, { "epoch": 1.1160633363544277, "grad_norm": 0.337835431098938, "learning_rate": 5.8134e-06, "loss": 0.004, "step": 20934 }, { "epoch": 1.1161166497840806, "grad_norm": 2.931022882461548, "learning_rate": 5.8132e-06, "loss": 0.0395, "step": 20935 }, { "epoch": 1.1161699632137336, "grad_norm": 0.9893785119056702, "learning_rate": 5.813000000000001e-06, "loss": 0.0073, "step": 20936 }, { "epoch": 1.1162232766433864, "grad_norm": 1.942409873008728, "learning_rate": 5.8128000000000005e-06, "loss": 0.0138, "step": 20937 }, { "epoch": 1.1162765900730394, "grad_norm": 0.720230758190155, "learning_rate": 5.8126e-06, "loss": 0.0126, "step": 20938 }, { "epoch": 1.1163299035026923, "grad_norm": 2.5651943683624268, "learning_rate": 5.8124000000000005e-06, "loss": 0.0484, "step": 20939 }, { "epoch": 1.1163832169323453, "grad_norm": 2.2474589347839355, "learning_rate": 5.8122e-06, "loss": 0.0186, "step": 20940 }, { "epoch": 1.116436530361998, "grad_norm": 1.706284999847412, "learning_rate": 5.812000000000001e-06, "loss": 0.0382, "step": 20941 }, { "epoch": 1.116489843791651, "grad_norm": 2.211855173110962, "learning_rate": 5.811800000000001e-06, "loss": 0.0146, "step": 20942 }, { "epoch": 1.116543157221304, "grad_norm": 2.3054487705230713, "learning_rate": 5.8116e-06, "loss": 0.03, "step": 20943 }, { "epoch": 1.116596470650957, "grad_norm": 1.5793368816375732, "learning_rate": 5.8114e-06, "loss": 0.0283, "step": 20944 }, { "epoch": 1.11664978408061, "grad_norm": 1.944638967514038, "learning_rate": 5.8112e-06, "loss": 0.0402, "step": 20945 }, { "epoch": 1.1167030975102628, "grad_norm": 1.8406485319137573, "learning_rate": 5.811e-06, "loss": 0.0156, "step": 20946 }, { "epoch": 1.1167564109399157, "grad_norm": 1.628861665725708, "learning_rate": 5.810800000000001e-06, "loss": 0.039, "step": 20947 }, { "epoch": 1.1168097243695687, "grad_norm": 0.5681256651878357, "learning_rate": 5.810600000000001e-06, "loss": 0.0063, "step": 20948 }, { "epoch": 1.1168630377992217, "grad_norm": 2.4314329624176025, "learning_rate": 5.8104e-06, "loss": 0.0266, "step": 20949 }, { "epoch": 1.1169163512288747, "grad_norm": 2.2985942363739014, "learning_rate": 5.8102e-06, "loss": 0.047, "step": 20950 }, { "epoch": 1.1169696646585274, "grad_norm": 0.8606764674186707, "learning_rate": 5.81e-06, "loss": 0.0161, "step": 20951 }, { "epoch": 1.1170229780881804, "grad_norm": 2.6983768939971924, "learning_rate": 5.809800000000001e-06, "loss": 0.0577, "step": 20952 }, { "epoch": 1.1170762915178334, "grad_norm": 3.6097218990325928, "learning_rate": 5.809600000000001e-06, "loss": 0.0347, "step": 20953 }, { "epoch": 1.1171296049474864, "grad_norm": 3.2586281299591064, "learning_rate": 5.809400000000001e-06, "loss": 0.023, "step": 20954 }, { "epoch": 1.1171829183771391, "grad_norm": 1.374288558959961, "learning_rate": 5.8092e-06, "loss": 0.0152, "step": 20955 }, { "epoch": 1.117236231806792, "grad_norm": 2.393043279647827, "learning_rate": 5.809e-06, "loss": 0.0368, "step": 20956 }, { "epoch": 1.117289545236445, "grad_norm": 1.1090646982192993, "learning_rate": 5.8088e-06, "loss": 0.0154, "step": 20957 }, { "epoch": 1.117342858666098, "grad_norm": 1.3856778144836426, "learning_rate": 5.808600000000001e-06, "loss": 0.0276, "step": 20958 }, { "epoch": 1.1173961720957508, "grad_norm": 1.5141055583953857, "learning_rate": 5.808400000000001e-06, "loss": 0.0355, "step": 20959 }, { "epoch": 1.1174494855254038, "grad_norm": 1.6258249282836914, "learning_rate": 5.8082000000000006e-06, "loss": 0.0268, "step": 20960 }, { "epoch": 1.1175027989550568, "grad_norm": 2.1987881660461426, "learning_rate": 5.808e-06, "loss": 0.0392, "step": 20961 }, { "epoch": 1.1175561123847098, "grad_norm": 2.2010574340820312, "learning_rate": 5.8078e-06, "loss": 0.0303, "step": 20962 }, { "epoch": 1.1176094258143627, "grad_norm": 2.316288471221924, "learning_rate": 5.807600000000001e-06, "loss": 0.0301, "step": 20963 }, { "epoch": 1.1176627392440155, "grad_norm": 2.2647759914398193, "learning_rate": 5.8074000000000005e-06, "loss": 0.0378, "step": 20964 }, { "epoch": 1.1177160526736685, "grad_norm": 1.5896540880203247, "learning_rate": 5.807200000000001e-06, "loss": 0.0242, "step": 20965 }, { "epoch": 1.1177693661033214, "grad_norm": 5.454209327697754, "learning_rate": 5.8070000000000005e-06, "loss": 0.0655, "step": 20966 }, { "epoch": 1.1178226795329744, "grad_norm": 1.0397874116897583, "learning_rate": 5.8068e-06, "loss": 0.0144, "step": 20967 }, { "epoch": 1.1178759929626272, "grad_norm": 1.9241994619369507, "learning_rate": 5.8066e-06, "loss": 0.0284, "step": 20968 }, { "epoch": 1.1179293063922802, "grad_norm": 1.0924232006072998, "learning_rate": 5.806400000000001e-06, "loss": 0.0255, "step": 20969 }, { "epoch": 1.1179826198219331, "grad_norm": 1.3144668340682983, "learning_rate": 5.8062e-06, "loss": 0.0182, "step": 20970 }, { "epoch": 1.1180359332515861, "grad_norm": 2.464913845062256, "learning_rate": 5.806000000000001e-06, "loss": 0.0322, "step": 20971 }, { "epoch": 1.118089246681239, "grad_norm": 1.1035876274108887, "learning_rate": 5.8058e-06, "loss": 0.0088, "step": 20972 }, { "epoch": 1.1181425601108919, "grad_norm": 1.6199434995651245, "learning_rate": 5.8056e-06, "loss": 0.0196, "step": 20973 }, { "epoch": 1.1181958735405448, "grad_norm": 0.9380237460136414, "learning_rate": 5.8053999999999995e-06, "loss": 0.007, "step": 20974 }, { "epoch": 1.1182491869701978, "grad_norm": 1.1510887145996094, "learning_rate": 5.805200000000001e-06, "loss": 0.0131, "step": 20975 }, { "epoch": 1.1183025003998508, "grad_norm": 2.941718578338623, "learning_rate": 5.805e-06, "loss": 0.0604, "step": 20976 }, { "epoch": 1.1183558138295036, "grad_norm": 2.4268996715545654, "learning_rate": 5.804800000000001e-06, "loss": 0.0387, "step": 20977 }, { "epoch": 1.1184091272591565, "grad_norm": 2.005047082901001, "learning_rate": 5.8046e-06, "loss": 0.0363, "step": 20978 }, { "epoch": 1.1184624406888095, "grad_norm": 1.2708817720413208, "learning_rate": 5.8044e-06, "loss": 0.0135, "step": 20979 }, { "epoch": 1.1185157541184625, "grad_norm": 0.9118470549583435, "learning_rate": 5.804200000000001e-06, "loss": 0.0092, "step": 20980 }, { "epoch": 1.1185690675481155, "grad_norm": 1.1918548345565796, "learning_rate": 5.804000000000001e-06, "loss": 0.0231, "step": 20981 }, { "epoch": 1.1186223809777682, "grad_norm": 1.435104250907898, "learning_rate": 5.8038e-06, "loss": 0.0192, "step": 20982 }, { "epoch": 1.1186756944074212, "grad_norm": 3.8092761039733887, "learning_rate": 5.803600000000001e-06, "loss": 0.0413, "step": 20983 }, { "epoch": 1.1187290078370742, "grad_norm": 2.066117286682129, "learning_rate": 5.8034e-06, "loss": 0.0214, "step": 20984 }, { "epoch": 1.1187823212667272, "grad_norm": 1.9222989082336426, "learning_rate": 5.8032e-06, "loss": 0.0235, "step": 20985 }, { "epoch": 1.11883563469638, "grad_norm": 1.3784968852996826, "learning_rate": 5.803000000000001e-06, "loss": 0.0211, "step": 20986 }, { "epoch": 1.118888948126033, "grad_norm": 0.9715885519981384, "learning_rate": 5.802800000000001e-06, "loss": 0.0076, "step": 20987 }, { "epoch": 1.1189422615556859, "grad_norm": 1.300176978111267, "learning_rate": 5.8026e-06, "loss": 0.0373, "step": 20988 }, { "epoch": 1.1189955749853389, "grad_norm": 2.26165509223938, "learning_rate": 5.8024000000000006e-06, "loss": 0.0108, "step": 20989 }, { "epoch": 1.1190488884149918, "grad_norm": 1.8501455783843994, "learning_rate": 5.8022e-06, "loss": 0.0391, "step": 20990 }, { "epoch": 1.1191022018446446, "grad_norm": 0.7113562226295471, "learning_rate": 5.802000000000001e-06, "loss": 0.0073, "step": 20991 }, { "epoch": 1.1191555152742976, "grad_norm": 2.020674467086792, "learning_rate": 5.801800000000001e-06, "loss": 0.0395, "step": 20992 }, { "epoch": 1.1192088287039506, "grad_norm": 1.4009895324707031, "learning_rate": 5.8016000000000005e-06, "loss": 0.0171, "step": 20993 }, { "epoch": 1.1192621421336035, "grad_norm": 3.7960426807403564, "learning_rate": 5.8014e-06, "loss": 0.0292, "step": 20994 }, { "epoch": 1.1193154555632563, "grad_norm": 2.959311008453369, "learning_rate": 5.8012000000000005e-06, "loss": 0.0343, "step": 20995 }, { "epoch": 1.1193687689929093, "grad_norm": 1.7147040367126465, "learning_rate": 5.801e-06, "loss": 0.0219, "step": 20996 }, { "epoch": 1.1194220824225622, "grad_norm": 1.5104293823242188, "learning_rate": 5.800800000000001e-06, "loss": 0.0218, "step": 20997 }, { "epoch": 1.1194753958522152, "grad_norm": 2.9120254516601562, "learning_rate": 5.800600000000001e-06, "loss": 0.0262, "step": 20998 }, { "epoch": 1.1195287092818682, "grad_norm": 0.7412688732147217, "learning_rate": 5.8004e-06, "loss": 0.0078, "step": 20999 }, { "epoch": 1.119582022711521, "grad_norm": 1.8653589487075806, "learning_rate": 5.8002e-06, "loss": 0.0411, "step": 21000 }, { "epoch": 1.119635336141174, "grad_norm": 1.5007684230804443, "learning_rate": 5.8e-06, "loss": 0.0218, "step": 21001 }, { "epoch": 1.119688649570827, "grad_norm": 0.5623663067817688, "learning_rate": 5.799800000000001e-06, "loss": 0.0057, "step": 21002 }, { "epoch": 1.11974196300048, "grad_norm": 1.2237648963928223, "learning_rate": 5.799600000000001e-06, "loss": 0.0127, "step": 21003 }, { "epoch": 1.1197952764301327, "grad_norm": 2.3828325271606445, "learning_rate": 5.799400000000001e-06, "loss": 0.0275, "step": 21004 }, { "epoch": 1.1198485898597856, "grad_norm": 3.3237833976745605, "learning_rate": 5.7992e-06, "loss": 0.0277, "step": 21005 }, { "epoch": 1.1199019032894386, "grad_norm": 1.683516502380371, "learning_rate": 5.799e-06, "loss": 0.0255, "step": 21006 }, { "epoch": 1.1199552167190916, "grad_norm": 2.3180019855499268, "learning_rate": 5.7988e-06, "loss": 0.0431, "step": 21007 }, { "epoch": 1.1200085301487444, "grad_norm": 1.2351531982421875, "learning_rate": 5.798600000000001e-06, "loss": 0.0217, "step": 21008 }, { "epoch": 1.1200618435783973, "grad_norm": 1.2998007535934448, "learning_rate": 5.798400000000001e-06, "loss": 0.0205, "step": 21009 }, { "epoch": 1.1201151570080503, "grad_norm": 1.2070034742355347, "learning_rate": 5.798200000000001e-06, "loss": 0.0087, "step": 21010 }, { "epoch": 1.1201684704377033, "grad_norm": 2.493530750274658, "learning_rate": 5.798e-06, "loss": 0.0406, "step": 21011 }, { "epoch": 1.1202217838673563, "grad_norm": 3.1567766666412354, "learning_rate": 5.7978e-06, "loss": 0.0182, "step": 21012 }, { "epoch": 1.120275097297009, "grad_norm": 1.2348072528839111, "learning_rate": 5.797600000000001e-06, "loss": 0.0262, "step": 21013 }, { "epoch": 1.120328410726662, "grad_norm": 0.7364248633384705, "learning_rate": 5.797400000000001e-06, "loss": 0.007, "step": 21014 }, { "epoch": 1.120381724156315, "grad_norm": 1.2109858989715576, "learning_rate": 5.797200000000001e-06, "loss": 0.0166, "step": 21015 }, { "epoch": 1.120435037585968, "grad_norm": 2.6556591987609863, "learning_rate": 5.7970000000000006e-06, "loss": 0.036, "step": 21016 }, { "epoch": 1.120488351015621, "grad_norm": 0.9154651165008545, "learning_rate": 5.7968e-06, "loss": 0.0076, "step": 21017 }, { "epoch": 1.1205416644452737, "grad_norm": 3.1180222034454346, "learning_rate": 5.7966e-06, "loss": 0.0322, "step": 21018 }, { "epoch": 1.1205949778749267, "grad_norm": 3.3580124378204346, "learning_rate": 5.796400000000001e-06, "loss": 0.0304, "step": 21019 }, { "epoch": 1.1206482913045797, "grad_norm": 0.9263734817504883, "learning_rate": 5.7962000000000005e-06, "loss": 0.0077, "step": 21020 }, { "epoch": 1.1207016047342326, "grad_norm": 2.590743064880371, "learning_rate": 5.796000000000001e-06, "loss": 0.0375, "step": 21021 }, { "epoch": 1.1207549181638854, "grad_norm": 1.552748680114746, "learning_rate": 5.7958000000000005e-06, "loss": 0.0189, "step": 21022 }, { "epoch": 1.1208082315935384, "grad_norm": 1.5433228015899658, "learning_rate": 5.7956e-06, "loss": 0.0327, "step": 21023 }, { "epoch": 1.1208615450231914, "grad_norm": 1.753443717956543, "learning_rate": 5.795400000000001e-06, "loss": 0.0184, "step": 21024 }, { "epoch": 1.1209148584528443, "grad_norm": 1.5757323503494263, "learning_rate": 5.795200000000001e-06, "loss": 0.0257, "step": 21025 }, { "epoch": 1.120968171882497, "grad_norm": 1.7221803665161133, "learning_rate": 5.795e-06, "loss": 0.0389, "step": 21026 }, { "epoch": 1.12102148531215, "grad_norm": 1.5655676126480103, "learning_rate": 5.794800000000001e-06, "loss": 0.0379, "step": 21027 }, { "epoch": 1.121074798741803, "grad_norm": 2.7653071880340576, "learning_rate": 5.7946e-06, "loss": 0.0415, "step": 21028 }, { "epoch": 1.121128112171456, "grad_norm": 1.2253130674362183, "learning_rate": 5.7944e-06, "loss": 0.0154, "step": 21029 }, { "epoch": 1.121181425601109, "grad_norm": 1.6737831830978394, "learning_rate": 5.794200000000001e-06, "loss": 0.0124, "step": 21030 }, { "epoch": 1.1212347390307618, "grad_norm": 1.3251144886016846, "learning_rate": 5.794000000000001e-06, "loss": 0.0239, "step": 21031 }, { "epoch": 1.1212880524604147, "grad_norm": 2.494706869125366, "learning_rate": 5.7938e-06, "loss": 0.0232, "step": 21032 }, { "epoch": 1.1213413658900677, "grad_norm": 1.8950533866882324, "learning_rate": 5.793600000000001e-06, "loss": 0.0357, "step": 21033 }, { "epoch": 1.1213946793197207, "grad_norm": 2.4889590740203857, "learning_rate": 5.7934e-06, "loss": 0.0193, "step": 21034 }, { "epoch": 1.1214479927493737, "grad_norm": 3.0145461559295654, "learning_rate": 5.7932e-06, "loss": 0.051, "step": 21035 }, { "epoch": 1.1215013061790264, "grad_norm": 3.3194518089294434, "learning_rate": 5.793000000000001e-06, "loss": 0.0223, "step": 21036 }, { "epoch": 1.1215546196086794, "grad_norm": 4.245042324066162, "learning_rate": 5.792800000000001e-06, "loss": 0.0455, "step": 21037 }, { "epoch": 1.1216079330383324, "grad_norm": 0.9281466603279114, "learning_rate": 5.7926e-06, "loss": 0.0075, "step": 21038 }, { "epoch": 1.1216612464679854, "grad_norm": 1.41974675655365, "learning_rate": 5.792400000000001e-06, "loss": 0.0198, "step": 21039 }, { "epoch": 1.1217145598976381, "grad_norm": 1.4377827644348145, "learning_rate": 5.7922e-06, "loss": 0.0184, "step": 21040 }, { "epoch": 1.121767873327291, "grad_norm": 0.9535238742828369, "learning_rate": 5.792000000000001e-06, "loss": 0.0185, "step": 21041 }, { "epoch": 1.121821186756944, "grad_norm": 2.0867583751678467, "learning_rate": 5.791800000000001e-06, "loss": 0.0493, "step": 21042 }, { "epoch": 1.121874500186597, "grad_norm": 2.4951677322387695, "learning_rate": 5.791600000000001e-06, "loss": 0.0312, "step": 21043 }, { "epoch": 1.1219278136162498, "grad_norm": 1.453772783279419, "learning_rate": 5.7914e-06, "loss": 0.0108, "step": 21044 }, { "epoch": 1.1219811270459028, "grad_norm": 2.4638006687164307, "learning_rate": 5.7912000000000006e-06, "loss": 0.0131, "step": 21045 }, { "epoch": 1.1220344404755558, "grad_norm": 1.6785547733306885, "learning_rate": 5.791e-06, "loss": 0.0152, "step": 21046 }, { "epoch": 1.1220877539052088, "grad_norm": 1.336134433746338, "learning_rate": 5.7908000000000005e-06, "loss": 0.0263, "step": 21047 }, { "epoch": 1.1221410673348617, "grad_norm": 1.6942017078399658, "learning_rate": 5.790600000000001e-06, "loss": 0.0171, "step": 21048 }, { "epoch": 1.1221943807645145, "grad_norm": 1.0199304819107056, "learning_rate": 5.7904000000000005e-06, "loss": 0.0146, "step": 21049 }, { "epoch": 1.1222476941941675, "grad_norm": 1.505918025970459, "learning_rate": 5.7902e-06, "loss": 0.0262, "step": 21050 }, { "epoch": 1.1223010076238205, "grad_norm": 2.9294042587280273, "learning_rate": 5.7900000000000005e-06, "loss": 0.0117, "step": 21051 }, { "epoch": 1.1223543210534734, "grad_norm": 2.2908565998077393, "learning_rate": 5.789800000000001e-06, "loss": 0.0479, "step": 21052 }, { "epoch": 1.1224076344831264, "grad_norm": 1.0707513093948364, "learning_rate": 5.7896000000000004e-06, "loss": 0.0139, "step": 21053 }, { "epoch": 1.1224609479127792, "grad_norm": 1.7792555093765259, "learning_rate": 5.789400000000001e-06, "loss": 0.0251, "step": 21054 }, { "epoch": 1.1225142613424322, "grad_norm": 1.1208348274230957, "learning_rate": 5.7892e-06, "loss": 0.0103, "step": 21055 }, { "epoch": 1.1225675747720851, "grad_norm": 0.6370444893836975, "learning_rate": 5.789e-06, "loss": 0.0066, "step": 21056 }, { "epoch": 1.122620888201738, "grad_norm": 1.6010686159133911, "learning_rate": 5.7888e-06, "loss": 0.0184, "step": 21057 }, { "epoch": 1.1226742016313909, "grad_norm": 1.0605082511901855, "learning_rate": 5.788600000000001e-06, "loss": 0.0293, "step": 21058 }, { "epoch": 1.1227275150610438, "grad_norm": 2.6672487258911133, "learning_rate": 5.7884e-06, "loss": 0.0445, "step": 21059 }, { "epoch": 1.1227808284906968, "grad_norm": 1.3908640146255493, "learning_rate": 5.788200000000001e-06, "loss": 0.0169, "step": 21060 }, { "epoch": 1.1228341419203498, "grad_norm": 1.968229055404663, "learning_rate": 5.788e-06, "loss": 0.017, "step": 21061 }, { "epoch": 1.1228874553500026, "grad_norm": 2.1478278636932373, "learning_rate": 5.7878e-06, "loss": 0.0346, "step": 21062 }, { "epoch": 1.1229407687796555, "grad_norm": 2.141509771347046, "learning_rate": 5.787600000000001e-06, "loss": 0.0448, "step": 21063 }, { "epoch": 1.1229940822093085, "grad_norm": 3.6992204189300537, "learning_rate": 5.787400000000001e-06, "loss": 0.0844, "step": 21064 }, { "epoch": 1.1230473956389615, "grad_norm": 1.9412239789962769, "learning_rate": 5.7872e-06, "loss": 0.0125, "step": 21065 }, { "epoch": 1.1231007090686145, "grad_norm": 1.7286672592163086, "learning_rate": 5.787000000000001e-06, "loss": 0.0253, "step": 21066 }, { "epoch": 1.1231540224982672, "grad_norm": 2.1160805225372314, "learning_rate": 5.7868e-06, "loss": 0.0318, "step": 21067 }, { "epoch": 1.1232073359279202, "grad_norm": 2.205578327178955, "learning_rate": 5.7866e-06, "loss": 0.0142, "step": 21068 }, { "epoch": 1.1232606493575732, "grad_norm": 2.1204047203063965, "learning_rate": 5.786400000000001e-06, "loss": 0.0225, "step": 21069 }, { "epoch": 1.1233139627872262, "grad_norm": 1.4998201131820679, "learning_rate": 5.786200000000001e-06, "loss": 0.0229, "step": 21070 }, { "epoch": 1.123367276216879, "grad_norm": 2.5331649780273438, "learning_rate": 5.786e-06, "loss": 0.026, "step": 21071 }, { "epoch": 1.123420589646532, "grad_norm": 1.7443156242370605, "learning_rate": 5.785800000000001e-06, "loss": 0.0149, "step": 21072 }, { "epoch": 1.1234739030761849, "grad_norm": 1.3700023889541626, "learning_rate": 5.7856e-06, "loss": 0.0153, "step": 21073 }, { "epoch": 1.1235272165058379, "grad_norm": 1.487587571144104, "learning_rate": 5.785400000000001e-06, "loss": 0.0294, "step": 21074 }, { "epoch": 1.1235805299354908, "grad_norm": 1.542446255683899, "learning_rate": 5.785200000000001e-06, "loss": 0.024, "step": 21075 }, { "epoch": 1.1236338433651436, "grad_norm": 2.289524793624878, "learning_rate": 5.7850000000000005e-06, "loss": 0.018, "step": 21076 }, { "epoch": 1.1236871567947966, "grad_norm": 3.4082181453704834, "learning_rate": 5.7848e-06, "loss": 0.0501, "step": 21077 }, { "epoch": 1.1237404702244496, "grad_norm": 1.2616044282913208, "learning_rate": 5.7846000000000005e-06, "loss": 0.0225, "step": 21078 }, { "epoch": 1.1237937836541025, "grad_norm": 1.6158241033554077, "learning_rate": 5.7844e-06, "loss": 0.0228, "step": 21079 }, { "epoch": 1.1238470970837553, "grad_norm": 3.8807640075683594, "learning_rate": 5.784200000000001e-06, "loss": 0.0252, "step": 21080 }, { "epoch": 1.1239004105134083, "grad_norm": 1.8907078504562378, "learning_rate": 5.784000000000001e-06, "loss": 0.0393, "step": 21081 }, { "epoch": 1.1239537239430613, "grad_norm": 3.0889151096343994, "learning_rate": 5.7838000000000004e-06, "loss": 0.0184, "step": 21082 }, { "epoch": 1.1240070373727142, "grad_norm": 2.4939539432525635, "learning_rate": 5.7836e-06, "loss": 0.0289, "step": 21083 }, { "epoch": 1.1240603508023672, "grad_norm": 1.8790265321731567, "learning_rate": 5.7834e-06, "loss": 0.0305, "step": 21084 }, { "epoch": 1.12411366423202, "grad_norm": 1.6436562538146973, "learning_rate": 5.7832e-06, "loss": 0.0204, "step": 21085 }, { "epoch": 1.124166977661673, "grad_norm": 2.594143867492676, "learning_rate": 5.783000000000001e-06, "loss": 0.0512, "step": 21086 }, { "epoch": 1.124220291091326, "grad_norm": 2.5992040634155273, "learning_rate": 5.782800000000001e-06, "loss": 0.0198, "step": 21087 }, { "epoch": 1.124273604520979, "grad_norm": 0.7992013096809387, "learning_rate": 5.7826e-06, "loss": 0.0062, "step": 21088 }, { "epoch": 1.1243269179506317, "grad_norm": 2.184705972671509, "learning_rate": 5.7824e-06, "loss": 0.0435, "step": 21089 }, { "epoch": 1.1243802313802846, "grad_norm": 1.5172662734985352, "learning_rate": 5.7822e-06, "loss": 0.0199, "step": 21090 }, { "epoch": 1.1244335448099376, "grad_norm": 1.0081075429916382, "learning_rate": 5.782000000000001e-06, "loss": 0.01, "step": 21091 }, { "epoch": 1.1244868582395906, "grad_norm": 1.8006242513656616, "learning_rate": 5.781800000000001e-06, "loss": 0.0153, "step": 21092 }, { "epoch": 1.1245401716692436, "grad_norm": 1.1555722951889038, "learning_rate": 5.781600000000001e-06, "loss": 0.0159, "step": 21093 }, { "epoch": 1.1245934850988963, "grad_norm": 1.9412187337875366, "learning_rate": 5.7814e-06, "loss": 0.0245, "step": 21094 }, { "epoch": 1.1246467985285493, "grad_norm": 2.6596508026123047, "learning_rate": 5.7812e-06, "loss": 0.0391, "step": 21095 }, { "epoch": 1.1247001119582023, "grad_norm": 1.9350950717926025, "learning_rate": 5.781e-06, "loss": 0.0357, "step": 21096 }, { "epoch": 1.1247534253878553, "grad_norm": 1.8547625541687012, "learning_rate": 5.780800000000001e-06, "loss": 0.0347, "step": 21097 }, { "epoch": 1.124806738817508, "grad_norm": 2.8411147594451904, "learning_rate": 5.780600000000001e-06, "loss": 0.0323, "step": 21098 }, { "epoch": 1.124860052247161, "grad_norm": 3.5081899166107178, "learning_rate": 5.780400000000001e-06, "loss": 0.0395, "step": 21099 }, { "epoch": 1.124913365676814, "grad_norm": 1.392502784729004, "learning_rate": 5.7802e-06, "loss": 0.0206, "step": 21100 }, { "epoch": 1.124966679106467, "grad_norm": 1.6106804609298706, "learning_rate": 5.78e-06, "loss": 0.0432, "step": 21101 }, { "epoch": 1.12501999253612, "grad_norm": 0.5947118401527405, "learning_rate": 5.779800000000001e-06, "loss": 0.0063, "step": 21102 }, { "epoch": 1.1250733059657727, "grad_norm": 1.397660732269287, "learning_rate": 5.7796000000000005e-06, "loss": 0.0217, "step": 21103 }, { "epoch": 1.1251266193954257, "grad_norm": 0.8285489678382874, "learning_rate": 5.779400000000001e-06, "loss": 0.0105, "step": 21104 }, { "epoch": 1.1251799328250787, "grad_norm": 1.563584804534912, "learning_rate": 5.7792000000000005e-06, "loss": 0.047, "step": 21105 }, { "epoch": 1.1252332462547316, "grad_norm": 1.721176028251648, "learning_rate": 5.779e-06, "loss": 0.0152, "step": 21106 }, { "epoch": 1.1252865596843844, "grad_norm": 1.7720707654953003, "learning_rate": 5.7788e-06, "loss": 0.0219, "step": 21107 }, { "epoch": 1.1253398731140374, "grad_norm": 3.594891309738159, "learning_rate": 5.778600000000001e-06, "loss": 0.0176, "step": 21108 }, { "epoch": 1.1253931865436904, "grad_norm": 1.5415687561035156, "learning_rate": 5.7784000000000004e-06, "loss": 0.0201, "step": 21109 }, { "epoch": 1.1254464999733433, "grad_norm": 1.2334872484207153, "learning_rate": 5.778200000000001e-06, "loss": 0.0239, "step": 21110 }, { "epoch": 1.125499813402996, "grad_norm": 2.544933319091797, "learning_rate": 5.778e-06, "loss": 0.0275, "step": 21111 }, { "epoch": 1.125553126832649, "grad_norm": 1.2258775234222412, "learning_rate": 5.7778e-06, "loss": 0.0252, "step": 21112 }, { "epoch": 1.125606440262302, "grad_norm": 1.5729864835739136, "learning_rate": 5.777600000000001e-06, "loss": 0.0504, "step": 21113 }, { "epoch": 1.125659753691955, "grad_norm": 2.8477602005004883, "learning_rate": 5.777400000000001e-06, "loss": 0.0457, "step": 21114 }, { "epoch": 1.125713067121608, "grad_norm": 1.0457175970077515, "learning_rate": 5.7772e-06, "loss": 0.0232, "step": 21115 }, { "epoch": 1.1257663805512608, "grad_norm": 1.2827668190002441, "learning_rate": 5.777000000000001e-06, "loss": 0.0182, "step": 21116 }, { "epoch": 1.1258196939809137, "grad_norm": 0.48208415508270264, "learning_rate": 5.7768e-06, "loss": 0.0034, "step": 21117 }, { "epoch": 1.1258730074105667, "grad_norm": 2.1464152336120605, "learning_rate": 5.7766e-06, "loss": 0.0184, "step": 21118 }, { "epoch": 1.1259263208402197, "grad_norm": 1.614132046699524, "learning_rate": 5.776400000000001e-06, "loss": 0.0217, "step": 21119 }, { "epoch": 1.1259796342698727, "grad_norm": 1.929394245147705, "learning_rate": 5.776200000000001e-06, "loss": 0.0163, "step": 21120 }, { "epoch": 1.1260329476995254, "grad_norm": 3.4675633907318115, "learning_rate": 5.776e-06, "loss": 0.037, "step": 21121 }, { "epoch": 1.1260862611291784, "grad_norm": 1.3976941108703613, "learning_rate": 5.775800000000001e-06, "loss": 0.0173, "step": 21122 }, { "epoch": 1.1261395745588314, "grad_norm": 1.1187387704849243, "learning_rate": 5.7756e-06, "loss": 0.0109, "step": 21123 }, { "epoch": 1.1261928879884844, "grad_norm": 2.718684673309326, "learning_rate": 5.775400000000001e-06, "loss": 0.029, "step": 21124 }, { "epoch": 1.1262462014181371, "grad_norm": 1.392027497291565, "learning_rate": 5.775200000000001e-06, "loss": 0.0185, "step": 21125 }, { "epoch": 1.1262995148477901, "grad_norm": 1.0997631549835205, "learning_rate": 5.775000000000001e-06, "loss": 0.0117, "step": 21126 }, { "epoch": 1.126352828277443, "grad_norm": 3.7464590072631836, "learning_rate": 5.7748e-06, "loss": 0.0361, "step": 21127 }, { "epoch": 1.126406141707096, "grad_norm": 3.232159376144409, "learning_rate": 5.774600000000001e-06, "loss": 0.0428, "step": 21128 }, { "epoch": 1.1264594551367488, "grad_norm": 1.9356389045715332, "learning_rate": 5.7744e-06, "loss": 0.0364, "step": 21129 }, { "epoch": 1.1265127685664018, "grad_norm": 1.1742733716964722, "learning_rate": 5.7742000000000006e-06, "loss": 0.0134, "step": 21130 }, { "epoch": 1.1265660819960548, "grad_norm": 0.9190565347671509, "learning_rate": 5.774000000000001e-06, "loss": 0.0109, "step": 21131 }, { "epoch": 1.1266193954257078, "grad_norm": 1.8193544149398804, "learning_rate": 5.7738000000000005e-06, "loss": 0.0525, "step": 21132 }, { "epoch": 1.1266727088553607, "grad_norm": 1.020452857017517, "learning_rate": 5.7736e-06, "loss": 0.0072, "step": 21133 }, { "epoch": 1.1267260222850135, "grad_norm": 1.3792338371276855, "learning_rate": 5.7734000000000005e-06, "loss": 0.0153, "step": 21134 }, { "epoch": 1.1267793357146665, "grad_norm": 1.2344917058944702, "learning_rate": 5.773200000000001e-06, "loss": 0.0266, "step": 21135 }, { "epoch": 1.1268326491443195, "grad_norm": 2.9062209129333496, "learning_rate": 5.7730000000000005e-06, "loss": 0.0145, "step": 21136 }, { "epoch": 1.1268859625739724, "grad_norm": 2.6638760566711426, "learning_rate": 5.772800000000001e-06, "loss": 0.0198, "step": 21137 }, { "epoch": 1.1269392760036254, "grad_norm": 2.003917694091797, "learning_rate": 5.7726000000000004e-06, "loss": 0.0232, "step": 21138 }, { "epoch": 1.1269925894332782, "grad_norm": 1.03122878074646, "learning_rate": 5.7724e-06, "loss": 0.0093, "step": 21139 }, { "epoch": 1.1270459028629312, "grad_norm": 0.7224220037460327, "learning_rate": 5.7722e-06, "loss": 0.0077, "step": 21140 }, { "epoch": 1.1270992162925841, "grad_norm": 1.8231502771377563, "learning_rate": 5.772000000000001e-06, "loss": 0.0302, "step": 21141 }, { "epoch": 1.1271525297222371, "grad_norm": 2.0709495544433594, "learning_rate": 5.7718e-06, "loss": 0.0305, "step": 21142 }, { "epoch": 1.1272058431518899, "grad_norm": 3.420011281967163, "learning_rate": 5.771600000000001e-06, "loss": 0.0475, "step": 21143 }, { "epoch": 1.1272591565815429, "grad_norm": 2.292001247406006, "learning_rate": 5.7714e-06, "loss": 0.0371, "step": 21144 }, { "epoch": 1.1273124700111958, "grad_norm": 1.9209312200546265, "learning_rate": 5.7712e-06, "loss": 0.0313, "step": 21145 }, { "epoch": 1.1273657834408488, "grad_norm": 2.781156539916992, "learning_rate": 5.771e-06, "loss": 0.039, "step": 21146 }, { "epoch": 1.1274190968705016, "grad_norm": 1.9695019721984863, "learning_rate": 5.770800000000001e-06, "loss": 0.0293, "step": 21147 }, { "epoch": 1.1274724103001545, "grad_norm": 0.5383977293968201, "learning_rate": 5.7706e-06, "loss": 0.0061, "step": 21148 }, { "epoch": 1.1275257237298075, "grad_norm": 2.5670101642608643, "learning_rate": 5.770400000000001e-06, "loss": 0.0212, "step": 21149 }, { "epoch": 1.1275790371594605, "grad_norm": 2.04665207862854, "learning_rate": 5.7702e-06, "loss": 0.026, "step": 21150 }, { "epoch": 1.1276323505891135, "grad_norm": 1.576403021812439, "learning_rate": 5.77e-06, "loss": 0.0226, "step": 21151 }, { "epoch": 1.1276856640187662, "grad_norm": 1.1142457723617554, "learning_rate": 5.769800000000001e-06, "loss": 0.0289, "step": 21152 }, { "epoch": 1.1277389774484192, "grad_norm": 1.6069121360778809, "learning_rate": 5.769600000000001e-06, "loss": 0.0337, "step": 21153 }, { "epoch": 1.1277922908780722, "grad_norm": 1.2729557752609253, "learning_rate": 5.7694e-06, "loss": 0.0221, "step": 21154 }, { "epoch": 1.1278456043077252, "grad_norm": 1.3494019508361816, "learning_rate": 5.769200000000001e-06, "loss": 0.0308, "step": 21155 }, { "epoch": 1.1278989177373782, "grad_norm": 1.661346197128296, "learning_rate": 5.769e-06, "loss": 0.0186, "step": 21156 }, { "epoch": 1.127952231167031, "grad_norm": 2.2277309894561768, "learning_rate": 5.7688e-06, "loss": 0.0569, "step": 21157 }, { "epoch": 1.128005544596684, "grad_norm": 1.6716877222061157, "learning_rate": 5.768600000000001e-06, "loss": 0.0242, "step": 21158 }, { "epoch": 1.1280588580263369, "grad_norm": 1.1020793914794922, "learning_rate": 5.7684000000000005e-06, "loss": 0.0221, "step": 21159 }, { "epoch": 1.1281121714559899, "grad_norm": 1.367844820022583, "learning_rate": 5.7682e-06, "loss": 0.0083, "step": 21160 }, { "epoch": 1.1281654848856426, "grad_norm": 3.0490829944610596, "learning_rate": 5.7680000000000005e-06, "loss": 0.0276, "step": 21161 }, { "epoch": 1.1282187983152956, "grad_norm": 1.4568125009536743, "learning_rate": 5.7678e-06, "loss": 0.0252, "step": 21162 }, { "epoch": 1.1282721117449486, "grad_norm": 1.551315426826477, "learning_rate": 5.767600000000001e-06, "loss": 0.064, "step": 21163 }, { "epoch": 1.1283254251746015, "grad_norm": 4.449973106384277, "learning_rate": 5.767400000000001e-06, "loss": 0.0474, "step": 21164 }, { "epoch": 1.1283787386042543, "grad_norm": 4.733387470245361, "learning_rate": 5.7672000000000004e-06, "loss": 0.0465, "step": 21165 }, { "epoch": 1.1284320520339073, "grad_norm": 2.2588682174682617, "learning_rate": 5.767e-06, "loss": 0.0318, "step": 21166 }, { "epoch": 1.1284853654635603, "grad_norm": 4.986124515533447, "learning_rate": 5.7668e-06, "loss": 0.0395, "step": 21167 }, { "epoch": 1.1285386788932132, "grad_norm": 1.842099666595459, "learning_rate": 5.7666e-06, "loss": 0.0292, "step": 21168 }, { "epoch": 1.1285919923228662, "grad_norm": 0.9968410730361938, "learning_rate": 5.766400000000001e-06, "loss": 0.0067, "step": 21169 }, { "epoch": 1.128645305752519, "grad_norm": 2.5255649089813232, "learning_rate": 5.766200000000001e-06, "loss": 0.0624, "step": 21170 }, { "epoch": 1.128698619182172, "grad_norm": 3.811692237854004, "learning_rate": 5.766e-06, "loss": 0.0362, "step": 21171 }, { "epoch": 1.128751932611825, "grad_norm": 0.6909372806549072, "learning_rate": 5.7658e-06, "loss": 0.0101, "step": 21172 }, { "epoch": 1.128805246041478, "grad_norm": 2.240224599838257, "learning_rate": 5.7656e-06, "loss": 0.0338, "step": 21173 }, { "epoch": 1.128858559471131, "grad_norm": 1.2704659700393677, "learning_rate": 5.765400000000001e-06, "loss": 0.0267, "step": 21174 }, { "epoch": 1.1289118729007837, "grad_norm": 2.1015117168426514, "learning_rate": 5.765200000000001e-06, "loss": 0.0235, "step": 21175 }, { "epoch": 1.1289651863304366, "grad_norm": 1.398785948753357, "learning_rate": 5.765000000000001e-06, "loss": 0.0152, "step": 21176 }, { "epoch": 1.1290184997600896, "grad_norm": 2.3283157348632812, "learning_rate": 5.7648e-06, "loss": 0.0314, "step": 21177 }, { "epoch": 1.1290718131897426, "grad_norm": 1.829857349395752, "learning_rate": 5.7646e-06, "loss": 0.058, "step": 21178 }, { "epoch": 1.1291251266193953, "grad_norm": 4.169927597045898, "learning_rate": 5.7644e-06, "loss": 0.036, "step": 21179 }, { "epoch": 1.1291784400490483, "grad_norm": 1.6838892698287964, "learning_rate": 5.764200000000001e-06, "loss": 0.0179, "step": 21180 }, { "epoch": 1.1292317534787013, "grad_norm": 0.7589810490608215, "learning_rate": 5.764000000000001e-06, "loss": 0.0083, "step": 21181 }, { "epoch": 1.1292850669083543, "grad_norm": 1.7522505521774292, "learning_rate": 5.763800000000001e-06, "loss": 0.0131, "step": 21182 }, { "epoch": 1.129338380338007, "grad_norm": 1.5625512599945068, "learning_rate": 5.7636e-06, "loss": 0.0193, "step": 21183 }, { "epoch": 1.12939169376766, "grad_norm": 2.0616533756256104, "learning_rate": 5.7634e-06, "loss": 0.017, "step": 21184 }, { "epoch": 1.129445007197313, "grad_norm": 1.3080493211746216, "learning_rate": 5.763200000000001e-06, "loss": 0.0212, "step": 21185 }, { "epoch": 1.129498320626966, "grad_norm": 1.2633633613586426, "learning_rate": 5.7630000000000006e-06, "loss": 0.0101, "step": 21186 }, { "epoch": 1.129551634056619, "grad_norm": 1.0257866382598877, "learning_rate": 5.762800000000001e-06, "loss": 0.0131, "step": 21187 }, { "epoch": 1.1296049474862717, "grad_norm": 1.5240724086761475, "learning_rate": 5.7626000000000005e-06, "loss": 0.0292, "step": 21188 }, { "epoch": 1.1296582609159247, "grad_norm": 3.4937427043914795, "learning_rate": 5.7624e-06, "loss": 0.0402, "step": 21189 }, { "epoch": 1.1297115743455777, "grad_norm": 2.145368814468384, "learning_rate": 5.7622e-06, "loss": 0.0149, "step": 21190 }, { "epoch": 1.1297648877752307, "grad_norm": 0.5991498231887817, "learning_rate": 5.762000000000001e-06, "loss": 0.0067, "step": 21191 }, { "epoch": 1.1298182012048836, "grad_norm": 2.102771759033203, "learning_rate": 5.7618000000000005e-06, "loss": 0.0378, "step": 21192 }, { "epoch": 1.1298715146345364, "grad_norm": 1.7839633226394653, "learning_rate": 5.761600000000001e-06, "loss": 0.0334, "step": 21193 }, { "epoch": 1.1299248280641894, "grad_norm": 1.2721209526062012, "learning_rate": 5.7614000000000004e-06, "loss": 0.0292, "step": 21194 }, { "epoch": 1.1299781414938423, "grad_norm": 2.4249398708343506, "learning_rate": 5.7612e-06, "loss": 0.02, "step": 21195 }, { "epoch": 1.130031454923495, "grad_norm": 1.4031776189804077, "learning_rate": 5.7609999999999996e-06, "loss": 0.0106, "step": 21196 }, { "epoch": 1.130084768353148, "grad_norm": 2.572873592376709, "learning_rate": 5.760800000000001e-06, "loss": 0.0374, "step": 21197 }, { "epoch": 1.130138081782801, "grad_norm": 0.9095403552055359, "learning_rate": 5.7606e-06, "loss": 0.0038, "step": 21198 }, { "epoch": 1.130191395212454, "grad_norm": 1.025847315788269, "learning_rate": 5.760400000000001e-06, "loss": 0.0176, "step": 21199 }, { "epoch": 1.130244708642107, "grad_norm": 1.3813281059265137, "learning_rate": 5.7602e-06, "loss": 0.0123, "step": 21200 }, { "epoch": 1.1302980220717598, "grad_norm": 2.5192604064941406, "learning_rate": 5.76e-06, "loss": 0.0349, "step": 21201 }, { "epoch": 1.1303513355014128, "grad_norm": 1.185125470161438, "learning_rate": 5.759800000000001e-06, "loss": 0.0086, "step": 21202 }, { "epoch": 1.1304046489310657, "grad_norm": 1.9442882537841797, "learning_rate": 5.759600000000001e-06, "loss": 0.0279, "step": 21203 }, { "epoch": 1.1304579623607187, "grad_norm": 2.997248888015747, "learning_rate": 5.7594e-06, "loss": 0.0537, "step": 21204 }, { "epoch": 1.1305112757903717, "grad_norm": 0.9374167919158936, "learning_rate": 5.759200000000001e-06, "loss": 0.0173, "step": 21205 }, { "epoch": 1.1305645892200245, "grad_norm": 3.5280630588531494, "learning_rate": 5.759e-06, "loss": 0.0682, "step": 21206 }, { "epoch": 1.1306179026496774, "grad_norm": 1.370174527168274, "learning_rate": 5.7588e-06, "loss": 0.0171, "step": 21207 }, { "epoch": 1.1306712160793304, "grad_norm": 2.608302116394043, "learning_rate": 5.758600000000001e-06, "loss": 0.0284, "step": 21208 }, { "epoch": 1.1307245295089834, "grad_norm": 5.441488742828369, "learning_rate": 5.758400000000001e-06, "loss": 0.0379, "step": 21209 }, { "epoch": 1.1307778429386361, "grad_norm": 1.6547290086746216, "learning_rate": 5.7582e-06, "loss": 0.018, "step": 21210 }, { "epoch": 1.1308311563682891, "grad_norm": 2.2900755405426025, "learning_rate": 5.758000000000001e-06, "loss": 0.0271, "step": 21211 }, { "epoch": 1.130884469797942, "grad_norm": 0.48240453004837036, "learning_rate": 5.7578e-06, "loss": 0.0046, "step": 21212 }, { "epoch": 1.130937783227595, "grad_norm": 2.2304232120513916, "learning_rate": 5.7576000000000006e-06, "loss": 0.0205, "step": 21213 }, { "epoch": 1.1309910966572478, "grad_norm": 1.5528769493103027, "learning_rate": 5.757400000000001e-06, "loss": 0.0116, "step": 21214 }, { "epoch": 1.1310444100869008, "grad_norm": 1.2796404361724854, "learning_rate": 5.7572000000000005e-06, "loss": 0.0137, "step": 21215 }, { "epoch": 1.1310977235165538, "grad_norm": 1.767677903175354, "learning_rate": 5.757e-06, "loss": 0.0122, "step": 21216 }, { "epoch": 1.1311510369462068, "grad_norm": 1.9333635568618774, "learning_rate": 5.7568000000000005e-06, "loss": 0.0255, "step": 21217 }, { "epoch": 1.1312043503758598, "grad_norm": 1.3068547248840332, "learning_rate": 5.7566e-06, "loss": 0.0155, "step": 21218 }, { "epoch": 1.1312576638055125, "grad_norm": 0.4151417016983032, "learning_rate": 5.7564000000000005e-06, "loss": 0.0038, "step": 21219 }, { "epoch": 1.1313109772351655, "grad_norm": 2.12382435798645, "learning_rate": 5.756200000000001e-06, "loss": 0.0243, "step": 21220 }, { "epoch": 1.1313642906648185, "grad_norm": 2.8922300338745117, "learning_rate": 5.7560000000000005e-06, "loss": 0.017, "step": 21221 }, { "epoch": 1.1314176040944715, "grad_norm": 1.694080114364624, "learning_rate": 5.7558e-06, "loss": 0.0199, "step": 21222 }, { "epoch": 1.1314709175241244, "grad_norm": 1.452379822731018, "learning_rate": 5.7556000000000004e-06, "loss": 0.0111, "step": 21223 }, { "epoch": 1.1315242309537772, "grad_norm": 2.3000893592834473, "learning_rate": 5.755400000000001e-06, "loss": 0.0188, "step": 21224 }, { "epoch": 1.1315775443834302, "grad_norm": 1.5381848812103271, "learning_rate": 5.7552e-06, "loss": 0.0196, "step": 21225 }, { "epoch": 1.1316308578130831, "grad_norm": 0.704563558101654, "learning_rate": 5.755000000000001e-06, "loss": 0.0038, "step": 21226 }, { "epoch": 1.1316841712427361, "grad_norm": 0.33014893531799316, "learning_rate": 5.7548e-06, "loss": 0.003, "step": 21227 }, { "epoch": 1.1317374846723889, "grad_norm": 1.4905542135238647, "learning_rate": 5.7546e-06, "loss": 0.0161, "step": 21228 }, { "epoch": 1.1317907981020419, "grad_norm": 2.7275638580322266, "learning_rate": 5.7544e-06, "loss": 0.0449, "step": 21229 }, { "epoch": 1.1318441115316948, "grad_norm": 1.5158077478408813, "learning_rate": 5.754200000000001e-06, "loss": 0.0317, "step": 21230 }, { "epoch": 1.1318974249613478, "grad_norm": 3.1952261924743652, "learning_rate": 5.754e-06, "loss": 0.0438, "step": 21231 }, { "epoch": 1.1319507383910006, "grad_norm": 1.7236018180847168, "learning_rate": 5.753800000000001e-06, "loss": 0.024, "step": 21232 }, { "epoch": 1.1320040518206536, "grad_norm": 3.0172502994537354, "learning_rate": 5.7536e-06, "loss": 0.0855, "step": 21233 }, { "epoch": 1.1320573652503065, "grad_norm": 0.9035359025001526, "learning_rate": 5.7534e-06, "loss": 0.0179, "step": 21234 }, { "epoch": 1.1321106786799595, "grad_norm": 0.9703618884086609, "learning_rate": 5.753200000000001e-06, "loss": 0.01, "step": 21235 }, { "epoch": 1.1321639921096125, "grad_norm": 2.5565762519836426, "learning_rate": 5.753000000000001e-06, "loss": 0.0283, "step": 21236 }, { "epoch": 1.1322173055392653, "grad_norm": 2.868300199508667, "learning_rate": 5.7528e-06, "loss": 0.0303, "step": 21237 }, { "epoch": 1.1322706189689182, "grad_norm": 1.6616696119308472, "learning_rate": 5.752600000000001e-06, "loss": 0.0103, "step": 21238 }, { "epoch": 1.1323239323985712, "grad_norm": 2.7764687538146973, "learning_rate": 5.7524e-06, "loss": 0.0142, "step": 21239 }, { "epoch": 1.1323772458282242, "grad_norm": 2.0408577919006348, "learning_rate": 5.7522e-06, "loss": 0.0272, "step": 21240 }, { "epoch": 1.1324305592578772, "grad_norm": 1.977690577507019, "learning_rate": 5.752000000000001e-06, "loss": 0.0199, "step": 21241 }, { "epoch": 1.13248387268753, "grad_norm": 3.193159818649292, "learning_rate": 5.7518000000000006e-06, "loss": 0.034, "step": 21242 }, { "epoch": 1.132537186117183, "grad_norm": 2.5482876300811768, "learning_rate": 5.7516e-06, "loss": 0.0329, "step": 21243 }, { "epoch": 1.1325904995468359, "grad_norm": 2.035717487335205, "learning_rate": 5.7514000000000005e-06, "loss": 0.0234, "step": 21244 }, { "epoch": 1.1326438129764889, "grad_norm": 1.731519103050232, "learning_rate": 5.7512e-06, "loss": 0.0099, "step": 21245 }, { "epoch": 1.1326971264061416, "grad_norm": 1.3249943256378174, "learning_rate": 5.751e-06, "loss": 0.0265, "step": 21246 }, { "epoch": 1.1327504398357946, "grad_norm": 1.7312878370285034, "learning_rate": 5.750800000000001e-06, "loss": 0.0264, "step": 21247 }, { "epoch": 1.1328037532654476, "grad_norm": 2.0065231323242188, "learning_rate": 5.7506000000000005e-06, "loss": 0.0388, "step": 21248 }, { "epoch": 1.1328570666951006, "grad_norm": 1.9116963148117065, "learning_rate": 5.7504e-06, "loss": 0.0301, "step": 21249 }, { "epoch": 1.1329103801247533, "grad_norm": 1.18462336063385, "learning_rate": 5.7502000000000004e-06, "loss": 0.0134, "step": 21250 }, { "epoch": 1.1329636935544063, "grad_norm": 2.1845881938934326, "learning_rate": 5.75e-06, "loss": 0.0312, "step": 21251 }, { "epoch": 1.1330170069840593, "grad_norm": 2.9592626094818115, "learning_rate": 5.749800000000001e-06, "loss": 0.0401, "step": 21252 }, { "epoch": 1.1330703204137123, "grad_norm": 2.2561798095703125, "learning_rate": 5.749600000000001e-06, "loss": 0.0142, "step": 21253 }, { "epoch": 1.1331236338433652, "grad_norm": 1.9668762683868408, "learning_rate": 5.7494e-06, "loss": 0.0069, "step": 21254 }, { "epoch": 1.133176947273018, "grad_norm": 1.205302357673645, "learning_rate": 5.7492e-06, "loss": 0.034, "step": 21255 }, { "epoch": 1.133230260702671, "grad_norm": 1.9020458459854126, "learning_rate": 5.749e-06, "loss": 0.0414, "step": 21256 }, { "epoch": 1.133283574132324, "grad_norm": 3.1663355827331543, "learning_rate": 5.7488e-06, "loss": 0.0459, "step": 21257 }, { "epoch": 1.133336887561977, "grad_norm": 2.162210702896118, "learning_rate": 5.748600000000001e-06, "loss": 0.035, "step": 21258 }, { "epoch": 1.13339020099163, "grad_norm": 2.6608002185821533, "learning_rate": 5.748400000000001e-06, "loss": 0.0288, "step": 21259 }, { "epoch": 1.1334435144212827, "grad_norm": 1.9914251565933228, "learning_rate": 5.7482e-06, "loss": 0.0231, "step": 21260 }, { "epoch": 1.1334968278509356, "grad_norm": 1.8823187351226807, "learning_rate": 5.748e-06, "loss": 0.0328, "step": 21261 }, { "epoch": 1.1335501412805886, "grad_norm": 3.3214564323425293, "learning_rate": 5.7478e-06, "loss": 0.0203, "step": 21262 }, { "epoch": 1.1336034547102416, "grad_norm": 2.377091646194458, "learning_rate": 5.747600000000001e-06, "loss": 0.0428, "step": 21263 }, { "epoch": 1.1336567681398944, "grad_norm": 2.1960701942443848, "learning_rate": 5.747400000000001e-06, "loss": 0.0437, "step": 21264 }, { "epoch": 1.1337100815695473, "grad_norm": 1.9384925365447998, "learning_rate": 5.747200000000001e-06, "loss": 0.028, "step": 21265 }, { "epoch": 1.1337633949992003, "grad_norm": 1.8285950422286987, "learning_rate": 5.747e-06, "loss": 0.0255, "step": 21266 }, { "epoch": 1.1338167084288533, "grad_norm": 1.3608145713806152, "learning_rate": 5.7468e-06, "loss": 0.0126, "step": 21267 }, { "epoch": 1.133870021858506, "grad_norm": 2.310673713684082, "learning_rate": 5.7466e-06, "loss": 0.0235, "step": 21268 }, { "epoch": 1.133923335288159, "grad_norm": 1.6478806734085083, "learning_rate": 5.746400000000001e-06, "loss": 0.0151, "step": 21269 }, { "epoch": 1.133976648717812, "grad_norm": 2.293142080307007, "learning_rate": 5.746200000000001e-06, "loss": 0.0209, "step": 21270 }, { "epoch": 1.134029962147465, "grad_norm": 0.8282652497291565, "learning_rate": 5.7460000000000006e-06, "loss": 0.0094, "step": 21271 }, { "epoch": 1.134083275577118, "grad_norm": 2.29914927482605, "learning_rate": 5.7458e-06, "loss": 0.0305, "step": 21272 }, { "epoch": 1.1341365890067707, "grad_norm": 1.6734788417816162, "learning_rate": 5.7456e-06, "loss": 0.0215, "step": 21273 }, { "epoch": 1.1341899024364237, "grad_norm": 1.6844037771224976, "learning_rate": 5.745400000000001e-06, "loss": 0.0328, "step": 21274 }, { "epoch": 1.1342432158660767, "grad_norm": 1.4467403888702393, "learning_rate": 5.7452000000000005e-06, "loss": 0.0492, "step": 21275 }, { "epoch": 1.1342965292957297, "grad_norm": 1.7538117170333862, "learning_rate": 5.745000000000001e-06, "loss": 0.0311, "step": 21276 }, { "epoch": 1.1343498427253826, "grad_norm": 1.8978564739227295, "learning_rate": 5.7448000000000005e-06, "loss": 0.0267, "step": 21277 }, { "epoch": 1.1344031561550354, "grad_norm": 2.095115900039673, "learning_rate": 5.7446e-06, "loss": 0.0501, "step": 21278 }, { "epoch": 1.1344564695846884, "grad_norm": 0.9146434664726257, "learning_rate": 5.7444e-06, "loss": 0.0231, "step": 21279 }, { "epoch": 1.1345097830143414, "grad_norm": 1.1293576955795288, "learning_rate": 5.744200000000001e-06, "loss": 0.0256, "step": 21280 }, { "epoch": 1.1345630964439943, "grad_norm": 1.538893222808838, "learning_rate": 5.744e-06, "loss": 0.0176, "step": 21281 }, { "epoch": 1.134616409873647, "grad_norm": 1.647316813468933, "learning_rate": 5.743800000000001e-06, "loss": 0.0323, "step": 21282 }, { "epoch": 1.1346697233033, "grad_norm": 0.8802505731582642, "learning_rate": 5.7436e-06, "loss": 0.0159, "step": 21283 }, { "epoch": 1.134723036732953, "grad_norm": 1.4365202188491821, "learning_rate": 5.7434e-06, "loss": 0.0202, "step": 21284 }, { "epoch": 1.134776350162606, "grad_norm": 0.9878151416778564, "learning_rate": 5.743200000000001e-06, "loss": 0.0136, "step": 21285 }, { "epoch": 1.1348296635922588, "grad_norm": 3.0618081092834473, "learning_rate": 5.743000000000001e-06, "loss": 0.0469, "step": 21286 }, { "epoch": 1.1348829770219118, "grad_norm": 0.6925046443939209, "learning_rate": 5.7428e-06, "loss": 0.0065, "step": 21287 }, { "epoch": 1.1349362904515647, "grad_norm": 1.7846111059188843, "learning_rate": 5.742600000000001e-06, "loss": 0.0287, "step": 21288 }, { "epoch": 1.1349896038812177, "grad_norm": 2.354945659637451, "learning_rate": 5.7424e-06, "loss": 0.0173, "step": 21289 }, { "epoch": 1.1350429173108707, "grad_norm": 2.0314295291900635, "learning_rate": 5.7422e-06, "loss": 0.0195, "step": 21290 }, { "epoch": 1.1350962307405235, "grad_norm": 1.4074021577835083, "learning_rate": 5.742000000000001e-06, "loss": 0.0377, "step": 21291 }, { "epoch": 1.1351495441701764, "grad_norm": 3.3148725032806396, "learning_rate": 5.741800000000001e-06, "loss": 0.0322, "step": 21292 }, { "epoch": 1.1352028575998294, "grad_norm": 0.4862357974052429, "learning_rate": 5.7416e-06, "loss": 0.0059, "step": 21293 }, { "epoch": 1.1352561710294824, "grad_norm": 1.3523427248001099, "learning_rate": 5.741400000000001e-06, "loss": 0.0107, "step": 21294 }, { "epoch": 1.1353094844591354, "grad_norm": 0.4409242272377014, "learning_rate": 5.7412e-06, "loss": 0.0054, "step": 21295 }, { "epoch": 1.1353627978887881, "grad_norm": 2.3115036487579346, "learning_rate": 5.741000000000001e-06, "loss": 0.0221, "step": 21296 }, { "epoch": 1.1354161113184411, "grad_norm": 1.9157052040100098, "learning_rate": 5.740800000000001e-06, "loss": 0.0219, "step": 21297 }, { "epoch": 1.135469424748094, "grad_norm": 1.2834118604660034, "learning_rate": 5.7406000000000006e-06, "loss": 0.0489, "step": 21298 }, { "epoch": 1.135522738177747, "grad_norm": 2.6432697772979736, "learning_rate": 5.7404e-06, "loss": 0.0423, "step": 21299 }, { "epoch": 1.1355760516073998, "grad_norm": 2.0591893196105957, "learning_rate": 5.7402000000000005e-06, "loss": 0.0278, "step": 21300 }, { "epoch": 1.1356293650370528, "grad_norm": 1.4232213497161865, "learning_rate": 5.74e-06, "loss": 0.0093, "step": 21301 }, { "epoch": 1.1356826784667058, "grad_norm": 1.4737499952316284, "learning_rate": 5.7398000000000005e-06, "loss": 0.0398, "step": 21302 }, { "epoch": 1.1357359918963588, "grad_norm": 2.654308795928955, "learning_rate": 5.739600000000001e-06, "loss": 0.0409, "step": 21303 }, { "epoch": 1.1357893053260115, "grad_norm": 2.2157158851623535, "learning_rate": 5.7394000000000005e-06, "loss": 0.0314, "step": 21304 }, { "epoch": 1.1358426187556645, "grad_norm": 0.6770294904708862, "learning_rate": 5.7392e-06, "loss": 0.0139, "step": 21305 }, { "epoch": 1.1358959321853175, "grad_norm": 1.4787025451660156, "learning_rate": 5.7390000000000004e-06, "loss": 0.0228, "step": 21306 }, { "epoch": 1.1359492456149705, "grad_norm": 2.8625094890594482, "learning_rate": 5.7388e-06, "loss": 0.018, "step": 21307 }, { "epoch": 1.1360025590446234, "grad_norm": 2.007323980331421, "learning_rate": 5.7386e-06, "loss": 0.0343, "step": 21308 }, { "epoch": 1.1360558724742762, "grad_norm": 1.6102060079574585, "learning_rate": 5.738400000000001e-06, "loss": 0.0361, "step": 21309 }, { "epoch": 1.1361091859039292, "grad_norm": 2.01686692237854, "learning_rate": 5.7382e-06, "loss": 0.0465, "step": 21310 }, { "epoch": 1.1361624993335822, "grad_norm": 2.141538143157959, "learning_rate": 5.738e-06, "loss": 0.0419, "step": 21311 }, { "epoch": 1.1362158127632351, "grad_norm": 1.7732033729553223, "learning_rate": 5.7378e-06, "loss": 0.017, "step": 21312 }, { "epoch": 1.1362691261928881, "grad_norm": 1.2491893768310547, "learning_rate": 5.737600000000001e-06, "loss": 0.0307, "step": 21313 }, { "epoch": 1.1363224396225409, "grad_norm": 2.35307240486145, "learning_rate": 5.7374e-06, "loss": 0.0252, "step": 21314 }, { "epoch": 1.1363757530521938, "grad_norm": 0.6198523640632629, "learning_rate": 5.737200000000001e-06, "loss": 0.005, "step": 21315 }, { "epoch": 1.1364290664818468, "grad_norm": 1.1310557126998901, "learning_rate": 5.737e-06, "loss": 0.0228, "step": 21316 }, { "epoch": 1.1364823799114996, "grad_norm": 2.691710948944092, "learning_rate": 5.7368e-06, "loss": 0.0247, "step": 21317 }, { "epoch": 1.1365356933411526, "grad_norm": 3.2256698608398438, "learning_rate": 5.7366e-06, "loss": 0.054, "step": 21318 }, { "epoch": 1.1365890067708055, "grad_norm": 2.664885997772217, "learning_rate": 5.736400000000001e-06, "loss": 0.0402, "step": 21319 }, { "epoch": 1.1366423202004585, "grad_norm": 1.5380131006240845, "learning_rate": 5.7362e-06, "loss": 0.045, "step": 21320 }, { "epoch": 1.1366956336301115, "grad_norm": 2.2840781211853027, "learning_rate": 5.736000000000001e-06, "loss": 0.0237, "step": 21321 }, { "epoch": 1.1367489470597643, "grad_norm": 0.9316827058792114, "learning_rate": 5.7358e-06, "loss": 0.0187, "step": 21322 }, { "epoch": 1.1368022604894172, "grad_norm": 2.2740399837493896, "learning_rate": 5.7356e-06, "loss": 0.0139, "step": 21323 }, { "epoch": 1.1368555739190702, "grad_norm": 1.7271127700805664, "learning_rate": 5.735400000000001e-06, "loss": 0.0253, "step": 21324 }, { "epoch": 1.1369088873487232, "grad_norm": 1.4200935363769531, "learning_rate": 5.735200000000001e-06, "loss": 0.0203, "step": 21325 }, { "epoch": 1.1369622007783762, "grad_norm": 0.8230589032173157, "learning_rate": 5.735e-06, "loss": 0.008, "step": 21326 }, { "epoch": 1.137015514208029, "grad_norm": 1.7803794145584106, "learning_rate": 5.7348000000000006e-06, "loss": 0.0157, "step": 21327 }, { "epoch": 1.137068827637682, "grad_norm": 1.2512072324752808, "learning_rate": 5.7346e-06, "loss": 0.0118, "step": 21328 }, { "epoch": 1.137122141067335, "grad_norm": 1.5443971157073975, "learning_rate": 5.7344e-06, "loss": 0.03, "step": 21329 }, { "epoch": 1.1371754544969879, "grad_norm": 1.3666595220565796, "learning_rate": 5.734200000000001e-06, "loss": 0.0125, "step": 21330 }, { "epoch": 1.1372287679266406, "grad_norm": 1.7766655683517456, "learning_rate": 5.7340000000000005e-06, "loss": 0.0264, "step": 21331 }, { "epoch": 1.1372820813562936, "grad_norm": 1.7658601999282837, "learning_rate": 5.7338e-06, "loss": 0.0199, "step": 21332 }, { "epoch": 1.1373353947859466, "grad_norm": 2.2509965896606445, "learning_rate": 5.7336000000000005e-06, "loss": 0.0142, "step": 21333 }, { "epoch": 1.1373887082155996, "grad_norm": 1.414336085319519, "learning_rate": 5.7334e-06, "loss": 0.0217, "step": 21334 }, { "epoch": 1.1374420216452523, "grad_norm": 0.6264423131942749, "learning_rate": 5.733200000000001e-06, "loss": 0.0072, "step": 21335 }, { "epoch": 1.1374953350749053, "grad_norm": 2.1506805419921875, "learning_rate": 5.733000000000001e-06, "loss": 0.0123, "step": 21336 }, { "epoch": 1.1375486485045583, "grad_norm": 1.7551778554916382, "learning_rate": 5.7328e-06, "loss": 0.0171, "step": 21337 }, { "epoch": 1.1376019619342113, "grad_norm": 2.937922716140747, "learning_rate": 5.7326e-06, "loss": 0.0506, "step": 21338 }, { "epoch": 1.1376552753638642, "grad_norm": 1.7723665237426758, "learning_rate": 5.7324e-06, "loss": 0.021, "step": 21339 }, { "epoch": 1.137708588793517, "grad_norm": 1.8230547904968262, "learning_rate": 5.7322e-06, "loss": 0.0207, "step": 21340 }, { "epoch": 1.13776190222317, "grad_norm": 1.4504139423370361, "learning_rate": 5.732000000000001e-06, "loss": 0.016, "step": 21341 }, { "epoch": 1.137815215652823, "grad_norm": 1.1472086906433105, "learning_rate": 5.731800000000001e-06, "loss": 0.0277, "step": 21342 }, { "epoch": 1.137868529082476, "grad_norm": 2.231189489364624, "learning_rate": 5.7316e-06, "loss": 0.0252, "step": 21343 }, { "epoch": 1.137921842512129, "grad_norm": 2.1129956245422363, "learning_rate": 5.7314e-06, "loss": 0.0147, "step": 21344 }, { "epoch": 1.1379751559417817, "grad_norm": 0.656905472278595, "learning_rate": 5.7312e-06, "loss": 0.0054, "step": 21345 }, { "epoch": 1.1380284693714346, "grad_norm": 2.488340377807617, "learning_rate": 5.731000000000001e-06, "loss": 0.0533, "step": 21346 }, { "epoch": 1.1380817828010876, "grad_norm": 2.770784616470337, "learning_rate": 5.730800000000001e-06, "loss": 0.0239, "step": 21347 }, { "epoch": 1.1381350962307406, "grad_norm": 2.810025215148926, "learning_rate": 5.730600000000001e-06, "loss": 0.031, "step": 21348 }, { "epoch": 1.1381884096603934, "grad_norm": 2.6420328617095947, "learning_rate": 5.7304e-06, "loss": 0.0754, "step": 21349 }, { "epoch": 1.1382417230900463, "grad_norm": 2.9961097240448, "learning_rate": 5.7302e-06, "loss": 0.0225, "step": 21350 }, { "epoch": 1.1382950365196993, "grad_norm": 4.14261531829834, "learning_rate": 5.73e-06, "loss": 0.0446, "step": 21351 }, { "epoch": 1.1383483499493523, "grad_norm": 1.0717389583587646, "learning_rate": 5.729800000000001e-06, "loss": 0.0208, "step": 21352 }, { "epoch": 1.138401663379005, "grad_norm": 1.2001718282699585, "learning_rate": 5.729600000000001e-06, "loss": 0.007, "step": 21353 }, { "epoch": 1.138454976808658, "grad_norm": 1.2538905143737793, "learning_rate": 5.7294000000000006e-06, "loss": 0.0231, "step": 21354 }, { "epoch": 1.138508290238311, "grad_norm": 2.7591192722320557, "learning_rate": 5.7292e-06, "loss": 0.0325, "step": 21355 }, { "epoch": 1.138561603667964, "grad_norm": 2.081979274749756, "learning_rate": 5.729e-06, "loss": 0.0158, "step": 21356 }, { "epoch": 1.138614917097617, "grad_norm": 1.3460208177566528, "learning_rate": 5.7288e-06, "loss": 0.0279, "step": 21357 }, { "epoch": 1.1386682305272697, "grad_norm": 1.9823813438415527, "learning_rate": 5.7286000000000005e-06, "loss": 0.0411, "step": 21358 }, { "epoch": 1.1387215439569227, "grad_norm": 0.8759903311729431, "learning_rate": 5.728400000000001e-06, "loss": 0.0071, "step": 21359 }, { "epoch": 1.1387748573865757, "grad_norm": 1.4928653240203857, "learning_rate": 5.7282000000000005e-06, "loss": 0.0114, "step": 21360 }, { "epoch": 1.1388281708162287, "grad_norm": 1.127232551574707, "learning_rate": 5.728e-06, "loss": 0.0193, "step": 21361 }, { "epoch": 1.1388814842458816, "grad_norm": 1.344980239868164, "learning_rate": 5.7278e-06, "loss": 0.0352, "step": 21362 }, { "epoch": 1.1389347976755344, "grad_norm": 1.2479954957962036, "learning_rate": 5.727600000000001e-06, "loss": 0.0196, "step": 21363 }, { "epoch": 1.1389881111051874, "grad_norm": 4.464632511138916, "learning_rate": 5.7274000000000004e-06, "loss": 0.0374, "step": 21364 }, { "epoch": 1.1390414245348404, "grad_norm": 1.523848056793213, "learning_rate": 5.727200000000001e-06, "loss": 0.0286, "step": 21365 }, { "epoch": 1.1390947379644933, "grad_norm": 2.5068459510803223, "learning_rate": 5.727e-06, "loss": 0.0205, "step": 21366 }, { "epoch": 1.139148051394146, "grad_norm": 1.1300359964370728, "learning_rate": 5.7268e-06, "loss": 0.0082, "step": 21367 }, { "epoch": 1.139201364823799, "grad_norm": 2.087064743041992, "learning_rate": 5.7265999999999995e-06, "loss": 0.0236, "step": 21368 }, { "epoch": 1.139254678253452, "grad_norm": 3.09273362159729, "learning_rate": 5.726400000000001e-06, "loss": 0.0467, "step": 21369 }, { "epoch": 1.139307991683105, "grad_norm": 4.467824459075928, "learning_rate": 5.7262e-06, "loss": 0.0427, "step": 21370 }, { "epoch": 1.1393613051127578, "grad_norm": 2.9893558025360107, "learning_rate": 5.726000000000001e-06, "loss": 0.038, "step": 21371 }, { "epoch": 1.1394146185424108, "grad_norm": 1.4761450290679932, "learning_rate": 5.7258e-06, "loss": 0.0218, "step": 21372 }, { "epoch": 1.1394679319720638, "grad_norm": 2.1678719520568848, "learning_rate": 5.7256e-06, "loss": 0.0255, "step": 21373 }, { "epoch": 1.1395212454017167, "grad_norm": 3.060633420944214, "learning_rate": 5.725400000000001e-06, "loss": 0.0648, "step": 21374 }, { "epoch": 1.1395745588313697, "grad_norm": 3.7410645484924316, "learning_rate": 5.725200000000001e-06, "loss": 0.039, "step": 21375 }, { "epoch": 1.1396278722610225, "grad_norm": 2.730868101119995, "learning_rate": 5.725e-06, "loss": 0.0483, "step": 21376 }, { "epoch": 1.1396811856906754, "grad_norm": 1.7494248151779175, "learning_rate": 5.724800000000001e-06, "loss": 0.0233, "step": 21377 }, { "epoch": 1.1397344991203284, "grad_norm": 2.741345167160034, "learning_rate": 5.7246e-06, "loss": 0.0284, "step": 21378 }, { "epoch": 1.1397878125499814, "grad_norm": 1.2772517204284668, "learning_rate": 5.7244e-06, "loss": 0.012, "step": 21379 }, { "epoch": 1.1398411259796344, "grad_norm": 4.228011131286621, "learning_rate": 5.724200000000001e-06, "loss": 0.065, "step": 21380 }, { "epoch": 1.1398944394092871, "grad_norm": 1.2990566492080688, "learning_rate": 5.724000000000001e-06, "loss": 0.0135, "step": 21381 }, { "epoch": 1.1399477528389401, "grad_norm": 1.3084977865219116, "learning_rate": 5.7238e-06, "loss": 0.0307, "step": 21382 }, { "epoch": 1.140001066268593, "grad_norm": 1.7377793788909912, "learning_rate": 5.7236000000000006e-06, "loss": 0.0266, "step": 21383 }, { "epoch": 1.140054379698246, "grad_norm": 1.6058927774429321, "learning_rate": 5.7234e-06, "loss": 0.0287, "step": 21384 }, { "epoch": 1.1401076931278988, "grad_norm": 1.4614378213882446, "learning_rate": 5.723200000000001e-06, "loss": 0.0283, "step": 21385 }, { "epoch": 1.1401610065575518, "grad_norm": 2.736335277557373, "learning_rate": 5.723000000000001e-06, "loss": 0.0334, "step": 21386 }, { "epoch": 1.1402143199872048, "grad_norm": 1.3853654861450195, "learning_rate": 5.7228000000000005e-06, "loss": 0.0332, "step": 21387 }, { "epoch": 1.1402676334168578, "grad_norm": 1.4497075080871582, "learning_rate": 5.7226e-06, "loss": 0.0222, "step": 21388 }, { "epoch": 1.1403209468465105, "grad_norm": 2.385460615158081, "learning_rate": 5.7224000000000005e-06, "loss": 0.0334, "step": 21389 }, { "epoch": 1.1403742602761635, "grad_norm": 2.775407552719116, "learning_rate": 5.7222e-06, "loss": 0.044, "step": 21390 }, { "epoch": 1.1404275737058165, "grad_norm": 1.6547040939331055, "learning_rate": 5.722000000000001e-06, "loss": 0.0181, "step": 21391 }, { "epoch": 1.1404808871354695, "grad_norm": 3.23958158493042, "learning_rate": 5.721800000000001e-06, "loss": 0.0255, "step": 21392 }, { "epoch": 1.1405342005651224, "grad_norm": 2.0171144008636475, "learning_rate": 5.7216e-06, "loss": 0.0201, "step": 21393 }, { "epoch": 1.1405875139947752, "grad_norm": 1.6610151529312134, "learning_rate": 5.7214e-06, "loss": 0.0438, "step": 21394 }, { "epoch": 1.1406408274244282, "grad_norm": 2.2521796226501465, "learning_rate": 5.7212e-06, "loss": 0.0298, "step": 21395 }, { "epoch": 1.1406941408540812, "grad_norm": 2.2811920642852783, "learning_rate": 5.721000000000001e-06, "loss": 0.0193, "step": 21396 }, { "epoch": 1.1407474542837341, "grad_norm": 1.3982208967208862, "learning_rate": 5.720800000000001e-06, "loss": 0.0258, "step": 21397 }, { "epoch": 1.1408007677133871, "grad_norm": 2.4320974349975586, "learning_rate": 5.720600000000001e-06, "loss": 0.0366, "step": 21398 }, { "epoch": 1.1408540811430399, "grad_norm": 2.11407732963562, "learning_rate": 5.7204e-06, "loss": 0.0442, "step": 21399 }, { "epoch": 1.1409073945726929, "grad_norm": 1.8162071704864502, "learning_rate": 5.7202e-06, "loss": 0.0312, "step": 21400 }, { "epoch": 1.1409607080023458, "grad_norm": 1.3413950204849243, "learning_rate": 5.72e-06, "loss": 0.0122, "step": 21401 }, { "epoch": 1.1410140214319988, "grad_norm": 0.48686179518699646, "learning_rate": 5.719800000000001e-06, "loss": 0.0053, "step": 21402 }, { "epoch": 1.1410673348616516, "grad_norm": 1.1363353729248047, "learning_rate": 5.719600000000001e-06, "loss": 0.0133, "step": 21403 }, { "epoch": 1.1411206482913046, "grad_norm": 1.0728144645690918, "learning_rate": 5.719400000000001e-06, "loss": 0.0084, "step": 21404 }, { "epoch": 1.1411739617209575, "grad_norm": 0.7949256896972656, "learning_rate": 5.7192e-06, "loss": 0.0172, "step": 21405 }, { "epoch": 1.1412272751506105, "grad_norm": 0.8819055557250977, "learning_rate": 5.719e-06, "loss": 0.0115, "step": 21406 }, { "epoch": 1.1412805885802633, "grad_norm": 1.7692278623580933, "learning_rate": 5.718800000000001e-06, "loss": 0.0416, "step": 21407 }, { "epoch": 1.1413339020099162, "grad_norm": 1.3186733722686768, "learning_rate": 5.718600000000001e-06, "loss": 0.0128, "step": 21408 }, { "epoch": 1.1413872154395692, "grad_norm": 1.5622013807296753, "learning_rate": 5.718400000000001e-06, "loss": 0.017, "step": 21409 }, { "epoch": 1.1414405288692222, "grad_norm": 0.4778570830821991, "learning_rate": 5.718200000000001e-06, "loss": 0.0054, "step": 21410 }, { "epoch": 1.1414938422988752, "grad_norm": 0.8588348627090454, "learning_rate": 5.718e-06, "loss": 0.0167, "step": 21411 }, { "epoch": 1.141547155728528, "grad_norm": 1.9376181364059448, "learning_rate": 5.7178e-06, "loss": 0.0529, "step": 21412 }, { "epoch": 1.141600469158181, "grad_norm": 2.2334516048431396, "learning_rate": 5.717600000000001e-06, "loss": 0.0159, "step": 21413 }, { "epoch": 1.141653782587834, "grad_norm": 0.9984862208366394, "learning_rate": 5.7174000000000005e-06, "loss": 0.0083, "step": 21414 }, { "epoch": 1.1417070960174869, "grad_norm": 1.3013744354248047, "learning_rate": 5.717200000000001e-06, "loss": 0.0092, "step": 21415 }, { "epoch": 1.1417604094471399, "grad_norm": 1.9913667440414429, "learning_rate": 5.7170000000000005e-06, "loss": 0.0387, "step": 21416 }, { "epoch": 1.1418137228767926, "grad_norm": 2.633897304534912, "learning_rate": 5.7168e-06, "loss": 0.0352, "step": 21417 }, { "epoch": 1.1418670363064456, "grad_norm": 2.886894702911377, "learning_rate": 5.7166e-06, "loss": 0.0264, "step": 21418 }, { "epoch": 1.1419203497360986, "grad_norm": 2.0596706867218018, "learning_rate": 5.716400000000001e-06, "loss": 0.0231, "step": 21419 }, { "epoch": 1.1419736631657513, "grad_norm": 1.3101640939712524, "learning_rate": 5.7162000000000004e-06, "loss": 0.0243, "step": 21420 }, { "epoch": 1.1420269765954043, "grad_norm": 1.8896552324295044, "learning_rate": 5.716000000000001e-06, "loss": 0.0315, "step": 21421 }, { "epoch": 1.1420802900250573, "grad_norm": 1.3942071199417114, "learning_rate": 5.7158e-06, "loss": 0.0381, "step": 21422 }, { "epoch": 1.1421336034547103, "grad_norm": 1.462368369102478, "learning_rate": 5.7156e-06, "loss": 0.0155, "step": 21423 }, { "epoch": 1.1421869168843632, "grad_norm": 1.8106391429901123, "learning_rate": 5.715400000000001e-06, "loss": 0.0244, "step": 21424 }, { "epoch": 1.142240230314016, "grad_norm": 3.2590677738189697, "learning_rate": 5.715200000000001e-06, "loss": 0.0374, "step": 21425 }, { "epoch": 1.142293543743669, "grad_norm": 1.221872329711914, "learning_rate": 5.715e-06, "loss": 0.0149, "step": 21426 }, { "epoch": 1.142346857173322, "grad_norm": 3.0035319328308105, "learning_rate": 5.714800000000001e-06, "loss": 0.0127, "step": 21427 }, { "epoch": 1.142400170602975, "grad_norm": 1.7428771257400513, "learning_rate": 5.7146e-06, "loss": 0.0397, "step": 21428 }, { "epoch": 1.142453484032628, "grad_norm": 3.0290944576263428, "learning_rate": 5.7144e-06, "loss": 0.0434, "step": 21429 }, { "epoch": 1.1425067974622807, "grad_norm": 1.7100157737731934, "learning_rate": 5.714200000000001e-06, "loss": 0.0153, "step": 21430 }, { "epoch": 1.1425601108919337, "grad_norm": 2.842963218688965, "learning_rate": 5.714000000000001e-06, "loss": 0.0329, "step": 21431 }, { "epoch": 1.1426134243215866, "grad_norm": 1.2977598905563354, "learning_rate": 5.7138e-06, "loss": 0.0196, "step": 21432 }, { "epoch": 1.1426667377512396, "grad_norm": 2.4136157035827637, "learning_rate": 5.713600000000001e-06, "loss": 0.0132, "step": 21433 }, { "epoch": 1.1427200511808924, "grad_norm": 1.0177894830703735, "learning_rate": 5.7134e-06, "loss": 0.0119, "step": 21434 }, { "epoch": 1.1427733646105454, "grad_norm": 2.7479336261749268, "learning_rate": 5.713200000000001e-06, "loss": 0.0248, "step": 21435 }, { "epoch": 1.1428266780401983, "grad_norm": 0.6720668077468872, "learning_rate": 5.713000000000001e-06, "loss": 0.0053, "step": 21436 }, { "epoch": 1.1428799914698513, "grad_norm": 1.725196361541748, "learning_rate": 5.712800000000001e-06, "loss": 0.0242, "step": 21437 }, { "epoch": 1.142933304899504, "grad_norm": 1.809792160987854, "learning_rate": 5.7126e-06, "loss": 0.0233, "step": 21438 }, { "epoch": 1.142986618329157, "grad_norm": 1.664462685585022, "learning_rate": 5.7124000000000006e-06, "loss": 0.0171, "step": 21439 }, { "epoch": 1.14303993175881, "grad_norm": 1.8914426565170288, "learning_rate": 5.7122e-06, "loss": 0.0265, "step": 21440 }, { "epoch": 1.143093245188463, "grad_norm": 2.432579517364502, "learning_rate": 5.7120000000000005e-06, "loss": 0.0133, "step": 21441 }, { "epoch": 1.143146558618116, "grad_norm": 0.8615626692771912, "learning_rate": 5.711800000000001e-06, "loss": 0.0055, "step": 21442 }, { "epoch": 1.1431998720477687, "grad_norm": 0.9998965263366699, "learning_rate": 5.7116000000000005e-06, "loss": 0.0136, "step": 21443 }, { "epoch": 1.1432531854774217, "grad_norm": 0.9988905191421509, "learning_rate": 5.7114e-06, "loss": 0.0161, "step": 21444 }, { "epoch": 1.1433064989070747, "grad_norm": 1.433827519416809, "learning_rate": 5.7112000000000005e-06, "loss": 0.0088, "step": 21445 }, { "epoch": 1.1433598123367277, "grad_norm": 1.0541096925735474, "learning_rate": 5.711000000000001e-06, "loss": 0.025, "step": 21446 }, { "epoch": 1.1434131257663807, "grad_norm": 0.6784167289733887, "learning_rate": 5.7108000000000004e-06, "loss": 0.0041, "step": 21447 }, { "epoch": 1.1434664391960334, "grad_norm": 1.3243321180343628, "learning_rate": 5.710600000000001e-06, "loss": 0.0273, "step": 21448 }, { "epoch": 1.1435197526256864, "grad_norm": 1.7969696521759033, "learning_rate": 5.7104e-06, "loss": 0.0229, "step": 21449 }, { "epoch": 1.1435730660553394, "grad_norm": 1.0230944156646729, "learning_rate": 5.7102e-06, "loss": 0.0138, "step": 21450 }, { "epoch": 1.1436263794849924, "grad_norm": 0.6520407795906067, "learning_rate": 5.71e-06, "loss": 0.0099, "step": 21451 }, { "epoch": 1.143679692914645, "grad_norm": 1.359133243560791, "learning_rate": 5.709800000000001e-06, "loss": 0.0164, "step": 21452 }, { "epoch": 1.143733006344298, "grad_norm": 2.047534704208374, "learning_rate": 5.7096e-06, "loss": 0.0531, "step": 21453 }, { "epoch": 1.143786319773951, "grad_norm": 1.3634098768234253, "learning_rate": 5.709400000000001e-06, "loss": 0.0112, "step": 21454 }, { "epoch": 1.143839633203604, "grad_norm": 1.0449726581573486, "learning_rate": 5.7092e-06, "loss": 0.0229, "step": 21455 }, { "epoch": 1.1438929466332568, "grad_norm": 1.8823323249816895, "learning_rate": 5.709e-06, "loss": 0.0146, "step": 21456 }, { "epoch": 1.1439462600629098, "grad_norm": 2.5931038856506348, "learning_rate": 5.708800000000001e-06, "loss": 0.0468, "step": 21457 }, { "epoch": 1.1439995734925628, "grad_norm": 2.4637467861175537, "learning_rate": 5.708600000000001e-06, "loss": 0.0226, "step": 21458 }, { "epoch": 1.1440528869222157, "grad_norm": 2.7225000858306885, "learning_rate": 5.7084e-06, "loss": 0.0506, "step": 21459 }, { "epoch": 1.1441062003518687, "grad_norm": 1.8652037382125854, "learning_rate": 5.708200000000001e-06, "loss": 0.0163, "step": 21460 }, { "epoch": 1.1441595137815215, "grad_norm": 1.3578007221221924, "learning_rate": 5.708e-06, "loss": 0.0176, "step": 21461 }, { "epoch": 1.1442128272111745, "grad_norm": 1.540808081626892, "learning_rate": 5.7078e-06, "loss": 0.0179, "step": 21462 }, { "epoch": 1.1442661406408274, "grad_norm": 1.760642170906067, "learning_rate": 5.707600000000001e-06, "loss": 0.0249, "step": 21463 }, { "epoch": 1.1443194540704804, "grad_norm": 2.729182004928589, "learning_rate": 5.707400000000001e-06, "loss": 0.0402, "step": 21464 }, { "epoch": 1.1443727675001334, "grad_norm": 1.6350266933441162, "learning_rate": 5.7072e-06, "loss": 0.0259, "step": 21465 }, { "epoch": 1.1444260809297861, "grad_norm": 0.950531005859375, "learning_rate": 5.707000000000001e-06, "loss": 0.0155, "step": 21466 }, { "epoch": 1.1444793943594391, "grad_norm": 2.564906120300293, "learning_rate": 5.7068e-06, "loss": 0.0397, "step": 21467 }, { "epoch": 1.144532707789092, "grad_norm": 2.1019723415374756, "learning_rate": 5.7066e-06, "loss": 0.039, "step": 21468 }, { "epoch": 1.144586021218745, "grad_norm": 2.620669364929199, "learning_rate": 5.706400000000001e-06, "loss": 0.0329, "step": 21469 }, { "epoch": 1.1446393346483978, "grad_norm": 2.83103084564209, "learning_rate": 5.7062000000000005e-06, "loss": 0.0254, "step": 21470 }, { "epoch": 1.1446926480780508, "grad_norm": 1.7760164737701416, "learning_rate": 5.706e-06, "loss": 0.0209, "step": 21471 }, { "epoch": 1.1447459615077038, "grad_norm": 1.8453985452651978, "learning_rate": 5.7058000000000005e-06, "loss": 0.0238, "step": 21472 }, { "epoch": 1.1447992749373568, "grad_norm": 1.6812164783477783, "learning_rate": 5.7056e-06, "loss": 0.0333, "step": 21473 }, { "epoch": 1.1448525883670095, "grad_norm": 1.7817580699920654, "learning_rate": 5.705400000000001e-06, "loss": 0.0481, "step": 21474 }, { "epoch": 1.1449059017966625, "grad_norm": 1.599800705909729, "learning_rate": 5.705200000000001e-06, "loss": 0.0307, "step": 21475 }, { "epoch": 1.1449592152263155, "grad_norm": 1.934181571006775, "learning_rate": 5.7050000000000004e-06, "loss": 0.0407, "step": 21476 }, { "epoch": 1.1450125286559685, "grad_norm": 2.5621721744537354, "learning_rate": 5.7048e-06, "loss": 0.032, "step": 21477 }, { "epoch": 1.1450658420856215, "grad_norm": 2.323587656021118, "learning_rate": 5.7046e-06, "loss": 0.0816, "step": 21478 }, { "epoch": 1.1451191555152742, "grad_norm": 1.1279442310333252, "learning_rate": 5.7044e-06, "loss": 0.0123, "step": 21479 }, { "epoch": 1.1451724689449272, "grad_norm": 1.2873908281326294, "learning_rate": 5.704200000000001e-06, "loss": 0.0302, "step": 21480 }, { "epoch": 1.1452257823745802, "grad_norm": 2.496697425842285, "learning_rate": 5.704000000000001e-06, "loss": 0.038, "step": 21481 }, { "epoch": 1.1452790958042331, "grad_norm": 1.7464796304702759, "learning_rate": 5.7038e-06, "loss": 0.0439, "step": 21482 }, { "epoch": 1.1453324092338861, "grad_norm": 1.1886365413665771, "learning_rate": 5.7036e-06, "loss": 0.0157, "step": 21483 }, { "epoch": 1.1453857226635389, "grad_norm": 1.7205007076263428, "learning_rate": 5.7034e-06, "loss": 0.0498, "step": 21484 }, { "epoch": 1.1454390360931919, "grad_norm": 2.0375709533691406, "learning_rate": 5.703200000000001e-06, "loss": 0.0337, "step": 21485 }, { "epoch": 1.1454923495228448, "grad_norm": 1.6546329259872437, "learning_rate": 5.703000000000001e-06, "loss": 0.0308, "step": 21486 }, { "epoch": 1.1455456629524978, "grad_norm": 1.9417803287506104, "learning_rate": 5.702800000000001e-06, "loss": 0.023, "step": 21487 }, { "epoch": 1.1455989763821506, "grad_norm": 0.6219380497932434, "learning_rate": 5.7026e-06, "loss": 0.0071, "step": 21488 }, { "epoch": 1.1456522898118036, "grad_norm": 2.506626605987549, "learning_rate": 5.7024e-06, "loss": 0.0361, "step": 21489 }, { "epoch": 1.1457056032414565, "grad_norm": 1.2359737157821655, "learning_rate": 5.7022e-06, "loss": 0.0235, "step": 21490 }, { "epoch": 1.1457589166711095, "grad_norm": 1.2862156629562378, "learning_rate": 5.702000000000001e-06, "loss": 0.0295, "step": 21491 }, { "epoch": 1.1458122301007623, "grad_norm": 2.710688591003418, "learning_rate": 5.701800000000001e-06, "loss": 0.046, "step": 21492 }, { "epoch": 1.1458655435304153, "grad_norm": 2.038195848464966, "learning_rate": 5.701600000000001e-06, "loss": 0.0314, "step": 21493 }, { "epoch": 1.1459188569600682, "grad_norm": 2.0995078086853027, "learning_rate": 5.7014e-06, "loss": 0.0415, "step": 21494 }, { "epoch": 1.1459721703897212, "grad_norm": 1.74305260181427, "learning_rate": 5.7012e-06, "loss": 0.0224, "step": 21495 }, { "epoch": 1.1460254838193742, "grad_norm": 3.4029486179351807, "learning_rate": 5.701000000000001e-06, "loss": 0.0289, "step": 21496 }, { "epoch": 1.146078797249027, "grad_norm": 2.224487543106079, "learning_rate": 5.7008000000000005e-06, "loss": 0.0338, "step": 21497 }, { "epoch": 1.14613211067868, "grad_norm": 1.8567620515823364, "learning_rate": 5.700600000000001e-06, "loss": 0.0167, "step": 21498 }, { "epoch": 1.146185424108333, "grad_norm": 2.363816976547241, "learning_rate": 5.7004000000000005e-06, "loss": 0.0471, "step": 21499 }, { "epoch": 1.1462387375379859, "grad_norm": 1.3940305709838867, "learning_rate": 5.7002e-06, "loss": 0.0181, "step": 21500 }, { "epoch": 1.1462920509676389, "grad_norm": 0.33373939990997314, "learning_rate": 5.7e-06, "loss": 0.0046, "step": 21501 }, { "epoch": 1.1463453643972916, "grad_norm": 2.0125961303710938, "learning_rate": 5.699800000000001e-06, "loss": 0.0341, "step": 21502 }, { "epoch": 1.1463986778269446, "grad_norm": 2.3439650535583496, "learning_rate": 5.6996000000000005e-06, "loss": 0.0339, "step": 21503 }, { "epoch": 1.1464519912565976, "grad_norm": 1.794111728668213, "learning_rate": 5.699400000000001e-06, "loss": 0.0397, "step": 21504 }, { "epoch": 1.1465053046862506, "grad_norm": 1.4279963970184326, "learning_rate": 5.6992e-06, "loss": 0.0226, "step": 21505 }, { "epoch": 1.1465586181159033, "grad_norm": 1.2897613048553467, "learning_rate": 5.699e-06, "loss": 0.0315, "step": 21506 }, { "epoch": 1.1466119315455563, "grad_norm": 2.862720251083374, "learning_rate": 5.698800000000001e-06, "loss": 0.0564, "step": 21507 }, { "epoch": 1.1466652449752093, "grad_norm": 0.7662340402603149, "learning_rate": 5.698600000000001e-06, "loss": 0.0084, "step": 21508 }, { "epoch": 1.1467185584048623, "grad_norm": 1.9205743074417114, "learning_rate": 5.6984e-06, "loss": 0.0265, "step": 21509 }, { "epoch": 1.146771871834515, "grad_norm": 1.4138091802597046, "learning_rate": 5.698200000000001e-06, "loss": 0.0287, "step": 21510 }, { "epoch": 1.146825185264168, "grad_norm": 1.62980318069458, "learning_rate": 5.698e-06, "loss": 0.0274, "step": 21511 }, { "epoch": 1.146878498693821, "grad_norm": 0.8350797295570374, "learning_rate": 5.6978e-06, "loss": 0.0166, "step": 21512 }, { "epoch": 1.146931812123474, "grad_norm": 0.8393279314041138, "learning_rate": 5.697600000000001e-06, "loss": 0.0091, "step": 21513 }, { "epoch": 1.146985125553127, "grad_norm": 0.8010069131851196, "learning_rate": 5.697400000000001e-06, "loss": 0.0186, "step": 21514 }, { "epoch": 1.1470384389827797, "grad_norm": 1.1576474905014038, "learning_rate": 5.6972e-06, "loss": 0.0215, "step": 21515 }, { "epoch": 1.1470917524124327, "grad_norm": 1.9110527038574219, "learning_rate": 5.697000000000001e-06, "loss": 0.0192, "step": 21516 }, { "epoch": 1.1471450658420856, "grad_norm": 2.344271421432495, "learning_rate": 5.6968e-06, "loss": 0.0327, "step": 21517 }, { "epoch": 1.1471983792717386, "grad_norm": 2.4030559062957764, "learning_rate": 5.696600000000001e-06, "loss": 0.0341, "step": 21518 }, { "epoch": 1.1472516927013916, "grad_norm": 3.114860773086548, "learning_rate": 5.696400000000001e-06, "loss": 0.018, "step": 21519 }, { "epoch": 1.1473050061310444, "grad_norm": 0.5239466428756714, "learning_rate": 5.696200000000001e-06, "loss": 0.0067, "step": 21520 }, { "epoch": 1.1473583195606973, "grad_norm": 1.4592000246047974, "learning_rate": 5.696e-06, "loss": 0.0123, "step": 21521 }, { "epoch": 1.1474116329903503, "grad_norm": 1.8568010330200195, "learning_rate": 5.695800000000001e-06, "loss": 0.0269, "step": 21522 }, { "epoch": 1.1474649464200033, "grad_norm": 1.5573333501815796, "learning_rate": 5.6956e-06, "loss": 0.0183, "step": 21523 }, { "epoch": 1.147518259849656, "grad_norm": 0.9243534803390503, "learning_rate": 5.6954000000000006e-06, "loss": 0.0243, "step": 21524 }, { "epoch": 1.147571573279309, "grad_norm": 2.279472827911377, "learning_rate": 5.695200000000001e-06, "loss": 0.0301, "step": 21525 }, { "epoch": 1.147624886708962, "grad_norm": 2.4745254516601562, "learning_rate": 5.6950000000000005e-06, "loss": 0.0386, "step": 21526 }, { "epoch": 1.147678200138615, "grad_norm": 0.9379280805587769, "learning_rate": 5.6948e-06, "loss": 0.0098, "step": 21527 }, { "epoch": 1.1477315135682677, "grad_norm": 1.7540574073791504, "learning_rate": 5.6946000000000005e-06, "loss": 0.0319, "step": 21528 }, { "epoch": 1.1477848269979207, "grad_norm": 2.2649805545806885, "learning_rate": 5.6944e-06, "loss": 0.042, "step": 21529 }, { "epoch": 1.1478381404275737, "grad_norm": 0.8906590342521667, "learning_rate": 5.6942000000000005e-06, "loss": 0.0212, "step": 21530 }, { "epoch": 1.1478914538572267, "grad_norm": 1.5154832601547241, "learning_rate": 5.694000000000001e-06, "loss": 0.0102, "step": 21531 }, { "epoch": 1.1479447672868797, "grad_norm": 2.7603352069854736, "learning_rate": 5.6938000000000004e-06, "loss": 0.043, "step": 21532 }, { "epoch": 1.1479980807165324, "grad_norm": 1.0868079662322998, "learning_rate": 5.6936e-06, "loss": 0.0117, "step": 21533 }, { "epoch": 1.1480513941461854, "grad_norm": 2.3559553623199463, "learning_rate": 5.6934e-06, "loss": 0.0284, "step": 21534 }, { "epoch": 1.1481047075758384, "grad_norm": 1.7930512428283691, "learning_rate": 5.693200000000001e-06, "loss": 0.016, "step": 21535 }, { "epoch": 1.1481580210054914, "grad_norm": 1.2076103687286377, "learning_rate": 5.693e-06, "loss": 0.0174, "step": 21536 }, { "epoch": 1.1482113344351443, "grad_norm": 1.7139298915863037, "learning_rate": 5.692800000000001e-06, "loss": 0.0114, "step": 21537 }, { "epoch": 1.148264647864797, "grad_norm": 2.0893256664276123, "learning_rate": 5.6926e-06, "loss": 0.0239, "step": 21538 }, { "epoch": 1.14831796129445, "grad_norm": 3.5364110469818115, "learning_rate": 5.6924e-06, "loss": 0.0534, "step": 21539 }, { "epoch": 1.148371274724103, "grad_norm": 4.8002495765686035, "learning_rate": 5.6922e-06, "loss": 0.0795, "step": 21540 }, { "epoch": 1.1484245881537558, "grad_norm": 2.6515557765960693, "learning_rate": 5.692000000000001e-06, "loss": 0.0263, "step": 21541 }, { "epoch": 1.1484779015834088, "grad_norm": 1.4736357927322388, "learning_rate": 5.6918e-06, "loss": 0.0163, "step": 21542 }, { "epoch": 1.1485312150130618, "grad_norm": 1.4512110948562622, "learning_rate": 5.691600000000001e-06, "loss": 0.0152, "step": 21543 }, { "epoch": 1.1485845284427147, "grad_norm": 1.2320045232772827, "learning_rate": 5.6914e-06, "loss": 0.0123, "step": 21544 }, { "epoch": 1.1486378418723677, "grad_norm": 1.662153720855713, "learning_rate": 5.6912e-06, "loss": 0.0214, "step": 21545 }, { "epoch": 1.1486911553020205, "grad_norm": 2.6429431438446045, "learning_rate": 5.691000000000001e-06, "loss": 0.0515, "step": 21546 }, { "epoch": 1.1487444687316735, "grad_norm": 1.2362797260284424, "learning_rate": 5.690800000000001e-06, "loss": 0.0126, "step": 21547 }, { "epoch": 1.1487977821613264, "grad_norm": 3.6858627796173096, "learning_rate": 5.6906e-06, "loss": 0.0519, "step": 21548 }, { "epoch": 1.1488510955909794, "grad_norm": 1.079882264137268, "learning_rate": 5.690400000000001e-06, "loss": 0.0191, "step": 21549 }, { "epoch": 1.1489044090206324, "grad_norm": 7.147906303405762, "learning_rate": 5.6902e-06, "loss": 0.041, "step": 21550 }, { "epoch": 1.1489577224502852, "grad_norm": 2.645134925842285, "learning_rate": 5.69e-06, "loss": 0.0382, "step": 21551 }, { "epoch": 1.1490110358799381, "grad_norm": 1.3712778091430664, "learning_rate": 5.689800000000001e-06, "loss": 0.0153, "step": 21552 }, { "epoch": 1.1490643493095911, "grad_norm": 1.875948429107666, "learning_rate": 5.6896000000000005e-06, "loss": 0.0187, "step": 21553 }, { "epoch": 1.149117662739244, "grad_norm": 1.1516400575637817, "learning_rate": 5.6894e-06, "loss": 0.0147, "step": 21554 }, { "epoch": 1.1491709761688969, "grad_norm": 2.946784257888794, "learning_rate": 5.6892000000000005e-06, "loss": 0.027, "step": 21555 }, { "epoch": 1.1492242895985498, "grad_norm": 1.0615919828414917, "learning_rate": 5.689e-06, "loss": 0.0157, "step": 21556 }, { "epoch": 1.1492776030282028, "grad_norm": 2.1576201915740967, "learning_rate": 5.688800000000001e-06, "loss": 0.0271, "step": 21557 }, { "epoch": 1.1493309164578558, "grad_norm": 3.1306629180908203, "learning_rate": 5.688600000000001e-06, "loss": 0.0297, "step": 21558 }, { "epoch": 1.1493842298875085, "grad_norm": 1.490830898284912, "learning_rate": 5.6884000000000005e-06, "loss": 0.0151, "step": 21559 }, { "epoch": 1.1494375433171615, "grad_norm": 1.1133630275726318, "learning_rate": 5.6882e-06, "loss": 0.0159, "step": 21560 }, { "epoch": 1.1494908567468145, "grad_norm": 2.389625310897827, "learning_rate": 5.6880000000000004e-06, "loss": 0.0284, "step": 21561 }, { "epoch": 1.1495441701764675, "grad_norm": 1.4997543096542358, "learning_rate": 5.6878e-06, "loss": 0.0136, "step": 21562 }, { "epoch": 1.1495974836061205, "grad_norm": 2.4535865783691406, "learning_rate": 5.687600000000001e-06, "loss": 0.0321, "step": 21563 }, { "epoch": 1.1496507970357732, "grad_norm": 2.0842182636260986, "learning_rate": 5.687400000000001e-06, "loss": 0.0322, "step": 21564 }, { "epoch": 1.1497041104654262, "grad_norm": 1.2868489027023315, "learning_rate": 5.6872e-06, "loss": 0.028, "step": 21565 }, { "epoch": 1.1497574238950792, "grad_norm": 2.36592173576355, "learning_rate": 5.687e-06, "loss": 0.0317, "step": 21566 }, { "epoch": 1.1498107373247322, "grad_norm": 1.196645736694336, "learning_rate": 5.6868e-06, "loss": 0.0106, "step": 21567 }, { "epoch": 1.1498640507543851, "grad_norm": 1.1665971279144287, "learning_rate": 5.686600000000001e-06, "loss": 0.0109, "step": 21568 }, { "epoch": 1.149917364184038, "grad_norm": 1.5131328105926514, "learning_rate": 5.686400000000001e-06, "loss": 0.014, "step": 21569 }, { "epoch": 1.1499706776136909, "grad_norm": 2.171686887741089, "learning_rate": 5.686200000000001e-06, "loss": 0.0376, "step": 21570 }, { "epoch": 1.1500239910433439, "grad_norm": 1.6185452938079834, "learning_rate": 5.686e-06, "loss": 0.0209, "step": 21571 }, { "epoch": 1.1500773044729968, "grad_norm": 1.8738856315612793, "learning_rate": 5.6858e-06, "loss": 0.0126, "step": 21572 }, { "epoch": 1.1501306179026496, "grad_norm": 2.2981410026550293, "learning_rate": 5.6856e-06, "loss": 0.0311, "step": 21573 }, { "epoch": 1.1501839313323026, "grad_norm": 0.764021098613739, "learning_rate": 5.685400000000001e-06, "loss": 0.0056, "step": 21574 }, { "epoch": 1.1502372447619555, "grad_norm": 2.966364860534668, "learning_rate": 5.685200000000001e-06, "loss": 0.0207, "step": 21575 }, { "epoch": 1.1502905581916085, "grad_norm": 1.8125354051589966, "learning_rate": 5.685000000000001e-06, "loss": 0.0229, "step": 21576 }, { "epoch": 1.1503438716212613, "grad_norm": 2.103813886642456, "learning_rate": 5.6848e-06, "loss": 0.0278, "step": 21577 }, { "epoch": 1.1503971850509143, "grad_norm": 1.2976629734039307, "learning_rate": 5.6846e-06, "loss": 0.0095, "step": 21578 }, { "epoch": 1.1504504984805672, "grad_norm": 2.256671190261841, "learning_rate": 5.6844e-06, "loss": 0.0384, "step": 21579 }, { "epoch": 1.1505038119102202, "grad_norm": 1.4471255540847778, "learning_rate": 5.6842000000000006e-06, "loss": 0.0197, "step": 21580 }, { "epoch": 1.1505571253398732, "grad_norm": 1.8086824417114258, "learning_rate": 5.684000000000001e-06, "loss": 0.0155, "step": 21581 }, { "epoch": 1.150610438769526, "grad_norm": 1.4039146900177002, "learning_rate": 5.6838000000000005e-06, "loss": 0.007, "step": 21582 }, { "epoch": 1.150663752199179, "grad_norm": 1.3176954984664917, "learning_rate": 5.6836e-06, "loss": 0.0118, "step": 21583 }, { "epoch": 1.150717065628832, "grad_norm": 1.960909366607666, "learning_rate": 5.6834e-06, "loss": 0.0263, "step": 21584 }, { "epoch": 1.150770379058485, "grad_norm": 0.7089688181877136, "learning_rate": 5.683200000000001e-06, "loss": 0.0048, "step": 21585 }, { "epoch": 1.1508236924881379, "grad_norm": 1.61549711227417, "learning_rate": 5.6830000000000005e-06, "loss": 0.0139, "step": 21586 }, { "epoch": 1.1508770059177906, "grad_norm": 1.9574880599975586, "learning_rate": 5.682800000000001e-06, "loss": 0.0274, "step": 21587 }, { "epoch": 1.1509303193474436, "grad_norm": 1.0714002847671509, "learning_rate": 5.6826000000000004e-06, "loss": 0.0096, "step": 21588 }, { "epoch": 1.1509836327770966, "grad_norm": 1.2084736824035645, "learning_rate": 5.6824e-06, "loss": 0.0075, "step": 21589 }, { "epoch": 1.1510369462067496, "grad_norm": 1.9342447519302368, "learning_rate": 5.6821999999999996e-06, "loss": 0.0259, "step": 21590 }, { "epoch": 1.1510902596364023, "grad_norm": 0.7319905757904053, "learning_rate": 5.682000000000001e-06, "loss": 0.0056, "step": 21591 }, { "epoch": 1.1511435730660553, "grad_norm": 3.6658945083618164, "learning_rate": 5.6818e-06, "loss": 0.0464, "step": 21592 }, { "epoch": 1.1511968864957083, "grad_norm": 2.657644271850586, "learning_rate": 5.681600000000001e-06, "loss": 0.03, "step": 21593 }, { "epoch": 1.1512501999253613, "grad_norm": 1.0925320386886597, "learning_rate": 5.6814e-06, "loss": 0.0064, "step": 21594 }, { "epoch": 1.151303513355014, "grad_norm": 2.5719525814056396, "learning_rate": 5.6812e-06, "loss": 0.0379, "step": 21595 }, { "epoch": 1.151356826784667, "grad_norm": 2.132546901702881, "learning_rate": 5.681000000000001e-06, "loss": 0.0127, "step": 21596 }, { "epoch": 1.15141014021432, "grad_norm": 1.349038004875183, "learning_rate": 5.680800000000001e-06, "loss": 0.0369, "step": 21597 }, { "epoch": 1.151463453643973, "grad_norm": 2.623020887374878, "learning_rate": 5.6806e-06, "loss": 0.0558, "step": 21598 }, { "epoch": 1.151516767073626, "grad_norm": 1.1573747396469116, "learning_rate": 5.680400000000001e-06, "loss": 0.0246, "step": 21599 }, { "epoch": 1.1515700805032787, "grad_norm": 1.1772526502609253, "learning_rate": 5.6802e-06, "loss": 0.0265, "step": 21600 }, { "epoch": 1.1516233939329317, "grad_norm": 1.8727822303771973, "learning_rate": 5.68e-06, "loss": 0.0255, "step": 21601 }, { "epoch": 1.1516767073625847, "grad_norm": 2.7616281509399414, "learning_rate": 5.679800000000001e-06, "loss": 0.0257, "step": 21602 }, { "epoch": 1.1517300207922376, "grad_norm": 3.256211280822754, "learning_rate": 5.679600000000001e-06, "loss": 0.049, "step": 21603 }, { "epoch": 1.1517833342218906, "grad_norm": 1.7630088329315186, "learning_rate": 5.6794e-06, "loss": 0.0137, "step": 21604 }, { "epoch": 1.1518366476515434, "grad_norm": 2.585395336151123, "learning_rate": 5.679200000000001e-06, "loss": 0.0152, "step": 21605 }, { "epoch": 1.1518899610811963, "grad_norm": 2.325392007827759, "learning_rate": 5.679e-06, "loss": 0.0304, "step": 21606 }, { "epoch": 1.1519432745108493, "grad_norm": 0.9518382549285889, "learning_rate": 5.678800000000001e-06, "loss": 0.0059, "step": 21607 }, { "epoch": 1.1519965879405023, "grad_norm": 0.49462878704071045, "learning_rate": 5.678600000000001e-06, "loss": 0.0046, "step": 21608 }, { "epoch": 1.152049901370155, "grad_norm": 1.409106731414795, "learning_rate": 5.6784000000000006e-06, "loss": 0.0099, "step": 21609 }, { "epoch": 1.152103214799808, "grad_norm": 4.524837017059326, "learning_rate": 5.6782e-06, "loss": 0.0554, "step": 21610 }, { "epoch": 1.152156528229461, "grad_norm": 2.8742856979370117, "learning_rate": 5.6780000000000005e-06, "loss": 0.0235, "step": 21611 }, { "epoch": 1.152209841659114, "grad_norm": 1.9286770820617676, "learning_rate": 5.6778e-06, "loss": 0.0106, "step": 21612 }, { "epoch": 1.1522631550887668, "grad_norm": 1.8942099809646606, "learning_rate": 5.6776000000000005e-06, "loss": 0.0197, "step": 21613 }, { "epoch": 1.1523164685184197, "grad_norm": 2.3602871894836426, "learning_rate": 5.677400000000001e-06, "loss": 0.023, "step": 21614 }, { "epoch": 1.1523697819480727, "grad_norm": 3.1349520683288574, "learning_rate": 5.6772000000000005e-06, "loss": 0.0273, "step": 21615 }, { "epoch": 1.1524230953777257, "grad_norm": 1.7318365573883057, "learning_rate": 5.677e-06, "loss": 0.0281, "step": 21616 }, { "epoch": 1.1524764088073787, "grad_norm": 2.553168535232544, "learning_rate": 5.6768000000000004e-06, "loss": 0.032, "step": 21617 }, { "epoch": 1.1525297222370314, "grad_norm": 1.2981979846954346, "learning_rate": 5.676600000000001e-06, "loss": 0.0187, "step": 21618 }, { "epoch": 1.1525830356666844, "grad_norm": 1.5811717510223389, "learning_rate": 5.6764e-06, "loss": 0.0295, "step": 21619 }, { "epoch": 1.1526363490963374, "grad_norm": 1.59185791015625, "learning_rate": 5.676200000000001e-06, "loss": 0.0352, "step": 21620 }, { "epoch": 1.1526896625259904, "grad_norm": 2.6536378860473633, "learning_rate": 5.676e-06, "loss": 0.0539, "step": 21621 }, { "epoch": 1.1527429759556433, "grad_norm": 3.0997657775878906, "learning_rate": 5.6758e-06, "loss": 0.0513, "step": 21622 }, { "epoch": 1.152796289385296, "grad_norm": 2.2123477458953857, "learning_rate": 5.6756e-06, "loss": 0.0375, "step": 21623 }, { "epoch": 1.152849602814949, "grad_norm": 2.7224843502044678, "learning_rate": 5.675400000000001e-06, "loss": 0.0306, "step": 21624 }, { "epoch": 1.152902916244602, "grad_norm": 4.811305522918701, "learning_rate": 5.6752e-06, "loss": 0.0338, "step": 21625 }, { "epoch": 1.152956229674255, "grad_norm": 1.1796479225158691, "learning_rate": 5.675000000000001e-06, "loss": 0.0201, "step": 21626 }, { "epoch": 1.1530095431039078, "grad_norm": 1.676823616027832, "learning_rate": 5.6748e-06, "loss": 0.0283, "step": 21627 }, { "epoch": 1.1530628565335608, "grad_norm": 1.7109529972076416, "learning_rate": 5.6746e-06, "loss": 0.0091, "step": 21628 }, { "epoch": 1.1531161699632138, "grad_norm": 2.2834787368774414, "learning_rate": 5.674400000000001e-06, "loss": 0.0228, "step": 21629 }, { "epoch": 1.1531694833928667, "grad_norm": 4.7921833992004395, "learning_rate": 5.674200000000001e-06, "loss": 0.0556, "step": 21630 }, { "epoch": 1.1532227968225195, "grad_norm": 2.5863921642303467, "learning_rate": 5.674e-06, "loss": 0.0217, "step": 21631 }, { "epoch": 1.1532761102521725, "grad_norm": 1.4723973274230957, "learning_rate": 5.673800000000001e-06, "loss": 0.0389, "step": 21632 }, { "epoch": 1.1533294236818254, "grad_norm": 0.8590759038925171, "learning_rate": 5.6736e-06, "loss": 0.0061, "step": 21633 }, { "epoch": 1.1533827371114784, "grad_norm": 2.5625083446502686, "learning_rate": 5.6734e-06, "loss": 0.0349, "step": 21634 }, { "epoch": 1.1534360505411314, "grad_norm": 1.4330146312713623, "learning_rate": 5.673200000000001e-06, "loss": 0.007, "step": 21635 }, { "epoch": 1.1534893639707842, "grad_norm": 3.1709423065185547, "learning_rate": 5.6730000000000006e-06, "loss": 0.0529, "step": 21636 }, { "epoch": 1.1535426774004371, "grad_norm": 2.328542709350586, "learning_rate": 5.6728e-06, "loss": 0.0202, "step": 21637 }, { "epoch": 1.1535959908300901, "grad_norm": 0.9317997694015503, "learning_rate": 5.6726000000000005e-06, "loss": 0.0219, "step": 21638 }, { "epoch": 1.153649304259743, "grad_norm": 2.3220250606536865, "learning_rate": 5.6724e-06, "loss": 0.0088, "step": 21639 }, { "epoch": 1.153702617689396, "grad_norm": 0.8587543964385986, "learning_rate": 5.6722e-06, "loss": 0.0179, "step": 21640 }, { "epoch": 1.1537559311190488, "grad_norm": 2.8934690952301025, "learning_rate": 5.672000000000001e-06, "loss": 0.0279, "step": 21641 }, { "epoch": 1.1538092445487018, "grad_norm": 3.4728410243988037, "learning_rate": 5.6718000000000005e-06, "loss": 0.0514, "step": 21642 }, { "epoch": 1.1538625579783548, "grad_norm": 1.5280508995056152, "learning_rate": 5.6716e-06, "loss": 0.033, "step": 21643 }, { "epoch": 1.1539158714080076, "grad_norm": 2.670639753341675, "learning_rate": 5.6714000000000005e-06, "loss": 0.0502, "step": 21644 }, { "epoch": 1.1539691848376605, "grad_norm": 1.202703833580017, "learning_rate": 5.6712e-06, "loss": 0.0103, "step": 21645 }, { "epoch": 1.1540224982673135, "grad_norm": 1.5619831085205078, "learning_rate": 5.671000000000001e-06, "loss": 0.0538, "step": 21646 }, { "epoch": 1.1540758116969665, "grad_norm": 2.372805118560791, "learning_rate": 5.670800000000001e-06, "loss": 0.0234, "step": 21647 }, { "epoch": 1.1541291251266195, "grad_norm": 1.8242697715759277, "learning_rate": 5.6706e-06, "loss": 0.0442, "step": 21648 }, { "epoch": 1.1541824385562722, "grad_norm": 1.62331223487854, "learning_rate": 5.6704e-06, "loss": 0.0162, "step": 21649 }, { "epoch": 1.1542357519859252, "grad_norm": 1.6025711297988892, "learning_rate": 5.6702e-06, "loss": 0.0271, "step": 21650 }, { "epoch": 1.1542890654155782, "grad_norm": 1.9244120121002197, "learning_rate": 5.67e-06, "loss": 0.0305, "step": 21651 }, { "epoch": 1.1543423788452312, "grad_norm": 0.6339203715324402, "learning_rate": 5.669800000000001e-06, "loss": 0.0074, "step": 21652 }, { "epoch": 1.1543956922748841, "grad_norm": 0.9599476456642151, "learning_rate": 5.669600000000001e-06, "loss": 0.0102, "step": 21653 }, { "epoch": 1.154449005704537, "grad_norm": 0.5712244510650635, "learning_rate": 5.6694e-06, "loss": 0.0068, "step": 21654 }, { "epoch": 1.1545023191341899, "grad_norm": 0.7527304887771606, "learning_rate": 5.6692e-06, "loss": 0.0126, "step": 21655 }, { "epoch": 1.1545556325638429, "grad_norm": 4.113846778869629, "learning_rate": 5.669e-06, "loss": 0.0334, "step": 21656 }, { "epoch": 1.1546089459934958, "grad_norm": 1.657859206199646, "learning_rate": 5.668800000000001e-06, "loss": 0.0262, "step": 21657 }, { "epoch": 1.1546622594231486, "grad_norm": 2.4544174671173096, "learning_rate": 5.668600000000001e-06, "loss": 0.036, "step": 21658 }, { "epoch": 1.1547155728528016, "grad_norm": 2.2815940380096436, "learning_rate": 5.668400000000001e-06, "loss": 0.027, "step": 21659 }, { "epoch": 1.1547688862824546, "grad_norm": 2.5457913875579834, "learning_rate": 5.6682e-06, "loss": 0.0268, "step": 21660 }, { "epoch": 1.1548221997121075, "grad_norm": 1.8805738687515259, "learning_rate": 5.668e-06, "loss": 0.045, "step": 21661 }, { "epoch": 1.1548755131417603, "grad_norm": 2.0998928546905518, "learning_rate": 5.6678e-06, "loss": 0.0384, "step": 21662 }, { "epoch": 1.1549288265714133, "grad_norm": 2.4272124767303467, "learning_rate": 5.667600000000001e-06, "loss": 0.0242, "step": 21663 }, { "epoch": 1.1549821400010662, "grad_norm": 1.8180501461029053, "learning_rate": 5.667400000000001e-06, "loss": 0.0139, "step": 21664 }, { "epoch": 1.1550354534307192, "grad_norm": 2.553912401199341, "learning_rate": 5.6672000000000006e-06, "loss": 0.0347, "step": 21665 }, { "epoch": 1.1550887668603722, "grad_norm": 0.6293579936027527, "learning_rate": 5.667e-06, "loss": 0.0061, "step": 21666 }, { "epoch": 1.155142080290025, "grad_norm": 2.012032985687256, "learning_rate": 5.6668e-06, "loss": 0.0105, "step": 21667 }, { "epoch": 1.155195393719678, "grad_norm": 1.4202133417129517, "learning_rate": 5.666600000000001e-06, "loss": 0.0126, "step": 21668 }, { "epoch": 1.155248707149331, "grad_norm": 2.1141884326934814, "learning_rate": 5.6664000000000005e-06, "loss": 0.0233, "step": 21669 }, { "epoch": 1.155302020578984, "grad_norm": 2.700066328048706, "learning_rate": 5.666200000000001e-06, "loss": 0.021, "step": 21670 }, { "epoch": 1.1553553340086369, "grad_norm": 4.620965480804443, "learning_rate": 5.6660000000000005e-06, "loss": 0.0383, "step": 21671 }, { "epoch": 1.1554086474382896, "grad_norm": 1.5876333713531494, "learning_rate": 5.6658e-06, "loss": 0.0279, "step": 21672 }, { "epoch": 1.1554619608679426, "grad_norm": 2.7708261013031006, "learning_rate": 5.6656e-06, "loss": 0.039, "step": 21673 }, { "epoch": 1.1555152742975956, "grad_norm": 2.2979137897491455, "learning_rate": 5.665400000000001e-06, "loss": 0.0185, "step": 21674 }, { "epoch": 1.1555685877272486, "grad_norm": 2.6296658515930176, "learning_rate": 5.6652e-06, "loss": 0.0313, "step": 21675 }, { "epoch": 1.1556219011569013, "grad_norm": 1.872787356376648, "learning_rate": 5.665000000000001e-06, "loss": 0.0311, "step": 21676 }, { "epoch": 1.1556752145865543, "grad_norm": 1.342705488204956, "learning_rate": 5.6648e-06, "loss": 0.0121, "step": 21677 }, { "epoch": 1.1557285280162073, "grad_norm": 1.2649149894714355, "learning_rate": 5.6646e-06, "loss": 0.0209, "step": 21678 }, { "epoch": 1.1557818414458603, "grad_norm": 2.8996334075927734, "learning_rate": 5.664400000000001e-06, "loss": 0.0308, "step": 21679 }, { "epoch": 1.155835154875513, "grad_norm": 0.6229035258293152, "learning_rate": 5.664200000000001e-06, "loss": 0.0052, "step": 21680 }, { "epoch": 1.155888468305166, "grad_norm": 1.7463877201080322, "learning_rate": 5.664e-06, "loss": 0.024, "step": 21681 }, { "epoch": 1.155941781734819, "grad_norm": 1.3588407039642334, "learning_rate": 5.663800000000001e-06, "loss": 0.0111, "step": 21682 }, { "epoch": 1.155995095164472, "grad_norm": 1.3150017261505127, "learning_rate": 5.6636e-06, "loss": 0.0216, "step": 21683 }, { "epoch": 1.156048408594125, "grad_norm": 2.9130914211273193, "learning_rate": 5.6634e-06, "loss": 0.0376, "step": 21684 }, { "epoch": 1.1561017220237777, "grad_norm": 1.6250759363174438, "learning_rate": 5.663200000000001e-06, "loss": 0.0345, "step": 21685 }, { "epoch": 1.1561550354534307, "grad_norm": 2.130378246307373, "learning_rate": 5.663000000000001e-06, "loss": 0.0612, "step": 21686 }, { "epoch": 1.1562083488830837, "grad_norm": 3.0044078826904297, "learning_rate": 5.6628e-06, "loss": 0.0329, "step": 21687 }, { "epoch": 1.1562616623127366, "grad_norm": 3.031122922897339, "learning_rate": 5.662600000000001e-06, "loss": 0.032, "step": 21688 }, { "epoch": 1.1563149757423896, "grad_norm": 3.5689504146575928, "learning_rate": 5.6624e-06, "loss": 0.0377, "step": 21689 }, { "epoch": 1.1563682891720424, "grad_norm": 2.104818344116211, "learning_rate": 5.6622e-06, "loss": 0.022, "step": 21690 }, { "epoch": 1.1564216026016954, "grad_norm": 3.9136979579925537, "learning_rate": 5.662000000000001e-06, "loss": 0.0519, "step": 21691 }, { "epoch": 1.1564749160313483, "grad_norm": 3.074676275253296, "learning_rate": 5.661800000000001e-06, "loss": 0.0214, "step": 21692 }, { "epoch": 1.1565282294610013, "grad_norm": 0.9655200839042664, "learning_rate": 5.6616e-06, "loss": 0.0163, "step": 21693 }, { "epoch": 1.156581542890654, "grad_norm": 1.4372445344924927, "learning_rate": 5.6614000000000005e-06, "loss": 0.0075, "step": 21694 }, { "epoch": 1.156634856320307, "grad_norm": 0.9174542427062988, "learning_rate": 5.6612e-06, "loss": 0.0078, "step": 21695 }, { "epoch": 1.15668816974996, "grad_norm": 2.0672388076782227, "learning_rate": 5.6610000000000005e-06, "loss": 0.0167, "step": 21696 }, { "epoch": 1.156741483179613, "grad_norm": 4.52235221862793, "learning_rate": 5.660800000000001e-06, "loss": 0.0467, "step": 21697 }, { "epoch": 1.1567947966092658, "grad_norm": 1.2173949480056763, "learning_rate": 5.6606000000000005e-06, "loss": 0.009, "step": 21698 }, { "epoch": 1.1568481100389187, "grad_norm": 2.9928324222564697, "learning_rate": 5.6604e-06, "loss": 0.0229, "step": 21699 }, { "epoch": 1.1569014234685717, "grad_norm": 1.868665337562561, "learning_rate": 5.6602000000000005e-06, "loss": 0.0127, "step": 21700 }, { "epoch": 1.1569547368982247, "grad_norm": 2.4275412559509277, "learning_rate": 5.66e-06, "loss": 0.0412, "step": 21701 }, { "epoch": 1.1570080503278777, "grad_norm": 2.1815590858459473, "learning_rate": 5.6598000000000004e-06, "loss": 0.018, "step": 21702 }, { "epoch": 1.1570613637575304, "grad_norm": 1.3494576215744019, "learning_rate": 5.659600000000001e-06, "loss": 0.0092, "step": 21703 }, { "epoch": 1.1571146771871834, "grad_norm": 1.921339750289917, "learning_rate": 5.6594e-06, "loss": 0.0196, "step": 21704 }, { "epoch": 1.1571679906168364, "grad_norm": 2.940859317779541, "learning_rate": 5.6592e-06, "loss": 0.0757, "step": 21705 }, { "epoch": 1.1572213040464894, "grad_norm": 2.1620266437530518, "learning_rate": 5.659e-06, "loss": 0.0411, "step": 21706 }, { "epoch": 1.1572746174761424, "grad_norm": 3.605419874191284, "learning_rate": 5.658800000000001e-06, "loss": 0.038, "step": 21707 }, { "epoch": 1.157327930905795, "grad_norm": 1.3191933631896973, "learning_rate": 5.6586e-06, "loss": 0.0152, "step": 21708 }, { "epoch": 1.157381244335448, "grad_norm": 2.4854612350463867, "learning_rate": 5.658400000000001e-06, "loss": 0.0405, "step": 21709 }, { "epoch": 1.157434557765101, "grad_norm": 0.7485052347183228, "learning_rate": 5.6582e-06, "loss": 0.0065, "step": 21710 }, { "epoch": 1.157487871194754, "grad_norm": 2.1173508167266846, "learning_rate": 5.658e-06, "loss": 0.0355, "step": 21711 }, { "epoch": 1.1575411846244068, "grad_norm": 1.605180025100708, "learning_rate": 5.6578e-06, "loss": 0.0339, "step": 21712 }, { "epoch": 1.1575944980540598, "grad_norm": 1.0766812562942505, "learning_rate": 5.657600000000001e-06, "loss": 0.0095, "step": 21713 }, { "epoch": 1.1576478114837128, "grad_norm": 3.1009268760681152, "learning_rate": 5.6574e-06, "loss": 0.0491, "step": 21714 }, { "epoch": 1.1577011249133657, "grad_norm": 1.1503666639328003, "learning_rate": 5.657200000000001e-06, "loss": 0.019, "step": 21715 }, { "epoch": 1.1577544383430185, "grad_norm": 2.0650575160980225, "learning_rate": 5.657e-06, "loss": 0.0139, "step": 21716 }, { "epoch": 1.1578077517726715, "grad_norm": 2.297731637954712, "learning_rate": 5.6568e-06, "loss": 0.0359, "step": 21717 }, { "epoch": 1.1578610652023245, "grad_norm": 1.688409447669983, "learning_rate": 5.656600000000001e-06, "loss": 0.0226, "step": 21718 }, { "epoch": 1.1579143786319774, "grad_norm": 2.713056802749634, "learning_rate": 5.656400000000001e-06, "loss": 0.014, "step": 21719 }, { "epoch": 1.1579676920616304, "grad_norm": 3.605968713760376, "learning_rate": 5.6562e-06, "loss": 0.0247, "step": 21720 }, { "epoch": 1.1580210054912832, "grad_norm": 1.0421394109725952, "learning_rate": 5.6560000000000006e-06, "loss": 0.0132, "step": 21721 }, { "epoch": 1.1580743189209362, "grad_norm": 1.2707992792129517, "learning_rate": 5.6558e-06, "loss": 0.0129, "step": 21722 }, { "epoch": 1.1581276323505891, "grad_norm": 1.3283414840698242, "learning_rate": 5.6556e-06, "loss": 0.0148, "step": 21723 }, { "epoch": 1.158180945780242, "grad_norm": 2.1478869915008545, "learning_rate": 5.655400000000001e-06, "loss": 0.0295, "step": 21724 }, { "epoch": 1.158234259209895, "grad_norm": 2.066277503967285, "learning_rate": 5.6552000000000005e-06, "loss": 0.0303, "step": 21725 }, { "epoch": 1.1582875726395478, "grad_norm": 0.6960160136222839, "learning_rate": 5.655e-06, "loss": 0.0059, "step": 21726 }, { "epoch": 1.1583408860692008, "grad_norm": 1.6403919458389282, "learning_rate": 5.6548000000000005e-06, "loss": 0.0219, "step": 21727 }, { "epoch": 1.1583941994988538, "grad_norm": 0.6824777126312256, "learning_rate": 5.6546e-06, "loss": 0.0045, "step": 21728 }, { "epoch": 1.1584475129285068, "grad_norm": 3.6710758209228516, "learning_rate": 5.654400000000001e-06, "loss": 0.037, "step": 21729 }, { "epoch": 1.1585008263581595, "grad_norm": 1.9908878803253174, "learning_rate": 5.654200000000001e-06, "loss": 0.0233, "step": 21730 }, { "epoch": 1.1585541397878125, "grad_norm": 1.4448621273040771, "learning_rate": 5.654e-06, "loss": 0.0151, "step": 21731 }, { "epoch": 1.1586074532174655, "grad_norm": 1.854927897453308, "learning_rate": 5.6538e-06, "loss": 0.0184, "step": 21732 }, { "epoch": 1.1586607666471185, "grad_norm": 0.5657631754875183, "learning_rate": 5.6536e-06, "loss": 0.004, "step": 21733 }, { "epoch": 1.1587140800767712, "grad_norm": 1.9210783243179321, "learning_rate": 5.6534e-06, "loss": 0.0243, "step": 21734 }, { "epoch": 1.1587673935064242, "grad_norm": 2.7856128215789795, "learning_rate": 5.653200000000001e-06, "loss": 0.04, "step": 21735 }, { "epoch": 1.1588207069360772, "grad_norm": 1.2417175769805908, "learning_rate": 5.653000000000001e-06, "loss": 0.0336, "step": 21736 }, { "epoch": 1.1588740203657302, "grad_norm": 1.39291250705719, "learning_rate": 5.6528e-06, "loss": 0.0192, "step": 21737 }, { "epoch": 1.1589273337953832, "grad_norm": 1.1201034784317017, "learning_rate": 5.6526e-06, "loss": 0.0099, "step": 21738 }, { "epoch": 1.158980647225036, "grad_norm": 0.8311677575111389, "learning_rate": 5.6524e-06, "loss": 0.0058, "step": 21739 }, { "epoch": 1.1590339606546889, "grad_norm": 0.8385199308395386, "learning_rate": 5.6522e-06, "loss": 0.0067, "step": 21740 }, { "epoch": 1.1590872740843419, "grad_norm": 1.6542975902557373, "learning_rate": 5.652000000000001e-06, "loss": 0.0269, "step": 21741 }, { "epoch": 1.1591405875139948, "grad_norm": 3.2849199771881104, "learning_rate": 5.651800000000001e-06, "loss": 0.0454, "step": 21742 }, { "epoch": 1.1591939009436478, "grad_norm": 1.9781030416488647, "learning_rate": 5.6516e-06, "loss": 0.0297, "step": 21743 }, { "epoch": 1.1592472143733006, "grad_norm": 2.288944959640503, "learning_rate": 5.6514e-06, "loss": 0.0245, "step": 21744 }, { "epoch": 1.1593005278029536, "grad_norm": 0.9661835432052612, "learning_rate": 5.6512e-06, "loss": 0.0096, "step": 21745 }, { "epoch": 1.1593538412326065, "grad_norm": 1.4644336700439453, "learning_rate": 5.651000000000001e-06, "loss": 0.0107, "step": 21746 }, { "epoch": 1.1594071546622593, "grad_norm": 1.9737385511398315, "learning_rate": 5.650800000000001e-06, "loss": 0.014, "step": 21747 }, { "epoch": 1.1594604680919123, "grad_norm": 1.130464792251587, "learning_rate": 5.650600000000001e-06, "loss": 0.011, "step": 21748 }, { "epoch": 1.1595137815215653, "grad_norm": 2.53131365776062, "learning_rate": 5.6504e-06, "loss": 0.0415, "step": 21749 }, { "epoch": 1.1595670949512182, "grad_norm": 3.0527637004852295, "learning_rate": 5.6502e-06, "loss": 0.018, "step": 21750 }, { "epoch": 1.1596204083808712, "grad_norm": 0.7945632934570312, "learning_rate": 5.65e-06, "loss": 0.006, "step": 21751 }, { "epoch": 1.159673721810524, "grad_norm": 2.596959352493286, "learning_rate": 5.6498000000000005e-06, "loss": 0.0214, "step": 21752 }, { "epoch": 1.159727035240177, "grad_norm": 2.41621470451355, "learning_rate": 5.649600000000001e-06, "loss": 0.0336, "step": 21753 }, { "epoch": 1.15978034866983, "grad_norm": 2.753725290298462, "learning_rate": 5.6494000000000005e-06, "loss": 0.0423, "step": 21754 }, { "epoch": 1.159833662099483, "grad_norm": 1.4400553703308105, "learning_rate": 5.6492e-06, "loss": 0.0111, "step": 21755 }, { "epoch": 1.1598869755291359, "grad_norm": 1.2908178567886353, "learning_rate": 5.649e-06, "loss": 0.033, "step": 21756 }, { "epoch": 1.1599402889587886, "grad_norm": 2.812124490737915, "learning_rate": 5.648800000000001e-06, "loss": 0.0447, "step": 21757 }, { "epoch": 1.1599936023884416, "grad_norm": 1.1653670072555542, "learning_rate": 5.6486000000000004e-06, "loss": 0.0073, "step": 21758 }, { "epoch": 1.1600469158180946, "grad_norm": 2.4823389053344727, "learning_rate": 5.648400000000001e-06, "loss": 0.0409, "step": 21759 }, { "epoch": 1.1601002292477476, "grad_norm": 3.1859500408172607, "learning_rate": 5.6482e-06, "loss": 0.0489, "step": 21760 }, { "epoch": 1.1601535426774003, "grad_norm": 0.2426280379295349, "learning_rate": 5.648e-06, "loss": 0.0025, "step": 21761 }, { "epoch": 1.1602068561070533, "grad_norm": 1.8821170330047607, "learning_rate": 5.6477999999999995e-06, "loss": 0.0417, "step": 21762 }, { "epoch": 1.1602601695367063, "grad_norm": 3.5523667335510254, "learning_rate": 5.647600000000001e-06, "loss": 0.0246, "step": 21763 }, { "epoch": 1.1603134829663593, "grad_norm": 2.507183313369751, "learning_rate": 5.6474e-06, "loss": 0.0358, "step": 21764 }, { "epoch": 1.160366796396012, "grad_norm": 2.2412099838256836, "learning_rate": 5.647200000000001e-06, "loss": 0.0528, "step": 21765 }, { "epoch": 1.160420109825665, "grad_norm": 1.6757982969284058, "learning_rate": 5.647e-06, "loss": 0.0284, "step": 21766 }, { "epoch": 1.160473423255318, "grad_norm": 1.9374281167984009, "learning_rate": 5.6468e-06, "loss": 0.0487, "step": 21767 }, { "epoch": 1.160526736684971, "grad_norm": 1.7167034149169922, "learning_rate": 5.646600000000001e-06, "loss": 0.0319, "step": 21768 }, { "epoch": 1.160580050114624, "grad_norm": 2.8088319301605225, "learning_rate": 5.646400000000001e-06, "loss": 0.0295, "step": 21769 }, { "epoch": 1.1606333635442767, "grad_norm": 1.5968050956726074, "learning_rate": 5.6462e-06, "loss": 0.0539, "step": 21770 }, { "epoch": 1.1606866769739297, "grad_norm": 1.739250659942627, "learning_rate": 5.646000000000001e-06, "loss": 0.0247, "step": 21771 }, { "epoch": 1.1607399904035827, "grad_norm": 3.1902174949645996, "learning_rate": 5.6458e-06, "loss": 0.042, "step": 21772 }, { "epoch": 1.1607933038332356, "grad_norm": 2.2369351387023926, "learning_rate": 5.6456e-06, "loss": 0.0208, "step": 21773 }, { "epoch": 1.1608466172628886, "grad_norm": 4.308271884918213, "learning_rate": 5.645400000000001e-06, "loss": 0.0437, "step": 21774 }, { "epoch": 1.1608999306925414, "grad_norm": 2.0187771320343018, "learning_rate": 5.645200000000001e-06, "loss": 0.0457, "step": 21775 }, { "epoch": 1.1609532441221944, "grad_norm": 3.81101131439209, "learning_rate": 5.645e-06, "loss": 0.0497, "step": 21776 }, { "epoch": 1.1610065575518473, "grad_norm": 3.2915546894073486, "learning_rate": 5.6448000000000006e-06, "loss": 0.0654, "step": 21777 }, { "epoch": 1.1610598709815003, "grad_norm": 0.8827807903289795, "learning_rate": 5.6446e-06, "loss": 0.0111, "step": 21778 }, { "epoch": 1.161113184411153, "grad_norm": 0.5253667831420898, "learning_rate": 5.644400000000001e-06, "loss": 0.0064, "step": 21779 }, { "epoch": 1.161166497840806, "grad_norm": 2.31488037109375, "learning_rate": 5.644200000000001e-06, "loss": 0.0264, "step": 21780 }, { "epoch": 1.161219811270459, "grad_norm": 2.431223154067993, "learning_rate": 5.6440000000000005e-06, "loss": 0.0467, "step": 21781 }, { "epoch": 1.161273124700112, "grad_norm": 1.2299482822418213, "learning_rate": 5.6438e-06, "loss": 0.0229, "step": 21782 }, { "epoch": 1.1613264381297648, "grad_norm": 2.118910789489746, "learning_rate": 5.6436000000000005e-06, "loss": 0.0259, "step": 21783 }, { "epoch": 1.1613797515594178, "grad_norm": 2.594749689102173, "learning_rate": 5.6434e-06, "loss": 0.0549, "step": 21784 }, { "epoch": 1.1614330649890707, "grad_norm": 2.1143980026245117, "learning_rate": 5.643200000000001e-06, "loss": 0.0373, "step": 21785 }, { "epoch": 1.1614863784187237, "grad_norm": 2.2770864963531494, "learning_rate": 5.643000000000001e-06, "loss": 0.0417, "step": 21786 }, { "epoch": 1.1615396918483767, "grad_norm": 1.3470330238342285, "learning_rate": 5.6428e-06, "loss": 0.0344, "step": 21787 }, { "epoch": 1.1615930052780294, "grad_norm": 1.8822375535964966, "learning_rate": 5.6426e-06, "loss": 0.0152, "step": 21788 }, { "epoch": 1.1616463187076824, "grad_norm": 1.9930822849273682, "learning_rate": 5.6424e-06, "loss": 0.0567, "step": 21789 }, { "epoch": 1.1616996321373354, "grad_norm": 2.5294578075408936, "learning_rate": 5.642200000000001e-06, "loss": 0.048, "step": 21790 }, { "epoch": 1.1617529455669884, "grad_norm": 4.358183860778809, "learning_rate": 5.642000000000001e-06, "loss": 0.0315, "step": 21791 }, { "epoch": 1.1618062589966414, "grad_norm": 1.3920246362686157, "learning_rate": 5.641800000000001e-06, "loss": 0.0158, "step": 21792 }, { "epoch": 1.1618595724262941, "grad_norm": 1.173858642578125, "learning_rate": 5.6416e-06, "loss": 0.0104, "step": 21793 }, { "epoch": 1.161912885855947, "grad_norm": 1.3802615404129028, "learning_rate": 5.6414e-06, "loss": 0.0097, "step": 21794 }, { "epoch": 1.1619661992856, "grad_norm": 1.6926885843276978, "learning_rate": 5.6412e-06, "loss": 0.0164, "step": 21795 }, { "epoch": 1.162019512715253, "grad_norm": 0.8769701719284058, "learning_rate": 5.641000000000001e-06, "loss": 0.0242, "step": 21796 }, { "epoch": 1.1620728261449058, "grad_norm": 1.2737658023834229, "learning_rate": 5.640800000000001e-06, "loss": 0.0155, "step": 21797 }, { "epoch": 1.1621261395745588, "grad_norm": 0.839714527130127, "learning_rate": 5.640600000000001e-06, "loss": 0.009, "step": 21798 }, { "epoch": 1.1621794530042118, "grad_norm": 1.4910238981246948, "learning_rate": 5.6404e-06, "loss": 0.0198, "step": 21799 }, { "epoch": 1.1622327664338648, "grad_norm": 2.563167095184326, "learning_rate": 5.6402e-06, "loss": 0.0475, "step": 21800 }, { "epoch": 1.1622860798635175, "grad_norm": 1.3601784706115723, "learning_rate": 5.64e-06, "loss": 0.0238, "step": 21801 }, { "epoch": 1.1623393932931705, "grad_norm": 0.7932131290435791, "learning_rate": 5.639800000000001e-06, "loss": 0.0209, "step": 21802 }, { "epoch": 1.1623927067228235, "grad_norm": 3.9889519214630127, "learning_rate": 5.639600000000001e-06, "loss": 0.0374, "step": 21803 }, { "epoch": 1.1624460201524764, "grad_norm": 1.7000912427902222, "learning_rate": 5.639400000000001e-06, "loss": 0.012, "step": 21804 }, { "epoch": 1.1624993335821294, "grad_norm": 0.5476977825164795, "learning_rate": 5.6392e-06, "loss": 0.0058, "step": 21805 }, { "epoch": 1.1625526470117822, "grad_norm": 1.4911854267120361, "learning_rate": 5.639e-06, "loss": 0.0136, "step": 21806 }, { "epoch": 1.1626059604414352, "grad_norm": 1.3233739137649536, "learning_rate": 5.638800000000001e-06, "loss": 0.0267, "step": 21807 }, { "epoch": 1.1626592738710881, "grad_norm": 1.1769843101501465, "learning_rate": 5.6386000000000005e-06, "loss": 0.0118, "step": 21808 }, { "epoch": 1.1627125873007411, "grad_norm": 1.7973440885543823, "learning_rate": 5.638400000000001e-06, "loss": 0.0174, "step": 21809 }, { "epoch": 1.162765900730394, "grad_norm": 4.981408596038818, "learning_rate": 5.6382000000000005e-06, "loss": 0.0506, "step": 21810 }, { "epoch": 1.1628192141600469, "grad_norm": 1.4816052913665771, "learning_rate": 5.638e-06, "loss": 0.0307, "step": 21811 }, { "epoch": 1.1628725275896998, "grad_norm": 1.9831939935684204, "learning_rate": 5.6378e-06, "loss": 0.0388, "step": 21812 }, { "epoch": 1.1629258410193528, "grad_norm": 1.5721136331558228, "learning_rate": 5.637600000000001e-06, "loss": 0.0131, "step": 21813 }, { "epoch": 1.1629791544490058, "grad_norm": 0.9621886610984802, "learning_rate": 5.6374000000000004e-06, "loss": 0.0181, "step": 21814 }, { "epoch": 1.1630324678786585, "grad_norm": 2.572854518890381, "learning_rate": 5.637200000000001e-06, "loss": 0.0475, "step": 21815 }, { "epoch": 1.1630857813083115, "grad_norm": 1.2596180438995361, "learning_rate": 5.637e-06, "loss": 0.0129, "step": 21816 }, { "epoch": 1.1631390947379645, "grad_norm": 1.4282149076461792, "learning_rate": 5.6368e-06, "loss": 0.0311, "step": 21817 }, { "epoch": 1.1631924081676175, "grad_norm": 2.2511801719665527, "learning_rate": 5.636600000000001e-06, "loss": 0.0337, "step": 21818 }, { "epoch": 1.1632457215972702, "grad_norm": 1.7558518648147583, "learning_rate": 5.636400000000001e-06, "loss": 0.0297, "step": 21819 }, { "epoch": 1.1632990350269232, "grad_norm": 1.3539067506790161, "learning_rate": 5.6362e-06, "loss": 0.0234, "step": 21820 }, { "epoch": 1.1633523484565762, "grad_norm": 2.016228199005127, "learning_rate": 5.636000000000001e-06, "loss": 0.021, "step": 21821 }, { "epoch": 1.1634056618862292, "grad_norm": 1.2633317708969116, "learning_rate": 5.6358e-06, "loss": 0.0107, "step": 21822 }, { "epoch": 1.1634589753158822, "grad_norm": 2.0493741035461426, "learning_rate": 5.6356e-06, "loss": 0.0256, "step": 21823 }, { "epoch": 1.163512288745535, "grad_norm": 2.531466007232666, "learning_rate": 5.635400000000001e-06, "loss": 0.0664, "step": 21824 }, { "epoch": 1.163565602175188, "grad_norm": 1.4136407375335693, "learning_rate": 5.635200000000001e-06, "loss": 0.0187, "step": 21825 }, { "epoch": 1.1636189156048409, "grad_norm": 2.080867290496826, "learning_rate": 5.635e-06, "loss": 0.0333, "step": 21826 }, { "epoch": 1.1636722290344939, "grad_norm": 0.701621413230896, "learning_rate": 5.634800000000001e-06, "loss": 0.011, "step": 21827 }, { "epoch": 1.1637255424641468, "grad_norm": 0.8382410407066345, "learning_rate": 5.6346e-06, "loss": 0.0188, "step": 21828 }, { "epoch": 1.1637788558937996, "grad_norm": 2.876605272293091, "learning_rate": 5.634400000000001e-06, "loss": 0.0527, "step": 21829 }, { "epoch": 1.1638321693234526, "grad_norm": 2.8704330921173096, "learning_rate": 5.634200000000001e-06, "loss": 0.0215, "step": 21830 }, { "epoch": 1.1638854827531055, "grad_norm": 2.827810049057007, "learning_rate": 5.634000000000001e-06, "loss": 0.0303, "step": 21831 }, { "epoch": 1.1639387961827585, "grad_norm": 1.4557125568389893, "learning_rate": 5.6338e-06, "loss": 0.0221, "step": 21832 }, { "epoch": 1.1639921096124113, "grad_norm": 2.0141823291778564, "learning_rate": 5.633600000000001e-06, "loss": 0.0178, "step": 21833 }, { "epoch": 1.1640454230420643, "grad_norm": 0.9730566740036011, "learning_rate": 5.6334e-06, "loss": 0.0081, "step": 21834 }, { "epoch": 1.1640987364717172, "grad_norm": 2.7713751792907715, "learning_rate": 5.6332000000000005e-06, "loss": 0.0553, "step": 21835 }, { "epoch": 1.1641520499013702, "grad_norm": 1.0834051370620728, "learning_rate": 5.633000000000001e-06, "loss": 0.0233, "step": 21836 }, { "epoch": 1.164205363331023, "grad_norm": 3.293262481689453, "learning_rate": 5.6328000000000005e-06, "loss": 0.0248, "step": 21837 }, { "epoch": 1.164258676760676, "grad_norm": 1.8528419733047485, "learning_rate": 5.6326e-06, "loss": 0.0422, "step": 21838 }, { "epoch": 1.164311990190329, "grad_norm": 2.066237211227417, "learning_rate": 5.6324000000000005e-06, "loss": 0.0144, "step": 21839 }, { "epoch": 1.164365303619982, "grad_norm": 1.7491648197174072, "learning_rate": 5.632200000000001e-06, "loss": 0.0403, "step": 21840 }, { "epoch": 1.164418617049635, "grad_norm": 1.3610364198684692, "learning_rate": 5.6320000000000005e-06, "loss": 0.0158, "step": 21841 }, { "epoch": 1.1644719304792877, "grad_norm": 1.0463298559188843, "learning_rate": 5.631800000000001e-06, "loss": 0.0109, "step": 21842 }, { "epoch": 1.1645252439089406, "grad_norm": 2.074554681777954, "learning_rate": 5.6316000000000004e-06, "loss": 0.0326, "step": 21843 }, { "epoch": 1.1645785573385936, "grad_norm": 1.9906467199325562, "learning_rate": 5.6314e-06, "loss": 0.0323, "step": 21844 }, { "epoch": 1.1646318707682466, "grad_norm": 3.201190710067749, "learning_rate": 5.6312e-06, "loss": 0.0635, "step": 21845 }, { "epoch": 1.1646851841978996, "grad_norm": 1.7187577486038208, "learning_rate": 5.631000000000001e-06, "loss": 0.0251, "step": 21846 }, { "epoch": 1.1647384976275523, "grad_norm": 3.0416018962860107, "learning_rate": 5.6308e-06, "loss": 0.0398, "step": 21847 }, { "epoch": 1.1647918110572053, "grad_norm": 0.8477566838264465, "learning_rate": 5.630600000000001e-06, "loss": 0.0064, "step": 21848 }, { "epoch": 1.1648451244868583, "grad_norm": 1.5895246267318726, "learning_rate": 5.6304e-06, "loss": 0.0415, "step": 21849 }, { "epoch": 1.1648984379165113, "grad_norm": 1.321744441986084, "learning_rate": 5.6302e-06, "loss": 0.022, "step": 21850 }, { "epoch": 1.164951751346164, "grad_norm": 3.290989875793457, "learning_rate": 5.63e-06, "loss": 0.0292, "step": 21851 }, { "epoch": 1.165005064775817, "grad_norm": 0.634315550327301, "learning_rate": 5.629800000000001e-06, "loss": 0.008, "step": 21852 }, { "epoch": 1.16505837820547, "grad_norm": 1.917876124382019, "learning_rate": 5.6296e-06, "loss": 0.025, "step": 21853 }, { "epoch": 1.165111691635123, "grad_norm": 1.6815643310546875, "learning_rate": 5.629400000000001e-06, "loss": 0.0211, "step": 21854 }, { "epoch": 1.1651650050647757, "grad_norm": 0.6051380634307861, "learning_rate": 5.6292e-06, "loss": 0.0069, "step": 21855 }, { "epoch": 1.1652183184944287, "grad_norm": 1.7745503187179565, "learning_rate": 5.629e-06, "loss": 0.0217, "step": 21856 }, { "epoch": 1.1652716319240817, "grad_norm": 1.5242900848388672, "learning_rate": 5.628800000000001e-06, "loss": 0.0205, "step": 21857 }, { "epoch": 1.1653249453537347, "grad_norm": 1.2354073524475098, "learning_rate": 5.628600000000001e-06, "loss": 0.0135, "step": 21858 }, { "epoch": 1.1653782587833876, "grad_norm": 1.417586326599121, "learning_rate": 5.6284e-06, "loss": 0.023, "step": 21859 }, { "epoch": 1.1654315722130404, "grad_norm": 2.239168643951416, "learning_rate": 5.628200000000001e-06, "loss": 0.0404, "step": 21860 }, { "epoch": 1.1654848856426934, "grad_norm": 1.5060553550720215, "learning_rate": 5.628e-06, "loss": 0.008, "step": 21861 }, { "epoch": 1.1655381990723463, "grad_norm": 1.5505937337875366, "learning_rate": 5.6278e-06, "loss": 0.0457, "step": 21862 }, { "epoch": 1.1655915125019993, "grad_norm": 1.5592414140701294, "learning_rate": 5.627600000000001e-06, "loss": 0.0155, "step": 21863 }, { "epoch": 1.1656448259316523, "grad_norm": 0.5217198133468628, "learning_rate": 5.6274000000000005e-06, "loss": 0.0055, "step": 21864 }, { "epoch": 1.165698139361305, "grad_norm": 1.1021602153778076, "learning_rate": 5.6272e-06, "loss": 0.0371, "step": 21865 }, { "epoch": 1.165751452790958, "grad_norm": 1.2321596145629883, "learning_rate": 5.6270000000000005e-06, "loss": 0.0069, "step": 21866 }, { "epoch": 1.165804766220611, "grad_norm": 2.1025359630584717, "learning_rate": 5.6268e-06, "loss": 0.0474, "step": 21867 }, { "epoch": 1.1658580796502638, "grad_norm": 2.237201452255249, "learning_rate": 5.626600000000001e-06, "loss": 0.0127, "step": 21868 }, { "epoch": 1.1659113930799168, "grad_norm": 1.3994702100753784, "learning_rate": 5.626400000000001e-06, "loss": 0.0145, "step": 21869 }, { "epoch": 1.1659647065095697, "grad_norm": 1.095780372619629, "learning_rate": 5.6262000000000004e-06, "loss": 0.0107, "step": 21870 }, { "epoch": 1.1660180199392227, "grad_norm": 1.8719055652618408, "learning_rate": 5.626e-06, "loss": 0.036, "step": 21871 }, { "epoch": 1.1660713333688757, "grad_norm": 1.863899827003479, "learning_rate": 5.6258e-06, "loss": 0.0419, "step": 21872 }, { "epoch": 1.1661246467985285, "grad_norm": 1.1729977130889893, "learning_rate": 5.6256e-06, "loss": 0.0227, "step": 21873 }, { "epoch": 1.1661779602281814, "grad_norm": 1.735230803489685, "learning_rate": 5.625400000000001e-06, "loss": 0.0213, "step": 21874 }, { "epoch": 1.1662312736578344, "grad_norm": 1.6907306909561157, "learning_rate": 5.625200000000001e-06, "loss": 0.0151, "step": 21875 }, { "epoch": 1.1662845870874874, "grad_norm": 1.9087222814559937, "learning_rate": 5.625e-06, "loss": 0.0408, "step": 21876 }, { "epoch": 1.1663379005171404, "grad_norm": 1.8400883674621582, "learning_rate": 5.6248e-06, "loss": 0.0163, "step": 21877 }, { "epoch": 1.1663912139467931, "grad_norm": 2.3644039630889893, "learning_rate": 5.6246e-06, "loss": 0.0401, "step": 21878 }, { "epoch": 1.166444527376446, "grad_norm": 1.6784332990646362, "learning_rate": 5.624400000000001e-06, "loss": 0.0316, "step": 21879 }, { "epoch": 1.166497840806099, "grad_norm": 4.258377552032471, "learning_rate": 5.624200000000001e-06, "loss": 0.0405, "step": 21880 }, { "epoch": 1.166551154235752, "grad_norm": 1.6629259586334229, "learning_rate": 5.624000000000001e-06, "loss": 0.014, "step": 21881 }, { "epoch": 1.1666044676654048, "grad_norm": 3.1492233276367188, "learning_rate": 5.6238e-06, "loss": 0.0291, "step": 21882 }, { "epoch": 1.1666577810950578, "grad_norm": 1.1095291376113892, "learning_rate": 5.6236e-06, "loss": 0.0138, "step": 21883 }, { "epoch": 1.1667110945247108, "grad_norm": 1.8182518482208252, "learning_rate": 5.6234e-06, "loss": 0.0221, "step": 21884 }, { "epoch": 1.1667644079543638, "grad_norm": 1.359352707862854, "learning_rate": 5.623200000000001e-06, "loss": 0.0098, "step": 21885 }, { "epoch": 1.1668177213840165, "grad_norm": 0.49065154790878296, "learning_rate": 5.623000000000001e-06, "loss": 0.004, "step": 21886 }, { "epoch": 1.1668710348136695, "grad_norm": 0.9104269742965698, "learning_rate": 5.622800000000001e-06, "loss": 0.007, "step": 21887 }, { "epoch": 1.1669243482433225, "grad_norm": 2.5036122798919678, "learning_rate": 5.6226e-06, "loss": 0.0347, "step": 21888 }, { "epoch": 1.1669776616729755, "grad_norm": 2.0181493759155273, "learning_rate": 5.6224e-06, "loss": 0.0191, "step": 21889 }, { "epoch": 1.1670309751026284, "grad_norm": 3.0479423999786377, "learning_rate": 5.622200000000001e-06, "loss": 0.0373, "step": 21890 }, { "epoch": 1.1670842885322812, "grad_norm": 1.648636817932129, "learning_rate": 5.6220000000000006e-06, "loss": 0.0241, "step": 21891 }, { "epoch": 1.1671376019619342, "grad_norm": 1.5197445154190063, "learning_rate": 5.621800000000001e-06, "loss": 0.0237, "step": 21892 }, { "epoch": 1.1671909153915871, "grad_norm": 1.752389907836914, "learning_rate": 5.6216000000000005e-06, "loss": 0.0338, "step": 21893 }, { "epoch": 1.1672442288212401, "grad_norm": 2.2427597045898438, "learning_rate": 5.6214e-06, "loss": 0.0287, "step": 21894 }, { "epoch": 1.167297542250893, "grad_norm": 2.8063650131225586, "learning_rate": 5.6212e-06, "loss": 0.0461, "step": 21895 }, { "epoch": 1.1673508556805459, "grad_norm": 2.6659576892852783, "learning_rate": 5.621000000000001e-06, "loss": 0.0371, "step": 21896 }, { "epoch": 1.1674041691101988, "grad_norm": 4.049905300140381, "learning_rate": 5.6208000000000005e-06, "loss": 0.0264, "step": 21897 }, { "epoch": 1.1674574825398518, "grad_norm": 4.8570451736450195, "learning_rate": 5.620600000000001e-06, "loss": 0.0678, "step": 21898 }, { "epoch": 1.1675107959695048, "grad_norm": 1.9305192232131958, "learning_rate": 5.6204000000000004e-06, "loss": 0.0284, "step": 21899 }, { "epoch": 1.1675641093991576, "grad_norm": 2.8695011138916016, "learning_rate": 5.6202e-06, "loss": 0.0304, "step": 21900 }, { "epoch": 1.1676174228288105, "grad_norm": 1.5644135475158691, "learning_rate": 5.620000000000001e-06, "loss": 0.0296, "step": 21901 }, { "epoch": 1.1676707362584635, "grad_norm": 1.4007192850112915, "learning_rate": 5.619800000000001e-06, "loss": 0.028, "step": 21902 }, { "epoch": 1.1677240496881165, "grad_norm": 1.5038657188415527, "learning_rate": 5.6196e-06, "loss": 0.0149, "step": 21903 }, { "epoch": 1.1677773631177693, "grad_norm": 1.422641634941101, "learning_rate": 5.619400000000001e-06, "loss": 0.026, "step": 21904 }, { "epoch": 1.1678306765474222, "grad_norm": 2.174306869506836, "learning_rate": 5.6192e-06, "loss": 0.0474, "step": 21905 }, { "epoch": 1.1678839899770752, "grad_norm": 1.9260364770889282, "learning_rate": 5.619e-06, "loss": 0.0266, "step": 21906 }, { "epoch": 1.1679373034067282, "grad_norm": 1.3768881559371948, "learning_rate": 5.618800000000001e-06, "loss": 0.0335, "step": 21907 }, { "epoch": 1.1679906168363812, "grad_norm": 2.3066468238830566, "learning_rate": 5.618600000000001e-06, "loss": 0.0241, "step": 21908 }, { "epoch": 1.168043930266034, "grad_norm": 2.605313777923584, "learning_rate": 5.6184e-06, "loss": 0.0282, "step": 21909 }, { "epoch": 1.168097243695687, "grad_norm": 3.0239670276641846, "learning_rate": 5.618200000000001e-06, "loss": 0.0326, "step": 21910 }, { "epoch": 1.1681505571253399, "grad_norm": 2.2527589797973633, "learning_rate": 5.618e-06, "loss": 0.0245, "step": 21911 }, { "epoch": 1.1682038705549929, "grad_norm": 2.0380117893218994, "learning_rate": 5.6178e-06, "loss": 0.0343, "step": 21912 }, { "epoch": 1.1682571839846458, "grad_norm": 2.00309419631958, "learning_rate": 5.617600000000001e-06, "loss": 0.048, "step": 21913 }, { "epoch": 1.1683104974142986, "grad_norm": 3.9052605628967285, "learning_rate": 5.617400000000001e-06, "loss": 0.0481, "step": 21914 }, { "epoch": 1.1683638108439516, "grad_norm": 1.3333160877227783, "learning_rate": 5.6172e-06, "loss": 0.0325, "step": 21915 }, { "epoch": 1.1684171242736046, "grad_norm": 1.3366365432739258, "learning_rate": 5.617000000000001e-06, "loss": 0.0277, "step": 21916 }, { "epoch": 1.1684704377032575, "grad_norm": 1.093003273010254, "learning_rate": 5.6168e-06, "loss": 0.0209, "step": 21917 }, { "epoch": 1.1685237511329103, "grad_norm": 1.3746232986450195, "learning_rate": 5.6166000000000006e-06, "loss": 0.0211, "step": 21918 }, { "epoch": 1.1685770645625633, "grad_norm": 1.6231948137283325, "learning_rate": 5.616400000000001e-06, "loss": 0.0336, "step": 21919 }, { "epoch": 1.1686303779922163, "grad_norm": 0.8803166747093201, "learning_rate": 5.6162000000000005e-06, "loss": 0.0081, "step": 21920 }, { "epoch": 1.1686836914218692, "grad_norm": 1.7867431640625, "learning_rate": 5.616e-06, "loss": 0.031, "step": 21921 }, { "epoch": 1.168737004851522, "grad_norm": 1.191089391708374, "learning_rate": 5.6158000000000005e-06, "loss": 0.0137, "step": 21922 }, { "epoch": 1.168790318281175, "grad_norm": 1.2812072038650513, "learning_rate": 5.6156e-06, "loss": 0.0111, "step": 21923 }, { "epoch": 1.168843631710828, "grad_norm": 1.335009217262268, "learning_rate": 5.6154000000000005e-06, "loss": 0.0312, "step": 21924 }, { "epoch": 1.168896945140481, "grad_norm": 1.1308234930038452, "learning_rate": 5.615200000000001e-06, "loss": 0.0156, "step": 21925 }, { "epoch": 1.168950258570134, "grad_norm": 2.2503926753997803, "learning_rate": 5.6150000000000005e-06, "loss": 0.0453, "step": 21926 }, { "epoch": 1.1690035719997867, "grad_norm": 2.5171875953674316, "learning_rate": 5.6148e-06, "loss": 0.056, "step": 21927 }, { "epoch": 1.1690568854294396, "grad_norm": 1.3550573587417603, "learning_rate": 5.6146e-06, "loss": 0.0296, "step": 21928 }, { "epoch": 1.1691101988590926, "grad_norm": 1.4280983209609985, "learning_rate": 5.614400000000001e-06, "loss": 0.0326, "step": 21929 }, { "epoch": 1.1691635122887456, "grad_norm": 1.608784794807434, "learning_rate": 5.6142e-06, "loss": 0.0241, "step": 21930 }, { "epoch": 1.1692168257183986, "grad_norm": 2.1398773193359375, "learning_rate": 5.614000000000001e-06, "loss": 0.0319, "step": 21931 }, { "epoch": 1.1692701391480513, "grad_norm": 1.6432610750198364, "learning_rate": 5.6138e-06, "loss": 0.0231, "step": 21932 }, { "epoch": 1.1693234525777043, "grad_norm": 1.5016111135482788, "learning_rate": 5.6136e-06, "loss": 0.0167, "step": 21933 }, { "epoch": 1.1693767660073573, "grad_norm": 1.494099736213684, "learning_rate": 5.6134e-06, "loss": 0.0263, "step": 21934 }, { "epoch": 1.1694300794370103, "grad_norm": 2.3392622470855713, "learning_rate": 5.613200000000001e-06, "loss": 0.0307, "step": 21935 }, { "epoch": 1.169483392866663, "grad_norm": 2.0599751472473145, "learning_rate": 5.613e-06, "loss": 0.0378, "step": 21936 }, { "epoch": 1.169536706296316, "grad_norm": 1.123820424079895, "learning_rate": 5.612800000000001e-06, "loss": 0.0144, "step": 21937 }, { "epoch": 1.169590019725969, "grad_norm": 2.081242799758911, "learning_rate": 5.6126e-06, "loss": 0.0544, "step": 21938 }, { "epoch": 1.169643333155622, "grad_norm": 1.8207932710647583, "learning_rate": 5.6124e-06, "loss": 0.0352, "step": 21939 }, { "epoch": 1.1696966465852747, "grad_norm": 1.6269147396087646, "learning_rate": 5.612200000000001e-06, "loss": 0.0167, "step": 21940 }, { "epoch": 1.1697499600149277, "grad_norm": 1.1919918060302734, "learning_rate": 5.612000000000001e-06, "loss": 0.0176, "step": 21941 }, { "epoch": 1.1698032734445807, "grad_norm": 1.3092067241668701, "learning_rate": 5.6118e-06, "loss": 0.0117, "step": 21942 }, { "epoch": 1.1698565868742337, "grad_norm": 2.4126367568969727, "learning_rate": 5.611600000000001e-06, "loss": 0.0183, "step": 21943 }, { "epoch": 1.1699099003038866, "grad_norm": 1.6754952669143677, "learning_rate": 5.6114e-06, "loss": 0.0444, "step": 21944 }, { "epoch": 1.1699632137335394, "grad_norm": 1.8906677961349487, "learning_rate": 5.6112e-06, "loss": 0.0276, "step": 21945 }, { "epoch": 1.1700165271631924, "grad_norm": 2.3250324726104736, "learning_rate": 5.611000000000001e-06, "loss": 0.0625, "step": 21946 }, { "epoch": 1.1700698405928454, "grad_norm": 2.2217533588409424, "learning_rate": 5.6108000000000006e-06, "loss": 0.0236, "step": 21947 }, { "epoch": 1.1701231540224983, "grad_norm": 2.216792583465576, "learning_rate": 5.6106e-06, "loss": 0.0578, "step": 21948 }, { "epoch": 1.1701764674521513, "grad_norm": 2.211796760559082, "learning_rate": 5.6104000000000005e-06, "loss": 0.0524, "step": 21949 }, { "epoch": 1.170229780881804, "grad_norm": 1.1399290561676025, "learning_rate": 5.6102e-06, "loss": 0.0259, "step": 21950 }, { "epoch": 1.170283094311457, "grad_norm": 1.8898026943206787, "learning_rate": 5.610000000000001e-06, "loss": 0.0251, "step": 21951 }, { "epoch": 1.17033640774111, "grad_norm": 2.7921297550201416, "learning_rate": 5.609800000000001e-06, "loss": 0.0597, "step": 21952 }, { "epoch": 1.170389721170763, "grad_norm": 1.9041035175323486, "learning_rate": 5.6096000000000005e-06, "loss": 0.0263, "step": 21953 }, { "epoch": 1.1704430346004158, "grad_norm": 1.5103358030319214, "learning_rate": 5.6094e-06, "loss": 0.0363, "step": 21954 }, { "epoch": 1.1704963480300687, "grad_norm": 0.933194100856781, "learning_rate": 5.6092000000000004e-06, "loss": 0.0071, "step": 21955 }, { "epoch": 1.1705496614597217, "grad_norm": 3.020911455154419, "learning_rate": 5.609e-06, "loss": 0.0769, "step": 21956 }, { "epoch": 1.1706029748893747, "grad_norm": 1.364457368850708, "learning_rate": 5.608800000000001e-06, "loss": 0.0188, "step": 21957 }, { "epoch": 1.1706562883190275, "grad_norm": 4.649115562438965, "learning_rate": 5.608600000000001e-06, "loss": 0.0531, "step": 21958 }, { "epoch": 1.1707096017486804, "grad_norm": 1.6664433479309082, "learning_rate": 5.6084e-06, "loss": 0.0181, "step": 21959 }, { "epoch": 1.1707629151783334, "grad_norm": 1.4392156600952148, "learning_rate": 5.6082e-06, "loss": 0.01, "step": 21960 }, { "epoch": 1.1708162286079864, "grad_norm": 1.9527106285095215, "learning_rate": 5.608e-06, "loss": 0.0411, "step": 21961 }, { "epoch": 1.1708695420376394, "grad_norm": 1.1938682794570923, "learning_rate": 5.6078e-06, "loss": 0.0127, "step": 21962 }, { "epoch": 1.1709228554672921, "grad_norm": 0.9599004983901978, "learning_rate": 5.607600000000001e-06, "loss": 0.0207, "step": 21963 }, { "epoch": 1.1709761688969451, "grad_norm": 1.7371671199798584, "learning_rate": 5.607400000000001e-06, "loss": 0.0344, "step": 21964 }, { "epoch": 1.171029482326598, "grad_norm": 1.1125563383102417, "learning_rate": 5.6072e-06, "loss": 0.0108, "step": 21965 }, { "epoch": 1.171082795756251, "grad_norm": 1.0729485750198364, "learning_rate": 5.607e-06, "loss": 0.0105, "step": 21966 }, { "epoch": 1.171136109185904, "grad_norm": 2.37379789352417, "learning_rate": 5.6068e-06, "loss": 0.0311, "step": 21967 }, { "epoch": 1.1711894226155568, "grad_norm": 2.0396275520324707, "learning_rate": 5.606600000000001e-06, "loss": 0.0261, "step": 21968 }, { "epoch": 1.1712427360452098, "grad_norm": 1.2666199207305908, "learning_rate": 5.606400000000001e-06, "loss": 0.0164, "step": 21969 }, { "epoch": 1.1712960494748628, "grad_norm": 2.91387677192688, "learning_rate": 5.606200000000001e-06, "loss": 0.0533, "step": 21970 }, { "epoch": 1.1713493629045155, "grad_norm": 2.3043899536132812, "learning_rate": 5.606e-06, "loss": 0.0182, "step": 21971 }, { "epoch": 1.1714026763341685, "grad_norm": 1.0091828107833862, "learning_rate": 5.6058e-06, "loss": 0.0105, "step": 21972 }, { "epoch": 1.1714559897638215, "grad_norm": 3.369572401046753, "learning_rate": 5.6056e-06, "loss": 0.0264, "step": 21973 }, { "epoch": 1.1715093031934745, "grad_norm": 1.6067923307418823, "learning_rate": 5.6054000000000006e-06, "loss": 0.0206, "step": 21974 }, { "epoch": 1.1715626166231274, "grad_norm": 1.7024130821228027, "learning_rate": 5.605200000000001e-06, "loss": 0.0368, "step": 21975 }, { "epoch": 1.1716159300527802, "grad_norm": 1.3331109285354614, "learning_rate": 5.6050000000000005e-06, "loss": 0.0332, "step": 21976 }, { "epoch": 1.1716692434824332, "grad_norm": 0.8901434540748596, "learning_rate": 5.6048e-06, "loss": 0.0122, "step": 21977 }, { "epoch": 1.1717225569120862, "grad_norm": 1.5393378734588623, "learning_rate": 5.6046e-06, "loss": 0.0157, "step": 21978 }, { "epoch": 1.1717758703417391, "grad_norm": 2.824333667755127, "learning_rate": 5.604400000000001e-06, "loss": 0.0395, "step": 21979 }, { "epoch": 1.1718291837713921, "grad_norm": 1.4810019731521606, "learning_rate": 5.6042000000000005e-06, "loss": 0.0146, "step": 21980 }, { "epoch": 1.1718824972010449, "grad_norm": 3.7698147296905518, "learning_rate": 5.604000000000001e-06, "loss": 0.0469, "step": 21981 }, { "epoch": 1.1719358106306978, "grad_norm": 1.8291946649551392, "learning_rate": 5.6038000000000005e-06, "loss": 0.0187, "step": 21982 }, { "epoch": 1.1719891240603508, "grad_norm": 1.0680814981460571, "learning_rate": 5.6036e-06, "loss": 0.0102, "step": 21983 }, { "epoch": 1.1720424374900038, "grad_norm": 1.004217267036438, "learning_rate": 5.6034e-06, "loss": 0.0149, "step": 21984 }, { "epoch": 1.1720957509196566, "grad_norm": 1.5015337467193604, "learning_rate": 5.603200000000001e-06, "loss": 0.0192, "step": 21985 }, { "epoch": 1.1721490643493095, "grad_norm": 2.4269118309020996, "learning_rate": 5.603e-06, "loss": 0.0083, "step": 21986 }, { "epoch": 1.1722023777789625, "grad_norm": 2.487156629562378, "learning_rate": 5.602800000000001e-06, "loss": 0.0474, "step": 21987 }, { "epoch": 1.1722556912086155, "grad_norm": 0.8390879034996033, "learning_rate": 5.6026e-06, "loss": 0.0117, "step": 21988 }, { "epoch": 1.1723090046382683, "grad_norm": 1.6236211061477661, "learning_rate": 5.6024e-06, "loss": 0.0231, "step": 21989 }, { "epoch": 1.1723623180679212, "grad_norm": 0.944103479385376, "learning_rate": 5.602200000000001e-06, "loss": 0.0064, "step": 21990 }, { "epoch": 1.1724156314975742, "grad_norm": 1.6575802564620972, "learning_rate": 5.602000000000001e-06, "loss": 0.0467, "step": 21991 }, { "epoch": 1.1724689449272272, "grad_norm": 1.651149868965149, "learning_rate": 5.6018e-06, "loss": 0.0363, "step": 21992 }, { "epoch": 1.1725222583568802, "grad_norm": 2.77105712890625, "learning_rate": 5.601600000000001e-06, "loss": 0.0263, "step": 21993 }, { "epoch": 1.172575571786533, "grad_norm": 2.300152063369751, "learning_rate": 5.6014e-06, "loss": 0.0377, "step": 21994 }, { "epoch": 1.172628885216186, "grad_norm": 2.280080795288086, "learning_rate": 5.6012e-06, "loss": 0.0087, "step": 21995 }, { "epoch": 1.172682198645839, "grad_norm": 0.9410639405250549, "learning_rate": 5.601000000000001e-06, "loss": 0.0137, "step": 21996 }, { "epoch": 1.1727355120754919, "grad_norm": 1.5240330696105957, "learning_rate": 5.600800000000001e-06, "loss": 0.0303, "step": 21997 }, { "epoch": 1.1727888255051448, "grad_norm": 3.2640044689178467, "learning_rate": 5.6006e-06, "loss": 0.0292, "step": 21998 }, { "epoch": 1.1728421389347976, "grad_norm": 2.427776336669922, "learning_rate": 5.600400000000001e-06, "loss": 0.0236, "step": 21999 }, { "epoch": 1.1728954523644506, "grad_norm": 2.079439163208008, "learning_rate": 5.6002e-06, "loss": 0.0311, "step": 22000 }, { "epoch": 1.1729487657941036, "grad_norm": 2.2537569999694824, "learning_rate": 5.600000000000001e-06, "loss": 0.0558, "step": 22001 }, { "epoch": 1.1730020792237565, "grad_norm": 2.9652650356292725, "learning_rate": 5.599800000000001e-06, "loss": 0.0305, "step": 22002 }, { "epoch": 1.1730553926534093, "grad_norm": 1.705336093902588, "learning_rate": 5.5996000000000006e-06, "loss": 0.0118, "step": 22003 }, { "epoch": 1.1731087060830623, "grad_norm": 1.083268642425537, "learning_rate": 5.5994e-06, "loss": 0.0083, "step": 22004 }, { "epoch": 1.1731620195127153, "grad_norm": 1.2127981185913086, "learning_rate": 5.5992000000000005e-06, "loss": 0.0141, "step": 22005 }, { "epoch": 1.1732153329423682, "grad_norm": 2.110964775085449, "learning_rate": 5.599e-06, "loss": 0.0289, "step": 22006 }, { "epoch": 1.173268646372021, "grad_norm": 2.028472900390625, "learning_rate": 5.5988000000000005e-06, "loss": 0.0139, "step": 22007 }, { "epoch": 1.173321959801674, "grad_norm": 1.3077392578125, "learning_rate": 5.598600000000001e-06, "loss": 0.0294, "step": 22008 }, { "epoch": 1.173375273231327, "grad_norm": 0.48705753684043884, "learning_rate": 5.5984000000000005e-06, "loss": 0.0043, "step": 22009 }, { "epoch": 1.17342858666098, "grad_norm": 1.8598692417144775, "learning_rate": 5.5982e-06, "loss": 0.0388, "step": 22010 }, { "epoch": 1.173481900090633, "grad_norm": 1.031828761100769, "learning_rate": 5.5980000000000004e-06, "loss": 0.0315, "step": 22011 }, { "epoch": 1.1735352135202857, "grad_norm": 0.5068263411521912, "learning_rate": 5.597800000000001e-06, "loss": 0.0046, "step": 22012 }, { "epoch": 1.1735885269499386, "grad_norm": 2.215322494506836, "learning_rate": 5.5976e-06, "loss": 0.043, "step": 22013 }, { "epoch": 1.1736418403795916, "grad_norm": 3.0630781650543213, "learning_rate": 5.597400000000001e-06, "loss": 0.0286, "step": 22014 }, { "epoch": 1.1736951538092446, "grad_norm": 2.3079617023468018, "learning_rate": 5.5972e-06, "loss": 0.056, "step": 22015 }, { "epoch": 1.1737484672388976, "grad_norm": 2.092432975769043, "learning_rate": 5.597e-06, "loss": 0.0184, "step": 22016 }, { "epoch": 1.1738017806685503, "grad_norm": 2.051753520965576, "learning_rate": 5.5968e-06, "loss": 0.0237, "step": 22017 }, { "epoch": 1.1738550940982033, "grad_norm": 3.7773571014404297, "learning_rate": 5.596600000000001e-06, "loss": 0.053, "step": 22018 }, { "epoch": 1.1739084075278563, "grad_norm": 0.9323849678039551, "learning_rate": 5.5964e-06, "loss": 0.0104, "step": 22019 }, { "epoch": 1.1739617209575093, "grad_norm": 1.5253850221633911, "learning_rate": 5.596200000000001e-06, "loss": 0.0238, "step": 22020 }, { "epoch": 1.174015034387162, "grad_norm": 0.6746487617492676, "learning_rate": 5.596e-06, "loss": 0.0085, "step": 22021 }, { "epoch": 1.174068347816815, "grad_norm": 1.280850887298584, "learning_rate": 5.5958e-06, "loss": 0.0195, "step": 22022 }, { "epoch": 1.174121661246468, "grad_norm": 1.1524205207824707, "learning_rate": 5.5956e-06, "loss": 0.0133, "step": 22023 }, { "epoch": 1.174174974676121, "grad_norm": 2.4195175170898438, "learning_rate": 5.595400000000001e-06, "loss": 0.0216, "step": 22024 }, { "epoch": 1.1742282881057737, "grad_norm": 1.3454811573028564, "learning_rate": 5.5952e-06, "loss": 0.0152, "step": 22025 }, { "epoch": 1.1742816015354267, "grad_norm": 1.1743003129959106, "learning_rate": 5.595000000000001e-06, "loss": 0.0147, "step": 22026 }, { "epoch": 1.1743349149650797, "grad_norm": 2.0044755935668945, "learning_rate": 5.5948e-06, "loss": 0.0341, "step": 22027 }, { "epoch": 1.1743882283947327, "grad_norm": 1.235261082649231, "learning_rate": 5.5946e-06, "loss": 0.0109, "step": 22028 }, { "epoch": 1.1744415418243856, "grad_norm": 0.686897873878479, "learning_rate": 5.594400000000001e-06, "loss": 0.006, "step": 22029 }, { "epoch": 1.1744948552540384, "grad_norm": 0.8300411701202393, "learning_rate": 5.594200000000001e-06, "loss": 0.0233, "step": 22030 }, { "epoch": 1.1745481686836914, "grad_norm": 2.076383113861084, "learning_rate": 5.594e-06, "loss": 0.0215, "step": 22031 }, { "epoch": 1.1746014821133444, "grad_norm": 1.4035307168960571, "learning_rate": 5.5938000000000006e-06, "loss": 0.0121, "step": 22032 }, { "epoch": 1.1746547955429973, "grad_norm": 1.743327021598816, "learning_rate": 5.5936e-06, "loss": 0.0137, "step": 22033 }, { "epoch": 1.1747081089726503, "grad_norm": 0.7874998450279236, "learning_rate": 5.5934e-06, "loss": 0.0058, "step": 22034 }, { "epoch": 1.174761422402303, "grad_norm": 1.0293960571289062, "learning_rate": 5.593200000000001e-06, "loss": 0.0245, "step": 22035 }, { "epoch": 1.174814735831956, "grad_norm": 1.376317024230957, "learning_rate": 5.5930000000000005e-06, "loss": 0.0203, "step": 22036 }, { "epoch": 1.174868049261609, "grad_norm": 1.3796063661575317, "learning_rate": 5.5928e-06, "loss": 0.0098, "step": 22037 }, { "epoch": 1.174921362691262, "grad_norm": 1.5726186037063599, "learning_rate": 5.5926000000000005e-06, "loss": 0.0139, "step": 22038 }, { "epoch": 1.1749746761209148, "grad_norm": 1.3148350715637207, "learning_rate": 5.5924e-06, "loss": 0.0292, "step": 22039 }, { "epoch": 1.1750279895505678, "grad_norm": 1.5520681142807007, "learning_rate": 5.592200000000001e-06, "loss": 0.0096, "step": 22040 }, { "epoch": 1.1750813029802207, "grad_norm": 3.859837055206299, "learning_rate": 5.592000000000001e-06, "loss": 0.0805, "step": 22041 }, { "epoch": 1.1751346164098737, "grad_norm": 1.5344618558883667, "learning_rate": 5.5918e-06, "loss": 0.0189, "step": 22042 }, { "epoch": 1.1751879298395265, "grad_norm": 1.3532707691192627, "learning_rate": 5.5916e-06, "loss": 0.0207, "step": 22043 }, { "epoch": 1.1752412432691794, "grad_norm": 2.3796329498291016, "learning_rate": 5.5914e-06, "loss": 0.0334, "step": 22044 }, { "epoch": 1.1752945566988324, "grad_norm": 1.437124490737915, "learning_rate": 5.5912e-06, "loss": 0.0117, "step": 22045 }, { "epoch": 1.1753478701284854, "grad_norm": 2.9301702976226807, "learning_rate": 5.591000000000001e-06, "loss": 0.0274, "step": 22046 }, { "epoch": 1.1754011835581384, "grad_norm": 2.640645980834961, "learning_rate": 5.590800000000001e-06, "loss": 0.0223, "step": 22047 }, { "epoch": 1.1754544969877911, "grad_norm": 1.7887656688690186, "learning_rate": 5.5906e-06, "loss": 0.016, "step": 22048 }, { "epoch": 1.1755078104174441, "grad_norm": 2.1306333541870117, "learning_rate": 5.5904e-06, "loss": 0.0338, "step": 22049 }, { "epoch": 1.175561123847097, "grad_norm": 0.969560444355011, "learning_rate": 5.5902e-06, "loss": 0.0063, "step": 22050 }, { "epoch": 1.17561443727675, "grad_norm": 5.495848655700684, "learning_rate": 5.590000000000001e-06, "loss": 0.0568, "step": 22051 }, { "epoch": 1.175667750706403, "grad_norm": 2.418285608291626, "learning_rate": 5.589800000000001e-06, "loss": 0.028, "step": 22052 }, { "epoch": 1.1757210641360558, "grad_norm": 1.4646645784378052, "learning_rate": 5.589600000000001e-06, "loss": 0.0156, "step": 22053 }, { "epoch": 1.1757743775657088, "grad_norm": 3.3704264163970947, "learning_rate": 5.5894e-06, "loss": 0.0462, "step": 22054 }, { "epoch": 1.1758276909953618, "grad_norm": 0.9415836930274963, "learning_rate": 5.5892e-06, "loss": 0.0051, "step": 22055 }, { "epoch": 1.1758810044250148, "grad_norm": 1.7078394889831543, "learning_rate": 5.589e-06, "loss": 0.0473, "step": 22056 }, { "epoch": 1.1759343178546675, "grad_norm": 0.414496511220932, "learning_rate": 5.588800000000001e-06, "loss": 0.0039, "step": 22057 }, { "epoch": 1.1759876312843205, "grad_norm": 1.6372933387756348, "learning_rate": 5.588600000000001e-06, "loss": 0.0136, "step": 22058 }, { "epoch": 1.1760409447139735, "grad_norm": 1.4454401731491089, "learning_rate": 5.5884000000000006e-06, "loss": 0.0187, "step": 22059 }, { "epoch": 1.1760942581436264, "grad_norm": 0.7669732570648193, "learning_rate": 5.5882e-06, "loss": 0.0055, "step": 22060 }, { "epoch": 1.1761475715732792, "grad_norm": 2.9467406272888184, "learning_rate": 5.588e-06, "loss": 0.0226, "step": 22061 }, { "epoch": 1.1762008850029322, "grad_norm": 1.3225033283233643, "learning_rate": 5.587800000000001e-06, "loss": 0.01, "step": 22062 }, { "epoch": 1.1762541984325852, "grad_norm": 2.349000930786133, "learning_rate": 5.5876000000000005e-06, "loss": 0.027, "step": 22063 }, { "epoch": 1.1763075118622381, "grad_norm": 2.5392298698425293, "learning_rate": 5.587400000000001e-06, "loss": 0.0202, "step": 22064 }, { "epoch": 1.1763608252918911, "grad_norm": 2.007234573364258, "learning_rate": 5.5872000000000005e-06, "loss": 0.017, "step": 22065 }, { "epoch": 1.1764141387215439, "grad_norm": 1.4656264781951904, "learning_rate": 5.587e-06, "loss": 0.0155, "step": 22066 }, { "epoch": 1.1764674521511969, "grad_norm": 3.114976406097412, "learning_rate": 5.5868e-06, "loss": 0.0441, "step": 22067 }, { "epoch": 1.1765207655808498, "grad_norm": 1.6236350536346436, "learning_rate": 5.586600000000001e-06, "loss": 0.0485, "step": 22068 }, { "epoch": 1.1765740790105028, "grad_norm": 1.5925294160842896, "learning_rate": 5.5864e-06, "loss": 0.0106, "step": 22069 }, { "epoch": 1.1766273924401558, "grad_norm": 4.849842548370361, "learning_rate": 5.586200000000001e-06, "loss": 0.0303, "step": 22070 }, { "epoch": 1.1766807058698086, "grad_norm": 1.5152350664138794, "learning_rate": 5.586e-06, "loss": 0.0306, "step": 22071 }, { "epoch": 1.1767340192994615, "grad_norm": 1.9690428972244263, "learning_rate": 5.5858e-06, "loss": 0.0407, "step": 22072 }, { "epoch": 1.1767873327291145, "grad_norm": 2.590052843093872, "learning_rate": 5.5855999999999995e-06, "loss": 0.051, "step": 22073 }, { "epoch": 1.1768406461587675, "grad_norm": 1.9356769323349, "learning_rate": 5.585400000000001e-06, "loss": 0.018, "step": 22074 }, { "epoch": 1.1768939595884202, "grad_norm": 1.8634165525436401, "learning_rate": 5.5852e-06, "loss": 0.0235, "step": 22075 }, { "epoch": 1.1769472730180732, "grad_norm": 1.3733643293380737, "learning_rate": 5.585000000000001e-06, "loss": 0.0113, "step": 22076 }, { "epoch": 1.1770005864477262, "grad_norm": 1.0202616453170776, "learning_rate": 5.5848e-06, "loss": 0.0088, "step": 22077 }, { "epoch": 1.1770538998773792, "grad_norm": 2.186391830444336, "learning_rate": 5.5846e-06, "loss": 0.0458, "step": 22078 }, { "epoch": 1.177107213307032, "grad_norm": 2.3816254138946533, "learning_rate": 5.584400000000001e-06, "loss": 0.031, "step": 22079 }, { "epoch": 1.177160526736685, "grad_norm": 1.6096909046173096, "learning_rate": 5.584200000000001e-06, "loss": 0.0282, "step": 22080 }, { "epoch": 1.177213840166338, "grad_norm": 1.66361665725708, "learning_rate": 5.584e-06, "loss": 0.0192, "step": 22081 }, { "epoch": 1.1772671535959909, "grad_norm": 0.6197630763053894, "learning_rate": 5.583800000000001e-06, "loss": 0.0049, "step": 22082 }, { "epoch": 1.1773204670256439, "grad_norm": 1.6927917003631592, "learning_rate": 5.5836e-06, "loss": 0.0228, "step": 22083 }, { "epoch": 1.1773737804552966, "grad_norm": 2.8875532150268555, "learning_rate": 5.5834e-06, "loss": 0.0513, "step": 22084 }, { "epoch": 1.1774270938849496, "grad_norm": 1.647881031036377, "learning_rate": 5.583200000000001e-06, "loss": 0.0149, "step": 22085 }, { "epoch": 1.1774804073146026, "grad_norm": 1.9496699571609497, "learning_rate": 5.583000000000001e-06, "loss": 0.018, "step": 22086 }, { "epoch": 1.1775337207442556, "grad_norm": 2.259694814682007, "learning_rate": 5.5828e-06, "loss": 0.0186, "step": 22087 }, { "epoch": 1.1775870341739085, "grad_norm": 5.748145580291748, "learning_rate": 5.5826000000000006e-06, "loss": 0.078, "step": 22088 }, { "epoch": 1.1776403476035613, "grad_norm": 1.6876250505447388, "learning_rate": 5.5824e-06, "loss": 0.0339, "step": 22089 }, { "epoch": 1.1776936610332143, "grad_norm": 1.6361161470413208, "learning_rate": 5.582200000000001e-06, "loss": 0.0282, "step": 22090 }, { "epoch": 1.1777469744628672, "grad_norm": 2.320751190185547, "learning_rate": 5.582000000000001e-06, "loss": 0.0224, "step": 22091 }, { "epoch": 1.17780028789252, "grad_norm": 2.01157546043396, "learning_rate": 5.5818000000000005e-06, "loss": 0.0431, "step": 22092 }, { "epoch": 1.177853601322173, "grad_norm": 2.0168724060058594, "learning_rate": 5.5816e-06, "loss": 0.0201, "step": 22093 }, { "epoch": 1.177906914751826, "grad_norm": 2.1352901458740234, "learning_rate": 5.5814000000000005e-06, "loss": 0.0225, "step": 22094 }, { "epoch": 1.177960228181479, "grad_norm": 1.0341289043426514, "learning_rate": 5.5812e-06, "loss": 0.0093, "step": 22095 }, { "epoch": 1.178013541611132, "grad_norm": 2.3073558807373047, "learning_rate": 5.581000000000001e-06, "loss": 0.0298, "step": 22096 }, { "epoch": 1.1780668550407847, "grad_norm": 2.295128107070923, "learning_rate": 5.580800000000001e-06, "loss": 0.0188, "step": 22097 }, { "epoch": 1.1781201684704377, "grad_norm": 2.0635921955108643, "learning_rate": 5.5806e-06, "loss": 0.0176, "step": 22098 }, { "epoch": 1.1781734819000906, "grad_norm": 2.5308523178100586, "learning_rate": 5.5804e-06, "loss": 0.058, "step": 22099 }, { "epoch": 1.1782267953297436, "grad_norm": 1.4731086492538452, "learning_rate": 5.5802e-06, "loss": 0.0383, "step": 22100 }, { "epoch": 1.1782801087593966, "grad_norm": 1.9806848764419556, "learning_rate": 5.580000000000001e-06, "loss": 0.0545, "step": 22101 }, { "epoch": 1.1783334221890494, "grad_norm": 7.485795974731445, "learning_rate": 5.579800000000001e-06, "loss": 0.0454, "step": 22102 }, { "epoch": 1.1783867356187023, "grad_norm": 2.0580615997314453, "learning_rate": 5.579600000000001e-06, "loss": 0.0213, "step": 22103 }, { "epoch": 1.1784400490483553, "grad_norm": 1.8537039756774902, "learning_rate": 5.5794e-06, "loss": 0.0321, "step": 22104 }, { "epoch": 1.1784933624780083, "grad_norm": 0.5095940232276917, "learning_rate": 5.5792e-06, "loss": 0.0049, "step": 22105 }, { "epoch": 1.178546675907661, "grad_norm": 0.4189568758010864, "learning_rate": 5.579e-06, "loss": 0.0039, "step": 22106 }, { "epoch": 1.178599989337314, "grad_norm": 1.4115568399429321, "learning_rate": 5.578800000000001e-06, "loss": 0.0264, "step": 22107 }, { "epoch": 1.178653302766967, "grad_norm": 1.4220610857009888, "learning_rate": 5.578600000000001e-06, "loss": 0.0304, "step": 22108 }, { "epoch": 1.17870661619662, "grad_norm": 2.868162155151367, "learning_rate": 5.578400000000001e-06, "loss": 0.0446, "step": 22109 }, { "epoch": 1.1787599296262727, "grad_norm": 2.9286417961120605, "learning_rate": 5.5782e-06, "loss": 0.0212, "step": 22110 }, { "epoch": 1.1788132430559257, "grad_norm": 3.2116737365722656, "learning_rate": 5.578e-06, "loss": 0.0719, "step": 22111 }, { "epoch": 1.1788665564855787, "grad_norm": 7.385957717895508, "learning_rate": 5.577800000000001e-06, "loss": 0.0334, "step": 22112 }, { "epoch": 1.1789198699152317, "grad_norm": 1.623515009880066, "learning_rate": 5.577600000000001e-06, "loss": 0.0228, "step": 22113 }, { "epoch": 1.1789731833448847, "grad_norm": 1.4328123331069946, "learning_rate": 5.577400000000001e-06, "loss": 0.0151, "step": 22114 }, { "epoch": 1.1790264967745374, "grad_norm": 1.9191241264343262, "learning_rate": 5.5772000000000006e-06, "loss": 0.017, "step": 22115 }, { "epoch": 1.1790798102041904, "grad_norm": 1.4184882640838623, "learning_rate": 5.577e-06, "loss": 0.0094, "step": 22116 }, { "epoch": 1.1791331236338434, "grad_norm": 1.489481806755066, "learning_rate": 5.5768e-06, "loss": 0.0259, "step": 22117 }, { "epoch": 1.1791864370634964, "grad_norm": 1.834512710571289, "learning_rate": 5.576600000000001e-06, "loss": 0.0384, "step": 22118 }, { "epoch": 1.1792397504931493, "grad_norm": 1.1483838558197021, "learning_rate": 5.5764000000000005e-06, "loss": 0.0213, "step": 22119 }, { "epoch": 1.179293063922802, "grad_norm": 1.0648956298828125, "learning_rate": 5.576200000000001e-06, "loss": 0.0131, "step": 22120 }, { "epoch": 1.179346377352455, "grad_norm": 1.7002034187316895, "learning_rate": 5.5760000000000005e-06, "loss": 0.034, "step": 22121 }, { "epoch": 1.179399690782108, "grad_norm": 2.2887301445007324, "learning_rate": 5.5758e-06, "loss": 0.0215, "step": 22122 }, { "epoch": 1.179453004211761, "grad_norm": 1.6022624969482422, "learning_rate": 5.5756e-06, "loss": 0.0298, "step": 22123 }, { "epoch": 1.1795063176414138, "grad_norm": 3.2589378356933594, "learning_rate": 5.575400000000001e-06, "loss": 0.0557, "step": 22124 }, { "epoch": 1.1795596310710668, "grad_norm": 2.156416416168213, "learning_rate": 5.5752000000000004e-06, "loss": 0.0118, "step": 22125 }, { "epoch": 1.1796129445007197, "grad_norm": 1.6896687746047974, "learning_rate": 5.575000000000001e-06, "loss": 0.02, "step": 22126 }, { "epoch": 1.1796662579303727, "grad_norm": 1.9476548433303833, "learning_rate": 5.5748e-06, "loss": 0.03, "step": 22127 }, { "epoch": 1.1797195713600255, "grad_norm": 4.792440891265869, "learning_rate": 5.5746e-06, "loss": 0.0274, "step": 22128 }, { "epoch": 1.1797728847896785, "grad_norm": 1.0518463850021362, "learning_rate": 5.574400000000001e-06, "loss": 0.011, "step": 22129 }, { "epoch": 1.1798261982193314, "grad_norm": 1.7486515045166016, "learning_rate": 5.574200000000001e-06, "loss": 0.0286, "step": 22130 }, { "epoch": 1.1798795116489844, "grad_norm": 1.5103238821029663, "learning_rate": 5.574e-06, "loss": 0.0254, "step": 22131 }, { "epoch": 1.1799328250786374, "grad_norm": 2.7535207271575928, "learning_rate": 5.573800000000001e-06, "loss": 0.042, "step": 22132 }, { "epoch": 1.1799861385082901, "grad_norm": 3.292371988296509, "learning_rate": 5.5736e-06, "loss": 0.0442, "step": 22133 }, { "epoch": 1.1800394519379431, "grad_norm": 1.446707010269165, "learning_rate": 5.5734e-06, "loss": 0.0229, "step": 22134 }, { "epoch": 1.180092765367596, "grad_norm": 2.0510458946228027, "learning_rate": 5.573200000000001e-06, "loss": 0.0285, "step": 22135 }, { "epoch": 1.180146078797249, "grad_norm": 0.9750659465789795, "learning_rate": 5.573000000000001e-06, "loss": 0.0117, "step": 22136 }, { "epoch": 1.180199392226902, "grad_norm": 0.7494639754295349, "learning_rate": 5.5728e-06, "loss": 0.0068, "step": 22137 }, { "epoch": 1.1802527056565548, "grad_norm": 1.1752965450286865, "learning_rate": 5.572600000000001e-06, "loss": 0.0214, "step": 22138 }, { "epoch": 1.1803060190862078, "grad_norm": 1.195759892463684, "learning_rate": 5.5724e-06, "loss": 0.0129, "step": 22139 }, { "epoch": 1.1803593325158608, "grad_norm": 1.6857906579971313, "learning_rate": 5.572200000000001e-06, "loss": 0.0234, "step": 22140 }, { "epoch": 1.1804126459455138, "grad_norm": 0.9575778245925903, "learning_rate": 5.572000000000001e-06, "loss": 0.0189, "step": 22141 }, { "epoch": 1.1804659593751665, "grad_norm": 1.6395108699798584, "learning_rate": 5.571800000000001e-06, "loss": 0.0243, "step": 22142 }, { "epoch": 1.1805192728048195, "grad_norm": 4.47572660446167, "learning_rate": 5.5716e-06, "loss": 0.0337, "step": 22143 }, { "epoch": 1.1805725862344725, "grad_norm": 1.5879197120666504, "learning_rate": 5.5714000000000006e-06, "loss": 0.0304, "step": 22144 }, { "epoch": 1.1806258996641255, "grad_norm": 2.4628021717071533, "learning_rate": 5.5712e-06, "loss": 0.0396, "step": 22145 }, { "epoch": 1.1806792130937782, "grad_norm": 0.8961477875709534, "learning_rate": 5.5710000000000005e-06, "loss": 0.0064, "step": 22146 }, { "epoch": 1.1807325265234312, "grad_norm": 2.3555517196655273, "learning_rate": 5.570800000000001e-06, "loss": 0.0465, "step": 22147 }, { "epoch": 1.1807858399530842, "grad_norm": 2.341965675354004, "learning_rate": 5.5706000000000005e-06, "loss": 0.0418, "step": 22148 }, { "epoch": 1.1808391533827372, "grad_norm": 1.3247474431991577, "learning_rate": 5.5704e-06, "loss": 0.0174, "step": 22149 }, { "epoch": 1.1808924668123901, "grad_norm": 5.797983169555664, "learning_rate": 5.5702000000000005e-06, "loss": 0.0384, "step": 22150 }, { "epoch": 1.1809457802420429, "grad_norm": 1.5310672521591187, "learning_rate": 5.570000000000001e-06, "loss": 0.0143, "step": 22151 }, { "epoch": 1.1809990936716959, "grad_norm": 1.2840989828109741, "learning_rate": 5.5698000000000004e-06, "loss": 0.0115, "step": 22152 }, { "epoch": 1.1810524071013488, "grad_norm": 2.7020981311798096, "learning_rate": 5.569600000000001e-06, "loss": 0.0269, "step": 22153 }, { "epoch": 1.1811057205310018, "grad_norm": 3.521026849746704, "learning_rate": 5.5694e-06, "loss": 0.0265, "step": 22154 }, { "epoch": 1.1811590339606548, "grad_norm": 2.303945302963257, "learning_rate": 5.5692e-06, "loss": 0.0286, "step": 22155 }, { "epoch": 1.1812123473903076, "grad_norm": 1.0694552659988403, "learning_rate": 5.569e-06, "loss": 0.0284, "step": 22156 }, { "epoch": 1.1812656608199605, "grad_norm": 1.79022216796875, "learning_rate": 5.568800000000001e-06, "loss": 0.0254, "step": 22157 }, { "epoch": 1.1813189742496135, "grad_norm": 2.4130587577819824, "learning_rate": 5.5686e-06, "loss": 0.0473, "step": 22158 }, { "epoch": 1.1813722876792665, "grad_norm": 1.966435432434082, "learning_rate": 5.568400000000001e-06, "loss": 0.0502, "step": 22159 }, { "epoch": 1.1814256011089193, "grad_norm": 2.0594241619110107, "learning_rate": 5.5682e-06, "loss": 0.0305, "step": 22160 }, { "epoch": 1.1814789145385722, "grad_norm": 1.8369554281234741, "learning_rate": 5.568e-06, "loss": 0.0209, "step": 22161 }, { "epoch": 1.1815322279682252, "grad_norm": 1.61106276512146, "learning_rate": 5.567800000000001e-06, "loss": 0.0254, "step": 22162 }, { "epoch": 1.1815855413978782, "grad_norm": 1.8141801357269287, "learning_rate": 5.567600000000001e-06, "loss": 0.0389, "step": 22163 }, { "epoch": 1.181638854827531, "grad_norm": 2.3594822883605957, "learning_rate": 5.5674e-06, "loss": 0.0228, "step": 22164 }, { "epoch": 1.181692168257184, "grad_norm": 0.7645822763442993, "learning_rate": 5.567200000000001e-06, "loss": 0.0083, "step": 22165 }, { "epoch": 1.181745481686837, "grad_norm": 2.058065414428711, "learning_rate": 5.567e-06, "loss": 0.0312, "step": 22166 }, { "epoch": 1.1817987951164899, "grad_norm": 2.8200786113739014, "learning_rate": 5.5668e-06, "loss": 0.0181, "step": 22167 }, { "epoch": 1.1818521085461429, "grad_norm": 1.2878121137619019, "learning_rate": 5.566600000000001e-06, "loss": 0.0154, "step": 22168 }, { "epoch": 1.1819054219757956, "grad_norm": 2.1214120388031006, "learning_rate": 5.566400000000001e-06, "loss": 0.0479, "step": 22169 }, { "epoch": 1.1819587354054486, "grad_norm": 1.7919782400131226, "learning_rate": 5.5662e-06, "loss": 0.0144, "step": 22170 }, { "epoch": 1.1820120488351016, "grad_norm": 1.6422563791275024, "learning_rate": 5.566000000000001e-06, "loss": 0.0198, "step": 22171 }, { "epoch": 1.1820653622647546, "grad_norm": 3.2352888584136963, "learning_rate": 5.5658e-06, "loss": 0.0289, "step": 22172 }, { "epoch": 1.1821186756944075, "grad_norm": 1.996052622795105, "learning_rate": 5.565600000000001e-06, "loss": 0.037, "step": 22173 }, { "epoch": 1.1821719891240603, "grad_norm": 0.9540532827377319, "learning_rate": 5.565400000000001e-06, "loss": 0.0091, "step": 22174 }, { "epoch": 1.1822253025537133, "grad_norm": 0.796342134475708, "learning_rate": 5.5652000000000005e-06, "loss": 0.0083, "step": 22175 }, { "epoch": 1.1822786159833663, "grad_norm": 1.2056759595870972, "learning_rate": 5.565e-06, "loss": 0.0176, "step": 22176 }, { "epoch": 1.1823319294130192, "grad_norm": 1.789567470550537, "learning_rate": 5.5648000000000005e-06, "loss": 0.0291, "step": 22177 }, { "epoch": 1.182385242842672, "grad_norm": 2.2620203495025635, "learning_rate": 5.5646e-06, "loss": 0.0362, "step": 22178 }, { "epoch": 1.182438556272325, "grad_norm": 1.7478086948394775, "learning_rate": 5.564400000000001e-06, "loss": 0.0306, "step": 22179 }, { "epoch": 1.182491869701978, "grad_norm": 1.6175684928894043, "learning_rate": 5.564200000000001e-06, "loss": 0.0144, "step": 22180 }, { "epoch": 1.182545183131631, "grad_norm": 0.7040393352508545, "learning_rate": 5.5640000000000004e-06, "loss": 0.0055, "step": 22181 }, { "epoch": 1.1825984965612837, "grad_norm": 2.0332071781158447, "learning_rate": 5.5638e-06, "loss": 0.0117, "step": 22182 }, { "epoch": 1.1826518099909367, "grad_norm": 1.0338350534439087, "learning_rate": 5.5636e-06, "loss": 0.0073, "step": 22183 }, { "epoch": 1.1827051234205896, "grad_norm": 2.609295606613159, "learning_rate": 5.5634e-06, "loss": 0.0262, "step": 22184 }, { "epoch": 1.1827584368502426, "grad_norm": 1.018829345703125, "learning_rate": 5.563200000000001e-06, "loss": 0.01, "step": 22185 }, { "epoch": 1.1828117502798956, "grad_norm": 2.9554646015167236, "learning_rate": 5.563000000000001e-06, "loss": 0.0442, "step": 22186 }, { "epoch": 1.1828650637095484, "grad_norm": 3.113781690597534, "learning_rate": 5.5628e-06, "loss": 0.0485, "step": 22187 }, { "epoch": 1.1829183771392013, "grad_norm": 1.7261736392974854, "learning_rate": 5.5626e-06, "loss": 0.0224, "step": 22188 }, { "epoch": 1.1829716905688543, "grad_norm": 2.2617061138153076, "learning_rate": 5.5624e-06, "loss": 0.0394, "step": 22189 }, { "epoch": 1.1830250039985073, "grad_norm": 1.4953653812408447, "learning_rate": 5.562200000000001e-06, "loss": 0.0214, "step": 22190 }, { "epoch": 1.1830783174281603, "grad_norm": 1.1056945323944092, "learning_rate": 5.562000000000001e-06, "loss": 0.0129, "step": 22191 }, { "epoch": 1.183131630857813, "grad_norm": 2.880192995071411, "learning_rate": 5.561800000000001e-06, "loss": 0.0383, "step": 22192 }, { "epoch": 1.183184944287466, "grad_norm": 1.7753849029541016, "learning_rate": 5.5616e-06, "loss": 0.0152, "step": 22193 }, { "epoch": 1.183238257717119, "grad_norm": 0.8696669340133667, "learning_rate": 5.5614e-06, "loss": 0.0099, "step": 22194 }, { "epoch": 1.1832915711467717, "grad_norm": 1.8173552751541138, "learning_rate": 5.5612e-06, "loss": 0.0107, "step": 22195 }, { "epoch": 1.1833448845764247, "grad_norm": 2.6568236351013184, "learning_rate": 5.561000000000001e-06, "loss": 0.018, "step": 22196 }, { "epoch": 1.1833981980060777, "grad_norm": 1.7471332550048828, "learning_rate": 5.560800000000001e-06, "loss": 0.0315, "step": 22197 }, { "epoch": 1.1834515114357307, "grad_norm": 2.339381694793701, "learning_rate": 5.560600000000001e-06, "loss": 0.0536, "step": 22198 }, { "epoch": 1.1835048248653837, "grad_norm": 3.981140613555908, "learning_rate": 5.5604e-06, "loss": 0.062, "step": 22199 }, { "epoch": 1.1835581382950364, "grad_norm": 2.4263317584991455, "learning_rate": 5.5602e-06, "loss": 0.0347, "step": 22200 }, { "epoch": 1.1836114517246894, "grad_norm": 1.7312263250350952, "learning_rate": 5.560000000000001e-06, "loss": 0.0191, "step": 22201 }, { "epoch": 1.1836647651543424, "grad_norm": 1.236971139907837, "learning_rate": 5.5598000000000005e-06, "loss": 0.0138, "step": 22202 }, { "epoch": 1.1837180785839954, "grad_norm": 1.5965890884399414, "learning_rate": 5.559600000000001e-06, "loss": 0.0193, "step": 22203 }, { "epoch": 1.1837713920136483, "grad_norm": 3.1231491565704346, "learning_rate": 5.5594000000000005e-06, "loss": 0.067, "step": 22204 }, { "epoch": 1.183824705443301, "grad_norm": 0.9053798913955688, "learning_rate": 5.5592e-06, "loss": 0.014, "step": 22205 }, { "epoch": 1.183878018872954, "grad_norm": 2.9910783767700195, "learning_rate": 5.559e-06, "loss": 0.0346, "step": 22206 }, { "epoch": 1.183931332302607, "grad_norm": 2.0770092010498047, "learning_rate": 5.558800000000001e-06, "loss": 0.0228, "step": 22207 }, { "epoch": 1.18398464573226, "grad_norm": 0.5628650784492493, "learning_rate": 5.5586000000000004e-06, "loss": 0.0054, "step": 22208 }, { "epoch": 1.1840379591619128, "grad_norm": 1.3112674951553345, "learning_rate": 5.558400000000001e-06, "loss": 0.0348, "step": 22209 }, { "epoch": 1.1840912725915658, "grad_norm": 2.370802164077759, "learning_rate": 5.5582e-06, "loss": 0.053, "step": 22210 }, { "epoch": 1.1841445860212187, "grad_norm": 2.3710784912109375, "learning_rate": 5.558e-06, "loss": 0.0262, "step": 22211 }, { "epoch": 1.1841978994508717, "grad_norm": 1.671547532081604, "learning_rate": 5.557800000000001e-06, "loss": 0.0253, "step": 22212 }, { "epoch": 1.1842512128805245, "grad_norm": 2.703314781188965, "learning_rate": 5.557600000000001e-06, "loss": 0.0378, "step": 22213 }, { "epoch": 1.1843045263101775, "grad_norm": 2.3633310794830322, "learning_rate": 5.5574e-06, "loss": 0.0221, "step": 22214 }, { "epoch": 1.1843578397398304, "grad_norm": 2.035850763320923, "learning_rate": 5.557200000000001e-06, "loss": 0.0218, "step": 22215 }, { "epoch": 1.1844111531694834, "grad_norm": 1.0083999633789062, "learning_rate": 5.557e-06, "loss": 0.0091, "step": 22216 }, { "epoch": 1.1844644665991364, "grad_norm": 0.8473044037818909, "learning_rate": 5.5568e-06, "loss": 0.0133, "step": 22217 }, { "epoch": 1.1845177800287892, "grad_norm": 2.0906012058258057, "learning_rate": 5.556600000000001e-06, "loss": 0.0324, "step": 22218 }, { "epoch": 1.1845710934584421, "grad_norm": 0.7421972751617432, "learning_rate": 5.556400000000001e-06, "loss": 0.0063, "step": 22219 }, { "epoch": 1.1846244068880951, "grad_norm": 1.9545494318008423, "learning_rate": 5.5562e-06, "loss": 0.0396, "step": 22220 }, { "epoch": 1.184677720317748, "grad_norm": 1.9235044717788696, "learning_rate": 5.556000000000001e-06, "loss": 0.0238, "step": 22221 }, { "epoch": 1.184731033747401, "grad_norm": 1.1717281341552734, "learning_rate": 5.5558e-06, "loss": 0.0118, "step": 22222 }, { "epoch": 1.1847843471770538, "grad_norm": 2.929938554763794, "learning_rate": 5.555600000000001e-06, "loss": 0.0471, "step": 22223 }, { "epoch": 1.1848376606067068, "grad_norm": 1.6671802997589111, "learning_rate": 5.555400000000001e-06, "loss": 0.0231, "step": 22224 }, { "epoch": 1.1848909740363598, "grad_norm": 0.8305699229240417, "learning_rate": 5.555200000000001e-06, "loss": 0.0319, "step": 22225 }, { "epoch": 1.1849442874660128, "grad_norm": 2.967019557952881, "learning_rate": 5.555e-06, "loss": 0.0321, "step": 22226 }, { "epoch": 1.1849976008956655, "grad_norm": 1.480569839477539, "learning_rate": 5.554800000000001e-06, "loss": 0.0214, "step": 22227 }, { "epoch": 1.1850509143253185, "grad_norm": 2.0319015979766846, "learning_rate": 5.5546e-06, "loss": 0.0306, "step": 22228 }, { "epoch": 1.1851042277549715, "grad_norm": 0.9121373891830444, "learning_rate": 5.5544000000000006e-06, "loss": 0.0159, "step": 22229 }, { "epoch": 1.1851575411846245, "grad_norm": 0.7221060991287231, "learning_rate": 5.554200000000001e-06, "loss": 0.0059, "step": 22230 }, { "epoch": 1.1852108546142772, "grad_norm": 1.685892939567566, "learning_rate": 5.5540000000000005e-06, "loss": 0.0301, "step": 22231 }, { "epoch": 1.1852641680439302, "grad_norm": 1.1940531730651855, "learning_rate": 5.5538e-06, "loss": 0.0098, "step": 22232 }, { "epoch": 1.1853174814735832, "grad_norm": 2.986586093902588, "learning_rate": 5.5536000000000005e-06, "loss": 0.0555, "step": 22233 }, { "epoch": 1.1853707949032362, "grad_norm": 2.014965295791626, "learning_rate": 5.5534e-06, "loss": 0.04, "step": 22234 }, { "epoch": 1.1854241083328891, "grad_norm": 2.4357402324676514, "learning_rate": 5.5532000000000005e-06, "loss": 0.0576, "step": 22235 }, { "epoch": 1.185477421762542, "grad_norm": 2.5284693241119385, "learning_rate": 5.553000000000001e-06, "loss": 0.0291, "step": 22236 }, { "epoch": 1.1855307351921949, "grad_norm": 3.1760473251342773, "learning_rate": 5.5528000000000004e-06, "loss": 0.0198, "step": 22237 }, { "epoch": 1.1855840486218479, "grad_norm": 2.367433547973633, "learning_rate": 5.5526e-06, "loss": 0.043, "step": 22238 }, { "epoch": 1.1856373620515008, "grad_norm": 1.739169955253601, "learning_rate": 5.5524e-06, "loss": 0.0299, "step": 22239 }, { "epoch": 1.1856906754811538, "grad_norm": 2.8542160987854004, "learning_rate": 5.552200000000001e-06, "loss": 0.0223, "step": 22240 }, { "epoch": 1.1857439889108066, "grad_norm": 2.466233491897583, "learning_rate": 5.552e-06, "loss": 0.0349, "step": 22241 }, { "epoch": 1.1857973023404595, "grad_norm": 1.1996738910675049, "learning_rate": 5.551800000000001e-06, "loss": 0.018, "step": 22242 }, { "epoch": 1.1858506157701125, "grad_norm": 1.6884665489196777, "learning_rate": 5.5516e-06, "loss": 0.0297, "step": 22243 }, { "epoch": 1.1859039291997655, "grad_norm": 1.903269648551941, "learning_rate": 5.5514e-06, "loss": 0.0287, "step": 22244 }, { "epoch": 1.1859572426294183, "grad_norm": 2.677511692047119, "learning_rate": 5.5512e-06, "loss": 0.047, "step": 22245 }, { "epoch": 1.1860105560590712, "grad_norm": 2.4261081218719482, "learning_rate": 5.551000000000001e-06, "loss": 0.0602, "step": 22246 }, { "epoch": 1.1860638694887242, "grad_norm": 1.3183683156967163, "learning_rate": 5.5508e-06, "loss": 0.0107, "step": 22247 }, { "epoch": 1.1861171829183772, "grad_norm": 1.94599449634552, "learning_rate": 5.550600000000001e-06, "loss": 0.0436, "step": 22248 }, { "epoch": 1.18617049634803, "grad_norm": 2.987577199935913, "learning_rate": 5.5504e-06, "loss": 0.0646, "step": 22249 }, { "epoch": 1.186223809777683, "grad_norm": 2.270620346069336, "learning_rate": 5.5502e-06, "loss": 0.0222, "step": 22250 }, { "epoch": 1.186277123207336, "grad_norm": 1.5813778638839722, "learning_rate": 5.550000000000001e-06, "loss": 0.014, "step": 22251 }, { "epoch": 1.186330436636989, "grad_norm": 2.156839370727539, "learning_rate": 5.549800000000001e-06, "loss": 0.0479, "step": 22252 }, { "epoch": 1.1863837500666419, "grad_norm": 2.371589422225952, "learning_rate": 5.5496e-06, "loss": 0.0249, "step": 22253 }, { "epoch": 1.1864370634962946, "grad_norm": 1.300119161605835, "learning_rate": 5.549400000000001e-06, "loss": 0.0196, "step": 22254 }, { "epoch": 1.1864903769259476, "grad_norm": 1.4385385513305664, "learning_rate": 5.5492e-06, "loss": 0.028, "step": 22255 }, { "epoch": 1.1865436903556006, "grad_norm": 1.323918342590332, "learning_rate": 5.549e-06, "loss": 0.0118, "step": 22256 }, { "epoch": 1.1865970037852536, "grad_norm": 2.112537145614624, "learning_rate": 5.548800000000001e-06, "loss": 0.0408, "step": 22257 }, { "epoch": 1.1866503172149065, "grad_norm": 2.5045738220214844, "learning_rate": 5.5486000000000005e-06, "loss": 0.0207, "step": 22258 }, { "epoch": 1.1867036306445593, "grad_norm": 1.701697826385498, "learning_rate": 5.5484e-06, "loss": 0.0556, "step": 22259 }, { "epoch": 1.1867569440742123, "grad_norm": 0.9785699844360352, "learning_rate": 5.5482000000000005e-06, "loss": 0.0152, "step": 22260 }, { "epoch": 1.1868102575038653, "grad_norm": 0.9755359888076782, "learning_rate": 5.548e-06, "loss": 0.0153, "step": 22261 }, { "epoch": 1.1868635709335182, "grad_norm": 2.8411991596221924, "learning_rate": 5.547800000000001e-06, "loss": 0.0614, "step": 22262 }, { "epoch": 1.186916884363171, "grad_norm": 1.8034578561782837, "learning_rate": 5.547600000000001e-06, "loss": 0.0409, "step": 22263 }, { "epoch": 1.186970197792824, "grad_norm": 2.2594752311706543, "learning_rate": 5.5474000000000005e-06, "loss": 0.0288, "step": 22264 }, { "epoch": 1.187023511222477, "grad_norm": 1.5786411762237549, "learning_rate": 5.5472e-06, "loss": 0.0157, "step": 22265 }, { "epoch": 1.18707682465213, "grad_norm": 2.3412413597106934, "learning_rate": 5.547e-06, "loss": 0.0196, "step": 22266 }, { "epoch": 1.1871301380817827, "grad_norm": 1.4820367097854614, "learning_rate": 5.5468e-06, "loss": 0.0202, "step": 22267 }, { "epoch": 1.1871834515114357, "grad_norm": 1.5832149982452393, "learning_rate": 5.546600000000001e-06, "loss": 0.0159, "step": 22268 }, { "epoch": 1.1872367649410887, "grad_norm": 1.0445002317428589, "learning_rate": 5.546400000000001e-06, "loss": 0.0098, "step": 22269 }, { "epoch": 1.1872900783707416, "grad_norm": 1.7688024044036865, "learning_rate": 5.5462e-06, "loss": 0.0227, "step": 22270 }, { "epoch": 1.1873433918003946, "grad_norm": 2.245958089828491, "learning_rate": 5.546e-06, "loss": 0.0195, "step": 22271 }, { "epoch": 1.1873967052300474, "grad_norm": 2.492466926574707, "learning_rate": 5.5458e-06, "loss": 0.037, "step": 22272 }, { "epoch": 1.1874500186597003, "grad_norm": 2.3220605850219727, "learning_rate": 5.545600000000001e-06, "loss": 0.0279, "step": 22273 }, { "epoch": 1.1875033320893533, "grad_norm": 2.177532196044922, "learning_rate": 5.545400000000001e-06, "loss": 0.0243, "step": 22274 }, { "epoch": 1.1875566455190063, "grad_norm": 1.5249173641204834, "learning_rate": 5.545200000000001e-06, "loss": 0.0251, "step": 22275 }, { "epoch": 1.1876099589486593, "grad_norm": 2.0551345348358154, "learning_rate": 5.545e-06, "loss": 0.0112, "step": 22276 }, { "epoch": 1.187663272378312, "grad_norm": 1.1071674823760986, "learning_rate": 5.5448e-06, "loss": 0.0129, "step": 22277 }, { "epoch": 1.187716585807965, "grad_norm": 1.5158252716064453, "learning_rate": 5.5446e-06, "loss": 0.0159, "step": 22278 }, { "epoch": 1.187769899237618, "grad_norm": 0.776157557964325, "learning_rate": 5.544400000000001e-06, "loss": 0.0064, "step": 22279 }, { "epoch": 1.187823212667271, "grad_norm": 3.1531641483306885, "learning_rate": 5.544200000000001e-06, "loss": 0.0171, "step": 22280 }, { "epoch": 1.1878765260969237, "grad_norm": 1.9950605630874634, "learning_rate": 5.544000000000001e-06, "loss": 0.0208, "step": 22281 }, { "epoch": 1.1879298395265767, "grad_norm": 0.8759029507637024, "learning_rate": 5.5438e-06, "loss": 0.0037, "step": 22282 }, { "epoch": 1.1879831529562297, "grad_norm": 1.8593056201934814, "learning_rate": 5.5436e-06, "loss": 0.0515, "step": 22283 }, { "epoch": 1.1880364663858827, "grad_norm": 0.8601017594337463, "learning_rate": 5.543400000000001e-06, "loss": 0.0157, "step": 22284 }, { "epoch": 1.1880897798155354, "grad_norm": 1.3253322839736938, "learning_rate": 5.5432000000000006e-06, "loss": 0.027, "step": 22285 }, { "epoch": 1.1881430932451884, "grad_norm": 4.443113803863525, "learning_rate": 5.543000000000001e-06, "loss": 0.0502, "step": 22286 }, { "epoch": 1.1881964066748414, "grad_norm": 2.0027947425842285, "learning_rate": 5.5428000000000005e-06, "loss": 0.0323, "step": 22287 }, { "epoch": 1.1882497201044944, "grad_norm": 2.4708778858184814, "learning_rate": 5.5426e-06, "loss": 0.0414, "step": 22288 }, { "epoch": 1.1883030335341473, "grad_norm": 1.6667534112930298, "learning_rate": 5.5424e-06, "loss": 0.0469, "step": 22289 }, { "epoch": 1.1883563469638, "grad_norm": 2.209353446960449, "learning_rate": 5.542200000000001e-06, "loss": 0.02, "step": 22290 }, { "epoch": 1.188409660393453, "grad_norm": 2.7537765502929688, "learning_rate": 5.5420000000000005e-06, "loss": 0.0336, "step": 22291 }, { "epoch": 1.188462973823106, "grad_norm": 4.891964912414551, "learning_rate": 5.541800000000001e-06, "loss": 0.0225, "step": 22292 }, { "epoch": 1.188516287252759, "grad_norm": 1.7706103324890137, "learning_rate": 5.5416000000000004e-06, "loss": 0.0273, "step": 22293 }, { "epoch": 1.188569600682412, "grad_norm": 2.002413749694824, "learning_rate": 5.5414e-06, "loss": 0.0324, "step": 22294 }, { "epoch": 1.1886229141120648, "grad_norm": 1.3240718841552734, "learning_rate": 5.5411999999999996e-06, "loss": 0.017, "step": 22295 }, { "epoch": 1.1886762275417178, "grad_norm": 1.0145872831344604, "learning_rate": 5.541000000000001e-06, "loss": 0.0099, "step": 22296 }, { "epoch": 1.1887295409713707, "grad_norm": 1.5739432573318481, "learning_rate": 5.5408e-06, "loss": 0.0223, "step": 22297 }, { "epoch": 1.1887828544010237, "grad_norm": 2.429835319519043, "learning_rate": 5.540600000000001e-06, "loss": 0.0245, "step": 22298 }, { "epoch": 1.1888361678306765, "grad_norm": 1.1705896854400635, "learning_rate": 5.5404e-06, "loss": 0.0193, "step": 22299 }, { "epoch": 1.1888894812603295, "grad_norm": 1.0862573385238647, "learning_rate": 5.5402e-06, "loss": 0.0093, "step": 22300 }, { "epoch": 1.1889427946899824, "grad_norm": 1.3591946363449097, "learning_rate": 5.540000000000001e-06, "loss": 0.0096, "step": 22301 }, { "epoch": 1.1889961081196354, "grad_norm": 0.5901730060577393, "learning_rate": 5.539800000000001e-06, "loss": 0.006, "step": 22302 }, { "epoch": 1.1890494215492882, "grad_norm": 1.1668882369995117, "learning_rate": 5.5396e-06, "loss": 0.0102, "step": 22303 }, { "epoch": 1.1891027349789411, "grad_norm": 1.6423546075820923, "learning_rate": 5.539400000000001e-06, "loss": 0.0288, "step": 22304 }, { "epoch": 1.1891560484085941, "grad_norm": 2.1342403888702393, "learning_rate": 5.5392e-06, "loss": 0.0274, "step": 22305 }, { "epoch": 1.189209361838247, "grad_norm": 1.3416939973831177, "learning_rate": 5.539e-06, "loss": 0.0213, "step": 22306 }, { "epoch": 1.1892626752679, "grad_norm": 0.4158673882484436, "learning_rate": 5.538800000000001e-06, "loss": 0.0043, "step": 22307 }, { "epoch": 1.1893159886975528, "grad_norm": 2.873518943786621, "learning_rate": 5.538600000000001e-06, "loss": 0.0461, "step": 22308 }, { "epoch": 1.1893693021272058, "grad_norm": 1.726180076599121, "learning_rate": 5.5384e-06, "loss": 0.0226, "step": 22309 }, { "epoch": 1.1894226155568588, "grad_norm": 1.8323262929916382, "learning_rate": 5.538200000000001e-06, "loss": 0.0349, "step": 22310 }, { "epoch": 1.1894759289865118, "grad_norm": 0.8252978324890137, "learning_rate": 5.538e-06, "loss": 0.0072, "step": 22311 }, { "epoch": 1.1895292424161648, "grad_norm": 0.7505606412887573, "learning_rate": 5.537800000000001e-06, "loss": 0.0095, "step": 22312 }, { "epoch": 1.1895825558458175, "grad_norm": 2.435508966445923, "learning_rate": 5.537600000000001e-06, "loss": 0.0455, "step": 22313 }, { "epoch": 1.1896358692754705, "grad_norm": 0.9380472302436829, "learning_rate": 5.5374000000000005e-06, "loss": 0.0186, "step": 22314 }, { "epoch": 1.1896891827051235, "grad_norm": 0.8610110282897949, "learning_rate": 5.5372e-06, "loss": 0.0084, "step": 22315 }, { "epoch": 1.1897424961347762, "grad_norm": 3.0038821697235107, "learning_rate": 5.5370000000000005e-06, "loss": 0.0318, "step": 22316 }, { "epoch": 1.1897958095644292, "grad_norm": 2.0322539806365967, "learning_rate": 5.5368e-06, "loss": 0.0371, "step": 22317 }, { "epoch": 1.1898491229940822, "grad_norm": 1.1613904237747192, "learning_rate": 5.5366000000000005e-06, "loss": 0.0101, "step": 22318 }, { "epoch": 1.1899024364237352, "grad_norm": 1.866976022720337, "learning_rate": 5.536400000000001e-06, "loss": 0.0331, "step": 22319 }, { "epoch": 1.1899557498533881, "grad_norm": 1.8502004146575928, "learning_rate": 5.5362000000000005e-06, "loss": 0.0126, "step": 22320 }, { "epoch": 1.190009063283041, "grad_norm": 1.7218252420425415, "learning_rate": 5.536e-06, "loss": 0.0356, "step": 22321 }, { "epoch": 1.1900623767126939, "grad_norm": 1.466142177581787, "learning_rate": 5.5358000000000004e-06, "loss": 0.015, "step": 22322 }, { "epoch": 1.1901156901423469, "grad_norm": 1.2251404523849487, "learning_rate": 5.535600000000001e-06, "loss": 0.01, "step": 22323 }, { "epoch": 1.1901690035719998, "grad_norm": 1.409672737121582, "learning_rate": 5.5354e-06, "loss": 0.0234, "step": 22324 }, { "epoch": 1.1902223170016528, "grad_norm": 1.9134522676467896, "learning_rate": 5.535200000000001e-06, "loss": 0.022, "step": 22325 }, { "epoch": 1.1902756304313056, "grad_norm": 1.728039264678955, "learning_rate": 5.535e-06, "loss": 0.0166, "step": 22326 }, { "epoch": 1.1903289438609586, "grad_norm": 2.759195327758789, "learning_rate": 5.5348e-06, "loss": 0.0547, "step": 22327 }, { "epoch": 1.1903822572906115, "grad_norm": 1.365250825881958, "learning_rate": 5.5346e-06, "loss": 0.039, "step": 22328 }, { "epoch": 1.1904355707202645, "grad_norm": 0.9738462567329407, "learning_rate": 5.534400000000001e-06, "loss": 0.0219, "step": 22329 }, { "epoch": 1.1904888841499173, "grad_norm": 2.1016528606414795, "learning_rate": 5.5342e-06, "loss": 0.0211, "step": 22330 }, { "epoch": 1.1905421975795702, "grad_norm": 2.726902484893799, "learning_rate": 5.534000000000001e-06, "loss": 0.0339, "step": 22331 }, { "epoch": 1.1905955110092232, "grad_norm": 1.5018608570098877, "learning_rate": 5.5338e-06, "loss": 0.0232, "step": 22332 }, { "epoch": 1.1906488244388762, "grad_norm": 1.5297688245773315, "learning_rate": 5.5336e-06, "loss": 0.0205, "step": 22333 }, { "epoch": 1.190702137868529, "grad_norm": 2.1045920848846436, "learning_rate": 5.533400000000001e-06, "loss": 0.0374, "step": 22334 }, { "epoch": 1.190755451298182, "grad_norm": 1.456886887550354, "learning_rate": 5.533200000000001e-06, "loss": 0.0213, "step": 22335 }, { "epoch": 1.190808764727835, "grad_norm": 2.124267578125, "learning_rate": 5.533e-06, "loss": 0.023, "step": 22336 }, { "epoch": 1.190862078157488, "grad_norm": 1.776940941810608, "learning_rate": 5.532800000000001e-06, "loss": 0.0401, "step": 22337 }, { "epoch": 1.1909153915871409, "grad_norm": 1.7684437036514282, "learning_rate": 5.5326e-06, "loss": 0.0172, "step": 22338 }, { "epoch": 1.1909687050167936, "grad_norm": 0.7936887145042419, "learning_rate": 5.5324e-06, "loss": 0.0125, "step": 22339 }, { "epoch": 1.1910220184464466, "grad_norm": 2.9990861415863037, "learning_rate": 5.532200000000001e-06, "loss": 0.0396, "step": 22340 }, { "epoch": 1.1910753318760996, "grad_norm": 1.7408503293991089, "learning_rate": 5.5320000000000006e-06, "loss": 0.0239, "step": 22341 }, { "epoch": 1.1911286453057526, "grad_norm": 1.9394551515579224, "learning_rate": 5.5318e-06, "loss": 0.0259, "step": 22342 }, { "epoch": 1.1911819587354056, "grad_norm": 1.6570549011230469, "learning_rate": 5.5316000000000005e-06, "loss": 0.0166, "step": 22343 }, { "epoch": 1.1912352721650583, "grad_norm": 0.936711847782135, "learning_rate": 5.5314e-06, "loss": 0.0081, "step": 22344 }, { "epoch": 1.1912885855947113, "grad_norm": 1.3665019273757935, "learning_rate": 5.5312e-06, "loss": 0.0155, "step": 22345 }, { "epoch": 1.1913418990243643, "grad_norm": 3.2144062519073486, "learning_rate": 5.531000000000001e-06, "loss": 0.0348, "step": 22346 }, { "epoch": 1.1913952124540172, "grad_norm": 1.336164116859436, "learning_rate": 5.5308000000000005e-06, "loss": 0.0256, "step": 22347 }, { "epoch": 1.19144852588367, "grad_norm": 1.198807716369629, "learning_rate": 5.5306e-06, "loss": 0.0213, "step": 22348 }, { "epoch": 1.191501839313323, "grad_norm": 1.4234040975570679, "learning_rate": 5.5304000000000004e-06, "loss": 0.0188, "step": 22349 }, { "epoch": 1.191555152742976, "grad_norm": 1.697060227394104, "learning_rate": 5.5302e-06, "loss": 0.0292, "step": 22350 }, { "epoch": 1.191608466172629, "grad_norm": 2.120278835296631, "learning_rate": 5.530000000000001e-06, "loss": 0.0163, "step": 22351 }, { "epoch": 1.1916617796022817, "grad_norm": 2.592898368835449, "learning_rate": 5.529800000000001e-06, "loss": 0.0354, "step": 22352 }, { "epoch": 1.1917150930319347, "grad_norm": 1.5610308647155762, "learning_rate": 5.5296e-06, "loss": 0.0165, "step": 22353 }, { "epoch": 1.1917684064615877, "grad_norm": 1.7869318723678589, "learning_rate": 5.5294e-06, "loss": 0.0305, "step": 22354 }, { "epoch": 1.1918217198912406, "grad_norm": 2.3330912590026855, "learning_rate": 5.5292e-06, "loss": 0.0186, "step": 22355 }, { "epoch": 1.1918750333208936, "grad_norm": 1.8901768922805786, "learning_rate": 5.529e-06, "loss": 0.0388, "step": 22356 }, { "epoch": 1.1919283467505464, "grad_norm": 1.6479147672653198, "learning_rate": 5.528800000000001e-06, "loss": 0.0353, "step": 22357 }, { "epoch": 1.1919816601801994, "grad_norm": 1.300178050994873, "learning_rate": 5.528600000000001e-06, "loss": 0.0171, "step": 22358 }, { "epoch": 1.1920349736098523, "grad_norm": 2.619619607925415, "learning_rate": 5.5284e-06, "loss": 0.025, "step": 22359 }, { "epoch": 1.1920882870395053, "grad_norm": 5.556741237640381, "learning_rate": 5.5282e-06, "loss": 0.0412, "step": 22360 }, { "epoch": 1.1921416004691583, "grad_norm": 1.6673429012298584, "learning_rate": 5.528e-06, "loss": 0.0139, "step": 22361 }, { "epoch": 1.192194913898811, "grad_norm": 2.6922006607055664, "learning_rate": 5.527800000000001e-06, "loss": 0.0156, "step": 22362 }, { "epoch": 1.192248227328464, "grad_norm": 1.956214189529419, "learning_rate": 5.527600000000001e-06, "loss": 0.0268, "step": 22363 }, { "epoch": 1.192301540758117, "grad_norm": 4.8180155754089355, "learning_rate": 5.527400000000001e-06, "loss": 0.0692, "step": 22364 }, { "epoch": 1.19235485418777, "grad_norm": 1.3662514686584473, "learning_rate": 5.5272e-06, "loss": 0.0195, "step": 22365 }, { "epoch": 1.1924081676174227, "grad_norm": 1.5627849102020264, "learning_rate": 5.527e-06, "loss": 0.0136, "step": 22366 }, { "epoch": 1.1924614810470757, "grad_norm": 2.235231876373291, "learning_rate": 5.5268e-06, "loss": 0.0151, "step": 22367 }, { "epoch": 1.1925147944767287, "grad_norm": 1.407765507698059, "learning_rate": 5.526600000000001e-06, "loss": 0.0209, "step": 22368 }, { "epoch": 1.1925681079063817, "grad_norm": 2.246593952178955, "learning_rate": 5.526400000000001e-06, "loss": 0.0218, "step": 22369 }, { "epoch": 1.1926214213360344, "grad_norm": 0.8278144598007202, "learning_rate": 5.5262000000000006e-06, "loss": 0.0169, "step": 22370 }, { "epoch": 1.1926747347656874, "grad_norm": 1.993829607963562, "learning_rate": 5.526e-06, "loss": 0.0336, "step": 22371 }, { "epoch": 1.1927280481953404, "grad_norm": 1.797163486480713, "learning_rate": 5.5258e-06, "loss": 0.0242, "step": 22372 }, { "epoch": 1.1927813616249934, "grad_norm": 1.125679850578308, "learning_rate": 5.525600000000001e-06, "loss": 0.0087, "step": 22373 }, { "epoch": 1.1928346750546464, "grad_norm": 0.6774839758872986, "learning_rate": 5.5254000000000005e-06, "loss": 0.0206, "step": 22374 }, { "epoch": 1.1928879884842991, "grad_norm": 2.493467092514038, "learning_rate": 5.525200000000001e-06, "loss": 0.0419, "step": 22375 }, { "epoch": 1.192941301913952, "grad_norm": 1.060982346534729, "learning_rate": 5.5250000000000005e-06, "loss": 0.0206, "step": 22376 }, { "epoch": 1.192994615343605, "grad_norm": 1.1526343822479248, "learning_rate": 5.5248e-06, "loss": 0.0079, "step": 22377 }, { "epoch": 1.193047928773258, "grad_norm": 0.6046289801597595, "learning_rate": 5.5246e-06, "loss": 0.0054, "step": 22378 }, { "epoch": 1.193101242202911, "grad_norm": 2.917821168899536, "learning_rate": 5.524400000000001e-06, "loss": 0.0551, "step": 22379 }, { "epoch": 1.1931545556325638, "grad_norm": 1.223933458328247, "learning_rate": 5.5242e-06, "loss": 0.008, "step": 22380 }, { "epoch": 1.1932078690622168, "grad_norm": 3.9647982120513916, "learning_rate": 5.524000000000001e-06, "loss": 0.0298, "step": 22381 }, { "epoch": 1.1932611824918697, "grad_norm": 0.9189419746398926, "learning_rate": 5.5238e-06, "loss": 0.0108, "step": 22382 }, { "epoch": 1.1933144959215227, "grad_norm": 3.6926777362823486, "learning_rate": 5.5236e-06, "loss": 0.0626, "step": 22383 }, { "epoch": 1.1933678093511755, "grad_norm": 2.588305711746216, "learning_rate": 5.523400000000001e-06, "loss": 0.0413, "step": 22384 }, { "epoch": 1.1934211227808285, "grad_norm": 0.8770157098770142, "learning_rate": 5.523200000000001e-06, "loss": 0.012, "step": 22385 }, { "epoch": 1.1934744362104814, "grad_norm": 1.5907834768295288, "learning_rate": 5.523e-06, "loss": 0.021, "step": 22386 }, { "epoch": 1.1935277496401344, "grad_norm": 1.4190218448638916, "learning_rate": 5.522800000000001e-06, "loss": 0.0153, "step": 22387 }, { "epoch": 1.1935810630697872, "grad_norm": 5.7587080001831055, "learning_rate": 5.5226e-06, "loss": 0.0377, "step": 22388 }, { "epoch": 1.1936343764994402, "grad_norm": 1.2400493621826172, "learning_rate": 5.5224e-06, "loss": 0.0112, "step": 22389 }, { "epoch": 1.1936876899290931, "grad_norm": 1.199762225151062, "learning_rate": 5.522200000000001e-06, "loss": 0.0094, "step": 22390 }, { "epoch": 1.1937410033587461, "grad_norm": 3.157139301300049, "learning_rate": 5.522000000000001e-06, "loss": 0.0353, "step": 22391 }, { "epoch": 1.193794316788399, "grad_norm": 1.9167836904525757, "learning_rate": 5.5218e-06, "loss": 0.0214, "step": 22392 }, { "epoch": 1.1938476302180518, "grad_norm": 1.7966898679733276, "learning_rate": 5.521600000000001e-06, "loss": 0.0221, "step": 22393 }, { "epoch": 1.1939009436477048, "grad_norm": 1.4985966682434082, "learning_rate": 5.5214e-06, "loss": 0.04, "step": 22394 }, { "epoch": 1.1939542570773578, "grad_norm": 2.6312639713287354, "learning_rate": 5.521200000000001e-06, "loss": 0.027, "step": 22395 }, { "epoch": 1.1940075705070108, "grad_norm": 2.366147518157959, "learning_rate": 5.521000000000001e-06, "loss": 0.0218, "step": 22396 }, { "epoch": 1.1940608839366638, "grad_norm": 1.8317970037460327, "learning_rate": 5.5208000000000006e-06, "loss": 0.0198, "step": 22397 }, { "epoch": 1.1941141973663165, "grad_norm": 1.4673160314559937, "learning_rate": 5.5206e-06, "loss": 0.0176, "step": 22398 }, { "epoch": 1.1941675107959695, "grad_norm": 2.0646212100982666, "learning_rate": 5.5204000000000005e-06, "loss": 0.0547, "step": 22399 }, { "epoch": 1.1942208242256225, "grad_norm": 2.018397808074951, "learning_rate": 5.5202e-06, "loss": 0.0281, "step": 22400 }, { "epoch": 1.1942741376552755, "grad_norm": 1.668957233428955, "learning_rate": 5.5200000000000005e-06, "loss": 0.018, "step": 22401 }, { "epoch": 1.1943274510849282, "grad_norm": 1.2825871706008911, "learning_rate": 5.519800000000001e-06, "loss": 0.0155, "step": 22402 }, { "epoch": 1.1943807645145812, "grad_norm": 1.6885604858398438, "learning_rate": 5.5196000000000005e-06, "loss": 0.0176, "step": 22403 }, { "epoch": 1.1944340779442342, "grad_norm": 2.5386955738067627, "learning_rate": 5.5194e-06, "loss": 0.0296, "step": 22404 }, { "epoch": 1.1944873913738872, "grad_norm": 2.2097325325012207, "learning_rate": 5.5192000000000005e-06, "loss": 0.0114, "step": 22405 }, { "epoch": 1.19454070480354, "grad_norm": 1.0555269718170166, "learning_rate": 5.519e-06, "loss": 0.0098, "step": 22406 }, { "epoch": 1.194594018233193, "grad_norm": 1.7717090845108032, "learning_rate": 5.5188e-06, "loss": 0.0136, "step": 22407 }, { "epoch": 1.1946473316628459, "grad_norm": 3.413923501968384, "learning_rate": 5.518600000000001e-06, "loss": 0.0356, "step": 22408 }, { "epoch": 1.1947006450924988, "grad_norm": 2.100975513458252, "learning_rate": 5.5184e-06, "loss": 0.03, "step": 22409 }, { "epoch": 1.1947539585221518, "grad_norm": 1.0052155256271362, "learning_rate": 5.5182e-06, "loss": 0.0063, "step": 22410 }, { "epoch": 1.1948072719518046, "grad_norm": 1.0053291320800781, "learning_rate": 5.518e-06, "loss": 0.0088, "step": 22411 }, { "epoch": 1.1948605853814576, "grad_norm": 0.5432486534118652, "learning_rate": 5.517800000000001e-06, "loss": 0.0033, "step": 22412 }, { "epoch": 1.1949138988111105, "grad_norm": 0.9783985018730164, "learning_rate": 5.5176e-06, "loss": 0.012, "step": 22413 }, { "epoch": 1.1949672122407635, "grad_norm": 1.5930883884429932, "learning_rate": 5.517400000000001e-06, "loss": 0.0402, "step": 22414 }, { "epoch": 1.1950205256704165, "grad_norm": 0.5767382383346558, "learning_rate": 5.5172e-06, "loss": 0.0076, "step": 22415 }, { "epoch": 1.1950738391000693, "grad_norm": 1.631091833114624, "learning_rate": 5.517e-06, "loss": 0.0178, "step": 22416 }, { "epoch": 1.1951271525297222, "grad_norm": 1.1991318464279175, "learning_rate": 5.5168e-06, "loss": 0.0062, "step": 22417 }, { "epoch": 1.1951804659593752, "grad_norm": 1.066927433013916, "learning_rate": 5.516600000000001e-06, "loss": 0.0077, "step": 22418 }, { "epoch": 1.195233779389028, "grad_norm": 0.7304882407188416, "learning_rate": 5.5164e-06, "loss": 0.0075, "step": 22419 }, { "epoch": 1.195287092818681, "grad_norm": 1.4179067611694336, "learning_rate": 5.516200000000001e-06, "loss": 0.0098, "step": 22420 }, { "epoch": 1.195340406248334, "grad_norm": 1.7051113843917847, "learning_rate": 5.516e-06, "loss": 0.0372, "step": 22421 }, { "epoch": 1.195393719677987, "grad_norm": 1.2773557901382446, "learning_rate": 5.5158e-06, "loss": 0.0092, "step": 22422 }, { "epoch": 1.19544703310764, "grad_norm": 1.4243433475494385, "learning_rate": 5.515600000000001e-06, "loss": 0.0351, "step": 22423 }, { "epoch": 1.1955003465372926, "grad_norm": 2.4759533405303955, "learning_rate": 5.515400000000001e-06, "loss": 0.0297, "step": 22424 }, { "epoch": 1.1955536599669456, "grad_norm": 3.0587663650512695, "learning_rate": 5.5152e-06, "loss": 0.043, "step": 22425 }, { "epoch": 1.1956069733965986, "grad_norm": 1.9452546834945679, "learning_rate": 5.5150000000000006e-06, "loss": 0.0299, "step": 22426 }, { "epoch": 1.1956602868262516, "grad_norm": 1.1995450258255005, "learning_rate": 5.5148e-06, "loss": 0.0189, "step": 22427 }, { "epoch": 1.1957136002559046, "grad_norm": 1.858167290687561, "learning_rate": 5.5146e-06, "loss": 0.0297, "step": 22428 }, { "epoch": 1.1957669136855573, "grad_norm": 1.084672212600708, "learning_rate": 5.514400000000001e-06, "loss": 0.0137, "step": 22429 }, { "epoch": 1.1958202271152103, "grad_norm": 1.8873980045318604, "learning_rate": 5.5142000000000005e-06, "loss": 0.0142, "step": 22430 }, { "epoch": 1.1958735405448633, "grad_norm": 1.4914305210113525, "learning_rate": 5.514e-06, "loss": 0.0289, "step": 22431 }, { "epoch": 1.1959268539745163, "grad_norm": 2.16571307182312, "learning_rate": 5.5138000000000005e-06, "loss": 0.0144, "step": 22432 }, { "epoch": 1.195980167404169, "grad_norm": 1.43674898147583, "learning_rate": 5.5136e-06, "loss": 0.0112, "step": 22433 }, { "epoch": 1.196033480833822, "grad_norm": 1.1101922988891602, "learning_rate": 5.513400000000001e-06, "loss": 0.0111, "step": 22434 }, { "epoch": 1.196086794263475, "grad_norm": 1.5405806303024292, "learning_rate": 5.513200000000001e-06, "loss": 0.0407, "step": 22435 }, { "epoch": 1.196140107693128, "grad_norm": 2.5263397693634033, "learning_rate": 5.513e-06, "loss": 0.0292, "step": 22436 }, { "epoch": 1.1961934211227807, "grad_norm": 4.817091941833496, "learning_rate": 5.5128e-06, "loss": 0.0341, "step": 22437 }, { "epoch": 1.1962467345524337, "grad_norm": 2.5047695636749268, "learning_rate": 5.5126e-06, "loss": 0.0208, "step": 22438 }, { "epoch": 1.1963000479820867, "grad_norm": 3.3909502029418945, "learning_rate": 5.5124e-06, "loss": 0.0543, "step": 22439 }, { "epoch": 1.1963533614117396, "grad_norm": 1.4829046726226807, "learning_rate": 5.512200000000001e-06, "loss": 0.0102, "step": 22440 }, { "epoch": 1.1964066748413926, "grad_norm": 1.3451282978057861, "learning_rate": 5.512000000000001e-06, "loss": 0.023, "step": 22441 }, { "epoch": 1.1964599882710454, "grad_norm": 0.671944797039032, "learning_rate": 5.5118e-06, "loss": 0.0048, "step": 22442 }, { "epoch": 1.1965133017006984, "grad_norm": 0.8789147138595581, "learning_rate": 5.5116e-06, "loss": 0.0135, "step": 22443 }, { "epoch": 1.1965666151303513, "grad_norm": 3.512134075164795, "learning_rate": 5.5114e-06, "loss": 0.0301, "step": 22444 }, { "epoch": 1.1966199285600043, "grad_norm": 1.4957921504974365, "learning_rate": 5.511200000000001e-06, "loss": 0.0105, "step": 22445 }, { "epoch": 1.1966732419896573, "grad_norm": 1.582521677017212, "learning_rate": 5.511000000000001e-06, "loss": 0.0195, "step": 22446 }, { "epoch": 1.19672655541931, "grad_norm": 1.8377177715301514, "learning_rate": 5.510800000000001e-06, "loss": 0.0401, "step": 22447 }, { "epoch": 1.196779868848963, "grad_norm": 1.9267528057098389, "learning_rate": 5.5106e-06, "loss": 0.0274, "step": 22448 }, { "epoch": 1.196833182278616, "grad_norm": 1.920950174331665, "learning_rate": 5.5104e-06, "loss": 0.0144, "step": 22449 }, { "epoch": 1.196886495708269, "grad_norm": 1.735236406326294, "learning_rate": 5.5102e-06, "loss": 0.0154, "step": 22450 }, { "epoch": 1.1969398091379218, "grad_norm": 1.336670160293579, "learning_rate": 5.510000000000001e-06, "loss": 0.0334, "step": 22451 }, { "epoch": 1.1969931225675747, "grad_norm": 0.7764013409614563, "learning_rate": 5.509800000000001e-06, "loss": 0.0066, "step": 22452 }, { "epoch": 1.1970464359972277, "grad_norm": 1.4557292461395264, "learning_rate": 5.509600000000001e-06, "loss": 0.0211, "step": 22453 }, { "epoch": 1.1970997494268807, "grad_norm": 1.9628876447677612, "learning_rate": 5.5094e-06, "loss": 0.0354, "step": 22454 }, { "epoch": 1.1971530628565334, "grad_norm": 1.675045371055603, "learning_rate": 5.5092e-06, "loss": 0.015, "step": 22455 }, { "epoch": 1.1972063762861864, "grad_norm": 1.7343872785568237, "learning_rate": 5.509e-06, "loss": 0.0232, "step": 22456 }, { "epoch": 1.1972596897158394, "grad_norm": 1.7583791017532349, "learning_rate": 5.5088000000000005e-06, "loss": 0.0213, "step": 22457 }, { "epoch": 1.1973130031454924, "grad_norm": 6.050923824310303, "learning_rate": 5.508600000000001e-06, "loss": 0.0326, "step": 22458 }, { "epoch": 1.1973663165751454, "grad_norm": 3.5617220401763916, "learning_rate": 5.5084000000000005e-06, "loss": 0.0533, "step": 22459 }, { "epoch": 1.1974196300047981, "grad_norm": 0.7871683835983276, "learning_rate": 5.5082e-06, "loss": 0.0064, "step": 22460 }, { "epoch": 1.197472943434451, "grad_norm": 1.814293622970581, "learning_rate": 5.508e-06, "loss": 0.0229, "step": 22461 }, { "epoch": 1.197526256864104, "grad_norm": 0.8056583404541016, "learning_rate": 5.507800000000001e-06, "loss": 0.0044, "step": 22462 }, { "epoch": 1.197579570293757, "grad_norm": 0.9928064346313477, "learning_rate": 5.5076000000000004e-06, "loss": 0.0166, "step": 22463 }, { "epoch": 1.19763288372341, "grad_norm": 1.854637622833252, "learning_rate": 5.507400000000001e-06, "loss": 0.0533, "step": 22464 }, { "epoch": 1.1976861971530628, "grad_norm": 1.2838882207870483, "learning_rate": 5.5072e-06, "loss": 0.0195, "step": 22465 }, { "epoch": 1.1977395105827158, "grad_norm": 1.0981711149215698, "learning_rate": 5.507e-06, "loss": 0.0243, "step": 22466 }, { "epoch": 1.1977928240123688, "grad_norm": 1.666806697845459, "learning_rate": 5.5067999999999995e-06, "loss": 0.0139, "step": 22467 }, { "epoch": 1.1978461374420217, "grad_norm": 2.4578213691711426, "learning_rate": 5.506600000000001e-06, "loss": 0.0234, "step": 22468 }, { "epoch": 1.1978994508716745, "grad_norm": 2.0702943801879883, "learning_rate": 5.5064e-06, "loss": 0.0315, "step": 22469 }, { "epoch": 1.1979527643013275, "grad_norm": 1.096953272819519, "learning_rate": 5.506200000000001e-06, "loss": 0.0159, "step": 22470 }, { "epoch": 1.1980060777309804, "grad_norm": 2.1086935997009277, "learning_rate": 5.506e-06, "loss": 0.0469, "step": 22471 }, { "epoch": 1.1980593911606334, "grad_norm": 1.4762372970581055, "learning_rate": 5.5058e-06, "loss": 0.0141, "step": 22472 }, { "epoch": 1.1981127045902862, "grad_norm": 1.4859871864318848, "learning_rate": 5.505600000000001e-06, "loss": 0.0145, "step": 22473 }, { "epoch": 1.1981660180199392, "grad_norm": 1.316530466079712, "learning_rate": 5.505400000000001e-06, "loss": 0.0295, "step": 22474 }, { "epoch": 1.1982193314495921, "grad_norm": 2.641371488571167, "learning_rate": 5.5052e-06, "loss": 0.0225, "step": 22475 }, { "epoch": 1.1982726448792451, "grad_norm": 1.6940960884094238, "learning_rate": 5.505000000000001e-06, "loss": 0.0195, "step": 22476 }, { "epoch": 1.198325958308898, "grad_norm": 1.9056732654571533, "learning_rate": 5.5048e-06, "loss": 0.0188, "step": 22477 }, { "epoch": 1.1983792717385509, "grad_norm": 2.2169790267944336, "learning_rate": 5.5046e-06, "loss": 0.0316, "step": 22478 }, { "epoch": 1.1984325851682038, "grad_norm": 2.728034734725952, "learning_rate": 5.504400000000001e-06, "loss": 0.0372, "step": 22479 }, { "epoch": 1.1984858985978568, "grad_norm": 2.5757031440734863, "learning_rate": 5.504200000000001e-06, "loss": 0.0179, "step": 22480 }, { "epoch": 1.1985392120275098, "grad_norm": 1.0801483392715454, "learning_rate": 5.504e-06, "loss": 0.0104, "step": 22481 }, { "epoch": 1.1985925254571628, "grad_norm": 0.6313643455505371, "learning_rate": 5.5038000000000006e-06, "loss": 0.0054, "step": 22482 }, { "epoch": 1.1986458388868155, "grad_norm": 1.7204577922821045, "learning_rate": 5.5036e-06, "loss": 0.025, "step": 22483 }, { "epoch": 1.1986991523164685, "grad_norm": 2.403775215148926, "learning_rate": 5.503400000000001e-06, "loss": 0.0212, "step": 22484 }, { "epoch": 1.1987524657461215, "grad_norm": 0.9533317685127258, "learning_rate": 5.503200000000001e-06, "loss": 0.0143, "step": 22485 }, { "epoch": 1.1988057791757745, "grad_norm": 1.0173567533493042, "learning_rate": 5.5030000000000005e-06, "loss": 0.0284, "step": 22486 }, { "epoch": 1.1988590926054272, "grad_norm": 1.841642141342163, "learning_rate": 5.5028e-06, "loss": 0.0141, "step": 22487 }, { "epoch": 1.1989124060350802, "grad_norm": 5.159369468688965, "learning_rate": 5.5026000000000005e-06, "loss": 0.0382, "step": 22488 }, { "epoch": 1.1989657194647332, "grad_norm": 1.9764635562896729, "learning_rate": 5.5024e-06, "loss": 0.0213, "step": 22489 }, { "epoch": 1.1990190328943862, "grad_norm": 1.0829963684082031, "learning_rate": 5.502200000000001e-06, "loss": 0.0069, "step": 22490 }, { "epoch": 1.199072346324039, "grad_norm": 1.5136901140213013, "learning_rate": 5.502000000000001e-06, "loss": 0.0359, "step": 22491 }, { "epoch": 1.199125659753692, "grad_norm": 1.7324520349502563, "learning_rate": 5.5018e-06, "loss": 0.019, "step": 22492 }, { "epoch": 1.1991789731833449, "grad_norm": 1.2265545129776, "learning_rate": 5.5016e-06, "loss": 0.0139, "step": 22493 }, { "epoch": 1.1992322866129979, "grad_norm": 0.9244004487991333, "learning_rate": 5.5014e-06, "loss": 0.0135, "step": 22494 }, { "epoch": 1.1992856000426508, "grad_norm": 1.5631729364395142, "learning_rate": 5.501200000000001e-06, "loss": 0.0378, "step": 22495 }, { "epoch": 1.1993389134723036, "grad_norm": 0.8551788926124573, "learning_rate": 5.501000000000001e-06, "loss": 0.0057, "step": 22496 }, { "epoch": 1.1993922269019566, "grad_norm": 0.9417417645454407, "learning_rate": 5.500800000000001e-06, "loss": 0.0068, "step": 22497 }, { "epoch": 1.1994455403316095, "grad_norm": 2.3951783180236816, "learning_rate": 5.5006e-06, "loss": 0.0452, "step": 22498 }, { "epoch": 1.1994988537612625, "grad_norm": 0.5509827733039856, "learning_rate": 5.5004e-06, "loss": 0.0054, "step": 22499 }, { "epoch": 1.1995521671909155, "grad_norm": 0.7094652652740479, "learning_rate": 5.5002e-06, "loss": 0.0056, "step": 22500 }, { "epoch": 1.1996054806205683, "grad_norm": 1.750252604484558, "learning_rate": 5.500000000000001e-06, "loss": 0.0493, "step": 22501 }, { "epoch": 1.1996587940502212, "grad_norm": 2.2069616317749023, "learning_rate": 5.499800000000001e-06, "loss": 0.0276, "step": 22502 }, { "epoch": 1.1997121074798742, "grad_norm": 2.274441719055176, "learning_rate": 5.499600000000001e-06, "loss": 0.0393, "step": 22503 }, { "epoch": 1.1997654209095272, "grad_norm": 2.2920360565185547, "learning_rate": 5.4994e-06, "loss": 0.0282, "step": 22504 }, { "epoch": 1.19981873433918, "grad_norm": 2.5574188232421875, "learning_rate": 5.4992e-06, "loss": 0.0439, "step": 22505 }, { "epoch": 1.199872047768833, "grad_norm": 3.470578908920288, "learning_rate": 5.499000000000001e-06, "loss": 0.039, "step": 22506 }, { "epoch": 1.199925361198486, "grad_norm": 0.8181002736091614, "learning_rate": 5.498800000000001e-06, "loss": 0.0083, "step": 22507 }, { "epoch": 1.199978674628139, "grad_norm": 1.1567397117614746, "learning_rate": 5.498600000000001e-06, "loss": 0.0096, "step": 22508 }, { "epoch": 1.2000319880577917, "grad_norm": 0.7352294921875, "learning_rate": 5.498400000000001e-06, "loss": 0.0077, "step": 22509 }, { "epoch": 1.2000853014874446, "grad_norm": 4.086512088775635, "learning_rate": 5.4982e-06, "loss": 0.0234, "step": 22510 }, { "epoch": 1.2001386149170976, "grad_norm": 1.7735395431518555, "learning_rate": 5.498e-06, "loss": 0.0297, "step": 22511 }, { "epoch": 1.2001919283467506, "grad_norm": 1.7733299732208252, "learning_rate": 5.497800000000001e-06, "loss": 0.0104, "step": 22512 }, { "epoch": 1.2002452417764036, "grad_norm": 1.39858078956604, "learning_rate": 5.4976000000000005e-06, "loss": 0.0412, "step": 22513 }, { "epoch": 1.2002985552060563, "grad_norm": 1.0189235210418701, "learning_rate": 5.497400000000001e-06, "loss": 0.0242, "step": 22514 }, { "epoch": 1.2003518686357093, "grad_norm": 0.9877577424049377, "learning_rate": 5.4972000000000005e-06, "loss": 0.0094, "step": 22515 }, { "epoch": 1.2004051820653623, "grad_norm": 0.8097975850105286, "learning_rate": 5.497e-06, "loss": 0.01, "step": 22516 }, { "epoch": 1.2004584954950153, "grad_norm": 0.4923185706138611, "learning_rate": 5.4968e-06, "loss": 0.0055, "step": 22517 }, { "epoch": 1.2005118089246682, "grad_norm": 1.0114389657974243, "learning_rate": 5.496600000000001e-06, "loss": 0.008, "step": 22518 }, { "epoch": 1.200565122354321, "grad_norm": 1.3193881511688232, "learning_rate": 5.4964000000000004e-06, "loss": 0.0244, "step": 22519 }, { "epoch": 1.200618435783974, "grad_norm": 1.66972017288208, "learning_rate": 5.496200000000001e-06, "loss": 0.0447, "step": 22520 }, { "epoch": 1.200671749213627, "grad_norm": 3.6695265769958496, "learning_rate": 5.496e-06, "loss": 0.014, "step": 22521 }, { "epoch": 1.20072506264328, "grad_norm": 1.8961610794067383, "learning_rate": 5.4958e-06, "loss": 0.022, "step": 22522 }, { "epoch": 1.2007783760729327, "grad_norm": 1.765526294708252, "learning_rate": 5.495600000000001e-06, "loss": 0.0383, "step": 22523 }, { "epoch": 1.2008316895025857, "grad_norm": 0.9902379512786865, "learning_rate": 5.495400000000001e-06, "loss": 0.0183, "step": 22524 }, { "epoch": 1.2008850029322387, "grad_norm": 2.117846727371216, "learning_rate": 5.4952e-06, "loss": 0.0403, "step": 22525 }, { "epoch": 1.2009383163618916, "grad_norm": 1.5451397895812988, "learning_rate": 5.495000000000001e-06, "loss": 0.0116, "step": 22526 }, { "epoch": 1.2009916297915444, "grad_norm": 2.131718158721924, "learning_rate": 5.4948e-06, "loss": 0.0214, "step": 22527 }, { "epoch": 1.2010449432211974, "grad_norm": 1.6670520305633545, "learning_rate": 5.4946e-06, "loss": 0.0253, "step": 22528 }, { "epoch": 1.2010982566508503, "grad_norm": 2.2367398738861084, "learning_rate": 5.494400000000001e-06, "loss": 0.0756, "step": 22529 }, { "epoch": 1.2011515700805033, "grad_norm": 1.6244841814041138, "learning_rate": 5.494200000000001e-06, "loss": 0.0305, "step": 22530 }, { "epoch": 1.2012048835101563, "grad_norm": 1.2766886949539185, "learning_rate": 5.494e-06, "loss": 0.0357, "step": 22531 }, { "epoch": 1.201258196939809, "grad_norm": 0.5625396966934204, "learning_rate": 5.493800000000001e-06, "loss": 0.007, "step": 22532 }, { "epoch": 1.201311510369462, "grad_norm": 2.066270589828491, "learning_rate": 5.4936e-06, "loss": 0.0206, "step": 22533 }, { "epoch": 1.201364823799115, "grad_norm": 1.488389492034912, "learning_rate": 5.493400000000001e-06, "loss": 0.0147, "step": 22534 }, { "epoch": 1.201418137228768, "grad_norm": 1.7076988220214844, "learning_rate": 5.493200000000001e-06, "loss": 0.0129, "step": 22535 }, { "epoch": 1.201471450658421, "grad_norm": 2.5773797035217285, "learning_rate": 5.493000000000001e-06, "loss": 0.0329, "step": 22536 }, { "epoch": 1.2015247640880737, "grad_norm": 1.5041821002960205, "learning_rate": 5.4928e-06, "loss": 0.0501, "step": 22537 }, { "epoch": 1.2015780775177267, "grad_norm": 4.019604206085205, "learning_rate": 5.4926000000000006e-06, "loss": 0.0378, "step": 22538 }, { "epoch": 1.2016313909473797, "grad_norm": 1.4503352642059326, "learning_rate": 5.4924e-06, "loss": 0.009, "step": 22539 }, { "epoch": 1.2016847043770325, "grad_norm": 1.3025379180908203, "learning_rate": 5.4922000000000005e-06, "loss": 0.0246, "step": 22540 }, { "epoch": 1.2017380178066854, "grad_norm": 1.5437098741531372, "learning_rate": 5.492000000000001e-06, "loss": 0.0134, "step": 22541 }, { "epoch": 1.2017913312363384, "grad_norm": 1.3868380784988403, "learning_rate": 5.4918000000000005e-06, "loss": 0.0105, "step": 22542 }, { "epoch": 1.2018446446659914, "grad_norm": 2.8550467491149902, "learning_rate": 5.4916e-06, "loss": 0.0321, "step": 22543 }, { "epoch": 1.2018979580956444, "grad_norm": 2.702272415161133, "learning_rate": 5.4914000000000005e-06, "loss": 0.041, "step": 22544 }, { "epoch": 1.2019512715252971, "grad_norm": 1.5804134607315063, "learning_rate": 5.491200000000001e-06, "loss": 0.0109, "step": 22545 }, { "epoch": 1.20200458495495, "grad_norm": 1.672764778137207, "learning_rate": 5.4910000000000005e-06, "loss": 0.0503, "step": 22546 }, { "epoch": 1.202057898384603, "grad_norm": 0.8754321932792664, "learning_rate": 5.490800000000001e-06, "loss": 0.0322, "step": 22547 }, { "epoch": 1.202111211814256, "grad_norm": 1.6410894393920898, "learning_rate": 5.4906e-06, "loss": 0.0374, "step": 22548 }, { "epoch": 1.202164525243909, "grad_norm": 1.6535227298736572, "learning_rate": 5.4904e-06, "loss": 0.0216, "step": 22549 }, { "epoch": 1.2022178386735618, "grad_norm": 1.0803122520446777, "learning_rate": 5.4902e-06, "loss": 0.0086, "step": 22550 }, { "epoch": 1.2022711521032148, "grad_norm": 2.0789244174957275, "learning_rate": 5.490000000000001e-06, "loss": 0.0327, "step": 22551 }, { "epoch": 1.2023244655328678, "grad_norm": 1.662900686264038, "learning_rate": 5.4898e-06, "loss": 0.0258, "step": 22552 }, { "epoch": 1.2023777789625207, "grad_norm": 2.157335042953491, "learning_rate": 5.489600000000001e-06, "loss": 0.0193, "step": 22553 }, { "epoch": 1.2024310923921735, "grad_norm": 1.596734881401062, "learning_rate": 5.4894e-06, "loss": 0.0215, "step": 22554 }, { "epoch": 1.2024844058218265, "grad_norm": 1.3609881401062012, "learning_rate": 5.4892e-06, "loss": 0.0083, "step": 22555 }, { "epoch": 1.2025377192514795, "grad_norm": 0.6108040809631348, "learning_rate": 5.489000000000001e-06, "loss": 0.007, "step": 22556 }, { "epoch": 1.2025910326811324, "grad_norm": 1.8728477954864502, "learning_rate": 5.488800000000001e-06, "loss": 0.022, "step": 22557 }, { "epoch": 1.2026443461107852, "grad_norm": 1.6020756959915161, "learning_rate": 5.4886e-06, "loss": 0.022, "step": 22558 }, { "epoch": 1.2026976595404382, "grad_norm": 2.7548232078552246, "learning_rate": 5.488400000000001e-06, "loss": 0.0774, "step": 22559 }, { "epoch": 1.2027509729700911, "grad_norm": 2.7458434104919434, "learning_rate": 5.4882e-06, "loss": 0.0252, "step": 22560 }, { "epoch": 1.2028042863997441, "grad_norm": 3.653211832046509, "learning_rate": 5.488e-06, "loss": 0.0237, "step": 22561 }, { "epoch": 1.202857599829397, "grad_norm": 1.7421772480010986, "learning_rate": 5.487800000000001e-06, "loss": 0.0439, "step": 22562 }, { "epoch": 1.2029109132590499, "grad_norm": 1.656583547592163, "learning_rate": 5.487600000000001e-06, "loss": 0.024, "step": 22563 }, { "epoch": 1.2029642266887028, "grad_norm": 2.8854408264160156, "learning_rate": 5.4874e-06, "loss": 0.0462, "step": 22564 }, { "epoch": 1.2030175401183558, "grad_norm": 1.9484680891036987, "learning_rate": 5.487200000000001e-06, "loss": 0.0162, "step": 22565 }, { "epoch": 1.2030708535480088, "grad_norm": 1.486438512802124, "learning_rate": 5.487e-06, "loss": 0.0116, "step": 22566 }, { "epoch": 1.2031241669776618, "grad_norm": 2.263819694519043, "learning_rate": 5.4868e-06, "loss": 0.042, "step": 22567 }, { "epoch": 1.2031774804073145, "grad_norm": 1.186996340751648, "learning_rate": 5.486600000000001e-06, "loss": 0.012, "step": 22568 }, { "epoch": 1.2032307938369675, "grad_norm": 2.547811269760132, "learning_rate": 5.4864000000000005e-06, "loss": 0.0311, "step": 22569 }, { "epoch": 1.2032841072666205, "grad_norm": 1.0612176656723022, "learning_rate": 5.4862e-06, "loss": 0.0176, "step": 22570 }, { "epoch": 1.2033374206962735, "grad_norm": 1.6666380167007446, "learning_rate": 5.4860000000000005e-06, "loss": 0.0271, "step": 22571 }, { "epoch": 1.2033907341259262, "grad_norm": 1.7163875102996826, "learning_rate": 5.4858e-06, "loss": 0.0193, "step": 22572 }, { "epoch": 1.2034440475555792, "grad_norm": 2.9843733310699463, "learning_rate": 5.485600000000001e-06, "loss": 0.0347, "step": 22573 }, { "epoch": 1.2034973609852322, "grad_norm": 1.4868760108947754, "learning_rate": 5.485400000000001e-06, "loss": 0.0107, "step": 22574 }, { "epoch": 1.2035506744148852, "grad_norm": 1.1820842027664185, "learning_rate": 5.4852000000000004e-06, "loss": 0.0112, "step": 22575 }, { "epoch": 1.203603987844538, "grad_norm": 1.4302386045455933, "learning_rate": 5.485e-06, "loss": 0.0293, "step": 22576 }, { "epoch": 1.203657301274191, "grad_norm": 0.8797045350074768, "learning_rate": 5.4848e-06, "loss": 0.0146, "step": 22577 }, { "epoch": 1.2037106147038439, "grad_norm": 2.7121574878692627, "learning_rate": 5.4846e-06, "loss": 0.0644, "step": 22578 }, { "epoch": 1.2037639281334969, "grad_norm": 0.8916963934898376, "learning_rate": 5.484400000000001e-06, "loss": 0.0118, "step": 22579 }, { "epoch": 1.2038172415631498, "grad_norm": 2.619790554046631, "learning_rate": 5.484200000000001e-06, "loss": 0.0104, "step": 22580 }, { "epoch": 1.2038705549928026, "grad_norm": 3.9631235599517822, "learning_rate": 5.484e-06, "loss": 0.0397, "step": 22581 }, { "epoch": 1.2039238684224556, "grad_norm": 2.147381544113159, "learning_rate": 5.4838e-06, "loss": 0.0251, "step": 22582 }, { "epoch": 1.2039771818521086, "grad_norm": 2.9302995204925537, "learning_rate": 5.4836e-06, "loss": 0.0163, "step": 22583 }, { "epoch": 1.2040304952817615, "grad_norm": 1.429568886756897, "learning_rate": 5.483400000000001e-06, "loss": 0.013, "step": 22584 }, { "epoch": 1.2040838087114145, "grad_norm": 3.528163433074951, "learning_rate": 5.483200000000001e-06, "loss": 0.0195, "step": 22585 }, { "epoch": 1.2041371221410673, "grad_norm": 1.502905011177063, "learning_rate": 5.483000000000001e-06, "loss": 0.0104, "step": 22586 }, { "epoch": 1.2041904355707203, "grad_norm": 2.751986026763916, "learning_rate": 5.4828e-06, "loss": 0.0289, "step": 22587 }, { "epoch": 1.2042437490003732, "grad_norm": 0.7454050183296204, "learning_rate": 5.4826e-06, "loss": 0.0046, "step": 22588 }, { "epoch": 1.2042970624300262, "grad_norm": 1.3962371349334717, "learning_rate": 5.4824e-06, "loss": 0.0253, "step": 22589 }, { "epoch": 1.204350375859679, "grad_norm": 2.8134329319000244, "learning_rate": 5.482200000000001e-06, "loss": 0.0414, "step": 22590 }, { "epoch": 1.204403689289332, "grad_norm": 1.148081660270691, "learning_rate": 5.482000000000001e-06, "loss": 0.0087, "step": 22591 }, { "epoch": 1.204457002718985, "grad_norm": 1.5843225717544556, "learning_rate": 5.481800000000001e-06, "loss": 0.0244, "step": 22592 }, { "epoch": 1.204510316148638, "grad_norm": 1.459051489830017, "learning_rate": 5.4816e-06, "loss": 0.0311, "step": 22593 }, { "epoch": 1.2045636295782907, "grad_norm": 3.192072629928589, "learning_rate": 5.4814e-06, "loss": 0.0547, "step": 22594 }, { "epoch": 1.2046169430079436, "grad_norm": 2.268289089202881, "learning_rate": 5.481200000000001e-06, "loss": 0.0506, "step": 22595 }, { "epoch": 1.2046702564375966, "grad_norm": 0.7482608556747437, "learning_rate": 5.4810000000000005e-06, "loss": 0.0063, "step": 22596 }, { "epoch": 1.2047235698672496, "grad_norm": 1.6535152196884155, "learning_rate": 5.480800000000001e-06, "loss": 0.0375, "step": 22597 }, { "epoch": 1.2047768832969026, "grad_norm": 2.849900722503662, "learning_rate": 5.4806000000000005e-06, "loss": 0.0225, "step": 22598 }, { "epoch": 1.2048301967265553, "grad_norm": 2.633704662322998, "learning_rate": 5.4804e-06, "loss": 0.0309, "step": 22599 }, { "epoch": 1.2048835101562083, "grad_norm": 4.0471954345703125, "learning_rate": 5.4802e-06, "loss": 0.0182, "step": 22600 }, { "epoch": 1.2049368235858613, "grad_norm": 2.3200302124023438, "learning_rate": 5.480000000000001e-06, "loss": 0.0399, "step": 22601 }, { "epoch": 1.2049901370155143, "grad_norm": 1.9872015714645386, "learning_rate": 5.4798000000000005e-06, "loss": 0.0279, "step": 22602 }, { "epoch": 1.2050434504451673, "grad_norm": 1.6225392818450928, "learning_rate": 5.479600000000001e-06, "loss": 0.0319, "step": 22603 }, { "epoch": 1.20509676387482, "grad_norm": 1.600595235824585, "learning_rate": 5.4794000000000004e-06, "loss": 0.0126, "step": 22604 }, { "epoch": 1.205150077304473, "grad_norm": 1.5175118446350098, "learning_rate": 5.4792e-06, "loss": 0.0128, "step": 22605 }, { "epoch": 1.205203390734126, "grad_norm": 2.8154091835021973, "learning_rate": 5.479000000000001e-06, "loss": 0.0313, "step": 22606 }, { "epoch": 1.205256704163779, "grad_norm": 2.7145609855651855, "learning_rate": 5.478800000000001e-06, "loss": 0.0313, "step": 22607 }, { "epoch": 1.2053100175934317, "grad_norm": 1.5977777242660522, "learning_rate": 5.4786e-06, "loss": 0.0278, "step": 22608 }, { "epoch": 1.2053633310230847, "grad_norm": 1.0779249668121338, "learning_rate": 5.478400000000001e-06, "loss": 0.0108, "step": 22609 }, { "epoch": 1.2054166444527377, "grad_norm": 1.49788498878479, "learning_rate": 5.4782e-06, "loss": 0.0298, "step": 22610 }, { "epoch": 1.2054699578823906, "grad_norm": 3.564958333969116, "learning_rate": 5.478e-06, "loss": 0.0252, "step": 22611 }, { "epoch": 1.2055232713120434, "grad_norm": 1.9348150491714478, "learning_rate": 5.477800000000001e-06, "loss": 0.0316, "step": 22612 }, { "epoch": 1.2055765847416964, "grad_norm": 1.8924903869628906, "learning_rate": 5.477600000000001e-06, "loss": 0.0095, "step": 22613 }, { "epoch": 1.2056298981713494, "grad_norm": 1.5641076564788818, "learning_rate": 5.4774e-06, "loss": 0.0274, "step": 22614 }, { "epoch": 1.2056832116010023, "grad_norm": 1.2802035808563232, "learning_rate": 5.477200000000001e-06, "loss": 0.0198, "step": 22615 }, { "epoch": 1.2057365250306553, "grad_norm": 1.220227599143982, "learning_rate": 5.477e-06, "loss": 0.0123, "step": 22616 }, { "epoch": 1.205789838460308, "grad_norm": 0.961613118648529, "learning_rate": 5.4768e-06, "loss": 0.0171, "step": 22617 }, { "epoch": 1.205843151889961, "grad_norm": 0.7727793455123901, "learning_rate": 5.476600000000001e-06, "loss": 0.0149, "step": 22618 }, { "epoch": 1.205896465319614, "grad_norm": 0.4839918315410614, "learning_rate": 5.476400000000001e-06, "loss": 0.0065, "step": 22619 }, { "epoch": 1.205949778749267, "grad_norm": 1.0554161071777344, "learning_rate": 5.4762e-06, "loss": 0.0184, "step": 22620 }, { "epoch": 1.20600309217892, "grad_norm": 2.2415711879730225, "learning_rate": 5.476000000000001e-06, "loss": 0.0559, "step": 22621 }, { "epoch": 1.2060564056085727, "grad_norm": 1.694159746170044, "learning_rate": 5.4758e-06, "loss": 0.0346, "step": 22622 }, { "epoch": 1.2061097190382257, "grad_norm": 1.412794589996338, "learning_rate": 5.4756000000000006e-06, "loss": 0.0305, "step": 22623 }, { "epoch": 1.2061630324678787, "grad_norm": 1.9709806442260742, "learning_rate": 5.475400000000001e-06, "loss": 0.0282, "step": 22624 }, { "epoch": 1.2062163458975317, "grad_norm": 0.8352911472320557, "learning_rate": 5.4752000000000005e-06, "loss": 0.0218, "step": 22625 }, { "epoch": 1.2062696593271844, "grad_norm": 2.022224187850952, "learning_rate": 5.475e-06, "loss": 0.0251, "step": 22626 }, { "epoch": 1.2063229727568374, "grad_norm": 1.1363494396209717, "learning_rate": 5.4748000000000005e-06, "loss": 0.0219, "step": 22627 }, { "epoch": 1.2063762861864904, "grad_norm": 0.8755514025688171, "learning_rate": 5.4746e-06, "loss": 0.0116, "step": 22628 }, { "epoch": 1.2064295996161434, "grad_norm": 0.8418327569961548, "learning_rate": 5.4744000000000005e-06, "loss": 0.0134, "step": 22629 }, { "epoch": 1.2064829130457961, "grad_norm": 1.7194145917892456, "learning_rate": 5.474200000000001e-06, "loss": 0.0426, "step": 22630 }, { "epoch": 1.2065362264754491, "grad_norm": 1.2320282459259033, "learning_rate": 5.4740000000000004e-06, "loss": 0.0213, "step": 22631 }, { "epoch": 1.206589539905102, "grad_norm": 0.7809626460075378, "learning_rate": 5.4738e-06, "loss": 0.0077, "step": 22632 }, { "epoch": 1.206642853334755, "grad_norm": 1.0689001083374023, "learning_rate": 5.4736e-06, "loss": 0.0131, "step": 22633 }, { "epoch": 1.206696166764408, "grad_norm": 3.043581247329712, "learning_rate": 5.473400000000001e-06, "loss": 0.0298, "step": 22634 }, { "epoch": 1.2067494801940608, "grad_norm": 1.87321138381958, "learning_rate": 5.4732e-06, "loss": 0.0271, "step": 22635 }, { "epoch": 1.2068027936237138, "grad_norm": 2.7889597415924072, "learning_rate": 5.473000000000001e-06, "loss": 0.0301, "step": 22636 }, { "epoch": 1.2068561070533668, "grad_norm": 2.5626909732818604, "learning_rate": 5.4728e-06, "loss": 0.0525, "step": 22637 }, { "epoch": 1.2069094204830197, "grad_norm": 2.0212881565093994, "learning_rate": 5.4726e-06, "loss": 0.0273, "step": 22638 }, { "epoch": 1.2069627339126727, "grad_norm": 2.4712977409362793, "learning_rate": 5.4724e-06, "loss": 0.0329, "step": 22639 }, { "epoch": 1.2070160473423255, "grad_norm": 2.3550071716308594, "learning_rate": 5.472200000000001e-06, "loss": 0.0655, "step": 22640 }, { "epoch": 1.2070693607719785, "grad_norm": 1.8308686017990112, "learning_rate": 5.472e-06, "loss": 0.0235, "step": 22641 }, { "epoch": 1.2071226742016314, "grad_norm": 0.9597731232643127, "learning_rate": 5.471800000000001e-06, "loss": 0.0112, "step": 22642 }, { "epoch": 1.2071759876312842, "grad_norm": 1.7844091653823853, "learning_rate": 5.4716e-06, "loss": 0.0096, "step": 22643 }, { "epoch": 1.2072293010609372, "grad_norm": 1.5717724561691284, "learning_rate": 5.4714e-06, "loss": 0.0156, "step": 22644 }, { "epoch": 1.2072826144905902, "grad_norm": 0.9578275084495544, "learning_rate": 5.471200000000001e-06, "loss": 0.0079, "step": 22645 }, { "epoch": 1.2073359279202431, "grad_norm": 3.0783627033233643, "learning_rate": 5.471000000000001e-06, "loss": 0.0405, "step": 22646 }, { "epoch": 1.2073892413498961, "grad_norm": 2.44115948677063, "learning_rate": 5.4708e-06, "loss": 0.0602, "step": 22647 }, { "epoch": 1.2074425547795489, "grad_norm": 2.9880435466766357, "learning_rate": 5.470600000000001e-06, "loss": 0.036, "step": 22648 }, { "epoch": 1.2074958682092019, "grad_norm": 1.252991795539856, "learning_rate": 5.4704e-06, "loss": 0.018, "step": 22649 }, { "epoch": 1.2075491816388548, "grad_norm": 1.9147157669067383, "learning_rate": 5.4702e-06, "loss": 0.0361, "step": 22650 }, { "epoch": 1.2076024950685078, "grad_norm": 1.034029245376587, "learning_rate": 5.470000000000001e-06, "loss": 0.0085, "step": 22651 }, { "epoch": 1.2076558084981608, "grad_norm": 2.285068988800049, "learning_rate": 5.4698000000000006e-06, "loss": 0.0401, "step": 22652 }, { "epoch": 1.2077091219278135, "grad_norm": 0.498091459274292, "learning_rate": 5.4696e-06, "loss": 0.0037, "step": 22653 }, { "epoch": 1.2077624353574665, "grad_norm": 2.5364906787872314, "learning_rate": 5.4694000000000005e-06, "loss": 0.0399, "step": 22654 }, { "epoch": 1.2078157487871195, "grad_norm": 0.7899717092514038, "learning_rate": 5.4692e-06, "loss": 0.0132, "step": 22655 }, { "epoch": 1.2078690622167725, "grad_norm": 1.092677354812622, "learning_rate": 5.469000000000001e-06, "loss": 0.0284, "step": 22656 }, { "epoch": 1.2079223756464252, "grad_norm": 2.1965017318725586, "learning_rate": 5.468800000000001e-06, "loss": 0.0426, "step": 22657 }, { "epoch": 1.2079756890760782, "grad_norm": 1.5669981241226196, "learning_rate": 5.4686000000000005e-06, "loss": 0.0358, "step": 22658 }, { "epoch": 1.2080290025057312, "grad_norm": 1.9876830577850342, "learning_rate": 5.4684e-06, "loss": 0.0397, "step": 22659 }, { "epoch": 1.2080823159353842, "grad_norm": 1.0246647596359253, "learning_rate": 5.4682000000000004e-06, "loss": 0.0136, "step": 22660 }, { "epoch": 1.208135629365037, "grad_norm": 2.3556761741638184, "learning_rate": 5.468e-06, "loss": 0.0435, "step": 22661 }, { "epoch": 1.20818894279469, "grad_norm": 0.9789211750030518, "learning_rate": 5.467800000000001e-06, "loss": 0.0176, "step": 22662 }, { "epoch": 1.208242256224343, "grad_norm": 2.1864876747131348, "learning_rate": 5.467600000000001e-06, "loss": 0.0277, "step": 22663 }, { "epoch": 1.2082955696539959, "grad_norm": 1.9703810214996338, "learning_rate": 5.4674e-06, "loss": 0.0394, "step": 22664 }, { "epoch": 1.2083488830836489, "grad_norm": 5.903076171875, "learning_rate": 5.4672e-06, "loss": 0.0338, "step": 22665 }, { "epoch": 1.2084021965133016, "grad_norm": 2.4467451572418213, "learning_rate": 5.467e-06, "loss": 0.0595, "step": 22666 }, { "epoch": 1.2084555099429546, "grad_norm": 1.551499605178833, "learning_rate": 5.466800000000001e-06, "loss": 0.0475, "step": 22667 }, { "epoch": 1.2085088233726076, "grad_norm": 0.892873227596283, "learning_rate": 5.466600000000001e-06, "loss": 0.0252, "step": 22668 }, { "epoch": 1.2085621368022605, "grad_norm": 2.1542861461639404, "learning_rate": 5.466400000000001e-06, "loss": 0.0343, "step": 22669 }, { "epoch": 1.2086154502319135, "grad_norm": 1.118640422821045, "learning_rate": 5.4662e-06, "loss": 0.022, "step": 22670 }, { "epoch": 1.2086687636615663, "grad_norm": 1.846751093864441, "learning_rate": 5.466e-06, "loss": 0.0472, "step": 22671 }, { "epoch": 1.2087220770912193, "grad_norm": 1.5207648277282715, "learning_rate": 5.4658e-06, "loss": 0.0324, "step": 22672 }, { "epoch": 1.2087753905208722, "grad_norm": 2.1725080013275146, "learning_rate": 5.465600000000001e-06, "loss": 0.0778, "step": 22673 }, { "epoch": 1.2088287039505252, "grad_norm": 1.3119477033615112, "learning_rate": 5.465400000000001e-06, "loss": 0.0116, "step": 22674 }, { "epoch": 1.208882017380178, "grad_norm": 1.006459355354309, "learning_rate": 5.465200000000001e-06, "loss": 0.0166, "step": 22675 }, { "epoch": 1.208935330809831, "grad_norm": 0.6200597286224365, "learning_rate": 5.465e-06, "loss": 0.0086, "step": 22676 }, { "epoch": 1.208988644239484, "grad_norm": 2.604973077774048, "learning_rate": 5.4648e-06, "loss": 0.0676, "step": 22677 }, { "epoch": 1.209041957669137, "grad_norm": 1.1267421245574951, "learning_rate": 5.4646e-06, "loss": 0.0103, "step": 22678 }, { "epoch": 1.2090952710987897, "grad_norm": 1.3867088556289673, "learning_rate": 5.4644000000000006e-06, "loss": 0.0313, "step": 22679 }, { "epoch": 1.2091485845284426, "grad_norm": 0.9016983509063721, "learning_rate": 5.464200000000001e-06, "loss": 0.0176, "step": 22680 }, { "epoch": 1.2092018979580956, "grad_norm": 1.5910176038742065, "learning_rate": 5.4640000000000005e-06, "loss": 0.0404, "step": 22681 }, { "epoch": 1.2092552113877486, "grad_norm": 1.824375033378601, "learning_rate": 5.4638e-06, "loss": 0.0166, "step": 22682 }, { "epoch": 1.2093085248174016, "grad_norm": 1.0852553844451904, "learning_rate": 5.4636e-06, "loss": 0.0152, "step": 22683 }, { "epoch": 1.2093618382470543, "grad_norm": 1.6895270347595215, "learning_rate": 5.463400000000001e-06, "loss": 0.0245, "step": 22684 }, { "epoch": 1.2094151516767073, "grad_norm": 1.3541079759597778, "learning_rate": 5.4632000000000005e-06, "loss": 0.0157, "step": 22685 }, { "epoch": 1.2094684651063603, "grad_norm": 1.9434095621109009, "learning_rate": 5.463000000000001e-06, "loss": 0.0377, "step": 22686 }, { "epoch": 1.2095217785360133, "grad_norm": 1.3341532945632935, "learning_rate": 5.4628000000000005e-06, "loss": 0.0338, "step": 22687 }, { "epoch": 1.2095750919656663, "grad_norm": 2.7625114917755127, "learning_rate": 5.4626e-06, "loss": 0.0509, "step": 22688 }, { "epoch": 1.209628405395319, "grad_norm": 2.377305269241333, "learning_rate": 5.4623999999999996e-06, "loss": 0.0432, "step": 22689 }, { "epoch": 1.209681718824972, "grad_norm": 2.2092032432556152, "learning_rate": 5.462200000000001e-06, "loss": 0.0173, "step": 22690 }, { "epoch": 1.209735032254625, "grad_norm": 1.6745446920394897, "learning_rate": 5.462e-06, "loss": 0.0157, "step": 22691 }, { "epoch": 1.209788345684278, "grad_norm": 1.7980257272720337, "learning_rate": 5.461800000000001e-06, "loss": 0.0368, "step": 22692 }, { "epoch": 1.2098416591139307, "grad_norm": 1.915537714958191, "learning_rate": 5.4616e-06, "loss": 0.0218, "step": 22693 }, { "epoch": 1.2098949725435837, "grad_norm": 1.248895287513733, "learning_rate": 5.4614e-06, "loss": 0.014, "step": 22694 }, { "epoch": 1.2099482859732367, "grad_norm": 0.8352068066596985, "learning_rate": 5.461200000000001e-06, "loss": 0.0106, "step": 22695 }, { "epoch": 1.2100015994028896, "grad_norm": 2.1646156311035156, "learning_rate": 5.461000000000001e-06, "loss": 0.0157, "step": 22696 }, { "epoch": 1.2100549128325424, "grad_norm": 2.742891311645508, "learning_rate": 5.4608e-06, "loss": 0.0242, "step": 22697 }, { "epoch": 1.2101082262621954, "grad_norm": 0.9273741245269775, "learning_rate": 5.460600000000001e-06, "loss": 0.0242, "step": 22698 }, { "epoch": 1.2101615396918484, "grad_norm": 1.7132319211959839, "learning_rate": 5.4604e-06, "loss": 0.0152, "step": 22699 }, { "epoch": 1.2102148531215013, "grad_norm": 1.199940800666809, "learning_rate": 5.4602e-06, "loss": 0.0098, "step": 22700 }, { "epoch": 1.2102681665511543, "grad_norm": 1.044192910194397, "learning_rate": 5.460000000000001e-06, "loss": 0.0373, "step": 22701 }, { "epoch": 1.210321479980807, "grad_norm": 0.25277361273765564, "learning_rate": 5.459800000000001e-06, "loss": 0.0031, "step": 22702 }, { "epoch": 1.21037479341046, "grad_norm": 1.6873713731765747, "learning_rate": 5.4596e-06, "loss": 0.0508, "step": 22703 }, { "epoch": 1.210428106840113, "grad_norm": 1.9408345222473145, "learning_rate": 5.459400000000001e-06, "loss": 0.0275, "step": 22704 }, { "epoch": 1.210481420269766, "grad_norm": 2.306896448135376, "learning_rate": 5.4592e-06, "loss": 0.0457, "step": 22705 }, { "epoch": 1.210534733699419, "grad_norm": 3.6971771717071533, "learning_rate": 5.459000000000001e-06, "loss": 0.0869, "step": 22706 }, { "epoch": 1.2105880471290718, "grad_norm": 0.7459957003593445, "learning_rate": 5.458800000000001e-06, "loss": 0.0178, "step": 22707 }, { "epoch": 1.2106413605587247, "grad_norm": 1.7726993560791016, "learning_rate": 5.4586000000000006e-06, "loss": 0.0136, "step": 22708 }, { "epoch": 1.2106946739883777, "grad_norm": 2.035991668701172, "learning_rate": 5.4584e-06, "loss": 0.0637, "step": 22709 }, { "epoch": 1.2107479874180307, "grad_norm": 1.2323007583618164, "learning_rate": 5.4582000000000005e-06, "loss": 0.0286, "step": 22710 }, { "epoch": 1.2108013008476834, "grad_norm": 1.5377569198608398, "learning_rate": 5.458e-06, "loss": 0.0323, "step": 22711 }, { "epoch": 1.2108546142773364, "grad_norm": 1.3913583755493164, "learning_rate": 5.4578000000000005e-06, "loss": 0.0324, "step": 22712 }, { "epoch": 1.2109079277069894, "grad_norm": 1.8957890272140503, "learning_rate": 5.457600000000001e-06, "loss": 0.0222, "step": 22713 }, { "epoch": 1.2109612411366424, "grad_norm": 1.4482154846191406, "learning_rate": 5.4574000000000005e-06, "loss": 0.022, "step": 22714 }, { "epoch": 1.2110145545662951, "grad_norm": 2.795208692550659, "learning_rate": 5.4572e-06, "loss": 0.0342, "step": 22715 }, { "epoch": 1.2110678679959481, "grad_norm": 2.2667133808135986, "learning_rate": 5.4570000000000004e-06, "loss": 0.0233, "step": 22716 }, { "epoch": 1.211121181425601, "grad_norm": 4.805214881896973, "learning_rate": 5.456800000000001e-06, "loss": 0.041, "step": 22717 }, { "epoch": 1.211174494855254, "grad_norm": 1.7597262859344482, "learning_rate": 5.4566e-06, "loss": 0.0211, "step": 22718 }, { "epoch": 1.211227808284907, "grad_norm": 1.8900753259658813, "learning_rate": 5.456400000000001e-06, "loss": 0.0319, "step": 22719 }, { "epoch": 1.2112811217145598, "grad_norm": 2.369314193725586, "learning_rate": 5.4562e-06, "loss": 0.0326, "step": 22720 }, { "epoch": 1.2113344351442128, "grad_norm": 1.6159648895263672, "learning_rate": 5.456e-06, "loss": 0.0216, "step": 22721 }, { "epoch": 1.2113877485738658, "grad_norm": 1.0767569541931152, "learning_rate": 5.4558e-06, "loss": 0.0314, "step": 22722 }, { "epoch": 1.2114410620035188, "grad_norm": 1.1914986371994019, "learning_rate": 5.455600000000001e-06, "loss": 0.0263, "step": 22723 }, { "epoch": 1.2114943754331717, "grad_norm": 0.6715402603149414, "learning_rate": 5.4554e-06, "loss": 0.0073, "step": 22724 }, { "epoch": 1.2115476888628245, "grad_norm": 0.807448148727417, "learning_rate": 5.455200000000001e-06, "loss": 0.0035, "step": 22725 }, { "epoch": 1.2116010022924775, "grad_norm": 1.8317714929580688, "learning_rate": 5.455e-06, "loss": 0.0274, "step": 22726 }, { "epoch": 1.2116543157221304, "grad_norm": 1.5500798225402832, "learning_rate": 5.4548e-06, "loss": 0.0211, "step": 22727 }, { "epoch": 1.2117076291517834, "grad_norm": 1.9180389642715454, "learning_rate": 5.4546e-06, "loss": 0.0337, "step": 22728 }, { "epoch": 1.2117609425814362, "grad_norm": 0.963966965675354, "learning_rate": 5.454400000000001e-06, "loss": 0.0412, "step": 22729 }, { "epoch": 1.2118142560110892, "grad_norm": 1.1895935535430908, "learning_rate": 5.4542e-06, "loss": 0.0226, "step": 22730 }, { "epoch": 1.2118675694407421, "grad_norm": 1.6462469100952148, "learning_rate": 5.454000000000001e-06, "loss": 0.0307, "step": 22731 }, { "epoch": 1.2119208828703951, "grad_norm": 2.0053040981292725, "learning_rate": 5.4538e-06, "loss": 0.0449, "step": 22732 }, { "epoch": 1.2119741963000479, "grad_norm": 0.5437227487564087, "learning_rate": 5.4536e-06, "loss": 0.0061, "step": 22733 }, { "epoch": 1.2120275097297009, "grad_norm": 1.866753101348877, "learning_rate": 5.453400000000001e-06, "loss": 0.0251, "step": 22734 }, { "epoch": 1.2120808231593538, "grad_norm": 1.710630178451538, "learning_rate": 5.4532000000000006e-06, "loss": 0.0176, "step": 22735 }, { "epoch": 1.2121341365890068, "grad_norm": 1.3325060606002808, "learning_rate": 5.453e-06, "loss": 0.0197, "step": 22736 }, { "epoch": 1.2121874500186598, "grad_norm": 0.9117926955223083, "learning_rate": 5.4528000000000005e-06, "loss": 0.0156, "step": 22737 }, { "epoch": 1.2122407634483126, "grad_norm": 1.5289109945297241, "learning_rate": 5.4526e-06, "loss": 0.014, "step": 22738 }, { "epoch": 1.2122940768779655, "grad_norm": 2.9830737113952637, "learning_rate": 5.4524e-06, "loss": 0.0459, "step": 22739 }, { "epoch": 1.2123473903076185, "grad_norm": 1.5904638767242432, "learning_rate": 5.452200000000001e-06, "loss": 0.0257, "step": 22740 }, { "epoch": 1.2124007037372715, "grad_norm": 1.6039079427719116, "learning_rate": 5.4520000000000005e-06, "loss": 0.013, "step": 22741 }, { "epoch": 1.2124540171669245, "grad_norm": 4.4885406494140625, "learning_rate": 5.4518e-06, "loss": 0.0255, "step": 22742 }, { "epoch": 1.2125073305965772, "grad_norm": 1.819010853767395, "learning_rate": 5.4516000000000005e-06, "loss": 0.0329, "step": 22743 }, { "epoch": 1.2125606440262302, "grad_norm": 5.237369060516357, "learning_rate": 5.4514e-06, "loss": 0.0478, "step": 22744 }, { "epoch": 1.2126139574558832, "grad_norm": 0.7502538561820984, "learning_rate": 5.451200000000001e-06, "loss": 0.0077, "step": 22745 }, { "epoch": 1.212667270885536, "grad_norm": 1.14519464969635, "learning_rate": 5.451000000000001e-06, "loss": 0.0168, "step": 22746 }, { "epoch": 1.212720584315189, "grad_norm": 1.6710251569747925, "learning_rate": 5.4508e-06, "loss": 0.0265, "step": 22747 }, { "epoch": 1.212773897744842, "grad_norm": 2.1387016773223877, "learning_rate": 5.4506e-06, "loss": 0.0277, "step": 22748 }, { "epoch": 1.2128272111744949, "grad_norm": 1.7941300868988037, "learning_rate": 5.4504e-06, "loss": 0.0138, "step": 22749 }, { "epoch": 1.2128805246041479, "grad_norm": 1.8546875715255737, "learning_rate": 5.4502e-06, "loss": 0.0444, "step": 22750 }, { "epoch": 1.2129338380338006, "grad_norm": 1.1202269792556763, "learning_rate": 5.450000000000001e-06, "loss": 0.0328, "step": 22751 }, { "epoch": 1.2129871514634536, "grad_norm": 1.874537467956543, "learning_rate": 5.449800000000001e-06, "loss": 0.0302, "step": 22752 }, { "epoch": 1.2130404648931066, "grad_norm": 0.36143946647644043, "learning_rate": 5.4496e-06, "loss": 0.0032, "step": 22753 }, { "epoch": 1.2130937783227596, "grad_norm": 3.76568603515625, "learning_rate": 5.4494e-06, "loss": 0.0889, "step": 22754 }, { "epoch": 1.2131470917524125, "grad_norm": 2.14031720161438, "learning_rate": 5.4492e-06, "loss": 0.0384, "step": 22755 }, { "epoch": 1.2132004051820653, "grad_norm": 1.532519817352295, "learning_rate": 5.449000000000001e-06, "loss": 0.0262, "step": 22756 }, { "epoch": 1.2132537186117183, "grad_norm": 3.1878490447998047, "learning_rate": 5.448800000000001e-06, "loss": 0.0489, "step": 22757 }, { "epoch": 1.2133070320413712, "grad_norm": 4.019355297088623, "learning_rate": 5.448600000000001e-06, "loss": 0.0213, "step": 22758 }, { "epoch": 1.2133603454710242, "grad_norm": 0.478584349155426, "learning_rate": 5.4484e-06, "loss": 0.0048, "step": 22759 }, { "epoch": 1.213413658900677, "grad_norm": 2.217695951461792, "learning_rate": 5.4482e-06, "loss": 0.0566, "step": 22760 }, { "epoch": 1.21346697233033, "grad_norm": 1.0147154331207275, "learning_rate": 5.448e-06, "loss": 0.0129, "step": 22761 }, { "epoch": 1.213520285759983, "grad_norm": 3.176649808883667, "learning_rate": 5.447800000000001e-06, "loss": 0.0374, "step": 22762 }, { "epoch": 1.213573599189636, "grad_norm": 1.6862785816192627, "learning_rate": 5.447600000000001e-06, "loss": 0.0477, "step": 22763 }, { "epoch": 1.2136269126192887, "grad_norm": 1.5757228136062622, "learning_rate": 5.4474000000000006e-06, "loss": 0.0144, "step": 22764 }, { "epoch": 1.2136802260489417, "grad_norm": 2.3306946754455566, "learning_rate": 5.4472e-06, "loss": 0.0438, "step": 22765 }, { "epoch": 1.2137335394785946, "grad_norm": 3.4564807415008545, "learning_rate": 5.447e-06, "loss": 0.0578, "step": 22766 }, { "epoch": 1.2137868529082476, "grad_norm": 1.2775325775146484, "learning_rate": 5.446800000000001e-06, "loss": 0.0144, "step": 22767 }, { "epoch": 1.2138401663379006, "grad_norm": 1.299065351486206, "learning_rate": 5.4466000000000005e-06, "loss": 0.0189, "step": 22768 }, { "epoch": 1.2138934797675534, "grad_norm": 1.4542547464370728, "learning_rate": 5.446400000000001e-06, "loss": 0.0146, "step": 22769 }, { "epoch": 1.2139467931972063, "grad_norm": 1.5854971408843994, "learning_rate": 5.4462000000000005e-06, "loss": 0.0303, "step": 22770 }, { "epoch": 1.2140001066268593, "grad_norm": 1.2517417669296265, "learning_rate": 5.446e-06, "loss": 0.0223, "step": 22771 }, { "epoch": 1.2140534200565123, "grad_norm": 1.1550028324127197, "learning_rate": 5.4458e-06, "loss": 0.0206, "step": 22772 }, { "epoch": 1.2141067334861653, "grad_norm": 2.8810338973999023, "learning_rate": 5.445600000000001e-06, "loss": 0.0133, "step": 22773 }, { "epoch": 1.214160046915818, "grad_norm": 1.594967246055603, "learning_rate": 5.4454e-06, "loss": 0.0181, "step": 22774 }, { "epoch": 1.214213360345471, "grad_norm": 1.1354845762252808, "learning_rate": 5.445200000000001e-06, "loss": 0.0091, "step": 22775 }, { "epoch": 1.214266673775124, "grad_norm": 0.4648788273334503, "learning_rate": 5.445e-06, "loss": 0.0062, "step": 22776 }, { "epoch": 1.214319987204777, "grad_norm": 1.1474007368087769, "learning_rate": 5.4448e-06, "loss": 0.0169, "step": 22777 }, { "epoch": 1.2143733006344297, "grad_norm": 2.959071159362793, "learning_rate": 5.444600000000001e-06, "loss": 0.0191, "step": 22778 }, { "epoch": 1.2144266140640827, "grad_norm": 0.80522221326828, "learning_rate": 5.444400000000001e-06, "loss": 0.0065, "step": 22779 }, { "epoch": 1.2144799274937357, "grad_norm": 3.1026418209075928, "learning_rate": 5.4442e-06, "loss": 0.068, "step": 22780 }, { "epoch": 1.2145332409233887, "grad_norm": 1.3582440614700317, "learning_rate": 5.444000000000001e-06, "loss": 0.0352, "step": 22781 }, { "epoch": 1.2145865543530414, "grad_norm": 2.7695658206939697, "learning_rate": 5.4438e-06, "loss": 0.0595, "step": 22782 }, { "epoch": 1.2146398677826944, "grad_norm": 1.732518196105957, "learning_rate": 5.4436e-06, "loss": 0.0169, "step": 22783 }, { "epoch": 1.2146931812123474, "grad_norm": 1.9107881784439087, "learning_rate": 5.443400000000001e-06, "loss": 0.0463, "step": 22784 }, { "epoch": 1.2147464946420004, "grad_norm": 2.0773985385894775, "learning_rate": 5.443200000000001e-06, "loss": 0.0562, "step": 22785 }, { "epoch": 1.2147998080716533, "grad_norm": 0.8857173323631287, "learning_rate": 5.443e-06, "loss": 0.0191, "step": 22786 }, { "epoch": 1.214853121501306, "grad_norm": 1.0977935791015625, "learning_rate": 5.442800000000001e-06, "loss": 0.0112, "step": 22787 }, { "epoch": 1.214906434930959, "grad_norm": 1.156049132347107, "learning_rate": 5.4426e-06, "loss": 0.011, "step": 22788 }, { "epoch": 1.214959748360612, "grad_norm": 1.7024117708206177, "learning_rate": 5.4424e-06, "loss": 0.0225, "step": 22789 }, { "epoch": 1.215013061790265, "grad_norm": 2.5443601608276367, "learning_rate": 5.442200000000001e-06, "loss": 0.0286, "step": 22790 }, { "epoch": 1.215066375219918, "grad_norm": 1.6011046171188354, "learning_rate": 5.442000000000001e-06, "loss": 0.0354, "step": 22791 }, { "epoch": 1.2151196886495708, "grad_norm": 1.7582041025161743, "learning_rate": 5.4418e-06, "loss": 0.0207, "step": 22792 }, { "epoch": 1.2151730020792237, "grad_norm": 2.1128716468811035, "learning_rate": 5.4416000000000006e-06, "loss": 0.0178, "step": 22793 }, { "epoch": 1.2152263155088767, "grad_norm": 1.7547330856323242, "learning_rate": 5.4414e-06, "loss": 0.0377, "step": 22794 }, { "epoch": 1.2152796289385297, "grad_norm": 3.226809501647949, "learning_rate": 5.4412000000000005e-06, "loss": 0.022, "step": 22795 }, { "epoch": 1.2153329423681825, "grad_norm": 2.0631067752838135, "learning_rate": 5.441000000000001e-06, "loss": 0.0147, "step": 22796 }, { "epoch": 1.2153862557978354, "grad_norm": 1.678086280822754, "learning_rate": 5.4408000000000005e-06, "loss": 0.0216, "step": 22797 }, { "epoch": 1.2154395692274884, "grad_norm": 2.421391010284424, "learning_rate": 5.4406e-06, "loss": 0.0332, "step": 22798 }, { "epoch": 1.2154928826571414, "grad_norm": 1.6374824047088623, "learning_rate": 5.4404000000000005e-06, "loss": 0.0373, "step": 22799 }, { "epoch": 1.2155461960867942, "grad_norm": 0.4900127947330475, "learning_rate": 5.4402e-06, "loss": 0.0052, "step": 22800 }, { "epoch": 1.2155995095164471, "grad_norm": 2.607085704803467, "learning_rate": 5.4400000000000004e-06, "loss": 0.0226, "step": 22801 }, { "epoch": 1.2156528229461, "grad_norm": 0.2650337517261505, "learning_rate": 5.439800000000001e-06, "loss": 0.0036, "step": 22802 }, { "epoch": 1.215706136375753, "grad_norm": 3.048665761947632, "learning_rate": 5.4396e-06, "loss": 0.0729, "step": 22803 }, { "epoch": 1.215759449805406, "grad_norm": 2.927760362625122, "learning_rate": 5.4394e-06, "loss": 0.0247, "step": 22804 }, { "epoch": 1.2158127632350588, "grad_norm": 2.1653316020965576, "learning_rate": 5.4392e-06, "loss": 0.0386, "step": 22805 }, { "epoch": 1.2158660766647118, "grad_norm": 1.542604923248291, "learning_rate": 5.439000000000001e-06, "loss": 0.0115, "step": 22806 }, { "epoch": 1.2159193900943648, "grad_norm": 1.7213668823242188, "learning_rate": 5.4388e-06, "loss": 0.0351, "step": 22807 }, { "epoch": 1.2159727035240178, "grad_norm": 1.0429048538208008, "learning_rate": 5.438600000000001e-06, "loss": 0.0323, "step": 22808 }, { "epoch": 1.2160260169536707, "grad_norm": 2.696712017059326, "learning_rate": 5.4384e-06, "loss": 0.0235, "step": 22809 }, { "epoch": 1.2160793303833235, "grad_norm": 0.9638636112213135, "learning_rate": 5.4382e-06, "loss": 0.0072, "step": 22810 }, { "epoch": 1.2161326438129765, "grad_norm": 1.1342300176620483, "learning_rate": 5.438e-06, "loss": 0.0125, "step": 22811 }, { "epoch": 1.2161859572426295, "grad_norm": 1.3794450759887695, "learning_rate": 5.437800000000001e-06, "loss": 0.0155, "step": 22812 }, { "epoch": 1.2162392706722824, "grad_norm": 3.1322970390319824, "learning_rate": 5.4376e-06, "loss": 0.0432, "step": 22813 }, { "epoch": 1.2162925841019352, "grad_norm": 1.5408669710159302, "learning_rate": 5.437400000000001e-06, "loss": 0.0438, "step": 22814 }, { "epoch": 1.2163458975315882, "grad_norm": 0.9586470127105713, "learning_rate": 5.4372e-06, "loss": 0.0077, "step": 22815 }, { "epoch": 1.2163992109612412, "grad_norm": 2.7672858238220215, "learning_rate": 5.437e-06, "loss": 0.0202, "step": 22816 }, { "epoch": 1.2164525243908941, "grad_norm": 1.6712110042572021, "learning_rate": 5.436800000000001e-06, "loss": 0.0179, "step": 22817 }, { "epoch": 1.2165058378205469, "grad_norm": 2.0499746799468994, "learning_rate": 5.436600000000001e-06, "loss": 0.0201, "step": 22818 }, { "epoch": 1.2165591512501999, "grad_norm": 4.247371673583984, "learning_rate": 5.4364e-06, "loss": 0.0335, "step": 22819 }, { "epoch": 1.2166124646798528, "grad_norm": 1.689160943031311, "learning_rate": 5.4362000000000006e-06, "loss": 0.0146, "step": 22820 }, { "epoch": 1.2166657781095058, "grad_norm": 1.6565968990325928, "learning_rate": 5.436e-06, "loss": 0.0113, "step": 22821 }, { "epoch": 1.2167190915391588, "grad_norm": 1.9417920112609863, "learning_rate": 5.4358e-06, "loss": 0.0387, "step": 22822 }, { "epoch": 1.2167724049688116, "grad_norm": 1.622451663017273, "learning_rate": 5.435600000000001e-06, "loss": 0.0138, "step": 22823 }, { "epoch": 1.2168257183984645, "grad_norm": 2.2914657592773438, "learning_rate": 5.4354000000000005e-06, "loss": 0.052, "step": 22824 }, { "epoch": 1.2168790318281175, "grad_norm": 2.2762796878814697, "learning_rate": 5.4352e-06, "loss": 0.0156, "step": 22825 }, { "epoch": 1.2169323452577705, "grad_norm": 1.3000928163528442, "learning_rate": 5.4350000000000005e-06, "loss": 0.0106, "step": 22826 }, { "epoch": 1.2169856586874235, "grad_norm": 1.0503780841827393, "learning_rate": 5.4348e-06, "loss": 0.0105, "step": 22827 }, { "epoch": 1.2170389721170762, "grad_norm": 1.46296226978302, "learning_rate": 5.434600000000001e-06, "loss": 0.0248, "step": 22828 }, { "epoch": 1.2170922855467292, "grad_norm": 2.8015494346618652, "learning_rate": 5.434400000000001e-06, "loss": 0.0336, "step": 22829 }, { "epoch": 1.2171455989763822, "grad_norm": 1.596100926399231, "learning_rate": 5.4342e-06, "loss": 0.0235, "step": 22830 }, { "epoch": 1.2171989124060352, "grad_norm": 0.8811159133911133, "learning_rate": 5.434e-06, "loss": 0.0142, "step": 22831 }, { "epoch": 1.217252225835688, "grad_norm": 1.7142951488494873, "learning_rate": 5.4338e-06, "loss": 0.0249, "step": 22832 }, { "epoch": 1.217305539265341, "grad_norm": 1.6064938306808472, "learning_rate": 5.4336e-06, "loss": 0.0159, "step": 22833 }, { "epoch": 1.2173588526949939, "grad_norm": 0.905714213848114, "learning_rate": 5.433400000000001e-06, "loss": 0.0181, "step": 22834 }, { "epoch": 1.2174121661246469, "grad_norm": 4.4942240715026855, "learning_rate": 5.433200000000001e-06, "loss": 0.0438, "step": 22835 }, { "epoch": 1.2174654795542996, "grad_norm": 1.096531867980957, "learning_rate": 5.433e-06, "loss": 0.0169, "step": 22836 }, { "epoch": 1.2175187929839526, "grad_norm": 1.2685823440551758, "learning_rate": 5.4328e-06, "loss": 0.0169, "step": 22837 }, { "epoch": 1.2175721064136056, "grad_norm": 1.8967359066009521, "learning_rate": 5.4326e-06, "loss": 0.0312, "step": 22838 }, { "epoch": 1.2176254198432586, "grad_norm": 1.52768874168396, "learning_rate": 5.4324e-06, "loss": 0.0087, "step": 22839 }, { "epoch": 1.2176787332729115, "grad_norm": 3.1537744998931885, "learning_rate": 5.432200000000001e-06, "loss": 0.0319, "step": 22840 }, { "epoch": 1.2177320467025643, "grad_norm": 1.6973729133605957, "learning_rate": 5.432000000000001e-06, "loss": 0.0485, "step": 22841 }, { "epoch": 1.2177853601322173, "grad_norm": 2.1492958068847656, "learning_rate": 5.4318e-06, "loss": 0.0336, "step": 22842 }, { "epoch": 1.2178386735618703, "grad_norm": 1.5224250555038452, "learning_rate": 5.4316e-06, "loss": 0.0562, "step": 22843 }, { "epoch": 1.2178919869915232, "grad_norm": 1.171338677406311, "learning_rate": 5.4314e-06, "loss": 0.024, "step": 22844 }, { "epoch": 1.2179453004211762, "grad_norm": 0.8708022236824036, "learning_rate": 5.431200000000001e-06, "loss": 0.0087, "step": 22845 }, { "epoch": 1.217998613850829, "grad_norm": 4.434363842010498, "learning_rate": 5.431000000000001e-06, "loss": 0.0303, "step": 22846 }, { "epoch": 1.218051927280482, "grad_norm": 1.5736711025238037, "learning_rate": 5.430800000000001e-06, "loss": 0.0411, "step": 22847 }, { "epoch": 1.218105240710135, "grad_norm": 0.8403488993644714, "learning_rate": 5.4306e-06, "loss": 0.0136, "step": 22848 }, { "epoch": 1.218158554139788, "grad_norm": 1.9918866157531738, "learning_rate": 5.4304e-06, "loss": 0.045, "step": 22849 }, { "epoch": 1.2182118675694407, "grad_norm": 1.078476905822754, "learning_rate": 5.4302e-06, "loss": 0.0148, "step": 22850 }, { "epoch": 1.2182651809990936, "grad_norm": 2.244720697402954, "learning_rate": 5.4300000000000005e-06, "loss": 0.0543, "step": 22851 }, { "epoch": 1.2183184944287466, "grad_norm": 1.234538197517395, "learning_rate": 5.429800000000001e-06, "loss": 0.0189, "step": 22852 }, { "epoch": 1.2183718078583996, "grad_norm": 2.3673059940338135, "learning_rate": 5.4296000000000005e-06, "loss": 0.0239, "step": 22853 }, { "epoch": 1.2184251212880524, "grad_norm": 1.079515814781189, "learning_rate": 5.4294e-06, "loss": 0.0052, "step": 22854 }, { "epoch": 1.2184784347177053, "grad_norm": 1.503360629081726, "learning_rate": 5.4292e-06, "loss": 0.0187, "step": 22855 }, { "epoch": 1.2185317481473583, "grad_norm": 0.9826406240463257, "learning_rate": 5.429000000000001e-06, "loss": 0.0115, "step": 22856 }, { "epoch": 1.2185850615770113, "grad_norm": 1.59535551071167, "learning_rate": 5.4288000000000004e-06, "loss": 0.011, "step": 22857 }, { "epoch": 1.2186383750066643, "grad_norm": 2.757183313369751, "learning_rate": 5.428600000000001e-06, "loss": 0.0397, "step": 22858 }, { "epoch": 1.218691688436317, "grad_norm": 1.2105772495269775, "learning_rate": 5.4284e-06, "loss": 0.0222, "step": 22859 }, { "epoch": 1.21874500186597, "grad_norm": 1.626358151435852, "learning_rate": 5.4282e-06, "loss": 0.0335, "step": 22860 }, { "epoch": 1.218798315295623, "grad_norm": 1.8961931467056274, "learning_rate": 5.4279999999999995e-06, "loss": 0.015, "step": 22861 }, { "epoch": 1.218851628725276, "grad_norm": 3.436279058456421, "learning_rate": 5.427800000000001e-06, "loss": 0.0331, "step": 22862 }, { "epoch": 1.218904942154929, "grad_norm": 1.8217909336090088, "learning_rate": 5.4276e-06, "loss": 0.0173, "step": 22863 }, { "epoch": 1.2189582555845817, "grad_norm": 2.6447222232818604, "learning_rate": 5.427400000000001e-06, "loss": 0.0224, "step": 22864 }, { "epoch": 1.2190115690142347, "grad_norm": 1.2067229747772217, "learning_rate": 5.4272e-06, "loss": 0.0181, "step": 22865 }, { "epoch": 1.2190648824438877, "grad_norm": 1.2371220588684082, "learning_rate": 5.427e-06, "loss": 0.0133, "step": 22866 }, { "epoch": 1.2191181958735404, "grad_norm": 2.2075531482696533, "learning_rate": 5.426800000000001e-06, "loss": 0.0257, "step": 22867 }, { "epoch": 1.2191715093031934, "grad_norm": 0.7323015332221985, "learning_rate": 5.426600000000001e-06, "loss": 0.0122, "step": 22868 }, { "epoch": 1.2192248227328464, "grad_norm": 0.7467960119247437, "learning_rate": 5.4264e-06, "loss": 0.006, "step": 22869 }, { "epoch": 1.2192781361624994, "grad_norm": 4.207606792449951, "learning_rate": 5.426200000000001e-06, "loss": 0.0424, "step": 22870 }, { "epoch": 1.2193314495921523, "grad_norm": 1.7543538808822632, "learning_rate": 5.426e-06, "loss": 0.0201, "step": 22871 }, { "epoch": 1.219384763021805, "grad_norm": 1.460073471069336, "learning_rate": 5.4258e-06, "loss": 0.0231, "step": 22872 }, { "epoch": 1.219438076451458, "grad_norm": 0.8256635069847107, "learning_rate": 5.425600000000001e-06, "loss": 0.0163, "step": 22873 }, { "epoch": 1.219491389881111, "grad_norm": 2.398118257522583, "learning_rate": 5.425400000000001e-06, "loss": 0.0215, "step": 22874 }, { "epoch": 1.219544703310764, "grad_norm": 2.5247912406921387, "learning_rate": 5.4252e-06, "loss": 0.0266, "step": 22875 }, { "epoch": 1.219598016740417, "grad_norm": 1.5217978954315186, "learning_rate": 5.4250000000000006e-06, "loss": 0.0238, "step": 22876 }, { "epoch": 1.2196513301700698, "grad_norm": 1.2331827878952026, "learning_rate": 5.4248e-06, "loss": 0.0259, "step": 22877 }, { "epoch": 1.2197046435997227, "grad_norm": 1.2780711650848389, "learning_rate": 5.424600000000001e-06, "loss": 0.0135, "step": 22878 }, { "epoch": 1.2197579570293757, "grad_norm": 1.483932375907898, "learning_rate": 5.424400000000001e-06, "loss": 0.0406, "step": 22879 }, { "epoch": 1.2198112704590287, "grad_norm": 1.7555925846099854, "learning_rate": 5.4242000000000005e-06, "loss": 0.0285, "step": 22880 }, { "epoch": 1.2198645838886815, "grad_norm": 2.823723316192627, "learning_rate": 5.424e-06, "loss": 0.029, "step": 22881 }, { "epoch": 1.2199178973183344, "grad_norm": 2.8754260540008545, "learning_rate": 5.4238000000000005e-06, "loss": 0.0296, "step": 22882 }, { "epoch": 1.2199712107479874, "grad_norm": 2.118321180343628, "learning_rate": 5.4236e-06, "loss": 0.0292, "step": 22883 }, { "epoch": 1.2200245241776404, "grad_norm": 2.1407511234283447, "learning_rate": 5.423400000000001e-06, "loss": 0.0135, "step": 22884 }, { "epoch": 1.2200778376072932, "grad_norm": 3.493807792663574, "learning_rate": 5.423200000000001e-06, "loss": 0.0581, "step": 22885 }, { "epoch": 1.2201311510369461, "grad_norm": 2.9364545345306396, "learning_rate": 5.4230000000000004e-06, "loss": 0.0337, "step": 22886 }, { "epoch": 1.2201844644665991, "grad_norm": 2.289761781692505, "learning_rate": 5.4228e-06, "loss": 0.0153, "step": 22887 }, { "epoch": 1.220237777896252, "grad_norm": 1.8695240020751953, "learning_rate": 5.4226e-06, "loss": 0.05, "step": 22888 }, { "epoch": 1.220291091325905, "grad_norm": 0.36391279101371765, "learning_rate": 5.422400000000001e-06, "loss": 0.0033, "step": 22889 }, { "epoch": 1.2203444047555578, "grad_norm": 2.477536201477051, "learning_rate": 5.422200000000001e-06, "loss": 0.0282, "step": 22890 }, { "epoch": 1.2203977181852108, "grad_norm": 1.729387879371643, "learning_rate": 5.422000000000001e-06, "loss": 0.0363, "step": 22891 }, { "epoch": 1.2204510316148638, "grad_norm": 2.566222667694092, "learning_rate": 5.4218e-06, "loss": 0.0303, "step": 22892 }, { "epoch": 1.2205043450445168, "grad_norm": 0.9883713722229004, "learning_rate": 5.4216e-06, "loss": 0.0122, "step": 22893 }, { "epoch": 1.2205576584741697, "grad_norm": 2.0906331539154053, "learning_rate": 5.4214e-06, "loss": 0.0463, "step": 22894 }, { "epoch": 1.2206109719038225, "grad_norm": 0.24514718353748322, "learning_rate": 5.421200000000001e-06, "loss": 0.0034, "step": 22895 }, { "epoch": 1.2206642853334755, "grad_norm": 1.405017375946045, "learning_rate": 5.421000000000001e-06, "loss": 0.023, "step": 22896 }, { "epoch": 1.2207175987631285, "grad_norm": 1.162445068359375, "learning_rate": 5.420800000000001e-06, "loss": 0.01, "step": 22897 }, { "epoch": 1.2207709121927814, "grad_norm": 2.0007026195526123, "learning_rate": 5.4206e-06, "loss": 0.025, "step": 22898 }, { "epoch": 1.2208242256224342, "grad_norm": 1.6927372217178345, "learning_rate": 5.4204e-06, "loss": 0.0224, "step": 22899 }, { "epoch": 1.2208775390520872, "grad_norm": 2.8751626014709473, "learning_rate": 5.4202e-06, "loss": 0.0427, "step": 22900 }, { "epoch": 1.2209308524817402, "grad_norm": 1.0255087614059448, "learning_rate": 5.420000000000001e-06, "loss": 0.0185, "step": 22901 }, { "epoch": 1.2209841659113931, "grad_norm": 1.3933360576629639, "learning_rate": 5.419800000000001e-06, "loss": 0.0188, "step": 22902 }, { "epoch": 1.221037479341046, "grad_norm": 2.728538990020752, "learning_rate": 5.419600000000001e-06, "loss": 0.0196, "step": 22903 }, { "epoch": 1.2210907927706989, "grad_norm": 1.3702541589736938, "learning_rate": 5.4194e-06, "loss": 0.0201, "step": 22904 }, { "epoch": 1.2211441062003519, "grad_norm": 1.2509264945983887, "learning_rate": 5.4192e-06, "loss": 0.0108, "step": 22905 }, { "epoch": 1.2211974196300048, "grad_norm": 2.1925625801086426, "learning_rate": 5.419000000000001e-06, "loss": 0.0404, "step": 22906 }, { "epoch": 1.2212507330596578, "grad_norm": 1.455553650856018, "learning_rate": 5.4188000000000005e-06, "loss": 0.0213, "step": 22907 }, { "epoch": 1.2213040464893106, "grad_norm": 1.9027901887893677, "learning_rate": 5.418600000000001e-06, "loss": 0.0368, "step": 22908 }, { "epoch": 1.2213573599189635, "grad_norm": 1.6993186473846436, "learning_rate": 5.4184000000000005e-06, "loss": 0.0281, "step": 22909 }, { "epoch": 1.2214106733486165, "grad_norm": 1.0258128643035889, "learning_rate": 5.4182e-06, "loss": 0.0117, "step": 22910 }, { "epoch": 1.2214639867782695, "grad_norm": 1.5222551822662354, "learning_rate": 5.418e-06, "loss": 0.0233, "step": 22911 }, { "epoch": 1.2215173002079225, "grad_norm": 1.9266287088394165, "learning_rate": 5.417800000000001e-06, "loss": 0.0169, "step": 22912 }, { "epoch": 1.2215706136375752, "grad_norm": 1.347033143043518, "learning_rate": 5.4176000000000004e-06, "loss": 0.0337, "step": 22913 }, { "epoch": 1.2216239270672282, "grad_norm": 1.0832003355026245, "learning_rate": 5.417400000000001e-06, "loss": 0.0137, "step": 22914 }, { "epoch": 1.2216772404968812, "grad_norm": 0.7468058466911316, "learning_rate": 5.4172e-06, "loss": 0.0072, "step": 22915 }, { "epoch": 1.2217305539265342, "grad_norm": 2.159135103225708, "learning_rate": 5.417e-06, "loss": 0.0214, "step": 22916 }, { "epoch": 1.221783867356187, "grad_norm": 1.7193783521652222, "learning_rate": 5.416800000000001e-06, "loss": 0.0175, "step": 22917 }, { "epoch": 1.22183718078584, "grad_norm": 1.9731314182281494, "learning_rate": 5.416600000000001e-06, "loss": 0.0353, "step": 22918 }, { "epoch": 1.221890494215493, "grad_norm": 1.9264179468154907, "learning_rate": 5.4164e-06, "loss": 0.0266, "step": 22919 }, { "epoch": 1.2219438076451459, "grad_norm": 1.2609820365905762, "learning_rate": 5.416200000000001e-06, "loss": 0.0101, "step": 22920 }, { "epoch": 1.2219971210747986, "grad_norm": 1.465503454208374, "learning_rate": 5.416e-06, "loss": 0.0112, "step": 22921 }, { "epoch": 1.2220504345044516, "grad_norm": 1.8214895725250244, "learning_rate": 5.4158e-06, "loss": 0.0162, "step": 22922 }, { "epoch": 1.2221037479341046, "grad_norm": 3.7290115356445312, "learning_rate": 5.415600000000001e-06, "loss": 0.0551, "step": 22923 }, { "epoch": 1.2221570613637576, "grad_norm": 1.6397571563720703, "learning_rate": 5.415400000000001e-06, "loss": 0.0364, "step": 22924 }, { "epoch": 1.2222103747934105, "grad_norm": 0.9531623721122742, "learning_rate": 5.4152e-06, "loss": 0.0186, "step": 22925 }, { "epoch": 1.2222636882230633, "grad_norm": 0.6103434562683105, "learning_rate": 5.415000000000001e-06, "loss": 0.0051, "step": 22926 }, { "epoch": 1.2223170016527163, "grad_norm": 1.9562302827835083, "learning_rate": 5.4148e-06, "loss": 0.015, "step": 22927 }, { "epoch": 1.2223703150823693, "grad_norm": 1.456624984741211, "learning_rate": 5.414600000000001e-06, "loss": 0.0112, "step": 22928 }, { "epoch": 1.2224236285120222, "grad_norm": 1.738729476928711, "learning_rate": 5.414400000000001e-06, "loss": 0.0315, "step": 22929 }, { "epoch": 1.2224769419416752, "grad_norm": 1.7577595710754395, "learning_rate": 5.414200000000001e-06, "loss": 0.0265, "step": 22930 }, { "epoch": 1.222530255371328, "grad_norm": 1.9330896139144897, "learning_rate": 5.414e-06, "loss": 0.0186, "step": 22931 }, { "epoch": 1.222583568800981, "grad_norm": 0.6622089743614197, "learning_rate": 5.413800000000001e-06, "loss": 0.0239, "step": 22932 }, { "epoch": 1.222636882230634, "grad_norm": 1.7261497974395752, "learning_rate": 5.4136e-06, "loss": 0.008, "step": 22933 }, { "epoch": 1.222690195660287, "grad_norm": 3.379059076309204, "learning_rate": 5.4134000000000006e-06, "loss": 0.019, "step": 22934 }, { "epoch": 1.2227435090899397, "grad_norm": 1.1271870136260986, "learning_rate": 5.413200000000001e-06, "loss": 0.0212, "step": 22935 }, { "epoch": 1.2227968225195927, "grad_norm": 1.4164350032806396, "learning_rate": 5.4130000000000005e-06, "loss": 0.0186, "step": 22936 }, { "epoch": 1.2228501359492456, "grad_norm": 1.8547431230545044, "learning_rate": 5.4128e-06, "loss": 0.0371, "step": 22937 }, { "epoch": 1.2229034493788986, "grad_norm": 0.8633695244789124, "learning_rate": 5.4126000000000005e-06, "loss": 0.0095, "step": 22938 }, { "epoch": 1.2229567628085514, "grad_norm": 1.5866554975509644, "learning_rate": 5.412400000000001e-06, "loss": 0.0201, "step": 22939 }, { "epoch": 1.2230100762382043, "grad_norm": 3.5383338928222656, "learning_rate": 5.4122000000000005e-06, "loss": 0.0164, "step": 22940 }, { "epoch": 1.2230633896678573, "grad_norm": 2.0668723583221436, "learning_rate": 5.412000000000001e-06, "loss": 0.0301, "step": 22941 }, { "epoch": 1.2231167030975103, "grad_norm": 1.4320948123931885, "learning_rate": 5.4118000000000004e-06, "loss": 0.0168, "step": 22942 }, { "epoch": 1.2231700165271633, "grad_norm": 3.9256811141967773, "learning_rate": 5.4116e-06, "loss": 0.0132, "step": 22943 }, { "epoch": 1.223223329956816, "grad_norm": 1.0506088733673096, "learning_rate": 5.4114e-06, "loss": 0.0145, "step": 22944 }, { "epoch": 1.223276643386469, "grad_norm": 1.0704548358917236, "learning_rate": 5.411200000000001e-06, "loss": 0.0083, "step": 22945 }, { "epoch": 1.223329956816122, "grad_norm": 1.3658647537231445, "learning_rate": 5.411e-06, "loss": 0.0177, "step": 22946 }, { "epoch": 1.223383270245775, "grad_norm": 1.5682482719421387, "learning_rate": 5.410800000000001e-06, "loss": 0.0326, "step": 22947 }, { "epoch": 1.223436583675428, "grad_norm": 1.815889596939087, "learning_rate": 5.4106e-06, "loss": 0.0261, "step": 22948 }, { "epoch": 1.2234898971050807, "grad_norm": 1.5929615497589111, "learning_rate": 5.4104e-06, "loss": 0.0321, "step": 22949 }, { "epoch": 1.2235432105347337, "grad_norm": 3.6978824138641357, "learning_rate": 5.4102e-06, "loss": 0.0327, "step": 22950 }, { "epoch": 1.2235965239643867, "grad_norm": 1.5674747228622437, "learning_rate": 5.410000000000001e-06, "loss": 0.0677, "step": 22951 }, { "epoch": 1.2236498373940397, "grad_norm": 1.5747783184051514, "learning_rate": 5.4098e-06, "loss": 0.029, "step": 22952 }, { "epoch": 1.2237031508236924, "grad_norm": 2.7871923446655273, "learning_rate": 5.409600000000001e-06, "loss": 0.0314, "step": 22953 }, { "epoch": 1.2237564642533454, "grad_norm": 0.9252521991729736, "learning_rate": 5.4094e-06, "loss": 0.0182, "step": 22954 }, { "epoch": 1.2238097776829984, "grad_norm": 1.3131229877471924, "learning_rate": 5.4092e-06, "loss": 0.0114, "step": 22955 }, { "epoch": 1.2238630911126513, "grad_norm": 3.5785412788391113, "learning_rate": 5.409000000000001e-06, "loss": 0.0293, "step": 22956 }, { "epoch": 1.223916404542304, "grad_norm": 3.058607816696167, "learning_rate": 5.408800000000001e-06, "loss": 0.012, "step": 22957 }, { "epoch": 1.223969717971957, "grad_norm": 1.6950572729110718, "learning_rate": 5.4086e-06, "loss": 0.0199, "step": 22958 }, { "epoch": 1.22402303140161, "grad_norm": 0.501686692237854, "learning_rate": 5.408400000000001e-06, "loss": 0.005, "step": 22959 }, { "epoch": 1.224076344831263, "grad_norm": 2.309055805206299, "learning_rate": 5.4082e-06, "loss": 0.054, "step": 22960 }, { "epoch": 1.224129658260916, "grad_norm": 1.7263699769973755, "learning_rate": 5.408e-06, "loss": 0.0218, "step": 22961 }, { "epoch": 1.2241829716905688, "grad_norm": 2.604403495788574, "learning_rate": 5.407800000000001e-06, "loss": 0.0297, "step": 22962 }, { "epoch": 1.2242362851202218, "grad_norm": 1.9377995729446411, "learning_rate": 5.4076000000000005e-06, "loss": 0.0265, "step": 22963 }, { "epoch": 1.2242895985498747, "grad_norm": 0.7730605602264404, "learning_rate": 5.4074e-06, "loss": 0.0055, "step": 22964 }, { "epoch": 1.2243429119795277, "grad_norm": 1.9050674438476562, "learning_rate": 5.4072000000000005e-06, "loss": 0.0359, "step": 22965 }, { "epoch": 1.2243962254091807, "grad_norm": 1.984468936920166, "learning_rate": 5.407e-06, "loss": 0.0272, "step": 22966 }, { "epoch": 1.2244495388388335, "grad_norm": 2.4618890285491943, "learning_rate": 5.406800000000001e-06, "loss": 0.0511, "step": 22967 }, { "epoch": 1.2245028522684864, "grad_norm": 3.0059099197387695, "learning_rate": 5.406600000000001e-06, "loss": 0.0341, "step": 22968 }, { "epoch": 1.2245561656981394, "grad_norm": 2.829911231994629, "learning_rate": 5.4064000000000004e-06, "loss": 0.0683, "step": 22969 }, { "epoch": 1.2246094791277922, "grad_norm": 1.8667181730270386, "learning_rate": 5.4062e-06, "loss": 0.0316, "step": 22970 }, { "epoch": 1.2246627925574451, "grad_norm": 1.2902148962020874, "learning_rate": 5.406e-06, "loss": 0.0111, "step": 22971 }, { "epoch": 1.2247161059870981, "grad_norm": 3.846245527267456, "learning_rate": 5.4058e-06, "loss": 0.0388, "step": 22972 }, { "epoch": 1.224769419416751, "grad_norm": 1.5290824174880981, "learning_rate": 5.405600000000001e-06, "loss": 0.0138, "step": 22973 }, { "epoch": 1.224822732846404, "grad_norm": 1.0141777992248535, "learning_rate": 5.405400000000001e-06, "loss": 0.0083, "step": 22974 }, { "epoch": 1.2248760462760568, "grad_norm": 1.6357665061950684, "learning_rate": 5.4052e-06, "loss": 0.0384, "step": 22975 }, { "epoch": 1.2249293597057098, "grad_norm": 1.045701503753662, "learning_rate": 5.405e-06, "loss": 0.0162, "step": 22976 }, { "epoch": 1.2249826731353628, "grad_norm": 1.7328557968139648, "learning_rate": 5.4048e-06, "loss": 0.0377, "step": 22977 }, { "epoch": 1.2250359865650158, "grad_norm": 1.3812601566314697, "learning_rate": 5.404600000000001e-06, "loss": 0.0107, "step": 22978 }, { "epoch": 1.2250892999946688, "grad_norm": 0.22486919164657593, "learning_rate": 5.404400000000001e-06, "loss": 0.0027, "step": 22979 }, { "epoch": 1.2251426134243215, "grad_norm": 2.468127965927124, "learning_rate": 5.404200000000001e-06, "loss": 0.036, "step": 22980 }, { "epoch": 1.2251959268539745, "grad_norm": 2.6292712688446045, "learning_rate": 5.404e-06, "loss": 0.0279, "step": 22981 }, { "epoch": 1.2252492402836275, "grad_norm": 1.1970947980880737, "learning_rate": 5.4038e-06, "loss": 0.0117, "step": 22982 }, { "epoch": 1.2253025537132805, "grad_norm": 1.6170769929885864, "learning_rate": 5.4036e-06, "loss": 0.0236, "step": 22983 }, { "epoch": 1.2253558671429332, "grad_norm": 1.4688183069229126, "learning_rate": 5.403400000000001e-06, "loss": 0.0177, "step": 22984 }, { "epoch": 1.2254091805725862, "grad_norm": 0.6903534531593323, "learning_rate": 5.403200000000001e-06, "loss": 0.0051, "step": 22985 }, { "epoch": 1.2254624940022392, "grad_norm": 0.28339433670043945, "learning_rate": 5.403000000000001e-06, "loss": 0.0037, "step": 22986 }, { "epoch": 1.2255158074318921, "grad_norm": 1.8495562076568604, "learning_rate": 5.4028e-06, "loss": 0.0112, "step": 22987 }, { "epoch": 1.225569120861545, "grad_norm": 1.2021070718765259, "learning_rate": 5.4026e-06, "loss": 0.0181, "step": 22988 }, { "epoch": 1.2256224342911979, "grad_norm": 2.806621551513672, "learning_rate": 5.402400000000001e-06, "loss": 0.0214, "step": 22989 }, { "epoch": 1.2256757477208509, "grad_norm": 2.0267951488494873, "learning_rate": 5.4022000000000006e-06, "loss": 0.0296, "step": 22990 }, { "epoch": 1.2257290611505038, "grad_norm": 1.066694736480713, "learning_rate": 5.402000000000001e-06, "loss": 0.0135, "step": 22991 }, { "epoch": 1.2257823745801568, "grad_norm": 2.573795795440674, "learning_rate": 5.4018000000000005e-06, "loss": 0.0453, "step": 22992 }, { "epoch": 1.2258356880098096, "grad_norm": 1.4846792221069336, "learning_rate": 5.4016e-06, "loss": 0.0149, "step": 22993 }, { "epoch": 1.2258890014394626, "grad_norm": 4.747918128967285, "learning_rate": 5.4014e-06, "loss": 0.0409, "step": 22994 }, { "epoch": 1.2259423148691155, "grad_norm": 1.8169065713882446, "learning_rate": 5.401200000000001e-06, "loss": 0.0244, "step": 22995 }, { "epoch": 1.2259956282987685, "grad_norm": 0.8629744052886963, "learning_rate": 5.4010000000000005e-06, "loss": 0.0181, "step": 22996 }, { "epoch": 1.2260489417284215, "grad_norm": 2.539945363998413, "learning_rate": 5.400800000000001e-06, "loss": 0.0582, "step": 22997 }, { "epoch": 1.2261022551580743, "grad_norm": 2.4847259521484375, "learning_rate": 5.4006000000000004e-06, "loss": 0.0204, "step": 22998 }, { "epoch": 1.2261555685877272, "grad_norm": 3.813243865966797, "learning_rate": 5.4004e-06, "loss": 0.0552, "step": 22999 }, { "epoch": 1.2262088820173802, "grad_norm": 2.35133957862854, "learning_rate": 5.4001999999999996e-06, "loss": 0.0316, "step": 23000 }, { "epoch": 1.2262621954470332, "grad_norm": 0.9812464118003845, "learning_rate": 5.400000000000001e-06, "loss": 0.0202, "step": 23001 }, { "epoch": 1.226315508876686, "grad_norm": 3.1441919803619385, "learning_rate": 5.3998e-06, "loss": 0.0312, "step": 23002 }, { "epoch": 1.226368822306339, "grad_norm": 1.8754633665084839, "learning_rate": 5.399600000000001e-06, "loss": 0.0263, "step": 23003 }, { "epoch": 1.226422135735992, "grad_norm": 1.7512545585632324, "learning_rate": 5.3994e-06, "loss": 0.0107, "step": 23004 }, { "epoch": 1.2264754491656449, "grad_norm": 1.651180624961853, "learning_rate": 5.3992e-06, "loss": 0.0188, "step": 23005 }, { "epoch": 1.2265287625952976, "grad_norm": 0.8436846137046814, "learning_rate": 5.399000000000001e-06, "loss": 0.0144, "step": 23006 }, { "epoch": 1.2265820760249506, "grad_norm": 1.3044041395187378, "learning_rate": 5.398800000000001e-06, "loss": 0.0123, "step": 23007 }, { "epoch": 1.2266353894546036, "grad_norm": 1.5807380676269531, "learning_rate": 5.3986e-06, "loss": 0.0191, "step": 23008 }, { "epoch": 1.2266887028842566, "grad_norm": 1.2588646411895752, "learning_rate": 5.398400000000001e-06, "loss": 0.0232, "step": 23009 }, { "epoch": 1.2267420163139096, "grad_norm": 2.2057228088378906, "learning_rate": 5.3982e-06, "loss": 0.0407, "step": 23010 }, { "epoch": 1.2267953297435623, "grad_norm": 1.3189283609390259, "learning_rate": 5.398e-06, "loss": 0.013, "step": 23011 }, { "epoch": 1.2268486431732153, "grad_norm": 1.6292181015014648, "learning_rate": 5.397800000000001e-06, "loss": 0.03, "step": 23012 }, { "epoch": 1.2269019566028683, "grad_norm": 2.192415237426758, "learning_rate": 5.397600000000001e-06, "loss": 0.0324, "step": 23013 }, { "epoch": 1.2269552700325213, "grad_norm": 1.067000389099121, "learning_rate": 5.3974e-06, "loss": 0.0112, "step": 23014 }, { "epoch": 1.2270085834621742, "grad_norm": 1.0139710903167725, "learning_rate": 5.397200000000001e-06, "loss": 0.0117, "step": 23015 }, { "epoch": 1.227061896891827, "grad_norm": 1.2435791492462158, "learning_rate": 5.397e-06, "loss": 0.0112, "step": 23016 }, { "epoch": 1.22711521032148, "grad_norm": 1.6716256141662598, "learning_rate": 5.3968000000000006e-06, "loss": 0.0223, "step": 23017 }, { "epoch": 1.227168523751133, "grad_norm": 3.242769479751587, "learning_rate": 5.396600000000001e-06, "loss": 0.0186, "step": 23018 }, { "epoch": 1.227221837180786, "grad_norm": 2.6476776599884033, "learning_rate": 5.3964000000000005e-06, "loss": 0.0171, "step": 23019 }, { "epoch": 1.2272751506104387, "grad_norm": 2.277686834335327, "learning_rate": 5.3962e-06, "loss": 0.0387, "step": 23020 }, { "epoch": 1.2273284640400917, "grad_norm": 2.7577712535858154, "learning_rate": 5.3960000000000005e-06, "loss": 0.0424, "step": 23021 }, { "epoch": 1.2273817774697446, "grad_norm": 2.809851884841919, "learning_rate": 5.3958e-06, "loss": 0.0549, "step": 23022 }, { "epoch": 1.2274350908993976, "grad_norm": 2.429971694946289, "learning_rate": 5.3956000000000005e-06, "loss": 0.0689, "step": 23023 }, { "epoch": 1.2274884043290504, "grad_norm": 1.207331657409668, "learning_rate": 5.395400000000001e-06, "loss": 0.0108, "step": 23024 }, { "epoch": 1.2275417177587034, "grad_norm": 1.064379096031189, "learning_rate": 5.3952000000000005e-06, "loss": 0.0059, "step": 23025 }, { "epoch": 1.2275950311883563, "grad_norm": 2.5117416381835938, "learning_rate": 5.395e-06, "loss": 0.043, "step": 23026 }, { "epoch": 1.2276483446180093, "grad_norm": 1.307430386543274, "learning_rate": 5.3948000000000004e-06, "loss": 0.0078, "step": 23027 }, { "epoch": 1.2277016580476623, "grad_norm": 2.3674142360687256, "learning_rate": 5.394600000000001e-06, "loss": 0.0099, "step": 23028 }, { "epoch": 1.227754971477315, "grad_norm": 2.701763391494751, "learning_rate": 5.3944e-06, "loss": 0.0453, "step": 23029 }, { "epoch": 1.227808284906968, "grad_norm": 2.7074122428894043, "learning_rate": 5.394200000000001e-06, "loss": 0.0501, "step": 23030 }, { "epoch": 1.227861598336621, "grad_norm": 2.4257097244262695, "learning_rate": 5.394e-06, "loss": 0.0413, "step": 23031 }, { "epoch": 1.227914911766274, "grad_norm": 2.405971050262451, "learning_rate": 5.3938e-06, "loss": 0.0131, "step": 23032 }, { "epoch": 1.227968225195927, "grad_norm": 1.1453508138656616, "learning_rate": 5.3936e-06, "loss": 0.009, "step": 23033 }, { "epoch": 1.2280215386255797, "grad_norm": 2.189829111099243, "learning_rate": 5.393400000000001e-06, "loss": 0.059, "step": 23034 }, { "epoch": 1.2280748520552327, "grad_norm": 2.394824504852295, "learning_rate": 5.3932e-06, "loss": 0.0402, "step": 23035 }, { "epoch": 1.2281281654848857, "grad_norm": 2.3742916584014893, "learning_rate": 5.393000000000001e-06, "loss": 0.0294, "step": 23036 }, { "epoch": 1.2281814789145387, "grad_norm": 1.0767004489898682, "learning_rate": 5.3928e-06, "loss": 0.0203, "step": 23037 }, { "epoch": 1.2282347923441914, "grad_norm": 1.6106727123260498, "learning_rate": 5.3926e-06, "loss": 0.0153, "step": 23038 }, { "epoch": 1.2282881057738444, "grad_norm": 1.1809712648391724, "learning_rate": 5.392400000000001e-06, "loss": 0.0102, "step": 23039 }, { "epoch": 1.2283414192034974, "grad_norm": 2.3787596225738525, "learning_rate": 5.392200000000001e-06, "loss": 0.0126, "step": 23040 }, { "epoch": 1.2283947326331504, "grad_norm": 3.351419687271118, "learning_rate": 5.392e-06, "loss": 0.0592, "step": 23041 }, { "epoch": 1.2284480460628031, "grad_norm": 1.0594912767410278, "learning_rate": 5.391800000000001e-06, "loss": 0.0106, "step": 23042 }, { "epoch": 1.228501359492456, "grad_norm": 2.3243699073791504, "learning_rate": 5.3916e-06, "loss": 0.0286, "step": 23043 }, { "epoch": 1.228554672922109, "grad_norm": 3.258085012435913, "learning_rate": 5.3914e-06, "loss": 0.039, "step": 23044 }, { "epoch": 1.228607986351762, "grad_norm": 1.9448403120040894, "learning_rate": 5.391200000000001e-06, "loss": 0.0313, "step": 23045 }, { "epoch": 1.228661299781415, "grad_norm": 1.0482420921325684, "learning_rate": 5.3910000000000006e-06, "loss": 0.0235, "step": 23046 }, { "epoch": 1.2287146132110678, "grad_norm": 2.4664580821990967, "learning_rate": 5.3908e-06, "loss": 0.0271, "step": 23047 }, { "epoch": 1.2287679266407208, "grad_norm": 1.7034248113632202, "learning_rate": 5.3906000000000005e-06, "loss": 0.0292, "step": 23048 }, { "epoch": 1.2288212400703737, "grad_norm": 1.5457884073257446, "learning_rate": 5.3904e-06, "loss": 0.0139, "step": 23049 }, { "epoch": 1.2288745535000267, "grad_norm": 0.7480394840240479, "learning_rate": 5.390200000000001e-06, "loss": 0.0246, "step": 23050 }, { "epoch": 1.2289278669296797, "grad_norm": 1.353724718093872, "learning_rate": 5.390000000000001e-06, "loss": 0.0131, "step": 23051 }, { "epoch": 1.2289811803593325, "grad_norm": 1.7391993999481201, "learning_rate": 5.3898000000000005e-06, "loss": 0.0189, "step": 23052 }, { "epoch": 1.2290344937889854, "grad_norm": 1.2706217765808105, "learning_rate": 5.3896e-06, "loss": 0.0331, "step": 23053 }, { "epoch": 1.2290878072186384, "grad_norm": 2.7804651260375977, "learning_rate": 5.3894000000000004e-06, "loss": 0.0636, "step": 23054 }, { "epoch": 1.2291411206482914, "grad_norm": 0.9971290826797485, "learning_rate": 5.3892e-06, "loss": 0.0242, "step": 23055 }, { "epoch": 1.2291944340779442, "grad_norm": 2.9658496379852295, "learning_rate": 5.389000000000001e-06, "loss": 0.0334, "step": 23056 }, { "epoch": 1.2292477475075971, "grad_norm": 2.6980018615722656, "learning_rate": 5.388800000000001e-06, "loss": 0.0142, "step": 23057 }, { "epoch": 1.2293010609372501, "grad_norm": 0.8019313216209412, "learning_rate": 5.3886e-06, "loss": 0.0108, "step": 23058 }, { "epoch": 1.229354374366903, "grad_norm": 1.842333197593689, "learning_rate": 5.3884e-06, "loss": 0.0346, "step": 23059 }, { "epoch": 1.2294076877965558, "grad_norm": 1.8876875638961792, "learning_rate": 5.3882e-06, "loss": 0.0111, "step": 23060 }, { "epoch": 1.2294610012262088, "grad_norm": 2.3889527320861816, "learning_rate": 5.388e-06, "loss": 0.024, "step": 23061 }, { "epoch": 1.2295143146558618, "grad_norm": 1.0803951025009155, "learning_rate": 5.387800000000001e-06, "loss": 0.0079, "step": 23062 }, { "epoch": 1.2295676280855148, "grad_norm": 1.6595609188079834, "learning_rate": 5.387600000000001e-06, "loss": 0.0106, "step": 23063 }, { "epoch": 1.2296209415151678, "grad_norm": 1.387824296951294, "learning_rate": 5.3874e-06, "loss": 0.0208, "step": 23064 }, { "epoch": 1.2296742549448205, "grad_norm": 2.433274507522583, "learning_rate": 5.3872e-06, "loss": 0.0348, "step": 23065 }, { "epoch": 1.2297275683744735, "grad_norm": 3.550088882446289, "learning_rate": 5.387e-06, "loss": 0.0238, "step": 23066 }, { "epoch": 1.2297808818041265, "grad_norm": 1.9568190574645996, "learning_rate": 5.386800000000001e-06, "loss": 0.0167, "step": 23067 }, { "epoch": 1.2298341952337795, "grad_norm": 2.0776994228363037, "learning_rate": 5.386600000000001e-06, "loss": 0.05, "step": 23068 }, { "epoch": 1.2298875086634324, "grad_norm": 2.519291639328003, "learning_rate": 5.386400000000001e-06, "loss": 0.0236, "step": 23069 }, { "epoch": 1.2299408220930852, "grad_norm": 1.185625433921814, "learning_rate": 5.3862e-06, "loss": 0.0208, "step": 23070 }, { "epoch": 1.2299941355227382, "grad_norm": 2.715853691101074, "learning_rate": 5.386e-06, "loss": 0.0365, "step": 23071 }, { "epoch": 1.2300474489523912, "grad_norm": 2.2794251441955566, "learning_rate": 5.3858e-06, "loss": 0.0175, "step": 23072 }, { "epoch": 1.2301007623820441, "grad_norm": 2.4919216632843018, "learning_rate": 5.385600000000001e-06, "loss": 0.0384, "step": 23073 }, { "epoch": 1.230154075811697, "grad_norm": 1.8524389266967773, "learning_rate": 5.385400000000001e-06, "loss": 0.0183, "step": 23074 }, { "epoch": 1.2302073892413499, "grad_norm": 0.9473516941070557, "learning_rate": 5.3852000000000005e-06, "loss": 0.0061, "step": 23075 }, { "epoch": 1.2302607026710028, "grad_norm": 1.2024009227752686, "learning_rate": 5.385e-06, "loss": 0.0121, "step": 23076 }, { "epoch": 1.2303140161006558, "grad_norm": 1.4296776056289673, "learning_rate": 5.3848e-06, "loss": 0.0145, "step": 23077 }, { "epoch": 1.2303673295303086, "grad_norm": 1.56759512424469, "learning_rate": 5.384600000000001e-06, "loss": 0.0204, "step": 23078 }, { "epoch": 1.2304206429599616, "grad_norm": 2.634186029434204, "learning_rate": 5.3844000000000005e-06, "loss": 0.0262, "step": 23079 }, { "epoch": 1.2304739563896145, "grad_norm": 2.497588634490967, "learning_rate": 5.384200000000001e-06, "loss": 0.0297, "step": 23080 }, { "epoch": 1.2305272698192675, "grad_norm": 2.193188190460205, "learning_rate": 5.3840000000000005e-06, "loss": 0.0511, "step": 23081 }, { "epoch": 1.2305805832489205, "grad_norm": 1.1415127515792847, "learning_rate": 5.3838e-06, "loss": 0.0085, "step": 23082 }, { "epoch": 1.2306338966785733, "grad_norm": 0.47426334023475647, "learning_rate": 5.3836e-06, "loss": 0.0043, "step": 23083 }, { "epoch": 1.2306872101082262, "grad_norm": 1.2906290292739868, "learning_rate": 5.383400000000001e-06, "loss": 0.0339, "step": 23084 }, { "epoch": 1.2307405235378792, "grad_norm": 3.5650954246520996, "learning_rate": 5.3832e-06, "loss": 0.0286, "step": 23085 }, { "epoch": 1.2307938369675322, "grad_norm": 2.333348035812378, "learning_rate": 5.383000000000001e-06, "loss": 0.0142, "step": 23086 }, { "epoch": 1.2308471503971852, "grad_norm": 1.653635859489441, "learning_rate": 5.3828e-06, "loss": 0.0276, "step": 23087 }, { "epoch": 1.230900463826838, "grad_norm": 1.1625889539718628, "learning_rate": 5.3826e-06, "loss": 0.0106, "step": 23088 }, { "epoch": 1.230953777256491, "grad_norm": 1.7284610271453857, "learning_rate": 5.382400000000001e-06, "loss": 0.0233, "step": 23089 }, { "epoch": 1.231007090686144, "grad_norm": 2.0016372203826904, "learning_rate": 5.382200000000001e-06, "loss": 0.0277, "step": 23090 }, { "epoch": 1.2310604041157966, "grad_norm": 1.383759617805481, "learning_rate": 5.382e-06, "loss": 0.0123, "step": 23091 }, { "epoch": 1.2311137175454496, "grad_norm": 3.207916498184204, "learning_rate": 5.381800000000001e-06, "loss": 0.0459, "step": 23092 }, { "epoch": 1.2311670309751026, "grad_norm": 5.215295791625977, "learning_rate": 5.3816e-06, "loss": 0.0359, "step": 23093 }, { "epoch": 1.2312203444047556, "grad_norm": 1.479993462562561, "learning_rate": 5.3814e-06, "loss": 0.0322, "step": 23094 }, { "epoch": 1.2312736578344086, "grad_norm": 3.2900218963623047, "learning_rate": 5.381200000000001e-06, "loss": 0.0254, "step": 23095 }, { "epoch": 1.2313269712640613, "grad_norm": 1.265703558921814, "learning_rate": 5.381000000000001e-06, "loss": 0.0099, "step": 23096 }, { "epoch": 1.2313802846937143, "grad_norm": 2.2372496128082275, "learning_rate": 5.3808e-06, "loss": 0.0301, "step": 23097 }, { "epoch": 1.2314335981233673, "grad_norm": 1.6821781396865845, "learning_rate": 5.380600000000001e-06, "loss": 0.0442, "step": 23098 }, { "epoch": 1.2314869115530203, "grad_norm": 0.8700587749481201, "learning_rate": 5.3804e-06, "loss": 0.0083, "step": 23099 }, { "epoch": 1.2315402249826732, "grad_norm": 3.0708110332489014, "learning_rate": 5.380200000000001e-06, "loss": 0.0174, "step": 23100 }, { "epoch": 1.231593538412326, "grad_norm": 2.57090163230896, "learning_rate": 5.380000000000001e-06, "loss": 0.0396, "step": 23101 }, { "epoch": 1.231646851841979, "grad_norm": 2.9578816890716553, "learning_rate": 5.3798000000000006e-06, "loss": 0.032, "step": 23102 }, { "epoch": 1.231700165271632, "grad_norm": 1.4693373441696167, "learning_rate": 5.3796e-06, "loss": 0.0256, "step": 23103 }, { "epoch": 1.231753478701285, "grad_norm": 2.3312573432922363, "learning_rate": 5.3794000000000005e-06, "loss": 0.0323, "step": 23104 }, { "epoch": 1.2318067921309377, "grad_norm": 3.562058925628662, "learning_rate": 5.3792e-06, "loss": 0.036, "step": 23105 }, { "epoch": 1.2318601055605907, "grad_norm": 1.6926673650741577, "learning_rate": 5.3790000000000005e-06, "loss": 0.032, "step": 23106 }, { "epoch": 1.2319134189902436, "grad_norm": 1.4953237771987915, "learning_rate": 5.378800000000001e-06, "loss": 0.0137, "step": 23107 }, { "epoch": 1.2319667324198966, "grad_norm": 1.3450947999954224, "learning_rate": 5.3786000000000005e-06, "loss": 0.0137, "step": 23108 }, { "epoch": 1.2320200458495494, "grad_norm": 0.8366996645927429, "learning_rate": 5.3784e-06, "loss": 0.007, "step": 23109 }, { "epoch": 1.2320733592792024, "grad_norm": 1.2241488695144653, "learning_rate": 5.3782000000000004e-06, "loss": 0.0174, "step": 23110 }, { "epoch": 1.2321266727088553, "grad_norm": 1.5868536233901978, "learning_rate": 5.378e-06, "loss": 0.0326, "step": 23111 }, { "epoch": 1.2321799861385083, "grad_norm": 1.6695563793182373, "learning_rate": 5.3778e-06, "loss": 0.0196, "step": 23112 }, { "epoch": 1.2322332995681613, "grad_norm": 1.757098913192749, "learning_rate": 5.377600000000001e-06, "loss": 0.0174, "step": 23113 }, { "epoch": 1.232286612997814, "grad_norm": 1.7170039415359497, "learning_rate": 5.3774e-06, "loss": 0.0391, "step": 23114 }, { "epoch": 1.232339926427467, "grad_norm": 2.6613993644714355, "learning_rate": 5.3772e-06, "loss": 0.0205, "step": 23115 }, { "epoch": 1.23239323985712, "grad_norm": 1.049389123916626, "learning_rate": 5.377e-06, "loss": 0.0117, "step": 23116 }, { "epoch": 1.232446553286773, "grad_norm": 2.053344249725342, "learning_rate": 5.376800000000001e-06, "loss": 0.0361, "step": 23117 }, { "epoch": 1.232499866716426, "grad_norm": 3.182260513305664, "learning_rate": 5.3766e-06, "loss": 0.0612, "step": 23118 }, { "epoch": 1.2325531801460787, "grad_norm": 0.9875069856643677, "learning_rate": 5.376400000000001e-06, "loss": 0.0061, "step": 23119 }, { "epoch": 1.2326064935757317, "grad_norm": 1.3229949474334717, "learning_rate": 5.3762e-06, "loss": 0.0148, "step": 23120 }, { "epoch": 1.2326598070053847, "grad_norm": 2.5907928943634033, "learning_rate": 5.376e-06, "loss": 0.0241, "step": 23121 }, { "epoch": 1.2327131204350377, "grad_norm": 1.1449806690216064, "learning_rate": 5.3758e-06, "loss": 0.025, "step": 23122 }, { "epoch": 1.2327664338646904, "grad_norm": 1.742754340171814, "learning_rate": 5.375600000000001e-06, "loss": 0.0272, "step": 23123 }, { "epoch": 1.2328197472943434, "grad_norm": 2.4838900566101074, "learning_rate": 5.3754e-06, "loss": 0.0203, "step": 23124 }, { "epoch": 1.2328730607239964, "grad_norm": 2.750917911529541, "learning_rate": 5.375200000000001e-06, "loss": 0.0267, "step": 23125 }, { "epoch": 1.2329263741536494, "grad_norm": 1.1232688426971436, "learning_rate": 5.375e-06, "loss": 0.0112, "step": 23126 }, { "epoch": 1.2329796875833021, "grad_norm": 1.1472854614257812, "learning_rate": 5.3748e-06, "loss": 0.0242, "step": 23127 }, { "epoch": 1.233033001012955, "grad_norm": 1.8800567388534546, "learning_rate": 5.374600000000001e-06, "loss": 0.0405, "step": 23128 }, { "epoch": 1.233086314442608, "grad_norm": 1.758323073387146, "learning_rate": 5.374400000000001e-06, "loss": 0.0393, "step": 23129 }, { "epoch": 1.233139627872261, "grad_norm": 0.8720673322677612, "learning_rate": 5.3742e-06, "loss": 0.0075, "step": 23130 }, { "epoch": 1.233192941301914, "grad_norm": 1.5003890991210938, "learning_rate": 5.3740000000000006e-06, "loss": 0.0293, "step": 23131 }, { "epoch": 1.2332462547315668, "grad_norm": 1.8700928688049316, "learning_rate": 5.3738e-06, "loss": 0.0099, "step": 23132 }, { "epoch": 1.2332995681612198, "grad_norm": 1.1398781538009644, "learning_rate": 5.3736e-06, "loss": 0.0125, "step": 23133 }, { "epoch": 1.2333528815908728, "grad_norm": 0.8633454442024231, "learning_rate": 5.373400000000001e-06, "loss": 0.0217, "step": 23134 }, { "epoch": 1.2334061950205257, "grad_norm": 2.0083234310150146, "learning_rate": 5.3732000000000005e-06, "loss": 0.0249, "step": 23135 }, { "epoch": 1.2334595084501787, "grad_norm": 1.2725499868392944, "learning_rate": 5.373e-06, "loss": 0.0334, "step": 23136 }, { "epoch": 1.2335128218798315, "grad_norm": 1.0888636112213135, "learning_rate": 5.3728000000000005e-06, "loss": 0.0059, "step": 23137 }, { "epoch": 1.2335661353094844, "grad_norm": 2.168473482131958, "learning_rate": 5.3726e-06, "loss": 0.0123, "step": 23138 }, { "epoch": 1.2336194487391374, "grad_norm": 2.3677735328674316, "learning_rate": 5.372400000000001e-06, "loss": 0.0353, "step": 23139 }, { "epoch": 1.2336727621687904, "grad_norm": 1.6764113903045654, "learning_rate": 5.372200000000001e-06, "loss": 0.023, "step": 23140 }, { "epoch": 1.2337260755984432, "grad_norm": 0.8529396057128906, "learning_rate": 5.372e-06, "loss": 0.0142, "step": 23141 }, { "epoch": 1.2337793890280961, "grad_norm": 6.758810997009277, "learning_rate": 5.3718e-06, "loss": 0.0484, "step": 23142 }, { "epoch": 1.2338327024577491, "grad_norm": 2.47818660736084, "learning_rate": 5.3716e-06, "loss": 0.0432, "step": 23143 }, { "epoch": 1.233886015887402, "grad_norm": 2.689804792404175, "learning_rate": 5.3714e-06, "loss": 0.0446, "step": 23144 }, { "epoch": 1.2339393293170549, "grad_norm": 0.45778366923332214, "learning_rate": 5.371200000000001e-06, "loss": 0.0052, "step": 23145 }, { "epoch": 1.2339926427467078, "grad_norm": 2.5061495304107666, "learning_rate": 5.371000000000001e-06, "loss": 0.0474, "step": 23146 }, { "epoch": 1.2340459561763608, "grad_norm": 2.151104211807251, "learning_rate": 5.3708e-06, "loss": 0.0223, "step": 23147 }, { "epoch": 1.2340992696060138, "grad_norm": 1.9111573696136475, "learning_rate": 5.3706e-06, "loss": 0.0263, "step": 23148 }, { "epoch": 1.2341525830356668, "grad_norm": 2.879155158996582, "learning_rate": 5.3704e-06, "loss": 0.0366, "step": 23149 }, { "epoch": 1.2342058964653195, "grad_norm": 2.930490732192993, "learning_rate": 5.370200000000001e-06, "loss": 0.0284, "step": 23150 }, { "epoch": 1.2342592098949725, "grad_norm": 1.2948209047317505, "learning_rate": 5.370000000000001e-06, "loss": 0.0146, "step": 23151 }, { "epoch": 1.2343125233246255, "grad_norm": 1.736657738685608, "learning_rate": 5.369800000000001e-06, "loss": 0.0178, "step": 23152 }, { "epoch": 1.2343658367542785, "grad_norm": 1.6478643417358398, "learning_rate": 5.3696e-06, "loss": 0.0195, "step": 23153 }, { "epoch": 1.2344191501839314, "grad_norm": 1.9791054725646973, "learning_rate": 5.3694e-06, "loss": 0.0222, "step": 23154 }, { "epoch": 1.2344724636135842, "grad_norm": 1.5778979063034058, "learning_rate": 5.3692e-06, "loss": 0.0241, "step": 23155 }, { "epoch": 1.2345257770432372, "grad_norm": 2.9619243144989014, "learning_rate": 5.369000000000001e-06, "loss": 0.0621, "step": 23156 }, { "epoch": 1.2345790904728902, "grad_norm": 1.3349041938781738, "learning_rate": 5.368800000000001e-06, "loss": 0.0134, "step": 23157 }, { "epoch": 1.2346324039025431, "grad_norm": 0.9582570791244507, "learning_rate": 5.3686000000000006e-06, "loss": 0.0113, "step": 23158 }, { "epoch": 1.234685717332196, "grad_norm": 1.76869535446167, "learning_rate": 5.3684e-06, "loss": 0.0162, "step": 23159 }, { "epoch": 1.2347390307618489, "grad_norm": 1.2499499320983887, "learning_rate": 5.3682e-06, "loss": 0.0225, "step": 23160 }, { "epoch": 1.2347923441915019, "grad_norm": 7.108985900878906, "learning_rate": 5.368000000000001e-06, "loss": 0.0638, "step": 23161 }, { "epoch": 1.2348456576211548, "grad_norm": 2.6495609283447266, "learning_rate": 5.3678000000000005e-06, "loss": 0.0382, "step": 23162 }, { "epoch": 1.2348989710508076, "grad_norm": 3.802412748336792, "learning_rate": 5.367600000000001e-06, "loss": 0.0324, "step": 23163 }, { "epoch": 1.2349522844804606, "grad_norm": 3.6012063026428223, "learning_rate": 5.3674000000000005e-06, "loss": 0.0367, "step": 23164 }, { "epoch": 1.2350055979101136, "grad_norm": 0.7778550386428833, "learning_rate": 5.3672e-06, "loss": 0.0182, "step": 23165 }, { "epoch": 1.2350589113397665, "grad_norm": 0.21047602593898773, "learning_rate": 5.367e-06, "loss": 0.0019, "step": 23166 }, { "epoch": 1.2351122247694195, "grad_norm": 0.7527613639831543, "learning_rate": 5.366800000000001e-06, "loss": 0.0195, "step": 23167 }, { "epoch": 1.2351655381990723, "grad_norm": 0.6822724938392639, "learning_rate": 5.3666e-06, "loss": 0.0035, "step": 23168 }, { "epoch": 1.2352188516287252, "grad_norm": 0.8671835064888, "learning_rate": 5.366400000000001e-06, "loss": 0.0057, "step": 23169 }, { "epoch": 1.2352721650583782, "grad_norm": 1.4538220167160034, "learning_rate": 5.3662e-06, "loss": 0.0221, "step": 23170 }, { "epoch": 1.2353254784880312, "grad_norm": 1.9749152660369873, "learning_rate": 5.366e-06, "loss": 0.0123, "step": 23171 }, { "epoch": 1.2353787919176842, "grad_norm": 0.9100317358970642, "learning_rate": 5.3657999999999995e-06, "loss": 0.018, "step": 23172 }, { "epoch": 1.235432105347337, "grad_norm": 2.5200750827789307, "learning_rate": 5.365600000000001e-06, "loss": 0.0314, "step": 23173 }, { "epoch": 1.23548541877699, "grad_norm": 3.6222493648529053, "learning_rate": 5.3654e-06, "loss": 0.0092, "step": 23174 }, { "epoch": 1.235538732206643, "grad_norm": 2.1742136478424072, "learning_rate": 5.365200000000001e-06, "loss": 0.0423, "step": 23175 }, { "epoch": 1.2355920456362959, "grad_norm": 0.8220149874687195, "learning_rate": 5.365e-06, "loss": 0.0056, "step": 23176 }, { "epoch": 1.2356453590659486, "grad_norm": 2.2164478302001953, "learning_rate": 5.3648e-06, "loss": 0.0174, "step": 23177 }, { "epoch": 1.2356986724956016, "grad_norm": 2.5895464420318604, "learning_rate": 5.364600000000001e-06, "loss": 0.0398, "step": 23178 }, { "epoch": 1.2357519859252546, "grad_norm": 1.7279428243637085, "learning_rate": 5.364400000000001e-06, "loss": 0.0228, "step": 23179 }, { "epoch": 1.2358052993549076, "grad_norm": 2.321937084197998, "learning_rate": 5.3642e-06, "loss": 0.0384, "step": 23180 }, { "epoch": 1.2358586127845603, "grad_norm": 2.319701910018921, "learning_rate": 5.364000000000001e-06, "loss": 0.0237, "step": 23181 }, { "epoch": 1.2359119262142133, "grad_norm": 1.181785225868225, "learning_rate": 5.3638e-06, "loss": 0.0123, "step": 23182 }, { "epoch": 1.2359652396438663, "grad_norm": 2.18179988861084, "learning_rate": 5.3636e-06, "loss": 0.0468, "step": 23183 }, { "epoch": 1.2360185530735193, "grad_norm": 1.3905194997787476, "learning_rate": 5.363400000000001e-06, "loss": 0.0141, "step": 23184 }, { "epoch": 1.2360718665031722, "grad_norm": 1.1414369344711304, "learning_rate": 5.363200000000001e-06, "loss": 0.0165, "step": 23185 }, { "epoch": 1.236125179932825, "grad_norm": 0.8133270740509033, "learning_rate": 5.363e-06, "loss": 0.006, "step": 23186 }, { "epoch": 1.236178493362478, "grad_norm": 3.3755645751953125, "learning_rate": 5.3628000000000006e-06, "loss": 0.0333, "step": 23187 }, { "epoch": 1.236231806792131, "grad_norm": 2.1497151851654053, "learning_rate": 5.3626e-06, "loss": 0.0379, "step": 23188 }, { "epoch": 1.236285120221784, "grad_norm": 1.5641534328460693, "learning_rate": 5.3624000000000005e-06, "loss": 0.0318, "step": 23189 }, { "epoch": 1.236338433651437, "grad_norm": 2.5212831497192383, "learning_rate": 5.362200000000001e-06, "loss": 0.0488, "step": 23190 }, { "epoch": 1.2363917470810897, "grad_norm": 1.5969823598861694, "learning_rate": 5.3620000000000005e-06, "loss": 0.0412, "step": 23191 }, { "epoch": 1.2364450605107427, "grad_norm": 2.096905469894409, "learning_rate": 5.3618e-06, "loss": 0.0532, "step": 23192 }, { "epoch": 1.2364983739403956, "grad_norm": 1.1207358837127686, "learning_rate": 5.3616000000000005e-06, "loss": 0.0098, "step": 23193 }, { "epoch": 1.2365516873700484, "grad_norm": 2.2870798110961914, "learning_rate": 5.3614e-06, "loss": 0.0308, "step": 23194 }, { "epoch": 1.2366050007997014, "grad_norm": 1.6261574029922485, "learning_rate": 5.3612000000000004e-06, "loss": 0.0142, "step": 23195 }, { "epoch": 1.2366583142293543, "grad_norm": 0.6611846089363098, "learning_rate": 5.361000000000001e-06, "loss": 0.0055, "step": 23196 }, { "epoch": 1.2367116276590073, "grad_norm": 0.8664914965629578, "learning_rate": 5.3608e-06, "loss": 0.0087, "step": 23197 }, { "epoch": 1.2367649410886603, "grad_norm": 1.9235247373580933, "learning_rate": 5.3606e-06, "loss": 0.0482, "step": 23198 }, { "epoch": 1.236818254518313, "grad_norm": 2.102790117263794, "learning_rate": 5.3604e-06, "loss": 0.0353, "step": 23199 }, { "epoch": 1.236871567947966, "grad_norm": 2.0936660766601562, "learning_rate": 5.360200000000001e-06, "loss": 0.0191, "step": 23200 }, { "epoch": 1.236924881377619, "grad_norm": 0.5910899639129639, "learning_rate": 5.36e-06, "loss": 0.0045, "step": 23201 }, { "epoch": 1.236978194807272, "grad_norm": 1.831555962562561, "learning_rate": 5.359800000000001e-06, "loss": 0.0107, "step": 23202 }, { "epoch": 1.237031508236925, "grad_norm": 2.2373080253601074, "learning_rate": 5.3596e-06, "loss": 0.0326, "step": 23203 }, { "epoch": 1.2370848216665777, "grad_norm": 3.842430830001831, "learning_rate": 5.3594e-06, "loss": 0.0281, "step": 23204 }, { "epoch": 1.2371381350962307, "grad_norm": 1.349426507949829, "learning_rate": 5.3592e-06, "loss": 0.0314, "step": 23205 }, { "epoch": 1.2371914485258837, "grad_norm": 1.2761281728744507, "learning_rate": 5.359000000000001e-06, "loss": 0.0208, "step": 23206 }, { "epoch": 1.2372447619555367, "grad_norm": 2.0119049549102783, "learning_rate": 5.3588e-06, "loss": 0.0143, "step": 23207 }, { "epoch": 1.2372980753851894, "grad_norm": 1.5294793844223022, "learning_rate": 5.358600000000001e-06, "loss": 0.0121, "step": 23208 }, { "epoch": 1.2373513888148424, "grad_norm": 4.544086933135986, "learning_rate": 5.3584e-06, "loss": 0.0485, "step": 23209 }, { "epoch": 1.2374047022444954, "grad_norm": 1.9848119020462036, "learning_rate": 5.3582e-06, "loss": 0.04, "step": 23210 }, { "epoch": 1.2374580156741484, "grad_norm": 1.5565844774246216, "learning_rate": 5.358000000000001e-06, "loss": 0.0268, "step": 23211 }, { "epoch": 1.2375113291038011, "grad_norm": 0.8813291788101196, "learning_rate": 5.357800000000001e-06, "loss": 0.0076, "step": 23212 }, { "epoch": 1.237564642533454, "grad_norm": 1.9356663227081299, "learning_rate": 5.3576e-06, "loss": 0.0161, "step": 23213 }, { "epoch": 1.237617955963107, "grad_norm": 1.7939547300338745, "learning_rate": 5.357400000000001e-06, "loss": 0.0175, "step": 23214 }, { "epoch": 1.23767126939276, "grad_norm": 1.4752272367477417, "learning_rate": 5.3572e-06, "loss": 0.0072, "step": 23215 }, { "epoch": 1.237724582822413, "grad_norm": 2.7587151527404785, "learning_rate": 5.357e-06, "loss": 0.0311, "step": 23216 }, { "epoch": 1.2377778962520658, "grad_norm": 5.311257362365723, "learning_rate": 5.356800000000001e-06, "loss": 0.034, "step": 23217 }, { "epoch": 1.2378312096817188, "grad_norm": 1.8451776504516602, "learning_rate": 5.3566000000000005e-06, "loss": 0.0415, "step": 23218 }, { "epoch": 1.2378845231113718, "grad_norm": 2.722829818725586, "learning_rate": 5.3564e-06, "loss": 0.0236, "step": 23219 }, { "epoch": 1.2379378365410247, "grad_norm": 1.2799047231674194, "learning_rate": 5.3562000000000005e-06, "loss": 0.0168, "step": 23220 }, { "epoch": 1.2379911499706777, "grad_norm": 1.3677374124526978, "learning_rate": 5.356e-06, "loss": 0.0164, "step": 23221 }, { "epoch": 1.2380444634003305, "grad_norm": 0.9198514819145203, "learning_rate": 5.3558e-06, "loss": 0.0113, "step": 23222 }, { "epoch": 1.2380977768299835, "grad_norm": 1.5102245807647705, "learning_rate": 5.355600000000001e-06, "loss": 0.0196, "step": 23223 }, { "epoch": 1.2381510902596364, "grad_norm": 2.901174306869507, "learning_rate": 5.3554000000000004e-06, "loss": 0.0272, "step": 23224 }, { "epoch": 1.2382044036892894, "grad_norm": 2.744720935821533, "learning_rate": 5.3552e-06, "loss": 0.0535, "step": 23225 }, { "epoch": 1.2382577171189422, "grad_norm": 0.9449831247329712, "learning_rate": 5.355e-06, "loss": 0.0086, "step": 23226 }, { "epoch": 1.2383110305485951, "grad_norm": 1.2918033599853516, "learning_rate": 5.3548e-06, "loss": 0.0109, "step": 23227 }, { "epoch": 1.2383643439782481, "grad_norm": 2.2689709663391113, "learning_rate": 5.354600000000001e-06, "loss": 0.0479, "step": 23228 }, { "epoch": 1.238417657407901, "grad_norm": 2.6255245208740234, "learning_rate": 5.354400000000001e-06, "loss": 0.0241, "step": 23229 }, { "epoch": 1.2384709708375539, "grad_norm": 1.0826115608215332, "learning_rate": 5.3542e-06, "loss": 0.0073, "step": 23230 }, { "epoch": 1.2385242842672068, "grad_norm": 2.226245641708374, "learning_rate": 5.354e-06, "loss": 0.0424, "step": 23231 }, { "epoch": 1.2385775976968598, "grad_norm": 1.9805071353912354, "learning_rate": 5.3538e-06, "loss": 0.0227, "step": 23232 }, { "epoch": 1.2386309111265128, "grad_norm": 2.788006544113159, "learning_rate": 5.3536e-06, "loss": 0.0149, "step": 23233 }, { "epoch": 1.2386842245561658, "grad_norm": 2.475203037261963, "learning_rate": 5.353400000000001e-06, "loss": 0.0275, "step": 23234 }, { "epoch": 1.2387375379858185, "grad_norm": 1.706325650215149, "learning_rate": 5.353200000000001e-06, "loss": 0.0109, "step": 23235 }, { "epoch": 1.2387908514154715, "grad_norm": 2.244278907775879, "learning_rate": 5.353e-06, "loss": 0.0185, "step": 23236 }, { "epoch": 1.2388441648451245, "grad_norm": 2.671233892440796, "learning_rate": 5.3528e-06, "loss": 0.023, "step": 23237 }, { "epoch": 1.2388974782747775, "grad_norm": 0.1656428575515747, "learning_rate": 5.3526e-06, "loss": 0.0017, "step": 23238 }, { "epoch": 1.2389507917044305, "grad_norm": 1.330448031425476, "learning_rate": 5.352400000000001e-06, "loss": 0.0541, "step": 23239 }, { "epoch": 1.2390041051340832, "grad_norm": 2.8257715702056885, "learning_rate": 5.352200000000001e-06, "loss": 0.0443, "step": 23240 }, { "epoch": 1.2390574185637362, "grad_norm": 1.6456202268600464, "learning_rate": 5.352000000000001e-06, "loss": 0.0233, "step": 23241 }, { "epoch": 1.2391107319933892, "grad_norm": 1.7198021411895752, "learning_rate": 5.3518e-06, "loss": 0.0227, "step": 23242 }, { "epoch": 1.2391640454230421, "grad_norm": 2.482285499572754, "learning_rate": 5.3516e-06, "loss": 0.0375, "step": 23243 }, { "epoch": 1.239217358852695, "grad_norm": 0.9564348459243774, "learning_rate": 5.3514e-06, "loss": 0.0148, "step": 23244 }, { "epoch": 1.2392706722823479, "grad_norm": 2.2501864433288574, "learning_rate": 5.3512000000000005e-06, "loss": 0.0529, "step": 23245 }, { "epoch": 1.2393239857120009, "grad_norm": 1.8613113164901733, "learning_rate": 5.351000000000001e-06, "loss": 0.0294, "step": 23246 }, { "epoch": 1.2393772991416538, "grad_norm": 1.505892038345337, "learning_rate": 5.3508000000000005e-06, "loss": 0.046, "step": 23247 }, { "epoch": 1.2394306125713066, "grad_norm": 1.8448278903961182, "learning_rate": 5.3506e-06, "loss": 0.022, "step": 23248 }, { "epoch": 1.2394839260009596, "grad_norm": 2.0538382530212402, "learning_rate": 5.3504e-06, "loss": 0.0387, "step": 23249 }, { "epoch": 1.2395372394306126, "grad_norm": 3.7109618186950684, "learning_rate": 5.350200000000001e-06, "loss": 0.0263, "step": 23250 }, { "epoch": 1.2395905528602655, "grad_norm": 2.428222894668579, "learning_rate": 5.3500000000000004e-06, "loss": 0.0303, "step": 23251 }, { "epoch": 1.2396438662899185, "grad_norm": 2.219132423400879, "learning_rate": 5.349800000000001e-06, "loss": 0.0273, "step": 23252 }, { "epoch": 1.2396971797195713, "grad_norm": 1.9126129150390625, "learning_rate": 5.3496e-06, "loss": 0.0132, "step": 23253 }, { "epoch": 1.2397504931492243, "grad_norm": 1.424443006515503, "learning_rate": 5.3494e-06, "loss": 0.0207, "step": 23254 }, { "epoch": 1.2398038065788772, "grad_norm": 0.9082673788070679, "learning_rate": 5.3491999999999995e-06, "loss": 0.008, "step": 23255 }, { "epoch": 1.2398571200085302, "grad_norm": 1.5552295446395874, "learning_rate": 5.349000000000001e-06, "loss": 0.0127, "step": 23256 }, { "epoch": 1.2399104334381832, "grad_norm": 0.9755330681800842, "learning_rate": 5.3488e-06, "loss": 0.0169, "step": 23257 }, { "epoch": 1.239963746867836, "grad_norm": 1.9482918977737427, "learning_rate": 5.348600000000001e-06, "loss": 0.0256, "step": 23258 }, { "epoch": 1.240017060297489, "grad_norm": 0.7264662384986877, "learning_rate": 5.3484e-06, "loss": 0.0143, "step": 23259 }, { "epoch": 1.240070373727142, "grad_norm": 1.7289005517959595, "learning_rate": 5.3482e-06, "loss": 0.0313, "step": 23260 }, { "epoch": 1.2401236871567949, "grad_norm": 0.1675957292318344, "learning_rate": 5.348000000000001e-06, "loss": 0.0019, "step": 23261 }, { "epoch": 1.2401770005864476, "grad_norm": 4.233246326446533, "learning_rate": 5.347800000000001e-06, "loss": 0.0339, "step": 23262 }, { "epoch": 1.2402303140161006, "grad_norm": 2.598207473754883, "learning_rate": 5.3476e-06, "loss": 0.0335, "step": 23263 }, { "epoch": 1.2402836274457536, "grad_norm": 2.0458199977874756, "learning_rate": 5.347400000000001e-06, "loss": 0.0329, "step": 23264 }, { "epoch": 1.2403369408754066, "grad_norm": 0.6424511671066284, "learning_rate": 5.3472e-06, "loss": 0.0128, "step": 23265 }, { "epoch": 1.2403902543050593, "grad_norm": 1.967066764831543, "learning_rate": 5.347e-06, "loss": 0.025, "step": 23266 }, { "epoch": 1.2404435677347123, "grad_norm": 1.586792230606079, "learning_rate": 5.346800000000001e-06, "loss": 0.013, "step": 23267 }, { "epoch": 1.2404968811643653, "grad_norm": 1.0688055753707886, "learning_rate": 5.346600000000001e-06, "loss": 0.0126, "step": 23268 }, { "epoch": 1.2405501945940183, "grad_norm": 2.0337512493133545, "learning_rate": 5.3464e-06, "loss": 0.0155, "step": 23269 }, { "epoch": 1.2406035080236713, "grad_norm": 1.666886329650879, "learning_rate": 5.346200000000001e-06, "loss": 0.0096, "step": 23270 }, { "epoch": 1.240656821453324, "grad_norm": 2.284348964691162, "learning_rate": 5.346e-06, "loss": 0.0333, "step": 23271 }, { "epoch": 1.240710134882977, "grad_norm": 1.2242456674575806, "learning_rate": 5.345800000000001e-06, "loss": 0.0111, "step": 23272 }, { "epoch": 1.24076344831263, "grad_norm": 3.3442351818084717, "learning_rate": 5.345600000000001e-06, "loss": 0.0351, "step": 23273 }, { "epoch": 1.240816761742283, "grad_norm": 3.832444667816162, "learning_rate": 5.3454000000000005e-06, "loss": 0.0445, "step": 23274 }, { "epoch": 1.240870075171936, "grad_norm": 1.3816291093826294, "learning_rate": 5.3452e-06, "loss": 0.0114, "step": 23275 }, { "epoch": 1.2409233886015887, "grad_norm": 1.1633597612380981, "learning_rate": 5.3450000000000005e-06, "loss": 0.0284, "step": 23276 }, { "epoch": 1.2409767020312417, "grad_norm": 0.5941356420516968, "learning_rate": 5.3448e-06, "loss": 0.006, "step": 23277 }, { "epoch": 1.2410300154608946, "grad_norm": 0.29858115315437317, "learning_rate": 5.344600000000001e-06, "loss": 0.0034, "step": 23278 }, { "epoch": 1.2410833288905476, "grad_norm": 1.1822870969772339, "learning_rate": 5.344400000000001e-06, "loss": 0.0069, "step": 23279 }, { "epoch": 1.2411366423202004, "grad_norm": 2.5879316329956055, "learning_rate": 5.3442000000000004e-06, "loss": 0.0217, "step": 23280 }, { "epoch": 1.2411899557498534, "grad_norm": 1.0013024806976318, "learning_rate": 5.344e-06, "loss": 0.0219, "step": 23281 }, { "epoch": 1.2412432691795063, "grad_norm": 1.3397787809371948, "learning_rate": 5.3438e-06, "loss": 0.0208, "step": 23282 }, { "epoch": 1.2412965826091593, "grad_norm": 3.0113515853881836, "learning_rate": 5.3436e-06, "loss": 0.0261, "step": 23283 }, { "epoch": 1.241349896038812, "grad_norm": 3.5424861907958984, "learning_rate": 5.343400000000001e-06, "loss": 0.0621, "step": 23284 }, { "epoch": 1.241403209468465, "grad_norm": 0.5262135863304138, "learning_rate": 5.343200000000001e-06, "loss": 0.0045, "step": 23285 }, { "epoch": 1.241456522898118, "grad_norm": 0.7173703908920288, "learning_rate": 5.343e-06, "loss": 0.0053, "step": 23286 }, { "epoch": 1.241509836327771, "grad_norm": 1.6612634658813477, "learning_rate": 5.3428e-06, "loss": 0.0224, "step": 23287 }, { "epoch": 1.241563149757424, "grad_norm": 1.8008954524993896, "learning_rate": 5.3426e-06, "loss": 0.015, "step": 23288 }, { "epoch": 1.2416164631870767, "grad_norm": 3.6236047744750977, "learning_rate": 5.342400000000001e-06, "loss": 0.0287, "step": 23289 }, { "epoch": 1.2416697766167297, "grad_norm": 2.156920909881592, "learning_rate": 5.342200000000001e-06, "loss": 0.0116, "step": 23290 }, { "epoch": 1.2417230900463827, "grad_norm": 3.0931754112243652, "learning_rate": 5.342000000000001e-06, "loss": 0.0837, "step": 23291 }, { "epoch": 1.2417764034760357, "grad_norm": 1.9025465250015259, "learning_rate": 5.3418e-06, "loss": 0.0357, "step": 23292 }, { "epoch": 1.2418297169056887, "grad_norm": 2.7434778213500977, "learning_rate": 5.3416e-06, "loss": 0.0183, "step": 23293 }, { "epoch": 1.2418830303353414, "grad_norm": 2.012971878051758, "learning_rate": 5.3414e-06, "loss": 0.0132, "step": 23294 }, { "epoch": 1.2419363437649944, "grad_norm": 1.9017908573150635, "learning_rate": 5.341200000000001e-06, "loss": 0.029, "step": 23295 }, { "epoch": 1.2419896571946474, "grad_norm": 1.112325668334961, "learning_rate": 5.341000000000001e-06, "loss": 0.0072, "step": 23296 }, { "epoch": 1.2420429706243004, "grad_norm": 2.0698318481445312, "learning_rate": 5.340800000000001e-06, "loss": 0.0199, "step": 23297 }, { "epoch": 1.2420962840539531, "grad_norm": 5.3180742263793945, "learning_rate": 5.3406e-06, "loss": 0.0236, "step": 23298 }, { "epoch": 1.242149597483606, "grad_norm": 1.5492056608200073, "learning_rate": 5.3404e-06, "loss": 0.0186, "step": 23299 }, { "epoch": 1.242202910913259, "grad_norm": 0.7288295030593872, "learning_rate": 5.340200000000001e-06, "loss": 0.0058, "step": 23300 }, { "epoch": 1.242256224342912, "grad_norm": 3.372119903564453, "learning_rate": 5.3400000000000005e-06, "loss": 0.0395, "step": 23301 }, { "epoch": 1.2423095377725648, "grad_norm": 1.1921738386154175, "learning_rate": 5.339800000000001e-06, "loss": 0.0079, "step": 23302 }, { "epoch": 1.2423628512022178, "grad_norm": 1.4335730075836182, "learning_rate": 5.3396000000000005e-06, "loss": 0.0367, "step": 23303 }, { "epoch": 1.2424161646318708, "grad_norm": 1.9877164363861084, "learning_rate": 5.3394e-06, "loss": 0.028, "step": 23304 }, { "epoch": 1.2424694780615237, "grad_norm": 1.6704641580581665, "learning_rate": 5.3392e-06, "loss": 0.0172, "step": 23305 }, { "epoch": 1.2425227914911767, "grad_norm": 2.0577428340911865, "learning_rate": 5.339000000000001e-06, "loss": 0.0155, "step": 23306 }, { "epoch": 1.2425761049208295, "grad_norm": 2.3278074264526367, "learning_rate": 5.3388000000000005e-06, "loss": 0.0129, "step": 23307 }, { "epoch": 1.2426294183504825, "grad_norm": 1.8433643579483032, "learning_rate": 5.338600000000001e-06, "loss": 0.0317, "step": 23308 }, { "epoch": 1.2426827317801354, "grad_norm": 2.3727245330810547, "learning_rate": 5.3384e-06, "loss": 0.0269, "step": 23309 }, { "epoch": 1.2427360452097884, "grad_norm": 0.7581536769866943, "learning_rate": 5.3382e-06, "loss": 0.0131, "step": 23310 }, { "epoch": 1.2427893586394414, "grad_norm": 2.8072640895843506, "learning_rate": 5.338000000000001e-06, "loss": 0.0422, "step": 23311 }, { "epoch": 1.2428426720690942, "grad_norm": 4.040125370025635, "learning_rate": 5.337800000000001e-06, "loss": 0.0108, "step": 23312 }, { "epoch": 1.2428959854987471, "grad_norm": 1.99699068069458, "learning_rate": 5.3376e-06, "loss": 0.0259, "step": 23313 }, { "epoch": 1.2429492989284001, "grad_norm": 1.398632526397705, "learning_rate": 5.337400000000001e-06, "loss": 0.0099, "step": 23314 }, { "epoch": 1.2430026123580529, "grad_norm": 2.852071762084961, "learning_rate": 5.3372e-06, "loss": 0.0511, "step": 23315 }, { "epoch": 1.2430559257877059, "grad_norm": 2.7378382682800293, "learning_rate": 5.337e-06, "loss": 0.0303, "step": 23316 }, { "epoch": 1.2431092392173588, "grad_norm": 2.361605644226074, "learning_rate": 5.336800000000001e-06, "loss": 0.024, "step": 23317 }, { "epoch": 1.2431625526470118, "grad_norm": 5.238580703735352, "learning_rate": 5.336600000000001e-06, "loss": 0.0306, "step": 23318 }, { "epoch": 1.2432158660766648, "grad_norm": 1.3419218063354492, "learning_rate": 5.3364e-06, "loss": 0.0121, "step": 23319 }, { "epoch": 1.2432691795063175, "grad_norm": 2.273089647293091, "learning_rate": 5.336200000000001e-06, "loss": 0.0257, "step": 23320 }, { "epoch": 1.2433224929359705, "grad_norm": 1.6393446922302246, "learning_rate": 5.336e-06, "loss": 0.0143, "step": 23321 }, { "epoch": 1.2433758063656235, "grad_norm": 1.361069917678833, "learning_rate": 5.335800000000001e-06, "loss": 0.0255, "step": 23322 }, { "epoch": 1.2434291197952765, "grad_norm": 3.8942360877990723, "learning_rate": 5.335600000000001e-06, "loss": 0.0406, "step": 23323 }, { "epoch": 1.2434824332249295, "grad_norm": 0.9170898199081421, "learning_rate": 5.335400000000001e-06, "loss": 0.0116, "step": 23324 }, { "epoch": 1.2435357466545822, "grad_norm": 1.7573186159133911, "learning_rate": 5.3352e-06, "loss": 0.029, "step": 23325 }, { "epoch": 1.2435890600842352, "grad_norm": 1.7795978784561157, "learning_rate": 5.335000000000001e-06, "loss": 0.0234, "step": 23326 }, { "epoch": 1.2436423735138882, "grad_norm": 0.7459715604782104, "learning_rate": 5.3348e-06, "loss": 0.006, "step": 23327 }, { "epoch": 1.2436956869435412, "grad_norm": 2.1671676635742188, "learning_rate": 5.3346000000000006e-06, "loss": 0.0228, "step": 23328 }, { "epoch": 1.243749000373194, "grad_norm": 1.358033299446106, "learning_rate": 5.334400000000001e-06, "loss": 0.013, "step": 23329 }, { "epoch": 1.243802313802847, "grad_norm": 1.8627527952194214, "learning_rate": 5.3342000000000005e-06, "loss": 0.0298, "step": 23330 }, { "epoch": 1.2438556272324999, "grad_norm": 2.2305426597595215, "learning_rate": 5.334e-06, "loss": 0.0195, "step": 23331 }, { "epoch": 1.2439089406621529, "grad_norm": 1.1822967529296875, "learning_rate": 5.3338000000000005e-06, "loss": 0.0096, "step": 23332 }, { "epoch": 1.2439622540918056, "grad_norm": 1.4713952541351318, "learning_rate": 5.3336e-06, "loss": 0.0197, "step": 23333 }, { "epoch": 1.2440155675214586, "grad_norm": 0.8371543884277344, "learning_rate": 5.3334000000000005e-06, "loss": 0.0253, "step": 23334 }, { "epoch": 1.2440688809511116, "grad_norm": 2.0575954914093018, "learning_rate": 5.333200000000001e-06, "loss": 0.0259, "step": 23335 }, { "epoch": 1.2441221943807645, "grad_norm": 0.3429349660873413, "learning_rate": 5.3330000000000004e-06, "loss": 0.0031, "step": 23336 }, { "epoch": 1.2441755078104175, "grad_norm": 1.2355372905731201, "learning_rate": 5.3328e-06, "loss": 0.0148, "step": 23337 }, { "epoch": 1.2442288212400703, "grad_norm": 2.1123483180999756, "learning_rate": 5.3326e-06, "loss": 0.0295, "step": 23338 }, { "epoch": 1.2442821346697233, "grad_norm": 1.1552391052246094, "learning_rate": 5.332400000000001e-06, "loss": 0.0133, "step": 23339 }, { "epoch": 1.2443354480993762, "grad_norm": 1.4529507160186768, "learning_rate": 5.3322e-06, "loss": 0.0084, "step": 23340 }, { "epoch": 1.2443887615290292, "grad_norm": 2.6203482151031494, "learning_rate": 5.332000000000001e-06, "loss": 0.0523, "step": 23341 }, { "epoch": 1.2444420749586822, "grad_norm": 2.5405375957489014, "learning_rate": 5.3318e-06, "loss": 0.0538, "step": 23342 }, { "epoch": 1.244495388388335, "grad_norm": 1.3329190015792847, "learning_rate": 5.3316e-06, "loss": 0.0226, "step": 23343 }, { "epoch": 1.244548701817988, "grad_norm": 2.025418281555176, "learning_rate": 5.3314e-06, "loss": 0.0391, "step": 23344 }, { "epoch": 1.244602015247641, "grad_norm": 2.139219045639038, "learning_rate": 5.331200000000001e-06, "loss": 0.0099, "step": 23345 }, { "epoch": 1.244655328677294, "grad_norm": 1.2253246307373047, "learning_rate": 5.331e-06, "loss": 0.0135, "step": 23346 }, { "epoch": 1.2447086421069466, "grad_norm": 1.4323606491088867, "learning_rate": 5.330800000000001e-06, "loss": 0.0198, "step": 23347 }, { "epoch": 1.2447619555365996, "grad_norm": 1.507800579071045, "learning_rate": 5.3306e-06, "loss": 0.0118, "step": 23348 }, { "epoch": 1.2448152689662526, "grad_norm": 0.7248032689094543, "learning_rate": 5.3304e-06, "loss": 0.0062, "step": 23349 }, { "epoch": 1.2448685823959056, "grad_norm": 1.0817346572875977, "learning_rate": 5.330200000000001e-06, "loss": 0.004, "step": 23350 }, { "epoch": 1.2449218958255583, "grad_norm": 3.19478702545166, "learning_rate": 5.330000000000001e-06, "loss": 0.045, "step": 23351 }, { "epoch": 1.2449752092552113, "grad_norm": 3.7699124813079834, "learning_rate": 5.3298e-06, "loss": 0.0269, "step": 23352 }, { "epoch": 1.2450285226848643, "grad_norm": 2.3686468601226807, "learning_rate": 5.329600000000001e-06, "loss": 0.0421, "step": 23353 }, { "epoch": 1.2450818361145173, "grad_norm": 3.0201175212860107, "learning_rate": 5.3294e-06, "loss": 0.0459, "step": 23354 }, { "epoch": 1.2451351495441703, "grad_norm": 2.5110528469085693, "learning_rate": 5.3292e-06, "loss": 0.0197, "step": 23355 }, { "epoch": 1.245188462973823, "grad_norm": 2.7269256114959717, "learning_rate": 5.329000000000001e-06, "loss": 0.0285, "step": 23356 }, { "epoch": 1.245241776403476, "grad_norm": 1.894864797592163, "learning_rate": 5.3288000000000005e-06, "loss": 0.042, "step": 23357 }, { "epoch": 1.245295089833129, "grad_norm": 3.5923049449920654, "learning_rate": 5.3286e-06, "loss": 0.0176, "step": 23358 }, { "epoch": 1.245348403262782, "grad_norm": 1.7297512292861938, "learning_rate": 5.3284000000000005e-06, "loss": 0.0201, "step": 23359 }, { "epoch": 1.245401716692435, "grad_norm": 1.415766954421997, "learning_rate": 5.3282e-06, "loss": 0.01, "step": 23360 }, { "epoch": 1.2454550301220877, "grad_norm": 2.498955488204956, "learning_rate": 5.328000000000001e-06, "loss": 0.0257, "step": 23361 }, { "epoch": 1.2455083435517407, "grad_norm": 1.8754210472106934, "learning_rate": 5.327800000000001e-06, "loss": 0.0487, "step": 23362 }, { "epoch": 1.2455616569813937, "grad_norm": 1.1082525253295898, "learning_rate": 5.3276000000000005e-06, "loss": 0.0172, "step": 23363 }, { "epoch": 1.2456149704110466, "grad_norm": 1.3464878797531128, "learning_rate": 5.3274e-06, "loss": 0.0227, "step": 23364 }, { "epoch": 1.2456682838406994, "grad_norm": 1.1230217218399048, "learning_rate": 5.3272000000000004e-06, "loss": 0.0154, "step": 23365 }, { "epoch": 1.2457215972703524, "grad_norm": 1.5944063663482666, "learning_rate": 5.327e-06, "loss": 0.0085, "step": 23366 }, { "epoch": 1.2457749107000053, "grad_norm": 2.2641282081604004, "learning_rate": 5.326800000000001e-06, "loss": 0.0374, "step": 23367 }, { "epoch": 1.2458282241296583, "grad_norm": 0.8423753380775452, "learning_rate": 5.326600000000001e-06, "loss": 0.0069, "step": 23368 }, { "epoch": 1.245881537559311, "grad_norm": 2.100452184677124, "learning_rate": 5.3264e-06, "loss": 0.0196, "step": 23369 }, { "epoch": 1.245934850988964, "grad_norm": 1.1136456727981567, "learning_rate": 5.3262e-06, "loss": 0.0064, "step": 23370 }, { "epoch": 1.245988164418617, "grad_norm": 2.2383298873901367, "learning_rate": 5.326e-06, "loss": 0.0101, "step": 23371 }, { "epoch": 1.24604147784827, "grad_norm": 1.364962100982666, "learning_rate": 5.325800000000001e-06, "loss": 0.0275, "step": 23372 }, { "epoch": 1.246094791277923, "grad_norm": 1.425439715385437, "learning_rate": 5.325600000000001e-06, "loss": 0.012, "step": 23373 }, { "epoch": 1.2461481047075758, "grad_norm": 2.0707268714904785, "learning_rate": 5.325400000000001e-06, "loss": 0.02, "step": 23374 }, { "epoch": 1.2462014181372287, "grad_norm": 0.7942960262298584, "learning_rate": 5.3252e-06, "loss": 0.0204, "step": 23375 }, { "epoch": 1.2462547315668817, "grad_norm": 1.0559779405593872, "learning_rate": 5.325e-06, "loss": 0.0269, "step": 23376 }, { "epoch": 1.2463080449965347, "grad_norm": 1.9020389318466187, "learning_rate": 5.3248e-06, "loss": 0.0167, "step": 23377 }, { "epoch": 1.2463613584261877, "grad_norm": 1.428027629852295, "learning_rate": 5.324600000000001e-06, "loss": 0.0201, "step": 23378 }, { "epoch": 1.2464146718558404, "grad_norm": 1.8472224473953247, "learning_rate": 5.324400000000001e-06, "loss": 0.0212, "step": 23379 }, { "epoch": 1.2464679852854934, "grad_norm": 1.6081639528274536, "learning_rate": 5.324200000000001e-06, "loss": 0.0248, "step": 23380 }, { "epoch": 1.2465212987151464, "grad_norm": 1.6877679824829102, "learning_rate": 5.324e-06, "loss": 0.0206, "step": 23381 }, { "epoch": 1.2465746121447994, "grad_norm": 1.1163197755813599, "learning_rate": 5.3238e-06, "loss": 0.0086, "step": 23382 }, { "epoch": 1.2466279255744521, "grad_norm": 3.918351888656616, "learning_rate": 5.323600000000001e-06, "loss": 0.0397, "step": 23383 }, { "epoch": 1.246681239004105, "grad_norm": 1.4131007194519043, "learning_rate": 5.3234000000000006e-06, "loss": 0.0131, "step": 23384 }, { "epoch": 1.246734552433758, "grad_norm": 2.3810184001922607, "learning_rate": 5.323200000000001e-06, "loss": 0.0394, "step": 23385 }, { "epoch": 1.246787865863411, "grad_norm": 2.145218849182129, "learning_rate": 5.3230000000000005e-06, "loss": 0.0301, "step": 23386 }, { "epoch": 1.2468411792930638, "grad_norm": 1.7368980646133423, "learning_rate": 5.3228e-06, "loss": 0.0176, "step": 23387 }, { "epoch": 1.2468944927227168, "grad_norm": 1.219578504562378, "learning_rate": 5.3226e-06, "loss": 0.0141, "step": 23388 }, { "epoch": 1.2469478061523698, "grad_norm": 2.4585087299346924, "learning_rate": 5.322400000000001e-06, "loss": 0.0312, "step": 23389 }, { "epoch": 1.2470011195820228, "grad_norm": 0.7936738133430481, "learning_rate": 5.3222000000000005e-06, "loss": 0.0086, "step": 23390 }, { "epoch": 1.2470544330116757, "grad_norm": 1.081229567527771, "learning_rate": 5.322000000000001e-06, "loss": 0.0179, "step": 23391 }, { "epoch": 1.2471077464413285, "grad_norm": 2.314582109451294, "learning_rate": 5.3218000000000004e-06, "loss": 0.0226, "step": 23392 }, { "epoch": 1.2471610598709815, "grad_norm": 1.1519588232040405, "learning_rate": 5.3216e-06, "loss": 0.0093, "step": 23393 }, { "epoch": 1.2472143733006344, "grad_norm": 2.3138880729675293, "learning_rate": 5.3213999999999996e-06, "loss": 0.0435, "step": 23394 }, { "epoch": 1.2472676867302874, "grad_norm": 1.036463975906372, "learning_rate": 5.321200000000001e-06, "loss": 0.0064, "step": 23395 }, { "epoch": 1.2473210001599404, "grad_norm": 1.7377854585647583, "learning_rate": 5.321e-06, "loss": 0.0108, "step": 23396 }, { "epoch": 1.2473743135895932, "grad_norm": 2.8193132877349854, "learning_rate": 5.320800000000001e-06, "loss": 0.0196, "step": 23397 }, { "epoch": 1.2474276270192461, "grad_norm": 0.6287119388580322, "learning_rate": 5.3206e-06, "loss": 0.0074, "step": 23398 }, { "epoch": 1.2474809404488991, "grad_norm": 2.9068403244018555, "learning_rate": 5.3204e-06, "loss": 0.0261, "step": 23399 }, { "epoch": 1.247534253878552, "grad_norm": 2.3131699562072754, "learning_rate": 5.320200000000001e-06, "loss": 0.0305, "step": 23400 }, { "epoch": 1.2475875673082049, "grad_norm": 2.3789753913879395, "learning_rate": 5.320000000000001e-06, "loss": 0.025, "step": 23401 }, { "epoch": 1.2476408807378578, "grad_norm": 1.863020420074463, "learning_rate": 5.3198e-06, "loss": 0.017, "step": 23402 }, { "epoch": 1.2476941941675108, "grad_norm": 1.6518772840499878, "learning_rate": 5.319600000000001e-06, "loss": 0.0149, "step": 23403 }, { "epoch": 1.2477475075971638, "grad_norm": 3.371814489364624, "learning_rate": 5.3194e-06, "loss": 0.0451, "step": 23404 }, { "epoch": 1.2478008210268166, "grad_norm": 1.9046235084533691, "learning_rate": 5.3192e-06, "loss": 0.0227, "step": 23405 }, { "epoch": 1.2478541344564695, "grad_norm": 1.4885706901550293, "learning_rate": 5.319000000000001e-06, "loss": 0.022, "step": 23406 }, { "epoch": 1.2479074478861225, "grad_norm": 2.534834623336792, "learning_rate": 5.318800000000001e-06, "loss": 0.0237, "step": 23407 }, { "epoch": 1.2479607613157755, "grad_norm": 1.824991226196289, "learning_rate": 5.3186e-06, "loss": 0.0163, "step": 23408 }, { "epoch": 1.2480140747454285, "grad_norm": 2.2606701850891113, "learning_rate": 5.318400000000001e-06, "loss": 0.0209, "step": 23409 }, { "epoch": 1.2480673881750812, "grad_norm": 4.493860721588135, "learning_rate": 5.3182e-06, "loss": 0.0301, "step": 23410 }, { "epoch": 1.2481207016047342, "grad_norm": 1.8617419004440308, "learning_rate": 5.318000000000001e-06, "loss": 0.0203, "step": 23411 }, { "epoch": 1.2481740150343872, "grad_norm": 3.2213551998138428, "learning_rate": 5.317800000000001e-06, "loss": 0.0423, "step": 23412 }, { "epoch": 1.2482273284640402, "grad_norm": 6.215190410614014, "learning_rate": 5.3176000000000006e-06, "loss": 0.0321, "step": 23413 }, { "epoch": 1.2482806418936931, "grad_norm": 1.892224907875061, "learning_rate": 5.3174e-06, "loss": 0.0133, "step": 23414 }, { "epoch": 1.248333955323346, "grad_norm": 1.5601136684417725, "learning_rate": 5.3172000000000005e-06, "loss": 0.0149, "step": 23415 }, { "epoch": 1.2483872687529989, "grad_norm": 1.9348682165145874, "learning_rate": 5.317e-06, "loss": 0.0189, "step": 23416 }, { "epoch": 1.2484405821826519, "grad_norm": 2.728760004043579, "learning_rate": 5.3168000000000005e-06, "loss": 0.0158, "step": 23417 }, { "epoch": 1.2484938956123046, "grad_norm": 2.2990469932556152, "learning_rate": 5.316600000000001e-06, "loss": 0.0178, "step": 23418 }, { "epoch": 1.2485472090419576, "grad_norm": 1.9506080150604248, "learning_rate": 5.3164000000000005e-06, "loss": 0.016, "step": 23419 }, { "epoch": 1.2486005224716106, "grad_norm": 2.3245694637298584, "learning_rate": 5.3162e-06, "loss": 0.0568, "step": 23420 }, { "epoch": 1.2486538359012636, "grad_norm": 1.3617451190948486, "learning_rate": 5.3160000000000004e-06, "loss": 0.0218, "step": 23421 }, { "epoch": 1.2487071493309165, "grad_norm": 0.38959604501724243, "learning_rate": 5.315800000000001e-06, "loss": 0.0023, "step": 23422 }, { "epoch": 1.2487604627605693, "grad_norm": 3.1564900875091553, "learning_rate": 5.3156e-06, "loss": 0.0629, "step": 23423 }, { "epoch": 1.2488137761902223, "grad_norm": 0.7439244985580444, "learning_rate": 5.315400000000001e-06, "loss": 0.0063, "step": 23424 }, { "epoch": 1.2488670896198752, "grad_norm": 2.292262315750122, "learning_rate": 5.3152e-06, "loss": 0.0232, "step": 23425 }, { "epoch": 1.2489204030495282, "grad_norm": 2.699937343597412, "learning_rate": 5.315e-06, "loss": 0.018, "step": 23426 }, { "epoch": 1.2489737164791812, "grad_norm": 1.5838401317596436, "learning_rate": 5.3148e-06, "loss": 0.0309, "step": 23427 }, { "epoch": 1.249027029908834, "grad_norm": 1.1976042985916138, "learning_rate": 5.314600000000001e-06, "loss": 0.03, "step": 23428 }, { "epoch": 1.249080343338487, "grad_norm": 1.8948010206222534, "learning_rate": 5.3144e-06, "loss": 0.0286, "step": 23429 }, { "epoch": 1.24913365676814, "grad_norm": 2.7590441703796387, "learning_rate": 5.314200000000001e-06, "loss": 0.0295, "step": 23430 }, { "epoch": 1.249186970197793, "grad_norm": 1.0221325159072876, "learning_rate": 5.314e-06, "loss": 0.0259, "step": 23431 }, { "epoch": 1.2492402836274457, "grad_norm": 1.679015040397644, "learning_rate": 5.3138e-06, "loss": 0.0239, "step": 23432 }, { "epoch": 1.2492935970570986, "grad_norm": 1.2270528078079224, "learning_rate": 5.313600000000001e-06, "loss": 0.0232, "step": 23433 }, { "epoch": 1.2493469104867516, "grad_norm": 2.038914442062378, "learning_rate": 5.313400000000001e-06, "loss": 0.014, "step": 23434 }, { "epoch": 1.2494002239164046, "grad_norm": 0.9191136956214905, "learning_rate": 5.3132e-06, "loss": 0.0117, "step": 23435 }, { "epoch": 1.2494535373460574, "grad_norm": 1.240010142326355, "learning_rate": 5.313000000000001e-06, "loss": 0.0124, "step": 23436 }, { "epoch": 1.2495068507757103, "grad_norm": 0.7177718877792358, "learning_rate": 5.3128e-06, "loss": 0.0072, "step": 23437 }, { "epoch": 1.2495601642053633, "grad_norm": 3.113691568374634, "learning_rate": 5.3126e-06, "loss": 0.0272, "step": 23438 }, { "epoch": 1.2496134776350163, "grad_norm": 1.3585729598999023, "learning_rate": 5.312400000000001e-06, "loss": 0.0151, "step": 23439 }, { "epoch": 1.2496667910646693, "grad_norm": 2.983200788497925, "learning_rate": 5.3122000000000006e-06, "loss": 0.0338, "step": 23440 }, { "epoch": 1.249720104494322, "grad_norm": 2.155726194381714, "learning_rate": 5.312e-06, "loss": 0.0357, "step": 23441 }, { "epoch": 1.249773417923975, "grad_norm": 1.880661129951477, "learning_rate": 5.3118000000000005e-06, "loss": 0.0189, "step": 23442 }, { "epoch": 1.249826731353628, "grad_norm": 1.57854163646698, "learning_rate": 5.3116e-06, "loss": 0.0285, "step": 23443 }, { "epoch": 1.249880044783281, "grad_norm": 1.753507137298584, "learning_rate": 5.3114e-06, "loss": 0.0193, "step": 23444 }, { "epoch": 1.249933358212934, "grad_norm": 2.673887252807617, "learning_rate": 5.311200000000001e-06, "loss": 0.0287, "step": 23445 }, { "epoch": 1.2499866716425867, "grad_norm": 2.328890800476074, "learning_rate": 5.3110000000000005e-06, "loss": 0.0343, "step": 23446 }, { "epoch": 1.2500399850722397, "grad_norm": 2.047745943069458, "learning_rate": 5.3108e-06, "loss": 0.032, "step": 23447 }, { "epoch": 1.2500932985018927, "grad_norm": 0.6133569478988647, "learning_rate": 5.3106000000000005e-06, "loss": 0.0068, "step": 23448 }, { "epoch": 1.2501466119315456, "grad_norm": 2.179654121398926, "learning_rate": 5.3104e-06, "loss": 0.0479, "step": 23449 }, { "epoch": 1.2501999253611986, "grad_norm": 3.1526777744293213, "learning_rate": 5.310200000000001e-06, "loss": 0.0329, "step": 23450 }, { "epoch": 1.2502532387908514, "grad_norm": 1.3435181379318237, "learning_rate": 5.310000000000001e-06, "loss": 0.0176, "step": 23451 }, { "epoch": 1.2503065522205044, "grad_norm": 5.407138347625732, "learning_rate": 5.3098e-06, "loss": 0.0582, "step": 23452 }, { "epoch": 1.2503598656501573, "grad_norm": 1.764404296875, "learning_rate": 5.3096e-06, "loss": 0.0416, "step": 23453 }, { "epoch": 1.25041317907981, "grad_norm": 2.3964545726776123, "learning_rate": 5.3094e-06, "loss": 0.041, "step": 23454 }, { "epoch": 1.250466492509463, "grad_norm": 4.3567280769348145, "learning_rate": 5.3092e-06, "loss": 0.0375, "step": 23455 }, { "epoch": 1.250519805939116, "grad_norm": 1.6351889371871948, "learning_rate": 5.309000000000001e-06, "loss": 0.028, "step": 23456 }, { "epoch": 1.250573119368769, "grad_norm": 2.3224239349365234, "learning_rate": 5.308800000000001e-06, "loss": 0.0251, "step": 23457 }, { "epoch": 1.250626432798422, "grad_norm": 3.476071357727051, "learning_rate": 5.3086e-06, "loss": 0.0229, "step": 23458 }, { "epoch": 1.2506797462280748, "grad_norm": 1.2378673553466797, "learning_rate": 5.3084e-06, "loss": 0.011, "step": 23459 }, { "epoch": 1.2507330596577277, "grad_norm": 1.2059842348098755, "learning_rate": 5.3082e-06, "loss": 0.01, "step": 23460 }, { "epoch": 1.2507863730873807, "grad_norm": 1.7936960458755493, "learning_rate": 5.308000000000001e-06, "loss": 0.0137, "step": 23461 }, { "epoch": 1.2508396865170337, "grad_norm": 2.7040603160858154, "learning_rate": 5.307800000000001e-06, "loss": 0.0463, "step": 23462 }, { "epoch": 1.2508929999466867, "grad_norm": 4.232309818267822, "learning_rate": 5.307600000000001e-06, "loss": 0.0201, "step": 23463 }, { "epoch": 1.2509463133763394, "grad_norm": 2.204979181289673, "learning_rate": 5.3074e-06, "loss": 0.014, "step": 23464 }, { "epoch": 1.2509996268059924, "grad_norm": 1.1430659294128418, "learning_rate": 5.3072e-06, "loss": 0.0209, "step": 23465 }, { "epoch": 1.2510529402356454, "grad_norm": 1.4306256771087646, "learning_rate": 5.307e-06, "loss": 0.0251, "step": 23466 }, { "epoch": 1.2511062536652982, "grad_norm": 1.4854408502578735, "learning_rate": 5.306800000000001e-06, "loss": 0.0119, "step": 23467 }, { "epoch": 1.2511595670949514, "grad_norm": 1.508421540260315, "learning_rate": 5.306600000000001e-06, "loss": 0.0137, "step": 23468 }, { "epoch": 1.251212880524604, "grad_norm": 1.7889810800552368, "learning_rate": 5.3064000000000006e-06, "loss": 0.0249, "step": 23469 }, { "epoch": 1.251266193954257, "grad_norm": 1.1121981143951416, "learning_rate": 5.3062e-06, "loss": 0.016, "step": 23470 }, { "epoch": 1.25131950738391, "grad_norm": 1.839442491531372, "learning_rate": 5.306e-06, "loss": 0.0273, "step": 23471 }, { "epoch": 1.2513728208135628, "grad_norm": 0.9967103600502014, "learning_rate": 5.305800000000001e-06, "loss": 0.0228, "step": 23472 }, { "epoch": 1.2514261342432158, "grad_norm": 1.5780094861984253, "learning_rate": 5.3056000000000005e-06, "loss": 0.0264, "step": 23473 }, { "epoch": 1.2514794476728688, "grad_norm": 1.9129726886749268, "learning_rate": 5.305400000000001e-06, "loss": 0.0248, "step": 23474 }, { "epoch": 1.2515327611025218, "grad_norm": 1.0209382772445679, "learning_rate": 5.3052000000000005e-06, "loss": 0.0254, "step": 23475 }, { "epoch": 1.2515860745321747, "grad_norm": 1.974562168121338, "learning_rate": 5.305e-06, "loss": 0.0296, "step": 23476 }, { "epoch": 1.2516393879618275, "grad_norm": 1.3148798942565918, "learning_rate": 5.3048e-06, "loss": 0.0114, "step": 23477 }, { "epoch": 1.2516927013914805, "grad_norm": 1.932450771331787, "learning_rate": 5.304600000000001e-06, "loss": 0.0231, "step": 23478 }, { "epoch": 1.2517460148211335, "grad_norm": 2.5191776752471924, "learning_rate": 5.3044e-06, "loss": 0.0232, "step": 23479 }, { "epoch": 1.2517993282507864, "grad_norm": 2.7663683891296387, "learning_rate": 5.304200000000001e-06, "loss": 0.0397, "step": 23480 }, { "epoch": 1.2518526416804394, "grad_norm": 2.645724296569824, "learning_rate": 5.304e-06, "loss": 0.0438, "step": 23481 }, { "epoch": 1.2519059551100922, "grad_norm": 3.1001548767089844, "learning_rate": 5.3038e-06, "loss": 0.0211, "step": 23482 }, { "epoch": 1.2519592685397452, "grad_norm": 2.303328275680542, "learning_rate": 5.303600000000001e-06, "loss": 0.0298, "step": 23483 }, { "epoch": 1.2520125819693981, "grad_norm": 1.9985297918319702, "learning_rate": 5.303400000000001e-06, "loss": 0.0239, "step": 23484 }, { "epoch": 1.2520658953990509, "grad_norm": 1.047291874885559, "learning_rate": 5.3032e-06, "loss": 0.0153, "step": 23485 }, { "epoch": 1.252119208828704, "grad_norm": 0.939665675163269, "learning_rate": 5.303000000000001e-06, "loss": 0.0212, "step": 23486 }, { "epoch": 1.2521725222583568, "grad_norm": 1.116877555847168, "learning_rate": 5.3028e-06, "loss": 0.008, "step": 23487 }, { "epoch": 1.2522258356880098, "grad_norm": 3.2806310653686523, "learning_rate": 5.3026e-06, "loss": 0.0522, "step": 23488 }, { "epoch": 1.2522791491176628, "grad_norm": 2.687087297439575, "learning_rate": 5.302400000000001e-06, "loss": 0.0295, "step": 23489 }, { "epoch": 1.2523324625473156, "grad_norm": 2.4849164485931396, "learning_rate": 5.302200000000001e-06, "loss": 0.016, "step": 23490 }, { "epoch": 1.2523857759769685, "grad_norm": 1.4289487600326538, "learning_rate": 5.302e-06, "loss": 0.0409, "step": 23491 }, { "epoch": 1.2524390894066215, "grad_norm": 0.7519370913505554, "learning_rate": 5.301800000000001e-06, "loss": 0.0086, "step": 23492 }, { "epoch": 1.2524924028362745, "grad_norm": 0.9309687614440918, "learning_rate": 5.3016e-06, "loss": 0.0087, "step": 23493 }, { "epoch": 1.2525457162659275, "grad_norm": 1.1890721321105957, "learning_rate": 5.3014e-06, "loss": 0.0242, "step": 23494 }, { "epoch": 1.2525990296955802, "grad_norm": 1.557367205619812, "learning_rate": 5.301200000000001e-06, "loss": 0.0117, "step": 23495 }, { "epoch": 1.2526523431252332, "grad_norm": 1.3687561750411987, "learning_rate": 5.3010000000000006e-06, "loss": 0.0306, "step": 23496 }, { "epoch": 1.2527056565548862, "grad_norm": 2.6263227462768555, "learning_rate": 5.3008e-06, "loss": 0.029, "step": 23497 }, { "epoch": 1.2527589699845392, "grad_norm": 1.4141052961349487, "learning_rate": 5.3006000000000005e-06, "loss": 0.0188, "step": 23498 }, { "epoch": 1.2528122834141922, "grad_norm": 1.8039276599884033, "learning_rate": 5.3004e-06, "loss": 0.0375, "step": 23499 }, { "epoch": 1.252865596843845, "grad_norm": 1.6138492822647095, "learning_rate": 5.3002000000000005e-06, "loss": 0.008, "step": 23500 }, { "epoch": 1.2529189102734979, "grad_norm": 1.099663257598877, "learning_rate": 5.300000000000001e-06, "loss": 0.0105, "step": 23501 }, { "epoch": 1.2529722237031509, "grad_norm": 1.2232892513275146, "learning_rate": 5.2998000000000005e-06, "loss": 0.0208, "step": 23502 }, { "epoch": 1.2530255371328036, "grad_norm": 1.4368233680725098, "learning_rate": 5.2996e-06, "loss": 0.0334, "step": 23503 }, { "epoch": 1.2530788505624566, "grad_norm": 1.6226427555084229, "learning_rate": 5.2994000000000005e-06, "loss": 0.0251, "step": 23504 }, { "epoch": 1.2531321639921096, "grad_norm": 4.942570686340332, "learning_rate": 5.2992e-06, "loss": 0.0793, "step": 23505 }, { "epoch": 1.2531854774217626, "grad_norm": 1.7025485038757324, "learning_rate": 5.2990000000000004e-06, "loss": 0.0327, "step": 23506 }, { "epoch": 1.2532387908514155, "grad_norm": 1.350259780883789, "learning_rate": 5.298800000000001e-06, "loss": 0.0219, "step": 23507 }, { "epoch": 1.2532921042810683, "grad_norm": 1.3981428146362305, "learning_rate": 5.2986e-06, "loss": 0.0094, "step": 23508 }, { "epoch": 1.2533454177107213, "grad_norm": 1.333046555519104, "learning_rate": 5.2984e-06, "loss": 0.0123, "step": 23509 }, { "epoch": 1.2533987311403743, "grad_norm": 1.471442699432373, "learning_rate": 5.2982e-06, "loss": 0.0182, "step": 23510 }, { "epoch": 1.2534520445700272, "grad_norm": 2.659162759780884, "learning_rate": 5.298000000000001e-06, "loss": 0.0241, "step": 23511 }, { "epoch": 1.2535053579996802, "grad_norm": 3.2590551376342773, "learning_rate": 5.2978e-06, "loss": 0.0342, "step": 23512 }, { "epoch": 1.253558671429333, "grad_norm": 1.344610333442688, "learning_rate": 5.297600000000001e-06, "loss": 0.0051, "step": 23513 }, { "epoch": 1.253611984858986, "grad_norm": 2.051177501678467, "learning_rate": 5.2974e-06, "loss": 0.028, "step": 23514 }, { "epoch": 1.253665298288639, "grad_norm": 1.822845458984375, "learning_rate": 5.2972e-06, "loss": 0.0313, "step": 23515 }, { "epoch": 1.253718611718292, "grad_norm": 0.9658756852149963, "learning_rate": 5.297e-06, "loss": 0.01, "step": 23516 }, { "epoch": 1.2537719251479449, "grad_norm": 0.7219016551971436, "learning_rate": 5.296800000000001e-06, "loss": 0.0082, "step": 23517 }, { "epoch": 1.2538252385775976, "grad_norm": 1.6373463869094849, "learning_rate": 5.2966e-06, "loss": 0.0372, "step": 23518 }, { "epoch": 1.2538785520072506, "grad_norm": 1.3240810632705688, "learning_rate": 5.296400000000001e-06, "loss": 0.0097, "step": 23519 }, { "epoch": 1.2539318654369036, "grad_norm": 4.659637928009033, "learning_rate": 5.2962e-06, "loss": 0.0323, "step": 23520 }, { "epoch": 1.2539851788665564, "grad_norm": 1.8356996774673462, "learning_rate": 5.296e-06, "loss": 0.011, "step": 23521 }, { "epoch": 1.2540384922962093, "grad_norm": 1.602485179901123, "learning_rate": 5.295800000000001e-06, "loss": 0.0147, "step": 23522 }, { "epoch": 1.2540918057258623, "grad_norm": 2.2725210189819336, "learning_rate": 5.295600000000001e-06, "loss": 0.0547, "step": 23523 }, { "epoch": 1.2541451191555153, "grad_norm": 1.512816309928894, "learning_rate": 5.2954e-06, "loss": 0.0157, "step": 23524 }, { "epoch": 1.2541984325851683, "grad_norm": 3.9902255535125732, "learning_rate": 5.2952000000000006e-06, "loss": 0.0568, "step": 23525 }, { "epoch": 1.254251746014821, "grad_norm": 2.340484380722046, "learning_rate": 5.295e-06, "loss": 0.0365, "step": 23526 }, { "epoch": 1.254305059444474, "grad_norm": 1.8644975423812866, "learning_rate": 5.2948e-06, "loss": 0.0169, "step": 23527 }, { "epoch": 1.254358372874127, "grad_norm": 1.4364806413650513, "learning_rate": 5.294600000000001e-06, "loss": 0.0219, "step": 23528 }, { "epoch": 1.25441168630378, "grad_norm": 2.9847614765167236, "learning_rate": 5.2944000000000005e-06, "loss": 0.0309, "step": 23529 }, { "epoch": 1.254464999733433, "grad_norm": 1.9154399633407593, "learning_rate": 5.2942e-06, "loss": 0.0258, "step": 23530 }, { "epoch": 1.2545183131630857, "grad_norm": 1.4086689949035645, "learning_rate": 5.2940000000000005e-06, "loss": 0.0293, "step": 23531 }, { "epoch": 1.2545716265927387, "grad_norm": 2.7973439693450928, "learning_rate": 5.2938e-06, "loss": 0.0551, "step": 23532 }, { "epoch": 1.2546249400223917, "grad_norm": 2.379164934158325, "learning_rate": 5.293600000000001e-06, "loss": 0.0384, "step": 23533 }, { "epoch": 1.2546782534520446, "grad_norm": 2.1378390789031982, "learning_rate": 5.293400000000001e-06, "loss": 0.0673, "step": 23534 }, { "epoch": 1.2547315668816976, "grad_norm": 1.22323477268219, "learning_rate": 5.2932e-06, "loss": 0.0062, "step": 23535 }, { "epoch": 1.2547848803113504, "grad_norm": 1.7777483463287354, "learning_rate": 5.293e-06, "loss": 0.0114, "step": 23536 }, { "epoch": 1.2548381937410034, "grad_norm": 2.2742927074432373, "learning_rate": 5.2928e-06, "loss": 0.0475, "step": 23537 }, { "epoch": 1.2548915071706563, "grad_norm": 1.3984060287475586, "learning_rate": 5.2926e-06, "loss": 0.0238, "step": 23538 }, { "epoch": 1.254944820600309, "grad_norm": 1.677153468132019, "learning_rate": 5.292400000000001e-06, "loss": 0.0174, "step": 23539 }, { "epoch": 1.254998134029962, "grad_norm": 2.2411351203918457, "learning_rate": 5.292200000000001e-06, "loss": 0.0182, "step": 23540 }, { "epoch": 1.255051447459615, "grad_norm": 2.851348638534546, "learning_rate": 5.292e-06, "loss": 0.0283, "step": 23541 }, { "epoch": 1.255104760889268, "grad_norm": 1.3543550968170166, "learning_rate": 5.2918e-06, "loss": 0.0135, "step": 23542 }, { "epoch": 1.255158074318921, "grad_norm": 1.6533399820327759, "learning_rate": 5.2916e-06, "loss": 0.0273, "step": 23543 }, { "epoch": 1.2552113877485738, "grad_norm": 0.6704700589179993, "learning_rate": 5.291400000000001e-06, "loss": 0.0076, "step": 23544 }, { "epoch": 1.2552647011782267, "grad_norm": 1.9659485816955566, "learning_rate": 5.291200000000001e-06, "loss": 0.0423, "step": 23545 }, { "epoch": 1.2553180146078797, "grad_norm": 1.6563265323638916, "learning_rate": 5.291000000000001e-06, "loss": 0.0147, "step": 23546 }, { "epoch": 1.2553713280375327, "grad_norm": 1.9873332977294922, "learning_rate": 5.2908e-06, "loss": 0.038, "step": 23547 }, { "epoch": 1.2554246414671857, "grad_norm": 2.7547688484191895, "learning_rate": 5.2906e-06, "loss": 0.0348, "step": 23548 }, { "epoch": 1.2554779548968384, "grad_norm": 0.7200754284858704, "learning_rate": 5.2904e-06, "loss": 0.0064, "step": 23549 }, { "epoch": 1.2555312683264914, "grad_norm": 1.4904191493988037, "learning_rate": 5.290200000000001e-06, "loss": 0.0211, "step": 23550 }, { "epoch": 1.2555845817561444, "grad_norm": 1.8633670806884766, "learning_rate": 5.290000000000001e-06, "loss": 0.0185, "step": 23551 }, { "epoch": 1.2556378951857974, "grad_norm": 1.7317876815795898, "learning_rate": 5.289800000000001e-06, "loss": 0.0135, "step": 23552 }, { "epoch": 1.2556912086154504, "grad_norm": 0.8108058571815491, "learning_rate": 5.2896e-06, "loss": 0.007, "step": 23553 }, { "epoch": 1.2557445220451031, "grad_norm": 0.5141022801399231, "learning_rate": 5.2894e-06, "loss": 0.0059, "step": 23554 }, { "epoch": 1.255797835474756, "grad_norm": 1.6954760551452637, "learning_rate": 5.2892e-06, "loss": 0.0275, "step": 23555 }, { "epoch": 1.255851148904409, "grad_norm": 0.7368233799934387, "learning_rate": 5.2890000000000005e-06, "loss": 0.0054, "step": 23556 }, { "epoch": 1.2559044623340618, "grad_norm": 1.2832165956497192, "learning_rate": 5.288800000000001e-06, "loss": 0.0157, "step": 23557 }, { "epoch": 1.2559577757637148, "grad_norm": 2.211061477661133, "learning_rate": 5.2886000000000005e-06, "loss": 0.0293, "step": 23558 }, { "epoch": 1.2560110891933678, "grad_norm": 2.828228712081909, "learning_rate": 5.2884e-06, "loss": 0.0239, "step": 23559 }, { "epoch": 1.2560644026230208, "grad_norm": 0.6478045582771301, "learning_rate": 5.2882e-06, "loss": 0.0052, "step": 23560 }, { "epoch": 1.2561177160526737, "grad_norm": 1.0146485567092896, "learning_rate": 5.288000000000001e-06, "loss": 0.0061, "step": 23561 }, { "epoch": 1.2561710294823265, "grad_norm": 0.5207919478416443, "learning_rate": 5.2878000000000004e-06, "loss": 0.0049, "step": 23562 }, { "epoch": 1.2562243429119795, "grad_norm": 1.280645728111267, "learning_rate": 5.287600000000001e-06, "loss": 0.0153, "step": 23563 }, { "epoch": 1.2562776563416325, "grad_norm": 1.043273687362671, "learning_rate": 5.2874e-06, "loss": 0.0119, "step": 23564 }, { "epoch": 1.2563309697712854, "grad_norm": 3.8729026317596436, "learning_rate": 5.2872e-06, "loss": 0.0337, "step": 23565 }, { "epoch": 1.2563842832009384, "grad_norm": 2.0177245140075684, "learning_rate": 5.2869999999999995e-06, "loss": 0.0226, "step": 23566 }, { "epoch": 1.2564375966305912, "grad_norm": 2.913144826889038, "learning_rate": 5.286800000000001e-06, "loss": 0.0817, "step": 23567 }, { "epoch": 1.2564909100602442, "grad_norm": 0.9499874711036682, "learning_rate": 5.2866e-06, "loss": 0.0171, "step": 23568 }, { "epoch": 1.2565442234898971, "grad_norm": 1.2782810926437378, "learning_rate": 5.286400000000001e-06, "loss": 0.0239, "step": 23569 }, { "epoch": 1.2565975369195501, "grad_norm": 1.9147788286209106, "learning_rate": 5.2862e-06, "loss": 0.0347, "step": 23570 }, { "epoch": 1.256650850349203, "grad_norm": 1.3717621564865112, "learning_rate": 5.286e-06, "loss": 0.013, "step": 23571 }, { "epoch": 1.2567041637788559, "grad_norm": 3.0867347717285156, "learning_rate": 5.285800000000001e-06, "loss": 0.0447, "step": 23572 }, { "epoch": 1.2567574772085088, "grad_norm": 1.2986459732055664, "learning_rate": 5.285600000000001e-06, "loss": 0.0142, "step": 23573 }, { "epoch": 1.2568107906381618, "grad_norm": 1.3517857789993286, "learning_rate": 5.2854e-06, "loss": 0.0244, "step": 23574 }, { "epoch": 1.2568641040678146, "grad_norm": 1.5543221235275269, "learning_rate": 5.285200000000001e-06, "loss": 0.0146, "step": 23575 }, { "epoch": 1.2569174174974675, "grad_norm": 1.4941052198410034, "learning_rate": 5.285e-06, "loss": 0.0263, "step": 23576 }, { "epoch": 1.2569707309271205, "grad_norm": 1.0316505432128906, "learning_rate": 5.2848e-06, "loss": 0.0104, "step": 23577 }, { "epoch": 1.2570240443567735, "grad_norm": 1.8922743797302246, "learning_rate": 5.284600000000001e-06, "loss": 0.0208, "step": 23578 }, { "epoch": 1.2570773577864265, "grad_norm": 1.9891040325164795, "learning_rate": 5.284400000000001e-06, "loss": 0.0326, "step": 23579 }, { "epoch": 1.2571306712160792, "grad_norm": 2.7534191608428955, "learning_rate": 5.2842e-06, "loss": 0.0193, "step": 23580 }, { "epoch": 1.2571839846457322, "grad_norm": 0.8623211979866028, "learning_rate": 5.2840000000000006e-06, "loss": 0.0077, "step": 23581 }, { "epoch": 1.2572372980753852, "grad_norm": 1.9557487964630127, "learning_rate": 5.2838e-06, "loss": 0.0199, "step": 23582 }, { "epoch": 1.2572906115050382, "grad_norm": 2.920943021774292, "learning_rate": 5.2836000000000005e-06, "loss": 0.0531, "step": 23583 }, { "epoch": 1.2573439249346912, "grad_norm": 1.5808976888656616, "learning_rate": 5.283400000000001e-06, "loss": 0.0133, "step": 23584 }, { "epoch": 1.257397238364344, "grad_norm": 1.7067830562591553, "learning_rate": 5.2832000000000005e-06, "loss": 0.0229, "step": 23585 }, { "epoch": 1.257450551793997, "grad_norm": 1.759940505027771, "learning_rate": 5.283e-06, "loss": 0.0232, "step": 23586 }, { "epoch": 1.2575038652236499, "grad_norm": 2.190526008605957, "learning_rate": 5.2828000000000005e-06, "loss": 0.0123, "step": 23587 }, { "epoch": 1.2575571786533026, "grad_norm": 0.9761170744895935, "learning_rate": 5.2826e-06, "loss": 0.0071, "step": 23588 }, { "epoch": 1.2576104920829558, "grad_norm": 2.272205352783203, "learning_rate": 5.2824000000000004e-06, "loss": 0.0412, "step": 23589 }, { "epoch": 1.2576638055126086, "grad_norm": 0.3220921754837036, "learning_rate": 5.282200000000001e-06, "loss": 0.0036, "step": 23590 }, { "epoch": 1.2577171189422616, "grad_norm": 2.516691207885742, "learning_rate": 5.282e-06, "loss": 0.0227, "step": 23591 }, { "epoch": 1.2577704323719145, "grad_norm": 2.531543016433716, "learning_rate": 5.2818e-06, "loss": 0.0251, "step": 23592 }, { "epoch": 1.2578237458015673, "grad_norm": 2.253241539001465, "learning_rate": 5.2816e-06, "loss": 0.0445, "step": 23593 }, { "epoch": 1.2578770592312203, "grad_norm": 1.6062637567520142, "learning_rate": 5.281400000000001e-06, "loss": 0.0241, "step": 23594 }, { "epoch": 1.2579303726608733, "grad_norm": 2.2560222148895264, "learning_rate": 5.2812e-06, "loss": 0.0249, "step": 23595 }, { "epoch": 1.2579836860905262, "grad_norm": 1.7797718048095703, "learning_rate": 5.281000000000001e-06, "loss": 0.0313, "step": 23596 }, { "epoch": 1.2580369995201792, "grad_norm": 0.5084954500198364, "learning_rate": 5.2808e-06, "loss": 0.0052, "step": 23597 }, { "epoch": 1.258090312949832, "grad_norm": 2.005971670150757, "learning_rate": 5.2806e-06, "loss": 0.0191, "step": 23598 }, { "epoch": 1.258143626379485, "grad_norm": 1.5816924571990967, "learning_rate": 5.2804e-06, "loss": 0.032, "step": 23599 }, { "epoch": 1.258196939809138, "grad_norm": 2.3852698802948, "learning_rate": 5.280200000000001e-06, "loss": 0.0226, "step": 23600 }, { "epoch": 1.258250253238791, "grad_norm": 2.2267446517944336, "learning_rate": 5.28e-06, "loss": 0.0331, "step": 23601 }, { "epoch": 1.258303566668444, "grad_norm": 2.4459102153778076, "learning_rate": 5.279800000000001e-06, "loss": 0.0444, "step": 23602 }, { "epoch": 1.2583568800980967, "grad_norm": 1.8168076276779175, "learning_rate": 5.2796e-06, "loss": 0.0123, "step": 23603 }, { "epoch": 1.2584101935277496, "grad_norm": 1.874467372894287, "learning_rate": 5.2794e-06, "loss": 0.0183, "step": 23604 }, { "epoch": 1.2584635069574026, "grad_norm": 1.6838020086288452, "learning_rate": 5.2792e-06, "loss": 0.0257, "step": 23605 }, { "epoch": 1.2585168203870554, "grad_norm": 1.8397101163864136, "learning_rate": 5.279000000000001e-06, "loss": 0.0141, "step": 23606 }, { "epoch": 1.2585701338167083, "grad_norm": 0.8074929118156433, "learning_rate": 5.2788e-06, "loss": 0.0069, "step": 23607 }, { "epoch": 1.2586234472463613, "grad_norm": 2.841776132583618, "learning_rate": 5.278600000000001e-06, "loss": 0.02, "step": 23608 }, { "epoch": 1.2586767606760143, "grad_norm": 1.4840679168701172, "learning_rate": 5.2784e-06, "loss": 0.0067, "step": 23609 }, { "epoch": 1.2587300741056673, "grad_norm": 3.8698554039001465, "learning_rate": 5.2782e-06, "loss": 0.0403, "step": 23610 }, { "epoch": 1.25878338753532, "grad_norm": 0.9839016199111938, "learning_rate": 5.278000000000001e-06, "loss": 0.0059, "step": 23611 }, { "epoch": 1.258836700964973, "grad_norm": 3.8804404735565186, "learning_rate": 5.2778000000000005e-06, "loss": 0.0541, "step": 23612 }, { "epoch": 1.258890014394626, "grad_norm": 2.0833353996276855, "learning_rate": 5.2776e-06, "loss": 0.0204, "step": 23613 }, { "epoch": 1.258943327824279, "grad_norm": 2.5664172172546387, "learning_rate": 5.2774000000000005e-06, "loss": 0.0282, "step": 23614 }, { "epoch": 1.258996641253932, "grad_norm": 2.6144044399261475, "learning_rate": 5.2772e-06, "loss": 0.0221, "step": 23615 }, { "epoch": 1.2590499546835847, "grad_norm": 1.5306150913238525, "learning_rate": 5.277e-06, "loss": 0.0173, "step": 23616 }, { "epoch": 1.2591032681132377, "grad_norm": 1.732001781463623, "learning_rate": 5.276800000000001e-06, "loss": 0.0308, "step": 23617 }, { "epoch": 1.2591565815428907, "grad_norm": 5.275287628173828, "learning_rate": 5.2766000000000004e-06, "loss": 0.0541, "step": 23618 }, { "epoch": 1.2592098949725437, "grad_norm": 2.0628693103790283, "learning_rate": 5.2764e-06, "loss": 0.0172, "step": 23619 }, { "epoch": 1.2592632084021966, "grad_norm": 1.8735166788101196, "learning_rate": 5.2762e-06, "loss": 0.0184, "step": 23620 }, { "epoch": 1.2593165218318494, "grad_norm": 2.6824259757995605, "learning_rate": 5.276e-06, "loss": 0.0415, "step": 23621 }, { "epoch": 1.2593698352615024, "grad_norm": 1.781577229499817, "learning_rate": 5.275800000000001e-06, "loss": 0.0135, "step": 23622 }, { "epoch": 1.2594231486911553, "grad_norm": 1.7291100025177002, "learning_rate": 5.275600000000001e-06, "loss": 0.0126, "step": 23623 }, { "epoch": 1.259476462120808, "grad_norm": 1.528930425643921, "learning_rate": 5.2754e-06, "loss": 0.0107, "step": 23624 }, { "epoch": 1.259529775550461, "grad_norm": 1.1973687410354614, "learning_rate": 5.2752e-06, "loss": 0.0175, "step": 23625 }, { "epoch": 1.259583088980114, "grad_norm": 1.1199754476547241, "learning_rate": 5.275e-06, "loss": 0.01, "step": 23626 }, { "epoch": 1.259636402409767, "grad_norm": 2.923292875289917, "learning_rate": 5.2748e-06, "loss": 0.0278, "step": 23627 }, { "epoch": 1.25968971583942, "grad_norm": 3.3095197677612305, "learning_rate": 5.274600000000001e-06, "loss": 0.0261, "step": 23628 }, { "epoch": 1.2597430292690728, "grad_norm": 1.2827781438827515, "learning_rate": 5.274400000000001e-06, "loss": 0.0112, "step": 23629 }, { "epoch": 1.2597963426987258, "grad_norm": 1.2220869064331055, "learning_rate": 5.2742e-06, "loss": 0.0104, "step": 23630 }, { "epoch": 1.2598496561283787, "grad_norm": 0.3985794484615326, "learning_rate": 5.274e-06, "loss": 0.0041, "step": 23631 }, { "epoch": 1.2599029695580317, "grad_norm": 1.0174181461334229, "learning_rate": 5.2738e-06, "loss": 0.0076, "step": 23632 }, { "epoch": 1.2599562829876847, "grad_norm": 0.7173537015914917, "learning_rate": 5.273600000000001e-06, "loss": 0.0052, "step": 23633 }, { "epoch": 1.2600095964173375, "grad_norm": 1.2068581581115723, "learning_rate": 5.273400000000001e-06, "loss": 0.0137, "step": 23634 }, { "epoch": 1.2600629098469904, "grad_norm": 1.4473267793655396, "learning_rate": 5.273200000000001e-06, "loss": 0.0094, "step": 23635 }, { "epoch": 1.2601162232766434, "grad_norm": 2.68066143989563, "learning_rate": 5.273e-06, "loss": 0.0469, "step": 23636 }, { "epoch": 1.2601695367062964, "grad_norm": 1.5209189653396606, "learning_rate": 5.2728e-06, "loss": 0.0133, "step": 23637 }, { "epoch": 1.2602228501359494, "grad_norm": 2.4769446849823, "learning_rate": 5.2726e-06, "loss": 0.0364, "step": 23638 }, { "epoch": 1.2602761635656021, "grad_norm": 0.937334418296814, "learning_rate": 5.2724000000000005e-06, "loss": 0.0062, "step": 23639 }, { "epoch": 1.260329476995255, "grad_norm": 1.0090782642364502, "learning_rate": 5.272200000000001e-06, "loss": 0.0061, "step": 23640 }, { "epoch": 1.260382790424908, "grad_norm": 1.0113211870193481, "learning_rate": 5.2720000000000005e-06, "loss": 0.0229, "step": 23641 }, { "epoch": 1.2604361038545608, "grad_norm": 1.3257039785385132, "learning_rate": 5.2718e-06, "loss": 0.0131, "step": 23642 }, { "epoch": 1.2604894172842138, "grad_norm": 1.269814133644104, "learning_rate": 5.2716e-06, "loss": 0.0113, "step": 23643 }, { "epoch": 1.2605427307138668, "grad_norm": 1.0120861530303955, "learning_rate": 5.271400000000001e-06, "loss": 0.0143, "step": 23644 }, { "epoch": 1.2605960441435198, "grad_norm": 1.6935416460037231, "learning_rate": 5.2712000000000005e-06, "loss": 0.0408, "step": 23645 }, { "epoch": 1.2606493575731728, "grad_norm": 2.5383026599884033, "learning_rate": 5.271000000000001e-06, "loss": 0.0553, "step": 23646 }, { "epoch": 1.2607026710028255, "grad_norm": 2.6027588844299316, "learning_rate": 5.2708000000000004e-06, "loss": 0.0308, "step": 23647 }, { "epoch": 1.2607559844324785, "grad_norm": 0.7045408487319946, "learning_rate": 5.2706e-06, "loss": 0.0074, "step": 23648 }, { "epoch": 1.2608092978621315, "grad_norm": 1.4922049045562744, "learning_rate": 5.2703999999999995e-06, "loss": 0.0085, "step": 23649 }, { "epoch": 1.2608626112917845, "grad_norm": 1.704514503479004, "learning_rate": 5.270200000000001e-06, "loss": 0.0203, "step": 23650 }, { "epoch": 1.2609159247214374, "grad_norm": 1.2071248292922974, "learning_rate": 5.27e-06, "loss": 0.011, "step": 23651 }, { "epoch": 1.2609692381510902, "grad_norm": 1.6189002990722656, "learning_rate": 5.269800000000001e-06, "loss": 0.0325, "step": 23652 }, { "epoch": 1.2610225515807432, "grad_norm": 1.6337052583694458, "learning_rate": 5.2696e-06, "loss": 0.0336, "step": 23653 }, { "epoch": 1.2610758650103961, "grad_norm": 2.4069297313690186, "learning_rate": 5.2694e-06, "loss": 0.0292, "step": 23654 }, { "epoch": 1.2611291784400491, "grad_norm": 0.9712591767311096, "learning_rate": 5.269200000000001e-06, "loss": 0.008, "step": 23655 }, { "epoch": 1.261182491869702, "grad_norm": 2.0661234855651855, "learning_rate": 5.269000000000001e-06, "loss": 0.0319, "step": 23656 }, { "epoch": 1.2612358052993549, "grad_norm": 2.779010534286499, "learning_rate": 5.2688e-06, "loss": 0.013, "step": 23657 }, { "epoch": 1.2612891187290078, "grad_norm": 2.066532850265503, "learning_rate": 5.268600000000001e-06, "loss": 0.0148, "step": 23658 }, { "epoch": 1.2613424321586608, "grad_norm": 0.42570817470550537, "learning_rate": 5.2684e-06, "loss": 0.0036, "step": 23659 }, { "epoch": 1.2613957455883136, "grad_norm": 1.5598291158676147, "learning_rate": 5.2682e-06, "loss": 0.0441, "step": 23660 }, { "epoch": 1.2614490590179666, "grad_norm": 1.2290054559707642, "learning_rate": 5.268000000000001e-06, "loss": 0.0086, "step": 23661 }, { "epoch": 1.2615023724476195, "grad_norm": 2.726797103881836, "learning_rate": 5.267800000000001e-06, "loss": 0.0364, "step": 23662 }, { "epoch": 1.2615556858772725, "grad_norm": 3.2280960083007812, "learning_rate": 5.2676e-06, "loss": 0.0486, "step": 23663 }, { "epoch": 1.2616089993069255, "grad_norm": 1.0237746238708496, "learning_rate": 5.267400000000001e-06, "loss": 0.0277, "step": 23664 }, { "epoch": 1.2616623127365783, "grad_norm": 1.8635629415512085, "learning_rate": 5.2672e-06, "loss": 0.0424, "step": 23665 }, { "epoch": 1.2617156261662312, "grad_norm": 1.6664460897445679, "learning_rate": 5.267e-06, "loss": 0.02, "step": 23666 }, { "epoch": 1.2617689395958842, "grad_norm": 0.8520922064781189, "learning_rate": 5.266800000000001e-06, "loss": 0.0248, "step": 23667 }, { "epoch": 1.2618222530255372, "grad_norm": 2.074594497680664, "learning_rate": 5.2666000000000005e-06, "loss": 0.0318, "step": 23668 }, { "epoch": 1.2618755664551902, "grad_norm": 1.9240896701812744, "learning_rate": 5.2664e-06, "loss": 0.0417, "step": 23669 }, { "epoch": 1.261928879884843, "grad_norm": 1.9768458604812622, "learning_rate": 5.2662000000000005e-06, "loss": 0.0426, "step": 23670 }, { "epoch": 1.261982193314496, "grad_norm": 1.8865077495574951, "learning_rate": 5.266e-06, "loss": 0.0183, "step": 23671 }, { "epoch": 1.2620355067441489, "grad_norm": 1.8680249452590942, "learning_rate": 5.265800000000001e-06, "loss": 0.0315, "step": 23672 }, { "epoch": 1.2620888201738019, "grad_norm": 2.248955249786377, "learning_rate": 5.265600000000001e-06, "loss": 0.0558, "step": 23673 }, { "epoch": 1.2621421336034548, "grad_norm": 2.947033166885376, "learning_rate": 5.2654000000000004e-06, "loss": 0.0161, "step": 23674 }, { "epoch": 1.2621954470331076, "grad_norm": 1.1684831380844116, "learning_rate": 5.2652e-06, "loss": 0.0181, "step": 23675 }, { "epoch": 1.2622487604627606, "grad_norm": 3.859757661819458, "learning_rate": 5.265e-06, "loss": 0.0279, "step": 23676 }, { "epoch": 1.2623020738924136, "grad_norm": 2.419412851333618, "learning_rate": 5.2648e-06, "loss": 0.0237, "step": 23677 }, { "epoch": 1.2623553873220663, "grad_norm": 1.6469069719314575, "learning_rate": 5.264600000000001e-06, "loss": 0.03, "step": 23678 }, { "epoch": 1.2624087007517193, "grad_norm": 0.8655005693435669, "learning_rate": 5.264400000000001e-06, "loss": 0.0077, "step": 23679 }, { "epoch": 1.2624620141813723, "grad_norm": 4.657641410827637, "learning_rate": 5.2642e-06, "loss": 0.0365, "step": 23680 }, { "epoch": 1.2625153276110253, "grad_norm": 1.0444916486740112, "learning_rate": 5.264e-06, "loss": 0.014, "step": 23681 }, { "epoch": 1.2625686410406782, "grad_norm": 1.4038165807724, "learning_rate": 5.2638e-06, "loss": 0.0152, "step": 23682 }, { "epoch": 1.262621954470331, "grad_norm": 2.6345462799072266, "learning_rate": 5.263600000000001e-06, "loss": 0.0167, "step": 23683 }, { "epoch": 1.262675267899984, "grad_norm": 1.3020697832107544, "learning_rate": 5.263400000000001e-06, "loss": 0.0232, "step": 23684 }, { "epoch": 1.262728581329637, "grad_norm": 0.7954035401344299, "learning_rate": 5.263200000000001e-06, "loss": 0.0041, "step": 23685 }, { "epoch": 1.26278189475929, "grad_norm": 1.5228097438812256, "learning_rate": 5.263e-06, "loss": 0.0083, "step": 23686 }, { "epoch": 1.262835208188943, "grad_norm": 1.261147141456604, "learning_rate": 5.2628e-06, "loss": 0.0194, "step": 23687 }, { "epoch": 1.2628885216185957, "grad_norm": 2.1710238456726074, "learning_rate": 5.2626e-06, "loss": 0.0281, "step": 23688 }, { "epoch": 1.2629418350482486, "grad_norm": 1.433085560798645, "learning_rate": 5.262400000000001e-06, "loss": 0.0161, "step": 23689 }, { "epoch": 1.2629951484779016, "grad_norm": 0.8021183013916016, "learning_rate": 5.262200000000001e-06, "loss": 0.0102, "step": 23690 }, { "epoch": 1.2630484619075544, "grad_norm": 1.655134677886963, "learning_rate": 5.262000000000001e-06, "loss": 0.0298, "step": 23691 }, { "epoch": 1.2631017753372076, "grad_norm": 1.5662466287612915, "learning_rate": 5.2618e-06, "loss": 0.0431, "step": 23692 }, { "epoch": 1.2631550887668603, "grad_norm": 3.027576446533203, "learning_rate": 5.2616e-06, "loss": 0.0223, "step": 23693 }, { "epoch": 1.2632084021965133, "grad_norm": 2.316815137863159, "learning_rate": 5.261400000000001e-06, "loss": 0.0233, "step": 23694 }, { "epoch": 1.2632617156261663, "grad_norm": 2.051575183868408, "learning_rate": 5.2612000000000006e-06, "loss": 0.0195, "step": 23695 }, { "epoch": 1.263315029055819, "grad_norm": 0.9729223251342773, "learning_rate": 5.261000000000001e-06, "loss": 0.0344, "step": 23696 }, { "epoch": 1.263368342485472, "grad_norm": 4.01963996887207, "learning_rate": 5.2608000000000005e-06, "loss": 0.0345, "step": 23697 }, { "epoch": 1.263421655915125, "grad_norm": 2.2809698581695557, "learning_rate": 5.2606e-06, "loss": 0.0266, "step": 23698 }, { "epoch": 1.263474969344778, "grad_norm": 2.070725202560425, "learning_rate": 5.2604e-06, "loss": 0.0252, "step": 23699 }, { "epoch": 1.263528282774431, "grad_norm": 3.2690072059631348, "learning_rate": 5.260200000000001e-06, "loss": 0.0324, "step": 23700 }, { "epoch": 1.2635815962040837, "grad_norm": 1.2416682243347168, "learning_rate": 5.2600000000000005e-06, "loss": 0.0187, "step": 23701 }, { "epoch": 1.2636349096337367, "grad_norm": 1.5051538944244385, "learning_rate": 5.259800000000001e-06, "loss": 0.0185, "step": 23702 }, { "epoch": 1.2636882230633897, "grad_norm": 1.2255889177322388, "learning_rate": 5.2596000000000004e-06, "loss": 0.0095, "step": 23703 }, { "epoch": 1.2637415364930427, "grad_norm": 1.065553903579712, "learning_rate": 5.2594e-06, "loss": 0.0276, "step": 23704 }, { "epoch": 1.2637948499226956, "grad_norm": 1.9453668594360352, "learning_rate": 5.259200000000001e-06, "loss": 0.0222, "step": 23705 }, { "epoch": 1.2638481633523484, "grad_norm": 1.820854902267456, "learning_rate": 5.259000000000001e-06, "loss": 0.0314, "step": 23706 }, { "epoch": 1.2639014767820014, "grad_norm": 1.8101409673690796, "learning_rate": 5.2588e-06, "loss": 0.0421, "step": 23707 }, { "epoch": 1.2639547902116544, "grad_norm": 2.8057639598846436, "learning_rate": 5.258600000000001e-06, "loss": 0.0449, "step": 23708 }, { "epoch": 1.2640081036413071, "grad_norm": 1.9198402166366577, "learning_rate": 5.2584e-06, "loss": 0.0284, "step": 23709 }, { "epoch": 1.26406141707096, "grad_norm": 1.0394344329833984, "learning_rate": 5.2582e-06, "loss": 0.0303, "step": 23710 }, { "epoch": 1.264114730500613, "grad_norm": 2.212289333343506, "learning_rate": 5.258000000000001e-06, "loss": 0.0354, "step": 23711 }, { "epoch": 1.264168043930266, "grad_norm": 2.8722431659698486, "learning_rate": 5.257800000000001e-06, "loss": 0.0328, "step": 23712 }, { "epoch": 1.264221357359919, "grad_norm": 2.736102819442749, "learning_rate": 5.2576e-06, "loss": 0.0371, "step": 23713 }, { "epoch": 1.2642746707895718, "grad_norm": 1.8343427181243896, "learning_rate": 5.257400000000001e-06, "loss": 0.0211, "step": 23714 }, { "epoch": 1.2643279842192248, "grad_norm": 0.9663007259368896, "learning_rate": 5.2572e-06, "loss": 0.0095, "step": 23715 }, { "epoch": 1.2643812976488777, "grad_norm": 2.559285879135132, "learning_rate": 5.257e-06, "loss": 0.0381, "step": 23716 }, { "epoch": 1.2644346110785307, "grad_norm": 2.3122267723083496, "learning_rate": 5.256800000000001e-06, "loss": 0.0275, "step": 23717 }, { "epoch": 1.2644879245081837, "grad_norm": 2.902956247329712, "learning_rate": 5.256600000000001e-06, "loss": 0.0229, "step": 23718 }, { "epoch": 1.2645412379378365, "grad_norm": 0.7981948852539062, "learning_rate": 5.2564e-06, "loss": 0.0081, "step": 23719 }, { "epoch": 1.2645945513674894, "grad_norm": 3.020801305770874, "learning_rate": 5.256200000000001e-06, "loss": 0.0474, "step": 23720 }, { "epoch": 1.2646478647971424, "grad_norm": 2.88845157623291, "learning_rate": 5.256e-06, "loss": 0.0284, "step": 23721 }, { "epoch": 1.2647011782267954, "grad_norm": 1.1547458171844482, "learning_rate": 5.2558000000000006e-06, "loss": 0.0121, "step": 23722 }, { "epoch": 1.2647544916564484, "grad_norm": 1.1686094999313354, "learning_rate": 5.255600000000001e-06, "loss": 0.0168, "step": 23723 }, { "epoch": 1.2648078050861011, "grad_norm": 2.1727259159088135, "learning_rate": 5.2554000000000005e-06, "loss": 0.0206, "step": 23724 }, { "epoch": 1.2648611185157541, "grad_norm": 2.244802236557007, "learning_rate": 5.2552e-06, "loss": 0.0193, "step": 23725 }, { "epoch": 1.264914431945407, "grad_norm": 1.1010055541992188, "learning_rate": 5.2550000000000005e-06, "loss": 0.0122, "step": 23726 }, { "epoch": 1.2649677453750598, "grad_norm": 0.8923413157463074, "learning_rate": 5.2548e-06, "loss": 0.0082, "step": 23727 }, { "epoch": 1.2650210588047128, "grad_norm": 2.774746894836426, "learning_rate": 5.2546000000000005e-06, "loss": 0.0269, "step": 23728 }, { "epoch": 1.2650743722343658, "grad_norm": 2.2856762409210205, "learning_rate": 5.254400000000001e-06, "loss": 0.0389, "step": 23729 }, { "epoch": 1.2651276856640188, "grad_norm": 1.9481321573257446, "learning_rate": 5.2542000000000004e-06, "loss": 0.0331, "step": 23730 }, { "epoch": 1.2651809990936718, "grad_norm": 1.6297893524169922, "learning_rate": 5.254e-06, "loss": 0.0132, "step": 23731 }, { "epoch": 1.2652343125233245, "grad_norm": 1.3787610530853271, "learning_rate": 5.2538e-06, "loss": 0.0117, "step": 23732 }, { "epoch": 1.2652876259529775, "grad_norm": 1.7256417274475098, "learning_rate": 5.253600000000001e-06, "loss": 0.0108, "step": 23733 }, { "epoch": 1.2653409393826305, "grad_norm": 1.0641095638275146, "learning_rate": 5.2534e-06, "loss": 0.0127, "step": 23734 }, { "epoch": 1.2653942528122835, "grad_norm": 3.1552734375, "learning_rate": 5.253200000000001e-06, "loss": 0.0375, "step": 23735 }, { "epoch": 1.2654475662419364, "grad_norm": 1.554855465888977, "learning_rate": 5.253e-06, "loss": 0.0237, "step": 23736 }, { "epoch": 1.2655008796715892, "grad_norm": 1.9092859029769897, "learning_rate": 5.2528e-06, "loss": 0.0211, "step": 23737 }, { "epoch": 1.2655541931012422, "grad_norm": 1.7569880485534668, "learning_rate": 5.2526e-06, "loss": 0.0336, "step": 23738 }, { "epoch": 1.2656075065308952, "grad_norm": 1.3996518850326538, "learning_rate": 5.252400000000001e-06, "loss": 0.0099, "step": 23739 }, { "epoch": 1.2656608199605481, "grad_norm": 1.9059470891952515, "learning_rate": 5.2522e-06, "loss": 0.0483, "step": 23740 }, { "epoch": 1.2657141333902011, "grad_norm": 1.4394272565841675, "learning_rate": 5.252000000000001e-06, "loss": 0.0375, "step": 23741 }, { "epoch": 1.2657674468198539, "grad_norm": 3.0706214904785156, "learning_rate": 5.2518e-06, "loss": 0.0214, "step": 23742 }, { "epoch": 1.2658207602495068, "grad_norm": 1.3897559642791748, "learning_rate": 5.2516e-06, "loss": 0.0354, "step": 23743 }, { "epoch": 1.2658740736791598, "grad_norm": 0.6268664598464966, "learning_rate": 5.251400000000001e-06, "loss": 0.0062, "step": 23744 }, { "epoch": 1.2659273871088126, "grad_norm": 0.746224582195282, "learning_rate": 5.251200000000001e-06, "loss": 0.0082, "step": 23745 }, { "epoch": 1.2659807005384656, "grad_norm": 3.4294166564941406, "learning_rate": 5.251e-06, "loss": 0.0359, "step": 23746 }, { "epoch": 1.2660340139681185, "grad_norm": 1.8863012790679932, "learning_rate": 5.250800000000001e-06, "loss": 0.0269, "step": 23747 }, { "epoch": 1.2660873273977715, "grad_norm": 1.9282455444335938, "learning_rate": 5.2506e-06, "loss": 0.0114, "step": 23748 }, { "epoch": 1.2661406408274245, "grad_norm": 1.2202147245407104, "learning_rate": 5.2504e-06, "loss": 0.0164, "step": 23749 }, { "epoch": 1.2661939542570773, "grad_norm": 2.1857359409332275, "learning_rate": 5.250200000000001e-06, "loss": 0.0176, "step": 23750 }, { "epoch": 1.2662472676867302, "grad_norm": 1.5264497995376587, "learning_rate": 5.2500000000000006e-06, "loss": 0.0264, "step": 23751 }, { "epoch": 1.2663005811163832, "grad_norm": 1.8217355012893677, "learning_rate": 5.2498e-06, "loss": 0.0097, "step": 23752 }, { "epoch": 1.2663538945460362, "grad_norm": 0.7649496793746948, "learning_rate": 5.2496000000000005e-06, "loss": 0.0054, "step": 23753 }, { "epoch": 1.2664072079756892, "grad_norm": 4.312803268432617, "learning_rate": 5.2494e-06, "loss": 0.0294, "step": 23754 }, { "epoch": 1.266460521405342, "grad_norm": 1.498507022857666, "learning_rate": 5.249200000000001e-06, "loss": 0.0086, "step": 23755 }, { "epoch": 1.266513834834995, "grad_norm": 1.8133949041366577, "learning_rate": 5.249000000000001e-06, "loss": 0.0362, "step": 23756 }, { "epoch": 1.266567148264648, "grad_norm": 1.2665061950683594, "learning_rate": 5.2488000000000005e-06, "loss": 0.0089, "step": 23757 }, { "epoch": 1.2666204616943009, "grad_norm": 1.5122201442718506, "learning_rate": 5.2486e-06, "loss": 0.0178, "step": 23758 }, { "epoch": 1.2666737751239538, "grad_norm": 0.8286768794059753, "learning_rate": 5.2484000000000004e-06, "loss": 0.0133, "step": 23759 }, { "epoch": 1.2667270885536066, "grad_norm": 2.4489219188690186, "learning_rate": 5.2482e-06, "loss": 0.0273, "step": 23760 }, { "epoch": 1.2667804019832596, "grad_norm": 2.927290201187134, "learning_rate": 5.248000000000001e-06, "loss": 0.0524, "step": 23761 }, { "epoch": 1.2668337154129126, "grad_norm": 0.6816241145133972, "learning_rate": 5.247800000000001e-06, "loss": 0.0059, "step": 23762 }, { "epoch": 1.2668870288425653, "grad_norm": 3.516752004623413, "learning_rate": 5.2476e-06, "loss": 0.0616, "step": 23763 }, { "epoch": 1.2669403422722183, "grad_norm": 1.2508859634399414, "learning_rate": 5.2474e-06, "loss": 0.0339, "step": 23764 }, { "epoch": 1.2669936557018713, "grad_norm": 1.3160176277160645, "learning_rate": 5.2472e-06, "loss": 0.0205, "step": 23765 }, { "epoch": 1.2670469691315243, "grad_norm": 0.9718391299247742, "learning_rate": 5.247000000000001e-06, "loss": 0.0108, "step": 23766 }, { "epoch": 1.2671002825611772, "grad_norm": 1.8826383352279663, "learning_rate": 5.246800000000001e-06, "loss": 0.0313, "step": 23767 }, { "epoch": 1.26715359599083, "grad_norm": 0.8361456990242004, "learning_rate": 5.246600000000001e-06, "loss": 0.0194, "step": 23768 }, { "epoch": 1.267206909420483, "grad_norm": 1.8037809133529663, "learning_rate": 5.2464e-06, "loss": 0.0193, "step": 23769 }, { "epoch": 1.267260222850136, "grad_norm": 1.8763208389282227, "learning_rate": 5.2462e-06, "loss": 0.0208, "step": 23770 }, { "epoch": 1.267313536279789, "grad_norm": 1.5453627109527588, "learning_rate": 5.246e-06, "loss": 0.0283, "step": 23771 }, { "epoch": 1.267366849709442, "grad_norm": 2.174898862838745, "learning_rate": 5.245800000000001e-06, "loss": 0.0147, "step": 23772 }, { "epoch": 1.2674201631390947, "grad_norm": 1.3376537561416626, "learning_rate": 5.245600000000001e-06, "loss": 0.0311, "step": 23773 }, { "epoch": 1.2674734765687476, "grad_norm": 0.8713042736053467, "learning_rate": 5.245400000000001e-06, "loss": 0.0084, "step": 23774 }, { "epoch": 1.2675267899984006, "grad_norm": 2.7000339031219482, "learning_rate": 5.2452e-06, "loss": 0.024, "step": 23775 }, { "epoch": 1.2675801034280536, "grad_norm": 1.6286649703979492, "learning_rate": 5.245e-06, "loss": 0.0327, "step": 23776 }, { "epoch": 1.2676334168577066, "grad_norm": 0.6920372843742371, "learning_rate": 5.2448e-06, "loss": 0.006, "step": 23777 }, { "epoch": 1.2676867302873593, "grad_norm": 0.8664634823799133, "learning_rate": 5.2446000000000006e-06, "loss": 0.0129, "step": 23778 }, { "epoch": 1.2677400437170123, "grad_norm": 1.0359231233596802, "learning_rate": 5.244400000000001e-06, "loss": 0.0181, "step": 23779 }, { "epoch": 1.2677933571466653, "grad_norm": 1.974990725517273, "learning_rate": 5.2442000000000005e-06, "loss": 0.0184, "step": 23780 }, { "epoch": 1.267846670576318, "grad_norm": 0.8198586106300354, "learning_rate": 5.244e-06, "loss": 0.0073, "step": 23781 }, { "epoch": 1.267899984005971, "grad_norm": 0.7365155816078186, "learning_rate": 5.2438e-06, "loss": 0.0031, "step": 23782 }, { "epoch": 1.267953297435624, "grad_norm": 1.6808449029922485, "learning_rate": 5.243600000000001e-06, "loss": 0.0204, "step": 23783 }, { "epoch": 1.268006610865277, "grad_norm": 1.1944193840026855, "learning_rate": 5.2434000000000005e-06, "loss": 0.0196, "step": 23784 }, { "epoch": 1.26805992429493, "grad_norm": 1.819359540939331, "learning_rate": 5.243200000000001e-06, "loss": 0.0144, "step": 23785 }, { "epoch": 1.2681132377245827, "grad_norm": 0.8760370016098022, "learning_rate": 5.2430000000000005e-06, "loss": 0.0432, "step": 23786 }, { "epoch": 1.2681665511542357, "grad_norm": 5.01428747177124, "learning_rate": 5.2428e-06, "loss": 0.0317, "step": 23787 }, { "epoch": 1.2682198645838887, "grad_norm": 3.1264407634735107, "learning_rate": 5.2426e-06, "loss": 0.0433, "step": 23788 }, { "epoch": 1.2682731780135417, "grad_norm": 1.176720380783081, "learning_rate": 5.242400000000001e-06, "loss": 0.0076, "step": 23789 }, { "epoch": 1.2683264914431946, "grad_norm": 1.3436377048492432, "learning_rate": 5.2422e-06, "loss": 0.0095, "step": 23790 }, { "epoch": 1.2683798048728474, "grad_norm": 1.8983564376831055, "learning_rate": 5.242000000000001e-06, "loss": 0.0299, "step": 23791 }, { "epoch": 1.2684331183025004, "grad_norm": 2.5191283226013184, "learning_rate": 5.2418e-06, "loss": 0.0342, "step": 23792 }, { "epoch": 1.2684864317321534, "grad_norm": 2.9224514961242676, "learning_rate": 5.2416e-06, "loss": 0.0237, "step": 23793 }, { "epoch": 1.2685397451618061, "grad_norm": 2.1303551197052, "learning_rate": 5.241400000000001e-06, "loss": 0.0205, "step": 23794 }, { "epoch": 1.2685930585914593, "grad_norm": 1.9875577688217163, "learning_rate": 5.241200000000001e-06, "loss": 0.0123, "step": 23795 }, { "epoch": 1.268646372021112, "grad_norm": 2.615859031677246, "learning_rate": 5.241e-06, "loss": 0.0325, "step": 23796 }, { "epoch": 1.268699685450765, "grad_norm": 1.926277995109558, "learning_rate": 5.240800000000001e-06, "loss": 0.0124, "step": 23797 }, { "epoch": 1.268752998880418, "grad_norm": 0.6943883895874023, "learning_rate": 5.2406e-06, "loss": 0.0055, "step": 23798 }, { "epoch": 1.2688063123100708, "grad_norm": 3.3369786739349365, "learning_rate": 5.2404e-06, "loss": 0.016, "step": 23799 }, { "epoch": 1.2688596257397238, "grad_norm": 0.9471617937088013, "learning_rate": 5.240200000000001e-06, "loss": 0.0067, "step": 23800 }, { "epoch": 1.2689129391693768, "grad_norm": 4.111340045928955, "learning_rate": 5.240000000000001e-06, "loss": 0.0306, "step": 23801 }, { "epoch": 1.2689662525990297, "grad_norm": 1.5594043731689453, "learning_rate": 5.2398e-06, "loss": 0.0141, "step": 23802 }, { "epoch": 1.2690195660286827, "grad_norm": 1.0100551843643188, "learning_rate": 5.239600000000001e-06, "loss": 0.012, "step": 23803 }, { "epoch": 1.2690728794583355, "grad_norm": 1.2031484842300415, "learning_rate": 5.2394e-06, "loss": 0.0218, "step": 23804 }, { "epoch": 1.2691261928879884, "grad_norm": 1.9240511655807495, "learning_rate": 5.239200000000001e-06, "loss": 0.0278, "step": 23805 }, { "epoch": 1.2691795063176414, "grad_norm": 3.5367770195007324, "learning_rate": 5.239000000000001e-06, "loss": 0.0103, "step": 23806 }, { "epoch": 1.2692328197472944, "grad_norm": 2.2098488807678223, "learning_rate": 5.2388000000000006e-06, "loss": 0.0237, "step": 23807 }, { "epoch": 1.2692861331769474, "grad_norm": 2.1339097023010254, "learning_rate": 5.2386e-06, "loss": 0.0148, "step": 23808 }, { "epoch": 1.2693394466066001, "grad_norm": 1.8838183879852295, "learning_rate": 5.2384000000000005e-06, "loss": 0.0163, "step": 23809 }, { "epoch": 1.2693927600362531, "grad_norm": 1.5327012538909912, "learning_rate": 5.2382e-06, "loss": 0.0287, "step": 23810 }, { "epoch": 1.269446073465906, "grad_norm": 2.864548921585083, "learning_rate": 5.2380000000000005e-06, "loss": 0.0537, "step": 23811 }, { "epoch": 1.2694993868955589, "grad_norm": 1.864924430847168, "learning_rate": 5.237800000000001e-06, "loss": 0.0317, "step": 23812 }, { "epoch": 1.269552700325212, "grad_norm": 1.4733699560165405, "learning_rate": 5.2376000000000005e-06, "loss": 0.0304, "step": 23813 }, { "epoch": 1.2696060137548648, "grad_norm": 2.674283504486084, "learning_rate": 5.2374e-06, "loss": 0.0688, "step": 23814 }, { "epoch": 1.2696593271845178, "grad_norm": 2.3832132816314697, "learning_rate": 5.2372000000000004e-06, "loss": 0.0526, "step": 23815 }, { "epoch": 1.2697126406141708, "grad_norm": 2.052682876586914, "learning_rate": 5.237000000000001e-06, "loss": 0.0401, "step": 23816 }, { "epoch": 1.2697659540438235, "grad_norm": 1.2356963157653809, "learning_rate": 5.2368e-06, "loss": 0.0118, "step": 23817 }, { "epoch": 1.2698192674734765, "grad_norm": 2.1714248657226562, "learning_rate": 5.236600000000001e-06, "loss": 0.0201, "step": 23818 }, { "epoch": 1.2698725809031295, "grad_norm": 1.2956011295318604, "learning_rate": 5.2364e-06, "loss": 0.0131, "step": 23819 }, { "epoch": 1.2699258943327825, "grad_norm": 1.7450670003890991, "learning_rate": 5.2362e-06, "loss": 0.0186, "step": 23820 }, { "epoch": 1.2699792077624354, "grad_norm": 3.757140874862671, "learning_rate": 5.236e-06, "loss": 0.0253, "step": 23821 }, { "epoch": 1.2700325211920882, "grad_norm": 1.6213915348052979, "learning_rate": 5.235800000000001e-06, "loss": 0.0137, "step": 23822 }, { "epoch": 1.2700858346217412, "grad_norm": 3.3252251148223877, "learning_rate": 5.2356e-06, "loss": 0.0322, "step": 23823 }, { "epoch": 1.2701391480513942, "grad_norm": 1.2592498064041138, "learning_rate": 5.235400000000001e-06, "loss": 0.0229, "step": 23824 }, { "epoch": 1.2701924614810471, "grad_norm": 5.188945293426514, "learning_rate": 5.2352e-06, "loss": 0.0465, "step": 23825 }, { "epoch": 1.2702457749107001, "grad_norm": 1.5635360479354858, "learning_rate": 5.235e-06, "loss": 0.0472, "step": 23826 }, { "epoch": 1.2702990883403529, "grad_norm": 1.18563973903656, "learning_rate": 5.2348e-06, "loss": 0.024, "step": 23827 }, { "epoch": 1.2703524017700059, "grad_norm": 0.8405763506889343, "learning_rate": 5.234600000000001e-06, "loss": 0.0059, "step": 23828 }, { "epoch": 1.2704057151996588, "grad_norm": 0.3972339332103729, "learning_rate": 5.2344e-06, "loss": 0.0047, "step": 23829 }, { "epoch": 1.2704590286293116, "grad_norm": 3.0087647438049316, "learning_rate": 5.234200000000001e-06, "loss": 0.0641, "step": 23830 }, { "epoch": 1.2705123420589646, "grad_norm": 1.5576391220092773, "learning_rate": 5.234e-06, "loss": 0.0265, "step": 23831 }, { "epoch": 1.2705656554886176, "grad_norm": 1.8797050714492798, "learning_rate": 5.2338e-06, "loss": 0.0118, "step": 23832 }, { "epoch": 1.2706189689182705, "grad_norm": 2.3173794746398926, "learning_rate": 5.233600000000001e-06, "loss": 0.0299, "step": 23833 }, { "epoch": 1.2706722823479235, "grad_norm": 1.7717499732971191, "learning_rate": 5.233400000000001e-06, "loss": 0.0139, "step": 23834 }, { "epoch": 1.2707255957775763, "grad_norm": 2.918800115585327, "learning_rate": 5.2332e-06, "loss": 0.0444, "step": 23835 }, { "epoch": 1.2707789092072292, "grad_norm": 2.445744037628174, "learning_rate": 5.2330000000000005e-06, "loss": 0.03, "step": 23836 }, { "epoch": 1.2708322226368822, "grad_norm": 3.0598387718200684, "learning_rate": 5.2328e-06, "loss": 0.0559, "step": 23837 }, { "epoch": 1.2708855360665352, "grad_norm": 1.5272448062896729, "learning_rate": 5.2326e-06, "loss": 0.016, "step": 23838 }, { "epoch": 1.2709388494961882, "grad_norm": 1.164792537689209, "learning_rate": 5.232400000000001e-06, "loss": 0.0108, "step": 23839 }, { "epoch": 1.270992162925841, "grad_norm": 1.6794826984405518, "learning_rate": 5.2322000000000005e-06, "loss": 0.0225, "step": 23840 }, { "epoch": 1.271045476355494, "grad_norm": 2.597684383392334, "learning_rate": 5.232e-06, "loss": 0.0505, "step": 23841 }, { "epoch": 1.271098789785147, "grad_norm": 2.83120059967041, "learning_rate": 5.2318000000000005e-06, "loss": 0.0216, "step": 23842 }, { "epoch": 1.2711521032147999, "grad_norm": 2.4614293575286865, "learning_rate": 5.2316e-06, "loss": 0.0502, "step": 23843 }, { "epoch": 1.2712054166444529, "grad_norm": 2.1696159839630127, "learning_rate": 5.231400000000001e-06, "loss": 0.0315, "step": 23844 }, { "epoch": 1.2712587300741056, "grad_norm": 1.1702699661254883, "learning_rate": 5.231200000000001e-06, "loss": 0.0096, "step": 23845 }, { "epoch": 1.2713120435037586, "grad_norm": 1.841280221939087, "learning_rate": 5.231e-06, "loss": 0.0147, "step": 23846 }, { "epoch": 1.2713653569334116, "grad_norm": 3.018449068069458, "learning_rate": 5.2308e-06, "loss": 0.0477, "step": 23847 }, { "epoch": 1.2714186703630643, "grad_norm": 1.5729107856750488, "learning_rate": 5.2306e-06, "loss": 0.0267, "step": 23848 }, { "epoch": 1.2714719837927173, "grad_norm": 1.0733643770217896, "learning_rate": 5.2304e-06, "loss": 0.0149, "step": 23849 }, { "epoch": 1.2715252972223703, "grad_norm": 2.109572172164917, "learning_rate": 5.230200000000001e-06, "loss": 0.0271, "step": 23850 }, { "epoch": 1.2715786106520233, "grad_norm": 0.9382333159446716, "learning_rate": 5.230000000000001e-06, "loss": 0.0207, "step": 23851 }, { "epoch": 1.2716319240816762, "grad_norm": 0.43339431285858154, "learning_rate": 5.2298e-06, "loss": 0.004, "step": 23852 }, { "epoch": 1.271685237511329, "grad_norm": 0.56166011095047, "learning_rate": 5.2296e-06, "loss": 0.0043, "step": 23853 }, { "epoch": 1.271738550940982, "grad_norm": 1.924686074256897, "learning_rate": 5.2294e-06, "loss": 0.0187, "step": 23854 }, { "epoch": 1.271791864370635, "grad_norm": 2.499264717102051, "learning_rate": 5.229200000000001e-06, "loss": 0.0279, "step": 23855 }, { "epoch": 1.271845177800288, "grad_norm": 1.469913125038147, "learning_rate": 5.229000000000001e-06, "loss": 0.0235, "step": 23856 }, { "epoch": 1.271898491229941, "grad_norm": 2.3970210552215576, "learning_rate": 5.228800000000001e-06, "loss": 0.0123, "step": 23857 }, { "epoch": 1.2719518046595937, "grad_norm": 1.5968672037124634, "learning_rate": 5.2286e-06, "loss": 0.0095, "step": 23858 }, { "epoch": 1.2720051180892467, "grad_norm": 1.3398516178131104, "learning_rate": 5.2284e-06, "loss": 0.0202, "step": 23859 }, { "epoch": 1.2720584315188996, "grad_norm": 1.648999810218811, "learning_rate": 5.2282e-06, "loss": 0.019, "step": 23860 }, { "epoch": 1.2721117449485526, "grad_norm": 1.5902472734451294, "learning_rate": 5.228000000000001e-06, "loss": 0.0106, "step": 23861 }, { "epoch": 1.2721650583782056, "grad_norm": 0.7738056182861328, "learning_rate": 5.227800000000001e-06, "loss": 0.0054, "step": 23862 }, { "epoch": 1.2722183718078584, "grad_norm": 1.094820499420166, "learning_rate": 5.2276000000000006e-06, "loss": 0.0209, "step": 23863 }, { "epoch": 1.2722716852375113, "grad_norm": 1.736170768737793, "learning_rate": 5.2274e-06, "loss": 0.02, "step": 23864 }, { "epoch": 1.2723249986671643, "grad_norm": 2.0004777908325195, "learning_rate": 5.2272e-06, "loss": 0.0345, "step": 23865 }, { "epoch": 1.272378312096817, "grad_norm": 1.4604350328445435, "learning_rate": 5.227000000000001e-06, "loss": 0.0072, "step": 23866 }, { "epoch": 1.27243162552647, "grad_norm": 0.9775598049163818, "learning_rate": 5.2268000000000005e-06, "loss": 0.0106, "step": 23867 }, { "epoch": 1.272484938956123, "grad_norm": 2.440516710281372, "learning_rate": 5.226600000000001e-06, "loss": 0.0297, "step": 23868 }, { "epoch": 1.272538252385776, "grad_norm": 1.8739523887634277, "learning_rate": 5.2264000000000005e-06, "loss": 0.0217, "step": 23869 }, { "epoch": 1.272591565815429, "grad_norm": 1.6586376428604126, "learning_rate": 5.2262e-06, "loss": 0.0103, "step": 23870 }, { "epoch": 1.2726448792450817, "grad_norm": 1.7924748659133911, "learning_rate": 5.226e-06, "loss": 0.0107, "step": 23871 }, { "epoch": 1.2726981926747347, "grad_norm": 2.05488920211792, "learning_rate": 5.225800000000001e-06, "loss": 0.0208, "step": 23872 }, { "epoch": 1.2727515061043877, "grad_norm": 0.32651013135910034, "learning_rate": 5.2256e-06, "loss": 0.0032, "step": 23873 }, { "epoch": 1.2728048195340407, "grad_norm": 3.032426357269287, "learning_rate": 5.225400000000001e-06, "loss": 0.0302, "step": 23874 }, { "epoch": 1.2728581329636937, "grad_norm": 0.6370463967323303, "learning_rate": 5.2252e-06, "loss": 0.0062, "step": 23875 }, { "epoch": 1.2729114463933464, "grad_norm": 1.7465494871139526, "learning_rate": 5.225e-06, "loss": 0.0148, "step": 23876 }, { "epoch": 1.2729647598229994, "grad_norm": 2.550462245941162, "learning_rate": 5.224800000000001e-06, "loss": 0.0332, "step": 23877 }, { "epoch": 1.2730180732526524, "grad_norm": 3.5892322063446045, "learning_rate": 5.224600000000001e-06, "loss": 0.066, "step": 23878 }, { "epoch": 1.2730713866823054, "grad_norm": 1.834904432296753, "learning_rate": 5.2244e-06, "loss": 0.0373, "step": 23879 }, { "epoch": 1.2731247001119583, "grad_norm": 1.8243085145950317, "learning_rate": 5.224200000000001e-06, "loss": 0.026, "step": 23880 }, { "epoch": 1.273178013541611, "grad_norm": 2.322299003601074, "learning_rate": 5.224e-06, "loss": 0.0138, "step": 23881 }, { "epoch": 1.273231326971264, "grad_norm": 2.765742540359497, "learning_rate": 5.2238e-06, "loss": 0.0512, "step": 23882 }, { "epoch": 1.273284640400917, "grad_norm": 1.556984543800354, "learning_rate": 5.223600000000001e-06, "loss": 0.0311, "step": 23883 }, { "epoch": 1.2733379538305698, "grad_norm": 1.3916109800338745, "learning_rate": 5.223400000000001e-06, "loss": 0.0102, "step": 23884 }, { "epoch": 1.2733912672602228, "grad_norm": 1.6153311729431152, "learning_rate": 5.2232e-06, "loss": 0.0152, "step": 23885 }, { "epoch": 1.2734445806898758, "grad_norm": 3.904420852661133, "learning_rate": 5.223000000000001e-06, "loss": 0.0414, "step": 23886 }, { "epoch": 1.2734978941195287, "grad_norm": 2.5703125, "learning_rate": 5.2228e-06, "loss": 0.0371, "step": 23887 }, { "epoch": 1.2735512075491817, "grad_norm": 1.1584818363189697, "learning_rate": 5.2226e-06, "loss": 0.007, "step": 23888 }, { "epoch": 1.2736045209788345, "grad_norm": 1.9212603569030762, "learning_rate": 5.222400000000001e-06, "loss": 0.0441, "step": 23889 }, { "epoch": 1.2736578344084875, "grad_norm": 1.527354121208191, "learning_rate": 5.222200000000001e-06, "loss": 0.0322, "step": 23890 }, { "epoch": 1.2737111478381404, "grad_norm": 2.608994245529175, "learning_rate": 5.222e-06, "loss": 0.0455, "step": 23891 }, { "epoch": 1.2737644612677934, "grad_norm": 1.1556040048599243, "learning_rate": 5.2218000000000006e-06, "loss": 0.0091, "step": 23892 }, { "epoch": 1.2738177746974464, "grad_norm": 4.575758934020996, "learning_rate": 5.2216e-06, "loss": 0.0521, "step": 23893 }, { "epoch": 1.2738710881270991, "grad_norm": 2.597418785095215, "learning_rate": 5.2214000000000005e-06, "loss": 0.0515, "step": 23894 }, { "epoch": 1.2739244015567521, "grad_norm": 2.578207492828369, "learning_rate": 5.221200000000001e-06, "loss": 0.0252, "step": 23895 }, { "epoch": 1.273977714986405, "grad_norm": 4.561751365661621, "learning_rate": 5.2210000000000005e-06, "loss": 0.0425, "step": 23896 }, { "epoch": 1.274031028416058, "grad_norm": 1.9483968019485474, "learning_rate": 5.2208e-06, "loss": 0.0476, "step": 23897 }, { "epoch": 1.274084341845711, "grad_norm": 1.4161282777786255, "learning_rate": 5.2206000000000005e-06, "loss": 0.0095, "step": 23898 }, { "epoch": 1.2741376552753638, "grad_norm": 1.78363835811615, "learning_rate": 5.2204e-06, "loss": 0.0288, "step": 23899 }, { "epoch": 1.2741909687050168, "grad_norm": 2.4683594703674316, "learning_rate": 5.2202000000000004e-06, "loss": 0.0409, "step": 23900 }, { "epoch": 1.2742442821346698, "grad_norm": 2.096456289291382, "learning_rate": 5.220000000000001e-06, "loss": 0.0339, "step": 23901 }, { "epoch": 1.2742975955643225, "grad_norm": 1.2563897371292114, "learning_rate": 5.2198e-06, "loss": 0.0152, "step": 23902 }, { "epoch": 1.2743509089939755, "grad_norm": 1.9676192998886108, "learning_rate": 5.2196e-06, "loss": 0.0268, "step": 23903 }, { "epoch": 1.2744042224236285, "grad_norm": 1.2764620780944824, "learning_rate": 5.2194e-06, "loss": 0.0105, "step": 23904 }, { "epoch": 1.2744575358532815, "grad_norm": 1.7073347568511963, "learning_rate": 5.219200000000001e-06, "loss": 0.0294, "step": 23905 }, { "epoch": 1.2745108492829345, "grad_norm": 2.272351026535034, "learning_rate": 5.219e-06, "loss": 0.0302, "step": 23906 }, { "epoch": 1.2745641627125872, "grad_norm": 0.9059606194496155, "learning_rate": 5.218800000000001e-06, "loss": 0.0089, "step": 23907 }, { "epoch": 1.2746174761422402, "grad_norm": 1.449959635734558, "learning_rate": 5.2186e-06, "loss": 0.0216, "step": 23908 }, { "epoch": 1.2746707895718932, "grad_norm": 0.7837150692939758, "learning_rate": 5.2184e-06, "loss": 0.0078, "step": 23909 }, { "epoch": 1.2747241030015461, "grad_norm": 1.9535129070281982, "learning_rate": 5.2182e-06, "loss": 0.0387, "step": 23910 }, { "epoch": 1.2747774164311991, "grad_norm": 1.4452265501022339, "learning_rate": 5.218000000000001e-06, "loss": 0.0259, "step": 23911 }, { "epoch": 1.2748307298608519, "grad_norm": 0.40618717670440674, "learning_rate": 5.2178e-06, "loss": 0.0048, "step": 23912 }, { "epoch": 1.2748840432905049, "grad_norm": 0.8414338231086731, "learning_rate": 5.217600000000001e-06, "loss": 0.0078, "step": 23913 }, { "epoch": 1.2749373567201578, "grad_norm": 1.772471308708191, "learning_rate": 5.2174e-06, "loss": 0.0567, "step": 23914 }, { "epoch": 1.2749906701498106, "grad_norm": 1.088987112045288, "learning_rate": 5.2172e-06, "loss": 0.024, "step": 23915 }, { "epoch": 1.2750439835794638, "grad_norm": 1.5809873342514038, "learning_rate": 5.217000000000001e-06, "loss": 0.0144, "step": 23916 }, { "epoch": 1.2750972970091166, "grad_norm": 1.7245049476623535, "learning_rate": 5.216800000000001e-06, "loss": 0.0117, "step": 23917 }, { "epoch": 1.2751506104387695, "grad_norm": 1.7393851280212402, "learning_rate": 5.2166e-06, "loss": 0.0392, "step": 23918 }, { "epoch": 1.2752039238684225, "grad_norm": 2.887507677078247, "learning_rate": 5.2164000000000006e-06, "loss": 0.042, "step": 23919 }, { "epoch": 1.2752572372980753, "grad_norm": 1.5467668771743774, "learning_rate": 5.2162e-06, "loss": 0.0354, "step": 23920 }, { "epoch": 1.2753105507277283, "grad_norm": 0.4261002838611603, "learning_rate": 5.216e-06, "loss": 0.0055, "step": 23921 }, { "epoch": 1.2753638641573812, "grad_norm": 2.9291234016418457, "learning_rate": 5.215800000000001e-06, "loss": 0.0171, "step": 23922 }, { "epoch": 1.2754171775870342, "grad_norm": 0.31676536798477173, "learning_rate": 5.2156000000000005e-06, "loss": 0.0039, "step": 23923 }, { "epoch": 1.2754704910166872, "grad_norm": 1.1132889986038208, "learning_rate": 5.2154e-06, "loss": 0.0087, "step": 23924 }, { "epoch": 1.27552380444634, "grad_norm": 1.6796486377716064, "learning_rate": 5.2152000000000005e-06, "loss": 0.0206, "step": 23925 }, { "epoch": 1.275577117875993, "grad_norm": 2.6438844203948975, "learning_rate": 5.215e-06, "loss": 0.0502, "step": 23926 }, { "epoch": 1.275630431305646, "grad_norm": 1.4655530452728271, "learning_rate": 5.214800000000001e-06, "loss": 0.019, "step": 23927 }, { "epoch": 1.2756837447352989, "grad_norm": 2.76523494720459, "learning_rate": 5.214600000000001e-06, "loss": 0.0496, "step": 23928 }, { "epoch": 1.2757370581649519, "grad_norm": 1.5739675760269165, "learning_rate": 5.2144e-06, "loss": 0.0198, "step": 23929 }, { "epoch": 1.2757903715946046, "grad_norm": 1.1917610168457031, "learning_rate": 5.2142e-06, "loss": 0.0145, "step": 23930 }, { "epoch": 1.2758436850242576, "grad_norm": 1.6893420219421387, "learning_rate": 5.214e-06, "loss": 0.0175, "step": 23931 }, { "epoch": 1.2758969984539106, "grad_norm": 1.1923338174819946, "learning_rate": 5.2138e-06, "loss": 0.0211, "step": 23932 }, { "epoch": 1.2759503118835633, "grad_norm": 3.1268656253814697, "learning_rate": 5.213600000000001e-06, "loss": 0.0298, "step": 23933 }, { "epoch": 1.2760036253132163, "grad_norm": 2.6988728046417236, "learning_rate": 5.213400000000001e-06, "loss": 0.0317, "step": 23934 }, { "epoch": 1.2760569387428693, "grad_norm": 1.8666751384735107, "learning_rate": 5.2132e-06, "loss": 0.0104, "step": 23935 }, { "epoch": 1.2761102521725223, "grad_norm": 1.187440276145935, "learning_rate": 5.213e-06, "loss": 0.0167, "step": 23936 }, { "epoch": 1.2761635656021753, "grad_norm": 1.5464534759521484, "learning_rate": 5.2128e-06, "loss": 0.0127, "step": 23937 }, { "epoch": 1.276216879031828, "grad_norm": 2.6068918704986572, "learning_rate": 5.2126e-06, "loss": 0.0334, "step": 23938 }, { "epoch": 1.276270192461481, "grad_norm": 1.8950711488723755, "learning_rate": 5.212400000000001e-06, "loss": 0.0348, "step": 23939 }, { "epoch": 1.276323505891134, "grad_norm": 1.839666485786438, "learning_rate": 5.212200000000001e-06, "loss": 0.0262, "step": 23940 }, { "epoch": 1.276376819320787, "grad_norm": 2.058164596557617, "learning_rate": 5.212e-06, "loss": 0.029, "step": 23941 }, { "epoch": 1.27643013275044, "grad_norm": 1.1556216478347778, "learning_rate": 5.2118e-06, "loss": 0.0139, "step": 23942 }, { "epoch": 1.2764834461800927, "grad_norm": 1.0930454730987549, "learning_rate": 5.2116e-06, "loss": 0.0093, "step": 23943 }, { "epoch": 1.2765367596097457, "grad_norm": 2.1546130180358887, "learning_rate": 5.211400000000001e-06, "loss": 0.0287, "step": 23944 }, { "epoch": 1.2765900730393986, "grad_norm": 1.6238611936569214, "learning_rate": 5.211200000000001e-06, "loss": 0.0238, "step": 23945 }, { "epoch": 1.2766433864690516, "grad_norm": 1.5068943500518799, "learning_rate": 5.211000000000001e-06, "loss": 0.0171, "step": 23946 }, { "epoch": 1.2766966998987046, "grad_norm": 2.2657315731048584, "learning_rate": 5.2108e-06, "loss": 0.0217, "step": 23947 }, { "epoch": 1.2767500133283574, "grad_norm": 1.7281150817871094, "learning_rate": 5.2106e-06, "loss": 0.0373, "step": 23948 }, { "epoch": 1.2768033267580103, "grad_norm": 1.5216848850250244, "learning_rate": 5.2104e-06, "loss": 0.02, "step": 23949 }, { "epoch": 1.2768566401876633, "grad_norm": 1.4419748783111572, "learning_rate": 5.2102000000000005e-06, "loss": 0.0161, "step": 23950 }, { "epoch": 1.276909953617316, "grad_norm": 1.3979206085205078, "learning_rate": 5.210000000000001e-06, "loss": 0.0133, "step": 23951 }, { "epoch": 1.276963267046969, "grad_norm": 0.7570196986198425, "learning_rate": 5.2098000000000005e-06, "loss": 0.0069, "step": 23952 }, { "epoch": 1.277016580476622, "grad_norm": 1.4166673421859741, "learning_rate": 5.2096e-06, "loss": 0.0281, "step": 23953 }, { "epoch": 1.277069893906275, "grad_norm": 1.7553958892822266, "learning_rate": 5.2094e-06, "loss": 0.0316, "step": 23954 }, { "epoch": 1.277123207335928, "grad_norm": 4.13803243637085, "learning_rate": 5.209200000000001e-06, "loss": 0.0295, "step": 23955 }, { "epoch": 1.2771765207655807, "grad_norm": 1.6641170978546143, "learning_rate": 5.2090000000000004e-06, "loss": 0.0241, "step": 23956 }, { "epoch": 1.2772298341952337, "grad_norm": 2.323894500732422, "learning_rate": 5.208800000000001e-06, "loss": 0.0302, "step": 23957 }, { "epoch": 1.2772831476248867, "grad_norm": 0.9508154392242432, "learning_rate": 5.2086e-06, "loss": 0.0193, "step": 23958 }, { "epoch": 1.2773364610545397, "grad_norm": 1.4986859560012817, "learning_rate": 5.2084e-06, "loss": 0.0181, "step": 23959 }, { "epoch": 1.2773897744841927, "grad_norm": 1.5171440839767456, "learning_rate": 5.2081999999999995e-06, "loss": 0.0303, "step": 23960 }, { "epoch": 1.2774430879138454, "grad_norm": 1.4039613008499146, "learning_rate": 5.208000000000001e-06, "loss": 0.0494, "step": 23961 }, { "epoch": 1.2774964013434984, "grad_norm": 1.6176649332046509, "learning_rate": 5.2078e-06, "loss": 0.0173, "step": 23962 }, { "epoch": 1.2775497147731514, "grad_norm": 1.939328670501709, "learning_rate": 5.207600000000001e-06, "loss": 0.0328, "step": 23963 }, { "epoch": 1.2776030282028044, "grad_norm": 1.5538265705108643, "learning_rate": 5.2074e-06, "loss": 0.0247, "step": 23964 }, { "epoch": 1.2776563416324573, "grad_norm": 0.8863388895988464, "learning_rate": 5.2072e-06, "loss": 0.011, "step": 23965 }, { "epoch": 1.27770965506211, "grad_norm": 0.6561778783798218, "learning_rate": 5.207000000000001e-06, "loss": 0.0056, "step": 23966 }, { "epoch": 1.277762968491763, "grad_norm": 2.5172786712646484, "learning_rate": 5.206800000000001e-06, "loss": 0.0264, "step": 23967 }, { "epoch": 1.277816281921416, "grad_norm": 1.6416083574295044, "learning_rate": 5.2066e-06, "loss": 0.0328, "step": 23968 }, { "epoch": 1.2778695953510688, "grad_norm": 2.8168394565582275, "learning_rate": 5.206400000000001e-06, "loss": 0.0249, "step": 23969 }, { "epoch": 1.2779229087807218, "grad_norm": 3.810298442840576, "learning_rate": 5.2062e-06, "loss": 0.0309, "step": 23970 }, { "epoch": 1.2779762222103748, "grad_norm": 1.4941216707229614, "learning_rate": 5.206e-06, "loss": 0.0142, "step": 23971 }, { "epoch": 1.2780295356400277, "grad_norm": 2.279250383377075, "learning_rate": 5.205800000000001e-06, "loss": 0.0359, "step": 23972 }, { "epoch": 1.2780828490696807, "grad_norm": 2.9258081912994385, "learning_rate": 5.205600000000001e-06, "loss": 0.0535, "step": 23973 }, { "epoch": 1.2781361624993335, "grad_norm": 2.5076794624328613, "learning_rate": 5.2054e-06, "loss": 0.0337, "step": 23974 }, { "epoch": 1.2781894759289865, "grad_norm": 2.0271825790405273, "learning_rate": 5.205200000000001e-06, "loss": 0.0338, "step": 23975 }, { "epoch": 1.2782427893586394, "grad_norm": 1.9054008722305298, "learning_rate": 5.205e-06, "loss": 0.0273, "step": 23976 }, { "epoch": 1.2782961027882924, "grad_norm": 1.7621668577194214, "learning_rate": 5.2048000000000005e-06, "loss": 0.015, "step": 23977 }, { "epoch": 1.2783494162179454, "grad_norm": 0.9601396322250366, "learning_rate": 5.204600000000001e-06, "loss": 0.0204, "step": 23978 }, { "epoch": 1.2784027296475982, "grad_norm": 1.8185415267944336, "learning_rate": 5.2044000000000005e-06, "loss": 0.0195, "step": 23979 }, { "epoch": 1.2784560430772511, "grad_norm": 2.0999300479888916, "learning_rate": 5.2042e-06, "loss": 0.0208, "step": 23980 }, { "epoch": 1.2785093565069041, "grad_norm": 1.9042640924453735, "learning_rate": 5.2040000000000005e-06, "loss": 0.0296, "step": 23981 }, { "epoch": 1.278562669936557, "grad_norm": 1.60841703414917, "learning_rate": 5.2038e-06, "loss": 0.026, "step": 23982 }, { "epoch": 1.27861598336621, "grad_norm": 0.5879234075546265, "learning_rate": 5.2036000000000005e-06, "loss": 0.0055, "step": 23983 }, { "epoch": 1.2786692967958628, "grad_norm": 2.3682427406311035, "learning_rate": 5.203400000000001e-06, "loss": 0.0174, "step": 23984 }, { "epoch": 1.2787226102255158, "grad_norm": 1.3871026039123535, "learning_rate": 5.2032000000000004e-06, "loss": 0.0232, "step": 23985 }, { "epoch": 1.2787759236551688, "grad_norm": 1.966505527496338, "learning_rate": 5.203e-06, "loss": 0.0349, "step": 23986 }, { "epoch": 1.2788292370848215, "grad_norm": 2.0314722061157227, "learning_rate": 5.2028e-06, "loss": 0.0322, "step": 23987 }, { "epoch": 1.2788825505144745, "grad_norm": 2.602365255355835, "learning_rate": 5.2026e-06, "loss": 0.0309, "step": 23988 }, { "epoch": 1.2789358639441275, "grad_norm": 1.2167726755142212, "learning_rate": 5.2024e-06, "loss": 0.0337, "step": 23989 }, { "epoch": 1.2789891773737805, "grad_norm": 1.8276156187057495, "learning_rate": 5.202200000000001e-06, "loss": 0.0179, "step": 23990 }, { "epoch": 1.2790424908034335, "grad_norm": 1.9652856588363647, "learning_rate": 5.202e-06, "loss": 0.027, "step": 23991 }, { "epoch": 1.2790958042330862, "grad_norm": 1.818933129310608, "learning_rate": 5.2018e-06, "loss": 0.0384, "step": 23992 }, { "epoch": 1.2791491176627392, "grad_norm": 1.7142488956451416, "learning_rate": 5.2016e-06, "loss": 0.0337, "step": 23993 }, { "epoch": 1.2792024310923922, "grad_norm": 2.043437957763672, "learning_rate": 5.201400000000001e-06, "loss": 0.0438, "step": 23994 }, { "epoch": 1.2792557445220452, "grad_norm": 2.865835428237915, "learning_rate": 5.2012e-06, "loss": 0.0451, "step": 23995 }, { "epoch": 1.2793090579516981, "grad_norm": 1.5285992622375488, "learning_rate": 5.201000000000001e-06, "loss": 0.0153, "step": 23996 }, { "epoch": 1.279362371381351, "grad_norm": 1.685421347618103, "learning_rate": 5.2008e-06, "loss": 0.0302, "step": 23997 }, { "epoch": 1.2794156848110039, "grad_norm": 1.5891706943511963, "learning_rate": 5.2006e-06, "loss": 0.0263, "step": 23998 }, { "epoch": 1.2794689982406569, "grad_norm": 5.558125972747803, "learning_rate": 5.2004e-06, "loss": 0.0843, "step": 23999 }, { "epoch": 1.2795223116703098, "grad_norm": 2.1351444721221924, "learning_rate": 5.200200000000001e-06, "loss": 0.0197, "step": 24000 }, { "epoch": 1.2795756250999628, "grad_norm": 1.5495142936706543, "learning_rate": 5.2e-06, "loss": 0.0189, "step": 24001 }, { "epoch": 1.2796289385296156, "grad_norm": 1.8659266233444214, "learning_rate": 5.199800000000001e-06, "loss": 0.0167, "step": 24002 }, { "epoch": 1.2796822519592685, "grad_norm": 1.721742033958435, "learning_rate": 5.1996e-06, "loss": 0.0211, "step": 24003 }, { "epoch": 1.2797355653889215, "grad_norm": 1.962750792503357, "learning_rate": 5.1994e-06, "loss": 0.0209, "step": 24004 }, { "epoch": 1.2797888788185743, "grad_norm": 2.1732523441314697, "learning_rate": 5.199200000000001e-06, "loss": 0.0241, "step": 24005 }, { "epoch": 1.2798421922482273, "grad_norm": 1.1609469652175903, "learning_rate": 5.1990000000000005e-06, "loss": 0.0183, "step": 24006 }, { "epoch": 1.2798955056778802, "grad_norm": 1.6877288818359375, "learning_rate": 5.1988e-06, "loss": 0.0167, "step": 24007 }, { "epoch": 1.2799488191075332, "grad_norm": 2.4857659339904785, "learning_rate": 5.1986000000000005e-06, "loss": 0.024, "step": 24008 }, { "epoch": 1.2800021325371862, "grad_norm": 1.5703496932983398, "learning_rate": 5.1984e-06, "loss": 0.037, "step": 24009 }, { "epoch": 1.280055445966839, "grad_norm": 2.1373867988586426, "learning_rate": 5.1982e-06, "loss": 0.0382, "step": 24010 }, { "epoch": 1.280108759396492, "grad_norm": 1.926303505897522, "learning_rate": 5.198000000000001e-06, "loss": 0.0144, "step": 24011 }, { "epoch": 1.280162072826145, "grad_norm": 1.0413916110992432, "learning_rate": 5.1978000000000004e-06, "loss": 0.0219, "step": 24012 }, { "epoch": 1.280215386255798, "grad_norm": 2.0356879234313965, "learning_rate": 5.1976e-06, "loss": 0.049, "step": 24013 }, { "epoch": 1.2802686996854509, "grad_norm": 5.60147762298584, "learning_rate": 5.1974e-06, "loss": 0.035, "step": 24014 }, { "epoch": 1.2803220131151036, "grad_norm": 3.0224082469940186, "learning_rate": 5.1972e-06, "loss": 0.0459, "step": 24015 }, { "epoch": 1.2803753265447566, "grad_norm": 0.9419175386428833, "learning_rate": 5.197000000000001e-06, "loss": 0.0203, "step": 24016 }, { "epoch": 1.2804286399744096, "grad_norm": 1.110853910446167, "learning_rate": 5.196800000000001e-06, "loss": 0.0063, "step": 24017 }, { "epoch": 1.2804819534040623, "grad_norm": 2.132744789123535, "learning_rate": 5.1966e-06, "loss": 0.0389, "step": 24018 }, { "epoch": 1.2805352668337155, "grad_norm": 1.712384581565857, "learning_rate": 5.1964e-06, "loss": 0.0331, "step": 24019 }, { "epoch": 1.2805885802633683, "grad_norm": 1.4944292306900024, "learning_rate": 5.1962e-06, "loss": 0.0444, "step": 24020 }, { "epoch": 1.2806418936930213, "grad_norm": 1.6291534900665283, "learning_rate": 5.196e-06, "loss": 0.0296, "step": 24021 }, { "epoch": 1.2806952071226743, "grad_norm": 1.5701824426651, "learning_rate": 5.195800000000001e-06, "loss": 0.0412, "step": 24022 }, { "epoch": 1.280748520552327, "grad_norm": 1.6547110080718994, "learning_rate": 5.195600000000001e-06, "loss": 0.019, "step": 24023 }, { "epoch": 1.28080183398198, "grad_norm": 0.9985641837120056, "learning_rate": 5.1954e-06, "loss": 0.0128, "step": 24024 }, { "epoch": 1.280855147411633, "grad_norm": 0.5333441495895386, "learning_rate": 5.1952e-06, "loss": 0.004, "step": 24025 }, { "epoch": 1.280908460841286, "grad_norm": 2.2145097255706787, "learning_rate": 5.195e-06, "loss": 0.0433, "step": 24026 }, { "epoch": 1.280961774270939, "grad_norm": 1.359688639640808, "learning_rate": 5.194800000000001e-06, "loss": 0.045, "step": 24027 }, { "epoch": 1.2810150877005917, "grad_norm": 1.81181001663208, "learning_rate": 5.194600000000001e-06, "loss": 0.0367, "step": 24028 }, { "epoch": 1.2810684011302447, "grad_norm": 5.875814437866211, "learning_rate": 5.194400000000001e-06, "loss": 0.0608, "step": 24029 }, { "epoch": 1.2811217145598977, "grad_norm": 1.8310612440109253, "learning_rate": 5.1942e-06, "loss": 0.024, "step": 24030 }, { "epoch": 1.2811750279895506, "grad_norm": 0.976153552532196, "learning_rate": 5.194e-06, "loss": 0.0123, "step": 24031 }, { "epoch": 1.2812283414192036, "grad_norm": 2.168269395828247, "learning_rate": 5.1938e-06, "loss": 0.0338, "step": 24032 }, { "epoch": 1.2812816548488564, "grad_norm": 1.2615699768066406, "learning_rate": 5.1936000000000006e-06, "loss": 0.0169, "step": 24033 }, { "epoch": 1.2813349682785093, "grad_norm": 1.3670036792755127, "learning_rate": 5.193400000000001e-06, "loss": 0.0358, "step": 24034 }, { "epoch": 1.2813882817081623, "grad_norm": 1.4740657806396484, "learning_rate": 5.1932000000000005e-06, "loss": 0.0144, "step": 24035 }, { "epoch": 1.281441595137815, "grad_norm": 2.03296160697937, "learning_rate": 5.193e-06, "loss": 0.024, "step": 24036 }, { "epoch": 1.2814949085674683, "grad_norm": 1.1073963642120361, "learning_rate": 5.1928e-06, "loss": 0.0214, "step": 24037 }, { "epoch": 1.281548221997121, "grad_norm": 0.9915016889572144, "learning_rate": 5.192600000000001e-06, "loss": 0.021, "step": 24038 }, { "epoch": 1.281601535426774, "grad_norm": 1.4152722358703613, "learning_rate": 5.1924000000000005e-06, "loss": 0.015, "step": 24039 }, { "epoch": 1.281654848856427, "grad_norm": 1.0691591501235962, "learning_rate": 5.192200000000001e-06, "loss": 0.0352, "step": 24040 }, { "epoch": 1.2817081622860798, "grad_norm": 1.5411401987075806, "learning_rate": 5.1920000000000004e-06, "loss": 0.0173, "step": 24041 }, { "epoch": 1.2817614757157327, "grad_norm": 1.821065068244934, "learning_rate": 5.1918e-06, "loss": 0.0287, "step": 24042 }, { "epoch": 1.2818147891453857, "grad_norm": 0.8184970021247864, "learning_rate": 5.1915999999999996e-06, "loss": 0.0096, "step": 24043 }, { "epoch": 1.2818681025750387, "grad_norm": 2.0289621353149414, "learning_rate": 5.191400000000001e-06, "loss": 0.0224, "step": 24044 }, { "epoch": 1.2819214160046917, "grad_norm": 1.6520254611968994, "learning_rate": 5.1912e-06, "loss": 0.0267, "step": 24045 }, { "epoch": 1.2819747294343444, "grad_norm": 2.3166275024414062, "learning_rate": 5.191000000000001e-06, "loss": 0.055, "step": 24046 }, { "epoch": 1.2820280428639974, "grad_norm": 1.2639917135238647, "learning_rate": 5.1908e-06, "loss": 0.0282, "step": 24047 }, { "epoch": 1.2820813562936504, "grad_norm": 1.65950608253479, "learning_rate": 5.1906e-06, "loss": 0.0187, "step": 24048 }, { "epoch": 1.2821346697233034, "grad_norm": 2.127800941467285, "learning_rate": 5.1903999999999995e-06, "loss": 0.0467, "step": 24049 }, { "epoch": 1.2821879831529563, "grad_norm": 0.8220576643943787, "learning_rate": 5.190200000000001e-06, "loss": 0.023, "step": 24050 }, { "epoch": 1.282241296582609, "grad_norm": 2.710526943206787, "learning_rate": 5.19e-06, "loss": 0.0255, "step": 24051 }, { "epoch": 1.282294610012262, "grad_norm": 0.8454148769378662, "learning_rate": 5.189800000000001e-06, "loss": 0.0095, "step": 24052 }, { "epoch": 1.282347923441915, "grad_norm": 1.9199957847595215, "learning_rate": 5.1896e-06, "loss": 0.0223, "step": 24053 }, { "epoch": 1.2824012368715678, "grad_norm": 1.3066296577453613, "learning_rate": 5.1894e-06, "loss": 0.0141, "step": 24054 }, { "epoch": 1.2824545503012208, "grad_norm": 1.2079585790634155, "learning_rate": 5.189200000000001e-06, "loss": 0.0267, "step": 24055 }, { "epoch": 1.2825078637308738, "grad_norm": 1.6997830867767334, "learning_rate": 5.189000000000001e-06, "loss": 0.0166, "step": 24056 }, { "epoch": 1.2825611771605268, "grad_norm": 2.136570453643799, "learning_rate": 5.1888e-06, "loss": 0.0451, "step": 24057 }, { "epoch": 1.2826144905901797, "grad_norm": 0.9412728548049927, "learning_rate": 5.188600000000001e-06, "loss": 0.0102, "step": 24058 }, { "epoch": 1.2826678040198325, "grad_norm": 1.7914338111877441, "learning_rate": 5.1884e-06, "loss": 0.0121, "step": 24059 }, { "epoch": 1.2827211174494855, "grad_norm": 0.798591136932373, "learning_rate": 5.1882e-06, "loss": 0.0083, "step": 24060 }, { "epoch": 1.2827744308791384, "grad_norm": 3.224520683288574, "learning_rate": 5.188000000000001e-06, "loss": 0.0215, "step": 24061 }, { "epoch": 1.2828277443087914, "grad_norm": 2.366271495819092, "learning_rate": 5.1878000000000005e-06, "loss": 0.0295, "step": 24062 }, { "epoch": 1.2828810577384444, "grad_norm": 2.4214165210723877, "learning_rate": 5.1876e-06, "loss": 0.0346, "step": 24063 }, { "epoch": 1.2829343711680972, "grad_norm": 0.5964232087135315, "learning_rate": 5.1874000000000005e-06, "loss": 0.0059, "step": 24064 }, { "epoch": 1.2829876845977501, "grad_norm": 2.3523125648498535, "learning_rate": 5.1872e-06, "loss": 0.0531, "step": 24065 }, { "epoch": 1.2830409980274031, "grad_norm": 2.608696222305298, "learning_rate": 5.187000000000001e-06, "loss": 0.0223, "step": 24066 }, { "epoch": 1.283094311457056, "grad_norm": 2.3299944400787354, "learning_rate": 5.186800000000001e-06, "loss": 0.0223, "step": 24067 }, { "epoch": 1.283147624886709, "grad_norm": 1.235066294670105, "learning_rate": 5.1866000000000005e-06, "loss": 0.0082, "step": 24068 }, { "epoch": 1.2832009383163618, "grad_norm": 2.465532064437866, "learning_rate": 5.1864e-06, "loss": 0.0351, "step": 24069 }, { "epoch": 1.2832542517460148, "grad_norm": 2.3778162002563477, "learning_rate": 5.1862e-06, "loss": 0.0163, "step": 24070 }, { "epoch": 1.2833075651756678, "grad_norm": 2.711007595062256, "learning_rate": 5.186e-06, "loss": 0.0591, "step": 24071 }, { "epoch": 1.2833608786053206, "grad_norm": 2.5281949043273926, "learning_rate": 5.185800000000001e-06, "loss": 0.0297, "step": 24072 }, { "epoch": 1.2834141920349735, "grad_norm": 0.9856096506118774, "learning_rate": 5.185600000000001e-06, "loss": 0.0083, "step": 24073 }, { "epoch": 1.2834675054646265, "grad_norm": 2.611851930618286, "learning_rate": 5.1854e-06, "loss": 0.0147, "step": 24074 }, { "epoch": 1.2835208188942795, "grad_norm": 3.1540238857269287, "learning_rate": 5.1852e-06, "loss": 0.0133, "step": 24075 }, { "epoch": 1.2835741323239325, "grad_norm": 2.649845838546753, "learning_rate": 5.185e-06, "loss": 0.0331, "step": 24076 }, { "epoch": 1.2836274457535852, "grad_norm": 2.9189558029174805, "learning_rate": 5.184800000000001e-06, "loss": 0.0539, "step": 24077 }, { "epoch": 1.2836807591832382, "grad_norm": 1.3256677389144897, "learning_rate": 5.184600000000001e-06, "loss": 0.0148, "step": 24078 }, { "epoch": 1.2837340726128912, "grad_norm": 1.1472173929214478, "learning_rate": 5.184400000000001e-06, "loss": 0.0103, "step": 24079 }, { "epoch": 1.2837873860425442, "grad_norm": 0.6489180326461792, "learning_rate": 5.1842e-06, "loss": 0.0071, "step": 24080 }, { "epoch": 1.2838406994721971, "grad_norm": 0.9521132111549377, "learning_rate": 5.184e-06, "loss": 0.0164, "step": 24081 }, { "epoch": 1.28389401290185, "grad_norm": 0.8523508310317993, "learning_rate": 5.1838e-06, "loss": 0.0084, "step": 24082 }, { "epoch": 1.2839473263315029, "grad_norm": 1.9075713157653809, "learning_rate": 5.183600000000001e-06, "loss": 0.0405, "step": 24083 }, { "epoch": 1.2840006397611559, "grad_norm": 0.31356576085090637, "learning_rate": 5.183400000000001e-06, "loss": 0.0034, "step": 24084 }, { "epoch": 1.2840539531908088, "grad_norm": 1.0249061584472656, "learning_rate": 5.183200000000001e-06, "loss": 0.0074, "step": 24085 }, { "epoch": 1.2841072666204618, "grad_norm": 1.4173794984817505, "learning_rate": 5.183e-06, "loss": 0.0242, "step": 24086 }, { "epoch": 1.2841605800501146, "grad_norm": 0.881632924079895, "learning_rate": 5.1828e-06, "loss": 0.0171, "step": 24087 }, { "epoch": 1.2842138934797676, "grad_norm": 1.411104679107666, "learning_rate": 5.182600000000001e-06, "loss": 0.0185, "step": 24088 }, { "epoch": 1.2842672069094205, "grad_norm": 1.7611393928527832, "learning_rate": 5.1824000000000006e-06, "loss": 0.0091, "step": 24089 }, { "epoch": 1.2843205203390733, "grad_norm": 2.347071409225464, "learning_rate": 5.182200000000001e-06, "loss": 0.0303, "step": 24090 }, { "epoch": 1.2843738337687263, "grad_norm": 1.3650026321411133, "learning_rate": 5.1820000000000005e-06, "loss": 0.0318, "step": 24091 }, { "epoch": 1.2844271471983792, "grad_norm": 2.840897798538208, "learning_rate": 5.1818e-06, "loss": 0.0305, "step": 24092 }, { "epoch": 1.2844804606280322, "grad_norm": 2.4048585891723633, "learning_rate": 5.1816e-06, "loss": 0.0428, "step": 24093 }, { "epoch": 1.2845337740576852, "grad_norm": 3.0047037601470947, "learning_rate": 5.181400000000001e-06, "loss": 0.0212, "step": 24094 }, { "epoch": 1.284587087487338, "grad_norm": 2.172447443008423, "learning_rate": 5.1812000000000005e-06, "loss": 0.0238, "step": 24095 }, { "epoch": 1.284640400916991, "grad_norm": 4.0542168617248535, "learning_rate": 5.181000000000001e-06, "loss": 0.0195, "step": 24096 }, { "epoch": 1.284693714346644, "grad_norm": 3.1425201892852783, "learning_rate": 5.1808000000000004e-06, "loss": 0.0331, "step": 24097 }, { "epoch": 1.284747027776297, "grad_norm": 1.8866565227508545, "learning_rate": 5.1806e-06, "loss": 0.0271, "step": 24098 }, { "epoch": 1.2848003412059499, "grad_norm": 3.6084961891174316, "learning_rate": 5.1803999999999996e-06, "loss": 0.0622, "step": 24099 }, { "epoch": 1.2848536546356026, "grad_norm": 1.4162888526916504, "learning_rate": 5.180200000000001e-06, "loss": 0.0327, "step": 24100 }, { "epoch": 1.2849069680652556, "grad_norm": 2.277409076690674, "learning_rate": 5.18e-06, "loss": 0.0153, "step": 24101 }, { "epoch": 1.2849602814949086, "grad_norm": 1.1404508352279663, "learning_rate": 5.179800000000001e-06, "loss": 0.027, "step": 24102 }, { "epoch": 1.2850135949245616, "grad_norm": 2.5036544799804688, "learning_rate": 5.1796e-06, "loss": 0.0239, "step": 24103 }, { "epoch": 1.2850669083542146, "grad_norm": 0.911605954170227, "learning_rate": 5.1794e-06, "loss": 0.0097, "step": 24104 }, { "epoch": 1.2851202217838673, "grad_norm": 1.8413825035095215, "learning_rate": 5.179200000000001e-06, "loss": 0.0301, "step": 24105 }, { "epoch": 1.2851735352135203, "grad_norm": 2.981215000152588, "learning_rate": 5.179000000000001e-06, "loss": 0.0218, "step": 24106 }, { "epoch": 1.2852268486431733, "grad_norm": 0.2576053738594055, "learning_rate": 5.1788e-06, "loss": 0.0028, "step": 24107 }, { "epoch": 1.285280162072826, "grad_norm": 1.5856298208236694, "learning_rate": 5.178600000000001e-06, "loss": 0.0229, "step": 24108 }, { "epoch": 1.285333475502479, "grad_norm": 2.320136070251465, "learning_rate": 5.1784e-06, "loss": 0.0401, "step": 24109 }, { "epoch": 1.285386788932132, "grad_norm": 1.0063589811325073, "learning_rate": 5.1782e-06, "loss": 0.0172, "step": 24110 }, { "epoch": 1.285440102361785, "grad_norm": 2.079256772994995, "learning_rate": 5.178000000000001e-06, "loss": 0.0421, "step": 24111 }, { "epoch": 1.285493415791438, "grad_norm": 2.6143786907196045, "learning_rate": 5.177800000000001e-06, "loss": 0.0281, "step": 24112 }, { "epoch": 1.2855467292210907, "grad_norm": 1.4927151203155518, "learning_rate": 5.1776e-06, "loss": 0.0406, "step": 24113 }, { "epoch": 1.2856000426507437, "grad_norm": 1.123278021812439, "learning_rate": 5.177400000000001e-06, "loss": 0.0176, "step": 24114 }, { "epoch": 1.2856533560803967, "grad_norm": 1.4810943603515625, "learning_rate": 5.1772e-06, "loss": 0.0374, "step": 24115 }, { "epoch": 1.2857066695100496, "grad_norm": 2.4961674213409424, "learning_rate": 5.177000000000001e-06, "loss": 0.054, "step": 24116 }, { "epoch": 1.2857599829397026, "grad_norm": 1.4028828144073486, "learning_rate": 5.176800000000001e-06, "loss": 0.0117, "step": 24117 }, { "epoch": 1.2858132963693554, "grad_norm": 2.464921236038208, "learning_rate": 5.1766000000000005e-06, "loss": 0.0586, "step": 24118 }, { "epoch": 1.2858666097990084, "grad_norm": 2.086448907852173, "learning_rate": 5.1764e-06, "loss": 0.0313, "step": 24119 }, { "epoch": 1.2859199232286613, "grad_norm": 0.9775481224060059, "learning_rate": 5.1762000000000005e-06, "loss": 0.0203, "step": 24120 }, { "epoch": 1.2859732366583143, "grad_norm": 2.0481090545654297, "learning_rate": 5.176e-06, "loss": 0.0149, "step": 24121 }, { "epoch": 1.2860265500879673, "grad_norm": 1.1297930479049683, "learning_rate": 5.1758000000000005e-06, "loss": 0.0185, "step": 24122 }, { "epoch": 1.28607986351762, "grad_norm": 0.9654023051261902, "learning_rate": 5.175600000000001e-06, "loss": 0.0221, "step": 24123 }, { "epoch": 1.286133176947273, "grad_norm": 1.1394984722137451, "learning_rate": 5.1754000000000005e-06, "loss": 0.0143, "step": 24124 }, { "epoch": 1.286186490376926, "grad_norm": 1.1003762483596802, "learning_rate": 5.1752e-06, "loss": 0.0263, "step": 24125 }, { "epoch": 1.2862398038065788, "grad_norm": 1.4195725917816162, "learning_rate": 5.1750000000000004e-06, "loss": 0.0178, "step": 24126 }, { "epoch": 1.2862931172362317, "grad_norm": 2.3035225868225098, "learning_rate": 5.174800000000001e-06, "loss": 0.0591, "step": 24127 }, { "epoch": 1.2863464306658847, "grad_norm": 1.5338064432144165, "learning_rate": 5.1746e-06, "loss": 0.0224, "step": 24128 }, { "epoch": 1.2863997440955377, "grad_norm": 1.0416316986083984, "learning_rate": 5.174400000000001e-06, "loss": 0.0153, "step": 24129 }, { "epoch": 1.2864530575251907, "grad_norm": 1.310748815536499, "learning_rate": 5.1742e-06, "loss": 0.0231, "step": 24130 }, { "epoch": 1.2865063709548434, "grad_norm": 0.941278338432312, "learning_rate": 5.174e-06, "loss": 0.0092, "step": 24131 }, { "epoch": 1.2865596843844964, "grad_norm": 1.1182758808135986, "learning_rate": 5.1738e-06, "loss": 0.0122, "step": 24132 }, { "epoch": 1.2866129978141494, "grad_norm": 1.8082396984100342, "learning_rate": 5.173600000000001e-06, "loss": 0.0343, "step": 24133 }, { "epoch": 1.2866663112438024, "grad_norm": 2.027573823928833, "learning_rate": 5.1734e-06, "loss": 0.0253, "step": 24134 }, { "epoch": 1.2867196246734554, "grad_norm": 0.6943413019180298, "learning_rate": 5.173200000000001e-06, "loss": 0.0088, "step": 24135 }, { "epoch": 1.286772938103108, "grad_norm": 1.554966926574707, "learning_rate": 5.173e-06, "loss": 0.0111, "step": 24136 }, { "epoch": 1.286826251532761, "grad_norm": 1.87021005153656, "learning_rate": 5.1728e-06, "loss": 0.0125, "step": 24137 }, { "epoch": 1.286879564962414, "grad_norm": 1.6633700132369995, "learning_rate": 5.172600000000001e-06, "loss": 0.0283, "step": 24138 }, { "epoch": 1.2869328783920668, "grad_norm": 4.104588508605957, "learning_rate": 5.172400000000001e-06, "loss": 0.026, "step": 24139 }, { "epoch": 1.28698619182172, "grad_norm": 2.9588840007781982, "learning_rate": 5.1722e-06, "loss": 0.0594, "step": 24140 }, { "epoch": 1.2870395052513728, "grad_norm": 1.817427396774292, "learning_rate": 5.172000000000001e-06, "loss": 0.0375, "step": 24141 }, { "epoch": 1.2870928186810258, "grad_norm": 1.7089991569519043, "learning_rate": 5.1718e-06, "loss": 0.0119, "step": 24142 }, { "epoch": 1.2871461321106787, "grad_norm": 0.9214848279953003, "learning_rate": 5.1716e-06, "loss": 0.0192, "step": 24143 }, { "epoch": 1.2871994455403315, "grad_norm": 1.6950825452804565, "learning_rate": 5.171400000000001e-06, "loss": 0.0228, "step": 24144 }, { "epoch": 1.2872527589699845, "grad_norm": 2.5103554725646973, "learning_rate": 5.1712000000000006e-06, "loss": 0.0309, "step": 24145 }, { "epoch": 1.2873060723996375, "grad_norm": 1.1819332838058472, "learning_rate": 5.171e-06, "loss": 0.0136, "step": 24146 }, { "epoch": 1.2873593858292904, "grad_norm": 0.6001232862472534, "learning_rate": 5.1708000000000005e-06, "loss": 0.0063, "step": 24147 }, { "epoch": 1.2874126992589434, "grad_norm": 2.3117330074310303, "learning_rate": 5.1706e-06, "loss": 0.0191, "step": 24148 }, { "epoch": 1.2874660126885962, "grad_norm": 1.6007533073425293, "learning_rate": 5.170400000000001e-06, "loss": 0.0268, "step": 24149 }, { "epoch": 1.2875193261182492, "grad_norm": 0.3860693871974945, "learning_rate": 5.170200000000001e-06, "loss": 0.0035, "step": 24150 }, { "epoch": 1.2875726395479021, "grad_norm": 1.7605140209197998, "learning_rate": 5.1700000000000005e-06, "loss": 0.0168, "step": 24151 }, { "epoch": 1.287625952977555, "grad_norm": 1.2577275037765503, "learning_rate": 5.1698e-06, "loss": 0.0195, "step": 24152 }, { "epoch": 1.287679266407208, "grad_norm": 1.8249589204788208, "learning_rate": 5.1696000000000004e-06, "loss": 0.0545, "step": 24153 }, { "epoch": 1.2877325798368608, "grad_norm": 2.8280951976776123, "learning_rate": 5.1694e-06, "loss": 0.0281, "step": 24154 }, { "epoch": 1.2877858932665138, "grad_norm": 1.6481788158416748, "learning_rate": 5.169200000000001e-06, "loss": 0.0188, "step": 24155 }, { "epoch": 1.2878392066961668, "grad_norm": 2.231239080429077, "learning_rate": 5.169000000000001e-06, "loss": 0.025, "step": 24156 }, { "epoch": 1.2878925201258196, "grad_norm": 1.181973934173584, "learning_rate": 5.1688e-06, "loss": 0.0154, "step": 24157 }, { "epoch": 1.2879458335554725, "grad_norm": 1.167422890663147, "learning_rate": 5.1686e-06, "loss": 0.0197, "step": 24158 }, { "epoch": 1.2879991469851255, "grad_norm": 1.237762451171875, "learning_rate": 5.1684e-06, "loss": 0.0071, "step": 24159 }, { "epoch": 1.2880524604147785, "grad_norm": 3.3004701137542725, "learning_rate": 5.1682e-06, "loss": 0.0329, "step": 24160 }, { "epoch": 1.2881057738444315, "grad_norm": 2.777263641357422, "learning_rate": 5.168000000000001e-06, "loss": 0.0191, "step": 24161 }, { "epoch": 1.2881590872740842, "grad_norm": 0.7989285588264465, "learning_rate": 5.167800000000001e-06, "loss": 0.0086, "step": 24162 }, { "epoch": 1.2882124007037372, "grad_norm": 0.9162140488624573, "learning_rate": 5.1676e-06, "loss": 0.0095, "step": 24163 }, { "epoch": 1.2882657141333902, "grad_norm": 2.3262288570404053, "learning_rate": 5.1674e-06, "loss": 0.0392, "step": 24164 }, { "epoch": 1.2883190275630432, "grad_norm": 0.9858334064483643, "learning_rate": 5.1672e-06, "loss": 0.0168, "step": 24165 }, { "epoch": 1.2883723409926962, "grad_norm": 1.5389363765716553, "learning_rate": 5.167000000000001e-06, "loss": 0.0184, "step": 24166 }, { "epoch": 1.288425654422349, "grad_norm": 1.2965164184570312, "learning_rate": 5.166800000000001e-06, "loss": 0.0213, "step": 24167 }, { "epoch": 1.2884789678520019, "grad_norm": 1.8913283348083496, "learning_rate": 5.166600000000001e-06, "loss": 0.042, "step": 24168 }, { "epoch": 1.2885322812816549, "grad_norm": 1.1903350353240967, "learning_rate": 5.1664e-06, "loss": 0.0183, "step": 24169 }, { "epoch": 1.2885855947113078, "grad_norm": 1.8119302988052368, "learning_rate": 5.1662e-06, "loss": 0.0294, "step": 24170 }, { "epoch": 1.2886389081409608, "grad_norm": 1.835839867591858, "learning_rate": 5.166e-06, "loss": 0.0121, "step": 24171 }, { "epoch": 1.2886922215706136, "grad_norm": 1.5436627864837646, "learning_rate": 5.165800000000001e-06, "loss": 0.0314, "step": 24172 }, { "epoch": 1.2887455350002666, "grad_norm": 0.8262811899185181, "learning_rate": 5.165600000000001e-06, "loss": 0.0081, "step": 24173 }, { "epoch": 1.2887988484299195, "grad_norm": 1.651471734046936, "learning_rate": 5.1654000000000006e-06, "loss": 0.012, "step": 24174 }, { "epoch": 1.2888521618595723, "grad_norm": 2.6629981994628906, "learning_rate": 5.1652e-06, "loss": 0.0472, "step": 24175 }, { "epoch": 1.2889054752892253, "grad_norm": 3.6717782020568848, "learning_rate": 5.165e-06, "loss": 0.0205, "step": 24176 }, { "epoch": 1.2889587887188783, "grad_norm": 2.5841612815856934, "learning_rate": 5.164800000000001e-06, "loss": 0.0396, "step": 24177 }, { "epoch": 1.2890121021485312, "grad_norm": 2.058875322341919, "learning_rate": 5.1646000000000005e-06, "loss": 0.0478, "step": 24178 }, { "epoch": 1.2890654155781842, "grad_norm": 1.1936737298965454, "learning_rate": 5.164400000000001e-06, "loss": 0.0229, "step": 24179 }, { "epoch": 1.289118729007837, "grad_norm": 2.08646559715271, "learning_rate": 5.1642000000000005e-06, "loss": 0.0411, "step": 24180 }, { "epoch": 1.28917204243749, "grad_norm": 1.0150338411331177, "learning_rate": 5.164e-06, "loss": 0.023, "step": 24181 }, { "epoch": 1.289225355867143, "grad_norm": 1.5625896453857422, "learning_rate": 5.1638e-06, "loss": 0.0306, "step": 24182 }, { "epoch": 1.289278669296796, "grad_norm": 1.8021974563598633, "learning_rate": 5.163600000000001e-06, "loss": 0.0511, "step": 24183 }, { "epoch": 1.2893319827264489, "grad_norm": 3.0441832542419434, "learning_rate": 5.1634e-06, "loss": 0.0312, "step": 24184 }, { "epoch": 1.2893852961561016, "grad_norm": 2.1431400775909424, "learning_rate": 5.163200000000001e-06, "loss": 0.0217, "step": 24185 }, { "epoch": 1.2894386095857546, "grad_norm": 4.221127986907959, "learning_rate": 5.163e-06, "loss": 0.073, "step": 24186 }, { "epoch": 1.2894919230154076, "grad_norm": 1.8761839866638184, "learning_rate": 5.1628e-06, "loss": 0.0477, "step": 24187 }, { "epoch": 1.2895452364450606, "grad_norm": 1.0758520364761353, "learning_rate": 5.162600000000001e-06, "loss": 0.02, "step": 24188 }, { "epoch": 1.2895985498747136, "grad_norm": 1.748288631439209, "learning_rate": 5.162400000000001e-06, "loss": 0.0306, "step": 24189 }, { "epoch": 1.2896518633043663, "grad_norm": 1.9400724172592163, "learning_rate": 5.1622e-06, "loss": 0.0315, "step": 24190 }, { "epoch": 1.2897051767340193, "grad_norm": 1.2050807476043701, "learning_rate": 5.162000000000001e-06, "loss": 0.0236, "step": 24191 }, { "epoch": 1.2897584901636723, "grad_norm": 0.8551350831985474, "learning_rate": 5.1618e-06, "loss": 0.0162, "step": 24192 }, { "epoch": 1.289811803593325, "grad_norm": 2.0813024044036865, "learning_rate": 5.1616e-06, "loss": 0.0516, "step": 24193 }, { "epoch": 1.289865117022978, "grad_norm": 3.5947396755218506, "learning_rate": 5.161400000000001e-06, "loss": 0.0298, "step": 24194 }, { "epoch": 1.289918430452631, "grad_norm": 2.606335163116455, "learning_rate": 5.161200000000001e-06, "loss": 0.0267, "step": 24195 }, { "epoch": 1.289971743882284, "grad_norm": 1.8069555759429932, "learning_rate": 5.161e-06, "loss": 0.0369, "step": 24196 }, { "epoch": 1.290025057311937, "grad_norm": 1.262055516242981, "learning_rate": 5.160800000000001e-06, "loss": 0.0199, "step": 24197 }, { "epoch": 1.2900783707415897, "grad_norm": 0.8185926675796509, "learning_rate": 5.1606e-06, "loss": 0.0137, "step": 24198 }, { "epoch": 1.2901316841712427, "grad_norm": 1.020983338356018, "learning_rate": 5.160400000000001e-06, "loss": 0.0162, "step": 24199 }, { "epoch": 1.2901849976008957, "grad_norm": 0.9756070375442505, "learning_rate": 5.160200000000001e-06, "loss": 0.0114, "step": 24200 }, { "epoch": 1.2902383110305486, "grad_norm": 1.5411399602890015, "learning_rate": 5.1600000000000006e-06, "loss": 0.0151, "step": 24201 }, { "epoch": 1.2902916244602016, "grad_norm": 1.8419052362442017, "learning_rate": 5.1598e-06, "loss": 0.0401, "step": 24202 }, { "epoch": 1.2903449378898544, "grad_norm": 1.6653865575790405, "learning_rate": 5.1596000000000005e-06, "loss": 0.0169, "step": 24203 }, { "epoch": 1.2903982513195074, "grad_norm": 2.2693262100219727, "learning_rate": 5.1594e-06, "loss": 0.0269, "step": 24204 }, { "epoch": 1.2904515647491603, "grad_norm": 2.6653079986572266, "learning_rate": 5.1592000000000005e-06, "loss": 0.069, "step": 24205 }, { "epoch": 1.2905048781788133, "grad_norm": 2.7679874897003174, "learning_rate": 5.159000000000001e-06, "loss": 0.0263, "step": 24206 }, { "epoch": 1.2905581916084663, "grad_norm": 1.7698618173599243, "learning_rate": 5.1588000000000005e-06, "loss": 0.0457, "step": 24207 }, { "epoch": 1.290611505038119, "grad_norm": 1.0060901641845703, "learning_rate": 5.1586e-06, "loss": 0.0154, "step": 24208 }, { "epoch": 1.290664818467772, "grad_norm": 2.0868308544158936, "learning_rate": 5.1584000000000005e-06, "loss": 0.0295, "step": 24209 }, { "epoch": 1.290718131897425, "grad_norm": 1.7151269912719727, "learning_rate": 5.1582e-06, "loss": 0.0311, "step": 24210 }, { "epoch": 1.2907714453270778, "grad_norm": 1.986854910850525, "learning_rate": 5.158e-06, "loss": 0.0139, "step": 24211 }, { "epoch": 1.2908247587567308, "grad_norm": 2.6188836097717285, "learning_rate": 5.157800000000001e-06, "loss": 0.0613, "step": 24212 }, { "epoch": 1.2908780721863837, "grad_norm": 0.5676846504211426, "learning_rate": 5.1576e-06, "loss": 0.0036, "step": 24213 }, { "epoch": 1.2909313856160367, "grad_norm": 3.1360907554626465, "learning_rate": 5.1574e-06, "loss": 0.0224, "step": 24214 }, { "epoch": 1.2909846990456897, "grad_norm": 1.124212622642517, "learning_rate": 5.1572e-06, "loss": 0.0292, "step": 24215 }, { "epoch": 1.2910380124753424, "grad_norm": 2.795382499694824, "learning_rate": 5.157000000000001e-06, "loss": 0.0429, "step": 24216 }, { "epoch": 1.2910913259049954, "grad_norm": 1.0958476066589355, "learning_rate": 5.1568e-06, "loss": 0.0114, "step": 24217 }, { "epoch": 1.2911446393346484, "grad_norm": 2.522301435470581, "learning_rate": 5.156600000000001e-06, "loss": 0.0365, "step": 24218 }, { "epoch": 1.2911979527643014, "grad_norm": 0.9457300305366516, "learning_rate": 5.1564e-06, "loss": 0.0146, "step": 24219 }, { "epoch": 1.2912512661939544, "grad_norm": 0.6332003474235535, "learning_rate": 5.1562e-06, "loss": 0.0037, "step": 24220 }, { "epoch": 1.2913045796236071, "grad_norm": 2.324733018875122, "learning_rate": 5.156e-06, "loss": 0.0118, "step": 24221 }, { "epoch": 1.29135789305326, "grad_norm": 0.6547454595565796, "learning_rate": 5.155800000000001e-06, "loss": 0.0063, "step": 24222 }, { "epoch": 1.291411206482913, "grad_norm": 1.4329115152359009, "learning_rate": 5.1556e-06, "loss": 0.0107, "step": 24223 }, { "epoch": 1.291464519912566, "grad_norm": 1.0037908554077148, "learning_rate": 5.155400000000001e-06, "loss": 0.0142, "step": 24224 }, { "epoch": 1.291517833342219, "grad_norm": 1.2091232538223267, "learning_rate": 5.1552e-06, "loss": 0.0162, "step": 24225 }, { "epoch": 1.2915711467718718, "grad_norm": 1.8955078125, "learning_rate": 5.155e-06, "loss": 0.0375, "step": 24226 }, { "epoch": 1.2916244602015248, "grad_norm": 2.525836944580078, "learning_rate": 5.154800000000001e-06, "loss": 0.0307, "step": 24227 }, { "epoch": 1.2916777736311778, "grad_norm": 1.1544876098632812, "learning_rate": 5.154600000000001e-06, "loss": 0.0194, "step": 24228 }, { "epoch": 1.2917310870608305, "grad_norm": 0.8877701163291931, "learning_rate": 5.1544e-06, "loss": 0.0197, "step": 24229 }, { "epoch": 1.2917844004904835, "grad_norm": 1.594890832901001, "learning_rate": 5.1542000000000006e-06, "loss": 0.0265, "step": 24230 }, { "epoch": 1.2918377139201365, "grad_norm": 0.7898606061935425, "learning_rate": 5.154e-06, "loss": 0.0152, "step": 24231 }, { "epoch": 1.2918910273497894, "grad_norm": 0.624711811542511, "learning_rate": 5.1538e-06, "loss": 0.0058, "step": 24232 }, { "epoch": 1.2919443407794424, "grad_norm": 0.8768122792243958, "learning_rate": 5.153600000000001e-06, "loss": 0.0104, "step": 24233 }, { "epoch": 1.2919976542090952, "grad_norm": 2.315211296081543, "learning_rate": 5.1534000000000005e-06, "loss": 0.0259, "step": 24234 }, { "epoch": 1.2920509676387482, "grad_norm": 2.608294725418091, "learning_rate": 5.1532e-06, "loss": 0.0354, "step": 24235 }, { "epoch": 1.2921042810684011, "grad_norm": 1.2074370384216309, "learning_rate": 5.1530000000000005e-06, "loss": 0.0191, "step": 24236 }, { "epoch": 1.2921575944980541, "grad_norm": 2.1706411838531494, "learning_rate": 5.1528e-06, "loss": 0.0509, "step": 24237 }, { "epoch": 1.292210907927707, "grad_norm": 1.098526954650879, "learning_rate": 5.152600000000001e-06, "loss": 0.0075, "step": 24238 }, { "epoch": 1.2922642213573599, "grad_norm": 1.137266993522644, "learning_rate": 5.152400000000001e-06, "loss": 0.0077, "step": 24239 }, { "epoch": 1.2923175347870128, "grad_norm": 2.060342788696289, "learning_rate": 5.1522e-06, "loss": 0.0167, "step": 24240 }, { "epoch": 1.2923708482166658, "grad_norm": 1.8534718751907349, "learning_rate": 5.152e-06, "loss": 0.0081, "step": 24241 }, { "epoch": 1.2924241616463186, "grad_norm": 0.642164409160614, "learning_rate": 5.1518e-06, "loss": 0.0066, "step": 24242 }, { "epoch": 1.2924774750759718, "grad_norm": 1.0215961933135986, "learning_rate": 5.1516e-06, "loss": 0.0088, "step": 24243 }, { "epoch": 1.2925307885056245, "grad_norm": 1.2375723123550415, "learning_rate": 5.151400000000001e-06, "loss": 0.0107, "step": 24244 }, { "epoch": 1.2925841019352775, "grad_norm": 0.8227031826972961, "learning_rate": 5.151200000000001e-06, "loss": 0.0153, "step": 24245 }, { "epoch": 1.2926374153649305, "grad_norm": 4.18785285949707, "learning_rate": 5.151e-06, "loss": 0.0876, "step": 24246 }, { "epoch": 1.2926907287945832, "grad_norm": 2.1876838207244873, "learning_rate": 5.1508e-06, "loss": 0.0344, "step": 24247 }, { "epoch": 1.2927440422242362, "grad_norm": 1.2837163209915161, "learning_rate": 5.1506e-06, "loss": 0.0255, "step": 24248 }, { "epoch": 1.2927973556538892, "grad_norm": 1.1046477556228638, "learning_rate": 5.150400000000001e-06, "loss": 0.0107, "step": 24249 }, { "epoch": 1.2928506690835422, "grad_norm": 1.0176268815994263, "learning_rate": 5.150200000000001e-06, "loss": 0.0071, "step": 24250 }, { "epoch": 1.2929039825131952, "grad_norm": 1.8259350061416626, "learning_rate": 5.150000000000001e-06, "loss": 0.0286, "step": 24251 }, { "epoch": 1.292957295942848, "grad_norm": 2.3429062366485596, "learning_rate": 5.1498e-06, "loss": 0.0315, "step": 24252 }, { "epoch": 1.293010609372501, "grad_norm": 1.7325681447982788, "learning_rate": 5.1496e-06, "loss": 0.0447, "step": 24253 }, { "epoch": 1.2930639228021539, "grad_norm": 2.0095131397247314, "learning_rate": 5.1494e-06, "loss": 0.0395, "step": 24254 }, { "epoch": 1.2931172362318069, "grad_norm": 1.7150368690490723, "learning_rate": 5.149200000000001e-06, "loss": 0.0079, "step": 24255 }, { "epoch": 1.2931705496614598, "grad_norm": 1.1893153190612793, "learning_rate": 5.149000000000001e-06, "loss": 0.0092, "step": 24256 }, { "epoch": 1.2932238630911126, "grad_norm": 1.3379791975021362, "learning_rate": 5.148800000000001e-06, "loss": 0.0259, "step": 24257 }, { "epoch": 1.2932771765207656, "grad_norm": 2.6096277236938477, "learning_rate": 5.1486e-06, "loss": 0.0395, "step": 24258 }, { "epoch": 1.2933304899504185, "grad_norm": 2.3021252155303955, "learning_rate": 5.1484e-06, "loss": 0.0324, "step": 24259 }, { "epoch": 1.2933838033800713, "grad_norm": 1.454909324645996, "learning_rate": 5.148200000000001e-06, "loss": 0.0358, "step": 24260 }, { "epoch": 1.2934371168097245, "grad_norm": 3.4244656562805176, "learning_rate": 5.1480000000000005e-06, "loss": 0.0225, "step": 24261 }, { "epoch": 1.2934904302393773, "grad_norm": 3.3846404552459717, "learning_rate": 5.147800000000001e-06, "loss": 0.0259, "step": 24262 }, { "epoch": 1.2935437436690302, "grad_norm": 2.163912534713745, "learning_rate": 5.1476000000000005e-06, "loss": 0.0192, "step": 24263 }, { "epoch": 1.2935970570986832, "grad_norm": 2.251378059387207, "learning_rate": 5.1474e-06, "loss": 0.0563, "step": 24264 }, { "epoch": 1.293650370528336, "grad_norm": 1.3717960119247437, "learning_rate": 5.1472e-06, "loss": 0.0182, "step": 24265 }, { "epoch": 1.293703683957989, "grad_norm": 1.1509904861450195, "learning_rate": 5.147000000000001e-06, "loss": 0.0131, "step": 24266 }, { "epoch": 1.293756997387642, "grad_norm": 0.4453045725822449, "learning_rate": 5.1468000000000004e-06, "loss": 0.004, "step": 24267 }, { "epoch": 1.293810310817295, "grad_norm": 1.6977099180221558, "learning_rate": 5.146600000000001e-06, "loss": 0.02, "step": 24268 }, { "epoch": 1.293863624246948, "grad_norm": 1.6630598306655884, "learning_rate": 5.1464e-06, "loss": 0.0144, "step": 24269 }, { "epoch": 1.2939169376766007, "grad_norm": 1.2109785079956055, "learning_rate": 5.1462e-06, "loss": 0.0115, "step": 24270 }, { "epoch": 1.2939702511062536, "grad_norm": 1.4893163442611694, "learning_rate": 5.1459999999999995e-06, "loss": 0.0256, "step": 24271 }, { "epoch": 1.2940235645359066, "grad_norm": 1.529811978340149, "learning_rate": 5.145800000000001e-06, "loss": 0.0227, "step": 24272 }, { "epoch": 1.2940768779655596, "grad_norm": 0.26081493496894836, "learning_rate": 5.1456e-06, "loss": 0.0029, "step": 24273 }, { "epoch": 1.2941301913952126, "grad_norm": 4.47418737411499, "learning_rate": 5.145400000000001e-06, "loss": 0.0479, "step": 24274 }, { "epoch": 1.2941835048248653, "grad_norm": 0.9374577403068542, "learning_rate": 5.1452e-06, "loss": 0.008, "step": 24275 }, { "epoch": 1.2942368182545183, "grad_norm": 0.7912304401397705, "learning_rate": 5.145e-06, "loss": 0.0139, "step": 24276 }, { "epoch": 1.2942901316841713, "grad_norm": 2.305037021636963, "learning_rate": 5.144800000000001e-06, "loss": 0.0192, "step": 24277 }, { "epoch": 1.294343445113824, "grad_norm": 1.569739818572998, "learning_rate": 5.144600000000001e-06, "loss": 0.0185, "step": 24278 }, { "epoch": 1.294396758543477, "grad_norm": 0.3885137438774109, "learning_rate": 5.1444e-06, "loss": 0.0038, "step": 24279 }, { "epoch": 1.29445007197313, "grad_norm": 2.89096999168396, "learning_rate": 5.144200000000001e-06, "loss": 0.0366, "step": 24280 }, { "epoch": 1.294503385402783, "grad_norm": 1.5397748947143555, "learning_rate": 5.144e-06, "loss": 0.0309, "step": 24281 }, { "epoch": 1.294556698832436, "grad_norm": 2.3694069385528564, "learning_rate": 5.1438e-06, "loss": 0.0209, "step": 24282 }, { "epoch": 1.2946100122620887, "grad_norm": 3.1257565021514893, "learning_rate": 5.143600000000001e-06, "loss": 0.0374, "step": 24283 }, { "epoch": 1.2946633256917417, "grad_norm": 1.9686208963394165, "learning_rate": 5.143400000000001e-06, "loss": 0.0315, "step": 24284 }, { "epoch": 1.2947166391213947, "grad_norm": 0.9966133832931519, "learning_rate": 5.1432e-06, "loss": 0.0067, "step": 24285 }, { "epoch": 1.2947699525510477, "grad_norm": 1.6682212352752686, "learning_rate": 5.1430000000000006e-06, "loss": 0.019, "step": 24286 }, { "epoch": 1.2948232659807006, "grad_norm": 0.9628849029541016, "learning_rate": 5.1428e-06, "loss": 0.0085, "step": 24287 }, { "epoch": 1.2948765794103534, "grad_norm": 1.3652355670928955, "learning_rate": 5.1426000000000005e-06, "loss": 0.0351, "step": 24288 }, { "epoch": 1.2949298928400064, "grad_norm": 1.0135197639465332, "learning_rate": 5.142400000000001e-06, "loss": 0.0086, "step": 24289 }, { "epoch": 1.2949832062696593, "grad_norm": 2.2697038650512695, "learning_rate": 5.1422000000000005e-06, "loss": 0.0254, "step": 24290 }, { "epoch": 1.2950365196993123, "grad_norm": 3.624467372894287, "learning_rate": 5.142e-06, "loss": 0.029, "step": 24291 }, { "epoch": 1.2950898331289653, "grad_norm": 1.4410876035690308, "learning_rate": 5.1418000000000005e-06, "loss": 0.0091, "step": 24292 }, { "epoch": 1.295143146558618, "grad_norm": 2.4590303897857666, "learning_rate": 5.1416e-06, "loss": 0.0193, "step": 24293 }, { "epoch": 1.295196459988271, "grad_norm": 2.0490200519561768, "learning_rate": 5.1414000000000004e-06, "loss": 0.0512, "step": 24294 }, { "epoch": 1.295249773417924, "grad_norm": 1.2423678636550903, "learning_rate": 5.141200000000001e-06, "loss": 0.0122, "step": 24295 }, { "epoch": 1.2953030868475768, "grad_norm": 1.5699045658111572, "learning_rate": 5.141e-06, "loss": 0.0142, "step": 24296 }, { "epoch": 1.2953564002772298, "grad_norm": 1.3657336235046387, "learning_rate": 5.1408e-06, "loss": 0.0182, "step": 24297 }, { "epoch": 1.2954097137068827, "grad_norm": 1.386178731918335, "learning_rate": 5.1406e-06, "loss": 0.0237, "step": 24298 }, { "epoch": 1.2954630271365357, "grad_norm": 1.3587836027145386, "learning_rate": 5.140400000000001e-06, "loss": 0.0258, "step": 24299 }, { "epoch": 1.2955163405661887, "grad_norm": 1.6274152994155884, "learning_rate": 5.1402e-06, "loss": 0.0214, "step": 24300 }, { "epoch": 1.2955696539958415, "grad_norm": 1.5810872316360474, "learning_rate": 5.140000000000001e-06, "loss": 0.0307, "step": 24301 }, { "epoch": 1.2956229674254944, "grad_norm": 3.0783326625823975, "learning_rate": 5.1398e-06, "loss": 0.0307, "step": 24302 }, { "epoch": 1.2956762808551474, "grad_norm": 1.4562183618545532, "learning_rate": 5.1396e-06, "loss": 0.0282, "step": 24303 }, { "epoch": 1.2957295942848004, "grad_norm": 1.3550983667373657, "learning_rate": 5.1394e-06, "loss": 0.0131, "step": 24304 }, { "epoch": 1.2957829077144534, "grad_norm": 1.6498239040374756, "learning_rate": 5.139200000000001e-06, "loss": 0.0103, "step": 24305 }, { "epoch": 1.2958362211441061, "grad_norm": 1.489492654800415, "learning_rate": 5.139e-06, "loss": 0.0219, "step": 24306 }, { "epoch": 1.295889534573759, "grad_norm": 0.9356285929679871, "learning_rate": 5.138800000000001e-06, "loss": 0.0155, "step": 24307 }, { "epoch": 1.295942848003412, "grad_norm": 0.76760333776474, "learning_rate": 5.1386e-06, "loss": 0.0065, "step": 24308 }, { "epoch": 1.295996161433065, "grad_norm": 1.9105592966079712, "learning_rate": 5.1384e-06, "loss": 0.0279, "step": 24309 }, { "epoch": 1.296049474862718, "grad_norm": 1.7430291175842285, "learning_rate": 5.138200000000001e-06, "loss": 0.0239, "step": 24310 }, { "epoch": 1.2961027882923708, "grad_norm": 1.9312748908996582, "learning_rate": 5.138000000000001e-06, "loss": 0.0333, "step": 24311 }, { "epoch": 1.2961561017220238, "grad_norm": 0.7482038736343384, "learning_rate": 5.1378e-06, "loss": 0.0056, "step": 24312 }, { "epoch": 1.2962094151516768, "grad_norm": 1.164182186126709, "learning_rate": 5.137600000000001e-06, "loss": 0.0166, "step": 24313 }, { "epoch": 1.2962627285813295, "grad_norm": 4.228209495544434, "learning_rate": 5.1374e-06, "loss": 0.0423, "step": 24314 }, { "epoch": 1.2963160420109825, "grad_norm": 2.3837850093841553, "learning_rate": 5.1372e-06, "loss": 0.0208, "step": 24315 }, { "epoch": 1.2963693554406355, "grad_norm": 2.674520254135132, "learning_rate": 5.137000000000001e-06, "loss": 0.0253, "step": 24316 }, { "epoch": 1.2964226688702885, "grad_norm": 2.8504669666290283, "learning_rate": 5.1368000000000005e-06, "loss": 0.018, "step": 24317 }, { "epoch": 1.2964759822999414, "grad_norm": 1.3853520154953003, "learning_rate": 5.1366e-06, "loss": 0.0147, "step": 24318 }, { "epoch": 1.2965292957295942, "grad_norm": 1.9216198921203613, "learning_rate": 5.1364000000000005e-06, "loss": 0.0243, "step": 24319 }, { "epoch": 1.2965826091592472, "grad_norm": 2.4637348651885986, "learning_rate": 5.1362e-06, "loss": 0.0295, "step": 24320 }, { "epoch": 1.2966359225889001, "grad_norm": 1.6179049015045166, "learning_rate": 5.136e-06, "loss": 0.0272, "step": 24321 }, { "epoch": 1.2966892360185531, "grad_norm": 0.4832450747489929, "learning_rate": 5.135800000000001e-06, "loss": 0.0052, "step": 24322 }, { "epoch": 1.296742549448206, "grad_norm": 0.4724774956703186, "learning_rate": 5.1356000000000004e-06, "loss": 0.0044, "step": 24323 }, { "epoch": 1.2967958628778589, "grad_norm": 2.981785535812378, "learning_rate": 5.1354e-06, "loss": 0.0132, "step": 24324 }, { "epoch": 1.2968491763075118, "grad_norm": 1.4786863327026367, "learning_rate": 5.1352e-06, "loss": 0.0089, "step": 24325 }, { "epoch": 1.2969024897371648, "grad_norm": 2.95873761177063, "learning_rate": 5.135e-06, "loss": 0.0555, "step": 24326 }, { "epoch": 1.2969558031668178, "grad_norm": 3.159022331237793, "learning_rate": 5.134800000000001e-06, "loss": 0.0597, "step": 24327 }, { "epoch": 1.2970091165964708, "grad_norm": 2.7262516021728516, "learning_rate": 5.134600000000001e-06, "loss": 0.0389, "step": 24328 }, { "epoch": 1.2970624300261235, "grad_norm": 2.216773271560669, "learning_rate": 5.1344e-06, "loss": 0.0381, "step": 24329 }, { "epoch": 1.2971157434557765, "grad_norm": 2.232501268386841, "learning_rate": 5.1342e-06, "loss": 0.0474, "step": 24330 }, { "epoch": 1.2971690568854295, "grad_norm": 2.435830593109131, "learning_rate": 5.134e-06, "loss": 0.0235, "step": 24331 }, { "epoch": 1.2972223703150823, "grad_norm": 2.097933530807495, "learning_rate": 5.1338e-06, "loss": 0.0253, "step": 24332 }, { "epoch": 1.2972756837447352, "grad_norm": 1.1074764728546143, "learning_rate": 5.133600000000001e-06, "loss": 0.0101, "step": 24333 }, { "epoch": 1.2973289971743882, "grad_norm": 3.6995365619659424, "learning_rate": 5.133400000000001e-06, "loss": 0.0287, "step": 24334 }, { "epoch": 1.2973823106040412, "grad_norm": 2.289760112762451, "learning_rate": 5.1332e-06, "loss": 0.0205, "step": 24335 }, { "epoch": 1.2974356240336942, "grad_norm": 1.5603820085525513, "learning_rate": 5.133e-06, "loss": 0.0152, "step": 24336 }, { "epoch": 1.297488937463347, "grad_norm": 1.1708450317382812, "learning_rate": 5.1328e-06, "loss": 0.0153, "step": 24337 }, { "epoch": 1.297542250893, "grad_norm": 1.913826823234558, "learning_rate": 5.132600000000001e-06, "loss": 0.0355, "step": 24338 }, { "epoch": 1.2975955643226529, "grad_norm": 1.8265511989593506, "learning_rate": 5.132400000000001e-06, "loss": 0.0147, "step": 24339 }, { "epoch": 1.2976488777523059, "grad_norm": 1.4034881591796875, "learning_rate": 5.132200000000001e-06, "loss": 0.0184, "step": 24340 }, { "epoch": 1.2977021911819588, "grad_norm": 1.382891297340393, "learning_rate": 5.132e-06, "loss": 0.0251, "step": 24341 }, { "epoch": 1.2977555046116116, "grad_norm": 6.143177509307861, "learning_rate": 5.1318e-06, "loss": 0.0468, "step": 24342 }, { "epoch": 1.2978088180412646, "grad_norm": 1.4963514804840088, "learning_rate": 5.1316e-06, "loss": 0.018, "step": 24343 }, { "epoch": 1.2978621314709176, "grad_norm": 1.2803586721420288, "learning_rate": 5.1314000000000005e-06, "loss": 0.0183, "step": 24344 }, { "epoch": 1.2979154449005705, "grad_norm": 1.8791635036468506, "learning_rate": 5.131200000000001e-06, "loss": 0.026, "step": 24345 }, { "epoch": 1.2979687583302235, "grad_norm": 0.3613169193267822, "learning_rate": 5.1310000000000005e-06, "loss": 0.0033, "step": 24346 }, { "epoch": 1.2980220717598763, "grad_norm": 1.7679373025894165, "learning_rate": 5.1308e-06, "loss": 0.0272, "step": 24347 }, { "epoch": 1.2980753851895293, "grad_norm": 1.7433407306671143, "learning_rate": 5.1306e-06, "loss": 0.0394, "step": 24348 }, { "epoch": 1.2981286986191822, "grad_norm": 2.0316827297210693, "learning_rate": 5.130400000000001e-06, "loss": 0.0238, "step": 24349 }, { "epoch": 1.298182012048835, "grad_norm": 1.1134212017059326, "learning_rate": 5.1302000000000004e-06, "loss": 0.0156, "step": 24350 }, { "epoch": 1.298235325478488, "grad_norm": 2.278036594390869, "learning_rate": 5.130000000000001e-06, "loss": 0.0682, "step": 24351 }, { "epoch": 1.298288638908141, "grad_norm": 1.199690580368042, "learning_rate": 5.1298e-06, "loss": 0.0167, "step": 24352 }, { "epoch": 1.298341952337794, "grad_norm": 1.3657994270324707, "learning_rate": 5.1296e-06, "loss": 0.0115, "step": 24353 }, { "epoch": 1.298395265767447, "grad_norm": 0.6668207049369812, "learning_rate": 5.1293999999999995e-06, "loss": 0.0062, "step": 24354 }, { "epoch": 1.2984485791970997, "grad_norm": 1.4240922927856445, "learning_rate": 5.129200000000001e-06, "loss": 0.0146, "step": 24355 }, { "epoch": 1.2985018926267526, "grad_norm": 1.8728777170181274, "learning_rate": 5.129e-06, "loss": 0.0198, "step": 24356 }, { "epoch": 1.2985552060564056, "grad_norm": 1.834965467453003, "learning_rate": 5.128800000000001e-06, "loss": 0.028, "step": 24357 }, { "epoch": 1.2986085194860586, "grad_norm": 1.4294201135635376, "learning_rate": 5.1286e-06, "loss": 0.0092, "step": 24358 }, { "epoch": 1.2986618329157116, "grad_norm": 0.4455427825450897, "learning_rate": 5.1284e-06, "loss": 0.0046, "step": 24359 }, { "epoch": 1.2987151463453643, "grad_norm": 1.3267009258270264, "learning_rate": 5.128200000000001e-06, "loss": 0.0117, "step": 24360 }, { "epoch": 1.2987684597750173, "grad_norm": 1.7884336709976196, "learning_rate": 5.128000000000001e-06, "loss": 0.0187, "step": 24361 }, { "epoch": 1.2988217732046703, "grad_norm": 0.8982952833175659, "learning_rate": 5.1278e-06, "loss": 0.0071, "step": 24362 }, { "epoch": 1.298875086634323, "grad_norm": 1.572050929069519, "learning_rate": 5.127600000000001e-06, "loss": 0.0318, "step": 24363 }, { "epoch": 1.2989284000639763, "grad_norm": 1.3801969289779663, "learning_rate": 5.1274e-06, "loss": 0.0538, "step": 24364 }, { "epoch": 1.298981713493629, "grad_norm": 2.375032901763916, "learning_rate": 5.1272e-06, "loss": 0.0625, "step": 24365 }, { "epoch": 1.299035026923282, "grad_norm": 1.3331600427627563, "learning_rate": 5.127000000000001e-06, "loss": 0.0097, "step": 24366 }, { "epoch": 1.299088340352935, "grad_norm": 0.6338685154914856, "learning_rate": 5.126800000000001e-06, "loss": 0.0064, "step": 24367 }, { "epoch": 1.2991416537825877, "grad_norm": 1.3868112564086914, "learning_rate": 5.1266e-06, "loss": 0.0193, "step": 24368 }, { "epoch": 1.2991949672122407, "grad_norm": 2.1968605518341064, "learning_rate": 5.126400000000001e-06, "loss": 0.0393, "step": 24369 }, { "epoch": 1.2992482806418937, "grad_norm": 0.8020139336585999, "learning_rate": 5.1262e-06, "loss": 0.0058, "step": 24370 }, { "epoch": 1.2993015940715467, "grad_norm": 2.112226963043213, "learning_rate": 5.126e-06, "loss": 0.0185, "step": 24371 }, { "epoch": 1.2993549075011996, "grad_norm": 2.190980911254883, "learning_rate": 5.125800000000001e-06, "loss": 0.0228, "step": 24372 }, { "epoch": 1.2994082209308524, "grad_norm": 1.6900993585586548, "learning_rate": 5.1256000000000005e-06, "loss": 0.0174, "step": 24373 }, { "epoch": 1.2994615343605054, "grad_norm": 1.5115550756454468, "learning_rate": 5.1254e-06, "loss": 0.0179, "step": 24374 }, { "epoch": 1.2995148477901584, "grad_norm": 1.161665678024292, "learning_rate": 5.1252000000000005e-06, "loss": 0.0181, "step": 24375 }, { "epoch": 1.2995681612198113, "grad_norm": 2.445573091506958, "learning_rate": 5.125e-06, "loss": 0.0188, "step": 24376 }, { "epoch": 1.2996214746494643, "grad_norm": 0.7711130976676941, "learning_rate": 5.1248000000000005e-06, "loss": 0.0142, "step": 24377 }, { "epoch": 1.299674788079117, "grad_norm": 1.1779298782348633, "learning_rate": 5.124600000000001e-06, "loss": 0.0188, "step": 24378 }, { "epoch": 1.29972810150877, "grad_norm": 1.0462672710418701, "learning_rate": 5.1244000000000004e-06, "loss": 0.0263, "step": 24379 }, { "epoch": 1.299781414938423, "grad_norm": 1.5967861413955688, "learning_rate": 5.1242e-06, "loss": 0.029, "step": 24380 }, { "epoch": 1.2998347283680758, "grad_norm": 2.3379650115966797, "learning_rate": 5.124e-06, "loss": 0.0178, "step": 24381 }, { "epoch": 1.2998880417977288, "grad_norm": 2.2498109340667725, "learning_rate": 5.1238e-06, "loss": 0.034, "step": 24382 }, { "epoch": 1.2999413552273817, "grad_norm": 1.5651065111160278, "learning_rate": 5.1236e-06, "loss": 0.0346, "step": 24383 }, { "epoch": 1.2999946686570347, "grad_norm": 1.4607146978378296, "learning_rate": 5.123400000000001e-06, "loss": 0.0244, "step": 24384 }, { "epoch": 1.3000479820866877, "grad_norm": 2.186866044998169, "learning_rate": 5.1232e-06, "loss": 0.0178, "step": 24385 }, { "epoch": 1.3001012955163405, "grad_norm": 2.3551416397094727, "learning_rate": 5.123e-06, "loss": 0.0475, "step": 24386 }, { "epoch": 1.3001546089459934, "grad_norm": 2.99994158744812, "learning_rate": 5.1228e-06, "loss": 0.0541, "step": 24387 }, { "epoch": 1.3002079223756464, "grad_norm": 1.4087034463882446, "learning_rate": 5.122600000000001e-06, "loss": 0.0222, "step": 24388 }, { "epoch": 1.3002612358052994, "grad_norm": 2.1122629642486572, "learning_rate": 5.1224e-06, "loss": 0.026, "step": 24389 }, { "epoch": 1.3003145492349524, "grad_norm": 1.3224740028381348, "learning_rate": 5.122200000000001e-06, "loss": 0.0167, "step": 24390 }, { "epoch": 1.3003678626646051, "grad_norm": 1.8409382104873657, "learning_rate": 5.122e-06, "loss": 0.0184, "step": 24391 }, { "epoch": 1.3004211760942581, "grad_norm": 0.9559870958328247, "learning_rate": 5.1218e-06, "loss": 0.0067, "step": 24392 }, { "epoch": 1.300474489523911, "grad_norm": 0.3898009955883026, "learning_rate": 5.1216e-06, "loss": 0.0039, "step": 24393 }, { "epoch": 1.300527802953564, "grad_norm": 1.941837191581726, "learning_rate": 5.121400000000001e-06, "loss": 0.0177, "step": 24394 }, { "epoch": 1.300581116383217, "grad_norm": 1.6257410049438477, "learning_rate": 5.1212e-06, "loss": 0.0124, "step": 24395 }, { "epoch": 1.3006344298128698, "grad_norm": 1.202805995941162, "learning_rate": 5.121000000000001e-06, "loss": 0.0151, "step": 24396 }, { "epoch": 1.3006877432425228, "grad_norm": 1.0092346668243408, "learning_rate": 5.1208e-06, "loss": 0.0211, "step": 24397 }, { "epoch": 1.3007410566721758, "grad_norm": 1.3695396184921265, "learning_rate": 5.1206e-06, "loss": 0.0226, "step": 24398 }, { "epoch": 1.3007943701018285, "grad_norm": 1.5441726446151733, "learning_rate": 5.120400000000001e-06, "loss": 0.018, "step": 24399 }, { "epoch": 1.3008476835314815, "grad_norm": 2.3158862590789795, "learning_rate": 5.1202000000000005e-06, "loss": 0.0264, "step": 24400 }, { "epoch": 1.3009009969611345, "grad_norm": 2.225146770477295, "learning_rate": 5.12e-06, "loss": 0.0309, "step": 24401 }, { "epoch": 1.3009543103907875, "grad_norm": 1.1048561334609985, "learning_rate": 5.1198000000000005e-06, "loss": 0.0126, "step": 24402 }, { "epoch": 1.3010076238204404, "grad_norm": 1.4298053979873657, "learning_rate": 5.1196e-06, "loss": 0.0271, "step": 24403 }, { "epoch": 1.3010609372500932, "grad_norm": 3.826483726501465, "learning_rate": 5.1194e-06, "loss": 0.0492, "step": 24404 }, { "epoch": 1.3011142506797462, "grad_norm": 1.0024760961532593, "learning_rate": 5.119200000000001e-06, "loss": 0.0063, "step": 24405 }, { "epoch": 1.3011675641093992, "grad_norm": 0.504395604133606, "learning_rate": 5.1190000000000005e-06, "loss": 0.0062, "step": 24406 }, { "epoch": 1.3012208775390521, "grad_norm": 0.8778071403503418, "learning_rate": 5.1188e-06, "loss": 0.0069, "step": 24407 }, { "epoch": 1.3012741909687051, "grad_norm": 1.588404655456543, "learning_rate": 5.1186000000000004e-06, "loss": 0.0237, "step": 24408 }, { "epoch": 1.3013275043983579, "grad_norm": 0.9186829328536987, "learning_rate": 5.1184e-06, "loss": 0.0081, "step": 24409 }, { "epoch": 1.3013808178280108, "grad_norm": 2.3219010829925537, "learning_rate": 5.118200000000001e-06, "loss": 0.0401, "step": 24410 }, { "epoch": 1.3014341312576638, "grad_norm": 0.7927256226539612, "learning_rate": 5.118000000000001e-06, "loss": 0.0064, "step": 24411 }, { "epoch": 1.3014874446873168, "grad_norm": 2.1307318210601807, "learning_rate": 5.1178e-06, "loss": 0.0462, "step": 24412 }, { "epoch": 1.3015407581169698, "grad_norm": 0.6901273727416992, "learning_rate": 5.1176e-06, "loss": 0.0059, "step": 24413 }, { "epoch": 1.3015940715466225, "grad_norm": 1.7074271440505981, "learning_rate": 5.1174e-06, "loss": 0.0217, "step": 24414 }, { "epoch": 1.3016473849762755, "grad_norm": 1.2244290113449097, "learning_rate": 5.1172e-06, "loss": 0.0119, "step": 24415 }, { "epoch": 1.3017006984059285, "grad_norm": 3.174482583999634, "learning_rate": 5.117000000000001e-06, "loss": 0.0357, "step": 24416 }, { "epoch": 1.3017540118355813, "grad_norm": 2.378357172012329, "learning_rate": 5.116800000000001e-06, "loss": 0.0094, "step": 24417 }, { "epoch": 1.3018073252652342, "grad_norm": 1.8848603963851929, "learning_rate": 5.1166e-06, "loss": 0.0182, "step": 24418 }, { "epoch": 1.3018606386948872, "grad_norm": 1.6472108364105225, "learning_rate": 5.1164e-06, "loss": 0.0309, "step": 24419 }, { "epoch": 1.3019139521245402, "grad_norm": 1.5335732698440552, "learning_rate": 5.1162e-06, "loss": 0.0409, "step": 24420 }, { "epoch": 1.3019672655541932, "grad_norm": 5.982068061828613, "learning_rate": 5.116000000000001e-06, "loss": 0.0261, "step": 24421 }, { "epoch": 1.302020578983846, "grad_norm": 2.3582468032836914, "learning_rate": 5.115800000000001e-06, "loss": 0.0335, "step": 24422 }, { "epoch": 1.302073892413499, "grad_norm": 2.02297306060791, "learning_rate": 5.115600000000001e-06, "loss": 0.0327, "step": 24423 }, { "epoch": 1.302127205843152, "grad_norm": 1.6538777351379395, "learning_rate": 5.1154e-06, "loss": 0.0254, "step": 24424 }, { "epoch": 1.3021805192728049, "grad_norm": 0.7511359453201294, "learning_rate": 5.1152e-06, "loss": 0.0058, "step": 24425 }, { "epoch": 1.3022338327024578, "grad_norm": 2.2234115600585938, "learning_rate": 5.115e-06, "loss": 0.0178, "step": 24426 }, { "epoch": 1.3022871461321106, "grad_norm": 1.4068429470062256, "learning_rate": 5.1148000000000006e-06, "loss": 0.0131, "step": 24427 }, { "epoch": 1.3023404595617636, "grad_norm": 2.4535675048828125, "learning_rate": 5.114600000000001e-06, "loss": 0.0384, "step": 24428 }, { "epoch": 1.3023937729914166, "grad_norm": 2.823611259460449, "learning_rate": 5.1144000000000005e-06, "loss": 0.0289, "step": 24429 }, { "epoch": 1.3024470864210695, "grad_norm": 1.983438491821289, "learning_rate": 5.1142e-06, "loss": 0.0132, "step": 24430 }, { "epoch": 1.3025003998507225, "grad_norm": 1.2101186513900757, "learning_rate": 5.114e-06, "loss": 0.0135, "step": 24431 }, { "epoch": 1.3025537132803753, "grad_norm": 1.9329158067703247, "learning_rate": 5.1138e-06, "loss": 0.0201, "step": 24432 }, { "epoch": 1.3026070267100283, "grad_norm": 2.124044418334961, "learning_rate": 5.1136000000000005e-06, "loss": 0.0445, "step": 24433 }, { "epoch": 1.3026603401396812, "grad_norm": 3.5613720417022705, "learning_rate": 5.113400000000001e-06, "loss": 0.0198, "step": 24434 }, { "epoch": 1.302713653569334, "grad_norm": 0.9470379948616028, "learning_rate": 5.1132000000000004e-06, "loss": 0.0198, "step": 24435 }, { "epoch": 1.302766966998987, "grad_norm": 1.5772230625152588, "learning_rate": 5.113e-06, "loss": 0.0246, "step": 24436 }, { "epoch": 1.30282028042864, "grad_norm": 1.6871119737625122, "learning_rate": 5.1127999999999996e-06, "loss": 0.0261, "step": 24437 }, { "epoch": 1.302873593858293, "grad_norm": 1.1518752574920654, "learning_rate": 5.112600000000001e-06, "loss": 0.0266, "step": 24438 }, { "epoch": 1.302926907287946, "grad_norm": 1.3897732496261597, "learning_rate": 5.1124e-06, "loss": 0.0103, "step": 24439 }, { "epoch": 1.3029802207175987, "grad_norm": 1.2061442136764526, "learning_rate": 5.112200000000001e-06, "loss": 0.0168, "step": 24440 }, { "epoch": 1.3030335341472516, "grad_norm": 4.597884178161621, "learning_rate": 5.112e-06, "loss": 0.0536, "step": 24441 }, { "epoch": 1.3030868475769046, "grad_norm": 2.6764893531799316, "learning_rate": 5.1118e-06, "loss": 0.0589, "step": 24442 }, { "epoch": 1.3031401610065576, "grad_norm": 2.4959044456481934, "learning_rate": 5.1115999999999995e-06, "loss": 0.0275, "step": 24443 }, { "epoch": 1.3031934744362106, "grad_norm": 1.6536870002746582, "learning_rate": 5.111400000000001e-06, "loss": 0.0101, "step": 24444 }, { "epoch": 1.3032467878658633, "grad_norm": 2.6273417472839355, "learning_rate": 5.1112e-06, "loss": 0.0161, "step": 24445 }, { "epoch": 1.3033001012955163, "grad_norm": 1.2323843240737915, "learning_rate": 5.111000000000001e-06, "loss": 0.0104, "step": 24446 }, { "epoch": 1.3033534147251693, "grad_norm": 1.9382874965667725, "learning_rate": 5.1108e-06, "loss": 0.0241, "step": 24447 }, { "epoch": 1.3034067281548223, "grad_norm": 1.7341476678848267, "learning_rate": 5.1106e-06, "loss": 0.0291, "step": 24448 }, { "epoch": 1.3034600415844753, "grad_norm": 1.692918300628662, "learning_rate": 5.110400000000001e-06, "loss": 0.0168, "step": 24449 }, { "epoch": 1.303513355014128, "grad_norm": 1.6227738857269287, "learning_rate": 5.110200000000001e-06, "loss": 0.0119, "step": 24450 }, { "epoch": 1.303566668443781, "grad_norm": 0.7696548700332642, "learning_rate": 5.11e-06, "loss": 0.0158, "step": 24451 }, { "epoch": 1.303619981873434, "grad_norm": 2.633944511413574, "learning_rate": 5.109800000000001e-06, "loss": 0.0182, "step": 24452 }, { "epoch": 1.3036732953030867, "grad_norm": 0.5441300868988037, "learning_rate": 5.1096e-06, "loss": 0.0063, "step": 24453 }, { "epoch": 1.3037266087327397, "grad_norm": 3.0739119052886963, "learning_rate": 5.1094e-06, "loss": 0.0767, "step": 24454 }, { "epoch": 1.3037799221623927, "grad_norm": 1.404551386833191, "learning_rate": 5.109200000000001e-06, "loss": 0.0176, "step": 24455 }, { "epoch": 1.3038332355920457, "grad_norm": 0.7460718750953674, "learning_rate": 5.1090000000000006e-06, "loss": 0.0068, "step": 24456 }, { "epoch": 1.3038865490216986, "grad_norm": 1.9654942750930786, "learning_rate": 5.1088e-06, "loss": 0.0257, "step": 24457 }, { "epoch": 1.3039398624513514, "grad_norm": 1.6059738397598267, "learning_rate": 5.1086000000000005e-06, "loss": 0.0148, "step": 24458 }, { "epoch": 1.3039931758810044, "grad_norm": 0.9483073949813843, "learning_rate": 5.1084e-06, "loss": 0.0065, "step": 24459 }, { "epoch": 1.3040464893106574, "grad_norm": 2.4589641094207764, "learning_rate": 5.108200000000001e-06, "loss": 0.0152, "step": 24460 }, { "epoch": 1.3040998027403103, "grad_norm": 0.5931034088134766, "learning_rate": 5.108000000000001e-06, "loss": 0.005, "step": 24461 }, { "epoch": 1.3041531161699633, "grad_norm": 0.7211852669715881, "learning_rate": 5.1078000000000005e-06, "loss": 0.0055, "step": 24462 }, { "epoch": 1.304206429599616, "grad_norm": 1.6119498014450073, "learning_rate": 5.1076e-06, "loss": 0.0097, "step": 24463 }, { "epoch": 1.304259743029269, "grad_norm": 0.29823336005210876, "learning_rate": 5.1074000000000004e-06, "loss": 0.0041, "step": 24464 }, { "epoch": 1.304313056458922, "grad_norm": 2.6175763607025146, "learning_rate": 5.1072e-06, "loss": 0.0152, "step": 24465 }, { "epoch": 1.3043663698885748, "grad_norm": 0.9164673089981079, "learning_rate": 5.107000000000001e-06, "loss": 0.0061, "step": 24466 }, { "epoch": 1.304419683318228, "grad_norm": 2.034466505050659, "learning_rate": 5.106800000000001e-06, "loss": 0.0439, "step": 24467 }, { "epoch": 1.3044729967478808, "grad_norm": 2.0763673782348633, "learning_rate": 5.1066e-06, "loss": 0.0235, "step": 24468 }, { "epoch": 1.3045263101775337, "grad_norm": 1.080284833908081, "learning_rate": 5.1064e-06, "loss": 0.018, "step": 24469 }, { "epoch": 1.3045796236071867, "grad_norm": 0.9093674421310425, "learning_rate": 5.1062e-06, "loss": 0.0231, "step": 24470 }, { "epoch": 1.3046329370368395, "grad_norm": 1.9425816535949707, "learning_rate": 5.106000000000001e-06, "loss": 0.0336, "step": 24471 }, { "epoch": 1.3046862504664924, "grad_norm": 1.3205338716506958, "learning_rate": 5.105800000000001e-06, "loss": 0.0285, "step": 24472 }, { "epoch": 1.3047395638961454, "grad_norm": 2.2719836235046387, "learning_rate": 5.105600000000001e-06, "loss": 0.0186, "step": 24473 }, { "epoch": 1.3047928773257984, "grad_norm": 0.9788926839828491, "learning_rate": 5.1054e-06, "loss": 0.0162, "step": 24474 }, { "epoch": 1.3048461907554514, "grad_norm": 2.6332807540893555, "learning_rate": 5.1052e-06, "loss": 0.0226, "step": 24475 }, { "epoch": 1.3048995041851041, "grad_norm": 3.6636369228363037, "learning_rate": 5.105e-06, "loss": 0.0213, "step": 24476 }, { "epoch": 1.3049528176147571, "grad_norm": 0.19874787330627441, "learning_rate": 5.104800000000001e-06, "loss": 0.0024, "step": 24477 }, { "epoch": 1.30500613104441, "grad_norm": 0.3416534662246704, "learning_rate": 5.104600000000001e-06, "loss": 0.0027, "step": 24478 }, { "epoch": 1.305059444474063, "grad_norm": 2.512411594390869, "learning_rate": 5.104400000000001e-06, "loss": 0.0217, "step": 24479 }, { "epoch": 1.305112757903716, "grad_norm": 1.38298761844635, "learning_rate": 5.1042e-06, "loss": 0.0287, "step": 24480 }, { "epoch": 1.3051660713333688, "grad_norm": 5.384045600891113, "learning_rate": 5.104e-06, "loss": 0.0563, "step": 24481 }, { "epoch": 1.3052193847630218, "grad_norm": 1.8937926292419434, "learning_rate": 5.1038e-06, "loss": 0.0276, "step": 24482 }, { "epoch": 1.3052726981926748, "grad_norm": 0.505452573299408, "learning_rate": 5.1036000000000006e-06, "loss": 0.0033, "step": 24483 }, { "epoch": 1.3053260116223275, "grad_norm": 1.2467063665390015, "learning_rate": 5.103400000000001e-06, "loss": 0.0347, "step": 24484 }, { "epoch": 1.3053793250519807, "grad_norm": 0.8792902231216431, "learning_rate": 5.1032000000000005e-06, "loss": 0.005, "step": 24485 }, { "epoch": 1.3054326384816335, "grad_norm": 4.970947742462158, "learning_rate": 5.103e-06, "loss": 0.0151, "step": 24486 }, { "epoch": 1.3054859519112865, "grad_norm": 1.6356765031814575, "learning_rate": 5.1028e-06, "loss": 0.0327, "step": 24487 }, { "epoch": 1.3055392653409394, "grad_norm": 1.3769341707229614, "learning_rate": 5.102600000000001e-06, "loss": 0.0125, "step": 24488 }, { "epoch": 1.3055925787705922, "grad_norm": 2.234564781188965, "learning_rate": 5.1024000000000005e-06, "loss": 0.026, "step": 24489 }, { "epoch": 1.3056458922002452, "grad_norm": 1.1125693321228027, "learning_rate": 5.102200000000001e-06, "loss": 0.0075, "step": 24490 }, { "epoch": 1.3056992056298982, "grad_norm": 1.4529117345809937, "learning_rate": 5.1020000000000004e-06, "loss": 0.0242, "step": 24491 }, { "epoch": 1.3057525190595511, "grad_norm": 0.9723875522613525, "learning_rate": 5.1018e-06, "loss": 0.017, "step": 24492 }, { "epoch": 1.3058058324892041, "grad_norm": 3.5282459259033203, "learning_rate": 5.1015999999999996e-06, "loss": 0.0137, "step": 24493 }, { "epoch": 1.3058591459188569, "grad_norm": 1.1609470844268799, "learning_rate": 5.101400000000001e-06, "loss": 0.027, "step": 24494 }, { "epoch": 1.3059124593485099, "grad_norm": 2.8706843852996826, "learning_rate": 5.1012e-06, "loss": 0.0536, "step": 24495 }, { "epoch": 1.3059657727781628, "grad_norm": 1.3659327030181885, "learning_rate": 5.101000000000001e-06, "loss": 0.022, "step": 24496 }, { "epoch": 1.3060190862078158, "grad_norm": 0.639160692691803, "learning_rate": 5.1008e-06, "loss": 0.0047, "step": 24497 }, { "epoch": 1.3060723996374688, "grad_norm": 3.878523349761963, "learning_rate": 5.1006e-06, "loss": 0.011, "step": 24498 }, { "epoch": 1.3061257130671216, "grad_norm": 1.2177367210388184, "learning_rate": 5.100400000000001e-06, "loss": 0.0154, "step": 24499 }, { "epoch": 1.3061790264967745, "grad_norm": 2.0875017642974854, "learning_rate": 5.100200000000001e-06, "loss": 0.0173, "step": 24500 }, { "epoch": 1.3062323399264275, "grad_norm": 2.067802667617798, "learning_rate": 5.1e-06, "loss": 0.0131, "step": 24501 }, { "epoch": 1.3062856533560803, "grad_norm": 2.1782875061035156, "learning_rate": 5.099800000000001e-06, "loss": 0.0225, "step": 24502 }, { "epoch": 1.3063389667857332, "grad_norm": 2.021329164505005, "learning_rate": 5.0996e-06, "loss": 0.0187, "step": 24503 }, { "epoch": 1.3063922802153862, "grad_norm": 1.3647396564483643, "learning_rate": 5.0994e-06, "loss": 0.0133, "step": 24504 }, { "epoch": 1.3064455936450392, "grad_norm": 1.489842176437378, "learning_rate": 5.099200000000001e-06, "loss": 0.0176, "step": 24505 }, { "epoch": 1.3064989070746922, "grad_norm": 0.9750438928604126, "learning_rate": 5.099000000000001e-06, "loss": 0.0121, "step": 24506 }, { "epoch": 1.306552220504345, "grad_norm": 1.4251270294189453, "learning_rate": 5.0988e-06, "loss": 0.009, "step": 24507 }, { "epoch": 1.306605533933998, "grad_norm": 1.738436222076416, "learning_rate": 5.098600000000001e-06, "loss": 0.028, "step": 24508 }, { "epoch": 1.306658847363651, "grad_norm": 1.242906093597412, "learning_rate": 5.0984e-06, "loss": 0.0072, "step": 24509 }, { "epoch": 1.3067121607933039, "grad_norm": 1.4487725496292114, "learning_rate": 5.098200000000001e-06, "loss": 0.031, "step": 24510 }, { "epoch": 1.3067654742229569, "grad_norm": 2.627199172973633, "learning_rate": 5.098000000000001e-06, "loss": 0.0272, "step": 24511 }, { "epoch": 1.3068187876526096, "grad_norm": 1.3760908842086792, "learning_rate": 5.0978000000000006e-06, "loss": 0.0263, "step": 24512 }, { "epoch": 1.3068721010822626, "grad_norm": 2.5636961460113525, "learning_rate": 5.0976e-06, "loss": 0.0523, "step": 24513 }, { "epoch": 1.3069254145119156, "grad_norm": 2.9919402599334717, "learning_rate": 5.0974000000000005e-06, "loss": 0.0423, "step": 24514 }, { "epoch": 1.3069787279415686, "grad_norm": 1.071582555770874, "learning_rate": 5.0972e-06, "loss": 0.0175, "step": 24515 }, { "epoch": 1.3070320413712215, "grad_norm": 3.9681661128997803, "learning_rate": 5.0970000000000005e-06, "loss": 0.0482, "step": 24516 }, { "epoch": 1.3070853548008743, "grad_norm": 1.4073503017425537, "learning_rate": 5.096800000000001e-06, "loss": 0.0374, "step": 24517 }, { "epoch": 1.3071386682305273, "grad_norm": 0.8492907881736755, "learning_rate": 5.0966000000000005e-06, "loss": 0.0061, "step": 24518 }, { "epoch": 1.3071919816601802, "grad_norm": 1.9846034049987793, "learning_rate": 5.0964e-06, "loss": 0.0328, "step": 24519 }, { "epoch": 1.307245295089833, "grad_norm": 1.3618206977844238, "learning_rate": 5.0962000000000004e-06, "loss": 0.0262, "step": 24520 }, { "epoch": 1.307298608519486, "grad_norm": 2.598613977432251, "learning_rate": 5.096000000000001e-06, "loss": 0.0305, "step": 24521 }, { "epoch": 1.307351921949139, "grad_norm": 0.58000248670578, "learning_rate": 5.0958e-06, "loss": 0.0039, "step": 24522 }, { "epoch": 1.307405235378792, "grad_norm": 1.7826818227767944, "learning_rate": 5.095600000000001e-06, "loss": 0.0086, "step": 24523 }, { "epoch": 1.307458548808445, "grad_norm": 1.9973747730255127, "learning_rate": 5.0954e-06, "loss": 0.0287, "step": 24524 }, { "epoch": 1.3075118622380977, "grad_norm": 1.7935408353805542, "learning_rate": 5.0952e-06, "loss": 0.0129, "step": 24525 }, { "epoch": 1.3075651756677507, "grad_norm": 1.3049288988113403, "learning_rate": 5.095e-06, "loss": 0.0194, "step": 24526 }, { "epoch": 1.3076184890974036, "grad_norm": 1.3139543533325195, "learning_rate": 5.094800000000001e-06, "loss": 0.0217, "step": 24527 }, { "epoch": 1.3076718025270566, "grad_norm": 5.171678066253662, "learning_rate": 5.0946e-06, "loss": 0.0401, "step": 24528 }, { "epoch": 1.3077251159567096, "grad_norm": 1.444513201713562, "learning_rate": 5.094400000000001e-06, "loss": 0.0081, "step": 24529 }, { "epoch": 1.3077784293863624, "grad_norm": 1.0383533239364624, "learning_rate": 5.0942e-06, "loss": 0.0119, "step": 24530 }, { "epoch": 1.3078317428160153, "grad_norm": 1.1931617259979248, "learning_rate": 5.094e-06, "loss": 0.0067, "step": 24531 }, { "epoch": 1.3078850562456683, "grad_norm": 2.358616828918457, "learning_rate": 5.093800000000001e-06, "loss": 0.0423, "step": 24532 }, { "epoch": 1.3079383696753213, "grad_norm": 0.9472013711929321, "learning_rate": 5.093600000000001e-06, "loss": 0.008, "step": 24533 }, { "epoch": 1.3079916831049743, "grad_norm": 4.093116283416748, "learning_rate": 5.0934e-06, "loss": 0.0278, "step": 24534 }, { "epoch": 1.308044996534627, "grad_norm": 1.9668457508087158, "learning_rate": 5.093200000000001e-06, "loss": 0.0136, "step": 24535 }, { "epoch": 1.30809830996428, "grad_norm": 1.3700450658798218, "learning_rate": 5.093e-06, "loss": 0.0262, "step": 24536 }, { "epoch": 1.308151623393933, "grad_norm": 1.6006759405136108, "learning_rate": 5.0928e-06, "loss": 0.0098, "step": 24537 }, { "epoch": 1.3082049368235857, "grad_norm": 1.9962859153747559, "learning_rate": 5.092600000000001e-06, "loss": 0.0266, "step": 24538 }, { "epoch": 1.3082582502532387, "grad_norm": 1.5398162603378296, "learning_rate": 5.0924000000000006e-06, "loss": 0.0234, "step": 24539 }, { "epoch": 1.3083115636828917, "grad_norm": 2.459444522857666, "learning_rate": 5.0922e-06, "loss": 0.0451, "step": 24540 }, { "epoch": 1.3083648771125447, "grad_norm": 2.249844789505005, "learning_rate": 5.0920000000000005e-06, "loss": 0.0288, "step": 24541 }, { "epoch": 1.3084181905421977, "grad_norm": 1.9236669540405273, "learning_rate": 5.0918e-06, "loss": 0.0118, "step": 24542 }, { "epoch": 1.3084715039718504, "grad_norm": 1.5053741931915283, "learning_rate": 5.0916e-06, "loss": 0.0131, "step": 24543 }, { "epoch": 1.3085248174015034, "grad_norm": 0.5917470455169678, "learning_rate": 5.091400000000001e-06, "loss": 0.0182, "step": 24544 }, { "epoch": 1.3085781308311564, "grad_norm": 1.7741611003875732, "learning_rate": 5.0912000000000005e-06, "loss": 0.0219, "step": 24545 }, { "epoch": 1.3086314442608094, "grad_norm": 1.5072604417800903, "learning_rate": 5.091e-06, "loss": 0.0099, "step": 24546 }, { "epoch": 1.3086847576904623, "grad_norm": 1.340279221534729, "learning_rate": 5.0908000000000005e-06, "loss": 0.024, "step": 24547 }, { "epoch": 1.308738071120115, "grad_norm": 3.0890212059020996, "learning_rate": 5.0906e-06, "loss": 0.0603, "step": 24548 }, { "epoch": 1.308791384549768, "grad_norm": 1.529968023300171, "learning_rate": 5.090400000000001e-06, "loss": 0.0303, "step": 24549 }, { "epoch": 1.308844697979421, "grad_norm": 2.1799705028533936, "learning_rate": 5.090200000000001e-06, "loss": 0.0235, "step": 24550 }, { "epoch": 1.308898011409074, "grad_norm": 1.6039072275161743, "learning_rate": 5.09e-06, "loss": 0.0076, "step": 24551 }, { "epoch": 1.308951324838727, "grad_norm": 1.4005019664764404, "learning_rate": 5.0898e-06, "loss": 0.027, "step": 24552 }, { "epoch": 1.3090046382683798, "grad_norm": 1.549182415008545, "learning_rate": 5.0896e-06, "loss": 0.0301, "step": 24553 }, { "epoch": 1.3090579516980327, "grad_norm": 1.074467420578003, "learning_rate": 5.0894e-06, "loss": 0.0297, "step": 24554 }, { "epoch": 1.3091112651276857, "grad_norm": 1.7154940366744995, "learning_rate": 5.089200000000001e-06, "loss": 0.0195, "step": 24555 }, { "epoch": 1.3091645785573385, "grad_norm": 1.8442209959030151, "learning_rate": 5.089000000000001e-06, "loss": 0.0258, "step": 24556 }, { "epoch": 1.3092178919869915, "grad_norm": 1.5930110216140747, "learning_rate": 5.0888e-06, "loss": 0.0232, "step": 24557 }, { "epoch": 1.3092712054166444, "grad_norm": 2.178492307662964, "learning_rate": 5.0886e-06, "loss": 0.0134, "step": 24558 }, { "epoch": 1.3093245188462974, "grad_norm": 1.2036848068237305, "learning_rate": 5.0884e-06, "loss": 0.047, "step": 24559 }, { "epoch": 1.3093778322759504, "grad_norm": 2.179131269454956, "learning_rate": 5.088200000000001e-06, "loss": 0.0383, "step": 24560 }, { "epoch": 1.3094311457056032, "grad_norm": 1.2539916038513184, "learning_rate": 5.088000000000001e-06, "loss": 0.0077, "step": 24561 }, { "epoch": 1.3094844591352561, "grad_norm": 1.9170480966567993, "learning_rate": 5.087800000000001e-06, "loss": 0.0175, "step": 24562 }, { "epoch": 1.309537772564909, "grad_norm": 0.9348545670509338, "learning_rate": 5.0876e-06, "loss": 0.0205, "step": 24563 }, { "epoch": 1.309591085994562, "grad_norm": 2.3933329582214355, "learning_rate": 5.0874e-06, "loss": 0.0406, "step": 24564 }, { "epoch": 1.309644399424215, "grad_norm": 3.1596457958221436, "learning_rate": 5.0872e-06, "loss": 0.0299, "step": 24565 }, { "epoch": 1.3096977128538678, "grad_norm": 2.9376747608184814, "learning_rate": 5.087000000000001e-06, "loss": 0.0435, "step": 24566 }, { "epoch": 1.3097510262835208, "grad_norm": 3.117159605026245, "learning_rate": 5.086800000000001e-06, "loss": 0.0365, "step": 24567 }, { "epoch": 1.3098043397131738, "grad_norm": 0.9928560256958008, "learning_rate": 5.0866000000000006e-06, "loss": 0.0227, "step": 24568 }, { "epoch": 1.3098576531428268, "grad_norm": 2.1007328033447266, "learning_rate": 5.0864e-06, "loss": 0.0392, "step": 24569 }, { "epoch": 1.3099109665724797, "grad_norm": 1.4014475345611572, "learning_rate": 5.0862e-06, "loss": 0.0119, "step": 24570 }, { "epoch": 1.3099642800021325, "grad_norm": 2.2341978549957275, "learning_rate": 5.086000000000001e-06, "loss": 0.0222, "step": 24571 }, { "epoch": 1.3100175934317855, "grad_norm": 2.13680362701416, "learning_rate": 5.0858000000000005e-06, "loss": 0.0319, "step": 24572 }, { "epoch": 1.3100709068614385, "grad_norm": 1.6818445920944214, "learning_rate": 5.085600000000001e-06, "loss": 0.0184, "step": 24573 }, { "epoch": 1.3101242202910912, "grad_norm": 0.7181435227394104, "learning_rate": 5.0854000000000005e-06, "loss": 0.007, "step": 24574 }, { "epoch": 1.3101775337207442, "grad_norm": 3.073901891708374, "learning_rate": 5.0852e-06, "loss": 0.0183, "step": 24575 }, { "epoch": 1.3102308471503972, "grad_norm": 1.2532347440719604, "learning_rate": 5.085e-06, "loss": 0.0145, "step": 24576 }, { "epoch": 1.3102841605800502, "grad_norm": 1.3006294965744019, "learning_rate": 5.084800000000001e-06, "loss": 0.0164, "step": 24577 }, { "epoch": 1.3103374740097031, "grad_norm": 0.967494785785675, "learning_rate": 5.0846e-06, "loss": 0.0055, "step": 24578 }, { "epoch": 1.3103907874393559, "grad_norm": 0.6928796172142029, "learning_rate": 5.084400000000001e-06, "loss": 0.0088, "step": 24579 }, { "epoch": 1.3104441008690089, "grad_norm": 0.8665429353713989, "learning_rate": 5.0842e-06, "loss": 0.0144, "step": 24580 }, { "epoch": 1.3104974142986618, "grad_norm": 1.7999801635742188, "learning_rate": 5.084e-06, "loss": 0.0367, "step": 24581 }, { "epoch": 1.3105507277283148, "grad_norm": 2.105285882949829, "learning_rate": 5.083800000000001e-06, "loss": 0.0513, "step": 24582 }, { "epoch": 1.3106040411579678, "grad_norm": 1.6412171125411987, "learning_rate": 5.083600000000001e-06, "loss": 0.0427, "step": 24583 }, { "epoch": 1.3106573545876206, "grad_norm": 1.4221282005310059, "learning_rate": 5.0834e-06, "loss": 0.0239, "step": 24584 }, { "epoch": 1.3107106680172735, "grad_norm": 1.4678577184677124, "learning_rate": 5.083200000000001e-06, "loss": 0.0198, "step": 24585 }, { "epoch": 1.3107639814469265, "grad_norm": 0.6313870549201965, "learning_rate": 5.083e-06, "loss": 0.0055, "step": 24586 }, { "epoch": 1.3108172948765793, "grad_norm": 2.151944398880005, "learning_rate": 5.0828e-06, "loss": 0.025, "step": 24587 }, { "epoch": 1.3108706083062325, "grad_norm": 1.324406623840332, "learning_rate": 5.082600000000001e-06, "loss": 0.0096, "step": 24588 }, { "epoch": 1.3109239217358852, "grad_norm": 1.2870171070098877, "learning_rate": 5.082400000000001e-06, "loss": 0.0212, "step": 24589 }, { "epoch": 1.3109772351655382, "grad_norm": 1.5807969570159912, "learning_rate": 5.0822e-06, "loss": 0.0109, "step": 24590 }, { "epoch": 1.3110305485951912, "grad_norm": 2.122150182723999, "learning_rate": 5.082000000000001e-06, "loss": 0.0294, "step": 24591 }, { "epoch": 1.311083862024844, "grad_norm": 2.754638433456421, "learning_rate": 5.0818e-06, "loss": 0.0679, "step": 24592 }, { "epoch": 1.311137175454497, "grad_norm": 2.215273380279541, "learning_rate": 5.0816e-06, "loss": 0.0189, "step": 24593 }, { "epoch": 1.31119048888415, "grad_norm": 1.114013433456421, "learning_rate": 5.081400000000001e-06, "loss": 0.0199, "step": 24594 }, { "epoch": 1.3112438023138029, "grad_norm": 0.8549004197120667, "learning_rate": 5.081200000000001e-06, "loss": 0.0092, "step": 24595 }, { "epoch": 1.3112971157434559, "grad_norm": 1.7437777519226074, "learning_rate": 5.081e-06, "loss": 0.0411, "step": 24596 }, { "epoch": 1.3113504291731086, "grad_norm": 2.375901460647583, "learning_rate": 5.0808000000000006e-06, "loss": 0.0232, "step": 24597 }, { "epoch": 1.3114037426027616, "grad_norm": 1.0361778736114502, "learning_rate": 5.0806e-06, "loss": 0.0135, "step": 24598 }, { "epoch": 1.3114570560324146, "grad_norm": 1.2314447164535522, "learning_rate": 5.0804000000000005e-06, "loss": 0.0238, "step": 24599 }, { "epoch": 1.3115103694620676, "grad_norm": 1.0781433582305908, "learning_rate": 5.080200000000001e-06, "loss": 0.015, "step": 24600 }, { "epoch": 1.3115636828917205, "grad_norm": 1.249617099761963, "learning_rate": 5.0800000000000005e-06, "loss": 0.0101, "step": 24601 }, { "epoch": 1.3116169963213733, "grad_norm": 1.4729397296905518, "learning_rate": 5.0798e-06, "loss": 0.0122, "step": 24602 }, { "epoch": 1.3116703097510263, "grad_norm": 3.755939245223999, "learning_rate": 5.0796000000000005e-06, "loss": 0.0345, "step": 24603 }, { "epoch": 1.3117236231806793, "grad_norm": 1.8825671672821045, "learning_rate": 5.0794e-06, "loss": 0.0226, "step": 24604 }, { "epoch": 1.311776936610332, "grad_norm": 1.3567798137664795, "learning_rate": 5.0792000000000004e-06, "loss": 0.0305, "step": 24605 }, { "epoch": 1.311830250039985, "grad_norm": 2.1711270809173584, "learning_rate": 5.079000000000001e-06, "loss": 0.0261, "step": 24606 }, { "epoch": 1.311883563469638, "grad_norm": 2.936253786087036, "learning_rate": 5.0788e-06, "loss": 0.0516, "step": 24607 }, { "epoch": 1.311936876899291, "grad_norm": 1.0410749912261963, "learning_rate": 5.0786e-06, "loss": 0.0166, "step": 24608 }, { "epoch": 1.311990190328944, "grad_norm": 0.9716783761978149, "learning_rate": 5.0784e-06, "loss": 0.0085, "step": 24609 }, { "epoch": 1.3120435037585967, "grad_norm": 2.583590030670166, "learning_rate": 5.078200000000001e-06, "loss": 0.0626, "step": 24610 }, { "epoch": 1.3120968171882497, "grad_norm": 1.287980318069458, "learning_rate": 5.078e-06, "loss": 0.0206, "step": 24611 }, { "epoch": 1.3121501306179026, "grad_norm": 1.6153638362884521, "learning_rate": 5.077800000000001e-06, "loss": 0.0321, "step": 24612 }, { "epoch": 1.3122034440475556, "grad_norm": 1.760898232460022, "learning_rate": 5.0776e-06, "loss": 0.021, "step": 24613 }, { "epoch": 1.3122567574772086, "grad_norm": 2.14347243309021, "learning_rate": 5.0774e-06, "loss": 0.0338, "step": 24614 }, { "epoch": 1.3123100709068614, "grad_norm": 1.4310791492462158, "learning_rate": 5.0772e-06, "loss": 0.0388, "step": 24615 }, { "epoch": 1.3123633843365143, "grad_norm": 1.520528793334961, "learning_rate": 5.077000000000001e-06, "loss": 0.0218, "step": 24616 }, { "epoch": 1.3124166977661673, "grad_norm": 2.9406068325042725, "learning_rate": 5.0768e-06, "loss": 0.0272, "step": 24617 }, { "epoch": 1.3124700111958203, "grad_norm": 1.9598909616470337, "learning_rate": 5.076600000000001e-06, "loss": 0.0246, "step": 24618 }, { "epoch": 1.3125233246254733, "grad_norm": 1.740094780921936, "learning_rate": 5.0764e-06, "loss": 0.0319, "step": 24619 }, { "epoch": 1.312576638055126, "grad_norm": 1.9571750164031982, "learning_rate": 5.0762e-06, "loss": 0.0125, "step": 24620 }, { "epoch": 1.312629951484779, "grad_norm": 1.765474557876587, "learning_rate": 5.076000000000001e-06, "loss": 0.0526, "step": 24621 }, { "epoch": 1.312683264914432, "grad_norm": 1.15009343624115, "learning_rate": 5.075800000000001e-06, "loss": 0.0202, "step": 24622 }, { "epoch": 1.3127365783440847, "grad_norm": 0.45223522186279297, "learning_rate": 5.0756e-06, "loss": 0.0046, "step": 24623 }, { "epoch": 1.3127898917737377, "grad_norm": 2.5072507858276367, "learning_rate": 5.0754000000000006e-06, "loss": 0.0298, "step": 24624 }, { "epoch": 1.3128432052033907, "grad_norm": 4.996823787689209, "learning_rate": 5.0752e-06, "loss": 0.0284, "step": 24625 }, { "epoch": 1.3128965186330437, "grad_norm": 0.6812435984611511, "learning_rate": 5.075e-06, "loss": 0.016, "step": 24626 }, { "epoch": 1.3129498320626967, "grad_norm": 2.4943675994873047, "learning_rate": 5.074800000000001e-06, "loss": 0.0493, "step": 24627 }, { "epoch": 1.3130031454923494, "grad_norm": 2.757477283477783, "learning_rate": 5.0746000000000005e-06, "loss": 0.0392, "step": 24628 }, { "epoch": 1.3130564589220024, "grad_norm": 3.419996976852417, "learning_rate": 5.0744e-06, "loss": 0.0173, "step": 24629 }, { "epoch": 1.3131097723516554, "grad_norm": 2.2498722076416016, "learning_rate": 5.0742000000000005e-06, "loss": 0.0178, "step": 24630 }, { "epoch": 1.3131630857813084, "grad_norm": 1.7518314123153687, "learning_rate": 5.074e-06, "loss": 0.0145, "step": 24631 }, { "epoch": 1.3132163992109613, "grad_norm": 1.3440492153167725, "learning_rate": 5.073800000000001e-06, "loss": 0.0114, "step": 24632 }, { "epoch": 1.313269712640614, "grad_norm": 1.0734095573425293, "learning_rate": 5.073600000000001e-06, "loss": 0.0205, "step": 24633 }, { "epoch": 1.313323026070267, "grad_norm": 1.8976932764053345, "learning_rate": 5.0734e-06, "loss": 0.0172, "step": 24634 }, { "epoch": 1.31337633949992, "grad_norm": 1.3153716325759888, "learning_rate": 5.0732e-06, "loss": 0.0159, "step": 24635 }, { "epoch": 1.313429652929573, "grad_norm": 2.3278648853302, "learning_rate": 5.073e-06, "loss": 0.0441, "step": 24636 }, { "epoch": 1.313482966359226, "grad_norm": 2.783202886581421, "learning_rate": 5.0728e-06, "loss": 0.0477, "step": 24637 }, { "epoch": 1.3135362797888788, "grad_norm": 2.209033489227295, "learning_rate": 5.072600000000001e-06, "loss": 0.0445, "step": 24638 }, { "epoch": 1.3135895932185317, "grad_norm": 1.9225258827209473, "learning_rate": 5.072400000000001e-06, "loss": 0.035, "step": 24639 }, { "epoch": 1.3136429066481847, "grad_norm": 0.9114144444465637, "learning_rate": 5.0722e-06, "loss": 0.0232, "step": 24640 }, { "epoch": 1.3136962200778375, "grad_norm": 0.6248989701271057, "learning_rate": 5.072e-06, "loss": 0.0139, "step": 24641 }, { "epoch": 1.3137495335074905, "grad_norm": 2.178044319152832, "learning_rate": 5.0718e-06, "loss": 0.029, "step": 24642 }, { "epoch": 1.3138028469371434, "grad_norm": 2.5719144344329834, "learning_rate": 5.071600000000001e-06, "loss": 0.0407, "step": 24643 }, { "epoch": 1.3138561603667964, "grad_norm": 1.8415297269821167, "learning_rate": 5.071400000000001e-06, "loss": 0.0366, "step": 24644 }, { "epoch": 1.3139094737964494, "grad_norm": 1.3627458810806274, "learning_rate": 5.071200000000001e-06, "loss": 0.017, "step": 24645 }, { "epoch": 1.3139627872261022, "grad_norm": 3.0013763904571533, "learning_rate": 5.071e-06, "loss": 0.0514, "step": 24646 }, { "epoch": 1.3140161006557551, "grad_norm": 0.8359755277633667, "learning_rate": 5.0708e-06, "loss": 0.0098, "step": 24647 }, { "epoch": 1.3140694140854081, "grad_norm": 0.9029554724693298, "learning_rate": 5.0706e-06, "loss": 0.0095, "step": 24648 }, { "epoch": 1.314122727515061, "grad_norm": 1.556888222694397, "learning_rate": 5.070400000000001e-06, "loss": 0.024, "step": 24649 }, { "epoch": 1.314176040944714, "grad_norm": 2.106393575668335, "learning_rate": 5.070200000000001e-06, "loss": 0.0163, "step": 24650 }, { "epoch": 1.3142293543743668, "grad_norm": 2.2334342002868652, "learning_rate": 5.070000000000001e-06, "loss": 0.0359, "step": 24651 }, { "epoch": 1.3142826678040198, "grad_norm": 0.6877889633178711, "learning_rate": 5.0698e-06, "loss": 0.0089, "step": 24652 }, { "epoch": 1.3143359812336728, "grad_norm": 2.0005593299865723, "learning_rate": 5.0696e-06, "loss": 0.0212, "step": 24653 }, { "epoch": 1.3143892946633258, "grad_norm": 1.6591750383377075, "learning_rate": 5.0694e-06, "loss": 0.0117, "step": 24654 }, { "epoch": 1.3144426080929787, "grad_norm": 2.164093017578125, "learning_rate": 5.0692000000000005e-06, "loss": 0.0395, "step": 24655 }, { "epoch": 1.3144959215226315, "grad_norm": 2.424450397491455, "learning_rate": 5.069000000000001e-06, "loss": 0.0409, "step": 24656 }, { "epoch": 1.3145492349522845, "grad_norm": 0.8533957600593567, "learning_rate": 5.0688000000000005e-06, "loss": 0.0117, "step": 24657 }, { "epoch": 1.3146025483819375, "grad_norm": 1.7824318408966064, "learning_rate": 5.0686e-06, "loss": 0.0093, "step": 24658 }, { "epoch": 1.3146558618115902, "grad_norm": 1.3441269397735596, "learning_rate": 5.0684e-06, "loss": 0.0116, "step": 24659 }, { "epoch": 1.3147091752412432, "grad_norm": 1.5065412521362305, "learning_rate": 5.068200000000001e-06, "loss": 0.0349, "step": 24660 }, { "epoch": 1.3147624886708962, "grad_norm": 1.0655609369277954, "learning_rate": 5.0680000000000004e-06, "loss": 0.0191, "step": 24661 }, { "epoch": 1.3148158021005492, "grad_norm": 1.0768946409225464, "learning_rate": 5.067800000000001e-06, "loss": 0.0281, "step": 24662 }, { "epoch": 1.3148691155302021, "grad_norm": 1.8451557159423828, "learning_rate": 5.0676e-06, "loss": 0.0081, "step": 24663 }, { "epoch": 1.314922428959855, "grad_norm": 2.3066506385803223, "learning_rate": 5.0674e-06, "loss": 0.0418, "step": 24664 }, { "epoch": 1.3149757423895079, "grad_norm": 2.1723110675811768, "learning_rate": 5.0671999999999995e-06, "loss": 0.0285, "step": 24665 }, { "epoch": 1.3150290558191609, "grad_norm": 2.0962157249450684, "learning_rate": 5.067000000000001e-06, "loss": 0.0161, "step": 24666 }, { "epoch": 1.3150823692488138, "grad_norm": 1.4582507610321045, "learning_rate": 5.0668e-06, "loss": 0.0143, "step": 24667 }, { "epoch": 1.3151356826784668, "grad_norm": 0.6842519640922546, "learning_rate": 5.066600000000001e-06, "loss": 0.0046, "step": 24668 }, { "epoch": 1.3151889961081196, "grad_norm": 2.160733699798584, "learning_rate": 5.0664e-06, "loss": 0.0404, "step": 24669 }, { "epoch": 1.3152423095377725, "grad_norm": 2.288801431655884, "learning_rate": 5.0662e-06, "loss": 0.0351, "step": 24670 }, { "epoch": 1.3152956229674255, "grad_norm": 3.0318338871002197, "learning_rate": 5.066000000000001e-06, "loss": 0.0405, "step": 24671 }, { "epoch": 1.3153489363970785, "grad_norm": 0.8489753007888794, "learning_rate": 5.065800000000001e-06, "loss": 0.0048, "step": 24672 }, { "epoch": 1.3154022498267315, "grad_norm": 3.0423216819763184, "learning_rate": 5.0656e-06, "loss": 0.023, "step": 24673 }, { "epoch": 1.3154555632563842, "grad_norm": 2.261085033416748, "learning_rate": 5.065400000000001e-06, "loss": 0.0417, "step": 24674 }, { "epoch": 1.3155088766860372, "grad_norm": 1.7525734901428223, "learning_rate": 5.0652e-06, "loss": 0.0151, "step": 24675 }, { "epoch": 1.3155621901156902, "grad_norm": 4.3048577308654785, "learning_rate": 5.065e-06, "loss": 0.0294, "step": 24676 }, { "epoch": 1.315615503545343, "grad_norm": 0.9433459639549255, "learning_rate": 5.064800000000001e-06, "loss": 0.0058, "step": 24677 }, { "epoch": 1.315668816974996, "grad_norm": 2.8442773818969727, "learning_rate": 5.064600000000001e-06, "loss": 0.0379, "step": 24678 }, { "epoch": 1.315722130404649, "grad_norm": 2.473536729812622, "learning_rate": 5.0644e-06, "loss": 0.0433, "step": 24679 }, { "epoch": 1.315775443834302, "grad_norm": 1.4899028539657593, "learning_rate": 5.0642000000000006e-06, "loss": 0.0074, "step": 24680 }, { "epoch": 1.3158287572639549, "grad_norm": 0.9500600695610046, "learning_rate": 5.064e-06, "loss": 0.0066, "step": 24681 }, { "epoch": 1.3158820706936076, "grad_norm": 0.9698001146316528, "learning_rate": 5.0638000000000005e-06, "loss": 0.0093, "step": 24682 }, { "epoch": 1.3159353841232606, "grad_norm": 1.1111836433410645, "learning_rate": 5.063600000000001e-06, "loss": 0.0151, "step": 24683 }, { "epoch": 1.3159886975529136, "grad_norm": 2.3992936611175537, "learning_rate": 5.0634000000000005e-06, "loss": 0.0157, "step": 24684 }, { "epoch": 1.3160420109825666, "grad_norm": 1.2134695053100586, "learning_rate": 5.0632e-06, "loss": 0.0179, "step": 24685 }, { "epoch": 1.3160953244122195, "grad_norm": 2.2861435413360596, "learning_rate": 5.0630000000000005e-06, "loss": 0.0167, "step": 24686 }, { "epoch": 1.3161486378418723, "grad_norm": 1.9800176620483398, "learning_rate": 5.0628e-06, "loss": 0.0271, "step": 24687 }, { "epoch": 1.3162019512715253, "grad_norm": 2.215644598007202, "learning_rate": 5.0626000000000005e-06, "loss": 0.0229, "step": 24688 }, { "epoch": 1.3162552647011783, "grad_norm": 0.5477946996688843, "learning_rate": 5.062400000000001e-06, "loss": 0.0032, "step": 24689 }, { "epoch": 1.316308578130831, "grad_norm": 2.61950945854187, "learning_rate": 5.0622e-06, "loss": 0.0216, "step": 24690 }, { "epoch": 1.3163618915604842, "grad_norm": 1.3041729927062988, "learning_rate": 5.062e-06, "loss": 0.0228, "step": 24691 }, { "epoch": 1.316415204990137, "grad_norm": 2.853830337524414, "learning_rate": 5.0618e-06, "loss": 0.0097, "step": 24692 }, { "epoch": 1.31646851841979, "grad_norm": 1.3570431470870972, "learning_rate": 5.061600000000001e-06, "loss": 0.0192, "step": 24693 }, { "epoch": 1.316521831849443, "grad_norm": 2.2511637210845947, "learning_rate": 5.0614e-06, "loss": 0.0292, "step": 24694 }, { "epoch": 1.3165751452790957, "grad_norm": 1.7360174655914307, "learning_rate": 5.061200000000001e-06, "loss": 0.0129, "step": 24695 }, { "epoch": 1.3166284587087487, "grad_norm": 1.308085560798645, "learning_rate": 5.061e-06, "loss": 0.0185, "step": 24696 }, { "epoch": 1.3166817721384017, "grad_norm": 2.9012012481689453, "learning_rate": 5.0608e-06, "loss": 0.0574, "step": 24697 }, { "epoch": 1.3167350855680546, "grad_norm": 7.3105621337890625, "learning_rate": 5.0606e-06, "loss": 0.0441, "step": 24698 }, { "epoch": 1.3167883989977076, "grad_norm": 2.5516610145568848, "learning_rate": 5.060400000000001e-06, "loss": 0.035, "step": 24699 }, { "epoch": 1.3168417124273604, "grad_norm": 2.7709994316101074, "learning_rate": 5.0602e-06, "loss": 0.0447, "step": 24700 }, { "epoch": 1.3168950258570133, "grad_norm": 4.058665752410889, "learning_rate": 5.060000000000001e-06, "loss": 0.0429, "step": 24701 }, { "epoch": 1.3169483392866663, "grad_norm": 5.677194118499756, "learning_rate": 5.0598e-06, "loss": 0.0493, "step": 24702 }, { "epoch": 1.3170016527163193, "grad_norm": 0.992691159248352, "learning_rate": 5.0596e-06, "loss": 0.0145, "step": 24703 }, { "epoch": 1.3170549661459723, "grad_norm": 2.3129477500915527, "learning_rate": 5.0594e-06, "loss": 0.0261, "step": 24704 }, { "epoch": 1.317108279575625, "grad_norm": 0.4053088426589966, "learning_rate": 5.059200000000001e-06, "loss": 0.0045, "step": 24705 }, { "epoch": 1.317161593005278, "grad_norm": 2.320265293121338, "learning_rate": 5.059e-06, "loss": 0.0126, "step": 24706 }, { "epoch": 1.317214906434931, "grad_norm": 0.6562234163284302, "learning_rate": 5.058800000000001e-06, "loss": 0.0195, "step": 24707 }, { "epoch": 1.3172682198645838, "grad_norm": 4.5061845779418945, "learning_rate": 5.0586e-06, "loss": 0.0397, "step": 24708 }, { "epoch": 1.3173215332942367, "grad_norm": 0.7332583665847778, "learning_rate": 5.0584e-06, "loss": 0.007, "step": 24709 }, { "epoch": 1.3173748467238897, "grad_norm": 3.4559221267700195, "learning_rate": 5.058200000000001e-06, "loss": 0.0399, "step": 24710 }, { "epoch": 1.3174281601535427, "grad_norm": 1.9592276811599731, "learning_rate": 5.0580000000000005e-06, "loss": 0.0298, "step": 24711 }, { "epoch": 1.3174814735831957, "grad_norm": 2.5688843727111816, "learning_rate": 5.0578e-06, "loss": 0.0227, "step": 24712 }, { "epoch": 1.3175347870128484, "grad_norm": 0.8437279462814331, "learning_rate": 5.0576000000000005e-06, "loss": 0.0083, "step": 24713 }, { "epoch": 1.3175881004425014, "grad_norm": 0.8350518941879272, "learning_rate": 5.0574e-06, "loss": 0.008, "step": 24714 }, { "epoch": 1.3176414138721544, "grad_norm": 3.714555025100708, "learning_rate": 5.0572e-06, "loss": 0.0341, "step": 24715 }, { "epoch": 1.3176947273018074, "grad_norm": 1.7517006397247314, "learning_rate": 5.057000000000001e-06, "loss": 0.0258, "step": 24716 }, { "epoch": 1.3177480407314603, "grad_norm": 2.3715357780456543, "learning_rate": 5.0568000000000004e-06, "loss": 0.0263, "step": 24717 }, { "epoch": 1.317801354161113, "grad_norm": 2.13192081451416, "learning_rate": 5.0566e-06, "loss": 0.0187, "step": 24718 }, { "epoch": 1.317854667590766, "grad_norm": 2.5064783096313477, "learning_rate": 5.0564e-06, "loss": 0.0211, "step": 24719 }, { "epoch": 1.317907981020419, "grad_norm": 1.3417339324951172, "learning_rate": 5.0562e-06, "loss": 0.0148, "step": 24720 }, { "epoch": 1.317961294450072, "grad_norm": 1.4852381944656372, "learning_rate": 5.056000000000001e-06, "loss": 0.0349, "step": 24721 }, { "epoch": 1.318014607879725, "grad_norm": 1.1596344709396362, "learning_rate": 5.055800000000001e-06, "loss": 0.0163, "step": 24722 }, { "epoch": 1.3180679213093778, "grad_norm": 1.1968722343444824, "learning_rate": 5.0556e-06, "loss": 0.009, "step": 24723 }, { "epoch": 1.3181212347390308, "grad_norm": 0.7969561219215393, "learning_rate": 5.0554e-06, "loss": 0.0064, "step": 24724 }, { "epoch": 1.3181745481686837, "grad_norm": 3.484250545501709, "learning_rate": 5.0552e-06, "loss": 0.0597, "step": 24725 }, { "epoch": 1.3182278615983365, "grad_norm": 1.227146863937378, "learning_rate": 5.055e-06, "loss": 0.0163, "step": 24726 }, { "epoch": 1.3182811750279895, "grad_norm": 1.7861249446868896, "learning_rate": 5.054800000000001e-06, "loss": 0.0241, "step": 24727 }, { "epoch": 1.3183344884576425, "grad_norm": 1.565984845161438, "learning_rate": 5.054600000000001e-06, "loss": 0.0261, "step": 24728 }, { "epoch": 1.3183878018872954, "grad_norm": 1.3448606729507446, "learning_rate": 5.0544e-06, "loss": 0.053, "step": 24729 }, { "epoch": 1.3184411153169484, "grad_norm": 2.118269920349121, "learning_rate": 5.0542e-06, "loss": 0.0256, "step": 24730 }, { "epoch": 1.3184944287466012, "grad_norm": 1.3476289510726929, "learning_rate": 5.054e-06, "loss": 0.0226, "step": 24731 }, { "epoch": 1.3185477421762541, "grad_norm": 2.947873830795288, "learning_rate": 5.053800000000001e-06, "loss": 0.0286, "step": 24732 }, { "epoch": 1.3186010556059071, "grad_norm": 1.708414077758789, "learning_rate": 5.053600000000001e-06, "loss": 0.0337, "step": 24733 }, { "epoch": 1.31865436903556, "grad_norm": 1.9991134405136108, "learning_rate": 5.053400000000001e-06, "loss": 0.0394, "step": 24734 }, { "epoch": 1.318707682465213, "grad_norm": 1.337405800819397, "learning_rate": 5.0532e-06, "loss": 0.0139, "step": 24735 }, { "epoch": 1.3187609958948658, "grad_norm": 2.6770050525665283, "learning_rate": 5.053e-06, "loss": 0.0662, "step": 24736 }, { "epoch": 1.3188143093245188, "grad_norm": 1.6850980520248413, "learning_rate": 5.0528e-06, "loss": 0.0197, "step": 24737 }, { "epoch": 1.3188676227541718, "grad_norm": 1.7386531829833984, "learning_rate": 5.0526000000000005e-06, "loss": 0.0166, "step": 24738 }, { "epoch": 1.3189209361838248, "grad_norm": 0.7265413403511047, "learning_rate": 5.052400000000001e-06, "loss": 0.0062, "step": 24739 }, { "epoch": 1.3189742496134778, "grad_norm": 1.2002675533294678, "learning_rate": 5.0522000000000005e-06, "loss": 0.0321, "step": 24740 }, { "epoch": 1.3190275630431305, "grad_norm": 2.351130485534668, "learning_rate": 5.052e-06, "loss": 0.0216, "step": 24741 }, { "epoch": 1.3190808764727835, "grad_norm": 1.1790177822113037, "learning_rate": 5.0518e-06, "loss": 0.016, "step": 24742 }, { "epoch": 1.3191341899024365, "grad_norm": 2.3348021507263184, "learning_rate": 5.051600000000001e-06, "loss": 0.0243, "step": 24743 }, { "epoch": 1.3191875033320892, "grad_norm": 1.6318538188934326, "learning_rate": 5.0514000000000005e-06, "loss": 0.0139, "step": 24744 }, { "epoch": 1.3192408167617422, "grad_norm": 3.6391353607177734, "learning_rate": 5.051200000000001e-06, "loss": 0.0198, "step": 24745 }, { "epoch": 1.3192941301913952, "grad_norm": 1.3925905227661133, "learning_rate": 5.0510000000000004e-06, "loss": 0.021, "step": 24746 }, { "epoch": 1.3193474436210482, "grad_norm": 3.035496473312378, "learning_rate": 5.0508e-06, "loss": 0.061, "step": 24747 }, { "epoch": 1.3194007570507011, "grad_norm": 1.723772406578064, "learning_rate": 5.0505999999999996e-06, "loss": 0.0328, "step": 24748 }, { "epoch": 1.319454070480354, "grad_norm": 1.6188960075378418, "learning_rate": 5.050400000000001e-06, "loss": 0.0235, "step": 24749 }, { "epoch": 1.3195073839100069, "grad_norm": 0.7735746502876282, "learning_rate": 5.0502e-06, "loss": 0.0079, "step": 24750 }, { "epoch": 1.3195606973396599, "grad_norm": 1.5546667575836182, "learning_rate": 5.050000000000001e-06, "loss": 0.0205, "step": 24751 }, { "epoch": 1.3196140107693128, "grad_norm": 0.83701491355896, "learning_rate": 5.0498e-06, "loss": 0.0082, "step": 24752 }, { "epoch": 1.3196673241989658, "grad_norm": 3.1303505897521973, "learning_rate": 5.0496e-06, "loss": 0.037, "step": 24753 }, { "epoch": 1.3197206376286186, "grad_norm": 3.6726410388946533, "learning_rate": 5.049400000000001e-06, "loss": 0.019, "step": 24754 }, { "epoch": 1.3197739510582716, "grad_norm": 3.8900606632232666, "learning_rate": 5.049200000000001e-06, "loss": 0.0186, "step": 24755 }, { "epoch": 1.3198272644879245, "grad_norm": 2.9179368019104004, "learning_rate": 5.049e-06, "loss": 0.023, "step": 24756 }, { "epoch": 1.3198805779175775, "grad_norm": 1.7225648164749146, "learning_rate": 5.048800000000001e-06, "loss": 0.0223, "step": 24757 }, { "epoch": 1.3199338913472305, "grad_norm": 2.578822135925293, "learning_rate": 5.0486e-06, "loss": 0.0287, "step": 24758 }, { "epoch": 1.3199872047768832, "grad_norm": 1.5044697523117065, "learning_rate": 5.0484e-06, "loss": 0.0132, "step": 24759 }, { "epoch": 1.3200405182065362, "grad_norm": 0.9998341202735901, "learning_rate": 5.048200000000001e-06, "loss": 0.0229, "step": 24760 }, { "epoch": 1.3200938316361892, "grad_norm": 2.07267165184021, "learning_rate": 5.048000000000001e-06, "loss": 0.0247, "step": 24761 }, { "epoch": 1.320147145065842, "grad_norm": 1.7772079706192017, "learning_rate": 5.0478e-06, "loss": 0.0253, "step": 24762 }, { "epoch": 1.320200458495495, "grad_norm": 1.8918366432189941, "learning_rate": 5.047600000000001e-06, "loss": 0.0276, "step": 24763 }, { "epoch": 1.320253771925148, "grad_norm": 1.0243667364120483, "learning_rate": 5.0474e-06, "loss": 0.01, "step": 24764 }, { "epoch": 1.320307085354801, "grad_norm": 2.0638270378112793, "learning_rate": 5.0472e-06, "loss": 0.0314, "step": 24765 }, { "epoch": 1.3203603987844539, "grad_norm": 1.4379817247390747, "learning_rate": 5.047000000000001e-06, "loss": 0.0277, "step": 24766 }, { "epoch": 1.3204137122141066, "grad_norm": 1.6623225212097168, "learning_rate": 5.0468000000000005e-06, "loss": 0.0183, "step": 24767 }, { "epoch": 1.3204670256437596, "grad_norm": 2.5771710872650146, "learning_rate": 5.0466e-06, "loss": 0.0491, "step": 24768 }, { "epoch": 1.3205203390734126, "grad_norm": 1.5650734901428223, "learning_rate": 5.0464000000000005e-06, "loss": 0.0363, "step": 24769 }, { "epoch": 1.3205736525030656, "grad_norm": 1.9359893798828125, "learning_rate": 5.0462e-06, "loss": 0.0435, "step": 24770 }, { "epoch": 1.3206269659327186, "grad_norm": 1.2443702220916748, "learning_rate": 5.0460000000000005e-06, "loss": 0.0106, "step": 24771 }, { "epoch": 1.3206802793623713, "grad_norm": 1.6955764293670654, "learning_rate": 5.045800000000001e-06, "loss": 0.0122, "step": 24772 }, { "epoch": 1.3207335927920243, "grad_norm": 0.8557663559913635, "learning_rate": 5.0456000000000004e-06, "loss": 0.0102, "step": 24773 }, { "epoch": 1.3207869062216773, "grad_norm": 2.184349536895752, "learning_rate": 5.0454e-06, "loss": 0.0572, "step": 24774 }, { "epoch": 1.3208402196513302, "grad_norm": 2.3456509113311768, "learning_rate": 5.0452e-06, "loss": 0.0386, "step": 24775 }, { "epoch": 1.3208935330809832, "grad_norm": 2.199049949645996, "learning_rate": 5.045e-06, "loss": 0.0192, "step": 24776 }, { "epoch": 1.320946846510636, "grad_norm": 2.0236833095550537, "learning_rate": 5.0448e-06, "loss": 0.0387, "step": 24777 }, { "epoch": 1.321000159940289, "grad_norm": 2.0302717685699463, "learning_rate": 5.044600000000001e-06, "loss": 0.0272, "step": 24778 }, { "epoch": 1.321053473369942, "grad_norm": 1.935858130455017, "learning_rate": 5.0444e-06, "loss": 0.0195, "step": 24779 }, { "epoch": 1.3211067867995947, "grad_norm": 1.2469791173934937, "learning_rate": 5.0442e-06, "loss": 0.0178, "step": 24780 }, { "epoch": 1.3211601002292477, "grad_norm": 1.7440403699874878, "learning_rate": 5.044e-06, "loss": 0.0473, "step": 24781 }, { "epoch": 1.3212134136589007, "grad_norm": 1.8874622583389282, "learning_rate": 5.043800000000001e-06, "loss": 0.0272, "step": 24782 }, { "epoch": 1.3212667270885536, "grad_norm": 0.5936170220375061, "learning_rate": 5.0436e-06, "loss": 0.0064, "step": 24783 }, { "epoch": 1.3213200405182066, "grad_norm": 1.2430390119552612, "learning_rate": 5.043400000000001e-06, "loss": 0.0085, "step": 24784 }, { "epoch": 1.3213733539478594, "grad_norm": 1.3697056770324707, "learning_rate": 5.0432e-06, "loss": 0.0215, "step": 24785 }, { "epoch": 1.3214266673775124, "grad_norm": 0.9879603385925293, "learning_rate": 5.043e-06, "loss": 0.0109, "step": 24786 }, { "epoch": 1.3214799808071653, "grad_norm": 0.750667154788971, "learning_rate": 5.0428e-06, "loss": 0.0077, "step": 24787 }, { "epoch": 1.3215332942368183, "grad_norm": 1.769303321838379, "learning_rate": 5.042600000000001e-06, "loss": 0.0174, "step": 24788 }, { "epoch": 1.3215866076664713, "grad_norm": 2.213773250579834, "learning_rate": 5.0424e-06, "loss": 0.0148, "step": 24789 }, { "epoch": 1.321639921096124, "grad_norm": 0.6053618788719177, "learning_rate": 5.042200000000001e-06, "loss": 0.0056, "step": 24790 }, { "epoch": 1.321693234525777, "grad_norm": 1.8673516511917114, "learning_rate": 5.042e-06, "loss": 0.0245, "step": 24791 }, { "epoch": 1.32174654795543, "grad_norm": 1.2161877155303955, "learning_rate": 5.0418e-06, "loss": 0.0217, "step": 24792 }, { "epoch": 1.3217998613850828, "grad_norm": 1.6365952491760254, "learning_rate": 5.041600000000001e-06, "loss": 0.0295, "step": 24793 }, { "epoch": 1.321853174814736, "grad_norm": 1.1590298414230347, "learning_rate": 5.0414000000000006e-06, "loss": 0.0163, "step": 24794 }, { "epoch": 1.3219064882443887, "grad_norm": 1.6469587087631226, "learning_rate": 5.0412e-06, "loss": 0.0162, "step": 24795 }, { "epoch": 1.3219598016740417, "grad_norm": 2.317535400390625, "learning_rate": 5.0410000000000005e-06, "loss": 0.0328, "step": 24796 }, { "epoch": 1.3220131151036947, "grad_norm": 1.5381183624267578, "learning_rate": 5.0408e-06, "loss": 0.0171, "step": 24797 }, { "epoch": 1.3220664285333474, "grad_norm": 1.759007215499878, "learning_rate": 5.0406e-06, "loss": 0.0302, "step": 24798 }, { "epoch": 1.3221197419630004, "grad_norm": 3.4312832355499268, "learning_rate": 5.040400000000001e-06, "loss": 0.0255, "step": 24799 }, { "epoch": 1.3221730553926534, "grad_norm": 2.086777687072754, "learning_rate": 5.0402000000000005e-06, "loss": 0.0345, "step": 24800 }, { "epoch": 1.3222263688223064, "grad_norm": 1.2651243209838867, "learning_rate": 5.04e-06, "loss": 0.0072, "step": 24801 }, { "epoch": 1.3222796822519594, "grad_norm": 1.530086636543274, "learning_rate": 5.0398000000000004e-06, "loss": 0.0309, "step": 24802 }, { "epoch": 1.3223329956816121, "grad_norm": 1.8765897750854492, "learning_rate": 5.0396e-06, "loss": 0.0088, "step": 24803 }, { "epoch": 1.322386309111265, "grad_norm": 1.4699904918670654, "learning_rate": 5.039400000000001e-06, "loss": 0.0247, "step": 24804 }, { "epoch": 1.322439622540918, "grad_norm": 1.8659223318099976, "learning_rate": 5.039200000000001e-06, "loss": 0.0242, "step": 24805 }, { "epoch": 1.322492935970571, "grad_norm": 1.5574204921722412, "learning_rate": 5.039e-06, "loss": 0.0109, "step": 24806 }, { "epoch": 1.322546249400224, "grad_norm": 1.133068561553955, "learning_rate": 5.0388e-06, "loss": 0.0084, "step": 24807 }, { "epoch": 1.3225995628298768, "grad_norm": 1.376624345779419, "learning_rate": 5.0386e-06, "loss": 0.0082, "step": 24808 }, { "epoch": 1.3226528762595298, "grad_norm": 1.5490645170211792, "learning_rate": 5.0384e-06, "loss": 0.0135, "step": 24809 }, { "epoch": 1.3227061896891827, "grad_norm": 3.928542375564575, "learning_rate": 5.038200000000001e-06, "loss": 0.0219, "step": 24810 }, { "epoch": 1.3227595031188355, "grad_norm": 2.098141670227051, "learning_rate": 5.038000000000001e-06, "loss": 0.0368, "step": 24811 }, { "epoch": 1.3228128165484887, "grad_norm": 1.9727166891098022, "learning_rate": 5.0378e-06, "loss": 0.0207, "step": 24812 }, { "epoch": 1.3228661299781415, "grad_norm": 3.1786718368530273, "learning_rate": 5.0376e-06, "loss": 0.0281, "step": 24813 }, { "epoch": 1.3229194434077944, "grad_norm": 2.2195398807525635, "learning_rate": 5.0374e-06, "loss": 0.0315, "step": 24814 }, { "epoch": 1.3229727568374474, "grad_norm": 2.0994205474853516, "learning_rate": 5.0372e-06, "loss": 0.0192, "step": 24815 }, { "epoch": 1.3230260702671002, "grad_norm": 0.3318636119365692, "learning_rate": 5.037000000000001e-06, "loss": 0.0031, "step": 24816 }, { "epoch": 1.3230793836967532, "grad_norm": 2.3673951625823975, "learning_rate": 5.036800000000001e-06, "loss": 0.0147, "step": 24817 }, { "epoch": 1.3231326971264061, "grad_norm": 3.3031234741210938, "learning_rate": 5.0366e-06, "loss": 0.0171, "step": 24818 }, { "epoch": 1.3231860105560591, "grad_norm": 3.8241491317749023, "learning_rate": 5.0364e-06, "loss": 0.0594, "step": 24819 }, { "epoch": 1.323239323985712, "grad_norm": 0.7395442724227905, "learning_rate": 5.0362e-06, "loss": 0.0059, "step": 24820 }, { "epoch": 1.3232926374153648, "grad_norm": 1.2733012437820435, "learning_rate": 5.0360000000000006e-06, "loss": 0.0075, "step": 24821 }, { "epoch": 1.3233459508450178, "grad_norm": 1.2290325164794922, "learning_rate": 5.035800000000001e-06, "loss": 0.007, "step": 24822 }, { "epoch": 1.3233992642746708, "grad_norm": 1.8361140489578247, "learning_rate": 5.0356000000000005e-06, "loss": 0.03, "step": 24823 }, { "epoch": 1.3234525777043238, "grad_norm": 1.6271226406097412, "learning_rate": 5.0354e-06, "loss": 0.0325, "step": 24824 }, { "epoch": 1.3235058911339768, "grad_norm": 1.6587426662445068, "learning_rate": 5.0352e-06, "loss": 0.0127, "step": 24825 }, { "epoch": 1.3235592045636295, "grad_norm": 1.7418060302734375, "learning_rate": 5.035e-06, "loss": 0.0076, "step": 24826 }, { "epoch": 1.3236125179932825, "grad_norm": 0.9653229117393494, "learning_rate": 5.0348000000000005e-06, "loss": 0.004, "step": 24827 }, { "epoch": 1.3236658314229355, "grad_norm": 1.594160556793213, "learning_rate": 5.034600000000001e-06, "loss": 0.0133, "step": 24828 }, { "epoch": 1.3237191448525882, "grad_norm": 1.488061785697937, "learning_rate": 5.0344000000000005e-06, "loss": 0.0208, "step": 24829 }, { "epoch": 1.3237724582822412, "grad_norm": 2.3366355895996094, "learning_rate": 5.0342e-06, "loss": 0.0321, "step": 24830 }, { "epoch": 1.3238257717118942, "grad_norm": 2.2573556900024414, "learning_rate": 5.0339999999999996e-06, "loss": 0.024, "step": 24831 }, { "epoch": 1.3238790851415472, "grad_norm": 1.9720563888549805, "learning_rate": 5.033800000000001e-06, "loss": 0.0206, "step": 24832 }, { "epoch": 1.3239323985712002, "grad_norm": 0.8894490003585815, "learning_rate": 5.0336e-06, "loss": 0.0066, "step": 24833 }, { "epoch": 1.323985712000853, "grad_norm": 1.5141767263412476, "learning_rate": 5.033400000000001e-06, "loss": 0.0184, "step": 24834 }, { "epoch": 1.324039025430506, "grad_norm": 1.7687331438064575, "learning_rate": 5.0332e-06, "loss": 0.0211, "step": 24835 }, { "epoch": 1.3240923388601589, "grad_norm": 0.718789279460907, "learning_rate": 5.033e-06, "loss": 0.0058, "step": 24836 }, { "epoch": 1.3241456522898118, "grad_norm": 0.8994971513748169, "learning_rate": 5.0327999999999995e-06, "loss": 0.0052, "step": 24837 }, { "epoch": 1.3241989657194648, "grad_norm": 1.092165470123291, "learning_rate": 5.032600000000001e-06, "loss": 0.0083, "step": 24838 }, { "epoch": 1.3242522791491176, "grad_norm": 2.244173765182495, "learning_rate": 5.0324e-06, "loss": 0.0165, "step": 24839 }, { "epoch": 1.3243055925787706, "grad_norm": 1.8004047870635986, "learning_rate": 5.032200000000001e-06, "loss": 0.0085, "step": 24840 }, { "epoch": 1.3243589060084235, "grad_norm": 0.6617466807365417, "learning_rate": 5.032e-06, "loss": 0.0057, "step": 24841 }, { "epoch": 1.3244122194380765, "grad_norm": 2.6238338947296143, "learning_rate": 5.0318e-06, "loss": 0.0211, "step": 24842 }, { "epoch": 1.3244655328677295, "grad_norm": 1.6285884380340576, "learning_rate": 5.031600000000001e-06, "loss": 0.0245, "step": 24843 }, { "epoch": 1.3245188462973823, "grad_norm": 1.0241063833236694, "learning_rate": 5.031400000000001e-06, "loss": 0.0068, "step": 24844 }, { "epoch": 1.3245721597270352, "grad_norm": 2.451993227005005, "learning_rate": 5.0312e-06, "loss": 0.016, "step": 24845 }, { "epoch": 1.3246254731566882, "grad_norm": 2.1932454109191895, "learning_rate": 5.031000000000001e-06, "loss": 0.0353, "step": 24846 }, { "epoch": 1.324678786586341, "grad_norm": 3.4985690116882324, "learning_rate": 5.0308e-06, "loss": 0.0532, "step": 24847 }, { "epoch": 1.324732100015994, "grad_norm": 1.1467509269714355, "learning_rate": 5.0306e-06, "loss": 0.0335, "step": 24848 }, { "epoch": 1.324785413445647, "grad_norm": 3.5923709869384766, "learning_rate": 5.030400000000001e-06, "loss": 0.0368, "step": 24849 }, { "epoch": 1.3248387268753, "grad_norm": 0.8036778569221497, "learning_rate": 5.0302000000000006e-06, "loss": 0.0042, "step": 24850 }, { "epoch": 1.324892040304953, "grad_norm": 0.17112304270267487, "learning_rate": 5.03e-06, "loss": 0.0017, "step": 24851 }, { "epoch": 1.3249453537346056, "grad_norm": 4.327471733093262, "learning_rate": 5.0298000000000005e-06, "loss": 0.043, "step": 24852 }, { "epoch": 1.3249986671642586, "grad_norm": 1.3587688207626343, "learning_rate": 5.0296e-06, "loss": 0.0181, "step": 24853 }, { "epoch": 1.3250519805939116, "grad_norm": 1.2252954244613647, "learning_rate": 5.029400000000001e-06, "loss": 0.0361, "step": 24854 }, { "epoch": 1.3251052940235646, "grad_norm": 2.0547964572906494, "learning_rate": 5.029200000000001e-06, "loss": 0.0224, "step": 24855 }, { "epoch": 1.3251586074532176, "grad_norm": 1.8722189664840698, "learning_rate": 5.0290000000000005e-06, "loss": 0.0101, "step": 24856 }, { "epoch": 1.3252119208828703, "grad_norm": 4.676142692565918, "learning_rate": 5.0288e-06, "loss": 0.0396, "step": 24857 }, { "epoch": 1.3252652343125233, "grad_norm": 1.0159411430358887, "learning_rate": 5.0286000000000004e-06, "loss": 0.0154, "step": 24858 }, { "epoch": 1.3253185477421763, "grad_norm": 1.7663023471832275, "learning_rate": 5.0284e-06, "loss": 0.0215, "step": 24859 }, { "epoch": 1.3253718611718293, "grad_norm": 3.0453548431396484, "learning_rate": 5.028200000000001e-06, "loss": 0.0528, "step": 24860 }, { "epoch": 1.3254251746014822, "grad_norm": 3.48110294342041, "learning_rate": 5.028000000000001e-06, "loss": 0.0389, "step": 24861 }, { "epoch": 1.325478488031135, "grad_norm": 1.4636287689208984, "learning_rate": 5.0278e-06, "loss": 0.0252, "step": 24862 }, { "epoch": 1.325531801460788, "grad_norm": 1.5293359756469727, "learning_rate": 5.0276e-06, "loss": 0.0164, "step": 24863 }, { "epoch": 1.325585114890441, "grad_norm": 1.9477360248565674, "learning_rate": 5.0274e-06, "loss": 0.0239, "step": 24864 }, { "epoch": 1.3256384283200937, "grad_norm": 3.100924491882324, "learning_rate": 5.0272e-06, "loss": 0.0489, "step": 24865 }, { "epoch": 1.3256917417497467, "grad_norm": 2.204998731613159, "learning_rate": 5.027000000000001e-06, "loss": 0.0355, "step": 24866 }, { "epoch": 1.3257450551793997, "grad_norm": 1.7852606773376465, "learning_rate": 5.026800000000001e-06, "loss": 0.0269, "step": 24867 }, { "epoch": 1.3257983686090526, "grad_norm": 1.364832878112793, "learning_rate": 5.0266e-06, "loss": 0.0147, "step": 24868 }, { "epoch": 1.3258516820387056, "grad_norm": 0.32174554467201233, "learning_rate": 5.0264e-06, "loss": 0.0027, "step": 24869 }, { "epoch": 1.3259049954683584, "grad_norm": 1.7889173030853271, "learning_rate": 5.0262e-06, "loss": 0.0143, "step": 24870 }, { "epoch": 1.3259583088980114, "grad_norm": 0.5397776961326599, "learning_rate": 5.026000000000001e-06, "loss": 0.0046, "step": 24871 }, { "epoch": 1.3260116223276643, "grad_norm": 1.207381248474121, "learning_rate": 5.025800000000001e-06, "loss": 0.0097, "step": 24872 }, { "epoch": 1.3260649357573173, "grad_norm": 1.7515572309494019, "learning_rate": 5.025600000000001e-06, "loss": 0.028, "step": 24873 }, { "epoch": 1.3261182491869703, "grad_norm": 0.3654806911945343, "learning_rate": 5.0254e-06, "loss": 0.004, "step": 24874 }, { "epoch": 1.326171562616623, "grad_norm": 1.4052163362503052, "learning_rate": 5.0252e-06, "loss": 0.0329, "step": 24875 }, { "epoch": 1.326224876046276, "grad_norm": 4.043872356414795, "learning_rate": 5.025e-06, "loss": 0.0307, "step": 24876 }, { "epoch": 1.326278189475929, "grad_norm": 1.3033236265182495, "learning_rate": 5.024800000000001e-06, "loss": 0.0103, "step": 24877 }, { "epoch": 1.326331502905582, "grad_norm": 0.682937741279602, "learning_rate": 5.024600000000001e-06, "loss": 0.0043, "step": 24878 }, { "epoch": 1.326384816335235, "grad_norm": 2.7448883056640625, "learning_rate": 5.0244000000000005e-06, "loss": 0.0417, "step": 24879 }, { "epoch": 1.3264381297648877, "grad_norm": 2.725966691970825, "learning_rate": 5.0242e-06, "loss": 0.0135, "step": 24880 }, { "epoch": 1.3264914431945407, "grad_norm": 2.5097479820251465, "learning_rate": 5.024e-06, "loss": 0.0188, "step": 24881 }, { "epoch": 1.3265447566241937, "grad_norm": 2.787794589996338, "learning_rate": 5.023800000000001e-06, "loss": 0.023, "step": 24882 }, { "epoch": 1.3265980700538464, "grad_norm": 1.4676809310913086, "learning_rate": 5.0236000000000005e-06, "loss": 0.0137, "step": 24883 }, { "epoch": 1.3266513834834994, "grad_norm": 0.7031314373016357, "learning_rate": 5.023400000000001e-06, "loss": 0.0044, "step": 24884 }, { "epoch": 1.3267046969131524, "grad_norm": 1.26425302028656, "learning_rate": 5.0232000000000005e-06, "loss": 0.0401, "step": 24885 }, { "epoch": 1.3267580103428054, "grad_norm": 1.6290767192840576, "learning_rate": 5.023e-06, "loss": 0.0338, "step": 24886 }, { "epoch": 1.3268113237724584, "grad_norm": 1.2846013307571411, "learning_rate": 5.0228e-06, "loss": 0.0121, "step": 24887 }, { "epoch": 1.3268646372021111, "grad_norm": 2.3501405715942383, "learning_rate": 5.022600000000001e-06, "loss": 0.0533, "step": 24888 }, { "epoch": 1.326917950631764, "grad_norm": 1.611162781715393, "learning_rate": 5.0224e-06, "loss": 0.0222, "step": 24889 }, { "epoch": 1.326971264061417, "grad_norm": 0.5400140881538391, "learning_rate": 5.022200000000001e-06, "loss": 0.0052, "step": 24890 }, { "epoch": 1.32702457749107, "grad_norm": 1.4426473379135132, "learning_rate": 5.022e-06, "loss": 0.023, "step": 24891 }, { "epoch": 1.327077890920723, "grad_norm": 2.2250776290893555, "learning_rate": 5.0218e-06, "loss": 0.0445, "step": 24892 }, { "epoch": 1.3271312043503758, "grad_norm": 2.4238345623016357, "learning_rate": 5.021600000000001e-06, "loss": 0.0334, "step": 24893 }, { "epoch": 1.3271845177800288, "grad_norm": 1.80473792552948, "learning_rate": 5.021400000000001e-06, "loss": 0.0637, "step": 24894 }, { "epoch": 1.3272378312096818, "grad_norm": 2.2090179920196533, "learning_rate": 5.0212e-06, "loss": 0.0123, "step": 24895 }, { "epoch": 1.3272911446393347, "grad_norm": 1.903082013130188, "learning_rate": 5.021000000000001e-06, "loss": 0.0256, "step": 24896 }, { "epoch": 1.3273444580689877, "grad_norm": 2.302290678024292, "learning_rate": 5.0208e-06, "loss": 0.0179, "step": 24897 }, { "epoch": 1.3273977714986405, "grad_norm": 2.481668472290039, "learning_rate": 5.0206e-06, "loss": 0.0284, "step": 24898 }, { "epoch": 1.3274510849282934, "grad_norm": 1.645792007446289, "learning_rate": 5.020400000000001e-06, "loss": 0.0217, "step": 24899 }, { "epoch": 1.3275043983579464, "grad_norm": 1.3524072170257568, "learning_rate": 5.020200000000001e-06, "loss": 0.012, "step": 24900 }, { "epoch": 1.3275577117875992, "grad_norm": 2.105865001678467, "learning_rate": 5.02e-06, "loss": 0.0154, "step": 24901 }, { "epoch": 1.3276110252172522, "grad_norm": 1.3940247297286987, "learning_rate": 5.019800000000001e-06, "loss": 0.0155, "step": 24902 }, { "epoch": 1.3276643386469051, "grad_norm": 3.560213327407837, "learning_rate": 5.0196e-06, "loss": 0.0212, "step": 24903 }, { "epoch": 1.3277176520765581, "grad_norm": 1.8968788385391235, "learning_rate": 5.019400000000001e-06, "loss": 0.0262, "step": 24904 }, { "epoch": 1.327770965506211, "grad_norm": 1.1757408380508423, "learning_rate": 5.019200000000001e-06, "loss": 0.0087, "step": 24905 }, { "epoch": 1.3278242789358639, "grad_norm": 2.4696812629699707, "learning_rate": 5.0190000000000006e-06, "loss": 0.0251, "step": 24906 }, { "epoch": 1.3278775923655168, "grad_norm": 1.2620961666107178, "learning_rate": 5.0188e-06, "loss": 0.0269, "step": 24907 }, { "epoch": 1.3279309057951698, "grad_norm": 1.333315134048462, "learning_rate": 5.0186000000000005e-06, "loss": 0.0148, "step": 24908 }, { "epoch": 1.3279842192248228, "grad_norm": 0.9014312028884888, "learning_rate": 5.0184e-06, "loss": 0.0202, "step": 24909 }, { "epoch": 1.3280375326544758, "grad_norm": 1.6445122957229614, "learning_rate": 5.0182000000000005e-06, "loss": 0.0283, "step": 24910 }, { "epoch": 1.3280908460841285, "grad_norm": 1.529833436012268, "learning_rate": 5.018000000000001e-06, "loss": 0.0122, "step": 24911 }, { "epoch": 1.3281441595137815, "grad_norm": 1.6877163648605347, "learning_rate": 5.0178000000000005e-06, "loss": 0.0159, "step": 24912 }, { "epoch": 1.3281974729434345, "grad_norm": 2.2776687145233154, "learning_rate": 5.0176e-06, "loss": 0.0271, "step": 24913 }, { "epoch": 1.3282507863730872, "grad_norm": 1.1983988285064697, "learning_rate": 5.0174000000000004e-06, "loss": 0.0238, "step": 24914 }, { "epoch": 1.3283040998027404, "grad_norm": 1.6310951709747314, "learning_rate": 5.017200000000001e-06, "loss": 0.0296, "step": 24915 }, { "epoch": 1.3283574132323932, "grad_norm": 2.465388298034668, "learning_rate": 5.017e-06, "loss": 0.0472, "step": 24916 }, { "epoch": 1.3284107266620462, "grad_norm": 1.8642303943634033, "learning_rate": 5.016800000000001e-06, "loss": 0.0288, "step": 24917 }, { "epoch": 1.3284640400916992, "grad_norm": 1.2447527647018433, "learning_rate": 5.0166e-06, "loss": 0.0085, "step": 24918 }, { "epoch": 1.328517353521352, "grad_norm": 1.5492783784866333, "learning_rate": 5.0164e-06, "loss": 0.0187, "step": 24919 }, { "epoch": 1.328570666951005, "grad_norm": 1.033671498298645, "learning_rate": 5.0162e-06, "loss": 0.0156, "step": 24920 }, { "epoch": 1.3286239803806579, "grad_norm": 0.5866990089416504, "learning_rate": 5.016000000000001e-06, "loss": 0.005, "step": 24921 }, { "epoch": 1.3286772938103109, "grad_norm": 1.4757916927337646, "learning_rate": 5.0158e-06, "loss": 0.0434, "step": 24922 }, { "epoch": 1.3287306072399638, "grad_norm": 3.0347161293029785, "learning_rate": 5.015600000000001e-06, "loss": 0.0442, "step": 24923 }, { "epoch": 1.3287839206696166, "grad_norm": 1.1178127527236938, "learning_rate": 5.0154e-06, "loss": 0.0075, "step": 24924 }, { "epoch": 1.3288372340992696, "grad_norm": 1.6912020444869995, "learning_rate": 5.0152e-06, "loss": 0.0435, "step": 24925 }, { "epoch": 1.3288905475289225, "grad_norm": 3.286210060119629, "learning_rate": 5.015e-06, "loss": 0.0123, "step": 24926 }, { "epoch": 1.3289438609585755, "grad_norm": 1.1881773471832275, "learning_rate": 5.014800000000001e-06, "loss": 0.0128, "step": 24927 }, { "epoch": 1.3289971743882285, "grad_norm": 0.683265745639801, "learning_rate": 5.0146e-06, "loss": 0.0079, "step": 24928 }, { "epoch": 1.3290504878178813, "grad_norm": 1.8609305620193481, "learning_rate": 5.014400000000001e-06, "loss": 0.0238, "step": 24929 }, { "epoch": 1.3291038012475342, "grad_norm": 2.105912685394287, "learning_rate": 5.0142e-06, "loss": 0.0192, "step": 24930 }, { "epoch": 1.3291571146771872, "grad_norm": 1.7518914937973022, "learning_rate": 5.014e-06, "loss": 0.0243, "step": 24931 }, { "epoch": 1.32921042810684, "grad_norm": 0.7801395058631897, "learning_rate": 5.013800000000001e-06, "loss": 0.0074, "step": 24932 }, { "epoch": 1.329263741536493, "grad_norm": 1.365423560142517, "learning_rate": 5.013600000000001e-06, "loss": 0.0153, "step": 24933 }, { "epoch": 1.329317054966146, "grad_norm": 1.5364738702774048, "learning_rate": 5.0134e-06, "loss": 0.0464, "step": 24934 }, { "epoch": 1.329370368395799, "grad_norm": 1.4761489629745483, "learning_rate": 5.0132000000000006e-06, "loss": 0.0135, "step": 24935 }, { "epoch": 1.329423681825452, "grad_norm": 1.5910414457321167, "learning_rate": 5.013e-06, "loss": 0.0448, "step": 24936 }, { "epoch": 1.3294769952551047, "grad_norm": 3.999505043029785, "learning_rate": 5.0128e-06, "loss": 0.0236, "step": 24937 }, { "epoch": 1.3295303086847576, "grad_norm": 2.475670099258423, "learning_rate": 5.012600000000001e-06, "loss": 0.0219, "step": 24938 }, { "epoch": 1.3295836221144106, "grad_norm": 3.921539306640625, "learning_rate": 5.0124000000000005e-06, "loss": 0.0251, "step": 24939 }, { "epoch": 1.3296369355440636, "grad_norm": 2.5413591861724854, "learning_rate": 5.0122e-06, "loss": 0.0428, "step": 24940 }, { "epoch": 1.3296902489737166, "grad_norm": 1.6209404468536377, "learning_rate": 5.0120000000000005e-06, "loss": 0.0105, "step": 24941 }, { "epoch": 1.3297435624033693, "grad_norm": 2.152743101119995, "learning_rate": 5.0118e-06, "loss": 0.0289, "step": 24942 }, { "epoch": 1.3297968758330223, "grad_norm": 1.1421666145324707, "learning_rate": 5.011600000000001e-06, "loss": 0.0166, "step": 24943 }, { "epoch": 1.3298501892626753, "grad_norm": 2.0579166412353516, "learning_rate": 5.011400000000001e-06, "loss": 0.0208, "step": 24944 }, { "epoch": 1.3299035026923283, "grad_norm": 1.9677739143371582, "learning_rate": 5.0112e-06, "loss": 0.011, "step": 24945 }, { "epoch": 1.3299568161219812, "grad_norm": 1.4169158935546875, "learning_rate": 5.011e-06, "loss": 0.0227, "step": 24946 }, { "epoch": 1.330010129551634, "grad_norm": 0.959520161151886, "learning_rate": 5.0108e-06, "loss": 0.0084, "step": 24947 }, { "epoch": 1.330063442981287, "grad_norm": 1.027294635772705, "learning_rate": 5.0106e-06, "loss": 0.0075, "step": 24948 }, { "epoch": 1.33011675641094, "grad_norm": 1.1337896585464478, "learning_rate": 5.010400000000001e-06, "loss": 0.015, "step": 24949 }, { "epoch": 1.3301700698405927, "grad_norm": 0.913223922252655, "learning_rate": 5.010200000000001e-06, "loss": 0.0065, "step": 24950 }, { "epoch": 1.3302233832702457, "grad_norm": 1.2630850076675415, "learning_rate": 5.01e-06, "loss": 0.0259, "step": 24951 }, { "epoch": 1.3302766966998987, "grad_norm": 1.8792099952697754, "learning_rate": 5.0098e-06, "loss": 0.0207, "step": 24952 }, { "epoch": 1.3303300101295517, "grad_norm": 2.3353188037872314, "learning_rate": 5.0096e-06, "loss": 0.0192, "step": 24953 }, { "epoch": 1.3303833235592046, "grad_norm": 2.74678111076355, "learning_rate": 5.009400000000001e-06, "loss": 0.0578, "step": 24954 }, { "epoch": 1.3304366369888574, "grad_norm": 2.1540768146514893, "learning_rate": 5.009200000000001e-06, "loss": 0.0145, "step": 24955 }, { "epoch": 1.3304899504185104, "grad_norm": 2.6530845165252686, "learning_rate": 5.009000000000001e-06, "loss": 0.045, "step": 24956 }, { "epoch": 1.3305432638481633, "grad_norm": 0.9522817134857178, "learning_rate": 5.0088e-06, "loss": 0.0194, "step": 24957 }, { "epoch": 1.3305965772778163, "grad_norm": 0.6409856677055359, "learning_rate": 5.0086e-06, "loss": 0.0054, "step": 24958 }, { "epoch": 1.3306498907074693, "grad_norm": 2.284630298614502, "learning_rate": 5.0084e-06, "loss": 0.0356, "step": 24959 }, { "epoch": 1.330703204137122, "grad_norm": 1.7492669820785522, "learning_rate": 5.008200000000001e-06, "loss": 0.0367, "step": 24960 }, { "epoch": 1.330756517566775, "grad_norm": 1.8124759197235107, "learning_rate": 5.008000000000001e-06, "loss": 0.0294, "step": 24961 }, { "epoch": 1.330809830996428, "grad_norm": 2.3316726684570312, "learning_rate": 5.0078000000000006e-06, "loss": 0.0391, "step": 24962 }, { "epoch": 1.330863144426081, "grad_norm": 1.5439401865005493, "learning_rate": 5.0076e-06, "loss": 0.0273, "step": 24963 }, { "epoch": 1.330916457855734, "grad_norm": 1.4788539409637451, "learning_rate": 5.0074e-06, "loss": 0.0151, "step": 24964 }, { "epoch": 1.3309697712853867, "grad_norm": 1.4758177995681763, "learning_rate": 5.007200000000001e-06, "loss": 0.0184, "step": 24965 }, { "epoch": 1.3310230847150397, "grad_norm": 1.3294731378555298, "learning_rate": 5.0070000000000005e-06, "loss": 0.0125, "step": 24966 }, { "epoch": 1.3310763981446927, "grad_norm": 1.9705810546875, "learning_rate": 5.006800000000001e-06, "loss": 0.0196, "step": 24967 }, { "epoch": 1.3311297115743455, "grad_norm": 1.5137784481048584, "learning_rate": 5.0066000000000005e-06, "loss": 0.0218, "step": 24968 }, { "epoch": 1.3311830250039984, "grad_norm": 3.2155942916870117, "learning_rate": 5.0064e-06, "loss": 0.0345, "step": 24969 }, { "epoch": 1.3312363384336514, "grad_norm": 2.6404802799224854, "learning_rate": 5.0062e-06, "loss": 0.0443, "step": 24970 }, { "epoch": 1.3312896518633044, "grad_norm": 1.6797138452529907, "learning_rate": 5.006000000000001e-06, "loss": 0.0263, "step": 24971 }, { "epoch": 1.3313429652929574, "grad_norm": 1.641701579093933, "learning_rate": 5.0058e-06, "loss": 0.0128, "step": 24972 }, { "epoch": 1.3313962787226101, "grad_norm": 0.9048247933387756, "learning_rate": 5.005600000000001e-06, "loss": 0.0064, "step": 24973 }, { "epoch": 1.331449592152263, "grad_norm": 1.259211540222168, "learning_rate": 5.0054e-06, "loss": 0.0204, "step": 24974 }, { "epoch": 1.331502905581916, "grad_norm": 1.7808400392532349, "learning_rate": 5.0052e-06, "loss": 0.016, "step": 24975 }, { "epoch": 1.331556219011569, "grad_norm": 2.966270923614502, "learning_rate": 5.0049999999999995e-06, "loss": 0.0258, "step": 24976 }, { "epoch": 1.331609532441222, "grad_norm": 2.0238707065582275, "learning_rate": 5.004800000000001e-06, "loss": 0.0429, "step": 24977 }, { "epoch": 1.3316628458708748, "grad_norm": 2.850579261779785, "learning_rate": 5.0046e-06, "loss": 0.0334, "step": 24978 }, { "epoch": 1.3317161593005278, "grad_norm": 2.1627955436706543, "learning_rate": 5.004400000000001e-06, "loss": 0.0239, "step": 24979 }, { "epoch": 1.3317694727301808, "grad_norm": 1.816210150718689, "learning_rate": 5.0042e-06, "loss": 0.0213, "step": 24980 }, { "epoch": 1.3318227861598337, "grad_norm": 1.3485280275344849, "learning_rate": 5.004e-06, "loss": 0.0174, "step": 24981 }, { "epoch": 1.3318760995894867, "grad_norm": 0.7441136240959167, "learning_rate": 5.003800000000001e-06, "loss": 0.0093, "step": 24982 }, { "epoch": 1.3319294130191395, "grad_norm": 0.6393241286277771, "learning_rate": 5.003600000000001e-06, "loss": 0.0053, "step": 24983 }, { "epoch": 1.3319827264487925, "grad_norm": 1.8882217407226562, "learning_rate": 5.0034e-06, "loss": 0.032, "step": 24984 }, { "epoch": 1.3320360398784454, "grad_norm": 1.9879711866378784, "learning_rate": 5.003200000000001e-06, "loss": 0.0097, "step": 24985 }, { "epoch": 1.3320893533080982, "grad_norm": 1.482535719871521, "learning_rate": 5.003e-06, "loss": 0.0131, "step": 24986 }, { "epoch": 1.3321426667377512, "grad_norm": 1.629746437072754, "learning_rate": 5.0028e-06, "loss": 0.0317, "step": 24987 }, { "epoch": 1.3321959801674041, "grad_norm": 1.128676176071167, "learning_rate": 5.002600000000001e-06, "loss": 0.006, "step": 24988 }, { "epoch": 1.3322492935970571, "grad_norm": 2.143090009689331, "learning_rate": 5.002400000000001e-06, "loss": 0.0243, "step": 24989 }, { "epoch": 1.33230260702671, "grad_norm": 0.7213862538337708, "learning_rate": 5.0022e-06, "loss": 0.0109, "step": 24990 }, { "epoch": 1.3323559204563629, "grad_norm": 3.698606014251709, "learning_rate": 5.0020000000000006e-06, "loss": 0.0455, "step": 24991 }, { "epoch": 1.3324092338860158, "grad_norm": 1.659428596496582, "learning_rate": 5.0018e-06, "loss": 0.0168, "step": 24992 }, { "epoch": 1.3324625473156688, "grad_norm": 1.4184561967849731, "learning_rate": 5.0016000000000005e-06, "loss": 0.0173, "step": 24993 }, { "epoch": 1.3325158607453218, "grad_norm": 1.2534838914871216, "learning_rate": 5.001400000000001e-06, "loss": 0.0148, "step": 24994 }, { "epoch": 1.3325691741749748, "grad_norm": 2.402864933013916, "learning_rate": 5.0012000000000005e-06, "loss": 0.0445, "step": 24995 }, { "epoch": 1.3326224876046275, "grad_norm": 4.181082248687744, "learning_rate": 5.001e-06, "loss": 0.0226, "step": 24996 }, { "epoch": 1.3326758010342805, "grad_norm": 0.8636621832847595, "learning_rate": 5.0008000000000005e-06, "loss": 0.0071, "step": 24997 }, { "epoch": 1.3327291144639335, "grad_norm": 1.794788122177124, "learning_rate": 5.0006e-06, "loss": 0.0151, "step": 24998 }, { "epoch": 1.3327824278935865, "grad_norm": 2.140428066253662, "learning_rate": 5.0004000000000004e-06, "loss": 0.0295, "step": 24999 }, { "epoch": 1.3328357413232395, "grad_norm": 1.114745855331421, "learning_rate": 5.000200000000001e-06, "loss": 0.0236, "step": 25000 }, { "epoch": 1.3328890547528922, "grad_norm": 3.5781474113464355, "learning_rate": 5e-06, "loss": 0.02, "step": 25001 }, { "epoch": 1.3329423681825452, "grad_norm": 1.541912317276001, "learning_rate": 4.9998e-06, "loss": 0.0123, "step": 25002 }, { "epoch": 1.3329956816121982, "grad_norm": 0.8264707326889038, "learning_rate": 4.9996e-06, "loss": 0.007, "step": 25003 }, { "epoch": 1.333048995041851, "grad_norm": 3.163236141204834, "learning_rate": 4.999400000000001e-06, "loss": 0.0431, "step": 25004 }, { "epoch": 1.333102308471504, "grad_norm": 2.582906723022461, "learning_rate": 4.9992e-06, "loss": 0.0388, "step": 25005 }, { "epoch": 1.3331556219011569, "grad_norm": 2.6560258865356445, "learning_rate": 4.999000000000001e-06, "loss": 0.0284, "step": 25006 }, { "epoch": 1.3332089353308099, "grad_norm": 2.506150245666504, "learning_rate": 4.9988e-06, "loss": 0.0449, "step": 25007 }, { "epoch": 1.3332622487604628, "grad_norm": 0.9185904860496521, "learning_rate": 4.998600000000001e-06, "loss": 0.0163, "step": 25008 }, { "epoch": 1.3333155621901156, "grad_norm": 1.508496642112732, "learning_rate": 4.9984e-06, "loss": 0.017, "step": 25009 }, { "epoch": 1.3333688756197686, "grad_norm": 0.6862544417381287, "learning_rate": 4.998200000000001e-06, "loss": 0.016, "step": 25010 }, { "epoch": 1.3334221890494216, "grad_norm": 1.5597567558288574, "learning_rate": 4.998e-06, "loss": 0.0273, "step": 25011 }, { "epoch": 1.3334755024790745, "grad_norm": 2.06728196144104, "learning_rate": 4.997800000000001e-06, "loss": 0.021, "step": 25012 }, { "epoch": 1.3335288159087275, "grad_norm": 2.866743564605713, "learning_rate": 4.9976e-06, "loss": 0.017, "step": 25013 }, { "epoch": 1.3335821293383803, "grad_norm": 1.5524358749389648, "learning_rate": 4.997400000000001e-06, "loss": 0.037, "step": 25014 }, { "epoch": 1.3336354427680333, "grad_norm": 1.0547213554382324, "learning_rate": 4.9972e-06, "loss": 0.0408, "step": 25015 }, { "epoch": 1.3336887561976862, "grad_norm": 1.3102189302444458, "learning_rate": 4.997000000000001e-06, "loss": 0.0278, "step": 25016 }, { "epoch": 1.333742069627339, "grad_norm": 1.4029239416122437, "learning_rate": 4.9968e-06, "loss": 0.0111, "step": 25017 }, { "epoch": 1.3337953830569922, "grad_norm": 3.5245113372802734, "learning_rate": 4.996600000000001e-06, "loss": 0.0294, "step": 25018 }, { "epoch": 1.333848696486645, "grad_norm": 1.7072702646255493, "learning_rate": 4.9964e-06, "loss": 0.025, "step": 25019 }, { "epoch": 1.333902009916298, "grad_norm": 2.9470434188842773, "learning_rate": 4.9962000000000005e-06, "loss": 0.0226, "step": 25020 }, { "epoch": 1.333955323345951, "grad_norm": 1.3163944482803345, "learning_rate": 4.996e-06, "loss": 0.0186, "step": 25021 }, { "epoch": 1.3340086367756037, "grad_norm": 2.474520444869995, "learning_rate": 4.9958000000000005e-06, "loss": 0.0206, "step": 25022 }, { "epoch": 1.3340619502052566, "grad_norm": 2.7072110176086426, "learning_rate": 4.9956e-06, "loss": 0.0503, "step": 25023 }, { "epoch": 1.3341152636349096, "grad_norm": 2.7059473991394043, "learning_rate": 4.9954000000000005e-06, "loss": 0.0433, "step": 25024 }, { "epoch": 1.3341685770645626, "grad_norm": 1.4098007678985596, "learning_rate": 4.995200000000001e-06, "loss": 0.0275, "step": 25025 }, { "epoch": 1.3342218904942156, "grad_norm": 1.335326910018921, "learning_rate": 4.9950000000000005e-06, "loss": 0.0148, "step": 25026 }, { "epoch": 1.3342752039238683, "grad_norm": 1.1662403345108032, "learning_rate": 4.9948e-06, "loss": 0.0266, "step": 25027 }, { "epoch": 1.3343285173535213, "grad_norm": 2.239518404006958, "learning_rate": 4.9946000000000004e-06, "loss": 0.0336, "step": 25028 }, { "epoch": 1.3343818307831743, "grad_norm": 2.121248960494995, "learning_rate": 4.9944e-06, "loss": 0.0365, "step": 25029 }, { "epoch": 1.3344351442128273, "grad_norm": 1.1666713953018188, "learning_rate": 4.9942e-06, "loss": 0.0115, "step": 25030 }, { "epoch": 1.3344884576424803, "grad_norm": 2.4284141063690186, "learning_rate": 4.994000000000001e-06, "loss": 0.0244, "step": 25031 }, { "epoch": 1.334541771072133, "grad_norm": 1.5923428535461426, "learning_rate": 4.9938e-06, "loss": 0.0127, "step": 25032 }, { "epoch": 1.334595084501786, "grad_norm": 1.2604074478149414, "learning_rate": 4.993600000000001e-06, "loss": 0.0095, "step": 25033 }, { "epoch": 1.334648397931439, "grad_norm": 1.169291377067566, "learning_rate": 4.9934e-06, "loss": 0.0198, "step": 25034 }, { "epoch": 1.3347017113610917, "grad_norm": 1.6319724321365356, "learning_rate": 4.9932e-06, "loss": 0.0112, "step": 25035 }, { "epoch": 1.334755024790745, "grad_norm": 2.134082794189453, "learning_rate": 4.993e-06, "loss": 0.0203, "step": 25036 }, { "epoch": 1.3348083382203977, "grad_norm": 1.452642798423767, "learning_rate": 4.992800000000001e-06, "loss": 0.0132, "step": 25037 }, { "epoch": 1.3348616516500507, "grad_norm": 3.4464170932769775, "learning_rate": 4.9926e-06, "loss": 0.0241, "step": 25038 }, { "epoch": 1.3349149650797036, "grad_norm": 0.4306653141975403, "learning_rate": 4.992400000000001e-06, "loss": 0.0042, "step": 25039 }, { "epoch": 1.3349682785093564, "grad_norm": 2.0276052951812744, "learning_rate": 4.9922e-06, "loss": 0.0244, "step": 25040 }, { "epoch": 1.3350215919390094, "grad_norm": 2.201061487197876, "learning_rate": 4.992e-06, "loss": 0.0366, "step": 25041 }, { "epoch": 1.3350749053686624, "grad_norm": 1.6791423559188843, "learning_rate": 4.9918e-06, "loss": 0.0268, "step": 25042 }, { "epoch": 1.3351282187983153, "grad_norm": 1.4314912557601929, "learning_rate": 4.991600000000001e-06, "loss": 0.0194, "step": 25043 }, { "epoch": 1.3351815322279683, "grad_norm": 2.128842830657959, "learning_rate": 4.9914e-06, "loss": 0.065, "step": 25044 }, { "epoch": 1.335234845657621, "grad_norm": 1.49504816532135, "learning_rate": 4.991200000000001e-06, "loss": 0.021, "step": 25045 }, { "epoch": 1.335288159087274, "grad_norm": 0.7942650318145752, "learning_rate": 4.991e-06, "loss": 0.0067, "step": 25046 }, { "epoch": 1.335341472516927, "grad_norm": 2.538477897644043, "learning_rate": 4.9908000000000006e-06, "loss": 0.0218, "step": 25047 }, { "epoch": 1.33539478594658, "grad_norm": 1.7089711427688599, "learning_rate": 4.9906e-06, "loss": 0.0376, "step": 25048 }, { "epoch": 1.335448099376233, "grad_norm": 2.5371460914611816, "learning_rate": 4.9904000000000005e-06, "loss": 0.0373, "step": 25049 }, { "epoch": 1.3355014128058857, "grad_norm": 1.089107871055603, "learning_rate": 4.990200000000001e-06, "loss": 0.0112, "step": 25050 }, { "epoch": 1.3355547262355387, "grad_norm": 3.53092098236084, "learning_rate": 4.9900000000000005e-06, "loss": 0.0362, "step": 25051 }, { "epoch": 1.3356080396651917, "grad_norm": 1.6774781942367554, "learning_rate": 4.9898e-06, "loss": 0.031, "step": 25052 }, { "epoch": 1.3356613530948445, "grad_norm": 0.9894500374794006, "learning_rate": 4.9896000000000005e-06, "loss": 0.0222, "step": 25053 }, { "epoch": 1.3357146665244974, "grad_norm": 2.141542434692383, "learning_rate": 4.9894e-06, "loss": 0.0132, "step": 25054 }, { "epoch": 1.3357679799541504, "grad_norm": 2.523005485534668, "learning_rate": 4.9892000000000004e-06, "loss": 0.0198, "step": 25055 }, { "epoch": 1.3358212933838034, "grad_norm": 1.539635181427002, "learning_rate": 4.989000000000001e-06, "loss": 0.0241, "step": 25056 }, { "epoch": 1.3358746068134564, "grad_norm": 1.2371418476104736, "learning_rate": 4.9888e-06, "loss": 0.0244, "step": 25057 }, { "epoch": 1.3359279202431091, "grad_norm": 0.5876587629318237, "learning_rate": 4.988600000000001e-06, "loss": 0.0057, "step": 25058 }, { "epoch": 1.3359812336727621, "grad_norm": 0.8159105777740479, "learning_rate": 4.9884e-06, "loss": 0.0082, "step": 25059 }, { "epoch": 1.336034547102415, "grad_norm": 1.2991567850112915, "learning_rate": 4.9882e-06, "loss": 0.014, "step": 25060 }, { "epoch": 1.336087860532068, "grad_norm": 1.927664041519165, "learning_rate": 4.988e-06, "loss": 0.0222, "step": 25061 }, { "epoch": 1.336141173961721, "grad_norm": 0.8723097443580627, "learning_rate": 4.987800000000001e-06, "loss": 0.0053, "step": 25062 }, { "epoch": 1.3361944873913738, "grad_norm": 2.5363001823425293, "learning_rate": 4.9876e-06, "loss": 0.0825, "step": 25063 }, { "epoch": 1.3362478008210268, "grad_norm": 2.3793632984161377, "learning_rate": 4.987400000000001e-06, "loss": 0.0511, "step": 25064 }, { "epoch": 1.3363011142506798, "grad_norm": 2.1512486934661865, "learning_rate": 4.9872e-06, "loss": 0.0156, "step": 25065 }, { "epoch": 1.3363544276803327, "grad_norm": 1.9932160377502441, "learning_rate": 4.987e-06, "loss": 0.0207, "step": 25066 }, { "epoch": 1.3364077411099857, "grad_norm": 3.270444869995117, "learning_rate": 4.9868e-06, "loss": 0.0627, "step": 25067 }, { "epoch": 1.3364610545396385, "grad_norm": 1.049134373664856, "learning_rate": 4.986600000000001e-06, "loss": 0.0131, "step": 25068 }, { "epoch": 1.3365143679692915, "grad_norm": 2.6988167762756348, "learning_rate": 4.986400000000001e-06, "loss": 0.0525, "step": 25069 }, { "epoch": 1.3365676813989444, "grad_norm": 1.7372422218322754, "learning_rate": 4.986200000000001e-06, "loss": 0.0361, "step": 25070 }, { "epoch": 1.3366209948285972, "grad_norm": 1.3895187377929688, "learning_rate": 4.986e-06, "loss": 0.0226, "step": 25071 }, { "epoch": 1.3366743082582502, "grad_norm": 2.717665910720825, "learning_rate": 4.985800000000001e-06, "loss": 0.0415, "step": 25072 }, { "epoch": 1.3367276216879032, "grad_norm": 4.178706169128418, "learning_rate": 4.9856e-06, "loss": 0.0271, "step": 25073 }, { "epoch": 1.3367809351175561, "grad_norm": 1.774014949798584, "learning_rate": 4.985400000000001e-06, "loss": 0.0256, "step": 25074 }, { "epoch": 1.3368342485472091, "grad_norm": 2.4591588973999023, "learning_rate": 4.985200000000001e-06, "loss": 0.0249, "step": 25075 }, { "epoch": 1.3368875619768619, "grad_norm": 1.4261912107467651, "learning_rate": 4.9850000000000006e-06, "loss": 0.0146, "step": 25076 }, { "epoch": 1.3369408754065149, "grad_norm": 1.728683590888977, "learning_rate": 4.9848e-06, "loss": 0.0277, "step": 25077 }, { "epoch": 1.3369941888361678, "grad_norm": 1.763109803199768, "learning_rate": 4.9846000000000005e-06, "loss": 0.0113, "step": 25078 }, { "epoch": 1.3370475022658208, "grad_norm": 1.4607422351837158, "learning_rate": 4.9844e-06, "loss": 0.0148, "step": 25079 }, { "epoch": 1.3371008156954738, "grad_norm": 0.977374792098999, "learning_rate": 4.9842000000000005e-06, "loss": 0.02, "step": 25080 }, { "epoch": 1.3371541291251265, "grad_norm": 2.566859006881714, "learning_rate": 4.984000000000001e-06, "loss": 0.0384, "step": 25081 }, { "epoch": 1.3372074425547795, "grad_norm": 1.554236650466919, "learning_rate": 4.9838000000000005e-06, "loss": 0.0129, "step": 25082 }, { "epoch": 1.3372607559844325, "grad_norm": 1.6826049089431763, "learning_rate": 4.983600000000001e-06, "loss": 0.0204, "step": 25083 }, { "epoch": 1.3373140694140855, "grad_norm": 0.8901071548461914, "learning_rate": 4.9834000000000004e-06, "loss": 0.014, "step": 25084 }, { "epoch": 1.3373673828437385, "grad_norm": 1.7708549499511719, "learning_rate": 4.9832e-06, "loss": 0.0151, "step": 25085 }, { "epoch": 1.3374206962733912, "grad_norm": 2.694884777069092, "learning_rate": 4.983e-06, "loss": 0.0376, "step": 25086 }, { "epoch": 1.3374740097030442, "grad_norm": 2.1874847412109375, "learning_rate": 4.982800000000001e-06, "loss": 0.0306, "step": 25087 }, { "epoch": 1.3375273231326972, "grad_norm": 2.810497999191284, "learning_rate": 4.9826e-06, "loss": 0.0498, "step": 25088 }, { "epoch": 1.33758063656235, "grad_norm": 2.001326322555542, "learning_rate": 4.982400000000001e-06, "loss": 0.0361, "step": 25089 }, { "epoch": 1.337633949992003, "grad_norm": 1.5913989543914795, "learning_rate": 4.9822e-06, "loss": 0.0225, "step": 25090 }, { "epoch": 1.337687263421656, "grad_norm": 1.6565542221069336, "learning_rate": 4.982e-06, "loss": 0.015, "step": 25091 }, { "epoch": 1.3377405768513089, "grad_norm": 1.6449378728866577, "learning_rate": 4.9818e-06, "loss": 0.0282, "step": 25092 }, { "epoch": 1.3377938902809619, "grad_norm": 3.394273281097412, "learning_rate": 4.981600000000001e-06, "loss": 0.0307, "step": 25093 }, { "epoch": 1.3378472037106146, "grad_norm": 1.6350457668304443, "learning_rate": 4.9814e-06, "loss": 0.0222, "step": 25094 }, { "epoch": 1.3379005171402676, "grad_norm": 1.513279914855957, "learning_rate": 4.981200000000001e-06, "loss": 0.0233, "step": 25095 }, { "epoch": 1.3379538305699206, "grad_norm": 1.1608773469924927, "learning_rate": 4.981e-06, "loss": 0.0147, "step": 25096 }, { "epoch": 1.3380071439995735, "grad_norm": 3.3036789894104004, "learning_rate": 4.980800000000001e-06, "loss": 0.0231, "step": 25097 }, { "epoch": 1.3380604574292265, "grad_norm": 2.3278088569641113, "learning_rate": 4.9806e-06, "loss": 0.0272, "step": 25098 }, { "epoch": 1.3381137708588793, "grad_norm": 1.0547831058502197, "learning_rate": 4.980400000000001e-06, "loss": 0.0185, "step": 25099 }, { "epoch": 1.3381670842885323, "grad_norm": 1.762580394744873, "learning_rate": 4.9802e-06, "loss": 0.0371, "step": 25100 }, { "epoch": 1.3382203977181852, "grad_norm": 1.90603768825531, "learning_rate": 4.980000000000001e-06, "loss": 0.0157, "step": 25101 }, { "epoch": 1.3382737111478382, "grad_norm": 1.298308253288269, "learning_rate": 4.9798e-06, "loss": 0.0132, "step": 25102 }, { "epoch": 1.3383270245774912, "grad_norm": 0.9842017292976379, "learning_rate": 4.9796000000000006e-06, "loss": 0.0232, "step": 25103 }, { "epoch": 1.338380338007144, "grad_norm": 2.5726046562194824, "learning_rate": 4.9794e-06, "loss": 0.0487, "step": 25104 }, { "epoch": 1.338433651436797, "grad_norm": 2.2987399101257324, "learning_rate": 4.9792000000000005e-06, "loss": 0.0484, "step": 25105 }, { "epoch": 1.33848696486645, "grad_norm": 1.971207618713379, "learning_rate": 4.979e-06, "loss": 0.0332, "step": 25106 }, { "epoch": 1.3385402782961027, "grad_norm": 0.7963091731071472, "learning_rate": 4.9788000000000005e-06, "loss": 0.0134, "step": 25107 }, { "epoch": 1.3385935917257556, "grad_norm": 1.893996238708496, "learning_rate": 4.978600000000001e-06, "loss": 0.0505, "step": 25108 }, { "epoch": 1.3386469051554086, "grad_norm": 1.2240371704101562, "learning_rate": 4.9784000000000005e-06, "loss": 0.0093, "step": 25109 }, { "epoch": 1.3387002185850616, "grad_norm": 0.6270745396614075, "learning_rate": 4.9782e-06, "loss": 0.0058, "step": 25110 }, { "epoch": 1.3387535320147146, "grad_norm": 0.8434290289878845, "learning_rate": 4.9780000000000005e-06, "loss": 0.0224, "step": 25111 }, { "epoch": 1.3388068454443673, "grad_norm": 1.790263056755066, "learning_rate": 4.9778e-06, "loss": 0.015, "step": 25112 }, { "epoch": 1.3388601588740203, "grad_norm": 1.9061423540115356, "learning_rate": 4.9776e-06, "loss": 0.0231, "step": 25113 }, { "epoch": 1.3389134723036733, "grad_norm": 1.1148687601089478, "learning_rate": 4.977400000000001e-06, "loss": 0.0251, "step": 25114 }, { "epoch": 1.3389667857333263, "grad_norm": 1.8728406429290771, "learning_rate": 4.9772e-06, "loss": 0.017, "step": 25115 }, { "epoch": 1.3390200991629793, "grad_norm": 1.3347699642181396, "learning_rate": 4.977e-06, "loss": 0.026, "step": 25116 }, { "epoch": 1.339073412592632, "grad_norm": 0.9963232278823853, "learning_rate": 4.9768e-06, "loss": 0.0163, "step": 25117 }, { "epoch": 1.339126726022285, "grad_norm": 1.5634628534317017, "learning_rate": 4.9766e-06, "loss": 0.0164, "step": 25118 }, { "epoch": 1.339180039451938, "grad_norm": 2.965528964996338, "learning_rate": 4.9764e-06, "loss": 0.0359, "step": 25119 }, { "epoch": 1.339233352881591, "grad_norm": 2.353728771209717, "learning_rate": 4.976200000000001e-06, "loss": 0.0161, "step": 25120 }, { "epoch": 1.339286666311244, "grad_norm": 0.8479993939399719, "learning_rate": 4.976e-06, "loss": 0.019, "step": 25121 }, { "epoch": 1.3393399797408967, "grad_norm": 1.380325436592102, "learning_rate": 4.975800000000001e-06, "loss": 0.0222, "step": 25122 }, { "epoch": 1.3393932931705497, "grad_norm": 1.850885033607483, "learning_rate": 4.9756e-06, "loss": 0.0178, "step": 25123 }, { "epoch": 1.3394466066002026, "grad_norm": 0.4477905035018921, "learning_rate": 4.9754e-06, "loss": 0.0044, "step": 25124 }, { "epoch": 1.3394999200298554, "grad_norm": 1.7664313316345215, "learning_rate": 4.9752e-06, "loss": 0.0382, "step": 25125 }, { "epoch": 1.3395532334595084, "grad_norm": 0.8378095030784607, "learning_rate": 4.975000000000001e-06, "loss": 0.0247, "step": 25126 }, { "epoch": 1.3396065468891614, "grad_norm": 1.056671142578125, "learning_rate": 4.9748e-06, "loss": 0.0082, "step": 25127 }, { "epoch": 1.3396598603188143, "grad_norm": 1.4910991191864014, "learning_rate": 4.974600000000001e-06, "loss": 0.0463, "step": 25128 }, { "epoch": 1.3397131737484673, "grad_norm": 1.5584723949432373, "learning_rate": 4.9744e-06, "loss": 0.0376, "step": 25129 }, { "epoch": 1.33976648717812, "grad_norm": 2.7718050479888916, "learning_rate": 4.9742e-06, "loss": 0.01, "step": 25130 }, { "epoch": 1.339819800607773, "grad_norm": 1.7154279947280884, "learning_rate": 4.974e-06, "loss": 0.023, "step": 25131 }, { "epoch": 1.339873114037426, "grad_norm": 2.0060794353485107, "learning_rate": 4.9738000000000006e-06, "loss": 0.0324, "step": 25132 }, { "epoch": 1.339926427467079, "grad_norm": 2.111928939819336, "learning_rate": 4.973600000000001e-06, "loss": 0.04, "step": 25133 }, { "epoch": 1.339979740896732, "grad_norm": 0.5749636888504028, "learning_rate": 4.9734000000000005e-06, "loss": 0.0056, "step": 25134 }, { "epoch": 1.3400330543263848, "grad_norm": 1.4713252782821655, "learning_rate": 4.9732e-06, "loss": 0.0116, "step": 25135 }, { "epoch": 1.3400863677560377, "grad_norm": 0.8114433884620667, "learning_rate": 4.9730000000000005e-06, "loss": 0.0068, "step": 25136 }, { "epoch": 1.3401396811856907, "grad_norm": 1.046118140220642, "learning_rate": 4.9728e-06, "loss": 0.0291, "step": 25137 }, { "epoch": 1.3401929946153435, "grad_norm": 1.509114384651184, "learning_rate": 4.9726000000000005e-06, "loss": 0.0233, "step": 25138 }, { "epoch": 1.3402463080449967, "grad_norm": 1.8378379344940186, "learning_rate": 4.972400000000001e-06, "loss": 0.0338, "step": 25139 }, { "epoch": 1.3402996214746494, "grad_norm": 3.0029685497283936, "learning_rate": 4.9722000000000004e-06, "loss": 0.0274, "step": 25140 }, { "epoch": 1.3403529349043024, "grad_norm": 1.4804601669311523, "learning_rate": 4.972e-06, "loss": 0.0133, "step": 25141 }, { "epoch": 1.3404062483339554, "grad_norm": 0.5105425119400024, "learning_rate": 4.9718e-06, "loss": 0.0055, "step": 25142 }, { "epoch": 1.3404595617636081, "grad_norm": 1.9242420196533203, "learning_rate": 4.9716e-06, "loss": 0.0243, "step": 25143 }, { "epoch": 1.3405128751932611, "grad_norm": 2.181421995162964, "learning_rate": 4.9714e-06, "loss": 0.025, "step": 25144 }, { "epoch": 1.340566188622914, "grad_norm": 1.7167302370071411, "learning_rate": 4.971200000000001e-06, "loss": 0.0187, "step": 25145 }, { "epoch": 1.340619502052567, "grad_norm": 1.524033784866333, "learning_rate": 4.971e-06, "loss": 0.0242, "step": 25146 }, { "epoch": 1.34067281548222, "grad_norm": 2.0996620655059814, "learning_rate": 4.970800000000001e-06, "loss": 0.0281, "step": 25147 }, { "epoch": 1.3407261289118728, "grad_norm": 0.9313476085662842, "learning_rate": 4.9706e-06, "loss": 0.0117, "step": 25148 }, { "epoch": 1.3407794423415258, "grad_norm": 1.6228452920913696, "learning_rate": 4.9704e-06, "loss": 0.0388, "step": 25149 }, { "epoch": 1.3408327557711788, "grad_norm": 1.9069862365722656, "learning_rate": 4.9702e-06, "loss": 0.0277, "step": 25150 }, { "epoch": 1.3408860692008318, "grad_norm": 3.6855809688568115, "learning_rate": 4.970000000000001e-06, "loss": 0.0377, "step": 25151 }, { "epoch": 1.3409393826304847, "grad_norm": 1.719280481338501, "learning_rate": 4.9698e-06, "loss": 0.0335, "step": 25152 }, { "epoch": 1.3409926960601375, "grad_norm": 1.0916849374771118, "learning_rate": 4.969600000000001e-06, "loss": 0.01, "step": 25153 }, { "epoch": 1.3410460094897905, "grad_norm": 2.7691235542297363, "learning_rate": 4.9694e-06, "loss": 0.0173, "step": 25154 }, { "epoch": 1.3410993229194434, "grad_norm": 0.6720699667930603, "learning_rate": 4.9692e-06, "loss": 0.0068, "step": 25155 }, { "epoch": 1.3411526363490962, "grad_norm": 1.1120198965072632, "learning_rate": 4.969e-06, "loss": 0.0183, "step": 25156 }, { "epoch": 1.3412059497787492, "grad_norm": 1.3061797618865967, "learning_rate": 4.968800000000001e-06, "loss": 0.027, "step": 25157 }, { "epoch": 1.3412592632084022, "grad_norm": 1.8298132419586182, "learning_rate": 4.968600000000001e-06, "loss": 0.0126, "step": 25158 }, { "epoch": 1.3413125766380551, "grad_norm": 1.9131821393966675, "learning_rate": 4.9684000000000006e-06, "loss": 0.0222, "step": 25159 }, { "epoch": 1.3413658900677081, "grad_norm": 2.6309664249420166, "learning_rate": 4.9682e-06, "loss": 0.0347, "step": 25160 }, { "epoch": 1.3414192034973609, "grad_norm": 1.1160708665847778, "learning_rate": 4.9680000000000005e-06, "loss": 0.0163, "step": 25161 }, { "epoch": 1.3414725169270139, "grad_norm": 1.6059905290603638, "learning_rate": 4.9678e-06, "loss": 0.0081, "step": 25162 }, { "epoch": 1.3415258303566668, "grad_norm": 1.6817415952682495, "learning_rate": 4.9676000000000005e-06, "loss": 0.0199, "step": 25163 }, { "epoch": 1.3415791437863198, "grad_norm": 1.1727335453033447, "learning_rate": 4.967400000000001e-06, "loss": 0.0072, "step": 25164 }, { "epoch": 1.3416324572159728, "grad_norm": 2.7345869541168213, "learning_rate": 4.9672000000000005e-06, "loss": 0.0356, "step": 25165 }, { "epoch": 1.3416857706456256, "grad_norm": 1.0773475170135498, "learning_rate": 4.967e-06, "loss": 0.0113, "step": 25166 }, { "epoch": 1.3417390840752785, "grad_norm": 0.6947882771492004, "learning_rate": 4.9668000000000005e-06, "loss": 0.0061, "step": 25167 }, { "epoch": 1.3417923975049315, "grad_norm": 0.9353252053260803, "learning_rate": 4.9666e-06, "loss": 0.0075, "step": 25168 }, { "epoch": 1.3418457109345845, "grad_norm": 3.1933372020721436, "learning_rate": 4.9664000000000004e-06, "loss": 0.0229, "step": 25169 }, { "epoch": 1.3418990243642375, "grad_norm": 2.065761089324951, "learning_rate": 4.966200000000001e-06, "loss": 0.0139, "step": 25170 }, { "epoch": 1.3419523377938902, "grad_norm": 1.606673240661621, "learning_rate": 4.966e-06, "loss": 0.0216, "step": 25171 }, { "epoch": 1.3420056512235432, "grad_norm": 3.1037397384643555, "learning_rate": 4.965800000000001e-06, "loss": 0.0473, "step": 25172 }, { "epoch": 1.3420589646531962, "grad_norm": 1.8278779983520508, "learning_rate": 4.9656e-06, "loss": 0.025, "step": 25173 }, { "epoch": 1.342112278082849, "grad_norm": 1.8317292928695679, "learning_rate": 4.9654e-06, "loss": 0.0211, "step": 25174 }, { "epoch": 1.342165591512502, "grad_norm": 1.7773380279541016, "learning_rate": 4.9652e-06, "loss": 0.0116, "step": 25175 }, { "epoch": 1.342218904942155, "grad_norm": 2.029326915740967, "learning_rate": 4.965000000000001e-06, "loss": 0.0173, "step": 25176 }, { "epoch": 1.3422722183718079, "grad_norm": 1.505615234375, "learning_rate": 4.9648e-06, "loss": 0.0068, "step": 25177 }, { "epoch": 1.3423255318014609, "grad_norm": 4.96890926361084, "learning_rate": 4.964600000000001e-06, "loss": 0.0251, "step": 25178 }, { "epoch": 1.3423788452311136, "grad_norm": 2.4706218242645264, "learning_rate": 4.9644e-06, "loss": 0.0215, "step": 25179 }, { "epoch": 1.3424321586607666, "grad_norm": 0.9873162508010864, "learning_rate": 4.9642e-06, "loss": 0.0156, "step": 25180 }, { "epoch": 1.3424854720904196, "grad_norm": 1.3930611610412598, "learning_rate": 4.964e-06, "loss": 0.0226, "step": 25181 }, { "epoch": 1.3425387855200726, "grad_norm": 2.6453945636749268, "learning_rate": 4.963800000000001e-06, "loss": 0.0364, "step": 25182 }, { "epoch": 1.3425920989497255, "grad_norm": 0.7426552176475525, "learning_rate": 4.963600000000001e-06, "loss": 0.0056, "step": 25183 }, { "epoch": 1.3426454123793783, "grad_norm": 1.0739418268203735, "learning_rate": 4.963400000000001e-06, "loss": 0.0195, "step": 25184 }, { "epoch": 1.3426987258090313, "grad_norm": 2.3229405879974365, "learning_rate": 4.9632e-06, "loss": 0.0354, "step": 25185 }, { "epoch": 1.3427520392386842, "grad_norm": 1.8041514158248901, "learning_rate": 4.963000000000001e-06, "loss": 0.039, "step": 25186 }, { "epoch": 1.3428053526683372, "grad_norm": 0.9805278182029724, "learning_rate": 4.9628e-06, "loss": 0.0137, "step": 25187 }, { "epoch": 1.3428586660979902, "grad_norm": 0.9547961354255676, "learning_rate": 4.9626000000000006e-06, "loss": 0.0194, "step": 25188 }, { "epoch": 1.342911979527643, "grad_norm": 1.0407495498657227, "learning_rate": 4.962400000000001e-06, "loss": 0.0074, "step": 25189 }, { "epoch": 1.342965292957296, "grad_norm": 0.8440626263618469, "learning_rate": 4.9622000000000005e-06, "loss": 0.0093, "step": 25190 }, { "epoch": 1.343018606386949, "grad_norm": 1.4773690700531006, "learning_rate": 4.962e-06, "loss": 0.0145, "step": 25191 }, { "epoch": 1.3430719198166017, "grad_norm": 2.005504846572876, "learning_rate": 4.9618000000000005e-06, "loss": 0.0146, "step": 25192 }, { "epoch": 1.3431252332462547, "grad_norm": 1.893417239189148, "learning_rate": 4.9616e-06, "loss": 0.0251, "step": 25193 }, { "epoch": 1.3431785466759076, "grad_norm": 1.519319772720337, "learning_rate": 4.9614000000000005e-06, "loss": 0.0104, "step": 25194 }, { "epoch": 1.3432318601055606, "grad_norm": 1.3047595024108887, "learning_rate": 4.961200000000001e-06, "loss": 0.0185, "step": 25195 }, { "epoch": 1.3432851735352136, "grad_norm": 2.958794116973877, "learning_rate": 4.9610000000000004e-06, "loss": 0.0172, "step": 25196 }, { "epoch": 1.3433384869648664, "grad_norm": 1.0322694778442383, "learning_rate": 4.960800000000001e-06, "loss": 0.02, "step": 25197 }, { "epoch": 1.3433918003945193, "grad_norm": 1.2260721921920776, "learning_rate": 4.9606e-06, "loss": 0.0091, "step": 25198 }, { "epoch": 1.3434451138241723, "grad_norm": 0.6581681370735168, "learning_rate": 4.9604e-06, "loss": 0.0044, "step": 25199 }, { "epoch": 1.3434984272538253, "grad_norm": 1.0811666250228882, "learning_rate": 4.9602e-06, "loss": 0.0362, "step": 25200 }, { "epoch": 1.3435517406834783, "grad_norm": 2.6731529235839844, "learning_rate": 4.960000000000001e-06, "loss": 0.0483, "step": 25201 }, { "epoch": 1.343605054113131, "grad_norm": 1.6356909275054932, "learning_rate": 4.9598e-06, "loss": 0.0088, "step": 25202 }, { "epoch": 1.343658367542784, "grad_norm": 1.6854194402694702, "learning_rate": 4.959600000000001e-06, "loss": 0.0184, "step": 25203 }, { "epoch": 1.343711680972437, "grad_norm": 1.4093742370605469, "learning_rate": 4.9594e-06, "loss": 0.021, "step": 25204 }, { "epoch": 1.34376499440209, "grad_norm": 1.725500226020813, "learning_rate": 4.959200000000001e-06, "loss": 0.0303, "step": 25205 }, { "epoch": 1.343818307831743, "grad_norm": 1.6534228324890137, "learning_rate": 4.959e-06, "loss": 0.0314, "step": 25206 }, { "epoch": 1.3438716212613957, "grad_norm": 0.7944475412368774, "learning_rate": 4.958800000000001e-06, "loss": 0.0069, "step": 25207 }, { "epoch": 1.3439249346910487, "grad_norm": 2.342033624649048, "learning_rate": 4.9586e-06, "loss": 0.0403, "step": 25208 }, { "epoch": 1.3439782481207017, "grad_norm": 0.8117013573646545, "learning_rate": 4.958400000000001e-06, "loss": 0.0059, "step": 25209 }, { "epoch": 1.3440315615503544, "grad_norm": 1.731886863708496, "learning_rate": 4.9582e-06, "loss": 0.025, "step": 25210 }, { "epoch": 1.3440848749800074, "grad_norm": 1.5368810892105103, "learning_rate": 4.958000000000001e-06, "loss": 0.0213, "step": 25211 }, { "epoch": 1.3441381884096604, "grad_norm": 1.9404988288879395, "learning_rate": 4.9578e-06, "loss": 0.0312, "step": 25212 }, { "epoch": 1.3441915018393134, "grad_norm": 1.9768385887145996, "learning_rate": 4.957600000000001e-06, "loss": 0.0289, "step": 25213 }, { "epoch": 1.3442448152689663, "grad_norm": 2.8354625701904297, "learning_rate": 4.9574e-06, "loss": 0.0267, "step": 25214 }, { "epoch": 1.344298128698619, "grad_norm": 1.993886947631836, "learning_rate": 4.957200000000001e-06, "loss": 0.0407, "step": 25215 }, { "epoch": 1.344351442128272, "grad_norm": 1.5106250047683716, "learning_rate": 4.957e-06, "loss": 0.0119, "step": 25216 }, { "epoch": 1.344404755557925, "grad_norm": 1.6350492238998413, "learning_rate": 4.9568000000000006e-06, "loss": 0.0133, "step": 25217 }, { "epoch": 1.344458068987578, "grad_norm": 5.8769426345825195, "learning_rate": 4.9566e-06, "loss": 0.0249, "step": 25218 }, { "epoch": 1.344511382417231, "grad_norm": 2.9603707790374756, "learning_rate": 4.9564000000000005e-06, "loss": 0.0267, "step": 25219 }, { "epoch": 1.3445646958468838, "grad_norm": 1.1090642213821411, "learning_rate": 4.9562e-06, "loss": 0.0086, "step": 25220 }, { "epoch": 1.3446180092765367, "grad_norm": 1.6170316934585571, "learning_rate": 4.9560000000000005e-06, "loss": 0.0143, "step": 25221 }, { "epoch": 1.3446713227061897, "grad_norm": 1.5030367374420166, "learning_rate": 4.955800000000001e-06, "loss": 0.0138, "step": 25222 }, { "epoch": 1.3447246361358427, "grad_norm": 1.5365114212036133, "learning_rate": 4.9556000000000005e-06, "loss": 0.0219, "step": 25223 }, { "epoch": 1.3447779495654957, "grad_norm": 1.3234058618545532, "learning_rate": 4.9554e-06, "loss": 0.0199, "step": 25224 }, { "epoch": 1.3448312629951484, "grad_norm": 2.8969717025756836, "learning_rate": 4.9552000000000004e-06, "loss": 0.0376, "step": 25225 }, { "epoch": 1.3448845764248014, "grad_norm": 1.673467993736267, "learning_rate": 4.955e-06, "loss": 0.0351, "step": 25226 }, { "epoch": 1.3449378898544544, "grad_norm": 2.266171455383301, "learning_rate": 4.9548e-06, "loss": 0.013, "step": 25227 }, { "epoch": 1.3449912032841072, "grad_norm": 2.5063512325286865, "learning_rate": 4.954600000000001e-06, "loss": 0.015, "step": 25228 }, { "epoch": 1.3450445167137601, "grad_norm": 1.3667563199996948, "learning_rate": 4.9544e-06, "loss": 0.0171, "step": 25229 }, { "epoch": 1.345097830143413, "grad_norm": 1.7176486253738403, "learning_rate": 4.954200000000001e-06, "loss": 0.0221, "step": 25230 }, { "epoch": 1.345151143573066, "grad_norm": 1.5284262895584106, "learning_rate": 4.954e-06, "loss": 0.011, "step": 25231 }, { "epoch": 1.345204457002719, "grad_norm": 0.7518265843391418, "learning_rate": 4.9538e-06, "loss": 0.0043, "step": 25232 }, { "epoch": 1.3452577704323718, "grad_norm": 2.8193068504333496, "learning_rate": 4.9536e-06, "loss": 0.0259, "step": 25233 }, { "epoch": 1.3453110838620248, "grad_norm": 3.491291046142578, "learning_rate": 4.953400000000001e-06, "loss": 0.0479, "step": 25234 }, { "epoch": 1.3453643972916778, "grad_norm": 1.830997109413147, "learning_rate": 4.9532e-06, "loss": 0.0102, "step": 25235 }, { "epoch": 1.3454177107213308, "grad_norm": 2.1016383171081543, "learning_rate": 4.953000000000001e-06, "loss": 0.0164, "step": 25236 }, { "epoch": 1.3454710241509837, "grad_norm": 1.476577639579773, "learning_rate": 4.9528e-06, "loss": 0.0119, "step": 25237 }, { "epoch": 1.3455243375806365, "grad_norm": 2.406316041946411, "learning_rate": 4.9526e-06, "loss": 0.0171, "step": 25238 }, { "epoch": 1.3455776510102895, "grad_norm": 2.583571195602417, "learning_rate": 4.9524e-06, "loss": 0.011, "step": 25239 }, { "epoch": 1.3456309644399425, "grad_norm": 2.8900444507598877, "learning_rate": 4.952200000000001e-06, "loss": 0.0337, "step": 25240 }, { "epoch": 1.3456842778695952, "grad_norm": 2.348642349243164, "learning_rate": 4.952e-06, "loss": 0.0309, "step": 25241 }, { "epoch": 1.3457375912992484, "grad_norm": 2.4567859172821045, "learning_rate": 4.951800000000001e-06, "loss": 0.018, "step": 25242 }, { "epoch": 1.3457909047289012, "grad_norm": 2.2572319507598877, "learning_rate": 4.9516e-06, "loss": 0.0186, "step": 25243 }, { "epoch": 1.3458442181585542, "grad_norm": 2.8011677265167236, "learning_rate": 4.9514000000000006e-06, "loss": 0.0229, "step": 25244 }, { "epoch": 1.3458975315882071, "grad_norm": 3.8917977809906006, "learning_rate": 4.9512e-06, "loss": 0.0283, "step": 25245 }, { "epoch": 1.3459508450178599, "grad_norm": 2.938336133956909, "learning_rate": 4.9510000000000005e-06, "loss": 0.0117, "step": 25246 }, { "epoch": 1.3460041584475129, "grad_norm": 2.1158907413482666, "learning_rate": 4.950800000000001e-06, "loss": 0.025, "step": 25247 }, { "epoch": 1.3460574718771658, "grad_norm": 2.686230182647705, "learning_rate": 4.9506000000000005e-06, "loss": 0.0417, "step": 25248 }, { "epoch": 1.3461107853068188, "grad_norm": 1.694549560546875, "learning_rate": 4.9504e-06, "loss": 0.0203, "step": 25249 }, { "epoch": 1.3461640987364718, "grad_norm": 1.7165642976760864, "learning_rate": 4.9502000000000005e-06, "loss": 0.041, "step": 25250 }, { "epoch": 1.3462174121661246, "grad_norm": 2.4297943115234375, "learning_rate": 4.95e-06, "loss": 0.0511, "step": 25251 }, { "epoch": 1.3462707255957775, "grad_norm": 1.3773349523544312, "learning_rate": 4.9498000000000004e-06, "loss": 0.013, "step": 25252 }, { "epoch": 1.3463240390254305, "grad_norm": 1.6053682565689087, "learning_rate": 4.949600000000001e-06, "loss": 0.0067, "step": 25253 }, { "epoch": 1.3463773524550835, "grad_norm": 1.7555394172668457, "learning_rate": 4.9494e-06, "loss": 0.0251, "step": 25254 }, { "epoch": 1.3464306658847365, "grad_norm": 0.8736181855201721, "learning_rate": 4.949200000000001e-06, "loss": 0.0072, "step": 25255 }, { "epoch": 1.3464839793143892, "grad_norm": 1.4536668062210083, "learning_rate": 4.949e-06, "loss": 0.0131, "step": 25256 }, { "epoch": 1.3465372927440422, "grad_norm": 1.4521417617797852, "learning_rate": 4.9488e-06, "loss": 0.0082, "step": 25257 }, { "epoch": 1.3465906061736952, "grad_norm": 1.4518511295318604, "learning_rate": 4.9486e-06, "loss": 0.0117, "step": 25258 }, { "epoch": 1.346643919603348, "grad_norm": 1.2637187242507935, "learning_rate": 4.948400000000001e-06, "loss": 0.0158, "step": 25259 }, { "epoch": 1.3466972330330012, "grad_norm": 1.9677180051803589, "learning_rate": 4.9482e-06, "loss": 0.0492, "step": 25260 }, { "epoch": 1.346750546462654, "grad_norm": 2.3723068237304688, "learning_rate": 4.948000000000001e-06, "loss": 0.0194, "step": 25261 }, { "epoch": 1.3468038598923069, "grad_norm": 2.644676685333252, "learning_rate": 4.9478e-06, "loss": 0.0173, "step": 25262 }, { "epoch": 1.3468571733219599, "grad_norm": 1.5951286554336548, "learning_rate": 4.9476e-06, "loss": 0.0625, "step": 25263 }, { "epoch": 1.3469104867516126, "grad_norm": 1.6715636253356934, "learning_rate": 4.9474e-06, "loss": 0.0314, "step": 25264 }, { "epoch": 1.3469638001812656, "grad_norm": 5.467575550079346, "learning_rate": 4.947200000000001e-06, "loss": 0.0421, "step": 25265 }, { "epoch": 1.3470171136109186, "grad_norm": 0.6289035677909851, "learning_rate": 4.947e-06, "loss": 0.0051, "step": 25266 }, { "epoch": 1.3470704270405716, "grad_norm": 1.3127720355987549, "learning_rate": 4.946800000000001e-06, "loss": 0.0272, "step": 25267 }, { "epoch": 1.3471237404702245, "grad_norm": 2.153127670288086, "learning_rate": 4.9466e-06, "loss": 0.0144, "step": 25268 }, { "epoch": 1.3471770538998773, "grad_norm": 2.4083633422851562, "learning_rate": 4.946400000000001e-06, "loss": 0.022, "step": 25269 }, { "epoch": 1.3472303673295303, "grad_norm": 0.6551005840301514, "learning_rate": 4.9462e-06, "loss": 0.0047, "step": 25270 }, { "epoch": 1.3472836807591833, "grad_norm": 2.1076924800872803, "learning_rate": 4.946000000000001e-06, "loss": 0.0373, "step": 25271 }, { "epoch": 1.3473369941888362, "grad_norm": 1.442094326019287, "learning_rate": 4.945800000000001e-06, "loss": 0.0348, "step": 25272 }, { "epoch": 1.3473903076184892, "grad_norm": 2.0182559490203857, "learning_rate": 4.9456000000000006e-06, "loss": 0.0379, "step": 25273 }, { "epoch": 1.347443621048142, "grad_norm": 1.4919312000274658, "learning_rate": 4.9454e-06, "loss": 0.0127, "step": 25274 }, { "epoch": 1.347496934477795, "grad_norm": 1.9346061944961548, "learning_rate": 4.9452000000000005e-06, "loss": 0.0411, "step": 25275 }, { "epoch": 1.347550247907448, "grad_norm": 1.7122623920440674, "learning_rate": 4.945e-06, "loss": 0.0218, "step": 25276 }, { "epoch": 1.3476035613371007, "grad_norm": 1.1870310306549072, "learning_rate": 4.9448000000000005e-06, "loss": 0.0206, "step": 25277 }, { "epoch": 1.3476568747667537, "grad_norm": 2.8088016510009766, "learning_rate": 4.944600000000001e-06, "loss": 0.0253, "step": 25278 }, { "epoch": 1.3477101881964066, "grad_norm": 2.6902360916137695, "learning_rate": 4.9444000000000005e-06, "loss": 0.043, "step": 25279 }, { "epoch": 1.3477635016260596, "grad_norm": 1.0708385705947876, "learning_rate": 4.944200000000001e-06, "loss": 0.0244, "step": 25280 }, { "epoch": 1.3478168150557126, "grad_norm": 2.2259681224823, "learning_rate": 4.9440000000000004e-06, "loss": 0.0382, "step": 25281 }, { "epoch": 1.3478701284853654, "grad_norm": 0.9915658831596375, "learning_rate": 4.9438e-06, "loss": 0.0165, "step": 25282 }, { "epoch": 1.3479234419150183, "grad_norm": 2.9083352088928223, "learning_rate": 4.9436e-06, "loss": 0.0197, "step": 25283 }, { "epoch": 1.3479767553446713, "grad_norm": 2.484769821166992, "learning_rate": 4.943400000000001e-06, "loss": 0.025, "step": 25284 }, { "epoch": 1.3480300687743243, "grad_norm": 1.2227972745895386, "learning_rate": 4.9432e-06, "loss": 0.0116, "step": 25285 }, { "epoch": 1.3480833822039773, "grad_norm": 2.2323074340820312, "learning_rate": 4.943000000000001e-06, "loss": 0.0132, "step": 25286 }, { "epoch": 1.34813669563363, "grad_norm": 1.6178370714187622, "learning_rate": 4.9428e-06, "loss": 0.0153, "step": 25287 }, { "epoch": 1.348190009063283, "grad_norm": 1.6706629991531372, "learning_rate": 4.9426e-06, "loss": 0.0107, "step": 25288 }, { "epoch": 1.348243322492936, "grad_norm": 2.366133213043213, "learning_rate": 4.9424e-06, "loss": 0.0146, "step": 25289 }, { "epoch": 1.348296635922589, "grad_norm": 1.52796471118927, "learning_rate": 4.942200000000001e-06, "loss": 0.0132, "step": 25290 }, { "epoch": 1.348349949352242, "grad_norm": 1.0093978643417358, "learning_rate": 4.942e-06, "loss": 0.0128, "step": 25291 }, { "epoch": 1.3484032627818947, "grad_norm": 0.6759973168373108, "learning_rate": 4.941800000000001e-06, "loss": 0.0083, "step": 25292 }, { "epoch": 1.3484565762115477, "grad_norm": 2.0554184913635254, "learning_rate": 4.9416e-06, "loss": 0.0219, "step": 25293 }, { "epoch": 1.3485098896412007, "grad_norm": 1.5780423879623413, "learning_rate": 4.941400000000001e-06, "loss": 0.019, "step": 25294 }, { "epoch": 1.3485632030708534, "grad_norm": 2.487071990966797, "learning_rate": 4.9412e-06, "loss": 0.0305, "step": 25295 }, { "epoch": 1.3486165165005064, "grad_norm": 1.6790395975112915, "learning_rate": 4.941000000000001e-06, "loss": 0.0215, "step": 25296 }, { "epoch": 1.3486698299301594, "grad_norm": 2.4704487323760986, "learning_rate": 4.9408e-06, "loss": 0.0426, "step": 25297 }, { "epoch": 1.3487231433598124, "grad_norm": 1.2192296981811523, "learning_rate": 4.940600000000001e-06, "loss": 0.0165, "step": 25298 }, { "epoch": 1.3487764567894653, "grad_norm": 1.547875165939331, "learning_rate": 4.9404e-06, "loss": 0.0334, "step": 25299 }, { "epoch": 1.348829770219118, "grad_norm": 3.7133584022521973, "learning_rate": 4.9402000000000006e-06, "loss": 0.0316, "step": 25300 }, { "epoch": 1.348883083648771, "grad_norm": 3.1828441619873047, "learning_rate": 4.94e-06, "loss": 0.0493, "step": 25301 }, { "epoch": 1.348936397078424, "grad_norm": 1.563136100769043, "learning_rate": 4.9398000000000005e-06, "loss": 0.0229, "step": 25302 }, { "epoch": 1.348989710508077, "grad_norm": 1.312830924987793, "learning_rate": 4.9396e-06, "loss": 0.0171, "step": 25303 }, { "epoch": 1.34904302393773, "grad_norm": 1.6880875825881958, "learning_rate": 4.9394000000000005e-06, "loss": 0.0187, "step": 25304 }, { "epoch": 1.3490963373673828, "grad_norm": 0.8344963788986206, "learning_rate": 4.939200000000001e-06, "loss": 0.0134, "step": 25305 }, { "epoch": 1.3491496507970357, "grad_norm": 2.354990005493164, "learning_rate": 4.9390000000000005e-06, "loss": 0.0299, "step": 25306 }, { "epoch": 1.3492029642266887, "grad_norm": 0.7984868884086609, "learning_rate": 4.9388e-06, "loss": 0.0144, "step": 25307 }, { "epoch": 1.3492562776563417, "grad_norm": 2.5808002948760986, "learning_rate": 4.9386000000000005e-06, "loss": 0.0231, "step": 25308 }, { "epoch": 1.3493095910859947, "grad_norm": 1.876117467880249, "learning_rate": 4.9384e-06, "loss": 0.0103, "step": 25309 }, { "epoch": 1.3493629045156474, "grad_norm": 3.2572333812713623, "learning_rate": 4.9382000000000004e-06, "loss": 0.0335, "step": 25310 }, { "epoch": 1.3494162179453004, "grad_norm": 1.4715920686721802, "learning_rate": 4.938000000000001e-06, "loss": 0.0325, "step": 25311 }, { "epoch": 1.3494695313749534, "grad_norm": 2.078021287918091, "learning_rate": 4.9378e-06, "loss": 0.0214, "step": 25312 }, { "epoch": 1.3495228448046062, "grad_norm": 2.55517315864563, "learning_rate": 4.9376e-06, "loss": 0.0239, "step": 25313 }, { "epoch": 1.3495761582342591, "grad_norm": 1.765435814857483, "learning_rate": 4.9374e-06, "loss": 0.0137, "step": 25314 }, { "epoch": 1.3496294716639121, "grad_norm": 1.7452418804168701, "learning_rate": 4.9372e-06, "loss": 0.0118, "step": 25315 }, { "epoch": 1.349682785093565, "grad_norm": 3.059068441390991, "learning_rate": 4.937e-06, "loss": 0.0466, "step": 25316 }, { "epoch": 1.349736098523218, "grad_norm": 1.137158989906311, "learning_rate": 4.936800000000001e-06, "loss": 0.0329, "step": 25317 }, { "epoch": 1.3497894119528708, "grad_norm": 1.3578863143920898, "learning_rate": 4.9366e-06, "loss": 0.0274, "step": 25318 }, { "epoch": 1.3498427253825238, "grad_norm": 1.8287838697433472, "learning_rate": 4.936400000000001e-06, "loss": 0.0208, "step": 25319 }, { "epoch": 1.3498960388121768, "grad_norm": 1.5309284925460815, "learning_rate": 4.9362e-06, "loss": 0.0198, "step": 25320 }, { "epoch": 1.3499493522418298, "grad_norm": 1.7058782577514648, "learning_rate": 4.936e-06, "loss": 0.0174, "step": 25321 }, { "epoch": 1.3500026656714827, "grad_norm": 1.5399459600448608, "learning_rate": 4.9358e-06, "loss": 0.022, "step": 25322 }, { "epoch": 1.3500559791011355, "grad_norm": 1.328855037689209, "learning_rate": 4.935600000000001e-06, "loss": 0.0083, "step": 25323 }, { "epoch": 1.3501092925307885, "grad_norm": 1.2665632963180542, "learning_rate": 4.9354e-06, "loss": 0.0099, "step": 25324 }, { "epoch": 1.3501626059604415, "grad_norm": 0.9480064511299133, "learning_rate": 4.935200000000001e-06, "loss": 0.0121, "step": 25325 }, { "epoch": 1.3502159193900944, "grad_norm": 2.181302070617676, "learning_rate": 4.935e-06, "loss": 0.0189, "step": 25326 }, { "epoch": 1.3502692328197474, "grad_norm": 2.1810302734375, "learning_rate": 4.9348e-06, "loss": 0.0121, "step": 25327 }, { "epoch": 1.3503225462494002, "grad_norm": 1.3194708824157715, "learning_rate": 4.9346e-06, "loss": 0.0224, "step": 25328 }, { "epoch": 1.3503758596790532, "grad_norm": 1.6220813989639282, "learning_rate": 4.9344000000000006e-06, "loss": 0.0404, "step": 25329 }, { "epoch": 1.3504291731087061, "grad_norm": 3.4761922359466553, "learning_rate": 4.934200000000001e-06, "loss": 0.0364, "step": 25330 }, { "epoch": 1.350482486538359, "grad_norm": 1.6802818775177002, "learning_rate": 4.9340000000000005e-06, "loss": 0.0259, "step": 25331 }, { "epoch": 1.3505357999680119, "grad_norm": 1.211235761642456, "learning_rate": 4.9338e-06, "loss": 0.0175, "step": 25332 }, { "epoch": 1.3505891133976649, "grad_norm": 1.8986679315567017, "learning_rate": 4.9336000000000005e-06, "loss": 0.026, "step": 25333 }, { "epoch": 1.3506424268273178, "grad_norm": 4.038094997406006, "learning_rate": 4.9334e-06, "loss": 0.0435, "step": 25334 }, { "epoch": 1.3506957402569708, "grad_norm": 2.6044559478759766, "learning_rate": 4.9332000000000005e-06, "loss": 0.0323, "step": 25335 }, { "epoch": 1.3507490536866236, "grad_norm": 0.6563793420791626, "learning_rate": 4.933000000000001e-06, "loss": 0.0048, "step": 25336 }, { "epoch": 1.3508023671162765, "grad_norm": 2.1143457889556885, "learning_rate": 4.9328000000000004e-06, "loss": 0.0344, "step": 25337 }, { "epoch": 1.3508556805459295, "grad_norm": 0.546286404132843, "learning_rate": 4.9326e-06, "loss": 0.0042, "step": 25338 }, { "epoch": 1.3509089939755825, "grad_norm": 0.7180734872817993, "learning_rate": 4.9324e-06, "loss": 0.0049, "step": 25339 }, { "epoch": 1.3509623074052355, "grad_norm": 1.0374865531921387, "learning_rate": 4.9322e-06, "loss": 0.0183, "step": 25340 }, { "epoch": 1.3510156208348882, "grad_norm": 1.4227546453475952, "learning_rate": 4.932e-06, "loss": 0.0169, "step": 25341 }, { "epoch": 1.3510689342645412, "grad_norm": 2.3670077323913574, "learning_rate": 4.931800000000001e-06, "loss": 0.0588, "step": 25342 }, { "epoch": 1.3511222476941942, "grad_norm": 1.375081181526184, "learning_rate": 4.9316e-06, "loss": 0.0258, "step": 25343 }, { "epoch": 1.3511755611238472, "grad_norm": 2.247541666030884, "learning_rate": 4.931400000000001e-06, "loss": 0.0517, "step": 25344 }, { "epoch": 1.3512288745535002, "grad_norm": 1.6095161437988281, "learning_rate": 4.9312e-06, "loss": 0.0306, "step": 25345 }, { "epoch": 1.351282187983153, "grad_norm": 1.5672593116760254, "learning_rate": 4.931e-06, "loss": 0.0318, "step": 25346 }, { "epoch": 1.351335501412806, "grad_norm": 2.5144917964935303, "learning_rate": 4.9308e-06, "loss": 0.0413, "step": 25347 }, { "epoch": 1.3513888148424589, "grad_norm": 1.4992506504058838, "learning_rate": 4.930600000000001e-06, "loss": 0.0187, "step": 25348 }, { "epoch": 1.3514421282721116, "grad_norm": 2.21105694770813, "learning_rate": 4.9304e-06, "loss": 0.027, "step": 25349 }, { "epoch": 1.3514954417017646, "grad_norm": 1.902896761894226, "learning_rate": 4.930200000000001e-06, "loss": 0.0217, "step": 25350 }, { "epoch": 1.3515487551314176, "grad_norm": 2.299882411956787, "learning_rate": 4.93e-06, "loss": 0.0312, "step": 25351 }, { "epoch": 1.3516020685610706, "grad_norm": 1.8709402084350586, "learning_rate": 4.9298e-06, "loss": 0.0183, "step": 25352 }, { "epoch": 1.3516553819907235, "grad_norm": 1.0215399265289307, "learning_rate": 4.9296e-06, "loss": 0.0123, "step": 25353 }, { "epoch": 1.3517086954203763, "grad_norm": 1.676068902015686, "learning_rate": 4.929400000000001e-06, "loss": 0.0254, "step": 25354 }, { "epoch": 1.3517620088500293, "grad_norm": 1.393121600151062, "learning_rate": 4.929200000000001e-06, "loss": 0.0096, "step": 25355 }, { "epoch": 1.3518153222796823, "grad_norm": 3.6643364429473877, "learning_rate": 4.929000000000001e-06, "loss": 0.0416, "step": 25356 }, { "epoch": 1.3518686357093352, "grad_norm": 1.160804033279419, "learning_rate": 4.9288e-06, "loss": 0.0115, "step": 25357 }, { "epoch": 1.3519219491389882, "grad_norm": 1.7653058767318726, "learning_rate": 4.9286000000000006e-06, "loss": 0.0395, "step": 25358 }, { "epoch": 1.351975262568641, "grad_norm": 1.6339178085327148, "learning_rate": 4.9284e-06, "loss": 0.0198, "step": 25359 }, { "epoch": 1.352028575998294, "grad_norm": 0.7692732214927673, "learning_rate": 4.9282000000000005e-06, "loss": 0.0106, "step": 25360 }, { "epoch": 1.352081889427947, "grad_norm": 4.138067245483398, "learning_rate": 4.928000000000001e-06, "loss": 0.0516, "step": 25361 }, { "epoch": 1.3521352028575997, "grad_norm": 1.759469985961914, "learning_rate": 4.9278000000000005e-06, "loss": 0.049, "step": 25362 }, { "epoch": 1.352188516287253, "grad_norm": 3.0331878662109375, "learning_rate": 4.9276e-06, "loss": 0.0466, "step": 25363 }, { "epoch": 1.3522418297169057, "grad_norm": 2.2630743980407715, "learning_rate": 4.9274000000000005e-06, "loss": 0.0341, "step": 25364 }, { "epoch": 1.3522951431465586, "grad_norm": 0.5824100375175476, "learning_rate": 4.9272e-06, "loss": 0.0052, "step": 25365 }, { "epoch": 1.3523484565762116, "grad_norm": 1.3176463842391968, "learning_rate": 4.9270000000000004e-06, "loss": 0.0191, "step": 25366 }, { "epoch": 1.3524017700058644, "grad_norm": 1.05362868309021, "learning_rate": 4.926800000000001e-06, "loss": 0.0128, "step": 25367 }, { "epoch": 1.3524550834355173, "grad_norm": 0.9110147356987, "learning_rate": 4.9266e-06, "loss": 0.0159, "step": 25368 }, { "epoch": 1.3525083968651703, "grad_norm": 2.6395227909088135, "learning_rate": 4.926400000000001e-06, "loss": 0.0501, "step": 25369 }, { "epoch": 1.3525617102948233, "grad_norm": 2.4686129093170166, "learning_rate": 4.9262e-06, "loss": 0.0503, "step": 25370 }, { "epoch": 1.3526150237244763, "grad_norm": 1.8037961721420288, "learning_rate": 4.926e-06, "loss": 0.0133, "step": 25371 }, { "epoch": 1.352668337154129, "grad_norm": 1.0344293117523193, "learning_rate": 4.9258e-06, "loss": 0.0198, "step": 25372 }, { "epoch": 1.352721650583782, "grad_norm": 1.3367654085159302, "learning_rate": 4.925600000000001e-06, "loss": 0.0254, "step": 25373 }, { "epoch": 1.352774964013435, "grad_norm": 0.9367170929908752, "learning_rate": 4.9254e-06, "loss": 0.0089, "step": 25374 }, { "epoch": 1.352828277443088, "grad_norm": 0.39903879165649414, "learning_rate": 4.925200000000001e-06, "loss": 0.0047, "step": 25375 }, { "epoch": 1.352881590872741, "grad_norm": 0.846912145614624, "learning_rate": 4.925e-06, "loss": 0.0241, "step": 25376 }, { "epoch": 1.3529349043023937, "grad_norm": 1.0617958307266235, "learning_rate": 4.9248e-06, "loss": 0.0159, "step": 25377 }, { "epoch": 1.3529882177320467, "grad_norm": 2.5189783573150635, "learning_rate": 4.9246e-06, "loss": 0.0528, "step": 25378 }, { "epoch": 1.3530415311616997, "grad_norm": 2.0559327602386475, "learning_rate": 4.924400000000001e-06, "loss": 0.0667, "step": 25379 }, { "epoch": 1.3530948445913524, "grad_norm": 1.9390552043914795, "learning_rate": 4.924200000000001e-06, "loss": 0.0199, "step": 25380 }, { "epoch": 1.3531481580210054, "grad_norm": 1.098495602607727, "learning_rate": 4.924000000000001e-06, "loss": 0.0156, "step": 25381 }, { "epoch": 1.3532014714506584, "grad_norm": 3.0148684978485107, "learning_rate": 4.9238e-06, "loss": 0.0382, "step": 25382 }, { "epoch": 1.3532547848803114, "grad_norm": 1.6312967538833618, "learning_rate": 4.923600000000001e-06, "loss": 0.0184, "step": 25383 }, { "epoch": 1.3533080983099643, "grad_norm": 2.304807186126709, "learning_rate": 4.9234e-06, "loss": 0.0373, "step": 25384 }, { "epoch": 1.353361411739617, "grad_norm": 1.2827794551849365, "learning_rate": 4.9232000000000006e-06, "loss": 0.0171, "step": 25385 }, { "epoch": 1.35341472516927, "grad_norm": 1.378680944442749, "learning_rate": 4.923000000000001e-06, "loss": 0.0138, "step": 25386 }, { "epoch": 1.353468038598923, "grad_norm": 1.8590034246444702, "learning_rate": 4.9228000000000005e-06, "loss": 0.0422, "step": 25387 }, { "epoch": 1.353521352028576, "grad_norm": 2.048196792602539, "learning_rate": 4.9226e-06, "loss": 0.0195, "step": 25388 }, { "epoch": 1.353574665458229, "grad_norm": 0.9720283150672913, "learning_rate": 4.9224000000000005e-06, "loss": 0.0105, "step": 25389 }, { "epoch": 1.3536279788878818, "grad_norm": 0.9642543196678162, "learning_rate": 4.9222e-06, "loss": 0.0077, "step": 25390 }, { "epoch": 1.3536812923175348, "grad_norm": 2.8918983936309814, "learning_rate": 4.9220000000000005e-06, "loss": 0.0322, "step": 25391 }, { "epoch": 1.3537346057471877, "grad_norm": 1.464599370956421, "learning_rate": 4.921800000000001e-06, "loss": 0.0194, "step": 25392 }, { "epoch": 1.3537879191768407, "grad_norm": 1.1712018251419067, "learning_rate": 4.9216000000000004e-06, "loss": 0.0147, "step": 25393 }, { "epoch": 1.3538412326064937, "grad_norm": 0.8652336597442627, "learning_rate": 4.921400000000001e-06, "loss": 0.0067, "step": 25394 }, { "epoch": 1.3538945460361465, "grad_norm": 1.9509105682373047, "learning_rate": 4.9212e-06, "loss": 0.0319, "step": 25395 }, { "epoch": 1.3539478594657994, "grad_norm": 1.2409477233886719, "learning_rate": 4.921e-06, "loss": 0.0177, "step": 25396 }, { "epoch": 1.3540011728954524, "grad_norm": 1.5872907638549805, "learning_rate": 4.9208e-06, "loss": 0.0375, "step": 25397 }, { "epoch": 1.3540544863251052, "grad_norm": 1.519838571548462, "learning_rate": 4.920600000000001e-06, "loss": 0.0298, "step": 25398 }, { "epoch": 1.3541077997547581, "grad_norm": 2.1282401084899902, "learning_rate": 4.9204e-06, "loss": 0.0367, "step": 25399 }, { "epoch": 1.3541611131844111, "grad_norm": 1.190896987915039, "learning_rate": 4.920200000000001e-06, "loss": 0.0165, "step": 25400 }, { "epoch": 1.354214426614064, "grad_norm": 1.3862943649291992, "learning_rate": 4.92e-06, "loss": 0.0375, "step": 25401 }, { "epoch": 1.354267740043717, "grad_norm": 1.3760193586349487, "learning_rate": 4.9198e-06, "loss": 0.0203, "step": 25402 }, { "epoch": 1.3543210534733698, "grad_norm": 1.5074859857559204, "learning_rate": 4.9196e-06, "loss": 0.0308, "step": 25403 }, { "epoch": 1.3543743669030228, "grad_norm": 1.6810741424560547, "learning_rate": 4.919400000000001e-06, "loss": 0.028, "step": 25404 }, { "epoch": 1.3544276803326758, "grad_norm": 0.9920302629470825, "learning_rate": 4.9192e-06, "loss": 0.0271, "step": 25405 }, { "epoch": 1.3544809937623288, "grad_norm": 1.7396759986877441, "learning_rate": 4.919000000000001e-06, "loss": 0.0434, "step": 25406 }, { "epoch": 1.3545343071919818, "grad_norm": 2.115469217300415, "learning_rate": 4.9188e-06, "loss": 0.0599, "step": 25407 }, { "epoch": 1.3545876206216345, "grad_norm": 1.3710432052612305, "learning_rate": 4.918600000000001e-06, "loss": 0.0159, "step": 25408 }, { "epoch": 1.3546409340512875, "grad_norm": 1.6367759704589844, "learning_rate": 4.9184e-06, "loss": 0.0432, "step": 25409 }, { "epoch": 1.3546942474809405, "grad_norm": 2.1553900241851807, "learning_rate": 4.918200000000001e-06, "loss": 0.0174, "step": 25410 }, { "epoch": 1.3547475609105935, "grad_norm": 0.5268833041191101, "learning_rate": 4.918e-06, "loss": 0.0056, "step": 25411 }, { "epoch": 1.3548008743402464, "grad_norm": 2.1210896968841553, "learning_rate": 4.917800000000001e-06, "loss": 0.0372, "step": 25412 }, { "epoch": 1.3548541877698992, "grad_norm": 2.648959159851074, "learning_rate": 4.9176e-06, "loss": 0.032, "step": 25413 }, { "epoch": 1.3549075011995522, "grad_norm": 1.4295203685760498, "learning_rate": 4.9174000000000006e-06, "loss": 0.0176, "step": 25414 }, { "epoch": 1.3549608146292051, "grad_norm": 1.9077131748199463, "learning_rate": 4.9172e-06, "loss": 0.0383, "step": 25415 }, { "epoch": 1.355014128058858, "grad_norm": 2.1106114387512207, "learning_rate": 4.9170000000000005e-06, "loss": 0.0243, "step": 25416 }, { "epoch": 1.3550674414885109, "grad_norm": 0.7413785457611084, "learning_rate": 4.9168e-06, "loss": 0.0122, "step": 25417 }, { "epoch": 1.3551207549181639, "grad_norm": 1.102094054222107, "learning_rate": 4.9166000000000005e-06, "loss": 0.0223, "step": 25418 }, { "epoch": 1.3551740683478168, "grad_norm": 1.2438957691192627, "learning_rate": 4.916400000000001e-06, "loss": 0.0123, "step": 25419 }, { "epoch": 1.3552273817774698, "grad_norm": 0.4480306804180145, "learning_rate": 4.9162000000000005e-06, "loss": 0.0044, "step": 25420 }, { "epoch": 1.3552806952071226, "grad_norm": 3.168057918548584, "learning_rate": 4.916e-06, "loss": 0.0383, "step": 25421 }, { "epoch": 1.3553340086367756, "grad_norm": 3.104992151260376, "learning_rate": 4.9158000000000004e-06, "loss": 0.0406, "step": 25422 }, { "epoch": 1.3553873220664285, "grad_norm": 0.31211069226264954, "learning_rate": 4.9156e-06, "loss": 0.0035, "step": 25423 }, { "epoch": 1.3554406354960815, "grad_norm": 0.8312842845916748, "learning_rate": 4.9154e-06, "loss": 0.0175, "step": 25424 }, { "epoch": 1.3554939489257345, "grad_norm": 0.6902141571044922, "learning_rate": 4.915200000000001e-06, "loss": 0.02, "step": 25425 }, { "epoch": 1.3555472623553873, "grad_norm": 1.401212453842163, "learning_rate": 4.915e-06, "loss": 0.0178, "step": 25426 }, { "epoch": 1.3556005757850402, "grad_norm": 1.3887308835983276, "learning_rate": 4.9148e-06, "loss": 0.016, "step": 25427 }, { "epoch": 1.3556538892146932, "grad_norm": 2.009603977203369, "learning_rate": 4.9146e-06, "loss": 0.0191, "step": 25428 }, { "epoch": 1.3557072026443462, "grad_norm": 2.2806265354156494, "learning_rate": 4.9144e-06, "loss": 0.0283, "step": 25429 }, { "epoch": 1.3557605160739992, "grad_norm": 0.8624289035797119, "learning_rate": 4.9142e-06, "loss": 0.0139, "step": 25430 }, { "epoch": 1.355813829503652, "grad_norm": 0.9042671322822571, "learning_rate": 4.914000000000001e-06, "loss": 0.0057, "step": 25431 }, { "epoch": 1.355867142933305, "grad_norm": 1.1579029560089111, "learning_rate": 4.9138e-06, "loss": 0.0247, "step": 25432 }, { "epoch": 1.3559204563629579, "grad_norm": 1.0106145143508911, "learning_rate": 4.913600000000001e-06, "loss": 0.0073, "step": 25433 }, { "epoch": 1.3559737697926106, "grad_norm": 1.2273962497711182, "learning_rate": 4.9134e-06, "loss": 0.013, "step": 25434 }, { "epoch": 1.3560270832222636, "grad_norm": 1.7503349781036377, "learning_rate": 4.9132e-06, "loss": 0.0193, "step": 25435 }, { "epoch": 1.3560803966519166, "grad_norm": 1.7643400430679321, "learning_rate": 4.913e-06, "loss": 0.017, "step": 25436 }, { "epoch": 1.3561337100815696, "grad_norm": 0.9399375915527344, "learning_rate": 4.912800000000001e-06, "loss": 0.0164, "step": 25437 }, { "epoch": 1.3561870235112226, "grad_norm": 1.908732533454895, "learning_rate": 4.9126e-06, "loss": 0.0236, "step": 25438 }, { "epoch": 1.3562403369408753, "grad_norm": 1.3776063919067383, "learning_rate": 4.912400000000001e-06, "loss": 0.0218, "step": 25439 }, { "epoch": 1.3562936503705283, "grad_norm": 1.3359105587005615, "learning_rate": 4.9122e-06, "loss": 0.009, "step": 25440 }, { "epoch": 1.3563469638001813, "grad_norm": 2.3204076290130615, "learning_rate": 4.9120000000000006e-06, "loss": 0.0227, "step": 25441 }, { "epoch": 1.3564002772298343, "grad_norm": 1.3967020511627197, "learning_rate": 4.9118e-06, "loss": 0.019, "step": 25442 }, { "epoch": 1.3564535906594872, "grad_norm": 1.7871726751327515, "learning_rate": 4.9116000000000005e-06, "loss": 0.0187, "step": 25443 }, { "epoch": 1.35650690408914, "grad_norm": 1.8678576946258545, "learning_rate": 4.911400000000001e-06, "loss": 0.0218, "step": 25444 }, { "epoch": 1.356560217518793, "grad_norm": 1.4438679218292236, "learning_rate": 4.9112000000000005e-06, "loss": 0.0097, "step": 25445 }, { "epoch": 1.356613530948446, "grad_norm": 1.500686764717102, "learning_rate": 4.911e-06, "loss": 0.0094, "step": 25446 }, { "epoch": 1.356666844378099, "grad_norm": 1.092023491859436, "learning_rate": 4.9108000000000005e-06, "loss": 0.0094, "step": 25447 }, { "epoch": 1.356720157807752, "grad_norm": 1.5574805736541748, "learning_rate": 4.9106e-06, "loss": 0.0181, "step": 25448 }, { "epoch": 1.3567734712374047, "grad_norm": 1.3149268627166748, "learning_rate": 4.9104000000000005e-06, "loss": 0.0124, "step": 25449 }, { "epoch": 1.3568267846670576, "grad_norm": 0.8696823120117188, "learning_rate": 4.910200000000001e-06, "loss": 0.0058, "step": 25450 }, { "epoch": 1.3568800980967106, "grad_norm": 2.2967638969421387, "learning_rate": 4.9100000000000004e-06, "loss": 0.0556, "step": 25451 }, { "epoch": 1.3569334115263634, "grad_norm": 1.9892349243164062, "learning_rate": 4.909800000000001e-06, "loss": 0.0284, "step": 25452 }, { "epoch": 1.3569867249560164, "grad_norm": 1.2667721509933472, "learning_rate": 4.9096e-06, "loss": 0.0244, "step": 25453 }, { "epoch": 1.3570400383856693, "grad_norm": 1.4763665199279785, "learning_rate": 4.9094e-06, "loss": 0.0172, "step": 25454 }, { "epoch": 1.3570933518153223, "grad_norm": 1.3457242250442505, "learning_rate": 4.9092e-06, "loss": 0.0113, "step": 25455 }, { "epoch": 1.3571466652449753, "grad_norm": 1.5948599576950073, "learning_rate": 4.909000000000001e-06, "loss": 0.0262, "step": 25456 }, { "epoch": 1.357199978674628, "grad_norm": 0.324726939201355, "learning_rate": 4.9088e-06, "loss": 0.0029, "step": 25457 }, { "epoch": 1.357253292104281, "grad_norm": 1.7664018869400024, "learning_rate": 4.908600000000001e-06, "loss": 0.0282, "step": 25458 }, { "epoch": 1.357306605533934, "grad_norm": 0.6665123105049133, "learning_rate": 4.9084e-06, "loss": 0.0195, "step": 25459 }, { "epoch": 1.357359918963587, "grad_norm": 0.5628122687339783, "learning_rate": 4.9082e-06, "loss": 0.004, "step": 25460 }, { "epoch": 1.35741323239324, "grad_norm": 1.066365122795105, "learning_rate": 4.908e-06, "loss": 0.0115, "step": 25461 }, { "epoch": 1.3574665458228927, "grad_norm": 5.825295925140381, "learning_rate": 4.907800000000001e-06, "loss": 0.0201, "step": 25462 }, { "epoch": 1.3575198592525457, "grad_norm": 2.404208183288574, "learning_rate": 4.9076e-06, "loss": 0.0088, "step": 25463 }, { "epoch": 1.3575731726821987, "grad_norm": 1.6634743213653564, "learning_rate": 4.907400000000001e-06, "loss": 0.017, "step": 25464 }, { "epoch": 1.3576264861118514, "grad_norm": 1.118307113647461, "learning_rate": 4.9072e-06, "loss": 0.01, "step": 25465 }, { "epoch": 1.3576797995415046, "grad_norm": 1.6679596900939941, "learning_rate": 4.907000000000001e-06, "loss": 0.0126, "step": 25466 }, { "epoch": 1.3577331129711574, "grad_norm": 3.5765419006347656, "learning_rate": 4.9068e-06, "loss": 0.0877, "step": 25467 }, { "epoch": 1.3577864264008104, "grad_norm": 2.4097280502319336, "learning_rate": 4.906600000000001e-06, "loss": 0.0282, "step": 25468 }, { "epoch": 1.3578397398304634, "grad_norm": 0.2552209794521332, "learning_rate": 4.906400000000001e-06, "loss": 0.0026, "step": 25469 }, { "epoch": 1.3578930532601161, "grad_norm": 1.8310937881469727, "learning_rate": 4.9062000000000006e-06, "loss": 0.0301, "step": 25470 }, { "epoch": 1.357946366689769, "grad_norm": 2.38580584526062, "learning_rate": 4.906e-06, "loss": 0.0239, "step": 25471 }, { "epoch": 1.357999680119422, "grad_norm": 4.9954071044921875, "learning_rate": 4.9058000000000005e-06, "loss": 0.0127, "step": 25472 }, { "epoch": 1.358052993549075, "grad_norm": 2.173612356185913, "learning_rate": 4.9056e-06, "loss": 0.0213, "step": 25473 }, { "epoch": 1.358106306978728, "grad_norm": 4.300520896911621, "learning_rate": 4.9054000000000005e-06, "loss": 0.0633, "step": 25474 }, { "epoch": 1.3581596204083808, "grad_norm": 2.1061296463012695, "learning_rate": 4.905200000000001e-06, "loss": 0.0306, "step": 25475 }, { "epoch": 1.3582129338380338, "grad_norm": 2.4133076667785645, "learning_rate": 4.9050000000000005e-06, "loss": 0.0387, "step": 25476 }, { "epoch": 1.3582662472676867, "grad_norm": 2.056619644165039, "learning_rate": 4.904800000000001e-06, "loss": 0.0285, "step": 25477 }, { "epoch": 1.3583195606973397, "grad_norm": 1.7784637212753296, "learning_rate": 4.9046000000000004e-06, "loss": 0.0109, "step": 25478 }, { "epoch": 1.3583728741269927, "grad_norm": 1.6470669507980347, "learning_rate": 4.9044e-06, "loss": 0.0075, "step": 25479 }, { "epoch": 1.3584261875566455, "grad_norm": 2.3379368782043457, "learning_rate": 4.9042e-06, "loss": 0.0235, "step": 25480 }, { "epoch": 1.3584795009862984, "grad_norm": 0.5313594937324524, "learning_rate": 4.904000000000001e-06, "loss": 0.0059, "step": 25481 }, { "epoch": 1.3585328144159514, "grad_norm": 2.3454065322875977, "learning_rate": 4.9038e-06, "loss": 0.044, "step": 25482 }, { "epoch": 1.3585861278456042, "grad_norm": 2.376232624053955, "learning_rate": 4.903600000000001e-06, "loss": 0.0571, "step": 25483 }, { "epoch": 1.3586394412752574, "grad_norm": 1.9781736135482788, "learning_rate": 4.9034e-06, "loss": 0.0301, "step": 25484 }, { "epoch": 1.3586927547049101, "grad_norm": 3.270998477935791, "learning_rate": 4.9032e-06, "loss": 0.0399, "step": 25485 }, { "epoch": 1.3587460681345631, "grad_norm": 3.121335506439209, "learning_rate": 4.903e-06, "loss": 0.0375, "step": 25486 }, { "epoch": 1.358799381564216, "grad_norm": 1.4300894737243652, "learning_rate": 4.902800000000001e-06, "loss": 0.0154, "step": 25487 }, { "epoch": 1.3588526949938688, "grad_norm": 1.972253441810608, "learning_rate": 4.9026e-06, "loss": 0.0405, "step": 25488 }, { "epoch": 1.3589060084235218, "grad_norm": 1.6802891492843628, "learning_rate": 4.902400000000001e-06, "loss": 0.0188, "step": 25489 }, { "epoch": 1.3589593218531748, "grad_norm": 3.7647182941436768, "learning_rate": 4.9022e-06, "loss": 0.0266, "step": 25490 }, { "epoch": 1.3590126352828278, "grad_norm": 2.464078903198242, "learning_rate": 4.902000000000001e-06, "loss": 0.0224, "step": 25491 }, { "epoch": 1.3590659487124808, "grad_norm": 2.5143678188323975, "learning_rate": 4.9018e-06, "loss": 0.0358, "step": 25492 }, { "epoch": 1.3591192621421335, "grad_norm": 1.1838529109954834, "learning_rate": 4.901600000000001e-06, "loss": 0.0099, "step": 25493 }, { "epoch": 1.3591725755717865, "grad_norm": 1.3581302165985107, "learning_rate": 4.9014e-06, "loss": 0.0088, "step": 25494 }, { "epoch": 1.3592258890014395, "grad_norm": 1.2417513132095337, "learning_rate": 4.901200000000001e-06, "loss": 0.0114, "step": 25495 }, { "epoch": 1.3592792024310925, "grad_norm": 2.925367832183838, "learning_rate": 4.901e-06, "loss": 0.0361, "step": 25496 }, { "epoch": 1.3593325158607454, "grad_norm": 1.8053336143493652, "learning_rate": 4.900800000000001e-06, "loss": 0.0116, "step": 25497 }, { "epoch": 1.3593858292903982, "grad_norm": 1.9688827991485596, "learning_rate": 4.9006e-06, "loss": 0.014, "step": 25498 }, { "epoch": 1.3594391427200512, "grad_norm": 2.981358051300049, "learning_rate": 4.9004000000000005e-06, "loss": 0.0426, "step": 25499 }, { "epoch": 1.3594924561497042, "grad_norm": 0.8187522888183594, "learning_rate": 4.9002e-06, "loss": 0.0095, "step": 25500 }, { "epoch": 1.359545769579357, "grad_norm": 1.295145034790039, "learning_rate": 4.9000000000000005e-06, "loss": 0.0253, "step": 25501 }, { "epoch": 1.35959908300901, "grad_norm": 0.8070035576820374, "learning_rate": 4.899800000000001e-06, "loss": 0.0069, "step": 25502 }, { "epoch": 1.3596523964386629, "grad_norm": 1.4697437286376953, "learning_rate": 4.8996000000000005e-06, "loss": 0.0237, "step": 25503 }, { "epoch": 1.3597057098683158, "grad_norm": 2.0457370281219482, "learning_rate": 4.8994e-06, "loss": 0.0249, "step": 25504 }, { "epoch": 1.3597590232979688, "grad_norm": 0.41715502738952637, "learning_rate": 4.8992000000000005e-06, "loss": 0.0055, "step": 25505 }, { "epoch": 1.3598123367276216, "grad_norm": 1.9859915971755981, "learning_rate": 4.899e-06, "loss": 0.0169, "step": 25506 }, { "epoch": 1.3598656501572746, "grad_norm": 1.5565205812454224, "learning_rate": 4.8988000000000004e-06, "loss": 0.0344, "step": 25507 }, { "epoch": 1.3599189635869275, "grad_norm": 1.959517002105713, "learning_rate": 4.898600000000001e-06, "loss": 0.0456, "step": 25508 }, { "epoch": 1.3599722770165805, "grad_norm": 2.8250749111175537, "learning_rate": 4.8984e-06, "loss": 0.0316, "step": 25509 }, { "epoch": 1.3600255904462335, "grad_norm": 1.87212336063385, "learning_rate": 4.8982e-06, "loss": 0.0265, "step": 25510 }, { "epoch": 1.3600789038758863, "grad_norm": 2.3303728103637695, "learning_rate": 4.898e-06, "loss": 0.0216, "step": 25511 }, { "epoch": 1.3601322173055392, "grad_norm": 3.9628734588623047, "learning_rate": 4.8978e-06, "loss": 0.04, "step": 25512 }, { "epoch": 1.3601855307351922, "grad_norm": 1.2309684753417969, "learning_rate": 4.8976e-06, "loss": 0.0231, "step": 25513 }, { "epoch": 1.3602388441648452, "grad_norm": 1.6396763324737549, "learning_rate": 4.897400000000001e-06, "loss": 0.0215, "step": 25514 }, { "epoch": 1.3602921575944982, "grad_norm": 0.6896354556083679, "learning_rate": 4.8972e-06, "loss": 0.0076, "step": 25515 }, { "epoch": 1.360345471024151, "grad_norm": 1.7834999561309814, "learning_rate": 4.897000000000001e-06, "loss": 0.04, "step": 25516 }, { "epoch": 1.360398784453804, "grad_norm": 0.4334390163421631, "learning_rate": 4.8968e-06, "loss": 0.0036, "step": 25517 }, { "epoch": 1.360452097883457, "grad_norm": 1.0628677606582642, "learning_rate": 4.8966e-06, "loss": 0.0145, "step": 25518 }, { "epoch": 1.3605054113131096, "grad_norm": 1.7909828424453735, "learning_rate": 4.8964e-06, "loss": 0.041, "step": 25519 }, { "epoch": 1.3605587247427626, "grad_norm": 2.57901668548584, "learning_rate": 4.896200000000001e-06, "loss": 0.0172, "step": 25520 }, { "epoch": 1.3606120381724156, "grad_norm": 1.2605990171432495, "learning_rate": 4.896e-06, "loss": 0.027, "step": 25521 }, { "epoch": 1.3606653516020686, "grad_norm": 1.2698436975479126, "learning_rate": 4.895800000000001e-06, "loss": 0.0209, "step": 25522 }, { "epoch": 1.3607186650317216, "grad_norm": 1.3656346797943115, "learning_rate": 4.8956e-06, "loss": 0.019, "step": 25523 }, { "epoch": 1.3607719784613743, "grad_norm": 1.4397650957107544, "learning_rate": 4.8954e-06, "loss": 0.0216, "step": 25524 }, { "epoch": 1.3608252918910273, "grad_norm": 2.5236973762512207, "learning_rate": 4.8952e-06, "loss": 0.0235, "step": 25525 }, { "epoch": 1.3608786053206803, "grad_norm": 1.8257625102996826, "learning_rate": 4.8950000000000006e-06, "loss": 0.0254, "step": 25526 }, { "epoch": 1.3609319187503333, "grad_norm": 1.4776594638824463, "learning_rate": 4.894800000000001e-06, "loss": 0.0194, "step": 25527 }, { "epoch": 1.3609852321799862, "grad_norm": 1.449608564376831, "learning_rate": 4.8946000000000005e-06, "loss": 0.0059, "step": 25528 }, { "epoch": 1.361038545609639, "grad_norm": 1.7080235481262207, "learning_rate": 4.8944e-06, "loss": 0.0415, "step": 25529 }, { "epoch": 1.361091859039292, "grad_norm": 2.9567947387695312, "learning_rate": 4.8942000000000005e-06, "loss": 0.0419, "step": 25530 }, { "epoch": 1.361145172468945, "grad_norm": 1.1969878673553467, "learning_rate": 4.894e-06, "loss": 0.013, "step": 25531 }, { "epoch": 1.361198485898598, "grad_norm": 1.0489048957824707, "learning_rate": 4.8938000000000005e-06, "loss": 0.0103, "step": 25532 }, { "epoch": 1.361251799328251, "grad_norm": 1.1839324235916138, "learning_rate": 4.893600000000001e-06, "loss": 0.0175, "step": 25533 }, { "epoch": 1.3613051127579037, "grad_norm": 2.2088663578033447, "learning_rate": 4.8934000000000004e-06, "loss": 0.0378, "step": 25534 }, { "epoch": 1.3613584261875566, "grad_norm": 1.34221613407135, "learning_rate": 4.8932e-06, "loss": 0.0148, "step": 25535 }, { "epoch": 1.3614117396172096, "grad_norm": 1.7808786630630493, "learning_rate": 4.893e-06, "loss": 0.0211, "step": 25536 }, { "epoch": 1.3614650530468624, "grad_norm": 3.08508038520813, "learning_rate": 4.8928e-06, "loss": 0.0215, "step": 25537 }, { "epoch": 1.3615183664765154, "grad_norm": 2.7062156200408936, "learning_rate": 4.8926e-06, "loss": 0.027, "step": 25538 }, { "epoch": 1.3615716799061683, "grad_norm": 0.40839678049087524, "learning_rate": 4.892400000000001e-06, "loss": 0.0044, "step": 25539 }, { "epoch": 1.3616249933358213, "grad_norm": 1.048867106437683, "learning_rate": 4.8922e-06, "loss": 0.0144, "step": 25540 }, { "epoch": 1.3616783067654743, "grad_norm": 2.9924087524414062, "learning_rate": 4.892000000000001e-06, "loss": 0.024, "step": 25541 }, { "epoch": 1.361731620195127, "grad_norm": 2.6951069831848145, "learning_rate": 4.8918e-06, "loss": 0.0617, "step": 25542 }, { "epoch": 1.36178493362478, "grad_norm": 0.8503648638725281, "learning_rate": 4.8916e-06, "loss": 0.0069, "step": 25543 }, { "epoch": 1.361838247054433, "grad_norm": 5.228115081787109, "learning_rate": 4.8914e-06, "loss": 0.0256, "step": 25544 }, { "epoch": 1.361891560484086, "grad_norm": 1.8300261497497559, "learning_rate": 4.891200000000001e-06, "loss": 0.033, "step": 25545 }, { "epoch": 1.361944873913739, "grad_norm": 1.6367943286895752, "learning_rate": 4.891e-06, "loss": 0.023, "step": 25546 }, { "epoch": 1.3619981873433917, "grad_norm": 1.878308653831482, "learning_rate": 4.890800000000001e-06, "loss": 0.0165, "step": 25547 }, { "epoch": 1.3620515007730447, "grad_norm": 1.0827912092208862, "learning_rate": 4.8906e-06, "loss": 0.0104, "step": 25548 }, { "epoch": 1.3621048142026977, "grad_norm": 0.461077481508255, "learning_rate": 4.8904e-06, "loss": 0.0031, "step": 25549 }, { "epoch": 1.3621581276323507, "grad_norm": 1.1488614082336426, "learning_rate": 4.8902e-06, "loss": 0.0204, "step": 25550 }, { "epoch": 1.3622114410620036, "grad_norm": 2.802793502807617, "learning_rate": 4.890000000000001e-06, "loss": 0.0352, "step": 25551 }, { "epoch": 1.3622647544916564, "grad_norm": 2.232269048690796, "learning_rate": 4.889800000000001e-06, "loss": 0.0205, "step": 25552 }, { "epoch": 1.3623180679213094, "grad_norm": 1.8400957584381104, "learning_rate": 4.889600000000001e-06, "loss": 0.0231, "step": 25553 }, { "epoch": 1.3623713813509624, "grad_norm": 3.034552812576294, "learning_rate": 4.8894e-06, "loss": 0.0638, "step": 25554 }, { "epoch": 1.3624246947806151, "grad_norm": 0.9495604634284973, "learning_rate": 4.8892000000000006e-06, "loss": 0.0188, "step": 25555 }, { "epoch": 1.362478008210268, "grad_norm": 0.8635851740837097, "learning_rate": 4.889e-06, "loss": 0.0069, "step": 25556 }, { "epoch": 1.362531321639921, "grad_norm": 2.9179606437683105, "learning_rate": 4.8888000000000005e-06, "loss": 0.015, "step": 25557 }, { "epoch": 1.362584635069574, "grad_norm": 2.308018207550049, "learning_rate": 4.888600000000001e-06, "loss": 0.0368, "step": 25558 }, { "epoch": 1.362637948499227, "grad_norm": 1.7406020164489746, "learning_rate": 4.8884000000000005e-06, "loss": 0.0185, "step": 25559 }, { "epoch": 1.3626912619288798, "grad_norm": 1.3694819211959839, "learning_rate": 4.8882e-06, "loss": 0.021, "step": 25560 }, { "epoch": 1.3627445753585328, "grad_norm": 1.7422391176223755, "learning_rate": 4.8880000000000005e-06, "loss": 0.0269, "step": 25561 }, { "epoch": 1.3627978887881858, "grad_norm": 1.7752294540405273, "learning_rate": 4.8878e-06, "loss": 0.0298, "step": 25562 }, { "epoch": 1.3628512022178387, "grad_norm": 1.9708974361419678, "learning_rate": 4.8876000000000004e-06, "loss": 0.0325, "step": 25563 }, { "epoch": 1.3629045156474917, "grad_norm": 2.1079888343811035, "learning_rate": 4.887400000000001e-06, "loss": 0.0419, "step": 25564 }, { "epoch": 1.3629578290771445, "grad_norm": 1.3126072883605957, "learning_rate": 4.8872e-06, "loss": 0.0323, "step": 25565 }, { "epoch": 1.3630111425067974, "grad_norm": 1.9287692308425903, "learning_rate": 4.887000000000001e-06, "loss": 0.0378, "step": 25566 }, { "epoch": 1.3630644559364504, "grad_norm": 2.9932639598846436, "learning_rate": 4.8868e-06, "loss": 0.0335, "step": 25567 }, { "epoch": 1.3631177693661034, "grad_norm": 3.1455740928649902, "learning_rate": 4.8866e-06, "loss": 0.0392, "step": 25568 }, { "epoch": 1.3631710827957564, "grad_norm": 2.5186636447906494, "learning_rate": 4.8864e-06, "loss": 0.0393, "step": 25569 }, { "epoch": 1.3632243962254091, "grad_norm": 1.4640214443206787, "learning_rate": 4.886200000000001e-06, "loss": 0.007, "step": 25570 }, { "epoch": 1.3632777096550621, "grad_norm": 3.725708246231079, "learning_rate": 4.886e-06, "loss": 0.0572, "step": 25571 }, { "epoch": 1.363331023084715, "grad_norm": 2.0496134757995605, "learning_rate": 4.885800000000001e-06, "loss": 0.0282, "step": 25572 }, { "epoch": 1.3633843365143679, "grad_norm": 2.5841052532196045, "learning_rate": 4.8856e-06, "loss": 0.0583, "step": 25573 }, { "epoch": 1.3634376499440208, "grad_norm": 0.3707776665687561, "learning_rate": 4.8854e-06, "loss": 0.0049, "step": 25574 }, { "epoch": 1.3634909633736738, "grad_norm": 1.533501148223877, "learning_rate": 4.8852e-06, "loss": 0.0201, "step": 25575 }, { "epoch": 1.3635442768033268, "grad_norm": 1.2141801118850708, "learning_rate": 4.885000000000001e-06, "loss": 0.0272, "step": 25576 }, { "epoch": 1.3635975902329798, "grad_norm": 2.5223495960235596, "learning_rate": 4.884800000000001e-06, "loss": 0.0126, "step": 25577 }, { "epoch": 1.3636509036626325, "grad_norm": 1.1865448951721191, "learning_rate": 4.884600000000001e-06, "loss": 0.0086, "step": 25578 }, { "epoch": 1.3637042170922855, "grad_norm": 1.8762810230255127, "learning_rate": 4.8844e-06, "loss": 0.0443, "step": 25579 }, { "epoch": 1.3637575305219385, "grad_norm": 1.6733014583587646, "learning_rate": 4.884200000000001e-06, "loss": 0.0298, "step": 25580 }, { "epoch": 1.3638108439515915, "grad_norm": 1.052233338356018, "learning_rate": 4.884e-06, "loss": 0.0095, "step": 25581 }, { "epoch": 1.3638641573812444, "grad_norm": 1.0659762620925903, "learning_rate": 4.8838000000000006e-06, "loss": 0.0119, "step": 25582 }, { "epoch": 1.3639174708108972, "grad_norm": 2.009369373321533, "learning_rate": 4.883600000000001e-06, "loss": 0.0155, "step": 25583 }, { "epoch": 1.3639707842405502, "grad_norm": 1.39436674118042, "learning_rate": 4.8834000000000005e-06, "loss": 0.015, "step": 25584 }, { "epoch": 1.3640240976702032, "grad_norm": 2.2302653789520264, "learning_rate": 4.8832e-06, "loss": 0.0375, "step": 25585 }, { "epoch": 1.364077411099856, "grad_norm": 1.1746858358383179, "learning_rate": 4.8830000000000005e-06, "loss": 0.0226, "step": 25586 }, { "epoch": 1.3641307245295091, "grad_norm": 1.4537246227264404, "learning_rate": 4.8828e-06, "loss": 0.0138, "step": 25587 }, { "epoch": 1.3641840379591619, "grad_norm": 0.9777140021324158, "learning_rate": 4.8826000000000005e-06, "loss": 0.0083, "step": 25588 }, { "epoch": 1.3642373513888149, "grad_norm": 1.639195203781128, "learning_rate": 4.882400000000001e-06, "loss": 0.0377, "step": 25589 }, { "epoch": 1.3642906648184678, "grad_norm": 2.2997028827667236, "learning_rate": 4.8822000000000005e-06, "loss": 0.0258, "step": 25590 }, { "epoch": 1.3643439782481206, "grad_norm": 1.866555094718933, "learning_rate": 4.882000000000001e-06, "loss": 0.051, "step": 25591 }, { "epoch": 1.3643972916777736, "grad_norm": 1.7307828664779663, "learning_rate": 4.8818e-06, "loss": 0.0271, "step": 25592 }, { "epoch": 1.3644506051074266, "grad_norm": 1.3229609727859497, "learning_rate": 4.8816e-06, "loss": 0.0118, "step": 25593 }, { "epoch": 1.3645039185370795, "grad_norm": 2.381115198135376, "learning_rate": 4.8814e-06, "loss": 0.0174, "step": 25594 }, { "epoch": 1.3645572319667325, "grad_norm": 2.2083323001861572, "learning_rate": 4.881200000000001e-06, "loss": 0.0502, "step": 25595 }, { "epoch": 1.3646105453963853, "grad_norm": 1.0830878019332886, "learning_rate": 4.881e-06, "loss": 0.0288, "step": 25596 }, { "epoch": 1.3646638588260382, "grad_norm": 2.8375296592712402, "learning_rate": 4.880800000000001e-06, "loss": 0.0615, "step": 25597 }, { "epoch": 1.3647171722556912, "grad_norm": 1.0613871812820435, "learning_rate": 4.8806e-06, "loss": 0.0108, "step": 25598 }, { "epoch": 1.3647704856853442, "grad_norm": 1.0720807313919067, "learning_rate": 4.8804e-06, "loss": 0.0145, "step": 25599 }, { "epoch": 1.3648237991149972, "grad_norm": 1.5697656869888306, "learning_rate": 4.8802e-06, "loss": 0.0194, "step": 25600 }, { "epoch": 1.36487711254465, "grad_norm": 1.8495169878005981, "learning_rate": 4.880000000000001e-06, "loss": 0.015, "step": 25601 }, { "epoch": 1.364930425974303, "grad_norm": 1.599792718887329, "learning_rate": 4.8798e-06, "loss": 0.0182, "step": 25602 }, { "epoch": 1.364983739403956, "grad_norm": 1.8803906440734863, "learning_rate": 4.879600000000001e-06, "loss": 0.0222, "step": 25603 }, { "epoch": 1.3650370528336087, "grad_norm": 0.8797175884246826, "learning_rate": 4.8794e-06, "loss": 0.0092, "step": 25604 }, { "epoch": 1.3650903662632616, "grad_norm": 3.16379976272583, "learning_rate": 4.879200000000001e-06, "loss": 0.0424, "step": 25605 }, { "epoch": 1.3651436796929146, "grad_norm": 1.004875659942627, "learning_rate": 4.879e-06, "loss": 0.0131, "step": 25606 }, { "epoch": 1.3651969931225676, "grad_norm": 1.4482734203338623, "learning_rate": 4.878800000000001e-06, "loss": 0.0234, "step": 25607 }, { "epoch": 1.3652503065522206, "grad_norm": 1.3391789197921753, "learning_rate": 4.8786e-06, "loss": 0.0223, "step": 25608 }, { "epoch": 1.3653036199818733, "grad_norm": 0.647639274597168, "learning_rate": 4.878400000000001e-06, "loss": 0.004, "step": 25609 }, { "epoch": 1.3653569334115263, "grad_norm": 1.6777245998382568, "learning_rate": 4.8782e-06, "loss": 0.0562, "step": 25610 }, { "epoch": 1.3654102468411793, "grad_norm": 1.9337177276611328, "learning_rate": 4.8780000000000006e-06, "loss": 0.049, "step": 25611 }, { "epoch": 1.3654635602708323, "grad_norm": 1.5876238346099854, "learning_rate": 4.8778e-06, "loss": 0.0187, "step": 25612 }, { "epoch": 1.3655168737004852, "grad_norm": 1.9922668933868408, "learning_rate": 4.8776000000000005e-06, "loss": 0.0205, "step": 25613 }, { "epoch": 1.365570187130138, "grad_norm": 1.891344428062439, "learning_rate": 4.8774e-06, "loss": 0.0329, "step": 25614 }, { "epoch": 1.365623500559791, "grad_norm": 1.9957343339920044, "learning_rate": 4.8772000000000005e-06, "loss": 0.041, "step": 25615 }, { "epoch": 1.365676813989444, "grad_norm": 1.106867790222168, "learning_rate": 4.877000000000001e-06, "loss": 0.0188, "step": 25616 }, { "epoch": 1.365730127419097, "grad_norm": 1.2583892345428467, "learning_rate": 4.8768000000000005e-06, "loss": 0.0138, "step": 25617 }, { "epoch": 1.36578344084875, "grad_norm": 0.7201752662658691, "learning_rate": 4.8766e-06, "loss": 0.0056, "step": 25618 }, { "epoch": 1.3658367542784027, "grad_norm": 1.5438621044158936, "learning_rate": 4.8764000000000004e-06, "loss": 0.0258, "step": 25619 }, { "epoch": 1.3658900677080557, "grad_norm": 1.8842324018478394, "learning_rate": 4.8762e-06, "loss": 0.022, "step": 25620 }, { "epoch": 1.3659433811377086, "grad_norm": 2.0196454524993896, "learning_rate": 4.876e-06, "loss": 0.0275, "step": 25621 }, { "epoch": 1.3659966945673614, "grad_norm": 3.194495916366577, "learning_rate": 4.875800000000001e-06, "loss": 0.0422, "step": 25622 }, { "epoch": 1.3660500079970144, "grad_norm": 1.248171091079712, "learning_rate": 4.8756e-06, "loss": 0.0144, "step": 25623 }, { "epoch": 1.3661033214266673, "grad_norm": 1.8415279388427734, "learning_rate": 4.8754e-06, "loss": 0.0206, "step": 25624 }, { "epoch": 1.3661566348563203, "grad_norm": 1.8590385913848877, "learning_rate": 4.8752e-06, "loss": 0.0146, "step": 25625 }, { "epoch": 1.3662099482859733, "grad_norm": 1.297311782836914, "learning_rate": 4.875e-06, "loss": 0.0205, "step": 25626 }, { "epoch": 1.366263261715626, "grad_norm": 1.6932861804962158, "learning_rate": 4.8748e-06, "loss": 0.0117, "step": 25627 }, { "epoch": 1.366316575145279, "grad_norm": 1.3981291055679321, "learning_rate": 4.874600000000001e-06, "loss": 0.0098, "step": 25628 }, { "epoch": 1.366369888574932, "grad_norm": 6.371492385864258, "learning_rate": 4.8744e-06, "loss": 0.0226, "step": 25629 }, { "epoch": 1.366423202004585, "grad_norm": 2.4922657012939453, "learning_rate": 4.874200000000001e-06, "loss": 0.03, "step": 25630 }, { "epoch": 1.366476515434238, "grad_norm": 2.309004545211792, "learning_rate": 4.874e-06, "loss": 0.0136, "step": 25631 }, { "epoch": 1.3665298288638907, "grad_norm": 3.0438201427459717, "learning_rate": 4.8738e-06, "loss": 0.0446, "step": 25632 }, { "epoch": 1.3665831422935437, "grad_norm": 1.5705803632736206, "learning_rate": 4.8736e-06, "loss": 0.0153, "step": 25633 }, { "epoch": 1.3666364557231967, "grad_norm": 2.647893190383911, "learning_rate": 4.873400000000001e-06, "loss": 0.0356, "step": 25634 }, { "epoch": 1.3666897691528497, "grad_norm": 2.7331929206848145, "learning_rate": 4.8732e-06, "loss": 0.0491, "step": 25635 }, { "epoch": 1.3667430825825027, "grad_norm": 1.2272473573684692, "learning_rate": 4.873000000000001e-06, "loss": 0.0098, "step": 25636 }, { "epoch": 1.3667963960121554, "grad_norm": 0.867246687412262, "learning_rate": 4.8728e-06, "loss": 0.0175, "step": 25637 }, { "epoch": 1.3668497094418084, "grad_norm": 3.2394943237304688, "learning_rate": 4.872600000000001e-06, "loss": 0.0203, "step": 25638 }, { "epoch": 1.3669030228714614, "grad_norm": 1.4292688369750977, "learning_rate": 4.8724e-06, "loss": 0.0088, "step": 25639 }, { "epoch": 1.3669563363011141, "grad_norm": 3.687992811203003, "learning_rate": 4.8722000000000005e-06, "loss": 0.0234, "step": 25640 }, { "epoch": 1.367009649730767, "grad_norm": 1.6687233448028564, "learning_rate": 4.872000000000001e-06, "loss": 0.0124, "step": 25641 }, { "epoch": 1.36706296316042, "grad_norm": 1.3364524841308594, "learning_rate": 4.8718000000000005e-06, "loss": 0.0241, "step": 25642 }, { "epoch": 1.367116276590073, "grad_norm": 1.1248984336853027, "learning_rate": 4.8716e-06, "loss": 0.0135, "step": 25643 }, { "epoch": 1.367169590019726, "grad_norm": 1.2795157432556152, "learning_rate": 4.8714000000000005e-06, "loss": 0.0084, "step": 25644 }, { "epoch": 1.3672229034493788, "grad_norm": 2.398571491241455, "learning_rate": 4.8712e-06, "loss": 0.0395, "step": 25645 }, { "epoch": 1.3672762168790318, "grad_norm": 2.5333452224731445, "learning_rate": 4.8710000000000005e-06, "loss": 0.0241, "step": 25646 }, { "epoch": 1.3673295303086848, "grad_norm": 1.1562165021896362, "learning_rate": 4.870800000000001e-06, "loss": 0.014, "step": 25647 }, { "epoch": 1.3673828437383377, "grad_norm": 2.640232801437378, "learning_rate": 4.8706000000000004e-06, "loss": 0.0374, "step": 25648 }, { "epoch": 1.3674361571679907, "grad_norm": 1.7571772336959839, "learning_rate": 4.8704e-06, "loss": 0.0268, "step": 25649 }, { "epoch": 1.3674894705976435, "grad_norm": 1.7918416261672974, "learning_rate": 4.8702e-06, "loss": 0.0387, "step": 25650 }, { "epoch": 1.3675427840272965, "grad_norm": 1.1003665924072266, "learning_rate": 4.87e-06, "loss": 0.0287, "step": 25651 }, { "epoch": 1.3675960974569494, "grad_norm": 0.29583150148391724, "learning_rate": 4.8698e-06, "loss": 0.003, "step": 25652 }, { "epoch": 1.3676494108866024, "grad_norm": 1.3384686708450317, "learning_rate": 4.869600000000001e-06, "loss": 0.0182, "step": 25653 }, { "epoch": 1.3677027243162554, "grad_norm": 1.6958237886428833, "learning_rate": 4.8694e-06, "loss": 0.0221, "step": 25654 }, { "epoch": 1.3677560377459081, "grad_norm": 0.7578903436660767, "learning_rate": 4.869200000000001e-06, "loss": 0.0062, "step": 25655 }, { "epoch": 1.3678093511755611, "grad_norm": 1.6488893032073975, "learning_rate": 4.869e-06, "loss": 0.0295, "step": 25656 }, { "epoch": 1.367862664605214, "grad_norm": 5.229986190795898, "learning_rate": 4.8688e-06, "loss": 0.0666, "step": 25657 }, { "epoch": 1.3679159780348669, "grad_norm": 1.882594347000122, "learning_rate": 4.8686e-06, "loss": 0.0195, "step": 25658 }, { "epoch": 1.3679692914645198, "grad_norm": 2.4289422035217285, "learning_rate": 4.868400000000001e-06, "loss": 0.0184, "step": 25659 }, { "epoch": 1.3680226048941728, "grad_norm": 1.8858201503753662, "learning_rate": 4.8682e-06, "loss": 0.0117, "step": 25660 }, { "epoch": 1.3680759183238258, "grad_norm": 2.352735996246338, "learning_rate": 4.868000000000001e-06, "loss": 0.0433, "step": 25661 }, { "epoch": 1.3681292317534788, "grad_norm": 1.4541505575180054, "learning_rate": 4.8678e-06, "loss": 0.0199, "step": 25662 }, { "epoch": 1.3681825451831315, "grad_norm": 1.4464118480682373, "learning_rate": 4.867600000000001e-06, "loss": 0.0103, "step": 25663 }, { "epoch": 1.3682358586127845, "grad_norm": 1.7730674743652344, "learning_rate": 4.8674e-06, "loss": 0.0142, "step": 25664 }, { "epoch": 1.3682891720424375, "grad_norm": 2.447415351867676, "learning_rate": 4.867200000000001e-06, "loss": 0.022, "step": 25665 }, { "epoch": 1.3683424854720905, "grad_norm": 2.5415260791778564, "learning_rate": 4.867000000000001e-06, "loss": 0.0148, "step": 25666 }, { "epoch": 1.3683957989017435, "grad_norm": 3.4037089347839355, "learning_rate": 4.8668000000000006e-06, "loss": 0.0175, "step": 25667 }, { "epoch": 1.3684491123313962, "grad_norm": 1.5235077142715454, "learning_rate": 4.8666e-06, "loss": 0.0092, "step": 25668 }, { "epoch": 1.3685024257610492, "grad_norm": 1.7754021883010864, "learning_rate": 4.8664000000000005e-06, "loss": 0.027, "step": 25669 }, { "epoch": 1.3685557391907022, "grad_norm": 2.3274905681610107, "learning_rate": 4.8662e-06, "loss": 0.0729, "step": 25670 }, { "epoch": 1.3686090526203551, "grad_norm": 1.0914231538772583, "learning_rate": 4.8660000000000005e-06, "loss": 0.0273, "step": 25671 }, { "epoch": 1.3686623660500081, "grad_norm": 1.7354888916015625, "learning_rate": 4.865800000000001e-06, "loss": 0.0113, "step": 25672 }, { "epoch": 1.3687156794796609, "grad_norm": 2.101926803588867, "learning_rate": 4.8656000000000005e-06, "loss": 0.0546, "step": 25673 }, { "epoch": 1.3687689929093139, "grad_norm": 0.6908646821975708, "learning_rate": 4.8654e-06, "loss": 0.0057, "step": 25674 }, { "epoch": 1.3688223063389668, "grad_norm": 1.0247148275375366, "learning_rate": 4.8652000000000004e-06, "loss": 0.0074, "step": 25675 }, { "epoch": 1.3688756197686196, "grad_norm": 1.0196192264556885, "learning_rate": 4.865e-06, "loss": 0.0157, "step": 25676 }, { "epoch": 1.3689289331982726, "grad_norm": 0.9771112203598022, "learning_rate": 4.8648e-06, "loss": 0.0096, "step": 25677 }, { "epoch": 1.3689822466279256, "grad_norm": 1.5481609106063843, "learning_rate": 4.864600000000001e-06, "loss": 0.0128, "step": 25678 }, { "epoch": 1.3690355600575785, "grad_norm": 0.9349567294120789, "learning_rate": 4.8644e-06, "loss": 0.0107, "step": 25679 }, { "epoch": 1.3690888734872315, "grad_norm": 2.0796399116516113, "learning_rate": 4.864200000000001e-06, "loss": 0.0227, "step": 25680 }, { "epoch": 1.3691421869168843, "grad_norm": 0.9899868965148926, "learning_rate": 4.864e-06, "loss": 0.0147, "step": 25681 }, { "epoch": 1.3691955003465373, "grad_norm": 2.536447048187256, "learning_rate": 4.8638e-06, "loss": 0.0382, "step": 25682 }, { "epoch": 1.3692488137761902, "grad_norm": 2.762651205062866, "learning_rate": 4.8636e-06, "loss": 0.0379, "step": 25683 }, { "epoch": 1.3693021272058432, "grad_norm": 1.7143524885177612, "learning_rate": 4.863400000000001e-06, "loss": 0.0337, "step": 25684 }, { "epoch": 1.3693554406354962, "grad_norm": 1.4440637826919556, "learning_rate": 4.8632e-06, "loss": 0.0095, "step": 25685 }, { "epoch": 1.369408754065149, "grad_norm": 2.0444161891937256, "learning_rate": 4.863000000000001e-06, "loss": 0.0221, "step": 25686 }, { "epoch": 1.369462067494802, "grad_norm": 1.7732573747634888, "learning_rate": 4.8628e-06, "loss": 0.0288, "step": 25687 }, { "epoch": 1.369515380924455, "grad_norm": 2.4226629734039307, "learning_rate": 4.862600000000001e-06, "loss": 0.0282, "step": 25688 }, { "epoch": 1.3695686943541077, "grad_norm": 2.4783341884613037, "learning_rate": 4.8624e-06, "loss": 0.0234, "step": 25689 }, { "epoch": 1.3696220077837609, "grad_norm": 1.2394182682037354, "learning_rate": 4.862200000000001e-06, "loss": 0.0236, "step": 25690 }, { "epoch": 1.3696753212134136, "grad_norm": 0.8376607298851013, "learning_rate": 4.862e-06, "loss": 0.0113, "step": 25691 }, { "epoch": 1.3697286346430666, "grad_norm": 1.9913573265075684, "learning_rate": 4.861800000000001e-06, "loss": 0.0359, "step": 25692 }, { "epoch": 1.3697819480727196, "grad_norm": 1.2144919633865356, "learning_rate": 4.8616e-06, "loss": 0.0222, "step": 25693 }, { "epoch": 1.3698352615023723, "grad_norm": 1.1064529418945312, "learning_rate": 4.861400000000001e-06, "loss": 0.0125, "step": 25694 }, { "epoch": 1.3698885749320253, "grad_norm": 2.9657881259918213, "learning_rate": 4.8612e-06, "loss": 0.0267, "step": 25695 }, { "epoch": 1.3699418883616783, "grad_norm": 1.7485079765319824, "learning_rate": 4.8610000000000006e-06, "loss": 0.0192, "step": 25696 }, { "epoch": 1.3699952017913313, "grad_norm": 1.2976553440093994, "learning_rate": 4.8608e-06, "loss": 0.055, "step": 25697 }, { "epoch": 1.3700485152209843, "grad_norm": 1.4988634586334229, "learning_rate": 4.8606000000000005e-06, "loss": 0.0219, "step": 25698 }, { "epoch": 1.370101828650637, "grad_norm": 1.2170615196228027, "learning_rate": 4.860400000000001e-06, "loss": 0.0136, "step": 25699 }, { "epoch": 1.37015514208029, "grad_norm": 1.3233063220977783, "learning_rate": 4.8602000000000005e-06, "loss": 0.0269, "step": 25700 }, { "epoch": 1.370208455509943, "grad_norm": 0.6378885507583618, "learning_rate": 4.86e-06, "loss": 0.0064, "step": 25701 }, { "epoch": 1.370261768939596, "grad_norm": 1.816702961921692, "learning_rate": 4.8598000000000005e-06, "loss": 0.0494, "step": 25702 }, { "epoch": 1.370315082369249, "grad_norm": 1.113223671913147, "learning_rate": 4.8596e-06, "loss": 0.0075, "step": 25703 }, { "epoch": 1.3703683957989017, "grad_norm": 1.7296656370162964, "learning_rate": 4.8594000000000004e-06, "loss": 0.0487, "step": 25704 }, { "epoch": 1.3704217092285547, "grad_norm": 1.236904501914978, "learning_rate": 4.859200000000001e-06, "loss": 0.0101, "step": 25705 }, { "epoch": 1.3704750226582076, "grad_norm": 2.3613667488098145, "learning_rate": 4.859e-06, "loss": 0.025, "step": 25706 }, { "epoch": 1.3705283360878604, "grad_norm": 1.1252233982086182, "learning_rate": 4.8588e-06, "loss": 0.0203, "step": 25707 }, { "epoch": 1.3705816495175134, "grad_norm": 0.8374514579772949, "learning_rate": 4.8586e-06, "loss": 0.007, "step": 25708 }, { "epoch": 1.3706349629471664, "grad_norm": 3.5618386268615723, "learning_rate": 4.8584e-06, "loss": 0.0415, "step": 25709 }, { "epoch": 1.3706882763768193, "grad_norm": 1.4646912813186646, "learning_rate": 4.8582e-06, "loss": 0.0093, "step": 25710 }, { "epoch": 1.3707415898064723, "grad_norm": 3.818549394607544, "learning_rate": 4.858000000000001e-06, "loss": 0.0293, "step": 25711 }, { "epoch": 1.370794903236125, "grad_norm": 1.5852724313735962, "learning_rate": 4.8578e-06, "loss": 0.0208, "step": 25712 }, { "epoch": 1.370848216665778, "grad_norm": 2.022355794906616, "learning_rate": 4.857600000000001e-06, "loss": 0.0101, "step": 25713 }, { "epoch": 1.370901530095431, "grad_norm": 2.1660373210906982, "learning_rate": 4.8574e-06, "loss": 0.0479, "step": 25714 }, { "epoch": 1.370954843525084, "grad_norm": 1.364829182624817, "learning_rate": 4.8572e-06, "loss": 0.0372, "step": 25715 }, { "epoch": 1.371008156954737, "grad_norm": 1.6154024600982666, "learning_rate": 4.857e-06, "loss": 0.0234, "step": 25716 }, { "epoch": 1.3710614703843897, "grad_norm": 0.9682797193527222, "learning_rate": 4.856800000000001e-06, "loss": 0.0177, "step": 25717 }, { "epoch": 1.3711147838140427, "grad_norm": 1.2727952003479004, "learning_rate": 4.8566e-06, "loss": 0.0082, "step": 25718 }, { "epoch": 1.3711680972436957, "grad_norm": 0.3989020586013794, "learning_rate": 4.856400000000001e-06, "loss": 0.0042, "step": 25719 }, { "epoch": 1.3712214106733487, "grad_norm": 0.8742486238479614, "learning_rate": 4.8562e-06, "loss": 0.0103, "step": 25720 }, { "epoch": 1.3712747241030017, "grad_norm": 0.7608031630516052, "learning_rate": 4.856e-06, "loss": 0.0062, "step": 25721 }, { "epoch": 1.3713280375326544, "grad_norm": 0.9508411884307861, "learning_rate": 4.8558e-06, "loss": 0.0056, "step": 25722 }, { "epoch": 1.3713813509623074, "grad_norm": 1.293806552886963, "learning_rate": 4.8556000000000006e-06, "loss": 0.0222, "step": 25723 }, { "epoch": 1.3714346643919604, "grad_norm": 1.6243467330932617, "learning_rate": 4.855400000000001e-06, "loss": 0.019, "step": 25724 }, { "epoch": 1.3714879778216131, "grad_norm": 0.9788157939910889, "learning_rate": 4.8552000000000005e-06, "loss": 0.008, "step": 25725 }, { "epoch": 1.3715412912512661, "grad_norm": 0.84771329164505, "learning_rate": 4.855e-06, "loss": 0.0196, "step": 25726 }, { "epoch": 1.371594604680919, "grad_norm": 1.4163427352905273, "learning_rate": 4.8548000000000005e-06, "loss": 0.0358, "step": 25727 }, { "epoch": 1.371647918110572, "grad_norm": 3.3805251121520996, "learning_rate": 4.8546e-06, "loss": 0.0333, "step": 25728 }, { "epoch": 1.371701231540225, "grad_norm": 2.422419548034668, "learning_rate": 4.8544000000000005e-06, "loss": 0.0157, "step": 25729 }, { "epoch": 1.3717545449698778, "grad_norm": 1.2422515153884888, "learning_rate": 4.854200000000001e-06, "loss": 0.0074, "step": 25730 }, { "epoch": 1.3718078583995308, "grad_norm": 1.5147777795791626, "learning_rate": 4.8540000000000005e-06, "loss": 0.0404, "step": 25731 }, { "epoch": 1.3718611718291838, "grad_norm": 0.5594527125358582, "learning_rate": 4.8538e-06, "loss": 0.0054, "step": 25732 }, { "epoch": 1.3719144852588367, "grad_norm": 2.4275147914886475, "learning_rate": 4.8536e-06, "loss": 0.0471, "step": 25733 }, { "epoch": 1.3719677986884897, "grad_norm": 1.2153276205062866, "learning_rate": 4.8534e-06, "loss": 0.0139, "step": 25734 }, { "epoch": 1.3720211121181425, "grad_norm": 0.28151923418045044, "learning_rate": 4.8532e-06, "loss": 0.0028, "step": 25735 }, { "epoch": 1.3720744255477955, "grad_norm": 1.842324137687683, "learning_rate": 4.853000000000001e-06, "loss": 0.0257, "step": 25736 }, { "epoch": 1.3721277389774484, "grad_norm": 1.2136890888214111, "learning_rate": 4.8528e-06, "loss": 0.0114, "step": 25737 }, { "epoch": 1.3721810524071014, "grad_norm": 1.4213913679122925, "learning_rate": 4.852600000000001e-06, "loss": 0.0352, "step": 25738 }, { "epoch": 1.3722343658367544, "grad_norm": 0.664201557636261, "learning_rate": 4.8524e-06, "loss": 0.0064, "step": 25739 }, { "epoch": 1.3722876792664072, "grad_norm": 2.4904799461364746, "learning_rate": 4.8522e-06, "loss": 0.0163, "step": 25740 }, { "epoch": 1.3723409926960601, "grad_norm": 1.1332381963729858, "learning_rate": 4.852e-06, "loss": 0.0187, "step": 25741 }, { "epoch": 1.3723943061257131, "grad_norm": 1.4565694332122803, "learning_rate": 4.851800000000001e-06, "loss": 0.0139, "step": 25742 }, { "epoch": 1.3724476195553659, "grad_norm": 2.025146245956421, "learning_rate": 4.8516e-06, "loss": 0.0181, "step": 25743 }, { "epoch": 1.3725009329850189, "grad_norm": 3.234767436981201, "learning_rate": 4.851400000000001e-06, "loss": 0.0342, "step": 25744 }, { "epoch": 1.3725542464146718, "grad_norm": 0.9825741648674011, "learning_rate": 4.8512e-06, "loss": 0.0108, "step": 25745 }, { "epoch": 1.3726075598443248, "grad_norm": 1.8138303756713867, "learning_rate": 4.851e-06, "loss": 0.0237, "step": 25746 }, { "epoch": 1.3726608732739778, "grad_norm": 1.4686219692230225, "learning_rate": 4.8508e-06, "loss": 0.0228, "step": 25747 }, { "epoch": 1.3727141867036305, "grad_norm": 0.4714204967021942, "learning_rate": 4.850600000000001e-06, "loss": 0.0049, "step": 25748 }, { "epoch": 1.3727675001332835, "grad_norm": 0.698977530002594, "learning_rate": 4.850400000000001e-06, "loss": 0.0071, "step": 25749 }, { "epoch": 1.3728208135629365, "grad_norm": 1.5857514142990112, "learning_rate": 4.850200000000001e-06, "loss": 0.0317, "step": 25750 }, { "epoch": 1.3728741269925895, "grad_norm": 1.5296826362609863, "learning_rate": 4.85e-06, "loss": 0.0314, "step": 25751 }, { "epoch": 1.3729274404222425, "grad_norm": 1.8385899066925049, "learning_rate": 4.8498000000000006e-06, "loss": 0.0142, "step": 25752 }, { "epoch": 1.3729807538518952, "grad_norm": 0.9403668642044067, "learning_rate": 4.8496e-06, "loss": 0.0158, "step": 25753 }, { "epoch": 1.3730340672815482, "grad_norm": 1.1373333930969238, "learning_rate": 4.8494000000000005e-06, "loss": 0.0158, "step": 25754 }, { "epoch": 1.3730873807112012, "grad_norm": 1.152016520500183, "learning_rate": 4.849200000000001e-06, "loss": 0.0158, "step": 25755 }, { "epoch": 1.3731406941408542, "grad_norm": 1.574378490447998, "learning_rate": 4.8490000000000005e-06, "loss": 0.0267, "step": 25756 }, { "epoch": 1.3731940075705071, "grad_norm": 0.3925306797027588, "learning_rate": 4.8488e-06, "loss": 0.0046, "step": 25757 }, { "epoch": 1.37324732100016, "grad_norm": 3.248603582382202, "learning_rate": 4.8486000000000005e-06, "loss": 0.0218, "step": 25758 }, { "epoch": 1.3733006344298129, "grad_norm": 1.2325469255447388, "learning_rate": 4.8484e-06, "loss": 0.0076, "step": 25759 }, { "epoch": 1.3733539478594659, "grad_norm": 1.6154454946517944, "learning_rate": 4.8482000000000004e-06, "loss": 0.0192, "step": 25760 }, { "epoch": 1.3734072612891186, "grad_norm": 0.8938968777656555, "learning_rate": 4.848000000000001e-06, "loss": 0.0088, "step": 25761 }, { "epoch": 1.3734605747187716, "grad_norm": 0.4267410933971405, "learning_rate": 4.8478e-06, "loss": 0.0038, "step": 25762 }, { "epoch": 1.3735138881484246, "grad_norm": 1.6137397289276123, "learning_rate": 4.847600000000001e-06, "loss": 0.024, "step": 25763 }, { "epoch": 1.3735672015780775, "grad_norm": 1.9862453937530518, "learning_rate": 4.8474e-06, "loss": 0.0129, "step": 25764 }, { "epoch": 1.3736205150077305, "grad_norm": 1.4532216787338257, "learning_rate": 4.8472e-06, "loss": 0.0422, "step": 25765 }, { "epoch": 1.3736738284373833, "grad_norm": 0.3552577793598175, "learning_rate": 4.847e-06, "loss": 0.0044, "step": 25766 }, { "epoch": 1.3737271418670363, "grad_norm": 1.3810575008392334, "learning_rate": 4.846800000000001e-06, "loss": 0.0161, "step": 25767 }, { "epoch": 1.3737804552966892, "grad_norm": 1.395390272140503, "learning_rate": 4.8466e-06, "loss": 0.0385, "step": 25768 }, { "epoch": 1.3738337687263422, "grad_norm": 2.8748841285705566, "learning_rate": 4.846400000000001e-06, "loss": 0.0254, "step": 25769 }, { "epoch": 1.3738870821559952, "grad_norm": 2.3911123275756836, "learning_rate": 4.8462e-06, "loss": 0.0292, "step": 25770 }, { "epoch": 1.373940395585648, "grad_norm": 1.8175530433654785, "learning_rate": 4.846e-06, "loss": 0.0337, "step": 25771 }, { "epoch": 1.373993709015301, "grad_norm": 2.2861669063568115, "learning_rate": 4.8458e-06, "loss": 0.0352, "step": 25772 }, { "epoch": 1.374047022444954, "grad_norm": 2.129643440246582, "learning_rate": 4.845600000000001e-06, "loss": 0.0338, "step": 25773 }, { "epoch": 1.374100335874607, "grad_norm": 3.9343621730804443, "learning_rate": 4.845400000000001e-06, "loss": 0.0279, "step": 25774 }, { "epoch": 1.3741536493042599, "grad_norm": 3.71079421043396, "learning_rate": 4.845200000000001e-06, "loss": 0.0241, "step": 25775 }, { "epoch": 1.3742069627339126, "grad_norm": 1.9294549226760864, "learning_rate": 4.845e-06, "loss": 0.0179, "step": 25776 }, { "epoch": 1.3742602761635656, "grad_norm": 2.44187331199646, "learning_rate": 4.844800000000001e-06, "loss": 0.0324, "step": 25777 }, { "epoch": 1.3743135895932186, "grad_norm": 2.8103811740875244, "learning_rate": 4.8446e-06, "loss": 0.0163, "step": 25778 }, { "epoch": 1.3743669030228713, "grad_norm": 1.6116019487380981, "learning_rate": 4.844400000000001e-06, "loss": 0.0264, "step": 25779 }, { "epoch": 1.3744202164525243, "grad_norm": 2.5613248348236084, "learning_rate": 4.844200000000001e-06, "loss": 0.0271, "step": 25780 }, { "epoch": 1.3744735298821773, "grad_norm": 3.245824098587036, "learning_rate": 4.8440000000000005e-06, "loss": 0.0498, "step": 25781 }, { "epoch": 1.3745268433118303, "grad_norm": 1.8165885210037231, "learning_rate": 4.8438e-06, "loss": 0.0304, "step": 25782 }, { "epoch": 1.3745801567414833, "grad_norm": 1.4242109060287476, "learning_rate": 4.8436000000000005e-06, "loss": 0.0335, "step": 25783 }, { "epoch": 1.374633470171136, "grad_norm": 0.4089294373989105, "learning_rate": 4.8434e-06, "loss": 0.0034, "step": 25784 }, { "epoch": 1.374686783600789, "grad_norm": 1.0046237707138062, "learning_rate": 4.8432000000000005e-06, "loss": 0.0212, "step": 25785 }, { "epoch": 1.374740097030442, "grad_norm": 3.008626937866211, "learning_rate": 4.843000000000001e-06, "loss": 0.0208, "step": 25786 }, { "epoch": 1.374793410460095, "grad_norm": 1.4215041399002075, "learning_rate": 4.8428000000000005e-06, "loss": 0.0147, "step": 25787 }, { "epoch": 1.374846723889748, "grad_norm": 1.0164880752563477, "learning_rate": 4.842600000000001e-06, "loss": 0.0102, "step": 25788 }, { "epoch": 1.3749000373194007, "grad_norm": 2.4736196994781494, "learning_rate": 4.8424000000000004e-06, "loss": 0.0272, "step": 25789 }, { "epoch": 1.3749533507490537, "grad_norm": 2.095024347305298, "learning_rate": 4.8422e-06, "loss": 0.0348, "step": 25790 }, { "epoch": 1.3750066641787067, "grad_norm": 0.9292197227478027, "learning_rate": 4.842e-06, "loss": 0.0155, "step": 25791 }, { "epoch": 1.3750599776083594, "grad_norm": 3.439640998840332, "learning_rate": 4.841800000000001e-06, "loss": 0.0381, "step": 25792 }, { "epoch": 1.3751132910380126, "grad_norm": 0.8364968299865723, "learning_rate": 4.8416e-06, "loss": 0.0054, "step": 25793 }, { "epoch": 1.3751666044676654, "grad_norm": 1.1880680322647095, "learning_rate": 4.841400000000001e-06, "loss": 0.0145, "step": 25794 }, { "epoch": 1.3752199178973183, "grad_norm": 2.467886447906494, "learning_rate": 4.8412e-06, "loss": 0.0271, "step": 25795 }, { "epoch": 1.3752732313269713, "grad_norm": 2.017613649368286, "learning_rate": 4.841e-06, "loss": 0.0314, "step": 25796 }, { "epoch": 1.375326544756624, "grad_norm": 2.061307907104492, "learning_rate": 4.8408e-06, "loss": 0.0344, "step": 25797 }, { "epoch": 1.375379858186277, "grad_norm": 1.7595323324203491, "learning_rate": 4.840600000000001e-06, "loss": 0.0107, "step": 25798 }, { "epoch": 1.37543317161593, "grad_norm": 1.6470588445663452, "learning_rate": 4.8404e-06, "loss": 0.0297, "step": 25799 }, { "epoch": 1.375486485045583, "grad_norm": 1.0648751258850098, "learning_rate": 4.840200000000001e-06, "loss": 0.0163, "step": 25800 }, { "epoch": 1.375539798475236, "grad_norm": 3.340212821960449, "learning_rate": 4.84e-06, "loss": 0.0305, "step": 25801 }, { "epoch": 1.3755931119048888, "grad_norm": 2.5766425132751465, "learning_rate": 4.839800000000001e-06, "loss": 0.0257, "step": 25802 }, { "epoch": 1.3756464253345417, "grad_norm": 2.522336959838867, "learning_rate": 4.8396e-06, "loss": 0.0462, "step": 25803 }, { "epoch": 1.3756997387641947, "grad_norm": 2.1619558334350586, "learning_rate": 4.839400000000001e-06, "loss": 0.021, "step": 25804 }, { "epoch": 1.3757530521938477, "grad_norm": 1.7579594850540161, "learning_rate": 4.8392e-06, "loss": 0.0309, "step": 25805 }, { "epoch": 1.3758063656235007, "grad_norm": 2.846064329147339, "learning_rate": 4.839000000000001e-06, "loss": 0.0256, "step": 25806 }, { "epoch": 1.3758596790531534, "grad_norm": 1.5373660326004028, "learning_rate": 4.8388e-06, "loss": 0.0469, "step": 25807 }, { "epoch": 1.3759129924828064, "grad_norm": 3.3474667072296143, "learning_rate": 4.8386000000000006e-06, "loss": 0.0259, "step": 25808 }, { "epoch": 1.3759663059124594, "grad_norm": 1.9489684104919434, "learning_rate": 4.8384e-06, "loss": 0.0155, "step": 25809 }, { "epoch": 1.3760196193421121, "grad_norm": 1.844961404800415, "learning_rate": 4.8382000000000005e-06, "loss": 0.0188, "step": 25810 }, { "epoch": 1.3760729327717653, "grad_norm": 1.034988522529602, "learning_rate": 4.838e-06, "loss": 0.0229, "step": 25811 }, { "epoch": 1.376126246201418, "grad_norm": 2.0359010696411133, "learning_rate": 4.8378000000000005e-06, "loss": 0.0279, "step": 25812 }, { "epoch": 1.376179559631071, "grad_norm": 2.007230758666992, "learning_rate": 4.837600000000001e-06, "loss": 0.0352, "step": 25813 }, { "epoch": 1.376232873060724, "grad_norm": 1.187827467918396, "learning_rate": 4.8374000000000005e-06, "loss": 0.014, "step": 25814 }, { "epoch": 1.3762861864903768, "grad_norm": 1.6953092813491821, "learning_rate": 4.8372e-06, "loss": 0.0265, "step": 25815 }, { "epoch": 1.3763394999200298, "grad_norm": 1.1572120189666748, "learning_rate": 4.8370000000000004e-06, "loss": 0.0155, "step": 25816 }, { "epoch": 1.3763928133496828, "grad_norm": 2.239511728286743, "learning_rate": 4.8368e-06, "loss": 0.0204, "step": 25817 }, { "epoch": 1.3764461267793358, "grad_norm": 1.5527846813201904, "learning_rate": 4.8366e-06, "loss": 0.0211, "step": 25818 }, { "epoch": 1.3764994402089887, "grad_norm": 0.6824601888656616, "learning_rate": 4.836400000000001e-06, "loss": 0.0187, "step": 25819 }, { "epoch": 1.3765527536386415, "grad_norm": 1.298523187637329, "learning_rate": 4.8362e-06, "loss": 0.0102, "step": 25820 }, { "epoch": 1.3766060670682945, "grad_norm": 2.3504223823547363, "learning_rate": 4.836e-06, "loss": 0.0404, "step": 25821 }, { "epoch": 1.3766593804979474, "grad_norm": 0.9266524314880371, "learning_rate": 4.8358e-06, "loss": 0.0198, "step": 25822 }, { "epoch": 1.3767126939276004, "grad_norm": 0.6016204953193665, "learning_rate": 4.8356e-06, "loss": 0.0059, "step": 25823 }, { "epoch": 1.3767660073572534, "grad_norm": 2.019545078277588, "learning_rate": 4.8354e-06, "loss": 0.032, "step": 25824 }, { "epoch": 1.3768193207869062, "grad_norm": 2.0466301441192627, "learning_rate": 4.835200000000001e-06, "loss": 0.02, "step": 25825 }, { "epoch": 1.3768726342165591, "grad_norm": 2.003458023071289, "learning_rate": 4.835e-06, "loss": 0.0339, "step": 25826 }, { "epoch": 1.3769259476462121, "grad_norm": 1.9435505867004395, "learning_rate": 4.834800000000001e-06, "loss": 0.0535, "step": 25827 }, { "epoch": 1.3769792610758649, "grad_norm": 1.7181828022003174, "learning_rate": 4.8346e-06, "loss": 0.027, "step": 25828 }, { "epoch": 1.3770325745055179, "grad_norm": 2.0942702293395996, "learning_rate": 4.8344e-06, "loss": 0.0483, "step": 25829 }, { "epoch": 1.3770858879351708, "grad_norm": 4.0355095863342285, "learning_rate": 4.8342e-06, "loss": 0.0351, "step": 25830 }, { "epoch": 1.3771392013648238, "grad_norm": 2.8603475093841553, "learning_rate": 4.834000000000001e-06, "loss": 0.0313, "step": 25831 }, { "epoch": 1.3771925147944768, "grad_norm": 1.0759855508804321, "learning_rate": 4.8338e-06, "loss": 0.0097, "step": 25832 }, { "epoch": 1.3772458282241296, "grad_norm": 2.698577642440796, "learning_rate": 4.833600000000001e-06, "loss": 0.0208, "step": 25833 }, { "epoch": 1.3772991416537825, "grad_norm": 2.1564576625823975, "learning_rate": 4.8334e-06, "loss": 0.0205, "step": 25834 }, { "epoch": 1.3773524550834355, "grad_norm": 1.742692470550537, "learning_rate": 4.833200000000001e-06, "loss": 0.0224, "step": 25835 }, { "epoch": 1.3774057685130885, "grad_norm": 1.1132365465164185, "learning_rate": 4.833e-06, "loss": 0.0254, "step": 25836 }, { "epoch": 1.3774590819427415, "grad_norm": 1.026230812072754, "learning_rate": 4.8328000000000006e-06, "loss": 0.0079, "step": 25837 }, { "epoch": 1.3775123953723942, "grad_norm": 2.382168769836426, "learning_rate": 4.832600000000001e-06, "loss": 0.0332, "step": 25838 }, { "epoch": 1.3775657088020472, "grad_norm": 1.3031024932861328, "learning_rate": 4.8324000000000005e-06, "loss": 0.0336, "step": 25839 }, { "epoch": 1.3776190222317002, "grad_norm": 4.644299507141113, "learning_rate": 4.8322e-06, "loss": 0.0668, "step": 25840 }, { "epoch": 1.3776723356613532, "grad_norm": 1.3159130811691284, "learning_rate": 4.8320000000000005e-06, "loss": 0.0145, "step": 25841 }, { "epoch": 1.3777256490910061, "grad_norm": 1.4877302646636963, "learning_rate": 4.8318e-06, "loss": 0.0292, "step": 25842 }, { "epoch": 1.377778962520659, "grad_norm": 1.0113948583602905, "learning_rate": 4.8316000000000005e-06, "loss": 0.0086, "step": 25843 }, { "epoch": 1.3778322759503119, "grad_norm": 1.272843360900879, "learning_rate": 4.831400000000001e-06, "loss": 0.0261, "step": 25844 }, { "epoch": 1.3778855893799649, "grad_norm": 0.7354918122291565, "learning_rate": 4.8312000000000004e-06, "loss": 0.0142, "step": 25845 }, { "epoch": 1.3779389028096176, "grad_norm": 2.4642629623413086, "learning_rate": 4.831e-06, "loss": 0.0219, "step": 25846 }, { "epoch": 1.3779922162392706, "grad_norm": 1.6049011945724487, "learning_rate": 4.8308e-06, "loss": 0.0118, "step": 25847 }, { "epoch": 1.3780455296689236, "grad_norm": 1.1828525066375732, "learning_rate": 4.8306e-06, "loss": 0.0105, "step": 25848 }, { "epoch": 1.3780988430985766, "grad_norm": 1.0960705280303955, "learning_rate": 4.8304e-06, "loss": 0.0143, "step": 25849 }, { "epoch": 1.3781521565282295, "grad_norm": 2.3457772731781006, "learning_rate": 4.830200000000001e-06, "loss": 0.0517, "step": 25850 }, { "epoch": 1.3782054699578823, "grad_norm": 0.7178844213485718, "learning_rate": 4.83e-06, "loss": 0.0068, "step": 25851 }, { "epoch": 1.3782587833875353, "grad_norm": 1.2222011089324951, "learning_rate": 4.829800000000001e-06, "loss": 0.0157, "step": 25852 }, { "epoch": 1.3783120968171882, "grad_norm": 1.710340142250061, "learning_rate": 4.8296e-06, "loss": 0.0252, "step": 25853 }, { "epoch": 1.3783654102468412, "grad_norm": 1.676823377609253, "learning_rate": 4.8294e-06, "loss": 0.012, "step": 25854 }, { "epoch": 1.3784187236764942, "grad_norm": 1.8668859004974365, "learning_rate": 4.8292e-06, "loss": 0.037, "step": 25855 }, { "epoch": 1.378472037106147, "grad_norm": 1.2931097745895386, "learning_rate": 4.829000000000001e-06, "loss": 0.0105, "step": 25856 }, { "epoch": 1.3785253505358, "grad_norm": 3.0159969329833984, "learning_rate": 4.8288e-06, "loss": 0.0343, "step": 25857 }, { "epoch": 1.378578663965453, "grad_norm": 1.628455638885498, "learning_rate": 4.828600000000001e-06, "loss": 0.0067, "step": 25858 }, { "epoch": 1.378631977395106, "grad_norm": 2.461428165435791, "learning_rate": 4.8284e-06, "loss": 0.0589, "step": 25859 }, { "epoch": 1.3786852908247589, "grad_norm": 1.0706766843795776, "learning_rate": 4.828200000000001e-06, "loss": 0.0085, "step": 25860 }, { "epoch": 1.3787386042544116, "grad_norm": 2.050330400466919, "learning_rate": 4.828e-06, "loss": 0.0141, "step": 25861 }, { "epoch": 1.3787919176840646, "grad_norm": 2.3671953678131104, "learning_rate": 4.827800000000001e-06, "loss": 0.0327, "step": 25862 }, { "epoch": 1.3788452311137176, "grad_norm": 1.5963404178619385, "learning_rate": 4.827600000000001e-06, "loss": 0.0087, "step": 25863 }, { "epoch": 1.3788985445433704, "grad_norm": 1.3758708238601685, "learning_rate": 4.8274000000000006e-06, "loss": 0.0111, "step": 25864 }, { "epoch": 1.3789518579730233, "grad_norm": 1.0224876403808594, "learning_rate": 4.8272e-06, "loss": 0.0252, "step": 25865 }, { "epoch": 1.3790051714026763, "grad_norm": 1.8862751722335815, "learning_rate": 4.8270000000000005e-06, "loss": 0.014, "step": 25866 }, { "epoch": 1.3790584848323293, "grad_norm": 1.519534707069397, "learning_rate": 4.8268e-06, "loss": 0.0217, "step": 25867 }, { "epoch": 1.3791117982619823, "grad_norm": 1.081447958946228, "learning_rate": 4.8266000000000005e-06, "loss": 0.0059, "step": 25868 }, { "epoch": 1.379165111691635, "grad_norm": 2.6638097763061523, "learning_rate": 4.826400000000001e-06, "loss": 0.0466, "step": 25869 }, { "epoch": 1.379218425121288, "grad_norm": 2.1648271083831787, "learning_rate": 4.8262000000000005e-06, "loss": 0.0236, "step": 25870 }, { "epoch": 1.379271738550941, "grad_norm": 0.999870240688324, "learning_rate": 4.826e-06, "loss": 0.0101, "step": 25871 }, { "epoch": 1.379325051980594, "grad_norm": 1.3349511623382568, "learning_rate": 4.8258000000000005e-06, "loss": 0.0203, "step": 25872 }, { "epoch": 1.379378365410247, "grad_norm": 1.9575270414352417, "learning_rate": 4.8256e-06, "loss": 0.0286, "step": 25873 }, { "epoch": 1.3794316788398997, "grad_norm": 1.0323189496994019, "learning_rate": 4.8254e-06, "loss": 0.0116, "step": 25874 }, { "epoch": 1.3794849922695527, "grad_norm": 1.8938517570495605, "learning_rate": 4.825200000000001e-06, "loss": 0.0156, "step": 25875 }, { "epoch": 1.3795383056992057, "grad_norm": 1.4046324491500854, "learning_rate": 4.825e-06, "loss": 0.0109, "step": 25876 }, { "epoch": 1.3795916191288586, "grad_norm": 2.108015298843384, "learning_rate": 4.824800000000001e-06, "loss": 0.0395, "step": 25877 }, { "epoch": 1.3796449325585116, "grad_norm": 2.3264880180358887, "learning_rate": 4.8246e-06, "loss": 0.0235, "step": 25878 }, { "epoch": 1.3796982459881644, "grad_norm": 1.2641721963882446, "learning_rate": 4.8244e-06, "loss": 0.0075, "step": 25879 }, { "epoch": 1.3797515594178174, "grad_norm": 2.6942331790924072, "learning_rate": 4.8242e-06, "loss": 0.0293, "step": 25880 }, { "epoch": 1.3798048728474703, "grad_norm": 0.851213276386261, "learning_rate": 4.824000000000001e-06, "loss": 0.0076, "step": 25881 }, { "epoch": 1.379858186277123, "grad_norm": 1.7171343564987183, "learning_rate": 4.8238e-06, "loss": 0.0096, "step": 25882 }, { "epoch": 1.379911499706776, "grad_norm": 0.44546639919281006, "learning_rate": 4.823600000000001e-06, "loss": 0.004, "step": 25883 }, { "epoch": 1.379964813136429, "grad_norm": 1.310301661491394, "learning_rate": 4.8234e-06, "loss": 0.0131, "step": 25884 }, { "epoch": 1.380018126566082, "grad_norm": 0.8592319488525391, "learning_rate": 4.823200000000001e-06, "loss": 0.0114, "step": 25885 }, { "epoch": 1.380071439995735, "grad_norm": 1.1625391244888306, "learning_rate": 4.823e-06, "loss": 0.0226, "step": 25886 }, { "epoch": 1.3801247534253878, "grad_norm": 1.889182448387146, "learning_rate": 4.822800000000001e-06, "loss": 0.0276, "step": 25887 }, { "epoch": 1.3801780668550407, "grad_norm": 0.9677252173423767, "learning_rate": 4.8226e-06, "loss": 0.0081, "step": 25888 }, { "epoch": 1.3802313802846937, "grad_norm": 1.3007233142852783, "learning_rate": 4.822400000000001e-06, "loss": 0.0206, "step": 25889 }, { "epoch": 1.3802846937143467, "grad_norm": 1.7660475969314575, "learning_rate": 4.8222e-06, "loss": 0.0378, "step": 25890 }, { "epoch": 1.3803380071439997, "grad_norm": 0.7231853008270264, "learning_rate": 4.822000000000001e-06, "loss": 0.0037, "step": 25891 }, { "epoch": 1.3803913205736524, "grad_norm": 3.9714910984039307, "learning_rate": 4.8218e-06, "loss": 0.0648, "step": 25892 }, { "epoch": 1.3804446340033054, "grad_norm": 0.5869459509849548, "learning_rate": 4.8216000000000006e-06, "loss": 0.0052, "step": 25893 }, { "epoch": 1.3804979474329584, "grad_norm": 1.1141159534454346, "learning_rate": 4.8214e-06, "loss": 0.0089, "step": 25894 }, { "epoch": 1.3805512608626114, "grad_norm": 2.395461082458496, "learning_rate": 4.8212000000000005e-06, "loss": 0.012, "step": 25895 }, { "epoch": 1.3806045742922644, "grad_norm": 1.7599257230758667, "learning_rate": 4.821e-06, "loss": 0.0348, "step": 25896 }, { "epoch": 1.380657887721917, "grad_norm": 4.397392749786377, "learning_rate": 4.8208000000000005e-06, "loss": 0.0261, "step": 25897 }, { "epoch": 1.38071120115157, "grad_norm": 3.118999481201172, "learning_rate": 4.8206e-06, "loss": 0.0264, "step": 25898 }, { "epoch": 1.380764514581223, "grad_norm": 6.015344619750977, "learning_rate": 4.8204000000000005e-06, "loss": 0.0559, "step": 25899 }, { "epoch": 1.3808178280108758, "grad_norm": 0.9168856143951416, "learning_rate": 4.8202e-06, "loss": 0.0165, "step": 25900 }, { "epoch": 1.3808711414405288, "grad_norm": 2.534169912338257, "learning_rate": 4.8200000000000004e-06, "loss": 0.0313, "step": 25901 }, { "epoch": 1.3809244548701818, "grad_norm": 1.9594314098358154, "learning_rate": 4.819800000000001e-06, "loss": 0.0251, "step": 25902 }, { "epoch": 1.3809777682998348, "grad_norm": 1.975001335144043, "learning_rate": 4.8196e-06, "loss": 0.0264, "step": 25903 }, { "epoch": 1.3810310817294877, "grad_norm": 1.7136379480361938, "learning_rate": 4.8194e-06, "loss": 0.0212, "step": 25904 }, { "epoch": 1.3810843951591405, "grad_norm": 1.0806219577789307, "learning_rate": 4.8192e-06, "loss": 0.0161, "step": 25905 }, { "epoch": 1.3811377085887935, "grad_norm": 2.6679041385650635, "learning_rate": 4.819e-06, "loss": 0.0432, "step": 25906 }, { "epoch": 1.3811910220184465, "grad_norm": 2.6039822101593018, "learning_rate": 4.8188e-06, "loss": 0.0237, "step": 25907 }, { "epoch": 1.3812443354480994, "grad_norm": 1.7007968425750732, "learning_rate": 4.818600000000001e-06, "loss": 0.0309, "step": 25908 }, { "epoch": 1.3812976488777524, "grad_norm": 1.893257737159729, "learning_rate": 4.8184e-06, "loss": 0.0254, "step": 25909 }, { "epoch": 1.3813509623074052, "grad_norm": 2.42976450920105, "learning_rate": 4.818200000000001e-06, "loss": 0.049, "step": 25910 }, { "epoch": 1.3814042757370582, "grad_norm": 1.9148293733596802, "learning_rate": 4.818e-06, "loss": 0.0191, "step": 25911 }, { "epoch": 1.3814575891667111, "grad_norm": 2.1017258167266846, "learning_rate": 4.8178e-06, "loss": 0.0201, "step": 25912 }, { "epoch": 1.3815109025963639, "grad_norm": 1.3181016445159912, "learning_rate": 4.8176e-06, "loss": 0.0174, "step": 25913 }, { "epoch": 1.381564216026017, "grad_norm": 2.7291173934936523, "learning_rate": 4.817400000000001e-06, "loss": 0.0447, "step": 25914 }, { "epoch": 1.3816175294556698, "grad_norm": 1.1011154651641846, "learning_rate": 4.8172e-06, "loss": 0.017, "step": 25915 }, { "epoch": 1.3816708428853228, "grad_norm": 0.921479344367981, "learning_rate": 4.817000000000001e-06, "loss": 0.0077, "step": 25916 }, { "epoch": 1.3817241563149758, "grad_norm": 0.9624288082122803, "learning_rate": 4.8168e-06, "loss": 0.015, "step": 25917 }, { "epoch": 1.3817774697446286, "grad_norm": 1.1942224502563477, "learning_rate": 4.8166e-06, "loss": 0.0209, "step": 25918 }, { "epoch": 1.3818307831742815, "grad_norm": 1.0812734365463257, "learning_rate": 4.8164e-06, "loss": 0.0218, "step": 25919 }, { "epoch": 1.3818840966039345, "grad_norm": 1.9320945739746094, "learning_rate": 4.8162000000000006e-06, "loss": 0.0384, "step": 25920 }, { "epoch": 1.3819374100335875, "grad_norm": 1.4538044929504395, "learning_rate": 4.816e-06, "loss": 0.027, "step": 25921 }, { "epoch": 1.3819907234632405, "grad_norm": 1.3416056632995605, "learning_rate": 4.8158000000000005e-06, "loss": 0.0184, "step": 25922 }, { "epoch": 1.3820440368928932, "grad_norm": 1.590273141860962, "learning_rate": 4.8156e-06, "loss": 0.0277, "step": 25923 }, { "epoch": 1.3820973503225462, "grad_norm": 1.4436451196670532, "learning_rate": 4.8154000000000005e-06, "loss": 0.0255, "step": 25924 }, { "epoch": 1.3821506637521992, "grad_norm": 1.1245272159576416, "learning_rate": 4.8152e-06, "loss": 0.0126, "step": 25925 }, { "epoch": 1.3822039771818522, "grad_norm": 1.6451395750045776, "learning_rate": 4.8150000000000005e-06, "loss": 0.0211, "step": 25926 }, { "epoch": 1.3822572906115052, "grad_norm": 2.789179563522339, "learning_rate": 4.814800000000001e-06, "loss": 0.0369, "step": 25927 }, { "epoch": 1.382310604041158, "grad_norm": 2.146787643432617, "learning_rate": 4.8146000000000005e-06, "loss": 0.0093, "step": 25928 }, { "epoch": 1.3823639174708109, "grad_norm": 1.4992403984069824, "learning_rate": 4.8144e-06, "loss": 0.0119, "step": 25929 }, { "epoch": 1.3824172309004639, "grad_norm": 1.1915910243988037, "learning_rate": 4.8142000000000004e-06, "loss": 0.0217, "step": 25930 }, { "epoch": 1.3824705443301166, "grad_norm": 1.6410672664642334, "learning_rate": 4.814e-06, "loss": 0.0133, "step": 25931 }, { "epoch": 1.3825238577597696, "grad_norm": 0.8303158283233643, "learning_rate": 4.8138e-06, "loss": 0.0175, "step": 25932 }, { "epoch": 1.3825771711894226, "grad_norm": 1.4573274850845337, "learning_rate": 4.813600000000001e-06, "loss": 0.0163, "step": 25933 }, { "epoch": 1.3826304846190756, "grad_norm": 1.0431967973709106, "learning_rate": 4.8134e-06, "loss": 0.0147, "step": 25934 }, { "epoch": 1.3826837980487285, "grad_norm": 1.2423326969146729, "learning_rate": 4.813200000000001e-06, "loss": 0.009, "step": 25935 }, { "epoch": 1.3827371114783813, "grad_norm": 0.9833760261535645, "learning_rate": 4.813e-06, "loss": 0.0122, "step": 25936 }, { "epoch": 1.3827904249080343, "grad_norm": 1.3090158700942993, "learning_rate": 4.8128e-06, "loss": 0.0099, "step": 25937 }, { "epoch": 1.3828437383376873, "grad_norm": 1.2975742816925049, "learning_rate": 4.8126e-06, "loss": 0.0238, "step": 25938 }, { "epoch": 1.3828970517673402, "grad_norm": 1.2540531158447266, "learning_rate": 4.812400000000001e-06, "loss": 0.0071, "step": 25939 }, { "epoch": 1.3829503651969932, "grad_norm": 1.0132601261138916, "learning_rate": 4.8122e-06, "loss": 0.0028, "step": 25940 }, { "epoch": 1.383003678626646, "grad_norm": 2.4713287353515625, "learning_rate": 4.812000000000001e-06, "loss": 0.0267, "step": 25941 }, { "epoch": 1.383056992056299, "grad_norm": 1.1685738563537598, "learning_rate": 4.8118e-06, "loss": 0.0209, "step": 25942 }, { "epoch": 1.383110305485952, "grad_norm": 1.2448253631591797, "learning_rate": 4.8116e-06, "loss": 0.0085, "step": 25943 }, { "epoch": 1.383163618915605, "grad_norm": 1.0822328329086304, "learning_rate": 4.8114e-06, "loss": 0.0289, "step": 25944 }, { "epoch": 1.3832169323452579, "grad_norm": 0.31238991022109985, "learning_rate": 4.811200000000001e-06, "loss": 0.0026, "step": 25945 }, { "epoch": 1.3832702457749106, "grad_norm": 1.8483192920684814, "learning_rate": 4.811000000000001e-06, "loss": 0.0316, "step": 25946 }, { "epoch": 1.3833235592045636, "grad_norm": 1.366591215133667, "learning_rate": 4.810800000000001e-06, "loss": 0.0077, "step": 25947 }, { "epoch": 1.3833768726342166, "grad_norm": 3.3152928352355957, "learning_rate": 4.8106e-06, "loss": 0.0379, "step": 25948 }, { "epoch": 1.3834301860638694, "grad_norm": 3.126762866973877, "learning_rate": 4.8104000000000006e-06, "loss": 0.0461, "step": 25949 }, { "epoch": 1.3834834994935223, "grad_norm": 1.7635668516159058, "learning_rate": 4.8102e-06, "loss": 0.0362, "step": 25950 }, { "epoch": 1.3835368129231753, "grad_norm": 2.495375156402588, "learning_rate": 4.8100000000000005e-06, "loss": 0.0327, "step": 25951 }, { "epoch": 1.3835901263528283, "grad_norm": 1.5393906831741333, "learning_rate": 4.809800000000001e-06, "loss": 0.0157, "step": 25952 }, { "epoch": 1.3836434397824813, "grad_norm": 0.6778236627578735, "learning_rate": 4.8096000000000005e-06, "loss": 0.0058, "step": 25953 }, { "epoch": 1.383696753212134, "grad_norm": 1.5048539638519287, "learning_rate": 4.8094e-06, "loss": 0.0074, "step": 25954 }, { "epoch": 1.383750066641787, "grad_norm": 2.6223435401916504, "learning_rate": 4.8092000000000005e-06, "loss": 0.0394, "step": 25955 }, { "epoch": 1.38380338007144, "grad_norm": 3.982842445373535, "learning_rate": 4.809e-06, "loss": 0.0267, "step": 25956 }, { "epoch": 1.383856693501093, "grad_norm": 2.480891466140747, "learning_rate": 4.8088000000000004e-06, "loss": 0.0302, "step": 25957 }, { "epoch": 1.383910006930746, "grad_norm": 2.264972448348999, "learning_rate": 4.808600000000001e-06, "loss": 0.0439, "step": 25958 }, { "epoch": 1.3839633203603987, "grad_norm": 2.4603254795074463, "learning_rate": 4.8084e-06, "loss": 0.0257, "step": 25959 }, { "epoch": 1.3840166337900517, "grad_norm": 1.8439527750015259, "learning_rate": 4.808200000000001e-06, "loss": 0.0103, "step": 25960 }, { "epoch": 1.3840699472197047, "grad_norm": 2.2248153686523438, "learning_rate": 4.808e-06, "loss": 0.0177, "step": 25961 }, { "epoch": 1.3841232606493576, "grad_norm": 2.147294759750366, "learning_rate": 4.8078e-06, "loss": 0.0127, "step": 25962 }, { "epoch": 1.3841765740790106, "grad_norm": 5.977863311767578, "learning_rate": 4.8076e-06, "loss": 0.0267, "step": 25963 }, { "epoch": 1.3842298875086634, "grad_norm": 1.045560598373413, "learning_rate": 4.807400000000001e-06, "loss": 0.0109, "step": 25964 }, { "epoch": 1.3842832009383164, "grad_norm": 1.2006957530975342, "learning_rate": 4.8072e-06, "loss": 0.0091, "step": 25965 }, { "epoch": 1.3843365143679693, "grad_norm": 1.7856491804122925, "learning_rate": 4.807000000000001e-06, "loss": 0.0139, "step": 25966 }, { "epoch": 1.384389827797622, "grad_norm": 2.387887716293335, "learning_rate": 4.8068e-06, "loss": 0.042, "step": 25967 }, { "epoch": 1.384443141227275, "grad_norm": 1.370388150215149, "learning_rate": 4.8066e-06, "loss": 0.0123, "step": 25968 }, { "epoch": 1.384496454656928, "grad_norm": 1.4853789806365967, "learning_rate": 4.8064e-06, "loss": 0.0441, "step": 25969 }, { "epoch": 1.384549768086581, "grad_norm": 1.0807076692581177, "learning_rate": 4.806200000000001e-06, "loss": 0.0193, "step": 25970 }, { "epoch": 1.384603081516234, "grad_norm": 0.5050712823867798, "learning_rate": 4.806000000000001e-06, "loss": 0.0052, "step": 25971 }, { "epoch": 1.3846563949458868, "grad_norm": 2.360889196395874, "learning_rate": 4.805800000000001e-06, "loss": 0.0271, "step": 25972 }, { "epoch": 1.3847097083755397, "grad_norm": 1.5834437608718872, "learning_rate": 4.8056e-06, "loss": 0.0186, "step": 25973 }, { "epoch": 1.3847630218051927, "grad_norm": 3.462432384490967, "learning_rate": 4.805400000000001e-06, "loss": 0.0444, "step": 25974 }, { "epoch": 1.3848163352348457, "grad_norm": 1.6205161809921265, "learning_rate": 4.8052e-06, "loss": 0.0092, "step": 25975 }, { "epoch": 1.3848696486644987, "grad_norm": 0.7538356184959412, "learning_rate": 4.805000000000001e-06, "loss": 0.0076, "step": 25976 }, { "epoch": 1.3849229620941514, "grad_norm": 1.8474026918411255, "learning_rate": 4.804800000000001e-06, "loss": 0.0088, "step": 25977 }, { "epoch": 1.3849762755238044, "grad_norm": 1.5032925605773926, "learning_rate": 4.8046000000000006e-06, "loss": 0.0073, "step": 25978 }, { "epoch": 1.3850295889534574, "grad_norm": 0.8260272741317749, "learning_rate": 4.8044e-06, "loss": 0.0073, "step": 25979 }, { "epoch": 1.3850829023831104, "grad_norm": 1.3884707689285278, "learning_rate": 4.8042000000000005e-06, "loss": 0.0257, "step": 25980 }, { "epoch": 1.3851362158127634, "grad_norm": 0.4884069859981537, "learning_rate": 4.804e-06, "loss": 0.0041, "step": 25981 }, { "epoch": 1.3851895292424161, "grad_norm": 2.175334930419922, "learning_rate": 4.8038000000000005e-06, "loss": 0.0325, "step": 25982 }, { "epoch": 1.385242842672069, "grad_norm": 2.950608968734741, "learning_rate": 4.803600000000001e-06, "loss": 0.0205, "step": 25983 }, { "epoch": 1.385296156101722, "grad_norm": 1.2773081064224243, "learning_rate": 4.8034000000000005e-06, "loss": 0.0091, "step": 25984 }, { "epoch": 1.3853494695313748, "grad_norm": 2.0012152194976807, "learning_rate": 4.803200000000001e-06, "loss": 0.0398, "step": 25985 }, { "epoch": 1.3854027829610278, "grad_norm": 1.9282280206680298, "learning_rate": 4.8030000000000004e-06, "loss": 0.0313, "step": 25986 }, { "epoch": 1.3854560963906808, "grad_norm": 1.3464102745056152, "learning_rate": 4.8028e-06, "loss": 0.0098, "step": 25987 }, { "epoch": 1.3855094098203338, "grad_norm": 2.059997320175171, "learning_rate": 4.8026e-06, "loss": 0.012, "step": 25988 }, { "epoch": 1.3855627232499867, "grad_norm": 1.4171957969665527, "learning_rate": 4.802400000000001e-06, "loss": 0.0251, "step": 25989 }, { "epoch": 1.3856160366796395, "grad_norm": 0.68204665184021, "learning_rate": 4.8022e-06, "loss": 0.0045, "step": 25990 }, { "epoch": 1.3856693501092925, "grad_norm": 2.1299800872802734, "learning_rate": 4.802000000000001e-06, "loss": 0.0303, "step": 25991 }, { "epoch": 1.3857226635389455, "grad_norm": 2.349574089050293, "learning_rate": 4.8018e-06, "loss": 0.0459, "step": 25992 }, { "epoch": 1.3857759769685984, "grad_norm": 2.2320520877838135, "learning_rate": 4.8016e-06, "loss": 0.0311, "step": 25993 }, { "epoch": 1.3858292903982514, "grad_norm": 1.9275226593017578, "learning_rate": 4.8014e-06, "loss": 0.0209, "step": 25994 }, { "epoch": 1.3858826038279042, "grad_norm": 5.322754859924316, "learning_rate": 4.801200000000001e-06, "loss": 0.0263, "step": 25995 }, { "epoch": 1.3859359172575572, "grad_norm": 2.356822967529297, "learning_rate": 4.801e-06, "loss": 0.035, "step": 25996 }, { "epoch": 1.3859892306872101, "grad_norm": 1.049817681312561, "learning_rate": 4.800800000000001e-06, "loss": 0.0103, "step": 25997 }, { "epoch": 1.3860425441168631, "grad_norm": 2.0931360721588135, "learning_rate": 4.8006e-06, "loss": 0.023, "step": 25998 }, { "epoch": 1.386095857546516, "grad_norm": 1.2069649696350098, "learning_rate": 4.800400000000001e-06, "loss": 0.0141, "step": 25999 }, { "epoch": 1.3861491709761689, "grad_norm": 1.6161259412765503, "learning_rate": 4.8002e-06, "loss": 0.0141, "step": 26000 }, { "epoch": 1.3862024844058218, "grad_norm": 1.7195699214935303, "learning_rate": 4.800000000000001e-06, "loss": 0.0286, "step": 26001 }, { "epoch": 1.3862557978354748, "grad_norm": 1.3412283658981323, "learning_rate": 4.7998e-06, "loss": 0.0229, "step": 26002 }, { "epoch": 1.3863091112651276, "grad_norm": 2.589174747467041, "learning_rate": 4.799600000000001e-06, "loss": 0.0215, "step": 26003 }, { "epoch": 1.3863624246947805, "grad_norm": 2.953456401824951, "learning_rate": 4.7994e-06, "loss": 0.0321, "step": 26004 }, { "epoch": 1.3864157381244335, "grad_norm": 2.2723541259765625, "learning_rate": 4.7992000000000006e-06, "loss": 0.0283, "step": 26005 }, { "epoch": 1.3864690515540865, "grad_norm": 1.0071669816970825, "learning_rate": 4.799e-06, "loss": 0.0075, "step": 26006 }, { "epoch": 1.3865223649837395, "grad_norm": 0.9369758367538452, "learning_rate": 4.7988000000000005e-06, "loss": 0.0079, "step": 26007 }, { "epoch": 1.3865756784133922, "grad_norm": 1.7640739679336548, "learning_rate": 4.7986e-06, "loss": 0.0082, "step": 26008 }, { "epoch": 1.3866289918430452, "grad_norm": 2.822052478790283, "learning_rate": 4.7984000000000005e-06, "loss": 0.0665, "step": 26009 }, { "epoch": 1.3866823052726982, "grad_norm": 1.7215969562530518, "learning_rate": 4.798200000000001e-06, "loss": 0.0105, "step": 26010 }, { "epoch": 1.3867356187023512, "grad_norm": 1.668578028678894, "learning_rate": 4.7980000000000005e-06, "loss": 0.0388, "step": 26011 }, { "epoch": 1.3867889321320042, "grad_norm": 3.493192672729492, "learning_rate": 4.7978e-06, "loss": 0.0134, "step": 26012 }, { "epoch": 1.386842245561657, "grad_norm": 3.005314826965332, "learning_rate": 4.7976000000000004e-06, "loss": 0.0358, "step": 26013 }, { "epoch": 1.38689555899131, "grad_norm": 2.242473602294922, "learning_rate": 4.7974e-06, "loss": 0.0085, "step": 26014 }, { "epoch": 1.3869488724209629, "grad_norm": 2.4130821228027344, "learning_rate": 4.7972e-06, "loss": 0.0324, "step": 26015 }, { "epoch": 1.3870021858506156, "grad_norm": 0.7969481348991394, "learning_rate": 4.797000000000001e-06, "loss": 0.0051, "step": 26016 }, { "epoch": 1.3870554992802688, "grad_norm": 1.4516645669937134, "learning_rate": 4.7968e-06, "loss": 0.0176, "step": 26017 }, { "epoch": 1.3871088127099216, "grad_norm": 1.9993864297866821, "learning_rate": 4.7966e-06, "loss": 0.0266, "step": 26018 }, { "epoch": 1.3871621261395746, "grad_norm": 2.5062525272369385, "learning_rate": 4.7964e-06, "loss": 0.0249, "step": 26019 }, { "epoch": 1.3872154395692275, "grad_norm": 2.594538688659668, "learning_rate": 4.7962e-06, "loss": 0.0305, "step": 26020 }, { "epoch": 1.3872687529988803, "grad_norm": 2.5127878189086914, "learning_rate": 4.796e-06, "loss": 0.0256, "step": 26021 }, { "epoch": 1.3873220664285333, "grad_norm": 1.4038714170455933, "learning_rate": 4.795800000000001e-06, "loss": 0.0292, "step": 26022 }, { "epoch": 1.3873753798581863, "grad_norm": 1.637432336807251, "learning_rate": 4.7956e-06, "loss": 0.0095, "step": 26023 }, { "epoch": 1.3874286932878392, "grad_norm": 0.5485535860061646, "learning_rate": 4.795400000000001e-06, "loss": 0.005, "step": 26024 }, { "epoch": 1.3874820067174922, "grad_norm": 1.2204878330230713, "learning_rate": 4.7952e-06, "loss": 0.0211, "step": 26025 }, { "epoch": 1.387535320147145, "grad_norm": 1.218612790107727, "learning_rate": 4.795e-06, "loss": 0.0169, "step": 26026 }, { "epoch": 1.387588633576798, "grad_norm": 1.430172324180603, "learning_rate": 4.7948e-06, "loss": 0.03, "step": 26027 }, { "epoch": 1.387641947006451, "grad_norm": 1.706844687461853, "learning_rate": 4.794600000000001e-06, "loss": 0.0064, "step": 26028 }, { "epoch": 1.387695260436104, "grad_norm": 1.6114418506622314, "learning_rate": 4.7944e-06, "loss": 0.033, "step": 26029 }, { "epoch": 1.387748573865757, "grad_norm": 0.5801518559455872, "learning_rate": 4.794200000000001e-06, "loss": 0.0049, "step": 26030 }, { "epoch": 1.3878018872954097, "grad_norm": 1.8822753429412842, "learning_rate": 4.794e-06, "loss": 0.0295, "step": 26031 }, { "epoch": 1.3878552007250626, "grad_norm": 1.5048540830612183, "learning_rate": 4.7938e-06, "loss": 0.0273, "step": 26032 }, { "epoch": 1.3879085141547156, "grad_norm": 2.5073487758636475, "learning_rate": 4.7936e-06, "loss": 0.043, "step": 26033 }, { "epoch": 1.3879618275843684, "grad_norm": 2.9296298027038574, "learning_rate": 4.7934000000000006e-06, "loss": 0.0255, "step": 26034 }, { "epoch": 1.3880151410140216, "grad_norm": 1.216858983039856, "learning_rate": 4.793200000000001e-06, "loss": 0.01, "step": 26035 }, { "epoch": 1.3880684544436743, "grad_norm": 2.2998054027557373, "learning_rate": 4.7930000000000005e-06, "loss": 0.0406, "step": 26036 }, { "epoch": 1.3881217678733273, "grad_norm": 1.6231333017349243, "learning_rate": 4.7928e-06, "loss": 0.015, "step": 26037 }, { "epoch": 1.3881750813029803, "grad_norm": 2.4860517978668213, "learning_rate": 4.7926000000000005e-06, "loss": 0.0221, "step": 26038 }, { "epoch": 1.388228394732633, "grad_norm": 1.8530267477035522, "learning_rate": 4.7924e-06, "loss": 0.0311, "step": 26039 }, { "epoch": 1.388281708162286, "grad_norm": 1.151472568511963, "learning_rate": 4.7922000000000005e-06, "loss": 0.0079, "step": 26040 }, { "epoch": 1.388335021591939, "grad_norm": 2.0799243450164795, "learning_rate": 4.792000000000001e-06, "loss": 0.0243, "step": 26041 }, { "epoch": 1.388388335021592, "grad_norm": 1.8107173442840576, "learning_rate": 4.7918000000000004e-06, "loss": 0.0167, "step": 26042 }, { "epoch": 1.388441648451245, "grad_norm": 0.961388349533081, "learning_rate": 4.7916e-06, "loss": 0.012, "step": 26043 }, { "epoch": 1.3884949618808977, "grad_norm": 0.6262496113777161, "learning_rate": 4.7914e-06, "loss": 0.0127, "step": 26044 }, { "epoch": 1.3885482753105507, "grad_norm": 1.15895676612854, "learning_rate": 4.7912e-06, "loss": 0.0186, "step": 26045 }, { "epoch": 1.3886015887402037, "grad_norm": 1.1508570909500122, "learning_rate": 4.791e-06, "loss": 0.0096, "step": 26046 }, { "epoch": 1.3886549021698567, "grad_norm": 1.4468605518341064, "learning_rate": 4.790800000000001e-06, "loss": 0.0164, "step": 26047 }, { "epoch": 1.3887082155995096, "grad_norm": 2.2576377391815186, "learning_rate": 4.7906e-06, "loss": 0.0272, "step": 26048 }, { "epoch": 1.3887615290291624, "grad_norm": 1.4739645719528198, "learning_rate": 4.790400000000001e-06, "loss": 0.0101, "step": 26049 }, { "epoch": 1.3888148424588154, "grad_norm": 1.1886394023895264, "learning_rate": 4.7902e-06, "loss": 0.0194, "step": 26050 }, { "epoch": 1.3888681558884683, "grad_norm": 1.4555720090866089, "learning_rate": 4.79e-06, "loss": 0.0093, "step": 26051 }, { "epoch": 1.388921469318121, "grad_norm": 1.917994499206543, "learning_rate": 4.7898e-06, "loss": 0.0313, "step": 26052 }, { "epoch": 1.388974782747774, "grad_norm": 1.5358102321624756, "learning_rate": 4.789600000000001e-06, "loss": 0.0234, "step": 26053 }, { "epoch": 1.389028096177427, "grad_norm": 2.3517181873321533, "learning_rate": 4.7894e-06, "loss": 0.0119, "step": 26054 }, { "epoch": 1.38908140960708, "grad_norm": 0.82254558801651, "learning_rate": 4.789200000000001e-06, "loss": 0.0067, "step": 26055 }, { "epoch": 1.389134723036733, "grad_norm": 1.833451271057129, "learning_rate": 4.789e-06, "loss": 0.0293, "step": 26056 }, { "epoch": 1.3891880364663858, "grad_norm": 2.1895499229431152, "learning_rate": 4.7888e-06, "loss": 0.0375, "step": 26057 }, { "epoch": 1.3892413498960388, "grad_norm": 1.0343348979949951, "learning_rate": 4.7886e-06, "loss": 0.0068, "step": 26058 }, { "epoch": 1.3892946633256917, "grad_norm": 1.362552285194397, "learning_rate": 4.788400000000001e-06, "loss": 0.0152, "step": 26059 }, { "epoch": 1.3893479767553447, "grad_norm": 2.344952344894409, "learning_rate": 4.788200000000001e-06, "loss": 0.0325, "step": 26060 }, { "epoch": 1.3894012901849977, "grad_norm": 0.3291350305080414, "learning_rate": 4.7880000000000006e-06, "loss": 0.0028, "step": 26061 }, { "epoch": 1.3894546036146505, "grad_norm": 2.420376777648926, "learning_rate": 4.7878e-06, "loss": 0.0356, "step": 26062 }, { "epoch": 1.3895079170443034, "grad_norm": 1.9391015768051147, "learning_rate": 4.7876000000000005e-06, "loss": 0.0264, "step": 26063 }, { "epoch": 1.3895612304739564, "grad_norm": 2.299762725830078, "learning_rate": 4.7874e-06, "loss": 0.0272, "step": 26064 }, { "epoch": 1.3896145439036094, "grad_norm": 3.1635797023773193, "learning_rate": 4.7872000000000005e-06, "loss": 0.069, "step": 26065 }, { "epoch": 1.3896678573332624, "grad_norm": 1.8204777240753174, "learning_rate": 4.787000000000001e-06, "loss": 0.0204, "step": 26066 }, { "epoch": 1.3897211707629151, "grad_norm": 0.828819751739502, "learning_rate": 4.7868000000000005e-06, "loss": 0.0059, "step": 26067 }, { "epoch": 1.389774484192568, "grad_norm": 2.993138074874878, "learning_rate": 4.7866e-06, "loss": 0.0154, "step": 26068 }, { "epoch": 1.389827797622221, "grad_norm": 3.7287256717681885, "learning_rate": 4.7864000000000005e-06, "loss": 0.0249, "step": 26069 }, { "epoch": 1.3898811110518738, "grad_norm": 3.094318151473999, "learning_rate": 4.7862e-06, "loss": 0.0327, "step": 26070 }, { "epoch": 1.3899344244815268, "grad_norm": 3.0649142265319824, "learning_rate": 4.7860000000000004e-06, "loss": 0.0374, "step": 26071 }, { "epoch": 1.3899877379111798, "grad_norm": 3.2108280658721924, "learning_rate": 4.785800000000001e-06, "loss": 0.0218, "step": 26072 }, { "epoch": 1.3900410513408328, "grad_norm": 1.1756031513214111, "learning_rate": 4.7856e-06, "loss": 0.01, "step": 26073 }, { "epoch": 1.3900943647704858, "grad_norm": 0.776531994342804, "learning_rate": 4.785400000000001e-06, "loss": 0.0055, "step": 26074 }, { "epoch": 1.3901476782001385, "grad_norm": 1.7549573183059692, "learning_rate": 4.7852e-06, "loss": 0.0253, "step": 26075 }, { "epoch": 1.3902009916297915, "grad_norm": 0.5109296441078186, "learning_rate": 4.785e-06, "loss": 0.003, "step": 26076 }, { "epoch": 1.3902543050594445, "grad_norm": 1.5918796062469482, "learning_rate": 4.7848e-06, "loss": 0.0233, "step": 26077 }, { "epoch": 1.3903076184890975, "grad_norm": 1.2664799690246582, "learning_rate": 4.784600000000001e-06, "loss": 0.0157, "step": 26078 }, { "epoch": 1.3903609319187504, "grad_norm": 0.9405783414840698, "learning_rate": 4.7844e-06, "loss": 0.0097, "step": 26079 }, { "epoch": 1.3904142453484032, "grad_norm": 2.3461639881134033, "learning_rate": 4.784200000000001e-06, "loss": 0.0083, "step": 26080 }, { "epoch": 1.3904675587780562, "grad_norm": 2.6413965225219727, "learning_rate": 4.784e-06, "loss": 0.0148, "step": 26081 }, { "epoch": 1.3905208722077091, "grad_norm": 2.0679280757904053, "learning_rate": 4.783800000000001e-06, "loss": 0.0285, "step": 26082 }, { "epoch": 1.3905741856373621, "grad_norm": 1.1572599411010742, "learning_rate": 4.7836e-06, "loss": 0.0188, "step": 26083 }, { "epoch": 1.390627499067015, "grad_norm": 2.3951492309570312, "learning_rate": 4.783400000000001e-06, "loss": 0.0402, "step": 26084 }, { "epoch": 1.3906808124966679, "grad_norm": 3.8952319622039795, "learning_rate": 4.7832e-06, "loss": 0.0387, "step": 26085 }, { "epoch": 1.3907341259263208, "grad_norm": 3.182439088821411, "learning_rate": 4.783000000000001e-06, "loss": 0.0763, "step": 26086 }, { "epoch": 1.3907874393559738, "grad_norm": 1.3055195808410645, "learning_rate": 4.7828e-06, "loss": 0.0082, "step": 26087 }, { "epoch": 1.3908407527856266, "grad_norm": 0.9449487328529358, "learning_rate": 4.782600000000001e-06, "loss": 0.0264, "step": 26088 }, { "epoch": 1.3908940662152796, "grad_norm": 0.9105178713798523, "learning_rate": 4.7824e-06, "loss": 0.0044, "step": 26089 }, { "epoch": 1.3909473796449325, "grad_norm": 0.7527629733085632, "learning_rate": 4.7822000000000006e-06, "loss": 0.0161, "step": 26090 }, { "epoch": 1.3910006930745855, "grad_norm": 0.227593332529068, "learning_rate": 4.782e-06, "loss": 0.0021, "step": 26091 }, { "epoch": 1.3910540065042385, "grad_norm": 1.585409164428711, "learning_rate": 4.7818000000000005e-06, "loss": 0.0094, "step": 26092 }, { "epoch": 1.3911073199338913, "grad_norm": 1.3621975183486938, "learning_rate": 4.7816e-06, "loss": 0.0192, "step": 26093 }, { "epoch": 1.3911606333635442, "grad_norm": 1.0212526321411133, "learning_rate": 4.7814000000000005e-06, "loss": 0.0081, "step": 26094 }, { "epoch": 1.3912139467931972, "grad_norm": 1.6342321634292603, "learning_rate": 4.7812e-06, "loss": 0.0419, "step": 26095 }, { "epoch": 1.3912672602228502, "grad_norm": 1.587707757949829, "learning_rate": 4.7810000000000005e-06, "loss": 0.009, "step": 26096 }, { "epoch": 1.3913205736525032, "grad_norm": 1.3437498807907104, "learning_rate": 4.7808e-06, "loss": 0.0146, "step": 26097 }, { "epoch": 1.391373887082156, "grad_norm": 3.897333860397339, "learning_rate": 4.7806000000000004e-06, "loss": 0.0191, "step": 26098 }, { "epoch": 1.391427200511809, "grad_norm": 1.9633512496948242, "learning_rate": 4.780400000000001e-06, "loss": 0.0105, "step": 26099 }, { "epoch": 1.3914805139414619, "grad_norm": 1.3968926668167114, "learning_rate": 4.7802e-06, "loss": 0.0111, "step": 26100 }, { "epoch": 1.3915338273711149, "grad_norm": 2.1820144653320312, "learning_rate": 4.78e-06, "loss": 0.0195, "step": 26101 }, { "epoch": 1.3915871408007678, "grad_norm": 0.5090839862823486, "learning_rate": 4.7798e-06, "loss": 0.0032, "step": 26102 }, { "epoch": 1.3916404542304206, "grad_norm": 1.9117567539215088, "learning_rate": 4.7796e-06, "loss": 0.0128, "step": 26103 }, { "epoch": 1.3916937676600736, "grad_norm": 3.025181531906128, "learning_rate": 4.7794e-06, "loss": 0.0266, "step": 26104 }, { "epoch": 1.3917470810897266, "grad_norm": 0.23592859506607056, "learning_rate": 4.779200000000001e-06, "loss": 0.0021, "step": 26105 }, { "epoch": 1.3918003945193793, "grad_norm": 1.12367582321167, "learning_rate": 4.779e-06, "loss": 0.0152, "step": 26106 }, { "epoch": 1.3918537079490323, "grad_norm": 0.4838033616542816, "learning_rate": 4.778800000000001e-06, "loss": 0.0039, "step": 26107 }, { "epoch": 1.3919070213786853, "grad_norm": 0.40825456380844116, "learning_rate": 4.7786e-06, "loss": 0.0035, "step": 26108 }, { "epoch": 1.3919603348083383, "grad_norm": 1.6963016986846924, "learning_rate": 4.7784e-06, "loss": 0.0277, "step": 26109 }, { "epoch": 1.3920136482379912, "grad_norm": 0.9371852278709412, "learning_rate": 4.7782e-06, "loss": 0.0117, "step": 26110 }, { "epoch": 1.392066961667644, "grad_norm": 2.795055866241455, "learning_rate": 4.778000000000001e-06, "loss": 0.0368, "step": 26111 }, { "epoch": 1.392120275097297, "grad_norm": 0.43643370270729065, "learning_rate": 4.7778e-06, "loss": 0.0026, "step": 26112 }, { "epoch": 1.39217358852695, "grad_norm": 2.004777431488037, "learning_rate": 4.777600000000001e-06, "loss": 0.0262, "step": 26113 }, { "epoch": 1.392226901956603, "grad_norm": 2.6958024501800537, "learning_rate": 4.7774e-06, "loss": 0.0315, "step": 26114 }, { "epoch": 1.392280215386256, "grad_norm": 1.9505882263183594, "learning_rate": 4.7772e-06, "loss": 0.0271, "step": 26115 }, { "epoch": 1.3923335288159087, "grad_norm": 1.359958529472351, "learning_rate": 4.777e-06, "loss": 0.0072, "step": 26116 }, { "epoch": 1.3923868422455616, "grad_norm": 1.9425139427185059, "learning_rate": 4.776800000000001e-06, "loss": 0.0441, "step": 26117 }, { "epoch": 1.3924401556752146, "grad_norm": 1.7705827951431274, "learning_rate": 4.7766e-06, "loss": 0.014, "step": 26118 }, { "epoch": 1.3924934691048676, "grad_norm": 2.715050458908081, "learning_rate": 4.7764000000000006e-06, "loss": 0.018, "step": 26119 }, { "epoch": 1.3925467825345206, "grad_norm": 4.410192489624023, "learning_rate": 4.7762e-06, "loss": 0.0447, "step": 26120 }, { "epoch": 1.3926000959641733, "grad_norm": 3.9610161781311035, "learning_rate": 4.7760000000000005e-06, "loss": 0.0188, "step": 26121 }, { "epoch": 1.3926534093938263, "grad_norm": 1.419327974319458, "learning_rate": 4.7758e-06, "loss": 0.0051, "step": 26122 }, { "epoch": 1.3927067228234793, "grad_norm": 3.19600772857666, "learning_rate": 4.7756000000000005e-06, "loss": 0.0223, "step": 26123 }, { "epoch": 1.392760036253132, "grad_norm": 0.3868352770805359, "learning_rate": 4.775400000000001e-06, "loss": 0.0023, "step": 26124 }, { "epoch": 1.392813349682785, "grad_norm": 1.6304540634155273, "learning_rate": 4.7752000000000005e-06, "loss": 0.0339, "step": 26125 }, { "epoch": 1.392866663112438, "grad_norm": 1.101580023765564, "learning_rate": 4.775e-06, "loss": 0.0043, "step": 26126 }, { "epoch": 1.392919976542091, "grad_norm": 1.542560338973999, "learning_rate": 4.7748000000000004e-06, "loss": 0.0186, "step": 26127 }, { "epoch": 1.392973289971744, "grad_norm": 0.9459385871887207, "learning_rate": 4.7746e-06, "loss": 0.0126, "step": 26128 }, { "epoch": 1.3930266034013967, "grad_norm": 1.290574073791504, "learning_rate": 4.7744e-06, "loss": 0.0127, "step": 26129 }, { "epoch": 1.3930799168310497, "grad_norm": 1.9399328231811523, "learning_rate": 4.774200000000001e-06, "loss": 0.0232, "step": 26130 }, { "epoch": 1.3931332302607027, "grad_norm": 1.807220697402954, "learning_rate": 4.774e-06, "loss": 0.0296, "step": 26131 }, { "epoch": 1.3931865436903557, "grad_norm": 2.1586973667144775, "learning_rate": 4.773800000000001e-06, "loss": 0.01, "step": 26132 }, { "epoch": 1.3932398571200086, "grad_norm": 0.9660349488258362, "learning_rate": 4.7736e-06, "loss": 0.0063, "step": 26133 }, { "epoch": 1.3932931705496614, "grad_norm": 2.4543747901916504, "learning_rate": 4.7734e-06, "loss": 0.0401, "step": 26134 }, { "epoch": 1.3933464839793144, "grad_norm": 2.376171827316284, "learning_rate": 4.7732e-06, "loss": 0.0405, "step": 26135 }, { "epoch": 1.3933997974089674, "grad_norm": 1.9570789337158203, "learning_rate": 4.773000000000001e-06, "loss": 0.0097, "step": 26136 }, { "epoch": 1.3934531108386201, "grad_norm": 1.746930480003357, "learning_rate": 4.7728e-06, "loss": 0.0191, "step": 26137 }, { "epoch": 1.3935064242682733, "grad_norm": 1.3153600692749023, "learning_rate": 4.772600000000001e-06, "loss": 0.0164, "step": 26138 }, { "epoch": 1.393559737697926, "grad_norm": 1.7192779779434204, "learning_rate": 4.7724e-06, "loss": 0.0135, "step": 26139 }, { "epoch": 1.393613051127579, "grad_norm": 2.823394536972046, "learning_rate": 4.7722e-06, "loss": 0.0116, "step": 26140 }, { "epoch": 1.393666364557232, "grad_norm": 4.069860458374023, "learning_rate": 4.772e-06, "loss": 0.0282, "step": 26141 }, { "epoch": 1.3937196779868848, "grad_norm": 2.8418548107147217, "learning_rate": 4.771800000000001e-06, "loss": 0.0222, "step": 26142 }, { "epoch": 1.3937729914165378, "grad_norm": 1.4568493366241455, "learning_rate": 4.7716e-06, "loss": 0.0197, "step": 26143 }, { "epoch": 1.3938263048461907, "grad_norm": 2.3631703853607178, "learning_rate": 4.771400000000001e-06, "loss": 0.0368, "step": 26144 }, { "epoch": 1.3938796182758437, "grad_norm": 2.1999216079711914, "learning_rate": 4.7712e-06, "loss": 0.0249, "step": 26145 }, { "epoch": 1.3939329317054967, "grad_norm": 1.377944827079773, "learning_rate": 4.7710000000000006e-06, "loss": 0.0143, "step": 26146 }, { "epoch": 1.3939862451351495, "grad_norm": 2.282554864883423, "learning_rate": 4.7708e-06, "loss": 0.0163, "step": 26147 }, { "epoch": 1.3940395585648024, "grad_norm": 0.5129877328872681, "learning_rate": 4.7706000000000005e-06, "loss": 0.0056, "step": 26148 }, { "epoch": 1.3940928719944554, "grad_norm": 1.4529359340667725, "learning_rate": 4.770400000000001e-06, "loss": 0.0182, "step": 26149 }, { "epoch": 1.3941461854241084, "grad_norm": 1.8273735046386719, "learning_rate": 4.7702000000000005e-06, "loss": 0.0299, "step": 26150 }, { "epoch": 1.3941994988537614, "grad_norm": 2.4096782207489014, "learning_rate": 4.77e-06, "loss": 0.0193, "step": 26151 }, { "epoch": 1.3942528122834141, "grad_norm": 2.233069896697998, "learning_rate": 4.7698000000000005e-06, "loss": 0.0366, "step": 26152 }, { "epoch": 1.3943061257130671, "grad_norm": 1.8701881170272827, "learning_rate": 4.7696e-06, "loss": 0.0217, "step": 26153 }, { "epoch": 1.39435943914272, "grad_norm": 1.6132303476333618, "learning_rate": 4.7694000000000004e-06, "loss": 0.0377, "step": 26154 }, { "epoch": 1.3944127525723728, "grad_norm": 3.4217581748962402, "learning_rate": 4.769200000000001e-06, "loss": 0.0258, "step": 26155 }, { "epoch": 1.3944660660020258, "grad_norm": 2.393620729446411, "learning_rate": 4.769e-06, "loss": 0.021, "step": 26156 }, { "epoch": 1.3945193794316788, "grad_norm": 1.519987940788269, "learning_rate": 4.768800000000001e-06, "loss": 0.0091, "step": 26157 }, { "epoch": 1.3945726928613318, "grad_norm": 2.44313383102417, "learning_rate": 4.7686e-06, "loss": 0.0332, "step": 26158 }, { "epoch": 1.3946260062909848, "grad_norm": 2.4092729091644287, "learning_rate": 4.7684e-06, "loss": 0.0198, "step": 26159 }, { "epoch": 1.3946793197206375, "grad_norm": 1.4141302108764648, "learning_rate": 4.7682e-06, "loss": 0.0317, "step": 26160 }, { "epoch": 1.3947326331502905, "grad_norm": 4.719844818115234, "learning_rate": 4.768000000000001e-06, "loss": 0.044, "step": 26161 }, { "epoch": 1.3947859465799435, "grad_norm": 0.5527552366256714, "learning_rate": 4.7678e-06, "loss": 0.0046, "step": 26162 }, { "epoch": 1.3948392600095965, "grad_norm": 1.6041244268417358, "learning_rate": 4.767600000000001e-06, "loss": 0.0198, "step": 26163 }, { "epoch": 1.3948925734392494, "grad_norm": 1.9384758472442627, "learning_rate": 4.7674e-06, "loss": 0.0321, "step": 26164 }, { "epoch": 1.3949458868689022, "grad_norm": 1.6233274936676025, "learning_rate": 4.7672e-06, "loss": 0.0184, "step": 26165 }, { "epoch": 1.3949992002985552, "grad_norm": 2.87154483795166, "learning_rate": 4.767e-06, "loss": 0.0323, "step": 26166 }, { "epoch": 1.3950525137282082, "grad_norm": 0.9601112008094788, "learning_rate": 4.766800000000001e-06, "loss": 0.0068, "step": 26167 }, { "epoch": 1.3951058271578611, "grad_norm": 1.000484585762024, "learning_rate": 4.7666e-06, "loss": 0.0287, "step": 26168 }, { "epoch": 1.3951591405875141, "grad_norm": 1.8168220520019531, "learning_rate": 4.766400000000001e-06, "loss": 0.0194, "step": 26169 }, { "epoch": 1.3952124540171669, "grad_norm": 0.9060620069503784, "learning_rate": 4.7662e-06, "loss": 0.0064, "step": 26170 }, { "epoch": 1.3952657674468198, "grad_norm": 2.067121982574463, "learning_rate": 4.766000000000001e-06, "loss": 0.0252, "step": 26171 }, { "epoch": 1.3953190808764728, "grad_norm": 3.278653621673584, "learning_rate": 4.7658e-06, "loss": 0.0209, "step": 26172 }, { "epoch": 1.3953723943061256, "grad_norm": 2.4651145935058594, "learning_rate": 4.765600000000001e-06, "loss": 0.035, "step": 26173 }, { "epoch": 1.3954257077357786, "grad_norm": 1.077256202697754, "learning_rate": 4.765400000000001e-06, "loss": 0.0079, "step": 26174 }, { "epoch": 1.3954790211654315, "grad_norm": 0.3443780839443207, "learning_rate": 4.7652000000000006e-06, "loss": 0.0027, "step": 26175 }, { "epoch": 1.3955323345950845, "grad_norm": 0.9711325764656067, "learning_rate": 4.765e-06, "loss": 0.0071, "step": 26176 }, { "epoch": 1.3955856480247375, "grad_norm": 2.9196078777313232, "learning_rate": 4.7648000000000005e-06, "loss": 0.0139, "step": 26177 }, { "epoch": 1.3956389614543903, "grad_norm": 1.9424476623535156, "learning_rate": 4.7646e-06, "loss": 0.0133, "step": 26178 }, { "epoch": 1.3956922748840432, "grad_norm": 1.4360405206680298, "learning_rate": 4.7644000000000005e-06, "loss": 0.0092, "step": 26179 }, { "epoch": 1.3957455883136962, "grad_norm": 2.4837260246276855, "learning_rate": 4.764200000000001e-06, "loss": 0.038, "step": 26180 }, { "epoch": 1.3957989017433492, "grad_norm": 1.5870556831359863, "learning_rate": 4.7640000000000005e-06, "loss": 0.0278, "step": 26181 }, { "epoch": 1.3958522151730022, "grad_norm": 3.3664710521698, "learning_rate": 4.763800000000001e-06, "loss": 0.0177, "step": 26182 }, { "epoch": 1.395905528602655, "grad_norm": 3.169023036956787, "learning_rate": 4.7636000000000004e-06, "loss": 0.0184, "step": 26183 }, { "epoch": 1.395958842032308, "grad_norm": 1.801687240600586, "learning_rate": 4.7634e-06, "loss": 0.0104, "step": 26184 }, { "epoch": 1.396012155461961, "grad_norm": 0.2657039761543274, "learning_rate": 4.7632e-06, "loss": 0.0025, "step": 26185 }, { "epoch": 1.3960654688916139, "grad_norm": 0.285836398601532, "learning_rate": 4.763000000000001e-06, "loss": 0.0032, "step": 26186 }, { "epoch": 1.3961187823212668, "grad_norm": 1.7829622030258179, "learning_rate": 4.7628e-06, "loss": 0.0402, "step": 26187 }, { "epoch": 1.3961720957509196, "grad_norm": 1.51905357837677, "learning_rate": 4.762600000000001e-06, "loss": 0.0172, "step": 26188 }, { "epoch": 1.3962254091805726, "grad_norm": 0.44283029437065125, "learning_rate": 4.7624e-06, "loss": 0.003, "step": 26189 }, { "epoch": 1.3962787226102256, "grad_norm": 2.7220518589019775, "learning_rate": 4.7622e-06, "loss": 0.0338, "step": 26190 }, { "epoch": 1.3963320360398783, "grad_norm": 1.2692444324493408, "learning_rate": 4.762e-06, "loss": 0.0188, "step": 26191 }, { "epoch": 1.3963853494695313, "grad_norm": 1.331634283065796, "learning_rate": 4.761800000000001e-06, "loss": 0.0367, "step": 26192 }, { "epoch": 1.3964386628991843, "grad_norm": 0.5112929940223694, "learning_rate": 4.7616e-06, "loss": 0.0047, "step": 26193 }, { "epoch": 1.3964919763288373, "grad_norm": 1.9610421657562256, "learning_rate": 4.761400000000001e-06, "loss": 0.0469, "step": 26194 }, { "epoch": 1.3965452897584902, "grad_norm": 2.0867972373962402, "learning_rate": 4.7612e-06, "loss": 0.0231, "step": 26195 }, { "epoch": 1.396598603188143, "grad_norm": 1.1047401428222656, "learning_rate": 4.761000000000001e-06, "loss": 0.0212, "step": 26196 }, { "epoch": 1.396651916617796, "grad_norm": 2.0128366947174072, "learning_rate": 4.7608e-06, "loss": 0.0417, "step": 26197 }, { "epoch": 1.396705230047449, "grad_norm": 1.0220986604690552, "learning_rate": 4.760600000000001e-06, "loss": 0.0071, "step": 26198 }, { "epoch": 1.396758543477102, "grad_norm": 1.966443657875061, "learning_rate": 4.7604e-06, "loss": 0.0084, "step": 26199 }, { "epoch": 1.396811856906755, "grad_norm": 1.9843175411224365, "learning_rate": 4.760200000000001e-06, "loss": 0.0303, "step": 26200 }, { "epoch": 1.3968651703364077, "grad_norm": 2.0213263034820557, "learning_rate": 4.76e-06, "loss": 0.0271, "step": 26201 }, { "epoch": 1.3969184837660606, "grad_norm": 1.1870158910751343, "learning_rate": 4.7598000000000006e-06, "loss": 0.0095, "step": 26202 }, { "epoch": 1.3969717971957136, "grad_norm": 1.8757517337799072, "learning_rate": 4.7596e-06, "loss": 0.0101, "step": 26203 }, { "epoch": 1.3970251106253666, "grad_norm": 1.3106820583343506, "learning_rate": 4.7594000000000005e-06, "loss": 0.0189, "step": 26204 }, { "epoch": 1.3970784240550196, "grad_norm": 0.46984025835990906, "learning_rate": 4.7592e-06, "loss": 0.0037, "step": 26205 }, { "epoch": 1.3971317374846723, "grad_norm": 0.9594415426254272, "learning_rate": 4.7590000000000005e-06, "loss": 0.0066, "step": 26206 }, { "epoch": 1.3971850509143253, "grad_norm": 1.999092698097229, "learning_rate": 4.758800000000001e-06, "loss": 0.0103, "step": 26207 }, { "epoch": 1.3972383643439783, "grad_norm": 4.125659465789795, "learning_rate": 4.7586000000000005e-06, "loss": 0.0327, "step": 26208 }, { "epoch": 1.397291677773631, "grad_norm": 3.480579376220703, "learning_rate": 4.7584e-06, "loss": 0.0264, "step": 26209 }, { "epoch": 1.397344991203284, "grad_norm": 2.862919569015503, "learning_rate": 4.7582000000000005e-06, "loss": 0.0255, "step": 26210 }, { "epoch": 1.397398304632937, "grad_norm": 2.430302143096924, "learning_rate": 4.758e-06, "loss": 0.0246, "step": 26211 }, { "epoch": 1.39745161806259, "grad_norm": 1.2368396520614624, "learning_rate": 4.7578000000000004e-06, "loss": 0.0235, "step": 26212 }, { "epoch": 1.397504931492243, "grad_norm": 2.1373724937438965, "learning_rate": 4.757600000000001e-06, "loss": 0.0449, "step": 26213 }, { "epoch": 1.3975582449218957, "grad_norm": 1.9586656093597412, "learning_rate": 4.7574e-06, "loss": 0.02, "step": 26214 }, { "epoch": 1.3976115583515487, "grad_norm": 3.217473030090332, "learning_rate": 4.7572e-06, "loss": 0.0321, "step": 26215 }, { "epoch": 1.3976648717812017, "grad_norm": 0.8732821941375732, "learning_rate": 4.757e-06, "loss": 0.0143, "step": 26216 }, { "epoch": 1.3977181852108547, "grad_norm": 0.7515830993652344, "learning_rate": 4.7568e-06, "loss": 0.022, "step": 26217 }, { "epoch": 1.3977714986405076, "grad_norm": 0.46589136123657227, "learning_rate": 4.7566e-06, "loss": 0.0044, "step": 26218 }, { "epoch": 1.3978248120701604, "grad_norm": 3.8999714851379395, "learning_rate": 4.756400000000001e-06, "loss": 0.0311, "step": 26219 }, { "epoch": 1.3978781254998134, "grad_norm": 2.586350202560425, "learning_rate": 4.7562e-06, "loss": 0.0306, "step": 26220 }, { "epoch": 1.3979314389294664, "grad_norm": 2.8187572956085205, "learning_rate": 4.756000000000001e-06, "loss": 0.051, "step": 26221 }, { "epoch": 1.3979847523591193, "grad_norm": 2.506075859069824, "learning_rate": 4.7558e-06, "loss": 0.0692, "step": 26222 }, { "epoch": 1.3980380657887723, "grad_norm": 1.613876223564148, "learning_rate": 4.7556e-06, "loss": 0.0352, "step": 26223 }, { "epoch": 1.398091379218425, "grad_norm": 1.8934133052825928, "learning_rate": 4.7554e-06, "loss": 0.0247, "step": 26224 }, { "epoch": 1.398144692648078, "grad_norm": 0.7424226999282837, "learning_rate": 4.755200000000001e-06, "loss": 0.0053, "step": 26225 }, { "epoch": 1.398198006077731, "grad_norm": 2.097050428390503, "learning_rate": 4.755e-06, "loss": 0.0262, "step": 26226 }, { "epoch": 1.3982513195073838, "grad_norm": 1.0190043449401855, "learning_rate": 4.754800000000001e-06, "loss": 0.01, "step": 26227 }, { "epoch": 1.3983046329370368, "grad_norm": 0.6658003330230713, "learning_rate": 4.7546e-06, "loss": 0.0058, "step": 26228 }, { "epoch": 1.3983579463666898, "grad_norm": 4.1610870361328125, "learning_rate": 4.7544e-06, "loss": 0.0392, "step": 26229 }, { "epoch": 1.3984112597963427, "grad_norm": 1.9718825817108154, "learning_rate": 4.7542e-06, "loss": 0.0301, "step": 26230 }, { "epoch": 1.3984645732259957, "grad_norm": 3.0585219860076904, "learning_rate": 4.7540000000000006e-06, "loss": 0.0283, "step": 26231 }, { "epoch": 1.3985178866556485, "grad_norm": 3.3348660469055176, "learning_rate": 4.753800000000001e-06, "loss": 0.0231, "step": 26232 }, { "epoch": 1.3985712000853014, "grad_norm": 1.063068151473999, "learning_rate": 4.7536000000000005e-06, "loss": 0.0056, "step": 26233 }, { "epoch": 1.3986245135149544, "grad_norm": 1.4814049005508423, "learning_rate": 4.7534e-06, "loss": 0.0277, "step": 26234 }, { "epoch": 1.3986778269446074, "grad_norm": 1.7908860445022583, "learning_rate": 4.7532000000000005e-06, "loss": 0.0208, "step": 26235 }, { "epoch": 1.3987311403742604, "grad_norm": 2.397040367126465, "learning_rate": 4.753e-06, "loss": 0.0129, "step": 26236 }, { "epoch": 1.3987844538039131, "grad_norm": 1.200396180152893, "learning_rate": 4.7528000000000005e-06, "loss": 0.0158, "step": 26237 }, { "epoch": 1.3988377672335661, "grad_norm": 0.6582721471786499, "learning_rate": 4.752600000000001e-06, "loss": 0.0169, "step": 26238 }, { "epoch": 1.398891080663219, "grad_norm": 0.7442411780357361, "learning_rate": 4.7524000000000004e-06, "loss": 0.0081, "step": 26239 }, { "epoch": 1.3989443940928719, "grad_norm": 1.5045371055603027, "learning_rate": 4.7522e-06, "loss": 0.0196, "step": 26240 }, { "epoch": 1.398997707522525, "grad_norm": 1.662163496017456, "learning_rate": 4.752e-06, "loss": 0.0178, "step": 26241 }, { "epoch": 1.3990510209521778, "grad_norm": 2.5044686794281006, "learning_rate": 4.7518e-06, "loss": 0.0261, "step": 26242 }, { "epoch": 1.3991043343818308, "grad_norm": 3.8670425415039062, "learning_rate": 4.7516e-06, "loss": 0.0291, "step": 26243 }, { "epoch": 1.3991576478114838, "grad_norm": 1.5964744091033936, "learning_rate": 4.751400000000001e-06, "loss": 0.0232, "step": 26244 }, { "epoch": 1.3992109612411365, "grad_norm": 2.0395331382751465, "learning_rate": 4.7512e-06, "loss": 0.0505, "step": 26245 }, { "epoch": 1.3992642746707895, "grad_norm": 1.84042489528656, "learning_rate": 4.751000000000001e-06, "loss": 0.0418, "step": 26246 }, { "epoch": 1.3993175881004425, "grad_norm": 2.1905100345611572, "learning_rate": 4.7508e-06, "loss": 0.0353, "step": 26247 }, { "epoch": 1.3993709015300955, "grad_norm": 2.4264681339263916, "learning_rate": 4.7506e-06, "loss": 0.0234, "step": 26248 }, { "epoch": 1.3994242149597484, "grad_norm": 1.2284067869186401, "learning_rate": 4.7504e-06, "loss": 0.0226, "step": 26249 }, { "epoch": 1.3994775283894012, "grad_norm": 2.4129340648651123, "learning_rate": 4.750200000000001e-06, "loss": 0.0227, "step": 26250 }, { "epoch": 1.3995308418190542, "grad_norm": 0.6082140803337097, "learning_rate": 4.75e-06, "loss": 0.0165, "step": 26251 }, { "epoch": 1.3995841552487072, "grad_norm": 1.5464322566986084, "learning_rate": 4.749800000000001e-06, "loss": 0.0206, "step": 26252 }, { "epoch": 1.3996374686783601, "grad_norm": 3.202165126800537, "learning_rate": 4.7496e-06, "loss": 0.0406, "step": 26253 }, { "epoch": 1.3996907821080131, "grad_norm": 5.178317546844482, "learning_rate": 4.7494e-06, "loss": 0.0399, "step": 26254 }, { "epoch": 1.3997440955376659, "grad_norm": 0.48857590556144714, "learning_rate": 4.7492e-06, "loss": 0.0046, "step": 26255 }, { "epoch": 1.3997974089673189, "grad_norm": 2.243464469909668, "learning_rate": 4.749000000000001e-06, "loss": 0.0449, "step": 26256 }, { "epoch": 1.3998507223969718, "grad_norm": 3.3633594512939453, "learning_rate": 4.748800000000001e-06, "loss": 0.0418, "step": 26257 }, { "epoch": 1.3999040358266246, "grad_norm": 1.0428240299224854, "learning_rate": 4.748600000000001e-06, "loss": 0.0085, "step": 26258 }, { "epoch": 1.3999573492562778, "grad_norm": 1.9986642599105835, "learning_rate": 4.7484e-06, "loss": 0.0271, "step": 26259 }, { "epoch": 1.4000106626859306, "grad_norm": 2.3088390827178955, "learning_rate": 4.7482000000000005e-06, "loss": 0.0293, "step": 26260 }, { "epoch": 1.4000639761155835, "grad_norm": 1.934382677078247, "learning_rate": 4.748e-06, "loss": 0.0347, "step": 26261 }, { "epoch": 1.4001172895452365, "grad_norm": 4.552154541015625, "learning_rate": 4.7478000000000005e-06, "loss": 0.0348, "step": 26262 }, { "epoch": 1.4001706029748893, "grad_norm": 1.331701397895813, "learning_rate": 4.747600000000001e-06, "loss": 0.0172, "step": 26263 }, { "epoch": 1.4002239164045422, "grad_norm": 0.5238704681396484, "learning_rate": 4.7474000000000005e-06, "loss": 0.0056, "step": 26264 }, { "epoch": 1.4002772298341952, "grad_norm": 1.9294456243515015, "learning_rate": 4.7472e-06, "loss": 0.0301, "step": 26265 }, { "epoch": 1.4003305432638482, "grad_norm": 2.3673908710479736, "learning_rate": 4.7470000000000005e-06, "loss": 0.0381, "step": 26266 }, { "epoch": 1.4003838566935012, "grad_norm": 2.4568393230438232, "learning_rate": 4.7468e-06, "loss": 0.0445, "step": 26267 }, { "epoch": 1.400437170123154, "grad_norm": 1.6349998712539673, "learning_rate": 4.7466000000000004e-06, "loss": 0.0301, "step": 26268 }, { "epoch": 1.400490483552807, "grad_norm": 2.4853053092956543, "learning_rate": 4.746400000000001e-06, "loss": 0.0388, "step": 26269 }, { "epoch": 1.40054379698246, "grad_norm": 2.1113944053649902, "learning_rate": 4.7462e-06, "loss": 0.0278, "step": 26270 }, { "epoch": 1.4005971104121129, "grad_norm": 1.374203085899353, "learning_rate": 4.746000000000001e-06, "loss": 0.0138, "step": 26271 }, { "epoch": 1.4006504238417659, "grad_norm": 2.237635850906372, "learning_rate": 4.7458e-06, "loss": 0.0157, "step": 26272 }, { "epoch": 1.4007037372714186, "grad_norm": 2.91161847114563, "learning_rate": 4.7456e-06, "loss": 0.025, "step": 26273 }, { "epoch": 1.4007570507010716, "grad_norm": 1.334873914718628, "learning_rate": 4.7454e-06, "loss": 0.0116, "step": 26274 }, { "epoch": 1.4008103641307246, "grad_norm": 2.9450082778930664, "learning_rate": 4.745200000000001e-06, "loss": 0.0596, "step": 26275 }, { "epoch": 1.4008636775603773, "grad_norm": 2.3311691284179688, "learning_rate": 4.745e-06, "loss": 0.0424, "step": 26276 }, { "epoch": 1.4009169909900303, "grad_norm": 1.6035617589950562, "learning_rate": 4.744800000000001e-06, "loss": 0.0297, "step": 26277 }, { "epoch": 1.4009703044196833, "grad_norm": 3.14302396774292, "learning_rate": 4.7446e-06, "loss": 0.0594, "step": 26278 }, { "epoch": 1.4010236178493363, "grad_norm": 3.099364995956421, "learning_rate": 4.7444e-06, "loss": 0.0514, "step": 26279 }, { "epoch": 1.4010769312789892, "grad_norm": 2.5710244178771973, "learning_rate": 4.7442e-06, "loss": 0.0436, "step": 26280 }, { "epoch": 1.401130244708642, "grad_norm": 1.2295417785644531, "learning_rate": 4.744000000000001e-06, "loss": 0.0252, "step": 26281 }, { "epoch": 1.401183558138295, "grad_norm": 1.7635140419006348, "learning_rate": 4.7438e-06, "loss": 0.0393, "step": 26282 }, { "epoch": 1.401236871567948, "grad_norm": 2.095989465713501, "learning_rate": 4.743600000000001e-06, "loss": 0.0363, "step": 26283 }, { "epoch": 1.401290184997601, "grad_norm": 1.5395735502243042, "learning_rate": 4.7434e-06, "loss": 0.0273, "step": 26284 }, { "epoch": 1.401343498427254, "grad_norm": 1.2732480764389038, "learning_rate": 4.743200000000001e-06, "loss": 0.0159, "step": 26285 }, { "epoch": 1.4013968118569067, "grad_norm": 2.2112154960632324, "learning_rate": 4.743e-06, "loss": 0.0292, "step": 26286 }, { "epoch": 1.4014501252865597, "grad_norm": 1.736651062965393, "learning_rate": 4.7428000000000006e-06, "loss": 0.0409, "step": 26287 }, { "epoch": 1.4015034387162126, "grad_norm": 1.9308862686157227, "learning_rate": 4.7426e-06, "loss": 0.0389, "step": 26288 }, { "epoch": 1.4015567521458656, "grad_norm": 2.343796968460083, "learning_rate": 4.7424000000000005e-06, "loss": 0.0209, "step": 26289 }, { "epoch": 1.4016100655755186, "grad_norm": 3.3388431072235107, "learning_rate": 4.7422e-06, "loss": 0.0393, "step": 26290 }, { "epoch": 1.4016633790051714, "grad_norm": 1.9150313138961792, "learning_rate": 4.7420000000000005e-06, "loss": 0.0413, "step": 26291 }, { "epoch": 1.4017166924348243, "grad_norm": 2.2281432151794434, "learning_rate": 4.7418e-06, "loss": 0.0277, "step": 26292 }, { "epoch": 1.4017700058644773, "grad_norm": 1.6249483823776245, "learning_rate": 4.7416000000000005e-06, "loss": 0.0205, "step": 26293 }, { "epoch": 1.40182331929413, "grad_norm": 1.4584100246429443, "learning_rate": 4.7414e-06, "loss": 0.034, "step": 26294 }, { "epoch": 1.401876632723783, "grad_norm": 1.7479546070098877, "learning_rate": 4.7412000000000004e-06, "loss": 0.0235, "step": 26295 }, { "epoch": 1.401929946153436, "grad_norm": 0.7038383483886719, "learning_rate": 4.741000000000001e-06, "loss": 0.0102, "step": 26296 }, { "epoch": 1.401983259583089, "grad_norm": 0.7454308867454529, "learning_rate": 4.7408e-06, "loss": 0.0077, "step": 26297 }, { "epoch": 1.402036573012742, "grad_norm": 0.46389350295066833, "learning_rate": 4.7406e-06, "loss": 0.0053, "step": 26298 }, { "epoch": 1.4020898864423947, "grad_norm": 0.9691567420959473, "learning_rate": 4.7404e-06, "loss": 0.0122, "step": 26299 }, { "epoch": 1.4021431998720477, "grad_norm": 1.9346002340316772, "learning_rate": 4.7402e-06, "loss": 0.026, "step": 26300 }, { "epoch": 1.4021965133017007, "grad_norm": 2.870767831802368, "learning_rate": 4.74e-06, "loss": 0.0457, "step": 26301 }, { "epoch": 1.4022498267313537, "grad_norm": 0.7400273084640503, "learning_rate": 4.739800000000001e-06, "loss": 0.0113, "step": 26302 }, { "epoch": 1.4023031401610067, "grad_norm": 1.3384900093078613, "learning_rate": 4.7396e-06, "loss": 0.0204, "step": 26303 }, { "epoch": 1.4023564535906594, "grad_norm": 2.0500385761260986, "learning_rate": 4.7394e-06, "loss": 0.0463, "step": 26304 }, { "epoch": 1.4024097670203124, "grad_norm": 1.0393210649490356, "learning_rate": 4.7392e-06, "loss": 0.0158, "step": 26305 }, { "epoch": 1.4024630804499654, "grad_norm": 1.2290385961532593, "learning_rate": 4.739e-06, "loss": 0.0236, "step": 26306 }, { "epoch": 1.4025163938796184, "grad_norm": 0.9446607232093811, "learning_rate": 4.7388e-06, "loss": 0.0171, "step": 26307 }, { "epoch": 1.4025697073092713, "grad_norm": 1.3566328287124634, "learning_rate": 4.738600000000001e-06, "loss": 0.0319, "step": 26308 }, { "epoch": 1.402623020738924, "grad_norm": 1.8305777311325073, "learning_rate": 4.7384e-06, "loss": 0.031, "step": 26309 }, { "epoch": 1.402676334168577, "grad_norm": 1.5428948402404785, "learning_rate": 4.738200000000001e-06, "loss": 0.026, "step": 26310 }, { "epoch": 1.40272964759823, "grad_norm": 1.758257269859314, "learning_rate": 4.738e-06, "loss": 0.0294, "step": 26311 }, { "epoch": 1.4027829610278828, "grad_norm": 1.0753774642944336, "learning_rate": 4.7378e-06, "loss": 0.0239, "step": 26312 }, { "epoch": 1.4028362744575358, "grad_norm": 1.7680573463439941, "learning_rate": 4.7376e-06, "loss": 0.053, "step": 26313 }, { "epoch": 1.4028895878871888, "grad_norm": 2.460808277130127, "learning_rate": 4.737400000000001e-06, "loss": 0.0388, "step": 26314 }, { "epoch": 1.4029429013168417, "grad_norm": 1.768226981163025, "learning_rate": 4.7372e-06, "loss": 0.0404, "step": 26315 }, { "epoch": 1.4029962147464947, "grad_norm": 2.6526613235473633, "learning_rate": 4.7370000000000006e-06, "loss": 0.0302, "step": 26316 }, { "epoch": 1.4030495281761475, "grad_norm": 3.706995725631714, "learning_rate": 4.7368e-06, "loss": 0.0627, "step": 26317 }, { "epoch": 1.4031028416058005, "grad_norm": 0.8945595622062683, "learning_rate": 4.7366000000000005e-06, "loss": 0.0152, "step": 26318 }, { "epoch": 1.4031561550354534, "grad_norm": 0.9583426117897034, "learning_rate": 4.7364e-06, "loss": 0.0147, "step": 26319 }, { "epoch": 1.4032094684651064, "grad_norm": 0.960034966468811, "learning_rate": 4.7362000000000005e-06, "loss": 0.0168, "step": 26320 }, { "epoch": 1.4032627818947594, "grad_norm": 0.8656779527664185, "learning_rate": 4.736000000000001e-06, "loss": 0.0093, "step": 26321 }, { "epoch": 1.4033160953244121, "grad_norm": 1.5894801616668701, "learning_rate": 4.7358000000000005e-06, "loss": 0.0202, "step": 26322 }, { "epoch": 1.4033694087540651, "grad_norm": 1.7458473443984985, "learning_rate": 4.7356e-06, "loss": 0.0397, "step": 26323 }, { "epoch": 1.403422722183718, "grad_norm": 1.3697223663330078, "learning_rate": 4.7354000000000004e-06, "loss": 0.0263, "step": 26324 }, { "epoch": 1.403476035613371, "grad_norm": 1.2247542142868042, "learning_rate": 4.7352e-06, "loss": 0.0163, "step": 26325 }, { "epoch": 1.403529349043024, "grad_norm": 0.9326573610305786, "learning_rate": 4.735e-06, "loss": 0.0083, "step": 26326 }, { "epoch": 1.4035826624726768, "grad_norm": 1.231924057006836, "learning_rate": 4.734800000000001e-06, "loss": 0.0295, "step": 26327 }, { "epoch": 1.4036359759023298, "grad_norm": 0.9926556348800659, "learning_rate": 4.7346e-06, "loss": 0.0109, "step": 26328 }, { "epoch": 1.4036892893319828, "grad_norm": 0.7356938719749451, "learning_rate": 4.734400000000001e-06, "loss": 0.017, "step": 26329 }, { "epoch": 1.4037426027616355, "grad_norm": 0.8290178179740906, "learning_rate": 4.7342e-06, "loss": 0.0094, "step": 26330 }, { "epoch": 1.4037959161912885, "grad_norm": 1.0929633378982544, "learning_rate": 4.734e-06, "loss": 0.0097, "step": 26331 }, { "epoch": 1.4038492296209415, "grad_norm": 3.419684886932373, "learning_rate": 4.7338e-06, "loss": 0.059, "step": 26332 }, { "epoch": 1.4039025430505945, "grad_norm": 1.3129009008407593, "learning_rate": 4.733600000000001e-06, "loss": 0.0085, "step": 26333 }, { "epoch": 1.4039558564802475, "grad_norm": 1.7590723037719727, "learning_rate": 4.7334e-06, "loss": 0.0247, "step": 26334 }, { "epoch": 1.4040091699099002, "grad_norm": 2.374799966812134, "learning_rate": 4.733200000000001e-06, "loss": 0.0475, "step": 26335 }, { "epoch": 1.4040624833395532, "grad_norm": 0.8496022820472717, "learning_rate": 4.733e-06, "loss": 0.0275, "step": 26336 }, { "epoch": 1.4041157967692062, "grad_norm": 1.5588810443878174, "learning_rate": 4.7328e-06, "loss": 0.0142, "step": 26337 }, { "epoch": 1.4041691101988591, "grad_norm": 1.2194197177886963, "learning_rate": 4.7326e-06, "loss": 0.023, "step": 26338 }, { "epoch": 1.4042224236285121, "grad_norm": 0.7519219517707825, "learning_rate": 4.732400000000001e-06, "loss": 0.0095, "step": 26339 }, { "epoch": 1.4042757370581649, "grad_norm": 1.0082310438156128, "learning_rate": 4.7322e-06, "loss": 0.0116, "step": 26340 }, { "epoch": 1.4043290504878179, "grad_norm": 2.5976009368896484, "learning_rate": 4.732000000000001e-06, "loss": 0.0398, "step": 26341 }, { "epoch": 1.4043823639174708, "grad_norm": 2.772461175918579, "learning_rate": 4.7318e-06, "loss": 0.0262, "step": 26342 }, { "epoch": 1.4044356773471238, "grad_norm": 1.5382553339004517, "learning_rate": 4.7316000000000006e-06, "loss": 0.0273, "step": 26343 }, { "epoch": 1.4044889907767768, "grad_norm": 2.5849967002868652, "learning_rate": 4.7314e-06, "loss": 0.018, "step": 26344 }, { "epoch": 1.4045423042064296, "grad_norm": 1.179070234298706, "learning_rate": 4.7312000000000005e-06, "loss": 0.0163, "step": 26345 }, { "epoch": 1.4045956176360825, "grad_norm": 1.980263113975525, "learning_rate": 4.731000000000001e-06, "loss": 0.0191, "step": 26346 }, { "epoch": 1.4046489310657355, "grad_norm": 1.4750310182571411, "learning_rate": 4.7308000000000005e-06, "loss": 0.0148, "step": 26347 }, { "epoch": 1.4047022444953883, "grad_norm": 1.4709088802337646, "learning_rate": 4.7306e-06, "loss": 0.0112, "step": 26348 }, { "epoch": 1.4047555579250413, "grad_norm": 0.6902149319648743, "learning_rate": 4.7304000000000005e-06, "loss": 0.0067, "step": 26349 }, { "epoch": 1.4048088713546942, "grad_norm": 2.4302115440368652, "learning_rate": 4.7302e-06, "loss": 0.0347, "step": 26350 }, { "epoch": 1.4048621847843472, "grad_norm": 1.7498815059661865, "learning_rate": 4.7300000000000005e-06, "loss": 0.0313, "step": 26351 }, { "epoch": 1.4049154982140002, "grad_norm": 2.120191812515259, "learning_rate": 4.729800000000001e-06, "loss": 0.025, "step": 26352 }, { "epoch": 1.404968811643653, "grad_norm": 0.6861202716827393, "learning_rate": 4.7296e-06, "loss": 0.0115, "step": 26353 }, { "epoch": 1.405022125073306, "grad_norm": 2.330785036087036, "learning_rate": 4.729400000000001e-06, "loss": 0.0185, "step": 26354 }, { "epoch": 1.405075438502959, "grad_norm": 1.5439120531082153, "learning_rate": 4.7292e-06, "loss": 0.0389, "step": 26355 }, { "epoch": 1.4051287519326119, "grad_norm": 2.6769332885742188, "learning_rate": 4.729e-06, "loss": 0.0255, "step": 26356 }, { "epoch": 1.4051820653622649, "grad_norm": 1.4334158897399902, "learning_rate": 4.7288e-06, "loss": 0.0243, "step": 26357 }, { "epoch": 1.4052353787919176, "grad_norm": 2.611058473587036, "learning_rate": 4.728600000000001e-06, "loss": 0.0406, "step": 26358 }, { "epoch": 1.4052886922215706, "grad_norm": 1.2145975828170776, "learning_rate": 4.7284e-06, "loss": 0.0157, "step": 26359 }, { "epoch": 1.4053420056512236, "grad_norm": 0.644121527671814, "learning_rate": 4.728200000000001e-06, "loss": 0.0062, "step": 26360 }, { "epoch": 1.4053953190808763, "grad_norm": 0.9551481008529663, "learning_rate": 4.728e-06, "loss": 0.0073, "step": 26361 }, { "epoch": 1.4054486325105295, "grad_norm": 1.2691010236740112, "learning_rate": 4.7278e-06, "loss": 0.0073, "step": 26362 }, { "epoch": 1.4055019459401823, "grad_norm": 2.2728195190429688, "learning_rate": 4.7276e-06, "loss": 0.0266, "step": 26363 }, { "epoch": 1.4055552593698353, "grad_norm": 2.624138593673706, "learning_rate": 4.727400000000001e-06, "loss": 0.0364, "step": 26364 }, { "epoch": 1.4056085727994883, "grad_norm": 1.2111055850982666, "learning_rate": 4.7272e-06, "loss": 0.0111, "step": 26365 }, { "epoch": 1.405661886229141, "grad_norm": 1.6858619451522827, "learning_rate": 4.727000000000001e-06, "loss": 0.016, "step": 26366 }, { "epoch": 1.405715199658794, "grad_norm": 1.8384318351745605, "learning_rate": 4.7268e-06, "loss": 0.0265, "step": 26367 }, { "epoch": 1.405768513088447, "grad_norm": 2.2351996898651123, "learning_rate": 4.726600000000001e-06, "loss": 0.0208, "step": 26368 }, { "epoch": 1.4058218265181, "grad_norm": 1.4516710042953491, "learning_rate": 4.7264e-06, "loss": 0.0171, "step": 26369 }, { "epoch": 1.405875139947753, "grad_norm": 4.109454154968262, "learning_rate": 4.726200000000001e-06, "loss": 0.0555, "step": 26370 }, { "epoch": 1.4059284533774057, "grad_norm": 2.3979640007019043, "learning_rate": 4.726000000000001e-06, "loss": 0.0308, "step": 26371 }, { "epoch": 1.4059817668070587, "grad_norm": 1.1588438749313354, "learning_rate": 4.7258000000000006e-06, "loss": 0.0088, "step": 26372 }, { "epoch": 1.4060350802367116, "grad_norm": 3.1917247772216797, "learning_rate": 4.7256e-06, "loss": 0.0384, "step": 26373 }, { "epoch": 1.4060883936663646, "grad_norm": 1.7510101795196533, "learning_rate": 4.7254000000000005e-06, "loss": 0.0252, "step": 26374 }, { "epoch": 1.4061417070960176, "grad_norm": 1.401816964149475, "learning_rate": 4.7252e-06, "loss": 0.0217, "step": 26375 }, { "epoch": 1.4061950205256704, "grad_norm": 1.5213749408721924, "learning_rate": 4.7250000000000005e-06, "loss": 0.0272, "step": 26376 }, { "epoch": 1.4062483339553233, "grad_norm": 1.0221034288406372, "learning_rate": 4.724800000000001e-06, "loss": 0.0223, "step": 26377 }, { "epoch": 1.4063016473849763, "grad_norm": 2.227342128753662, "learning_rate": 4.7246000000000005e-06, "loss": 0.0333, "step": 26378 }, { "epoch": 1.406354960814629, "grad_norm": 1.5395601987838745, "learning_rate": 4.724400000000001e-06, "loss": 0.0368, "step": 26379 }, { "epoch": 1.406408274244282, "grad_norm": 1.488844394683838, "learning_rate": 4.7242000000000004e-06, "loss": 0.0198, "step": 26380 }, { "epoch": 1.406461587673935, "grad_norm": 1.5608755350112915, "learning_rate": 4.724e-06, "loss": 0.0258, "step": 26381 }, { "epoch": 1.406514901103588, "grad_norm": 1.5129257440567017, "learning_rate": 4.7238e-06, "loss": 0.0203, "step": 26382 }, { "epoch": 1.406568214533241, "grad_norm": 4.07929801940918, "learning_rate": 4.723600000000001e-06, "loss": 0.0543, "step": 26383 }, { "epoch": 1.4066215279628937, "grad_norm": 1.0648256540298462, "learning_rate": 4.7234e-06, "loss": 0.0207, "step": 26384 }, { "epoch": 1.4066748413925467, "grad_norm": 2.6591689586639404, "learning_rate": 4.723200000000001e-06, "loss": 0.0284, "step": 26385 }, { "epoch": 1.4067281548221997, "grad_norm": 0.7707247138023376, "learning_rate": 4.723e-06, "loss": 0.0197, "step": 26386 }, { "epoch": 1.4067814682518527, "grad_norm": 1.3772234916687012, "learning_rate": 4.7228e-06, "loss": 0.0266, "step": 26387 }, { "epoch": 1.4068347816815057, "grad_norm": 0.6518570184707642, "learning_rate": 4.7226e-06, "loss": 0.0141, "step": 26388 }, { "epoch": 1.4068880951111584, "grad_norm": 1.1701393127441406, "learning_rate": 4.722400000000001e-06, "loss": 0.0367, "step": 26389 }, { "epoch": 1.4069414085408114, "grad_norm": 1.2940186262130737, "learning_rate": 4.7222e-06, "loss": 0.0352, "step": 26390 }, { "epoch": 1.4069947219704644, "grad_norm": 2.189317464828491, "learning_rate": 4.722000000000001e-06, "loss": 0.0414, "step": 26391 }, { "epoch": 1.4070480354001174, "grad_norm": 1.4320310354232788, "learning_rate": 4.7218e-06, "loss": 0.0254, "step": 26392 }, { "epoch": 1.4071013488297703, "grad_norm": 1.3077771663665771, "learning_rate": 4.721600000000001e-06, "loss": 0.0234, "step": 26393 }, { "epoch": 1.407154662259423, "grad_norm": 1.750255823135376, "learning_rate": 4.7214e-06, "loss": 0.0266, "step": 26394 }, { "epoch": 1.407207975689076, "grad_norm": 2.0497751235961914, "learning_rate": 4.721200000000001e-06, "loss": 0.0348, "step": 26395 }, { "epoch": 1.407261289118729, "grad_norm": 3.0921730995178223, "learning_rate": 4.721e-06, "loss": 0.0446, "step": 26396 }, { "epoch": 1.4073146025483818, "grad_norm": 2.234158754348755, "learning_rate": 4.720800000000001e-06, "loss": 0.0479, "step": 26397 }, { "epoch": 1.4073679159780348, "grad_norm": 1.6741669178009033, "learning_rate": 4.7206e-06, "loss": 0.015, "step": 26398 }, { "epoch": 1.4074212294076878, "grad_norm": 1.0948593616485596, "learning_rate": 4.720400000000001e-06, "loss": 0.0146, "step": 26399 }, { "epoch": 1.4074745428373407, "grad_norm": 2.1849491596221924, "learning_rate": 4.7202e-06, "loss": 0.057, "step": 26400 }, { "epoch": 1.4075278562669937, "grad_norm": 1.059287428855896, "learning_rate": 4.7200000000000005e-06, "loss": 0.0238, "step": 26401 }, { "epoch": 1.4075811696966465, "grad_norm": 1.9711090326309204, "learning_rate": 4.7198e-06, "loss": 0.0243, "step": 26402 }, { "epoch": 1.4076344831262995, "grad_norm": 1.4699006080627441, "learning_rate": 4.7196000000000005e-06, "loss": 0.0113, "step": 26403 }, { "epoch": 1.4076877965559524, "grad_norm": 1.8236788511276245, "learning_rate": 4.719400000000001e-06, "loss": 0.059, "step": 26404 }, { "epoch": 1.4077411099856054, "grad_norm": 1.721130132675171, "learning_rate": 4.7192000000000005e-06, "loss": 0.0304, "step": 26405 }, { "epoch": 1.4077944234152584, "grad_norm": 1.3475871086120605, "learning_rate": 4.719e-06, "loss": 0.0215, "step": 26406 }, { "epoch": 1.4078477368449112, "grad_norm": 1.9147086143493652, "learning_rate": 4.7188000000000005e-06, "loss": 0.0352, "step": 26407 }, { "epoch": 1.4079010502745641, "grad_norm": 2.1602118015289307, "learning_rate": 4.7186e-06, "loss": 0.0501, "step": 26408 }, { "epoch": 1.4079543637042171, "grad_norm": 2.77876615524292, "learning_rate": 4.7184000000000004e-06, "loss": 0.0393, "step": 26409 }, { "epoch": 1.40800767713387, "grad_norm": 1.6370446681976318, "learning_rate": 4.718200000000001e-06, "loss": 0.0199, "step": 26410 }, { "epoch": 1.408060990563523, "grad_norm": 1.3813530206680298, "learning_rate": 4.718e-06, "loss": 0.0315, "step": 26411 }, { "epoch": 1.4081143039931758, "grad_norm": 1.0475038290023804, "learning_rate": 4.7178e-06, "loss": 0.0204, "step": 26412 }, { "epoch": 1.4081676174228288, "grad_norm": 0.44287532567977905, "learning_rate": 4.7176e-06, "loss": 0.0037, "step": 26413 }, { "epoch": 1.4082209308524818, "grad_norm": 1.0699745416641235, "learning_rate": 4.7174e-06, "loss": 0.0111, "step": 26414 }, { "epoch": 1.4082742442821345, "grad_norm": 3.151972770690918, "learning_rate": 4.7172e-06, "loss": 0.0314, "step": 26415 }, { "epoch": 1.4083275577117875, "grad_norm": 1.0879665613174438, "learning_rate": 4.717000000000001e-06, "loss": 0.013, "step": 26416 }, { "epoch": 1.4083808711414405, "grad_norm": 2.340928792953491, "learning_rate": 4.7168e-06, "loss": 0.0248, "step": 26417 }, { "epoch": 1.4084341845710935, "grad_norm": 1.3845583200454712, "learning_rate": 4.716600000000001e-06, "loss": 0.0123, "step": 26418 }, { "epoch": 1.4084874980007465, "grad_norm": 1.8379647731781006, "learning_rate": 4.7164e-06, "loss": 0.0164, "step": 26419 }, { "epoch": 1.4085408114303992, "grad_norm": 2.112269878387451, "learning_rate": 4.7162e-06, "loss": 0.0187, "step": 26420 }, { "epoch": 1.4085941248600522, "grad_norm": 1.9638084173202515, "learning_rate": 4.716e-06, "loss": 0.0344, "step": 26421 }, { "epoch": 1.4086474382897052, "grad_norm": 0.5706282258033752, "learning_rate": 4.715800000000001e-06, "loss": 0.0058, "step": 26422 }, { "epoch": 1.4087007517193582, "grad_norm": 2.7595980167388916, "learning_rate": 4.7156e-06, "loss": 0.047, "step": 26423 }, { "epoch": 1.4087540651490111, "grad_norm": 1.7731660604476929, "learning_rate": 4.715400000000001e-06, "loss": 0.0305, "step": 26424 }, { "epoch": 1.408807378578664, "grad_norm": 1.8253159523010254, "learning_rate": 4.7152e-06, "loss": 0.0368, "step": 26425 }, { "epoch": 1.4088606920083169, "grad_norm": 1.032052993774414, "learning_rate": 4.715e-06, "loss": 0.0113, "step": 26426 }, { "epoch": 1.4089140054379699, "grad_norm": 2.8161678314208984, "learning_rate": 4.7148e-06, "loss": 0.0343, "step": 26427 }, { "epoch": 1.4089673188676228, "grad_norm": 1.4951653480529785, "learning_rate": 4.7146000000000006e-06, "loss": 0.0259, "step": 26428 }, { "epoch": 1.4090206322972758, "grad_norm": 1.806875228881836, "learning_rate": 4.714400000000001e-06, "loss": 0.0273, "step": 26429 }, { "epoch": 1.4090739457269286, "grad_norm": 2.061288356781006, "learning_rate": 4.7142000000000005e-06, "loss": 0.0248, "step": 26430 }, { "epoch": 1.4091272591565815, "grad_norm": 1.7471940517425537, "learning_rate": 4.714e-06, "loss": 0.0297, "step": 26431 }, { "epoch": 1.4091805725862345, "grad_norm": 1.3924219608306885, "learning_rate": 4.7138000000000005e-06, "loss": 0.0166, "step": 26432 }, { "epoch": 1.4092338860158873, "grad_norm": 1.6990561485290527, "learning_rate": 4.7136e-06, "loss": 0.0246, "step": 26433 }, { "epoch": 1.4092871994455403, "grad_norm": 2.4449737071990967, "learning_rate": 4.7134000000000005e-06, "loss": 0.0161, "step": 26434 }, { "epoch": 1.4093405128751932, "grad_norm": 1.8710260391235352, "learning_rate": 4.713200000000001e-06, "loss": 0.0335, "step": 26435 }, { "epoch": 1.4093938263048462, "grad_norm": 1.2288517951965332, "learning_rate": 4.7130000000000004e-06, "loss": 0.0211, "step": 26436 }, { "epoch": 1.4094471397344992, "grad_norm": 0.6569162607192993, "learning_rate": 4.7128e-06, "loss": 0.0057, "step": 26437 }, { "epoch": 1.409500453164152, "grad_norm": 2.707507371902466, "learning_rate": 4.7126e-06, "loss": 0.0429, "step": 26438 }, { "epoch": 1.409553766593805, "grad_norm": 1.6911592483520508, "learning_rate": 4.7124e-06, "loss": 0.026, "step": 26439 }, { "epoch": 1.409607080023458, "grad_norm": 1.5083768367767334, "learning_rate": 4.7122e-06, "loss": 0.019, "step": 26440 }, { "epoch": 1.409660393453111, "grad_norm": 1.7764474153518677, "learning_rate": 4.712000000000001e-06, "loss": 0.0374, "step": 26441 }, { "epoch": 1.4097137068827639, "grad_norm": 2.0194814205169678, "learning_rate": 4.7118e-06, "loss": 0.0321, "step": 26442 }, { "epoch": 1.4097670203124166, "grad_norm": 1.1723010540008545, "learning_rate": 4.711600000000001e-06, "loss": 0.0217, "step": 26443 }, { "epoch": 1.4098203337420696, "grad_norm": 1.3028558492660522, "learning_rate": 4.7114e-06, "loss": 0.0138, "step": 26444 }, { "epoch": 1.4098736471717226, "grad_norm": 0.8652210235595703, "learning_rate": 4.7112e-06, "loss": 0.008, "step": 26445 }, { "epoch": 1.4099269606013756, "grad_norm": 1.9159570932388306, "learning_rate": 4.711e-06, "loss": 0.0465, "step": 26446 }, { "epoch": 1.4099802740310285, "grad_norm": 2.363342523574829, "learning_rate": 4.710800000000001e-06, "loss": 0.0262, "step": 26447 }, { "epoch": 1.4100335874606813, "grad_norm": 1.9740408658981323, "learning_rate": 4.7106e-06, "loss": 0.02, "step": 26448 }, { "epoch": 1.4100869008903343, "grad_norm": 1.4228521585464478, "learning_rate": 4.710400000000001e-06, "loss": 0.0136, "step": 26449 }, { "epoch": 1.4101402143199873, "grad_norm": 1.6208045482635498, "learning_rate": 4.7102e-06, "loss": 0.0142, "step": 26450 }, { "epoch": 1.41019352774964, "grad_norm": 2.1985318660736084, "learning_rate": 4.71e-06, "loss": 0.0323, "step": 26451 }, { "epoch": 1.410246841179293, "grad_norm": 1.884441614151001, "learning_rate": 4.7098e-06, "loss": 0.025, "step": 26452 }, { "epoch": 1.410300154608946, "grad_norm": 1.4694560766220093, "learning_rate": 4.709600000000001e-06, "loss": 0.0132, "step": 26453 }, { "epoch": 1.410353468038599, "grad_norm": 0.7567402720451355, "learning_rate": 4.709400000000001e-06, "loss": 0.0066, "step": 26454 }, { "epoch": 1.410406781468252, "grad_norm": 1.9923832416534424, "learning_rate": 4.709200000000001e-06, "loss": 0.0242, "step": 26455 }, { "epoch": 1.4104600948979047, "grad_norm": 2.7582709789276123, "learning_rate": 4.709e-06, "loss": 0.0422, "step": 26456 }, { "epoch": 1.4105134083275577, "grad_norm": 2.054781675338745, "learning_rate": 4.7088000000000006e-06, "loss": 0.0109, "step": 26457 }, { "epoch": 1.4105667217572107, "grad_norm": 2.177793502807617, "learning_rate": 4.7086e-06, "loss": 0.0283, "step": 26458 }, { "epoch": 1.4106200351868636, "grad_norm": 2.7971532344818115, "learning_rate": 4.7084000000000005e-06, "loss": 0.0444, "step": 26459 }, { "epoch": 1.4106733486165166, "grad_norm": 2.4551494121551514, "learning_rate": 4.708200000000001e-06, "loss": 0.034, "step": 26460 }, { "epoch": 1.4107266620461694, "grad_norm": 1.8756139278411865, "learning_rate": 4.7080000000000005e-06, "loss": 0.025, "step": 26461 }, { "epoch": 1.4107799754758223, "grad_norm": 1.3844550848007202, "learning_rate": 4.7078e-06, "loss": 0.0335, "step": 26462 }, { "epoch": 1.4108332889054753, "grad_norm": 3.157398223876953, "learning_rate": 4.7076000000000005e-06, "loss": 0.0175, "step": 26463 }, { "epoch": 1.410886602335128, "grad_norm": 2.801724672317505, "learning_rate": 4.7074e-06, "loss": 0.0299, "step": 26464 }, { "epoch": 1.4109399157647813, "grad_norm": 1.9778051376342773, "learning_rate": 4.7072000000000004e-06, "loss": 0.0419, "step": 26465 }, { "epoch": 1.410993229194434, "grad_norm": 1.5554914474487305, "learning_rate": 4.707000000000001e-06, "loss": 0.0073, "step": 26466 }, { "epoch": 1.411046542624087, "grad_norm": 2.7148468494415283, "learning_rate": 4.7068e-06, "loss": 0.023, "step": 26467 }, { "epoch": 1.41109985605374, "grad_norm": 3.2638800144195557, "learning_rate": 4.706600000000001e-06, "loss": 0.0147, "step": 26468 }, { "epoch": 1.4111531694833928, "grad_norm": 0.5013144612312317, "learning_rate": 4.7064e-06, "loss": 0.0057, "step": 26469 }, { "epoch": 1.4112064829130457, "grad_norm": 1.2499691247940063, "learning_rate": 4.7062e-06, "loss": 0.0293, "step": 26470 }, { "epoch": 1.4112597963426987, "grad_norm": 3.711693286895752, "learning_rate": 4.706e-06, "loss": 0.0551, "step": 26471 }, { "epoch": 1.4113131097723517, "grad_norm": 0.7274110913276672, "learning_rate": 4.705800000000001e-06, "loss": 0.0057, "step": 26472 }, { "epoch": 1.4113664232020047, "grad_norm": 1.767987847328186, "learning_rate": 4.7056e-06, "loss": 0.0378, "step": 26473 }, { "epoch": 1.4114197366316574, "grad_norm": 1.9390716552734375, "learning_rate": 4.705400000000001e-06, "loss": 0.0183, "step": 26474 }, { "epoch": 1.4114730500613104, "grad_norm": 1.3615833520889282, "learning_rate": 4.7052e-06, "loss": 0.0094, "step": 26475 }, { "epoch": 1.4115263634909634, "grad_norm": 1.5966894626617432, "learning_rate": 4.705e-06, "loss": 0.0153, "step": 26476 }, { "epoch": 1.4115796769206164, "grad_norm": 0.7721390724182129, "learning_rate": 4.7048e-06, "loss": 0.0125, "step": 26477 }, { "epoch": 1.4116329903502693, "grad_norm": 1.7266570329666138, "learning_rate": 4.704600000000001e-06, "loss": 0.0285, "step": 26478 }, { "epoch": 1.411686303779922, "grad_norm": 0.8499940633773804, "learning_rate": 4.7044e-06, "loss": 0.0053, "step": 26479 }, { "epoch": 1.411739617209575, "grad_norm": 3.3800055980682373, "learning_rate": 4.704200000000001e-06, "loss": 0.0626, "step": 26480 }, { "epoch": 1.411792930639228, "grad_norm": 1.448104739189148, "learning_rate": 4.704e-06, "loss": 0.012, "step": 26481 }, { "epoch": 1.4118462440688808, "grad_norm": 1.4295880794525146, "learning_rate": 4.703800000000001e-06, "loss": 0.0165, "step": 26482 }, { "epoch": 1.411899557498534, "grad_norm": 5.944217205047607, "learning_rate": 4.7036e-06, "loss": 0.0332, "step": 26483 }, { "epoch": 1.4119528709281868, "grad_norm": 0.9077149033546448, "learning_rate": 4.7034000000000006e-06, "loss": 0.0101, "step": 26484 }, { "epoch": 1.4120061843578398, "grad_norm": 1.604219913482666, "learning_rate": 4.7032e-06, "loss": 0.0332, "step": 26485 }, { "epoch": 1.4120594977874927, "grad_norm": 1.5857009887695312, "learning_rate": 4.7030000000000005e-06, "loss": 0.0131, "step": 26486 }, { "epoch": 1.4121128112171455, "grad_norm": 2.341346025466919, "learning_rate": 4.7028e-06, "loss": 0.0338, "step": 26487 }, { "epoch": 1.4121661246467985, "grad_norm": 0.9858493208885193, "learning_rate": 4.7026000000000005e-06, "loss": 0.0073, "step": 26488 }, { "epoch": 1.4122194380764514, "grad_norm": 0.31940722465515137, "learning_rate": 4.7024e-06, "loss": 0.0033, "step": 26489 }, { "epoch": 1.4122727515061044, "grad_norm": 1.6439367532730103, "learning_rate": 4.7022000000000005e-06, "loss": 0.0162, "step": 26490 }, { "epoch": 1.4123260649357574, "grad_norm": 1.0004454851150513, "learning_rate": 4.702e-06, "loss": 0.0079, "step": 26491 }, { "epoch": 1.4123793783654102, "grad_norm": 0.6293317079544067, "learning_rate": 4.7018000000000005e-06, "loss": 0.0058, "step": 26492 }, { "epoch": 1.4124326917950631, "grad_norm": 0.5111232399940491, "learning_rate": 4.701600000000001e-06, "loss": 0.0044, "step": 26493 }, { "epoch": 1.4124860052247161, "grad_norm": 3.0931508541107178, "learning_rate": 4.7014e-06, "loss": 0.0293, "step": 26494 }, { "epoch": 1.412539318654369, "grad_norm": 0.8257556557655334, "learning_rate": 4.7012e-06, "loss": 0.0186, "step": 26495 }, { "epoch": 1.412592632084022, "grad_norm": 0.6459243893623352, "learning_rate": 4.701e-06, "loss": 0.0036, "step": 26496 }, { "epoch": 1.4126459455136748, "grad_norm": 2.6970505714416504, "learning_rate": 4.7008e-06, "loss": 0.0469, "step": 26497 }, { "epoch": 1.4126992589433278, "grad_norm": 0.8997096419334412, "learning_rate": 4.7006e-06, "loss": 0.0128, "step": 26498 }, { "epoch": 1.4127525723729808, "grad_norm": 2.054584264755249, "learning_rate": 4.700400000000001e-06, "loss": 0.0192, "step": 26499 }, { "epoch": 1.4128058858026336, "grad_norm": 2.66695499420166, "learning_rate": 4.7002e-06, "loss": 0.0542, "step": 26500 }, { "epoch": 1.4128591992322865, "grad_norm": 1.3734629154205322, "learning_rate": 4.7e-06, "loss": 0.0072, "step": 26501 }, { "epoch": 1.4129125126619395, "grad_norm": 3.34324049949646, "learning_rate": 4.6998e-06, "loss": 0.0259, "step": 26502 }, { "epoch": 1.4129658260915925, "grad_norm": 2.513474464416504, "learning_rate": 4.6996e-06, "loss": 0.011, "step": 26503 }, { "epoch": 1.4130191395212455, "grad_norm": 1.2376573085784912, "learning_rate": 4.6994e-06, "loss": 0.0099, "step": 26504 }, { "epoch": 1.4130724529508982, "grad_norm": 1.9565372467041016, "learning_rate": 4.699200000000001e-06, "loss": 0.0139, "step": 26505 }, { "epoch": 1.4131257663805512, "grad_norm": 1.1589022874832153, "learning_rate": 4.699e-06, "loss": 0.0146, "step": 26506 }, { "epoch": 1.4131790798102042, "grad_norm": 1.371636152267456, "learning_rate": 4.698800000000001e-06, "loss": 0.0194, "step": 26507 }, { "epoch": 1.4132323932398572, "grad_norm": 1.1066226959228516, "learning_rate": 4.6986e-06, "loss": 0.0136, "step": 26508 }, { "epoch": 1.4132857066695101, "grad_norm": 0.5576534271240234, "learning_rate": 4.6984e-06, "loss": 0.003, "step": 26509 }, { "epoch": 1.413339020099163, "grad_norm": 1.8248788118362427, "learning_rate": 4.6982e-06, "loss": 0.03, "step": 26510 }, { "epoch": 1.4133923335288159, "grad_norm": 1.1676359176635742, "learning_rate": 4.698000000000001e-06, "loss": 0.0178, "step": 26511 }, { "epoch": 1.4134456469584689, "grad_norm": 0.6785576939582825, "learning_rate": 4.6978e-06, "loss": 0.0127, "step": 26512 }, { "epoch": 1.4134989603881218, "grad_norm": 1.1054980754852295, "learning_rate": 4.6976000000000006e-06, "loss": 0.0052, "step": 26513 }, { "epoch": 1.4135522738177748, "grad_norm": 1.7910114526748657, "learning_rate": 4.6974e-06, "loss": 0.0139, "step": 26514 }, { "epoch": 1.4136055872474276, "grad_norm": 0.8137811422348022, "learning_rate": 4.6972000000000005e-06, "loss": 0.0103, "step": 26515 }, { "epoch": 1.4136589006770806, "grad_norm": 0.9143823981285095, "learning_rate": 4.697e-06, "loss": 0.0121, "step": 26516 }, { "epoch": 1.4137122141067335, "grad_norm": 0.8084458708763123, "learning_rate": 4.6968000000000005e-06, "loss": 0.0168, "step": 26517 }, { "epoch": 1.4137655275363863, "grad_norm": 2.8706088066101074, "learning_rate": 4.696600000000001e-06, "loss": 0.0471, "step": 26518 }, { "epoch": 1.4138188409660393, "grad_norm": 2.7450408935546875, "learning_rate": 4.6964000000000005e-06, "loss": 0.0384, "step": 26519 }, { "epoch": 1.4138721543956922, "grad_norm": 1.8183408975601196, "learning_rate": 4.6962e-06, "loss": 0.0135, "step": 26520 }, { "epoch": 1.4139254678253452, "grad_norm": 3.879204750061035, "learning_rate": 4.6960000000000004e-06, "loss": 0.0315, "step": 26521 }, { "epoch": 1.4139787812549982, "grad_norm": 1.099513053894043, "learning_rate": 4.6958e-06, "loss": 0.0131, "step": 26522 }, { "epoch": 1.414032094684651, "grad_norm": 3.145578145980835, "learning_rate": 4.6956e-06, "loss": 0.0603, "step": 26523 }, { "epoch": 1.414085408114304, "grad_norm": 2.6693058013916016, "learning_rate": 4.695400000000001e-06, "loss": 0.0268, "step": 26524 }, { "epoch": 1.414138721543957, "grad_norm": 1.470774531364441, "learning_rate": 4.6952e-06, "loss": 0.0123, "step": 26525 }, { "epoch": 1.41419203497361, "grad_norm": 3.5268421173095703, "learning_rate": 4.695e-06, "loss": 0.0452, "step": 26526 }, { "epoch": 1.4142453484032629, "grad_norm": 1.250075340270996, "learning_rate": 4.6948e-06, "loss": 0.0114, "step": 26527 }, { "epoch": 1.4142986618329156, "grad_norm": 2.71171236038208, "learning_rate": 4.6946e-06, "loss": 0.0259, "step": 26528 }, { "epoch": 1.4143519752625686, "grad_norm": 3.0369064807891846, "learning_rate": 4.6944e-06, "loss": 0.0249, "step": 26529 }, { "epoch": 1.4144052886922216, "grad_norm": 2.194679021835327, "learning_rate": 4.694200000000001e-06, "loss": 0.025, "step": 26530 }, { "epoch": 1.4144586021218746, "grad_norm": 1.203961730003357, "learning_rate": 4.694e-06, "loss": 0.0201, "step": 26531 }, { "epoch": 1.4145119155515276, "grad_norm": 1.0329160690307617, "learning_rate": 4.693800000000001e-06, "loss": 0.0196, "step": 26532 }, { "epoch": 1.4145652289811803, "grad_norm": 1.9000691175460815, "learning_rate": 4.6936e-06, "loss": 0.0128, "step": 26533 }, { "epoch": 1.4146185424108333, "grad_norm": 2.1867446899414062, "learning_rate": 4.6934e-06, "loss": 0.0256, "step": 26534 }, { "epoch": 1.4146718558404863, "grad_norm": 2.373141050338745, "learning_rate": 4.6932e-06, "loss": 0.0325, "step": 26535 }, { "epoch": 1.414725169270139, "grad_norm": 1.780838131904602, "learning_rate": 4.693000000000001e-06, "loss": 0.0186, "step": 26536 }, { "epoch": 1.414778482699792, "grad_norm": 2.045444965362549, "learning_rate": 4.6928e-06, "loss": 0.0468, "step": 26537 }, { "epoch": 1.414831796129445, "grad_norm": 3.1147024631500244, "learning_rate": 4.692600000000001e-06, "loss": 0.0218, "step": 26538 }, { "epoch": 1.414885109559098, "grad_norm": 1.5783100128173828, "learning_rate": 4.6924e-06, "loss": 0.0214, "step": 26539 }, { "epoch": 1.414938422988751, "grad_norm": 3.4543838500976562, "learning_rate": 4.692200000000001e-06, "loss": 0.0288, "step": 26540 }, { "epoch": 1.4149917364184037, "grad_norm": 1.5221643447875977, "learning_rate": 4.692e-06, "loss": 0.0178, "step": 26541 }, { "epoch": 1.4150450498480567, "grad_norm": 0.6363475918769836, "learning_rate": 4.6918000000000005e-06, "loss": 0.007, "step": 26542 }, { "epoch": 1.4150983632777097, "grad_norm": 2.0309529304504395, "learning_rate": 4.691600000000001e-06, "loss": 0.0304, "step": 26543 }, { "epoch": 1.4151516767073626, "grad_norm": 1.6733331680297852, "learning_rate": 4.6914000000000005e-06, "loss": 0.0355, "step": 26544 }, { "epoch": 1.4152049901370156, "grad_norm": 1.0662314891815186, "learning_rate": 4.6912e-06, "loss": 0.01, "step": 26545 }, { "epoch": 1.4152583035666684, "grad_norm": 1.9775310754776, "learning_rate": 4.6910000000000005e-06, "loss": 0.0291, "step": 26546 }, { "epoch": 1.4153116169963214, "grad_norm": 1.5930312871932983, "learning_rate": 4.6908e-06, "loss": 0.014, "step": 26547 }, { "epoch": 1.4153649304259743, "grad_norm": 2.3228251934051514, "learning_rate": 4.6906000000000005e-06, "loss": 0.0191, "step": 26548 }, { "epoch": 1.4154182438556273, "grad_norm": 1.1307399272918701, "learning_rate": 4.690400000000001e-06, "loss": 0.0121, "step": 26549 }, { "epoch": 1.4154715572852803, "grad_norm": 1.5072253942489624, "learning_rate": 4.6902000000000004e-06, "loss": 0.0184, "step": 26550 }, { "epoch": 1.415524870714933, "grad_norm": 1.4556281566619873, "learning_rate": 4.69e-06, "loss": 0.0322, "step": 26551 }, { "epoch": 1.415578184144586, "grad_norm": 0.48760315775871277, "learning_rate": 4.6898e-06, "loss": 0.0058, "step": 26552 }, { "epoch": 1.415631497574239, "grad_norm": 1.2153226137161255, "learning_rate": 4.6896e-06, "loss": 0.0101, "step": 26553 }, { "epoch": 1.4156848110038918, "grad_norm": 2.5111215114593506, "learning_rate": 4.6894e-06, "loss": 0.0268, "step": 26554 }, { "epoch": 1.4157381244335447, "grad_norm": 1.576493740081787, "learning_rate": 4.689200000000001e-06, "loss": 0.02, "step": 26555 }, { "epoch": 1.4157914378631977, "grad_norm": 2.9348514080047607, "learning_rate": 4.689e-06, "loss": 0.0548, "step": 26556 }, { "epoch": 1.4158447512928507, "grad_norm": 1.140436053276062, "learning_rate": 4.688800000000001e-06, "loss": 0.0132, "step": 26557 }, { "epoch": 1.4158980647225037, "grad_norm": 1.7876158952713013, "learning_rate": 4.6886e-06, "loss": 0.0204, "step": 26558 }, { "epoch": 1.4159513781521564, "grad_norm": 2.567519426345825, "learning_rate": 4.6884e-06, "loss": 0.042, "step": 26559 }, { "epoch": 1.4160046915818094, "grad_norm": 1.1673574447631836, "learning_rate": 4.6882e-06, "loss": 0.0147, "step": 26560 }, { "epoch": 1.4160580050114624, "grad_norm": 1.810363531112671, "learning_rate": 4.688000000000001e-06, "loss": 0.045, "step": 26561 }, { "epoch": 1.4161113184411154, "grad_norm": 1.9272115230560303, "learning_rate": 4.6878e-06, "loss": 0.0241, "step": 26562 }, { "epoch": 1.4161646318707684, "grad_norm": 2.61948561668396, "learning_rate": 4.687600000000001e-06, "loss": 0.0209, "step": 26563 }, { "epoch": 1.416217945300421, "grad_norm": 1.0470293760299683, "learning_rate": 4.6874e-06, "loss": 0.0159, "step": 26564 }, { "epoch": 1.416271258730074, "grad_norm": 2.63444185256958, "learning_rate": 4.687200000000001e-06, "loss": 0.033, "step": 26565 }, { "epoch": 1.416324572159727, "grad_norm": 0.6754116415977478, "learning_rate": 4.687e-06, "loss": 0.0067, "step": 26566 }, { "epoch": 1.41637788558938, "grad_norm": 1.5698442459106445, "learning_rate": 4.686800000000001e-06, "loss": 0.0315, "step": 26567 }, { "epoch": 1.416431199019033, "grad_norm": 0.8251237273216248, "learning_rate": 4.686600000000001e-06, "loss": 0.0089, "step": 26568 }, { "epoch": 1.4164845124486858, "grad_norm": 1.0888222455978394, "learning_rate": 4.6864000000000006e-06, "loss": 0.0064, "step": 26569 }, { "epoch": 1.4165378258783388, "grad_norm": 3.7738888263702393, "learning_rate": 4.6862e-06, "loss": 0.0428, "step": 26570 }, { "epoch": 1.4165911393079917, "grad_norm": 1.8786612749099731, "learning_rate": 4.6860000000000005e-06, "loss": 0.0233, "step": 26571 }, { "epoch": 1.4166444527376445, "grad_norm": 1.8154151439666748, "learning_rate": 4.6858e-06, "loss": 0.0174, "step": 26572 }, { "epoch": 1.4166977661672975, "grad_norm": 1.9841055870056152, "learning_rate": 4.6856000000000005e-06, "loss": 0.0376, "step": 26573 }, { "epoch": 1.4167510795969505, "grad_norm": 2.2820239067077637, "learning_rate": 4.685400000000001e-06, "loss": 0.0217, "step": 26574 }, { "epoch": 1.4168043930266034, "grad_norm": 1.7946765422821045, "learning_rate": 4.6852000000000005e-06, "loss": 0.0367, "step": 26575 }, { "epoch": 1.4168577064562564, "grad_norm": 3.455944776535034, "learning_rate": 4.685000000000001e-06, "loss": 0.0423, "step": 26576 }, { "epoch": 1.4169110198859092, "grad_norm": 2.0824294090270996, "learning_rate": 4.6848000000000004e-06, "loss": 0.0123, "step": 26577 }, { "epoch": 1.4169643333155622, "grad_norm": 1.3664838075637817, "learning_rate": 4.6846e-06, "loss": 0.021, "step": 26578 }, { "epoch": 1.4170176467452151, "grad_norm": 0.9310824275016785, "learning_rate": 4.6844e-06, "loss": 0.012, "step": 26579 }, { "epoch": 1.417070960174868, "grad_norm": 2.0109500885009766, "learning_rate": 4.684200000000001e-06, "loss": 0.0482, "step": 26580 }, { "epoch": 1.417124273604521, "grad_norm": 2.0938756465911865, "learning_rate": 4.684e-06, "loss": 0.0124, "step": 26581 }, { "epoch": 1.4171775870341738, "grad_norm": 3.0166237354278564, "learning_rate": 4.683800000000001e-06, "loss": 0.061, "step": 26582 }, { "epoch": 1.4172309004638268, "grad_norm": 1.6837880611419678, "learning_rate": 4.6836e-06, "loss": 0.0156, "step": 26583 }, { "epoch": 1.4172842138934798, "grad_norm": 1.2854894399642944, "learning_rate": 4.6834e-06, "loss": 0.0137, "step": 26584 }, { "epoch": 1.4173375273231326, "grad_norm": 1.5207290649414062, "learning_rate": 4.6832e-06, "loss": 0.0231, "step": 26585 }, { "epoch": 1.4173908407527858, "grad_norm": 1.2648191452026367, "learning_rate": 4.683000000000001e-06, "loss": 0.0134, "step": 26586 }, { "epoch": 1.4174441541824385, "grad_norm": 1.9615229368209839, "learning_rate": 4.6828e-06, "loss": 0.045, "step": 26587 }, { "epoch": 1.4174974676120915, "grad_norm": 2.6754019260406494, "learning_rate": 4.682600000000001e-06, "loss": 0.017, "step": 26588 }, { "epoch": 1.4175507810417445, "grad_norm": 1.590712070465088, "learning_rate": 4.6824e-06, "loss": 0.0348, "step": 26589 }, { "epoch": 1.4176040944713972, "grad_norm": 0.7109774947166443, "learning_rate": 4.682200000000001e-06, "loss": 0.0088, "step": 26590 }, { "epoch": 1.4176574079010502, "grad_norm": 2.456570863723755, "learning_rate": 4.682e-06, "loss": 0.0345, "step": 26591 }, { "epoch": 1.4177107213307032, "grad_norm": 1.997300386428833, "learning_rate": 4.681800000000001e-06, "loss": 0.0381, "step": 26592 }, { "epoch": 1.4177640347603562, "grad_norm": 1.1586153507232666, "learning_rate": 4.6816e-06, "loss": 0.0113, "step": 26593 }, { "epoch": 1.4178173481900092, "grad_norm": 2.2810354232788086, "learning_rate": 4.681400000000001e-06, "loss": 0.0181, "step": 26594 }, { "epoch": 1.417870661619662, "grad_norm": 1.5348296165466309, "learning_rate": 4.6812e-06, "loss": 0.0293, "step": 26595 }, { "epoch": 1.4179239750493149, "grad_norm": 1.2882335186004639, "learning_rate": 4.681000000000001e-06, "loss": 0.0177, "step": 26596 }, { "epoch": 1.4179772884789679, "grad_norm": 1.5648870468139648, "learning_rate": 4.6808e-06, "loss": 0.0117, "step": 26597 }, { "epoch": 1.4180306019086208, "grad_norm": 0.8113123178482056, "learning_rate": 4.6806000000000006e-06, "loss": 0.0072, "step": 26598 }, { "epoch": 1.4180839153382738, "grad_norm": 1.940569519996643, "learning_rate": 4.6804e-06, "loss": 0.0271, "step": 26599 }, { "epoch": 1.4181372287679266, "grad_norm": 1.4026200771331787, "learning_rate": 4.6802000000000005e-06, "loss": 0.0233, "step": 26600 }, { "epoch": 1.4181905421975796, "grad_norm": 1.1574678421020508, "learning_rate": 4.680000000000001e-06, "loss": 0.0112, "step": 26601 }, { "epoch": 1.4182438556272325, "grad_norm": 1.7563543319702148, "learning_rate": 4.6798000000000005e-06, "loss": 0.0091, "step": 26602 }, { "epoch": 1.4182971690568853, "grad_norm": 0.900654673576355, "learning_rate": 4.6796e-06, "loss": 0.0189, "step": 26603 }, { "epoch": 1.4183504824865383, "grad_norm": 1.3572920560836792, "learning_rate": 4.6794000000000005e-06, "loss": 0.0338, "step": 26604 }, { "epoch": 1.4184037959161913, "grad_norm": 1.2879751920700073, "learning_rate": 4.6792e-06, "loss": 0.0051, "step": 26605 }, { "epoch": 1.4184571093458442, "grad_norm": 0.2280586212873459, "learning_rate": 4.6790000000000004e-06, "loss": 0.0021, "step": 26606 }, { "epoch": 1.4185104227754972, "grad_norm": 4.3511881828308105, "learning_rate": 4.678800000000001e-06, "loss": 0.0423, "step": 26607 }, { "epoch": 1.41856373620515, "grad_norm": 1.4450191259384155, "learning_rate": 4.6786e-06, "loss": 0.0147, "step": 26608 }, { "epoch": 1.418617049634803, "grad_norm": 1.0854514837265015, "learning_rate": 4.6784e-06, "loss": 0.0116, "step": 26609 }, { "epoch": 1.418670363064456, "grad_norm": 2.2548868656158447, "learning_rate": 4.6782e-06, "loss": 0.0443, "step": 26610 }, { "epoch": 1.418723676494109, "grad_norm": 1.5513643026351929, "learning_rate": 4.678e-06, "loss": 0.0257, "step": 26611 }, { "epoch": 1.4187769899237619, "grad_norm": 1.8641704320907593, "learning_rate": 4.6778e-06, "loss": 0.0271, "step": 26612 }, { "epoch": 1.4188303033534146, "grad_norm": 0.6306460499763489, "learning_rate": 4.677600000000001e-06, "loss": 0.0089, "step": 26613 }, { "epoch": 1.4188836167830676, "grad_norm": 1.8816269636154175, "learning_rate": 4.6774e-06, "loss": 0.026, "step": 26614 }, { "epoch": 1.4189369302127206, "grad_norm": 1.4650782346725464, "learning_rate": 4.677200000000001e-06, "loss": 0.0125, "step": 26615 }, { "epoch": 1.4189902436423736, "grad_norm": 2.2230803966522217, "learning_rate": 4.677e-06, "loss": 0.0312, "step": 26616 }, { "epoch": 1.4190435570720266, "grad_norm": 1.2698172330856323, "learning_rate": 4.6768e-06, "loss": 0.0159, "step": 26617 }, { "epoch": 1.4190968705016793, "grad_norm": 3.948385238647461, "learning_rate": 4.6766e-06, "loss": 0.0304, "step": 26618 }, { "epoch": 1.4191501839313323, "grad_norm": 1.7178471088409424, "learning_rate": 4.676400000000001e-06, "loss": 0.0265, "step": 26619 }, { "epoch": 1.4192034973609853, "grad_norm": 3.5014357566833496, "learning_rate": 4.6762e-06, "loss": 0.0404, "step": 26620 }, { "epoch": 1.419256810790638, "grad_norm": 2.8012473583221436, "learning_rate": 4.676000000000001e-06, "loss": 0.0282, "step": 26621 }, { "epoch": 1.419310124220291, "grad_norm": 1.6890170574188232, "learning_rate": 4.6758e-06, "loss": 0.0154, "step": 26622 }, { "epoch": 1.419363437649944, "grad_norm": 1.9090659618377686, "learning_rate": 4.6756e-06, "loss": 0.0238, "step": 26623 }, { "epoch": 1.419416751079597, "grad_norm": 1.373477816581726, "learning_rate": 4.6754e-06, "loss": 0.019, "step": 26624 }, { "epoch": 1.41947006450925, "grad_norm": 2.034536361694336, "learning_rate": 4.6752000000000006e-06, "loss": 0.0132, "step": 26625 }, { "epoch": 1.4195233779389027, "grad_norm": 1.9284383058547974, "learning_rate": 4.675000000000001e-06, "loss": 0.0547, "step": 26626 }, { "epoch": 1.4195766913685557, "grad_norm": 1.8532100915908813, "learning_rate": 4.6748000000000005e-06, "loss": 0.0126, "step": 26627 }, { "epoch": 1.4196300047982087, "grad_norm": 1.4090749025344849, "learning_rate": 4.6746e-06, "loss": 0.0163, "step": 26628 }, { "epoch": 1.4196833182278616, "grad_norm": 1.8324165344238281, "learning_rate": 4.6744000000000005e-06, "loss": 0.0416, "step": 26629 }, { "epoch": 1.4197366316575146, "grad_norm": 1.6489248275756836, "learning_rate": 4.6742e-06, "loss": 0.0297, "step": 26630 }, { "epoch": 1.4197899450871674, "grad_norm": 1.5973420143127441, "learning_rate": 4.6740000000000005e-06, "loss": 0.0138, "step": 26631 }, { "epoch": 1.4198432585168204, "grad_norm": 1.8452799320220947, "learning_rate": 4.673800000000001e-06, "loss": 0.0143, "step": 26632 }, { "epoch": 1.4198965719464733, "grad_norm": 4.093749046325684, "learning_rate": 4.6736000000000005e-06, "loss": 0.0262, "step": 26633 }, { "epoch": 1.4199498853761263, "grad_norm": 1.920354962348938, "learning_rate": 4.6734e-06, "loss": 0.0338, "step": 26634 }, { "epoch": 1.4200031988057793, "grad_norm": 1.96952223777771, "learning_rate": 4.6732e-06, "loss": 0.0406, "step": 26635 }, { "epoch": 1.420056512235432, "grad_norm": 1.8132030963897705, "learning_rate": 4.673e-06, "loss": 0.0352, "step": 26636 }, { "epoch": 1.420109825665085, "grad_norm": 3.957277774810791, "learning_rate": 4.6728e-06, "loss": 0.0366, "step": 26637 }, { "epoch": 1.420163139094738, "grad_norm": 2.4599575996398926, "learning_rate": 4.672600000000001e-06, "loss": 0.0317, "step": 26638 }, { "epoch": 1.4202164525243908, "grad_norm": 1.9447307586669922, "learning_rate": 4.6724e-06, "loss": 0.0131, "step": 26639 }, { "epoch": 1.4202697659540438, "grad_norm": 1.5094592571258545, "learning_rate": 4.672200000000001e-06, "loss": 0.0129, "step": 26640 }, { "epoch": 1.4203230793836967, "grad_norm": 1.0130795240402222, "learning_rate": 4.672e-06, "loss": 0.018, "step": 26641 }, { "epoch": 1.4203763928133497, "grad_norm": 3.173680305480957, "learning_rate": 4.6718e-06, "loss": 0.0374, "step": 26642 }, { "epoch": 1.4204297062430027, "grad_norm": 2.514110803604126, "learning_rate": 4.6716e-06, "loss": 0.0296, "step": 26643 }, { "epoch": 1.4204830196726554, "grad_norm": 2.1729094982147217, "learning_rate": 4.671400000000001e-06, "loss": 0.0317, "step": 26644 }, { "epoch": 1.4205363331023084, "grad_norm": 1.5048489570617676, "learning_rate": 4.6712e-06, "loss": 0.0331, "step": 26645 }, { "epoch": 1.4205896465319614, "grad_norm": 1.4057174921035767, "learning_rate": 4.671000000000001e-06, "loss": 0.0381, "step": 26646 }, { "epoch": 1.4206429599616144, "grad_norm": 2.147522211074829, "learning_rate": 4.6708e-06, "loss": 0.0585, "step": 26647 }, { "epoch": 1.4206962733912674, "grad_norm": 2.282388687133789, "learning_rate": 4.6706e-06, "loss": 0.0329, "step": 26648 }, { "epoch": 1.4207495868209201, "grad_norm": 1.2358804941177368, "learning_rate": 4.6704e-06, "loss": 0.0167, "step": 26649 }, { "epoch": 1.420802900250573, "grad_norm": 1.795288324356079, "learning_rate": 4.670200000000001e-06, "loss": 0.0318, "step": 26650 }, { "epoch": 1.420856213680226, "grad_norm": 1.9428367614746094, "learning_rate": 4.670000000000001e-06, "loss": 0.037, "step": 26651 }, { "epoch": 1.420909527109879, "grad_norm": 1.372044324874878, "learning_rate": 4.669800000000001e-06, "loss": 0.0091, "step": 26652 }, { "epoch": 1.420962840539532, "grad_norm": 1.282517433166504, "learning_rate": 4.6696e-06, "loss": 0.0117, "step": 26653 }, { "epoch": 1.4210161539691848, "grad_norm": 2.830348491668701, "learning_rate": 4.6694000000000006e-06, "loss": 0.0241, "step": 26654 }, { "epoch": 1.4210694673988378, "grad_norm": 1.744473934173584, "learning_rate": 4.6692e-06, "loss": 0.0182, "step": 26655 }, { "epoch": 1.4211227808284908, "grad_norm": 1.4512758255004883, "learning_rate": 4.6690000000000005e-06, "loss": 0.0208, "step": 26656 }, { "epoch": 1.4211760942581435, "grad_norm": 1.435348629951477, "learning_rate": 4.668800000000001e-06, "loss": 0.0254, "step": 26657 }, { "epoch": 1.4212294076877965, "grad_norm": 1.7184545993804932, "learning_rate": 4.6686000000000005e-06, "loss": 0.0194, "step": 26658 }, { "epoch": 1.4212827211174495, "grad_norm": 0.6403416991233826, "learning_rate": 4.6684e-06, "loss": 0.0062, "step": 26659 }, { "epoch": 1.4213360345471024, "grad_norm": 4.812218189239502, "learning_rate": 4.6682000000000005e-06, "loss": 0.0558, "step": 26660 }, { "epoch": 1.4213893479767554, "grad_norm": 1.2988001108169556, "learning_rate": 4.668e-06, "loss": 0.0248, "step": 26661 }, { "epoch": 1.4214426614064082, "grad_norm": 1.761112093925476, "learning_rate": 4.6678000000000004e-06, "loss": 0.0416, "step": 26662 }, { "epoch": 1.4214959748360612, "grad_norm": 1.81768000125885, "learning_rate": 4.667600000000001e-06, "loss": 0.0145, "step": 26663 }, { "epoch": 1.4215492882657141, "grad_norm": 2.2820372581481934, "learning_rate": 4.6674e-06, "loss": 0.0279, "step": 26664 }, { "epoch": 1.4216026016953671, "grad_norm": 1.8889909982681274, "learning_rate": 4.667200000000001e-06, "loss": 0.0209, "step": 26665 }, { "epoch": 1.42165591512502, "grad_norm": 4.761124134063721, "learning_rate": 4.667e-06, "loss": 0.04, "step": 26666 }, { "epoch": 1.4217092285546729, "grad_norm": 0.940123975276947, "learning_rate": 4.6668e-06, "loss": 0.012, "step": 26667 }, { "epoch": 1.4217625419843258, "grad_norm": 1.4165772199630737, "learning_rate": 4.6666e-06, "loss": 0.0244, "step": 26668 }, { "epoch": 1.4218158554139788, "grad_norm": 1.3348407745361328, "learning_rate": 4.666400000000001e-06, "loss": 0.0365, "step": 26669 }, { "epoch": 1.4218691688436318, "grad_norm": 0.8570552468299866, "learning_rate": 4.6662e-06, "loss": 0.0059, "step": 26670 }, { "epoch": 1.4219224822732848, "grad_norm": 1.1426150798797607, "learning_rate": 4.666000000000001e-06, "loss": 0.0265, "step": 26671 }, { "epoch": 1.4219757957029375, "grad_norm": 1.4990692138671875, "learning_rate": 4.6658e-06, "loss": 0.0214, "step": 26672 }, { "epoch": 1.4220291091325905, "grad_norm": 1.1811296939849854, "learning_rate": 4.6656e-06, "loss": 0.0118, "step": 26673 }, { "epoch": 1.4220824225622435, "grad_norm": 1.2509913444519043, "learning_rate": 4.6654e-06, "loss": 0.0099, "step": 26674 }, { "epoch": 1.4221357359918962, "grad_norm": 1.653067946434021, "learning_rate": 4.665200000000001e-06, "loss": 0.0217, "step": 26675 }, { "epoch": 1.4221890494215492, "grad_norm": 1.7442342042922974, "learning_rate": 4.665e-06, "loss": 0.0153, "step": 26676 }, { "epoch": 1.4222423628512022, "grad_norm": 1.333763599395752, "learning_rate": 4.664800000000001e-06, "loss": 0.018, "step": 26677 }, { "epoch": 1.4222956762808552, "grad_norm": 2.6540424823760986, "learning_rate": 4.6646e-06, "loss": 0.0338, "step": 26678 }, { "epoch": 1.4223489897105082, "grad_norm": 1.0818572044372559, "learning_rate": 4.664400000000001e-06, "loss": 0.0092, "step": 26679 }, { "epoch": 1.422402303140161, "grad_norm": 1.7570072412490845, "learning_rate": 4.6642e-06, "loss": 0.0238, "step": 26680 }, { "epoch": 1.422455616569814, "grad_norm": 1.5309908390045166, "learning_rate": 4.664000000000001e-06, "loss": 0.0285, "step": 26681 }, { "epoch": 1.4225089299994669, "grad_norm": 2.2923684120178223, "learning_rate": 4.6638e-06, "loss": 0.0548, "step": 26682 }, { "epoch": 1.4225622434291199, "grad_norm": 1.4127010107040405, "learning_rate": 4.6636000000000005e-06, "loss": 0.0326, "step": 26683 }, { "epoch": 1.4226155568587728, "grad_norm": 2.0331790447235107, "learning_rate": 4.6634e-06, "loss": 0.0268, "step": 26684 }, { "epoch": 1.4226688702884256, "grad_norm": 1.1953680515289307, "learning_rate": 4.6632000000000005e-06, "loss": 0.0179, "step": 26685 }, { "epoch": 1.4227221837180786, "grad_norm": 2.078251361846924, "learning_rate": 4.663e-06, "loss": 0.026, "step": 26686 }, { "epoch": 1.4227754971477315, "grad_norm": 1.5084519386291504, "learning_rate": 4.6628000000000005e-06, "loss": 0.0233, "step": 26687 }, { "epoch": 1.4228288105773843, "grad_norm": 2.4105899333953857, "learning_rate": 4.6626e-06, "loss": 0.0393, "step": 26688 }, { "epoch": 1.4228821240070375, "grad_norm": 1.6341651678085327, "learning_rate": 4.6624000000000005e-06, "loss": 0.0257, "step": 26689 }, { "epoch": 1.4229354374366903, "grad_norm": 1.892109990119934, "learning_rate": 4.662200000000001e-06, "loss": 0.0269, "step": 26690 }, { "epoch": 1.4229887508663432, "grad_norm": 1.9473897218704224, "learning_rate": 4.6620000000000004e-06, "loss": 0.0425, "step": 26691 }, { "epoch": 1.4230420642959962, "grad_norm": 1.8130760192871094, "learning_rate": 4.6618e-06, "loss": 0.0245, "step": 26692 }, { "epoch": 1.423095377725649, "grad_norm": 1.9729444980621338, "learning_rate": 4.6616e-06, "loss": 0.0277, "step": 26693 }, { "epoch": 1.423148691155302, "grad_norm": 1.173761010169983, "learning_rate": 4.6614e-06, "loss": 0.0115, "step": 26694 }, { "epoch": 1.423202004584955, "grad_norm": 2.0750458240509033, "learning_rate": 4.6612e-06, "loss": 0.0239, "step": 26695 }, { "epoch": 1.423255318014608, "grad_norm": 1.216989517211914, "learning_rate": 4.661000000000001e-06, "loss": 0.0215, "step": 26696 }, { "epoch": 1.423308631444261, "grad_norm": 2.8904170989990234, "learning_rate": 4.6608e-06, "loss": 0.0386, "step": 26697 }, { "epoch": 1.4233619448739137, "grad_norm": 2.6641716957092285, "learning_rate": 4.6606e-06, "loss": 0.0281, "step": 26698 }, { "epoch": 1.4234152583035666, "grad_norm": 1.062669277191162, "learning_rate": 4.6604e-06, "loss": 0.0078, "step": 26699 }, { "epoch": 1.4234685717332196, "grad_norm": 1.8403037786483765, "learning_rate": 4.6602e-06, "loss": 0.0182, "step": 26700 }, { "epoch": 1.4235218851628726, "grad_norm": 1.12553071975708, "learning_rate": 4.66e-06, "loss": 0.0158, "step": 26701 }, { "epoch": 1.4235751985925256, "grad_norm": 2.543104648590088, "learning_rate": 4.659800000000001e-06, "loss": 0.0388, "step": 26702 }, { "epoch": 1.4236285120221783, "grad_norm": 1.082245111465454, "learning_rate": 4.6596e-06, "loss": 0.018, "step": 26703 }, { "epoch": 1.4236818254518313, "grad_norm": 2.055295705795288, "learning_rate": 4.659400000000001e-06, "loss": 0.0248, "step": 26704 }, { "epoch": 1.4237351388814843, "grad_norm": 1.8120002746582031, "learning_rate": 4.6592e-06, "loss": 0.032, "step": 26705 }, { "epoch": 1.423788452311137, "grad_norm": 1.4750640392303467, "learning_rate": 4.659e-06, "loss": 0.0424, "step": 26706 }, { "epoch": 1.42384176574079, "grad_norm": 0.9482927918434143, "learning_rate": 4.6588e-06, "loss": 0.0091, "step": 26707 }, { "epoch": 1.423895079170443, "grad_norm": 1.2554614543914795, "learning_rate": 4.658600000000001e-06, "loss": 0.0111, "step": 26708 }, { "epoch": 1.423948392600096, "grad_norm": 0.9865960478782654, "learning_rate": 4.6584e-06, "loss": 0.008, "step": 26709 }, { "epoch": 1.424001706029749, "grad_norm": 2.05432391166687, "learning_rate": 4.6582000000000006e-06, "loss": 0.0335, "step": 26710 }, { "epoch": 1.4240550194594017, "grad_norm": 1.1181578636169434, "learning_rate": 4.658e-06, "loss": 0.0147, "step": 26711 }, { "epoch": 1.4241083328890547, "grad_norm": 2.200864553451538, "learning_rate": 4.6578000000000005e-06, "loss": 0.0484, "step": 26712 }, { "epoch": 1.4241616463187077, "grad_norm": 1.9032210111618042, "learning_rate": 4.6576e-06, "loss": 0.022, "step": 26713 }, { "epoch": 1.4242149597483607, "grad_norm": 2.3798751831054688, "learning_rate": 4.6574000000000005e-06, "loss": 0.042, "step": 26714 }, { "epoch": 1.4242682731780136, "grad_norm": 2.344576835632324, "learning_rate": 4.657200000000001e-06, "loss": 0.0276, "step": 26715 }, { "epoch": 1.4243215866076664, "grad_norm": 1.5313607454299927, "learning_rate": 4.6570000000000005e-06, "loss": 0.0231, "step": 26716 }, { "epoch": 1.4243749000373194, "grad_norm": 2.052778720855713, "learning_rate": 4.6568e-06, "loss": 0.0109, "step": 26717 }, { "epoch": 1.4244282134669723, "grad_norm": 0.6099914312362671, "learning_rate": 4.6566000000000004e-06, "loss": 0.0057, "step": 26718 }, { "epoch": 1.4244815268966253, "grad_norm": 1.2783151865005493, "learning_rate": 4.6564e-06, "loss": 0.0231, "step": 26719 }, { "epoch": 1.4245348403262783, "grad_norm": 1.2249770164489746, "learning_rate": 4.6562e-06, "loss": 0.0096, "step": 26720 }, { "epoch": 1.424588153755931, "grad_norm": 1.1792354583740234, "learning_rate": 4.656000000000001e-06, "loss": 0.0119, "step": 26721 }, { "epoch": 1.424641467185584, "grad_norm": 3.5501866340637207, "learning_rate": 4.6558e-06, "loss": 0.0518, "step": 26722 }, { "epoch": 1.424694780615237, "grad_norm": 1.5756044387817383, "learning_rate": 4.6556e-06, "loss": 0.0269, "step": 26723 }, { "epoch": 1.4247480940448898, "grad_norm": 2.0071730613708496, "learning_rate": 4.6554e-06, "loss": 0.0129, "step": 26724 }, { "epoch": 1.4248014074745428, "grad_norm": 1.0338060855865479, "learning_rate": 4.6552e-06, "loss": 0.0097, "step": 26725 }, { "epoch": 1.4248547209041957, "grad_norm": 1.601919174194336, "learning_rate": 4.655e-06, "loss": 0.0233, "step": 26726 }, { "epoch": 1.4249080343338487, "grad_norm": 1.6734156608581543, "learning_rate": 4.654800000000001e-06, "loss": 0.0184, "step": 26727 }, { "epoch": 1.4249613477635017, "grad_norm": 2.075232744216919, "learning_rate": 4.6546e-06, "loss": 0.0172, "step": 26728 }, { "epoch": 1.4250146611931545, "grad_norm": 1.6729865074157715, "learning_rate": 4.654400000000001e-06, "loss": 0.0128, "step": 26729 }, { "epoch": 1.4250679746228074, "grad_norm": 1.629256010055542, "learning_rate": 4.6542e-06, "loss": 0.0297, "step": 26730 }, { "epoch": 1.4251212880524604, "grad_norm": 1.8102682828903198, "learning_rate": 4.654e-06, "loss": 0.0286, "step": 26731 }, { "epoch": 1.4251746014821134, "grad_norm": 1.3932076692581177, "learning_rate": 4.6538e-06, "loss": 0.0339, "step": 26732 }, { "epoch": 1.4252279149117664, "grad_norm": 1.405389666557312, "learning_rate": 4.653600000000001e-06, "loss": 0.0093, "step": 26733 }, { "epoch": 1.4252812283414191, "grad_norm": 2.4908804893493652, "learning_rate": 4.6534e-06, "loss": 0.0419, "step": 26734 }, { "epoch": 1.425334541771072, "grad_norm": 2.0424110889434814, "learning_rate": 4.653200000000001e-06, "loss": 0.0165, "step": 26735 }, { "epoch": 1.425387855200725, "grad_norm": 1.793928861618042, "learning_rate": 4.653e-06, "loss": 0.0245, "step": 26736 }, { "epoch": 1.425441168630378, "grad_norm": 1.4647046327590942, "learning_rate": 4.652800000000001e-06, "loss": 0.036, "step": 26737 }, { "epoch": 1.425494482060031, "grad_norm": 2.526614189147949, "learning_rate": 4.6526e-06, "loss": 0.0394, "step": 26738 }, { "epoch": 1.4255477954896838, "grad_norm": 1.4748713970184326, "learning_rate": 4.6524000000000006e-06, "loss": 0.0304, "step": 26739 }, { "epoch": 1.4256011089193368, "grad_norm": 4.066226005554199, "learning_rate": 4.652200000000001e-06, "loss": 0.029, "step": 26740 }, { "epoch": 1.4256544223489898, "grad_norm": 2.160397529602051, "learning_rate": 4.6520000000000005e-06, "loss": 0.0418, "step": 26741 }, { "epoch": 1.4257077357786425, "grad_norm": 1.5090230703353882, "learning_rate": 4.6518e-06, "loss": 0.0083, "step": 26742 }, { "epoch": 1.4257610492082955, "grad_norm": 2.1665031909942627, "learning_rate": 4.6516000000000005e-06, "loss": 0.015, "step": 26743 }, { "epoch": 1.4258143626379485, "grad_norm": 2.0923831462860107, "learning_rate": 4.6514e-06, "loss": 0.0317, "step": 26744 }, { "epoch": 1.4258676760676015, "grad_norm": 1.7690284252166748, "learning_rate": 4.6512000000000005e-06, "loss": 0.0297, "step": 26745 }, { "epoch": 1.4259209894972544, "grad_norm": 2.9583911895751953, "learning_rate": 4.651000000000001e-06, "loss": 0.0254, "step": 26746 }, { "epoch": 1.4259743029269072, "grad_norm": 0.9457014203071594, "learning_rate": 4.6508000000000004e-06, "loss": 0.0156, "step": 26747 }, { "epoch": 1.4260276163565602, "grad_norm": 3.445810317993164, "learning_rate": 4.6506e-06, "loss": 0.0362, "step": 26748 }, { "epoch": 1.4260809297862131, "grad_norm": 2.5768373012542725, "learning_rate": 4.6504e-06, "loss": 0.035, "step": 26749 }, { "epoch": 1.4261342432158661, "grad_norm": 0.7678387761116028, "learning_rate": 4.6502e-06, "loss": 0.0067, "step": 26750 }, { "epoch": 1.426187556645519, "grad_norm": 2.519141912460327, "learning_rate": 4.65e-06, "loss": 0.021, "step": 26751 }, { "epoch": 1.4262408700751719, "grad_norm": 1.1592859029769897, "learning_rate": 4.649800000000001e-06, "loss": 0.0169, "step": 26752 }, { "epoch": 1.4262941835048248, "grad_norm": 2.754847288131714, "learning_rate": 4.6496e-06, "loss": 0.0292, "step": 26753 }, { "epoch": 1.4263474969344778, "grad_norm": 4.174229621887207, "learning_rate": 4.649400000000001e-06, "loss": 0.0366, "step": 26754 }, { "epoch": 1.4264008103641308, "grad_norm": 0.3082829713821411, "learning_rate": 4.6492e-06, "loss": 0.0031, "step": 26755 }, { "epoch": 1.4264541237937838, "grad_norm": 1.453187346458435, "learning_rate": 4.649e-06, "loss": 0.0206, "step": 26756 }, { "epoch": 1.4265074372234365, "grad_norm": 0.6480652689933777, "learning_rate": 4.6488e-06, "loss": 0.0047, "step": 26757 }, { "epoch": 1.4265607506530895, "grad_norm": 1.1122452020645142, "learning_rate": 4.648600000000001e-06, "loss": 0.0109, "step": 26758 }, { "epoch": 1.4266140640827425, "grad_norm": 1.6436010599136353, "learning_rate": 4.6484e-06, "loss": 0.023, "step": 26759 }, { "epoch": 1.4266673775123953, "grad_norm": 1.3173946142196655, "learning_rate": 4.648200000000001e-06, "loss": 0.0229, "step": 26760 }, { "epoch": 1.4267206909420482, "grad_norm": 1.6591590642929077, "learning_rate": 4.648e-06, "loss": 0.021, "step": 26761 }, { "epoch": 1.4267740043717012, "grad_norm": 2.928359031677246, "learning_rate": 4.647800000000001e-06, "loss": 0.0458, "step": 26762 }, { "epoch": 1.4268273178013542, "grad_norm": 1.3425027132034302, "learning_rate": 4.6476e-06, "loss": 0.0116, "step": 26763 }, { "epoch": 1.4268806312310072, "grad_norm": 1.6826250553131104, "learning_rate": 4.647400000000001e-06, "loss": 0.0227, "step": 26764 }, { "epoch": 1.42693394466066, "grad_norm": 1.2931864261627197, "learning_rate": 4.647200000000001e-06, "loss": 0.0131, "step": 26765 }, { "epoch": 1.426987258090313, "grad_norm": 1.2431890964508057, "learning_rate": 4.6470000000000006e-06, "loss": 0.0088, "step": 26766 }, { "epoch": 1.4270405715199659, "grad_norm": 1.0059659481048584, "learning_rate": 4.6468e-06, "loss": 0.0079, "step": 26767 }, { "epoch": 1.4270938849496189, "grad_norm": 0.7827427387237549, "learning_rate": 4.6466000000000005e-06, "loss": 0.0078, "step": 26768 }, { "epoch": 1.4271471983792718, "grad_norm": 2.038412570953369, "learning_rate": 4.6464e-06, "loss": 0.0243, "step": 26769 }, { "epoch": 1.4272005118089246, "grad_norm": 1.708743691444397, "learning_rate": 4.6462000000000005e-06, "loss": 0.0223, "step": 26770 }, { "epoch": 1.4272538252385776, "grad_norm": 0.7788985371589661, "learning_rate": 4.646000000000001e-06, "loss": 0.0074, "step": 26771 }, { "epoch": 1.4273071386682306, "grad_norm": 3.5521531105041504, "learning_rate": 4.6458000000000005e-06, "loss": 0.0355, "step": 26772 }, { "epoch": 1.4273604520978835, "grad_norm": 1.716140627861023, "learning_rate": 4.6456e-06, "loss": 0.0108, "step": 26773 }, { "epoch": 1.4274137655275365, "grad_norm": 2.020209312438965, "learning_rate": 4.6454000000000004e-06, "loss": 0.0201, "step": 26774 }, { "epoch": 1.4274670789571893, "grad_norm": 1.750599980354309, "learning_rate": 4.6452e-06, "loss": 0.0143, "step": 26775 }, { "epoch": 1.4275203923868423, "grad_norm": 2.973517894744873, "learning_rate": 4.645e-06, "loss": 0.0243, "step": 26776 }, { "epoch": 1.4275737058164952, "grad_norm": 0.5163434743881226, "learning_rate": 4.644800000000001e-06, "loss": 0.0047, "step": 26777 }, { "epoch": 1.427627019246148, "grad_norm": 2.065448522567749, "learning_rate": 4.6446e-06, "loss": 0.0316, "step": 26778 }, { "epoch": 1.427680332675801, "grad_norm": 1.3037376403808594, "learning_rate": 4.644400000000001e-06, "loss": 0.0467, "step": 26779 }, { "epoch": 1.427733646105454, "grad_norm": 2.055509090423584, "learning_rate": 4.6442e-06, "loss": 0.0496, "step": 26780 }, { "epoch": 1.427786959535107, "grad_norm": 1.755273699760437, "learning_rate": 4.644e-06, "loss": 0.0234, "step": 26781 }, { "epoch": 1.42784027296476, "grad_norm": 1.1270689964294434, "learning_rate": 4.6438e-06, "loss": 0.008, "step": 26782 }, { "epoch": 1.4278935863944127, "grad_norm": 1.1013412475585938, "learning_rate": 4.643600000000001e-06, "loss": 0.0068, "step": 26783 }, { "epoch": 1.4279468998240656, "grad_norm": 2.280647039413452, "learning_rate": 4.6434e-06, "loss": 0.021, "step": 26784 }, { "epoch": 1.4280002132537186, "grad_norm": 1.6497143507003784, "learning_rate": 4.643200000000001e-06, "loss": 0.0165, "step": 26785 }, { "epoch": 1.4280535266833716, "grad_norm": 0.9364615082740784, "learning_rate": 4.643e-06, "loss": 0.0063, "step": 26786 }, { "epoch": 1.4281068401130246, "grad_norm": 1.9890331029891968, "learning_rate": 4.642800000000001e-06, "loss": 0.0338, "step": 26787 }, { "epoch": 1.4281601535426773, "grad_norm": 0.4554603397846222, "learning_rate": 4.6426e-06, "loss": 0.0044, "step": 26788 }, { "epoch": 1.4282134669723303, "grad_norm": 1.5327647924423218, "learning_rate": 4.642400000000001e-06, "loss": 0.0231, "step": 26789 }, { "epoch": 1.4282667804019833, "grad_norm": 1.1331895589828491, "learning_rate": 4.6422e-06, "loss": 0.0183, "step": 26790 }, { "epoch": 1.428320093831636, "grad_norm": 2.42038893699646, "learning_rate": 4.642000000000001e-06, "loss": 0.0442, "step": 26791 }, { "epoch": 1.4283734072612893, "grad_norm": 2.6462771892547607, "learning_rate": 4.6418e-06, "loss": 0.052, "step": 26792 }, { "epoch": 1.428426720690942, "grad_norm": 0.86607825756073, "learning_rate": 4.641600000000001e-06, "loss": 0.0172, "step": 26793 }, { "epoch": 1.428480034120595, "grad_norm": 2.9278903007507324, "learning_rate": 4.6414e-06, "loss": 0.0328, "step": 26794 }, { "epoch": 1.428533347550248, "grad_norm": 1.9865702390670776, "learning_rate": 4.6412000000000006e-06, "loss": 0.0153, "step": 26795 }, { "epoch": 1.4285866609799007, "grad_norm": 2.5620510578155518, "learning_rate": 4.641e-06, "loss": 0.0186, "step": 26796 }, { "epoch": 1.4286399744095537, "grad_norm": 2.2171342372894287, "learning_rate": 4.6408000000000005e-06, "loss": 0.018, "step": 26797 }, { "epoch": 1.4286932878392067, "grad_norm": 1.8748425245285034, "learning_rate": 4.6406e-06, "loss": 0.0137, "step": 26798 }, { "epoch": 1.4287466012688597, "grad_norm": 2.4708292484283447, "learning_rate": 4.6404000000000005e-06, "loss": 0.0361, "step": 26799 }, { "epoch": 1.4287999146985126, "grad_norm": 1.7699860334396362, "learning_rate": 4.6402e-06, "loss": 0.0218, "step": 26800 }, { "epoch": 1.4288532281281654, "grad_norm": 1.6521639823913574, "learning_rate": 4.6400000000000005e-06, "loss": 0.0216, "step": 26801 }, { "epoch": 1.4289065415578184, "grad_norm": 2.5508501529693604, "learning_rate": 4.6398e-06, "loss": 0.0518, "step": 26802 }, { "epoch": 1.4289598549874714, "grad_norm": 1.6605359315872192, "learning_rate": 4.6396000000000004e-06, "loss": 0.0153, "step": 26803 }, { "epoch": 1.4290131684171243, "grad_norm": 1.7977415323257446, "learning_rate": 4.639400000000001e-06, "loss": 0.0335, "step": 26804 }, { "epoch": 1.4290664818467773, "grad_norm": 1.4921709299087524, "learning_rate": 4.6392e-06, "loss": 0.0164, "step": 26805 }, { "epoch": 1.42911979527643, "grad_norm": 2.321601152420044, "learning_rate": 4.639e-06, "loss": 0.0221, "step": 26806 }, { "epoch": 1.429173108706083, "grad_norm": 1.4982244968414307, "learning_rate": 4.6388e-06, "loss": 0.0267, "step": 26807 }, { "epoch": 1.429226422135736, "grad_norm": 3.0187559127807617, "learning_rate": 4.6386e-06, "loss": 0.0475, "step": 26808 }, { "epoch": 1.4292797355653888, "grad_norm": 1.3524816036224365, "learning_rate": 4.6384e-06, "loss": 0.0284, "step": 26809 }, { "epoch": 1.429333048995042, "grad_norm": 1.0213438272476196, "learning_rate": 4.638200000000001e-06, "loss": 0.0179, "step": 26810 }, { "epoch": 1.4293863624246947, "grad_norm": 1.094768762588501, "learning_rate": 4.638e-06, "loss": 0.0101, "step": 26811 }, { "epoch": 1.4294396758543477, "grad_norm": 3.2809391021728516, "learning_rate": 4.637800000000001e-06, "loss": 0.0411, "step": 26812 }, { "epoch": 1.4294929892840007, "grad_norm": 1.4487873315811157, "learning_rate": 4.6376e-06, "loss": 0.0137, "step": 26813 }, { "epoch": 1.4295463027136535, "grad_norm": 2.8118560314178467, "learning_rate": 4.6374e-06, "loss": 0.0263, "step": 26814 }, { "epoch": 1.4295996161433064, "grad_norm": 1.291739821434021, "learning_rate": 4.6372e-06, "loss": 0.0336, "step": 26815 }, { "epoch": 1.4296529295729594, "grad_norm": 1.3774546384811401, "learning_rate": 4.637000000000001e-06, "loss": 0.0351, "step": 26816 }, { "epoch": 1.4297062430026124, "grad_norm": 2.257002592086792, "learning_rate": 4.6368e-06, "loss": 0.0307, "step": 26817 }, { "epoch": 1.4297595564322654, "grad_norm": 1.5988695621490479, "learning_rate": 4.636600000000001e-06, "loss": 0.0178, "step": 26818 }, { "epoch": 1.4298128698619181, "grad_norm": 3.1195762157440186, "learning_rate": 4.6364e-06, "loss": 0.0236, "step": 26819 }, { "epoch": 1.4298661832915711, "grad_norm": 1.689460277557373, "learning_rate": 4.6362e-06, "loss": 0.0101, "step": 26820 }, { "epoch": 1.429919496721224, "grad_norm": 0.8601554036140442, "learning_rate": 4.636e-06, "loss": 0.0078, "step": 26821 }, { "epoch": 1.429972810150877, "grad_norm": 3.8354809284210205, "learning_rate": 4.6358000000000006e-06, "loss": 0.0349, "step": 26822 }, { "epoch": 1.43002612358053, "grad_norm": 1.3203275203704834, "learning_rate": 4.635600000000001e-06, "loss": 0.0209, "step": 26823 }, { "epoch": 1.4300794370101828, "grad_norm": 1.4757184982299805, "learning_rate": 4.6354000000000005e-06, "loss": 0.024, "step": 26824 }, { "epoch": 1.4301327504398358, "grad_norm": 2.4617996215820312, "learning_rate": 4.6352e-06, "loss": 0.0408, "step": 26825 }, { "epoch": 1.4301860638694888, "grad_norm": 1.229678988456726, "learning_rate": 4.6350000000000005e-06, "loss": 0.0129, "step": 26826 }, { "epoch": 1.4302393772991415, "grad_norm": 1.3731582164764404, "learning_rate": 4.6348e-06, "loss": 0.0372, "step": 26827 }, { "epoch": 1.4302926907287945, "grad_norm": 0.9133375287055969, "learning_rate": 4.6346000000000005e-06, "loss": 0.0078, "step": 26828 }, { "epoch": 1.4303460041584475, "grad_norm": 2.7578353881835938, "learning_rate": 4.634400000000001e-06, "loss": 0.0421, "step": 26829 }, { "epoch": 1.4303993175881005, "grad_norm": 1.029800534248352, "learning_rate": 4.6342000000000005e-06, "loss": 0.0114, "step": 26830 }, { "epoch": 1.4304526310177534, "grad_norm": 0.9606483578681946, "learning_rate": 4.634e-06, "loss": 0.0077, "step": 26831 }, { "epoch": 1.4305059444474062, "grad_norm": 1.0787243843078613, "learning_rate": 4.6338000000000004e-06, "loss": 0.0083, "step": 26832 }, { "epoch": 1.4305592578770592, "grad_norm": 2.4200689792633057, "learning_rate": 4.6336e-06, "loss": 0.0306, "step": 26833 }, { "epoch": 1.4306125713067122, "grad_norm": 1.0208945274353027, "learning_rate": 4.6334e-06, "loss": 0.0066, "step": 26834 }, { "epoch": 1.4306658847363651, "grad_norm": 1.243501901626587, "learning_rate": 4.633200000000001e-06, "loss": 0.0147, "step": 26835 }, { "epoch": 1.4307191981660181, "grad_norm": 1.8000773191452026, "learning_rate": 4.633e-06, "loss": 0.0303, "step": 26836 }, { "epoch": 1.4307725115956709, "grad_norm": 1.3896698951721191, "learning_rate": 4.632800000000001e-06, "loss": 0.0243, "step": 26837 }, { "epoch": 1.4308258250253238, "grad_norm": 1.2112798690795898, "learning_rate": 4.6326e-06, "loss": 0.017, "step": 26838 }, { "epoch": 1.4308791384549768, "grad_norm": 3.3906497955322266, "learning_rate": 4.6324e-06, "loss": 0.0192, "step": 26839 }, { "epoch": 1.4309324518846298, "grad_norm": 1.5580624341964722, "learning_rate": 4.6322e-06, "loss": 0.0218, "step": 26840 }, { "epoch": 1.4309857653142828, "grad_norm": 1.5983847379684448, "learning_rate": 4.632000000000001e-06, "loss": 0.0383, "step": 26841 }, { "epoch": 1.4310390787439355, "grad_norm": 1.9297165870666504, "learning_rate": 4.6318e-06, "loss": 0.0373, "step": 26842 }, { "epoch": 1.4310923921735885, "grad_norm": 0.92364901304245, "learning_rate": 4.631600000000001e-06, "loss": 0.0086, "step": 26843 }, { "epoch": 1.4311457056032415, "grad_norm": 1.6365996599197388, "learning_rate": 4.6314e-06, "loss": 0.022, "step": 26844 }, { "epoch": 1.4311990190328943, "grad_norm": 2.05307674407959, "learning_rate": 4.6312e-06, "loss": 0.0249, "step": 26845 }, { "epoch": 1.4312523324625472, "grad_norm": 1.230016827583313, "learning_rate": 4.631e-06, "loss": 0.0234, "step": 26846 }, { "epoch": 1.4313056458922002, "grad_norm": 2.133837938308716, "learning_rate": 4.630800000000001e-06, "loss": 0.0327, "step": 26847 }, { "epoch": 1.4313589593218532, "grad_norm": 1.461546778678894, "learning_rate": 4.630600000000001e-06, "loss": 0.0282, "step": 26848 }, { "epoch": 1.4314122727515062, "grad_norm": 1.4618109464645386, "learning_rate": 4.630400000000001e-06, "loss": 0.0171, "step": 26849 }, { "epoch": 1.431465586181159, "grad_norm": 0.8958249688148499, "learning_rate": 4.6302e-06, "loss": 0.0087, "step": 26850 }, { "epoch": 1.431518899610812, "grad_norm": 4.479893684387207, "learning_rate": 4.6300000000000006e-06, "loss": 0.0242, "step": 26851 }, { "epoch": 1.431572213040465, "grad_norm": 1.7787048816680908, "learning_rate": 4.6298e-06, "loss": 0.0341, "step": 26852 }, { "epoch": 1.4316255264701179, "grad_norm": 0.8413974046707153, "learning_rate": 4.6296000000000005e-06, "loss": 0.0163, "step": 26853 }, { "epoch": 1.4316788398997708, "grad_norm": 2.286975383758545, "learning_rate": 4.629400000000001e-06, "loss": 0.0276, "step": 26854 }, { "epoch": 1.4317321533294236, "grad_norm": 1.1319912672042847, "learning_rate": 4.6292000000000005e-06, "loss": 0.0172, "step": 26855 }, { "epoch": 1.4317854667590766, "grad_norm": 1.889813780784607, "learning_rate": 4.629e-06, "loss": 0.0421, "step": 26856 }, { "epoch": 1.4318387801887296, "grad_norm": 1.3886798620224, "learning_rate": 4.6288000000000005e-06, "loss": 0.0353, "step": 26857 }, { "epoch": 1.4318920936183825, "grad_norm": 1.4202938079833984, "learning_rate": 4.6286e-06, "loss": 0.0269, "step": 26858 }, { "epoch": 1.4319454070480355, "grad_norm": 0.5227537155151367, "learning_rate": 4.6284000000000004e-06, "loss": 0.0043, "step": 26859 }, { "epoch": 1.4319987204776883, "grad_norm": 5.300121784210205, "learning_rate": 4.628200000000001e-06, "loss": 0.019, "step": 26860 }, { "epoch": 1.4320520339073413, "grad_norm": 1.8174465894699097, "learning_rate": 4.628e-06, "loss": 0.0235, "step": 26861 }, { "epoch": 1.4321053473369942, "grad_norm": 1.9498567581176758, "learning_rate": 4.627800000000001e-06, "loss": 0.02, "step": 26862 }, { "epoch": 1.432158660766647, "grad_norm": 1.9122642278671265, "learning_rate": 4.6276e-06, "loss": 0.0192, "step": 26863 }, { "epoch": 1.4322119741963, "grad_norm": 2.6180038452148438, "learning_rate": 4.6274e-06, "loss": 0.0215, "step": 26864 }, { "epoch": 1.432265287625953, "grad_norm": 1.3056632280349731, "learning_rate": 4.6272e-06, "loss": 0.0245, "step": 26865 }, { "epoch": 1.432318601055606, "grad_norm": 1.364149808883667, "learning_rate": 4.627000000000001e-06, "loss": 0.014, "step": 26866 }, { "epoch": 1.432371914485259, "grad_norm": 0.4823068082332611, "learning_rate": 4.6268e-06, "loss": 0.0045, "step": 26867 }, { "epoch": 1.4324252279149117, "grad_norm": 1.3566091060638428, "learning_rate": 4.626600000000001e-06, "loss": 0.0169, "step": 26868 }, { "epoch": 1.4324785413445646, "grad_norm": 1.552601933479309, "learning_rate": 4.6264e-06, "loss": 0.0308, "step": 26869 }, { "epoch": 1.4325318547742176, "grad_norm": 1.804710865020752, "learning_rate": 4.6262e-06, "loss": 0.0233, "step": 26870 }, { "epoch": 1.4325851682038706, "grad_norm": 2.075270414352417, "learning_rate": 4.626e-06, "loss": 0.0107, "step": 26871 }, { "epoch": 1.4326384816335236, "grad_norm": 1.3675981760025024, "learning_rate": 4.625800000000001e-06, "loss": 0.0096, "step": 26872 }, { "epoch": 1.4326917950631763, "grad_norm": 1.2844980955123901, "learning_rate": 4.6256e-06, "loss": 0.0186, "step": 26873 }, { "epoch": 1.4327451084928293, "grad_norm": 1.4304167032241821, "learning_rate": 4.625400000000001e-06, "loss": 0.027, "step": 26874 }, { "epoch": 1.4327984219224823, "grad_norm": 1.9541212320327759, "learning_rate": 4.6252e-06, "loss": 0.0137, "step": 26875 }, { "epoch": 1.4328517353521353, "grad_norm": 1.4657058715820312, "learning_rate": 4.625000000000001e-06, "loss": 0.0428, "step": 26876 }, { "epoch": 1.4329050487817883, "grad_norm": 0.6054814457893372, "learning_rate": 4.6248e-06, "loss": 0.0036, "step": 26877 }, { "epoch": 1.432958362211441, "grad_norm": 1.0587514638900757, "learning_rate": 4.624600000000001e-06, "loss": 0.0149, "step": 26878 }, { "epoch": 1.433011675641094, "grad_norm": 1.1713974475860596, "learning_rate": 4.6244e-06, "loss": 0.0236, "step": 26879 }, { "epoch": 1.433064989070747, "grad_norm": 1.5303997993469238, "learning_rate": 4.6242000000000006e-06, "loss": 0.0327, "step": 26880 }, { "epoch": 1.4331183025003997, "grad_norm": 1.2153093814849854, "learning_rate": 4.624e-06, "loss": 0.0099, "step": 26881 }, { "epoch": 1.4331716159300527, "grad_norm": 2.885758638381958, "learning_rate": 4.6238000000000005e-06, "loss": 0.0205, "step": 26882 }, { "epoch": 1.4332249293597057, "grad_norm": 1.2372348308563232, "learning_rate": 4.6236e-06, "loss": 0.0214, "step": 26883 }, { "epoch": 1.4332782427893587, "grad_norm": 1.4686744213104248, "learning_rate": 4.6234000000000005e-06, "loss": 0.0144, "step": 26884 }, { "epoch": 1.4333315562190116, "grad_norm": 1.553735375404358, "learning_rate": 4.6232e-06, "loss": 0.0297, "step": 26885 }, { "epoch": 1.4333848696486644, "grad_norm": 2.4875102043151855, "learning_rate": 4.6230000000000005e-06, "loss": 0.0119, "step": 26886 }, { "epoch": 1.4334381830783174, "grad_norm": 3.0789599418640137, "learning_rate": 4.622800000000001e-06, "loss": 0.0556, "step": 26887 }, { "epoch": 1.4334914965079704, "grad_norm": 1.4304076433181763, "learning_rate": 4.6226000000000004e-06, "loss": 0.0182, "step": 26888 }, { "epoch": 1.4335448099376233, "grad_norm": 1.604805588722229, "learning_rate": 4.6224e-06, "loss": 0.0097, "step": 26889 }, { "epoch": 1.4335981233672763, "grad_norm": 1.2391576766967773, "learning_rate": 4.6222e-06, "loss": 0.0074, "step": 26890 }, { "epoch": 1.433651436796929, "grad_norm": 1.6770738363265991, "learning_rate": 4.622e-06, "loss": 0.0248, "step": 26891 }, { "epoch": 1.433704750226582, "grad_norm": 1.0023750066757202, "learning_rate": 4.6218e-06, "loss": 0.0181, "step": 26892 }, { "epoch": 1.433758063656235, "grad_norm": 1.5270872116088867, "learning_rate": 4.621600000000001e-06, "loss": 0.014, "step": 26893 }, { "epoch": 1.433811377085888, "grad_norm": 2.0766854286193848, "learning_rate": 4.6214e-06, "loss": 0.0239, "step": 26894 }, { "epoch": 1.433864690515541, "grad_norm": 2.333111047744751, "learning_rate": 4.6212e-06, "loss": 0.0471, "step": 26895 }, { "epoch": 1.4339180039451938, "grad_norm": 2.113206624984741, "learning_rate": 4.621e-06, "loss": 0.0273, "step": 26896 }, { "epoch": 1.4339713173748467, "grad_norm": 1.278135895729065, "learning_rate": 4.6208e-06, "loss": 0.0125, "step": 26897 }, { "epoch": 1.4340246308044997, "grad_norm": 0.752465009689331, "learning_rate": 4.6206e-06, "loss": 0.0042, "step": 26898 }, { "epoch": 1.4340779442341525, "grad_norm": 3.2656266689300537, "learning_rate": 4.620400000000001e-06, "loss": 0.0265, "step": 26899 }, { "epoch": 1.4341312576638054, "grad_norm": 1.8537198305130005, "learning_rate": 4.6202e-06, "loss": 0.0196, "step": 26900 }, { "epoch": 1.4341845710934584, "grad_norm": 1.7840214967727661, "learning_rate": 4.620000000000001e-06, "loss": 0.0238, "step": 26901 }, { "epoch": 1.4342378845231114, "grad_norm": 1.7015799283981323, "learning_rate": 4.6198e-06, "loss": 0.0122, "step": 26902 }, { "epoch": 1.4342911979527644, "grad_norm": 1.4031368494033813, "learning_rate": 4.6196e-06, "loss": 0.0173, "step": 26903 }, { "epoch": 1.4343445113824171, "grad_norm": 2.170271635055542, "learning_rate": 4.6194e-06, "loss": 0.0283, "step": 26904 }, { "epoch": 1.4343978248120701, "grad_norm": 2.0866899490356445, "learning_rate": 4.619200000000001e-06, "loss": 0.024, "step": 26905 }, { "epoch": 1.434451138241723, "grad_norm": 2.319314956665039, "learning_rate": 4.619e-06, "loss": 0.0458, "step": 26906 }, { "epoch": 1.434504451671376, "grad_norm": 0.8715386986732483, "learning_rate": 4.6188000000000006e-06, "loss": 0.0084, "step": 26907 }, { "epoch": 1.434557765101029, "grad_norm": 1.7528856992721558, "learning_rate": 4.6186e-06, "loss": 0.0169, "step": 26908 }, { "epoch": 1.4346110785306818, "grad_norm": 0.6648578643798828, "learning_rate": 4.6184e-06, "loss": 0.0055, "step": 26909 }, { "epoch": 1.4346643919603348, "grad_norm": 2.238161563873291, "learning_rate": 4.6182e-06, "loss": 0.023, "step": 26910 }, { "epoch": 1.4347177053899878, "grad_norm": 1.7309991121292114, "learning_rate": 4.6180000000000005e-06, "loss": 0.017, "step": 26911 }, { "epoch": 1.4347710188196405, "grad_norm": 2.9382119178771973, "learning_rate": 4.617800000000001e-06, "loss": 0.0406, "step": 26912 }, { "epoch": 1.4348243322492937, "grad_norm": 2.2535388469696045, "learning_rate": 4.6176000000000005e-06, "loss": 0.0142, "step": 26913 }, { "epoch": 1.4348776456789465, "grad_norm": 1.8825405836105347, "learning_rate": 4.6174e-06, "loss": 0.0304, "step": 26914 }, { "epoch": 1.4349309591085995, "grad_norm": 1.6985820531845093, "learning_rate": 4.6172000000000004e-06, "loss": 0.0238, "step": 26915 }, { "epoch": 1.4349842725382524, "grad_norm": 2.9210002422332764, "learning_rate": 4.617e-06, "loss": 0.0236, "step": 26916 }, { "epoch": 1.4350375859679052, "grad_norm": 2.098637580871582, "learning_rate": 4.6168e-06, "loss": 0.0118, "step": 26917 }, { "epoch": 1.4350908993975582, "grad_norm": 1.6166881322860718, "learning_rate": 4.616600000000001e-06, "loss": 0.0197, "step": 26918 }, { "epoch": 1.4351442128272112, "grad_norm": 1.4518872499465942, "learning_rate": 4.6164e-06, "loss": 0.0168, "step": 26919 }, { "epoch": 1.4351975262568641, "grad_norm": 0.6170364022254944, "learning_rate": 4.6162e-06, "loss": 0.0059, "step": 26920 }, { "epoch": 1.4352508396865171, "grad_norm": 1.840742588043213, "learning_rate": 4.616e-06, "loss": 0.0405, "step": 26921 }, { "epoch": 1.4353041531161699, "grad_norm": 3.871929407119751, "learning_rate": 4.6158e-06, "loss": 0.0316, "step": 26922 }, { "epoch": 1.4353574665458229, "grad_norm": 1.3310573101043701, "learning_rate": 4.6156e-06, "loss": 0.027, "step": 26923 }, { "epoch": 1.4354107799754758, "grad_norm": 2.8831233978271484, "learning_rate": 4.615400000000001e-06, "loss": 0.0102, "step": 26924 }, { "epoch": 1.4354640934051288, "grad_norm": 1.1771249771118164, "learning_rate": 4.6152e-06, "loss": 0.0085, "step": 26925 }, { "epoch": 1.4355174068347818, "grad_norm": 0.9740510582923889, "learning_rate": 4.615000000000001e-06, "loss": 0.024, "step": 26926 }, { "epoch": 1.4355707202644346, "grad_norm": 0.4175662398338318, "learning_rate": 4.6148e-06, "loss": 0.0038, "step": 26927 }, { "epoch": 1.4356240336940875, "grad_norm": 1.7828326225280762, "learning_rate": 4.6146e-06, "loss": 0.0101, "step": 26928 }, { "epoch": 1.4356773471237405, "grad_norm": 1.5634024143218994, "learning_rate": 4.6144e-06, "loss": 0.0304, "step": 26929 }, { "epoch": 1.4357306605533933, "grad_norm": 2.4837706089019775, "learning_rate": 4.614200000000001e-06, "loss": 0.0255, "step": 26930 }, { "epoch": 1.4357839739830462, "grad_norm": 2.1869089603424072, "learning_rate": 4.614e-06, "loss": 0.031, "step": 26931 }, { "epoch": 1.4358372874126992, "grad_norm": 1.549281358718872, "learning_rate": 4.613800000000001e-06, "loss": 0.0104, "step": 26932 }, { "epoch": 1.4358906008423522, "grad_norm": 1.6457716226577759, "learning_rate": 4.6136e-06, "loss": 0.0217, "step": 26933 }, { "epoch": 1.4359439142720052, "grad_norm": 2.313642740249634, "learning_rate": 4.6134e-06, "loss": 0.0416, "step": 26934 }, { "epoch": 1.435997227701658, "grad_norm": 0.5374698042869568, "learning_rate": 4.6132e-06, "loss": 0.0034, "step": 26935 }, { "epoch": 1.436050541131311, "grad_norm": 1.269640326499939, "learning_rate": 4.6130000000000006e-06, "loss": 0.0248, "step": 26936 }, { "epoch": 1.436103854560964, "grad_norm": 1.3331882953643799, "learning_rate": 4.612800000000001e-06, "loss": 0.0124, "step": 26937 }, { "epoch": 1.4361571679906169, "grad_norm": 2.0780692100524902, "learning_rate": 4.6126000000000005e-06, "loss": 0.0489, "step": 26938 }, { "epoch": 1.4362104814202699, "grad_norm": 4.3202738761901855, "learning_rate": 4.6124e-06, "loss": 0.0488, "step": 26939 }, { "epoch": 1.4362637948499226, "grad_norm": 2.3651883602142334, "learning_rate": 4.6122000000000005e-06, "loss": 0.0247, "step": 26940 }, { "epoch": 1.4363171082795756, "grad_norm": 2.5441462993621826, "learning_rate": 4.612e-06, "loss": 0.0247, "step": 26941 }, { "epoch": 1.4363704217092286, "grad_norm": 2.567946434020996, "learning_rate": 4.6118000000000005e-06, "loss": 0.022, "step": 26942 }, { "epoch": 1.4364237351388816, "grad_norm": 1.4209941625595093, "learning_rate": 4.611600000000001e-06, "loss": 0.0388, "step": 26943 }, { "epoch": 1.4364770485685345, "grad_norm": 2.5850391387939453, "learning_rate": 4.6114000000000004e-06, "loss": 0.0179, "step": 26944 }, { "epoch": 1.4365303619981873, "grad_norm": 0.8964341282844543, "learning_rate": 4.6112e-06, "loss": 0.013, "step": 26945 }, { "epoch": 1.4365836754278403, "grad_norm": 0.8358694911003113, "learning_rate": 4.611e-06, "loss": 0.0132, "step": 26946 }, { "epoch": 1.4366369888574932, "grad_norm": 2.402158260345459, "learning_rate": 4.6108e-06, "loss": 0.0309, "step": 26947 }, { "epoch": 1.436690302287146, "grad_norm": 3.422320604324341, "learning_rate": 4.6106e-06, "loss": 0.0186, "step": 26948 }, { "epoch": 1.436743615716799, "grad_norm": 2.58249568939209, "learning_rate": 4.610400000000001e-06, "loss": 0.043, "step": 26949 }, { "epoch": 1.436796929146452, "grad_norm": 1.715416431427002, "learning_rate": 4.6102e-06, "loss": 0.0171, "step": 26950 }, { "epoch": 1.436850242576105, "grad_norm": 1.5956430435180664, "learning_rate": 4.610000000000001e-06, "loss": 0.0219, "step": 26951 }, { "epoch": 1.436903556005758, "grad_norm": 1.212633728981018, "learning_rate": 4.6098e-06, "loss": 0.0255, "step": 26952 }, { "epoch": 1.4369568694354107, "grad_norm": 2.2791364192962646, "learning_rate": 4.6096e-06, "loss": 0.0256, "step": 26953 }, { "epoch": 1.4370101828650637, "grad_norm": 1.3125325441360474, "learning_rate": 4.6094e-06, "loss": 0.0267, "step": 26954 }, { "epoch": 1.4370634962947166, "grad_norm": 1.1406162977218628, "learning_rate": 4.609200000000001e-06, "loss": 0.0081, "step": 26955 }, { "epoch": 1.4371168097243696, "grad_norm": 1.04442298412323, "learning_rate": 4.609e-06, "loss": 0.0287, "step": 26956 }, { "epoch": 1.4371701231540226, "grad_norm": 1.5012658834457397, "learning_rate": 4.608800000000001e-06, "loss": 0.0329, "step": 26957 }, { "epoch": 1.4372234365836754, "grad_norm": 1.0169752836227417, "learning_rate": 4.6086e-06, "loss": 0.007, "step": 26958 }, { "epoch": 1.4372767500133283, "grad_norm": 1.309512972831726, "learning_rate": 4.608400000000001e-06, "loss": 0.0319, "step": 26959 }, { "epoch": 1.4373300634429813, "grad_norm": 1.6376793384552002, "learning_rate": 4.6082e-06, "loss": 0.0094, "step": 26960 }, { "epoch": 1.4373833768726343, "grad_norm": 1.9911555051803589, "learning_rate": 4.608000000000001e-06, "loss": 0.0261, "step": 26961 }, { "epoch": 1.4374366903022873, "grad_norm": 1.454261064529419, "learning_rate": 4.607800000000001e-06, "loss": 0.0159, "step": 26962 }, { "epoch": 1.43749000373194, "grad_norm": 1.5313032865524292, "learning_rate": 4.6076000000000006e-06, "loss": 0.0169, "step": 26963 }, { "epoch": 1.437543317161593, "grad_norm": 1.9949766397476196, "learning_rate": 4.6074e-06, "loss": 0.0329, "step": 26964 }, { "epoch": 1.437596630591246, "grad_norm": 1.475831389427185, "learning_rate": 4.6072000000000005e-06, "loss": 0.024, "step": 26965 }, { "epoch": 1.4376499440208987, "grad_norm": 1.3464607000350952, "learning_rate": 4.607e-06, "loss": 0.0267, "step": 26966 }, { "epoch": 1.4377032574505517, "grad_norm": 1.6135576963424683, "learning_rate": 4.6068000000000005e-06, "loss": 0.0328, "step": 26967 }, { "epoch": 1.4377565708802047, "grad_norm": 1.665074110031128, "learning_rate": 4.606600000000001e-06, "loss": 0.0297, "step": 26968 }, { "epoch": 1.4378098843098577, "grad_norm": 2.407485246658325, "learning_rate": 4.6064000000000005e-06, "loss": 0.0263, "step": 26969 }, { "epoch": 1.4378631977395107, "grad_norm": 3.865138530731201, "learning_rate": 4.6062e-06, "loss": 0.0561, "step": 26970 }, { "epoch": 1.4379165111691634, "grad_norm": 0.9538931250572205, "learning_rate": 4.6060000000000005e-06, "loss": 0.0114, "step": 26971 }, { "epoch": 1.4379698245988164, "grad_norm": 2.325469970703125, "learning_rate": 4.6058e-06, "loss": 0.0159, "step": 26972 }, { "epoch": 1.4380231380284694, "grad_norm": 1.1307343244552612, "learning_rate": 4.6056000000000004e-06, "loss": 0.0181, "step": 26973 }, { "epoch": 1.4380764514581224, "grad_norm": 1.8947885036468506, "learning_rate": 4.605400000000001e-06, "loss": 0.0351, "step": 26974 }, { "epoch": 1.4381297648877753, "grad_norm": 2.6771483421325684, "learning_rate": 4.6052e-06, "loss": 0.0431, "step": 26975 }, { "epoch": 1.438183078317428, "grad_norm": 4.115346431732178, "learning_rate": 4.605000000000001e-06, "loss": 0.0133, "step": 26976 }, { "epoch": 1.438236391747081, "grad_norm": 1.529662847518921, "learning_rate": 4.6048e-06, "loss": 0.0228, "step": 26977 }, { "epoch": 1.438289705176734, "grad_norm": 2.153917074203491, "learning_rate": 4.6046e-06, "loss": 0.0217, "step": 26978 }, { "epoch": 1.438343018606387, "grad_norm": 2.0213816165924072, "learning_rate": 4.6044e-06, "loss": 0.0322, "step": 26979 }, { "epoch": 1.43839633203604, "grad_norm": 2.5603058338165283, "learning_rate": 4.604200000000001e-06, "loss": 0.0206, "step": 26980 }, { "epoch": 1.4384496454656928, "grad_norm": 1.177695631980896, "learning_rate": 4.604e-06, "loss": 0.009, "step": 26981 }, { "epoch": 1.4385029588953457, "grad_norm": 1.1258376836776733, "learning_rate": 4.603800000000001e-06, "loss": 0.0118, "step": 26982 }, { "epoch": 1.4385562723249987, "grad_norm": 0.9227504730224609, "learning_rate": 4.6036e-06, "loss": 0.0181, "step": 26983 }, { "epoch": 1.4386095857546515, "grad_norm": 4.3704328536987305, "learning_rate": 4.603400000000001e-06, "loss": 0.0421, "step": 26984 }, { "epoch": 1.4386628991843045, "grad_norm": 1.6055163145065308, "learning_rate": 4.6032e-06, "loss": 0.028, "step": 26985 }, { "epoch": 1.4387162126139574, "grad_norm": 1.6981565952301025, "learning_rate": 4.603000000000001e-06, "loss": 0.0323, "step": 26986 }, { "epoch": 1.4387695260436104, "grad_norm": 2.379775047302246, "learning_rate": 4.6028e-06, "loss": 0.0459, "step": 26987 }, { "epoch": 1.4388228394732634, "grad_norm": 1.688250184059143, "learning_rate": 4.602600000000001e-06, "loss": 0.0136, "step": 26988 }, { "epoch": 1.4388761529029162, "grad_norm": 0.9808082580566406, "learning_rate": 4.6024e-06, "loss": 0.0286, "step": 26989 }, { "epoch": 1.4389294663325691, "grad_norm": 1.4698911905288696, "learning_rate": 4.602200000000001e-06, "loss": 0.0111, "step": 26990 }, { "epoch": 1.438982779762222, "grad_norm": 1.4781928062438965, "learning_rate": 4.602e-06, "loss": 0.0227, "step": 26991 }, { "epoch": 1.439036093191875, "grad_norm": 2.1931583881378174, "learning_rate": 4.6018000000000006e-06, "loss": 0.0423, "step": 26992 }, { "epoch": 1.439089406621528, "grad_norm": 1.5857141017913818, "learning_rate": 4.6016e-06, "loss": 0.0107, "step": 26993 }, { "epoch": 1.4391427200511808, "grad_norm": 2.7902920246124268, "learning_rate": 4.6014000000000005e-06, "loss": 0.018, "step": 26994 }, { "epoch": 1.4391960334808338, "grad_norm": 3.8632752895355225, "learning_rate": 4.6012e-06, "loss": 0.0703, "step": 26995 }, { "epoch": 1.4392493469104868, "grad_norm": 1.329968810081482, "learning_rate": 4.6010000000000005e-06, "loss": 0.0224, "step": 26996 }, { "epoch": 1.4393026603401398, "grad_norm": 1.825403094291687, "learning_rate": 4.6008e-06, "loss": 0.0209, "step": 26997 }, { "epoch": 1.4393559737697927, "grad_norm": 1.351246953010559, "learning_rate": 4.6006000000000005e-06, "loss": 0.0126, "step": 26998 }, { "epoch": 1.4394092871994455, "grad_norm": 1.8150649070739746, "learning_rate": 4.6004e-06, "loss": 0.023, "step": 26999 }, { "epoch": 1.4394626006290985, "grad_norm": 1.4704537391662598, "learning_rate": 4.6002000000000004e-06, "loss": 0.0116, "step": 27000 }, { "epoch": 1.4395159140587515, "grad_norm": 0.46179553866386414, "learning_rate": 4.600000000000001e-06, "loss": 0.0042, "step": 27001 }, { "epoch": 1.4395692274884042, "grad_norm": 2.3121750354766846, "learning_rate": 4.5998e-06, "loss": 0.0251, "step": 27002 }, { "epoch": 1.4396225409180572, "grad_norm": 2.0910532474517822, "learning_rate": 4.5996e-06, "loss": 0.0115, "step": 27003 }, { "epoch": 1.4396758543477102, "grad_norm": 3.8810791969299316, "learning_rate": 4.5994e-06, "loss": 0.0427, "step": 27004 }, { "epoch": 1.4397291677773632, "grad_norm": 1.5251396894454956, "learning_rate": 4.5992e-06, "loss": 0.0193, "step": 27005 }, { "epoch": 1.4397824812070161, "grad_norm": 0.6206153035163879, "learning_rate": 4.599e-06, "loss": 0.0056, "step": 27006 }, { "epoch": 1.4398357946366689, "grad_norm": 1.8052610158920288, "learning_rate": 4.598800000000001e-06, "loss": 0.0328, "step": 27007 }, { "epoch": 1.4398891080663219, "grad_norm": 0.8943889141082764, "learning_rate": 4.5986e-06, "loss": 0.0076, "step": 27008 }, { "epoch": 1.4399424214959748, "grad_norm": 1.0066112279891968, "learning_rate": 4.598400000000001e-06, "loss": 0.0089, "step": 27009 }, { "epoch": 1.4399957349256278, "grad_norm": 0.634775698184967, "learning_rate": 4.5982e-06, "loss": 0.0049, "step": 27010 }, { "epoch": 1.4400490483552808, "grad_norm": 2.0906877517700195, "learning_rate": 4.598e-06, "loss": 0.0112, "step": 27011 }, { "epoch": 1.4401023617849336, "grad_norm": 1.268505334854126, "learning_rate": 4.5978e-06, "loss": 0.0118, "step": 27012 }, { "epoch": 1.4401556752145865, "grad_norm": 9.638397216796875, "learning_rate": 4.597600000000001e-06, "loss": 0.036, "step": 27013 }, { "epoch": 1.4402089886442395, "grad_norm": 1.9980521202087402, "learning_rate": 4.5974e-06, "loss": 0.0301, "step": 27014 }, { "epoch": 1.4402623020738923, "grad_norm": 1.593662977218628, "learning_rate": 4.597200000000001e-06, "loss": 0.0196, "step": 27015 }, { "epoch": 1.4403156155035455, "grad_norm": 1.7555413246154785, "learning_rate": 4.597e-06, "loss": 0.012, "step": 27016 }, { "epoch": 1.4403689289331982, "grad_norm": 2.6003682613372803, "learning_rate": 4.5968e-06, "loss": 0.0071, "step": 27017 }, { "epoch": 1.4404222423628512, "grad_norm": 0.4602723717689514, "learning_rate": 4.5966e-06, "loss": 0.0041, "step": 27018 }, { "epoch": 1.4404755557925042, "grad_norm": 0.8363035321235657, "learning_rate": 4.596400000000001e-06, "loss": 0.004, "step": 27019 }, { "epoch": 1.440528869222157, "grad_norm": 0.7212132811546326, "learning_rate": 4.5962e-06, "loss": 0.0224, "step": 27020 }, { "epoch": 1.44058218265181, "grad_norm": 1.3953018188476562, "learning_rate": 4.5960000000000006e-06, "loss": 0.0186, "step": 27021 }, { "epoch": 1.440635496081463, "grad_norm": 2.0752453804016113, "learning_rate": 4.5958e-06, "loss": 0.0272, "step": 27022 }, { "epoch": 1.4406888095111159, "grad_norm": 1.9032855033874512, "learning_rate": 4.5956000000000005e-06, "loss": 0.0288, "step": 27023 }, { "epoch": 1.4407421229407689, "grad_norm": 2.2353339195251465, "learning_rate": 4.5954e-06, "loss": 0.0153, "step": 27024 }, { "epoch": 1.4407954363704216, "grad_norm": 2.6938021183013916, "learning_rate": 4.5952000000000005e-06, "loss": 0.035, "step": 27025 }, { "epoch": 1.4408487498000746, "grad_norm": 1.925140142440796, "learning_rate": 4.595000000000001e-06, "loss": 0.0542, "step": 27026 }, { "epoch": 1.4409020632297276, "grad_norm": 2.1166834831237793, "learning_rate": 4.5948000000000005e-06, "loss": 0.0158, "step": 27027 }, { "epoch": 1.4409553766593806, "grad_norm": 3.9370834827423096, "learning_rate": 4.5946e-06, "loss": 0.032, "step": 27028 }, { "epoch": 1.4410086900890335, "grad_norm": 1.7880102396011353, "learning_rate": 4.5944000000000004e-06, "loss": 0.0091, "step": 27029 }, { "epoch": 1.4410620035186863, "grad_norm": 1.0272685289382935, "learning_rate": 4.5942e-06, "loss": 0.0174, "step": 27030 }, { "epoch": 1.4411153169483393, "grad_norm": 2.2333168983459473, "learning_rate": 4.594e-06, "loss": 0.013, "step": 27031 }, { "epoch": 1.4411686303779923, "grad_norm": 2.248291015625, "learning_rate": 4.593800000000001e-06, "loss": 0.0187, "step": 27032 }, { "epoch": 1.441221943807645, "grad_norm": 1.860929012298584, "learning_rate": 4.5936e-06, "loss": 0.0105, "step": 27033 }, { "epoch": 1.4412752572372982, "grad_norm": 2.1375374794006348, "learning_rate": 4.593400000000001e-06, "loss": 0.0184, "step": 27034 }, { "epoch": 1.441328570666951, "grad_norm": 1.4128748178482056, "learning_rate": 4.5932e-06, "loss": 0.009, "step": 27035 }, { "epoch": 1.441381884096604, "grad_norm": 4.617677211761475, "learning_rate": 4.593e-06, "loss": 0.0403, "step": 27036 }, { "epoch": 1.441435197526257, "grad_norm": 1.8390365839004517, "learning_rate": 4.5928e-06, "loss": 0.0255, "step": 27037 }, { "epoch": 1.4414885109559097, "grad_norm": 2.6527223587036133, "learning_rate": 4.592600000000001e-06, "loss": 0.0476, "step": 27038 }, { "epoch": 1.4415418243855627, "grad_norm": 2.343561887741089, "learning_rate": 4.5924e-06, "loss": 0.0337, "step": 27039 }, { "epoch": 1.4415951378152156, "grad_norm": 0.8978525996208191, "learning_rate": 4.592200000000001e-06, "loss": 0.0081, "step": 27040 }, { "epoch": 1.4416484512448686, "grad_norm": 2.399207353591919, "learning_rate": 4.592e-06, "loss": 0.0212, "step": 27041 }, { "epoch": 1.4417017646745216, "grad_norm": 1.3581255674362183, "learning_rate": 4.5918e-06, "loss": 0.0138, "step": 27042 }, { "epoch": 1.4417550781041744, "grad_norm": 1.7982254028320312, "learning_rate": 4.5916e-06, "loss": 0.0083, "step": 27043 }, { "epoch": 1.4418083915338273, "grad_norm": 2.7161693572998047, "learning_rate": 4.591400000000001e-06, "loss": 0.0536, "step": 27044 }, { "epoch": 1.4418617049634803, "grad_norm": 1.6514636278152466, "learning_rate": 4.5912e-06, "loss": 0.0351, "step": 27045 }, { "epoch": 1.4419150183931333, "grad_norm": 2.561544179916382, "learning_rate": 4.591000000000001e-06, "loss": 0.0651, "step": 27046 }, { "epoch": 1.4419683318227863, "grad_norm": 2.3685379028320312, "learning_rate": 4.5908e-06, "loss": 0.017, "step": 27047 }, { "epoch": 1.442021645252439, "grad_norm": 1.2347416877746582, "learning_rate": 4.5906000000000006e-06, "loss": 0.0342, "step": 27048 }, { "epoch": 1.442074958682092, "grad_norm": 1.132306694984436, "learning_rate": 4.5904e-06, "loss": 0.0139, "step": 27049 }, { "epoch": 1.442128272111745, "grad_norm": 1.437225580215454, "learning_rate": 4.5902000000000005e-06, "loss": 0.0226, "step": 27050 }, { "epoch": 1.4421815855413977, "grad_norm": 1.654579520225525, "learning_rate": 4.590000000000001e-06, "loss": 0.0329, "step": 27051 }, { "epoch": 1.4422348989710507, "grad_norm": 1.5961858034133911, "learning_rate": 4.5898000000000005e-06, "loss": 0.0273, "step": 27052 }, { "epoch": 1.4422882124007037, "grad_norm": 1.2405872344970703, "learning_rate": 4.5896e-06, "loss": 0.0261, "step": 27053 }, { "epoch": 1.4423415258303567, "grad_norm": 0.8721116185188293, "learning_rate": 4.5894000000000005e-06, "loss": 0.0219, "step": 27054 }, { "epoch": 1.4423948392600097, "grad_norm": 1.4777034521102905, "learning_rate": 4.5892e-06, "loss": 0.0283, "step": 27055 }, { "epoch": 1.4424481526896624, "grad_norm": 1.0313674211502075, "learning_rate": 4.5890000000000004e-06, "loss": 0.0254, "step": 27056 }, { "epoch": 1.4425014661193154, "grad_norm": 1.388558030128479, "learning_rate": 4.588800000000001e-06, "loss": 0.0177, "step": 27057 }, { "epoch": 1.4425547795489684, "grad_norm": 2.05657958984375, "learning_rate": 4.5886e-06, "loss": 0.0204, "step": 27058 }, { "epoch": 1.4426080929786214, "grad_norm": 2.995567560195923, "learning_rate": 4.588400000000001e-06, "loss": 0.052, "step": 27059 }, { "epoch": 1.4426614064082743, "grad_norm": 2.071603298187256, "learning_rate": 4.5882e-06, "loss": 0.0227, "step": 27060 }, { "epoch": 1.442714719837927, "grad_norm": 0.7324721813201904, "learning_rate": 4.588e-06, "loss": 0.0082, "step": 27061 }, { "epoch": 1.44276803326758, "grad_norm": 3.9171204566955566, "learning_rate": 4.5878e-06, "loss": 0.0276, "step": 27062 }, { "epoch": 1.442821346697233, "grad_norm": 1.9625484943389893, "learning_rate": 4.587600000000001e-06, "loss": 0.0146, "step": 27063 }, { "epoch": 1.442874660126886, "grad_norm": 1.3289486169815063, "learning_rate": 4.5874e-06, "loss": 0.0127, "step": 27064 }, { "epoch": 1.442927973556539, "grad_norm": 1.4878009557724, "learning_rate": 4.587200000000001e-06, "loss": 0.0221, "step": 27065 }, { "epoch": 1.4429812869861918, "grad_norm": 1.5376839637756348, "learning_rate": 4.587e-06, "loss": 0.0132, "step": 27066 }, { "epoch": 1.4430346004158447, "grad_norm": 0.722083568572998, "learning_rate": 4.5868e-06, "loss": 0.0053, "step": 27067 }, { "epoch": 1.4430879138454977, "grad_norm": 3.1471750736236572, "learning_rate": 4.5866e-06, "loss": 0.0296, "step": 27068 }, { "epoch": 1.4431412272751505, "grad_norm": 0.8887377977371216, "learning_rate": 4.586400000000001e-06, "loss": 0.008, "step": 27069 }, { "epoch": 1.4431945407048035, "grad_norm": 0.5933064222335815, "learning_rate": 4.5862e-06, "loss": 0.0057, "step": 27070 }, { "epoch": 1.4432478541344564, "grad_norm": 1.0778142213821411, "learning_rate": 4.586000000000001e-06, "loss": 0.0074, "step": 27071 }, { "epoch": 1.4433011675641094, "grad_norm": 1.7809680700302124, "learning_rate": 4.5858e-06, "loss": 0.0247, "step": 27072 }, { "epoch": 1.4433544809937624, "grad_norm": 2.091933488845825, "learning_rate": 4.585600000000001e-06, "loss": 0.0249, "step": 27073 }, { "epoch": 1.4434077944234152, "grad_norm": 1.9181872606277466, "learning_rate": 4.5854e-06, "loss": 0.042, "step": 27074 }, { "epoch": 1.4434611078530681, "grad_norm": 0.9152936935424805, "learning_rate": 4.585200000000001e-06, "loss": 0.0164, "step": 27075 }, { "epoch": 1.4435144212827211, "grad_norm": 2.8405017852783203, "learning_rate": 4.585e-06, "loss": 0.0501, "step": 27076 }, { "epoch": 1.443567734712374, "grad_norm": 0.9559242725372314, "learning_rate": 4.5848000000000006e-06, "loss": 0.0084, "step": 27077 }, { "epoch": 1.443621048142027, "grad_norm": 0.3881029784679413, "learning_rate": 4.5846e-06, "loss": 0.0037, "step": 27078 }, { "epoch": 1.4436743615716798, "grad_norm": 0.6707257032394409, "learning_rate": 4.5844000000000005e-06, "loss": 0.0047, "step": 27079 }, { "epoch": 1.4437276750013328, "grad_norm": 0.5423628091812134, "learning_rate": 4.5842e-06, "loss": 0.0048, "step": 27080 }, { "epoch": 1.4437809884309858, "grad_norm": 2.769876480102539, "learning_rate": 4.5840000000000005e-06, "loss": 0.023, "step": 27081 }, { "epoch": 1.4438343018606388, "grad_norm": 1.358574628829956, "learning_rate": 4.5838e-06, "loss": 0.0147, "step": 27082 }, { "epoch": 1.4438876152902917, "grad_norm": 1.1558303833007812, "learning_rate": 4.5836000000000005e-06, "loss": 0.0071, "step": 27083 }, { "epoch": 1.4439409287199445, "grad_norm": 1.9920001029968262, "learning_rate": 4.583400000000001e-06, "loss": 0.019, "step": 27084 }, { "epoch": 1.4439942421495975, "grad_norm": 1.091775894165039, "learning_rate": 4.5832000000000004e-06, "loss": 0.0094, "step": 27085 }, { "epoch": 1.4440475555792505, "grad_norm": 0.6406012773513794, "learning_rate": 4.583e-06, "loss": 0.0044, "step": 27086 }, { "epoch": 1.4441008690089032, "grad_norm": 2.8029725551605225, "learning_rate": 4.5828e-06, "loss": 0.0733, "step": 27087 }, { "epoch": 1.4441541824385562, "grad_norm": 3.3601608276367188, "learning_rate": 4.5826e-06, "loss": 0.0443, "step": 27088 }, { "epoch": 1.4442074958682092, "grad_norm": 1.2725169658660889, "learning_rate": 4.5824e-06, "loss": 0.0188, "step": 27089 }, { "epoch": 1.4442608092978622, "grad_norm": 1.1145946979522705, "learning_rate": 4.582200000000001e-06, "loss": 0.0124, "step": 27090 }, { "epoch": 1.4443141227275151, "grad_norm": 1.1708204746246338, "learning_rate": 4.582e-06, "loss": 0.0172, "step": 27091 }, { "epoch": 1.444367436157168, "grad_norm": 1.7616380453109741, "learning_rate": 4.5818e-06, "loss": 0.0148, "step": 27092 }, { "epoch": 1.4444207495868209, "grad_norm": 3.6256964206695557, "learning_rate": 4.5816e-06, "loss": 0.0383, "step": 27093 }, { "epoch": 1.4444740630164739, "grad_norm": 1.7930165529251099, "learning_rate": 4.5814e-06, "loss": 0.0116, "step": 27094 }, { "epoch": 1.4445273764461268, "grad_norm": 1.2061481475830078, "learning_rate": 4.5812e-06, "loss": 0.0091, "step": 27095 }, { "epoch": 1.4445806898757798, "grad_norm": 1.7606561183929443, "learning_rate": 4.581000000000001e-06, "loss": 0.0139, "step": 27096 }, { "epoch": 1.4446340033054326, "grad_norm": 1.915880560874939, "learning_rate": 4.5808e-06, "loss": 0.0248, "step": 27097 }, { "epoch": 1.4446873167350855, "grad_norm": 0.8967562317848206, "learning_rate": 4.580600000000001e-06, "loss": 0.0187, "step": 27098 }, { "epoch": 1.4447406301647385, "grad_norm": 0.6860774755477905, "learning_rate": 4.5804e-06, "loss": 0.0027, "step": 27099 }, { "epoch": 1.4447939435943915, "grad_norm": 4.542736530303955, "learning_rate": 4.5802e-06, "loss": 0.0145, "step": 27100 }, { "epoch": 1.4448472570240445, "grad_norm": 3.674729347229004, "learning_rate": 4.58e-06, "loss": 0.0283, "step": 27101 }, { "epoch": 1.4449005704536972, "grad_norm": 2.7980496883392334, "learning_rate": 4.579800000000001e-06, "loss": 0.0319, "step": 27102 }, { "epoch": 1.4449538838833502, "grad_norm": 1.8986831903457642, "learning_rate": 4.5796e-06, "loss": 0.0412, "step": 27103 }, { "epoch": 1.4450071973130032, "grad_norm": 1.9984074831008911, "learning_rate": 4.5794000000000006e-06, "loss": 0.0353, "step": 27104 }, { "epoch": 1.445060510742656, "grad_norm": 1.9808330535888672, "learning_rate": 4.5792e-06, "loss": 0.0497, "step": 27105 }, { "epoch": 1.445113824172309, "grad_norm": 1.074181318283081, "learning_rate": 4.579e-06, "loss": 0.0088, "step": 27106 }, { "epoch": 1.445167137601962, "grad_norm": 2.2737998962402344, "learning_rate": 4.5788e-06, "loss": 0.0125, "step": 27107 }, { "epoch": 1.445220451031615, "grad_norm": 2.6732177734375, "learning_rate": 4.5786000000000005e-06, "loss": 0.0256, "step": 27108 }, { "epoch": 1.4452737644612679, "grad_norm": 1.0992590188980103, "learning_rate": 4.578400000000001e-06, "loss": 0.0177, "step": 27109 }, { "epoch": 1.4453270778909206, "grad_norm": 1.7435904741287231, "learning_rate": 4.5782000000000005e-06, "loss": 0.0617, "step": 27110 }, { "epoch": 1.4453803913205736, "grad_norm": 1.0924259424209595, "learning_rate": 4.578e-06, "loss": 0.0089, "step": 27111 }, { "epoch": 1.4454337047502266, "grad_norm": 2.702622652053833, "learning_rate": 4.5778000000000005e-06, "loss": 0.0242, "step": 27112 }, { "epoch": 1.4454870181798796, "grad_norm": 0.8016819953918457, "learning_rate": 4.5776e-06, "loss": 0.0117, "step": 27113 }, { "epoch": 1.4455403316095325, "grad_norm": 1.5047188997268677, "learning_rate": 4.5774e-06, "loss": 0.0126, "step": 27114 }, { "epoch": 1.4455936450391853, "grad_norm": 2.4924747943878174, "learning_rate": 4.577200000000001e-06, "loss": 0.0489, "step": 27115 }, { "epoch": 1.4456469584688383, "grad_norm": 1.9976627826690674, "learning_rate": 4.577e-06, "loss": 0.021, "step": 27116 }, { "epoch": 1.4457002718984913, "grad_norm": 1.0486552715301514, "learning_rate": 4.5768e-06, "loss": 0.0095, "step": 27117 }, { "epoch": 1.4457535853281442, "grad_norm": 3.5459094047546387, "learning_rate": 4.5766e-06, "loss": 0.0457, "step": 27118 }, { "epoch": 1.4458068987577972, "grad_norm": 1.8607462644577026, "learning_rate": 4.5764e-06, "loss": 0.0362, "step": 27119 }, { "epoch": 1.44586021218745, "grad_norm": 2.419438362121582, "learning_rate": 4.5762e-06, "loss": 0.0291, "step": 27120 }, { "epoch": 1.445913525617103, "grad_norm": 1.214385986328125, "learning_rate": 4.576000000000001e-06, "loss": 0.0411, "step": 27121 }, { "epoch": 1.445966839046756, "grad_norm": 2.5804452896118164, "learning_rate": 4.5758e-06, "loss": 0.035, "step": 27122 }, { "epoch": 1.4460201524764087, "grad_norm": 2.0812199115753174, "learning_rate": 4.575600000000001e-06, "loss": 0.0353, "step": 27123 }, { "epoch": 1.4460734659060617, "grad_norm": 1.6871618032455444, "learning_rate": 4.5754e-06, "loss": 0.0261, "step": 27124 }, { "epoch": 1.4461267793357147, "grad_norm": 1.6376782655715942, "learning_rate": 4.5752e-06, "loss": 0.0163, "step": 27125 }, { "epoch": 1.4461800927653676, "grad_norm": 1.1291227340698242, "learning_rate": 4.575e-06, "loss": 0.0159, "step": 27126 }, { "epoch": 1.4462334061950206, "grad_norm": 0.8451598882675171, "learning_rate": 4.574800000000001e-06, "loss": 0.0161, "step": 27127 }, { "epoch": 1.4462867196246734, "grad_norm": 1.5479243993759155, "learning_rate": 4.5746e-06, "loss": 0.0211, "step": 27128 }, { "epoch": 1.4463400330543263, "grad_norm": 1.6102324724197388, "learning_rate": 4.574400000000001e-06, "loss": 0.0139, "step": 27129 }, { "epoch": 1.4463933464839793, "grad_norm": 1.6047567129135132, "learning_rate": 4.5742e-06, "loss": 0.0231, "step": 27130 }, { "epoch": 1.4464466599136323, "grad_norm": 2.564939022064209, "learning_rate": 4.574e-06, "loss": 0.0385, "step": 27131 }, { "epoch": 1.4464999733432853, "grad_norm": 3.733067035675049, "learning_rate": 4.5738e-06, "loss": 0.0524, "step": 27132 }, { "epoch": 1.446553286772938, "grad_norm": 2.0633277893066406, "learning_rate": 4.5736000000000006e-06, "loss": 0.0303, "step": 27133 }, { "epoch": 1.446606600202591, "grad_norm": 1.214268445968628, "learning_rate": 4.573400000000001e-06, "loss": 0.0237, "step": 27134 }, { "epoch": 1.446659913632244, "grad_norm": 2.0213778018951416, "learning_rate": 4.5732000000000005e-06, "loss": 0.0251, "step": 27135 }, { "epoch": 1.4467132270618968, "grad_norm": 1.5446478128433228, "learning_rate": 4.573e-06, "loss": 0.0131, "step": 27136 }, { "epoch": 1.44676654049155, "grad_norm": 1.4332412481307983, "learning_rate": 4.5728000000000005e-06, "loss": 0.0304, "step": 27137 }, { "epoch": 1.4468198539212027, "grad_norm": 1.035895824432373, "learning_rate": 4.5726e-06, "loss": 0.0212, "step": 27138 }, { "epoch": 1.4468731673508557, "grad_norm": 1.4473246335983276, "learning_rate": 4.5724000000000005e-06, "loss": 0.0173, "step": 27139 }, { "epoch": 1.4469264807805087, "grad_norm": 1.4717628955841064, "learning_rate": 4.572200000000001e-06, "loss": 0.0295, "step": 27140 }, { "epoch": 1.4469797942101614, "grad_norm": 2.3100528717041016, "learning_rate": 4.5720000000000004e-06, "loss": 0.0176, "step": 27141 }, { "epoch": 1.4470331076398144, "grad_norm": 2.1177661418914795, "learning_rate": 4.5718e-06, "loss": 0.0328, "step": 27142 }, { "epoch": 1.4470864210694674, "grad_norm": 1.3331751823425293, "learning_rate": 4.5716e-06, "loss": 0.0225, "step": 27143 }, { "epoch": 1.4471397344991204, "grad_norm": 0.9035755395889282, "learning_rate": 4.5714e-06, "loss": 0.0123, "step": 27144 }, { "epoch": 1.4471930479287733, "grad_norm": 1.3375051021575928, "learning_rate": 4.5712e-06, "loss": 0.0142, "step": 27145 }, { "epoch": 1.447246361358426, "grad_norm": 4.247287273406982, "learning_rate": 4.571000000000001e-06, "loss": 0.0332, "step": 27146 }, { "epoch": 1.447299674788079, "grad_norm": 1.9153683185577393, "learning_rate": 4.5708e-06, "loss": 0.0452, "step": 27147 }, { "epoch": 1.447352988217732, "grad_norm": 1.4028730392456055, "learning_rate": 4.570600000000001e-06, "loss": 0.025, "step": 27148 }, { "epoch": 1.447406301647385, "grad_norm": 1.5920722484588623, "learning_rate": 4.5704e-06, "loss": 0.0166, "step": 27149 }, { "epoch": 1.447459615077038, "grad_norm": 0.7993588447570801, "learning_rate": 4.5702e-06, "loss": 0.0215, "step": 27150 }, { "epoch": 1.4475129285066908, "grad_norm": 1.036134123802185, "learning_rate": 4.57e-06, "loss": 0.0272, "step": 27151 }, { "epoch": 1.4475662419363438, "grad_norm": 1.6597346067428589, "learning_rate": 4.569800000000001e-06, "loss": 0.0225, "step": 27152 }, { "epoch": 1.4476195553659967, "grad_norm": 2.755894422531128, "learning_rate": 4.5696e-06, "loss": 0.0325, "step": 27153 }, { "epoch": 1.4476728687956495, "grad_norm": 2.765549421310425, "learning_rate": 4.569400000000001e-06, "loss": 0.0131, "step": 27154 }, { "epoch": 1.4477261822253025, "grad_norm": 1.0037285089492798, "learning_rate": 4.5692e-06, "loss": 0.0101, "step": 27155 }, { "epoch": 1.4477794956549555, "grad_norm": 1.6682548522949219, "learning_rate": 4.569e-06, "loss": 0.0128, "step": 27156 }, { "epoch": 1.4478328090846084, "grad_norm": 2.8916878700256348, "learning_rate": 4.5688e-06, "loss": 0.0533, "step": 27157 }, { "epoch": 1.4478861225142614, "grad_norm": 1.6098194122314453, "learning_rate": 4.568600000000001e-06, "loss": 0.0213, "step": 27158 }, { "epoch": 1.4479394359439142, "grad_norm": 1.8385423421859741, "learning_rate": 4.568400000000001e-06, "loss": 0.0191, "step": 27159 }, { "epoch": 1.4479927493735671, "grad_norm": 2.2460687160491943, "learning_rate": 4.568200000000001e-06, "loss": 0.0364, "step": 27160 }, { "epoch": 1.4480460628032201, "grad_norm": 0.78389573097229, "learning_rate": 4.568e-06, "loss": 0.0063, "step": 27161 }, { "epoch": 1.448099376232873, "grad_norm": 0.956202507019043, "learning_rate": 4.5678000000000005e-06, "loss": 0.0083, "step": 27162 }, { "epoch": 1.448152689662526, "grad_norm": 1.7798762321472168, "learning_rate": 4.5676e-06, "loss": 0.0339, "step": 27163 }, { "epoch": 1.4482060030921788, "grad_norm": 1.1381422281265259, "learning_rate": 4.5674000000000005e-06, "loss": 0.0144, "step": 27164 }, { "epoch": 1.4482593165218318, "grad_norm": 4.347867965698242, "learning_rate": 4.567200000000001e-06, "loss": 0.0323, "step": 27165 }, { "epoch": 1.4483126299514848, "grad_norm": 4.177096366882324, "learning_rate": 4.5670000000000005e-06, "loss": 0.0383, "step": 27166 }, { "epoch": 1.4483659433811378, "grad_norm": 1.984925389289856, "learning_rate": 4.5668e-06, "loss": 0.0288, "step": 27167 }, { "epoch": 1.4484192568107908, "grad_norm": 1.1454651355743408, "learning_rate": 4.5666000000000005e-06, "loss": 0.0247, "step": 27168 }, { "epoch": 1.4484725702404435, "grad_norm": 1.3134090900421143, "learning_rate": 4.5664e-06, "loss": 0.0149, "step": 27169 }, { "epoch": 1.4485258836700965, "grad_norm": 1.210557222366333, "learning_rate": 4.5662000000000004e-06, "loss": 0.0099, "step": 27170 }, { "epoch": 1.4485791970997495, "grad_norm": 3.5916666984558105, "learning_rate": 4.566000000000001e-06, "loss": 0.0139, "step": 27171 }, { "epoch": 1.4486325105294022, "grad_norm": 1.670263409614563, "learning_rate": 4.5658e-06, "loss": 0.0151, "step": 27172 }, { "epoch": 1.4486858239590552, "grad_norm": 2.7858078479766846, "learning_rate": 4.565600000000001e-06, "loss": 0.0187, "step": 27173 }, { "epoch": 1.4487391373887082, "grad_norm": 1.6646453142166138, "learning_rate": 4.5654e-06, "loss": 0.0219, "step": 27174 }, { "epoch": 1.4487924508183612, "grad_norm": 2.0365052223205566, "learning_rate": 4.5652e-06, "loss": 0.02, "step": 27175 }, { "epoch": 1.4488457642480141, "grad_norm": 2.2712130546569824, "learning_rate": 4.565e-06, "loss": 0.0259, "step": 27176 }, { "epoch": 1.448899077677667, "grad_norm": 2.6541481018066406, "learning_rate": 4.564800000000001e-06, "loss": 0.0255, "step": 27177 }, { "epoch": 1.4489523911073199, "grad_norm": 2.355189323425293, "learning_rate": 4.5646e-06, "loss": 0.0139, "step": 27178 }, { "epoch": 1.4490057045369729, "grad_norm": 3.5907790660858154, "learning_rate": 4.564400000000001e-06, "loss": 0.0436, "step": 27179 }, { "epoch": 1.4490590179666258, "grad_norm": 1.4952884912490845, "learning_rate": 4.5642e-06, "loss": 0.0171, "step": 27180 }, { "epoch": 1.4491123313962788, "grad_norm": 2.5877723693847656, "learning_rate": 4.564e-06, "loss": 0.055, "step": 27181 }, { "epoch": 1.4491656448259316, "grad_norm": 0.6075007319450378, "learning_rate": 4.5638e-06, "loss": 0.0067, "step": 27182 }, { "epoch": 1.4492189582555846, "grad_norm": 1.7033168077468872, "learning_rate": 4.563600000000001e-06, "loss": 0.0277, "step": 27183 }, { "epoch": 1.4492722716852375, "grad_norm": 0.8587889075279236, "learning_rate": 4.5634e-06, "loss": 0.0069, "step": 27184 }, { "epoch": 1.4493255851148905, "grad_norm": 1.008893609046936, "learning_rate": 4.563200000000001e-06, "loss": 0.0071, "step": 27185 }, { "epoch": 1.4493788985445435, "grad_norm": 2.4087576866149902, "learning_rate": 4.563e-06, "loss": 0.0335, "step": 27186 }, { "epoch": 1.4494322119741962, "grad_norm": 2.126457452774048, "learning_rate": 4.562800000000001e-06, "loss": 0.0327, "step": 27187 }, { "epoch": 1.4494855254038492, "grad_norm": 3.0276622772216797, "learning_rate": 4.5626e-06, "loss": 0.027, "step": 27188 }, { "epoch": 1.4495388388335022, "grad_norm": 2.731945276260376, "learning_rate": 4.5624000000000006e-06, "loss": 0.0191, "step": 27189 }, { "epoch": 1.449592152263155, "grad_norm": 1.167489767074585, "learning_rate": 4.5622e-06, "loss": 0.0078, "step": 27190 }, { "epoch": 1.449645465692808, "grad_norm": 0.4851764738559723, "learning_rate": 4.5620000000000005e-06, "loss": 0.0037, "step": 27191 }, { "epoch": 1.449698779122461, "grad_norm": 1.8410451412200928, "learning_rate": 4.5618e-06, "loss": 0.0327, "step": 27192 }, { "epoch": 1.449752092552114, "grad_norm": 1.035624623298645, "learning_rate": 4.5616000000000005e-06, "loss": 0.0287, "step": 27193 }, { "epoch": 1.4498054059817669, "grad_norm": 1.4348030090332031, "learning_rate": 4.5614e-06, "loss": 0.0263, "step": 27194 }, { "epoch": 1.4498587194114196, "grad_norm": 3.0439770221710205, "learning_rate": 4.5612000000000005e-06, "loss": 0.0336, "step": 27195 }, { "epoch": 1.4499120328410726, "grad_norm": 0.2941574156284332, "learning_rate": 4.561e-06, "loss": 0.0031, "step": 27196 }, { "epoch": 1.4499653462707256, "grad_norm": 1.0228358507156372, "learning_rate": 4.5608000000000004e-06, "loss": 0.0094, "step": 27197 }, { "epoch": 1.4500186597003786, "grad_norm": 1.604475498199463, "learning_rate": 4.560600000000001e-06, "loss": 0.0255, "step": 27198 }, { "epoch": 1.4500719731300316, "grad_norm": 0.9196762442588806, "learning_rate": 4.5604e-06, "loss": 0.0085, "step": 27199 }, { "epoch": 1.4501252865596843, "grad_norm": 1.7042005062103271, "learning_rate": 4.5602e-06, "loss": 0.0196, "step": 27200 }, { "epoch": 1.4501785999893373, "grad_norm": 3.5672659873962402, "learning_rate": 4.56e-06, "loss": 0.0426, "step": 27201 }, { "epoch": 1.4502319134189903, "grad_norm": 1.5494083166122437, "learning_rate": 4.5598e-06, "loss": 0.0057, "step": 27202 }, { "epoch": 1.4502852268486432, "grad_norm": 1.2064526081085205, "learning_rate": 4.5596e-06, "loss": 0.0174, "step": 27203 }, { "epoch": 1.4503385402782962, "grad_norm": 1.696992039680481, "learning_rate": 4.559400000000001e-06, "loss": 0.0217, "step": 27204 }, { "epoch": 1.450391853707949, "grad_norm": 1.397190809249878, "learning_rate": 4.5592e-06, "loss": 0.0322, "step": 27205 }, { "epoch": 1.450445167137602, "grad_norm": 2.237607002258301, "learning_rate": 4.559000000000001e-06, "loss": 0.0382, "step": 27206 }, { "epoch": 1.450498480567255, "grad_norm": 1.391629695892334, "learning_rate": 4.5588e-06, "loss": 0.0212, "step": 27207 }, { "epoch": 1.4505517939969077, "grad_norm": 1.4467484951019287, "learning_rate": 4.5586e-06, "loss": 0.0081, "step": 27208 }, { "epoch": 1.4506051074265607, "grad_norm": 0.5905008316040039, "learning_rate": 4.5584e-06, "loss": 0.0057, "step": 27209 }, { "epoch": 1.4506584208562137, "grad_norm": 2.7079246044158936, "learning_rate": 4.558200000000001e-06, "loss": 0.0496, "step": 27210 }, { "epoch": 1.4507117342858666, "grad_norm": 1.3555245399475098, "learning_rate": 4.558e-06, "loss": 0.0132, "step": 27211 }, { "epoch": 1.4507650477155196, "grad_norm": 5.9975080490112305, "learning_rate": 4.557800000000001e-06, "loss": 0.0445, "step": 27212 }, { "epoch": 1.4508183611451724, "grad_norm": 2.0859885215759277, "learning_rate": 4.5576e-06, "loss": 0.0546, "step": 27213 }, { "epoch": 1.4508716745748254, "grad_norm": 2.951359510421753, "learning_rate": 4.5574e-06, "loss": 0.0135, "step": 27214 }, { "epoch": 1.4509249880044783, "grad_norm": 2.9726603031158447, "learning_rate": 4.5572e-06, "loss": 0.0463, "step": 27215 }, { "epoch": 1.4509783014341313, "grad_norm": 2.342418909072876, "learning_rate": 4.557000000000001e-06, "loss": 0.0204, "step": 27216 }, { "epoch": 1.4510316148637843, "grad_norm": 2.330395221710205, "learning_rate": 4.5568e-06, "loss": 0.0504, "step": 27217 }, { "epoch": 1.451084928293437, "grad_norm": 0.9415706396102905, "learning_rate": 4.5566000000000006e-06, "loss": 0.0087, "step": 27218 }, { "epoch": 1.45113824172309, "grad_norm": 0.5311141610145569, "learning_rate": 4.5564e-06, "loss": 0.0052, "step": 27219 }, { "epoch": 1.451191555152743, "grad_norm": 2.2855916023254395, "learning_rate": 4.5562000000000005e-06, "loss": 0.0259, "step": 27220 }, { "epoch": 1.451244868582396, "grad_norm": 1.4740018844604492, "learning_rate": 4.556e-06, "loss": 0.016, "step": 27221 }, { "epoch": 1.451298182012049, "grad_norm": 1.200260043144226, "learning_rate": 4.5558000000000005e-06, "loss": 0.01, "step": 27222 }, { "epoch": 1.4513514954417017, "grad_norm": 0.26782116293907166, "learning_rate": 4.555600000000001e-06, "loss": 0.0033, "step": 27223 }, { "epoch": 1.4514048088713547, "grad_norm": 2.326669692993164, "learning_rate": 4.5554000000000005e-06, "loss": 0.0286, "step": 27224 }, { "epoch": 1.4514581223010077, "grad_norm": 1.567578673362732, "learning_rate": 4.5552e-06, "loss": 0.0287, "step": 27225 }, { "epoch": 1.4515114357306604, "grad_norm": 1.6840522289276123, "learning_rate": 4.5550000000000004e-06, "loss": 0.0149, "step": 27226 }, { "epoch": 1.4515647491603134, "grad_norm": 1.1794705390930176, "learning_rate": 4.5548e-06, "loss": 0.021, "step": 27227 }, { "epoch": 1.4516180625899664, "grad_norm": 1.4275946617126465, "learning_rate": 4.5546e-06, "loss": 0.014, "step": 27228 }, { "epoch": 1.4516713760196194, "grad_norm": 1.5858250856399536, "learning_rate": 4.554400000000001e-06, "loss": 0.0182, "step": 27229 }, { "epoch": 1.4517246894492724, "grad_norm": 2.596815347671509, "learning_rate": 4.5542e-06, "loss": 0.0365, "step": 27230 }, { "epoch": 1.4517780028789251, "grad_norm": 1.4183052778244019, "learning_rate": 4.554000000000001e-06, "loss": 0.015, "step": 27231 }, { "epoch": 1.451831316308578, "grad_norm": 1.5957845449447632, "learning_rate": 4.5538e-06, "loss": 0.0108, "step": 27232 }, { "epoch": 1.451884629738231, "grad_norm": 2.349992275238037, "learning_rate": 4.5536e-06, "loss": 0.0234, "step": 27233 }, { "epoch": 1.451937943167884, "grad_norm": 2.097412586212158, "learning_rate": 4.5534e-06, "loss": 0.0303, "step": 27234 }, { "epoch": 1.451991256597537, "grad_norm": 0.8149724006652832, "learning_rate": 4.553200000000001e-06, "loss": 0.0076, "step": 27235 }, { "epoch": 1.4520445700271898, "grad_norm": 1.1350994110107422, "learning_rate": 4.553e-06, "loss": 0.0074, "step": 27236 }, { "epoch": 1.4520978834568428, "grad_norm": 1.8166109323501587, "learning_rate": 4.552800000000001e-06, "loss": 0.0329, "step": 27237 }, { "epoch": 1.4521511968864957, "grad_norm": 1.3271201848983765, "learning_rate": 4.5526e-06, "loss": 0.0108, "step": 27238 }, { "epoch": 1.4522045103161485, "grad_norm": 1.53790283203125, "learning_rate": 4.5524e-06, "loss": 0.0133, "step": 27239 }, { "epoch": 1.4522578237458017, "grad_norm": 2.786834478378296, "learning_rate": 4.5522e-06, "loss": 0.034, "step": 27240 }, { "epoch": 1.4523111371754545, "grad_norm": 2.172046422958374, "learning_rate": 4.552000000000001e-06, "loss": 0.0328, "step": 27241 }, { "epoch": 1.4523644506051074, "grad_norm": 2.089470148086548, "learning_rate": 4.5518e-06, "loss": 0.0445, "step": 27242 }, { "epoch": 1.4524177640347604, "grad_norm": 1.4030494689941406, "learning_rate": 4.551600000000001e-06, "loss": 0.027, "step": 27243 }, { "epoch": 1.4524710774644132, "grad_norm": 0.8762133717536926, "learning_rate": 4.5514e-06, "loss": 0.0161, "step": 27244 }, { "epoch": 1.4525243908940662, "grad_norm": 1.3998348712921143, "learning_rate": 4.5512000000000006e-06, "loss": 0.0285, "step": 27245 }, { "epoch": 1.4525777043237191, "grad_norm": 2.2083308696746826, "learning_rate": 4.551e-06, "loss": 0.0281, "step": 27246 }, { "epoch": 1.4526310177533721, "grad_norm": 1.1517448425292969, "learning_rate": 4.5508000000000005e-06, "loss": 0.0105, "step": 27247 }, { "epoch": 1.452684331183025, "grad_norm": 0.8453173637390137, "learning_rate": 4.550600000000001e-06, "loss": 0.0142, "step": 27248 }, { "epoch": 1.4527376446126778, "grad_norm": 2.7462682723999023, "learning_rate": 4.5504000000000005e-06, "loss": 0.0343, "step": 27249 }, { "epoch": 1.4527909580423308, "grad_norm": 1.3119806051254272, "learning_rate": 4.5502e-06, "loss": 0.0243, "step": 27250 }, { "epoch": 1.4528442714719838, "grad_norm": 1.5864675045013428, "learning_rate": 4.5500000000000005e-06, "loss": 0.0104, "step": 27251 }, { "epoch": 1.4528975849016368, "grad_norm": 1.3879419565200806, "learning_rate": 4.5498e-06, "loss": 0.0256, "step": 27252 }, { "epoch": 1.4529508983312898, "grad_norm": 1.7866284847259521, "learning_rate": 4.5496000000000005e-06, "loss": 0.0209, "step": 27253 }, { "epoch": 1.4530042117609425, "grad_norm": 1.9945809841156006, "learning_rate": 4.549400000000001e-06, "loss": 0.0273, "step": 27254 }, { "epoch": 1.4530575251905955, "grad_norm": 1.8778512477874756, "learning_rate": 4.5492e-06, "loss": 0.0273, "step": 27255 }, { "epoch": 1.4531108386202485, "grad_norm": 1.9246656894683838, "learning_rate": 4.549000000000001e-06, "loss": 0.0108, "step": 27256 }, { "epoch": 1.4531641520499012, "grad_norm": 6.04340934753418, "learning_rate": 4.5488e-06, "loss": 0.0387, "step": 27257 }, { "epoch": 1.4532174654795544, "grad_norm": 0.9819827079772949, "learning_rate": 4.5486e-06, "loss": 0.0117, "step": 27258 }, { "epoch": 1.4532707789092072, "grad_norm": 2.0895824432373047, "learning_rate": 4.5484e-06, "loss": 0.0189, "step": 27259 }, { "epoch": 1.4533240923388602, "grad_norm": 3.1820127964019775, "learning_rate": 4.548200000000001e-06, "loss": 0.0138, "step": 27260 }, { "epoch": 1.4533774057685132, "grad_norm": 7.447568893432617, "learning_rate": 4.548e-06, "loss": 0.0252, "step": 27261 }, { "epoch": 1.453430719198166, "grad_norm": 2.2067275047302246, "learning_rate": 4.547800000000001e-06, "loss": 0.0347, "step": 27262 }, { "epoch": 1.453484032627819, "grad_norm": 0.7365875840187073, "learning_rate": 4.5476e-06, "loss": 0.0058, "step": 27263 }, { "epoch": 1.4535373460574719, "grad_norm": 2.4948770999908447, "learning_rate": 4.5474e-06, "loss": 0.0462, "step": 27264 }, { "epoch": 1.4535906594871248, "grad_norm": 3.731544256210327, "learning_rate": 4.5472e-06, "loss": 0.0648, "step": 27265 }, { "epoch": 1.4536439729167778, "grad_norm": 0.984058678150177, "learning_rate": 4.547000000000001e-06, "loss": 0.0256, "step": 27266 }, { "epoch": 1.4536972863464306, "grad_norm": 1.9827319383621216, "learning_rate": 4.5468e-06, "loss": 0.0239, "step": 27267 }, { "epoch": 1.4537505997760836, "grad_norm": 1.0641955137252808, "learning_rate": 4.546600000000001e-06, "loss": 0.0283, "step": 27268 }, { "epoch": 1.4538039132057365, "grad_norm": 2.5816493034362793, "learning_rate": 4.5464e-06, "loss": 0.0296, "step": 27269 }, { "epoch": 1.4538572266353895, "grad_norm": 2.299351215362549, "learning_rate": 4.546200000000001e-06, "loss": 0.025, "step": 27270 }, { "epoch": 1.4539105400650425, "grad_norm": 6.934542179107666, "learning_rate": 4.546e-06, "loss": 0.0611, "step": 27271 }, { "epoch": 1.4539638534946953, "grad_norm": 1.9817322492599487, "learning_rate": 4.545800000000001e-06, "loss": 0.0352, "step": 27272 }, { "epoch": 1.4540171669243482, "grad_norm": 3.211315870285034, "learning_rate": 4.5456e-06, "loss": 0.0343, "step": 27273 }, { "epoch": 1.4540704803540012, "grad_norm": 3.0902411937713623, "learning_rate": 4.5454000000000006e-06, "loss": 0.0295, "step": 27274 }, { "epoch": 1.454123793783654, "grad_norm": 2.9225716590881348, "learning_rate": 4.5452e-06, "loss": 0.0202, "step": 27275 }, { "epoch": 1.454177107213307, "grad_norm": 1.295351266860962, "learning_rate": 4.5450000000000005e-06, "loss": 0.0289, "step": 27276 }, { "epoch": 1.45423042064296, "grad_norm": 1.843961238861084, "learning_rate": 4.5448e-06, "loss": 0.0137, "step": 27277 }, { "epoch": 1.454283734072613, "grad_norm": 1.376572608947754, "learning_rate": 4.5446000000000005e-06, "loss": 0.0126, "step": 27278 }, { "epoch": 1.454337047502266, "grad_norm": 2.735931634902954, "learning_rate": 4.5444e-06, "loss": 0.0296, "step": 27279 }, { "epoch": 1.4543903609319186, "grad_norm": 2.1700031757354736, "learning_rate": 4.5442000000000005e-06, "loss": 0.0204, "step": 27280 }, { "epoch": 1.4544436743615716, "grad_norm": 2.6962730884552, "learning_rate": 4.544000000000001e-06, "loss": 0.0448, "step": 27281 }, { "epoch": 1.4544969877912246, "grad_norm": 3.63814115524292, "learning_rate": 4.5438000000000004e-06, "loss": 0.0347, "step": 27282 }, { "epoch": 1.4545503012208776, "grad_norm": 6.795348167419434, "learning_rate": 4.5436e-06, "loss": 0.0325, "step": 27283 }, { "epoch": 1.4546036146505306, "grad_norm": 2.0460493564605713, "learning_rate": 4.5434e-06, "loss": 0.0254, "step": 27284 }, { "epoch": 1.4546569280801833, "grad_norm": 0.7149524092674255, "learning_rate": 4.5432e-06, "loss": 0.0117, "step": 27285 }, { "epoch": 1.4547102415098363, "grad_norm": 1.0808684825897217, "learning_rate": 4.543e-06, "loss": 0.0118, "step": 27286 }, { "epoch": 1.4547635549394893, "grad_norm": 0.9233586192131042, "learning_rate": 4.542800000000001e-06, "loss": 0.0252, "step": 27287 }, { "epoch": 1.4548168683691423, "grad_norm": 1.0026272535324097, "learning_rate": 4.5426e-06, "loss": 0.0114, "step": 27288 }, { "epoch": 1.4548701817987952, "grad_norm": 1.9357974529266357, "learning_rate": 4.5424e-06, "loss": 0.0171, "step": 27289 }, { "epoch": 1.454923495228448, "grad_norm": 2.7766356468200684, "learning_rate": 4.5422e-06, "loss": 0.0239, "step": 27290 }, { "epoch": 1.454976808658101, "grad_norm": 0.763095498085022, "learning_rate": 4.542e-06, "loss": 0.0048, "step": 27291 }, { "epoch": 1.455030122087754, "grad_norm": 1.723605990409851, "learning_rate": 4.5418e-06, "loss": 0.0316, "step": 27292 }, { "epoch": 1.4550834355174067, "grad_norm": 3.259140968322754, "learning_rate": 4.541600000000001e-06, "loss": 0.0335, "step": 27293 }, { "epoch": 1.4551367489470597, "grad_norm": 1.4780939817428589, "learning_rate": 4.5414e-06, "loss": 0.0097, "step": 27294 }, { "epoch": 1.4551900623767127, "grad_norm": 2.2580723762512207, "learning_rate": 4.541200000000001e-06, "loss": 0.0255, "step": 27295 }, { "epoch": 1.4552433758063656, "grad_norm": 2.677668571472168, "learning_rate": 4.541e-06, "loss": 0.046, "step": 27296 }, { "epoch": 1.4552966892360186, "grad_norm": 0.770390510559082, "learning_rate": 4.5408e-06, "loss": 0.0055, "step": 27297 }, { "epoch": 1.4553500026656714, "grad_norm": 0.7089943289756775, "learning_rate": 4.5406e-06, "loss": 0.0046, "step": 27298 }, { "epoch": 1.4554033160953244, "grad_norm": 0.4540695250034332, "learning_rate": 4.540400000000001e-06, "loss": 0.0049, "step": 27299 }, { "epoch": 1.4554566295249773, "grad_norm": 4.586103916168213, "learning_rate": 4.5402e-06, "loss": 0.0394, "step": 27300 }, { "epoch": 1.4555099429546303, "grad_norm": 1.5668219327926636, "learning_rate": 4.540000000000001e-06, "loss": 0.0148, "step": 27301 }, { "epoch": 1.4555632563842833, "grad_norm": 2.4298017024993896, "learning_rate": 4.5398e-06, "loss": 0.0271, "step": 27302 }, { "epoch": 1.455616569813936, "grad_norm": 0.8389967083930969, "learning_rate": 4.5396e-06, "loss": 0.019, "step": 27303 }, { "epoch": 1.455669883243589, "grad_norm": 2.4693920612335205, "learning_rate": 4.5394e-06, "loss": 0.0405, "step": 27304 }, { "epoch": 1.455723196673242, "grad_norm": 1.5196757316589355, "learning_rate": 4.5392000000000005e-06, "loss": 0.0145, "step": 27305 }, { "epoch": 1.455776510102895, "grad_norm": 1.6417322158813477, "learning_rate": 4.539000000000001e-06, "loss": 0.012, "step": 27306 }, { "epoch": 1.455829823532548, "grad_norm": 1.710607647895813, "learning_rate": 4.5388000000000005e-06, "loss": 0.0177, "step": 27307 }, { "epoch": 1.4558831369622007, "grad_norm": 3.18241548538208, "learning_rate": 4.5386e-06, "loss": 0.0272, "step": 27308 }, { "epoch": 1.4559364503918537, "grad_norm": 2.2499301433563232, "learning_rate": 4.5384000000000005e-06, "loss": 0.0233, "step": 27309 }, { "epoch": 1.4559897638215067, "grad_norm": 2.8930704593658447, "learning_rate": 4.5382e-06, "loss": 0.0193, "step": 27310 }, { "epoch": 1.4560430772511594, "grad_norm": 3.4547994136810303, "learning_rate": 4.5380000000000004e-06, "loss": 0.033, "step": 27311 }, { "epoch": 1.4560963906808124, "grad_norm": 2.5577685832977295, "learning_rate": 4.537800000000001e-06, "loss": 0.0403, "step": 27312 }, { "epoch": 1.4561497041104654, "grad_norm": 2.289222478866577, "learning_rate": 4.5376e-06, "loss": 0.0137, "step": 27313 }, { "epoch": 1.4562030175401184, "grad_norm": 1.2184604406356812, "learning_rate": 4.5374e-06, "loss": 0.0205, "step": 27314 }, { "epoch": 1.4562563309697714, "grad_norm": 0.7639411687850952, "learning_rate": 4.5372e-06, "loss": 0.0194, "step": 27315 }, { "epoch": 1.4563096443994241, "grad_norm": 1.0420472621917725, "learning_rate": 4.537e-06, "loss": 0.0109, "step": 27316 }, { "epoch": 1.456362957829077, "grad_norm": 1.4824107885360718, "learning_rate": 4.5368e-06, "loss": 0.0233, "step": 27317 }, { "epoch": 1.45641627125873, "grad_norm": 1.691324234008789, "learning_rate": 4.536600000000001e-06, "loss": 0.0156, "step": 27318 }, { "epoch": 1.456469584688383, "grad_norm": 3.061112880706787, "learning_rate": 4.5364e-06, "loss": 0.0584, "step": 27319 }, { "epoch": 1.456522898118036, "grad_norm": 1.1514692306518555, "learning_rate": 4.536200000000001e-06, "loss": 0.012, "step": 27320 }, { "epoch": 1.4565762115476888, "grad_norm": 2.055124044418335, "learning_rate": 4.536e-06, "loss": 0.0511, "step": 27321 }, { "epoch": 1.4566295249773418, "grad_norm": 1.5852782726287842, "learning_rate": 4.5358e-06, "loss": 0.0177, "step": 27322 }, { "epoch": 1.4566828384069948, "grad_norm": 2.399550676345825, "learning_rate": 4.5356e-06, "loss": 0.0477, "step": 27323 }, { "epoch": 1.4567361518366477, "grad_norm": 2.746147871017456, "learning_rate": 4.535400000000001e-06, "loss": 0.0366, "step": 27324 }, { "epoch": 1.4567894652663007, "grad_norm": 1.1034842729568481, "learning_rate": 4.5352e-06, "loss": 0.0164, "step": 27325 }, { "epoch": 1.4568427786959535, "grad_norm": 4.584836006164551, "learning_rate": 4.535000000000001e-06, "loss": 0.0482, "step": 27326 }, { "epoch": 1.4568960921256064, "grad_norm": 1.3008781671524048, "learning_rate": 4.5348e-06, "loss": 0.0247, "step": 27327 }, { "epoch": 1.4569494055552594, "grad_norm": 1.8025448322296143, "learning_rate": 4.5346e-06, "loss": 0.0161, "step": 27328 }, { "epoch": 1.4570027189849122, "grad_norm": 2.183173179626465, "learning_rate": 4.5344e-06, "loss": 0.0192, "step": 27329 }, { "epoch": 1.4570560324145652, "grad_norm": 2.677889585494995, "learning_rate": 4.5342000000000006e-06, "loss": 0.0296, "step": 27330 }, { "epoch": 1.4571093458442181, "grad_norm": 1.8163182735443115, "learning_rate": 4.534000000000001e-06, "loss": 0.0141, "step": 27331 }, { "epoch": 1.4571626592738711, "grad_norm": 2.5814435482025146, "learning_rate": 4.5338000000000005e-06, "loss": 0.0754, "step": 27332 }, { "epoch": 1.457215972703524, "grad_norm": 2.6801905632019043, "learning_rate": 4.5336e-06, "loss": 0.0191, "step": 27333 }, { "epoch": 1.4572692861331769, "grad_norm": 1.272806167602539, "learning_rate": 4.5334000000000005e-06, "loss": 0.0119, "step": 27334 }, { "epoch": 1.4573225995628298, "grad_norm": 2.585719347000122, "learning_rate": 4.5332e-06, "loss": 0.0158, "step": 27335 }, { "epoch": 1.4573759129924828, "grad_norm": 1.3963253498077393, "learning_rate": 4.5330000000000005e-06, "loss": 0.0177, "step": 27336 }, { "epoch": 1.4574292264221358, "grad_norm": 1.040190577507019, "learning_rate": 4.532800000000001e-06, "loss": 0.016, "step": 27337 }, { "epoch": 1.4574825398517888, "grad_norm": 1.6288844347000122, "learning_rate": 4.5326000000000004e-06, "loss": 0.0322, "step": 27338 }, { "epoch": 1.4575358532814415, "grad_norm": 1.0440524816513062, "learning_rate": 4.5324e-06, "loss": 0.0064, "step": 27339 }, { "epoch": 1.4575891667110945, "grad_norm": 1.882453203201294, "learning_rate": 4.5322e-06, "loss": 0.0321, "step": 27340 }, { "epoch": 1.4576424801407475, "grad_norm": 1.9714903831481934, "learning_rate": 4.532e-06, "loss": 0.0236, "step": 27341 }, { "epoch": 1.4576957935704005, "grad_norm": 2.1022958755493164, "learning_rate": 4.5318e-06, "loss": 0.0324, "step": 27342 }, { "epoch": 1.4577491070000534, "grad_norm": 1.860393762588501, "learning_rate": 4.531600000000001e-06, "loss": 0.0325, "step": 27343 }, { "epoch": 1.4578024204297062, "grad_norm": 1.0101308822631836, "learning_rate": 4.5314e-06, "loss": 0.0177, "step": 27344 }, { "epoch": 1.4578557338593592, "grad_norm": 1.7398045063018799, "learning_rate": 4.531200000000001e-06, "loss": 0.0183, "step": 27345 }, { "epoch": 1.4579090472890122, "grad_norm": 0.9678041338920593, "learning_rate": 4.531e-06, "loss": 0.0084, "step": 27346 }, { "epoch": 1.457962360718665, "grad_norm": 0.5442507863044739, "learning_rate": 4.5308e-06, "loss": 0.0053, "step": 27347 }, { "epoch": 1.458015674148318, "grad_norm": 1.372922658920288, "learning_rate": 4.5306e-06, "loss": 0.0087, "step": 27348 }, { "epoch": 1.4580689875779709, "grad_norm": 1.7407422065734863, "learning_rate": 4.530400000000001e-06, "loss": 0.0295, "step": 27349 }, { "epoch": 1.4581223010076239, "grad_norm": 0.5662082433700562, "learning_rate": 4.5302e-06, "loss": 0.0052, "step": 27350 }, { "epoch": 1.4581756144372768, "grad_norm": 0.8195590376853943, "learning_rate": 4.530000000000001e-06, "loss": 0.0146, "step": 27351 }, { "epoch": 1.4582289278669296, "grad_norm": 2.263885259628296, "learning_rate": 4.5298e-06, "loss": 0.0299, "step": 27352 }, { "epoch": 1.4582822412965826, "grad_norm": 1.6972084045410156, "learning_rate": 4.5296e-06, "loss": 0.0316, "step": 27353 }, { "epoch": 1.4583355547262356, "grad_norm": 1.5855097770690918, "learning_rate": 4.5294e-06, "loss": 0.0163, "step": 27354 }, { "epoch": 1.4583888681558885, "grad_norm": 1.1777628660202026, "learning_rate": 4.529200000000001e-06, "loss": 0.0238, "step": 27355 }, { "epoch": 1.4584421815855415, "grad_norm": 1.161980390548706, "learning_rate": 4.529000000000001e-06, "loss": 0.0148, "step": 27356 }, { "epoch": 1.4584954950151943, "grad_norm": 1.5709978342056274, "learning_rate": 4.528800000000001e-06, "loss": 0.0106, "step": 27357 }, { "epoch": 1.4585488084448472, "grad_norm": 1.6881282329559326, "learning_rate": 4.5286e-06, "loss": 0.0161, "step": 27358 }, { "epoch": 1.4586021218745002, "grad_norm": 1.3426358699798584, "learning_rate": 4.5284000000000006e-06, "loss": 0.0085, "step": 27359 }, { "epoch": 1.458655435304153, "grad_norm": 1.5847758054733276, "learning_rate": 4.5282e-06, "loss": 0.0385, "step": 27360 }, { "epoch": 1.4587087487338062, "grad_norm": 2.672917366027832, "learning_rate": 4.5280000000000005e-06, "loss": 0.0145, "step": 27361 }, { "epoch": 1.458762062163459, "grad_norm": 1.6905865669250488, "learning_rate": 4.527800000000001e-06, "loss": 0.0114, "step": 27362 }, { "epoch": 1.458815375593112, "grad_norm": 1.2930208444595337, "learning_rate": 4.5276000000000005e-06, "loss": 0.0106, "step": 27363 }, { "epoch": 1.458868689022765, "grad_norm": 2.6446030139923096, "learning_rate": 4.5274e-06, "loss": 0.0273, "step": 27364 }, { "epoch": 1.4589220024524177, "grad_norm": 1.2164725065231323, "learning_rate": 4.5272000000000005e-06, "loss": 0.0094, "step": 27365 }, { "epoch": 1.4589753158820706, "grad_norm": 2.9991614818573, "learning_rate": 4.527e-06, "loss": 0.0196, "step": 27366 }, { "epoch": 1.4590286293117236, "grad_norm": 2.117384672164917, "learning_rate": 4.5268000000000004e-06, "loss": 0.0269, "step": 27367 }, { "epoch": 1.4590819427413766, "grad_norm": 0.717108428478241, "learning_rate": 4.526600000000001e-06, "loss": 0.009, "step": 27368 }, { "epoch": 1.4591352561710296, "grad_norm": 0.8477725982666016, "learning_rate": 4.5264e-06, "loss": 0.0071, "step": 27369 }, { "epoch": 1.4591885696006823, "grad_norm": 1.372747778892517, "learning_rate": 4.526200000000001e-06, "loss": 0.0129, "step": 27370 }, { "epoch": 1.4592418830303353, "grad_norm": 1.8845975399017334, "learning_rate": 4.526e-06, "loss": 0.0314, "step": 27371 }, { "epoch": 1.4592951964599883, "grad_norm": 2.4158034324645996, "learning_rate": 4.5258e-06, "loss": 0.0166, "step": 27372 }, { "epoch": 1.4593485098896413, "grad_norm": 1.2636754512786865, "learning_rate": 4.5256e-06, "loss": 0.008, "step": 27373 }, { "epoch": 1.4594018233192942, "grad_norm": 2.305588722229004, "learning_rate": 4.525400000000001e-06, "loss": 0.0365, "step": 27374 }, { "epoch": 1.459455136748947, "grad_norm": 0.9741323590278625, "learning_rate": 4.5252e-06, "loss": 0.0126, "step": 27375 }, { "epoch": 1.4595084501786, "grad_norm": 0.7831408977508545, "learning_rate": 4.525000000000001e-06, "loss": 0.0058, "step": 27376 }, { "epoch": 1.459561763608253, "grad_norm": 1.2818692922592163, "learning_rate": 4.5248e-06, "loss": 0.043, "step": 27377 }, { "epoch": 1.4596150770379057, "grad_norm": 1.2417471408843994, "learning_rate": 4.5246e-06, "loss": 0.0219, "step": 27378 }, { "epoch": 1.4596683904675587, "grad_norm": 1.5384738445281982, "learning_rate": 4.5244e-06, "loss": 0.0139, "step": 27379 }, { "epoch": 1.4597217038972117, "grad_norm": 1.0065786838531494, "learning_rate": 4.524200000000001e-06, "loss": 0.0083, "step": 27380 }, { "epoch": 1.4597750173268647, "grad_norm": 2.3448007106781006, "learning_rate": 4.524e-06, "loss": 0.025, "step": 27381 }, { "epoch": 1.4598283307565176, "grad_norm": 2.396819829940796, "learning_rate": 4.523800000000001e-06, "loss": 0.0395, "step": 27382 }, { "epoch": 1.4598816441861704, "grad_norm": 1.959389090538025, "learning_rate": 4.5236e-06, "loss": 0.0177, "step": 27383 }, { "epoch": 1.4599349576158234, "grad_norm": 1.8935751914978027, "learning_rate": 4.523400000000001e-06, "loss": 0.0178, "step": 27384 }, { "epoch": 1.4599882710454763, "grad_norm": 0.7219380140304565, "learning_rate": 4.5232e-06, "loss": 0.009, "step": 27385 }, { "epoch": 1.4600415844751293, "grad_norm": 1.3836041688919067, "learning_rate": 4.5230000000000006e-06, "loss": 0.0085, "step": 27386 }, { "epoch": 1.4600948979047823, "grad_norm": 1.3243318796157837, "learning_rate": 4.5228e-06, "loss": 0.0187, "step": 27387 }, { "epoch": 1.460148211334435, "grad_norm": 1.721915602684021, "learning_rate": 4.5226000000000005e-06, "loss": 0.0302, "step": 27388 }, { "epoch": 1.460201524764088, "grad_norm": 0.6671591997146606, "learning_rate": 4.5224e-06, "loss": 0.0039, "step": 27389 }, { "epoch": 1.460254838193741, "grad_norm": 3.2493109703063965, "learning_rate": 4.5222000000000005e-06, "loss": 0.025, "step": 27390 }, { "epoch": 1.460308151623394, "grad_norm": 2.157214879989624, "learning_rate": 4.522e-06, "loss": 0.0198, "step": 27391 }, { "epoch": 1.460361465053047, "grad_norm": 1.2812894582748413, "learning_rate": 4.5218000000000005e-06, "loss": 0.0105, "step": 27392 }, { "epoch": 1.4604147784826997, "grad_norm": 3.024137496948242, "learning_rate": 4.5216e-06, "loss": 0.0182, "step": 27393 }, { "epoch": 1.4604680919123527, "grad_norm": 0.5077845454216003, "learning_rate": 4.5214000000000005e-06, "loss": 0.0033, "step": 27394 }, { "epoch": 1.4605214053420057, "grad_norm": 1.893665075302124, "learning_rate": 4.521200000000001e-06, "loss": 0.0118, "step": 27395 }, { "epoch": 1.4605747187716585, "grad_norm": 1.5953294038772583, "learning_rate": 4.521e-06, "loss": 0.0073, "step": 27396 }, { "epoch": 1.4606280322013114, "grad_norm": 1.7998757362365723, "learning_rate": 4.5208e-06, "loss": 0.0201, "step": 27397 }, { "epoch": 1.4606813456309644, "grad_norm": 0.8798307180404663, "learning_rate": 4.5206e-06, "loss": 0.0115, "step": 27398 }, { "epoch": 1.4607346590606174, "grad_norm": 3.4852240085601807, "learning_rate": 4.5204e-06, "loss": 0.0277, "step": 27399 }, { "epoch": 1.4607879724902704, "grad_norm": 0.27753493189811707, "learning_rate": 4.5202e-06, "loss": 0.0023, "step": 27400 }, { "epoch": 1.4608412859199231, "grad_norm": 1.8458610773086548, "learning_rate": 4.520000000000001e-06, "loss": 0.0201, "step": 27401 }, { "epoch": 1.460894599349576, "grad_norm": 0.9586969614028931, "learning_rate": 4.5198e-06, "loss": 0.0145, "step": 27402 }, { "epoch": 1.460947912779229, "grad_norm": 2.1399919986724854, "learning_rate": 4.5196e-06, "loss": 0.0178, "step": 27403 }, { "epoch": 1.461001226208882, "grad_norm": 0.7661452889442444, "learning_rate": 4.5194e-06, "loss": 0.0041, "step": 27404 }, { "epoch": 1.461054539638535, "grad_norm": 0.8234255313873291, "learning_rate": 4.5192e-06, "loss": 0.0056, "step": 27405 }, { "epoch": 1.4611078530681878, "grad_norm": 1.1107553243637085, "learning_rate": 4.519e-06, "loss": 0.0145, "step": 27406 }, { "epoch": 1.4611611664978408, "grad_norm": 1.403002142906189, "learning_rate": 4.518800000000001e-06, "loss": 0.0248, "step": 27407 }, { "epoch": 1.4612144799274938, "grad_norm": 1.3327980041503906, "learning_rate": 4.5186e-06, "loss": 0.018, "step": 27408 }, { "epoch": 1.4612677933571467, "grad_norm": 0.6492727994918823, "learning_rate": 4.518400000000001e-06, "loss": 0.0058, "step": 27409 }, { "epoch": 1.4613211067867997, "grad_norm": 1.455309271812439, "learning_rate": 4.5182e-06, "loss": 0.0114, "step": 27410 }, { "epoch": 1.4613744202164525, "grad_norm": 1.310361385345459, "learning_rate": 4.518e-06, "loss": 0.0067, "step": 27411 }, { "epoch": 1.4614277336461055, "grad_norm": 2.1545259952545166, "learning_rate": 4.5178e-06, "loss": 0.0387, "step": 27412 }, { "epoch": 1.4614810470757584, "grad_norm": 0.9917660355567932, "learning_rate": 4.517600000000001e-06, "loss": 0.017, "step": 27413 }, { "epoch": 1.4615343605054112, "grad_norm": 2.28690242767334, "learning_rate": 4.5174e-06, "loss": 0.0634, "step": 27414 }, { "epoch": 1.4615876739350642, "grad_norm": 2.425948143005371, "learning_rate": 4.5172000000000006e-06, "loss": 0.0122, "step": 27415 }, { "epoch": 1.4616409873647171, "grad_norm": 1.6980469226837158, "learning_rate": 4.517e-06, "loss": 0.0151, "step": 27416 }, { "epoch": 1.4616943007943701, "grad_norm": 1.6354382038116455, "learning_rate": 4.5168000000000005e-06, "loss": 0.019, "step": 27417 }, { "epoch": 1.461747614224023, "grad_norm": 1.8709813356399536, "learning_rate": 4.5166e-06, "loss": 0.0131, "step": 27418 }, { "epoch": 1.4618009276536759, "grad_norm": 2.2643589973449707, "learning_rate": 4.5164000000000005e-06, "loss": 0.0118, "step": 27419 }, { "epoch": 1.4618542410833288, "grad_norm": 1.1468640565872192, "learning_rate": 4.516200000000001e-06, "loss": 0.0136, "step": 27420 }, { "epoch": 1.4619075545129818, "grad_norm": 3.520848274230957, "learning_rate": 4.5160000000000005e-06, "loss": 0.0175, "step": 27421 }, { "epoch": 1.4619608679426348, "grad_norm": 2.556868076324463, "learning_rate": 4.5158e-06, "loss": 0.0323, "step": 27422 }, { "epoch": 1.4620141813722878, "grad_norm": 1.245365023612976, "learning_rate": 4.5156000000000004e-06, "loss": 0.0057, "step": 27423 }, { "epoch": 1.4620674948019405, "grad_norm": 1.5835813283920288, "learning_rate": 4.5154e-06, "loss": 0.0272, "step": 27424 }, { "epoch": 1.4621208082315935, "grad_norm": 2.02095103263855, "learning_rate": 4.5152e-06, "loss": 0.0209, "step": 27425 }, { "epoch": 1.4621741216612465, "grad_norm": 0.732404351234436, "learning_rate": 4.515000000000001e-06, "loss": 0.0048, "step": 27426 }, { "epoch": 1.4622274350908995, "grad_norm": 2.1099674701690674, "learning_rate": 4.5148e-06, "loss": 0.029, "step": 27427 }, { "epoch": 1.4622807485205525, "grad_norm": 1.202919363975525, "learning_rate": 4.5146e-06, "loss": 0.0183, "step": 27428 }, { "epoch": 1.4623340619502052, "grad_norm": 3.6832339763641357, "learning_rate": 4.5144e-06, "loss": 0.019, "step": 27429 }, { "epoch": 1.4623873753798582, "grad_norm": 1.629345417022705, "learning_rate": 4.5142e-06, "loss": 0.0184, "step": 27430 }, { "epoch": 1.4624406888095112, "grad_norm": 3.0125019550323486, "learning_rate": 4.514e-06, "loss": 0.0148, "step": 27431 }, { "epoch": 1.462494002239164, "grad_norm": 1.823890209197998, "learning_rate": 4.513800000000001e-06, "loss": 0.0231, "step": 27432 }, { "epoch": 1.462547315668817, "grad_norm": 2.326507329940796, "learning_rate": 4.5136e-06, "loss": 0.026, "step": 27433 }, { "epoch": 1.4626006290984699, "grad_norm": 1.1043390035629272, "learning_rate": 4.513400000000001e-06, "loss": 0.0087, "step": 27434 }, { "epoch": 1.4626539425281229, "grad_norm": 3.7340800762176514, "learning_rate": 4.5132e-06, "loss": 0.0719, "step": 27435 }, { "epoch": 1.4627072559577758, "grad_norm": 2.26588773727417, "learning_rate": 4.513e-06, "loss": 0.0217, "step": 27436 }, { "epoch": 1.4627605693874286, "grad_norm": 3.5315446853637695, "learning_rate": 4.5128e-06, "loss": 0.0399, "step": 27437 }, { "epoch": 1.4628138828170816, "grad_norm": 1.2572773694992065, "learning_rate": 4.512600000000001e-06, "loss": 0.0117, "step": 27438 }, { "epoch": 1.4628671962467346, "grad_norm": 2.227980852127075, "learning_rate": 4.5124e-06, "loss": 0.0144, "step": 27439 }, { "epoch": 1.4629205096763875, "grad_norm": 2.338627338409424, "learning_rate": 4.512200000000001e-06, "loss": 0.0197, "step": 27440 }, { "epoch": 1.4629738231060405, "grad_norm": 3.1377339363098145, "learning_rate": 4.512e-06, "loss": 0.0334, "step": 27441 }, { "epoch": 1.4630271365356933, "grad_norm": 1.8089237213134766, "learning_rate": 4.511800000000001e-06, "loss": 0.0233, "step": 27442 }, { "epoch": 1.4630804499653463, "grad_norm": 1.731408953666687, "learning_rate": 4.5116e-06, "loss": 0.0144, "step": 27443 }, { "epoch": 1.4631337633949992, "grad_norm": 2.469224214553833, "learning_rate": 4.5114000000000005e-06, "loss": 0.0219, "step": 27444 }, { "epoch": 1.4631870768246522, "grad_norm": 1.6172034740447998, "learning_rate": 4.511200000000001e-06, "loss": 0.0189, "step": 27445 }, { "epoch": 1.4632403902543052, "grad_norm": 1.564087986946106, "learning_rate": 4.5110000000000005e-06, "loss": 0.0097, "step": 27446 }, { "epoch": 1.463293703683958, "grad_norm": 0.717492401599884, "learning_rate": 4.5108e-06, "loss": 0.007, "step": 27447 }, { "epoch": 1.463347017113611, "grad_norm": 2.8634095191955566, "learning_rate": 4.5106000000000005e-06, "loss": 0.0344, "step": 27448 }, { "epoch": 1.463400330543264, "grad_norm": 1.2391972541809082, "learning_rate": 4.5104e-06, "loss": 0.0138, "step": 27449 }, { "epoch": 1.4634536439729167, "grad_norm": 1.4805938005447388, "learning_rate": 4.5102000000000005e-06, "loss": 0.0169, "step": 27450 }, { "epoch": 1.4635069574025696, "grad_norm": 1.5108604431152344, "learning_rate": 4.510000000000001e-06, "loss": 0.0152, "step": 27451 }, { "epoch": 1.4635602708322226, "grad_norm": 0.9572159051895142, "learning_rate": 4.5098000000000004e-06, "loss": 0.0133, "step": 27452 }, { "epoch": 1.4636135842618756, "grad_norm": 4.799041748046875, "learning_rate": 4.509600000000001e-06, "loss": 0.0669, "step": 27453 }, { "epoch": 1.4636668976915286, "grad_norm": 1.9154213666915894, "learning_rate": 4.5094e-06, "loss": 0.0211, "step": 27454 }, { "epoch": 1.4637202111211813, "grad_norm": 2.779632329940796, "learning_rate": 4.5092e-06, "loss": 0.0402, "step": 27455 }, { "epoch": 1.4637735245508343, "grad_norm": 1.9177513122558594, "learning_rate": 4.509e-06, "loss": 0.0214, "step": 27456 }, { "epoch": 1.4638268379804873, "grad_norm": 2.249128818511963, "learning_rate": 4.508800000000001e-06, "loss": 0.0502, "step": 27457 }, { "epoch": 1.4638801514101403, "grad_norm": 1.0917353630065918, "learning_rate": 4.5086e-06, "loss": 0.0096, "step": 27458 }, { "epoch": 1.4639334648397933, "grad_norm": 2.0419981479644775, "learning_rate": 4.508400000000001e-06, "loss": 0.0391, "step": 27459 }, { "epoch": 1.463986778269446, "grad_norm": 2.372647285461426, "learning_rate": 4.5082e-06, "loss": 0.021, "step": 27460 }, { "epoch": 1.464040091699099, "grad_norm": 3.0637032985687256, "learning_rate": 4.508e-06, "loss": 0.0309, "step": 27461 }, { "epoch": 1.464093405128752, "grad_norm": 1.7324693202972412, "learning_rate": 4.5078e-06, "loss": 0.0151, "step": 27462 }, { "epoch": 1.4641467185584047, "grad_norm": 2.903916120529175, "learning_rate": 4.507600000000001e-06, "loss": 0.0419, "step": 27463 }, { "epoch": 1.464200031988058, "grad_norm": 1.2751749753952026, "learning_rate": 4.5074e-06, "loss": 0.0207, "step": 27464 }, { "epoch": 1.4642533454177107, "grad_norm": 1.399485468864441, "learning_rate": 4.507200000000001e-06, "loss": 0.0105, "step": 27465 }, { "epoch": 1.4643066588473637, "grad_norm": 1.8061450719833374, "learning_rate": 4.507e-06, "loss": 0.0249, "step": 27466 }, { "epoch": 1.4643599722770166, "grad_norm": 1.4278972148895264, "learning_rate": 4.506800000000001e-06, "loss": 0.0213, "step": 27467 }, { "epoch": 1.4644132857066694, "grad_norm": 2.0163094997406006, "learning_rate": 4.5066e-06, "loss": 0.0319, "step": 27468 }, { "epoch": 1.4644665991363224, "grad_norm": 4.289654731750488, "learning_rate": 4.506400000000001e-06, "loss": 0.0465, "step": 27469 }, { "epoch": 1.4645199125659754, "grad_norm": 3.075767993927002, "learning_rate": 4.5062e-06, "loss": 0.0445, "step": 27470 }, { "epoch": 1.4645732259956283, "grad_norm": 0.8690380454063416, "learning_rate": 4.5060000000000006e-06, "loss": 0.008, "step": 27471 }, { "epoch": 1.4646265394252813, "grad_norm": 2.1853771209716797, "learning_rate": 4.5058e-06, "loss": 0.0184, "step": 27472 }, { "epoch": 1.464679852854934, "grad_norm": 1.6080371141433716, "learning_rate": 4.5056000000000005e-06, "loss": 0.0247, "step": 27473 }, { "epoch": 1.464733166284587, "grad_norm": 1.5294467210769653, "learning_rate": 4.5054e-06, "loss": 0.0378, "step": 27474 }, { "epoch": 1.46478647971424, "grad_norm": 0.8523341417312622, "learning_rate": 4.5052000000000005e-06, "loss": 0.014, "step": 27475 }, { "epoch": 1.464839793143893, "grad_norm": 0.548957109451294, "learning_rate": 4.505e-06, "loss": 0.0042, "step": 27476 }, { "epoch": 1.464893106573546, "grad_norm": 1.8680100440979004, "learning_rate": 4.5048000000000005e-06, "loss": 0.0178, "step": 27477 }, { "epoch": 1.4649464200031987, "grad_norm": 0.8935102224349976, "learning_rate": 4.504600000000001e-06, "loss": 0.0085, "step": 27478 }, { "epoch": 1.4649997334328517, "grad_norm": 1.4898691177368164, "learning_rate": 4.5044000000000004e-06, "loss": 0.0198, "step": 27479 }, { "epoch": 1.4650530468625047, "grad_norm": 0.7981232404708862, "learning_rate": 4.5042e-06, "loss": 0.004, "step": 27480 }, { "epoch": 1.4651063602921575, "grad_norm": 1.7425886392593384, "learning_rate": 4.504e-06, "loss": 0.0244, "step": 27481 }, { "epoch": 1.4651596737218107, "grad_norm": 1.413464069366455, "learning_rate": 4.5038e-06, "loss": 0.0136, "step": 27482 }, { "epoch": 1.4652129871514634, "grad_norm": 1.8793336153030396, "learning_rate": 4.5036e-06, "loss": 0.0315, "step": 27483 }, { "epoch": 1.4652663005811164, "grad_norm": 1.1385291814804077, "learning_rate": 4.503400000000001e-06, "loss": 0.0085, "step": 27484 }, { "epoch": 1.4653196140107694, "grad_norm": 2.64802622795105, "learning_rate": 4.5032e-06, "loss": 0.0375, "step": 27485 }, { "epoch": 1.4653729274404221, "grad_norm": 1.9484403133392334, "learning_rate": 4.503e-06, "loss": 0.0356, "step": 27486 }, { "epoch": 1.4654262408700751, "grad_norm": 1.6137295961380005, "learning_rate": 4.5028e-06, "loss": 0.0267, "step": 27487 }, { "epoch": 1.465479554299728, "grad_norm": 2.592496871948242, "learning_rate": 4.5026e-06, "loss": 0.0226, "step": 27488 }, { "epoch": 1.465532867729381, "grad_norm": 1.5491089820861816, "learning_rate": 4.5024e-06, "loss": 0.0178, "step": 27489 }, { "epoch": 1.465586181159034, "grad_norm": 1.1856324672698975, "learning_rate": 4.502200000000001e-06, "loss": 0.0216, "step": 27490 }, { "epoch": 1.4656394945886868, "grad_norm": 1.2614620923995972, "learning_rate": 4.502e-06, "loss": 0.0092, "step": 27491 }, { "epoch": 1.4656928080183398, "grad_norm": 2.908579111099243, "learning_rate": 4.501800000000001e-06, "loss": 0.0718, "step": 27492 }, { "epoch": 1.4657461214479928, "grad_norm": 2.064084529876709, "learning_rate": 4.5016e-06, "loss": 0.0549, "step": 27493 }, { "epoch": 1.4657994348776457, "grad_norm": 4.251945972442627, "learning_rate": 4.5014e-06, "loss": 0.0281, "step": 27494 }, { "epoch": 1.4658527483072987, "grad_norm": 2.2892038822174072, "learning_rate": 4.5012e-06, "loss": 0.0148, "step": 27495 }, { "epoch": 1.4659060617369515, "grad_norm": 2.14119553565979, "learning_rate": 4.501000000000001e-06, "loss": 0.0194, "step": 27496 }, { "epoch": 1.4659593751666045, "grad_norm": 1.2533164024353027, "learning_rate": 4.5008e-06, "loss": 0.0115, "step": 27497 }, { "epoch": 1.4660126885962574, "grad_norm": 1.1575196981430054, "learning_rate": 4.500600000000001e-06, "loss": 0.0082, "step": 27498 }, { "epoch": 1.4660660020259102, "grad_norm": 1.4174988269805908, "learning_rate": 4.5004e-06, "loss": 0.0208, "step": 27499 }, { "epoch": 1.4661193154555632, "grad_norm": 1.3076887130737305, "learning_rate": 4.5002e-06, "loss": 0.0197, "step": 27500 }, { "epoch": 1.4661726288852162, "grad_norm": 2.8478469848632812, "learning_rate": 4.5e-06, "loss": 0.0335, "step": 27501 }, { "epoch": 1.4662259423148691, "grad_norm": 0.5182252526283264, "learning_rate": 4.4998000000000005e-06, "loss": 0.0044, "step": 27502 }, { "epoch": 1.4662792557445221, "grad_norm": 1.5181983709335327, "learning_rate": 4.499600000000001e-06, "loss": 0.0154, "step": 27503 }, { "epoch": 1.4663325691741749, "grad_norm": 2.070892333984375, "learning_rate": 4.4994000000000005e-06, "loss": 0.0334, "step": 27504 }, { "epoch": 1.4663858826038279, "grad_norm": 1.5448102951049805, "learning_rate": 4.4992e-06, "loss": 0.0151, "step": 27505 }, { "epoch": 1.4664391960334808, "grad_norm": 1.0177010297775269, "learning_rate": 4.4990000000000005e-06, "loss": 0.0187, "step": 27506 }, { "epoch": 1.4664925094631338, "grad_norm": 1.859560489654541, "learning_rate": 4.4988e-06, "loss": 0.0153, "step": 27507 }, { "epoch": 1.4665458228927868, "grad_norm": 1.3354911804199219, "learning_rate": 4.4986000000000004e-06, "loss": 0.019, "step": 27508 }, { "epoch": 1.4665991363224395, "grad_norm": 1.890325903892517, "learning_rate": 4.498400000000001e-06, "loss": 0.0235, "step": 27509 }, { "epoch": 1.4666524497520925, "grad_norm": 2.636953115463257, "learning_rate": 4.4982e-06, "loss": 0.0456, "step": 27510 }, { "epoch": 1.4667057631817455, "grad_norm": 0.9696246385574341, "learning_rate": 4.498e-06, "loss": 0.0147, "step": 27511 }, { "epoch": 1.4667590766113985, "grad_norm": 0.6037735342979431, "learning_rate": 4.4978e-06, "loss": 0.0051, "step": 27512 }, { "epoch": 1.4668123900410515, "grad_norm": 1.0024312734603882, "learning_rate": 4.4976e-06, "loss": 0.0099, "step": 27513 }, { "epoch": 1.4668657034707042, "grad_norm": 1.4628725051879883, "learning_rate": 4.4974e-06, "loss": 0.016, "step": 27514 }, { "epoch": 1.4669190169003572, "grad_norm": 0.8308153748512268, "learning_rate": 4.497200000000001e-06, "loss": 0.0195, "step": 27515 }, { "epoch": 1.4669723303300102, "grad_norm": 1.337232232093811, "learning_rate": 4.497e-06, "loss": 0.0103, "step": 27516 }, { "epoch": 1.467025643759663, "grad_norm": 1.1524115800857544, "learning_rate": 4.496800000000001e-06, "loss": 0.018, "step": 27517 }, { "epoch": 1.467078957189316, "grad_norm": 2.975935697555542, "learning_rate": 4.4966e-06, "loss": 0.0382, "step": 27518 }, { "epoch": 1.467132270618969, "grad_norm": 1.6140124797821045, "learning_rate": 4.4964e-06, "loss": 0.0121, "step": 27519 }, { "epoch": 1.4671855840486219, "grad_norm": 1.7665629386901855, "learning_rate": 4.4962e-06, "loss": 0.0377, "step": 27520 }, { "epoch": 1.4672388974782749, "grad_norm": 1.8338543176651, "learning_rate": 4.496000000000001e-06, "loss": 0.018, "step": 27521 }, { "epoch": 1.4672922109079276, "grad_norm": 1.811651349067688, "learning_rate": 4.4958e-06, "loss": 0.0211, "step": 27522 }, { "epoch": 1.4673455243375806, "grad_norm": 1.8556278944015503, "learning_rate": 4.495600000000001e-06, "loss": 0.026, "step": 27523 }, { "epoch": 1.4673988377672336, "grad_norm": 2.212764263153076, "learning_rate": 4.4954e-06, "loss": 0.0458, "step": 27524 }, { "epoch": 1.4674521511968865, "grad_norm": 1.364972710609436, "learning_rate": 4.4952e-06, "loss": 0.0113, "step": 27525 }, { "epoch": 1.4675054646265395, "grad_norm": 2.4316864013671875, "learning_rate": 4.495e-06, "loss": 0.0315, "step": 27526 }, { "epoch": 1.4675587780561923, "grad_norm": 1.3522895574569702, "learning_rate": 4.4948000000000006e-06, "loss": 0.0135, "step": 27527 }, { "epoch": 1.4676120914858453, "grad_norm": 1.8039170503616333, "learning_rate": 4.494600000000001e-06, "loss": 0.0091, "step": 27528 }, { "epoch": 1.4676654049154982, "grad_norm": 0.8807884454727173, "learning_rate": 4.4944000000000005e-06, "loss": 0.0076, "step": 27529 }, { "epoch": 1.4677187183451512, "grad_norm": 1.6950651407241821, "learning_rate": 4.4942e-06, "loss": 0.0273, "step": 27530 }, { "epoch": 1.4677720317748042, "grad_norm": 0.9693045616149902, "learning_rate": 4.4940000000000005e-06, "loss": 0.0056, "step": 27531 }, { "epoch": 1.467825345204457, "grad_norm": 1.2759884595870972, "learning_rate": 4.4938e-06, "loss": 0.0302, "step": 27532 }, { "epoch": 1.46787865863411, "grad_norm": 1.4030365943908691, "learning_rate": 4.4936000000000005e-06, "loss": 0.0124, "step": 27533 }, { "epoch": 1.467931972063763, "grad_norm": 0.4494141936302185, "learning_rate": 4.493400000000001e-06, "loss": 0.0047, "step": 27534 }, { "epoch": 1.4679852854934157, "grad_norm": 1.3710559606552124, "learning_rate": 4.4932000000000005e-06, "loss": 0.0129, "step": 27535 }, { "epoch": 1.4680385989230686, "grad_norm": 1.4909226894378662, "learning_rate": 4.493e-06, "loss": 0.0287, "step": 27536 }, { "epoch": 1.4680919123527216, "grad_norm": 2.001175880432129, "learning_rate": 4.4928e-06, "loss": 0.0352, "step": 27537 }, { "epoch": 1.4681452257823746, "grad_norm": 1.692400574684143, "learning_rate": 4.4926e-06, "loss": 0.0093, "step": 27538 }, { "epoch": 1.4681985392120276, "grad_norm": 2.1276397705078125, "learning_rate": 4.4924e-06, "loss": 0.0232, "step": 27539 }, { "epoch": 1.4682518526416803, "grad_norm": 0.409603476524353, "learning_rate": 4.492200000000001e-06, "loss": 0.004, "step": 27540 }, { "epoch": 1.4683051660713333, "grad_norm": 2.2265236377716064, "learning_rate": 4.492e-06, "loss": 0.035, "step": 27541 }, { "epoch": 1.4683584795009863, "grad_norm": 0.7952156662940979, "learning_rate": 4.491800000000001e-06, "loss": 0.0116, "step": 27542 }, { "epoch": 1.4684117929306393, "grad_norm": 0.6275676488876343, "learning_rate": 4.4916e-06, "loss": 0.0067, "step": 27543 }, { "epoch": 1.4684651063602923, "grad_norm": 2.211986780166626, "learning_rate": 4.4914e-06, "loss": 0.0087, "step": 27544 }, { "epoch": 1.468518419789945, "grad_norm": 0.8157031536102295, "learning_rate": 4.4912e-06, "loss": 0.0068, "step": 27545 }, { "epoch": 1.468571733219598, "grad_norm": 1.5391333103179932, "learning_rate": 4.491000000000001e-06, "loss": 0.0202, "step": 27546 }, { "epoch": 1.468625046649251, "grad_norm": 0.8855512738227844, "learning_rate": 4.4908e-06, "loss": 0.0053, "step": 27547 }, { "epoch": 1.468678360078904, "grad_norm": 2.5377566814422607, "learning_rate": 4.490600000000001e-06, "loss": 0.0113, "step": 27548 }, { "epoch": 1.468731673508557, "grad_norm": 0.30937349796295166, "learning_rate": 4.4904e-06, "loss": 0.0031, "step": 27549 }, { "epoch": 1.4687849869382097, "grad_norm": 2.201009511947632, "learning_rate": 4.4902e-06, "loss": 0.0326, "step": 27550 }, { "epoch": 1.4688383003678627, "grad_norm": 1.856069564819336, "learning_rate": 4.49e-06, "loss": 0.0239, "step": 27551 }, { "epoch": 1.4688916137975156, "grad_norm": 1.0579153299331665, "learning_rate": 4.489800000000001e-06, "loss": 0.0081, "step": 27552 }, { "epoch": 1.4689449272271684, "grad_norm": 0.9029604196548462, "learning_rate": 4.489600000000001e-06, "loss": 0.0093, "step": 27553 }, { "epoch": 1.4689982406568214, "grad_norm": 2.2024037837982178, "learning_rate": 4.489400000000001e-06, "loss": 0.0354, "step": 27554 }, { "epoch": 1.4690515540864744, "grad_norm": 0.937031626701355, "learning_rate": 4.4892e-06, "loss": 0.0068, "step": 27555 }, { "epoch": 1.4691048675161273, "grad_norm": 1.1125245094299316, "learning_rate": 4.4890000000000006e-06, "loss": 0.0107, "step": 27556 }, { "epoch": 1.4691581809457803, "grad_norm": 1.1785145998001099, "learning_rate": 4.4888e-06, "loss": 0.0101, "step": 27557 }, { "epoch": 1.469211494375433, "grad_norm": 2.92885422706604, "learning_rate": 4.4886000000000005e-06, "loss": 0.0378, "step": 27558 }, { "epoch": 1.469264807805086, "grad_norm": 1.5238667726516724, "learning_rate": 4.488400000000001e-06, "loss": 0.0195, "step": 27559 }, { "epoch": 1.469318121234739, "grad_norm": 0.6653645634651184, "learning_rate": 4.4882000000000005e-06, "loss": 0.0061, "step": 27560 }, { "epoch": 1.469371434664392, "grad_norm": 0.8032288551330566, "learning_rate": 4.488e-06, "loss": 0.0065, "step": 27561 }, { "epoch": 1.469424748094045, "grad_norm": 1.4650843143463135, "learning_rate": 4.4878000000000005e-06, "loss": 0.014, "step": 27562 }, { "epoch": 1.4694780615236978, "grad_norm": 0.5178963541984558, "learning_rate": 4.4876e-06, "loss": 0.0054, "step": 27563 }, { "epoch": 1.4695313749533507, "grad_norm": 2.033285140991211, "learning_rate": 4.4874000000000004e-06, "loss": 0.0138, "step": 27564 }, { "epoch": 1.4695846883830037, "grad_norm": 0.4610355496406555, "learning_rate": 4.487200000000001e-06, "loss": 0.003, "step": 27565 }, { "epoch": 1.4696380018126567, "grad_norm": 1.6171869039535522, "learning_rate": 4.487e-06, "loss": 0.0186, "step": 27566 }, { "epoch": 1.4696913152423097, "grad_norm": 1.1895939111709595, "learning_rate": 4.486800000000001e-06, "loss": 0.0061, "step": 27567 }, { "epoch": 1.4697446286719624, "grad_norm": 1.9895222187042236, "learning_rate": 4.4866e-06, "loss": 0.0079, "step": 27568 }, { "epoch": 1.4697979421016154, "grad_norm": 4.390725135803223, "learning_rate": 4.4864e-06, "loss": 0.0617, "step": 27569 }, { "epoch": 1.4698512555312684, "grad_norm": 3.0310468673706055, "learning_rate": 4.4862e-06, "loss": 0.0397, "step": 27570 }, { "epoch": 1.4699045689609211, "grad_norm": 1.1234419345855713, "learning_rate": 4.486000000000001e-06, "loss": 0.0096, "step": 27571 }, { "epoch": 1.4699578823905741, "grad_norm": 1.0018380880355835, "learning_rate": 4.4858e-06, "loss": 0.018, "step": 27572 }, { "epoch": 1.470011195820227, "grad_norm": 1.506426215171814, "learning_rate": 4.485600000000001e-06, "loss": 0.0093, "step": 27573 }, { "epoch": 1.47006450924988, "grad_norm": 3.924111843109131, "learning_rate": 4.4854e-06, "loss": 0.0238, "step": 27574 }, { "epoch": 1.470117822679533, "grad_norm": 3.9856433868408203, "learning_rate": 4.4852e-06, "loss": 0.0346, "step": 27575 }, { "epoch": 1.4701711361091858, "grad_norm": 1.8097255229949951, "learning_rate": 4.485e-06, "loss": 0.0501, "step": 27576 }, { "epoch": 1.4702244495388388, "grad_norm": 3.0993893146514893, "learning_rate": 4.484800000000001e-06, "loss": 0.0605, "step": 27577 }, { "epoch": 1.4702777629684918, "grad_norm": 0.9460579752922058, "learning_rate": 4.4846e-06, "loss": 0.0073, "step": 27578 }, { "epoch": 1.4703310763981448, "grad_norm": 1.662042260169983, "learning_rate": 4.484400000000001e-06, "loss": 0.0571, "step": 27579 }, { "epoch": 1.4703843898277977, "grad_norm": 1.3895100355148315, "learning_rate": 4.4842e-06, "loss": 0.0161, "step": 27580 }, { "epoch": 1.4704377032574505, "grad_norm": 2.353527784347534, "learning_rate": 4.484000000000001e-06, "loss": 0.0274, "step": 27581 }, { "epoch": 1.4704910166871035, "grad_norm": 3.660046100616455, "learning_rate": 4.4838e-06, "loss": 0.0352, "step": 27582 }, { "epoch": 1.4705443301167564, "grad_norm": 4.437188148498535, "learning_rate": 4.4836000000000006e-06, "loss": 0.0407, "step": 27583 }, { "epoch": 1.4705976435464092, "grad_norm": 2.7585856914520264, "learning_rate": 4.4834e-06, "loss": 0.022, "step": 27584 }, { "epoch": 1.4706509569760624, "grad_norm": 3.0782783031463623, "learning_rate": 4.4832000000000005e-06, "loss": 0.0436, "step": 27585 }, { "epoch": 1.4707042704057152, "grad_norm": 1.0299707651138306, "learning_rate": 4.483e-06, "loss": 0.0078, "step": 27586 }, { "epoch": 1.4707575838353681, "grad_norm": 0.9760283827781677, "learning_rate": 4.4828000000000005e-06, "loss": 0.0117, "step": 27587 }, { "epoch": 1.4708108972650211, "grad_norm": 1.7781003713607788, "learning_rate": 4.4826e-06, "loss": 0.0082, "step": 27588 }, { "epoch": 1.4708642106946739, "grad_norm": 2.01967453956604, "learning_rate": 4.4824000000000005e-06, "loss": 0.0206, "step": 27589 }, { "epoch": 1.4709175241243269, "grad_norm": 1.252862811088562, "learning_rate": 4.4822e-06, "loss": 0.0083, "step": 27590 }, { "epoch": 1.4709708375539798, "grad_norm": 2.5075931549072266, "learning_rate": 4.4820000000000005e-06, "loss": 0.0475, "step": 27591 }, { "epoch": 1.4710241509836328, "grad_norm": 1.5386043787002563, "learning_rate": 4.481800000000001e-06, "loss": 0.0158, "step": 27592 }, { "epoch": 1.4710774644132858, "grad_norm": 3.225475311279297, "learning_rate": 4.4816000000000004e-06, "loss": 0.0356, "step": 27593 }, { "epoch": 1.4711307778429386, "grad_norm": 2.734403133392334, "learning_rate": 4.4814e-06, "loss": 0.0421, "step": 27594 }, { "epoch": 1.4711840912725915, "grad_norm": 0.9754893779754639, "learning_rate": 4.4812e-06, "loss": 0.0199, "step": 27595 }, { "epoch": 1.4712374047022445, "grad_norm": 1.650366187095642, "learning_rate": 4.481e-06, "loss": 0.0142, "step": 27596 }, { "epoch": 1.4712907181318975, "grad_norm": 2.2346909046173096, "learning_rate": 4.4808e-06, "loss": 0.0267, "step": 27597 }, { "epoch": 1.4713440315615505, "grad_norm": 1.8972889184951782, "learning_rate": 4.480600000000001e-06, "loss": 0.0463, "step": 27598 }, { "epoch": 1.4713973449912032, "grad_norm": 2.5063886642456055, "learning_rate": 4.4804e-06, "loss": 0.0369, "step": 27599 }, { "epoch": 1.4714506584208562, "grad_norm": 1.0672845840454102, "learning_rate": 4.4802e-06, "loss": 0.0119, "step": 27600 }, { "epoch": 1.4715039718505092, "grad_norm": 0.6421589255332947, "learning_rate": 4.48e-06, "loss": 0.0053, "step": 27601 }, { "epoch": 1.471557285280162, "grad_norm": 2.100487232208252, "learning_rate": 4.4798e-06, "loss": 0.0254, "step": 27602 }, { "epoch": 1.471610598709815, "grad_norm": 0.8822843432426453, "learning_rate": 4.4796e-06, "loss": 0.0101, "step": 27603 }, { "epoch": 1.471663912139468, "grad_norm": 3.404995918273926, "learning_rate": 4.479400000000001e-06, "loss": 0.0334, "step": 27604 }, { "epoch": 1.4717172255691209, "grad_norm": 3.356708526611328, "learning_rate": 4.4792e-06, "loss": 0.0411, "step": 27605 }, { "epoch": 1.4717705389987739, "grad_norm": 3.8490359783172607, "learning_rate": 4.479000000000001e-06, "loss": 0.024, "step": 27606 }, { "epoch": 1.4718238524284266, "grad_norm": 2.7227542400360107, "learning_rate": 4.4788e-06, "loss": 0.0389, "step": 27607 }, { "epoch": 1.4718771658580796, "grad_norm": 1.664414882659912, "learning_rate": 4.4786e-06, "loss": 0.0161, "step": 27608 }, { "epoch": 1.4719304792877326, "grad_norm": 0.6210213303565979, "learning_rate": 4.4784e-06, "loss": 0.0053, "step": 27609 }, { "epoch": 1.4719837927173856, "grad_norm": 2.402283191680908, "learning_rate": 4.478200000000001e-06, "loss": 0.0372, "step": 27610 }, { "epoch": 1.4720371061470385, "grad_norm": 0.6219018697738647, "learning_rate": 4.478e-06, "loss": 0.0065, "step": 27611 }, { "epoch": 1.4720904195766913, "grad_norm": 4.515679359436035, "learning_rate": 4.4778000000000006e-06, "loss": 0.0273, "step": 27612 }, { "epoch": 1.4721437330063443, "grad_norm": 0.9618120789527893, "learning_rate": 4.4776e-06, "loss": 0.0098, "step": 27613 }, { "epoch": 1.4721970464359972, "grad_norm": 3.3652191162109375, "learning_rate": 4.4774000000000005e-06, "loss": 0.022, "step": 27614 }, { "epoch": 1.4722503598656502, "grad_norm": 1.1500656604766846, "learning_rate": 4.4772e-06, "loss": 0.0087, "step": 27615 }, { "epoch": 1.4723036732953032, "grad_norm": 2.307607889175415, "learning_rate": 4.4770000000000005e-06, "loss": 0.0087, "step": 27616 }, { "epoch": 1.472356986724956, "grad_norm": 0.9497547149658203, "learning_rate": 4.476800000000001e-06, "loss": 0.0064, "step": 27617 }, { "epoch": 1.472410300154609, "grad_norm": 1.7067005634307861, "learning_rate": 4.4766000000000005e-06, "loss": 0.0235, "step": 27618 }, { "epoch": 1.472463613584262, "grad_norm": 1.7323187589645386, "learning_rate": 4.4764e-06, "loss": 0.0318, "step": 27619 }, { "epoch": 1.4725169270139147, "grad_norm": 3.0985019207000732, "learning_rate": 4.4762000000000004e-06, "loss": 0.0702, "step": 27620 }, { "epoch": 1.4725702404435677, "grad_norm": 0.5958526730537415, "learning_rate": 4.476e-06, "loss": 0.0039, "step": 27621 }, { "epoch": 1.4726235538732206, "grad_norm": 1.8566261529922485, "learning_rate": 4.4758e-06, "loss": 0.0292, "step": 27622 }, { "epoch": 1.4726768673028736, "grad_norm": 1.2528156042099, "learning_rate": 4.475600000000001e-06, "loss": 0.0144, "step": 27623 }, { "epoch": 1.4727301807325266, "grad_norm": 2.9686336517333984, "learning_rate": 4.4754e-06, "loss": 0.0298, "step": 27624 }, { "epoch": 1.4727834941621794, "grad_norm": 0.8225787281990051, "learning_rate": 4.4752e-06, "loss": 0.0049, "step": 27625 }, { "epoch": 1.4728368075918323, "grad_norm": 2.259195566177368, "learning_rate": 4.475e-06, "loss": 0.0284, "step": 27626 }, { "epoch": 1.4728901210214853, "grad_norm": 3.3912158012390137, "learning_rate": 4.4748e-06, "loss": 0.0271, "step": 27627 }, { "epoch": 1.4729434344511383, "grad_norm": 1.220275640487671, "learning_rate": 4.4746e-06, "loss": 0.0288, "step": 27628 }, { "epoch": 1.4729967478807913, "grad_norm": 1.5357365608215332, "learning_rate": 4.474400000000001e-06, "loss": 0.0294, "step": 27629 }, { "epoch": 1.473050061310444, "grad_norm": 1.821997046470642, "learning_rate": 4.4742e-06, "loss": 0.0228, "step": 27630 }, { "epoch": 1.473103374740097, "grad_norm": 2.3196277618408203, "learning_rate": 4.474000000000001e-06, "loss": 0.013, "step": 27631 }, { "epoch": 1.47315668816975, "grad_norm": 1.8444933891296387, "learning_rate": 4.4738e-06, "loss": 0.0167, "step": 27632 }, { "epoch": 1.473210001599403, "grad_norm": 1.6405562162399292, "learning_rate": 4.4736e-06, "loss": 0.0559, "step": 27633 }, { "epoch": 1.473263315029056, "grad_norm": 1.61183500289917, "learning_rate": 4.4734e-06, "loss": 0.0109, "step": 27634 }, { "epoch": 1.4733166284587087, "grad_norm": 1.572681188583374, "learning_rate": 4.473200000000001e-06, "loss": 0.0285, "step": 27635 }, { "epoch": 1.4733699418883617, "grad_norm": 1.4927324056625366, "learning_rate": 4.473e-06, "loss": 0.0134, "step": 27636 }, { "epoch": 1.4734232553180147, "grad_norm": 1.7383009195327759, "learning_rate": 4.472800000000001e-06, "loss": 0.0402, "step": 27637 }, { "epoch": 1.4734765687476674, "grad_norm": 1.9967834949493408, "learning_rate": 4.4726e-06, "loss": 0.0298, "step": 27638 }, { "epoch": 1.4735298821773204, "grad_norm": 2.5696523189544678, "learning_rate": 4.472400000000001e-06, "loss": 0.0308, "step": 27639 }, { "epoch": 1.4735831956069734, "grad_norm": 1.4817041158676147, "learning_rate": 4.4722e-06, "loss": 0.0285, "step": 27640 }, { "epoch": 1.4736365090366264, "grad_norm": 0.7041000723838806, "learning_rate": 4.4720000000000006e-06, "loss": 0.0082, "step": 27641 }, { "epoch": 1.4736898224662793, "grad_norm": 0.7060155868530273, "learning_rate": 4.471800000000001e-06, "loss": 0.0158, "step": 27642 }, { "epoch": 1.473743135895932, "grad_norm": 0.8418468236923218, "learning_rate": 4.4716000000000005e-06, "loss": 0.0232, "step": 27643 }, { "epoch": 1.473796449325585, "grad_norm": 0.9709331393241882, "learning_rate": 4.4714e-06, "loss": 0.0143, "step": 27644 }, { "epoch": 1.473849762755238, "grad_norm": 1.722679615020752, "learning_rate": 4.4712000000000005e-06, "loss": 0.0118, "step": 27645 }, { "epoch": 1.473903076184891, "grad_norm": 1.1469449996948242, "learning_rate": 4.471e-06, "loss": 0.0118, "step": 27646 }, { "epoch": 1.473956389614544, "grad_norm": 0.6730573773384094, "learning_rate": 4.4708000000000005e-06, "loss": 0.0203, "step": 27647 }, { "epoch": 1.4740097030441968, "grad_norm": 1.3318325281143188, "learning_rate": 4.470600000000001e-06, "loss": 0.0215, "step": 27648 }, { "epoch": 1.4740630164738497, "grad_norm": 0.6628910303115845, "learning_rate": 4.4704000000000004e-06, "loss": 0.0064, "step": 27649 }, { "epoch": 1.4741163299035027, "grad_norm": 2.006838798522949, "learning_rate": 4.4702e-06, "loss": 0.0288, "step": 27650 }, { "epoch": 1.4741696433331557, "grad_norm": 2.084468126296997, "learning_rate": 4.47e-06, "loss": 0.0283, "step": 27651 }, { "epoch": 1.4742229567628087, "grad_norm": 2.2430074214935303, "learning_rate": 4.4698e-06, "loss": 0.0366, "step": 27652 }, { "epoch": 1.4742762701924614, "grad_norm": 2.906653642654419, "learning_rate": 4.4696e-06, "loss": 0.0247, "step": 27653 }, { "epoch": 1.4743295836221144, "grad_norm": 1.547553539276123, "learning_rate": 4.469400000000001e-06, "loss": 0.0169, "step": 27654 }, { "epoch": 1.4743828970517674, "grad_norm": 1.5657204389572144, "learning_rate": 4.4692e-06, "loss": 0.0188, "step": 27655 }, { "epoch": 1.4744362104814202, "grad_norm": 0.8705225586891174, "learning_rate": 4.469000000000001e-06, "loss": 0.0065, "step": 27656 }, { "epoch": 1.4744895239110731, "grad_norm": 4.82808256149292, "learning_rate": 4.4688e-06, "loss": 0.0518, "step": 27657 }, { "epoch": 1.474542837340726, "grad_norm": 1.3204559087753296, "learning_rate": 4.4686e-06, "loss": 0.0105, "step": 27658 }, { "epoch": 1.474596150770379, "grad_norm": 2.017451763153076, "learning_rate": 4.4684e-06, "loss": 0.0429, "step": 27659 }, { "epoch": 1.474649464200032, "grad_norm": 2.271435260772705, "learning_rate": 4.468200000000001e-06, "loss": 0.0338, "step": 27660 }, { "epoch": 1.4747027776296848, "grad_norm": 1.618240475654602, "learning_rate": 4.468e-06, "loss": 0.0191, "step": 27661 }, { "epoch": 1.4747560910593378, "grad_norm": 1.3101089000701904, "learning_rate": 4.467800000000001e-06, "loss": 0.0175, "step": 27662 }, { "epoch": 1.4748094044889908, "grad_norm": 1.2078895568847656, "learning_rate": 4.4676e-06, "loss": 0.0115, "step": 27663 }, { "epoch": 1.4748627179186438, "grad_norm": 1.2261954545974731, "learning_rate": 4.467400000000001e-06, "loss": 0.0217, "step": 27664 }, { "epoch": 1.4749160313482967, "grad_norm": 1.515752911567688, "learning_rate": 4.4672e-06, "loss": 0.0189, "step": 27665 }, { "epoch": 1.4749693447779495, "grad_norm": 1.1323333978652954, "learning_rate": 4.467000000000001e-06, "loss": 0.0195, "step": 27666 }, { "epoch": 1.4750226582076025, "grad_norm": 3.2972660064697266, "learning_rate": 4.4668e-06, "loss": 0.0352, "step": 27667 }, { "epoch": 1.4750759716372555, "grad_norm": 0.47589558362960815, "learning_rate": 4.4666000000000006e-06, "loss": 0.0039, "step": 27668 }, { "epoch": 1.4751292850669084, "grad_norm": 1.4794719219207764, "learning_rate": 4.4664e-06, "loss": 0.0358, "step": 27669 }, { "epoch": 1.4751825984965614, "grad_norm": 1.5728543996810913, "learning_rate": 4.4662000000000005e-06, "loss": 0.0333, "step": 27670 }, { "epoch": 1.4752359119262142, "grad_norm": 1.6062567234039307, "learning_rate": 4.466e-06, "loss": 0.0259, "step": 27671 }, { "epoch": 1.4752892253558672, "grad_norm": 2.0410983562469482, "learning_rate": 4.4658000000000005e-06, "loss": 0.029, "step": 27672 }, { "epoch": 1.4753425387855201, "grad_norm": 1.7760695219039917, "learning_rate": 4.4656e-06, "loss": 0.0119, "step": 27673 }, { "epoch": 1.4753958522151729, "grad_norm": 1.4494147300720215, "learning_rate": 4.4654000000000005e-06, "loss": 0.0242, "step": 27674 }, { "epoch": 1.4754491656448259, "grad_norm": 1.6426289081573486, "learning_rate": 4.4652e-06, "loss": 0.0154, "step": 27675 }, { "epoch": 1.4755024790744788, "grad_norm": 0.7006494998931885, "learning_rate": 4.4650000000000004e-06, "loss": 0.0064, "step": 27676 }, { "epoch": 1.4755557925041318, "grad_norm": 0.9760328531265259, "learning_rate": 4.4648e-06, "loss": 0.007, "step": 27677 }, { "epoch": 1.4756091059337848, "grad_norm": 0.7478740811347961, "learning_rate": 4.4646e-06, "loss": 0.0045, "step": 27678 }, { "epoch": 1.4756624193634376, "grad_norm": 3.2104079723358154, "learning_rate": 4.4644e-06, "loss": 0.0187, "step": 27679 }, { "epoch": 1.4757157327930905, "grad_norm": 2.101548910140991, "learning_rate": 4.4642e-06, "loss": 0.0263, "step": 27680 }, { "epoch": 1.4757690462227435, "grad_norm": 3.014047384262085, "learning_rate": 4.464000000000001e-06, "loss": 0.0539, "step": 27681 }, { "epoch": 1.4758223596523965, "grad_norm": 2.8257482051849365, "learning_rate": 4.4638e-06, "loss": 0.0313, "step": 27682 }, { "epoch": 1.4758756730820495, "grad_norm": 1.2836366891860962, "learning_rate": 4.4636e-06, "loss": 0.0205, "step": 27683 }, { "epoch": 1.4759289865117022, "grad_norm": 0.9013651013374329, "learning_rate": 4.4634e-06, "loss": 0.0188, "step": 27684 }, { "epoch": 1.4759822999413552, "grad_norm": 1.1670794486999512, "learning_rate": 4.4632e-06, "loss": 0.0189, "step": 27685 }, { "epoch": 1.4760356133710082, "grad_norm": 1.081655502319336, "learning_rate": 4.463e-06, "loss": 0.0124, "step": 27686 }, { "epoch": 1.476088926800661, "grad_norm": 0.4017464220523834, "learning_rate": 4.462800000000001e-06, "loss": 0.0032, "step": 27687 }, { "epoch": 1.4761422402303142, "grad_norm": 2.1333019733428955, "learning_rate": 4.4626e-06, "loss": 0.0235, "step": 27688 }, { "epoch": 1.476195553659967, "grad_norm": 0.7706368565559387, "learning_rate": 4.462400000000001e-06, "loss": 0.01, "step": 27689 }, { "epoch": 1.4762488670896199, "grad_norm": 2.0935966968536377, "learning_rate": 4.4622e-06, "loss": 0.014, "step": 27690 }, { "epoch": 1.4763021805192729, "grad_norm": 1.7247488498687744, "learning_rate": 4.462e-06, "loss": 0.0335, "step": 27691 }, { "epoch": 1.4763554939489256, "grad_norm": 2.0580315589904785, "learning_rate": 4.4618e-06, "loss": 0.0157, "step": 27692 }, { "epoch": 1.4764088073785786, "grad_norm": 1.5132861137390137, "learning_rate": 4.461600000000001e-06, "loss": 0.011, "step": 27693 }, { "epoch": 1.4764621208082316, "grad_norm": 2.1917848587036133, "learning_rate": 4.4614e-06, "loss": 0.0142, "step": 27694 }, { "epoch": 1.4765154342378846, "grad_norm": 3.4401304721832275, "learning_rate": 4.461200000000001e-06, "loss": 0.0382, "step": 27695 }, { "epoch": 1.4765687476675375, "grad_norm": 2.256875514984131, "learning_rate": 4.461e-06, "loss": 0.0531, "step": 27696 }, { "epoch": 1.4766220610971903, "grad_norm": 1.281887173652649, "learning_rate": 4.4608e-06, "loss": 0.0247, "step": 27697 }, { "epoch": 1.4766753745268433, "grad_norm": 1.1885462999343872, "learning_rate": 4.4606e-06, "loss": 0.0226, "step": 27698 }, { "epoch": 1.4767286879564963, "grad_norm": 2.66408371925354, "learning_rate": 4.4604000000000005e-06, "loss": 0.022, "step": 27699 }, { "epoch": 1.4767820013861492, "grad_norm": 0.5903801321983337, "learning_rate": 4.460200000000001e-06, "loss": 0.0051, "step": 27700 }, { "epoch": 1.4768353148158022, "grad_norm": 2.88218355178833, "learning_rate": 4.4600000000000005e-06, "loss": 0.0289, "step": 27701 }, { "epoch": 1.476888628245455, "grad_norm": 1.6006370782852173, "learning_rate": 4.4598e-06, "loss": 0.0222, "step": 27702 }, { "epoch": 1.476941941675108, "grad_norm": 1.6946231126785278, "learning_rate": 4.4596000000000005e-06, "loss": 0.0212, "step": 27703 }, { "epoch": 1.476995255104761, "grad_norm": 2.5534961223602295, "learning_rate": 4.4594e-06, "loss": 0.0417, "step": 27704 }, { "epoch": 1.4770485685344137, "grad_norm": 1.319519281387329, "learning_rate": 4.4592000000000004e-06, "loss": 0.0109, "step": 27705 }, { "epoch": 1.4771018819640667, "grad_norm": 1.1901605129241943, "learning_rate": 4.459000000000001e-06, "loss": 0.0155, "step": 27706 }, { "epoch": 1.4771551953937196, "grad_norm": 1.9199535846710205, "learning_rate": 4.4588e-06, "loss": 0.0219, "step": 27707 }, { "epoch": 1.4772085088233726, "grad_norm": 3.0616366863250732, "learning_rate": 4.4586e-06, "loss": 0.0118, "step": 27708 }, { "epoch": 1.4772618222530256, "grad_norm": 0.6885648965835571, "learning_rate": 4.4584e-06, "loss": 0.0047, "step": 27709 }, { "epoch": 1.4773151356826784, "grad_norm": 1.3464399576187134, "learning_rate": 4.4582e-06, "loss": 0.0288, "step": 27710 }, { "epoch": 1.4773684491123313, "grad_norm": 1.8678739070892334, "learning_rate": 4.458e-06, "loss": 0.0358, "step": 27711 }, { "epoch": 1.4774217625419843, "grad_norm": 2.43554425239563, "learning_rate": 4.457800000000001e-06, "loss": 0.0206, "step": 27712 }, { "epoch": 1.4774750759716373, "grad_norm": 1.1746976375579834, "learning_rate": 4.4576e-06, "loss": 0.011, "step": 27713 }, { "epoch": 1.4775283894012903, "grad_norm": 2.442193031311035, "learning_rate": 4.457400000000001e-06, "loss": 0.0448, "step": 27714 }, { "epoch": 1.477581702830943, "grad_norm": 1.2378541231155396, "learning_rate": 4.4572e-06, "loss": 0.0105, "step": 27715 }, { "epoch": 1.477635016260596, "grad_norm": 0.986535370349884, "learning_rate": 4.457e-06, "loss": 0.0206, "step": 27716 }, { "epoch": 1.477688329690249, "grad_norm": 1.6240369081497192, "learning_rate": 4.4568e-06, "loss": 0.0096, "step": 27717 }, { "epoch": 1.477741643119902, "grad_norm": 0.9974445700645447, "learning_rate": 4.456600000000001e-06, "loss": 0.0187, "step": 27718 }, { "epoch": 1.477794956549555, "grad_norm": 1.4003185033798218, "learning_rate": 4.4564e-06, "loss": 0.0188, "step": 27719 }, { "epoch": 1.4778482699792077, "grad_norm": 1.1825754642486572, "learning_rate": 4.456200000000001e-06, "loss": 0.0122, "step": 27720 }, { "epoch": 1.4779015834088607, "grad_norm": 1.3819011449813843, "learning_rate": 4.456e-06, "loss": 0.0271, "step": 27721 }, { "epoch": 1.4779548968385137, "grad_norm": 2.2068889141082764, "learning_rate": 4.4558e-06, "loss": 0.0236, "step": 27722 }, { "epoch": 1.4780082102681664, "grad_norm": 1.089618444442749, "learning_rate": 4.4556e-06, "loss": 0.0143, "step": 27723 }, { "epoch": 1.4780615236978194, "grad_norm": 0.980655312538147, "learning_rate": 4.4554000000000006e-06, "loss": 0.0105, "step": 27724 }, { "epoch": 1.4781148371274724, "grad_norm": 1.8146007061004639, "learning_rate": 4.455200000000001e-06, "loss": 0.0334, "step": 27725 }, { "epoch": 1.4781681505571254, "grad_norm": 1.4338401556015015, "learning_rate": 4.4550000000000005e-06, "loss": 0.025, "step": 27726 }, { "epoch": 1.4782214639867783, "grad_norm": 1.8862000703811646, "learning_rate": 4.4548e-06, "loss": 0.029, "step": 27727 }, { "epoch": 1.478274777416431, "grad_norm": 1.6261191368103027, "learning_rate": 4.4546000000000005e-06, "loss": 0.0219, "step": 27728 }, { "epoch": 1.478328090846084, "grad_norm": 0.9669682383537292, "learning_rate": 4.4544e-06, "loss": 0.0082, "step": 27729 }, { "epoch": 1.478381404275737, "grad_norm": 1.598747968673706, "learning_rate": 4.4542000000000005e-06, "loss": 0.0266, "step": 27730 }, { "epoch": 1.47843471770539, "grad_norm": 1.8463870286941528, "learning_rate": 4.454000000000001e-06, "loss": 0.0265, "step": 27731 }, { "epoch": 1.478488031135043, "grad_norm": 0.20638269186019897, "learning_rate": 4.4538000000000005e-06, "loss": 0.0024, "step": 27732 }, { "epoch": 1.4785413445646958, "grad_norm": 1.3872718811035156, "learning_rate": 4.4536e-06, "loss": 0.0207, "step": 27733 }, { "epoch": 1.4785946579943487, "grad_norm": 1.2814226150512695, "learning_rate": 4.4534000000000004e-06, "loss": 0.0132, "step": 27734 }, { "epoch": 1.4786479714240017, "grad_norm": 1.385905146598816, "learning_rate": 4.4532e-06, "loss": 0.0233, "step": 27735 }, { "epoch": 1.4787012848536547, "grad_norm": 1.4791756868362427, "learning_rate": 4.453e-06, "loss": 0.0122, "step": 27736 }, { "epoch": 1.4787545982833077, "grad_norm": 0.8803401589393616, "learning_rate": 4.452800000000001e-06, "loss": 0.0086, "step": 27737 }, { "epoch": 1.4788079117129604, "grad_norm": 1.5194032192230225, "learning_rate": 4.4526e-06, "loss": 0.0097, "step": 27738 }, { "epoch": 1.4788612251426134, "grad_norm": 1.8037062883377075, "learning_rate": 4.452400000000001e-06, "loss": 0.0166, "step": 27739 }, { "epoch": 1.4789145385722664, "grad_norm": 1.616859793663025, "learning_rate": 4.4522e-06, "loss": 0.0125, "step": 27740 }, { "epoch": 1.4789678520019192, "grad_norm": 1.3114696741104126, "learning_rate": 4.452e-06, "loss": 0.0296, "step": 27741 }, { "epoch": 1.4790211654315721, "grad_norm": 3.1110892295837402, "learning_rate": 4.4518e-06, "loss": 0.0129, "step": 27742 }, { "epoch": 1.4790744788612251, "grad_norm": 1.2809562683105469, "learning_rate": 4.451600000000001e-06, "loss": 0.0092, "step": 27743 }, { "epoch": 1.479127792290878, "grad_norm": 0.7508512139320374, "learning_rate": 4.4514e-06, "loss": 0.0054, "step": 27744 }, { "epoch": 1.479181105720531, "grad_norm": 3.6350367069244385, "learning_rate": 4.451200000000001e-06, "loss": 0.0648, "step": 27745 }, { "epoch": 1.4792344191501838, "grad_norm": 0.9897677302360535, "learning_rate": 4.451e-06, "loss": 0.009, "step": 27746 }, { "epoch": 1.4792877325798368, "grad_norm": 1.2173686027526855, "learning_rate": 4.4508e-06, "loss": 0.0201, "step": 27747 }, { "epoch": 1.4793410460094898, "grad_norm": 2.3463287353515625, "learning_rate": 4.4506e-06, "loss": 0.0256, "step": 27748 }, { "epoch": 1.4793943594391428, "grad_norm": 1.2956669330596924, "learning_rate": 4.450400000000001e-06, "loss": 0.0129, "step": 27749 }, { "epoch": 1.4794476728687957, "grad_norm": 1.391347885131836, "learning_rate": 4.450200000000001e-06, "loss": 0.0258, "step": 27750 }, { "epoch": 1.4795009862984485, "grad_norm": 1.0317660570144653, "learning_rate": 4.450000000000001e-06, "loss": 0.0117, "step": 27751 }, { "epoch": 1.4795542997281015, "grad_norm": 1.032868504524231, "learning_rate": 4.4498e-06, "loss": 0.0071, "step": 27752 }, { "epoch": 1.4796076131577545, "grad_norm": 1.1395695209503174, "learning_rate": 4.4496000000000006e-06, "loss": 0.0121, "step": 27753 }, { "epoch": 1.4796609265874074, "grad_norm": 1.865777850151062, "learning_rate": 4.4494e-06, "loss": 0.0369, "step": 27754 }, { "epoch": 1.4797142400170604, "grad_norm": 2.6757991313934326, "learning_rate": 4.4492000000000005e-06, "loss": 0.0193, "step": 27755 }, { "epoch": 1.4797675534467132, "grad_norm": 1.0066096782684326, "learning_rate": 4.449000000000001e-06, "loss": 0.0084, "step": 27756 }, { "epoch": 1.4798208668763662, "grad_norm": 1.6548595428466797, "learning_rate": 4.4488000000000005e-06, "loss": 0.032, "step": 27757 }, { "epoch": 1.4798741803060191, "grad_norm": 2.4227638244628906, "learning_rate": 4.4486e-06, "loss": 0.0369, "step": 27758 }, { "epoch": 1.479927493735672, "grad_norm": 1.328005313873291, "learning_rate": 4.4484000000000005e-06, "loss": 0.0215, "step": 27759 }, { "epoch": 1.4799808071653249, "grad_norm": 1.585056185722351, "learning_rate": 4.4482e-06, "loss": 0.018, "step": 27760 }, { "epoch": 1.4800341205949779, "grad_norm": 0.5841772556304932, "learning_rate": 4.4480000000000004e-06, "loss": 0.0056, "step": 27761 }, { "epoch": 1.4800874340246308, "grad_norm": 2.27599835395813, "learning_rate": 4.447800000000001e-06, "loss": 0.0289, "step": 27762 }, { "epoch": 1.4801407474542838, "grad_norm": 1.735996961593628, "learning_rate": 4.4476e-06, "loss": 0.0127, "step": 27763 }, { "epoch": 1.4801940608839366, "grad_norm": 0.5267900228500366, "learning_rate": 4.447400000000001e-06, "loss": 0.0043, "step": 27764 }, { "epoch": 1.4802473743135895, "grad_norm": 0.5811172127723694, "learning_rate": 4.4472e-06, "loss": 0.0037, "step": 27765 }, { "epoch": 1.4803006877432425, "grad_norm": 3.665165901184082, "learning_rate": 4.447e-06, "loss": 0.0276, "step": 27766 }, { "epoch": 1.4803540011728955, "grad_norm": 2.205895185470581, "learning_rate": 4.4468e-06, "loss": 0.0337, "step": 27767 }, { "epoch": 1.4804073146025485, "grad_norm": 2.6297011375427246, "learning_rate": 4.446600000000001e-06, "loss": 0.0191, "step": 27768 }, { "epoch": 1.4804606280322012, "grad_norm": 2.1662356853485107, "learning_rate": 4.4464e-06, "loss": 0.0247, "step": 27769 }, { "epoch": 1.4805139414618542, "grad_norm": 1.6791244745254517, "learning_rate": 4.446200000000001e-06, "loss": 0.0111, "step": 27770 }, { "epoch": 1.4805672548915072, "grad_norm": 3.199981689453125, "learning_rate": 4.446e-06, "loss": 0.0688, "step": 27771 }, { "epoch": 1.4806205683211602, "grad_norm": 2.1549081802368164, "learning_rate": 4.4458e-06, "loss": 0.0425, "step": 27772 }, { "epoch": 1.4806738817508132, "grad_norm": 1.2968056201934814, "learning_rate": 4.4456e-06, "loss": 0.0058, "step": 27773 }, { "epoch": 1.480727195180466, "grad_norm": 0.9140135049819946, "learning_rate": 4.445400000000001e-06, "loss": 0.0155, "step": 27774 }, { "epoch": 1.480780508610119, "grad_norm": 1.1143625974655151, "learning_rate": 4.4452e-06, "loss": 0.022, "step": 27775 }, { "epoch": 1.4808338220397719, "grad_norm": 0.4987565279006958, "learning_rate": 4.445000000000001e-06, "loss": 0.0047, "step": 27776 }, { "epoch": 1.4808871354694246, "grad_norm": 1.621717095375061, "learning_rate": 4.4448e-06, "loss": 0.0443, "step": 27777 }, { "epoch": 1.4809404488990776, "grad_norm": 1.69040048122406, "learning_rate": 4.444600000000001e-06, "loss": 0.0225, "step": 27778 }, { "epoch": 1.4809937623287306, "grad_norm": 1.7481200695037842, "learning_rate": 4.4444e-06, "loss": 0.014, "step": 27779 }, { "epoch": 1.4810470757583836, "grad_norm": 1.487210988998413, "learning_rate": 4.444200000000001e-06, "loss": 0.0303, "step": 27780 }, { "epoch": 1.4811003891880365, "grad_norm": 1.7113245725631714, "learning_rate": 4.444e-06, "loss": 0.0103, "step": 27781 }, { "epoch": 1.4811537026176893, "grad_norm": 1.5726031064987183, "learning_rate": 4.4438000000000006e-06, "loss": 0.028, "step": 27782 }, { "epoch": 1.4812070160473423, "grad_norm": 1.633680820465088, "learning_rate": 4.4436e-06, "loss": 0.0332, "step": 27783 }, { "epoch": 1.4812603294769953, "grad_norm": 0.8322984576225281, "learning_rate": 4.4434000000000005e-06, "loss": 0.009, "step": 27784 }, { "epoch": 1.4813136429066482, "grad_norm": 0.7154317498207092, "learning_rate": 4.4432e-06, "loss": 0.0053, "step": 27785 }, { "epoch": 1.4813669563363012, "grad_norm": 1.6907289028167725, "learning_rate": 4.4430000000000005e-06, "loss": 0.0291, "step": 27786 }, { "epoch": 1.481420269765954, "grad_norm": 1.5243239402770996, "learning_rate": 4.4428e-06, "loss": 0.0083, "step": 27787 }, { "epoch": 1.481473583195607, "grad_norm": 1.9596949815750122, "learning_rate": 4.4426000000000005e-06, "loss": 0.0185, "step": 27788 }, { "epoch": 1.48152689662526, "grad_norm": 2.08035945892334, "learning_rate": 4.442400000000001e-06, "loss": 0.0206, "step": 27789 }, { "epoch": 1.4815802100549127, "grad_norm": 2.7907235622406006, "learning_rate": 4.4422000000000004e-06, "loss": 0.0154, "step": 27790 }, { "epoch": 1.481633523484566, "grad_norm": 2.303072214126587, "learning_rate": 4.442e-06, "loss": 0.0207, "step": 27791 }, { "epoch": 1.4816868369142187, "grad_norm": 1.4268051385879517, "learning_rate": 4.4418e-06, "loss": 0.0176, "step": 27792 }, { "epoch": 1.4817401503438716, "grad_norm": 0.870439350605011, "learning_rate": 4.4416e-06, "loss": 0.0098, "step": 27793 }, { "epoch": 1.4817934637735246, "grad_norm": 1.7473787069320679, "learning_rate": 4.4414e-06, "loss": 0.0265, "step": 27794 }, { "epoch": 1.4818467772031774, "grad_norm": 1.5632238388061523, "learning_rate": 4.441200000000001e-06, "loss": 0.0217, "step": 27795 }, { "epoch": 1.4819000906328303, "grad_norm": 1.6626945734024048, "learning_rate": 4.441e-06, "loss": 0.0221, "step": 27796 }, { "epoch": 1.4819534040624833, "grad_norm": 2.074760675430298, "learning_rate": 4.4408e-06, "loss": 0.0186, "step": 27797 }, { "epoch": 1.4820067174921363, "grad_norm": 3.0133402347564697, "learning_rate": 4.4406e-06, "loss": 0.011, "step": 27798 }, { "epoch": 1.4820600309217893, "grad_norm": 2.02005934715271, "learning_rate": 4.4404e-06, "loss": 0.0253, "step": 27799 }, { "epoch": 1.482113344351442, "grad_norm": 3.697944402694702, "learning_rate": 4.4402e-06, "loss": 0.0293, "step": 27800 }, { "epoch": 1.482166657781095, "grad_norm": 1.7433425188064575, "learning_rate": 4.440000000000001e-06, "loss": 0.0225, "step": 27801 }, { "epoch": 1.482219971210748, "grad_norm": 1.7451690435409546, "learning_rate": 4.4398e-06, "loss": 0.0251, "step": 27802 }, { "epoch": 1.482273284640401, "grad_norm": 1.1227113008499146, "learning_rate": 4.439600000000001e-06, "loss": 0.0169, "step": 27803 }, { "epoch": 1.482326598070054, "grad_norm": 2.1934547424316406, "learning_rate": 4.4394e-06, "loss": 0.0306, "step": 27804 }, { "epoch": 1.4823799114997067, "grad_norm": 0.8473060727119446, "learning_rate": 4.4392e-06, "loss": 0.0052, "step": 27805 }, { "epoch": 1.4824332249293597, "grad_norm": 4.005594253540039, "learning_rate": 4.439e-06, "loss": 0.0839, "step": 27806 }, { "epoch": 1.4824865383590127, "grad_norm": 2.8727471828460693, "learning_rate": 4.438800000000001e-06, "loss": 0.054, "step": 27807 }, { "epoch": 1.4825398517886654, "grad_norm": 1.740970253944397, "learning_rate": 4.4386e-06, "loss": 0.0308, "step": 27808 }, { "epoch": 1.4825931652183186, "grad_norm": 1.7615896463394165, "learning_rate": 4.4384000000000006e-06, "loss": 0.0164, "step": 27809 }, { "epoch": 1.4826464786479714, "grad_norm": 0.8578489422798157, "learning_rate": 4.4382e-06, "loss": 0.0062, "step": 27810 }, { "epoch": 1.4826997920776244, "grad_norm": 1.7324965000152588, "learning_rate": 4.438e-06, "loss": 0.0232, "step": 27811 }, { "epoch": 1.4827531055072773, "grad_norm": 0.5495540499687195, "learning_rate": 4.4378e-06, "loss": 0.0059, "step": 27812 }, { "epoch": 1.48280641893693, "grad_norm": 1.4636385440826416, "learning_rate": 4.4376000000000005e-06, "loss": 0.024, "step": 27813 }, { "epoch": 1.482859732366583, "grad_norm": 2.5896456241607666, "learning_rate": 4.437400000000001e-06, "loss": 0.0285, "step": 27814 }, { "epoch": 1.482913045796236, "grad_norm": 1.4859238862991333, "learning_rate": 4.4372000000000005e-06, "loss": 0.0093, "step": 27815 }, { "epoch": 1.482966359225889, "grad_norm": 2.433995008468628, "learning_rate": 4.437e-06, "loss": 0.0145, "step": 27816 }, { "epoch": 1.483019672655542, "grad_norm": 0.715212345123291, "learning_rate": 4.4368000000000004e-06, "loss": 0.0059, "step": 27817 }, { "epoch": 1.4830729860851948, "grad_norm": 2.09454607963562, "learning_rate": 4.4366e-06, "loss": 0.0272, "step": 27818 }, { "epoch": 1.4831262995148478, "grad_norm": 0.7415609359741211, "learning_rate": 4.4364e-06, "loss": 0.0169, "step": 27819 }, { "epoch": 1.4831796129445007, "grad_norm": 2.0644679069519043, "learning_rate": 4.436200000000001e-06, "loss": 0.0221, "step": 27820 }, { "epoch": 1.4832329263741537, "grad_norm": 2.335993766784668, "learning_rate": 4.436e-06, "loss": 0.0218, "step": 27821 }, { "epoch": 1.4832862398038067, "grad_norm": 2.1400327682495117, "learning_rate": 4.4358e-06, "loss": 0.0288, "step": 27822 }, { "epoch": 1.4833395532334595, "grad_norm": 1.5884928703308105, "learning_rate": 4.4356e-06, "loss": 0.0224, "step": 27823 }, { "epoch": 1.4833928666631124, "grad_norm": 2.2839725017547607, "learning_rate": 4.4354e-06, "loss": 0.0183, "step": 27824 }, { "epoch": 1.4834461800927654, "grad_norm": 0.9376510381698608, "learning_rate": 4.4352e-06, "loss": 0.0105, "step": 27825 }, { "epoch": 1.4834994935224182, "grad_norm": 1.1980620622634888, "learning_rate": 4.435000000000001e-06, "loss": 0.0175, "step": 27826 }, { "epoch": 1.4835528069520711, "grad_norm": 1.8188570737838745, "learning_rate": 4.4348e-06, "loss": 0.0235, "step": 27827 }, { "epoch": 1.4836061203817241, "grad_norm": 2.218402624130249, "learning_rate": 4.434600000000001e-06, "loss": 0.0218, "step": 27828 }, { "epoch": 1.483659433811377, "grad_norm": 0.990602433681488, "learning_rate": 4.4344e-06, "loss": 0.0117, "step": 27829 }, { "epoch": 1.48371274724103, "grad_norm": 3.648656129837036, "learning_rate": 4.4342e-06, "loss": 0.0385, "step": 27830 }, { "epoch": 1.4837660606706828, "grad_norm": 1.8634451627731323, "learning_rate": 4.434e-06, "loss": 0.0368, "step": 27831 }, { "epoch": 1.4838193741003358, "grad_norm": 2.5439765453338623, "learning_rate": 4.433800000000001e-06, "loss": 0.0485, "step": 27832 }, { "epoch": 1.4838726875299888, "grad_norm": 0.5934998989105225, "learning_rate": 4.4336e-06, "loss": 0.0052, "step": 27833 }, { "epoch": 1.4839260009596418, "grad_norm": 2.9778401851654053, "learning_rate": 4.433400000000001e-06, "loss": 0.0305, "step": 27834 }, { "epoch": 1.4839793143892948, "grad_norm": 2.1370303630828857, "learning_rate": 4.4332e-06, "loss": 0.0189, "step": 27835 }, { "epoch": 1.4840326278189475, "grad_norm": 1.794332504272461, "learning_rate": 4.433000000000001e-06, "loss": 0.0343, "step": 27836 }, { "epoch": 1.4840859412486005, "grad_norm": 2.406526803970337, "learning_rate": 4.4328e-06, "loss": 0.0295, "step": 27837 }, { "epoch": 1.4841392546782535, "grad_norm": 1.2823822498321533, "learning_rate": 4.4326000000000006e-06, "loss": 0.0073, "step": 27838 }, { "epoch": 1.4841925681079065, "grad_norm": 1.5829473733901978, "learning_rate": 4.432400000000001e-06, "loss": 0.0164, "step": 27839 }, { "epoch": 1.4842458815375594, "grad_norm": 0.48390209674835205, "learning_rate": 4.4322000000000005e-06, "loss": 0.005, "step": 27840 }, { "epoch": 1.4842991949672122, "grad_norm": 2.2690517902374268, "learning_rate": 4.432e-06, "loss": 0.0543, "step": 27841 }, { "epoch": 1.4843525083968652, "grad_norm": 1.461458683013916, "learning_rate": 4.4318000000000005e-06, "loss": 0.0258, "step": 27842 }, { "epoch": 1.4844058218265181, "grad_norm": 1.877748727798462, "learning_rate": 4.4316e-06, "loss": 0.0273, "step": 27843 }, { "epoch": 1.484459135256171, "grad_norm": 1.3398321866989136, "learning_rate": 4.4314000000000005e-06, "loss": 0.0077, "step": 27844 }, { "epoch": 1.4845124486858239, "grad_norm": 1.521944284439087, "learning_rate": 4.431200000000001e-06, "loss": 0.0409, "step": 27845 }, { "epoch": 1.4845657621154769, "grad_norm": 2.478233814239502, "learning_rate": 4.4310000000000004e-06, "loss": 0.0517, "step": 27846 }, { "epoch": 1.4846190755451298, "grad_norm": 3.4227213859558105, "learning_rate": 4.4308e-06, "loss": 0.0227, "step": 27847 }, { "epoch": 1.4846723889747828, "grad_norm": 1.850514531135559, "learning_rate": 4.4306e-06, "loss": 0.0183, "step": 27848 }, { "epoch": 1.4847257024044356, "grad_norm": 1.5553076267242432, "learning_rate": 4.4304e-06, "loss": 0.0134, "step": 27849 }, { "epoch": 1.4847790158340886, "grad_norm": 1.3632347583770752, "learning_rate": 4.4302e-06, "loss": 0.0167, "step": 27850 }, { "epoch": 1.4848323292637415, "grad_norm": 0.7272554039955139, "learning_rate": 4.430000000000001e-06, "loss": 0.0059, "step": 27851 }, { "epoch": 1.4848856426933945, "grad_norm": 1.0088601112365723, "learning_rate": 4.4298e-06, "loss": 0.0106, "step": 27852 }, { "epoch": 1.4849389561230475, "grad_norm": 1.8062387704849243, "learning_rate": 4.429600000000001e-06, "loss": 0.019, "step": 27853 }, { "epoch": 1.4849922695527003, "grad_norm": 2.253070592880249, "learning_rate": 4.4294e-06, "loss": 0.039, "step": 27854 }, { "epoch": 1.4850455829823532, "grad_norm": 1.8642737865447998, "learning_rate": 4.4292e-06, "loss": 0.0197, "step": 27855 }, { "epoch": 1.4850988964120062, "grad_norm": 2.163943290710449, "learning_rate": 4.429e-06, "loss": 0.0246, "step": 27856 }, { "epoch": 1.4851522098416592, "grad_norm": 2.2395732402801514, "learning_rate": 4.428800000000001e-06, "loss": 0.0061, "step": 27857 }, { "epoch": 1.4852055232713122, "grad_norm": 3.68076491355896, "learning_rate": 4.4286e-06, "loss": 0.034, "step": 27858 }, { "epoch": 1.485258836700965, "grad_norm": 2.046241283416748, "learning_rate": 4.428400000000001e-06, "loss": 0.0308, "step": 27859 }, { "epoch": 1.485312150130618, "grad_norm": 2.4188950061798096, "learning_rate": 4.4282e-06, "loss": 0.0342, "step": 27860 }, { "epoch": 1.4853654635602709, "grad_norm": 3.5053813457489014, "learning_rate": 4.428000000000001e-06, "loss": 0.0262, "step": 27861 }, { "epoch": 1.4854187769899236, "grad_norm": 2.1749520301818848, "learning_rate": 4.4278e-06, "loss": 0.034, "step": 27862 }, { "epoch": 1.4854720904195766, "grad_norm": 2.495471239089966, "learning_rate": 4.427600000000001e-06, "loss": 0.0368, "step": 27863 }, { "epoch": 1.4855254038492296, "grad_norm": 1.3429540395736694, "learning_rate": 4.4274e-06, "loss": 0.013, "step": 27864 }, { "epoch": 1.4855787172788826, "grad_norm": 3.6066410541534424, "learning_rate": 4.4272000000000006e-06, "loss": 0.0359, "step": 27865 }, { "epoch": 1.4856320307085356, "grad_norm": 1.2926896810531616, "learning_rate": 4.427e-06, "loss": 0.0078, "step": 27866 }, { "epoch": 1.4856853441381883, "grad_norm": 1.2065541744232178, "learning_rate": 4.4268000000000005e-06, "loss": 0.0234, "step": 27867 }, { "epoch": 1.4857386575678413, "grad_norm": 0.600413978099823, "learning_rate": 4.4266e-06, "loss": 0.0054, "step": 27868 }, { "epoch": 1.4857919709974943, "grad_norm": 0.9711365699768066, "learning_rate": 4.4264000000000005e-06, "loss": 0.0129, "step": 27869 }, { "epoch": 1.4858452844271473, "grad_norm": 1.6877224445343018, "learning_rate": 4.4262e-06, "loss": 0.0286, "step": 27870 }, { "epoch": 1.4858985978568002, "grad_norm": 1.8333756923675537, "learning_rate": 4.4260000000000005e-06, "loss": 0.0326, "step": 27871 }, { "epoch": 1.485951911286453, "grad_norm": 1.0876734256744385, "learning_rate": 4.4258e-06, "loss": 0.0166, "step": 27872 }, { "epoch": 1.486005224716106, "grad_norm": 0.6117380857467651, "learning_rate": 4.4256000000000005e-06, "loss": 0.0061, "step": 27873 }, { "epoch": 1.486058538145759, "grad_norm": 0.9998865723609924, "learning_rate": 4.4254e-06, "loss": 0.0136, "step": 27874 }, { "epoch": 1.486111851575412, "grad_norm": 1.4639577865600586, "learning_rate": 4.4252000000000004e-06, "loss": 0.0346, "step": 27875 }, { "epoch": 1.486165165005065, "grad_norm": 1.8439699411392212, "learning_rate": 4.425e-06, "loss": 0.0409, "step": 27876 }, { "epoch": 1.4862184784347177, "grad_norm": 1.2787744998931885, "learning_rate": 4.4248e-06, "loss": 0.0241, "step": 27877 }, { "epoch": 1.4862717918643706, "grad_norm": 3.551041603088379, "learning_rate": 4.424600000000001e-06, "loss": 0.0273, "step": 27878 }, { "epoch": 1.4863251052940236, "grad_norm": 0.8069705367088318, "learning_rate": 4.4244e-06, "loss": 0.0074, "step": 27879 }, { "epoch": 1.4863784187236764, "grad_norm": 1.7317243814468384, "learning_rate": 4.4242e-06, "loss": 0.0211, "step": 27880 }, { "epoch": 1.4864317321533294, "grad_norm": 1.4185558557510376, "learning_rate": 4.424e-06, "loss": 0.0233, "step": 27881 }, { "epoch": 1.4864850455829823, "grad_norm": 1.472139596939087, "learning_rate": 4.4238e-06, "loss": 0.0183, "step": 27882 }, { "epoch": 1.4865383590126353, "grad_norm": 2.6434433460235596, "learning_rate": 4.4236e-06, "loss": 0.0407, "step": 27883 }, { "epoch": 1.4865916724422883, "grad_norm": 2.1720499992370605, "learning_rate": 4.423400000000001e-06, "loss": 0.033, "step": 27884 }, { "epoch": 1.486644985871941, "grad_norm": 1.829032301902771, "learning_rate": 4.4232e-06, "loss": 0.0259, "step": 27885 }, { "epoch": 1.486698299301594, "grad_norm": 0.8005388975143433, "learning_rate": 4.423000000000001e-06, "loss": 0.0073, "step": 27886 }, { "epoch": 1.486751612731247, "grad_norm": 1.183526635169983, "learning_rate": 4.4228e-06, "loss": 0.0119, "step": 27887 }, { "epoch": 1.4868049261609, "grad_norm": 2.0180087089538574, "learning_rate": 4.4226e-06, "loss": 0.0238, "step": 27888 }, { "epoch": 1.486858239590553, "grad_norm": 2.5315375328063965, "learning_rate": 4.4224e-06, "loss": 0.0321, "step": 27889 }, { "epoch": 1.4869115530202057, "grad_norm": 2.4400711059570312, "learning_rate": 4.422200000000001e-06, "loss": 0.0218, "step": 27890 }, { "epoch": 1.4869648664498587, "grad_norm": 2.341172933578491, "learning_rate": 4.422e-06, "loss": 0.03, "step": 27891 }, { "epoch": 1.4870181798795117, "grad_norm": 1.7612792253494263, "learning_rate": 4.421800000000001e-06, "loss": 0.0208, "step": 27892 }, { "epoch": 1.4870714933091647, "grad_norm": 2.812248468399048, "learning_rate": 4.4216e-06, "loss": 0.0194, "step": 27893 }, { "epoch": 1.4871248067388176, "grad_norm": 1.4760112762451172, "learning_rate": 4.4214e-06, "loss": 0.0108, "step": 27894 }, { "epoch": 1.4871781201684704, "grad_norm": 1.0832833051681519, "learning_rate": 4.4212e-06, "loss": 0.0081, "step": 27895 }, { "epoch": 1.4872314335981234, "grad_norm": 3.2341551780700684, "learning_rate": 4.4210000000000005e-06, "loss": 0.0269, "step": 27896 }, { "epoch": 1.4872847470277764, "grad_norm": 1.2714967727661133, "learning_rate": 4.4208e-06, "loss": 0.013, "step": 27897 }, { "epoch": 1.4873380604574291, "grad_norm": 3.0347347259521484, "learning_rate": 4.4206000000000005e-06, "loss": 0.0391, "step": 27898 }, { "epoch": 1.487391373887082, "grad_norm": 0.9059211015701294, "learning_rate": 4.4204e-06, "loss": 0.0093, "step": 27899 }, { "epoch": 1.487444687316735, "grad_norm": 1.3252418041229248, "learning_rate": 4.4202000000000005e-06, "loss": 0.0173, "step": 27900 }, { "epoch": 1.487498000746388, "grad_norm": 1.5094237327575684, "learning_rate": 4.42e-06, "loss": 0.0252, "step": 27901 }, { "epoch": 1.487551314176041, "grad_norm": 1.4201663732528687, "learning_rate": 4.4198000000000004e-06, "loss": 0.0114, "step": 27902 }, { "epoch": 1.4876046276056938, "grad_norm": 1.395500659942627, "learning_rate": 4.419600000000001e-06, "loss": 0.0208, "step": 27903 }, { "epoch": 1.4876579410353468, "grad_norm": 1.3148850202560425, "learning_rate": 4.4194e-06, "loss": 0.0131, "step": 27904 }, { "epoch": 1.4877112544649997, "grad_norm": 1.444239854812622, "learning_rate": 4.4192e-06, "loss": 0.0448, "step": 27905 }, { "epoch": 1.4877645678946527, "grad_norm": 1.2594560384750366, "learning_rate": 4.419e-06, "loss": 0.0064, "step": 27906 }, { "epoch": 1.4878178813243057, "grad_norm": 2.2956278324127197, "learning_rate": 4.4188e-06, "loss": 0.0206, "step": 27907 }, { "epoch": 1.4878711947539585, "grad_norm": 1.7275748252868652, "learning_rate": 4.4186e-06, "loss": 0.0303, "step": 27908 }, { "epoch": 1.4879245081836114, "grad_norm": 1.8753050565719604, "learning_rate": 4.418400000000001e-06, "loss": 0.044, "step": 27909 }, { "epoch": 1.4879778216132644, "grad_norm": 0.8652607798576355, "learning_rate": 4.4182e-06, "loss": 0.0072, "step": 27910 }, { "epoch": 1.4880311350429172, "grad_norm": 2.9834485054016113, "learning_rate": 4.418000000000001e-06, "loss": 0.0286, "step": 27911 }, { "epoch": 1.4880844484725704, "grad_norm": 0.9783639311790466, "learning_rate": 4.4178e-06, "loss": 0.011, "step": 27912 }, { "epoch": 1.4881377619022231, "grad_norm": 1.8222379684448242, "learning_rate": 4.4176e-06, "loss": 0.0397, "step": 27913 }, { "epoch": 1.4881910753318761, "grad_norm": 1.5260716676712036, "learning_rate": 4.4174e-06, "loss": 0.0252, "step": 27914 }, { "epoch": 1.488244388761529, "grad_norm": 0.6057001948356628, "learning_rate": 4.417200000000001e-06, "loss": 0.0062, "step": 27915 }, { "epoch": 1.4882977021911818, "grad_norm": 2.3023786544799805, "learning_rate": 4.417e-06, "loss": 0.0246, "step": 27916 }, { "epoch": 1.4883510156208348, "grad_norm": 2.48724365234375, "learning_rate": 4.416800000000001e-06, "loss": 0.032, "step": 27917 }, { "epoch": 1.4884043290504878, "grad_norm": 2.088897466659546, "learning_rate": 4.4166e-06, "loss": 0.0312, "step": 27918 }, { "epoch": 1.4884576424801408, "grad_norm": 1.7651121616363525, "learning_rate": 4.4164e-06, "loss": 0.0194, "step": 27919 }, { "epoch": 1.4885109559097938, "grad_norm": 2.071152925491333, "learning_rate": 4.4162e-06, "loss": 0.0341, "step": 27920 }, { "epoch": 1.4885642693394465, "grad_norm": 1.4410333633422852, "learning_rate": 4.416000000000001e-06, "loss": 0.0262, "step": 27921 }, { "epoch": 1.4886175827690995, "grad_norm": 0.9913375973701477, "learning_rate": 4.4158e-06, "loss": 0.0095, "step": 27922 }, { "epoch": 1.4886708961987525, "grad_norm": 2.0802180767059326, "learning_rate": 4.4156000000000005e-06, "loss": 0.0279, "step": 27923 }, { "epoch": 1.4887242096284055, "grad_norm": 1.118270993232727, "learning_rate": 4.4154e-06, "loss": 0.017, "step": 27924 }, { "epoch": 1.4887775230580584, "grad_norm": 1.8032536506652832, "learning_rate": 4.4152000000000005e-06, "loss": 0.0281, "step": 27925 }, { "epoch": 1.4888308364877112, "grad_norm": 1.8911761045455933, "learning_rate": 4.415e-06, "loss": 0.015, "step": 27926 }, { "epoch": 1.4888841499173642, "grad_norm": 3.918342351913452, "learning_rate": 4.4148000000000005e-06, "loss": 0.0158, "step": 27927 }, { "epoch": 1.4889374633470172, "grad_norm": 1.353421688079834, "learning_rate": 4.414600000000001e-06, "loss": 0.0126, "step": 27928 }, { "epoch": 1.48899077677667, "grad_norm": 1.0479763746261597, "learning_rate": 4.4144000000000005e-06, "loss": 0.0083, "step": 27929 }, { "epoch": 1.489044090206323, "grad_norm": 1.8593066930770874, "learning_rate": 4.4142e-06, "loss": 0.0201, "step": 27930 }, { "epoch": 1.4890974036359759, "grad_norm": 1.6877623796463013, "learning_rate": 4.4140000000000004e-06, "loss": 0.0391, "step": 27931 }, { "epoch": 1.4891507170656288, "grad_norm": 1.7869292497634888, "learning_rate": 4.4138e-06, "loss": 0.0207, "step": 27932 }, { "epoch": 1.4892040304952818, "grad_norm": 2.5780820846557617, "learning_rate": 4.4136e-06, "loss": 0.0284, "step": 27933 }, { "epoch": 1.4892573439249346, "grad_norm": 1.4964741468429565, "learning_rate": 4.413400000000001e-06, "loss": 0.0158, "step": 27934 }, { "epoch": 1.4893106573545876, "grad_norm": 2.017439126968384, "learning_rate": 4.4132e-06, "loss": 0.0097, "step": 27935 }, { "epoch": 1.4893639707842405, "grad_norm": 0.7360092401504517, "learning_rate": 4.413000000000001e-06, "loss": 0.0063, "step": 27936 }, { "epoch": 1.4894172842138935, "grad_norm": 0.9778534173965454, "learning_rate": 4.4128e-06, "loss": 0.0127, "step": 27937 }, { "epoch": 1.4894705976435465, "grad_norm": 2.403261423110962, "learning_rate": 4.4126e-06, "loss": 0.0297, "step": 27938 }, { "epoch": 1.4895239110731993, "grad_norm": 2.3917975425720215, "learning_rate": 4.4124e-06, "loss": 0.0324, "step": 27939 }, { "epoch": 1.4895772245028522, "grad_norm": 6.978355407714844, "learning_rate": 4.412200000000001e-06, "loss": 0.0479, "step": 27940 }, { "epoch": 1.4896305379325052, "grad_norm": 1.6258732080459595, "learning_rate": 4.412e-06, "loss": 0.0134, "step": 27941 }, { "epoch": 1.4896838513621582, "grad_norm": 0.9020527005195618, "learning_rate": 4.411800000000001e-06, "loss": 0.0093, "step": 27942 }, { "epoch": 1.4897371647918112, "grad_norm": 1.882101058959961, "learning_rate": 4.4116e-06, "loss": 0.0165, "step": 27943 }, { "epoch": 1.489790478221464, "grad_norm": 1.2632954120635986, "learning_rate": 4.4114e-06, "loss": 0.0109, "step": 27944 }, { "epoch": 1.489843791651117, "grad_norm": 2.255265951156616, "learning_rate": 4.4112e-06, "loss": 0.0307, "step": 27945 }, { "epoch": 1.48989710508077, "grad_norm": 3.2900376319885254, "learning_rate": 4.411000000000001e-06, "loss": 0.0399, "step": 27946 }, { "epoch": 1.4899504185104226, "grad_norm": 1.3220547437667847, "learning_rate": 4.410800000000001e-06, "loss": 0.0187, "step": 27947 }, { "epoch": 1.4900037319400756, "grad_norm": 1.1791861057281494, "learning_rate": 4.410600000000001e-06, "loss": 0.0083, "step": 27948 }, { "epoch": 1.4900570453697286, "grad_norm": 1.8968793153762817, "learning_rate": 4.4104e-06, "loss": 0.0252, "step": 27949 }, { "epoch": 1.4901103587993816, "grad_norm": 2.0644428730010986, "learning_rate": 4.4102000000000006e-06, "loss": 0.0332, "step": 27950 }, { "epoch": 1.4901636722290346, "grad_norm": 2.216271162033081, "learning_rate": 4.41e-06, "loss": 0.0347, "step": 27951 }, { "epoch": 1.4902169856586873, "grad_norm": 1.2433093786239624, "learning_rate": 4.4098000000000005e-06, "loss": 0.0268, "step": 27952 }, { "epoch": 1.4902702990883403, "grad_norm": 1.0809541940689087, "learning_rate": 4.409600000000001e-06, "loss": 0.0299, "step": 27953 }, { "epoch": 1.4903236125179933, "grad_norm": 1.3834208250045776, "learning_rate": 4.4094000000000005e-06, "loss": 0.0093, "step": 27954 }, { "epoch": 1.4903769259476463, "grad_norm": 3.468190908432007, "learning_rate": 4.4092e-06, "loss": 0.0156, "step": 27955 }, { "epoch": 1.4904302393772992, "grad_norm": 1.6719566583633423, "learning_rate": 4.4090000000000005e-06, "loss": 0.018, "step": 27956 }, { "epoch": 1.490483552806952, "grad_norm": 2.981818199157715, "learning_rate": 4.4088e-06, "loss": 0.0158, "step": 27957 }, { "epoch": 1.490536866236605, "grad_norm": 1.070353388786316, "learning_rate": 4.4086000000000004e-06, "loss": 0.0133, "step": 27958 }, { "epoch": 1.490590179666258, "grad_norm": 1.1368695497512817, "learning_rate": 4.408400000000001e-06, "loss": 0.0082, "step": 27959 }, { "epoch": 1.490643493095911, "grad_norm": 0.9711059927940369, "learning_rate": 4.4082e-06, "loss": 0.0106, "step": 27960 }, { "epoch": 1.490696806525564, "grad_norm": 1.3415775299072266, "learning_rate": 4.408000000000001e-06, "loss": 0.0274, "step": 27961 }, { "epoch": 1.4907501199552167, "grad_norm": 0.9400132894515991, "learning_rate": 4.4078e-06, "loss": 0.0089, "step": 27962 }, { "epoch": 1.4908034333848696, "grad_norm": 2.2689695358276367, "learning_rate": 4.4076e-06, "loss": 0.027, "step": 27963 }, { "epoch": 1.4908567468145226, "grad_norm": 1.4608038663864136, "learning_rate": 4.4074e-06, "loss": 0.0158, "step": 27964 }, { "epoch": 1.4909100602441754, "grad_norm": 1.0921739339828491, "learning_rate": 4.407200000000001e-06, "loss": 0.0166, "step": 27965 }, { "epoch": 1.4909633736738284, "grad_norm": 1.2547211647033691, "learning_rate": 4.407e-06, "loss": 0.022, "step": 27966 }, { "epoch": 1.4910166871034813, "grad_norm": 3.849276542663574, "learning_rate": 4.406800000000001e-06, "loss": 0.023, "step": 27967 }, { "epoch": 1.4910700005331343, "grad_norm": 2.8197786808013916, "learning_rate": 4.4066e-06, "loss": 0.0352, "step": 27968 }, { "epoch": 1.4911233139627873, "grad_norm": 1.4317035675048828, "learning_rate": 4.4064e-06, "loss": 0.012, "step": 27969 }, { "epoch": 1.49117662739244, "grad_norm": 1.0381547212600708, "learning_rate": 4.4062e-06, "loss": 0.0176, "step": 27970 }, { "epoch": 1.491229940822093, "grad_norm": 2.1443095207214355, "learning_rate": 4.406000000000001e-06, "loss": 0.0248, "step": 27971 }, { "epoch": 1.491283254251746, "grad_norm": 2.0909860134124756, "learning_rate": 4.4058e-06, "loss": 0.0188, "step": 27972 }, { "epoch": 1.491336567681399, "grad_norm": 3.7511775493621826, "learning_rate": 4.405600000000001e-06, "loss": 0.0514, "step": 27973 }, { "epoch": 1.491389881111052, "grad_norm": 1.6813088655471802, "learning_rate": 4.4054e-06, "loss": 0.0232, "step": 27974 }, { "epoch": 1.4914431945407047, "grad_norm": 1.987594723701477, "learning_rate": 4.405200000000001e-06, "loss": 0.0096, "step": 27975 }, { "epoch": 1.4914965079703577, "grad_norm": 1.1195900440216064, "learning_rate": 4.405e-06, "loss": 0.0311, "step": 27976 }, { "epoch": 1.4915498214000107, "grad_norm": 1.7353435754776, "learning_rate": 4.404800000000001e-06, "loss": 0.0144, "step": 27977 }, { "epoch": 1.4916031348296637, "grad_norm": 1.9404915571212769, "learning_rate": 4.4046e-06, "loss": 0.0153, "step": 27978 }, { "epoch": 1.4916564482593166, "grad_norm": 0.7341712713241577, "learning_rate": 4.4044000000000006e-06, "loss": 0.0037, "step": 27979 }, { "epoch": 1.4917097616889694, "grad_norm": 2.2409937381744385, "learning_rate": 4.4042e-06, "loss": 0.0346, "step": 27980 }, { "epoch": 1.4917630751186224, "grad_norm": 0.9800257682800293, "learning_rate": 4.4040000000000005e-06, "loss": 0.0059, "step": 27981 }, { "epoch": 1.4918163885482754, "grad_norm": 5.084105968475342, "learning_rate": 4.4038e-06, "loss": 0.0221, "step": 27982 }, { "epoch": 1.4918697019779281, "grad_norm": 0.9790161848068237, "learning_rate": 4.4036000000000005e-06, "loss": 0.0061, "step": 27983 }, { "epoch": 1.491923015407581, "grad_norm": 1.896723985671997, "learning_rate": 4.4034e-06, "loss": 0.0151, "step": 27984 }, { "epoch": 1.491976328837234, "grad_norm": 1.3242285251617432, "learning_rate": 4.4032000000000005e-06, "loss": 0.0164, "step": 27985 }, { "epoch": 1.492029642266887, "grad_norm": 1.0829718112945557, "learning_rate": 4.403000000000001e-06, "loss": 0.0261, "step": 27986 }, { "epoch": 1.49208295569654, "grad_norm": 1.2690582275390625, "learning_rate": 4.4028000000000004e-06, "loss": 0.0124, "step": 27987 }, { "epoch": 1.4921362691261928, "grad_norm": 1.440203309059143, "learning_rate": 4.4026e-06, "loss": 0.0279, "step": 27988 }, { "epoch": 1.4921895825558458, "grad_norm": 0.4377610683441162, "learning_rate": 4.4024e-06, "loss": 0.0046, "step": 27989 }, { "epoch": 1.4922428959854988, "grad_norm": 1.9986960887908936, "learning_rate": 4.4022e-06, "loss": 0.0117, "step": 27990 }, { "epoch": 1.4922962094151517, "grad_norm": 1.3791981935501099, "learning_rate": 4.402e-06, "loss": 0.0176, "step": 27991 }, { "epoch": 1.4923495228448047, "grad_norm": 2.10379958152771, "learning_rate": 4.401800000000001e-06, "loss": 0.0292, "step": 27992 }, { "epoch": 1.4924028362744575, "grad_norm": 3.380580425262451, "learning_rate": 4.4016e-06, "loss": 0.047, "step": 27993 }, { "epoch": 1.4924561497041104, "grad_norm": 0.8739624619483948, "learning_rate": 4.4014e-06, "loss": 0.0071, "step": 27994 }, { "epoch": 1.4925094631337634, "grad_norm": 4.301042556762695, "learning_rate": 4.4012e-06, "loss": 0.0284, "step": 27995 }, { "epoch": 1.4925627765634164, "grad_norm": 1.2346638441085815, "learning_rate": 4.401e-06, "loss": 0.0136, "step": 27996 }, { "epoch": 1.4926160899930694, "grad_norm": 1.6986831426620483, "learning_rate": 4.4008e-06, "loss": 0.0245, "step": 27997 }, { "epoch": 1.4926694034227221, "grad_norm": 3.9173333644866943, "learning_rate": 4.400600000000001e-06, "loss": 0.0536, "step": 27998 }, { "epoch": 1.4927227168523751, "grad_norm": 0.8542678356170654, "learning_rate": 4.4004e-06, "loss": 0.0159, "step": 27999 }, { "epoch": 1.492776030282028, "grad_norm": 1.6744214296340942, "learning_rate": 4.400200000000001e-06, "loss": 0.0075, "step": 28000 }, { "epoch": 1.4928293437116809, "grad_norm": 2.8801767826080322, "learning_rate": 4.4e-06, "loss": 0.0694, "step": 28001 }, { "epoch": 1.4928826571413338, "grad_norm": 1.5867482423782349, "learning_rate": 4.3998e-06, "loss": 0.016, "step": 28002 }, { "epoch": 1.4929359705709868, "grad_norm": 1.321629285812378, "learning_rate": 4.3996e-06, "loss": 0.0143, "step": 28003 }, { "epoch": 1.4929892840006398, "grad_norm": 1.9780231714248657, "learning_rate": 4.399400000000001e-06, "loss": 0.0409, "step": 28004 }, { "epoch": 1.4930425974302928, "grad_norm": 0.984300434589386, "learning_rate": 4.3992e-06, "loss": 0.0156, "step": 28005 }, { "epoch": 1.4930959108599455, "grad_norm": 2.597109079360962, "learning_rate": 4.3990000000000006e-06, "loss": 0.0333, "step": 28006 }, { "epoch": 1.4931492242895985, "grad_norm": 1.4766091108322144, "learning_rate": 4.3988e-06, "loss": 0.015, "step": 28007 }, { "epoch": 1.4932025377192515, "grad_norm": 2.659743547439575, "learning_rate": 4.3986e-06, "loss": 0.0247, "step": 28008 }, { "epoch": 1.4932558511489045, "grad_norm": 1.4041521549224854, "learning_rate": 4.3984e-06, "loss": 0.0173, "step": 28009 }, { "epoch": 1.4933091645785574, "grad_norm": 6.004517555236816, "learning_rate": 4.3982000000000005e-06, "loss": 0.0421, "step": 28010 }, { "epoch": 1.4933624780082102, "grad_norm": 3.7915172576904297, "learning_rate": 4.398000000000001e-06, "loss": 0.0449, "step": 28011 }, { "epoch": 1.4934157914378632, "grad_norm": 1.0676143169403076, "learning_rate": 4.3978000000000005e-06, "loss": 0.0118, "step": 28012 }, { "epoch": 1.4934691048675162, "grad_norm": 3.1467485427856445, "learning_rate": 4.3976e-06, "loss": 0.0429, "step": 28013 }, { "epoch": 1.493522418297169, "grad_norm": 2.905625820159912, "learning_rate": 4.3974000000000005e-06, "loss": 0.0345, "step": 28014 }, { "epoch": 1.4935757317268221, "grad_norm": 1.1934003829956055, "learning_rate": 4.3972e-06, "loss": 0.0062, "step": 28015 }, { "epoch": 1.4936290451564749, "grad_norm": 1.3644999265670776, "learning_rate": 4.397e-06, "loss": 0.0123, "step": 28016 }, { "epoch": 1.4936823585861279, "grad_norm": 2.2463197708129883, "learning_rate": 4.396800000000001e-06, "loss": 0.0422, "step": 28017 }, { "epoch": 1.4937356720157808, "grad_norm": 0.6989614367485046, "learning_rate": 4.3966e-06, "loss": 0.0053, "step": 28018 }, { "epoch": 1.4937889854454336, "grad_norm": 2.200403928756714, "learning_rate": 4.3964e-06, "loss": 0.0445, "step": 28019 }, { "epoch": 1.4938422988750866, "grad_norm": 0.9454531669616699, "learning_rate": 4.3962e-06, "loss": 0.0215, "step": 28020 }, { "epoch": 1.4938956123047396, "grad_norm": 0.5994604229927063, "learning_rate": 4.396e-06, "loss": 0.0061, "step": 28021 }, { "epoch": 1.4939489257343925, "grad_norm": 0.7701525688171387, "learning_rate": 4.3958e-06, "loss": 0.0075, "step": 28022 }, { "epoch": 1.4940022391640455, "grad_norm": 1.7308106422424316, "learning_rate": 4.395600000000001e-06, "loss": 0.0316, "step": 28023 }, { "epoch": 1.4940555525936983, "grad_norm": 1.881422519683838, "learning_rate": 4.3954e-06, "loss": 0.037, "step": 28024 }, { "epoch": 1.4941088660233512, "grad_norm": 1.0099685192108154, "learning_rate": 4.395200000000001e-06, "loss": 0.0232, "step": 28025 }, { "epoch": 1.4941621794530042, "grad_norm": 0.9333674907684326, "learning_rate": 4.395e-06, "loss": 0.0116, "step": 28026 }, { "epoch": 1.4942154928826572, "grad_norm": 1.6067887544631958, "learning_rate": 4.3948e-06, "loss": 0.0278, "step": 28027 }, { "epoch": 1.4942688063123102, "grad_norm": 1.975311279296875, "learning_rate": 4.3946e-06, "loss": 0.0223, "step": 28028 }, { "epoch": 1.494322119741963, "grad_norm": 2.4095585346221924, "learning_rate": 4.394400000000001e-06, "loss": 0.0396, "step": 28029 }, { "epoch": 1.494375433171616, "grad_norm": 2.3439881801605225, "learning_rate": 4.3942e-06, "loss": 0.0249, "step": 28030 }, { "epoch": 1.494428746601269, "grad_norm": 1.1535463333129883, "learning_rate": 4.394000000000001e-06, "loss": 0.0205, "step": 28031 }, { "epoch": 1.4944820600309217, "grad_norm": 1.244889497756958, "learning_rate": 4.3938e-06, "loss": 0.0194, "step": 28032 }, { "epoch": 1.4945353734605749, "grad_norm": 0.7462995052337646, "learning_rate": 4.3936e-06, "loss": 0.0062, "step": 28033 }, { "epoch": 1.4945886868902276, "grad_norm": 0.9229497909545898, "learning_rate": 4.3934e-06, "loss": 0.0079, "step": 28034 }, { "epoch": 1.4946420003198806, "grad_norm": 0.9978472590446472, "learning_rate": 4.3932000000000006e-06, "loss": 0.0122, "step": 28035 }, { "epoch": 1.4946953137495336, "grad_norm": 3.485602378845215, "learning_rate": 4.393000000000001e-06, "loss": 0.0309, "step": 28036 }, { "epoch": 1.4947486271791863, "grad_norm": 1.975350022315979, "learning_rate": 4.3928000000000005e-06, "loss": 0.0391, "step": 28037 }, { "epoch": 1.4948019406088393, "grad_norm": 1.0117851495742798, "learning_rate": 4.3926e-06, "loss": 0.0163, "step": 28038 }, { "epoch": 1.4948552540384923, "grad_norm": 1.0527963638305664, "learning_rate": 4.3924000000000005e-06, "loss": 0.0108, "step": 28039 }, { "epoch": 1.4949085674681453, "grad_norm": 3.745727300643921, "learning_rate": 4.3922e-06, "loss": 0.0558, "step": 28040 }, { "epoch": 1.4949618808977982, "grad_norm": 1.7854959964752197, "learning_rate": 4.3920000000000005e-06, "loss": 0.0256, "step": 28041 }, { "epoch": 1.495015194327451, "grad_norm": 2.1031322479248047, "learning_rate": 4.391800000000001e-06, "loss": 0.018, "step": 28042 }, { "epoch": 1.495068507757104, "grad_norm": 0.5518845319747925, "learning_rate": 4.3916000000000004e-06, "loss": 0.0055, "step": 28043 }, { "epoch": 1.495121821186757, "grad_norm": 1.5009877681732178, "learning_rate": 4.3914e-06, "loss": 0.0289, "step": 28044 }, { "epoch": 1.49517513461641, "grad_norm": 2.303219795227051, "learning_rate": 4.3912e-06, "loss": 0.0241, "step": 28045 }, { "epoch": 1.495228448046063, "grad_norm": 1.3293724060058594, "learning_rate": 4.391e-06, "loss": 0.016, "step": 28046 }, { "epoch": 1.4952817614757157, "grad_norm": 1.2839754819869995, "learning_rate": 4.3908e-06, "loss": 0.015, "step": 28047 }, { "epoch": 1.4953350749053687, "grad_norm": 1.3783637285232544, "learning_rate": 4.390600000000001e-06, "loss": 0.0107, "step": 28048 }, { "epoch": 1.4953883883350216, "grad_norm": 3.109805107116699, "learning_rate": 4.3904e-06, "loss": 0.0328, "step": 28049 }, { "epoch": 1.4954417017646744, "grad_norm": 1.8251992464065552, "learning_rate": 4.390200000000001e-06, "loss": 0.0337, "step": 28050 }, { "epoch": 1.4954950151943274, "grad_norm": 1.9243192672729492, "learning_rate": 4.39e-06, "loss": 0.0247, "step": 28051 }, { "epoch": 1.4955483286239803, "grad_norm": 1.0318231582641602, "learning_rate": 4.3898e-06, "loss": 0.0167, "step": 28052 }, { "epoch": 1.4956016420536333, "grad_norm": 1.9219651222229004, "learning_rate": 4.3896e-06, "loss": 0.0479, "step": 28053 }, { "epoch": 1.4956549554832863, "grad_norm": 1.8147011995315552, "learning_rate": 4.389400000000001e-06, "loss": 0.0244, "step": 28054 }, { "epoch": 1.495708268912939, "grad_norm": 0.4919692277908325, "learning_rate": 4.3892e-06, "loss": 0.0046, "step": 28055 }, { "epoch": 1.495761582342592, "grad_norm": 1.7086652517318726, "learning_rate": 4.389000000000001e-06, "loss": 0.0225, "step": 28056 }, { "epoch": 1.495814895772245, "grad_norm": 3.6537835597991943, "learning_rate": 4.3888e-06, "loss": 0.0265, "step": 28057 }, { "epoch": 1.495868209201898, "grad_norm": 2.0612380504608154, "learning_rate": 4.3886e-06, "loss": 0.0573, "step": 28058 }, { "epoch": 1.495921522631551, "grad_norm": 1.5751514434814453, "learning_rate": 4.3884e-06, "loss": 0.0142, "step": 28059 }, { "epoch": 1.4959748360612037, "grad_norm": 0.3157632052898407, "learning_rate": 4.388200000000001e-06, "loss": 0.0034, "step": 28060 }, { "epoch": 1.4960281494908567, "grad_norm": 0.963158369064331, "learning_rate": 4.388e-06, "loss": 0.0062, "step": 28061 }, { "epoch": 1.4960814629205097, "grad_norm": 1.278113603591919, "learning_rate": 4.387800000000001e-06, "loss": 0.0159, "step": 28062 }, { "epoch": 1.4961347763501627, "grad_norm": 2.040510892868042, "learning_rate": 4.3876e-06, "loss": 0.0327, "step": 28063 }, { "epoch": 1.4961880897798157, "grad_norm": 1.015692949295044, "learning_rate": 4.3874000000000005e-06, "loss": 0.0225, "step": 28064 }, { "epoch": 1.4962414032094684, "grad_norm": 1.100766658782959, "learning_rate": 4.3872e-06, "loss": 0.0136, "step": 28065 }, { "epoch": 1.4962947166391214, "grad_norm": 0.6088535189628601, "learning_rate": 4.3870000000000005e-06, "loss": 0.0061, "step": 28066 }, { "epoch": 1.4963480300687744, "grad_norm": 1.4897576570510864, "learning_rate": 4.3868e-06, "loss": 0.0176, "step": 28067 }, { "epoch": 1.4964013434984271, "grad_norm": 1.6318379640579224, "learning_rate": 4.3866000000000005e-06, "loss": 0.0286, "step": 28068 }, { "epoch": 1.49645465692808, "grad_norm": 8.891934394836426, "learning_rate": 4.3864e-06, "loss": 0.05, "step": 28069 }, { "epoch": 1.496507970357733, "grad_norm": 0.7732200622558594, "learning_rate": 4.3862000000000005e-06, "loss": 0.0069, "step": 28070 }, { "epoch": 1.496561283787386, "grad_norm": 4.008398532867432, "learning_rate": 4.386e-06, "loss": 0.0422, "step": 28071 }, { "epoch": 1.496614597217039, "grad_norm": 0.9763007164001465, "learning_rate": 4.3858000000000004e-06, "loss": 0.0108, "step": 28072 }, { "epoch": 1.4966679106466918, "grad_norm": 1.0460132360458374, "learning_rate": 4.3856e-06, "loss": 0.015, "step": 28073 }, { "epoch": 1.4967212240763448, "grad_norm": 2.2624495029449463, "learning_rate": 4.3854e-06, "loss": 0.0441, "step": 28074 }, { "epoch": 1.4967745375059978, "grad_norm": 1.6806985139846802, "learning_rate": 4.385200000000001e-06, "loss": 0.0236, "step": 28075 }, { "epoch": 1.4968278509356507, "grad_norm": 2.004324436187744, "learning_rate": 4.385e-06, "loss": 0.0297, "step": 28076 }, { "epoch": 1.4968811643653037, "grad_norm": 2.625643253326416, "learning_rate": 4.3848e-06, "loss": 0.0374, "step": 28077 }, { "epoch": 1.4969344777949565, "grad_norm": 0.9888501167297363, "learning_rate": 4.3846e-06, "loss": 0.0138, "step": 28078 }, { "epoch": 1.4969877912246095, "grad_norm": 1.120693564414978, "learning_rate": 4.3844e-06, "loss": 0.0227, "step": 28079 }, { "epoch": 1.4970411046542624, "grad_norm": 3.0459797382354736, "learning_rate": 4.3842e-06, "loss": 0.0371, "step": 28080 }, { "epoch": 1.4970944180839154, "grad_norm": 2.0334672927856445, "learning_rate": 4.384000000000001e-06, "loss": 0.0171, "step": 28081 }, { "epoch": 1.4971477315135684, "grad_norm": 2.135446310043335, "learning_rate": 4.3838e-06, "loss": 0.0339, "step": 28082 }, { "epoch": 1.4972010449432211, "grad_norm": 1.9660016298294067, "learning_rate": 4.383600000000001e-06, "loss": 0.0217, "step": 28083 }, { "epoch": 1.4972543583728741, "grad_norm": 0.26291292905807495, "learning_rate": 4.3834e-06, "loss": 0.0027, "step": 28084 }, { "epoch": 1.497307671802527, "grad_norm": 0.24532918632030487, "learning_rate": 4.3832e-06, "loss": 0.0027, "step": 28085 }, { "epoch": 1.4973609852321799, "grad_norm": 2.3355042934417725, "learning_rate": 4.383e-06, "loss": 0.0176, "step": 28086 }, { "epoch": 1.4974142986618328, "grad_norm": 2.410551071166992, "learning_rate": 4.382800000000001e-06, "loss": 0.0173, "step": 28087 }, { "epoch": 1.4974676120914858, "grad_norm": 1.7380154132843018, "learning_rate": 4.3826e-06, "loss": 0.0248, "step": 28088 }, { "epoch": 1.4975209255211388, "grad_norm": 1.7563410997390747, "learning_rate": 4.382400000000001e-06, "loss": 0.016, "step": 28089 }, { "epoch": 1.4975742389507918, "grad_norm": 0.31464511156082153, "learning_rate": 4.3822e-06, "loss": 0.0036, "step": 28090 }, { "epoch": 1.4976275523804445, "grad_norm": 2.8064093589782715, "learning_rate": 4.382e-06, "loss": 0.0144, "step": 28091 }, { "epoch": 1.4976808658100975, "grad_norm": 1.4720314741134644, "learning_rate": 4.3818e-06, "loss": 0.0142, "step": 28092 }, { "epoch": 1.4977341792397505, "grad_norm": 1.475919246673584, "learning_rate": 4.3816000000000005e-06, "loss": 0.0272, "step": 28093 }, { "epoch": 1.4977874926694035, "grad_norm": 3.5022411346435547, "learning_rate": 4.3814e-06, "loss": 0.0219, "step": 28094 }, { "epoch": 1.4978408060990565, "grad_norm": 0.9846099019050598, "learning_rate": 4.3812000000000005e-06, "loss": 0.0086, "step": 28095 }, { "epoch": 1.4978941195287092, "grad_norm": 1.7543001174926758, "learning_rate": 4.381e-06, "loss": 0.021, "step": 28096 }, { "epoch": 1.4979474329583622, "grad_norm": 2.007115602493286, "learning_rate": 4.3808000000000005e-06, "loss": 0.019, "step": 28097 }, { "epoch": 1.4980007463880152, "grad_norm": 1.4239528179168701, "learning_rate": 4.3806e-06, "loss": 0.0272, "step": 28098 }, { "epoch": 1.4980540598176681, "grad_norm": 1.2130537033081055, "learning_rate": 4.3804000000000004e-06, "loss": 0.0113, "step": 28099 }, { "epoch": 1.4981073732473211, "grad_norm": 1.0886344909667969, "learning_rate": 4.380200000000001e-06, "loss": 0.0058, "step": 28100 }, { "epoch": 1.4981606866769739, "grad_norm": 0.978756308555603, "learning_rate": 4.38e-06, "loss": 0.0079, "step": 28101 }, { "epoch": 1.4982140001066269, "grad_norm": 2.583012819290161, "learning_rate": 4.3798e-06, "loss": 0.0166, "step": 28102 }, { "epoch": 1.4982673135362798, "grad_norm": 1.1124640703201294, "learning_rate": 4.3796e-06, "loss": 0.0062, "step": 28103 }, { "epoch": 1.4983206269659326, "grad_norm": 2.1368565559387207, "learning_rate": 4.3794e-06, "loss": 0.0298, "step": 28104 }, { "epoch": 1.4983739403955856, "grad_norm": 1.8685860633850098, "learning_rate": 4.3792e-06, "loss": 0.0112, "step": 28105 }, { "epoch": 1.4984272538252386, "grad_norm": 1.078102946281433, "learning_rate": 4.379000000000001e-06, "loss": 0.0139, "step": 28106 }, { "epoch": 1.4984805672548915, "grad_norm": 1.822539210319519, "learning_rate": 4.3788e-06, "loss": 0.0091, "step": 28107 }, { "epoch": 1.4985338806845445, "grad_norm": 2.9056718349456787, "learning_rate": 4.378600000000001e-06, "loss": 0.0257, "step": 28108 }, { "epoch": 1.4985871941141973, "grad_norm": 0.6821184158325195, "learning_rate": 4.3784e-06, "loss": 0.0059, "step": 28109 }, { "epoch": 1.4986405075438503, "grad_norm": 1.4583773612976074, "learning_rate": 4.3782e-06, "loss": 0.0286, "step": 28110 }, { "epoch": 1.4986938209735032, "grad_norm": 2.4101924896240234, "learning_rate": 4.378e-06, "loss": 0.046, "step": 28111 }, { "epoch": 1.4987471344031562, "grad_norm": 0.5424550175666809, "learning_rate": 4.377800000000001e-06, "loss": 0.0044, "step": 28112 }, { "epoch": 1.4988004478328092, "grad_norm": 2.6500346660614014, "learning_rate": 4.3776e-06, "loss": 0.013, "step": 28113 }, { "epoch": 1.498853761262462, "grad_norm": 1.9384280443191528, "learning_rate": 4.377400000000001e-06, "loss": 0.024, "step": 28114 }, { "epoch": 1.498907074692115, "grad_norm": 0.8820634484291077, "learning_rate": 4.3772e-06, "loss": 0.0033, "step": 28115 }, { "epoch": 1.498960388121768, "grad_norm": 1.8804512023925781, "learning_rate": 4.377e-06, "loss": 0.0255, "step": 28116 }, { "epoch": 1.4990137015514209, "grad_norm": 3.969327926635742, "learning_rate": 4.3768e-06, "loss": 0.0432, "step": 28117 }, { "epoch": 1.4990670149810739, "grad_norm": 4.043468475341797, "learning_rate": 4.376600000000001e-06, "loss": 0.0436, "step": 28118 }, { "epoch": 1.4991203284107266, "grad_norm": 0.9276353120803833, "learning_rate": 4.3764e-06, "loss": 0.0083, "step": 28119 }, { "epoch": 1.4991736418403796, "grad_norm": 1.2230643033981323, "learning_rate": 4.3762000000000006e-06, "loss": 0.0065, "step": 28120 }, { "epoch": 1.4992269552700326, "grad_norm": 0.8009188771247864, "learning_rate": 4.376e-06, "loss": 0.0048, "step": 28121 }, { "epoch": 1.4992802686996853, "grad_norm": 2.079957962036133, "learning_rate": 4.3758000000000005e-06, "loss": 0.0249, "step": 28122 }, { "epoch": 1.4993335821293383, "grad_norm": 1.878968358039856, "learning_rate": 4.3756e-06, "loss": 0.0208, "step": 28123 }, { "epoch": 1.4993868955589913, "grad_norm": 1.1626018285751343, "learning_rate": 4.3754000000000005e-06, "loss": 0.0156, "step": 28124 }, { "epoch": 1.4994402089886443, "grad_norm": 1.9631949663162231, "learning_rate": 4.375200000000001e-06, "loss": 0.0303, "step": 28125 }, { "epoch": 1.4994935224182973, "grad_norm": 2.3381855487823486, "learning_rate": 4.3750000000000005e-06, "loss": 0.0585, "step": 28126 }, { "epoch": 1.49954683584795, "grad_norm": 1.1503616571426392, "learning_rate": 4.3748e-06, "loss": 0.0314, "step": 28127 }, { "epoch": 1.499600149277603, "grad_norm": 0.9754261374473572, "learning_rate": 4.3746000000000004e-06, "loss": 0.0041, "step": 28128 }, { "epoch": 1.499653462707256, "grad_norm": 2.149174928665161, "learning_rate": 4.3744e-06, "loss": 0.0179, "step": 28129 }, { "epoch": 1.499706776136909, "grad_norm": 1.4788944721221924, "learning_rate": 4.3742e-06, "loss": 0.0262, "step": 28130 }, { "epoch": 1.499760089566562, "grad_norm": 1.77283775806427, "learning_rate": 4.374000000000001e-06, "loss": 0.0183, "step": 28131 }, { "epoch": 1.4998134029962147, "grad_norm": 1.2574336528778076, "learning_rate": 4.3738e-06, "loss": 0.0309, "step": 28132 }, { "epoch": 1.4998667164258677, "grad_norm": 1.0798852443695068, "learning_rate": 4.373600000000001e-06, "loss": 0.0079, "step": 28133 }, { "epoch": 1.4999200298555206, "grad_norm": 2.464686870574951, "learning_rate": 4.3734e-06, "loss": 0.0329, "step": 28134 }, { "epoch": 1.4999733432851734, "grad_norm": 6.078603267669678, "learning_rate": 4.3732e-06, "loss": 0.0414, "step": 28135 }, { "epoch": 1.5000266567148266, "grad_norm": 0.3917997181415558, "learning_rate": 4.373e-06, "loss": 0.0038, "step": 28136 }, { "epoch": 1.5000799701444794, "grad_norm": 0.6920256018638611, "learning_rate": 4.372800000000001e-06, "loss": 0.0071, "step": 28137 }, { "epoch": 1.5001332835741323, "grad_norm": 3.1584296226501465, "learning_rate": 4.3726e-06, "loss": 0.0305, "step": 28138 }, { "epoch": 1.5001865970037853, "grad_norm": 1.9269180297851562, "learning_rate": 4.372400000000001e-06, "loss": 0.0191, "step": 28139 }, { "epoch": 1.500239910433438, "grad_norm": 3.314988136291504, "learning_rate": 4.3722e-06, "loss": 0.0503, "step": 28140 }, { "epoch": 1.5002932238630913, "grad_norm": 0.511941134929657, "learning_rate": 4.372e-06, "loss": 0.0041, "step": 28141 }, { "epoch": 1.500346537292744, "grad_norm": 4.668336391448975, "learning_rate": 4.3718e-06, "loss": 0.0195, "step": 28142 }, { "epoch": 1.500399850722397, "grad_norm": 1.612580418586731, "learning_rate": 4.371600000000001e-06, "loss": 0.0149, "step": 28143 }, { "epoch": 1.50045316415205, "grad_norm": 0.9211073517799377, "learning_rate": 4.3714e-06, "loss": 0.0052, "step": 28144 }, { "epoch": 1.5005064775817027, "grad_norm": 2.9279727935791016, "learning_rate": 4.371200000000001e-06, "loss": 0.0446, "step": 28145 }, { "epoch": 1.5005597910113557, "grad_norm": 4.0976972579956055, "learning_rate": 4.371e-06, "loss": 0.0793, "step": 28146 }, { "epoch": 1.5006131044410087, "grad_norm": 2.0847792625427246, "learning_rate": 4.3708000000000006e-06, "loss": 0.0343, "step": 28147 }, { "epoch": 1.5006664178706615, "grad_norm": 1.4996094703674316, "learning_rate": 4.3706e-06, "loss": 0.0158, "step": 28148 }, { "epoch": 1.5007197313003147, "grad_norm": 1.3592040538787842, "learning_rate": 4.3704000000000005e-06, "loss": 0.0098, "step": 28149 }, { "epoch": 1.5007730447299674, "grad_norm": 1.5325273275375366, "learning_rate": 4.370200000000001e-06, "loss": 0.0261, "step": 28150 }, { "epoch": 1.5008263581596204, "grad_norm": 3.1449413299560547, "learning_rate": 4.3700000000000005e-06, "loss": 0.0207, "step": 28151 }, { "epoch": 1.5008796715892734, "grad_norm": 1.5077341794967651, "learning_rate": 4.3698e-06, "loss": 0.0156, "step": 28152 }, { "epoch": 1.5009329850189261, "grad_norm": 1.998100757598877, "learning_rate": 4.3696000000000005e-06, "loss": 0.0259, "step": 28153 }, { "epoch": 1.5009862984485793, "grad_norm": 2.36301589012146, "learning_rate": 4.3694e-06, "loss": 0.0332, "step": 28154 }, { "epoch": 1.501039611878232, "grad_norm": 2.0083723068237305, "learning_rate": 4.3692000000000005e-06, "loss": 0.0276, "step": 28155 }, { "epoch": 1.501092925307885, "grad_norm": 1.4752013683319092, "learning_rate": 4.369000000000001e-06, "loss": 0.0122, "step": 28156 }, { "epoch": 1.501146238737538, "grad_norm": 4.48253059387207, "learning_rate": 4.3688e-06, "loss": 0.0169, "step": 28157 }, { "epoch": 1.5011995521671908, "grad_norm": 0.9610234498977661, "learning_rate": 4.368600000000001e-06, "loss": 0.021, "step": 28158 }, { "epoch": 1.501252865596844, "grad_norm": 2.3486099243164062, "learning_rate": 4.3684e-06, "loss": 0.0236, "step": 28159 }, { "epoch": 1.5013061790264968, "grad_norm": 1.1559324264526367, "learning_rate": 4.3682e-06, "loss": 0.0147, "step": 28160 }, { "epoch": 1.5013594924561497, "grad_norm": 2.094027519226074, "learning_rate": 4.368e-06, "loss": 0.0418, "step": 28161 }, { "epoch": 1.5014128058858027, "grad_norm": 2.7740275859832764, "learning_rate": 4.367800000000001e-06, "loss": 0.0184, "step": 28162 }, { "epoch": 1.5014661193154555, "grad_norm": 1.1237705945968628, "learning_rate": 4.3676e-06, "loss": 0.01, "step": 28163 }, { "epoch": 1.5015194327451085, "grad_norm": 5.041398048400879, "learning_rate": 4.367400000000001e-06, "loss": 0.0561, "step": 28164 }, { "epoch": 1.5015727461747614, "grad_norm": 2.217658519744873, "learning_rate": 4.3672e-06, "loss": 0.0324, "step": 28165 }, { "epoch": 1.5016260596044142, "grad_norm": 2.701596736907959, "learning_rate": 4.367e-06, "loss": 0.015, "step": 28166 }, { "epoch": 1.5016793730340674, "grad_norm": 1.5193259716033936, "learning_rate": 4.3668e-06, "loss": 0.0213, "step": 28167 }, { "epoch": 1.5017326864637202, "grad_norm": 1.0700163841247559, "learning_rate": 4.366600000000001e-06, "loss": 0.0064, "step": 28168 }, { "epoch": 1.5017859998933731, "grad_norm": 1.3130857944488525, "learning_rate": 4.3664e-06, "loss": 0.0117, "step": 28169 }, { "epoch": 1.5018393133230261, "grad_norm": 3.5830459594726562, "learning_rate": 4.366200000000001e-06, "loss": 0.0369, "step": 28170 }, { "epoch": 1.5018926267526789, "grad_norm": 1.254407286643982, "learning_rate": 4.366e-06, "loss": 0.0257, "step": 28171 }, { "epoch": 1.501945940182332, "grad_norm": 1.418838620185852, "learning_rate": 4.365800000000001e-06, "loss": 0.0184, "step": 28172 }, { "epoch": 1.5019992536119848, "grad_norm": 1.2967723608016968, "learning_rate": 4.3656e-06, "loss": 0.016, "step": 28173 }, { "epoch": 1.5020525670416378, "grad_norm": 0.2391601800918579, "learning_rate": 4.365400000000001e-06, "loss": 0.0029, "step": 28174 }, { "epoch": 1.5021058804712908, "grad_norm": 1.823659062385559, "learning_rate": 4.3652e-06, "loss": 0.0348, "step": 28175 }, { "epoch": 1.5021591939009435, "grad_norm": 1.0284311771392822, "learning_rate": 4.3650000000000006e-06, "loss": 0.0073, "step": 28176 }, { "epoch": 1.5022125073305965, "grad_norm": 2.2961950302124023, "learning_rate": 4.3648e-06, "loss": 0.0337, "step": 28177 }, { "epoch": 1.5022658207602495, "grad_norm": 1.0850173234939575, "learning_rate": 4.3646000000000005e-06, "loss": 0.0094, "step": 28178 }, { "epoch": 1.5023191341899025, "grad_norm": 2.10861873626709, "learning_rate": 4.3644e-06, "loss": 0.017, "step": 28179 }, { "epoch": 1.5023724476195555, "grad_norm": 1.0330039262771606, "learning_rate": 4.3642000000000005e-06, "loss": 0.0165, "step": 28180 }, { "epoch": 1.5024257610492082, "grad_norm": 1.383540391921997, "learning_rate": 4.364e-06, "loss": 0.0114, "step": 28181 }, { "epoch": 1.5024790744788612, "grad_norm": 1.7635248899459839, "learning_rate": 4.3638000000000005e-06, "loss": 0.0448, "step": 28182 }, { "epoch": 1.5025323879085142, "grad_norm": 2.1600100994110107, "learning_rate": 4.363600000000001e-06, "loss": 0.0182, "step": 28183 }, { "epoch": 1.502585701338167, "grad_norm": 2.704133987426758, "learning_rate": 4.3634000000000004e-06, "loss": 0.0374, "step": 28184 }, { "epoch": 1.5026390147678201, "grad_norm": 1.4707762002944946, "learning_rate": 4.3632e-06, "loss": 0.0217, "step": 28185 }, { "epoch": 1.502692328197473, "grad_norm": 5.117799758911133, "learning_rate": 4.363e-06, "loss": 0.0305, "step": 28186 }, { "epoch": 1.5027456416271259, "grad_norm": 1.7563636302947998, "learning_rate": 4.3628e-06, "loss": 0.021, "step": 28187 }, { "epoch": 1.5027989550567789, "grad_norm": 1.3713783025741577, "learning_rate": 4.3626e-06, "loss": 0.018, "step": 28188 }, { "epoch": 1.5028522684864316, "grad_norm": 0.8365375995635986, "learning_rate": 4.362400000000001e-06, "loss": 0.0128, "step": 28189 }, { "epoch": 1.5029055819160848, "grad_norm": 1.1869922876358032, "learning_rate": 4.3622e-06, "loss": 0.0131, "step": 28190 }, { "epoch": 1.5029588953457376, "grad_norm": 1.734387755393982, "learning_rate": 4.362e-06, "loss": 0.0272, "step": 28191 }, { "epoch": 1.5030122087753905, "grad_norm": 2.3318560123443604, "learning_rate": 4.3618e-06, "loss": 0.0303, "step": 28192 }, { "epoch": 1.5030655222050435, "grad_norm": 1.6052067279815674, "learning_rate": 4.3616e-06, "loss": 0.034, "step": 28193 }, { "epoch": 1.5031188356346963, "grad_norm": 1.8647799491882324, "learning_rate": 4.3614e-06, "loss": 0.0547, "step": 28194 }, { "epoch": 1.5031721490643493, "grad_norm": 0.8331997990608215, "learning_rate": 4.361200000000001e-06, "loss": 0.0071, "step": 28195 }, { "epoch": 1.5032254624940022, "grad_norm": 1.9564871788024902, "learning_rate": 4.361e-06, "loss": 0.0141, "step": 28196 }, { "epoch": 1.5032787759236552, "grad_norm": 1.1955020427703857, "learning_rate": 4.360800000000001e-06, "loss": 0.0087, "step": 28197 }, { "epoch": 1.5033320893533082, "grad_norm": 1.097306489944458, "learning_rate": 4.3606e-06, "loss": 0.0151, "step": 28198 }, { "epoch": 1.503385402782961, "grad_norm": 1.3501285314559937, "learning_rate": 4.3604e-06, "loss": 0.0225, "step": 28199 }, { "epoch": 1.503438716212614, "grad_norm": 0.8835560083389282, "learning_rate": 4.3602e-06, "loss": 0.0078, "step": 28200 }, { "epoch": 1.503492029642267, "grad_norm": 1.0693840980529785, "learning_rate": 4.360000000000001e-06, "loss": 0.0096, "step": 28201 }, { "epoch": 1.5035453430719197, "grad_norm": 2.919875383377075, "learning_rate": 4.3598e-06, "loss": 0.0148, "step": 28202 }, { "epoch": 1.5035986565015729, "grad_norm": 1.7189185619354248, "learning_rate": 4.359600000000001e-06, "loss": 0.012, "step": 28203 }, { "epoch": 1.5036519699312256, "grad_norm": 1.978287696838379, "learning_rate": 4.3594e-06, "loss": 0.0196, "step": 28204 }, { "epoch": 1.5037052833608786, "grad_norm": 0.5692324638366699, "learning_rate": 4.3592e-06, "loss": 0.004, "step": 28205 }, { "epoch": 1.5037585967905316, "grad_norm": 0.8999506235122681, "learning_rate": 4.359e-06, "loss": 0.0066, "step": 28206 }, { "epoch": 1.5038119102201843, "grad_norm": 0.6497961282730103, "learning_rate": 4.3588000000000005e-06, "loss": 0.0046, "step": 28207 }, { "epoch": 1.5038652236498375, "grad_norm": 2.1370322704315186, "learning_rate": 4.358600000000001e-06, "loss": 0.0282, "step": 28208 }, { "epoch": 1.5039185370794903, "grad_norm": 0.5937811136245728, "learning_rate": 4.3584000000000005e-06, "loss": 0.0052, "step": 28209 }, { "epoch": 1.5039718505091433, "grad_norm": 1.3959347009658813, "learning_rate": 4.3582e-06, "loss": 0.0136, "step": 28210 }, { "epoch": 1.5040251639387963, "grad_norm": 2.423002004623413, "learning_rate": 4.3580000000000005e-06, "loss": 0.0175, "step": 28211 }, { "epoch": 1.504078477368449, "grad_norm": 2.5605297088623047, "learning_rate": 4.3578e-06, "loss": 0.039, "step": 28212 }, { "epoch": 1.504131790798102, "grad_norm": 0.9713126420974731, "learning_rate": 4.3576000000000004e-06, "loss": 0.007, "step": 28213 }, { "epoch": 1.504185104227755, "grad_norm": 2.291257619857788, "learning_rate": 4.357400000000001e-06, "loss": 0.0382, "step": 28214 }, { "epoch": 1.504238417657408, "grad_norm": 2.192915201187134, "learning_rate": 4.3572e-06, "loss": 0.0099, "step": 28215 }, { "epoch": 1.504291731087061, "grad_norm": 1.2764228582382202, "learning_rate": 4.357e-06, "loss": 0.0087, "step": 28216 }, { "epoch": 1.5043450445167137, "grad_norm": 2.066358804702759, "learning_rate": 4.3568e-06, "loss": 0.0256, "step": 28217 }, { "epoch": 1.5043983579463667, "grad_norm": 2.5874135494232178, "learning_rate": 4.3566e-06, "loss": 0.0541, "step": 28218 }, { "epoch": 1.5044516713760197, "grad_norm": 1.888506293296814, "learning_rate": 4.3564e-06, "loss": 0.0296, "step": 28219 }, { "epoch": 1.5045049848056724, "grad_norm": 1.1778920888900757, "learning_rate": 4.356200000000001e-06, "loss": 0.0112, "step": 28220 }, { "epoch": 1.5045582982353256, "grad_norm": 0.9928310513496399, "learning_rate": 4.356e-06, "loss": 0.0164, "step": 28221 }, { "epoch": 1.5046116116649784, "grad_norm": 2.5863213539123535, "learning_rate": 4.355800000000001e-06, "loss": 0.0211, "step": 28222 }, { "epoch": 1.5046649250946313, "grad_norm": 4.998314380645752, "learning_rate": 4.3556e-06, "loss": 0.0288, "step": 28223 }, { "epoch": 1.5047182385242843, "grad_norm": 2.388692855834961, "learning_rate": 4.3554e-06, "loss": 0.0241, "step": 28224 }, { "epoch": 1.504771551953937, "grad_norm": 2.148745059967041, "learning_rate": 4.3552e-06, "loss": 0.0231, "step": 28225 }, { "epoch": 1.5048248653835903, "grad_norm": 2.622319221496582, "learning_rate": 4.355000000000001e-06, "loss": 0.0291, "step": 28226 }, { "epoch": 1.504878178813243, "grad_norm": 1.6070820093154907, "learning_rate": 4.3548e-06, "loss": 0.0135, "step": 28227 }, { "epoch": 1.504931492242896, "grad_norm": 0.9780400395393372, "learning_rate": 4.354600000000001e-06, "loss": 0.0066, "step": 28228 }, { "epoch": 1.504984805672549, "grad_norm": 1.116389274597168, "learning_rate": 4.3544e-06, "loss": 0.0335, "step": 28229 }, { "epoch": 1.5050381191022018, "grad_norm": 4.88291072845459, "learning_rate": 4.3542e-06, "loss": 0.0284, "step": 28230 }, { "epoch": 1.5050914325318547, "grad_norm": 2.051074266433716, "learning_rate": 4.354e-06, "loss": 0.023, "step": 28231 }, { "epoch": 1.5051447459615077, "grad_norm": 2.1689505577087402, "learning_rate": 4.3538000000000006e-06, "loss": 0.0417, "step": 28232 }, { "epoch": 1.5051980593911605, "grad_norm": 0.9660688042640686, "learning_rate": 4.353600000000001e-06, "loss": 0.0086, "step": 28233 }, { "epoch": 1.5052513728208137, "grad_norm": 2.5726263523101807, "learning_rate": 4.3534000000000005e-06, "loss": 0.0295, "step": 28234 }, { "epoch": 1.5053046862504664, "grad_norm": 2.356255054473877, "learning_rate": 4.3532e-06, "loss": 0.0332, "step": 28235 }, { "epoch": 1.5053579996801194, "grad_norm": 1.912062644958496, "learning_rate": 4.3530000000000005e-06, "loss": 0.0192, "step": 28236 }, { "epoch": 1.5054113131097724, "grad_norm": 1.5690878629684448, "learning_rate": 4.3528e-06, "loss": 0.0256, "step": 28237 }, { "epoch": 1.5054646265394251, "grad_norm": 1.332122564315796, "learning_rate": 4.3526000000000005e-06, "loss": 0.0092, "step": 28238 }, { "epoch": 1.5055179399690783, "grad_norm": 1.491141438484192, "learning_rate": 4.352400000000001e-06, "loss": 0.0075, "step": 28239 }, { "epoch": 1.505571253398731, "grad_norm": 1.1195013523101807, "learning_rate": 4.3522000000000004e-06, "loss": 0.0196, "step": 28240 }, { "epoch": 1.505624566828384, "grad_norm": 1.1659376621246338, "learning_rate": 4.352e-06, "loss": 0.0131, "step": 28241 }, { "epoch": 1.505677880258037, "grad_norm": 1.5160529613494873, "learning_rate": 4.3518e-06, "loss": 0.0185, "step": 28242 }, { "epoch": 1.5057311936876898, "grad_norm": 0.4157038927078247, "learning_rate": 4.3516e-06, "loss": 0.0036, "step": 28243 }, { "epoch": 1.505784507117343, "grad_norm": 2.745220422744751, "learning_rate": 4.3514e-06, "loss": 0.021, "step": 28244 }, { "epoch": 1.5058378205469958, "grad_norm": 0.7153242826461792, "learning_rate": 4.351200000000001e-06, "loss": 0.0065, "step": 28245 }, { "epoch": 1.5058911339766488, "grad_norm": 1.2122083902359009, "learning_rate": 4.351e-06, "loss": 0.023, "step": 28246 }, { "epoch": 1.5059444474063017, "grad_norm": 1.16048002243042, "learning_rate": 4.350800000000001e-06, "loss": 0.0099, "step": 28247 }, { "epoch": 1.5059977608359545, "grad_norm": 1.2698513269424438, "learning_rate": 4.3506e-06, "loss": 0.0138, "step": 28248 }, { "epoch": 1.5060510742656075, "grad_norm": 1.453423023223877, "learning_rate": 4.3504e-06, "loss": 0.0168, "step": 28249 }, { "epoch": 1.5061043876952604, "grad_norm": 2.178129196166992, "learning_rate": 4.3502e-06, "loss": 0.0426, "step": 28250 }, { "epoch": 1.5061577011249132, "grad_norm": 0.32712361216545105, "learning_rate": 4.350000000000001e-06, "loss": 0.0031, "step": 28251 }, { "epoch": 1.5062110145545664, "grad_norm": 0.8965964317321777, "learning_rate": 4.3498e-06, "loss": 0.0065, "step": 28252 }, { "epoch": 1.5062643279842192, "grad_norm": 0.5557238459587097, "learning_rate": 4.349600000000001e-06, "loss": 0.0044, "step": 28253 }, { "epoch": 1.5063176414138721, "grad_norm": 1.6079964637756348, "learning_rate": 4.3494e-06, "loss": 0.0142, "step": 28254 }, { "epoch": 1.5063709548435251, "grad_norm": 1.1468867063522339, "learning_rate": 4.3492e-06, "loss": 0.0195, "step": 28255 }, { "epoch": 1.5064242682731779, "grad_norm": 2.3411693572998047, "learning_rate": 4.349e-06, "loss": 0.0122, "step": 28256 }, { "epoch": 1.506477581702831, "grad_norm": 1.309706211090088, "learning_rate": 4.348800000000001e-06, "loss": 0.0259, "step": 28257 }, { "epoch": 1.5065308951324838, "grad_norm": 1.0730152130126953, "learning_rate": 4.348600000000001e-06, "loss": 0.0057, "step": 28258 }, { "epoch": 1.5065842085621368, "grad_norm": 1.541675090789795, "learning_rate": 4.348400000000001e-06, "loss": 0.0266, "step": 28259 }, { "epoch": 1.5066375219917898, "grad_norm": 1.0590221881866455, "learning_rate": 4.3482e-06, "loss": 0.0123, "step": 28260 }, { "epoch": 1.5066908354214426, "grad_norm": 2.3516948223114014, "learning_rate": 4.3480000000000006e-06, "loss": 0.0235, "step": 28261 }, { "epoch": 1.5067441488510958, "grad_norm": 1.563705325126648, "learning_rate": 4.3478e-06, "loss": 0.0227, "step": 28262 }, { "epoch": 1.5067974622807485, "grad_norm": 0.8100070953369141, "learning_rate": 4.3476000000000005e-06, "loss": 0.0243, "step": 28263 }, { "epoch": 1.5068507757104015, "grad_norm": 2.1735951900482178, "learning_rate": 4.347400000000001e-06, "loss": 0.0249, "step": 28264 }, { "epoch": 1.5069040891400545, "grad_norm": 0.9739595055580139, "learning_rate": 4.3472000000000005e-06, "loss": 0.007, "step": 28265 }, { "epoch": 1.5069574025697072, "grad_norm": 1.0959954261779785, "learning_rate": 4.347e-06, "loss": 0.0257, "step": 28266 }, { "epoch": 1.5070107159993602, "grad_norm": 2.43296480178833, "learning_rate": 4.3468000000000005e-06, "loss": 0.0249, "step": 28267 }, { "epoch": 1.5070640294290132, "grad_norm": 1.750177264213562, "learning_rate": 4.3466e-06, "loss": 0.0211, "step": 28268 }, { "epoch": 1.507117342858666, "grad_norm": 2.019573211669922, "learning_rate": 4.3464000000000004e-06, "loss": 0.0229, "step": 28269 }, { "epoch": 1.5071706562883191, "grad_norm": 1.5767452716827393, "learning_rate": 4.346200000000001e-06, "loss": 0.0218, "step": 28270 }, { "epoch": 1.507223969717972, "grad_norm": 3.9847633838653564, "learning_rate": 4.346e-06, "loss": 0.0456, "step": 28271 }, { "epoch": 1.5072772831476249, "grad_norm": 2.540687084197998, "learning_rate": 4.345800000000001e-06, "loss": 0.0323, "step": 28272 }, { "epoch": 1.5073305965772779, "grad_norm": 0.9019609689712524, "learning_rate": 4.3456e-06, "loss": 0.0057, "step": 28273 }, { "epoch": 1.5073839100069306, "grad_norm": 2.3263213634490967, "learning_rate": 4.3454e-06, "loss": 0.0155, "step": 28274 }, { "epoch": 1.5074372234365838, "grad_norm": 1.1911730766296387, "learning_rate": 4.3452e-06, "loss": 0.0187, "step": 28275 }, { "epoch": 1.5074905368662366, "grad_norm": 2.6778080463409424, "learning_rate": 4.345000000000001e-06, "loss": 0.0192, "step": 28276 }, { "epoch": 1.5075438502958896, "grad_norm": 1.865474820137024, "learning_rate": 4.3448e-06, "loss": 0.0329, "step": 28277 }, { "epoch": 1.5075971637255425, "grad_norm": 2.305377721786499, "learning_rate": 4.344600000000001e-06, "loss": 0.0155, "step": 28278 }, { "epoch": 1.5076504771551953, "grad_norm": 1.0522807836532593, "learning_rate": 4.3444e-06, "loss": 0.0121, "step": 28279 }, { "epoch": 1.5077037905848483, "grad_norm": 3.191734552383423, "learning_rate": 4.3442e-06, "loss": 0.014, "step": 28280 }, { "epoch": 1.5077571040145012, "grad_norm": 2.7463059425354004, "learning_rate": 4.344e-06, "loss": 0.0279, "step": 28281 }, { "epoch": 1.5078104174441542, "grad_norm": 2.4374637603759766, "learning_rate": 4.343800000000001e-06, "loss": 0.0153, "step": 28282 }, { "epoch": 1.5078637308738072, "grad_norm": 1.6843883991241455, "learning_rate": 4.3436e-06, "loss": 0.0218, "step": 28283 }, { "epoch": 1.50791704430346, "grad_norm": 1.3316705226898193, "learning_rate": 4.343400000000001e-06, "loss": 0.0106, "step": 28284 }, { "epoch": 1.507970357733113, "grad_norm": 1.479071021080017, "learning_rate": 4.3432e-06, "loss": 0.0174, "step": 28285 }, { "epoch": 1.508023671162766, "grad_norm": 1.3900595903396606, "learning_rate": 4.343000000000001e-06, "loss": 0.0132, "step": 28286 }, { "epoch": 1.5080769845924187, "grad_norm": 2.4173812866210938, "learning_rate": 4.3428e-06, "loss": 0.0229, "step": 28287 }, { "epoch": 1.5081302980220719, "grad_norm": 2.354081392288208, "learning_rate": 4.3426000000000006e-06, "loss": 0.007, "step": 28288 }, { "epoch": 1.5081836114517246, "grad_norm": 1.3755298852920532, "learning_rate": 4.3424e-06, "loss": 0.0136, "step": 28289 }, { "epoch": 1.5082369248813776, "grad_norm": 2.670132637023926, "learning_rate": 4.3422000000000005e-06, "loss": 0.0254, "step": 28290 }, { "epoch": 1.5082902383110306, "grad_norm": 1.3627086877822876, "learning_rate": 4.342e-06, "loss": 0.0156, "step": 28291 }, { "epoch": 1.5083435517406834, "grad_norm": 1.6818205118179321, "learning_rate": 4.3418000000000005e-06, "loss": 0.0095, "step": 28292 }, { "epoch": 1.5083968651703366, "grad_norm": 0.7794238924980164, "learning_rate": 4.3416e-06, "loss": 0.017, "step": 28293 }, { "epoch": 1.5084501785999893, "grad_norm": 3.0558457374572754, "learning_rate": 4.3414000000000005e-06, "loss": 0.0467, "step": 28294 }, { "epoch": 1.5085034920296423, "grad_norm": 0.30379199981689453, "learning_rate": 4.3412e-06, "loss": 0.0032, "step": 28295 }, { "epoch": 1.5085568054592953, "grad_norm": 1.1653351783752441, "learning_rate": 4.3410000000000005e-06, "loss": 0.0224, "step": 28296 }, { "epoch": 1.508610118888948, "grad_norm": 1.3479996919631958, "learning_rate": 4.340800000000001e-06, "loss": 0.0095, "step": 28297 }, { "epoch": 1.508663432318601, "grad_norm": 2.2797884941101074, "learning_rate": 4.3406e-06, "loss": 0.0163, "step": 28298 }, { "epoch": 1.508716745748254, "grad_norm": 1.2817796468734741, "learning_rate": 4.3404e-06, "loss": 0.0106, "step": 28299 }, { "epoch": 1.508770059177907, "grad_norm": 2.036992311477661, "learning_rate": 4.3402e-06, "loss": 0.0242, "step": 28300 }, { "epoch": 1.50882337260756, "grad_norm": 1.8353897333145142, "learning_rate": 4.34e-06, "loss": 0.0145, "step": 28301 }, { "epoch": 1.5088766860372127, "grad_norm": 1.3999090194702148, "learning_rate": 4.3398e-06, "loss": 0.0116, "step": 28302 }, { "epoch": 1.5089299994668657, "grad_norm": 2.7569313049316406, "learning_rate": 4.339600000000001e-06, "loss": 0.02, "step": 28303 }, { "epoch": 1.5089833128965187, "grad_norm": 2.2114365100860596, "learning_rate": 4.3394e-06, "loss": 0.0124, "step": 28304 }, { "epoch": 1.5090366263261714, "grad_norm": 1.1602418422698975, "learning_rate": 4.3392e-06, "loss": 0.018, "step": 28305 }, { "epoch": 1.5090899397558246, "grad_norm": 1.1497299671173096, "learning_rate": 4.339e-06, "loss": 0.0181, "step": 28306 }, { "epoch": 1.5091432531854774, "grad_norm": 0.5877043008804321, "learning_rate": 4.3388e-06, "loss": 0.0041, "step": 28307 }, { "epoch": 1.5091965666151304, "grad_norm": 1.9918516874313354, "learning_rate": 4.3386e-06, "loss": 0.0285, "step": 28308 }, { "epoch": 1.5092498800447833, "grad_norm": 0.32938727736473083, "learning_rate": 4.338400000000001e-06, "loss": 0.0025, "step": 28309 }, { "epoch": 1.509303193474436, "grad_norm": 3.4156224727630615, "learning_rate": 4.3382e-06, "loss": 0.022, "step": 28310 }, { "epoch": 1.5093565069040893, "grad_norm": 1.980366826057434, "learning_rate": 4.338000000000001e-06, "loss": 0.0361, "step": 28311 }, { "epoch": 1.509409820333742, "grad_norm": 3.2080135345458984, "learning_rate": 4.3378e-06, "loss": 0.0104, "step": 28312 }, { "epoch": 1.509463133763395, "grad_norm": 3.1326706409454346, "learning_rate": 4.3376e-06, "loss": 0.0246, "step": 28313 }, { "epoch": 1.509516447193048, "grad_norm": 0.7743433713912964, "learning_rate": 4.3374e-06, "loss": 0.005, "step": 28314 }, { "epoch": 1.5095697606227008, "grad_norm": 2.1048200130462646, "learning_rate": 4.337200000000001e-06, "loss": 0.03, "step": 28315 }, { "epoch": 1.5096230740523537, "grad_norm": 3.3370306491851807, "learning_rate": 4.337e-06, "loss": 0.0136, "step": 28316 }, { "epoch": 1.5096763874820067, "grad_norm": 2.1574788093566895, "learning_rate": 4.3368000000000006e-06, "loss": 0.0292, "step": 28317 }, { "epoch": 1.5097297009116597, "grad_norm": 1.4279175996780396, "learning_rate": 4.3366e-06, "loss": 0.0065, "step": 28318 }, { "epoch": 1.5097830143413127, "grad_norm": 1.0999455451965332, "learning_rate": 4.3364000000000005e-06, "loss": 0.0159, "step": 28319 }, { "epoch": 1.5098363277709654, "grad_norm": 1.746520757675171, "learning_rate": 4.3362e-06, "loss": 0.0272, "step": 28320 }, { "epoch": 1.5098896412006184, "grad_norm": 0.46218612790107727, "learning_rate": 4.3360000000000005e-06, "loss": 0.003, "step": 28321 }, { "epoch": 1.5099429546302714, "grad_norm": 2.682281494140625, "learning_rate": 4.335800000000001e-06, "loss": 0.0229, "step": 28322 }, { "epoch": 1.5099962680599242, "grad_norm": 1.4174869060516357, "learning_rate": 4.3356000000000005e-06, "loss": 0.0228, "step": 28323 }, { "epoch": 1.5100495814895774, "grad_norm": 2.2248525619506836, "learning_rate": 4.3354e-06, "loss": 0.0202, "step": 28324 }, { "epoch": 1.51010289491923, "grad_norm": 1.9830043315887451, "learning_rate": 4.3352000000000004e-06, "loss": 0.017, "step": 28325 }, { "epoch": 1.510156208348883, "grad_norm": 2.249159097671509, "learning_rate": 4.335e-06, "loss": 0.0116, "step": 28326 }, { "epoch": 1.510209521778536, "grad_norm": 0.7013457417488098, "learning_rate": 4.3348e-06, "loss": 0.0053, "step": 28327 }, { "epoch": 1.5102628352081888, "grad_norm": 3.537189245223999, "learning_rate": 4.334600000000001e-06, "loss": 0.0406, "step": 28328 }, { "epoch": 1.510316148637842, "grad_norm": 2.666750907897949, "learning_rate": 4.3344e-06, "loss": 0.0316, "step": 28329 }, { "epoch": 1.5103694620674948, "grad_norm": 1.7422162294387817, "learning_rate": 4.334200000000001e-06, "loss": 0.0108, "step": 28330 }, { "epoch": 1.5104227754971478, "grad_norm": 1.4975017309188843, "learning_rate": 4.334e-06, "loss": 0.0145, "step": 28331 }, { "epoch": 1.5104760889268007, "grad_norm": 1.8791567087173462, "learning_rate": 4.3338e-06, "loss": 0.0246, "step": 28332 }, { "epoch": 1.5105294023564535, "grad_norm": 1.1117472648620605, "learning_rate": 4.3336e-06, "loss": 0.0071, "step": 28333 }, { "epoch": 1.5105827157861065, "grad_norm": 1.6066954135894775, "learning_rate": 4.333400000000001e-06, "loss": 0.0147, "step": 28334 }, { "epoch": 1.5106360292157595, "grad_norm": 2.0958945751190186, "learning_rate": 4.3332e-06, "loss": 0.0156, "step": 28335 }, { "epoch": 1.5106893426454122, "grad_norm": 2.2346866130828857, "learning_rate": 4.333000000000001e-06, "loss": 0.029, "step": 28336 }, { "epoch": 1.5107426560750654, "grad_norm": 4.365134239196777, "learning_rate": 4.3328e-06, "loss": 0.0515, "step": 28337 }, { "epoch": 1.5107959695047182, "grad_norm": 1.8071476221084595, "learning_rate": 4.3326e-06, "loss": 0.0106, "step": 28338 }, { "epoch": 1.5108492829343712, "grad_norm": 1.4475271701812744, "learning_rate": 4.3324e-06, "loss": 0.0218, "step": 28339 }, { "epoch": 1.5109025963640241, "grad_norm": 1.9524506330490112, "learning_rate": 4.332200000000001e-06, "loss": 0.0157, "step": 28340 }, { "epoch": 1.5109559097936769, "grad_norm": 0.8491844534873962, "learning_rate": 4.332e-06, "loss": 0.0051, "step": 28341 }, { "epoch": 1.51100922322333, "grad_norm": 1.4183028936386108, "learning_rate": 4.331800000000001e-06, "loss": 0.0218, "step": 28342 }, { "epoch": 1.5110625366529828, "grad_norm": 0.8074012398719788, "learning_rate": 4.3316e-06, "loss": 0.0043, "step": 28343 }, { "epoch": 1.5111158500826358, "grad_norm": 2.228858232498169, "learning_rate": 4.3314000000000006e-06, "loss": 0.0367, "step": 28344 }, { "epoch": 1.5111691635122888, "grad_norm": 2.072748899459839, "learning_rate": 4.3312e-06, "loss": 0.0188, "step": 28345 }, { "epoch": 1.5112224769419416, "grad_norm": 1.713228702545166, "learning_rate": 4.3310000000000005e-06, "loss": 0.0499, "step": 28346 }, { "epoch": 1.5112757903715948, "grad_norm": 1.1989281177520752, "learning_rate": 4.330800000000001e-06, "loss": 0.0104, "step": 28347 }, { "epoch": 1.5113291038012475, "grad_norm": 2.473827362060547, "learning_rate": 4.3306000000000005e-06, "loss": 0.0445, "step": 28348 }, { "epoch": 1.5113824172309005, "grad_norm": 0.8028651475906372, "learning_rate": 4.3304e-06, "loss": 0.0047, "step": 28349 }, { "epoch": 1.5114357306605535, "grad_norm": 2.6018457412719727, "learning_rate": 4.3302000000000005e-06, "loss": 0.0195, "step": 28350 }, { "epoch": 1.5114890440902062, "grad_norm": 2.442439317703247, "learning_rate": 4.33e-06, "loss": 0.0472, "step": 28351 }, { "epoch": 1.5115423575198592, "grad_norm": 1.5013724565505981, "learning_rate": 4.3298000000000005e-06, "loss": 0.0172, "step": 28352 }, { "epoch": 1.5115956709495122, "grad_norm": 2.374110698699951, "learning_rate": 4.329600000000001e-06, "loss": 0.011, "step": 28353 }, { "epoch": 1.511648984379165, "grad_norm": 1.3199000358581543, "learning_rate": 4.3294000000000004e-06, "loss": 0.0121, "step": 28354 }, { "epoch": 1.5117022978088182, "grad_norm": 1.4384679794311523, "learning_rate": 4.329200000000001e-06, "loss": 0.0173, "step": 28355 }, { "epoch": 1.511755611238471, "grad_norm": 1.020296335220337, "learning_rate": 4.329e-06, "loss": 0.0056, "step": 28356 }, { "epoch": 1.5118089246681239, "grad_norm": 5.7003889083862305, "learning_rate": 4.3288e-06, "loss": 0.0212, "step": 28357 }, { "epoch": 1.5118622380977769, "grad_norm": 1.76059889793396, "learning_rate": 4.3286e-06, "loss": 0.0103, "step": 28358 }, { "epoch": 1.5119155515274296, "grad_norm": 0.6314737796783447, "learning_rate": 4.328400000000001e-06, "loss": 0.0146, "step": 28359 }, { "epoch": 1.5119688649570828, "grad_norm": 1.1307803392410278, "learning_rate": 4.3282e-06, "loss": 0.0067, "step": 28360 }, { "epoch": 1.5120221783867356, "grad_norm": 3.4026718139648438, "learning_rate": 4.328000000000001e-06, "loss": 0.0515, "step": 28361 }, { "epoch": 1.5120754918163886, "grad_norm": 0.4515669047832489, "learning_rate": 4.3278e-06, "loss": 0.0037, "step": 28362 }, { "epoch": 1.5121288052460415, "grad_norm": 1.1190639734268188, "learning_rate": 4.3276e-06, "loss": 0.0053, "step": 28363 }, { "epoch": 1.5121821186756943, "grad_norm": 1.0916221141815186, "learning_rate": 4.3274e-06, "loss": 0.0092, "step": 28364 }, { "epoch": 1.5122354321053475, "grad_norm": 1.375748634338379, "learning_rate": 4.327200000000001e-06, "loss": 0.0082, "step": 28365 }, { "epoch": 1.5122887455350003, "grad_norm": 3.16251540184021, "learning_rate": 4.327e-06, "loss": 0.0369, "step": 28366 }, { "epoch": 1.5123420589646532, "grad_norm": 1.5241608619689941, "learning_rate": 4.326800000000001e-06, "loss": 0.022, "step": 28367 }, { "epoch": 1.5123953723943062, "grad_norm": 1.225693702697754, "learning_rate": 4.3266e-06, "loss": 0.0087, "step": 28368 }, { "epoch": 1.512448685823959, "grad_norm": 1.1104176044464111, "learning_rate": 4.326400000000001e-06, "loss": 0.0237, "step": 28369 }, { "epoch": 1.512501999253612, "grad_norm": 0.9499928951263428, "learning_rate": 4.3262e-06, "loss": 0.0058, "step": 28370 }, { "epoch": 1.512555312683265, "grad_norm": 1.763863444328308, "learning_rate": 4.326000000000001e-06, "loss": 0.0163, "step": 28371 }, { "epoch": 1.5126086261129177, "grad_norm": 2.0896475315093994, "learning_rate": 4.3258e-06, "loss": 0.0205, "step": 28372 }, { "epoch": 1.5126619395425709, "grad_norm": 1.9420921802520752, "learning_rate": 4.3256000000000006e-06, "loss": 0.0195, "step": 28373 }, { "epoch": 1.5127152529722236, "grad_norm": 1.7242765426635742, "learning_rate": 4.3254e-06, "loss": 0.0258, "step": 28374 }, { "epoch": 1.5127685664018766, "grad_norm": 1.0962483882904053, "learning_rate": 4.3252000000000005e-06, "loss": 0.0201, "step": 28375 }, { "epoch": 1.5128218798315296, "grad_norm": 1.252663254737854, "learning_rate": 4.325e-06, "loss": 0.0093, "step": 28376 }, { "epoch": 1.5128751932611824, "grad_norm": 1.8020066022872925, "learning_rate": 4.3248000000000005e-06, "loss": 0.0183, "step": 28377 }, { "epoch": 1.5129285066908356, "grad_norm": 4.749237060546875, "learning_rate": 4.3246e-06, "loss": 0.0332, "step": 28378 }, { "epoch": 1.5129818201204883, "grad_norm": 0.9137521982192993, "learning_rate": 4.3244000000000005e-06, "loss": 0.0128, "step": 28379 }, { "epoch": 1.5130351335501413, "grad_norm": 1.6163649559020996, "learning_rate": 4.324200000000001e-06, "loss": 0.0096, "step": 28380 }, { "epoch": 1.5130884469797943, "grad_norm": 3.092149019241333, "learning_rate": 4.3240000000000004e-06, "loss": 0.0414, "step": 28381 }, { "epoch": 1.513141760409447, "grad_norm": 1.1550824642181396, "learning_rate": 4.3238e-06, "loss": 0.0086, "step": 28382 }, { "epoch": 1.5131950738391002, "grad_norm": 0.9096105694770813, "learning_rate": 4.3236e-06, "loss": 0.019, "step": 28383 }, { "epoch": 1.513248387268753, "grad_norm": 1.4442830085754395, "learning_rate": 4.3234e-06, "loss": 0.0269, "step": 28384 }, { "epoch": 1.513301700698406, "grad_norm": 1.8843919038772583, "learning_rate": 4.3232e-06, "loss": 0.0285, "step": 28385 }, { "epoch": 1.513355014128059, "grad_norm": 1.9432289600372314, "learning_rate": 4.323000000000001e-06, "loss": 0.0257, "step": 28386 }, { "epoch": 1.5134083275577117, "grad_norm": 1.6494078636169434, "learning_rate": 4.3228e-06, "loss": 0.0278, "step": 28387 }, { "epoch": 1.5134616409873647, "grad_norm": 2.0334200859069824, "learning_rate": 4.3226e-06, "loss": 0.0239, "step": 28388 }, { "epoch": 1.5135149544170177, "grad_norm": 2.4865589141845703, "learning_rate": 4.3224e-06, "loss": 0.0141, "step": 28389 }, { "epoch": 1.5135682678466704, "grad_norm": 1.785909652709961, "learning_rate": 4.3222e-06, "loss": 0.0142, "step": 28390 }, { "epoch": 1.5136215812763236, "grad_norm": 1.2506146430969238, "learning_rate": 4.322e-06, "loss": 0.0123, "step": 28391 }, { "epoch": 1.5136748947059764, "grad_norm": 3.2374188899993896, "learning_rate": 4.321800000000001e-06, "loss": 0.0224, "step": 28392 }, { "epoch": 1.5137282081356294, "grad_norm": 2.075822591781616, "learning_rate": 4.3216e-06, "loss": 0.0265, "step": 28393 }, { "epoch": 1.5137815215652823, "grad_norm": 1.3700840473175049, "learning_rate": 4.321400000000001e-06, "loss": 0.0131, "step": 28394 }, { "epoch": 1.513834834994935, "grad_norm": 2.007908821105957, "learning_rate": 4.3212e-06, "loss": 0.0319, "step": 28395 }, { "epoch": 1.5138881484245883, "grad_norm": 2.431684732437134, "learning_rate": 4.321e-06, "loss": 0.0287, "step": 28396 }, { "epoch": 1.513941461854241, "grad_norm": 1.7198330163955688, "learning_rate": 4.3208e-06, "loss": 0.0138, "step": 28397 }, { "epoch": 1.513994775283894, "grad_norm": 1.26475191116333, "learning_rate": 4.320600000000001e-06, "loss": 0.0087, "step": 28398 }, { "epoch": 1.514048088713547, "grad_norm": 1.094612956047058, "learning_rate": 4.3204e-06, "loss": 0.0083, "step": 28399 }, { "epoch": 1.5141014021431998, "grad_norm": 3.0957283973693848, "learning_rate": 4.320200000000001e-06, "loss": 0.0361, "step": 28400 }, { "epoch": 1.5141547155728527, "grad_norm": 0.5804803371429443, "learning_rate": 4.32e-06, "loss": 0.0049, "step": 28401 }, { "epoch": 1.5142080290025057, "grad_norm": 1.4205602407455444, "learning_rate": 4.3198e-06, "loss": 0.0151, "step": 28402 }, { "epoch": 1.5142613424321587, "grad_norm": 1.4958336353302002, "learning_rate": 4.3196e-06, "loss": 0.0181, "step": 28403 }, { "epoch": 1.5143146558618117, "grad_norm": 1.8373725414276123, "learning_rate": 4.3194000000000005e-06, "loss": 0.0189, "step": 28404 }, { "epoch": 1.5143679692914644, "grad_norm": 1.4506711959838867, "learning_rate": 4.319200000000001e-06, "loss": 0.0235, "step": 28405 }, { "epoch": 1.5144212827211174, "grad_norm": 3.8859951496124268, "learning_rate": 4.3190000000000005e-06, "loss": 0.0498, "step": 28406 }, { "epoch": 1.5144745961507704, "grad_norm": 1.347967267036438, "learning_rate": 4.3188e-06, "loss": 0.0095, "step": 28407 }, { "epoch": 1.5145279095804232, "grad_norm": 2.8036677837371826, "learning_rate": 4.3186000000000005e-06, "loss": 0.025, "step": 28408 }, { "epoch": 1.5145812230100764, "grad_norm": 0.9537436962127686, "learning_rate": 4.3184e-06, "loss": 0.0073, "step": 28409 }, { "epoch": 1.5146345364397291, "grad_norm": 2.171858549118042, "learning_rate": 4.3182000000000004e-06, "loss": 0.0143, "step": 28410 }, { "epoch": 1.514687849869382, "grad_norm": 1.5172193050384521, "learning_rate": 4.318000000000001e-06, "loss": 0.0295, "step": 28411 }, { "epoch": 1.514741163299035, "grad_norm": 1.6059197187423706, "learning_rate": 4.3178e-06, "loss": 0.0107, "step": 28412 }, { "epoch": 1.5147944767286878, "grad_norm": 1.8527214527130127, "learning_rate": 4.3176e-06, "loss": 0.0197, "step": 28413 }, { "epoch": 1.514847790158341, "grad_norm": 0.8014809489250183, "learning_rate": 4.3174e-06, "loss": 0.006, "step": 28414 }, { "epoch": 1.5149011035879938, "grad_norm": 1.3468369245529175, "learning_rate": 4.3172e-06, "loss": 0.0211, "step": 28415 }, { "epoch": 1.5149544170176468, "grad_norm": 0.88497394323349, "learning_rate": 4.317e-06, "loss": 0.0188, "step": 28416 }, { "epoch": 1.5150077304472997, "grad_norm": 0.41212400794029236, "learning_rate": 4.316800000000001e-06, "loss": 0.0033, "step": 28417 }, { "epoch": 1.5150610438769525, "grad_norm": 2.331056594848633, "learning_rate": 4.3166e-06, "loss": 0.0203, "step": 28418 }, { "epoch": 1.5151143573066055, "grad_norm": 0.4018782079219818, "learning_rate": 4.316400000000001e-06, "loss": 0.0023, "step": 28419 }, { "epoch": 1.5151676707362585, "grad_norm": 3.0014491081237793, "learning_rate": 4.3162e-06, "loss": 0.0559, "step": 28420 }, { "epoch": 1.5152209841659114, "grad_norm": 1.0892492532730103, "learning_rate": 4.316e-06, "loss": 0.0079, "step": 28421 }, { "epoch": 1.5152742975955644, "grad_norm": 1.5944275856018066, "learning_rate": 4.3158e-06, "loss": 0.0139, "step": 28422 }, { "epoch": 1.5153276110252172, "grad_norm": 1.2390480041503906, "learning_rate": 4.315600000000001e-06, "loss": 0.0106, "step": 28423 }, { "epoch": 1.5153809244548702, "grad_norm": 2.6119048595428467, "learning_rate": 4.3154e-06, "loss": 0.0196, "step": 28424 }, { "epoch": 1.5154342378845231, "grad_norm": 1.926694631576538, "learning_rate": 4.315200000000001e-06, "loss": 0.0216, "step": 28425 }, { "epoch": 1.515487551314176, "grad_norm": 2.161612033843994, "learning_rate": 4.315e-06, "loss": 0.041, "step": 28426 }, { "epoch": 1.515540864743829, "grad_norm": 2.996432304382324, "learning_rate": 4.3148e-06, "loss": 0.0195, "step": 28427 }, { "epoch": 1.5155941781734819, "grad_norm": 1.9672496318817139, "learning_rate": 4.3146e-06, "loss": 0.0075, "step": 28428 }, { "epoch": 1.5156474916031348, "grad_norm": 2.703667402267456, "learning_rate": 4.3144000000000006e-06, "loss": 0.0275, "step": 28429 }, { "epoch": 1.5157008050327878, "grad_norm": 1.5325316190719604, "learning_rate": 4.314200000000001e-06, "loss": 0.0254, "step": 28430 }, { "epoch": 1.5157541184624406, "grad_norm": 0.39157912135124207, "learning_rate": 4.3140000000000005e-06, "loss": 0.003, "step": 28431 }, { "epoch": 1.5158074318920938, "grad_norm": 0.7873908281326294, "learning_rate": 4.3138e-06, "loss": 0.0052, "step": 28432 }, { "epoch": 1.5158607453217465, "grad_norm": 1.3602768182754517, "learning_rate": 4.3136000000000005e-06, "loss": 0.0091, "step": 28433 }, { "epoch": 1.5159140587513995, "grad_norm": 1.8565938472747803, "learning_rate": 4.3134e-06, "loss": 0.0073, "step": 28434 }, { "epoch": 1.5159673721810525, "grad_norm": 2.463031530380249, "learning_rate": 4.3132000000000005e-06, "loss": 0.0479, "step": 28435 }, { "epoch": 1.5160206856107052, "grad_norm": 0.9133405685424805, "learning_rate": 4.313000000000001e-06, "loss": 0.0039, "step": 28436 }, { "epoch": 1.5160739990403582, "grad_norm": 2.312150478363037, "learning_rate": 4.3128000000000004e-06, "loss": 0.0304, "step": 28437 }, { "epoch": 1.5161273124700112, "grad_norm": 1.372363567352295, "learning_rate": 4.3126e-06, "loss": 0.0097, "step": 28438 }, { "epoch": 1.5161806258996642, "grad_norm": 0.4367942810058594, "learning_rate": 4.3124e-06, "loss": 0.0033, "step": 28439 }, { "epoch": 1.5162339393293172, "grad_norm": 2.2244956493377686, "learning_rate": 4.3122e-06, "loss": 0.0103, "step": 28440 }, { "epoch": 1.51628725275897, "grad_norm": 1.7980148792266846, "learning_rate": 4.312e-06, "loss": 0.0321, "step": 28441 }, { "epoch": 1.516340566188623, "grad_norm": 2.685992956161499, "learning_rate": 4.311800000000001e-06, "loss": 0.0466, "step": 28442 }, { "epoch": 1.5163938796182759, "grad_norm": 1.5760672092437744, "learning_rate": 4.3116e-06, "loss": 0.0159, "step": 28443 }, { "epoch": 1.5164471930479286, "grad_norm": 1.6270593404769897, "learning_rate": 4.311400000000001e-06, "loss": 0.0184, "step": 28444 }, { "epoch": 1.5165005064775818, "grad_norm": 2.5919113159179688, "learning_rate": 4.3112e-06, "loss": 0.0374, "step": 28445 }, { "epoch": 1.5165538199072346, "grad_norm": 1.2589385509490967, "learning_rate": 4.311e-06, "loss": 0.0123, "step": 28446 }, { "epoch": 1.5166071333368876, "grad_norm": 0.8866077065467834, "learning_rate": 4.3108e-06, "loss": 0.0077, "step": 28447 }, { "epoch": 1.5166604467665405, "grad_norm": 1.7717267274856567, "learning_rate": 4.310600000000001e-06, "loss": 0.0155, "step": 28448 }, { "epoch": 1.5167137601961933, "grad_norm": 2.399738311767578, "learning_rate": 4.3104e-06, "loss": 0.0508, "step": 28449 }, { "epoch": 1.5167670736258465, "grad_norm": 1.7357333898544312, "learning_rate": 4.310200000000001e-06, "loss": 0.0147, "step": 28450 }, { "epoch": 1.5168203870554993, "grad_norm": 1.494524359703064, "learning_rate": 4.31e-06, "loss": 0.0083, "step": 28451 }, { "epoch": 1.5168737004851522, "grad_norm": 2.656952142715454, "learning_rate": 4.3098e-06, "loss": 0.027, "step": 28452 }, { "epoch": 1.5169270139148052, "grad_norm": 3.970062732696533, "learning_rate": 4.3096e-06, "loss": 0.0305, "step": 28453 }, { "epoch": 1.516980327344458, "grad_norm": 2.637199640274048, "learning_rate": 4.309400000000001e-06, "loss": 0.0255, "step": 28454 }, { "epoch": 1.517033640774111, "grad_norm": 3.6630077362060547, "learning_rate": 4.309200000000001e-06, "loss": 0.0196, "step": 28455 }, { "epoch": 1.517086954203764, "grad_norm": 0.5402437448501587, "learning_rate": 4.309000000000001e-06, "loss": 0.0045, "step": 28456 }, { "epoch": 1.5171402676334167, "grad_norm": 0.7023674845695496, "learning_rate": 4.3088e-06, "loss": 0.0044, "step": 28457 }, { "epoch": 1.51719358106307, "grad_norm": 0.7901784181594849, "learning_rate": 4.3086000000000006e-06, "loss": 0.016, "step": 28458 }, { "epoch": 1.5172468944927227, "grad_norm": 3.644599437713623, "learning_rate": 4.3084e-06, "loss": 0.0183, "step": 28459 }, { "epoch": 1.5173002079223756, "grad_norm": 0.8401057720184326, "learning_rate": 4.3082000000000005e-06, "loss": 0.0058, "step": 28460 }, { "epoch": 1.5173535213520286, "grad_norm": 1.0304462909698486, "learning_rate": 4.308000000000001e-06, "loss": 0.009, "step": 28461 }, { "epoch": 1.5174068347816814, "grad_norm": 0.6297824382781982, "learning_rate": 4.3078000000000005e-06, "loss": 0.0051, "step": 28462 }, { "epoch": 1.5174601482113346, "grad_norm": 2.6637630462646484, "learning_rate": 4.3076e-06, "loss": 0.0243, "step": 28463 }, { "epoch": 1.5175134616409873, "grad_norm": 2.487240791320801, "learning_rate": 4.3074000000000005e-06, "loss": 0.0337, "step": 28464 }, { "epoch": 1.5175667750706403, "grad_norm": 2.7207727432250977, "learning_rate": 4.3072e-06, "loss": 0.0366, "step": 28465 }, { "epoch": 1.5176200885002933, "grad_norm": 2.4599967002868652, "learning_rate": 4.3070000000000004e-06, "loss": 0.0304, "step": 28466 }, { "epoch": 1.517673401929946, "grad_norm": 2.4249274730682373, "learning_rate": 4.306800000000001e-06, "loss": 0.0284, "step": 28467 }, { "epoch": 1.5177267153595992, "grad_norm": 2.8921167850494385, "learning_rate": 4.3066e-06, "loss": 0.0451, "step": 28468 }, { "epoch": 1.517780028789252, "grad_norm": 0.8578444719314575, "learning_rate": 4.306400000000001e-06, "loss": 0.0053, "step": 28469 }, { "epoch": 1.517833342218905, "grad_norm": 1.0342453718185425, "learning_rate": 4.3062e-06, "loss": 0.0058, "step": 28470 }, { "epoch": 1.517886655648558, "grad_norm": 2.5362911224365234, "learning_rate": 4.306e-06, "loss": 0.0166, "step": 28471 }, { "epoch": 1.5179399690782107, "grad_norm": 1.5335925817489624, "learning_rate": 4.3058e-06, "loss": 0.0119, "step": 28472 }, { "epoch": 1.5179932825078637, "grad_norm": 1.084405779838562, "learning_rate": 4.305600000000001e-06, "loss": 0.0063, "step": 28473 }, { "epoch": 1.5180465959375167, "grad_norm": 1.5865795612335205, "learning_rate": 4.3054e-06, "loss": 0.0315, "step": 28474 }, { "epoch": 1.5180999093671694, "grad_norm": 0.644709587097168, "learning_rate": 4.305200000000001e-06, "loss": 0.0044, "step": 28475 }, { "epoch": 1.5181532227968226, "grad_norm": 2.4258017539978027, "learning_rate": 4.305e-06, "loss": 0.0152, "step": 28476 }, { "epoch": 1.5182065362264754, "grad_norm": 1.6332584619522095, "learning_rate": 4.3048e-06, "loss": 0.0168, "step": 28477 }, { "epoch": 1.5182598496561284, "grad_norm": 5.739921569824219, "learning_rate": 4.3046e-06, "loss": 0.042, "step": 28478 }, { "epoch": 1.5183131630857813, "grad_norm": 1.435233473777771, "learning_rate": 4.304400000000001e-06, "loss": 0.0321, "step": 28479 }, { "epoch": 1.518366476515434, "grad_norm": 1.6694908142089844, "learning_rate": 4.3042e-06, "loss": 0.0227, "step": 28480 }, { "epoch": 1.5184197899450873, "grad_norm": 1.8475303649902344, "learning_rate": 4.304000000000001e-06, "loss": 0.0273, "step": 28481 }, { "epoch": 1.51847310337474, "grad_norm": 3.0495119094848633, "learning_rate": 4.3038e-06, "loss": 0.0579, "step": 28482 }, { "epoch": 1.518526416804393, "grad_norm": 1.266148567199707, "learning_rate": 4.303600000000001e-06, "loss": 0.0137, "step": 28483 }, { "epoch": 1.518579730234046, "grad_norm": 1.3873305320739746, "learning_rate": 4.3034e-06, "loss": 0.0093, "step": 28484 }, { "epoch": 1.5186330436636988, "grad_norm": 0.41809409856796265, "learning_rate": 4.3032000000000006e-06, "loss": 0.004, "step": 28485 }, { "epoch": 1.518686357093352, "grad_norm": 2.717515230178833, "learning_rate": 4.303e-06, "loss": 0.0169, "step": 28486 }, { "epoch": 1.5187396705230047, "grad_norm": 3.4816501140594482, "learning_rate": 4.3028000000000005e-06, "loss": 0.0583, "step": 28487 }, { "epoch": 1.5187929839526577, "grad_norm": 1.797716736793518, "learning_rate": 4.3026e-06, "loss": 0.0207, "step": 28488 }, { "epoch": 1.5188462973823107, "grad_norm": 2.0233421325683594, "learning_rate": 4.3024000000000005e-06, "loss": 0.0372, "step": 28489 }, { "epoch": 1.5188996108119635, "grad_norm": 1.7292261123657227, "learning_rate": 4.3022e-06, "loss": 0.0207, "step": 28490 }, { "epoch": 1.5189529242416164, "grad_norm": 2.075568437576294, "learning_rate": 4.3020000000000005e-06, "loss": 0.0271, "step": 28491 }, { "epoch": 1.5190062376712694, "grad_norm": 1.3908637762069702, "learning_rate": 4.3018e-06, "loss": 0.0132, "step": 28492 }, { "epoch": 1.5190595511009222, "grad_norm": 2.677499771118164, "learning_rate": 4.3016000000000005e-06, "loss": 0.0217, "step": 28493 }, { "epoch": 1.5191128645305754, "grad_norm": 3.541012763977051, "learning_rate": 4.301400000000001e-06, "loss": 0.0637, "step": 28494 }, { "epoch": 1.5191661779602281, "grad_norm": 2.702101945877075, "learning_rate": 4.3012000000000004e-06, "loss": 0.0149, "step": 28495 }, { "epoch": 1.519219491389881, "grad_norm": 2.260021686553955, "learning_rate": 4.301e-06, "loss": 0.0156, "step": 28496 }, { "epoch": 1.519272804819534, "grad_norm": 1.628650188446045, "learning_rate": 4.3008e-06, "loss": 0.0258, "step": 28497 }, { "epoch": 1.5193261182491868, "grad_norm": 0.86063551902771, "learning_rate": 4.3006e-06, "loss": 0.0058, "step": 28498 }, { "epoch": 1.51937943167884, "grad_norm": 1.3858169317245483, "learning_rate": 4.3004e-06, "loss": 0.0099, "step": 28499 }, { "epoch": 1.5194327451084928, "grad_norm": 2.164635419845581, "learning_rate": 4.300200000000001e-06, "loss": 0.0172, "step": 28500 }, { "epoch": 1.5194860585381458, "grad_norm": 0.8484154939651489, "learning_rate": 4.3e-06, "loss": 0.0055, "step": 28501 }, { "epoch": 1.5195393719677988, "grad_norm": 0.7257756590843201, "learning_rate": 4.2998e-06, "loss": 0.0068, "step": 28502 }, { "epoch": 1.5195926853974515, "grad_norm": 1.5056159496307373, "learning_rate": 4.2996e-06, "loss": 0.0192, "step": 28503 }, { "epoch": 1.5196459988271045, "grad_norm": 1.513461947441101, "learning_rate": 4.2994e-06, "loss": 0.0156, "step": 28504 }, { "epoch": 1.5196993122567575, "grad_norm": 0.3808887302875519, "learning_rate": 4.2992e-06, "loss": 0.0033, "step": 28505 }, { "epoch": 1.5197526256864105, "grad_norm": 1.3726191520690918, "learning_rate": 4.299000000000001e-06, "loss": 0.0355, "step": 28506 }, { "epoch": 1.5198059391160634, "grad_norm": 2.6469271183013916, "learning_rate": 4.2988e-06, "loss": 0.0588, "step": 28507 }, { "epoch": 1.5198592525457162, "grad_norm": 1.89893639087677, "learning_rate": 4.298600000000001e-06, "loss": 0.0143, "step": 28508 }, { "epoch": 1.5199125659753692, "grad_norm": 0.8426436185836792, "learning_rate": 4.2984e-06, "loss": 0.0069, "step": 28509 }, { "epoch": 1.5199658794050221, "grad_norm": 2.4552242755889893, "learning_rate": 4.2982e-06, "loss": 0.0281, "step": 28510 }, { "epoch": 1.520019192834675, "grad_norm": 0.8963773846626282, "learning_rate": 4.298e-06, "loss": 0.0148, "step": 28511 }, { "epoch": 1.520072506264328, "grad_norm": 1.6705750226974487, "learning_rate": 4.297800000000001e-06, "loss": 0.0188, "step": 28512 }, { "epoch": 1.5201258196939809, "grad_norm": 1.2690845727920532, "learning_rate": 4.2976e-06, "loss": 0.01, "step": 28513 }, { "epoch": 1.5201791331236338, "grad_norm": 1.3112469911575317, "learning_rate": 4.2974000000000006e-06, "loss": 0.007, "step": 28514 }, { "epoch": 1.5202324465532868, "grad_norm": 1.783621072769165, "learning_rate": 4.2972e-06, "loss": 0.0353, "step": 28515 }, { "epoch": 1.5202857599829396, "grad_norm": 1.2845104932785034, "learning_rate": 4.2970000000000005e-06, "loss": 0.017, "step": 28516 }, { "epoch": 1.5203390734125928, "grad_norm": 1.5278332233428955, "learning_rate": 4.2968e-06, "loss": 0.0181, "step": 28517 }, { "epoch": 1.5203923868422455, "grad_norm": 2.5787806510925293, "learning_rate": 4.2966000000000005e-06, "loss": 0.0314, "step": 28518 }, { "epoch": 1.5204457002718985, "grad_norm": 1.0010398626327515, "learning_rate": 4.296400000000001e-06, "loss": 0.0101, "step": 28519 }, { "epoch": 1.5204990137015515, "grad_norm": 3.0320701599121094, "learning_rate": 4.2962000000000005e-06, "loss": 0.0394, "step": 28520 }, { "epoch": 1.5205523271312043, "grad_norm": 1.9684394598007202, "learning_rate": 4.296e-06, "loss": 0.0329, "step": 28521 }, { "epoch": 1.5206056405608572, "grad_norm": 3.2489688396453857, "learning_rate": 4.2958000000000004e-06, "loss": 0.0279, "step": 28522 }, { "epoch": 1.5206589539905102, "grad_norm": 1.2688016891479492, "learning_rate": 4.2956e-06, "loss": 0.0321, "step": 28523 }, { "epoch": 1.5207122674201632, "grad_norm": 1.8357995748519897, "learning_rate": 4.2954e-06, "loss": 0.0258, "step": 28524 }, { "epoch": 1.5207655808498162, "grad_norm": 0.9146616458892822, "learning_rate": 4.295200000000001e-06, "loss": 0.0126, "step": 28525 }, { "epoch": 1.520818894279469, "grad_norm": 1.3313803672790527, "learning_rate": 4.295e-06, "loss": 0.0115, "step": 28526 }, { "epoch": 1.520872207709122, "grad_norm": 4.038977146148682, "learning_rate": 4.2948e-06, "loss": 0.051, "step": 28527 }, { "epoch": 1.5209255211387749, "grad_norm": 1.6916203498840332, "learning_rate": 4.2946e-06, "loss": 0.0082, "step": 28528 }, { "epoch": 1.5209788345684276, "grad_norm": 2.2136852741241455, "learning_rate": 4.2944e-06, "loss": 0.0278, "step": 28529 }, { "epoch": 1.5210321479980808, "grad_norm": 1.4192136526107788, "learning_rate": 4.2942e-06, "loss": 0.0133, "step": 28530 }, { "epoch": 1.5210854614277336, "grad_norm": 3.042607307434082, "learning_rate": 4.294000000000001e-06, "loss": 0.0297, "step": 28531 }, { "epoch": 1.5211387748573866, "grad_norm": 1.057134747505188, "learning_rate": 4.2938e-06, "loss": 0.0168, "step": 28532 }, { "epoch": 1.5211920882870396, "grad_norm": 2.587616205215454, "learning_rate": 4.293600000000001e-06, "loss": 0.0275, "step": 28533 }, { "epoch": 1.5212454017166923, "grad_norm": 2.6283278465270996, "learning_rate": 4.2934e-06, "loss": 0.019, "step": 28534 }, { "epoch": 1.5212987151463455, "grad_norm": 2.6664719581604004, "learning_rate": 4.2932e-06, "loss": 0.0262, "step": 28535 }, { "epoch": 1.5213520285759983, "grad_norm": 2.2042253017425537, "learning_rate": 4.293e-06, "loss": 0.0356, "step": 28536 }, { "epoch": 1.5214053420056513, "grad_norm": 1.498703122138977, "learning_rate": 4.292800000000001e-06, "loss": 0.0069, "step": 28537 }, { "epoch": 1.5214586554353042, "grad_norm": 2.7058165073394775, "learning_rate": 4.2926e-06, "loss": 0.0393, "step": 28538 }, { "epoch": 1.521511968864957, "grad_norm": 1.3230677843093872, "learning_rate": 4.292400000000001e-06, "loss": 0.0184, "step": 28539 }, { "epoch": 1.52156528229461, "grad_norm": 1.98709237575531, "learning_rate": 4.2922e-06, "loss": 0.016, "step": 28540 }, { "epoch": 1.521618595724263, "grad_norm": 2.923912763595581, "learning_rate": 4.292000000000001e-06, "loss": 0.026, "step": 28541 }, { "epoch": 1.521671909153916, "grad_norm": 0.6877889633178711, "learning_rate": 4.2918e-06, "loss": 0.0064, "step": 28542 }, { "epoch": 1.521725222583569, "grad_norm": 6.084771633148193, "learning_rate": 4.2916000000000006e-06, "loss": 0.0458, "step": 28543 }, { "epoch": 1.5217785360132217, "grad_norm": 1.2113817930221558, "learning_rate": 4.291400000000001e-06, "loss": 0.0084, "step": 28544 }, { "epoch": 1.5218318494428746, "grad_norm": 3.310187578201294, "learning_rate": 4.2912000000000005e-06, "loss": 0.0447, "step": 28545 }, { "epoch": 1.5218851628725276, "grad_norm": 2.2656896114349365, "learning_rate": 4.291e-06, "loss": 0.0395, "step": 28546 }, { "epoch": 1.5219384763021804, "grad_norm": 1.9579334259033203, "learning_rate": 4.2908000000000005e-06, "loss": 0.0197, "step": 28547 }, { "epoch": 1.5219917897318336, "grad_norm": 1.4635120630264282, "learning_rate": 4.2906e-06, "loss": 0.0228, "step": 28548 }, { "epoch": 1.5220451031614863, "grad_norm": 0.5611939430236816, "learning_rate": 4.2904000000000005e-06, "loss": 0.0046, "step": 28549 }, { "epoch": 1.5220984165911393, "grad_norm": 2.703179359436035, "learning_rate": 4.290200000000001e-06, "loss": 0.0136, "step": 28550 }, { "epoch": 1.5221517300207923, "grad_norm": 1.962648630142212, "learning_rate": 4.2900000000000004e-06, "loss": 0.0295, "step": 28551 }, { "epoch": 1.522205043450445, "grad_norm": 1.557833194732666, "learning_rate": 4.2898e-06, "loss": 0.0092, "step": 28552 }, { "epoch": 1.5222583568800983, "grad_norm": 0.6523405909538269, "learning_rate": 4.2896e-06, "loss": 0.0159, "step": 28553 }, { "epoch": 1.522311670309751, "grad_norm": 1.7451891899108887, "learning_rate": 4.2894e-06, "loss": 0.0078, "step": 28554 }, { "epoch": 1.522364983739404, "grad_norm": 3.238511800765991, "learning_rate": 4.2892e-06, "loss": 0.0448, "step": 28555 }, { "epoch": 1.522418297169057, "grad_norm": 3.3190910816192627, "learning_rate": 4.289000000000001e-06, "loss": 0.0415, "step": 28556 }, { "epoch": 1.5224716105987097, "grad_norm": 1.6951184272766113, "learning_rate": 4.2888e-06, "loss": 0.0335, "step": 28557 }, { "epoch": 1.5225249240283627, "grad_norm": 1.265960454940796, "learning_rate": 4.288600000000001e-06, "loss": 0.0196, "step": 28558 }, { "epoch": 1.5225782374580157, "grad_norm": 1.811712384223938, "learning_rate": 4.2884e-06, "loss": 0.0157, "step": 28559 }, { "epoch": 1.5226315508876684, "grad_norm": 1.5812379121780396, "learning_rate": 4.2882e-06, "loss": 0.0107, "step": 28560 }, { "epoch": 1.5226848643173216, "grad_norm": 1.9641889333724976, "learning_rate": 4.288e-06, "loss": 0.0314, "step": 28561 }, { "epoch": 1.5227381777469744, "grad_norm": 0.5977951288223267, "learning_rate": 4.287800000000001e-06, "loss": 0.0055, "step": 28562 }, { "epoch": 1.5227914911766274, "grad_norm": 2.829585313796997, "learning_rate": 4.2876e-06, "loss": 0.042, "step": 28563 }, { "epoch": 1.5228448046062804, "grad_norm": 1.1355787515640259, "learning_rate": 4.287400000000001e-06, "loss": 0.0097, "step": 28564 }, { "epoch": 1.5228981180359331, "grad_norm": 2.3227322101593018, "learning_rate": 4.2872e-06, "loss": 0.0274, "step": 28565 }, { "epoch": 1.5229514314655863, "grad_norm": 2.019662618637085, "learning_rate": 4.287000000000001e-06, "loss": 0.0248, "step": 28566 }, { "epoch": 1.523004744895239, "grad_norm": 2.095393180847168, "learning_rate": 4.2868e-06, "loss": 0.0436, "step": 28567 }, { "epoch": 1.523058058324892, "grad_norm": 1.3499886989593506, "learning_rate": 4.286600000000001e-06, "loss": 0.0138, "step": 28568 }, { "epoch": 1.523111371754545, "grad_norm": 1.4972553253173828, "learning_rate": 4.2864e-06, "loss": 0.0108, "step": 28569 }, { "epoch": 1.5231646851841978, "grad_norm": 1.3559874296188354, "learning_rate": 4.2862000000000006e-06, "loss": 0.0376, "step": 28570 }, { "epoch": 1.523217998613851, "grad_norm": 2.6714422702789307, "learning_rate": 4.286e-06, "loss": 0.0178, "step": 28571 }, { "epoch": 1.5232713120435037, "grad_norm": 3.0235984325408936, "learning_rate": 4.2858000000000005e-06, "loss": 0.0236, "step": 28572 }, { "epoch": 1.5233246254731567, "grad_norm": 1.8399478197097778, "learning_rate": 4.2856e-06, "loss": 0.0477, "step": 28573 }, { "epoch": 1.5233779389028097, "grad_norm": 2.276210069656372, "learning_rate": 4.2854000000000005e-06, "loss": 0.0347, "step": 28574 }, { "epoch": 1.5234312523324625, "grad_norm": 1.4649553298950195, "learning_rate": 4.2852e-06, "loss": 0.0371, "step": 28575 }, { "epoch": 1.5234845657621154, "grad_norm": 2.48699688911438, "learning_rate": 4.2850000000000005e-06, "loss": 0.0395, "step": 28576 }, { "epoch": 1.5235378791917684, "grad_norm": 1.7345508337020874, "learning_rate": 4.284800000000001e-06, "loss": 0.0216, "step": 28577 }, { "epoch": 1.5235911926214212, "grad_norm": 1.799674391746521, "learning_rate": 4.2846000000000004e-06, "loss": 0.0296, "step": 28578 }, { "epoch": 1.5236445060510744, "grad_norm": 1.530005693435669, "learning_rate": 4.2844e-06, "loss": 0.0157, "step": 28579 }, { "epoch": 1.5236978194807271, "grad_norm": 1.405945062637329, "learning_rate": 4.2842e-06, "loss": 0.0258, "step": 28580 }, { "epoch": 1.5237511329103801, "grad_norm": 3.250067710876465, "learning_rate": 4.284e-06, "loss": 0.0338, "step": 28581 }, { "epoch": 1.523804446340033, "grad_norm": 0.5556212067604065, "learning_rate": 4.2838e-06, "loss": 0.0053, "step": 28582 }, { "epoch": 1.5238577597696858, "grad_norm": 2.705864667892456, "learning_rate": 4.283600000000001e-06, "loss": 0.018, "step": 28583 }, { "epoch": 1.523911073199339, "grad_norm": 1.7773158550262451, "learning_rate": 4.2834e-06, "loss": 0.0296, "step": 28584 }, { "epoch": 1.5239643866289918, "grad_norm": 2.1332497596740723, "learning_rate": 4.2832e-06, "loss": 0.0289, "step": 28585 }, { "epoch": 1.5240177000586448, "grad_norm": 2.3176534175872803, "learning_rate": 4.283e-06, "loss": 0.0385, "step": 28586 }, { "epoch": 1.5240710134882978, "grad_norm": 1.052381157875061, "learning_rate": 4.2828e-06, "loss": 0.0088, "step": 28587 }, { "epoch": 1.5241243269179505, "grad_norm": 1.3791227340698242, "learning_rate": 4.2826e-06, "loss": 0.0188, "step": 28588 }, { "epoch": 1.5241776403476037, "grad_norm": 1.8228789567947388, "learning_rate": 4.282400000000001e-06, "loss": 0.0355, "step": 28589 }, { "epoch": 1.5242309537772565, "grad_norm": 1.88065505027771, "learning_rate": 4.2822e-06, "loss": 0.039, "step": 28590 }, { "epoch": 1.5242842672069095, "grad_norm": 1.1901355981826782, "learning_rate": 4.282000000000001e-06, "loss": 0.0299, "step": 28591 }, { "epoch": 1.5243375806365624, "grad_norm": 1.5199445486068726, "learning_rate": 4.2818e-06, "loss": 0.0185, "step": 28592 }, { "epoch": 1.5243908940662152, "grad_norm": 2.143768072128296, "learning_rate": 4.2816e-06, "loss": 0.0511, "step": 28593 }, { "epoch": 1.5244442074958682, "grad_norm": 1.18974769115448, "learning_rate": 4.2814e-06, "loss": 0.0153, "step": 28594 }, { "epoch": 1.5244975209255212, "grad_norm": 2.1650571823120117, "learning_rate": 4.281200000000001e-06, "loss": 0.0224, "step": 28595 }, { "epoch": 1.524550834355174, "grad_norm": 1.960364580154419, "learning_rate": 4.281e-06, "loss": 0.0181, "step": 28596 }, { "epoch": 1.5246041477848271, "grad_norm": 1.401134729385376, "learning_rate": 4.280800000000001e-06, "loss": 0.0245, "step": 28597 }, { "epoch": 1.5246574612144799, "grad_norm": 2.130368709564209, "learning_rate": 4.2806e-06, "loss": 0.0202, "step": 28598 }, { "epoch": 1.5247107746441328, "grad_norm": 2.4457178115844727, "learning_rate": 4.2804e-06, "loss": 0.0365, "step": 28599 }, { "epoch": 1.5247640880737858, "grad_norm": 2.6109213829040527, "learning_rate": 4.2802e-06, "loss": 0.0211, "step": 28600 }, { "epoch": 1.5248174015034386, "grad_norm": 2.0245683193206787, "learning_rate": 4.2800000000000005e-06, "loss": 0.0305, "step": 28601 }, { "epoch": 1.5248707149330918, "grad_norm": 1.857753038406372, "learning_rate": 4.279800000000001e-06, "loss": 0.0153, "step": 28602 }, { "epoch": 1.5249240283627445, "grad_norm": 2.065526247024536, "learning_rate": 4.2796000000000005e-06, "loss": 0.0263, "step": 28603 }, { "epoch": 1.5249773417923975, "grad_norm": 0.7751126289367676, "learning_rate": 4.2794e-06, "loss": 0.0067, "step": 28604 }, { "epoch": 1.5250306552220505, "grad_norm": 0.8577225208282471, "learning_rate": 4.2792000000000005e-06, "loss": 0.0071, "step": 28605 }, { "epoch": 1.5250839686517033, "grad_norm": 2.155695676803589, "learning_rate": 4.279e-06, "loss": 0.0205, "step": 28606 }, { "epoch": 1.5251372820813565, "grad_norm": 1.7305183410644531, "learning_rate": 4.2788000000000004e-06, "loss": 0.0087, "step": 28607 }, { "epoch": 1.5251905955110092, "grad_norm": 1.9717681407928467, "learning_rate": 4.278600000000001e-06, "loss": 0.04, "step": 28608 }, { "epoch": 1.5252439089406622, "grad_norm": 1.6806437969207764, "learning_rate": 4.2784e-06, "loss": 0.0077, "step": 28609 }, { "epoch": 1.5252972223703152, "grad_norm": 1.8092637062072754, "learning_rate": 4.2782e-06, "loss": 0.0191, "step": 28610 }, { "epoch": 1.525350535799968, "grad_norm": 0.9748755097389221, "learning_rate": 4.278e-06, "loss": 0.0251, "step": 28611 }, { "epoch": 1.525403849229621, "grad_norm": 3.8658785820007324, "learning_rate": 4.2778e-06, "loss": 0.0259, "step": 28612 }, { "epoch": 1.525457162659274, "grad_norm": 1.0211431980133057, "learning_rate": 4.2776e-06, "loss": 0.0187, "step": 28613 }, { "epoch": 1.5255104760889266, "grad_norm": 0.24993333220481873, "learning_rate": 4.277400000000001e-06, "loss": 0.0027, "step": 28614 }, { "epoch": 1.5255637895185798, "grad_norm": 1.2173476219177246, "learning_rate": 4.2772e-06, "loss": 0.016, "step": 28615 }, { "epoch": 1.5256171029482326, "grad_norm": 1.3260457515716553, "learning_rate": 4.277000000000001e-06, "loss": 0.0189, "step": 28616 }, { "epoch": 1.5256704163778856, "grad_norm": 1.5688003301620483, "learning_rate": 4.2768e-06, "loss": 0.0473, "step": 28617 }, { "epoch": 1.5257237298075386, "grad_norm": 1.7902263402938843, "learning_rate": 4.2766e-06, "loss": 0.0234, "step": 28618 }, { "epoch": 1.5257770432371913, "grad_norm": 1.146764874458313, "learning_rate": 4.2764e-06, "loss": 0.0215, "step": 28619 }, { "epoch": 1.5258303566668445, "grad_norm": 1.257736325263977, "learning_rate": 4.276200000000001e-06, "loss": 0.0215, "step": 28620 }, { "epoch": 1.5258836700964973, "grad_norm": 2.9166259765625, "learning_rate": 4.276e-06, "loss": 0.0308, "step": 28621 }, { "epoch": 1.5259369835261503, "grad_norm": 2.261861801147461, "learning_rate": 4.275800000000001e-06, "loss": 0.0376, "step": 28622 }, { "epoch": 1.5259902969558032, "grad_norm": 2.3633084297180176, "learning_rate": 4.2756e-06, "loss": 0.0181, "step": 28623 }, { "epoch": 1.526043610385456, "grad_norm": 1.7256683111190796, "learning_rate": 4.2754e-06, "loss": 0.0211, "step": 28624 }, { "epoch": 1.526096923815109, "grad_norm": 2.632213830947876, "learning_rate": 4.2752e-06, "loss": 0.0158, "step": 28625 }, { "epoch": 1.526150237244762, "grad_norm": 2.8489131927490234, "learning_rate": 4.2750000000000006e-06, "loss": 0.0333, "step": 28626 }, { "epoch": 1.526203550674415, "grad_norm": 2.2923648357391357, "learning_rate": 4.274800000000001e-06, "loss": 0.0175, "step": 28627 }, { "epoch": 1.526256864104068, "grad_norm": 2.0270307064056396, "learning_rate": 4.2746000000000005e-06, "loss": 0.0123, "step": 28628 }, { "epoch": 1.5263101775337207, "grad_norm": 2.3042845726013184, "learning_rate": 4.2744e-06, "loss": 0.0109, "step": 28629 }, { "epoch": 1.5263634909633736, "grad_norm": 1.4132258892059326, "learning_rate": 4.2742000000000005e-06, "loss": 0.0071, "step": 28630 }, { "epoch": 1.5264168043930266, "grad_norm": 1.9965094327926636, "learning_rate": 4.274e-06, "loss": 0.0493, "step": 28631 }, { "epoch": 1.5264701178226794, "grad_norm": 2.5508687496185303, "learning_rate": 4.2738000000000005e-06, "loss": 0.0283, "step": 28632 }, { "epoch": 1.5265234312523326, "grad_norm": 1.9576070308685303, "learning_rate": 4.273600000000001e-06, "loss": 0.0316, "step": 28633 }, { "epoch": 1.5265767446819853, "grad_norm": 1.76043701171875, "learning_rate": 4.2734000000000005e-06, "loss": 0.0252, "step": 28634 }, { "epoch": 1.5266300581116383, "grad_norm": 1.1818383932113647, "learning_rate": 4.2732e-06, "loss": 0.0218, "step": 28635 }, { "epoch": 1.5266833715412913, "grad_norm": 0.5668198466300964, "learning_rate": 4.2730000000000004e-06, "loss": 0.0034, "step": 28636 }, { "epoch": 1.526736684970944, "grad_norm": 5.727201461791992, "learning_rate": 4.2728e-06, "loss": 0.0358, "step": 28637 }, { "epoch": 1.5267899984005973, "grad_norm": 1.2172645330429077, "learning_rate": 4.2726e-06, "loss": 0.0138, "step": 28638 }, { "epoch": 1.52684331183025, "grad_norm": 1.639163851737976, "learning_rate": 4.272400000000001e-06, "loss": 0.0383, "step": 28639 }, { "epoch": 1.526896625259903, "grad_norm": 2.4609642028808594, "learning_rate": 4.2722e-06, "loss": 0.0426, "step": 28640 }, { "epoch": 1.526949938689556, "grad_norm": 2.9716384410858154, "learning_rate": 4.272000000000001e-06, "loss": 0.0164, "step": 28641 }, { "epoch": 1.5270032521192087, "grad_norm": 1.5925934314727783, "learning_rate": 4.2718e-06, "loss": 0.01, "step": 28642 }, { "epoch": 1.5270565655488617, "grad_norm": 2.099257230758667, "learning_rate": 4.2716e-06, "loss": 0.025, "step": 28643 }, { "epoch": 1.5271098789785147, "grad_norm": 2.422635078430176, "learning_rate": 4.2714e-06, "loss": 0.0217, "step": 28644 }, { "epoch": 1.5271631924081677, "grad_norm": 0.6234891414642334, "learning_rate": 4.271200000000001e-06, "loss": 0.0056, "step": 28645 }, { "epoch": 1.5272165058378206, "grad_norm": 3.3272383213043213, "learning_rate": 4.271e-06, "loss": 0.0392, "step": 28646 }, { "epoch": 1.5272698192674734, "grad_norm": 4.489732265472412, "learning_rate": 4.270800000000001e-06, "loss": 0.0294, "step": 28647 }, { "epoch": 1.5273231326971264, "grad_norm": 3.629096746444702, "learning_rate": 4.2706e-06, "loss": 0.0349, "step": 28648 }, { "epoch": 1.5273764461267794, "grad_norm": 2.258106231689453, "learning_rate": 4.2704e-06, "loss": 0.028, "step": 28649 }, { "epoch": 1.5274297595564321, "grad_norm": 1.067888855934143, "learning_rate": 4.2702e-06, "loss": 0.0201, "step": 28650 }, { "epoch": 1.5274830729860853, "grad_norm": 1.9663074016571045, "learning_rate": 4.270000000000001e-06, "loss": 0.0117, "step": 28651 }, { "epoch": 1.527536386415738, "grad_norm": 1.2418543100357056, "learning_rate": 4.269800000000001e-06, "loss": 0.021, "step": 28652 }, { "epoch": 1.527589699845391, "grad_norm": 2.2700326442718506, "learning_rate": 4.269600000000001e-06, "loss": 0.0304, "step": 28653 }, { "epoch": 1.527643013275044, "grad_norm": 2.233768939971924, "learning_rate": 4.2694e-06, "loss": 0.014, "step": 28654 }, { "epoch": 1.5276963267046968, "grad_norm": 1.4006993770599365, "learning_rate": 4.2692000000000006e-06, "loss": 0.0131, "step": 28655 }, { "epoch": 1.52774964013435, "grad_norm": 2.7464962005615234, "learning_rate": 4.269e-06, "loss": 0.0279, "step": 28656 }, { "epoch": 1.5278029535640028, "grad_norm": 2.368987560272217, "learning_rate": 4.2688000000000005e-06, "loss": 0.039, "step": 28657 }, { "epoch": 1.5278562669936557, "grad_norm": 4.493856430053711, "learning_rate": 4.268600000000001e-06, "loss": 0.0302, "step": 28658 }, { "epoch": 1.5279095804233087, "grad_norm": 2.8686091899871826, "learning_rate": 4.2684000000000005e-06, "loss": 0.0445, "step": 28659 }, { "epoch": 1.5279628938529615, "grad_norm": 2.2573258876800537, "learning_rate": 4.2682e-06, "loss": 0.0281, "step": 28660 }, { "epoch": 1.5280162072826144, "grad_norm": 2.126901865005493, "learning_rate": 4.2680000000000005e-06, "loss": 0.061, "step": 28661 }, { "epoch": 1.5280695207122674, "grad_norm": 1.1802165508270264, "learning_rate": 4.2678e-06, "loss": 0.0203, "step": 28662 }, { "epoch": 1.5281228341419202, "grad_norm": 0.7548635005950928, "learning_rate": 4.2676000000000004e-06, "loss": 0.0057, "step": 28663 }, { "epoch": 1.5281761475715734, "grad_norm": 0.5674963593482971, "learning_rate": 4.267400000000001e-06, "loss": 0.0063, "step": 28664 }, { "epoch": 1.5282294610012261, "grad_norm": 1.171289324760437, "learning_rate": 4.2672e-06, "loss": 0.021, "step": 28665 }, { "epoch": 1.5282827744308791, "grad_norm": 2.372501850128174, "learning_rate": 4.267000000000001e-06, "loss": 0.0427, "step": 28666 }, { "epoch": 1.528336087860532, "grad_norm": 2.0968689918518066, "learning_rate": 4.2668e-06, "loss": 0.0238, "step": 28667 }, { "epoch": 1.5283894012901849, "grad_norm": 2.150113105773926, "learning_rate": 4.2666e-06, "loss": 0.0168, "step": 28668 }, { "epoch": 1.528442714719838, "grad_norm": 1.2713289260864258, "learning_rate": 4.2664e-06, "loss": 0.0274, "step": 28669 }, { "epoch": 1.5284960281494908, "grad_norm": 2.5995242595672607, "learning_rate": 4.266200000000001e-06, "loss": 0.0322, "step": 28670 }, { "epoch": 1.5285493415791438, "grad_norm": 1.9574775695800781, "learning_rate": 4.266e-06, "loss": 0.032, "step": 28671 }, { "epoch": 1.5286026550087968, "grad_norm": 1.1199241876602173, "learning_rate": 4.265800000000001e-06, "loss": 0.0098, "step": 28672 }, { "epoch": 1.5286559684384495, "grad_norm": 2.11740779876709, "learning_rate": 4.2656e-06, "loss": 0.0191, "step": 28673 }, { "epoch": 1.5287092818681027, "grad_norm": 1.4775612354278564, "learning_rate": 4.2654e-06, "loss": 0.0319, "step": 28674 }, { "epoch": 1.5287625952977555, "grad_norm": 1.4979536533355713, "learning_rate": 4.2652e-06, "loss": 0.0305, "step": 28675 }, { "epoch": 1.5288159087274085, "grad_norm": 2.559548854827881, "learning_rate": 4.265000000000001e-06, "loss": 0.0338, "step": 28676 }, { "epoch": 1.5288692221570614, "grad_norm": 1.4953879117965698, "learning_rate": 4.2648e-06, "loss": 0.0242, "step": 28677 }, { "epoch": 1.5289225355867142, "grad_norm": 1.7313001155853271, "learning_rate": 4.264600000000001e-06, "loss": 0.0174, "step": 28678 }, { "epoch": 1.5289758490163672, "grad_norm": 1.460589051246643, "learning_rate": 4.2644e-06, "loss": 0.009, "step": 28679 }, { "epoch": 1.5290291624460202, "grad_norm": 1.4422390460968018, "learning_rate": 4.264200000000001e-06, "loss": 0.024, "step": 28680 }, { "epoch": 1.529082475875673, "grad_norm": 1.2527801990509033, "learning_rate": 4.264e-06, "loss": 0.0202, "step": 28681 }, { "epoch": 1.5291357893053261, "grad_norm": 2.418680191040039, "learning_rate": 4.263800000000001e-06, "loss": 0.0327, "step": 28682 }, { "epoch": 1.5291891027349789, "grad_norm": 1.5370358228683472, "learning_rate": 4.2636e-06, "loss": 0.0077, "step": 28683 }, { "epoch": 1.5292424161646319, "grad_norm": 1.974591851234436, "learning_rate": 4.2634000000000005e-06, "loss": 0.0526, "step": 28684 }, { "epoch": 1.5292957295942848, "grad_norm": 1.350054383277893, "learning_rate": 4.2632e-06, "loss": 0.0194, "step": 28685 }, { "epoch": 1.5293490430239376, "grad_norm": 2.1202611923217773, "learning_rate": 4.2630000000000005e-06, "loss": 0.0183, "step": 28686 }, { "epoch": 1.5294023564535908, "grad_norm": 1.2322375774383545, "learning_rate": 4.2628e-06, "loss": 0.0206, "step": 28687 }, { "epoch": 1.5294556698832436, "grad_norm": 2.8144547939300537, "learning_rate": 4.2626000000000005e-06, "loss": 0.0225, "step": 28688 }, { "epoch": 1.5295089833128965, "grad_norm": 2.3140220642089844, "learning_rate": 4.2624e-06, "loss": 0.0309, "step": 28689 }, { "epoch": 1.5295622967425495, "grad_norm": 2.0357019901275635, "learning_rate": 4.2622000000000005e-06, "loss": 0.0111, "step": 28690 }, { "epoch": 1.5296156101722023, "grad_norm": 0.9152907729148865, "learning_rate": 4.262000000000001e-06, "loss": 0.0061, "step": 28691 }, { "epoch": 1.5296689236018555, "grad_norm": 1.9977896213531494, "learning_rate": 4.2618000000000004e-06, "loss": 0.02, "step": 28692 }, { "epoch": 1.5297222370315082, "grad_norm": 1.2890983819961548, "learning_rate": 4.2616e-06, "loss": 0.0188, "step": 28693 }, { "epoch": 1.5297755504611612, "grad_norm": 1.4496551752090454, "learning_rate": 4.2614e-06, "loss": 0.0135, "step": 28694 }, { "epoch": 1.5298288638908142, "grad_norm": 1.239974021911621, "learning_rate": 4.2612e-06, "loss": 0.0083, "step": 28695 }, { "epoch": 1.529882177320467, "grad_norm": 2.2999768257141113, "learning_rate": 4.261e-06, "loss": 0.0416, "step": 28696 }, { "epoch": 1.52993549075012, "grad_norm": 1.0734682083129883, "learning_rate": 4.260800000000001e-06, "loss": 0.0089, "step": 28697 }, { "epoch": 1.529988804179773, "grad_norm": 1.8751789331436157, "learning_rate": 4.2606e-06, "loss": 0.0421, "step": 28698 }, { "epoch": 1.5300421176094257, "grad_norm": 0.6640763282775879, "learning_rate": 4.2604e-06, "loss": 0.0068, "step": 28699 }, { "epoch": 1.5300954310390789, "grad_norm": 1.6105754375457764, "learning_rate": 4.2602e-06, "loss": 0.0143, "step": 28700 }, { "epoch": 1.5301487444687316, "grad_norm": 1.1403803825378418, "learning_rate": 4.26e-06, "loss": 0.0076, "step": 28701 }, { "epoch": 1.5302020578983846, "grad_norm": 1.5070809125900269, "learning_rate": 4.2598e-06, "loss": 0.0197, "step": 28702 }, { "epoch": 1.5302553713280376, "grad_norm": 0.9257000088691711, "learning_rate": 4.259600000000001e-06, "loss": 0.0182, "step": 28703 }, { "epoch": 1.5303086847576903, "grad_norm": 1.8366413116455078, "learning_rate": 4.2594e-06, "loss": 0.0144, "step": 28704 }, { "epoch": 1.5303619981873435, "grad_norm": 1.0878753662109375, "learning_rate": 4.259200000000001e-06, "loss": 0.0067, "step": 28705 }, { "epoch": 1.5304153116169963, "grad_norm": 1.646634578704834, "learning_rate": 4.259e-06, "loss": 0.0346, "step": 28706 }, { "epoch": 1.5304686250466493, "grad_norm": 1.2506890296936035, "learning_rate": 4.2588e-06, "loss": 0.018, "step": 28707 }, { "epoch": 1.5305219384763022, "grad_norm": 1.4286398887634277, "learning_rate": 4.2586e-06, "loss": 0.0226, "step": 28708 }, { "epoch": 1.530575251905955, "grad_norm": 2.402590036392212, "learning_rate": 4.258400000000001e-06, "loss": 0.0182, "step": 28709 }, { "epoch": 1.5306285653356082, "grad_norm": 0.7584518790245056, "learning_rate": 4.2582e-06, "loss": 0.013, "step": 28710 }, { "epoch": 1.530681878765261, "grad_norm": 0.26116666197776794, "learning_rate": 4.2580000000000006e-06, "loss": 0.0024, "step": 28711 }, { "epoch": 1.530735192194914, "grad_norm": 2.4635331630706787, "learning_rate": 4.2578e-06, "loss": 0.0256, "step": 28712 }, { "epoch": 1.530788505624567, "grad_norm": 1.2098559141159058, "learning_rate": 4.2576000000000005e-06, "loss": 0.01, "step": 28713 }, { "epoch": 1.5308418190542197, "grad_norm": 1.699631690979004, "learning_rate": 4.2574e-06, "loss": 0.0278, "step": 28714 }, { "epoch": 1.5308951324838727, "grad_norm": 1.2622019052505493, "learning_rate": 4.2572000000000005e-06, "loss": 0.0074, "step": 28715 }, { "epoch": 1.5309484459135256, "grad_norm": 2.278641700744629, "learning_rate": 4.257000000000001e-06, "loss": 0.017, "step": 28716 }, { "epoch": 1.5310017593431784, "grad_norm": 2.719529867172241, "learning_rate": 4.2568000000000005e-06, "loss": 0.0243, "step": 28717 }, { "epoch": 1.5310550727728316, "grad_norm": 1.7421132326126099, "learning_rate": 4.2566e-06, "loss": 0.0164, "step": 28718 }, { "epoch": 1.5311083862024844, "grad_norm": 0.8461847901344299, "learning_rate": 4.2564000000000004e-06, "loss": 0.0072, "step": 28719 }, { "epoch": 1.5311616996321373, "grad_norm": 4.210729598999023, "learning_rate": 4.2562e-06, "loss": 0.0648, "step": 28720 }, { "epoch": 1.5312150130617903, "grad_norm": 1.0529733896255493, "learning_rate": 4.256e-06, "loss": 0.0098, "step": 28721 }, { "epoch": 1.531268326491443, "grad_norm": 1.1631882190704346, "learning_rate": 4.255800000000001e-06, "loss": 0.0148, "step": 28722 }, { "epoch": 1.5313216399210963, "grad_norm": 2.413877487182617, "learning_rate": 4.2556e-06, "loss": 0.0252, "step": 28723 }, { "epoch": 1.531374953350749, "grad_norm": 1.9928550720214844, "learning_rate": 4.2554e-06, "loss": 0.0203, "step": 28724 }, { "epoch": 1.531428266780402, "grad_norm": 1.6222766637802124, "learning_rate": 4.2552e-06, "loss": 0.0223, "step": 28725 }, { "epoch": 1.531481580210055, "grad_norm": 1.4374420642852783, "learning_rate": 4.255e-06, "loss": 0.0094, "step": 28726 }, { "epoch": 1.5315348936397077, "grad_norm": 2.2076733112335205, "learning_rate": 4.2548e-06, "loss": 0.0136, "step": 28727 }, { "epoch": 1.5315882070693607, "grad_norm": 1.9214190244674683, "learning_rate": 4.254600000000001e-06, "loss": 0.02, "step": 28728 }, { "epoch": 1.5316415204990137, "grad_norm": 1.1257116794586182, "learning_rate": 4.2544e-06, "loss": 0.0061, "step": 28729 }, { "epoch": 1.5316948339286667, "grad_norm": 3.2363064289093018, "learning_rate": 4.254200000000001e-06, "loss": 0.0347, "step": 28730 }, { "epoch": 1.5317481473583197, "grad_norm": 2.9806840419769287, "learning_rate": 4.254e-06, "loss": 0.0248, "step": 28731 }, { "epoch": 1.5318014607879724, "grad_norm": 0.4021672010421753, "learning_rate": 4.2538e-06, "loss": 0.0029, "step": 28732 }, { "epoch": 1.5318547742176254, "grad_norm": 1.207128882408142, "learning_rate": 4.2536e-06, "loss": 0.0262, "step": 28733 }, { "epoch": 1.5319080876472784, "grad_norm": 0.8553943634033203, "learning_rate": 4.253400000000001e-06, "loss": 0.0221, "step": 28734 }, { "epoch": 1.5319614010769311, "grad_norm": 1.5276575088500977, "learning_rate": 4.2532e-06, "loss": 0.0192, "step": 28735 }, { "epoch": 1.5320147145065843, "grad_norm": 2.0990376472473145, "learning_rate": 4.253000000000001e-06, "loss": 0.0179, "step": 28736 }, { "epoch": 1.532068027936237, "grad_norm": 1.6196287870407104, "learning_rate": 4.2528e-06, "loss": 0.0252, "step": 28737 }, { "epoch": 1.53212134136589, "grad_norm": 2.801542282104492, "learning_rate": 4.252600000000001e-06, "loss": 0.0195, "step": 28738 }, { "epoch": 1.532174654795543, "grad_norm": 1.9691458940505981, "learning_rate": 4.2524e-06, "loss": 0.0107, "step": 28739 }, { "epoch": 1.5322279682251958, "grad_norm": 1.4991847276687622, "learning_rate": 4.2522000000000006e-06, "loss": 0.0084, "step": 28740 }, { "epoch": 1.532281281654849, "grad_norm": 0.662703812122345, "learning_rate": 4.252000000000001e-06, "loss": 0.0049, "step": 28741 }, { "epoch": 1.5323345950845018, "grad_norm": 1.2045204639434814, "learning_rate": 4.2518000000000005e-06, "loss": 0.0307, "step": 28742 }, { "epoch": 1.5323879085141547, "grad_norm": 1.3577003479003906, "learning_rate": 4.2516e-06, "loss": 0.0249, "step": 28743 }, { "epoch": 1.5324412219438077, "grad_norm": 1.480543851852417, "learning_rate": 4.2514000000000005e-06, "loss": 0.0082, "step": 28744 }, { "epoch": 1.5324945353734605, "grad_norm": 2.740870475769043, "learning_rate": 4.2512e-06, "loss": 0.0475, "step": 28745 }, { "epoch": 1.5325478488031135, "grad_norm": 1.9230616092681885, "learning_rate": 4.2510000000000005e-06, "loss": 0.0293, "step": 28746 }, { "epoch": 1.5326011622327664, "grad_norm": 1.1348484754562378, "learning_rate": 4.250800000000001e-06, "loss": 0.0121, "step": 28747 }, { "epoch": 1.5326544756624194, "grad_norm": 4.329961776733398, "learning_rate": 4.2506000000000004e-06, "loss": 0.0498, "step": 28748 }, { "epoch": 1.5327077890920724, "grad_norm": 1.946866512298584, "learning_rate": 4.2504e-06, "loss": 0.0361, "step": 28749 }, { "epoch": 1.5327611025217251, "grad_norm": 3.739285469055176, "learning_rate": 4.2502e-06, "loss": 0.0313, "step": 28750 }, { "epoch": 1.5328144159513781, "grad_norm": 1.3118259906768799, "learning_rate": 4.25e-06, "loss": 0.0176, "step": 28751 }, { "epoch": 1.532867729381031, "grad_norm": 1.052530288696289, "learning_rate": 4.2498e-06, "loss": 0.0128, "step": 28752 }, { "epoch": 1.5329210428106839, "grad_norm": 3.3628857135772705, "learning_rate": 4.249600000000001e-06, "loss": 0.0277, "step": 28753 }, { "epoch": 1.532974356240337, "grad_norm": 1.2082083225250244, "learning_rate": 4.2494e-06, "loss": 0.0092, "step": 28754 }, { "epoch": 1.5330276696699898, "grad_norm": 1.575445294380188, "learning_rate": 4.249200000000001e-06, "loss": 0.0076, "step": 28755 }, { "epoch": 1.5330809830996428, "grad_norm": 1.418117880821228, "learning_rate": 4.249e-06, "loss": 0.0162, "step": 28756 }, { "epoch": 1.5331342965292958, "grad_norm": 1.3228802680969238, "learning_rate": 4.2488e-06, "loss": 0.0087, "step": 28757 }, { "epoch": 1.5331876099589485, "grad_norm": 1.4471572637557983, "learning_rate": 4.2486e-06, "loss": 0.0224, "step": 28758 }, { "epoch": 1.5332409233886017, "grad_norm": 1.3923159837722778, "learning_rate": 4.248400000000001e-06, "loss": 0.013, "step": 28759 }, { "epoch": 1.5332942368182545, "grad_norm": 0.9765669107437134, "learning_rate": 4.2482e-06, "loss": 0.0105, "step": 28760 }, { "epoch": 1.5333475502479075, "grad_norm": 1.319934368133545, "learning_rate": 4.248000000000001e-06, "loss": 0.0248, "step": 28761 }, { "epoch": 1.5334008636775605, "grad_norm": 2.2134299278259277, "learning_rate": 4.2478e-06, "loss": 0.0252, "step": 28762 }, { "epoch": 1.5334541771072132, "grad_norm": 1.1193355321884155, "learning_rate": 4.247600000000001e-06, "loss": 0.0095, "step": 28763 }, { "epoch": 1.5335074905368662, "grad_norm": 2.3368778228759766, "learning_rate": 4.2474e-06, "loss": 0.0287, "step": 28764 }, { "epoch": 1.5335608039665192, "grad_norm": 1.9335495233535767, "learning_rate": 4.247200000000001e-06, "loss": 0.0434, "step": 28765 }, { "epoch": 1.5336141173961721, "grad_norm": 1.5681616067886353, "learning_rate": 4.247e-06, "loss": 0.0275, "step": 28766 }, { "epoch": 1.5336674308258251, "grad_norm": 1.6832754611968994, "learning_rate": 4.2468000000000006e-06, "loss": 0.0377, "step": 28767 }, { "epoch": 1.5337207442554779, "grad_norm": 3.1926329135894775, "learning_rate": 4.2466e-06, "loss": 0.0385, "step": 28768 }, { "epoch": 1.5337740576851309, "grad_norm": 1.481615424156189, "learning_rate": 4.2464000000000005e-06, "loss": 0.0245, "step": 28769 }, { "epoch": 1.5338273711147838, "grad_norm": 2.7824277877807617, "learning_rate": 4.2462e-06, "loss": 0.042, "step": 28770 }, { "epoch": 1.5338806845444366, "grad_norm": 1.3809562921524048, "learning_rate": 4.2460000000000005e-06, "loss": 0.0271, "step": 28771 }, { "epoch": 1.5339339979740898, "grad_norm": 1.3459066152572632, "learning_rate": 4.2458e-06, "loss": 0.0169, "step": 28772 }, { "epoch": 1.5339873114037426, "grad_norm": 1.816271424293518, "learning_rate": 4.2456000000000005e-06, "loss": 0.0142, "step": 28773 }, { "epoch": 1.5340406248333955, "grad_norm": 3.3188536167144775, "learning_rate": 4.2454e-06, "loss": 0.0453, "step": 28774 }, { "epoch": 1.5340939382630485, "grad_norm": 1.467474102973938, "learning_rate": 4.2452000000000005e-06, "loss": 0.0189, "step": 28775 }, { "epoch": 1.5341472516927013, "grad_norm": 2.0858328342437744, "learning_rate": 4.245e-06, "loss": 0.0127, "step": 28776 }, { "epoch": 1.5342005651223545, "grad_norm": 2.9400534629821777, "learning_rate": 4.2448e-06, "loss": 0.0596, "step": 28777 }, { "epoch": 1.5342538785520072, "grad_norm": 1.8299847841262817, "learning_rate": 4.2446e-06, "loss": 0.0359, "step": 28778 }, { "epoch": 1.5343071919816602, "grad_norm": 0.9123448729515076, "learning_rate": 4.2444e-06, "loss": 0.0115, "step": 28779 }, { "epoch": 1.5343605054113132, "grad_norm": 0.9128018021583557, "learning_rate": 4.244200000000001e-06, "loss": 0.0069, "step": 28780 }, { "epoch": 1.534413818840966, "grad_norm": 0.9775870442390442, "learning_rate": 4.244e-06, "loss": 0.0122, "step": 28781 }, { "epoch": 1.534467132270619, "grad_norm": 1.138428807258606, "learning_rate": 4.2438e-06, "loss": 0.0082, "step": 28782 }, { "epoch": 1.534520445700272, "grad_norm": 1.7369626760482788, "learning_rate": 4.2436e-06, "loss": 0.0307, "step": 28783 }, { "epoch": 1.5345737591299247, "grad_norm": 3.2023568153381348, "learning_rate": 4.2434e-06, "loss": 0.035, "step": 28784 }, { "epoch": 1.5346270725595779, "grad_norm": 1.917600393295288, "learning_rate": 4.2432e-06, "loss": 0.0239, "step": 28785 }, { "epoch": 1.5346803859892306, "grad_norm": 3.8581018447875977, "learning_rate": 4.243000000000001e-06, "loss": 0.0162, "step": 28786 }, { "epoch": 1.5347336994188836, "grad_norm": 1.8058708906173706, "learning_rate": 4.2428e-06, "loss": 0.0294, "step": 28787 }, { "epoch": 1.5347870128485366, "grad_norm": 1.8360215425491333, "learning_rate": 4.242600000000001e-06, "loss": 0.0231, "step": 28788 }, { "epoch": 1.5348403262781893, "grad_norm": 1.9550570249557495, "learning_rate": 4.2424e-06, "loss": 0.0146, "step": 28789 }, { "epoch": 1.5348936397078425, "grad_norm": 1.7562766075134277, "learning_rate": 4.2422e-06, "loss": 0.0175, "step": 28790 }, { "epoch": 1.5349469531374953, "grad_norm": 0.9885984659194946, "learning_rate": 4.242e-06, "loss": 0.0182, "step": 28791 }, { "epoch": 1.5350002665671483, "grad_norm": 1.4289871454238892, "learning_rate": 4.241800000000001e-06, "loss": 0.0214, "step": 28792 }, { "epoch": 1.5350535799968013, "grad_norm": 1.4221028089523315, "learning_rate": 4.2416e-06, "loss": 0.0171, "step": 28793 }, { "epoch": 1.535106893426454, "grad_norm": 3.1997740268707275, "learning_rate": 4.241400000000001e-06, "loss": 0.0313, "step": 28794 }, { "epoch": 1.5351602068561072, "grad_norm": 2.215729236602783, "learning_rate": 4.2412e-06, "loss": 0.0269, "step": 28795 }, { "epoch": 1.53521352028576, "grad_norm": 0.4763267934322357, "learning_rate": 4.241e-06, "loss": 0.0043, "step": 28796 }, { "epoch": 1.535266833715413, "grad_norm": 1.00151526927948, "learning_rate": 4.2408e-06, "loss": 0.0119, "step": 28797 }, { "epoch": 1.535320147145066, "grad_norm": 1.3307774066925049, "learning_rate": 4.2406000000000005e-06, "loss": 0.008, "step": 28798 }, { "epoch": 1.5353734605747187, "grad_norm": 1.396652102470398, "learning_rate": 4.2404e-06, "loss": 0.0108, "step": 28799 }, { "epoch": 1.5354267740043717, "grad_norm": 1.3531664609909058, "learning_rate": 4.2402000000000005e-06, "loss": 0.0219, "step": 28800 }, { "epoch": 1.5354800874340246, "grad_norm": 0.6018425822257996, "learning_rate": 4.24e-06, "loss": 0.0051, "step": 28801 }, { "epoch": 1.5355334008636774, "grad_norm": 0.5968867540359497, "learning_rate": 4.2398000000000005e-06, "loss": 0.006, "step": 28802 }, { "epoch": 1.5355867142933306, "grad_norm": 1.542195439338684, "learning_rate": 4.2396e-06, "loss": 0.022, "step": 28803 }, { "epoch": 1.5356400277229834, "grad_norm": 1.5192214250564575, "learning_rate": 4.2394000000000004e-06, "loss": 0.0154, "step": 28804 }, { "epoch": 1.5356933411526363, "grad_norm": 1.7163785696029663, "learning_rate": 4.239200000000001e-06, "loss": 0.024, "step": 28805 }, { "epoch": 1.5357466545822893, "grad_norm": 1.081247329711914, "learning_rate": 4.239e-06, "loss": 0.009, "step": 28806 }, { "epoch": 1.535799968011942, "grad_norm": 1.765626072883606, "learning_rate": 4.2388e-06, "loss": 0.0237, "step": 28807 }, { "epoch": 1.5358532814415953, "grad_norm": 0.7141140103340149, "learning_rate": 4.2386e-06, "loss": 0.0229, "step": 28808 }, { "epoch": 1.535906594871248, "grad_norm": 2.0988998413085938, "learning_rate": 4.2384e-06, "loss": 0.0386, "step": 28809 }, { "epoch": 1.535959908300901, "grad_norm": 0.8410583734512329, "learning_rate": 4.2382e-06, "loss": 0.0078, "step": 28810 }, { "epoch": 1.536013221730554, "grad_norm": 1.2226364612579346, "learning_rate": 4.238000000000001e-06, "loss": 0.0053, "step": 28811 }, { "epoch": 1.5360665351602067, "grad_norm": 4.227477550506592, "learning_rate": 4.2378e-06, "loss": 0.0253, "step": 28812 }, { "epoch": 1.53611984858986, "grad_norm": 1.2796550989151, "learning_rate": 4.237600000000001e-06, "loss": 0.0111, "step": 28813 }, { "epoch": 1.5361731620195127, "grad_norm": 5.879891395568848, "learning_rate": 4.2374e-06, "loss": 0.0306, "step": 28814 }, { "epoch": 1.5362264754491657, "grad_norm": 2.529315233230591, "learning_rate": 4.2372e-06, "loss": 0.046, "step": 28815 }, { "epoch": 1.5362797888788187, "grad_norm": 1.280704379081726, "learning_rate": 4.237e-06, "loss": 0.0069, "step": 28816 }, { "epoch": 1.5363331023084714, "grad_norm": 1.4002079963684082, "learning_rate": 4.236800000000001e-06, "loss": 0.0207, "step": 28817 }, { "epoch": 1.5363864157381244, "grad_norm": 1.6957457065582275, "learning_rate": 4.2366e-06, "loss": 0.0388, "step": 28818 }, { "epoch": 1.5364397291677774, "grad_norm": 1.468113660812378, "learning_rate": 4.236400000000001e-06, "loss": 0.0273, "step": 28819 }, { "epoch": 1.5364930425974301, "grad_norm": 1.2137138843536377, "learning_rate": 4.2362e-06, "loss": 0.0207, "step": 28820 }, { "epoch": 1.5365463560270833, "grad_norm": 2.4086644649505615, "learning_rate": 4.236e-06, "loss": 0.012, "step": 28821 }, { "epoch": 1.536599669456736, "grad_norm": 1.5401809215545654, "learning_rate": 4.2358e-06, "loss": 0.0083, "step": 28822 }, { "epoch": 1.536652982886389, "grad_norm": 3.651257038116455, "learning_rate": 4.235600000000001e-06, "loss": 0.0378, "step": 28823 }, { "epoch": 1.536706296316042, "grad_norm": 3.9957668781280518, "learning_rate": 4.235400000000001e-06, "loss": 0.0685, "step": 28824 }, { "epoch": 1.5367596097456948, "grad_norm": 2.275947332382202, "learning_rate": 4.2352000000000005e-06, "loss": 0.0318, "step": 28825 }, { "epoch": 1.536812923175348, "grad_norm": 1.8758083581924438, "learning_rate": 4.235e-06, "loss": 0.031, "step": 28826 }, { "epoch": 1.5368662366050008, "grad_norm": 1.7383760213851929, "learning_rate": 4.2348000000000005e-06, "loss": 0.0061, "step": 28827 }, { "epoch": 1.5369195500346537, "grad_norm": 1.234533429145813, "learning_rate": 4.2346e-06, "loss": 0.007, "step": 28828 }, { "epoch": 1.5369728634643067, "grad_norm": 1.4942617416381836, "learning_rate": 4.2344000000000005e-06, "loss": 0.025, "step": 28829 }, { "epoch": 1.5370261768939595, "grad_norm": 1.9826014041900635, "learning_rate": 4.234200000000001e-06, "loss": 0.0427, "step": 28830 }, { "epoch": 1.5370794903236125, "grad_norm": 1.6450802087783813, "learning_rate": 4.2340000000000005e-06, "loss": 0.0239, "step": 28831 }, { "epoch": 1.5371328037532654, "grad_norm": 1.1552224159240723, "learning_rate": 4.2338e-06, "loss": 0.0121, "step": 28832 }, { "epoch": 1.5371861171829184, "grad_norm": 1.5581495761871338, "learning_rate": 4.2336000000000004e-06, "loss": 0.013, "step": 28833 }, { "epoch": 1.5372394306125714, "grad_norm": 3.4921014308929443, "learning_rate": 4.2334e-06, "loss": 0.0713, "step": 28834 }, { "epoch": 1.5372927440422242, "grad_norm": 2.294049024581909, "learning_rate": 4.2332e-06, "loss": 0.0414, "step": 28835 }, { "epoch": 1.5373460574718771, "grad_norm": 1.0464929342269897, "learning_rate": 4.233000000000001e-06, "loss": 0.023, "step": 28836 }, { "epoch": 1.5373993709015301, "grad_norm": 2.4091341495513916, "learning_rate": 4.2328e-06, "loss": 0.0354, "step": 28837 }, { "epoch": 1.5374526843311829, "grad_norm": 2.0185298919677734, "learning_rate": 4.232600000000001e-06, "loss": 0.0192, "step": 28838 }, { "epoch": 1.537505997760836, "grad_norm": 4.261380672454834, "learning_rate": 4.2324e-06, "loss": 0.0257, "step": 28839 }, { "epoch": 1.5375593111904888, "grad_norm": 1.0704822540283203, "learning_rate": 4.2322e-06, "loss": 0.0185, "step": 28840 }, { "epoch": 1.5376126246201418, "grad_norm": 2.957681179046631, "learning_rate": 4.232e-06, "loss": 0.0395, "step": 28841 }, { "epoch": 1.5376659380497948, "grad_norm": 0.9880802035331726, "learning_rate": 4.231800000000001e-06, "loss": 0.0104, "step": 28842 }, { "epoch": 1.5377192514794475, "grad_norm": 0.8978012204170227, "learning_rate": 4.2316e-06, "loss": 0.0079, "step": 28843 }, { "epoch": 1.5377725649091007, "grad_norm": 1.523437738418579, "learning_rate": 4.231400000000001e-06, "loss": 0.025, "step": 28844 }, { "epoch": 1.5378258783387535, "grad_norm": 0.956173300743103, "learning_rate": 4.2312e-06, "loss": 0.008, "step": 28845 }, { "epoch": 1.5378791917684065, "grad_norm": 0.9460346102714539, "learning_rate": 4.231e-06, "loss": 0.0068, "step": 28846 }, { "epoch": 1.5379325051980595, "grad_norm": 1.1954445838928223, "learning_rate": 4.2308e-06, "loss": 0.0216, "step": 28847 }, { "epoch": 1.5379858186277122, "grad_norm": 0.7474053502082825, "learning_rate": 4.230600000000001e-06, "loss": 0.0043, "step": 28848 }, { "epoch": 1.5380391320573652, "grad_norm": 1.5550904273986816, "learning_rate": 4.230400000000001e-06, "loss": 0.0333, "step": 28849 }, { "epoch": 1.5380924454870182, "grad_norm": 1.0759592056274414, "learning_rate": 4.230200000000001e-06, "loss": 0.0242, "step": 28850 }, { "epoch": 1.5381457589166712, "grad_norm": 1.3727102279663086, "learning_rate": 4.23e-06, "loss": 0.0316, "step": 28851 }, { "epoch": 1.5381990723463241, "grad_norm": 2.310635566711426, "learning_rate": 4.2298000000000006e-06, "loss": 0.0595, "step": 28852 }, { "epoch": 1.538252385775977, "grad_norm": 2.037665605545044, "learning_rate": 4.2296e-06, "loss": 0.0302, "step": 28853 }, { "epoch": 1.5383056992056299, "grad_norm": 2.0145673751831055, "learning_rate": 4.2294000000000005e-06, "loss": 0.0255, "step": 28854 }, { "epoch": 1.5383590126352829, "grad_norm": 2.146989583969116, "learning_rate": 4.229200000000001e-06, "loss": 0.0123, "step": 28855 }, { "epoch": 1.5384123260649356, "grad_norm": 1.3280766010284424, "learning_rate": 4.2290000000000005e-06, "loss": 0.0261, "step": 28856 }, { "epoch": 1.5384656394945888, "grad_norm": 1.9026681184768677, "learning_rate": 4.2288e-06, "loss": 0.0548, "step": 28857 }, { "epoch": 1.5385189529242416, "grad_norm": 0.946290910243988, "learning_rate": 4.2286000000000005e-06, "loss": 0.0061, "step": 28858 }, { "epoch": 1.5385722663538945, "grad_norm": 0.9816213250160217, "learning_rate": 4.2284e-06, "loss": 0.0074, "step": 28859 }, { "epoch": 1.5386255797835475, "grad_norm": 1.512730360031128, "learning_rate": 4.2282000000000004e-06, "loss": 0.0193, "step": 28860 }, { "epoch": 1.5386788932132003, "grad_norm": 2.816044807434082, "learning_rate": 4.228000000000001e-06, "loss": 0.0346, "step": 28861 }, { "epoch": 1.5387322066428535, "grad_norm": 1.6965415477752686, "learning_rate": 4.2278e-06, "loss": 0.0228, "step": 28862 }, { "epoch": 1.5387855200725062, "grad_norm": 1.385806918144226, "learning_rate": 4.227600000000001e-06, "loss": 0.0089, "step": 28863 }, { "epoch": 1.5388388335021592, "grad_norm": 1.1037448644638062, "learning_rate": 4.2274e-06, "loss": 0.0166, "step": 28864 }, { "epoch": 1.5388921469318122, "grad_norm": 2.520612955093384, "learning_rate": 4.2272e-06, "loss": 0.025, "step": 28865 }, { "epoch": 1.538945460361465, "grad_norm": 1.2611980438232422, "learning_rate": 4.227e-06, "loss": 0.0096, "step": 28866 }, { "epoch": 1.538998773791118, "grad_norm": 1.357958436012268, "learning_rate": 4.226800000000001e-06, "loss": 0.0142, "step": 28867 }, { "epoch": 1.539052087220771, "grad_norm": 1.031006097793579, "learning_rate": 4.2266e-06, "loss": 0.0122, "step": 28868 }, { "epoch": 1.539105400650424, "grad_norm": 0.5431146621704102, "learning_rate": 4.226400000000001e-06, "loss": 0.0061, "step": 28869 }, { "epoch": 1.5391587140800769, "grad_norm": 1.2575253248214722, "learning_rate": 4.2262e-06, "loss": 0.0177, "step": 28870 }, { "epoch": 1.5392120275097296, "grad_norm": 1.7305487394332886, "learning_rate": 4.226e-06, "loss": 0.0536, "step": 28871 }, { "epoch": 1.5392653409393826, "grad_norm": 1.1523131132125854, "learning_rate": 4.2258e-06, "loss": 0.0126, "step": 28872 }, { "epoch": 1.5393186543690356, "grad_norm": 2.3658409118652344, "learning_rate": 4.225600000000001e-06, "loss": 0.0225, "step": 28873 }, { "epoch": 1.5393719677986883, "grad_norm": 1.2227450609207153, "learning_rate": 4.2254e-06, "loss": 0.0305, "step": 28874 }, { "epoch": 1.5394252812283415, "grad_norm": 1.1834720373153687, "learning_rate": 4.225200000000001e-06, "loss": 0.0151, "step": 28875 }, { "epoch": 1.5394785946579943, "grad_norm": 1.22003173828125, "learning_rate": 4.225e-06, "loss": 0.0086, "step": 28876 }, { "epoch": 1.5395319080876473, "grad_norm": 3.385207414627075, "learning_rate": 4.224800000000001e-06, "loss": 0.0657, "step": 28877 }, { "epoch": 1.5395852215173003, "grad_norm": 2.4354088306427, "learning_rate": 4.2246e-06, "loss": 0.0202, "step": 28878 }, { "epoch": 1.539638534946953, "grad_norm": 1.9939931631088257, "learning_rate": 4.224400000000001e-06, "loss": 0.034, "step": 28879 }, { "epoch": 1.5396918483766062, "grad_norm": 1.455534815788269, "learning_rate": 4.2242e-06, "loss": 0.0408, "step": 28880 }, { "epoch": 1.539745161806259, "grad_norm": 1.31309175491333, "learning_rate": 4.2240000000000006e-06, "loss": 0.023, "step": 28881 }, { "epoch": 1.539798475235912, "grad_norm": 1.4520626068115234, "learning_rate": 4.2238e-06, "loss": 0.02, "step": 28882 }, { "epoch": 1.539851788665565, "grad_norm": 2.5157406330108643, "learning_rate": 4.2236000000000005e-06, "loss": 0.0482, "step": 28883 }, { "epoch": 1.5399051020952177, "grad_norm": 1.569138526916504, "learning_rate": 4.2234e-06, "loss": 0.0368, "step": 28884 }, { "epoch": 1.5399584155248707, "grad_norm": 1.015255331993103, "learning_rate": 4.2232000000000005e-06, "loss": 0.0213, "step": 28885 }, { "epoch": 1.5400117289545237, "grad_norm": 2.5674846172332764, "learning_rate": 4.223e-06, "loss": 0.0253, "step": 28886 }, { "epoch": 1.5400650423841764, "grad_norm": 1.0935907363891602, "learning_rate": 4.2228000000000005e-06, "loss": 0.0091, "step": 28887 }, { "epoch": 1.5401183558138296, "grad_norm": 1.8839969635009766, "learning_rate": 4.222600000000001e-06, "loss": 0.0174, "step": 28888 }, { "epoch": 1.5401716692434824, "grad_norm": 2.800539255142212, "learning_rate": 4.2224000000000004e-06, "loss": 0.0442, "step": 28889 }, { "epoch": 1.5402249826731353, "grad_norm": 1.7171223163604736, "learning_rate": 4.2222e-06, "loss": 0.028, "step": 28890 }, { "epoch": 1.5402782961027883, "grad_norm": 1.3149685859680176, "learning_rate": 4.222e-06, "loss": 0.0223, "step": 28891 }, { "epoch": 1.540331609532441, "grad_norm": 2.270582675933838, "learning_rate": 4.2218e-06, "loss": 0.0267, "step": 28892 }, { "epoch": 1.5403849229620943, "grad_norm": 0.9849756360054016, "learning_rate": 4.2216e-06, "loss": 0.0211, "step": 28893 }, { "epoch": 1.540438236391747, "grad_norm": 3.182790756225586, "learning_rate": 4.221400000000001e-06, "loss": 0.0119, "step": 28894 }, { "epoch": 1.5404915498214, "grad_norm": 1.7092714309692383, "learning_rate": 4.2212e-06, "loss": 0.0372, "step": 28895 }, { "epoch": 1.540544863251053, "grad_norm": 1.2569897174835205, "learning_rate": 4.221e-06, "loss": 0.0172, "step": 28896 }, { "epoch": 1.5405981766807058, "grad_norm": 1.1543351411819458, "learning_rate": 4.2208e-06, "loss": 0.0319, "step": 28897 }, { "epoch": 1.540651490110359, "grad_norm": 0.950128972530365, "learning_rate": 4.2206e-06, "loss": 0.0066, "step": 28898 }, { "epoch": 1.5407048035400117, "grad_norm": 1.7685725688934326, "learning_rate": 4.2204e-06, "loss": 0.031, "step": 28899 }, { "epoch": 1.5407581169696647, "grad_norm": 3.450251817703247, "learning_rate": 4.220200000000001e-06, "loss": 0.0272, "step": 28900 }, { "epoch": 1.5408114303993177, "grad_norm": 1.4055266380310059, "learning_rate": 4.22e-06, "loss": 0.0166, "step": 28901 }, { "epoch": 1.5408647438289704, "grad_norm": 1.6885393857955933, "learning_rate": 4.219800000000001e-06, "loss": 0.0221, "step": 28902 }, { "epoch": 1.5409180572586234, "grad_norm": 1.8637514114379883, "learning_rate": 4.2196e-06, "loss": 0.0352, "step": 28903 }, { "epoch": 1.5409713706882764, "grad_norm": 0.9194411039352417, "learning_rate": 4.2194e-06, "loss": 0.0115, "step": 28904 }, { "epoch": 1.5410246841179291, "grad_norm": 0.9284473061561584, "learning_rate": 4.2192e-06, "loss": 0.0156, "step": 28905 }, { "epoch": 1.5410779975475823, "grad_norm": 2.5135209560394287, "learning_rate": 4.219000000000001e-06, "loss": 0.0207, "step": 28906 }, { "epoch": 1.541131310977235, "grad_norm": 0.8563086986541748, "learning_rate": 4.2188e-06, "loss": 0.0085, "step": 28907 }, { "epoch": 1.541184624406888, "grad_norm": 1.6932119131088257, "learning_rate": 4.2186000000000006e-06, "loss": 0.0204, "step": 28908 }, { "epoch": 1.541237937836541, "grad_norm": 0.5530611276626587, "learning_rate": 4.2184e-06, "loss": 0.0044, "step": 28909 }, { "epoch": 1.5412912512661938, "grad_norm": 1.7924749851226807, "learning_rate": 4.2182e-06, "loss": 0.0251, "step": 28910 }, { "epoch": 1.541344564695847, "grad_norm": 1.570513367652893, "learning_rate": 4.218e-06, "loss": 0.0105, "step": 28911 }, { "epoch": 1.5413978781254998, "grad_norm": 2.7015492916107178, "learning_rate": 4.2178000000000005e-06, "loss": 0.0471, "step": 28912 }, { "epoch": 1.5414511915551528, "grad_norm": 1.2348748445510864, "learning_rate": 4.217600000000001e-06, "loss": 0.0123, "step": 28913 }, { "epoch": 1.5415045049848057, "grad_norm": 2.0514583587646484, "learning_rate": 4.2174000000000005e-06, "loss": 0.0427, "step": 28914 }, { "epoch": 1.5415578184144585, "grad_norm": 0.8756312131881714, "learning_rate": 4.2172e-06, "loss": 0.0121, "step": 28915 }, { "epoch": 1.5416111318441117, "grad_norm": 2.3031537532806396, "learning_rate": 4.2170000000000005e-06, "loss": 0.0463, "step": 28916 }, { "epoch": 1.5416644452737644, "grad_norm": 1.6711146831512451, "learning_rate": 4.2168e-06, "loss": 0.0206, "step": 28917 }, { "epoch": 1.5417177587034174, "grad_norm": 1.3372074365615845, "learning_rate": 4.2166e-06, "loss": 0.0129, "step": 28918 }, { "epoch": 1.5417710721330704, "grad_norm": 1.6690471172332764, "learning_rate": 4.216400000000001e-06, "loss": 0.0065, "step": 28919 }, { "epoch": 1.5418243855627232, "grad_norm": 2.160142183303833, "learning_rate": 4.2162e-06, "loss": 0.017, "step": 28920 }, { "epoch": 1.5418776989923761, "grad_norm": 1.5603840351104736, "learning_rate": 4.216e-06, "loss": 0.0148, "step": 28921 }, { "epoch": 1.5419310124220291, "grad_norm": 2.0681233406066895, "learning_rate": 4.2158e-06, "loss": 0.0256, "step": 28922 }, { "epoch": 1.5419843258516819, "grad_norm": 2.751702070236206, "learning_rate": 4.2156e-06, "loss": 0.0375, "step": 28923 }, { "epoch": 1.542037639281335, "grad_norm": 1.5900804996490479, "learning_rate": 4.2154e-06, "loss": 0.0135, "step": 28924 }, { "epoch": 1.5420909527109878, "grad_norm": 1.237686276435852, "learning_rate": 4.215200000000001e-06, "loss": 0.0188, "step": 28925 }, { "epoch": 1.5421442661406408, "grad_norm": 1.2655997276306152, "learning_rate": 4.215e-06, "loss": 0.031, "step": 28926 }, { "epoch": 1.5421975795702938, "grad_norm": 1.9695183038711548, "learning_rate": 4.214800000000001e-06, "loss": 0.0061, "step": 28927 }, { "epoch": 1.5422508929999466, "grad_norm": 3.4057912826538086, "learning_rate": 4.2146e-06, "loss": 0.0156, "step": 28928 }, { "epoch": 1.5423042064295998, "grad_norm": 1.3786160945892334, "learning_rate": 4.2144e-06, "loss": 0.0106, "step": 28929 }, { "epoch": 1.5423575198592525, "grad_norm": 1.8569422960281372, "learning_rate": 4.2142e-06, "loss": 0.0211, "step": 28930 }, { "epoch": 1.5424108332889055, "grad_norm": 1.8503642082214355, "learning_rate": 4.214000000000001e-06, "loss": 0.0083, "step": 28931 }, { "epoch": 1.5424641467185585, "grad_norm": 2.093740701675415, "learning_rate": 4.2138e-06, "loss": 0.0199, "step": 28932 }, { "epoch": 1.5425174601482112, "grad_norm": 2.681704521179199, "learning_rate": 4.213600000000001e-06, "loss": 0.049, "step": 28933 }, { "epoch": 1.5425707735778644, "grad_norm": 4.53609561920166, "learning_rate": 4.2134e-06, "loss": 0.0333, "step": 28934 }, { "epoch": 1.5426240870075172, "grad_norm": 4.683626174926758, "learning_rate": 4.2132e-06, "loss": 0.0463, "step": 28935 }, { "epoch": 1.5426774004371702, "grad_norm": 1.410813570022583, "learning_rate": 4.213e-06, "loss": 0.0161, "step": 28936 }, { "epoch": 1.5427307138668231, "grad_norm": 1.8530031442642212, "learning_rate": 4.2128000000000006e-06, "loss": 0.0161, "step": 28937 }, { "epoch": 1.542784027296476, "grad_norm": 1.4623206853866577, "learning_rate": 4.212600000000001e-06, "loss": 0.0079, "step": 28938 }, { "epoch": 1.5428373407261289, "grad_norm": 1.4518632888793945, "learning_rate": 4.2124000000000005e-06, "loss": 0.0384, "step": 28939 }, { "epoch": 1.5428906541557819, "grad_norm": 1.6393017768859863, "learning_rate": 4.2122e-06, "loss": 0.0219, "step": 28940 }, { "epoch": 1.5429439675854346, "grad_norm": 5.12058162689209, "learning_rate": 4.2120000000000005e-06, "loss": 0.0263, "step": 28941 }, { "epoch": 1.5429972810150878, "grad_norm": 2.721787452697754, "learning_rate": 4.2118e-06, "loss": 0.026, "step": 28942 }, { "epoch": 1.5430505944447406, "grad_norm": 2.2272653579711914, "learning_rate": 4.2116000000000005e-06, "loss": 0.0229, "step": 28943 }, { "epoch": 1.5431039078743936, "grad_norm": 2.1248083114624023, "learning_rate": 4.211400000000001e-06, "loss": 0.0226, "step": 28944 }, { "epoch": 1.5431572213040465, "grad_norm": 1.746556282043457, "learning_rate": 4.2112000000000004e-06, "loss": 0.0247, "step": 28945 }, { "epoch": 1.5432105347336993, "grad_norm": 1.3304387331008911, "learning_rate": 4.211e-06, "loss": 0.0116, "step": 28946 }, { "epoch": 1.5432638481633525, "grad_norm": 1.763840913772583, "learning_rate": 4.2108e-06, "loss": 0.0063, "step": 28947 }, { "epoch": 1.5433171615930052, "grad_norm": 1.1401296854019165, "learning_rate": 4.2106e-06, "loss": 0.0134, "step": 28948 }, { "epoch": 1.5433704750226582, "grad_norm": 2.050173044204712, "learning_rate": 4.2104e-06, "loss": 0.0264, "step": 28949 }, { "epoch": 1.5434237884523112, "grad_norm": 2.7187554836273193, "learning_rate": 4.210200000000001e-06, "loss": 0.0392, "step": 28950 }, { "epoch": 1.543477101881964, "grad_norm": 2.3167226314544678, "learning_rate": 4.21e-06, "loss": 0.0091, "step": 28951 }, { "epoch": 1.543530415311617, "grad_norm": 2.0693576335906982, "learning_rate": 4.209800000000001e-06, "loss": 0.055, "step": 28952 }, { "epoch": 1.54358372874127, "grad_norm": 1.7453471422195435, "learning_rate": 4.2096e-06, "loss": 0.0223, "step": 28953 }, { "epoch": 1.543637042170923, "grad_norm": 1.4177173376083374, "learning_rate": 4.2094e-06, "loss": 0.0226, "step": 28954 }, { "epoch": 1.5436903556005759, "grad_norm": 1.15599524974823, "learning_rate": 4.2092e-06, "loss": 0.0143, "step": 28955 }, { "epoch": 1.5437436690302286, "grad_norm": 2.250298500061035, "learning_rate": 4.209000000000001e-06, "loss": 0.0184, "step": 28956 }, { "epoch": 1.5437969824598816, "grad_norm": 1.1918500661849976, "learning_rate": 4.2088e-06, "loss": 0.0167, "step": 28957 }, { "epoch": 1.5438502958895346, "grad_norm": 1.8011890649795532, "learning_rate": 4.208600000000001e-06, "loss": 0.0369, "step": 28958 }, { "epoch": 1.5439036093191874, "grad_norm": 2.2000515460968018, "learning_rate": 4.2084e-06, "loss": 0.0235, "step": 28959 }, { "epoch": 1.5439569227488406, "grad_norm": 1.0437949895858765, "learning_rate": 4.208200000000001e-06, "loss": 0.008, "step": 28960 }, { "epoch": 1.5440102361784933, "grad_norm": 1.5096410512924194, "learning_rate": 4.208e-06, "loss": 0.0064, "step": 28961 }, { "epoch": 1.5440635496081463, "grad_norm": 0.841382622718811, "learning_rate": 4.207800000000001e-06, "loss": 0.0158, "step": 28962 }, { "epoch": 1.5441168630377993, "grad_norm": 0.8183812499046326, "learning_rate": 4.2076e-06, "loss": 0.012, "step": 28963 }, { "epoch": 1.544170176467452, "grad_norm": 1.3024080991744995, "learning_rate": 4.207400000000001e-06, "loss": 0.011, "step": 28964 }, { "epoch": 1.5442234898971052, "grad_norm": 1.4119482040405273, "learning_rate": 4.2072e-06, "loss": 0.0054, "step": 28965 }, { "epoch": 1.544276803326758, "grad_norm": 1.1408289670944214, "learning_rate": 4.2070000000000005e-06, "loss": 0.0169, "step": 28966 }, { "epoch": 1.544330116756411, "grad_norm": 0.7289817929267883, "learning_rate": 4.2068e-06, "loss": 0.0171, "step": 28967 }, { "epoch": 1.544383430186064, "grad_norm": 2.1919450759887695, "learning_rate": 4.2066000000000005e-06, "loss": 0.0103, "step": 28968 }, { "epoch": 1.5444367436157167, "grad_norm": 1.1179019212722778, "learning_rate": 4.2064e-06, "loss": 0.028, "step": 28969 }, { "epoch": 1.5444900570453697, "grad_norm": 1.311890959739685, "learning_rate": 4.2062000000000005e-06, "loss": 0.0209, "step": 28970 }, { "epoch": 1.5445433704750227, "grad_norm": 1.317328691482544, "learning_rate": 4.206e-06, "loss": 0.0203, "step": 28971 }, { "epoch": 1.5445966839046756, "grad_norm": 1.3629367351531982, "learning_rate": 4.2058000000000005e-06, "loss": 0.0158, "step": 28972 }, { "epoch": 1.5446499973343286, "grad_norm": 0.28693631291389465, "learning_rate": 4.2056e-06, "loss": 0.0029, "step": 28973 }, { "epoch": 1.5447033107639814, "grad_norm": 2.0650405883789062, "learning_rate": 4.2054000000000004e-06, "loss": 0.0254, "step": 28974 }, { "epoch": 1.5447566241936344, "grad_norm": 1.2911875247955322, "learning_rate": 4.2052e-06, "loss": 0.0301, "step": 28975 }, { "epoch": 1.5448099376232873, "grad_norm": 1.5351617336273193, "learning_rate": 4.205e-06, "loss": 0.0194, "step": 28976 }, { "epoch": 1.54486325105294, "grad_norm": 0.7183861136436462, "learning_rate": 4.204800000000001e-06, "loss": 0.0059, "step": 28977 }, { "epoch": 1.5449165644825933, "grad_norm": 0.7721753120422363, "learning_rate": 4.2046e-06, "loss": 0.0123, "step": 28978 }, { "epoch": 1.544969877912246, "grad_norm": 1.8414024114608765, "learning_rate": 4.2044e-06, "loss": 0.0311, "step": 28979 }, { "epoch": 1.545023191341899, "grad_norm": 1.0552324056625366, "learning_rate": 4.2042e-06, "loss": 0.0178, "step": 28980 }, { "epoch": 1.545076504771552, "grad_norm": 1.5795613527297974, "learning_rate": 4.204e-06, "loss": 0.0241, "step": 28981 }, { "epoch": 1.5451298182012048, "grad_norm": 0.5773969292640686, "learning_rate": 4.2038e-06, "loss": 0.0061, "step": 28982 }, { "epoch": 1.545183131630858, "grad_norm": 1.582626223564148, "learning_rate": 4.203600000000001e-06, "loss": 0.0153, "step": 28983 }, { "epoch": 1.5452364450605107, "grad_norm": 1.3440569639205933, "learning_rate": 4.2034e-06, "loss": 0.0235, "step": 28984 }, { "epoch": 1.5452897584901637, "grad_norm": 1.8609144687652588, "learning_rate": 4.203200000000001e-06, "loss": 0.0316, "step": 28985 }, { "epoch": 1.5453430719198167, "grad_norm": 1.243751049041748, "learning_rate": 4.203e-06, "loss": 0.014, "step": 28986 }, { "epoch": 1.5453963853494694, "grad_norm": 1.0296635627746582, "learning_rate": 4.2028e-06, "loss": 0.0063, "step": 28987 }, { "epoch": 1.5454496987791224, "grad_norm": 0.30391380190849304, "learning_rate": 4.2026e-06, "loss": 0.0032, "step": 28988 }, { "epoch": 1.5455030122087754, "grad_norm": 3.0530269145965576, "learning_rate": 4.202400000000001e-06, "loss": 0.0391, "step": 28989 }, { "epoch": 1.5455563256384284, "grad_norm": 0.9202097654342651, "learning_rate": 4.2022e-06, "loss": 0.0158, "step": 28990 }, { "epoch": 1.5456096390680814, "grad_norm": 1.5140321254730225, "learning_rate": 4.202000000000001e-06, "loss": 0.0228, "step": 28991 }, { "epoch": 1.545662952497734, "grad_norm": 2.3191986083984375, "learning_rate": 4.2018e-06, "loss": 0.0103, "step": 28992 }, { "epoch": 1.545716265927387, "grad_norm": 1.6321139335632324, "learning_rate": 4.2016e-06, "loss": 0.0292, "step": 28993 }, { "epoch": 1.54576957935704, "grad_norm": 0.9537072777748108, "learning_rate": 4.2014e-06, "loss": 0.0082, "step": 28994 }, { "epoch": 1.5458228927866928, "grad_norm": 0.960580587387085, "learning_rate": 4.2012000000000005e-06, "loss": 0.0177, "step": 28995 }, { "epoch": 1.545876206216346, "grad_norm": 1.022364616394043, "learning_rate": 4.201e-06, "loss": 0.0098, "step": 28996 }, { "epoch": 1.5459295196459988, "grad_norm": 2.1184000968933105, "learning_rate": 4.2008000000000005e-06, "loss": 0.0243, "step": 28997 }, { "epoch": 1.5459828330756518, "grad_norm": 0.8546081185340881, "learning_rate": 4.2006e-06, "loss": 0.0194, "step": 28998 }, { "epoch": 1.5460361465053047, "grad_norm": 1.6077582836151123, "learning_rate": 4.2004000000000005e-06, "loss": 0.0312, "step": 28999 }, { "epoch": 1.5460894599349575, "grad_norm": 1.6010686159133911, "learning_rate": 4.2002e-06, "loss": 0.0238, "step": 29000 }, { "epoch": 1.5461427733646107, "grad_norm": 0.6920756697654724, "learning_rate": 4.2000000000000004e-06, "loss": 0.0042, "step": 29001 }, { "epoch": 1.5461960867942635, "grad_norm": 2.2400269508361816, "learning_rate": 4.199800000000001e-06, "loss": 0.0143, "step": 29002 }, { "epoch": 1.5462494002239164, "grad_norm": 2.837989568710327, "learning_rate": 4.1996e-06, "loss": 0.0285, "step": 29003 }, { "epoch": 1.5463027136535694, "grad_norm": 3.55228853225708, "learning_rate": 4.1994e-06, "loss": 0.0399, "step": 29004 }, { "epoch": 1.5463560270832222, "grad_norm": 2.320166826248169, "learning_rate": 4.1992e-06, "loss": 0.0236, "step": 29005 }, { "epoch": 1.5464093405128752, "grad_norm": 4.753217697143555, "learning_rate": 4.199e-06, "loss": 0.066, "step": 29006 }, { "epoch": 1.5464626539425281, "grad_norm": 1.6841862201690674, "learning_rate": 4.1988e-06, "loss": 0.0218, "step": 29007 }, { "epoch": 1.5465159673721809, "grad_norm": 2.0957131385803223, "learning_rate": 4.198600000000001e-06, "loss": 0.013, "step": 29008 }, { "epoch": 1.546569280801834, "grad_norm": 2.3006341457366943, "learning_rate": 4.1984e-06, "loss": 0.0198, "step": 29009 }, { "epoch": 1.5466225942314868, "grad_norm": 1.7116957902908325, "learning_rate": 4.198200000000001e-06, "loss": 0.0161, "step": 29010 }, { "epoch": 1.5466759076611398, "grad_norm": 1.6615904569625854, "learning_rate": 4.198e-06, "loss": 0.0202, "step": 29011 }, { "epoch": 1.5467292210907928, "grad_norm": 2.1391873359680176, "learning_rate": 4.1978e-06, "loss": 0.0508, "step": 29012 }, { "epoch": 1.5467825345204456, "grad_norm": 1.5791243314743042, "learning_rate": 4.1976e-06, "loss": 0.0099, "step": 29013 }, { "epoch": 1.5468358479500988, "grad_norm": 2.4751100540161133, "learning_rate": 4.197400000000001e-06, "loss": 0.0131, "step": 29014 }, { "epoch": 1.5468891613797515, "grad_norm": 2.372891664505005, "learning_rate": 4.1972e-06, "loss": 0.04, "step": 29015 }, { "epoch": 1.5469424748094045, "grad_norm": 2.002747058868408, "learning_rate": 4.197000000000001e-06, "loss": 0.0153, "step": 29016 }, { "epoch": 1.5469957882390575, "grad_norm": 2.7444047927856445, "learning_rate": 4.1968e-06, "loss": 0.0364, "step": 29017 }, { "epoch": 1.5470491016687102, "grad_norm": 1.6053191423416138, "learning_rate": 4.1966e-06, "loss": 0.042, "step": 29018 }, { "epoch": 1.5471024150983634, "grad_norm": 1.4910423755645752, "learning_rate": 4.1964e-06, "loss": 0.0106, "step": 29019 }, { "epoch": 1.5471557285280162, "grad_norm": 1.4504215717315674, "learning_rate": 4.196200000000001e-06, "loss": 0.0284, "step": 29020 }, { "epoch": 1.5472090419576692, "grad_norm": 1.3084917068481445, "learning_rate": 4.196e-06, "loss": 0.0091, "step": 29021 }, { "epoch": 1.5472623553873222, "grad_norm": 0.9849470257759094, "learning_rate": 4.1958000000000006e-06, "loss": 0.0101, "step": 29022 }, { "epoch": 1.547315668816975, "grad_norm": 1.4845675230026245, "learning_rate": 4.1956e-06, "loss": 0.0142, "step": 29023 }, { "epoch": 1.5473689822466279, "grad_norm": 1.861608624458313, "learning_rate": 4.1954000000000005e-06, "loss": 0.011, "step": 29024 }, { "epoch": 1.5474222956762809, "grad_norm": 1.209081768989563, "learning_rate": 4.1952e-06, "loss": 0.0336, "step": 29025 }, { "epoch": 1.5474756091059336, "grad_norm": 3.109740972518921, "learning_rate": 4.1950000000000005e-06, "loss": 0.0199, "step": 29026 }, { "epoch": 1.5475289225355868, "grad_norm": 1.6347670555114746, "learning_rate": 4.194800000000001e-06, "loss": 0.0174, "step": 29027 }, { "epoch": 1.5475822359652396, "grad_norm": 0.8318462371826172, "learning_rate": 4.1946000000000005e-06, "loss": 0.0076, "step": 29028 }, { "epoch": 1.5476355493948926, "grad_norm": 1.8131290674209595, "learning_rate": 4.1944e-06, "loss": 0.021, "step": 29029 }, { "epoch": 1.5476888628245455, "grad_norm": 1.6560404300689697, "learning_rate": 4.1942000000000004e-06, "loss": 0.0211, "step": 29030 }, { "epoch": 1.5477421762541983, "grad_norm": 1.7917289733886719, "learning_rate": 4.194e-06, "loss": 0.0091, "step": 29031 }, { "epoch": 1.5477954896838515, "grad_norm": 0.49711617827415466, "learning_rate": 4.1938e-06, "loss": 0.0024, "step": 29032 }, { "epoch": 1.5478488031135043, "grad_norm": 1.7502334117889404, "learning_rate": 4.193600000000001e-06, "loss": 0.0167, "step": 29033 }, { "epoch": 1.5479021165431572, "grad_norm": 1.7332829236984253, "learning_rate": 4.1934e-06, "loss": 0.0164, "step": 29034 }, { "epoch": 1.5479554299728102, "grad_norm": 1.8027094602584839, "learning_rate": 4.193200000000001e-06, "loss": 0.0167, "step": 29035 }, { "epoch": 1.548008743402463, "grad_norm": 0.9198253750801086, "learning_rate": 4.193e-06, "loss": 0.0094, "step": 29036 }, { "epoch": 1.5480620568321162, "grad_norm": 1.1142829656600952, "learning_rate": 4.1928e-06, "loss": 0.0106, "step": 29037 }, { "epoch": 1.548115370261769, "grad_norm": 2.0796961784362793, "learning_rate": 4.1926e-06, "loss": 0.0345, "step": 29038 }, { "epoch": 1.548168683691422, "grad_norm": 2.146028518676758, "learning_rate": 4.192400000000001e-06, "loss": 0.0327, "step": 29039 }, { "epoch": 1.5482219971210749, "grad_norm": 1.700440526008606, "learning_rate": 4.1922e-06, "loss": 0.0192, "step": 29040 }, { "epoch": 1.5482753105507276, "grad_norm": 0.9331300258636475, "learning_rate": 4.192000000000001e-06, "loss": 0.0118, "step": 29041 }, { "epoch": 1.5483286239803806, "grad_norm": 2.757976770401001, "learning_rate": 4.1918e-06, "loss": 0.0449, "step": 29042 }, { "epoch": 1.5483819374100336, "grad_norm": 1.1090749502182007, "learning_rate": 4.1916e-06, "loss": 0.0088, "step": 29043 }, { "epoch": 1.5484352508396864, "grad_norm": 3.2306692600250244, "learning_rate": 4.1914e-06, "loss": 0.0269, "step": 29044 }, { "epoch": 1.5484885642693396, "grad_norm": 1.6800496578216553, "learning_rate": 4.191200000000001e-06, "loss": 0.0161, "step": 29045 }, { "epoch": 1.5485418776989923, "grad_norm": 2.3335814476013184, "learning_rate": 4.191e-06, "loss": 0.0547, "step": 29046 }, { "epoch": 1.5485951911286453, "grad_norm": 2.5576391220092773, "learning_rate": 4.190800000000001e-06, "loss": 0.0201, "step": 29047 }, { "epoch": 1.5486485045582983, "grad_norm": 1.064894437789917, "learning_rate": 4.1906e-06, "loss": 0.0079, "step": 29048 }, { "epoch": 1.548701817987951, "grad_norm": 0.9747065901756287, "learning_rate": 4.1904000000000006e-06, "loss": 0.0167, "step": 29049 }, { "epoch": 1.5487551314176042, "grad_norm": 0.49671563506126404, "learning_rate": 4.1902e-06, "loss": 0.0045, "step": 29050 }, { "epoch": 1.548808444847257, "grad_norm": 1.20022714138031, "learning_rate": 4.1900000000000005e-06, "loss": 0.0276, "step": 29051 }, { "epoch": 1.54886175827691, "grad_norm": 0.46575435996055603, "learning_rate": 4.189800000000001e-06, "loss": 0.0047, "step": 29052 }, { "epoch": 1.548915071706563, "grad_norm": 1.4157720804214478, "learning_rate": 4.1896000000000005e-06, "loss": 0.0142, "step": 29053 }, { "epoch": 1.5489683851362157, "grad_norm": 1.979011058807373, "learning_rate": 4.1894e-06, "loss": 0.0295, "step": 29054 }, { "epoch": 1.5490216985658687, "grad_norm": 0.7206957340240479, "learning_rate": 4.1892000000000005e-06, "loss": 0.0056, "step": 29055 }, { "epoch": 1.5490750119955217, "grad_norm": 1.852990984916687, "learning_rate": 4.189e-06, "loss": 0.0244, "step": 29056 }, { "epoch": 1.5491283254251746, "grad_norm": 0.5870771408081055, "learning_rate": 4.1888000000000005e-06, "loss": 0.0059, "step": 29057 }, { "epoch": 1.5491816388548276, "grad_norm": 1.6975493431091309, "learning_rate": 4.188600000000001e-06, "loss": 0.0275, "step": 29058 }, { "epoch": 1.5492349522844804, "grad_norm": 1.6419198513031006, "learning_rate": 4.1884e-06, "loss": 0.0212, "step": 29059 }, { "epoch": 1.5492882657141334, "grad_norm": 0.3923719525337219, "learning_rate": 4.188200000000001e-06, "loss": 0.004, "step": 29060 }, { "epoch": 1.5493415791437863, "grad_norm": 2.2250659465789795, "learning_rate": 4.188e-06, "loss": 0.0192, "step": 29061 }, { "epoch": 1.549394892573439, "grad_norm": 2.034919023513794, "learning_rate": 4.1878e-06, "loss": 0.0243, "step": 29062 }, { "epoch": 1.5494482060030923, "grad_norm": 1.8546127080917358, "learning_rate": 4.1876e-06, "loss": 0.0282, "step": 29063 }, { "epoch": 1.549501519432745, "grad_norm": 0.9741160869598389, "learning_rate": 4.187400000000001e-06, "loss": 0.0173, "step": 29064 }, { "epoch": 1.549554832862398, "grad_norm": 4.871732711791992, "learning_rate": 4.1872e-06, "loss": 0.0238, "step": 29065 }, { "epoch": 1.549608146292051, "grad_norm": 0.8918915390968323, "learning_rate": 4.187000000000001e-06, "loss": 0.005, "step": 29066 }, { "epoch": 1.5496614597217038, "grad_norm": 0.36436545848846436, "learning_rate": 4.1868e-06, "loss": 0.0045, "step": 29067 }, { "epoch": 1.549714773151357, "grad_norm": 2.9678447246551514, "learning_rate": 4.1866e-06, "loss": 0.0474, "step": 29068 }, { "epoch": 1.5497680865810097, "grad_norm": 1.802133321762085, "learning_rate": 4.1864e-06, "loss": 0.0261, "step": 29069 }, { "epoch": 1.5498214000106627, "grad_norm": 2.5074007511138916, "learning_rate": 4.186200000000001e-06, "loss": 0.0296, "step": 29070 }, { "epoch": 1.5498747134403157, "grad_norm": 2.2997214794158936, "learning_rate": 4.186e-06, "loss": 0.0131, "step": 29071 }, { "epoch": 1.5499280268699684, "grad_norm": 2.5023210048675537, "learning_rate": 4.185800000000001e-06, "loss": 0.0362, "step": 29072 }, { "epoch": 1.5499813402996214, "grad_norm": 1.6139339208602905, "learning_rate": 4.1856e-06, "loss": 0.0122, "step": 29073 }, { "epoch": 1.5500346537292744, "grad_norm": 1.2349928617477417, "learning_rate": 4.185400000000001e-06, "loss": 0.0185, "step": 29074 }, { "epoch": 1.5500879671589274, "grad_norm": 1.063153624534607, "learning_rate": 4.1852e-06, "loss": 0.0211, "step": 29075 }, { "epoch": 1.5501412805885804, "grad_norm": 0.5603138208389282, "learning_rate": 4.185000000000001e-06, "loss": 0.0052, "step": 29076 }, { "epoch": 1.5501945940182331, "grad_norm": 2.0131757259368896, "learning_rate": 4.1848e-06, "loss": 0.0362, "step": 29077 }, { "epoch": 1.550247907447886, "grad_norm": 1.628807544708252, "learning_rate": 4.1846000000000006e-06, "loss": 0.0095, "step": 29078 }, { "epoch": 1.550301220877539, "grad_norm": 2.8652255535125732, "learning_rate": 4.1844e-06, "loss": 0.0348, "step": 29079 }, { "epoch": 1.5503545343071918, "grad_norm": 1.26472008228302, "learning_rate": 4.1842000000000005e-06, "loss": 0.009, "step": 29080 }, { "epoch": 1.550407847736845, "grad_norm": 1.2001707553863525, "learning_rate": 4.184e-06, "loss": 0.0246, "step": 29081 }, { "epoch": 1.5504611611664978, "grad_norm": 1.404873251914978, "learning_rate": 4.1838000000000005e-06, "loss": 0.0158, "step": 29082 }, { "epoch": 1.5505144745961508, "grad_norm": 0.6339133977890015, "learning_rate": 4.1836e-06, "loss": 0.0052, "step": 29083 }, { "epoch": 1.5505677880258038, "grad_norm": 1.449077844619751, "learning_rate": 4.1834000000000005e-06, "loss": 0.0152, "step": 29084 }, { "epoch": 1.5506211014554565, "grad_norm": 2.1327455043792725, "learning_rate": 4.183200000000001e-06, "loss": 0.042, "step": 29085 }, { "epoch": 1.5506744148851097, "grad_norm": 1.497879147529602, "learning_rate": 4.1830000000000004e-06, "loss": 0.0115, "step": 29086 }, { "epoch": 1.5507277283147625, "grad_norm": 2.2384707927703857, "learning_rate": 4.1828e-06, "loss": 0.0183, "step": 29087 }, { "epoch": 1.5507810417444154, "grad_norm": 1.7306588888168335, "learning_rate": 4.1826e-06, "loss": 0.0148, "step": 29088 }, { "epoch": 1.5508343551740684, "grad_norm": 1.427754521369934, "learning_rate": 4.1824e-06, "loss": 0.0101, "step": 29089 }, { "epoch": 1.5508876686037212, "grad_norm": 1.4288960695266724, "learning_rate": 4.1822e-06, "loss": 0.0257, "step": 29090 }, { "epoch": 1.5509409820333742, "grad_norm": 1.013541579246521, "learning_rate": 4.182000000000001e-06, "loss": 0.0056, "step": 29091 }, { "epoch": 1.5509942954630271, "grad_norm": 3.5558509826660156, "learning_rate": 4.1818e-06, "loss": 0.0269, "step": 29092 }, { "epoch": 1.5510476088926801, "grad_norm": 1.327260136604309, "learning_rate": 4.1816e-06, "loss": 0.0144, "step": 29093 }, { "epoch": 1.551100922322333, "grad_norm": 1.0727794170379639, "learning_rate": 4.1814e-06, "loss": 0.0371, "step": 29094 }, { "epoch": 1.5511542357519859, "grad_norm": 1.8052253723144531, "learning_rate": 4.1812e-06, "loss": 0.0207, "step": 29095 }, { "epoch": 1.5512075491816388, "grad_norm": 1.7523375749588013, "learning_rate": 4.181e-06, "loss": 0.0249, "step": 29096 }, { "epoch": 1.5512608626112918, "grad_norm": 0.7992540597915649, "learning_rate": 4.180800000000001e-06, "loss": 0.006, "step": 29097 }, { "epoch": 1.5513141760409446, "grad_norm": 2.5555953979492188, "learning_rate": 4.1806e-06, "loss": 0.0269, "step": 29098 }, { "epoch": 1.5513674894705978, "grad_norm": 1.8309921026229858, "learning_rate": 4.180400000000001e-06, "loss": 0.0187, "step": 29099 }, { "epoch": 1.5514208029002505, "grad_norm": 1.5321966409683228, "learning_rate": 4.1802e-06, "loss": 0.0219, "step": 29100 }, { "epoch": 1.5514741163299035, "grad_norm": 0.944710373878479, "learning_rate": 4.18e-06, "loss": 0.0113, "step": 29101 }, { "epoch": 1.5515274297595565, "grad_norm": 2.3051095008850098, "learning_rate": 4.1798e-06, "loss": 0.025, "step": 29102 }, { "epoch": 1.5515807431892092, "grad_norm": 1.2009860277175903, "learning_rate": 4.179600000000001e-06, "loss": 0.0126, "step": 29103 }, { "epoch": 1.5516340566188624, "grad_norm": 2.0641844272613525, "learning_rate": 4.1794e-06, "loss": 0.0193, "step": 29104 }, { "epoch": 1.5516873700485152, "grad_norm": 1.9361400604248047, "learning_rate": 4.179200000000001e-06, "loss": 0.0087, "step": 29105 }, { "epoch": 1.5517406834781682, "grad_norm": 1.6527727842330933, "learning_rate": 4.179e-06, "loss": 0.0395, "step": 29106 }, { "epoch": 1.5517939969078212, "grad_norm": 1.9297258853912354, "learning_rate": 4.1788e-06, "loss": 0.0197, "step": 29107 }, { "epoch": 1.551847310337474, "grad_norm": 1.9695019721984863, "learning_rate": 4.1786e-06, "loss": 0.0478, "step": 29108 }, { "epoch": 1.551900623767127, "grad_norm": 3.952803611755371, "learning_rate": 4.1784000000000005e-06, "loss": 0.0499, "step": 29109 }, { "epoch": 1.5519539371967799, "grad_norm": 2.8887927532196045, "learning_rate": 4.178200000000001e-06, "loss": 0.0261, "step": 29110 }, { "epoch": 1.5520072506264326, "grad_norm": 1.1255524158477783, "learning_rate": 4.1780000000000005e-06, "loss": 0.0227, "step": 29111 }, { "epoch": 1.5520605640560858, "grad_norm": 3.138622999191284, "learning_rate": 4.1778e-06, "loss": 0.0182, "step": 29112 }, { "epoch": 1.5521138774857386, "grad_norm": 1.8297518491744995, "learning_rate": 4.1776000000000005e-06, "loss": 0.0244, "step": 29113 }, { "epoch": 1.5521671909153916, "grad_norm": 5.085094928741455, "learning_rate": 4.1774e-06, "loss": 0.0354, "step": 29114 }, { "epoch": 1.5522205043450445, "grad_norm": 0.7799007892608643, "learning_rate": 4.1772000000000004e-06, "loss": 0.0062, "step": 29115 }, { "epoch": 1.5522738177746973, "grad_norm": 1.903788447380066, "learning_rate": 4.177000000000001e-06, "loss": 0.0345, "step": 29116 }, { "epoch": 1.5523271312043505, "grad_norm": 1.5138713121414185, "learning_rate": 4.1768e-06, "loss": 0.0127, "step": 29117 }, { "epoch": 1.5523804446340033, "grad_norm": 3.5911006927490234, "learning_rate": 4.1766e-06, "loss": 0.0202, "step": 29118 }, { "epoch": 1.5524337580636562, "grad_norm": 3.49644136428833, "learning_rate": 4.1764e-06, "loss": 0.0252, "step": 29119 }, { "epoch": 1.5524870714933092, "grad_norm": 3.878169536590576, "learning_rate": 4.1762e-06, "loss": 0.0347, "step": 29120 }, { "epoch": 1.552540384922962, "grad_norm": 3.7188870906829834, "learning_rate": 4.176e-06, "loss": 0.034, "step": 29121 }, { "epoch": 1.5525936983526152, "grad_norm": 1.881390929222107, "learning_rate": 4.175800000000001e-06, "loss": 0.0275, "step": 29122 }, { "epoch": 1.552647011782268, "grad_norm": 2.1497647762298584, "learning_rate": 4.1756e-06, "loss": 0.0362, "step": 29123 }, { "epoch": 1.552700325211921, "grad_norm": 0.44983750581741333, "learning_rate": 4.175400000000001e-06, "loss": 0.0051, "step": 29124 }, { "epoch": 1.552753638641574, "grad_norm": 2.8226284980773926, "learning_rate": 4.1752e-06, "loss": 0.0478, "step": 29125 }, { "epoch": 1.5528069520712267, "grad_norm": 2.9986653327941895, "learning_rate": 4.175e-06, "loss": 0.0383, "step": 29126 }, { "epoch": 1.5528602655008796, "grad_norm": 2.2577037811279297, "learning_rate": 4.1748e-06, "loss": 0.0495, "step": 29127 }, { "epoch": 1.5529135789305326, "grad_norm": 1.0647351741790771, "learning_rate": 4.174600000000001e-06, "loss": 0.0128, "step": 29128 }, { "epoch": 1.5529668923601854, "grad_norm": 1.8323100805282593, "learning_rate": 4.1744e-06, "loss": 0.0425, "step": 29129 }, { "epoch": 1.5530202057898386, "grad_norm": 2.7628469467163086, "learning_rate": 4.174200000000001e-06, "loss": 0.023, "step": 29130 }, { "epoch": 1.5530735192194913, "grad_norm": 1.758029580116272, "learning_rate": 4.174e-06, "loss": 0.0256, "step": 29131 }, { "epoch": 1.5531268326491443, "grad_norm": 2.1032867431640625, "learning_rate": 4.1738e-06, "loss": 0.0402, "step": 29132 }, { "epoch": 1.5531801460787973, "grad_norm": 0.8463279604911804, "learning_rate": 4.1736e-06, "loss": 0.0158, "step": 29133 }, { "epoch": 1.55323345950845, "grad_norm": 0.981412947177887, "learning_rate": 4.1734000000000006e-06, "loss": 0.0115, "step": 29134 }, { "epoch": 1.5532867729381032, "grad_norm": 2.226308822631836, "learning_rate": 4.173200000000001e-06, "loss": 0.0212, "step": 29135 }, { "epoch": 1.553340086367756, "grad_norm": 1.607729196548462, "learning_rate": 4.1730000000000005e-06, "loss": 0.0292, "step": 29136 }, { "epoch": 1.553393399797409, "grad_norm": 2.786773681640625, "learning_rate": 4.1728e-06, "loss": 0.0196, "step": 29137 }, { "epoch": 1.553446713227062, "grad_norm": 1.6010594367980957, "learning_rate": 4.1726000000000005e-06, "loss": 0.0141, "step": 29138 }, { "epoch": 1.5535000266567147, "grad_norm": 2.4196290969848633, "learning_rate": 4.1724e-06, "loss": 0.0252, "step": 29139 }, { "epoch": 1.553553340086368, "grad_norm": 2.290250301361084, "learning_rate": 4.1722000000000005e-06, "loss": 0.0239, "step": 29140 }, { "epoch": 1.5536066535160207, "grad_norm": 0.9036821722984314, "learning_rate": 4.172000000000001e-06, "loss": 0.0178, "step": 29141 }, { "epoch": 1.5536599669456737, "grad_norm": 1.5280859470367432, "learning_rate": 4.1718000000000004e-06, "loss": 0.0102, "step": 29142 }, { "epoch": 1.5537132803753266, "grad_norm": 1.946010708808899, "learning_rate": 4.1716e-06, "loss": 0.017, "step": 29143 }, { "epoch": 1.5537665938049794, "grad_norm": 1.097114086151123, "learning_rate": 4.1714e-06, "loss": 0.011, "step": 29144 }, { "epoch": 1.5538199072346324, "grad_norm": 0.9667210578918457, "learning_rate": 4.1712e-06, "loss": 0.0215, "step": 29145 }, { "epoch": 1.5538732206642853, "grad_norm": 0.9951795935630798, "learning_rate": 4.171e-06, "loss": 0.023, "step": 29146 }, { "epoch": 1.553926534093938, "grad_norm": 1.3165053129196167, "learning_rate": 4.170800000000001e-06, "loss": 0.0193, "step": 29147 }, { "epoch": 1.5539798475235913, "grad_norm": 1.8423879146575928, "learning_rate": 4.1706e-06, "loss": 0.0279, "step": 29148 }, { "epoch": 1.554033160953244, "grad_norm": 1.8694645166397095, "learning_rate": 4.170400000000001e-06, "loss": 0.0269, "step": 29149 }, { "epoch": 1.554086474382897, "grad_norm": 1.9666613340377808, "learning_rate": 4.1702e-06, "loss": 0.0199, "step": 29150 }, { "epoch": 1.55413978781255, "grad_norm": 2.355477809906006, "learning_rate": 4.17e-06, "loss": 0.0213, "step": 29151 }, { "epoch": 1.5541931012422028, "grad_norm": 1.3877623081207275, "learning_rate": 4.1698e-06, "loss": 0.0168, "step": 29152 }, { "epoch": 1.554246414671856, "grad_norm": 1.4650601148605347, "learning_rate": 4.169600000000001e-06, "loss": 0.0238, "step": 29153 }, { "epoch": 1.5542997281015087, "grad_norm": 0.7733505964279175, "learning_rate": 4.1694e-06, "loss": 0.0107, "step": 29154 }, { "epoch": 1.5543530415311617, "grad_norm": 1.519897222518921, "learning_rate": 4.169200000000001e-06, "loss": 0.0105, "step": 29155 }, { "epoch": 1.5544063549608147, "grad_norm": 3.2468130588531494, "learning_rate": 4.169e-06, "loss": 0.0528, "step": 29156 }, { "epoch": 1.5544596683904675, "grad_norm": 2.20052433013916, "learning_rate": 4.1688e-06, "loss": 0.0211, "step": 29157 }, { "epoch": 1.5545129818201207, "grad_norm": 2.0272483825683594, "learning_rate": 4.1686e-06, "loss": 0.0285, "step": 29158 }, { "epoch": 1.5545662952497734, "grad_norm": 1.0247676372528076, "learning_rate": 4.168400000000001e-06, "loss": 0.0074, "step": 29159 }, { "epoch": 1.5546196086794264, "grad_norm": 0.5785267353057861, "learning_rate": 4.1682e-06, "loss": 0.0041, "step": 29160 }, { "epoch": 1.5546729221090794, "grad_norm": 2.7185332775115967, "learning_rate": 4.168000000000001e-06, "loss": 0.0189, "step": 29161 }, { "epoch": 1.5547262355387321, "grad_norm": 1.6591461896896362, "learning_rate": 4.1678e-06, "loss": 0.0166, "step": 29162 }, { "epoch": 1.554779548968385, "grad_norm": 0.47998711466789246, "learning_rate": 4.1676000000000006e-06, "loss": 0.0048, "step": 29163 }, { "epoch": 1.554832862398038, "grad_norm": 1.436682105064392, "learning_rate": 4.1674e-06, "loss": 0.03, "step": 29164 }, { "epoch": 1.5548861758276908, "grad_norm": 1.9649944305419922, "learning_rate": 4.1672000000000005e-06, "loss": 0.0251, "step": 29165 }, { "epoch": 1.554939489257344, "grad_norm": 2.34877872467041, "learning_rate": 4.167e-06, "loss": 0.0145, "step": 29166 }, { "epoch": 1.5549928026869968, "grad_norm": 3.784221887588501, "learning_rate": 4.1668000000000005e-06, "loss": 0.0179, "step": 29167 }, { "epoch": 1.5550461161166498, "grad_norm": 2.5835938453674316, "learning_rate": 4.1666e-06, "loss": 0.0401, "step": 29168 }, { "epoch": 1.5550994295463028, "grad_norm": 1.0158268213272095, "learning_rate": 4.1664000000000005e-06, "loss": 0.0081, "step": 29169 }, { "epoch": 1.5551527429759555, "grad_norm": 2.3757050037384033, "learning_rate": 4.1662e-06, "loss": 0.0325, "step": 29170 }, { "epoch": 1.5552060564056087, "grad_norm": 1.8382112979888916, "learning_rate": 4.1660000000000004e-06, "loss": 0.0149, "step": 29171 }, { "epoch": 1.5552593698352615, "grad_norm": 1.3028185367584229, "learning_rate": 4.1658e-06, "loss": 0.0103, "step": 29172 }, { "epoch": 1.5553126832649145, "grad_norm": 2.4310457706451416, "learning_rate": 4.1656e-06, "loss": 0.0438, "step": 29173 }, { "epoch": 1.5553659966945674, "grad_norm": 0.700596272945404, "learning_rate": 4.165400000000001e-06, "loss": 0.0041, "step": 29174 }, { "epoch": 1.5554193101242202, "grad_norm": 2.0573642253875732, "learning_rate": 4.1652e-06, "loss": 0.0137, "step": 29175 }, { "epoch": 1.5554726235538732, "grad_norm": 3.3725125789642334, "learning_rate": 4.165e-06, "loss": 0.0455, "step": 29176 }, { "epoch": 1.5555259369835261, "grad_norm": 1.8567192554473877, "learning_rate": 4.1648e-06, "loss": 0.0402, "step": 29177 }, { "epoch": 1.5555792504131791, "grad_norm": 0.18671119213104248, "learning_rate": 4.1646e-06, "loss": 0.0022, "step": 29178 }, { "epoch": 1.555632563842832, "grad_norm": 2.4737792015075684, "learning_rate": 4.1644e-06, "loss": 0.022, "step": 29179 }, { "epoch": 1.5556858772724849, "grad_norm": 0.20084518194198608, "learning_rate": 4.164200000000001e-06, "loss": 0.002, "step": 29180 }, { "epoch": 1.5557391907021378, "grad_norm": 1.4032176733016968, "learning_rate": 4.164e-06, "loss": 0.0191, "step": 29181 }, { "epoch": 1.5557925041317908, "grad_norm": 1.6432183980941772, "learning_rate": 4.1638e-06, "loss": 0.0197, "step": 29182 }, { "epoch": 1.5558458175614436, "grad_norm": 1.8595294952392578, "learning_rate": 4.1636e-06, "loss": 0.0276, "step": 29183 }, { "epoch": 1.5558991309910968, "grad_norm": 1.7300018072128296, "learning_rate": 4.1634e-06, "loss": 0.0176, "step": 29184 }, { "epoch": 1.5559524444207495, "grad_norm": 1.1142185926437378, "learning_rate": 4.1632e-06, "loss": 0.0127, "step": 29185 }, { "epoch": 1.5560057578504025, "grad_norm": 2.8899824619293213, "learning_rate": 4.163000000000001e-06, "loss": 0.0091, "step": 29186 }, { "epoch": 1.5560590712800555, "grad_norm": 1.2348958253860474, "learning_rate": 4.1628e-06, "loss": 0.0239, "step": 29187 }, { "epoch": 1.5561123847097083, "grad_norm": 1.1186954975128174, "learning_rate": 4.162600000000001e-06, "loss": 0.0177, "step": 29188 }, { "epoch": 1.5561656981393615, "grad_norm": 2.972654104232788, "learning_rate": 4.1624e-06, "loss": 0.0434, "step": 29189 }, { "epoch": 1.5562190115690142, "grad_norm": 2.266145944595337, "learning_rate": 4.1622e-06, "loss": 0.0122, "step": 29190 }, { "epoch": 1.5562723249986672, "grad_norm": 1.3169161081314087, "learning_rate": 4.162e-06, "loss": 0.0132, "step": 29191 }, { "epoch": 1.5563256384283202, "grad_norm": 1.5961329936981201, "learning_rate": 4.1618000000000005e-06, "loss": 0.0177, "step": 29192 }, { "epoch": 1.556378951857973, "grad_norm": 0.9583738446235657, "learning_rate": 4.1616e-06, "loss": 0.0057, "step": 29193 }, { "epoch": 1.556432265287626, "grad_norm": 1.3172154426574707, "learning_rate": 4.1614000000000005e-06, "loss": 0.0231, "step": 29194 }, { "epoch": 1.5564855787172789, "grad_norm": 1.6446897983551025, "learning_rate": 4.1612e-06, "loss": 0.015, "step": 29195 }, { "epoch": 1.5565388921469319, "grad_norm": 2.731459379196167, "learning_rate": 4.1610000000000005e-06, "loss": 0.0228, "step": 29196 }, { "epoch": 1.5565922055765848, "grad_norm": 0.5362735986709595, "learning_rate": 4.1608e-06, "loss": 0.0041, "step": 29197 }, { "epoch": 1.5566455190062376, "grad_norm": 0.9624195694923401, "learning_rate": 4.1606000000000004e-06, "loss": 0.0067, "step": 29198 }, { "epoch": 1.5566988324358906, "grad_norm": 2.0337698459625244, "learning_rate": 4.160400000000001e-06, "loss": 0.0424, "step": 29199 }, { "epoch": 1.5567521458655436, "grad_norm": 1.6756240129470825, "learning_rate": 4.1602e-06, "loss": 0.021, "step": 29200 }, { "epoch": 1.5568054592951963, "grad_norm": 1.0851328372955322, "learning_rate": 4.16e-06, "loss": 0.0061, "step": 29201 }, { "epoch": 1.5568587727248495, "grad_norm": 1.4048495292663574, "learning_rate": 4.1598e-06, "loss": 0.0091, "step": 29202 }, { "epoch": 1.5569120861545023, "grad_norm": 1.6528583765029907, "learning_rate": 4.1596e-06, "loss": 0.0136, "step": 29203 }, { "epoch": 1.5569653995841553, "grad_norm": 3.167752504348755, "learning_rate": 4.1594e-06, "loss": 0.0311, "step": 29204 }, { "epoch": 1.5570187130138082, "grad_norm": 2.309676170349121, "learning_rate": 4.159200000000001e-06, "loss": 0.0184, "step": 29205 }, { "epoch": 1.557072026443461, "grad_norm": 1.742425799369812, "learning_rate": 4.159e-06, "loss": 0.0378, "step": 29206 }, { "epoch": 1.5571253398731142, "grad_norm": 1.0535731315612793, "learning_rate": 4.158800000000001e-06, "loss": 0.0104, "step": 29207 }, { "epoch": 1.557178653302767, "grad_norm": 1.7314951419830322, "learning_rate": 4.1586e-06, "loss": 0.0423, "step": 29208 }, { "epoch": 1.55723196673242, "grad_norm": 1.685020923614502, "learning_rate": 4.1584e-06, "loss": 0.033, "step": 29209 }, { "epoch": 1.557285280162073, "grad_norm": 1.1436454057693481, "learning_rate": 4.1582e-06, "loss": 0.0075, "step": 29210 }, { "epoch": 1.5573385935917257, "grad_norm": 1.3005973100662231, "learning_rate": 4.158000000000001e-06, "loss": 0.0059, "step": 29211 }, { "epoch": 1.5573919070213786, "grad_norm": 1.9031529426574707, "learning_rate": 4.1578e-06, "loss": 0.0412, "step": 29212 }, { "epoch": 1.5574452204510316, "grad_norm": 1.2953401803970337, "learning_rate": 4.157600000000001e-06, "loss": 0.0069, "step": 29213 }, { "epoch": 1.5574985338806846, "grad_norm": 0.9254883527755737, "learning_rate": 4.1574e-06, "loss": 0.0107, "step": 29214 }, { "epoch": 1.5575518473103376, "grad_norm": 0.470794677734375, "learning_rate": 4.1572e-06, "loss": 0.0045, "step": 29215 }, { "epoch": 1.5576051607399903, "grad_norm": 1.4651414155960083, "learning_rate": 4.157e-06, "loss": 0.0179, "step": 29216 }, { "epoch": 1.5576584741696433, "grad_norm": 3.304227828979492, "learning_rate": 4.156800000000001e-06, "loss": 0.0475, "step": 29217 }, { "epoch": 1.5577117875992963, "grad_norm": 2.0923104286193848, "learning_rate": 4.1566e-06, "loss": 0.0265, "step": 29218 }, { "epoch": 1.557765101028949, "grad_norm": 0.9298110604286194, "learning_rate": 4.1564000000000006e-06, "loss": 0.0084, "step": 29219 }, { "epoch": 1.5578184144586023, "grad_norm": 1.5814132690429688, "learning_rate": 4.1562e-06, "loss": 0.0446, "step": 29220 }, { "epoch": 1.557871727888255, "grad_norm": 1.802972435951233, "learning_rate": 4.1560000000000005e-06, "loss": 0.0312, "step": 29221 }, { "epoch": 1.557925041317908, "grad_norm": 0.7244411110877991, "learning_rate": 4.1558e-06, "loss": 0.017, "step": 29222 }, { "epoch": 1.557978354747561, "grad_norm": 1.7746599912643433, "learning_rate": 4.1556000000000005e-06, "loss": 0.0106, "step": 29223 }, { "epoch": 1.5580316681772137, "grad_norm": 1.4108426570892334, "learning_rate": 4.155400000000001e-06, "loss": 0.0208, "step": 29224 }, { "epoch": 1.558084981606867, "grad_norm": 4.1901326179504395, "learning_rate": 4.1552000000000005e-06, "loss": 0.0322, "step": 29225 }, { "epoch": 1.5581382950365197, "grad_norm": 0.6650437712669373, "learning_rate": 4.155e-06, "loss": 0.0096, "step": 29226 }, { "epoch": 1.5581916084661727, "grad_norm": 0.5823489427566528, "learning_rate": 4.1548000000000004e-06, "loss": 0.0055, "step": 29227 }, { "epoch": 1.5582449218958256, "grad_norm": 0.17077748477458954, "learning_rate": 4.1546e-06, "loss": 0.0016, "step": 29228 }, { "epoch": 1.5582982353254784, "grad_norm": 3.2153637409210205, "learning_rate": 4.1544e-06, "loss": 0.0686, "step": 29229 }, { "epoch": 1.5583515487551314, "grad_norm": 1.1416364908218384, "learning_rate": 4.154200000000001e-06, "loss": 0.0173, "step": 29230 }, { "epoch": 1.5584048621847844, "grad_norm": 0.6637257933616638, "learning_rate": 4.154e-06, "loss": 0.0048, "step": 29231 }, { "epoch": 1.5584581756144371, "grad_norm": 1.7187310457229614, "learning_rate": 4.153800000000001e-06, "loss": 0.0325, "step": 29232 }, { "epoch": 1.5585114890440903, "grad_norm": 3.1629586219787598, "learning_rate": 4.1536e-06, "loss": 0.0244, "step": 29233 }, { "epoch": 1.558564802473743, "grad_norm": 1.4344451427459717, "learning_rate": 4.1534e-06, "loss": 0.0148, "step": 29234 }, { "epoch": 1.558618115903396, "grad_norm": 2.1123344898223877, "learning_rate": 4.1532e-06, "loss": 0.0284, "step": 29235 }, { "epoch": 1.558671429333049, "grad_norm": 1.5090960264205933, "learning_rate": 4.153000000000001e-06, "loss": 0.0235, "step": 29236 }, { "epoch": 1.5587247427627018, "grad_norm": 1.88081693649292, "learning_rate": 4.1528e-06, "loss": 0.0145, "step": 29237 }, { "epoch": 1.558778056192355, "grad_norm": 1.0262402296066284, "learning_rate": 4.152600000000001e-06, "loss": 0.015, "step": 29238 }, { "epoch": 1.5588313696220077, "grad_norm": 0.826097846031189, "learning_rate": 4.1524e-06, "loss": 0.007, "step": 29239 }, { "epoch": 1.5588846830516607, "grad_norm": 1.022570013999939, "learning_rate": 4.1522e-06, "loss": 0.0104, "step": 29240 }, { "epoch": 1.5589379964813137, "grad_norm": 1.0329293012619019, "learning_rate": 4.152e-06, "loss": 0.0082, "step": 29241 }, { "epoch": 1.5589913099109665, "grad_norm": 1.4929213523864746, "learning_rate": 4.151800000000001e-06, "loss": 0.0372, "step": 29242 }, { "epoch": 1.5590446233406197, "grad_norm": 0.556397557258606, "learning_rate": 4.1516e-06, "loss": 0.0054, "step": 29243 }, { "epoch": 1.5590979367702724, "grad_norm": 1.73679518699646, "learning_rate": 4.151400000000001e-06, "loss": 0.0355, "step": 29244 }, { "epoch": 1.5591512501999254, "grad_norm": 1.6018208265304565, "learning_rate": 4.1512e-06, "loss": 0.0117, "step": 29245 }, { "epoch": 1.5592045636295784, "grad_norm": 1.20982027053833, "learning_rate": 4.1510000000000006e-06, "loss": 0.008, "step": 29246 }, { "epoch": 1.5592578770592311, "grad_norm": 1.979584813117981, "learning_rate": 4.1508e-06, "loss": 0.0339, "step": 29247 }, { "epoch": 1.5593111904888841, "grad_norm": 0.6391663551330566, "learning_rate": 4.1506000000000005e-06, "loss": 0.0059, "step": 29248 }, { "epoch": 1.559364503918537, "grad_norm": 0.9524428248405457, "learning_rate": 4.150400000000001e-06, "loss": 0.0064, "step": 29249 }, { "epoch": 1.5594178173481898, "grad_norm": 2.2423653602600098, "learning_rate": 4.1502000000000005e-06, "loss": 0.0084, "step": 29250 }, { "epoch": 1.559471130777843, "grad_norm": 1.0449233055114746, "learning_rate": 4.15e-06, "loss": 0.0188, "step": 29251 }, { "epoch": 1.5595244442074958, "grad_norm": 2.1146240234375, "learning_rate": 4.1498000000000005e-06, "loss": 0.036, "step": 29252 }, { "epoch": 1.5595777576371488, "grad_norm": 1.0451858043670654, "learning_rate": 4.1496e-06, "loss": 0.0201, "step": 29253 }, { "epoch": 1.5596310710668018, "grad_norm": 2.5762369632720947, "learning_rate": 4.1494000000000005e-06, "loss": 0.0151, "step": 29254 }, { "epoch": 1.5596843844964545, "grad_norm": 2.4979002475738525, "learning_rate": 4.149200000000001e-06, "loss": 0.0223, "step": 29255 }, { "epoch": 1.5597376979261077, "grad_norm": 1.0399713516235352, "learning_rate": 4.1490000000000004e-06, "loss": 0.0163, "step": 29256 }, { "epoch": 1.5597910113557605, "grad_norm": 1.5815317630767822, "learning_rate": 4.148800000000001e-06, "loss": 0.0203, "step": 29257 }, { "epoch": 1.5598443247854135, "grad_norm": 2.414132833480835, "learning_rate": 4.1486e-06, "loss": 0.0392, "step": 29258 }, { "epoch": 1.5598976382150664, "grad_norm": 2.663001537322998, "learning_rate": 4.1484e-06, "loss": 0.0121, "step": 29259 }, { "epoch": 1.5599509516447192, "grad_norm": 1.554227590560913, "learning_rate": 4.1482e-06, "loss": 0.0286, "step": 29260 }, { "epoch": 1.5600042650743724, "grad_norm": 1.0066688060760498, "learning_rate": 4.148000000000001e-06, "loss": 0.0131, "step": 29261 }, { "epoch": 1.5600575785040252, "grad_norm": 2.3340365886688232, "learning_rate": 4.1478e-06, "loss": 0.0342, "step": 29262 }, { "epoch": 1.5601108919336781, "grad_norm": 0.6751165986061096, "learning_rate": 4.147600000000001e-06, "loss": 0.0101, "step": 29263 }, { "epoch": 1.5601642053633311, "grad_norm": 1.1519861221313477, "learning_rate": 4.1474e-06, "loss": 0.012, "step": 29264 }, { "epoch": 1.5602175187929839, "grad_norm": 1.94758141040802, "learning_rate": 4.1472e-06, "loss": 0.0284, "step": 29265 }, { "epoch": 1.5602708322226368, "grad_norm": 1.8410080671310425, "learning_rate": 4.147e-06, "loss": 0.0208, "step": 29266 }, { "epoch": 1.5603241456522898, "grad_norm": 1.4623721837997437, "learning_rate": 4.146800000000001e-06, "loss": 0.035, "step": 29267 }, { "epoch": 1.5603774590819426, "grad_norm": 1.6693476438522339, "learning_rate": 4.1466e-06, "loss": 0.0372, "step": 29268 }, { "epoch": 1.5604307725115958, "grad_norm": 1.6609737873077393, "learning_rate": 4.146400000000001e-06, "loss": 0.0228, "step": 29269 }, { "epoch": 1.5604840859412485, "grad_norm": 5.134648323059082, "learning_rate": 4.1462e-06, "loss": 0.0273, "step": 29270 }, { "epoch": 1.5605373993709015, "grad_norm": 1.6378494501113892, "learning_rate": 4.146000000000001e-06, "loss": 0.0114, "step": 29271 }, { "epoch": 1.5605907128005545, "grad_norm": 1.0216056108474731, "learning_rate": 4.1458e-06, "loss": 0.0086, "step": 29272 }, { "epoch": 1.5606440262302073, "grad_norm": 2.197850227355957, "learning_rate": 4.145600000000001e-06, "loss": 0.0334, "step": 29273 }, { "epoch": 1.5606973396598605, "grad_norm": 3.050017833709717, "learning_rate": 4.1454e-06, "loss": 0.0444, "step": 29274 }, { "epoch": 1.5607506530895132, "grad_norm": 1.4725490808486938, "learning_rate": 4.1452000000000006e-06, "loss": 0.0168, "step": 29275 }, { "epoch": 1.5608039665191662, "grad_norm": 1.776039958000183, "learning_rate": 4.145e-06, "loss": 0.0453, "step": 29276 }, { "epoch": 1.5608572799488192, "grad_norm": 1.6509755849838257, "learning_rate": 4.1448000000000005e-06, "loss": 0.0191, "step": 29277 }, { "epoch": 1.560910593378472, "grad_norm": 1.7033700942993164, "learning_rate": 4.1446e-06, "loss": 0.0351, "step": 29278 }, { "epoch": 1.560963906808125, "grad_norm": 1.752454161643982, "learning_rate": 4.1444000000000005e-06, "loss": 0.0224, "step": 29279 }, { "epoch": 1.561017220237778, "grad_norm": 2.8940391540527344, "learning_rate": 4.1442e-06, "loss": 0.0131, "step": 29280 }, { "epoch": 1.5610705336674309, "grad_norm": 2.2745766639709473, "learning_rate": 4.1440000000000005e-06, "loss": 0.0487, "step": 29281 }, { "epoch": 1.5611238470970838, "grad_norm": 4.108312129974365, "learning_rate": 4.143800000000001e-06, "loss": 0.0343, "step": 29282 }, { "epoch": 1.5611771605267366, "grad_norm": 0.9091366529464722, "learning_rate": 4.1436000000000004e-06, "loss": 0.0061, "step": 29283 }, { "epoch": 1.5612304739563896, "grad_norm": 1.3491384983062744, "learning_rate": 4.1434e-06, "loss": 0.0097, "step": 29284 }, { "epoch": 1.5612837873860426, "grad_norm": 1.3937796354293823, "learning_rate": 4.1432e-06, "loss": 0.0324, "step": 29285 }, { "epoch": 1.5613371008156953, "grad_norm": 1.1556644439697266, "learning_rate": 4.143e-06, "loss": 0.0108, "step": 29286 }, { "epoch": 1.5613904142453485, "grad_norm": 1.2539902925491333, "learning_rate": 4.1428e-06, "loss": 0.0161, "step": 29287 }, { "epoch": 1.5614437276750013, "grad_norm": 1.1383508443832397, "learning_rate": 4.142600000000001e-06, "loss": 0.0181, "step": 29288 }, { "epoch": 1.5614970411046543, "grad_norm": 1.1656861305236816, "learning_rate": 4.1424e-06, "loss": 0.02, "step": 29289 }, { "epoch": 1.5615503545343072, "grad_norm": 1.7067735195159912, "learning_rate": 4.1422e-06, "loss": 0.0522, "step": 29290 }, { "epoch": 1.56160366796396, "grad_norm": 0.9858579039573669, "learning_rate": 4.142e-06, "loss": 0.0065, "step": 29291 }, { "epoch": 1.5616569813936132, "grad_norm": 2.570014715194702, "learning_rate": 4.1418e-06, "loss": 0.0193, "step": 29292 }, { "epoch": 1.561710294823266, "grad_norm": 2.594897747039795, "learning_rate": 4.1416e-06, "loss": 0.0182, "step": 29293 }, { "epoch": 1.561763608252919, "grad_norm": 2.770071268081665, "learning_rate": 4.141400000000001e-06, "loss": 0.0173, "step": 29294 }, { "epoch": 1.561816921682572, "grad_norm": 1.2352941036224365, "learning_rate": 4.1412e-06, "loss": 0.0104, "step": 29295 }, { "epoch": 1.5618702351122247, "grad_norm": 2.173393726348877, "learning_rate": 4.141000000000001e-06, "loss": 0.0152, "step": 29296 }, { "epoch": 1.5619235485418776, "grad_norm": 1.7561357021331787, "learning_rate": 4.1408e-06, "loss": 0.0306, "step": 29297 }, { "epoch": 1.5619768619715306, "grad_norm": 2.8213398456573486, "learning_rate": 4.1406e-06, "loss": 0.0364, "step": 29298 }, { "epoch": 1.5620301754011836, "grad_norm": 3.0412092208862305, "learning_rate": 4.1404e-06, "loss": 0.0137, "step": 29299 }, { "epoch": 1.5620834888308366, "grad_norm": 1.3000504970550537, "learning_rate": 4.140200000000001e-06, "loss": 0.016, "step": 29300 }, { "epoch": 1.5621368022604893, "grad_norm": 1.271365761756897, "learning_rate": 4.14e-06, "loss": 0.0319, "step": 29301 }, { "epoch": 1.5621901156901423, "grad_norm": 2.2227132320404053, "learning_rate": 4.139800000000001e-06, "loss": 0.0361, "step": 29302 }, { "epoch": 1.5622434291197953, "grad_norm": 0.5923055410385132, "learning_rate": 4.1396e-06, "loss": 0.0036, "step": 29303 }, { "epoch": 1.562296742549448, "grad_norm": 1.276578664779663, "learning_rate": 4.1394e-06, "loss": 0.0356, "step": 29304 }, { "epoch": 1.5623500559791013, "grad_norm": 1.689955711364746, "learning_rate": 4.1392e-06, "loss": 0.0183, "step": 29305 }, { "epoch": 1.562403369408754, "grad_norm": 1.842573881149292, "learning_rate": 4.1390000000000005e-06, "loss": 0.0137, "step": 29306 }, { "epoch": 1.562456682838407, "grad_norm": 2.3771960735321045, "learning_rate": 4.138800000000001e-06, "loss": 0.0387, "step": 29307 }, { "epoch": 1.56250999626806, "grad_norm": 1.7810724973678589, "learning_rate": 4.1386000000000005e-06, "loss": 0.0143, "step": 29308 }, { "epoch": 1.5625633096977127, "grad_norm": 0.5892951488494873, "learning_rate": 4.1384e-06, "loss": 0.0052, "step": 29309 }, { "epoch": 1.562616623127366, "grad_norm": 2.3691394329071045, "learning_rate": 4.1382000000000005e-06, "loss": 0.0409, "step": 29310 }, { "epoch": 1.5626699365570187, "grad_norm": 1.3291950225830078, "learning_rate": 4.138e-06, "loss": 0.008, "step": 29311 }, { "epoch": 1.5627232499866717, "grad_norm": 2.5892231464385986, "learning_rate": 4.1378000000000004e-06, "loss": 0.0101, "step": 29312 }, { "epoch": 1.5627765634163246, "grad_norm": 2.991556167602539, "learning_rate": 4.137600000000001e-06, "loss": 0.0262, "step": 29313 }, { "epoch": 1.5628298768459774, "grad_norm": 2.0076377391815186, "learning_rate": 4.1374e-06, "loss": 0.0292, "step": 29314 }, { "epoch": 1.5628831902756304, "grad_norm": 0.584931492805481, "learning_rate": 4.1372e-06, "loss": 0.0048, "step": 29315 }, { "epoch": 1.5629365037052834, "grad_norm": 1.4218759536743164, "learning_rate": 4.137e-06, "loss": 0.0291, "step": 29316 }, { "epoch": 1.5629898171349363, "grad_norm": 1.2676115036010742, "learning_rate": 4.1368e-06, "loss": 0.0179, "step": 29317 }, { "epoch": 1.5630431305645893, "grad_norm": 1.600935935974121, "learning_rate": 4.1366e-06, "loss": 0.0135, "step": 29318 }, { "epoch": 1.563096443994242, "grad_norm": 0.7129480838775635, "learning_rate": 4.136400000000001e-06, "loss": 0.0101, "step": 29319 }, { "epoch": 1.563149757423895, "grad_norm": 1.705376386642456, "learning_rate": 4.1362e-06, "loss": 0.0303, "step": 29320 }, { "epoch": 1.563203070853548, "grad_norm": 1.4702205657958984, "learning_rate": 4.136000000000001e-06, "loss": 0.0108, "step": 29321 }, { "epoch": 1.5632563842832008, "grad_norm": 2.252875804901123, "learning_rate": 4.1358e-06, "loss": 0.0336, "step": 29322 }, { "epoch": 1.563309697712854, "grad_norm": 2.771010398864746, "learning_rate": 4.1356e-06, "loss": 0.0146, "step": 29323 }, { "epoch": 1.5633630111425068, "grad_norm": 0.9563685059547424, "learning_rate": 4.1354e-06, "loss": 0.0205, "step": 29324 }, { "epoch": 1.5634163245721597, "grad_norm": 1.4025884866714478, "learning_rate": 4.135200000000001e-06, "loss": 0.0205, "step": 29325 }, { "epoch": 1.5634696380018127, "grad_norm": 0.6624090671539307, "learning_rate": 4.135e-06, "loss": 0.0073, "step": 29326 }, { "epoch": 1.5635229514314655, "grad_norm": 1.508858561515808, "learning_rate": 4.134800000000001e-06, "loss": 0.0153, "step": 29327 }, { "epoch": 1.5635762648611187, "grad_norm": 1.8877708911895752, "learning_rate": 4.1346e-06, "loss": 0.0248, "step": 29328 }, { "epoch": 1.5636295782907714, "grad_norm": 1.7961570024490356, "learning_rate": 4.1344e-06, "loss": 0.015, "step": 29329 }, { "epoch": 1.5636828917204244, "grad_norm": 3.23103928565979, "learning_rate": 4.1342e-06, "loss": 0.0915, "step": 29330 }, { "epoch": 1.5637362051500774, "grad_norm": 0.6398186087608337, "learning_rate": 4.1340000000000006e-06, "loss": 0.0157, "step": 29331 }, { "epoch": 1.5637895185797301, "grad_norm": 0.940924346446991, "learning_rate": 4.133800000000001e-06, "loss": 0.0154, "step": 29332 }, { "epoch": 1.5638428320093831, "grad_norm": 1.0944010019302368, "learning_rate": 4.1336000000000005e-06, "loss": 0.0102, "step": 29333 }, { "epoch": 1.563896145439036, "grad_norm": 2.320729970932007, "learning_rate": 4.1334e-06, "loss": 0.0278, "step": 29334 }, { "epoch": 1.5639494588686889, "grad_norm": 0.47605663537979126, "learning_rate": 4.1332000000000005e-06, "loss": 0.0038, "step": 29335 }, { "epoch": 1.564002772298342, "grad_norm": 1.0785248279571533, "learning_rate": 4.133e-06, "loss": 0.0235, "step": 29336 }, { "epoch": 1.5640560857279948, "grad_norm": 3.31891131401062, "learning_rate": 4.1328000000000005e-06, "loss": 0.0258, "step": 29337 }, { "epoch": 1.5641093991576478, "grad_norm": 2.351750135421753, "learning_rate": 4.132600000000001e-06, "loss": 0.0185, "step": 29338 }, { "epoch": 1.5641627125873008, "grad_norm": 2.7954306602478027, "learning_rate": 4.1324000000000004e-06, "loss": 0.0449, "step": 29339 }, { "epoch": 1.5642160260169535, "grad_norm": 0.712736189365387, "learning_rate": 4.1322e-06, "loss": 0.0062, "step": 29340 }, { "epoch": 1.5642693394466067, "grad_norm": 1.221765160560608, "learning_rate": 4.132e-06, "loss": 0.0158, "step": 29341 }, { "epoch": 1.5643226528762595, "grad_norm": 1.4409964084625244, "learning_rate": 4.1318e-06, "loss": 0.0186, "step": 29342 }, { "epoch": 1.5643759663059125, "grad_norm": 0.2663099467754364, "learning_rate": 4.1316e-06, "loss": 0.0033, "step": 29343 }, { "epoch": 1.5644292797355654, "grad_norm": 1.4869447946548462, "learning_rate": 4.131400000000001e-06, "loss": 0.0274, "step": 29344 }, { "epoch": 1.5644825931652182, "grad_norm": 1.8343937397003174, "learning_rate": 4.1312e-06, "loss": 0.018, "step": 29345 }, { "epoch": 1.5645359065948714, "grad_norm": 1.301445484161377, "learning_rate": 4.131000000000001e-06, "loss": 0.0124, "step": 29346 }, { "epoch": 1.5645892200245242, "grad_norm": 2.5837483406066895, "learning_rate": 4.1308e-06, "loss": 0.0338, "step": 29347 }, { "epoch": 1.5646425334541771, "grad_norm": 1.5902236700057983, "learning_rate": 4.1306e-06, "loss": 0.0128, "step": 29348 }, { "epoch": 1.5646958468838301, "grad_norm": 1.1444990634918213, "learning_rate": 4.1304e-06, "loss": 0.0189, "step": 29349 }, { "epoch": 1.5647491603134829, "grad_norm": 1.0090389251708984, "learning_rate": 4.130200000000001e-06, "loss": 0.0077, "step": 29350 }, { "epoch": 1.5648024737431359, "grad_norm": 1.2842814922332764, "learning_rate": 4.13e-06, "loss": 0.015, "step": 29351 }, { "epoch": 1.5648557871727888, "grad_norm": 0.7783821225166321, "learning_rate": 4.129800000000001e-06, "loss": 0.0085, "step": 29352 }, { "epoch": 1.5649091006024416, "grad_norm": 1.065280556678772, "learning_rate": 4.1296e-06, "loss": 0.0121, "step": 29353 }, { "epoch": 1.5649624140320948, "grad_norm": 1.3792861700057983, "learning_rate": 4.1294e-06, "loss": 0.0108, "step": 29354 }, { "epoch": 1.5650157274617476, "grad_norm": 2.2601478099823, "learning_rate": 4.1292e-06, "loss": 0.0211, "step": 29355 }, { "epoch": 1.5650690408914005, "grad_norm": 0.44330301880836487, "learning_rate": 4.129000000000001e-06, "loss": 0.0033, "step": 29356 }, { "epoch": 1.5651223543210535, "grad_norm": 2.0520615577697754, "learning_rate": 4.1288e-06, "loss": 0.0365, "step": 29357 }, { "epoch": 1.5651756677507063, "grad_norm": 1.717248797416687, "learning_rate": 4.128600000000001e-06, "loss": 0.0252, "step": 29358 }, { "epoch": 1.5652289811803595, "grad_norm": 1.5289907455444336, "learning_rate": 4.1284e-06, "loss": 0.0218, "step": 29359 }, { "epoch": 1.5652822946100122, "grad_norm": 1.8267338275909424, "learning_rate": 4.1282000000000006e-06, "loss": 0.0184, "step": 29360 }, { "epoch": 1.5653356080396652, "grad_norm": 1.4657330513000488, "learning_rate": 4.128e-06, "loss": 0.0287, "step": 29361 }, { "epoch": 1.5653889214693182, "grad_norm": 2.0662362575531006, "learning_rate": 4.1278000000000005e-06, "loss": 0.0454, "step": 29362 }, { "epoch": 1.565442234898971, "grad_norm": 1.2378350496292114, "learning_rate": 4.1276e-06, "loss": 0.0104, "step": 29363 }, { "epoch": 1.5654955483286241, "grad_norm": 2.034524440765381, "learning_rate": 4.1274000000000005e-06, "loss": 0.0234, "step": 29364 }, { "epoch": 1.565548861758277, "grad_norm": 2.021289348602295, "learning_rate": 4.1272e-06, "loss": 0.0216, "step": 29365 }, { "epoch": 1.5656021751879299, "grad_norm": 1.9097896814346313, "learning_rate": 4.1270000000000005e-06, "loss": 0.0133, "step": 29366 }, { "epoch": 1.5656554886175829, "grad_norm": 1.98789644241333, "learning_rate": 4.1268e-06, "loss": 0.0232, "step": 29367 }, { "epoch": 1.5657088020472356, "grad_norm": 1.6339137554168701, "learning_rate": 4.1266000000000004e-06, "loss": 0.0218, "step": 29368 }, { "epoch": 1.5657621154768886, "grad_norm": 1.4183744192123413, "learning_rate": 4.1264e-06, "loss": 0.0158, "step": 29369 }, { "epoch": 1.5658154289065416, "grad_norm": 2.936345100402832, "learning_rate": 4.1262e-06, "loss": 0.0559, "step": 29370 }, { "epoch": 1.5658687423361943, "grad_norm": 0.9861579537391663, "learning_rate": 4.126000000000001e-06, "loss": 0.0258, "step": 29371 }, { "epoch": 1.5659220557658475, "grad_norm": 2.3199942111968994, "learning_rate": 4.1258e-06, "loss": 0.0301, "step": 29372 }, { "epoch": 1.5659753691955003, "grad_norm": 2.1308577060699463, "learning_rate": 4.1256e-06, "loss": 0.0211, "step": 29373 }, { "epoch": 1.5660286826251533, "grad_norm": 1.7094264030456543, "learning_rate": 4.1254e-06, "loss": 0.0234, "step": 29374 }, { "epoch": 1.5660819960548062, "grad_norm": 1.10053288936615, "learning_rate": 4.1252e-06, "loss": 0.0084, "step": 29375 }, { "epoch": 1.566135309484459, "grad_norm": 1.6025071144104004, "learning_rate": 4.125e-06, "loss": 0.023, "step": 29376 }, { "epoch": 1.5661886229141122, "grad_norm": 0.898942768573761, "learning_rate": 4.124800000000001e-06, "loss": 0.0064, "step": 29377 }, { "epoch": 1.566241936343765, "grad_norm": 1.16847825050354, "learning_rate": 4.1246e-06, "loss": 0.0183, "step": 29378 }, { "epoch": 1.566295249773418, "grad_norm": 2.6261703968048096, "learning_rate": 4.1244e-06, "loss": 0.0277, "step": 29379 }, { "epoch": 1.566348563203071, "grad_norm": 2.1058146953582764, "learning_rate": 4.1242e-06, "loss": 0.0311, "step": 29380 }, { "epoch": 1.5664018766327237, "grad_norm": 1.0095791816711426, "learning_rate": 4.124e-06, "loss": 0.007, "step": 29381 }, { "epoch": 1.5664551900623769, "grad_norm": 1.1874446868896484, "learning_rate": 4.1238e-06, "loss": 0.0204, "step": 29382 }, { "epoch": 1.5665085034920296, "grad_norm": 2.210568904876709, "learning_rate": 4.123600000000001e-06, "loss": 0.012, "step": 29383 }, { "epoch": 1.5665618169216826, "grad_norm": 1.5868489742279053, "learning_rate": 4.1234e-06, "loss": 0.0172, "step": 29384 }, { "epoch": 1.5666151303513356, "grad_norm": 0.866720974445343, "learning_rate": 4.123200000000001e-06, "loss": 0.0165, "step": 29385 }, { "epoch": 1.5666684437809884, "grad_norm": 1.7412948608398438, "learning_rate": 4.123e-06, "loss": 0.0307, "step": 29386 }, { "epoch": 1.5667217572106413, "grad_norm": 1.0934208631515503, "learning_rate": 4.1228e-06, "loss": 0.0109, "step": 29387 }, { "epoch": 1.5667750706402943, "grad_norm": 2.052579879760742, "learning_rate": 4.1226e-06, "loss": 0.0333, "step": 29388 }, { "epoch": 1.566828384069947, "grad_norm": 1.9154009819030762, "learning_rate": 4.1224000000000005e-06, "loss": 0.0209, "step": 29389 }, { "epoch": 1.5668816974996003, "grad_norm": 2.4064924716949463, "learning_rate": 4.1222e-06, "loss": 0.0196, "step": 29390 }, { "epoch": 1.566935010929253, "grad_norm": 1.5659188032150269, "learning_rate": 4.1220000000000005e-06, "loss": 0.0169, "step": 29391 }, { "epoch": 1.566988324358906, "grad_norm": 1.648050308227539, "learning_rate": 4.1218e-06, "loss": 0.038, "step": 29392 }, { "epoch": 1.567041637788559, "grad_norm": 2.040719509124756, "learning_rate": 4.1216000000000005e-06, "loss": 0.0402, "step": 29393 }, { "epoch": 1.5670949512182117, "grad_norm": 1.6281901597976685, "learning_rate": 4.1214e-06, "loss": 0.0097, "step": 29394 }, { "epoch": 1.567148264647865, "grad_norm": 2.0886502265930176, "learning_rate": 4.1212000000000005e-06, "loss": 0.0457, "step": 29395 }, { "epoch": 1.5672015780775177, "grad_norm": 2.510948657989502, "learning_rate": 4.121000000000001e-06, "loss": 0.0179, "step": 29396 }, { "epoch": 1.5672548915071707, "grad_norm": 0.6214072704315186, "learning_rate": 4.1208000000000004e-06, "loss": 0.0046, "step": 29397 }, { "epoch": 1.5673082049368237, "grad_norm": 1.628459095954895, "learning_rate": 4.1206e-06, "loss": 0.0335, "step": 29398 }, { "epoch": 1.5673615183664764, "grad_norm": 3.7247836589813232, "learning_rate": 4.1204e-06, "loss": 0.0365, "step": 29399 }, { "epoch": 1.5674148317961294, "grad_norm": 1.7297509908676147, "learning_rate": 4.1202e-06, "loss": 0.0138, "step": 29400 }, { "epoch": 1.5674681452257824, "grad_norm": 3.1098504066467285, "learning_rate": 4.12e-06, "loss": 0.0177, "step": 29401 }, { "epoch": 1.5675214586554354, "grad_norm": 1.1298015117645264, "learning_rate": 4.119800000000001e-06, "loss": 0.0413, "step": 29402 }, { "epoch": 1.5675747720850883, "grad_norm": 1.7775713205337524, "learning_rate": 4.1196e-06, "loss": 0.024, "step": 29403 }, { "epoch": 1.567628085514741, "grad_norm": 1.7301104068756104, "learning_rate": 4.1194e-06, "loss": 0.014, "step": 29404 }, { "epoch": 1.567681398944394, "grad_norm": 0.9362897872924805, "learning_rate": 4.1192e-06, "loss": 0.0117, "step": 29405 }, { "epoch": 1.567734712374047, "grad_norm": 4.981347560882568, "learning_rate": 4.119e-06, "loss": 0.016, "step": 29406 }, { "epoch": 1.5677880258036998, "grad_norm": 1.0618104934692383, "learning_rate": 4.1188e-06, "loss": 0.0159, "step": 29407 }, { "epoch": 1.567841339233353, "grad_norm": 1.1150720119476318, "learning_rate": 4.118600000000001e-06, "loss": 0.0239, "step": 29408 }, { "epoch": 1.5678946526630058, "grad_norm": 1.5013048648834229, "learning_rate": 4.1184e-06, "loss": 0.0246, "step": 29409 }, { "epoch": 1.5679479660926587, "grad_norm": 3.5942132472991943, "learning_rate": 4.118200000000001e-06, "loss": 0.0592, "step": 29410 }, { "epoch": 1.5680012795223117, "grad_norm": 1.8923972845077515, "learning_rate": 4.118e-06, "loss": 0.0134, "step": 29411 }, { "epoch": 1.5680545929519645, "grad_norm": 3.061060667037964, "learning_rate": 4.1178e-06, "loss": 0.0203, "step": 29412 }, { "epoch": 1.5681079063816177, "grad_norm": 2.9640960693359375, "learning_rate": 4.1176e-06, "loss": 0.0354, "step": 29413 }, { "epoch": 1.5681612198112704, "grad_norm": 3.619372606277466, "learning_rate": 4.117400000000001e-06, "loss": 0.0299, "step": 29414 }, { "epoch": 1.5682145332409234, "grad_norm": 1.4549760818481445, "learning_rate": 4.1172e-06, "loss": 0.0123, "step": 29415 }, { "epoch": 1.5682678466705764, "grad_norm": 0.6462332606315613, "learning_rate": 4.1170000000000006e-06, "loss": 0.0064, "step": 29416 }, { "epoch": 1.5683211601002292, "grad_norm": 1.1994588375091553, "learning_rate": 4.1168e-06, "loss": 0.0318, "step": 29417 }, { "epoch": 1.5683744735298821, "grad_norm": 3.0186760425567627, "learning_rate": 4.1166000000000005e-06, "loss": 0.0426, "step": 29418 }, { "epoch": 1.568427786959535, "grad_norm": 1.4315881729125977, "learning_rate": 4.1164e-06, "loss": 0.0168, "step": 29419 }, { "epoch": 1.568481100389188, "grad_norm": 1.0676032304763794, "learning_rate": 4.1162000000000005e-06, "loss": 0.0106, "step": 29420 }, { "epoch": 1.568534413818841, "grad_norm": 2.3750193119049072, "learning_rate": 4.116000000000001e-06, "loss": 0.0198, "step": 29421 }, { "epoch": 1.5685877272484938, "grad_norm": 4.896330833435059, "learning_rate": 4.1158000000000005e-06, "loss": 0.047, "step": 29422 }, { "epoch": 1.5686410406781468, "grad_norm": 1.9412622451782227, "learning_rate": 4.1156e-06, "loss": 0.0239, "step": 29423 }, { "epoch": 1.5686943541077998, "grad_norm": 1.594053030014038, "learning_rate": 4.1154000000000004e-06, "loss": 0.0119, "step": 29424 }, { "epoch": 1.5687476675374525, "grad_norm": 2.9537878036499023, "learning_rate": 4.1152e-06, "loss": 0.0206, "step": 29425 }, { "epoch": 1.5688009809671057, "grad_norm": 1.3531278371810913, "learning_rate": 4.115e-06, "loss": 0.0206, "step": 29426 }, { "epoch": 1.5688542943967585, "grad_norm": 1.9299180507659912, "learning_rate": 4.114800000000001e-06, "loss": 0.0283, "step": 29427 }, { "epoch": 1.5689076078264115, "grad_norm": 0.7685971260070801, "learning_rate": 4.1146e-06, "loss": 0.0065, "step": 29428 }, { "epoch": 1.5689609212560645, "grad_norm": 1.2250609397888184, "learning_rate": 4.1144e-06, "loss": 0.0179, "step": 29429 }, { "epoch": 1.5690142346857172, "grad_norm": 0.27437421679496765, "learning_rate": 4.1142e-06, "loss": 0.0033, "step": 29430 }, { "epoch": 1.5690675481153704, "grad_norm": 2.0323679447174072, "learning_rate": 4.114e-06, "loss": 0.0402, "step": 29431 }, { "epoch": 1.5691208615450232, "grad_norm": 2.9988231658935547, "learning_rate": 4.1138e-06, "loss": 0.0349, "step": 29432 }, { "epoch": 1.5691741749746762, "grad_norm": 0.7718617916107178, "learning_rate": 4.113600000000001e-06, "loss": 0.0086, "step": 29433 }, { "epoch": 1.5692274884043291, "grad_norm": 0.9729618430137634, "learning_rate": 4.1134e-06, "loss": 0.0158, "step": 29434 }, { "epoch": 1.5692808018339819, "grad_norm": 3.114074945449829, "learning_rate": 4.113200000000001e-06, "loss": 0.0641, "step": 29435 }, { "epoch": 1.5693341152636349, "grad_norm": 1.222257137298584, "learning_rate": 4.113e-06, "loss": 0.0095, "step": 29436 }, { "epoch": 1.5693874286932878, "grad_norm": 1.776613712310791, "learning_rate": 4.1128e-06, "loss": 0.0304, "step": 29437 }, { "epoch": 1.5694407421229408, "grad_norm": 1.8381844758987427, "learning_rate": 4.1126e-06, "loss": 0.0248, "step": 29438 }, { "epoch": 1.5694940555525938, "grad_norm": 1.5037440061569214, "learning_rate": 4.112400000000001e-06, "loss": 0.0171, "step": 29439 }, { "epoch": 1.5695473689822466, "grad_norm": 1.3335562944412231, "learning_rate": 4.1122e-06, "loss": 0.0117, "step": 29440 }, { "epoch": 1.5696006824118995, "grad_norm": 3.6030497550964355, "learning_rate": 4.112000000000001e-06, "loss": 0.0255, "step": 29441 }, { "epoch": 1.5696539958415525, "grad_norm": 2.313462018966675, "learning_rate": 4.1118e-06, "loss": 0.0438, "step": 29442 }, { "epoch": 1.5697073092712053, "grad_norm": 1.3096098899841309, "learning_rate": 4.111600000000001e-06, "loss": 0.0179, "step": 29443 }, { "epoch": 1.5697606227008585, "grad_norm": 0.8348095417022705, "learning_rate": 4.1114e-06, "loss": 0.0141, "step": 29444 }, { "epoch": 1.5698139361305112, "grad_norm": 2.0755085945129395, "learning_rate": 4.1112000000000006e-06, "loss": 0.03, "step": 29445 }, { "epoch": 1.5698672495601642, "grad_norm": 0.9009491205215454, "learning_rate": 4.111000000000001e-06, "loss": 0.0142, "step": 29446 }, { "epoch": 1.5699205629898172, "grad_norm": 1.1815553903579712, "learning_rate": 4.1108000000000005e-06, "loss": 0.0167, "step": 29447 }, { "epoch": 1.56997387641947, "grad_norm": 2.0888307094573975, "learning_rate": 4.1106e-06, "loss": 0.0261, "step": 29448 }, { "epoch": 1.5700271898491232, "grad_norm": 2.897021532058716, "learning_rate": 4.1104000000000005e-06, "loss": 0.0166, "step": 29449 }, { "epoch": 1.570080503278776, "grad_norm": 1.1643203496932983, "learning_rate": 4.1102e-06, "loss": 0.0227, "step": 29450 }, { "epoch": 1.5701338167084289, "grad_norm": 1.455723762512207, "learning_rate": 4.1100000000000005e-06, "loss": 0.0247, "step": 29451 }, { "epoch": 1.5701871301380819, "grad_norm": 1.5204486846923828, "learning_rate": 4.109800000000001e-06, "loss": 0.0178, "step": 29452 }, { "epoch": 1.5702404435677346, "grad_norm": 1.7466858625411987, "learning_rate": 4.1096000000000004e-06, "loss": 0.0236, "step": 29453 }, { "epoch": 1.5702937569973876, "grad_norm": 2.343353509902954, "learning_rate": 4.109400000000001e-06, "loss": 0.0359, "step": 29454 }, { "epoch": 1.5703470704270406, "grad_norm": 1.4870613813400269, "learning_rate": 4.1092e-06, "loss": 0.0214, "step": 29455 }, { "epoch": 1.5704003838566933, "grad_norm": 0.5931997299194336, "learning_rate": 4.109e-06, "loss": 0.0068, "step": 29456 }, { "epoch": 1.5704536972863465, "grad_norm": 1.320989966392517, "learning_rate": 4.1088e-06, "loss": 0.0217, "step": 29457 }, { "epoch": 1.5705070107159993, "grad_norm": 1.4831650257110596, "learning_rate": 4.108600000000001e-06, "loss": 0.0295, "step": 29458 }, { "epoch": 1.5705603241456523, "grad_norm": 0.5780879855155945, "learning_rate": 4.1084e-06, "loss": 0.0063, "step": 29459 }, { "epoch": 1.5706136375753053, "grad_norm": 0.9712973833084106, "learning_rate": 4.108200000000001e-06, "loss": 0.0094, "step": 29460 }, { "epoch": 1.570666951004958, "grad_norm": 1.2903399467468262, "learning_rate": 4.108e-06, "loss": 0.0097, "step": 29461 }, { "epoch": 1.5707202644346112, "grad_norm": 2.184174060821533, "learning_rate": 4.1078e-06, "loss": 0.03, "step": 29462 }, { "epoch": 1.570773577864264, "grad_norm": 0.9883662462234497, "learning_rate": 4.1076e-06, "loss": 0.0111, "step": 29463 }, { "epoch": 1.570826891293917, "grad_norm": 1.7309892177581787, "learning_rate": 4.107400000000001e-06, "loss": 0.019, "step": 29464 }, { "epoch": 1.57088020472357, "grad_norm": 1.3687679767608643, "learning_rate": 4.1072e-06, "loss": 0.0277, "step": 29465 }, { "epoch": 1.5709335181532227, "grad_norm": 1.3905491828918457, "learning_rate": 4.107000000000001e-06, "loss": 0.0104, "step": 29466 }, { "epoch": 1.5709868315828759, "grad_norm": 1.1656454801559448, "learning_rate": 4.1068e-06, "loss": 0.0105, "step": 29467 }, { "epoch": 1.5710401450125286, "grad_norm": 1.482508897781372, "learning_rate": 4.106600000000001e-06, "loss": 0.0133, "step": 29468 }, { "epoch": 1.5710934584421816, "grad_norm": 1.7500056028366089, "learning_rate": 4.1064e-06, "loss": 0.0266, "step": 29469 }, { "epoch": 1.5711467718718346, "grad_norm": 2.6723203659057617, "learning_rate": 4.106200000000001e-06, "loss": 0.0427, "step": 29470 }, { "epoch": 1.5712000853014874, "grad_norm": 1.7689645290374756, "learning_rate": 4.106e-06, "loss": 0.0294, "step": 29471 }, { "epoch": 1.5712533987311403, "grad_norm": 2.111130475997925, "learning_rate": 4.1058000000000006e-06, "loss": 0.0316, "step": 29472 }, { "epoch": 1.5713067121607933, "grad_norm": 0.7927535772323608, "learning_rate": 4.1056e-06, "loss": 0.0049, "step": 29473 }, { "epoch": 1.571360025590446, "grad_norm": 2.2759053707122803, "learning_rate": 4.1054000000000005e-06, "loss": 0.0355, "step": 29474 }, { "epoch": 1.5714133390200993, "grad_norm": 1.9944812059402466, "learning_rate": 4.1052e-06, "loss": 0.0474, "step": 29475 }, { "epoch": 1.571466652449752, "grad_norm": 1.1100918054580688, "learning_rate": 4.1050000000000005e-06, "loss": 0.0209, "step": 29476 }, { "epoch": 1.571519965879405, "grad_norm": 2.143939971923828, "learning_rate": 4.1048e-06, "loss": 0.018, "step": 29477 }, { "epoch": 1.571573279309058, "grad_norm": 1.2343366146087646, "learning_rate": 4.1046000000000005e-06, "loss": 0.0231, "step": 29478 }, { "epoch": 1.5716265927387107, "grad_norm": 3.1046159267425537, "learning_rate": 4.104400000000001e-06, "loss": 0.025, "step": 29479 }, { "epoch": 1.571679906168364, "grad_norm": 1.8319027423858643, "learning_rate": 4.1042000000000004e-06, "loss": 0.0261, "step": 29480 }, { "epoch": 1.5717332195980167, "grad_norm": 1.4848108291625977, "learning_rate": 4.104e-06, "loss": 0.0121, "step": 29481 }, { "epoch": 1.5717865330276697, "grad_norm": 1.6483919620513916, "learning_rate": 4.1038e-06, "loss": 0.0149, "step": 29482 }, { "epoch": 1.5718398464573227, "grad_norm": 3.1707382202148438, "learning_rate": 4.1036e-06, "loss": 0.0289, "step": 29483 }, { "epoch": 1.5718931598869754, "grad_norm": 1.1777188777923584, "learning_rate": 4.1034e-06, "loss": 0.0108, "step": 29484 }, { "epoch": 1.5719464733166286, "grad_norm": 2.068946123123169, "learning_rate": 4.103200000000001e-06, "loss": 0.0245, "step": 29485 }, { "epoch": 1.5719997867462814, "grad_norm": 2.0466411113739014, "learning_rate": 4.103e-06, "loss": 0.021, "step": 29486 }, { "epoch": 1.5720531001759344, "grad_norm": 0.48386821150779724, "learning_rate": 4.1028e-06, "loss": 0.004, "step": 29487 }, { "epoch": 1.5721064136055873, "grad_norm": 2.7781026363372803, "learning_rate": 4.1026e-06, "loss": 0.0196, "step": 29488 }, { "epoch": 1.57215972703524, "grad_norm": 1.7462197542190552, "learning_rate": 4.1024e-06, "loss": 0.0093, "step": 29489 }, { "epoch": 1.572213040464893, "grad_norm": 2.0192036628723145, "learning_rate": 4.1022e-06, "loss": 0.0235, "step": 29490 }, { "epoch": 1.572266353894546, "grad_norm": 1.8709803819656372, "learning_rate": 4.102000000000001e-06, "loss": 0.0313, "step": 29491 }, { "epoch": 1.5723196673241988, "grad_norm": 1.5053513050079346, "learning_rate": 4.1018e-06, "loss": 0.0158, "step": 29492 }, { "epoch": 1.572372980753852, "grad_norm": 1.6963484287261963, "learning_rate": 4.101600000000001e-06, "loss": 0.0222, "step": 29493 }, { "epoch": 1.5724262941835048, "grad_norm": 0.6863819360733032, "learning_rate": 4.1014e-06, "loss": 0.0058, "step": 29494 }, { "epoch": 1.5724796076131577, "grad_norm": 2.201721668243408, "learning_rate": 4.1012e-06, "loss": 0.0367, "step": 29495 }, { "epoch": 1.5725329210428107, "grad_norm": 2.1961917877197266, "learning_rate": 4.101e-06, "loss": 0.0443, "step": 29496 }, { "epoch": 1.5725862344724635, "grad_norm": 2.447632312774658, "learning_rate": 4.100800000000001e-06, "loss": 0.0083, "step": 29497 }, { "epoch": 1.5726395479021167, "grad_norm": 0.7652131915092468, "learning_rate": 4.1006e-06, "loss": 0.0075, "step": 29498 }, { "epoch": 1.5726928613317694, "grad_norm": 2.667320489883423, "learning_rate": 4.100400000000001e-06, "loss": 0.0205, "step": 29499 }, { "epoch": 1.5727461747614224, "grad_norm": 1.1678768396377563, "learning_rate": 4.1002e-06, "loss": 0.0137, "step": 29500 }, { "epoch": 1.5727994881910754, "grad_norm": 2.0644001960754395, "learning_rate": 4.1e-06, "loss": 0.0233, "step": 29501 }, { "epoch": 1.5728528016207282, "grad_norm": 2.5217015743255615, "learning_rate": 4.0998e-06, "loss": 0.0139, "step": 29502 }, { "epoch": 1.5729061150503811, "grad_norm": 1.7206027507781982, "learning_rate": 4.0996000000000005e-06, "loss": 0.0122, "step": 29503 }, { "epoch": 1.5729594284800341, "grad_norm": 2.317979574203491, "learning_rate": 4.099400000000001e-06, "loss": 0.0234, "step": 29504 }, { "epoch": 1.573012741909687, "grad_norm": 0.8525561094284058, "learning_rate": 4.0992000000000005e-06, "loss": 0.0057, "step": 29505 }, { "epoch": 1.57306605533934, "grad_norm": 0.886228621006012, "learning_rate": 4.099e-06, "loss": 0.0087, "step": 29506 }, { "epoch": 1.5731193687689928, "grad_norm": 2.2276670932769775, "learning_rate": 4.0988000000000005e-06, "loss": 0.0081, "step": 29507 }, { "epoch": 1.5731726821986458, "grad_norm": 1.7161142826080322, "learning_rate": 4.0986e-06, "loss": 0.037, "step": 29508 }, { "epoch": 1.5732259956282988, "grad_norm": 1.6256319284439087, "learning_rate": 4.0984000000000004e-06, "loss": 0.0109, "step": 29509 }, { "epoch": 1.5732793090579515, "grad_norm": 1.8924401998519897, "learning_rate": 4.098200000000001e-06, "loss": 0.0236, "step": 29510 }, { "epoch": 1.5733326224876047, "grad_norm": 1.1184381246566772, "learning_rate": 4.098e-06, "loss": 0.0147, "step": 29511 }, { "epoch": 1.5733859359172575, "grad_norm": 2.812788724899292, "learning_rate": 4.0978e-06, "loss": 0.0275, "step": 29512 }, { "epoch": 1.5734392493469105, "grad_norm": 1.5953489542007446, "learning_rate": 4.0976e-06, "loss": 0.0134, "step": 29513 }, { "epoch": 1.5734925627765635, "grad_norm": 2.5245020389556885, "learning_rate": 4.0974e-06, "loss": 0.0241, "step": 29514 }, { "epoch": 1.5735458762062162, "grad_norm": 1.7356823682785034, "learning_rate": 4.0972e-06, "loss": 0.0139, "step": 29515 }, { "epoch": 1.5735991896358694, "grad_norm": 2.27104115486145, "learning_rate": 4.097000000000001e-06, "loss": 0.0111, "step": 29516 }, { "epoch": 1.5736525030655222, "grad_norm": 0.5679888129234314, "learning_rate": 4.0968e-06, "loss": 0.0055, "step": 29517 }, { "epoch": 1.5737058164951752, "grad_norm": 0.5907500386238098, "learning_rate": 4.096600000000001e-06, "loss": 0.0052, "step": 29518 }, { "epoch": 1.5737591299248281, "grad_norm": 1.0980901718139648, "learning_rate": 4.0964e-06, "loss": 0.0076, "step": 29519 }, { "epoch": 1.573812443354481, "grad_norm": 1.8733841180801392, "learning_rate": 4.0962e-06, "loss": 0.0092, "step": 29520 }, { "epoch": 1.5738657567841339, "grad_norm": 2.050075054168701, "learning_rate": 4.096e-06, "loss": 0.0343, "step": 29521 }, { "epoch": 1.5739190702137869, "grad_norm": 1.7984024286270142, "learning_rate": 4.095800000000001e-06, "loss": 0.0344, "step": 29522 }, { "epoch": 1.5739723836434398, "grad_norm": 2.4711356163024902, "learning_rate": 4.0956e-06, "loss": 0.0472, "step": 29523 }, { "epoch": 1.5740256970730928, "grad_norm": 2.6225671768188477, "learning_rate": 4.095400000000001e-06, "loss": 0.0166, "step": 29524 }, { "epoch": 1.5740790105027456, "grad_norm": 1.8813977241516113, "learning_rate": 4.0952e-06, "loss": 0.0162, "step": 29525 }, { "epoch": 1.5741323239323985, "grad_norm": 3.6305992603302, "learning_rate": 4.095e-06, "loss": 0.0322, "step": 29526 }, { "epoch": 1.5741856373620515, "grad_norm": 2.489912271499634, "learning_rate": 4.0948e-06, "loss": 0.0111, "step": 29527 }, { "epoch": 1.5742389507917043, "grad_norm": 1.4029744863510132, "learning_rate": 4.0946000000000006e-06, "loss": 0.0213, "step": 29528 }, { "epoch": 1.5742922642213575, "grad_norm": 1.4180923700332642, "learning_rate": 4.094400000000001e-06, "loss": 0.0127, "step": 29529 }, { "epoch": 1.5743455776510102, "grad_norm": 3.4782025814056396, "learning_rate": 4.0942000000000005e-06, "loss": 0.0402, "step": 29530 }, { "epoch": 1.5743988910806632, "grad_norm": 2.291903018951416, "learning_rate": 4.094e-06, "loss": 0.0083, "step": 29531 }, { "epoch": 1.5744522045103162, "grad_norm": 3.5652713775634766, "learning_rate": 4.0938000000000005e-06, "loss": 0.0463, "step": 29532 }, { "epoch": 1.574505517939969, "grad_norm": 1.5384483337402344, "learning_rate": 4.0936e-06, "loss": 0.0062, "step": 29533 }, { "epoch": 1.5745588313696222, "grad_norm": 2.5705573558807373, "learning_rate": 4.0934000000000005e-06, "loss": 0.0312, "step": 29534 }, { "epoch": 1.574612144799275, "grad_norm": 0.6478308439254761, "learning_rate": 4.093200000000001e-06, "loss": 0.0058, "step": 29535 }, { "epoch": 1.574665458228928, "grad_norm": 1.5311837196350098, "learning_rate": 4.0930000000000005e-06, "loss": 0.0141, "step": 29536 }, { "epoch": 1.5747187716585809, "grad_norm": 0.3819452226161957, "learning_rate": 4.0928e-06, "loss": 0.0032, "step": 29537 }, { "epoch": 1.5747720850882336, "grad_norm": 4.930693626403809, "learning_rate": 4.0926e-06, "loss": 0.0673, "step": 29538 }, { "epoch": 1.5748253985178866, "grad_norm": 1.226569414138794, "learning_rate": 4.0924e-06, "loss": 0.0113, "step": 29539 }, { "epoch": 1.5748787119475396, "grad_norm": 2.3620827198028564, "learning_rate": 4.0922e-06, "loss": 0.0488, "step": 29540 }, { "epoch": 1.5749320253771926, "grad_norm": 2.4218485355377197, "learning_rate": 4.092000000000001e-06, "loss": 0.0065, "step": 29541 }, { "epoch": 1.5749853388068455, "grad_norm": 2.07682204246521, "learning_rate": 4.0918e-06, "loss": 0.0382, "step": 29542 }, { "epoch": 1.5750386522364983, "grad_norm": 1.4485325813293457, "learning_rate": 4.091600000000001e-06, "loss": 0.0183, "step": 29543 }, { "epoch": 1.5750919656661513, "grad_norm": 1.3355295658111572, "learning_rate": 4.0914e-06, "loss": 0.0093, "step": 29544 }, { "epoch": 1.5751452790958043, "grad_norm": 3.2965145111083984, "learning_rate": 4.0912e-06, "loss": 0.044, "step": 29545 }, { "epoch": 1.575198592525457, "grad_norm": 1.7372229099273682, "learning_rate": 4.091e-06, "loss": 0.0247, "step": 29546 }, { "epoch": 1.5752519059551102, "grad_norm": 2.0611190795898438, "learning_rate": 4.090800000000001e-06, "loss": 0.0151, "step": 29547 }, { "epoch": 1.575305219384763, "grad_norm": 1.639481782913208, "learning_rate": 4.0906e-06, "loss": 0.0255, "step": 29548 }, { "epoch": 1.575358532814416, "grad_norm": 1.6154487133026123, "learning_rate": 4.090400000000001e-06, "loss": 0.0198, "step": 29549 }, { "epoch": 1.575411846244069, "grad_norm": 0.8658483624458313, "learning_rate": 4.0902e-06, "loss": 0.0059, "step": 29550 }, { "epoch": 1.5754651596737217, "grad_norm": 1.9990999698638916, "learning_rate": 4.09e-06, "loss": 0.0118, "step": 29551 }, { "epoch": 1.575518473103375, "grad_norm": 1.2313125133514404, "learning_rate": 4.0898e-06, "loss": 0.0067, "step": 29552 }, { "epoch": 1.5755717865330277, "grad_norm": 1.9224128723144531, "learning_rate": 4.089600000000001e-06, "loss": 0.0342, "step": 29553 }, { "epoch": 1.5756250999626806, "grad_norm": 1.559193730354309, "learning_rate": 4.0894e-06, "loss": 0.0119, "step": 29554 }, { "epoch": 1.5756784133923336, "grad_norm": 1.823352336883545, "learning_rate": 4.089200000000001e-06, "loss": 0.0175, "step": 29555 }, { "epoch": 1.5757317268219864, "grad_norm": 3.197064161300659, "learning_rate": 4.089e-06, "loss": 0.0268, "step": 29556 }, { "epoch": 1.5757850402516393, "grad_norm": 0.6993100047111511, "learning_rate": 4.0888000000000006e-06, "loss": 0.0056, "step": 29557 }, { "epoch": 1.5758383536812923, "grad_norm": 0.9921775460243225, "learning_rate": 4.0886e-06, "loss": 0.0191, "step": 29558 }, { "epoch": 1.575891667110945, "grad_norm": 1.980181336402893, "learning_rate": 4.0884000000000005e-06, "loss": 0.0408, "step": 29559 }, { "epoch": 1.5759449805405983, "grad_norm": 2.9081718921661377, "learning_rate": 4.0882e-06, "loss": 0.0481, "step": 29560 }, { "epoch": 1.575998293970251, "grad_norm": 1.183716058731079, "learning_rate": 4.0880000000000005e-06, "loss": 0.0159, "step": 29561 }, { "epoch": 1.576051607399904, "grad_norm": 0.3194381594657898, "learning_rate": 4.0878e-06, "loss": 0.0037, "step": 29562 }, { "epoch": 1.576104920829557, "grad_norm": 1.8890515565872192, "learning_rate": 4.0876000000000005e-06, "loss": 0.007, "step": 29563 }, { "epoch": 1.5761582342592098, "grad_norm": 1.5104069709777832, "learning_rate": 4.0874e-06, "loss": 0.0115, "step": 29564 }, { "epoch": 1.576211547688863, "grad_norm": 1.142371654510498, "learning_rate": 4.0872000000000004e-06, "loss": 0.0133, "step": 29565 }, { "epoch": 1.5762648611185157, "grad_norm": 2.130204439163208, "learning_rate": 4.087e-06, "loss": 0.0293, "step": 29566 }, { "epoch": 1.5763181745481687, "grad_norm": 0.5998784303665161, "learning_rate": 4.0868e-06, "loss": 0.0058, "step": 29567 }, { "epoch": 1.5763714879778217, "grad_norm": 1.7871586084365845, "learning_rate": 4.086600000000001e-06, "loss": 0.0171, "step": 29568 }, { "epoch": 1.5764248014074744, "grad_norm": 0.8381256461143494, "learning_rate": 4.0864e-06, "loss": 0.0196, "step": 29569 }, { "epoch": 1.5764781148371276, "grad_norm": 1.6537209749221802, "learning_rate": 4.0862e-06, "loss": 0.0162, "step": 29570 }, { "epoch": 1.5765314282667804, "grad_norm": 1.7182140350341797, "learning_rate": 4.086e-06, "loss": 0.0157, "step": 29571 }, { "epoch": 1.5765847416964334, "grad_norm": 1.6778628826141357, "learning_rate": 4.0858e-06, "loss": 0.0137, "step": 29572 }, { "epoch": 1.5766380551260863, "grad_norm": 0.9763011336326599, "learning_rate": 4.0856e-06, "loss": 0.0057, "step": 29573 }, { "epoch": 1.576691368555739, "grad_norm": 3.012174606323242, "learning_rate": 4.085400000000001e-06, "loss": 0.0221, "step": 29574 }, { "epoch": 1.576744681985392, "grad_norm": 1.8922936916351318, "learning_rate": 4.0852e-06, "loss": 0.0104, "step": 29575 }, { "epoch": 1.576797995415045, "grad_norm": 1.490105152130127, "learning_rate": 4.085e-06, "loss": 0.0077, "step": 29576 }, { "epoch": 1.5768513088446978, "grad_norm": 1.6838784217834473, "learning_rate": 4.0848e-06, "loss": 0.0142, "step": 29577 }, { "epoch": 1.576904622274351, "grad_norm": 1.1271966695785522, "learning_rate": 4.0846e-06, "loss": 0.0111, "step": 29578 }, { "epoch": 1.5769579357040038, "grad_norm": 1.9771783351898193, "learning_rate": 4.0844e-06, "loss": 0.04, "step": 29579 }, { "epoch": 1.5770112491336568, "grad_norm": 1.837666630744934, "learning_rate": 4.084200000000001e-06, "loss": 0.0295, "step": 29580 }, { "epoch": 1.5770645625633097, "grad_norm": 1.4494510889053345, "learning_rate": 4.084e-06, "loss": 0.0075, "step": 29581 }, { "epoch": 1.5771178759929625, "grad_norm": 1.47854745388031, "learning_rate": 4.083800000000001e-06, "loss": 0.0109, "step": 29582 }, { "epoch": 1.5771711894226157, "grad_norm": 1.6510071754455566, "learning_rate": 4.0836e-06, "loss": 0.0127, "step": 29583 }, { "epoch": 1.5772245028522685, "grad_norm": 1.547372579574585, "learning_rate": 4.0834e-06, "loss": 0.0246, "step": 29584 }, { "epoch": 1.5772778162819214, "grad_norm": 2.3572144508361816, "learning_rate": 4.0832e-06, "loss": 0.0435, "step": 29585 }, { "epoch": 1.5773311297115744, "grad_norm": 3.0223491191864014, "learning_rate": 4.0830000000000005e-06, "loss": 0.0443, "step": 29586 }, { "epoch": 1.5773844431412272, "grad_norm": 2.5801684856414795, "learning_rate": 4.0828e-06, "loss": 0.0345, "step": 29587 }, { "epoch": 1.5774377565708804, "grad_norm": 2.021857738494873, "learning_rate": 4.0826000000000005e-06, "loss": 0.0124, "step": 29588 }, { "epoch": 1.5774910700005331, "grad_norm": 1.3595194816589355, "learning_rate": 4.0824e-06, "loss": 0.0287, "step": 29589 }, { "epoch": 1.577544383430186, "grad_norm": 1.6260075569152832, "learning_rate": 4.0822000000000005e-06, "loss": 0.0125, "step": 29590 }, { "epoch": 1.577597696859839, "grad_norm": 3.3211514949798584, "learning_rate": 4.082e-06, "loss": 0.0363, "step": 29591 }, { "epoch": 1.5776510102894918, "grad_norm": 1.6110727787017822, "learning_rate": 4.0818000000000005e-06, "loss": 0.0073, "step": 29592 }, { "epoch": 1.5777043237191448, "grad_norm": 0.7912937998771667, "learning_rate": 4.081600000000001e-06, "loss": 0.0048, "step": 29593 }, { "epoch": 1.5777576371487978, "grad_norm": 1.7064484357833862, "learning_rate": 4.0814000000000004e-06, "loss": 0.0469, "step": 29594 }, { "epoch": 1.5778109505784506, "grad_norm": 1.3834669589996338, "learning_rate": 4.0812e-06, "loss": 0.023, "step": 29595 }, { "epoch": 1.5778642640081038, "grad_norm": 2.1122806072235107, "learning_rate": 4.081e-06, "loss": 0.0116, "step": 29596 }, { "epoch": 1.5779175774377565, "grad_norm": 0.425281286239624, "learning_rate": 4.0808e-06, "loss": 0.0042, "step": 29597 }, { "epoch": 1.5779708908674095, "grad_norm": 1.3055065870285034, "learning_rate": 4.0806e-06, "loss": 0.011, "step": 29598 }, { "epoch": 1.5780242042970625, "grad_norm": 2.1527864933013916, "learning_rate": 4.080400000000001e-06, "loss": 0.0166, "step": 29599 }, { "epoch": 1.5780775177267152, "grad_norm": 2.4412343502044678, "learning_rate": 4.0802e-06, "loss": 0.0152, "step": 29600 }, { "epoch": 1.5781308311563684, "grad_norm": 1.1759297847747803, "learning_rate": 4.08e-06, "loss": 0.0099, "step": 29601 }, { "epoch": 1.5781841445860212, "grad_norm": 0.7784696221351624, "learning_rate": 4.0798e-06, "loss": 0.0055, "step": 29602 }, { "epoch": 1.5782374580156742, "grad_norm": 0.8438495993614197, "learning_rate": 4.0796e-06, "loss": 0.0063, "step": 29603 }, { "epoch": 1.5782907714453271, "grad_norm": 1.426735520362854, "learning_rate": 4.0794e-06, "loss": 0.0157, "step": 29604 }, { "epoch": 1.57834408487498, "grad_norm": 2.3641812801361084, "learning_rate": 4.079200000000001e-06, "loss": 0.0363, "step": 29605 }, { "epoch": 1.578397398304633, "grad_norm": 2.379277229309082, "learning_rate": 4.079e-06, "loss": 0.0172, "step": 29606 }, { "epoch": 1.5784507117342859, "grad_norm": 3.511889696121216, "learning_rate": 4.078800000000001e-06, "loss": 0.0222, "step": 29607 }, { "epoch": 1.5785040251639388, "grad_norm": 2.094357967376709, "learning_rate": 4.0786e-06, "loss": 0.0145, "step": 29608 }, { "epoch": 1.5785573385935918, "grad_norm": 1.7716343402862549, "learning_rate": 4.0784e-06, "loss": 0.024, "step": 29609 }, { "epoch": 1.5786106520232446, "grad_norm": 3.7413806915283203, "learning_rate": 4.0782e-06, "loss": 0.0302, "step": 29610 }, { "epoch": 1.5786639654528976, "grad_norm": 1.5238094329833984, "learning_rate": 4.078000000000001e-06, "loss": 0.0315, "step": 29611 }, { "epoch": 1.5787172788825505, "grad_norm": 3.2994141578674316, "learning_rate": 4.0778e-06, "loss": 0.0301, "step": 29612 }, { "epoch": 1.5787705923122033, "grad_norm": 0.4583258330821991, "learning_rate": 4.0776000000000006e-06, "loss": 0.0031, "step": 29613 }, { "epoch": 1.5788239057418565, "grad_norm": 1.7704410552978516, "learning_rate": 4.0774e-06, "loss": 0.0117, "step": 29614 }, { "epoch": 1.5788772191715092, "grad_norm": 1.5259218215942383, "learning_rate": 4.0772000000000005e-06, "loss": 0.0219, "step": 29615 }, { "epoch": 1.5789305326011622, "grad_norm": 1.6701761484146118, "learning_rate": 4.077e-06, "loss": 0.0346, "step": 29616 }, { "epoch": 1.5789838460308152, "grad_norm": 2.489014148712158, "learning_rate": 4.0768000000000005e-06, "loss": 0.0269, "step": 29617 }, { "epoch": 1.579037159460468, "grad_norm": 2.2532646656036377, "learning_rate": 4.076600000000001e-06, "loss": 0.0251, "step": 29618 }, { "epoch": 1.5790904728901212, "grad_norm": 0.7940890789031982, "learning_rate": 4.0764000000000005e-06, "loss": 0.0062, "step": 29619 }, { "epoch": 1.579143786319774, "grad_norm": 4.718399524688721, "learning_rate": 4.0762e-06, "loss": 0.0445, "step": 29620 }, { "epoch": 1.579197099749427, "grad_norm": 2.841182231903076, "learning_rate": 4.0760000000000004e-06, "loss": 0.0345, "step": 29621 }, { "epoch": 1.5792504131790799, "grad_norm": 0.6530570983886719, "learning_rate": 4.0758e-06, "loss": 0.005, "step": 29622 }, { "epoch": 1.5793037266087326, "grad_norm": 3.969705581665039, "learning_rate": 4.0756e-06, "loss": 0.0155, "step": 29623 }, { "epoch": 1.5793570400383856, "grad_norm": 1.5803040266036987, "learning_rate": 4.075400000000001e-06, "loss": 0.0158, "step": 29624 }, { "epoch": 1.5794103534680386, "grad_norm": 2.069082021713257, "learning_rate": 4.0752e-06, "loss": 0.0165, "step": 29625 }, { "epoch": 1.5794636668976916, "grad_norm": 1.685926079750061, "learning_rate": 4.075e-06, "loss": 0.0456, "step": 29626 }, { "epoch": 1.5795169803273446, "grad_norm": 1.1637223958969116, "learning_rate": 4.0748e-06, "loss": 0.0111, "step": 29627 }, { "epoch": 1.5795702937569973, "grad_norm": 2.107855796813965, "learning_rate": 4.0746e-06, "loss": 0.032, "step": 29628 }, { "epoch": 1.5796236071866503, "grad_norm": 2.1939656734466553, "learning_rate": 4.0744e-06, "loss": 0.0512, "step": 29629 }, { "epoch": 1.5796769206163033, "grad_norm": 0.7034400105476379, "learning_rate": 4.074200000000001e-06, "loss": 0.0056, "step": 29630 }, { "epoch": 1.579730234045956, "grad_norm": 1.5528512001037598, "learning_rate": 4.074e-06, "loss": 0.0085, "step": 29631 }, { "epoch": 1.5797835474756092, "grad_norm": 1.4151438474655151, "learning_rate": 4.073800000000001e-06, "loss": 0.0415, "step": 29632 }, { "epoch": 1.579836860905262, "grad_norm": 2.5294971466064453, "learning_rate": 4.0736e-06, "loss": 0.0133, "step": 29633 }, { "epoch": 1.579890174334915, "grad_norm": 2.6073617935180664, "learning_rate": 4.0734e-06, "loss": 0.0421, "step": 29634 }, { "epoch": 1.579943487764568, "grad_norm": 1.9792423248291016, "learning_rate": 4.0732e-06, "loss": 0.0285, "step": 29635 }, { "epoch": 1.5799968011942207, "grad_norm": 2.4191458225250244, "learning_rate": 4.073000000000001e-06, "loss": 0.0208, "step": 29636 }, { "epoch": 1.580050114623874, "grad_norm": 1.3659404516220093, "learning_rate": 4.0728e-06, "loss": 0.0135, "step": 29637 }, { "epoch": 1.5801034280535267, "grad_norm": 1.7098497152328491, "learning_rate": 4.072600000000001e-06, "loss": 0.0328, "step": 29638 }, { "epoch": 1.5801567414831796, "grad_norm": 1.1604459285736084, "learning_rate": 4.0724e-06, "loss": 0.011, "step": 29639 }, { "epoch": 1.5802100549128326, "grad_norm": 1.9493354558944702, "learning_rate": 4.072200000000001e-06, "loss": 0.0174, "step": 29640 }, { "epoch": 1.5802633683424854, "grad_norm": 3.168959379196167, "learning_rate": 4.072e-06, "loss": 0.0559, "step": 29641 }, { "epoch": 1.5803166817721384, "grad_norm": 1.973761796951294, "learning_rate": 4.0718000000000006e-06, "loss": 0.0175, "step": 29642 }, { "epoch": 1.5803699952017913, "grad_norm": 0.49061280488967896, "learning_rate": 4.071600000000001e-06, "loss": 0.0035, "step": 29643 }, { "epoch": 1.5804233086314443, "grad_norm": 1.4355533123016357, "learning_rate": 4.0714000000000005e-06, "loss": 0.017, "step": 29644 }, { "epoch": 1.5804766220610973, "grad_norm": 1.7790770530700684, "learning_rate": 4.0712e-06, "loss": 0.0394, "step": 29645 }, { "epoch": 1.58052993549075, "grad_norm": 4.1327948570251465, "learning_rate": 4.0710000000000005e-06, "loss": 0.0482, "step": 29646 }, { "epoch": 1.580583248920403, "grad_norm": 2.5317869186401367, "learning_rate": 4.0708e-06, "loss": 0.0228, "step": 29647 }, { "epoch": 1.580636562350056, "grad_norm": 1.6084009408950806, "learning_rate": 4.0706000000000005e-06, "loss": 0.0069, "step": 29648 }, { "epoch": 1.5806898757797088, "grad_norm": 2.4776861667633057, "learning_rate": 4.070400000000001e-06, "loss": 0.0231, "step": 29649 }, { "epoch": 1.580743189209362, "grad_norm": 2.585089921951294, "learning_rate": 4.0702000000000004e-06, "loss": 0.0146, "step": 29650 }, { "epoch": 1.5807965026390147, "grad_norm": 1.5162043571472168, "learning_rate": 4.07e-06, "loss": 0.0196, "step": 29651 }, { "epoch": 1.5808498160686677, "grad_norm": 1.7794700860977173, "learning_rate": 4.0698e-06, "loss": 0.0307, "step": 29652 }, { "epoch": 1.5809031294983207, "grad_norm": 3.0215392112731934, "learning_rate": 4.0696e-06, "loss": 0.0328, "step": 29653 }, { "epoch": 1.5809564429279734, "grad_norm": 2.510633945465088, "learning_rate": 4.0694e-06, "loss": 0.0293, "step": 29654 }, { "epoch": 1.5810097563576266, "grad_norm": 1.6919246912002563, "learning_rate": 4.069200000000001e-06, "loss": 0.0165, "step": 29655 }, { "epoch": 1.5810630697872794, "grad_norm": 1.4137554168701172, "learning_rate": 4.069e-06, "loss": 0.0324, "step": 29656 }, { "epoch": 1.5811163832169324, "grad_norm": 1.547829270362854, "learning_rate": 4.068800000000001e-06, "loss": 0.0374, "step": 29657 }, { "epoch": 1.5811696966465854, "grad_norm": 1.3398083448410034, "learning_rate": 4.0686e-06, "loss": 0.0118, "step": 29658 }, { "epoch": 1.5812230100762381, "grad_norm": 2.415067195892334, "learning_rate": 4.0684e-06, "loss": 0.0291, "step": 29659 }, { "epoch": 1.581276323505891, "grad_norm": 1.09109365940094, "learning_rate": 4.0682e-06, "loss": 0.0087, "step": 29660 }, { "epoch": 1.581329636935544, "grad_norm": 0.7754001021385193, "learning_rate": 4.068000000000001e-06, "loss": 0.0121, "step": 29661 }, { "epoch": 1.5813829503651968, "grad_norm": 1.4044256210327148, "learning_rate": 4.0678e-06, "loss": 0.0192, "step": 29662 }, { "epoch": 1.58143626379485, "grad_norm": 2.175776958465576, "learning_rate": 4.067600000000001e-06, "loss": 0.0245, "step": 29663 }, { "epoch": 1.5814895772245028, "grad_norm": 2.103952407836914, "learning_rate": 4.0674e-06, "loss": 0.0112, "step": 29664 }, { "epoch": 1.5815428906541558, "grad_norm": 1.1482007503509521, "learning_rate": 4.067200000000001e-06, "loss": 0.0243, "step": 29665 }, { "epoch": 1.5815962040838087, "grad_norm": 3.4134457111358643, "learning_rate": 4.067e-06, "loss": 0.0605, "step": 29666 }, { "epoch": 1.5816495175134615, "grad_norm": 0.8344548940658569, "learning_rate": 4.066800000000001e-06, "loss": 0.0088, "step": 29667 }, { "epoch": 1.5817028309431147, "grad_norm": 1.7837533950805664, "learning_rate": 4.0666e-06, "loss": 0.0313, "step": 29668 }, { "epoch": 1.5817561443727675, "grad_norm": 1.4418433904647827, "learning_rate": 4.0664000000000006e-06, "loss": 0.0224, "step": 29669 }, { "epoch": 1.5818094578024204, "grad_norm": 1.8966931104660034, "learning_rate": 4.0662e-06, "loss": 0.0234, "step": 29670 }, { "epoch": 1.5818627712320734, "grad_norm": 1.605757474899292, "learning_rate": 4.0660000000000005e-06, "loss": 0.0143, "step": 29671 }, { "epoch": 1.5819160846617262, "grad_norm": 0.7015153765678406, "learning_rate": 4.0658e-06, "loss": 0.0069, "step": 29672 }, { "epoch": 1.5819693980913794, "grad_norm": 1.3545417785644531, "learning_rate": 4.0656000000000005e-06, "loss": 0.0101, "step": 29673 }, { "epoch": 1.5820227115210321, "grad_norm": 0.755809485912323, "learning_rate": 4.0654e-06, "loss": 0.0078, "step": 29674 }, { "epoch": 1.5820760249506851, "grad_norm": 1.6819275617599487, "learning_rate": 4.0652000000000005e-06, "loss": 0.0312, "step": 29675 }, { "epoch": 1.582129338380338, "grad_norm": 2.509209632873535, "learning_rate": 4.065e-06, "loss": 0.0272, "step": 29676 }, { "epoch": 1.5821826518099908, "grad_norm": 1.1986808776855469, "learning_rate": 4.0648000000000005e-06, "loss": 0.016, "step": 29677 }, { "epoch": 1.5822359652396438, "grad_norm": 1.2928352355957031, "learning_rate": 4.0646e-06, "loss": 0.0216, "step": 29678 }, { "epoch": 1.5822892786692968, "grad_norm": 2.0017271041870117, "learning_rate": 4.0644e-06, "loss": 0.0189, "step": 29679 }, { "epoch": 1.5823425920989496, "grad_norm": 1.9124053716659546, "learning_rate": 4.0642e-06, "loss": 0.0254, "step": 29680 }, { "epoch": 1.5823959055286028, "grad_norm": 0.5251853466033936, "learning_rate": 4.064e-06, "loss": 0.0048, "step": 29681 }, { "epoch": 1.5824492189582555, "grad_norm": 1.5803600549697876, "learning_rate": 4.063800000000001e-06, "loss": 0.0265, "step": 29682 }, { "epoch": 1.5825025323879085, "grad_norm": 3.3709027767181396, "learning_rate": 4.0636e-06, "loss": 0.0513, "step": 29683 }, { "epoch": 1.5825558458175615, "grad_norm": 0.8556879758834839, "learning_rate": 4.0634e-06, "loss": 0.0135, "step": 29684 }, { "epoch": 1.5826091592472142, "grad_norm": 1.6849517822265625, "learning_rate": 4.0632e-06, "loss": 0.0212, "step": 29685 }, { "epoch": 1.5826624726768674, "grad_norm": 1.652264952659607, "learning_rate": 4.063e-06, "loss": 0.0155, "step": 29686 }, { "epoch": 1.5827157861065202, "grad_norm": 1.1750595569610596, "learning_rate": 4.0628e-06, "loss": 0.0189, "step": 29687 }, { "epoch": 1.5827690995361732, "grad_norm": 4.911782741546631, "learning_rate": 4.062600000000001e-06, "loss": 0.0282, "step": 29688 }, { "epoch": 1.5828224129658262, "grad_norm": 1.7424203157424927, "learning_rate": 4.0624e-06, "loss": 0.023, "step": 29689 }, { "epoch": 1.582875726395479, "grad_norm": 0.42547813057899475, "learning_rate": 4.062200000000001e-06, "loss": 0.0035, "step": 29690 }, { "epoch": 1.5829290398251321, "grad_norm": 0.8558177351951599, "learning_rate": 4.062e-06, "loss": 0.0076, "step": 29691 }, { "epoch": 1.5829823532547849, "grad_norm": 0.8947435617446899, "learning_rate": 4.0618e-06, "loss": 0.0065, "step": 29692 }, { "epoch": 1.5830356666844378, "grad_norm": 3.487382650375366, "learning_rate": 4.0616e-06, "loss": 0.0371, "step": 29693 }, { "epoch": 1.5830889801140908, "grad_norm": 1.3970013856887817, "learning_rate": 4.061400000000001e-06, "loss": 0.0288, "step": 29694 }, { "epoch": 1.5831422935437436, "grad_norm": 2.7064003944396973, "learning_rate": 4.0612e-06, "loss": 0.0639, "step": 29695 }, { "epoch": 1.5831956069733966, "grad_norm": 2.0430591106414795, "learning_rate": 4.061000000000001e-06, "loss": 0.019, "step": 29696 }, { "epoch": 1.5832489204030495, "grad_norm": 1.765061855316162, "learning_rate": 4.0608e-06, "loss": 0.0126, "step": 29697 }, { "epoch": 1.5833022338327023, "grad_norm": 1.1539244651794434, "learning_rate": 4.0606e-06, "loss": 0.017, "step": 29698 }, { "epoch": 1.5833555472623555, "grad_norm": 1.6903644800186157, "learning_rate": 4.0604e-06, "loss": 0.0268, "step": 29699 }, { "epoch": 1.5834088606920083, "grad_norm": 1.7285112142562866, "learning_rate": 4.0602000000000005e-06, "loss": 0.028, "step": 29700 }, { "epoch": 1.5834621741216612, "grad_norm": 1.6654192209243774, "learning_rate": 4.060000000000001e-06, "loss": 0.0292, "step": 29701 }, { "epoch": 1.5835154875513142, "grad_norm": 1.76859712600708, "learning_rate": 4.0598000000000005e-06, "loss": 0.0391, "step": 29702 }, { "epoch": 1.583568800980967, "grad_norm": 2.1283538341522217, "learning_rate": 4.0596e-06, "loss": 0.05, "step": 29703 }, { "epoch": 1.5836221144106202, "grad_norm": 1.3769561052322388, "learning_rate": 4.0594000000000005e-06, "loss": 0.009, "step": 29704 }, { "epoch": 1.583675427840273, "grad_norm": 1.3455251455307007, "learning_rate": 4.0592e-06, "loss": 0.0183, "step": 29705 }, { "epoch": 1.583728741269926, "grad_norm": 2.4309873580932617, "learning_rate": 4.0590000000000004e-06, "loss": 0.0202, "step": 29706 }, { "epoch": 1.583782054699579, "grad_norm": 1.7456670999526978, "learning_rate": 4.058800000000001e-06, "loss": 0.0577, "step": 29707 }, { "epoch": 1.5838353681292316, "grad_norm": 0.7261773943901062, "learning_rate": 4.0586e-06, "loss": 0.01, "step": 29708 }, { "epoch": 1.5838886815588848, "grad_norm": 2.5515987873077393, "learning_rate": 4.0584e-06, "loss": 0.0165, "step": 29709 }, { "epoch": 1.5839419949885376, "grad_norm": 1.5744471549987793, "learning_rate": 4.0582e-06, "loss": 0.0107, "step": 29710 }, { "epoch": 1.5839953084181906, "grad_norm": 1.422153115272522, "learning_rate": 4.058e-06, "loss": 0.0128, "step": 29711 }, { "epoch": 1.5840486218478436, "grad_norm": 3.7887909412384033, "learning_rate": 4.0578e-06, "loss": 0.0304, "step": 29712 }, { "epoch": 1.5841019352774963, "grad_norm": 1.2739440202713013, "learning_rate": 4.057600000000001e-06, "loss": 0.0353, "step": 29713 }, { "epoch": 1.5841552487071493, "grad_norm": 1.521118402481079, "learning_rate": 4.0574e-06, "loss": 0.0118, "step": 29714 }, { "epoch": 1.5842085621368023, "grad_norm": 0.3510149121284485, "learning_rate": 4.057200000000001e-06, "loss": 0.0039, "step": 29715 }, { "epoch": 1.584261875566455, "grad_norm": 1.4808422327041626, "learning_rate": 4.057e-06, "loss": 0.0345, "step": 29716 }, { "epoch": 1.5843151889961082, "grad_norm": 1.0817950963974, "learning_rate": 4.0568e-06, "loss": 0.0137, "step": 29717 }, { "epoch": 1.584368502425761, "grad_norm": 1.5805000066757202, "learning_rate": 4.0566e-06, "loss": 0.0257, "step": 29718 }, { "epoch": 1.584421815855414, "grad_norm": 2.2354114055633545, "learning_rate": 4.056400000000001e-06, "loss": 0.0418, "step": 29719 }, { "epoch": 1.584475129285067, "grad_norm": 3.2099854946136475, "learning_rate": 4.0562e-06, "loss": 0.0596, "step": 29720 }, { "epoch": 1.5845284427147197, "grad_norm": 1.1964246034622192, "learning_rate": 4.056000000000001e-06, "loss": 0.0207, "step": 29721 }, { "epoch": 1.584581756144373, "grad_norm": 2.23452091217041, "learning_rate": 4.0558e-06, "loss": 0.0411, "step": 29722 }, { "epoch": 1.5846350695740257, "grad_norm": 1.6813491582870483, "learning_rate": 4.0556e-06, "loss": 0.0182, "step": 29723 }, { "epoch": 1.5846883830036786, "grad_norm": 1.0545742511749268, "learning_rate": 4.0554e-06, "loss": 0.0124, "step": 29724 }, { "epoch": 1.5847416964333316, "grad_norm": 1.1893045902252197, "learning_rate": 4.055200000000001e-06, "loss": 0.0094, "step": 29725 }, { "epoch": 1.5847950098629844, "grad_norm": 1.280138373374939, "learning_rate": 4.055000000000001e-06, "loss": 0.0094, "step": 29726 }, { "epoch": 1.5848483232926374, "grad_norm": 0.769537627696991, "learning_rate": 4.0548000000000005e-06, "loss": 0.0086, "step": 29727 }, { "epoch": 1.5849016367222903, "grad_norm": 1.5650986433029175, "learning_rate": 4.0546e-06, "loss": 0.0156, "step": 29728 }, { "epoch": 1.5849549501519433, "grad_norm": 1.2995240688323975, "learning_rate": 4.0544000000000005e-06, "loss": 0.0274, "step": 29729 }, { "epoch": 1.5850082635815963, "grad_norm": 1.5465905666351318, "learning_rate": 4.0542e-06, "loss": 0.0084, "step": 29730 }, { "epoch": 1.585061577011249, "grad_norm": 2.092949867248535, "learning_rate": 4.0540000000000005e-06, "loss": 0.0238, "step": 29731 }, { "epoch": 1.585114890440902, "grad_norm": 0.7727630734443665, "learning_rate": 4.053800000000001e-06, "loss": 0.0061, "step": 29732 }, { "epoch": 1.585168203870555, "grad_norm": 0.832449197769165, "learning_rate": 4.0536000000000005e-06, "loss": 0.0183, "step": 29733 }, { "epoch": 1.5852215173002078, "grad_norm": 1.093945026397705, "learning_rate": 4.0534e-06, "loss": 0.0072, "step": 29734 }, { "epoch": 1.585274830729861, "grad_norm": 1.2073918581008911, "learning_rate": 4.0532000000000004e-06, "loss": 0.0222, "step": 29735 }, { "epoch": 1.5853281441595137, "grad_norm": 3.4061224460601807, "learning_rate": 4.053e-06, "loss": 0.0374, "step": 29736 }, { "epoch": 1.5853814575891667, "grad_norm": 0.5427353382110596, "learning_rate": 4.0528e-06, "loss": 0.0048, "step": 29737 }, { "epoch": 1.5854347710188197, "grad_norm": 1.715840220451355, "learning_rate": 4.052600000000001e-06, "loss": 0.0121, "step": 29738 }, { "epoch": 1.5854880844484724, "grad_norm": 0.9242352247238159, "learning_rate": 4.0524e-06, "loss": 0.0065, "step": 29739 }, { "epoch": 1.5855413978781256, "grad_norm": 1.828287124633789, "learning_rate": 4.052200000000001e-06, "loss": 0.0154, "step": 29740 }, { "epoch": 1.5855947113077784, "grad_norm": 0.9899628758430481, "learning_rate": 4.052e-06, "loss": 0.0079, "step": 29741 }, { "epoch": 1.5856480247374314, "grad_norm": 1.5567822456359863, "learning_rate": 4.0518e-06, "loss": 0.011, "step": 29742 }, { "epoch": 1.5857013381670844, "grad_norm": 1.8479301929473877, "learning_rate": 4.0516e-06, "loss": 0.017, "step": 29743 }, { "epoch": 1.5857546515967371, "grad_norm": 1.35990571975708, "learning_rate": 4.051400000000001e-06, "loss": 0.0096, "step": 29744 }, { "epoch": 1.58580796502639, "grad_norm": 2.017315149307251, "learning_rate": 4.0512e-06, "loss": 0.0174, "step": 29745 }, { "epoch": 1.585861278456043, "grad_norm": 2.4602296352386475, "learning_rate": 4.051000000000001e-06, "loss": 0.0183, "step": 29746 }, { "epoch": 1.585914591885696, "grad_norm": 1.7102402448654175, "learning_rate": 4.0508e-06, "loss": 0.0199, "step": 29747 }, { "epoch": 1.585967905315349, "grad_norm": 1.8079233169555664, "learning_rate": 4.0506e-06, "loss": 0.0323, "step": 29748 }, { "epoch": 1.5860212187450018, "grad_norm": 3.5711963176727295, "learning_rate": 4.0504e-06, "loss": 0.0185, "step": 29749 }, { "epoch": 1.5860745321746548, "grad_norm": 1.0386340618133545, "learning_rate": 4.050200000000001e-06, "loss": 0.0201, "step": 29750 }, { "epoch": 1.5861278456043078, "grad_norm": 3.416961908340454, "learning_rate": 4.05e-06, "loss": 0.034, "step": 29751 }, { "epoch": 1.5861811590339605, "grad_norm": 1.794057846069336, "learning_rate": 4.049800000000001e-06, "loss": 0.0372, "step": 29752 }, { "epoch": 1.5862344724636137, "grad_norm": 1.104809284210205, "learning_rate": 4.0496e-06, "loss": 0.0088, "step": 29753 }, { "epoch": 1.5862877858932665, "grad_norm": 2.154204845428467, "learning_rate": 4.0494000000000006e-06, "loss": 0.0282, "step": 29754 }, { "epoch": 1.5863410993229194, "grad_norm": 2.7264387607574463, "learning_rate": 4.0492e-06, "loss": 0.02, "step": 29755 }, { "epoch": 1.5863944127525724, "grad_norm": 3.397963523864746, "learning_rate": 4.0490000000000005e-06, "loss": 0.0216, "step": 29756 }, { "epoch": 1.5864477261822252, "grad_norm": 1.6515954732894897, "learning_rate": 4.0488e-06, "loss": 0.0202, "step": 29757 }, { "epoch": 1.5865010396118784, "grad_norm": 2.6994998455047607, "learning_rate": 4.0486000000000005e-06, "loss": 0.0253, "step": 29758 }, { "epoch": 1.5865543530415311, "grad_norm": 2.5398271083831787, "learning_rate": 4.0484e-06, "loss": 0.0402, "step": 29759 }, { "epoch": 1.5866076664711841, "grad_norm": 3.0418763160705566, "learning_rate": 4.0482000000000005e-06, "loss": 0.0374, "step": 29760 }, { "epoch": 1.586660979900837, "grad_norm": 1.033890724182129, "learning_rate": 4.048e-06, "loss": 0.0266, "step": 29761 }, { "epoch": 1.5867142933304899, "grad_norm": 1.4129277467727661, "learning_rate": 4.0478000000000004e-06, "loss": 0.0095, "step": 29762 }, { "epoch": 1.5867676067601428, "grad_norm": 1.0054811239242554, "learning_rate": 4.0476e-06, "loss": 0.008, "step": 29763 }, { "epoch": 1.5868209201897958, "grad_norm": 2.752037286758423, "learning_rate": 4.0474e-06, "loss": 0.0388, "step": 29764 }, { "epoch": 1.5868742336194488, "grad_norm": 1.2844985723495483, "learning_rate": 4.047200000000001e-06, "loss": 0.0184, "step": 29765 }, { "epoch": 1.5869275470491018, "grad_norm": 0.6777371168136597, "learning_rate": 4.047e-06, "loss": 0.0056, "step": 29766 }, { "epoch": 1.5869808604787545, "grad_norm": 1.5635900497436523, "learning_rate": 4.0468e-06, "loss": 0.0204, "step": 29767 }, { "epoch": 1.5870341739084075, "grad_norm": 2.595146417617798, "learning_rate": 4.0466e-06, "loss": 0.0128, "step": 29768 }, { "epoch": 1.5870874873380605, "grad_norm": 0.5208448171615601, "learning_rate": 4.0464e-06, "loss": 0.0055, "step": 29769 }, { "epoch": 1.5871408007677132, "grad_norm": 1.4553416967391968, "learning_rate": 4.0462e-06, "loss": 0.0132, "step": 29770 }, { "epoch": 1.5871941141973664, "grad_norm": 1.1493760347366333, "learning_rate": 4.046000000000001e-06, "loss": 0.0236, "step": 29771 }, { "epoch": 1.5872474276270192, "grad_norm": 0.5923435091972351, "learning_rate": 4.0458e-06, "loss": 0.004, "step": 29772 }, { "epoch": 1.5873007410566722, "grad_norm": 1.903563141822815, "learning_rate": 4.0456e-06, "loss": 0.0248, "step": 29773 }, { "epoch": 1.5873540544863252, "grad_norm": 0.8758534789085388, "learning_rate": 4.0454e-06, "loss": 0.0179, "step": 29774 }, { "epoch": 1.587407367915978, "grad_norm": 2.2730579376220703, "learning_rate": 4.0452e-06, "loss": 0.023, "step": 29775 }, { "epoch": 1.5874606813456311, "grad_norm": 0.900691032409668, "learning_rate": 4.045e-06, "loss": 0.008, "step": 29776 }, { "epoch": 1.5875139947752839, "grad_norm": 1.0143524408340454, "learning_rate": 4.044800000000001e-06, "loss": 0.0211, "step": 29777 }, { "epoch": 1.5875673082049369, "grad_norm": 2.2493903636932373, "learning_rate": 4.0446e-06, "loss": 0.0245, "step": 29778 }, { "epoch": 1.5876206216345898, "grad_norm": 0.991528332233429, "learning_rate": 4.044400000000001e-06, "loss": 0.0147, "step": 29779 }, { "epoch": 1.5876739350642426, "grad_norm": 1.686383605003357, "learning_rate": 4.0442e-06, "loss": 0.0249, "step": 29780 }, { "epoch": 1.5877272484938956, "grad_norm": 2.1795296669006348, "learning_rate": 4.044e-06, "loss": 0.0278, "step": 29781 }, { "epoch": 1.5877805619235486, "grad_norm": 0.6447858214378357, "learning_rate": 4.0438e-06, "loss": 0.0051, "step": 29782 }, { "epoch": 1.5878338753532013, "grad_norm": 2.4892048835754395, "learning_rate": 4.0436000000000006e-06, "loss": 0.0168, "step": 29783 }, { "epoch": 1.5878871887828545, "grad_norm": 2.335981607437134, "learning_rate": 4.0434e-06, "loss": 0.0291, "step": 29784 }, { "epoch": 1.5879405022125073, "grad_norm": 1.1734588146209717, "learning_rate": 4.0432000000000005e-06, "loss": 0.0201, "step": 29785 }, { "epoch": 1.5879938156421602, "grad_norm": 2.5005295276641846, "learning_rate": 4.043e-06, "loss": 0.0328, "step": 29786 }, { "epoch": 1.5880471290718132, "grad_norm": 0.6148765087127686, "learning_rate": 4.0428e-06, "loss": 0.0055, "step": 29787 }, { "epoch": 1.588100442501466, "grad_norm": 2.8918259143829346, "learning_rate": 4.0426e-06, "loss": 0.03, "step": 29788 }, { "epoch": 1.5881537559311192, "grad_norm": 0.9855176210403442, "learning_rate": 4.0424000000000005e-06, "loss": 0.0072, "step": 29789 }, { "epoch": 1.588207069360772, "grad_norm": 0.3324702978134155, "learning_rate": 4.042200000000001e-06, "loss": 0.0031, "step": 29790 }, { "epoch": 1.588260382790425, "grad_norm": 0.8522728085517883, "learning_rate": 4.0420000000000004e-06, "loss": 0.0071, "step": 29791 }, { "epoch": 1.588313696220078, "grad_norm": 2.0529608726501465, "learning_rate": 4.0418e-06, "loss": 0.0301, "step": 29792 }, { "epoch": 1.5883670096497307, "grad_norm": 1.7434245347976685, "learning_rate": 4.0416e-06, "loss": 0.0222, "step": 29793 }, { "epoch": 1.5884203230793839, "grad_norm": 1.54630446434021, "learning_rate": 4.0414e-06, "loss": 0.0195, "step": 29794 }, { "epoch": 1.5884736365090366, "grad_norm": 1.7185851335525513, "learning_rate": 4.0412e-06, "loss": 0.0352, "step": 29795 }, { "epoch": 1.5885269499386896, "grad_norm": 2.726954221725464, "learning_rate": 4.041000000000001e-06, "loss": 0.0244, "step": 29796 }, { "epoch": 1.5885802633683426, "grad_norm": 2.307222843170166, "learning_rate": 4.0408e-06, "loss": 0.024, "step": 29797 }, { "epoch": 1.5886335767979953, "grad_norm": 1.7925399541854858, "learning_rate": 4.0406e-06, "loss": 0.0076, "step": 29798 }, { "epoch": 1.5886868902276483, "grad_norm": 3.6884427070617676, "learning_rate": 4.0404e-06, "loss": 0.0147, "step": 29799 }, { "epoch": 1.5887402036573013, "grad_norm": 1.6728507280349731, "learning_rate": 4.0402e-06, "loss": 0.0204, "step": 29800 }, { "epoch": 1.588793517086954, "grad_norm": 1.2585369348526, "learning_rate": 4.04e-06, "loss": 0.0205, "step": 29801 }, { "epoch": 1.5888468305166072, "grad_norm": 1.8949408531188965, "learning_rate": 4.039800000000001e-06, "loss": 0.0256, "step": 29802 }, { "epoch": 1.58890014394626, "grad_norm": 1.9358478784561157, "learning_rate": 4.0396e-06, "loss": 0.032, "step": 29803 }, { "epoch": 1.588953457375913, "grad_norm": 1.6088470220565796, "learning_rate": 4.039400000000001e-06, "loss": 0.018, "step": 29804 }, { "epoch": 1.589006770805566, "grad_norm": 1.3862437009811401, "learning_rate": 4.0392e-06, "loss": 0.0085, "step": 29805 }, { "epoch": 1.5890600842352187, "grad_norm": 0.2048768401145935, "learning_rate": 4.039e-06, "loss": 0.0028, "step": 29806 }, { "epoch": 1.589113397664872, "grad_norm": 1.5977097749710083, "learning_rate": 4.0388e-06, "loss": 0.0167, "step": 29807 }, { "epoch": 1.5891667110945247, "grad_norm": 1.8676211833953857, "learning_rate": 4.038600000000001e-06, "loss": 0.0228, "step": 29808 }, { "epoch": 1.5892200245241777, "grad_norm": 2.155308485031128, "learning_rate": 4.0384e-06, "loss": 0.019, "step": 29809 }, { "epoch": 1.5892733379538306, "grad_norm": 0.9978672862052917, "learning_rate": 4.0382000000000006e-06, "loss": 0.0081, "step": 29810 }, { "epoch": 1.5893266513834834, "grad_norm": 0.6457788944244385, "learning_rate": 4.038e-06, "loss": 0.0045, "step": 29811 }, { "epoch": 1.5893799648131366, "grad_norm": 1.0003348588943481, "learning_rate": 4.0378e-06, "loss": 0.0142, "step": 29812 }, { "epoch": 1.5894332782427893, "grad_norm": 2.0677096843719482, "learning_rate": 4.0376e-06, "loss": 0.0151, "step": 29813 }, { "epoch": 1.5894865916724423, "grad_norm": 1.1338247060775757, "learning_rate": 4.0374000000000005e-06, "loss": 0.0168, "step": 29814 }, { "epoch": 1.5895399051020953, "grad_norm": 3.78547739982605, "learning_rate": 4.037200000000001e-06, "loss": 0.0203, "step": 29815 }, { "epoch": 1.589593218531748, "grad_norm": 1.550426959991455, "learning_rate": 4.0370000000000005e-06, "loss": 0.0135, "step": 29816 }, { "epoch": 1.589646531961401, "grad_norm": 1.0684231519699097, "learning_rate": 4.0368e-06, "loss": 0.0162, "step": 29817 }, { "epoch": 1.589699845391054, "grad_norm": 1.8777544498443604, "learning_rate": 4.0366000000000005e-06, "loss": 0.0525, "step": 29818 }, { "epoch": 1.5897531588207068, "grad_norm": 3.5035183429718018, "learning_rate": 4.0364e-06, "loss": 0.0175, "step": 29819 }, { "epoch": 1.58980647225036, "grad_norm": 1.425494909286499, "learning_rate": 4.0362e-06, "loss": 0.0174, "step": 29820 }, { "epoch": 1.5898597856800127, "grad_norm": 1.951146125793457, "learning_rate": 4.036000000000001e-06, "loss": 0.0196, "step": 29821 }, { "epoch": 1.5899130991096657, "grad_norm": 2.5800912380218506, "learning_rate": 4.0358e-06, "loss": 0.0204, "step": 29822 }, { "epoch": 1.5899664125393187, "grad_norm": 3.342156171798706, "learning_rate": 4.0356e-06, "loss": 0.0299, "step": 29823 }, { "epoch": 1.5900197259689715, "grad_norm": 1.556412696838379, "learning_rate": 4.0354e-06, "loss": 0.0158, "step": 29824 }, { "epoch": 1.5900730393986247, "grad_norm": 1.4395068883895874, "learning_rate": 4.0352e-06, "loss": 0.0219, "step": 29825 }, { "epoch": 1.5901263528282774, "grad_norm": 0.8843359351158142, "learning_rate": 4.035e-06, "loss": 0.0092, "step": 29826 }, { "epoch": 1.5901796662579304, "grad_norm": 2.5357589721679688, "learning_rate": 4.034800000000001e-06, "loss": 0.0319, "step": 29827 }, { "epoch": 1.5902329796875834, "grad_norm": 1.3158314228057861, "learning_rate": 4.0346e-06, "loss": 0.0101, "step": 29828 }, { "epoch": 1.5902862931172361, "grad_norm": 1.7918857336044312, "learning_rate": 4.034400000000001e-06, "loss": 0.0305, "step": 29829 }, { "epoch": 1.590339606546889, "grad_norm": 0.9560837745666504, "learning_rate": 4.0342e-06, "loss": 0.008, "step": 29830 }, { "epoch": 1.590392919976542, "grad_norm": 3.0957722663879395, "learning_rate": 4.034e-06, "loss": 0.0353, "step": 29831 }, { "epoch": 1.590446233406195, "grad_norm": 1.8646894693374634, "learning_rate": 4.0338e-06, "loss": 0.01, "step": 29832 }, { "epoch": 1.590499546835848, "grad_norm": 2.2818403244018555, "learning_rate": 4.033600000000001e-06, "loss": 0.0337, "step": 29833 }, { "epoch": 1.5905528602655008, "grad_norm": 3.0791664123535156, "learning_rate": 4.0334e-06, "loss": 0.0378, "step": 29834 }, { "epoch": 1.5906061736951538, "grad_norm": 3.9442648887634277, "learning_rate": 4.033200000000001e-06, "loss": 0.0447, "step": 29835 }, { "epoch": 1.5906594871248068, "grad_norm": 1.570043683052063, "learning_rate": 4.033e-06, "loss": 0.015, "step": 29836 }, { "epoch": 1.5907128005544595, "grad_norm": 1.762783169746399, "learning_rate": 4.032800000000001e-06, "loss": 0.0164, "step": 29837 }, { "epoch": 1.5907661139841127, "grad_norm": 1.2666884660720825, "learning_rate": 4.0326e-06, "loss": 0.0107, "step": 29838 }, { "epoch": 1.5908194274137655, "grad_norm": 2.0034148693084717, "learning_rate": 4.0324000000000006e-06, "loss": 0.0365, "step": 29839 }, { "epoch": 1.5908727408434185, "grad_norm": 1.4681426286697388, "learning_rate": 4.032200000000001e-06, "loss": 0.0053, "step": 29840 }, { "epoch": 1.5909260542730714, "grad_norm": 1.3661444187164307, "learning_rate": 4.0320000000000005e-06, "loss": 0.0084, "step": 29841 }, { "epoch": 1.5909793677027242, "grad_norm": 0.981075644493103, "learning_rate": 4.0318e-06, "loss": 0.0094, "step": 29842 }, { "epoch": 1.5910326811323774, "grad_norm": 1.72395658493042, "learning_rate": 4.0316000000000005e-06, "loss": 0.0365, "step": 29843 }, { "epoch": 1.5910859945620301, "grad_norm": 2.9865827560424805, "learning_rate": 4.0314e-06, "loss": 0.0123, "step": 29844 }, { "epoch": 1.5911393079916831, "grad_norm": 0.8735945224761963, "learning_rate": 4.0312000000000005e-06, "loss": 0.021, "step": 29845 }, { "epoch": 1.591192621421336, "grad_norm": 0.28049975633621216, "learning_rate": 4.031000000000001e-06, "loss": 0.0022, "step": 29846 }, { "epoch": 1.5912459348509889, "grad_norm": 2.9422082901000977, "learning_rate": 4.0308000000000004e-06, "loss": 0.0635, "step": 29847 }, { "epoch": 1.5912992482806418, "grad_norm": 1.1287237405776978, "learning_rate": 4.0306e-06, "loss": 0.0121, "step": 29848 }, { "epoch": 1.5913525617102948, "grad_norm": 0.20651598274707794, "learning_rate": 4.0304e-06, "loss": 0.0023, "step": 29849 }, { "epoch": 1.5914058751399478, "grad_norm": 1.6626875400543213, "learning_rate": 4.0302e-06, "loss": 0.0226, "step": 29850 }, { "epoch": 1.5914591885696008, "grad_norm": 1.5599136352539062, "learning_rate": 4.03e-06, "loss": 0.0079, "step": 29851 }, { "epoch": 1.5915125019992535, "grad_norm": 0.6548381447792053, "learning_rate": 4.029800000000001e-06, "loss": 0.004, "step": 29852 }, { "epoch": 1.5915658154289065, "grad_norm": 3.31099009513855, "learning_rate": 4.0296e-06, "loss": 0.0312, "step": 29853 }, { "epoch": 1.5916191288585595, "grad_norm": 2.474515914916992, "learning_rate": 4.029400000000001e-06, "loss": 0.0305, "step": 29854 }, { "epoch": 1.5916724422882123, "grad_norm": 1.6863471269607544, "learning_rate": 4.0292e-06, "loss": 0.0291, "step": 29855 }, { "epoch": 1.5917257557178655, "grad_norm": 0.6457720994949341, "learning_rate": 4.029e-06, "loss": 0.0082, "step": 29856 }, { "epoch": 1.5917790691475182, "grad_norm": 1.0441815853118896, "learning_rate": 4.0288e-06, "loss": 0.0142, "step": 29857 }, { "epoch": 1.5918323825771712, "grad_norm": 0.17150196433067322, "learning_rate": 4.028600000000001e-06, "loss": 0.0021, "step": 29858 }, { "epoch": 1.5918856960068242, "grad_norm": 1.1768988370895386, "learning_rate": 4.0284e-06, "loss": 0.0095, "step": 29859 }, { "epoch": 1.591939009436477, "grad_norm": 1.2754148244857788, "learning_rate": 4.028200000000001e-06, "loss": 0.0136, "step": 29860 }, { "epoch": 1.5919923228661301, "grad_norm": 1.3653477430343628, "learning_rate": 4.028e-06, "loss": 0.0073, "step": 29861 }, { "epoch": 1.5920456362957829, "grad_norm": 2.0716872215270996, "learning_rate": 4.027800000000001e-06, "loss": 0.0269, "step": 29862 }, { "epoch": 1.5920989497254359, "grad_norm": 1.9853953123092651, "learning_rate": 4.0276e-06, "loss": 0.0103, "step": 29863 }, { "epoch": 1.5921522631550888, "grad_norm": 2.445303440093994, "learning_rate": 4.027400000000001e-06, "loss": 0.0325, "step": 29864 }, { "epoch": 1.5922055765847416, "grad_norm": 2.4940741062164307, "learning_rate": 4.0272e-06, "loss": 0.0638, "step": 29865 }, { "epoch": 1.5922588900143946, "grad_norm": 2.717243194580078, "learning_rate": 4.027000000000001e-06, "loss": 0.0188, "step": 29866 }, { "epoch": 1.5923122034440476, "grad_norm": 2.0715792179107666, "learning_rate": 4.0268e-06, "loss": 0.0183, "step": 29867 }, { "epoch": 1.5923655168737005, "grad_norm": 1.7435517311096191, "learning_rate": 4.0266000000000005e-06, "loss": 0.0153, "step": 29868 }, { "epoch": 1.5924188303033535, "grad_norm": 0.9983465671539307, "learning_rate": 4.0264e-06, "loss": 0.0086, "step": 29869 }, { "epoch": 1.5924721437330063, "grad_norm": 0.8552958965301514, "learning_rate": 4.0262000000000005e-06, "loss": 0.0064, "step": 29870 }, { "epoch": 1.5925254571626593, "grad_norm": 0.7571386694908142, "learning_rate": 4.026e-06, "loss": 0.0056, "step": 29871 }, { "epoch": 1.5925787705923122, "grad_norm": 2.000244140625, "learning_rate": 4.0258000000000005e-06, "loss": 0.0109, "step": 29872 }, { "epoch": 1.592632084021965, "grad_norm": 1.9033797979354858, "learning_rate": 4.0256e-06, "loss": 0.0139, "step": 29873 }, { "epoch": 1.5926853974516182, "grad_norm": 3.315427780151367, "learning_rate": 4.0254000000000005e-06, "loss": 0.0266, "step": 29874 }, { "epoch": 1.592738710881271, "grad_norm": 0.881257951259613, "learning_rate": 4.0252e-06, "loss": 0.0147, "step": 29875 }, { "epoch": 1.592792024310924, "grad_norm": 0.9383065104484558, "learning_rate": 4.0250000000000004e-06, "loss": 0.015, "step": 29876 }, { "epoch": 1.592845337740577, "grad_norm": 1.3534553050994873, "learning_rate": 4.0248e-06, "loss": 0.0101, "step": 29877 }, { "epoch": 1.5928986511702297, "grad_norm": 0.29251977801322937, "learning_rate": 4.0246e-06, "loss": 0.0035, "step": 29878 }, { "epoch": 1.5929519645998829, "grad_norm": 1.375710129737854, "learning_rate": 4.024400000000001e-06, "loss": 0.0117, "step": 29879 }, { "epoch": 1.5930052780295356, "grad_norm": 2.609563112258911, "learning_rate": 4.0242e-06, "loss": 0.0374, "step": 29880 }, { "epoch": 1.5930585914591886, "grad_norm": 1.914791464805603, "learning_rate": 4.024e-06, "loss": 0.0146, "step": 29881 }, { "epoch": 1.5931119048888416, "grad_norm": 0.8432157635688782, "learning_rate": 4.0238e-06, "loss": 0.0153, "step": 29882 }, { "epoch": 1.5931652183184943, "grad_norm": 0.7880335450172424, "learning_rate": 4.0236e-06, "loss": 0.0055, "step": 29883 }, { "epoch": 1.5932185317481473, "grad_norm": 2.514169692993164, "learning_rate": 4.0234e-06, "loss": 0.024, "step": 29884 }, { "epoch": 1.5932718451778003, "grad_norm": 1.3678349256515503, "learning_rate": 4.023200000000001e-06, "loss": 0.0305, "step": 29885 }, { "epoch": 1.593325158607453, "grad_norm": 1.278701663017273, "learning_rate": 4.023e-06, "loss": 0.0241, "step": 29886 }, { "epoch": 1.5933784720371063, "grad_norm": 2.7244012355804443, "learning_rate": 4.022800000000001e-06, "loss": 0.0362, "step": 29887 }, { "epoch": 1.593431785466759, "grad_norm": 0.8795855045318604, "learning_rate": 4.0226e-06, "loss": 0.0179, "step": 29888 }, { "epoch": 1.593485098896412, "grad_norm": 0.9444965720176697, "learning_rate": 4.0224e-06, "loss": 0.0079, "step": 29889 }, { "epoch": 1.593538412326065, "grad_norm": 1.6612999439239502, "learning_rate": 4.0222e-06, "loss": 0.0204, "step": 29890 }, { "epoch": 1.5935917257557177, "grad_norm": 2.419626235961914, "learning_rate": 4.022000000000001e-06, "loss": 0.0206, "step": 29891 }, { "epoch": 1.593645039185371, "grad_norm": 2.114109516143799, "learning_rate": 4.0218e-06, "loss": 0.0168, "step": 29892 }, { "epoch": 1.5936983526150237, "grad_norm": 2.4761486053466797, "learning_rate": 4.021600000000001e-06, "loss": 0.0538, "step": 29893 }, { "epoch": 1.5937516660446767, "grad_norm": 1.1888768672943115, "learning_rate": 4.0214e-06, "loss": 0.0066, "step": 29894 }, { "epoch": 1.5938049794743296, "grad_norm": 2.1470859050750732, "learning_rate": 4.0212e-06, "loss": 0.0097, "step": 29895 }, { "epoch": 1.5938582929039824, "grad_norm": 1.3850234746932983, "learning_rate": 4.021e-06, "loss": 0.0097, "step": 29896 }, { "epoch": 1.5939116063336356, "grad_norm": 2.244462728500366, "learning_rate": 4.0208000000000005e-06, "loss": 0.0123, "step": 29897 }, { "epoch": 1.5939649197632884, "grad_norm": 3.3056082725524902, "learning_rate": 4.0206e-06, "loss": 0.022, "step": 29898 }, { "epoch": 1.5940182331929413, "grad_norm": 2.3179538249969482, "learning_rate": 4.0204000000000005e-06, "loss": 0.0213, "step": 29899 }, { "epoch": 1.5940715466225943, "grad_norm": 1.2533760070800781, "learning_rate": 4.0202e-06, "loss": 0.0232, "step": 29900 }, { "epoch": 1.594124860052247, "grad_norm": 3.952807664871216, "learning_rate": 4.0200000000000005e-06, "loss": 0.0409, "step": 29901 }, { "epoch": 1.5941781734819, "grad_norm": 1.7668870687484741, "learning_rate": 4.0198e-06, "loss": 0.0099, "step": 29902 }, { "epoch": 1.594231486911553, "grad_norm": 3.0132787227630615, "learning_rate": 4.0196000000000004e-06, "loss": 0.0128, "step": 29903 }, { "epoch": 1.5942848003412058, "grad_norm": 2.267538070678711, "learning_rate": 4.019400000000001e-06, "loss": 0.02, "step": 29904 }, { "epoch": 1.594338113770859, "grad_norm": 0.9089950323104858, "learning_rate": 4.0192e-06, "loss": 0.0237, "step": 29905 }, { "epoch": 1.5943914272005117, "grad_norm": 3.755396842956543, "learning_rate": 4.019e-06, "loss": 0.0333, "step": 29906 }, { "epoch": 1.5944447406301647, "grad_norm": 1.5667670965194702, "learning_rate": 4.0188e-06, "loss": 0.0358, "step": 29907 }, { "epoch": 1.5944980540598177, "grad_norm": 1.7465736865997314, "learning_rate": 4.0186e-06, "loss": 0.0347, "step": 29908 }, { "epoch": 1.5945513674894705, "grad_norm": 1.1509032249450684, "learning_rate": 4.0184e-06, "loss": 0.0082, "step": 29909 }, { "epoch": 1.5946046809191237, "grad_norm": 1.363508701324463, "learning_rate": 4.018200000000001e-06, "loss": 0.0152, "step": 29910 }, { "epoch": 1.5946579943487764, "grad_norm": 1.571494221687317, "learning_rate": 4.018e-06, "loss": 0.0445, "step": 29911 }, { "epoch": 1.5947113077784294, "grad_norm": 1.1058261394500732, "learning_rate": 4.017800000000001e-06, "loss": 0.0224, "step": 29912 }, { "epoch": 1.5947646212080824, "grad_norm": 0.713675320148468, "learning_rate": 4.0176e-06, "loss": 0.0031, "step": 29913 }, { "epoch": 1.5948179346377351, "grad_norm": 0.6418922543525696, "learning_rate": 4.0174e-06, "loss": 0.0041, "step": 29914 }, { "epoch": 1.5948712480673883, "grad_norm": 1.8965175151824951, "learning_rate": 4.0172e-06, "loss": 0.0305, "step": 29915 }, { "epoch": 1.594924561497041, "grad_norm": 2.4810221195220947, "learning_rate": 4.017000000000001e-06, "loss": 0.0389, "step": 29916 }, { "epoch": 1.594977874926694, "grad_norm": 1.345254898071289, "learning_rate": 4.0168e-06, "loss": 0.0124, "step": 29917 }, { "epoch": 1.595031188356347, "grad_norm": 3.9391510486602783, "learning_rate": 4.016600000000001e-06, "loss": 0.0228, "step": 29918 }, { "epoch": 1.5950845017859998, "grad_norm": 0.5890215039253235, "learning_rate": 4.0164e-06, "loss": 0.0046, "step": 29919 }, { "epoch": 1.5951378152156528, "grad_norm": 1.9141438007354736, "learning_rate": 4.0162e-06, "loss": 0.017, "step": 29920 }, { "epoch": 1.5951911286453058, "grad_norm": 3.625380039215088, "learning_rate": 4.016e-06, "loss": 0.0289, "step": 29921 }, { "epoch": 1.5952444420749585, "grad_norm": 0.7478002905845642, "learning_rate": 4.015800000000001e-06, "loss": 0.0066, "step": 29922 }, { "epoch": 1.5952977555046117, "grad_norm": 1.8471348285675049, "learning_rate": 4.0156e-06, "loss": 0.0237, "step": 29923 }, { "epoch": 1.5953510689342645, "grad_norm": 2.393442392349243, "learning_rate": 4.0154000000000006e-06, "loss": 0.0172, "step": 29924 }, { "epoch": 1.5954043823639175, "grad_norm": 1.4949194192886353, "learning_rate": 4.0152e-06, "loss": 0.0255, "step": 29925 }, { "epoch": 1.5954576957935704, "grad_norm": 1.4200352430343628, "learning_rate": 4.0150000000000005e-06, "loss": 0.0255, "step": 29926 }, { "epoch": 1.5955110092232232, "grad_norm": 3.303339958190918, "learning_rate": 4.0148e-06, "loss": 0.0371, "step": 29927 }, { "epoch": 1.5955643226528764, "grad_norm": 2.6761245727539062, "learning_rate": 4.0146000000000005e-06, "loss": 0.0307, "step": 29928 }, { "epoch": 1.5956176360825292, "grad_norm": 0.5590291619300842, "learning_rate": 4.014400000000001e-06, "loss": 0.0051, "step": 29929 }, { "epoch": 1.5956709495121821, "grad_norm": 2.486975908279419, "learning_rate": 4.0142000000000005e-06, "loss": 0.032, "step": 29930 }, { "epoch": 1.5957242629418351, "grad_norm": 3.6316311359405518, "learning_rate": 4.014e-06, "loss": 0.0511, "step": 29931 }, { "epoch": 1.5957775763714879, "grad_norm": 1.9891576766967773, "learning_rate": 4.0138000000000004e-06, "loss": 0.0219, "step": 29932 }, { "epoch": 1.595830889801141, "grad_norm": 1.3390512466430664, "learning_rate": 4.0136e-06, "loss": 0.0205, "step": 29933 }, { "epoch": 1.5958842032307938, "grad_norm": 1.192927360534668, "learning_rate": 4.0134e-06, "loss": 0.0091, "step": 29934 }, { "epoch": 1.5959375166604468, "grad_norm": 2.2520358562469482, "learning_rate": 4.013200000000001e-06, "loss": 0.031, "step": 29935 }, { "epoch": 1.5959908300900998, "grad_norm": 2.494677782058716, "learning_rate": 4.013e-06, "loss": 0.043, "step": 29936 }, { "epoch": 1.5960441435197525, "grad_norm": 1.8398863077163696, "learning_rate": 4.012800000000001e-06, "loss": 0.0315, "step": 29937 }, { "epoch": 1.5960974569494055, "grad_norm": 5.014688968658447, "learning_rate": 4.0126e-06, "loss": 0.0162, "step": 29938 }, { "epoch": 1.5961507703790585, "grad_norm": 3.4402458667755127, "learning_rate": 4.0124e-06, "loss": 0.03, "step": 29939 }, { "epoch": 1.5962040838087113, "grad_norm": 3.9015679359436035, "learning_rate": 4.0122e-06, "loss": 0.0564, "step": 29940 }, { "epoch": 1.5962573972383645, "grad_norm": 1.692960500717163, "learning_rate": 4.012000000000001e-06, "loss": 0.0074, "step": 29941 }, { "epoch": 1.5963107106680172, "grad_norm": 3.32373046875, "learning_rate": 4.0118e-06, "loss": 0.0367, "step": 29942 }, { "epoch": 1.5963640240976702, "grad_norm": 2.3113224506378174, "learning_rate": 4.011600000000001e-06, "loss": 0.0522, "step": 29943 }, { "epoch": 1.5964173375273232, "grad_norm": 1.7523906230926514, "learning_rate": 4.0114e-06, "loss": 0.0158, "step": 29944 }, { "epoch": 1.596470650956976, "grad_norm": 1.0622673034667969, "learning_rate": 4.0112e-06, "loss": 0.0101, "step": 29945 }, { "epoch": 1.5965239643866291, "grad_norm": 1.6500710248947144, "learning_rate": 4.011e-06, "loss": 0.0359, "step": 29946 }, { "epoch": 1.596577277816282, "grad_norm": 1.6735726594924927, "learning_rate": 4.010800000000001e-06, "loss": 0.0163, "step": 29947 }, { "epoch": 1.5966305912459349, "grad_norm": 1.8376141786575317, "learning_rate": 4.0106e-06, "loss": 0.0118, "step": 29948 }, { "epoch": 1.5966839046755879, "grad_norm": 1.5259056091308594, "learning_rate": 4.010400000000001e-06, "loss": 0.0198, "step": 29949 }, { "epoch": 1.5967372181052406, "grad_norm": 1.2384082078933716, "learning_rate": 4.0102e-06, "loss": 0.0168, "step": 29950 }, { "epoch": 1.5967905315348936, "grad_norm": 4.823496341705322, "learning_rate": 4.0100000000000006e-06, "loss": 0.0215, "step": 29951 }, { "epoch": 1.5968438449645466, "grad_norm": 3.860607624053955, "learning_rate": 4.0098e-06, "loss": 0.047, "step": 29952 }, { "epoch": 1.5968971583941995, "grad_norm": 1.4813027381896973, "learning_rate": 4.0096000000000005e-06, "loss": 0.0289, "step": 29953 }, { "epoch": 1.5969504718238525, "grad_norm": 0.7355514168739319, "learning_rate": 4.0094e-06, "loss": 0.0052, "step": 29954 }, { "epoch": 1.5970037852535053, "grad_norm": 2.4247288703918457, "learning_rate": 4.0092000000000005e-06, "loss": 0.0464, "step": 29955 }, { "epoch": 1.5970570986831583, "grad_norm": 2.625594139099121, "learning_rate": 4.009e-06, "loss": 0.0373, "step": 29956 }, { "epoch": 1.5971104121128112, "grad_norm": 1.2633930444717407, "learning_rate": 4.0088000000000005e-06, "loss": 0.0186, "step": 29957 }, { "epoch": 1.597163725542464, "grad_norm": 0.8966225385665894, "learning_rate": 4.0086e-06, "loss": 0.0067, "step": 29958 }, { "epoch": 1.5972170389721172, "grad_norm": 1.9379743337631226, "learning_rate": 4.0084000000000005e-06, "loss": 0.0283, "step": 29959 }, { "epoch": 1.59727035240177, "grad_norm": 1.04655921459198, "learning_rate": 4.0082e-06, "loss": 0.009, "step": 29960 }, { "epoch": 1.597323665831423, "grad_norm": 1.2859042882919312, "learning_rate": 4.008e-06, "loss": 0.0288, "step": 29961 }, { "epoch": 1.597376979261076, "grad_norm": 2.0149078369140625, "learning_rate": 4.007800000000001e-06, "loss": 0.0334, "step": 29962 }, { "epoch": 1.5974302926907287, "grad_norm": 0.9475131034851074, "learning_rate": 4.0076e-06, "loss": 0.0057, "step": 29963 }, { "epoch": 1.5974836061203819, "grad_norm": 2.4130630493164062, "learning_rate": 4.0074e-06, "loss": 0.0175, "step": 29964 }, { "epoch": 1.5975369195500346, "grad_norm": 1.7387112379074097, "learning_rate": 4.0072e-06, "loss": 0.027, "step": 29965 }, { "epoch": 1.5975902329796876, "grad_norm": 0.7697656750679016, "learning_rate": 4.007e-06, "loss": 0.0061, "step": 29966 }, { "epoch": 1.5976435464093406, "grad_norm": 2.1047208309173584, "learning_rate": 4.0068e-06, "loss": 0.0496, "step": 29967 }, { "epoch": 1.5976968598389933, "grad_norm": 0.9340207576751709, "learning_rate": 4.006600000000001e-06, "loss": 0.0037, "step": 29968 }, { "epoch": 1.5977501732686463, "grad_norm": 1.1567281484603882, "learning_rate": 4.0064e-06, "loss": 0.0213, "step": 29969 }, { "epoch": 1.5978034866982993, "grad_norm": 3.0925092697143555, "learning_rate": 4.0062e-06, "loss": 0.0385, "step": 29970 }, { "epoch": 1.5978568001279523, "grad_norm": 1.6009138822555542, "learning_rate": 4.006e-06, "loss": 0.0186, "step": 29971 }, { "epoch": 1.5979101135576053, "grad_norm": 2.4875330924987793, "learning_rate": 4.0058e-06, "loss": 0.034, "step": 29972 }, { "epoch": 1.597963426987258, "grad_norm": 4.562503337860107, "learning_rate": 4.0056e-06, "loss": 0.0282, "step": 29973 }, { "epoch": 1.598016740416911, "grad_norm": 2.2460286617279053, "learning_rate": 4.005400000000001e-06, "loss": 0.0298, "step": 29974 }, { "epoch": 1.598070053846564, "grad_norm": 1.9164609909057617, "learning_rate": 4.0052e-06, "loss": 0.0172, "step": 29975 }, { "epoch": 1.5981233672762167, "grad_norm": 1.3335177898406982, "learning_rate": 4.005000000000001e-06, "loss": 0.0285, "step": 29976 }, { "epoch": 1.59817668070587, "grad_norm": 1.586244821548462, "learning_rate": 4.0048e-06, "loss": 0.0144, "step": 29977 }, { "epoch": 1.5982299941355227, "grad_norm": 2.6094274520874023, "learning_rate": 4.0046e-06, "loss": 0.0225, "step": 29978 }, { "epoch": 1.5982833075651757, "grad_norm": 2.448636531829834, "learning_rate": 4.0044e-06, "loss": 0.0295, "step": 29979 }, { "epoch": 1.5983366209948286, "grad_norm": 1.8398218154907227, "learning_rate": 4.0042000000000006e-06, "loss": 0.02, "step": 29980 }, { "epoch": 1.5983899344244814, "grad_norm": 1.7599091529846191, "learning_rate": 4.004e-06, "loss": 0.0103, "step": 29981 }, { "epoch": 1.5984432478541346, "grad_norm": 2.568854570388794, "learning_rate": 4.0038000000000005e-06, "loss": 0.0377, "step": 29982 }, { "epoch": 1.5984965612837874, "grad_norm": 0.6402220726013184, "learning_rate": 4.0036e-06, "loss": 0.0067, "step": 29983 }, { "epoch": 1.5985498747134403, "grad_norm": 1.670351505279541, "learning_rate": 4.0034e-06, "loss": 0.0339, "step": 29984 }, { "epoch": 1.5986031881430933, "grad_norm": 1.3627756834030151, "learning_rate": 4.0032e-06, "loss": 0.0104, "step": 29985 }, { "epoch": 1.598656501572746, "grad_norm": 3.013976812362671, "learning_rate": 4.0030000000000005e-06, "loss": 0.0125, "step": 29986 }, { "epoch": 1.598709815002399, "grad_norm": 1.0749073028564453, "learning_rate": 4.002800000000001e-06, "loss": 0.0084, "step": 29987 }, { "epoch": 1.598763128432052, "grad_norm": 2.3102035522460938, "learning_rate": 4.0026000000000004e-06, "loss": 0.0352, "step": 29988 }, { "epoch": 1.598816441861705, "grad_norm": 0.9297365546226501, "learning_rate": 4.0024e-06, "loss": 0.024, "step": 29989 }, { "epoch": 1.598869755291358, "grad_norm": 1.6384451389312744, "learning_rate": 4.0022e-06, "loss": 0.0364, "step": 29990 }, { "epoch": 1.5989230687210108, "grad_norm": 0.939516544342041, "learning_rate": 4.002e-06, "loss": 0.0059, "step": 29991 }, { "epoch": 1.5989763821506637, "grad_norm": 1.6322630643844604, "learning_rate": 4.0018e-06, "loss": 0.0087, "step": 29992 }, { "epoch": 1.5990296955803167, "grad_norm": 1.5784177780151367, "learning_rate": 4.001600000000001e-06, "loss": 0.0402, "step": 29993 }, { "epoch": 1.5990830090099695, "grad_norm": 0.9822506904602051, "learning_rate": 4.0014e-06, "loss": 0.0083, "step": 29994 }, { "epoch": 1.5991363224396227, "grad_norm": 1.8684412240982056, "learning_rate": 4.0012e-06, "loss": 0.0208, "step": 29995 }, { "epoch": 1.5991896358692754, "grad_norm": 0.49796316027641296, "learning_rate": 4.001e-06, "loss": 0.0038, "step": 29996 }, { "epoch": 1.5992429492989284, "grad_norm": 2.5642335414886475, "learning_rate": 4.0008e-06, "loss": 0.0205, "step": 29997 }, { "epoch": 1.5992962627285814, "grad_norm": 0.18039405345916748, "learning_rate": 4.0006e-06, "loss": 0.0019, "step": 29998 }, { "epoch": 1.5993495761582341, "grad_norm": 2.1060309410095215, "learning_rate": 4.000400000000001e-06, "loss": 0.0281, "step": 29999 }, { "epoch": 1.5994028895878873, "grad_norm": 1.0597838163375854, "learning_rate": 4.0002e-06, "loss": 0.0064, "step": 30000 }, { "epoch": 1.59945620301754, "grad_norm": 2.281851053237915, "learning_rate": 4.000000000000001e-06, "loss": 0.0299, "step": 30001 }, { "epoch": 1.599509516447193, "grad_norm": 3.8021998405456543, "learning_rate": 3.9998e-06, "loss": 0.0342, "step": 30002 }, { "epoch": 1.599562829876846, "grad_norm": 0.8262332677841187, "learning_rate": 3.9996e-06, "loss": 0.0053, "step": 30003 }, { "epoch": 1.5996161433064988, "grad_norm": 2.956691026687622, "learning_rate": 3.9994e-06, "loss": 0.0391, "step": 30004 }, { "epoch": 1.5996694567361518, "grad_norm": 1.0390846729278564, "learning_rate": 3.999200000000001e-06, "loss": 0.0158, "step": 30005 }, { "epoch": 1.5997227701658048, "grad_norm": 2.106962203979492, "learning_rate": 3.999e-06, "loss": 0.0255, "step": 30006 }, { "epoch": 1.5997760835954575, "grad_norm": 1.8450963497161865, "learning_rate": 3.9988000000000006e-06, "loss": 0.0378, "step": 30007 }, { "epoch": 1.5998293970251107, "grad_norm": 1.7763274908065796, "learning_rate": 3.9986e-06, "loss": 0.0183, "step": 30008 }, { "epoch": 1.5998827104547635, "grad_norm": 0.4970032274723053, "learning_rate": 3.9984e-06, "loss": 0.005, "step": 30009 }, { "epoch": 1.5999360238844165, "grad_norm": 1.226523756980896, "learning_rate": 3.9982e-06, "loss": 0.0245, "step": 30010 }, { "epoch": 1.5999893373140694, "grad_norm": 0.9942443370819092, "learning_rate": 3.9980000000000005e-06, "loss": 0.0059, "step": 30011 }, { "epoch": 1.6000426507437222, "grad_norm": 1.776383638381958, "learning_rate": 3.997800000000001e-06, "loss": 0.0252, "step": 30012 }, { "epoch": 1.6000959641733754, "grad_norm": 1.5018396377563477, "learning_rate": 3.9976000000000005e-06, "loss": 0.0115, "step": 30013 }, { "epoch": 1.6001492776030282, "grad_norm": 2.827383041381836, "learning_rate": 3.9974e-06, "loss": 0.0247, "step": 30014 }, { "epoch": 1.6002025910326811, "grad_norm": 1.9299955368041992, "learning_rate": 3.9972000000000005e-06, "loss": 0.0219, "step": 30015 }, { "epoch": 1.6002559044623341, "grad_norm": 1.8874744176864624, "learning_rate": 3.997e-06, "loss": 0.0223, "step": 30016 }, { "epoch": 1.6003092178919869, "grad_norm": 2.565781593322754, "learning_rate": 3.9968000000000004e-06, "loss": 0.0236, "step": 30017 }, { "epoch": 1.60036253132164, "grad_norm": 1.8665541410446167, "learning_rate": 3.996600000000001e-06, "loss": 0.0326, "step": 30018 }, { "epoch": 1.6004158447512928, "grad_norm": 1.263763666152954, "learning_rate": 3.9964e-06, "loss": 0.0096, "step": 30019 }, { "epoch": 1.6004691581809458, "grad_norm": 1.7285538911819458, "learning_rate": 3.9962e-06, "loss": 0.0138, "step": 30020 }, { "epoch": 1.6005224716105988, "grad_norm": 3.4392881393432617, "learning_rate": 3.996e-06, "loss": 0.0372, "step": 30021 }, { "epoch": 1.6005757850402516, "grad_norm": 1.9714936017990112, "learning_rate": 3.9958e-06, "loss": 0.0126, "step": 30022 }, { "epoch": 1.6006290984699045, "grad_norm": 1.5238196849822998, "learning_rate": 3.9956e-06, "loss": 0.0123, "step": 30023 }, { "epoch": 1.6006824118995575, "grad_norm": 2.1310675144195557, "learning_rate": 3.995400000000001e-06, "loss": 0.0655, "step": 30024 }, { "epoch": 1.6007357253292103, "grad_norm": 2.815194845199585, "learning_rate": 3.9952e-06, "loss": 0.0274, "step": 30025 }, { "epoch": 1.6007890387588635, "grad_norm": 0.9777740240097046, "learning_rate": 3.995000000000001e-06, "loss": 0.0094, "step": 30026 }, { "epoch": 1.6008423521885162, "grad_norm": 2.7678580284118652, "learning_rate": 3.9948e-06, "loss": 0.0276, "step": 30027 }, { "epoch": 1.6008956656181692, "grad_norm": 3.2500619888305664, "learning_rate": 3.9946e-06, "loss": 0.0147, "step": 30028 }, { "epoch": 1.6009489790478222, "grad_norm": 1.9547114372253418, "learning_rate": 3.9944e-06, "loss": 0.0187, "step": 30029 }, { "epoch": 1.601002292477475, "grad_norm": 1.6395496129989624, "learning_rate": 3.994200000000001e-06, "loss": 0.0219, "step": 30030 }, { "epoch": 1.6010556059071281, "grad_norm": 2.925287961959839, "learning_rate": 3.994e-06, "loss": 0.0339, "step": 30031 }, { "epoch": 1.601108919336781, "grad_norm": 1.5250672101974487, "learning_rate": 3.993800000000001e-06, "loss": 0.0264, "step": 30032 }, { "epoch": 1.6011622327664339, "grad_norm": 1.0220056772232056, "learning_rate": 3.9936e-06, "loss": 0.0097, "step": 30033 }, { "epoch": 1.6012155461960869, "grad_norm": 0.6726827621459961, "learning_rate": 3.9934e-06, "loss": 0.0059, "step": 30034 }, { "epoch": 1.6012688596257396, "grad_norm": 2.2434396743774414, "learning_rate": 3.9932e-06, "loss": 0.0279, "step": 30035 }, { "epoch": 1.6013221730553928, "grad_norm": 3.9513370990753174, "learning_rate": 3.9930000000000006e-06, "loss": 0.073, "step": 30036 }, { "epoch": 1.6013754864850456, "grad_norm": 1.3768237829208374, "learning_rate": 3.992800000000001e-06, "loss": 0.0134, "step": 30037 }, { "epoch": 1.6014287999146986, "grad_norm": 2.5443339347839355, "learning_rate": 3.9926000000000005e-06, "loss": 0.0406, "step": 30038 }, { "epoch": 1.6014821133443515, "grad_norm": 1.6117719411849976, "learning_rate": 3.9924e-06, "loss": 0.0117, "step": 30039 }, { "epoch": 1.6015354267740043, "grad_norm": 1.556749939918518, "learning_rate": 3.9922000000000005e-06, "loss": 0.0201, "step": 30040 }, { "epoch": 1.6015887402036573, "grad_norm": 2.4377787113189697, "learning_rate": 3.992e-06, "loss": 0.0382, "step": 30041 }, { "epoch": 1.6016420536333102, "grad_norm": 0.43628841638565063, "learning_rate": 3.9918000000000005e-06, "loss": 0.0037, "step": 30042 }, { "epoch": 1.601695367062963, "grad_norm": 3.283674478530884, "learning_rate": 3.991600000000001e-06, "loss": 0.0232, "step": 30043 }, { "epoch": 1.6017486804926162, "grad_norm": 1.94212007522583, "learning_rate": 3.9914000000000004e-06, "loss": 0.0227, "step": 30044 }, { "epoch": 1.601801993922269, "grad_norm": 2.129789113998413, "learning_rate": 3.9912e-06, "loss": 0.021, "step": 30045 }, { "epoch": 1.601855307351922, "grad_norm": 3.2414004802703857, "learning_rate": 3.991e-06, "loss": 0.0453, "step": 30046 }, { "epoch": 1.601908620781575, "grad_norm": 2.1008129119873047, "learning_rate": 3.9908e-06, "loss": 0.0139, "step": 30047 }, { "epoch": 1.6019619342112277, "grad_norm": 3.0483806133270264, "learning_rate": 3.9906e-06, "loss": 0.0291, "step": 30048 }, { "epoch": 1.6020152476408809, "grad_norm": 1.6710593700408936, "learning_rate": 3.990400000000001e-06, "loss": 0.0195, "step": 30049 }, { "epoch": 1.6020685610705336, "grad_norm": 1.0388084650039673, "learning_rate": 3.9902e-06, "loss": 0.0289, "step": 30050 }, { "epoch": 1.6021218745001866, "grad_norm": 1.696746587753296, "learning_rate": 3.990000000000001e-06, "loss": 0.0105, "step": 30051 }, { "epoch": 1.6021751879298396, "grad_norm": 2.5579779148101807, "learning_rate": 3.9898e-06, "loss": 0.0237, "step": 30052 }, { "epoch": 1.6022285013594924, "grad_norm": 2.1574747562408447, "learning_rate": 3.9896e-06, "loss": 0.0255, "step": 30053 }, { "epoch": 1.6022818147891453, "grad_norm": 1.2222086191177368, "learning_rate": 3.9894e-06, "loss": 0.0092, "step": 30054 }, { "epoch": 1.6023351282187983, "grad_norm": 2.0266404151916504, "learning_rate": 3.989200000000001e-06, "loss": 0.0322, "step": 30055 }, { "epoch": 1.6023884416484513, "grad_norm": 2.580265522003174, "learning_rate": 3.989e-06, "loss": 0.0412, "step": 30056 }, { "epoch": 1.6024417550781043, "grad_norm": 2.287203550338745, "learning_rate": 3.988800000000001e-06, "loss": 0.0127, "step": 30057 }, { "epoch": 1.602495068507757, "grad_norm": 3.1499502658843994, "learning_rate": 3.9886e-06, "loss": 0.0337, "step": 30058 }, { "epoch": 1.60254838193741, "grad_norm": 2.27858829498291, "learning_rate": 3.9884e-06, "loss": 0.016, "step": 30059 }, { "epoch": 1.602601695367063, "grad_norm": 1.599523663520813, "learning_rate": 3.9882e-06, "loss": 0.0249, "step": 30060 }, { "epoch": 1.6026550087967157, "grad_norm": 1.059396505355835, "learning_rate": 3.988000000000001e-06, "loss": 0.0061, "step": 30061 }, { "epoch": 1.602708322226369, "grad_norm": 0.9515233039855957, "learning_rate": 3.9878e-06, "loss": 0.0096, "step": 30062 }, { "epoch": 1.6027616356560217, "grad_norm": 0.9806729555130005, "learning_rate": 3.987600000000001e-06, "loss": 0.0063, "step": 30063 }, { "epoch": 1.6028149490856747, "grad_norm": 1.1960208415985107, "learning_rate": 3.9874e-06, "loss": 0.0243, "step": 30064 }, { "epoch": 1.6028682625153277, "grad_norm": 1.796478271484375, "learning_rate": 3.9872000000000006e-06, "loss": 0.0083, "step": 30065 }, { "epoch": 1.6029215759449804, "grad_norm": 1.6641408205032349, "learning_rate": 3.987e-06, "loss": 0.0174, "step": 30066 }, { "epoch": 1.6029748893746336, "grad_norm": 2.1268081665039062, "learning_rate": 3.9868000000000005e-06, "loss": 0.0423, "step": 30067 }, { "epoch": 1.6030282028042864, "grad_norm": 2.039025068283081, "learning_rate": 3.9866e-06, "loss": 0.0248, "step": 30068 }, { "epoch": 1.6030815162339394, "grad_norm": 2.2637109756469727, "learning_rate": 3.9864000000000005e-06, "loss": 0.0253, "step": 30069 }, { "epoch": 1.6031348296635923, "grad_norm": 1.9306608438491821, "learning_rate": 3.9862e-06, "loss": 0.0296, "step": 30070 }, { "epoch": 1.603188143093245, "grad_norm": 2.2258949279785156, "learning_rate": 3.9860000000000005e-06, "loss": 0.0317, "step": 30071 }, { "epoch": 1.603241456522898, "grad_norm": 2.971646547317505, "learning_rate": 3.9858e-06, "loss": 0.0151, "step": 30072 }, { "epoch": 1.603294769952551, "grad_norm": 1.319791555404663, "learning_rate": 3.9856000000000004e-06, "loss": 0.015, "step": 30073 }, { "epoch": 1.603348083382204, "grad_norm": 1.2221815586090088, "learning_rate": 3.9854e-06, "loss": 0.0228, "step": 30074 }, { "epoch": 1.603401396811857, "grad_norm": 1.4000730514526367, "learning_rate": 3.9852e-06, "loss": 0.0167, "step": 30075 }, { "epoch": 1.6034547102415098, "grad_norm": 1.5983998775482178, "learning_rate": 3.985000000000001e-06, "loss": 0.0204, "step": 30076 }, { "epoch": 1.6035080236711627, "grad_norm": 2.2282357215881348, "learning_rate": 3.9848e-06, "loss": 0.0146, "step": 30077 }, { "epoch": 1.6035613371008157, "grad_norm": 2.1510202884674072, "learning_rate": 3.9846e-06, "loss": 0.0138, "step": 30078 }, { "epoch": 1.6036146505304685, "grad_norm": 6.297044277191162, "learning_rate": 3.9844e-06, "loss": 0.0398, "step": 30079 }, { "epoch": 1.6036679639601217, "grad_norm": 3.53558087348938, "learning_rate": 3.9842e-06, "loss": 0.0254, "step": 30080 }, { "epoch": 1.6037212773897744, "grad_norm": 2.5960493087768555, "learning_rate": 3.984e-06, "loss": 0.0272, "step": 30081 }, { "epoch": 1.6037745908194274, "grad_norm": 2.006218910217285, "learning_rate": 3.983800000000001e-06, "loss": 0.0081, "step": 30082 }, { "epoch": 1.6038279042490804, "grad_norm": 1.2247767448425293, "learning_rate": 3.9836e-06, "loss": 0.0104, "step": 30083 }, { "epoch": 1.6038812176787332, "grad_norm": 2.0575621128082275, "learning_rate": 3.983400000000001e-06, "loss": 0.0218, "step": 30084 }, { "epoch": 1.6039345311083864, "grad_norm": 1.6879236698150635, "learning_rate": 3.9832e-06, "loss": 0.0187, "step": 30085 }, { "epoch": 1.603987844538039, "grad_norm": 1.1810672283172607, "learning_rate": 3.983e-06, "loss": 0.0083, "step": 30086 }, { "epoch": 1.604041157967692, "grad_norm": 0.535922110080719, "learning_rate": 3.9828e-06, "loss": 0.0038, "step": 30087 }, { "epoch": 1.604094471397345, "grad_norm": 1.0104737281799316, "learning_rate": 3.982600000000001e-06, "loss": 0.0116, "step": 30088 }, { "epoch": 1.6041477848269978, "grad_norm": 1.2077958583831787, "learning_rate": 3.9824e-06, "loss": 0.0095, "step": 30089 }, { "epoch": 1.6042010982566508, "grad_norm": 1.2740352153778076, "learning_rate": 3.982200000000001e-06, "loss": 0.0088, "step": 30090 }, { "epoch": 1.6042544116863038, "grad_norm": 1.268513798713684, "learning_rate": 3.982e-06, "loss": 0.025, "step": 30091 }, { "epoch": 1.6043077251159568, "grad_norm": 1.8704702854156494, "learning_rate": 3.9818e-06, "loss": 0.0137, "step": 30092 }, { "epoch": 1.6043610385456097, "grad_norm": 1.2168327569961548, "learning_rate": 3.9816e-06, "loss": 0.0075, "step": 30093 }, { "epoch": 1.6044143519752625, "grad_norm": 1.5662906169891357, "learning_rate": 3.9814000000000005e-06, "loss": 0.0102, "step": 30094 }, { "epoch": 1.6044676654049155, "grad_norm": 1.1483778953552246, "learning_rate": 3.9812e-06, "loss": 0.0093, "step": 30095 }, { "epoch": 1.6045209788345685, "grad_norm": 1.2433024644851685, "learning_rate": 3.9810000000000005e-06, "loss": 0.025, "step": 30096 }, { "epoch": 1.6045742922642212, "grad_norm": 1.6588860750198364, "learning_rate": 3.9808e-06, "loss": 0.0256, "step": 30097 }, { "epoch": 1.6046276056938744, "grad_norm": 1.5525116920471191, "learning_rate": 3.9806000000000005e-06, "loss": 0.0181, "step": 30098 }, { "epoch": 1.6046809191235272, "grad_norm": 1.4429078102111816, "learning_rate": 3.9804e-06, "loss": 0.0099, "step": 30099 }, { "epoch": 1.6047342325531802, "grad_norm": 2.1857266426086426, "learning_rate": 3.9802000000000004e-06, "loss": 0.0253, "step": 30100 }, { "epoch": 1.6047875459828331, "grad_norm": 1.7701419591903687, "learning_rate": 3.980000000000001e-06, "loss": 0.017, "step": 30101 }, { "epoch": 1.6048408594124859, "grad_norm": 0.601865828037262, "learning_rate": 3.9798e-06, "loss": 0.0039, "step": 30102 }, { "epoch": 1.604894172842139, "grad_norm": 3.706498861312866, "learning_rate": 3.9796e-06, "loss": 0.0509, "step": 30103 }, { "epoch": 1.6049474862717918, "grad_norm": 2.926039695739746, "learning_rate": 3.9794e-06, "loss": 0.0225, "step": 30104 }, { "epoch": 1.6050007997014448, "grad_norm": 3.1208550930023193, "learning_rate": 3.9792e-06, "loss": 0.0341, "step": 30105 }, { "epoch": 1.6050541131310978, "grad_norm": 3.4811673164367676, "learning_rate": 3.979e-06, "loss": 0.0279, "step": 30106 }, { "epoch": 1.6051074265607506, "grad_norm": 2.5743205547332764, "learning_rate": 3.978800000000001e-06, "loss": 0.0174, "step": 30107 }, { "epoch": 1.6051607399904035, "grad_norm": 2.2278668880462646, "learning_rate": 3.9786e-06, "loss": 0.0148, "step": 30108 }, { "epoch": 1.6052140534200565, "grad_norm": 1.6070019006729126, "learning_rate": 3.978400000000001e-06, "loss": 0.0111, "step": 30109 }, { "epoch": 1.6052673668497093, "grad_norm": 1.5614250898361206, "learning_rate": 3.9782e-06, "loss": 0.0116, "step": 30110 }, { "epoch": 1.6053206802793625, "grad_norm": 3.2252464294433594, "learning_rate": 3.978e-06, "loss": 0.0204, "step": 30111 }, { "epoch": 1.6053739937090152, "grad_norm": 1.6018269062042236, "learning_rate": 3.9778e-06, "loss": 0.0293, "step": 30112 }, { "epoch": 1.6054273071386682, "grad_norm": 1.6157574653625488, "learning_rate": 3.977600000000001e-06, "loss": 0.0261, "step": 30113 }, { "epoch": 1.6054806205683212, "grad_norm": 2.608412504196167, "learning_rate": 3.9774e-06, "loss": 0.0204, "step": 30114 }, { "epoch": 1.605533933997974, "grad_norm": 1.9740451574325562, "learning_rate": 3.977200000000001e-06, "loss": 0.0244, "step": 30115 }, { "epoch": 1.6055872474276272, "grad_norm": 1.6801916360855103, "learning_rate": 3.977e-06, "loss": 0.0282, "step": 30116 }, { "epoch": 1.60564056085728, "grad_norm": 2.099327802658081, "learning_rate": 3.9768e-06, "loss": 0.0402, "step": 30117 }, { "epoch": 1.6056938742869329, "grad_norm": 0.4791824221611023, "learning_rate": 3.9766e-06, "loss": 0.0039, "step": 30118 }, { "epoch": 1.6057471877165859, "grad_norm": 1.2548253536224365, "learning_rate": 3.976400000000001e-06, "loss": 0.0092, "step": 30119 }, { "epoch": 1.6058005011462386, "grad_norm": 2.456453800201416, "learning_rate": 3.9762e-06, "loss": 0.0271, "step": 30120 }, { "epoch": 1.6058538145758918, "grad_norm": 2.0994040966033936, "learning_rate": 3.9760000000000006e-06, "loss": 0.0255, "step": 30121 }, { "epoch": 1.6059071280055446, "grad_norm": 1.5535327196121216, "learning_rate": 3.9758e-06, "loss": 0.0058, "step": 30122 }, { "epoch": 1.6059604414351976, "grad_norm": 1.409280776977539, "learning_rate": 3.9756000000000005e-06, "loss": 0.0178, "step": 30123 }, { "epoch": 1.6060137548648505, "grad_norm": 1.7281297445297241, "learning_rate": 3.9754e-06, "loss": 0.0195, "step": 30124 }, { "epoch": 1.6060670682945033, "grad_norm": 1.9828402996063232, "learning_rate": 3.9752000000000005e-06, "loss": 0.0122, "step": 30125 }, { "epoch": 1.6061203817241563, "grad_norm": 2.5969066619873047, "learning_rate": 3.975000000000001e-06, "loss": 0.023, "step": 30126 }, { "epoch": 1.6061736951538093, "grad_norm": 1.2617342472076416, "learning_rate": 3.9748000000000005e-06, "loss": 0.0092, "step": 30127 }, { "epoch": 1.606227008583462, "grad_norm": 1.3454933166503906, "learning_rate": 3.9746e-06, "loss": 0.0104, "step": 30128 }, { "epoch": 1.6062803220131152, "grad_norm": 2.036468029022217, "learning_rate": 3.9744000000000004e-06, "loss": 0.0279, "step": 30129 }, { "epoch": 1.606333635442768, "grad_norm": 1.916813611984253, "learning_rate": 3.9742e-06, "loss": 0.0079, "step": 30130 }, { "epoch": 1.606386948872421, "grad_norm": 2.010089635848999, "learning_rate": 3.974e-06, "loss": 0.0122, "step": 30131 }, { "epoch": 1.606440262302074, "grad_norm": 1.9884588718414307, "learning_rate": 3.973800000000001e-06, "loss": 0.027, "step": 30132 }, { "epoch": 1.6064935757317267, "grad_norm": 2.352476119995117, "learning_rate": 3.9736e-06, "loss": 0.0108, "step": 30133 }, { "epoch": 1.6065468891613799, "grad_norm": 1.6081538200378418, "learning_rate": 3.973400000000001e-06, "loss": 0.0184, "step": 30134 }, { "epoch": 1.6066002025910326, "grad_norm": 2.989168643951416, "learning_rate": 3.9732e-06, "loss": 0.0118, "step": 30135 }, { "epoch": 1.6066535160206856, "grad_norm": 1.3276987075805664, "learning_rate": 3.973e-06, "loss": 0.0084, "step": 30136 }, { "epoch": 1.6067068294503386, "grad_norm": 2.0100185871124268, "learning_rate": 3.9728e-06, "loss": 0.0152, "step": 30137 }, { "epoch": 1.6067601428799914, "grad_norm": 4.761928558349609, "learning_rate": 3.972600000000001e-06, "loss": 0.0173, "step": 30138 }, { "epoch": 1.6068134563096446, "grad_norm": 2.1808359622955322, "learning_rate": 3.9724e-06, "loss": 0.0215, "step": 30139 }, { "epoch": 1.6068667697392973, "grad_norm": 2.1104352474212646, "learning_rate": 3.972200000000001e-06, "loss": 0.0121, "step": 30140 }, { "epoch": 1.6069200831689503, "grad_norm": 3.103914976119995, "learning_rate": 3.972e-06, "loss": 0.0367, "step": 30141 }, { "epoch": 1.6069733965986033, "grad_norm": 0.36548659205436707, "learning_rate": 3.9718e-06, "loss": 0.004, "step": 30142 }, { "epoch": 1.607026710028256, "grad_norm": 1.9905059337615967, "learning_rate": 3.9716e-06, "loss": 0.02, "step": 30143 }, { "epoch": 1.607080023457909, "grad_norm": 3.6474297046661377, "learning_rate": 3.971400000000001e-06, "loss": 0.0226, "step": 30144 }, { "epoch": 1.607133336887562, "grad_norm": 1.9872664213180542, "learning_rate": 3.9712e-06, "loss": 0.0234, "step": 30145 }, { "epoch": 1.6071866503172147, "grad_norm": 4.3842034339904785, "learning_rate": 3.971000000000001e-06, "loss": 0.0118, "step": 30146 }, { "epoch": 1.607239963746868, "grad_norm": 2.1059510707855225, "learning_rate": 3.9708e-06, "loss": 0.0367, "step": 30147 }, { "epoch": 1.6072932771765207, "grad_norm": 1.936069369316101, "learning_rate": 3.9706000000000006e-06, "loss": 0.0212, "step": 30148 }, { "epoch": 1.6073465906061737, "grad_norm": 1.9847739934921265, "learning_rate": 3.9704e-06, "loss": 0.023, "step": 30149 }, { "epoch": 1.6073999040358267, "grad_norm": 1.7320889234542847, "learning_rate": 3.9702000000000005e-06, "loss": 0.0296, "step": 30150 }, { "epoch": 1.6074532174654794, "grad_norm": 1.4576702117919922, "learning_rate": 3.97e-06, "loss": 0.0241, "step": 30151 }, { "epoch": 1.6075065308951326, "grad_norm": 1.1958311796188354, "learning_rate": 3.9698000000000005e-06, "loss": 0.0107, "step": 30152 }, { "epoch": 1.6075598443247854, "grad_norm": 1.592726707458496, "learning_rate": 3.9696e-06, "loss": 0.0122, "step": 30153 }, { "epoch": 1.6076131577544384, "grad_norm": 2.0697782039642334, "learning_rate": 3.9694000000000005e-06, "loss": 0.0363, "step": 30154 }, { "epoch": 1.6076664711840913, "grad_norm": 2.084322452545166, "learning_rate": 3.9692e-06, "loss": 0.0477, "step": 30155 }, { "epoch": 1.607719784613744, "grad_norm": 0.4119987189769745, "learning_rate": 3.9690000000000005e-06, "loss": 0.0033, "step": 30156 }, { "epoch": 1.6077730980433973, "grad_norm": 2.379807710647583, "learning_rate": 3.9688e-06, "loss": 0.0229, "step": 30157 }, { "epoch": 1.60782641147305, "grad_norm": 1.4780956506729126, "learning_rate": 3.9686000000000004e-06, "loss": 0.0136, "step": 30158 }, { "epoch": 1.607879724902703, "grad_norm": 1.8526489734649658, "learning_rate": 3.968400000000001e-06, "loss": 0.0297, "step": 30159 }, { "epoch": 1.607933038332356, "grad_norm": 3.054483413696289, "learning_rate": 3.9682e-06, "loss": 0.0171, "step": 30160 }, { "epoch": 1.6079863517620088, "grad_norm": 2.4204587936401367, "learning_rate": 3.968e-06, "loss": 0.0193, "step": 30161 }, { "epoch": 1.6080396651916617, "grad_norm": 3.247358560562134, "learning_rate": 3.9678e-06, "loss": 0.0165, "step": 30162 }, { "epoch": 1.6080929786213147, "grad_norm": 1.1748135089874268, "learning_rate": 3.9676e-06, "loss": 0.0146, "step": 30163 }, { "epoch": 1.6081462920509675, "grad_norm": 0.9767690300941467, "learning_rate": 3.9674e-06, "loss": 0.0067, "step": 30164 }, { "epoch": 1.6081996054806207, "grad_norm": 1.064293622970581, "learning_rate": 3.967200000000001e-06, "loss": 0.006, "step": 30165 }, { "epoch": 1.6082529189102734, "grad_norm": 1.1896528005599976, "learning_rate": 3.967e-06, "loss": 0.0146, "step": 30166 }, { "epoch": 1.6083062323399264, "grad_norm": 1.904099941253662, "learning_rate": 3.9668e-06, "loss": 0.0262, "step": 30167 }, { "epoch": 1.6083595457695794, "grad_norm": 1.8257689476013184, "learning_rate": 3.9666e-06, "loss": 0.0241, "step": 30168 }, { "epoch": 1.6084128591992322, "grad_norm": 0.4694141745567322, "learning_rate": 3.9664e-06, "loss": 0.0051, "step": 30169 }, { "epoch": 1.6084661726288854, "grad_norm": 1.4893403053283691, "learning_rate": 3.9662e-06, "loss": 0.008, "step": 30170 }, { "epoch": 1.6085194860585381, "grad_norm": 1.7159123420715332, "learning_rate": 3.966000000000001e-06, "loss": 0.0181, "step": 30171 }, { "epoch": 1.608572799488191, "grad_norm": 2.325855255126953, "learning_rate": 3.9658e-06, "loss": 0.0415, "step": 30172 }, { "epoch": 1.608626112917844, "grad_norm": 1.6881853342056274, "learning_rate": 3.965600000000001e-06, "loss": 0.0101, "step": 30173 }, { "epoch": 1.6086794263474968, "grad_norm": 2.256588935852051, "learning_rate": 3.9654e-06, "loss": 0.0212, "step": 30174 }, { "epoch": 1.6087327397771498, "grad_norm": 3.280250072479248, "learning_rate": 3.9652e-06, "loss": 0.0536, "step": 30175 }, { "epoch": 1.6087860532068028, "grad_norm": 1.7238861322402954, "learning_rate": 3.965e-06, "loss": 0.029, "step": 30176 }, { "epoch": 1.6088393666364558, "grad_norm": 2.016993284225464, "learning_rate": 3.9648000000000006e-06, "loss": 0.016, "step": 30177 }, { "epoch": 1.6088926800661087, "grad_norm": 1.3476166725158691, "learning_rate": 3.9646e-06, "loss": 0.0249, "step": 30178 }, { "epoch": 1.6089459934957615, "grad_norm": 0.966170608997345, "learning_rate": 3.9644000000000005e-06, "loss": 0.0049, "step": 30179 }, { "epoch": 1.6089993069254145, "grad_norm": 0.9749820828437805, "learning_rate": 3.9642e-06, "loss": 0.0052, "step": 30180 }, { "epoch": 1.6090526203550675, "grad_norm": 1.4881681203842163, "learning_rate": 3.964e-06, "loss": 0.0084, "step": 30181 }, { "epoch": 1.6091059337847202, "grad_norm": 2.3093910217285156, "learning_rate": 3.9638e-06, "loss": 0.019, "step": 30182 }, { "epoch": 1.6091592472143734, "grad_norm": 2.413551092147827, "learning_rate": 3.9636000000000005e-06, "loss": 0.0433, "step": 30183 }, { "epoch": 1.6092125606440262, "grad_norm": 2.5666518211364746, "learning_rate": 3.963400000000001e-06, "loss": 0.0265, "step": 30184 }, { "epoch": 1.6092658740736792, "grad_norm": 1.5605412721633911, "learning_rate": 3.9632000000000004e-06, "loss": 0.0125, "step": 30185 }, { "epoch": 1.6093191875033321, "grad_norm": 2.0680348873138428, "learning_rate": 3.963e-06, "loss": 0.0247, "step": 30186 }, { "epoch": 1.609372500932985, "grad_norm": 4.674697399139404, "learning_rate": 3.9628e-06, "loss": 0.0509, "step": 30187 }, { "epoch": 1.609425814362638, "grad_norm": 3.33244252204895, "learning_rate": 3.9626e-06, "loss": 0.0312, "step": 30188 }, { "epoch": 1.6094791277922909, "grad_norm": 0.8390137553215027, "learning_rate": 3.9624e-06, "loss": 0.0057, "step": 30189 }, { "epoch": 1.6095324412219438, "grad_norm": 3.156446933746338, "learning_rate": 3.962200000000001e-06, "loss": 0.0268, "step": 30190 }, { "epoch": 1.6095857546515968, "grad_norm": 2.8102991580963135, "learning_rate": 3.962e-06, "loss": 0.0144, "step": 30191 }, { "epoch": 1.6096390680812496, "grad_norm": 2.6517081260681152, "learning_rate": 3.9618e-06, "loss": 0.0488, "step": 30192 }, { "epoch": 1.6096923815109025, "grad_norm": 2.3955280780792236, "learning_rate": 3.9616e-06, "loss": 0.0238, "step": 30193 }, { "epoch": 1.6097456949405555, "grad_norm": 1.017683982849121, "learning_rate": 3.9614e-06, "loss": 0.0062, "step": 30194 }, { "epoch": 1.6097990083702085, "grad_norm": 2.9371564388275146, "learning_rate": 3.9612e-06, "loss": 0.0543, "step": 30195 }, { "epoch": 1.6098523217998615, "grad_norm": 1.3430991172790527, "learning_rate": 3.961000000000001e-06, "loss": 0.0193, "step": 30196 }, { "epoch": 1.6099056352295142, "grad_norm": 1.8588114976882935, "learning_rate": 3.9608e-06, "loss": 0.0156, "step": 30197 }, { "epoch": 1.6099589486591672, "grad_norm": 0.7354118824005127, "learning_rate": 3.960600000000001e-06, "loss": 0.0047, "step": 30198 }, { "epoch": 1.6100122620888202, "grad_norm": 3.5192580223083496, "learning_rate": 3.9604e-06, "loss": 0.0486, "step": 30199 }, { "epoch": 1.610065575518473, "grad_norm": 2.121042490005493, "learning_rate": 3.9602e-06, "loss": 0.0366, "step": 30200 }, { "epoch": 1.6101188889481262, "grad_norm": 2.2306430339813232, "learning_rate": 3.96e-06, "loss": 0.0247, "step": 30201 }, { "epoch": 1.610172202377779, "grad_norm": 1.9477331638336182, "learning_rate": 3.959800000000001e-06, "loss": 0.0383, "step": 30202 }, { "epoch": 1.610225515807432, "grad_norm": 1.3280688524246216, "learning_rate": 3.9596e-06, "loss": 0.0225, "step": 30203 }, { "epoch": 1.6102788292370849, "grad_norm": 1.719424843788147, "learning_rate": 3.959400000000001e-06, "loss": 0.0183, "step": 30204 }, { "epoch": 1.6103321426667376, "grad_norm": 1.1270509958267212, "learning_rate": 3.9592e-06, "loss": 0.0207, "step": 30205 }, { "epoch": 1.6103854560963908, "grad_norm": 3.2794837951660156, "learning_rate": 3.959e-06, "loss": 0.0639, "step": 30206 }, { "epoch": 1.6104387695260436, "grad_norm": 3.9714128971099854, "learning_rate": 3.9588e-06, "loss": 0.0163, "step": 30207 }, { "epoch": 1.6104920829556966, "grad_norm": 0.8388650417327881, "learning_rate": 3.9586000000000005e-06, "loss": 0.0078, "step": 30208 }, { "epoch": 1.6105453963853495, "grad_norm": 1.9546996355056763, "learning_rate": 3.958400000000001e-06, "loss": 0.0142, "step": 30209 }, { "epoch": 1.6105987098150023, "grad_norm": 3.30947208404541, "learning_rate": 3.9582000000000005e-06, "loss": 0.044, "step": 30210 }, { "epoch": 1.6106520232446553, "grad_norm": 2.2492194175720215, "learning_rate": 3.958e-06, "loss": 0.0247, "step": 30211 }, { "epoch": 1.6107053366743083, "grad_norm": 1.6064932346343994, "learning_rate": 3.9578000000000005e-06, "loss": 0.0126, "step": 30212 }, { "epoch": 1.6107586501039612, "grad_norm": 3.957188367843628, "learning_rate": 3.9576e-06, "loss": 0.0149, "step": 30213 }, { "epoch": 1.6108119635336142, "grad_norm": 2.149670124053955, "learning_rate": 3.9574000000000004e-06, "loss": 0.0278, "step": 30214 }, { "epoch": 1.610865276963267, "grad_norm": 2.7868330478668213, "learning_rate": 3.957200000000001e-06, "loss": 0.0191, "step": 30215 }, { "epoch": 1.61091859039292, "grad_norm": 1.916698694229126, "learning_rate": 3.957e-06, "loss": 0.0232, "step": 30216 }, { "epoch": 1.610971903822573, "grad_norm": 2.0278384685516357, "learning_rate": 3.9568e-06, "loss": 0.0291, "step": 30217 }, { "epoch": 1.6110252172522257, "grad_norm": 1.0936297178268433, "learning_rate": 3.9566e-06, "loss": 0.0138, "step": 30218 }, { "epoch": 1.611078530681879, "grad_norm": 2.1496975421905518, "learning_rate": 3.9564e-06, "loss": 0.0373, "step": 30219 }, { "epoch": 1.6111318441115317, "grad_norm": 1.783806324005127, "learning_rate": 3.9562e-06, "loss": 0.0212, "step": 30220 }, { "epoch": 1.6111851575411846, "grad_norm": 3.196152687072754, "learning_rate": 3.956000000000001e-06, "loss": 0.05, "step": 30221 }, { "epoch": 1.6112384709708376, "grad_norm": 3.114210605621338, "learning_rate": 3.9558e-06, "loss": 0.0148, "step": 30222 }, { "epoch": 1.6112917844004904, "grad_norm": 2.1970150470733643, "learning_rate": 3.955600000000001e-06, "loss": 0.0366, "step": 30223 }, { "epoch": 1.6113450978301436, "grad_norm": 2.664377212524414, "learning_rate": 3.9554e-06, "loss": 0.0418, "step": 30224 }, { "epoch": 1.6113984112597963, "grad_norm": 2.5007922649383545, "learning_rate": 3.9552e-06, "loss": 0.0303, "step": 30225 }, { "epoch": 1.6114517246894493, "grad_norm": 1.3415398597717285, "learning_rate": 3.955e-06, "loss": 0.0241, "step": 30226 }, { "epoch": 1.6115050381191023, "grad_norm": 3.235659599304199, "learning_rate": 3.954800000000001e-06, "loss": 0.0172, "step": 30227 }, { "epoch": 1.611558351548755, "grad_norm": 0.8990520238876343, "learning_rate": 3.9546e-06, "loss": 0.0083, "step": 30228 }, { "epoch": 1.611611664978408, "grad_norm": 0.7551781535148621, "learning_rate": 3.954400000000001e-06, "loss": 0.0081, "step": 30229 }, { "epoch": 1.611664978408061, "grad_norm": 1.5062514543533325, "learning_rate": 3.9542e-06, "loss": 0.0152, "step": 30230 }, { "epoch": 1.6117182918377138, "grad_norm": 1.9921855926513672, "learning_rate": 3.954e-06, "loss": 0.0477, "step": 30231 }, { "epoch": 1.611771605267367, "grad_norm": 1.5535483360290527, "learning_rate": 3.9538e-06, "loss": 0.0131, "step": 30232 }, { "epoch": 1.6118249186970197, "grad_norm": 2.6782469749450684, "learning_rate": 3.9536000000000006e-06, "loss": 0.0411, "step": 30233 }, { "epoch": 1.6118782321266727, "grad_norm": 0.38225501775741577, "learning_rate": 3.953400000000001e-06, "loss": 0.0036, "step": 30234 }, { "epoch": 1.6119315455563257, "grad_norm": 2.537247896194458, "learning_rate": 3.9532000000000005e-06, "loss": 0.0481, "step": 30235 }, { "epoch": 1.6119848589859784, "grad_norm": 0.32157087326049805, "learning_rate": 3.953e-06, "loss": 0.0034, "step": 30236 }, { "epoch": 1.6120381724156316, "grad_norm": 2.5759804248809814, "learning_rate": 3.9528000000000005e-06, "loss": 0.0353, "step": 30237 }, { "epoch": 1.6120914858452844, "grad_norm": 1.4668325185775757, "learning_rate": 3.9526e-06, "loss": 0.0235, "step": 30238 }, { "epoch": 1.6121447992749374, "grad_norm": 0.7976971864700317, "learning_rate": 3.9524000000000005e-06, "loss": 0.0061, "step": 30239 }, { "epoch": 1.6121981127045903, "grad_norm": 0.33582937717437744, "learning_rate": 3.952200000000001e-06, "loss": 0.0039, "step": 30240 }, { "epoch": 1.612251426134243, "grad_norm": 1.5551364421844482, "learning_rate": 3.9520000000000004e-06, "loss": 0.0324, "step": 30241 }, { "epoch": 1.6123047395638963, "grad_norm": 2.0965824127197266, "learning_rate": 3.9518e-06, "loss": 0.0212, "step": 30242 }, { "epoch": 1.612358052993549, "grad_norm": 1.049414873123169, "learning_rate": 3.9516e-06, "loss": 0.0165, "step": 30243 }, { "epoch": 1.612411366423202, "grad_norm": 1.8778531551361084, "learning_rate": 3.9514e-06, "loss": 0.026, "step": 30244 }, { "epoch": 1.612464679852855, "grad_norm": 1.6122815608978271, "learning_rate": 3.9512e-06, "loss": 0.0158, "step": 30245 }, { "epoch": 1.6125179932825078, "grad_norm": 2.4793570041656494, "learning_rate": 3.951000000000001e-06, "loss": 0.0318, "step": 30246 }, { "epoch": 1.6125713067121608, "grad_norm": 2.395012378692627, "learning_rate": 3.9508e-06, "loss": 0.0261, "step": 30247 }, { "epoch": 1.6126246201418137, "grad_norm": 2.165241241455078, "learning_rate": 3.950600000000001e-06, "loss": 0.0141, "step": 30248 }, { "epoch": 1.6126779335714665, "grad_norm": 2.7853994369506836, "learning_rate": 3.9504e-06, "loss": 0.0276, "step": 30249 }, { "epoch": 1.6127312470011197, "grad_norm": 1.2145315408706665, "learning_rate": 3.9502e-06, "loss": 0.0174, "step": 30250 }, { "epoch": 1.6127845604307725, "grad_norm": 1.4624541997909546, "learning_rate": 3.95e-06, "loss": 0.0239, "step": 30251 }, { "epoch": 1.6128378738604254, "grad_norm": 1.752016305923462, "learning_rate": 3.949800000000001e-06, "loss": 0.0376, "step": 30252 }, { "epoch": 1.6128911872900784, "grad_norm": 1.3711806535720825, "learning_rate": 3.9496e-06, "loss": 0.012, "step": 30253 }, { "epoch": 1.6129445007197312, "grad_norm": 0.7015097737312317, "learning_rate": 3.949400000000001e-06, "loss": 0.0075, "step": 30254 }, { "epoch": 1.6129978141493844, "grad_norm": 1.6008003950119019, "learning_rate": 3.9492e-06, "loss": 0.0264, "step": 30255 }, { "epoch": 1.6130511275790371, "grad_norm": 0.5030843019485474, "learning_rate": 3.949e-06, "loss": 0.0057, "step": 30256 }, { "epoch": 1.61310444100869, "grad_norm": 4.195922374725342, "learning_rate": 3.9488e-06, "loss": 0.0755, "step": 30257 }, { "epoch": 1.613157754438343, "grad_norm": 1.8132487535476685, "learning_rate": 3.948600000000001e-06, "loss": 0.0243, "step": 30258 }, { "epoch": 1.6132110678679958, "grad_norm": 2.3567779064178467, "learning_rate": 3.9484e-06, "loss": 0.0485, "step": 30259 }, { "epoch": 1.613264381297649, "grad_norm": 1.1776357889175415, "learning_rate": 3.948200000000001e-06, "loss": 0.0283, "step": 30260 }, { "epoch": 1.6133176947273018, "grad_norm": 1.3102655410766602, "learning_rate": 3.948e-06, "loss": 0.02, "step": 30261 }, { "epoch": 1.6133710081569548, "grad_norm": 1.1270350217819214, "learning_rate": 3.9478000000000006e-06, "loss": 0.0075, "step": 30262 }, { "epoch": 1.6134243215866078, "grad_norm": 2.2896039485931396, "learning_rate": 3.9476e-06, "loss": 0.0227, "step": 30263 }, { "epoch": 1.6134776350162605, "grad_norm": 2.2466049194335938, "learning_rate": 3.9474000000000005e-06, "loss": 0.0158, "step": 30264 }, { "epoch": 1.6135309484459135, "grad_norm": 1.265226125717163, "learning_rate": 3.9472e-06, "loss": 0.0122, "step": 30265 }, { "epoch": 1.6135842618755665, "grad_norm": 0.7311401963233948, "learning_rate": 3.9470000000000005e-06, "loss": 0.009, "step": 30266 }, { "epoch": 1.6136375753052192, "grad_norm": 2.2927730083465576, "learning_rate": 3.9468e-06, "loss": 0.0274, "step": 30267 }, { "epoch": 1.6136908887348724, "grad_norm": 1.5587869882583618, "learning_rate": 3.9466000000000005e-06, "loss": 0.0288, "step": 30268 }, { "epoch": 1.6137442021645252, "grad_norm": 1.4589200019836426, "learning_rate": 3.9464e-06, "loss": 0.0115, "step": 30269 }, { "epoch": 1.6137975155941782, "grad_norm": 0.9831502437591553, "learning_rate": 3.9462000000000004e-06, "loss": 0.0122, "step": 30270 }, { "epoch": 1.6138508290238311, "grad_norm": 1.6598230600357056, "learning_rate": 3.946e-06, "loss": 0.026, "step": 30271 }, { "epoch": 1.613904142453484, "grad_norm": 1.9613252878189087, "learning_rate": 3.9458e-06, "loss": 0.0157, "step": 30272 }, { "epoch": 1.613957455883137, "grad_norm": 3.0554404258728027, "learning_rate": 3.945600000000001e-06, "loss": 0.0368, "step": 30273 }, { "epoch": 1.6140107693127899, "grad_norm": 0.809343159198761, "learning_rate": 3.9454e-06, "loss": 0.0089, "step": 30274 }, { "epoch": 1.6140640827424428, "grad_norm": 1.5196130275726318, "learning_rate": 3.9452e-06, "loss": 0.0149, "step": 30275 }, { "epoch": 1.6141173961720958, "grad_norm": 1.2380032539367676, "learning_rate": 3.945e-06, "loss": 0.0143, "step": 30276 }, { "epoch": 1.6141707096017486, "grad_norm": 1.3036296367645264, "learning_rate": 3.9448e-06, "loss": 0.0237, "step": 30277 }, { "epoch": 1.6142240230314016, "grad_norm": 1.3219259977340698, "learning_rate": 3.9446e-06, "loss": 0.0088, "step": 30278 }, { "epoch": 1.6142773364610545, "grad_norm": 2.1984071731567383, "learning_rate": 3.944400000000001e-06, "loss": 0.0261, "step": 30279 }, { "epoch": 1.6143306498907075, "grad_norm": 1.9979681968688965, "learning_rate": 3.9442e-06, "loss": 0.0257, "step": 30280 }, { "epoch": 1.6143839633203605, "grad_norm": 1.2868202924728394, "learning_rate": 3.944e-06, "loss": 0.0072, "step": 30281 }, { "epoch": 1.6144372767500133, "grad_norm": 1.468161940574646, "learning_rate": 3.9438e-06, "loss": 0.014, "step": 30282 }, { "epoch": 1.6144905901796662, "grad_norm": 0.9728143215179443, "learning_rate": 3.9436e-06, "loss": 0.0076, "step": 30283 }, { "epoch": 1.6145439036093192, "grad_norm": 1.1627769470214844, "learning_rate": 3.9434e-06, "loss": 0.008, "step": 30284 }, { "epoch": 1.614597217038972, "grad_norm": 2.9147446155548096, "learning_rate": 3.943200000000001e-06, "loss": 0.0446, "step": 30285 }, { "epoch": 1.6146505304686252, "grad_norm": 3.056797504425049, "learning_rate": 3.943e-06, "loss": 0.0302, "step": 30286 }, { "epoch": 1.614703843898278, "grad_norm": 2.5510036945343018, "learning_rate": 3.942800000000001e-06, "loss": 0.0133, "step": 30287 }, { "epoch": 1.614757157327931, "grad_norm": 0.831023097038269, "learning_rate": 3.9426e-06, "loss": 0.0082, "step": 30288 }, { "epoch": 1.6148104707575839, "grad_norm": 1.0925624370574951, "learning_rate": 3.9424e-06, "loss": 0.0151, "step": 30289 }, { "epoch": 1.6148637841872366, "grad_norm": 1.5306930541992188, "learning_rate": 3.9422e-06, "loss": 0.0242, "step": 30290 }, { "epoch": 1.6149170976168898, "grad_norm": 1.240466833114624, "learning_rate": 3.9420000000000005e-06, "loss": 0.0134, "step": 30291 }, { "epoch": 1.6149704110465426, "grad_norm": 2.638005256652832, "learning_rate": 3.9418e-06, "loss": 0.0352, "step": 30292 }, { "epoch": 1.6150237244761956, "grad_norm": 2.041294813156128, "learning_rate": 3.9416000000000005e-06, "loss": 0.061, "step": 30293 }, { "epoch": 1.6150770379058486, "grad_norm": 2.717848062515259, "learning_rate": 3.9414e-06, "loss": 0.0372, "step": 30294 }, { "epoch": 1.6151303513355013, "grad_norm": 3.450263500213623, "learning_rate": 3.9412000000000005e-06, "loss": 0.0315, "step": 30295 }, { "epoch": 1.6151836647651543, "grad_norm": 2.190120220184326, "learning_rate": 3.941e-06, "loss": 0.021, "step": 30296 }, { "epoch": 1.6152369781948073, "grad_norm": 2.3639636039733887, "learning_rate": 3.9408000000000005e-06, "loss": 0.0168, "step": 30297 }, { "epoch": 1.6152902916244603, "grad_norm": 2.139415740966797, "learning_rate": 3.940600000000001e-06, "loss": 0.0144, "step": 30298 }, { "epoch": 1.6153436050541132, "grad_norm": 0.6502702236175537, "learning_rate": 3.9404e-06, "loss": 0.0053, "step": 30299 }, { "epoch": 1.615396918483766, "grad_norm": 3.2564857006073, "learning_rate": 3.9402e-06, "loss": 0.0352, "step": 30300 }, { "epoch": 1.615450231913419, "grad_norm": 1.1836031675338745, "learning_rate": 3.94e-06, "loss": 0.0136, "step": 30301 }, { "epoch": 1.615503545343072, "grad_norm": 1.7425436973571777, "learning_rate": 3.9398e-06, "loss": 0.0309, "step": 30302 }, { "epoch": 1.6155568587727247, "grad_norm": 2.503918170928955, "learning_rate": 3.9396e-06, "loss": 0.013, "step": 30303 }, { "epoch": 1.615610172202378, "grad_norm": 0.48248353600502014, "learning_rate": 3.939400000000001e-06, "loss": 0.0042, "step": 30304 }, { "epoch": 1.6156634856320307, "grad_norm": 5.631697177886963, "learning_rate": 3.9392e-06, "loss": 0.0402, "step": 30305 }, { "epoch": 1.6157167990616836, "grad_norm": 0.7316255569458008, "learning_rate": 3.939e-06, "loss": 0.0157, "step": 30306 }, { "epoch": 1.6157701124913366, "grad_norm": 1.9673672914505005, "learning_rate": 3.9388e-06, "loss": 0.027, "step": 30307 }, { "epoch": 1.6158234259209894, "grad_norm": 1.9854172468185425, "learning_rate": 3.9386e-06, "loss": 0.0256, "step": 30308 }, { "epoch": 1.6158767393506426, "grad_norm": 1.9036614894866943, "learning_rate": 3.9384e-06, "loss": 0.0278, "step": 30309 }, { "epoch": 1.6159300527802953, "grad_norm": 1.2984808683395386, "learning_rate": 3.938200000000001e-06, "loss": 0.0177, "step": 30310 }, { "epoch": 1.6159833662099483, "grad_norm": 0.94756680727005, "learning_rate": 3.938e-06, "loss": 0.0096, "step": 30311 }, { "epoch": 1.6160366796396013, "grad_norm": 1.6093132495880127, "learning_rate": 3.937800000000001e-06, "loss": 0.0169, "step": 30312 }, { "epoch": 1.616089993069254, "grad_norm": 1.5661680698394775, "learning_rate": 3.9376e-06, "loss": 0.0156, "step": 30313 }, { "epoch": 1.616143306498907, "grad_norm": 2.4665932655334473, "learning_rate": 3.9374e-06, "loss": 0.0236, "step": 30314 }, { "epoch": 1.61619661992856, "grad_norm": 1.7569584846496582, "learning_rate": 3.9372e-06, "loss": 0.0282, "step": 30315 }, { "epoch": 1.616249933358213, "grad_norm": 1.1966502666473389, "learning_rate": 3.937000000000001e-06, "loss": 0.0146, "step": 30316 }, { "epoch": 1.616303246787866, "grad_norm": 0.8268707990646362, "learning_rate": 3.9368e-06, "loss": 0.0046, "step": 30317 }, { "epoch": 1.6163565602175187, "grad_norm": 0.3330429792404175, "learning_rate": 3.9366000000000006e-06, "loss": 0.0037, "step": 30318 }, { "epoch": 1.6164098736471717, "grad_norm": 4.3418049812316895, "learning_rate": 3.9364e-06, "loss": 0.0191, "step": 30319 }, { "epoch": 1.6164631870768247, "grad_norm": 4.213395118713379, "learning_rate": 3.9362000000000005e-06, "loss": 0.0307, "step": 30320 }, { "epoch": 1.6165165005064774, "grad_norm": 2.754973888397217, "learning_rate": 3.936e-06, "loss": 0.0483, "step": 30321 }, { "epoch": 1.6165698139361306, "grad_norm": 2.647001266479492, "learning_rate": 3.9358000000000005e-06, "loss": 0.0203, "step": 30322 }, { "epoch": 1.6166231273657834, "grad_norm": 1.660452961921692, "learning_rate": 3.935600000000001e-06, "loss": 0.0172, "step": 30323 }, { "epoch": 1.6166764407954364, "grad_norm": 1.4947842359542847, "learning_rate": 3.9354000000000005e-06, "loss": 0.0085, "step": 30324 }, { "epoch": 1.6167297542250894, "grad_norm": 0.9836134910583496, "learning_rate": 3.9352e-06, "loss": 0.0125, "step": 30325 }, { "epoch": 1.6167830676547421, "grad_norm": 0.7845046520233154, "learning_rate": 3.9350000000000004e-06, "loss": 0.0213, "step": 30326 }, { "epoch": 1.6168363810843953, "grad_norm": 1.624634027481079, "learning_rate": 3.9348e-06, "loss": 0.0161, "step": 30327 }, { "epoch": 1.616889694514048, "grad_norm": 0.7905993461608887, "learning_rate": 3.9346e-06, "loss": 0.0073, "step": 30328 }, { "epoch": 1.616943007943701, "grad_norm": 1.363465428352356, "learning_rate": 3.934400000000001e-06, "loss": 0.0137, "step": 30329 }, { "epoch": 1.616996321373354, "grad_norm": 1.2890156507492065, "learning_rate": 3.9342e-06, "loss": 0.0319, "step": 30330 }, { "epoch": 1.6170496348030068, "grad_norm": 2.3875043392181396, "learning_rate": 3.934000000000001e-06, "loss": 0.0188, "step": 30331 }, { "epoch": 1.6171029482326598, "grad_norm": 0.2545822262763977, "learning_rate": 3.9338e-06, "loss": 0.0026, "step": 30332 }, { "epoch": 1.6171562616623127, "grad_norm": 0.8505792617797852, "learning_rate": 3.9336e-06, "loss": 0.0092, "step": 30333 }, { "epoch": 1.6172095750919655, "grad_norm": 2.169898748397827, "learning_rate": 3.9334e-06, "loss": 0.0477, "step": 30334 }, { "epoch": 1.6172628885216187, "grad_norm": 1.4452080726623535, "learning_rate": 3.933200000000001e-06, "loss": 0.0297, "step": 30335 }, { "epoch": 1.6173162019512715, "grad_norm": 0.7969890832901001, "learning_rate": 3.933e-06, "loss": 0.0082, "step": 30336 }, { "epoch": 1.6173695153809244, "grad_norm": 1.3772625923156738, "learning_rate": 3.932800000000001e-06, "loss": 0.0263, "step": 30337 }, { "epoch": 1.6174228288105774, "grad_norm": 4.07362699508667, "learning_rate": 3.9326e-06, "loss": 0.0405, "step": 30338 }, { "epoch": 1.6174761422402302, "grad_norm": 1.0284299850463867, "learning_rate": 3.9324e-06, "loss": 0.0088, "step": 30339 }, { "epoch": 1.6175294556698834, "grad_norm": 1.4462568759918213, "learning_rate": 3.9322e-06, "loss": 0.0132, "step": 30340 }, { "epoch": 1.6175827690995361, "grad_norm": 3.11482310295105, "learning_rate": 3.932000000000001e-06, "loss": 0.0316, "step": 30341 }, { "epoch": 1.6176360825291891, "grad_norm": 3.0876357555389404, "learning_rate": 3.9318e-06, "loss": 0.016, "step": 30342 }, { "epoch": 1.617689395958842, "grad_norm": 0.4372852146625519, "learning_rate": 3.931600000000001e-06, "loss": 0.0046, "step": 30343 }, { "epoch": 1.6177427093884948, "grad_norm": 0.8950645923614502, "learning_rate": 3.9314e-06, "loss": 0.014, "step": 30344 }, { "epoch": 1.617796022818148, "grad_norm": 2.872912883758545, "learning_rate": 3.931200000000001e-06, "loss": 0.0265, "step": 30345 }, { "epoch": 1.6178493362478008, "grad_norm": 1.933576226234436, "learning_rate": 3.931e-06, "loss": 0.0121, "step": 30346 }, { "epoch": 1.6179026496774538, "grad_norm": 2.1109395027160645, "learning_rate": 3.9308000000000005e-06, "loss": 0.0122, "step": 30347 }, { "epoch": 1.6179559631071068, "grad_norm": 0.9299259185791016, "learning_rate": 3.9306e-06, "loss": 0.005, "step": 30348 }, { "epoch": 1.6180092765367595, "grad_norm": 1.4819629192352295, "learning_rate": 3.9304000000000005e-06, "loss": 0.0306, "step": 30349 }, { "epoch": 1.6180625899664125, "grad_norm": 1.2833369970321655, "learning_rate": 3.9302e-06, "loss": 0.022, "step": 30350 }, { "epoch": 1.6181159033960655, "grad_norm": 1.4314115047454834, "learning_rate": 3.9300000000000005e-06, "loss": 0.0219, "step": 30351 }, { "epoch": 1.6181692168257182, "grad_norm": 2.165116548538208, "learning_rate": 3.9298e-06, "loss": 0.0146, "step": 30352 }, { "epoch": 1.6182225302553714, "grad_norm": 1.1150304079055786, "learning_rate": 3.9296000000000005e-06, "loss": 0.0136, "step": 30353 }, { "epoch": 1.6182758436850242, "grad_norm": 1.2575513124465942, "learning_rate": 3.9294e-06, "loss": 0.0101, "step": 30354 }, { "epoch": 1.6183291571146772, "grad_norm": 0.9100730419158936, "learning_rate": 3.9292000000000004e-06, "loss": 0.0062, "step": 30355 }, { "epoch": 1.6183824705443302, "grad_norm": 0.3829054534435272, "learning_rate": 3.929000000000001e-06, "loss": 0.0043, "step": 30356 }, { "epoch": 1.618435783973983, "grad_norm": 1.9382156133651733, "learning_rate": 3.9288e-06, "loss": 0.0306, "step": 30357 }, { "epoch": 1.6184890974036361, "grad_norm": 0.9469442367553711, "learning_rate": 3.9286e-06, "loss": 0.0069, "step": 30358 }, { "epoch": 1.6185424108332889, "grad_norm": 0.2968924343585968, "learning_rate": 3.9284e-06, "loss": 0.0029, "step": 30359 }, { "epoch": 1.6185957242629418, "grad_norm": 1.601022481918335, "learning_rate": 3.9282e-06, "loss": 0.0145, "step": 30360 }, { "epoch": 1.6186490376925948, "grad_norm": 1.225260615348816, "learning_rate": 3.928e-06, "loss": 0.0104, "step": 30361 }, { "epoch": 1.6187023511222476, "grad_norm": 1.6320043802261353, "learning_rate": 3.927800000000001e-06, "loss": 0.0293, "step": 30362 }, { "epoch": 1.6187556645519008, "grad_norm": 1.1944442987442017, "learning_rate": 3.9276e-06, "loss": 0.0057, "step": 30363 }, { "epoch": 1.6188089779815535, "grad_norm": 1.3323220014572144, "learning_rate": 3.9274e-06, "loss": 0.0074, "step": 30364 }, { "epoch": 1.6188622914112065, "grad_norm": 0.7717491984367371, "learning_rate": 3.9272e-06, "loss": 0.0064, "step": 30365 }, { "epoch": 1.6189156048408595, "grad_norm": 1.778100609779358, "learning_rate": 3.927e-06, "loss": 0.0223, "step": 30366 }, { "epoch": 1.6189689182705123, "grad_norm": 1.0170540809631348, "learning_rate": 3.9268e-06, "loss": 0.0056, "step": 30367 }, { "epoch": 1.6190222317001652, "grad_norm": 1.0684183835983276, "learning_rate": 3.926600000000001e-06, "loss": 0.0397, "step": 30368 }, { "epoch": 1.6190755451298182, "grad_norm": 1.5156503915786743, "learning_rate": 3.9264e-06, "loss": 0.0117, "step": 30369 }, { "epoch": 1.619128858559471, "grad_norm": 1.337410569190979, "learning_rate": 3.926200000000001e-06, "loss": 0.0218, "step": 30370 }, { "epoch": 1.6191821719891242, "grad_norm": 2.5165023803710938, "learning_rate": 3.926e-06, "loss": 0.029, "step": 30371 }, { "epoch": 1.619235485418777, "grad_norm": 1.3482118844985962, "learning_rate": 3.9258e-06, "loss": 0.0311, "step": 30372 }, { "epoch": 1.61928879884843, "grad_norm": 2.3322911262512207, "learning_rate": 3.9256e-06, "loss": 0.0201, "step": 30373 }, { "epoch": 1.619342112278083, "grad_norm": 1.528347373008728, "learning_rate": 3.9254000000000006e-06, "loss": 0.0154, "step": 30374 }, { "epoch": 1.6193954257077356, "grad_norm": 1.5037829875946045, "learning_rate": 3.9252e-06, "loss": 0.0209, "step": 30375 }, { "epoch": 1.6194487391373888, "grad_norm": 4.974376201629639, "learning_rate": 3.9250000000000005e-06, "loss": 0.036, "step": 30376 }, { "epoch": 1.6195020525670416, "grad_norm": 1.1589683294296265, "learning_rate": 3.9248e-06, "loss": 0.0092, "step": 30377 }, { "epoch": 1.6195553659966946, "grad_norm": 0.8011161684989929, "learning_rate": 3.9246e-06, "loss": 0.0074, "step": 30378 }, { "epoch": 1.6196086794263476, "grad_norm": 2.5901215076446533, "learning_rate": 3.9244e-06, "loss": 0.0331, "step": 30379 }, { "epoch": 1.6196619928560003, "grad_norm": 1.8422123193740845, "learning_rate": 3.9242000000000005e-06, "loss": 0.0115, "step": 30380 }, { "epoch": 1.6197153062856535, "grad_norm": 0.7345472574234009, "learning_rate": 3.924000000000001e-06, "loss": 0.006, "step": 30381 }, { "epoch": 1.6197686197153063, "grad_norm": 2.028920888900757, "learning_rate": 3.9238000000000004e-06, "loss": 0.0138, "step": 30382 }, { "epoch": 1.6198219331449593, "grad_norm": 1.5434095859527588, "learning_rate": 3.9236e-06, "loss": 0.031, "step": 30383 }, { "epoch": 1.6198752465746122, "grad_norm": 1.091603398323059, "learning_rate": 3.9234e-06, "loss": 0.0094, "step": 30384 }, { "epoch": 1.619928560004265, "grad_norm": 0.6070486903190613, "learning_rate": 3.9232e-06, "loss": 0.0032, "step": 30385 }, { "epoch": 1.619981873433918, "grad_norm": 2.002713441848755, "learning_rate": 3.923e-06, "loss": 0.0166, "step": 30386 }, { "epoch": 1.620035186863571, "grad_norm": 1.172849178314209, "learning_rate": 3.922800000000001e-06, "loss": 0.0066, "step": 30387 }, { "epoch": 1.6200885002932237, "grad_norm": 1.7890514135360718, "learning_rate": 3.9226e-06, "loss": 0.0408, "step": 30388 }, { "epoch": 1.620141813722877, "grad_norm": 1.6589850187301636, "learning_rate": 3.9224e-06, "loss": 0.0222, "step": 30389 }, { "epoch": 1.6201951271525297, "grad_norm": 0.6670497059822083, "learning_rate": 3.9222e-06, "loss": 0.006, "step": 30390 }, { "epoch": 1.6202484405821826, "grad_norm": 1.192296028137207, "learning_rate": 3.922e-06, "loss": 0.0302, "step": 30391 }, { "epoch": 1.6203017540118356, "grad_norm": 1.4231774806976318, "learning_rate": 3.9218e-06, "loss": 0.0094, "step": 30392 }, { "epoch": 1.6203550674414884, "grad_norm": 0.5319398045539856, "learning_rate": 3.921600000000001e-06, "loss": 0.0038, "step": 30393 }, { "epoch": 1.6204083808711416, "grad_norm": 1.8431544303894043, "learning_rate": 3.9214e-06, "loss": 0.034, "step": 30394 }, { "epoch": 1.6204616943007943, "grad_norm": 3.5077295303344727, "learning_rate": 3.921200000000001e-06, "loss": 0.0336, "step": 30395 }, { "epoch": 1.6205150077304473, "grad_norm": 1.6736139059066772, "learning_rate": 3.921e-06, "loss": 0.0221, "step": 30396 }, { "epoch": 1.6205683211601003, "grad_norm": 1.9623631238937378, "learning_rate": 3.9208e-06, "loss": 0.054, "step": 30397 }, { "epoch": 1.620621634589753, "grad_norm": 1.7488106489181519, "learning_rate": 3.9206e-06, "loss": 0.0164, "step": 30398 }, { "epoch": 1.620674948019406, "grad_norm": 1.181705355644226, "learning_rate": 3.920400000000001e-06, "loss": 0.0211, "step": 30399 }, { "epoch": 1.620728261449059, "grad_norm": 2.2845921516418457, "learning_rate": 3.9202e-06, "loss": 0.0349, "step": 30400 }, { "epoch": 1.620781574878712, "grad_norm": 1.571735143661499, "learning_rate": 3.920000000000001e-06, "loss": 0.0127, "step": 30401 }, { "epoch": 1.620834888308365, "grad_norm": 1.8605139255523682, "learning_rate": 3.9198e-06, "loss": 0.0132, "step": 30402 }, { "epoch": 1.6208882017380177, "grad_norm": 4.046838283538818, "learning_rate": 3.9196e-06, "loss": 0.0537, "step": 30403 }, { "epoch": 1.6209415151676707, "grad_norm": 1.450192928314209, "learning_rate": 3.9194e-06, "loss": 0.0068, "step": 30404 }, { "epoch": 1.6209948285973237, "grad_norm": 1.563624620437622, "learning_rate": 3.9192000000000005e-06, "loss": 0.0141, "step": 30405 }, { "epoch": 1.6210481420269764, "grad_norm": 0.30902692675590515, "learning_rate": 3.919000000000001e-06, "loss": 0.002, "step": 30406 }, { "epoch": 1.6211014554566296, "grad_norm": 1.6561026573181152, "learning_rate": 3.9188000000000005e-06, "loss": 0.0279, "step": 30407 }, { "epoch": 1.6211547688862824, "grad_norm": 2.2846012115478516, "learning_rate": 3.9186e-06, "loss": 0.0126, "step": 30408 }, { "epoch": 1.6212080823159354, "grad_norm": 1.3183820247650146, "learning_rate": 3.9184000000000005e-06, "loss": 0.0079, "step": 30409 }, { "epoch": 1.6212613957455884, "grad_norm": 1.398742914199829, "learning_rate": 3.9182e-06, "loss": 0.0095, "step": 30410 }, { "epoch": 1.6213147091752411, "grad_norm": 3.4113550186157227, "learning_rate": 3.9180000000000004e-06, "loss": 0.0265, "step": 30411 }, { "epoch": 1.6213680226048943, "grad_norm": 2.584409713745117, "learning_rate": 3.917800000000001e-06, "loss": 0.0149, "step": 30412 }, { "epoch": 1.621421336034547, "grad_norm": 0.8993369340896606, "learning_rate": 3.9176e-06, "loss": 0.0074, "step": 30413 }, { "epoch": 1.6214746494642, "grad_norm": 1.273126244544983, "learning_rate": 3.9174e-06, "loss": 0.0211, "step": 30414 }, { "epoch": 1.621527962893853, "grad_norm": 3.134243965148926, "learning_rate": 3.9172e-06, "loss": 0.0451, "step": 30415 }, { "epoch": 1.6215812763235058, "grad_norm": 1.579696774482727, "learning_rate": 3.917e-06, "loss": 0.0268, "step": 30416 }, { "epoch": 1.6216345897531588, "grad_norm": 2.1917126178741455, "learning_rate": 3.9168e-06, "loss": 0.0449, "step": 30417 }, { "epoch": 1.6216879031828118, "grad_norm": 2.283155918121338, "learning_rate": 3.916600000000001e-06, "loss": 0.0277, "step": 30418 }, { "epoch": 1.6217412166124647, "grad_norm": 1.5197137594223022, "learning_rate": 3.9164e-06, "loss": 0.0165, "step": 30419 }, { "epoch": 1.6217945300421177, "grad_norm": 1.8167310953140259, "learning_rate": 3.916200000000001e-06, "loss": 0.0225, "step": 30420 }, { "epoch": 1.6218478434717705, "grad_norm": 2.133491039276123, "learning_rate": 3.916e-06, "loss": 0.0352, "step": 30421 }, { "epoch": 1.6219011569014234, "grad_norm": 2.2801690101623535, "learning_rate": 3.9158e-06, "loss": 0.0247, "step": 30422 }, { "epoch": 1.6219544703310764, "grad_norm": 1.2587230205535889, "learning_rate": 3.9156e-06, "loss": 0.012, "step": 30423 }, { "epoch": 1.6220077837607292, "grad_norm": 1.900554895401001, "learning_rate": 3.915400000000001e-06, "loss": 0.0205, "step": 30424 }, { "epoch": 1.6220610971903824, "grad_norm": 3.1141459941864014, "learning_rate": 3.9152e-06, "loss": 0.0347, "step": 30425 }, { "epoch": 1.6221144106200351, "grad_norm": 0.3863426148891449, "learning_rate": 3.915000000000001e-06, "loss": 0.0027, "step": 30426 }, { "epoch": 1.6221677240496881, "grad_norm": 1.4153999090194702, "learning_rate": 3.9148e-06, "loss": 0.0163, "step": 30427 }, { "epoch": 1.622221037479341, "grad_norm": 2.0371246337890625, "learning_rate": 3.9146e-06, "loss": 0.0172, "step": 30428 }, { "epoch": 1.6222743509089939, "grad_norm": 1.0084338188171387, "learning_rate": 3.9144e-06, "loss": 0.0183, "step": 30429 }, { "epoch": 1.622327664338647, "grad_norm": 0.15836429595947266, "learning_rate": 3.9142000000000006e-06, "loss": 0.0017, "step": 30430 }, { "epoch": 1.6223809777682998, "grad_norm": 1.3472894430160522, "learning_rate": 3.914000000000001e-06, "loss": 0.01, "step": 30431 }, { "epoch": 1.6224342911979528, "grad_norm": 1.2503911256790161, "learning_rate": 3.9138000000000005e-06, "loss": 0.0156, "step": 30432 }, { "epoch": 1.6224876046276058, "grad_norm": 1.7772443294525146, "learning_rate": 3.9136e-06, "loss": 0.0284, "step": 30433 }, { "epoch": 1.6225409180572585, "grad_norm": 1.6266716718673706, "learning_rate": 3.9134000000000005e-06, "loss": 0.0114, "step": 30434 }, { "epoch": 1.6225942314869115, "grad_norm": 1.5687460899353027, "learning_rate": 3.9132e-06, "loss": 0.022, "step": 30435 }, { "epoch": 1.6226475449165645, "grad_norm": 1.4753377437591553, "learning_rate": 3.9130000000000005e-06, "loss": 0.0257, "step": 30436 }, { "epoch": 1.6227008583462175, "grad_norm": 1.9508259296417236, "learning_rate": 3.912800000000001e-06, "loss": 0.0138, "step": 30437 }, { "epoch": 1.6227541717758704, "grad_norm": 2.4107487201690674, "learning_rate": 3.9126000000000005e-06, "loss": 0.0201, "step": 30438 }, { "epoch": 1.6228074852055232, "grad_norm": 1.1005018949508667, "learning_rate": 3.9124e-06, "loss": 0.0079, "step": 30439 }, { "epoch": 1.6228607986351762, "grad_norm": 0.9834099411964417, "learning_rate": 3.9122e-06, "loss": 0.0055, "step": 30440 }, { "epoch": 1.6229141120648292, "grad_norm": 1.948180913925171, "learning_rate": 3.912e-06, "loss": 0.0246, "step": 30441 }, { "epoch": 1.622967425494482, "grad_norm": 2.5619547367095947, "learning_rate": 3.9118e-06, "loss": 0.0596, "step": 30442 }, { "epoch": 1.6230207389241351, "grad_norm": 0.9622629880905151, "learning_rate": 3.911600000000001e-06, "loss": 0.0088, "step": 30443 }, { "epoch": 1.6230740523537879, "grad_norm": 0.96567302942276, "learning_rate": 3.9114e-06, "loss": 0.0073, "step": 30444 }, { "epoch": 1.6231273657834409, "grad_norm": 2.266286611557007, "learning_rate": 3.911200000000001e-06, "loss": 0.0204, "step": 30445 }, { "epoch": 1.6231806792130938, "grad_norm": 1.0910006761550903, "learning_rate": 3.911e-06, "loss": 0.032, "step": 30446 }, { "epoch": 1.6232339926427466, "grad_norm": 1.4385498762130737, "learning_rate": 3.9108e-06, "loss": 0.0259, "step": 30447 }, { "epoch": 1.6232873060723998, "grad_norm": 1.8414607048034668, "learning_rate": 3.9106e-06, "loss": 0.0165, "step": 30448 }, { "epoch": 1.6233406195020526, "grad_norm": 3.255319118499756, "learning_rate": 3.910400000000001e-06, "loss": 0.0469, "step": 30449 }, { "epoch": 1.6233939329317055, "grad_norm": 1.1579214334487915, "learning_rate": 3.9102e-06, "loss": 0.0127, "step": 30450 }, { "epoch": 1.6234472463613585, "grad_norm": 2.187687635421753, "learning_rate": 3.910000000000001e-06, "loss": 0.0111, "step": 30451 }, { "epoch": 1.6235005597910113, "grad_norm": 0.7657272219657898, "learning_rate": 3.9098e-06, "loss": 0.0069, "step": 30452 }, { "epoch": 1.6235538732206642, "grad_norm": 1.6520423889160156, "learning_rate": 3.9096e-06, "loss": 0.0165, "step": 30453 }, { "epoch": 1.6236071866503172, "grad_norm": 1.006160855293274, "learning_rate": 3.9094e-06, "loss": 0.0164, "step": 30454 }, { "epoch": 1.62366050007997, "grad_norm": 0.8355415463447571, "learning_rate": 3.909200000000001e-06, "loss": 0.0037, "step": 30455 }, { "epoch": 1.6237138135096232, "grad_norm": 1.2218577861785889, "learning_rate": 3.909e-06, "loss": 0.0074, "step": 30456 }, { "epoch": 1.623767126939276, "grad_norm": 1.8224631547927856, "learning_rate": 3.908800000000001e-06, "loss": 0.0114, "step": 30457 }, { "epoch": 1.623820440368929, "grad_norm": 1.0717295408248901, "learning_rate": 3.9086e-06, "loss": 0.0075, "step": 30458 }, { "epoch": 1.623873753798582, "grad_norm": 1.5701831579208374, "learning_rate": 3.9084000000000006e-06, "loss": 0.0193, "step": 30459 }, { "epoch": 1.6239270672282347, "grad_norm": 0.635970413684845, "learning_rate": 3.9082e-06, "loss": 0.0069, "step": 30460 }, { "epoch": 1.6239803806578879, "grad_norm": 2.745335340499878, "learning_rate": 3.9080000000000005e-06, "loss": 0.0315, "step": 30461 }, { "epoch": 1.6240336940875406, "grad_norm": 1.7707242965698242, "learning_rate": 3.9078e-06, "loss": 0.0265, "step": 30462 }, { "epoch": 1.6240870075171936, "grad_norm": 1.6198347806930542, "learning_rate": 3.9076000000000005e-06, "loss": 0.029, "step": 30463 }, { "epoch": 1.6241403209468466, "grad_norm": 0.5280024409294128, "learning_rate": 3.9074e-06, "loss": 0.004, "step": 30464 }, { "epoch": 1.6241936343764993, "grad_norm": 2.0707788467407227, "learning_rate": 3.9072000000000005e-06, "loss": 0.0205, "step": 30465 }, { "epoch": 1.6242469478061525, "grad_norm": 0.5910230278968811, "learning_rate": 3.907e-06, "loss": 0.0045, "step": 30466 }, { "epoch": 1.6243002612358053, "grad_norm": 1.7831147909164429, "learning_rate": 3.9068000000000004e-06, "loss": 0.0401, "step": 30467 }, { "epoch": 1.6243535746654583, "grad_norm": 1.2136119604110718, "learning_rate": 3.9066e-06, "loss": 0.0093, "step": 30468 }, { "epoch": 1.6244068880951112, "grad_norm": 2.22141695022583, "learning_rate": 3.9064e-06, "loss": 0.0341, "step": 30469 }, { "epoch": 1.624460201524764, "grad_norm": 2.053767442703247, "learning_rate": 3.906200000000001e-06, "loss": 0.0193, "step": 30470 }, { "epoch": 1.624513514954417, "grad_norm": 3.0300943851470947, "learning_rate": 3.906e-06, "loss": 0.0321, "step": 30471 }, { "epoch": 1.62456682838407, "grad_norm": 0.7792418599128723, "learning_rate": 3.9058e-06, "loss": 0.0035, "step": 30472 }, { "epoch": 1.6246201418137227, "grad_norm": 1.155293345451355, "learning_rate": 3.9056e-06, "loss": 0.0169, "step": 30473 }, { "epoch": 1.624673455243376, "grad_norm": 1.4910162687301636, "learning_rate": 3.9054e-06, "loss": 0.0347, "step": 30474 }, { "epoch": 1.6247267686730287, "grad_norm": 1.0632119178771973, "learning_rate": 3.9052e-06, "loss": 0.0068, "step": 30475 }, { "epoch": 1.6247800821026817, "grad_norm": 1.049498200416565, "learning_rate": 3.905000000000001e-06, "loss": 0.0125, "step": 30476 }, { "epoch": 1.6248333955323346, "grad_norm": 0.6068239212036133, "learning_rate": 3.9048e-06, "loss": 0.0043, "step": 30477 }, { "epoch": 1.6248867089619874, "grad_norm": 1.771632432937622, "learning_rate": 3.9046e-06, "loss": 0.0349, "step": 30478 }, { "epoch": 1.6249400223916406, "grad_norm": 2.0593318939208984, "learning_rate": 3.9044e-06, "loss": 0.0225, "step": 30479 }, { "epoch": 1.6249933358212933, "grad_norm": 2.394243001937866, "learning_rate": 3.9042e-06, "loss": 0.0314, "step": 30480 }, { "epoch": 1.6250466492509463, "grad_norm": 1.5450855493545532, "learning_rate": 3.904e-06, "loss": 0.0188, "step": 30481 }, { "epoch": 1.6250999626805993, "grad_norm": 2.9623100757598877, "learning_rate": 3.903800000000001e-06, "loss": 0.0383, "step": 30482 }, { "epoch": 1.625153276110252, "grad_norm": 0.5493078231811523, "learning_rate": 3.9036e-06, "loss": 0.0046, "step": 30483 }, { "epoch": 1.6252065895399053, "grad_norm": 0.7219256162643433, "learning_rate": 3.903400000000001e-06, "loss": 0.0062, "step": 30484 }, { "epoch": 1.625259902969558, "grad_norm": 2.1258840560913086, "learning_rate": 3.9032e-06, "loss": 0.0244, "step": 30485 }, { "epoch": 1.625313216399211, "grad_norm": 3.942910671234131, "learning_rate": 3.903e-06, "loss": 0.0332, "step": 30486 }, { "epoch": 1.625366529828864, "grad_norm": 0.6974414587020874, "learning_rate": 3.9028e-06, "loss": 0.0061, "step": 30487 }, { "epoch": 1.6254198432585167, "grad_norm": 1.8830280303955078, "learning_rate": 3.9026000000000005e-06, "loss": 0.0209, "step": 30488 }, { "epoch": 1.6254731566881697, "grad_norm": 1.422257900238037, "learning_rate": 3.9024e-06, "loss": 0.0112, "step": 30489 }, { "epoch": 1.6255264701178227, "grad_norm": 1.0799574851989746, "learning_rate": 3.9022000000000005e-06, "loss": 0.0159, "step": 30490 }, { "epoch": 1.6255797835474755, "grad_norm": 0.7838120460510254, "learning_rate": 3.902e-06, "loss": 0.0068, "step": 30491 }, { "epoch": 1.6256330969771287, "grad_norm": 1.8598353862762451, "learning_rate": 3.9018000000000005e-06, "loss": 0.0235, "step": 30492 }, { "epoch": 1.6256864104067814, "grad_norm": 2.4772722721099854, "learning_rate": 3.9016e-06, "loss": 0.0102, "step": 30493 }, { "epoch": 1.6257397238364344, "grad_norm": 3.0353240966796875, "learning_rate": 3.9014000000000005e-06, "loss": 0.0355, "step": 30494 }, { "epoch": 1.6257930372660874, "grad_norm": 0.8721808791160583, "learning_rate": 3.901200000000001e-06, "loss": 0.0058, "step": 30495 }, { "epoch": 1.6258463506957401, "grad_norm": 1.1104018688201904, "learning_rate": 3.9010000000000004e-06, "loss": 0.0077, "step": 30496 }, { "epoch": 1.6258996641253933, "grad_norm": 1.1475006341934204, "learning_rate": 3.9008e-06, "loss": 0.009, "step": 30497 }, { "epoch": 1.625952977555046, "grad_norm": 0.7785984873771667, "learning_rate": 3.9006e-06, "loss": 0.0064, "step": 30498 }, { "epoch": 1.626006290984699, "grad_norm": 0.9494768381118774, "learning_rate": 3.9004e-06, "loss": 0.0273, "step": 30499 }, { "epoch": 1.626059604414352, "grad_norm": 1.6780331134796143, "learning_rate": 3.9002e-06, "loss": 0.0289, "step": 30500 }, { "epoch": 1.6261129178440048, "grad_norm": 1.461952805519104, "learning_rate": 3.900000000000001e-06, "loss": 0.0112, "step": 30501 }, { "epoch": 1.6261662312736578, "grad_norm": 1.924904704093933, "learning_rate": 3.8998e-06, "loss": 0.0267, "step": 30502 }, { "epoch": 1.6262195447033108, "grad_norm": 1.215971827507019, "learning_rate": 3.8996e-06, "loss": 0.0044, "step": 30503 }, { "epoch": 1.6262728581329637, "grad_norm": 2.2731590270996094, "learning_rate": 3.8994e-06, "loss": 0.0385, "step": 30504 }, { "epoch": 1.6263261715626167, "grad_norm": 2.376911163330078, "learning_rate": 3.8992e-06, "loss": 0.0085, "step": 30505 }, { "epoch": 1.6263794849922695, "grad_norm": 2.592893123626709, "learning_rate": 3.899e-06, "loss": 0.0242, "step": 30506 }, { "epoch": 1.6264327984219225, "grad_norm": 1.524038314819336, "learning_rate": 3.898800000000001e-06, "loss": 0.01, "step": 30507 }, { "epoch": 1.6264861118515754, "grad_norm": 1.962403655052185, "learning_rate": 3.8986e-06, "loss": 0.0281, "step": 30508 }, { "epoch": 1.6265394252812282, "grad_norm": 0.8587412238121033, "learning_rate": 3.898400000000001e-06, "loss": 0.0039, "step": 30509 }, { "epoch": 1.6265927387108814, "grad_norm": 1.8187965154647827, "learning_rate": 3.8982e-06, "loss": 0.0201, "step": 30510 }, { "epoch": 1.6266460521405341, "grad_norm": 2.075901508331299, "learning_rate": 3.898e-06, "loss": 0.0191, "step": 30511 }, { "epoch": 1.6266993655701871, "grad_norm": 2.585909128189087, "learning_rate": 3.8978e-06, "loss": 0.0211, "step": 30512 }, { "epoch": 1.62675267899984, "grad_norm": 2.049534559249878, "learning_rate": 3.897600000000001e-06, "loss": 0.0264, "step": 30513 }, { "epoch": 1.6268059924294929, "grad_norm": 2.4641904830932617, "learning_rate": 3.8974e-06, "loss": 0.0296, "step": 30514 }, { "epoch": 1.626859305859146, "grad_norm": 1.003035545349121, "learning_rate": 3.8972000000000006e-06, "loss": 0.0199, "step": 30515 }, { "epoch": 1.6269126192887988, "grad_norm": 1.1946158409118652, "learning_rate": 3.897e-06, "loss": 0.0146, "step": 30516 }, { "epoch": 1.6269659327184518, "grad_norm": 1.3786864280700684, "learning_rate": 3.8968000000000005e-06, "loss": 0.0214, "step": 30517 }, { "epoch": 1.6270192461481048, "grad_norm": 4.6906819343566895, "learning_rate": 3.8966e-06, "loss": 0.0118, "step": 30518 }, { "epoch": 1.6270725595777575, "grad_norm": 2.674391984939575, "learning_rate": 3.8964000000000005e-06, "loss": 0.0253, "step": 30519 }, { "epoch": 1.6271258730074105, "grad_norm": 1.5726547241210938, "learning_rate": 3.896200000000001e-06, "loss": 0.0123, "step": 30520 }, { "epoch": 1.6271791864370635, "grad_norm": 1.9633979797363281, "learning_rate": 3.8960000000000005e-06, "loss": 0.0218, "step": 30521 }, { "epoch": 1.6272324998667165, "grad_norm": 1.3079265356063843, "learning_rate": 3.8958e-06, "loss": 0.0126, "step": 30522 }, { "epoch": 1.6272858132963695, "grad_norm": 1.7982629537582397, "learning_rate": 3.8956000000000004e-06, "loss": 0.0119, "step": 30523 }, { "epoch": 1.6273391267260222, "grad_norm": 1.1651421785354614, "learning_rate": 3.8954e-06, "loss": 0.0074, "step": 30524 }, { "epoch": 1.6273924401556752, "grad_norm": 4.134943008422852, "learning_rate": 3.8952e-06, "loss": 0.0227, "step": 30525 }, { "epoch": 1.6274457535853282, "grad_norm": 1.8585866689682007, "learning_rate": 3.895000000000001e-06, "loss": 0.0214, "step": 30526 }, { "epoch": 1.627499067014981, "grad_norm": 4.481227397918701, "learning_rate": 3.8948e-06, "loss": 0.0298, "step": 30527 }, { "epoch": 1.6275523804446341, "grad_norm": 3.820136308670044, "learning_rate": 3.8946e-06, "loss": 0.0269, "step": 30528 }, { "epoch": 1.6276056938742869, "grad_norm": 2.1436233520507812, "learning_rate": 3.8944e-06, "loss": 0.0273, "step": 30529 }, { "epoch": 1.6276590073039399, "grad_norm": 1.279455542564392, "learning_rate": 3.8942e-06, "loss": 0.0211, "step": 30530 }, { "epoch": 1.6277123207335928, "grad_norm": 2.357478618621826, "learning_rate": 3.894e-06, "loss": 0.0136, "step": 30531 }, { "epoch": 1.6277656341632456, "grad_norm": 0.9462786912918091, "learning_rate": 3.893800000000001e-06, "loss": 0.0068, "step": 30532 }, { "epoch": 1.6278189475928988, "grad_norm": 0.15439285337924957, "learning_rate": 3.8936e-06, "loss": 0.0015, "step": 30533 }, { "epoch": 1.6278722610225516, "grad_norm": 2.7430264949798584, "learning_rate": 3.893400000000001e-06, "loss": 0.0318, "step": 30534 }, { "epoch": 1.6279255744522045, "grad_norm": 3.9026527404785156, "learning_rate": 3.8932e-06, "loss": 0.0364, "step": 30535 }, { "epoch": 1.6279788878818575, "grad_norm": 1.5132129192352295, "learning_rate": 3.893e-06, "loss": 0.0446, "step": 30536 }, { "epoch": 1.6280322013115103, "grad_norm": 2.4741384983062744, "learning_rate": 3.8928e-06, "loss": 0.027, "step": 30537 }, { "epoch": 1.6280855147411633, "grad_norm": 1.7789232730865479, "learning_rate": 3.892600000000001e-06, "loss": 0.01, "step": 30538 }, { "epoch": 1.6281388281708162, "grad_norm": 1.7342195510864258, "learning_rate": 3.8924e-06, "loss": 0.0445, "step": 30539 }, { "epoch": 1.6281921416004692, "grad_norm": 3.6139345169067383, "learning_rate": 3.892200000000001e-06, "loss": 0.0124, "step": 30540 }, { "epoch": 1.6282454550301222, "grad_norm": 2.770641803741455, "learning_rate": 3.892e-06, "loss": 0.046, "step": 30541 }, { "epoch": 1.628298768459775, "grad_norm": 1.5071080923080444, "learning_rate": 3.891800000000001e-06, "loss": 0.0092, "step": 30542 }, { "epoch": 1.628352081889428, "grad_norm": 0.2543189525604248, "learning_rate": 3.8916e-06, "loss": 0.0025, "step": 30543 }, { "epoch": 1.628405395319081, "grad_norm": 2.065717935562134, "learning_rate": 3.8914000000000006e-06, "loss": 0.0301, "step": 30544 }, { "epoch": 1.6284587087487337, "grad_norm": 2.3780205249786377, "learning_rate": 3.8912e-06, "loss": 0.028, "step": 30545 }, { "epoch": 1.6285120221783869, "grad_norm": 1.982316017150879, "learning_rate": 3.8910000000000005e-06, "loss": 0.025, "step": 30546 }, { "epoch": 1.6285653356080396, "grad_norm": 1.4980353116989136, "learning_rate": 3.8908e-06, "loss": 0.0174, "step": 30547 }, { "epoch": 1.6286186490376926, "grad_norm": 1.4383195638656616, "learning_rate": 3.8906000000000005e-06, "loss": 0.0199, "step": 30548 }, { "epoch": 1.6286719624673456, "grad_norm": 1.4425241947174072, "learning_rate": 3.8904e-06, "loss": 0.0175, "step": 30549 }, { "epoch": 1.6287252758969983, "grad_norm": 1.13016676902771, "learning_rate": 3.8902000000000005e-06, "loss": 0.0083, "step": 30550 }, { "epoch": 1.6287785893266515, "grad_norm": 0.7130062580108643, "learning_rate": 3.89e-06, "loss": 0.0047, "step": 30551 }, { "epoch": 1.6288319027563043, "grad_norm": 1.984676480293274, "learning_rate": 3.8898000000000004e-06, "loss": 0.0085, "step": 30552 }, { "epoch": 1.6288852161859573, "grad_norm": 1.0950307846069336, "learning_rate": 3.8896e-06, "loss": 0.0167, "step": 30553 }, { "epoch": 1.6289385296156103, "grad_norm": 1.9883290529251099, "learning_rate": 3.8894e-06, "loss": 0.0209, "step": 30554 }, { "epoch": 1.628991843045263, "grad_norm": 1.322186827659607, "learning_rate": 3.8892e-06, "loss": 0.0232, "step": 30555 }, { "epoch": 1.629045156474916, "grad_norm": 0.7122488021850586, "learning_rate": 3.889e-06, "loss": 0.0066, "step": 30556 }, { "epoch": 1.629098469904569, "grad_norm": 1.4826529026031494, "learning_rate": 3.8888e-06, "loss": 0.0192, "step": 30557 }, { "epoch": 1.6291517833342217, "grad_norm": 3.4915614128112793, "learning_rate": 3.8886e-06, "loss": 0.0304, "step": 30558 }, { "epoch": 1.629205096763875, "grad_norm": 0.7238450050354004, "learning_rate": 3.888400000000001e-06, "loss": 0.004, "step": 30559 }, { "epoch": 1.6292584101935277, "grad_norm": 2.8503293991088867, "learning_rate": 3.8882e-06, "loss": 0.0275, "step": 30560 }, { "epoch": 1.6293117236231807, "grad_norm": 0.6788390278816223, "learning_rate": 3.888e-06, "loss": 0.0058, "step": 30561 }, { "epoch": 1.6293650370528336, "grad_norm": 1.2086620330810547, "learning_rate": 3.8878e-06, "loss": 0.0048, "step": 30562 }, { "epoch": 1.6294183504824864, "grad_norm": 1.794166922569275, "learning_rate": 3.8876e-06, "loss": 0.0317, "step": 30563 }, { "epoch": 1.6294716639121396, "grad_norm": 1.024389386177063, "learning_rate": 3.8874e-06, "loss": 0.0062, "step": 30564 }, { "epoch": 1.6295249773417924, "grad_norm": 1.1362024545669556, "learning_rate": 3.887200000000001e-06, "loss": 0.0149, "step": 30565 }, { "epoch": 1.6295782907714453, "grad_norm": 3.0980544090270996, "learning_rate": 3.887e-06, "loss": 0.0202, "step": 30566 }, { "epoch": 1.6296316042010983, "grad_norm": 1.7773464918136597, "learning_rate": 3.886800000000001e-06, "loss": 0.0127, "step": 30567 }, { "epoch": 1.629684917630751, "grad_norm": 1.799588918685913, "learning_rate": 3.8866e-06, "loss": 0.0335, "step": 30568 }, { "epoch": 1.6297382310604043, "grad_norm": 1.488699197769165, "learning_rate": 3.8864e-06, "loss": 0.0183, "step": 30569 }, { "epoch": 1.629791544490057, "grad_norm": 1.237660527229309, "learning_rate": 3.8862e-06, "loss": 0.0106, "step": 30570 }, { "epoch": 1.62984485791971, "grad_norm": 1.044006109237671, "learning_rate": 3.8860000000000006e-06, "loss": 0.0203, "step": 30571 }, { "epoch": 1.629898171349363, "grad_norm": 2.034540891647339, "learning_rate": 3.8858e-06, "loss": 0.0139, "step": 30572 }, { "epoch": 1.6299514847790157, "grad_norm": 1.2844407558441162, "learning_rate": 3.8856000000000005e-06, "loss": 0.0063, "step": 30573 }, { "epoch": 1.6300047982086687, "grad_norm": 3.5230679512023926, "learning_rate": 3.8854e-06, "loss": 0.0469, "step": 30574 }, { "epoch": 1.6300581116383217, "grad_norm": 1.029448390007019, "learning_rate": 3.8852e-06, "loss": 0.0063, "step": 30575 }, { "epoch": 1.6301114250679745, "grad_norm": 1.3585739135742188, "learning_rate": 3.885e-06, "loss": 0.0132, "step": 30576 }, { "epoch": 1.6301647384976277, "grad_norm": 1.394797921180725, "learning_rate": 3.8848000000000005e-06, "loss": 0.0102, "step": 30577 }, { "epoch": 1.6302180519272804, "grad_norm": 3.6792263984680176, "learning_rate": 3.884600000000001e-06, "loss": 0.028, "step": 30578 }, { "epoch": 1.6302713653569334, "grad_norm": 2.146766424179077, "learning_rate": 3.8844000000000005e-06, "loss": 0.0205, "step": 30579 }, { "epoch": 1.6303246787865864, "grad_norm": 0.9443175196647644, "learning_rate": 3.8842e-06, "loss": 0.0117, "step": 30580 }, { "epoch": 1.6303779922162391, "grad_norm": 1.2788333892822266, "learning_rate": 3.884e-06, "loss": 0.0113, "step": 30581 }, { "epoch": 1.6304313056458923, "grad_norm": 1.9977085590362549, "learning_rate": 3.8838e-06, "loss": 0.0252, "step": 30582 }, { "epoch": 1.630484619075545, "grad_norm": 1.6152931451797485, "learning_rate": 3.8836e-06, "loss": 0.0294, "step": 30583 }, { "epoch": 1.630537932505198, "grad_norm": 1.517998456954956, "learning_rate": 3.883400000000001e-06, "loss": 0.0156, "step": 30584 }, { "epoch": 1.630591245934851, "grad_norm": 2.381208896636963, "learning_rate": 3.8832e-06, "loss": 0.021, "step": 30585 }, { "epoch": 1.6306445593645038, "grad_norm": 0.7891136407852173, "learning_rate": 3.883e-06, "loss": 0.0053, "step": 30586 }, { "epoch": 1.630697872794157, "grad_norm": 2.075310468673706, "learning_rate": 3.8828e-06, "loss": 0.0344, "step": 30587 }, { "epoch": 1.6307511862238098, "grad_norm": 3.0977113246917725, "learning_rate": 3.8826e-06, "loss": 0.0319, "step": 30588 }, { "epoch": 1.6308044996534627, "grad_norm": 2.167595386505127, "learning_rate": 3.8824e-06, "loss": 0.0227, "step": 30589 }, { "epoch": 1.6308578130831157, "grad_norm": 2.5131821632385254, "learning_rate": 3.882200000000001e-06, "loss": 0.0199, "step": 30590 }, { "epoch": 1.6309111265127685, "grad_norm": 2.6804769039154053, "learning_rate": 3.882e-06, "loss": 0.0229, "step": 30591 }, { "epoch": 1.6309644399424215, "grad_norm": 2.3366611003875732, "learning_rate": 3.881800000000001e-06, "loss": 0.0216, "step": 30592 }, { "epoch": 1.6310177533720744, "grad_norm": 4.699663162231445, "learning_rate": 3.8816e-06, "loss": 0.0458, "step": 30593 }, { "epoch": 1.6310710668017272, "grad_norm": 2.3746399879455566, "learning_rate": 3.8814e-06, "loss": 0.0336, "step": 30594 }, { "epoch": 1.6311243802313804, "grad_norm": 2.447784662246704, "learning_rate": 3.8812e-06, "loss": 0.0152, "step": 30595 }, { "epoch": 1.6311776936610332, "grad_norm": 1.4763387441635132, "learning_rate": 3.881000000000001e-06, "loss": 0.0178, "step": 30596 }, { "epoch": 1.6312310070906861, "grad_norm": 2.327855110168457, "learning_rate": 3.8808e-06, "loss": 0.0299, "step": 30597 }, { "epoch": 1.6312843205203391, "grad_norm": 1.0224683284759521, "learning_rate": 3.880600000000001e-06, "loss": 0.0095, "step": 30598 }, { "epoch": 1.6313376339499919, "grad_norm": 2.337137222290039, "learning_rate": 3.8804e-06, "loss": 0.0194, "step": 30599 }, { "epoch": 1.631390947379645, "grad_norm": 2.2321062088012695, "learning_rate": 3.8802e-06, "loss": 0.0278, "step": 30600 }, { "epoch": 1.6314442608092978, "grad_norm": 1.358018398284912, "learning_rate": 3.88e-06, "loss": 0.0187, "step": 30601 }, { "epoch": 1.6314975742389508, "grad_norm": 3.5646543502807617, "learning_rate": 3.8798000000000005e-06, "loss": 0.0245, "step": 30602 }, { "epoch": 1.6315508876686038, "grad_norm": 2.334672451019287, "learning_rate": 3.879600000000001e-06, "loss": 0.0324, "step": 30603 }, { "epoch": 1.6316042010982565, "grad_norm": 1.2430235147476196, "learning_rate": 3.8794000000000005e-06, "loss": 0.0207, "step": 30604 }, { "epoch": 1.6316575145279095, "grad_norm": 1.6093229055404663, "learning_rate": 3.8792e-06, "loss": 0.0233, "step": 30605 }, { "epoch": 1.6317108279575625, "grad_norm": 1.9963405132293701, "learning_rate": 3.8790000000000005e-06, "loss": 0.0109, "step": 30606 }, { "epoch": 1.6317641413872155, "grad_norm": 3.2102279663085938, "learning_rate": 3.8788e-06, "loss": 0.0372, "step": 30607 }, { "epoch": 1.6318174548168685, "grad_norm": 2.410702705383301, "learning_rate": 3.8786000000000004e-06, "loss": 0.0199, "step": 30608 }, { "epoch": 1.6318707682465212, "grad_norm": 1.9609522819519043, "learning_rate": 3.878400000000001e-06, "loss": 0.0356, "step": 30609 }, { "epoch": 1.6319240816761742, "grad_norm": 0.956951916217804, "learning_rate": 3.8782e-06, "loss": 0.0186, "step": 30610 }, { "epoch": 1.6319773951058272, "grad_norm": 1.5935471057891846, "learning_rate": 3.878e-06, "loss": 0.0134, "step": 30611 }, { "epoch": 1.63203070853548, "grad_norm": 1.6572411060333252, "learning_rate": 3.8778e-06, "loss": 0.0089, "step": 30612 }, { "epoch": 1.6320840219651331, "grad_norm": 2.139749526977539, "learning_rate": 3.8776e-06, "loss": 0.0279, "step": 30613 }, { "epoch": 1.632137335394786, "grad_norm": 2.576925754547119, "learning_rate": 3.8774e-06, "loss": 0.0225, "step": 30614 }, { "epoch": 1.6321906488244389, "grad_norm": 0.6172784566879272, "learning_rate": 3.877200000000001e-06, "loss": 0.0045, "step": 30615 }, { "epoch": 1.6322439622540919, "grad_norm": 6.39256477355957, "learning_rate": 3.877e-06, "loss": 0.0147, "step": 30616 }, { "epoch": 1.6322972756837446, "grad_norm": 0.6926705241203308, "learning_rate": 3.876800000000001e-06, "loss": 0.0175, "step": 30617 }, { "epoch": 1.6323505891133978, "grad_norm": 1.22100830078125, "learning_rate": 3.8766e-06, "loss": 0.0107, "step": 30618 }, { "epoch": 1.6324039025430506, "grad_norm": 1.3994288444519043, "learning_rate": 3.8764e-06, "loss": 0.0223, "step": 30619 }, { "epoch": 1.6324572159727035, "grad_norm": 1.7035168409347534, "learning_rate": 3.8762e-06, "loss": 0.023, "step": 30620 }, { "epoch": 1.6325105294023565, "grad_norm": 1.0569723844528198, "learning_rate": 3.876000000000001e-06, "loss": 0.0069, "step": 30621 }, { "epoch": 1.6325638428320093, "grad_norm": 0.7425852417945862, "learning_rate": 3.8758e-06, "loss": 0.0153, "step": 30622 }, { "epoch": 1.6326171562616623, "grad_norm": 2.2045018672943115, "learning_rate": 3.875600000000001e-06, "loss": 0.0126, "step": 30623 }, { "epoch": 1.6326704696913152, "grad_norm": 3.2047393321990967, "learning_rate": 3.8754e-06, "loss": 0.0272, "step": 30624 }, { "epoch": 1.6327237831209682, "grad_norm": 0.6525850892066956, "learning_rate": 3.8752e-06, "loss": 0.0042, "step": 30625 }, { "epoch": 1.6327770965506212, "grad_norm": 1.6973659992218018, "learning_rate": 3.875e-06, "loss": 0.0355, "step": 30626 }, { "epoch": 1.632830409980274, "grad_norm": 2.3737070560455322, "learning_rate": 3.874800000000001e-06, "loss": 0.0323, "step": 30627 }, { "epoch": 1.632883723409927, "grad_norm": 2.201129198074341, "learning_rate": 3.874600000000001e-06, "loss": 0.0187, "step": 30628 }, { "epoch": 1.63293703683958, "grad_norm": 0.736431360244751, "learning_rate": 3.8744000000000005e-06, "loss": 0.0046, "step": 30629 }, { "epoch": 1.6329903502692327, "grad_norm": 1.4672837257385254, "learning_rate": 3.8742e-06, "loss": 0.0125, "step": 30630 }, { "epoch": 1.6330436636988859, "grad_norm": 1.4240529537200928, "learning_rate": 3.8740000000000005e-06, "loss": 0.0292, "step": 30631 }, { "epoch": 1.6330969771285386, "grad_norm": 1.4052242040634155, "learning_rate": 3.8738e-06, "loss": 0.0191, "step": 30632 }, { "epoch": 1.6331502905581916, "grad_norm": 2.620121717453003, "learning_rate": 3.8736000000000005e-06, "loss": 0.0278, "step": 30633 }, { "epoch": 1.6332036039878446, "grad_norm": 0.8357731103897095, "learning_rate": 3.873400000000001e-06, "loss": 0.0075, "step": 30634 }, { "epoch": 1.6332569174174973, "grad_norm": 1.7946079969406128, "learning_rate": 3.8732000000000005e-06, "loss": 0.0241, "step": 30635 }, { "epoch": 1.6333102308471505, "grad_norm": 1.182384729385376, "learning_rate": 3.873e-06, "loss": 0.0158, "step": 30636 }, { "epoch": 1.6333635442768033, "grad_norm": 3.7313108444213867, "learning_rate": 3.8728000000000004e-06, "loss": 0.0184, "step": 30637 }, { "epoch": 1.6334168577064563, "grad_norm": 2.8380651473999023, "learning_rate": 3.8726e-06, "loss": 0.0186, "step": 30638 }, { "epoch": 1.6334701711361093, "grad_norm": 2.852843761444092, "learning_rate": 3.8724e-06, "loss": 0.014, "step": 30639 }, { "epoch": 1.633523484565762, "grad_norm": 0.775017499923706, "learning_rate": 3.872200000000001e-06, "loss": 0.0055, "step": 30640 }, { "epoch": 1.633576797995415, "grad_norm": 1.5776036977767944, "learning_rate": 3.872e-06, "loss": 0.0179, "step": 30641 }, { "epoch": 1.633630111425068, "grad_norm": 1.6442255973815918, "learning_rate": 3.871800000000001e-06, "loss": 0.0114, "step": 30642 }, { "epoch": 1.633683424854721, "grad_norm": 1.8659368753433228, "learning_rate": 3.8716e-06, "loss": 0.0362, "step": 30643 }, { "epoch": 1.633736738284374, "grad_norm": 0.4248153269290924, "learning_rate": 3.8714e-06, "loss": 0.0037, "step": 30644 }, { "epoch": 1.6337900517140267, "grad_norm": 0.8382940292358398, "learning_rate": 3.8712e-06, "loss": 0.0047, "step": 30645 }, { "epoch": 1.6338433651436797, "grad_norm": 0.8875148892402649, "learning_rate": 3.871000000000001e-06, "loss": 0.0057, "step": 30646 }, { "epoch": 1.6338966785733327, "grad_norm": 0.6067337989807129, "learning_rate": 3.8708e-06, "loss": 0.0043, "step": 30647 }, { "epoch": 1.6339499920029854, "grad_norm": 1.373414158821106, "learning_rate": 3.870600000000001e-06, "loss": 0.0311, "step": 30648 }, { "epoch": 1.6340033054326386, "grad_norm": 1.0393751859664917, "learning_rate": 3.8704e-06, "loss": 0.0057, "step": 30649 }, { "epoch": 1.6340566188622914, "grad_norm": 0.9816669821739197, "learning_rate": 3.8702e-06, "loss": 0.0064, "step": 30650 }, { "epoch": 1.6341099322919443, "grad_norm": 1.1681236028671265, "learning_rate": 3.87e-06, "loss": 0.0083, "step": 30651 }, { "epoch": 1.6341632457215973, "grad_norm": 1.8444589376449585, "learning_rate": 3.869800000000001e-06, "loss": 0.0365, "step": 30652 }, { "epoch": 1.63421655915125, "grad_norm": 3.4552783966064453, "learning_rate": 3.8696e-06, "loss": 0.0472, "step": 30653 }, { "epoch": 1.6342698725809033, "grad_norm": 3.3869307041168213, "learning_rate": 3.869400000000001e-06, "loss": 0.0394, "step": 30654 }, { "epoch": 1.634323186010556, "grad_norm": 1.1127749681472778, "learning_rate": 3.8692e-06, "loss": 0.0395, "step": 30655 }, { "epoch": 1.634376499440209, "grad_norm": 1.1106042861938477, "learning_rate": 3.8690000000000006e-06, "loss": 0.0197, "step": 30656 }, { "epoch": 1.634429812869862, "grad_norm": 2.1935014724731445, "learning_rate": 3.8688e-06, "loss": 0.025, "step": 30657 }, { "epoch": 1.6344831262995148, "grad_norm": 1.0891972780227661, "learning_rate": 3.8686000000000005e-06, "loss": 0.0161, "step": 30658 }, { "epoch": 1.6345364397291677, "grad_norm": 1.4223419427871704, "learning_rate": 3.8684e-06, "loss": 0.019, "step": 30659 }, { "epoch": 1.6345897531588207, "grad_norm": 2.0006825923919678, "learning_rate": 3.8682000000000005e-06, "loss": 0.0146, "step": 30660 }, { "epoch": 1.6346430665884735, "grad_norm": 1.8110277652740479, "learning_rate": 3.868e-06, "loss": 0.0234, "step": 30661 }, { "epoch": 1.6346963800181267, "grad_norm": 1.270390510559082, "learning_rate": 3.8678000000000005e-06, "loss": 0.0081, "step": 30662 }, { "epoch": 1.6347496934477794, "grad_norm": 1.7161306142807007, "learning_rate": 3.8676e-06, "loss": 0.0169, "step": 30663 }, { "epoch": 1.6348030068774324, "grad_norm": 2.9389045238494873, "learning_rate": 3.8674000000000004e-06, "loss": 0.0253, "step": 30664 }, { "epoch": 1.6348563203070854, "grad_norm": 2.095337390899658, "learning_rate": 3.8672e-06, "loss": 0.0336, "step": 30665 }, { "epoch": 1.6349096337367381, "grad_norm": 1.7783799171447754, "learning_rate": 3.867e-06, "loss": 0.0468, "step": 30666 }, { "epoch": 1.6349629471663913, "grad_norm": 1.5208969116210938, "learning_rate": 3.866800000000001e-06, "loss": 0.0268, "step": 30667 }, { "epoch": 1.635016260596044, "grad_norm": 1.8590211868286133, "learning_rate": 3.8666e-06, "loss": 0.0134, "step": 30668 }, { "epoch": 1.635069574025697, "grad_norm": 4.992746353149414, "learning_rate": 3.8664e-06, "loss": 0.0341, "step": 30669 }, { "epoch": 1.63512288745535, "grad_norm": 1.5221905708312988, "learning_rate": 3.8662e-06, "loss": 0.0165, "step": 30670 }, { "epoch": 1.6351762008850028, "grad_norm": 2.381805896759033, "learning_rate": 3.866e-06, "loss": 0.0158, "step": 30671 }, { "epoch": 1.635229514314656, "grad_norm": 2.704282283782959, "learning_rate": 3.8658e-06, "loss": 0.0334, "step": 30672 }, { "epoch": 1.6352828277443088, "grad_norm": 2.169926643371582, "learning_rate": 3.865600000000001e-06, "loss": 0.0287, "step": 30673 }, { "epoch": 1.6353361411739618, "grad_norm": 2.125561237335205, "learning_rate": 3.8654e-06, "loss": 0.0217, "step": 30674 }, { "epoch": 1.6353894546036147, "grad_norm": 0.9457805752754211, "learning_rate": 3.8652e-06, "loss": 0.0166, "step": 30675 }, { "epoch": 1.6354427680332675, "grad_norm": 3.0630364418029785, "learning_rate": 3.865e-06, "loss": 0.0328, "step": 30676 }, { "epoch": 1.6354960814629205, "grad_norm": 0.9770970940589905, "learning_rate": 3.8648e-06, "loss": 0.0068, "step": 30677 }, { "epoch": 1.6355493948925734, "grad_norm": 1.4185545444488525, "learning_rate": 3.8646e-06, "loss": 0.0187, "step": 30678 }, { "epoch": 1.6356027083222262, "grad_norm": 1.4402618408203125, "learning_rate": 3.864400000000001e-06, "loss": 0.014, "step": 30679 }, { "epoch": 1.6356560217518794, "grad_norm": 3.570796012878418, "learning_rate": 3.8642e-06, "loss": 0.0425, "step": 30680 }, { "epoch": 1.6357093351815322, "grad_norm": 0.8432741761207581, "learning_rate": 3.864000000000001e-06, "loss": 0.0069, "step": 30681 }, { "epoch": 1.6357626486111851, "grad_norm": 0.5379603505134583, "learning_rate": 3.8638e-06, "loss": 0.0048, "step": 30682 }, { "epoch": 1.6358159620408381, "grad_norm": 0.6773507595062256, "learning_rate": 3.8636e-06, "loss": 0.0143, "step": 30683 }, { "epoch": 1.6358692754704909, "grad_norm": 0.8678470253944397, "learning_rate": 3.8634e-06, "loss": 0.0147, "step": 30684 }, { "epoch": 1.635922588900144, "grad_norm": 2.3493173122406006, "learning_rate": 3.8632000000000006e-06, "loss": 0.0258, "step": 30685 }, { "epoch": 1.6359759023297968, "grad_norm": 2.0533604621887207, "learning_rate": 3.863e-06, "loss": 0.0493, "step": 30686 }, { "epoch": 1.6360292157594498, "grad_norm": 0.654902458190918, "learning_rate": 3.8628000000000005e-06, "loss": 0.0052, "step": 30687 }, { "epoch": 1.6360825291891028, "grad_norm": 0.6594573855400085, "learning_rate": 3.8626e-06, "loss": 0.0085, "step": 30688 }, { "epoch": 1.6361358426187556, "grad_norm": 1.8317911624908447, "learning_rate": 3.8624000000000005e-06, "loss": 0.0239, "step": 30689 }, { "epoch": 1.6361891560484088, "grad_norm": 2.152270555496216, "learning_rate": 3.8622e-06, "loss": 0.0333, "step": 30690 }, { "epoch": 1.6362424694780615, "grad_norm": 0.39639419317245483, "learning_rate": 3.8620000000000005e-06, "loss": 0.0034, "step": 30691 }, { "epoch": 1.6362957829077145, "grad_norm": 2.141893148422241, "learning_rate": 3.861800000000001e-06, "loss": 0.0242, "step": 30692 }, { "epoch": 1.6363490963373675, "grad_norm": 2.111016273498535, "learning_rate": 3.8616000000000004e-06, "loss": 0.0393, "step": 30693 }, { "epoch": 1.6364024097670202, "grad_norm": 1.5402320623397827, "learning_rate": 3.8614e-06, "loss": 0.0169, "step": 30694 }, { "epoch": 1.6364557231966732, "grad_norm": 0.6218575239181519, "learning_rate": 3.8612e-06, "loss": 0.0049, "step": 30695 }, { "epoch": 1.6365090366263262, "grad_norm": 1.550429344177246, "learning_rate": 3.861e-06, "loss": 0.027, "step": 30696 }, { "epoch": 1.636562350055979, "grad_norm": 1.3900364637374878, "learning_rate": 3.8608e-06, "loss": 0.015, "step": 30697 }, { "epoch": 1.6366156634856321, "grad_norm": 1.3369908332824707, "learning_rate": 3.860600000000001e-06, "loss": 0.0198, "step": 30698 }, { "epoch": 1.636668976915285, "grad_norm": 1.3229268789291382, "learning_rate": 3.8604e-06, "loss": 0.0103, "step": 30699 }, { "epoch": 1.6367222903449379, "grad_norm": 1.679588794708252, "learning_rate": 3.8602e-06, "loss": 0.0289, "step": 30700 }, { "epoch": 1.6367756037745909, "grad_norm": 1.678393006324768, "learning_rate": 3.86e-06, "loss": 0.0146, "step": 30701 }, { "epoch": 1.6368289172042436, "grad_norm": 1.6205216646194458, "learning_rate": 3.8598e-06, "loss": 0.0323, "step": 30702 }, { "epoch": 1.6368822306338968, "grad_norm": 2.0438270568847656, "learning_rate": 3.8596e-06, "loss": 0.0281, "step": 30703 }, { "epoch": 1.6369355440635496, "grad_norm": 2.5347273349761963, "learning_rate": 3.859400000000001e-06, "loss": 0.0156, "step": 30704 }, { "epoch": 1.6369888574932026, "grad_norm": 1.924303650856018, "learning_rate": 3.8592e-06, "loss": 0.0288, "step": 30705 }, { "epoch": 1.6370421709228555, "grad_norm": 1.5508499145507812, "learning_rate": 3.859000000000001e-06, "loss": 0.0203, "step": 30706 }, { "epoch": 1.6370954843525083, "grad_norm": 2.224961996078491, "learning_rate": 3.8588e-06, "loss": 0.017, "step": 30707 }, { "epoch": 1.6371487977821615, "grad_norm": 1.322113037109375, "learning_rate": 3.8586e-06, "loss": 0.0318, "step": 30708 }, { "epoch": 1.6372021112118142, "grad_norm": 1.021201252937317, "learning_rate": 3.8584e-06, "loss": 0.0089, "step": 30709 }, { "epoch": 1.6372554246414672, "grad_norm": 2.6145761013031006, "learning_rate": 3.858200000000001e-06, "loss": 0.0332, "step": 30710 }, { "epoch": 1.6373087380711202, "grad_norm": 1.6599664688110352, "learning_rate": 3.858e-06, "loss": 0.031, "step": 30711 }, { "epoch": 1.637362051500773, "grad_norm": 3.787875175476074, "learning_rate": 3.8578000000000006e-06, "loss": 0.0154, "step": 30712 }, { "epoch": 1.637415364930426, "grad_norm": 1.1379859447479248, "learning_rate": 3.8576e-06, "loss": 0.0089, "step": 30713 }, { "epoch": 1.637468678360079, "grad_norm": 2.4063422679901123, "learning_rate": 3.8574000000000005e-06, "loss": 0.0235, "step": 30714 }, { "epoch": 1.6375219917897317, "grad_norm": 1.4605321884155273, "learning_rate": 3.8572e-06, "loss": 0.0124, "step": 30715 }, { "epoch": 1.6375753052193849, "grad_norm": 1.5219894647598267, "learning_rate": 3.8570000000000005e-06, "loss": 0.0289, "step": 30716 }, { "epoch": 1.6376286186490376, "grad_norm": 1.8150978088378906, "learning_rate": 3.856800000000001e-06, "loss": 0.0223, "step": 30717 }, { "epoch": 1.6376819320786906, "grad_norm": 2.7301156520843506, "learning_rate": 3.8566000000000005e-06, "loss": 0.0332, "step": 30718 }, { "epoch": 1.6377352455083436, "grad_norm": 0.7102379202842712, "learning_rate": 3.8564e-06, "loss": 0.0162, "step": 30719 }, { "epoch": 1.6377885589379964, "grad_norm": 2.144036293029785, "learning_rate": 3.8562000000000005e-06, "loss": 0.0095, "step": 30720 }, { "epoch": 1.6378418723676496, "grad_norm": 2.294466018676758, "learning_rate": 3.856e-06, "loss": 0.0522, "step": 30721 }, { "epoch": 1.6378951857973023, "grad_norm": 2.592122793197632, "learning_rate": 3.8558e-06, "loss": 0.0204, "step": 30722 }, { "epoch": 1.6379484992269553, "grad_norm": 2.7783634662628174, "learning_rate": 3.855600000000001e-06, "loss": 0.0189, "step": 30723 }, { "epoch": 1.6380018126566083, "grad_norm": 2.591876268386841, "learning_rate": 3.8554e-06, "loss": 0.0258, "step": 30724 }, { "epoch": 1.638055126086261, "grad_norm": 3.2181811332702637, "learning_rate": 3.8552e-06, "loss": 0.0137, "step": 30725 }, { "epoch": 1.638108439515914, "grad_norm": 1.197547435760498, "learning_rate": 3.855e-06, "loss": 0.0207, "step": 30726 }, { "epoch": 1.638161752945567, "grad_norm": 1.561856985092163, "learning_rate": 3.8548e-06, "loss": 0.0173, "step": 30727 }, { "epoch": 1.63821506637522, "grad_norm": 2.44429612159729, "learning_rate": 3.8546e-06, "loss": 0.031, "step": 30728 }, { "epoch": 1.638268379804873, "grad_norm": 1.2669165134429932, "learning_rate": 3.854400000000001e-06, "loss": 0.0163, "step": 30729 }, { "epoch": 1.6383216932345257, "grad_norm": 1.140801191329956, "learning_rate": 3.8542e-06, "loss": 0.025, "step": 30730 }, { "epoch": 1.6383750066641787, "grad_norm": 1.3272732496261597, "learning_rate": 3.854000000000001e-06, "loss": 0.0233, "step": 30731 }, { "epoch": 1.6384283200938317, "grad_norm": 1.3410074710845947, "learning_rate": 3.8538e-06, "loss": 0.0145, "step": 30732 }, { "epoch": 1.6384816335234844, "grad_norm": 1.5684608221054077, "learning_rate": 3.8536e-06, "loss": 0.0295, "step": 30733 }, { "epoch": 1.6385349469531376, "grad_norm": 1.6173791885375977, "learning_rate": 3.8534e-06, "loss": 0.0346, "step": 30734 }, { "epoch": 1.6385882603827904, "grad_norm": 1.6393136978149414, "learning_rate": 3.853200000000001e-06, "loss": 0.0101, "step": 30735 }, { "epoch": 1.6386415738124434, "grad_norm": 2.58001708984375, "learning_rate": 3.853e-06, "loss": 0.0519, "step": 30736 }, { "epoch": 1.6386948872420963, "grad_norm": 0.5776179432868958, "learning_rate": 3.852800000000001e-06, "loss": 0.0049, "step": 30737 }, { "epoch": 1.638748200671749, "grad_norm": 1.2367106676101685, "learning_rate": 3.8526e-06, "loss": 0.0269, "step": 30738 }, { "epoch": 1.6388015141014023, "grad_norm": 0.6938506960868835, "learning_rate": 3.852400000000001e-06, "loss": 0.0054, "step": 30739 }, { "epoch": 1.638854827531055, "grad_norm": 0.7990328669548035, "learning_rate": 3.8522e-06, "loss": 0.0137, "step": 30740 }, { "epoch": 1.638908140960708, "grad_norm": 0.3224376440048218, "learning_rate": 3.8520000000000006e-06, "loss": 0.0037, "step": 30741 }, { "epoch": 1.638961454390361, "grad_norm": 1.007562279701233, "learning_rate": 3.8518e-06, "loss": 0.0128, "step": 30742 }, { "epoch": 1.6390147678200138, "grad_norm": 0.8475773930549622, "learning_rate": 3.8516000000000005e-06, "loss": 0.0064, "step": 30743 }, { "epoch": 1.6390680812496667, "grad_norm": 3.2492878437042236, "learning_rate": 3.8514e-06, "loss": 0.0236, "step": 30744 }, { "epoch": 1.6391213946793197, "grad_norm": 4.244361400604248, "learning_rate": 3.8512000000000005e-06, "loss": 0.0456, "step": 30745 }, { "epoch": 1.6391747081089727, "grad_norm": 1.2784759998321533, "learning_rate": 3.851e-06, "loss": 0.0121, "step": 30746 }, { "epoch": 1.6392280215386257, "grad_norm": 2.0759241580963135, "learning_rate": 3.8508000000000005e-06, "loss": 0.0383, "step": 30747 }, { "epoch": 1.6392813349682784, "grad_norm": 2.4066638946533203, "learning_rate": 3.8506e-06, "loss": 0.0195, "step": 30748 }, { "epoch": 1.6393346483979314, "grad_norm": 2.052945852279663, "learning_rate": 3.8504000000000004e-06, "loss": 0.0162, "step": 30749 }, { "epoch": 1.6393879618275844, "grad_norm": 2.1781280040740967, "learning_rate": 3.8502e-06, "loss": 0.0244, "step": 30750 }, { "epoch": 1.6394412752572372, "grad_norm": 1.9703634977340698, "learning_rate": 3.85e-06, "loss": 0.0228, "step": 30751 }, { "epoch": 1.6394945886868904, "grad_norm": 4.9364824295043945, "learning_rate": 3.8498e-06, "loss": 0.0249, "step": 30752 }, { "epoch": 1.639547902116543, "grad_norm": 1.3538175821304321, "learning_rate": 3.8496e-06, "loss": 0.024, "step": 30753 }, { "epoch": 1.639601215546196, "grad_norm": 2.2680716514587402, "learning_rate": 3.8494e-06, "loss": 0.0448, "step": 30754 }, { "epoch": 1.639654528975849, "grad_norm": 1.5513064861297607, "learning_rate": 3.8492e-06, "loss": 0.012, "step": 30755 }, { "epoch": 1.6397078424055018, "grad_norm": 1.3196474313735962, "learning_rate": 3.849000000000001e-06, "loss": 0.0106, "step": 30756 }, { "epoch": 1.639761155835155, "grad_norm": 3.949793815612793, "learning_rate": 3.8488e-06, "loss": 0.0161, "step": 30757 }, { "epoch": 1.6398144692648078, "grad_norm": 1.9875415563583374, "learning_rate": 3.8486e-06, "loss": 0.0256, "step": 30758 }, { "epoch": 1.6398677826944608, "grad_norm": 0.5931718945503235, "learning_rate": 3.8484e-06, "loss": 0.0038, "step": 30759 }, { "epoch": 1.6399210961241137, "grad_norm": 1.0512561798095703, "learning_rate": 3.8482e-06, "loss": 0.0114, "step": 30760 }, { "epoch": 1.6399744095537665, "grad_norm": 1.28273606300354, "learning_rate": 3.848e-06, "loss": 0.0128, "step": 30761 }, { "epoch": 1.6400277229834195, "grad_norm": 1.844979166984558, "learning_rate": 3.847800000000001e-06, "loss": 0.027, "step": 30762 }, { "epoch": 1.6400810364130725, "grad_norm": 1.5981477499008179, "learning_rate": 3.8476e-06, "loss": 0.0192, "step": 30763 }, { "epoch": 1.6401343498427254, "grad_norm": 1.7592849731445312, "learning_rate": 3.847400000000001e-06, "loss": 0.0114, "step": 30764 }, { "epoch": 1.6401876632723784, "grad_norm": 2.2301652431488037, "learning_rate": 3.8472e-06, "loss": 0.0272, "step": 30765 }, { "epoch": 1.6402409767020312, "grad_norm": 2.0694921016693115, "learning_rate": 3.847e-06, "loss": 0.0311, "step": 30766 }, { "epoch": 1.6402942901316842, "grad_norm": 1.3921300172805786, "learning_rate": 3.8468e-06, "loss": 0.0216, "step": 30767 }, { "epoch": 1.6403476035613371, "grad_norm": 2.1404311656951904, "learning_rate": 3.8466000000000006e-06, "loss": 0.0252, "step": 30768 }, { "epoch": 1.6404009169909899, "grad_norm": 1.663044810295105, "learning_rate": 3.8464e-06, "loss": 0.0241, "step": 30769 }, { "epoch": 1.640454230420643, "grad_norm": 2.3611655235290527, "learning_rate": 3.8462000000000005e-06, "loss": 0.0244, "step": 30770 }, { "epoch": 1.6405075438502958, "grad_norm": 1.3679795265197754, "learning_rate": 3.846e-06, "loss": 0.0094, "step": 30771 }, { "epoch": 1.6405608572799488, "grad_norm": 1.747767686843872, "learning_rate": 3.8458e-06, "loss": 0.0121, "step": 30772 }, { "epoch": 1.6406141707096018, "grad_norm": 2.567556619644165, "learning_rate": 3.8456e-06, "loss": 0.0258, "step": 30773 }, { "epoch": 1.6406674841392546, "grad_norm": 2.1740500926971436, "learning_rate": 3.8454000000000005e-06, "loss": 0.0516, "step": 30774 }, { "epoch": 1.6407207975689078, "grad_norm": 1.3346112966537476, "learning_rate": 3.8452e-06, "loss": 0.02, "step": 30775 }, { "epoch": 1.6407741109985605, "grad_norm": 1.4730054140090942, "learning_rate": 3.8450000000000005e-06, "loss": 0.0102, "step": 30776 }, { "epoch": 1.6408274244282135, "grad_norm": 1.3069661855697632, "learning_rate": 3.8448e-06, "loss": 0.0186, "step": 30777 }, { "epoch": 1.6408807378578665, "grad_norm": 1.5991559028625488, "learning_rate": 3.8446000000000004e-06, "loss": 0.0102, "step": 30778 }, { "epoch": 1.6409340512875192, "grad_norm": 3.1483497619628906, "learning_rate": 3.8444e-06, "loss": 0.0206, "step": 30779 }, { "epoch": 1.6409873647171722, "grad_norm": 1.2267802953720093, "learning_rate": 3.8442e-06, "loss": 0.0173, "step": 30780 }, { "epoch": 1.6410406781468252, "grad_norm": 2.2321112155914307, "learning_rate": 3.844000000000001e-06, "loss": 0.0232, "step": 30781 }, { "epoch": 1.641093991576478, "grad_norm": 1.8322534561157227, "learning_rate": 3.8438e-06, "loss": 0.0266, "step": 30782 }, { "epoch": 1.6411473050061312, "grad_norm": 1.5038200616836548, "learning_rate": 3.8436e-06, "loss": 0.0202, "step": 30783 }, { "epoch": 1.641200618435784, "grad_norm": 2.336745023727417, "learning_rate": 3.8434e-06, "loss": 0.0227, "step": 30784 }, { "epoch": 1.6412539318654369, "grad_norm": 2.657356023788452, "learning_rate": 3.8432e-06, "loss": 0.043, "step": 30785 }, { "epoch": 1.6413072452950899, "grad_norm": 1.4454904794692993, "learning_rate": 3.843e-06, "loss": 0.0331, "step": 30786 }, { "epoch": 1.6413605587247426, "grad_norm": 2.0369064807891846, "learning_rate": 3.842800000000001e-06, "loss": 0.0353, "step": 30787 }, { "epoch": 1.6414138721543958, "grad_norm": 0.9056922793388367, "learning_rate": 3.8426e-06, "loss": 0.0152, "step": 30788 }, { "epoch": 1.6414671855840486, "grad_norm": 2.286257028579712, "learning_rate": 3.842400000000001e-06, "loss": 0.0278, "step": 30789 }, { "epoch": 1.6415204990137016, "grad_norm": 3.271606922149658, "learning_rate": 3.8422e-06, "loss": 0.0365, "step": 30790 }, { "epoch": 1.6415738124433545, "grad_norm": 2.709735870361328, "learning_rate": 3.842e-06, "loss": 0.0314, "step": 30791 }, { "epoch": 1.6416271258730073, "grad_norm": 1.2696982622146606, "learning_rate": 3.8418e-06, "loss": 0.0105, "step": 30792 }, { "epoch": 1.6416804393026605, "grad_norm": 1.4795485734939575, "learning_rate": 3.841600000000001e-06, "loss": 0.0182, "step": 30793 }, { "epoch": 1.6417337527323133, "grad_norm": 0.9652391672134399, "learning_rate": 3.8414e-06, "loss": 0.0067, "step": 30794 }, { "epoch": 1.6417870661619662, "grad_norm": 0.9704864621162415, "learning_rate": 3.841200000000001e-06, "loss": 0.0085, "step": 30795 }, { "epoch": 1.6418403795916192, "grad_norm": 1.2204550504684448, "learning_rate": 3.841e-06, "loss": 0.0116, "step": 30796 }, { "epoch": 1.641893693021272, "grad_norm": 2.522524356842041, "learning_rate": 3.8408e-06, "loss": 0.0196, "step": 30797 }, { "epoch": 1.641947006450925, "grad_norm": 1.3507730960845947, "learning_rate": 3.8406e-06, "loss": 0.0104, "step": 30798 }, { "epoch": 1.642000319880578, "grad_norm": 1.109993577003479, "learning_rate": 3.8404000000000005e-06, "loss": 0.0091, "step": 30799 }, { "epoch": 1.6420536333102307, "grad_norm": 1.5521563291549683, "learning_rate": 3.8402e-06, "loss": 0.0143, "step": 30800 }, { "epoch": 1.6421069467398839, "grad_norm": 1.9274471998214722, "learning_rate": 3.8400000000000005e-06, "loss": 0.0266, "step": 30801 }, { "epoch": 1.6421602601695366, "grad_norm": 0.9278640747070312, "learning_rate": 3.8398e-06, "loss": 0.0063, "step": 30802 }, { "epoch": 1.6422135735991896, "grad_norm": 1.6996142864227295, "learning_rate": 3.8396000000000005e-06, "loss": 0.0335, "step": 30803 }, { "epoch": 1.6422668870288426, "grad_norm": 1.9753717184066772, "learning_rate": 3.8394e-06, "loss": 0.0145, "step": 30804 }, { "epoch": 1.6423202004584954, "grad_norm": 3.8250033855438232, "learning_rate": 3.8392000000000004e-06, "loss": 0.0306, "step": 30805 }, { "epoch": 1.6423735138881486, "grad_norm": 2.6990065574645996, "learning_rate": 3.839000000000001e-06, "loss": 0.0223, "step": 30806 }, { "epoch": 1.6424268273178013, "grad_norm": 1.4693273305892944, "learning_rate": 3.8388e-06, "loss": 0.0132, "step": 30807 }, { "epoch": 1.6424801407474543, "grad_norm": 2.1191272735595703, "learning_rate": 3.8386e-06, "loss": 0.0092, "step": 30808 }, { "epoch": 1.6425334541771073, "grad_norm": 0.9005045294761658, "learning_rate": 3.8384e-06, "loss": 0.0075, "step": 30809 }, { "epoch": 1.64258676760676, "grad_norm": 2.298665761947632, "learning_rate": 3.8382e-06, "loss": 0.0083, "step": 30810 }, { "epoch": 1.6426400810364132, "grad_norm": 1.2514408826828003, "learning_rate": 3.838e-06, "loss": 0.0098, "step": 30811 }, { "epoch": 1.642693394466066, "grad_norm": 2.653430461883545, "learning_rate": 3.837800000000001e-06, "loss": 0.0406, "step": 30812 }, { "epoch": 1.642746707895719, "grad_norm": 2.1006226539611816, "learning_rate": 3.8376e-06, "loss": 0.0207, "step": 30813 }, { "epoch": 1.642800021325372, "grad_norm": 2.236640691757202, "learning_rate": 3.837400000000001e-06, "loss": 0.0242, "step": 30814 }, { "epoch": 1.6428533347550247, "grad_norm": 1.8273738622665405, "learning_rate": 3.8372e-06, "loss": 0.013, "step": 30815 }, { "epoch": 1.6429066481846777, "grad_norm": 1.72412109375, "learning_rate": 3.837e-06, "loss": 0.0098, "step": 30816 }, { "epoch": 1.6429599616143307, "grad_norm": 4.081607818603516, "learning_rate": 3.8368e-06, "loss": 0.0168, "step": 30817 }, { "epoch": 1.6430132750439834, "grad_norm": 2.3344504833221436, "learning_rate": 3.836600000000001e-06, "loss": 0.017, "step": 30818 }, { "epoch": 1.6430665884736366, "grad_norm": 1.1875033378601074, "learning_rate": 3.8364e-06, "loss": 0.0065, "step": 30819 }, { "epoch": 1.6431199019032894, "grad_norm": 1.1454776525497437, "learning_rate": 3.836200000000001e-06, "loss": 0.0078, "step": 30820 }, { "epoch": 1.6431732153329424, "grad_norm": 2.818713903427124, "learning_rate": 3.836e-06, "loss": 0.0616, "step": 30821 }, { "epoch": 1.6432265287625953, "grad_norm": 2.910653829574585, "learning_rate": 3.8358e-06, "loss": 0.0168, "step": 30822 }, { "epoch": 1.643279842192248, "grad_norm": 1.6127960681915283, "learning_rate": 3.8356e-06, "loss": 0.0142, "step": 30823 }, { "epoch": 1.6433331556219013, "grad_norm": 6.627423286437988, "learning_rate": 3.835400000000001e-06, "loss": 0.0178, "step": 30824 }, { "epoch": 1.643386469051554, "grad_norm": 1.1846445798873901, "learning_rate": 3.835200000000001e-06, "loss": 0.0124, "step": 30825 }, { "epoch": 1.643439782481207, "grad_norm": 1.8929864168167114, "learning_rate": 3.8350000000000006e-06, "loss": 0.0176, "step": 30826 }, { "epoch": 1.64349309591086, "grad_norm": 2.8736300468444824, "learning_rate": 3.8348e-06, "loss": 0.0677, "step": 30827 }, { "epoch": 1.6435464093405128, "grad_norm": 2.333660840988159, "learning_rate": 3.8346000000000005e-06, "loss": 0.009, "step": 30828 }, { "epoch": 1.6435997227701657, "grad_norm": 2.4543628692626953, "learning_rate": 3.8344e-06, "loss": 0.0335, "step": 30829 }, { "epoch": 1.6436530361998187, "grad_norm": 2.7082910537719727, "learning_rate": 3.8342000000000005e-06, "loss": 0.0253, "step": 30830 }, { "epoch": 1.6437063496294717, "grad_norm": 1.3451930284500122, "learning_rate": 3.834000000000001e-06, "loss": 0.0103, "step": 30831 }, { "epoch": 1.6437596630591247, "grad_norm": 1.8035389184951782, "learning_rate": 3.8338000000000005e-06, "loss": 0.0536, "step": 30832 }, { "epoch": 1.6438129764887774, "grad_norm": 0.9653276205062866, "learning_rate": 3.8336e-06, "loss": 0.0254, "step": 30833 }, { "epoch": 1.6438662899184304, "grad_norm": 2.44453763961792, "learning_rate": 3.8334000000000004e-06, "loss": 0.0097, "step": 30834 }, { "epoch": 1.6439196033480834, "grad_norm": 3.26495361328125, "learning_rate": 3.8332e-06, "loss": 0.0509, "step": 30835 }, { "epoch": 1.6439729167777362, "grad_norm": 1.0697014331817627, "learning_rate": 3.833e-06, "loss": 0.0292, "step": 30836 }, { "epoch": 1.6440262302073894, "grad_norm": 2.7429137229919434, "learning_rate": 3.832800000000001e-06, "loss": 0.0547, "step": 30837 }, { "epoch": 1.6440795436370421, "grad_norm": 2.263594150543213, "learning_rate": 3.8326e-06, "loss": 0.0231, "step": 30838 }, { "epoch": 1.644132857066695, "grad_norm": 2.8096764087677, "learning_rate": 3.832400000000001e-06, "loss": 0.0229, "step": 30839 }, { "epoch": 1.644186170496348, "grad_norm": 2.7070658206939697, "learning_rate": 3.8322e-06, "loss": 0.0181, "step": 30840 }, { "epoch": 1.6442394839260008, "grad_norm": 1.6336889266967773, "learning_rate": 3.832e-06, "loss": 0.0337, "step": 30841 }, { "epoch": 1.644292797355654, "grad_norm": 2.72314715385437, "learning_rate": 3.8318e-06, "loss": 0.048, "step": 30842 }, { "epoch": 1.6443461107853068, "grad_norm": 3.554433584213257, "learning_rate": 3.831600000000001e-06, "loss": 0.0207, "step": 30843 }, { "epoch": 1.6443994242149598, "grad_norm": 7.413410186767578, "learning_rate": 3.8314e-06, "loss": 0.0334, "step": 30844 }, { "epoch": 1.6444527376446127, "grad_norm": 3.4426534175872803, "learning_rate": 3.831200000000001e-06, "loss": 0.0142, "step": 30845 }, { "epoch": 1.6445060510742655, "grad_norm": 2.7670907974243164, "learning_rate": 3.831e-06, "loss": 0.054, "step": 30846 }, { "epoch": 1.6445593645039185, "grad_norm": 1.2674332857131958, "learning_rate": 3.8308e-06, "loss": 0.0204, "step": 30847 }, { "epoch": 1.6446126779335715, "grad_norm": 1.9195362329483032, "learning_rate": 3.8306e-06, "loss": 0.0302, "step": 30848 }, { "epoch": 1.6446659913632244, "grad_norm": 1.2637847661972046, "learning_rate": 3.830400000000001e-06, "loss": 0.0096, "step": 30849 }, { "epoch": 1.6447193047928774, "grad_norm": 0.8299224376678467, "learning_rate": 3.8302e-06, "loss": 0.0308, "step": 30850 }, { "epoch": 1.6447726182225302, "grad_norm": 1.7447565793991089, "learning_rate": 3.830000000000001e-06, "loss": 0.0282, "step": 30851 }, { "epoch": 1.6448259316521832, "grad_norm": 0.6269776225090027, "learning_rate": 3.8298e-06, "loss": 0.0057, "step": 30852 }, { "epoch": 1.6448792450818361, "grad_norm": 0.799822211265564, "learning_rate": 3.8296000000000006e-06, "loss": 0.0096, "step": 30853 }, { "epoch": 1.644932558511489, "grad_norm": 1.4238700866699219, "learning_rate": 3.8294e-06, "loss": 0.031, "step": 30854 }, { "epoch": 1.644985871941142, "grad_norm": 3.2196409702301025, "learning_rate": 3.8292000000000005e-06, "loss": 0.0633, "step": 30855 }, { "epoch": 1.6450391853707949, "grad_norm": 1.600221872329712, "learning_rate": 3.829e-06, "loss": 0.02, "step": 30856 }, { "epoch": 1.6450924988004478, "grad_norm": 1.083005666732788, "learning_rate": 3.8288000000000005e-06, "loss": 0.0131, "step": 30857 }, { "epoch": 1.6451458122301008, "grad_norm": 1.8163397312164307, "learning_rate": 3.8286e-06, "loss": 0.0522, "step": 30858 }, { "epoch": 1.6451991256597536, "grad_norm": 1.8349945545196533, "learning_rate": 3.8284000000000005e-06, "loss": 0.0305, "step": 30859 }, { "epoch": 1.6452524390894068, "grad_norm": 1.1654917001724243, "learning_rate": 3.8282e-06, "loss": 0.02, "step": 30860 }, { "epoch": 1.6453057525190595, "grad_norm": 1.6790709495544434, "learning_rate": 3.8280000000000004e-06, "loss": 0.0261, "step": 30861 }, { "epoch": 1.6453590659487125, "grad_norm": 1.8486534357070923, "learning_rate": 3.8278e-06, "loss": 0.0206, "step": 30862 }, { "epoch": 1.6454123793783655, "grad_norm": 2.0055909156799316, "learning_rate": 3.8276e-06, "loss": 0.0261, "step": 30863 }, { "epoch": 1.6454656928080182, "grad_norm": 1.6185554265975952, "learning_rate": 3.827400000000001e-06, "loss": 0.02, "step": 30864 }, { "epoch": 1.6455190062376712, "grad_norm": 2.252631664276123, "learning_rate": 3.8272e-06, "loss": 0.0288, "step": 30865 }, { "epoch": 1.6455723196673242, "grad_norm": 1.2271809577941895, "learning_rate": 3.827e-06, "loss": 0.013, "step": 30866 }, { "epoch": 1.6456256330969772, "grad_norm": 1.5729186534881592, "learning_rate": 3.8268e-06, "loss": 0.013, "step": 30867 }, { "epoch": 1.6456789465266302, "grad_norm": 1.427336573600769, "learning_rate": 3.8266e-06, "loss": 0.009, "step": 30868 }, { "epoch": 1.645732259956283, "grad_norm": 1.7664299011230469, "learning_rate": 3.8264e-06, "loss": 0.017, "step": 30869 }, { "epoch": 1.645785573385936, "grad_norm": 2.136425733566284, "learning_rate": 3.826200000000001e-06, "loss": 0.0252, "step": 30870 }, { "epoch": 1.6458388868155889, "grad_norm": 1.7867919206619263, "learning_rate": 3.826e-06, "loss": 0.0272, "step": 30871 }, { "epoch": 1.6458922002452416, "grad_norm": 1.2655621767044067, "learning_rate": 3.8258e-06, "loss": 0.0151, "step": 30872 }, { "epoch": 1.6459455136748948, "grad_norm": 1.861760139465332, "learning_rate": 3.8256e-06, "loss": 0.0124, "step": 30873 }, { "epoch": 1.6459988271045476, "grad_norm": 0.7743854522705078, "learning_rate": 3.8254e-06, "loss": 0.0092, "step": 30874 }, { "epoch": 1.6460521405342006, "grad_norm": 1.7033960819244385, "learning_rate": 3.8252e-06, "loss": 0.0137, "step": 30875 }, { "epoch": 1.6461054539638535, "grad_norm": 1.3233672380447388, "learning_rate": 3.825000000000001e-06, "loss": 0.0089, "step": 30876 }, { "epoch": 1.6461587673935063, "grad_norm": 1.4202957153320312, "learning_rate": 3.8248e-06, "loss": 0.0283, "step": 30877 }, { "epoch": 1.6462120808231595, "grad_norm": 1.1444083452224731, "learning_rate": 3.824600000000001e-06, "loss": 0.0074, "step": 30878 }, { "epoch": 1.6462653942528123, "grad_norm": 0.8696302771568298, "learning_rate": 3.8244e-06, "loss": 0.0053, "step": 30879 }, { "epoch": 1.6463187076824652, "grad_norm": 1.5215373039245605, "learning_rate": 3.8242e-06, "loss": 0.0323, "step": 30880 }, { "epoch": 1.6463720211121182, "grad_norm": 0.982747495174408, "learning_rate": 3.824e-06, "loss": 0.0069, "step": 30881 }, { "epoch": 1.646425334541771, "grad_norm": 0.9762700796127319, "learning_rate": 3.8238000000000006e-06, "loss": 0.014, "step": 30882 }, { "epoch": 1.646478647971424, "grad_norm": 0.9482643008232117, "learning_rate": 3.8236e-06, "loss": 0.006, "step": 30883 }, { "epoch": 1.646531961401077, "grad_norm": 1.7130149602890015, "learning_rate": 3.8234000000000005e-06, "loss": 0.007, "step": 30884 }, { "epoch": 1.6465852748307297, "grad_norm": 2.6124260425567627, "learning_rate": 3.8232e-06, "loss": 0.0234, "step": 30885 }, { "epoch": 1.646638588260383, "grad_norm": 0.7468337416648865, "learning_rate": 3.823e-06, "loss": 0.0046, "step": 30886 }, { "epoch": 1.6466919016900357, "grad_norm": 2.6652965545654297, "learning_rate": 3.8228e-06, "loss": 0.0452, "step": 30887 }, { "epoch": 1.6467452151196886, "grad_norm": 1.196342945098877, "learning_rate": 3.8226000000000005e-06, "loss": 0.0233, "step": 30888 }, { "epoch": 1.6467985285493416, "grad_norm": 2.0587894916534424, "learning_rate": 3.822400000000001e-06, "loss": 0.0104, "step": 30889 }, { "epoch": 1.6468518419789944, "grad_norm": 1.3448147773742676, "learning_rate": 3.8222000000000004e-06, "loss": 0.0149, "step": 30890 }, { "epoch": 1.6469051554086476, "grad_norm": 3.716294050216675, "learning_rate": 3.822e-06, "loss": 0.0475, "step": 30891 }, { "epoch": 1.6469584688383003, "grad_norm": 1.3088065385818481, "learning_rate": 3.8218e-06, "loss": 0.0135, "step": 30892 }, { "epoch": 1.6470117822679533, "grad_norm": 2.6555562019348145, "learning_rate": 3.8216e-06, "loss": 0.0164, "step": 30893 }, { "epoch": 1.6470650956976063, "grad_norm": 1.5572118759155273, "learning_rate": 3.8214e-06, "loss": 0.0235, "step": 30894 }, { "epoch": 1.647118409127259, "grad_norm": 1.8508093357086182, "learning_rate": 3.821200000000001e-06, "loss": 0.0326, "step": 30895 }, { "epoch": 1.6471717225569122, "grad_norm": 1.558178424835205, "learning_rate": 3.821e-06, "loss": 0.0228, "step": 30896 }, { "epoch": 1.647225035986565, "grad_norm": 2.8859400749206543, "learning_rate": 3.8208e-06, "loss": 0.0299, "step": 30897 }, { "epoch": 1.647278349416218, "grad_norm": 2.305722951889038, "learning_rate": 3.8206e-06, "loss": 0.0254, "step": 30898 }, { "epoch": 1.647331662845871, "grad_norm": 2.022340774536133, "learning_rate": 3.8204e-06, "loss": 0.0235, "step": 30899 }, { "epoch": 1.6473849762755237, "grad_norm": 2.5743136405944824, "learning_rate": 3.8202e-06, "loss": 0.0471, "step": 30900 }, { "epoch": 1.6474382897051767, "grad_norm": 5.557651996612549, "learning_rate": 3.820000000000001e-06, "loss": 0.0431, "step": 30901 }, { "epoch": 1.6474916031348297, "grad_norm": 2.230652332305908, "learning_rate": 3.8198e-06, "loss": 0.0135, "step": 30902 }, { "epoch": 1.6475449165644824, "grad_norm": 4.100808620452881, "learning_rate": 3.819600000000001e-06, "loss": 0.0158, "step": 30903 }, { "epoch": 1.6475982299941356, "grad_norm": 1.3926059007644653, "learning_rate": 3.8194e-06, "loss": 0.0185, "step": 30904 }, { "epoch": 1.6476515434237884, "grad_norm": 1.2010003328323364, "learning_rate": 3.8192e-06, "loss": 0.0077, "step": 30905 }, { "epoch": 1.6477048568534414, "grad_norm": 1.7656782865524292, "learning_rate": 3.819e-06, "loss": 0.026, "step": 30906 }, { "epoch": 1.6477581702830943, "grad_norm": 1.9690155982971191, "learning_rate": 3.818800000000001e-06, "loss": 0.0176, "step": 30907 }, { "epoch": 1.647811483712747, "grad_norm": 0.6886759996414185, "learning_rate": 3.8186e-06, "loss": 0.0078, "step": 30908 }, { "epoch": 1.6478647971424003, "grad_norm": 1.2731711864471436, "learning_rate": 3.8184000000000006e-06, "loss": 0.0195, "step": 30909 }, { "epoch": 1.647918110572053, "grad_norm": 1.8851730823516846, "learning_rate": 3.8182e-06, "loss": 0.0353, "step": 30910 }, { "epoch": 1.647971424001706, "grad_norm": 1.0663824081420898, "learning_rate": 3.818e-06, "loss": 0.0071, "step": 30911 }, { "epoch": 1.648024737431359, "grad_norm": 1.892530918121338, "learning_rate": 3.8178e-06, "loss": 0.022, "step": 30912 }, { "epoch": 1.6480780508610118, "grad_norm": 1.7780495882034302, "learning_rate": 3.8176000000000005e-06, "loss": 0.0122, "step": 30913 }, { "epoch": 1.648131364290665, "grad_norm": 1.4075058698654175, "learning_rate": 3.817400000000001e-06, "loss": 0.0335, "step": 30914 }, { "epoch": 1.6481846777203177, "grad_norm": 1.388949990272522, "learning_rate": 3.8172000000000005e-06, "loss": 0.0078, "step": 30915 }, { "epoch": 1.6482379911499707, "grad_norm": 1.2969468832015991, "learning_rate": 3.817e-06, "loss": 0.0225, "step": 30916 }, { "epoch": 1.6482913045796237, "grad_norm": 1.4690930843353271, "learning_rate": 3.8168000000000005e-06, "loss": 0.0225, "step": 30917 }, { "epoch": 1.6483446180092765, "grad_norm": 0.9688432812690735, "learning_rate": 3.8166e-06, "loss": 0.0078, "step": 30918 }, { "epoch": 1.6483979314389294, "grad_norm": 2.857328414916992, "learning_rate": 3.8164000000000004e-06, "loss": 0.0378, "step": 30919 }, { "epoch": 1.6484512448685824, "grad_norm": 2.3521618843078613, "learning_rate": 3.816200000000001e-06, "loss": 0.0179, "step": 30920 }, { "epoch": 1.6485045582982352, "grad_norm": 1.536049723625183, "learning_rate": 3.816e-06, "loss": 0.0205, "step": 30921 }, { "epoch": 1.6485578717278884, "grad_norm": 4.235261917114258, "learning_rate": 3.8158e-06, "loss": 0.0271, "step": 30922 }, { "epoch": 1.6486111851575411, "grad_norm": 2.3546547889709473, "learning_rate": 3.8156e-06, "loss": 0.0368, "step": 30923 }, { "epoch": 1.648664498587194, "grad_norm": 1.4930094480514526, "learning_rate": 3.8154e-06, "loss": 0.0206, "step": 30924 }, { "epoch": 1.648717812016847, "grad_norm": 2.2371020317077637, "learning_rate": 3.8152e-06, "loss": 0.0302, "step": 30925 }, { "epoch": 1.6487711254464998, "grad_norm": 1.5980616807937622, "learning_rate": 3.815000000000001e-06, "loss": 0.0217, "step": 30926 }, { "epoch": 1.648824438876153, "grad_norm": 1.8178120851516724, "learning_rate": 3.8148e-06, "loss": 0.02, "step": 30927 }, { "epoch": 1.6488777523058058, "grad_norm": 1.928130030632019, "learning_rate": 3.8146000000000007e-06, "loss": 0.0129, "step": 30928 }, { "epoch": 1.6489310657354588, "grad_norm": 2.7134130001068115, "learning_rate": 3.8144000000000003e-06, "loss": 0.0322, "step": 30929 }, { "epoch": 1.6489843791651118, "grad_norm": 0.691916823387146, "learning_rate": 3.8142000000000003e-06, "loss": 0.0054, "step": 30930 }, { "epoch": 1.6490376925947645, "grad_norm": 3.036693811416626, "learning_rate": 3.8140000000000007e-06, "loss": 0.0252, "step": 30931 }, { "epoch": 1.6490910060244177, "grad_norm": 2.197953462600708, "learning_rate": 3.8138000000000002e-06, "loss": 0.0385, "step": 30932 }, { "epoch": 1.6491443194540705, "grad_norm": 0.31298166513442993, "learning_rate": 3.8136000000000002e-06, "loss": 0.0035, "step": 30933 }, { "epoch": 1.6491976328837235, "grad_norm": 1.1765753030776978, "learning_rate": 3.8134000000000006e-06, "loss": 0.0149, "step": 30934 }, { "epoch": 1.6492509463133764, "grad_norm": 1.3843228816986084, "learning_rate": 3.8132e-06, "loss": 0.0185, "step": 30935 }, { "epoch": 1.6493042597430292, "grad_norm": 1.6063716411590576, "learning_rate": 3.813e-06, "loss": 0.016, "step": 30936 }, { "epoch": 1.6493575731726822, "grad_norm": 2.349172592163086, "learning_rate": 3.8128000000000006e-06, "loss": 0.023, "step": 30937 }, { "epoch": 1.6494108866023351, "grad_norm": 2.3278989791870117, "learning_rate": 3.8126e-06, "loss": 0.0487, "step": 30938 }, { "epoch": 1.649464200031988, "grad_norm": 1.5206302404403687, "learning_rate": 3.8124000000000005e-06, "loss": 0.0171, "step": 30939 }, { "epoch": 1.649517513461641, "grad_norm": 0.8885787129402161, "learning_rate": 3.8122000000000005e-06, "loss": 0.0096, "step": 30940 }, { "epoch": 1.6495708268912939, "grad_norm": 3.716331720352173, "learning_rate": 3.812e-06, "loss": 0.0445, "step": 30941 }, { "epoch": 1.6496241403209468, "grad_norm": 0.7170659303665161, "learning_rate": 3.8118000000000005e-06, "loss": 0.0074, "step": 30942 }, { "epoch": 1.6496774537505998, "grad_norm": 1.519935965538025, "learning_rate": 3.8116000000000005e-06, "loss": 0.031, "step": 30943 }, { "epoch": 1.6497307671802526, "grad_norm": 0.4249371290206909, "learning_rate": 3.8114e-06, "loss": 0.0032, "step": 30944 }, { "epoch": 1.6497840806099058, "grad_norm": 2.257293701171875, "learning_rate": 3.8112000000000005e-06, "loss": 0.0242, "step": 30945 }, { "epoch": 1.6498373940395585, "grad_norm": 1.5542223453521729, "learning_rate": 3.8110000000000004e-06, "loss": 0.0234, "step": 30946 }, { "epoch": 1.6498907074692115, "grad_norm": 1.277255654335022, "learning_rate": 3.8108e-06, "loss": 0.0104, "step": 30947 }, { "epoch": 1.6499440208988645, "grad_norm": 1.967207908630371, "learning_rate": 3.8106000000000004e-06, "loss": 0.0203, "step": 30948 }, { "epoch": 1.6499973343285173, "grad_norm": 2.379101276397705, "learning_rate": 3.8104000000000004e-06, "loss": 0.0212, "step": 30949 }, { "epoch": 1.6500506477581702, "grad_norm": 1.1543480157852173, "learning_rate": 3.8102000000000004e-06, "loss": 0.0178, "step": 30950 }, { "epoch": 1.6501039611878232, "grad_norm": 1.19754958152771, "learning_rate": 3.8100000000000004e-06, "loss": 0.0135, "step": 30951 }, { "epoch": 1.6501572746174762, "grad_norm": 1.618332028388977, "learning_rate": 3.8098000000000004e-06, "loss": 0.0354, "step": 30952 }, { "epoch": 1.6502105880471292, "grad_norm": 1.372450828552246, "learning_rate": 3.8096000000000003e-06, "loss": 0.021, "step": 30953 }, { "epoch": 1.650263901476782, "grad_norm": 2.4357290267944336, "learning_rate": 3.8094000000000003e-06, "loss": 0.0168, "step": 30954 }, { "epoch": 1.650317214906435, "grad_norm": 1.5055850744247437, "learning_rate": 3.8092000000000003e-06, "loss": 0.0099, "step": 30955 }, { "epoch": 1.6503705283360879, "grad_norm": 2.256347179412842, "learning_rate": 3.8090000000000003e-06, "loss": 0.0376, "step": 30956 }, { "epoch": 1.6504238417657406, "grad_norm": 1.9295458793640137, "learning_rate": 3.8088000000000003e-06, "loss": 0.0235, "step": 30957 }, { "epoch": 1.6504771551953938, "grad_norm": 1.1694613695144653, "learning_rate": 3.8086000000000003e-06, "loss": 0.0164, "step": 30958 }, { "epoch": 1.6505304686250466, "grad_norm": 5.010017395019531, "learning_rate": 3.8084000000000002e-06, "loss": 0.0225, "step": 30959 }, { "epoch": 1.6505837820546996, "grad_norm": 1.4994570016860962, "learning_rate": 3.8082000000000002e-06, "loss": 0.009, "step": 30960 }, { "epoch": 1.6506370954843526, "grad_norm": 1.0969839096069336, "learning_rate": 3.8080000000000006e-06, "loss": 0.0076, "step": 30961 }, { "epoch": 1.6506904089140053, "grad_norm": 0.6531780362129211, "learning_rate": 3.8078e-06, "loss": 0.0056, "step": 30962 }, { "epoch": 1.6507437223436585, "grad_norm": 1.4932321310043335, "learning_rate": 3.8076e-06, "loss": 0.0116, "step": 30963 }, { "epoch": 1.6507970357733113, "grad_norm": 2.5952515602111816, "learning_rate": 3.8074000000000006e-06, "loss": 0.0445, "step": 30964 }, { "epoch": 1.6508503492029643, "grad_norm": 2.8341307640075684, "learning_rate": 3.8072e-06, "loss": 0.0377, "step": 30965 }, { "epoch": 1.6509036626326172, "grad_norm": 2.018728733062744, "learning_rate": 3.807e-06, "loss": 0.0351, "step": 30966 }, { "epoch": 1.65095697606227, "grad_norm": 1.3134269714355469, "learning_rate": 3.8068000000000006e-06, "loss": 0.0166, "step": 30967 }, { "epoch": 1.651010289491923, "grad_norm": 2.2982676029205322, "learning_rate": 3.8066e-06, "loss": 0.0226, "step": 30968 }, { "epoch": 1.651063602921576, "grad_norm": 2.8264291286468506, "learning_rate": 3.8064e-06, "loss": 0.0205, "step": 30969 }, { "epoch": 1.651116916351229, "grad_norm": 1.4191467761993408, "learning_rate": 3.8062000000000005e-06, "loss": 0.0418, "step": 30970 }, { "epoch": 1.651170229780882, "grad_norm": 2.016159772872925, "learning_rate": 3.806e-06, "loss": 0.0338, "step": 30971 }, { "epoch": 1.6512235432105347, "grad_norm": 2.770984172821045, "learning_rate": 3.8058e-06, "loss": 0.026, "step": 30972 }, { "epoch": 1.6512768566401876, "grad_norm": 3.2164101600646973, "learning_rate": 3.8056000000000005e-06, "loss": 0.0224, "step": 30973 }, { "epoch": 1.6513301700698406, "grad_norm": 1.8674092292785645, "learning_rate": 3.8054e-06, "loss": 0.0128, "step": 30974 }, { "epoch": 1.6513834834994934, "grad_norm": 2.0761821269989014, "learning_rate": 3.8052000000000004e-06, "loss": 0.0283, "step": 30975 }, { "epoch": 1.6514367969291466, "grad_norm": 1.6183857917785645, "learning_rate": 3.8050000000000004e-06, "loss": 0.0283, "step": 30976 }, { "epoch": 1.6514901103587993, "grad_norm": 3.057508945465088, "learning_rate": 3.8048e-06, "loss": 0.0516, "step": 30977 }, { "epoch": 1.6515434237884523, "grad_norm": 1.8685550689697266, "learning_rate": 3.8046000000000004e-06, "loss": 0.0139, "step": 30978 }, { "epoch": 1.6515967372181053, "grad_norm": 1.0824159383773804, "learning_rate": 3.8044000000000004e-06, "loss": 0.0173, "step": 30979 }, { "epoch": 1.651650050647758, "grad_norm": 1.2960313558578491, "learning_rate": 3.8042e-06, "loss": 0.011, "step": 30980 }, { "epoch": 1.6517033640774113, "grad_norm": 1.7164050340652466, "learning_rate": 3.8040000000000003e-06, "loss": 0.016, "step": 30981 }, { "epoch": 1.651756677507064, "grad_norm": 1.5089466571807861, "learning_rate": 3.8038000000000003e-06, "loss": 0.0105, "step": 30982 }, { "epoch": 1.651809990936717, "grad_norm": 2.7002692222595215, "learning_rate": 3.8036e-06, "loss": 0.0272, "step": 30983 }, { "epoch": 1.65186330436637, "grad_norm": 1.4961097240447998, "learning_rate": 3.8034000000000003e-06, "loss": 0.0142, "step": 30984 }, { "epoch": 1.6519166177960227, "grad_norm": 2.946624279022217, "learning_rate": 3.8032000000000003e-06, "loss": 0.0302, "step": 30985 }, { "epoch": 1.6519699312256757, "grad_norm": 2.171506881713867, "learning_rate": 3.8030000000000007e-06, "loss": 0.0204, "step": 30986 }, { "epoch": 1.6520232446553287, "grad_norm": 2.008005380630493, "learning_rate": 3.8028000000000003e-06, "loss": 0.0187, "step": 30987 }, { "epoch": 1.6520765580849817, "grad_norm": 1.720080852508545, "learning_rate": 3.8026000000000002e-06, "loss": 0.0146, "step": 30988 }, { "epoch": 1.6521298715146346, "grad_norm": 2.1200833320617676, "learning_rate": 3.8024000000000006e-06, "loss": 0.0401, "step": 30989 }, { "epoch": 1.6521831849442874, "grad_norm": 2.030226230621338, "learning_rate": 3.8022e-06, "loss": 0.0165, "step": 30990 }, { "epoch": 1.6522364983739404, "grad_norm": 2.8819708824157715, "learning_rate": 3.802e-06, "loss": 0.0512, "step": 30991 }, { "epoch": 1.6522898118035934, "grad_norm": 1.493580937385559, "learning_rate": 3.8018000000000006e-06, "loss": 0.0121, "step": 30992 }, { "epoch": 1.6523431252332461, "grad_norm": 1.7827281951904297, "learning_rate": 3.8016e-06, "loss": 0.031, "step": 30993 }, { "epoch": 1.6523964386628993, "grad_norm": 1.7139036655426025, "learning_rate": 3.8014e-06, "loss": 0.023, "step": 30994 }, { "epoch": 1.652449752092552, "grad_norm": 1.5187084674835205, "learning_rate": 3.8012000000000006e-06, "loss": 0.0103, "step": 30995 }, { "epoch": 1.652503065522205, "grad_norm": 0.5482864379882812, "learning_rate": 3.801e-06, "loss": 0.0048, "step": 30996 }, { "epoch": 1.652556378951858, "grad_norm": 0.8937660455703735, "learning_rate": 3.8008e-06, "loss": 0.0116, "step": 30997 }, { "epoch": 1.6526096923815108, "grad_norm": 1.498771071434021, "learning_rate": 3.8006000000000005e-06, "loss": 0.024, "step": 30998 }, { "epoch": 1.652663005811164, "grad_norm": 1.4959361553192139, "learning_rate": 3.8004e-06, "loss": 0.0178, "step": 30999 }, { "epoch": 1.6527163192408167, "grad_norm": 1.9746798276901245, "learning_rate": 3.8002000000000005e-06, "loss": 0.0354, "step": 31000 }, { "epoch": 1.6527696326704697, "grad_norm": 4.7297844886779785, "learning_rate": 3.8000000000000005e-06, "loss": 0.0288, "step": 31001 }, { "epoch": 1.6528229461001227, "grad_norm": 4.885672569274902, "learning_rate": 3.7998e-06, "loss": 0.0479, "step": 31002 }, { "epoch": 1.6528762595297755, "grad_norm": 1.8580238819122314, "learning_rate": 3.7996000000000004e-06, "loss": 0.0324, "step": 31003 }, { "epoch": 1.6529295729594284, "grad_norm": 1.8622525930404663, "learning_rate": 3.7994000000000004e-06, "loss": 0.0284, "step": 31004 }, { "epoch": 1.6529828863890814, "grad_norm": 1.4241939783096313, "learning_rate": 3.7992e-06, "loss": 0.0104, "step": 31005 }, { "epoch": 1.6530361998187342, "grad_norm": 4.145415782928467, "learning_rate": 3.7990000000000004e-06, "loss": 0.0252, "step": 31006 }, { "epoch": 1.6530895132483874, "grad_norm": 2.394944190979004, "learning_rate": 3.7988000000000004e-06, "loss": 0.0325, "step": 31007 }, { "epoch": 1.6531428266780401, "grad_norm": 1.9288781881332397, "learning_rate": 3.7986e-06, "loss": 0.0326, "step": 31008 }, { "epoch": 1.6531961401076931, "grad_norm": 2.329643726348877, "learning_rate": 3.7984000000000003e-06, "loss": 0.0265, "step": 31009 }, { "epoch": 1.653249453537346, "grad_norm": 1.5417115688323975, "learning_rate": 3.7982000000000003e-06, "loss": 0.0292, "step": 31010 }, { "epoch": 1.6533027669669988, "grad_norm": 2.5988240242004395, "learning_rate": 3.7980000000000007e-06, "loss": 0.0245, "step": 31011 }, { "epoch": 1.653356080396652, "grad_norm": 1.787752389907837, "learning_rate": 3.7978000000000003e-06, "loss": 0.0324, "step": 31012 }, { "epoch": 1.6534093938263048, "grad_norm": 0.5904029607772827, "learning_rate": 3.7976000000000003e-06, "loss": 0.0048, "step": 31013 }, { "epoch": 1.6534627072559578, "grad_norm": 1.8968250751495361, "learning_rate": 3.7974000000000007e-06, "loss": 0.0158, "step": 31014 }, { "epoch": 1.6535160206856108, "grad_norm": 1.4564523696899414, "learning_rate": 3.7972000000000003e-06, "loss": 0.0177, "step": 31015 }, { "epoch": 1.6535693341152635, "grad_norm": 1.5240705013275146, "learning_rate": 3.7970000000000002e-06, "loss": 0.0179, "step": 31016 }, { "epoch": 1.6536226475449167, "grad_norm": 1.0511500835418701, "learning_rate": 3.7968000000000006e-06, "loss": 0.0296, "step": 31017 }, { "epoch": 1.6536759609745695, "grad_norm": 0.39918434619903564, "learning_rate": 3.7966e-06, "loss": 0.0047, "step": 31018 }, { "epoch": 1.6537292744042225, "grad_norm": 1.9582886695861816, "learning_rate": 3.7964e-06, "loss": 0.0204, "step": 31019 }, { "epoch": 1.6537825878338754, "grad_norm": 1.260014533996582, "learning_rate": 3.7962000000000006e-06, "loss": 0.012, "step": 31020 }, { "epoch": 1.6538359012635282, "grad_norm": 1.8678734302520752, "learning_rate": 3.796e-06, "loss": 0.0265, "step": 31021 }, { "epoch": 1.6538892146931812, "grad_norm": 1.7087525129318237, "learning_rate": 3.7958e-06, "loss": 0.0113, "step": 31022 }, { "epoch": 1.6539425281228342, "grad_norm": 0.2306603044271469, "learning_rate": 3.7956000000000006e-06, "loss": 0.003, "step": 31023 }, { "epoch": 1.653995841552487, "grad_norm": 0.9881938099861145, "learning_rate": 3.7954e-06, "loss": 0.011, "step": 31024 }, { "epoch": 1.6540491549821401, "grad_norm": 1.3009191751480103, "learning_rate": 3.7952000000000005e-06, "loss": 0.0253, "step": 31025 }, { "epoch": 1.6541024684117929, "grad_norm": 1.093919038772583, "learning_rate": 3.7950000000000005e-06, "loss": 0.0278, "step": 31026 }, { "epoch": 1.6541557818414458, "grad_norm": 2.0961554050445557, "learning_rate": 3.7948e-06, "loss": 0.0347, "step": 31027 }, { "epoch": 1.6542090952710988, "grad_norm": 0.7009280920028687, "learning_rate": 3.7946000000000005e-06, "loss": 0.0061, "step": 31028 }, { "epoch": 1.6542624087007516, "grad_norm": 2.3724944591522217, "learning_rate": 3.7944000000000005e-06, "loss": 0.0265, "step": 31029 }, { "epoch": 1.6543157221304048, "grad_norm": 1.8348945379257202, "learning_rate": 3.7942e-06, "loss": 0.0185, "step": 31030 }, { "epoch": 1.6543690355600575, "grad_norm": 1.5666989088058472, "learning_rate": 3.7940000000000004e-06, "loss": 0.0481, "step": 31031 }, { "epoch": 1.6544223489897105, "grad_norm": 1.314929485321045, "learning_rate": 3.7938000000000004e-06, "loss": 0.0093, "step": 31032 }, { "epoch": 1.6544756624193635, "grad_norm": 1.8416097164154053, "learning_rate": 3.7936e-06, "loss": 0.0224, "step": 31033 }, { "epoch": 1.6545289758490163, "grad_norm": 1.7004142999649048, "learning_rate": 3.7934000000000004e-06, "loss": 0.0307, "step": 31034 }, { "epoch": 1.6545822892786695, "grad_norm": 1.4794849157333374, "learning_rate": 3.7932000000000004e-06, "loss": 0.0259, "step": 31035 }, { "epoch": 1.6546356027083222, "grad_norm": 1.4156509637832642, "learning_rate": 3.7930000000000004e-06, "loss": 0.0099, "step": 31036 }, { "epoch": 1.6546889161379752, "grad_norm": 1.6891072988510132, "learning_rate": 3.7928000000000003e-06, "loss": 0.0141, "step": 31037 }, { "epoch": 1.6547422295676282, "grad_norm": 1.4216923713684082, "learning_rate": 3.7926000000000003e-06, "loss": 0.0172, "step": 31038 }, { "epoch": 1.654795542997281, "grad_norm": 1.2569011449813843, "learning_rate": 3.7924000000000003e-06, "loss": 0.0099, "step": 31039 }, { "epoch": 1.654848856426934, "grad_norm": 1.8547394275665283, "learning_rate": 3.7922000000000003e-06, "loss": 0.0338, "step": 31040 }, { "epoch": 1.654902169856587, "grad_norm": 2.359239101409912, "learning_rate": 3.7920000000000003e-06, "loss": 0.0255, "step": 31041 }, { "epoch": 1.6549554832862396, "grad_norm": 1.9916167259216309, "learning_rate": 3.7918000000000003e-06, "loss": 0.0121, "step": 31042 }, { "epoch": 1.6550087967158928, "grad_norm": 2.1370935440063477, "learning_rate": 3.7916000000000003e-06, "loss": 0.0148, "step": 31043 }, { "epoch": 1.6550621101455456, "grad_norm": 1.689659833908081, "learning_rate": 3.7914000000000002e-06, "loss": 0.0276, "step": 31044 }, { "epoch": 1.6551154235751986, "grad_norm": 1.4510688781738281, "learning_rate": 3.7912000000000002e-06, "loss": 0.031, "step": 31045 }, { "epoch": 1.6551687370048516, "grad_norm": 1.6359426975250244, "learning_rate": 3.791e-06, "loss": 0.0157, "step": 31046 }, { "epoch": 1.6552220504345043, "grad_norm": 0.859893798828125, "learning_rate": 3.7908e-06, "loss": 0.0179, "step": 31047 }, { "epoch": 1.6552753638641575, "grad_norm": 1.8492555618286133, "learning_rate": 3.7906e-06, "loss": 0.0263, "step": 31048 }, { "epoch": 1.6553286772938103, "grad_norm": 2.516690254211426, "learning_rate": 3.7904e-06, "loss": 0.0182, "step": 31049 }, { "epoch": 1.6553819907234633, "grad_norm": 0.6390131711959839, "learning_rate": 3.7902000000000006e-06, "loss": 0.0158, "step": 31050 }, { "epoch": 1.6554353041531162, "grad_norm": 1.1584655046463013, "learning_rate": 3.79e-06, "loss": 0.0206, "step": 31051 }, { "epoch": 1.655488617582769, "grad_norm": 1.1553499698638916, "learning_rate": 3.7898e-06, "loss": 0.0076, "step": 31052 }, { "epoch": 1.655541931012422, "grad_norm": 1.2791025638580322, "learning_rate": 3.7896000000000005e-06, "loss": 0.0085, "step": 31053 }, { "epoch": 1.655595244442075, "grad_norm": 1.5323635339736938, "learning_rate": 3.7894e-06, "loss": 0.0179, "step": 31054 }, { "epoch": 1.655648557871728, "grad_norm": 0.7762773036956787, "learning_rate": 3.7892e-06, "loss": 0.0038, "step": 31055 }, { "epoch": 1.655701871301381, "grad_norm": 1.5350030660629272, "learning_rate": 3.7890000000000005e-06, "loss": 0.0394, "step": 31056 }, { "epoch": 1.6557551847310337, "grad_norm": 2.722538709640503, "learning_rate": 3.7888e-06, "loss": 0.051, "step": 31057 }, { "epoch": 1.6558084981606866, "grad_norm": 0.8948620557785034, "learning_rate": 3.7886e-06, "loss": 0.0059, "step": 31058 }, { "epoch": 1.6558618115903396, "grad_norm": 0.41298574209213257, "learning_rate": 3.7884000000000004e-06, "loss": 0.0046, "step": 31059 }, { "epoch": 1.6559151250199924, "grad_norm": 1.4314299821853638, "learning_rate": 3.7882e-06, "loss": 0.0237, "step": 31060 }, { "epoch": 1.6559684384496456, "grad_norm": 1.8386220932006836, "learning_rate": 3.7880000000000004e-06, "loss": 0.0351, "step": 31061 }, { "epoch": 1.6560217518792983, "grad_norm": 1.543021559715271, "learning_rate": 3.7878000000000004e-06, "loss": 0.0176, "step": 31062 }, { "epoch": 1.6560750653089513, "grad_norm": 2.5271005630493164, "learning_rate": 3.7876e-06, "loss": 0.0393, "step": 31063 }, { "epoch": 1.6561283787386043, "grad_norm": 2.140080690383911, "learning_rate": 3.7874000000000004e-06, "loss": 0.0269, "step": 31064 }, { "epoch": 1.656181692168257, "grad_norm": 1.2203830480575562, "learning_rate": 3.7872000000000003e-06, "loss": 0.0263, "step": 31065 }, { "epoch": 1.6562350055979103, "grad_norm": 1.188217282295227, "learning_rate": 3.787e-06, "loss": 0.0096, "step": 31066 }, { "epoch": 1.656288319027563, "grad_norm": 1.3398723602294922, "learning_rate": 3.7868000000000003e-06, "loss": 0.0124, "step": 31067 }, { "epoch": 1.656341632457216, "grad_norm": 1.0337556600570679, "learning_rate": 3.7866000000000003e-06, "loss": 0.0225, "step": 31068 }, { "epoch": 1.656394945886869, "grad_norm": 2.0694329738616943, "learning_rate": 3.7864e-06, "loss": 0.0243, "step": 31069 }, { "epoch": 1.6564482593165217, "grad_norm": 2.247586250305176, "learning_rate": 3.7862000000000003e-06, "loss": 0.0154, "step": 31070 }, { "epoch": 1.6565015727461747, "grad_norm": 1.8587446212768555, "learning_rate": 3.7860000000000003e-06, "loss": 0.0156, "step": 31071 }, { "epoch": 1.6565548861758277, "grad_norm": 1.6855117082595825, "learning_rate": 3.7858000000000007e-06, "loss": 0.0131, "step": 31072 }, { "epoch": 1.6566081996054807, "grad_norm": 0.42842575907707214, "learning_rate": 3.7856000000000002e-06, "loss": 0.004, "step": 31073 }, { "epoch": 1.6566615130351336, "grad_norm": 0.5862998366355896, "learning_rate": 3.7854000000000002e-06, "loss": 0.0047, "step": 31074 }, { "epoch": 1.6567148264647864, "grad_norm": 2.1860737800598145, "learning_rate": 3.7852000000000006e-06, "loss": 0.0351, "step": 31075 }, { "epoch": 1.6567681398944394, "grad_norm": 1.4803842306137085, "learning_rate": 3.785e-06, "loss": 0.0168, "step": 31076 }, { "epoch": 1.6568214533240924, "grad_norm": 2.4171531200408936, "learning_rate": 3.7848e-06, "loss": 0.0314, "step": 31077 }, { "epoch": 1.6568747667537451, "grad_norm": 0.7334781289100647, "learning_rate": 3.7846000000000006e-06, "loss": 0.0073, "step": 31078 }, { "epoch": 1.6569280801833983, "grad_norm": 5.993711471557617, "learning_rate": 3.7844e-06, "loss": 0.0457, "step": 31079 }, { "epoch": 1.656981393613051, "grad_norm": 0.38243117928504944, "learning_rate": 3.7842e-06, "loss": 0.0032, "step": 31080 }, { "epoch": 1.657034707042704, "grad_norm": 0.9571734070777893, "learning_rate": 3.7840000000000005e-06, "loss": 0.0218, "step": 31081 }, { "epoch": 1.657088020472357, "grad_norm": 1.716691493988037, "learning_rate": 3.7838e-06, "loss": 0.0156, "step": 31082 }, { "epoch": 1.6571413339020098, "grad_norm": 1.8171601295471191, "learning_rate": 3.7836e-06, "loss": 0.0273, "step": 31083 }, { "epoch": 1.657194647331663, "grad_norm": 0.3196347951889038, "learning_rate": 3.7834000000000005e-06, "loss": 0.0031, "step": 31084 }, { "epoch": 1.6572479607613158, "grad_norm": 2.975952625274658, "learning_rate": 3.7832e-06, "loss": 0.0527, "step": 31085 }, { "epoch": 1.6573012741909687, "grad_norm": 2.4423885345458984, "learning_rate": 3.7830000000000005e-06, "loss": 0.0456, "step": 31086 }, { "epoch": 1.6573545876206217, "grad_norm": 1.5952824354171753, "learning_rate": 3.7828000000000004e-06, "loss": 0.0198, "step": 31087 }, { "epoch": 1.6574079010502745, "grad_norm": 2.019286632537842, "learning_rate": 3.7826e-06, "loss": 0.027, "step": 31088 }, { "epoch": 1.6574612144799274, "grad_norm": 0.7135692834854126, "learning_rate": 3.7824000000000004e-06, "loss": 0.0042, "step": 31089 }, { "epoch": 1.6575145279095804, "grad_norm": 2.3612208366394043, "learning_rate": 3.7822000000000004e-06, "loss": 0.02, "step": 31090 }, { "epoch": 1.6575678413392334, "grad_norm": 2.468336343765259, "learning_rate": 3.782e-06, "loss": 0.0113, "step": 31091 }, { "epoch": 1.6576211547688864, "grad_norm": 2.1151673793792725, "learning_rate": 3.7818000000000004e-06, "loss": 0.0222, "step": 31092 }, { "epoch": 1.6576744681985391, "grad_norm": 1.6883779764175415, "learning_rate": 3.7816000000000004e-06, "loss": 0.0204, "step": 31093 }, { "epoch": 1.6577277816281921, "grad_norm": 0.9675889611244202, "learning_rate": 3.7814e-06, "loss": 0.0168, "step": 31094 }, { "epoch": 1.657781095057845, "grad_norm": 1.6765086650848389, "learning_rate": 3.7812000000000003e-06, "loss": 0.0077, "step": 31095 }, { "epoch": 1.6578344084874979, "grad_norm": 2.0012941360473633, "learning_rate": 3.7810000000000003e-06, "loss": 0.013, "step": 31096 }, { "epoch": 1.657887721917151, "grad_norm": 0.6174514889717102, "learning_rate": 3.7808000000000007e-06, "loss": 0.0045, "step": 31097 }, { "epoch": 1.6579410353468038, "grad_norm": 2.04901385307312, "learning_rate": 3.7806000000000003e-06, "loss": 0.0148, "step": 31098 }, { "epoch": 1.6579943487764568, "grad_norm": 1.826479434967041, "learning_rate": 3.7804000000000003e-06, "loss": 0.0258, "step": 31099 }, { "epoch": 1.6580476622061098, "grad_norm": 1.0821648836135864, "learning_rate": 3.7802000000000007e-06, "loss": 0.0067, "step": 31100 }, { "epoch": 1.6581009756357625, "grad_norm": 3.4874517917633057, "learning_rate": 3.7800000000000002e-06, "loss": 0.0123, "step": 31101 }, { "epoch": 1.6581542890654157, "grad_norm": 0.9606499075889587, "learning_rate": 3.7798000000000002e-06, "loss": 0.0242, "step": 31102 }, { "epoch": 1.6582076024950685, "grad_norm": 1.3136168718338013, "learning_rate": 3.7796000000000006e-06, "loss": 0.0226, "step": 31103 }, { "epoch": 1.6582609159247215, "grad_norm": 0.3390035033226013, "learning_rate": 3.7794e-06, "loss": 0.0032, "step": 31104 }, { "epoch": 1.6583142293543744, "grad_norm": 1.9904063940048218, "learning_rate": 3.7792e-06, "loss": 0.0304, "step": 31105 }, { "epoch": 1.6583675427840272, "grad_norm": 1.548007607460022, "learning_rate": 3.7790000000000006e-06, "loss": 0.0138, "step": 31106 }, { "epoch": 1.6584208562136802, "grad_norm": 3.459355115890503, "learning_rate": 3.7788e-06, "loss": 0.0499, "step": 31107 }, { "epoch": 1.6584741696433332, "grad_norm": 2.889171600341797, "learning_rate": 3.7786e-06, "loss": 0.0317, "step": 31108 }, { "epoch": 1.658527483072986, "grad_norm": 1.5611482858657837, "learning_rate": 3.7784000000000005e-06, "loss": 0.0359, "step": 31109 }, { "epoch": 1.6585807965026391, "grad_norm": 1.993411660194397, "learning_rate": 3.7782e-06, "loss": 0.0114, "step": 31110 }, { "epoch": 1.6586341099322919, "grad_norm": 2.351924180984497, "learning_rate": 3.7780000000000005e-06, "loss": 0.028, "step": 31111 }, { "epoch": 1.6586874233619449, "grad_norm": 0.822417676448822, "learning_rate": 3.7778000000000005e-06, "loss": 0.0038, "step": 31112 }, { "epoch": 1.6587407367915978, "grad_norm": 1.345994234085083, "learning_rate": 3.7776e-06, "loss": 0.0154, "step": 31113 }, { "epoch": 1.6587940502212506, "grad_norm": 1.2668217420578003, "learning_rate": 3.7774000000000005e-06, "loss": 0.0125, "step": 31114 }, { "epoch": 1.6588473636509038, "grad_norm": 2.1430978775024414, "learning_rate": 3.7772000000000004e-06, "loss": 0.0209, "step": 31115 }, { "epoch": 1.6589006770805566, "grad_norm": 2.5595345497131348, "learning_rate": 3.777e-06, "loss": 0.0192, "step": 31116 }, { "epoch": 1.6589539905102095, "grad_norm": 1.7332426309585571, "learning_rate": 3.7768000000000004e-06, "loss": 0.0222, "step": 31117 }, { "epoch": 1.6590073039398625, "grad_norm": 1.024718999862671, "learning_rate": 3.7766000000000004e-06, "loss": 0.0074, "step": 31118 }, { "epoch": 1.6590606173695153, "grad_norm": 1.6456035375595093, "learning_rate": 3.7764e-06, "loss": 0.0265, "step": 31119 }, { "epoch": 1.6591139307991685, "grad_norm": 2.5337929725646973, "learning_rate": 3.7762000000000004e-06, "loss": 0.0298, "step": 31120 }, { "epoch": 1.6591672442288212, "grad_norm": 2.0071122646331787, "learning_rate": 3.7760000000000004e-06, "loss": 0.0418, "step": 31121 }, { "epoch": 1.6592205576584742, "grad_norm": 3.25048828125, "learning_rate": 3.7758000000000008e-06, "loss": 0.0296, "step": 31122 }, { "epoch": 1.6592738710881272, "grad_norm": 2.4970011711120605, "learning_rate": 3.7756000000000003e-06, "loss": 0.0258, "step": 31123 }, { "epoch": 1.65932718451778, "grad_norm": 2.7868359088897705, "learning_rate": 3.7754000000000003e-06, "loss": 0.0207, "step": 31124 }, { "epoch": 1.659380497947433, "grad_norm": 1.2938258647918701, "learning_rate": 3.7752000000000007e-06, "loss": 0.0113, "step": 31125 }, { "epoch": 1.659433811377086, "grad_norm": 1.2933624982833862, "learning_rate": 3.7750000000000003e-06, "loss": 0.0073, "step": 31126 }, { "epoch": 1.6594871248067387, "grad_norm": 1.8175289630889893, "learning_rate": 3.7748000000000003e-06, "loss": 0.0138, "step": 31127 }, { "epoch": 1.6595404382363919, "grad_norm": 1.4459279775619507, "learning_rate": 3.7746000000000007e-06, "loss": 0.0117, "step": 31128 }, { "epoch": 1.6595937516660446, "grad_norm": 0.367042601108551, "learning_rate": 3.7744000000000002e-06, "loss": 0.004, "step": 31129 }, { "epoch": 1.6596470650956976, "grad_norm": 1.1495729684829712, "learning_rate": 3.7742000000000002e-06, "loss": 0.0168, "step": 31130 }, { "epoch": 1.6597003785253506, "grad_norm": 1.8962790966033936, "learning_rate": 3.7740000000000006e-06, "loss": 0.0261, "step": 31131 }, { "epoch": 1.6597536919550033, "grad_norm": 1.7434475421905518, "learning_rate": 3.7738e-06, "loss": 0.0144, "step": 31132 }, { "epoch": 1.6598070053846565, "grad_norm": 1.4126195907592773, "learning_rate": 3.7736e-06, "loss": 0.0129, "step": 31133 }, { "epoch": 1.6598603188143093, "grad_norm": 0.46722960472106934, "learning_rate": 3.7734000000000006e-06, "loss": 0.0042, "step": 31134 }, { "epoch": 1.6599136322439623, "grad_norm": 2.1710727214813232, "learning_rate": 3.7732e-06, "loss": 0.0398, "step": 31135 }, { "epoch": 1.6599669456736152, "grad_norm": 2.6336724758148193, "learning_rate": 3.7730000000000006e-06, "loss": 0.0131, "step": 31136 }, { "epoch": 1.660020259103268, "grad_norm": 1.0505462884902954, "learning_rate": 3.7728000000000005e-06, "loss": 0.0076, "step": 31137 }, { "epoch": 1.6600735725329212, "grad_norm": 1.1202017068862915, "learning_rate": 3.7726e-06, "loss": 0.0175, "step": 31138 }, { "epoch": 1.660126885962574, "grad_norm": 1.7964670658111572, "learning_rate": 3.7724000000000005e-06, "loss": 0.0312, "step": 31139 }, { "epoch": 1.660180199392227, "grad_norm": 1.8620362281799316, "learning_rate": 3.7722000000000005e-06, "loss": 0.0263, "step": 31140 }, { "epoch": 1.66023351282188, "grad_norm": 1.5470746755599976, "learning_rate": 3.772e-06, "loss": 0.0219, "step": 31141 }, { "epoch": 1.6602868262515327, "grad_norm": 1.1054922342300415, "learning_rate": 3.7718000000000005e-06, "loss": 0.0065, "step": 31142 }, { "epoch": 1.6603401396811857, "grad_norm": 2.0910351276397705, "learning_rate": 3.7716000000000004e-06, "loss": 0.0313, "step": 31143 }, { "epoch": 1.6603934531108386, "grad_norm": 1.1757590770721436, "learning_rate": 3.7714e-06, "loss": 0.0117, "step": 31144 }, { "epoch": 1.6604467665404914, "grad_norm": 0.9442638754844666, "learning_rate": 3.7712000000000004e-06, "loss": 0.0146, "step": 31145 }, { "epoch": 1.6605000799701446, "grad_norm": 1.3463739156723022, "learning_rate": 3.7710000000000004e-06, "loss": 0.0096, "step": 31146 }, { "epoch": 1.6605533933997974, "grad_norm": 0.44460272789001465, "learning_rate": 3.7708000000000004e-06, "loss": 0.0041, "step": 31147 }, { "epoch": 1.6606067068294503, "grad_norm": 1.2580305337905884, "learning_rate": 3.7706000000000004e-06, "loss": 0.011, "step": 31148 }, { "epoch": 1.6606600202591033, "grad_norm": 2.4734249114990234, "learning_rate": 3.7704000000000004e-06, "loss": 0.051, "step": 31149 }, { "epoch": 1.660713333688756, "grad_norm": 1.7155674695968628, "learning_rate": 3.7702000000000003e-06, "loss": 0.0384, "step": 31150 }, { "epoch": 1.6607666471184093, "grad_norm": 2.53359317779541, "learning_rate": 3.7700000000000003e-06, "loss": 0.018, "step": 31151 }, { "epoch": 1.660819960548062, "grad_norm": 1.1998401880264282, "learning_rate": 3.7698000000000003e-06, "loss": 0.004, "step": 31152 }, { "epoch": 1.660873273977715, "grad_norm": 1.8448123931884766, "learning_rate": 3.7696000000000003e-06, "loss": 0.0354, "step": 31153 }, { "epoch": 1.660926587407368, "grad_norm": 2.5806496143341064, "learning_rate": 3.7694000000000003e-06, "loss": 0.0287, "step": 31154 }, { "epoch": 1.6609799008370207, "grad_norm": 2.5242512226104736, "learning_rate": 3.7692000000000003e-06, "loss": 0.0184, "step": 31155 }, { "epoch": 1.661033214266674, "grad_norm": 2.076476812362671, "learning_rate": 3.7690000000000003e-06, "loss": 0.0221, "step": 31156 }, { "epoch": 1.6610865276963267, "grad_norm": 0.6246517300605774, "learning_rate": 3.7688000000000002e-06, "loss": 0.0051, "step": 31157 }, { "epoch": 1.6611398411259797, "grad_norm": 3.287656545639038, "learning_rate": 3.7686000000000002e-06, "loss": 0.0394, "step": 31158 }, { "epoch": 1.6611931545556327, "grad_norm": 3.0006461143493652, "learning_rate": 3.7684e-06, "loss": 0.0192, "step": 31159 }, { "epoch": 1.6612464679852854, "grad_norm": 2.235997438430786, "learning_rate": 3.7682e-06, "loss": 0.0372, "step": 31160 }, { "epoch": 1.6612997814149384, "grad_norm": 3.7564055919647217, "learning_rate": 3.7680000000000006e-06, "loss": 0.0239, "step": 31161 }, { "epoch": 1.6613530948445914, "grad_norm": 2.8946614265441895, "learning_rate": 3.7678e-06, "loss": 0.0229, "step": 31162 }, { "epoch": 1.6614064082742441, "grad_norm": 1.3088523149490356, "learning_rate": 3.7676e-06, "loss": 0.0154, "step": 31163 }, { "epoch": 1.6614597217038973, "grad_norm": 1.213950276374817, "learning_rate": 3.7674000000000006e-06, "loss": 0.0287, "step": 31164 }, { "epoch": 1.66151303513355, "grad_norm": 1.5427143573760986, "learning_rate": 3.7672e-06, "loss": 0.0201, "step": 31165 }, { "epoch": 1.661566348563203, "grad_norm": 2.1190245151519775, "learning_rate": 3.767e-06, "loss": 0.0295, "step": 31166 }, { "epoch": 1.661619661992856, "grad_norm": 2.7049052715301514, "learning_rate": 3.7668000000000005e-06, "loss": 0.0179, "step": 31167 }, { "epoch": 1.6616729754225088, "grad_norm": 1.2721703052520752, "learning_rate": 3.7666e-06, "loss": 0.0273, "step": 31168 }, { "epoch": 1.661726288852162, "grad_norm": 1.0399855375289917, "learning_rate": 3.7664e-06, "loss": 0.0082, "step": 31169 }, { "epoch": 1.6617796022818148, "grad_norm": 2.8826863765716553, "learning_rate": 3.7662000000000005e-06, "loss": 0.0232, "step": 31170 }, { "epoch": 1.6618329157114677, "grad_norm": 1.616347312927246, "learning_rate": 3.766e-06, "loss": 0.0093, "step": 31171 }, { "epoch": 1.6618862291411207, "grad_norm": 0.7453994154930115, "learning_rate": 3.7658000000000004e-06, "loss": 0.0042, "step": 31172 }, { "epoch": 1.6619395425707735, "grad_norm": 1.4760626554489136, "learning_rate": 3.7656000000000004e-06, "loss": 0.0166, "step": 31173 }, { "epoch": 1.6619928560004265, "grad_norm": 1.9987298250198364, "learning_rate": 3.7654e-06, "loss": 0.041, "step": 31174 }, { "epoch": 1.6620461694300794, "grad_norm": 0.8975518941879272, "learning_rate": 3.7652000000000004e-06, "loss": 0.0146, "step": 31175 }, { "epoch": 1.6620994828597324, "grad_norm": 1.5348249673843384, "learning_rate": 3.7650000000000004e-06, "loss": 0.0317, "step": 31176 }, { "epoch": 1.6621527962893854, "grad_norm": 1.3971706628799438, "learning_rate": 3.7648e-06, "loss": 0.0068, "step": 31177 }, { "epoch": 1.6622061097190381, "grad_norm": 1.2950140237808228, "learning_rate": 3.7646000000000003e-06, "loss": 0.0245, "step": 31178 }, { "epoch": 1.6622594231486911, "grad_norm": 1.8795844316482544, "learning_rate": 3.7644000000000003e-06, "loss": 0.0369, "step": 31179 }, { "epoch": 1.662312736578344, "grad_norm": 1.0954909324645996, "learning_rate": 3.7642e-06, "loss": 0.0218, "step": 31180 }, { "epoch": 1.6623660500079969, "grad_norm": 1.6280522346496582, "learning_rate": 3.7640000000000003e-06, "loss": 0.0365, "step": 31181 }, { "epoch": 1.66241936343765, "grad_norm": 2.0288631916046143, "learning_rate": 3.7638000000000003e-06, "loss": 0.0333, "step": 31182 }, { "epoch": 1.6624726768673028, "grad_norm": 0.8734285235404968, "learning_rate": 3.7636e-06, "loss": 0.0157, "step": 31183 }, { "epoch": 1.6625259902969558, "grad_norm": 4.351791858673096, "learning_rate": 3.7634000000000003e-06, "loss": 0.0184, "step": 31184 }, { "epoch": 1.6625793037266088, "grad_norm": 1.5557847023010254, "learning_rate": 3.7632000000000002e-06, "loss": 0.0122, "step": 31185 }, { "epoch": 1.6626326171562615, "grad_norm": 1.5466575622558594, "learning_rate": 3.7630000000000006e-06, "loss": 0.0074, "step": 31186 }, { "epoch": 1.6626859305859147, "grad_norm": 2.442270278930664, "learning_rate": 3.7628e-06, "loss": 0.0679, "step": 31187 }, { "epoch": 1.6627392440155675, "grad_norm": 1.6202223300933838, "learning_rate": 3.7626e-06, "loss": 0.0283, "step": 31188 }, { "epoch": 1.6627925574452205, "grad_norm": 1.3039449453353882, "learning_rate": 3.7624000000000006e-06, "loss": 0.0094, "step": 31189 }, { "epoch": 1.6628458708748735, "grad_norm": 2.2620179653167725, "learning_rate": 3.7622e-06, "loss": 0.0605, "step": 31190 }, { "epoch": 1.6628991843045262, "grad_norm": 2.679459571838379, "learning_rate": 3.762e-06, "loss": 0.0284, "step": 31191 }, { "epoch": 1.6629524977341792, "grad_norm": 2.6580605506896973, "learning_rate": 3.7618000000000006e-06, "loss": 0.068, "step": 31192 }, { "epoch": 1.6630058111638322, "grad_norm": 1.6104172468185425, "learning_rate": 3.7616e-06, "loss": 0.0101, "step": 31193 }, { "epoch": 1.6630591245934851, "grad_norm": 1.095051646232605, "learning_rate": 3.7614e-06, "loss": 0.0084, "step": 31194 }, { "epoch": 1.6631124380231381, "grad_norm": 1.4105041027069092, "learning_rate": 3.7612000000000005e-06, "loss": 0.0166, "step": 31195 }, { "epoch": 1.6631657514527909, "grad_norm": 1.9530330896377563, "learning_rate": 3.761e-06, "loss": 0.0089, "step": 31196 }, { "epoch": 1.6632190648824439, "grad_norm": 0.9955086708068848, "learning_rate": 3.7608000000000005e-06, "loss": 0.017, "step": 31197 }, { "epoch": 1.6632723783120968, "grad_norm": 1.31621253490448, "learning_rate": 3.7606000000000005e-06, "loss": 0.0525, "step": 31198 }, { "epoch": 1.6633256917417496, "grad_norm": 2.695038080215454, "learning_rate": 3.7604e-06, "loss": 0.0469, "step": 31199 }, { "epoch": 1.6633790051714028, "grad_norm": 2.673487901687622, "learning_rate": 3.7602000000000004e-06, "loss": 0.0334, "step": 31200 }, { "epoch": 1.6634323186010556, "grad_norm": 1.823184847831726, "learning_rate": 3.7600000000000004e-06, "loss": 0.0174, "step": 31201 }, { "epoch": 1.6634856320307085, "grad_norm": 1.2124665975570679, "learning_rate": 3.7598e-06, "loss": 0.027, "step": 31202 }, { "epoch": 1.6635389454603615, "grad_norm": 1.5438215732574463, "learning_rate": 3.7596000000000004e-06, "loss": 0.0124, "step": 31203 }, { "epoch": 1.6635922588900143, "grad_norm": 1.1401464939117432, "learning_rate": 3.7594000000000004e-06, "loss": 0.0113, "step": 31204 }, { "epoch": 1.6636455723196675, "grad_norm": 1.755327820777893, "learning_rate": 3.7592e-06, "loss": 0.0205, "step": 31205 }, { "epoch": 1.6636988857493202, "grad_norm": 1.286167860031128, "learning_rate": 3.7590000000000003e-06, "loss": 0.0204, "step": 31206 }, { "epoch": 1.6637521991789732, "grad_norm": 4.209733009338379, "learning_rate": 3.7588000000000003e-06, "loss": 0.036, "step": 31207 }, { "epoch": 1.6638055126086262, "grad_norm": 1.426430106163025, "learning_rate": 3.7586000000000007e-06, "loss": 0.0319, "step": 31208 }, { "epoch": 1.663858826038279, "grad_norm": 1.6563094854354858, "learning_rate": 3.7584000000000003e-06, "loss": 0.0301, "step": 31209 }, { "epoch": 1.663912139467932, "grad_norm": 1.0443964004516602, "learning_rate": 3.7582000000000003e-06, "loss": 0.0123, "step": 31210 }, { "epoch": 1.663965452897585, "grad_norm": 1.2619601488113403, "learning_rate": 3.7580000000000007e-06, "loss": 0.0295, "step": 31211 }, { "epoch": 1.6640187663272379, "grad_norm": 1.553687334060669, "learning_rate": 3.7578000000000003e-06, "loss": 0.0262, "step": 31212 }, { "epoch": 1.6640720797568909, "grad_norm": 2.0400280952453613, "learning_rate": 3.7576000000000002e-06, "loss": 0.0507, "step": 31213 }, { "epoch": 1.6641253931865436, "grad_norm": 2.19303560256958, "learning_rate": 3.7574000000000007e-06, "loss": 0.0114, "step": 31214 }, { "epoch": 1.6641787066161966, "grad_norm": 2.7320291996002197, "learning_rate": 3.7572000000000002e-06, "loss": 0.0396, "step": 31215 }, { "epoch": 1.6642320200458496, "grad_norm": 2.9661552906036377, "learning_rate": 3.757e-06, "loss": 0.0375, "step": 31216 }, { "epoch": 1.6642853334755023, "grad_norm": 1.4928418397903442, "learning_rate": 3.7568000000000006e-06, "loss": 0.0117, "step": 31217 }, { "epoch": 1.6643386469051555, "grad_norm": 1.6089609861373901, "learning_rate": 3.7566e-06, "loss": 0.0355, "step": 31218 }, { "epoch": 1.6643919603348083, "grad_norm": 1.5158231258392334, "learning_rate": 3.7564e-06, "loss": 0.0123, "step": 31219 }, { "epoch": 1.6644452737644613, "grad_norm": 1.6754565238952637, "learning_rate": 3.7562000000000006e-06, "loss": 0.0175, "step": 31220 }, { "epoch": 1.6644985871941143, "grad_norm": 1.1898220777511597, "learning_rate": 3.756e-06, "loss": 0.0137, "step": 31221 }, { "epoch": 1.664551900623767, "grad_norm": 0.47442686557769775, "learning_rate": 3.7558000000000005e-06, "loss": 0.0051, "step": 31222 }, { "epoch": 1.6646052140534202, "grad_norm": 1.877144694328308, "learning_rate": 3.7556000000000005e-06, "loss": 0.0113, "step": 31223 }, { "epoch": 1.664658527483073, "grad_norm": 1.3965752124786377, "learning_rate": 3.7554e-06, "loss": 0.0109, "step": 31224 }, { "epoch": 1.664711840912726, "grad_norm": 1.3347833156585693, "learning_rate": 3.7552000000000005e-06, "loss": 0.0165, "step": 31225 }, { "epoch": 1.664765154342379, "grad_norm": 1.9011324644088745, "learning_rate": 3.7550000000000005e-06, "loss": 0.0541, "step": 31226 }, { "epoch": 1.6648184677720317, "grad_norm": 1.4981269836425781, "learning_rate": 3.7548e-06, "loss": 0.0153, "step": 31227 }, { "epoch": 1.6648717812016847, "grad_norm": 1.916077733039856, "learning_rate": 3.7546000000000004e-06, "loss": 0.0162, "step": 31228 }, { "epoch": 1.6649250946313376, "grad_norm": 1.9988480806350708, "learning_rate": 3.7544000000000004e-06, "loss": 0.0196, "step": 31229 }, { "epoch": 1.6649784080609904, "grad_norm": 1.3581684827804565, "learning_rate": 3.7542e-06, "loss": 0.0231, "step": 31230 }, { "epoch": 1.6650317214906436, "grad_norm": 2.719975471496582, "learning_rate": 3.7540000000000004e-06, "loss": 0.0168, "step": 31231 }, { "epoch": 1.6650850349202964, "grad_norm": 0.8721626400947571, "learning_rate": 3.7538000000000004e-06, "loss": 0.0111, "step": 31232 }, { "epoch": 1.6651383483499493, "grad_norm": 1.9888523817062378, "learning_rate": 3.7536000000000004e-06, "loss": 0.0472, "step": 31233 }, { "epoch": 1.6651916617796023, "grad_norm": 1.826460599899292, "learning_rate": 3.7534000000000004e-06, "loss": 0.0297, "step": 31234 }, { "epoch": 1.665244975209255, "grad_norm": 1.4585466384887695, "learning_rate": 3.7532000000000003e-06, "loss": 0.0353, "step": 31235 }, { "epoch": 1.6652982886389083, "grad_norm": 1.2360719442367554, "learning_rate": 3.7530000000000003e-06, "loss": 0.0142, "step": 31236 }, { "epoch": 1.665351602068561, "grad_norm": 2.052006721496582, "learning_rate": 3.7528000000000003e-06, "loss": 0.0164, "step": 31237 }, { "epoch": 1.665404915498214, "grad_norm": 1.2302675247192383, "learning_rate": 3.7526000000000003e-06, "loss": 0.0126, "step": 31238 }, { "epoch": 1.665458228927867, "grad_norm": 1.4594711065292358, "learning_rate": 3.7524000000000003e-06, "loss": 0.0181, "step": 31239 }, { "epoch": 1.6655115423575197, "grad_norm": 1.3665879964828491, "learning_rate": 3.7522000000000003e-06, "loss": 0.0263, "step": 31240 }, { "epoch": 1.665564855787173, "grad_norm": 1.2075018882751465, "learning_rate": 3.7520000000000002e-06, "loss": 0.0127, "step": 31241 }, { "epoch": 1.6656181692168257, "grad_norm": 0.8803612589836121, "learning_rate": 3.7518000000000002e-06, "loss": 0.007, "step": 31242 }, { "epoch": 1.6656714826464787, "grad_norm": 0.5656929016113281, "learning_rate": 3.7516000000000002e-06, "loss": 0.0037, "step": 31243 }, { "epoch": 1.6657247960761317, "grad_norm": 1.8073773384094238, "learning_rate": 3.7514e-06, "loss": 0.0344, "step": 31244 }, { "epoch": 1.6657781095057844, "grad_norm": 2.9722917079925537, "learning_rate": 3.7512e-06, "loss": 0.0574, "step": 31245 }, { "epoch": 1.6658314229354374, "grad_norm": 0.9809895753860474, "learning_rate": 3.751e-06, "loss": 0.01, "step": 31246 }, { "epoch": 1.6658847363650904, "grad_norm": 1.442887783050537, "learning_rate": 3.7508000000000006e-06, "loss": 0.0224, "step": 31247 }, { "epoch": 1.6659380497947431, "grad_norm": 1.4227405786514282, "learning_rate": 3.7506e-06, "loss": 0.0176, "step": 31248 }, { "epoch": 1.6659913632243963, "grad_norm": 1.1296306848526, "learning_rate": 3.7504e-06, "loss": 0.0218, "step": 31249 }, { "epoch": 1.666044676654049, "grad_norm": 2.4029600620269775, "learning_rate": 3.7502000000000005e-06, "loss": 0.0229, "step": 31250 }, { "epoch": 1.666097990083702, "grad_norm": 1.0166974067687988, "learning_rate": 3.7500000000000005e-06, "loss": 0.0094, "step": 31251 }, { "epoch": 1.666151303513355, "grad_norm": 1.5056918859481812, "learning_rate": 3.7498e-06, "loss": 0.0289, "step": 31252 }, { "epoch": 1.6662046169430078, "grad_norm": 1.7691680192947388, "learning_rate": 3.7496000000000005e-06, "loss": 0.029, "step": 31253 }, { "epoch": 1.666257930372661, "grad_norm": 1.3373879194259644, "learning_rate": 3.7494000000000005e-06, "loss": 0.0092, "step": 31254 }, { "epoch": 1.6663112438023138, "grad_norm": 2.250033378601074, "learning_rate": 3.7492e-06, "loss": 0.0114, "step": 31255 }, { "epoch": 1.6663645572319667, "grad_norm": 1.9672448635101318, "learning_rate": 3.7490000000000004e-06, "loss": 0.0156, "step": 31256 }, { "epoch": 1.6664178706616197, "grad_norm": 2.282343626022339, "learning_rate": 3.7488000000000004e-06, "loss": 0.0339, "step": 31257 }, { "epoch": 1.6664711840912725, "grad_norm": 1.4318609237670898, "learning_rate": 3.7486000000000004e-06, "loss": 0.0145, "step": 31258 }, { "epoch": 1.6665244975209257, "grad_norm": 2.4277122020721436, "learning_rate": 3.7484000000000004e-06, "loss": 0.0238, "step": 31259 }, { "epoch": 1.6665778109505784, "grad_norm": 0.7505232095718384, "learning_rate": 3.7482000000000004e-06, "loss": 0.0059, "step": 31260 }, { "epoch": 1.6666311243802314, "grad_norm": 0.5527081489562988, "learning_rate": 3.7480000000000004e-06, "loss": 0.0048, "step": 31261 }, { "epoch": 1.6666844378098844, "grad_norm": 2.000431776046753, "learning_rate": 3.7478000000000004e-06, "loss": 0.0236, "step": 31262 }, { "epoch": 1.6667377512395372, "grad_norm": 1.5624390840530396, "learning_rate": 3.7476000000000003e-06, "loss": 0.024, "step": 31263 }, { "epoch": 1.6667910646691901, "grad_norm": 3.4226346015930176, "learning_rate": 3.7474000000000003e-06, "loss": 0.044, "step": 31264 }, { "epoch": 1.6668443780988431, "grad_norm": 2.1037678718566895, "learning_rate": 3.7472000000000003e-06, "loss": 0.0376, "step": 31265 }, { "epoch": 1.6668976915284959, "grad_norm": 1.0612725019454956, "learning_rate": 3.7470000000000003e-06, "loss": 0.0134, "step": 31266 }, { "epoch": 1.666951004958149, "grad_norm": 1.4530596733093262, "learning_rate": 3.7468000000000003e-06, "loss": 0.0177, "step": 31267 }, { "epoch": 1.6670043183878018, "grad_norm": 0.2775772511959076, "learning_rate": 3.7466000000000003e-06, "loss": 0.0022, "step": 31268 }, { "epoch": 1.6670576318174548, "grad_norm": 4.118900299072266, "learning_rate": 3.7464000000000002e-06, "loss": 0.0232, "step": 31269 }, { "epoch": 1.6671109452471078, "grad_norm": 2.685029983520508, "learning_rate": 3.7462000000000002e-06, "loss": 0.0267, "step": 31270 }, { "epoch": 1.6671642586767605, "grad_norm": 2.05444073677063, "learning_rate": 3.7460000000000002e-06, "loss": 0.0245, "step": 31271 }, { "epoch": 1.6672175721064137, "grad_norm": 1.4974573850631714, "learning_rate": 3.7458000000000006e-06, "loss": 0.0089, "step": 31272 }, { "epoch": 1.6672708855360665, "grad_norm": 2.246272087097168, "learning_rate": 3.7456e-06, "loss": 0.0203, "step": 31273 }, { "epoch": 1.6673241989657195, "grad_norm": 2.2166457176208496, "learning_rate": 3.7454e-06, "loss": 0.0122, "step": 31274 }, { "epoch": 1.6673775123953725, "grad_norm": 2.1302895545959473, "learning_rate": 3.7452000000000006e-06, "loss": 0.0206, "step": 31275 }, { "epoch": 1.6674308258250252, "grad_norm": 5.89599084854126, "learning_rate": 3.745e-06, "loss": 0.0716, "step": 31276 }, { "epoch": 1.6674841392546782, "grad_norm": 1.6653094291687012, "learning_rate": 3.7448e-06, "loss": 0.0286, "step": 31277 }, { "epoch": 1.6675374526843312, "grad_norm": 1.0374531745910645, "learning_rate": 3.7446000000000005e-06, "loss": 0.0157, "step": 31278 }, { "epoch": 1.6675907661139842, "grad_norm": 2.098802328109741, "learning_rate": 3.7444e-06, "loss": 0.0215, "step": 31279 }, { "epoch": 1.6676440795436371, "grad_norm": 1.3517801761627197, "learning_rate": 3.7442e-06, "loss": 0.0363, "step": 31280 }, { "epoch": 1.66769739297329, "grad_norm": 1.4111963510513306, "learning_rate": 3.7440000000000005e-06, "loss": 0.0124, "step": 31281 }, { "epoch": 1.6677507064029429, "grad_norm": 1.6370888948440552, "learning_rate": 3.7438e-06, "loss": 0.0224, "step": 31282 }, { "epoch": 1.6678040198325959, "grad_norm": 2.5080270767211914, "learning_rate": 3.7436000000000005e-06, "loss": 0.0198, "step": 31283 }, { "epoch": 1.6678573332622486, "grad_norm": 1.205781102180481, "learning_rate": 3.7434000000000004e-06, "loss": 0.0246, "step": 31284 }, { "epoch": 1.6679106466919018, "grad_norm": 1.6989467144012451, "learning_rate": 3.7432e-06, "loss": 0.0195, "step": 31285 }, { "epoch": 1.6679639601215546, "grad_norm": 0.31003904342651367, "learning_rate": 3.7430000000000004e-06, "loss": 0.0032, "step": 31286 }, { "epoch": 1.6680172735512075, "grad_norm": 1.1769993305206299, "learning_rate": 3.7428000000000004e-06, "loss": 0.0193, "step": 31287 }, { "epoch": 1.6680705869808605, "grad_norm": 2.9627091884613037, "learning_rate": 3.7426e-06, "loss": 0.022, "step": 31288 }, { "epoch": 1.6681239004105133, "grad_norm": 2.0000548362731934, "learning_rate": 3.7424000000000004e-06, "loss": 0.0491, "step": 31289 }, { "epoch": 1.6681772138401665, "grad_norm": 1.987165927886963, "learning_rate": 3.7422000000000004e-06, "loss": 0.0372, "step": 31290 }, { "epoch": 1.6682305272698192, "grad_norm": 2.3858838081359863, "learning_rate": 3.742e-06, "loss": 0.0178, "step": 31291 }, { "epoch": 1.6682838406994722, "grad_norm": 1.4617358446121216, "learning_rate": 3.7418000000000003e-06, "loss": 0.0126, "step": 31292 }, { "epoch": 1.6683371541291252, "grad_norm": 1.8988667726516724, "learning_rate": 3.7416000000000003e-06, "loss": 0.0148, "step": 31293 }, { "epoch": 1.668390467558778, "grad_norm": 1.5712884664535522, "learning_rate": 3.7414e-06, "loss": 0.0097, "step": 31294 }, { "epoch": 1.668443780988431, "grad_norm": 2.5685582160949707, "learning_rate": 3.7412000000000003e-06, "loss": 0.0356, "step": 31295 }, { "epoch": 1.668497094418084, "grad_norm": 0.7234439849853516, "learning_rate": 3.7410000000000003e-06, "loss": 0.0078, "step": 31296 }, { "epoch": 1.668550407847737, "grad_norm": 2.1473324298858643, "learning_rate": 3.7408000000000007e-06, "loss": 0.0433, "step": 31297 }, { "epoch": 1.6686037212773899, "grad_norm": 0.7244402766227722, "learning_rate": 3.7406000000000002e-06, "loss": 0.0068, "step": 31298 }, { "epoch": 1.6686570347070426, "grad_norm": 1.2768020629882812, "learning_rate": 3.7404000000000002e-06, "loss": 0.0218, "step": 31299 }, { "epoch": 1.6687103481366956, "grad_norm": 0.7823410034179688, "learning_rate": 3.7402000000000006e-06, "loss": 0.0082, "step": 31300 }, { "epoch": 1.6687636615663486, "grad_norm": 0.8305178284645081, "learning_rate": 3.74e-06, "loss": 0.0073, "step": 31301 }, { "epoch": 1.6688169749960013, "grad_norm": 0.29658499360084534, "learning_rate": 3.7398e-06, "loss": 0.0026, "step": 31302 }, { "epoch": 1.6688702884256545, "grad_norm": 1.3442118167877197, "learning_rate": 3.7396000000000006e-06, "loss": 0.0216, "step": 31303 }, { "epoch": 1.6689236018553073, "grad_norm": 1.4891746044158936, "learning_rate": 3.7394e-06, "loss": 0.0136, "step": 31304 }, { "epoch": 1.6689769152849603, "grad_norm": 1.0896793603897095, "learning_rate": 3.7392e-06, "loss": 0.0182, "step": 31305 }, { "epoch": 1.6690302287146133, "grad_norm": 1.5827898979187012, "learning_rate": 3.7390000000000005e-06, "loss": 0.0251, "step": 31306 }, { "epoch": 1.669083542144266, "grad_norm": 1.2844040393829346, "learning_rate": 3.7388e-06, "loss": 0.0199, "step": 31307 }, { "epoch": 1.6691368555739192, "grad_norm": 1.6134536266326904, "learning_rate": 3.7386000000000005e-06, "loss": 0.0193, "step": 31308 }, { "epoch": 1.669190169003572, "grad_norm": 2.0105206966400146, "learning_rate": 3.7384000000000005e-06, "loss": 0.0237, "step": 31309 }, { "epoch": 1.669243482433225, "grad_norm": 2.581116199493408, "learning_rate": 3.7382e-06, "loss": 0.0364, "step": 31310 }, { "epoch": 1.669296795862878, "grad_norm": 1.1659427881240845, "learning_rate": 3.7380000000000005e-06, "loss": 0.0297, "step": 31311 }, { "epoch": 1.6693501092925307, "grad_norm": 2.9817569255828857, "learning_rate": 3.7378000000000004e-06, "loss": 0.0284, "step": 31312 }, { "epoch": 1.6694034227221837, "grad_norm": 1.159570574760437, "learning_rate": 3.7376e-06, "loss": 0.0104, "step": 31313 }, { "epoch": 1.6694567361518367, "grad_norm": 0.5773642659187317, "learning_rate": 3.7374000000000004e-06, "loss": 0.0072, "step": 31314 }, { "epoch": 1.6695100495814896, "grad_norm": 3.2416024208068848, "learning_rate": 3.7372000000000004e-06, "loss": 0.0501, "step": 31315 }, { "epoch": 1.6695633630111426, "grad_norm": 0.9990352392196655, "learning_rate": 3.737e-06, "loss": 0.0059, "step": 31316 }, { "epoch": 1.6696166764407954, "grad_norm": 3.929091691970825, "learning_rate": 3.7368000000000004e-06, "loss": 0.032, "step": 31317 }, { "epoch": 1.6696699898704483, "grad_norm": 1.1292804479599, "learning_rate": 3.7366000000000004e-06, "loss": 0.0141, "step": 31318 }, { "epoch": 1.6697233033001013, "grad_norm": 1.768540859222412, "learning_rate": 3.7364000000000008e-06, "loss": 0.0395, "step": 31319 }, { "epoch": 1.669776616729754, "grad_norm": 0.6586485505104065, "learning_rate": 3.7362000000000003e-06, "loss": 0.006, "step": 31320 }, { "epoch": 1.6698299301594073, "grad_norm": 2.2999303340911865, "learning_rate": 3.7360000000000003e-06, "loss": 0.0235, "step": 31321 }, { "epoch": 1.66988324358906, "grad_norm": 2.4537806510925293, "learning_rate": 3.7358000000000007e-06, "loss": 0.0224, "step": 31322 }, { "epoch": 1.669936557018713, "grad_norm": 1.366281270980835, "learning_rate": 3.7356000000000003e-06, "loss": 0.0137, "step": 31323 }, { "epoch": 1.669989870448366, "grad_norm": 3.638796091079712, "learning_rate": 3.7354000000000003e-06, "loss": 0.0112, "step": 31324 }, { "epoch": 1.6700431838780188, "grad_norm": 0.8897467255592346, "learning_rate": 3.7352000000000007e-06, "loss": 0.0055, "step": 31325 }, { "epoch": 1.670096497307672, "grad_norm": 2.097693920135498, "learning_rate": 3.7350000000000002e-06, "loss": 0.0293, "step": 31326 }, { "epoch": 1.6701498107373247, "grad_norm": 0.8484635353088379, "learning_rate": 3.7348000000000002e-06, "loss": 0.0064, "step": 31327 }, { "epoch": 1.6702031241669777, "grad_norm": 1.3105157613754272, "learning_rate": 3.7346000000000006e-06, "loss": 0.0199, "step": 31328 }, { "epoch": 1.6702564375966307, "grad_norm": 2.486560821533203, "learning_rate": 3.7344e-06, "loss": 0.0172, "step": 31329 }, { "epoch": 1.6703097510262834, "grad_norm": 1.3201979398727417, "learning_rate": 3.7342e-06, "loss": 0.0137, "step": 31330 }, { "epoch": 1.6703630644559364, "grad_norm": 3.9601426124572754, "learning_rate": 3.7340000000000006e-06, "loss": 0.0343, "step": 31331 }, { "epoch": 1.6704163778855894, "grad_norm": 2.3295040130615234, "learning_rate": 3.7338e-06, "loss": 0.0393, "step": 31332 }, { "epoch": 1.6704696913152421, "grad_norm": 0.3724411725997925, "learning_rate": 3.7336000000000006e-06, "loss": 0.0034, "step": 31333 }, { "epoch": 1.6705230047448953, "grad_norm": 0.7156040072441101, "learning_rate": 3.7334000000000005e-06, "loss": 0.0123, "step": 31334 }, { "epoch": 1.670576318174548, "grad_norm": 1.8300482034683228, "learning_rate": 3.7332e-06, "loss": 0.0347, "step": 31335 }, { "epoch": 1.670629631604201, "grad_norm": 0.4771328568458557, "learning_rate": 3.7330000000000005e-06, "loss": 0.0042, "step": 31336 }, { "epoch": 1.670682945033854, "grad_norm": 1.6001579761505127, "learning_rate": 3.7328000000000005e-06, "loss": 0.0147, "step": 31337 }, { "epoch": 1.6707362584635068, "grad_norm": 1.8655704259872437, "learning_rate": 3.7326e-06, "loss": 0.0125, "step": 31338 }, { "epoch": 1.67078957189316, "grad_norm": 0.7480342984199524, "learning_rate": 3.7324000000000005e-06, "loss": 0.0055, "step": 31339 }, { "epoch": 1.6708428853228128, "grad_norm": 2.54770565032959, "learning_rate": 3.7322000000000005e-06, "loss": 0.0654, "step": 31340 }, { "epoch": 1.6708961987524658, "grad_norm": 1.0400398969650269, "learning_rate": 3.732e-06, "loss": 0.0158, "step": 31341 }, { "epoch": 1.6709495121821187, "grad_norm": 1.2084203958511353, "learning_rate": 3.7318000000000004e-06, "loss": 0.0071, "step": 31342 }, { "epoch": 1.6710028256117715, "grad_norm": 1.1344032287597656, "learning_rate": 3.7316000000000004e-06, "loss": 0.0245, "step": 31343 }, { "epoch": 1.6710561390414247, "grad_norm": 1.3191477060317993, "learning_rate": 3.7314000000000004e-06, "loss": 0.0178, "step": 31344 }, { "epoch": 1.6711094524710775, "grad_norm": 1.2527583837509155, "learning_rate": 3.7312000000000004e-06, "loss": 0.0057, "step": 31345 }, { "epoch": 1.6711627659007304, "grad_norm": 1.0573028326034546, "learning_rate": 3.7310000000000004e-06, "loss": 0.0206, "step": 31346 }, { "epoch": 1.6712160793303834, "grad_norm": 1.3634670972824097, "learning_rate": 3.7308000000000003e-06, "loss": 0.0224, "step": 31347 }, { "epoch": 1.6712693927600362, "grad_norm": 3.1113288402557373, "learning_rate": 3.7306000000000003e-06, "loss": 0.0266, "step": 31348 }, { "epoch": 1.6713227061896891, "grad_norm": 1.846343755722046, "learning_rate": 3.7304000000000003e-06, "loss": 0.0125, "step": 31349 }, { "epoch": 1.6713760196193421, "grad_norm": 1.910599708557129, "learning_rate": 3.7302000000000003e-06, "loss": 0.0082, "step": 31350 }, { "epoch": 1.6714293330489949, "grad_norm": 2.595750331878662, "learning_rate": 3.7300000000000003e-06, "loss": 0.0302, "step": 31351 }, { "epoch": 1.671482646478648, "grad_norm": 2.855203628540039, "learning_rate": 3.7298000000000003e-06, "loss": 0.0299, "step": 31352 }, { "epoch": 1.6715359599083008, "grad_norm": 2.1656548976898193, "learning_rate": 3.7296000000000003e-06, "loss": 0.0327, "step": 31353 }, { "epoch": 1.6715892733379538, "grad_norm": 0.986103355884552, "learning_rate": 3.7294000000000002e-06, "loss": 0.0048, "step": 31354 }, { "epoch": 1.6716425867676068, "grad_norm": 1.8719350099563599, "learning_rate": 3.7292000000000002e-06, "loss": 0.0345, "step": 31355 }, { "epoch": 1.6716959001972596, "grad_norm": 2.0394489765167236, "learning_rate": 3.729e-06, "loss": 0.0196, "step": 31356 }, { "epoch": 1.6717492136269128, "grad_norm": 2.2815003395080566, "learning_rate": 3.7288e-06, "loss": 0.0164, "step": 31357 }, { "epoch": 1.6718025270565655, "grad_norm": 5.683127403259277, "learning_rate": 3.7286000000000006e-06, "loss": 0.0482, "step": 31358 }, { "epoch": 1.6718558404862185, "grad_norm": 5.103538513183594, "learning_rate": 3.7284e-06, "loss": 0.0159, "step": 31359 }, { "epoch": 1.6719091539158715, "grad_norm": 1.1681796312332153, "learning_rate": 3.7282e-06, "loss": 0.0092, "step": 31360 }, { "epoch": 1.6719624673455242, "grad_norm": 1.1424651145935059, "learning_rate": 3.7280000000000006e-06, "loss": 0.0066, "step": 31361 }, { "epoch": 1.6720157807751774, "grad_norm": 2.6602187156677246, "learning_rate": 3.7278e-06, "loss": 0.0416, "step": 31362 }, { "epoch": 1.6720690942048302, "grad_norm": 1.6646206378936768, "learning_rate": 3.7276e-06, "loss": 0.0199, "step": 31363 }, { "epoch": 1.6721224076344832, "grad_norm": 1.8744754791259766, "learning_rate": 3.7274000000000005e-06, "loss": 0.0315, "step": 31364 }, { "epoch": 1.6721757210641361, "grad_norm": 0.2782853841781616, "learning_rate": 3.7272e-06, "loss": 0.0028, "step": 31365 }, { "epoch": 1.672229034493789, "grad_norm": 0.4635827839374542, "learning_rate": 3.727e-06, "loss": 0.0049, "step": 31366 }, { "epoch": 1.6722823479234419, "grad_norm": 0.3139401376247406, "learning_rate": 3.7268000000000005e-06, "loss": 0.0031, "step": 31367 }, { "epoch": 1.6723356613530949, "grad_norm": 2.069234848022461, "learning_rate": 3.7266e-06, "loss": 0.0331, "step": 31368 }, { "epoch": 1.6723889747827476, "grad_norm": 1.3256443738937378, "learning_rate": 3.7264000000000004e-06, "loss": 0.0114, "step": 31369 }, { "epoch": 1.6724422882124008, "grad_norm": 1.4286843538284302, "learning_rate": 3.7262000000000004e-06, "loss": 0.0099, "step": 31370 }, { "epoch": 1.6724956016420536, "grad_norm": 1.5232915878295898, "learning_rate": 3.726e-06, "loss": 0.0167, "step": 31371 }, { "epoch": 1.6725489150717066, "grad_norm": 1.0759893655776978, "learning_rate": 3.7258000000000004e-06, "loss": 0.0175, "step": 31372 }, { "epoch": 1.6726022285013595, "grad_norm": 1.9526320695877075, "learning_rate": 3.7256000000000004e-06, "loss": 0.0134, "step": 31373 }, { "epoch": 1.6726555419310123, "grad_norm": 3.146268606185913, "learning_rate": 3.7254e-06, "loss": 0.0198, "step": 31374 }, { "epoch": 1.6727088553606655, "grad_norm": 1.198149561882019, "learning_rate": 3.7252000000000003e-06, "loss": 0.0224, "step": 31375 }, { "epoch": 1.6727621687903182, "grad_norm": 2.5728631019592285, "learning_rate": 3.7250000000000003e-06, "loss": 0.0167, "step": 31376 }, { "epoch": 1.6728154822199712, "grad_norm": 1.453486442565918, "learning_rate": 3.7248e-06, "loss": 0.0082, "step": 31377 }, { "epoch": 1.6728687956496242, "grad_norm": 0.6966763734817505, "learning_rate": 3.7246000000000003e-06, "loss": 0.0034, "step": 31378 }, { "epoch": 1.672922109079277, "grad_norm": 5.397365093231201, "learning_rate": 3.7244000000000003e-06, "loss": 0.0475, "step": 31379 }, { "epoch": 1.6729754225089302, "grad_norm": 0.2527550458908081, "learning_rate": 3.7242e-06, "loss": 0.0024, "step": 31380 }, { "epoch": 1.673028735938583, "grad_norm": 2.708099603652954, "learning_rate": 3.7240000000000003e-06, "loss": 0.028, "step": 31381 }, { "epoch": 1.673082049368236, "grad_norm": 2.76773738861084, "learning_rate": 3.7238000000000002e-06, "loss": 0.0193, "step": 31382 }, { "epoch": 1.6731353627978889, "grad_norm": 1.9709510803222656, "learning_rate": 3.7236000000000007e-06, "loss": 0.0195, "step": 31383 }, { "epoch": 1.6731886762275416, "grad_norm": 1.9608001708984375, "learning_rate": 3.7234000000000002e-06, "loss": 0.0412, "step": 31384 }, { "epoch": 1.6732419896571946, "grad_norm": 2.0315160751342773, "learning_rate": 3.7232e-06, "loss": 0.0368, "step": 31385 }, { "epoch": 1.6732953030868476, "grad_norm": 1.5576480627059937, "learning_rate": 3.7230000000000006e-06, "loss": 0.0156, "step": 31386 }, { "epoch": 1.6733486165165004, "grad_norm": 1.765841007232666, "learning_rate": 3.7228e-06, "loss": 0.0454, "step": 31387 }, { "epoch": 1.6734019299461536, "grad_norm": 1.606603980064392, "learning_rate": 3.7226e-06, "loss": 0.0111, "step": 31388 }, { "epoch": 1.6734552433758063, "grad_norm": 2.1978068351745605, "learning_rate": 3.7224000000000006e-06, "loss": 0.0149, "step": 31389 }, { "epoch": 1.6735085568054593, "grad_norm": 2.794542074203491, "learning_rate": 3.7222e-06, "loss": 0.0493, "step": 31390 }, { "epoch": 1.6735618702351123, "grad_norm": 1.5247150659561157, "learning_rate": 3.722e-06, "loss": 0.014, "step": 31391 }, { "epoch": 1.673615183664765, "grad_norm": 2.0408928394317627, "learning_rate": 3.7218000000000005e-06, "loss": 0.0142, "step": 31392 }, { "epoch": 1.6736684970944182, "grad_norm": 2.227898120880127, "learning_rate": 3.7216e-06, "loss": 0.0122, "step": 31393 }, { "epoch": 1.673721810524071, "grad_norm": 1.7312121391296387, "learning_rate": 3.7214000000000005e-06, "loss": 0.0168, "step": 31394 }, { "epoch": 1.673775123953724, "grad_norm": 2.0576367378234863, "learning_rate": 3.7212000000000005e-06, "loss": 0.0586, "step": 31395 }, { "epoch": 1.673828437383377, "grad_norm": 1.387189269065857, "learning_rate": 3.721e-06, "loss": 0.0335, "step": 31396 }, { "epoch": 1.6738817508130297, "grad_norm": 2.060405731201172, "learning_rate": 3.7208000000000004e-06, "loss": 0.038, "step": 31397 }, { "epoch": 1.6739350642426827, "grad_norm": 1.864571213722229, "learning_rate": 3.7206000000000004e-06, "loss": 0.0081, "step": 31398 }, { "epoch": 1.6739883776723357, "grad_norm": 6.652781963348389, "learning_rate": 3.7204e-06, "loss": 0.0434, "step": 31399 }, { "epoch": 1.6740416911019886, "grad_norm": 1.675276517868042, "learning_rate": 3.7202000000000004e-06, "loss": 0.0129, "step": 31400 }, { "epoch": 1.6740950045316416, "grad_norm": 1.8371350765228271, "learning_rate": 3.7200000000000004e-06, "loss": 0.0246, "step": 31401 }, { "epoch": 1.6741483179612944, "grad_norm": 1.7947057485580444, "learning_rate": 3.7198e-06, "loss": 0.0223, "step": 31402 }, { "epoch": 1.6742016313909474, "grad_norm": 1.1697797775268555, "learning_rate": 3.7196000000000004e-06, "loss": 0.0099, "step": 31403 }, { "epoch": 1.6742549448206003, "grad_norm": 1.6982910633087158, "learning_rate": 3.7194000000000003e-06, "loss": 0.0265, "step": 31404 }, { "epoch": 1.674308258250253, "grad_norm": 1.9264733791351318, "learning_rate": 3.7192e-06, "loss": 0.0141, "step": 31405 }, { "epoch": 1.6743615716799063, "grad_norm": 2.7913663387298584, "learning_rate": 3.7190000000000003e-06, "loss": 0.044, "step": 31406 }, { "epoch": 1.674414885109559, "grad_norm": 0.8308544754981995, "learning_rate": 3.7188000000000003e-06, "loss": 0.0207, "step": 31407 }, { "epoch": 1.674468198539212, "grad_norm": 1.1209322214126587, "learning_rate": 3.7186000000000007e-06, "loss": 0.0177, "step": 31408 }, { "epoch": 1.674521511968865, "grad_norm": 1.9250837564468384, "learning_rate": 3.7184000000000003e-06, "loss": 0.0195, "step": 31409 }, { "epoch": 1.6745748253985178, "grad_norm": 1.694865107536316, "learning_rate": 3.7182000000000002e-06, "loss": 0.0262, "step": 31410 }, { "epoch": 1.674628138828171, "grad_norm": 1.9508755207061768, "learning_rate": 3.7180000000000007e-06, "loss": 0.0315, "step": 31411 }, { "epoch": 1.6746814522578237, "grad_norm": 0.9456890821456909, "learning_rate": 3.7178000000000002e-06, "loss": 0.0114, "step": 31412 }, { "epoch": 1.6747347656874767, "grad_norm": 0.2857581377029419, "learning_rate": 3.7176e-06, "loss": 0.0028, "step": 31413 }, { "epoch": 1.6747880791171297, "grad_norm": 0.8656890392303467, "learning_rate": 3.7174000000000006e-06, "loss": 0.0045, "step": 31414 }, { "epoch": 1.6748413925467824, "grad_norm": 2.6269619464874268, "learning_rate": 3.7172e-06, "loss": 0.014, "step": 31415 }, { "epoch": 1.6748947059764354, "grad_norm": 1.5048513412475586, "learning_rate": 3.717e-06, "loss": 0.016, "step": 31416 }, { "epoch": 1.6749480194060884, "grad_norm": 1.1105458736419678, "learning_rate": 3.7168000000000006e-06, "loss": 0.0091, "step": 31417 }, { "epoch": 1.6750013328357414, "grad_norm": 0.530462920665741, "learning_rate": 3.7166e-06, "loss": 0.0042, "step": 31418 }, { "epoch": 1.6750546462653944, "grad_norm": 3.9659550189971924, "learning_rate": 3.7164000000000005e-06, "loss": 0.0323, "step": 31419 }, { "epoch": 1.675107959695047, "grad_norm": 1.4065592288970947, "learning_rate": 3.7162000000000005e-06, "loss": 0.0103, "step": 31420 }, { "epoch": 1.6751612731247, "grad_norm": 2.11725115776062, "learning_rate": 3.716e-06, "loss": 0.0305, "step": 31421 }, { "epoch": 1.675214586554353, "grad_norm": 1.4355705976486206, "learning_rate": 3.7158000000000005e-06, "loss": 0.0318, "step": 31422 }, { "epoch": 1.6752678999840058, "grad_norm": 0.717988133430481, "learning_rate": 3.7156000000000005e-06, "loss": 0.0081, "step": 31423 }, { "epoch": 1.675321213413659, "grad_norm": 1.2931897640228271, "learning_rate": 3.7154e-06, "loss": 0.0186, "step": 31424 }, { "epoch": 1.6753745268433118, "grad_norm": 1.9375561475753784, "learning_rate": 3.7152000000000004e-06, "loss": 0.0238, "step": 31425 }, { "epoch": 1.6754278402729648, "grad_norm": 0.7874304056167603, "learning_rate": 3.7150000000000004e-06, "loss": 0.013, "step": 31426 }, { "epoch": 1.6754811537026177, "grad_norm": 2.9111578464508057, "learning_rate": 3.7148e-06, "loss": 0.0537, "step": 31427 }, { "epoch": 1.6755344671322705, "grad_norm": 1.0895016193389893, "learning_rate": 3.7146000000000004e-06, "loss": 0.0136, "step": 31428 }, { "epoch": 1.6755877805619237, "grad_norm": 0.6183072328567505, "learning_rate": 3.7144000000000004e-06, "loss": 0.0071, "step": 31429 }, { "epoch": 1.6756410939915765, "grad_norm": 1.1284654140472412, "learning_rate": 3.7142e-06, "loss": 0.0076, "step": 31430 }, { "epoch": 1.6756944074212294, "grad_norm": 2.5955986976623535, "learning_rate": 3.7140000000000004e-06, "loss": 0.0116, "step": 31431 }, { "epoch": 1.6757477208508824, "grad_norm": 0.9731549024581909, "learning_rate": 3.7138000000000003e-06, "loss": 0.0084, "step": 31432 }, { "epoch": 1.6758010342805352, "grad_norm": 1.7640928030014038, "learning_rate": 3.7136000000000007e-06, "loss": 0.0347, "step": 31433 }, { "epoch": 1.6758543477101882, "grad_norm": 3.647578239440918, "learning_rate": 3.7134000000000003e-06, "loss": 0.0477, "step": 31434 }, { "epoch": 1.6759076611398411, "grad_norm": 1.2393553256988525, "learning_rate": 3.7132000000000003e-06, "loss": 0.0099, "step": 31435 }, { "epoch": 1.675960974569494, "grad_norm": 0.6867849230766296, "learning_rate": 3.7130000000000007e-06, "loss": 0.0068, "step": 31436 }, { "epoch": 1.676014287999147, "grad_norm": 1.0886608362197876, "learning_rate": 3.7128000000000003e-06, "loss": 0.0083, "step": 31437 }, { "epoch": 1.6760676014287998, "grad_norm": 1.2962697744369507, "learning_rate": 3.7126000000000002e-06, "loss": 0.0115, "step": 31438 }, { "epoch": 1.6761209148584528, "grad_norm": 2.198530673980713, "learning_rate": 3.7124000000000007e-06, "loss": 0.0388, "step": 31439 }, { "epoch": 1.6761742282881058, "grad_norm": 0.7659013271331787, "learning_rate": 3.7122000000000002e-06, "loss": 0.0077, "step": 31440 }, { "epoch": 1.6762275417177586, "grad_norm": 2.426661729812622, "learning_rate": 3.712e-06, "loss": 0.0288, "step": 31441 }, { "epoch": 1.6762808551474118, "grad_norm": 0.7266098856925964, "learning_rate": 3.7118000000000006e-06, "loss": 0.014, "step": 31442 }, { "epoch": 1.6763341685770645, "grad_norm": 2.103858232498169, "learning_rate": 3.7116e-06, "loss": 0.0456, "step": 31443 }, { "epoch": 1.6763874820067175, "grad_norm": 0.6717432737350464, "learning_rate": 3.7114000000000006e-06, "loss": 0.0061, "step": 31444 }, { "epoch": 1.6764407954363705, "grad_norm": 0.932441234588623, "learning_rate": 3.7112000000000006e-06, "loss": 0.0157, "step": 31445 }, { "epoch": 1.6764941088660232, "grad_norm": 1.228561520576477, "learning_rate": 3.711e-06, "loss": 0.0115, "step": 31446 }, { "epoch": 1.6765474222956764, "grad_norm": 3.498436212539673, "learning_rate": 3.7108000000000005e-06, "loss": 0.0336, "step": 31447 }, { "epoch": 1.6766007357253292, "grad_norm": 1.6824052333831787, "learning_rate": 3.7106000000000005e-06, "loss": 0.0117, "step": 31448 }, { "epoch": 1.6766540491549822, "grad_norm": 2.534989595413208, "learning_rate": 3.7104e-06, "loss": 0.0181, "step": 31449 }, { "epoch": 1.6767073625846352, "grad_norm": 3.382389783859253, "learning_rate": 3.7102000000000005e-06, "loss": 0.0201, "step": 31450 }, { "epoch": 1.676760676014288, "grad_norm": 3.0644497871398926, "learning_rate": 3.7100000000000005e-06, "loss": 0.0226, "step": 31451 }, { "epoch": 1.6768139894439409, "grad_norm": 3.144655466079712, "learning_rate": 3.7098e-06, "loss": 0.0171, "step": 31452 }, { "epoch": 1.6768673028735939, "grad_norm": 5.7886762619018555, "learning_rate": 3.7096000000000004e-06, "loss": 0.0326, "step": 31453 }, { "epoch": 1.6769206163032466, "grad_norm": 1.8326252698898315, "learning_rate": 3.7094000000000004e-06, "loss": 0.0125, "step": 31454 }, { "epoch": 1.6769739297328998, "grad_norm": 3.4639670848846436, "learning_rate": 3.7092000000000004e-06, "loss": 0.0275, "step": 31455 }, { "epoch": 1.6770272431625526, "grad_norm": 0.9506192207336426, "learning_rate": 3.7090000000000004e-06, "loss": 0.0078, "step": 31456 }, { "epoch": 1.6770805565922056, "grad_norm": 0.7320274710655212, "learning_rate": 3.7088000000000004e-06, "loss": 0.017, "step": 31457 }, { "epoch": 1.6771338700218585, "grad_norm": 0.8592442870140076, "learning_rate": 3.7086000000000004e-06, "loss": 0.0172, "step": 31458 }, { "epoch": 1.6771871834515113, "grad_norm": 0.6434141397476196, "learning_rate": 3.7084000000000004e-06, "loss": 0.0154, "step": 31459 }, { "epoch": 1.6772404968811645, "grad_norm": 3.137357234954834, "learning_rate": 3.7082000000000003e-06, "loss": 0.0138, "step": 31460 }, { "epoch": 1.6772938103108173, "grad_norm": 2.1485447883605957, "learning_rate": 3.7080000000000003e-06, "loss": 0.0291, "step": 31461 }, { "epoch": 1.6773471237404702, "grad_norm": 2.7486255168914795, "learning_rate": 3.7078000000000003e-06, "loss": 0.0462, "step": 31462 }, { "epoch": 1.6774004371701232, "grad_norm": 2.5315208435058594, "learning_rate": 3.7076000000000003e-06, "loss": 0.0527, "step": 31463 }, { "epoch": 1.677453750599776, "grad_norm": 2.022679567337036, "learning_rate": 3.7074000000000003e-06, "loss": 0.0125, "step": 31464 }, { "epoch": 1.6775070640294292, "grad_norm": 1.8080354928970337, "learning_rate": 3.7072000000000003e-06, "loss": 0.0133, "step": 31465 }, { "epoch": 1.677560377459082, "grad_norm": 3.398510694503784, "learning_rate": 3.7070000000000003e-06, "loss": 0.0255, "step": 31466 }, { "epoch": 1.677613690888735, "grad_norm": 0.4214892089366913, "learning_rate": 3.7068000000000002e-06, "loss": 0.0034, "step": 31467 }, { "epoch": 1.6776670043183879, "grad_norm": 1.4908729791641235, "learning_rate": 3.7066000000000002e-06, "loss": 0.0127, "step": 31468 }, { "epoch": 1.6777203177480406, "grad_norm": 1.093953013420105, "learning_rate": 3.7064000000000006e-06, "loss": 0.0092, "step": 31469 }, { "epoch": 1.6777736311776936, "grad_norm": 2.0991404056549072, "learning_rate": 3.7062e-06, "loss": 0.0322, "step": 31470 }, { "epoch": 1.6778269446073466, "grad_norm": 1.3591845035552979, "learning_rate": 3.706e-06, "loss": 0.007, "step": 31471 }, { "epoch": 1.6778802580369994, "grad_norm": 1.7107410430908203, "learning_rate": 3.7058000000000006e-06, "loss": 0.0322, "step": 31472 }, { "epoch": 1.6779335714666526, "grad_norm": 1.803299069404602, "learning_rate": 3.7056e-06, "loss": 0.0425, "step": 31473 }, { "epoch": 1.6779868848963053, "grad_norm": 1.9999644756317139, "learning_rate": 3.7054e-06, "loss": 0.0298, "step": 31474 }, { "epoch": 1.6780401983259583, "grad_norm": 1.6499701738357544, "learning_rate": 3.7052000000000005e-06, "loss": 0.018, "step": 31475 }, { "epoch": 1.6780935117556113, "grad_norm": 2.4692418575286865, "learning_rate": 3.705e-06, "loss": 0.0126, "step": 31476 }, { "epoch": 1.678146825185264, "grad_norm": 4.572486877441406, "learning_rate": 3.7048e-06, "loss": 0.06, "step": 31477 }, { "epoch": 1.6782001386149172, "grad_norm": 2.862093448638916, "learning_rate": 3.7046000000000005e-06, "loss": 0.0566, "step": 31478 }, { "epoch": 1.67825345204457, "grad_norm": 0.9509868025779724, "learning_rate": 3.7044e-06, "loss": 0.007, "step": 31479 }, { "epoch": 1.678306765474223, "grad_norm": 2.1228179931640625, "learning_rate": 3.7042000000000005e-06, "loss": 0.0176, "step": 31480 }, { "epoch": 1.678360078903876, "grad_norm": 1.5336765050888062, "learning_rate": 3.7040000000000005e-06, "loss": 0.0185, "step": 31481 }, { "epoch": 1.6784133923335287, "grad_norm": 1.3831068277359009, "learning_rate": 3.7038e-06, "loss": 0.0197, "step": 31482 }, { "epoch": 1.678466705763182, "grad_norm": 1.929319977760315, "learning_rate": 3.7036000000000004e-06, "loss": 0.0217, "step": 31483 }, { "epoch": 1.6785200191928347, "grad_norm": 1.1245126724243164, "learning_rate": 3.7034000000000004e-06, "loss": 0.0126, "step": 31484 }, { "epoch": 1.6785733326224876, "grad_norm": 2.360473394393921, "learning_rate": 3.7032e-06, "loss": 0.0594, "step": 31485 }, { "epoch": 1.6786266460521406, "grad_norm": 1.4565483331680298, "learning_rate": 3.7030000000000004e-06, "loss": 0.0111, "step": 31486 }, { "epoch": 1.6786799594817934, "grad_norm": 1.5736911296844482, "learning_rate": 3.7028000000000004e-06, "loss": 0.0106, "step": 31487 }, { "epoch": 1.6787332729114464, "grad_norm": 0.9770815372467041, "learning_rate": 3.7026e-06, "loss": 0.0099, "step": 31488 }, { "epoch": 1.6787865863410993, "grad_norm": 3.6720921993255615, "learning_rate": 3.7024000000000003e-06, "loss": 0.0167, "step": 31489 }, { "epoch": 1.678839899770752, "grad_norm": 2.229551076889038, "learning_rate": 3.7022000000000003e-06, "loss": 0.0155, "step": 31490 }, { "epoch": 1.6788932132004053, "grad_norm": 2.2220699787139893, "learning_rate": 3.702e-06, "loss": 0.0419, "step": 31491 }, { "epoch": 1.678946526630058, "grad_norm": 2.550917863845825, "learning_rate": 3.7018000000000003e-06, "loss": 0.0163, "step": 31492 }, { "epoch": 1.678999840059711, "grad_norm": 2.6200296878814697, "learning_rate": 3.7016000000000003e-06, "loss": 0.0139, "step": 31493 }, { "epoch": 1.679053153489364, "grad_norm": 2.361123561859131, "learning_rate": 3.7014000000000007e-06, "loss": 0.0167, "step": 31494 }, { "epoch": 1.6791064669190168, "grad_norm": 1.7210670709609985, "learning_rate": 3.7012000000000002e-06, "loss": 0.0263, "step": 31495 }, { "epoch": 1.67915978034867, "grad_norm": 4.3622026443481445, "learning_rate": 3.7010000000000002e-06, "loss": 0.0323, "step": 31496 }, { "epoch": 1.6792130937783227, "grad_norm": 1.6131552457809448, "learning_rate": 3.7008000000000006e-06, "loss": 0.0091, "step": 31497 }, { "epoch": 1.6792664072079757, "grad_norm": 0.5538471937179565, "learning_rate": 3.7006e-06, "loss": 0.0038, "step": 31498 }, { "epoch": 1.6793197206376287, "grad_norm": 1.37044358253479, "learning_rate": 3.7004e-06, "loss": 0.0173, "step": 31499 }, { "epoch": 1.6793730340672814, "grad_norm": 2.2239596843719482, "learning_rate": 3.7002000000000006e-06, "loss": 0.0211, "step": 31500 }, { "epoch": 1.6794263474969344, "grad_norm": 2.3621490001678467, "learning_rate": 3.7e-06, "loss": 0.0114, "step": 31501 }, { "epoch": 1.6794796609265874, "grad_norm": 0.6097853779792786, "learning_rate": 3.6998e-06, "loss": 0.0051, "step": 31502 }, { "epoch": 1.6795329743562404, "grad_norm": 2.2399137020111084, "learning_rate": 3.6996000000000005e-06, "loss": 0.0054, "step": 31503 }, { "epoch": 1.6795862877858934, "grad_norm": 0.990597128868103, "learning_rate": 3.6994e-06, "loss": 0.0182, "step": 31504 }, { "epoch": 1.6796396012155461, "grad_norm": 2.0376482009887695, "learning_rate": 3.6992000000000005e-06, "loss": 0.0447, "step": 31505 }, { "epoch": 1.679692914645199, "grad_norm": 0.662632942199707, "learning_rate": 3.6990000000000005e-06, "loss": 0.0056, "step": 31506 }, { "epoch": 1.679746228074852, "grad_norm": 1.0970617532730103, "learning_rate": 3.6988e-06, "loss": 0.0182, "step": 31507 }, { "epoch": 1.6797995415045048, "grad_norm": 2.957141637802124, "learning_rate": 3.6986000000000005e-06, "loss": 0.0357, "step": 31508 }, { "epoch": 1.679852854934158, "grad_norm": 1.972184658050537, "learning_rate": 3.6984000000000005e-06, "loss": 0.0314, "step": 31509 }, { "epoch": 1.6799061683638108, "grad_norm": 1.018847942352295, "learning_rate": 3.6982e-06, "loss": 0.0195, "step": 31510 }, { "epoch": 1.6799594817934638, "grad_norm": 1.2449413537979126, "learning_rate": 3.6980000000000004e-06, "loss": 0.0113, "step": 31511 }, { "epoch": 1.6800127952231168, "grad_norm": 0.8643954396247864, "learning_rate": 3.6978000000000004e-06, "loss": 0.0231, "step": 31512 }, { "epoch": 1.6800661086527695, "grad_norm": 1.6099905967712402, "learning_rate": 3.6976e-06, "loss": 0.0064, "step": 31513 }, { "epoch": 1.6801194220824227, "grad_norm": 0.7567967772483826, "learning_rate": 3.6974000000000004e-06, "loss": 0.0048, "step": 31514 }, { "epoch": 1.6801727355120755, "grad_norm": 2.9254841804504395, "learning_rate": 3.6972000000000004e-06, "loss": 0.0386, "step": 31515 }, { "epoch": 1.6802260489417284, "grad_norm": 1.70314621925354, "learning_rate": 3.697e-06, "loss": 0.0294, "step": 31516 }, { "epoch": 1.6802793623713814, "grad_norm": 2.7349464893341064, "learning_rate": 3.6968000000000003e-06, "loss": 0.0386, "step": 31517 }, { "epoch": 1.6803326758010342, "grad_norm": 2.799023389816284, "learning_rate": 3.6966000000000003e-06, "loss": 0.0414, "step": 31518 }, { "epoch": 1.6803859892306872, "grad_norm": 2.03303861618042, "learning_rate": 3.6964000000000007e-06, "loss": 0.0278, "step": 31519 }, { "epoch": 1.6804393026603401, "grad_norm": 0.6597929000854492, "learning_rate": 3.6962000000000003e-06, "loss": 0.0057, "step": 31520 }, { "epoch": 1.6804926160899931, "grad_norm": 0.9348212480545044, "learning_rate": 3.6960000000000003e-06, "loss": 0.0077, "step": 31521 }, { "epoch": 1.680545929519646, "grad_norm": 3.3774919509887695, "learning_rate": 3.6958000000000007e-06, "loss": 0.0481, "step": 31522 }, { "epoch": 1.6805992429492989, "grad_norm": 1.138100266456604, "learning_rate": 3.6956000000000002e-06, "loss": 0.0178, "step": 31523 }, { "epoch": 1.6806525563789518, "grad_norm": 1.4042094945907593, "learning_rate": 3.6954000000000002e-06, "loss": 0.0142, "step": 31524 }, { "epoch": 1.6807058698086048, "grad_norm": 3.2890841960906982, "learning_rate": 3.6952000000000006e-06, "loss": 0.0411, "step": 31525 }, { "epoch": 1.6807591832382576, "grad_norm": 3.151477336883545, "learning_rate": 3.695e-06, "loss": 0.0135, "step": 31526 }, { "epoch": 1.6808124966679108, "grad_norm": 3.4237053394317627, "learning_rate": 3.6948e-06, "loss": 0.0635, "step": 31527 }, { "epoch": 1.6808658100975635, "grad_norm": 0.8877303004264832, "learning_rate": 3.6946000000000006e-06, "loss": 0.0075, "step": 31528 }, { "epoch": 1.6809191235272165, "grad_norm": 1.2766327857971191, "learning_rate": 3.6944e-06, "loss": 0.0121, "step": 31529 }, { "epoch": 1.6809724369568695, "grad_norm": 2.1476309299468994, "learning_rate": 3.6942000000000006e-06, "loss": 0.0244, "step": 31530 }, { "epoch": 1.6810257503865222, "grad_norm": 1.0518895387649536, "learning_rate": 3.6940000000000005e-06, "loss": 0.0085, "step": 31531 }, { "epoch": 1.6810790638161754, "grad_norm": 1.2430845499038696, "learning_rate": 3.6938e-06, "loss": 0.0238, "step": 31532 }, { "epoch": 1.6811323772458282, "grad_norm": 2.5066137313842773, "learning_rate": 3.6936000000000005e-06, "loss": 0.0338, "step": 31533 }, { "epoch": 1.6811856906754812, "grad_norm": 0.7800951600074768, "learning_rate": 3.6934000000000005e-06, "loss": 0.0057, "step": 31534 }, { "epoch": 1.6812390041051342, "grad_norm": 0.3412298560142517, "learning_rate": 3.6932e-06, "loss": 0.0028, "step": 31535 }, { "epoch": 1.681292317534787, "grad_norm": 1.6143805980682373, "learning_rate": 3.6930000000000005e-06, "loss": 0.0164, "step": 31536 }, { "epoch": 1.68134563096444, "grad_norm": 1.2374106645584106, "learning_rate": 3.6928000000000005e-06, "loss": 0.0261, "step": 31537 }, { "epoch": 1.6813989443940929, "grad_norm": 1.799304723739624, "learning_rate": 3.6926e-06, "loss": 0.0199, "step": 31538 }, { "epoch": 1.6814522578237459, "grad_norm": 1.9891501665115356, "learning_rate": 3.6924000000000004e-06, "loss": 0.0262, "step": 31539 }, { "epoch": 1.6815055712533988, "grad_norm": 2.2492117881774902, "learning_rate": 3.6922000000000004e-06, "loss": 0.0333, "step": 31540 }, { "epoch": 1.6815588846830516, "grad_norm": 0.9623028635978699, "learning_rate": 3.692e-06, "loss": 0.0196, "step": 31541 }, { "epoch": 1.6816121981127046, "grad_norm": 1.7031261920928955, "learning_rate": 3.6918000000000004e-06, "loss": 0.0231, "step": 31542 }, { "epoch": 1.6816655115423575, "grad_norm": 2.2487730979919434, "learning_rate": 3.6916000000000004e-06, "loss": 0.0268, "step": 31543 }, { "epoch": 1.6817188249720103, "grad_norm": 0.8676590323448181, "learning_rate": 3.6914000000000004e-06, "loss": 0.014, "step": 31544 }, { "epoch": 1.6817721384016635, "grad_norm": 2.892965078353882, "learning_rate": 3.6912000000000003e-06, "loss": 0.0136, "step": 31545 }, { "epoch": 1.6818254518313163, "grad_norm": 1.98591148853302, "learning_rate": 3.6910000000000003e-06, "loss": 0.0376, "step": 31546 }, { "epoch": 1.6818787652609692, "grad_norm": 0.3019786775112152, "learning_rate": 3.6908000000000003e-06, "loss": 0.0033, "step": 31547 }, { "epoch": 1.6819320786906222, "grad_norm": 1.404992699623108, "learning_rate": 3.6906000000000003e-06, "loss": 0.0113, "step": 31548 }, { "epoch": 1.681985392120275, "grad_norm": 1.6191718578338623, "learning_rate": 3.6904000000000003e-06, "loss": 0.019, "step": 31549 }, { "epoch": 1.6820387055499282, "grad_norm": 1.462629795074463, "learning_rate": 3.6902000000000003e-06, "loss": 0.0239, "step": 31550 }, { "epoch": 1.682092018979581, "grad_norm": 1.7808581590652466, "learning_rate": 3.6900000000000002e-06, "loss": 0.02, "step": 31551 }, { "epoch": 1.682145332409234, "grad_norm": 0.6144549250602722, "learning_rate": 3.6898000000000002e-06, "loss": 0.0054, "step": 31552 }, { "epoch": 1.682198645838887, "grad_norm": 2.1713333129882812, "learning_rate": 3.6896000000000002e-06, "loss": 0.048, "step": 31553 }, { "epoch": 1.6822519592685397, "grad_norm": 1.5033804178237915, "learning_rate": 3.6894e-06, "loss": 0.0169, "step": 31554 }, { "epoch": 1.6823052726981926, "grad_norm": 2.5305161476135254, "learning_rate": 3.6892000000000006e-06, "loss": 0.0204, "step": 31555 }, { "epoch": 1.6823585861278456, "grad_norm": 1.3633601665496826, "learning_rate": 3.689e-06, "loss": 0.0164, "step": 31556 }, { "epoch": 1.6824118995574984, "grad_norm": 1.8768348693847656, "learning_rate": 3.6888e-06, "loss": 0.0171, "step": 31557 }, { "epoch": 1.6824652129871516, "grad_norm": 1.3351188898086548, "learning_rate": 3.6886000000000006e-06, "loss": 0.0159, "step": 31558 }, { "epoch": 1.6825185264168043, "grad_norm": 1.6805527210235596, "learning_rate": 3.6884e-06, "loss": 0.0203, "step": 31559 }, { "epoch": 1.6825718398464573, "grad_norm": 2.608823776245117, "learning_rate": 3.6882e-06, "loss": 0.0216, "step": 31560 }, { "epoch": 1.6826251532761103, "grad_norm": 2.189598321914673, "learning_rate": 3.6880000000000005e-06, "loss": 0.0134, "step": 31561 }, { "epoch": 1.682678466705763, "grad_norm": 2.5788493156433105, "learning_rate": 3.6878e-06, "loss": 0.0271, "step": 31562 }, { "epoch": 1.6827317801354162, "grad_norm": 1.6847286224365234, "learning_rate": 3.6876e-06, "loss": 0.0107, "step": 31563 }, { "epoch": 1.682785093565069, "grad_norm": 1.6534112691879272, "learning_rate": 3.6874000000000005e-06, "loss": 0.0173, "step": 31564 }, { "epoch": 1.682838406994722, "grad_norm": 0.8402137160301208, "learning_rate": 3.6872e-06, "loss": 0.0072, "step": 31565 }, { "epoch": 1.682891720424375, "grad_norm": 1.9261119365692139, "learning_rate": 3.6870000000000004e-06, "loss": 0.038, "step": 31566 }, { "epoch": 1.6829450338540277, "grad_norm": 1.5666056871414185, "learning_rate": 3.6868000000000004e-06, "loss": 0.0111, "step": 31567 }, { "epoch": 1.682998347283681, "grad_norm": 3.039767265319824, "learning_rate": 3.6866e-06, "loss": 0.0436, "step": 31568 }, { "epoch": 1.6830516607133337, "grad_norm": 3.029763698577881, "learning_rate": 3.6864000000000004e-06, "loss": 0.0432, "step": 31569 }, { "epoch": 1.6831049741429867, "grad_norm": 0.9231353998184204, "learning_rate": 3.6862000000000004e-06, "loss": 0.0105, "step": 31570 }, { "epoch": 1.6831582875726396, "grad_norm": 1.5827140808105469, "learning_rate": 3.686e-06, "loss": 0.0167, "step": 31571 }, { "epoch": 1.6832116010022924, "grad_norm": 2.82316255569458, "learning_rate": 3.6858000000000004e-06, "loss": 0.0267, "step": 31572 }, { "epoch": 1.6832649144319454, "grad_norm": 3.8887181282043457, "learning_rate": 3.6856000000000003e-06, "loss": 0.0267, "step": 31573 }, { "epoch": 1.6833182278615983, "grad_norm": 0.7696664929389954, "learning_rate": 3.6854e-06, "loss": 0.0075, "step": 31574 }, { "epoch": 1.683371541291251, "grad_norm": 2.0526208877563477, "learning_rate": 3.6852000000000003e-06, "loss": 0.0276, "step": 31575 }, { "epoch": 1.6834248547209043, "grad_norm": 0.8098406791687012, "learning_rate": 3.6850000000000003e-06, "loss": 0.0052, "step": 31576 }, { "epoch": 1.683478168150557, "grad_norm": 4.325530529022217, "learning_rate": 3.6848e-06, "loss": 0.0199, "step": 31577 }, { "epoch": 1.68353148158021, "grad_norm": 1.0385394096374512, "learning_rate": 3.6846000000000003e-06, "loss": 0.0205, "step": 31578 }, { "epoch": 1.683584795009863, "grad_norm": 2.6611015796661377, "learning_rate": 3.6844000000000002e-06, "loss": 0.0518, "step": 31579 }, { "epoch": 1.6836381084395158, "grad_norm": 2.193699598312378, "learning_rate": 3.6842000000000007e-06, "loss": 0.0205, "step": 31580 }, { "epoch": 1.683691421869169, "grad_norm": 1.3034392595291138, "learning_rate": 3.6840000000000002e-06, "loss": 0.0197, "step": 31581 }, { "epoch": 1.6837447352988217, "grad_norm": 1.989705204963684, "learning_rate": 3.6838e-06, "loss": 0.0283, "step": 31582 }, { "epoch": 1.6837980487284747, "grad_norm": 3.3835175037384033, "learning_rate": 3.6836000000000006e-06, "loss": 0.0294, "step": 31583 }, { "epoch": 1.6838513621581277, "grad_norm": 1.8554295301437378, "learning_rate": 3.6834e-06, "loss": 0.0199, "step": 31584 }, { "epoch": 1.6839046755877805, "grad_norm": 2.3971610069274902, "learning_rate": 3.6832e-06, "loss": 0.0259, "step": 31585 }, { "epoch": 1.6839579890174337, "grad_norm": 1.9033257961273193, "learning_rate": 3.6830000000000006e-06, "loss": 0.0267, "step": 31586 }, { "epoch": 1.6840113024470864, "grad_norm": 6.015224933624268, "learning_rate": 3.6828e-06, "loss": 0.0343, "step": 31587 }, { "epoch": 1.6840646158767394, "grad_norm": 2.237977981567383, "learning_rate": 3.6826e-06, "loss": 0.015, "step": 31588 }, { "epoch": 1.6841179293063924, "grad_norm": 2.4838671684265137, "learning_rate": 3.6824000000000005e-06, "loss": 0.019, "step": 31589 }, { "epoch": 1.6841712427360451, "grad_norm": 1.2035844326019287, "learning_rate": 3.6822e-06, "loss": 0.0111, "step": 31590 }, { "epoch": 1.684224556165698, "grad_norm": 3.8586513996124268, "learning_rate": 3.6820000000000005e-06, "loss": 0.0249, "step": 31591 }, { "epoch": 1.684277869595351, "grad_norm": 0.3982928693294525, "learning_rate": 3.6818000000000005e-06, "loss": 0.0042, "step": 31592 }, { "epoch": 1.6843311830250038, "grad_norm": 1.3349560499191284, "learning_rate": 3.6816e-06, "loss": 0.01, "step": 31593 }, { "epoch": 1.684384496454657, "grad_norm": 1.5986013412475586, "learning_rate": 3.6814000000000004e-06, "loss": 0.0213, "step": 31594 }, { "epoch": 1.6844378098843098, "grad_norm": 1.789198637008667, "learning_rate": 3.6812000000000004e-06, "loss": 0.0165, "step": 31595 }, { "epoch": 1.6844911233139628, "grad_norm": 0.8474578261375427, "learning_rate": 3.681e-06, "loss": 0.0073, "step": 31596 }, { "epoch": 1.6845444367436158, "grad_norm": 1.3054065704345703, "learning_rate": 3.6808000000000004e-06, "loss": 0.0051, "step": 31597 }, { "epoch": 1.6845977501732685, "grad_norm": 1.8236806392669678, "learning_rate": 3.6806000000000004e-06, "loss": 0.0184, "step": 31598 }, { "epoch": 1.6846510636029217, "grad_norm": 1.218283772468567, "learning_rate": 3.6804e-06, "loss": 0.007, "step": 31599 }, { "epoch": 1.6847043770325745, "grad_norm": 1.0309988260269165, "learning_rate": 3.6802000000000004e-06, "loss": 0.0036, "step": 31600 }, { "epoch": 1.6847576904622275, "grad_norm": 1.5840479135513306, "learning_rate": 3.6800000000000003e-06, "loss": 0.0227, "step": 31601 }, { "epoch": 1.6848110038918804, "grad_norm": 2.115354299545288, "learning_rate": 3.6798e-06, "loss": 0.0186, "step": 31602 }, { "epoch": 1.6848643173215332, "grad_norm": 1.4280847311019897, "learning_rate": 3.6796000000000003e-06, "loss": 0.0067, "step": 31603 }, { "epoch": 1.6849176307511862, "grad_norm": 1.2063748836517334, "learning_rate": 3.6794000000000003e-06, "loss": 0.015, "step": 31604 }, { "epoch": 1.6849709441808391, "grad_norm": 3.151094913482666, "learning_rate": 3.6792000000000007e-06, "loss": 0.0123, "step": 31605 }, { "epoch": 1.6850242576104921, "grad_norm": 3.6812987327575684, "learning_rate": 3.6790000000000003e-06, "loss": 0.0451, "step": 31606 }, { "epoch": 1.685077571040145, "grad_norm": 2.6385140419006348, "learning_rate": 3.6788000000000003e-06, "loss": 0.0278, "step": 31607 }, { "epoch": 1.6851308844697979, "grad_norm": 1.2847923040390015, "learning_rate": 3.6786000000000007e-06, "loss": 0.0132, "step": 31608 }, { "epoch": 1.6851841978994508, "grad_norm": 2.1792078018188477, "learning_rate": 3.6784000000000002e-06, "loss": 0.0239, "step": 31609 }, { "epoch": 1.6852375113291038, "grad_norm": 1.9597454071044922, "learning_rate": 3.6782e-06, "loss": 0.0245, "step": 31610 }, { "epoch": 1.6852908247587566, "grad_norm": 2.318728446960449, "learning_rate": 3.6780000000000006e-06, "loss": 0.0157, "step": 31611 }, { "epoch": 1.6853441381884098, "grad_norm": 1.3875608444213867, "learning_rate": 3.6778e-06, "loss": 0.0127, "step": 31612 }, { "epoch": 1.6853974516180625, "grad_norm": 1.8635573387145996, "learning_rate": 3.6776e-06, "loss": 0.0213, "step": 31613 }, { "epoch": 1.6854507650477155, "grad_norm": 2.254678249359131, "learning_rate": 3.6774000000000006e-06, "loss": 0.0374, "step": 31614 }, { "epoch": 1.6855040784773685, "grad_norm": 1.5045084953308105, "learning_rate": 3.6772e-06, "loss": 0.0264, "step": 31615 }, { "epoch": 1.6855573919070213, "grad_norm": 1.108317494392395, "learning_rate": 3.6770000000000005e-06, "loss": 0.0202, "step": 31616 }, { "epoch": 1.6856107053366745, "grad_norm": 1.1505738496780396, "learning_rate": 3.6768000000000005e-06, "loss": 0.0065, "step": 31617 }, { "epoch": 1.6856640187663272, "grad_norm": 0.4859435260295868, "learning_rate": 3.6766e-06, "loss": 0.0034, "step": 31618 }, { "epoch": 1.6857173321959802, "grad_norm": 1.585147738456726, "learning_rate": 3.6764000000000005e-06, "loss": 0.0276, "step": 31619 }, { "epoch": 1.6857706456256332, "grad_norm": 1.6730417013168335, "learning_rate": 3.6762000000000005e-06, "loss": 0.0411, "step": 31620 }, { "epoch": 1.685823959055286, "grad_norm": 4.9491658210754395, "learning_rate": 3.676e-06, "loss": 0.0223, "step": 31621 }, { "epoch": 1.685877272484939, "grad_norm": 1.1255967617034912, "learning_rate": 3.6758000000000004e-06, "loss": 0.017, "step": 31622 }, { "epoch": 1.6859305859145919, "grad_norm": 2.354330062866211, "learning_rate": 3.6756000000000004e-06, "loss": 0.0411, "step": 31623 }, { "epoch": 1.6859838993442449, "grad_norm": 0.9628570675849915, "learning_rate": 3.6754e-06, "loss": 0.0154, "step": 31624 }, { "epoch": 1.6860372127738978, "grad_norm": 0.7894566655158997, "learning_rate": 3.6752000000000004e-06, "loss": 0.0035, "step": 31625 }, { "epoch": 1.6860905262035506, "grad_norm": 1.313318133354187, "learning_rate": 3.6750000000000004e-06, "loss": 0.0045, "step": 31626 }, { "epoch": 1.6861438396332036, "grad_norm": 0.9494872689247131, "learning_rate": 3.6748e-06, "loss": 0.0076, "step": 31627 }, { "epoch": 1.6861971530628566, "grad_norm": 1.3368136882781982, "learning_rate": 3.6746000000000004e-06, "loss": 0.0163, "step": 31628 }, { "epoch": 1.6862504664925093, "grad_norm": 2.5541298389434814, "learning_rate": 3.6744000000000003e-06, "loss": 0.0387, "step": 31629 }, { "epoch": 1.6863037799221625, "grad_norm": 2.3196470737457275, "learning_rate": 3.6742000000000008e-06, "loss": 0.0418, "step": 31630 }, { "epoch": 1.6863570933518153, "grad_norm": 1.4556725025177002, "learning_rate": 3.6740000000000003e-06, "loss": 0.0092, "step": 31631 }, { "epoch": 1.6864104067814683, "grad_norm": 1.2795778512954712, "learning_rate": 3.6738000000000003e-06, "loss": 0.0365, "step": 31632 }, { "epoch": 1.6864637202111212, "grad_norm": 0.6971402168273926, "learning_rate": 3.6736000000000007e-06, "loss": 0.005, "step": 31633 }, { "epoch": 1.686517033640774, "grad_norm": 2.3284473419189453, "learning_rate": 3.6734000000000003e-06, "loss": 0.0166, "step": 31634 }, { "epoch": 1.6865703470704272, "grad_norm": 1.3236191272735596, "learning_rate": 3.6732000000000003e-06, "loss": 0.0114, "step": 31635 }, { "epoch": 1.68662366050008, "grad_norm": 2.9290640354156494, "learning_rate": 3.6730000000000007e-06, "loss": 0.0402, "step": 31636 }, { "epoch": 1.686676973929733, "grad_norm": 1.636587381362915, "learning_rate": 3.6728000000000002e-06, "loss": 0.017, "step": 31637 }, { "epoch": 1.686730287359386, "grad_norm": 1.276282787322998, "learning_rate": 3.6726e-06, "loss": 0.0114, "step": 31638 }, { "epoch": 1.6867836007890387, "grad_norm": 1.168354868888855, "learning_rate": 3.6724000000000006e-06, "loss": 0.0071, "step": 31639 }, { "epoch": 1.6868369142186916, "grad_norm": 0.26585444808006287, "learning_rate": 3.6722e-06, "loss": 0.0036, "step": 31640 }, { "epoch": 1.6868902276483446, "grad_norm": 3.142622709274292, "learning_rate": 3.6720000000000006e-06, "loss": 0.0304, "step": 31641 }, { "epoch": 1.6869435410779976, "grad_norm": 2.990483522415161, "learning_rate": 3.6718000000000006e-06, "loss": 0.014, "step": 31642 }, { "epoch": 1.6869968545076506, "grad_norm": 1.107749104499817, "learning_rate": 3.6716e-06, "loss": 0.015, "step": 31643 }, { "epoch": 1.6870501679373033, "grad_norm": 1.088463306427002, "learning_rate": 3.6714000000000005e-06, "loss": 0.007, "step": 31644 }, { "epoch": 1.6871034813669563, "grad_norm": 1.2497141361236572, "learning_rate": 3.6712000000000005e-06, "loss": 0.0089, "step": 31645 }, { "epoch": 1.6871567947966093, "grad_norm": 1.455244541168213, "learning_rate": 3.671e-06, "loss": 0.0235, "step": 31646 }, { "epoch": 1.687210108226262, "grad_norm": 0.8488483428955078, "learning_rate": 3.6708000000000005e-06, "loss": 0.0049, "step": 31647 }, { "epoch": 1.6872634216559153, "grad_norm": 1.697247862815857, "learning_rate": 3.6706000000000005e-06, "loss": 0.016, "step": 31648 }, { "epoch": 1.687316735085568, "grad_norm": 1.3056459426879883, "learning_rate": 3.6704e-06, "loss": 0.0175, "step": 31649 }, { "epoch": 1.687370048515221, "grad_norm": 1.0334157943725586, "learning_rate": 3.6702000000000005e-06, "loss": 0.0153, "step": 31650 }, { "epoch": 1.687423361944874, "grad_norm": 1.2180500030517578, "learning_rate": 3.6700000000000004e-06, "loss": 0.0255, "step": 31651 }, { "epoch": 1.6874766753745267, "grad_norm": 1.7814453840255737, "learning_rate": 3.6698e-06, "loss": 0.0257, "step": 31652 }, { "epoch": 1.68752998880418, "grad_norm": 3.5966148376464844, "learning_rate": 3.6696000000000004e-06, "loss": 0.0432, "step": 31653 }, { "epoch": 1.6875833022338327, "grad_norm": 3.251127004623413, "learning_rate": 3.6694000000000004e-06, "loss": 0.0517, "step": 31654 }, { "epoch": 1.6876366156634857, "grad_norm": 2.2250635623931885, "learning_rate": 3.6692000000000004e-06, "loss": 0.018, "step": 31655 }, { "epoch": 1.6876899290931386, "grad_norm": 0.7684573531150818, "learning_rate": 3.6690000000000004e-06, "loss": 0.01, "step": 31656 }, { "epoch": 1.6877432425227914, "grad_norm": 4.198429584503174, "learning_rate": 3.6688000000000003e-06, "loss": 0.0391, "step": 31657 }, { "epoch": 1.6877965559524444, "grad_norm": 1.2283865213394165, "learning_rate": 3.6686000000000003e-06, "loss": 0.0112, "step": 31658 }, { "epoch": 1.6878498693820974, "grad_norm": 2.4903879165649414, "learning_rate": 3.6684000000000003e-06, "loss": 0.026, "step": 31659 }, { "epoch": 1.6879031828117501, "grad_norm": 2.274183750152588, "learning_rate": 3.6682000000000003e-06, "loss": 0.015, "step": 31660 }, { "epoch": 1.6879564962414033, "grad_norm": 1.4225683212280273, "learning_rate": 3.6680000000000003e-06, "loss": 0.0235, "step": 31661 }, { "epoch": 1.688009809671056, "grad_norm": 1.007631540298462, "learning_rate": 3.6678000000000003e-06, "loss": 0.0081, "step": 31662 }, { "epoch": 1.688063123100709, "grad_norm": 2.0097784996032715, "learning_rate": 3.6676000000000003e-06, "loss": 0.0123, "step": 31663 }, { "epoch": 1.688116436530362, "grad_norm": 0.7010523676872253, "learning_rate": 3.6674000000000002e-06, "loss": 0.0063, "step": 31664 }, { "epoch": 1.6881697499600148, "grad_norm": 1.8090975284576416, "learning_rate": 3.6672000000000002e-06, "loss": 0.0074, "step": 31665 }, { "epoch": 1.688223063389668, "grad_norm": 3.332085609436035, "learning_rate": 3.6670000000000006e-06, "loss": 0.0364, "step": 31666 }, { "epoch": 1.6882763768193207, "grad_norm": 3.2959234714508057, "learning_rate": 3.6668e-06, "loss": 0.0418, "step": 31667 }, { "epoch": 1.6883296902489737, "grad_norm": 1.202651858329773, "learning_rate": 3.6666e-06, "loss": 0.0177, "step": 31668 }, { "epoch": 1.6883830036786267, "grad_norm": 1.2013260126113892, "learning_rate": 3.6664000000000006e-06, "loss": 0.0112, "step": 31669 }, { "epoch": 1.6884363171082795, "grad_norm": 2.759737968444824, "learning_rate": 3.6662e-06, "loss": 0.0282, "step": 31670 }, { "epoch": 1.6884896305379327, "grad_norm": 1.854541540145874, "learning_rate": 3.666e-06, "loss": 0.0292, "step": 31671 }, { "epoch": 1.6885429439675854, "grad_norm": 0.6495073437690735, "learning_rate": 3.6658000000000005e-06, "loss": 0.0047, "step": 31672 }, { "epoch": 1.6885962573972384, "grad_norm": 1.6394894123077393, "learning_rate": 3.6656e-06, "loss": 0.0356, "step": 31673 }, { "epoch": 1.6886495708268914, "grad_norm": 1.1121469736099243, "learning_rate": 3.6654e-06, "loss": 0.0057, "step": 31674 }, { "epoch": 1.6887028842565441, "grad_norm": 1.3016728162765503, "learning_rate": 3.6652000000000005e-06, "loss": 0.0054, "step": 31675 }, { "epoch": 1.6887561976861971, "grad_norm": 2.184370756149292, "learning_rate": 3.665e-06, "loss": 0.0199, "step": 31676 }, { "epoch": 1.68880951111585, "grad_norm": 2.537973642349243, "learning_rate": 3.6648e-06, "loss": 0.055, "step": 31677 }, { "epoch": 1.6888628245455028, "grad_norm": 3.935795545578003, "learning_rate": 3.6646000000000005e-06, "loss": 0.0127, "step": 31678 }, { "epoch": 1.688916137975156, "grad_norm": 1.324910283088684, "learning_rate": 3.6644e-06, "loss": 0.0193, "step": 31679 }, { "epoch": 1.6889694514048088, "grad_norm": 2.38438081741333, "learning_rate": 3.6642000000000004e-06, "loss": 0.0114, "step": 31680 }, { "epoch": 1.6890227648344618, "grad_norm": 2.989288806915283, "learning_rate": 3.6640000000000004e-06, "loss": 0.0132, "step": 31681 }, { "epoch": 1.6890760782641148, "grad_norm": 1.4760583639144897, "learning_rate": 3.6638e-06, "loss": 0.0063, "step": 31682 }, { "epoch": 1.6891293916937675, "grad_norm": 0.79505854845047, "learning_rate": 3.6636000000000004e-06, "loss": 0.0042, "step": 31683 }, { "epoch": 1.6891827051234207, "grad_norm": 2.0590178966522217, "learning_rate": 3.6634000000000004e-06, "loss": 0.0168, "step": 31684 }, { "epoch": 1.6892360185530735, "grad_norm": 2.0731794834136963, "learning_rate": 3.6632e-06, "loss": 0.0487, "step": 31685 }, { "epoch": 1.6892893319827265, "grad_norm": 2.7027015686035156, "learning_rate": 3.6630000000000003e-06, "loss": 0.0358, "step": 31686 }, { "epoch": 1.6893426454123794, "grad_norm": 2.7778518199920654, "learning_rate": 3.6628000000000003e-06, "loss": 0.0262, "step": 31687 }, { "epoch": 1.6893959588420322, "grad_norm": 2.306943893432617, "learning_rate": 3.6626e-06, "loss": 0.0374, "step": 31688 }, { "epoch": 1.6894492722716854, "grad_norm": 1.5923752784729004, "learning_rate": 3.6624000000000003e-06, "loss": 0.0272, "step": 31689 }, { "epoch": 1.6895025857013382, "grad_norm": 1.7534046173095703, "learning_rate": 3.6622000000000003e-06, "loss": 0.0515, "step": 31690 }, { "epoch": 1.6895558991309911, "grad_norm": 1.1797778606414795, "learning_rate": 3.6620000000000007e-06, "loss": 0.0077, "step": 31691 }, { "epoch": 1.6896092125606441, "grad_norm": 0.5142296552658081, "learning_rate": 3.6618000000000002e-06, "loss": 0.0051, "step": 31692 }, { "epoch": 1.6896625259902969, "grad_norm": 1.5340697765350342, "learning_rate": 3.6616000000000002e-06, "loss": 0.0312, "step": 31693 }, { "epoch": 1.6897158394199498, "grad_norm": 2.411414384841919, "learning_rate": 3.6614000000000006e-06, "loss": 0.0236, "step": 31694 }, { "epoch": 1.6897691528496028, "grad_norm": 2.5715365409851074, "learning_rate": 3.6612e-06, "loss": 0.0222, "step": 31695 }, { "epoch": 1.6898224662792556, "grad_norm": 2.5150437355041504, "learning_rate": 3.661e-06, "loss": 0.021, "step": 31696 }, { "epoch": 1.6898757797089088, "grad_norm": 2.165271043777466, "learning_rate": 3.6608000000000006e-06, "loss": 0.0154, "step": 31697 }, { "epoch": 1.6899290931385615, "grad_norm": 1.906384825706482, "learning_rate": 3.6606e-06, "loss": 0.0185, "step": 31698 }, { "epoch": 1.6899824065682145, "grad_norm": 1.4746818542480469, "learning_rate": 3.6604e-06, "loss": 0.0105, "step": 31699 }, { "epoch": 1.6900357199978675, "grad_norm": 1.4420710802078247, "learning_rate": 3.6602000000000005e-06, "loss": 0.0213, "step": 31700 }, { "epoch": 1.6900890334275203, "grad_norm": 1.0106830596923828, "learning_rate": 3.66e-06, "loss": 0.0128, "step": 31701 }, { "epoch": 1.6901423468571735, "grad_norm": 1.11806321144104, "learning_rate": 3.6598000000000005e-06, "loss": 0.0294, "step": 31702 }, { "epoch": 1.6901956602868262, "grad_norm": 1.6760528087615967, "learning_rate": 3.6596000000000005e-06, "loss": 0.0165, "step": 31703 }, { "epoch": 1.6902489737164792, "grad_norm": 4.399754047393799, "learning_rate": 3.6594e-06, "loss": 0.0345, "step": 31704 }, { "epoch": 1.6903022871461322, "grad_norm": 1.5001373291015625, "learning_rate": 3.6592000000000005e-06, "loss": 0.0299, "step": 31705 }, { "epoch": 1.690355600575785, "grad_norm": 0.9983890056610107, "learning_rate": 3.6590000000000005e-06, "loss": 0.0102, "step": 31706 }, { "epoch": 1.6904089140054381, "grad_norm": 3.0504791736602783, "learning_rate": 3.6588e-06, "loss": 0.0204, "step": 31707 }, { "epoch": 1.690462227435091, "grad_norm": 1.0537610054016113, "learning_rate": 3.6586000000000004e-06, "loss": 0.0192, "step": 31708 }, { "epoch": 1.6905155408647439, "grad_norm": 0.5411566495895386, "learning_rate": 3.6584000000000004e-06, "loss": 0.0047, "step": 31709 }, { "epoch": 1.6905688542943968, "grad_norm": 1.255853295326233, "learning_rate": 3.6582e-06, "loss": 0.0155, "step": 31710 }, { "epoch": 1.6906221677240496, "grad_norm": 0.9371275901794434, "learning_rate": 3.6580000000000004e-06, "loss": 0.0065, "step": 31711 }, { "epoch": 1.6906754811537026, "grad_norm": 0.9523279070854187, "learning_rate": 3.6578000000000004e-06, "loss": 0.0122, "step": 31712 }, { "epoch": 1.6907287945833556, "grad_norm": 0.8835485577583313, "learning_rate": 3.6576e-06, "loss": 0.006, "step": 31713 }, { "epoch": 1.6907821080130083, "grad_norm": 0.636279821395874, "learning_rate": 3.6574000000000003e-06, "loss": 0.0033, "step": 31714 }, { "epoch": 1.6908354214426615, "grad_norm": 1.4814224243164062, "learning_rate": 3.6572000000000003e-06, "loss": 0.0051, "step": 31715 }, { "epoch": 1.6908887348723143, "grad_norm": 1.8008286952972412, "learning_rate": 3.6570000000000007e-06, "loss": 0.0073, "step": 31716 }, { "epoch": 1.6909420483019673, "grad_norm": 1.5913612842559814, "learning_rate": 3.6568000000000003e-06, "loss": 0.0303, "step": 31717 }, { "epoch": 1.6909953617316202, "grad_norm": 1.0245150327682495, "learning_rate": 3.6566000000000003e-06, "loss": 0.0212, "step": 31718 }, { "epoch": 1.691048675161273, "grad_norm": 2.764166831970215, "learning_rate": 3.6564000000000007e-06, "loss": 0.0292, "step": 31719 }, { "epoch": 1.6911019885909262, "grad_norm": 3.8982958793640137, "learning_rate": 3.6562000000000002e-06, "loss": 0.0706, "step": 31720 }, { "epoch": 1.691155302020579, "grad_norm": 2.152860403060913, "learning_rate": 3.6560000000000002e-06, "loss": 0.032, "step": 31721 }, { "epoch": 1.691208615450232, "grad_norm": 1.0041601657867432, "learning_rate": 3.6558000000000006e-06, "loss": 0.0104, "step": 31722 }, { "epoch": 1.691261928879885, "grad_norm": 1.940549612045288, "learning_rate": 3.6556e-06, "loss": 0.0088, "step": 31723 }, { "epoch": 1.6913152423095377, "grad_norm": 1.5154839754104614, "learning_rate": 3.6554e-06, "loss": 0.027, "step": 31724 }, { "epoch": 1.6913685557391906, "grad_norm": 1.6173317432403564, "learning_rate": 3.6552000000000006e-06, "loss": 0.0172, "step": 31725 }, { "epoch": 1.6914218691688436, "grad_norm": 0.8309431672096252, "learning_rate": 3.655e-06, "loss": 0.0081, "step": 31726 }, { "epoch": 1.6914751825984966, "grad_norm": 2.6409127712249756, "learning_rate": 3.6548000000000006e-06, "loss": 0.0248, "step": 31727 }, { "epoch": 1.6915284960281496, "grad_norm": 1.883902907371521, "learning_rate": 3.6546000000000006e-06, "loss": 0.0106, "step": 31728 }, { "epoch": 1.6915818094578023, "grad_norm": 2.8654658794403076, "learning_rate": 3.6544e-06, "loss": 0.0249, "step": 31729 }, { "epoch": 1.6916351228874553, "grad_norm": 2.8769166469573975, "learning_rate": 3.6542000000000005e-06, "loss": 0.0115, "step": 31730 }, { "epoch": 1.6916884363171083, "grad_norm": 1.6037840843200684, "learning_rate": 3.6540000000000005e-06, "loss": 0.0293, "step": 31731 }, { "epoch": 1.691741749746761, "grad_norm": 1.9228413105010986, "learning_rate": 3.6538e-06, "loss": 0.0183, "step": 31732 }, { "epoch": 1.6917950631764143, "grad_norm": 0.6699302196502686, "learning_rate": 3.6536000000000005e-06, "loss": 0.0059, "step": 31733 }, { "epoch": 1.691848376606067, "grad_norm": 3.111236572265625, "learning_rate": 3.6534000000000005e-06, "loss": 0.03, "step": 31734 }, { "epoch": 1.69190169003572, "grad_norm": 0.9622103571891785, "learning_rate": 3.6532e-06, "loss": 0.0069, "step": 31735 }, { "epoch": 1.691955003465373, "grad_norm": 2.428748369216919, "learning_rate": 3.6530000000000004e-06, "loss": 0.035, "step": 31736 }, { "epoch": 1.6920083168950257, "grad_norm": 1.0949662923812866, "learning_rate": 3.6528000000000004e-06, "loss": 0.0098, "step": 31737 }, { "epoch": 1.692061630324679, "grad_norm": 3.9559719562530518, "learning_rate": 3.6526e-06, "loss": 0.0473, "step": 31738 }, { "epoch": 1.6921149437543317, "grad_norm": 1.6911685466766357, "learning_rate": 3.6524000000000004e-06, "loss": 0.0249, "step": 31739 }, { "epoch": 1.6921682571839847, "grad_norm": 1.5525215864181519, "learning_rate": 3.6522000000000004e-06, "loss": 0.0175, "step": 31740 }, { "epoch": 1.6922215706136376, "grad_norm": 8.669851303100586, "learning_rate": 3.6520000000000004e-06, "loss": 0.0308, "step": 31741 }, { "epoch": 1.6922748840432904, "grad_norm": 2.236192226409912, "learning_rate": 3.6518000000000003e-06, "loss": 0.0432, "step": 31742 }, { "epoch": 1.6923281974729434, "grad_norm": 2.1134731769561768, "learning_rate": 3.6516000000000003e-06, "loss": 0.0147, "step": 31743 }, { "epoch": 1.6923815109025964, "grad_norm": 1.0314600467681885, "learning_rate": 3.6514000000000003e-06, "loss": 0.0158, "step": 31744 }, { "epoch": 1.6924348243322493, "grad_norm": 1.01271390914917, "learning_rate": 3.6512000000000003e-06, "loss": 0.0199, "step": 31745 }, { "epoch": 1.6924881377619023, "grad_norm": 2.249462127685547, "learning_rate": 3.6510000000000003e-06, "loss": 0.0443, "step": 31746 }, { "epoch": 1.692541451191555, "grad_norm": 2.073474407196045, "learning_rate": 3.6508000000000003e-06, "loss": 0.069, "step": 31747 }, { "epoch": 1.692594764621208, "grad_norm": 3.083153486251831, "learning_rate": 3.6506000000000003e-06, "loss": 0.0158, "step": 31748 }, { "epoch": 1.692648078050861, "grad_norm": 1.0702970027923584, "learning_rate": 3.6504000000000002e-06, "loss": 0.0182, "step": 31749 }, { "epoch": 1.6927013914805138, "grad_norm": 2.644425392150879, "learning_rate": 3.6502000000000002e-06, "loss": 0.0132, "step": 31750 }, { "epoch": 1.692754704910167, "grad_norm": 2.6662325859069824, "learning_rate": 3.65e-06, "loss": 0.0191, "step": 31751 }, { "epoch": 1.6928080183398198, "grad_norm": 1.7276756763458252, "learning_rate": 3.6498000000000006e-06, "loss": 0.0212, "step": 31752 }, { "epoch": 1.6928613317694727, "grad_norm": 0.4589056372642517, "learning_rate": 3.6496e-06, "loss": 0.0042, "step": 31753 }, { "epoch": 1.6929146451991257, "grad_norm": 3.101271867752075, "learning_rate": 3.6494e-06, "loss": 0.0559, "step": 31754 }, { "epoch": 1.6929679586287785, "grad_norm": 3.879211902618408, "learning_rate": 3.6492000000000006e-06, "loss": 0.0501, "step": 31755 }, { "epoch": 1.6930212720584317, "grad_norm": 0.5822928547859192, "learning_rate": 3.649e-06, "loss": 0.0041, "step": 31756 }, { "epoch": 1.6930745854880844, "grad_norm": 1.470857858657837, "learning_rate": 3.6488e-06, "loss": 0.0224, "step": 31757 }, { "epoch": 1.6931278989177374, "grad_norm": 1.9422117471694946, "learning_rate": 3.6486000000000005e-06, "loss": 0.0191, "step": 31758 }, { "epoch": 1.6931812123473904, "grad_norm": 3.0708811283111572, "learning_rate": 3.6484e-06, "loss": 0.0355, "step": 31759 }, { "epoch": 1.6932345257770431, "grad_norm": 1.0101643800735474, "learning_rate": 3.6482e-06, "loss": 0.0066, "step": 31760 }, { "epoch": 1.6932878392066961, "grad_norm": 2.152932643890381, "learning_rate": 3.6480000000000005e-06, "loss": 0.0219, "step": 31761 }, { "epoch": 1.693341152636349, "grad_norm": 1.2745920419692993, "learning_rate": 3.6478e-06, "loss": 0.0148, "step": 31762 }, { "epoch": 1.693394466066002, "grad_norm": 3.538356304168701, "learning_rate": 3.6476e-06, "loss": 0.0353, "step": 31763 }, { "epoch": 1.693447779495655, "grad_norm": 2.691720962524414, "learning_rate": 3.6474000000000004e-06, "loss": 0.0291, "step": 31764 }, { "epoch": 1.6935010929253078, "grad_norm": 0.8050362467765808, "learning_rate": 3.6472e-06, "loss": 0.0219, "step": 31765 }, { "epoch": 1.6935544063549608, "grad_norm": 1.456723928451538, "learning_rate": 3.6470000000000004e-06, "loss": 0.0199, "step": 31766 }, { "epoch": 1.6936077197846138, "grad_norm": 1.4805995225906372, "learning_rate": 3.6468000000000004e-06, "loss": 0.0133, "step": 31767 }, { "epoch": 1.6936610332142665, "grad_norm": 1.9681732654571533, "learning_rate": 3.6466e-06, "loss": 0.0149, "step": 31768 }, { "epoch": 1.6937143466439197, "grad_norm": 1.8006293773651123, "learning_rate": 3.6464000000000004e-06, "loss": 0.025, "step": 31769 }, { "epoch": 1.6937676600735725, "grad_norm": 3.4141714572906494, "learning_rate": 3.6462000000000003e-06, "loss": 0.0111, "step": 31770 }, { "epoch": 1.6938209735032255, "grad_norm": 1.6898831129074097, "learning_rate": 3.646e-06, "loss": 0.0395, "step": 31771 }, { "epoch": 1.6938742869328784, "grad_norm": 1.4240024089813232, "learning_rate": 3.6458000000000003e-06, "loss": 0.0147, "step": 31772 }, { "epoch": 1.6939276003625312, "grad_norm": 1.9371215105056763, "learning_rate": 3.6456000000000003e-06, "loss": 0.032, "step": 31773 }, { "epoch": 1.6939809137921844, "grad_norm": 2.2812047004699707, "learning_rate": 3.6454e-06, "loss": 0.0163, "step": 31774 }, { "epoch": 1.6940342272218372, "grad_norm": 2.7281084060668945, "learning_rate": 3.6452000000000003e-06, "loss": 0.029, "step": 31775 }, { "epoch": 1.6940875406514901, "grad_norm": 2.4908993244171143, "learning_rate": 3.6450000000000003e-06, "loss": 0.0227, "step": 31776 }, { "epoch": 1.6941408540811431, "grad_norm": 2.0057013034820557, "learning_rate": 3.6448000000000007e-06, "loss": 0.01, "step": 31777 }, { "epoch": 1.6941941675107959, "grad_norm": 0.44848933815956116, "learning_rate": 3.6446000000000002e-06, "loss": 0.0038, "step": 31778 }, { "epoch": 1.6942474809404489, "grad_norm": 1.2159688472747803, "learning_rate": 3.6444e-06, "loss": 0.0285, "step": 31779 }, { "epoch": 1.6943007943701018, "grad_norm": 3.232557535171509, "learning_rate": 3.6442000000000006e-06, "loss": 0.014, "step": 31780 }, { "epoch": 1.6943541077997546, "grad_norm": 1.149093508720398, "learning_rate": 3.644e-06, "loss": 0.0128, "step": 31781 }, { "epoch": 1.6944074212294078, "grad_norm": 0.9880579710006714, "learning_rate": 3.6438e-06, "loss": 0.0153, "step": 31782 }, { "epoch": 1.6944607346590606, "grad_norm": 2.872471809387207, "learning_rate": 3.6436000000000006e-06, "loss": 0.0534, "step": 31783 }, { "epoch": 1.6945140480887135, "grad_norm": 2.852022171020508, "learning_rate": 3.6434e-06, "loss": 0.0301, "step": 31784 }, { "epoch": 1.6945673615183665, "grad_norm": 2.457885980606079, "learning_rate": 3.6432e-06, "loss": 0.0288, "step": 31785 }, { "epoch": 1.6946206749480193, "grad_norm": 2.504781484603882, "learning_rate": 3.6430000000000005e-06, "loss": 0.0378, "step": 31786 }, { "epoch": 1.6946739883776725, "grad_norm": 0.9214878082275391, "learning_rate": 3.6428e-06, "loss": 0.0081, "step": 31787 }, { "epoch": 1.6947273018073252, "grad_norm": 1.2089242935180664, "learning_rate": 3.6426e-06, "loss": 0.0071, "step": 31788 }, { "epoch": 1.6947806152369782, "grad_norm": 1.771878957748413, "learning_rate": 3.6424000000000005e-06, "loss": 0.0248, "step": 31789 }, { "epoch": 1.6948339286666312, "grad_norm": 2.138596773147583, "learning_rate": 3.6422e-06, "loss": 0.0203, "step": 31790 }, { "epoch": 1.694887242096284, "grad_norm": 1.6122713088989258, "learning_rate": 3.6420000000000005e-06, "loss": 0.0304, "step": 31791 }, { "epoch": 1.6949405555259371, "grad_norm": 1.4266138076782227, "learning_rate": 3.6418000000000004e-06, "loss": 0.0299, "step": 31792 }, { "epoch": 1.69499386895559, "grad_norm": 2.1196584701538086, "learning_rate": 3.6416e-06, "loss": 0.0268, "step": 31793 }, { "epoch": 1.6950471823852429, "grad_norm": 0.3563535809516907, "learning_rate": 3.6414000000000004e-06, "loss": 0.0038, "step": 31794 }, { "epoch": 1.6951004958148959, "grad_norm": 1.7975220680236816, "learning_rate": 3.6412000000000004e-06, "loss": 0.0299, "step": 31795 }, { "epoch": 1.6951538092445486, "grad_norm": 0.9617748260498047, "learning_rate": 3.641e-06, "loss": 0.0172, "step": 31796 }, { "epoch": 1.6952071226742016, "grad_norm": 0.35785049200057983, "learning_rate": 3.6408000000000004e-06, "loss": 0.0032, "step": 31797 }, { "epoch": 1.6952604361038546, "grad_norm": 3.5416665077209473, "learning_rate": 3.6406000000000003e-06, "loss": 0.0389, "step": 31798 }, { "epoch": 1.6953137495335073, "grad_norm": 3.961556911468506, "learning_rate": 3.6404e-06, "loss": 0.0345, "step": 31799 }, { "epoch": 1.6953670629631605, "grad_norm": 1.3182228803634644, "learning_rate": 3.6402000000000003e-06, "loss": 0.043, "step": 31800 }, { "epoch": 1.6954203763928133, "grad_norm": 1.4913833141326904, "learning_rate": 3.6400000000000003e-06, "loss": 0.0108, "step": 31801 }, { "epoch": 1.6954736898224663, "grad_norm": 1.756583571434021, "learning_rate": 3.6398000000000007e-06, "loss": 0.0263, "step": 31802 }, { "epoch": 1.6955270032521192, "grad_norm": 1.465795636177063, "learning_rate": 3.6396000000000003e-06, "loss": 0.0166, "step": 31803 }, { "epoch": 1.695580316681772, "grad_norm": 1.2959725856781006, "learning_rate": 3.6394000000000003e-06, "loss": 0.0065, "step": 31804 }, { "epoch": 1.6956336301114252, "grad_norm": 1.988589882850647, "learning_rate": 3.6392000000000007e-06, "loss": 0.025, "step": 31805 }, { "epoch": 1.695686943541078, "grad_norm": 1.2314969301223755, "learning_rate": 3.6390000000000002e-06, "loss": 0.0068, "step": 31806 }, { "epoch": 1.695740256970731, "grad_norm": 1.0024690628051758, "learning_rate": 3.6388e-06, "loss": 0.0111, "step": 31807 }, { "epoch": 1.695793570400384, "grad_norm": 1.4091718196868896, "learning_rate": 3.6386000000000006e-06, "loss": 0.0353, "step": 31808 }, { "epoch": 1.6958468838300367, "grad_norm": 1.156313419342041, "learning_rate": 3.6384e-06, "loss": 0.0195, "step": 31809 }, { "epoch": 1.6959001972596899, "grad_norm": 0.9718109965324402, "learning_rate": 3.6382e-06, "loss": 0.0286, "step": 31810 }, { "epoch": 1.6959535106893426, "grad_norm": 1.6021192073822021, "learning_rate": 3.6380000000000006e-06, "loss": 0.0212, "step": 31811 }, { "epoch": 1.6960068241189956, "grad_norm": 0.6088162064552307, "learning_rate": 3.6378e-06, "loss": 0.0054, "step": 31812 }, { "epoch": 1.6960601375486486, "grad_norm": 1.5996443033218384, "learning_rate": 3.6376e-06, "loss": 0.0074, "step": 31813 }, { "epoch": 1.6961134509783014, "grad_norm": 1.95249342918396, "learning_rate": 3.6374000000000005e-06, "loss": 0.0085, "step": 31814 }, { "epoch": 1.6961667644079543, "grad_norm": 2.0906708240509033, "learning_rate": 3.6372e-06, "loss": 0.0142, "step": 31815 }, { "epoch": 1.6962200778376073, "grad_norm": 2.2676286697387695, "learning_rate": 3.6370000000000005e-06, "loss": 0.0205, "step": 31816 }, { "epoch": 1.69627339126726, "grad_norm": 1.4114198684692383, "learning_rate": 3.6368000000000005e-06, "loss": 0.0165, "step": 31817 }, { "epoch": 1.6963267046969133, "grad_norm": 1.2646056413650513, "learning_rate": 3.6366e-06, "loss": 0.0134, "step": 31818 }, { "epoch": 1.696380018126566, "grad_norm": 0.8915017247200012, "learning_rate": 3.6364000000000005e-06, "loss": 0.009, "step": 31819 }, { "epoch": 1.696433331556219, "grad_norm": 1.771235466003418, "learning_rate": 3.6362000000000004e-06, "loss": 0.0429, "step": 31820 }, { "epoch": 1.696486644985872, "grad_norm": 2.3570902347564697, "learning_rate": 3.636e-06, "loss": 0.0374, "step": 31821 }, { "epoch": 1.6965399584155247, "grad_norm": 1.6352618932724, "learning_rate": 3.6358000000000004e-06, "loss": 0.0114, "step": 31822 }, { "epoch": 1.696593271845178, "grad_norm": 1.8519781827926636, "learning_rate": 3.6356000000000004e-06, "loss": 0.0124, "step": 31823 }, { "epoch": 1.6966465852748307, "grad_norm": 1.636501669883728, "learning_rate": 3.6354e-06, "loss": 0.0283, "step": 31824 }, { "epoch": 1.6966998987044837, "grad_norm": 2.05739426612854, "learning_rate": 3.6352000000000004e-06, "loss": 0.0219, "step": 31825 }, { "epoch": 1.6967532121341367, "grad_norm": 1.3727490901947021, "learning_rate": 3.6350000000000003e-06, "loss": 0.0133, "step": 31826 }, { "epoch": 1.6968065255637894, "grad_norm": 1.6725400686264038, "learning_rate": 3.6348000000000008e-06, "loss": 0.0199, "step": 31827 }, { "epoch": 1.6968598389934424, "grad_norm": 0.23416465520858765, "learning_rate": 3.6346000000000003e-06, "loss": 0.0022, "step": 31828 }, { "epoch": 1.6969131524230954, "grad_norm": 2.3356435298919678, "learning_rate": 3.6344000000000003e-06, "loss": 0.0318, "step": 31829 }, { "epoch": 1.6969664658527484, "grad_norm": 1.433744192123413, "learning_rate": 3.6342000000000007e-06, "loss": 0.0164, "step": 31830 }, { "epoch": 1.6970197792824013, "grad_norm": 1.396213412284851, "learning_rate": 3.6340000000000003e-06, "loss": 0.0129, "step": 31831 }, { "epoch": 1.697073092712054, "grad_norm": 2.8349363803863525, "learning_rate": 3.6338000000000003e-06, "loss": 0.0398, "step": 31832 }, { "epoch": 1.697126406141707, "grad_norm": 1.952640414237976, "learning_rate": 3.6336000000000007e-06, "loss": 0.0226, "step": 31833 }, { "epoch": 1.69717971957136, "grad_norm": 2.302260160446167, "learning_rate": 3.6334000000000002e-06, "loss": 0.032, "step": 31834 }, { "epoch": 1.6972330330010128, "grad_norm": 6.895534992218018, "learning_rate": 3.6332000000000002e-06, "loss": 0.0375, "step": 31835 }, { "epoch": 1.697286346430666, "grad_norm": 1.9815294742584229, "learning_rate": 3.6330000000000006e-06, "loss": 0.0106, "step": 31836 }, { "epoch": 1.6973396598603188, "grad_norm": 1.4564532041549683, "learning_rate": 3.6328e-06, "loss": 0.014, "step": 31837 }, { "epoch": 1.6973929732899717, "grad_norm": 1.6870949268341064, "learning_rate": 3.6326000000000006e-06, "loss": 0.0147, "step": 31838 }, { "epoch": 1.6974462867196247, "grad_norm": 2.904773235321045, "learning_rate": 3.6324000000000006e-06, "loss": 0.0179, "step": 31839 }, { "epoch": 1.6974996001492775, "grad_norm": 3.3277223110198975, "learning_rate": 3.6322e-06, "loss": 0.0408, "step": 31840 }, { "epoch": 1.6975529135789307, "grad_norm": 0.7628690004348755, "learning_rate": 3.6320000000000005e-06, "loss": 0.0065, "step": 31841 }, { "epoch": 1.6976062270085834, "grad_norm": 1.8648548126220703, "learning_rate": 3.6318000000000005e-06, "loss": 0.0178, "step": 31842 }, { "epoch": 1.6976595404382364, "grad_norm": 2.3793370723724365, "learning_rate": 3.6316e-06, "loss": 0.0174, "step": 31843 }, { "epoch": 1.6977128538678894, "grad_norm": 0.6284881830215454, "learning_rate": 3.6314000000000005e-06, "loss": 0.0043, "step": 31844 }, { "epoch": 1.6977661672975422, "grad_norm": 2.1822352409362793, "learning_rate": 3.6312000000000005e-06, "loss": 0.0138, "step": 31845 }, { "epoch": 1.6978194807271951, "grad_norm": 1.392377495765686, "learning_rate": 3.631e-06, "loss": 0.0324, "step": 31846 }, { "epoch": 1.697872794156848, "grad_norm": 0.297389417886734, "learning_rate": 3.6308000000000005e-06, "loss": 0.0029, "step": 31847 }, { "epoch": 1.697926107586501, "grad_norm": 1.8651442527770996, "learning_rate": 3.6306000000000004e-06, "loss": 0.0176, "step": 31848 }, { "epoch": 1.697979421016154, "grad_norm": 1.7827776670455933, "learning_rate": 3.6304e-06, "loss": 0.0155, "step": 31849 }, { "epoch": 1.6980327344458068, "grad_norm": 1.3305147886276245, "learning_rate": 3.6302000000000004e-06, "loss": 0.0288, "step": 31850 }, { "epoch": 1.6980860478754598, "grad_norm": 2.486281394958496, "learning_rate": 3.6300000000000004e-06, "loss": 0.0271, "step": 31851 }, { "epoch": 1.6981393613051128, "grad_norm": 1.6510999202728271, "learning_rate": 3.6298000000000004e-06, "loss": 0.0248, "step": 31852 }, { "epoch": 1.6981926747347655, "grad_norm": 1.7019360065460205, "learning_rate": 3.6296000000000004e-06, "loss": 0.0238, "step": 31853 }, { "epoch": 1.6982459881644187, "grad_norm": 1.065605878829956, "learning_rate": 3.6294000000000004e-06, "loss": 0.0234, "step": 31854 }, { "epoch": 1.6982993015940715, "grad_norm": 1.9428602457046509, "learning_rate": 3.6292000000000003e-06, "loss": 0.0376, "step": 31855 }, { "epoch": 1.6983526150237245, "grad_norm": 1.6186227798461914, "learning_rate": 3.6290000000000003e-06, "loss": 0.0145, "step": 31856 }, { "epoch": 1.6984059284533775, "grad_norm": 0.9535903334617615, "learning_rate": 3.6288000000000003e-06, "loss": 0.0146, "step": 31857 }, { "epoch": 1.6984592418830302, "grad_norm": 0.7192803621292114, "learning_rate": 3.6286000000000003e-06, "loss": 0.006, "step": 31858 }, { "epoch": 1.6985125553126834, "grad_norm": 1.8519060611724854, "learning_rate": 3.6284000000000003e-06, "loss": 0.0162, "step": 31859 }, { "epoch": 1.6985658687423362, "grad_norm": 2.328944444656372, "learning_rate": 3.6282000000000003e-06, "loss": 0.0219, "step": 31860 }, { "epoch": 1.6986191821719892, "grad_norm": 1.1100530624389648, "learning_rate": 3.6280000000000002e-06, "loss": 0.0067, "step": 31861 }, { "epoch": 1.6986724956016421, "grad_norm": 1.5435247421264648, "learning_rate": 3.6278000000000002e-06, "loss": 0.0126, "step": 31862 }, { "epoch": 1.6987258090312949, "grad_norm": 3.052757501602173, "learning_rate": 3.6276000000000006e-06, "loss": 0.024, "step": 31863 }, { "epoch": 1.6987791224609479, "grad_norm": 3.108931064605713, "learning_rate": 3.6274e-06, "loss": 0.0184, "step": 31864 }, { "epoch": 1.6988324358906008, "grad_norm": 1.665269136428833, "learning_rate": 3.6272e-06, "loss": 0.0132, "step": 31865 }, { "epoch": 1.6988857493202538, "grad_norm": 0.48106858134269714, "learning_rate": 3.6270000000000006e-06, "loss": 0.0041, "step": 31866 }, { "epoch": 1.6989390627499068, "grad_norm": 1.8683058023452759, "learning_rate": 3.6268e-06, "loss": 0.0144, "step": 31867 }, { "epoch": 1.6989923761795596, "grad_norm": 2.9616308212280273, "learning_rate": 3.6266e-06, "loss": 0.0199, "step": 31868 }, { "epoch": 1.6990456896092125, "grad_norm": 1.7062907218933105, "learning_rate": 3.6264000000000005e-06, "loss": 0.0199, "step": 31869 }, { "epoch": 1.6990990030388655, "grad_norm": 1.62494957447052, "learning_rate": 3.6262e-06, "loss": 0.0118, "step": 31870 }, { "epoch": 1.6991523164685183, "grad_norm": 5.506199359893799, "learning_rate": 3.626e-06, "loss": 0.0455, "step": 31871 }, { "epoch": 1.6992056298981715, "grad_norm": 1.6703678369522095, "learning_rate": 3.6258000000000005e-06, "loss": 0.0214, "step": 31872 }, { "epoch": 1.6992589433278242, "grad_norm": 0.44710686802864075, "learning_rate": 3.6256e-06, "loss": 0.0037, "step": 31873 }, { "epoch": 1.6993122567574772, "grad_norm": 1.3737443685531616, "learning_rate": 3.6254e-06, "loss": 0.0125, "step": 31874 }, { "epoch": 1.6993655701871302, "grad_norm": 2.1072983741760254, "learning_rate": 3.6252000000000005e-06, "loss": 0.0308, "step": 31875 }, { "epoch": 1.699418883616783, "grad_norm": 1.9980312585830688, "learning_rate": 3.625e-06, "loss": 0.0471, "step": 31876 }, { "epoch": 1.6994721970464362, "grad_norm": 2.039863109588623, "learning_rate": 3.6248000000000004e-06, "loss": 0.0242, "step": 31877 }, { "epoch": 1.699525510476089, "grad_norm": 1.518931269645691, "learning_rate": 3.6246000000000004e-06, "loss": 0.0173, "step": 31878 }, { "epoch": 1.6995788239057419, "grad_norm": 2.247929573059082, "learning_rate": 3.6244e-06, "loss": 0.0257, "step": 31879 }, { "epoch": 1.6996321373353949, "grad_norm": 2.209190607070923, "learning_rate": 3.6242000000000004e-06, "loss": 0.0268, "step": 31880 }, { "epoch": 1.6996854507650476, "grad_norm": 3.863455057144165, "learning_rate": 3.6240000000000004e-06, "loss": 0.0219, "step": 31881 }, { "epoch": 1.6997387641947006, "grad_norm": 2.716085195541382, "learning_rate": 3.6238e-06, "loss": 0.0199, "step": 31882 }, { "epoch": 1.6997920776243536, "grad_norm": 1.495345115661621, "learning_rate": 3.6236000000000003e-06, "loss": 0.0144, "step": 31883 }, { "epoch": 1.6998453910540063, "grad_norm": 1.3907004594802856, "learning_rate": 3.6234000000000003e-06, "loss": 0.0053, "step": 31884 }, { "epoch": 1.6998987044836595, "grad_norm": 2.0208992958068848, "learning_rate": 3.6232e-06, "loss": 0.0266, "step": 31885 }, { "epoch": 1.6999520179133123, "grad_norm": 1.5401009321212769, "learning_rate": 3.6230000000000003e-06, "loss": 0.0123, "step": 31886 }, { "epoch": 1.7000053313429653, "grad_norm": 0.9189444184303284, "learning_rate": 3.6228000000000003e-06, "loss": 0.007, "step": 31887 }, { "epoch": 1.7000586447726183, "grad_norm": 1.219277262687683, "learning_rate": 3.6226000000000007e-06, "loss": 0.0179, "step": 31888 }, { "epoch": 1.700111958202271, "grad_norm": 1.1745949983596802, "learning_rate": 3.6224000000000002e-06, "loss": 0.0177, "step": 31889 }, { "epoch": 1.7001652716319242, "grad_norm": 0.4708542227745056, "learning_rate": 3.6222000000000002e-06, "loss": 0.0042, "step": 31890 }, { "epoch": 1.700218585061577, "grad_norm": 1.4191906452178955, "learning_rate": 3.6220000000000006e-06, "loss": 0.0238, "step": 31891 }, { "epoch": 1.70027189849123, "grad_norm": 1.201543927192688, "learning_rate": 3.6218e-06, "loss": 0.0163, "step": 31892 }, { "epoch": 1.700325211920883, "grad_norm": 1.864209771156311, "learning_rate": 3.6216e-06, "loss": 0.0318, "step": 31893 }, { "epoch": 1.7003785253505357, "grad_norm": 1.0355643033981323, "learning_rate": 3.6214000000000006e-06, "loss": 0.0063, "step": 31894 }, { "epoch": 1.7004318387801889, "grad_norm": 1.20413076877594, "learning_rate": 3.6212e-06, "loss": 0.0082, "step": 31895 }, { "epoch": 1.7004851522098416, "grad_norm": 1.5972226858139038, "learning_rate": 3.621e-06, "loss": 0.0106, "step": 31896 }, { "epoch": 1.7005384656394946, "grad_norm": 1.613599419593811, "learning_rate": 3.6208000000000006e-06, "loss": 0.0297, "step": 31897 }, { "epoch": 1.7005917790691476, "grad_norm": 2.213013172149658, "learning_rate": 3.6206e-06, "loss": 0.0626, "step": 31898 }, { "epoch": 1.7006450924988004, "grad_norm": 2.9034857749938965, "learning_rate": 3.6204e-06, "loss": 0.0228, "step": 31899 }, { "epoch": 1.7006984059284533, "grad_norm": 2.821361541748047, "learning_rate": 3.6202000000000005e-06, "loss": 0.0166, "step": 31900 }, { "epoch": 1.7007517193581063, "grad_norm": 0.7785534858703613, "learning_rate": 3.62e-06, "loss": 0.0044, "step": 31901 }, { "epoch": 1.700805032787759, "grad_norm": 1.9399311542510986, "learning_rate": 3.6198000000000005e-06, "loss": 0.0232, "step": 31902 }, { "epoch": 1.7008583462174123, "grad_norm": 2.3908252716064453, "learning_rate": 3.6196000000000005e-06, "loss": 0.0253, "step": 31903 }, { "epoch": 1.700911659647065, "grad_norm": 0.8372933864593506, "learning_rate": 3.6194e-06, "loss": 0.0107, "step": 31904 }, { "epoch": 1.700964973076718, "grad_norm": 0.479915589094162, "learning_rate": 3.6192000000000004e-06, "loss": 0.0039, "step": 31905 }, { "epoch": 1.701018286506371, "grad_norm": 1.5745183229446411, "learning_rate": 3.6190000000000004e-06, "loss": 0.0088, "step": 31906 }, { "epoch": 1.7010715999360237, "grad_norm": 1.1713547706604004, "learning_rate": 3.6188e-06, "loss": 0.0083, "step": 31907 }, { "epoch": 1.701124913365677, "grad_norm": 3.471454381942749, "learning_rate": 3.6186000000000004e-06, "loss": 0.0349, "step": 31908 }, { "epoch": 1.7011782267953297, "grad_norm": 1.2123010158538818, "learning_rate": 3.6184000000000004e-06, "loss": 0.0078, "step": 31909 }, { "epoch": 1.7012315402249827, "grad_norm": 2.4811630249023438, "learning_rate": 3.6182e-06, "loss": 0.0289, "step": 31910 }, { "epoch": 1.7012848536546357, "grad_norm": 0.33926162123680115, "learning_rate": 3.6180000000000003e-06, "loss": 0.0028, "step": 31911 }, { "epoch": 1.7013381670842884, "grad_norm": 0.9037438631057739, "learning_rate": 3.6178000000000003e-06, "loss": 0.0164, "step": 31912 }, { "epoch": 1.7013914805139416, "grad_norm": 2.239445686340332, "learning_rate": 3.6176000000000007e-06, "loss": 0.0582, "step": 31913 }, { "epoch": 1.7014447939435944, "grad_norm": 2.2426702976226807, "learning_rate": 3.6174000000000003e-06, "loss": 0.0134, "step": 31914 }, { "epoch": 1.7014981073732474, "grad_norm": 2.0729339122772217, "learning_rate": 3.6172000000000003e-06, "loss": 0.0158, "step": 31915 }, { "epoch": 1.7015514208029003, "grad_norm": 1.9525327682495117, "learning_rate": 3.6170000000000007e-06, "loss": 0.0235, "step": 31916 }, { "epoch": 1.701604734232553, "grad_norm": 2.7113521099090576, "learning_rate": 3.6168000000000003e-06, "loss": 0.0529, "step": 31917 }, { "epoch": 1.701658047662206, "grad_norm": 2.1716225147247314, "learning_rate": 3.6166000000000002e-06, "loss": 0.0219, "step": 31918 }, { "epoch": 1.701711361091859, "grad_norm": 0.32361075282096863, "learning_rate": 3.6164000000000006e-06, "loss": 0.0031, "step": 31919 }, { "epoch": 1.7017646745215118, "grad_norm": 2.8165910243988037, "learning_rate": 3.6162e-06, "loss": 0.0175, "step": 31920 }, { "epoch": 1.701817987951165, "grad_norm": 1.3608790636062622, "learning_rate": 3.616e-06, "loss": 0.0193, "step": 31921 }, { "epoch": 1.7018713013808178, "grad_norm": 1.2037876844406128, "learning_rate": 3.6158000000000006e-06, "loss": 0.012, "step": 31922 }, { "epoch": 1.7019246148104707, "grad_norm": 3.2184805870056152, "learning_rate": 3.6156e-06, "loss": 0.0328, "step": 31923 }, { "epoch": 1.7019779282401237, "grad_norm": 2.290489912033081, "learning_rate": 3.6154e-06, "loss": 0.0256, "step": 31924 }, { "epoch": 1.7020312416697765, "grad_norm": 2.350621461868286, "learning_rate": 3.6152000000000006e-06, "loss": 0.0233, "step": 31925 }, { "epoch": 1.7020845550994297, "grad_norm": 1.3836557865142822, "learning_rate": 3.615e-06, "loss": 0.0242, "step": 31926 }, { "epoch": 1.7021378685290824, "grad_norm": 1.8004367351531982, "learning_rate": 3.6148000000000005e-06, "loss": 0.0132, "step": 31927 }, { "epoch": 1.7021911819587354, "grad_norm": 2.023909091949463, "learning_rate": 3.6146000000000005e-06, "loss": 0.0349, "step": 31928 }, { "epoch": 1.7022444953883884, "grad_norm": 1.5440336465835571, "learning_rate": 3.6144e-06, "loss": 0.0105, "step": 31929 }, { "epoch": 1.7022978088180412, "grad_norm": 2.9801793098449707, "learning_rate": 3.6142000000000005e-06, "loss": 0.0324, "step": 31930 }, { "epoch": 1.7023511222476944, "grad_norm": 1.1606078147888184, "learning_rate": 3.6140000000000005e-06, "loss": 0.0158, "step": 31931 }, { "epoch": 1.7024044356773471, "grad_norm": 3.381453037261963, "learning_rate": 3.6138e-06, "loss": 0.0174, "step": 31932 }, { "epoch": 1.702457749107, "grad_norm": 2.307130813598633, "learning_rate": 3.6136000000000004e-06, "loss": 0.0114, "step": 31933 }, { "epoch": 1.702511062536653, "grad_norm": 1.7996026277542114, "learning_rate": 3.6134000000000004e-06, "loss": 0.028, "step": 31934 }, { "epoch": 1.7025643759663058, "grad_norm": 2.197105646133423, "learning_rate": 3.6132e-06, "loss": 0.0283, "step": 31935 }, { "epoch": 1.7026176893959588, "grad_norm": 1.906683087348938, "learning_rate": 3.6130000000000004e-06, "loss": 0.0146, "step": 31936 }, { "epoch": 1.7026710028256118, "grad_norm": 2.0115392208099365, "learning_rate": 3.6128000000000004e-06, "loss": 0.0349, "step": 31937 }, { "epoch": 1.7027243162552645, "grad_norm": 0.8515500426292419, "learning_rate": 3.6126000000000004e-06, "loss": 0.0057, "step": 31938 }, { "epoch": 1.7027776296849177, "grad_norm": 1.1305145025253296, "learning_rate": 3.6124000000000003e-06, "loss": 0.015, "step": 31939 }, { "epoch": 1.7028309431145705, "grad_norm": 2.470752477645874, "learning_rate": 3.6122000000000003e-06, "loss": 0.0542, "step": 31940 }, { "epoch": 1.7028842565442235, "grad_norm": 3.316718578338623, "learning_rate": 3.6120000000000003e-06, "loss": 0.0326, "step": 31941 }, { "epoch": 1.7029375699738765, "grad_norm": 2.1340060234069824, "learning_rate": 3.6118000000000003e-06, "loss": 0.0413, "step": 31942 }, { "epoch": 1.7029908834035292, "grad_norm": 1.7708090543746948, "learning_rate": 3.6116000000000003e-06, "loss": 0.0415, "step": 31943 }, { "epoch": 1.7030441968331824, "grad_norm": 1.3651951551437378, "learning_rate": 3.6114000000000003e-06, "loss": 0.0146, "step": 31944 }, { "epoch": 1.7030975102628352, "grad_norm": 1.8096450567245483, "learning_rate": 3.6112000000000003e-06, "loss": 0.0393, "step": 31945 }, { "epoch": 1.7031508236924882, "grad_norm": 2.1583034992218018, "learning_rate": 3.6110000000000002e-06, "loss": 0.0342, "step": 31946 }, { "epoch": 1.7032041371221411, "grad_norm": 2.0771377086639404, "learning_rate": 3.6108000000000002e-06, "loss": 0.0162, "step": 31947 }, { "epoch": 1.703257450551794, "grad_norm": 1.7421714067459106, "learning_rate": 3.6106e-06, "loss": 0.014, "step": 31948 }, { "epoch": 1.7033107639814469, "grad_norm": 3.202545642852783, "learning_rate": 3.6104000000000006e-06, "loss": 0.0222, "step": 31949 }, { "epoch": 1.7033640774110999, "grad_norm": 2.986710786819458, "learning_rate": 3.6102e-06, "loss": 0.0524, "step": 31950 }, { "epoch": 1.7034173908407528, "grad_norm": 3.1252341270446777, "learning_rate": 3.61e-06, "loss": 0.0389, "step": 31951 }, { "epoch": 1.7034707042704058, "grad_norm": 2.7859058380126953, "learning_rate": 3.6098000000000006e-06, "loss": 0.0131, "step": 31952 }, { "epoch": 1.7035240177000586, "grad_norm": 1.515696406364441, "learning_rate": 3.6096e-06, "loss": 0.0176, "step": 31953 }, { "epoch": 1.7035773311297115, "grad_norm": 0.8404545783996582, "learning_rate": 3.6094e-06, "loss": 0.0073, "step": 31954 }, { "epoch": 1.7036306445593645, "grad_norm": 2.4634618759155273, "learning_rate": 3.6092000000000005e-06, "loss": 0.0313, "step": 31955 }, { "epoch": 1.7036839579890173, "grad_norm": 2.2418160438537598, "learning_rate": 3.609e-06, "loss": 0.0195, "step": 31956 }, { "epoch": 1.7037372714186705, "grad_norm": 1.5481891632080078, "learning_rate": 3.6088e-06, "loss": 0.009, "step": 31957 }, { "epoch": 1.7037905848483232, "grad_norm": 1.3346034288406372, "learning_rate": 3.6086000000000005e-06, "loss": 0.0076, "step": 31958 }, { "epoch": 1.7038438982779762, "grad_norm": 0.8955364227294922, "learning_rate": 3.6084e-06, "loss": 0.0134, "step": 31959 }, { "epoch": 1.7038972117076292, "grad_norm": 1.4287989139556885, "learning_rate": 3.6082e-06, "loss": 0.0177, "step": 31960 }, { "epoch": 1.703950525137282, "grad_norm": 0.58418208360672, "learning_rate": 3.6080000000000004e-06, "loss": 0.006, "step": 31961 }, { "epoch": 1.7040038385669352, "grad_norm": 2.697239398956299, "learning_rate": 3.6078e-06, "loss": 0.0602, "step": 31962 }, { "epoch": 1.704057151996588, "grad_norm": 1.8309606313705444, "learning_rate": 3.6076000000000004e-06, "loss": 0.0163, "step": 31963 }, { "epoch": 1.704110465426241, "grad_norm": 1.3980683088302612, "learning_rate": 3.6074000000000004e-06, "loss": 0.0105, "step": 31964 }, { "epoch": 1.7041637788558939, "grad_norm": 2.4504218101501465, "learning_rate": 3.6072e-06, "loss": 0.0568, "step": 31965 }, { "epoch": 1.7042170922855466, "grad_norm": 1.6565542221069336, "learning_rate": 3.6070000000000004e-06, "loss": 0.026, "step": 31966 }, { "epoch": 1.7042704057151996, "grad_norm": 1.4237558841705322, "learning_rate": 3.6068000000000003e-06, "loss": 0.0196, "step": 31967 }, { "epoch": 1.7043237191448526, "grad_norm": 0.8156107068061829, "learning_rate": 3.6066e-06, "loss": 0.0074, "step": 31968 }, { "epoch": 1.7043770325745056, "grad_norm": 1.4883676767349243, "learning_rate": 3.6064000000000003e-06, "loss": 0.0253, "step": 31969 }, { "epoch": 1.7044303460041585, "grad_norm": 1.7545843124389648, "learning_rate": 3.6062000000000003e-06, "loss": 0.0287, "step": 31970 }, { "epoch": 1.7044836594338113, "grad_norm": 1.8205527067184448, "learning_rate": 3.606e-06, "loss": 0.0287, "step": 31971 }, { "epoch": 1.7045369728634643, "grad_norm": 0.8545125126838684, "learning_rate": 3.6058000000000003e-06, "loss": 0.0072, "step": 31972 }, { "epoch": 1.7045902862931173, "grad_norm": 1.3219919204711914, "learning_rate": 3.6056000000000003e-06, "loss": 0.0189, "step": 31973 }, { "epoch": 1.70464359972277, "grad_norm": 1.8563363552093506, "learning_rate": 3.6054000000000007e-06, "loss": 0.0114, "step": 31974 }, { "epoch": 1.7046969131524232, "grad_norm": 0.4784048795700073, "learning_rate": 3.6052000000000002e-06, "loss": 0.005, "step": 31975 }, { "epoch": 1.704750226582076, "grad_norm": 1.1073403358459473, "learning_rate": 3.6050000000000002e-06, "loss": 0.0115, "step": 31976 }, { "epoch": 1.704803540011729, "grad_norm": 1.4298241138458252, "learning_rate": 3.6048000000000006e-06, "loss": 0.0408, "step": 31977 }, { "epoch": 1.704856853441382, "grad_norm": 0.9492141604423523, "learning_rate": 3.6046e-06, "loss": 0.0086, "step": 31978 }, { "epoch": 1.7049101668710347, "grad_norm": 2.743603229522705, "learning_rate": 3.6044e-06, "loss": 0.0141, "step": 31979 }, { "epoch": 1.704963480300688, "grad_norm": 0.9272220730781555, "learning_rate": 3.6042000000000006e-06, "loss": 0.0172, "step": 31980 }, { "epoch": 1.7050167937303407, "grad_norm": 1.6435186862945557, "learning_rate": 3.604e-06, "loss": 0.0296, "step": 31981 }, { "epoch": 1.7050701071599936, "grad_norm": 0.9674464464187622, "learning_rate": 3.6038e-06, "loss": 0.0092, "step": 31982 }, { "epoch": 1.7051234205896466, "grad_norm": 1.1811048984527588, "learning_rate": 3.6036000000000005e-06, "loss": 0.0255, "step": 31983 }, { "epoch": 1.7051767340192994, "grad_norm": 1.2077242136001587, "learning_rate": 3.6034e-06, "loss": 0.0193, "step": 31984 }, { "epoch": 1.7052300474489523, "grad_norm": 3.9378631114959717, "learning_rate": 3.6032e-06, "loss": 0.0389, "step": 31985 }, { "epoch": 1.7052833608786053, "grad_norm": 1.653064250946045, "learning_rate": 3.6030000000000005e-06, "loss": 0.0191, "step": 31986 }, { "epoch": 1.7053366743082583, "grad_norm": 1.4271652698516846, "learning_rate": 3.6028e-06, "loss": 0.0099, "step": 31987 }, { "epoch": 1.7053899877379113, "grad_norm": 2.1914918422698975, "learning_rate": 3.6026000000000005e-06, "loss": 0.0453, "step": 31988 }, { "epoch": 1.705443301167564, "grad_norm": 2.5882210731506348, "learning_rate": 3.6024000000000004e-06, "loss": 0.0384, "step": 31989 }, { "epoch": 1.705496614597217, "grad_norm": 1.3593209981918335, "learning_rate": 3.6022e-06, "loss": 0.021, "step": 31990 }, { "epoch": 1.70554992802687, "grad_norm": 1.2863023281097412, "learning_rate": 3.6020000000000004e-06, "loss": 0.0084, "step": 31991 }, { "epoch": 1.7056032414565228, "grad_norm": 1.7614763975143433, "learning_rate": 3.6018000000000004e-06, "loss": 0.0219, "step": 31992 }, { "epoch": 1.705656554886176, "grad_norm": 3.041053295135498, "learning_rate": 3.6016e-06, "loss": 0.0171, "step": 31993 }, { "epoch": 1.7057098683158287, "grad_norm": 3.116619110107422, "learning_rate": 3.6014000000000004e-06, "loss": 0.0215, "step": 31994 }, { "epoch": 1.7057631817454817, "grad_norm": 3.395559310913086, "learning_rate": 3.6012000000000004e-06, "loss": 0.0324, "step": 31995 }, { "epoch": 1.7058164951751347, "grad_norm": 5.3986639976501465, "learning_rate": 3.601e-06, "loss": 0.0478, "step": 31996 }, { "epoch": 1.7058698086047874, "grad_norm": 1.2409005165100098, "learning_rate": 3.6008000000000003e-06, "loss": 0.0105, "step": 31997 }, { "epoch": 1.7059231220344406, "grad_norm": 1.0157480239868164, "learning_rate": 3.6006000000000003e-06, "loss": 0.0325, "step": 31998 }, { "epoch": 1.7059764354640934, "grad_norm": 2.663161039352417, "learning_rate": 3.6004000000000007e-06, "loss": 0.0169, "step": 31999 }, { "epoch": 1.7060297488937464, "grad_norm": 0.588437020778656, "learning_rate": 3.6002000000000003e-06, "loss": 0.005, "step": 32000 }, { "epoch": 1.7060830623233993, "grad_norm": 1.3692944049835205, "learning_rate": 3.6000000000000003e-06, "loss": 0.0093, "step": 32001 }, { "epoch": 1.706136375753052, "grad_norm": 0.24679355323314667, "learning_rate": 3.5998000000000007e-06, "loss": 0.0028, "step": 32002 }, { "epoch": 1.706189689182705, "grad_norm": 1.2131379842758179, "learning_rate": 3.5996000000000002e-06, "loss": 0.0152, "step": 32003 }, { "epoch": 1.706243002612358, "grad_norm": 1.1735827922821045, "learning_rate": 3.5994000000000002e-06, "loss": 0.026, "step": 32004 }, { "epoch": 1.7062963160420108, "grad_norm": 0.9357248544692993, "learning_rate": 3.5992000000000006e-06, "loss": 0.0094, "step": 32005 }, { "epoch": 1.706349629471664, "grad_norm": 1.676845908164978, "learning_rate": 3.599e-06, "loss": 0.0229, "step": 32006 }, { "epoch": 1.7064029429013168, "grad_norm": 1.4702268838882446, "learning_rate": 3.5988e-06, "loss": 0.0056, "step": 32007 }, { "epoch": 1.7064562563309698, "grad_norm": 0.9025877714157104, "learning_rate": 3.5986000000000006e-06, "loss": 0.0164, "step": 32008 }, { "epoch": 1.7065095697606227, "grad_norm": 1.1701303720474243, "learning_rate": 3.5984e-06, "loss": 0.0221, "step": 32009 }, { "epoch": 1.7065628831902755, "grad_norm": 1.4030934572219849, "learning_rate": 3.5982e-06, "loss": 0.0069, "step": 32010 }, { "epoch": 1.7066161966199287, "grad_norm": 0.6356446743011475, "learning_rate": 3.5980000000000005e-06, "loss": 0.0065, "step": 32011 }, { "epoch": 1.7066695100495815, "grad_norm": 2.288579225540161, "learning_rate": 3.5978e-06, "loss": 0.025, "step": 32012 }, { "epoch": 1.7067228234792344, "grad_norm": 2.143097162246704, "learning_rate": 3.5976000000000005e-06, "loss": 0.0178, "step": 32013 }, { "epoch": 1.7067761369088874, "grad_norm": 2.4468045234680176, "learning_rate": 3.5974000000000005e-06, "loss": 0.0158, "step": 32014 }, { "epoch": 1.7068294503385402, "grad_norm": 0.9187846183776855, "learning_rate": 3.5972e-06, "loss": 0.008, "step": 32015 }, { "epoch": 1.7068827637681934, "grad_norm": 1.7634230852127075, "learning_rate": 3.5970000000000005e-06, "loss": 0.0096, "step": 32016 }, { "epoch": 1.7069360771978461, "grad_norm": 1.1078776121139526, "learning_rate": 3.5968000000000004e-06, "loss": 0.0196, "step": 32017 }, { "epoch": 1.706989390627499, "grad_norm": 2.3982887268066406, "learning_rate": 3.5966e-06, "loss": 0.0134, "step": 32018 }, { "epoch": 1.707042704057152, "grad_norm": 0.7579617500305176, "learning_rate": 3.5964000000000004e-06, "loss": 0.006, "step": 32019 }, { "epoch": 1.7070960174868048, "grad_norm": 1.8074910640716553, "learning_rate": 3.5962000000000004e-06, "loss": 0.0122, "step": 32020 }, { "epoch": 1.7071493309164578, "grad_norm": 2.032615900039673, "learning_rate": 3.596e-06, "loss": 0.0234, "step": 32021 }, { "epoch": 1.7072026443461108, "grad_norm": 1.228733777999878, "learning_rate": 3.5958000000000004e-06, "loss": 0.0146, "step": 32022 }, { "epoch": 1.7072559577757636, "grad_norm": 1.0254325866699219, "learning_rate": 3.5956000000000004e-06, "loss": 0.0177, "step": 32023 }, { "epoch": 1.7073092712054168, "grad_norm": 1.0698657035827637, "learning_rate": 3.5954000000000008e-06, "loss": 0.0057, "step": 32024 }, { "epoch": 1.7073625846350695, "grad_norm": 2.4380884170532227, "learning_rate": 3.5952000000000003e-06, "loss": 0.0134, "step": 32025 }, { "epoch": 1.7074158980647225, "grad_norm": 1.7673609256744385, "learning_rate": 3.5950000000000003e-06, "loss": 0.014, "step": 32026 }, { "epoch": 1.7074692114943755, "grad_norm": 1.6927725076675415, "learning_rate": 3.5948000000000007e-06, "loss": 0.0388, "step": 32027 }, { "epoch": 1.7075225249240282, "grad_norm": 0.7835673093795776, "learning_rate": 3.5946000000000003e-06, "loss": 0.0044, "step": 32028 }, { "epoch": 1.7075758383536814, "grad_norm": 0.6587985754013062, "learning_rate": 3.5944000000000003e-06, "loss": 0.0038, "step": 32029 }, { "epoch": 1.7076291517833342, "grad_norm": 2.034501791000366, "learning_rate": 3.5942000000000007e-06, "loss": 0.0084, "step": 32030 }, { "epoch": 1.7076824652129872, "grad_norm": 0.31817659735679626, "learning_rate": 3.5940000000000002e-06, "loss": 0.0027, "step": 32031 }, { "epoch": 1.7077357786426401, "grad_norm": 3.0757715702056885, "learning_rate": 3.5938000000000002e-06, "loss": 0.0313, "step": 32032 }, { "epoch": 1.707789092072293, "grad_norm": 0.2710801959037781, "learning_rate": 3.5936000000000006e-06, "loss": 0.0025, "step": 32033 }, { "epoch": 1.707842405501946, "grad_norm": 1.420121192932129, "learning_rate": 3.5934e-06, "loss": 0.0054, "step": 32034 }, { "epoch": 1.7078957189315989, "grad_norm": 2.0222880840301514, "learning_rate": 3.5932e-06, "loss": 0.0515, "step": 32035 }, { "epoch": 1.7079490323612518, "grad_norm": 3.2744767665863037, "learning_rate": 3.5930000000000006e-06, "loss": 0.0485, "step": 32036 }, { "epoch": 1.7080023457909048, "grad_norm": 3.188687562942505, "learning_rate": 3.5928e-06, "loss": 0.0595, "step": 32037 }, { "epoch": 1.7080556592205576, "grad_norm": 1.3338372707366943, "learning_rate": 3.5926000000000006e-06, "loss": 0.0204, "step": 32038 }, { "epoch": 1.7081089726502106, "grad_norm": 1.2483683824539185, "learning_rate": 3.5924000000000005e-06, "loss": 0.0148, "step": 32039 }, { "epoch": 1.7081622860798635, "grad_norm": 1.441834807395935, "learning_rate": 3.5922e-06, "loss": 0.0167, "step": 32040 }, { "epoch": 1.7082155995095163, "grad_norm": 0.6157761812210083, "learning_rate": 3.5920000000000005e-06, "loss": 0.0043, "step": 32041 }, { "epoch": 1.7082689129391695, "grad_norm": 2.6955208778381348, "learning_rate": 3.5918000000000005e-06, "loss": 0.0183, "step": 32042 }, { "epoch": 1.7083222263688222, "grad_norm": 1.8361502885818481, "learning_rate": 3.5916e-06, "loss": 0.012, "step": 32043 }, { "epoch": 1.7083755397984752, "grad_norm": 3.0265400409698486, "learning_rate": 3.5914000000000005e-06, "loss": 0.0248, "step": 32044 }, { "epoch": 1.7084288532281282, "grad_norm": 0.48784127831459045, "learning_rate": 3.5912000000000004e-06, "loss": 0.0032, "step": 32045 }, { "epoch": 1.708482166657781, "grad_norm": 1.4305552244186401, "learning_rate": 3.591e-06, "loss": 0.0052, "step": 32046 }, { "epoch": 1.7085354800874342, "grad_norm": 1.4063105583190918, "learning_rate": 3.5908000000000004e-06, "loss": 0.0211, "step": 32047 }, { "epoch": 1.708588793517087, "grad_norm": 1.797302484512329, "learning_rate": 3.5906000000000004e-06, "loss": 0.0103, "step": 32048 }, { "epoch": 1.70864210694674, "grad_norm": 2.5303955078125, "learning_rate": 3.5904000000000004e-06, "loss": 0.0122, "step": 32049 }, { "epoch": 1.7086954203763929, "grad_norm": 1.3262090682983398, "learning_rate": 3.5902000000000004e-06, "loss": 0.0239, "step": 32050 }, { "epoch": 1.7087487338060456, "grad_norm": 2.293673515319824, "learning_rate": 3.5900000000000004e-06, "loss": 0.035, "step": 32051 }, { "epoch": 1.7088020472356986, "grad_norm": 1.1831414699554443, "learning_rate": 3.5898000000000003e-06, "loss": 0.0188, "step": 32052 }, { "epoch": 1.7088553606653516, "grad_norm": 1.4633280038833618, "learning_rate": 3.5896000000000003e-06, "loss": 0.0131, "step": 32053 }, { "epoch": 1.7089086740950046, "grad_norm": 0.8978869318962097, "learning_rate": 3.5894000000000003e-06, "loss": 0.007, "step": 32054 }, { "epoch": 1.7089619875246576, "grad_norm": 1.451790690422058, "learning_rate": 3.5892000000000003e-06, "loss": 0.0279, "step": 32055 }, { "epoch": 1.7090153009543103, "grad_norm": 1.793035864830017, "learning_rate": 3.5890000000000003e-06, "loss": 0.0129, "step": 32056 }, { "epoch": 1.7090686143839633, "grad_norm": 3.448819160461426, "learning_rate": 3.5888000000000003e-06, "loss": 0.028, "step": 32057 }, { "epoch": 1.7091219278136163, "grad_norm": 2.1162564754486084, "learning_rate": 3.5886000000000003e-06, "loss": 0.0188, "step": 32058 }, { "epoch": 1.709175241243269, "grad_norm": 1.2717130184173584, "learning_rate": 3.5884000000000002e-06, "loss": 0.0056, "step": 32059 }, { "epoch": 1.7092285546729222, "grad_norm": 2.7558131217956543, "learning_rate": 3.5882000000000002e-06, "loss": 0.015, "step": 32060 }, { "epoch": 1.709281868102575, "grad_norm": 1.1736624240875244, "learning_rate": 3.588e-06, "loss": 0.0091, "step": 32061 }, { "epoch": 1.709335181532228, "grad_norm": 0.919400155544281, "learning_rate": 3.5878e-06, "loss": 0.0071, "step": 32062 }, { "epoch": 1.709388494961881, "grad_norm": 2.5277321338653564, "learning_rate": 3.5876000000000006e-06, "loss": 0.0332, "step": 32063 }, { "epoch": 1.7094418083915337, "grad_norm": 1.8652629852294922, "learning_rate": 3.5874e-06, "loss": 0.0128, "step": 32064 }, { "epoch": 1.709495121821187, "grad_norm": 1.921165943145752, "learning_rate": 3.5872e-06, "loss": 0.0173, "step": 32065 }, { "epoch": 1.7095484352508397, "grad_norm": 1.7958996295928955, "learning_rate": 3.5870000000000006e-06, "loss": 0.0581, "step": 32066 }, { "epoch": 1.7096017486804926, "grad_norm": 4.1766533851623535, "learning_rate": 3.5868e-06, "loss": 0.0219, "step": 32067 }, { "epoch": 1.7096550621101456, "grad_norm": 2.2340810298919678, "learning_rate": 3.5866e-06, "loss": 0.0207, "step": 32068 }, { "epoch": 1.7097083755397984, "grad_norm": 0.9549571871757507, "learning_rate": 3.5864000000000005e-06, "loss": 0.0049, "step": 32069 }, { "epoch": 1.7097616889694514, "grad_norm": 2.1999757289886475, "learning_rate": 3.5862e-06, "loss": 0.0191, "step": 32070 }, { "epoch": 1.7098150023991043, "grad_norm": 2.9558849334716797, "learning_rate": 3.586e-06, "loss": 0.0285, "step": 32071 }, { "epoch": 1.7098683158287573, "grad_norm": 3.0901987552642822, "learning_rate": 3.5858000000000005e-06, "loss": 0.0282, "step": 32072 }, { "epoch": 1.7099216292584103, "grad_norm": 2.135594606399536, "learning_rate": 3.5856e-06, "loss": 0.012, "step": 32073 }, { "epoch": 1.709974942688063, "grad_norm": 1.6404988765716553, "learning_rate": 3.5854000000000004e-06, "loss": 0.0158, "step": 32074 }, { "epoch": 1.710028256117716, "grad_norm": 1.078490972518921, "learning_rate": 3.5852000000000004e-06, "loss": 0.0097, "step": 32075 }, { "epoch": 1.710081569547369, "grad_norm": 1.3559751510620117, "learning_rate": 3.585e-06, "loss": 0.0096, "step": 32076 }, { "epoch": 1.7101348829770218, "grad_norm": 3.3798041343688965, "learning_rate": 3.5848000000000004e-06, "loss": 0.0187, "step": 32077 }, { "epoch": 1.710188196406675, "grad_norm": 0.9542175531387329, "learning_rate": 3.5846000000000004e-06, "loss": 0.0142, "step": 32078 }, { "epoch": 1.7102415098363277, "grad_norm": 1.5427378416061401, "learning_rate": 3.5844e-06, "loss": 0.0109, "step": 32079 }, { "epoch": 1.7102948232659807, "grad_norm": 3.1026558876037598, "learning_rate": 3.5842000000000003e-06, "loss": 0.0232, "step": 32080 }, { "epoch": 1.7103481366956337, "grad_norm": 3.01863694190979, "learning_rate": 3.5840000000000003e-06, "loss": 0.0349, "step": 32081 }, { "epoch": 1.7104014501252864, "grad_norm": 0.959104597568512, "learning_rate": 3.5838e-06, "loss": 0.0219, "step": 32082 }, { "epoch": 1.7104547635549396, "grad_norm": 3.496335744857788, "learning_rate": 3.5836000000000003e-06, "loss": 0.053, "step": 32083 }, { "epoch": 1.7105080769845924, "grad_norm": 1.98921799659729, "learning_rate": 3.5834000000000003e-06, "loss": 0.0462, "step": 32084 }, { "epoch": 1.7105613904142454, "grad_norm": 3.81040620803833, "learning_rate": 3.5832000000000007e-06, "loss": 0.0316, "step": 32085 }, { "epoch": 1.7106147038438984, "grad_norm": 2.515601873397827, "learning_rate": 3.5830000000000003e-06, "loss": 0.0258, "step": 32086 }, { "epoch": 1.7106680172735511, "grad_norm": 3.4671900272369385, "learning_rate": 3.5828000000000002e-06, "loss": 0.0248, "step": 32087 }, { "epoch": 1.710721330703204, "grad_norm": 1.6352471113204956, "learning_rate": 3.5826000000000006e-06, "loss": 0.0222, "step": 32088 }, { "epoch": 1.710774644132857, "grad_norm": 2.8557446002960205, "learning_rate": 3.5824e-06, "loss": 0.0281, "step": 32089 }, { "epoch": 1.71082795756251, "grad_norm": 1.3249061107635498, "learning_rate": 3.5822e-06, "loss": 0.0238, "step": 32090 }, { "epoch": 1.710881270992163, "grad_norm": 2.1696274280548096, "learning_rate": 3.5820000000000006e-06, "loss": 0.0338, "step": 32091 }, { "epoch": 1.7109345844218158, "grad_norm": 2.55568528175354, "learning_rate": 3.5818e-06, "loss": 0.0373, "step": 32092 }, { "epoch": 1.7109878978514688, "grad_norm": 2.240633010864258, "learning_rate": 3.5816e-06, "loss": 0.0133, "step": 32093 }, { "epoch": 1.7110412112811217, "grad_norm": 3.4619245529174805, "learning_rate": 3.5814000000000006e-06, "loss": 0.0104, "step": 32094 }, { "epoch": 1.7110945247107745, "grad_norm": 1.5062239170074463, "learning_rate": 3.5812e-06, "loss": 0.0145, "step": 32095 }, { "epoch": 1.7111478381404277, "grad_norm": 2.110926628112793, "learning_rate": 3.581e-06, "loss": 0.0097, "step": 32096 }, { "epoch": 1.7112011515700805, "grad_norm": 4.675487041473389, "learning_rate": 3.5808000000000005e-06, "loss": 0.0601, "step": 32097 }, { "epoch": 1.7112544649997334, "grad_norm": 2.4202301502227783, "learning_rate": 3.5806e-06, "loss": 0.0333, "step": 32098 }, { "epoch": 1.7113077784293864, "grad_norm": 2.3003156185150146, "learning_rate": 3.5804000000000005e-06, "loss": 0.0277, "step": 32099 }, { "epoch": 1.7113610918590392, "grad_norm": 2.237666368484497, "learning_rate": 3.5802000000000005e-06, "loss": 0.0192, "step": 32100 }, { "epoch": 1.7114144052886924, "grad_norm": 1.704795241355896, "learning_rate": 3.58e-06, "loss": 0.014, "step": 32101 }, { "epoch": 1.7114677187183451, "grad_norm": 1.1599717140197754, "learning_rate": 3.5798000000000004e-06, "loss": 0.0079, "step": 32102 }, { "epoch": 1.7115210321479981, "grad_norm": 0.36105725169181824, "learning_rate": 3.5796000000000004e-06, "loss": 0.0024, "step": 32103 }, { "epoch": 1.711574345577651, "grad_norm": 1.9507802724838257, "learning_rate": 3.5794e-06, "loss": 0.0299, "step": 32104 }, { "epoch": 1.7116276590073038, "grad_norm": 1.4704980850219727, "learning_rate": 3.5792000000000004e-06, "loss": 0.0365, "step": 32105 }, { "epoch": 1.7116809724369568, "grad_norm": 1.8277561664581299, "learning_rate": 3.5790000000000004e-06, "loss": 0.0282, "step": 32106 }, { "epoch": 1.7117342858666098, "grad_norm": 1.5627319812774658, "learning_rate": 3.5788e-06, "loss": 0.0108, "step": 32107 }, { "epoch": 1.7117875992962626, "grad_norm": 0.3403753638267517, "learning_rate": 3.5786000000000003e-06, "loss": 0.0026, "step": 32108 }, { "epoch": 1.7118409127259158, "grad_norm": 1.525570034980774, "learning_rate": 3.5784000000000003e-06, "loss": 0.0238, "step": 32109 }, { "epoch": 1.7118942261555685, "grad_norm": 2.1676788330078125, "learning_rate": 3.5782000000000007e-06, "loss": 0.0322, "step": 32110 }, { "epoch": 1.7119475395852215, "grad_norm": 1.4685430526733398, "learning_rate": 3.5780000000000003e-06, "loss": 0.0275, "step": 32111 }, { "epoch": 1.7120008530148745, "grad_norm": 2.013302803039551, "learning_rate": 3.5778000000000003e-06, "loss": 0.0165, "step": 32112 }, { "epoch": 1.7120541664445272, "grad_norm": 1.630934476852417, "learning_rate": 3.5776000000000007e-06, "loss": 0.0351, "step": 32113 }, { "epoch": 1.7121074798741804, "grad_norm": 3.6650378704071045, "learning_rate": 3.5774000000000003e-06, "loss": 0.0232, "step": 32114 }, { "epoch": 1.7121607933038332, "grad_norm": 1.799185037612915, "learning_rate": 3.5772000000000002e-06, "loss": 0.0189, "step": 32115 }, { "epoch": 1.7122141067334862, "grad_norm": 1.7542872428894043, "learning_rate": 3.5770000000000007e-06, "loss": 0.0163, "step": 32116 }, { "epoch": 1.7122674201631392, "grad_norm": 1.5117582082748413, "learning_rate": 3.5768e-06, "loss": 0.0128, "step": 32117 }, { "epoch": 1.712320733592792, "grad_norm": 2.6684460639953613, "learning_rate": 3.5766e-06, "loss": 0.0295, "step": 32118 }, { "epoch": 1.7123740470224451, "grad_norm": 2.254082441329956, "learning_rate": 3.5764000000000006e-06, "loss": 0.0593, "step": 32119 }, { "epoch": 1.7124273604520979, "grad_norm": 1.8146946430206299, "learning_rate": 3.5762e-06, "loss": 0.0273, "step": 32120 }, { "epoch": 1.7124806738817508, "grad_norm": 0.9819472432136536, "learning_rate": 3.576e-06, "loss": 0.0117, "step": 32121 }, { "epoch": 1.7125339873114038, "grad_norm": 0.9526290893554688, "learning_rate": 3.5758000000000006e-06, "loss": 0.009, "step": 32122 }, { "epoch": 1.7125873007410566, "grad_norm": 3.387700319290161, "learning_rate": 3.5756e-06, "loss": 0.036, "step": 32123 }, { "epoch": 1.7126406141707096, "grad_norm": 0.9676618576049805, "learning_rate": 3.5754000000000005e-06, "loss": 0.0133, "step": 32124 }, { "epoch": 1.7126939276003625, "grad_norm": 1.7936911582946777, "learning_rate": 3.5752000000000005e-06, "loss": 0.0191, "step": 32125 }, { "epoch": 1.7127472410300153, "grad_norm": 0.7471303939819336, "learning_rate": 3.575e-06, "loss": 0.0057, "step": 32126 }, { "epoch": 1.7128005544596685, "grad_norm": 1.6109941005706787, "learning_rate": 3.5748000000000005e-06, "loss": 0.011, "step": 32127 }, { "epoch": 1.7128538678893213, "grad_norm": 2.1221745014190674, "learning_rate": 3.5746000000000005e-06, "loss": 0.0385, "step": 32128 }, { "epoch": 1.7129071813189742, "grad_norm": 1.1863821744918823, "learning_rate": 3.5744e-06, "loss": 0.0087, "step": 32129 }, { "epoch": 1.7129604947486272, "grad_norm": 1.1081421375274658, "learning_rate": 3.5742000000000004e-06, "loss": 0.0187, "step": 32130 }, { "epoch": 1.71301380817828, "grad_norm": 1.7765973806381226, "learning_rate": 3.5740000000000004e-06, "loss": 0.0201, "step": 32131 }, { "epoch": 1.7130671216079332, "grad_norm": 2.650930404663086, "learning_rate": 3.5738e-06, "loss": 0.0473, "step": 32132 }, { "epoch": 1.713120435037586, "grad_norm": 2.3397858142852783, "learning_rate": 3.5736000000000004e-06, "loss": 0.0177, "step": 32133 }, { "epoch": 1.713173748467239, "grad_norm": 2.7302629947662354, "learning_rate": 3.5734000000000004e-06, "loss": 0.0233, "step": 32134 }, { "epoch": 1.713227061896892, "grad_norm": 2.426198720932007, "learning_rate": 3.5732000000000004e-06, "loss": 0.0202, "step": 32135 }, { "epoch": 1.7132803753265446, "grad_norm": 2.435349702835083, "learning_rate": 3.5730000000000003e-06, "loss": 0.0208, "step": 32136 }, { "epoch": 1.7133336887561978, "grad_norm": 1.5194400548934937, "learning_rate": 3.5728000000000003e-06, "loss": 0.028, "step": 32137 }, { "epoch": 1.7133870021858506, "grad_norm": 2.15971302986145, "learning_rate": 3.5726000000000003e-06, "loss": 0.0108, "step": 32138 }, { "epoch": 1.7134403156155036, "grad_norm": 3.808866262435913, "learning_rate": 3.5724000000000003e-06, "loss": 0.0503, "step": 32139 }, { "epoch": 1.7134936290451566, "grad_norm": 1.9871498346328735, "learning_rate": 3.5722000000000003e-06, "loss": 0.0266, "step": 32140 }, { "epoch": 1.7135469424748093, "grad_norm": 2.321969747543335, "learning_rate": 3.5720000000000003e-06, "loss": 0.0369, "step": 32141 }, { "epoch": 1.7136002559044623, "grad_norm": 0.9434393644332886, "learning_rate": 3.5718000000000003e-06, "loss": 0.0199, "step": 32142 }, { "epoch": 1.7136535693341153, "grad_norm": 1.7467950582504272, "learning_rate": 3.5716000000000002e-06, "loss": 0.0249, "step": 32143 }, { "epoch": 1.713706882763768, "grad_norm": 2.1285195350646973, "learning_rate": 3.5714000000000002e-06, "loss": 0.0127, "step": 32144 }, { "epoch": 1.7137601961934212, "grad_norm": 0.4929566979408264, "learning_rate": 3.5712000000000002e-06, "loss": 0.0057, "step": 32145 }, { "epoch": 1.713813509623074, "grad_norm": 1.2018355131149292, "learning_rate": 3.571e-06, "loss": 0.0112, "step": 32146 }, { "epoch": 1.713866823052727, "grad_norm": 1.2602626085281372, "learning_rate": 3.5708e-06, "loss": 0.0355, "step": 32147 }, { "epoch": 1.71392013648238, "grad_norm": 1.4047094583511353, "learning_rate": 3.5706e-06, "loss": 0.034, "step": 32148 }, { "epoch": 1.7139734499120327, "grad_norm": 2.1038334369659424, "learning_rate": 3.5704000000000006e-06, "loss": 0.0329, "step": 32149 }, { "epoch": 1.714026763341686, "grad_norm": 1.125442385673523, "learning_rate": 3.5702e-06, "loss": 0.0084, "step": 32150 }, { "epoch": 1.7140800767713387, "grad_norm": 1.3244324922561646, "learning_rate": 3.57e-06, "loss": 0.0081, "step": 32151 }, { "epoch": 1.7141333902009916, "grad_norm": 1.7527379989624023, "learning_rate": 3.5698000000000005e-06, "loss": 0.0406, "step": 32152 }, { "epoch": 1.7141867036306446, "grad_norm": 1.5714589357376099, "learning_rate": 3.5696e-06, "loss": 0.0225, "step": 32153 }, { "epoch": 1.7142400170602974, "grad_norm": 3.5112380981445312, "learning_rate": 3.5694e-06, "loss": 0.0377, "step": 32154 }, { "epoch": 1.7142933304899506, "grad_norm": 1.3707892894744873, "learning_rate": 3.5692000000000005e-06, "loss": 0.0059, "step": 32155 }, { "epoch": 1.7143466439196033, "grad_norm": 0.6368324160575867, "learning_rate": 3.569e-06, "loss": 0.0056, "step": 32156 }, { "epoch": 1.7143999573492563, "grad_norm": 1.5678943395614624, "learning_rate": 3.5688e-06, "loss": 0.0148, "step": 32157 }, { "epoch": 1.7144532707789093, "grad_norm": 3.28067684173584, "learning_rate": 3.5686000000000004e-06, "loss": 0.0316, "step": 32158 }, { "epoch": 1.714506584208562, "grad_norm": 1.087999701499939, "learning_rate": 3.5684e-06, "loss": 0.006, "step": 32159 }, { "epoch": 1.714559897638215, "grad_norm": 1.5568408966064453, "learning_rate": 3.5682000000000004e-06, "loss": 0.0252, "step": 32160 }, { "epoch": 1.714613211067868, "grad_norm": 1.7645334005355835, "learning_rate": 3.5680000000000004e-06, "loss": 0.0248, "step": 32161 }, { "epoch": 1.7146665244975208, "grad_norm": 1.3188380002975464, "learning_rate": 3.5678e-06, "loss": 0.0214, "step": 32162 }, { "epoch": 1.714719837927174, "grad_norm": 0.7039214968681335, "learning_rate": 3.5676000000000004e-06, "loss": 0.0056, "step": 32163 }, { "epoch": 1.7147731513568267, "grad_norm": 1.947953224182129, "learning_rate": 3.5674000000000004e-06, "loss": 0.0216, "step": 32164 }, { "epoch": 1.7148264647864797, "grad_norm": 1.8431652784347534, "learning_rate": 3.5672e-06, "loss": 0.0458, "step": 32165 }, { "epoch": 1.7148797782161327, "grad_norm": 0.94940584897995, "learning_rate": 3.5670000000000003e-06, "loss": 0.0072, "step": 32166 }, { "epoch": 1.7149330916457854, "grad_norm": 0.7516040205955505, "learning_rate": 3.5668000000000003e-06, "loss": 0.0118, "step": 32167 }, { "epoch": 1.7149864050754386, "grad_norm": 2.269526243209839, "learning_rate": 3.5666e-06, "loss": 0.035, "step": 32168 }, { "epoch": 1.7150397185050914, "grad_norm": 0.4153324365615845, "learning_rate": 3.5664000000000003e-06, "loss": 0.0018, "step": 32169 }, { "epoch": 1.7150930319347444, "grad_norm": 1.210653305053711, "learning_rate": 3.5662000000000003e-06, "loss": 0.0316, "step": 32170 }, { "epoch": 1.7151463453643974, "grad_norm": 1.3845943212509155, "learning_rate": 3.566e-06, "loss": 0.0235, "step": 32171 }, { "epoch": 1.7151996587940501, "grad_norm": 1.3218344449996948, "learning_rate": 3.5658000000000002e-06, "loss": 0.0154, "step": 32172 }, { "epoch": 1.715252972223703, "grad_norm": 1.50263249874115, "learning_rate": 3.5656000000000002e-06, "loss": 0.013, "step": 32173 }, { "epoch": 1.715306285653356, "grad_norm": 1.5011709928512573, "learning_rate": 3.5654000000000006e-06, "loss": 0.0225, "step": 32174 }, { "epoch": 1.715359599083009, "grad_norm": 1.2392096519470215, "learning_rate": 3.5652e-06, "loss": 0.008, "step": 32175 }, { "epoch": 1.715412912512662, "grad_norm": 2.4360291957855225, "learning_rate": 3.565e-06, "loss": 0.0401, "step": 32176 }, { "epoch": 1.7154662259423148, "grad_norm": 1.165501594543457, "learning_rate": 3.5648000000000006e-06, "loss": 0.0102, "step": 32177 }, { "epoch": 1.7155195393719678, "grad_norm": 1.433996558189392, "learning_rate": 3.5646e-06, "loss": 0.007, "step": 32178 }, { "epoch": 1.7155728528016208, "grad_norm": 1.77642822265625, "learning_rate": 3.5644e-06, "loss": 0.0228, "step": 32179 }, { "epoch": 1.7156261662312735, "grad_norm": 2.1132090091705322, "learning_rate": 3.5642000000000005e-06, "loss": 0.0326, "step": 32180 }, { "epoch": 1.7156794796609267, "grad_norm": 1.6476508378982544, "learning_rate": 3.564e-06, "loss": 0.0244, "step": 32181 }, { "epoch": 1.7157327930905795, "grad_norm": 2.517671823501587, "learning_rate": 3.5638e-06, "loss": 0.036, "step": 32182 }, { "epoch": 1.7157861065202324, "grad_norm": 2.1244254112243652, "learning_rate": 3.5636000000000005e-06, "loss": 0.0414, "step": 32183 }, { "epoch": 1.7158394199498854, "grad_norm": 1.0712701082229614, "learning_rate": 3.5634e-06, "loss": 0.0078, "step": 32184 }, { "epoch": 1.7158927333795382, "grad_norm": 2.4316604137420654, "learning_rate": 3.5632000000000005e-06, "loss": 0.0291, "step": 32185 }, { "epoch": 1.7159460468091914, "grad_norm": 1.9849305152893066, "learning_rate": 3.5630000000000004e-06, "loss": 0.0122, "step": 32186 }, { "epoch": 1.7159993602388441, "grad_norm": 1.1557387113571167, "learning_rate": 3.5628e-06, "loss": 0.0102, "step": 32187 }, { "epoch": 1.7160526736684971, "grad_norm": 2.2742910385131836, "learning_rate": 3.5626000000000004e-06, "loss": 0.0347, "step": 32188 }, { "epoch": 1.71610598709815, "grad_norm": 1.194305419921875, "learning_rate": 3.5624000000000004e-06, "loss": 0.0083, "step": 32189 }, { "epoch": 1.7161593005278029, "grad_norm": 1.5990740060806274, "learning_rate": 3.5622e-06, "loss": 0.0234, "step": 32190 }, { "epoch": 1.7162126139574558, "grad_norm": 1.0197104215621948, "learning_rate": 3.5620000000000004e-06, "loss": 0.0172, "step": 32191 }, { "epoch": 1.7162659273871088, "grad_norm": 3.8111584186553955, "learning_rate": 3.5618000000000004e-06, "loss": 0.0326, "step": 32192 }, { "epoch": 1.7163192408167618, "grad_norm": 1.1819735765457153, "learning_rate": 3.5616e-06, "loss": 0.0183, "step": 32193 }, { "epoch": 1.7163725542464148, "grad_norm": 0.836356520652771, "learning_rate": 3.5614000000000003e-06, "loss": 0.007, "step": 32194 }, { "epoch": 1.7164258676760675, "grad_norm": 0.989311158657074, "learning_rate": 3.5612000000000003e-06, "loss": 0.0186, "step": 32195 }, { "epoch": 1.7164791811057205, "grad_norm": 2.0311057567596436, "learning_rate": 3.5610000000000007e-06, "loss": 0.0396, "step": 32196 }, { "epoch": 1.7165324945353735, "grad_norm": 0.9055057168006897, "learning_rate": 3.5608000000000003e-06, "loss": 0.0064, "step": 32197 }, { "epoch": 1.7165858079650262, "grad_norm": 0.8398497700691223, "learning_rate": 3.5606000000000003e-06, "loss": 0.0088, "step": 32198 }, { "epoch": 1.7166391213946794, "grad_norm": 1.9055112600326538, "learning_rate": 3.5604000000000007e-06, "loss": 0.0265, "step": 32199 }, { "epoch": 1.7166924348243322, "grad_norm": 2.5934720039367676, "learning_rate": 3.5602000000000002e-06, "loss": 0.0216, "step": 32200 }, { "epoch": 1.7167457482539852, "grad_norm": 1.2495309114456177, "learning_rate": 3.5600000000000002e-06, "loss": 0.0107, "step": 32201 }, { "epoch": 1.7167990616836382, "grad_norm": 1.5199341773986816, "learning_rate": 3.5598000000000006e-06, "loss": 0.0191, "step": 32202 }, { "epoch": 1.716852375113291, "grad_norm": 1.6521642208099365, "learning_rate": 3.5596e-06, "loss": 0.0223, "step": 32203 }, { "epoch": 1.7169056885429441, "grad_norm": 1.4523588418960571, "learning_rate": 3.5594e-06, "loss": 0.0207, "step": 32204 }, { "epoch": 1.7169590019725969, "grad_norm": 1.3224599361419678, "learning_rate": 3.5592000000000006e-06, "loss": 0.0158, "step": 32205 }, { "epoch": 1.7170123154022499, "grad_norm": 2.1259653568267822, "learning_rate": 3.559e-06, "loss": 0.0242, "step": 32206 }, { "epoch": 1.7170656288319028, "grad_norm": 1.2520016431808472, "learning_rate": 3.5588e-06, "loss": 0.0245, "step": 32207 }, { "epoch": 1.7171189422615556, "grad_norm": 1.2103134393692017, "learning_rate": 3.5586000000000005e-06, "loss": 0.0161, "step": 32208 }, { "epoch": 1.7171722556912086, "grad_norm": 2.040114164352417, "learning_rate": 3.5584e-06, "loss": 0.0178, "step": 32209 }, { "epoch": 1.7172255691208616, "grad_norm": 2.0209145545959473, "learning_rate": 3.5582000000000005e-06, "loss": 0.0313, "step": 32210 }, { "epoch": 1.7172788825505145, "grad_norm": 1.1832900047302246, "learning_rate": 3.5580000000000005e-06, "loss": 0.0152, "step": 32211 }, { "epoch": 1.7173321959801675, "grad_norm": 2.3409535884857178, "learning_rate": 3.5578e-06, "loss": 0.0462, "step": 32212 }, { "epoch": 1.7173855094098203, "grad_norm": 1.8773714303970337, "learning_rate": 3.5576000000000005e-06, "loss": 0.0095, "step": 32213 }, { "epoch": 1.7174388228394732, "grad_norm": 2.3205769062042236, "learning_rate": 3.5574000000000004e-06, "loss": 0.0332, "step": 32214 }, { "epoch": 1.7174921362691262, "grad_norm": 2.0822677612304688, "learning_rate": 3.5572e-06, "loss": 0.0134, "step": 32215 }, { "epoch": 1.717545449698779, "grad_norm": 1.954052209854126, "learning_rate": 3.5570000000000004e-06, "loss": 0.0142, "step": 32216 }, { "epoch": 1.7175987631284322, "grad_norm": 1.8056403398513794, "learning_rate": 3.5568000000000004e-06, "loss": 0.0206, "step": 32217 }, { "epoch": 1.717652076558085, "grad_norm": 1.4008264541625977, "learning_rate": 3.5566e-06, "loss": 0.0084, "step": 32218 }, { "epoch": 1.717705389987738, "grad_norm": 2.652651309967041, "learning_rate": 3.5564000000000004e-06, "loss": 0.0408, "step": 32219 }, { "epoch": 1.717758703417391, "grad_norm": 1.4227837324142456, "learning_rate": 3.5562000000000004e-06, "loss": 0.0164, "step": 32220 }, { "epoch": 1.7178120168470437, "grad_norm": 2.4365479946136475, "learning_rate": 3.5560000000000008e-06, "loss": 0.0167, "step": 32221 }, { "epoch": 1.7178653302766969, "grad_norm": 1.4857709407806396, "learning_rate": 3.5558000000000003e-06, "loss": 0.0152, "step": 32222 }, { "epoch": 1.7179186437063496, "grad_norm": 1.7845906019210815, "learning_rate": 3.5556000000000003e-06, "loss": 0.0191, "step": 32223 }, { "epoch": 1.7179719571360026, "grad_norm": 1.1237796545028687, "learning_rate": 3.5554000000000007e-06, "loss": 0.0122, "step": 32224 }, { "epoch": 1.7180252705656556, "grad_norm": 1.379961371421814, "learning_rate": 3.5552000000000003e-06, "loss": 0.0127, "step": 32225 }, { "epoch": 1.7180785839953083, "grad_norm": 2.018667221069336, "learning_rate": 3.5550000000000003e-06, "loss": 0.0384, "step": 32226 }, { "epoch": 1.7181318974249613, "grad_norm": 1.9330347776412964, "learning_rate": 3.5548000000000007e-06, "loss": 0.0216, "step": 32227 }, { "epoch": 1.7181852108546143, "grad_norm": 1.086106777191162, "learning_rate": 3.5546000000000002e-06, "loss": 0.0164, "step": 32228 }, { "epoch": 1.718238524284267, "grad_norm": 4.548486232757568, "learning_rate": 3.5544000000000002e-06, "loss": 0.0512, "step": 32229 }, { "epoch": 1.7182918377139202, "grad_norm": 2.291008234024048, "learning_rate": 3.5542000000000006e-06, "loss": 0.0513, "step": 32230 }, { "epoch": 1.718345151143573, "grad_norm": 1.4677562713623047, "learning_rate": 3.554e-06, "loss": 0.0131, "step": 32231 }, { "epoch": 1.718398464573226, "grad_norm": 0.8431943655014038, "learning_rate": 3.5538e-06, "loss": 0.0167, "step": 32232 }, { "epoch": 1.718451778002879, "grad_norm": 4.591429233551025, "learning_rate": 3.5536000000000006e-06, "loss": 0.0203, "step": 32233 }, { "epoch": 1.7185050914325317, "grad_norm": 1.5292954444885254, "learning_rate": 3.5534e-06, "loss": 0.009, "step": 32234 }, { "epoch": 1.718558404862185, "grad_norm": 1.5620423555374146, "learning_rate": 3.5532000000000006e-06, "loss": 0.0056, "step": 32235 }, { "epoch": 1.7186117182918377, "grad_norm": 1.475334644317627, "learning_rate": 3.5530000000000005e-06, "loss": 0.0181, "step": 32236 }, { "epoch": 1.7186650317214907, "grad_norm": 2.151641845703125, "learning_rate": 3.5528e-06, "loss": 0.0105, "step": 32237 }, { "epoch": 1.7187183451511436, "grad_norm": 0.637032151222229, "learning_rate": 3.5526000000000005e-06, "loss": 0.0075, "step": 32238 }, { "epoch": 1.7187716585807964, "grad_norm": 1.9930901527404785, "learning_rate": 3.5524000000000005e-06, "loss": 0.044, "step": 32239 }, { "epoch": 1.7188249720104496, "grad_norm": 2.5551085472106934, "learning_rate": 3.5522e-06, "loss": 0.0425, "step": 32240 }, { "epoch": 1.7188782854401023, "grad_norm": 0.9537179470062256, "learning_rate": 3.5520000000000005e-06, "loss": 0.0069, "step": 32241 }, { "epoch": 1.7189315988697553, "grad_norm": 4.135716915130615, "learning_rate": 3.5518000000000005e-06, "loss": 0.0292, "step": 32242 }, { "epoch": 1.7189849122994083, "grad_norm": 1.719437599182129, "learning_rate": 3.5516e-06, "loss": 0.0214, "step": 32243 }, { "epoch": 1.719038225729061, "grad_norm": 1.1064035892486572, "learning_rate": 3.5514000000000004e-06, "loss": 0.0094, "step": 32244 }, { "epoch": 1.719091539158714, "grad_norm": 1.9572654962539673, "learning_rate": 3.5512000000000004e-06, "loss": 0.0213, "step": 32245 }, { "epoch": 1.719144852588367, "grad_norm": 2.1040186882019043, "learning_rate": 3.5510000000000004e-06, "loss": 0.0308, "step": 32246 }, { "epoch": 1.7191981660180198, "grad_norm": 2.0262486934661865, "learning_rate": 3.5508000000000004e-06, "loss": 0.019, "step": 32247 }, { "epoch": 1.719251479447673, "grad_norm": 1.5621055364608765, "learning_rate": 3.5506000000000004e-06, "loss": 0.0273, "step": 32248 }, { "epoch": 1.7193047928773257, "grad_norm": 1.3175408840179443, "learning_rate": 3.5504000000000003e-06, "loss": 0.0129, "step": 32249 }, { "epoch": 1.7193581063069787, "grad_norm": 0.15762412548065186, "learning_rate": 3.5502000000000003e-06, "loss": 0.0017, "step": 32250 }, { "epoch": 1.7194114197366317, "grad_norm": 1.241140604019165, "learning_rate": 3.5500000000000003e-06, "loss": 0.0108, "step": 32251 }, { "epoch": 1.7194647331662845, "grad_norm": 1.3940101861953735, "learning_rate": 3.5498000000000003e-06, "loss": 0.0055, "step": 32252 }, { "epoch": 1.7195180465959377, "grad_norm": 0.9988586902618408, "learning_rate": 3.5496000000000003e-06, "loss": 0.0174, "step": 32253 }, { "epoch": 1.7195713600255904, "grad_norm": 1.6304351091384888, "learning_rate": 3.5494000000000003e-06, "loss": 0.0186, "step": 32254 }, { "epoch": 1.7196246734552434, "grad_norm": 1.4699944257736206, "learning_rate": 3.5492000000000003e-06, "loss": 0.0112, "step": 32255 }, { "epoch": 1.7196779868848964, "grad_norm": 2.892827033996582, "learning_rate": 3.5490000000000002e-06, "loss": 0.0342, "step": 32256 }, { "epoch": 1.7197313003145491, "grad_norm": 0.8423906564712524, "learning_rate": 3.5488000000000002e-06, "loss": 0.0056, "step": 32257 }, { "epoch": 1.7197846137442023, "grad_norm": 2.2987632751464844, "learning_rate": 3.5486e-06, "loss": 0.0233, "step": 32258 }, { "epoch": 1.719837927173855, "grad_norm": 3.8095250129699707, "learning_rate": 3.5484e-06, "loss": 0.0358, "step": 32259 }, { "epoch": 1.719891240603508, "grad_norm": 4.164650917053223, "learning_rate": 3.5482000000000006e-06, "loss": 0.0385, "step": 32260 }, { "epoch": 1.719944554033161, "grad_norm": 1.1006625890731812, "learning_rate": 3.548e-06, "loss": 0.0069, "step": 32261 }, { "epoch": 1.7199978674628138, "grad_norm": 0.8504328727722168, "learning_rate": 3.5478e-06, "loss": 0.0056, "step": 32262 }, { "epoch": 1.7200511808924668, "grad_norm": 0.9186826944351196, "learning_rate": 3.5476000000000006e-06, "loss": 0.0072, "step": 32263 }, { "epoch": 1.7201044943221198, "grad_norm": 1.7760781049728394, "learning_rate": 3.5474e-06, "loss": 0.0156, "step": 32264 }, { "epoch": 1.7201578077517725, "grad_norm": 1.8337279558181763, "learning_rate": 3.5472e-06, "loss": 0.033, "step": 32265 }, { "epoch": 1.7202111211814257, "grad_norm": 1.746921181678772, "learning_rate": 3.5470000000000005e-06, "loss": 0.018, "step": 32266 }, { "epoch": 1.7202644346110785, "grad_norm": 0.8388088345527649, "learning_rate": 3.5468e-06, "loss": 0.0059, "step": 32267 }, { "epoch": 1.7203177480407315, "grad_norm": 1.5361874103546143, "learning_rate": 3.5466e-06, "loss": 0.0122, "step": 32268 }, { "epoch": 1.7203710614703844, "grad_norm": 0.8999568819999695, "learning_rate": 3.5464000000000005e-06, "loss": 0.0088, "step": 32269 }, { "epoch": 1.7204243749000372, "grad_norm": 1.7778892517089844, "learning_rate": 3.5462e-06, "loss": 0.03, "step": 32270 }, { "epoch": 1.7204776883296904, "grad_norm": 2.0238139629364014, "learning_rate": 3.5460000000000004e-06, "loss": 0.0411, "step": 32271 }, { "epoch": 1.7205310017593431, "grad_norm": 0.9782993197441101, "learning_rate": 3.5458000000000004e-06, "loss": 0.0191, "step": 32272 }, { "epoch": 1.7205843151889961, "grad_norm": 1.027483582496643, "learning_rate": 3.5456e-06, "loss": 0.0056, "step": 32273 }, { "epoch": 1.720637628618649, "grad_norm": 1.382020354270935, "learning_rate": 3.5454000000000004e-06, "loss": 0.0125, "step": 32274 }, { "epoch": 1.7206909420483019, "grad_norm": 2.7368855476379395, "learning_rate": 3.5452000000000004e-06, "loss": 0.0407, "step": 32275 }, { "epoch": 1.7207442554779548, "grad_norm": 0.6484876871109009, "learning_rate": 3.545e-06, "loss": 0.0049, "step": 32276 }, { "epoch": 1.7207975689076078, "grad_norm": 4.901939392089844, "learning_rate": 3.5448000000000003e-06, "loss": 0.059, "step": 32277 }, { "epoch": 1.7208508823372608, "grad_norm": 1.3837748765945435, "learning_rate": 3.5446000000000003e-06, "loss": 0.0159, "step": 32278 }, { "epoch": 1.7209041957669138, "grad_norm": 1.8619678020477295, "learning_rate": 3.5444e-06, "loss": 0.0149, "step": 32279 }, { "epoch": 1.7209575091965665, "grad_norm": 1.782714605331421, "learning_rate": 3.5442000000000003e-06, "loss": 0.0172, "step": 32280 }, { "epoch": 1.7210108226262195, "grad_norm": 2.463261365890503, "learning_rate": 3.5440000000000003e-06, "loss": 0.0325, "step": 32281 }, { "epoch": 1.7210641360558725, "grad_norm": 3.684819221496582, "learning_rate": 3.5438e-06, "loss": 0.0243, "step": 32282 }, { "epoch": 1.7211174494855253, "grad_norm": 2.1817221641540527, "learning_rate": 3.5436000000000003e-06, "loss": 0.0449, "step": 32283 }, { "epoch": 1.7211707629151785, "grad_norm": 1.1406526565551758, "learning_rate": 3.5434000000000002e-06, "loss": 0.0068, "step": 32284 }, { "epoch": 1.7212240763448312, "grad_norm": 1.9258191585540771, "learning_rate": 3.5432000000000007e-06, "loss": 0.0254, "step": 32285 }, { "epoch": 1.7212773897744842, "grad_norm": 1.5548495054244995, "learning_rate": 3.5430000000000002e-06, "loss": 0.005, "step": 32286 }, { "epoch": 1.7213307032041372, "grad_norm": 1.7333793640136719, "learning_rate": 3.5428e-06, "loss": 0.01, "step": 32287 }, { "epoch": 1.72138401663379, "grad_norm": 1.1451126337051392, "learning_rate": 3.5426000000000006e-06, "loss": 0.0162, "step": 32288 }, { "epoch": 1.7214373300634431, "grad_norm": 2.353666067123413, "learning_rate": 3.5424e-06, "loss": 0.0151, "step": 32289 }, { "epoch": 1.7214906434930959, "grad_norm": 3.7923765182495117, "learning_rate": 3.5422e-06, "loss": 0.0588, "step": 32290 }, { "epoch": 1.7215439569227489, "grad_norm": 2.2135543823242188, "learning_rate": 3.5420000000000006e-06, "loss": 0.0258, "step": 32291 }, { "epoch": 1.7215972703524018, "grad_norm": 1.4054120779037476, "learning_rate": 3.5418e-06, "loss": 0.0123, "step": 32292 }, { "epoch": 1.7216505837820546, "grad_norm": 0.8688257336616516, "learning_rate": 3.5416e-06, "loss": 0.014, "step": 32293 }, { "epoch": 1.7217038972117076, "grad_norm": 2.2525532245635986, "learning_rate": 3.5414000000000005e-06, "loss": 0.0279, "step": 32294 }, { "epoch": 1.7217572106413606, "grad_norm": 2.1381614208221436, "learning_rate": 3.5412e-06, "loss": 0.0403, "step": 32295 }, { "epoch": 1.7218105240710135, "grad_norm": 1.5035043954849243, "learning_rate": 3.5410000000000005e-06, "loss": 0.0186, "step": 32296 }, { "epoch": 1.7218638375006665, "grad_norm": 1.9588061571121216, "learning_rate": 3.5408000000000005e-06, "loss": 0.0456, "step": 32297 }, { "epoch": 1.7219171509303193, "grad_norm": 1.8017935752868652, "learning_rate": 3.5406e-06, "loss": 0.0217, "step": 32298 }, { "epoch": 1.7219704643599723, "grad_norm": 1.7645748853683472, "learning_rate": 3.5404000000000004e-06, "loss": 0.0191, "step": 32299 }, { "epoch": 1.7220237777896252, "grad_norm": 1.4998501539230347, "learning_rate": 3.5402000000000004e-06, "loss": 0.016, "step": 32300 }, { "epoch": 1.722077091219278, "grad_norm": 0.8571667075157166, "learning_rate": 3.54e-06, "loss": 0.007, "step": 32301 }, { "epoch": 1.7221304046489312, "grad_norm": 1.8575780391693115, "learning_rate": 3.5398000000000004e-06, "loss": 0.0084, "step": 32302 }, { "epoch": 1.722183718078584, "grad_norm": 1.4153200387954712, "learning_rate": 3.5396000000000004e-06, "loss": 0.0121, "step": 32303 }, { "epoch": 1.722237031508237, "grad_norm": 0.3204807937145233, "learning_rate": 3.5394e-06, "loss": 0.003, "step": 32304 }, { "epoch": 1.72229034493789, "grad_norm": 0.5803675651550293, "learning_rate": 3.5392000000000004e-06, "loss": 0.0037, "step": 32305 }, { "epoch": 1.7223436583675427, "grad_norm": 0.9351170063018799, "learning_rate": 3.5390000000000003e-06, "loss": 0.012, "step": 32306 }, { "epoch": 1.7223969717971959, "grad_norm": 1.4720498323440552, "learning_rate": 3.5388e-06, "loss": 0.0199, "step": 32307 }, { "epoch": 1.7224502852268486, "grad_norm": 3.598680019378662, "learning_rate": 3.5386000000000003e-06, "loss": 0.0502, "step": 32308 }, { "epoch": 1.7225035986565016, "grad_norm": 1.4491559267044067, "learning_rate": 3.5384000000000003e-06, "loss": 0.0143, "step": 32309 }, { "epoch": 1.7225569120861546, "grad_norm": 1.5827008485794067, "learning_rate": 3.5382000000000007e-06, "loss": 0.0193, "step": 32310 }, { "epoch": 1.7226102255158073, "grad_norm": 0.8629134893417358, "learning_rate": 3.5380000000000003e-06, "loss": 0.0145, "step": 32311 }, { "epoch": 1.7226635389454603, "grad_norm": 0.3758174479007721, "learning_rate": 3.5378000000000002e-06, "loss": 0.004, "step": 32312 }, { "epoch": 1.7227168523751133, "grad_norm": 1.1192471981048584, "learning_rate": 3.5376000000000007e-06, "loss": 0.02, "step": 32313 }, { "epoch": 1.7227701658047663, "grad_norm": 1.5209062099456787, "learning_rate": 3.5374000000000002e-06, "loss": 0.0075, "step": 32314 }, { "epoch": 1.7228234792344193, "grad_norm": 4.211877346038818, "learning_rate": 3.5372e-06, "loss": 0.0227, "step": 32315 }, { "epoch": 1.722876792664072, "grad_norm": 0.8089483976364136, "learning_rate": 3.5370000000000006e-06, "loss": 0.0074, "step": 32316 }, { "epoch": 1.722930106093725, "grad_norm": 2.3155441284179688, "learning_rate": 3.5368e-06, "loss": 0.0418, "step": 32317 }, { "epoch": 1.722983419523378, "grad_norm": 2.6293444633483887, "learning_rate": 3.5366e-06, "loss": 0.0217, "step": 32318 }, { "epoch": 1.7230367329530307, "grad_norm": 1.3587039709091187, "learning_rate": 3.5364000000000006e-06, "loss": 0.0271, "step": 32319 }, { "epoch": 1.723090046382684, "grad_norm": 1.1131181716918945, "learning_rate": 3.5362e-06, "loss": 0.0097, "step": 32320 }, { "epoch": 1.7231433598123367, "grad_norm": 1.4246386289596558, "learning_rate": 3.5360000000000005e-06, "loss": 0.0207, "step": 32321 }, { "epoch": 1.7231966732419897, "grad_norm": 1.3507046699523926, "learning_rate": 3.5358000000000005e-06, "loss": 0.0235, "step": 32322 }, { "epoch": 1.7232499866716426, "grad_norm": 3.114264488220215, "learning_rate": 3.5356e-06, "loss": 0.0382, "step": 32323 }, { "epoch": 1.7233033001012954, "grad_norm": 1.5535671710968018, "learning_rate": 3.5354000000000005e-06, "loss": 0.0216, "step": 32324 }, { "epoch": 1.7233566135309486, "grad_norm": 1.878494143486023, "learning_rate": 3.5352000000000005e-06, "loss": 0.0086, "step": 32325 }, { "epoch": 1.7234099269606014, "grad_norm": 2.034271717071533, "learning_rate": 3.535e-06, "loss": 0.0272, "step": 32326 }, { "epoch": 1.7234632403902543, "grad_norm": 2.0391790866851807, "learning_rate": 3.5348000000000004e-06, "loss": 0.0234, "step": 32327 }, { "epoch": 1.7235165538199073, "grad_norm": 1.8727548122406006, "learning_rate": 3.5346000000000004e-06, "loss": 0.0094, "step": 32328 }, { "epoch": 1.72356986724956, "grad_norm": 1.698839783668518, "learning_rate": 3.5344e-06, "loss": 0.0271, "step": 32329 }, { "epoch": 1.723623180679213, "grad_norm": 1.363963007926941, "learning_rate": 3.5342000000000004e-06, "loss": 0.0115, "step": 32330 }, { "epoch": 1.723676494108866, "grad_norm": 1.9192723035812378, "learning_rate": 3.5340000000000004e-06, "loss": 0.0191, "step": 32331 }, { "epoch": 1.7237298075385188, "grad_norm": 1.251396656036377, "learning_rate": 3.5338000000000004e-06, "loss": 0.0218, "step": 32332 }, { "epoch": 1.723783120968172, "grad_norm": 1.3578897714614868, "learning_rate": 3.5336000000000004e-06, "loss": 0.0176, "step": 32333 }, { "epoch": 1.7238364343978247, "grad_norm": 1.991421103477478, "learning_rate": 3.5334000000000003e-06, "loss": 0.0107, "step": 32334 }, { "epoch": 1.7238897478274777, "grad_norm": 0.7418562769889832, "learning_rate": 3.5332000000000003e-06, "loss": 0.005, "step": 32335 }, { "epoch": 1.7239430612571307, "grad_norm": 1.8480628728866577, "learning_rate": 3.5330000000000003e-06, "loss": 0.0089, "step": 32336 }, { "epoch": 1.7239963746867835, "grad_norm": 2.089045524597168, "learning_rate": 3.5328000000000003e-06, "loss": 0.0254, "step": 32337 }, { "epoch": 1.7240496881164367, "grad_norm": 1.6444170475006104, "learning_rate": 3.5326000000000003e-06, "loss": 0.0234, "step": 32338 }, { "epoch": 1.7241030015460894, "grad_norm": 1.5629392862319946, "learning_rate": 3.5324000000000003e-06, "loss": 0.0225, "step": 32339 }, { "epoch": 1.7241563149757424, "grad_norm": 5.019041538238525, "learning_rate": 3.5322000000000002e-06, "loss": 0.0741, "step": 32340 }, { "epoch": 1.7242096284053954, "grad_norm": 2.1399941444396973, "learning_rate": 3.5320000000000002e-06, "loss": 0.0188, "step": 32341 }, { "epoch": 1.7242629418350481, "grad_norm": 0.8518815040588379, "learning_rate": 3.5318000000000002e-06, "loss": 0.0063, "step": 32342 }, { "epoch": 1.7243162552647013, "grad_norm": 1.572648525238037, "learning_rate": 3.5316e-06, "loss": 0.0186, "step": 32343 }, { "epoch": 1.724369568694354, "grad_norm": 2.581711530685425, "learning_rate": 3.5314e-06, "loss": 0.0229, "step": 32344 }, { "epoch": 1.724422882124007, "grad_norm": 3.1987497806549072, "learning_rate": 3.5312e-06, "loss": 0.0212, "step": 32345 }, { "epoch": 1.72447619555366, "grad_norm": 5.4526753425598145, "learning_rate": 3.5310000000000006e-06, "loss": 0.0615, "step": 32346 }, { "epoch": 1.7245295089833128, "grad_norm": 1.5039421319961548, "learning_rate": 3.5308e-06, "loss": 0.0181, "step": 32347 }, { "epoch": 1.7245828224129658, "grad_norm": 1.971954345703125, "learning_rate": 3.5306e-06, "loss": 0.0141, "step": 32348 }, { "epoch": 1.7246361358426188, "grad_norm": 2.937406539916992, "learning_rate": 3.5304000000000005e-06, "loss": 0.0322, "step": 32349 }, { "epoch": 1.7246894492722715, "grad_norm": 1.7770798206329346, "learning_rate": 3.5302e-06, "loss": 0.0268, "step": 32350 }, { "epoch": 1.7247427627019247, "grad_norm": 1.5777475833892822, "learning_rate": 3.53e-06, "loss": 0.0313, "step": 32351 }, { "epoch": 1.7247960761315775, "grad_norm": 0.742053747177124, "learning_rate": 3.5298000000000005e-06, "loss": 0.0063, "step": 32352 }, { "epoch": 1.7248493895612305, "grad_norm": 1.9233472347259521, "learning_rate": 3.5296e-06, "loss": 0.0177, "step": 32353 }, { "epoch": 1.7249027029908834, "grad_norm": 2.6969869136810303, "learning_rate": 3.5294e-06, "loss": 0.0399, "step": 32354 }, { "epoch": 1.7249560164205362, "grad_norm": 2.2943010330200195, "learning_rate": 3.5292000000000004e-06, "loss": 0.0109, "step": 32355 }, { "epoch": 1.7250093298501894, "grad_norm": 4.229480266571045, "learning_rate": 3.529e-06, "loss": 0.021, "step": 32356 }, { "epoch": 1.7250626432798422, "grad_norm": 5.220544815063477, "learning_rate": 3.5288000000000004e-06, "loss": 0.0232, "step": 32357 }, { "epoch": 1.7251159567094951, "grad_norm": 2.136760950088501, "learning_rate": 3.5286000000000004e-06, "loss": 0.0149, "step": 32358 }, { "epoch": 1.7251692701391481, "grad_norm": 1.0055984258651733, "learning_rate": 3.5284e-06, "loss": 0.0071, "step": 32359 }, { "epoch": 1.7252225835688009, "grad_norm": 1.6656858921051025, "learning_rate": 3.5282000000000004e-06, "loss": 0.0067, "step": 32360 }, { "epoch": 1.725275896998454, "grad_norm": 2.2013649940490723, "learning_rate": 3.5280000000000004e-06, "loss": 0.0404, "step": 32361 }, { "epoch": 1.7253292104281068, "grad_norm": 1.7197593450546265, "learning_rate": 3.5278e-06, "loss": 0.031, "step": 32362 }, { "epoch": 1.7253825238577598, "grad_norm": 1.1765393018722534, "learning_rate": 3.5276000000000003e-06, "loss": 0.0179, "step": 32363 }, { "epoch": 1.7254358372874128, "grad_norm": 1.6292953491210938, "learning_rate": 3.5274000000000003e-06, "loss": 0.0338, "step": 32364 }, { "epoch": 1.7254891507170655, "grad_norm": 1.8044180870056152, "learning_rate": 3.5272e-06, "loss": 0.0091, "step": 32365 }, { "epoch": 1.7255424641467185, "grad_norm": 1.1153713464736938, "learning_rate": 3.5270000000000003e-06, "loss": 0.0272, "step": 32366 }, { "epoch": 1.7255957775763715, "grad_norm": 1.8210093975067139, "learning_rate": 3.5268000000000003e-06, "loss": 0.0084, "step": 32367 }, { "epoch": 1.7256490910060243, "grad_norm": 1.7515873908996582, "learning_rate": 3.5266e-06, "loss": 0.0331, "step": 32368 }, { "epoch": 1.7257024044356775, "grad_norm": 2.8552980422973633, "learning_rate": 3.5264000000000002e-06, "loss": 0.0701, "step": 32369 }, { "epoch": 1.7257557178653302, "grad_norm": 1.1617194414138794, "learning_rate": 3.5262000000000002e-06, "loss": 0.0141, "step": 32370 }, { "epoch": 1.7258090312949832, "grad_norm": 0.9751423001289368, "learning_rate": 3.5260000000000006e-06, "loss": 0.0054, "step": 32371 }, { "epoch": 1.7258623447246362, "grad_norm": 1.4498010873794556, "learning_rate": 3.5258e-06, "loss": 0.0131, "step": 32372 }, { "epoch": 1.725915658154289, "grad_norm": 0.7511988282203674, "learning_rate": 3.5256e-06, "loss": 0.0047, "step": 32373 }, { "epoch": 1.7259689715839421, "grad_norm": 2.1178364753723145, "learning_rate": 3.5254000000000006e-06, "loss": 0.0237, "step": 32374 }, { "epoch": 1.726022285013595, "grad_norm": 1.5424952507019043, "learning_rate": 3.5252e-06, "loss": 0.0241, "step": 32375 }, { "epoch": 1.7260755984432479, "grad_norm": 2.2960257530212402, "learning_rate": 3.525e-06, "loss": 0.0358, "step": 32376 }, { "epoch": 1.7261289118729009, "grad_norm": 1.036887764930725, "learning_rate": 3.5248000000000005e-06, "loss": 0.026, "step": 32377 }, { "epoch": 1.7261822253025536, "grad_norm": 1.4660818576812744, "learning_rate": 3.5246e-06, "loss": 0.0101, "step": 32378 }, { "epoch": 1.7262355387322068, "grad_norm": 1.5574402809143066, "learning_rate": 3.5244e-06, "loss": 0.0212, "step": 32379 }, { "epoch": 1.7262888521618596, "grad_norm": 2.6631128787994385, "learning_rate": 3.5242000000000005e-06, "loss": 0.0155, "step": 32380 }, { "epoch": 1.7263421655915125, "grad_norm": 1.2702765464782715, "learning_rate": 3.524e-06, "loss": 0.0095, "step": 32381 }, { "epoch": 1.7263954790211655, "grad_norm": 1.3594608306884766, "learning_rate": 3.5238000000000005e-06, "loss": 0.0251, "step": 32382 }, { "epoch": 1.7264487924508183, "grad_norm": 2.874704360961914, "learning_rate": 3.5236000000000004e-06, "loss": 0.0266, "step": 32383 }, { "epoch": 1.7265021058804713, "grad_norm": 0.9406580328941345, "learning_rate": 3.5234e-06, "loss": 0.0068, "step": 32384 }, { "epoch": 1.7265554193101242, "grad_norm": 0.9153061509132385, "learning_rate": 3.5232000000000004e-06, "loss": 0.0074, "step": 32385 }, { "epoch": 1.726608732739777, "grad_norm": 2.009958267211914, "learning_rate": 3.5230000000000004e-06, "loss": 0.0202, "step": 32386 }, { "epoch": 1.7266620461694302, "grad_norm": 2.286167860031128, "learning_rate": 3.5228e-06, "loss": 0.0246, "step": 32387 }, { "epoch": 1.726715359599083, "grad_norm": 0.4845812916755676, "learning_rate": 3.5226000000000004e-06, "loss": 0.0046, "step": 32388 }, { "epoch": 1.726768673028736, "grad_norm": 1.7280362844467163, "learning_rate": 3.5224000000000004e-06, "loss": 0.0139, "step": 32389 }, { "epoch": 1.726821986458389, "grad_norm": 1.9574981927871704, "learning_rate": 3.5222e-06, "loss": 0.0224, "step": 32390 }, { "epoch": 1.7268752998880417, "grad_norm": 1.8546032905578613, "learning_rate": 3.5220000000000003e-06, "loss": 0.0085, "step": 32391 }, { "epoch": 1.7269286133176949, "grad_norm": 2.2743494510650635, "learning_rate": 3.5218000000000003e-06, "loss": 0.0279, "step": 32392 }, { "epoch": 1.7269819267473476, "grad_norm": 2.2471697330474854, "learning_rate": 3.5216e-06, "loss": 0.0336, "step": 32393 }, { "epoch": 1.7270352401770006, "grad_norm": 1.8609908819198608, "learning_rate": 3.5214000000000003e-06, "loss": 0.0291, "step": 32394 }, { "epoch": 1.7270885536066536, "grad_norm": 1.7582740783691406, "learning_rate": 3.5212000000000003e-06, "loss": 0.014, "step": 32395 }, { "epoch": 1.7271418670363063, "grad_norm": 1.037002682685852, "learning_rate": 3.5210000000000007e-06, "loss": 0.0224, "step": 32396 }, { "epoch": 1.7271951804659593, "grad_norm": 1.183056116104126, "learning_rate": 3.5208000000000002e-06, "loss": 0.009, "step": 32397 }, { "epoch": 1.7272484938956123, "grad_norm": 1.1095303297042847, "learning_rate": 3.5206000000000002e-06, "loss": 0.0042, "step": 32398 }, { "epoch": 1.7273018073252653, "grad_norm": 0.5006802082061768, "learning_rate": 3.5204000000000006e-06, "loss": 0.0041, "step": 32399 }, { "epoch": 1.7273551207549183, "grad_norm": 3.701979875564575, "learning_rate": 3.5202e-06, "loss": 0.0239, "step": 32400 }, { "epoch": 1.727408434184571, "grad_norm": 0.4982808530330658, "learning_rate": 3.52e-06, "loss": 0.004, "step": 32401 }, { "epoch": 1.727461747614224, "grad_norm": 0.8784568309783936, "learning_rate": 3.5198000000000006e-06, "loss": 0.0099, "step": 32402 }, { "epoch": 1.727515061043877, "grad_norm": 1.0765894651412964, "learning_rate": 3.5196e-06, "loss": 0.0092, "step": 32403 }, { "epoch": 1.7275683744735297, "grad_norm": 1.6933987140655518, "learning_rate": 3.5194e-06, "loss": 0.0218, "step": 32404 }, { "epoch": 1.727621687903183, "grad_norm": 0.9249609708786011, "learning_rate": 3.5192000000000005e-06, "loss": 0.0125, "step": 32405 }, { "epoch": 1.7276750013328357, "grad_norm": 1.7789595127105713, "learning_rate": 3.519e-06, "loss": 0.0326, "step": 32406 }, { "epoch": 1.7277283147624887, "grad_norm": 2.062268018722534, "learning_rate": 3.5188000000000005e-06, "loss": 0.0242, "step": 32407 }, { "epoch": 1.7277816281921416, "grad_norm": 2.962883949279785, "learning_rate": 3.5186000000000005e-06, "loss": 0.031, "step": 32408 }, { "epoch": 1.7278349416217944, "grad_norm": 2.325744867324829, "learning_rate": 3.5184e-06, "loss": 0.0287, "step": 32409 }, { "epoch": 1.7278882550514476, "grad_norm": 0.9272204041481018, "learning_rate": 3.5182000000000005e-06, "loss": 0.0056, "step": 32410 }, { "epoch": 1.7279415684811004, "grad_norm": 0.7693414092063904, "learning_rate": 3.5180000000000005e-06, "loss": 0.0053, "step": 32411 }, { "epoch": 1.7279948819107533, "grad_norm": 0.5914518237113953, "learning_rate": 3.5178e-06, "loss": 0.0056, "step": 32412 }, { "epoch": 1.7280481953404063, "grad_norm": 1.955382227897644, "learning_rate": 3.5176000000000004e-06, "loss": 0.0358, "step": 32413 }, { "epoch": 1.728101508770059, "grad_norm": 1.6796621084213257, "learning_rate": 3.5174000000000004e-06, "loss": 0.0094, "step": 32414 }, { "epoch": 1.728154822199712, "grad_norm": 2.2922258377075195, "learning_rate": 3.5172e-06, "loss": 0.0106, "step": 32415 }, { "epoch": 1.728208135629365, "grad_norm": 1.8022491931915283, "learning_rate": 3.5170000000000004e-06, "loss": 0.0261, "step": 32416 }, { "epoch": 1.728261449059018, "grad_norm": 0.275870144367218, "learning_rate": 3.5168000000000004e-06, "loss": 0.0024, "step": 32417 }, { "epoch": 1.728314762488671, "grad_norm": 2.4396629333496094, "learning_rate": 3.5166e-06, "loss": 0.0142, "step": 32418 }, { "epoch": 1.7283680759183238, "grad_norm": 2.5138297080993652, "learning_rate": 3.5164000000000003e-06, "loss": 0.0463, "step": 32419 }, { "epoch": 1.7284213893479767, "grad_norm": 1.1217790842056274, "learning_rate": 3.5162000000000003e-06, "loss": 0.0061, "step": 32420 }, { "epoch": 1.7284747027776297, "grad_norm": 2.1618824005126953, "learning_rate": 3.5160000000000007e-06, "loss": 0.0212, "step": 32421 }, { "epoch": 1.7285280162072825, "grad_norm": 1.4268286228179932, "learning_rate": 3.5158000000000003e-06, "loss": 0.0107, "step": 32422 }, { "epoch": 1.7285813296369357, "grad_norm": 1.7668737173080444, "learning_rate": 3.5156000000000003e-06, "loss": 0.0297, "step": 32423 }, { "epoch": 1.7286346430665884, "grad_norm": 1.663688063621521, "learning_rate": 3.5154000000000007e-06, "loss": 0.0175, "step": 32424 }, { "epoch": 1.7286879564962414, "grad_norm": 0.8233082294464111, "learning_rate": 3.5152000000000002e-06, "loss": 0.0058, "step": 32425 }, { "epoch": 1.7287412699258944, "grad_norm": 1.6785409450531006, "learning_rate": 3.5150000000000002e-06, "loss": 0.0124, "step": 32426 }, { "epoch": 1.7287945833555471, "grad_norm": 2.0359702110290527, "learning_rate": 3.5148000000000006e-06, "loss": 0.0282, "step": 32427 }, { "epoch": 1.7288478967852003, "grad_norm": 2.94357967376709, "learning_rate": 3.5146e-06, "loss": 0.0155, "step": 32428 }, { "epoch": 1.728901210214853, "grad_norm": 1.8187800645828247, "learning_rate": 3.5144e-06, "loss": 0.0136, "step": 32429 }, { "epoch": 1.728954523644506, "grad_norm": 1.559631109237671, "learning_rate": 3.5142000000000006e-06, "loss": 0.0187, "step": 32430 }, { "epoch": 1.729007837074159, "grad_norm": 3.0496792793273926, "learning_rate": 3.514e-06, "loss": 0.0218, "step": 32431 }, { "epoch": 1.7290611505038118, "grad_norm": 1.1052141189575195, "learning_rate": 3.5138000000000006e-06, "loss": 0.0053, "step": 32432 }, { "epoch": 1.7291144639334648, "grad_norm": 1.7319624423980713, "learning_rate": 3.5136000000000005e-06, "loss": 0.0266, "step": 32433 }, { "epoch": 1.7291677773631178, "grad_norm": 1.6926078796386719, "learning_rate": 3.5134e-06, "loss": 0.023, "step": 32434 }, { "epoch": 1.7292210907927708, "grad_norm": 1.5079748630523682, "learning_rate": 3.5132000000000005e-06, "loss": 0.0174, "step": 32435 }, { "epoch": 1.7292744042224237, "grad_norm": 1.8506286144256592, "learning_rate": 3.5130000000000005e-06, "loss": 0.0389, "step": 32436 }, { "epoch": 1.7293277176520765, "grad_norm": 2.719628095626831, "learning_rate": 3.5128e-06, "loss": 0.0334, "step": 32437 }, { "epoch": 1.7293810310817295, "grad_norm": 1.8280959129333496, "learning_rate": 3.5126000000000005e-06, "loss": 0.0241, "step": 32438 }, { "epoch": 1.7294343445113824, "grad_norm": 2.89007568359375, "learning_rate": 3.5124000000000005e-06, "loss": 0.0141, "step": 32439 }, { "epoch": 1.7294876579410352, "grad_norm": 3.1031463146209717, "learning_rate": 3.5122e-06, "loss": 0.0359, "step": 32440 }, { "epoch": 1.7295409713706884, "grad_norm": 1.812389850616455, "learning_rate": 3.5120000000000004e-06, "loss": 0.0463, "step": 32441 }, { "epoch": 1.7295942848003412, "grad_norm": 2.3413283824920654, "learning_rate": 3.5118000000000004e-06, "loss": 0.0181, "step": 32442 }, { "epoch": 1.7296475982299941, "grad_norm": 2.1405718326568604, "learning_rate": 3.5116000000000004e-06, "loss": 0.0211, "step": 32443 }, { "epoch": 1.7297009116596471, "grad_norm": 1.8729339838027954, "learning_rate": 3.5114000000000004e-06, "loss": 0.0148, "step": 32444 }, { "epoch": 1.7297542250892999, "grad_norm": 1.9681440591812134, "learning_rate": 3.5112000000000004e-06, "loss": 0.0171, "step": 32445 }, { "epoch": 1.729807538518953, "grad_norm": 1.6801015138626099, "learning_rate": 3.5110000000000004e-06, "loss": 0.0059, "step": 32446 }, { "epoch": 1.7298608519486058, "grad_norm": 1.0461403131484985, "learning_rate": 3.5108000000000003e-06, "loss": 0.0078, "step": 32447 }, { "epoch": 1.7299141653782588, "grad_norm": 2.1813294887542725, "learning_rate": 3.5106000000000003e-06, "loss": 0.0251, "step": 32448 }, { "epoch": 1.7299674788079118, "grad_norm": 2.3773064613342285, "learning_rate": 3.5104000000000003e-06, "loss": 0.0489, "step": 32449 }, { "epoch": 1.7300207922375646, "grad_norm": 3.7100212574005127, "learning_rate": 3.5102000000000003e-06, "loss": 0.0445, "step": 32450 }, { "epoch": 1.7300741056672175, "grad_norm": 1.2593297958374023, "learning_rate": 3.5100000000000003e-06, "loss": 0.0192, "step": 32451 }, { "epoch": 1.7301274190968705, "grad_norm": 5.7328715324401855, "learning_rate": 3.5098000000000003e-06, "loss": 0.0434, "step": 32452 }, { "epoch": 1.7301807325265233, "grad_norm": 1.1053876876831055, "learning_rate": 3.5096000000000002e-06, "loss": 0.0135, "step": 32453 }, { "epoch": 1.7302340459561765, "grad_norm": 3.8040127754211426, "learning_rate": 3.5094000000000002e-06, "loss": 0.0455, "step": 32454 }, { "epoch": 1.7302873593858292, "grad_norm": 2.2212367057800293, "learning_rate": 3.5092000000000002e-06, "loss": 0.0189, "step": 32455 }, { "epoch": 1.7303406728154822, "grad_norm": 1.8533170223236084, "learning_rate": 3.509e-06, "loss": 0.02, "step": 32456 }, { "epoch": 1.7303939862451352, "grad_norm": 4.101535320281982, "learning_rate": 3.5088000000000006e-06, "loss": 0.0186, "step": 32457 }, { "epoch": 1.730447299674788, "grad_norm": 0.9190285801887512, "learning_rate": 3.5086e-06, "loss": 0.007, "step": 32458 }, { "epoch": 1.7305006131044411, "grad_norm": 2.2048919200897217, "learning_rate": 3.5084e-06, "loss": 0.0093, "step": 32459 }, { "epoch": 1.730553926534094, "grad_norm": 2.419574022293091, "learning_rate": 3.5082000000000006e-06, "loss": 0.0199, "step": 32460 }, { "epoch": 1.7306072399637469, "grad_norm": 1.9720646142959595, "learning_rate": 3.508e-06, "loss": 0.0291, "step": 32461 }, { "epoch": 1.7306605533933999, "grad_norm": 2.2590110301971436, "learning_rate": 3.5078e-06, "loss": 0.0445, "step": 32462 }, { "epoch": 1.7307138668230526, "grad_norm": 1.8123546838760376, "learning_rate": 3.5076000000000005e-06, "loss": 0.0343, "step": 32463 }, { "epoch": 1.7307671802527058, "grad_norm": 2.2849884033203125, "learning_rate": 3.5074e-06, "loss": 0.0277, "step": 32464 }, { "epoch": 1.7308204936823586, "grad_norm": 1.6742273569107056, "learning_rate": 3.5072e-06, "loss": 0.0303, "step": 32465 }, { "epoch": 1.7308738071120116, "grad_norm": 1.76168954372406, "learning_rate": 3.5070000000000005e-06, "loss": 0.0271, "step": 32466 }, { "epoch": 1.7309271205416645, "grad_norm": 4.079547882080078, "learning_rate": 3.5068e-06, "loss": 0.0388, "step": 32467 }, { "epoch": 1.7309804339713173, "grad_norm": 2.308755874633789, "learning_rate": 3.5066000000000004e-06, "loss": 0.054, "step": 32468 }, { "epoch": 1.7310337474009703, "grad_norm": 2.350846529006958, "learning_rate": 3.5064000000000004e-06, "loss": 0.0277, "step": 32469 }, { "epoch": 1.7310870608306232, "grad_norm": 1.5059846639633179, "learning_rate": 3.5062e-06, "loss": 0.0115, "step": 32470 }, { "epoch": 1.731140374260276, "grad_norm": 1.0239715576171875, "learning_rate": 3.5060000000000004e-06, "loss": 0.0114, "step": 32471 }, { "epoch": 1.7311936876899292, "grad_norm": 2.0009572505950928, "learning_rate": 3.5058000000000004e-06, "loss": 0.0344, "step": 32472 }, { "epoch": 1.731247001119582, "grad_norm": 2.7871286869049072, "learning_rate": 3.5056e-06, "loss": 0.0395, "step": 32473 }, { "epoch": 1.731300314549235, "grad_norm": 2.6201655864715576, "learning_rate": 3.5054000000000004e-06, "loss": 0.0299, "step": 32474 }, { "epoch": 1.731353627978888, "grad_norm": 0.6584342122077942, "learning_rate": 3.5052000000000003e-06, "loss": 0.007, "step": 32475 }, { "epoch": 1.7314069414085407, "grad_norm": 1.529294490814209, "learning_rate": 3.505e-06, "loss": 0.0226, "step": 32476 }, { "epoch": 1.7314602548381939, "grad_norm": 2.6164393424987793, "learning_rate": 3.5048000000000003e-06, "loss": 0.0354, "step": 32477 }, { "epoch": 1.7315135682678466, "grad_norm": 2.0503695011138916, "learning_rate": 3.5046000000000003e-06, "loss": 0.023, "step": 32478 }, { "epoch": 1.7315668816974996, "grad_norm": 2.0518176555633545, "learning_rate": 3.5044e-06, "loss": 0.0225, "step": 32479 }, { "epoch": 1.7316201951271526, "grad_norm": 1.9756747484207153, "learning_rate": 3.5042000000000003e-06, "loss": 0.0162, "step": 32480 }, { "epoch": 1.7316735085568054, "grad_norm": 1.0865432024002075, "learning_rate": 3.5040000000000002e-06, "loss": 0.0114, "step": 32481 }, { "epoch": 1.7317268219864586, "grad_norm": 4.440706729888916, "learning_rate": 3.5038000000000007e-06, "loss": 0.0212, "step": 32482 }, { "epoch": 1.7317801354161113, "grad_norm": 1.113756537437439, "learning_rate": 3.5036000000000002e-06, "loss": 0.0249, "step": 32483 }, { "epoch": 1.7318334488457643, "grad_norm": 2.4095916748046875, "learning_rate": 3.5034e-06, "loss": 0.0153, "step": 32484 }, { "epoch": 1.7318867622754173, "grad_norm": 1.1202434301376343, "learning_rate": 3.5032000000000006e-06, "loss": 0.0168, "step": 32485 }, { "epoch": 1.73194007570507, "grad_norm": 1.6602740287780762, "learning_rate": 3.503e-06, "loss": 0.0228, "step": 32486 }, { "epoch": 1.731993389134723, "grad_norm": 1.2957998514175415, "learning_rate": 3.5028e-06, "loss": 0.0145, "step": 32487 }, { "epoch": 1.732046702564376, "grad_norm": 1.171539068222046, "learning_rate": 3.5026000000000006e-06, "loss": 0.0093, "step": 32488 }, { "epoch": 1.7321000159940287, "grad_norm": 1.6324400901794434, "learning_rate": 3.5024e-06, "loss": 0.0289, "step": 32489 }, { "epoch": 1.732153329423682, "grad_norm": 1.6875536441802979, "learning_rate": 3.5022e-06, "loss": 0.0136, "step": 32490 }, { "epoch": 1.7322066428533347, "grad_norm": 6.127553462982178, "learning_rate": 3.5020000000000005e-06, "loss": 0.0587, "step": 32491 }, { "epoch": 1.7322599562829877, "grad_norm": 2.566053628921509, "learning_rate": 3.5018e-06, "loss": 0.0335, "step": 32492 }, { "epoch": 1.7323132697126407, "grad_norm": 2.1007895469665527, "learning_rate": 3.5016000000000005e-06, "loss": 0.0199, "step": 32493 }, { "epoch": 1.7323665831422934, "grad_norm": 1.2373062372207642, "learning_rate": 3.5014000000000005e-06, "loss": 0.018, "step": 32494 }, { "epoch": 1.7324198965719466, "grad_norm": 1.0375827550888062, "learning_rate": 3.5012e-06, "loss": 0.0166, "step": 32495 }, { "epoch": 1.7324732100015994, "grad_norm": 0.5506250262260437, "learning_rate": 3.5010000000000004e-06, "loss": 0.0049, "step": 32496 }, { "epoch": 1.7325265234312524, "grad_norm": 1.4869744777679443, "learning_rate": 3.5008000000000004e-06, "loss": 0.0236, "step": 32497 }, { "epoch": 1.7325798368609053, "grad_norm": 3.355696201324463, "learning_rate": 3.5006e-06, "loss": 0.0251, "step": 32498 }, { "epoch": 1.732633150290558, "grad_norm": 1.232722520828247, "learning_rate": 3.5004000000000004e-06, "loss": 0.0094, "step": 32499 }, { "epoch": 1.732686463720211, "grad_norm": 1.1628106832504272, "learning_rate": 3.5002000000000004e-06, "loss": 0.0077, "step": 32500 }, { "epoch": 1.732739777149864, "grad_norm": 1.3486037254333496, "learning_rate": 3.5e-06, "loss": 0.0167, "step": 32501 }, { "epoch": 1.732793090579517, "grad_norm": 1.2189964056015015, "learning_rate": 3.4998000000000004e-06, "loss": 0.026, "step": 32502 }, { "epoch": 1.73284640400917, "grad_norm": 1.4175593852996826, "learning_rate": 3.4996000000000003e-06, "loss": 0.0132, "step": 32503 }, { "epoch": 1.7328997174388228, "grad_norm": 1.856846570968628, "learning_rate": 3.4994e-06, "loss": 0.0106, "step": 32504 }, { "epoch": 1.7329530308684757, "grad_norm": 0.7239310145378113, "learning_rate": 3.4992000000000003e-06, "loss": 0.0042, "step": 32505 }, { "epoch": 1.7330063442981287, "grad_norm": 1.9669467210769653, "learning_rate": 3.4990000000000003e-06, "loss": 0.0204, "step": 32506 }, { "epoch": 1.7330596577277815, "grad_norm": 1.107100248336792, "learning_rate": 3.4988000000000007e-06, "loss": 0.0086, "step": 32507 }, { "epoch": 1.7331129711574347, "grad_norm": 2.4523632526397705, "learning_rate": 3.4986000000000003e-06, "loss": 0.0456, "step": 32508 }, { "epoch": 1.7331662845870874, "grad_norm": 0.8589680194854736, "learning_rate": 3.4984000000000003e-06, "loss": 0.006, "step": 32509 }, { "epoch": 1.7332195980167404, "grad_norm": 1.1592018604278564, "learning_rate": 3.4982000000000007e-06, "loss": 0.0102, "step": 32510 }, { "epoch": 1.7332729114463934, "grad_norm": 1.1733324527740479, "learning_rate": 3.4980000000000002e-06, "loss": 0.0096, "step": 32511 }, { "epoch": 1.7333262248760462, "grad_norm": 0.4602912962436676, "learning_rate": 3.4978e-06, "loss": 0.0045, "step": 32512 }, { "epoch": 1.7333795383056994, "grad_norm": 1.8362897634506226, "learning_rate": 3.4976000000000006e-06, "loss": 0.0269, "step": 32513 }, { "epoch": 1.733432851735352, "grad_norm": 2.3790221214294434, "learning_rate": 3.4974e-06, "loss": 0.033, "step": 32514 }, { "epoch": 1.733486165165005, "grad_norm": 3.760354518890381, "learning_rate": 3.4972e-06, "loss": 0.0495, "step": 32515 }, { "epoch": 1.733539478594658, "grad_norm": 1.8232307434082031, "learning_rate": 3.4970000000000006e-06, "loss": 0.0109, "step": 32516 }, { "epoch": 1.7335927920243108, "grad_norm": 1.032118558883667, "learning_rate": 3.4968e-06, "loss": 0.0069, "step": 32517 }, { "epoch": 1.7336461054539638, "grad_norm": 1.7055193185806274, "learning_rate": 3.4966000000000005e-06, "loss": 0.027, "step": 32518 }, { "epoch": 1.7336994188836168, "grad_norm": 2.121677875518799, "learning_rate": 3.4964000000000005e-06, "loss": 0.0183, "step": 32519 }, { "epoch": 1.7337527323132698, "grad_norm": 1.8290570974349976, "learning_rate": 3.4962e-06, "loss": 0.0244, "step": 32520 }, { "epoch": 1.7338060457429227, "grad_norm": 1.0134185552597046, "learning_rate": 3.4960000000000005e-06, "loss": 0.0085, "step": 32521 }, { "epoch": 1.7338593591725755, "grad_norm": 2.3526611328125, "learning_rate": 3.4958000000000005e-06, "loss": 0.0132, "step": 32522 }, { "epoch": 1.7339126726022285, "grad_norm": 1.4237045049667358, "learning_rate": 3.4956e-06, "loss": 0.0068, "step": 32523 }, { "epoch": 1.7339659860318815, "grad_norm": 1.9969412088394165, "learning_rate": 3.4954000000000004e-06, "loss": 0.0291, "step": 32524 }, { "epoch": 1.7340192994615342, "grad_norm": 1.6446703672409058, "learning_rate": 3.4952000000000004e-06, "loss": 0.0134, "step": 32525 }, { "epoch": 1.7340726128911874, "grad_norm": 1.5612205266952515, "learning_rate": 3.495e-06, "loss": 0.0081, "step": 32526 }, { "epoch": 1.7341259263208402, "grad_norm": 1.2658894062042236, "learning_rate": 3.4948000000000004e-06, "loss": 0.0082, "step": 32527 }, { "epoch": 1.7341792397504932, "grad_norm": 2.7465145587921143, "learning_rate": 3.4946000000000004e-06, "loss": 0.0129, "step": 32528 }, { "epoch": 1.7342325531801461, "grad_norm": 4.536001682281494, "learning_rate": 3.4944e-06, "loss": 0.0265, "step": 32529 }, { "epoch": 1.7342858666097989, "grad_norm": 1.5161778926849365, "learning_rate": 3.4942000000000004e-06, "loss": 0.0116, "step": 32530 }, { "epoch": 1.734339180039452, "grad_norm": 2.0360987186431885, "learning_rate": 3.4940000000000003e-06, "loss": 0.0087, "step": 32531 }, { "epoch": 1.7343924934691048, "grad_norm": 1.7948517799377441, "learning_rate": 3.4938000000000003e-06, "loss": 0.0432, "step": 32532 }, { "epoch": 1.7344458068987578, "grad_norm": 0.3673042058944702, "learning_rate": 3.4936000000000003e-06, "loss": 0.0028, "step": 32533 }, { "epoch": 1.7344991203284108, "grad_norm": 1.556308627128601, "learning_rate": 3.4934000000000003e-06, "loss": 0.006, "step": 32534 }, { "epoch": 1.7345524337580636, "grad_norm": 1.2754617929458618, "learning_rate": 3.4932000000000003e-06, "loss": 0.0102, "step": 32535 }, { "epoch": 1.7346057471877165, "grad_norm": 2.4094393253326416, "learning_rate": 3.4930000000000003e-06, "loss": 0.0137, "step": 32536 }, { "epoch": 1.7346590606173695, "grad_norm": 0.8437501192092896, "learning_rate": 3.4928000000000003e-06, "loss": 0.0026, "step": 32537 }, { "epoch": 1.7347123740470225, "grad_norm": 1.7506420612335205, "learning_rate": 3.4926000000000002e-06, "loss": 0.0328, "step": 32538 }, { "epoch": 1.7347656874766755, "grad_norm": 1.1228636503219604, "learning_rate": 3.4924000000000002e-06, "loss": 0.0046, "step": 32539 }, { "epoch": 1.7348190009063282, "grad_norm": 0.9034739136695862, "learning_rate": 3.4922e-06, "loss": 0.0121, "step": 32540 }, { "epoch": 1.7348723143359812, "grad_norm": 1.6143229007720947, "learning_rate": 3.492e-06, "loss": 0.0149, "step": 32541 }, { "epoch": 1.7349256277656342, "grad_norm": 1.4217770099639893, "learning_rate": 3.4918e-06, "loss": 0.008, "step": 32542 }, { "epoch": 1.734978941195287, "grad_norm": 2.044999599456787, "learning_rate": 3.4916000000000006e-06, "loss": 0.0397, "step": 32543 }, { "epoch": 1.7350322546249402, "grad_norm": 2.6945719718933105, "learning_rate": 3.4914e-06, "loss": 0.0526, "step": 32544 }, { "epoch": 1.735085568054593, "grad_norm": 1.7311911582946777, "learning_rate": 3.4912e-06, "loss": 0.0382, "step": 32545 }, { "epoch": 1.7351388814842459, "grad_norm": 1.4110603332519531, "learning_rate": 3.4910000000000005e-06, "loss": 0.0197, "step": 32546 }, { "epoch": 1.7351921949138989, "grad_norm": 3.471952199935913, "learning_rate": 3.4908e-06, "loss": 0.0243, "step": 32547 }, { "epoch": 1.7352455083435516, "grad_norm": 1.9427489042282104, "learning_rate": 3.4906e-06, "loss": 0.0196, "step": 32548 }, { "epoch": 1.7352988217732048, "grad_norm": 1.6125383377075195, "learning_rate": 3.4904000000000005e-06, "loss": 0.0138, "step": 32549 }, { "epoch": 1.7353521352028576, "grad_norm": 1.6575742959976196, "learning_rate": 3.4902e-06, "loss": 0.0165, "step": 32550 }, { "epoch": 1.7354054486325106, "grad_norm": 2.488316297531128, "learning_rate": 3.49e-06, "loss": 0.0401, "step": 32551 }, { "epoch": 1.7354587620621635, "grad_norm": 1.4063361883163452, "learning_rate": 3.4898000000000005e-06, "loss": 0.0312, "step": 32552 }, { "epoch": 1.7355120754918163, "grad_norm": 1.5210766792297363, "learning_rate": 3.4896e-06, "loss": 0.0144, "step": 32553 }, { "epoch": 1.7355653889214693, "grad_norm": 2.3813295364379883, "learning_rate": 3.4894e-06, "loss": 0.0347, "step": 32554 }, { "epoch": 1.7356187023511223, "grad_norm": 1.4943102598190308, "learning_rate": 3.4892000000000004e-06, "loss": 0.0124, "step": 32555 }, { "epoch": 1.735672015780775, "grad_norm": 1.7724395990371704, "learning_rate": 3.489e-06, "loss": 0.0279, "step": 32556 }, { "epoch": 1.7357253292104282, "grad_norm": 1.3658331632614136, "learning_rate": 3.4888000000000004e-06, "loss": 0.0083, "step": 32557 }, { "epoch": 1.735778642640081, "grad_norm": 1.5044418573379517, "learning_rate": 3.4886000000000004e-06, "loss": 0.0257, "step": 32558 }, { "epoch": 1.735831956069734, "grad_norm": 3.623274087905884, "learning_rate": 3.4884e-06, "loss": 0.0281, "step": 32559 }, { "epoch": 1.735885269499387, "grad_norm": 2.559464693069458, "learning_rate": 3.4882000000000003e-06, "loss": 0.0228, "step": 32560 }, { "epoch": 1.7359385829290397, "grad_norm": 2.4745171070098877, "learning_rate": 3.4880000000000003e-06, "loss": 0.0134, "step": 32561 }, { "epoch": 1.7359918963586929, "grad_norm": 3.6879072189331055, "learning_rate": 3.4878e-06, "loss": 0.0518, "step": 32562 }, { "epoch": 1.7360452097883456, "grad_norm": 0.9145585894584656, "learning_rate": 3.4876000000000003e-06, "loss": 0.0061, "step": 32563 }, { "epoch": 1.7360985232179986, "grad_norm": 1.4883003234863281, "learning_rate": 3.4874000000000003e-06, "loss": 0.0115, "step": 32564 }, { "epoch": 1.7361518366476516, "grad_norm": 0.958954930305481, "learning_rate": 3.4872e-06, "loss": 0.0084, "step": 32565 }, { "epoch": 1.7362051500773044, "grad_norm": 1.6970988512039185, "learning_rate": 3.4870000000000002e-06, "loss": 0.0252, "step": 32566 }, { "epoch": 1.7362584635069576, "grad_norm": 2.2253310680389404, "learning_rate": 3.4868000000000002e-06, "loss": 0.0481, "step": 32567 }, { "epoch": 1.7363117769366103, "grad_norm": 2.934227228164673, "learning_rate": 3.4866000000000006e-06, "loss": 0.0465, "step": 32568 }, { "epoch": 1.7363650903662633, "grad_norm": 1.1892703771591187, "learning_rate": 3.4864e-06, "loss": 0.0177, "step": 32569 }, { "epoch": 1.7364184037959163, "grad_norm": 1.1457414627075195, "learning_rate": 3.4862e-06, "loss": 0.0088, "step": 32570 }, { "epoch": 1.736471717225569, "grad_norm": 1.396661400794983, "learning_rate": 3.4860000000000006e-06, "loss": 0.0142, "step": 32571 }, { "epoch": 1.736525030655222, "grad_norm": 0.4533054232597351, "learning_rate": 3.4858e-06, "loss": 0.0039, "step": 32572 }, { "epoch": 1.736578344084875, "grad_norm": 1.7300995588302612, "learning_rate": 3.4856e-06, "loss": 0.0084, "step": 32573 }, { "epoch": 1.7366316575145277, "grad_norm": 2.335996150970459, "learning_rate": 3.4854000000000005e-06, "loss": 0.0327, "step": 32574 }, { "epoch": 1.736684970944181, "grad_norm": 1.3778048753738403, "learning_rate": 3.4852e-06, "loss": 0.019, "step": 32575 }, { "epoch": 1.7367382843738337, "grad_norm": 2.338191032409668, "learning_rate": 3.485e-06, "loss": 0.0117, "step": 32576 }, { "epoch": 1.7367915978034867, "grad_norm": 1.80243718624115, "learning_rate": 3.4848000000000005e-06, "loss": 0.0274, "step": 32577 }, { "epoch": 1.7368449112331397, "grad_norm": 3.0362651348114014, "learning_rate": 3.4846e-06, "loss": 0.0207, "step": 32578 }, { "epoch": 1.7368982246627924, "grad_norm": 1.0437036752700806, "learning_rate": 3.4844000000000005e-06, "loss": 0.0122, "step": 32579 }, { "epoch": 1.7369515380924456, "grad_norm": 1.4603327512741089, "learning_rate": 3.4842000000000005e-06, "loss": 0.011, "step": 32580 }, { "epoch": 1.7370048515220984, "grad_norm": 1.4919281005859375, "learning_rate": 3.484e-06, "loss": 0.0183, "step": 32581 }, { "epoch": 1.7370581649517514, "grad_norm": 2.0363707542419434, "learning_rate": 3.4838000000000004e-06, "loss": 0.0176, "step": 32582 }, { "epoch": 1.7371114783814043, "grad_norm": 2.985361099243164, "learning_rate": 3.4836000000000004e-06, "loss": 0.0304, "step": 32583 }, { "epoch": 1.737164791811057, "grad_norm": 1.553916335105896, "learning_rate": 3.4834e-06, "loss": 0.0107, "step": 32584 }, { "epoch": 1.7372181052407103, "grad_norm": 0.9992796182632446, "learning_rate": 3.4832000000000004e-06, "loss": 0.0129, "step": 32585 }, { "epoch": 1.737271418670363, "grad_norm": 6.3611016273498535, "learning_rate": 3.4830000000000004e-06, "loss": 0.0236, "step": 32586 }, { "epoch": 1.737324732100016, "grad_norm": 2.062213182449341, "learning_rate": 3.4828e-06, "loss": 0.0201, "step": 32587 }, { "epoch": 1.737378045529669, "grad_norm": 1.6621187925338745, "learning_rate": 3.4826000000000003e-06, "loss": 0.0097, "step": 32588 }, { "epoch": 1.7374313589593218, "grad_norm": 2.1511390209198, "learning_rate": 3.4824000000000003e-06, "loss": 0.0285, "step": 32589 }, { "epoch": 1.7374846723889747, "grad_norm": 1.3900688886642456, "learning_rate": 3.4822e-06, "loss": 0.0106, "step": 32590 }, { "epoch": 1.7375379858186277, "grad_norm": 2.9438118934631348, "learning_rate": 3.4820000000000003e-06, "loss": 0.03, "step": 32591 }, { "epoch": 1.7375912992482805, "grad_norm": 1.0859739780426025, "learning_rate": 3.4818000000000003e-06, "loss": 0.0074, "step": 32592 }, { "epoch": 1.7376446126779337, "grad_norm": 1.1003997325897217, "learning_rate": 3.4816000000000007e-06, "loss": 0.0058, "step": 32593 }, { "epoch": 1.7376979261075864, "grad_norm": 1.1021971702575684, "learning_rate": 3.4814000000000002e-06, "loss": 0.0041, "step": 32594 }, { "epoch": 1.7377512395372394, "grad_norm": 0.9210807681083679, "learning_rate": 3.4812000000000002e-06, "loss": 0.0129, "step": 32595 }, { "epoch": 1.7378045529668924, "grad_norm": 1.8915659189224243, "learning_rate": 3.4810000000000006e-06, "loss": 0.0243, "step": 32596 }, { "epoch": 1.7378578663965452, "grad_norm": 5.440916538238525, "learning_rate": 3.4808e-06, "loss": 0.0538, "step": 32597 }, { "epoch": 1.7379111798261984, "grad_norm": 1.9542794227600098, "learning_rate": 3.4806e-06, "loss": 0.0307, "step": 32598 }, { "epoch": 1.7379644932558511, "grad_norm": 2.2269060611724854, "learning_rate": 3.4804000000000006e-06, "loss": 0.0332, "step": 32599 }, { "epoch": 1.738017806685504, "grad_norm": 1.890236258506775, "learning_rate": 3.4802e-06, "loss": 0.0352, "step": 32600 }, { "epoch": 1.738071120115157, "grad_norm": 1.692596197128296, "learning_rate": 3.48e-06, "loss": 0.0094, "step": 32601 }, { "epoch": 1.7381244335448098, "grad_norm": 1.1735191345214844, "learning_rate": 3.4798000000000005e-06, "loss": 0.0177, "step": 32602 }, { "epoch": 1.7381777469744628, "grad_norm": 1.3482921123504639, "learning_rate": 3.4796e-06, "loss": 0.0087, "step": 32603 }, { "epoch": 1.7382310604041158, "grad_norm": 1.912743091583252, "learning_rate": 3.4794000000000005e-06, "loss": 0.024, "step": 32604 }, { "epoch": 1.7382843738337688, "grad_norm": 1.9397112131118774, "learning_rate": 3.4792000000000005e-06, "loss": 0.0343, "step": 32605 }, { "epoch": 1.7383376872634217, "grad_norm": 1.1923335790634155, "learning_rate": 3.479e-06, "loss": 0.0146, "step": 32606 }, { "epoch": 1.7383910006930745, "grad_norm": 1.944357991218567, "learning_rate": 3.4788000000000005e-06, "loss": 0.0155, "step": 32607 }, { "epoch": 1.7384443141227275, "grad_norm": 1.2465392351150513, "learning_rate": 3.4786000000000005e-06, "loss": 0.0193, "step": 32608 }, { "epoch": 1.7384976275523805, "grad_norm": 1.800397276878357, "learning_rate": 3.4784e-06, "loss": 0.0157, "step": 32609 }, { "epoch": 1.7385509409820332, "grad_norm": 1.748405933380127, "learning_rate": 3.4782000000000004e-06, "loss": 0.0164, "step": 32610 }, { "epoch": 1.7386042544116864, "grad_norm": 2.0375871658325195, "learning_rate": 3.4780000000000004e-06, "loss": 0.0227, "step": 32611 }, { "epoch": 1.7386575678413392, "grad_norm": 2.639361619949341, "learning_rate": 3.4778e-06, "loss": 0.0335, "step": 32612 }, { "epoch": 1.7387108812709922, "grad_norm": 0.8635587096214294, "learning_rate": 3.4776000000000004e-06, "loss": 0.0059, "step": 32613 }, { "epoch": 1.7387641947006451, "grad_norm": 5.225762367248535, "learning_rate": 3.4774000000000004e-06, "loss": 0.0366, "step": 32614 }, { "epoch": 1.738817508130298, "grad_norm": 1.8344887495040894, "learning_rate": 3.4772e-06, "loss": 0.0352, "step": 32615 }, { "epoch": 1.738870821559951, "grad_norm": 1.7826924324035645, "learning_rate": 3.4770000000000003e-06, "loss": 0.0318, "step": 32616 }, { "epoch": 1.7389241349896039, "grad_norm": 1.0685230493545532, "learning_rate": 3.4768000000000003e-06, "loss": 0.0336, "step": 32617 }, { "epoch": 1.7389774484192568, "grad_norm": 1.0241457223892212, "learning_rate": 3.4766000000000007e-06, "loss": 0.0071, "step": 32618 }, { "epoch": 1.7390307618489098, "grad_norm": 1.7026866674423218, "learning_rate": 3.4764000000000003e-06, "loss": 0.0139, "step": 32619 }, { "epoch": 1.7390840752785626, "grad_norm": 1.5943801403045654, "learning_rate": 3.4762000000000003e-06, "loss": 0.0119, "step": 32620 }, { "epoch": 1.7391373887082155, "grad_norm": 1.9257183074951172, "learning_rate": 3.4760000000000007e-06, "loss": 0.013, "step": 32621 }, { "epoch": 1.7391907021378685, "grad_norm": 2.7427940368652344, "learning_rate": 3.4758000000000002e-06, "loss": 0.0289, "step": 32622 }, { "epoch": 1.7392440155675215, "grad_norm": 1.42477548122406, "learning_rate": 3.4756000000000002e-06, "loss": 0.0131, "step": 32623 }, { "epoch": 1.7392973289971745, "grad_norm": 1.046575903892517, "learning_rate": 3.4754000000000006e-06, "loss": 0.0138, "step": 32624 }, { "epoch": 1.7393506424268272, "grad_norm": 1.256202220916748, "learning_rate": 3.4752e-06, "loss": 0.0295, "step": 32625 }, { "epoch": 1.7394039558564802, "grad_norm": 1.410272479057312, "learning_rate": 3.475e-06, "loss": 0.0105, "step": 32626 }, { "epoch": 1.7394572692861332, "grad_norm": 3.5611157417297363, "learning_rate": 3.4748000000000006e-06, "loss": 0.024, "step": 32627 }, { "epoch": 1.739510582715786, "grad_norm": 1.2279372215270996, "learning_rate": 3.4746e-06, "loss": 0.0201, "step": 32628 }, { "epoch": 1.7395638961454392, "grad_norm": 1.3112810850143433, "learning_rate": 3.4744000000000006e-06, "loss": 0.0124, "step": 32629 }, { "epoch": 1.739617209575092, "grad_norm": 2.0235002040863037, "learning_rate": 3.4742000000000006e-06, "loss": 0.0101, "step": 32630 }, { "epoch": 1.739670523004745, "grad_norm": 2.7251501083374023, "learning_rate": 3.474e-06, "loss": 0.032, "step": 32631 }, { "epoch": 1.7397238364343979, "grad_norm": 1.885664701461792, "learning_rate": 3.4738000000000005e-06, "loss": 0.031, "step": 32632 }, { "epoch": 1.7397771498640506, "grad_norm": 1.163216471672058, "learning_rate": 3.4736000000000005e-06, "loss": 0.0099, "step": 32633 }, { "epoch": 1.7398304632937038, "grad_norm": 3.0272367000579834, "learning_rate": 3.4734e-06, "loss": 0.0284, "step": 32634 }, { "epoch": 1.7398837767233566, "grad_norm": 1.5150142908096313, "learning_rate": 3.4732000000000005e-06, "loss": 0.0181, "step": 32635 }, { "epoch": 1.7399370901530096, "grad_norm": 1.2210016250610352, "learning_rate": 3.4730000000000005e-06, "loss": 0.0115, "step": 32636 }, { "epoch": 1.7399904035826625, "grad_norm": 4.052363395690918, "learning_rate": 3.4728e-06, "loss": 0.0177, "step": 32637 }, { "epoch": 1.7400437170123153, "grad_norm": 2.692927360534668, "learning_rate": 3.4726000000000004e-06, "loss": 0.0198, "step": 32638 }, { "epoch": 1.7400970304419683, "grad_norm": 2.542731761932373, "learning_rate": 3.4724000000000004e-06, "loss": 0.0156, "step": 32639 }, { "epoch": 1.7401503438716213, "grad_norm": 1.3590545654296875, "learning_rate": 3.4722e-06, "loss": 0.0127, "step": 32640 }, { "epoch": 1.7402036573012742, "grad_norm": 2.631289005279541, "learning_rate": 3.4720000000000004e-06, "loss": 0.0101, "step": 32641 }, { "epoch": 1.7402569707309272, "grad_norm": 2.053095579147339, "learning_rate": 3.4718000000000004e-06, "loss": 0.0184, "step": 32642 }, { "epoch": 1.74031028416058, "grad_norm": 2.4560699462890625, "learning_rate": 3.4716000000000004e-06, "loss": 0.0241, "step": 32643 }, { "epoch": 1.740363597590233, "grad_norm": 1.0560659170150757, "learning_rate": 3.4714000000000003e-06, "loss": 0.0154, "step": 32644 }, { "epoch": 1.740416911019886, "grad_norm": 2.2906057834625244, "learning_rate": 3.4712000000000003e-06, "loss": 0.014, "step": 32645 }, { "epoch": 1.7404702244495387, "grad_norm": 2.358076333999634, "learning_rate": 3.4710000000000003e-06, "loss": 0.0221, "step": 32646 }, { "epoch": 1.740523537879192, "grad_norm": 2.104560613632202, "learning_rate": 3.4708000000000003e-06, "loss": 0.0337, "step": 32647 }, { "epoch": 1.7405768513088447, "grad_norm": 1.3088784217834473, "learning_rate": 3.4706000000000003e-06, "loss": 0.0117, "step": 32648 }, { "epoch": 1.7406301647384976, "grad_norm": 3.1424150466918945, "learning_rate": 3.4704000000000003e-06, "loss": 0.0333, "step": 32649 }, { "epoch": 1.7406834781681506, "grad_norm": 2.614039659500122, "learning_rate": 3.4702000000000002e-06, "loss": 0.0216, "step": 32650 }, { "epoch": 1.7407367915978034, "grad_norm": 1.2108567953109741, "learning_rate": 3.4700000000000002e-06, "loss": 0.0082, "step": 32651 }, { "epoch": 1.7407901050274566, "grad_norm": 1.7867108583450317, "learning_rate": 3.4698000000000002e-06, "loss": 0.0116, "step": 32652 }, { "epoch": 1.7408434184571093, "grad_norm": 0.4369516968727112, "learning_rate": 3.4696e-06, "loss": 0.0036, "step": 32653 }, { "epoch": 1.7408967318867623, "grad_norm": 1.6616058349609375, "learning_rate": 3.4694000000000006e-06, "loss": 0.0112, "step": 32654 }, { "epoch": 1.7409500453164153, "grad_norm": 1.3138043880462646, "learning_rate": 3.4692e-06, "loss": 0.0063, "step": 32655 }, { "epoch": 1.741003358746068, "grad_norm": 1.2477656602859497, "learning_rate": 3.469e-06, "loss": 0.0062, "step": 32656 }, { "epoch": 1.741056672175721, "grad_norm": 0.35149455070495605, "learning_rate": 3.4688000000000006e-06, "loss": 0.0033, "step": 32657 }, { "epoch": 1.741109985605374, "grad_norm": 1.6967508792877197, "learning_rate": 3.4686e-06, "loss": 0.0222, "step": 32658 }, { "epoch": 1.7411632990350268, "grad_norm": 1.8456501960754395, "learning_rate": 3.4684e-06, "loss": 0.0269, "step": 32659 }, { "epoch": 1.74121661246468, "grad_norm": 0.1450355052947998, "learning_rate": 3.4682000000000005e-06, "loss": 0.0017, "step": 32660 }, { "epoch": 1.7412699258943327, "grad_norm": 1.6987066268920898, "learning_rate": 3.468e-06, "loss": 0.0106, "step": 32661 }, { "epoch": 1.7413232393239857, "grad_norm": 0.2682054936885834, "learning_rate": 3.4678e-06, "loss": 0.0022, "step": 32662 }, { "epoch": 1.7413765527536387, "grad_norm": 1.7257386445999146, "learning_rate": 3.4676000000000005e-06, "loss": 0.027, "step": 32663 }, { "epoch": 1.7414298661832914, "grad_norm": 1.9431755542755127, "learning_rate": 3.4674e-06, "loss": 0.017, "step": 32664 }, { "epoch": 1.7414831796129446, "grad_norm": 1.4385215044021606, "learning_rate": 3.4672e-06, "loss": 0.0166, "step": 32665 }, { "epoch": 1.7415364930425974, "grad_norm": 2.6355507373809814, "learning_rate": 3.4670000000000004e-06, "loss": 0.0353, "step": 32666 }, { "epoch": 1.7415898064722504, "grad_norm": 0.4229637086391449, "learning_rate": 3.4668e-06, "loss": 0.0027, "step": 32667 }, { "epoch": 1.7416431199019033, "grad_norm": 0.9459633231163025, "learning_rate": 3.4666000000000004e-06, "loss": 0.0056, "step": 32668 }, { "epoch": 1.741696433331556, "grad_norm": 1.5378984212875366, "learning_rate": 3.4664000000000004e-06, "loss": 0.0224, "step": 32669 }, { "epoch": 1.7417497467612093, "grad_norm": 1.9524654150009155, "learning_rate": 3.4662e-06, "loss": 0.0075, "step": 32670 }, { "epoch": 1.741803060190862, "grad_norm": 3.256268262863159, "learning_rate": 3.4660000000000004e-06, "loss": 0.0499, "step": 32671 }, { "epoch": 1.741856373620515, "grad_norm": 2.260606050491333, "learning_rate": 3.4658000000000003e-06, "loss": 0.0192, "step": 32672 }, { "epoch": 1.741909687050168, "grad_norm": 0.9448930621147156, "learning_rate": 3.4656e-06, "loss": 0.005, "step": 32673 }, { "epoch": 1.7419630004798208, "grad_norm": 1.590372085571289, "learning_rate": 3.4654000000000003e-06, "loss": 0.01, "step": 32674 }, { "epoch": 1.7420163139094738, "grad_norm": 1.1212764978408813, "learning_rate": 3.4652000000000003e-06, "loss": 0.0246, "step": 32675 }, { "epoch": 1.7420696273391267, "grad_norm": 2.5029451847076416, "learning_rate": 3.465e-06, "loss": 0.0455, "step": 32676 }, { "epoch": 1.7421229407687795, "grad_norm": 2.5714399814605713, "learning_rate": 3.4648000000000003e-06, "loss": 0.0382, "step": 32677 }, { "epoch": 1.7421762541984327, "grad_norm": 2.6660847663879395, "learning_rate": 3.4646000000000003e-06, "loss": 0.042, "step": 32678 }, { "epoch": 1.7422295676280855, "grad_norm": 3.297045946121216, "learning_rate": 3.4644000000000007e-06, "loss": 0.0294, "step": 32679 }, { "epoch": 1.7422828810577384, "grad_norm": 2.4070279598236084, "learning_rate": 3.4642000000000002e-06, "loss": 0.0302, "step": 32680 }, { "epoch": 1.7423361944873914, "grad_norm": 2.499465227127075, "learning_rate": 3.464e-06, "loss": 0.0181, "step": 32681 }, { "epoch": 1.7423895079170442, "grad_norm": 3.468705892562866, "learning_rate": 3.4638000000000006e-06, "loss": 0.0423, "step": 32682 }, { "epoch": 1.7424428213466974, "grad_norm": 3.4190759658813477, "learning_rate": 3.4636e-06, "loss": 0.0571, "step": 32683 }, { "epoch": 1.7424961347763501, "grad_norm": 2.858680486679077, "learning_rate": 3.4634e-06, "loss": 0.024, "step": 32684 }, { "epoch": 1.742549448206003, "grad_norm": 1.872307300567627, "learning_rate": 3.4632000000000006e-06, "loss": 0.0298, "step": 32685 }, { "epoch": 1.742602761635656, "grad_norm": 1.6171303987503052, "learning_rate": 3.463e-06, "loss": 0.0095, "step": 32686 }, { "epoch": 1.7426560750653088, "grad_norm": 1.6973718404769897, "learning_rate": 3.4628e-06, "loss": 0.0111, "step": 32687 }, { "epoch": 1.742709388494962, "grad_norm": 0.8298865556716919, "learning_rate": 3.4626000000000005e-06, "loss": 0.0064, "step": 32688 }, { "epoch": 1.7427627019246148, "grad_norm": 3.250764846801758, "learning_rate": 3.4624e-06, "loss": 0.0288, "step": 32689 }, { "epoch": 1.7428160153542678, "grad_norm": 1.5078362226486206, "learning_rate": 3.4622000000000005e-06, "loss": 0.012, "step": 32690 }, { "epoch": 1.7428693287839208, "grad_norm": 0.7384145259857178, "learning_rate": 3.4620000000000005e-06, "loss": 0.0063, "step": 32691 }, { "epoch": 1.7429226422135735, "grad_norm": 4.139425277709961, "learning_rate": 3.4618e-06, "loss": 0.0291, "step": 32692 }, { "epoch": 1.7429759556432265, "grad_norm": 1.0972651243209839, "learning_rate": 3.4616000000000005e-06, "loss": 0.0064, "step": 32693 }, { "epoch": 1.7430292690728795, "grad_norm": 1.682530164718628, "learning_rate": 3.4614000000000004e-06, "loss": 0.0075, "step": 32694 }, { "epoch": 1.7430825825025322, "grad_norm": 1.6202404499053955, "learning_rate": 3.4612e-06, "loss": 0.012, "step": 32695 }, { "epoch": 1.7431358959321854, "grad_norm": 1.3429601192474365, "learning_rate": 3.4610000000000004e-06, "loss": 0.0196, "step": 32696 }, { "epoch": 1.7431892093618382, "grad_norm": 2.2621750831604004, "learning_rate": 3.4608000000000004e-06, "loss": 0.0305, "step": 32697 }, { "epoch": 1.7432425227914912, "grad_norm": 0.7993468046188354, "learning_rate": 3.4606e-06, "loss": 0.0078, "step": 32698 }, { "epoch": 1.7432958362211441, "grad_norm": 2.3773229122161865, "learning_rate": 3.4604000000000004e-06, "loss": 0.0241, "step": 32699 }, { "epoch": 1.743349149650797, "grad_norm": 0.13371078670024872, "learning_rate": 3.4602000000000003e-06, "loss": 0.0016, "step": 32700 }, { "epoch": 1.74340246308045, "grad_norm": 3.413111448287964, "learning_rate": 3.46e-06, "loss": 0.0354, "step": 32701 }, { "epoch": 1.7434557765101029, "grad_norm": 1.0246503353118896, "learning_rate": 3.4598000000000003e-06, "loss": 0.0175, "step": 32702 }, { "epoch": 1.7435090899397558, "grad_norm": 1.6738243103027344, "learning_rate": 3.4596000000000003e-06, "loss": 0.0239, "step": 32703 }, { "epoch": 1.7435624033694088, "grad_norm": 0.3853732943534851, "learning_rate": 3.4594000000000007e-06, "loss": 0.0037, "step": 32704 }, { "epoch": 1.7436157167990616, "grad_norm": 1.7134732007980347, "learning_rate": 3.4592000000000003e-06, "loss": 0.0248, "step": 32705 }, { "epoch": 1.7436690302287148, "grad_norm": 0.6137282252311707, "learning_rate": 3.4590000000000003e-06, "loss": 0.0045, "step": 32706 }, { "epoch": 1.7437223436583675, "grad_norm": 2.482534170150757, "learning_rate": 3.4588000000000007e-06, "loss": 0.0163, "step": 32707 }, { "epoch": 1.7437756570880205, "grad_norm": 2.044799566268921, "learning_rate": 3.4586000000000002e-06, "loss": 0.0456, "step": 32708 }, { "epoch": 1.7438289705176735, "grad_norm": 1.168601632118225, "learning_rate": 3.4584e-06, "loss": 0.0111, "step": 32709 }, { "epoch": 1.7438822839473263, "grad_norm": 0.8587895631790161, "learning_rate": 3.4582000000000006e-06, "loss": 0.0069, "step": 32710 }, { "epoch": 1.7439355973769792, "grad_norm": 4.100735187530518, "learning_rate": 3.458e-06, "loss": 0.0242, "step": 32711 }, { "epoch": 1.7439889108066322, "grad_norm": 1.5591506958007812, "learning_rate": 3.4578e-06, "loss": 0.022, "step": 32712 }, { "epoch": 1.744042224236285, "grad_norm": 1.132341742515564, "learning_rate": 3.4576000000000006e-06, "loss": 0.009, "step": 32713 }, { "epoch": 1.7440955376659382, "grad_norm": 1.0835596323013306, "learning_rate": 3.4574e-06, "loss": 0.0095, "step": 32714 }, { "epoch": 1.744148851095591, "grad_norm": 1.2859634160995483, "learning_rate": 3.4572000000000005e-06, "loss": 0.0051, "step": 32715 }, { "epoch": 1.744202164525244, "grad_norm": 2.5214273929595947, "learning_rate": 3.4570000000000005e-06, "loss": 0.0154, "step": 32716 }, { "epoch": 1.7442554779548969, "grad_norm": 1.9713480472564697, "learning_rate": 3.4568e-06, "loss": 0.0174, "step": 32717 }, { "epoch": 1.7443087913845496, "grad_norm": 1.9293040037155151, "learning_rate": 3.4566000000000005e-06, "loss": 0.0093, "step": 32718 }, { "epoch": 1.7443621048142028, "grad_norm": 1.6317394971847534, "learning_rate": 3.4564000000000005e-06, "loss": 0.0355, "step": 32719 }, { "epoch": 1.7444154182438556, "grad_norm": 2.7501513957977295, "learning_rate": 3.4562e-06, "loss": 0.0112, "step": 32720 }, { "epoch": 1.7444687316735086, "grad_norm": 3.4323182106018066, "learning_rate": 3.4560000000000005e-06, "loss": 0.0081, "step": 32721 }, { "epoch": 1.7445220451031616, "grad_norm": 0.4470668435096741, "learning_rate": 3.4558000000000004e-06, "loss": 0.0028, "step": 32722 }, { "epoch": 1.7445753585328143, "grad_norm": 1.7109867334365845, "learning_rate": 3.4556e-06, "loss": 0.0184, "step": 32723 }, { "epoch": 1.7446286719624673, "grad_norm": 2.261813163757324, "learning_rate": 3.4554000000000004e-06, "loss": 0.0408, "step": 32724 }, { "epoch": 1.7446819853921203, "grad_norm": 2.8728435039520264, "learning_rate": 3.4552000000000004e-06, "loss": 0.0375, "step": 32725 }, { "epoch": 1.7447352988217733, "grad_norm": 0.7300335168838501, "learning_rate": 3.455e-06, "loss": 0.0053, "step": 32726 }, { "epoch": 1.7447886122514262, "grad_norm": 1.06304132938385, "learning_rate": 3.4548000000000004e-06, "loss": 0.0153, "step": 32727 }, { "epoch": 1.744841925681079, "grad_norm": 0.33024272322654724, "learning_rate": 3.4546000000000003e-06, "loss": 0.0036, "step": 32728 }, { "epoch": 1.744895239110732, "grad_norm": 0.6566274762153625, "learning_rate": 3.4544000000000003e-06, "loss": 0.0058, "step": 32729 }, { "epoch": 1.744948552540385, "grad_norm": 2.4078705310821533, "learning_rate": 3.4542000000000003e-06, "loss": 0.0208, "step": 32730 }, { "epoch": 1.7450018659700377, "grad_norm": 1.9958778619766235, "learning_rate": 3.4540000000000003e-06, "loss": 0.0125, "step": 32731 }, { "epoch": 1.745055179399691, "grad_norm": 1.3367786407470703, "learning_rate": 3.4538000000000003e-06, "loss": 0.0236, "step": 32732 }, { "epoch": 1.7451084928293437, "grad_norm": 1.442847728729248, "learning_rate": 3.4536000000000003e-06, "loss": 0.014, "step": 32733 }, { "epoch": 1.7451618062589966, "grad_norm": 2.3419649600982666, "learning_rate": 3.4534000000000003e-06, "loss": 0.0292, "step": 32734 }, { "epoch": 1.7452151196886496, "grad_norm": 1.1986490488052368, "learning_rate": 3.4532000000000002e-06, "loss": 0.0076, "step": 32735 }, { "epoch": 1.7452684331183024, "grad_norm": 1.9378547668457031, "learning_rate": 3.4530000000000002e-06, "loss": 0.0137, "step": 32736 }, { "epoch": 1.7453217465479556, "grad_norm": 1.2221782207489014, "learning_rate": 3.4528000000000002e-06, "loss": 0.0089, "step": 32737 }, { "epoch": 1.7453750599776083, "grad_norm": 2.098231077194214, "learning_rate": 3.4526e-06, "loss": 0.0418, "step": 32738 }, { "epoch": 1.7454283734072613, "grad_norm": 1.2560628652572632, "learning_rate": 3.4524e-06, "loss": 0.0184, "step": 32739 }, { "epoch": 1.7454816868369143, "grad_norm": 3.9584083557128906, "learning_rate": 3.4522000000000006e-06, "loss": 0.0404, "step": 32740 }, { "epoch": 1.745535000266567, "grad_norm": 0.9178186655044556, "learning_rate": 3.452e-06, "loss": 0.018, "step": 32741 }, { "epoch": 1.74558831369622, "grad_norm": 1.5934323072433472, "learning_rate": 3.4518e-06, "loss": 0.0156, "step": 32742 }, { "epoch": 1.745641627125873, "grad_norm": 3.441895008087158, "learning_rate": 3.4516000000000005e-06, "loss": 0.0243, "step": 32743 }, { "epoch": 1.745694940555526, "grad_norm": 2.567946195602417, "learning_rate": 3.4514e-06, "loss": 0.0293, "step": 32744 }, { "epoch": 1.745748253985179, "grad_norm": 1.6233896017074585, "learning_rate": 3.4512e-06, "loss": 0.0073, "step": 32745 }, { "epoch": 1.7458015674148317, "grad_norm": 1.4271327257156372, "learning_rate": 3.4510000000000005e-06, "loss": 0.0184, "step": 32746 }, { "epoch": 1.7458548808444847, "grad_norm": 2.0548911094665527, "learning_rate": 3.4508e-06, "loss": 0.0316, "step": 32747 }, { "epoch": 1.7459081942741377, "grad_norm": 1.7581689357757568, "learning_rate": 3.4506e-06, "loss": 0.0258, "step": 32748 }, { "epoch": 1.7459615077037904, "grad_norm": 2.4550750255584717, "learning_rate": 3.4504000000000005e-06, "loss": 0.0355, "step": 32749 }, { "epoch": 1.7460148211334436, "grad_norm": 0.9272783994674683, "learning_rate": 3.4502e-06, "loss": 0.0048, "step": 32750 }, { "epoch": 1.7460681345630964, "grad_norm": 1.8134965896606445, "learning_rate": 3.45e-06, "loss": 0.0351, "step": 32751 }, { "epoch": 1.7461214479927494, "grad_norm": 1.0810885429382324, "learning_rate": 3.4498000000000004e-06, "loss": 0.0051, "step": 32752 }, { "epoch": 1.7461747614224024, "grad_norm": 1.8582783937454224, "learning_rate": 3.4496e-06, "loss": 0.0201, "step": 32753 }, { "epoch": 1.7462280748520551, "grad_norm": 0.6034343242645264, "learning_rate": 3.4494000000000004e-06, "loss": 0.0047, "step": 32754 }, { "epoch": 1.7462813882817083, "grad_norm": 2.3687374591827393, "learning_rate": 3.4492000000000004e-06, "loss": 0.055, "step": 32755 }, { "epoch": 1.746334701711361, "grad_norm": 2.105443239212036, "learning_rate": 3.449e-06, "loss": 0.0204, "step": 32756 }, { "epoch": 1.746388015141014, "grad_norm": 1.2568540573120117, "learning_rate": 3.4488000000000003e-06, "loss": 0.0072, "step": 32757 }, { "epoch": 1.746441328570667, "grad_norm": 2.699192523956299, "learning_rate": 3.4486000000000003e-06, "loss": 0.0444, "step": 32758 }, { "epoch": 1.7464946420003198, "grad_norm": 2.5543808937072754, "learning_rate": 3.4484e-06, "loss": 0.0117, "step": 32759 }, { "epoch": 1.7465479554299728, "grad_norm": 7.503362655639648, "learning_rate": 3.4482000000000003e-06, "loss": 0.0229, "step": 32760 }, { "epoch": 1.7466012688596257, "grad_norm": 2.100519895553589, "learning_rate": 3.4480000000000003e-06, "loss": 0.0398, "step": 32761 }, { "epoch": 1.7466545822892787, "grad_norm": 0.5864757299423218, "learning_rate": 3.4478e-06, "loss": 0.0054, "step": 32762 }, { "epoch": 1.7467078957189317, "grad_norm": 1.1049305200576782, "learning_rate": 3.4476000000000002e-06, "loss": 0.0093, "step": 32763 }, { "epoch": 1.7467612091485845, "grad_norm": 1.309113621711731, "learning_rate": 3.4474000000000002e-06, "loss": 0.016, "step": 32764 }, { "epoch": 1.7468145225782374, "grad_norm": 1.4536412954330444, "learning_rate": 3.4472000000000006e-06, "loss": 0.0385, "step": 32765 }, { "epoch": 1.7468678360078904, "grad_norm": 1.4705039262771606, "learning_rate": 3.447e-06, "loss": 0.0077, "step": 32766 }, { "epoch": 1.7469211494375432, "grad_norm": 1.9692764282226562, "learning_rate": 3.4468e-06, "loss": 0.0117, "step": 32767 }, { "epoch": 1.7469744628671964, "grad_norm": 0.6287222504615784, "learning_rate": 3.4466000000000006e-06, "loss": 0.0061, "step": 32768 }, { "epoch": 1.7470277762968491, "grad_norm": 2.6948912143707275, "learning_rate": 3.4464e-06, "loss": 0.0184, "step": 32769 }, { "epoch": 1.7470810897265021, "grad_norm": 1.775250792503357, "learning_rate": 3.4462e-06, "loss": 0.0427, "step": 32770 }, { "epoch": 1.747134403156155, "grad_norm": 1.3594950437545776, "learning_rate": 3.4460000000000005e-06, "loss": 0.0087, "step": 32771 }, { "epoch": 1.7471877165858078, "grad_norm": 5.037747383117676, "learning_rate": 3.4458e-06, "loss": 0.0335, "step": 32772 }, { "epoch": 1.747241030015461, "grad_norm": 1.7210242748260498, "learning_rate": 3.4456e-06, "loss": 0.021, "step": 32773 }, { "epoch": 1.7472943434451138, "grad_norm": 1.70933997631073, "learning_rate": 3.4454000000000005e-06, "loss": 0.0309, "step": 32774 }, { "epoch": 1.7473476568747668, "grad_norm": 1.2039496898651123, "learning_rate": 3.4452e-06, "loss": 0.0293, "step": 32775 }, { "epoch": 1.7474009703044198, "grad_norm": 1.8224823474884033, "learning_rate": 3.445e-06, "loss": 0.0321, "step": 32776 }, { "epoch": 1.7474542837340725, "grad_norm": 1.1047090291976929, "learning_rate": 3.4448000000000005e-06, "loss": 0.0063, "step": 32777 }, { "epoch": 1.7475075971637255, "grad_norm": 1.2221814393997192, "learning_rate": 3.4446e-06, "loss": 0.015, "step": 32778 }, { "epoch": 1.7475609105933785, "grad_norm": 2.4770283699035645, "learning_rate": 3.4444000000000004e-06, "loss": 0.0335, "step": 32779 }, { "epoch": 1.7476142240230312, "grad_norm": 1.279815673828125, "learning_rate": 3.4442000000000004e-06, "loss": 0.0084, "step": 32780 }, { "epoch": 1.7476675374526844, "grad_norm": 2.783344268798828, "learning_rate": 3.444e-06, "loss": 0.0232, "step": 32781 }, { "epoch": 1.7477208508823372, "grad_norm": 1.4341681003570557, "learning_rate": 3.4438000000000004e-06, "loss": 0.016, "step": 32782 }, { "epoch": 1.7477741643119902, "grad_norm": 0.9755116701126099, "learning_rate": 3.4436000000000004e-06, "loss": 0.0213, "step": 32783 }, { "epoch": 1.7478274777416432, "grad_norm": 1.4908862113952637, "learning_rate": 3.4434e-06, "loss": 0.0154, "step": 32784 }, { "epoch": 1.747880791171296, "grad_norm": 1.6836144924163818, "learning_rate": 3.4432000000000003e-06, "loss": 0.0186, "step": 32785 }, { "epoch": 1.7479341046009491, "grad_norm": 2.8116865158081055, "learning_rate": 3.4430000000000003e-06, "loss": 0.0321, "step": 32786 }, { "epoch": 1.7479874180306019, "grad_norm": 1.1143418550491333, "learning_rate": 3.4428e-06, "loss": 0.0122, "step": 32787 }, { "epoch": 1.7480407314602548, "grad_norm": 0.2690654397010803, "learning_rate": 3.4426000000000003e-06, "loss": 0.0026, "step": 32788 }, { "epoch": 1.7480940448899078, "grad_norm": 2.510643720626831, "learning_rate": 3.4424000000000003e-06, "loss": 0.0268, "step": 32789 }, { "epoch": 1.7481473583195606, "grad_norm": 1.999029517173767, "learning_rate": 3.4422000000000007e-06, "loss": 0.0179, "step": 32790 }, { "epoch": 1.7482006717492138, "grad_norm": 0.9765827059745789, "learning_rate": 3.4420000000000002e-06, "loss": 0.0179, "step": 32791 }, { "epoch": 1.7482539851788665, "grad_norm": 1.230143427848816, "learning_rate": 3.4418000000000002e-06, "loss": 0.0065, "step": 32792 }, { "epoch": 1.7483072986085195, "grad_norm": 0.9986783862113953, "learning_rate": 3.4416000000000006e-06, "loss": 0.0069, "step": 32793 }, { "epoch": 1.7483606120381725, "grad_norm": 1.4072591066360474, "learning_rate": 3.4414e-06, "loss": 0.0084, "step": 32794 }, { "epoch": 1.7484139254678253, "grad_norm": 2.5487754344940186, "learning_rate": 3.4412e-06, "loss": 0.0204, "step": 32795 }, { "epoch": 1.7484672388974782, "grad_norm": 2.7660655975341797, "learning_rate": 3.4410000000000006e-06, "loss": 0.0282, "step": 32796 }, { "epoch": 1.7485205523271312, "grad_norm": 0.9405093193054199, "learning_rate": 3.4408e-06, "loss": 0.0102, "step": 32797 }, { "epoch": 1.748573865756784, "grad_norm": 0.8787694573402405, "learning_rate": 3.4406e-06, "loss": 0.0133, "step": 32798 }, { "epoch": 1.7486271791864372, "grad_norm": 1.2375253438949585, "learning_rate": 3.4404000000000006e-06, "loss": 0.0085, "step": 32799 }, { "epoch": 1.74868049261609, "grad_norm": 1.5826336145401, "learning_rate": 3.4402e-06, "loss": 0.0301, "step": 32800 }, { "epoch": 1.748733806045743, "grad_norm": 2.338779926300049, "learning_rate": 3.44e-06, "loss": 0.0207, "step": 32801 }, { "epoch": 1.748787119475396, "grad_norm": 2.1057398319244385, "learning_rate": 3.4398000000000005e-06, "loss": 0.0406, "step": 32802 }, { "epoch": 1.7488404329050486, "grad_norm": 1.2134404182434082, "learning_rate": 3.4396e-06, "loss": 0.0143, "step": 32803 }, { "epoch": 1.7488937463347018, "grad_norm": 1.5835964679718018, "learning_rate": 3.4394000000000005e-06, "loss": 0.0292, "step": 32804 }, { "epoch": 1.7489470597643546, "grad_norm": 0.7227838635444641, "learning_rate": 3.4392000000000005e-06, "loss": 0.0069, "step": 32805 }, { "epoch": 1.7490003731940076, "grad_norm": 1.3913111686706543, "learning_rate": 3.439e-06, "loss": 0.0078, "step": 32806 }, { "epoch": 1.7490536866236606, "grad_norm": 1.6585055589675903, "learning_rate": 3.4388000000000004e-06, "loss": 0.0206, "step": 32807 }, { "epoch": 1.7491070000533133, "grad_norm": 1.223691701889038, "learning_rate": 3.4386000000000004e-06, "loss": 0.0139, "step": 32808 }, { "epoch": 1.7491603134829665, "grad_norm": 1.4510952234268188, "learning_rate": 3.4384e-06, "loss": 0.0203, "step": 32809 }, { "epoch": 1.7492136269126193, "grad_norm": 1.363365650177002, "learning_rate": 3.4382000000000004e-06, "loss": 0.0077, "step": 32810 }, { "epoch": 1.7492669403422723, "grad_norm": 1.207268476486206, "learning_rate": 3.4380000000000004e-06, "loss": 0.0179, "step": 32811 }, { "epoch": 1.7493202537719252, "grad_norm": 3.695181369781494, "learning_rate": 3.4378e-06, "loss": 0.0152, "step": 32812 }, { "epoch": 1.749373567201578, "grad_norm": 1.6752849817276, "learning_rate": 3.4376000000000003e-06, "loss": 0.0251, "step": 32813 }, { "epoch": 1.749426880631231, "grad_norm": 1.781278371810913, "learning_rate": 3.4374000000000003e-06, "loss": 0.0095, "step": 32814 }, { "epoch": 1.749480194060884, "grad_norm": 1.870163083076477, "learning_rate": 3.4372000000000007e-06, "loss": 0.0239, "step": 32815 }, { "epoch": 1.7495335074905367, "grad_norm": 0.9139050841331482, "learning_rate": 3.4370000000000003e-06, "loss": 0.0062, "step": 32816 }, { "epoch": 1.74958682092019, "grad_norm": 0.7019979953765869, "learning_rate": 3.4368000000000003e-06, "loss": 0.0126, "step": 32817 }, { "epoch": 1.7496401343498427, "grad_norm": 0.7587732076644897, "learning_rate": 3.4366000000000007e-06, "loss": 0.0166, "step": 32818 }, { "epoch": 1.7496934477794956, "grad_norm": 0.6336883902549744, "learning_rate": 3.4364000000000003e-06, "loss": 0.0051, "step": 32819 }, { "epoch": 1.7497467612091486, "grad_norm": 1.5387647151947021, "learning_rate": 3.4362000000000002e-06, "loss": 0.0139, "step": 32820 }, { "epoch": 1.7498000746388014, "grad_norm": 1.2638499736785889, "learning_rate": 3.4360000000000006e-06, "loss": 0.0411, "step": 32821 }, { "epoch": 1.7498533880684546, "grad_norm": 2.4270665645599365, "learning_rate": 3.4358e-06, "loss": 0.0182, "step": 32822 }, { "epoch": 1.7499067014981073, "grad_norm": 1.0137667655944824, "learning_rate": 3.4356e-06, "loss": 0.0069, "step": 32823 }, { "epoch": 1.7499600149277603, "grad_norm": 2.084817886352539, "learning_rate": 3.4354000000000006e-06, "loss": 0.0104, "step": 32824 }, { "epoch": 1.7500133283574133, "grad_norm": 1.3996609449386597, "learning_rate": 3.4352e-06, "loss": 0.0277, "step": 32825 }, { "epoch": 1.750066641787066, "grad_norm": 2.5238044261932373, "learning_rate": 3.4350000000000006e-06, "loss": 0.022, "step": 32826 }, { "epoch": 1.750119955216719, "grad_norm": 1.2309956550598145, "learning_rate": 3.4348000000000006e-06, "loss": 0.0146, "step": 32827 }, { "epoch": 1.750173268646372, "grad_norm": 2.2534635066986084, "learning_rate": 3.4346e-06, "loss": 0.0405, "step": 32828 }, { "epoch": 1.750226582076025, "grad_norm": 1.2810958623886108, "learning_rate": 3.4344000000000005e-06, "loss": 0.0074, "step": 32829 }, { "epoch": 1.750279895505678, "grad_norm": 1.6388157606124878, "learning_rate": 3.4342000000000005e-06, "loss": 0.0239, "step": 32830 }, { "epoch": 1.7503332089353307, "grad_norm": 1.894938588142395, "learning_rate": 3.434e-06, "loss": 0.0187, "step": 32831 }, { "epoch": 1.7503865223649837, "grad_norm": 1.5026816129684448, "learning_rate": 3.4338000000000005e-06, "loss": 0.0185, "step": 32832 }, { "epoch": 1.7504398357946367, "grad_norm": 0.672590970993042, "learning_rate": 3.4336000000000005e-06, "loss": 0.0046, "step": 32833 }, { "epoch": 1.7504931492242894, "grad_norm": 1.426755428314209, "learning_rate": 3.4334e-06, "loss": 0.0237, "step": 32834 }, { "epoch": 1.7505464626539426, "grad_norm": 2.320690155029297, "learning_rate": 3.4332000000000004e-06, "loss": 0.0277, "step": 32835 }, { "epoch": 1.7505997760835954, "grad_norm": 0.5872399806976318, "learning_rate": 3.4330000000000004e-06, "loss": 0.0059, "step": 32836 }, { "epoch": 1.7506530895132484, "grad_norm": 1.5703544616699219, "learning_rate": 3.4328e-06, "loss": 0.0322, "step": 32837 }, { "epoch": 1.7507064029429014, "grad_norm": 1.722877860069275, "learning_rate": 3.4326000000000004e-06, "loss": 0.0283, "step": 32838 }, { "epoch": 1.7507597163725541, "grad_norm": 1.0885921716690063, "learning_rate": 3.4324000000000004e-06, "loss": 0.0092, "step": 32839 }, { "epoch": 1.7508130298022073, "grad_norm": 1.5995997190475464, "learning_rate": 3.4322000000000004e-06, "loss": 0.0234, "step": 32840 }, { "epoch": 1.75086634323186, "grad_norm": 2.11879825592041, "learning_rate": 3.4320000000000003e-06, "loss": 0.0124, "step": 32841 }, { "epoch": 1.750919656661513, "grad_norm": 1.004421591758728, "learning_rate": 3.4318000000000003e-06, "loss": 0.0086, "step": 32842 }, { "epoch": 1.750972970091166, "grad_norm": 2.043168783187866, "learning_rate": 3.4316000000000003e-06, "loss": 0.0145, "step": 32843 }, { "epoch": 1.7510262835208188, "grad_norm": 1.4480788707733154, "learning_rate": 3.4314000000000003e-06, "loss": 0.0114, "step": 32844 }, { "epoch": 1.7510795969504718, "grad_norm": 3.4869916439056396, "learning_rate": 3.4312000000000003e-06, "loss": 0.0252, "step": 32845 }, { "epoch": 1.7511329103801248, "grad_norm": 2.2330803871154785, "learning_rate": 3.4310000000000003e-06, "loss": 0.0135, "step": 32846 }, { "epoch": 1.7511862238097777, "grad_norm": 5.825451374053955, "learning_rate": 3.4308000000000003e-06, "loss": 0.0375, "step": 32847 }, { "epoch": 1.7512395372394307, "grad_norm": 4.659270286560059, "learning_rate": 3.4306000000000002e-06, "loss": 0.0341, "step": 32848 }, { "epoch": 1.7512928506690835, "grad_norm": 1.4254831075668335, "learning_rate": 3.4304000000000002e-06, "loss": 0.0193, "step": 32849 }, { "epoch": 1.7513461640987364, "grad_norm": 2.699066400527954, "learning_rate": 3.4302e-06, "loss": 0.0187, "step": 32850 }, { "epoch": 1.7513994775283894, "grad_norm": 2.3808088302612305, "learning_rate": 3.4300000000000006e-06, "loss": 0.0411, "step": 32851 }, { "epoch": 1.7514527909580422, "grad_norm": 1.6394795179367065, "learning_rate": 3.4298e-06, "loss": 0.022, "step": 32852 }, { "epoch": 1.7515061043876954, "grad_norm": 1.0962790250778198, "learning_rate": 3.4296e-06, "loss": 0.0121, "step": 32853 }, { "epoch": 1.7515594178173481, "grad_norm": 2.417893171310425, "learning_rate": 3.4294000000000006e-06, "loss": 0.0243, "step": 32854 }, { "epoch": 1.7516127312470011, "grad_norm": 0.2523041069507599, "learning_rate": 3.4292e-06, "loss": 0.0022, "step": 32855 }, { "epoch": 1.751666044676654, "grad_norm": 2.035909652709961, "learning_rate": 3.429e-06, "loss": 0.0424, "step": 32856 }, { "epoch": 1.7517193581063069, "grad_norm": 1.3349888324737549, "learning_rate": 3.4288000000000005e-06, "loss": 0.0106, "step": 32857 }, { "epoch": 1.75177267153596, "grad_norm": 0.5956181287765503, "learning_rate": 3.4286e-06, "loss": 0.0045, "step": 32858 }, { "epoch": 1.7518259849656128, "grad_norm": 0.5607179403305054, "learning_rate": 3.4284e-06, "loss": 0.0046, "step": 32859 }, { "epoch": 1.7518792983952658, "grad_norm": 1.1929595470428467, "learning_rate": 3.4282000000000005e-06, "loss": 0.03, "step": 32860 }, { "epoch": 1.7519326118249188, "grad_norm": 0.6267541646957397, "learning_rate": 3.428e-06, "loss": 0.0043, "step": 32861 }, { "epoch": 1.7519859252545715, "grad_norm": 1.454083800315857, "learning_rate": 3.4278e-06, "loss": 0.0311, "step": 32862 }, { "epoch": 1.7520392386842245, "grad_norm": 2.493736505508423, "learning_rate": 3.4276000000000004e-06, "loss": 0.0092, "step": 32863 }, { "epoch": 1.7520925521138775, "grad_norm": 3.248020648956299, "learning_rate": 3.4274e-06, "loss": 0.0167, "step": 32864 }, { "epoch": 1.7521458655435305, "grad_norm": 2.388516426086426, "learning_rate": 3.4272000000000004e-06, "loss": 0.0525, "step": 32865 }, { "epoch": 1.7521991789731834, "grad_norm": 0.5986006855964661, "learning_rate": 3.4270000000000004e-06, "loss": 0.0046, "step": 32866 }, { "epoch": 1.7522524924028362, "grad_norm": 1.4306138753890991, "learning_rate": 3.4268e-06, "loss": 0.0071, "step": 32867 }, { "epoch": 1.7523058058324892, "grad_norm": 1.287064790725708, "learning_rate": 3.4266000000000004e-06, "loss": 0.0095, "step": 32868 }, { "epoch": 1.7523591192621422, "grad_norm": 1.489366888999939, "learning_rate": 3.4264000000000003e-06, "loss": 0.0451, "step": 32869 }, { "epoch": 1.752412432691795, "grad_norm": 1.4117392301559448, "learning_rate": 3.4262e-06, "loss": 0.0275, "step": 32870 }, { "epoch": 1.7524657461214481, "grad_norm": 0.5287721753120422, "learning_rate": 3.4260000000000003e-06, "loss": 0.0038, "step": 32871 }, { "epoch": 1.7525190595511009, "grad_norm": 3.416841745376587, "learning_rate": 3.4258000000000003e-06, "loss": 0.0436, "step": 32872 }, { "epoch": 1.7525723729807539, "grad_norm": 1.5540390014648438, "learning_rate": 3.4256e-06, "loss": 0.0069, "step": 32873 }, { "epoch": 1.7526256864104068, "grad_norm": 2.590881586074829, "learning_rate": 3.4254000000000003e-06, "loss": 0.0465, "step": 32874 }, { "epoch": 1.7526789998400596, "grad_norm": 1.1029374599456787, "learning_rate": 3.4252000000000003e-06, "loss": 0.0082, "step": 32875 }, { "epoch": 1.7527323132697128, "grad_norm": 1.9655152559280396, "learning_rate": 3.4250000000000007e-06, "loss": 0.0307, "step": 32876 }, { "epoch": 1.7527856266993656, "grad_norm": 0.9577469825744629, "learning_rate": 3.4248000000000002e-06, "loss": 0.0125, "step": 32877 }, { "epoch": 1.7528389401290185, "grad_norm": 1.1477068662643433, "learning_rate": 3.4246000000000002e-06, "loss": 0.0079, "step": 32878 }, { "epoch": 1.7528922535586715, "grad_norm": 2.337472677230835, "learning_rate": 3.4244000000000006e-06, "loss": 0.0389, "step": 32879 }, { "epoch": 1.7529455669883243, "grad_norm": 1.4191957712173462, "learning_rate": 3.4242e-06, "loss": 0.0195, "step": 32880 }, { "epoch": 1.7529988804179772, "grad_norm": 0.9597765207290649, "learning_rate": 3.424e-06, "loss": 0.0084, "step": 32881 }, { "epoch": 1.7530521938476302, "grad_norm": 1.6620898246765137, "learning_rate": 3.4238000000000006e-06, "loss": 0.0162, "step": 32882 }, { "epoch": 1.753105507277283, "grad_norm": 1.6183829307556152, "learning_rate": 3.4236e-06, "loss": 0.0133, "step": 32883 }, { "epoch": 1.7531588207069362, "grad_norm": 1.1831315755844116, "learning_rate": 3.4234e-06, "loss": 0.0192, "step": 32884 }, { "epoch": 1.753212134136589, "grad_norm": 1.1664447784423828, "learning_rate": 3.4232000000000005e-06, "loss": 0.0095, "step": 32885 }, { "epoch": 1.753265447566242, "grad_norm": 3.7451629638671875, "learning_rate": 3.423e-06, "loss": 0.0528, "step": 32886 }, { "epoch": 1.753318760995895, "grad_norm": 1.0441032648086548, "learning_rate": 3.4228e-06, "loss": 0.0079, "step": 32887 }, { "epoch": 1.7533720744255477, "grad_norm": 1.5706565380096436, "learning_rate": 3.4226000000000005e-06, "loss": 0.0212, "step": 32888 }, { "epoch": 1.7534253878552009, "grad_norm": 0.5927859544754028, "learning_rate": 3.4224e-06, "loss": 0.0053, "step": 32889 }, { "epoch": 1.7534787012848536, "grad_norm": 2.8581931591033936, "learning_rate": 3.4222000000000005e-06, "loss": 0.0236, "step": 32890 }, { "epoch": 1.7535320147145066, "grad_norm": 0.740848958492279, "learning_rate": 3.4220000000000004e-06, "loss": 0.0059, "step": 32891 }, { "epoch": 1.7535853281441596, "grad_norm": 0.7308562397956848, "learning_rate": 3.4218e-06, "loss": 0.0041, "step": 32892 }, { "epoch": 1.7536386415738123, "grad_norm": 2.1852426528930664, "learning_rate": 3.4216000000000004e-06, "loss": 0.0173, "step": 32893 }, { "epoch": 1.7536919550034655, "grad_norm": 1.3574910163879395, "learning_rate": 3.4214000000000004e-06, "loss": 0.0066, "step": 32894 }, { "epoch": 1.7537452684331183, "grad_norm": 1.7200541496276855, "learning_rate": 3.4212e-06, "loss": 0.0101, "step": 32895 }, { "epoch": 1.7537985818627713, "grad_norm": 1.1978057622909546, "learning_rate": 3.4210000000000004e-06, "loss": 0.0067, "step": 32896 }, { "epoch": 1.7538518952924242, "grad_norm": 1.44558584690094, "learning_rate": 3.4208000000000004e-06, "loss": 0.0141, "step": 32897 }, { "epoch": 1.753905208722077, "grad_norm": 1.827834129333496, "learning_rate": 3.4206e-06, "loss": 0.0093, "step": 32898 }, { "epoch": 1.75395852215173, "grad_norm": 0.8844761252403259, "learning_rate": 3.4204000000000003e-06, "loss": 0.0041, "step": 32899 }, { "epoch": 1.754011835581383, "grad_norm": 0.9429665207862854, "learning_rate": 3.4202000000000003e-06, "loss": 0.0052, "step": 32900 }, { "epoch": 1.7540651490110357, "grad_norm": 1.644413948059082, "learning_rate": 3.4200000000000007e-06, "loss": 0.0293, "step": 32901 }, { "epoch": 1.754118462440689, "grad_norm": 1.774314045906067, "learning_rate": 3.4198000000000003e-06, "loss": 0.0114, "step": 32902 }, { "epoch": 1.7541717758703417, "grad_norm": 1.7432582378387451, "learning_rate": 3.4196000000000003e-06, "loss": 0.0133, "step": 32903 }, { "epoch": 1.7542250892999947, "grad_norm": 1.2352099418640137, "learning_rate": 3.4194000000000007e-06, "loss": 0.0164, "step": 32904 }, { "epoch": 1.7542784027296476, "grad_norm": 0.30842941999435425, "learning_rate": 3.4192000000000002e-06, "loss": 0.0022, "step": 32905 }, { "epoch": 1.7543317161593004, "grad_norm": 0.3645654320716858, "learning_rate": 3.4190000000000002e-06, "loss": 0.003, "step": 32906 }, { "epoch": 1.7543850295889536, "grad_norm": 2.304368019104004, "learning_rate": 3.4188000000000006e-06, "loss": 0.0264, "step": 32907 }, { "epoch": 1.7544383430186063, "grad_norm": 1.9441715478897095, "learning_rate": 3.4186e-06, "loss": 0.0299, "step": 32908 }, { "epoch": 1.7544916564482593, "grad_norm": 1.3003342151641846, "learning_rate": 3.4184e-06, "loss": 0.0177, "step": 32909 }, { "epoch": 1.7545449698779123, "grad_norm": 1.1312612295150757, "learning_rate": 3.4182000000000006e-06, "loss": 0.0057, "step": 32910 }, { "epoch": 1.754598283307565, "grad_norm": 1.417527198791504, "learning_rate": 3.418e-06, "loss": 0.0059, "step": 32911 }, { "epoch": 1.7546515967372183, "grad_norm": 2.5066730976104736, "learning_rate": 3.4178e-06, "loss": 0.012, "step": 32912 }, { "epoch": 1.754704910166871, "grad_norm": 1.0722556114196777, "learning_rate": 3.4176000000000005e-06, "loss": 0.0088, "step": 32913 }, { "epoch": 1.754758223596524, "grad_norm": 0.8971322178840637, "learning_rate": 3.4174e-06, "loss": 0.0062, "step": 32914 }, { "epoch": 1.754811537026177, "grad_norm": 0.9498016834259033, "learning_rate": 3.4172000000000005e-06, "loss": 0.0055, "step": 32915 }, { "epoch": 1.7548648504558297, "grad_norm": 1.7792962789535522, "learning_rate": 3.4170000000000005e-06, "loss": 0.0244, "step": 32916 }, { "epoch": 1.7549181638854827, "grad_norm": 1.3295843601226807, "learning_rate": 3.4168e-06, "loss": 0.0181, "step": 32917 }, { "epoch": 1.7549714773151357, "grad_norm": 1.3007476329803467, "learning_rate": 3.4166000000000005e-06, "loss": 0.0057, "step": 32918 }, { "epoch": 1.7550247907447885, "grad_norm": 1.6301004886627197, "learning_rate": 3.4164000000000004e-06, "loss": 0.0196, "step": 32919 }, { "epoch": 1.7550781041744417, "grad_norm": 1.7506519556045532, "learning_rate": 3.4162e-06, "loss": 0.0159, "step": 32920 }, { "epoch": 1.7551314176040944, "grad_norm": 4.654012203216553, "learning_rate": 3.4160000000000004e-06, "loss": 0.0599, "step": 32921 }, { "epoch": 1.7551847310337474, "grad_norm": 1.5338646173477173, "learning_rate": 3.4158000000000004e-06, "loss": 0.0145, "step": 32922 }, { "epoch": 1.7552380444634004, "grad_norm": 1.6279387474060059, "learning_rate": 3.4156e-06, "loss": 0.0204, "step": 32923 }, { "epoch": 1.7552913578930531, "grad_norm": 1.1979504823684692, "learning_rate": 3.4154000000000004e-06, "loss": 0.0082, "step": 32924 }, { "epoch": 1.7553446713227063, "grad_norm": 3.4464187622070312, "learning_rate": 3.4152000000000004e-06, "loss": 0.0626, "step": 32925 }, { "epoch": 1.755397984752359, "grad_norm": 1.0065058469772339, "learning_rate": 3.4150000000000003e-06, "loss": 0.005, "step": 32926 }, { "epoch": 1.755451298182012, "grad_norm": 2.094298839569092, "learning_rate": 3.4148000000000003e-06, "loss": 0.0293, "step": 32927 }, { "epoch": 1.755504611611665, "grad_norm": 1.6861820220947266, "learning_rate": 3.4146000000000003e-06, "loss": 0.0318, "step": 32928 }, { "epoch": 1.7555579250413178, "grad_norm": 2.1256356239318848, "learning_rate": 3.4144000000000003e-06, "loss": 0.0145, "step": 32929 }, { "epoch": 1.755611238470971, "grad_norm": 1.068746566772461, "learning_rate": 3.4142000000000003e-06, "loss": 0.0239, "step": 32930 }, { "epoch": 1.7556645519006238, "grad_norm": 0.9204761981964111, "learning_rate": 3.4140000000000003e-06, "loss": 0.0063, "step": 32931 }, { "epoch": 1.7557178653302767, "grad_norm": 1.9024053812026978, "learning_rate": 3.4138000000000002e-06, "loss": 0.0252, "step": 32932 }, { "epoch": 1.7557711787599297, "grad_norm": 2.0659613609313965, "learning_rate": 3.4136000000000002e-06, "loss": 0.0222, "step": 32933 }, { "epoch": 1.7558244921895825, "grad_norm": 1.7761236429214478, "learning_rate": 3.4134000000000002e-06, "loss": 0.0106, "step": 32934 }, { "epoch": 1.7558778056192355, "grad_norm": 1.4707554578781128, "learning_rate": 3.4132e-06, "loss": 0.0094, "step": 32935 }, { "epoch": 1.7559311190488884, "grad_norm": 1.5754585266113281, "learning_rate": 3.413e-06, "loss": 0.0089, "step": 32936 }, { "epoch": 1.7559844324785412, "grad_norm": 2.232635498046875, "learning_rate": 3.4128e-06, "loss": 0.0234, "step": 32937 }, { "epoch": 1.7560377459081944, "grad_norm": 2.7939276695251465, "learning_rate": 3.4126e-06, "loss": 0.0154, "step": 32938 }, { "epoch": 1.7560910593378471, "grad_norm": 3.146348237991333, "learning_rate": 3.4124e-06, "loss": 0.0304, "step": 32939 }, { "epoch": 1.7561443727675001, "grad_norm": 0.7229036092758179, "learning_rate": 3.4122000000000006e-06, "loss": 0.0037, "step": 32940 }, { "epoch": 1.756197686197153, "grad_norm": 3.6047003269195557, "learning_rate": 3.412e-06, "loss": 0.0197, "step": 32941 }, { "epoch": 1.7562509996268059, "grad_norm": 1.854400873184204, "learning_rate": 3.4118e-06, "loss": 0.0235, "step": 32942 }, { "epoch": 1.756304313056459, "grad_norm": 1.441298246383667, "learning_rate": 3.4116000000000005e-06, "loss": 0.0115, "step": 32943 }, { "epoch": 1.7563576264861118, "grad_norm": 0.5980070233345032, "learning_rate": 3.4114e-06, "loss": 0.0041, "step": 32944 }, { "epoch": 1.7564109399157648, "grad_norm": 1.8020457029342651, "learning_rate": 3.4112e-06, "loss": 0.0368, "step": 32945 }, { "epoch": 1.7564642533454178, "grad_norm": 1.6780805587768555, "learning_rate": 3.4110000000000005e-06, "loss": 0.019, "step": 32946 }, { "epoch": 1.7565175667750705, "grad_norm": 2.2090635299682617, "learning_rate": 3.4108e-06, "loss": 0.0151, "step": 32947 }, { "epoch": 1.7565708802047235, "grad_norm": 1.0442297458648682, "learning_rate": 3.4106e-06, "loss": 0.0245, "step": 32948 }, { "epoch": 1.7566241936343765, "grad_norm": 1.0989032983779907, "learning_rate": 3.4104000000000004e-06, "loss": 0.0068, "step": 32949 }, { "epoch": 1.7566775070640295, "grad_norm": 1.0794970989227295, "learning_rate": 3.4102e-06, "loss": 0.0068, "step": 32950 }, { "epoch": 1.7567308204936825, "grad_norm": 0.6432391405105591, "learning_rate": 3.4100000000000004e-06, "loss": 0.0035, "step": 32951 }, { "epoch": 1.7567841339233352, "grad_norm": 2.2315165996551514, "learning_rate": 3.4098000000000004e-06, "loss": 0.0158, "step": 32952 }, { "epoch": 1.7568374473529882, "grad_norm": 1.782670021057129, "learning_rate": 3.4096e-06, "loss": 0.0226, "step": 32953 }, { "epoch": 1.7568907607826412, "grad_norm": 2.7923359870910645, "learning_rate": 3.4094000000000003e-06, "loss": 0.024, "step": 32954 }, { "epoch": 1.756944074212294, "grad_norm": 0.579059898853302, "learning_rate": 3.4092000000000003e-06, "loss": 0.0034, "step": 32955 }, { "epoch": 1.7569973876419471, "grad_norm": 0.47857537865638733, "learning_rate": 3.409e-06, "loss": 0.0037, "step": 32956 }, { "epoch": 1.7570507010715999, "grad_norm": 2.596865653991699, "learning_rate": 3.4088000000000003e-06, "loss": 0.0272, "step": 32957 }, { "epoch": 1.7571040145012529, "grad_norm": 0.9058281183242798, "learning_rate": 3.4086000000000003e-06, "loss": 0.019, "step": 32958 }, { "epoch": 1.7571573279309058, "grad_norm": 1.5295664072036743, "learning_rate": 3.4084e-06, "loss": 0.0329, "step": 32959 }, { "epoch": 1.7572106413605586, "grad_norm": 1.8049633502960205, "learning_rate": 3.4082000000000003e-06, "loss": 0.0163, "step": 32960 }, { "epoch": 1.7572639547902118, "grad_norm": 1.524210810661316, "learning_rate": 3.4080000000000002e-06, "loss": 0.0094, "step": 32961 }, { "epoch": 1.7573172682198646, "grad_norm": 0.9478982090950012, "learning_rate": 3.4078000000000006e-06, "loss": 0.0062, "step": 32962 }, { "epoch": 1.7573705816495175, "grad_norm": 2.5209693908691406, "learning_rate": 3.4076e-06, "loss": 0.0342, "step": 32963 }, { "epoch": 1.7574238950791705, "grad_norm": 2.326477527618408, "learning_rate": 3.4074e-06, "loss": 0.0138, "step": 32964 }, { "epoch": 1.7574772085088233, "grad_norm": 3.1795248985290527, "learning_rate": 3.4072000000000006e-06, "loss": 0.0474, "step": 32965 }, { "epoch": 1.7575305219384763, "grad_norm": 0.41398757696151733, "learning_rate": 3.407e-06, "loss": 0.003, "step": 32966 }, { "epoch": 1.7575838353681292, "grad_norm": 0.4917050898075104, "learning_rate": 3.4068e-06, "loss": 0.0045, "step": 32967 }, { "epoch": 1.7576371487977822, "grad_norm": 2.240135908126831, "learning_rate": 3.4066000000000006e-06, "loss": 0.0428, "step": 32968 }, { "epoch": 1.7576904622274352, "grad_norm": 1.2329070568084717, "learning_rate": 3.4064e-06, "loss": 0.0113, "step": 32969 }, { "epoch": 1.757743775657088, "grad_norm": 1.732251763343811, "learning_rate": 3.4062e-06, "loss": 0.0089, "step": 32970 }, { "epoch": 1.757797089086741, "grad_norm": 1.0272654294967651, "learning_rate": 3.4060000000000005e-06, "loss": 0.0053, "step": 32971 }, { "epoch": 1.757850402516394, "grad_norm": 3.655775785446167, "learning_rate": 3.4058e-06, "loss": 0.041, "step": 32972 }, { "epoch": 1.7579037159460467, "grad_norm": 1.8496594429016113, "learning_rate": 3.4056e-06, "loss": 0.0087, "step": 32973 }, { "epoch": 1.7579570293756999, "grad_norm": 2.994811773300171, "learning_rate": 3.4054000000000005e-06, "loss": 0.0338, "step": 32974 }, { "epoch": 1.7580103428053526, "grad_norm": 0.4418926239013672, "learning_rate": 3.4052e-06, "loss": 0.0033, "step": 32975 }, { "epoch": 1.7580636562350056, "grad_norm": 1.8731054067611694, "learning_rate": 3.4050000000000004e-06, "loss": 0.0103, "step": 32976 }, { "epoch": 1.7581169696646586, "grad_norm": 5.864201068878174, "learning_rate": 3.4048000000000004e-06, "loss": 0.0215, "step": 32977 }, { "epoch": 1.7581702830943113, "grad_norm": 1.655397891998291, "learning_rate": 3.4046e-06, "loss": 0.0078, "step": 32978 }, { "epoch": 1.7582235965239645, "grad_norm": 1.3733725547790527, "learning_rate": 3.4044000000000004e-06, "loss": 0.0267, "step": 32979 }, { "epoch": 1.7582769099536173, "grad_norm": 1.3916635513305664, "learning_rate": 3.4042000000000004e-06, "loss": 0.0079, "step": 32980 }, { "epoch": 1.7583302233832703, "grad_norm": 1.460171103477478, "learning_rate": 3.404e-06, "loss": 0.0213, "step": 32981 }, { "epoch": 1.7583835368129233, "grad_norm": 1.623528003692627, "learning_rate": 3.4038000000000003e-06, "loss": 0.0201, "step": 32982 }, { "epoch": 1.758436850242576, "grad_norm": 1.5869988203048706, "learning_rate": 3.4036000000000003e-06, "loss": 0.0099, "step": 32983 }, { "epoch": 1.758490163672229, "grad_norm": 2.17596435546875, "learning_rate": 3.4034e-06, "loss": 0.0188, "step": 32984 }, { "epoch": 1.758543477101882, "grad_norm": 2.4244768619537354, "learning_rate": 3.4032000000000003e-06, "loss": 0.0192, "step": 32985 }, { "epoch": 1.758596790531535, "grad_norm": 3.3170814514160156, "learning_rate": 3.4030000000000003e-06, "loss": 0.0283, "step": 32986 }, { "epoch": 1.758650103961188, "grad_norm": 1.4970998764038086, "learning_rate": 3.4028000000000007e-06, "loss": 0.0149, "step": 32987 }, { "epoch": 1.7587034173908407, "grad_norm": 1.8642247915267944, "learning_rate": 3.4026000000000003e-06, "loss": 0.0186, "step": 32988 }, { "epoch": 1.7587567308204937, "grad_norm": 1.3452619314193726, "learning_rate": 3.4024000000000002e-06, "loss": 0.0206, "step": 32989 }, { "epoch": 1.7588100442501466, "grad_norm": 2.0888831615448, "learning_rate": 3.4022000000000006e-06, "loss": 0.026, "step": 32990 }, { "epoch": 1.7588633576797994, "grad_norm": 2.2318170070648193, "learning_rate": 3.402e-06, "loss": 0.0234, "step": 32991 }, { "epoch": 1.7589166711094526, "grad_norm": 1.8620717525482178, "learning_rate": 3.4018e-06, "loss": 0.0149, "step": 32992 }, { "epoch": 1.7589699845391054, "grad_norm": 0.8603575229644775, "learning_rate": 3.4016000000000006e-06, "loss": 0.0045, "step": 32993 }, { "epoch": 1.7590232979687583, "grad_norm": 1.4803146123886108, "learning_rate": 3.4014e-06, "loss": 0.0109, "step": 32994 }, { "epoch": 1.7590766113984113, "grad_norm": 2.899784803390503, "learning_rate": 3.4012e-06, "loss": 0.0199, "step": 32995 }, { "epoch": 1.759129924828064, "grad_norm": 2.261481523513794, "learning_rate": 3.4010000000000006e-06, "loss": 0.0261, "step": 32996 }, { "epoch": 1.7591832382577173, "grad_norm": 0.8971364498138428, "learning_rate": 3.4008e-06, "loss": 0.0163, "step": 32997 }, { "epoch": 1.75923655168737, "grad_norm": 2.9139609336853027, "learning_rate": 3.4006e-06, "loss": 0.0473, "step": 32998 }, { "epoch": 1.759289865117023, "grad_norm": 1.5047348737716675, "learning_rate": 3.4004000000000005e-06, "loss": 0.0409, "step": 32999 }, { "epoch": 1.759343178546676, "grad_norm": 0.7607646584510803, "learning_rate": 3.4002e-06, "loss": 0.0051, "step": 33000 }, { "epoch": 1.7593964919763287, "grad_norm": 2.889963150024414, "learning_rate": 3.4000000000000005e-06, "loss": 0.0178, "step": 33001 }, { "epoch": 1.7594498054059817, "grad_norm": 1.4953614473342896, "learning_rate": 3.3998000000000005e-06, "loss": 0.0156, "step": 33002 }, { "epoch": 1.7595031188356347, "grad_norm": 1.0235826969146729, "learning_rate": 3.3996e-06, "loss": 0.0055, "step": 33003 }, { "epoch": 1.7595564322652875, "grad_norm": 2.6534528732299805, "learning_rate": 3.3994000000000004e-06, "loss": 0.0104, "step": 33004 }, { "epoch": 1.7596097456949407, "grad_norm": 1.4771828651428223, "learning_rate": 3.3992000000000004e-06, "loss": 0.0095, "step": 33005 }, { "epoch": 1.7596630591245934, "grad_norm": 0.8062801957130432, "learning_rate": 3.399e-06, "loss": 0.0066, "step": 33006 }, { "epoch": 1.7597163725542464, "grad_norm": 2.2737412452697754, "learning_rate": 3.3988000000000004e-06, "loss": 0.013, "step": 33007 }, { "epoch": 1.7597696859838994, "grad_norm": 1.967447280883789, "learning_rate": 3.3986000000000004e-06, "loss": 0.0254, "step": 33008 }, { "epoch": 1.7598229994135521, "grad_norm": 0.9692703485488892, "learning_rate": 3.3984e-06, "loss": 0.0052, "step": 33009 }, { "epoch": 1.7598763128432053, "grad_norm": 1.9909850358963013, "learning_rate": 3.3982000000000003e-06, "loss": 0.0155, "step": 33010 }, { "epoch": 1.759929626272858, "grad_norm": 2.8205413818359375, "learning_rate": 3.3980000000000003e-06, "loss": 0.0336, "step": 33011 }, { "epoch": 1.759982939702511, "grad_norm": 0.5270527005195618, "learning_rate": 3.3978000000000007e-06, "loss": 0.0032, "step": 33012 }, { "epoch": 1.760036253132164, "grad_norm": 0.7256880402565002, "learning_rate": 3.3976000000000003e-06, "loss": 0.0035, "step": 33013 }, { "epoch": 1.7600895665618168, "grad_norm": 0.5065855979919434, "learning_rate": 3.3974000000000003e-06, "loss": 0.0031, "step": 33014 }, { "epoch": 1.76014287999147, "grad_norm": 1.8247909545898438, "learning_rate": 3.3972000000000007e-06, "loss": 0.0102, "step": 33015 }, { "epoch": 1.7601961934211228, "grad_norm": 1.457410216331482, "learning_rate": 3.3970000000000003e-06, "loss": 0.0082, "step": 33016 }, { "epoch": 1.7602495068507757, "grad_norm": 2.142312526702881, "learning_rate": 3.3968000000000002e-06, "loss": 0.0555, "step": 33017 }, { "epoch": 1.7603028202804287, "grad_norm": 2.313410997390747, "learning_rate": 3.3966000000000006e-06, "loss": 0.0423, "step": 33018 }, { "epoch": 1.7603561337100815, "grad_norm": 1.9363362789154053, "learning_rate": 3.3964e-06, "loss": 0.0202, "step": 33019 }, { "epoch": 1.7604094471397345, "grad_norm": 1.1686491966247559, "learning_rate": 3.3962e-06, "loss": 0.0065, "step": 33020 }, { "epoch": 1.7604627605693874, "grad_norm": 1.2906877994537354, "learning_rate": 3.3960000000000006e-06, "loss": 0.0478, "step": 33021 }, { "epoch": 1.7605160739990402, "grad_norm": 2.6522581577301025, "learning_rate": 3.3958e-06, "loss": 0.0249, "step": 33022 }, { "epoch": 1.7605693874286934, "grad_norm": 0.6794684529304504, "learning_rate": 3.3956e-06, "loss": 0.0084, "step": 33023 }, { "epoch": 1.7606227008583462, "grad_norm": 2.6270270347595215, "learning_rate": 3.3954000000000006e-06, "loss": 0.0411, "step": 33024 }, { "epoch": 1.7606760142879991, "grad_norm": 3.5678985118865967, "learning_rate": 3.3952e-06, "loss": 0.0479, "step": 33025 }, { "epoch": 1.7607293277176521, "grad_norm": 2.3249588012695312, "learning_rate": 3.3950000000000005e-06, "loss": 0.0329, "step": 33026 }, { "epoch": 1.7607826411473049, "grad_norm": 1.2212234735488892, "learning_rate": 3.3948000000000005e-06, "loss": 0.0122, "step": 33027 }, { "epoch": 1.760835954576958, "grad_norm": 1.2129664421081543, "learning_rate": 3.3946e-06, "loss": 0.0117, "step": 33028 }, { "epoch": 1.7608892680066108, "grad_norm": 2.088536262512207, "learning_rate": 3.3944000000000005e-06, "loss": 0.0169, "step": 33029 }, { "epoch": 1.7609425814362638, "grad_norm": 2.5005691051483154, "learning_rate": 3.3942000000000005e-06, "loss": 0.0194, "step": 33030 }, { "epoch": 1.7609958948659168, "grad_norm": 2.079399824142456, "learning_rate": 3.394e-06, "loss": 0.011, "step": 33031 }, { "epoch": 1.7610492082955695, "grad_norm": 1.5582026243209839, "learning_rate": 3.3938000000000004e-06, "loss": 0.0102, "step": 33032 }, { "epoch": 1.7611025217252227, "grad_norm": 8.106145858764648, "learning_rate": 3.3936000000000004e-06, "loss": 0.0157, "step": 33033 }, { "epoch": 1.7611558351548755, "grad_norm": 0.8715832829475403, "learning_rate": 3.3934e-06, "loss": 0.0067, "step": 33034 }, { "epoch": 1.7612091485845285, "grad_norm": 1.3837615251541138, "learning_rate": 3.3932000000000004e-06, "loss": 0.0231, "step": 33035 }, { "epoch": 1.7612624620141815, "grad_norm": 2.642312526702881, "learning_rate": 3.3930000000000004e-06, "loss": 0.0373, "step": 33036 }, { "epoch": 1.7613157754438342, "grad_norm": 1.842003345489502, "learning_rate": 3.3928000000000004e-06, "loss": 0.0117, "step": 33037 }, { "epoch": 1.7613690888734872, "grad_norm": 2.0234463214874268, "learning_rate": 3.3926000000000003e-06, "loss": 0.0137, "step": 33038 }, { "epoch": 1.7614224023031402, "grad_norm": 1.7760554552078247, "learning_rate": 3.3924000000000003e-06, "loss": 0.0158, "step": 33039 }, { "epoch": 1.761475715732793, "grad_norm": 1.2005908489227295, "learning_rate": 3.3922000000000003e-06, "loss": 0.0245, "step": 33040 }, { "epoch": 1.7615290291624461, "grad_norm": 1.7969486713409424, "learning_rate": 3.3920000000000003e-06, "loss": 0.0286, "step": 33041 }, { "epoch": 1.761582342592099, "grad_norm": 1.3171608448028564, "learning_rate": 3.3918000000000003e-06, "loss": 0.0247, "step": 33042 }, { "epoch": 1.7616356560217519, "grad_norm": 1.461471676826477, "learning_rate": 3.3916000000000003e-06, "loss": 0.0142, "step": 33043 }, { "epoch": 1.7616889694514049, "grad_norm": 3.8309950828552246, "learning_rate": 3.3914000000000003e-06, "loss": 0.0589, "step": 33044 }, { "epoch": 1.7617422828810576, "grad_norm": 1.8447980880737305, "learning_rate": 3.3912000000000002e-06, "loss": 0.0222, "step": 33045 }, { "epoch": 1.7617955963107108, "grad_norm": 2.7630398273468018, "learning_rate": 3.3910000000000002e-06, "loss": 0.056, "step": 33046 }, { "epoch": 1.7618489097403636, "grad_norm": 1.6201331615447998, "learning_rate": 3.3908000000000002e-06, "loss": 0.0272, "step": 33047 }, { "epoch": 1.7619022231700165, "grad_norm": 3.159301996231079, "learning_rate": 3.3906e-06, "loss": 0.0502, "step": 33048 }, { "epoch": 1.7619555365996695, "grad_norm": 1.9607497453689575, "learning_rate": 3.3904e-06, "loss": 0.0159, "step": 33049 }, { "epoch": 1.7620088500293223, "grad_norm": 1.4466779232025146, "learning_rate": 3.3902e-06, "loss": 0.0081, "step": 33050 }, { "epoch": 1.7620621634589753, "grad_norm": 1.0184638500213623, "learning_rate": 3.3900000000000006e-06, "loss": 0.0085, "step": 33051 }, { "epoch": 1.7621154768886282, "grad_norm": 3.0249688625335693, "learning_rate": 3.3898e-06, "loss": 0.0306, "step": 33052 }, { "epoch": 1.7621687903182812, "grad_norm": 0.915023922920227, "learning_rate": 3.3896e-06, "loss": 0.013, "step": 33053 }, { "epoch": 1.7622221037479342, "grad_norm": 1.9002482891082764, "learning_rate": 3.3894000000000005e-06, "loss": 0.0126, "step": 33054 }, { "epoch": 1.762275417177587, "grad_norm": 2.0414319038391113, "learning_rate": 3.3892e-06, "loss": 0.0291, "step": 33055 }, { "epoch": 1.76232873060724, "grad_norm": 3.0401744842529297, "learning_rate": 3.389e-06, "loss": 0.0173, "step": 33056 }, { "epoch": 1.762382044036893, "grad_norm": 2.484367609024048, "learning_rate": 3.3888000000000005e-06, "loss": 0.0251, "step": 33057 }, { "epoch": 1.7624353574665457, "grad_norm": 2.749631881713867, "learning_rate": 3.3886e-06, "loss": 0.0188, "step": 33058 }, { "epoch": 1.7624886708961989, "grad_norm": 2.0015666484832764, "learning_rate": 3.3884e-06, "loss": 0.0369, "step": 33059 }, { "epoch": 1.7625419843258516, "grad_norm": 1.3558909893035889, "learning_rate": 3.3882000000000004e-06, "loss": 0.01, "step": 33060 }, { "epoch": 1.7625952977555046, "grad_norm": 2.734571933746338, "learning_rate": 3.388e-06, "loss": 0.0537, "step": 33061 }, { "epoch": 1.7626486111851576, "grad_norm": 0.8138459920883179, "learning_rate": 3.3878000000000004e-06, "loss": 0.0174, "step": 33062 }, { "epoch": 1.7627019246148103, "grad_norm": 1.9434871673583984, "learning_rate": 3.3876000000000004e-06, "loss": 0.019, "step": 33063 }, { "epoch": 1.7627552380444635, "grad_norm": 1.5566068887710571, "learning_rate": 3.3874e-06, "loss": 0.0074, "step": 33064 }, { "epoch": 1.7628085514741163, "grad_norm": 1.936570167541504, "learning_rate": 3.3872000000000004e-06, "loss": 0.0249, "step": 33065 }, { "epoch": 1.7628618649037693, "grad_norm": 1.5207887887954712, "learning_rate": 3.3870000000000004e-06, "loss": 0.028, "step": 33066 }, { "epoch": 1.7629151783334223, "grad_norm": 2.185941219329834, "learning_rate": 3.3868e-06, "loss": 0.0414, "step": 33067 }, { "epoch": 1.762968491763075, "grad_norm": 0.85476154088974, "learning_rate": 3.3866000000000003e-06, "loss": 0.0068, "step": 33068 }, { "epoch": 1.763021805192728, "grad_norm": 0.9610926508903503, "learning_rate": 3.3864000000000003e-06, "loss": 0.0115, "step": 33069 }, { "epoch": 1.763075118622381, "grad_norm": 1.4292621612548828, "learning_rate": 3.3862e-06, "loss": 0.0141, "step": 33070 }, { "epoch": 1.763128432052034, "grad_norm": 2.900163173675537, "learning_rate": 3.3860000000000003e-06, "loss": 0.0302, "step": 33071 }, { "epoch": 1.763181745481687, "grad_norm": 2.0418508052825928, "learning_rate": 3.3858000000000003e-06, "loss": 0.0273, "step": 33072 }, { "epoch": 1.7632350589113397, "grad_norm": 2.1697933673858643, "learning_rate": 3.3856000000000007e-06, "loss": 0.0191, "step": 33073 }, { "epoch": 1.7632883723409927, "grad_norm": 2.2486472129821777, "learning_rate": 3.3854000000000002e-06, "loss": 0.0239, "step": 33074 }, { "epoch": 1.7633416857706457, "grad_norm": 1.7007414102554321, "learning_rate": 3.3852000000000002e-06, "loss": 0.0151, "step": 33075 }, { "epoch": 1.7633949992002984, "grad_norm": 1.8612858057022095, "learning_rate": 3.3850000000000006e-06, "loss": 0.0366, "step": 33076 }, { "epoch": 1.7634483126299516, "grad_norm": 0.9552936553955078, "learning_rate": 3.3848e-06, "loss": 0.0186, "step": 33077 }, { "epoch": 1.7635016260596044, "grad_norm": 2.3211004734039307, "learning_rate": 3.3846e-06, "loss": 0.0209, "step": 33078 }, { "epoch": 1.7635549394892573, "grad_norm": 1.3413424491882324, "learning_rate": 3.3844000000000006e-06, "loss": 0.033, "step": 33079 }, { "epoch": 1.7636082529189103, "grad_norm": 1.6003148555755615, "learning_rate": 3.3842e-06, "loss": 0.0158, "step": 33080 }, { "epoch": 1.763661566348563, "grad_norm": 1.3958079814910889, "learning_rate": 3.384e-06, "loss": 0.0169, "step": 33081 }, { "epoch": 1.7637148797782163, "grad_norm": 0.760094165802002, "learning_rate": 3.3838000000000005e-06, "loss": 0.0058, "step": 33082 }, { "epoch": 1.763768193207869, "grad_norm": 2.3126862049102783, "learning_rate": 3.3836e-06, "loss": 0.0201, "step": 33083 }, { "epoch": 1.763821506637522, "grad_norm": 1.5862799882888794, "learning_rate": 3.3834e-06, "loss": 0.0202, "step": 33084 }, { "epoch": 1.763874820067175, "grad_norm": 2.5115737915039062, "learning_rate": 3.3832000000000005e-06, "loss": 0.0273, "step": 33085 }, { "epoch": 1.7639281334968278, "grad_norm": 1.709045171737671, "learning_rate": 3.383e-06, "loss": 0.0274, "step": 33086 }, { "epoch": 1.7639814469264807, "grad_norm": 0.7191128730773926, "learning_rate": 3.3828000000000005e-06, "loss": 0.0184, "step": 33087 }, { "epoch": 1.7640347603561337, "grad_norm": 3.6528031826019287, "learning_rate": 3.3826000000000004e-06, "loss": 0.0825, "step": 33088 }, { "epoch": 1.7640880737857867, "grad_norm": 2.9257705211639404, "learning_rate": 3.3824e-06, "loss": 0.0304, "step": 33089 }, { "epoch": 1.7641413872154397, "grad_norm": 1.9910014867782593, "learning_rate": 3.3822000000000004e-06, "loss": 0.0149, "step": 33090 }, { "epoch": 1.7641947006450924, "grad_norm": 2.751960039138794, "learning_rate": 3.3820000000000004e-06, "loss": 0.022, "step": 33091 }, { "epoch": 1.7642480140747454, "grad_norm": 0.6148139238357544, "learning_rate": 3.3818e-06, "loss": 0.0064, "step": 33092 }, { "epoch": 1.7643013275043984, "grad_norm": 1.7834362983703613, "learning_rate": 3.3816000000000004e-06, "loss": 0.0136, "step": 33093 }, { "epoch": 1.7643546409340511, "grad_norm": 3.2125918865203857, "learning_rate": 3.3814000000000004e-06, "loss": 0.036, "step": 33094 }, { "epoch": 1.7644079543637043, "grad_norm": 0.4881817400455475, "learning_rate": 3.3812e-06, "loss": 0.0052, "step": 33095 }, { "epoch": 1.764461267793357, "grad_norm": 2.1958978176116943, "learning_rate": 3.3810000000000003e-06, "loss": 0.023, "step": 33096 }, { "epoch": 1.76451458122301, "grad_norm": 1.1938365697860718, "learning_rate": 3.3808000000000003e-06, "loss": 0.0176, "step": 33097 }, { "epoch": 1.764567894652663, "grad_norm": 1.1254167556762695, "learning_rate": 3.3806000000000007e-06, "loss": 0.0203, "step": 33098 }, { "epoch": 1.7646212080823158, "grad_norm": 1.8148270845413208, "learning_rate": 3.3804000000000003e-06, "loss": 0.0191, "step": 33099 }, { "epoch": 1.764674521511969, "grad_norm": 1.4653106927871704, "learning_rate": 3.3802000000000003e-06, "loss": 0.0216, "step": 33100 }, { "epoch": 1.7647278349416218, "grad_norm": 1.328012228012085, "learning_rate": 3.3800000000000007e-06, "loss": 0.015, "step": 33101 }, { "epoch": 1.7647811483712748, "grad_norm": 1.6548655033111572, "learning_rate": 3.3798000000000002e-06, "loss": 0.0228, "step": 33102 }, { "epoch": 1.7648344618009277, "grad_norm": 1.1913350820541382, "learning_rate": 3.3796000000000002e-06, "loss": 0.018, "step": 33103 }, { "epoch": 1.7648877752305805, "grad_norm": 1.2043170928955078, "learning_rate": 3.3794000000000006e-06, "loss": 0.0044, "step": 33104 }, { "epoch": 1.7649410886602335, "grad_norm": 2.1368842124938965, "learning_rate": 3.3792e-06, "loss": 0.0154, "step": 33105 }, { "epoch": 1.7649944020898864, "grad_norm": 0.8462626934051514, "learning_rate": 3.379e-06, "loss": 0.0054, "step": 33106 }, { "epoch": 1.7650477155195392, "grad_norm": 1.7171411514282227, "learning_rate": 3.3788000000000006e-06, "loss": 0.0373, "step": 33107 }, { "epoch": 1.7651010289491924, "grad_norm": 0.34497955441474915, "learning_rate": 3.3786e-06, "loss": 0.0029, "step": 33108 }, { "epoch": 1.7651543423788452, "grad_norm": 1.4376722574234009, "learning_rate": 3.3784e-06, "loss": 0.0195, "step": 33109 }, { "epoch": 1.7652076558084981, "grad_norm": 1.8090898990631104, "learning_rate": 3.3782000000000005e-06, "loss": 0.0152, "step": 33110 }, { "epoch": 1.7652609692381511, "grad_norm": 2.0185675621032715, "learning_rate": 3.378e-06, "loss": 0.0193, "step": 33111 }, { "epoch": 1.7653142826678039, "grad_norm": 1.7373201847076416, "learning_rate": 3.3778000000000005e-06, "loss": 0.0117, "step": 33112 }, { "epoch": 1.765367596097457, "grad_norm": 1.0042203664779663, "learning_rate": 3.3776000000000005e-06, "loss": 0.0132, "step": 33113 }, { "epoch": 1.7654209095271098, "grad_norm": 1.771458387374878, "learning_rate": 3.3774e-06, "loss": 0.0228, "step": 33114 }, { "epoch": 1.7654742229567628, "grad_norm": 2.115276575088501, "learning_rate": 3.3772000000000005e-06, "loss": 0.0302, "step": 33115 }, { "epoch": 1.7655275363864158, "grad_norm": 0.9237523674964905, "learning_rate": 3.3770000000000004e-06, "loss": 0.0166, "step": 33116 }, { "epoch": 1.7655808498160686, "grad_norm": 1.849899172782898, "learning_rate": 3.3768e-06, "loss": 0.0126, "step": 33117 }, { "epoch": 1.7656341632457218, "grad_norm": 1.583980917930603, "learning_rate": 3.3766000000000004e-06, "loss": 0.0242, "step": 33118 }, { "epoch": 1.7656874766753745, "grad_norm": 0.7736246585845947, "learning_rate": 3.3764000000000004e-06, "loss": 0.0047, "step": 33119 }, { "epoch": 1.7657407901050275, "grad_norm": 1.5287917852401733, "learning_rate": 3.3762e-06, "loss": 0.0107, "step": 33120 }, { "epoch": 1.7657941035346805, "grad_norm": 1.4393621683120728, "learning_rate": 3.3760000000000004e-06, "loss": 0.0099, "step": 33121 }, { "epoch": 1.7658474169643332, "grad_norm": 2.1859655380249023, "learning_rate": 3.3758000000000004e-06, "loss": 0.0162, "step": 33122 }, { "epoch": 1.7659007303939862, "grad_norm": 2.004302740097046, "learning_rate": 3.3756000000000003e-06, "loss": 0.0143, "step": 33123 }, { "epoch": 1.7659540438236392, "grad_norm": 1.5328532457351685, "learning_rate": 3.3754000000000003e-06, "loss": 0.0266, "step": 33124 }, { "epoch": 1.766007357253292, "grad_norm": 1.567049503326416, "learning_rate": 3.3752000000000003e-06, "loss": 0.0117, "step": 33125 }, { "epoch": 1.7660606706829451, "grad_norm": 1.32016122341156, "learning_rate": 3.3750000000000003e-06, "loss": 0.0261, "step": 33126 }, { "epoch": 1.766113984112598, "grad_norm": 2.237600564956665, "learning_rate": 3.3748000000000003e-06, "loss": 0.0122, "step": 33127 }, { "epoch": 1.7661672975422509, "grad_norm": 2.143296957015991, "learning_rate": 3.3746000000000003e-06, "loss": 0.014, "step": 33128 }, { "epoch": 1.7662206109719039, "grad_norm": 2.2706875801086426, "learning_rate": 3.3744000000000003e-06, "loss": 0.0499, "step": 33129 }, { "epoch": 1.7662739244015566, "grad_norm": 2.584364414215088, "learning_rate": 3.3742000000000002e-06, "loss": 0.0259, "step": 33130 }, { "epoch": 1.7663272378312098, "grad_norm": 2.6085939407348633, "learning_rate": 3.3740000000000002e-06, "loss": 0.0113, "step": 33131 }, { "epoch": 1.7663805512608626, "grad_norm": 1.5215834379196167, "learning_rate": 3.3738e-06, "loss": 0.0144, "step": 33132 }, { "epoch": 1.7664338646905156, "grad_norm": 3.4752163887023926, "learning_rate": 3.3736e-06, "loss": 0.0226, "step": 33133 }, { "epoch": 1.7664871781201685, "grad_norm": 1.9112470149993896, "learning_rate": 3.3734e-06, "loss": 0.0132, "step": 33134 }, { "epoch": 1.7665404915498213, "grad_norm": 0.35414281487464905, "learning_rate": 3.3732e-06, "loss": 0.0031, "step": 33135 }, { "epoch": 1.7665938049794745, "grad_norm": 1.8440450429916382, "learning_rate": 3.373e-06, "loss": 0.019, "step": 33136 }, { "epoch": 1.7666471184091272, "grad_norm": 3.3526675701141357, "learning_rate": 3.3728000000000006e-06, "loss": 0.0399, "step": 33137 }, { "epoch": 1.7667004318387802, "grad_norm": 0.8234958648681641, "learning_rate": 3.3726e-06, "loss": 0.0062, "step": 33138 }, { "epoch": 1.7667537452684332, "grad_norm": 0.8892975449562073, "learning_rate": 3.3724e-06, "loss": 0.0063, "step": 33139 }, { "epoch": 1.766807058698086, "grad_norm": 1.543575406074524, "learning_rate": 3.3722000000000005e-06, "loss": 0.0342, "step": 33140 }, { "epoch": 1.766860372127739, "grad_norm": 2.4776813983917236, "learning_rate": 3.372e-06, "loss": 0.0433, "step": 33141 }, { "epoch": 1.766913685557392, "grad_norm": 2.06561541557312, "learning_rate": 3.3718e-06, "loss": 0.0249, "step": 33142 }, { "epoch": 1.7669669989870447, "grad_norm": 0.3260422945022583, "learning_rate": 3.3716000000000005e-06, "loss": 0.0022, "step": 33143 }, { "epoch": 1.7670203124166979, "grad_norm": 2.8500711917877197, "learning_rate": 3.3714e-06, "loss": 0.027, "step": 33144 }, { "epoch": 1.7670736258463506, "grad_norm": 1.4534612894058228, "learning_rate": 3.3712e-06, "loss": 0.0104, "step": 33145 }, { "epoch": 1.7671269392760036, "grad_norm": 1.507022500038147, "learning_rate": 3.3710000000000004e-06, "loss": 0.0118, "step": 33146 }, { "epoch": 1.7671802527056566, "grad_norm": 1.0625910758972168, "learning_rate": 3.3708e-06, "loss": 0.0167, "step": 33147 }, { "epoch": 1.7672335661353094, "grad_norm": 1.8912371397018433, "learning_rate": 3.3706000000000004e-06, "loss": 0.0231, "step": 33148 }, { "epoch": 1.7672868795649626, "grad_norm": 0.5846272110939026, "learning_rate": 3.3704000000000004e-06, "loss": 0.0046, "step": 33149 }, { "epoch": 1.7673401929946153, "grad_norm": 1.34175705909729, "learning_rate": 3.3702e-06, "loss": 0.0067, "step": 33150 }, { "epoch": 1.7673935064242683, "grad_norm": 2.9191668033599854, "learning_rate": 3.3700000000000003e-06, "loss": 0.0288, "step": 33151 }, { "epoch": 1.7674468198539213, "grad_norm": 1.7393417358398438, "learning_rate": 3.3698000000000003e-06, "loss": 0.0357, "step": 33152 }, { "epoch": 1.767500133283574, "grad_norm": 0.3908035159111023, "learning_rate": 3.3696e-06, "loss": 0.0042, "step": 33153 }, { "epoch": 1.7675534467132272, "grad_norm": 1.3898930549621582, "learning_rate": 3.3694000000000003e-06, "loss": 0.007, "step": 33154 }, { "epoch": 1.76760676014288, "grad_norm": 2.2727880477905273, "learning_rate": 3.3692000000000003e-06, "loss": 0.0222, "step": 33155 }, { "epoch": 1.767660073572533, "grad_norm": 0.9304391145706177, "learning_rate": 3.369e-06, "loss": 0.0176, "step": 33156 }, { "epoch": 1.767713387002186, "grad_norm": 1.1821861267089844, "learning_rate": 3.3688000000000003e-06, "loss": 0.0103, "step": 33157 }, { "epoch": 1.7677667004318387, "grad_norm": 2.151777744293213, "learning_rate": 3.3686000000000002e-06, "loss": 0.0336, "step": 33158 }, { "epoch": 1.7678200138614917, "grad_norm": 0.9255915284156799, "learning_rate": 3.3684e-06, "loss": 0.0137, "step": 33159 }, { "epoch": 1.7678733272911447, "grad_norm": 2.8674986362457275, "learning_rate": 3.3682e-06, "loss": 0.0391, "step": 33160 }, { "epoch": 1.7679266407207974, "grad_norm": 1.9116073846817017, "learning_rate": 3.368e-06, "loss": 0.035, "step": 33161 }, { "epoch": 1.7679799541504506, "grad_norm": 2.605239152908325, "learning_rate": 3.3678000000000006e-06, "loss": 0.0328, "step": 33162 }, { "epoch": 1.7680332675801034, "grad_norm": 1.5666453838348389, "learning_rate": 3.3676e-06, "loss": 0.0072, "step": 33163 }, { "epoch": 1.7680865810097564, "grad_norm": 2.3866491317749023, "learning_rate": 3.3674e-06, "loss": 0.0122, "step": 33164 }, { "epoch": 1.7681398944394093, "grad_norm": 2.7623910903930664, "learning_rate": 3.3672000000000006e-06, "loss": 0.0269, "step": 33165 }, { "epoch": 1.768193207869062, "grad_norm": 1.5612373352050781, "learning_rate": 3.367e-06, "loss": 0.0217, "step": 33166 }, { "epoch": 1.7682465212987153, "grad_norm": 1.4886798858642578, "learning_rate": 3.3668e-06, "loss": 0.0294, "step": 33167 }, { "epoch": 1.768299834728368, "grad_norm": 0.9731857180595398, "learning_rate": 3.3666000000000005e-06, "loss": 0.0133, "step": 33168 }, { "epoch": 1.768353148158021, "grad_norm": 1.8539682626724243, "learning_rate": 3.3664e-06, "loss": 0.013, "step": 33169 }, { "epoch": 1.768406461587674, "grad_norm": 0.9609204530715942, "learning_rate": 3.3662e-06, "loss": 0.0051, "step": 33170 }, { "epoch": 1.7684597750173268, "grad_norm": 2.6774845123291016, "learning_rate": 3.3660000000000005e-06, "loss": 0.0235, "step": 33171 }, { "epoch": 1.7685130884469797, "grad_norm": 4.436313629150391, "learning_rate": 3.3658e-06, "loss": 0.0496, "step": 33172 }, { "epoch": 1.7685664018766327, "grad_norm": 2.832282066345215, "learning_rate": 3.3656000000000004e-06, "loss": 0.0102, "step": 33173 }, { "epoch": 1.7686197153062857, "grad_norm": 2.3062474727630615, "learning_rate": 3.3654000000000004e-06, "loss": 0.0246, "step": 33174 }, { "epoch": 1.7686730287359387, "grad_norm": 2.952427625656128, "learning_rate": 3.3652e-06, "loss": 0.0233, "step": 33175 }, { "epoch": 1.7687263421655914, "grad_norm": 1.0831468105316162, "learning_rate": 3.3650000000000004e-06, "loss": 0.0137, "step": 33176 }, { "epoch": 1.7687796555952444, "grad_norm": 2.504728078842163, "learning_rate": 3.3648000000000004e-06, "loss": 0.0481, "step": 33177 }, { "epoch": 1.7688329690248974, "grad_norm": 1.6429436206817627, "learning_rate": 3.3646e-06, "loss": 0.0093, "step": 33178 }, { "epoch": 1.7688862824545502, "grad_norm": 1.6060831546783447, "learning_rate": 3.3644000000000003e-06, "loss": 0.024, "step": 33179 }, { "epoch": 1.7689395958842034, "grad_norm": 1.5269865989685059, "learning_rate": 3.3642000000000003e-06, "loss": 0.0217, "step": 33180 }, { "epoch": 1.768992909313856, "grad_norm": 2.945688009262085, "learning_rate": 3.364e-06, "loss": 0.0163, "step": 33181 }, { "epoch": 1.769046222743509, "grad_norm": 1.4427605867385864, "learning_rate": 3.3638000000000003e-06, "loss": 0.0088, "step": 33182 }, { "epoch": 1.769099536173162, "grad_norm": 4.273681163787842, "learning_rate": 3.3636000000000003e-06, "loss": 0.0208, "step": 33183 }, { "epoch": 1.7691528496028148, "grad_norm": 0.5287516713142395, "learning_rate": 3.3634e-06, "loss": 0.0038, "step": 33184 }, { "epoch": 1.769206163032468, "grad_norm": 0.8016275763511658, "learning_rate": 3.3632000000000003e-06, "loss": 0.0063, "step": 33185 }, { "epoch": 1.7692594764621208, "grad_norm": 0.902324378490448, "learning_rate": 3.3630000000000002e-06, "loss": 0.0221, "step": 33186 }, { "epoch": 1.7693127898917738, "grad_norm": 2.4610671997070312, "learning_rate": 3.3628000000000007e-06, "loss": 0.0295, "step": 33187 }, { "epoch": 1.7693661033214267, "grad_norm": 2.196442127227783, "learning_rate": 3.3626000000000002e-06, "loss": 0.0273, "step": 33188 }, { "epoch": 1.7694194167510795, "grad_norm": 1.3965582847595215, "learning_rate": 3.3624e-06, "loss": 0.0075, "step": 33189 }, { "epoch": 1.7694727301807325, "grad_norm": 1.0706136226654053, "learning_rate": 3.3622000000000006e-06, "loss": 0.0188, "step": 33190 }, { "epoch": 1.7695260436103855, "grad_norm": 2.220839262008667, "learning_rate": 3.362e-06, "loss": 0.013, "step": 33191 }, { "epoch": 1.7695793570400384, "grad_norm": 2.4405317306518555, "learning_rate": 3.3618e-06, "loss": 0.0186, "step": 33192 }, { "epoch": 1.7696326704696914, "grad_norm": 1.4357521533966064, "learning_rate": 3.3616000000000006e-06, "loss": 0.0224, "step": 33193 }, { "epoch": 1.7696859838993442, "grad_norm": 2.1104025840759277, "learning_rate": 3.3614e-06, "loss": 0.04, "step": 33194 }, { "epoch": 1.7697392973289972, "grad_norm": 1.1238104104995728, "learning_rate": 3.3612e-06, "loss": 0.0246, "step": 33195 }, { "epoch": 1.7697926107586501, "grad_norm": 1.614183783531189, "learning_rate": 3.3610000000000005e-06, "loss": 0.0077, "step": 33196 }, { "epoch": 1.7698459241883029, "grad_norm": 3.4612057209014893, "learning_rate": 3.3608e-06, "loss": 0.0397, "step": 33197 }, { "epoch": 1.769899237617956, "grad_norm": 3.7566239833831787, "learning_rate": 3.3606000000000005e-06, "loss": 0.0316, "step": 33198 }, { "epoch": 1.7699525510476088, "grad_norm": 1.440739393234253, "learning_rate": 3.3604000000000005e-06, "loss": 0.0323, "step": 33199 }, { "epoch": 1.7700058644772618, "grad_norm": 0.7740601897239685, "learning_rate": 3.3602e-06, "loss": 0.0078, "step": 33200 }, { "epoch": 1.7700591779069148, "grad_norm": 1.1836119890213013, "learning_rate": 3.3600000000000004e-06, "loss": 0.0067, "step": 33201 }, { "epoch": 1.7701124913365676, "grad_norm": 1.229863166809082, "learning_rate": 3.3598000000000004e-06, "loss": 0.0083, "step": 33202 }, { "epoch": 1.7701658047662208, "grad_norm": 2.990560293197632, "learning_rate": 3.3596e-06, "loss": 0.0351, "step": 33203 }, { "epoch": 1.7702191181958735, "grad_norm": 1.2510699033737183, "learning_rate": 3.3594000000000004e-06, "loss": 0.018, "step": 33204 }, { "epoch": 1.7702724316255265, "grad_norm": 0.8612498641014099, "learning_rate": 3.3592000000000004e-06, "loss": 0.0069, "step": 33205 }, { "epoch": 1.7703257450551795, "grad_norm": 1.3609552383422852, "learning_rate": 3.359e-06, "loss": 0.019, "step": 33206 }, { "epoch": 1.7703790584848322, "grad_norm": 1.423353910446167, "learning_rate": 3.3588000000000004e-06, "loss": 0.0136, "step": 33207 }, { "epoch": 1.7704323719144852, "grad_norm": 2.333280086517334, "learning_rate": 3.3586000000000003e-06, "loss": 0.0383, "step": 33208 }, { "epoch": 1.7704856853441382, "grad_norm": 1.6163856983184814, "learning_rate": 3.3584000000000007e-06, "loss": 0.0314, "step": 33209 }, { "epoch": 1.7705389987737912, "grad_norm": 1.013404130935669, "learning_rate": 3.3582000000000003e-06, "loss": 0.0096, "step": 33210 }, { "epoch": 1.7705923122034442, "grad_norm": 2.1968204975128174, "learning_rate": 3.3580000000000003e-06, "loss": 0.0262, "step": 33211 }, { "epoch": 1.770645625633097, "grad_norm": 0.9557900428771973, "learning_rate": 3.3578000000000007e-06, "loss": 0.0092, "step": 33212 }, { "epoch": 1.7706989390627499, "grad_norm": 1.309053897857666, "learning_rate": 3.3576000000000003e-06, "loss": 0.0142, "step": 33213 }, { "epoch": 1.7707522524924029, "grad_norm": 1.6331851482391357, "learning_rate": 3.3574000000000002e-06, "loss": 0.0229, "step": 33214 }, { "epoch": 1.7708055659220556, "grad_norm": 1.3827487230300903, "learning_rate": 3.3572000000000007e-06, "loss": 0.0328, "step": 33215 }, { "epoch": 1.7708588793517088, "grad_norm": 1.4386630058288574, "learning_rate": 3.3570000000000002e-06, "loss": 0.0124, "step": 33216 }, { "epoch": 1.7709121927813616, "grad_norm": 2.3089993000030518, "learning_rate": 3.3568e-06, "loss": 0.0471, "step": 33217 }, { "epoch": 1.7709655062110146, "grad_norm": 0.9768412113189697, "learning_rate": 3.3566000000000006e-06, "loss": 0.0087, "step": 33218 }, { "epoch": 1.7710188196406675, "grad_norm": 2.8334100246429443, "learning_rate": 3.3564e-06, "loss": 0.0149, "step": 33219 }, { "epoch": 1.7710721330703203, "grad_norm": 1.7260231971740723, "learning_rate": 3.3562e-06, "loss": 0.0199, "step": 33220 }, { "epoch": 1.7711254464999735, "grad_norm": 0.7941545248031616, "learning_rate": 3.3560000000000006e-06, "loss": 0.0065, "step": 33221 }, { "epoch": 1.7711787599296263, "grad_norm": 0.8426867127418518, "learning_rate": 3.3558e-06, "loss": 0.0046, "step": 33222 }, { "epoch": 1.7712320733592792, "grad_norm": 1.1568303108215332, "learning_rate": 3.3556000000000005e-06, "loss": 0.0094, "step": 33223 }, { "epoch": 1.7712853867889322, "grad_norm": 1.1335152387619019, "learning_rate": 3.3554000000000005e-06, "loss": 0.012, "step": 33224 }, { "epoch": 1.771338700218585, "grad_norm": 1.7577730417251587, "learning_rate": 3.3552e-06, "loss": 0.0139, "step": 33225 }, { "epoch": 1.771392013648238, "grad_norm": 1.2155253887176514, "learning_rate": 3.3550000000000005e-06, "loss": 0.0158, "step": 33226 }, { "epoch": 1.771445327077891, "grad_norm": 2.037752866744995, "learning_rate": 3.3548000000000005e-06, "loss": 0.0174, "step": 33227 }, { "epoch": 1.7714986405075437, "grad_norm": 2.1909680366516113, "learning_rate": 3.3546e-06, "loss": 0.0193, "step": 33228 }, { "epoch": 1.7715519539371969, "grad_norm": 1.0246485471725464, "learning_rate": 3.3544000000000004e-06, "loss": 0.0062, "step": 33229 }, { "epoch": 1.7716052673668496, "grad_norm": 1.0336858034133911, "learning_rate": 3.3542000000000004e-06, "loss": 0.0154, "step": 33230 }, { "epoch": 1.7716585807965026, "grad_norm": 1.547478437423706, "learning_rate": 3.354e-06, "loss": 0.0155, "step": 33231 }, { "epoch": 1.7717118942261556, "grad_norm": 2.0399229526519775, "learning_rate": 3.3538000000000004e-06, "loss": 0.0232, "step": 33232 }, { "epoch": 1.7717652076558084, "grad_norm": 0.9688271284103394, "learning_rate": 3.3536000000000004e-06, "loss": 0.0062, "step": 33233 }, { "epoch": 1.7718185210854616, "grad_norm": 1.4825363159179688, "learning_rate": 3.3534000000000004e-06, "loss": 0.007, "step": 33234 }, { "epoch": 1.7718718345151143, "grad_norm": 1.1329145431518555, "learning_rate": 3.3532000000000004e-06, "loss": 0.0081, "step": 33235 }, { "epoch": 1.7719251479447673, "grad_norm": 1.1681628227233887, "learning_rate": 3.3530000000000003e-06, "loss": 0.0072, "step": 33236 }, { "epoch": 1.7719784613744203, "grad_norm": 1.1615922451019287, "learning_rate": 3.3528000000000003e-06, "loss": 0.0088, "step": 33237 }, { "epoch": 1.772031774804073, "grad_norm": 0.7417298555374146, "learning_rate": 3.3526000000000003e-06, "loss": 0.0056, "step": 33238 }, { "epoch": 1.7720850882337262, "grad_norm": 1.7450764179229736, "learning_rate": 3.3524000000000003e-06, "loss": 0.0466, "step": 33239 }, { "epoch": 1.772138401663379, "grad_norm": 1.9239262342453003, "learning_rate": 3.3522000000000003e-06, "loss": 0.03, "step": 33240 }, { "epoch": 1.772191715093032, "grad_norm": 2.069260597229004, "learning_rate": 3.3520000000000003e-06, "loss": 0.0392, "step": 33241 }, { "epoch": 1.772245028522685, "grad_norm": 0.8455613255500793, "learning_rate": 3.3518000000000002e-06, "loss": 0.0133, "step": 33242 }, { "epoch": 1.7722983419523377, "grad_norm": 1.4092369079589844, "learning_rate": 3.3516000000000002e-06, "loss": 0.0206, "step": 33243 }, { "epoch": 1.7723516553819907, "grad_norm": 0.9669029712677002, "learning_rate": 3.3514000000000002e-06, "loss": 0.0146, "step": 33244 }, { "epoch": 1.7724049688116437, "grad_norm": 0.8391407132148743, "learning_rate": 3.3512e-06, "loss": 0.0183, "step": 33245 }, { "epoch": 1.7724582822412964, "grad_norm": 1.2461905479431152, "learning_rate": 3.351e-06, "loss": 0.0201, "step": 33246 }, { "epoch": 1.7725115956709496, "grad_norm": 1.338470697402954, "learning_rate": 3.3508e-06, "loss": 0.014, "step": 33247 }, { "epoch": 1.7725649091006024, "grad_norm": 1.6258738040924072, "learning_rate": 3.3506000000000006e-06, "loss": 0.0205, "step": 33248 }, { "epoch": 1.7726182225302554, "grad_norm": 1.7028214931488037, "learning_rate": 3.3504e-06, "loss": 0.0323, "step": 33249 }, { "epoch": 1.7726715359599083, "grad_norm": 2.150297164916992, "learning_rate": 3.3502e-06, "loss": 0.0167, "step": 33250 }, { "epoch": 1.772724849389561, "grad_norm": 0.8242124319076538, "learning_rate": 3.3500000000000005e-06, "loss": 0.0113, "step": 33251 }, { "epoch": 1.7727781628192143, "grad_norm": 2.352710485458374, "learning_rate": 3.3498e-06, "loss": 0.0236, "step": 33252 }, { "epoch": 1.772831476248867, "grad_norm": 4.38372278213501, "learning_rate": 3.3496e-06, "loss": 0.0486, "step": 33253 }, { "epoch": 1.77288478967852, "grad_norm": 2.626897096633911, "learning_rate": 3.3494000000000005e-06, "loss": 0.0184, "step": 33254 }, { "epoch": 1.772938103108173, "grad_norm": 1.7902337312698364, "learning_rate": 3.3492e-06, "loss": 0.0163, "step": 33255 }, { "epoch": 1.7729914165378258, "grad_norm": 2.0013012886047363, "learning_rate": 3.349e-06, "loss": 0.0159, "step": 33256 }, { "epoch": 1.773044729967479, "grad_norm": 1.0600390434265137, "learning_rate": 3.3488000000000004e-06, "loss": 0.0065, "step": 33257 }, { "epoch": 1.7730980433971317, "grad_norm": 2.173403024673462, "learning_rate": 3.3486e-06, "loss": 0.0283, "step": 33258 }, { "epoch": 1.7731513568267847, "grad_norm": 2.962912082672119, "learning_rate": 3.3484000000000004e-06, "loss": 0.0177, "step": 33259 }, { "epoch": 1.7732046702564377, "grad_norm": 1.6579755544662476, "learning_rate": 3.3482000000000004e-06, "loss": 0.0086, "step": 33260 }, { "epoch": 1.7732579836860904, "grad_norm": 1.3807382583618164, "learning_rate": 3.348e-06, "loss": 0.0075, "step": 33261 }, { "epoch": 1.7733112971157434, "grad_norm": 2.9344840049743652, "learning_rate": 3.3478000000000004e-06, "loss": 0.0304, "step": 33262 }, { "epoch": 1.7733646105453964, "grad_norm": 1.8302751779556274, "learning_rate": 3.3476000000000004e-06, "loss": 0.039, "step": 33263 }, { "epoch": 1.7734179239750492, "grad_norm": 0.851334273815155, "learning_rate": 3.3474e-06, "loss": 0.005, "step": 33264 }, { "epoch": 1.7734712374047024, "grad_norm": 1.5796284675598145, "learning_rate": 3.3472000000000003e-06, "loss": 0.0066, "step": 33265 }, { "epoch": 1.7735245508343551, "grad_norm": 2.415313482284546, "learning_rate": 3.3470000000000003e-06, "loss": 0.0575, "step": 33266 }, { "epoch": 1.773577864264008, "grad_norm": 0.8841471672058105, "learning_rate": 3.3468e-06, "loss": 0.0065, "step": 33267 }, { "epoch": 1.773631177693661, "grad_norm": 2.6561813354492188, "learning_rate": 3.3466000000000003e-06, "loss": 0.0146, "step": 33268 }, { "epoch": 1.7736844911233138, "grad_norm": 0.31972965598106384, "learning_rate": 3.3464000000000003e-06, "loss": 0.0025, "step": 33269 }, { "epoch": 1.773737804552967, "grad_norm": 2.319063425064087, "learning_rate": 3.3462e-06, "loss": 0.0159, "step": 33270 }, { "epoch": 1.7737911179826198, "grad_norm": 0.7403860092163086, "learning_rate": 3.3460000000000002e-06, "loss": 0.0058, "step": 33271 }, { "epoch": 1.7738444314122728, "grad_norm": 2.986708879470825, "learning_rate": 3.3458000000000002e-06, "loss": 0.0453, "step": 33272 }, { "epoch": 1.7738977448419257, "grad_norm": 1.7729164361953735, "learning_rate": 3.3456000000000006e-06, "loss": 0.0122, "step": 33273 }, { "epoch": 1.7739510582715785, "grad_norm": 2.0637543201446533, "learning_rate": 3.3454e-06, "loss": 0.0225, "step": 33274 }, { "epoch": 1.7740043717012315, "grad_norm": 1.4184993505477905, "learning_rate": 3.3452e-06, "loss": 0.0202, "step": 33275 }, { "epoch": 1.7740576851308845, "grad_norm": 3.057007074356079, "learning_rate": 3.3450000000000006e-06, "loss": 0.0376, "step": 33276 }, { "epoch": 1.7741109985605374, "grad_norm": 2.1721103191375732, "learning_rate": 3.3448e-06, "loss": 0.0263, "step": 33277 }, { "epoch": 1.7741643119901904, "grad_norm": 2.1882665157318115, "learning_rate": 3.3446e-06, "loss": 0.0107, "step": 33278 }, { "epoch": 1.7742176254198432, "grad_norm": 1.4140896797180176, "learning_rate": 3.3444000000000005e-06, "loss": 0.0169, "step": 33279 }, { "epoch": 1.7742709388494962, "grad_norm": 0.8914178609848022, "learning_rate": 3.3442e-06, "loss": 0.0094, "step": 33280 }, { "epoch": 1.7743242522791491, "grad_norm": 1.9793781042099, "learning_rate": 3.344e-06, "loss": 0.0232, "step": 33281 }, { "epoch": 1.774377565708802, "grad_norm": 2.4497172832489014, "learning_rate": 3.3438000000000005e-06, "loss": 0.0104, "step": 33282 }, { "epoch": 1.774430879138455, "grad_norm": 0.5371397137641907, "learning_rate": 3.3436e-06, "loss": 0.0049, "step": 33283 }, { "epoch": 1.7744841925681079, "grad_norm": 1.1674234867095947, "learning_rate": 3.3434000000000005e-06, "loss": 0.0065, "step": 33284 }, { "epoch": 1.7745375059977608, "grad_norm": 1.646267294883728, "learning_rate": 3.3432000000000004e-06, "loss": 0.0189, "step": 33285 }, { "epoch": 1.7745908194274138, "grad_norm": 0.9921587109565735, "learning_rate": 3.343e-06, "loss": 0.0075, "step": 33286 }, { "epoch": 1.7746441328570666, "grad_norm": 1.6698887348175049, "learning_rate": 3.3428000000000004e-06, "loss": 0.0131, "step": 33287 }, { "epoch": 1.7746974462867198, "grad_norm": 1.0130136013031006, "learning_rate": 3.3426000000000004e-06, "loss": 0.0057, "step": 33288 }, { "epoch": 1.7747507597163725, "grad_norm": 3.310626268386841, "learning_rate": 3.3424e-06, "loss": 0.0248, "step": 33289 }, { "epoch": 1.7748040731460255, "grad_norm": 0.9988544583320618, "learning_rate": 3.3422000000000004e-06, "loss": 0.0161, "step": 33290 }, { "epoch": 1.7748573865756785, "grad_norm": 0.6673033237457275, "learning_rate": 3.3420000000000004e-06, "loss": 0.0129, "step": 33291 }, { "epoch": 1.7749107000053312, "grad_norm": 5.093782901763916, "learning_rate": 3.3418e-06, "loss": 0.0291, "step": 33292 }, { "epoch": 1.7749640134349842, "grad_norm": 1.4719988107681274, "learning_rate": 3.3416000000000003e-06, "loss": 0.0222, "step": 33293 }, { "epoch": 1.7750173268646372, "grad_norm": 2.6191766262054443, "learning_rate": 3.3414000000000003e-06, "loss": 0.03, "step": 33294 }, { "epoch": 1.7750706402942902, "grad_norm": 2.374070644378662, "learning_rate": 3.3412e-06, "loss": 0.0161, "step": 33295 }, { "epoch": 1.7751239537239432, "grad_norm": 2.3184545040130615, "learning_rate": 3.3410000000000003e-06, "loss": 0.0358, "step": 33296 }, { "epoch": 1.775177267153596, "grad_norm": 3.4697604179382324, "learning_rate": 3.3408000000000003e-06, "loss": 0.0363, "step": 33297 }, { "epoch": 1.775230580583249, "grad_norm": 1.567218542098999, "learning_rate": 3.3406000000000007e-06, "loss": 0.0172, "step": 33298 }, { "epoch": 1.7752838940129019, "grad_norm": 1.5877522230148315, "learning_rate": 3.3404000000000002e-06, "loss": 0.0342, "step": 33299 }, { "epoch": 1.7753372074425546, "grad_norm": 4.611313819885254, "learning_rate": 3.3402000000000002e-06, "loss": 0.0336, "step": 33300 }, { "epoch": 1.7753905208722078, "grad_norm": 2.199471950531006, "learning_rate": 3.3400000000000006e-06, "loss": 0.0182, "step": 33301 }, { "epoch": 1.7754438343018606, "grad_norm": 1.5381181240081787, "learning_rate": 3.3398e-06, "loss": 0.0142, "step": 33302 }, { "epoch": 1.7754971477315136, "grad_norm": 1.609952449798584, "learning_rate": 3.3396e-06, "loss": 0.0128, "step": 33303 }, { "epoch": 1.7755504611611665, "grad_norm": 1.1798535585403442, "learning_rate": 3.3394000000000006e-06, "loss": 0.0078, "step": 33304 }, { "epoch": 1.7756037745908193, "grad_norm": 1.7647453546524048, "learning_rate": 3.3392e-06, "loss": 0.0131, "step": 33305 }, { "epoch": 1.7756570880204725, "grad_norm": 1.871639609336853, "learning_rate": 3.339e-06, "loss": 0.0128, "step": 33306 }, { "epoch": 1.7757104014501253, "grad_norm": 4.900589942932129, "learning_rate": 3.3388000000000005e-06, "loss": 0.0191, "step": 33307 }, { "epoch": 1.7757637148797782, "grad_norm": 3.0816354751586914, "learning_rate": 3.3386e-06, "loss": 0.0223, "step": 33308 }, { "epoch": 1.7758170283094312, "grad_norm": 1.2957911491394043, "learning_rate": 3.3384000000000005e-06, "loss": 0.0208, "step": 33309 }, { "epoch": 1.775870341739084, "grad_norm": 1.387634038925171, "learning_rate": 3.3382000000000005e-06, "loss": 0.0232, "step": 33310 }, { "epoch": 1.775923655168737, "grad_norm": 1.0146641731262207, "learning_rate": 3.338e-06, "loss": 0.0121, "step": 33311 }, { "epoch": 1.77597696859839, "grad_norm": 0.8988946676254272, "learning_rate": 3.3378000000000005e-06, "loss": 0.0052, "step": 33312 }, { "epoch": 1.776030282028043, "grad_norm": 2.7168660163879395, "learning_rate": 3.3376000000000005e-06, "loss": 0.0129, "step": 33313 }, { "epoch": 1.776083595457696, "grad_norm": 1.7275291681289673, "learning_rate": 3.3374e-06, "loss": 0.0296, "step": 33314 }, { "epoch": 1.7761369088873487, "grad_norm": 2.8340952396392822, "learning_rate": 3.3372000000000004e-06, "loss": 0.0137, "step": 33315 }, { "epoch": 1.7761902223170016, "grad_norm": 1.0251973867416382, "learning_rate": 3.3370000000000004e-06, "loss": 0.0102, "step": 33316 }, { "epoch": 1.7762435357466546, "grad_norm": 2.799145460128784, "learning_rate": 3.3368e-06, "loss": 0.0623, "step": 33317 }, { "epoch": 1.7762968491763074, "grad_norm": 1.6607584953308105, "learning_rate": 3.3366000000000004e-06, "loss": 0.0316, "step": 33318 }, { "epoch": 1.7763501626059606, "grad_norm": 1.0344794988632202, "learning_rate": 3.3364000000000004e-06, "loss": 0.0196, "step": 33319 }, { "epoch": 1.7764034760356133, "grad_norm": 1.9980566501617432, "learning_rate": 3.3362000000000003e-06, "loss": 0.0198, "step": 33320 }, { "epoch": 1.7764567894652663, "grad_norm": 2.421806573867798, "learning_rate": 3.3360000000000003e-06, "loss": 0.0306, "step": 33321 }, { "epoch": 1.7765101028949193, "grad_norm": 0.7501905560493469, "learning_rate": 3.3358000000000003e-06, "loss": 0.0213, "step": 33322 }, { "epoch": 1.776563416324572, "grad_norm": 1.847288727760315, "learning_rate": 3.3356000000000003e-06, "loss": 0.0239, "step": 33323 }, { "epoch": 1.7766167297542252, "grad_norm": 0.707274317741394, "learning_rate": 3.3354000000000003e-06, "loss": 0.0071, "step": 33324 }, { "epoch": 1.776670043183878, "grad_norm": 2.089667558670044, "learning_rate": 3.3352000000000003e-06, "loss": 0.0162, "step": 33325 }, { "epoch": 1.776723356613531, "grad_norm": 4.424038410186768, "learning_rate": 3.3350000000000003e-06, "loss": 0.0219, "step": 33326 }, { "epoch": 1.776776670043184, "grad_norm": 1.381917119026184, "learning_rate": 3.3348000000000002e-06, "loss": 0.0161, "step": 33327 }, { "epoch": 1.7768299834728367, "grad_norm": 1.8590025901794434, "learning_rate": 3.3346000000000002e-06, "loss": 0.0256, "step": 33328 }, { "epoch": 1.7768832969024897, "grad_norm": 1.4338544607162476, "learning_rate": 3.3344e-06, "loss": 0.0094, "step": 33329 }, { "epoch": 1.7769366103321427, "grad_norm": 0.7178535461425781, "learning_rate": 3.3342e-06, "loss": 0.007, "step": 33330 }, { "epoch": 1.7769899237617954, "grad_norm": 2.2486143112182617, "learning_rate": 3.334e-06, "loss": 0.0203, "step": 33331 }, { "epoch": 1.7770432371914486, "grad_norm": 3.0266599655151367, "learning_rate": 3.3338e-06, "loss": 0.0315, "step": 33332 }, { "epoch": 1.7770965506211014, "grad_norm": 1.5552088022232056, "learning_rate": 3.3336e-06, "loss": 0.0179, "step": 33333 }, { "epoch": 1.7771498640507544, "grad_norm": 2.3372673988342285, "learning_rate": 3.3334000000000006e-06, "loss": 0.0168, "step": 33334 }, { "epoch": 1.7772031774804073, "grad_norm": 4.074284553527832, "learning_rate": 3.3332e-06, "loss": 0.0341, "step": 33335 }, { "epoch": 1.77725649091006, "grad_norm": 2.9207262992858887, "learning_rate": 3.333e-06, "loss": 0.0428, "step": 33336 }, { "epoch": 1.7773098043397133, "grad_norm": 1.7930958271026611, "learning_rate": 3.3328000000000005e-06, "loss": 0.0346, "step": 33337 }, { "epoch": 1.777363117769366, "grad_norm": 1.9969419240951538, "learning_rate": 3.3326e-06, "loss": 0.0227, "step": 33338 }, { "epoch": 1.777416431199019, "grad_norm": 2.5314712524414062, "learning_rate": 3.3324e-06, "loss": 0.0339, "step": 33339 }, { "epoch": 1.777469744628672, "grad_norm": 6.881876468658447, "learning_rate": 3.3322000000000005e-06, "loss": 0.0406, "step": 33340 }, { "epoch": 1.7775230580583248, "grad_norm": 2.431657075881958, "learning_rate": 3.332e-06, "loss": 0.0391, "step": 33341 }, { "epoch": 1.777576371487978, "grad_norm": 2.471705198287964, "learning_rate": 3.3318e-06, "loss": 0.0477, "step": 33342 }, { "epoch": 1.7776296849176307, "grad_norm": 2.0801143646240234, "learning_rate": 3.3316000000000004e-06, "loss": 0.0338, "step": 33343 }, { "epoch": 1.7776829983472837, "grad_norm": 1.5063488483428955, "learning_rate": 3.3314e-06, "loss": 0.0213, "step": 33344 }, { "epoch": 1.7777363117769367, "grad_norm": 1.1852176189422607, "learning_rate": 3.3312000000000004e-06, "loss": 0.0089, "step": 33345 }, { "epoch": 1.7777896252065895, "grad_norm": 1.1202640533447266, "learning_rate": 3.3310000000000004e-06, "loss": 0.0079, "step": 33346 }, { "epoch": 1.7778429386362424, "grad_norm": 3.2138848304748535, "learning_rate": 3.3308e-06, "loss": 0.028, "step": 33347 }, { "epoch": 1.7778962520658954, "grad_norm": 2.138068914413452, "learning_rate": 3.3306000000000003e-06, "loss": 0.0391, "step": 33348 }, { "epoch": 1.7779495654955482, "grad_norm": 1.2688634395599365, "learning_rate": 3.3304000000000003e-06, "loss": 0.015, "step": 33349 }, { "epoch": 1.7780028789252014, "grad_norm": 1.2125663757324219, "learning_rate": 3.3302e-06, "loss": 0.0146, "step": 33350 }, { "epoch": 1.7780561923548541, "grad_norm": 1.238415241241455, "learning_rate": 3.3300000000000003e-06, "loss": 0.0154, "step": 33351 }, { "epoch": 1.778109505784507, "grad_norm": 2.1709952354431152, "learning_rate": 3.3298000000000003e-06, "loss": 0.0149, "step": 33352 }, { "epoch": 1.77816281921416, "grad_norm": 2.0861971378326416, "learning_rate": 3.3296e-06, "loss": 0.0192, "step": 33353 }, { "epoch": 1.7782161326438128, "grad_norm": 1.4919880628585815, "learning_rate": 3.3294000000000003e-06, "loss": 0.0144, "step": 33354 }, { "epoch": 1.778269446073466, "grad_norm": 0.8816691040992737, "learning_rate": 3.3292000000000002e-06, "loss": 0.0076, "step": 33355 }, { "epoch": 1.7783227595031188, "grad_norm": 2.0112907886505127, "learning_rate": 3.329e-06, "loss": 0.0074, "step": 33356 }, { "epoch": 1.7783760729327718, "grad_norm": 1.1078605651855469, "learning_rate": 3.3288000000000002e-06, "loss": 0.0078, "step": 33357 }, { "epoch": 1.7784293863624248, "grad_norm": 2.2804667949676514, "learning_rate": 3.3286e-06, "loss": 0.0412, "step": 33358 }, { "epoch": 1.7784826997920775, "grad_norm": 1.4721928834915161, "learning_rate": 3.3284000000000006e-06, "loss": 0.0222, "step": 33359 }, { "epoch": 1.7785360132217307, "grad_norm": 0.511845052242279, "learning_rate": 3.3282e-06, "loss": 0.0048, "step": 33360 }, { "epoch": 1.7785893266513835, "grad_norm": 2.1589109897613525, "learning_rate": 3.328e-06, "loss": 0.0213, "step": 33361 }, { "epoch": 1.7786426400810365, "grad_norm": 1.63961923122406, "learning_rate": 3.3278000000000006e-06, "loss": 0.0122, "step": 33362 }, { "epoch": 1.7786959535106894, "grad_norm": 2.368435859680176, "learning_rate": 3.3276e-06, "loss": 0.0319, "step": 33363 }, { "epoch": 1.7787492669403422, "grad_norm": 1.437135100364685, "learning_rate": 3.3274e-06, "loss": 0.0194, "step": 33364 }, { "epoch": 1.7788025803699952, "grad_norm": 3.022472381591797, "learning_rate": 3.3272000000000005e-06, "loss": 0.0331, "step": 33365 }, { "epoch": 1.7788558937996481, "grad_norm": 2.4384491443634033, "learning_rate": 3.327e-06, "loss": 0.0193, "step": 33366 }, { "epoch": 1.778909207229301, "grad_norm": 0.6038036942481995, "learning_rate": 3.3268e-06, "loss": 0.007, "step": 33367 }, { "epoch": 1.778962520658954, "grad_norm": 2.4188101291656494, "learning_rate": 3.3266000000000005e-06, "loss": 0.0152, "step": 33368 }, { "epoch": 1.7790158340886069, "grad_norm": 1.9105730056762695, "learning_rate": 3.3264e-06, "loss": 0.0326, "step": 33369 }, { "epoch": 1.7790691475182598, "grad_norm": 1.4733062982559204, "learning_rate": 3.3262000000000004e-06, "loss": 0.0195, "step": 33370 }, { "epoch": 1.7791224609479128, "grad_norm": 1.5202748775482178, "learning_rate": 3.3260000000000004e-06, "loss": 0.0301, "step": 33371 }, { "epoch": 1.7791757743775656, "grad_norm": 0.5268367528915405, "learning_rate": 3.3258e-06, "loss": 0.0048, "step": 33372 }, { "epoch": 1.7792290878072188, "grad_norm": 2.4134020805358887, "learning_rate": 3.3256000000000004e-06, "loss": 0.0231, "step": 33373 }, { "epoch": 1.7792824012368715, "grad_norm": 1.6330156326293945, "learning_rate": 3.3254000000000004e-06, "loss": 0.0247, "step": 33374 }, { "epoch": 1.7793357146665245, "grad_norm": 1.8275214433670044, "learning_rate": 3.3252e-06, "loss": 0.0201, "step": 33375 }, { "epoch": 1.7793890280961775, "grad_norm": 1.4631036520004272, "learning_rate": 3.3250000000000004e-06, "loss": 0.0167, "step": 33376 }, { "epoch": 1.7794423415258303, "grad_norm": 1.3746964931488037, "learning_rate": 3.3248000000000003e-06, "loss": 0.0075, "step": 33377 }, { "epoch": 1.7794956549554835, "grad_norm": 1.4706180095672607, "learning_rate": 3.3246e-06, "loss": 0.0117, "step": 33378 }, { "epoch": 1.7795489683851362, "grad_norm": 1.2160271406173706, "learning_rate": 3.3244000000000003e-06, "loss": 0.0268, "step": 33379 }, { "epoch": 1.7796022818147892, "grad_norm": 3.6354475021362305, "learning_rate": 3.3242000000000003e-06, "loss": 0.0164, "step": 33380 }, { "epoch": 1.7796555952444422, "grad_norm": 1.7354658842086792, "learning_rate": 3.324e-06, "loss": 0.0295, "step": 33381 }, { "epoch": 1.779708908674095, "grad_norm": 1.4829171895980835, "learning_rate": 3.3238000000000003e-06, "loss": 0.015, "step": 33382 }, { "epoch": 1.779762222103748, "grad_norm": 1.673182487487793, "learning_rate": 3.3236000000000002e-06, "loss": 0.0092, "step": 33383 }, { "epoch": 1.7798155355334009, "grad_norm": 1.6171232461929321, "learning_rate": 3.3234000000000007e-06, "loss": 0.012, "step": 33384 }, { "epoch": 1.7798688489630536, "grad_norm": 1.3125642538070679, "learning_rate": 3.3232000000000002e-06, "loss": 0.0089, "step": 33385 }, { "epoch": 1.7799221623927068, "grad_norm": 2.5182383060455322, "learning_rate": 3.323e-06, "loss": 0.0286, "step": 33386 }, { "epoch": 1.7799754758223596, "grad_norm": 2.161512613296509, "learning_rate": 3.3228000000000006e-06, "loss": 0.0538, "step": 33387 }, { "epoch": 1.7800287892520126, "grad_norm": 0.7203165888786316, "learning_rate": 3.3226e-06, "loss": 0.0045, "step": 33388 }, { "epoch": 1.7800821026816656, "grad_norm": 1.405177116394043, "learning_rate": 3.3224e-06, "loss": 0.0129, "step": 33389 }, { "epoch": 1.7801354161113183, "grad_norm": 1.7468793392181396, "learning_rate": 3.3222000000000006e-06, "loss": 0.0168, "step": 33390 }, { "epoch": 1.7801887295409715, "grad_norm": 0.5877453684806824, "learning_rate": 3.322e-06, "loss": 0.0044, "step": 33391 }, { "epoch": 1.7802420429706243, "grad_norm": 0.6657013297080994, "learning_rate": 3.3218e-06, "loss": 0.006, "step": 33392 }, { "epoch": 1.7802953564002773, "grad_norm": 1.3299462795257568, "learning_rate": 3.3216000000000005e-06, "loss": 0.013, "step": 33393 }, { "epoch": 1.7803486698299302, "grad_norm": 1.1024481058120728, "learning_rate": 3.3214e-06, "loss": 0.0068, "step": 33394 }, { "epoch": 1.780401983259583, "grad_norm": 2.045088052749634, "learning_rate": 3.3212000000000005e-06, "loss": 0.0329, "step": 33395 }, { "epoch": 1.780455296689236, "grad_norm": 2.0558531284332275, "learning_rate": 3.3210000000000005e-06, "loss": 0.0167, "step": 33396 }, { "epoch": 1.780508610118889, "grad_norm": 1.3760045766830444, "learning_rate": 3.3208e-06, "loss": 0.0164, "step": 33397 }, { "epoch": 1.780561923548542, "grad_norm": 4.395427227020264, "learning_rate": 3.3206000000000004e-06, "loss": 0.0316, "step": 33398 }, { "epoch": 1.780615236978195, "grad_norm": 2.005009174346924, "learning_rate": 3.3204000000000004e-06, "loss": 0.0224, "step": 33399 }, { "epoch": 1.7806685504078477, "grad_norm": 1.7015056610107422, "learning_rate": 3.3202e-06, "loss": 0.0211, "step": 33400 }, { "epoch": 1.7807218638375006, "grad_norm": 2.3897716999053955, "learning_rate": 3.3200000000000004e-06, "loss": 0.0247, "step": 33401 }, { "epoch": 1.7807751772671536, "grad_norm": 1.650179147720337, "learning_rate": 3.3198000000000004e-06, "loss": 0.0323, "step": 33402 }, { "epoch": 1.7808284906968064, "grad_norm": 1.0296744108200073, "learning_rate": 3.3196e-06, "loss": 0.0057, "step": 33403 }, { "epoch": 1.7808818041264596, "grad_norm": 1.0908033847808838, "learning_rate": 3.3194000000000004e-06, "loss": 0.0071, "step": 33404 }, { "epoch": 1.7809351175561123, "grad_norm": 1.4966661930084229, "learning_rate": 3.3192000000000003e-06, "loss": 0.0318, "step": 33405 }, { "epoch": 1.7809884309857653, "grad_norm": 2.049278974533081, "learning_rate": 3.319e-06, "loss": 0.0249, "step": 33406 }, { "epoch": 1.7810417444154183, "grad_norm": 3.2578818798065186, "learning_rate": 3.3188000000000003e-06, "loss": 0.044, "step": 33407 }, { "epoch": 1.781095057845071, "grad_norm": 0.6272794008255005, "learning_rate": 3.3186000000000003e-06, "loss": 0.0047, "step": 33408 }, { "epoch": 1.7811483712747243, "grad_norm": 1.599028468132019, "learning_rate": 3.3184000000000007e-06, "loss": 0.0177, "step": 33409 }, { "epoch": 1.781201684704377, "grad_norm": 1.329595923423767, "learning_rate": 3.3182000000000003e-06, "loss": 0.0086, "step": 33410 }, { "epoch": 1.78125499813403, "grad_norm": 2.7373087406158447, "learning_rate": 3.3180000000000003e-06, "loss": 0.0493, "step": 33411 }, { "epoch": 1.781308311563683, "grad_norm": 0.7622202634811401, "learning_rate": 3.3178000000000007e-06, "loss": 0.0061, "step": 33412 }, { "epoch": 1.7813616249933357, "grad_norm": 2.1217572689056396, "learning_rate": 3.3176000000000002e-06, "loss": 0.0194, "step": 33413 }, { "epoch": 1.7814149384229887, "grad_norm": 1.511802315711975, "learning_rate": 3.3174e-06, "loss": 0.0052, "step": 33414 }, { "epoch": 1.7814682518526417, "grad_norm": 2.4644601345062256, "learning_rate": 3.3172000000000006e-06, "loss": 0.0369, "step": 33415 }, { "epoch": 1.7815215652822947, "grad_norm": 0.9873844981193542, "learning_rate": 3.317e-06, "loss": 0.0059, "step": 33416 }, { "epoch": 1.7815748787119476, "grad_norm": 1.72589910030365, "learning_rate": 3.3168e-06, "loss": 0.0106, "step": 33417 }, { "epoch": 1.7816281921416004, "grad_norm": 2.298766613006592, "learning_rate": 3.3166000000000006e-06, "loss": 0.0171, "step": 33418 }, { "epoch": 1.7816815055712534, "grad_norm": 0.5462881326675415, "learning_rate": 3.3164e-06, "loss": 0.0046, "step": 33419 }, { "epoch": 1.7817348190009064, "grad_norm": 1.1448121070861816, "learning_rate": 3.3162000000000005e-06, "loss": 0.0091, "step": 33420 }, { "epoch": 1.7817881324305591, "grad_norm": 0.8117387294769287, "learning_rate": 3.3160000000000005e-06, "loss": 0.0042, "step": 33421 }, { "epoch": 1.7818414458602123, "grad_norm": 1.4929221868515015, "learning_rate": 3.3158e-06, "loss": 0.0195, "step": 33422 }, { "epoch": 1.781894759289865, "grad_norm": 2.3363616466522217, "learning_rate": 3.3156000000000005e-06, "loss": 0.0274, "step": 33423 }, { "epoch": 1.781948072719518, "grad_norm": 2.9080114364624023, "learning_rate": 3.3154000000000005e-06, "loss": 0.0164, "step": 33424 }, { "epoch": 1.782001386149171, "grad_norm": 1.5523810386657715, "learning_rate": 3.3152e-06, "loss": 0.0281, "step": 33425 }, { "epoch": 1.7820546995788238, "grad_norm": 1.2817553281784058, "learning_rate": 3.3150000000000004e-06, "loss": 0.0191, "step": 33426 }, { "epoch": 1.782108013008477, "grad_norm": 2.0706613063812256, "learning_rate": 3.3148000000000004e-06, "loss": 0.0152, "step": 33427 }, { "epoch": 1.7821613264381297, "grad_norm": 1.6408213376998901, "learning_rate": 3.3146e-06, "loss": 0.0185, "step": 33428 }, { "epoch": 1.7822146398677827, "grad_norm": 0.7391634583473206, "learning_rate": 3.3144000000000004e-06, "loss": 0.0033, "step": 33429 }, { "epoch": 1.7822679532974357, "grad_norm": 1.3463243246078491, "learning_rate": 3.3142000000000004e-06, "loss": 0.0132, "step": 33430 }, { "epoch": 1.7823212667270885, "grad_norm": 1.175154209136963, "learning_rate": 3.314e-06, "loss": 0.0126, "step": 33431 }, { "epoch": 1.7823745801567414, "grad_norm": 3.220428705215454, "learning_rate": 3.3138000000000004e-06, "loss": 0.0206, "step": 33432 }, { "epoch": 1.7824278935863944, "grad_norm": 0.646202027797699, "learning_rate": 3.3136000000000003e-06, "loss": 0.0048, "step": 33433 }, { "epoch": 1.7824812070160474, "grad_norm": 1.356486439704895, "learning_rate": 3.3134000000000003e-06, "loss": 0.0063, "step": 33434 }, { "epoch": 1.7825345204457004, "grad_norm": 2.3504042625427246, "learning_rate": 3.3132000000000003e-06, "loss": 0.0277, "step": 33435 }, { "epoch": 1.7825878338753531, "grad_norm": 1.2072126865386963, "learning_rate": 3.3130000000000003e-06, "loss": 0.007, "step": 33436 }, { "epoch": 1.7826411473050061, "grad_norm": 1.9348011016845703, "learning_rate": 3.3128000000000003e-06, "loss": 0.0243, "step": 33437 }, { "epoch": 1.782694460734659, "grad_norm": 2.359010696411133, "learning_rate": 3.3126000000000003e-06, "loss": 0.0156, "step": 33438 }, { "epoch": 1.7827477741643118, "grad_norm": 1.3298007249832153, "learning_rate": 3.3124000000000003e-06, "loss": 0.0202, "step": 33439 }, { "epoch": 1.782801087593965, "grad_norm": 2.441685438156128, "learning_rate": 3.3122000000000002e-06, "loss": 0.0109, "step": 33440 }, { "epoch": 1.7828544010236178, "grad_norm": 0.989741325378418, "learning_rate": 3.3120000000000002e-06, "loss": 0.0173, "step": 33441 }, { "epoch": 1.7829077144532708, "grad_norm": 2.6253299713134766, "learning_rate": 3.3118e-06, "loss": 0.0284, "step": 33442 }, { "epoch": 1.7829610278829238, "grad_norm": 1.9744164943695068, "learning_rate": 3.3116e-06, "loss": 0.016, "step": 33443 }, { "epoch": 1.7830143413125765, "grad_norm": 1.815569281578064, "learning_rate": 3.3114e-06, "loss": 0.0234, "step": 33444 }, { "epoch": 1.7830676547422297, "grad_norm": 1.2510682344436646, "learning_rate": 3.3112000000000006e-06, "loss": 0.0245, "step": 33445 }, { "epoch": 1.7831209681718825, "grad_norm": 3.1356074810028076, "learning_rate": 3.311e-06, "loss": 0.0341, "step": 33446 }, { "epoch": 1.7831742816015355, "grad_norm": 1.2781888246536255, "learning_rate": 3.3108e-06, "loss": 0.0146, "step": 33447 }, { "epoch": 1.7832275950311884, "grad_norm": 1.2633904218673706, "learning_rate": 3.3106000000000005e-06, "loss": 0.0089, "step": 33448 }, { "epoch": 1.7832809084608412, "grad_norm": 2.02994441986084, "learning_rate": 3.3104e-06, "loss": 0.0238, "step": 33449 }, { "epoch": 1.7833342218904942, "grad_norm": 0.9464934468269348, "learning_rate": 3.3102e-06, "loss": 0.0062, "step": 33450 }, { "epoch": 1.7833875353201472, "grad_norm": 0.5158078670501709, "learning_rate": 3.3100000000000005e-06, "loss": 0.0049, "step": 33451 }, { "epoch": 1.7834408487498, "grad_norm": 1.196842074394226, "learning_rate": 3.3098e-06, "loss": 0.0279, "step": 33452 }, { "epoch": 1.7834941621794531, "grad_norm": 2.590357780456543, "learning_rate": 3.3096e-06, "loss": 0.0375, "step": 33453 }, { "epoch": 1.7835474756091059, "grad_norm": 1.809411883354187, "learning_rate": 3.3094000000000005e-06, "loss": 0.0404, "step": 33454 }, { "epoch": 1.7836007890387588, "grad_norm": 1.8075294494628906, "learning_rate": 3.3092e-06, "loss": 0.0294, "step": 33455 }, { "epoch": 1.7836541024684118, "grad_norm": 5.409849643707275, "learning_rate": 3.3090000000000004e-06, "loss": 0.0284, "step": 33456 }, { "epoch": 1.7837074158980646, "grad_norm": 1.4083201885223389, "learning_rate": 3.3088000000000004e-06, "loss": 0.0394, "step": 33457 }, { "epoch": 1.7837607293277178, "grad_norm": 3.0602333545684814, "learning_rate": 3.3086e-06, "loss": 0.0256, "step": 33458 }, { "epoch": 1.7838140427573705, "grad_norm": 2.013275146484375, "learning_rate": 3.3084000000000004e-06, "loss": 0.0229, "step": 33459 }, { "epoch": 1.7838673561870235, "grad_norm": 0.8468778133392334, "learning_rate": 3.3082000000000004e-06, "loss": 0.0049, "step": 33460 }, { "epoch": 1.7839206696166765, "grad_norm": 0.6410591006278992, "learning_rate": 3.308e-06, "loss": 0.0047, "step": 33461 }, { "epoch": 1.7839739830463293, "grad_norm": 2.0375559329986572, "learning_rate": 3.3078000000000003e-06, "loss": 0.0311, "step": 33462 }, { "epoch": 1.7840272964759825, "grad_norm": 2.68487548828125, "learning_rate": 3.3076000000000003e-06, "loss": 0.0213, "step": 33463 }, { "epoch": 1.7840806099056352, "grad_norm": 0.5603327751159668, "learning_rate": 3.3074e-06, "loss": 0.0042, "step": 33464 }, { "epoch": 1.7841339233352882, "grad_norm": 5.828818321228027, "learning_rate": 3.3072000000000003e-06, "loss": 0.038, "step": 33465 }, { "epoch": 1.7841872367649412, "grad_norm": 1.6939728260040283, "learning_rate": 3.3070000000000003e-06, "loss": 0.0142, "step": 33466 }, { "epoch": 1.784240550194594, "grad_norm": 1.6809608936309814, "learning_rate": 3.3068e-06, "loss": 0.0232, "step": 33467 }, { "epoch": 1.784293863624247, "grad_norm": 1.2792834043502808, "learning_rate": 3.3066000000000002e-06, "loss": 0.0178, "step": 33468 }, { "epoch": 1.7843471770539, "grad_norm": 3.148689031600952, "learning_rate": 3.3064000000000002e-06, "loss": 0.018, "step": 33469 }, { "epoch": 1.7844004904835526, "grad_norm": 0.4000178277492523, "learning_rate": 3.3062000000000006e-06, "loss": 0.0042, "step": 33470 }, { "epoch": 1.7844538039132058, "grad_norm": 1.663236141204834, "learning_rate": 3.306e-06, "loss": 0.0229, "step": 33471 }, { "epoch": 1.7845071173428586, "grad_norm": 1.3341639041900635, "learning_rate": 3.3058e-06, "loss": 0.0138, "step": 33472 }, { "epoch": 1.7845604307725116, "grad_norm": 4.321421146392822, "learning_rate": 3.3056000000000006e-06, "loss": 0.0213, "step": 33473 }, { "epoch": 1.7846137442021646, "grad_norm": 1.8089905977249146, "learning_rate": 3.3054e-06, "loss": 0.0301, "step": 33474 }, { "epoch": 1.7846670576318173, "grad_norm": 0.9073534607887268, "learning_rate": 3.3052e-06, "loss": 0.0186, "step": 33475 }, { "epoch": 1.7847203710614705, "grad_norm": 2.0969905853271484, "learning_rate": 3.3050000000000005e-06, "loss": 0.0329, "step": 33476 }, { "epoch": 1.7847736844911233, "grad_norm": 1.2165734767913818, "learning_rate": 3.3048e-06, "loss": 0.0174, "step": 33477 }, { "epoch": 1.7848269979207763, "grad_norm": 0.4284593164920807, "learning_rate": 3.3046e-06, "loss": 0.0037, "step": 33478 }, { "epoch": 1.7848803113504292, "grad_norm": 1.9301621913909912, "learning_rate": 3.3044000000000005e-06, "loss": 0.0528, "step": 33479 }, { "epoch": 1.784933624780082, "grad_norm": 1.7539069652557373, "learning_rate": 3.3042e-06, "loss": 0.0252, "step": 33480 }, { "epoch": 1.7849869382097352, "grad_norm": 1.2540459632873535, "learning_rate": 3.3040000000000005e-06, "loss": 0.0205, "step": 33481 }, { "epoch": 1.785040251639388, "grad_norm": 0.9041250348091125, "learning_rate": 3.3038000000000005e-06, "loss": 0.012, "step": 33482 }, { "epoch": 1.785093565069041, "grad_norm": 1.2611942291259766, "learning_rate": 3.3036e-06, "loss": 0.0279, "step": 33483 }, { "epoch": 1.785146878498694, "grad_norm": 2.161062240600586, "learning_rate": 3.3034000000000004e-06, "loss": 0.0111, "step": 33484 }, { "epoch": 1.7852001919283467, "grad_norm": 2.431035041809082, "learning_rate": 3.3032000000000004e-06, "loss": 0.0213, "step": 33485 }, { "epoch": 1.7852535053579996, "grad_norm": 0.9630917310714722, "learning_rate": 3.303e-06, "loss": 0.0145, "step": 33486 }, { "epoch": 1.7853068187876526, "grad_norm": 1.0037769079208374, "learning_rate": 3.3028000000000004e-06, "loss": 0.0203, "step": 33487 }, { "epoch": 1.7853601322173054, "grad_norm": 1.3564870357513428, "learning_rate": 3.3026000000000004e-06, "loss": 0.0078, "step": 33488 }, { "epoch": 1.7854134456469586, "grad_norm": 0.8118298649787903, "learning_rate": 3.3024e-06, "loss": 0.0076, "step": 33489 }, { "epoch": 1.7854667590766113, "grad_norm": 1.1332099437713623, "learning_rate": 3.3022000000000003e-06, "loss": 0.0152, "step": 33490 }, { "epoch": 1.7855200725062643, "grad_norm": 1.4529093503952026, "learning_rate": 3.3020000000000003e-06, "loss": 0.0141, "step": 33491 }, { "epoch": 1.7855733859359173, "grad_norm": 1.5372262001037598, "learning_rate": 3.3018e-06, "loss": 0.0367, "step": 33492 }, { "epoch": 1.78562669936557, "grad_norm": 0.9808405637741089, "learning_rate": 3.3016000000000003e-06, "loss": 0.0113, "step": 33493 }, { "epoch": 1.7856800127952233, "grad_norm": 1.0078186988830566, "learning_rate": 3.3014000000000003e-06, "loss": 0.0218, "step": 33494 }, { "epoch": 1.785733326224876, "grad_norm": 3.5380334854125977, "learning_rate": 3.3012000000000007e-06, "loss": 0.0261, "step": 33495 }, { "epoch": 1.785786639654529, "grad_norm": 1.874860167503357, "learning_rate": 3.3010000000000002e-06, "loss": 0.0264, "step": 33496 }, { "epoch": 1.785839953084182, "grad_norm": 4.1996355056762695, "learning_rate": 3.3008000000000002e-06, "loss": 0.0323, "step": 33497 }, { "epoch": 1.7858932665138347, "grad_norm": 2.873352527618408, "learning_rate": 3.3006000000000006e-06, "loss": 0.0301, "step": 33498 }, { "epoch": 1.7859465799434877, "grad_norm": 1.9861667156219482, "learning_rate": 3.3004e-06, "loss": 0.0253, "step": 33499 }, { "epoch": 1.7859998933731407, "grad_norm": 1.2563053369522095, "learning_rate": 3.3002e-06, "loss": 0.0082, "step": 33500 }, { "epoch": 1.7860532068027937, "grad_norm": 0.5004172921180725, "learning_rate": 3.3000000000000006e-06, "loss": 0.0036, "step": 33501 }, { "epoch": 1.7861065202324466, "grad_norm": 3.1463499069213867, "learning_rate": 3.2998e-06, "loss": 0.0462, "step": 33502 }, { "epoch": 1.7861598336620994, "grad_norm": 1.6130093336105347, "learning_rate": 3.2996e-06, "loss": 0.0433, "step": 33503 }, { "epoch": 1.7862131470917524, "grad_norm": 0.3748498558998108, "learning_rate": 3.2994000000000005e-06, "loss": 0.0035, "step": 33504 }, { "epoch": 1.7862664605214054, "grad_norm": 2.071789264678955, "learning_rate": 3.2992e-06, "loss": 0.0229, "step": 33505 }, { "epoch": 1.7863197739510581, "grad_norm": 1.2190345525741577, "learning_rate": 3.2990000000000005e-06, "loss": 0.0078, "step": 33506 }, { "epoch": 1.7863730873807113, "grad_norm": 1.7259955406188965, "learning_rate": 3.2988000000000005e-06, "loss": 0.0189, "step": 33507 }, { "epoch": 1.786426400810364, "grad_norm": 1.8967827558517456, "learning_rate": 3.2986e-06, "loss": 0.0237, "step": 33508 }, { "epoch": 1.786479714240017, "grad_norm": 4.8963117599487305, "learning_rate": 3.2984000000000005e-06, "loss": 0.0456, "step": 33509 }, { "epoch": 1.78653302766967, "grad_norm": 1.3788442611694336, "learning_rate": 3.2982000000000005e-06, "loss": 0.0205, "step": 33510 }, { "epoch": 1.7865863410993228, "grad_norm": 2.002501964569092, "learning_rate": 3.298e-06, "loss": 0.0412, "step": 33511 }, { "epoch": 1.786639654528976, "grad_norm": 1.5225523710250854, "learning_rate": 3.2978000000000004e-06, "loss": 0.0162, "step": 33512 }, { "epoch": 1.7866929679586288, "grad_norm": 2.024631977081299, "learning_rate": 3.2976000000000004e-06, "loss": 0.0165, "step": 33513 }, { "epoch": 1.7867462813882817, "grad_norm": 2.1127004623413086, "learning_rate": 3.2974e-06, "loss": 0.0509, "step": 33514 }, { "epoch": 1.7867995948179347, "grad_norm": 1.4962836503982544, "learning_rate": 3.2972000000000004e-06, "loss": 0.0093, "step": 33515 }, { "epoch": 1.7868529082475875, "grad_norm": 1.7120716571807861, "learning_rate": 3.2970000000000004e-06, "loss": 0.0209, "step": 33516 }, { "epoch": 1.7869062216772404, "grad_norm": 2.4107143878936768, "learning_rate": 3.2968e-06, "loss": 0.0274, "step": 33517 }, { "epoch": 1.7869595351068934, "grad_norm": 0.8167492747306824, "learning_rate": 3.2966000000000003e-06, "loss": 0.0079, "step": 33518 }, { "epoch": 1.7870128485365464, "grad_norm": 3.017005205154419, "learning_rate": 3.2964000000000003e-06, "loss": 0.0303, "step": 33519 }, { "epoch": 1.7870661619661994, "grad_norm": 1.9276819229125977, "learning_rate": 3.2962000000000003e-06, "loss": 0.0242, "step": 33520 }, { "epoch": 1.7871194753958521, "grad_norm": 0.8364802002906799, "learning_rate": 3.2960000000000003e-06, "loss": 0.0168, "step": 33521 }, { "epoch": 1.7871727888255051, "grad_norm": 1.4444572925567627, "learning_rate": 3.2958000000000003e-06, "loss": 0.0131, "step": 33522 }, { "epoch": 1.787226102255158, "grad_norm": 1.2767513990402222, "learning_rate": 3.2956000000000003e-06, "loss": 0.0172, "step": 33523 }, { "epoch": 1.7872794156848109, "grad_norm": 2.137373447418213, "learning_rate": 3.2954000000000002e-06, "loss": 0.0401, "step": 33524 }, { "epoch": 1.787332729114464, "grad_norm": 1.5791336297988892, "learning_rate": 3.2952000000000002e-06, "loss": 0.0107, "step": 33525 }, { "epoch": 1.7873860425441168, "grad_norm": 1.4652551412582397, "learning_rate": 3.2950000000000002e-06, "loss": 0.0102, "step": 33526 }, { "epoch": 1.7874393559737698, "grad_norm": 0.678852379322052, "learning_rate": 3.2948e-06, "loss": 0.0058, "step": 33527 }, { "epoch": 1.7874926694034228, "grad_norm": 1.86611807346344, "learning_rate": 3.2946e-06, "loss": 0.0152, "step": 33528 }, { "epoch": 1.7875459828330755, "grad_norm": 1.1838197708129883, "learning_rate": 3.2944e-06, "loss": 0.0166, "step": 33529 }, { "epoch": 1.7875992962627287, "grad_norm": 0.6877594590187073, "learning_rate": 3.2942e-06, "loss": 0.0063, "step": 33530 }, { "epoch": 1.7876526096923815, "grad_norm": 1.2585413455963135, "learning_rate": 3.2940000000000006e-06, "loss": 0.0176, "step": 33531 }, { "epoch": 1.7877059231220345, "grad_norm": 1.2999078035354614, "learning_rate": 3.2938e-06, "loss": 0.0074, "step": 33532 }, { "epoch": 1.7877592365516874, "grad_norm": 1.7305493354797363, "learning_rate": 3.2936e-06, "loss": 0.0233, "step": 33533 }, { "epoch": 1.7878125499813402, "grad_norm": 2.5400073528289795, "learning_rate": 3.2934000000000005e-06, "loss": 0.0414, "step": 33534 }, { "epoch": 1.7878658634109932, "grad_norm": 0.6171594858169556, "learning_rate": 3.2932e-06, "loss": 0.0045, "step": 33535 }, { "epoch": 1.7879191768406462, "grad_norm": 0.18397682905197144, "learning_rate": 3.293e-06, "loss": 0.0021, "step": 33536 }, { "epoch": 1.7879724902702991, "grad_norm": 0.8792884349822998, "learning_rate": 3.2928000000000005e-06, "loss": 0.009, "step": 33537 }, { "epoch": 1.7880258036999521, "grad_norm": 1.6378490924835205, "learning_rate": 3.2926e-06, "loss": 0.0271, "step": 33538 }, { "epoch": 1.7880791171296049, "grad_norm": 0.8179196119308472, "learning_rate": 3.2924e-06, "loss": 0.0075, "step": 33539 }, { "epoch": 1.7881324305592579, "grad_norm": 2.003431558609009, "learning_rate": 3.2922000000000004e-06, "loss": 0.0254, "step": 33540 }, { "epoch": 1.7881857439889108, "grad_norm": 1.7289901971817017, "learning_rate": 3.292e-06, "loss": 0.0356, "step": 33541 }, { "epoch": 1.7882390574185636, "grad_norm": 1.294614553451538, "learning_rate": 3.2918e-06, "loss": 0.0136, "step": 33542 }, { "epoch": 1.7882923708482168, "grad_norm": 1.5063765048980713, "learning_rate": 3.2916000000000004e-06, "loss": 0.0213, "step": 33543 }, { "epoch": 1.7883456842778696, "grad_norm": 0.7906374931335449, "learning_rate": 3.2914e-06, "loss": 0.0074, "step": 33544 }, { "epoch": 1.7883989977075225, "grad_norm": 1.2525606155395508, "learning_rate": 3.2912000000000004e-06, "loss": 0.0229, "step": 33545 }, { "epoch": 1.7884523111371755, "grad_norm": 0.8070080876350403, "learning_rate": 3.2910000000000003e-06, "loss": 0.0084, "step": 33546 }, { "epoch": 1.7885056245668283, "grad_norm": 4.1196699142456055, "learning_rate": 3.2908e-06, "loss": 0.0263, "step": 33547 }, { "epoch": 1.7885589379964815, "grad_norm": 2.2476911544799805, "learning_rate": 3.2906000000000003e-06, "loss": 0.0082, "step": 33548 }, { "epoch": 1.7886122514261342, "grad_norm": 0.9758096933364868, "learning_rate": 3.2904000000000003e-06, "loss": 0.0096, "step": 33549 }, { "epoch": 1.7886655648557872, "grad_norm": 1.0439261198043823, "learning_rate": 3.2902e-06, "loss": 0.005, "step": 33550 }, { "epoch": 1.7887188782854402, "grad_norm": 2.2148845195770264, "learning_rate": 3.2900000000000003e-06, "loss": 0.0242, "step": 33551 }, { "epoch": 1.788772191715093, "grad_norm": 1.5468335151672363, "learning_rate": 3.2898000000000002e-06, "loss": 0.0113, "step": 33552 }, { "epoch": 1.788825505144746, "grad_norm": 0.5395554304122925, "learning_rate": 3.2896e-06, "loss": 0.0045, "step": 33553 }, { "epoch": 1.788878818574399, "grad_norm": 2.0770649909973145, "learning_rate": 3.2894000000000002e-06, "loss": 0.0156, "step": 33554 }, { "epoch": 1.7889321320040517, "grad_norm": 0.9041592478752136, "learning_rate": 3.2892e-06, "loss": 0.0047, "step": 33555 }, { "epoch": 1.7889854454337049, "grad_norm": 0.9736970663070679, "learning_rate": 3.2890000000000006e-06, "loss": 0.0182, "step": 33556 }, { "epoch": 1.7890387588633576, "grad_norm": 1.5663154125213623, "learning_rate": 3.2888e-06, "loss": 0.0272, "step": 33557 }, { "epoch": 1.7890920722930106, "grad_norm": 1.445750117301941, "learning_rate": 3.2886e-06, "loss": 0.0127, "step": 33558 }, { "epoch": 1.7891453857226636, "grad_norm": 1.3570690155029297, "learning_rate": 3.2884000000000006e-06, "loss": 0.0093, "step": 33559 }, { "epoch": 1.7891986991523163, "grad_norm": 0.7036188840866089, "learning_rate": 3.2882e-06, "loss": 0.0176, "step": 33560 }, { "epoch": 1.7892520125819695, "grad_norm": 1.8663146495819092, "learning_rate": 3.288e-06, "loss": 0.032, "step": 33561 }, { "epoch": 1.7893053260116223, "grad_norm": 2.8591043949127197, "learning_rate": 3.2878000000000005e-06, "loss": 0.0404, "step": 33562 }, { "epoch": 1.7893586394412753, "grad_norm": 1.4238290786743164, "learning_rate": 3.2876e-06, "loss": 0.0089, "step": 33563 }, { "epoch": 1.7894119528709282, "grad_norm": 1.0871944427490234, "learning_rate": 3.2874e-06, "loss": 0.0182, "step": 33564 }, { "epoch": 1.789465266300581, "grad_norm": 1.3699973821640015, "learning_rate": 3.2872000000000005e-06, "loss": 0.0097, "step": 33565 }, { "epoch": 1.7895185797302342, "grad_norm": 1.1676896810531616, "learning_rate": 3.287e-06, "loss": 0.0185, "step": 33566 }, { "epoch": 1.789571893159887, "grad_norm": 1.713639259338379, "learning_rate": 3.2868000000000004e-06, "loss": 0.0286, "step": 33567 }, { "epoch": 1.78962520658954, "grad_norm": 1.7116812467575073, "learning_rate": 3.2866000000000004e-06, "loss": 0.009, "step": 33568 }, { "epoch": 1.789678520019193, "grad_norm": 1.988553524017334, "learning_rate": 3.2864e-06, "loss": 0.0124, "step": 33569 }, { "epoch": 1.7897318334488457, "grad_norm": 2.8596274852752686, "learning_rate": 3.2862000000000004e-06, "loss": 0.0241, "step": 33570 }, { "epoch": 1.7897851468784987, "grad_norm": 2.15877628326416, "learning_rate": 3.2860000000000004e-06, "loss": 0.0392, "step": 33571 }, { "epoch": 1.7898384603081516, "grad_norm": 1.6444305181503296, "learning_rate": 3.2858e-06, "loss": 0.0117, "step": 33572 }, { "epoch": 1.7898917737378044, "grad_norm": 0.6495144963264465, "learning_rate": 3.2856000000000004e-06, "loss": 0.0041, "step": 33573 }, { "epoch": 1.7899450871674576, "grad_norm": 2.7331221103668213, "learning_rate": 3.2854000000000003e-06, "loss": 0.0329, "step": 33574 }, { "epoch": 1.7899984005971104, "grad_norm": 2.8300294876098633, "learning_rate": 3.2852e-06, "loss": 0.0068, "step": 33575 }, { "epoch": 1.7900517140267633, "grad_norm": 1.8672475814819336, "learning_rate": 3.2850000000000003e-06, "loss": 0.0157, "step": 33576 }, { "epoch": 1.7901050274564163, "grad_norm": 0.8772729635238647, "learning_rate": 3.2848000000000003e-06, "loss": 0.0041, "step": 33577 }, { "epoch": 1.790158340886069, "grad_norm": 1.0757237672805786, "learning_rate": 3.2846e-06, "loss": 0.0057, "step": 33578 }, { "epoch": 1.7902116543157223, "grad_norm": 2.0270657539367676, "learning_rate": 3.2844000000000003e-06, "loss": 0.0112, "step": 33579 }, { "epoch": 1.790264967745375, "grad_norm": 2.9488401412963867, "learning_rate": 3.2842000000000003e-06, "loss": 0.035, "step": 33580 }, { "epoch": 1.790318281175028, "grad_norm": 1.6396909952163696, "learning_rate": 3.2840000000000007e-06, "loss": 0.0085, "step": 33581 }, { "epoch": 1.790371594604681, "grad_norm": 1.8132941722869873, "learning_rate": 3.2838000000000002e-06, "loss": 0.0289, "step": 33582 }, { "epoch": 1.7904249080343337, "grad_norm": 1.4767022132873535, "learning_rate": 3.2836e-06, "loss": 0.0126, "step": 33583 }, { "epoch": 1.790478221463987, "grad_norm": 2.3976941108703613, "learning_rate": 3.2834000000000006e-06, "loss": 0.0086, "step": 33584 }, { "epoch": 1.7905315348936397, "grad_norm": 1.302223801612854, "learning_rate": 3.2832e-06, "loss": 0.0123, "step": 33585 }, { "epoch": 1.7905848483232927, "grad_norm": 1.9497764110565186, "learning_rate": 3.283e-06, "loss": 0.0325, "step": 33586 }, { "epoch": 1.7906381617529457, "grad_norm": 3.021268367767334, "learning_rate": 3.2828000000000006e-06, "loss": 0.0351, "step": 33587 }, { "epoch": 1.7906914751825984, "grad_norm": 2.770524263381958, "learning_rate": 3.2826e-06, "loss": 0.0456, "step": 33588 }, { "epoch": 1.7907447886122514, "grad_norm": 1.487316608428955, "learning_rate": 3.2824e-06, "loss": 0.0075, "step": 33589 }, { "epoch": 1.7907981020419044, "grad_norm": 5.755264759063721, "learning_rate": 3.2822000000000005e-06, "loss": 0.0353, "step": 33590 }, { "epoch": 1.7908514154715571, "grad_norm": 1.270453691482544, "learning_rate": 3.282e-06, "loss": 0.0272, "step": 33591 }, { "epoch": 1.7909047289012103, "grad_norm": 1.7892519235610962, "learning_rate": 3.2818000000000005e-06, "loss": 0.0156, "step": 33592 }, { "epoch": 1.790958042330863, "grad_norm": 1.70553457736969, "learning_rate": 3.2816000000000005e-06, "loss": 0.0306, "step": 33593 }, { "epoch": 1.791011355760516, "grad_norm": 1.6841961145401, "learning_rate": 3.2814e-06, "loss": 0.032, "step": 33594 }, { "epoch": 1.791064669190169, "grad_norm": 0.8363085985183716, "learning_rate": 3.2812000000000004e-06, "loss": 0.0055, "step": 33595 }, { "epoch": 1.7911179826198218, "grad_norm": 1.6711547374725342, "learning_rate": 3.2810000000000004e-06, "loss": 0.0153, "step": 33596 }, { "epoch": 1.791171296049475, "grad_norm": 2.5621390342712402, "learning_rate": 3.2808e-06, "loss": 0.0429, "step": 33597 }, { "epoch": 1.7912246094791278, "grad_norm": 1.1915775537490845, "learning_rate": 3.2806000000000004e-06, "loss": 0.034, "step": 33598 }, { "epoch": 1.7912779229087807, "grad_norm": 0.4789223372936249, "learning_rate": 3.2804000000000004e-06, "loss": 0.003, "step": 33599 }, { "epoch": 1.7913312363384337, "grad_norm": 0.5223381519317627, "learning_rate": 3.2802e-06, "loss": 0.0041, "step": 33600 }, { "epoch": 1.7913845497680865, "grad_norm": 1.6568493843078613, "learning_rate": 3.2800000000000004e-06, "loss": 0.0241, "step": 33601 }, { "epoch": 1.7914378631977395, "grad_norm": 3.3530921936035156, "learning_rate": 3.2798000000000003e-06, "loss": 0.0326, "step": 33602 }, { "epoch": 1.7914911766273924, "grad_norm": 1.229162335395813, "learning_rate": 3.2796e-06, "loss": 0.0118, "step": 33603 }, { "epoch": 1.7915444900570454, "grad_norm": 1.6413521766662598, "learning_rate": 3.2794000000000003e-06, "loss": 0.0361, "step": 33604 }, { "epoch": 1.7915978034866984, "grad_norm": 1.5758271217346191, "learning_rate": 3.2792000000000003e-06, "loss": 0.0065, "step": 33605 }, { "epoch": 1.7916511169163511, "grad_norm": 1.9072248935699463, "learning_rate": 3.2790000000000007e-06, "loss": 0.0177, "step": 33606 }, { "epoch": 1.7917044303460041, "grad_norm": 1.605270266532898, "learning_rate": 3.2788000000000003e-06, "loss": 0.0079, "step": 33607 }, { "epoch": 1.791757743775657, "grad_norm": 1.0886086225509644, "learning_rate": 3.2786000000000003e-06, "loss": 0.0052, "step": 33608 }, { "epoch": 1.7918110572053099, "grad_norm": 2.041548490524292, "learning_rate": 3.2784000000000007e-06, "loss": 0.0176, "step": 33609 }, { "epoch": 1.791864370634963, "grad_norm": 1.157775640487671, "learning_rate": 3.2782000000000002e-06, "loss": 0.0242, "step": 33610 }, { "epoch": 1.7919176840646158, "grad_norm": 2.6931955814361572, "learning_rate": 3.278e-06, "loss": 0.0085, "step": 33611 }, { "epoch": 1.7919709974942688, "grad_norm": 2.7770001888275146, "learning_rate": 3.2778000000000006e-06, "loss": 0.0446, "step": 33612 }, { "epoch": 1.7920243109239218, "grad_norm": 2.265016555786133, "learning_rate": 3.2776e-06, "loss": 0.0285, "step": 33613 }, { "epoch": 1.7920776243535745, "grad_norm": 2.246467351913452, "learning_rate": 3.2774e-06, "loss": 0.0378, "step": 33614 }, { "epoch": 1.7921309377832277, "grad_norm": 0.3898361027240753, "learning_rate": 3.2772000000000006e-06, "loss": 0.0029, "step": 33615 }, { "epoch": 1.7921842512128805, "grad_norm": 1.3349438905715942, "learning_rate": 3.277e-06, "loss": 0.013, "step": 33616 }, { "epoch": 1.7922375646425335, "grad_norm": 1.4183335304260254, "learning_rate": 3.2768000000000005e-06, "loss": 0.0107, "step": 33617 }, { "epoch": 1.7922908780721865, "grad_norm": 2.193476915359497, "learning_rate": 3.2766000000000005e-06, "loss": 0.0333, "step": 33618 }, { "epoch": 1.7923441915018392, "grad_norm": 0.822479784488678, "learning_rate": 3.2764e-06, "loss": 0.0072, "step": 33619 }, { "epoch": 1.7923975049314922, "grad_norm": 1.1712980270385742, "learning_rate": 3.2762000000000005e-06, "loss": 0.0083, "step": 33620 }, { "epoch": 1.7924508183611452, "grad_norm": 0.274137407541275, "learning_rate": 3.2760000000000005e-06, "loss": 0.0015, "step": 33621 }, { "epoch": 1.7925041317907981, "grad_norm": 1.7565432786941528, "learning_rate": 3.2758e-06, "loss": 0.0251, "step": 33622 }, { "epoch": 1.7925574452204511, "grad_norm": 0.9195685386657715, "learning_rate": 3.2756000000000005e-06, "loss": 0.0134, "step": 33623 }, { "epoch": 1.7926107586501039, "grad_norm": 1.5776413679122925, "learning_rate": 3.2754000000000004e-06, "loss": 0.016, "step": 33624 }, { "epoch": 1.7926640720797569, "grad_norm": 0.14938884973526, "learning_rate": 3.2752e-06, "loss": 0.0014, "step": 33625 }, { "epoch": 1.7927173855094098, "grad_norm": 2.204531669616699, "learning_rate": 3.2750000000000004e-06, "loss": 0.0225, "step": 33626 }, { "epoch": 1.7927706989390626, "grad_norm": 1.5036046504974365, "learning_rate": 3.2748000000000004e-06, "loss": 0.0059, "step": 33627 }, { "epoch": 1.7928240123687158, "grad_norm": 1.4381834268569946, "learning_rate": 3.2746e-06, "loss": 0.0148, "step": 33628 }, { "epoch": 1.7928773257983686, "grad_norm": 1.9700788259506226, "learning_rate": 3.2744000000000004e-06, "loss": 0.0208, "step": 33629 }, { "epoch": 1.7929306392280215, "grad_norm": 1.7220594882965088, "learning_rate": 3.2742000000000003e-06, "loss": 0.0205, "step": 33630 }, { "epoch": 1.7929839526576745, "grad_norm": 7.183686256408691, "learning_rate": 3.2740000000000003e-06, "loss": 0.05, "step": 33631 }, { "epoch": 1.7930372660873273, "grad_norm": 1.9342914819717407, "learning_rate": 3.2738000000000003e-06, "loss": 0.0286, "step": 33632 }, { "epoch": 1.7930905795169805, "grad_norm": 1.4633171558380127, "learning_rate": 3.2736000000000003e-06, "loss": 0.0209, "step": 33633 }, { "epoch": 1.7931438929466332, "grad_norm": 2.3600785732269287, "learning_rate": 3.2734000000000003e-06, "loss": 0.0102, "step": 33634 }, { "epoch": 1.7931972063762862, "grad_norm": 2.3143317699432373, "learning_rate": 3.2732000000000003e-06, "loss": 0.0209, "step": 33635 }, { "epoch": 1.7932505198059392, "grad_norm": 1.3734713792800903, "learning_rate": 3.2730000000000003e-06, "loss": 0.0133, "step": 33636 }, { "epoch": 1.793303833235592, "grad_norm": 1.1100239753723145, "learning_rate": 3.2728000000000002e-06, "loss": 0.0077, "step": 33637 }, { "epoch": 1.793357146665245, "grad_norm": 1.2003315687179565, "learning_rate": 3.2726000000000002e-06, "loss": 0.0408, "step": 33638 }, { "epoch": 1.793410460094898, "grad_norm": 2.992267608642578, "learning_rate": 3.2724000000000002e-06, "loss": 0.0204, "step": 33639 }, { "epoch": 1.7934637735245509, "grad_norm": 3.1310417652130127, "learning_rate": 3.2722e-06, "loss": 0.0196, "step": 33640 }, { "epoch": 1.7935170869542039, "grad_norm": 2.793816089630127, "learning_rate": 3.272e-06, "loss": 0.0306, "step": 33641 }, { "epoch": 1.7935704003838566, "grad_norm": 1.5554132461547852, "learning_rate": 3.2718000000000006e-06, "loss": 0.0075, "step": 33642 }, { "epoch": 1.7936237138135096, "grad_norm": 2.016880512237549, "learning_rate": 3.2716e-06, "loss": 0.0416, "step": 33643 }, { "epoch": 1.7936770272431626, "grad_norm": 0.4529995620250702, "learning_rate": 3.2714e-06, "loss": 0.0035, "step": 33644 }, { "epoch": 1.7937303406728153, "grad_norm": 2.7097713947296143, "learning_rate": 3.2712000000000005e-06, "loss": 0.0153, "step": 33645 }, { "epoch": 1.7937836541024685, "grad_norm": 3.723322868347168, "learning_rate": 3.271e-06, "loss": 0.0139, "step": 33646 }, { "epoch": 1.7938369675321213, "grad_norm": 1.9936280250549316, "learning_rate": 3.2708e-06, "loss": 0.0474, "step": 33647 }, { "epoch": 1.7938902809617743, "grad_norm": 3.2107553482055664, "learning_rate": 3.2706000000000005e-06, "loss": 0.0376, "step": 33648 }, { "epoch": 1.7939435943914273, "grad_norm": 1.0369501113891602, "learning_rate": 3.2704e-06, "loss": 0.0056, "step": 33649 }, { "epoch": 1.79399690782108, "grad_norm": 2.1366727352142334, "learning_rate": 3.2702e-06, "loss": 0.0537, "step": 33650 }, { "epoch": 1.7940502212507332, "grad_norm": 1.8751780986785889, "learning_rate": 3.2700000000000005e-06, "loss": 0.0288, "step": 33651 }, { "epoch": 1.794103534680386, "grad_norm": 0.9845770001411438, "learning_rate": 3.2698e-06, "loss": 0.0148, "step": 33652 }, { "epoch": 1.794156848110039, "grad_norm": 1.8172047138214111, "learning_rate": 3.2696e-06, "loss": 0.0286, "step": 33653 }, { "epoch": 1.794210161539692, "grad_norm": 2.0055153369903564, "learning_rate": 3.2694000000000004e-06, "loss": 0.0212, "step": 33654 }, { "epoch": 1.7942634749693447, "grad_norm": 0.918249785900116, "learning_rate": 3.2692e-06, "loss": 0.0045, "step": 33655 }, { "epoch": 1.7943167883989977, "grad_norm": 2.27897572517395, "learning_rate": 3.2690000000000004e-06, "loss": 0.0218, "step": 33656 }, { "epoch": 1.7943701018286506, "grad_norm": 0.31876081228256226, "learning_rate": 3.2688000000000004e-06, "loss": 0.0032, "step": 33657 }, { "epoch": 1.7944234152583034, "grad_norm": 1.9902405738830566, "learning_rate": 3.2686e-06, "loss": 0.0497, "step": 33658 }, { "epoch": 1.7944767286879566, "grad_norm": 1.2663060426712036, "learning_rate": 3.2684000000000003e-06, "loss": 0.009, "step": 33659 }, { "epoch": 1.7945300421176094, "grad_norm": 2.0116288661956787, "learning_rate": 3.2682000000000003e-06, "loss": 0.015, "step": 33660 }, { "epoch": 1.7945833555472623, "grad_norm": 2.3839149475097656, "learning_rate": 3.268e-06, "loss": 0.0146, "step": 33661 }, { "epoch": 1.7946366689769153, "grad_norm": 2.126305341720581, "learning_rate": 3.2678000000000003e-06, "loss": 0.0111, "step": 33662 }, { "epoch": 1.794689982406568, "grad_norm": 1.136905312538147, "learning_rate": 3.2676000000000003e-06, "loss": 0.0155, "step": 33663 }, { "epoch": 1.7947432958362213, "grad_norm": 2.4301936626434326, "learning_rate": 3.2674e-06, "loss": 0.0135, "step": 33664 }, { "epoch": 1.794796609265874, "grad_norm": 3.4104816913604736, "learning_rate": 3.2672000000000002e-06, "loss": 0.0308, "step": 33665 }, { "epoch": 1.794849922695527, "grad_norm": 1.265006422996521, "learning_rate": 3.2670000000000002e-06, "loss": 0.0074, "step": 33666 }, { "epoch": 1.79490323612518, "grad_norm": 1.4377734661102295, "learning_rate": 3.2668000000000006e-06, "loss": 0.0069, "step": 33667 }, { "epoch": 1.7949565495548327, "grad_norm": 1.2798177003860474, "learning_rate": 3.2666e-06, "loss": 0.0082, "step": 33668 }, { "epoch": 1.795009862984486, "grad_norm": 1.2062299251556396, "learning_rate": 3.2664e-06, "loss": 0.0069, "step": 33669 }, { "epoch": 1.7950631764141387, "grad_norm": 1.7356388568878174, "learning_rate": 3.2662000000000006e-06, "loss": 0.0257, "step": 33670 }, { "epoch": 1.7951164898437917, "grad_norm": 2.9754953384399414, "learning_rate": 3.266e-06, "loss": 0.0506, "step": 33671 }, { "epoch": 1.7951698032734447, "grad_norm": 2.291339159011841, "learning_rate": 3.2658e-06, "loss": 0.0457, "step": 33672 }, { "epoch": 1.7952231167030974, "grad_norm": 3.722299098968506, "learning_rate": 3.2656000000000005e-06, "loss": 0.0295, "step": 33673 }, { "epoch": 1.7952764301327504, "grad_norm": 2.3531172275543213, "learning_rate": 3.2654e-06, "loss": 0.0348, "step": 33674 }, { "epoch": 1.7953297435624034, "grad_norm": 1.9180899858474731, "learning_rate": 3.2652e-06, "loss": 0.024, "step": 33675 }, { "epoch": 1.7953830569920561, "grad_norm": 2.226827383041382, "learning_rate": 3.2650000000000005e-06, "loss": 0.0211, "step": 33676 }, { "epoch": 1.7954363704217093, "grad_norm": 1.2279235124588013, "learning_rate": 3.2648e-06, "loss": 0.0162, "step": 33677 }, { "epoch": 1.795489683851362, "grad_norm": 1.9952930212020874, "learning_rate": 3.2646e-06, "loss": 0.0179, "step": 33678 }, { "epoch": 1.795542997281015, "grad_norm": 2.936551332473755, "learning_rate": 3.2644000000000005e-06, "loss": 0.0345, "step": 33679 }, { "epoch": 1.795596310710668, "grad_norm": 0.423100084066391, "learning_rate": 3.2642e-06, "loss": 0.0046, "step": 33680 }, { "epoch": 1.7956496241403208, "grad_norm": 2.9878296852111816, "learning_rate": 3.2640000000000004e-06, "loss": 0.035, "step": 33681 }, { "epoch": 1.795702937569974, "grad_norm": 1.748307466506958, "learning_rate": 3.2638000000000004e-06, "loss": 0.0085, "step": 33682 }, { "epoch": 1.7957562509996268, "grad_norm": 3.4572243690490723, "learning_rate": 3.2636e-06, "loss": 0.0222, "step": 33683 }, { "epoch": 1.7958095644292797, "grad_norm": 1.472967267036438, "learning_rate": 3.2634000000000004e-06, "loss": 0.0137, "step": 33684 }, { "epoch": 1.7958628778589327, "grad_norm": 2.4794368743896484, "learning_rate": 3.2632000000000004e-06, "loss": 0.0176, "step": 33685 }, { "epoch": 1.7959161912885855, "grad_norm": 1.8647229671478271, "learning_rate": 3.263e-06, "loss": 0.0316, "step": 33686 }, { "epoch": 1.7959695047182387, "grad_norm": 1.6865205764770508, "learning_rate": 3.2628000000000003e-06, "loss": 0.0194, "step": 33687 }, { "epoch": 1.7960228181478914, "grad_norm": 1.5263746976852417, "learning_rate": 3.2626000000000003e-06, "loss": 0.0143, "step": 33688 }, { "epoch": 1.7960761315775444, "grad_norm": 5.996790409088135, "learning_rate": 3.2624e-06, "loss": 0.0217, "step": 33689 }, { "epoch": 1.7961294450071974, "grad_norm": 2.6711809635162354, "learning_rate": 3.2622000000000003e-06, "loss": 0.03, "step": 33690 }, { "epoch": 1.7961827584368502, "grad_norm": 1.1387256383895874, "learning_rate": 3.2620000000000003e-06, "loss": 0.0214, "step": 33691 }, { "epoch": 1.7962360718665031, "grad_norm": 0.27629634737968445, "learning_rate": 3.2618000000000007e-06, "loss": 0.0026, "step": 33692 }, { "epoch": 1.7962893852961561, "grad_norm": 2.5625839233398438, "learning_rate": 3.2616000000000002e-06, "loss": 0.035, "step": 33693 }, { "epoch": 1.7963426987258089, "grad_norm": 0.9915938377380371, "learning_rate": 3.2614000000000002e-06, "loss": 0.0125, "step": 33694 }, { "epoch": 1.796396012155462, "grad_norm": 2.847654104232788, "learning_rate": 3.2612000000000006e-06, "loss": 0.0577, "step": 33695 }, { "epoch": 1.7964493255851148, "grad_norm": 0.9160472750663757, "learning_rate": 3.261e-06, "loss": 0.0055, "step": 33696 }, { "epoch": 1.7965026390147678, "grad_norm": 1.3854442834854126, "learning_rate": 3.2608e-06, "loss": 0.0252, "step": 33697 }, { "epoch": 1.7965559524444208, "grad_norm": 2.361149549484253, "learning_rate": 3.2606000000000006e-06, "loss": 0.0154, "step": 33698 }, { "epoch": 1.7966092658740735, "grad_norm": 1.9719626903533936, "learning_rate": 3.2604e-06, "loss": 0.0304, "step": 33699 }, { "epoch": 1.7966625793037267, "grad_norm": 2.4414703845977783, "learning_rate": 3.2602e-06, "loss": 0.0357, "step": 33700 }, { "epoch": 1.7967158927333795, "grad_norm": 1.3857383728027344, "learning_rate": 3.2600000000000006e-06, "loss": 0.0089, "step": 33701 }, { "epoch": 1.7967692061630325, "grad_norm": 1.2565217018127441, "learning_rate": 3.2598e-06, "loss": 0.0223, "step": 33702 }, { "epoch": 1.7968225195926855, "grad_norm": 3.5087833404541016, "learning_rate": 3.2596000000000005e-06, "loss": 0.0441, "step": 33703 }, { "epoch": 1.7968758330223382, "grad_norm": 0.7562631964683533, "learning_rate": 3.2594000000000005e-06, "loss": 0.0067, "step": 33704 }, { "epoch": 1.7969291464519914, "grad_norm": 1.6577848196029663, "learning_rate": 3.2592e-06, "loss": 0.0299, "step": 33705 }, { "epoch": 1.7969824598816442, "grad_norm": 2.0854837894439697, "learning_rate": 3.2590000000000005e-06, "loss": 0.0189, "step": 33706 }, { "epoch": 1.7970357733112972, "grad_norm": 2.0075206756591797, "learning_rate": 3.2588000000000005e-06, "loss": 0.0208, "step": 33707 }, { "epoch": 1.7970890867409501, "grad_norm": 1.4486231803894043, "learning_rate": 3.2586e-06, "loss": 0.0171, "step": 33708 }, { "epoch": 1.797142400170603, "grad_norm": 2.1639647483825684, "learning_rate": 3.2584000000000004e-06, "loss": 0.0313, "step": 33709 }, { "epoch": 1.7971957136002559, "grad_norm": 1.2330611944198608, "learning_rate": 3.2582000000000004e-06, "loss": 0.0211, "step": 33710 }, { "epoch": 1.7972490270299089, "grad_norm": 1.8777209520339966, "learning_rate": 3.258e-06, "loss": 0.0151, "step": 33711 }, { "epoch": 1.7973023404595616, "grad_norm": 1.9042651653289795, "learning_rate": 3.2578000000000004e-06, "loss": 0.0559, "step": 33712 }, { "epoch": 1.7973556538892148, "grad_norm": 3.4391820430755615, "learning_rate": 3.2576000000000004e-06, "loss": 0.0323, "step": 33713 }, { "epoch": 1.7974089673188676, "grad_norm": 1.9776582717895508, "learning_rate": 3.2574e-06, "loss": 0.0401, "step": 33714 }, { "epoch": 1.7974622807485205, "grad_norm": 2.7606072425842285, "learning_rate": 3.2572000000000003e-06, "loss": 0.0286, "step": 33715 }, { "epoch": 1.7975155941781735, "grad_norm": 4.147571563720703, "learning_rate": 3.2570000000000003e-06, "loss": 0.0263, "step": 33716 }, { "epoch": 1.7975689076078263, "grad_norm": 1.102302074432373, "learning_rate": 3.2568000000000003e-06, "loss": 0.0098, "step": 33717 }, { "epoch": 1.7976222210374795, "grad_norm": 1.6915867328643799, "learning_rate": 3.2566000000000003e-06, "loss": 0.0096, "step": 33718 }, { "epoch": 1.7976755344671322, "grad_norm": 2.383465528488159, "learning_rate": 3.2564000000000003e-06, "loss": 0.0179, "step": 33719 }, { "epoch": 1.7977288478967852, "grad_norm": 2.7985129356384277, "learning_rate": 3.2562000000000003e-06, "loss": 0.03, "step": 33720 }, { "epoch": 1.7977821613264382, "grad_norm": 1.487136721611023, "learning_rate": 3.2560000000000003e-06, "loss": 0.0122, "step": 33721 }, { "epoch": 1.797835474756091, "grad_norm": 3.1457555294036865, "learning_rate": 3.2558000000000002e-06, "loss": 0.0372, "step": 33722 }, { "epoch": 1.797888788185744, "grad_norm": 2.070641279220581, "learning_rate": 3.2556000000000002e-06, "loss": 0.0179, "step": 33723 }, { "epoch": 1.797942101615397, "grad_norm": 1.742599368095398, "learning_rate": 3.2554e-06, "loss": 0.025, "step": 33724 }, { "epoch": 1.79799541504505, "grad_norm": 0.6503349542617798, "learning_rate": 3.2552e-06, "loss": 0.0061, "step": 33725 }, { "epoch": 1.7980487284747029, "grad_norm": 1.4155155420303345, "learning_rate": 3.255e-06, "loss": 0.0283, "step": 33726 }, { "epoch": 1.7981020419043556, "grad_norm": 0.4110739529132843, "learning_rate": 3.2548e-06, "loss": 0.005, "step": 33727 }, { "epoch": 1.7981553553340086, "grad_norm": 2.960336208343506, "learning_rate": 3.2546000000000006e-06, "loss": 0.0374, "step": 33728 }, { "epoch": 1.7982086687636616, "grad_norm": 2.3237464427948, "learning_rate": 3.2544e-06, "loss": 0.0166, "step": 33729 }, { "epoch": 1.7982619821933143, "grad_norm": 2.2984793186187744, "learning_rate": 3.2542e-06, "loss": 0.0349, "step": 33730 }, { "epoch": 1.7983152956229675, "grad_norm": 0.6099970936775208, "learning_rate": 3.2540000000000005e-06, "loss": 0.0056, "step": 33731 }, { "epoch": 1.7983686090526203, "grad_norm": 1.5159354209899902, "learning_rate": 3.2538e-06, "loss": 0.0114, "step": 33732 }, { "epoch": 1.7984219224822733, "grad_norm": 1.6208429336547852, "learning_rate": 3.2536e-06, "loss": 0.0273, "step": 33733 }, { "epoch": 1.7984752359119263, "grad_norm": 1.002622127532959, "learning_rate": 3.2534000000000005e-06, "loss": 0.0103, "step": 33734 }, { "epoch": 1.798528549341579, "grad_norm": 1.0213693380355835, "learning_rate": 3.2532e-06, "loss": 0.0068, "step": 33735 }, { "epoch": 1.7985818627712322, "grad_norm": 1.5064949989318848, "learning_rate": 3.253e-06, "loss": 0.0267, "step": 33736 }, { "epoch": 1.798635176200885, "grad_norm": 2.090958833694458, "learning_rate": 3.2528000000000004e-06, "loss": 0.0286, "step": 33737 }, { "epoch": 1.798688489630538, "grad_norm": 1.1256452798843384, "learning_rate": 3.2526e-06, "loss": 0.0107, "step": 33738 }, { "epoch": 1.798741803060191, "grad_norm": 2.4923925399780273, "learning_rate": 3.2524e-06, "loss": 0.0479, "step": 33739 }, { "epoch": 1.7987951164898437, "grad_norm": 2.595529317855835, "learning_rate": 3.2522000000000004e-06, "loss": 0.0571, "step": 33740 }, { "epoch": 1.7988484299194967, "grad_norm": 0.8335418701171875, "learning_rate": 3.252e-06, "loss": 0.0073, "step": 33741 }, { "epoch": 1.7989017433491497, "grad_norm": 1.9556268453598022, "learning_rate": 3.2518000000000004e-06, "loss": 0.0102, "step": 33742 }, { "epoch": 1.7989550567788026, "grad_norm": 1.9091362953186035, "learning_rate": 3.2516000000000003e-06, "loss": 0.0283, "step": 33743 }, { "epoch": 1.7990083702084556, "grad_norm": 1.8671718835830688, "learning_rate": 3.2514e-06, "loss": 0.0254, "step": 33744 }, { "epoch": 1.7990616836381084, "grad_norm": 1.7565001249313354, "learning_rate": 3.2512000000000003e-06, "loss": 0.0403, "step": 33745 }, { "epoch": 1.7991149970677613, "grad_norm": 2.426295518875122, "learning_rate": 3.2510000000000003e-06, "loss": 0.0269, "step": 33746 }, { "epoch": 1.7991683104974143, "grad_norm": 1.840049386024475, "learning_rate": 3.2508e-06, "loss": 0.0311, "step": 33747 }, { "epoch": 1.799221623927067, "grad_norm": 3.0148606300354004, "learning_rate": 3.2506000000000003e-06, "loss": 0.0288, "step": 33748 }, { "epoch": 1.7992749373567203, "grad_norm": 2.3921825885772705, "learning_rate": 3.2504000000000003e-06, "loss": 0.0146, "step": 33749 }, { "epoch": 1.799328250786373, "grad_norm": 1.4509342908859253, "learning_rate": 3.2502e-06, "loss": 0.0147, "step": 33750 }, { "epoch": 1.799381564216026, "grad_norm": 1.5024166107177734, "learning_rate": 3.2500000000000002e-06, "loss": 0.0328, "step": 33751 }, { "epoch": 1.799434877645679, "grad_norm": 0.9105391502380371, "learning_rate": 3.2498e-06, "loss": 0.0205, "step": 33752 }, { "epoch": 1.7994881910753318, "grad_norm": 1.914088249206543, "learning_rate": 3.2496000000000006e-06, "loss": 0.0261, "step": 33753 }, { "epoch": 1.799541504504985, "grad_norm": 2.1290969848632812, "learning_rate": 3.2494e-06, "loss": 0.0446, "step": 33754 }, { "epoch": 1.7995948179346377, "grad_norm": 1.8135812282562256, "learning_rate": 3.2492e-06, "loss": 0.0115, "step": 33755 }, { "epoch": 1.7996481313642907, "grad_norm": 1.172554612159729, "learning_rate": 3.2490000000000006e-06, "loss": 0.0217, "step": 33756 }, { "epoch": 1.7997014447939437, "grad_norm": 1.2378724813461304, "learning_rate": 3.2488e-06, "loss": 0.0222, "step": 33757 }, { "epoch": 1.7997547582235964, "grad_norm": 2.6979520320892334, "learning_rate": 3.2486e-06, "loss": 0.0356, "step": 33758 }, { "epoch": 1.7998080716532494, "grad_norm": 1.1948437690734863, "learning_rate": 3.2484000000000005e-06, "loss": 0.016, "step": 33759 }, { "epoch": 1.7998613850829024, "grad_norm": 1.8293662071228027, "learning_rate": 3.2482e-06, "loss": 0.0145, "step": 33760 }, { "epoch": 1.7999146985125554, "grad_norm": 1.946602463722229, "learning_rate": 3.248e-06, "loss": 0.0271, "step": 33761 }, { "epoch": 1.7999680119422083, "grad_norm": 1.4243135452270508, "learning_rate": 3.2478000000000005e-06, "loss": 0.0159, "step": 33762 }, { "epoch": 1.800021325371861, "grad_norm": 1.0007383823394775, "learning_rate": 3.2476e-06, "loss": 0.0077, "step": 33763 }, { "epoch": 1.800074638801514, "grad_norm": 0.8948562145233154, "learning_rate": 3.2474e-06, "loss": 0.0074, "step": 33764 }, { "epoch": 1.800127952231167, "grad_norm": 1.4166511297225952, "learning_rate": 3.2472000000000004e-06, "loss": 0.014, "step": 33765 }, { "epoch": 1.8001812656608198, "grad_norm": 2.328854560852051, "learning_rate": 3.247e-06, "loss": 0.0235, "step": 33766 }, { "epoch": 1.800234579090473, "grad_norm": 0.8592893481254578, "learning_rate": 3.2468000000000004e-06, "loss": 0.0088, "step": 33767 }, { "epoch": 1.8002878925201258, "grad_norm": 0.9470330476760864, "learning_rate": 3.2466000000000004e-06, "loss": 0.0083, "step": 33768 }, { "epoch": 1.8003412059497788, "grad_norm": 1.3238492012023926, "learning_rate": 3.2464e-06, "loss": 0.0211, "step": 33769 }, { "epoch": 1.8003945193794317, "grad_norm": 1.4209129810333252, "learning_rate": 3.2462000000000004e-06, "loss": 0.0355, "step": 33770 }, { "epoch": 1.8004478328090845, "grad_norm": 2.082347869873047, "learning_rate": 3.2460000000000003e-06, "loss": 0.0472, "step": 33771 }, { "epoch": 1.8005011462387377, "grad_norm": 1.2690646648406982, "learning_rate": 3.2458e-06, "loss": 0.012, "step": 33772 }, { "epoch": 1.8005544596683905, "grad_norm": 2.411616802215576, "learning_rate": 3.2456000000000003e-06, "loss": 0.0458, "step": 33773 }, { "epoch": 1.8006077730980434, "grad_norm": 2.057184934616089, "learning_rate": 3.2454000000000003e-06, "loss": 0.0312, "step": 33774 }, { "epoch": 1.8006610865276964, "grad_norm": 1.1093716621398926, "learning_rate": 3.2452e-06, "loss": 0.0188, "step": 33775 }, { "epoch": 1.8007143999573492, "grad_norm": 1.0992964506149292, "learning_rate": 3.2450000000000003e-06, "loss": 0.0184, "step": 33776 }, { "epoch": 1.8007677133870021, "grad_norm": 1.2463173866271973, "learning_rate": 3.2448000000000003e-06, "loss": 0.0124, "step": 33777 }, { "epoch": 1.8008210268166551, "grad_norm": 1.4893373250961304, "learning_rate": 3.2446000000000007e-06, "loss": 0.0297, "step": 33778 }, { "epoch": 1.8008743402463079, "grad_norm": 2.889293909072876, "learning_rate": 3.2444000000000002e-06, "loss": 0.0195, "step": 33779 }, { "epoch": 1.800927653675961, "grad_norm": 1.4553042650222778, "learning_rate": 3.2442e-06, "loss": 0.0071, "step": 33780 }, { "epoch": 1.8009809671056138, "grad_norm": 0.9567527770996094, "learning_rate": 3.2440000000000006e-06, "loss": 0.0114, "step": 33781 }, { "epoch": 1.8010342805352668, "grad_norm": 1.0132228136062622, "learning_rate": 3.2438e-06, "loss": 0.0116, "step": 33782 }, { "epoch": 1.8010875939649198, "grad_norm": 1.6426138877868652, "learning_rate": 3.2436e-06, "loss": 0.0156, "step": 33783 }, { "epoch": 1.8011409073945726, "grad_norm": 3.8767497539520264, "learning_rate": 3.2434000000000006e-06, "loss": 0.0325, "step": 33784 }, { "epoch": 1.8011942208242258, "grad_norm": 3.170194387435913, "learning_rate": 3.2432e-06, "loss": 0.0293, "step": 33785 }, { "epoch": 1.8012475342538785, "grad_norm": 1.4399375915527344, "learning_rate": 3.243e-06, "loss": 0.0099, "step": 33786 }, { "epoch": 1.8013008476835315, "grad_norm": 0.7081854343414307, "learning_rate": 3.2428000000000005e-06, "loss": 0.0048, "step": 33787 }, { "epoch": 1.8013541611131845, "grad_norm": 1.203481912612915, "learning_rate": 3.2426e-06, "loss": 0.008, "step": 33788 }, { "epoch": 1.8014074745428372, "grad_norm": 2.1161985397338867, "learning_rate": 3.2424e-06, "loss": 0.0439, "step": 33789 }, { "epoch": 1.8014607879724904, "grad_norm": 0.39982500672340393, "learning_rate": 3.2422000000000005e-06, "loss": 0.0038, "step": 33790 }, { "epoch": 1.8015141014021432, "grad_norm": 1.432891607284546, "learning_rate": 3.242e-06, "loss": 0.0149, "step": 33791 }, { "epoch": 1.8015674148317962, "grad_norm": 1.696897029876709, "learning_rate": 3.2418000000000005e-06, "loss": 0.025, "step": 33792 }, { "epoch": 1.8016207282614491, "grad_norm": 1.0177055597305298, "learning_rate": 3.2416000000000004e-06, "loss": 0.0151, "step": 33793 }, { "epoch": 1.801674041691102, "grad_norm": 1.2827839851379395, "learning_rate": 3.2414e-06, "loss": 0.0197, "step": 33794 }, { "epoch": 1.8017273551207549, "grad_norm": 0.6538798809051514, "learning_rate": 3.2412000000000004e-06, "loss": 0.0048, "step": 33795 }, { "epoch": 1.8017806685504079, "grad_norm": 2.630423069000244, "learning_rate": 3.2410000000000004e-06, "loss": 0.0378, "step": 33796 }, { "epoch": 1.8018339819800606, "grad_norm": 2.333172559738159, "learning_rate": 3.2408e-06, "loss": 0.0243, "step": 33797 }, { "epoch": 1.8018872954097138, "grad_norm": 0.5209029912948608, "learning_rate": 3.2406000000000004e-06, "loss": 0.0054, "step": 33798 }, { "epoch": 1.8019406088393666, "grad_norm": 1.3649649620056152, "learning_rate": 3.2404000000000003e-06, "loss": 0.0124, "step": 33799 }, { "epoch": 1.8019939222690196, "grad_norm": 1.1779792308807373, "learning_rate": 3.2402e-06, "loss": 0.0137, "step": 33800 }, { "epoch": 1.8020472356986725, "grad_norm": 0.6310086846351624, "learning_rate": 3.2400000000000003e-06, "loss": 0.0044, "step": 33801 }, { "epoch": 1.8021005491283253, "grad_norm": 1.5007259845733643, "learning_rate": 3.2398000000000003e-06, "loss": 0.0197, "step": 33802 }, { "epoch": 1.8021538625579785, "grad_norm": 1.749208688735962, "learning_rate": 3.2396000000000007e-06, "loss": 0.0129, "step": 33803 }, { "epoch": 1.8022071759876312, "grad_norm": 1.9748988151550293, "learning_rate": 3.2394000000000003e-06, "loss": 0.0298, "step": 33804 }, { "epoch": 1.8022604894172842, "grad_norm": 3.0366978645324707, "learning_rate": 3.2392000000000003e-06, "loss": 0.0235, "step": 33805 }, { "epoch": 1.8023138028469372, "grad_norm": 1.3659833669662476, "learning_rate": 3.2390000000000007e-06, "loss": 0.0271, "step": 33806 }, { "epoch": 1.80236711627659, "grad_norm": 1.11345374584198, "learning_rate": 3.2388000000000002e-06, "loss": 0.0153, "step": 33807 }, { "epoch": 1.8024204297062432, "grad_norm": 1.772839069366455, "learning_rate": 3.2386000000000002e-06, "loss": 0.0459, "step": 33808 }, { "epoch": 1.802473743135896, "grad_norm": 1.8874218463897705, "learning_rate": 3.2384000000000006e-06, "loss": 0.0146, "step": 33809 }, { "epoch": 1.802527056565549, "grad_norm": 2.0748229026794434, "learning_rate": 3.2382e-06, "loss": 0.0557, "step": 33810 }, { "epoch": 1.8025803699952019, "grad_norm": 2.123842477798462, "learning_rate": 3.238e-06, "loss": 0.039, "step": 33811 }, { "epoch": 1.8026336834248546, "grad_norm": 1.4234081506729126, "learning_rate": 3.2378000000000006e-06, "loss": 0.0077, "step": 33812 }, { "epoch": 1.8026869968545076, "grad_norm": 1.902348518371582, "learning_rate": 3.2376e-06, "loss": 0.0269, "step": 33813 }, { "epoch": 1.8027403102841606, "grad_norm": 0.6844799518585205, "learning_rate": 3.2374000000000005e-06, "loss": 0.0041, "step": 33814 }, { "epoch": 1.8027936237138134, "grad_norm": 1.4181666374206543, "learning_rate": 3.2372000000000005e-06, "loss": 0.0107, "step": 33815 }, { "epoch": 1.8028469371434666, "grad_norm": 0.7625568509101868, "learning_rate": 3.237e-06, "loss": 0.0073, "step": 33816 }, { "epoch": 1.8029002505731193, "grad_norm": 1.3712968826293945, "learning_rate": 3.2368000000000005e-06, "loss": 0.0144, "step": 33817 }, { "epoch": 1.8029535640027723, "grad_norm": 1.8142715692520142, "learning_rate": 3.2366000000000005e-06, "loss": 0.0163, "step": 33818 }, { "epoch": 1.8030068774324253, "grad_norm": 2.4535927772521973, "learning_rate": 3.2364e-06, "loss": 0.0138, "step": 33819 }, { "epoch": 1.803060190862078, "grad_norm": 2.23099422454834, "learning_rate": 3.2362000000000005e-06, "loss": 0.0264, "step": 33820 }, { "epoch": 1.8031135042917312, "grad_norm": 1.8953614234924316, "learning_rate": 3.2360000000000004e-06, "loss": 0.0347, "step": 33821 }, { "epoch": 1.803166817721384, "grad_norm": 1.6623884439468384, "learning_rate": 3.2358e-06, "loss": 0.0141, "step": 33822 }, { "epoch": 1.803220131151037, "grad_norm": 2.650728702545166, "learning_rate": 3.2356000000000004e-06, "loss": 0.0215, "step": 33823 }, { "epoch": 1.80327344458069, "grad_norm": 1.7971408367156982, "learning_rate": 3.2354000000000004e-06, "loss": 0.0236, "step": 33824 }, { "epoch": 1.8033267580103427, "grad_norm": 1.9312446117401123, "learning_rate": 3.2352e-06, "loss": 0.0149, "step": 33825 }, { "epoch": 1.8033800714399957, "grad_norm": 1.9781608581542969, "learning_rate": 3.2350000000000004e-06, "loss": 0.0255, "step": 33826 }, { "epoch": 1.8034333848696487, "grad_norm": 1.0179102420806885, "learning_rate": 3.2348000000000004e-06, "loss": 0.0071, "step": 33827 }, { "epoch": 1.8034866982993016, "grad_norm": 2.564207077026367, "learning_rate": 3.2346000000000003e-06, "loss": 0.015, "step": 33828 }, { "epoch": 1.8035400117289546, "grad_norm": 1.116994023323059, "learning_rate": 3.2344000000000003e-06, "loss": 0.0058, "step": 33829 }, { "epoch": 1.8035933251586074, "grad_norm": 2.164644241333008, "learning_rate": 3.2342000000000003e-06, "loss": 0.0151, "step": 33830 }, { "epoch": 1.8036466385882604, "grad_norm": 2.0309102535247803, "learning_rate": 3.2340000000000003e-06, "loss": 0.0208, "step": 33831 }, { "epoch": 1.8036999520179133, "grad_norm": 1.3602396249771118, "learning_rate": 3.2338000000000003e-06, "loss": 0.0079, "step": 33832 }, { "epoch": 1.803753265447566, "grad_norm": 2.361284017562866, "learning_rate": 3.2336000000000003e-06, "loss": 0.0116, "step": 33833 }, { "epoch": 1.8038065788772193, "grad_norm": 2.362804889678955, "learning_rate": 3.2334000000000002e-06, "loss": 0.0123, "step": 33834 }, { "epoch": 1.803859892306872, "grad_norm": 1.890030860900879, "learning_rate": 3.2332000000000002e-06, "loss": 0.0139, "step": 33835 }, { "epoch": 1.803913205736525, "grad_norm": 1.30953848361969, "learning_rate": 3.2330000000000002e-06, "loss": 0.0238, "step": 33836 }, { "epoch": 1.803966519166178, "grad_norm": 2.0599255561828613, "learning_rate": 3.2328e-06, "loss": 0.0083, "step": 33837 }, { "epoch": 1.8040198325958308, "grad_norm": 1.5267622470855713, "learning_rate": 3.2326e-06, "loss": 0.0159, "step": 33838 }, { "epoch": 1.804073146025484, "grad_norm": 1.900502324104309, "learning_rate": 3.2324000000000006e-06, "loss": 0.0121, "step": 33839 }, { "epoch": 1.8041264594551367, "grad_norm": 2.3192827701568604, "learning_rate": 3.2322e-06, "loss": 0.0325, "step": 33840 }, { "epoch": 1.8041797728847897, "grad_norm": 1.56510329246521, "learning_rate": 3.232e-06, "loss": 0.0082, "step": 33841 }, { "epoch": 1.8042330863144427, "grad_norm": 1.3057950735092163, "learning_rate": 3.2318000000000006e-06, "loss": 0.0091, "step": 33842 }, { "epoch": 1.8042863997440954, "grad_norm": 2.097642183303833, "learning_rate": 3.2316e-06, "loss": 0.0267, "step": 33843 }, { "epoch": 1.8043397131737484, "grad_norm": 1.4879555702209473, "learning_rate": 3.2314e-06, "loss": 0.0302, "step": 33844 }, { "epoch": 1.8043930266034014, "grad_norm": 2.3891208171844482, "learning_rate": 3.2312000000000005e-06, "loss": 0.0159, "step": 33845 }, { "epoch": 1.8044463400330544, "grad_norm": 1.8768938779830933, "learning_rate": 3.231e-06, "loss": 0.0122, "step": 33846 }, { "epoch": 1.8044996534627074, "grad_norm": 2.1756062507629395, "learning_rate": 3.2308e-06, "loss": 0.0105, "step": 33847 }, { "epoch": 1.80455296689236, "grad_norm": 1.349117398262024, "learning_rate": 3.2306000000000005e-06, "loss": 0.0094, "step": 33848 }, { "epoch": 1.804606280322013, "grad_norm": 2.5644898414611816, "learning_rate": 3.2304e-06, "loss": 0.0319, "step": 33849 }, { "epoch": 1.804659593751666, "grad_norm": 0.44671735167503357, "learning_rate": 3.2302e-06, "loss": 0.0022, "step": 33850 }, { "epoch": 1.8047129071813188, "grad_norm": 0.9514836072921753, "learning_rate": 3.2300000000000004e-06, "loss": 0.0053, "step": 33851 }, { "epoch": 1.804766220610972, "grad_norm": 2.1945810317993164, "learning_rate": 3.2298e-06, "loss": 0.0109, "step": 33852 }, { "epoch": 1.8048195340406248, "grad_norm": 1.1940737962722778, "learning_rate": 3.2296000000000004e-06, "loss": 0.0091, "step": 33853 }, { "epoch": 1.8048728474702778, "grad_norm": 0.8892451524734497, "learning_rate": 3.2294000000000004e-06, "loss": 0.0057, "step": 33854 }, { "epoch": 1.8049261608999307, "grad_norm": 1.3105798959732056, "learning_rate": 3.2292e-06, "loss": 0.009, "step": 33855 }, { "epoch": 1.8049794743295835, "grad_norm": 1.687182068824768, "learning_rate": 3.2290000000000003e-06, "loss": 0.0119, "step": 33856 }, { "epoch": 1.8050327877592367, "grad_norm": 1.2222262620925903, "learning_rate": 3.2288000000000003e-06, "loss": 0.0127, "step": 33857 }, { "epoch": 1.8050861011888895, "grad_norm": 1.9221925735473633, "learning_rate": 3.2286e-06, "loss": 0.0284, "step": 33858 }, { "epoch": 1.8051394146185424, "grad_norm": 2.189302921295166, "learning_rate": 3.2284000000000003e-06, "loss": 0.0096, "step": 33859 }, { "epoch": 1.8051927280481954, "grad_norm": 0.8602927327156067, "learning_rate": 3.2282000000000003e-06, "loss": 0.0131, "step": 33860 }, { "epoch": 1.8052460414778482, "grad_norm": 4.4243927001953125, "learning_rate": 3.228e-06, "loss": 0.0698, "step": 33861 }, { "epoch": 1.8052993549075012, "grad_norm": 1.5658730268478394, "learning_rate": 3.2278000000000002e-06, "loss": 0.016, "step": 33862 }, { "epoch": 1.8053526683371541, "grad_norm": 2.2074198722839355, "learning_rate": 3.2276000000000002e-06, "loss": 0.0431, "step": 33863 }, { "epoch": 1.805405981766807, "grad_norm": 0.7007256150245667, "learning_rate": 3.2274000000000006e-06, "loss": 0.0046, "step": 33864 }, { "epoch": 1.80545929519646, "grad_norm": 2.6059792041778564, "learning_rate": 3.2272e-06, "loss": 0.031, "step": 33865 }, { "epoch": 1.8055126086261128, "grad_norm": 1.6328760385513306, "learning_rate": 3.227e-06, "loss": 0.0153, "step": 33866 }, { "epoch": 1.8055659220557658, "grad_norm": 2.385035753250122, "learning_rate": 3.2268000000000006e-06, "loss": 0.0123, "step": 33867 }, { "epoch": 1.8056192354854188, "grad_norm": 0.3673168122768402, "learning_rate": 3.2266e-06, "loss": 0.0028, "step": 33868 }, { "epoch": 1.8056725489150716, "grad_norm": 1.450059175491333, "learning_rate": 3.2264e-06, "loss": 0.0114, "step": 33869 }, { "epoch": 1.8057258623447248, "grad_norm": 2.179258108139038, "learning_rate": 3.2262000000000006e-06, "loss": 0.0199, "step": 33870 }, { "epoch": 1.8057791757743775, "grad_norm": 2.8390111923217773, "learning_rate": 3.226e-06, "loss": 0.0468, "step": 33871 }, { "epoch": 1.8058324892040305, "grad_norm": 1.5556628704071045, "learning_rate": 3.2258e-06, "loss": 0.0054, "step": 33872 }, { "epoch": 1.8058858026336835, "grad_norm": 2.2092530727386475, "learning_rate": 3.2256000000000005e-06, "loss": 0.0218, "step": 33873 }, { "epoch": 1.8059391160633362, "grad_norm": 0.5099337100982666, "learning_rate": 3.2254e-06, "loss": 0.0025, "step": 33874 }, { "epoch": 1.8059924294929894, "grad_norm": 2.562985420227051, "learning_rate": 3.2252e-06, "loss": 0.0222, "step": 33875 }, { "epoch": 1.8060457429226422, "grad_norm": 1.2168294191360474, "learning_rate": 3.2250000000000005e-06, "loss": 0.0054, "step": 33876 }, { "epoch": 1.8060990563522952, "grad_norm": 2.9128506183624268, "learning_rate": 3.2248e-06, "loss": 0.0455, "step": 33877 }, { "epoch": 1.8061523697819482, "grad_norm": 1.6933363676071167, "learning_rate": 3.2246000000000004e-06, "loss": 0.028, "step": 33878 }, { "epoch": 1.806205683211601, "grad_norm": 1.8707925081253052, "learning_rate": 3.2244000000000004e-06, "loss": 0.0225, "step": 33879 }, { "epoch": 1.8062589966412539, "grad_norm": 1.465154767036438, "learning_rate": 3.2242e-06, "loss": 0.0106, "step": 33880 }, { "epoch": 1.8063123100709069, "grad_norm": 0.9692429304122925, "learning_rate": 3.2240000000000004e-06, "loss": 0.0049, "step": 33881 }, { "epoch": 1.8063656235005596, "grad_norm": 1.9737275838851929, "learning_rate": 3.2238000000000004e-06, "loss": 0.0233, "step": 33882 }, { "epoch": 1.8064189369302128, "grad_norm": 2.4350132942199707, "learning_rate": 3.2236e-06, "loss": 0.0164, "step": 33883 }, { "epoch": 1.8064722503598656, "grad_norm": 1.3562262058258057, "learning_rate": 3.2234000000000003e-06, "loss": 0.0099, "step": 33884 }, { "epoch": 1.8065255637895186, "grad_norm": 2.8615989685058594, "learning_rate": 3.2232000000000003e-06, "loss": 0.0248, "step": 33885 }, { "epoch": 1.8065788772191715, "grad_norm": 0.8981835246086121, "learning_rate": 3.223e-06, "loss": 0.0072, "step": 33886 }, { "epoch": 1.8066321906488243, "grad_norm": 1.8582701683044434, "learning_rate": 3.2228000000000003e-06, "loss": 0.0401, "step": 33887 }, { "epoch": 1.8066855040784775, "grad_norm": 0.6123285889625549, "learning_rate": 3.2226000000000003e-06, "loss": 0.0057, "step": 33888 }, { "epoch": 1.8067388175081303, "grad_norm": 1.6788427829742432, "learning_rate": 3.2224000000000007e-06, "loss": 0.0246, "step": 33889 }, { "epoch": 1.8067921309377832, "grad_norm": 1.685214877128601, "learning_rate": 3.2222000000000003e-06, "loss": 0.0098, "step": 33890 }, { "epoch": 1.8068454443674362, "grad_norm": 4.85205078125, "learning_rate": 3.2220000000000002e-06, "loss": 0.0264, "step": 33891 }, { "epoch": 1.806898757797089, "grad_norm": 1.0085898637771606, "learning_rate": 3.2218000000000006e-06, "loss": 0.0078, "step": 33892 }, { "epoch": 1.8069520712267422, "grad_norm": 2.1444859504699707, "learning_rate": 3.2216e-06, "loss": 0.0156, "step": 33893 }, { "epoch": 1.807005384656395, "grad_norm": 2.346641778945923, "learning_rate": 3.2214e-06, "loss": 0.0185, "step": 33894 }, { "epoch": 1.807058698086048, "grad_norm": 2.062178134918213, "learning_rate": 3.2212000000000006e-06, "loss": 0.0324, "step": 33895 }, { "epoch": 1.8071120115157009, "grad_norm": 2.531522035598755, "learning_rate": 3.221e-06, "loss": 0.0285, "step": 33896 }, { "epoch": 1.8071653249453536, "grad_norm": 1.143987774848938, "learning_rate": 3.2208e-06, "loss": 0.0068, "step": 33897 }, { "epoch": 1.8072186383750066, "grad_norm": 3.0419628620147705, "learning_rate": 3.2206000000000006e-06, "loss": 0.0402, "step": 33898 }, { "epoch": 1.8072719518046596, "grad_norm": 1.7056909799575806, "learning_rate": 3.2204e-06, "loss": 0.0196, "step": 33899 }, { "epoch": 1.8073252652343124, "grad_norm": 1.8856725692749023, "learning_rate": 3.2202e-06, "loss": 0.0151, "step": 33900 }, { "epoch": 1.8073785786639656, "grad_norm": 3.302560567855835, "learning_rate": 3.2200000000000005e-06, "loss": 0.0358, "step": 33901 }, { "epoch": 1.8074318920936183, "grad_norm": 1.2237663269042969, "learning_rate": 3.2198e-06, "loss": 0.0106, "step": 33902 }, { "epoch": 1.8074852055232713, "grad_norm": 1.97292959690094, "learning_rate": 3.2196000000000005e-06, "loss": 0.0213, "step": 33903 }, { "epoch": 1.8075385189529243, "grad_norm": 1.2732899188995361, "learning_rate": 3.2194000000000005e-06, "loss": 0.0064, "step": 33904 }, { "epoch": 1.807591832382577, "grad_norm": 2.444113254547119, "learning_rate": 3.2192e-06, "loss": 0.0317, "step": 33905 }, { "epoch": 1.8076451458122302, "grad_norm": 2.0309977531433105, "learning_rate": 3.2190000000000004e-06, "loss": 0.0256, "step": 33906 }, { "epoch": 1.807698459241883, "grad_norm": 2.2979042530059814, "learning_rate": 3.2188000000000004e-06, "loss": 0.0251, "step": 33907 }, { "epoch": 1.807751772671536, "grad_norm": 2.127419948577881, "learning_rate": 3.2186e-06, "loss": 0.0151, "step": 33908 }, { "epoch": 1.807805086101189, "grad_norm": 0.7667500376701355, "learning_rate": 3.2184000000000004e-06, "loss": 0.0026, "step": 33909 }, { "epoch": 1.8078583995308417, "grad_norm": 2.163317918777466, "learning_rate": 3.2182000000000004e-06, "loss": 0.0289, "step": 33910 }, { "epoch": 1.807911712960495, "grad_norm": 1.1810053586959839, "learning_rate": 3.218e-06, "loss": 0.0074, "step": 33911 }, { "epoch": 1.8079650263901477, "grad_norm": 1.111981749534607, "learning_rate": 3.2178000000000003e-06, "loss": 0.0232, "step": 33912 }, { "epoch": 1.8080183398198006, "grad_norm": 5.46804141998291, "learning_rate": 3.2176000000000003e-06, "loss": 0.0124, "step": 33913 }, { "epoch": 1.8080716532494536, "grad_norm": 1.101504921913147, "learning_rate": 3.2174000000000003e-06, "loss": 0.007, "step": 33914 }, { "epoch": 1.8081249666791064, "grad_norm": 1.814378261566162, "learning_rate": 3.2172000000000003e-06, "loss": 0.0286, "step": 33915 }, { "epoch": 1.8081782801087594, "grad_norm": 0.46701177954673767, "learning_rate": 3.2170000000000003e-06, "loss": 0.0032, "step": 33916 }, { "epoch": 1.8082315935384123, "grad_norm": 2.317141532897949, "learning_rate": 3.2168000000000003e-06, "loss": 0.0149, "step": 33917 }, { "epoch": 1.808284906968065, "grad_norm": 1.5662691593170166, "learning_rate": 3.2166000000000003e-06, "loss": 0.0487, "step": 33918 }, { "epoch": 1.8083382203977183, "grad_norm": 2.106555461883545, "learning_rate": 3.2164000000000002e-06, "loss": 0.0334, "step": 33919 }, { "epoch": 1.808391533827371, "grad_norm": 0.7247359752655029, "learning_rate": 3.2162000000000002e-06, "loss": 0.004, "step": 33920 }, { "epoch": 1.808444847257024, "grad_norm": 1.5038340091705322, "learning_rate": 3.216e-06, "loss": 0.0177, "step": 33921 }, { "epoch": 1.808498160686677, "grad_norm": 0.9984986782073975, "learning_rate": 3.2158e-06, "loss": 0.0102, "step": 33922 }, { "epoch": 1.8085514741163298, "grad_norm": 1.114933967590332, "learning_rate": 3.2156e-06, "loss": 0.0115, "step": 33923 }, { "epoch": 1.808604787545983, "grad_norm": 1.9266146421432495, "learning_rate": 3.2154e-06, "loss": 0.0412, "step": 33924 }, { "epoch": 1.8086581009756357, "grad_norm": 1.530163288116455, "learning_rate": 3.2152e-06, "loss": 0.0179, "step": 33925 }, { "epoch": 1.8087114144052887, "grad_norm": 2.323977470397949, "learning_rate": 3.215e-06, "loss": 0.0197, "step": 33926 }, { "epoch": 1.8087647278349417, "grad_norm": 2.269260883331299, "learning_rate": 3.2148e-06, "loss": 0.0514, "step": 33927 }, { "epoch": 1.8088180412645944, "grad_norm": 1.10464608669281, "learning_rate": 3.2146000000000005e-06, "loss": 0.0197, "step": 33928 }, { "epoch": 1.8088713546942476, "grad_norm": 1.193960428237915, "learning_rate": 3.2144e-06, "loss": 0.0102, "step": 33929 }, { "epoch": 1.8089246681239004, "grad_norm": 2.168773651123047, "learning_rate": 3.2142e-06, "loss": 0.0263, "step": 33930 }, { "epoch": 1.8089779815535534, "grad_norm": 2.3228297233581543, "learning_rate": 3.2140000000000005e-06, "loss": 0.0282, "step": 33931 }, { "epoch": 1.8090312949832064, "grad_norm": 0.8794313073158264, "learning_rate": 3.2138e-06, "loss": 0.007, "step": 33932 }, { "epoch": 1.8090846084128591, "grad_norm": 3.6490533351898193, "learning_rate": 3.2136e-06, "loss": 0.0326, "step": 33933 }, { "epoch": 1.809137921842512, "grad_norm": 1.4924890995025635, "learning_rate": 3.2134000000000004e-06, "loss": 0.0133, "step": 33934 }, { "epoch": 1.809191235272165, "grad_norm": 1.4274992942810059, "learning_rate": 3.2132e-06, "loss": 0.011, "step": 33935 }, { "epoch": 1.8092445487018178, "grad_norm": 1.355988621711731, "learning_rate": 3.213e-06, "loss": 0.0067, "step": 33936 }, { "epoch": 1.809297862131471, "grad_norm": 2.901843309402466, "learning_rate": 3.2128000000000004e-06, "loss": 0.0386, "step": 33937 }, { "epoch": 1.8093511755611238, "grad_norm": 2.998269557952881, "learning_rate": 3.2126e-06, "loss": 0.0262, "step": 33938 }, { "epoch": 1.8094044889907768, "grad_norm": 1.0264374017715454, "learning_rate": 3.2124000000000004e-06, "loss": 0.016, "step": 33939 }, { "epoch": 1.8094578024204298, "grad_norm": 1.4571361541748047, "learning_rate": 3.2122000000000003e-06, "loss": 0.0288, "step": 33940 }, { "epoch": 1.8095111158500825, "grad_norm": 1.9130584001541138, "learning_rate": 3.212e-06, "loss": 0.0044, "step": 33941 }, { "epoch": 1.8095644292797357, "grad_norm": 1.4217963218688965, "learning_rate": 3.2118000000000003e-06, "loss": 0.0136, "step": 33942 }, { "epoch": 1.8096177427093885, "grad_norm": 3.079078435897827, "learning_rate": 3.2116000000000003e-06, "loss": 0.0208, "step": 33943 }, { "epoch": 1.8096710561390414, "grad_norm": 2.5818185806274414, "learning_rate": 3.2114e-06, "loss": 0.0297, "step": 33944 }, { "epoch": 1.8097243695686944, "grad_norm": 2.0384230613708496, "learning_rate": 3.2112000000000003e-06, "loss": 0.0304, "step": 33945 }, { "epoch": 1.8097776829983472, "grad_norm": 0.9311423301696777, "learning_rate": 3.2110000000000003e-06, "loss": 0.0049, "step": 33946 }, { "epoch": 1.8098309964280002, "grad_norm": 3.7185516357421875, "learning_rate": 3.2108e-06, "loss": 0.0332, "step": 33947 }, { "epoch": 1.8098843098576531, "grad_norm": 1.8508819341659546, "learning_rate": 3.2106000000000002e-06, "loss": 0.0138, "step": 33948 }, { "epoch": 1.8099376232873061, "grad_norm": 1.7501825094223022, "learning_rate": 3.2104000000000002e-06, "loss": 0.0379, "step": 33949 }, { "epoch": 1.809990936716959, "grad_norm": 1.165260910987854, "learning_rate": 3.2102000000000006e-06, "loss": 0.0042, "step": 33950 }, { "epoch": 1.8100442501466119, "grad_norm": 1.6624701023101807, "learning_rate": 3.21e-06, "loss": 0.0216, "step": 33951 }, { "epoch": 1.8100975635762648, "grad_norm": 1.3738110065460205, "learning_rate": 3.2098e-06, "loss": 0.0083, "step": 33952 }, { "epoch": 1.8101508770059178, "grad_norm": 0.7992273569107056, "learning_rate": 3.2096000000000006e-06, "loss": 0.0148, "step": 33953 }, { "epoch": 1.8102041904355706, "grad_norm": 2.4632580280303955, "learning_rate": 3.2094e-06, "loss": 0.0593, "step": 33954 }, { "epoch": 1.8102575038652238, "grad_norm": 1.545745849609375, "learning_rate": 3.2092e-06, "loss": 0.0317, "step": 33955 }, { "epoch": 1.8103108172948765, "grad_norm": 1.4611320495605469, "learning_rate": 3.2090000000000005e-06, "loss": 0.0154, "step": 33956 }, { "epoch": 1.8103641307245295, "grad_norm": 1.7568039894104004, "learning_rate": 3.2088e-06, "loss": 0.0212, "step": 33957 }, { "epoch": 1.8104174441541825, "grad_norm": 2.1246676445007324, "learning_rate": 3.2086e-06, "loss": 0.0115, "step": 33958 }, { "epoch": 1.8104707575838352, "grad_norm": 0.1909513920545578, "learning_rate": 3.2084000000000005e-06, "loss": 0.002, "step": 33959 }, { "epoch": 1.8105240710134884, "grad_norm": 0.25259116291999817, "learning_rate": 3.2082e-06, "loss": 0.0028, "step": 33960 }, { "epoch": 1.8105773844431412, "grad_norm": 1.4842082262039185, "learning_rate": 3.208e-06, "loss": 0.0093, "step": 33961 }, { "epoch": 1.8106306978727942, "grad_norm": 0.9495204091072083, "learning_rate": 3.2078000000000004e-06, "loss": 0.0156, "step": 33962 }, { "epoch": 1.8106840113024472, "grad_norm": 3.878112554550171, "learning_rate": 3.2076e-06, "loss": 0.0321, "step": 33963 }, { "epoch": 1.8107373247321, "grad_norm": 1.691514253616333, "learning_rate": 3.2074000000000004e-06, "loss": 0.0204, "step": 33964 }, { "epoch": 1.810790638161753, "grad_norm": 0.6178457140922546, "learning_rate": 3.2072000000000004e-06, "loss": 0.0049, "step": 33965 }, { "epoch": 1.8108439515914059, "grad_norm": 2.8627560138702393, "learning_rate": 3.207e-06, "loss": 0.0188, "step": 33966 }, { "epoch": 1.8108972650210589, "grad_norm": 1.625406265258789, "learning_rate": 3.2068000000000004e-06, "loss": 0.0255, "step": 33967 }, { "epoch": 1.8109505784507118, "grad_norm": 5.710720062255859, "learning_rate": 3.2066000000000004e-06, "loss": 0.0352, "step": 33968 }, { "epoch": 1.8110038918803646, "grad_norm": 2.559443235397339, "learning_rate": 3.2064e-06, "loss": 0.0301, "step": 33969 }, { "epoch": 1.8110572053100176, "grad_norm": 2.3131580352783203, "learning_rate": 3.2062000000000003e-06, "loss": 0.0291, "step": 33970 }, { "epoch": 1.8111105187396705, "grad_norm": 2.0068278312683105, "learning_rate": 3.2060000000000003e-06, "loss": 0.03, "step": 33971 }, { "epoch": 1.8111638321693233, "grad_norm": 0.7871988415718079, "learning_rate": 3.2058e-06, "loss": 0.0068, "step": 33972 }, { "epoch": 1.8112171455989765, "grad_norm": 1.8158090114593506, "learning_rate": 3.2056000000000003e-06, "loss": 0.0274, "step": 33973 }, { "epoch": 1.8112704590286293, "grad_norm": 2.537193775177002, "learning_rate": 3.2054000000000003e-06, "loss": 0.0185, "step": 33974 }, { "epoch": 1.8113237724582822, "grad_norm": 2.281341314315796, "learning_rate": 3.2052000000000007e-06, "loss": 0.0372, "step": 33975 }, { "epoch": 1.8113770858879352, "grad_norm": 1.8466943502426147, "learning_rate": 3.2050000000000002e-06, "loss": 0.0092, "step": 33976 }, { "epoch": 1.811430399317588, "grad_norm": 1.3974794149398804, "learning_rate": 3.2048000000000002e-06, "loss": 0.029, "step": 33977 }, { "epoch": 1.8114837127472412, "grad_norm": 1.6991785764694214, "learning_rate": 3.2046000000000006e-06, "loss": 0.0364, "step": 33978 }, { "epoch": 1.811537026176894, "grad_norm": 2.1075592041015625, "learning_rate": 3.2044e-06, "loss": 0.0185, "step": 33979 }, { "epoch": 1.811590339606547, "grad_norm": 1.9241492748260498, "learning_rate": 3.2042e-06, "loss": 0.016, "step": 33980 }, { "epoch": 1.8116436530362, "grad_norm": 1.7863619327545166, "learning_rate": 3.2040000000000006e-06, "loss": 0.0143, "step": 33981 }, { "epoch": 1.8116969664658527, "grad_norm": 1.2507187128067017, "learning_rate": 3.2038e-06, "loss": 0.0246, "step": 33982 }, { "epoch": 1.8117502798955056, "grad_norm": 2.1435444355010986, "learning_rate": 3.2036e-06, "loss": 0.0137, "step": 33983 }, { "epoch": 1.8118035933251586, "grad_norm": 1.9386869668960571, "learning_rate": 3.2034000000000005e-06, "loss": 0.0338, "step": 33984 }, { "epoch": 1.8118569067548116, "grad_norm": 2.159367561340332, "learning_rate": 3.2032e-06, "loss": 0.0155, "step": 33985 }, { "epoch": 1.8119102201844646, "grad_norm": 1.3896979093551636, "learning_rate": 3.203e-06, "loss": 0.0202, "step": 33986 }, { "epoch": 1.8119635336141173, "grad_norm": 1.4158384799957275, "learning_rate": 3.2028000000000005e-06, "loss": 0.0147, "step": 33987 }, { "epoch": 1.8120168470437703, "grad_norm": 0.8241666555404663, "learning_rate": 3.2026e-06, "loss": 0.0094, "step": 33988 }, { "epoch": 1.8120701604734233, "grad_norm": 1.6381381750106812, "learning_rate": 3.2024000000000005e-06, "loss": 0.0387, "step": 33989 }, { "epoch": 1.812123473903076, "grad_norm": 2.1916518211364746, "learning_rate": 3.2022000000000004e-06, "loss": 0.0351, "step": 33990 }, { "epoch": 1.8121767873327292, "grad_norm": 0.5871955752372742, "learning_rate": 3.202e-06, "loss": 0.0034, "step": 33991 }, { "epoch": 1.812230100762382, "grad_norm": 0.7297532558441162, "learning_rate": 3.2018000000000004e-06, "loss": 0.0049, "step": 33992 }, { "epoch": 1.812283414192035, "grad_norm": 1.2735013961791992, "learning_rate": 3.2016000000000004e-06, "loss": 0.0118, "step": 33993 }, { "epoch": 1.812336727621688, "grad_norm": 1.4826358556747437, "learning_rate": 3.2014e-06, "loss": 0.0221, "step": 33994 }, { "epoch": 1.8123900410513407, "grad_norm": 1.8712961673736572, "learning_rate": 3.2012000000000004e-06, "loss": 0.0305, "step": 33995 }, { "epoch": 1.812443354480994, "grad_norm": 2.6596431732177734, "learning_rate": 3.2010000000000004e-06, "loss": 0.0611, "step": 33996 }, { "epoch": 1.8124966679106467, "grad_norm": 2.089203357696533, "learning_rate": 3.2008e-06, "loss": 0.0195, "step": 33997 }, { "epoch": 1.8125499813402997, "grad_norm": 2.3124277591705322, "learning_rate": 3.2006000000000003e-06, "loss": 0.0263, "step": 33998 }, { "epoch": 1.8126032947699526, "grad_norm": 1.7690421342849731, "learning_rate": 3.2004000000000003e-06, "loss": 0.03, "step": 33999 }, { "epoch": 1.8126566081996054, "grad_norm": 1.4110838174819946, "learning_rate": 3.2002000000000007e-06, "loss": 0.0082, "step": 34000 }, { "epoch": 1.8127099216292584, "grad_norm": 2.1216766834259033, "learning_rate": 3.2000000000000003e-06, "loss": 0.0303, "step": 34001 }, { "epoch": 1.8127632350589113, "grad_norm": 0.5312390923500061, "learning_rate": 3.1998000000000003e-06, "loss": 0.0046, "step": 34002 }, { "epoch": 1.812816548488564, "grad_norm": 0.5425786972045898, "learning_rate": 3.1996000000000007e-06, "loss": 0.0049, "step": 34003 }, { "epoch": 1.8128698619182173, "grad_norm": 2.2420341968536377, "learning_rate": 3.1994000000000002e-06, "loss": 0.0265, "step": 34004 }, { "epoch": 1.81292317534787, "grad_norm": 4.211817264556885, "learning_rate": 3.1992000000000002e-06, "loss": 0.0278, "step": 34005 }, { "epoch": 1.812976488777523, "grad_norm": 1.7001951932907104, "learning_rate": 3.1990000000000006e-06, "loss": 0.0108, "step": 34006 }, { "epoch": 1.813029802207176, "grad_norm": 1.61758291721344, "learning_rate": 3.1988e-06, "loss": 0.0195, "step": 34007 }, { "epoch": 1.8130831156368288, "grad_norm": 0.9229175448417664, "learning_rate": 3.1986e-06, "loss": 0.0069, "step": 34008 }, { "epoch": 1.813136429066482, "grad_norm": 2.3781094551086426, "learning_rate": 3.1984000000000006e-06, "loss": 0.022, "step": 34009 }, { "epoch": 1.8131897424961347, "grad_norm": 1.3484026193618774, "learning_rate": 3.1982e-06, "loss": 0.0088, "step": 34010 }, { "epoch": 1.8132430559257877, "grad_norm": 2.1074016094207764, "learning_rate": 3.198e-06, "loss": 0.0401, "step": 34011 }, { "epoch": 1.8132963693554407, "grad_norm": 1.4462755918502808, "learning_rate": 3.1978000000000005e-06, "loss": 0.0154, "step": 34012 }, { "epoch": 1.8133496827850935, "grad_norm": 1.9123201370239258, "learning_rate": 3.1976e-06, "loss": 0.0309, "step": 34013 }, { "epoch": 1.8134029962147467, "grad_norm": 1.1820298433303833, "learning_rate": 3.1974000000000005e-06, "loss": 0.0182, "step": 34014 }, { "epoch": 1.8134563096443994, "grad_norm": 2.322549343109131, "learning_rate": 3.1972000000000005e-06, "loss": 0.0127, "step": 34015 }, { "epoch": 1.8135096230740524, "grad_norm": 7.095150470733643, "learning_rate": 3.197e-06, "loss": 0.0524, "step": 34016 }, { "epoch": 1.8135629365037054, "grad_norm": 2.20192813873291, "learning_rate": 3.1968000000000005e-06, "loss": 0.0257, "step": 34017 }, { "epoch": 1.8136162499333581, "grad_norm": 1.2035804986953735, "learning_rate": 3.1966000000000004e-06, "loss": 0.0167, "step": 34018 }, { "epoch": 1.813669563363011, "grad_norm": 1.5062187910079956, "learning_rate": 3.1964e-06, "loss": 0.0095, "step": 34019 }, { "epoch": 1.813722876792664, "grad_norm": 2.2252449989318848, "learning_rate": 3.1962000000000004e-06, "loss": 0.0301, "step": 34020 }, { "epoch": 1.8137761902223168, "grad_norm": 1.480105996131897, "learning_rate": 3.1960000000000004e-06, "loss": 0.0248, "step": 34021 }, { "epoch": 1.81382950365197, "grad_norm": 2.1862525939941406, "learning_rate": 3.1958e-06, "loss": 0.0188, "step": 34022 }, { "epoch": 1.8138828170816228, "grad_norm": 2.8788983821868896, "learning_rate": 3.1956000000000004e-06, "loss": 0.0239, "step": 34023 }, { "epoch": 1.8139361305112758, "grad_norm": 2.105145215988159, "learning_rate": 3.1954000000000004e-06, "loss": 0.0222, "step": 34024 }, { "epoch": 1.8139894439409288, "grad_norm": 1.2702299356460571, "learning_rate": 3.1952000000000003e-06, "loss": 0.0111, "step": 34025 }, { "epoch": 1.8140427573705815, "grad_norm": 1.073396921157837, "learning_rate": 3.1950000000000003e-06, "loss": 0.0138, "step": 34026 }, { "epoch": 1.8140960708002347, "grad_norm": 1.1194472312927246, "learning_rate": 3.1948000000000003e-06, "loss": 0.0079, "step": 34027 }, { "epoch": 1.8141493842298875, "grad_norm": 1.5846270322799683, "learning_rate": 3.1946000000000003e-06, "loss": 0.0157, "step": 34028 }, { "epoch": 1.8142026976595405, "grad_norm": 2.2822134494781494, "learning_rate": 3.1944000000000003e-06, "loss": 0.0201, "step": 34029 }, { "epoch": 1.8142560110891934, "grad_norm": 3.5579745769500732, "learning_rate": 3.1942000000000003e-06, "loss": 0.0161, "step": 34030 }, { "epoch": 1.8143093245188462, "grad_norm": 1.3851689100265503, "learning_rate": 3.1940000000000003e-06, "loss": 0.0265, "step": 34031 }, { "epoch": 1.8143626379484994, "grad_norm": 2.488065242767334, "learning_rate": 3.1938000000000002e-06, "loss": 0.0244, "step": 34032 }, { "epoch": 1.8144159513781521, "grad_norm": 2.276370048522949, "learning_rate": 3.1936000000000002e-06, "loss": 0.0307, "step": 34033 }, { "epoch": 1.8144692648078051, "grad_norm": 2.6859734058380127, "learning_rate": 3.1934e-06, "loss": 0.062, "step": 34034 }, { "epoch": 1.814522578237458, "grad_norm": 1.9306567907333374, "learning_rate": 3.1932e-06, "loss": 0.0238, "step": 34035 }, { "epoch": 1.8145758916671109, "grad_norm": 1.3173705339431763, "learning_rate": 3.193e-06, "loss": 0.0111, "step": 34036 }, { "epoch": 1.8146292050967638, "grad_norm": 0.8611413836479187, "learning_rate": 3.1928e-06, "loss": 0.0078, "step": 34037 }, { "epoch": 1.8146825185264168, "grad_norm": 0.7591823935508728, "learning_rate": 3.1926e-06, "loss": 0.0087, "step": 34038 }, { "epoch": 1.8147358319560696, "grad_norm": 3.1447672843933105, "learning_rate": 3.1924000000000006e-06, "loss": 0.0366, "step": 34039 }, { "epoch": 1.8147891453857228, "grad_norm": 2.3682267665863037, "learning_rate": 3.1922e-06, "loss": 0.0304, "step": 34040 }, { "epoch": 1.8148424588153755, "grad_norm": 1.8253720998764038, "learning_rate": 3.192e-06, "loss": 0.0117, "step": 34041 }, { "epoch": 1.8148957722450285, "grad_norm": 1.607652187347412, "learning_rate": 3.1918000000000005e-06, "loss": 0.0304, "step": 34042 }, { "epoch": 1.8149490856746815, "grad_norm": 2.6215624809265137, "learning_rate": 3.1916e-06, "loss": 0.0181, "step": 34043 }, { "epoch": 1.8150023991043343, "grad_norm": 1.2565982341766357, "learning_rate": 3.1914e-06, "loss": 0.0154, "step": 34044 }, { "epoch": 1.8150557125339875, "grad_norm": 0.5512575507164001, "learning_rate": 3.1912000000000005e-06, "loss": 0.0038, "step": 34045 }, { "epoch": 1.8151090259636402, "grad_norm": 2.7160584926605225, "learning_rate": 3.191e-06, "loss": 0.0436, "step": 34046 }, { "epoch": 1.8151623393932932, "grad_norm": 1.159989833831787, "learning_rate": 3.1908e-06, "loss": 0.0139, "step": 34047 }, { "epoch": 1.8152156528229462, "grad_norm": 1.3293578624725342, "learning_rate": 3.1906000000000004e-06, "loss": 0.02, "step": 34048 }, { "epoch": 1.815268966252599, "grad_norm": 1.7324910163879395, "learning_rate": 3.1904e-06, "loss": 0.0062, "step": 34049 }, { "epoch": 1.815322279682252, "grad_norm": 0.5930607318878174, "learning_rate": 3.1902000000000004e-06, "loss": 0.0045, "step": 34050 }, { "epoch": 1.8153755931119049, "grad_norm": 1.3689136505126953, "learning_rate": 3.1900000000000004e-06, "loss": 0.0173, "step": 34051 }, { "epoch": 1.8154289065415579, "grad_norm": 1.552385687828064, "learning_rate": 3.1898e-06, "loss": 0.0171, "step": 34052 }, { "epoch": 1.8154822199712108, "grad_norm": 2.2535343170166016, "learning_rate": 3.1896000000000003e-06, "loss": 0.0264, "step": 34053 }, { "epoch": 1.8155355334008636, "grad_norm": 1.3976714611053467, "learning_rate": 3.1894000000000003e-06, "loss": 0.0209, "step": 34054 }, { "epoch": 1.8155888468305166, "grad_norm": 1.9868005514144897, "learning_rate": 3.1892e-06, "loss": 0.0383, "step": 34055 }, { "epoch": 1.8156421602601696, "grad_norm": 1.2574222087860107, "learning_rate": 3.1890000000000003e-06, "loss": 0.022, "step": 34056 }, { "epoch": 1.8156954736898223, "grad_norm": 1.3597502708435059, "learning_rate": 3.1888000000000003e-06, "loss": 0.0081, "step": 34057 }, { "epoch": 1.8157487871194755, "grad_norm": 0.5327225923538208, "learning_rate": 3.1886e-06, "loss": 0.0059, "step": 34058 }, { "epoch": 1.8158021005491283, "grad_norm": 2.0116679668426514, "learning_rate": 3.1884000000000003e-06, "loss": 0.0223, "step": 34059 }, { "epoch": 1.8158554139787813, "grad_norm": 2.978496551513672, "learning_rate": 3.1882000000000002e-06, "loss": 0.0139, "step": 34060 }, { "epoch": 1.8159087274084342, "grad_norm": 1.7116994857788086, "learning_rate": 3.188e-06, "loss": 0.0149, "step": 34061 }, { "epoch": 1.815962040838087, "grad_norm": 2.368302583694458, "learning_rate": 3.1878e-06, "loss": 0.0314, "step": 34062 }, { "epoch": 1.8160153542677402, "grad_norm": 1.4142025709152222, "learning_rate": 3.1876e-06, "loss": 0.0116, "step": 34063 }, { "epoch": 1.816068667697393, "grad_norm": 1.201339840888977, "learning_rate": 3.1874000000000006e-06, "loss": 0.0071, "step": 34064 }, { "epoch": 1.816121981127046, "grad_norm": 2.1274471282958984, "learning_rate": 3.1872e-06, "loss": 0.0291, "step": 34065 }, { "epoch": 1.816175294556699, "grad_norm": 1.3282123804092407, "learning_rate": 3.187e-06, "loss": 0.0226, "step": 34066 }, { "epoch": 1.8162286079863517, "grad_norm": 2.258955955505371, "learning_rate": 3.1868000000000006e-06, "loss": 0.0147, "step": 34067 }, { "epoch": 1.8162819214160046, "grad_norm": 1.4895401000976562, "learning_rate": 3.1866e-06, "loss": 0.0125, "step": 34068 }, { "epoch": 1.8163352348456576, "grad_norm": 1.368064284324646, "learning_rate": 3.1864e-06, "loss": 0.0097, "step": 34069 }, { "epoch": 1.8163885482753106, "grad_norm": 2.9421401023864746, "learning_rate": 3.1862000000000005e-06, "loss": 0.0166, "step": 34070 }, { "epoch": 1.8164418617049636, "grad_norm": 1.8597338199615479, "learning_rate": 3.186e-06, "loss": 0.0205, "step": 34071 }, { "epoch": 1.8164951751346163, "grad_norm": 2.0319876670837402, "learning_rate": 3.1858e-06, "loss": 0.0353, "step": 34072 }, { "epoch": 1.8165484885642693, "grad_norm": 1.2778834104537964, "learning_rate": 3.1856000000000005e-06, "loss": 0.0063, "step": 34073 }, { "epoch": 1.8166018019939223, "grad_norm": 1.3716726303100586, "learning_rate": 3.1854e-06, "loss": 0.0238, "step": 34074 }, { "epoch": 1.816655115423575, "grad_norm": 0.7746679782867432, "learning_rate": 3.1852000000000004e-06, "loss": 0.0039, "step": 34075 }, { "epoch": 1.8167084288532283, "grad_norm": 3.9716269969940186, "learning_rate": 3.1850000000000004e-06, "loss": 0.0159, "step": 34076 }, { "epoch": 1.816761742282881, "grad_norm": 2.0818755626678467, "learning_rate": 3.1848e-06, "loss": 0.0318, "step": 34077 }, { "epoch": 1.816815055712534, "grad_norm": 1.5976074934005737, "learning_rate": 3.1846000000000004e-06, "loss": 0.0089, "step": 34078 }, { "epoch": 1.816868369142187, "grad_norm": 2.836951971054077, "learning_rate": 3.1844000000000004e-06, "loss": 0.0342, "step": 34079 }, { "epoch": 1.8169216825718397, "grad_norm": 2.3380987644195557, "learning_rate": 3.1842e-06, "loss": 0.0207, "step": 34080 }, { "epoch": 1.816974996001493, "grad_norm": 1.9516878128051758, "learning_rate": 3.1840000000000003e-06, "loss": 0.0261, "step": 34081 }, { "epoch": 1.8170283094311457, "grad_norm": 0.7278692126274109, "learning_rate": 3.1838000000000003e-06, "loss": 0.0057, "step": 34082 }, { "epoch": 1.8170816228607987, "grad_norm": 1.1807615756988525, "learning_rate": 3.1836e-06, "loss": 0.0068, "step": 34083 }, { "epoch": 1.8171349362904516, "grad_norm": 1.0351300239562988, "learning_rate": 3.1834000000000003e-06, "loss": 0.0159, "step": 34084 }, { "epoch": 1.8171882497201044, "grad_norm": 2.293280839920044, "learning_rate": 3.1832000000000003e-06, "loss": 0.0256, "step": 34085 }, { "epoch": 1.8172415631497574, "grad_norm": 0.7637033462524414, "learning_rate": 3.1830000000000007e-06, "loss": 0.0054, "step": 34086 }, { "epoch": 1.8172948765794104, "grad_norm": 0.6243593692779541, "learning_rate": 3.1828000000000003e-06, "loss": 0.0046, "step": 34087 }, { "epoch": 1.8173481900090633, "grad_norm": 1.7451083660125732, "learning_rate": 3.1826000000000002e-06, "loss": 0.0366, "step": 34088 }, { "epoch": 1.8174015034387163, "grad_norm": 1.4259912967681885, "learning_rate": 3.1824000000000007e-06, "loss": 0.0285, "step": 34089 }, { "epoch": 1.817454816868369, "grad_norm": 0.7829698920249939, "learning_rate": 3.1822000000000002e-06, "loss": 0.0057, "step": 34090 }, { "epoch": 1.817508130298022, "grad_norm": 2.1037518978118896, "learning_rate": 3.182e-06, "loss": 0.0137, "step": 34091 }, { "epoch": 1.817561443727675, "grad_norm": 1.869667649269104, "learning_rate": 3.1818000000000006e-06, "loss": 0.0078, "step": 34092 }, { "epoch": 1.8176147571573278, "grad_norm": 1.9895743131637573, "learning_rate": 3.1816e-06, "loss": 0.0239, "step": 34093 }, { "epoch": 1.817668070586981, "grad_norm": 1.0970523357391357, "learning_rate": 3.1814e-06, "loss": 0.019, "step": 34094 }, { "epoch": 1.8177213840166337, "grad_norm": 0.8217806220054626, "learning_rate": 3.1812000000000006e-06, "loss": 0.0126, "step": 34095 }, { "epoch": 1.8177746974462867, "grad_norm": 1.3816006183624268, "learning_rate": 3.181e-06, "loss": 0.0085, "step": 34096 }, { "epoch": 1.8178280108759397, "grad_norm": 1.6472125053405762, "learning_rate": 3.1808e-06, "loss": 0.0106, "step": 34097 }, { "epoch": 1.8178813243055925, "grad_norm": 1.596553087234497, "learning_rate": 3.1806000000000005e-06, "loss": 0.0349, "step": 34098 }, { "epoch": 1.8179346377352457, "grad_norm": 2.2679405212402344, "learning_rate": 3.1804e-06, "loss": 0.0191, "step": 34099 }, { "epoch": 1.8179879511648984, "grad_norm": 1.2217812538146973, "learning_rate": 3.1802000000000005e-06, "loss": 0.0196, "step": 34100 }, { "epoch": 1.8180412645945514, "grad_norm": 1.649072289466858, "learning_rate": 3.1800000000000005e-06, "loss": 0.0226, "step": 34101 }, { "epoch": 1.8180945780242044, "grad_norm": 0.4453139305114746, "learning_rate": 3.1798e-06, "loss": 0.0036, "step": 34102 }, { "epoch": 1.8181478914538571, "grad_norm": 1.6873847246170044, "learning_rate": 3.1796000000000004e-06, "loss": 0.0267, "step": 34103 }, { "epoch": 1.8182012048835101, "grad_norm": 0.38716983795166016, "learning_rate": 3.1794000000000004e-06, "loss": 0.0035, "step": 34104 }, { "epoch": 1.818254518313163, "grad_norm": 0.7569637894630432, "learning_rate": 3.1792e-06, "loss": 0.0068, "step": 34105 }, { "epoch": 1.8183078317428158, "grad_norm": 1.225511908531189, "learning_rate": 3.1790000000000004e-06, "loss": 0.0226, "step": 34106 }, { "epoch": 1.818361145172469, "grad_norm": 1.9382646083831787, "learning_rate": 3.1788000000000004e-06, "loss": 0.0141, "step": 34107 }, { "epoch": 1.8184144586021218, "grad_norm": 2.5381813049316406, "learning_rate": 3.1786e-06, "loss": 0.0416, "step": 34108 }, { "epoch": 1.8184677720317748, "grad_norm": 1.7876001596450806, "learning_rate": 3.1784000000000003e-06, "loss": 0.0172, "step": 34109 }, { "epoch": 1.8185210854614278, "grad_norm": 1.7154951095581055, "learning_rate": 3.1782000000000003e-06, "loss": 0.0189, "step": 34110 }, { "epoch": 1.8185743988910805, "grad_norm": 1.381089687347412, "learning_rate": 3.1780000000000003e-06, "loss": 0.0364, "step": 34111 }, { "epoch": 1.8186277123207337, "grad_norm": 0.5363684892654419, "learning_rate": 3.1778000000000003e-06, "loss": 0.0049, "step": 34112 }, { "epoch": 1.8186810257503865, "grad_norm": 1.537216067314148, "learning_rate": 3.1776000000000003e-06, "loss": 0.0175, "step": 34113 }, { "epoch": 1.8187343391800395, "grad_norm": 2.3184025287628174, "learning_rate": 3.1774000000000003e-06, "loss": 0.017, "step": 34114 }, { "epoch": 1.8187876526096924, "grad_norm": 1.74771249294281, "learning_rate": 3.1772000000000003e-06, "loss": 0.0323, "step": 34115 }, { "epoch": 1.8188409660393452, "grad_norm": 1.2320923805236816, "learning_rate": 3.1770000000000002e-06, "loss": 0.017, "step": 34116 }, { "epoch": 1.8188942794689984, "grad_norm": 1.241478443145752, "learning_rate": 3.1768000000000002e-06, "loss": 0.007, "step": 34117 }, { "epoch": 1.8189475928986512, "grad_norm": 1.842592477798462, "learning_rate": 3.1766000000000002e-06, "loss": 0.0132, "step": 34118 }, { "epoch": 1.8190009063283041, "grad_norm": 1.2441520690917969, "learning_rate": 3.1764e-06, "loss": 0.0141, "step": 34119 }, { "epoch": 1.8190542197579571, "grad_norm": 2.2826812267303467, "learning_rate": 3.1762e-06, "loss": 0.0109, "step": 34120 }, { "epoch": 1.8191075331876099, "grad_norm": 0.9601035118103027, "learning_rate": 3.176e-06, "loss": 0.0055, "step": 34121 }, { "epoch": 1.8191608466172628, "grad_norm": 1.7382172346115112, "learning_rate": 3.1758e-06, "loss": 0.0187, "step": 34122 }, { "epoch": 1.8192141600469158, "grad_norm": 3.5921740531921387, "learning_rate": 3.1756e-06, "loss": 0.0185, "step": 34123 }, { "epoch": 1.8192674734765686, "grad_norm": 1.0927883386611938, "learning_rate": 3.1754e-06, "loss": 0.0111, "step": 34124 }, { "epoch": 1.8193207869062218, "grad_norm": 1.148589015007019, "learning_rate": 3.1752000000000005e-06, "loss": 0.0064, "step": 34125 }, { "epoch": 1.8193741003358745, "grad_norm": 1.1759707927703857, "learning_rate": 3.175e-06, "loss": 0.009, "step": 34126 }, { "epoch": 1.8194274137655275, "grad_norm": 0.5047231316566467, "learning_rate": 3.1748e-06, "loss": 0.0037, "step": 34127 }, { "epoch": 1.8194807271951805, "grad_norm": 2.733527898788452, "learning_rate": 3.1746000000000005e-06, "loss": 0.0183, "step": 34128 }, { "epoch": 1.8195340406248333, "grad_norm": 1.94855535030365, "learning_rate": 3.1744e-06, "loss": 0.0425, "step": 34129 }, { "epoch": 1.8195873540544865, "grad_norm": 1.4235820770263672, "learning_rate": 3.1742e-06, "loss": 0.0033, "step": 34130 }, { "epoch": 1.8196406674841392, "grad_norm": 2.184870719909668, "learning_rate": 3.1740000000000004e-06, "loss": 0.0179, "step": 34131 }, { "epoch": 1.8196939809137922, "grad_norm": 4.7884931564331055, "learning_rate": 3.1738e-06, "loss": 0.0598, "step": 34132 }, { "epoch": 1.8197472943434452, "grad_norm": 1.781282901763916, "learning_rate": 3.1736e-06, "loss": 0.0171, "step": 34133 }, { "epoch": 1.819800607773098, "grad_norm": 2.081723928451538, "learning_rate": 3.1734000000000004e-06, "loss": 0.0115, "step": 34134 }, { "epoch": 1.8198539212027511, "grad_norm": 3.7819786071777344, "learning_rate": 3.1732e-06, "loss": 0.0255, "step": 34135 }, { "epoch": 1.819907234632404, "grad_norm": 2.5769753456115723, "learning_rate": 3.1730000000000004e-06, "loss": 0.0231, "step": 34136 }, { "epoch": 1.8199605480620569, "grad_norm": 2.7011077404022217, "learning_rate": 3.1728000000000004e-06, "loss": 0.0414, "step": 34137 }, { "epoch": 1.8200138614917099, "grad_norm": 0.4897739589214325, "learning_rate": 3.1726e-06, "loss": 0.004, "step": 34138 }, { "epoch": 1.8200671749213626, "grad_norm": 1.4770278930664062, "learning_rate": 3.1724000000000003e-06, "loss": 0.0091, "step": 34139 }, { "epoch": 1.8201204883510156, "grad_norm": 1.4391603469848633, "learning_rate": 3.1722000000000003e-06, "loss": 0.0234, "step": 34140 }, { "epoch": 1.8201738017806686, "grad_norm": 2.9268033504486084, "learning_rate": 3.172e-06, "loss": 0.0274, "step": 34141 }, { "epoch": 1.8202271152103213, "grad_norm": 0.38709646463394165, "learning_rate": 3.1718000000000003e-06, "loss": 0.0021, "step": 34142 }, { "epoch": 1.8202804286399745, "grad_norm": 2.905268907546997, "learning_rate": 3.1716000000000003e-06, "loss": 0.013, "step": 34143 }, { "epoch": 1.8203337420696273, "grad_norm": 1.8166393041610718, "learning_rate": 3.1714e-06, "loss": 0.0167, "step": 34144 }, { "epoch": 1.8203870554992803, "grad_norm": 2.2376506328582764, "learning_rate": 3.1712000000000002e-06, "loss": 0.0423, "step": 34145 }, { "epoch": 1.8204403689289332, "grad_norm": 2.149505853652954, "learning_rate": 3.1710000000000002e-06, "loss": 0.0253, "step": 34146 }, { "epoch": 1.820493682358586, "grad_norm": 3.2679340839385986, "learning_rate": 3.1707999999999998e-06, "loss": 0.0323, "step": 34147 }, { "epoch": 1.8205469957882392, "grad_norm": 3.0383002758026123, "learning_rate": 3.1706e-06, "loss": 0.0206, "step": 34148 }, { "epoch": 1.820600309217892, "grad_norm": 0.4209291636943817, "learning_rate": 3.1704e-06, "loss": 0.0037, "step": 34149 }, { "epoch": 1.820653622647545, "grad_norm": 2.5231940746307373, "learning_rate": 3.1702000000000006e-06, "loss": 0.0151, "step": 34150 }, { "epoch": 1.820706936077198, "grad_norm": 2.0560929775238037, "learning_rate": 3.17e-06, "loss": 0.0183, "step": 34151 }, { "epoch": 1.8207602495068507, "grad_norm": 3.1310291290283203, "learning_rate": 3.1698e-06, "loss": 0.0294, "step": 34152 }, { "epoch": 1.8208135629365039, "grad_norm": 2.211146831512451, "learning_rate": 3.1696000000000005e-06, "loss": 0.0206, "step": 34153 }, { "epoch": 1.8208668763661566, "grad_norm": 1.5940771102905273, "learning_rate": 3.1694e-06, "loss": 0.0178, "step": 34154 }, { "epoch": 1.8209201897958096, "grad_norm": 3.1067917346954346, "learning_rate": 3.1692e-06, "loss": 0.0239, "step": 34155 }, { "epoch": 1.8209735032254626, "grad_norm": 2.3935322761535645, "learning_rate": 3.1690000000000005e-06, "loss": 0.0185, "step": 34156 }, { "epoch": 1.8210268166551153, "grad_norm": 1.735842227935791, "learning_rate": 3.1688e-06, "loss": 0.0133, "step": 34157 }, { "epoch": 1.8210801300847683, "grad_norm": 3.1010005474090576, "learning_rate": 3.1686e-06, "loss": 0.0512, "step": 34158 }, { "epoch": 1.8211334435144213, "grad_norm": 0.8498355746269226, "learning_rate": 3.1684000000000004e-06, "loss": 0.0057, "step": 34159 }, { "epoch": 1.821186756944074, "grad_norm": 1.868878722190857, "learning_rate": 3.1682e-06, "loss": 0.0218, "step": 34160 }, { "epoch": 1.8212400703737273, "grad_norm": 0.6536182165145874, "learning_rate": 3.1680000000000004e-06, "loss": 0.005, "step": 34161 }, { "epoch": 1.82129338380338, "grad_norm": 0.6943528652191162, "learning_rate": 3.1678000000000004e-06, "loss": 0.0059, "step": 34162 }, { "epoch": 1.821346697233033, "grad_norm": 1.290446400642395, "learning_rate": 3.1676e-06, "loss": 0.0211, "step": 34163 }, { "epoch": 1.821400010662686, "grad_norm": 1.6449021100997925, "learning_rate": 3.1674000000000004e-06, "loss": 0.0181, "step": 34164 }, { "epoch": 1.8214533240923387, "grad_norm": 1.6372992992401123, "learning_rate": 3.1672000000000004e-06, "loss": 0.0133, "step": 34165 }, { "epoch": 1.821506637521992, "grad_norm": 1.2257534265518188, "learning_rate": 3.167e-06, "loss": 0.0093, "step": 34166 }, { "epoch": 1.8215599509516447, "grad_norm": 1.4942781925201416, "learning_rate": 3.1668000000000003e-06, "loss": 0.0157, "step": 34167 }, { "epoch": 1.8216132643812977, "grad_norm": 2.2141714096069336, "learning_rate": 3.1666000000000003e-06, "loss": 0.0212, "step": 34168 }, { "epoch": 1.8216665778109506, "grad_norm": 3.625509023666382, "learning_rate": 3.1664e-06, "loss": 0.0328, "step": 34169 }, { "epoch": 1.8217198912406034, "grad_norm": 4.481875896453857, "learning_rate": 3.1662000000000003e-06, "loss": 0.0297, "step": 34170 }, { "epoch": 1.8217732046702564, "grad_norm": 0.5660043954849243, "learning_rate": 3.1660000000000003e-06, "loss": 0.0034, "step": 34171 }, { "epoch": 1.8218265180999094, "grad_norm": 2.578423023223877, "learning_rate": 3.1658e-06, "loss": 0.037, "step": 34172 }, { "epoch": 1.8218798315295623, "grad_norm": 0.9085827469825745, "learning_rate": 3.1656000000000002e-06, "loss": 0.0073, "step": 34173 }, { "epoch": 1.8219331449592153, "grad_norm": 1.2721956968307495, "learning_rate": 3.1654000000000002e-06, "loss": 0.0198, "step": 34174 }, { "epoch": 1.821986458388868, "grad_norm": 1.121165156364441, "learning_rate": 3.1652000000000006e-06, "loss": 0.0064, "step": 34175 }, { "epoch": 1.822039771818521, "grad_norm": 1.4425935745239258, "learning_rate": 3.165e-06, "loss": 0.0085, "step": 34176 }, { "epoch": 1.822093085248174, "grad_norm": 2.8427302837371826, "learning_rate": 3.1648e-06, "loss": 0.0184, "step": 34177 }, { "epoch": 1.8221463986778268, "grad_norm": 1.881986379623413, "learning_rate": 3.1646000000000006e-06, "loss": 0.0148, "step": 34178 }, { "epoch": 1.82219971210748, "grad_norm": 2.528493642807007, "learning_rate": 3.1644e-06, "loss": 0.0206, "step": 34179 }, { "epoch": 1.8222530255371328, "grad_norm": 2.198941469192505, "learning_rate": 3.1642e-06, "loss": 0.0116, "step": 34180 }, { "epoch": 1.8223063389667857, "grad_norm": 1.8289262056350708, "learning_rate": 3.1640000000000005e-06, "loss": 0.0217, "step": 34181 }, { "epoch": 1.8223596523964387, "grad_norm": 2.0152242183685303, "learning_rate": 3.1638e-06, "loss": 0.0227, "step": 34182 }, { "epoch": 1.8224129658260915, "grad_norm": 4.717819690704346, "learning_rate": 3.1636e-06, "loss": 0.0369, "step": 34183 }, { "epoch": 1.8224662792557447, "grad_norm": 1.779145359992981, "learning_rate": 3.1634000000000005e-06, "loss": 0.0139, "step": 34184 }, { "epoch": 1.8225195926853974, "grad_norm": 7.136096954345703, "learning_rate": 3.1632e-06, "loss": 0.035, "step": 34185 }, { "epoch": 1.8225729061150504, "grad_norm": 1.3502479791641235, "learning_rate": 3.1630000000000005e-06, "loss": 0.0184, "step": 34186 }, { "epoch": 1.8226262195447034, "grad_norm": 2.5825400352478027, "learning_rate": 3.1628000000000004e-06, "loss": 0.028, "step": 34187 }, { "epoch": 1.8226795329743561, "grad_norm": 1.3928388357162476, "learning_rate": 3.1626e-06, "loss": 0.0073, "step": 34188 }, { "epoch": 1.8227328464040091, "grad_norm": 1.742136836051941, "learning_rate": 3.1624000000000004e-06, "loss": 0.0334, "step": 34189 }, { "epoch": 1.822786159833662, "grad_norm": 1.9744840860366821, "learning_rate": 3.1622000000000004e-06, "loss": 0.027, "step": 34190 }, { "epoch": 1.822839473263315, "grad_norm": 1.7946864366531372, "learning_rate": 3.162e-06, "loss": 0.0254, "step": 34191 }, { "epoch": 1.822892786692968, "grad_norm": 2.680384635925293, "learning_rate": 3.1618000000000004e-06, "loss": 0.0185, "step": 34192 }, { "epoch": 1.8229461001226208, "grad_norm": 1.076055884361267, "learning_rate": 3.1616000000000004e-06, "loss": 0.0061, "step": 34193 }, { "epoch": 1.8229994135522738, "grad_norm": 1.270212173461914, "learning_rate": 3.1614e-06, "loss": 0.0118, "step": 34194 }, { "epoch": 1.8230527269819268, "grad_norm": 1.5169169902801514, "learning_rate": 3.1612000000000003e-06, "loss": 0.0173, "step": 34195 }, { "epoch": 1.8231060404115795, "grad_norm": 2.441103935241699, "learning_rate": 3.1610000000000003e-06, "loss": 0.0205, "step": 34196 }, { "epoch": 1.8231593538412327, "grad_norm": 1.3205431699752808, "learning_rate": 3.1608000000000007e-06, "loss": 0.0067, "step": 34197 }, { "epoch": 1.8232126672708855, "grad_norm": 4.680997848510742, "learning_rate": 3.1606000000000003e-06, "loss": 0.0178, "step": 34198 }, { "epoch": 1.8232659807005385, "grad_norm": 2.0156314373016357, "learning_rate": 3.1604000000000003e-06, "loss": 0.0129, "step": 34199 }, { "epoch": 1.8233192941301914, "grad_norm": 3.112421989440918, "learning_rate": 3.1602000000000007e-06, "loss": 0.0418, "step": 34200 }, { "epoch": 1.8233726075598442, "grad_norm": 2.702268123626709, "learning_rate": 3.1600000000000002e-06, "loss": 0.0277, "step": 34201 }, { "epoch": 1.8234259209894974, "grad_norm": 2.7270145416259766, "learning_rate": 3.1598000000000002e-06, "loss": 0.0318, "step": 34202 }, { "epoch": 1.8234792344191502, "grad_norm": 3.760590076446533, "learning_rate": 3.1596000000000006e-06, "loss": 0.0208, "step": 34203 }, { "epoch": 1.8235325478488031, "grad_norm": 1.9568395614624023, "learning_rate": 3.1594e-06, "loss": 0.0085, "step": 34204 }, { "epoch": 1.8235858612784561, "grad_norm": 0.7077215313911438, "learning_rate": 3.1592e-06, "loss": 0.0059, "step": 34205 }, { "epoch": 1.8236391747081089, "grad_norm": 1.4904714822769165, "learning_rate": 3.1590000000000006e-06, "loss": 0.0234, "step": 34206 }, { "epoch": 1.8236924881377619, "grad_norm": 1.9682893753051758, "learning_rate": 3.1588e-06, "loss": 0.0212, "step": 34207 }, { "epoch": 1.8237458015674148, "grad_norm": 2.051051378250122, "learning_rate": 3.1586e-06, "loss": 0.0241, "step": 34208 }, { "epoch": 1.8237991149970678, "grad_norm": 1.190411925315857, "learning_rate": 3.1584000000000005e-06, "loss": 0.0074, "step": 34209 }, { "epoch": 1.8238524284267208, "grad_norm": 1.647722601890564, "learning_rate": 3.1582e-06, "loss": 0.0226, "step": 34210 }, { "epoch": 1.8239057418563736, "grad_norm": 2.948556661605835, "learning_rate": 3.1580000000000005e-06, "loss": 0.0254, "step": 34211 }, { "epoch": 1.8239590552860265, "grad_norm": 2.2493338584899902, "learning_rate": 3.1578000000000005e-06, "loss": 0.0138, "step": 34212 }, { "epoch": 1.8240123687156795, "grad_norm": 2.8772003650665283, "learning_rate": 3.1576e-06, "loss": 0.0307, "step": 34213 }, { "epoch": 1.8240656821453323, "grad_norm": 3.3118131160736084, "learning_rate": 3.1574000000000005e-06, "loss": 0.0245, "step": 34214 }, { "epoch": 1.8241189955749855, "grad_norm": 1.0182591676712036, "learning_rate": 3.1572000000000005e-06, "loss": 0.0084, "step": 34215 }, { "epoch": 1.8241723090046382, "grad_norm": 3.064127206802368, "learning_rate": 3.157e-06, "loss": 0.0316, "step": 34216 }, { "epoch": 1.8242256224342912, "grad_norm": 1.1400346755981445, "learning_rate": 3.1568000000000004e-06, "loss": 0.0122, "step": 34217 }, { "epoch": 1.8242789358639442, "grad_norm": 1.4105576276779175, "learning_rate": 3.1566000000000004e-06, "loss": 0.016, "step": 34218 }, { "epoch": 1.824332249293597, "grad_norm": 1.7798445224761963, "learning_rate": 3.1564e-06, "loss": 0.0158, "step": 34219 }, { "epoch": 1.8243855627232501, "grad_norm": 1.7160048484802246, "learning_rate": 3.1562000000000004e-06, "loss": 0.0118, "step": 34220 }, { "epoch": 1.824438876152903, "grad_norm": 2.3009958267211914, "learning_rate": 3.1560000000000004e-06, "loss": 0.0674, "step": 34221 }, { "epoch": 1.8244921895825559, "grad_norm": 0.9757697582244873, "learning_rate": 3.1558000000000003e-06, "loss": 0.0204, "step": 34222 }, { "epoch": 1.8245455030122089, "grad_norm": 1.7592633962631226, "learning_rate": 3.1556000000000003e-06, "loss": 0.0387, "step": 34223 }, { "epoch": 1.8245988164418616, "grad_norm": 1.6891721487045288, "learning_rate": 3.1554000000000003e-06, "loss": 0.0193, "step": 34224 }, { "epoch": 1.8246521298715146, "grad_norm": 2.3691155910491943, "learning_rate": 3.1552000000000003e-06, "loss": 0.0156, "step": 34225 }, { "epoch": 1.8247054433011676, "grad_norm": 2.510248899459839, "learning_rate": 3.1550000000000003e-06, "loss": 0.0311, "step": 34226 }, { "epoch": 1.8247587567308203, "grad_norm": 1.2456996440887451, "learning_rate": 3.1548000000000003e-06, "loss": 0.0078, "step": 34227 }, { "epoch": 1.8248120701604735, "grad_norm": 1.1348360776901245, "learning_rate": 3.1546000000000003e-06, "loss": 0.0074, "step": 34228 }, { "epoch": 1.8248653835901263, "grad_norm": 0.26311954855918884, "learning_rate": 3.1544000000000002e-06, "loss": 0.0021, "step": 34229 }, { "epoch": 1.8249186970197793, "grad_norm": 1.6235450506210327, "learning_rate": 3.1542000000000002e-06, "loss": 0.0071, "step": 34230 }, { "epoch": 1.8249720104494322, "grad_norm": 1.8715245723724365, "learning_rate": 3.154e-06, "loss": 0.0111, "step": 34231 }, { "epoch": 1.825025323879085, "grad_norm": 1.727000117301941, "learning_rate": 3.1538e-06, "loss": 0.0205, "step": 34232 }, { "epoch": 1.8250786373087382, "grad_norm": 2.44045352935791, "learning_rate": 3.1536e-06, "loss": 0.0131, "step": 34233 }, { "epoch": 1.825131950738391, "grad_norm": 1.9711718559265137, "learning_rate": 3.1534e-06, "loss": 0.0065, "step": 34234 }, { "epoch": 1.825185264168044, "grad_norm": 0.9845688939094543, "learning_rate": 3.1532e-06, "loss": 0.0057, "step": 34235 }, { "epoch": 1.825238577597697, "grad_norm": 1.2648895978927612, "learning_rate": 3.1530000000000006e-06, "loss": 0.0087, "step": 34236 }, { "epoch": 1.8252918910273497, "grad_norm": 3.1893699169158936, "learning_rate": 3.1528e-06, "loss": 0.0457, "step": 34237 }, { "epoch": 1.8253452044570029, "grad_norm": 0.9599589705467224, "learning_rate": 3.1526e-06, "loss": 0.0216, "step": 34238 }, { "epoch": 1.8253985178866556, "grad_norm": 4.002843856811523, "learning_rate": 3.1524000000000005e-06, "loss": 0.0188, "step": 34239 }, { "epoch": 1.8254518313163086, "grad_norm": 0.9843303561210632, "learning_rate": 3.1522e-06, "loss": 0.0038, "step": 34240 }, { "epoch": 1.8255051447459616, "grad_norm": 2.9653539657592773, "learning_rate": 3.152e-06, "loss": 0.0129, "step": 34241 }, { "epoch": 1.8255584581756144, "grad_norm": 3.926523208618164, "learning_rate": 3.1518000000000005e-06, "loss": 0.0311, "step": 34242 }, { "epoch": 1.8256117716052673, "grad_norm": 1.3303778171539307, "learning_rate": 3.1516e-06, "loss": 0.0201, "step": 34243 }, { "epoch": 1.8256650850349203, "grad_norm": 1.1246016025543213, "learning_rate": 3.1514e-06, "loss": 0.0135, "step": 34244 }, { "epoch": 1.825718398464573, "grad_norm": 1.9770888090133667, "learning_rate": 3.1512000000000004e-06, "loss": 0.0309, "step": 34245 }, { "epoch": 1.8257717118942263, "grad_norm": 2.618095874786377, "learning_rate": 3.151e-06, "loss": 0.0222, "step": 34246 }, { "epoch": 1.825825025323879, "grad_norm": 2.35402250289917, "learning_rate": 3.1508000000000004e-06, "loss": 0.023, "step": 34247 }, { "epoch": 1.825878338753532, "grad_norm": 3.1194093227386475, "learning_rate": 3.1506000000000004e-06, "loss": 0.0124, "step": 34248 }, { "epoch": 1.825931652183185, "grad_norm": 0.950304388999939, "learning_rate": 3.1504e-06, "loss": 0.0036, "step": 34249 }, { "epoch": 1.8259849656128377, "grad_norm": 2.952606439590454, "learning_rate": 3.1502000000000003e-06, "loss": 0.0358, "step": 34250 }, { "epoch": 1.826038279042491, "grad_norm": 1.833072304725647, "learning_rate": 3.1500000000000003e-06, "loss": 0.0091, "step": 34251 }, { "epoch": 1.8260915924721437, "grad_norm": 1.3214828968048096, "learning_rate": 3.1498e-06, "loss": 0.0171, "step": 34252 }, { "epoch": 1.8261449059017967, "grad_norm": 3.219151496887207, "learning_rate": 3.1496000000000003e-06, "loss": 0.0526, "step": 34253 }, { "epoch": 1.8261982193314497, "grad_norm": 1.1176234483718872, "learning_rate": 3.1494000000000003e-06, "loss": 0.0188, "step": 34254 }, { "epoch": 1.8262515327611024, "grad_norm": 1.9841117858886719, "learning_rate": 3.1492e-06, "loss": 0.0246, "step": 34255 }, { "epoch": 1.8263048461907556, "grad_norm": 2.6249873638153076, "learning_rate": 3.1490000000000003e-06, "loss": 0.0388, "step": 34256 }, { "epoch": 1.8263581596204084, "grad_norm": 2.016932487487793, "learning_rate": 3.1488000000000002e-06, "loss": 0.0244, "step": 34257 }, { "epoch": 1.8264114730500614, "grad_norm": 1.8149358034133911, "learning_rate": 3.1486e-06, "loss": 0.0284, "step": 34258 }, { "epoch": 1.8264647864797143, "grad_norm": 0.8713234663009644, "learning_rate": 3.1484000000000002e-06, "loss": 0.0174, "step": 34259 }, { "epoch": 1.826518099909367, "grad_norm": 3.0820038318634033, "learning_rate": 3.1482e-06, "loss": 0.0619, "step": 34260 }, { "epoch": 1.82657141333902, "grad_norm": 2.6834371089935303, "learning_rate": 3.1480000000000006e-06, "loss": 0.0427, "step": 34261 }, { "epoch": 1.826624726768673, "grad_norm": 2.4977927207946777, "learning_rate": 3.1478e-06, "loss": 0.0252, "step": 34262 }, { "epoch": 1.8266780401983258, "grad_norm": 1.853365421295166, "learning_rate": 3.1476e-06, "loss": 0.012, "step": 34263 }, { "epoch": 1.826731353627979, "grad_norm": 1.0773853063583374, "learning_rate": 3.1474000000000006e-06, "loss": 0.0187, "step": 34264 }, { "epoch": 1.8267846670576318, "grad_norm": 1.786392331123352, "learning_rate": 3.1472e-06, "loss": 0.0256, "step": 34265 }, { "epoch": 1.8268379804872847, "grad_norm": 1.7648621797561646, "learning_rate": 3.147e-06, "loss": 0.0149, "step": 34266 }, { "epoch": 1.8268912939169377, "grad_norm": 1.0381828546524048, "learning_rate": 3.1468000000000005e-06, "loss": 0.0082, "step": 34267 }, { "epoch": 1.8269446073465905, "grad_norm": 2.2872912883758545, "learning_rate": 3.1466e-06, "loss": 0.0201, "step": 34268 }, { "epoch": 1.8269979207762437, "grad_norm": 0.628935694694519, "learning_rate": 3.1464e-06, "loss": 0.0055, "step": 34269 }, { "epoch": 1.8270512342058964, "grad_norm": 2.0753092765808105, "learning_rate": 3.1462000000000005e-06, "loss": 0.0238, "step": 34270 }, { "epoch": 1.8271045476355494, "grad_norm": 1.304868459701538, "learning_rate": 3.146e-06, "loss": 0.0211, "step": 34271 }, { "epoch": 1.8271578610652024, "grad_norm": 3.0320050716400146, "learning_rate": 3.1458000000000004e-06, "loss": 0.0467, "step": 34272 }, { "epoch": 1.8272111744948552, "grad_norm": 2.2238924503326416, "learning_rate": 3.1456000000000004e-06, "loss": 0.0321, "step": 34273 }, { "epoch": 1.8272644879245081, "grad_norm": 2.3849129676818848, "learning_rate": 3.1454e-06, "loss": 0.0362, "step": 34274 }, { "epoch": 1.827317801354161, "grad_norm": 1.6276968717575073, "learning_rate": 3.1452000000000004e-06, "loss": 0.0204, "step": 34275 }, { "epoch": 1.827371114783814, "grad_norm": 1.856465458869934, "learning_rate": 3.1450000000000004e-06, "loss": 0.0165, "step": 34276 }, { "epoch": 1.827424428213467, "grad_norm": 2.6668241024017334, "learning_rate": 3.1448e-06, "loss": 0.0256, "step": 34277 }, { "epoch": 1.8274777416431198, "grad_norm": 2.4583652019500732, "learning_rate": 3.1446000000000004e-06, "loss": 0.0461, "step": 34278 }, { "epoch": 1.8275310550727728, "grad_norm": 1.604269027709961, "learning_rate": 3.1444000000000003e-06, "loss": 0.0128, "step": 34279 }, { "epoch": 1.8275843685024258, "grad_norm": 1.5000933408737183, "learning_rate": 3.1442e-06, "loss": 0.0197, "step": 34280 }, { "epoch": 1.8276376819320785, "grad_norm": 1.9698965549468994, "learning_rate": 3.1440000000000003e-06, "loss": 0.0435, "step": 34281 }, { "epoch": 1.8276909953617317, "grad_norm": 4.46846342086792, "learning_rate": 3.1438000000000003e-06, "loss": 0.024, "step": 34282 }, { "epoch": 1.8277443087913845, "grad_norm": 1.8839104175567627, "learning_rate": 3.1436e-06, "loss": 0.0157, "step": 34283 }, { "epoch": 1.8277976222210375, "grad_norm": 5.128926753997803, "learning_rate": 3.1434000000000003e-06, "loss": 0.0074, "step": 34284 }, { "epoch": 1.8278509356506905, "grad_norm": 1.6575533151626587, "learning_rate": 3.1432000000000002e-06, "loss": 0.0127, "step": 34285 }, { "epoch": 1.8279042490803432, "grad_norm": 1.6755458116531372, "learning_rate": 3.1430000000000007e-06, "loss": 0.0093, "step": 34286 }, { "epoch": 1.8279575625099964, "grad_norm": 1.4318814277648926, "learning_rate": 3.1428000000000002e-06, "loss": 0.0145, "step": 34287 }, { "epoch": 1.8280108759396492, "grad_norm": 1.1856695413589478, "learning_rate": 3.1426e-06, "loss": 0.0125, "step": 34288 }, { "epoch": 1.8280641893693022, "grad_norm": 0.8692744374275208, "learning_rate": 3.1424000000000006e-06, "loss": 0.0234, "step": 34289 }, { "epoch": 1.8281175027989551, "grad_norm": 1.2722911834716797, "learning_rate": 3.1422e-06, "loss": 0.0045, "step": 34290 }, { "epoch": 1.8281708162286079, "grad_norm": 2.1679635047912598, "learning_rate": 3.142e-06, "loss": 0.0331, "step": 34291 }, { "epoch": 1.8282241296582609, "grad_norm": 0.38462236523628235, "learning_rate": 3.1418000000000006e-06, "loss": 0.0033, "step": 34292 }, { "epoch": 1.8282774430879138, "grad_norm": 1.0460255146026611, "learning_rate": 3.1416e-06, "loss": 0.0313, "step": 34293 }, { "epoch": 1.8283307565175668, "grad_norm": 0.6321573853492737, "learning_rate": 3.1414e-06, "loss": 0.0049, "step": 34294 }, { "epoch": 1.8283840699472198, "grad_norm": 0.44298255443573, "learning_rate": 3.1412000000000005e-06, "loss": 0.0025, "step": 34295 }, { "epoch": 1.8284373833768726, "grad_norm": 2.6306662559509277, "learning_rate": 3.141e-06, "loss": 0.0254, "step": 34296 }, { "epoch": 1.8284906968065255, "grad_norm": 2.042170524597168, "learning_rate": 3.1408000000000005e-06, "loss": 0.0141, "step": 34297 }, { "epoch": 1.8285440102361785, "grad_norm": 2.231752634048462, "learning_rate": 3.1406000000000005e-06, "loss": 0.0331, "step": 34298 }, { "epoch": 1.8285973236658313, "grad_norm": 2.531445264816284, "learning_rate": 3.1404e-06, "loss": 0.0238, "step": 34299 }, { "epoch": 1.8286506370954845, "grad_norm": 0.5522449016571045, "learning_rate": 3.1402000000000004e-06, "loss": 0.005, "step": 34300 }, { "epoch": 1.8287039505251372, "grad_norm": 2.190891981124878, "learning_rate": 3.1400000000000004e-06, "loss": 0.017, "step": 34301 }, { "epoch": 1.8287572639547902, "grad_norm": 0.5879572629928589, "learning_rate": 3.1398e-06, "loss": 0.005, "step": 34302 }, { "epoch": 1.8288105773844432, "grad_norm": 2.295985221862793, "learning_rate": 3.1396000000000004e-06, "loss": 0.0308, "step": 34303 }, { "epoch": 1.828863890814096, "grad_norm": 1.8989555835723877, "learning_rate": 3.1394000000000004e-06, "loss": 0.0163, "step": 34304 }, { "epoch": 1.8289172042437492, "grad_norm": 2.098435640335083, "learning_rate": 3.1392e-06, "loss": 0.0209, "step": 34305 }, { "epoch": 1.828970517673402, "grad_norm": 2.009507894515991, "learning_rate": 3.1390000000000004e-06, "loss": 0.0308, "step": 34306 }, { "epoch": 1.8290238311030549, "grad_norm": 0.3841712474822998, "learning_rate": 3.1388000000000003e-06, "loss": 0.0036, "step": 34307 }, { "epoch": 1.8290771445327079, "grad_norm": 1.2354779243469238, "learning_rate": 3.1386e-06, "loss": 0.017, "step": 34308 }, { "epoch": 1.8291304579623606, "grad_norm": 2.6441924571990967, "learning_rate": 3.1384000000000003e-06, "loss": 0.0265, "step": 34309 }, { "epoch": 1.8291837713920136, "grad_norm": 4.065212249755859, "learning_rate": 3.1382000000000003e-06, "loss": 0.0258, "step": 34310 }, { "epoch": 1.8292370848216666, "grad_norm": 2.5901811122894287, "learning_rate": 3.1380000000000003e-06, "loss": 0.0196, "step": 34311 }, { "epoch": 1.8292903982513196, "grad_norm": 1.462323784828186, "learning_rate": 3.1378000000000003e-06, "loss": 0.0131, "step": 34312 }, { "epoch": 1.8293437116809725, "grad_norm": 1.8559486865997314, "learning_rate": 3.1376000000000002e-06, "loss": 0.028, "step": 34313 }, { "epoch": 1.8293970251106253, "grad_norm": 1.5361666679382324, "learning_rate": 3.1374000000000002e-06, "loss": 0.01, "step": 34314 }, { "epoch": 1.8294503385402783, "grad_norm": 2.962700605392456, "learning_rate": 3.1372000000000002e-06, "loss": 0.0329, "step": 34315 }, { "epoch": 1.8295036519699313, "grad_norm": 1.4003127813339233, "learning_rate": 3.137e-06, "loss": 0.0109, "step": 34316 }, { "epoch": 1.829556965399584, "grad_norm": 2.2747533321380615, "learning_rate": 3.1368e-06, "loss": 0.0285, "step": 34317 }, { "epoch": 1.8296102788292372, "grad_norm": 2.7559072971343994, "learning_rate": 3.1366e-06, "loss": 0.0324, "step": 34318 }, { "epoch": 1.82966359225889, "grad_norm": 2.205380439758301, "learning_rate": 3.1364e-06, "loss": 0.0263, "step": 34319 }, { "epoch": 1.829716905688543, "grad_norm": 2.796363353729248, "learning_rate": 3.1362e-06, "loss": 0.0327, "step": 34320 }, { "epoch": 1.829770219118196, "grad_norm": 1.0191242694854736, "learning_rate": 3.136e-06, "loss": 0.0068, "step": 34321 }, { "epoch": 1.8298235325478487, "grad_norm": 0.6119276881217957, "learning_rate": 3.1358000000000005e-06, "loss": 0.0045, "step": 34322 }, { "epoch": 1.8298768459775019, "grad_norm": 1.234956979751587, "learning_rate": 3.1356e-06, "loss": 0.0082, "step": 34323 }, { "epoch": 1.8299301594071546, "grad_norm": 4.321816921234131, "learning_rate": 3.1354e-06, "loss": 0.0142, "step": 34324 }, { "epoch": 1.8299834728368076, "grad_norm": 1.9991143941879272, "learning_rate": 3.1352000000000005e-06, "loss": 0.0249, "step": 34325 }, { "epoch": 1.8300367862664606, "grad_norm": 2.404787302017212, "learning_rate": 3.135e-06, "loss": 0.0159, "step": 34326 }, { "epoch": 1.8300900996961134, "grad_norm": 1.8757745027542114, "learning_rate": 3.1348e-06, "loss": 0.0343, "step": 34327 }, { "epoch": 1.8301434131257663, "grad_norm": 2.028977632522583, "learning_rate": 3.1346000000000004e-06, "loss": 0.0399, "step": 34328 }, { "epoch": 1.8301967265554193, "grad_norm": 2.2357301712036133, "learning_rate": 3.1344e-06, "loss": 0.025, "step": 34329 }, { "epoch": 1.830250039985072, "grad_norm": 1.3002337217330933, "learning_rate": 3.1342e-06, "loss": 0.0208, "step": 34330 }, { "epoch": 1.8303033534147253, "grad_norm": 2.86742901802063, "learning_rate": 3.1340000000000004e-06, "loss": 0.0341, "step": 34331 }, { "epoch": 1.830356666844378, "grad_norm": 3.662630558013916, "learning_rate": 3.1338e-06, "loss": 0.0229, "step": 34332 }, { "epoch": 1.830409980274031, "grad_norm": 1.8273210525512695, "learning_rate": 3.1336000000000004e-06, "loss": 0.0145, "step": 34333 }, { "epoch": 1.830463293703684, "grad_norm": 1.6813366413116455, "learning_rate": 3.1334000000000004e-06, "loss": 0.0132, "step": 34334 }, { "epoch": 1.8305166071333367, "grad_norm": 0.8044698238372803, "learning_rate": 3.1332e-06, "loss": 0.0065, "step": 34335 }, { "epoch": 1.83056992056299, "grad_norm": 2.5162148475646973, "learning_rate": 3.1330000000000003e-06, "loss": 0.0295, "step": 34336 }, { "epoch": 1.8306232339926427, "grad_norm": 1.5656282901763916, "learning_rate": 3.1328000000000003e-06, "loss": 0.0108, "step": 34337 }, { "epoch": 1.8306765474222957, "grad_norm": 1.7772694826126099, "learning_rate": 3.1326e-06, "loss": 0.0116, "step": 34338 }, { "epoch": 1.8307298608519487, "grad_norm": 2.162794828414917, "learning_rate": 3.1324000000000003e-06, "loss": 0.0279, "step": 34339 }, { "epoch": 1.8307831742816014, "grad_norm": 1.15422785282135, "learning_rate": 3.1322000000000003e-06, "loss": 0.0199, "step": 34340 }, { "epoch": 1.8308364877112546, "grad_norm": 1.4087399244308472, "learning_rate": 3.132e-06, "loss": 0.0168, "step": 34341 }, { "epoch": 1.8308898011409074, "grad_norm": 1.2170675992965698, "learning_rate": 3.1318000000000002e-06, "loss": 0.0133, "step": 34342 }, { "epoch": 1.8309431145705604, "grad_norm": 1.9296802282333374, "learning_rate": 3.1316000000000002e-06, "loss": 0.0237, "step": 34343 }, { "epoch": 1.8309964280002133, "grad_norm": 2.1001555919647217, "learning_rate": 3.1314e-06, "loss": 0.0178, "step": 34344 }, { "epoch": 1.831049741429866, "grad_norm": 1.5568300485610962, "learning_rate": 3.1312e-06, "loss": 0.0137, "step": 34345 }, { "epoch": 1.831103054859519, "grad_norm": 1.7566641569137573, "learning_rate": 3.131e-06, "loss": 0.0153, "step": 34346 }, { "epoch": 1.831156368289172, "grad_norm": 2.9764111042022705, "learning_rate": 3.1308000000000006e-06, "loss": 0.0297, "step": 34347 }, { "epoch": 1.8312096817188248, "grad_norm": 2.785668134689331, "learning_rate": 3.1306e-06, "loss": 0.0373, "step": 34348 }, { "epoch": 1.831262995148478, "grad_norm": 1.781165599822998, "learning_rate": 3.1304e-06, "loss": 0.0216, "step": 34349 }, { "epoch": 1.8313163085781308, "grad_norm": 1.8582494258880615, "learning_rate": 3.1302000000000005e-06, "loss": 0.0215, "step": 34350 }, { "epoch": 1.8313696220077837, "grad_norm": 0.8720346093177795, "learning_rate": 3.13e-06, "loss": 0.0054, "step": 34351 }, { "epoch": 1.8314229354374367, "grad_norm": 1.017115831375122, "learning_rate": 3.1298e-06, "loss": 0.0049, "step": 34352 }, { "epoch": 1.8314762488670895, "grad_norm": 1.9100189208984375, "learning_rate": 3.1296000000000005e-06, "loss": 0.029, "step": 34353 }, { "epoch": 1.8315295622967427, "grad_norm": 1.2063982486724854, "learning_rate": 3.1294e-06, "loss": 0.0145, "step": 34354 }, { "epoch": 1.8315828757263954, "grad_norm": 2.8952062129974365, "learning_rate": 3.1292e-06, "loss": 0.0233, "step": 34355 }, { "epoch": 1.8316361891560484, "grad_norm": 3.525325059890747, "learning_rate": 3.1290000000000005e-06, "loss": 0.0345, "step": 34356 }, { "epoch": 1.8316895025857014, "grad_norm": 1.608432650566101, "learning_rate": 3.1288e-06, "loss": 0.0107, "step": 34357 }, { "epoch": 1.8317428160153542, "grad_norm": 1.1330770254135132, "learning_rate": 3.1286000000000004e-06, "loss": 0.0155, "step": 34358 }, { "epoch": 1.8317961294450074, "grad_norm": 0.9877059459686279, "learning_rate": 3.1284000000000004e-06, "loss": 0.0114, "step": 34359 }, { "epoch": 1.8318494428746601, "grad_norm": 1.6181678771972656, "learning_rate": 3.1282e-06, "loss": 0.0271, "step": 34360 }, { "epoch": 1.831902756304313, "grad_norm": 2.0565543174743652, "learning_rate": 3.1280000000000004e-06, "loss": 0.0166, "step": 34361 }, { "epoch": 1.831956069733966, "grad_norm": 0.6927987337112427, "learning_rate": 3.1278000000000004e-06, "loss": 0.0172, "step": 34362 }, { "epoch": 1.8320093831636188, "grad_norm": 0.18094880878925323, "learning_rate": 3.1276e-06, "loss": 0.0018, "step": 34363 }, { "epoch": 1.8320626965932718, "grad_norm": 0.535858154296875, "learning_rate": 3.1274000000000003e-06, "loss": 0.004, "step": 34364 }, { "epoch": 1.8321160100229248, "grad_norm": 2.1058382987976074, "learning_rate": 3.1272000000000003e-06, "loss": 0.0085, "step": 34365 }, { "epoch": 1.8321693234525775, "grad_norm": 1.482903003692627, "learning_rate": 3.127e-06, "loss": 0.0309, "step": 34366 }, { "epoch": 1.8322226368822307, "grad_norm": 1.9809703826904297, "learning_rate": 3.1268000000000003e-06, "loss": 0.0252, "step": 34367 }, { "epoch": 1.8322759503118835, "grad_norm": 1.496099829673767, "learning_rate": 3.1266000000000003e-06, "loss": 0.0174, "step": 34368 }, { "epoch": 1.8323292637415365, "grad_norm": 2.0865745544433594, "learning_rate": 3.1264e-06, "loss": 0.0423, "step": 34369 }, { "epoch": 1.8323825771711895, "grad_norm": 1.4251781702041626, "learning_rate": 3.1262000000000002e-06, "loss": 0.0147, "step": 34370 }, { "epoch": 1.8324358906008422, "grad_norm": 2.0906741619110107, "learning_rate": 3.1260000000000002e-06, "loss": 0.0364, "step": 34371 }, { "epoch": 1.8324892040304954, "grad_norm": 1.7536827325820923, "learning_rate": 3.1258000000000006e-06, "loss": 0.0286, "step": 34372 }, { "epoch": 1.8325425174601482, "grad_norm": 2.7429726123809814, "learning_rate": 3.1256e-06, "loss": 0.0378, "step": 34373 }, { "epoch": 1.8325958308898012, "grad_norm": 1.336106538772583, "learning_rate": 3.1254e-06, "loss": 0.0119, "step": 34374 }, { "epoch": 1.8326491443194541, "grad_norm": 1.1748160123825073, "learning_rate": 3.1252000000000006e-06, "loss": 0.0083, "step": 34375 }, { "epoch": 1.832702457749107, "grad_norm": 1.2825257778167725, "learning_rate": 3.125e-06, "loss": 0.0265, "step": 34376 }, { "epoch": 1.83275577117876, "grad_norm": 3.470484733581543, "learning_rate": 3.1248e-06, "loss": 0.0441, "step": 34377 }, { "epoch": 1.8328090846084129, "grad_norm": 1.5658071041107178, "learning_rate": 3.1246000000000005e-06, "loss": 0.0273, "step": 34378 }, { "epoch": 1.8328623980380658, "grad_norm": 1.4861189126968384, "learning_rate": 3.1244e-06, "loss": 0.0356, "step": 34379 }, { "epoch": 1.8329157114677188, "grad_norm": 2.0314202308654785, "learning_rate": 3.1242e-06, "loss": 0.0308, "step": 34380 }, { "epoch": 1.8329690248973716, "grad_norm": 4.5740742683410645, "learning_rate": 3.1240000000000005e-06, "loss": 0.0194, "step": 34381 }, { "epoch": 1.8330223383270245, "grad_norm": 2.1760690212249756, "learning_rate": 3.1238e-06, "loss": 0.009, "step": 34382 }, { "epoch": 1.8330756517566775, "grad_norm": 3.217064380645752, "learning_rate": 3.1236000000000005e-06, "loss": 0.0176, "step": 34383 }, { "epoch": 1.8331289651863303, "grad_norm": 1.622461199760437, "learning_rate": 3.1234000000000005e-06, "loss": 0.0298, "step": 34384 }, { "epoch": 1.8331822786159835, "grad_norm": 3.1471943855285645, "learning_rate": 3.1232e-06, "loss": 0.0377, "step": 34385 }, { "epoch": 1.8332355920456362, "grad_norm": 0.7889164090156555, "learning_rate": 3.1230000000000004e-06, "loss": 0.0101, "step": 34386 }, { "epoch": 1.8332889054752892, "grad_norm": 2.1378486156463623, "learning_rate": 3.1228000000000004e-06, "loss": 0.0299, "step": 34387 }, { "epoch": 1.8333422189049422, "grad_norm": 3.6115756034851074, "learning_rate": 3.1226e-06, "loss": 0.0531, "step": 34388 }, { "epoch": 1.833395532334595, "grad_norm": 3.261627435684204, "learning_rate": 3.1224000000000004e-06, "loss": 0.0387, "step": 34389 }, { "epoch": 1.8334488457642482, "grad_norm": 1.0388143062591553, "learning_rate": 3.1222000000000004e-06, "loss": 0.0074, "step": 34390 }, { "epoch": 1.833502159193901, "grad_norm": 3.071641683578491, "learning_rate": 3.122e-06, "loss": 0.0263, "step": 34391 }, { "epoch": 1.833555472623554, "grad_norm": 0.8618255853652954, "learning_rate": 3.1218000000000003e-06, "loss": 0.0075, "step": 34392 }, { "epoch": 1.8336087860532069, "grad_norm": 1.8362383842468262, "learning_rate": 3.1216000000000003e-06, "loss": 0.0115, "step": 34393 }, { "epoch": 1.8336620994828596, "grad_norm": 1.0286822319030762, "learning_rate": 3.1214e-06, "loss": 0.0089, "step": 34394 }, { "epoch": 1.8337154129125126, "grad_norm": 1.789434552192688, "learning_rate": 3.1212000000000003e-06, "loss": 0.0145, "step": 34395 }, { "epoch": 1.8337687263421656, "grad_norm": 1.3987535238265991, "learning_rate": 3.1210000000000003e-06, "loss": 0.0218, "step": 34396 }, { "epoch": 1.8338220397718186, "grad_norm": 1.1258161067962646, "learning_rate": 3.1208000000000007e-06, "loss": 0.0137, "step": 34397 }, { "epoch": 1.8338753532014715, "grad_norm": 2.3146872520446777, "learning_rate": 3.1206000000000002e-06, "loss": 0.0363, "step": 34398 }, { "epoch": 1.8339286666311243, "grad_norm": 1.2593051195144653, "learning_rate": 3.1204000000000002e-06, "loss": 0.0268, "step": 34399 }, { "epoch": 1.8339819800607773, "grad_norm": 1.567953109741211, "learning_rate": 3.1202000000000006e-06, "loss": 0.0289, "step": 34400 }, { "epoch": 1.8340352934904303, "grad_norm": 0.9647206664085388, "learning_rate": 3.12e-06, "loss": 0.0215, "step": 34401 }, { "epoch": 1.834088606920083, "grad_norm": 3.1872663497924805, "learning_rate": 3.1198e-06, "loss": 0.0295, "step": 34402 }, { "epoch": 1.8341419203497362, "grad_norm": 0.7724505662918091, "learning_rate": 3.1196000000000006e-06, "loss": 0.0055, "step": 34403 }, { "epoch": 1.834195233779389, "grad_norm": 1.598741054534912, "learning_rate": 3.1194e-06, "loss": 0.0159, "step": 34404 }, { "epoch": 1.834248547209042, "grad_norm": 0.7030660510063171, "learning_rate": 3.1192e-06, "loss": 0.0061, "step": 34405 }, { "epoch": 1.834301860638695, "grad_norm": 0.895896315574646, "learning_rate": 3.1190000000000005e-06, "loss": 0.0057, "step": 34406 }, { "epoch": 1.8343551740683477, "grad_norm": 2.197495698928833, "learning_rate": 3.1188e-06, "loss": 0.0147, "step": 34407 }, { "epoch": 1.834408487498001, "grad_norm": 1.1522959470748901, "learning_rate": 3.1186000000000005e-06, "loss": 0.008, "step": 34408 }, { "epoch": 1.8344618009276537, "grad_norm": 1.6447948217391968, "learning_rate": 3.1184000000000005e-06, "loss": 0.0185, "step": 34409 }, { "epoch": 1.8345151143573066, "grad_norm": 1.1796748638153076, "learning_rate": 3.1182e-06, "loss": 0.0213, "step": 34410 }, { "epoch": 1.8345684277869596, "grad_norm": 1.5029330253601074, "learning_rate": 3.1180000000000005e-06, "loss": 0.0207, "step": 34411 }, { "epoch": 1.8346217412166124, "grad_norm": 1.6138525009155273, "learning_rate": 3.1178000000000005e-06, "loss": 0.0149, "step": 34412 }, { "epoch": 1.8346750546462653, "grad_norm": 1.5887531042099, "learning_rate": 3.1176e-06, "loss": 0.0433, "step": 34413 }, { "epoch": 1.8347283680759183, "grad_norm": 0.7074728012084961, "learning_rate": 3.1174000000000004e-06, "loss": 0.0069, "step": 34414 }, { "epoch": 1.8347816815055713, "grad_norm": 2.150864601135254, "learning_rate": 3.1172000000000004e-06, "loss": 0.0158, "step": 34415 }, { "epoch": 1.8348349949352243, "grad_norm": 1.9826182126998901, "learning_rate": 3.117e-06, "loss": 0.0257, "step": 34416 }, { "epoch": 1.834888308364877, "grad_norm": 2.22947096824646, "learning_rate": 3.1168000000000004e-06, "loss": 0.021, "step": 34417 }, { "epoch": 1.83494162179453, "grad_norm": 1.938970923423767, "learning_rate": 3.1166000000000004e-06, "loss": 0.0347, "step": 34418 }, { "epoch": 1.834994935224183, "grad_norm": 2.6875052452087402, "learning_rate": 3.1164e-06, "loss": 0.0218, "step": 34419 }, { "epoch": 1.8350482486538358, "grad_norm": 1.3232605457305908, "learning_rate": 3.1162000000000003e-06, "loss": 0.0215, "step": 34420 }, { "epoch": 1.835101562083489, "grad_norm": 6.4241814613342285, "learning_rate": 3.1160000000000003e-06, "loss": 0.0318, "step": 34421 }, { "epoch": 1.8351548755131417, "grad_norm": 1.4921191930770874, "learning_rate": 3.1158000000000003e-06, "loss": 0.0172, "step": 34422 }, { "epoch": 1.8352081889427947, "grad_norm": 0.4974929988384247, "learning_rate": 3.1156000000000003e-06, "loss": 0.0041, "step": 34423 }, { "epoch": 1.8352615023724477, "grad_norm": 1.0563727617263794, "learning_rate": 3.1154000000000003e-06, "loss": 0.0062, "step": 34424 }, { "epoch": 1.8353148158021004, "grad_norm": 2.8131675720214844, "learning_rate": 3.1152000000000003e-06, "loss": 0.025, "step": 34425 }, { "epoch": 1.8353681292317536, "grad_norm": 0.7282764315605164, "learning_rate": 3.1150000000000002e-06, "loss": 0.0147, "step": 34426 }, { "epoch": 1.8354214426614064, "grad_norm": 0.51995450258255, "learning_rate": 3.1148000000000002e-06, "loss": 0.004, "step": 34427 }, { "epoch": 1.8354747560910594, "grad_norm": 1.670893907546997, "learning_rate": 3.1146000000000002e-06, "loss": 0.0168, "step": 34428 }, { "epoch": 1.8355280695207123, "grad_norm": 0.9296658635139465, "learning_rate": 3.1144e-06, "loss": 0.0054, "step": 34429 }, { "epoch": 1.835581382950365, "grad_norm": 3.0104503631591797, "learning_rate": 3.1142e-06, "loss": 0.0213, "step": 34430 }, { "epoch": 1.835634696380018, "grad_norm": 1.9639627933502197, "learning_rate": 3.114e-06, "loss": 0.054, "step": 34431 }, { "epoch": 1.835688009809671, "grad_norm": 2.499267816543579, "learning_rate": 3.1138e-06, "loss": 0.031, "step": 34432 }, { "epoch": 1.835741323239324, "grad_norm": 7.145566940307617, "learning_rate": 3.1136000000000006e-06, "loss": 0.0423, "step": 34433 }, { "epoch": 1.835794636668977, "grad_norm": 0.9706305265426636, "learning_rate": 3.1134e-06, "loss": 0.0045, "step": 34434 }, { "epoch": 1.8358479500986298, "grad_norm": 1.2235506772994995, "learning_rate": 3.1132e-06, "loss": 0.0075, "step": 34435 }, { "epoch": 1.8359012635282828, "grad_norm": 4.034278869628906, "learning_rate": 3.1130000000000005e-06, "loss": 0.0234, "step": 34436 }, { "epoch": 1.8359545769579357, "grad_norm": 3.188814640045166, "learning_rate": 3.1128e-06, "loss": 0.0145, "step": 34437 }, { "epoch": 1.8360078903875885, "grad_norm": 2.563044786453247, "learning_rate": 3.1126e-06, "loss": 0.0174, "step": 34438 }, { "epoch": 1.8360612038172417, "grad_norm": 2.119276762008667, "learning_rate": 3.1124000000000005e-06, "loss": 0.0245, "step": 34439 }, { "epoch": 1.8361145172468945, "grad_norm": 1.3152010440826416, "learning_rate": 3.1122e-06, "loss": 0.009, "step": 34440 }, { "epoch": 1.8361678306765474, "grad_norm": 1.0388140678405762, "learning_rate": 3.112e-06, "loss": 0.0063, "step": 34441 }, { "epoch": 1.8362211441062004, "grad_norm": 0.9255484938621521, "learning_rate": 3.1118000000000004e-06, "loss": 0.0073, "step": 34442 }, { "epoch": 1.8362744575358532, "grad_norm": 1.3677171468734741, "learning_rate": 3.1116e-06, "loss": 0.0099, "step": 34443 }, { "epoch": 1.8363277709655064, "grad_norm": 8.091550827026367, "learning_rate": 3.1114000000000004e-06, "loss": 0.0369, "step": 34444 }, { "epoch": 1.8363810843951591, "grad_norm": 6.316157817840576, "learning_rate": 3.1112000000000004e-06, "loss": 0.0203, "step": 34445 }, { "epoch": 1.836434397824812, "grad_norm": 1.7431813478469849, "learning_rate": 3.111e-06, "loss": 0.0217, "step": 34446 }, { "epoch": 1.836487711254465, "grad_norm": 2.3775787353515625, "learning_rate": 3.1108000000000004e-06, "loss": 0.0298, "step": 34447 }, { "epoch": 1.8365410246841178, "grad_norm": 1.8860788345336914, "learning_rate": 3.1106000000000003e-06, "loss": 0.0192, "step": 34448 }, { "epoch": 1.8365943381137708, "grad_norm": 1.1034421920776367, "learning_rate": 3.1104e-06, "loss": 0.0214, "step": 34449 }, { "epoch": 1.8366476515434238, "grad_norm": 2.179959297180176, "learning_rate": 3.1102000000000003e-06, "loss": 0.0213, "step": 34450 }, { "epoch": 1.8367009649730766, "grad_norm": 2.184998035430908, "learning_rate": 3.1100000000000003e-06, "loss": 0.0147, "step": 34451 }, { "epoch": 1.8367542784027298, "grad_norm": 1.123696208000183, "learning_rate": 3.1098e-06, "loss": 0.0129, "step": 34452 }, { "epoch": 1.8368075918323825, "grad_norm": 1.3659590482711792, "learning_rate": 3.1096000000000003e-06, "loss": 0.0308, "step": 34453 }, { "epoch": 1.8368609052620355, "grad_norm": 0.764531672000885, "learning_rate": 3.1094000000000002e-06, "loss": 0.0045, "step": 34454 }, { "epoch": 1.8369142186916885, "grad_norm": 1.7915219068527222, "learning_rate": 3.1092e-06, "loss": 0.0228, "step": 34455 }, { "epoch": 1.8369675321213412, "grad_norm": 0.7304606437683105, "learning_rate": 3.1090000000000002e-06, "loss": 0.0046, "step": 34456 }, { "epoch": 1.8370208455509944, "grad_norm": 0.5390100479125977, "learning_rate": 3.1088e-06, "loss": 0.0036, "step": 34457 }, { "epoch": 1.8370741589806472, "grad_norm": 1.345786690711975, "learning_rate": 3.1086000000000006e-06, "loss": 0.0214, "step": 34458 }, { "epoch": 1.8371274724103002, "grad_norm": 0.9988690614700317, "learning_rate": 3.1084e-06, "loss": 0.0215, "step": 34459 }, { "epoch": 1.8371807858399531, "grad_norm": 1.3702564239501953, "learning_rate": 3.1082e-06, "loss": 0.0285, "step": 34460 }, { "epoch": 1.837234099269606, "grad_norm": 2.3627030849456787, "learning_rate": 3.1080000000000006e-06, "loss": 0.0303, "step": 34461 }, { "epoch": 1.837287412699259, "grad_norm": 2.2170615196228027, "learning_rate": 3.1078e-06, "loss": 0.035, "step": 34462 }, { "epoch": 1.8373407261289119, "grad_norm": 1.5889822244644165, "learning_rate": 3.1076e-06, "loss": 0.0272, "step": 34463 }, { "epoch": 1.8373940395585648, "grad_norm": 1.9772357940673828, "learning_rate": 3.1074000000000005e-06, "loss": 0.0148, "step": 34464 }, { "epoch": 1.8374473529882178, "grad_norm": 1.788458228111267, "learning_rate": 3.1072e-06, "loss": 0.0211, "step": 34465 }, { "epoch": 1.8375006664178706, "grad_norm": 1.5820902585983276, "learning_rate": 3.107e-06, "loss": 0.0107, "step": 34466 }, { "epoch": 1.8375539798475236, "grad_norm": 2.512787103652954, "learning_rate": 3.1068000000000005e-06, "loss": 0.0396, "step": 34467 }, { "epoch": 1.8376072932771765, "grad_norm": 1.3503119945526123, "learning_rate": 3.1066e-06, "loss": 0.0094, "step": 34468 }, { "epoch": 1.8376606067068293, "grad_norm": 1.6947052478790283, "learning_rate": 3.1064000000000004e-06, "loss": 0.0071, "step": 34469 }, { "epoch": 1.8377139201364825, "grad_norm": 2.1710426807403564, "learning_rate": 3.1062000000000004e-06, "loss": 0.0215, "step": 34470 }, { "epoch": 1.8377672335661352, "grad_norm": 5.73272180557251, "learning_rate": 3.106e-06, "loss": 0.0194, "step": 34471 }, { "epoch": 1.8378205469957882, "grad_norm": 2.0608699321746826, "learning_rate": 3.1058000000000004e-06, "loss": 0.0426, "step": 34472 }, { "epoch": 1.8378738604254412, "grad_norm": 1.5299965143203735, "learning_rate": 3.1056000000000004e-06, "loss": 0.0071, "step": 34473 }, { "epoch": 1.837927173855094, "grad_norm": 0.9809420108795166, "learning_rate": 3.1054e-06, "loss": 0.0075, "step": 34474 }, { "epoch": 1.8379804872847472, "grad_norm": 0.4677985906600952, "learning_rate": 3.1052000000000004e-06, "loss": 0.0035, "step": 34475 }, { "epoch": 1.8380338007144, "grad_norm": 3.2314796447753906, "learning_rate": 3.1050000000000003e-06, "loss": 0.016, "step": 34476 }, { "epoch": 1.838087114144053, "grad_norm": 2.5442659854888916, "learning_rate": 3.1048e-06, "loss": 0.024, "step": 34477 }, { "epoch": 1.8381404275737059, "grad_norm": 1.6754685640335083, "learning_rate": 3.1046000000000003e-06, "loss": 0.0219, "step": 34478 }, { "epoch": 1.8381937410033586, "grad_norm": 0.7464643120765686, "learning_rate": 3.1044000000000003e-06, "loss": 0.0056, "step": 34479 }, { "epoch": 1.8382470544330118, "grad_norm": 2.345999240875244, "learning_rate": 3.1042e-06, "loss": 0.0498, "step": 34480 }, { "epoch": 1.8383003678626646, "grad_norm": 1.3713500499725342, "learning_rate": 3.1040000000000003e-06, "loss": 0.0176, "step": 34481 }, { "epoch": 1.8383536812923176, "grad_norm": 1.2504650354385376, "learning_rate": 3.1038000000000003e-06, "loss": 0.007, "step": 34482 }, { "epoch": 1.8384069947219706, "grad_norm": 2.4974374771118164, "learning_rate": 3.1036000000000007e-06, "loss": 0.0245, "step": 34483 }, { "epoch": 1.8384603081516233, "grad_norm": 2.1411783695220947, "learning_rate": 3.1034000000000002e-06, "loss": 0.0114, "step": 34484 }, { "epoch": 1.8385136215812763, "grad_norm": 0.3877056837081909, "learning_rate": 3.1032e-06, "loss": 0.0032, "step": 34485 }, { "epoch": 1.8385669350109293, "grad_norm": 2.6106467247009277, "learning_rate": 3.1030000000000006e-06, "loss": 0.0361, "step": 34486 }, { "epoch": 1.838620248440582, "grad_norm": 1.8410077095031738, "learning_rate": 3.1028e-06, "loss": 0.0263, "step": 34487 }, { "epoch": 1.8386735618702352, "grad_norm": 3.1719746589660645, "learning_rate": 3.1026e-06, "loss": 0.0335, "step": 34488 }, { "epoch": 1.838726875299888, "grad_norm": 1.7791229486465454, "learning_rate": 3.1024000000000006e-06, "loss": 0.0219, "step": 34489 }, { "epoch": 1.838780188729541, "grad_norm": 2.1795685291290283, "learning_rate": 3.1022e-06, "loss": 0.03, "step": 34490 }, { "epoch": 1.838833502159194, "grad_norm": 1.1569995880126953, "learning_rate": 3.102e-06, "loss": 0.0118, "step": 34491 }, { "epoch": 1.8388868155888467, "grad_norm": 2.408125638961792, "learning_rate": 3.1018000000000005e-06, "loss": 0.0296, "step": 34492 }, { "epoch": 1.8389401290185, "grad_norm": 2.138800621032715, "learning_rate": 3.1016e-06, "loss": 0.0159, "step": 34493 }, { "epoch": 1.8389934424481527, "grad_norm": 1.4737075567245483, "learning_rate": 3.1014000000000005e-06, "loss": 0.0106, "step": 34494 }, { "epoch": 1.8390467558778056, "grad_norm": 2.274888515472412, "learning_rate": 3.1012000000000005e-06, "loss": 0.0105, "step": 34495 }, { "epoch": 1.8391000693074586, "grad_norm": 0.7198038101196289, "learning_rate": 3.101e-06, "loss": 0.0063, "step": 34496 }, { "epoch": 1.8391533827371114, "grad_norm": 1.9728986024856567, "learning_rate": 3.1008000000000004e-06, "loss": 0.0162, "step": 34497 }, { "epoch": 1.8392066961667644, "grad_norm": 2.197134256362915, "learning_rate": 3.1006000000000004e-06, "loss": 0.0295, "step": 34498 }, { "epoch": 1.8392600095964173, "grad_norm": 3.1793246269226074, "learning_rate": 3.1004e-06, "loss": 0.0274, "step": 34499 }, { "epoch": 1.8393133230260703, "grad_norm": 1.1633373498916626, "learning_rate": 3.1002000000000004e-06, "loss": 0.0267, "step": 34500 }, { "epoch": 1.8393666364557233, "grad_norm": 1.2987996339797974, "learning_rate": 3.1000000000000004e-06, "loss": 0.0096, "step": 34501 }, { "epoch": 1.839419949885376, "grad_norm": 2.320711612701416, "learning_rate": 3.0998e-06, "loss": 0.0164, "step": 34502 }, { "epoch": 1.839473263315029, "grad_norm": 0.9178284406661987, "learning_rate": 3.0996000000000004e-06, "loss": 0.0159, "step": 34503 }, { "epoch": 1.839526576744682, "grad_norm": 2.142958402633667, "learning_rate": 3.0994000000000003e-06, "loss": 0.0205, "step": 34504 }, { "epoch": 1.8395798901743348, "grad_norm": 1.7048895359039307, "learning_rate": 3.0992e-06, "loss": 0.0225, "step": 34505 }, { "epoch": 1.839633203603988, "grad_norm": 2.12709379196167, "learning_rate": 3.0990000000000003e-06, "loss": 0.0134, "step": 34506 }, { "epoch": 1.8396865170336407, "grad_norm": 1.871856689453125, "learning_rate": 3.0988000000000003e-06, "loss": 0.0096, "step": 34507 }, { "epoch": 1.8397398304632937, "grad_norm": 1.3435757160186768, "learning_rate": 3.0986000000000007e-06, "loss": 0.0231, "step": 34508 }, { "epoch": 1.8397931438929467, "grad_norm": 2.105618476867676, "learning_rate": 3.0984000000000003e-06, "loss": 0.0236, "step": 34509 }, { "epoch": 1.8398464573225994, "grad_norm": 2.0878260135650635, "learning_rate": 3.0982000000000003e-06, "loss": 0.0284, "step": 34510 }, { "epoch": 1.8398997707522526, "grad_norm": 1.5212137699127197, "learning_rate": 3.0980000000000007e-06, "loss": 0.0174, "step": 34511 }, { "epoch": 1.8399530841819054, "grad_norm": 1.7770822048187256, "learning_rate": 3.0978000000000002e-06, "loss": 0.0212, "step": 34512 }, { "epoch": 1.8400063976115584, "grad_norm": 2.154095411300659, "learning_rate": 3.0976e-06, "loss": 0.0176, "step": 34513 }, { "epoch": 1.8400597110412114, "grad_norm": 2.065768003463745, "learning_rate": 3.0974000000000006e-06, "loss": 0.0204, "step": 34514 }, { "epoch": 1.8401130244708641, "grad_norm": 2.8494536876678467, "learning_rate": 3.0972e-06, "loss": 0.0121, "step": 34515 }, { "epoch": 1.840166337900517, "grad_norm": 1.0587687492370605, "learning_rate": 3.097e-06, "loss": 0.0035, "step": 34516 }, { "epoch": 1.84021965133017, "grad_norm": 4.31882905960083, "learning_rate": 3.0968000000000006e-06, "loss": 0.0192, "step": 34517 }, { "epoch": 1.840272964759823, "grad_norm": 2.6265223026275635, "learning_rate": 3.0966e-06, "loss": 0.0142, "step": 34518 }, { "epoch": 1.840326278189476, "grad_norm": 4.494375228881836, "learning_rate": 3.0964000000000005e-06, "loss": 0.0255, "step": 34519 }, { "epoch": 1.8403795916191288, "grad_norm": 2.9458250999450684, "learning_rate": 3.0962000000000005e-06, "loss": 0.02, "step": 34520 }, { "epoch": 1.8404329050487818, "grad_norm": 2.6899828910827637, "learning_rate": 3.096e-06, "loss": 0.0282, "step": 34521 }, { "epoch": 1.8404862184784347, "grad_norm": 2.6703765392303467, "learning_rate": 3.0958000000000005e-06, "loss": 0.0251, "step": 34522 }, { "epoch": 1.8405395319080875, "grad_norm": 2.3472585678100586, "learning_rate": 3.0956000000000005e-06, "loss": 0.0334, "step": 34523 }, { "epoch": 1.8405928453377407, "grad_norm": 1.8459408283233643, "learning_rate": 3.0954e-06, "loss": 0.0182, "step": 34524 }, { "epoch": 1.8406461587673935, "grad_norm": 1.0437822341918945, "learning_rate": 3.0952000000000005e-06, "loss": 0.0062, "step": 34525 }, { "epoch": 1.8406994721970464, "grad_norm": 1.5269534587860107, "learning_rate": 3.0950000000000004e-06, "loss": 0.0082, "step": 34526 }, { "epoch": 1.8407527856266994, "grad_norm": 1.5469443798065186, "learning_rate": 3.0948e-06, "loss": 0.0235, "step": 34527 }, { "epoch": 1.8408060990563522, "grad_norm": 3.9328863620758057, "learning_rate": 3.0946000000000004e-06, "loss": 0.0266, "step": 34528 }, { "epoch": 1.8408594124860054, "grad_norm": 1.4194939136505127, "learning_rate": 3.0944000000000004e-06, "loss": 0.0122, "step": 34529 }, { "epoch": 1.8409127259156581, "grad_norm": 1.7206662893295288, "learning_rate": 3.0942e-06, "loss": 0.0079, "step": 34530 }, { "epoch": 1.8409660393453111, "grad_norm": 1.063563346862793, "learning_rate": 3.0940000000000004e-06, "loss": 0.013, "step": 34531 }, { "epoch": 1.841019352774964, "grad_norm": 3.4505903720855713, "learning_rate": 3.0938000000000003e-06, "loss": 0.0256, "step": 34532 }, { "epoch": 1.8410726662046168, "grad_norm": 1.252862572669983, "learning_rate": 3.0936000000000003e-06, "loss": 0.0127, "step": 34533 }, { "epoch": 1.8411259796342698, "grad_norm": 1.7707690000534058, "learning_rate": 3.0934000000000003e-06, "loss": 0.0326, "step": 34534 }, { "epoch": 1.8411792930639228, "grad_norm": 1.4640750885009766, "learning_rate": 3.0932000000000003e-06, "loss": 0.0083, "step": 34535 }, { "epoch": 1.8412326064935758, "grad_norm": 1.2739325761795044, "learning_rate": 3.0930000000000003e-06, "loss": 0.0114, "step": 34536 }, { "epoch": 1.8412859199232288, "grad_norm": 2.61129093170166, "learning_rate": 3.0928000000000003e-06, "loss": 0.0295, "step": 34537 }, { "epoch": 1.8413392333528815, "grad_norm": 2.147043466567993, "learning_rate": 3.0926000000000003e-06, "loss": 0.0265, "step": 34538 }, { "epoch": 1.8413925467825345, "grad_norm": 1.7184847593307495, "learning_rate": 3.0924000000000002e-06, "loss": 0.0098, "step": 34539 }, { "epoch": 1.8414458602121875, "grad_norm": 2.538060188293457, "learning_rate": 3.0922000000000002e-06, "loss": 0.0349, "step": 34540 }, { "epoch": 1.8414991736418402, "grad_norm": 1.4002281427383423, "learning_rate": 3.092e-06, "loss": 0.0163, "step": 34541 }, { "epoch": 1.8415524870714934, "grad_norm": 1.4508696794509888, "learning_rate": 3.0918e-06, "loss": 0.0087, "step": 34542 }, { "epoch": 1.8416058005011462, "grad_norm": 0.32472875714302063, "learning_rate": 3.0916e-06, "loss": 0.0032, "step": 34543 }, { "epoch": 1.8416591139307992, "grad_norm": 1.226444125175476, "learning_rate": 3.0914000000000006e-06, "loss": 0.0044, "step": 34544 }, { "epoch": 1.8417124273604522, "grad_norm": 0.8749450445175171, "learning_rate": 3.0912e-06, "loss": 0.0068, "step": 34545 }, { "epoch": 1.841765740790105, "grad_norm": 2.952899217605591, "learning_rate": 3.091e-06, "loss": 0.0225, "step": 34546 }, { "epoch": 1.8418190542197581, "grad_norm": 2.045950174331665, "learning_rate": 3.0908000000000005e-06, "loss": 0.0284, "step": 34547 }, { "epoch": 1.8418723676494109, "grad_norm": 1.8050036430358887, "learning_rate": 3.0906e-06, "loss": 0.0313, "step": 34548 }, { "epoch": 1.8419256810790638, "grad_norm": 2.810502052307129, "learning_rate": 3.0904e-06, "loss": 0.0222, "step": 34549 }, { "epoch": 1.8419789945087168, "grad_norm": 1.2949233055114746, "learning_rate": 3.0902000000000005e-06, "loss": 0.0177, "step": 34550 }, { "epoch": 1.8420323079383696, "grad_norm": 1.9128310680389404, "learning_rate": 3.09e-06, "loss": 0.0162, "step": 34551 }, { "epoch": 1.8420856213680226, "grad_norm": 1.7753287553787231, "learning_rate": 3.0898e-06, "loss": 0.0255, "step": 34552 }, { "epoch": 1.8421389347976755, "grad_norm": 1.7196928262710571, "learning_rate": 3.0896000000000005e-06, "loss": 0.0364, "step": 34553 }, { "epoch": 1.8421922482273283, "grad_norm": 2.9241180419921875, "learning_rate": 3.0894e-06, "loss": 0.0226, "step": 34554 }, { "epoch": 1.8422455616569815, "grad_norm": 2.247823715209961, "learning_rate": 3.0892e-06, "loss": 0.0243, "step": 34555 }, { "epoch": 1.8422988750866343, "grad_norm": 1.4984970092773438, "learning_rate": 3.0890000000000004e-06, "loss": 0.0197, "step": 34556 }, { "epoch": 1.8423521885162872, "grad_norm": 3.4522323608398438, "learning_rate": 3.0888e-06, "loss": 0.0134, "step": 34557 }, { "epoch": 1.8424055019459402, "grad_norm": 1.8611658811569214, "learning_rate": 3.0886000000000004e-06, "loss": 0.023, "step": 34558 }, { "epoch": 1.842458815375593, "grad_norm": 4.929976940155029, "learning_rate": 3.0884000000000004e-06, "loss": 0.0487, "step": 34559 }, { "epoch": 1.8425121288052462, "grad_norm": 3.2478232383728027, "learning_rate": 3.0882e-06, "loss": 0.0305, "step": 34560 }, { "epoch": 1.842565442234899, "grad_norm": 1.3043575286865234, "learning_rate": 3.0880000000000003e-06, "loss": 0.0147, "step": 34561 }, { "epoch": 1.842618755664552, "grad_norm": 1.901993989944458, "learning_rate": 3.0878000000000003e-06, "loss": 0.0146, "step": 34562 }, { "epoch": 1.842672069094205, "grad_norm": 0.7128130197525024, "learning_rate": 3.0876e-06, "loss": 0.0053, "step": 34563 }, { "epoch": 1.8427253825238576, "grad_norm": 1.0107102394104004, "learning_rate": 3.0874000000000003e-06, "loss": 0.0054, "step": 34564 }, { "epoch": 1.8427786959535108, "grad_norm": 0.803350567817688, "learning_rate": 3.0872000000000003e-06, "loss": 0.0038, "step": 34565 }, { "epoch": 1.8428320093831636, "grad_norm": 2.0042126178741455, "learning_rate": 3.087e-06, "loss": 0.0349, "step": 34566 }, { "epoch": 1.8428853228128166, "grad_norm": 0.9257439374923706, "learning_rate": 3.0868000000000002e-06, "loss": 0.0126, "step": 34567 }, { "epoch": 1.8429386362424696, "grad_norm": 1.1450031995773315, "learning_rate": 3.0866000000000002e-06, "loss": 0.0054, "step": 34568 }, { "epoch": 1.8429919496721223, "grad_norm": 1.5219868421554565, "learning_rate": 3.0864000000000006e-06, "loss": 0.0088, "step": 34569 }, { "epoch": 1.8430452631017753, "grad_norm": 1.708433985710144, "learning_rate": 3.0862e-06, "loss": 0.0218, "step": 34570 }, { "epoch": 1.8430985765314283, "grad_norm": 2.2645697593688965, "learning_rate": 3.086e-06, "loss": 0.0213, "step": 34571 }, { "epoch": 1.843151889961081, "grad_norm": 2.3659262657165527, "learning_rate": 3.0858000000000006e-06, "loss": 0.0215, "step": 34572 }, { "epoch": 1.8432052033907342, "grad_norm": 0.2321990579366684, "learning_rate": 3.0856e-06, "loss": 0.0023, "step": 34573 }, { "epoch": 1.843258516820387, "grad_norm": 1.2512441873550415, "learning_rate": 3.0854e-06, "loss": 0.0136, "step": 34574 }, { "epoch": 1.84331183025004, "grad_norm": 2.319356918334961, "learning_rate": 3.0852000000000005e-06, "loss": 0.0113, "step": 34575 }, { "epoch": 1.843365143679693, "grad_norm": 1.3111932277679443, "learning_rate": 3.085e-06, "loss": 0.0307, "step": 34576 }, { "epoch": 1.8434184571093457, "grad_norm": 1.661081075668335, "learning_rate": 3.0848e-06, "loss": 0.0377, "step": 34577 }, { "epoch": 1.843471770538999, "grad_norm": 0.4440298080444336, "learning_rate": 3.0846000000000005e-06, "loss": 0.0029, "step": 34578 }, { "epoch": 1.8435250839686517, "grad_norm": 4.308072566986084, "learning_rate": 3.0844e-06, "loss": 0.0205, "step": 34579 }, { "epoch": 1.8435783973983046, "grad_norm": 0.9741210341453552, "learning_rate": 3.0842000000000005e-06, "loss": 0.0137, "step": 34580 }, { "epoch": 1.8436317108279576, "grad_norm": 2.1362674236297607, "learning_rate": 3.0840000000000005e-06, "loss": 0.0127, "step": 34581 }, { "epoch": 1.8436850242576104, "grad_norm": 2.100332498550415, "learning_rate": 3.0838e-06, "loss": 0.0229, "step": 34582 }, { "epoch": 1.8437383376872636, "grad_norm": 1.4238901138305664, "learning_rate": 3.0836000000000004e-06, "loss": 0.0101, "step": 34583 }, { "epoch": 1.8437916511169163, "grad_norm": 0.5727807879447937, "learning_rate": 3.0834000000000004e-06, "loss": 0.005, "step": 34584 }, { "epoch": 1.8438449645465693, "grad_norm": 4.559518337249756, "learning_rate": 3.0832e-06, "loss": 0.0239, "step": 34585 }, { "epoch": 1.8438982779762223, "grad_norm": 0.8698602914810181, "learning_rate": 3.0830000000000004e-06, "loss": 0.0042, "step": 34586 }, { "epoch": 1.843951591405875, "grad_norm": 1.5550765991210938, "learning_rate": 3.0828000000000004e-06, "loss": 0.0248, "step": 34587 }, { "epoch": 1.844004904835528, "grad_norm": 2.1985924243927, "learning_rate": 3.0826e-06, "loss": 0.0245, "step": 34588 }, { "epoch": 1.844058218265181, "grad_norm": 1.327318787574768, "learning_rate": 3.0824000000000003e-06, "loss": 0.0067, "step": 34589 }, { "epoch": 1.8441115316948338, "grad_norm": 3.6658823490142822, "learning_rate": 3.0822000000000003e-06, "loss": 0.0366, "step": 34590 }, { "epoch": 1.844164845124487, "grad_norm": 0.8001952171325684, "learning_rate": 3.082e-06, "loss": 0.0057, "step": 34591 }, { "epoch": 1.8442181585541397, "grad_norm": 1.6851348876953125, "learning_rate": 3.0818000000000003e-06, "loss": 0.0194, "step": 34592 }, { "epoch": 1.8442714719837927, "grad_norm": 1.800913691520691, "learning_rate": 3.0816000000000003e-06, "loss": 0.0261, "step": 34593 }, { "epoch": 1.8443247854134457, "grad_norm": 3.375858783721924, "learning_rate": 3.0814000000000007e-06, "loss": 0.0333, "step": 34594 }, { "epoch": 1.8443780988430984, "grad_norm": 1.7769684791564941, "learning_rate": 3.0812000000000002e-06, "loss": 0.0137, "step": 34595 }, { "epoch": 1.8444314122727516, "grad_norm": 1.1453431844711304, "learning_rate": 3.0810000000000002e-06, "loss": 0.0122, "step": 34596 }, { "epoch": 1.8444847257024044, "grad_norm": 1.3393161296844482, "learning_rate": 3.0808000000000006e-06, "loss": 0.0102, "step": 34597 }, { "epoch": 1.8445380391320574, "grad_norm": 1.25807785987854, "learning_rate": 3.0806e-06, "loss": 0.0055, "step": 34598 }, { "epoch": 1.8445913525617104, "grad_norm": 1.545352578163147, "learning_rate": 3.0804e-06, "loss": 0.0062, "step": 34599 }, { "epoch": 1.8446446659913631, "grad_norm": 2.7956035137176514, "learning_rate": 3.0802000000000006e-06, "loss": 0.0286, "step": 34600 }, { "epoch": 1.844697979421016, "grad_norm": 0.7944750189781189, "learning_rate": 3.08e-06, "loss": 0.0122, "step": 34601 }, { "epoch": 1.844751292850669, "grad_norm": 1.3855900764465332, "learning_rate": 3.0798e-06, "loss": 0.0167, "step": 34602 }, { "epoch": 1.844804606280322, "grad_norm": 1.056206464767456, "learning_rate": 3.0796000000000006e-06, "loss": 0.0046, "step": 34603 }, { "epoch": 1.844857919709975, "grad_norm": 4.182547092437744, "learning_rate": 3.0794e-06, "loss": 0.0571, "step": 34604 }, { "epoch": 1.8449112331396278, "grad_norm": 2.144925355911255, "learning_rate": 3.0792000000000005e-06, "loss": 0.0088, "step": 34605 }, { "epoch": 1.8449645465692808, "grad_norm": 0.49795693159103394, "learning_rate": 3.0790000000000005e-06, "loss": 0.0036, "step": 34606 }, { "epoch": 1.8450178599989338, "grad_norm": 1.168821930885315, "learning_rate": 3.0788e-06, "loss": 0.0056, "step": 34607 }, { "epoch": 1.8450711734285865, "grad_norm": 1.7556018829345703, "learning_rate": 3.0786000000000005e-06, "loss": 0.0227, "step": 34608 }, { "epoch": 1.8451244868582397, "grad_norm": 2.672292470932007, "learning_rate": 3.0784000000000005e-06, "loss": 0.0425, "step": 34609 }, { "epoch": 1.8451778002878925, "grad_norm": 1.361978530883789, "learning_rate": 3.0782e-06, "loss": 0.0071, "step": 34610 }, { "epoch": 1.8452311137175454, "grad_norm": 0.39998695254325867, "learning_rate": 3.0780000000000004e-06, "loss": 0.0033, "step": 34611 }, { "epoch": 1.8452844271471984, "grad_norm": 4.063852787017822, "learning_rate": 3.0778000000000004e-06, "loss": 0.0284, "step": 34612 }, { "epoch": 1.8453377405768512, "grad_norm": 0.9655703902244568, "learning_rate": 3.0776e-06, "loss": 0.0147, "step": 34613 }, { "epoch": 1.8453910540065044, "grad_norm": 2.2819321155548096, "learning_rate": 3.0774000000000004e-06, "loss": 0.0135, "step": 34614 }, { "epoch": 1.8454443674361571, "grad_norm": 1.5940935611724854, "learning_rate": 3.0772000000000004e-06, "loss": 0.0116, "step": 34615 }, { "epoch": 1.8454976808658101, "grad_norm": 2.60652756690979, "learning_rate": 3.077e-06, "loss": 0.0475, "step": 34616 }, { "epoch": 1.845550994295463, "grad_norm": 1.3428014516830444, "learning_rate": 3.0768000000000003e-06, "loss": 0.0119, "step": 34617 }, { "epoch": 1.8456043077251159, "grad_norm": 2.1684348583221436, "learning_rate": 3.0766000000000003e-06, "loss": 0.0375, "step": 34618 }, { "epoch": 1.8456576211547688, "grad_norm": 2.4523983001708984, "learning_rate": 3.0764000000000003e-06, "loss": 0.037, "step": 34619 }, { "epoch": 1.8457109345844218, "grad_norm": 1.7724168300628662, "learning_rate": 3.0762000000000003e-06, "loss": 0.0186, "step": 34620 }, { "epoch": 1.8457642480140748, "grad_norm": 0.569364070892334, "learning_rate": 3.0760000000000003e-06, "loss": 0.0042, "step": 34621 }, { "epoch": 1.8458175614437278, "grad_norm": 2.452112913131714, "learning_rate": 3.0758000000000003e-06, "loss": 0.0468, "step": 34622 }, { "epoch": 1.8458708748733805, "grad_norm": 2.6843740940093994, "learning_rate": 3.0756000000000003e-06, "loss": 0.0224, "step": 34623 }, { "epoch": 1.8459241883030335, "grad_norm": 0.8519481420516968, "learning_rate": 3.0754000000000002e-06, "loss": 0.006, "step": 34624 }, { "epoch": 1.8459775017326865, "grad_norm": 4.08636474609375, "learning_rate": 3.0752000000000002e-06, "loss": 0.0286, "step": 34625 }, { "epoch": 1.8460308151623392, "grad_norm": 2.3497273921966553, "learning_rate": 3.075e-06, "loss": 0.0175, "step": 34626 }, { "epoch": 1.8460841285919924, "grad_norm": 0.9512211084365845, "learning_rate": 3.0748e-06, "loss": 0.008, "step": 34627 }, { "epoch": 1.8461374420216452, "grad_norm": 1.8049025535583496, "learning_rate": 3.0746e-06, "loss": 0.0407, "step": 34628 }, { "epoch": 1.8461907554512982, "grad_norm": 2.0178937911987305, "learning_rate": 3.0744e-06, "loss": 0.0368, "step": 34629 }, { "epoch": 1.8462440688809512, "grad_norm": 0.7804178595542908, "learning_rate": 3.0742000000000006e-06, "loss": 0.0061, "step": 34630 }, { "epoch": 1.846297382310604, "grad_norm": 2.0145857334136963, "learning_rate": 3.074e-06, "loss": 0.0192, "step": 34631 }, { "epoch": 1.8463506957402571, "grad_norm": 2.4499690532684326, "learning_rate": 3.0738e-06, "loss": 0.0108, "step": 34632 }, { "epoch": 1.8464040091699099, "grad_norm": 1.9168621301651, "learning_rate": 3.0736000000000005e-06, "loss": 0.0158, "step": 34633 }, { "epoch": 1.8464573225995629, "grad_norm": 2.1614344120025635, "learning_rate": 3.0734e-06, "loss": 0.0481, "step": 34634 }, { "epoch": 1.8465106360292158, "grad_norm": 2.098801374435425, "learning_rate": 3.0732e-06, "loss": 0.0351, "step": 34635 }, { "epoch": 1.8465639494588686, "grad_norm": 1.5502243041992188, "learning_rate": 3.0730000000000005e-06, "loss": 0.0066, "step": 34636 }, { "epoch": 1.8466172628885216, "grad_norm": 0.7356614470481873, "learning_rate": 3.0728e-06, "loss": 0.0063, "step": 34637 }, { "epoch": 1.8466705763181746, "grad_norm": 2.7493245601654053, "learning_rate": 3.0726e-06, "loss": 0.0335, "step": 34638 }, { "epoch": 1.8467238897478275, "grad_norm": 0.876605212688446, "learning_rate": 3.0724000000000004e-06, "loss": 0.0057, "step": 34639 }, { "epoch": 1.8467772031774805, "grad_norm": 0.26627856492996216, "learning_rate": 3.0722e-06, "loss": 0.0028, "step": 34640 }, { "epoch": 1.8468305166071333, "grad_norm": 2.6061363220214844, "learning_rate": 3.072e-06, "loss": 0.0161, "step": 34641 }, { "epoch": 1.8468838300367862, "grad_norm": 0.7220683693885803, "learning_rate": 3.0718000000000004e-06, "loss": 0.006, "step": 34642 }, { "epoch": 1.8469371434664392, "grad_norm": 1.6636672019958496, "learning_rate": 3.0716e-06, "loss": 0.015, "step": 34643 }, { "epoch": 1.846990456896092, "grad_norm": 2.1023099422454834, "learning_rate": 3.0714000000000004e-06, "loss": 0.0196, "step": 34644 }, { "epoch": 1.8470437703257452, "grad_norm": 2.5441906452178955, "learning_rate": 3.0712000000000003e-06, "loss": 0.0423, "step": 34645 }, { "epoch": 1.847097083755398, "grad_norm": 2.250174045562744, "learning_rate": 3.071e-06, "loss": 0.0267, "step": 34646 }, { "epoch": 1.847150397185051, "grad_norm": 1.1307038068771362, "learning_rate": 3.0708000000000003e-06, "loss": 0.0107, "step": 34647 }, { "epoch": 1.847203710614704, "grad_norm": 1.975797176361084, "learning_rate": 3.0706000000000003e-06, "loss": 0.0132, "step": 34648 }, { "epoch": 1.8472570240443567, "grad_norm": 1.2012406587600708, "learning_rate": 3.0704e-06, "loss": 0.0093, "step": 34649 }, { "epoch": 1.8473103374740099, "grad_norm": 1.5280592441558838, "learning_rate": 3.0702000000000003e-06, "loss": 0.0163, "step": 34650 }, { "epoch": 1.8473636509036626, "grad_norm": 1.1988658905029297, "learning_rate": 3.0700000000000003e-06, "loss": 0.0237, "step": 34651 }, { "epoch": 1.8474169643333156, "grad_norm": 0.7258983254432678, "learning_rate": 3.0698e-06, "loss": 0.0062, "step": 34652 }, { "epoch": 1.8474702777629686, "grad_norm": 1.1182936429977417, "learning_rate": 3.0696000000000002e-06, "loss": 0.0285, "step": 34653 }, { "epoch": 1.8475235911926213, "grad_norm": 2.2455403804779053, "learning_rate": 3.0694e-06, "loss": 0.0276, "step": 34654 }, { "epoch": 1.8475769046222743, "grad_norm": 0.58152836561203, "learning_rate": 3.0692000000000006e-06, "loss": 0.0042, "step": 34655 }, { "epoch": 1.8476302180519273, "grad_norm": 3.5606741905212402, "learning_rate": 3.069e-06, "loss": 0.0159, "step": 34656 }, { "epoch": 1.84768353148158, "grad_norm": 2.1163430213928223, "learning_rate": 3.0688e-06, "loss": 0.0214, "step": 34657 }, { "epoch": 1.8477368449112332, "grad_norm": 1.0770150423049927, "learning_rate": 3.0686000000000006e-06, "loss": 0.0072, "step": 34658 }, { "epoch": 1.847790158340886, "grad_norm": 3.5351390838623047, "learning_rate": 3.0684e-06, "loss": 0.0365, "step": 34659 }, { "epoch": 1.847843471770539, "grad_norm": 1.3690499067306519, "learning_rate": 3.0682e-06, "loss": 0.0334, "step": 34660 }, { "epoch": 1.847896785200192, "grad_norm": 1.1829615831375122, "learning_rate": 3.0680000000000005e-06, "loss": 0.017, "step": 34661 }, { "epoch": 1.8479500986298447, "grad_norm": 1.1905947923660278, "learning_rate": 3.0678e-06, "loss": 0.0083, "step": 34662 }, { "epoch": 1.848003412059498, "grad_norm": 1.8635510206222534, "learning_rate": 3.0676e-06, "loss": 0.016, "step": 34663 }, { "epoch": 1.8480567254891507, "grad_norm": 1.5116007328033447, "learning_rate": 3.0674000000000005e-06, "loss": 0.0142, "step": 34664 }, { "epoch": 1.8481100389188037, "grad_norm": 2.122194766998291, "learning_rate": 3.0672e-06, "loss": 0.0234, "step": 34665 }, { "epoch": 1.8481633523484566, "grad_norm": 0.3732774555683136, "learning_rate": 3.067e-06, "loss": 0.0023, "step": 34666 }, { "epoch": 1.8482166657781094, "grad_norm": 1.5266485214233398, "learning_rate": 3.0668000000000004e-06, "loss": 0.0279, "step": 34667 }, { "epoch": 1.8482699792077626, "grad_norm": 1.992873191833496, "learning_rate": 3.0666e-06, "loss": 0.0127, "step": 34668 }, { "epoch": 1.8483232926374153, "grad_norm": 1.1509642601013184, "learning_rate": 3.0664000000000004e-06, "loss": 0.0197, "step": 34669 }, { "epoch": 1.8483766060670683, "grad_norm": 1.390001654624939, "learning_rate": 3.0662000000000004e-06, "loss": 0.0317, "step": 34670 }, { "epoch": 1.8484299194967213, "grad_norm": 0.6640561819076538, "learning_rate": 3.066e-06, "loss": 0.0058, "step": 34671 }, { "epoch": 1.848483232926374, "grad_norm": 1.3829199075698853, "learning_rate": 3.0658000000000004e-06, "loss": 0.007, "step": 34672 }, { "epoch": 1.848536546356027, "grad_norm": 2.2258312702178955, "learning_rate": 3.0656000000000003e-06, "loss": 0.0108, "step": 34673 }, { "epoch": 1.84858985978568, "grad_norm": 2.0422298908233643, "learning_rate": 3.0654e-06, "loss": 0.0158, "step": 34674 }, { "epoch": 1.8486431732153328, "grad_norm": 3.0284640789031982, "learning_rate": 3.0652000000000003e-06, "loss": 0.0432, "step": 34675 }, { "epoch": 1.848696486644986, "grad_norm": 2.6202032566070557, "learning_rate": 3.0650000000000003e-06, "loss": 0.0353, "step": 34676 }, { "epoch": 1.8487498000746387, "grad_norm": 1.6587296724319458, "learning_rate": 3.0648e-06, "loss": 0.0137, "step": 34677 }, { "epoch": 1.8488031135042917, "grad_norm": 1.7322866916656494, "learning_rate": 3.0646000000000003e-06, "loss": 0.0182, "step": 34678 }, { "epoch": 1.8488564269339447, "grad_norm": 0.34898829460144043, "learning_rate": 3.0644000000000003e-06, "loss": 0.0029, "step": 34679 }, { "epoch": 1.8489097403635975, "grad_norm": 3.17284893989563, "learning_rate": 3.0642000000000007e-06, "loss": 0.0253, "step": 34680 }, { "epoch": 1.8489630537932507, "grad_norm": 1.5044264793395996, "learning_rate": 3.0640000000000002e-06, "loss": 0.0128, "step": 34681 }, { "epoch": 1.8490163672229034, "grad_norm": 1.5215914249420166, "learning_rate": 3.0638e-06, "loss": 0.0057, "step": 34682 }, { "epoch": 1.8490696806525564, "grad_norm": 4.041991710662842, "learning_rate": 3.0636000000000006e-06, "loss": 0.0406, "step": 34683 }, { "epoch": 1.8491229940822094, "grad_norm": 2.277676582336426, "learning_rate": 3.0634e-06, "loss": 0.0164, "step": 34684 }, { "epoch": 1.8491763075118621, "grad_norm": 1.8652857542037964, "learning_rate": 3.0632e-06, "loss": 0.0216, "step": 34685 }, { "epoch": 1.8492296209415153, "grad_norm": 3.0327060222625732, "learning_rate": 3.0630000000000006e-06, "loss": 0.0361, "step": 34686 }, { "epoch": 1.849282934371168, "grad_norm": 4.9765214920043945, "learning_rate": 3.0628e-06, "loss": 0.0233, "step": 34687 }, { "epoch": 1.849336247800821, "grad_norm": 1.854183316230774, "learning_rate": 3.0626e-06, "loss": 0.0169, "step": 34688 }, { "epoch": 1.849389561230474, "grad_norm": 1.8495311737060547, "learning_rate": 3.0624000000000005e-06, "loss": 0.0096, "step": 34689 }, { "epoch": 1.8494428746601268, "grad_norm": 1.8981471061706543, "learning_rate": 3.0622e-06, "loss": 0.0224, "step": 34690 }, { "epoch": 1.8494961880897798, "grad_norm": 6.216803550720215, "learning_rate": 3.0620000000000005e-06, "loss": 0.0484, "step": 34691 }, { "epoch": 1.8495495015194328, "grad_norm": 1.4220741987228394, "learning_rate": 3.0618000000000005e-06, "loss": 0.0088, "step": 34692 }, { "epoch": 1.8496028149490855, "grad_norm": 1.0197412967681885, "learning_rate": 3.0616e-06, "loss": 0.0266, "step": 34693 }, { "epoch": 1.8496561283787387, "grad_norm": 2.879497766494751, "learning_rate": 3.0614000000000005e-06, "loss": 0.0421, "step": 34694 }, { "epoch": 1.8497094418083915, "grad_norm": 3.78304123878479, "learning_rate": 3.0612000000000004e-06, "loss": 0.0324, "step": 34695 }, { "epoch": 1.8497627552380445, "grad_norm": 0.9367305040359497, "learning_rate": 3.061e-06, "loss": 0.0134, "step": 34696 }, { "epoch": 1.8498160686676974, "grad_norm": 2.46740984916687, "learning_rate": 3.0608000000000004e-06, "loss": 0.0359, "step": 34697 }, { "epoch": 1.8498693820973502, "grad_norm": 4.090514183044434, "learning_rate": 3.0606000000000004e-06, "loss": 0.0138, "step": 34698 }, { "epoch": 1.8499226955270034, "grad_norm": 1.7677457332611084, "learning_rate": 3.0604e-06, "loss": 0.0227, "step": 34699 }, { "epoch": 1.8499760089566561, "grad_norm": 2.6556437015533447, "learning_rate": 3.0602000000000004e-06, "loss": 0.0399, "step": 34700 }, { "epoch": 1.8500293223863091, "grad_norm": 2.1072235107421875, "learning_rate": 3.0600000000000003e-06, "loss": 0.0196, "step": 34701 }, { "epoch": 1.850082635815962, "grad_norm": 2.4681856632232666, "learning_rate": 3.0598e-06, "loss": 0.0252, "step": 34702 }, { "epoch": 1.8501359492456149, "grad_norm": 3.2748687267303467, "learning_rate": 3.0596000000000003e-06, "loss": 0.0228, "step": 34703 }, { "epoch": 1.850189262675268, "grad_norm": 1.7242220640182495, "learning_rate": 3.0594000000000003e-06, "loss": 0.0265, "step": 34704 }, { "epoch": 1.8502425761049208, "grad_norm": 2.19087815284729, "learning_rate": 3.0592000000000007e-06, "loss": 0.024, "step": 34705 }, { "epoch": 1.8502958895345738, "grad_norm": 1.5506037473678589, "learning_rate": 3.0590000000000003e-06, "loss": 0.0091, "step": 34706 }, { "epoch": 1.8503492029642268, "grad_norm": 0.9449397921562195, "learning_rate": 3.0588000000000003e-06, "loss": 0.007, "step": 34707 }, { "epoch": 1.8504025163938795, "grad_norm": 0.7856807112693787, "learning_rate": 3.0586000000000007e-06, "loss": 0.0063, "step": 34708 }, { "epoch": 1.8504558298235325, "grad_norm": 2.3757970333099365, "learning_rate": 3.0584000000000002e-06, "loss": 0.0297, "step": 34709 }, { "epoch": 1.8505091432531855, "grad_norm": 1.577226161956787, "learning_rate": 3.0582000000000002e-06, "loss": 0.0116, "step": 34710 }, { "epoch": 1.8505624566828383, "grad_norm": 0.21444077789783478, "learning_rate": 3.0580000000000006e-06, "loss": 0.0018, "step": 34711 }, { "epoch": 1.8506157701124915, "grad_norm": 1.844854474067688, "learning_rate": 3.0578e-06, "loss": 0.0184, "step": 34712 }, { "epoch": 1.8506690835421442, "grad_norm": 0.6580719947814941, "learning_rate": 3.0576e-06, "loss": 0.0038, "step": 34713 }, { "epoch": 1.8507223969717972, "grad_norm": 2.4086966514587402, "learning_rate": 3.0574000000000006e-06, "loss": 0.0117, "step": 34714 }, { "epoch": 1.8507757104014502, "grad_norm": 2.813448667526245, "learning_rate": 3.0572e-06, "loss": 0.0369, "step": 34715 }, { "epoch": 1.850829023831103, "grad_norm": 3.4191505908966064, "learning_rate": 3.0570000000000005e-06, "loss": 0.0209, "step": 34716 }, { "epoch": 1.8508823372607561, "grad_norm": 1.53862726688385, "learning_rate": 3.0568000000000005e-06, "loss": 0.009, "step": 34717 }, { "epoch": 1.8509356506904089, "grad_norm": 3.7820801734924316, "learning_rate": 3.0566e-06, "loss": 0.0274, "step": 34718 }, { "epoch": 1.8509889641200619, "grad_norm": 0.33461105823516846, "learning_rate": 3.0564000000000005e-06, "loss": 0.0035, "step": 34719 }, { "epoch": 1.8510422775497148, "grad_norm": 1.622017502784729, "learning_rate": 3.0562000000000005e-06, "loss": 0.0367, "step": 34720 }, { "epoch": 1.8510955909793676, "grad_norm": 2.5605273246765137, "learning_rate": 3.056e-06, "loss": 0.0217, "step": 34721 }, { "epoch": 1.8511489044090206, "grad_norm": 2.403947114944458, "learning_rate": 3.0558000000000005e-06, "loss": 0.0212, "step": 34722 }, { "epoch": 1.8512022178386736, "grad_norm": 0.6788837313652039, "learning_rate": 3.0556000000000004e-06, "loss": 0.0042, "step": 34723 }, { "epoch": 1.8512555312683265, "grad_norm": 1.7126696109771729, "learning_rate": 3.0554e-06, "loss": 0.0332, "step": 34724 }, { "epoch": 1.8513088446979795, "grad_norm": 2.8966517448425293, "learning_rate": 3.0552000000000004e-06, "loss": 0.0146, "step": 34725 }, { "epoch": 1.8513621581276323, "grad_norm": 2.043839454650879, "learning_rate": 3.0550000000000004e-06, "loss": 0.0279, "step": 34726 }, { "epoch": 1.8514154715572853, "grad_norm": 3.6738643646240234, "learning_rate": 3.0548e-06, "loss": 0.0282, "step": 34727 }, { "epoch": 1.8514687849869382, "grad_norm": 2.083961009979248, "learning_rate": 3.0546000000000004e-06, "loss": 0.0165, "step": 34728 }, { "epoch": 1.851522098416591, "grad_norm": 0.9401972889900208, "learning_rate": 3.0544000000000004e-06, "loss": 0.006, "step": 34729 }, { "epoch": 1.8515754118462442, "grad_norm": 2.8586552143096924, "learning_rate": 3.0542000000000003e-06, "loss": 0.0436, "step": 34730 }, { "epoch": 1.851628725275897, "grad_norm": 2.3817288875579834, "learning_rate": 3.0540000000000003e-06, "loss": 0.0233, "step": 34731 }, { "epoch": 1.85168203870555, "grad_norm": 1.2694615125656128, "learning_rate": 3.0538000000000003e-06, "loss": 0.012, "step": 34732 }, { "epoch": 1.851735352135203, "grad_norm": 4.034708499908447, "learning_rate": 3.0536000000000003e-06, "loss": 0.0163, "step": 34733 }, { "epoch": 1.8517886655648557, "grad_norm": 1.8876525163650513, "learning_rate": 3.0534000000000003e-06, "loss": 0.0147, "step": 34734 }, { "epoch": 1.8518419789945089, "grad_norm": 0.8939098715782166, "learning_rate": 3.0532000000000003e-06, "loss": 0.0051, "step": 34735 }, { "epoch": 1.8518952924241616, "grad_norm": 1.1553195714950562, "learning_rate": 3.0530000000000002e-06, "loss": 0.017, "step": 34736 }, { "epoch": 1.8519486058538146, "grad_norm": 1.3300412893295288, "learning_rate": 3.0528000000000002e-06, "loss": 0.0156, "step": 34737 }, { "epoch": 1.8520019192834676, "grad_norm": 0.7862852215766907, "learning_rate": 3.0526000000000002e-06, "loss": 0.005, "step": 34738 }, { "epoch": 1.8520552327131203, "grad_norm": 1.8497968912124634, "learning_rate": 3.0524e-06, "loss": 0.0134, "step": 34739 }, { "epoch": 1.8521085461427733, "grad_norm": 2.010633707046509, "learning_rate": 3.0522e-06, "loss": 0.0386, "step": 34740 }, { "epoch": 1.8521618595724263, "grad_norm": 1.7520549297332764, "learning_rate": 3.0520000000000006e-06, "loss": 0.0408, "step": 34741 }, { "epoch": 1.8522151730020793, "grad_norm": 1.5391961336135864, "learning_rate": 3.0518e-06, "loss": 0.0231, "step": 34742 }, { "epoch": 1.8522684864317323, "grad_norm": 1.6342644691467285, "learning_rate": 3.0516e-06, "loss": 0.0292, "step": 34743 }, { "epoch": 1.852321799861385, "grad_norm": 1.6519895792007446, "learning_rate": 3.0514000000000005e-06, "loss": 0.0247, "step": 34744 }, { "epoch": 1.852375113291038, "grad_norm": 1.7623777389526367, "learning_rate": 3.0512e-06, "loss": 0.0111, "step": 34745 }, { "epoch": 1.852428426720691, "grad_norm": 1.54262113571167, "learning_rate": 3.051e-06, "loss": 0.0057, "step": 34746 }, { "epoch": 1.8524817401503437, "grad_norm": 2.2281229496002197, "learning_rate": 3.0508000000000005e-06, "loss": 0.0185, "step": 34747 }, { "epoch": 1.852535053579997, "grad_norm": 2.3496668338775635, "learning_rate": 3.0506e-06, "loss": 0.0246, "step": 34748 }, { "epoch": 1.8525883670096497, "grad_norm": 1.2260494232177734, "learning_rate": 3.0504e-06, "loss": 0.0115, "step": 34749 }, { "epoch": 1.8526416804393027, "grad_norm": 1.1493381261825562, "learning_rate": 3.0502000000000005e-06, "loss": 0.0168, "step": 34750 }, { "epoch": 1.8526949938689556, "grad_norm": 2.0419607162475586, "learning_rate": 3.05e-06, "loss": 0.0276, "step": 34751 }, { "epoch": 1.8527483072986084, "grad_norm": 2.3605968952178955, "learning_rate": 3.0498e-06, "loss": 0.009, "step": 34752 }, { "epoch": 1.8528016207282616, "grad_norm": 1.0892449617385864, "learning_rate": 3.0496000000000004e-06, "loss": 0.0134, "step": 34753 }, { "epoch": 1.8528549341579144, "grad_norm": 1.4415829181671143, "learning_rate": 3.0494e-06, "loss": 0.0306, "step": 34754 }, { "epoch": 1.8529082475875673, "grad_norm": 3.846292734146118, "learning_rate": 3.0492000000000004e-06, "loss": 0.0454, "step": 34755 }, { "epoch": 1.8529615610172203, "grad_norm": 3.006056785583496, "learning_rate": 3.0490000000000004e-06, "loss": 0.0421, "step": 34756 }, { "epoch": 1.853014874446873, "grad_norm": 3.0580127239227295, "learning_rate": 3.0488e-06, "loss": 0.0557, "step": 34757 }, { "epoch": 1.853068187876526, "grad_norm": 2.725811004638672, "learning_rate": 3.0486000000000003e-06, "loss": 0.0244, "step": 34758 }, { "epoch": 1.853121501306179, "grad_norm": 0.7033855319023132, "learning_rate": 3.0484000000000003e-06, "loss": 0.0052, "step": 34759 }, { "epoch": 1.853174814735832, "grad_norm": 3.175264358520508, "learning_rate": 3.0482e-06, "loss": 0.0578, "step": 34760 }, { "epoch": 1.853228128165485, "grad_norm": 3.2487807273864746, "learning_rate": 3.0480000000000003e-06, "loss": 0.0116, "step": 34761 }, { "epoch": 1.8532814415951377, "grad_norm": 1.1652085781097412, "learning_rate": 3.0478000000000003e-06, "loss": 0.0068, "step": 34762 }, { "epoch": 1.8533347550247907, "grad_norm": 2.5551435947418213, "learning_rate": 3.0476e-06, "loss": 0.0377, "step": 34763 }, { "epoch": 1.8533880684544437, "grad_norm": 2.3633816242218018, "learning_rate": 3.0474000000000002e-06, "loss": 0.0314, "step": 34764 }, { "epoch": 1.8534413818840965, "grad_norm": 1.7352063655853271, "learning_rate": 3.0472000000000002e-06, "loss": 0.0111, "step": 34765 }, { "epoch": 1.8534946953137497, "grad_norm": 0.4765022397041321, "learning_rate": 3.0470000000000006e-06, "loss": 0.0062, "step": 34766 }, { "epoch": 1.8535480087434024, "grad_norm": 2.20481538772583, "learning_rate": 3.0468e-06, "loss": 0.0207, "step": 34767 }, { "epoch": 1.8536013221730554, "grad_norm": 2.0363171100616455, "learning_rate": 3.0466e-06, "loss": 0.0116, "step": 34768 }, { "epoch": 1.8536546356027084, "grad_norm": 1.61104416847229, "learning_rate": 3.0464000000000006e-06, "loss": 0.0182, "step": 34769 }, { "epoch": 1.8537079490323611, "grad_norm": 1.2383675575256348, "learning_rate": 3.0462e-06, "loss": 0.0245, "step": 34770 }, { "epoch": 1.8537612624620143, "grad_norm": 0.9863483309745789, "learning_rate": 3.046e-06, "loss": 0.009, "step": 34771 }, { "epoch": 1.853814575891667, "grad_norm": 2.356175422668457, "learning_rate": 3.0458000000000006e-06, "loss": 0.0365, "step": 34772 }, { "epoch": 1.85386788932132, "grad_norm": 2.121248722076416, "learning_rate": 3.0456e-06, "loss": 0.011, "step": 34773 }, { "epoch": 1.853921202750973, "grad_norm": 2.6516475677490234, "learning_rate": 3.0454e-06, "loss": 0.0252, "step": 34774 }, { "epoch": 1.8539745161806258, "grad_norm": 0.8968327641487122, "learning_rate": 3.0452000000000005e-06, "loss": 0.0125, "step": 34775 }, { "epoch": 1.8540278296102788, "grad_norm": 1.6011028289794922, "learning_rate": 3.045e-06, "loss": 0.0182, "step": 34776 }, { "epoch": 1.8540811430399318, "grad_norm": 1.7143313884735107, "learning_rate": 3.0448e-06, "loss": 0.0208, "step": 34777 }, { "epoch": 1.8541344564695845, "grad_norm": 0.8166210055351257, "learning_rate": 3.0446000000000005e-06, "loss": 0.0091, "step": 34778 }, { "epoch": 1.8541877698992377, "grad_norm": 0.6907415986061096, "learning_rate": 3.0444e-06, "loss": 0.0042, "step": 34779 }, { "epoch": 1.8542410833288905, "grad_norm": 1.3347501754760742, "learning_rate": 3.0442000000000004e-06, "loss": 0.0172, "step": 34780 }, { "epoch": 1.8542943967585435, "grad_norm": 1.4989099502563477, "learning_rate": 3.0440000000000004e-06, "loss": 0.0117, "step": 34781 }, { "epoch": 1.8543477101881964, "grad_norm": 1.1984925270080566, "learning_rate": 3.0438e-06, "loss": 0.0115, "step": 34782 }, { "epoch": 1.8544010236178492, "grad_norm": 2.4894635677337646, "learning_rate": 3.0436000000000004e-06, "loss": 0.0301, "step": 34783 }, { "epoch": 1.8544543370475024, "grad_norm": 1.9916552305221558, "learning_rate": 3.0434000000000004e-06, "loss": 0.033, "step": 34784 }, { "epoch": 1.8545076504771552, "grad_norm": 1.4765450954437256, "learning_rate": 3.0432e-06, "loss": 0.0118, "step": 34785 }, { "epoch": 1.8545609639068081, "grad_norm": 2.4254910945892334, "learning_rate": 3.0430000000000003e-06, "loss": 0.0631, "step": 34786 }, { "epoch": 1.8546142773364611, "grad_norm": 2.3180932998657227, "learning_rate": 3.0428000000000003e-06, "loss": 0.0303, "step": 34787 }, { "epoch": 1.8546675907661139, "grad_norm": 1.1727885007858276, "learning_rate": 3.0426e-06, "loss": 0.0087, "step": 34788 }, { "epoch": 1.854720904195767, "grad_norm": 1.5288726091384888, "learning_rate": 3.0424000000000003e-06, "loss": 0.0269, "step": 34789 }, { "epoch": 1.8547742176254198, "grad_norm": 2.518770217895508, "learning_rate": 3.0422000000000003e-06, "loss": 0.0286, "step": 34790 }, { "epoch": 1.8548275310550728, "grad_norm": 1.0348093509674072, "learning_rate": 3.0420000000000007e-06, "loss": 0.0148, "step": 34791 }, { "epoch": 1.8548808444847258, "grad_norm": 3.059701681137085, "learning_rate": 3.0418000000000003e-06, "loss": 0.0252, "step": 34792 }, { "epoch": 1.8549341579143785, "grad_norm": 1.5504733324050903, "learning_rate": 3.0416000000000002e-06, "loss": 0.0108, "step": 34793 }, { "epoch": 1.8549874713440315, "grad_norm": 1.948988437652588, "learning_rate": 3.0414000000000006e-06, "loss": 0.0296, "step": 34794 }, { "epoch": 1.8550407847736845, "grad_norm": 2.274900197982788, "learning_rate": 3.0412e-06, "loss": 0.0337, "step": 34795 }, { "epoch": 1.8550940982033373, "grad_norm": 1.7929893732070923, "learning_rate": 3.041e-06, "loss": 0.0222, "step": 34796 }, { "epoch": 1.8551474116329905, "grad_norm": 0.7158706188201904, "learning_rate": 3.0408000000000006e-06, "loss": 0.0045, "step": 34797 }, { "epoch": 1.8552007250626432, "grad_norm": 2.0022990703582764, "learning_rate": 3.0406e-06, "loss": 0.0169, "step": 34798 }, { "epoch": 1.8552540384922962, "grad_norm": 1.3983845710754395, "learning_rate": 3.0404e-06, "loss": 0.0192, "step": 34799 }, { "epoch": 1.8553073519219492, "grad_norm": 1.267361044883728, "learning_rate": 3.0402000000000006e-06, "loss": 0.0149, "step": 34800 }, { "epoch": 1.855360665351602, "grad_norm": 1.5195621252059937, "learning_rate": 3.04e-06, "loss": 0.0265, "step": 34801 }, { "epoch": 1.8554139787812551, "grad_norm": 2.995483875274658, "learning_rate": 3.0398e-06, "loss": 0.0376, "step": 34802 }, { "epoch": 1.855467292210908, "grad_norm": 2.791846752166748, "learning_rate": 3.0396000000000005e-06, "loss": 0.0193, "step": 34803 }, { "epoch": 1.8555206056405609, "grad_norm": 0.3114989995956421, "learning_rate": 3.0394e-06, "loss": 0.0029, "step": 34804 }, { "epoch": 1.8555739190702139, "grad_norm": 1.9238535165786743, "learning_rate": 3.0392000000000005e-06, "loss": 0.0138, "step": 34805 }, { "epoch": 1.8556272324998666, "grad_norm": 0.8671085834503174, "learning_rate": 3.0390000000000005e-06, "loss": 0.0121, "step": 34806 }, { "epoch": 1.8556805459295198, "grad_norm": 2.4492461681365967, "learning_rate": 3.0388e-06, "loss": 0.0216, "step": 34807 }, { "epoch": 1.8557338593591726, "grad_norm": 1.9071035385131836, "learning_rate": 3.0386000000000004e-06, "loss": 0.0132, "step": 34808 }, { "epoch": 1.8557871727888255, "grad_norm": 0.9440749883651733, "learning_rate": 3.0384000000000004e-06, "loss": 0.0061, "step": 34809 }, { "epoch": 1.8558404862184785, "grad_norm": 0.7552129030227661, "learning_rate": 3.0382e-06, "loss": 0.0152, "step": 34810 }, { "epoch": 1.8558937996481313, "grad_norm": 1.2572650909423828, "learning_rate": 3.0380000000000004e-06, "loss": 0.0288, "step": 34811 }, { "epoch": 1.8559471130777843, "grad_norm": 1.3704196214675903, "learning_rate": 3.0378000000000004e-06, "loss": 0.0192, "step": 34812 }, { "epoch": 1.8560004265074372, "grad_norm": 3.828885078430176, "learning_rate": 3.0376e-06, "loss": 0.0134, "step": 34813 }, { "epoch": 1.85605373993709, "grad_norm": 1.456399917602539, "learning_rate": 3.0374000000000003e-06, "loss": 0.0178, "step": 34814 }, { "epoch": 1.8561070533667432, "grad_norm": 1.6813640594482422, "learning_rate": 3.0372000000000003e-06, "loss": 0.0158, "step": 34815 }, { "epoch": 1.856160366796396, "grad_norm": 1.9092601537704468, "learning_rate": 3.0370000000000003e-06, "loss": 0.0167, "step": 34816 }, { "epoch": 1.856213680226049, "grad_norm": 2.263887643814087, "learning_rate": 3.0368000000000003e-06, "loss": 0.0186, "step": 34817 }, { "epoch": 1.856266993655702, "grad_norm": 2.209368944168091, "learning_rate": 3.0366000000000003e-06, "loss": 0.0267, "step": 34818 }, { "epoch": 1.8563203070853547, "grad_norm": 1.1560875177383423, "learning_rate": 3.0364000000000003e-06, "loss": 0.0049, "step": 34819 }, { "epoch": 1.8563736205150079, "grad_norm": 1.8932934999465942, "learning_rate": 3.0362000000000003e-06, "loss": 0.0112, "step": 34820 }, { "epoch": 1.8564269339446606, "grad_norm": 3.5271942615509033, "learning_rate": 3.0360000000000002e-06, "loss": 0.0109, "step": 34821 }, { "epoch": 1.8564802473743136, "grad_norm": 2.283446788787842, "learning_rate": 3.0358000000000002e-06, "loss": 0.01, "step": 34822 }, { "epoch": 1.8565335608039666, "grad_norm": 1.9370427131652832, "learning_rate": 3.0356e-06, "loss": 0.0103, "step": 34823 }, { "epoch": 1.8565868742336193, "grad_norm": 1.163331389427185, "learning_rate": 3.0354e-06, "loss": 0.0046, "step": 34824 }, { "epoch": 1.8566401876632723, "grad_norm": 1.1047126054763794, "learning_rate": 3.0352e-06, "loss": 0.0062, "step": 34825 }, { "epoch": 1.8566935010929253, "grad_norm": 2.5781075954437256, "learning_rate": 3.035e-06, "loss": 0.0468, "step": 34826 }, { "epoch": 1.8567468145225783, "grad_norm": 1.363236665725708, "learning_rate": 3.0348000000000006e-06, "loss": 0.0295, "step": 34827 }, { "epoch": 1.8568001279522313, "grad_norm": 1.6445883512496948, "learning_rate": 3.0346e-06, "loss": 0.0366, "step": 34828 }, { "epoch": 1.856853441381884, "grad_norm": 1.6532589197158813, "learning_rate": 3.0344e-06, "loss": 0.0345, "step": 34829 }, { "epoch": 1.856906754811537, "grad_norm": 2.0653038024902344, "learning_rate": 3.0342000000000005e-06, "loss": 0.0315, "step": 34830 }, { "epoch": 1.85696006824119, "grad_norm": 1.672360897064209, "learning_rate": 3.034e-06, "loss": 0.0319, "step": 34831 }, { "epoch": 1.8570133816708427, "grad_norm": 1.9390181303024292, "learning_rate": 3.0338e-06, "loss": 0.0269, "step": 34832 }, { "epoch": 1.857066695100496, "grad_norm": 0.2858848571777344, "learning_rate": 3.0336000000000005e-06, "loss": 0.0022, "step": 34833 }, { "epoch": 1.8571200085301487, "grad_norm": 2.502857208251953, "learning_rate": 3.0334e-06, "loss": 0.0217, "step": 34834 }, { "epoch": 1.8571733219598017, "grad_norm": 2.1724801063537598, "learning_rate": 3.0332e-06, "loss": 0.0305, "step": 34835 }, { "epoch": 1.8572266353894546, "grad_norm": 1.6412525177001953, "learning_rate": 3.0330000000000004e-06, "loss": 0.0285, "step": 34836 }, { "epoch": 1.8572799488191074, "grad_norm": 1.1161478757858276, "learning_rate": 3.0328e-06, "loss": 0.0064, "step": 34837 }, { "epoch": 1.8573332622487606, "grad_norm": 0.3019779324531555, "learning_rate": 3.0326e-06, "loss": 0.0028, "step": 34838 }, { "epoch": 1.8573865756784134, "grad_norm": 4.818792819976807, "learning_rate": 3.0324000000000004e-06, "loss": 0.0339, "step": 34839 }, { "epoch": 1.8574398891080663, "grad_norm": 2.9778952598571777, "learning_rate": 3.0322e-06, "loss": 0.0214, "step": 34840 }, { "epoch": 1.8574932025377193, "grad_norm": 1.2994035482406616, "learning_rate": 3.0320000000000004e-06, "loss": 0.0084, "step": 34841 }, { "epoch": 1.857546515967372, "grad_norm": 3.1180577278137207, "learning_rate": 3.0318000000000003e-06, "loss": 0.0217, "step": 34842 }, { "epoch": 1.857599829397025, "grad_norm": 1.8825165033340454, "learning_rate": 3.0316e-06, "loss": 0.0198, "step": 34843 }, { "epoch": 1.857653142826678, "grad_norm": 2.0916216373443604, "learning_rate": 3.0314000000000003e-06, "loss": 0.0293, "step": 34844 }, { "epoch": 1.857706456256331, "grad_norm": 4.319029331207275, "learning_rate": 3.0312000000000003e-06, "loss": 0.0385, "step": 34845 }, { "epoch": 1.857759769685984, "grad_norm": 1.1546045541763306, "learning_rate": 3.031e-06, "loss": 0.0077, "step": 34846 }, { "epoch": 1.8578130831156368, "grad_norm": 0.9576960802078247, "learning_rate": 3.0308000000000003e-06, "loss": 0.0075, "step": 34847 }, { "epoch": 1.8578663965452897, "grad_norm": 2.2170798778533936, "learning_rate": 3.0306000000000003e-06, "loss": 0.0301, "step": 34848 }, { "epoch": 1.8579197099749427, "grad_norm": 0.9506741762161255, "learning_rate": 3.0304e-06, "loss": 0.0056, "step": 34849 }, { "epoch": 1.8579730234045955, "grad_norm": 1.6213384866714478, "learning_rate": 3.0302000000000002e-06, "loss": 0.0264, "step": 34850 }, { "epoch": 1.8580263368342487, "grad_norm": 0.7665724754333496, "learning_rate": 3.0300000000000002e-06, "loss": 0.0051, "step": 34851 }, { "epoch": 1.8580796502639014, "grad_norm": 0.8476464152336121, "learning_rate": 3.0298000000000006e-06, "loss": 0.0144, "step": 34852 }, { "epoch": 1.8581329636935544, "grad_norm": 1.1329396963119507, "learning_rate": 3.0296e-06, "loss": 0.0279, "step": 34853 }, { "epoch": 1.8581862771232074, "grad_norm": 2.6054632663726807, "learning_rate": 3.0294e-06, "loss": 0.016, "step": 34854 }, { "epoch": 1.8582395905528601, "grad_norm": 0.5146498084068298, "learning_rate": 3.0292000000000006e-06, "loss": 0.0064, "step": 34855 }, { "epoch": 1.8582929039825133, "grad_norm": 1.2297013998031616, "learning_rate": 3.029e-06, "loss": 0.0299, "step": 34856 }, { "epoch": 1.858346217412166, "grad_norm": 2.1713197231292725, "learning_rate": 3.0288e-06, "loss": 0.0266, "step": 34857 }, { "epoch": 1.858399530841819, "grad_norm": 2.1763947010040283, "learning_rate": 3.0286000000000005e-06, "loss": 0.0207, "step": 34858 }, { "epoch": 1.858452844271472, "grad_norm": 1.6781851053237915, "learning_rate": 3.0284e-06, "loss": 0.0231, "step": 34859 }, { "epoch": 1.8585061577011248, "grad_norm": 1.1667441129684448, "learning_rate": 3.0282e-06, "loss": 0.0146, "step": 34860 }, { "epoch": 1.8585594711307778, "grad_norm": 2.6551177501678467, "learning_rate": 3.0280000000000005e-06, "loss": 0.0219, "step": 34861 }, { "epoch": 1.8586127845604308, "grad_norm": 1.615740180015564, "learning_rate": 3.0278e-06, "loss": 0.0347, "step": 34862 }, { "epoch": 1.8586660979900838, "grad_norm": 1.6593782901763916, "learning_rate": 3.0276e-06, "loss": 0.0191, "step": 34863 }, { "epoch": 1.8587194114197367, "grad_norm": 1.7728575468063354, "learning_rate": 3.0274000000000004e-06, "loss": 0.0221, "step": 34864 }, { "epoch": 1.8587727248493895, "grad_norm": 2.2156646251678467, "learning_rate": 3.0272e-06, "loss": 0.0405, "step": 34865 }, { "epoch": 1.8588260382790425, "grad_norm": 1.062086820602417, "learning_rate": 3.0270000000000004e-06, "loss": 0.0066, "step": 34866 }, { "epoch": 1.8588793517086954, "grad_norm": 1.599753975868225, "learning_rate": 3.0268000000000004e-06, "loss": 0.0183, "step": 34867 }, { "epoch": 1.8589326651383482, "grad_norm": 0.4995630383491516, "learning_rate": 3.0266e-06, "loss": 0.0037, "step": 34868 }, { "epoch": 1.8589859785680014, "grad_norm": 1.4791382551193237, "learning_rate": 3.0264000000000004e-06, "loss": 0.0243, "step": 34869 }, { "epoch": 1.8590392919976542, "grad_norm": 2.378002882003784, "learning_rate": 3.0262000000000004e-06, "loss": 0.0371, "step": 34870 }, { "epoch": 1.8590926054273071, "grad_norm": 1.4133951663970947, "learning_rate": 3.026e-06, "loss": 0.0141, "step": 34871 }, { "epoch": 1.8591459188569601, "grad_norm": 1.50496506690979, "learning_rate": 3.0258000000000003e-06, "loss": 0.0113, "step": 34872 }, { "epoch": 1.8591992322866129, "grad_norm": 1.2220808267593384, "learning_rate": 3.0256000000000003e-06, "loss": 0.0077, "step": 34873 }, { "epoch": 1.859252545716266, "grad_norm": 2.2788116931915283, "learning_rate": 3.0254e-06, "loss": 0.0261, "step": 34874 }, { "epoch": 1.8593058591459188, "grad_norm": 1.0981676578521729, "learning_rate": 3.0252000000000003e-06, "loss": 0.0146, "step": 34875 }, { "epoch": 1.8593591725755718, "grad_norm": 0.6061828136444092, "learning_rate": 3.0250000000000003e-06, "loss": 0.0055, "step": 34876 }, { "epoch": 1.8594124860052248, "grad_norm": 1.2321159839630127, "learning_rate": 3.0248000000000007e-06, "loss": 0.011, "step": 34877 }, { "epoch": 1.8594657994348776, "grad_norm": 1.0711476802825928, "learning_rate": 3.0246000000000002e-06, "loss": 0.0191, "step": 34878 }, { "epoch": 1.8595191128645305, "grad_norm": 1.3343679904937744, "learning_rate": 3.0244000000000002e-06, "loss": 0.0117, "step": 34879 }, { "epoch": 1.8595724262941835, "grad_norm": 0.44160693883895874, "learning_rate": 3.0242000000000006e-06, "loss": 0.0034, "step": 34880 }, { "epoch": 1.8596257397238363, "grad_norm": 1.7678784132003784, "learning_rate": 3.024e-06, "loss": 0.0274, "step": 34881 }, { "epoch": 1.8596790531534895, "grad_norm": 0.8651376366615295, "learning_rate": 3.0238e-06, "loss": 0.0065, "step": 34882 }, { "epoch": 1.8597323665831422, "grad_norm": 3.9403443336486816, "learning_rate": 3.0236000000000006e-06, "loss": 0.0335, "step": 34883 }, { "epoch": 1.8597856800127952, "grad_norm": 1.6314116716384888, "learning_rate": 3.0234e-06, "loss": 0.0134, "step": 34884 }, { "epoch": 1.8598389934424482, "grad_norm": 2.083085060119629, "learning_rate": 3.0232e-06, "loss": 0.0361, "step": 34885 }, { "epoch": 1.859892306872101, "grad_norm": 1.991731882095337, "learning_rate": 3.0230000000000005e-06, "loss": 0.0184, "step": 34886 }, { "epoch": 1.8599456203017541, "grad_norm": 1.0958316326141357, "learning_rate": 3.0228e-06, "loss": 0.0114, "step": 34887 }, { "epoch": 1.859998933731407, "grad_norm": 0.710371196269989, "learning_rate": 3.0226e-06, "loss": 0.0045, "step": 34888 }, { "epoch": 1.8600522471610599, "grad_norm": 0.6413308382034302, "learning_rate": 3.0224000000000005e-06, "loss": 0.0057, "step": 34889 }, { "epoch": 1.8601055605907129, "grad_norm": 0.8565671443939209, "learning_rate": 3.0222e-06, "loss": 0.0065, "step": 34890 }, { "epoch": 1.8601588740203656, "grad_norm": 2.395223617553711, "learning_rate": 3.0220000000000005e-06, "loss": 0.0203, "step": 34891 }, { "epoch": 1.8602121874500188, "grad_norm": 1.2219605445861816, "learning_rate": 3.0218000000000004e-06, "loss": 0.0048, "step": 34892 }, { "epoch": 1.8602655008796716, "grad_norm": 0.3173521161079407, "learning_rate": 3.0216e-06, "loss": 0.0034, "step": 34893 }, { "epoch": 1.8603188143093246, "grad_norm": 1.8613125085830688, "learning_rate": 3.0214000000000004e-06, "loss": 0.0141, "step": 34894 }, { "epoch": 1.8603721277389775, "grad_norm": 1.3338545560836792, "learning_rate": 3.0212000000000004e-06, "loss": 0.0102, "step": 34895 }, { "epoch": 1.8604254411686303, "grad_norm": 1.5478744506835938, "learning_rate": 3.021e-06, "loss": 0.0084, "step": 34896 }, { "epoch": 1.8604787545982833, "grad_norm": 0.7228395938873291, "learning_rate": 3.0208000000000004e-06, "loss": 0.0047, "step": 34897 }, { "epoch": 1.8605320680279362, "grad_norm": 1.3014546632766724, "learning_rate": 3.0206000000000004e-06, "loss": 0.0184, "step": 34898 }, { "epoch": 1.860585381457589, "grad_norm": 2.44956636428833, "learning_rate": 3.0204e-06, "loss": 0.0281, "step": 34899 }, { "epoch": 1.8606386948872422, "grad_norm": 2.9419350624084473, "learning_rate": 3.0202000000000003e-06, "loss": 0.0129, "step": 34900 }, { "epoch": 1.860692008316895, "grad_norm": 2.1062517166137695, "learning_rate": 3.0200000000000003e-06, "loss": 0.0316, "step": 34901 }, { "epoch": 1.860745321746548, "grad_norm": 1.4454379081726074, "learning_rate": 3.0198000000000007e-06, "loss": 0.0072, "step": 34902 }, { "epoch": 1.860798635176201, "grad_norm": 1.0585660934448242, "learning_rate": 3.0196000000000003e-06, "loss": 0.0252, "step": 34903 }, { "epoch": 1.8608519486058537, "grad_norm": 1.9537551403045654, "learning_rate": 3.0194000000000003e-06, "loss": 0.024, "step": 34904 }, { "epoch": 1.8609052620355069, "grad_norm": 1.5493345260620117, "learning_rate": 3.0192000000000007e-06, "loss": 0.0259, "step": 34905 }, { "epoch": 1.8609585754651596, "grad_norm": 1.0568538904190063, "learning_rate": 3.0190000000000002e-06, "loss": 0.0067, "step": 34906 }, { "epoch": 1.8610118888948126, "grad_norm": 2.061117649078369, "learning_rate": 3.0188000000000002e-06, "loss": 0.0315, "step": 34907 }, { "epoch": 1.8610652023244656, "grad_norm": 0.9273372292518616, "learning_rate": 3.0186000000000006e-06, "loss": 0.018, "step": 34908 }, { "epoch": 1.8611185157541184, "grad_norm": 1.9718881845474243, "learning_rate": 3.0184e-06, "loss": 0.0243, "step": 34909 }, { "epoch": 1.8611718291837716, "grad_norm": 2.613438367843628, "learning_rate": 3.0182e-06, "loss": 0.0298, "step": 34910 }, { "epoch": 1.8612251426134243, "grad_norm": 1.7334877252578735, "learning_rate": 3.0180000000000006e-06, "loss": 0.0164, "step": 34911 }, { "epoch": 1.8612784560430773, "grad_norm": 1.591562032699585, "learning_rate": 3.0178e-06, "loss": 0.0088, "step": 34912 }, { "epoch": 1.8613317694727303, "grad_norm": 1.8108104467391968, "learning_rate": 3.0176e-06, "loss": 0.0195, "step": 34913 }, { "epoch": 1.861385082902383, "grad_norm": 1.4664109945297241, "learning_rate": 3.0174000000000005e-06, "loss": 0.0098, "step": 34914 }, { "epoch": 1.861438396332036, "grad_norm": 1.9040132761001587, "learning_rate": 3.0172e-06, "loss": 0.0096, "step": 34915 }, { "epoch": 1.861491709761689, "grad_norm": 3.6771790981292725, "learning_rate": 3.0170000000000005e-06, "loss": 0.0291, "step": 34916 }, { "epoch": 1.8615450231913417, "grad_norm": 1.102766513824463, "learning_rate": 3.0168000000000005e-06, "loss": 0.0143, "step": 34917 }, { "epoch": 1.861598336620995, "grad_norm": 2.759716272354126, "learning_rate": 3.0166e-06, "loss": 0.0127, "step": 34918 }, { "epoch": 1.8616516500506477, "grad_norm": 1.4305247068405151, "learning_rate": 3.0164000000000005e-06, "loss": 0.0233, "step": 34919 }, { "epoch": 1.8617049634803007, "grad_norm": 1.5686914920806885, "learning_rate": 3.0162000000000004e-06, "loss": 0.0115, "step": 34920 }, { "epoch": 1.8617582769099537, "grad_norm": 2.5966222286224365, "learning_rate": 3.016e-06, "loss": 0.0205, "step": 34921 }, { "epoch": 1.8618115903396064, "grad_norm": 1.3928654193878174, "learning_rate": 3.0158000000000004e-06, "loss": 0.0235, "step": 34922 }, { "epoch": 1.8618649037692596, "grad_norm": 2.10313081741333, "learning_rate": 3.0156000000000004e-06, "loss": 0.0141, "step": 34923 }, { "epoch": 1.8619182171989124, "grad_norm": 1.8978632688522339, "learning_rate": 3.0154e-06, "loss": 0.0226, "step": 34924 }, { "epoch": 1.8619715306285654, "grad_norm": 2.220614433288574, "learning_rate": 3.0152000000000004e-06, "loss": 0.035, "step": 34925 }, { "epoch": 1.8620248440582183, "grad_norm": 0.6507828235626221, "learning_rate": 3.0150000000000004e-06, "loss": 0.0048, "step": 34926 }, { "epoch": 1.862078157487871, "grad_norm": 0.6527957320213318, "learning_rate": 3.0148000000000003e-06, "loss": 0.0035, "step": 34927 }, { "epoch": 1.8621314709175243, "grad_norm": 2.1194489002227783, "learning_rate": 3.0146000000000003e-06, "loss": 0.0136, "step": 34928 }, { "epoch": 1.862184784347177, "grad_norm": 1.6074663400650024, "learning_rate": 3.0144000000000003e-06, "loss": 0.0135, "step": 34929 }, { "epoch": 1.86223809777683, "grad_norm": 1.8367947340011597, "learning_rate": 3.0142000000000003e-06, "loss": 0.0098, "step": 34930 }, { "epoch": 1.862291411206483, "grad_norm": 0.6925288438796997, "learning_rate": 3.0140000000000003e-06, "loss": 0.0047, "step": 34931 }, { "epoch": 1.8623447246361358, "grad_norm": 1.4807614088058472, "learning_rate": 3.0138000000000003e-06, "loss": 0.0068, "step": 34932 }, { "epoch": 1.8623980380657887, "grad_norm": 1.6390539407730103, "learning_rate": 3.0136000000000003e-06, "loss": 0.0188, "step": 34933 }, { "epoch": 1.8624513514954417, "grad_norm": 0.9725381731987, "learning_rate": 3.0134000000000002e-06, "loss": 0.0065, "step": 34934 }, { "epoch": 1.8625046649250945, "grad_norm": 2.526063919067383, "learning_rate": 3.0132000000000002e-06, "loss": 0.0209, "step": 34935 }, { "epoch": 1.8625579783547477, "grad_norm": 0.7183970212936401, "learning_rate": 3.013e-06, "loss": 0.0043, "step": 34936 }, { "epoch": 1.8626112917844004, "grad_norm": 1.5586185455322266, "learning_rate": 3.0128e-06, "loss": 0.0171, "step": 34937 }, { "epoch": 1.8626646052140534, "grad_norm": 2.117643356323242, "learning_rate": 3.0126e-06, "loss": 0.0225, "step": 34938 }, { "epoch": 1.8627179186437064, "grad_norm": 1.1830343008041382, "learning_rate": 3.0124e-06, "loss": 0.0053, "step": 34939 }, { "epoch": 1.8627712320733592, "grad_norm": 3.5041964054107666, "learning_rate": 3.0122e-06, "loss": 0.0175, "step": 34940 }, { "epoch": 1.8628245455030124, "grad_norm": 0.8377828001976013, "learning_rate": 3.0120000000000006e-06, "loss": 0.0095, "step": 34941 }, { "epoch": 1.862877858932665, "grad_norm": 1.332563042640686, "learning_rate": 3.0118e-06, "loss": 0.0146, "step": 34942 }, { "epoch": 1.862931172362318, "grad_norm": 1.9311761856079102, "learning_rate": 3.0116e-06, "loss": 0.0261, "step": 34943 }, { "epoch": 1.862984485791971, "grad_norm": 2.1168277263641357, "learning_rate": 3.0114000000000005e-06, "loss": 0.0175, "step": 34944 }, { "epoch": 1.8630377992216238, "grad_norm": 2.6298305988311768, "learning_rate": 3.0112e-06, "loss": 0.0606, "step": 34945 }, { "epoch": 1.8630911126512768, "grad_norm": 0.6970399618148804, "learning_rate": 3.011e-06, "loss": 0.0033, "step": 34946 }, { "epoch": 1.8631444260809298, "grad_norm": 0.45693254470825195, "learning_rate": 3.0108000000000005e-06, "loss": 0.0035, "step": 34947 }, { "epoch": 1.8631977395105828, "grad_norm": 2.0838353633880615, "learning_rate": 3.0106e-06, "loss": 0.0271, "step": 34948 }, { "epoch": 1.8632510529402357, "grad_norm": 1.2114155292510986, "learning_rate": 3.0104e-06, "loss": 0.0262, "step": 34949 }, { "epoch": 1.8633043663698885, "grad_norm": 2.7271997928619385, "learning_rate": 3.0102000000000004e-06, "loss": 0.042, "step": 34950 }, { "epoch": 1.8633576797995415, "grad_norm": 1.8478658199310303, "learning_rate": 3.01e-06, "loss": 0.0112, "step": 34951 }, { "epoch": 1.8634109932291945, "grad_norm": 3.7132654190063477, "learning_rate": 3.0098000000000004e-06, "loss": 0.0159, "step": 34952 }, { "epoch": 1.8634643066588472, "grad_norm": 2.0579111576080322, "learning_rate": 3.0096000000000004e-06, "loss": 0.011, "step": 34953 }, { "epoch": 1.8635176200885004, "grad_norm": 2.0675926208496094, "learning_rate": 3.0094e-06, "loss": 0.015, "step": 34954 }, { "epoch": 1.8635709335181532, "grad_norm": 1.9128018617630005, "learning_rate": 3.0092000000000003e-06, "loss": 0.0145, "step": 34955 }, { "epoch": 1.8636242469478062, "grad_norm": 4.437968730926514, "learning_rate": 3.0090000000000003e-06, "loss": 0.0195, "step": 34956 }, { "epoch": 1.8636775603774591, "grad_norm": 2.6062076091766357, "learning_rate": 3.0088e-06, "loss": 0.0211, "step": 34957 }, { "epoch": 1.8637308738071119, "grad_norm": 3.148637533187866, "learning_rate": 3.0086000000000003e-06, "loss": 0.0583, "step": 34958 }, { "epoch": 1.863784187236765, "grad_norm": 1.4154425859451294, "learning_rate": 3.0084000000000003e-06, "loss": 0.0111, "step": 34959 }, { "epoch": 1.8638375006664178, "grad_norm": 0.3306897282600403, "learning_rate": 3.0082e-06, "loss": 0.0032, "step": 34960 }, { "epoch": 1.8638908140960708, "grad_norm": 1.9927114248275757, "learning_rate": 3.0080000000000003e-06, "loss": 0.0291, "step": 34961 }, { "epoch": 1.8639441275257238, "grad_norm": 1.7708494663238525, "learning_rate": 3.0078000000000002e-06, "loss": 0.0112, "step": 34962 }, { "epoch": 1.8639974409553766, "grad_norm": 1.1432082653045654, "learning_rate": 3.0076000000000006e-06, "loss": 0.0075, "step": 34963 }, { "epoch": 1.8640507543850295, "grad_norm": 2.051203727722168, "learning_rate": 3.0074e-06, "loss": 0.0186, "step": 34964 }, { "epoch": 1.8641040678146825, "grad_norm": 2.9856629371643066, "learning_rate": 3.0072e-06, "loss": 0.0166, "step": 34965 }, { "epoch": 1.8641573812443355, "grad_norm": 2.448612689971924, "learning_rate": 3.0070000000000006e-06, "loss": 0.0088, "step": 34966 }, { "epoch": 1.8642106946739885, "grad_norm": 1.0310157537460327, "learning_rate": 3.0068e-06, "loss": 0.0096, "step": 34967 }, { "epoch": 1.8642640081036412, "grad_norm": 3.5578806400299072, "learning_rate": 3.0066e-06, "loss": 0.0241, "step": 34968 }, { "epoch": 1.8643173215332942, "grad_norm": 0.5580341219902039, "learning_rate": 3.0064000000000006e-06, "loss": 0.0029, "step": 34969 }, { "epoch": 1.8643706349629472, "grad_norm": 1.111628532409668, "learning_rate": 3.0062e-06, "loss": 0.012, "step": 34970 }, { "epoch": 1.8644239483926, "grad_norm": 0.18933767080307007, "learning_rate": 3.006e-06, "loss": 0.0016, "step": 34971 }, { "epoch": 1.8644772618222532, "grad_norm": 0.6756196022033691, "learning_rate": 3.0058000000000005e-06, "loss": 0.0049, "step": 34972 }, { "epoch": 1.864530575251906, "grad_norm": 1.3136652708053589, "learning_rate": 3.0056e-06, "loss": 0.0222, "step": 34973 }, { "epoch": 1.8645838886815589, "grad_norm": 0.3294104337692261, "learning_rate": 3.0054e-06, "loss": 0.0021, "step": 34974 }, { "epoch": 1.8646372021112119, "grad_norm": 1.6950030326843262, "learning_rate": 3.0052000000000005e-06, "loss": 0.0108, "step": 34975 }, { "epoch": 1.8646905155408646, "grad_norm": 1.6989970207214355, "learning_rate": 3.005e-06, "loss": 0.0175, "step": 34976 }, { "epoch": 1.8647438289705178, "grad_norm": 0.6941924095153809, "learning_rate": 3.0048000000000004e-06, "loss": 0.0036, "step": 34977 }, { "epoch": 1.8647971424001706, "grad_norm": 0.37236881256103516, "learning_rate": 3.0046000000000004e-06, "loss": 0.0029, "step": 34978 }, { "epoch": 1.8648504558298236, "grad_norm": 1.7743395566940308, "learning_rate": 3.0044e-06, "loss": 0.0249, "step": 34979 }, { "epoch": 1.8649037692594765, "grad_norm": 1.2346878051757812, "learning_rate": 3.0042000000000004e-06, "loss": 0.0193, "step": 34980 }, { "epoch": 1.8649570826891293, "grad_norm": 0.0874391570687294, "learning_rate": 3.0040000000000004e-06, "loss": 0.0009, "step": 34981 }, { "epoch": 1.8650103961187823, "grad_norm": 1.6469035148620605, "learning_rate": 3.0038e-06, "loss": 0.0285, "step": 34982 }, { "epoch": 1.8650637095484353, "grad_norm": 2.589317798614502, "learning_rate": 3.0036000000000003e-06, "loss": 0.0376, "step": 34983 }, { "epoch": 1.8651170229780882, "grad_norm": 1.634852409362793, "learning_rate": 3.0034000000000003e-06, "loss": 0.0074, "step": 34984 }, { "epoch": 1.8651703364077412, "grad_norm": 1.453137993812561, "learning_rate": 3.0032e-06, "loss": 0.0122, "step": 34985 }, { "epoch": 1.865223649837394, "grad_norm": 1.4675554037094116, "learning_rate": 3.0030000000000003e-06, "loss": 0.0192, "step": 34986 }, { "epoch": 1.865276963267047, "grad_norm": 2.114671468734741, "learning_rate": 3.0028000000000003e-06, "loss": 0.008, "step": 34987 }, { "epoch": 1.8653302766967, "grad_norm": 2.672571897506714, "learning_rate": 3.0026000000000007e-06, "loss": 0.0228, "step": 34988 }, { "epoch": 1.8653835901263527, "grad_norm": 4.725177764892578, "learning_rate": 3.0024000000000003e-06, "loss": 0.0173, "step": 34989 }, { "epoch": 1.8654369035560059, "grad_norm": 1.409127950668335, "learning_rate": 3.0022000000000002e-06, "loss": 0.0259, "step": 34990 }, { "epoch": 1.8654902169856586, "grad_norm": 2.280860424041748, "learning_rate": 3.0020000000000006e-06, "loss": 0.0219, "step": 34991 }, { "epoch": 1.8655435304153116, "grad_norm": 3.103360891342163, "learning_rate": 3.0018e-06, "loss": 0.0146, "step": 34992 }, { "epoch": 1.8655968438449646, "grad_norm": 2.318021535873413, "learning_rate": 3.0016e-06, "loss": 0.0216, "step": 34993 }, { "epoch": 1.8656501572746174, "grad_norm": 2.1950690746307373, "learning_rate": 3.0014000000000006e-06, "loss": 0.0157, "step": 34994 }, { "epoch": 1.8657034707042706, "grad_norm": 0.44661280512809753, "learning_rate": 3.0012e-06, "loss": 0.0043, "step": 34995 }, { "epoch": 1.8657567841339233, "grad_norm": 2.8802318572998047, "learning_rate": 3.001e-06, "loss": 0.0321, "step": 34996 }, { "epoch": 1.8658100975635763, "grad_norm": 1.4405913352966309, "learning_rate": 3.0008000000000006e-06, "loss": 0.0075, "step": 34997 }, { "epoch": 1.8658634109932293, "grad_norm": 4.124904155731201, "learning_rate": 3.0006e-06, "loss": 0.0407, "step": 34998 }, { "epoch": 1.865916724422882, "grad_norm": 0.8995631337165833, "learning_rate": 3.0004e-06, "loss": 0.0046, "step": 34999 }, { "epoch": 1.865970037852535, "grad_norm": 1.8864880800247192, "learning_rate": 3.0002000000000005e-06, "loss": 0.018, "step": 35000 }, { "epoch": 1.866023351282188, "grad_norm": 3.662287712097168, "learning_rate": 3e-06, "loss": 0.0705, "step": 35001 }, { "epoch": 1.8660766647118407, "grad_norm": 2.2800018787384033, "learning_rate": 2.9998000000000005e-06, "loss": 0.0243, "step": 35002 }, { "epoch": 1.866129978141494, "grad_norm": 1.607728362083435, "learning_rate": 2.9996000000000005e-06, "loss": 0.0164, "step": 35003 }, { "epoch": 1.8661832915711467, "grad_norm": 0.9908381104469299, "learning_rate": 2.9994e-06, "loss": 0.0048, "step": 35004 }, { "epoch": 1.8662366050007997, "grad_norm": 2.2572779655456543, "learning_rate": 2.9992000000000004e-06, "loss": 0.0241, "step": 35005 }, { "epoch": 1.8662899184304527, "grad_norm": 1.7094323635101318, "learning_rate": 2.9990000000000004e-06, "loss": 0.0286, "step": 35006 }, { "epoch": 1.8663432318601054, "grad_norm": 2.603564500808716, "learning_rate": 2.9988e-06, "loss": 0.0405, "step": 35007 }, { "epoch": 1.8663965452897586, "grad_norm": 1.0059436559677124, "learning_rate": 2.9986000000000004e-06, "loss": 0.0181, "step": 35008 }, { "epoch": 1.8664498587194114, "grad_norm": 1.7030773162841797, "learning_rate": 2.9984000000000004e-06, "loss": 0.0218, "step": 35009 }, { "epoch": 1.8665031721490644, "grad_norm": 1.1445896625518799, "learning_rate": 2.9982e-06, "loss": 0.0207, "step": 35010 }, { "epoch": 1.8665564855787173, "grad_norm": 1.7634128332138062, "learning_rate": 2.9980000000000003e-06, "loss": 0.0155, "step": 35011 }, { "epoch": 1.86660979900837, "grad_norm": 1.53548002243042, "learning_rate": 2.9978000000000003e-06, "loss": 0.0077, "step": 35012 }, { "epoch": 1.8666631124380233, "grad_norm": 2.0436227321624756, "learning_rate": 2.9976000000000003e-06, "loss": 0.0234, "step": 35013 }, { "epoch": 1.866716425867676, "grad_norm": 1.7996859550476074, "learning_rate": 2.9974000000000003e-06, "loss": 0.0231, "step": 35014 }, { "epoch": 1.866769739297329, "grad_norm": 1.9355154037475586, "learning_rate": 2.9972000000000003e-06, "loss": 0.0122, "step": 35015 }, { "epoch": 1.866823052726982, "grad_norm": 1.3547226190567017, "learning_rate": 2.9970000000000003e-06, "loss": 0.0171, "step": 35016 }, { "epoch": 1.8668763661566348, "grad_norm": 2.1915416717529297, "learning_rate": 2.9968000000000003e-06, "loss": 0.0149, "step": 35017 }, { "epoch": 1.8669296795862877, "grad_norm": 2.5800588130950928, "learning_rate": 2.9966000000000002e-06, "loss": 0.0161, "step": 35018 }, { "epoch": 1.8669829930159407, "grad_norm": 0.8737025260925293, "learning_rate": 2.9964000000000002e-06, "loss": 0.0078, "step": 35019 }, { "epoch": 1.8670363064455935, "grad_norm": 2.204599380493164, "learning_rate": 2.9962000000000002e-06, "loss": 0.013, "step": 35020 }, { "epoch": 1.8670896198752467, "grad_norm": 0.7216417789459229, "learning_rate": 2.996e-06, "loss": 0.0057, "step": 35021 }, { "epoch": 1.8671429333048994, "grad_norm": 2.208012819290161, "learning_rate": 2.9958e-06, "loss": 0.0351, "step": 35022 }, { "epoch": 1.8671962467345524, "grad_norm": 1.6957043409347534, "learning_rate": 2.9956e-06, "loss": 0.0222, "step": 35023 }, { "epoch": 1.8672495601642054, "grad_norm": 2.939842462539673, "learning_rate": 2.9954e-06, "loss": 0.0382, "step": 35024 }, { "epoch": 1.8673028735938582, "grad_norm": 1.735733985900879, "learning_rate": 2.9952e-06, "loss": 0.0233, "step": 35025 }, { "epoch": 1.8673561870235114, "grad_norm": 2.781297206878662, "learning_rate": 2.995e-06, "loss": 0.0324, "step": 35026 }, { "epoch": 1.8674095004531641, "grad_norm": 3.027306318283081, "learning_rate": 2.9948000000000005e-06, "loss": 0.0363, "step": 35027 }, { "epoch": 1.867462813882817, "grad_norm": 1.226285696029663, "learning_rate": 2.9946e-06, "loss": 0.0063, "step": 35028 }, { "epoch": 1.86751612731247, "grad_norm": 1.4689033031463623, "learning_rate": 2.9944e-06, "loss": 0.0104, "step": 35029 }, { "epoch": 1.8675694407421228, "grad_norm": 0.950344443321228, "learning_rate": 2.9942000000000005e-06, "loss": 0.0091, "step": 35030 }, { "epoch": 1.867622754171776, "grad_norm": 0.21321552991867065, "learning_rate": 2.994e-06, "loss": 0.0021, "step": 35031 }, { "epoch": 1.8676760676014288, "grad_norm": 2.0292229652404785, "learning_rate": 2.9938e-06, "loss": 0.0236, "step": 35032 }, { "epoch": 1.8677293810310818, "grad_norm": 1.5243866443634033, "learning_rate": 2.9936000000000004e-06, "loss": 0.0099, "step": 35033 }, { "epoch": 1.8677826944607347, "grad_norm": 1.552381157875061, "learning_rate": 2.9934e-06, "loss": 0.0099, "step": 35034 }, { "epoch": 1.8678360078903875, "grad_norm": 0.5526003837585449, "learning_rate": 2.9932e-06, "loss": 0.0039, "step": 35035 }, { "epoch": 1.8678893213200405, "grad_norm": 0.7029317021369934, "learning_rate": 2.9930000000000004e-06, "loss": 0.0046, "step": 35036 }, { "epoch": 1.8679426347496935, "grad_norm": 0.4549051821231842, "learning_rate": 2.9928e-06, "loss": 0.0034, "step": 35037 }, { "epoch": 1.8679959481793462, "grad_norm": 0.9136626124382019, "learning_rate": 2.9926000000000004e-06, "loss": 0.0075, "step": 35038 }, { "epoch": 1.8680492616089994, "grad_norm": 2.799628734588623, "learning_rate": 2.9924000000000004e-06, "loss": 0.03, "step": 35039 }, { "epoch": 1.8681025750386522, "grad_norm": 1.4953373670578003, "learning_rate": 2.9922e-06, "loss": 0.0112, "step": 35040 }, { "epoch": 1.8681558884683052, "grad_norm": 1.7518508434295654, "learning_rate": 2.9920000000000003e-06, "loss": 0.0104, "step": 35041 }, { "epoch": 1.8682092018979581, "grad_norm": 1.639394998550415, "learning_rate": 2.9918000000000003e-06, "loss": 0.0144, "step": 35042 }, { "epoch": 1.868262515327611, "grad_norm": 2.3542816638946533, "learning_rate": 2.9916e-06, "loss": 0.0268, "step": 35043 }, { "epoch": 1.868315828757264, "grad_norm": 0.3410474956035614, "learning_rate": 2.9914000000000003e-06, "loss": 0.0029, "step": 35044 }, { "epoch": 1.8683691421869169, "grad_norm": 1.7776654958724976, "learning_rate": 2.9912000000000003e-06, "loss": 0.023, "step": 35045 }, { "epoch": 1.8684224556165698, "grad_norm": 0.5764462947845459, "learning_rate": 2.991e-06, "loss": 0.0033, "step": 35046 }, { "epoch": 1.8684757690462228, "grad_norm": 1.0376302003860474, "learning_rate": 2.9908000000000002e-06, "loss": 0.0157, "step": 35047 }, { "epoch": 1.8685290824758756, "grad_norm": 0.46400776505470276, "learning_rate": 2.9906000000000002e-06, "loss": 0.0027, "step": 35048 }, { "epoch": 1.8685823959055285, "grad_norm": 3.076699733734131, "learning_rate": 2.9903999999999998e-06, "loss": 0.0373, "step": 35049 }, { "epoch": 1.8686357093351815, "grad_norm": 1.350696325302124, "learning_rate": 2.9902e-06, "loss": 0.0077, "step": 35050 }, { "epoch": 1.8686890227648345, "grad_norm": 1.9142338037490845, "learning_rate": 2.99e-06, "loss": 0.0415, "step": 35051 }, { "epoch": 1.8687423361944875, "grad_norm": 1.179461121559143, "learning_rate": 2.9898000000000006e-06, "loss": 0.0251, "step": 35052 }, { "epoch": 1.8687956496241402, "grad_norm": 1.962743878364563, "learning_rate": 2.9896e-06, "loss": 0.0222, "step": 35053 }, { "epoch": 1.8688489630537932, "grad_norm": 1.1413791179656982, "learning_rate": 2.9894e-06, "loss": 0.0064, "step": 35054 }, { "epoch": 1.8689022764834462, "grad_norm": 1.8745485544204712, "learning_rate": 2.9892000000000005e-06, "loss": 0.0124, "step": 35055 }, { "epoch": 1.868955589913099, "grad_norm": 1.79294753074646, "learning_rate": 2.989e-06, "loss": 0.01, "step": 35056 }, { "epoch": 1.8690089033427522, "grad_norm": 2.74698805809021, "learning_rate": 2.9888e-06, "loss": 0.0145, "step": 35057 }, { "epoch": 1.869062216772405, "grad_norm": 3.2385332584381104, "learning_rate": 2.9886000000000005e-06, "loss": 0.0339, "step": 35058 }, { "epoch": 1.869115530202058, "grad_norm": 2.308490037918091, "learning_rate": 2.9884e-06, "loss": 0.0124, "step": 35059 }, { "epoch": 1.8691688436317109, "grad_norm": 2.682494878768921, "learning_rate": 2.9882e-06, "loss": 0.018, "step": 35060 }, { "epoch": 1.8692221570613636, "grad_norm": 1.3173925876617432, "learning_rate": 2.9880000000000004e-06, "loss": 0.0076, "step": 35061 }, { "epoch": 1.8692754704910168, "grad_norm": 1.620560884475708, "learning_rate": 2.9878e-06, "loss": 0.0103, "step": 35062 }, { "epoch": 1.8693287839206696, "grad_norm": 2.6143009662628174, "learning_rate": 2.9876000000000004e-06, "loss": 0.0132, "step": 35063 }, { "epoch": 1.8693820973503226, "grad_norm": 1.059017300605774, "learning_rate": 2.9874000000000004e-06, "loss": 0.0148, "step": 35064 }, { "epoch": 1.8694354107799755, "grad_norm": 1.8733986616134644, "learning_rate": 2.9872e-06, "loss": 0.0103, "step": 35065 }, { "epoch": 1.8694887242096283, "grad_norm": 2.422917366027832, "learning_rate": 2.9870000000000004e-06, "loss": 0.0124, "step": 35066 }, { "epoch": 1.8695420376392813, "grad_norm": 0.9937905073165894, "learning_rate": 2.9868000000000004e-06, "loss": 0.0056, "step": 35067 }, { "epoch": 1.8695953510689343, "grad_norm": 1.1406866312026978, "learning_rate": 2.9866e-06, "loss": 0.025, "step": 35068 }, { "epoch": 1.8696486644985872, "grad_norm": 2.564199209213257, "learning_rate": 2.9864000000000003e-06, "loss": 0.0332, "step": 35069 }, { "epoch": 1.8697019779282402, "grad_norm": 1.582757592201233, "learning_rate": 2.9862000000000003e-06, "loss": 0.0093, "step": 35070 }, { "epoch": 1.869755291357893, "grad_norm": 2.2455787658691406, "learning_rate": 2.986e-06, "loss": 0.0463, "step": 35071 }, { "epoch": 1.869808604787546, "grad_norm": 3.4759998321533203, "learning_rate": 2.9858000000000003e-06, "loss": 0.0366, "step": 35072 }, { "epoch": 1.869861918217199, "grad_norm": 2.4063594341278076, "learning_rate": 2.9856000000000003e-06, "loss": 0.0489, "step": 35073 }, { "epoch": 1.8699152316468517, "grad_norm": 1.661819338798523, "learning_rate": 2.9854000000000007e-06, "loss": 0.0146, "step": 35074 }, { "epoch": 1.869968545076505, "grad_norm": 0.5454072952270508, "learning_rate": 2.9852000000000002e-06, "loss": 0.0032, "step": 35075 }, { "epoch": 1.8700218585061577, "grad_norm": 1.4562697410583496, "learning_rate": 2.9850000000000002e-06, "loss": 0.0054, "step": 35076 }, { "epoch": 1.8700751719358106, "grad_norm": 4.202325820922852, "learning_rate": 2.9848000000000006e-06, "loss": 0.0221, "step": 35077 }, { "epoch": 1.8701284853654636, "grad_norm": 3.236664056777954, "learning_rate": 2.9846e-06, "loss": 0.0225, "step": 35078 }, { "epoch": 1.8701817987951164, "grad_norm": 1.7129995822906494, "learning_rate": 2.9844e-06, "loss": 0.0204, "step": 35079 }, { "epoch": 1.8702351122247696, "grad_norm": 3.046963930130005, "learning_rate": 2.9842000000000006e-06, "loss": 0.0166, "step": 35080 }, { "epoch": 1.8702884256544223, "grad_norm": 3.028064012527466, "learning_rate": 2.984e-06, "loss": 0.0374, "step": 35081 }, { "epoch": 1.8703417390840753, "grad_norm": 1.5792028903961182, "learning_rate": 2.9838e-06, "loss": 0.011, "step": 35082 }, { "epoch": 1.8703950525137283, "grad_norm": 1.236289143562317, "learning_rate": 2.9836000000000005e-06, "loss": 0.0061, "step": 35083 }, { "epoch": 1.870448365943381, "grad_norm": 1.3496711254119873, "learning_rate": 2.9834e-06, "loss": 0.0253, "step": 35084 }, { "epoch": 1.870501679373034, "grad_norm": 0.5514702796936035, "learning_rate": 2.9832e-06, "loss": 0.0043, "step": 35085 }, { "epoch": 1.870554992802687, "grad_norm": 1.129787802696228, "learning_rate": 2.9830000000000005e-06, "loss": 0.0174, "step": 35086 }, { "epoch": 1.87060830623234, "grad_norm": 0.46016576886177063, "learning_rate": 2.9828e-06, "loss": 0.0031, "step": 35087 }, { "epoch": 1.870661619661993, "grad_norm": 2.7523419857025146, "learning_rate": 2.9826000000000005e-06, "loss": 0.0384, "step": 35088 }, { "epoch": 1.8707149330916457, "grad_norm": 1.6947633028030396, "learning_rate": 2.9824000000000004e-06, "loss": 0.0091, "step": 35089 }, { "epoch": 1.8707682465212987, "grad_norm": 0.8616848587989807, "learning_rate": 2.9822e-06, "loss": 0.0052, "step": 35090 }, { "epoch": 1.8708215599509517, "grad_norm": 2.775441884994507, "learning_rate": 2.9820000000000004e-06, "loss": 0.0258, "step": 35091 }, { "epoch": 1.8708748733806044, "grad_norm": 2.0946688652038574, "learning_rate": 2.9818000000000004e-06, "loss": 0.0213, "step": 35092 }, { "epoch": 1.8709281868102576, "grad_norm": 2.687204599380493, "learning_rate": 2.9816e-06, "loss": 0.0104, "step": 35093 }, { "epoch": 1.8709815002399104, "grad_norm": 2.6218008995056152, "learning_rate": 2.9814000000000004e-06, "loss": 0.012, "step": 35094 }, { "epoch": 1.8710348136695634, "grad_norm": 2.770207405090332, "learning_rate": 2.9812000000000004e-06, "loss": 0.0269, "step": 35095 }, { "epoch": 1.8710881270992163, "grad_norm": 0.45791590213775635, "learning_rate": 2.981e-06, "loss": 0.0022, "step": 35096 }, { "epoch": 1.871141440528869, "grad_norm": 2.76102352142334, "learning_rate": 2.9808000000000003e-06, "loss": 0.0109, "step": 35097 }, { "epoch": 1.8711947539585223, "grad_norm": 3.213718891143799, "learning_rate": 2.9806000000000003e-06, "loss": 0.0154, "step": 35098 }, { "epoch": 1.871248067388175, "grad_norm": 2.4208786487579346, "learning_rate": 2.9804000000000007e-06, "loss": 0.01, "step": 35099 }, { "epoch": 1.871301380817828, "grad_norm": 1.550227403640747, "learning_rate": 2.9802000000000003e-06, "loss": 0.0067, "step": 35100 }, { "epoch": 1.871354694247481, "grad_norm": 2.5938880443573, "learning_rate": 2.9800000000000003e-06, "loss": 0.0429, "step": 35101 }, { "epoch": 1.8714080076771338, "grad_norm": 2.1334481239318848, "learning_rate": 2.9798000000000007e-06, "loss": 0.0274, "step": 35102 }, { "epoch": 1.8714613211067868, "grad_norm": 1.8260730504989624, "learning_rate": 2.9796000000000002e-06, "loss": 0.0095, "step": 35103 }, { "epoch": 1.8715146345364397, "grad_norm": 0.9501602649688721, "learning_rate": 2.9794000000000002e-06, "loss": 0.0054, "step": 35104 }, { "epoch": 1.8715679479660925, "grad_norm": 1.8467698097229004, "learning_rate": 2.9792000000000006e-06, "loss": 0.0145, "step": 35105 }, { "epoch": 1.8716212613957457, "grad_norm": 1.346724510192871, "learning_rate": 2.979e-06, "loss": 0.0084, "step": 35106 }, { "epoch": 1.8716745748253985, "grad_norm": 2.292813301086426, "learning_rate": 2.9788e-06, "loss": 0.0134, "step": 35107 }, { "epoch": 1.8717278882550514, "grad_norm": 1.947372317314148, "learning_rate": 2.9786000000000006e-06, "loss": 0.0226, "step": 35108 }, { "epoch": 1.8717812016847044, "grad_norm": 2.553483009338379, "learning_rate": 2.9784e-06, "loss": 0.019, "step": 35109 }, { "epoch": 1.8718345151143572, "grad_norm": 1.7616814374923706, "learning_rate": 2.9782e-06, "loss": 0.0064, "step": 35110 }, { "epoch": 1.8718878285440104, "grad_norm": 1.4221850633621216, "learning_rate": 2.9780000000000005e-06, "loss": 0.0218, "step": 35111 }, { "epoch": 1.8719411419736631, "grad_norm": 3.0049946308135986, "learning_rate": 2.9778e-06, "loss": 0.0192, "step": 35112 }, { "epoch": 1.871994455403316, "grad_norm": 4.640854835510254, "learning_rate": 2.9776000000000005e-06, "loss": 0.0359, "step": 35113 }, { "epoch": 1.872047768832969, "grad_norm": 1.4154850244522095, "learning_rate": 2.9774000000000005e-06, "loss": 0.0073, "step": 35114 }, { "epoch": 1.8721010822626218, "grad_norm": 1.4328172206878662, "learning_rate": 2.9772e-06, "loss": 0.0243, "step": 35115 }, { "epoch": 1.872154395692275, "grad_norm": 0.6585014462471008, "learning_rate": 2.9770000000000005e-06, "loss": 0.0035, "step": 35116 }, { "epoch": 1.8722077091219278, "grad_norm": 2.005470037460327, "learning_rate": 2.9768000000000005e-06, "loss": 0.0239, "step": 35117 }, { "epoch": 1.8722610225515808, "grad_norm": 1.0986441373825073, "learning_rate": 2.9766e-06, "loss": 0.0176, "step": 35118 }, { "epoch": 1.8723143359812338, "grad_norm": 1.756649374961853, "learning_rate": 2.9764000000000004e-06, "loss": 0.0347, "step": 35119 }, { "epoch": 1.8723676494108865, "grad_norm": 0.384181410074234, "learning_rate": 2.9762000000000004e-06, "loss": 0.0032, "step": 35120 }, { "epoch": 1.8724209628405395, "grad_norm": 1.8403486013412476, "learning_rate": 2.976e-06, "loss": 0.0427, "step": 35121 }, { "epoch": 1.8724742762701925, "grad_norm": 2.0243282318115234, "learning_rate": 2.9758000000000004e-06, "loss": 0.025, "step": 35122 }, { "epoch": 1.8725275896998452, "grad_norm": 4.909881114959717, "learning_rate": 2.9756000000000004e-06, "loss": 0.0236, "step": 35123 }, { "epoch": 1.8725809031294984, "grad_norm": 1.9406704902648926, "learning_rate": 2.9754000000000003e-06, "loss": 0.0107, "step": 35124 }, { "epoch": 1.8726342165591512, "grad_norm": 3.1654796600341797, "learning_rate": 2.9752000000000003e-06, "loss": 0.0146, "step": 35125 }, { "epoch": 1.8726875299888042, "grad_norm": 0.4613792300224304, "learning_rate": 2.9750000000000003e-06, "loss": 0.0029, "step": 35126 }, { "epoch": 1.8727408434184571, "grad_norm": 3.848201036453247, "learning_rate": 2.9748000000000003e-06, "loss": 0.0162, "step": 35127 }, { "epoch": 1.87279415684811, "grad_norm": 2.3137505054473877, "learning_rate": 2.9746000000000003e-06, "loss": 0.024, "step": 35128 }, { "epoch": 1.872847470277763, "grad_norm": 1.911783218383789, "learning_rate": 2.9744000000000003e-06, "loss": 0.0076, "step": 35129 }, { "epoch": 1.8729007837074159, "grad_norm": 3.847670078277588, "learning_rate": 2.9742000000000003e-06, "loss": 0.015, "step": 35130 }, { "epoch": 1.8729540971370688, "grad_norm": 2.1341073513031006, "learning_rate": 2.9740000000000002e-06, "loss": 0.0136, "step": 35131 }, { "epoch": 1.8730074105667218, "grad_norm": 0.6919502019882202, "learning_rate": 2.9738000000000002e-06, "loss": 0.0074, "step": 35132 }, { "epoch": 1.8730607239963746, "grad_norm": 1.7227275371551514, "learning_rate": 2.9736e-06, "loss": 0.0197, "step": 35133 }, { "epoch": 1.8731140374260278, "grad_norm": 1.6599611043930054, "learning_rate": 2.9734e-06, "loss": 0.0215, "step": 35134 }, { "epoch": 1.8731673508556805, "grad_norm": 1.851453185081482, "learning_rate": 2.9732e-06, "loss": 0.0355, "step": 35135 }, { "epoch": 1.8732206642853335, "grad_norm": 0.8190633654594421, "learning_rate": 2.973e-06, "loss": 0.0053, "step": 35136 }, { "epoch": 1.8732739777149865, "grad_norm": 1.2248122692108154, "learning_rate": 2.9728e-06, "loss": 0.0161, "step": 35137 }, { "epoch": 1.8733272911446393, "grad_norm": 0.2661401927471161, "learning_rate": 2.9726000000000006e-06, "loss": 0.0024, "step": 35138 }, { "epoch": 1.8733806045742922, "grad_norm": 1.4753103256225586, "learning_rate": 2.9724e-06, "loss": 0.0195, "step": 35139 }, { "epoch": 1.8734339180039452, "grad_norm": 1.2299374341964722, "learning_rate": 2.9722e-06, "loss": 0.0074, "step": 35140 }, { "epoch": 1.873487231433598, "grad_norm": 2.7421159744262695, "learning_rate": 2.9720000000000005e-06, "loss": 0.0424, "step": 35141 }, { "epoch": 1.8735405448632512, "grad_norm": 1.1677647829055786, "learning_rate": 2.9718e-06, "loss": 0.0124, "step": 35142 }, { "epoch": 1.873593858292904, "grad_norm": 2.6489503383636475, "learning_rate": 2.9716e-06, "loss": 0.0118, "step": 35143 }, { "epoch": 1.873647171722557, "grad_norm": 2.400820732116699, "learning_rate": 2.9714000000000005e-06, "loss": 0.0121, "step": 35144 }, { "epoch": 1.8737004851522099, "grad_norm": 2.904085874557495, "learning_rate": 2.9712e-06, "loss": 0.032, "step": 35145 }, { "epoch": 1.8737537985818626, "grad_norm": 1.4722319841384888, "learning_rate": 2.971e-06, "loss": 0.0089, "step": 35146 }, { "epoch": 1.8738071120115158, "grad_norm": 2.6829330921173096, "learning_rate": 2.9708000000000004e-06, "loss": 0.0229, "step": 35147 }, { "epoch": 1.8738604254411686, "grad_norm": 2.0174643993377686, "learning_rate": 2.9706e-06, "loss": 0.0113, "step": 35148 }, { "epoch": 1.8739137388708216, "grad_norm": 2.0204291343688965, "learning_rate": 2.9704000000000004e-06, "loss": 0.0444, "step": 35149 }, { "epoch": 1.8739670523004746, "grad_norm": 1.603479266166687, "learning_rate": 2.9702000000000004e-06, "loss": 0.0288, "step": 35150 }, { "epoch": 1.8740203657301273, "grad_norm": 2.1711957454681396, "learning_rate": 2.97e-06, "loss": 0.0078, "step": 35151 }, { "epoch": 1.8740736791597805, "grad_norm": 1.3010066747665405, "learning_rate": 2.9698000000000003e-06, "loss": 0.0071, "step": 35152 }, { "epoch": 1.8741269925894333, "grad_norm": 2.3243203163146973, "learning_rate": 2.9696000000000003e-06, "loss": 0.0431, "step": 35153 }, { "epoch": 1.8741803060190863, "grad_norm": 3.645174503326416, "learning_rate": 2.9694e-06, "loss": 0.0342, "step": 35154 }, { "epoch": 1.8742336194487392, "grad_norm": 2.2382965087890625, "learning_rate": 2.9692000000000003e-06, "loss": 0.0221, "step": 35155 }, { "epoch": 1.874286932878392, "grad_norm": 1.2678825855255127, "learning_rate": 2.9690000000000003e-06, "loss": 0.0097, "step": 35156 }, { "epoch": 1.874340246308045, "grad_norm": 2.994635820388794, "learning_rate": 2.9688e-06, "loss": 0.0296, "step": 35157 }, { "epoch": 1.874393559737698, "grad_norm": 1.3449040651321411, "learning_rate": 2.9686000000000003e-06, "loss": 0.0134, "step": 35158 }, { "epoch": 1.8744468731673507, "grad_norm": 3.4987521171569824, "learning_rate": 2.9684000000000002e-06, "loss": 0.036, "step": 35159 }, { "epoch": 1.874500186597004, "grad_norm": 2.084111452102661, "learning_rate": 2.9682e-06, "loss": 0.0114, "step": 35160 }, { "epoch": 1.8745535000266567, "grad_norm": 1.4713715314865112, "learning_rate": 2.9680000000000002e-06, "loss": 0.0185, "step": 35161 }, { "epoch": 1.8746068134563096, "grad_norm": 2.340322494506836, "learning_rate": 2.9678e-06, "loss": 0.0131, "step": 35162 }, { "epoch": 1.8746601268859626, "grad_norm": 2.1550984382629395, "learning_rate": 2.9676000000000006e-06, "loss": 0.025, "step": 35163 }, { "epoch": 1.8747134403156154, "grad_norm": 2.3741374015808105, "learning_rate": 2.9674e-06, "loss": 0.0324, "step": 35164 }, { "epoch": 1.8747667537452686, "grad_norm": 1.3617076873779297, "learning_rate": 2.9672e-06, "loss": 0.0078, "step": 35165 }, { "epoch": 1.8748200671749213, "grad_norm": 1.1358591318130493, "learning_rate": 2.9670000000000006e-06, "loss": 0.0155, "step": 35166 }, { "epoch": 1.8748733806045743, "grad_norm": 0.9261285662651062, "learning_rate": 2.9668e-06, "loss": 0.0049, "step": 35167 }, { "epoch": 1.8749266940342273, "grad_norm": 0.9862844944000244, "learning_rate": 2.9666e-06, "loss": 0.0148, "step": 35168 }, { "epoch": 1.87498000746388, "grad_norm": 2.246227979660034, "learning_rate": 2.9664000000000005e-06, "loss": 0.0176, "step": 35169 }, { "epoch": 1.875033320893533, "grad_norm": 6.043281078338623, "learning_rate": 2.9662e-06, "loss": 0.0375, "step": 35170 }, { "epoch": 1.875086634323186, "grad_norm": 1.615881085395813, "learning_rate": 2.966e-06, "loss": 0.0157, "step": 35171 }, { "epoch": 1.875139947752839, "grad_norm": 2.5854828357696533, "learning_rate": 2.9658000000000005e-06, "loss": 0.0409, "step": 35172 }, { "epoch": 1.875193261182492, "grad_norm": 2.7032968997955322, "learning_rate": 2.9656e-06, "loss": 0.0286, "step": 35173 }, { "epoch": 1.8752465746121447, "grad_norm": 1.1194034814834595, "learning_rate": 2.9654000000000004e-06, "loss": 0.0048, "step": 35174 }, { "epoch": 1.8752998880417977, "grad_norm": 0.41949746012687683, "learning_rate": 2.9652000000000004e-06, "loss": 0.0038, "step": 35175 }, { "epoch": 1.8753532014714507, "grad_norm": 1.8535698652267456, "learning_rate": 2.965e-06, "loss": 0.0147, "step": 35176 }, { "epoch": 1.8754065149011034, "grad_norm": 4.142727375030518, "learning_rate": 2.9648000000000004e-06, "loss": 0.0475, "step": 35177 }, { "epoch": 1.8754598283307566, "grad_norm": 1.311421275138855, "learning_rate": 2.9646000000000004e-06, "loss": 0.0123, "step": 35178 }, { "epoch": 1.8755131417604094, "grad_norm": 1.3432438373565674, "learning_rate": 2.9644e-06, "loss": 0.008, "step": 35179 }, { "epoch": 1.8755664551900624, "grad_norm": 0.5118815898895264, "learning_rate": 2.9642000000000004e-06, "loss": 0.0044, "step": 35180 }, { "epoch": 1.8756197686197154, "grad_norm": 1.973150610923767, "learning_rate": 2.9640000000000003e-06, "loss": 0.0201, "step": 35181 }, { "epoch": 1.8756730820493681, "grad_norm": 0.2857774496078491, "learning_rate": 2.9638e-06, "loss": 0.0027, "step": 35182 }, { "epoch": 1.8757263954790213, "grad_norm": 1.826995849609375, "learning_rate": 2.9636000000000003e-06, "loss": 0.0169, "step": 35183 }, { "epoch": 1.875779708908674, "grad_norm": 0.9974491000175476, "learning_rate": 2.9634000000000003e-06, "loss": 0.0242, "step": 35184 }, { "epoch": 1.875833022338327, "grad_norm": 2.8308539390563965, "learning_rate": 2.9632e-06, "loss": 0.0219, "step": 35185 }, { "epoch": 1.87588633576798, "grad_norm": 2.3153836727142334, "learning_rate": 2.9630000000000003e-06, "loss": 0.036, "step": 35186 }, { "epoch": 1.8759396491976328, "grad_norm": 1.3298572301864624, "learning_rate": 2.9628000000000002e-06, "loss": 0.0296, "step": 35187 }, { "epoch": 1.8759929626272858, "grad_norm": 2.0524661540985107, "learning_rate": 2.9626000000000007e-06, "loss": 0.0216, "step": 35188 }, { "epoch": 1.8760462760569387, "grad_norm": 2.0522868633270264, "learning_rate": 2.9624000000000002e-06, "loss": 0.0328, "step": 35189 }, { "epoch": 1.8760995894865917, "grad_norm": 0.7732462286949158, "learning_rate": 2.9622e-06, "loss": 0.0053, "step": 35190 }, { "epoch": 1.8761529029162447, "grad_norm": 1.2511177062988281, "learning_rate": 2.9620000000000006e-06, "loss": 0.0097, "step": 35191 }, { "epoch": 1.8762062163458975, "grad_norm": 2.3137619495391846, "learning_rate": 2.9618e-06, "loss": 0.0284, "step": 35192 }, { "epoch": 1.8762595297755504, "grad_norm": 1.1533275842666626, "learning_rate": 2.9616e-06, "loss": 0.02, "step": 35193 }, { "epoch": 1.8763128432052034, "grad_norm": 1.460810661315918, "learning_rate": 2.9614000000000006e-06, "loss": 0.0157, "step": 35194 }, { "epoch": 1.8763661566348562, "grad_norm": 2.399732828140259, "learning_rate": 2.9612e-06, "loss": 0.0465, "step": 35195 }, { "epoch": 1.8764194700645094, "grad_norm": 2.044191598892212, "learning_rate": 2.961e-06, "loss": 0.029, "step": 35196 }, { "epoch": 1.8764727834941621, "grad_norm": 0.918002724647522, "learning_rate": 2.9608000000000005e-06, "loss": 0.0049, "step": 35197 }, { "epoch": 1.8765260969238151, "grad_norm": 2.1026744842529297, "learning_rate": 2.9606e-06, "loss": 0.0112, "step": 35198 }, { "epoch": 1.876579410353468, "grad_norm": 1.4529409408569336, "learning_rate": 2.9604000000000005e-06, "loss": 0.0187, "step": 35199 }, { "epoch": 1.8766327237831208, "grad_norm": 3.384836196899414, "learning_rate": 2.9602000000000005e-06, "loss": 0.029, "step": 35200 }, { "epoch": 1.876686037212774, "grad_norm": 2.3466408252716064, "learning_rate": 2.96e-06, "loss": 0.0188, "step": 35201 }, { "epoch": 1.8767393506424268, "grad_norm": 0.8784071803092957, "learning_rate": 2.9598000000000004e-06, "loss": 0.0055, "step": 35202 }, { "epoch": 1.8767926640720798, "grad_norm": 1.1634820699691772, "learning_rate": 2.9596000000000004e-06, "loss": 0.0089, "step": 35203 }, { "epoch": 1.8768459775017328, "grad_norm": 1.8446377515792847, "learning_rate": 2.9594e-06, "loss": 0.0086, "step": 35204 }, { "epoch": 1.8768992909313855, "grad_norm": 0.9846927523612976, "learning_rate": 2.9592000000000004e-06, "loss": 0.0144, "step": 35205 }, { "epoch": 1.8769526043610385, "grad_norm": 0.1785542070865631, "learning_rate": 2.9590000000000004e-06, "loss": 0.0018, "step": 35206 }, { "epoch": 1.8770059177906915, "grad_norm": 2.6369001865386963, "learning_rate": 2.9588e-06, "loss": 0.0343, "step": 35207 }, { "epoch": 1.8770592312203445, "grad_norm": 1.9392553567886353, "learning_rate": 2.9586000000000004e-06, "loss": 0.0138, "step": 35208 }, { "epoch": 1.8771125446499974, "grad_norm": 1.4289134740829468, "learning_rate": 2.9584000000000003e-06, "loss": 0.0194, "step": 35209 }, { "epoch": 1.8771658580796502, "grad_norm": 0.9398199915885925, "learning_rate": 2.9582000000000003e-06, "loss": 0.0188, "step": 35210 }, { "epoch": 1.8772191715093032, "grad_norm": 1.9785255193710327, "learning_rate": 2.9580000000000003e-06, "loss": 0.0129, "step": 35211 }, { "epoch": 1.8772724849389562, "grad_norm": 1.1670372486114502, "learning_rate": 2.9578000000000003e-06, "loss": 0.0066, "step": 35212 }, { "epoch": 1.877325798368609, "grad_norm": 2.238800048828125, "learning_rate": 2.9576000000000003e-06, "loss": 0.0306, "step": 35213 }, { "epoch": 1.8773791117982621, "grad_norm": 2.7845852375030518, "learning_rate": 2.9574000000000003e-06, "loss": 0.0133, "step": 35214 }, { "epoch": 1.8774324252279149, "grad_norm": 0.44996514916419983, "learning_rate": 2.9572000000000002e-06, "loss": 0.0034, "step": 35215 }, { "epoch": 1.8774857386575678, "grad_norm": 2.4621329307556152, "learning_rate": 2.9570000000000002e-06, "loss": 0.0185, "step": 35216 }, { "epoch": 1.8775390520872208, "grad_norm": 2.035928249359131, "learning_rate": 2.9568000000000002e-06, "loss": 0.0169, "step": 35217 }, { "epoch": 1.8775923655168736, "grad_norm": 1.9275944232940674, "learning_rate": 2.9566e-06, "loss": 0.0373, "step": 35218 }, { "epoch": 1.8776456789465268, "grad_norm": 3.2297251224517822, "learning_rate": 2.9564e-06, "loss": 0.0166, "step": 35219 }, { "epoch": 1.8776989923761795, "grad_norm": 1.5637876987457275, "learning_rate": 2.9562e-06, "loss": 0.0414, "step": 35220 }, { "epoch": 1.8777523058058325, "grad_norm": 0.3820067048072815, "learning_rate": 2.956e-06, "loss": 0.0042, "step": 35221 }, { "epoch": 1.8778056192354855, "grad_norm": 1.3677785396575928, "learning_rate": 2.9558e-06, "loss": 0.0075, "step": 35222 }, { "epoch": 1.8778589326651383, "grad_norm": 1.1116936206817627, "learning_rate": 2.9556e-06, "loss": 0.0126, "step": 35223 }, { "epoch": 1.8779122460947912, "grad_norm": 0.8928027749061584, "learning_rate": 2.9554000000000005e-06, "loss": 0.0128, "step": 35224 }, { "epoch": 1.8779655595244442, "grad_norm": 1.6642714738845825, "learning_rate": 2.9552e-06, "loss": 0.0149, "step": 35225 }, { "epoch": 1.878018872954097, "grad_norm": 3.236987352371216, "learning_rate": 2.955e-06, "loss": 0.0449, "step": 35226 }, { "epoch": 1.8780721863837502, "grad_norm": 1.1340514421463013, "learning_rate": 2.9548000000000005e-06, "loss": 0.0075, "step": 35227 }, { "epoch": 1.878125499813403, "grad_norm": 2.1225574016571045, "learning_rate": 2.9546e-06, "loss": 0.0354, "step": 35228 }, { "epoch": 1.878178813243056, "grad_norm": 2.5419654846191406, "learning_rate": 2.9544e-06, "loss": 0.0297, "step": 35229 }, { "epoch": 1.878232126672709, "grad_norm": 1.7342449426651, "learning_rate": 2.9542000000000004e-06, "loss": 0.0089, "step": 35230 }, { "epoch": 1.8782854401023616, "grad_norm": 2.3129169940948486, "learning_rate": 2.954e-06, "loss": 0.0181, "step": 35231 }, { "epoch": 1.8783387535320148, "grad_norm": 1.7764947414398193, "learning_rate": 2.9538e-06, "loss": 0.0143, "step": 35232 }, { "epoch": 1.8783920669616676, "grad_norm": 1.1055376529693604, "learning_rate": 2.9536000000000004e-06, "loss": 0.0082, "step": 35233 }, { "epoch": 1.8784453803913206, "grad_norm": 2.6899502277374268, "learning_rate": 2.9534e-06, "loss": 0.0316, "step": 35234 }, { "epoch": 1.8784986938209736, "grad_norm": 0.8400440216064453, "learning_rate": 2.9532000000000004e-06, "loss": 0.0052, "step": 35235 }, { "epoch": 1.8785520072506263, "grad_norm": 2.071749687194824, "learning_rate": 2.9530000000000004e-06, "loss": 0.019, "step": 35236 }, { "epoch": 1.8786053206802795, "grad_norm": 1.3743709325790405, "learning_rate": 2.9528e-06, "loss": 0.0174, "step": 35237 }, { "epoch": 1.8786586341099323, "grad_norm": 2.7963271141052246, "learning_rate": 2.9526000000000003e-06, "loss": 0.0187, "step": 35238 }, { "epoch": 1.8787119475395853, "grad_norm": 2.7610952854156494, "learning_rate": 2.9524000000000003e-06, "loss": 0.0369, "step": 35239 }, { "epoch": 1.8787652609692382, "grad_norm": 1.079405426979065, "learning_rate": 2.9522e-06, "loss": 0.019, "step": 35240 }, { "epoch": 1.878818574398891, "grad_norm": 2.555826187133789, "learning_rate": 2.9520000000000003e-06, "loss": 0.019, "step": 35241 }, { "epoch": 1.878871887828544, "grad_norm": 1.4489437341690063, "learning_rate": 2.9518000000000003e-06, "loss": 0.0082, "step": 35242 }, { "epoch": 1.878925201258197, "grad_norm": 3.8577792644500732, "learning_rate": 2.9516e-06, "loss": 0.0458, "step": 35243 }, { "epoch": 1.8789785146878497, "grad_norm": 1.8071857690811157, "learning_rate": 2.9514000000000002e-06, "loss": 0.0335, "step": 35244 }, { "epoch": 1.879031828117503, "grad_norm": 1.7245880365371704, "learning_rate": 2.9512000000000002e-06, "loss": 0.0138, "step": 35245 }, { "epoch": 1.8790851415471557, "grad_norm": 3.05955171585083, "learning_rate": 2.9509999999999998e-06, "loss": 0.0195, "step": 35246 }, { "epoch": 1.8791384549768086, "grad_norm": 1.0784636735916138, "learning_rate": 2.9508e-06, "loss": 0.0332, "step": 35247 }, { "epoch": 1.8791917684064616, "grad_norm": 1.7679449319839478, "learning_rate": 2.9506e-06, "loss": 0.0255, "step": 35248 }, { "epoch": 1.8792450818361144, "grad_norm": 2.7134811878204346, "learning_rate": 2.9504000000000006e-06, "loss": 0.0396, "step": 35249 }, { "epoch": 1.8792983952657676, "grad_norm": 1.8193761110305786, "learning_rate": 2.9502e-06, "loss": 0.0126, "step": 35250 }, { "epoch": 1.8793517086954203, "grad_norm": 1.3621174097061157, "learning_rate": 2.95e-06, "loss": 0.0195, "step": 35251 }, { "epoch": 1.8794050221250733, "grad_norm": 4.284528732299805, "learning_rate": 2.9498000000000005e-06, "loss": 0.0107, "step": 35252 }, { "epoch": 1.8794583355547263, "grad_norm": 1.731351375579834, "learning_rate": 2.9496e-06, "loss": 0.0324, "step": 35253 }, { "epoch": 1.879511648984379, "grad_norm": 1.510320782661438, "learning_rate": 2.9494e-06, "loss": 0.0103, "step": 35254 }, { "epoch": 1.8795649624140323, "grad_norm": 2.2467217445373535, "learning_rate": 2.9492000000000005e-06, "loss": 0.0168, "step": 35255 }, { "epoch": 1.879618275843685, "grad_norm": 0.9403516054153442, "learning_rate": 2.949e-06, "loss": 0.0062, "step": 35256 }, { "epoch": 1.879671589273338, "grad_norm": 1.0095746517181396, "learning_rate": 2.9488e-06, "loss": 0.014, "step": 35257 }, { "epoch": 1.879724902702991, "grad_norm": 1.685760498046875, "learning_rate": 2.9486000000000004e-06, "loss": 0.0164, "step": 35258 }, { "epoch": 1.8797782161326437, "grad_norm": 0.6317241191864014, "learning_rate": 2.9484e-06, "loss": 0.0049, "step": 35259 }, { "epoch": 1.8798315295622967, "grad_norm": 1.1169354915618896, "learning_rate": 2.9482000000000004e-06, "loss": 0.0085, "step": 35260 }, { "epoch": 1.8798848429919497, "grad_norm": 1.2544535398483276, "learning_rate": 2.9480000000000004e-06, "loss": 0.016, "step": 35261 }, { "epoch": 1.8799381564216024, "grad_norm": 1.6929231882095337, "learning_rate": 2.9478e-06, "loss": 0.0221, "step": 35262 }, { "epoch": 1.8799914698512556, "grad_norm": 1.2078124284744263, "learning_rate": 2.9476000000000004e-06, "loss": 0.0141, "step": 35263 }, { "epoch": 1.8800447832809084, "grad_norm": 2.5339150428771973, "learning_rate": 2.9474000000000004e-06, "loss": 0.0255, "step": 35264 }, { "epoch": 1.8800980967105614, "grad_norm": 2.108431816101074, "learning_rate": 2.9472e-06, "loss": 0.0232, "step": 35265 }, { "epoch": 1.8801514101402144, "grad_norm": 0.9340084791183472, "learning_rate": 2.9470000000000003e-06, "loss": 0.0186, "step": 35266 }, { "epoch": 1.8802047235698671, "grad_norm": 4.639756679534912, "learning_rate": 2.9468000000000003e-06, "loss": 0.0304, "step": 35267 }, { "epoch": 1.8802580369995203, "grad_norm": 1.6091740131378174, "learning_rate": 2.9466e-06, "loss": 0.0197, "step": 35268 }, { "epoch": 1.880311350429173, "grad_norm": 0.3320983350276947, "learning_rate": 2.9464000000000003e-06, "loss": 0.0027, "step": 35269 }, { "epoch": 1.880364663858826, "grad_norm": 2.367466688156128, "learning_rate": 2.9462000000000003e-06, "loss": 0.019, "step": 35270 }, { "epoch": 1.880417977288479, "grad_norm": 3.4212515354156494, "learning_rate": 2.946e-06, "loss": 0.0093, "step": 35271 }, { "epoch": 1.8804712907181318, "grad_norm": 1.6620450019836426, "learning_rate": 2.9458000000000002e-06, "loss": 0.0119, "step": 35272 }, { "epoch": 1.8805246041477848, "grad_norm": 2.2879488468170166, "learning_rate": 2.9456000000000002e-06, "loss": 0.0337, "step": 35273 }, { "epoch": 1.8805779175774378, "grad_norm": 1.8919013738632202, "learning_rate": 2.9454000000000006e-06, "loss": 0.0401, "step": 35274 }, { "epoch": 1.8806312310070907, "grad_norm": 1.3806482553482056, "learning_rate": 2.9452e-06, "loss": 0.0233, "step": 35275 }, { "epoch": 1.8806845444367437, "grad_norm": 1.063550591468811, "learning_rate": 2.945e-06, "loss": 0.0158, "step": 35276 }, { "epoch": 1.8807378578663965, "grad_norm": 0.8194193840026855, "learning_rate": 2.9448000000000006e-06, "loss": 0.0045, "step": 35277 }, { "epoch": 1.8807911712960494, "grad_norm": 2.3276987075805664, "learning_rate": 2.9446e-06, "loss": 0.0327, "step": 35278 }, { "epoch": 1.8808444847257024, "grad_norm": 0.5241284370422363, "learning_rate": 2.9444e-06, "loss": 0.0038, "step": 35279 }, { "epoch": 1.8808977981553552, "grad_norm": 2.202821731567383, "learning_rate": 2.9442000000000005e-06, "loss": 0.0227, "step": 35280 }, { "epoch": 1.8809511115850084, "grad_norm": 3.068477153778076, "learning_rate": 2.944e-06, "loss": 0.0285, "step": 35281 }, { "epoch": 1.8810044250146611, "grad_norm": 0.3185766041278839, "learning_rate": 2.9438e-06, "loss": 0.0031, "step": 35282 }, { "epoch": 1.8810577384443141, "grad_norm": 0.8853524923324585, "learning_rate": 2.9436000000000005e-06, "loss": 0.0055, "step": 35283 }, { "epoch": 1.881111051873967, "grad_norm": 1.7126199007034302, "learning_rate": 2.9434e-06, "loss": 0.0221, "step": 35284 }, { "epoch": 1.8811643653036199, "grad_norm": 0.7706205248832703, "learning_rate": 2.9432000000000005e-06, "loss": 0.0052, "step": 35285 }, { "epoch": 1.881217678733273, "grad_norm": 0.4413754642009735, "learning_rate": 2.9430000000000005e-06, "loss": 0.0041, "step": 35286 }, { "epoch": 1.8812709921629258, "grad_norm": 2.332509994506836, "learning_rate": 2.9428e-06, "loss": 0.0215, "step": 35287 }, { "epoch": 1.8813243055925788, "grad_norm": 2.1826889514923096, "learning_rate": 2.9426000000000004e-06, "loss": 0.0133, "step": 35288 }, { "epoch": 1.8813776190222318, "grad_norm": 2.2184925079345703, "learning_rate": 2.9424000000000004e-06, "loss": 0.0379, "step": 35289 }, { "epoch": 1.8814309324518845, "grad_norm": 0.5727997422218323, "learning_rate": 2.9422e-06, "loss": 0.0047, "step": 35290 }, { "epoch": 1.8814842458815375, "grad_norm": 2.2389183044433594, "learning_rate": 2.9420000000000004e-06, "loss": 0.0102, "step": 35291 }, { "epoch": 1.8815375593111905, "grad_norm": 2.0218279361724854, "learning_rate": 2.9418000000000004e-06, "loss": 0.0234, "step": 35292 }, { "epoch": 1.8815908727408435, "grad_norm": 1.4223461151123047, "learning_rate": 2.9416e-06, "loss": 0.0095, "step": 35293 }, { "epoch": 1.8816441861704964, "grad_norm": 1.9020326137542725, "learning_rate": 2.9414000000000003e-06, "loss": 0.0242, "step": 35294 }, { "epoch": 1.8816974996001492, "grad_norm": 2.2735867500305176, "learning_rate": 2.9412000000000003e-06, "loss": 0.0411, "step": 35295 }, { "epoch": 1.8817508130298022, "grad_norm": 1.0435184240341187, "learning_rate": 2.941e-06, "loss": 0.0162, "step": 35296 }, { "epoch": 1.8818041264594552, "grad_norm": 0.6973111629486084, "learning_rate": 2.9408000000000003e-06, "loss": 0.0043, "step": 35297 }, { "epoch": 1.881857439889108, "grad_norm": 3.1097888946533203, "learning_rate": 2.9406000000000003e-06, "loss": 0.0471, "step": 35298 }, { "epoch": 1.8819107533187611, "grad_norm": 0.29258930683135986, "learning_rate": 2.9404000000000007e-06, "loss": 0.003, "step": 35299 }, { "epoch": 1.8819640667484139, "grad_norm": 1.9362012147903442, "learning_rate": 2.9402000000000002e-06, "loss": 0.0349, "step": 35300 }, { "epoch": 1.8820173801780669, "grad_norm": 1.4521158933639526, "learning_rate": 2.9400000000000002e-06, "loss": 0.0154, "step": 35301 }, { "epoch": 1.8820706936077198, "grad_norm": 2.2044999599456787, "learning_rate": 2.9398000000000006e-06, "loss": 0.0407, "step": 35302 }, { "epoch": 1.8821240070373726, "grad_norm": 1.2554593086242676, "learning_rate": 2.9396e-06, "loss": 0.0259, "step": 35303 }, { "epoch": 1.8821773204670258, "grad_norm": 0.3996405303478241, "learning_rate": 2.9394e-06, "loss": 0.0039, "step": 35304 }, { "epoch": 1.8822306338966786, "grad_norm": 2.1094841957092285, "learning_rate": 2.9392000000000006e-06, "loss": 0.0136, "step": 35305 }, { "epoch": 1.8822839473263315, "grad_norm": 0.7456294298171997, "learning_rate": 2.939e-06, "loss": 0.0065, "step": 35306 }, { "epoch": 1.8823372607559845, "grad_norm": 1.964760184288025, "learning_rate": 2.9388e-06, "loss": 0.0121, "step": 35307 }, { "epoch": 1.8823905741856373, "grad_norm": 2.0369582176208496, "learning_rate": 2.9386000000000005e-06, "loss": 0.031, "step": 35308 }, { "epoch": 1.8824438876152902, "grad_norm": 3.4744935035705566, "learning_rate": 2.9384e-06, "loss": 0.0239, "step": 35309 }, { "epoch": 1.8824972010449432, "grad_norm": 1.65497887134552, "learning_rate": 2.9382000000000005e-06, "loss": 0.0228, "step": 35310 }, { "epoch": 1.8825505144745962, "grad_norm": 1.6960012912750244, "learning_rate": 2.9380000000000005e-06, "loss": 0.0244, "step": 35311 }, { "epoch": 1.8826038279042492, "grad_norm": 1.113205075263977, "learning_rate": 2.9378e-06, "loss": 0.0131, "step": 35312 }, { "epoch": 1.882657141333902, "grad_norm": 2.3551480770111084, "learning_rate": 2.9376000000000005e-06, "loss": 0.0433, "step": 35313 }, { "epoch": 1.882710454763555, "grad_norm": 2.718404531478882, "learning_rate": 2.9374000000000005e-06, "loss": 0.0343, "step": 35314 }, { "epoch": 1.882763768193208, "grad_norm": 2.150935173034668, "learning_rate": 2.9372e-06, "loss": 0.0221, "step": 35315 }, { "epoch": 1.8828170816228607, "grad_norm": 1.1436986923217773, "learning_rate": 2.9370000000000004e-06, "loss": 0.0099, "step": 35316 }, { "epoch": 1.8828703950525139, "grad_norm": 2.093801498413086, "learning_rate": 2.9368000000000004e-06, "loss": 0.0119, "step": 35317 }, { "epoch": 1.8829237084821666, "grad_norm": 0.9068362712860107, "learning_rate": 2.9366e-06, "loss": 0.0071, "step": 35318 }, { "epoch": 1.8829770219118196, "grad_norm": 1.2817795276641846, "learning_rate": 2.9364000000000004e-06, "loss": 0.0067, "step": 35319 }, { "epoch": 1.8830303353414726, "grad_norm": 3.075298309326172, "learning_rate": 2.9362000000000004e-06, "loss": 0.0273, "step": 35320 }, { "epoch": 1.8830836487711253, "grad_norm": 0.8832865357398987, "learning_rate": 2.9360000000000003e-06, "loss": 0.0073, "step": 35321 }, { "epoch": 1.8831369622007785, "grad_norm": 1.4679213762283325, "learning_rate": 2.9358000000000003e-06, "loss": 0.0149, "step": 35322 }, { "epoch": 1.8831902756304313, "grad_norm": 1.8229619264602661, "learning_rate": 2.9356000000000003e-06, "loss": 0.0418, "step": 35323 }, { "epoch": 1.8832435890600843, "grad_norm": 2.4148526191711426, "learning_rate": 2.9354000000000003e-06, "loss": 0.0114, "step": 35324 }, { "epoch": 1.8832969024897372, "grad_norm": 2.76729416847229, "learning_rate": 2.9352000000000003e-06, "loss": 0.0185, "step": 35325 }, { "epoch": 1.88335021591939, "grad_norm": 1.4435434341430664, "learning_rate": 2.9350000000000003e-06, "loss": 0.006, "step": 35326 }, { "epoch": 1.883403529349043, "grad_norm": 1.7391232252120972, "learning_rate": 2.9348000000000003e-06, "loss": 0.0398, "step": 35327 }, { "epoch": 1.883456842778696, "grad_norm": 1.2273356914520264, "learning_rate": 2.9346000000000002e-06, "loss": 0.016, "step": 35328 }, { "epoch": 1.8835101562083487, "grad_norm": 1.5402624607086182, "learning_rate": 2.9344000000000002e-06, "loss": 0.021, "step": 35329 }, { "epoch": 1.883563469638002, "grad_norm": 2.6995067596435547, "learning_rate": 2.9342000000000002e-06, "loss": 0.0278, "step": 35330 }, { "epoch": 1.8836167830676547, "grad_norm": 1.5092750787734985, "learning_rate": 2.934e-06, "loss": 0.035, "step": 35331 }, { "epoch": 1.8836700964973077, "grad_norm": 0.7157077193260193, "learning_rate": 2.9338e-06, "loss": 0.0058, "step": 35332 }, { "epoch": 1.8837234099269606, "grad_norm": 0.6899946928024292, "learning_rate": 2.9336e-06, "loss": 0.0052, "step": 35333 }, { "epoch": 1.8837767233566134, "grad_norm": 1.0127463340759277, "learning_rate": 2.9334e-06, "loss": 0.0124, "step": 35334 }, { "epoch": 1.8838300367862666, "grad_norm": 2.4545047283172607, "learning_rate": 2.9332000000000006e-06, "loss": 0.0282, "step": 35335 }, { "epoch": 1.8838833502159194, "grad_norm": 1.0343713760375977, "learning_rate": 2.933e-06, "loss": 0.0152, "step": 35336 }, { "epoch": 1.8839366636455723, "grad_norm": 1.850626826286316, "learning_rate": 2.9328e-06, "loss": 0.0194, "step": 35337 }, { "epoch": 1.8839899770752253, "grad_norm": 1.2082420587539673, "learning_rate": 2.9326000000000005e-06, "loss": 0.0169, "step": 35338 }, { "epoch": 1.884043290504878, "grad_norm": 2.3638663291931152, "learning_rate": 2.9324e-06, "loss": 0.029, "step": 35339 }, { "epoch": 1.8840966039345313, "grad_norm": 1.3657913208007812, "learning_rate": 2.9322e-06, "loss": 0.0114, "step": 35340 }, { "epoch": 1.884149917364184, "grad_norm": 1.514499545097351, "learning_rate": 2.9320000000000005e-06, "loss": 0.0127, "step": 35341 }, { "epoch": 1.884203230793837, "grad_norm": 1.363036036491394, "learning_rate": 2.9318e-06, "loss": 0.0214, "step": 35342 }, { "epoch": 1.88425654422349, "grad_norm": 2.395303249359131, "learning_rate": 2.9316e-06, "loss": 0.0148, "step": 35343 }, { "epoch": 1.8843098576531427, "grad_norm": 1.9271684885025024, "learning_rate": 2.9314000000000004e-06, "loss": 0.0326, "step": 35344 }, { "epoch": 1.8843631710827957, "grad_norm": 2.7429747581481934, "learning_rate": 2.9312e-06, "loss": 0.0185, "step": 35345 }, { "epoch": 1.8844164845124487, "grad_norm": 2.3153798580169678, "learning_rate": 2.9310000000000004e-06, "loss": 0.041, "step": 35346 }, { "epoch": 1.8844697979421015, "grad_norm": 1.2517449855804443, "learning_rate": 2.9308000000000004e-06, "loss": 0.0256, "step": 35347 }, { "epoch": 1.8845231113717547, "grad_norm": 4.309194564819336, "learning_rate": 2.9306e-06, "loss": 0.0326, "step": 35348 }, { "epoch": 1.8845764248014074, "grad_norm": 0.72170090675354, "learning_rate": 2.9304000000000004e-06, "loss": 0.005, "step": 35349 }, { "epoch": 1.8846297382310604, "grad_norm": 0.6793705821037292, "learning_rate": 2.9302000000000003e-06, "loss": 0.0053, "step": 35350 }, { "epoch": 1.8846830516607134, "grad_norm": 2.3960025310516357, "learning_rate": 2.93e-06, "loss": 0.0555, "step": 35351 }, { "epoch": 1.8847363650903661, "grad_norm": 1.4271190166473389, "learning_rate": 2.9298000000000003e-06, "loss": 0.0087, "step": 35352 }, { "epoch": 1.8847896785200193, "grad_norm": 1.823972225189209, "learning_rate": 2.9296000000000003e-06, "loss": 0.0085, "step": 35353 }, { "epoch": 1.884842991949672, "grad_norm": 2.3459393978118896, "learning_rate": 2.9294e-06, "loss": 0.0302, "step": 35354 }, { "epoch": 1.884896305379325, "grad_norm": 4.378702163696289, "learning_rate": 2.9292000000000003e-06, "loss": 0.0318, "step": 35355 }, { "epoch": 1.884949618808978, "grad_norm": 0.37602347135543823, "learning_rate": 2.9290000000000002e-06, "loss": 0.0033, "step": 35356 }, { "epoch": 1.8850029322386308, "grad_norm": 2.121690273284912, "learning_rate": 2.9288e-06, "loss": 0.0099, "step": 35357 }, { "epoch": 1.885056245668284, "grad_norm": 1.5357121229171753, "learning_rate": 2.9286000000000002e-06, "loss": 0.009, "step": 35358 }, { "epoch": 1.8851095590979368, "grad_norm": 1.1657288074493408, "learning_rate": 2.9284e-06, "loss": 0.0096, "step": 35359 }, { "epoch": 1.8851628725275897, "grad_norm": 1.249485731124878, "learning_rate": 2.9282000000000006e-06, "loss": 0.0229, "step": 35360 }, { "epoch": 1.8852161859572427, "grad_norm": 1.4631662368774414, "learning_rate": 2.928e-06, "loss": 0.0179, "step": 35361 }, { "epoch": 1.8852694993868955, "grad_norm": 2.3318586349487305, "learning_rate": 2.9278e-06, "loss": 0.0271, "step": 35362 }, { "epoch": 1.8853228128165485, "grad_norm": 2.372866630554199, "learning_rate": 2.9276000000000006e-06, "loss": 0.0147, "step": 35363 }, { "epoch": 1.8853761262462014, "grad_norm": 1.3436931371688843, "learning_rate": 2.9274e-06, "loss": 0.0092, "step": 35364 }, { "epoch": 1.8854294396758542, "grad_norm": 0.5632618069648743, "learning_rate": 2.9272e-06, "loss": 0.0058, "step": 35365 }, { "epoch": 1.8854827531055074, "grad_norm": 1.6882647275924683, "learning_rate": 2.9270000000000005e-06, "loss": 0.0294, "step": 35366 }, { "epoch": 1.8855360665351601, "grad_norm": 1.3169755935668945, "learning_rate": 2.9268e-06, "loss": 0.0082, "step": 35367 }, { "epoch": 1.8855893799648131, "grad_norm": 1.0184581279754639, "learning_rate": 2.9266e-06, "loss": 0.0148, "step": 35368 }, { "epoch": 1.885642693394466, "grad_norm": 1.3404896259307861, "learning_rate": 2.9264000000000005e-06, "loss": 0.0218, "step": 35369 }, { "epoch": 1.8856960068241189, "grad_norm": 0.6619217991828918, "learning_rate": 2.9262e-06, "loss": 0.0043, "step": 35370 }, { "epoch": 1.885749320253772, "grad_norm": 0.6127753257751465, "learning_rate": 2.9260000000000004e-06, "loss": 0.0039, "step": 35371 }, { "epoch": 1.8858026336834248, "grad_norm": 2.4696784019470215, "learning_rate": 2.9258000000000004e-06, "loss": 0.0206, "step": 35372 }, { "epoch": 1.8858559471130778, "grad_norm": 1.7802531719207764, "learning_rate": 2.9256e-06, "loss": 0.0243, "step": 35373 }, { "epoch": 1.8859092605427308, "grad_norm": 1.7188706398010254, "learning_rate": 2.9254000000000004e-06, "loss": 0.0249, "step": 35374 }, { "epoch": 1.8859625739723835, "grad_norm": 1.9322898387908936, "learning_rate": 2.9252000000000004e-06, "loss": 0.0243, "step": 35375 }, { "epoch": 1.8860158874020367, "grad_norm": 1.919858455657959, "learning_rate": 2.925e-06, "loss": 0.0455, "step": 35376 }, { "epoch": 1.8860692008316895, "grad_norm": 1.1903451681137085, "learning_rate": 2.9248000000000004e-06, "loss": 0.0135, "step": 35377 }, { "epoch": 1.8861225142613425, "grad_norm": 1.363328218460083, "learning_rate": 2.9246000000000003e-06, "loss": 0.0179, "step": 35378 }, { "epoch": 1.8861758276909955, "grad_norm": 1.6322011947631836, "learning_rate": 2.9244e-06, "loss": 0.0199, "step": 35379 }, { "epoch": 1.8862291411206482, "grad_norm": 1.5231508016586304, "learning_rate": 2.9242000000000003e-06, "loss": 0.017, "step": 35380 }, { "epoch": 1.8862824545503012, "grad_norm": 0.8251993656158447, "learning_rate": 2.9240000000000003e-06, "loss": 0.006, "step": 35381 }, { "epoch": 1.8863357679799542, "grad_norm": 2.0406181812286377, "learning_rate": 2.9238e-06, "loss": 0.0221, "step": 35382 }, { "epoch": 1.886389081409607, "grad_norm": 0.4127649962902069, "learning_rate": 2.9236000000000003e-06, "loss": 0.003, "step": 35383 }, { "epoch": 1.8864423948392601, "grad_norm": 2.4301750659942627, "learning_rate": 2.9234000000000003e-06, "loss": 0.0355, "step": 35384 }, { "epoch": 1.8864957082689129, "grad_norm": 3.317365884780884, "learning_rate": 2.9232000000000007e-06, "loss": 0.0333, "step": 35385 }, { "epoch": 1.8865490216985659, "grad_norm": 2.180924415588379, "learning_rate": 2.9230000000000002e-06, "loss": 0.0255, "step": 35386 }, { "epoch": 1.8866023351282188, "grad_norm": 1.8274869918823242, "learning_rate": 2.9228e-06, "loss": 0.0178, "step": 35387 }, { "epoch": 1.8866556485578716, "grad_norm": 2.4580259323120117, "learning_rate": 2.9226000000000006e-06, "loss": 0.0241, "step": 35388 }, { "epoch": 1.8867089619875248, "grad_norm": 1.9519262313842773, "learning_rate": 2.9224e-06, "loss": 0.0139, "step": 35389 }, { "epoch": 1.8867622754171776, "grad_norm": 2.299926280975342, "learning_rate": 2.9222e-06, "loss": 0.0344, "step": 35390 }, { "epoch": 1.8868155888468305, "grad_norm": 1.2960656881332397, "learning_rate": 2.9220000000000006e-06, "loss": 0.0213, "step": 35391 }, { "epoch": 1.8868689022764835, "grad_norm": 1.386741280555725, "learning_rate": 2.9218e-06, "loss": 0.0133, "step": 35392 }, { "epoch": 1.8869222157061363, "grad_norm": 0.8301623463630676, "learning_rate": 2.9216e-06, "loss": 0.0141, "step": 35393 }, { "epoch": 1.8869755291357893, "grad_norm": 1.3446191549301147, "learning_rate": 2.9214000000000005e-06, "loss": 0.0102, "step": 35394 }, { "epoch": 1.8870288425654422, "grad_norm": 1.0736290216445923, "learning_rate": 2.9212e-06, "loss": 0.0122, "step": 35395 }, { "epoch": 1.8870821559950952, "grad_norm": 1.0730522871017456, "learning_rate": 2.9210000000000005e-06, "loss": 0.0204, "step": 35396 }, { "epoch": 1.8871354694247482, "grad_norm": 1.1864460706710815, "learning_rate": 2.9208000000000005e-06, "loss": 0.0065, "step": 35397 }, { "epoch": 1.887188782854401, "grad_norm": 1.4890321493148804, "learning_rate": 2.9206e-06, "loss": 0.0138, "step": 35398 }, { "epoch": 1.887242096284054, "grad_norm": 0.7070789337158203, "learning_rate": 2.9204000000000004e-06, "loss": 0.011, "step": 35399 }, { "epoch": 1.887295409713707, "grad_norm": 0.9335952401161194, "learning_rate": 2.9202000000000004e-06, "loss": 0.0065, "step": 35400 }, { "epoch": 1.8873487231433597, "grad_norm": 1.830202579498291, "learning_rate": 2.92e-06, "loss": 0.0392, "step": 35401 }, { "epoch": 1.8874020365730129, "grad_norm": 1.912122130393982, "learning_rate": 2.9198000000000004e-06, "loss": 0.0321, "step": 35402 }, { "epoch": 1.8874553500026656, "grad_norm": 1.5319565534591675, "learning_rate": 2.9196000000000004e-06, "loss": 0.0272, "step": 35403 }, { "epoch": 1.8875086634323186, "grad_norm": 2.1686596870422363, "learning_rate": 2.9194e-06, "loss": 0.0372, "step": 35404 }, { "epoch": 1.8875619768619716, "grad_norm": 0.7383586764335632, "learning_rate": 2.9192000000000004e-06, "loss": 0.0047, "step": 35405 }, { "epoch": 1.8876152902916243, "grad_norm": 0.43479621410369873, "learning_rate": 2.9190000000000003e-06, "loss": 0.0039, "step": 35406 }, { "epoch": 1.8876686037212775, "grad_norm": 0.9172763228416443, "learning_rate": 2.9188e-06, "loss": 0.0173, "step": 35407 }, { "epoch": 1.8877219171509303, "grad_norm": 2.997248888015747, "learning_rate": 2.9186000000000003e-06, "loss": 0.0263, "step": 35408 }, { "epoch": 1.8877752305805833, "grad_norm": 0.9227194786071777, "learning_rate": 2.9184000000000003e-06, "loss": 0.0075, "step": 35409 }, { "epoch": 1.8878285440102363, "grad_norm": 3.021848440170288, "learning_rate": 2.9182000000000003e-06, "loss": 0.0356, "step": 35410 }, { "epoch": 1.887881857439889, "grad_norm": 1.6452194452285767, "learning_rate": 2.9180000000000003e-06, "loss": 0.0269, "step": 35411 }, { "epoch": 1.887935170869542, "grad_norm": 1.0821670293807983, "learning_rate": 2.9178000000000003e-06, "loss": 0.0101, "step": 35412 }, { "epoch": 1.887988484299195, "grad_norm": 6.142338752746582, "learning_rate": 2.9176000000000002e-06, "loss": 0.0203, "step": 35413 }, { "epoch": 1.888041797728848, "grad_norm": 2.3735716342926025, "learning_rate": 2.9174000000000002e-06, "loss": 0.0263, "step": 35414 }, { "epoch": 1.888095111158501, "grad_norm": 1.4427894353866577, "learning_rate": 2.9172e-06, "loss": 0.0092, "step": 35415 }, { "epoch": 1.8881484245881537, "grad_norm": 1.7637827396392822, "learning_rate": 2.917e-06, "loss": 0.0412, "step": 35416 }, { "epoch": 1.8882017380178067, "grad_norm": 2.3864405155181885, "learning_rate": 2.9168e-06, "loss": 0.0125, "step": 35417 }, { "epoch": 1.8882550514474596, "grad_norm": 2.2607553005218506, "learning_rate": 2.9166e-06, "loss": 0.0245, "step": 35418 }, { "epoch": 1.8883083648771124, "grad_norm": 3.6303551197052, "learning_rate": 2.9164e-06, "loss": 0.039, "step": 35419 }, { "epoch": 1.8883616783067656, "grad_norm": 3.7754006385803223, "learning_rate": 2.9162e-06, "loss": 0.0346, "step": 35420 }, { "epoch": 1.8884149917364184, "grad_norm": 1.0397160053253174, "learning_rate": 2.9160000000000005e-06, "loss": 0.013, "step": 35421 }, { "epoch": 1.8884683051660713, "grad_norm": 1.5767230987548828, "learning_rate": 2.9158e-06, "loss": 0.0181, "step": 35422 }, { "epoch": 1.8885216185957243, "grad_norm": 1.3255640268325806, "learning_rate": 2.9156e-06, "loss": 0.0263, "step": 35423 }, { "epoch": 1.888574932025377, "grad_norm": 0.3047512173652649, "learning_rate": 2.9154000000000005e-06, "loss": 0.0027, "step": 35424 }, { "epoch": 1.8886282454550303, "grad_norm": 1.9713382720947266, "learning_rate": 2.9152e-06, "loss": 0.0078, "step": 35425 }, { "epoch": 1.888681558884683, "grad_norm": 0.7764063477516174, "learning_rate": 2.915e-06, "loss": 0.0043, "step": 35426 }, { "epoch": 1.888734872314336, "grad_norm": 0.9166805744171143, "learning_rate": 2.9148000000000005e-06, "loss": 0.0067, "step": 35427 }, { "epoch": 1.888788185743989, "grad_norm": 2.255079984664917, "learning_rate": 2.9146e-06, "loss": 0.0444, "step": 35428 }, { "epoch": 1.8888414991736417, "grad_norm": 1.1710413694381714, "learning_rate": 2.9144e-06, "loss": 0.0159, "step": 35429 }, { "epoch": 1.8888948126032947, "grad_norm": 1.4584712982177734, "learning_rate": 2.9142000000000004e-06, "loss": 0.0148, "step": 35430 }, { "epoch": 1.8889481260329477, "grad_norm": 1.9562733173370361, "learning_rate": 2.914e-06, "loss": 0.0181, "step": 35431 }, { "epoch": 1.8890014394626007, "grad_norm": 1.3397244215011597, "learning_rate": 2.9138e-06, "loss": 0.0165, "step": 35432 }, { "epoch": 1.8890547528922537, "grad_norm": 1.1691750288009644, "learning_rate": 2.9136000000000004e-06, "loss": 0.0244, "step": 35433 }, { "epoch": 1.8891080663219064, "grad_norm": 0.8016042709350586, "learning_rate": 2.9134e-06, "loss": 0.0069, "step": 35434 }, { "epoch": 1.8891613797515594, "grad_norm": 1.6435120105743408, "learning_rate": 2.9132000000000003e-06, "loss": 0.0278, "step": 35435 }, { "epoch": 1.8892146931812124, "grad_norm": 1.4024296998977661, "learning_rate": 2.9130000000000003e-06, "loss": 0.0058, "step": 35436 }, { "epoch": 1.8892680066108651, "grad_norm": 1.6066094636917114, "learning_rate": 2.9128e-06, "loss": 0.0089, "step": 35437 }, { "epoch": 1.8893213200405183, "grad_norm": 1.6026792526245117, "learning_rate": 2.9126000000000003e-06, "loss": 0.0276, "step": 35438 }, { "epoch": 1.889374633470171, "grad_norm": 1.6983036994934082, "learning_rate": 2.9124000000000003e-06, "loss": 0.0119, "step": 35439 }, { "epoch": 1.889427946899824, "grad_norm": 0.3173117935657501, "learning_rate": 2.9122e-06, "loss": 0.0029, "step": 35440 }, { "epoch": 1.889481260329477, "grad_norm": 2.324876070022583, "learning_rate": 2.9120000000000002e-06, "loss": 0.0355, "step": 35441 }, { "epoch": 1.8895345737591298, "grad_norm": 1.103387713432312, "learning_rate": 2.9118000000000002e-06, "loss": 0.0112, "step": 35442 }, { "epoch": 1.889587887188783, "grad_norm": 2.0785398483276367, "learning_rate": 2.9115999999999998e-06, "loss": 0.0143, "step": 35443 }, { "epoch": 1.8896412006184358, "grad_norm": 1.9022700786590576, "learning_rate": 2.9114e-06, "loss": 0.0103, "step": 35444 }, { "epoch": 1.8896945140480887, "grad_norm": 1.1303645372390747, "learning_rate": 2.9112e-06, "loss": 0.0076, "step": 35445 }, { "epoch": 1.8897478274777417, "grad_norm": 0.5096498727798462, "learning_rate": 2.9110000000000006e-06, "loss": 0.0041, "step": 35446 }, { "epoch": 1.8898011409073945, "grad_norm": 0.7025494575500488, "learning_rate": 2.9108e-06, "loss": 0.0062, "step": 35447 }, { "epoch": 1.8898544543370475, "grad_norm": 0.7806470990180969, "learning_rate": 2.9106e-06, "loss": 0.011, "step": 35448 }, { "epoch": 1.8899077677667004, "grad_norm": 2.271355152130127, "learning_rate": 2.9104000000000005e-06, "loss": 0.0417, "step": 35449 }, { "epoch": 1.8899610811963532, "grad_norm": 3.121319055557251, "learning_rate": 2.9102e-06, "loss": 0.0096, "step": 35450 }, { "epoch": 1.8900143946260064, "grad_norm": 1.624709129333496, "learning_rate": 2.91e-06, "loss": 0.021, "step": 35451 }, { "epoch": 1.8900677080556592, "grad_norm": 2.2862839698791504, "learning_rate": 2.9098000000000005e-06, "loss": 0.0133, "step": 35452 }, { "epoch": 1.8901210214853121, "grad_norm": 0.9816376566886902, "learning_rate": 2.9096e-06, "loss": 0.0113, "step": 35453 }, { "epoch": 1.8901743349149651, "grad_norm": 5.013660430908203, "learning_rate": 2.9094e-06, "loss": 0.0525, "step": 35454 }, { "epoch": 1.8902276483446179, "grad_norm": 1.2316370010375977, "learning_rate": 2.9092000000000005e-06, "loss": 0.0055, "step": 35455 }, { "epoch": 1.890280961774271, "grad_norm": 0.8504544496536255, "learning_rate": 2.909e-06, "loss": 0.0055, "step": 35456 }, { "epoch": 1.8903342752039238, "grad_norm": 1.6832795143127441, "learning_rate": 2.9088000000000004e-06, "loss": 0.0211, "step": 35457 }, { "epoch": 1.8903875886335768, "grad_norm": 1.426499605178833, "learning_rate": 2.9086000000000004e-06, "loss": 0.0248, "step": 35458 }, { "epoch": 1.8904409020632298, "grad_norm": 1.6600919961929321, "learning_rate": 2.9084e-06, "loss": 0.0257, "step": 35459 }, { "epoch": 1.8904942154928825, "grad_norm": 1.1551343202590942, "learning_rate": 2.9082000000000004e-06, "loss": 0.007, "step": 35460 }, { "epoch": 1.8905475289225357, "grad_norm": 1.3428574800491333, "learning_rate": 2.9080000000000004e-06, "loss": 0.0329, "step": 35461 }, { "epoch": 1.8906008423521885, "grad_norm": 0.3913615942001343, "learning_rate": 2.9078e-06, "loss": 0.0043, "step": 35462 }, { "epoch": 1.8906541557818415, "grad_norm": 1.6026290655136108, "learning_rate": 2.9076000000000003e-06, "loss": 0.0223, "step": 35463 }, { "epoch": 1.8907074692114945, "grad_norm": 1.5749921798706055, "learning_rate": 2.9074000000000003e-06, "loss": 0.0078, "step": 35464 }, { "epoch": 1.8907607826411472, "grad_norm": 3.365285873413086, "learning_rate": 2.9072e-06, "loss": 0.0535, "step": 35465 }, { "epoch": 1.8908140960708002, "grad_norm": 1.8414868116378784, "learning_rate": 2.9070000000000003e-06, "loss": 0.0262, "step": 35466 }, { "epoch": 1.8908674095004532, "grad_norm": 1.085056185722351, "learning_rate": 2.9068000000000003e-06, "loss": 0.0072, "step": 35467 }, { "epoch": 1.890920722930106, "grad_norm": 1.8738398551940918, "learning_rate": 2.9066e-06, "loss": 0.0145, "step": 35468 }, { "epoch": 1.8909740363597591, "grad_norm": 2.0918798446655273, "learning_rate": 2.9064000000000002e-06, "loss": 0.0237, "step": 35469 }, { "epoch": 1.891027349789412, "grad_norm": 0.8178650736808777, "learning_rate": 2.9062000000000002e-06, "loss": 0.0052, "step": 35470 }, { "epoch": 1.8910806632190649, "grad_norm": 3.0425570011138916, "learning_rate": 2.9060000000000006e-06, "loss": 0.0296, "step": 35471 }, { "epoch": 1.8911339766487179, "grad_norm": 0.355746865272522, "learning_rate": 2.9058e-06, "loss": 0.0029, "step": 35472 }, { "epoch": 1.8911872900783706, "grad_norm": 1.343483567237854, "learning_rate": 2.9056e-06, "loss": 0.0086, "step": 35473 }, { "epoch": 1.8912406035080238, "grad_norm": 2.052720546722412, "learning_rate": 2.9054000000000006e-06, "loss": 0.0393, "step": 35474 }, { "epoch": 1.8912939169376766, "grad_norm": 1.1396269798278809, "learning_rate": 2.9052e-06, "loss": 0.0191, "step": 35475 }, { "epoch": 1.8913472303673295, "grad_norm": 2.1893668174743652, "learning_rate": 2.905e-06, "loss": 0.0149, "step": 35476 }, { "epoch": 1.8914005437969825, "grad_norm": 1.6031737327575684, "learning_rate": 2.9048000000000005e-06, "loss": 0.0333, "step": 35477 }, { "epoch": 1.8914538572266353, "grad_norm": 1.3271528482437134, "learning_rate": 2.9046e-06, "loss": 0.0139, "step": 35478 }, { "epoch": 1.8915071706562885, "grad_norm": 1.4404922723770142, "learning_rate": 2.9044e-06, "loss": 0.0103, "step": 35479 }, { "epoch": 1.8915604840859412, "grad_norm": 2.5184428691864014, "learning_rate": 2.9042000000000005e-06, "loss": 0.0245, "step": 35480 }, { "epoch": 1.8916137975155942, "grad_norm": 1.8665357828140259, "learning_rate": 2.904e-06, "loss": 0.0185, "step": 35481 }, { "epoch": 1.8916671109452472, "grad_norm": 1.87025785446167, "learning_rate": 2.9038000000000005e-06, "loss": 0.0253, "step": 35482 }, { "epoch": 1.8917204243749, "grad_norm": 1.2811044454574585, "learning_rate": 2.9036000000000005e-06, "loss": 0.0216, "step": 35483 }, { "epoch": 1.891773737804553, "grad_norm": 1.7729593515396118, "learning_rate": 2.9034e-06, "loss": 0.0205, "step": 35484 }, { "epoch": 1.891827051234206, "grad_norm": 2.0757904052734375, "learning_rate": 2.9032000000000004e-06, "loss": 0.0136, "step": 35485 }, { "epoch": 1.8918803646638587, "grad_norm": 1.4289237260818481, "learning_rate": 2.9030000000000004e-06, "loss": 0.0107, "step": 35486 }, { "epoch": 1.8919336780935119, "grad_norm": 1.9633477926254272, "learning_rate": 2.9028e-06, "loss": 0.0116, "step": 35487 }, { "epoch": 1.8919869915231646, "grad_norm": 0.5233371257781982, "learning_rate": 2.9026000000000004e-06, "loss": 0.0046, "step": 35488 }, { "epoch": 1.8920403049528176, "grad_norm": 2.5290794372558594, "learning_rate": 2.9024000000000004e-06, "loss": 0.0307, "step": 35489 }, { "epoch": 1.8920936183824706, "grad_norm": 2.261106014251709, "learning_rate": 2.9022e-06, "loss": 0.0361, "step": 35490 }, { "epoch": 1.8921469318121233, "grad_norm": 1.15287184715271, "learning_rate": 2.9020000000000003e-06, "loss": 0.0075, "step": 35491 }, { "epoch": 1.8922002452417765, "grad_norm": 2.638662815093994, "learning_rate": 2.9018000000000003e-06, "loss": 0.0275, "step": 35492 }, { "epoch": 1.8922535586714293, "grad_norm": 1.7717880010604858, "learning_rate": 2.9016e-06, "loss": 0.0142, "step": 35493 }, { "epoch": 1.8923068721010823, "grad_norm": 2.468858242034912, "learning_rate": 2.9014000000000003e-06, "loss": 0.0188, "step": 35494 }, { "epoch": 1.8923601855307353, "grad_norm": 1.7334657907485962, "learning_rate": 2.9012000000000003e-06, "loss": 0.0093, "step": 35495 }, { "epoch": 1.892413498960388, "grad_norm": 2.2074437141418457, "learning_rate": 2.9010000000000007e-06, "loss": 0.0432, "step": 35496 }, { "epoch": 1.892466812390041, "grad_norm": 0.7814084887504578, "learning_rate": 2.9008000000000002e-06, "loss": 0.0221, "step": 35497 }, { "epoch": 1.892520125819694, "grad_norm": 1.459797978401184, "learning_rate": 2.9006000000000002e-06, "loss": 0.0312, "step": 35498 }, { "epoch": 1.892573439249347, "grad_norm": 1.7278380393981934, "learning_rate": 2.9004000000000006e-06, "loss": 0.0117, "step": 35499 }, { "epoch": 1.892626752679, "grad_norm": 2.071765422821045, "learning_rate": 2.9002e-06, "loss": 0.0374, "step": 35500 }, { "epoch": 1.8926800661086527, "grad_norm": 2.697551965713501, "learning_rate": 2.9e-06, "loss": 0.0312, "step": 35501 }, { "epoch": 1.8927333795383057, "grad_norm": 2.119210958480835, "learning_rate": 2.8998000000000006e-06, "loss": 0.014, "step": 35502 }, { "epoch": 1.8927866929679587, "grad_norm": 2.7109384536743164, "learning_rate": 2.8996e-06, "loss": 0.0418, "step": 35503 }, { "epoch": 1.8928400063976114, "grad_norm": 2.782184600830078, "learning_rate": 2.8994e-06, "loss": 0.0332, "step": 35504 }, { "epoch": 1.8928933198272646, "grad_norm": 1.293047308921814, "learning_rate": 2.8992000000000005e-06, "loss": 0.0204, "step": 35505 }, { "epoch": 1.8929466332569174, "grad_norm": 2.360515594482422, "learning_rate": 2.899e-06, "loss": 0.0236, "step": 35506 }, { "epoch": 1.8929999466865703, "grad_norm": 3.792363405227661, "learning_rate": 2.8988000000000005e-06, "loss": 0.0235, "step": 35507 }, { "epoch": 1.8930532601162233, "grad_norm": 1.1871284246444702, "learning_rate": 2.8986000000000005e-06, "loss": 0.0093, "step": 35508 }, { "epoch": 1.893106573545876, "grad_norm": 1.3021471500396729, "learning_rate": 2.8984e-06, "loss": 0.0081, "step": 35509 }, { "epoch": 1.8931598869755293, "grad_norm": 4.3537516593933105, "learning_rate": 2.8982000000000005e-06, "loss": 0.025, "step": 35510 }, { "epoch": 1.893213200405182, "grad_norm": 1.8177270889282227, "learning_rate": 2.8980000000000005e-06, "loss": 0.0187, "step": 35511 }, { "epoch": 1.893266513834835, "grad_norm": 1.680026888847351, "learning_rate": 2.8978e-06, "loss": 0.0391, "step": 35512 }, { "epoch": 1.893319827264488, "grad_norm": 1.6575103998184204, "learning_rate": 2.8976000000000004e-06, "loss": 0.0191, "step": 35513 }, { "epoch": 1.8933731406941408, "grad_norm": 1.5663973093032837, "learning_rate": 2.8974000000000004e-06, "loss": 0.011, "step": 35514 }, { "epoch": 1.8934264541237937, "grad_norm": 1.6900454759597778, "learning_rate": 2.8972e-06, "loss": 0.0376, "step": 35515 }, { "epoch": 1.8934797675534467, "grad_norm": 2.124610424041748, "learning_rate": 2.8970000000000004e-06, "loss": 0.0283, "step": 35516 }, { "epoch": 1.8935330809830997, "grad_norm": 0.5207533240318298, "learning_rate": 2.8968000000000004e-06, "loss": 0.0021, "step": 35517 }, { "epoch": 1.8935863944127527, "grad_norm": 1.4869325160980225, "learning_rate": 2.8966e-06, "loss": 0.0111, "step": 35518 }, { "epoch": 1.8936397078424054, "grad_norm": 2.830423593521118, "learning_rate": 2.8964000000000003e-06, "loss": 0.0255, "step": 35519 }, { "epoch": 1.8936930212720584, "grad_norm": 1.4667413234710693, "learning_rate": 2.8962000000000003e-06, "loss": 0.024, "step": 35520 }, { "epoch": 1.8937463347017114, "grad_norm": 1.527057409286499, "learning_rate": 2.8960000000000003e-06, "loss": 0.0066, "step": 35521 }, { "epoch": 1.8937996481313641, "grad_norm": 2.64227032661438, "learning_rate": 2.8958000000000003e-06, "loss": 0.0317, "step": 35522 }, { "epoch": 1.8938529615610173, "grad_norm": 0.48166900873184204, "learning_rate": 2.8956000000000003e-06, "loss": 0.0044, "step": 35523 }, { "epoch": 1.89390627499067, "grad_norm": 1.2724930047988892, "learning_rate": 2.8954000000000003e-06, "loss": 0.0071, "step": 35524 }, { "epoch": 1.893959588420323, "grad_norm": 3.2899017333984375, "learning_rate": 2.8952000000000002e-06, "loss": 0.0464, "step": 35525 }, { "epoch": 1.894012901849976, "grad_norm": 2.5048654079437256, "learning_rate": 2.8950000000000002e-06, "loss": 0.0355, "step": 35526 }, { "epoch": 1.8940662152796288, "grad_norm": 1.1996643543243408, "learning_rate": 2.8948000000000002e-06, "loss": 0.017, "step": 35527 }, { "epoch": 1.894119528709282, "grad_norm": 1.2152279615402222, "learning_rate": 2.8946e-06, "loss": 0.0183, "step": 35528 }, { "epoch": 1.8941728421389348, "grad_norm": 1.3506979942321777, "learning_rate": 2.8944e-06, "loss": 0.0229, "step": 35529 }, { "epoch": 1.8942261555685878, "grad_norm": 2.726567268371582, "learning_rate": 2.8942e-06, "loss": 0.0397, "step": 35530 }, { "epoch": 1.8942794689982407, "grad_norm": 1.511999249458313, "learning_rate": 2.894e-06, "loss": 0.0187, "step": 35531 }, { "epoch": 1.8943327824278935, "grad_norm": 1.3118904829025269, "learning_rate": 2.8938000000000006e-06, "loss": 0.011, "step": 35532 }, { "epoch": 1.8943860958575465, "grad_norm": 1.6915098428726196, "learning_rate": 2.8936e-06, "loss": 0.0123, "step": 35533 }, { "epoch": 1.8944394092871994, "grad_norm": 1.5851714611053467, "learning_rate": 2.8934e-06, "loss": 0.0144, "step": 35534 }, { "epoch": 1.8944927227168524, "grad_norm": 0.9733940958976746, "learning_rate": 2.8932000000000005e-06, "loss": 0.0081, "step": 35535 }, { "epoch": 1.8945460361465054, "grad_norm": 2.4586775302886963, "learning_rate": 2.893e-06, "loss": 0.0327, "step": 35536 }, { "epoch": 1.8945993495761582, "grad_norm": 1.5261298418045044, "learning_rate": 2.8928e-06, "loss": 0.0223, "step": 35537 }, { "epoch": 1.8946526630058111, "grad_norm": 1.3846570253372192, "learning_rate": 2.8926000000000005e-06, "loss": 0.0184, "step": 35538 }, { "epoch": 1.8947059764354641, "grad_norm": 0.9612830877304077, "learning_rate": 2.8924e-06, "loss": 0.0189, "step": 35539 }, { "epoch": 1.8947592898651169, "grad_norm": 1.1816262006759644, "learning_rate": 2.8922e-06, "loss": 0.0126, "step": 35540 }, { "epoch": 1.89481260329477, "grad_norm": 2.2263784408569336, "learning_rate": 2.8920000000000004e-06, "loss": 0.0242, "step": 35541 }, { "epoch": 1.8948659167244228, "grad_norm": 2.4361636638641357, "learning_rate": 2.8918e-06, "loss": 0.0226, "step": 35542 }, { "epoch": 1.8949192301540758, "grad_norm": 1.4000526666641235, "learning_rate": 2.8916e-06, "loss": 0.0111, "step": 35543 }, { "epoch": 1.8949725435837288, "grad_norm": 2.14332914352417, "learning_rate": 2.8914000000000004e-06, "loss": 0.0248, "step": 35544 }, { "epoch": 1.8950258570133816, "grad_norm": 1.1928625106811523, "learning_rate": 2.8912e-06, "loss": 0.0106, "step": 35545 }, { "epoch": 1.8950791704430348, "grad_norm": 2.7134859561920166, "learning_rate": 2.8910000000000004e-06, "loss": 0.0339, "step": 35546 }, { "epoch": 1.8951324838726875, "grad_norm": 1.5296889543533325, "learning_rate": 2.8908000000000003e-06, "loss": 0.0269, "step": 35547 }, { "epoch": 1.8951857973023405, "grad_norm": 1.5386396646499634, "learning_rate": 2.8906e-06, "loss": 0.0166, "step": 35548 }, { "epoch": 1.8952391107319935, "grad_norm": 4.6598920822143555, "learning_rate": 2.8904000000000003e-06, "loss": 0.0244, "step": 35549 }, { "epoch": 1.8952924241616462, "grad_norm": 1.6451358795166016, "learning_rate": 2.8902000000000003e-06, "loss": 0.0236, "step": 35550 }, { "epoch": 1.8953457375912992, "grad_norm": 1.0673009157180786, "learning_rate": 2.89e-06, "loss": 0.0088, "step": 35551 }, { "epoch": 1.8953990510209522, "grad_norm": 1.0205914974212646, "learning_rate": 2.8898000000000003e-06, "loss": 0.0111, "step": 35552 }, { "epoch": 1.895452364450605, "grad_norm": 1.415065050125122, "learning_rate": 2.8896000000000003e-06, "loss": 0.0113, "step": 35553 }, { "epoch": 1.8955056778802581, "grad_norm": 1.7247453927993774, "learning_rate": 2.8894e-06, "loss": 0.0116, "step": 35554 }, { "epoch": 1.895558991309911, "grad_norm": 1.9621639251708984, "learning_rate": 2.8892000000000002e-06, "loss": 0.0155, "step": 35555 }, { "epoch": 1.8956123047395639, "grad_norm": 0.404876172542572, "learning_rate": 2.889e-06, "loss": 0.0038, "step": 35556 }, { "epoch": 1.8956656181692169, "grad_norm": 2.3577771186828613, "learning_rate": 2.8888000000000006e-06, "loss": 0.0214, "step": 35557 }, { "epoch": 1.8957189315988696, "grad_norm": 3.3252933025360107, "learning_rate": 2.8886e-06, "loss": 0.0527, "step": 35558 }, { "epoch": 1.8957722450285228, "grad_norm": 2.0174412727355957, "learning_rate": 2.8884e-06, "loss": 0.0176, "step": 35559 }, { "epoch": 1.8958255584581756, "grad_norm": 1.2934818267822266, "learning_rate": 2.8882000000000006e-06, "loss": 0.0063, "step": 35560 }, { "epoch": 1.8958788718878286, "grad_norm": 1.6221041679382324, "learning_rate": 2.888e-06, "loss": 0.0202, "step": 35561 }, { "epoch": 1.8959321853174815, "grad_norm": 1.3716169595718384, "learning_rate": 2.8878e-06, "loss": 0.0229, "step": 35562 }, { "epoch": 1.8959854987471343, "grad_norm": 2.2838685512542725, "learning_rate": 2.8876000000000005e-06, "loss": 0.0268, "step": 35563 }, { "epoch": 1.8960388121767875, "grad_norm": 1.7630341053009033, "learning_rate": 2.8874e-06, "loss": 0.0118, "step": 35564 }, { "epoch": 1.8960921256064402, "grad_norm": 3.1848700046539307, "learning_rate": 2.8872e-06, "loss": 0.0228, "step": 35565 }, { "epoch": 1.8961454390360932, "grad_norm": 2.1184284687042236, "learning_rate": 2.8870000000000005e-06, "loss": 0.0259, "step": 35566 }, { "epoch": 1.8961987524657462, "grad_norm": 4.120291233062744, "learning_rate": 2.8868e-06, "loss": 0.0191, "step": 35567 }, { "epoch": 1.896252065895399, "grad_norm": 3.73048734664917, "learning_rate": 2.8866000000000005e-06, "loss": 0.0452, "step": 35568 }, { "epoch": 1.896305379325052, "grad_norm": 2.7110729217529297, "learning_rate": 2.8864000000000004e-06, "loss": 0.0303, "step": 35569 }, { "epoch": 1.896358692754705, "grad_norm": 2.875657796859741, "learning_rate": 2.8862e-06, "loss": 0.0187, "step": 35570 }, { "epoch": 1.8964120061843577, "grad_norm": 1.2497209310531616, "learning_rate": 2.8860000000000004e-06, "loss": 0.0121, "step": 35571 }, { "epoch": 1.8964653196140109, "grad_norm": 2.7895779609680176, "learning_rate": 2.8858000000000004e-06, "loss": 0.0302, "step": 35572 }, { "epoch": 1.8965186330436636, "grad_norm": 1.6595076322555542, "learning_rate": 2.8856e-06, "loss": 0.0137, "step": 35573 }, { "epoch": 1.8965719464733166, "grad_norm": 2.6453301906585693, "learning_rate": 2.8854000000000004e-06, "loss": 0.0502, "step": 35574 }, { "epoch": 1.8966252599029696, "grad_norm": 1.8934801816940308, "learning_rate": 2.8852000000000003e-06, "loss": 0.0288, "step": 35575 }, { "epoch": 1.8966785733326224, "grad_norm": 0.8795080780982971, "learning_rate": 2.885e-06, "loss": 0.0105, "step": 35576 }, { "epoch": 1.8967318867622756, "grad_norm": 3.0503957271575928, "learning_rate": 2.8848000000000003e-06, "loss": 0.0355, "step": 35577 }, { "epoch": 1.8967852001919283, "grad_norm": 1.5289850234985352, "learning_rate": 2.8846000000000003e-06, "loss": 0.0257, "step": 35578 }, { "epoch": 1.8968385136215813, "grad_norm": 1.5241777896881104, "learning_rate": 2.8844e-06, "loss": 0.0128, "step": 35579 }, { "epoch": 1.8968918270512343, "grad_norm": 1.555587649345398, "learning_rate": 2.8842000000000003e-06, "loss": 0.008, "step": 35580 }, { "epoch": 1.896945140480887, "grad_norm": 1.9810935258865356, "learning_rate": 2.8840000000000003e-06, "loss": 0.0295, "step": 35581 }, { "epoch": 1.8969984539105402, "grad_norm": 3.442054033279419, "learning_rate": 2.8838000000000007e-06, "loss": 0.0143, "step": 35582 }, { "epoch": 1.897051767340193, "grad_norm": 1.2233943939208984, "learning_rate": 2.8836000000000002e-06, "loss": 0.0085, "step": 35583 }, { "epoch": 1.897105080769846, "grad_norm": 3.7854788303375244, "learning_rate": 2.8834e-06, "loss": 0.0533, "step": 35584 }, { "epoch": 1.897158394199499, "grad_norm": 2.31522536277771, "learning_rate": 2.8832000000000006e-06, "loss": 0.0281, "step": 35585 }, { "epoch": 1.8972117076291517, "grad_norm": 2.1430516242980957, "learning_rate": 2.883e-06, "loss": 0.0169, "step": 35586 }, { "epoch": 1.8972650210588047, "grad_norm": 1.9564439058303833, "learning_rate": 2.8828e-06, "loss": 0.0254, "step": 35587 }, { "epoch": 1.8973183344884577, "grad_norm": 3.1680490970611572, "learning_rate": 2.8826000000000006e-06, "loss": 0.0567, "step": 35588 }, { "epoch": 1.8973716479181104, "grad_norm": 2.6804800033569336, "learning_rate": 2.8824e-06, "loss": 0.0203, "step": 35589 }, { "epoch": 1.8974249613477636, "grad_norm": 1.3971689939498901, "learning_rate": 2.8822e-06, "loss": 0.0183, "step": 35590 }, { "epoch": 1.8974782747774164, "grad_norm": 1.2505687475204468, "learning_rate": 2.8820000000000005e-06, "loss": 0.0063, "step": 35591 }, { "epoch": 1.8975315882070694, "grad_norm": 2.0730810165405273, "learning_rate": 2.8818e-06, "loss": 0.0225, "step": 35592 }, { "epoch": 1.8975849016367223, "grad_norm": 0.9056774377822876, "learning_rate": 2.8816000000000005e-06, "loss": 0.0198, "step": 35593 }, { "epoch": 1.897638215066375, "grad_norm": 1.4382883310317993, "learning_rate": 2.8814000000000005e-06, "loss": 0.0219, "step": 35594 }, { "epoch": 1.8976915284960283, "grad_norm": 4.017679214477539, "learning_rate": 2.8812e-06, "loss": 0.0537, "step": 35595 }, { "epoch": 1.897744841925681, "grad_norm": 1.4148938655853271, "learning_rate": 2.8810000000000005e-06, "loss": 0.0098, "step": 35596 }, { "epoch": 1.897798155355334, "grad_norm": 1.8663967847824097, "learning_rate": 2.8808000000000004e-06, "loss": 0.0118, "step": 35597 }, { "epoch": 1.897851468784987, "grad_norm": 1.7777048349380493, "learning_rate": 2.8806e-06, "loss": 0.0306, "step": 35598 }, { "epoch": 1.8979047822146398, "grad_norm": 2.4699859619140625, "learning_rate": 2.8804000000000004e-06, "loss": 0.0225, "step": 35599 }, { "epoch": 1.8979580956442927, "grad_norm": 1.3699722290039062, "learning_rate": 2.8802000000000004e-06, "loss": 0.0137, "step": 35600 }, { "epoch": 1.8980114090739457, "grad_norm": 2.6212470531463623, "learning_rate": 2.88e-06, "loss": 0.0375, "step": 35601 }, { "epoch": 1.8980647225035987, "grad_norm": 1.9576950073242188, "learning_rate": 2.8798000000000004e-06, "loss": 0.0127, "step": 35602 }, { "epoch": 1.8981180359332517, "grad_norm": 1.25632905960083, "learning_rate": 2.8796000000000003e-06, "loss": 0.0175, "step": 35603 }, { "epoch": 1.8981713493629044, "grad_norm": 1.1831223964691162, "learning_rate": 2.8794e-06, "loss": 0.0141, "step": 35604 }, { "epoch": 1.8982246627925574, "grad_norm": 5.67254114151001, "learning_rate": 2.8792000000000003e-06, "loss": 0.036, "step": 35605 }, { "epoch": 1.8982779762222104, "grad_norm": 1.660468578338623, "learning_rate": 2.8790000000000003e-06, "loss": 0.0212, "step": 35606 }, { "epoch": 1.8983312896518632, "grad_norm": 0.3639134168624878, "learning_rate": 2.8788000000000003e-06, "loss": 0.0038, "step": 35607 }, { "epoch": 1.8983846030815164, "grad_norm": 1.5712931156158447, "learning_rate": 2.8786000000000003e-06, "loss": 0.0308, "step": 35608 }, { "epoch": 1.898437916511169, "grad_norm": 1.4065934419631958, "learning_rate": 2.8784000000000003e-06, "loss": 0.0148, "step": 35609 }, { "epoch": 1.898491229940822, "grad_norm": 2.1900408267974854, "learning_rate": 2.8782000000000002e-06, "loss": 0.0266, "step": 35610 }, { "epoch": 1.898544543370475, "grad_norm": 1.4290125370025635, "learning_rate": 2.8780000000000002e-06, "loss": 0.024, "step": 35611 }, { "epoch": 1.8985978568001278, "grad_norm": 1.1102899312973022, "learning_rate": 2.8778000000000002e-06, "loss": 0.0204, "step": 35612 }, { "epoch": 1.898651170229781, "grad_norm": 0.6083199381828308, "learning_rate": 2.8776e-06, "loss": 0.005, "step": 35613 }, { "epoch": 1.8987044836594338, "grad_norm": 0.7525560855865479, "learning_rate": 2.8774e-06, "loss": 0.0065, "step": 35614 }, { "epoch": 1.8987577970890868, "grad_norm": 2.79537296295166, "learning_rate": 2.8772e-06, "loss": 0.0306, "step": 35615 }, { "epoch": 1.8988111105187397, "grad_norm": 2.2261600494384766, "learning_rate": 2.877e-06, "loss": 0.0107, "step": 35616 }, { "epoch": 1.8988644239483925, "grad_norm": 1.3434032201766968, "learning_rate": 2.8768e-06, "loss": 0.014, "step": 35617 }, { "epoch": 1.8989177373780455, "grad_norm": 1.3187503814697266, "learning_rate": 2.8766000000000005e-06, "loss": 0.0217, "step": 35618 }, { "epoch": 1.8989710508076985, "grad_norm": 1.4114274978637695, "learning_rate": 2.8764e-06, "loss": 0.0123, "step": 35619 }, { "epoch": 1.8990243642373514, "grad_norm": 0.6850783824920654, "learning_rate": 2.8762e-06, "loss": 0.0045, "step": 35620 }, { "epoch": 1.8990776776670044, "grad_norm": 0.3445276618003845, "learning_rate": 2.8760000000000005e-06, "loss": 0.0033, "step": 35621 }, { "epoch": 1.8991309910966572, "grad_norm": 1.272650957107544, "learning_rate": 2.8758e-06, "loss": 0.0084, "step": 35622 }, { "epoch": 1.8991843045263102, "grad_norm": 1.5406665802001953, "learning_rate": 2.8756e-06, "loss": 0.0222, "step": 35623 }, { "epoch": 1.8992376179559631, "grad_norm": 1.5615665912628174, "learning_rate": 2.8754000000000005e-06, "loss": 0.0091, "step": 35624 }, { "epoch": 1.8992909313856159, "grad_norm": 2.247921943664551, "learning_rate": 2.8752e-06, "loss": 0.0297, "step": 35625 }, { "epoch": 1.899344244815269, "grad_norm": 2.9018383026123047, "learning_rate": 2.875e-06, "loss": 0.0371, "step": 35626 }, { "epoch": 1.8993975582449218, "grad_norm": 0.3930249810218811, "learning_rate": 2.8748000000000004e-06, "loss": 0.0033, "step": 35627 }, { "epoch": 1.8994508716745748, "grad_norm": 2.038350820541382, "learning_rate": 2.8746e-06, "loss": 0.0224, "step": 35628 }, { "epoch": 1.8995041851042278, "grad_norm": 2.427785634994507, "learning_rate": 2.8744e-06, "loss": 0.0123, "step": 35629 }, { "epoch": 1.8995574985338806, "grad_norm": 1.5599766969680786, "learning_rate": 2.8742000000000004e-06, "loss": 0.0109, "step": 35630 }, { "epoch": 1.8996108119635338, "grad_norm": 0.6966772079467773, "learning_rate": 2.874e-06, "loss": 0.0058, "step": 35631 }, { "epoch": 1.8996641253931865, "grad_norm": 1.561735987663269, "learning_rate": 2.8738000000000003e-06, "loss": 0.008, "step": 35632 }, { "epoch": 1.8997174388228395, "grad_norm": 4.805883407592773, "learning_rate": 2.8736000000000003e-06, "loss": 0.0161, "step": 35633 }, { "epoch": 1.8997707522524925, "grad_norm": 1.5761247873306274, "learning_rate": 2.8734e-06, "loss": 0.0058, "step": 35634 }, { "epoch": 1.8998240656821452, "grad_norm": 1.7502212524414062, "learning_rate": 2.8732000000000003e-06, "loss": 0.0154, "step": 35635 }, { "epoch": 1.8998773791117982, "grad_norm": 2.5533206462860107, "learning_rate": 2.8730000000000003e-06, "loss": 0.0206, "step": 35636 }, { "epoch": 1.8999306925414512, "grad_norm": 3.276254177093506, "learning_rate": 2.8728e-06, "loss": 0.0461, "step": 35637 }, { "epoch": 1.8999840059711042, "grad_norm": 0.3338921368122101, "learning_rate": 2.8726000000000002e-06, "loss": 0.002, "step": 35638 }, { "epoch": 1.9000373194007572, "grad_norm": 1.8902182579040527, "learning_rate": 2.8724000000000002e-06, "loss": 0.0064, "step": 35639 }, { "epoch": 1.90009063283041, "grad_norm": 0.7261807918548584, "learning_rate": 2.8722e-06, "loss": 0.0087, "step": 35640 }, { "epoch": 1.9001439462600629, "grad_norm": 1.3437724113464355, "learning_rate": 2.872e-06, "loss": 0.0234, "step": 35641 }, { "epoch": 1.9001972596897159, "grad_norm": 1.7238755226135254, "learning_rate": 2.8718e-06, "loss": 0.0188, "step": 35642 }, { "epoch": 1.9002505731193686, "grad_norm": 1.0826178789138794, "learning_rate": 2.8716000000000006e-06, "loss": 0.0133, "step": 35643 }, { "epoch": 1.9003038865490218, "grad_norm": 1.0449166297912598, "learning_rate": 2.8714e-06, "loss": 0.0145, "step": 35644 }, { "epoch": 1.9003571999786746, "grad_norm": 1.4500571489334106, "learning_rate": 2.8712e-06, "loss": 0.0103, "step": 35645 }, { "epoch": 1.9004105134083276, "grad_norm": 0.9669681191444397, "learning_rate": 2.8710000000000005e-06, "loss": 0.0186, "step": 35646 }, { "epoch": 1.9004638268379805, "grad_norm": 1.576032042503357, "learning_rate": 2.8708e-06, "loss": 0.0261, "step": 35647 }, { "epoch": 1.9005171402676333, "grad_norm": 1.5711944103240967, "learning_rate": 2.8706e-06, "loss": 0.0391, "step": 35648 }, { "epoch": 1.9005704536972865, "grad_norm": 2.2991631031036377, "learning_rate": 2.8704000000000005e-06, "loss": 0.0219, "step": 35649 }, { "epoch": 1.9006237671269393, "grad_norm": 2.252483367919922, "learning_rate": 2.8702e-06, "loss": 0.0421, "step": 35650 }, { "epoch": 1.9006770805565922, "grad_norm": 0.8304519057273865, "learning_rate": 2.87e-06, "loss": 0.0032, "step": 35651 }, { "epoch": 1.9007303939862452, "grad_norm": 1.1111994981765747, "learning_rate": 2.8698000000000005e-06, "loss": 0.0281, "step": 35652 }, { "epoch": 1.900783707415898, "grad_norm": 1.57956063747406, "learning_rate": 2.8696e-06, "loss": 0.0322, "step": 35653 }, { "epoch": 1.900837020845551, "grad_norm": 0.2917792797088623, "learning_rate": 2.8694e-06, "loss": 0.0032, "step": 35654 }, { "epoch": 1.900890334275204, "grad_norm": 2.0537173748016357, "learning_rate": 2.8692000000000004e-06, "loss": 0.0184, "step": 35655 }, { "epoch": 1.9009436477048567, "grad_norm": 3.7266697883605957, "learning_rate": 2.869e-06, "loss": 0.028, "step": 35656 }, { "epoch": 1.9009969611345099, "grad_norm": 1.4286521673202515, "learning_rate": 2.8688000000000004e-06, "loss": 0.0351, "step": 35657 }, { "epoch": 1.9010502745641626, "grad_norm": 1.8847733736038208, "learning_rate": 2.8686000000000004e-06, "loss": 0.018, "step": 35658 }, { "epoch": 1.9011035879938156, "grad_norm": 3.333786725997925, "learning_rate": 2.8684e-06, "loss": 0.0451, "step": 35659 }, { "epoch": 1.9011569014234686, "grad_norm": 0.8475077152252197, "learning_rate": 2.8682000000000003e-06, "loss": 0.0103, "step": 35660 }, { "epoch": 1.9012102148531214, "grad_norm": 2.663316011428833, "learning_rate": 2.8680000000000003e-06, "loss": 0.0292, "step": 35661 }, { "epoch": 1.9012635282827746, "grad_norm": 2.015796422958374, "learning_rate": 2.8678e-06, "loss": 0.0262, "step": 35662 }, { "epoch": 1.9013168417124273, "grad_norm": 2.789048910140991, "learning_rate": 2.8676000000000003e-06, "loss": 0.0288, "step": 35663 }, { "epoch": 1.9013701551420803, "grad_norm": 1.8725072145462036, "learning_rate": 2.8674000000000003e-06, "loss": 0.0145, "step": 35664 }, { "epoch": 1.9014234685717333, "grad_norm": 0.7451960444450378, "learning_rate": 2.8672e-06, "loss": 0.0054, "step": 35665 }, { "epoch": 1.901476782001386, "grad_norm": 2.9244325160980225, "learning_rate": 2.8670000000000002e-06, "loss": 0.0333, "step": 35666 }, { "epoch": 1.9015300954310392, "grad_norm": 0.9853352308273315, "learning_rate": 2.8668000000000002e-06, "loss": 0.0181, "step": 35667 }, { "epoch": 1.901583408860692, "grad_norm": 3.4721426963806152, "learning_rate": 2.8666000000000006e-06, "loss": 0.0576, "step": 35668 }, { "epoch": 1.901636722290345, "grad_norm": 2.2722113132476807, "learning_rate": 2.8664e-06, "loss": 0.0277, "step": 35669 }, { "epoch": 1.901690035719998, "grad_norm": 1.8483389616012573, "learning_rate": 2.8662e-06, "loss": 0.0062, "step": 35670 }, { "epoch": 1.9017433491496507, "grad_norm": 0.6647769212722778, "learning_rate": 2.8660000000000006e-06, "loss": 0.0063, "step": 35671 }, { "epoch": 1.9017966625793037, "grad_norm": 1.5300363302230835, "learning_rate": 2.8658e-06, "loss": 0.0189, "step": 35672 }, { "epoch": 1.9018499760089567, "grad_norm": 0.8312861919403076, "learning_rate": 2.8656e-06, "loss": 0.0091, "step": 35673 }, { "epoch": 1.9019032894386094, "grad_norm": 1.5328565835952759, "learning_rate": 2.8654000000000006e-06, "loss": 0.017, "step": 35674 }, { "epoch": 1.9019566028682626, "grad_norm": 2.5432071685791016, "learning_rate": 2.8652e-06, "loss": 0.0386, "step": 35675 }, { "epoch": 1.9020099162979154, "grad_norm": 0.44617006182670593, "learning_rate": 2.865e-06, "loss": 0.0044, "step": 35676 }, { "epoch": 1.9020632297275684, "grad_norm": 1.6351938247680664, "learning_rate": 2.8648000000000005e-06, "loss": 0.0156, "step": 35677 }, { "epoch": 1.9021165431572213, "grad_norm": 1.99537992477417, "learning_rate": 2.8646e-06, "loss": 0.0195, "step": 35678 }, { "epoch": 1.902169856586874, "grad_norm": 1.4657407999038696, "learning_rate": 2.8644e-06, "loss": 0.0072, "step": 35679 }, { "epoch": 1.9022231700165273, "grad_norm": 2.4082446098327637, "learning_rate": 2.8642000000000005e-06, "loss": 0.033, "step": 35680 }, { "epoch": 1.90227648344618, "grad_norm": 2.2391271591186523, "learning_rate": 2.864e-06, "loss": 0.0358, "step": 35681 }, { "epoch": 1.902329796875833, "grad_norm": 0.5411193370819092, "learning_rate": 2.8638000000000004e-06, "loss": 0.0039, "step": 35682 }, { "epoch": 1.902383110305486, "grad_norm": 2.2671101093292236, "learning_rate": 2.8636000000000004e-06, "loss": 0.0449, "step": 35683 }, { "epoch": 1.9024364237351388, "grad_norm": 3.342207193374634, "learning_rate": 2.8634e-06, "loss": 0.0395, "step": 35684 }, { "epoch": 1.902489737164792, "grad_norm": 1.0371956825256348, "learning_rate": 2.8632000000000004e-06, "loss": 0.0059, "step": 35685 }, { "epoch": 1.9025430505944447, "grad_norm": 2.446709632873535, "learning_rate": 2.8630000000000004e-06, "loss": 0.042, "step": 35686 }, { "epoch": 1.9025963640240977, "grad_norm": 2.6204192638397217, "learning_rate": 2.8628e-06, "loss": 0.0249, "step": 35687 }, { "epoch": 1.9026496774537507, "grad_norm": 1.2938810586929321, "learning_rate": 2.8626000000000003e-06, "loss": 0.0128, "step": 35688 }, { "epoch": 1.9027029908834034, "grad_norm": 1.464173674583435, "learning_rate": 2.8624000000000003e-06, "loss": 0.0091, "step": 35689 }, { "epoch": 1.9027563043130564, "grad_norm": 1.7532345056533813, "learning_rate": 2.8622e-06, "loss": 0.035, "step": 35690 }, { "epoch": 1.9028096177427094, "grad_norm": 1.1653625965118408, "learning_rate": 2.8620000000000003e-06, "loss": 0.0131, "step": 35691 }, { "epoch": 1.9028629311723622, "grad_norm": 1.6058945655822754, "learning_rate": 2.8618000000000003e-06, "loss": 0.0505, "step": 35692 }, { "epoch": 1.9029162446020154, "grad_norm": 1.9226903915405273, "learning_rate": 2.8616000000000007e-06, "loss": 0.0318, "step": 35693 }, { "epoch": 1.9029695580316681, "grad_norm": 1.9624500274658203, "learning_rate": 2.8614000000000003e-06, "loss": 0.0123, "step": 35694 }, { "epoch": 1.903022871461321, "grad_norm": 2.1083011627197266, "learning_rate": 2.8612000000000002e-06, "loss": 0.0114, "step": 35695 }, { "epoch": 1.903076184890974, "grad_norm": 3.3191680908203125, "learning_rate": 2.8610000000000006e-06, "loss": 0.0235, "step": 35696 }, { "epoch": 1.9031294983206268, "grad_norm": 1.0440564155578613, "learning_rate": 2.8608e-06, "loss": 0.0081, "step": 35697 }, { "epoch": 1.90318281175028, "grad_norm": 2.249464750289917, "learning_rate": 2.8606e-06, "loss": 0.022, "step": 35698 }, { "epoch": 1.9032361251799328, "grad_norm": 2.650341272354126, "learning_rate": 2.8604000000000006e-06, "loss": 0.0137, "step": 35699 }, { "epoch": 1.9032894386095858, "grad_norm": 2.7503812313079834, "learning_rate": 2.8602e-06, "loss": 0.0329, "step": 35700 }, { "epoch": 1.9033427520392387, "grad_norm": 1.902462124824524, "learning_rate": 2.86e-06, "loss": 0.0296, "step": 35701 }, { "epoch": 1.9033960654688915, "grad_norm": 1.3354684114456177, "learning_rate": 2.8598000000000006e-06, "loss": 0.0122, "step": 35702 }, { "epoch": 1.9034493788985447, "grad_norm": 0.9805665612220764, "learning_rate": 2.8596e-06, "loss": 0.0151, "step": 35703 }, { "epoch": 1.9035026923281975, "grad_norm": 0.9590801000595093, "learning_rate": 2.8594000000000005e-06, "loss": 0.0235, "step": 35704 }, { "epoch": 1.9035560057578504, "grad_norm": 3.3272790908813477, "learning_rate": 2.8592000000000005e-06, "loss": 0.0499, "step": 35705 }, { "epoch": 1.9036093191875034, "grad_norm": 1.549665093421936, "learning_rate": 2.859e-06, "loss": 0.0166, "step": 35706 }, { "epoch": 1.9036626326171562, "grad_norm": 1.4125083684921265, "learning_rate": 2.8588000000000005e-06, "loss": 0.0269, "step": 35707 }, { "epoch": 1.9037159460468092, "grad_norm": 0.9732150435447693, "learning_rate": 2.8586000000000005e-06, "loss": 0.0152, "step": 35708 }, { "epoch": 1.9037692594764621, "grad_norm": 0.9305469393730164, "learning_rate": 2.8584e-06, "loss": 0.0101, "step": 35709 }, { "epoch": 1.903822572906115, "grad_norm": 2.3019344806671143, "learning_rate": 2.8582000000000004e-06, "loss": 0.0322, "step": 35710 }, { "epoch": 1.903875886335768, "grad_norm": 2.0790364742279053, "learning_rate": 2.8580000000000004e-06, "loss": 0.0415, "step": 35711 }, { "epoch": 1.9039291997654209, "grad_norm": 1.4091496467590332, "learning_rate": 2.8578e-06, "loss": 0.0082, "step": 35712 }, { "epoch": 1.9039825131950738, "grad_norm": 1.5298010110855103, "learning_rate": 2.8576000000000004e-06, "loss": 0.024, "step": 35713 }, { "epoch": 1.9040358266247268, "grad_norm": 2.405870199203491, "learning_rate": 2.8574000000000004e-06, "loss": 0.0113, "step": 35714 }, { "epoch": 1.9040891400543796, "grad_norm": 0.39943912625312805, "learning_rate": 2.8572e-06, "loss": 0.0052, "step": 35715 }, { "epoch": 1.9041424534840328, "grad_norm": 1.628888726234436, "learning_rate": 2.8570000000000003e-06, "loss": 0.0223, "step": 35716 }, { "epoch": 1.9041957669136855, "grad_norm": 0.8253592848777771, "learning_rate": 2.8568000000000003e-06, "loss": 0.0069, "step": 35717 }, { "epoch": 1.9042490803433385, "grad_norm": 0.9727990031242371, "learning_rate": 2.8566000000000003e-06, "loss": 0.0123, "step": 35718 }, { "epoch": 1.9043023937729915, "grad_norm": 0.6490659117698669, "learning_rate": 2.8564000000000003e-06, "loss": 0.0147, "step": 35719 }, { "epoch": 1.9043557072026442, "grad_norm": 2.1403791904449463, "learning_rate": 2.8562000000000003e-06, "loss": 0.019, "step": 35720 }, { "epoch": 1.9044090206322972, "grad_norm": 4.882648468017578, "learning_rate": 2.8560000000000003e-06, "loss": 0.0331, "step": 35721 }, { "epoch": 1.9044623340619502, "grad_norm": 0.8145787119865417, "learning_rate": 2.8558000000000003e-06, "loss": 0.0058, "step": 35722 }, { "epoch": 1.9045156474916032, "grad_norm": 1.4409772157669067, "learning_rate": 2.8556000000000002e-06, "loss": 0.008, "step": 35723 }, { "epoch": 1.9045689609212562, "grad_norm": 2.566148042678833, "learning_rate": 2.8554000000000002e-06, "loss": 0.0356, "step": 35724 }, { "epoch": 1.904622274350909, "grad_norm": 2.018923759460449, "learning_rate": 2.8552e-06, "loss": 0.0136, "step": 35725 }, { "epoch": 1.904675587780562, "grad_norm": 0.36229565739631653, "learning_rate": 2.855e-06, "loss": 0.0032, "step": 35726 }, { "epoch": 1.9047289012102149, "grad_norm": 2.54587984085083, "learning_rate": 2.8548e-06, "loss": 0.0162, "step": 35727 }, { "epoch": 1.9047822146398676, "grad_norm": 2.5714688301086426, "learning_rate": 2.8546e-06, "loss": 0.0254, "step": 35728 }, { "epoch": 1.9048355280695208, "grad_norm": 2.406770706176758, "learning_rate": 2.8544000000000006e-06, "loss": 0.0421, "step": 35729 }, { "epoch": 1.9048888414991736, "grad_norm": 0.4559333920478821, "learning_rate": 2.8542e-06, "loss": 0.0046, "step": 35730 }, { "epoch": 1.9049421549288266, "grad_norm": 1.598392128944397, "learning_rate": 2.854e-06, "loss": 0.0173, "step": 35731 }, { "epoch": 1.9049954683584795, "grad_norm": 1.1258094310760498, "learning_rate": 2.8538000000000005e-06, "loss": 0.009, "step": 35732 }, { "epoch": 1.9050487817881323, "grad_norm": 0.6277611255645752, "learning_rate": 2.8536e-06, "loss": 0.0047, "step": 35733 }, { "epoch": 1.9051020952177855, "grad_norm": 1.7816883325576782, "learning_rate": 2.8534e-06, "loss": 0.0282, "step": 35734 }, { "epoch": 1.9051554086474383, "grad_norm": 3.0103671550750732, "learning_rate": 2.8532000000000005e-06, "loss": 0.0399, "step": 35735 }, { "epoch": 1.9052087220770912, "grad_norm": 2.220825433731079, "learning_rate": 2.853e-06, "loss": 0.0431, "step": 35736 }, { "epoch": 1.9052620355067442, "grad_norm": 2.599553346633911, "learning_rate": 2.8528e-06, "loss": 0.0209, "step": 35737 }, { "epoch": 1.905315348936397, "grad_norm": 2.4623804092407227, "learning_rate": 2.8526000000000004e-06, "loss": 0.0292, "step": 35738 }, { "epoch": 1.90536866236605, "grad_norm": 2.0294346809387207, "learning_rate": 2.8524e-06, "loss": 0.0399, "step": 35739 }, { "epoch": 1.905421975795703, "grad_norm": 3.604853391647339, "learning_rate": 2.8522e-06, "loss": 0.0408, "step": 35740 }, { "epoch": 1.905475289225356, "grad_norm": 2.017493724822998, "learning_rate": 2.8520000000000004e-06, "loss": 0.0527, "step": 35741 }, { "epoch": 1.905528602655009, "grad_norm": 2.5391156673431396, "learning_rate": 2.8518e-06, "loss": 0.015, "step": 35742 }, { "epoch": 1.9055819160846617, "grad_norm": 1.0472004413604736, "learning_rate": 2.8516000000000004e-06, "loss": 0.0095, "step": 35743 }, { "epoch": 1.9056352295143146, "grad_norm": 1.240780234336853, "learning_rate": 2.8514000000000003e-06, "loss": 0.0079, "step": 35744 }, { "epoch": 1.9056885429439676, "grad_norm": 0.9396709203720093, "learning_rate": 2.8512e-06, "loss": 0.0058, "step": 35745 }, { "epoch": 1.9057418563736204, "grad_norm": 1.5478055477142334, "learning_rate": 2.8510000000000003e-06, "loss": 0.0148, "step": 35746 }, { "epoch": 1.9057951698032736, "grad_norm": 1.7374897003173828, "learning_rate": 2.8508000000000003e-06, "loss": 0.0205, "step": 35747 }, { "epoch": 1.9058484832329263, "grad_norm": 1.0785359144210815, "learning_rate": 2.8506e-06, "loss": 0.0143, "step": 35748 }, { "epoch": 1.9059017966625793, "grad_norm": 1.7197006940841675, "learning_rate": 2.8504000000000003e-06, "loss": 0.0371, "step": 35749 }, { "epoch": 1.9059551100922323, "grad_norm": 0.1742996722459793, "learning_rate": 2.8502000000000003e-06, "loss": 0.0019, "step": 35750 }, { "epoch": 1.906008423521885, "grad_norm": 1.2190719842910767, "learning_rate": 2.85e-06, "loss": 0.0071, "step": 35751 }, { "epoch": 1.9060617369515382, "grad_norm": 2.3619391918182373, "learning_rate": 2.8498000000000002e-06, "loss": 0.0368, "step": 35752 }, { "epoch": 1.906115050381191, "grad_norm": 0.807023823261261, "learning_rate": 2.8496e-06, "loss": 0.0159, "step": 35753 }, { "epoch": 1.906168363810844, "grad_norm": 1.856187105178833, "learning_rate": 2.8494000000000006e-06, "loss": 0.0093, "step": 35754 }, { "epoch": 1.906221677240497, "grad_norm": 2.967200994491577, "learning_rate": 2.8492e-06, "loss": 0.029, "step": 35755 }, { "epoch": 1.9062749906701497, "grad_norm": 2.3952624797821045, "learning_rate": 2.849e-06, "loss": 0.0231, "step": 35756 }, { "epoch": 1.9063283040998027, "grad_norm": 1.9686157703399658, "learning_rate": 2.8488000000000006e-06, "loss": 0.0304, "step": 35757 }, { "epoch": 1.9063816175294557, "grad_norm": 0.9201948642730713, "learning_rate": 2.8486e-06, "loss": 0.0178, "step": 35758 }, { "epoch": 1.9064349309591087, "grad_norm": 2.9498631954193115, "learning_rate": 2.8484e-06, "loss": 0.0228, "step": 35759 }, { "epoch": 1.9064882443887616, "grad_norm": 2.084073781967163, "learning_rate": 2.8482000000000005e-06, "loss": 0.0153, "step": 35760 }, { "epoch": 1.9065415578184144, "grad_norm": 0.2869436740875244, "learning_rate": 2.848e-06, "loss": 0.0031, "step": 35761 }, { "epoch": 1.9065948712480674, "grad_norm": 2.1397340297698975, "learning_rate": 2.8478e-06, "loss": 0.0236, "step": 35762 }, { "epoch": 1.9066481846777203, "grad_norm": 2.3726677894592285, "learning_rate": 2.8476000000000005e-06, "loss": 0.0332, "step": 35763 }, { "epoch": 1.906701498107373, "grad_norm": 2.35514235496521, "learning_rate": 2.8474e-06, "loss": 0.0111, "step": 35764 }, { "epoch": 1.9067548115370263, "grad_norm": 2.0620365142822266, "learning_rate": 2.8472e-06, "loss": 0.0246, "step": 35765 }, { "epoch": 1.906808124966679, "grad_norm": 1.3993194103240967, "learning_rate": 2.8470000000000004e-06, "loss": 0.007, "step": 35766 }, { "epoch": 1.906861438396332, "grad_norm": 2.805837631225586, "learning_rate": 2.8468e-06, "loss": 0.0345, "step": 35767 }, { "epoch": 1.906914751825985, "grad_norm": 2.027817487716675, "learning_rate": 2.8466000000000004e-06, "loss": 0.0252, "step": 35768 }, { "epoch": 1.9069680652556378, "grad_norm": 2.375725269317627, "learning_rate": 2.8464000000000004e-06, "loss": 0.0253, "step": 35769 }, { "epoch": 1.907021378685291, "grad_norm": 1.7395046949386597, "learning_rate": 2.8462e-06, "loss": 0.015, "step": 35770 }, { "epoch": 1.9070746921149437, "grad_norm": 1.529044270515442, "learning_rate": 2.8460000000000004e-06, "loss": 0.0111, "step": 35771 }, { "epoch": 1.9071280055445967, "grad_norm": 4.187777042388916, "learning_rate": 2.8458000000000003e-06, "loss": 0.0469, "step": 35772 }, { "epoch": 1.9071813189742497, "grad_norm": 1.4217981100082397, "learning_rate": 2.8456e-06, "loss": 0.0068, "step": 35773 }, { "epoch": 1.9072346324039025, "grad_norm": 2.857405424118042, "learning_rate": 2.8454000000000003e-06, "loss": 0.0144, "step": 35774 }, { "epoch": 1.9072879458335554, "grad_norm": 2.7178781032562256, "learning_rate": 2.8452000000000003e-06, "loss": 0.038, "step": 35775 }, { "epoch": 1.9073412592632084, "grad_norm": 1.0118181705474854, "learning_rate": 2.845e-06, "loss": 0.0086, "step": 35776 }, { "epoch": 1.9073945726928612, "grad_norm": 0.7716131806373596, "learning_rate": 2.8448000000000003e-06, "loss": 0.0084, "step": 35777 }, { "epoch": 1.9074478861225144, "grad_norm": 1.5051383972167969, "learning_rate": 2.8446000000000003e-06, "loss": 0.0107, "step": 35778 }, { "epoch": 1.9075011995521671, "grad_norm": 1.6394098997116089, "learning_rate": 2.8444000000000007e-06, "loss": 0.0292, "step": 35779 }, { "epoch": 1.90755451298182, "grad_norm": 1.5445075035095215, "learning_rate": 2.8442000000000002e-06, "loss": 0.0252, "step": 35780 }, { "epoch": 1.907607826411473, "grad_norm": 0.5612103939056396, "learning_rate": 2.8440000000000002e-06, "loss": 0.0057, "step": 35781 }, { "epoch": 1.9076611398411258, "grad_norm": 1.880132794380188, "learning_rate": 2.8438000000000006e-06, "loss": 0.049, "step": 35782 }, { "epoch": 1.907714453270779, "grad_norm": 1.294877529144287, "learning_rate": 2.8436e-06, "loss": 0.0189, "step": 35783 }, { "epoch": 1.9077677667004318, "grad_norm": 1.8164663314819336, "learning_rate": 2.8434e-06, "loss": 0.0135, "step": 35784 }, { "epoch": 1.9078210801300848, "grad_norm": 2.8194947242736816, "learning_rate": 2.8432000000000006e-06, "loss": 0.0245, "step": 35785 }, { "epoch": 1.9078743935597378, "grad_norm": 2.542229413986206, "learning_rate": 2.843e-06, "loss": 0.0311, "step": 35786 }, { "epoch": 1.9079277069893905, "grad_norm": 0.601635217666626, "learning_rate": 2.8428e-06, "loss": 0.0041, "step": 35787 }, { "epoch": 1.9079810204190437, "grad_norm": 3.786109447479248, "learning_rate": 2.8426000000000005e-06, "loss": 0.0443, "step": 35788 }, { "epoch": 1.9080343338486965, "grad_norm": 0.7044629454612732, "learning_rate": 2.8424e-06, "loss": 0.0151, "step": 35789 }, { "epoch": 1.9080876472783495, "grad_norm": 2.4513494968414307, "learning_rate": 2.8422e-06, "loss": 0.0303, "step": 35790 }, { "epoch": 1.9081409607080024, "grad_norm": 0.6960993409156799, "learning_rate": 2.8420000000000005e-06, "loss": 0.0053, "step": 35791 }, { "epoch": 1.9081942741376552, "grad_norm": 1.3591444492340088, "learning_rate": 2.8418e-06, "loss": 0.0194, "step": 35792 }, { "epoch": 1.9082475875673082, "grad_norm": 1.7521703243255615, "learning_rate": 2.8416000000000005e-06, "loss": 0.0157, "step": 35793 }, { "epoch": 1.9083009009969611, "grad_norm": 1.6020938158035278, "learning_rate": 2.8414000000000004e-06, "loss": 0.0218, "step": 35794 }, { "epoch": 1.908354214426614, "grad_norm": 2.6508448123931885, "learning_rate": 2.8412e-06, "loss": 0.0244, "step": 35795 }, { "epoch": 1.908407527856267, "grad_norm": 1.082728624343872, "learning_rate": 2.8410000000000004e-06, "loss": 0.0048, "step": 35796 }, { "epoch": 1.9084608412859199, "grad_norm": 4.1888227462768555, "learning_rate": 2.8408000000000004e-06, "loss": 0.0385, "step": 35797 }, { "epoch": 1.9085141547155728, "grad_norm": 2.0832977294921875, "learning_rate": 2.8406e-06, "loss": 0.0172, "step": 35798 }, { "epoch": 1.9085674681452258, "grad_norm": 1.8240126371383667, "learning_rate": 2.8404000000000004e-06, "loss": 0.0241, "step": 35799 }, { "epoch": 1.9086207815748786, "grad_norm": 2.1882762908935547, "learning_rate": 2.8402000000000004e-06, "loss": 0.0468, "step": 35800 }, { "epoch": 1.9086740950045318, "grad_norm": 0.970846951007843, "learning_rate": 2.84e-06, "loss": 0.0089, "step": 35801 }, { "epoch": 1.9087274084341845, "grad_norm": 1.1745548248291016, "learning_rate": 2.8398000000000003e-06, "loss": 0.0099, "step": 35802 }, { "epoch": 1.9087807218638375, "grad_norm": 1.527844786643982, "learning_rate": 2.8396000000000003e-06, "loss": 0.0193, "step": 35803 }, { "epoch": 1.9088340352934905, "grad_norm": 3.4114198684692383, "learning_rate": 2.8394000000000003e-06, "loss": 0.0121, "step": 35804 }, { "epoch": 1.9088873487231433, "grad_norm": 3.8374531269073486, "learning_rate": 2.8392000000000003e-06, "loss": 0.0555, "step": 35805 }, { "epoch": 1.9089406621527965, "grad_norm": 1.3822605609893799, "learning_rate": 2.8390000000000003e-06, "loss": 0.021, "step": 35806 }, { "epoch": 1.9089939755824492, "grad_norm": 1.972420334815979, "learning_rate": 2.8388000000000002e-06, "loss": 0.022, "step": 35807 }, { "epoch": 1.9090472890121022, "grad_norm": 3.2308852672576904, "learning_rate": 2.8386000000000002e-06, "loss": 0.0136, "step": 35808 }, { "epoch": 1.9091006024417552, "grad_norm": 2.878865957260132, "learning_rate": 2.8384000000000002e-06, "loss": 0.0243, "step": 35809 }, { "epoch": 1.909153915871408, "grad_norm": 3.271956205368042, "learning_rate": 2.8382e-06, "loss": 0.0296, "step": 35810 }, { "epoch": 1.909207229301061, "grad_norm": 2.509697198867798, "learning_rate": 2.838e-06, "loss": 0.0217, "step": 35811 }, { "epoch": 1.9092605427307139, "grad_norm": 2.280200719833374, "learning_rate": 2.8378e-06, "loss": 0.0204, "step": 35812 }, { "epoch": 1.9093138561603666, "grad_norm": 1.9824419021606445, "learning_rate": 2.8376e-06, "loss": 0.0271, "step": 35813 }, { "epoch": 1.9093671695900198, "grad_norm": 0.5542270541191101, "learning_rate": 2.8374e-06, "loss": 0.0059, "step": 35814 }, { "epoch": 1.9094204830196726, "grad_norm": 1.5667109489440918, "learning_rate": 2.8372000000000006e-06, "loss": 0.01, "step": 35815 }, { "epoch": 1.9094737964493256, "grad_norm": 4.691005706787109, "learning_rate": 2.837e-06, "loss": 0.0386, "step": 35816 }, { "epoch": 1.9095271098789786, "grad_norm": 1.6754634380340576, "learning_rate": 2.8368e-06, "loss": 0.0227, "step": 35817 }, { "epoch": 1.9095804233086313, "grad_norm": 1.399743676185608, "learning_rate": 2.8366000000000005e-06, "loss": 0.0102, "step": 35818 }, { "epoch": 1.9096337367382845, "grad_norm": 0.8806791305541992, "learning_rate": 2.8364e-06, "loss": 0.0072, "step": 35819 }, { "epoch": 1.9096870501679373, "grad_norm": 0.8962180614471436, "learning_rate": 2.8362e-06, "loss": 0.0056, "step": 35820 }, { "epoch": 1.9097403635975903, "grad_norm": 0.9625412225723267, "learning_rate": 2.8360000000000005e-06, "loss": 0.0334, "step": 35821 }, { "epoch": 1.9097936770272432, "grad_norm": 2.026771068572998, "learning_rate": 2.8358e-06, "loss": 0.0554, "step": 35822 }, { "epoch": 1.909846990456896, "grad_norm": 1.6848174333572388, "learning_rate": 2.8356e-06, "loss": 0.0285, "step": 35823 }, { "epoch": 1.909900303886549, "grad_norm": 1.2063000202178955, "learning_rate": 2.8354000000000004e-06, "loss": 0.0057, "step": 35824 }, { "epoch": 1.909953617316202, "grad_norm": 1.3673347234725952, "learning_rate": 2.8352e-06, "loss": 0.0094, "step": 35825 }, { "epoch": 1.910006930745855, "grad_norm": 1.4095040559768677, "learning_rate": 2.835e-06, "loss": 0.0042, "step": 35826 }, { "epoch": 1.910060244175508, "grad_norm": 1.3543803691864014, "learning_rate": 2.8348000000000004e-06, "loss": 0.0149, "step": 35827 }, { "epoch": 1.9101135576051607, "grad_norm": 4.096478462219238, "learning_rate": 2.8346e-06, "loss": 0.0541, "step": 35828 }, { "epoch": 1.9101668710348136, "grad_norm": 1.0315130949020386, "learning_rate": 2.8344000000000003e-06, "loss": 0.0073, "step": 35829 }, { "epoch": 1.9102201844644666, "grad_norm": 5.170164585113525, "learning_rate": 2.8342000000000003e-06, "loss": 0.0365, "step": 35830 }, { "epoch": 1.9102734978941194, "grad_norm": 1.9381023645401, "learning_rate": 2.834e-06, "loss": 0.041, "step": 35831 }, { "epoch": 1.9103268113237726, "grad_norm": 2.4188268184661865, "learning_rate": 2.8338000000000003e-06, "loss": 0.0165, "step": 35832 }, { "epoch": 1.9103801247534253, "grad_norm": 2.808138847351074, "learning_rate": 2.8336000000000003e-06, "loss": 0.0445, "step": 35833 }, { "epoch": 1.9104334381830783, "grad_norm": 5.095404624938965, "learning_rate": 2.8334e-06, "loss": 0.0243, "step": 35834 }, { "epoch": 1.9104867516127313, "grad_norm": 1.5451020002365112, "learning_rate": 2.8332000000000002e-06, "loss": 0.0119, "step": 35835 }, { "epoch": 1.910540065042384, "grad_norm": 2.0229692459106445, "learning_rate": 2.8330000000000002e-06, "loss": 0.0224, "step": 35836 }, { "epoch": 1.9105933784720373, "grad_norm": 1.5091503858566284, "learning_rate": 2.8328e-06, "loss": 0.0148, "step": 35837 }, { "epoch": 1.91064669190169, "grad_norm": 1.2169350385665894, "learning_rate": 2.8326e-06, "loss": 0.0136, "step": 35838 }, { "epoch": 1.910700005331343, "grad_norm": 2.0835349559783936, "learning_rate": 2.8324e-06, "loss": 0.0283, "step": 35839 }, { "epoch": 1.910753318760996, "grad_norm": 1.2371323108673096, "learning_rate": 2.8322000000000006e-06, "loss": 0.0074, "step": 35840 }, { "epoch": 1.9108066321906487, "grad_norm": 3.510702133178711, "learning_rate": 2.832e-06, "loss": 0.0384, "step": 35841 }, { "epoch": 1.9108599456203017, "grad_norm": 0.9571854472160339, "learning_rate": 2.8318e-06, "loss": 0.0082, "step": 35842 }, { "epoch": 1.9109132590499547, "grad_norm": 2.526217460632324, "learning_rate": 2.8316000000000006e-06, "loss": 0.0201, "step": 35843 }, { "epoch": 1.9109665724796077, "grad_norm": 0.835654079914093, "learning_rate": 2.8314e-06, "loss": 0.0065, "step": 35844 }, { "epoch": 1.9110198859092606, "grad_norm": 1.2941112518310547, "learning_rate": 2.8312e-06, "loss": 0.0137, "step": 35845 }, { "epoch": 1.9110731993389134, "grad_norm": 0.8789567947387695, "learning_rate": 2.8310000000000005e-06, "loss": 0.0125, "step": 35846 }, { "epoch": 1.9111265127685664, "grad_norm": 1.1776334047317505, "learning_rate": 2.8308e-06, "loss": 0.0214, "step": 35847 }, { "epoch": 1.9111798261982194, "grad_norm": 1.5408997535705566, "learning_rate": 2.8306e-06, "loss": 0.0102, "step": 35848 }, { "epoch": 1.9112331396278721, "grad_norm": 0.6374318599700928, "learning_rate": 2.8304000000000005e-06, "loss": 0.0054, "step": 35849 }, { "epoch": 1.9112864530575253, "grad_norm": 2.5700745582580566, "learning_rate": 2.8302e-06, "loss": 0.0186, "step": 35850 }, { "epoch": 1.911339766487178, "grad_norm": 1.4362164735794067, "learning_rate": 2.83e-06, "loss": 0.0188, "step": 35851 }, { "epoch": 1.911393079916831, "grad_norm": 2.3638317584991455, "learning_rate": 2.8298000000000004e-06, "loss": 0.0188, "step": 35852 }, { "epoch": 1.911446393346484, "grad_norm": 0.8606613874435425, "learning_rate": 2.8296e-06, "loss": 0.0118, "step": 35853 }, { "epoch": 1.9114997067761368, "grad_norm": 1.4932994842529297, "learning_rate": 2.8294000000000004e-06, "loss": 0.0109, "step": 35854 }, { "epoch": 1.91155302020579, "grad_norm": 1.1784754991531372, "learning_rate": 2.8292000000000004e-06, "loss": 0.0086, "step": 35855 }, { "epoch": 1.9116063336354427, "grad_norm": 1.477313756942749, "learning_rate": 2.829e-06, "loss": 0.0281, "step": 35856 }, { "epoch": 1.9116596470650957, "grad_norm": 1.376854419708252, "learning_rate": 2.8288000000000003e-06, "loss": 0.0128, "step": 35857 }, { "epoch": 1.9117129604947487, "grad_norm": 1.2267895936965942, "learning_rate": 2.8286000000000003e-06, "loss": 0.0218, "step": 35858 }, { "epoch": 1.9117662739244015, "grad_norm": 7.512252330780029, "learning_rate": 2.8284e-06, "loss": 0.0517, "step": 35859 }, { "epoch": 1.9118195873540544, "grad_norm": 3.2053959369659424, "learning_rate": 2.8282000000000003e-06, "loss": 0.0435, "step": 35860 }, { "epoch": 1.9118729007837074, "grad_norm": 1.311344861984253, "learning_rate": 2.8280000000000003e-06, "loss": 0.0155, "step": 35861 }, { "epoch": 1.9119262142133604, "grad_norm": 0.28715693950653076, "learning_rate": 2.8278e-06, "loss": 0.0026, "step": 35862 }, { "epoch": 1.9119795276430134, "grad_norm": 2.025932550430298, "learning_rate": 2.8276000000000003e-06, "loss": 0.0442, "step": 35863 }, { "epoch": 1.9120328410726661, "grad_norm": 0.34222304821014404, "learning_rate": 2.8274000000000002e-06, "loss": 0.0023, "step": 35864 }, { "epoch": 1.9120861545023191, "grad_norm": 1.6187858581542969, "learning_rate": 2.8272000000000006e-06, "loss": 0.0142, "step": 35865 }, { "epoch": 1.912139467931972, "grad_norm": 2.118424892425537, "learning_rate": 2.827e-06, "loss": 0.0144, "step": 35866 }, { "epoch": 1.9121927813616248, "grad_norm": 2.2208287715911865, "learning_rate": 2.8268e-06, "loss": 0.0097, "step": 35867 }, { "epoch": 1.912246094791278, "grad_norm": 3.0534658432006836, "learning_rate": 2.8266000000000006e-06, "loss": 0.0657, "step": 35868 }, { "epoch": 1.9122994082209308, "grad_norm": 2.058758020401001, "learning_rate": 2.8264e-06, "loss": 0.0454, "step": 35869 }, { "epoch": 1.9123527216505838, "grad_norm": 0.33692294359207153, "learning_rate": 2.8262e-06, "loss": 0.0031, "step": 35870 }, { "epoch": 1.9124060350802368, "grad_norm": 1.34148108959198, "learning_rate": 2.8260000000000006e-06, "loss": 0.0229, "step": 35871 }, { "epoch": 1.9124593485098895, "grad_norm": 1.2451739311218262, "learning_rate": 2.8258e-06, "loss": 0.0231, "step": 35872 }, { "epoch": 1.9125126619395427, "grad_norm": 1.2411940097808838, "learning_rate": 2.8256e-06, "loss": 0.0089, "step": 35873 }, { "epoch": 1.9125659753691955, "grad_norm": 1.1662458181381226, "learning_rate": 2.8254000000000005e-06, "loss": 0.0085, "step": 35874 }, { "epoch": 1.9126192887988485, "grad_norm": 0.6289665102958679, "learning_rate": 2.8252e-06, "loss": 0.0045, "step": 35875 }, { "epoch": 1.9126726022285014, "grad_norm": 1.818215250968933, "learning_rate": 2.825e-06, "loss": 0.0053, "step": 35876 }, { "epoch": 1.9127259156581542, "grad_norm": 0.7439497113227844, "learning_rate": 2.8248000000000005e-06, "loss": 0.0041, "step": 35877 }, { "epoch": 1.9127792290878072, "grad_norm": 3.0752344131469727, "learning_rate": 2.8246e-06, "loss": 0.0104, "step": 35878 }, { "epoch": 1.9128325425174602, "grad_norm": 0.9978509545326233, "learning_rate": 2.8244000000000004e-06, "loss": 0.0057, "step": 35879 }, { "epoch": 1.912885855947113, "grad_norm": 1.2021812200546265, "learning_rate": 2.8242000000000004e-06, "loss": 0.0054, "step": 35880 }, { "epoch": 1.9129391693767661, "grad_norm": 1.2528095245361328, "learning_rate": 2.824e-06, "loss": 0.0173, "step": 35881 }, { "epoch": 1.9129924828064189, "grad_norm": 3.041719675064087, "learning_rate": 2.8238000000000004e-06, "loss": 0.0093, "step": 35882 }, { "epoch": 1.9130457962360718, "grad_norm": 0.8851122260093689, "learning_rate": 2.8236000000000004e-06, "loss": 0.0107, "step": 35883 }, { "epoch": 1.9130991096657248, "grad_norm": 0.37275439500808716, "learning_rate": 2.8234e-06, "loss": 0.003, "step": 35884 }, { "epoch": 1.9131524230953776, "grad_norm": 1.5217009782791138, "learning_rate": 2.8232000000000003e-06, "loss": 0.0076, "step": 35885 }, { "epoch": 1.9132057365250308, "grad_norm": 1.0860159397125244, "learning_rate": 2.8230000000000003e-06, "loss": 0.0124, "step": 35886 }, { "epoch": 1.9132590499546835, "grad_norm": 2.18963360786438, "learning_rate": 2.8228e-06, "loss": 0.0189, "step": 35887 }, { "epoch": 1.9133123633843365, "grad_norm": 0.8748178482055664, "learning_rate": 2.8226000000000003e-06, "loss": 0.028, "step": 35888 }, { "epoch": 1.9133656768139895, "grad_norm": 2.3250439167022705, "learning_rate": 2.8224000000000003e-06, "loss": 0.0362, "step": 35889 }, { "epoch": 1.9134189902436423, "grad_norm": 0.4691040515899658, "learning_rate": 2.8222000000000007e-06, "loss": 0.0043, "step": 35890 }, { "epoch": 1.9134723036732955, "grad_norm": 0.7535061240196228, "learning_rate": 2.8220000000000003e-06, "loss": 0.0145, "step": 35891 }, { "epoch": 1.9135256171029482, "grad_norm": 2.1033127307891846, "learning_rate": 2.8218000000000002e-06, "loss": 0.024, "step": 35892 }, { "epoch": 1.9135789305326012, "grad_norm": 1.7351608276367188, "learning_rate": 2.8216000000000006e-06, "loss": 0.0337, "step": 35893 }, { "epoch": 1.9136322439622542, "grad_norm": 2.451000452041626, "learning_rate": 2.8214e-06, "loss": 0.0282, "step": 35894 }, { "epoch": 1.913685557391907, "grad_norm": 2.3641092777252197, "learning_rate": 2.8212e-06, "loss": 0.0225, "step": 35895 }, { "epoch": 1.91373887082156, "grad_norm": 1.373757243156433, "learning_rate": 2.8210000000000006e-06, "loss": 0.0192, "step": 35896 }, { "epoch": 1.913792184251213, "grad_norm": 3.9151971340179443, "learning_rate": 2.8208e-06, "loss": 0.0316, "step": 35897 }, { "epoch": 1.9138454976808656, "grad_norm": 0.7161871790885925, "learning_rate": 2.8206e-06, "loss": 0.0041, "step": 35898 }, { "epoch": 1.9138988111105188, "grad_norm": 0.5490087270736694, "learning_rate": 2.8204000000000006e-06, "loss": 0.0042, "step": 35899 }, { "epoch": 1.9139521245401716, "grad_norm": 1.540767788887024, "learning_rate": 2.8202e-06, "loss": 0.0116, "step": 35900 }, { "epoch": 1.9140054379698246, "grad_norm": 2.5179603099823, "learning_rate": 2.82e-06, "loss": 0.0432, "step": 35901 }, { "epoch": 1.9140587513994776, "grad_norm": 1.9721708297729492, "learning_rate": 2.8198000000000005e-06, "loss": 0.0292, "step": 35902 }, { "epoch": 1.9141120648291303, "grad_norm": 1.0520329475402832, "learning_rate": 2.8196e-06, "loss": 0.0067, "step": 35903 }, { "epoch": 1.9141653782587835, "grad_norm": 1.4407023191452026, "learning_rate": 2.8194000000000005e-06, "loss": 0.0094, "step": 35904 }, { "epoch": 1.9142186916884363, "grad_norm": 1.4804586172103882, "learning_rate": 2.8192000000000005e-06, "loss": 0.0217, "step": 35905 }, { "epoch": 1.9142720051180893, "grad_norm": 0.6022367477416992, "learning_rate": 2.819e-06, "loss": 0.005, "step": 35906 }, { "epoch": 1.9143253185477422, "grad_norm": 1.1599376201629639, "learning_rate": 2.8188000000000004e-06, "loss": 0.0131, "step": 35907 }, { "epoch": 1.914378631977395, "grad_norm": 0.5805956125259399, "learning_rate": 2.8186000000000004e-06, "loss": 0.0048, "step": 35908 }, { "epoch": 1.9144319454070482, "grad_norm": 0.5445152521133423, "learning_rate": 2.8184e-06, "loss": 0.0037, "step": 35909 }, { "epoch": 1.914485258836701, "grad_norm": 1.1353614330291748, "learning_rate": 2.8182000000000004e-06, "loss": 0.0126, "step": 35910 }, { "epoch": 1.914538572266354, "grad_norm": 2.035902976989746, "learning_rate": 2.8180000000000004e-06, "loss": 0.0202, "step": 35911 }, { "epoch": 1.914591885696007, "grad_norm": 1.9398564100265503, "learning_rate": 2.8178e-06, "loss": 0.0078, "step": 35912 }, { "epoch": 1.9146451991256597, "grad_norm": 2.7281088829040527, "learning_rate": 2.8176000000000003e-06, "loss": 0.0225, "step": 35913 }, { "epoch": 1.9146985125553126, "grad_norm": 4.010985374450684, "learning_rate": 2.8174000000000003e-06, "loss": 0.0439, "step": 35914 }, { "epoch": 1.9147518259849656, "grad_norm": 1.014414668083191, "learning_rate": 2.8172000000000003e-06, "loss": 0.0078, "step": 35915 }, { "epoch": 1.9148051394146184, "grad_norm": 3.0995826721191406, "learning_rate": 2.8170000000000003e-06, "loss": 0.0197, "step": 35916 }, { "epoch": 1.9148584528442716, "grad_norm": 1.4454686641693115, "learning_rate": 2.8168000000000003e-06, "loss": 0.0083, "step": 35917 }, { "epoch": 1.9149117662739243, "grad_norm": 1.695807933807373, "learning_rate": 2.8166000000000003e-06, "loss": 0.0094, "step": 35918 }, { "epoch": 1.9149650797035773, "grad_norm": 1.9116121530532837, "learning_rate": 2.8164000000000003e-06, "loss": 0.0187, "step": 35919 }, { "epoch": 1.9150183931332303, "grad_norm": 2.4656009674072266, "learning_rate": 2.8162000000000002e-06, "loss": 0.0265, "step": 35920 }, { "epoch": 1.915071706562883, "grad_norm": 1.3692132234573364, "learning_rate": 2.8160000000000002e-06, "loss": 0.0054, "step": 35921 }, { "epoch": 1.9151250199925363, "grad_norm": 2.4814212322235107, "learning_rate": 2.8158000000000002e-06, "loss": 0.0184, "step": 35922 }, { "epoch": 1.915178333422189, "grad_norm": 2.882612705230713, "learning_rate": 2.8156e-06, "loss": 0.0186, "step": 35923 }, { "epoch": 1.915231646851842, "grad_norm": 0.38925614953041077, "learning_rate": 2.8154e-06, "loss": 0.0039, "step": 35924 }, { "epoch": 1.915284960281495, "grad_norm": 2.7227556705474854, "learning_rate": 2.8152e-06, "loss": 0.0076, "step": 35925 }, { "epoch": 1.9153382737111477, "grad_norm": 2.3090295791625977, "learning_rate": 2.815e-06, "loss": 0.0327, "step": 35926 }, { "epoch": 1.915391587140801, "grad_norm": 2.552661657333374, "learning_rate": 2.8148e-06, "loss": 0.0194, "step": 35927 }, { "epoch": 1.9154449005704537, "grad_norm": 1.4528683423995972, "learning_rate": 2.8146e-06, "loss": 0.0062, "step": 35928 }, { "epoch": 1.9154982140001067, "grad_norm": 0.6040320992469788, "learning_rate": 2.8144000000000005e-06, "loss": 0.0054, "step": 35929 }, { "epoch": 1.9155515274297596, "grad_norm": 2.222939968109131, "learning_rate": 2.8142e-06, "loss": 0.0157, "step": 35930 }, { "epoch": 1.9156048408594124, "grad_norm": 2.2690088748931885, "learning_rate": 2.814e-06, "loss": 0.0181, "step": 35931 }, { "epoch": 1.9156581542890654, "grad_norm": 2.783219337463379, "learning_rate": 2.8138000000000005e-06, "loss": 0.03, "step": 35932 }, { "epoch": 1.9157114677187184, "grad_norm": 1.2296863794326782, "learning_rate": 2.8136e-06, "loss": 0.0145, "step": 35933 }, { "epoch": 1.9157647811483711, "grad_norm": 3.0814199447631836, "learning_rate": 2.8134e-06, "loss": 0.0433, "step": 35934 }, { "epoch": 1.9158180945780243, "grad_norm": 1.9993014335632324, "learning_rate": 2.8132000000000004e-06, "loss": 0.0301, "step": 35935 }, { "epoch": 1.915871408007677, "grad_norm": 2.149226427078247, "learning_rate": 2.813e-06, "loss": 0.0136, "step": 35936 }, { "epoch": 1.91592472143733, "grad_norm": 2.9468963146209717, "learning_rate": 2.8128e-06, "loss": 0.0227, "step": 35937 }, { "epoch": 1.915978034866983, "grad_norm": 1.0523104667663574, "learning_rate": 2.8126000000000004e-06, "loss": 0.0068, "step": 35938 }, { "epoch": 1.9160313482966358, "grad_norm": 2.5318989753723145, "learning_rate": 2.8124e-06, "loss": 0.0447, "step": 35939 }, { "epoch": 1.916084661726289, "grad_norm": 1.49588143825531, "learning_rate": 2.8122000000000004e-06, "loss": 0.0134, "step": 35940 }, { "epoch": 1.9161379751559418, "grad_norm": 1.4482945203781128, "learning_rate": 2.8120000000000004e-06, "loss": 0.0103, "step": 35941 }, { "epoch": 1.9161912885855947, "grad_norm": 1.5118930339813232, "learning_rate": 2.8118e-06, "loss": 0.0169, "step": 35942 }, { "epoch": 1.9162446020152477, "grad_norm": 0.9857409596443176, "learning_rate": 2.8116000000000003e-06, "loss": 0.0222, "step": 35943 }, { "epoch": 1.9162979154449005, "grad_norm": 2.7176764011383057, "learning_rate": 2.8114000000000003e-06, "loss": 0.0183, "step": 35944 }, { "epoch": 1.9163512288745534, "grad_norm": 2.6162445545196533, "learning_rate": 2.8112e-06, "loss": 0.0253, "step": 35945 }, { "epoch": 1.9164045423042064, "grad_norm": 2.569169521331787, "learning_rate": 2.8110000000000003e-06, "loss": 0.0355, "step": 35946 }, { "epoch": 1.9164578557338594, "grad_norm": 1.9432015419006348, "learning_rate": 2.8108000000000003e-06, "loss": 0.0103, "step": 35947 }, { "epoch": 1.9165111691635124, "grad_norm": 2.8702476024627686, "learning_rate": 2.8106e-06, "loss": 0.0297, "step": 35948 }, { "epoch": 1.9165644825931651, "grad_norm": 3.477992534637451, "learning_rate": 2.8104000000000002e-06, "loss": 0.0278, "step": 35949 }, { "epoch": 1.9166177960228181, "grad_norm": 3.2218520641326904, "learning_rate": 2.8102000000000002e-06, "loss": 0.0285, "step": 35950 }, { "epoch": 1.916671109452471, "grad_norm": 1.341483473777771, "learning_rate": 2.8100000000000006e-06, "loss": 0.0089, "step": 35951 }, { "epoch": 1.9167244228821239, "grad_norm": 3.1285219192504883, "learning_rate": 2.8098e-06, "loss": 0.0186, "step": 35952 }, { "epoch": 1.916777736311777, "grad_norm": 1.9235460758209229, "learning_rate": 2.8096e-06, "loss": 0.0137, "step": 35953 }, { "epoch": 1.9168310497414298, "grad_norm": 0.4114643633365631, "learning_rate": 2.8094000000000006e-06, "loss": 0.0027, "step": 35954 }, { "epoch": 1.9168843631710828, "grad_norm": 3.9165265560150146, "learning_rate": 2.8092e-06, "loss": 0.0375, "step": 35955 }, { "epoch": 1.9169376766007358, "grad_norm": 1.396558403968811, "learning_rate": 2.809e-06, "loss": 0.0126, "step": 35956 }, { "epoch": 1.9169909900303885, "grad_norm": 3.243539333343506, "learning_rate": 2.8088000000000005e-06, "loss": 0.0307, "step": 35957 }, { "epoch": 1.9170443034600417, "grad_norm": 1.460976243019104, "learning_rate": 2.8086e-06, "loss": 0.0076, "step": 35958 }, { "epoch": 1.9170976168896945, "grad_norm": 0.609427273273468, "learning_rate": 2.8084e-06, "loss": 0.0046, "step": 35959 }, { "epoch": 1.9171509303193475, "grad_norm": 1.0033059120178223, "learning_rate": 2.8082000000000005e-06, "loss": 0.0045, "step": 35960 }, { "epoch": 1.9172042437490004, "grad_norm": 1.2095375061035156, "learning_rate": 2.808e-06, "loss": 0.0292, "step": 35961 }, { "epoch": 1.9172575571786532, "grad_norm": 2.548708438873291, "learning_rate": 2.8078e-06, "loss": 0.0422, "step": 35962 }, { "epoch": 1.9173108706083062, "grad_norm": 0.7810842394828796, "learning_rate": 2.8076000000000004e-06, "loss": 0.006, "step": 35963 }, { "epoch": 1.9173641840379592, "grad_norm": 1.5883597135543823, "learning_rate": 2.8074e-06, "loss": 0.0167, "step": 35964 }, { "epoch": 1.9174174974676121, "grad_norm": 0.7528931498527527, "learning_rate": 2.8072000000000004e-06, "loss": 0.0167, "step": 35965 }, { "epoch": 1.9174708108972651, "grad_norm": 1.6033539772033691, "learning_rate": 2.8070000000000004e-06, "loss": 0.0203, "step": 35966 }, { "epoch": 1.9175241243269179, "grad_norm": 3.5961596965789795, "learning_rate": 2.8068e-06, "loss": 0.0618, "step": 35967 }, { "epoch": 1.9175774377565709, "grad_norm": 0.563141942024231, "learning_rate": 2.8066000000000004e-06, "loss": 0.0038, "step": 35968 }, { "epoch": 1.9176307511862238, "grad_norm": 2.337425470352173, "learning_rate": 2.8064000000000004e-06, "loss": 0.0182, "step": 35969 }, { "epoch": 1.9176840646158766, "grad_norm": 2.6575698852539062, "learning_rate": 2.8062e-06, "loss": 0.0324, "step": 35970 }, { "epoch": 1.9177373780455298, "grad_norm": 3.294447422027588, "learning_rate": 2.8060000000000003e-06, "loss": 0.0358, "step": 35971 }, { "epoch": 1.9177906914751826, "grad_norm": 2.679257392883301, "learning_rate": 2.8058000000000003e-06, "loss": 0.048, "step": 35972 }, { "epoch": 1.9178440049048355, "grad_norm": 0.25284072756767273, "learning_rate": 2.8056e-06, "loss": 0.0025, "step": 35973 }, { "epoch": 1.9178973183344885, "grad_norm": 0.901375949382782, "learning_rate": 2.8054000000000003e-06, "loss": 0.0061, "step": 35974 }, { "epoch": 1.9179506317641413, "grad_norm": 0.6646541357040405, "learning_rate": 2.8052000000000003e-06, "loss": 0.0144, "step": 35975 }, { "epoch": 1.9180039451937945, "grad_norm": 1.6641371250152588, "learning_rate": 2.8050000000000007e-06, "loss": 0.0211, "step": 35976 }, { "epoch": 1.9180572586234472, "grad_norm": 1.3746418952941895, "learning_rate": 2.8048000000000002e-06, "loss": 0.008, "step": 35977 }, { "epoch": 1.9181105720531002, "grad_norm": 3.0463759899139404, "learning_rate": 2.8046000000000002e-06, "loss": 0.046, "step": 35978 }, { "epoch": 1.9181638854827532, "grad_norm": 1.543060541152954, "learning_rate": 2.8044000000000006e-06, "loss": 0.024, "step": 35979 }, { "epoch": 1.918217198912406, "grad_norm": 1.1064631938934326, "learning_rate": 2.8042e-06, "loss": 0.0117, "step": 35980 }, { "epoch": 1.918270512342059, "grad_norm": 1.11276113986969, "learning_rate": 2.804e-06, "loss": 0.0222, "step": 35981 }, { "epoch": 1.918323825771712, "grad_norm": 1.363890528678894, "learning_rate": 2.8038000000000006e-06, "loss": 0.0081, "step": 35982 }, { "epoch": 1.9183771392013649, "grad_norm": 1.664326548576355, "learning_rate": 2.8036e-06, "loss": 0.0141, "step": 35983 }, { "epoch": 1.9184304526310179, "grad_norm": 1.1956820487976074, "learning_rate": 2.8034e-06, "loss": 0.0082, "step": 35984 }, { "epoch": 1.9184837660606706, "grad_norm": 1.5809944868087769, "learning_rate": 2.8032000000000005e-06, "loss": 0.0298, "step": 35985 }, { "epoch": 1.9185370794903236, "grad_norm": 2.0669896602630615, "learning_rate": 2.803e-06, "loss": 0.0186, "step": 35986 }, { "epoch": 1.9185903929199766, "grad_norm": 1.2436615228652954, "learning_rate": 2.8028e-06, "loss": 0.0252, "step": 35987 }, { "epoch": 1.9186437063496293, "grad_norm": 2.4657437801361084, "learning_rate": 2.8026000000000005e-06, "loss": 0.0145, "step": 35988 }, { "epoch": 1.9186970197792825, "grad_norm": 0.4244079887866974, "learning_rate": 2.8024e-06, "loss": 0.0043, "step": 35989 }, { "epoch": 1.9187503332089353, "grad_norm": 4.3236165046691895, "learning_rate": 2.8022000000000005e-06, "loss": 0.0329, "step": 35990 }, { "epoch": 1.9188036466385883, "grad_norm": 1.9518965482711792, "learning_rate": 2.8020000000000004e-06, "loss": 0.0356, "step": 35991 }, { "epoch": 1.9188569600682412, "grad_norm": 2.2686495780944824, "learning_rate": 2.8018e-06, "loss": 0.0208, "step": 35992 }, { "epoch": 1.918910273497894, "grad_norm": 0.7071531414985657, "learning_rate": 2.8016000000000004e-06, "loss": 0.0046, "step": 35993 }, { "epoch": 1.9189635869275472, "grad_norm": 1.880557894706726, "learning_rate": 2.8014000000000004e-06, "loss": 0.0099, "step": 35994 }, { "epoch": 1.9190169003572, "grad_norm": 1.9458978176116943, "learning_rate": 2.8012e-06, "loss": 0.0147, "step": 35995 }, { "epoch": 1.919070213786853, "grad_norm": 1.2008494138717651, "learning_rate": 2.8010000000000004e-06, "loss": 0.0064, "step": 35996 }, { "epoch": 1.919123527216506, "grad_norm": 2.9242985248565674, "learning_rate": 2.8008000000000004e-06, "loss": 0.0206, "step": 35997 }, { "epoch": 1.9191768406461587, "grad_norm": 1.1268301010131836, "learning_rate": 2.8006e-06, "loss": 0.009, "step": 35998 }, { "epoch": 1.9192301540758117, "grad_norm": 2.7118446826934814, "learning_rate": 2.8004000000000003e-06, "loss": 0.0528, "step": 35999 }, { "epoch": 1.9192834675054646, "grad_norm": 2.4440197944641113, "learning_rate": 2.8002000000000003e-06, "loss": 0.047, "step": 36000 }, { "epoch": 1.9193367809351174, "grad_norm": 1.8417145013809204, "learning_rate": 2.8000000000000003e-06, "loss": 0.0181, "step": 36001 }, { "epoch": 1.9193900943647706, "grad_norm": 2.184446096420288, "learning_rate": 2.7998000000000003e-06, "loss": 0.0357, "step": 36002 }, { "epoch": 1.9194434077944234, "grad_norm": 2.076033592224121, "learning_rate": 2.7996000000000003e-06, "loss": 0.0332, "step": 36003 }, { "epoch": 1.9194967212240763, "grad_norm": 1.1885825395584106, "learning_rate": 2.7994000000000003e-06, "loss": 0.0247, "step": 36004 }, { "epoch": 1.9195500346537293, "grad_norm": 1.416890263557434, "learning_rate": 2.7992000000000002e-06, "loss": 0.0272, "step": 36005 }, { "epoch": 1.919603348083382, "grad_norm": 1.6663364171981812, "learning_rate": 2.7990000000000002e-06, "loss": 0.0434, "step": 36006 }, { "epoch": 1.9196566615130353, "grad_norm": 2.1318984031677246, "learning_rate": 2.7988e-06, "loss": 0.0275, "step": 36007 }, { "epoch": 1.919709974942688, "grad_norm": 1.4622985124588013, "learning_rate": 2.7986e-06, "loss": 0.0209, "step": 36008 }, { "epoch": 1.919763288372341, "grad_norm": 2.4533071517944336, "learning_rate": 2.7984e-06, "loss": 0.0105, "step": 36009 }, { "epoch": 1.919816601801994, "grad_norm": 1.625429391860962, "learning_rate": 2.7982e-06, "loss": 0.031, "step": 36010 }, { "epoch": 1.9198699152316467, "grad_norm": 1.6810076236724854, "learning_rate": 2.798e-06, "loss": 0.0178, "step": 36011 }, { "epoch": 1.9199232286613, "grad_norm": 1.8060568571090698, "learning_rate": 2.7978e-06, "loss": 0.0123, "step": 36012 }, { "epoch": 1.9199765420909527, "grad_norm": 2.2096168994903564, "learning_rate": 2.7976e-06, "loss": 0.0278, "step": 36013 }, { "epoch": 1.9200298555206057, "grad_norm": 2.2933168411254883, "learning_rate": 2.7974e-06, "loss": 0.0182, "step": 36014 }, { "epoch": 1.9200831689502587, "grad_norm": 4.917923450469971, "learning_rate": 2.7972000000000005e-06, "loss": 0.0403, "step": 36015 }, { "epoch": 1.9201364823799114, "grad_norm": 0.9810953140258789, "learning_rate": 2.797e-06, "loss": 0.0094, "step": 36016 }, { "epoch": 1.9201897958095644, "grad_norm": 1.0472526550292969, "learning_rate": 2.7968e-06, "loss": 0.0099, "step": 36017 }, { "epoch": 1.9202431092392174, "grad_norm": 2.8112893104553223, "learning_rate": 2.7966000000000005e-06, "loss": 0.0214, "step": 36018 }, { "epoch": 1.9202964226688701, "grad_norm": 0.9540557265281677, "learning_rate": 2.7964e-06, "loss": 0.006, "step": 36019 }, { "epoch": 1.9203497360985233, "grad_norm": 0.7007162570953369, "learning_rate": 2.7962e-06, "loss": 0.0061, "step": 36020 }, { "epoch": 1.920403049528176, "grad_norm": 1.6716668605804443, "learning_rate": 2.7960000000000004e-06, "loss": 0.0218, "step": 36021 }, { "epoch": 1.920456362957829, "grad_norm": 0.9822875261306763, "learning_rate": 2.7958e-06, "loss": 0.0162, "step": 36022 }, { "epoch": 1.920509676387482, "grad_norm": 1.1579270362854004, "learning_rate": 2.7956e-06, "loss": 0.0219, "step": 36023 }, { "epoch": 1.9205629898171348, "grad_norm": 1.503735065460205, "learning_rate": 2.7954000000000004e-06, "loss": 0.0166, "step": 36024 }, { "epoch": 1.920616303246788, "grad_norm": 1.272532343864441, "learning_rate": 2.7952e-06, "loss": 0.0381, "step": 36025 }, { "epoch": 1.9206696166764408, "grad_norm": 0.16773498058319092, "learning_rate": 2.7950000000000003e-06, "loss": 0.0021, "step": 36026 }, { "epoch": 1.9207229301060937, "grad_norm": 0.4970606565475464, "learning_rate": 2.7948000000000003e-06, "loss": 0.0029, "step": 36027 }, { "epoch": 1.9207762435357467, "grad_norm": 1.338472843170166, "learning_rate": 2.7946e-06, "loss": 0.0061, "step": 36028 }, { "epoch": 1.9208295569653995, "grad_norm": 5.668334007263184, "learning_rate": 2.7944000000000003e-06, "loss": 0.0426, "step": 36029 }, { "epoch": 1.9208828703950527, "grad_norm": 2.8123257160186768, "learning_rate": 2.7942000000000003e-06, "loss": 0.0212, "step": 36030 }, { "epoch": 1.9209361838247054, "grad_norm": 0.7769496440887451, "learning_rate": 2.794e-06, "loss": 0.0036, "step": 36031 }, { "epoch": 1.9209894972543584, "grad_norm": 0.9357452988624573, "learning_rate": 2.7938000000000003e-06, "loss": 0.0068, "step": 36032 }, { "epoch": 1.9210428106840114, "grad_norm": 0.780634880065918, "learning_rate": 2.7936000000000002e-06, "loss": 0.0039, "step": 36033 }, { "epoch": 1.9210961241136641, "grad_norm": 0.4738702178001404, "learning_rate": 2.7934e-06, "loss": 0.0032, "step": 36034 }, { "epoch": 1.9211494375433171, "grad_norm": 1.340907335281372, "learning_rate": 2.7932e-06, "loss": 0.0142, "step": 36035 }, { "epoch": 1.92120275097297, "grad_norm": 1.269321322441101, "learning_rate": 2.793e-06, "loss": 0.0075, "step": 36036 }, { "epoch": 1.9212560644026229, "grad_norm": 0.34260210394859314, "learning_rate": 2.7927999999999998e-06, "loss": 0.0027, "step": 36037 }, { "epoch": 1.921309377832276, "grad_norm": 1.7267409563064575, "learning_rate": 2.7926e-06, "loss": 0.0334, "step": 36038 }, { "epoch": 1.9213626912619288, "grad_norm": 1.9365553855895996, "learning_rate": 2.7924e-06, "loss": 0.0078, "step": 36039 }, { "epoch": 1.9214160046915818, "grad_norm": 3.03505539894104, "learning_rate": 2.7922000000000006e-06, "loss": 0.0165, "step": 36040 }, { "epoch": 1.9214693181212348, "grad_norm": 0.7816510200500488, "learning_rate": 2.792e-06, "loss": 0.006, "step": 36041 }, { "epoch": 1.9215226315508875, "grad_norm": 1.527730107307434, "learning_rate": 2.7918e-06, "loss": 0.0346, "step": 36042 }, { "epoch": 1.9215759449805407, "grad_norm": 2.2113850116729736, "learning_rate": 2.7916000000000005e-06, "loss": 0.0176, "step": 36043 }, { "epoch": 1.9216292584101935, "grad_norm": 1.0285491943359375, "learning_rate": 2.7914e-06, "loss": 0.0182, "step": 36044 }, { "epoch": 1.9216825718398465, "grad_norm": 1.333314299583435, "learning_rate": 2.7912e-06, "loss": 0.0289, "step": 36045 }, { "epoch": 1.9217358852694995, "grad_norm": 1.3749475479125977, "learning_rate": 2.7910000000000005e-06, "loss": 0.0158, "step": 36046 }, { "epoch": 1.9217891986991522, "grad_norm": 1.7403364181518555, "learning_rate": 2.7908e-06, "loss": 0.0304, "step": 36047 }, { "epoch": 1.9218425121288052, "grad_norm": 0.5762293934822083, "learning_rate": 2.7906e-06, "loss": 0.0037, "step": 36048 }, { "epoch": 1.9218958255584582, "grad_norm": 1.728559970855713, "learning_rate": 2.7904000000000004e-06, "loss": 0.0114, "step": 36049 }, { "epoch": 1.9219491389881111, "grad_norm": 1.5522602796554565, "learning_rate": 2.7902e-06, "loss": 0.0308, "step": 36050 }, { "epoch": 1.9220024524177641, "grad_norm": 1.6498169898986816, "learning_rate": 2.7900000000000004e-06, "loss": 0.021, "step": 36051 }, { "epoch": 1.9220557658474169, "grad_norm": 1.824936032295227, "learning_rate": 2.7898000000000004e-06, "loss": 0.0207, "step": 36052 }, { "epoch": 1.9221090792770699, "grad_norm": 3.052647829055786, "learning_rate": 2.7896e-06, "loss": 0.0367, "step": 36053 }, { "epoch": 1.9221623927067228, "grad_norm": 3.118522882461548, "learning_rate": 2.7894000000000003e-06, "loss": 0.0208, "step": 36054 }, { "epoch": 1.9222157061363756, "grad_norm": 2.268433094024658, "learning_rate": 2.7892000000000003e-06, "loss": 0.0354, "step": 36055 }, { "epoch": 1.9222690195660288, "grad_norm": 0.755260705947876, "learning_rate": 2.789e-06, "loss": 0.0071, "step": 36056 }, { "epoch": 1.9223223329956816, "grad_norm": 3.0594582557678223, "learning_rate": 2.7888000000000003e-06, "loss": 0.0155, "step": 36057 }, { "epoch": 1.9223756464253345, "grad_norm": 1.477699875831604, "learning_rate": 2.7886000000000003e-06, "loss": 0.0129, "step": 36058 }, { "epoch": 1.9224289598549875, "grad_norm": 1.445611596107483, "learning_rate": 2.7884e-06, "loss": 0.0098, "step": 36059 }, { "epoch": 1.9224822732846403, "grad_norm": 2.061843156814575, "learning_rate": 2.7882000000000003e-06, "loss": 0.0173, "step": 36060 }, { "epoch": 1.9225355867142935, "grad_norm": 2.233645439147949, "learning_rate": 2.7880000000000002e-06, "loss": 0.0226, "step": 36061 }, { "epoch": 1.9225889001439462, "grad_norm": 3.2791810035705566, "learning_rate": 2.7878e-06, "loss": 0.0183, "step": 36062 }, { "epoch": 1.9226422135735992, "grad_norm": 2.9552221298217773, "learning_rate": 2.7876000000000002e-06, "loss": 0.0154, "step": 36063 }, { "epoch": 1.9226955270032522, "grad_norm": 0.41489967703819275, "learning_rate": 2.7874e-06, "loss": 0.0039, "step": 36064 }, { "epoch": 1.922748840432905, "grad_norm": 1.6916764974594116, "learning_rate": 2.7872000000000006e-06, "loss": 0.0201, "step": 36065 }, { "epoch": 1.922802153862558, "grad_norm": 4.281238079071045, "learning_rate": 2.787e-06, "loss": 0.0238, "step": 36066 }, { "epoch": 1.922855467292211, "grad_norm": 0.5296254754066467, "learning_rate": 2.7868e-06, "loss": 0.0044, "step": 36067 }, { "epoch": 1.9229087807218639, "grad_norm": 1.1673442125320435, "learning_rate": 2.7866000000000006e-06, "loss": 0.0099, "step": 36068 }, { "epoch": 1.9229620941515169, "grad_norm": 0.48519763350486755, "learning_rate": 2.7864e-06, "loss": 0.0037, "step": 36069 }, { "epoch": 1.9230154075811696, "grad_norm": 2.0833284854888916, "learning_rate": 2.7862e-06, "loss": 0.0091, "step": 36070 }, { "epoch": 1.9230687210108226, "grad_norm": 1.1109799146652222, "learning_rate": 2.7860000000000005e-06, "loss": 0.0062, "step": 36071 }, { "epoch": 1.9231220344404756, "grad_norm": 2.3864591121673584, "learning_rate": 2.7858e-06, "loss": 0.028, "step": 36072 }, { "epoch": 1.9231753478701283, "grad_norm": 0.8027107119560242, "learning_rate": 2.7856e-06, "loss": 0.0042, "step": 36073 }, { "epoch": 1.9232286612997815, "grad_norm": 0.8890224099159241, "learning_rate": 2.7854000000000005e-06, "loss": 0.0135, "step": 36074 }, { "epoch": 1.9232819747294343, "grad_norm": 2.3604719638824463, "learning_rate": 2.7852e-06, "loss": 0.0191, "step": 36075 }, { "epoch": 1.9233352881590873, "grad_norm": 1.4429521560668945, "learning_rate": 2.7850000000000004e-06, "loss": 0.0232, "step": 36076 }, { "epoch": 1.9233886015887403, "grad_norm": 2.416579008102417, "learning_rate": 2.7848000000000004e-06, "loss": 0.0263, "step": 36077 }, { "epoch": 1.923441915018393, "grad_norm": 2.3630242347717285, "learning_rate": 2.7846e-06, "loss": 0.0187, "step": 36078 }, { "epoch": 1.9234952284480462, "grad_norm": 0.7053790092468262, "learning_rate": 2.7844000000000004e-06, "loss": 0.0028, "step": 36079 }, { "epoch": 1.923548541877699, "grad_norm": 2.66817569732666, "learning_rate": 2.7842000000000004e-06, "loss": 0.072, "step": 36080 }, { "epoch": 1.923601855307352, "grad_norm": 1.0191644430160522, "learning_rate": 2.784e-06, "loss": 0.0178, "step": 36081 }, { "epoch": 1.923655168737005, "grad_norm": 2.50541353225708, "learning_rate": 2.7838000000000004e-06, "loss": 0.0315, "step": 36082 }, { "epoch": 1.9237084821666577, "grad_norm": 0.8568880558013916, "learning_rate": 2.7836000000000003e-06, "loss": 0.0053, "step": 36083 }, { "epoch": 1.9237617955963107, "grad_norm": 1.0948498249053955, "learning_rate": 2.7834e-06, "loss": 0.0059, "step": 36084 }, { "epoch": 1.9238151090259636, "grad_norm": 1.3702847957611084, "learning_rate": 2.7832000000000003e-06, "loss": 0.0109, "step": 36085 }, { "epoch": 1.9238684224556166, "grad_norm": 0.698796808719635, "learning_rate": 2.7830000000000003e-06, "loss": 0.0047, "step": 36086 }, { "epoch": 1.9239217358852696, "grad_norm": 1.1294299364089966, "learning_rate": 2.7828000000000007e-06, "loss": 0.0124, "step": 36087 }, { "epoch": 1.9239750493149224, "grad_norm": 3.8445889949798584, "learning_rate": 2.7826000000000003e-06, "loss": 0.0311, "step": 36088 }, { "epoch": 1.9240283627445753, "grad_norm": 1.594892978668213, "learning_rate": 2.7824000000000002e-06, "loss": 0.0104, "step": 36089 }, { "epoch": 1.9240816761742283, "grad_norm": 2.5905091762542725, "learning_rate": 2.7822000000000007e-06, "loss": 0.0328, "step": 36090 }, { "epoch": 1.924134989603881, "grad_norm": 1.5586130619049072, "learning_rate": 2.7820000000000002e-06, "loss": 0.0071, "step": 36091 }, { "epoch": 1.9241883030335343, "grad_norm": 1.3695451021194458, "learning_rate": 2.7818e-06, "loss": 0.0073, "step": 36092 }, { "epoch": 1.924241616463187, "grad_norm": 1.3170771598815918, "learning_rate": 2.7816000000000006e-06, "loss": 0.016, "step": 36093 }, { "epoch": 1.92429492989284, "grad_norm": 1.50661039352417, "learning_rate": 2.7814e-06, "loss": 0.0103, "step": 36094 }, { "epoch": 1.924348243322493, "grad_norm": 1.6893320083618164, "learning_rate": 2.7812e-06, "loss": 0.0099, "step": 36095 }, { "epoch": 1.9244015567521457, "grad_norm": 1.9406259059906006, "learning_rate": 2.7810000000000006e-06, "loss": 0.011, "step": 36096 }, { "epoch": 1.924454870181799, "grad_norm": 2.377215623855591, "learning_rate": 2.7808e-06, "loss": 0.0316, "step": 36097 }, { "epoch": 1.9245081836114517, "grad_norm": 1.800013542175293, "learning_rate": 2.7806e-06, "loss": 0.0118, "step": 36098 }, { "epoch": 1.9245614970411047, "grad_norm": 4.742552757263184, "learning_rate": 2.7804000000000005e-06, "loss": 0.012, "step": 36099 }, { "epoch": 1.9246148104707577, "grad_norm": 2.2448997497558594, "learning_rate": 2.7802e-06, "loss": 0.015, "step": 36100 }, { "epoch": 1.9246681239004104, "grad_norm": 1.319154143333435, "learning_rate": 2.7800000000000005e-06, "loss": 0.0248, "step": 36101 }, { "epoch": 1.9247214373300634, "grad_norm": 1.2979964017868042, "learning_rate": 2.7798000000000005e-06, "loss": 0.0088, "step": 36102 }, { "epoch": 1.9247747507597164, "grad_norm": 1.7760844230651855, "learning_rate": 2.7796e-06, "loss": 0.0194, "step": 36103 }, { "epoch": 1.9248280641893691, "grad_norm": 2.9615042209625244, "learning_rate": 2.7794000000000004e-06, "loss": 0.0248, "step": 36104 }, { "epoch": 1.9248813776190223, "grad_norm": 0.33858364820480347, "learning_rate": 2.7792000000000004e-06, "loss": 0.0036, "step": 36105 }, { "epoch": 1.924934691048675, "grad_norm": 1.933010220527649, "learning_rate": 2.779e-06, "loss": 0.0359, "step": 36106 }, { "epoch": 1.924988004478328, "grad_norm": 1.101179838180542, "learning_rate": 2.7788000000000004e-06, "loss": 0.026, "step": 36107 }, { "epoch": 1.925041317907981, "grad_norm": 1.9435094594955444, "learning_rate": 2.7786000000000004e-06, "loss": 0.0297, "step": 36108 }, { "epoch": 1.9250946313376338, "grad_norm": 1.5915372371673584, "learning_rate": 2.7784e-06, "loss": 0.0292, "step": 36109 }, { "epoch": 1.925147944767287, "grad_norm": 2.3281075954437256, "learning_rate": 2.7782000000000004e-06, "loss": 0.0265, "step": 36110 }, { "epoch": 1.9252012581969398, "grad_norm": 2.264436721801758, "learning_rate": 2.7780000000000003e-06, "loss": 0.0219, "step": 36111 }, { "epoch": 1.9252545716265927, "grad_norm": 0.8547400236129761, "learning_rate": 2.7778000000000003e-06, "loss": 0.0038, "step": 36112 }, { "epoch": 1.9253078850562457, "grad_norm": 3.3634023666381836, "learning_rate": 2.7776000000000003e-06, "loss": 0.0225, "step": 36113 }, { "epoch": 1.9253611984858985, "grad_norm": 1.2311218976974487, "learning_rate": 2.7774000000000003e-06, "loss": 0.0068, "step": 36114 }, { "epoch": 1.9254145119155517, "grad_norm": 1.2011603116989136, "learning_rate": 2.7772000000000003e-06, "loss": 0.0235, "step": 36115 }, { "epoch": 1.9254678253452044, "grad_norm": 1.1682322025299072, "learning_rate": 2.7770000000000003e-06, "loss": 0.0061, "step": 36116 }, { "epoch": 1.9255211387748574, "grad_norm": 1.7540209293365479, "learning_rate": 2.7768000000000002e-06, "loss": 0.0358, "step": 36117 }, { "epoch": 1.9255744522045104, "grad_norm": 1.633716344833374, "learning_rate": 2.7766000000000002e-06, "loss": 0.0149, "step": 36118 }, { "epoch": 1.9256277656341632, "grad_norm": 2.4811651706695557, "learning_rate": 2.7764000000000002e-06, "loss": 0.01, "step": 36119 }, { "epoch": 1.9256810790638161, "grad_norm": 0.8238120675086975, "learning_rate": 2.7762e-06, "loss": 0.0042, "step": 36120 }, { "epoch": 1.9257343924934691, "grad_norm": 7.263576984405518, "learning_rate": 2.776e-06, "loss": 0.0397, "step": 36121 }, { "epoch": 1.9257877059231219, "grad_norm": 1.2724413871765137, "learning_rate": 2.7758e-06, "loss": 0.0316, "step": 36122 }, { "epoch": 1.925841019352775, "grad_norm": 2.187664270401001, "learning_rate": 2.7756e-06, "loss": 0.0264, "step": 36123 }, { "epoch": 1.9258943327824278, "grad_norm": 1.225266933441162, "learning_rate": 2.7754e-06, "loss": 0.0076, "step": 36124 }, { "epoch": 1.9259476462120808, "grad_norm": 2.237307071685791, "learning_rate": 2.7752e-06, "loss": 0.0291, "step": 36125 }, { "epoch": 1.9260009596417338, "grad_norm": 4.227133274078369, "learning_rate": 2.7750000000000005e-06, "loss": 0.0104, "step": 36126 }, { "epoch": 1.9260542730713865, "grad_norm": 2.9126410484313965, "learning_rate": 2.7748e-06, "loss": 0.0481, "step": 36127 }, { "epoch": 1.9261075865010397, "grad_norm": 1.4245201349258423, "learning_rate": 2.7746e-06, "loss": 0.0306, "step": 36128 }, { "epoch": 1.9261608999306925, "grad_norm": 2.610056161880493, "learning_rate": 2.7744000000000005e-06, "loss": 0.0297, "step": 36129 }, { "epoch": 1.9262142133603455, "grad_norm": 0.4131240248680115, "learning_rate": 2.7742e-06, "loss": 0.004, "step": 36130 }, { "epoch": 1.9262675267899985, "grad_norm": 2.554849624633789, "learning_rate": 2.774e-06, "loss": 0.0446, "step": 36131 }, { "epoch": 1.9263208402196512, "grad_norm": 1.8808566331863403, "learning_rate": 2.7738000000000004e-06, "loss": 0.0195, "step": 36132 }, { "epoch": 1.9263741536493044, "grad_norm": 2.030095100402832, "learning_rate": 2.7736e-06, "loss": 0.0278, "step": 36133 }, { "epoch": 1.9264274670789572, "grad_norm": 1.4220727682113647, "learning_rate": 2.7734e-06, "loss": 0.0172, "step": 36134 }, { "epoch": 1.9264807805086102, "grad_norm": 0.6914474368095398, "learning_rate": 2.7732000000000004e-06, "loss": 0.0053, "step": 36135 }, { "epoch": 1.9265340939382631, "grad_norm": 2.522045850753784, "learning_rate": 2.773e-06, "loss": 0.0197, "step": 36136 }, { "epoch": 1.926587407367916, "grad_norm": 3.446361780166626, "learning_rate": 2.7728000000000004e-06, "loss": 0.0248, "step": 36137 }, { "epoch": 1.9266407207975689, "grad_norm": 1.4739937782287598, "learning_rate": 2.7726000000000004e-06, "loss": 0.0184, "step": 36138 }, { "epoch": 1.9266940342272219, "grad_norm": 2.095515251159668, "learning_rate": 2.7724e-06, "loss": 0.0175, "step": 36139 }, { "epoch": 1.9267473476568746, "grad_norm": 1.2123022079467773, "learning_rate": 2.7722000000000003e-06, "loss": 0.0074, "step": 36140 }, { "epoch": 1.9268006610865278, "grad_norm": 2.288300037384033, "learning_rate": 2.7720000000000003e-06, "loss": 0.0299, "step": 36141 }, { "epoch": 1.9268539745161806, "grad_norm": 2.3866798877716064, "learning_rate": 2.7718e-06, "loss": 0.0124, "step": 36142 }, { "epoch": 1.9269072879458335, "grad_norm": 1.1051955223083496, "learning_rate": 2.7716000000000003e-06, "loss": 0.0077, "step": 36143 }, { "epoch": 1.9269606013754865, "grad_norm": 2.393836259841919, "learning_rate": 2.7714000000000003e-06, "loss": 0.0246, "step": 36144 }, { "epoch": 1.9270139148051393, "grad_norm": 2.2464547157287598, "learning_rate": 2.7712e-06, "loss": 0.0381, "step": 36145 }, { "epoch": 1.9270672282347925, "grad_norm": 2.0619893074035645, "learning_rate": 2.7710000000000002e-06, "loss": 0.0165, "step": 36146 }, { "epoch": 1.9271205416644452, "grad_norm": 2.411245584487915, "learning_rate": 2.7708000000000002e-06, "loss": 0.0436, "step": 36147 }, { "epoch": 1.9271738550940982, "grad_norm": 0.41312018036842346, "learning_rate": 2.7705999999999998e-06, "loss": 0.0038, "step": 36148 }, { "epoch": 1.9272271685237512, "grad_norm": 1.4775004386901855, "learning_rate": 2.7704e-06, "loss": 0.0197, "step": 36149 }, { "epoch": 1.927280481953404, "grad_norm": 2.2825286388397217, "learning_rate": 2.7702e-06, "loss": 0.0147, "step": 36150 }, { "epoch": 1.9273337953830572, "grad_norm": 2.0556631088256836, "learning_rate": 2.7700000000000006e-06, "loss": 0.0139, "step": 36151 }, { "epoch": 1.92738710881271, "grad_norm": 2.0819950103759766, "learning_rate": 2.7698e-06, "loss": 0.0112, "step": 36152 }, { "epoch": 1.927440422242363, "grad_norm": 1.231802225112915, "learning_rate": 2.7696e-06, "loss": 0.0089, "step": 36153 }, { "epoch": 1.9274937356720159, "grad_norm": 1.1626198291778564, "learning_rate": 2.7694000000000005e-06, "loss": 0.0096, "step": 36154 }, { "epoch": 1.9275470491016686, "grad_norm": 3.1894185543060303, "learning_rate": 2.7692e-06, "loss": 0.0179, "step": 36155 }, { "epoch": 1.9276003625313216, "grad_norm": 1.730418086051941, "learning_rate": 2.769e-06, "loss": 0.0304, "step": 36156 }, { "epoch": 1.9276536759609746, "grad_norm": 3.395880699157715, "learning_rate": 2.7688000000000005e-06, "loss": 0.021, "step": 36157 }, { "epoch": 1.9277069893906273, "grad_norm": 1.42877197265625, "learning_rate": 2.7686e-06, "loss": 0.0071, "step": 36158 }, { "epoch": 1.9277603028202805, "grad_norm": 1.7062294483184814, "learning_rate": 2.7684e-06, "loss": 0.0263, "step": 36159 }, { "epoch": 1.9278136162499333, "grad_norm": 2.302011251449585, "learning_rate": 2.7682000000000004e-06, "loss": 0.0375, "step": 36160 }, { "epoch": 1.9278669296795863, "grad_norm": 0.22921545803546906, "learning_rate": 2.768e-06, "loss": 0.0019, "step": 36161 }, { "epoch": 1.9279202431092393, "grad_norm": 2.08715558052063, "learning_rate": 2.7678000000000004e-06, "loss": 0.0187, "step": 36162 }, { "epoch": 1.927973556538892, "grad_norm": 1.2108556032180786, "learning_rate": 2.7676000000000004e-06, "loss": 0.0218, "step": 36163 }, { "epoch": 1.9280268699685452, "grad_norm": 0.8825803995132446, "learning_rate": 2.7674e-06, "loss": 0.0049, "step": 36164 }, { "epoch": 1.928080183398198, "grad_norm": 1.3276363611221313, "learning_rate": 2.7672000000000004e-06, "loss": 0.0186, "step": 36165 }, { "epoch": 1.928133496827851, "grad_norm": 2.3092474937438965, "learning_rate": 2.7670000000000004e-06, "loss": 0.0219, "step": 36166 }, { "epoch": 1.928186810257504, "grad_norm": 1.0916447639465332, "learning_rate": 2.7668e-06, "loss": 0.0225, "step": 36167 }, { "epoch": 1.9282401236871567, "grad_norm": 1.3669981956481934, "learning_rate": 2.7666000000000003e-06, "loss": 0.0108, "step": 36168 }, { "epoch": 1.9282934371168097, "grad_norm": 2.171147108078003, "learning_rate": 2.7664000000000003e-06, "loss": 0.0448, "step": 36169 }, { "epoch": 1.9283467505464627, "grad_norm": 1.853022575378418, "learning_rate": 2.7662e-06, "loss": 0.0203, "step": 36170 }, { "epoch": 1.9284000639761156, "grad_norm": 1.4107952117919922, "learning_rate": 2.7660000000000003e-06, "loss": 0.0223, "step": 36171 }, { "epoch": 1.9284533774057686, "grad_norm": 1.6312702894210815, "learning_rate": 2.7658000000000003e-06, "loss": 0.0233, "step": 36172 }, { "epoch": 1.9285066908354214, "grad_norm": 3.759091854095459, "learning_rate": 2.7656e-06, "loss": 0.0224, "step": 36173 }, { "epoch": 1.9285600042650743, "grad_norm": 2.7879714965820312, "learning_rate": 2.7654000000000002e-06, "loss": 0.0249, "step": 36174 }, { "epoch": 1.9286133176947273, "grad_norm": 1.7020434141159058, "learning_rate": 2.7652000000000002e-06, "loss": 0.0067, "step": 36175 }, { "epoch": 1.92866663112438, "grad_norm": 4.654016971588135, "learning_rate": 2.7650000000000006e-06, "loss": 0.0447, "step": 36176 }, { "epoch": 1.9287199445540333, "grad_norm": 2.6855809688568115, "learning_rate": 2.7648e-06, "loss": 0.0136, "step": 36177 }, { "epoch": 1.928773257983686, "grad_norm": 2.930572509765625, "learning_rate": 2.7646e-06, "loss": 0.0188, "step": 36178 }, { "epoch": 1.928826571413339, "grad_norm": 1.1976597309112549, "learning_rate": 2.7644000000000006e-06, "loss": 0.0155, "step": 36179 }, { "epoch": 1.928879884842992, "grad_norm": 2.431027889251709, "learning_rate": 2.7642e-06, "loss": 0.0171, "step": 36180 }, { "epoch": 1.9289331982726448, "grad_norm": 3.282315969467163, "learning_rate": 2.764e-06, "loss": 0.0125, "step": 36181 }, { "epoch": 1.928986511702298, "grad_norm": 1.4009360074996948, "learning_rate": 2.7638000000000005e-06, "loss": 0.0086, "step": 36182 }, { "epoch": 1.9290398251319507, "grad_norm": 2.3637869358062744, "learning_rate": 2.7636e-06, "loss": 0.0472, "step": 36183 }, { "epoch": 1.9290931385616037, "grad_norm": 1.669196367263794, "learning_rate": 2.7634e-06, "loss": 0.0319, "step": 36184 }, { "epoch": 1.9291464519912567, "grad_norm": 2.0003693103790283, "learning_rate": 2.7632000000000005e-06, "loss": 0.0117, "step": 36185 }, { "epoch": 1.9291997654209094, "grad_norm": 1.3911707401275635, "learning_rate": 2.763e-06, "loss": 0.0131, "step": 36186 }, { "epoch": 1.9292530788505624, "grad_norm": 0.3627566397190094, "learning_rate": 2.7628000000000005e-06, "loss": 0.0031, "step": 36187 }, { "epoch": 1.9293063922802154, "grad_norm": 1.945072054862976, "learning_rate": 2.7626000000000005e-06, "loss": 0.0129, "step": 36188 }, { "epoch": 1.9293597057098684, "grad_norm": 0.35375291109085083, "learning_rate": 2.7624e-06, "loss": 0.0026, "step": 36189 }, { "epoch": 1.9294130191395213, "grad_norm": 1.9249110221862793, "learning_rate": 2.7622000000000004e-06, "loss": 0.008, "step": 36190 }, { "epoch": 1.929466332569174, "grad_norm": 2.623432159423828, "learning_rate": 2.7620000000000004e-06, "loss": 0.0137, "step": 36191 }, { "epoch": 1.929519645998827, "grad_norm": 4.343924045562744, "learning_rate": 2.7618e-06, "loss": 0.0202, "step": 36192 }, { "epoch": 1.92957295942848, "grad_norm": 0.8801380395889282, "learning_rate": 2.7616000000000004e-06, "loss": 0.0061, "step": 36193 }, { "epoch": 1.9296262728581328, "grad_norm": 0.36620911955833435, "learning_rate": 2.7614000000000004e-06, "loss": 0.0024, "step": 36194 }, { "epoch": 1.929679586287786, "grad_norm": 3.3494820594787598, "learning_rate": 2.7612e-06, "loss": 0.0221, "step": 36195 }, { "epoch": 1.9297328997174388, "grad_norm": 3.0539708137512207, "learning_rate": 2.7610000000000003e-06, "loss": 0.0324, "step": 36196 }, { "epoch": 1.9297862131470918, "grad_norm": 1.2214118242263794, "learning_rate": 2.7608000000000003e-06, "loss": 0.0155, "step": 36197 }, { "epoch": 1.9298395265767447, "grad_norm": 0.8854307532310486, "learning_rate": 2.7606000000000003e-06, "loss": 0.0145, "step": 36198 }, { "epoch": 1.9298928400063975, "grad_norm": 1.8737516403198242, "learning_rate": 2.7604000000000003e-06, "loss": 0.0238, "step": 36199 }, { "epoch": 1.9299461534360507, "grad_norm": 2.995272159576416, "learning_rate": 2.7602000000000003e-06, "loss": 0.0223, "step": 36200 }, { "epoch": 1.9299994668657035, "grad_norm": 4.311161994934082, "learning_rate": 2.7600000000000003e-06, "loss": 0.0308, "step": 36201 }, { "epoch": 1.9300527802953564, "grad_norm": 2.669275999069214, "learning_rate": 2.7598000000000002e-06, "loss": 0.0383, "step": 36202 }, { "epoch": 1.9301060937250094, "grad_norm": 1.0095980167388916, "learning_rate": 2.7596000000000002e-06, "loss": 0.0048, "step": 36203 }, { "epoch": 1.9301594071546622, "grad_norm": 3.5410985946655273, "learning_rate": 2.7594e-06, "loss": 0.0398, "step": 36204 }, { "epoch": 1.9302127205843151, "grad_norm": 1.8823970556259155, "learning_rate": 2.7592e-06, "loss": 0.0171, "step": 36205 }, { "epoch": 1.9302660340139681, "grad_norm": 2.445533037185669, "learning_rate": 2.759e-06, "loss": 0.0259, "step": 36206 }, { "epoch": 1.930319347443621, "grad_norm": 3.4559874534606934, "learning_rate": 2.7588e-06, "loss": 0.0162, "step": 36207 }, { "epoch": 1.930372660873274, "grad_norm": 1.7987785339355469, "learning_rate": 2.7586e-06, "loss": 0.0349, "step": 36208 }, { "epoch": 1.9304259743029268, "grad_norm": 1.330963373184204, "learning_rate": 2.7584e-06, "loss": 0.0189, "step": 36209 }, { "epoch": 1.9304792877325798, "grad_norm": 2.5408971309661865, "learning_rate": 2.7582e-06, "loss": 0.0197, "step": 36210 }, { "epoch": 1.9305326011622328, "grad_norm": 1.469610571861267, "learning_rate": 2.758e-06, "loss": 0.0089, "step": 36211 }, { "epoch": 1.9305859145918856, "grad_norm": 1.0427489280700684, "learning_rate": 2.7578000000000005e-06, "loss": 0.0092, "step": 36212 }, { "epoch": 1.9306392280215388, "grad_norm": 2.023543119430542, "learning_rate": 2.7576e-06, "loss": 0.0143, "step": 36213 }, { "epoch": 1.9306925414511915, "grad_norm": 2.5044281482696533, "learning_rate": 2.7574e-06, "loss": 0.0167, "step": 36214 }, { "epoch": 1.9307458548808445, "grad_norm": 1.4088044166564941, "learning_rate": 2.7572000000000005e-06, "loss": 0.0079, "step": 36215 }, { "epoch": 1.9307991683104975, "grad_norm": 2.40859317779541, "learning_rate": 2.757e-06, "loss": 0.0452, "step": 36216 }, { "epoch": 1.9308524817401502, "grad_norm": 1.489300012588501, "learning_rate": 2.7568e-06, "loss": 0.0189, "step": 36217 }, { "epoch": 1.9309057951698034, "grad_norm": 1.2369877099990845, "learning_rate": 2.7566000000000004e-06, "loss": 0.0116, "step": 36218 }, { "epoch": 1.9309591085994562, "grad_norm": 1.3898143768310547, "learning_rate": 2.7564e-06, "loss": 0.0202, "step": 36219 }, { "epoch": 1.9310124220291092, "grad_norm": 1.9247982501983643, "learning_rate": 2.7562e-06, "loss": 0.0164, "step": 36220 }, { "epoch": 1.9310657354587621, "grad_norm": 2.962521553039551, "learning_rate": 2.7560000000000004e-06, "loss": 0.0157, "step": 36221 }, { "epoch": 1.931119048888415, "grad_norm": 0.9099115133285522, "learning_rate": 2.7558e-06, "loss": 0.0044, "step": 36222 }, { "epoch": 1.9311723623180679, "grad_norm": 2.6715962886810303, "learning_rate": 2.7556000000000003e-06, "loss": 0.0277, "step": 36223 }, { "epoch": 1.9312256757477209, "grad_norm": 2.248180866241455, "learning_rate": 2.7554000000000003e-06, "loss": 0.016, "step": 36224 }, { "epoch": 1.9312789891773736, "grad_norm": 2.0015993118286133, "learning_rate": 2.7552e-06, "loss": 0.0105, "step": 36225 }, { "epoch": 1.9313323026070268, "grad_norm": 2.3847570419311523, "learning_rate": 2.7550000000000003e-06, "loss": 0.0218, "step": 36226 }, { "epoch": 1.9313856160366796, "grad_norm": 1.9371941089630127, "learning_rate": 2.7548000000000003e-06, "loss": 0.0237, "step": 36227 }, { "epoch": 1.9314389294663326, "grad_norm": 0.43536511063575745, "learning_rate": 2.7546e-06, "loss": 0.0028, "step": 36228 }, { "epoch": 1.9314922428959855, "grad_norm": 1.3074299097061157, "learning_rate": 2.7544000000000003e-06, "loss": 0.0222, "step": 36229 }, { "epoch": 1.9315455563256383, "grad_norm": 1.7028794288635254, "learning_rate": 2.7542000000000002e-06, "loss": 0.0084, "step": 36230 }, { "epoch": 1.9315988697552915, "grad_norm": 1.7529693841934204, "learning_rate": 2.754e-06, "loss": 0.0459, "step": 36231 }, { "epoch": 1.9316521831849442, "grad_norm": 5.025115489959717, "learning_rate": 2.7538000000000002e-06, "loss": 0.0261, "step": 36232 }, { "epoch": 1.9317054966145972, "grad_norm": 1.7941049337387085, "learning_rate": 2.7536e-06, "loss": 0.0367, "step": 36233 }, { "epoch": 1.9317588100442502, "grad_norm": 0.642193615436554, "learning_rate": 2.7533999999999998e-06, "loss": 0.0043, "step": 36234 }, { "epoch": 1.931812123473903, "grad_norm": 2.759577751159668, "learning_rate": 2.7532e-06, "loss": 0.0113, "step": 36235 }, { "epoch": 1.9318654369035562, "grad_norm": 4.391677379608154, "learning_rate": 2.753e-06, "loss": 0.0242, "step": 36236 }, { "epoch": 1.931918750333209, "grad_norm": 0.5039169192314148, "learning_rate": 2.7528000000000006e-06, "loss": 0.0041, "step": 36237 }, { "epoch": 1.931972063762862, "grad_norm": 1.5599766969680786, "learning_rate": 2.7526e-06, "loss": 0.0119, "step": 36238 }, { "epoch": 1.9320253771925149, "grad_norm": 0.8980355262756348, "learning_rate": 2.7524e-06, "loss": 0.017, "step": 36239 }, { "epoch": 1.9320786906221676, "grad_norm": 1.642728567123413, "learning_rate": 2.7522000000000005e-06, "loss": 0.0087, "step": 36240 }, { "epoch": 1.9321320040518206, "grad_norm": 0.3554241955280304, "learning_rate": 2.752e-06, "loss": 0.0017, "step": 36241 }, { "epoch": 1.9321853174814736, "grad_norm": 3.486517906188965, "learning_rate": 2.7518e-06, "loss": 0.0302, "step": 36242 }, { "epoch": 1.9322386309111264, "grad_norm": 0.9268601536750793, "learning_rate": 2.7516000000000005e-06, "loss": 0.0158, "step": 36243 }, { "epoch": 1.9322919443407796, "grad_norm": 3.3704137802124023, "learning_rate": 2.7514e-06, "loss": 0.024, "step": 36244 }, { "epoch": 1.9323452577704323, "grad_norm": 1.8872077465057373, "learning_rate": 2.7512e-06, "loss": 0.0296, "step": 36245 }, { "epoch": 1.9323985712000853, "grad_norm": 1.8007943630218506, "learning_rate": 2.7510000000000004e-06, "loss": 0.0079, "step": 36246 }, { "epoch": 1.9324518846297383, "grad_norm": 0.8321388959884644, "learning_rate": 2.7508e-06, "loss": 0.0053, "step": 36247 }, { "epoch": 1.932505198059391, "grad_norm": 1.7162398099899292, "learning_rate": 2.7506000000000004e-06, "loss": 0.0303, "step": 36248 }, { "epoch": 1.9325585114890442, "grad_norm": 1.627479910850525, "learning_rate": 2.7504000000000004e-06, "loss": 0.0173, "step": 36249 }, { "epoch": 1.932611824918697, "grad_norm": 0.4493265151977539, "learning_rate": 2.7502e-06, "loss": 0.0035, "step": 36250 }, { "epoch": 1.93266513834835, "grad_norm": 1.0746536254882812, "learning_rate": 2.7500000000000004e-06, "loss": 0.01, "step": 36251 }, { "epoch": 1.932718451778003, "grad_norm": 1.7640506029129028, "learning_rate": 2.7498000000000003e-06, "loss": 0.0184, "step": 36252 }, { "epoch": 1.9327717652076557, "grad_norm": 2.376678943634033, "learning_rate": 2.7496e-06, "loss": 0.0192, "step": 36253 }, { "epoch": 1.932825078637309, "grad_norm": 0.6230000853538513, "learning_rate": 2.7494000000000003e-06, "loss": 0.0049, "step": 36254 }, { "epoch": 1.9328783920669617, "grad_norm": 2.2645390033721924, "learning_rate": 2.7492000000000003e-06, "loss": 0.0191, "step": 36255 }, { "epoch": 1.9329317054966146, "grad_norm": 1.4768550395965576, "learning_rate": 2.749e-06, "loss": 0.0121, "step": 36256 }, { "epoch": 1.9329850189262676, "grad_norm": 2.849093437194824, "learning_rate": 2.7488000000000003e-06, "loss": 0.028, "step": 36257 }, { "epoch": 1.9330383323559204, "grad_norm": 2.3351595401763916, "learning_rate": 2.7486000000000002e-06, "loss": 0.0101, "step": 36258 }, { "epoch": 1.9330916457855734, "grad_norm": 1.9991594552993774, "learning_rate": 2.7484e-06, "loss": 0.0364, "step": 36259 }, { "epoch": 1.9331449592152263, "grad_norm": 1.9354238510131836, "learning_rate": 2.7482000000000002e-06, "loss": 0.0441, "step": 36260 }, { "epoch": 1.933198272644879, "grad_norm": 1.5181419849395752, "learning_rate": 2.748e-06, "loss": 0.0072, "step": 36261 }, { "epoch": 1.9332515860745323, "grad_norm": 2.8661463260650635, "learning_rate": 2.7478000000000006e-06, "loss": 0.0198, "step": 36262 }, { "epoch": 1.933304899504185, "grad_norm": 1.8451646566390991, "learning_rate": 2.7476e-06, "loss": 0.032, "step": 36263 }, { "epoch": 1.933358212933838, "grad_norm": 1.3504997491836548, "learning_rate": 2.7474e-06, "loss": 0.0107, "step": 36264 }, { "epoch": 1.933411526363491, "grad_norm": 1.0210795402526855, "learning_rate": 2.7472000000000006e-06, "loss": 0.0054, "step": 36265 }, { "epoch": 1.9334648397931438, "grad_norm": 2.575373649597168, "learning_rate": 2.747e-06, "loss": 0.0388, "step": 36266 }, { "epoch": 1.933518153222797, "grad_norm": 2.507143497467041, "learning_rate": 2.7468e-06, "loss": 0.018, "step": 36267 }, { "epoch": 1.9335714666524497, "grad_norm": 0.7237115502357483, "learning_rate": 2.7466000000000005e-06, "loss": 0.0048, "step": 36268 }, { "epoch": 1.9336247800821027, "grad_norm": 1.7267428636550903, "learning_rate": 2.7464e-06, "loss": 0.0079, "step": 36269 }, { "epoch": 1.9336780935117557, "grad_norm": 1.8557789325714111, "learning_rate": 2.7462e-06, "loss": 0.0305, "step": 36270 }, { "epoch": 1.9337314069414084, "grad_norm": 1.5815472602844238, "learning_rate": 2.7460000000000005e-06, "loss": 0.0104, "step": 36271 }, { "epoch": 1.9337847203710614, "grad_norm": 0.844390869140625, "learning_rate": 2.7458e-06, "loss": 0.0043, "step": 36272 }, { "epoch": 1.9338380338007144, "grad_norm": 1.681167483329773, "learning_rate": 2.7456000000000004e-06, "loss": 0.0229, "step": 36273 }, { "epoch": 1.9338913472303674, "grad_norm": 0.7825378775596619, "learning_rate": 2.7454000000000004e-06, "loss": 0.0055, "step": 36274 }, { "epoch": 1.9339446606600204, "grad_norm": 1.0944448709487915, "learning_rate": 2.7452e-06, "loss": 0.0366, "step": 36275 }, { "epoch": 1.933997974089673, "grad_norm": 1.3762046098709106, "learning_rate": 2.7450000000000004e-06, "loss": 0.0195, "step": 36276 }, { "epoch": 1.934051287519326, "grad_norm": 1.5351336002349854, "learning_rate": 2.7448000000000004e-06, "loss": 0.0147, "step": 36277 }, { "epoch": 1.934104600948979, "grad_norm": 2.014728307723999, "learning_rate": 2.7446e-06, "loss": 0.0216, "step": 36278 }, { "epoch": 1.9341579143786318, "grad_norm": 0.9808340072631836, "learning_rate": 2.7444000000000004e-06, "loss": 0.0056, "step": 36279 }, { "epoch": 1.934211227808285, "grad_norm": 1.5598198175430298, "learning_rate": 2.7442000000000003e-06, "loss": 0.0109, "step": 36280 }, { "epoch": 1.9342645412379378, "grad_norm": 1.683920979499817, "learning_rate": 2.744e-06, "loss": 0.0224, "step": 36281 }, { "epoch": 1.9343178546675908, "grad_norm": 0.9860266447067261, "learning_rate": 2.7438000000000003e-06, "loss": 0.0199, "step": 36282 }, { "epoch": 1.9343711680972437, "grad_norm": 1.617728590965271, "learning_rate": 2.7436000000000003e-06, "loss": 0.0075, "step": 36283 }, { "epoch": 1.9344244815268965, "grad_norm": 1.806555986404419, "learning_rate": 2.7434e-06, "loss": 0.0149, "step": 36284 }, { "epoch": 1.9344777949565497, "grad_norm": 2.1418020725250244, "learning_rate": 2.7432000000000003e-06, "loss": 0.0293, "step": 36285 }, { "epoch": 1.9345311083862025, "grad_norm": 1.25564444065094, "learning_rate": 2.7430000000000002e-06, "loss": 0.0047, "step": 36286 }, { "epoch": 1.9345844218158554, "grad_norm": 0.6902274489402771, "learning_rate": 2.7428000000000007e-06, "loss": 0.0047, "step": 36287 }, { "epoch": 1.9346377352455084, "grad_norm": 1.3768386840820312, "learning_rate": 2.7426000000000002e-06, "loss": 0.0102, "step": 36288 }, { "epoch": 1.9346910486751612, "grad_norm": 3.6538078784942627, "learning_rate": 2.7424e-06, "loss": 0.0356, "step": 36289 }, { "epoch": 1.9347443621048142, "grad_norm": 2.264232873916626, "learning_rate": 2.7422000000000006e-06, "loss": 0.0365, "step": 36290 }, { "epoch": 1.9347976755344671, "grad_norm": 1.8459433317184448, "learning_rate": 2.742e-06, "loss": 0.0495, "step": 36291 }, { "epoch": 1.93485098896412, "grad_norm": 3.0404977798461914, "learning_rate": 2.7418e-06, "loss": 0.0547, "step": 36292 }, { "epoch": 1.934904302393773, "grad_norm": 1.598648190498352, "learning_rate": 2.7416000000000006e-06, "loss": 0.0166, "step": 36293 }, { "epoch": 1.9349576158234258, "grad_norm": 0.5539284348487854, "learning_rate": 2.7414e-06, "loss": 0.0034, "step": 36294 }, { "epoch": 1.9350109292530788, "grad_norm": 1.1023331880569458, "learning_rate": 2.7412e-06, "loss": 0.0089, "step": 36295 }, { "epoch": 1.9350642426827318, "grad_norm": 0.9073600172996521, "learning_rate": 2.7410000000000005e-06, "loss": 0.0163, "step": 36296 }, { "epoch": 1.9351175561123846, "grad_norm": 1.8975157737731934, "learning_rate": 2.7408e-06, "loss": 0.0222, "step": 36297 }, { "epoch": 1.9351708695420378, "grad_norm": 2.206789970397949, "learning_rate": 2.7406000000000005e-06, "loss": 0.0113, "step": 36298 }, { "epoch": 1.9352241829716905, "grad_norm": 1.2001816034317017, "learning_rate": 2.7404000000000005e-06, "loss": 0.0089, "step": 36299 }, { "epoch": 1.9352774964013435, "grad_norm": 6.447390079498291, "learning_rate": 2.7402e-06, "loss": 0.0529, "step": 36300 }, { "epoch": 1.9353308098309965, "grad_norm": 1.7675789594650269, "learning_rate": 2.7400000000000004e-06, "loss": 0.0154, "step": 36301 }, { "epoch": 1.9353841232606492, "grad_norm": 0.9391762614250183, "learning_rate": 2.7398000000000004e-06, "loss": 0.0058, "step": 36302 }, { "epoch": 1.9354374366903024, "grad_norm": 1.3479199409484863, "learning_rate": 2.7396e-06, "loss": 0.0133, "step": 36303 }, { "epoch": 1.9354907501199552, "grad_norm": 1.8526984453201294, "learning_rate": 2.7394000000000004e-06, "loss": 0.0107, "step": 36304 }, { "epoch": 1.9355440635496082, "grad_norm": 1.295828104019165, "learning_rate": 2.7392000000000004e-06, "loss": 0.0154, "step": 36305 }, { "epoch": 1.9355973769792612, "grad_norm": 2.1966731548309326, "learning_rate": 2.739e-06, "loss": 0.0177, "step": 36306 }, { "epoch": 1.935650690408914, "grad_norm": 2.434303045272827, "learning_rate": 2.7388000000000004e-06, "loss": 0.0264, "step": 36307 }, { "epoch": 1.9357040038385669, "grad_norm": 1.233256220817566, "learning_rate": 2.7386000000000003e-06, "loss": 0.0183, "step": 36308 }, { "epoch": 1.9357573172682199, "grad_norm": 1.7635138034820557, "learning_rate": 2.7384e-06, "loss": 0.028, "step": 36309 }, { "epoch": 1.9358106306978728, "grad_norm": 0.5084347724914551, "learning_rate": 2.7382000000000003e-06, "loss": 0.0041, "step": 36310 }, { "epoch": 1.9358639441275258, "grad_norm": 2.0023889541625977, "learning_rate": 2.7380000000000003e-06, "loss": 0.0244, "step": 36311 }, { "epoch": 1.9359172575571786, "grad_norm": 1.9671992063522339, "learning_rate": 2.7378000000000003e-06, "loss": 0.0183, "step": 36312 }, { "epoch": 1.9359705709868316, "grad_norm": 1.1970478296279907, "learning_rate": 2.7376000000000003e-06, "loss": 0.0143, "step": 36313 }, { "epoch": 1.9360238844164845, "grad_norm": 2.6126160621643066, "learning_rate": 2.7374000000000003e-06, "loss": 0.0303, "step": 36314 }, { "epoch": 1.9360771978461373, "grad_norm": 1.4854793548583984, "learning_rate": 2.7372000000000002e-06, "loss": 0.0117, "step": 36315 }, { "epoch": 1.9361305112757905, "grad_norm": 1.0709785223007202, "learning_rate": 2.7370000000000002e-06, "loss": 0.0163, "step": 36316 }, { "epoch": 1.9361838247054433, "grad_norm": 1.8126729726791382, "learning_rate": 2.7368e-06, "loss": 0.0262, "step": 36317 }, { "epoch": 1.9362371381350962, "grad_norm": 1.2055939435958862, "learning_rate": 2.7366e-06, "loss": 0.0062, "step": 36318 }, { "epoch": 1.9362904515647492, "grad_norm": 2.0030434131622314, "learning_rate": 2.7364e-06, "loss": 0.0333, "step": 36319 }, { "epoch": 1.936343764994402, "grad_norm": 3.4285032749176025, "learning_rate": 2.7362e-06, "loss": 0.0139, "step": 36320 }, { "epoch": 1.9363970784240552, "grad_norm": 2.050752639770508, "learning_rate": 2.736e-06, "loss": 0.0372, "step": 36321 }, { "epoch": 1.936450391853708, "grad_norm": 1.5708012580871582, "learning_rate": 2.7358e-06, "loss": 0.0201, "step": 36322 }, { "epoch": 1.936503705283361, "grad_norm": 1.263901948928833, "learning_rate": 2.7356000000000005e-06, "loss": 0.0144, "step": 36323 }, { "epoch": 1.936557018713014, "grad_norm": 2.0267117023468018, "learning_rate": 2.7354e-06, "loss": 0.0142, "step": 36324 }, { "epoch": 1.9366103321426666, "grad_norm": 1.1801997423171997, "learning_rate": 2.7352e-06, "loss": 0.0091, "step": 36325 }, { "epoch": 1.9366636455723196, "grad_norm": 0.5010950565338135, "learning_rate": 2.7350000000000005e-06, "loss": 0.0042, "step": 36326 }, { "epoch": 1.9367169590019726, "grad_norm": 1.303080677986145, "learning_rate": 2.7348e-06, "loss": 0.0071, "step": 36327 }, { "epoch": 1.9367702724316254, "grad_norm": 1.923352599143982, "learning_rate": 2.7346e-06, "loss": 0.024, "step": 36328 }, { "epoch": 1.9368235858612786, "grad_norm": 0.45957526564598083, "learning_rate": 2.7344000000000005e-06, "loss": 0.0034, "step": 36329 }, { "epoch": 1.9368768992909313, "grad_norm": 1.5453479290008545, "learning_rate": 2.7342e-06, "loss": 0.0236, "step": 36330 }, { "epoch": 1.9369302127205843, "grad_norm": 2.9043784141540527, "learning_rate": 2.734e-06, "loss": 0.0341, "step": 36331 }, { "epoch": 1.9369835261502373, "grad_norm": 3.302027463912964, "learning_rate": 2.7338000000000004e-06, "loss": 0.0254, "step": 36332 }, { "epoch": 1.93703683957989, "grad_norm": 1.7898917198181152, "learning_rate": 2.7336e-06, "loss": 0.0124, "step": 36333 }, { "epoch": 1.9370901530095432, "grad_norm": 3.394411087036133, "learning_rate": 2.7334000000000004e-06, "loss": 0.0485, "step": 36334 }, { "epoch": 1.937143466439196, "grad_norm": 1.8219318389892578, "learning_rate": 2.7332000000000004e-06, "loss": 0.0192, "step": 36335 }, { "epoch": 1.937196779868849, "grad_norm": 1.591764211654663, "learning_rate": 2.733e-06, "loss": 0.007, "step": 36336 }, { "epoch": 1.937250093298502, "grad_norm": 0.9598318934440613, "learning_rate": 2.7328000000000003e-06, "loss": 0.0069, "step": 36337 }, { "epoch": 1.9373034067281547, "grad_norm": 5.744882106781006, "learning_rate": 2.7326000000000003e-06, "loss": 0.0357, "step": 36338 }, { "epoch": 1.937356720157808, "grad_norm": 1.577561616897583, "learning_rate": 2.7324e-06, "loss": 0.0393, "step": 36339 }, { "epoch": 1.9374100335874607, "grad_norm": 2.6097824573516846, "learning_rate": 2.7322000000000003e-06, "loss": 0.0335, "step": 36340 }, { "epoch": 1.9374633470171136, "grad_norm": 1.7839998006820679, "learning_rate": 2.7320000000000003e-06, "loss": 0.0081, "step": 36341 }, { "epoch": 1.9375166604467666, "grad_norm": 0.6094604134559631, "learning_rate": 2.7318e-06, "loss": 0.0038, "step": 36342 }, { "epoch": 1.9375699738764194, "grad_norm": 1.8626829385757446, "learning_rate": 2.7316000000000002e-06, "loss": 0.0279, "step": 36343 }, { "epoch": 1.9376232873060724, "grad_norm": 1.036596655845642, "learning_rate": 2.7314000000000002e-06, "loss": 0.0092, "step": 36344 }, { "epoch": 1.9376766007357253, "grad_norm": 1.8131089210510254, "learning_rate": 2.7311999999999998e-06, "loss": 0.0193, "step": 36345 }, { "epoch": 1.937729914165378, "grad_norm": 0.98084557056427, "learning_rate": 2.731e-06, "loss": 0.0122, "step": 36346 }, { "epoch": 1.9377832275950313, "grad_norm": 1.1163631677627563, "learning_rate": 2.7308e-06, "loss": 0.009, "step": 36347 }, { "epoch": 1.937836541024684, "grad_norm": 1.1393046379089355, "learning_rate": 2.7306000000000006e-06, "loss": 0.0172, "step": 36348 }, { "epoch": 1.937889854454337, "grad_norm": 1.824999213218689, "learning_rate": 2.7304e-06, "loss": 0.0094, "step": 36349 }, { "epoch": 1.93794316788399, "grad_norm": 1.2576615810394287, "learning_rate": 2.7302e-06, "loss": 0.0134, "step": 36350 }, { "epoch": 1.9379964813136428, "grad_norm": 1.7639890909194946, "learning_rate": 2.7300000000000005e-06, "loss": 0.0118, "step": 36351 }, { "epoch": 1.938049794743296, "grad_norm": 0.5530527234077454, "learning_rate": 2.7298e-06, "loss": 0.0039, "step": 36352 }, { "epoch": 1.9381031081729487, "grad_norm": 1.150049090385437, "learning_rate": 2.7296e-06, "loss": 0.0068, "step": 36353 }, { "epoch": 1.9381564216026017, "grad_norm": 1.1825424432754517, "learning_rate": 2.7294000000000005e-06, "loss": 0.0083, "step": 36354 }, { "epoch": 1.9382097350322547, "grad_norm": 0.6719200015068054, "learning_rate": 2.7292e-06, "loss": 0.0047, "step": 36355 }, { "epoch": 1.9382630484619074, "grad_norm": 1.2563623189926147, "learning_rate": 2.729e-06, "loss": 0.01, "step": 36356 }, { "epoch": 1.9383163618915606, "grad_norm": 1.1169861555099487, "learning_rate": 2.7288000000000005e-06, "loss": 0.025, "step": 36357 }, { "epoch": 1.9383696753212134, "grad_norm": 2.3778603076934814, "learning_rate": 2.7286e-06, "loss": 0.0347, "step": 36358 }, { "epoch": 1.9384229887508664, "grad_norm": 1.6354355812072754, "learning_rate": 2.7284000000000004e-06, "loss": 0.0245, "step": 36359 }, { "epoch": 1.9384763021805194, "grad_norm": 2.6834282875061035, "learning_rate": 2.7282000000000004e-06, "loss": 0.0299, "step": 36360 }, { "epoch": 1.9385296156101721, "grad_norm": 1.4090343713760376, "learning_rate": 2.728e-06, "loss": 0.012, "step": 36361 }, { "epoch": 1.938582929039825, "grad_norm": 1.6523473262786865, "learning_rate": 2.7278000000000004e-06, "loss": 0.01, "step": 36362 }, { "epoch": 1.938636242469478, "grad_norm": 2.7367427349090576, "learning_rate": 2.7276000000000004e-06, "loss": 0.0097, "step": 36363 }, { "epoch": 1.9386895558991308, "grad_norm": 1.4514720439910889, "learning_rate": 2.7274e-06, "loss": 0.0048, "step": 36364 }, { "epoch": 1.938742869328784, "grad_norm": 2.2244017124176025, "learning_rate": 2.7272000000000003e-06, "loss": 0.0237, "step": 36365 }, { "epoch": 1.9387961827584368, "grad_norm": 1.038726806640625, "learning_rate": 2.7270000000000003e-06, "loss": 0.0038, "step": 36366 }, { "epoch": 1.9388494961880898, "grad_norm": 1.8427906036376953, "learning_rate": 2.7268e-06, "loss": 0.0109, "step": 36367 }, { "epoch": 1.9389028096177428, "grad_norm": 1.0117906332015991, "learning_rate": 2.7266000000000003e-06, "loss": 0.023, "step": 36368 }, { "epoch": 1.9389561230473955, "grad_norm": 1.788610577583313, "learning_rate": 2.7264000000000003e-06, "loss": 0.009, "step": 36369 }, { "epoch": 1.9390094364770487, "grad_norm": 2.572692394256592, "learning_rate": 2.7262e-06, "loss": 0.0334, "step": 36370 }, { "epoch": 1.9390627499067015, "grad_norm": 0.3448497951030731, "learning_rate": 2.7260000000000002e-06, "loss": 0.0025, "step": 36371 }, { "epoch": 1.9391160633363544, "grad_norm": 2.311847686767578, "learning_rate": 2.7258000000000002e-06, "loss": 0.0217, "step": 36372 }, { "epoch": 1.9391693767660074, "grad_norm": 0.5601124167442322, "learning_rate": 2.7256000000000006e-06, "loss": 0.0045, "step": 36373 }, { "epoch": 1.9392226901956602, "grad_norm": 1.0930966138839722, "learning_rate": 2.7254e-06, "loss": 0.0083, "step": 36374 }, { "epoch": 1.9392760036253134, "grad_norm": 2.7014412879943848, "learning_rate": 2.7252e-06, "loss": 0.0423, "step": 36375 }, { "epoch": 1.9393293170549661, "grad_norm": 0.4677676260471344, "learning_rate": 2.7250000000000006e-06, "loss": 0.0039, "step": 36376 }, { "epoch": 1.9393826304846191, "grad_norm": 0.5324591994285583, "learning_rate": 2.7248e-06, "loss": 0.0026, "step": 36377 }, { "epoch": 1.939435943914272, "grad_norm": 1.1662158966064453, "learning_rate": 2.7246e-06, "loss": 0.0064, "step": 36378 }, { "epoch": 1.9394892573439249, "grad_norm": 1.9398388862609863, "learning_rate": 2.7244000000000005e-06, "loss": 0.0361, "step": 36379 }, { "epoch": 1.9395425707735778, "grad_norm": 0.9807959198951721, "learning_rate": 2.7242e-06, "loss": 0.0038, "step": 36380 }, { "epoch": 1.9395958842032308, "grad_norm": 2.0926215648651123, "learning_rate": 2.724e-06, "loss": 0.0513, "step": 36381 }, { "epoch": 1.9396491976328836, "grad_norm": 2.0030410289764404, "learning_rate": 2.7238000000000005e-06, "loss": 0.0387, "step": 36382 }, { "epoch": 1.9397025110625368, "grad_norm": 2.19622802734375, "learning_rate": 2.7236e-06, "loss": 0.0246, "step": 36383 }, { "epoch": 1.9397558244921895, "grad_norm": 1.3011276721954346, "learning_rate": 2.7234000000000005e-06, "loss": 0.0077, "step": 36384 }, { "epoch": 1.9398091379218425, "grad_norm": 4.5993757247924805, "learning_rate": 2.7232000000000005e-06, "loss": 0.0447, "step": 36385 }, { "epoch": 1.9398624513514955, "grad_norm": 1.562103033065796, "learning_rate": 2.723e-06, "loss": 0.0199, "step": 36386 }, { "epoch": 1.9399157647811482, "grad_norm": 1.60011887550354, "learning_rate": 2.7228000000000004e-06, "loss": 0.0106, "step": 36387 }, { "epoch": 1.9399690782108014, "grad_norm": 2.9474315643310547, "learning_rate": 2.7226000000000004e-06, "loss": 0.0207, "step": 36388 }, { "epoch": 1.9400223916404542, "grad_norm": 2.5589706897735596, "learning_rate": 2.7224e-06, "loss": 0.0254, "step": 36389 }, { "epoch": 1.9400757050701072, "grad_norm": 1.7481136322021484, "learning_rate": 2.7222000000000004e-06, "loss": 0.0107, "step": 36390 }, { "epoch": 1.9401290184997602, "grad_norm": 1.6184744834899902, "learning_rate": 2.7220000000000004e-06, "loss": 0.0125, "step": 36391 }, { "epoch": 1.940182331929413, "grad_norm": 0.5236237645149231, "learning_rate": 2.7218e-06, "loss": 0.0051, "step": 36392 }, { "epoch": 1.940235645359066, "grad_norm": 1.3011205196380615, "learning_rate": 2.7216000000000003e-06, "loss": 0.0198, "step": 36393 }, { "epoch": 1.9402889587887189, "grad_norm": 1.3143926858901978, "learning_rate": 2.7214000000000003e-06, "loss": 0.0083, "step": 36394 }, { "epoch": 1.9403422722183719, "grad_norm": 1.1106085777282715, "learning_rate": 2.7212e-06, "loss": 0.0092, "step": 36395 }, { "epoch": 1.9403955856480248, "grad_norm": 2.309300184249878, "learning_rate": 2.7210000000000003e-06, "loss": 0.0225, "step": 36396 }, { "epoch": 1.9404488990776776, "grad_norm": 3.5735318660736084, "learning_rate": 2.7208000000000003e-06, "loss": 0.0586, "step": 36397 }, { "epoch": 1.9405022125073306, "grad_norm": 2.6690359115600586, "learning_rate": 2.7206000000000003e-06, "loss": 0.0424, "step": 36398 }, { "epoch": 1.9405555259369835, "grad_norm": 3.2859814167022705, "learning_rate": 2.7204000000000002e-06, "loss": 0.0093, "step": 36399 }, { "epoch": 1.9406088393666363, "grad_norm": 1.9523961544036865, "learning_rate": 2.7202000000000002e-06, "loss": 0.0544, "step": 36400 }, { "epoch": 1.9406621527962895, "grad_norm": 3.401930809020996, "learning_rate": 2.7200000000000002e-06, "loss": 0.0432, "step": 36401 }, { "epoch": 1.9407154662259423, "grad_norm": 1.618281602859497, "learning_rate": 2.7198e-06, "loss": 0.0077, "step": 36402 }, { "epoch": 1.9407687796555952, "grad_norm": 6.061779022216797, "learning_rate": 2.7196e-06, "loss": 0.0241, "step": 36403 }, { "epoch": 1.9408220930852482, "grad_norm": 1.3974370956420898, "learning_rate": 2.7194e-06, "loss": 0.0141, "step": 36404 }, { "epoch": 1.940875406514901, "grad_norm": 1.6406238079071045, "learning_rate": 2.7192e-06, "loss": 0.0109, "step": 36405 }, { "epoch": 1.9409287199445542, "grad_norm": 0.8830018639564514, "learning_rate": 2.719e-06, "loss": 0.0064, "step": 36406 }, { "epoch": 1.940982033374207, "grad_norm": 1.8246926069259644, "learning_rate": 2.7188e-06, "loss": 0.0141, "step": 36407 }, { "epoch": 1.94103534680386, "grad_norm": 1.7362077236175537, "learning_rate": 2.7186e-06, "loss": 0.0297, "step": 36408 }, { "epoch": 1.941088660233513, "grad_norm": 4.087296962738037, "learning_rate": 2.7184000000000005e-06, "loss": 0.0157, "step": 36409 }, { "epoch": 1.9411419736631657, "grad_norm": 2.785775899887085, "learning_rate": 2.7182e-06, "loss": 0.0196, "step": 36410 }, { "epoch": 1.9411952870928186, "grad_norm": 2.302481174468994, "learning_rate": 2.718e-06, "loss": 0.0211, "step": 36411 }, { "epoch": 1.9412486005224716, "grad_norm": 2.8078460693359375, "learning_rate": 2.7178000000000005e-06, "loss": 0.0237, "step": 36412 }, { "epoch": 1.9413019139521246, "grad_norm": 2.7145817279815674, "learning_rate": 2.7176e-06, "loss": 0.0488, "step": 36413 }, { "epoch": 1.9413552273817776, "grad_norm": 3.130091667175293, "learning_rate": 2.7174e-06, "loss": 0.0267, "step": 36414 }, { "epoch": 1.9414085408114303, "grad_norm": 2.2436306476593018, "learning_rate": 2.7172000000000004e-06, "loss": 0.0449, "step": 36415 }, { "epoch": 1.9414618542410833, "grad_norm": 0.9572210907936096, "learning_rate": 2.717e-06, "loss": 0.0052, "step": 36416 }, { "epoch": 1.9415151676707363, "grad_norm": 1.5276660919189453, "learning_rate": 2.7168e-06, "loss": 0.0145, "step": 36417 }, { "epoch": 1.941568481100389, "grad_norm": 2.529008626937866, "learning_rate": 2.7166000000000004e-06, "loss": 0.035, "step": 36418 }, { "epoch": 1.9416217945300422, "grad_norm": 1.464491844177246, "learning_rate": 2.7164e-06, "loss": 0.0081, "step": 36419 }, { "epoch": 1.941675107959695, "grad_norm": 3.373399257659912, "learning_rate": 2.7162e-06, "loss": 0.0366, "step": 36420 }, { "epoch": 1.941728421389348, "grad_norm": 1.8443337678909302, "learning_rate": 2.7160000000000003e-06, "loss": 0.0438, "step": 36421 }, { "epoch": 1.941781734819001, "grad_norm": 3.533651113510132, "learning_rate": 2.7158e-06, "loss": 0.0459, "step": 36422 }, { "epoch": 1.9418350482486537, "grad_norm": 0.957464337348938, "learning_rate": 2.7156000000000003e-06, "loss": 0.0175, "step": 36423 }, { "epoch": 1.941888361678307, "grad_norm": 1.6857112646102905, "learning_rate": 2.7154000000000003e-06, "loss": 0.0249, "step": 36424 }, { "epoch": 1.9419416751079597, "grad_norm": 1.7375307083129883, "learning_rate": 2.7152e-06, "loss": 0.0257, "step": 36425 }, { "epoch": 1.9419949885376127, "grad_norm": 1.3018324375152588, "learning_rate": 2.7150000000000003e-06, "loss": 0.014, "step": 36426 }, { "epoch": 1.9420483019672656, "grad_norm": 1.9578485488891602, "learning_rate": 2.7148000000000002e-06, "loss": 0.0156, "step": 36427 }, { "epoch": 1.9421016153969184, "grad_norm": 1.7549948692321777, "learning_rate": 2.7146e-06, "loss": 0.0188, "step": 36428 }, { "epoch": 1.9421549288265714, "grad_norm": 1.0237295627593994, "learning_rate": 2.7144000000000002e-06, "loss": 0.0085, "step": 36429 }, { "epoch": 1.9422082422562243, "grad_norm": 4.031291961669922, "learning_rate": 2.7142e-06, "loss": 0.0239, "step": 36430 }, { "epoch": 1.9422615556858773, "grad_norm": 1.608749270439148, "learning_rate": 2.7139999999999998e-06, "loss": 0.0241, "step": 36431 }, { "epoch": 1.9423148691155303, "grad_norm": 1.202684998512268, "learning_rate": 2.7138e-06, "loss": 0.0118, "step": 36432 }, { "epoch": 1.942368182545183, "grad_norm": 2.225750684738159, "learning_rate": 2.7136e-06, "loss": 0.0273, "step": 36433 }, { "epoch": 1.942421495974836, "grad_norm": 1.363966941833496, "learning_rate": 2.7134000000000006e-06, "loss": 0.0083, "step": 36434 }, { "epoch": 1.942474809404489, "grad_norm": 0.9875699281692505, "learning_rate": 2.7132e-06, "loss": 0.0055, "step": 36435 }, { "epoch": 1.9425281228341418, "grad_norm": 1.4290242195129395, "learning_rate": 2.713e-06, "loss": 0.0079, "step": 36436 }, { "epoch": 1.942581436263795, "grad_norm": 1.3486192226409912, "learning_rate": 2.7128000000000005e-06, "loss": 0.0049, "step": 36437 }, { "epoch": 1.9426347496934477, "grad_norm": 2.244276285171509, "learning_rate": 2.7126e-06, "loss": 0.0228, "step": 36438 }, { "epoch": 1.9426880631231007, "grad_norm": 2.04316782951355, "learning_rate": 2.7124e-06, "loss": 0.023, "step": 36439 }, { "epoch": 1.9427413765527537, "grad_norm": 1.9671940803527832, "learning_rate": 2.7122000000000005e-06, "loss": 0.0214, "step": 36440 }, { "epoch": 1.9427946899824065, "grad_norm": 1.3269747495651245, "learning_rate": 2.712e-06, "loss": 0.0161, "step": 36441 }, { "epoch": 1.9428480034120597, "grad_norm": 1.278016209602356, "learning_rate": 2.7118e-06, "loss": 0.0098, "step": 36442 }, { "epoch": 1.9429013168417124, "grad_norm": 2.779820442199707, "learning_rate": 2.7116000000000004e-06, "loss": 0.022, "step": 36443 }, { "epoch": 1.9429546302713654, "grad_norm": 2.046079158782959, "learning_rate": 2.7114e-06, "loss": 0.0175, "step": 36444 }, { "epoch": 1.9430079437010184, "grad_norm": 2.4526846408843994, "learning_rate": 2.7112000000000004e-06, "loss": 0.023, "step": 36445 }, { "epoch": 1.9430612571306711, "grad_norm": 0.9596168398857117, "learning_rate": 2.7110000000000004e-06, "loss": 0.0052, "step": 36446 }, { "epoch": 1.943114570560324, "grad_norm": 1.0715898275375366, "learning_rate": 2.7108e-06, "loss": 0.0173, "step": 36447 }, { "epoch": 1.943167883989977, "grad_norm": 2.729370594024658, "learning_rate": 2.7106000000000004e-06, "loss": 0.0382, "step": 36448 }, { "epoch": 1.9432211974196298, "grad_norm": 1.4584332704544067, "learning_rate": 2.7104000000000003e-06, "loss": 0.0198, "step": 36449 }, { "epoch": 1.943274510849283, "grad_norm": 1.8298931121826172, "learning_rate": 2.7102e-06, "loss": 0.0135, "step": 36450 }, { "epoch": 1.9433278242789358, "grad_norm": 1.4875394105911255, "learning_rate": 2.7100000000000003e-06, "loss": 0.0126, "step": 36451 }, { "epoch": 1.9433811377085888, "grad_norm": 1.7406033277511597, "learning_rate": 2.7098000000000003e-06, "loss": 0.0276, "step": 36452 }, { "epoch": 1.9434344511382418, "grad_norm": 2.207306146621704, "learning_rate": 2.7096e-06, "loss": 0.0223, "step": 36453 }, { "epoch": 1.9434877645678945, "grad_norm": 1.3294976949691772, "learning_rate": 2.7094000000000003e-06, "loss": 0.0096, "step": 36454 }, { "epoch": 1.9435410779975477, "grad_norm": 2.6859538555145264, "learning_rate": 2.7092000000000003e-06, "loss": 0.0354, "step": 36455 }, { "epoch": 1.9435943914272005, "grad_norm": 2.381044387817383, "learning_rate": 2.709e-06, "loss": 0.0177, "step": 36456 }, { "epoch": 1.9436477048568535, "grad_norm": 3.441849708557129, "learning_rate": 2.7088000000000002e-06, "loss": 0.0175, "step": 36457 }, { "epoch": 1.9437010182865064, "grad_norm": 6.364443778991699, "learning_rate": 2.7086e-06, "loss": 0.0184, "step": 36458 }, { "epoch": 1.9437543317161592, "grad_norm": 2.4527127742767334, "learning_rate": 2.7084000000000006e-06, "loss": 0.0159, "step": 36459 }, { "epoch": 1.9438076451458124, "grad_norm": 2.8643898963928223, "learning_rate": 2.7082e-06, "loss": 0.0181, "step": 36460 }, { "epoch": 1.9438609585754651, "grad_norm": 1.2688273191452026, "learning_rate": 2.708e-06, "loss": 0.0274, "step": 36461 }, { "epoch": 1.9439142720051181, "grad_norm": 1.210968255996704, "learning_rate": 2.7078000000000006e-06, "loss": 0.0088, "step": 36462 }, { "epoch": 1.943967585434771, "grad_norm": 1.8728772401809692, "learning_rate": 2.7076e-06, "loss": 0.0468, "step": 36463 }, { "epoch": 1.9440208988644239, "grad_norm": 1.9029494524002075, "learning_rate": 2.7074e-06, "loss": 0.0391, "step": 36464 }, { "epoch": 1.9440742122940768, "grad_norm": 2.2443106174468994, "learning_rate": 2.7072000000000005e-06, "loss": 0.033, "step": 36465 }, { "epoch": 1.9441275257237298, "grad_norm": 2.128725528717041, "learning_rate": 2.707e-06, "loss": 0.0451, "step": 36466 }, { "epoch": 1.9441808391533826, "grad_norm": 0.5622510313987732, "learning_rate": 2.7068e-06, "loss": 0.0028, "step": 36467 }, { "epoch": 1.9442341525830358, "grad_norm": 1.2879695892333984, "learning_rate": 2.7066000000000005e-06, "loss": 0.0126, "step": 36468 }, { "epoch": 1.9442874660126885, "grad_norm": 0.9368675351142883, "learning_rate": 2.7064e-06, "loss": 0.0064, "step": 36469 }, { "epoch": 1.9443407794423415, "grad_norm": 1.7647740840911865, "learning_rate": 2.7062000000000004e-06, "loss": 0.0131, "step": 36470 }, { "epoch": 1.9443940928719945, "grad_norm": 3.5622448921203613, "learning_rate": 2.7060000000000004e-06, "loss": 0.0083, "step": 36471 }, { "epoch": 1.9444474063016473, "grad_norm": 3.3230690956115723, "learning_rate": 2.7058e-06, "loss": 0.0138, "step": 36472 }, { "epoch": 1.9445007197313005, "grad_norm": 2.0014150142669678, "learning_rate": 2.7056000000000004e-06, "loss": 0.041, "step": 36473 }, { "epoch": 1.9445540331609532, "grad_norm": 1.9360591173171997, "learning_rate": 2.7054000000000004e-06, "loss": 0.0131, "step": 36474 }, { "epoch": 1.9446073465906062, "grad_norm": 1.9549692869186401, "learning_rate": 2.7052e-06, "loss": 0.0134, "step": 36475 }, { "epoch": 1.9446606600202592, "grad_norm": 3.0345306396484375, "learning_rate": 2.7050000000000004e-06, "loss": 0.0145, "step": 36476 }, { "epoch": 1.944713973449912, "grad_norm": 2.0258493423461914, "learning_rate": 2.7048000000000003e-06, "loss": 0.0209, "step": 36477 }, { "epoch": 1.9447672868795651, "grad_norm": 1.1198798418045044, "learning_rate": 2.7046e-06, "loss": 0.0085, "step": 36478 }, { "epoch": 1.9448206003092179, "grad_norm": 1.2341612577438354, "learning_rate": 2.7044000000000003e-06, "loss": 0.0135, "step": 36479 }, { "epoch": 1.9448739137388709, "grad_norm": 2.0897040367126465, "learning_rate": 2.7042000000000003e-06, "loss": 0.0331, "step": 36480 }, { "epoch": 1.9449272271685238, "grad_norm": 2.8418638706207275, "learning_rate": 2.704e-06, "loss": 0.0203, "step": 36481 }, { "epoch": 1.9449805405981766, "grad_norm": 1.1142696142196655, "learning_rate": 2.7038000000000003e-06, "loss": 0.0074, "step": 36482 }, { "epoch": 1.9450338540278296, "grad_norm": 1.425436019897461, "learning_rate": 2.7036000000000003e-06, "loss": 0.011, "step": 36483 }, { "epoch": 1.9450871674574826, "grad_norm": 2.7505486011505127, "learning_rate": 2.7034000000000007e-06, "loss": 0.0161, "step": 36484 }, { "epoch": 1.9451404808871353, "grad_norm": 2.018195390701294, "learning_rate": 2.7032000000000002e-06, "loss": 0.0194, "step": 36485 }, { "epoch": 1.9451937943167885, "grad_norm": 1.8938251733779907, "learning_rate": 2.703e-06, "loss": 0.0163, "step": 36486 }, { "epoch": 1.9452471077464413, "grad_norm": 0.7486369609832764, "learning_rate": 2.7028000000000006e-06, "loss": 0.0149, "step": 36487 }, { "epoch": 1.9453004211760943, "grad_norm": 1.4244186878204346, "learning_rate": 2.7026e-06, "loss": 0.0138, "step": 36488 }, { "epoch": 1.9453537346057472, "grad_norm": 3.7157435417175293, "learning_rate": 2.7024e-06, "loss": 0.0362, "step": 36489 }, { "epoch": 1.9454070480354, "grad_norm": 1.0218827724456787, "learning_rate": 2.7022000000000006e-06, "loss": 0.007, "step": 36490 }, { "epoch": 1.9454603614650532, "grad_norm": 2.1732234954833984, "learning_rate": 2.702e-06, "loss": 0.02, "step": 36491 }, { "epoch": 1.945513674894706, "grad_norm": 0.6272554993629456, "learning_rate": 2.7018e-06, "loss": 0.0048, "step": 36492 }, { "epoch": 1.945566988324359, "grad_norm": 2.545192003250122, "learning_rate": 2.7016000000000005e-06, "loss": 0.0306, "step": 36493 }, { "epoch": 1.945620301754012, "grad_norm": 2.493915319442749, "learning_rate": 2.7014e-06, "loss": 0.0143, "step": 36494 }, { "epoch": 1.9456736151836647, "grad_norm": 6.601734638214111, "learning_rate": 2.7012000000000005e-06, "loss": 0.0471, "step": 36495 }, { "epoch": 1.9457269286133176, "grad_norm": 1.1442495584487915, "learning_rate": 2.7010000000000005e-06, "loss": 0.0074, "step": 36496 }, { "epoch": 1.9457802420429706, "grad_norm": 1.4158318042755127, "learning_rate": 2.7008e-06, "loss": 0.0158, "step": 36497 }, { "epoch": 1.9458335554726236, "grad_norm": 2.193732738494873, "learning_rate": 2.7006000000000005e-06, "loss": 0.0189, "step": 36498 }, { "epoch": 1.9458868689022766, "grad_norm": 1.1772403717041016, "learning_rate": 2.7004000000000004e-06, "loss": 0.0059, "step": 36499 }, { "epoch": 1.9459401823319293, "grad_norm": 2.2809982299804688, "learning_rate": 2.7002e-06, "loss": 0.0278, "step": 36500 }, { "epoch": 1.9459934957615823, "grad_norm": 1.4116337299346924, "learning_rate": 2.7000000000000004e-06, "loss": 0.0059, "step": 36501 }, { "epoch": 1.9460468091912353, "grad_norm": 1.9618629217147827, "learning_rate": 2.6998000000000004e-06, "loss": 0.019, "step": 36502 }, { "epoch": 1.946100122620888, "grad_norm": 2.3647360801696777, "learning_rate": 2.6996e-06, "loss": 0.0209, "step": 36503 }, { "epoch": 1.9461534360505413, "grad_norm": 0.8217053413391113, "learning_rate": 2.6994000000000004e-06, "loss": 0.0224, "step": 36504 }, { "epoch": 1.946206749480194, "grad_norm": 1.2511749267578125, "learning_rate": 2.6992000000000003e-06, "loss": 0.0065, "step": 36505 }, { "epoch": 1.946260062909847, "grad_norm": 2.485222339630127, "learning_rate": 2.699e-06, "loss": 0.0113, "step": 36506 }, { "epoch": 1.9463133763395, "grad_norm": 1.384714126586914, "learning_rate": 2.6988000000000003e-06, "loss": 0.0088, "step": 36507 }, { "epoch": 1.9463666897691527, "grad_norm": 2.8940389156341553, "learning_rate": 2.6986000000000003e-06, "loss": 0.0252, "step": 36508 }, { "epoch": 1.946420003198806, "grad_norm": 2.174363613128662, "learning_rate": 2.6984000000000003e-06, "loss": 0.0202, "step": 36509 }, { "epoch": 1.9464733166284587, "grad_norm": 1.963932752609253, "learning_rate": 2.6982000000000003e-06, "loss": 0.0149, "step": 36510 }, { "epoch": 1.9465266300581117, "grad_norm": 2.7996718883514404, "learning_rate": 2.6980000000000003e-06, "loss": 0.0199, "step": 36511 }, { "epoch": 1.9465799434877646, "grad_norm": 2.0692520141601562, "learning_rate": 2.6978000000000002e-06, "loss": 0.0265, "step": 36512 }, { "epoch": 1.9466332569174174, "grad_norm": 2.341958999633789, "learning_rate": 2.6976000000000002e-06, "loss": 0.0167, "step": 36513 }, { "epoch": 1.9466865703470704, "grad_norm": 2.188375949859619, "learning_rate": 2.6974000000000002e-06, "loss": 0.0181, "step": 36514 }, { "epoch": 1.9467398837767234, "grad_norm": 2.496269464492798, "learning_rate": 2.6972e-06, "loss": 0.0128, "step": 36515 }, { "epoch": 1.9467931972063763, "grad_norm": 0.8466004729270935, "learning_rate": 2.697e-06, "loss": 0.004, "step": 36516 }, { "epoch": 1.9468465106360293, "grad_norm": 3.4628467559814453, "learning_rate": 2.6968e-06, "loss": 0.0255, "step": 36517 }, { "epoch": 1.946899824065682, "grad_norm": 1.928581714630127, "learning_rate": 2.6966e-06, "loss": 0.0106, "step": 36518 }, { "epoch": 1.946953137495335, "grad_norm": 1.7157692909240723, "learning_rate": 2.6964e-06, "loss": 0.0217, "step": 36519 }, { "epoch": 1.947006450924988, "grad_norm": 1.9023135900497437, "learning_rate": 2.6962000000000005e-06, "loss": 0.0076, "step": 36520 }, { "epoch": 1.9470597643546408, "grad_norm": 1.4474725723266602, "learning_rate": 2.696e-06, "loss": 0.0291, "step": 36521 }, { "epoch": 1.947113077784294, "grad_norm": 1.8610637187957764, "learning_rate": 2.6958e-06, "loss": 0.0227, "step": 36522 }, { "epoch": 1.9471663912139467, "grad_norm": 0.40663158893585205, "learning_rate": 2.6956000000000005e-06, "loss": 0.0037, "step": 36523 }, { "epoch": 1.9472197046435997, "grad_norm": 1.777129888534546, "learning_rate": 2.6954e-06, "loss": 0.027, "step": 36524 }, { "epoch": 1.9472730180732527, "grad_norm": 3.5564451217651367, "learning_rate": 2.6952e-06, "loss": 0.0314, "step": 36525 }, { "epoch": 1.9473263315029055, "grad_norm": 1.3599863052368164, "learning_rate": 2.6950000000000005e-06, "loss": 0.0051, "step": 36526 }, { "epoch": 1.9473796449325587, "grad_norm": 3.6991021633148193, "learning_rate": 2.6948e-06, "loss": 0.0351, "step": 36527 }, { "epoch": 1.9474329583622114, "grad_norm": 2.9822676181793213, "learning_rate": 2.6946e-06, "loss": 0.0298, "step": 36528 }, { "epoch": 1.9474862717918644, "grad_norm": 3.551628589630127, "learning_rate": 2.6944000000000004e-06, "loss": 0.0278, "step": 36529 }, { "epoch": 1.9475395852215174, "grad_norm": 4.846217632293701, "learning_rate": 2.6942e-06, "loss": 0.0306, "step": 36530 }, { "epoch": 1.9475928986511701, "grad_norm": 3.7437243461608887, "learning_rate": 2.694e-06, "loss": 0.0183, "step": 36531 }, { "epoch": 1.9476462120808231, "grad_norm": 1.4629464149475098, "learning_rate": 2.6938000000000004e-06, "loss": 0.0071, "step": 36532 }, { "epoch": 1.947699525510476, "grad_norm": 1.7498053312301636, "learning_rate": 2.6936e-06, "loss": 0.0104, "step": 36533 }, { "epoch": 1.947752838940129, "grad_norm": 2.759133815765381, "learning_rate": 2.6934000000000003e-06, "loss": 0.0235, "step": 36534 }, { "epoch": 1.947806152369782, "grad_norm": 1.7860602140426636, "learning_rate": 2.6932000000000003e-06, "loss": 0.0087, "step": 36535 }, { "epoch": 1.9478594657994348, "grad_norm": 6.437680721282959, "learning_rate": 2.693e-06, "loss": 0.0155, "step": 36536 }, { "epoch": 1.9479127792290878, "grad_norm": 2.0920543670654297, "learning_rate": 2.6928000000000003e-06, "loss": 0.0178, "step": 36537 }, { "epoch": 1.9479660926587408, "grad_norm": 3.4042420387268066, "learning_rate": 2.6926000000000003e-06, "loss": 0.0223, "step": 36538 }, { "epoch": 1.9480194060883935, "grad_norm": 1.6740657091140747, "learning_rate": 2.6924e-06, "loss": 0.0384, "step": 36539 }, { "epoch": 1.9480727195180467, "grad_norm": 1.7263555526733398, "learning_rate": 2.6922000000000002e-06, "loss": 0.0259, "step": 36540 }, { "epoch": 1.9481260329476995, "grad_norm": 3.026289939880371, "learning_rate": 2.6920000000000002e-06, "loss": 0.0145, "step": 36541 }, { "epoch": 1.9481793463773525, "grad_norm": 0.5397199988365173, "learning_rate": 2.6918e-06, "loss": 0.0039, "step": 36542 }, { "epoch": 1.9482326598070054, "grad_norm": 1.591154932975769, "learning_rate": 2.6916e-06, "loss": 0.0175, "step": 36543 }, { "epoch": 1.9482859732366582, "grad_norm": 2.211107015609741, "learning_rate": 2.6914e-06, "loss": 0.0113, "step": 36544 }, { "epoch": 1.9483392866663114, "grad_norm": 1.4545087814331055, "learning_rate": 2.6912000000000006e-06, "loss": 0.0153, "step": 36545 }, { "epoch": 1.9483926000959642, "grad_norm": 1.4099206924438477, "learning_rate": 2.691e-06, "loss": 0.0138, "step": 36546 }, { "epoch": 1.9484459135256171, "grad_norm": 2.9050886631011963, "learning_rate": 2.6908e-06, "loss": 0.0159, "step": 36547 }, { "epoch": 1.9484992269552701, "grad_norm": 3.073162317276001, "learning_rate": 2.6906000000000005e-06, "loss": 0.027, "step": 36548 }, { "epoch": 1.9485525403849229, "grad_norm": 1.6011937856674194, "learning_rate": 2.6904e-06, "loss": 0.0081, "step": 36549 }, { "epoch": 1.9486058538145759, "grad_norm": 2.6072731018066406, "learning_rate": 2.6902e-06, "loss": 0.0224, "step": 36550 }, { "epoch": 1.9486591672442288, "grad_norm": 1.4671028852462769, "learning_rate": 2.6900000000000005e-06, "loss": 0.0137, "step": 36551 }, { "epoch": 1.9487124806738816, "grad_norm": 1.0645359754562378, "learning_rate": 2.6898e-06, "loss": 0.0071, "step": 36552 }, { "epoch": 1.9487657941035348, "grad_norm": 0.49146392941474915, "learning_rate": 2.6896e-06, "loss": 0.0037, "step": 36553 }, { "epoch": 1.9488191075331875, "grad_norm": 1.8699345588684082, "learning_rate": 2.6894000000000005e-06, "loss": 0.0187, "step": 36554 }, { "epoch": 1.9488724209628405, "grad_norm": 1.071433663368225, "learning_rate": 2.6892e-06, "loss": 0.0113, "step": 36555 }, { "epoch": 1.9489257343924935, "grad_norm": 1.1823610067367554, "learning_rate": 2.689e-06, "loss": 0.0156, "step": 36556 }, { "epoch": 1.9489790478221463, "grad_norm": 2.5105178356170654, "learning_rate": 2.6888000000000004e-06, "loss": 0.0228, "step": 36557 }, { "epoch": 1.9490323612517995, "grad_norm": 3.422049045562744, "learning_rate": 2.6886e-06, "loss": 0.0115, "step": 36558 }, { "epoch": 1.9490856746814522, "grad_norm": 1.9223403930664062, "learning_rate": 2.6884000000000004e-06, "loss": 0.0077, "step": 36559 }, { "epoch": 1.9491389881111052, "grad_norm": 1.5647163391113281, "learning_rate": 2.6882000000000004e-06, "loss": 0.0164, "step": 36560 }, { "epoch": 1.9491923015407582, "grad_norm": 4.816629409790039, "learning_rate": 2.688e-06, "loss": 0.0329, "step": 36561 }, { "epoch": 1.949245614970411, "grad_norm": 3.5436739921569824, "learning_rate": 2.6878000000000003e-06, "loss": 0.0183, "step": 36562 }, { "epoch": 1.9492989284000641, "grad_norm": 1.7871124744415283, "learning_rate": 2.6876000000000003e-06, "loss": 0.0099, "step": 36563 }, { "epoch": 1.949352241829717, "grad_norm": 1.4695616960525513, "learning_rate": 2.6874e-06, "loss": 0.0148, "step": 36564 }, { "epoch": 1.9494055552593699, "grad_norm": 7.114877700805664, "learning_rate": 2.6872000000000003e-06, "loss": 0.074, "step": 36565 }, { "epoch": 1.9494588686890229, "grad_norm": 1.3655954599380493, "learning_rate": 2.6870000000000003e-06, "loss": 0.0211, "step": 36566 }, { "epoch": 1.9495121821186756, "grad_norm": 0.35226354002952576, "learning_rate": 2.6868e-06, "loss": 0.0029, "step": 36567 }, { "epoch": 1.9495654955483286, "grad_norm": 1.1447937488555908, "learning_rate": 2.6866000000000002e-06, "loss": 0.0251, "step": 36568 }, { "epoch": 1.9496188089779816, "grad_norm": 2.890915632247925, "learning_rate": 2.6864000000000002e-06, "loss": 0.0298, "step": 36569 }, { "epoch": 1.9496721224076343, "grad_norm": 3.1005825996398926, "learning_rate": 2.6862000000000006e-06, "loss": 0.0205, "step": 36570 }, { "epoch": 1.9497254358372875, "grad_norm": 1.4389568567276, "learning_rate": 2.686e-06, "loss": 0.0059, "step": 36571 }, { "epoch": 1.9497787492669403, "grad_norm": 2.351013660430908, "learning_rate": 2.6858e-06, "loss": 0.0075, "step": 36572 }, { "epoch": 1.9498320626965933, "grad_norm": 1.8343806266784668, "learning_rate": 2.6856000000000006e-06, "loss": 0.0221, "step": 36573 }, { "epoch": 1.9498853761262462, "grad_norm": 0.970063328742981, "learning_rate": 2.6854e-06, "loss": 0.0077, "step": 36574 }, { "epoch": 1.949938689555899, "grad_norm": 1.9234169721603394, "learning_rate": 2.6852e-06, "loss": 0.02, "step": 36575 }, { "epoch": 1.9499920029855522, "grad_norm": 0.6145341396331787, "learning_rate": 2.6850000000000006e-06, "loss": 0.0047, "step": 36576 }, { "epoch": 1.950045316415205, "grad_norm": 2.99387788772583, "learning_rate": 2.6848e-06, "loss": 0.0284, "step": 36577 }, { "epoch": 1.950098629844858, "grad_norm": 1.4490548372268677, "learning_rate": 2.6846e-06, "loss": 0.0079, "step": 36578 }, { "epoch": 1.950151943274511, "grad_norm": 1.3610725402832031, "learning_rate": 2.6844000000000005e-06, "loss": 0.0162, "step": 36579 }, { "epoch": 1.9502052567041637, "grad_norm": 1.7819632291793823, "learning_rate": 2.6842e-06, "loss": 0.017, "step": 36580 }, { "epoch": 1.9502585701338169, "grad_norm": 2.8652191162109375, "learning_rate": 2.6840000000000005e-06, "loss": 0.0375, "step": 36581 }, { "epoch": 1.9503118835634696, "grad_norm": 1.6948230266571045, "learning_rate": 2.6838000000000005e-06, "loss": 0.0073, "step": 36582 }, { "epoch": 1.9503651969931226, "grad_norm": 2.0804922580718994, "learning_rate": 2.6836e-06, "loss": 0.0248, "step": 36583 }, { "epoch": 1.9504185104227756, "grad_norm": 1.1164242029190063, "learning_rate": 2.6834000000000004e-06, "loss": 0.0103, "step": 36584 }, { "epoch": 1.9504718238524283, "grad_norm": 0.6342836618423462, "learning_rate": 2.6832000000000004e-06, "loss": 0.005, "step": 36585 }, { "epoch": 1.9505251372820813, "grad_norm": 2.455467700958252, "learning_rate": 2.683e-06, "loss": 0.0216, "step": 36586 }, { "epoch": 1.9505784507117343, "grad_norm": 1.9190524816513062, "learning_rate": 2.6828000000000004e-06, "loss": 0.006, "step": 36587 }, { "epoch": 1.950631764141387, "grad_norm": 2.2064383029937744, "learning_rate": 2.6826000000000004e-06, "loss": 0.0268, "step": 36588 }, { "epoch": 1.9506850775710403, "grad_norm": 1.270675778388977, "learning_rate": 2.6824e-06, "loss": 0.0108, "step": 36589 }, { "epoch": 1.950738391000693, "grad_norm": 0.8958601951599121, "learning_rate": 2.6822000000000003e-06, "loss": 0.0046, "step": 36590 }, { "epoch": 1.950791704430346, "grad_norm": 2.7169246673583984, "learning_rate": 2.6820000000000003e-06, "loss": 0.0181, "step": 36591 }, { "epoch": 1.950845017859999, "grad_norm": 2.1043806076049805, "learning_rate": 2.6818e-06, "loss": 0.0256, "step": 36592 }, { "epoch": 1.9508983312896517, "grad_norm": 2.348889112472534, "learning_rate": 2.6816000000000003e-06, "loss": 0.0144, "step": 36593 }, { "epoch": 1.950951644719305, "grad_norm": 1.5525833368301392, "learning_rate": 2.6814000000000003e-06, "loss": 0.0052, "step": 36594 }, { "epoch": 1.9510049581489577, "grad_norm": 3.0322062969207764, "learning_rate": 2.6812000000000003e-06, "loss": 0.0132, "step": 36595 }, { "epoch": 1.9510582715786107, "grad_norm": 1.5736554861068726, "learning_rate": 2.6810000000000003e-06, "loss": 0.0065, "step": 36596 }, { "epoch": 1.9511115850082636, "grad_norm": 2.477452516555786, "learning_rate": 2.6808000000000002e-06, "loss": 0.0493, "step": 36597 }, { "epoch": 1.9511648984379164, "grad_norm": 1.979008674621582, "learning_rate": 2.6806000000000002e-06, "loss": 0.02, "step": 36598 }, { "epoch": 1.9512182118675694, "grad_norm": 1.7652462720870972, "learning_rate": 2.6804e-06, "loss": 0.0215, "step": 36599 }, { "epoch": 1.9512715252972224, "grad_norm": 1.6344407796859741, "learning_rate": 2.6802e-06, "loss": 0.0328, "step": 36600 }, { "epoch": 1.9513248387268753, "grad_norm": 1.2290806770324707, "learning_rate": 2.68e-06, "loss": 0.014, "step": 36601 }, { "epoch": 1.9513781521565283, "grad_norm": 1.9415677785873413, "learning_rate": 2.6798e-06, "loss": 0.0233, "step": 36602 }, { "epoch": 1.951431465586181, "grad_norm": 1.7748111486434937, "learning_rate": 2.6796e-06, "loss": 0.0173, "step": 36603 }, { "epoch": 1.951484779015834, "grad_norm": 2.82759165763855, "learning_rate": 2.6794e-06, "loss": 0.014, "step": 36604 }, { "epoch": 1.951538092445487, "grad_norm": 2.4860775470733643, "learning_rate": 2.6792e-06, "loss": 0.0254, "step": 36605 }, { "epoch": 1.9515914058751398, "grad_norm": 0.8912295699119568, "learning_rate": 2.6790000000000005e-06, "loss": 0.0045, "step": 36606 }, { "epoch": 1.951644719304793, "grad_norm": 3.0742297172546387, "learning_rate": 2.6788e-06, "loss": 0.0244, "step": 36607 }, { "epoch": 1.9516980327344458, "grad_norm": 1.9311935901641846, "learning_rate": 2.6786e-06, "loss": 0.0067, "step": 36608 }, { "epoch": 1.9517513461640987, "grad_norm": 0.22791661322116852, "learning_rate": 2.6784000000000005e-06, "loss": 0.0023, "step": 36609 }, { "epoch": 1.9518046595937517, "grad_norm": 1.2542874813079834, "learning_rate": 2.6782e-06, "loss": 0.0233, "step": 36610 }, { "epoch": 1.9518579730234045, "grad_norm": 1.8797686100006104, "learning_rate": 2.678e-06, "loss": 0.0205, "step": 36611 }, { "epoch": 1.9519112864530577, "grad_norm": 2.320571184158325, "learning_rate": 2.6778000000000004e-06, "loss": 0.0367, "step": 36612 }, { "epoch": 1.9519645998827104, "grad_norm": 1.633867621421814, "learning_rate": 2.6776e-06, "loss": 0.0106, "step": 36613 }, { "epoch": 1.9520179133123634, "grad_norm": 2.537583351135254, "learning_rate": 2.6774e-06, "loss": 0.0373, "step": 36614 }, { "epoch": 1.9520712267420164, "grad_norm": 1.7524546384811401, "learning_rate": 2.6772000000000004e-06, "loss": 0.0192, "step": 36615 }, { "epoch": 1.9521245401716691, "grad_norm": 2.0702297687530518, "learning_rate": 2.677e-06, "loss": 0.0139, "step": 36616 }, { "epoch": 1.9521778536013221, "grad_norm": 3.5718700885772705, "learning_rate": 2.6768e-06, "loss": 0.0509, "step": 36617 }, { "epoch": 1.952231167030975, "grad_norm": 2.5593581199645996, "learning_rate": 2.6766000000000003e-06, "loss": 0.0353, "step": 36618 }, { "epoch": 1.952284480460628, "grad_norm": 2.802670955657959, "learning_rate": 2.6764e-06, "loss": 0.0157, "step": 36619 }, { "epoch": 1.952337793890281, "grad_norm": 1.7758450508117676, "learning_rate": 2.6762000000000003e-06, "loss": 0.0165, "step": 36620 }, { "epoch": 1.9523911073199338, "grad_norm": 0.9688701033592224, "learning_rate": 2.6760000000000003e-06, "loss": 0.0066, "step": 36621 }, { "epoch": 1.9524444207495868, "grad_norm": 1.8381596803665161, "learning_rate": 2.6758e-06, "loss": 0.0088, "step": 36622 }, { "epoch": 1.9524977341792398, "grad_norm": 1.402347207069397, "learning_rate": 2.6756000000000003e-06, "loss": 0.0066, "step": 36623 }, { "epoch": 1.9525510476088925, "grad_norm": 1.3398635387420654, "learning_rate": 2.6754000000000003e-06, "loss": 0.0071, "step": 36624 }, { "epoch": 1.9526043610385457, "grad_norm": 2.345724582672119, "learning_rate": 2.6752e-06, "loss": 0.039, "step": 36625 }, { "epoch": 1.9526576744681985, "grad_norm": 1.9182039499282837, "learning_rate": 2.6750000000000002e-06, "loss": 0.026, "step": 36626 }, { "epoch": 1.9527109878978515, "grad_norm": 2.64184308052063, "learning_rate": 2.6748e-06, "loss": 0.0234, "step": 36627 }, { "epoch": 1.9527643013275044, "grad_norm": 2.547086238861084, "learning_rate": 2.6745999999999998e-06, "loss": 0.0421, "step": 36628 }, { "epoch": 1.9528176147571572, "grad_norm": 1.0598641633987427, "learning_rate": 2.6744e-06, "loss": 0.0065, "step": 36629 }, { "epoch": 1.9528709281868104, "grad_norm": 1.0397549867630005, "learning_rate": 2.6742e-06, "loss": 0.0073, "step": 36630 }, { "epoch": 1.9529242416164632, "grad_norm": 1.416371464729309, "learning_rate": 2.6740000000000006e-06, "loss": 0.0188, "step": 36631 }, { "epoch": 1.9529775550461161, "grad_norm": 2.6353530883789062, "learning_rate": 2.6738e-06, "loss": 0.022, "step": 36632 }, { "epoch": 1.9530308684757691, "grad_norm": 1.6979409456253052, "learning_rate": 2.6736e-06, "loss": 0.0083, "step": 36633 }, { "epoch": 1.9530841819054219, "grad_norm": 1.5947096347808838, "learning_rate": 2.6734000000000005e-06, "loss": 0.0123, "step": 36634 }, { "epoch": 1.9531374953350749, "grad_norm": 1.2314621210098267, "learning_rate": 2.6732e-06, "loss": 0.0203, "step": 36635 }, { "epoch": 1.9531908087647278, "grad_norm": 2.562753438949585, "learning_rate": 2.673e-06, "loss": 0.0109, "step": 36636 }, { "epoch": 1.9532441221943808, "grad_norm": 1.2737431526184082, "learning_rate": 2.6728000000000005e-06, "loss": 0.0152, "step": 36637 }, { "epoch": 1.9532974356240338, "grad_norm": 1.9237996339797974, "learning_rate": 2.6726e-06, "loss": 0.035, "step": 36638 }, { "epoch": 1.9533507490536866, "grad_norm": 2.1284103393554688, "learning_rate": 2.6724e-06, "loss": 0.0099, "step": 36639 }, { "epoch": 1.9534040624833395, "grad_norm": 0.37285470962524414, "learning_rate": 2.6722000000000004e-06, "loss": 0.0031, "step": 36640 }, { "epoch": 1.9534573759129925, "grad_norm": 3.964733839035034, "learning_rate": 2.672e-06, "loss": 0.0167, "step": 36641 }, { "epoch": 1.9535106893426453, "grad_norm": 2.7942514419555664, "learning_rate": 2.6718e-06, "loss": 0.0486, "step": 36642 }, { "epoch": 1.9535640027722985, "grad_norm": 1.149242877960205, "learning_rate": 2.6716000000000004e-06, "loss": 0.0112, "step": 36643 }, { "epoch": 1.9536173162019512, "grad_norm": 1.888932466506958, "learning_rate": 2.6714e-06, "loss": 0.0159, "step": 36644 }, { "epoch": 1.9536706296316042, "grad_norm": 1.9762274026870728, "learning_rate": 2.6712000000000004e-06, "loss": 0.013, "step": 36645 }, { "epoch": 1.9537239430612572, "grad_norm": 1.9759418964385986, "learning_rate": 2.6710000000000003e-06, "loss": 0.0106, "step": 36646 }, { "epoch": 1.95377725649091, "grad_norm": 0.46281698346138, "learning_rate": 2.6708e-06, "loss": 0.0044, "step": 36647 }, { "epoch": 1.9538305699205631, "grad_norm": 0.3664036989212036, "learning_rate": 2.6706000000000003e-06, "loss": 0.0029, "step": 36648 }, { "epoch": 1.953883883350216, "grad_norm": 2.1112148761749268, "learning_rate": 2.6704000000000003e-06, "loss": 0.0222, "step": 36649 }, { "epoch": 1.9539371967798689, "grad_norm": 0.9915508031845093, "learning_rate": 2.6702e-06, "loss": 0.0063, "step": 36650 }, { "epoch": 1.9539905102095219, "grad_norm": 1.4929840564727783, "learning_rate": 2.6700000000000003e-06, "loss": 0.0092, "step": 36651 }, { "epoch": 1.9540438236391746, "grad_norm": 3.4972939491271973, "learning_rate": 2.6698000000000003e-06, "loss": 0.0236, "step": 36652 }, { "epoch": 1.9540971370688276, "grad_norm": 2.5067811012268066, "learning_rate": 2.6696e-06, "loss": 0.0231, "step": 36653 }, { "epoch": 1.9541504504984806, "grad_norm": 0.5905159115791321, "learning_rate": 2.6694000000000002e-06, "loss": 0.0039, "step": 36654 }, { "epoch": 1.9542037639281333, "grad_norm": 2.4246788024902344, "learning_rate": 2.6692e-06, "loss": 0.0219, "step": 36655 }, { "epoch": 1.9542570773577865, "grad_norm": 2.1571974754333496, "learning_rate": 2.6690000000000006e-06, "loss": 0.0101, "step": 36656 }, { "epoch": 1.9543103907874393, "grad_norm": 2.878530502319336, "learning_rate": 2.6688e-06, "loss": 0.0153, "step": 36657 }, { "epoch": 1.9543637042170923, "grad_norm": 1.3134477138519287, "learning_rate": 2.6686e-06, "loss": 0.0203, "step": 36658 }, { "epoch": 1.9544170176467452, "grad_norm": 2.2881555557250977, "learning_rate": 2.6684000000000006e-06, "loss": 0.0222, "step": 36659 }, { "epoch": 1.954470331076398, "grad_norm": 1.4351190328598022, "learning_rate": 2.6682e-06, "loss": 0.013, "step": 36660 }, { "epoch": 1.9545236445060512, "grad_norm": 1.7844282388687134, "learning_rate": 2.668e-06, "loss": 0.0271, "step": 36661 }, { "epoch": 1.954576957935704, "grad_norm": 2.135354995727539, "learning_rate": 2.6678000000000005e-06, "loss": 0.0064, "step": 36662 }, { "epoch": 1.954630271365357, "grad_norm": 1.3441206216812134, "learning_rate": 2.6676e-06, "loss": 0.0089, "step": 36663 }, { "epoch": 1.95468358479501, "grad_norm": 1.4110020399093628, "learning_rate": 2.6674e-06, "loss": 0.0147, "step": 36664 }, { "epoch": 1.9547368982246627, "grad_norm": 0.4223317801952362, "learning_rate": 2.6672000000000005e-06, "loss": 0.0035, "step": 36665 }, { "epoch": 1.9547902116543159, "grad_norm": 1.9164835214614868, "learning_rate": 2.667e-06, "loss": 0.0248, "step": 36666 }, { "epoch": 1.9548435250839686, "grad_norm": 2.0443437099456787, "learning_rate": 2.6668e-06, "loss": 0.0152, "step": 36667 }, { "epoch": 1.9548968385136216, "grad_norm": 0.6461368203163147, "learning_rate": 2.6666000000000004e-06, "loss": 0.0043, "step": 36668 }, { "epoch": 1.9549501519432746, "grad_norm": 1.9915214776992798, "learning_rate": 2.6664e-06, "loss": 0.0229, "step": 36669 }, { "epoch": 1.9550034653729274, "grad_norm": 0.21944482624530792, "learning_rate": 2.6662000000000004e-06, "loss": 0.0015, "step": 36670 }, { "epoch": 1.9550567788025803, "grad_norm": 3.0179176330566406, "learning_rate": 2.6660000000000004e-06, "loss": 0.0262, "step": 36671 }, { "epoch": 1.9551100922322333, "grad_norm": 2.933077812194824, "learning_rate": 2.6658e-06, "loss": 0.0206, "step": 36672 }, { "epoch": 1.955163405661886, "grad_norm": 0.2505846917629242, "learning_rate": 2.6656000000000004e-06, "loss": 0.0025, "step": 36673 }, { "epoch": 1.9552167190915393, "grad_norm": 0.8760579824447632, "learning_rate": 2.6654000000000003e-06, "loss": 0.0045, "step": 36674 }, { "epoch": 1.955270032521192, "grad_norm": 1.8123283386230469, "learning_rate": 2.6652e-06, "loss": 0.0202, "step": 36675 }, { "epoch": 1.955323345950845, "grad_norm": 1.9044787883758545, "learning_rate": 2.6650000000000003e-06, "loss": 0.0191, "step": 36676 }, { "epoch": 1.955376659380498, "grad_norm": 1.3606678247451782, "learning_rate": 2.6648000000000003e-06, "loss": 0.0078, "step": 36677 }, { "epoch": 1.9554299728101507, "grad_norm": 0.4584263265132904, "learning_rate": 2.6646e-06, "loss": 0.0032, "step": 36678 }, { "epoch": 1.955483286239804, "grad_norm": 0.28684934973716736, "learning_rate": 2.6644000000000003e-06, "loss": 0.0031, "step": 36679 }, { "epoch": 1.9555365996694567, "grad_norm": 1.9489401578903198, "learning_rate": 2.6642000000000003e-06, "loss": 0.0352, "step": 36680 }, { "epoch": 1.9555899130991097, "grad_norm": 2.0879039764404297, "learning_rate": 2.6640000000000007e-06, "loss": 0.0262, "step": 36681 }, { "epoch": 1.9556432265287627, "grad_norm": 4.333647727966309, "learning_rate": 2.6638000000000002e-06, "loss": 0.032, "step": 36682 }, { "epoch": 1.9556965399584154, "grad_norm": 3.1209254264831543, "learning_rate": 2.6636000000000002e-06, "loss": 0.0326, "step": 36683 }, { "epoch": 1.9557498533880686, "grad_norm": 2.496037006378174, "learning_rate": 2.6634000000000006e-06, "loss": 0.028, "step": 36684 }, { "epoch": 1.9558031668177214, "grad_norm": 1.0083805322647095, "learning_rate": 2.6632e-06, "loss": 0.0098, "step": 36685 }, { "epoch": 1.9558564802473744, "grad_norm": 1.8246350288391113, "learning_rate": 2.663e-06, "loss": 0.0194, "step": 36686 }, { "epoch": 1.9559097936770273, "grad_norm": 3.3225934505462646, "learning_rate": 2.6628000000000006e-06, "loss": 0.0183, "step": 36687 }, { "epoch": 1.95596310710668, "grad_norm": 1.0653338432312012, "learning_rate": 2.6626e-06, "loss": 0.0183, "step": 36688 }, { "epoch": 1.956016420536333, "grad_norm": 1.312109112739563, "learning_rate": 2.6624e-06, "loss": 0.0045, "step": 36689 }, { "epoch": 1.956069733965986, "grad_norm": 1.6619120836257935, "learning_rate": 2.6622000000000005e-06, "loss": 0.0139, "step": 36690 }, { "epoch": 1.9561230473956388, "grad_norm": 0.782535970211029, "learning_rate": 2.662e-06, "loss": 0.0139, "step": 36691 }, { "epoch": 1.956176360825292, "grad_norm": 2.5551064014434814, "learning_rate": 2.6618000000000005e-06, "loss": 0.0275, "step": 36692 }, { "epoch": 1.9562296742549448, "grad_norm": 4.3256516456604, "learning_rate": 2.6616000000000005e-06, "loss": 0.0533, "step": 36693 }, { "epoch": 1.9562829876845977, "grad_norm": 2.880467414855957, "learning_rate": 2.6614e-06, "loss": 0.0452, "step": 36694 }, { "epoch": 1.9563363011142507, "grad_norm": 1.1664379835128784, "learning_rate": 2.6612000000000005e-06, "loss": 0.0059, "step": 36695 }, { "epoch": 1.9563896145439035, "grad_norm": 0.8492236137390137, "learning_rate": 2.6610000000000004e-06, "loss": 0.0053, "step": 36696 }, { "epoch": 1.9564429279735567, "grad_norm": 2.2084760665893555, "learning_rate": 2.6608e-06, "loss": 0.0096, "step": 36697 }, { "epoch": 1.9564962414032094, "grad_norm": 2.0998051166534424, "learning_rate": 2.6606000000000004e-06, "loss": 0.036, "step": 36698 }, { "epoch": 1.9565495548328624, "grad_norm": 0.9588031768798828, "learning_rate": 2.6604000000000004e-06, "loss": 0.0051, "step": 36699 }, { "epoch": 1.9566028682625154, "grad_norm": 1.6179158687591553, "learning_rate": 2.6602e-06, "loss": 0.0184, "step": 36700 }, { "epoch": 1.9566561816921682, "grad_norm": 0.6102080941200256, "learning_rate": 2.6600000000000004e-06, "loss": 0.0052, "step": 36701 }, { "epoch": 1.9567094951218214, "grad_norm": 3.748593330383301, "learning_rate": 2.6598000000000004e-06, "loss": 0.0169, "step": 36702 }, { "epoch": 1.956762808551474, "grad_norm": 0.9952247142791748, "learning_rate": 2.6596e-06, "loss": 0.0048, "step": 36703 }, { "epoch": 1.956816121981127, "grad_norm": 2.125779628753662, "learning_rate": 2.6594000000000003e-06, "loss": 0.0229, "step": 36704 }, { "epoch": 1.95686943541078, "grad_norm": 1.7557432651519775, "learning_rate": 2.6592000000000003e-06, "loss": 0.012, "step": 36705 }, { "epoch": 1.9569227488404328, "grad_norm": 1.952865481376648, "learning_rate": 2.6590000000000003e-06, "loss": 0.0307, "step": 36706 }, { "epoch": 1.9569760622700858, "grad_norm": 2.0728890895843506, "learning_rate": 2.6588000000000003e-06, "loss": 0.0274, "step": 36707 }, { "epoch": 1.9570293756997388, "grad_norm": 1.3891252279281616, "learning_rate": 2.6586000000000003e-06, "loss": 0.0054, "step": 36708 }, { "epoch": 1.9570826891293915, "grad_norm": 5.213091850280762, "learning_rate": 2.6584000000000002e-06, "loss": 0.0102, "step": 36709 }, { "epoch": 1.9571360025590447, "grad_norm": 1.1411182880401611, "learning_rate": 2.6582000000000002e-06, "loss": 0.0106, "step": 36710 }, { "epoch": 1.9571893159886975, "grad_norm": 0.17610324919223785, "learning_rate": 2.6580000000000002e-06, "loss": 0.0019, "step": 36711 }, { "epoch": 1.9572426294183505, "grad_norm": 1.9040054082870483, "learning_rate": 2.6578e-06, "loss": 0.0141, "step": 36712 }, { "epoch": 1.9572959428480035, "grad_norm": 1.020323634147644, "learning_rate": 2.6576e-06, "loss": 0.0186, "step": 36713 }, { "epoch": 1.9573492562776562, "grad_norm": 1.8740689754486084, "learning_rate": 2.6574e-06, "loss": 0.0144, "step": 36714 }, { "epoch": 1.9574025697073094, "grad_norm": 2.2669615745544434, "learning_rate": 2.6572e-06, "loss": 0.0221, "step": 36715 }, { "epoch": 1.9574558831369622, "grad_norm": 0.9152878522872925, "learning_rate": 2.657e-06, "loss": 0.0095, "step": 36716 }, { "epoch": 1.9575091965666152, "grad_norm": 0.5293028354644775, "learning_rate": 2.6568000000000005e-06, "loss": 0.0032, "step": 36717 }, { "epoch": 1.9575625099962681, "grad_norm": 1.8961526155471802, "learning_rate": 2.6566e-06, "loss": 0.0095, "step": 36718 }, { "epoch": 1.9576158234259209, "grad_norm": 1.7291630506515503, "learning_rate": 2.6564e-06, "loss": 0.0079, "step": 36719 }, { "epoch": 1.9576691368555739, "grad_norm": 0.8415662050247192, "learning_rate": 2.6562000000000005e-06, "loss": 0.0151, "step": 36720 }, { "epoch": 1.9577224502852268, "grad_norm": 1.545723557472229, "learning_rate": 2.656e-06, "loss": 0.0324, "step": 36721 }, { "epoch": 1.9577757637148798, "grad_norm": 0.3713358938694, "learning_rate": 2.6558e-06, "loss": 0.0037, "step": 36722 }, { "epoch": 1.9578290771445328, "grad_norm": 3.1906909942626953, "learning_rate": 2.6556000000000005e-06, "loss": 0.0302, "step": 36723 }, { "epoch": 1.9578823905741856, "grad_norm": 2.301377058029175, "learning_rate": 2.6554e-06, "loss": 0.0656, "step": 36724 }, { "epoch": 1.9579357040038385, "grad_norm": 1.3529787063598633, "learning_rate": 2.6552e-06, "loss": 0.021, "step": 36725 }, { "epoch": 1.9579890174334915, "grad_norm": 1.0844013690948486, "learning_rate": 2.6550000000000004e-06, "loss": 0.0158, "step": 36726 }, { "epoch": 1.9580423308631443, "grad_norm": 1.9051839113235474, "learning_rate": 2.6548e-06, "loss": 0.0257, "step": 36727 }, { "epoch": 1.9580956442927975, "grad_norm": 1.83748459815979, "learning_rate": 2.6546e-06, "loss": 0.0115, "step": 36728 }, { "epoch": 1.9581489577224502, "grad_norm": 2.2676775455474854, "learning_rate": 2.6544000000000004e-06, "loss": 0.024, "step": 36729 }, { "epoch": 1.9582022711521032, "grad_norm": 2.2529666423797607, "learning_rate": 2.6542e-06, "loss": 0.0164, "step": 36730 }, { "epoch": 1.9582555845817562, "grad_norm": 2.482581853866577, "learning_rate": 2.6540000000000003e-06, "loss": 0.0145, "step": 36731 }, { "epoch": 1.958308898011409, "grad_norm": 1.3645751476287842, "learning_rate": 2.6538000000000003e-06, "loss": 0.0242, "step": 36732 }, { "epoch": 1.9583622114410622, "grad_norm": 1.2144653797149658, "learning_rate": 2.6536e-06, "loss": 0.0062, "step": 36733 }, { "epoch": 1.958415524870715, "grad_norm": 0.6061936020851135, "learning_rate": 2.6534000000000003e-06, "loss": 0.0049, "step": 36734 }, { "epoch": 1.9584688383003679, "grad_norm": 1.5195322036743164, "learning_rate": 2.6532000000000003e-06, "loss": 0.0218, "step": 36735 }, { "epoch": 1.9585221517300209, "grad_norm": 3.5662384033203125, "learning_rate": 2.653e-06, "loss": 0.0203, "step": 36736 }, { "epoch": 1.9585754651596736, "grad_norm": 2.1509175300598145, "learning_rate": 2.6528000000000002e-06, "loss": 0.0286, "step": 36737 }, { "epoch": 1.9586287785893266, "grad_norm": 0.7713494896888733, "learning_rate": 2.6526000000000002e-06, "loss": 0.0042, "step": 36738 }, { "epoch": 1.9586820920189796, "grad_norm": 1.4076271057128906, "learning_rate": 2.6524e-06, "loss": 0.0072, "step": 36739 }, { "epoch": 1.9587354054486326, "grad_norm": 1.742021918296814, "learning_rate": 2.6522e-06, "loss": 0.033, "step": 36740 }, { "epoch": 1.9587887188782855, "grad_norm": 1.1333866119384766, "learning_rate": 2.652e-06, "loss": 0.015, "step": 36741 }, { "epoch": 1.9588420323079383, "grad_norm": 2.5155575275421143, "learning_rate": 2.6518000000000006e-06, "loss": 0.0109, "step": 36742 }, { "epoch": 1.9588953457375913, "grad_norm": 2.0225777626037598, "learning_rate": 2.6516e-06, "loss": 0.0314, "step": 36743 }, { "epoch": 1.9589486591672443, "grad_norm": 1.6487886905670166, "learning_rate": 2.6514e-06, "loss": 0.0476, "step": 36744 }, { "epoch": 1.959001972596897, "grad_norm": 1.6578023433685303, "learning_rate": 2.6512000000000006e-06, "loss": 0.0174, "step": 36745 }, { "epoch": 1.9590552860265502, "grad_norm": 0.3837728798389435, "learning_rate": 2.651e-06, "loss": 0.0032, "step": 36746 }, { "epoch": 1.959108599456203, "grad_norm": 1.1321280002593994, "learning_rate": 2.6508e-06, "loss": 0.0138, "step": 36747 }, { "epoch": 1.959161912885856, "grad_norm": 1.683467984199524, "learning_rate": 2.6506000000000005e-06, "loss": 0.0075, "step": 36748 }, { "epoch": 1.959215226315509, "grad_norm": 1.1806942224502563, "learning_rate": 2.6504e-06, "loss": 0.0196, "step": 36749 }, { "epoch": 1.9592685397451617, "grad_norm": 2.9305126667022705, "learning_rate": 2.6502e-06, "loss": 0.0232, "step": 36750 }, { "epoch": 1.9593218531748149, "grad_norm": 2.483067274093628, "learning_rate": 2.6500000000000005e-06, "loss": 0.0147, "step": 36751 }, { "epoch": 1.9593751666044676, "grad_norm": 0.746056079864502, "learning_rate": 2.6498e-06, "loss": 0.0064, "step": 36752 }, { "epoch": 1.9594284800341206, "grad_norm": 1.938107967376709, "learning_rate": 2.6496e-06, "loss": 0.0278, "step": 36753 }, { "epoch": 1.9594817934637736, "grad_norm": 2.4329004287719727, "learning_rate": 2.6494000000000004e-06, "loss": 0.0478, "step": 36754 }, { "epoch": 1.9595351068934264, "grad_norm": 1.5365883111953735, "learning_rate": 2.6492e-06, "loss": 0.0377, "step": 36755 }, { "epoch": 1.9595884203230793, "grad_norm": 2.050020456314087, "learning_rate": 2.6490000000000004e-06, "loss": 0.0134, "step": 36756 }, { "epoch": 1.9596417337527323, "grad_norm": 1.4417897462844849, "learning_rate": 2.6488000000000004e-06, "loss": 0.012, "step": 36757 }, { "epoch": 1.9596950471823853, "grad_norm": 1.1384267807006836, "learning_rate": 2.6486e-06, "loss": 0.0088, "step": 36758 }, { "epoch": 1.9597483606120383, "grad_norm": 0.9720152020454407, "learning_rate": 2.6484000000000003e-06, "loss": 0.0082, "step": 36759 }, { "epoch": 1.959801674041691, "grad_norm": 1.1794695854187012, "learning_rate": 2.6482000000000003e-06, "loss": 0.0112, "step": 36760 }, { "epoch": 1.959854987471344, "grad_norm": 3.889599323272705, "learning_rate": 2.648e-06, "loss": 0.0253, "step": 36761 }, { "epoch": 1.959908300900997, "grad_norm": 2.8802735805511475, "learning_rate": 2.6478000000000003e-06, "loss": 0.0291, "step": 36762 }, { "epoch": 1.9599616143306497, "grad_norm": 2.1502935886383057, "learning_rate": 2.6476000000000003e-06, "loss": 0.0182, "step": 36763 }, { "epoch": 1.960014927760303, "grad_norm": 3.084246873855591, "learning_rate": 2.6474e-06, "loss": 0.0399, "step": 36764 }, { "epoch": 1.9600682411899557, "grad_norm": 2.128653049468994, "learning_rate": 2.6472000000000003e-06, "loss": 0.0095, "step": 36765 }, { "epoch": 1.9601215546196087, "grad_norm": 2.19337797164917, "learning_rate": 2.6470000000000002e-06, "loss": 0.0199, "step": 36766 }, { "epoch": 1.9601748680492617, "grad_norm": 1.761264443397522, "learning_rate": 2.6468000000000006e-06, "loss": 0.0124, "step": 36767 }, { "epoch": 1.9602281814789144, "grad_norm": 1.7021173238754272, "learning_rate": 2.6466e-06, "loss": 0.0097, "step": 36768 }, { "epoch": 1.9602814949085676, "grad_norm": 1.7659906148910522, "learning_rate": 2.6464e-06, "loss": 0.0119, "step": 36769 }, { "epoch": 1.9603348083382204, "grad_norm": 1.1227458715438843, "learning_rate": 2.6462000000000006e-06, "loss": 0.0272, "step": 36770 }, { "epoch": 1.9603881217678734, "grad_norm": 0.7798946499824524, "learning_rate": 2.646e-06, "loss": 0.0046, "step": 36771 }, { "epoch": 1.9604414351975263, "grad_norm": 1.8210841417312622, "learning_rate": 2.6458e-06, "loss": 0.0203, "step": 36772 }, { "epoch": 1.960494748627179, "grad_norm": 1.649709701538086, "learning_rate": 2.6456000000000006e-06, "loss": 0.0091, "step": 36773 }, { "epoch": 1.960548062056832, "grad_norm": 1.5854201316833496, "learning_rate": 2.6454e-06, "loss": 0.0128, "step": 36774 }, { "epoch": 1.960601375486485, "grad_norm": 3.806743621826172, "learning_rate": 2.6452e-06, "loss": 0.0191, "step": 36775 }, { "epoch": 1.9606546889161378, "grad_norm": 6.142251014709473, "learning_rate": 2.6450000000000005e-06, "loss": 0.0294, "step": 36776 }, { "epoch": 1.960708002345791, "grad_norm": 1.4664407968521118, "learning_rate": 2.6448e-06, "loss": 0.0243, "step": 36777 }, { "epoch": 1.9607613157754438, "grad_norm": 2.899303436279297, "learning_rate": 2.6446e-06, "loss": 0.0229, "step": 36778 }, { "epoch": 1.9608146292050967, "grad_norm": 3.5079612731933594, "learning_rate": 2.6444000000000005e-06, "loss": 0.0147, "step": 36779 }, { "epoch": 1.9608679426347497, "grad_norm": 1.1358064413070679, "learning_rate": 2.6442e-06, "loss": 0.0151, "step": 36780 }, { "epoch": 1.9609212560644025, "grad_norm": 1.1090646982192993, "learning_rate": 2.6440000000000004e-06, "loss": 0.0239, "step": 36781 }, { "epoch": 1.9609745694940557, "grad_norm": 1.806002140045166, "learning_rate": 2.6438000000000004e-06, "loss": 0.0192, "step": 36782 }, { "epoch": 1.9610278829237084, "grad_norm": 2.3486475944519043, "learning_rate": 2.6436e-06, "loss": 0.0144, "step": 36783 }, { "epoch": 1.9610811963533614, "grad_norm": 1.7811347246170044, "learning_rate": 2.6434000000000004e-06, "loss": 0.0283, "step": 36784 }, { "epoch": 1.9611345097830144, "grad_norm": 1.7156188488006592, "learning_rate": 2.6432000000000004e-06, "loss": 0.0172, "step": 36785 }, { "epoch": 1.9611878232126672, "grad_norm": 0.8580625653266907, "learning_rate": 2.643e-06, "loss": 0.0043, "step": 36786 }, { "epoch": 1.9612411366423204, "grad_norm": 2.597907543182373, "learning_rate": 2.6428000000000003e-06, "loss": 0.0167, "step": 36787 }, { "epoch": 1.9612944500719731, "grad_norm": 1.3312710523605347, "learning_rate": 2.6426000000000003e-06, "loss": 0.0089, "step": 36788 }, { "epoch": 1.961347763501626, "grad_norm": 1.713089108467102, "learning_rate": 2.6424e-06, "loss": 0.0208, "step": 36789 }, { "epoch": 1.961401076931279, "grad_norm": 1.4042912721633911, "learning_rate": 2.6422000000000003e-06, "loss": 0.01, "step": 36790 }, { "epoch": 1.9614543903609318, "grad_norm": 1.695387601852417, "learning_rate": 2.6420000000000003e-06, "loss": 0.0064, "step": 36791 }, { "epoch": 1.9615077037905848, "grad_norm": 5.009525775909424, "learning_rate": 2.6418000000000003e-06, "loss": 0.0429, "step": 36792 }, { "epoch": 1.9615610172202378, "grad_norm": 0.6469501256942749, "learning_rate": 2.6416000000000003e-06, "loss": 0.0042, "step": 36793 }, { "epoch": 1.9616143306498905, "grad_norm": 5.35708475112915, "learning_rate": 2.6414000000000002e-06, "loss": 0.0309, "step": 36794 }, { "epoch": 1.9616676440795437, "grad_norm": 1.5451377630233765, "learning_rate": 2.6412000000000002e-06, "loss": 0.0168, "step": 36795 }, { "epoch": 1.9617209575091965, "grad_norm": 2.275554895401001, "learning_rate": 2.641e-06, "loss": 0.0272, "step": 36796 }, { "epoch": 1.9617742709388495, "grad_norm": 0.6377214789390564, "learning_rate": 2.6408e-06, "loss": 0.0043, "step": 36797 }, { "epoch": 1.9618275843685025, "grad_norm": 2.3748679161071777, "learning_rate": 2.6406e-06, "loss": 0.0295, "step": 36798 }, { "epoch": 1.9618808977981552, "grad_norm": 0.9714193940162659, "learning_rate": 2.6404e-06, "loss": 0.0063, "step": 36799 }, { "epoch": 1.9619342112278084, "grad_norm": 2.0240955352783203, "learning_rate": 2.6402e-06, "loss": 0.0075, "step": 36800 }, { "epoch": 1.9619875246574612, "grad_norm": 1.320327877998352, "learning_rate": 2.64e-06, "loss": 0.0255, "step": 36801 }, { "epoch": 1.9620408380871142, "grad_norm": 0.18795499205589294, "learning_rate": 2.6398e-06, "loss": 0.0017, "step": 36802 }, { "epoch": 1.9620941515167671, "grad_norm": 1.4406797885894775, "learning_rate": 2.6396e-06, "loss": 0.0168, "step": 36803 }, { "epoch": 1.96214746494642, "grad_norm": 1.9980792999267578, "learning_rate": 2.6394e-06, "loss": 0.0143, "step": 36804 }, { "epoch": 1.962200778376073, "grad_norm": 1.4033411741256714, "learning_rate": 2.6392e-06, "loss": 0.0195, "step": 36805 }, { "epoch": 1.9622540918057259, "grad_norm": 3.62734317779541, "learning_rate": 2.6390000000000005e-06, "loss": 0.0262, "step": 36806 }, { "epoch": 1.9623074052353788, "grad_norm": 0.5381782054901123, "learning_rate": 2.6388e-06, "loss": 0.0036, "step": 36807 }, { "epoch": 1.9623607186650318, "grad_norm": 2.8857295513153076, "learning_rate": 2.6386e-06, "loss": 0.0261, "step": 36808 }, { "epoch": 1.9624140320946846, "grad_norm": 0.9172243475914001, "learning_rate": 2.6384000000000004e-06, "loss": 0.0042, "step": 36809 }, { "epoch": 1.9624673455243375, "grad_norm": 1.0885957479476929, "learning_rate": 2.6382e-06, "loss": 0.0067, "step": 36810 }, { "epoch": 1.9625206589539905, "grad_norm": 1.7279077768325806, "learning_rate": 2.638e-06, "loss": 0.0169, "step": 36811 }, { "epoch": 1.9625739723836433, "grad_norm": 0.9230815172195435, "learning_rate": 2.6378000000000004e-06, "loss": 0.0049, "step": 36812 }, { "epoch": 1.9626272858132965, "grad_norm": 1.835587739944458, "learning_rate": 2.6376e-06, "loss": 0.0066, "step": 36813 }, { "epoch": 1.9626805992429492, "grad_norm": 3.1594369411468506, "learning_rate": 2.6374e-06, "loss": 0.0182, "step": 36814 }, { "epoch": 1.9627339126726022, "grad_norm": 0.6535626649856567, "learning_rate": 2.6372000000000003e-06, "loss": 0.0044, "step": 36815 }, { "epoch": 1.9627872261022552, "grad_norm": 2.3054909706115723, "learning_rate": 2.637e-06, "loss": 0.0083, "step": 36816 }, { "epoch": 1.962840539531908, "grad_norm": 3.27827787399292, "learning_rate": 2.6368000000000003e-06, "loss": 0.0527, "step": 36817 }, { "epoch": 1.9628938529615612, "grad_norm": 2.9127817153930664, "learning_rate": 2.6366000000000003e-06, "loss": 0.0075, "step": 36818 }, { "epoch": 1.962947166391214, "grad_norm": 1.8818150758743286, "learning_rate": 2.6364e-06, "loss": 0.0071, "step": 36819 }, { "epoch": 1.963000479820867, "grad_norm": 2.904555320739746, "learning_rate": 2.6362000000000003e-06, "loss": 0.0228, "step": 36820 }, { "epoch": 1.9630537932505199, "grad_norm": 2.385484457015991, "learning_rate": 2.6360000000000003e-06, "loss": 0.0233, "step": 36821 }, { "epoch": 1.9631071066801726, "grad_norm": 1.6651420593261719, "learning_rate": 2.6358e-06, "loss": 0.0263, "step": 36822 }, { "epoch": 1.9631604201098256, "grad_norm": 2.6307625770568848, "learning_rate": 2.6356000000000002e-06, "loss": 0.0358, "step": 36823 }, { "epoch": 1.9632137335394786, "grad_norm": 2.978280782699585, "learning_rate": 2.6354000000000002e-06, "loss": 0.0252, "step": 36824 }, { "epoch": 1.9632670469691316, "grad_norm": 1.0064109563827515, "learning_rate": 2.6351999999999998e-06, "loss": 0.0065, "step": 36825 }, { "epoch": 1.9633203603987845, "grad_norm": 2.6249587535858154, "learning_rate": 2.635e-06, "loss": 0.0192, "step": 36826 }, { "epoch": 1.9633736738284373, "grad_norm": 2.9725589752197266, "learning_rate": 2.6348e-06, "loss": 0.0241, "step": 36827 }, { "epoch": 1.9634269872580903, "grad_norm": 1.5992449522018433, "learning_rate": 2.6346000000000006e-06, "loss": 0.0211, "step": 36828 }, { "epoch": 1.9634803006877433, "grad_norm": 1.8879979848861694, "learning_rate": 2.6344e-06, "loss": 0.0107, "step": 36829 }, { "epoch": 1.963533614117396, "grad_norm": 2.586498975753784, "learning_rate": 2.6342e-06, "loss": 0.0269, "step": 36830 }, { "epoch": 1.9635869275470492, "grad_norm": 0.7283133864402771, "learning_rate": 2.6340000000000005e-06, "loss": 0.015, "step": 36831 }, { "epoch": 1.963640240976702, "grad_norm": 1.9996557235717773, "learning_rate": 2.6338e-06, "loss": 0.0426, "step": 36832 }, { "epoch": 1.963693554406355, "grad_norm": 1.2747403383255005, "learning_rate": 2.6336e-06, "loss": 0.0086, "step": 36833 }, { "epoch": 1.963746867836008, "grad_norm": 1.8395628929138184, "learning_rate": 2.6334000000000005e-06, "loss": 0.0078, "step": 36834 }, { "epoch": 1.9638001812656607, "grad_norm": 2.64054536819458, "learning_rate": 2.6332e-06, "loss": 0.0189, "step": 36835 }, { "epoch": 1.963853494695314, "grad_norm": 1.5113728046417236, "learning_rate": 2.633e-06, "loss": 0.0066, "step": 36836 }, { "epoch": 1.9639068081249667, "grad_norm": 1.9150407314300537, "learning_rate": 2.6328000000000004e-06, "loss": 0.0329, "step": 36837 }, { "epoch": 1.9639601215546196, "grad_norm": 1.0542495250701904, "learning_rate": 2.6326e-06, "loss": 0.0109, "step": 36838 }, { "epoch": 1.9640134349842726, "grad_norm": 1.055434226989746, "learning_rate": 2.6324e-06, "loss": 0.0043, "step": 36839 }, { "epoch": 1.9640667484139254, "grad_norm": 2.0379817485809326, "learning_rate": 2.6322000000000004e-06, "loss": 0.0404, "step": 36840 }, { "epoch": 1.9641200618435783, "grad_norm": 3.615081548690796, "learning_rate": 2.632e-06, "loss": 0.0237, "step": 36841 }, { "epoch": 1.9641733752732313, "grad_norm": 2.319852590560913, "learning_rate": 2.6318000000000004e-06, "loss": 0.0103, "step": 36842 }, { "epoch": 1.9642266887028843, "grad_norm": 1.452177882194519, "learning_rate": 2.6316000000000004e-06, "loss": 0.023, "step": 36843 }, { "epoch": 1.9642800021325373, "grad_norm": 1.3824801445007324, "learning_rate": 2.6314e-06, "loss": 0.0066, "step": 36844 }, { "epoch": 1.96433331556219, "grad_norm": 0.5498030781745911, "learning_rate": 2.6312000000000003e-06, "loss": 0.0045, "step": 36845 }, { "epoch": 1.964386628991843, "grad_norm": 1.6715768575668335, "learning_rate": 2.6310000000000003e-06, "loss": 0.015, "step": 36846 }, { "epoch": 1.964439942421496, "grad_norm": 0.6935515999794006, "learning_rate": 2.6308e-06, "loss": 0.0051, "step": 36847 }, { "epoch": 1.9644932558511488, "grad_norm": 3.5601742267608643, "learning_rate": 2.6306000000000003e-06, "loss": 0.0197, "step": 36848 }, { "epoch": 1.964546569280802, "grad_norm": 1.8201816082000732, "learning_rate": 2.6304000000000003e-06, "loss": 0.0133, "step": 36849 }, { "epoch": 1.9645998827104547, "grad_norm": 2.773965358734131, "learning_rate": 2.6302e-06, "loss": 0.0322, "step": 36850 }, { "epoch": 1.9646531961401077, "grad_norm": 2.4037089347839355, "learning_rate": 2.6300000000000002e-06, "loss": 0.0117, "step": 36851 }, { "epoch": 1.9647065095697607, "grad_norm": 0.40519067645072937, "learning_rate": 2.6298000000000002e-06, "loss": 0.0026, "step": 36852 }, { "epoch": 1.9647598229994134, "grad_norm": 2.0016674995422363, "learning_rate": 2.6296000000000006e-06, "loss": 0.016, "step": 36853 }, { "epoch": 1.9648131364290666, "grad_norm": 2.6194233894348145, "learning_rate": 2.6294e-06, "loss": 0.0078, "step": 36854 }, { "epoch": 1.9648664498587194, "grad_norm": 0.3797225058078766, "learning_rate": 2.6292e-06, "loss": 0.003, "step": 36855 }, { "epoch": 1.9649197632883724, "grad_norm": 2.6922833919525146, "learning_rate": 2.6290000000000006e-06, "loss": 0.0226, "step": 36856 }, { "epoch": 1.9649730767180253, "grad_norm": 2.8007888793945312, "learning_rate": 2.6288e-06, "loss": 0.0149, "step": 36857 }, { "epoch": 1.965026390147678, "grad_norm": 0.9797818064689636, "learning_rate": 2.6286e-06, "loss": 0.0164, "step": 36858 }, { "epoch": 1.965079703577331, "grad_norm": 5.646793842315674, "learning_rate": 2.6284000000000005e-06, "loss": 0.0457, "step": 36859 }, { "epoch": 1.965133017006984, "grad_norm": 0.9527395367622375, "learning_rate": 2.6282e-06, "loss": 0.0048, "step": 36860 }, { "epoch": 1.965186330436637, "grad_norm": 1.2688125371932983, "learning_rate": 2.628e-06, "loss": 0.0075, "step": 36861 }, { "epoch": 1.96523964386629, "grad_norm": 2.5276288986206055, "learning_rate": 2.6278000000000005e-06, "loss": 0.0201, "step": 36862 }, { "epoch": 1.9652929572959428, "grad_norm": 0.23364783823490143, "learning_rate": 2.6276e-06, "loss": 0.0022, "step": 36863 }, { "epoch": 1.9653462707255958, "grad_norm": 1.006257176399231, "learning_rate": 2.6274e-06, "loss": 0.0142, "step": 36864 }, { "epoch": 1.9653995841552487, "grad_norm": 1.15315842628479, "learning_rate": 2.6272000000000004e-06, "loss": 0.0144, "step": 36865 }, { "epoch": 1.9654528975849015, "grad_norm": 0.3636193871498108, "learning_rate": 2.627e-06, "loss": 0.0029, "step": 36866 }, { "epoch": 1.9655062110145547, "grad_norm": 1.2208662033081055, "learning_rate": 2.6268000000000004e-06, "loss": 0.0044, "step": 36867 }, { "epoch": 1.9655595244442075, "grad_norm": 0.3659379780292511, "learning_rate": 2.6266000000000004e-06, "loss": 0.0019, "step": 36868 }, { "epoch": 1.9656128378738604, "grad_norm": 5.216156482696533, "learning_rate": 2.6264e-06, "loss": 0.048, "step": 36869 }, { "epoch": 1.9656661513035134, "grad_norm": 1.302699089050293, "learning_rate": 2.6262000000000004e-06, "loss": 0.0048, "step": 36870 }, { "epoch": 1.9657194647331662, "grad_norm": 1.784411907196045, "learning_rate": 2.6260000000000004e-06, "loss": 0.0362, "step": 36871 }, { "epoch": 1.9657727781628194, "grad_norm": 1.0047070980072021, "learning_rate": 2.6258e-06, "loss": 0.0044, "step": 36872 }, { "epoch": 1.9658260915924721, "grad_norm": 1.3523716926574707, "learning_rate": 2.6256000000000003e-06, "loss": 0.0174, "step": 36873 }, { "epoch": 1.965879405022125, "grad_norm": 1.419871211051941, "learning_rate": 2.6254000000000003e-06, "loss": 0.0219, "step": 36874 }, { "epoch": 1.965932718451778, "grad_norm": 2.1232855319976807, "learning_rate": 2.6252e-06, "loss": 0.01, "step": 36875 }, { "epoch": 1.9659860318814308, "grad_norm": 0.40746259689331055, "learning_rate": 2.6250000000000003e-06, "loss": 0.0034, "step": 36876 }, { "epoch": 1.9660393453110838, "grad_norm": 2.401942014694214, "learning_rate": 2.6248000000000003e-06, "loss": 0.0126, "step": 36877 }, { "epoch": 1.9660926587407368, "grad_norm": 2.4441542625427246, "learning_rate": 2.6246000000000007e-06, "loss": 0.03, "step": 36878 }, { "epoch": 1.9661459721703896, "grad_norm": 2.8360519409179688, "learning_rate": 2.6244000000000002e-06, "loss": 0.0125, "step": 36879 }, { "epoch": 1.9661992856000428, "grad_norm": 1.9167906045913696, "learning_rate": 2.6242000000000002e-06, "loss": 0.0234, "step": 36880 }, { "epoch": 1.9662525990296955, "grad_norm": 5.948599338531494, "learning_rate": 2.6240000000000006e-06, "loss": 0.0396, "step": 36881 }, { "epoch": 1.9663059124593485, "grad_norm": 5.072942733764648, "learning_rate": 2.6238e-06, "loss": 0.0207, "step": 36882 }, { "epoch": 1.9663592258890015, "grad_norm": 2.5743300914764404, "learning_rate": 2.6236e-06, "loss": 0.0172, "step": 36883 }, { "epoch": 1.9664125393186542, "grad_norm": 1.9532712697982788, "learning_rate": 2.6234000000000006e-06, "loss": 0.0263, "step": 36884 }, { "epoch": 1.9664658527483074, "grad_norm": 0.9969794750213623, "learning_rate": 2.6232e-06, "loss": 0.0066, "step": 36885 }, { "epoch": 1.9665191661779602, "grad_norm": 2.444723606109619, "learning_rate": 2.623e-06, "loss": 0.0345, "step": 36886 }, { "epoch": 1.9665724796076132, "grad_norm": 0.7959157824516296, "learning_rate": 2.6228000000000005e-06, "loss": 0.0147, "step": 36887 }, { "epoch": 1.9666257930372661, "grad_norm": 0.8213690519332886, "learning_rate": 2.6226e-06, "loss": 0.0137, "step": 36888 }, { "epoch": 1.966679106466919, "grad_norm": 1.1097928285598755, "learning_rate": 2.6224e-06, "loss": 0.0044, "step": 36889 }, { "epoch": 1.966732419896572, "grad_norm": 0.8902462720870972, "learning_rate": 2.6222000000000005e-06, "loss": 0.0055, "step": 36890 }, { "epoch": 1.9667857333262249, "grad_norm": 1.0093055963516235, "learning_rate": 2.622e-06, "loss": 0.0054, "step": 36891 }, { "epoch": 1.9668390467558778, "grad_norm": 1.738654375076294, "learning_rate": 2.6218000000000005e-06, "loss": 0.009, "step": 36892 }, { "epoch": 1.9668923601855308, "grad_norm": 2.436185598373413, "learning_rate": 2.6216000000000004e-06, "loss": 0.0404, "step": 36893 }, { "epoch": 1.9669456736151836, "grad_norm": 2.2103376388549805, "learning_rate": 2.6214e-06, "loss": 0.0179, "step": 36894 }, { "epoch": 1.9669989870448366, "grad_norm": 1.3052232265472412, "learning_rate": 2.6212000000000004e-06, "loss": 0.0083, "step": 36895 }, { "epoch": 1.9670523004744895, "grad_norm": 1.3478039503097534, "learning_rate": 2.6210000000000004e-06, "loss": 0.0205, "step": 36896 }, { "epoch": 1.9671056139041423, "grad_norm": 1.0557479858398438, "learning_rate": 2.6208e-06, "loss": 0.0176, "step": 36897 }, { "epoch": 1.9671589273337955, "grad_norm": 2.243462085723877, "learning_rate": 2.6206000000000004e-06, "loss": 0.0303, "step": 36898 }, { "epoch": 1.9672122407634482, "grad_norm": 5.775481224060059, "learning_rate": 2.6204000000000004e-06, "loss": 0.0217, "step": 36899 }, { "epoch": 1.9672655541931012, "grad_norm": 0.9511850476264954, "learning_rate": 2.6202e-06, "loss": 0.0043, "step": 36900 }, { "epoch": 1.9673188676227542, "grad_norm": 1.3032541275024414, "learning_rate": 2.6200000000000003e-06, "loss": 0.0071, "step": 36901 }, { "epoch": 1.967372181052407, "grad_norm": 0.8302825689315796, "learning_rate": 2.6198000000000003e-06, "loss": 0.0032, "step": 36902 }, { "epoch": 1.9674254944820602, "grad_norm": 2.6322379112243652, "learning_rate": 2.6196000000000003e-06, "loss": 0.0154, "step": 36903 }, { "epoch": 1.967478807911713, "grad_norm": 1.2532427310943604, "learning_rate": 2.6194000000000003e-06, "loss": 0.017, "step": 36904 }, { "epoch": 1.967532121341366, "grad_norm": 1.5266607999801636, "learning_rate": 2.6192000000000003e-06, "loss": 0.0098, "step": 36905 }, { "epoch": 1.9675854347710189, "grad_norm": 1.1147156953811646, "learning_rate": 2.6190000000000003e-06, "loss": 0.0066, "step": 36906 }, { "epoch": 1.9676387482006716, "grad_norm": 2.0228118896484375, "learning_rate": 2.6188000000000002e-06, "loss": 0.0411, "step": 36907 }, { "epoch": 1.9676920616303248, "grad_norm": 2.9769914150238037, "learning_rate": 2.6186000000000002e-06, "loss": 0.0203, "step": 36908 }, { "epoch": 1.9677453750599776, "grad_norm": 2.994767189025879, "learning_rate": 2.6184e-06, "loss": 0.0161, "step": 36909 }, { "epoch": 1.9677986884896306, "grad_norm": 1.2538433074951172, "learning_rate": 2.6182e-06, "loss": 0.0063, "step": 36910 }, { "epoch": 1.9678520019192836, "grad_norm": 2.4523115158081055, "learning_rate": 2.618e-06, "loss": 0.036, "step": 36911 }, { "epoch": 1.9679053153489363, "grad_norm": 1.8292429447174072, "learning_rate": 2.6178e-06, "loss": 0.0125, "step": 36912 }, { "epoch": 1.9679586287785893, "grad_norm": 1.4880356788635254, "learning_rate": 2.6176e-06, "loss": 0.0329, "step": 36913 }, { "epoch": 1.9680119422082423, "grad_norm": 1.6672706604003906, "learning_rate": 2.6174e-06, "loss": 0.019, "step": 36914 }, { "epoch": 1.968065255637895, "grad_norm": 2.853126049041748, "learning_rate": 2.6172e-06, "loss": 0.0207, "step": 36915 }, { "epoch": 1.9681185690675482, "grad_norm": 4.6334075927734375, "learning_rate": 2.617e-06, "loss": 0.059, "step": 36916 }, { "epoch": 1.968171882497201, "grad_norm": 0.7014897465705872, "learning_rate": 2.6168000000000005e-06, "loss": 0.0056, "step": 36917 }, { "epoch": 1.968225195926854, "grad_norm": 0.7209856510162354, "learning_rate": 2.6166e-06, "loss": 0.0054, "step": 36918 }, { "epoch": 1.968278509356507, "grad_norm": 2.882331132888794, "learning_rate": 2.6164e-06, "loss": 0.0317, "step": 36919 }, { "epoch": 1.9683318227861597, "grad_norm": 2.5505878925323486, "learning_rate": 2.6162000000000005e-06, "loss": 0.0139, "step": 36920 }, { "epoch": 1.968385136215813, "grad_norm": 2.8237369060516357, "learning_rate": 2.616e-06, "loss": 0.0217, "step": 36921 }, { "epoch": 1.9684384496454657, "grad_norm": 4.022904872894287, "learning_rate": 2.6158e-06, "loss": 0.042, "step": 36922 }, { "epoch": 1.9684917630751186, "grad_norm": 0.5162436962127686, "learning_rate": 2.6156000000000004e-06, "loss": 0.0033, "step": 36923 }, { "epoch": 1.9685450765047716, "grad_norm": 1.6375426054000854, "learning_rate": 2.6154e-06, "loss": 0.011, "step": 36924 }, { "epoch": 1.9685983899344244, "grad_norm": 2.604060173034668, "learning_rate": 2.6152e-06, "loss": 0.0331, "step": 36925 }, { "epoch": 1.9686517033640776, "grad_norm": 3.433366060256958, "learning_rate": 2.6150000000000004e-06, "loss": 0.0193, "step": 36926 }, { "epoch": 1.9687050167937303, "grad_norm": 1.18492591381073, "learning_rate": 2.6148e-06, "loss": 0.0089, "step": 36927 }, { "epoch": 1.9687583302233833, "grad_norm": 0.8390995264053345, "learning_rate": 2.6146000000000003e-06, "loss": 0.0054, "step": 36928 }, { "epoch": 1.9688116436530363, "grad_norm": 2.3311500549316406, "learning_rate": 2.6144000000000003e-06, "loss": 0.0142, "step": 36929 }, { "epoch": 1.968864957082689, "grad_norm": 1.5439772605895996, "learning_rate": 2.6142e-06, "loss": 0.025, "step": 36930 }, { "epoch": 1.968918270512342, "grad_norm": 1.1844009160995483, "learning_rate": 2.6140000000000003e-06, "loss": 0.0261, "step": 36931 }, { "epoch": 1.968971583941995, "grad_norm": 1.085147500038147, "learning_rate": 2.6138000000000003e-06, "loss": 0.004, "step": 36932 }, { "epoch": 1.9690248973716478, "grad_norm": 3.1505537033081055, "learning_rate": 2.6136e-06, "loss": 0.0195, "step": 36933 }, { "epoch": 1.969078210801301, "grad_norm": 1.2448049783706665, "learning_rate": 2.6134000000000003e-06, "loss": 0.0209, "step": 36934 }, { "epoch": 1.9691315242309537, "grad_norm": 3.088433265686035, "learning_rate": 2.6132000000000002e-06, "loss": 0.0325, "step": 36935 }, { "epoch": 1.9691848376606067, "grad_norm": 1.8621615171432495, "learning_rate": 2.613e-06, "loss": 0.0049, "step": 36936 }, { "epoch": 1.9692381510902597, "grad_norm": 2.1718590259552, "learning_rate": 2.6128e-06, "loss": 0.0129, "step": 36937 }, { "epoch": 1.9692914645199124, "grad_norm": 2.3078038692474365, "learning_rate": 2.6126e-06, "loss": 0.0245, "step": 36938 }, { "epoch": 1.9693447779495656, "grad_norm": 2.019900321960449, "learning_rate": 2.6124000000000006e-06, "loss": 0.028, "step": 36939 }, { "epoch": 1.9693980913792184, "grad_norm": 1.7520201206207275, "learning_rate": 2.6122e-06, "loss": 0.0275, "step": 36940 }, { "epoch": 1.9694514048088714, "grad_norm": 2.895658493041992, "learning_rate": 2.612e-06, "loss": 0.0171, "step": 36941 }, { "epoch": 1.9695047182385244, "grad_norm": 2.3378050327301025, "learning_rate": 2.6118000000000006e-06, "loss": 0.0099, "step": 36942 }, { "epoch": 1.9695580316681771, "grad_norm": 1.534867525100708, "learning_rate": 2.6116e-06, "loss": 0.0148, "step": 36943 }, { "epoch": 1.96961134509783, "grad_norm": 1.6270136833190918, "learning_rate": 2.6114e-06, "loss": 0.0084, "step": 36944 }, { "epoch": 1.969664658527483, "grad_norm": 4.158285140991211, "learning_rate": 2.6112000000000005e-06, "loss": 0.0424, "step": 36945 }, { "epoch": 1.969717971957136, "grad_norm": 1.1746646165847778, "learning_rate": 2.611e-06, "loss": 0.0072, "step": 36946 }, { "epoch": 1.969771285386789, "grad_norm": 0.6266760230064392, "learning_rate": 2.6108e-06, "loss": 0.004, "step": 36947 }, { "epoch": 1.9698245988164418, "grad_norm": 5.467632293701172, "learning_rate": 2.6106000000000005e-06, "loss": 0.0311, "step": 36948 }, { "epoch": 1.9698779122460948, "grad_norm": 1.6618086099624634, "learning_rate": 2.6104e-06, "loss": 0.0335, "step": 36949 }, { "epoch": 1.9699312256757477, "grad_norm": 2.1078689098358154, "learning_rate": 2.6102e-06, "loss": 0.0118, "step": 36950 }, { "epoch": 1.9699845391054005, "grad_norm": 6.340975284576416, "learning_rate": 2.6100000000000004e-06, "loss": 0.0553, "step": 36951 }, { "epoch": 1.9700378525350537, "grad_norm": 0.4678748548030853, "learning_rate": 2.6098e-06, "loss": 0.0045, "step": 36952 }, { "epoch": 1.9700911659647065, "grad_norm": 1.4319980144500732, "learning_rate": 2.6096000000000004e-06, "loss": 0.0088, "step": 36953 }, { "epoch": 1.9701444793943594, "grad_norm": 1.038435697555542, "learning_rate": 2.6094000000000004e-06, "loss": 0.0058, "step": 36954 }, { "epoch": 1.9701977928240124, "grad_norm": 1.8985576629638672, "learning_rate": 2.6092e-06, "loss": 0.0097, "step": 36955 }, { "epoch": 1.9702511062536652, "grad_norm": 1.1494742631912231, "learning_rate": 2.6090000000000003e-06, "loss": 0.0204, "step": 36956 }, { "epoch": 1.9703044196833184, "grad_norm": 1.605072021484375, "learning_rate": 2.6088000000000003e-06, "loss": 0.0085, "step": 36957 }, { "epoch": 1.9703577331129711, "grad_norm": 2.7935218811035156, "learning_rate": 2.6086e-06, "loss": 0.0253, "step": 36958 }, { "epoch": 1.9704110465426241, "grad_norm": 4.992126941680908, "learning_rate": 2.6084000000000003e-06, "loss": 0.0226, "step": 36959 }, { "epoch": 1.970464359972277, "grad_norm": 0.9912683367729187, "learning_rate": 2.6082000000000003e-06, "loss": 0.0092, "step": 36960 }, { "epoch": 1.9705176734019298, "grad_norm": 0.4168887734413147, "learning_rate": 2.608e-06, "loss": 0.0022, "step": 36961 }, { "epoch": 1.9705709868315828, "grad_norm": 2.6254804134368896, "learning_rate": 2.6078000000000003e-06, "loss": 0.0168, "step": 36962 }, { "epoch": 1.9706243002612358, "grad_norm": 1.5568256378173828, "learning_rate": 2.6076000000000002e-06, "loss": 0.0295, "step": 36963 }, { "epoch": 1.9706776136908888, "grad_norm": 2.470914363861084, "learning_rate": 2.6074000000000007e-06, "loss": 0.0582, "step": 36964 }, { "epoch": 1.9707309271205418, "grad_norm": 2.1795074939727783, "learning_rate": 2.6072e-06, "loss": 0.0196, "step": 36965 }, { "epoch": 1.9707842405501945, "grad_norm": 2.8838164806365967, "learning_rate": 2.607e-06, "loss": 0.0253, "step": 36966 }, { "epoch": 1.9708375539798475, "grad_norm": 1.9723420143127441, "learning_rate": 2.6068000000000006e-06, "loss": 0.0223, "step": 36967 }, { "epoch": 1.9708908674095005, "grad_norm": 1.1923733949661255, "learning_rate": 2.6066e-06, "loss": 0.0242, "step": 36968 }, { "epoch": 1.9709441808391532, "grad_norm": 1.8089202642440796, "learning_rate": 2.6064e-06, "loss": 0.0105, "step": 36969 }, { "epoch": 1.9709974942688064, "grad_norm": 3.1417837142944336, "learning_rate": 2.6062000000000006e-06, "loss": 0.0377, "step": 36970 }, { "epoch": 1.9710508076984592, "grad_norm": 1.2073665857315063, "learning_rate": 2.606e-06, "loss": 0.0103, "step": 36971 }, { "epoch": 1.9711041211281122, "grad_norm": 1.238875150680542, "learning_rate": 2.6058e-06, "loss": 0.0111, "step": 36972 }, { "epoch": 1.9711574345577652, "grad_norm": 0.8476841449737549, "learning_rate": 2.6056000000000005e-06, "loss": 0.0065, "step": 36973 }, { "epoch": 1.971210747987418, "grad_norm": 2.7923097610473633, "learning_rate": 2.6054e-06, "loss": 0.0404, "step": 36974 }, { "epoch": 1.9712640614170711, "grad_norm": 1.719577670097351, "learning_rate": 2.6052e-06, "loss": 0.0064, "step": 36975 }, { "epoch": 1.9713173748467239, "grad_norm": 2.298170328140259, "learning_rate": 2.6050000000000005e-06, "loss": 0.0108, "step": 36976 }, { "epoch": 1.9713706882763768, "grad_norm": 1.4494884014129639, "learning_rate": 2.6048e-06, "loss": 0.0146, "step": 36977 }, { "epoch": 1.9714240017060298, "grad_norm": 1.4877427816390991, "learning_rate": 2.6046000000000004e-06, "loss": 0.0237, "step": 36978 }, { "epoch": 1.9714773151356826, "grad_norm": 1.3873647451400757, "learning_rate": 2.6044000000000004e-06, "loss": 0.0249, "step": 36979 }, { "epoch": 1.9715306285653356, "grad_norm": 1.8213069438934326, "learning_rate": 2.6042e-06, "loss": 0.0177, "step": 36980 }, { "epoch": 1.9715839419949885, "grad_norm": 2.575925827026367, "learning_rate": 2.6040000000000004e-06, "loss": 0.0296, "step": 36981 }, { "epoch": 1.9716372554246415, "grad_norm": 1.337538480758667, "learning_rate": 2.6038000000000004e-06, "loss": 0.0324, "step": 36982 }, { "epoch": 1.9716905688542945, "grad_norm": 1.9641804695129395, "learning_rate": 2.6036e-06, "loss": 0.0148, "step": 36983 }, { "epoch": 1.9717438822839473, "grad_norm": 1.2424752712249756, "learning_rate": 2.6034000000000003e-06, "loss": 0.0085, "step": 36984 }, { "epoch": 1.9717971957136002, "grad_norm": 1.2569390535354614, "learning_rate": 2.6032000000000003e-06, "loss": 0.0189, "step": 36985 }, { "epoch": 1.9718505091432532, "grad_norm": 0.929449737071991, "learning_rate": 2.603e-06, "loss": 0.015, "step": 36986 }, { "epoch": 1.971903822572906, "grad_norm": 1.303841471672058, "learning_rate": 2.6028000000000003e-06, "loss": 0.0106, "step": 36987 }, { "epoch": 1.9719571360025592, "grad_norm": 5.966055393218994, "learning_rate": 2.6026000000000003e-06, "loss": 0.0197, "step": 36988 }, { "epoch": 1.972010449432212, "grad_norm": 2.5211493968963623, "learning_rate": 2.6024000000000003e-06, "loss": 0.0254, "step": 36989 }, { "epoch": 1.972063762861865, "grad_norm": 4.677289962768555, "learning_rate": 2.6022000000000003e-06, "loss": 0.0541, "step": 36990 }, { "epoch": 1.972117076291518, "grad_norm": 3.1141891479492188, "learning_rate": 2.6020000000000002e-06, "loss": 0.0267, "step": 36991 }, { "epoch": 1.9721703897211706, "grad_norm": 2.5056498050689697, "learning_rate": 2.6018000000000002e-06, "loss": 0.0489, "step": 36992 }, { "epoch": 1.9722237031508238, "grad_norm": 4.10745906829834, "learning_rate": 2.6016000000000002e-06, "loss": 0.0222, "step": 36993 }, { "epoch": 1.9722770165804766, "grad_norm": 1.7275761365890503, "learning_rate": 2.6014e-06, "loss": 0.0092, "step": 36994 }, { "epoch": 1.9723303300101296, "grad_norm": 1.5150842666625977, "learning_rate": 2.6012e-06, "loss": 0.0245, "step": 36995 }, { "epoch": 1.9723836434397826, "grad_norm": 0.7485079169273376, "learning_rate": 2.601e-06, "loss": 0.0155, "step": 36996 }, { "epoch": 1.9724369568694353, "grad_norm": 1.4077582359313965, "learning_rate": 2.6008e-06, "loss": 0.0335, "step": 36997 }, { "epoch": 1.9724902702990883, "grad_norm": 0.993230938911438, "learning_rate": 2.6006e-06, "loss": 0.0067, "step": 36998 }, { "epoch": 1.9725435837287413, "grad_norm": 1.822727084159851, "learning_rate": 2.6004e-06, "loss": 0.0217, "step": 36999 }, { "epoch": 1.972596897158394, "grad_norm": 1.2093535661697388, "learning_rate": 2.6002e-06, "loss": 0.0094, "step": 37000 }, { "epoch": 1.9726502105880472, "grad_norm": 1.6774094104766846, "learning_rate": 2.6e-06, "loss": 0.0136, "step": 37001 }, { "epoch": 1.9727035240177, "grad_norm": 0.9310871362686157, "learning_rate": 2.5998e-06, "loss": 0.0066, "step": 37002 }, { "epoch": 1.972756837447353, "grad_norm": 1.2419439554214478, "learning_rate": 2.5996000000000005e-06, "loss": 0.0092, "step": 37003 }, { "epoch": 1.972810150877006, "grad_norm": 1.5886762142181396, "learning_rate": 2.5994e-06, "loss": 0.0287, "step": 37004 }, { "epoch": 1.9728634643066587, "grad_norm": 0.7532532215118408, "learning_rate": 2.5992e-06, "loss": 0.005, "step": 37005 }, { "epoch": 1.972916777736312, "grad_norm": 1.9779452085494995, "learning_rate": 2.5990000000000004e-06, "loss": 0.0163, "step": 37006 }, { "epoch": 1.9729700911659647, "grad_norm": 0.3471885621547699, "learning_rate": 2.5988e-06, "loss": 0.0033, "step": 37007 }, { "epoch": 1.9730234045956176, "grad_norm": 1.308448076248169, "learning_rate": 2.5986e-06, "loss": 0.0254, "step": 37008 }, { "epoch": 1.9730767180252706, "grad_norm": 0.8752438426017761, "learning_rate": 2.5984000000000004e-06, "loss": 0.0196, "step": 37009 }, { "epoch": 1.9731300314549234, "grad_norm": 3.4463894367218018, "learning_rate": 2.5982e-06, "loss": 0.0484, "step": 37010 }, { "epoch": 1.9731833448845766, "grad_norm": 0.5169739127159119, "learning_rate": 2.598e-06, "loss": 0.0042, "step": 37011 }, { "epoch": 1.9732366583142293, "grad_norm": 1.3606895208358765, "learning_rate": 2.5978000000000004e-06, "loss": 0.0159, "step": 37012 }, { "epoch": 1.9732899717438823, "grad_norm": 1.9672698974609375, "learning_rate": 2.5976e-06, "loss": 0.0243, "step": 37013 }, { "epoch": 1.9733432851735353, "grad_norm": 2.153913736343384, "learning_rate": 2.5974000000000003e-06, "loss": 0.0291, "step": 37014 }, { "epoch": 1.973396598603188, "grad_norm": 1.340875267982483, "learning_rate": 2.5972000000000003e-06, "loss": 0.0347, "step": 37015 }, { "epoch": 1.973449912032841, "grad_norm": 1.4975274801254272, "learning_rate": 2.597e-06, "loss": 0.0196, "step": 37016 }, { "epoch": 1.973503225462494, "grad_norm": 0.631940484046936, "learning_rate": 2.5968000000000003e-06, "loss": 0.0042, "step": 37017 }, { "epoch": 1.9735565388921468, "grad_norm": 4.019840717315674, "learning_rate": 2.5966000000000003e-06, "loss": 0.0159, "step": 37018 }, { "epoch": 1.9736098523218, "grad_norm": 2.7695116996765137, "learning_rate": 2.5964e-06, "loss": 0.0295, "step": 37019 }, { "epoch": 1.9736631657514527, "grad_norm": 1.352439284324646, "learning_rate": 2.5962000000000002e-06, "loss": 0.016, "step": 37020 }, { "epoch": 1.9737164791811057, "grad_norm": 2.436372756958008, "learning_rate": 2.5960000000000002e-06, "loss": 0.0302, "step": 37021 }, { "epoch": 1.9737697926107587, "grad_norm": 1.1942740678787231, "learning_rate": 2.5957999999999998e-06, "loss": 0.0082, "step": 37022 }, { "epoch": 1.9738231060404114, "grad_norm": 1.7383686304092407, "learning_rate": 2.5956e-06, "loss": 0.0139, "step": 37023 }, { "epoch": 1.9738764194700646, "grad_norm": 3.1007089614868164, "learning_rate": 2.5954e-06, "loss": 0.0598, "step": 37024 }, { "epoch": 1.9739297328997174, "grad_norm": 1.0866471529006958, "learning_rate": 2.5951999999999997e-06, "loss": 0.0085, "step": 37025 }, { "epoch": 1.9739830463293704, "grad_norm": 1.334878921508789, "learning_rate": 2.595e-06, "loss": 0.0201, "step": 37026 }, { "epoch": 1.9740363597590234, "grad_norm": 1.348591685295105, "learning_rate": 2.5948e-06, "loss": 0.0113, "step": 37027 }, { "epoch": 1.9740896731886761, "grad_norm": 0.6895847916603088, "learning_rate": 2.5946000000000005e-06, "loss": 0.0062, "step": 37028 }, { "epoch": 1.9741429866183293, "grad_norm": 0.7101878523826599, "learning_rate": 2.5944e-06, "loss": 0.0066, "step": 37029 }, { "epoch": 1.974196300047982, "grad_norm": 3.582691192626953, "learning_rate": 2.5942e-06, "loss": 0.0354, "step": 37030 }, { "epoch": 1.974249613477635, "grad_norm": 1.7525861263275146, "learning_rate": 2.5940000000000005e-06, "loss": 0.0191, "step": 37031 }, { "epoch": 1.974302926907288, "grad_norm": 1.3913575410842896, "learning_rate": 2.5938e-06, "loss": 0.0231, "step": 37032 }, { "epoch": 1.9743562403369408, "grad_norm": 1.0513392686843872, "learning_rate": 2.5936e-06, "loss": 0.0111, "step": 37033 }, { "epoch": 1.9744095537665938, "grad_norm": 3.247382402420044, "learning_rate": 2.5934000000000004e-06, "loss": 0.0305, "step": 37034 }, { "epoch": 1.9744628671962468, "grad_norm": 1.8390120267868042, "learning_rate": 2.5932e-06, "loss": 0.0281, "step": 37035 }, { "epoch": 1.9745161806258995, "grad_norm": 2.44858455657959, "learning_rate": 2.593e-06, "loss": 0.0149, "step": 37036 }, { "epoch": 1.9745694940555527, "grad_norm": 0.40889623761177063, "learning_rate": 2.5928000000000004e-06, "loss": 0.0033, "step": 37037 }, { "epoch": 1.9746228074852055, "grad_norm": 1.4387919902801514, "learning_rate": 2.5926e-06, "loss": 0.024, "step": 37038 }, { "epoch": 1.9746761209148584, "grad_norm": 0.24906012415885925, "learning_rate": 2.5924000000000004e-06, "loss": 0.0029, "step": 37039 }, { "epoch": 1.9747294343445114, "grad_norm": 0.5169524550437927, "learning_rate": 2.5922000000000004e-06, "loss": 0.0036, "step": 37040 }, { "epoch": 1.9747827477741642, "grad_norm": 1.4586610794067383, "learning_rate": 2.592e-06, "loss": 0.0099, "step": 37041 }, { "epoch": 1.9748360612038174, "grad_norm": 0.6910485625267029, "learning_rate": 2.5918000000000003e-06, "loss": 0.0101, "step": 37042 }, { "epoch": 1.9748893746334701, "grad_norm": 1.0177128314971924, "learning_rate": 2.5916000000000003e-06, "loss": 0.0179, "step": 37043 }, { "epoch": 1.9749426880631231, "grad_norm": 4.017190933227539, "learning_rate": 2.5914e-06, "loss": 0.0183, "step": 37044 }, { "epoch": 1.974996001492776, "grad_norm": 2.4330368041992188, "learning_rate": 2.5912000000000003e-06, "loss": 0.0409, "step": 37045 }, { "epoch": 1.9750493149224289, "grad_norm": 1.382900595664978, "learning_rate": 2.5910000000000003e-06, "loss": 0.0071, "step": 37046 }, { "epoch": 1.9751026283520818, "grad_norm": 1.680471420288086, "learning_rate": 2.5908e-06, "loss": 0.0194, "step": 37047 }, { "epoch": 1.9751559417817348, "grad_norm": 2.2798240184783936, "learning_rate": 2.5906000000000002e-06, "loss": 0.0459, "step": 37048 }, { "epoch": 1.9752092552113878, "grad_norm": 1.8658652305603027, "learning_rate": 2.5904000000000002e-06, "loss": 0.0242, "step": 37049 }, { "epoch": 1.9752625686410408, "grad_norm": 2.019521713256836, "learning_rate": 2.5901999999999998e-06, "loss": 0.0191, "step": 37050 }, { "epoch": 1.9753158820706935, "grad_norm": 0.5391836166381836, "learning_rate": 2.59e-06, "loss": 0.0044, "step": 37051 }, { "epoch": 1.9753691955003465, "grad_norm": 1.9864236116409302, "learning_rate": 2.5898e-06, "loss": 0.0189, "step": 37052 }, { "epoch": 1.9754225089299995, "grad_norm": 1.8933217525482178, "learning_rate": 2.5896000000000006e-06, "loss": 0.0141, "step": 37053 }, { "epoch": 1.9754758223596522, "grad_norm": 1.307059645652771, "learning_rate": 2.5894e-06, "loss": 0.011, "step": 37054 }, { "epoch": 1.9755291357893054, "grad_norm": 1.343375325202942, "learning_rate": 2.5892e-06, "loss": 0.0168, "step": 37055 }, { "epoch": 1.9755824492189582, "grad_norm": 2.5497028827667236, "learning_rate": 2.5890000000000005e-06, "loss": 0.0318, "step": 37056 }, { "epoch": 1.9756357626486112, "grad_norm": 0.9836516976356506, "learning_rate": 2.5888e-06, "loss": 0.0063, "step": 37057 }, { "epoch": 1.9756890760782642, "grad_norm": 3.0400147438049316, "learning_rate": 2.5886e-06, "loss": 0.0151, "step": 37058 }, { "epoch": 1.975742389507917, "grad_norm": 1.4315515756607056, "learning_rate": 2.5884000000000005e-06, "loss": 0.0102, "step": 37059 }, { "epoch": 1.9757957029375701, "grad_norm": 1.7809125185012817, "learning_rate": 2.5882e-06, "loss": 0.0361, "step": 37060 }, { "epoch": 1.9758490163672229, "grad_norm": 2.4982354640960693, "learning_rate": 2.588e-06, "loss": 0.0129, "step": 37061 }, { "epoch": 1.9759023297968759, "grad_norm": 1.0212557315826416, "learning_rate": 2.5878000000000004e-06, "loss": 0.0188, "step": 37062 }, { "epoch": 1.9759556432265288, "grad_norm": 3.3077731132507324, "learning_rate": 2.5876e-06, "loss": 0.0475, "step": 37063 }, { "epoch": 1.9760089566561816, "grad_norm": 3.3876311779022217, "learning_rate": 2.5874000000000004e-06, "loss": 0.0295, "step": 37064 }, { "epoch": 1.9760622700858346, "grad_norm": 1.532293438911438, "learning_rate": 2.5872000000000004e-06, "loss": 0.0224, "step": 37065 }, { "epoch": 1.9761155835154876, "grad_norm": 1.51225745677948, "learning_rate": 2.587e-06, "loss": 0.0239, "step": 37066 }, { "epoch": 1.9761688969451405, "grad_norm": 0.9631864428520203, "learning_rate": 2.5868000000000004e-06, "loss": 0.0056, "step": 37067 }, { "epoch": 1.9762222103747935, "grad_norm": 1.2743149995803833, "learning_rate": 2.5866000000000004e-06, "loss": 0.0141, "step": 37068 }, { "epoch": 1.9762755238044463, "grad_norm": 2.428068161010742, "learning_rate": 2.5864e-06, "loss": 0.0223, "step": 37069 }, { "epoch": 1.9763288372340992, "grad_norm": 1.5346951484680176, "learning_rate": 2.5862000000000003e-06, "loss": 0.0159, "step": 37070 }, { "epoch": 1.9763821506637522, "grad_norm": 0.8068009614944458, "learning_rate": 2.5860000000000003e-06, "loss": 0.0131, "step": 37071 }, { "epoch": 1.976435464093405, "grad_norm": 1.37323796749115, "learning_rate": 2.5858e-06, "loss": 0.0276, "step": 37072 }, { "epoch": 1.9764887775230582, "grad_norm": 1.782901644706726, "learning_rate": 2.5856000000000003e-06, "loss": 0.0195, "step": 37073 }, { "epoch": 1.976542090952711, "grad_norm": 1.3154778480529785, "learning_rate": 2.5854000000000003e-06, "loss": 0.0173, "step": 37074 }, { "epoch": 1.976595404382364, "grad_norm": 2.076531171798706, "learning_rate": 2.5852000000000007e-06, "loss": 0.0619, "step": 37075 }, { "epoch": 1.976648717812017, "grad_norm": 1.2576738595962524, "learning_rate": 2.5850000000000002e-06, "loss": 0.0104, "step": 37076 }, { "epoch": 1.9767020312416697, "grad_norm": 1.2041871547698975, "learning_rate": 2.5848000000000002e-06, "loss": 0.0071, "step": 37077 }, { "epoch": 1.9767553446713229, "grad_norm": 2.1288583278656006, "learning_rate": 2.5846000000000006e-06, "loss": 0.0317, "step": 37078 }, { "epoch": 1.9768086581009756, "grad_norm": 1.0697962045669556, "learning_rate": 2.5844e-06, "loss": 0.0176, "step": 37079 }, { "epoch": 1.9768619715306286, "grad_norm": 4.222796440124512, "learning_rate": 2.5842e-06, "loss": 0.0455, "step": 37080 }, { "epoch": 1.9769152849602816, "grad_norm": 1.5254102945327759, "learning_rate": 2.5840000000000006e-06, "loss": 0.0268, "step": 37081 }, { "epoch": 1.9769685983899343, "grad_norm": 4.435204029083252, "learning_rate": 2.5838e-06, "loss": 0.0242, "step": 37082 }, { "epoch": 1.9770219118195873, "grad_norm": 0.8694161176681519, "learning_rate": 2.5836e-06, "loss": 0.0046, "step": 37083 }, { "epoch": 1.9770752252492403, "grad_norm": 1.2177214622497559, "learning_rate": 2.5834000000000005e-06, "loss": 0.022, "step": 37084 }, { "epoch": 1.9771285386788933, "grad_norm": 2.1712334156036377, "learning_rate": 2.5832e-06, "loss": 0.0086, "step": 37085 }, { "epoch": 1.9771818521085462, "grad_norm": 1.2521522045135498, "learning_rate": 2.583e-06, "loss": 0.0165, "step": 37086 }, { "epoch": 1.977235165538199, "grad_norm": 1.8124854564666748, "learning_rate": 2.5828000000000005e-06, "loss": 0.0245, "step": 37087 }, { "epoch": 1.977288478967852, "grad_norm": 0.9261752963066101, "learning_rate": 2.5826e-06, "loss": 0.0081, "step": 37088 }, { "epoch": 1.977341792397505, "grad_norm": 2.403783082962036, "learning_rate": 2.5824000000000005e-06, "loss": 0.0215, "step": 37089 }, { "epoch": 1.9773951058271577, "grad_norm": 1.5501986742019653, "learning_rate": 2.5822000000000005e-06, "loss": 0.011, "step": 37090 }, { "epoch": 1.977448419256811, "grad_norm": 3.295741081237793, "learning_rate": 2.582e-06, "loss": 0.0259, "step": 37091 }, { "epoch": 1.9775017326864637, "grad_norm": 1.7120623588562012, "learning_rate": 2.5818000000000004e-06, "loss": 0.0236, "step": 37092 }, { "epoch": 1.9775550461161167, "grad_norm": 0.9059597849845886, "learning_rate": 2.5816000000000004e-06, "loss": 0.0049, "step": 37093 }, { "epoch": 1.9776083595457696, "grad_norm": 2.4239232540130615, "learning_rate": 2.5814e-06, "loss": 0.0345, "step": 37094 }, { "epoch": 1.9776616729754224, "grad_norm": 2.8950111865997314, "learning_rate": 2.5812000000000004e-06, "loss": 0.0269, "step": 37095 }, { "epoch": 1.9777149864050756, "grad_norm": 1.4264087677001953, "learning_rate": 2.5810000000000004e-06, "loss": 0.0278, "step": 37096 }, { "epoch": 1.9777682998347283, "grad_norm": 2.111043930053711, "learning_rate": 2.5808e-06, "loss": 0.0166, "step": 37097 }, { "epoch": 1.9778216132643813, "grad_norm": 2.206780195236206, "learning_rate": 2.5806000000000003e-06, "loss": 0.0198, "step": 37098 }, { "epoch": 1.9778749266940343, "grad_norm": 0.5871451497077942, "learning_rate": 2.5804000000000003e-06, "loss": 0.0045, "step": 37099 }, { "epoch": 1.977928240123687, "grad_norm": 1.9514094591140747, "learning_rate": 2.5802000000000003e-06, "loss": 0.0233, "step": 37100 }, { "epoch": 1.97798155355334, "grad_norm": 1.88583242893219, "learning_rate": 2.5800000000000003e-06, "loss": 0.0381, "step": 37101 }, { "epoch": 1.978034866982993, "grad_norm": 1.0125335454940796, "learning_rate": 2.5798000000000003e-06, "loss": 0.0057, "step": 37102 }, { "epoch": 1.9780881804126458, "grad_norm": 0.325602650642395, "learning_rate": 2.5796000000000003e-06, "loss": 0.0026, "step": 37103 }, { "epoch": 1.978141493842299, "grad_norm": 1.5118218660354614, "learning_rate": 2.5794000000000002e-06, "loss": 0.0126, "step": 37104 }, { "epoch": 1.9781948072719517, "grad_norm": 0.6783971786499023, "learning_rate": 2.5792000000000002e-06, "loss": 0.0134, "step": 37105 }, { "epoch": 1.9782481207016047, "grad_norm": 1.256469488143921, "learning_rate": 2.579e-06, "loss": 0.0062, "step": 37106 }, { "epoch": 1.9783014341312577, "grad_norm": 1.6391838788986206, "learning_rate": 2.5788e-06, "loss": 0.0282, "step": 37107 }, { "epoch": 1.9783547475609105, "grad_norm": 1.9256445169448853, "learning_rate": 2.5786e-06, "loss": 0.035, "step": 37108 }, { "epoch": 1.9784080609905637, "grad_norm": 1.3839350938796997, "learning_rate": 2.5784e-06, "loss": 0.01, "step": 37109 }, { "epoch": 1.9784613744202164, "grad_norm": 1.4837101697921753, "learning_rate": 2.5782e-06, "loss": 0.0076, "step": 37110 }, { "epoch": 1.9785146878498694, "grad_norm": 1.5413814783096313, "learning_rate": 2.578e-06, "loss": 0.0059, "step": 37111 }, { "epoch": 1.9785680012795224, "grad_norm": 2.2440083026885986, "learning_rate": 2.5778e-06, "loss": 0.0312, "step": 37112 }, { "epoch": 1.9786213147091751, "grad_norm": 2.446007013320923, "learning_rate": 2.5776e-06, "loss": 0.0335, "step": 37113 }, { "epoch": 1.9786746281388283, "grad_norm": 1.7384408712387085, "learning_rate": 2.5774000000000005e-06, "loss": 0.0235, "step": 37114 }, { "epoch": 1.978727941568481, "grad_norm": 1.3882888555526733, "learning_rate": 2.5772e-06, "loss": 0.0262, "step": 37115 }, { "epoch": 1.978781254998134, "grad_norm": 2.0077741146087646, "learning_rate": 2.577e-06, "loss": 0.0199, "step": 37116 }, { "epoch": 1.978834568427787, "grad_norm": 2.2512989044189453, "learning_rate": 2.5768000000000005e-06, "loss": 0.0339, "step": 37117 }, { "epoch": 1.9788878818574398, "grad_norm": 2.012453079223633, "learning_rate": 2.5766e-06, "loss": 0.0195, "step": 37118 }, { "epoch": 1.9789411952870928, "grad_norm": 2.8775970935821533, "learning_rate": 2.5764e-06, "loss": 0.0112, "step": 37119 }, { "epoch": 1.9789945087167458, "grad_norm": 1.7937896251678467, "learning_rate": 2.5762000000000004e-06, "loss": 0.0067, "step": 37120 }, { "epoch": 1.9790478221463985, "grad_norm": 1.5272226333618164, "learning_rate": 2.576e-06, "loss": 0.0191, "step": 37121 }, { "epoch": 1.9791011355760517, "grad_norm": 1.843294620513916, "learning_rate": 2.5758e-06, "loss": 0.0211, "step": 37122 }, { "epoch": 1.9791544490057045, "grad_norm": 1.803432583808899, "learning_rate": 2.5756000000000004e-06, "loss": 0.0093, "step": 37123 }, { "epoch": 1.9792077624353575, "grad_norm": 1.286765694618225, "learning_rate": 2.5754e-06, "loss": 0.0071, "step": 37124 }, { "epoch": 1.9792610758650104, "grad_norm": 3.4387426376342773, "learning_rate": 2.5752000000000003e-06, "loss": 0.0269, "step": 37125 }, { "epoch": 1.9793143892946632, "grad_norm": 1.0128793716430664, "learning_rate": 2.5750000000000003e-06, "loss": 0.0057, "step": 37126 }, { "epoch": 1.9793677027243164, "grad_norm": 1.3658051490783691, "learning_rate": 2.5748e-06, "loss": 0.0111, "step": 37127 }, { "epoch": 1.9794210161539691, "grad_norm": 2.927485227584839, "learning_rate": 2.5746000000000003e-06, "loss": 0.0287, "step": 37128 }, { "epoch": 1.9794743295836221, "grad_norm": 1.2609410285949707, "learning_rate": 2.5744000000000003e-06, "loss": 0.0246, "step": 37129 }, { "epoch": 1.979527643013275, "grad_norm": 1.6615140438079834, "learning_rate": 2.5742e-06, "loss": 0.0128, "step": 37130 }, { "epoch": 1.9795809564429279, "grad_norm": 1.6335642337799072, "learning_rate": 2.5740000000000003e-06, "loss": 0.0417, "step": 37131 }, { "epoch": 1.979634269872581, "grad_norm": 1.4363552331924438, "learning_rate": 2.5738000000000002e-06, "loss": 0.0165, "step": 37132 }, { "epoch": 1.9796875833022338, "grad_norm": 1.6108431816101074, "learning_rate": 2.5736e-06, "loss": 0.0179, "step": 37133 }, { "epoch": 1.9797408967318868, "grad_norm": 1.5136462450027466, "learning_rate": 2.5734000000000002e-06, "loss": 0.0062, "step": 37134 }, { "epoch": 1.9797942101615398, "grad_norm": 1.655332088470459, "learning_rate": 2.5732e-06, "loss": 0.009, "step": 37135 }, { "epoch": 1.9798475235911925, "grad_norm": 1.5495141744613647, "learning_rate": 2.5729999999999998e-06, "loss": 0.0077, "step": 37136 }, { "epoch": 1.9799008370208455, "grad_norm": 1.8684362173080444, "learning_rate": 2.5728e-06, "loss": 0.0295, "step": 37137 }, { "epoch": 1.9799541504504985, "grad_norm": 2.0628511905670166, "learning_rate": 2.5726e-06, "loss": 0.02, "step": 37138 }, { "epoch": 1.9800074638801513, "grad_norm": 6.579087734222412, "learning_rate": 2.5724000000000006e-06, "loss": 0.0144, "step": 37139 }, { "epoch": 1.9800607773098045, "grad_norm": 1.3871331214904785, "learning_rate": 2.5722e-06, "loss": 0.0081, "step": 37140 }, { "epoch": 1.9801140907394572, "grad_norm": 1.556943655014038, "learning_rate": 2.572e-06, "loss": 0.0216, "step": 37141 }, { "epoch": 1.9801674041691102, "grad_norm": 0.8247766494750977, "learning_rate": 2.5718000000000005e-06, "loss": 0.0061, "step": 37142 }, { "epoch": 1.9802207175987632, "grad_norm": 3.2616233825683594, "learning_rate": 2.5716e-06, "loss": 0.0183, "step": 37143 }, { "epoch": 1.980274031028416, "grad_norm": 0.7388932108879089, "learning_rate": 2.5714e-06, "loss": 0.0027, "step": 37144 }, { "epoch": 1.9803273444580691, "grad_norm": 1.513895869255066, "learning_rate": 2.5712000000000005e-06, "loss": 0.0205, "step": 37145 }, { "epoch": 1.9803806578877219, "grad_norm": 1.6959260702133179, "learning_rate": 2.571e-06, "loss": 0.0096, "step": 37146 }, { "epoch": 1.9804339713173749, "grad_norm": 3.116542339324951, "learning_rate": 2.5708e-06, "loss": 0.0455, "step": 37147 }, { "epoch": 1.9804872847470278, "grad_norm": 0.9572179317474365, "learning_rate": 2.5706000000000004e-06, "loss": 0.0265, "step": 37148 }, { "epoch": 1.9805405981766806, "grad_norm": 1.1162092685699463, "learning_rate": 2.5704e-06, "loss": 0.0068, "step": 37149 }, { "epoch": 1.9805939116063338, "grad_norm": 0.838312566280365, "learning_rate": 2.5702000000000004e-06, "loss": 0.0113, "step": 37150 }, { "epoch": 1.9806472250359866, "grad_norm": 1.278722882270813, "learning_rate": 2.5700000000000004e-06, "loss": 0.0067, "step": 37151 }, { "epoch": 1.9807005384656395, "grad_norm": 2.3015403747558594, "learning_rate": 2.5698e-06, "loss": 0.0178, "step": 37152 }, { "epoch": 1.9807538518952925, "grad_norm": 1.8348671197891235, "learning_rate": 2.5696000000000004e-06, "loss": 0.0086, "step": 37153 }, { "epoch": 1.9808071653249453, "grad_norm": 1.7922831773757935, "learning_rate": 2.5694000000000003e-06, "loss": 0.0231, "step": 37154 }, { "epoch": 1.9808604787545983, "grad_norm": 1.2051032781600952, "learning_rate": 2.5692e-06, "loss": 0.0068, "step": 37155 }, { "epoch": 1.9809137921842512, "grad_norm": 0.44236308336257935, "learning_rate": 2.5690000000000003e-06, "loss": 0.0024, "step": 37156 }, { "epoch": 1.980967105613904, "grad_norm": 0.8066707849502563, "learning_rate": 2.5688000000000003e-06, "loss": 0.0147, "step": 37157 }, { "epoch": 1.9810204190435572, "grad_norm": 2.067516803741455, "learning_rate": 2.5686e-06, "loss": 0.0102, "step": 37158 }, { "epoch": 1.98107373247321, "grad_norm": 2.8972837924957275, "learning_rate": 2.5684000000000003e-06, "loss": 0.0078, "step": 37159 }, { "epoch": 1.981127045902863, "grad_norm": 0.6127104759216309, "learning_rate": 2.5682000000000002e-06, "loss": 0.0101, "step": 37160 }, { "epoch": 1.981180359332516, "grad_norm": 1.795594573020935, "learning_rate": 2.568e-06, "loss": 0.0194, "step": 37161 }, { "epoch": 1.9812336727621687, "grad_norm": 4.431382656097412, "learning_rate": 2.5678000000000002e-06, "loss": 0.0335, "step": 37162 }, { "epoch": 1.9812869861918219, "grad_norm": 2.33634877204895, "learning_rate": 2.5676e-06, "loss": 0.0532, "step": 37163 }, { "epoch": 1.9813402996214746, "grad_norm": 2.4503448009490967, "learning_rate": 2.5674000000000006e-06, "loss": 0.0183, "step": 37164 }, { "epoch": 1.9813936130511276, "grad_norm": 1.3863710165023804, "learning_rate": 2.5672e-06, "loss": 0.0193, "step": 37165 }, { "epoch": 1.9814469264807806, "grad_norm": 0.6703457832336426, "learning_rate": 2.567e-06, "loss": 0.0043, "step": 37166 }, { "epoch": 1.9815002399104333, "grad_norm": 1.208238124847412, "learning_rate": 2.5668000000000006e-06, "loss": 0.0058, "step": 37167 }, { "epoch": 1.9815535533400863, "grad_norm": 2.0918383598327637, "learning_rate": 2.5666e-06, "loss": 0.0099, "step": 37168 }, { "epoch": 1.9816068667697393, "grad_norm": 3.995922088623047, "learning_rate": 2.5664e-06, "loss": 0.0376, "step": 37169 }, { "epoch": 1.9816601801993923, "grad_norm": 2.3657989501953125, "learning_rate": 2.5662000000000005e-06, "loss": 0.0116, "step": 37170 }, { "epoch": 1.9817134936290453, "grad_norm": 0.9876607060432434, "learning_rate": 2.566e-06, "loss": 0.0072, "step": 37171 }, { "epoch": 1.981766807058698, "grad_norm": 2.571420669555664, "learning_rate": 2.5658e-06, "loss": 0.0156, "step": 37172 }, { "epoch": 1.981820120488351, "grad_norm": 1.3871666193008423, "learning_rate": 2.5656000000000005e-06, "loss": 0.0099, "step": 37173 }, { "epoch": 1.981873433918004, "grad_norm": 1.3146870136260986, "learning_rate": 2.5654e-06, "loss": 0.0292, "step": 37174 }, { "epoch": 1.9819267473476567, "grad_norm": 3.0390100479125977, "learning_rate": 2.5652000000000004e-06, "loss": 0.0156, "step": 37175 }, { "epoch": 1.98198006077731, "grad_norm": 2.580918788909912, "learning_rate": 2.5650000000000004e-06, "loss": 0.0222, "step": 37176 }, { "epoch": 1.9820333742069627, "grad_norm": 2.040088176727295, "learning_rate": 2.5648e-06, "loss": 0.0481, "step": 37177 }, { "epoch": 1.9820866876366157, "grad_norm": 7.3647050857543945, "learning_rate": 2.5646000000000004e-06, "loss": 0.0341, "step": 37178 }, { "epoch": 1.9821400010662686, "grad_norm": 0.7419022917747498, "learning_rate": 2.5644000000000004e-06, "loss": 0.0045, "step": 37179 }, { "epoch": 1.9821933144959214, "grad_norm": 0.43681809306144714, "learning_rate": 2.5642e-06, "loss": 0.0027, "step": 37180 }, { "epoch": 1.9822466279255746, "grad_norm": 3.4497835636138916, "learning_rate": 2.5640000000000004e-06, "loss": 0.0156, "step": 37181 }, { "epoch": 1.9822999413552274, "grad_norm": 2.4535787105560303, "learning_rate": 2.5638000000000003e-06, "loss": 0.0115, "step": 37182 }, { "epoch": 1.9823532547848803, "grad_norm": 1.9947324991226196, "learning_rate": 2.5636e-06, "loss": 0.0218, "step": 37183 }, { "epoch": 1.9824065682145333, "grad_norm": 1.6029657125473022, "learning_rate": 2.5634000000000003e-06, "loss": 0.007, "step": 37184 }, { "epoch": 1.982459881644186, "grad_norm": 1.4147629737854004, "learning_rate": 2.5632000000000003e-06, "loss": 0.0297, "step": 37185 }, { "epoch": 1.982513195073839, "grad_norm": 0.8205862045288086, "learning_rate": 2.563e-06, "loss": 0.0061, "step": 37186 }, { "epoch": 1.982566508503492, "grad_norm": 1.3724720478057861, "learning_rate": 2.5628000000000003e-06, "loss": 0.0109, "step": 37187 }, { "epoch": 1.982619821933145, "grad_norm": 1.576857089996338, "learning_rate": 2.5626000000000002e-06, "loss": 0.0133, "step": 37188 }, { "epoch": 1.982673135362798, "grad_norm": 1.9257768392562866, "learning_rate": 2.5624000000000002e-06, "loss": 0.0213, "step": 37189 }, { "epoch": 1.9827264487924507, "grad_norm": 3.043442487716675, "learning_rate": 2.5622000000000002e-06, "loss": 0.017, "step": 37190 }, { "epoch": 1.9827797622221037, "grad_norm": 2.042667865753174, "learning_rate": 2.562e-06, "loss": 0.0146, "step": 37191 }, { "epoch": 1.9828330756517567, "grad_norm": 0.8726944923400879, "learning_rate": 2.5618e-06, "loss": 0.0041, "step": 37192 }, { "epoch": 1.9828863890814095, "grad_norm": 1.088061809539795, "learning_rate": 2.5616e-06, "loss": 0.0069, "step": 37193 }, { "epoch": 1.9829397025110627, "grad_norm": 0.8675934672355652, "learning_rate": 2.5614e-06, "loss": 0.0045, "step": 37194 }, { "epoch": 1.9829930159407154, "grad_norm": 1.8068757057189941, "learning_rate": 2.5612e-06, "loss": 0.0278, "step": 37195 }, { "epoch": 1.9830463293703684, "grad_norm": 1.1893537044525146, "learning_rate": 2.561e-06, "loss": 0.0035, "step": 37196 }, { "epoch": 1.9830996428000214, "grad_norm": 1.6054742336273193, "learning_rate": 2.5608e-06, "loss": 0.0068, "step": 37197 }, { "epoch": 1.9831529562296741, "grad_norm": 2.5294029712677, "learning_rate": 2.5606e-06, "loss": 0.0274, "step": 37198 }, { "epoch": 1.9832062696593273, "grad_norm": 1.8347793817520142, "learning_rate": 2.5604e-06, "loss": 0.0185, "step": 37199 }, { "epoch": 1.98325958308898, "grad_norm": 2.6406967639923096, "learning_rate": 2.5602000000000005e-06, "loss": 0.0462, "step": 37200 }, { "epoch": 1.983312896518633, "grad_norm": 1.2699145078659058, "learning_rate": 2.56e-06, "loss": 0.0094, "step": 37201 }, { "epoch": 1.983366209948286, "grad_norm": 2.742243766784668, "learning_rate": 2.5598e-06, "loss": 0.0166, "step": 37202 }, { "epoch": 1.9834195233779388, "grad_norm": 0.986099362373352, "learning_rate": 2.5596000000000004e-06, "loss": 0.0063, "step": 37203 }, { "epoch": 1.9834728368075918, "grad_norm": 2.177048683166504, "learning_rate": 2.5594e-06, "loss": 0.0197, "step": 37204 }, { "epoch": 1.9835261502372448, "grad_norm": 1.6123063564300537, "learning_rate": 2.5592e-06, "loss": 0.0101, "step": 37205 }, { "epoch": 1.9835794636668977, "grad_norm": 1.7874386310577393, "learning_rate": 2.5590000000000004e-06, "loss": 0.0114, "step": 37206 }, { "epoch": 1.9836327770965507, "grad_norm": 0.5933977365493774, "learning_rate": 2.5588e-06, "loss": 0.004, "step": 37207 }, { "epoch": 1.9836860905262035, "grad_norm": 0.8415785431861877, "learning_rate": 2.5586e-06, "loss": 0.0037, "step": 37208 }, { "epoch": 1.9837394039558565, "grad_norm": 1.546421766281128, "learning_rate": 2.5584000000000004e-06, "loss": 0.0147, "step": 37209 }, { "epoch": 1.9837927173855094, "grad_norm": 2.1754276752471924, "learning_rate": 2.5582e-06, "loss": 0.0279, "step": 37210 }, { "epoch": 1.9838460308151622, "grad_norm": 2.1388821601867676, "learning_rate": 2.5580000000000003e-06, "loss": 0.0223, "step": 37211 }, { "epoch": 1.9838993442448154, "grad_norm": 1.5045394897460938, "learning_rate": 2.5578000000000003e-06, "loss": 0.0105, "step": 37212 }, { "epoch": 1.9839526576744682, "grad_norm": 1.7177345752716064, "learning_rate": 2.5576e-06, "loss": 0.006, "step": 37213 }, { "epoch": 1.9840059711041211, "grad_norm": 2.8284554481506348, "learning_rate": 2.5574000000000003e-06, "loss": 0.0236, "step": 37214 }, { "epoch": 1.9840592845337741, "grad_norm": 1.7930303812026978, "learning_rate": 2.5572000000000003e-06, "loss": 0.0146, "step": 37215 }, { "epoch": 1.9841125979634269, "grad_norm": 1.7668014764785767, "learning_rate": 2.557e-06, "loss": 0.0109, "step": 37216 }, { "epoch": 1.98416591139308, "grad_norm": 2.2582130432128906, "learning_rate": 2.5568000000000002e-06, "loss": 0.0138, "step": 37217 }, { "epoch": 1.9842192248227328, "grad_norm": 2.390321731567383, "learning_rate": 2.5566000000000002e-06, "loss": 0.0225, "step": 37218 }, { "epoch": 1.9842725382523858, "grad_norm": 3.374436855316162, "learning_rate": 2.5563999999999998e-06, "loss": 0.0282, "step": 37219 }, { "epoch": 1.9843258516820388, "grad_norm": 1.9034825563430786, "learning_rate": 2.5562e-06, "loss": 0.0311, "step": 37220 }, { "epoch": 1.9843791651116915, "grad_norm": 1.2719969749450684, "learning_rate": 2.556e-06, "loss": 0.0117, "step": 37221 }, { "epoch": 1.9844324785413445, "grad_norm": 2.0114965438842773, "learning_rate": 2.5557999999999997e-06, "loss": 0.015, "step": 37222 }, { "epoch": 1.9844857919709975, "grad_norm": 1.1900497674942017, "learning_rate": 2.5556e-06, "loss": 0.0091, "step": 37223 }, { "epoch": 1.9845391054006503, "grad_norm": 2.5549120903015137, "learning_rate": 2.5554e-06, "loss": 0.0408, "step": 37224 }, { "epoch": 1.9845924188303035, "grad_norm": 2.4841394424438477, "learning_rate": 2.5552000000000005e-06, "loss": 0.013, "step": 37225 }, { "epoch": 1.9846457322599562, "grad_norm": 1.0206884145736694, "learning_rate": 2.555e-06, "loss": 0.0167, "step": 37226 }, { "epoch": 1.9846990456896092, "grad_norm": 3.369859457015991, "learning_rate": 2.5548e-06, "loss": 0.0174, "step": 37227 }, { "epoch": 1.9847523591192622, "grad_norm": 1.0367460250854492, "learning_rate": 2.5546000000000005e-06, "loss": 0.0088, "step": 37228 }, { "epoch": 1.984805672548915, "grad_norm": 2.556201696395874, "learning_rate": 2.5544e-06, "loss": 0.0189, "step": 37229 }, { "epoch": 1.9848589859785681, "grad_norm": 0.3352048993110657, "learning_rate": 2.5542e-06, "loss": 0.003, "step": 37230 }, { "epoch": 1.984912299408221, "grad_norm": 2.1643669605255127, "learning_rate": 2.5540000000000004e-06, "loss": 0.0185, "step": 37231 }, { "epoch": 1.9849656128378739, "grad_norm": 0.307888001203537, "learning_rate": 2.5538e-06, "loss": 0.0025, "step": 37232 }, { "epoch": 1.9850189262675269, "grad_norm": 2.6147584915161133, "learning_rate": 2.5536e-06, "loss": 0.0328, "step": 37233 }, { "epoch": 1.9850722396971796, "grad_norm": 1.2062851190567017, "learning_rate": 2.5534000000000004e-06, "loss": 0.0217, "step": 37234 }, { "epoch": 1.9851255531268328, "grad_norm": 1.9405686855316162, "learning_rate": 2.5532e-06, "loss": 0.0243, "step": 37235 }, { "epoch": 1.9851788665564856, "grad_norm": 1.6724023818969727, "learning_rate": 2.5530000000000004e-06, "loss": 0.0125, "step": 37236 }, { "epoch": 1.9852321799861385, "grad_norm": 2.182983636856079, "learning_rate": 2.5528000000000004e-06, "loss": 0.035, "step": 37237 }, { "epoch": 1.9852854934157915, "grad_norm": 1.0555661916732788, "learning_rate": 2.5526e-06, "loss": 0.0058, "step": 37238 }, { "epoch": 1.9853388068454443, "grad_norm": 1.7858525514602661, "learning_rate": 2.5524000000000003e-06, "loss": 0.0183, "step": 37239 }, { "epoch": 1.9853921202750973, "grad_norm": 1.3704867362976074, "learning_rate": 2.5522000000000003e-06, "loss": 0.01, "step": 37240 }, { "epoch": 1.9854454337047502, "grad_norm": 2.053359270095825, "learning_rate": 2.552e-06, "loss": 0.0269, "step": 37241 }, { "epoch": 1.985498747134403, "grad_norm": 2.145664691925049, "learning_rate": 2.5518000000000003e-06, "loss": 0.0179, "step": 37242 }, { "epoch": 1.9855520605640562, "grad_norm": 1.7371402978897095, "learning_rate": 2.5516000000000003e-06, "loss": 0.0081, "step": 37243 }, { "epoch": 1.985605373993709, "grad_norm": 4.158320426940918, "learning_rate": 2.5514e-06, "loss": 0.0304, "step": 37244 }, { "epoch": 1.985658687423362, "grad_norm": 0.20631618797779083, "learning_rate": 2.5512000000000002e-06, "loss": 0.0017, "step": 37245 }, { "epoch": 1.985712000853015, "grad_norm": 0.4954100549221039, "learning_rate": 2.5510000000000002e-06, "loss": 0.0033, "step": 37246 }, { "epoch": 1.9857653142826677, "grad_norm": 2.2250897884368896, "learning_rate": 2.5507999999999998e-06, "loss": 0.027, "step": 37247 }, { "epoch": 1.9858186277123209, "grad_norm": 0.6801096796989441, "learning_rate": 2.5506e-06, "loss": 0.0044, "step": 37248 }, { "epoch": 1.9858719411419736, "grad_norm": 1.006070852279663, "learning_rate": 2.5504e-06, "loss": 0.0088, "step": 37249 }, { "epoch": 1.9859252545716266, "grad_norm": 3.7479658126831055, "learning_rate": 2.5502000000000006e-06, "loss": 0.0262, "step": 37250 }, { "epoch": 1.9859785680012796, "grad_norm": 1.906037449836731, "learning_rate": 2.55e-06, "loss": 0.0441, "step": 37251 }, { "epoch": 1.9860318814309323, "grad_norm": 1.3839603662490845, "learning_rate": 2.5498e-06, "loss": 0.0154, "step": 37252 }, { "epoch": 1.9860851948605855, "grad_norm": 2.2267065048217773, "learning_rate": 2.5496000000000005e-06, "loss": 0.0314, "step": 37253 }, { "epoch": 1.9861385082902383, "grad_norm": 3.0550670623779297, "learning_rate": 2.5494e-06, "loss": 0.0281, "step": 37254 }, { "epoch": 1.9861918217198913, "grad_norm": 1.5338135957717896, "learning_rate": 2.5492e-06, "loss": 0.0073, "step": 37255 }, { "epoch": 1.9862451351495443, "grad_norm": 1.0557715892791748, "learning_rate": 2.5490000000000005e-06, "loss": 0.0135, "step": 37256 }, { "epoch": 1.986298448579197, "grad_norm": 0.4545800983905792, "learning_rate": 2.5488e-06, "loss": 0.0025, "step": 37257 }, { "epoch": 1.98635176200885, "grad_norm": 1.932015299797058, "learning_rate": 2.5486e-06, "loss": 0.0143, "step": 37258 }, { "epoch": 1.986405075438503, "grad_norm": 1.133826494216919, "learning_rate": 2.5484000000000005e-06, "loss": 0.0074, "step": 37259 }, { "epoch": 1.9864583888681557, "grad_norm": 2.036112070083618, "learning_rate": 2.5482e-06, "loss": 0.0274, "step": 37260 }, { "epoch": 1.986511702297809, "grad_norm": 1.8105907440185547, "learning_rate": 2.5480000000000004e-06, "loss": 0.0135, "step": 37261 }, { "epoch": 1.9865650157274617, "grad_norm": 0.46105295419692993, "learning_rate": 2.5478000000000004e-06, "loss": 0.0032, "step": 37262 }, { "epoch": 1.9866183291571147, "grad_norm": 1.8733822107315063, "learning_rate": 2.5476e-06, "loss": 0.0266, "step": 37263 }, { "epoch": 1.9866716425867676, "grad_norm": 1.6697487831115723, "learning_rate": 2.5474000000000004e-06, "loss": 0.0074, "step": 37264 }, { "epoch": 1.9867249560164204, "grad_norm": 1.9701942205429077, "learning_rate": 2.5472000000000004e-06, "loss": 0.0113, "step": 37265 }, { "epoch": 1.9867782694460736, "grad_norm": 2.6570329666137695, "learning_rate": 2.547e-06, "loss": 0.0459, "step": 37266 }, { "epoch": 1.9868315828757264, "grad_norm": 1.5393867492675781, "learning_rate": 2.5468000000000003e-06, "loss": 0.0243, "step": 37267 }, { "epoch": 1.9868848963053793, "grad_norm": 2.163583755493164, "learning_rate": 2.5466000000000003e-06, "loss": 0.0363, "step": 37268 }, { "epoch": 1.9869382097350323, "grad_norm": 2.201828956604004, "learning_rate": 2.5464e-06, "loss": 0.0092, "step": 37269 }, { "epoch": 1.986991523164685, "grad_norm": 0.6494336724281311, "learning_rate": 2.5462000000000003e-06, "loss": 0.0046, "step": 37270 }, { "epoch": 1.987044836594338, "grad_norm": 0.9998847842216492, "learning_rate": 2.5460000000000003e-06, "loss": 0.0066, "step": 37271 }, { "epoch": 1.987098150023991, "grad_norm": 2.835171937942505, "learning_rate": 2.5458e-06, "loss": 0.0194, "step": 37272 }, { "epoch": 1.987151463453644, "grad_norm": 2.10827898979187, "learning_rate": 2.5456000000000002e-06, "loss": 0.0322, "step": 37273 }, { "epoch": 1.987204776883297, "grad_norm": 2.6661479473114014, "learning_rate": 2.5454000000000002e-06, "loss": 0.0189, "step": 37274 }, { "epoch": 1.9872580903129498, "grad_norm": 2.1389238834381104, "learning_rate": 2.5452000000000006e-06, "loss": 0.0131, "step": 37275 }, { "epoch": 1.9873114037426027, "grad_norm": 3.7332468032836914, "learning_rate": 2.545e-06, "loss": 0.0607, "step": 37276 }, { "epoch": 1.9873647171722557, "grad_norm": 1.8827377557754517, "learning_rate": 2.5448e-06, "loss": 0.0276, "step": 37277 }, { "epoch": 1.9874180306019085, "grad_norm": 2.1538844108581543, "learning_rate": 2.5446000000000006e-06, "loss": 0.0173, "step": 37278 }, { "epoch": 1.9874713440315617, "grad_norm": 1.2648375034332275, "learning_rate": 2.5444e-06, "loss": 0.0183, "step": 37279 }, { "epoch": 1.9875246574612144, "grad_norm": 3.0226550102233887, "learning_rate": 2.5442e-06, "loss": 0.0215, "step": 37280 }, { "epoch": 1.9875779708908674, "grad_norm": 2.054297685623169, "learning_rate": 2.5440000000000005e-06, "loss": 0.0106, "step": 37281 }, { "epoch": 1.9876312843205204, "grad_norm": 1.7282487154006958, "learning_rate": 2.5438e-06, "loss": 0.0122, "step": 37282 }, { "epoch": 1.9876845977501731, "grad_norm": 2.508225917816162, "learning_rate": 2.5436e-06, "loss": 0.0192, "step": 37283 }, { "epoch": 1.9877379111798263, "grad_norm": 3.206080913543701, "learning_rate": 2.5434000000000005e-06, "loss": 0.0228, "step": 37284 }, { "epoch": 1.987791224609479, "grad_norm": 6.599134922027588, "learning_rate": 2.5432e-06, "loss": 0.0447, "step": 37285 }, { "epoch": 1.987844538039132, "grad_norm": 3.1444056034088135, "learning_rate": 2.5430000000000005e-06, "loss": 0.0467, "step": 37286 }, { "epoch": 1.987897851468785, "grad_norm": 1.672835350036621, "learning_rate": 2.5428000000000005e-06, "loss": 0.0198, "step": 37287 }, { "epoch": 1.9879511648984378, "grad_norm": 5.1511101722717285, "learning_rate": 2.5426e-06, "loss": 0.028, "step": 37288 }, { "epoch": 1.9880044783280908, "grad_norm": 1.406882405281067, "learning_rate": 2.5424000000000004e-06, "loss": 0.0069, "step": 37289 }, { "epoch": 1.9880577917577438, "grad_norm": 1.9002774953842163, "learning_rate": 2.5422000000000004e-06, "loss": 0.0139, "step": 37290 }, { "epoch": 1.9881111051873968, "grad_norm": 1.7321815490722656, "learning_rate": 2.542e-06, "loss": 0.0171, "step": 37291 }, { "epoch": 1.9881644186170497, "grad_norm": 1.2788972854614258, "learning_rate": 2.5418000000000004e-06, "loss": 0.0195, "step": 37292 }, { "epoch": 1.9882177320467025, "grad_norm": 1.2674002647399902, "learning_rate": 2.5416000000000004e-06, "loss": 0.0189, "step": 37293 }, { "epoch": 1.9882710454763555, "grad_norm": 1.907925009727478, "learning_rate": 2.5414e-06, "loss": 0.033, "step": 37294 }, { "epoch": 1.9883243589060084, "grad_norm": 1.5459387302398682, "learning_rate": 2.5412000000000003e-06, "loss": 0.0145, "step": 37295 }, { "epoch": 1.9883776723356612, "grad_norm": 1.3168988227844238, "learning_rate": 2.5410000000000003e-06, "loss": 0.0065, "step": 37296 }, { "epoch": 1.9884309857653144, "grad_norm": 1.0178866386413574, "learning_rate": 2.5408e-06, "loss": 0.0077, "step": 37297 }, { "epoch": 1.9884842991949672, "grad_norm": 1.8825231790542603, "learning_rate": 2.5406000000000003e-06, "loss": 0.0259, "step": 37298 }, { "epoch": 1.9885376126246201, "grad_norm": 2.02546763420105, "learning_rate": 2.5404000000000003e-06, "loss": 0.0134, "step": 37299 }, { "epoch": 1.9885909260542731, "grad_norm": 0.9105144739151001, "learning_rate": 2.5402000000000003e-06, "loss": 0.0134, "step": 37300 }, { "epoch": 1.9886442394839259, "grad_norm": 2.082659959793091, "learning_rate": 2.5400000000000002e-06, "loss": 0.0324, "step": 37301 }, { "epoch": 1.988697552913579, "grad_norm": 0.6831741333007812, "learning_rate": 2.5398000000000002e-06, "loss": 0.0056, "step": 37302 }, { "epoch": 1.9887508663432318, "grad_norm": 0.8261899352073669, "learning_rate": 2.5396000000000002e-06, "loss": 0.005, "step": 37303 }, { "epoch": 1.9888041797728848, "grad_norm": 1.208900809288025, "learning_rate": 2.5394e-06, "loss": 0.0264, "step": 37304 }, { "epoch": 1.9888574932025378, "grad_norm": 1.7882658243179321, "learning_rate": 2.5392e-06, "loss": 0.02, "step": 37305 }, { "epoch": 1.9889108066321906, "grad_norm": 2.054818868637085, "learning_rate": 2.539e-06, "loss": 0.02, "step": 37306 }, { "epoch": 1.9889641200618435, "grad_norm": 1.5656229257583618, "learning_rate": 2.5388e-06, "loss": 0.011, "step": 37307 }, { "epoch": 1.9890174334914965, "grad_norm": 1.1755188703536987, "learning_rate": 2.5386e-06, "loss": 0.0138, "step": 37308 }, { "epoch": 1.9890707469211495, "grad_norm": 3.148270845413208, "learning_rate": 2.5384e-06, "loss": 0.0244, "step": 37309 }, { "epoch": 1.9891240603508025, "grad_norm": 0.3446989059448242, "learning_rate": 2.5382e-06, "loss": 0.0021, "step": 37310 }, { "epoch": 1.9891773737804552, "grad_norm": 2.5921268463134766, "learning_rate": 2.5380000000000005e-06, "loss": 0.0116, "step": 37311 }, { "epoch": 1.9892306872101082, "grad_norm": 2.7811334133148193, "learning_rate": 2.5378e-06, "loss": 0.0491, "step": 37312 }, { "epoch": 1.9892840006397612, "grad_norm": 1.1360065937042236, "learning_rate": 2.5376e-06, "loss": 0.0105, "step": 37313 }, { "epoch": 1.989337314069414, "grad_norm": 2.4514572620391846, "learning_rate": 2.5374000000000005e-06, "loss": 0.0094, "step": 37314 }, { "epoch": 1.9893906274990671, "grad_norm": 1.55598783493042, "learning_rate": 2.5372e-06, "loss": 0.0115, "step": 37315 }, { "epoch": 1.98944394092872, "grad_norm": 2.501962900161743, "learning_rate": 2.537e-06, "loss": 0.0245, "step": 37316 }, { "epoch": 1.9894972543583729, "grad_norm": 2.646963596343994, "learning_rate": 2.5368000000000004e-06, "loss": 0.0112, "step": 37317 }, { "epoch": 1.9895505677880259, "grad_norm": 2.397496223449707, "learning_rate": 2.5366e-06, "loss": 0.0285, "step": 37318 }, { "epoch": 1.9896038812176786, "grad_norm": 1.2927190065383911, "learning_rate": 2.5364e-06, "loss": 0.0136, "step": 37319 }, { "epoch": 1.9896571946473318, "grad_norm": 2.9973526000976562, "learning_rate": 2.5362000000000004e-06, "loss": 0.0319, "step": 37320 }, { "epoch": 1.9897105080769846, "grad_norm": 1.6292364597320557, "learning_rate": 2.536e-06, "loss": 0.0074, "step": 37321 }, { "epoch": 1.9897638215066376, "grad_norm": 0.8734554648399353, "learning_rate": 2.5358000000000004e-06, "loss": 0.0051, "step": 37322 }, { "epoch": 1.9898171349362905, "grad_norm": 1.568823218345642, "learning_rate": 2.5356000000000003e-06, "loss": 0.0246, "step": 37323 }, { "epoch": 1.9898704483659433, "grad_norm": 0.8164288401603699, "learning_rate": 2.5354e-06, "loss": 0.0066, "step": 37324 }, { "epoch": 1.9899237617955963, "grad_norm": 1.7059706449508667, "learning_rate": 2.5352000000000003e-06, "loss": 0.0252, "step": 37325 }, { "epoch": 1.9899770752252492, "grad_norm": 2.3544466495513916, "learning_rate": 2.5350000000000003e-06, "loss": 0.0099, "step": 37326 }, { "epoch": 1.990030388654902, "grad_norm": 1.5123566389083862, "learning_rate": 2.5348e-06, "loss": 0.0141, "step": 37327 }, { "epoch": 1.9900837020845552, "grad_norm": 2.4861252307891846, "learning_rate": 2.5346000000000003e-06, "loss": 0.0165, "step": 37328 }, { "epoch": 1.990137015514208, "grad_norm": 1.0995113849639893, "learning_rate": 2.5344000000000002e-06, "loss": 0.0236, "step": 37329 }, { "epoch": 1.990190328943861, "grad_norm": 1.2406045198440552, "learning_rate": 2.5342e-06, "loss": 0.0156, "step": 37330 }, { "epoch": 1.990243642373514, "grad_norm": 1.6168909072875977, "learning_rate": 2.5340000000000002e-06, "loss": 0.0187, "step": 37331 }, { "epoch": 1.9902969558031667, "grad_norm": 0.8813455700874329, "learning_rate": 2.5338e-06, "loss": 0.0051, "step": 37332 }, { "epoch": 1.9903502692328199, "grad_norm": 1.52980375289917, "learning_rate": 2.5335999999999998e-06, "loss": 0.0077, "step": 37333 }, { "epoch": 1.9904035826624726, "grad_norm": 1.26847505569458, "learning_rate": 2.5334e-06, "loss": 0.0136, "step": 37334 }, { "epoch": 1.9904568960921256, "grad_norm": 1.8753036260604858, "learning_rate": 2.5332e-06, "loss": 0.0387, "step": 37335 }, { "epoch": 1.9905102095217786, "grad_norm": 6.155189514160156, "learning_rate": 2.5330000000000006e-06, "loss": 0.0491, "step": 37336 }, { "epoch": 1.9905635229514314, "grad_norm": 2.303633213043213, "learning_rate": 2.5328e-06, "loss": 0.0138, "step": 37337 }, { "epoch": 1.9906168363810846, "grad_norm": 1.1616674661636353, "learning_rate": 2.5326e-06, "loss": 0.0281, "step": 37338 }, { "epoch": 1.9906701498107373, "grad_norm": 2.09563946723938, "learning_rate": 2.5324000000000005e-06, "loss": 0.0335, "step": 37339 }, { "epoch": 1.9907234632403903, "grad_norm": 0.32537418603897095, "learning_rate": 2.5322e-06, "loss": 0.0023, "step": 37340 }, { "epoch": 1.9907767766700433, "grad_norm": 1.786888599395752, "learning_rate": 2.532e-06, "loss": 0.0167, "step": 37341 }, { "epoch": 1.990830090099696, "grad_norm": 1.5207600593566895, "learning_rate": 2.5318000000000005e-06, "loss": 0.0083, "step": 37342 }, { "epoch": 1.990883403529349, "grad_norm": 1.3085262775421143, "learning_rate": 2.5316e-06, "loss": 0.0249, "step": 37343 }, { "epoch": 1.990936716959002, "grad_norm": 2.354323387145996, "learning_rate": 2.5314e-06, "loss": 0.0186, "step": 37344 }, { "epoch": 1.9909900303886547, "grad_norm": 2.8197035789489746, "learning_rate": 2.5312000000000004e-06, "loss": 0.0541, "step": 37345 }, { "epoch": 1.991043343818308, "grad_norm": 0.7855698466300964, "learning_rate": 2.531e-06, "loss": 0.0057, "step": 37346 }, { "epoch": 1.9910966572479607, "grad_norm": 2.1690478324890137, "learning_rate": 2.5308000000000004e-06, "loss": 0.0079, "step": 37347 }, { "epoch": 1.9911499706776137, "grad_norm": 2.376851797103882, "learning_rate": 2.5306000000000004e-06, "loss": 0.0194, "step": 37348 }, { "epoch": 1.9912032841072667, "grad_norm": 2.1321563720703125, "learning_rate": 2.5304e-06, "loss": 0.0336, "step": 37349 }, { "epoch": 1.9912565975369194, "grad_norm": 2.0408084392547607, "learning_rate": 2.5302000000000004e-06, "loss": 0.0302, "step": 37350 }, { "epoch": 1.9913099109665726, "grad_norm": 1.1186275482177734, "learning_rate": 2.5300000000000003e-06, "loss": 0.009, "step": 37351 }, { "epoch": 1.9913632243962254, "grad_norm": 2.3675143718719482, "learning_rate": 2.5298e-06, "loss": 0.0235, "step": 37352 }, { "epoch": 1.9914165378258784, "grad_norm": 1.7423127889633179, "learning_rate": 2.5296000000000003e-06, "loss": 0.0338, "step": 37353 }, { "epoch": 1.9914698512555313, "grad_norm": 1.7395968437194824, "learning_rate": 2.5294000000000003e-06, "loss": 0.0138, "step": 37354 }, { "epoch": 1.991523164685184, "grad_norm": 1.2518171072006226, "learning_rate": 2.5292e-06, "loss": 0.0114, "step": 37355 }, { "epoch": 1.9915764781148373, "grad_norm": 2.16523814201355, "learning_rate": 2.5290000000000003e-06, "loss": 0.0182, "step": 37356 }, { "epoch": 1.99162979154449, "grad_norm": 2.9074528217315674, "learning_rate": 2.5288000000000003e-06, "loss": 0.0302, "step": 37357 }, { "epoch": 1.991683104974143, "grad_norm": 1.9665228128433228, "learning_rate": 2.5286e-06, "loss": 0.0367, "step": 37358 }, { "epoch": 1.991736418403796, "grad_norm": 2.2570135593414307, "learning_rate": 2.5284000000000002e-06, "loss": 0.0058, "step": 37359 }, { "epoch": 1.9917897318334488, "grad_norm": 3.8232479095458984, "learning_rate": 2.5282e-06, "loss": 0.0294, "step": 37360 }, { "epoch": 1.9918430452631017, "grad_norm": 1.6888513565063477, "learning_rate": 2.5280000000000006e-06, "loss": 0.0148, "step": 37361 }, { "epoch": 1.9918963586927547, "grad_norm": 1.0834195613861084, "learning_rate": 2.5278e-06, "loss": 0.0073, "step": 37362 }, { "epoch": 1.9919496721224075, "grad_norm": 0.17542211711406708, "learning_rate": 2.5276e-06, "loss": 0.0022, "step": 37363 }, { "epoch": 1.9920029855520607, "grad_norm": 2.349540948867798, "learning_rate": 2.5274000000000006e-06, "loss": 0.0131, "step": 37364 }, { "epoch": 1.9920562989817134, "grad_norm": 1.3505961894989014, "learning_rate": 2.5272e-06, "loss": 0.0078, "step": 37365 }, { "epoch": 1.9921096124113664, "grad_norm": 0.42792797088623047, "learning_rate": 2.527e-06, "loss": 0.0042, "step": 37366 }, { "epoch": 1.9921629258410194, "grad_norm": 1.743802785873413, "learning_rate": 2.5268000000000005e-06, "loss": 0.0172, "step": 37367 }, { "epoch": 1.9922162392706722, "grad_norm": 4.599204063415527, "learning_rate": 2.5266e-06, "loss": 0.0328, "step": 37368 }, { "epoch": 1.9922695527003254, "grad_norm": 2.1150262355804443, "learning_rate": 2.5264e-06, "loss": 0.0185, "step": 37369 }, { "epoch": 1.992322866129978, "grad_norm": 1.820464015007019, "learning_rate": 2.5262000000000005e-06, "loss": 0.0096, "step": 37370 }, { "epoch": 1.992376179559631, "grad_norm": 3.0691962242126465, "learning_rate": 2.526e-06, "loss": 0.0242, "step": 37371 }, { "epoch": 1.992429492989284, "grad_norm": 1.9569709300994873, "learning_rate": 2.5258000000000004e-06, "loss": 0.0292, "step": 37372 }, { "epoch": 1.9924828064189368, "grad_norm": 1.8499009609222412, "learning_rate": 2.5256000000000004e-06, "loss": 0.0124, "step": 37373 }, { "epoch": 1.99253611984859, "grad_norm": 4.699991703033447, "learning_rate": 2.5254e-06, "loss": 0.0397, "step": 37374 }, { "epoch": 1.9925894332782428, "grad_norm": 1.3659404516220093, "learning_rate": 2.5252000000000004e-06, "loss": 0.0183, "step": 37375 }, { "epoch": 1.9926427467078958, "grad_norm": 1.5653791427612305, "learning_rate": 2.5250000000000004e-06, "loss": 0.0108, "step": 37376 }, { "epoch": 1.9926960601375487, "grad_norm": 3.165262222290039, "learning_rate": 2.5248e-06, "loss": 0.0245, "step": 37377 }, { "epoch": 1.9927493735672015, "grad_norm": 3.1625900268554688, "learning_rate": 2.5246000000000004e-06, "loss": 0.0519, "step": 37378 }, { "epoch": 1.9928026869968545, "grad_norm": 2.5087435245513916, "learning_rate": 2.5244000000000003e-06, "loss": 0.0438, "step": 37379 }, { "epoch": 1.9928560004265075, "grad_norm": 0.9165576696395874, "learning_rate": 2.5242e-06, "loss": 0.006, "step": 37380 }, { "epoch": 1.9929093138561602, "grad_norm": 2.5797977447509766, "learning_rate": 2.5240000000000003e-06, "loss": 0.0464, "step": 37381 }, { "epoch": 1.9929626272858134, "grad_norm": 3.034801959991455, "learning_rate": 2.5238000000000003e-06, "loss": 0.0063, "step": 37382 }, { "epoch": 1.9930159407154662, "grad_norm": 0.9492202997207642, "learning_rate": 2.5236e-06, "loss": 0.0071, "step": 37383 }, { "epoch": 1.9930692541451192, "grad_norm": 3.0613296031951904, "learning_rate": 2.5234000000000003e-06, "loss": 0.024, "step": 37384 }, { "epoch": 1.9931225675747721, "grad_norm": 0.9645405411720276, "learning_rate": 2.5232000000000003e-06, "loss": 0.0066, "step": 37385 }, { "epoch": 1.9931758810044249, "grad_norm": 3.1348628997802734, "learning_rate": 2.5230000000000002e-06, "loss": 0.0282, "step": 37386 }, { "epoch": 1.993229194434078, "grad_norm": 0.7313545942306519, "learning_rate": 2.5228000000000002e-06, "loss": 0.0058, "step": 37387 }, { "epoch": 1.9932825078637308, "grad_norm": 1.6412030458450317, "learning_rate": 2.5226e-06, "loss": 0.0212, "step": 37388 }, { "epoch": 1.9933358212933838, "grad_norm": 1.3658748865127563, "learning_rate": 2.5224e-06, "loss": 0.0067, "step": 37389 }, { "epoch": 1.9933891347230368, "grad_norm": 2.969252347946167, "learning_rate": 2.5222e-06, "loss": 0.0254, "step": 37390 }, { "epoch": 1.9934424481526896, "grad_norm": 1.9544446468353271, "learning_rate": 2.522e-06, "loss": 0.0434, "step": 37391 }, { "epoch": 1.9934957615823425, "grad_norm": 4.290657043457031, "learning_rate": 2.5218e-06, "loss": 0.0162, "step": 37392 }, { "epoch": 1.9935490750119955, "grad_norm": 2.046504497528076, "learning_rate": 2.5216e-06, "loss": 0.0211, "step": 37393 }, { "epoch": 1.9936023884416485, "grad_norm": 1.6857746839523315, "learning_rate": 2.5214e-06, "loss": 0.0101, "step": 37394 }, { "epoch": 1.9936557018713015, "grad_norm": 2.0508766174316406, "learning_rate": 2.5212e-06, "loss": 0.0357, "step": 37395 }, { "epoch": 1.9937090153009542, "grad_norm": 1.8943055868148804, "learning_rate": 2.521e-06, "loss": 0.0091, "step": 37396 }, { "epoch": 1.9937623287306072, "grad_norm": 2.205981492996216, "learning_rate": 2.5208000000000005e-06, "loss": 0.0222, "step": 37397 }, { "epoch": 1.9938156421602602, "grad_norm": 1.075198769569397, "learning_rate": 2.5206e-06, "loss": 0.0122, "step": 37398 }, { "epoch": 1.993868955589913, "grad_norm": 2.548924207687378, "learning_rate": 2.5204e-06, "loss": 0.0134, "step": 37399 }, { "epoch": 1.9939222690195662, "grad_norm": 1.9343268871307373, "learning_rate": 2.5202000000000005e-06, "loss": 0.0127, "step": 37400 }, { "epoch": 1.993975582449219, "grad_norm": 2.2753396034240723, "learning_rate": 2.52e-06, "loss": 0.0138, "step": 37401 }, { "epoch": 1.9940288958788719, "grad_norm": 2.6886279582977295, "learning_rate": 2.5198e-06, "loss": 0.0205, "step": 37402 }, { "epoch": 1.9940822093085249, "grad_norm": 1.8425562381744385, "learning_rate": 2.5196000000000004e-06, "loss": 0.0273, "step": 37403 }, { "epoch": 1.9941355227381776, "grad_norm": 2.8552260398864746, "learning_rate": 2.5194e-06, "loss": 0.0319, "step": 37404 }, { "epoch": 1.9941888361678308, "grad_norm": 2.9158236980438232, "learning_rate": 2.5192e-06, "loss": 0.0172, "step": 37405 }, { "epoch": 1.9942421495974836, "grad_norm": 1.510105013847351, "learning_rate": 2.5190000000000004e-06, "loss": 0.0105, "step": 37406 }, { "epoch": 1.9942954630271366, "grad_norm": 1.5780532360076904, "learning_rate": 2.5188e-06, "loss": 0.0203, "step": 37407 }, { "epoch": 1.9943487764567895, "grad_norm": 1.1130188703536987, "learning_rate": 2.5186e-06, "loss": 0.0164, "step": 37408 }, { "epoch": 1.9944020898864423, "grad_norm": 3.972466230392456, "learning_rate": 2.5184000000000003e-06, "loss": 0.0212, "step": 37409 }, { "epoch": 1.9944554033160953, "grad_norm": 1.1046085357666016, "learning_rate": 2.5182e-06, "loss": 0.0147, "step": 37410 }, { "epoch": 1.9945087167457483, "grad_norm": 0.9511897563934326, "learning_rate": 2.5180000000000003e-06, "loss": 0.0071, "step": 37411 }, { "epoch": 1.9945620301754012, "grad_norm": 1.8414756059646606, "learning_rate": 2.5178000000000003e-06, "loss": 0.0186, "step": 37412 }, { "epoch": 1.9946153436050542, "grad_norm": 1.6557151079177856, "learning_rate": 2.5176e-06, "loss": 0.0212, "step": 37413 }, { "epoch": 1.994668657034707, "grad_norm": 0.28990843892097473, "learning_rate": 2.5174000000000002e-06, "loss": 0.002, "step": 37414 }, { "epoch": 1.99472197046436, "grad_norm": 1.0428370237350464, "learning_rate": 2.5172000000000002e-06, "loss": 0.0274, "step": 37415 }, { "epoch": 1.994775283894013, "grad_norm": 2.365262508392334, "learning_rate": 2.5169999999999998e-06, "loss": 0.0228, "step": 37416 }, { "epoch": 1.9948285973236657, "grad_norm": 2.280433177947998, "learning_rate": 2.5168e-06, "loss": 0.0288, "step": 37417 }, { "epoch": 1.9948819107533189, "grad_norm": 0.1899801641702652, "learning_rate": 2.5166e-06, "loss": 0.0019, "step": 37418 }, { "epoch": 1.9949352241829716, "grad_norm": 2.5499520301818848, "learning_rate": 2.5163999999999997e-06, "loss": 0.0293, "step": 37419 }, { "epoch": 1.9949885376126246, "grad_norm": 1.8660814762115479, "learning_rate": 2.5162e-06, "loss": 0.0169, "step": 37420 }, { "epoch": 1.9950418510422776, "grad_norm": 2.26369571685791, "learning_rate": 2.516e-06, "loss": 0.011, "step": 37421 }, { "epoch": 1.9950951644719304, "grad_norm": 0.9692542552947998, "learning_rate": 2.5158000000000005e-06, "loss": 0.0231, "step": 37422 }, { "epoch": 1.9951484779015836, "grad_norm": 0.304161936044693, "learning_rate": 2.5156e-06, "loss": 0.0027, "step": 37423 }, { "epoch": 1.9952017913312363, "grad_norm": 3.943523645401001, "learning_rate": 2.5154e-06, "loss": 0.0186, "step": 37424 }, { "epoch": 1.9952551047608893, "grad_norm": 1.7760589122772217, "learning_rate": 2.5152000000000005e-06, "loss": 0.0238, "step": 37425 }, { "epoch": 1.9953084181905423, "grad_norm": 1.6511672735214233, "learning_rate": 2.515e-06, "loss": 0.0358, "step": 37426 }, { "epoch": 1.995361731620195, "grad_norm": 2.596435070037842, "learning_rate": 2.5148e-06, "loss": 0.018, "step": 37427 }, { "epoch": 1.995415045049848, "grad_norm": 1.9586385488510132, "learning_rate": 2.5146000000000005e-06, "loss": 0.0126, "step": 37428 }, { "epoch": 1.995468358479501, "grad_norm": 1.4111636877059937, "learning_rate": 2.5144e-06, "loss": 0.0269, "step": 37429 }, { "epoch": 1.995521671909154, "grad_norm": 1.6064931154251099, "learning_rate": 2.5142e-06, "loss": 0.0333, "step": 37430 }, { "epoch": 1.995574985338807, "grad_norm": 0.9464184045791626, "learning_rate": 2.5140000000000004e-06, "loss": 0.0053, "step": 37431 }, { "epoch": 1.9956282987684597, "grad_norm": 1.5048521757125854, "learning_rate": 2.5138e-06, "loss": 0.0273, "step": 37432 }, { "epoch": 1.9956816121981127, "grad_norm": 2.277452230453491, "learning_rate": 2.5136e-06, "loss": 0.0337, "step": 37433 }, { "epoch": 1.9957349256277657, "grad_norm": 1.4698320627212524, "learning_rate": 2.5134000000000004e-06, "loss": 0.0114, "step": 37434 }, { "epoch": 1.9957882390574184, "grad_norm": 1.1803772449493408, "learning_rate": 2.5132e-06, "loss": 0.0158, "step": 37435 }, { "epoch": 1.9958415524870716, "grad_norm": 3.0603411197662354, "learning_rate": 2.5130000000000003e-06, "loss": 0.0297, "step": 37436 }, { "epoch": 1.9958948659167244, "grad_norm": 3.3776800632476807, "learning_rate": 2.5128000000000003e-06, "loss": 0.0143, "step": 37437 }, { "epoch": 1.9959481793463774, "grad_norm": 1.2850357294082642, "learning_rate": 2.5126e-06, "loss": 0.0078, "step": 37438 }, { "epoch": 1.9960014927760303, "grad_norm": 0.37182363867759705, "learning_rate": 2.5124000000000003e-06, "loss": 0.0029, "step": 37439 }, { "epoch": 1.996054806205683, "grad_norm": 1.6879891157150269, "learning_rate": 2.5122000000000003e-06, "loss": 0.0074, "step": 37440 }, { "epoch": 1.9961081196353363, "grad_norm": 1.4803498983383179, "learning_rate": 2.512e-06, "loss": 0.0143, "step": 37441 }, { "epoch": 1.996161433064989, "grad_norm": 1.916538953781128, "learning_rate": 2.5118000000000002e-06, "loss": 0.0303, "step": 37442 }, { "epoch": 1.996214746494642, "grad_norm": 1.5820001363754272, "learning_rate": 2.5116000000000002e-06, "loss": 0.0101, "step": 37443 }, { "epoch": 1.996268059924295, "grad_norm": 0.3808204233646393, "learning_rate": 2.5114e-06, "loss": 0.0031, "step": 37444 }, { "epoch": 1.9963213733539478, "grad_norm": 2.326211929321289, "learning_rate": 2.5112e-06, "loss": 0.0192, "step": 37445 }, { "epoch": 1.9963746867836007, "grad_norm": 1.7045738697052002, "learning_rate": 2.511e-06, "loss": 0.0308, "step": 37446 }, { "epoch": 1.9964280002132537, "grad_norm": 1.38687264919281, "learning_rate": 2.5108000000000006e-06, "loss": 0.0234, "step": 37447 }, { "epoch": 1.9964813136429065, "grad_norm": 1.1576652526855469, "learning_rate": 2.5106e-06, "loss": 0.0172, "step": 37448 }, { "epoch": 1.9965346270725597, "grad_norm": 2.3201310634613037, "learning_rate": 2.5104e-06, "loss": 0.0287, "step": 37449 }, { "epoch": 1.9965879405022124, "grad_norm": 2.2610700130462646, "learning_rate": 2.5102000000000005e-06, "loss": 0.0254, "step": 37450 }, { "epoch": 1.9966412539318654, "grad_norm": 0.7672127485275269, "learning_rate": 2.51e-06, "loss": 0.0046, "step": 37451 }, { "epoch": 1.9966945673615184, "grad_norm": 6.671769142150879, "learning_rate": 2.5098e-06, "loss": 0.0318, "step": 37452 }, { "epoch": 1.9967478807911712, "grad_norm": 2.2972824573516846, "learning_rate": 2.5096000000000005e-06, "loss": 0.0094, "step": 37453 }, { "epoch": 1.9968011942208244, "grad_norm": 1.7320175170898438, "learning_rate": 2.5094e-06, "loss": 0.0239, "step": 37454 }, { "epoch": 1.9968545076504771, "grad_norm": 2.7509262561798096, "learning_rate": 2.5092e-06, "loss": 0.0111, "step": 37455 }, { "epoch": 1.99690782108013, "grad_norm": 0.831056535243988, "learning_rate": 2.5090000000000005e-06, "loss": 0.0054, "step": 37456 }, { "epoch": 1.996961134509783, "grad_norm": 1.6905357837677002, "learning_rate": 2.5088e-06, "loss": 0.0154, "step": 37457 }, { "epoch": 1.9970144479394358, "grad_norm": 0.5336081385612488, "learning_rate": 2.5086000000000004e-06, "loss": 0.0044, "step": 37458 }, { "epoch": 1.997067761369089, "grad_norm": 2.1103756427764893, "learning_rate": 2.5084000000000004e-06, "loss": 0.0323, "step": 37459 }, { "epoch": 1.9971210747987418, "grad_norm": 1.0818893909454346, "learning_rate": 2.5082e-06, "loss": 0.0083, "step": 37460 }, { "epoch": 1.9971743882283948, "grad_norm": 0.5433468222618103, "learning_rate": 2.5080000000000004e-06, "loss": 0.0041, "step": 37461 }, { "epoch": 1.9972277016580477, "grad_norm": 1.8655245304107666, "learning_rate": 2.5078000000000004e-06, "loss": 0.0253, "step": 37462 }, { "epoch": 1.9972810150877005, "grad_norm": 2.054784059524536, "learning_rate": 2.5076e-06, "loss": 0.0092, "step": 37463 }, { "epoch": 1.9973343285173535, "grad_norm": 1.371449589729309, "learning_rate": 2.5074000000000003e-06, "loss": 0.0046, "step": 37464 }, { "epoch": 1.9973876419470065, "grad_norm": 1.8582271337509155, "learning_rate": 2.5072000000000003e-06, "loss": 0.0246, "step": 37465 }, { "epoch": 1.9974409553766592, "grad_norm": 1.490369200706482, "learning_rate": 2.507e-06, "loss": 0.0256, "step": 37466 }, { "epoch": 1.9974942688063124, "grad_norm": 1.9592370986938477, "learning_rate": 2.5068000000000003e-06, "loss": 0.0133, "step": 37467 }, { "epoch": 1.9975475822359652, "grad_norm": 1.6038798093795776, "learning_rate": 2.5066000000000003e-06, "loss": 0.0172, "step": 37468 }, { "epoch": 1.9976008956656182, "grad_norm": 0.27871274948120117, "learning_rate": 2.5064e-06, "loss": 0.0026, "step": 37469 }, { "epoch": 1.9976542090952711, "grad_norm": 1.5832794904708862, "learning_rate": 2.5062000000000002e-06, "loss": 0.0149, "step": 37470 }, { "epoch": 1.997707522524924, "grad_norm": 1.9745579957962036, "learning_rate": 2.5060000000000002e-06, "loss": 0.0195, "step": 37471 }, { "epoch": 1.997760835954577, "grad_norm": 1.9547053575515747, "learning_rate": 2.5058000000000006e-06, "loss": 0.0254, "step": 37472 }, { "epoch": 1.9978141493842299, "grad_norm": 2.5735373497009277, "learning_rate": 2.5056e-06, "loss": 0.0187, "step": 37473 }, { "epoch": 1.9978674628138828, "grad_norm": 1.155563235282898, "learning_rate": 2.5054e-06, "loss": 0.0056, "step": 37474 }, { "epoch": 1.9979207762435358, "grad_norm": 1.3579410314559937, "learning_rate": 2.5052000000000006e-06, "loss": 0.0075, "step": 37475 }, { "epoch": 1.9979740896731886, "grad_norm": 0.583634614944458, "learning_rate": 2.505e-06, "loss": 0.0035, "step": 37476 }, { "epoch": 1.9980274031028418, "grad_norm": 0.6055499911308289, "learning_rate": 2.5048e-06, "loss": 0.0058, "step": 37477 }, { "epoch": 1.9980807165324945, "grad_norm": 1.3931567668914795, "learning_rate": 2.5046000000000006e-06, "loss": 0.0348, "step": 37478 }, { "epoch": 1.9981340299621475, "grad_norm": 2.2921645641326904, "learning_rate": 2.5044e-06, "loss": 0.0191, "step": 37479 }, { "epoch": 1.9981873433918005, "grad_norm": 2.508801221847534, "learning_rate": 2.5042e-06, "loss": 0.0258, "step": 37480 }, { "epoch": 1.9982406568214532, "grad_norm": 0.25693079829216003, "learning_rate": 2.5040000000000005e-06, "loss": 0.0024, "step": 37481 }, { "epoch": 1.9982939702511062, "grad_norm": 1.1209402084350586, "learning_rate": 2.5038e-06, "loss": 0.0048, "step": 37482 }, { "epoch": 1.9983472836807592, "grad_norm": 2.2340285778045654, "learning_rate": 2.5036000000000005e-06, "loss": 0.0278, "step": 37483 }, { "epoch": 1.998400597110412, "grad_norm": 0.5697845816612244, "learning_rate": 2.5034000000000005e-06, "loss": 0.0041, "step": 37484 }, { "epoch": 1.9984539105400652, "grad_norm": 2.3113179206848145, "learning_rate": 2.5032e-06, "loss": 0.0388, "step": 37485 }, { "epoch": 1.998507223969718, "grad_norm": 1.6749430894851685, "learning_rate": 2.5030000000000004e-06, "loss": 0.0078, "step": 37486 }, { "epoch": 1.998560537399371, "grad_norm": 0.3679007887840271, "learning_rate": 2.5028000000000004e-06, "loss": 0.0034, "step": 37487 }, { "epoch": 1.9986138508290239, "grad_norm": 2.8343088626861572, "learning_rate": 2.5026e-06, "loss": 0.0175, "step": 37488 }, { "epoch": 1.9986671642586766, "grad_norm": 1.8513076305389404, "learning_rate": 2.5024000000000004e-06, "loss": 0.0317, "step": 37489 }, { "epoch": 1.9987204776883298, "grad_norm": 0.45964184403419495, "learning_rate": 2.5022000000000004e-06, "loss": 0.0031, "step": 37490 }, { "epoch": 1.9987737911179826, "grad_norm": 2.757420063018799, "learning_rate": 2.502e-06, "loss": 0.0298, "step": 37491 }, { "epoch": 1.9988271045476356, "grad_norm": 2.6621134281158447, "learning_rate": 2.5018000000000003e-06, "loss": 0.0194, "step": 37492 }, { "epoch": 1.9988804179772885, "grad_norm": 1.2919948101043701, "learning_rate": 2.5016000000000003e-06, "loss": 0.0152, "step": 37493 }, { "epoch": 1.9989337314069413, "grad_norm": 1.2640953063964844, "learning_rate": 2.5014e-06, "loss": 0.0053, "step": 37494 }, { "epoch": 1.9989870448365943, "grad_norm": 2.2625315189361572, "learning_rate": 2.5012000000000003e-06, "loss": 0.0161, "step": 37495 }, { "epoch": 1.9990403582662473, "grad_norm": 1.9925888776779175, "learning_rate": 2.5010000000000003e-06, "loss": 0.0076, "step": 37496 }, { "epoch": 1.9990936716959002, "grad_norm": 1.8034697771072388, "learning_rate": 2.5008000000000003e-06, "loss": 0.0097, "step": 37497 }, { "epoch": 1.9991469851255532, "grad_norm": 0.8560008406639099, "learning_rate": 2.5006000000000002e-06, "loss": 0.005, "step": 37498 }, { "epoch": 1.999200298555206, "grad_norm": 3.013312339782715, "learning_rate": 2.5004000000000002e-06, "loss": 0.0384, "step": 37499 }, { "epoch": 1.999253611984859, "grad_norm": 1.4342149496078491, "learning_rate": 2.5002000000000002e-06, "loss": 0.0098, "step": 37500 }, { "epoch": 1.999306925414512, "grad_norm": 1.186591625213623, "learning_rate": 2.5e-06, "loss": 0.0058, "step": 37501 }, { "epoch": 1.9993602388441647, "grad_norm": 2.843653917312622, "learning_rate": 2.4998e-06, "loss": 0.0313, "step": 37502 }, { "epoch": 1.999413552273818, "grad_norm": 0.4101564586162567, "learning_rate": 2.4996e-06, "loss": 0.0032, "step": 37503 }, { "epoch": 1.9994668657034707, "grad_norm": 5.534938335418701, "learning_rate": 2.4994e-06, "loss": 0.0255, "step": 37504 }, { "epoch": 1.9995201791331236, "grad_norm": 0.4278987944126129, "learning_rate": 2.4992e-06, "loss": 0.0038, "step": 37505 }, { "epoch": 1.9995734925627766, "grad_norm": 4.058761119842529, "learning_rate": 2.499e-06, "loss": 0.0226, "step": 37506 }, { "epoch": 1.9996268059924294, "grad_norm": 2.559772253036499, "learning_rate": 2.4988e-06, "loss": 0.0411, "step": 37507 }, { "epoch": 1.9996801194220826, "grad_norm": 1.7755635976791382, "learning_rate": 2.4986e-06, "loss": 0.0146, "step": 37508 }, { "epoch": 1.9997334328517353, "grad_norm": 4.068098545074463, "learning_rate": 2.4984e-06, "loss": 0.0162, "step": 37509 }, { "epoch": 1.9997867462813883, "grad_norm": 2.6380209922790527, "learning_rate": 2.4982e-06, "loss": 0.0129, "step": 37510 }, { "epoch": 1.9998400597110413, "grad_norm": 1.7418323755264282, "learning_rate": 2.498e-06, "loss": 0.0118, "step": 37511 }, { "epoch": 1.999893373140694, "grad_norm": 2.7471141815185547, "learning_rate": 2.4978e-06, "loss": 0.0137, "step": 37512 }, { "epoch": 1.999946686570347, "grad_norm": 0.5683687329292297, "learning_rate": 2.4976000000000004e-06, "loss": 0.0046, "step": 37513 }, { "epoch": 2.0, "grad_norm": 0.6329697370529175, "learning_rate": 2.4974e-06, "loss": 0.0046, "step": 37514 }, { "epoch": 2.0000533134296528, "grad_norm": 1.3002647161483765, "learning_rate": 2.4972e-06, "loss": 0.008, "step": 37515 }, { "epoch": 2.000106626859306, "grad_norm": 2.386641025543213, "learning_rate": 2.4970000000000004e-06, "loss": 0.0328, "step": 37516 }, { "epoch": 2.0001599402889587, "grad_norm": 2.5056748390197754, "learning_rate": 2.4968000000000004e-06, "loss": 0.0265, "step": 37517 }, { "epoch": 2.000213253718612, "grad_norm": 0.49055778980255127, "learning_rate": 2.4966e-06, "loss": 0.0024, "step": 37518 }, { "epoch": 2.0002665671482647, "grad_norm": 1.056771993637085, "learning_rate": 2.4964000000000004e-06, "loss": 0.0079, "step": 37519 }, { "epoch": 2.0003198805779174, "grad_norm": 0.30883079767227173, "learning_rate": 2.4962000000000003e-06, "loss": 0.0023, "step": 37520 }, { "epoch": 2.0003731940075706, "grad_norm": 2.5020337104797363, "learning_rate": 2.496e-06, "loss": 0.0124, "step": 37521 }, { "epoch": 2.0004265074372234, "grad_norm": 1.8476859331130981, "learning_rate": 2.4958000000000003e-06, "loss": 0.0082, "step": 37522 }, { "epoch": 2.000479820866876, "grad_norm": 1.5254530906677246, "learning_rate": 2.4956000000000003e-06, "loss": 0.0076, "step": 37523 }, { "epoch": 2.0005331342965293, "grad_norm": 2.091844081878662, "learning_rate": 2.4954000000000003e-06, "loss": 0.0502, "step": 37524 }, { "epoch": 2.000586447726182, "grad_norm": 2.126340866088867, "learning_rate": 2.4952000000000003e-06, "loss": 0.0185, "step": 37525 }, { "epoch": 2.0006397611558353, "grad_norm": 1.224623203277588, "learning_rate": 2.4950000000000003e-06, "loss": 0.0183, "step": 37526 }, { "epoch": 2.000693074585488, "grad_norm": 2.85353946685791, "learning_rate": 2.4948000000000002e-06, "loss": 0.0438, "step": 37527 }, { "epoch": 2.000746388015141, "grad_norm": 1.5537489652633667, "learning_rate": 2.4946000000000002e-06, "loss": 0.0099, "step": 37528 }, { "epoch": 2.000799701444794, "grad_norm": 0.3308592438697815, "learning_rate": 2.4944e-06, "loss": 0.0027, "step": 37529 }, { "epoch": 2.000853014874447, "grad_norm": 2.6524486541748047, "learning_rate": 2.4942e-06, "loss": 0.049, "step": 37530 }, { "epoch": 2.0009063283041, "grad_norm": 1.775801420211792, "learning_rate": 2.494e-06, "loss": 0.0184, "step": 37531 }, { "epoch": 2.0009596417337527, "grad_norm": 0.2797192633152008, "learning_rate": 2.4938e-06, "loss": 0.0024, "step": 37532 }, { "epoch": 2.0010129551634055, "grad_norm": 2.791916608810425, "learning_rate": 2.4936e-06, "loss": 0.0121, "step": 37533 }, { "epoch": 2.0010662685930587, "grad_norm": 5.211215496063232, "learning_rate": 2.4934e-06, "loss": 0.0103, "step": 37534 }, { "epoch": 2.0011195820227115, "grad_norm": 2.352921724319458, "learning_rate": 2.4932000000000005e-06, "loss": 0.0449, "step": 37535 }, { "epoch": 2.0011728954523647, "grad_norm": 0.2663072347640991, "learning_rate": 2.493e-06, "loss": 0.002, "step": 37536 }, { "epoch": 2.0012262088820174, "grad_norm": 0.7263460755348206, "learning_rate": 2.4928e-06, "loss": 0.0044, "step": 37537 }, { "epoch": 2.00127952231167, "grad_norm": 1.756913661956787, "learning_rate": 2.4926000000000005e-06, "loss": 0.0291, "step": 37538 }, { "epoch": 2.0013328357413234, "grad_norm": 3.1862192153930664, "learning_rate": 2.4924e-06, "loss": 0.0521, "step": 37539 }, { "epoch": 2.001386149170976, "grad_norm": 2.9031286239624023, "learning_rate": 2.4922e-06, "loss": 0.0367, "step": 37540 }, { "epoch": 2.001439462600629, "grad_norm": 0.2209583818912506, "learning_rate": 2.4920000000000005e-06, "loss": 0.0017, "step": 37541 }, { "epoch": 2.001492776030282, "grad_norm": 0.859575092792511, "learning_rate": 2.4918000000000004e-06, "loss": 0.0063, "step": 37542 }, { "epoch": 2.001546089459935, "grad_norm": 3.400601863861084, "learning_rate": 2.4916e-06, "loss": 0.0306, "step": 37543 }, { "epoch": 2.001599402889588, "grad_norm": 0.9487320184707642, "learning_rate": 2.4914000000000004e-06, "loss": 0.0042, "step": 37544 }, { "epoch": 2.001652716319241, "grad_norm": 1.5115141868591309, "learning_rate": 2.4912000000000004e-06, "loss": 0.0256, "step": 37545 }, { "epoch": 2.0017060297488936, "grad_norm": 1.147570013999939, "learning_rate": 2.491e-06, "loss": 0.0044, "step": 37546 }, { "epoch": 2.0017593431785468, "grad_norm": 1.49701988697052, "learning_rate": 2.4908000000000004e-06, "loss": 0.0167, "step": 37547 }, { "epoch": 2.0018126566081995, "grad_norm": 2.2370243072509766, "learning_rate": 2.4906000000000003e-06, "loss": 0.0137, "step": 37548 }, { "epoch": 2.0018659700378527, "grad_norm": 1.6907953023910522, "learning_rate": 2.4904000000000003e-06, "loss": 0.0131, "step": 37549 }, { "epoch": 2.0019192834675055, "grad_norm": 2.8237926959991455, "learning_rate": 2.4902000000000003e-06, "loss": 0.0084, "step": 37550 }, { "epoch": 2.0019725968971582, "grad_norm": 1.6067909002304077, "learning_rate": 2.4900000000000003e-06, "loss": 0.0294, "step": 37551 }, { "epoch": 2.0020259103268114, "grad_norm": 0.48358917236328125, "learning_rate": 2.4898000000000003e-06, "loss": 0.0037, "step": 37552 }, { "epoch": 2.002079223756464, "grad_norm": 1.80034339427948, "learning_rate": 2.4896000000000003e-06, "loss": 0.0132, "step": 37553 }, { "epoch": 2.0021325371861174, "grad_norm": 2.259105920791626, "learning_rate": 2.4894000000000003e-06, "loss": 0.0216, "step": 37554 }, { "epoch": 2.00218585061577, "grad_norm": 1.042196273803711, "learning_rate": 2.4892000000000002e-06, "loss": 0.0064, "step": 37555 }, { "epoch": 2.002239164045423, "grad_norm": 1.8615933656692505, "learning_rate": 2.4890000000000002e-06, "loss": 0.0312, "step": 37556 }, { "epoch": 2.002292477475076, "grad_norm": 0.6549168229103088, "learning_rate": 2.4888e-06, "loss": 0.0051, "step": 37557 }, { "epoch": 2.002345790904729, "grad_norm": 2.4858126640319824, "learning_rate": 2.4886e-06, "loss": 0.0188, "step": 37558 }, { "epoch": 2.0023991043343816, "grad_norm": 1.4902986288070679, "learning_rate": 2.4884e-06, "loss": 0.0062, "step": 37559 }, { "epoch": 2.002452417764035, "grad_norm": 1.7585291862487793, "learning_rate": 2.4882e-06, "loss": 0.0228, "step": 37560 }, { "epoch": 2.0025057311936876, "grad_norm": 1.049760341644287, "learning_rate": 2.488e-06, "loss": 0.0059, "step": 37561 }, { "epoch": 2.002559044623341, "grad_norm": 1.0660290718078613, "learning_rate": 2.4878e-06, "loss": 0.0045, "step": 37562 }, { "epoch": 2.0026123580529935, "grad_norm": 3.0038154125213623, "learning_rate": 2.4876e-06, "loss": 0.0135, "step": 37563 }, { "epoch": 2.0026656714826463, "grad_norm": 1.3753389120101929, "learning_rate": 2.4874e-06, "loss": 0.0048, "step": 37564 }, { "epoch": 2.0027189849122995, "grad_norm": 1.712520718574524, "learning_rate": 2.4872e-06, "loss": 0.027, "step": 37565 }, { "epoch": 2.0027722983419523, "grad_norm": 2.3130669593811035, "learning_rate": 2.487e-06, "loss": 0.0206, "step": 37566 }, { "epoch": 2.0028256117716055, "grad_norm": 2.101813316345215, "learning_rate": 2.4868000000000005e-06, "loss": 0.0388, "step": 37567 }, { "epoch": 2.002878925201258, "grad_norm": 0.6655169725418091, "learning_rate": 2.4866e-06, "loss": 0.0043, "step": 37568 }, { "epoch": 2.002932238630911, "grad_norm": 1.2467883825302124, "learning_rate": 2.4864e-06, "loss": 0.0069, "step": 37569 }, { "epoch": 2.002985552060564, "grad_norm": 3.926558256149292, "learning_rate": 2.4862000000000004e-06, "loss": 0.031, "step": 37570 }, { "epoch": 2.003038865490217, "grad_norm": 1.337628960609436, "learning_rate": 2.486e-06, "loss": 0.0118, "step": 37571 }, { "epoch": 2.00309217891987, "grad_norm": 1.4124250411987305, "learning_rate": 2.4858e-06, "loss": 0.0271, "step": 37572 }, { "epoch": 2.003145492349523, "grad_norm": 2.2938392162323, "learning_rate": 2.4856000000000004e-06, "loss": 0.0091, "step": 37573 }, { "epoch": 2.0031988057791756, "grad_norm": 1.460447907447815, "learning_rate": 2.4854000000000004e-06, "loss": 0.013, "step": 37574 }, { "epoch": 2.003252119208829, "grad_norm": 1.7956548929214478, "learning_rate": 2.4852e-06, "loss": 0.0533, "step": 37575 }, { "epoch": 2.0033054326384816, "grad_norm": 1.4197263717651367, "learning_rate": 2.4850000000000003e-06, "loss": 0.0196, "step": 37576 }, { "epoch": 2.0033587460681344, "grad_norm": 1.5543574094772339, "learning_rate": 2.4848000000000003e-06, "loss": 0.0061, "step": 37577 }, { "epoch": 2.0034120594977876, "grad_norm": 1.8847875595092773, "learning_rate": 2.4846e-06, "loss": 0.013, "step": 37578 }, { "epoch": 2.0034653729274403, "grad_norm": 2.1747710704803467, "learning_rate": 2.4844000000000003e-06, "loss": 0.0094, "step": 37579 }, { "epoch": 2.0035186863570935, "grad_norm": 2.496339797973633, "learning_rate": 2.4842000000000003e-06, "loss": 0.0182, "step": 37580 }, { "epoch": 2.0035719997867463, "grad_norm": 1.7601336240768433, "learning_rate": 2.4840000000000003e-06, "loss": 0.0168, "step": 37581 }, { "epoch": 2.003625313216399, "grad_norm": 1.5326987504959106, "learning_rate": 2.4838000000000003e-06, "loss": 0.0081, "step": 37582 }, { "epoch": 2.0036786266460522, "grad_norm": 0.854741632938385, "learning_rate": 2.4836000000000002e-06, "loss": 0.0057, "step": 37583 }, { "epoch": 2.003731940075705, "grad_norm": 1.4852664470672607, "learning_rate": 2.4834000000000002e-06, "loss": 0.013, "step": 37584 }, { "epoch": 2.003785253505358, "grad_norm": 0.5174044370651245, "learning_rate": 2.4832000000000002e-06, "loss": 0.0026, "step": 37585 }, { "epoch": 2.003838566935011, "grad_norm": 1.2763607501983643, "learning_rate": 2.483e-06, "loss": 0.0076, "step": 37586 }, { "epoch": 2.0038918803646637, "grad_norm": 2.905345916748047, "learning_rate": 2.4828e-06, "loss": 0.0135, "step": 37587 }, { "epoch": 2.003945193794317, "grad_norm": 2.892252206802368, "learning_rate": 2.4826e-06, "loss": 0.0435, "step": 37588 }, { "epoch": 2.0039985072239697, "grad_norm": 1.7186628580093384, "learning_rate": 2.4824e-06, "loss": 0.0198, "step": 37589 }, { "epoch": 2.004051820653623, "grad_norm": 1.148203730583191, "learning_rate": 2.4822e-06, "loss": 0.022, "step": 37590 }, { "epoch": 2.0041051340832756, "grad_norm": 1.6233406066894531, "learning_rate": 2.482e-06, "loss": 0.0081, "step": 37591 }, { "epoch": 2.0041584475129284, "grad_norm": 4.103846073150635, "learning_rate": 2.4818000000000005e-06, "loss": 0.0763, "step": 37592 }, { "epoch": 2.0042117609425816, "grad_norm": 0.41909217834472656, "learning_rate": 2.4816e-06, "loss": 0.0022, "step": 37593 }, { "epoch": 2.0042650743722343, "grad_norm": 1.5153429508209229, "learning_rate": 2.4814e-06, "loss": 0.0083, "step": 37594 }, { "epoch": 2.004318387801887, "grad_norm": 2.629385471343994, "learning_rate": 2.4812000000000005e-06, "loss": 0.0224, "step": 37595 }, { "epoch": 2.0043717012315403, "grad_norm": 1.8174177408218384, "learning_rate": 2.481e-06, "loss": 0.0298, "step": 37596 }, { "epoch": 2.004425014661193, "grad_norm": 1.390325665473938, "learning_rate": 2.4808e-06, "loss": 0.0154, "step": 37597 }, { "epoch": 2.0044783280908463, "grad_norm": 3.0418856143951416, "learning_rate": 2.4806000000000004e-06, "loss": 0.0342, "step": 37598 }, { "epoch": 2.004531641520499, "grad_norm": 0.886569082736969, "learning_rate": 2.4804000000000004e-06, "loss": 0.0035, "step": 37599 }, { "epoch": 2.0045849549501518, "grad_norm": 1.6006630659103394, "learning_rate": 2.4802e-06, "loss": 0.0122, "step": 37600 }, { "epoch": 2.004638268379805, "grad_norm": 1.7807456254959106, "learning_rate": 2.4800000000000004e-06, "loss": 0.0273, "step": 37601 }, { "epoch": 2.0046915818094577, "grad_norm": 4.7921552658081055, "learning_rate": 2.4798000000000004e-06, "loss": 0.0176, "step": 37602 }, { "epoch": 2.004744895239111, "grad_norm": 1.5327593088150024, "learning_rate": 2.4796000000000004e-06, "loss": 0.0078, "step": 37603 }, { "epoch": 2.0047982086687637, "grad_norm": 0.6070032715797424, "learning_rate": 2.4794000000000004e-06, "loss": 0.0038, "step": 37604 }, { "epoch": 2.0048515220984164, "grad_norm": 1.7100989818572998, "learning_rate": 2.4792000000000003e-06, "loss": 0.013, "step": 37605 }, { "epoch": 2.0049048355280696, "grad_norm": 0.5678879022598267, "learning_rate": 2.4790000000000003e-06, "loss": 0.0054, "step": 37606 }, { "epoch": 2.0049581489577224, "grad_norm": 1.8412878513336182, "learning_rate": 2.4788000000000003e-06, "loss": 0.0111, "step": 37607 }, { "epoch": 2.0050114623873756, "grad_norm": 3.3373641967773438, "learning_rate": 2.4786000000000003e-06, "loss": 0.0334, "step": 37608 }, { "epoch": 2.0050647758170284, "grad_norm": 2.2444722652435303, "learning_rate": 2.4784000000000003e-06, "loss": 0.0423, "step": 37609 }, { "epoch": 2.005118089246681, "grad_norm": 0.3094807267189026, "learning_rate": 2.4782000000000003e-06, "loss": 0.0026, "step": 37610 }, { "epoch": 2.0051714026763343, "grad_norm": 2.050419807434082, "learning_rate": 2.4780000000000002e-06, "loss": 0.0118, "step": 37611 }, { "epoch": 2.005224716105987, "grad_norm": 1.700322151184082, "learning_rate": 2.4778000000000002e-06, "loss": 0.0249, "step": 37612 }, { "epoch": 2.00527802953564, "grad_norm": 1.407295823097229, "learning_rate": 2.4776000000000002e-06, "loss": 0.0098, "step": 37613 }, { "epoch": 2.005331342965293, "grad_norm": 1.0483813285827637, "learning_rate": 2.4774e-06, "loss": 0.0047, "step": 37614 }, { "epoch": 2.005384656394946, "grad_norm": 1.6786653995513916, "learning_rate": 2.4772e-06, "loss": 0.0087, "step": 37615 }, { "epoch": 2.005437969824599, "grad_norm": 1.2322933673858643, "learning_rate": 2.477e-06, "loss": 0.0057, "step": 37616 }, { "epoch": 2.0054912832542517, "grad_norm": 2.3923473358154297, "learning_rate": 2.4768e-06, "loss": 0.0251, "step": 37617 }, { "epoch": 2.0055445966839045, "grad_norm": 1.0934258699417114, "learning_rate": 2.4766e-06, "loss": 0.016, "step": 37618 }, { "epoch": 2.0055979101135577, "grad_norm": 0.36544564366340637, "learning_rate": 2.4764e-06, "loss": 0.0024, "step": 37619 }, { "epoch": 2.0056512235432105, "grad_norm": 3.9722251892089844, "learning_rate": 2.4762e-06, "loss": 0.0336, "step": 37620 }, { "epoch": 2.0057045369728637, "grad_norm": 1.4205856323242188, "learning_rate": 2.476e-06, "loss": 0.0105, "step": 37621 }, { "epoch": 2.0057578504025164, "grad_norm": 1.9334279298782349, "learning_rate": 2.4758e-06, "loss": 0.0278, "step": 37622 }, { "epoch": 2.005811163832169, "grad_norm": 1.53982675075531, "learning_rate": 2.4756e-06, "loss": 0.0184, "step": 37623 }, { "epoch": 2.0058644772618224, "grad_norm": 2.1950430870056152, "learning_rate": 2.4754000000000005e-06, "loss": 0.0184, "step": 37624 }, { "epoch": 2.005917790691475, "grad_norm": 2.421787977218628, "learning_rate": 2.4752e-06, "loss": 0.0052, "step": 37625 }, { "epoch": 2.0059711041211283, "grad_norm": 2.2180397510528564, "learning_rate": 2.475e-06, "loss": 0.0221, "step": 37626 }, { "epoch": 2.006024417550781, "grad_norm": 2.728839159011841, "learning_rate": 2.4748000000000004e-06, "loss": 0.043, "step": 37627 }, { "epoch": 2.006077730980434, "grad_norm": 1.983066439628601, "learning_rate": 2.4746000000000004e-06, "loss": 0.0254, "step": 37628 }, { "epoch": 2.006131044410087, "grad_norm": 2.5444889068603516, "learning_rate": 2.4744e-06, "loss": 0.0216, "step": 37629 }, { "epoch": 2.00618435783974, "grad_norm": 2.680392265319824, "learning_rate": 2.4742000000000004e-06, "loss": 0.0416, "step": 37630 }, { "epoch": 2.0062376712693926, "grad_norm": 1.588643193244934, "learning_rate": 2.4740000000000004e-06, "loss": 0.0179, "step": 37631 }, { "epoch": 2.0062909846990458, "grad_norm": 1.8558266162872314, "learning_rate": 2.4738e-06, "loss": 0.0155, "step": 37632 }, { "epoch": 2.0063442981286985, "grad_norm": 0.8917793035507202, "learning_rate": 2.4736000000000003e-06, "loss": 0.0052, "step": 37633 }, { "epoch": 2.0063976115583517, "grad_norm": 1.4163775444030762, "learning_rate": 2.4734000000000003e-06, "loss": 0.0074, "step": 37634 }, { "epoch": 2.0064509249880045, "grad_norm": 0.9461763501167297, "learning_rate": 2.4732000000000003e-06, "loss": 0.006, "step": 37635 }, { "epoch": 2.0065042384176572, "grad_norm": 2.495784044265747, "learning_rate": 2.4730000000000003e-06, "loss": 0.0311, "step": 37636 }, { "epoch": 2.0065575518473104, "grad_norm": 2.110262155532837, "learning_rate": 2.4728000000000003e-06, "loss": 0.0196, "step": 37637 }, { "epoch": 2.006610865276963, "grad_norm": 0.8159860968589783, "learning_rate": 2.4726000000000003e-06, "loss": 0.0107, "step": 37638 }, { "epoch": 2.0066641787066164, "grad_norm": 2.7270610332489014, "learning_rate": 2.4724000000000002e-06, "loss": 0.018, "step": 37639 }, { "epoch": 2.006717492136269, "grad_norm": 1.5828278064727783, "learning_rate": 2.4722000000000002e-06, "loss": 0.009, "step": 37640 }, { "epoch": 2.006770805565922, "grad_norm": 3.072566032409668, "learning_rate": 2.4720000000000002e-06, "loss": 0.0125, "step": 37641 }, { "epoch": 2.006824118995575, "grad_norm": 1.7408512830734253, "learning_rate": 2.4718e-06, "loss": 0.0173, "step": 37642 }, { "epoch": 2.006877432425228, "grad_norm": 3.003833770751953, "learning_rate": 2.4716e-06, "loss": 0.0135, "step": 37643 }, { "epoch": 2.0069307458548806, "grad_norm": 1.6479073762893677, "learning_rate": 2.4714e-06, "loss": 0.0118, "step": 37644 }, { "epoch": 2.006984059284534, "grad_norm": 2.0739362239837646, "learning_rate": 2.4712e-06, "loss": 0.0098, "step": 37645 }, { "epoch": 2.0070373727141866, "grad_norm": 0.9306696057319641, "learning_rate": 2.471e-06, "loss": 0.0057, "step": 37646 }, { "epoch": 2.00709068614384, "grad_norm": 1.3310606479644775, "learning_rate": 2.4708e-06, "loss": 0.0189, "step": 37647 }, { "epoch": 2.0071439995734925, "grad_norm": 0.4615537226200104, "learning_rate": 2.4706e-06, "loss": 0.0034, "step": 37648 }, { "epoch": 2.0071973130031453, "grad_norm": 2.106016159057617, "learning_rate": 2.4704e-06, "loss": 0.0185, "step": 37649 }, { "epoch": 2.0072506264327985, "grad_norm": 3.0120608806610107, "learning_rate": 2.4702e-06, "loss": 0.0232, "step": 37650 }, { "epoch": 2.0073039398624513, "grad_norm": 1.9102990627288818, "learning_rate": 2.47e-06, "loss": 0.0133, "step": 37651 }, { "epoch": 2.0073572532921045, "grad_norm": 1.8383551836013794, "learning_rate": 2.4698e-06, "loss": 0.0106, "step": 37652 }, { "epoch": 2.007410566721757, "grad_norm": 1.4588027000427246, "learning_rate": 2.4696000000000005e-06, "loss": 0.0122, "step": 37653 }, { "epoch": 2.00746388015141, "grad_norm": 0.8823608160018921, "learning_rate": 2.4694e-06, "loss": 0.0214, "step": 37654 }, { "epoch": 2.007517193581063, "grad_norm": 3.3199245929718018, "learning_rate": 2.4692e-06, "loss": 0.0198, "step": 37655 }, { "epoch": 2.007570507010716, "grad_norm": 2.3084027767181396, "learning_rate": 2.4690000000000004e-06, "loss": 0.0241, "step": 37656 }, { "epoch": 2.007623820440369, "grad_norm": 0.28880390524864197, "learning_rate": 2.4688e-06, "loss": 0.0021, "step": 37657 }, { "epoch": 2.007677133870022, "grad_norm": 2.3161182403564453, "learning_rate": 2.4686e-06, "loss": 0.022, "step": 37658 }, { "epoch": 2.0077304472996746, "grad_norm": 2.0303690433502197, "learning_rate": 2.4684000000000004e-06, "loss": 0.0214, "step": 37659 }, { "epoch": 2.007783760729328, "grad_norm": 0.743575930595398, "learning_rate": 2.4682000000000004e-06, "loss": 0.0178, "step": 37660 }, { "epoch": 2.0078370741589806, "grad_norm": 1.9592941999435425, "learning_rate": 2.468e-06, "loss": 0.0126, "step": 37661 }, { "epoch": 2.0078903875886334, "grad_norm": 2.0051488876342773, "learning_rate": 2.4678000000000003e-06, "loss": 0.0388, "step": 37662 }, { "epoch": 2.0079437010182866, "grad_norm": 0.41512447595596313, "learning_rate": 2.4676000000000003e-06, "loss": 0.0035, "step": 37663 }, { "epoch": 2.0079970144479393, "grad_norm": 0.23598815500736237, "learning_rate": 2.4674e-06, "loss": 0.0019, "step": 37664 }, { "epoch": 2.0080503278775925, "grad_norm": 2.2012784481048584, "learning_rate": 2.4672000000000003e-06, "loss": 0.0103, "step": 37665 }, { "epoch": 2.0081036413072453, "grad_norm": 2.5171191692352295, "learning_rate": 2.4670000000000003e-06, "loss": 0.0191, "step": 37666 }, { "epoch": 2.008156954736898, "grad_norm": 3.271650552749634, "learning_rate": 2.4668000000000003e-06, "loss": 0.0431, "step": 37667 }, { "epoch": 2.0082102681665512, "grad_norm": 1.0827580690383911, "learning_rate": 2.4666000000000002e-06, "loss": 0.0129, "step": 37668 }, { "epoch": 2.008263581596204, "grad_norm": 1.7622116804122925, "learning_rate": 2.4664000000000002e-06, "loss": 0.026, "step": 37669 }, { "epoch": 2.008316895025857, "grad_norm": 1.5747010707855225, "learning_rate": 2.4662e-06, "loss": 0.008, "step": 37670 }, { "epoch": 2.00837020845551, "grad_norm": 0.9965158700942993, "learning_rate": 2.466e-06, "loss": 0.0121, "step": 37671 }, { "epoch": 2.0084235218851627, "grad_norm": 1.9219334125518799, "learning_rate": 2.4658e-06, "loss": 0.0151, "step": 37672 }, { "epoch": 2.008476835314816, "grad_norm": 1.9683973789215088, "learning_rate": 2.4656e-06, "loss": 0.0113, "step": 37673 }, { "epoch": 2.0085301487444687, "grad_norm": 1.436223030090332, "learning_rate": 2.4654e-06, "loss": 0.0129, "step": 37674 }, { "epoch": 2.008583462174122, "grad_norm": 1.2797056436538696, "learning_rate": 2.4652e-06, "loss": 0.0101, "step": 37675 }, { "epoch": 2.0086367756037746, "grad_norm": 1.594146490097046, "learning_rate": 2.465e-06, "loss": 0.0123, "step": 37676 }, { "epoch": 2.0086900890334274, "grad_norm": 1.8224974870681763, "learning_rate": 2.4648e-06, "loss": 0.0236, "step": 37677 }, { "epoch": 2.0087434024630806, "grad_norm": 0.7383468151092529, "learning_rate": 2.4646000000000005e-06, "loss": 0.0049, "step": 37678 }, { "epoch": 2.0087967158927333, "grad_norm": 1.873405933380127, "learning_rate": 2.4644e-06, "loss": 0.0248, "step": 37679 }, { "epoch": 2.008850029322386, "grad_norm": 1.316999912261963, "learning_rate": 2.4642e-06, "loss": 0.0244, "step": 37680 }, { "epoch": 2.0089033427520393, "grad_norm": 1.0115634202957153, "learning_rate": 2.4640000000000005e-06, "loss": 0.018, "step": 37681 }, { "epoch": 2.008956656181692, "grad_norm": 1.7738215923309326, "learning_rate": 2.4638e-06, "loss": 0.0146, "step": 37682 }, { "epoch": 2.0090099696113453, "grad_norm": 0.6292365193367004, "learning_rate": 2.4636e-06, "loss": 0.0042, "step": 37683 }, { "epoch": 2.009063283040998, "grad_norm": 3.4716811180114746, "learning_rate": 2.4634000000000004e-06, "loss": 0.0224, "step": 37684 }, { "epoch": 2.0091165964706508, "grad_norm": 2.266798973083496, "learning_rate": 2.4632000000000004e-06, "loss": 0.0354, "step": 37685 }, { "epoch": 2.009169909900304, "grad_norm": 0.6068475246429443, "learning_rate": 2.463e-06, "loss": 0.0036, "step": 37686 }, { "epoch": 2.0092232233299567, "grad_norm": 1.4758912324905396, "learning_rate": 2.4628000000000004e-06, "loss": 0.0078, "step": 37687 }, { "epoch": 2.00927653675961, "grad_norm": 2.459195613861084, "learning_rate": 2.4626000000000004e-06, "loss": 0.0097, "step": 37688 }, { "epoch": 2.0093298501892627, "grad_norm": 2.331922769546509, "learning_rate": 2.4624e-06, "loss": 0.0107, "step": 37689 }, { "epoch": 2.0093831636189154, "grad_norm": 1.7988017797470093, "learning_rate": 2.4622000000000003e-06, "loss": 0.0232, "step": 37690 }, { "epoch": 2.0094364770485686, "grad_norm": 1.3069243431091309, "learning_rate": 2.4620000000000003e-06, "loss": 0.0069, "step": 37691 }, { "epoch": 2.0094897904782214, "grad_norm": 1.0859018564224243, "learning_rate": 2.4618000000000003e-06, "loss": 0.0264, "step": 37692 }, { "epoch": 2.0095431039078746, "grad_norm": 1.2404663562774658, "learning_rate": 2.4616000000000003e-06, "loss": 0.0065, "step": 37693 }, { "epoch": 2.0095964173375274, "grad_norm": 1.9236996173858643, "learning_rate": 2.4614000000000003e-06, "loss": 0.032, "step": 37694 }, { "epoch": 2.00964973076718, "grad_norm": 2.5449984073638916, "learning_rate": 2.4612000000000003e-06, "loss": 0.0196, "step": 37695 }, { "epoch": 2.0097030441968333, "grad_norm": 2.943669319152832, "learning_rate": 2.4610000000000002e-06, "loss": 0.0297, "step": 37696 }, { "epoch": 2.009756357626486, "grad_norm": 1.2475621700286865, "learning_rate": 2.4608000000000002e-06, "loss": 0.0096, "step": 37697 }, { "epoch": 2.009809671056139, "grad_norm": 3.5575971603393555, "learning_rate": 2.4606e-06, "loss": 0.0354, "step": 37698 }, { "epoch": 2.009862984485792, "grad_norm": 1.7591484785079956, "learning_rate": 2.4604e-06, "loss": 0.02, "step": 37699 }, { "epoch": 2.009916297915445, "grad_norm": 1.996659755706787, "learning_rate": 2.4602e-06, "loss": 0.0225, "step": 37700 }, { "epoch": 2.009969611345098, "grad_norm": 1.4957467317581177, "learning_rate": 2.46e-06, "loss": 0.0108, "step": 37701 }, { "epoch": 2.0100229247747508, "grad_norm": 2.5370335578918457, "learning_rate": 2.4598e-06, "loss": 0.0178, "step": 37702 }, { "epoch": 2.0100762382044035, "grad_norm": 2.01698637008667, "learning_rate": 2.4596e-06, "loss": 0.007, "step": 37703 }, { "epoch": 2.0101295516340567, "grad_norm": 1.2682985067367554, "learning_rate": 2.4594e-06, "loss": 0.0075, "step": 37704 }, { "epoch": 2.0101828650637095, "grad_norm": 3.247279167175293, "learning_rate": 2.4592e-06, "loss": 0.0232, "step": 37705 }, { "epoch": 2.0102361784933627, "grad_norm": 1.5616930723190308, "learning_rate": 2.459e-06, "loss": 0.0392, "step": 37706 }, { "epoch": 2.0102894919230154, "grad_norm": 2.259092092514038, "learning_rate": 2.4588e-06, "loss": 0.0205, "step": 37707 }, { "epoch": 2.010342805352668, "grad_norm": 2.9285922050476074, "learning_rate": 2.4586e-06, "loss": 0.0233, "step": 37708 }, { "epoch": 2.0103961187823214, "grad_norm": 2.440319299697876, "learning_rate": 2.4584e-06, "loss": 0.0276, "step": 37709 }, { "epoch": 2.010449432211974, "grad_norm": 0.5327923893928528, "learning_rate": 2.4582000000000005e-06, "loss": 0.0036, "step": 37710 }, { "epoch": 2.0105027456416273, "grad_norm": 2.2311646938323975, "learning_rate": 2.458e-06, "loss": 0.0521, "step": 37711 }, { "epoch": 2.01055605907128, "grad_norm": 2.6241471767425537, "learning_rate": 2.4578e-06, "loss": 0.0076, "step": 37712 }, { "epoch": 2.010609372500933, "grad_norm": 1.1640833616256714, "learning_rate": 2.4576000000000004e-06, "loss": 0.0305, "step": 37713 }, { "epoch": 2.010662685930586, "grad_norm": 3.2188608646392822, "learning_rate": 2.4574e-06, "loss": 0.0351, "step": 37714 }, { "epoch": 2.010715999360239, "grad_norm": 2.273592472076416, "learning_rate": 2.4572e-06, "loss": 0.0088, "step": 37715 }, { "epoch": 2.0107693127898916, "grad_norm": 2.4541077613830566, "learning_rate": 2.4570000000000004e-06, "loss": 0.0404, "step": 37716 }, { "epoch": 2.0108226262195448, "grad_norm": 3.605696678161621, "learning_rate": 2.4568000000000003e-06, "loss": 0.0309, "step": 37717 }, { "epoch": 2.0108759396491975, "grad_norm": 1.1459771394729614, "learning_rate": 2.4566e-06, "loss": 0.022, "step": 37718 }, { "epoch": 2.0109292530788507, "grad_norm": 1.9164741039276123, "learning_rate": 2.4564000000000003e-06, "loss": 0.0059, "step": 37719 }, { "epoch": 2.0109825665085035, "grad_norm": 2.936047315597534, "learning_rate": 2.4562000000000003e-06, "loss": 0.0337, "step": 37720 }, { "epoch": 2.0110358799381562, "grad_norm": 0.7128024697303772, "learning_rate": 2.4560000000000003e-06, "loss": 0.0033, "step": 37721 }, { "epoch": 2.0110891933678094, "grad_norm": 2.2112250328063965, "learning_rate": 2.4558000000000003e-06, "loss": 0.0183, "step": 37722 }, { "epoch": 2.011142506797462, "grad_norm": 2.160212278366089, "learning_rate": 2.4556000000000003e-06, "loss": 0.0157, "step": 37723 }, { "epoch": 2.0111958202271154, "grad_norm": 2.2010011672973633, "learning_rate": 2.4554000000000002e-06, "loss": 0.0106, "step": 37724 }, { "epoch": 2.011249133656768, "grad_norm": 1.7744853496551514, "learning_rate": 2.4552000000000002e-06, "loss": 0.0087, "step": 37725 }, { "epoch": 2.011302447086421, "grad_norm": 0.9662525653839111, "learning_rate": 2.4550000000000002e-06, "loss": 0.0232, "step": 37726 }, { "epoch": 2.011355760516074, "grad_norm": 1.556024193763733, "learning_rate": 2.4548e-06, "loss": 0.0097, "step": 37727 }, { "epoch": 2.011409073945727, "grad_norm": 1.4165483713150024, "learning_rate": 2.4546e-06, "loss": 0.011, "step": 37728 }, { "epoch": 2.01146238737538, "grad_norm": 2.229884147644043, "learning_rate": 2.4544e-06, "loss": 0.0347, "step": 37729 }, { "epoch": 2.011515700805033, "grad_norm": 2.140225648880005, "learning_rate": 2.4542e-06, "loss": 0.041, "step": 37730 }, { "epoch": 2.0115690142346856, "grad_norm": 2.0965561866760254, "learning_rate": 2.454e-06, "loss": 0.0142, "step": 37731 }, { "epoch": 2.011622327664339, "grad_norm": 0.507835865020752, "learning_rate": 2.4538e-06, "loss": 0.005, "step": 37732 }, { "epoch": 2.0116756410939916, "grad_norm": 3.5947699546813965, "learning_rate": 2.4536e-06, "loss": 0.0213, "step": 37733 }, { "epoch": 2.0117289545236443, "grad_norm": 3.1452977657318115, "learning_rate": 2.4534e-06, "loss": 0.0207, "step": 37734 }, { "epoch": 2.0117822679532975, "grad_norm": 2.3344075679779053, "learning_rate": 2.4532000000000005e-06, "loss": 0.012, "step": 37735 }, { "epoch": 2.0118355813829503, "grad_norm": 0.41226866841316223, "learning_rate": 2.453e-06, "loss": 0.0031, "step": 37736 }, { "epoch": 2.0118888948126035, "grad_norm": 1.5979660749435425, "learning_rate": 2.4528e-06, "loss": 0.0237, "step": 37737 }, { "epoch": 2.0119422082422562, "grad_norm": 1.2059732675552368, "learning_rate": 2.4526000000000005e-06, "loss": 0.0196, "step": 37738 }, { "epoch": 2.011995521671909, "grad_norm": 2.907902240753174, "learning_rate": 2.4524000000000004e-06, "loss": 0.0379, "step": 37739 }, { "epoch": 2.012048835101562, "grad_norm": 3.6131906509399414, "learning_rate": 2.4522e-06, "loss": 0.0388, "step": 37740 }, { "epoch": 2.012102148531215, "grad_norm": 0.445009708404541, "learning_rate": 2.4520000000000004e-06, "loss": 0.0035, "step": 37741 }, { "epoch": 2.012155461960868, "grad_norm": 4.368621349334717, "learning_rate": 2.4518000000000004e-06, "loss": 0.0411, "step": 37742 }, { "epoch": 2.012208775390521, "grad_norm": 2.8648862838745117, "learning_rate": 2.4516e-06, "loss": 0.0326, "step": 37743 }, { "epoch": 2.0122620888201737, "grad_norm": 1.1810007095336914, "learning_rate": 2.4514000000000004e-06, "loss": 0.0061, "step": 37744 }, { "epoch": 2.012315402249827, "grad_norm": 2.6227471828460693, "learning_rate": 2.4512000000000003e-06, "loss": 0.0371, "step": 37745 }, { "epoch": 2.0123687156794796, "grad_norm": 2.205169916152954, "learning_rate": 2.4510000000000003e-06, "loss": 0.0452, "step": 37746 }, { "epoch": 2.0124220291091324, "grad_norm": 4.324921607971191, "learning_rate": 2.4508000000000003e-06, "loss": 0.0365, "step": 37747 }, { "epoch": 2.0124753425387856, "grad_norm": 4.534027099609375, "learning_rate": 2.4506000000000003e-06, "loss": 0.0409, "step": 37748 }, { "epoch": 2.0125286559684383, "grad_norm": 1.0204895734786987, "learning_rate": 2.4504000000000003e-06, "loss": 0.0073, "step": 37749 }, { "epoch": 2.0125819693980915, "grad_norm": 2.3719725608825684, "learning_rate": 2.4502000000000003e-06, "loss": 0.0287, "step": 37750 }, { "epoch": 2.0126352828277443, "grad_norm": 1.700278401374817, "learning_rate": 2.4500000000000003e-06, "loss": 0.0134, "step": 37751 }, { "epoch": 2.012688596257397, "grad_norm": 1.9580047130584717, "learning_rate": 2.4498000000000002e-06, "loss": 0.0201, "step": 37752 }, { "epoch": 2.0127419096870502, "grad_norm": 3.2601547241210938, "learning_rate": 2.4496000000000002e-06, "loss": 0.0117, "step": 37753 }, { "epoch": 2.012795223116703, "grad_norm": 0.8964531421661377, "learning_rate": 2.4494000000000002e-06, "loss": 0.0058, "step": 37754 }, { "epoch": 2.012848536546356, "grad_norm": 2.122704029083252, "learning_rate": 2.4492e-06, "loss": 0.0371, "step": 37755 }, { "epoch": 2.012901849976009, "grad_norm": 0.8077322244644165, "learning_rate": 2.449e-06, "loss": 0.0056, "step": 37756 }, { "epoch": 2.0129551634056617, "grad_norm": 3.1659188270568848, "learning_rate": 2.4488e-06, "loss": 0.0347, "step": 37757 }, { "epoch": 2.013008476835315, "grad_norm": 1.9139281511306763, "learning_rate": 2.4486e-06, "loss": 0.0142, "step": 37758 }, { "epoch": 2.0130617902649677, "grad_norm": 2.134467363357544, "learning_rate": 2.4484e-06, "loss": 0.0246, "step": 37759 }, { "epoch": 2.013115103694621, "grad_norm": 1.813396692276001, "learning_rate": 2.4482e-06, "loss": 0.0127, "step": 37760 }, { "epoch": 2.0131684171242736, "grad_norm": 2.2417008876800537, "learning_rate": 2.448e-06, "loss": 0.0247, "step": 37761 }, { "epoch": 2.0132217305539264, "grad_norm": 1.1476629972457886, "learning_rate": 2.4478e-06, "loss": 0.0182, "step": 37762 }, { "epoch": 2.0132750439835796, "grad_norm": 1.26820707321167, "learning_rate": 2.4476e-06, "loss": 0.0079, "step": 37763 }, { "epoch": 2.0133283574132324, "grad_norm": 1.8157858848571777, "learning_rate": 2.4474000000000005e-06, "loss": 0.0134, "step": 37764 }, { "epoch": 2.013381670842885, "grad_norm": 0.5386273860931396, "learning_rate": 2.4472e-06, "loss": 0.0047, "step": 37765 }, { "epoch": 2.0134349842725383, "grad_norm": 2.1472556591033936, "learning_rate": 2.447e-06, "loss": 0.0228, "step": 37766 }, { "epoch": 2.013488297702191, "grad_norm": 2.4232523441314697, "learning_rate": 2.4468000000000004e-06, "loss": 0.0386, "step": 37767 }, { "epoch": 2.0135416111318443, "grad_norm": 2.1611833572387695, "learning_rate": 2.4466e-06, "loss": 0.0144, "step": 37768 }, { "epoch": 2.013594924561497, "grad_norm": 1.5414018630981445, "learning_rate": 2.4464e-06, "loss": 0.0086, "step": 37769 }, { "epoch": 2.01364823799115, "grad_norm": 1.0378950834274292, "learning_rate": 2.4462000000000004e-06, "loss": 0.0072, "step": 37770 }, { "epoch": 2.013701551420803, "grad_norm": 0.2647620439529419, "learning_rate": 2.4460000000000004e-06, "loss": 0.002, "step": 37771 }, { "epoch": 2.0137548648504557, "grad_norm": 1.2436774969100952, "learning_rate": 2.4458e-06, "loss": 0.0298, "step": 37772 }, { "epoch": 2.013808178280109, "grad_norm": 1.3463304042816162, "learning_rate": 2.4456000000000004e-06, "loss": 0.0211, "step": 37773 }, { "epoch": 2.0138614917097617, "grad_norm": 2.323040723800659, "learning_rate": 2.4454000000000003e-06, "loss": 0.0197, "step": 37774 }, { "epoch": 2.0139148051394145, "grad_norm": 1.6113924980163574, "learning_rate": 2.4452e-06, "loss": 0.019, "step": 37775 }, { "epoch": 2.0139681185690677, "grad_norm": 0.11191767454147339, "learning_rate": 2.4450000000000003e-06, "loss": 0.0012, "step": 37776 }, { "epoch": 2.0140214319987204, "grad_norm": 1.1796088218688965, "learning_rate": 2.4448000000000003e-06, "loss": 0.0171, "step": 37777 }, { "epoch": 2.0140747454283736, "grad_norm": 0.9362438321113586, "learning_rate": 2.4446000000000003e-06, "loss": 0.004, "step": 37778 }, { "epoch": 2.0141280588580264, "grad_norm": 2.7766757011413574, "learning_rate": 2.4444000000000003e-06, "loss": 0.0434, "step": 37779 }, { "epoch": 2.014181372287679, "grad_norm": 1.1705349683761597, "learning_rate": 2.4442000000000002e-06, "loss": 0.0089, "step": 37780 }, { "epoch": 2.0142346857173323, "grad_norm": 0.29686439037323, "learning_rate": 2.4440000000000002e-06, "loss": 0.0027, "step": 37781 }, { "epoch": 2.014287999146985, "grad_norm": 0.428537517786026, "learning_rate": 2.4438000000000002e-06, "loss": 0.0035, "step": 37782 }, { "epoch": 2.014341312576638, "grad_norm": 1.1583009958267212, "learning_rate": 2.4436e-06, "loss": 0.021, "step": 37783 }, { "epoch": 2.014394626006291, "grad_norm": 1.2315928936004639, "learning_rate": 2.4434e-06, "loss": 0.0233, "step": 37784 }, { "epoch": 2.014447939435944, "grad_norm": 3.5322282314300537, "learning_rate": 2.4432e-06, "loss": 0.0499, "step": 37785 }, { "epoch": 2.014501252865597, "grad_norm": 0.4054642617702484, "learning_rate": 2.443e-06, "loss": 0.0032, "step": 37786 }, { "epoch": 2.0145545662952498, "grad_norm": 3.029749870300293, "learning_rate": 2.4428e-06, "loss": 0.0442, "step": 37787 }, { "epoch": 2.0146078797249025, "grad_norm": 1.7898519039154053, "learning_rate": 2.4426e-06, "loss": 0.0198, "step": 37788 }, { "epoch": 2.0146611931545557, "grad_norm": 2.4314968585968018, "learning_rate": 2.4424000000000005e-06, "loss": 0.0293, "step": 37789 }, { "epoch": 2.0147145065842085, "grad_norm": 1.4886903762817383, "learning_rate": 2.4422e-06, "loss": 0.0189, "step": 37790 }, { "epoch": 2.0147678200138617, "grad_norm": 1.3983516693115234, "learning_rate": 2.442e-06, "loss": 0.0181, "step": 37791 }, { "epoch": 2.0148211334435144, "grad_norm": 1.641243815422058, "learning_rate": 2.4418000000000005e-06, "loss": 0.0155, "step": 37792 }, { "epoch": 2.014874446873167, "grad_norm": 0.9889720678329468, "learning_rate": 2.4416e-06, "loss": 0.0063, "step": 37793 }, { "epoch": 2.0149277603028204, "grad_norm": 4.327112197875977, "learning_rate": 2.4414e-06, "loss": 0.0095, "step": 37794 }, { "epoch": 2.014981073732473, "grad_norm": 1.9830044507980347, "learning_rate": 2.4412000000000004e-06, "loss": 0.021, "step": 37795 }, { "epoch": 2.0150343871621264, "grad_norm": 2.749128580093384, "learning_rate": 2.4410000000000004e-06, "loss": 0.0113, "step": 37796 }, { "epoch": 2.015087700591779, "grad_norm": 3.9414052963256836, "learning_rate": 2.4408e-06, "loss": 0.0384, "step": 37797 }, { "epoch": 2.015141014021432, "grad_norm": 2.2227296829223633, "learning_rate": 2.4406000000000004e-06, "loss": 0.0347, "step": 37798 }, { "epoch": 2.015194327451085, "grad_norm": 2.3465936183929443, "learning_rate": 2.4404000000000004e-06, "loss": 0.018, "step": 37799 }, { "epoch": 2.015247640880738, "grad_norm": 2.217007637023926, "learning_rate": 2.4402e-06, "loss": 0.0104, "step": 37800 }, { "epoch": 2.0153009543103906, "grad_norm": 3.293351411819458, "learning_rate": 2.4400000000000004e-06, "loss": 0.0166, "step": 37801 }, { "epoch": 2.015354267740044, "grad_norm": 1.220786452293396, "learning_rate": 2.4398000000000003e-06, "loss": 0.0117, "step": 37802 }, { "epoch": 2.0154075811696965, "grad_norm": 1.1681036949157715, "learning_rate": 2.4396000000000003e-06, "loss": 0.0066, "step": 37803 }, { "epoch": 2.0154608945993497, "grad_norm": 3.1097757816314697, "learning_rate": 2.4394000000000003e-06, "loss": 0.0295, "step": 37804 }, { "epoch": 2.0155142080290025, "grad_norm": 1.8439102172851562, "learning_rate": 2.4392000000000003e-06, "loss": 0.0293, "step": 37805 }, { "epoch": 2.0155675214586553, "grad_norm": 1.0812464952468872, "learning_rate": 2.4390000000000003e-06, "loss": 0.0147, "step": 37806 }, { "epoch": 2.0156208348883085, "grad_norm": 1.924088954925537, "learning_rate": 2.4388000000000003e-06, "loss": 0.0191, "step": 37807 }, { "epoch": 2.015674148317961, "grad_norm": 2.2323617935180664, "learning_rate": 2.4386000000000003e-06, "loss": 0.0226, "step": 37808 }, { "epoch": 2.0157274617476144, "grad_norm": 0.9171981811523438, "learning_rate": 2.4384000000000002e-06, "loss": 0.006, "step": 37809 }, { "epoch": 2.015780775177267, "grad_norm": 0.8022749423980713, "learning_rate": 2.4382000000000002e-06, "loss": 0.0042, "step": 37810 }, { "epoch": 2.01583408860692, "grad_norm": 2.0085859298706055, "learning_rate": 2.438e-06, "loss": 0.0078, "step": 37811 }, { "epoch": 2.015887402036573, "grad_norm": 1.4697903394699097, "learning_rate": 2.4378e-06, "loss": 0.0163, "step": 37812 }, { "epoch": 2.015940715466226, "grad_norm": 2.001342535018921, "learning_rate": 2.4376e-06, "loss": 0.0396, "step": 37813 }, { "epoch": 2.015994028895879, "grad_norm": 1.942413330078125, "learning_rate": 2.4374e-06, "loss": 0.0212, "step": 37814 }, { "epoch": 2.016047342325532, "grad_norm": 2.2179272174835205, "learning_rate": 2.4372e-06, "loss": 0.0547, "step": 37815 }, { "epoch": 2.0161006557551846, "grad_norm": 1.2321960926055908, "learning_rate": 2.437e-06, "loss": 0.0077, "step": 37816 }, { "epoch": 2.016153969184838, "grad_norm": 1.910272240638733, "learning_rate": 2.4368e-06, "loss": 0.0248, "step": 37817 }, { "epoch": 2.0162072826144906, "grad_norm": 2.755634069442749, "learning_rate": 2.4366e-06, "loss": 0.0153, "step": 37818 }, { "epoch": 2.0162605960441433, "grad_norm": 2.2562105655670166, "learning_rate": 2.4364e-06, "loss": 0.0237, "step": 37819 }, { "epoch": 2.0163139094737965, "grad_norm": 1.9111641645431519, "learning_rate": 2.4362e-06, "loss": 0.0091, "step": 37820 }, { "epoch": 2.0163672229034493, "grad_norm": 0.527640700340271, "learning_rate": 2.4360000000000005e-06, "loss": 0.003, "step": 37821 }, { "epoch": 2.0164205363331025, "grad_norm": 1.134962558746338, "learning_rate": 2.4358e-06, "loss": 0.0065, "step": 37822 }, { "epoch": 2.0164738497627552, "grad_norm": 6.831202507019043, "learning_rate": 2.4356e-06, "loss": 0.0145, "step": 37823 }, { "epoch": 2.016527163192408, "grad_norm": 1.7124723196029663, "learning_rate": 2.4354000000000004e-06, "loss": 0.0248, "step": 37824 }, { "epoch": 2.016580476622061, "grad_norm": 0.6784241795539856, "learning_rate": 2.4352e-06, "loss": 0.0045, "step": 37825 }, { "epoch": 2.016633790051714, "grad_norm": 1.5616401433944702, "learning_rate": 2.435e-06, "loss": 0.0068, "step": 37826 }, { "epoch": 2.016687103481367, "grad_norm": 3.284055233001709, "learning_rate": 2.4348000000000004e-06, "loss": 0.0221, "step": 37827 }, { "epoch": 2.01674041691102, "grad_norm": 2.387418270111084, "learning_rate": 2.4346000000000004e-06, "loss": 0.0365, "step": 37828 }, { "epoch": 2.0167937303406727, "grad_norm": 2.594597101211548, "learning_rate": 2.4344e-06, "loss": 0.0189, "step": 37829 }, { "epoch": 2.016847043770326, "grad_norm": 1.2815603017807007, "learning_rate": 2.4342000000000003e-06, "loss": 0.0288, "step": 37830 }, { "epoch": 2.0169003571999786, "grad_norm": 0.23682042956352234, "learning_rate": 2.4340000000000003e-06, "loss": 0.0024, "step": 37831 }, { "epoch": 2.016953670629632, "grad_norm": 2.4746735095977783, "learning_rate": 2.4338000000000003e-06, "loss": 0.0249, "step": 37832 }, { "epoch": 2.0170069840592846, "grad_norm": 0.20251064002513885, "learning_rate": 2.4336000000000003e-06, "loss": 0.0019, "step": 37833 }, { "epoch": 2.0170602974889373, "grad_norm": 1.0429834127426147, "learning_rate": 2.4334000000000003e-06, "loss": 0.0059, "step": 37834 }, { "epoch": 2.0171136109185905, "grad_norm": 1.5074787139892578, "learning_rate": 2.4332000000000003e-06, "loss": 0.0105, "step": 37835 }, { "epoch": 2.0171669243482433, "grad_norm": 1.4444984197616577, "learning_rate": 2.4330000000000003e-06, "loss": 0.016, "step": 37836 }, { "epoch": 2.017220237777896, "grad_norm": 0.7743112444877625, "learning_rate": 2.4328000000000002e-06, "loss": 0.0055, "step": 37837 }, { "epoch": 2.0172735512075493, "grad_norm": 1.1770210266113281, "learning_rate": 2.4326000000000002e-06, "loss": 0.0221, "step": 37838 }, { "epoch": 2.017326864637202, "grad_norm": 2.688204288482666, "learning_rate": 2.4324e-06, "loss": 0.013, "step": 37839 }, { "epoch": 2.017380178066855, "grad_norm": 2.95843768119812, "learning_rate": 2.4322e-06, "loss": 0.0162, "step": 37840 }, { "epoch": 2.017433491496508, "grad_norm": 1.4840277433395386, "learning_rate": 2.432e-06, "loss": 0.0192, "step": 37841 }, { "epoch": 2.0174868049261607, "grad_norm": 1.3115813732147217, "learning_rate": 2.4318e-06, "loss": 0.0066, "step": 37842 }, { "epoch": 2.017540118355814, "grad_norm": 2.8816304206848145, "learning_rate": 2.4316e-06, "loss": 0.026, "step": 37843 }, { "epoch": 2.0175934317854667, "grad_norm": 1.5580271482467651, "learning_rate": 2.4314e-06, "loss": 0.0142, "step": 37844 }, { "epoch": 2.01764674521512, "grad_norm": 2.1814732551574707, "learning_rate": 2.4312e-06, "loss": 0.0083, "step": 37845 }, { "epoch": 2.0177000586447726, "grad_norm": 1.4918986558914185, "learning_rate": 2.431e-06, "loss": 0.021, "step": 37846 }, { "epoch": 2.0177533720744254, "grad_norm": 4.49531364440918, "learning_rate": 2.4308e-06, "loss": 0.0103, "step": 37847 }, { "epoch": 2.0178066855040786, "grad_norm": 1.5992835760116577, "learning_rate": 2.4306e-06, "loss": 0.0086, "step": 37848 }, { "epoch": 2.0178599989337314, "grad_norm": 4.188082218170166, "learning_rate": 2.4304e-06, "loss": 0.037, "step": 37849 }, { "epoch": 2.017913312363384, "grad_norm": 2.067686080932617, "learning_rate": 2.4302000000000005e-06, "loss": 0.041, "step": 37850 }, { "epoch": 2.0179666257930373, "grad_norm": 2.3324663639068604, "learning_rate": 2.43e-06, "loss": 0.022, "step": 37851 }, { "epoch": 2.01801993922269, "grad_norm": 1.8823490142822266, "learning_rate": 2.4298e-06, "loss": 0.0317, "step": 37852 }, { "epoch": 2.0180732526523433, "grad_norm": 3.157431125640869, "learning_rate": 2.4296000000000004e-06, "loss": 0.0283, "step": 37853 }, { "epoch": 2.018126566081996, "grad_norm": 2.262606382369995, "learning_rate": 2.4294e-06, "loss": 0.0317, "step": 37854 }, { "epoch": 2.018179879511649, "grad_norm": 0.34778258204460144, "learning_rate": 2.4292e-06, "loss": 0.0025, "step": 37855 }, { "epoch": 2.018233192941302, "grad_norm": 2.1028060913085938, "learning_rate": 2.4290000000000004e-06, "loss": 0.0336, "step": 37856 }, { "epoch": 2.0182865063709547, "grad_norm": 1.263616919517517, "learning_rate": 2.4288000000000004e-06, "loss": 0.0256, "step": 37857 }, { "epoch": 2.018339819800608, "grad_norm": 0.9823987483978271, "learning_rate": 2.4286e-06, "loss": 0.0079, "step": 37858 }, { "epoch": 2.0183931332302607, "grad_norm": 3.330108642578125, "learning_rate": 2.4284000000000003e-06, "loss": 0.0316, "step": 37859 }, { "epoch": 2.0184464466599135, "grad_norm": 1.8313767910003662, "learning_rate": 2.4282000000000003e-06, "loss": 0.033, "step": 37860 }, { "epoch": 2.0184997600895667, "grad_norm": 3.7328102588653564, "learning_rate": 2.428e-06, "loss": 0.015, "step": 37861 }, { "epoch": 2.0185530735192194, "grad_norm": 1.3027292490005493, "learning_rate": 2.4278000000000003e-06, "loss": 0.0057, "step": 37862 }, { "epoch": 2.0186063869488726, "grad_norm": 5.0803375244140625, "learning_rate": 2.4276000000000003e-06, "loss": 0.0265, "step": 37863 }, { "epoch": 2.0186597003785254, "grad_norm": 1.6449029445648193, "learning_rate": 2.4274000000000003e-06, "loss": 0.0197, "step": 37864 }, { "epoch": 2.018713013808178, "grad_norm": 2.427689790725708, "learning_rate": 2.4272000000000002e-06, "loss": 0.012, "step": 37865 }, { "epoch": 2.0187663272378313, "grad_norm": 1.045833706855774, "learning_rate": 2.4270000000000002e-06, "loss": 0.0049, "step": 37866 }, { "epoch": 2.018819640667484, "grad_norm": 5.839596271514893, "learning_rate": 2.4268e-06, "loss": 0.0414, "step": 37867 }, { "epoch": 2.018872954097137, "grad_norm": 4.117335796356201, "learning_rate": 2.4266e-06, "loss": 0.0178, "step": 37868 }, { "epoch": 2.01892626752679, "grad_norm": 1.0708730220794678, "learning_rate": 2.4264e-06, "loss": 0.011, "step": 37869 }, { "epoch": 2.018979580956443, "grad_norm": 1.2512059211730957, "learning_rate": 2.4262e-06, "loss": 0.0116, "step": 37870 }, { "epoch": 2.019032894386096, "grad_norm": 3.4775331020355225, "learning_rate": 2.426e-06, "loss": 0.035, "step": 37871 }, { "epoch": 2.0190862078157488, "grad_norm": 1.3899600505828857, "learning_rate": 2.4258e-06, "loss": 0.0185, "step": 37872 }, { "epoch": 2.0191395212454015, "grad_norm": 1.3407584428787231, "learning_rate": 2.4256e-06, "loss": 0.0061, "step": 37873 }, { "epoch": 2.0191928346750547, "grad_norm": 0.980945348739624, "learning_rate": 2.4254e-06, "loss": 0.0071, "step": 37874 }, { "epoch": 2.0192461481047075, "grad_norm": 2.0483927726745605, "learning_rate": 2.4252000000000005e-06, "loss": 0.0207, "step": 37875 }, { "epoch": 2.0192994615343607, "grad_norm": 0.6314841508865356, "learning_rate": 2.425e-06, "loss": 0.0049, "step": 37876 }, { "epoch": 2.0193527749640134, "grad_norm": 1.82216477394104, "learning_rate": 2.4248e-06, "loss": 0.0166, "step": 37877 }, { "epoch": 2.019406088393666, "grad_norm": 1.1374931335449219, "learning_rate": 2.4246000000000005e-06, "loss": 0.007, "step": 37878 }, { "epoch": 2.0194594018233194, "grad_norm": 1.6168824434280396, "learning_rate": 2.4244e-06, "loss": 0.0344, "step": 37879 }, { "epoch": 2.019512715252972, "grad_norm": 1.4994492530822754, "learning_rate": 2.4242e-06, "loss": 0.0148, "step": 37880 }, { "epoch": 2.0195660286826254, "grad_norm": 0.824781596660614, "learning_rate": 2.4240000000000004e-06, "loss": 0.0065, "step": 37881 }, { "epoch": 2.019619342112278, "grad_norm": 1.6344600915908813, "learning_rate": 2.4238000000000004e-06, "loss": 0.0171, "step": 37882 }, { "epoch": 2.019672655541931, "grad_norm": 3.9875376224517822, "learning_rate": 2.4236e-06, "loss": 0.0192, "step": 37883 }, { "epoch": 2.019725968971584, "grad_norm": 1.4774783849716187, "learning_rate": 2.4234000000000004e-06, "loss": 0.0228, "step": 37884 }, { "epoch": 2.019779282401237, "grad_norm": 2.0753931999206543, "learning_rate": 2.4232000000000004e-06, "loss": 0.0256, "step": 37885 }, { "epoch": 2.0198325958308896, "grad_norm": 1.0841987133026123, "learning_rate": 2.423e-06, "loss": 0.0091, "step": 37886 }, { "epoch": 2.019885909260543, "grad_norm": 1.032637596130371, "learning_rate": 2.4228000000000003e-06, "loss": 0.0134, "step": 37887 }, { "epoch": 2.0199392226901955, "grad_norm": 4.094786167144775, "learning_rate": 2.4226000000000003e-06, "loss": 0.0343, "step": 37888 }, { "epoch": 2.0199925361198487, "grad_norm": 1.5006541013717651, "learning_rate": 2.4224000000000003e-06, "loss": 0.0164, "step": 37889 }, { "epoch": 2.0200458495495015, "grad_norm": 3.713639974594116, "learning_rate": 2.4222000000000003e-06, "loss": 0.0185, "step": 37890 }, { "epoch": 2.0200991629791543, "grad_norm": 0.25914156436920166, "learning_rate": 2.4220000000000003e-06, "loss": 0.0023, "step": 37891 }, { "epoch": 2.0201524764088075, "grad_norm": 1.9817079305648804, "learning_rate": 2.4218000000000003e-06, "loss": 0.0274, "step": 37892 }, { "epoch": 2.02020578983846, "grad_norm": 1.9070990085601807, "learning_rate": 2.4216000000000002e-06, "loss": 0.016, "step": 37893 }, { "epoch": 2.0202591032681134, "grad_norm": 2.6126654148101807, "learning_rate": 2.4214000000000002e-06, "loss": 0.0166, "step": 37894 }, { "epoch": 2.020312416697766, "grad_norm": 2.4781417846679688, "learning_rate": 2.4212000000000002e-06, "loss": 0.0432, "step": 37895 }, { "epoch": 2.020365730127419, "grad_norm": 4.391024112701416, "learning_rate": 2.421e-06, "loss": 0.0377, "step": 37896 }, { "epoch": 2.020419043557072, "grad_norm": 1.4314767122268677, "learning_rate": 2.4208e-06, "loss": 0.0094, "step": 37897 }, { "epoch": 2.020472356986725, "grad_norm": 1.8111788034439087, "learning_rate": 2.4206e-06, "loss": 0.0129, "step": 37898 }, { "epoch": 2.020525670416378, "grad_norm": 3.327338933944702, "learning_rate": 2.4204e-06, "loss": 0.0153, "step": 37899 }, { "epoch": 2.020578983846031, "grad_norm": 1.2292330265045166, "learning_rate": 2.4202e-06, "loss": 0.0273, "step": 37900 }, { "epoch": 2.0206322972756836, "grad_norm": 1.0662147998809814, "learning_rate": 2.42e-06, "loss": 0.0121, "step": 37901 }, { "epoch": 2.020685610705337, "grad_norm": 1.6570457220077515, "learning_rate": 2.4198e-06, "loss": 0.0186, "step": 37902 }, { "epoch": 2.0207389241349896, "grad_norm": 4.0089240074157715, "learning_rate": 2.4196e-06, "loss": 0.0281, "step": 37903 }, { "epoch": 2.0207922375646423, "grad_norm": 1.4925631284713745, "learning_rate": 2.4194e-06, "loss": 0.0195, "step": 37904 }, { "epoch": 2.0208455509942955, "grad_norm": 0.9546477794647217, "learning_rate": 2.4192e-06, "loss": 0.0032, "step": 37905 }, { "epoch": 2.0208988644239483, "grad_norm": 1.4049848318099976, "learning_rate": 2.419e-06, "loss": 0.0091, "step": 37906 }, { "epoch": 2.0209521778536015, "grad_norm": 2.3257510662078857, "learning_rate": 2.4188000000000005e-06, "loss": 0.008, "step": 37907 }, { "epoch": 2.0210054912832542, "grad_norm": 1.9033883810043335, "learning_rate": 2.4186e-06, "loss": 0.0137, "step": 37908 }, { "epoch": 2.021058804712907, "grad_norm": 2.150825023651123, "learning_rate": 2.4184e-06, "loss": 0.0224, "step": 37909 }, { "epoch": 2.02111211814256, "grad_norm": 1.4149771928787231, "learning_rate": 2.4182000000000004e-06, "loss": 0.0325, "step": 37910 }, { "epoch": 2.021165431572213, "grad_norm": 0.9667254090309143, "learning_rate": 2.418e-06, "loss": 0.0193, "step": 37911 }, { "epoch": 2.021218745001866, "grad_norm": 1.930375576019287, "learning_rate": 2.4178e-06, "loss": 0.0057, "step": 37912 }, { "epoch": 2.021272058431519, "grad_norm": 1.8384596109390259, "learning_rate": 2.4176000000000004e-06, "loss": 0.0433, "step": 37913 }, { "epoch": 2.0213253718611717, "grad_norm": 1.3645899295806885, "learning_rate": 2.4174000000000004e-06, "loss": 0.0039, "step": 37914 }, { "epoch": 2.021378685290825, "grad_norm": 0.23648138344287872, "learning_rate": 2.4172e-06, "loss": 0.0023, "step": 37915 }, { "epoch": 2.0214319987204776, "grad_norm": 1.3005903959274292, "learning_rate": 2.4170000000000003e-06, "loss": 0.0061, "step": 37916 }, { "epoch": 2.021485312150131, "grad_norm": 1.900524377822876, "learning_rate": 2.4168000000000003e-06, "loss": 0.0164, "step": 37917 }, { "epoch": 2.0215386255797836, "grad_norm": 1.9098531007766724, "learning_rate": 2.4166000000000003e-06, "loss": 0.0129, "step": 37918 }, { "epoch": 2.0215919390094363, "grad_norm": 1.7800382375717163, "learning_rate": 2.4164000000000003e-06, "loss": 0.0185, "step": 37919 }, { "epoch": 2.0216452524390895, "grad_norm": 1.2453359365463257, "learning_rate": 2.4162000000000003e-06, "loss": 0.0197, "step": 37920 }, { "epoch": 2.0216985658687423, "grad_norm": 2.6067445278167725, "learning_rate": 2.4160000000000002e-06, "loss": 0.0386, "step": 37921 }, { "epoch": 2.021751879298395, "grad_norm": 7.319512367248535, "learning_rate": 2.4158000000000002e-06, "loss": 0.0183, "step": 37922 }, { "epoch": 2.0218051927280483, "grad_norm": 1.861445426940918, "learning_rate": 2.4156000000000002e-06, "loss": 0.0207, "step": 37923 }, { "epoch": 2.021858506157701, "grad_norm": 1.666983962059021, "learning_rate": 2.4154e-06, "loss": 0.0409, "step": 37924 }, { "epoch": 2.021911819587354, "grad_norm": 1.9209603071212769, "learning_rate": 2.4152e-06, "loss": 0.0305, "step": 37925 }, { "epoch": 2.021965133017007, "grad_norm": 1.9671608209609985, "learning_rate": 2.415e-06, "loss": 0.024, "step": 37926 }, { "epoch": 2.0220184464466597, "grad_norm": 2.402337074279785, "learning_rate": 2.4148e-06, "loss": 0.0211, "step": 37927 }, { "epoch": 2.022071759876313, "grad_norm": 0.4675268828868866, "learning_rate": 2.4146e-06, "loss": 0.0031, "step": 37928 }, { "epoch": 2.0221250733059657, "grad_norm": 1.3041529655456543, "learning_rate": 2.4144e-06, "loss": 0.0131, "step": 37929 }, { "epoch": 2.022178386735619, "grad_norm": 2.661318778991699, "learning_rate": 2.4142e-06, "loss": 0.0459, "step": 37930 }, { "epoch": 2.0222317001652717, "grad_norm": 5.797555446624756, "learning_rate": 2.414e-06, "loss": 0.0342, "step": 37931 }, { "epoch": 2.0222850135949244, "grad_norm": 1.264296054840088, "learning_rate": 2.4138000000000005e-06, "loss": 0.0148, "step": 37932 }, { "epoch": 2.0223383270245776, "grad_norm": 2.028667449951172, "learning_rate": 2.4136e-06, "loss": 0.012, "step": 37933 }, { "epoch": 2.0223916404542304, "grad_norm": 2.435046672821045, "learning_rate": 2.4134e-06, "loss": 0.0295, "step": 37934 }, { "epoch": 2.0224449538838836, "grad_norm": 0.8846035599708557, "learning_rate": 2.4132000000000005e-06, "loss": 0.0043, "step": 37935 }, { "epoch": 2.0224982673135363, "grad_norm": 1.0394574403762817, "learning_rate": 2.413e-06, "loss": 0.007, "step": 37936 }, { "epoch": 2.022551580743189, "grad_norm": 2.903407573699951, "learning_rate": 2.4128e-06, "loss": 0.076, "step": 37937 }, { "epoch": 2.0226048941728423, "grad_norm": 1.9639054536819458, "learning_rate": 2.4126000000000004e-06, "loss": 0.0113, "step": 37938 }, { "epoch": 2.022658207602495, "grad_norm": 3.9639737606048584, "learning_rate": 2.4124000000000004e-06, "loss": 0.0306, "step": 37939 }, { "epoch": 2.022711521032148, "grad_norm": 1.7980715036392212, "learning_rate": 2.4122e-06, "loss": 0.022, "step": 37940 }, { "epoch": 2.022764834461801, "grad_norm": 0.9436585307121277, "learning_rate": 2.4120000000000004e-06, "loss": 0.0053, "step": 37941 }, { "epoch": 2.0228181478914538, "grad_norm": 1.4678369760513306, "learning_rate": 2.4118000000000004e-06, "loss": 0.0152, "step": 37942 }, { "epoch": 2.022871461321107, "grad_norm": 0.5483549237251282, "learning_rate": 2.4116000000000003e-06, "loss": 0.0048, "step": 37943 }, { "epoch": 2.0229247747507597, "grad_norm": 1.092794418334961, "learning_rate": 2.4114000000000003e-06, "loss": 0.0113, "step": 37944 }, { "epoch": 2.0229780881804125, "grad_norm": 1.216103434562683, "learning_rate": 2.4112000000000003e-06, "loss": 0.0088, "step": 37945 }, { "epoch": 2.0230314016100657, "grad_norm": 0.7272494435310364, "learning_rate": 2.4110000000000003e-06, "loss": 0.0034, "step": 37946 }, { "epoch": 2.0230847150397184, "grad_norm": 0.6993880867958069, "learning_rate": 2.4108000000000003e-06, "loss": 0.0157, "step": 37947 }, { "epoch": 2.0231380284693716, "grad_norm": 2.2303168773651123, "learning_rate": 2.4106000000000003e-06, "loss": 0.0332, "step": 37948 }, { "epoch": 2.0231913418990244, "grad_norm": 1.3127946853637695, "learning_rate": 2.4104000000000002e-06, "loss": 0.02, "step": 37949 }, { "epoch": 2.023244655328677, "grad_norm": 1.5910533666610718, "learning_rate": 2.4102000000000002e-06, "loss": 0.0218, "step": 37950 }, { "epoch": 2.0232979687583303, "grad_norm": 2.134704351425171, "learning_rate": 2.4100000000000002e-06, "loss": 0.0226, "step": 37951 }, { "epoch": 2.023351282187983, "grad_norm": 1.6182851791381836, "learning_rate": 2.4098e-06, "loss": 0.0167, "step": 37952 }, { "epoch": 2.0234045956176363, "grad_norm": 1.3507037162780762, "learning_rate": 2.4096e-06, "loss": 0.0065, "step": 37953 }, { "epoch": 2.023457909047289, "grad_norm": 1.9209835529327393, "learning_rate": 2.4094e-06, "loss": 0.0267, "step": 37954 }, { "epoch": 2.023511222476942, "grad_norm": 0.5879570841789246, "learning_rate": 2.4092e-06, "loss": 0.0051, "step": 37955 }, { "epoch": 2.023564535906595, "grad_norm": 3.2592153549194336, "learning_rate": 2.409e-06, "loss": 0.0363, "step": 37956 }, { "epoch": 2.0236178493362478, "grad_norm": 2.6754915714263916, "learning_rate": 2.4088e-06, "loss": 0.022, "step": 37957 }, { "epoch": 2.0236711627659005, "grad_norm": 4.0536370277404785, "learning_rate": 2.4086e-06, "loss": 0.0556, "step": 37958 }, { "epoch": 2.0237244761955537, "grad_norm": 2.8319356441497803, "learning_rate": 2.4084e-06, "loss": 0.0121, "step": 37959 }, { "epoch": 2.0237777896252065, "grad_norm": 1.4311884641647339, "learning_rate": 2.4082e-06, "loss": 0.0077, "step": 37960 }, { "epoch": 2.0238311030548597, "grad_norm": 2.2271556854248047, "learning_rate": 2.408e-06, "loss": 0.0235, "step": 37961 }, { "epoch": 2.0238844164845124, "grad_norm": 2.200380325317383, "learning_rate": 2.4078e-06, "loss": 0.01, "step": 37962 }, { "epoch": 2.023937729914165, "grad_norm": 2.294916868209839, "learning_rate": 2.4076e-06, "loss": 0.0201, "step": 37963 }, { "epoch": 2.0239910433438184, "grad_norm": 2.0493881702423096, "learning_rate": 2.4074000000000004e-06, "loss": 0.0173, "step": 37964 }, { "epoch": 2.024044356773471, "grad_norm": 0.6144147515296936, "learning_rate": 2.4072e-06, "loss": 0.0048, "step": 37965 }, { "epoch": 2.0240976702031244, "grad_norm": 1.6306480169296265, "learning_rate": 2.407e-06, "loss": 0.0261, "step": 37966 }, { "epoch": 2.024150983632777, "grad_norm": 2.260132074356079, "learning_rate": 2.4068000000000004e-06, "loss": 0.0226, "step": 37967 }, { "epoch": 2.02420429706243, "grad_norm": 1.1977717876434326, "learning_rate": 2.4066000000000004e-06, "loss": 0.0063, "step": 37968 }, { "epoch": 2.024257610492083, "grad_norm": 1.393995761871338, "learning_rate": 2.4064e-06, "loss": 0.0174, "step": 37969 }, { "epoch": 2.024310923921736, "grad_norm": 2.4083874225616455, "learning_rate": 2.4062000000000004e-06, "loss": 0.029, "step": 37970 }, { "epoch": 2.0243642373513886, "grad_norm": 1.9668822288513184, "learning_rate": 2.4060000000000003e-06, "loss": 0.0087, "step": 37971 }, { "epoch": 2.024417550781042, "grad_norm": 1.4345978498458862, "learning_rate": 2.4058e-06, "loss": 0.0265, "step": 37972 }, { "epoch": 2.0244708642106946, "grad_norm": 1.9315656423568726, "learning_rate": 2.4056000000000003e-06, "loss": 0.0141, "step": 37973 }, { "epoch": 2.0245241776403478, "grad_norm": 1.5377864837646484, "learning_rate": 2.4054000000000003e-06, "loss": 0.0374, "step": 37974 }, { "epoch": 2.0245774910700005, "grad_norm": 2.2737762928009033, "learning_rate": 2.4052000000000003e-06, "loss": 0.0223, "step": 37975 }, { "epoch": 2.0246308044996533, "grad_norm": 1.3756498098373413, "learning_rate": 2.4050000000000003e-06, "loss": 0.0129, "step": 37976 }, { "epoch": 2.0246841179293065, "grad_norm": 1.2457568645477295, "learning_rate": 2.4048000000000003e-06, "loss": 0.0052, "step": 37977 }, { "epoch": 2.0247374313589592, "grad_norm": 3.609807014465332, "learning_rate": 2.4046000000000002e-06, "loss": 0.0304, "step": 37978 }, { "epoch": 2.0247907447886124, "grad_norm": 1.0880669355392456, "learning_rate": 2.4044000000000002e-06, "loss": 0.006, "step": 37979 }, { "epoch": 2.024844058218265, "grad_norm": 2.710012197494507, "learning_rate": 2.4042e-06, "loss": 0.0087, "step": 37980 }, { "epoch": 2.024897371647918, "grad_norm": 0.9457641839981079, "learning_rate": 2.404e-06, "loss": 0.0061, "step": 37981 }, { "epoch": 2.024950685077571, "grad_norm": 1.8918176889419556, "learning_rate": 2.4038e-06, "loss": 0.0136, "step": 37982 }, { "epoch": 2.025003998507224, "grad_norm": 2.4573147296905518, "learning_rate": 2.4036e-06, "loss": 0.0203, "step": 37983 }, { "epoch": 2.025057311936877, "grad_norm": 1.6739683151245117, "learning_rate": 2.4034e-06, "loss": 0.0144, "step": 37984 }, { "epoch": 2.02511062536653, "grad_norm": 0.6120180487632751, "learning_rate": 2.4032e-06, "loss": 0.0051, "step": 37985 }, { "epoch": 2.0251639387961826, "grad_norm": 2.6444432735443115, "learning_rate": 2.4030000000000005e-06, "loss": 0.0169, "step": 37986 }, { "epoch": 2.025217252225836, "grad_norm": 1.5411754846572876, "learning_rate": 2.4028e-06, "loss": 0.0194, "step": 37987 }, { "epoch": 2.0252705656554886, "grad_norm": 0.7004634141921997, "learning_rate": 2.4026e-06, "loss": 0.0046, "step": 37988 }, { "epoch": 2.0253238790851413, "grad_norm": 1.1699780225753784, "learning_rate": 2.4024000000000005e-06, "loss": 0.0224, "step": 37989 }, { "epoch": 2.0253771925147945, "grad_norm": 0.5358380675315857, "learning_rate": 2.4022e-06, "loss": 0.004, "step": 37990 }, { "epoch": 2.0254305059444473, "grad_norm": 0.40882933139801025, "learning_rate": 2.402e-06, "loss": 0.0039, "step": 37991 }, { "epoch": 2.0254838193741005, "grad_norm": 2.6766040325164795, "learning_rate": 2.4018000000000005e-06, "loss": 0.028, "step": 37992 }, { "epoch": 2.0255371328037532, "grad_norm": 2.631976366043091, "learning_rate": 2.4016000000000004e-06, "loss": 0.021, "step": 37993 }, { "epoch": 2.025590446233406, "grad_norm": 2.3375608921051025, "learning_rate": 2.4014e-06, "loss": 0.0118, "step": 37994 }, { "epoch": 2.025643759663059, "grad_norm": 1.2491123676300049, "learning_rate": 2.4012000000000004e-06, "loss": 0.0162, "step": 37995 }, { "epoch": 2.025697073092712, "grad_norm": 1.9025415182113647, "learning_rate": 2.4010000000000004e-06, "loss": 0.0192, "step": 37996 }, { "epoch": 2.025750386522365, "grad_norm": 0.6880384683609009, "learning_rate": 2.4008e-06, "loss": 0.005, "step": 37997 }, { "epoch": 2.025803699952018, "grad_norm": 0.9379808902740479, "learning_rate": 2.4006000000000004e-06, "loss": 0.007, "step": 37998 }, { "epoch": 2.0258570133816707, "grad_norm": 2.1341729164123535, "learning_rate": 2.4004000000000003e-06, "loss": 0.0263, "step": 37999 }, { "epoch": 2.025910326811324, "grad_norm": 1.0003167390823364, "learning_rate": 2.4002000000000003e-06, "loss": 0.0138, "step": 38000 }, { "epoch": 2.0259636402409766, "grad_norm": 1.1605347394943237, "learning_rate": 2.4000000000000003e-06, "loss": 0.0154, "step": 38001 }, { "epoch": 2.02601695367063, "grad_norm": 0.8527666926383972, "learning_rate": 2.3998000000000003e-06, "loss": 0.0171, "step": 38002 }, { "epoch": 2.0260702671002826, "grad_norm": 1.8497962951660156, "learning_rate": 2.3996000000000003e-06, "loss": 0.0485, "step": 38003 }, { "epoch": 2.0261235805299354, "grad_norm": 2.310814380645752, "learning_rate": 2.3994000000000003e-06, "loss": 0.034, "step": 38004 }, { "epoch": 2.0261768939595886, "grad_norm": 1.899733304977417, "learning_rate": 2.3992000000000003e-06, "loss": 0.0155, "step": 38005 }, { "epoch": 2.0262302073892413, "grad_norm": 1.2985811233520508, "learning_rate": 2.3990000000000002e-06, "loss": 0.0113, "step": 38006 }, { "epoch": 2.026283520818894, "grad_norm": 1.0752357244491577, "learning_rate": 2.3988000000000002e-06, "loss": 0.0106, "step": 38007 }, { "epoch": 2.0263368342485473, "grad_norm": 1.6822844743728638, "learning_rate": 2.3986e-06, "loss": 0.0285, "step": 38008 }, { "epoch": 2.0263901476782, "grad_norm": 2.183603048324585, "learning_rate": 2.3984e-06, "loss": 0.0121, "step": 38009 }, { "epoch": 2.0264434611078532, "grad_norm": 0.744408905506134, "learning_rate": 2.3982e-06, "loss": 0.0035, "step": 38010 }, { "epoch": 2.026496774537506, "grad_norm": 1.9700030088424683, "learning_rate": 2.398e-06, "loss": 0.0205, "step": 38011 }, { "epoch": 2.0265500879671587, "grad_norm": 1.5581716299057007, "learning_rate": 2.3978e-06, "loss": 0.0084, "step": 38012 }, { "epoch": 2.026603401396812, "grad_norm": 1.6629283428192139, "learning_rate": 2.3976e-06, "loss": 0.0104, "step": 38013 }, { "epoch": 2.0266567148264647, "grad_norm": 1.9895182847976685, "learning_rate": 2.3974e-06, "loss": 0.0268, "step": 38014 }, { "epoch": 2.026710028256118, "grad_norm": 1.9708597660064697, "learning_rate": 2.3972e-06, "loss": 0.0176, "step": 38015 }, { "epoch": 2.0267633416857707, "grad_norm": 1.8240790367126465, "learning_rate": 2.397e-06, "loss": 0.0319, "step": 38016 }, { "epoch": 2.0268166551154234, "grad_norm": 1.6438632011413574, "learning_rate": 2.3968e-06, "loss": 0.0075, "step": 38017 }, { "epoch": 2.0268699685450766, "grad_norm": 1.9220117330551147, "learning_rate": 2.3966000000000005e-06, "loss": 0.0245, "step": 38018 }, { "epoch": 2.0269232819747294, "grad_norm": 2.568258285522461, "learning_rate": 2.3964e-06, "loss": 0.0411, "step": 38019 }, { "epoch": 2.0269765954043826, "grad_norm": 2.3662850856781006, "learning_rate": 2.3962e-06, "loss": 0.0407, "step": 38020 }, { "epoch": 2.0270299088340353, "grad_norm": 1.0540653467178345, "learning_rate": 2.3960000000000004e-06, "loss": 0.0087, "step": 38021 }, { "epoch": 2.027083222263688, "grad_norm": 1.7270721197128296, "learning_rate": 2.3958e-06, "loss": 0.0136, "step": 38022 }, { "epoch": 2.0271365356933413, "grad_norm": 1.0264655351638794, "learning_rate": 2.3956e-06, "loss": 0.0055, "step": 38023 }, { "epoch": 2.027189849122994, "grad_norm": 3.2869253158569336, "learning_rate": 2.3954000000000004e-06, "loss": 0.0194, "step": 38024 }, { "epoch": 2.027243162552647, "grad_norm": 1.7132372856140137, "learning_rate": 2.3952000000000004e-06, "loss": 0.0114, "step": 38025 }, { "epoch": 2.0272964759823, "grad_norm": 1.5341472625732422, "learning_rate": 2.395e-06, "loss": 0.0204, "step": 38026 }, { "epoch": 2.0273497894119528, "grad_norm": 3.099688768386841, "learning_rate": 2.3948000000000003e-06, "loss": 0.036, "step": 38027 }, { "epoch": 2.027403102841606, "grad_norm": 0.812232494354248, "learning_rate": 2.3946000000000003e-06, "loss": 0.0179, "step": 38028 }, { "epoch": 2.0274564162712587, "grad_norm": 0.8108469843864441, "learning_rate": 2.3944e-06, "loss": 0.0048, "step": 38029 }, { "epoch": 2.0275097297009115, "grad_norm": 0.43079838156700134, "learning_rate": 2.3942000000000003e-06, "loss": 0.0032, "step": 38030 }, { "epoch": 2.0275630431305647, "grad_norm": 2.4461240768432617, "learning_rate": 2.3940000000000003e-06, "loss": 0.0157, "step": 38031 }, { "epoch": 2.0276163565602174, "grad_norm": 1.9336172342300415, "learning_rate": 2.3938000000000003e-06, "loss": 0.0222, "step": 38032 }, { "epoch": 2.0276696699898706, "grad_norm": 0.8260687589645386, "learning_rate": 2.3936000000000003e-06, "loss": 0.0036, "step": 38033 }, { "epoch": 2.0277229834195234, "grad_norm": 1.7909796237945557, "learning_rate": 2.3934000000000002e-06, "loss": 0.0173, "step": 38034 }, { "epoch": 2.027776296849176, "grad_norm": 1.6893460750579834, "learning_rate": 2.3932000000000002e-06, "loss": 0.0089, "step": 38035 }, { "epoch": 2.0278296102788294, "grad_norm": 3.3586130142211914, "learning_rate": 2.3930000000000002e-06, "loss": 0.0386, "step": 38036 }, { "epoch": 2.027882923708482, "grad_norm": 4.314655303955078, "learning_rate": 2.3928e-06, "loss": 0.028, "step": 38037 }, { "epoch": 2.0279362371381353, "grad_norm": 0.28261616826057434, "learning_rate": 2.3926e-06, "loss": 0.0028, "step": 38038 }, { "epoch": 2.027989550567788, "grad_norm": 1.5421372652053833, "learning_rate": 2.3924e-06, "loss": 0.0084, "step": 38039 }, { "epoch": 2.028042863997441, "grad_norm": 1.7162820100784302, "learning_rate": 2.3922e-06, "loss": 0.0193, "step": 38040 }, { "epoch": 2.028096177427094, "grad_norm": 1.5044150352478027, "learning_rate": 2.392e-06, "loss": 0.0126, "step": 38041 }, { "epoch": 2.028149490856747, "grad_norm": 2.0593512058258057, "learning_rate": 2.3918e-06, "loss": 0.0111, "step": 38042 }, { "epoch": 2.0282028042863995, "grad_norm": 2.242692470550537, "learning_rate": 2.3916e-06, "loss": 0.0228, "step": 38043 }, { "epoch": 2.0282561177160527, "grad_norm": 1.5413410663604736, "learning_rate": 2.3914e-06, "loss": 0.0243, "step": 38044 }, { "epoch": 2.0283094311457055, "grad_norm": 3.352742910385132, "learning_rate": 2.3912e-06, "loss": 0.0098, "step": 38045 }, { "epoch": 2.0283627445753587, "grad_norm": 2.5092616081237793, "learning_rate": 2.391e-06, "loss": 0.0292, "step": 38046 }, { "epoch": 2.0284160580050115, "grad_norm": 1.2895623445510864, "learning_rate": 2.3908e-06, "loss": 0.0173, "step": 38047 }, { "epoch": 2.028469371434664, "grad_norm": 3.4482202529907227, "learning_rate": 2.3906e-06, "loss": 0.0411, "step": 38048 }, { "epoch": 2.0285226848643174, "grad_norm": 1.072468876838684, "learning_rate": 2.3904e-06, "loss": 0.0059, "step": 38049 }, { "epoch": 2.02857599829397, "grad_norm": 4.346550464630127, "learning_rate": 2.3902000000000004e-06, "loss": 0.0183, "step": 38050 }, { "epoch": 2.0286293117236234, "grad_norm": 1.3702391386032104, "learning_rate": 2.39e-06, "loss": 0.0077, "step": 38051 }, { "epoch": 2.028682625153276, "grad_norm": 2.873932123184204, "learning_rate": 2.3898e-06, "loss": 0.0328, "step": 38052 }, { "epoch": 2.028735938582929, "grad_norm": 1.3299529552459717, "learning_rate": 2.3896000000000004e-06, "loss": 0.0058, "step": 38053 }, { "epoch": 2.028789252012582, "grad_norm": 1.8010953664779663, "learning_rate": 2.3894000000000004e-06, "loss": 0.0292, "step": 38054 }, { "epoch": 2.028842565442235, "grad_norm": 1.711768388748169, "learning_rate": 2.3892e-06, "loss": 0.0199, "step": 38055 }, { "epoch": 2.028895878871888, "grad_norm": 3.0475869178771973, "learning_rate": 2.3890000000000003e-06, "loss": 0.0369, "step": 38056 }, { "epoch": 2.028949192301541, "grad_norm": 1.7442803382873535, "learning_rate": 2.3888000000000003e-06, "loss": 0.016, "step": 38057 }, { "epoch": 2.0290025057311936, "grad_norm": 4.607677936553955, "learning_rate": 2.3886e-06, "loss": 0.0346, "step": 38058 }, { "epoch": 2.0290558191608468, "grad_norm": 3.0788984298706055, "learning_rate": 2.3884000000000003e-06, "loss": 0.0241, "step": 38059 }, { "epoch": 2.0291091325904995, "grad_norm": 1.574647068977356, "learning_rate": 2.3882000000000003e-06, "loss": 0.0184, "step": 38060 }, { "epoch": 2.0291624460201523, "grad_norm": 0.797535240650177, "learning_rate": 2.3880000000000003e-06, "loss": 0.0061, "step": 38061 }, { "epoch": 2.0292157594498055, "grad_norm": 3.8031868934631348, "learning_rate": 2.3878000000000002e-06, "loss": 0.0175, "step": 38062 }, { "epoch": 2.0292690728794582, "grad_norm": 3.172973155975342, "learning_rate": 2.3876000000000002e-06, "loss": 0.0211, "step": 38063 }, { "epoch": 2.0293223863091114, "grad_norm": 0.9822993278503418, "learning_rate": 2.3874000000000002e-06, "loss": 0.0089, "step": 38064 }, { "epoch": 2.029375699738764, "grad_norm": 1.7786494493484497, "learning_rate": 2.3872e-06, "loss": 0.0403, "step": 38065 }, { "epoch": 2.029429013168417, "grad_norm": 0.7641165256500244, "learning_rate": 2.387e-06, "loss": 0.0126, "step": 38066 }, { "epoch": 2.02948232659807, "grad_norm": 1.6597838401794434, "learning_rate": 2.3868e-06, "loss": 0.008, "step": 38067 }, { "epoch": 2.029535640027723, "grad_norm": 0.938219428062439, "learning_rate": 2.3866e-06, "loss": 0.0196, "step": 38068 }, { "epoch": 2.029588953457376, "grad_norm": 2.6596648693084717, "learning_rate": 2.3864e-06, "loss": 0.0266, "step": 38069 }, { "epoch": 2.029642266887029, "grad_norm": 2.7246146202087402, "learning_rate": 2.3862e-06, "loss": 0.032, "step": 38070 }, { "epoch": 2.0296955803166816, "grad_norm": 1.4839612245559692, "learning_rate": 2.386e-06, "loss": 0.012, "step": 38071 }, { "epoch": 2.029748893746335, "grad_norm": 0.9870781302452087, "learning_rate": 2.3858e-06, "loss": 0.0059, "step": 38072 }, { "epoch": 2.0298022071759876, "grad_norm": 2.131114959716797, "learning_rate": 2.3856e-06, "loss": 0.0364, "step": 38073 }, { "epoch": 2.029855520605641, "grad_norm": 3.192204713821411, "learning_rate": 2.3854e-06, "loss": 0.0231, "step": 38074 }, { "epoch": 2.0299088340352935, "grad_norm": 2.019242286682129, "learning_rate": 2.3852000000000005e-06, "loss": 0.0212, "step": 38075 }, { "epoch": 2.0299621474649463, "grad_norm": 2.361591339111328, "learning_rate": 2.385e-06, "loss": 0.0243, "step": 38076 }, { "epoch": 2.0300154608945995, "grad_norm": 1.6387393474578857, "learning_rate": 2.3848e-06, "loss": 0.0188, "step": 38077 }, { "epoch": 2.0300687743242523, "grad_norm": 0.43498390913009644, "learning_rate": 2.3846000000000004e-06, "loss": 0.004, "step": 38078 }, { "epoch": 2.030122087753905, "grad_norm": 0.9814514517784119, "learning_rate": 2.3844000000000004e-06, "loss": 0.0057, "step": 38079 }, { "epoch": 2.030175401183558, "grad_norm": 2.2983474731445312, "learning_rate": 2.3842e-06, "loss": 0.0157, "step": 38080 }, { "epoch": 2.030228714613211, "grad_norm": 3.863870620727539, "learning_rate": 2.3840000000000004e-06, "loss": 0.0433, "step": 38081 }, { "epoch": 2.030282028042864, "grad_norm": 1.941423773765564, "learning_rate": 2.3838000000000004e-06, "loss": 0.0258, "step": 38082 }, { "epoch": 2.030335341472517, "grad_norm": 3.0393738746643066, "learning_rate": 2.3836e-06, "loss": 0.0228, "step": 38083 }, { "epoch": 2.0303886549021697, "grad_norm": 3.3292441368103027, "learning_rate": 2.3834000000000003e-06, "loss": 0.0168, "step": 38084 }, { "epoch": 2.030441968331823, "grad_norm": 0.9951634407043457, "learning_rate": 2.3832000000000003e-06, "loss": 0.013, "step": 38085 }, { "epoch": 2.0304952817614756, "grad_norm": 1.1883389949798584, "learning_rate": 2.3830000000000003e-06, "loss": 0.0236, "step": 38086 }, { "epoch": 2.030548595191129, "grad_norm": 1.0827950239181519, "learning_rate": 2.3828000000000003e-06, "loss": 0.007, "step": 38087 }, { "epoch": 2.0306019086207816, "grad_norm": 1.0987271070480347, "learning_rate": 2.3826000000000003e-06, "loss": 0.0077, "step": 38088 }, { "epoch": 2.0306552220504344, "grad_norm": 1.3413194417953491, "learning_rate": 2.3824000000000003e-06, "loss": 0.0081, "step": 38089 }, { "epoch": 2.0307085354800876, "grad_norm": 1.4857177734375, "learning_rate": 2.3822000000000002e-06, "loss": 0.0156, "step": 38090 }, { "epoch": 2.0307618489097403, "grad_norm": 3.6595003604888916, "learning_rate": 2.3820000000000002e-06, "loss": 0.0114, "step": 38091 }, { "epoch": 2.030815162339393, "grad_norm": 1.8341816663742065, "learning_rate": 2.3818000000000002e-06, "loss": 0.026, "step": 38092 }, { "epoch": 2.0308684757690463, "grad_norm": 3.059732437133789, "learning_rate": 2.3816e-06, "loss": 0.0122, "step": 38093 }, { "epoch": 2.030921789198699, "grad_norm": 0.582207202911377, "learning_rate": 2.3814e-06, "loss": 0.0042, "step": 38094 }, { "epoch": 2.0309751026283522, "grad_norm": 0.5962238311767578, "learning_rate": 2.3812e-06, "loss": 0.0039, "step": 38095 }, { "epoch": 2.031028416058005, "grad_norm": 2.1524064540863037, "learning_rate": 2.381e-06, "loss": 0.0229, "step": 38096 }, { "epoch": 2.0310817294876577, "grad_norm": 1.2938019037246704, "learning_rate": 2.3808e-06, "loss": 0.0191, "step": 38097 }, { "epoch": 2.031135042917311, "grad_norm": 3.0832552909851074, "learning_rate": 2.3806e-06, "loss": 0.0284, "step": 38098 }, { "epoch": 2.0311883563469637, "grad_norm": 0.255344420671463, "learning_rate": 2.3804e-06, "loss": 0.0026, "step": 38099 }, { "epoch": 2.031241669776617, "grad_norm": 3.957210063934326, "learning_rate": 2.3802e-06, "loss": 0.0319, "step": 38100 }, { "epoch": 2.0312949832062697, "grad_norm": 1.3691192865371704, "learning_rate": 2.38e-06, "loss": 0.0097, "step": 38101 }, { "epoch": 2.0313482966359224, "grad_norm": 0.6343227028846741, "learning_rate": 2.3798e-06, "loss": 0.0041, "step": 38102 }, { "epoch": 2.0314016100655756, "grad_norm": 0.3511869013309479, "learning_rate": 2.3796e-06, "loss": 0.0024, "step": 38103 }, { "epoch": 2.0314549234952284, "grad_norm": 1.1460055112838745, "learning_rate": 2.3794000000000005e-06, "loss": 0.0059, "step": 38104 }, { "epoch": 2.0315082369248816, "grad_norm": 1.9423335790634155, "learning_rate": 2.3792e-06, "loss": 0.0343, "step": 38105 }, { "epoch": 2.0315615503545343, "grad_norm": 1.0715769529342651, "learning_rate": 2.379e-06, "loss": 0.0217, "step": 38106 }, { "epoch": 2.031614863784187, "grad_norm": 1.7543749809265137, "learning_rate": 2.3788000000000004e-06, "loss": 0.015, "step": 38107 }, { "epoch": 2.0316681772138403, "grad_norm": 0.9921100735664368, "learning_rate": 2.3786e-06, "loss": 0.0042, "step": 38108 }, { "epoch": 2.031721490643493, "grad_norm": 2.8729894161224365, "learning_rate": 2.3784e-06, "loss": 0.0199, "step": 38109 }, { "epoch": 2.031774804073146, "grad_norm": 1.1423317193984985, "learning_rate": 2.3782000000000004e-06, "loss": 0.0242, "step": 38110 }, { "epoch": 2.031828117502799, "grad_norm": 1.7750219106674194, "learning_rate": 2.3780000000000004e-06, "loss": 0.0189, "step": 38111 }, { "epoch": 2.0318814309324518, "grad_norm": 2.5209243297576904, "learning_rate": 2.3778e-06, "loss": 0.0325, "step": 38112 }, { "epoch": 2.031934744362105, "grad_norm": 1.4159173965454102, "learning_rate": 2.3776000000000003e-06, "loss": 0.0093, "step": 38113 }, { "epoch": 2.0319880577917577, "grad_norm": 2.3220105171203613, "learning_rate": 2.3774000000000003e-06, "loss": 0.0172, "step": 38114 }, { "epoch": 2.0320413712214105, "grad_norm": 2.2859947681427, "learning_rate": 2.3772e-06, "loss": 0.049, "step": 38115 }, { "epoch": 2.0320946846510637, "grad_norm": 1.5446258783340454, "learning_rate": 2.3770000000000003e-06, "loss": 0.0143, "step": 38116 }, { "epoch": 2.0321479980807164, "grad_norm": 1.2052496671676636, "learning_rate": 2.3768000000000003e-06, "loss": 0.006, "step": 38117 }, { "epoch": 2.0322013115103696, "grad_norm": 0.763683557510376, "learning_rate": 2.3766000000000003e-06, "loss": 0.0056, "step": 38118 }, { "epoch": 2.0322546249400224, "grad_norm": 0.8432938456535339, "learning_rate": 2.3764000000000002e-06, "loss": 0.0103, "step": 38119 }, { "epoch": 2.032307938369675, "grad_norm": 1.3778903484344482, "learning_rate": 2.3762000000000002e-06, "loss": 0.018, "step": 38120 }, { "epoch": 2.0323612517993284, "grad_norm": 1.018754482269287, "learning_rate": 2.376e-06, "loss": 0.0061, "step": 38121 }, { "epoch": 2.032414565228981, "grad_norm": 1.6823914051055908, "learning_rate": 2.3758e-06, "loss": 0.0276, "step": 38122 }, { "epoch": 2.0324678786586343, "grad_norm": 3.0377023220062256, "learning_rate": 2.3756e-06, "loss": 0.0222, "step": 38123 }, { "epoch": 2.032521192088287, "grad_norm": 1.434432029724121, "learning_rate": 2.3754e-06, "loss": 0.0167, "step": 38124 }, { "epoch": 2.03257450551794, "grad_norm": 3.751254081726074, "learning_rate": 2.3752e-06, "loss": 0.0325, "step": 38125 }, { "epoch": 2.032627818947593, "grad_norm": 2.371953010559082, "learning_rate": 2.375e-06, "loss": 0.0315, "step": 38126 }, { "epoch": 2.032681132377246, "grad_norm": 1.062720537185669, "learning_rate": 2.3748e-06, "loss": 0.0145, "step": 38127 }, { "epoch": 2.0327344458068985, "grad_norm": 1.9642709493637085, "learning_rate": 2.3746e-06, "loss": 0.0238, "step": 38128 }, { "epoch": 2.0327877592365518, "grad_norm": 1.823169231414795, "learning_rate": 2.3744000000000005e-06, "loss": 0.0085, "step": 38129 }, { "epoch": 2.0328410726662045, "grad_norm": 1.366546869277954, "learning_rate": 2.3742e-06, "loss": 0.0124, "step": 38130 }, { "epoch": 2.0328943860958577, "grad_norm": 1.2994030714035034, "learning_rate": 2.374e-06, "loss": 0.0082, "step": 38131 }, { "epoch": 2.0329476995255105, "grad_norm": 2.7560603618621826, "learning_rate": 2.3738000000000005e-06, "loss": 0.0129, "step": 38132 }, { "epoch": 2.0330010129551632, "grad_norm": 1.490844964981079, "learning_rate": 2.3736e-06, "loss": 0.0126, "step": 38133 }, { "epoch": 2.0330543263848164, "grad_norm": 1.6868880987167358, "learning_rate": 2.3734e-06, "loss": 0.0225, "step": 38134 }, { "epoch": 2.033107639814469, "grad_norm": 1.5647435188293457, "learning_rate": 2.3732000000000004e-06, "loss": 0.0105, "step": 38135 }, { "epoch": 2.0331609532441224, "grad_norm": 3.0907366275787354, "learning_rate": 2.3730000000000004e-06, "loss": 0.0335, "step": 38136 }, { "epoch": 2.033214266673775, "grad_norm": 0.4088283181190491, "learning_rate": 2.3728e-06, "loss": 0.0025, "step": 38137 }, { "epoch": 2.033267580103428, "grad_norm": 4.261227130889893, "learning_rate": 2.3726000000000004e-06, "loss": 0.0248, "step": 38138 }, { "epoch": 2.033320893533081, "grad_norm": 2.5989484786987305, "learning_rate": 2.3724000000000004e-06, "loss": 0.0224, "step": 38139 }, { "epoch": 2.033374206962734, "grad_norm": 2.4115891456604004, "learning_rate": 2.3722e-06, "loss": 0.0117, "step": 38140 }, { "epoch": 2.033427520392387, "grad_norm": 2.0663838386535645, "learning_rate": 2.3720000000000003e-06, "loss": 0.024, "step": 38141 }, { "epoch": 2.03348083382204, "grad_norm": 1.617781162261963, "learning_rate": 2.3718000000000003e-06, "loss": 0.035, "step": 38142 }, { "epoch": 2.0335341472516926, "grad_norm": 0.4295131266117096, "learning_rate": 2.3716000000000003e-06, "loss": 0.0035, "step": 38143 }, { "epoch": 2.0335874606813458, "grad_norm": 3.7123522758483887, "learning_rate": 2.3714000000000003e-06, "loss": 0.0269, "step": 38144 }, { "epoch": 2.0336407741109985, "grad_norm": 2.4906387329101562, "learning_rate": 2.3712000000000003e-06, "loss": 0.0412, "step": 38145 }, { "epoch": 2.0336940875406513, "grad_norm": 2.0123186111450195, "learning_rate": 2.3710000000000003e-06, "loss": 0.0218, "step": 38146 }, { "epoch": 2.0337474009703045, "grad_norm": 0.5776020884513855, "learning_rate": 2.3708000000000002e-06, "loss": 0.0044, "step": 38147 }, { "epoch": 2.0338007143999572, "grad_norm": 1.3373942375183105, "learning_rate": 2.3706000000000002e-06, "loss": 0.0238, "step": 38148 }, { "epoch": 2.0338540278296104, "grad_norm": 1.4729645252227783, "learning_rate": 2.3704e-06, "loss": 0.0123, "step": 38149 }, { "epoch": 2.033907341259263, "grad_norm": 2.250572443008423, "learning_rate": 2.3702e-06, "loss": 0.0231, "step": 38150 }, { "epoch": 2.033960654688916, "grad_norm": 1.9476079940795898, "learning_rate": 2.37e-06, "loss": 0.0342, "step": 38151 }, { "epoch": 2.034013968118569, "grad_norm": 0.7974580526351929, "learning_rate": 2.3698e-06, "loss": 0.004, "step": 38152 }, { "epoch": 2.034067281548222, "grad_norm": 1.7612074613571167, "learning_rate": 2.3696e-06, "loss": 0.0237, "step": 38153 }, { "epoch": 2.034120594977875, "grad_norm": 0.9258235096931458, "learning_rate": 2.3694e-06, "loss": 0.0053, "step": 38154 }, { "epoch": 2.034173908407528, "grad_norm": 2.1136958599090576, "learning_rate": 2.3692e-06, "loss": 0.0152, "step": 38155 }, { "epoch": 2.0342272218371806, "grad_norm": 0.5231662392616272, "learning_rate": 2.369e-06, "loss": 0.0045, "step": 38156 }, { "epoch": 2.034280535266834, "grad_norm": 1.4185227155685425, "learning_rate": 2.3688e-06, "loss": 0.0192, "step": 38157 }, { "epoch": 2.0343338486964866, "grad_norm": 1.9199692010879517, "learning_rate": 2.3686e-06, "loss": 0.0214, "step": 38158 }, { "epoch": 2.03438716212614, "grad_norm": 1.5889219045639038, "learning_rate": 2.3684e-06, "loss": 0.0164, "step": 38159 }, { "epoch": 2.0344404755557925, "grad_norm": 1.2798771858215332, "learning_rate": 2.3682e-06, "loss": 0.0151, "step": 38160 }, { "epoch": 2.0344937889854453, "grad_norm": 1.4537125825881958, "learning_rate": 2.3680000000000005e-06, "loss": 0.0185, "step": 38161 }, { "epoch": 2.0345471024150985, "grad_norm": 3.025270462036133, "learning_rate": 2.3678e-06, "loss": 0.014, "step": 38162 }, { "epoch": 2.0346004158447513, "grad_norm": 0.8138257265090942, "learning_rate": 2.3676e-06, "loss": 0.0051, "step": 38163 }, { "epoch": 2.034653729274404, "grad_norm": 1.1065407991409302, "learning_rate": 2.3674000000000004e-06, "loss": 0.0154, "step": 38164 }, { "epoch": 2.0347070427040572, "grad_norm": 1.735129475593567, "learning_rate": 2.3672000000000004e-06, "loss": 0.0098, "step": 38165 }, { "epoch": 2.03476035613371, "grad_norm": 1.1042975187301636, "learning_rate": 2.367e-06, "loss": 0.0182, "step": 38166 }, { "epoch": 2.034813669563363, "grad_norm": 1.8485848903656006, "learning_rate": 2.3668000000000004e-06, "loss": 0.0083, "step": 38167 }, { "epoch": 2.034866982993016, "grad_norm": 2.433915376663208, "learning_rate": 2.3666000000000003e-06, "loss": 0.0338, "step": 38168 }, { "epoch": 2.0349202964226687, "grad_norm": 1.5772404670715332, "learning_rate": 2.3664e-06, "loss": 0.0085, "step": 38169 }, { "epoch": 2.034973609852322, "grad_norm": 2.7080140113830566, "learning_rate": 2.3662000000000003e-06, "loss": 0.0194, "step": 38170 }, { "epoch": 2.0350269232819747, "grad_norm": 2.372528076171875, "learning_rate": 2.3660000000000003e-06, "loss": 0.0155, "step": 38171 }, { "epoch": 2.035080236711628, "grad_norm": 1.5077766180038452, "learning_rate": 2.3658000000000003e-06, "loss": 0.0121, "step": 38172 }, { "epoch": 2.0351335501412806, "grad_norm": 3.111783027648926, "learning_rate": 2.3656000000000003e-06, "loss": 0.0314, "step": 38173 }, { "epoch": 2.0351868635709334, "grad_norm": 1.8334227800369263, "learning_rate": 2.3654000000000003e-06, "loss": 0.0051, "step": 38174 }, { "epoch": 2.0352401770005866, "grad_norm": 0.2595764100551605, "learning_rate": 2.3652000000000002e-06, "loss": 0.0022, "step": 38175 }, { "epoch": 2.0352934904302393, "grad_norm": 1.92026686668396, "learning_rate": 2.3650000000000002e-06, "loss": 0.0098, "step": 38176 }, { "epoch": 2.035346803859892, "grad_norm": 2.9497580528259277, "learning_rate": 2.3648e-06, "loss": 0.0431, "step": 38177 }, { "epoch": 2.0354001172895453, "grad_norm": 3.0452232360839844, "learning_rate": 2.3646e-06, "loss": 0.0151, "step": 38178 }, { "epoch": 2.035453430719198, "grad_norm": 1.425032615661621, "learning_rate": 2.3644e-06, "loss": 0.0058, "step": 38179 }, { "epoch": 2.0355067441488512, "grad_norm": 5.405126094818115, "learning_rate": 2.3642e-06, "loss": 0.0186, "step": 38180 }, { "epoch": 2.035560057578504, "grad_norm": 0.6706025004386902, "learning_rate": 2.364e-06, "loss": 0.0031, "step": 38181 }, { "epoch": 2.0356133710081568, "grad_norm": 1.8704888820648193, "learning_rate": 2.3638e-06, "loss": 0.0289, "step": 38182 }, { "epoch": 2.03566668443781, "grad_norm": 1.1923224925994873, "learning_rate": 2.3636e-06, "loss": 0.0074, "step": 38183 }, { "epoch": 2.0357199978674627, "grad_norm": 3.644862174987793, "learning_rate": 2.3634e-06, "loss": 0.0198, "step": 38184 }, { "epoch": 2.035773311297116, "grad_norm": 1.491096019744873, "learning_rate": 2.3632e-06, "loss": 0.0281, "step": 38185 }, { "epoch": 2.0358266247267687, "grad_norm": 1.1251367330551147, "learning_rate": 2.3630000000000005e-06, "loss": 0.0142, "step": 38186 }, { "epoch": 2.0358799381564214, "grad_norm": 0.9245527386665344, "learning_rate": 2.3628e-06, "loss": 0.0055, "step": 38187 }, { "epoch": 2.0359332515860746, "grad_norm": 1.3265790939331055, "learning_rate": 2.3626e-06, "loss": 0.0224, "step": 38188 }, { "epoch": 2.0359865650157274, "grad_norm": 2.5012412071228027, "learning_rate": 2.3624000000000005e-06, "loss": 0.0203, "step": 38189 }, { "epoch": 2.0360398784453806, "grad_norm": 4.193736553192139, "learning_rate": 2.3622000000000004e-06, "loss": 0.0216, "step": 38190 }, { "epoch": 2.0360931918750333, "grad_norm": 2.387755870819092, "learning_rate": 2.362e-06, "loss": 0.0137, "step": 38191 }, { "epoch": 2.036146505304686, "grad_norm": 3.3190128803253174, "learning_rate": 2.3618000000000004e-06, "loss": 0.0134, "step": 38192 }, { "epoch": 2.0361998187343393, "grad_norm": 1.8944562673568726, "learning_rate": 2.3616000000000004e-06, "loss": 0.0101, "step": 38193 }, { "epoch": 2.036253132163992, "grad_norm": 1.0444871187210083, "learning_rate": 2.3614e-06, "loss": 0.0083, "step": 38194 }, { "epoch": 2.0363064455936453, "grad_norm": 1.9244669675827026, "learning_rate": 2.3612000000000004e-06, "loss": 0.0096, "step": 38195 }, { "epoch": 2.036359759023298, "grad_norm": 1.022162914276123, "learning_rate": 2.3610000000000003e-06, "loss": 0.0057, "step": 38196 }, { "epoch": 2.036413072452951, "grad_norm": 0.8622463941574097, "learning_rate": 2.3608000000000003e-06, "loss": 0.0057, "step": 38197 }, { "epoch": 2.036466385882604, "grad_norm": 2.109938144683838, "learning_rate": 2.3606000000000003e-06, "loss": 0.0261, "step": 38198 }, { "epoch": 2.0365196993122567, "grad_norm": 2.087806463241577, "learning_rate": 2.3604000000000003e-06, "loss": 0.0172, "step": 38199 }, { "epoch": 2.0365730127419095, "grad_norm": 3.543689250946045, "learning_rate": 2.3602000000000003e-06, "loss": 0.0225, "step": 38200 }, { "epoch": 2.0366263261715627, "grad_norm": 1.2869430780410767, "learning_rate": 2.3600000000000003e-06, "loss": 0.0158, "step": 38201 }, { "epoch": 2.0366796396012155, "grad_norm": 0.9398661851882935, "learning_rate": 2.3598000000000003e-06, "loss": 0.0117, "step": 38202 }, { "epoch": 2.0367329530308687, "grad_norm": 1.785677194595337, "learning_rate": 2.3596000000000002e-06, "loss": 0.0078, "step": 38203 }, { "epoch": 2.0367862664605214, "grad_norm": 2.3305463790893555, "learning_rate": 2.3594000000000002e-06, "loss": 0.0501, "step": 38204 }, { "epoch": 2.036839579890174, "grad_norm": 2.1524765491485596, "learning_rate": 2.3592000000000002e-06, "loss": 0.0281, "step": 38205 }, { "epoch": 2.0368928933198274, "grad_norm": 1.4175602197647095, "learning_rate": 2.359e-06, "loss": 0.0223, "step": 38206 }, { "epoch": 2.03694620674948, "grad_norm": 1.8807477951049805, "learning_rate": 2.3588e-06, "loss": 0.0189, "step": 38207 }, { "epoch": 2.0369995201791333, "grad_norm": 2.4211792945861816, "learning_rate": 2.3586e-06, "loss": 0.0268, "step": 38208 }, { "epoch": 2.037052833608786, "grad_norm": 1.2040393352508545, "learning_rate": 2.3584e-06, "loss": 0.0233, "step": 38209 }, { "epoch": 2.037106147038439, "grad_norm": 0.7841507792472839, "learning_rate": 2.3582e-06, "loss": 0.0068, "step": 38210 }, { "epoch": 2.037159460468092, "grad_norm": 1.9339481592178345, "learning_rate": 2.358e-06, "loss": 0.0194, "step": 38211 }, { "epoch": 2.037212773897745, "grad_norm": 1.0703409910202026, "learning_rate": 2.3578e-06, "loss": 0.0108, "step": 38212 }, { "epoch": 2.0372660873273976, "grad_norm": 1.9051939249038696, "learning_rate": 2.3576e-06, "loss": 0.0075, "step": 38213 }, { "epoch": 2.0373194007570508, "grad_norm": 1.8171379566192627, "learning_rate": 2.3574e-06, "loss": 0.0465, "step": 38214 }, { "epoch": 2.0373727141867035, "grad_norm": 0.8638606667518616, "learning_rate": 2.3572000000000005e-06, "loss": 0.004, "step": 38215 }, { "epoch": 2.0374260276163567, "grad_norm": 2.8522300720214844, "learning_rate": 2.357e-06, "loss": 0.0276, "step": 38216 }, { "epoch": 2.0374793410460095, "grad_norm": 0.3714761734008789, "learning_rate": 2.3568e-06, "loss": 0.0029, "step": 38217 }, { "epoch": 2.0375326544756622, "grad_norm": 1.06011962890625, "learning_rate": 2.3566000000000004e-06, "loss": 0.0067, "step": 38218 }, { "epoch": 2.0375859679053154, "grad_norm": 2.3793270587921143, "learning_rate": 2.3564e-06, "loss": 0.0757, "step": 38219 }, { "epoch": 2.037639281334968, "grad_norm": 1.7252302169799805, "learning_rate": 2.3562e-06, "loss": 0.0083, "step": 38220 }, { "epoch": 2.0376925947646214, "grad_norm": 2.2920584678649902, "learning_rate": 2.3560000000000004e-06, "loss": 0.0154, "step": 38221 }, { "epoch": 2.037745908194274, "grad_norm": 1.1855939626693726, "learning_rate": 2.3558000000000004e-06, "loss": 0.0064, "step": 38222 }, { "epoch": 2.037799221623927, "grad_norm": 1.1605198383331299, "learning_rate": 2.3556e-06, "loss": 0.0144, "step": 38223 }, { "epoch": 2.03785253505358, "grad_norm": 3.2622857093811035, "learning_rate": 2.3554000000000004e-06, "loss": 0.0261, "step": 38224 }, { "epoch": 2.037905848483233, "grad_norm": 0.43633049726486206, "learning_rate": 2.3552000000000003e-06, "loss": 0.0028, "step": 38225 }, { "epoch": 2.037959161912886, "grad_norm": 1.1632460355758667, "learning_rate": 2.355e-06, "loss": 0.0079, "step": 38226 }, { "epoch": 2.038012475342539, "grad_norm": 1.8263002634048462, "learning_rate": 2.3548000000000003e-06, "loss": 0.0338, "step": 38227 }, { "epoch": 2.0380657887721916, "grad_norm": 1.7339146137237549, "learning_rate": 2.3546000000000003e-06, "loss": 0.0139, "step": 38228 }, { "epoch": 2.038119102201845, "grad_norm": 7.212408542633057, "learning_rate": 2.3544000000000003e-06, "loss": 0.0216, "step": 38229 }, { "epoch": 2.0381724156314975, "grad_norm": 0.32829195261001587, "learning_rate": 2.3542000000000003e-06, "loss": 0.0033, "step": 38230 }, { "epoch": 2.0382257290611503, "grad_norm": 4.861959934234619, "learning_rate": 2.3540000000000002e-06, "loss": 0.0151, "step": 38231 }, { "epoch": 2.0382790424908035, "grad_norm": 1.0872447490692139, "learning_rate": 2.3538000000000002e-06, "loss": 0.0062, "step": 38232 }, { "epoch": 2.0383323559204563, "grad_norm": 2.2338979244232178, "learning_rate": 2.3536000000000002e-06, "loss": 0.0209, "step": 38233 }, { "epoch": 2.0383856693501095, "grad_norm": 1.7815093994140625, "learning_rate": 2.3534e-06, "loss": 0.0207, "step": 38234 }, { "epoch": 2.038438982779762, "grad_norm": 0.4225637912750244, "learning_rate": 2.3532e-06, "loss": 0.0042, "step": 38235 }, { "epoch": 2.038492296209415, "grad_norm": 2.2470104694366455, "learning_rate": 2.353e-06, "loss": 0.0263, "step": 38236 }, { "epoch": 2.038545609639068, "grad_norm": 1.894311547279358, "learning_rate": 2.3528e-06, "loss": 0.0058, "step": 38237 }, { "epoch": 2.038598923068721, "grad_norm": 3.7188122272491455, "learning_rate": 2.3526e-06, "loss": 0.0219, "step": 38238 }, { "epoch": 2.038652236498374, "grad_norm": 2.8383805751800537, "learning_rate": 2.3524e-06, "loss": 0.0269, "step": 38239 }, { "epoch": 2.038705549928027, "grad_norm": 2.0414273738861084, "learning_rate": 2.3522e-06, "loss": 0.0195, "step": 38240 }, { "epoch": 2.0387588633576796, "grad_norm": 0.8799633979797363, "learning_rate": 2.352e-06, "loss": 0.023, "step": 38241 }, { "epoch": 2.038812176787333, "grad_norm": 2.2050492763519287, "learning_rate": 2.3518e-06, "loss": 0.0291, "step": 38242 }, { "epoch": 2.0388654902169856, "grad_norm": 1.9531967639923096, "learning_rate": 2.3516e-06, "loss": 0.0203, "step": 38243 }, { "epoch": 2.038918803646639, "grad_norm": 2.36055064201355, "learning_rate": 2.3514e-06, "loss": 0.0236, "step": 38244 }, { "epoch": 2.0389721170762916, "grad_norm": 1.6789791584014893, "learning_rate": 2.3512e-06, "loss": 0.0288, "step": 38245 }, { "epoch": 2.0390254305059443, "grad_norm": 2.9406728744506836, "learning_rate": 2.351e-06, "loss": 0.0257, "step": 38246 }, { "epoch": 2.0390787439355975, "grad_norm": 0.785830557346344, "learning_rate": 2.3508000000000004e-06, "loss": 0.0044, "step": 38247 }, { "epoch": 2.0391320573652503, "grad_norm": 1.0085301399230957, "learning_rate": 2.3506e-06, "loss": 0.0049, "step": 38248 }, { "epoch": 2.039185370794903, "grad_norm": 1.9220387935638428, "learning_rate": 2.3504e-06, "loss": 0.049, "step": 38249 }, { "epoch": 2.0392386842245562, "grad_norm": 0.8709104061126709, "learning_rate": 2.3502000000000004e-06, "loss": 0.006, "step": 38250 }, { "epoch": 2.039291997654209, "grad_norm": 1.6913725137710571, "learning_rate": 2.35e-06, "loss": 0.0076, "step": 38251 }, { "epoch": 2.039345311083862, "grad_norm": 1.6648104190826416, "learning_rate": 2.3498e-06, "loss": 0.0102, "step": 38252 }, { "epoch": 2.039398624513515, "grad_norm": 1.9314507246017456, "learning_rate": 2.3496000000000003e-06, "loss": 0.0381, "step": 38253 }, { "epoch": 2.0394519379431677, "grad_norm": 1.0410343408584595, "learning_rate": 2.3494000000000003e-06, "loss": 0.0069, "step": 38254 }, { "epoch": 2.039505251372821, "grad_norm": 4.676257610321045, "learning_rate": 2.3492e-06, "loss": 0.0475, "step": 38255 }, { "epoch": 2.0395585648024737, "grad_norm": 0.678824782371521, "learning_rate": 2.3490000000000003e-06, "loss": 0.0041, "step": 38256 }, { "epoch": 2.039611878232127, "grad_norm": 1.4905339479446411, "learning_rate": 2.3488000000000003e-06, "loss": 0.0247, "step": 38257 }, { "epoch": 2.0396651916617796, "grad_norm": 2.9313526153564453, "learning_rate": 2.3486000000000003e-06, "loss": 0.0386, "step": 38258 }, { "epoch": 2.0397185050914324, "grad_norm": 1.3893368244171143, "learning_rate": 2.3484000000000002e-06, "loss": 0.013, "step": 38259 }, { "epoch": 2.0397718185210856, "grad_norm": 1.0325982570648193, "learning_rate": 2.3482000000000002e-06, "loss": 0.0054, "step": 38260 }, { "epoch": 2.0398251319507383, "grad_norm": 2.258924722671509, "learning_rate": 2.3480000000000002e-06, "loss": 0.0348, "step": 38261 }, { "epoch": 2.0398784453803915, "grad_norm": 1.8639742136001587, "learning_rate": 2.3478e-06, "loss": 0.0204, "step": 38262 }, { "epoch": 2.0399317588100443, "grad_norm": 2.1778993606567383, "learning_rate": 2.3476e-06, "loss": 0.0186, "step": 38263 }, { "epoch": 2.039985072239697, "grad_norm": 1.5554695129394531, "learning_rate": 2.3474e-06, "loss": 0.0413, "step": 38264 }, { "epoch": 2.0400383856693503, "grad_norm": 1.5587347745895386, "learning_rate": 2.3472e-06, "loss": 0.0157, "step": 38265 }, { "epoch": 2.040091699099003, "grad_norm": 2.0860705375671387, "learning_rate": 2.347e-06, "loss": 0.0206, "step": 38266 }, { "epoch": 2.0401450125286558, "grad_norm": 1.141618013381958, "learning_rate": 2.3468e-06, "loss": 0.0059, "step": 38267 }, { "epoch": 2.040198325958309, "grad_norm": 0.9135096669197083, "learning_rate": 2.3466e-06, "loss": 0.0063, "step": 38268 }, { "epoch": 2.0402516393879617, "grad_norm": 1.450449824333191, "learning_rate": 2.3464e-06, "loss": 0.0261, "step": 38269 }, { "epoch": 2.040304952817615, "grad_norm": 1.3169020414352417, "learning_rate": 2.3462e-06, "loss": 0.0095, "step": 38270 }, { "epoch": 2.0403582662472677, "grad_norm": 1.813586711883545, "learning_rate": 2.346e-06, "loss": 0.0143, "step": 38271 }, { "epoch": 2.0404115796769204, "grad_norm": 1.6234253644943237, "learning_rate": 2.3458000000000005e-06, "loss": 0.0153, "step": 38272 }, { "epoch": 2.0404648931065736, "grad_norm": 1.5543580055236816, "learning_rate": 2.3456e-06, "loss": 0.0113, "step": 38273 }, { "epoch": 2.0405182065362264, "grad_norm": 1.3919143676757812, "learning_rate": 2.3454e-06, "loss": 0.0126, "step": 38274 }, { "epoch": 2.0405715199658796, "grad_norm": 1.5348619222640991, "learning_rate": 2.3452000000000004e-06, "loss": 0.0068, "step": 38275 }, { "epoch": 2.0406248333955324, "grad_norm": 4.248289585113525, "learning_rate": 2.345e-06, "loss": 0.0198, "step": 38276 }, { "epoch": 2.040678146825185, "grad_norm": 2.2590720653533936, "learning_rate": 2.3448e-06, "loss": 0.0153, "step": 38277 }, { "epoch": 2.0407314602548383, "grad_norm": 1.9577432870864868, "learning_rate": 2.3446000000000004e-06, "loss": 0.0179, "step": 38278 }, { "epoch": 2.040784773684491, "grad_norm": 1.4348613023757935, "learning_rate": 2.3444000000000004e-06, "loss": 0.007, "step": 38279 }, { "epoch": 2.0408380871141443, "grad_norm": 1.2489901781082153, "learning_rate": 2.3442e-06, "loss": 0.021, "step": 38280 }, { "epoch": 2.040891400543797, "grad_norm": 1.4263304471969604, "learning_rate": 2.3440000000000003e-06, "loss": 0.017, "step": 38281 }, { "epoch": 2.04094471397345, "grad_norm": 5.350898265838623, "learning_rate": 2.3438000000000003e-06, "loss": 0.0133, "step": 38282 }, { "epoch": 2.040998027403103, "grad_norm": 1.262943148612976, "learning_rate": 2.3436000000000003e-06, "loss": 0.006, "step": 38283 }, { "epoch": 2.0410513408327557, "grad_norm": 1.012179970741272, "learning_rate": 2.3434000000000003e-06, "loss": 0.0072, "step": 38284 }, { "epoch": 2.0411046542624085, "grad_norm": 1.7154607772827148, "learning_rate": 2.3432000000000003e-06, "loss": 0.0292, "step": 38285 }, { "epoch": 2.0411579676920617, "grad_norm": 1.0490065813064575, "learning_rate": 2.3430000000000003e-06, "loss": 0.0207, "step": 38286 }, { "epoch": 2.0412112811217145, "grad_norm": 2.349158763885498, "learning_rate": 2.3428000000000003e-06, "loss": 0.0097, "step": 38287 }, { "epoch": 2.0412645945513677, "grad_norm": 0.953983724117279, "learning_rate": 2.3426000000000002e-06, "loss": 0.0054, "step": 38288 }, { "epoch": 2.0413179079810204, "grad_norm": 1.0910292863845825, "learning_rate": 2.3424000000000002e-06, "loss": 0.0039, "step": 38289 }, { "epoch": 2.041371221410673, "grad_norm": 0.5281804800033569, "learning_rate": 2.3422e-06, "loss": 0.0039, "step": 38290 }, { "epoch": 2.0414245348403264, "grad_norm": 1.5105968713760376, "learning_rate": 2.342e-06, "loss": 0.0155, "step": 38291 }, { "epoch": 2.041477848269979, "grad_norm": 3.06668758392334, "learning_rate": 2.3418e-06, "loss": 0.0236, "step": 38292 }, { "epoch": 2.0415311616996323, "grad_norm": 0.9670555591583252, "learning_rate": 2.3416e-06, "loss": 0.0042, "step": 38293 }, { "epoch": 2.041584475129285, "grad_norm": 2.52260684967041, "learning_rate": 2.3414e-06, "loss": 0.0361, "step": 38294 }, { "epoch": 2.041637788558938, "grad_norm": 0.891287624835968, "learning_rate": 2.3412e-06, "loss": 0.0063, "step": 38295 }, { "epoch": 2.041691101988591, "grad_norm": 1.4200915098190308, "learning_rate": 2.341e-06, "loss": 0.01, "step": 38296 }, { "epoch": 2.041744415418244, "grad_norm": 1.4627104997634888, "learning_rate": 2.3408e-06, "loss": 0.0245, "step": 38297 }, { "epoch": 2.0417977288478966, "grad_norm": 2.0230889320373535, "learning_rate": 2.3406e-06, "loss": 0.012, "step": 38298 }, { "epoch": 2.0418510422775498, "grad_norm": 1.583500623703003, "learning_rate": 2.3404e-06, "loss": 0.0088, "step": 38299 }, { "epoch": 2.0419043557072025, "grad_norm": 1.3245315551757812, "learning_rate": 2.3402e-06, "loss": 0.0027, "step": 38300 }, { "epoch": 2.0419576691368557, "grad_norm": 3.2446787357330322, "learning_rate": 2.3400000000000005e-06, "loss": 0.0419, "step": 38301 }, { "epoch": 2.0420109825665085, "grad_norm": 2.0372262001037598, "learning_rate": 2.3398e-06, "loss": 0.0188, "step": 38302 }, { "epoch": 2.0420642959961612, "grad_norm": 1.715680480003357, "learning_rate": 2.3396e-06, "loss": 0.0062, "step": 38303 }, { "epoch": 2.0421176094258144, "grad_norm": 3.812500238418579, "learning_rate": 2.3394000000000004e-06, "loss": 0.0103, "step": 38304 }, { "epoch": 2.042170922855467, "grad_norm": 1.7724685668945312, "learning_rate": 2.3392e-06, "loss": 0.0097, "step": 38305 }, { "epoch": 2.0422242362851204, "grad_norm": 1.6896005868911743, "learning_rate": 2.339e-06, "loss": 0.0074, "step": 38306 }, { "epoch": 2.042277549714773, "grad_norm": 0.8105652332305908, "learning_rate": 2.3388000000000004e-06, "loss": 0.0065, "step": 38307 }, { "epoch": 2.042330863144426, "grad_norm": 2.440986156463623, "learning_rate": 2.3386000000000004e-06, "loss": 0.0168, "step": 38308 }, { "epoch": 2.042384176574079, "grad_norm": 2.087852716445923, "learning_rate": 2.3384e-06, "loss": 0.006, "step": 38309 }, { "epoch": 2.042437490003732, "grad_norm": 0.5686489939689636, "learning_rate": 2.3382000000000003e-06, "loss": 0.0039, "step": 38310 }, { "epoch": 2.042490803433385, "grad_norm": 0.5519202351570129, "learning_rate": 2.3380000000000003e-06, "loss": 0.0028, "step": 38311 }, { "epoch": 2.042544116863038, "grad_norm": 2.3042726516723633, "learning_rate": 2.3378e-06, "loss": 0.0124, "step": 38312 }, { "epoch": 2.0425974302926906, "grad_norm": 0.3261988162994385, "learning_rate": 2.3376000000000003e-06, "loss": 0.002, "step": 38313 }, { "epoch": 2.042650743722344, "grad_norm": 1.670423150062561, "learning_rate": 2.3374000000000003e-06, "loss": 0.0387, "step": 38314 }, { "epoch": 2.0427040571519965, "grad_norm": 1.9887011051177979, "learning_rate": 2.3372000000000003e-06, "loss": 0.0143, "step": 38315 }, { "epoch": 2.0427573705816493, "grad_norm": 2.036919116973877, "learning_rate": 2.3370000000000002e-06, "loss": 0.039, "step": 38316 }, { "epoch": 2.0428106840113025, "grad_norm": 1.4825454950332642, "learning_rate": 2.3368000000000002e-06, "loss": 0.0073, "step": 38317 }, { "epoch": 2.0428639974409553, "grad_norm": 0.35118305683135986, "learning_rate": 2.3366e-06, "loss": 0.0023, "step": 38318 }, { "epoch": 2.0429173108706085, "grad_norm": 1.3536627292633057, "learning_rate": 2.3364e-06, "loss": 0.0086, "step": 38319 }, { "epoch": 2.042970624300261, "grad_norm": 1.3848950862884521, "learning_rate": 2.3362e-06, "loss": 0.0133, "step": 38320 }, { "epoch": 2.043023937729914, "grad_norm": 1.283551812171936, "learning_rate": 2.336e-06, "loss": 0.006, "step": 38321 }, { "epoch": 2.043077251159567, "grad_norm": 1.563080906867981, "learning_rate": 2.3358e-06, "loss": 0.0123, "step": 38322 }, { "epoch": 2.04313056458922, "grad_norm": 0.49944600462913513, "learning_rate": 2.3356e-06, "loss": 0.0038, "step": 38323 }, { "epoch": 2.043183878018873, "grad_norm": 1.4393815994262695, "learning_rate": 2.3354e-06, "loss": 0.0073, "step": 38324 }, { "epoch": 2.043237191448526, "grad_norm": 4.144497394561768, "learning_rate": 2.3352e-06, "loss": 0.0559, "step": 38325 }, { "epoch": 2.0432905048781786, "grad_norm": 1.0732476711273193, "learning_rate": 2.3350000000000005e-06, "loss": 0.0191, "step": 38326 }, { "epoch": 2.043343818307832, "grad_norm": 2.266554117202759, "learning_rate": 2.3348e-06, "loss": 0.0176, "step": 38327 }, { "epoch": 2.0433971317374846, "grad_norm": 0.40262341499328613, "learning_rate": 2.3346e-06, "loss": 0.0027, "step": 38328 }, { "epoch": 2.043450445167138, "grad_norm": 3.215900421142578, "learning_rate": 2.3344000000000005e-06, "loss": 0.0155, "step": 38329 }, { "epoch": 2.0435037585967906, "grad_norm": 2.497556686401367, "learning_rate": 2.3342e-06, "loss": 0.0087, "step": 38330 }, { "epoch": 2.0435570720264433, "grad_norm": 1.5025744438171387, "learning_rate": 2.334e-06, "loss": 0.0172, "step": 38331 }, { "epoch": 2.0436103854560965, "grad_norm": 6.599753379821777, "learning_rate": 2.3338000000000004e-06, "loss": 0.017, "step": 38332 }, { "epoch": 2.0436636988857493, "grad_norm": 2.107276439666748, "learning_rate": 2.3336000000000004e-06, "loss": 0.0198, "step": 38333 }, { "epoch": 2.043717012315402, "grad_norm": 0.438889741897583, "learning_rate": 2.3334e-06, "loss": 0.0028, "step": 38334 }, { "epoch": 2.0437703257450552, "grad_norm": 1.5734670162200928, "learning_rate": 2.3332000000000004e-06, "loss": 0.0097, "step": 38335 }, { "epoch": 2.043823639174708, "grad_norm": 1.0331887006759644, "learning_rate": 2.3330000000000004e-06, "loss": 0.0281, "step": 38336 }, { "epoch": 2.043876952604361, "grad_norm": 2.8338146209716797, "learning_rate": 2.3328e-06, "loss": 0.0275, "step": 38337 }, { "epoch": 2.043930266034014, "grad_norm": 1.456357717514038, "learning_rate": 2.3326000000000003e-06, "loss": 0.0047, "step": 38338 }, { "epoch": 2.0439835794636667, "grad_norm": 2.7911345958709717, "learning_rate": 2.3324000000000003e-06, "loss": 0.0317, "step": 38339 }, { "epoch": 2.04403689289332, "grad_norm": 0.3212393522262573, "learning_rate": 2.3322000000000003e-06, "loss": 0.002, "step": 38340 }, { "epoch": 2.0440902063229727, "grad_norm": 0.8010120391845703, "learning_rate": 2.3320000000000003e-06, "loss": 0.0035, "step": 38341 }, { "epoch": 2.044143519752626, "grad_norm": 1.6512950658798218, "learning_rate": 2.3318000000000003e-06, "loss": 0.0213, "step": 38342 }, { "epoch": 2.0441968331822786, "grad_norm": 1.0490472316741943, "learning_rate": 2.3316000000000003e-06, "loss": 0.0149, "step": 38343 }, { "epoch": 2.0442501466119314, "grad_norm": 1.4165630340576172, "learning_rate": 2.3314000000000002e-06, "loss": 0.0071, "step": 38344 }, { "epoch": 2.0443034600415846, "grad_norm": 2.320314645767212, "learning_rate": 2.3312000000000002e-06, "loss": 0.0271, "step": 38345 }, { "epoch": 2.0443567734712373, "grad_norm": 1.2872064113616943, "learning_rate": 2.3310000000000002e-06, "loss": 0.0225, "step": 38346 }, { "epoch": 2.0444100869008905, "grad_norm": 1.9546936750411987, "learning_rate": 2.3308e-06, "loss": 0.0209, "step": 38347 }, { "epoch": 2.0444634003305433, "grad_norm": 1.0197055339813232, "learning_rate": 2.3306e-06, "loss": 0.0187, "step": 38348 }, { "epoch": 2.044516713760196, "grad_norm": 0.8197763562202454, "learning_rate": 2.3304e-06, "loss": 0.0043, "step": 38349 }, { "epoch": 2.0445700271898493, "grad_norm": 3.581026792526245, "learning_rate": 2.3302e-06, "loss": 0.0286, "step": 38350 }, { "epoch": 2.044623340619502, "grad_norm": 0.7765318751335144, "learning_rate": 2.33e-06, "loss": 0.0036, "step": 38351 }, { "epoch": 2.0446766540491548, "grad_norm": 0.44453227519989014, "learning_rate": 2.3298e-06, "loss": 0.003, "step": 38352 }, { "epoch": 2.044729967478808, "grad_norm": 1.1292415857315063, "learning_rate": 2.3296e-06, "loss": 0.0036, "step": 38353 }, { "epoch": 2.0447832809084607, "grad_norm": 0.5502690672874451, "learning_rate": 2.3294e-06, "loss": 0.0034, "step": 38354 }, { "epoch": 2.044836594338114, "grad_norm": 3.2219955921173096, "learning_rate": 2.3292e-06, "loss": 0.0348, "step": 38355 }, { "epoch": 2.0448899077677667, "grad_norm": 1.2492995262145996, "learning_rate": 2.329e-06, "loss": 0.0363, "step": 38356 }, { "epoch": 2.0449432211974194, "grad_norm": 2.3361260890960693, "learning_rate": 2.3288e-06, "loss": 0.0247, "step": 38357 }, { "epoch": 2.0449965346270726, "grad_norm": 4.32905912399292, "learning_rate": 2.3286000000000005e-06, "loss": 0.0203, "step": 38358 }, { "epoch": 2.0450498480567254, "grad_norm": 2.4959256649017334, "learning_rate": 2.3284e-06, "loss": 0.0275, "step": 38359 }, { "epoch": 2.0451031614863786, "grad_norm": 1.1609324216842651, "learning_rate": 2.3282e-06, "loss": 0.0053, "step": 38360 }, { "epoch": 2.0451564749160314, "grad_norm": 1.8195446729660034, "learning_rate": 2.3280000000000004e-06, "loss": 0.0111, "step": 38361 }, { "epoch": 2.045209788345684, "grad_norm": 0.6994348168373108, "learning_rate": 2.3278e-06, "loss": 0.0037, "step": 38362 }, { "epoch": 2.0452631017753373, "grad_norm": 1.2750760316848755, "learning_rate": 2.3276e-06, "loss": 0.0056, "step": 38363 }, { "epoch": 2.04531641520499, "grad_norm": 3.33634352684021, "learning_rate": 2.3274000000000004e-06, "loss": 0.0172, "step": 38364 }, { "epoch": 2.0453697286346433, "grad_norm": 3.266646146774292, "learning_rate": 2.3272000000000004e-06, "loss": 0.0478, "step": 38365 }, { "epoch": 2.045423042064296, "grad_norm": 0.5209976434707642, "learning_rate": 2.327e-06, "loss": 0.0032, "step": 38366 }, { "epoch": 2.045476355493949, "grad_norm": 1.9100687503814697, "learning_rate": 2.3268000000000003e-06, "loss": 0.0117, "step": 38367 }, { "epoch": 2.045529668923602, "grad_norm": 2.7658369541168213, "learning_rate": 2.3266000000000003e-06, "loss": 0.0251, "step": 38368 }, { "epoch": 2.0455829823532548, "grad_norm": 2.3770933151245117, "learning_rate": 2.3264000000000003e-06, "loss": 0.0376, "step": 38369 }, { "epoch": 2.0456362957829075, "grad_norm": 2.0439541339874268, "learning_rate": 2.3262000000000003e-06, "loss": 0.0079, "step": 38370 }, { "epoch": 2.0456896092125607, "grad_norm": 3.389827013015747, "learning_rate": 2.3260000000000003e-06, "loss": 0.0177, "step": 38371 }, { "epoch": 2.0457429226422135, "grad_norm": 2.2679409980773926, "learning_rate": 2.3258000000000002e-06, "loss": 0.0235, "step": 38372 }, { "epoch": 2.0457962360718667, "grad_norm": 0.6958780884742737, "learning_rate": 2.3256000000000002e-06, "loss": 0.0035, "step": 38373 }, { "epoch": 2.0458495495015194, "grad_norm": 1.2574323415756226, "learning_rate": 2.3254000000000002e-06, "loss": 0.0084, "step": 38374 }, { "epoch": 2.045902862931172, "grad_norm": 1.5502252578735352, "learning_rate": 2.3252e-06, "loss": 0.0153, "step": 38375 }, { "epoch": 2.0459561763608254, "grad_norm": 0.7258808016777039, "learning_rate": 2.325e-06, "loss": 0.0039, "step": 38376 }, { "epoch": 2.046009489790478, "grad_norm": 4.50067663192749, "learning_rate": 2.3248e-06, "loss": 0.0372, "step": 38377 }, { "epoch": 2.0460628032201313, "grad_norm": 2.363597869873047, "learning_rate": 2.3246e-06, "loss": 0.0166, "step": 38378 }, { "epoch": 2.046116116649784, "grad_norm": 1.0670735836029053, "learning_rate": 2.3244e-06, "loss": 0.0059, "step": 38379 }, { "epoch": 2.046169430079437, "grad_norm": 2.134507179260254, "learning_rate": 2.3242e-06, "loss": 0.0119, "step": 38380 }, { "epoch": 2.04622274350909, "grad_norm": 1.5061073303222656, "learning_rate": 2.324e-06, "loss": 0.0218, "step": 38381 }, { "epoch": 2.046276056938743, "grad_norm": 4.6016035079956055, "learning_rate": 2.3238e-06, "loss": 0.0177, "step": 38382 }, { "epoch": 2.046329370368396, "grad_norm": 6.455255031585693, "learning_rate": 2.3236000000000005e-06, "loss": 0.0317, "step": 38383 }, { "epoch": 2.0463826837980488, "grad_norm": 1.2372715473175049, "learning_rate": 2.3234e-06, "loss": 0.0081, "step": 38384 }, { "epoch": 2.0464359972277015, "grad_norm": 1.6485815048217773, "learning_rate": 2.3232e-06, "loss": 0.0211, "step": 38385 }, { "epoch": 2.0464893106573547, "grad_norm": 1.0871913433074951, "learning_rate": 2.3230000000000005e-06, "loss": 0.005, "step": 38386 }, { "epoch": 2.0465426240870075, "grad_norm": 1.3082702159881592, "learning_rate": 2.3228e-06, "loss": 0.0113, "step": 38387 }, { "epoch": 2.0465959375166602, "grad_norm": 1.73450767993927, "learning_rate": 2.3226e-06, "loss": 0.0164, "step": 38388 }, { "epoch": 2.0466492509463134, "grad_norm": 1.7226579189300537, "learning_rate": 2.3224000000000004e-06, "loss": 0.0096, "step": 38389 }, { "epoch": 2.046702564375966, "grad_norm": 0.9749143719673157, "learning_rate": 2.3222000000000004e-06, "loss": 0.0029, "step": 38390 }, { "epoch": 2.0467558778056194, "grad_norm": 4.177850723266602, "learning_rate": 2.322e-06, "loss": 0.0105, "step": 38391 }, { "epoch": 2.046809191235272, "grad_norm": 0.6854287385940552, "learning_rate": 2.3218000000000004e-06, "loss": 0.0033, "step": 38392 }, { "epoch": 2.046862504664925, "grad_norm": 1.607915997505188, "learning_rate": 2.3216000000000004e-06, "loss": 0.0084, "step": 38393 }, { "epoch": 2.046915818094578, "grad_norm": 1.1561483144760132, "learning_rate": 2.3214000000000003e-06, "loss": 0.0036, "step": 38394 }, { "epoch": 2.046969131524231, "grad_norm": 3.3001956939697266, "learning_rate": 2.3212000000000003e-06, "loss": 0.0187, "step": 38395 }, { "epoch": 2.047022444953884, "grad_norm": 1.597872018814087, "learning_rate": 2.3210000000000003e-06, "loss": 0.0211, "step": 38396 }, { "epoch": 2.047075758383537, "grad_norm": 2.141636610031128, "learning_rate": 2.3208000000000003e-06, "loss": 0.0333, "step": 38397 }, { "epoch": 2.0471290718131896, "grad_norm": 2.5667543411254883, "learning_rate": 2.3206000000000003e-06, "loss": 0.0333, "step": 38398 }, { "epoch": 2.047182385242843, "grad_norm": 1.3402111530303955, "learning_rate": 2.3204000000000003e-06, "loss": 0.005, "step": 38399 }, { "epoch": 2.0472356986724956, "grad_norm": 1.4256004095077515, "learning_rate": 2.3202000000000002e-06, "loss": 0.0196, "step": 38400 }, { "epoch": 2.0472890121021488, "grad_norm": 1.292229413986206, "learning_rate": 2.3200000000000002e-06, "loss": 0.0068, "step": 38401 }, { "epoch": 2.0473423255318015, "grad_norm": 2.3122260570526123, "learning_rate": 2.3198000000000002e-06, "loss": 0.0257, "step": 38402 }, { "epoch": 2.0473956389614543, "grad_norm": 3.0147149562835693, "learning_rate": 2.3196e-06, "loss": 0.0291, "step": 38403 }, { "epoch": 2.0474489523911075, "grad_norm": 0.8313416838645935, "learning_rate": 2.3194e-06, "loss": 0.0037, "step": 38404 }, { "epoch": 2.0475022658207602, "grad_norm": 1.515525460243225, "learning_rate": 2.3192e-06, "loss": 0.0173, "step": 38405 }, { "epoch": 2.047555579250413, "grad_norm": 1.266995906829834, "learning_rate": 2.319e-06, "loss": 0.0069, "step": 38406 }, { "epoch": 2.047608892680066, "grad_norm": 3.4655163288116455, "learning_rate": 2.3188e-06, "loss": 0.0098, "step": 38407 }, { "epoch": 2.047662206109719, "grad_norm": 1.9115169048309326, "learning_rate": 2.3186e-06, "loss": 0.0095, "step": 38408 }, { "epoch": 2.047715519539372, "grad_norm": 3.7251803874969482, "learning_rate": 2.3184e-06, "loss": 0.0422, "step": 38409 }, { "epoch": 2.047768832969025, "grad_norm": 0.5361828804016113, "learning_rate": 2.3182e-06, "loss": 0.0043, "step": 38410 }, { "epoch": 2.0478221463986777, "grad_norm": 1.0592464208602905, "learning_rate": 2.318e-06, "loss": 0.0186, "step": 38411 }, { "epoch": 2.047875459828331, "grad_norm": 1.7138359546661377, "learning_rate": 2.3178000000000005e-06, "loss": 0.0088, "step": 38412 }, { "epoch": 2.0479287732579836, "grad_norm": 3.669189214706421, "learning_rate": 2.3176e-06, "loss": 0.0187, "step": 38413 }, { "epoch": 2.047982086687637, "grad_norm": 0.44425463676452637, "learning_rate": 2.3174e-06, "loss": 0.0035, "step": 38414 }, { "epoch": 2.0480354001172896, "grad_norm": 0.6580918431282043, "learning_rate": 2.3172000000000004e-06, "loss": 0.0038, "step": 38415 }, { "epoch": 2.0480887135469423, "grad_norm": 1.3277767896652222, "learning_rate": 2.317e-06, "loss": 0.0058, "step": 38416 }, { "epoch": 2.0481420269765955, "grad_norm": 4.562785625457764, "learning_rate": 2.3168e-06, "loss": 0.0286, "step": 38417 }, { "epoch": 2.0481953404062483, "grad_norm": 1.8415923118591309, "learning_rate": 2.3166000000000004e-06, "loss": 0.0186, "step": 38418 }, { "epoch": 2.048248653835901, "grad_norm": 2.6585867404937744, "learning_rate": 2.3164000000000004e-06, "loss": 0.0262, "step": 38419 }, { "epoch": 2.0483019672655542, "grad_norm": 2.780635356903076, "learning_rate": 2.3162e-06, "loss": 0.0106, "step": 38420 }, { "epoch": 2.048355280695207, "grad_norm": 0.7097552418708801, "learning_rate": 2.3160000000000004e-06, "loss": 0.0022, "step": 38421 }, { "epoch": 2.04840859412486, "grad_norm": 5.225423336029053, "learning_rate": 2.3158000000000003e-06, "loss": 0.0289, "step": 38422 }, { "epoch": 2.048461907554513, "grad_norm": 0.4951128363609314, "learning_rate": 2.3156e-06, "loss": 0.0027, "step": 38423 }, { "epoch": 2.0485152209841657, "grad_norm": 2.4927804470062256, "learning_rate": 2.3154000000000003e-06, "loss": 0.0282, "step": 38424 }, { "epoch": 2.048568534413819, "grad_norm": 0.59955233335495, "learning_rate": 2.3152000000000003e-06, "loss": 0.0033, "step": 38425 }, { "epoch": 2.0486218478434717, "grad_norm": 4.405542850494385, "learning_rate": 2.3150000000000003e-06, "loss": 0.0235, "step": 38426 }, { "epoch": 2.048675161273125, "grad_norm": 1.8082573413848877, "learning_rate": 2.3148000000000003e-06, "loss": 0.012, "step": 38427 }, { "epoch": 2.0487284747027776, "grad_norm": 2.0130486488342285, "learning_rate": 2.3146000000000003e-06, "loss": 0.0235, "step": 38428 }, { "epoch": 2.0487817881324304, "grad_norm": 2.8264846801757812, "learning_rate": 2.3144000000000002e-06, "loss": 0.0049, "step": 38429 }, { "epoch": 2.0488351015620836, "grad_norm": 0.7300777435302734, "learning_rate": 2.3142000000000002e-06, "loss": 0.0023, "step": 38430 }, { "epoch": 2.0488884149917364, "grad_norm": 1.6334245204925537, "learning_rate": 2.314e-06, "loss": 0.0252, "step": 38431 }, { "epoch": 2.0489417284213896, "grad_norm": 5.966933250427246, "learning_rate": 2.3138e-06, "loss": 0.0513, "step": 38432 }, { "epoch": 2.0489950418510423, "grad_norm": 3.0518476963043213, "learning_rate": 2.3136e-06, "loss": 0.0577, "step": 38433 }, { "epoch": 2.049048355280695, "grad_norm": 1.5466889142990112, "learning_rate": 2.3134e-06, "loss": 0.0212, "step": 38434 }, { "epoch": 2.0491016687103483, "grad_norm": 0.7254412174224854, "learning_rate": 2.3132e-06, "loss": 0.0038, "step": 38435 }, { "epoch": 2.049154982140001, "grad_norm": 1.3950551748275757, "learning_rate": 2.313e-06, "loss": 0.008, "step": 38436 }, { "epoch": 2.049208295569654, "grad_norm": 2.5737576484680176, "learning_rate": 2.3128e-06, "loss": 0.0458, "step": 38437 }, { "epoch": 2.049261608999307, "grad_norm": 1.2890682220458984, "learning_rate": 2.3126e-06, "loss": 0.0195, "step": 38438 }, { "epoch": 2.0493149224289597, "grad_norm": 2.7027478218078613, "learning_rate": 2.3124e-06, "loss": 0.0344, "step": 38439 }, { "epoch": 2.049368235858613, "grad_norm": 1.2663185596466064, "learning_rate": 2.3122e-06, "loss": 0.0175, "step": 38440 }, { "epoch": 2.0494215492882657, "grad_norm": 2.0200841426849365, "learning_rate": 2.312e-06, "loss": 0.0219, "step": 38441 }, { "epoch": 2.0494748627179185, "grad_norm": 2.2393767833709717, "learning_rate": 2.3118e-06, "loss": 0.0198, "step": 38442 }, { "epoch": 2.0495281761475717, "grad_norm": 1.6664128303527832, "learning_rate": 2.3116e-06, "loss": 0.0163, "step": 38443 }, { "epoch": 2.0495814895772244, "grad_norm": 1.7320550680160522, "learning_rate": 2.3114000000000004e-06, "loss": 0.0166, "step": 38444 }, { "epoch": 2.0496348030068776, "grad_norm": 2.6239449977874756, "learning_rate": 2.3112e-06, "loss": 0.0079, "step": 38445 }, { "epoch": 2.0496881164365304, "grad_norm": 1.2230610847473145, "learning_rate": 2.311e-06, "loss": 0.0155, "step": 38446 }, { "epoch": 2.049741429866183, "grad_norm": 2.4326364994049072, "learning_rate": 2.3108000000000004e-06, "loss": 0.0263, "step": 38447 }, { "epoch": 2.0497947432958363, "grad_norm": 2.165886402130127, "learning_rate": 2.3106e-06, "loss": 0.0295, "step": 38448 }, { "epoch": 2.049848056725489, "grad_norm": 2.369546413421631, "learning_rate": 2.3104e-06, "loss": 0.0284, "step": 38449 }, { "epoch": 2.0499013701551423, "grad_norm": 2.26951265335083, "learning_rate": 2.3102000000000003e-06, "loss": 0.0165, "step": 38450 }, { "epoch": 2.049954683584795, "grad_norm": 4.452753067016602, "learning_rate": 2.3100000000000003e-06, "loss": 0.0168, "step": 38451 }, { "epoch": 2.050007997014448, "grad_norm": 2.7881407737731934, "learning_rate": 2.3098e-06, "loss": 0.0331, "step": 38452 }, { "epoch": 2.050061310444101, "grad_norm": 2.3867385387420654, "learning_rate": 2.3096000000000003e-06, "loss": 0.0113, "step": 38453 }, { "epoch": 2.0501146238737538, "grad_norm": 4.051498889923096, "learning_rate": 2.3094000000000003e-06, "loss": 0.038, "step": 38454 }, { "epoch": 2.0501679373034065, "grad_norm": 2.7364485263824463, "learning_rate": 2.3092e-06, "loss": 0.0197, "step": 38455 }, { "epoch": 2.0502212507330597, "grad_norm": 0.831173837184906, "learning_rate": 2.3090000000000003e-06, "loss": 0.0055, "step": 38456 }, { "epoch": 2.0502745641627125, "grad_norm": 1.4140700101852417, "learning_rate": 2.3088000000000002e-06, "loss": 0.0082, "step": 38457 }, { "epoch": 2.0503278775923657, "grad_norm": 2.6592061519622803, "learning_rate": 2.3086000000000002e-06, "loss": 0.0384, "step": 38458 }, { "epoch": 2.0503811910220184, "grad_norm": 1.442407488822937, "learning_rate": 2.3084e-06, "loss": 0.0138, "step": 38459 }, { "epoch": 2.050434504451671, "grad_norm": 1.7915723323822021, "learning_rate": 2.3082e-06, "loss": 0.0079, "step": 38460 }, { "epoch": 2.0504878178813244, "grad_norm": 2.7974071502685547, "learning_rate": 2.308e-06, "loss": 0.0386, "step": 38461 }, { "epoch": 2.050541131310977, "grad_norm": 1.626493215560913, "learning_rate": 2.3078e-06, "loss": 0.0251, "step": 38462 }, { "epoch": 2.0505944447406304, "grad_norm": 1.1409261226654053, "learning_rate": 2.3076e-06, "loss": 0.0048, "step": 38463 }, { "epoch": 2.050647758170283, "grad_norm": 0.9735533595085144, "learning_rate": 2.3074e-06, "loss": 0.0058, "step": 38464 }, { "epoch": 2.050701071599936, "grad_norm": 0.6221168041229248, "learning_rate": 2.3072e-06, "loss": 0.0055, "step": 38465 }, { "epoch": 2.050754385029589, "grad_norm": 1.4006842374801636, "learning_rate": 2.307e-06, "loss": 0.0048, "step": 38466 }, { "epoch": 2.050807698459242, "grad_norm": 1.3433749675750732, "learning_rate": 2.3068e-06, "loss": 0.0147, "step": 38467 }, { "epoch": 2.050861011888895, "grad_norm": 2.2906899452209473, "learning_rate": 2.3066e-06, "loss": 0.0155, "step": 38468 }, { "epoch": 2.050914325318548, "grad_norm": 1.662534236907959, "learning_rate": 2.3064000000000005e-06, "loss": 0.0066, "step": 38469 }, { "epoch": 2.0509676387482005, "grad_norm": 2.2998108863830566, "learning_rate": 2.3062e-06, "loss": 0.0102, "step": 38470 }, { "epoch": 2.0510209521778537, "grad_norm": 0.4820075035095215, "learning_rate": 2.306e-06, "loss": 0.0035, "step": 38471 }, { "epoch": 2.0510742656075065, "grad_norm": 2.5965521335601807, "learning_rate": 2.3058000000000004e-06, "loss": 0.0164, "step": 38472 }, { "epoch": 2.0511275790371593, "grad_norm": 1.7866332530975342, "learning_rate": 2.3056e-06, "loss": 0.0295, "step": 38473 }, { "epoch": 2.0511808924668125, "grad_norm": 1.0891896486282349, "learning_rate": 2.3054e-06, "loss": 0.0128, "step": 38474 }, { "epoch": 2.051234205896465, "grad_norm": 4.607202053070068, "learning_rate": 2.3052000000000004e-06, "loss": 0.0699, "step": 38475 }, { "epoch": 2.0512875193261184, "grad_norm": 2.190584659576416, "learning_rate": 2.3050000000000004e-06, "loss": 0.0764, "step": 38476 }, { "epoch": 2.051340832755771, "grad_norm": 1.4921529293060303, "learning_rate": 2.3048e-06, "loss": 0.021, "step": 38477 }, { "epoch": 2.051394146185424, "grad_norm": 1.8193752765655518, "learning_rate": 2.3046000000000003e-06, "loss": 0.0196, "step": 38478 }, { "epoch": 2.051447459615077, "grad_norm": 2.3246140480041504, "learning_rate": 2.3044000000000003e-06, "loss": 0.0102, "step": 38479 }, { "epoch": 2.05150077304473, "grad_norm": 3.0404582023620605, "learning_rate": 2.3042000000000003e-06, "loss": 0.0428, "step": 38480 }, { "epoch": 2.051554086474383, "grad_norm": 2.169684886932373, "learning_rate": 2.3040000000000003e-06, "loss": 0.0168, "step": 38481 }, { "epoch": 2.051607399904036, "grad_norm": 0.9656902551651001, "learning_rate": 2.3038000000000003e-06, "loss": 0.005, "step": 38482 }, { "epoch": 2.0516607133336886, "grad_norm": 1.8272356986999512, "learning_rate": 2.3036000000000003e-06, "loss": 0.0128, "step": 38483 }, { "epoch": 2.051714026763342, "grad_norm": 1.531861424446106, "learning_rate": 2.3034000000000003e-06, "loss": 0.0303, "step": 38484 }, { "epoch": 2.0517673401929946, "grad_norm": 1.1570141315460205, "learning_rate": 2.3032000000000002e-06, "loss": 0.0107, "step": 38485 }, { "epoch": 2.0518206536226478, "grad_norm": 0.6669639945030212, "learning_rate": 2.3030000000000002e-06, "loss": 0.0055, "step": 38486 }, { "epoch": 2.0518739670523005, "grad_norm": 2.313685178756714, "learning_rate": 2.3028000000000002e-06, "loss": 0.0197, "step": 38487 }, { "epoch": 2.0519272804819533, "grad_norm": 1.7992831468582153, "learning_rate": 2.3026e-06, "loss": 0.011, "step": 38488 }, { "epoch": 2.0519805939116065, "grad_norm": 5.203640937805176, "learning_rate": 2.3024e-06, "loss": 0.021, "step": 38489 }, { "epoch": 2.0520339073412592, "grad_norm": 1.87994384765625, "learning_rate": 2.3022e-06, "loss": 0.0184, "step": 38490 }, { "epoch": 2.052087220770912, "grad_norm": 3.664301872253418, "learning_rate": 2.302e-06, "loss": 0.0472, "step": 38491 }, { "epoch": 2.052140534200565, "grad_norm": 0.8361335396766663, "learning_rate": 2.3018e-06, "loss": 0.0134, "step": 38492 }, { "epoch": 2.052193847630218, "grad_norm": 0.9076293110847473, "learning_rate": 2.3016e-06, "loss": 0.0053, "step": 38493 }, { "epoch": 2.052247161059871, "grad_norm": 2.717627763748169, "learning_rate": 2.3014e-06, "loss": 0.0143, "step": 38494 }, { "epoch": 2.052300474489524, "grad_norm": 1.2736133337020874, "learning_rate": 2.3012e-06, "loss": 0.0088, "step": 38495 }, { "epoch": 2.0523537879191767, "grad_norm": 1.4058325290679932, "learning_rate": 2.301e-06, "loss": 0.0182, "step": 38496 }, { "epoch": 2.05240710134883, "grad_norm": 1.4241887331008911, "learning_rate": 2.3008e-06, "loss": 0.011, "step": 38497 }, { "epoch": 2.0524604147784826, "grad_norm": 1.5826215744018555, "learning_rate": 2.3006e-06, "loss": 0.0262, "step": 38498 }, { "epoch": 2.052513728208136, "grad_norm": 1.9109959602355957, "learning_rate": 2.3004e-06, "loss": 0.0073, "step": 38499 }, { "epoch": 2.0525670416377886, "grad_norm": 4.038867473602295, "learning_rate": 2.3002e-06, "loss": 0.0372, "step": 38500 }, { "epoch": 2.0526203550674413, "grad_norm": 2.437345266342163, "learning_rate": 2.3000000000000004e-06, "loss": 0.052, "step": 38501 }, { "epoch": 2.0526736684970945, "grad_norm": 1.2591007947921753, "learning_rate": 2.2998e-06, "loss": 0.0073, "step": 38502 }, { "epoch": 2.0527269819267473, "grad_norm": 2.2634530067443848, "learning_rate": 2.2996e-06, "loss": 0.0196, "step": 38503 }, { "epoch": 2.0527802953564, "grad_norm": 1.3285073041915894, "learning_rate": 2.2994000000000004e-06, "loss": 0.0123, "step": 38504 }, { "epoch": 2.0528336087860533, "grad_norm": 1.2611844539642334, "learning_rate": 2.2992000000000004e-06, "loss": 0.0057, "step": 38505 }, { "epoch": 2.052886922215706, "grad_norm": 1.8329945802688599, "learning_rate": 2.299e-06, "loss": 0.0203, "step": 38506 }, { "epoch": 2.052940235645359, "grad_norm": 1.6079710721969604, "learning_rate": 2.2988000000000003e-06, "loss": 0.0114, "step": 38507 }, { "epoch": 2.052993549075012, "grad_norm": 3.4743459224700928, "learning_rate": 2.2986000000000003e-06, "loss": 0.0223, "step": 38508 }, { "epoch": 2.0530468625046647, "grad_norm": 1.9138920307159424, "learning_rate": 2.2984e-06, "loss": 0.0357, "step": 38509 }, { "epoch": 2.053100175934318, "grad_norm": 2.0004489421844482, "learning_rate": 2.2982000000000003e-06, "loss": 0.0206, "step": 38510 }, { "epoch": 2.0531534893639707, "grad_norm": 0.8447085022926331, "learning_rate": 2.2980000000000003e-06, "loss": 0.0049, "step": 38511 }, { "epoch": 2.053206802793624, "grad_norm": 4.0009613037109375, "learning_rate": 2.2978000000000003e-06, "loss": 0.0448, "step": 38512 }, { "epoch": 2.0532601162232766, "grad_norm": 1.1994646787643433, "learning_rate": 2.2976000000000002e-06, "loss": 0.0068, "step": 38513 }, { "epoch": 2.0533134296529294, "grad_norm": 2.432401418685913, "learning_rate": 2.2974000000000002e-06, "loss": 0.0132, "step": 38514 }, { "epoch": 2.0533667430825826, "grad_norm": 1.1105647087097168, "learning_rate": 2.2972000000000002e-06, "loss": 0.0116, "step": 38515 }, { "epoch": 2.0534200565122354, "grad_norm": 0.7830763459205627, "learning_rate": 2.297e-06, "loss": 0.0055, "step": 38516 }, { "epoch": 2.0534733699418886, "grad_norm": 1.7617876529693604, "learning_rate": 2.2968e-06, "loss": 0.0181, "step": 38517 }, { "epoch": 2.0535266833715413, "grad_norm": 3.3602521419525146, "learning_rate": 2.2966e-06, "loss": 0.0262, "step": 38518 }, { "epoch": 2.053579996801194, "grad_norm": 3.999717950820923, "learning_rate": 2.2964e-06, "loss": 0.0125, "step": 38519 }, { "epoch": 2.0536333102308473, "grad_norm": 0.4530872106552124, "learning_rate": 2.2962e-06, "loss": 0.0027, "step": 38520 }, { "epoch": 2.0536866236605, "grad_norm": 1.3501530885696411, "learning_rate": 2.296e-06, "loss": 0.0281, "step": 38521 }, { "epoch": 2.0537399370901532, "grad_norm": 0.37131932377815247, "learning_rate": 2.2958e-06, "loss": 0.0028, "step": 38522 }, { "epoch": 2.053793250519806, "grad_norm": 0.6460448503494263, "learning_rate": 2.2956e-06, "loss": 0.0038, "step": 38523 }, { "epoch": 2.0538465639494587, "grad_norm": 0.941858172416687, "learning_rate": 2.2954e-06, "loss": 0.0067, "step": 38524 }, { "epoch": 2.053899877379112, "grad_norm": 2.2084929943084717, "learning_rate": 2.2952e-06, "loss": 0.0124, "step": 38525 }, { "epoch": 2.0539531908087647, "grad_norm": 1.338113784790039, "learning_rate": 2.2950000000000005e-06, "loss": 0.0185, "step": 38526 }, { "epoch": 2.0540065042384175, "grad_norm": 2.28851580619812, "learning_rate": 2.2948e-06, "loss": 0.0192, "step": 38527 }, { "epoch": 2.0540598176680707, "grad_norm": 1.472198724746704, "learning_rate": 2.2946e-06, "loss": 0.0227, "step": 38528 }, { "epoch": 2.0541131310977234, "grad_norm": 1.713337779045105, "learning_rate": 2.2944000000000004e-06, "loss": 0.0069, "step": 38529 }, { "epoch": 2.0541664445273766, "grad_norm": 1.507818341255188, "learning_rate": 2.2942000000000004e-06, "loss": 0.0092, "step": 38530 }, { "epoch": 2.0542197579570294, "grad_norm": 1.3246510028839111, "learning_rate": 2.294e-06, "loss": 0.0154, "step": 38531 }, { "epoch": 2.054273071386682, "grad_norm": 1.3280898332595825, "learning_rate": 2.2938000000000004e-06, "loss": 0.0146, "step": 38532 }, { "epoch": 2.0543263848163353, "grad_norm": 0.3697285056114197, "learning_rate": 2.2936000000000004e-06, "loss": 0.0037, "step": 38533 }, { "epoch": 2.054379698245988, "grad_norm": 1.8920704126358032, "learning_rate": 2.2934e-06, "loss": 0.0337, "step": 38534 }, { "epoch": 2.0544330116756413, "grad_norm": 1.6003799438476562, "learning_rate": 2.2932000000000003e-06, "loss": 0.0088, "step": 38535 }, { "epoch": 2.054486325105294, "grad_norm": 3.0615930557250977, "learning_rate": 2.2930000000000003e-06, "loss": 0.0238, "step": 38536 }, { "epoch": 2.054539638534947, "grad_norm": 1.5091420412063599, "learning_rate": 2.2928000000000003e-06, "loss": 0.0053, "step": 38537 }, { "epoch": 2.0545929519646, "grad_norm": 2.846975803375244, "learning_rate": 2.2926000000000003e-06, "loss": 0.0355, "step": 38538 }, { "epoch": 2.0546462653942528, "grad_norm": 0.6165858507156372, "learning_rate": 2.2924000000000003e-06, "loss": 0.0043, "step": 38539 }, { "epoch": 2.0546995788239055, "grad_norm": 1.995816946029663, "learning_rate": 2.2922000000000003e-06, "loss": 0.013, "step": 38540 }, { "epoch": 2.0547528922535587, "grad_norm": 2.076803207397461, "learning_rate": 2.2920000000000002e-06, "loss": 0.0234, "step": 38541 }, { "epoch": 2.0548062056832115, "grad_norm": 1.1578751802444458, "learning_rate": 2.2918000000000002e-06, "loss": 0.0052, "step": 38542 }, { "epoch": 2.0548595191128647, "grad_norm": 1.0127705335617065, "learning_rate": 2.2916000000000002e-06, "loss": 0.0117, "step": 38543 }, { "epoch": 2.0549128325425174, "grad_norm": 1.0102670192718506, "learning_rate": 2.2914e-06, "loss": 0.0156, "step": 38544 }, { "epoch": 2.05496614597217, "grad_norm": 1.5728622674942017, "learning_rate": 2.2912e-06, "loss": 0.0067, "step": 38545 }, { "epoch": 2.0550194594018234, "grad_norm": 1.3322982788085938, "learning_rate": 2.291e-06, "loss": 0.0051, "step": 38546 }, { "epoch": 2.055072772831476, "grad_norm": 1.9977574348449707, "learning_rate": 2.2908e-06, "loss": 0.0115, "step": 38547 }, { "epoch": 2.0551260862611294, "grad_norm": 2.371680974960327, "learning_rate": 2.2906e-06, "loss": 0.0208, "step": 38548 }, { "epoch": 2.055179399690782, "grad_norm": 1.284767985343933, "learning_rate": 2.2904e-06, "loss": 0.0055, "step": 38549 }, { "epoch": 2.055232713120435, "grad_norm": 4.378462791442871, "learning_rate": 2.2902e-06, "loss": 0.0074, "step": 38550 }, { "epoch": 2.055286026550088, "grad_norm": 1.5728535652160645, "learning_rate": 2.29e-06, "loss": 0.0462, "step": 38551 }, { "epoch": 2.055339339979741, "grad_norm": 1.7131882905960083, "learning_rate": 2.2898e-06, "loss": 0.0209, "step": 38552 }, { "epoch": 2.055392653409394, "grad_norm": 1.252445936203003, "learning_rate": 2.2896e-06, "loss": 0.0251, "step": 38553 }, { "epoch": 2.055445966839047, "grad_norm": 0.8804505467414856, "learning_rate": 2.2894e-06, "loss": 0.0056, "step": 38554 }, { "epoch": 2.0554992802686995, "grad_norm": 2.165771007537842, "learning_rate": 2.2892000000000005e-06, "loss": 0.0495, "step": 38555 }, { "epoch": 2.0555525936983527, "grad_norm": 1.5309616327285767, "learning_rate": 2.289e-06, "loss": 0.0076, "step": 38556 }, { "epoch": 2.0556059071280055, "grad_norm": 1.1924161911010742, "learning_rate": 2.2888e-06, "loss": 0.0034, "step": 38557 }, { "epoch": 2.0556592205576583, "grad_norm": 3.351261615753174, "learning_rate": 2.2886000000000004e-06, "loss": 0.0097, "step": 38558 }, { "epoch": 2.0557125339873115, "grad_norm": 2.7100932598114014, "learning_rate": 2.2884e-06, "loss": 0.0247, "step": 38559 }, { "epoch": 2.055765847416964, "grad_norm": 4.461274147033691, "learning_rate": 2.2882e-06, "loss": 0.0308, "step": 38560 }, { "epoch": 2.0558191608466174, "grad_norm": 1.0644850730895996, "learning_rate": 2.2880000000000004e-06, "loss": 0.0146, "step": 38561 }, { "epoch": 2.05587247427627, "grad_norm": 1.108364224433899, "learning_rate": 2.2878000000000004e-06, "loss": 0.0058, "step": 38562 }, { "epoch": 2.055925787705923, "grad_norm": 1.1210047006607056, "learning_rate": 2.2876e-06, "loss": 0.023, "step": 38563 }, { "epoch": 2.055979101135576, "grad_norm": 1.66243577003479, "learning_rate": 2.2874000000000003e-06, "loss": 0.0297, "step": 38564 }, { "epoch": 2.056032414565229, "grad_norm": 2.3281466960906982, "learning_rate": 2.2872000000000003e-06, "loss": 0.0285, "step": 38565 }, { "epoch": 2.056085727994882, "grad_norm": 1.5290610790252686, "learning_rate": 2.287e-06, "loss": 0.0086, "step": 38566 }, { "epoch": 2.056139041424535, "grad_norm": 1.5962166786193848, "learning_rate": 2.2868000000000003e-06, "loss": 0.0087, "step": 38567 }, { "epoch": 2.0561923548541876, "grad_norm": 1.0871875286102295, "learning_rate": 2.2866000000000003e-06, "loss": 0.0118, "step": 38568 }, { "epoch": 2.056245668283841, "grad_norm": 1.9284566640853882, "learning_rate": 2.2864000000000003e-06, "loss": 0.0058, "step": 38569 }, { "epoch": 2.0562989817134936, "grad_norm": 0.7806461453437805, "learning_rate": 2.2862000000000002e-06, "loss": 0.0055, "step": 38570 }, { "epoch": 2.0563522951431468, "grad_norm": 0.7726860046386719, "learning_rate": 2.2860000000000002e-06, "loss": 0.0032, "step": 38571 }, { "epoch": 2.0564056085727995, "grad_norm": 3.659757375717163, "learning_rate": 2.2858e-06, "loss": 0.0633, "step": 38572 }, { "epoch": 2.0564589220024523, "grad_norm": 1.6623766422271729, "learning_rate": 2.2856e-06, "loss": 0.0115, "step": 38573 }, { "epoch": 2.0565122354321055, "grad_norm": 1.3854296207427979, "learning_rate": 2.2854e-06, "loss": 0.0291, "step": 38574 }, { "epoch": 2.0565655488617582, "grad_norm": 4.872340202331543, "learning_rate": 2.2852e-06, "loss": 0.0355, "step": 38575 }, { "epoch": 2.056618862291411, "grad_norm": 1.6744369268417358, "learning_rate": 2.285e-06, "loss": 0.0071, "step": 38576 }, { "epoch": 2.056672175721064, "grad_norm": 1.0888335704803467, "learning_rate": 2.2848e-06, "loss": 0.0207, "step": 38577 }, { "epoch": 2.056725489150717, "grad_norm": 0.9196985960006714, "learning_rate": 2.2846e-06, "loss": 0.0051, "step": 38578 }, { "epoch": 2.05677880258037, "grad_norm": 1.3392847776412964, "learning_rate": 2.2844e-06, "loss": 0.005, "step": 38579 }, { "epoch": 2.056832116010023, "grad_norm": 1.5012311935424805, "learning_rate": 2.2842000000000005e-06, "loss": 0.0254, "step": 38580 }, { "epoch": 2.0568854294396757, "grad_norm": 3.0962772369384766, "learning_rate": 2.284e-06, "loss": 0.0281, "step": 38581 }, { "epoch": 2.056938742869329, "grad_norm": 2.0200254917144775, "learning_rate": 2.2838e-06, "loss": 0.0106, "step": 38582 }, { "epoch": 2.0569920562989816, "grad_norm": 3.1232128143310547, "learning_rate": 2.2836000000000005e-06, "loss": 0.0312, "step": 38583 }, { "epoch": 2.057045369728635, "grad_norm": 1.6873939037322998, "learning_rate": 2.2834e-06, "loss": 0.0118, "step": 38584 }, { "epoch": 2.0570986831582876, "grad_norm": 2.675765037536621, "learning_rate": 2.2832e-06, "loss": 0.0205, "step": 38585 }, { "epoch": 2.0571519965879403, "grad_norm": 2.6096746921539307, "learning_rate": 2.2830000000000004e-06, "loss": 0.0454, "step": 38586 }, { "epoch": 2.0572053100175935, "grad_norm": 1.9347608089447021, "learning_rate": 2.2828000000000004e-06, "loss": 0.0094, "step": 38587 }, { "epoch": 2.0572586234472463, "grad_norm": 2.763660430908203, "learning_rate": 2.2826e-06, "loss": 0.0322, "step": 38588 }, { "epoch": 2.0573119368768995, "grad_norm": 5.956470489501953, "learning_rate": 2.2824000000000004e-06, "loss": 0.0141, "step": 38589 }, { "epoch": 2.0573652503065523, "grad_norm": 1.8945553302764893, "learning_rate": 2.2822000000000004e-06, "loss": 0.0222, "step": 38590 }, { "epoch": 2.057418563736205, "grad_norm": 1.140000581741333, "learning_rate": 2.282e-06, "loss": 0.0045, "step": 38591 }, { "epoch": 2.057471877165858, "grad_norm": 5.313379287719727, "learning_rate": 2.2818000000000003e-06, "loss": 0.0556, "step": 38592 }, { "epoch": 2.057525190595511, "grad_norm": 3.072202205657959, "learning_rate": 2.2816000000000003e-06, "loss": 0.0246, "step": 38593 }, { "epoch": 2.0575785040251637, "grad_norm": 0.9981867074966431, "learning_rate": 2.2814000000000003e-06, "loss": 0.0055, "step": 38594 }, { "epoch": 2.057631817454817, "grad_norm": 0.7046234011650085, "learning_rate": 2.2812000000000003e-06, "loss": 0.0039, "step": 38595 }, { "epoch": 2.0576851308844697, "grad_norm": 1.6067334413528442, "learning_rate": 2.2810000000000003e-06, "loss": 0.026, "step": 38596 }, { "epoch": 2.057738444314123, "grad_norm": 1.4231778383255005, "learning_rate": 2.2808000000000003e-06, "loss": 0.0173, "step": 38597 }, { "epoch": 2.0577917577437757, "grad_norm": 1.6927118301391602, "learning_rate": 2.2806000000000002e-06, "loss": 0.0262, "step": 38598 }, { "epoch": 2.0578450711734284, "grad_norm": 1.626587986946106, "learning_rate": 2.2804000000000002e-06, "loss": 0.0342, "step": 38599 }, { "epoch": 2.0578983846030816, "grad_norm": 1.9580944776535034, "learning_rate": 2.2802e-06, "loss": 0.0379, "step": 38600 }, { "epoch": 2.0579516980327344, "grad_norm": 2.2838022708892822, "learning_rate": 2.28e-06, "loss": 0.0075, "step": 38601 }, { "epoch": 2.0580050114623876, "grad_norm": 0.6019868850708008, "learning_rate": 2.2798e-06, "loss": 0.004, "step": 38602 }, { "epoch": 2.0580583248920403, "grad_norm": 2.8764567375183105, "learning_rate": 2.2796e-06, "loss": 0.0326, "step": 38603 }, { "epoch": 2.058111638321693, "grad_norm": 1.5444341897964478, "learning_rate": 2.2794e-06, "loss": 0.0054, "step": 38604 }, { "epoch": 2.0581649517513463, "grad_norm": 0.9726371169090271, "learning_rate": 2.2792e-06, "loss": 0.0052, "step": 38605 }, { "epoch": 2.058218265180999, "grad_norm": 1.5028823614120483, "learning_rate": 2.279e-06, "loss": 0.0101, "step": 38606 }, { "epoch": 2.0582715786106522, "grad_norm": 1.5151221752166748, "learning_rate": 2.2788e-06, "loss": 0.047, "step": 38607 }, { "epoch": 2.058324892040305, "grad_norm": 2.932471513748169, "learning_rate": 2.2786e-06, "loss": 0.0409, "step": 38608 }, { "epoch": 2.0583782054699578, "grad_norm": 2.671172857284546, "learning_rate": 2.2784e-06, "loss": 0.0137, "step": 38609 }, { "epoch": 2.058431518899611, "grad_norm": 1.0221351385116577, "learning_rate": 2.2782e-06, "loss": 0.0125, "step": 38610 }, { "epoch": 2.0584848323292637, "grad_norm": 4.324580192565918, "learning_rate": 2.278e-06, "loss": 0.0289, "step": 38611 }, { "epoch": 2.0585381457589165, "grad_norm": 0.5408445000648499, "learning_rate": 2.2778000000000005e-06, "loss": 0.0041, "step": 38612 }, { "epoch": 2.0585914591885697, "grad_norm": 1.4439871311187744, "learning_rate": 2.2776e-06, "loss": 0.0218, "step": 38613 }, { "epoch": 2.0586447726182224, "grad_norm": 2.4975638389587402, "learning_rate": 2.2774e-06, "loss": 0.0255, "step": 38614 }, { "epoch": 2.0586980860478756, "grad_norm": 1.6450536251068115, "learning_rate": 2.2772000000000004e-06, "loss": 0.0105, "step": 38615 }, { "epoch": 2.0587513994775284, "grad_norm": 3.280130386352539, "learning_rate": 2.2770000000000004e-06, "loss": 0.0335, "step": 38616 }, { "epoch": 2.058804712907181, "grad_norm": 2.52032470703125, "learning_rate": 2.2768e-06, "loss": 0.0305, "step": 38617 }, { "epoch": 2.0588580263368343, "grad_norm": 0.3328924775123596, "learning_rate": 2.2766000000000004e-06, "loss": 0.0032, "step": 38618 }, { "epoch": 2.058911339766487, "grad_norm": 3.9387381076812744, "learning_rate": 2.2764000000000003e-06, "loss": 0.0643, "step": 38619 }, { "epoch": 2.0589646531961403, "grad_norm": 0.6856793761253357, "learning_rate": 2.2762e-06, "loss": 0.0052, "step": 38620 }, { "epoch": 2.059017966625793, "grad_norm": 0.9649622440338135, "learning_rate": 2.2760000000000003e-06, "loss": 0.0057, "step": 38621 }, { "epoch": 2.059071280055446, "grad_norm": 2.138038158416748, "learning_rate": 2.2758000000000003e-06, "loss": 0.0344, "step": 38622 }, { "epoch": 2.059124593485099, "grad_norm": 2.1355128288269043, "learning_rate": 2.2756000000000003e-06, "loss": 0.0179, "step": 38623 }, { "epoch": 2.0591779069147518, "grad_norm": 1.2630096673965454, "learning_rate": 2.2754000000000003e-06, "loss": 0.0167, "step": 38624 }, { "epoch": 2.0592312203444045, "grad_norm": 0.5441550016403198, "learning_rate": 2.2752000000000003e-06, "loss": 0.0048, "step": 38625 }, { "epoch": 2.0592845337740577, "grad_norm": 1.040964126586914, "learning_rate": 2.2750000000000002e-06, "loss": 0.0081, "step": 38626 }, { "epoch": 2.0593378472037105, "grad_norm": 4.091915130615234, "learning_rate": 2.2748000000000002e-06, "loss": 0.0191, "step": 38627 }, { "epoch": 2.0593911606333637, "grad_norm": 2.2221477031707764, "learning_rate": 2.2746e-06, "loss": 0.015, "step": 38628 }, { "epoch": 2.0594444740630165, "grad_norm": 1.1615413427352905, "learning_rate": 2.2744e-06, "loss": 0.0084, "step": 38629 }, { "epoch": 2.059497787492669, "grad_norm": 1.5280100107192993, "learning_rate": 2.2742e-06, "loss": 0.0216, "step": 38630 }, { "epoch": 2.0595511009223224, "grad_norm": 2.423069953918457, "learning_rate": 2.274e-06, "loss": 0.029, "step": 38631 }, { "epoch": 2.059604414351975, "grad_norm": 1.3731005191802979, "learning_rate": 2.2738e-06, "loss": 0.025, "step": 38632 }, { "epoch": 2.0596577277816284, "grad_norm": 0.9040996432304382, "learning_rate": 2.2736e-06, "loss": 0.0068, "step": 38633 }, { "epoch": 2.059711041211281, "grad_norm": 1.3352612257003784, "learning_rate": 2.2734e-06, "loss": 0.0142, "step": 38634 }, { "epoch": 2.059764354640934, "grad_norm": 1.4304044246673584, "learning_rate": 2.2732e-06, "loss": 0.0154, "step": 38635 }, { "epoch": 2.059817668070587, "grad_norm": 1.3831164836883545, "learning_rate": 2.273e-06, "loss": 0.0101, "step": 38636 }, { "epoch": 2.05987098150024, "grad_norm": 2.1067557334899902, "learning_rate": 2.2728e-06, "loss": 0.0152, "step": 38637 }, { "epoch": 2.059924294929893, "grad_norm": 3.4926886558532715, "learning_rate": 2.2726e-06, "loss": 0.034, "step": 38638 }, { "epoch": 2.059977608359546, "grad_norm": 0.49747434258461, "learning_rate": 2.2724e-06, "loss": 0.0038, "step": 38639 }, { "epoch": 2.0600309217891986, "grad_norm": 2.223132610321045, "learning_rate": 2.2722e-06, "loss": 0.0247, "step": 38640 }, { "epoch": 2.0600842352188518, "grad_norm": 0.5301066040992737, "learning_rate": 2.2720000000000004e-06, "loss": 0.004, "step": 38641 }, { "epoch": 2.0601375486485045, "grad_norm": 1.740944504737854, "learning_rate": 2.2718e-06, "loss": 0.0108, "step": 38642 }, { "epoch": 2.0601908620781573, "grad_norm": 2.1780474185943604, "learning_rate": 2.2716e-06, "loss": 0.0204, "step": 38643 }, { "epoch": 2.0602441755078105, "grad_norm": 1.8524726629257202, "learning_rate": 2.2714000000000004e-06, "loss": 0.0396, "step": 38644 }, { "epoch": 2.0602974889374632, "grad_norm": 1.3853696584701538, "learning_rate": 2.2712e-06, "loss": 0.0143, "step": 38645 }, { "epoch": 2.0603508023671164, "grad_norm": 1.3134326934814453, "learning_rate": 2.271e-06, "loss": 0.005, "step": 38646 }, { "epoch": 2.060404115796769, "grad_norm": 2.8823740482330322, "learning_rate": 2.2708000000000003e-06, "loss": 0.0255, "step": 38647 }, { "epoch": 2.060457429226422, "grad_norm": 1.9807056188583374, "learning_rate": 2.2706000000000003e-06, "loss": 0.016, "step": 38648 }, { "epoch": 2.060510742656075, "grad_norm": 2.300429105758667, "learning_rate": 2.2704e-06, "loss": 0.0212, "step": 38649 }, { "epoch": 2.060564056085728, "grad_norm": 2.6128811836242676, "learning_rate": 2.2702000000000003e-06, "loss": 0.021, "step": 38650 }, { "epoch": 2.060617369515381, "grad_norm": 1.0588949918746948, "learning_rate": 2.2700000000000003e-06, "loss": 0.0143, "step": 38651 }, { "epoch": 2.060670682945034, "grad_norm": 1.9796491861343384, "learning_rate": 2.2698e-06, "loss": 0.0143, "step": 38652 }, { "epoch": 2.0607239963746866, "grad_norm": 0.41571730375289917, "learning_rate": 2.2696000000000003e-06, "loss": 0.0036, "step": 38653 }, { "epoch": 2.06077730980434, "grad_norm": 1.472765326499939, "learning_rate": 2.2694000000000002e-06, "loss": 0.0057, "step": 38654 }, { "epoch": 2.0608306232339926, "grad_norm": 4.101919174194336, "learning_rate": 2.2692000000000002e-06, "loss": 0.025, "step": 38655 }, { "epoch": 2.0608839366636458, "grad_norm": 1.5598994493484497, "learning_rate": 2.2690000000000002e-06, "loss": 0.0163, "step": 38656 }, { "epoch": 2.0609372500932985, "grad_norm": 0.7877026796340942, "learning_rate": 2.2688e-06, "loss": 0.007, "step": 38657 }, { "epoch": 2.0609905635229513, "grad_norm": 1.626364827156067, "learning_rate": 2.2686e-06, "loss": 0.0113, "step": 38658 }, { "epoch": 2.0610438769526045, "grad_norm": 1.5311453342437744, "learning_rate": 2.2684e-06, "loss": 0.0131, "step": 38659 }, { "epoch": 2.0610971903822572, "grad_norm": 0.8084088563919067, "learning_rate": 2.2682e-06, "loss": 0.0045, "step": 38660 }, { "epoch": 2.06115050381191, "grad_norm": 2.4773447513580322, "learning_rate": 2.268e-06, "loss": 0.041, "step": 38661 }, { "epoch": 2.061203817241563, "grad_norm": 1.5170979499816895, "learning_rate": 2.2678e-06, "loss": 0.0295, "step": 38662 }, { "epoch": 2.061257130671216, "grad_norm": 1.2892366647720337, "learning_rate": 2.2676e-06, "loss": 0.0111, "step": 38663 }, { "epoch": 2.061310444100869, "grad_norm": 2.2597262859344482, "learning_rate": 2.2674e-06, "loss": 0.0186, "step": 38664 }, { "epoch": 2.061363757530522, "grad_norm": 2.569945812225342, "learning_rate": 2.2672e-06, "loss": 0.0254, "step": 38665 }, { "epoch": 2.0614170709601747, "grad_norm": 1.5534354448318481, "learning_rate": 2.2670000000000005e-06, "loss": 0.0264, "step": 38666 }, { "epoch": 2.061470384389828, "grad_norm": 0.5913477540016174, "learning_rate": 2.2668e-06, "loss": 0.0031, "step": 38667 }, { "epoch": 2.0615236978194806, "grad_norm": 1.4422091245651245, "learning_rate": 2.2666e-06, "loss": 0.0076, "step": 38668 }, { "epoch": 2.061577011249134, "grad_norm": 1.4947208166122437, "learning_rate": 2.2664000000000004e-06, "loss": 0.035, "step": 38669 }, { "epoch": 2.0616303246787866, "grad_norm": 1.9564288854599, "learning_rate": 2.2662e-06, "loss": 0.0094, "step": 38670 }, { "epoch": 2.0616836381084394, "grad_norm": 0.8601798415184021, "learning_rate": 2.266e-06, "loss": 0.005, "step": 38671 }, { "epoch": 2.0617369515380926, "grad_norm": 2.1223819255828857, "learning_rate": 2.2658000000000004e-06, "loss": 0.0351, "step": 38672 }, { "epoch": 2.0617902649677453, "grad_norm": 2.001116991043091, "learning_rate": 2.2656000000000004e-06, "loss": 0.0139, "step": 38673 }, { "epoch": 2.0618435783973985, "grad_norm": 1.6483365297317505, "learning_rate": 2.2654e-06, "loss": 0.0242, "step": 38674 }, { "epoch": 2.0618968918270513, "grad_norm": 1.0503889322280884, "learning_rate": 2.2652000000000004e-06, "loss": 0.0072, "step": 38675 }, { "epoch": 2.061950205256704, "grad_norm": 0.9588408470153809, "learning_rate": 2.2650000000000003e-06, "loss": 0.0056, "step": 38676 }, { "epoch": 2.0620035186863572, "grad_norm": 1.7316457033157349, "learning_rate": 2.2648e-06, "loss": 0.0083, "step": 38677 }, { "epoch": 2.06205683211601, "grad_norm": 2.152773857116699, "learning_rate": 2.2646000000000003e-06, "loss": 0.0174, "step": 38678 }, { "epoch": 2.0621101455456627, "grad_norm": 3.4676949977874756, "learning_rate": 2.2644000000000003e-06, "loss": 0.0224, "step": 38679 }, { "epoch": 2.062163458975316, "grad_norm": 2.668243885040283, "learning_rate": 2.2642000000000003e-06, "loss": 0.0229, "step": 38680 }, { "epoch": 2.0622167724049687, "grad_norm": 1.0862483978271484, "learning_rate": 2.2640000000000003e-06, "loss": 0.0149, "step": 38681 }, { "epoch": 2.062270085834622, "grad_norm": 0.9960184097290039, "learning_rate": 2.2638000000000002e-06, "loss": 0.005, "step": 38682 }, { "epoch": 2.0623233992642747, "grad_norm": 3.1661486625671387, "learning_rate": 2.2636000000000002e-06, "loss": 0.0145, "step": 38683 }, { "epoch": 2.0623767126939274, "grad_norm": 4.916634559631348, "learning_rate": 2.2634000000000002e-06, "loss": 0.0165, "step": 38684 }, { "epoch": 2.0624300261235806, "grad_norm": 1.740097165107727, "learning_rate": 2.2632e-06, "loss": 0.0193, "step": 38685 }, { "epoch": 2.0624833395532334, "grad_norm": 2.1389529705047607, "learning_rate": 2.263e-06, "loss": 0.0108, "step": 38686 }, { "epoch": 2.0625366529828866, "grad_norm": 3.003814220428467, "learning_rate": 2.2628e-06, "loss": 0.0404, "step": 38687 }, { "epoch": 2.0625899664125393, "grad_norm": 2.9484665393829346, "learning_rate": 2.2626e-06, "loss": 0.0547, "step": 38688 }, { "epoch": 2.062643279842192, "grad_norm": 1.5328277349472046, "learning_rate": 2.2624e-06, "loss": 0.0073, "step": 38689 }, { "epoch": 2.0626965932718453, "grad_norm": 0.9766458868980408, "learning_rate": 2.2622e-06, "loss": 0.005, "step": 38690 }, { "epoch": 2.062749906701498, "grad_norm": 1.483013391494751, "learning_rate": 2.262e-06, "loss": 0.0067, "step": 38691 }, { "epoch": 2.0628032201311512, "grad_norm": 3.107225179672241, "learning_rate": 2.2618e-06, "loss": 0.0143, "step": 38692 }, { "epoch": 2.062856533560804, "grad_norm": 1.4528905153274536, "learning_rate": 2.2616e-06, "loss": 0.0222, "step": 38693 }, { "epoch": 2.0629098469904568, "grad_norm": 4.016918659210205, "learning_rate": 2.2614e-06, "loss": 0.0193, "step": 38694 }, { "epoch": 2.06296316042011, "grad_norm": 1.328493356704712, "learning_rate": 2.2612e-06, "loss": 0.0079, "step": 38695 }, { "epoch": 2.0630164738497627, "grad_norm": 1.655013084411621, "learning_rate": 2.261e-06, "loss": 0.021, "step": 38696 }, { "epoch": 2.0630697872794155, "grad_norm": 1.1823041439056396, "learning_rate": 2.2608e-06, "loss": 0.0127, "step": 38697 }, { "epoch": 2.0631231007090687, "grad_norm": 2.111290216445923, "learning_rate": 2.2606000000000004e-06, "loss": 0.0328, "step": 38698 }, { "epoch": 2.0631764141387214, "grad_norm": 1.9816280603408813, "learning_rate": 2.2604e-06, "loss": 0.018, "step": 38699 }, { "epoch": 2.0632297275683746, "grad_norm": 2.210153818130493, "learning_rate": 2.2602e-06, "loss": 0.0298, "step": 38700 }, { "epoch": 2.0632830409980274, "grad_norm": 1.3082351684570312, "learning_rate": 2.2600000000000004e-06, "loss": 0.0074, "step": 38701 }, { "epoch": 2.06333635442768, "grad_norm": 1.7491580247879028, "learning_rate": 2.2598e-06, "loss": 0.0144, "step": 38702 }, { "epoch": 2.0633896678573334, "grad_norm": 1.5794743299484253, "learning_rate": 2.2596e-06, "loss": 0.0135, "step": 38703 }, { "epoch": 2.063442981286986, "grad_norm": 0.43409961462020874, "learning_rate": 2.2594000000000003e-06, "loss": 0.0038, "step": 38704 }, { "epoch": 2.0634962947166393, "grad_norm": 0.515568733215332, "learning_rate": 2.2592000000000003e-06, "loss": 0.0037, "step": 38705 }, { "epoch": 2.063549608146292, "grad_norm": 2.1916189193725586, "learning_rate": 2.259e-06, "loss": 0.0112, "step": 38706 }, { "epoch": 2.063602921575945, "grad_norm": 1.86972177028656, "learning_rate": 2.2588000000000003e-06, "loss": 0.0105, "step": 38707 }, { "epoch": 2.063656235005598, "grad_norm": 1.5019816160202026, "learning_rate": 2.2586000000000003e-06, "loss": 0.0107, "step": 38708 }, { "epoch": 2.063709548435251, "grad_norm": 1.5708415508270264, "learning_rate": 2.2584000000000003e-06, "loss": 0.0339, "step": 38709 }, { "epoch": 2.0637628618649035, "grad_norm": 1.8811575174331665, "learning_rate": 2.2582000000000002e-06, "loss": 0.0213, "step": 38710 }, { "epoch": 2.0638161752945567, "grad_norm": 3.036928653717041, "learning_rate": 2.2580000000000002e-06, "loss": 0.0305, "step": 38711 }, { "epoch": 2.0638694887242095, "grad_norm": 2.5721793174743652, "learning_rate": 2.2578000000000002e-06, "loss": 0.0228, "step": 38712 }, { "epoch": 2.0639228021538627, "grad_norm": 2.185767412185669, "learning_rate": 2.2576e-06, "loss": 0.0333, "step": 38713 }, { "epoch": 2.0639761155835155, "grad_norm": 2.9872586727142334, "learning_rate": 2.2574e-06, "loss": 0.019, "step": 38714 }, { "epoch": 2.064029429013168, "grad_norm": 1.738828182220459, "learning_rate": 2.2572e-06, "loss": 0.0094, "step": 38715 }, { "epoch": 2.0640827424428214, "grad_norm": 2.1508243083953857, "learning_rate": 2.257e-06, "loss": 0.0174, "step": 38716 }, { "epoch": 2.064136055872474, "grad_norm": 1.1384515762329102, "learning_rate": 2.2568e-06, "loss": 0.0056, "step": 38717 }, { "epoch": 2.0641893693021274, "grad_norm": 1.7361078262329102, "learning_rate": 2.2566e-06, "loss": 0.0124, "step": 38718 }, { "epoch": 2.06424268273178, "grad_norm": 3.8066978454589844, "learning_rate": 2.2564e-06, "loss": 0.0192, "step": 38719 }, { "epoch": 2.064295996161433, "grad_norm": 2.3797097206115723, "learning_rate": 2.2562e-06, "loss": 0.0159, "step": 38720 }, { "epoch": 2.064349309591086, "grad_norm": 1.9675421714782715, "learning_rate": 2.256e-06, "loss": 0.0196, "step": 38721 }, { "epoch": 2.064402623020739, "grad_norm": 2.3978567123413086, "learning_rate": 2.2558e-06, "loss": 0.0254, "step": 38722 }, { "epoch": 2.064455936450392, "grad_norm": 1.065573811531067, "learning_rate": 2.2556000000000005e-06, "loss": 0.006, "step": 38723 }, { "epoch": 2.064509249880045, "grad_norm": 1.844976782798767, "learning_rate": 2.2554e-06, "loss": 0.0125, "step": 38724 }, { "epoch": 2.0645625633096976, "grad_norm": 0.4797643721103668, "learning_rate": 2.2552e-06, "loss": 0.0028, "step": 38725 }, { "epoch": 2.0646158767393508, "grad_norm": 2.058793306350708, "learning_rate": 2.2550000000000004e-06, "loss": 0.0241, "step": 38726 }, { "epoch": 2.0646691901690035, "grad_norm": 1.8841890096664429, "learning_rate": 2.2548000000000004e-06, "loss": 0.0182, "step": 38727 }, { "epoch": 2.0647225035986567, "grad_norm": 6.859442234039307, "learning_rate": 2.2546e-06, "loss": 0.0242, "step": 38728 }, { "epoch": 2.0647758170283095, "grad_norm": 1.415946364402771, "learning_rate": 2.2544000000000004e-06, "loss": 0.0205, "step": 38729 }, { "epoch": 2.0648291304579622, "grad_norm": 2.086531639099121, "learning_rate": 2.2542000000000004e-06, "loss": 0.0129, "step": 38730 }, { "epoch": 2.0648824438876154, "grad_norm": 2.762741804122925, "learning_rate": 2.254e-06, "loss": 0.0118, "step": 38731 }, { "epoch": 2.064935757317268, "grad_norm": 2.2326645851135254, "learning_rate": 2.2538000000000003e-06, "loss": 0.029, "step": 38732 }, { "epoch": 2.064989070746921, "grad_norm": 0.5835099220275879, "learning_rate": 2.2536000000000003e-06, "loss": 0.003, "step": 38733 }, { "epoch": 2.065042384176574, "grad_norm": 2.250730514526367, "learning_rate": 2.2534000000000003e-06, "loss": 0.0256, "step": 38734 }, { "epoch": 2.065095697606227, "grad_norm": 3.6530508995056152, "learning_rate": 2.2532000000000003e-06, "loss": 0.0127, "step": 38735 }, { "epoch": 2.06514901103588, "grad_norm": 3.2247886657714844, "learning_rate": 2.2530000000000003e-06, "loss": 0.026, "step": 38736 }, { "epoch": 2.065202324465533, "grad_norm": 1.9754022359848022, "learning_rate": 2.2528000000000003e-06, "loss": 0.0333, "step": 38737 }, { "epoch": 2.0652556378951856, "grad_norm": 0.9254150986671448, "learning_rate": 2.2526000000000003e-06, "loss": 0.0037, "step": 38738 }, { "epoch": 2.065308951324839, "grad_norm": 1.7347030639648438, "learning_rate": 2.2524000000000002e-06, "loss": 0.0217, "step": 38739 }, { "epoch": 2.0653622647544916, "grad_norm": 1.465490460395813, "learning_rate": 2.2522000000000002e-06, "loss": 0.0384, "step": 38740 }, { "epoch": 2.065415578184145, "grad_norm": 0.6735720634460449, "learning_rate": 2.252e-06, "loss": 0.0055, "step": 38741 }, { "epoch": 2.0654688916137975, "grad_norm": 3.9016594886779785, "learning_rate": 2.2518e-06, "loss": 0.0529, "step": 38742 }, { "epoch": 2.0655222050434503, "grad_norm": 0.5389813184738159, "learning_rate": 2.2516e-06, "loss": 0.0039, "step": 38743 }, { "epoch": 2.0655755184731035, "grad_norm": 4.242701053619385, "learning_rate": 2.2514e-06, "loss": 0.0425, "step": 38744 }, { "epoch": 2.0656288319027563, "grad_norm": 1.265586018562317, "learning_rate": 2.2512e-06, "loss": 0.0151, "step": 38745 }, { "epoch": 2.065682145332409, "grad_norm": 1.9202027320861816, "learning_rate": 2.251e-06, "loss": 0.0097, "step": 38746 }, { "epoch": 2.065735458762062, "grad_norm": 0.6280908584594727, "learning_rate": 2.2508e-06, "loss": 0.0051, "step": 38747 }, { "epoch": 2.065788772191715, "grad_norm": 0.5575144290924072, "learning_rate": 2.2506e-06, "loss": 0.004, "step": 38748 }, { "epoch": 2.065842085621368, "grad_norm": 2.1225128173828125, "learning_rate": 2.2504e-06, "loss": 0.0165, "step": 38749 }, { "epoch": 2.065895399051021, "grad_norm": 1.9354888200759888, "learning_rate": 2.2502e-06, "loss": 0.0256, "step": 38750 }, { "epoch": 2.0659487124806737, "grad_norm": 1.5072216987609863, "learning_rate": 2.25e-06, "loss": 0.0076, "step": 38751 }, { "epoch": 2.066002025910327, "grad_norm": 1.3522976636886597, "learning_rate": 2.2498000000000005e-06, "loss": 0.0081, "step": 38752 }, { "epoch": 2.0660553393399796, "grad_norm": 2.541883945465088, "learning_rate": 2.2496e-06, "loss": 0.0133, "step": 38753 }, { "epoch": 2.066108652769633, "grad_norm": 1.6534063816070557, "learning_rate": 2.2494e-06, "loss": 0.0066, "step": 38754 }, { "epoch": 2.0661619661992856, "grad_norm": 2.526977777481079, "learning_rate": 2.2492000000000004e-06, "loss": 0.0144, "step": 38755 }, { "epoch": 2.0662152796289384, "grad_norm": 2.097790241241455, "learning_rate": 2.249e-06, "loss": 0.0246, "step": 38756 }, { "epoch": 2.0662685930585916, "grad_norm": 1.2887651920318604, "learning_rate": 2.2488e-06, "loss": 0.0137, "step": 38757 }, { "epoch": 2.0663219064882443, "grad_norm": 1.2818937301635742, "learning_rate": 2.2486000000000004e-06, "loss": 0.0051, "step": 38758 }, { "epoch": 2.0663752199178975, "grad_norm": 0.22906522452831268, "learning_rate": 2.2484000000000004e-06, "loss": 0.0024, "step": 38759 }, { "epoch": 2.0664285333475503, "grad_norm": 1.5204018354415894, "learning_rate": 2.2482e-06, "loss": 0.0098, "step": 38760 }, { "epoch": 2.066481846777203, "grad_norm": 2.867827892303467, "learning_rate": 2.2480000000000003e-06, "loss": 0.0352, "step": 38761 }, { "epoch": 2.0665351602068562, "grad_norm": 0.29876092076301575, "learning_rate": 2.2478000000000003e-06, "loss": 0.0026, "step": 38762 }, { "epoch": 2.066588473636509, "grad_norm": 2.0666561126708984, "learning_rate": 2.2476e-06, "loss": 0.0078, "step": 38763 }, { "epoch": 2.066641787066162, "grad_norm": 0.8897128701210022, "learning_rate": 2.2474000000000003e-06, "loss": 0.015, "step": 38764 }, { "epoch": 2.066695100495815, "grad_norm": 1.9402681589126587, "learning_rate": 2.2472000000000003e-06, "loss": 0.0094, "step": 38765 }, { "epoch": 2.0667484139254677, "grad_norm": 1.938780665397644, "learning_rate": 2.2470000000000003e-06, "loss": 0.0064, "step": 38766 }, { "epoch": 2.066801727355121, "grad_norm": 2.304446220397949, "learning_rate": 2.2468000000000002e-06, "loss": 0.0217, "step": 38767 }, { "epoch": 2.0668550407847737, "grad_norm": 1.90867280960083, "learning_rate": 2.2466000000000002e-06, "loss": 0.0135, "step": 38768 }, { "epoch": 2.0669083542144264, "grad_norm": 1.7491930723190308, "learning_rate": 2.2464e-06, "loss": 0.0192, "step": 38769 }, { "epoch": 2.0669616676440796, "grad_norm": 2.709317445755005, "learning_rate": 2.2462e-06, "loss": 0.0258, "step": 38770 }, { "epoch": 2.0670149810737324, "grad_norm": 2.0054781436920166, "learning_rate": 2.246e-06, "loss": 0.0075, "step": 38771 }, { "epoch": 2.0670682945033856, "grad_norm": 1.287423014640808, "learning_rate": 2.2458e-06, "loss": 0.0067, "step": 38772 }, { "epoch": 2.0671216079330383, "grad_norm": 4.231491565704346, "learning_rate": 2.2456e-06, "loss": 0.0398, "step": 38773 }, { "epoch": 2.067174921362691, "grad_norm": 1.4469830989837646, "learning_rate": 2.2454e-06, "loss": 0.0102, "step": 38774 }, { "epoch": 2.0672282347923443, "grad_norm": 1.5518356561660767, "learning_rate": 2.2452e-06, "loss": 0.0077, "step": 38775 }, { "epoch": 2.067281548221997, "grad_norm": 2.7356672286987305, "learning_rate": 2.245e-06, "loss": 0.0427, "step": 38776 }, { "epoch": 2.0673348616516503, "grad_norm": 2.6809165477752686, "learning_rate": 2.2448000000000005e-06, "loss": 0.0221, "step": 38777 }, { "epoch": 2.067388175081303, "grad_norm": 2.0084869861602783, "learning_rate": 2.2446e-06, "loss": 0.0154, "step": 38778 }, { "epoch": 2.0674414885109558, "grad_norm": 0.8068354725837708, "learning_rate": 2.2444e-06, "loss": 0.0036, "step": 38779 }, { "epoch": 2.067494801940609, "grad_norm": 2.6241915225982666, "learning_rate": 2.2442000000000005e-06, "loss": 0.0181, "step": 38780 }, { "epoch": 2.0675481153702617, "grad_norm": 2.334581136703491, "learning_rate": 2.244e-06, "loss": 0.0218, "step": 38781 }, { "epoch": 2.0676014287999145, "grad_norm": 2.1035497188568115, "learning_rate": 2.2438e-06, "loss": 0.0105, "step": 38782 }, { "epoch": 2.0676547422295677, "grad_norm": 2.95560884475708, "learning_rate": 2.2436000000000004e-06, "loss": 0.0183, "step": 38783 }, { "epoch": 2.0677080556592204, "grad_norm": 1.0048714876174927, "learning_rate": 2.2434000000000004e-06, "loss": 0.014, "step": 38784 }, { "epoch": 2.0677613690888736, "grad_norm": 0.9446617960929871, "learning_rate": 2.2432e-06, "loss": 0.0042, "step": 38785 }, { "epoch": 2.0678146825185264, "grad_norm": 1.2359609603881836, "learning_rate": 2.2430000000000004e-06, "loss": 0.0051, "step": 38786 }, { "epoch": 2.067867995948179, "grad_norm": 1.721469759941101, "learning_rate": 2.2428000000000004e-06, "loss": 0.0161, "step": 38787 }, { "epoch": 2.0679213093778324, "grad_norm": 2.3056304454803467, "learning_rate": 2.2426e-06, "loss": 0.0058, "step": 38788 }, { "epoch": 2.067974622807485, "grad_norm": 1.5803104639053345, "learning_rate": 2.2424000000000003e-06, "loss": 0.0152, "step": 38789 }, { "epoch": 2.0680279362371383, "grad_norm": 1.44821035861969, "learning_rate": 2.2422000000000003e-06, "loss": 0.0075, "step": 38790 }, { "epoch": 2.068081249666791, "grad_norm": 3.3670389652252197, "learning_rate": 2.2420000000000003e-06, "loss": 0.0144, "step": 38791 }, { "epoch": 2.068134563096444, "grad_norm": 2.004765510559082, "learning_rate": 2.2418000000000003e-06, "loss": 0.0228, "step": 38792 }, { "epoch": 2.068187876526097, "grad_norm": 2.67391037940979, "learning_rate": 2.2416000000000003e-06, "loss": 0.0247, "step": 38793 }, { "epoch": 2.06824118995575, "grad_norm": 2.5767390727996826, "learning_rate": 2.2414000000000003e-06, "loss": 0.0523, "step": 38794 }, { "epoch": 2.068294503385403, "grad_norm": 1.8152949810028076, "learning_rate": 2.2412000000000002e-06, "loss": 0.0143, "step": 38795 }, { "epoch": 2.0683478168150558, "grad_norm": 1.363427758216858, "learning_rate": 2.2410000000000002e-06, "loss": 0.0089, "step": 38796 }, { "epoch": 2.0684011302447085, "grad_norm": 0.38196906447410583, "learning_rate": 2.2408000000000002e-06, "loss": 0.0029, "step": 38797 }, { "epoch": 2.0684544436743617, "grad_norm": 1.9131067991256714, "learning_rate": 2.2406e-06, "loss": 0.0214, "step": 38798 }, { "epoch": 2.0685077571040145, "grad_norm": 1.6333909034729004, "learning_rate": 2.2404e-06, "loss": 0.0407, "step": 38799 }, { "epoch": 2.0685610705336672, "grad_norm": 1.9329092502593994, "learning_rate": 2.2402e-06, "loss": 0.013, "step": 38800 }, { "epoch": 2.0686143839633204, "grad_norm": 1.5429584980010986, "learning_rate": 2.24e-06, "loss": 0.0092, "step": 38801 }, { "epoch": 2.068667697392973, "grad_norm": 3.089189291000366, "learning_rate": 2.2398e-06, "loss": 0.0131, "step": 38802 }, { "epoch": 2.0687210108226264, "grad_norm": 4.099119663238525, "learning_rate": 2.2396e-06, "loss": 0.0235, "step": 38803 }, { "epoch": 2.068774324252279, "grad_norm": 1.9087656736373901, "learning_rate": 2.2394e-06, "loss": 0.0085, "step": 38804 }, { "epoch": 2.068827637681932, "grad_norm": 1.307895302772522, "learning_rate": 2.2392e-06, "loss": 0.0064, "step": 38805 }, { "epoch": 2.068880951111585, "grad_norm": 2.5127711296081543, "learning_rate": 2.239e-06, "loss": 0.0139, "step": 38806 }, { "epoch": 2.068934264541238, "grad_norm": 1.679021954536438, "learning_rate": 2.2388e-06, "loss": 0.0117, "step": 38807 }, { "epoch": 2.068987577970891, "grad_norm": 1.6746984720230103, "learning_rate": 2.2386e-06, "loss": 0.0071, "step": 38808 }, { "epoch": 2.069040891400544, "grad_norm": 1.6570765972137451, "learning_rate": 2.2384000000000005e-06, "loss": 0.0074, "step": 38809 }, { "epoch": 2.0690942048301966, "grad_norm": 1.2117496728897095, "learning_rate": 2.2382e-06, "loss": 0.0116, "step": 38810 }, { "epoch": 2.0691475182598498, "grad_norm": 1.5666791200637817, "learning_rate": 2.238e-06, "loss": 0.0147, "step": 38811 }, { "epoch": 2.0692008316895025, "grad_norm": 3.0836915969848633, "learning_rate": 2.2378000000000004e-06, "loss": 0.0301, "step": 38812 }, { "epoch": 2.0692541451191557, "grad_norm": 3.134023427963257, "learning_rate": 2.2376e-06, "loss": 0.0137, "step": 38813 }, { "epoch": 2.0693074585488085, "grad_norm": 3.0885379314422607, "learning_rate": 2.2374e-06, "loss": 0.0439, "step": 38814 }, { "epoch": 2.0693607719784612, "grad_norm": 0.45012524724006653, "learning_rate": 2.2372000000000004e-06, "loss": 0.003, "step": 38815 }, { "epoch": 2.0694140854081144, "grad_norm": 1.932837963104248, "learning_rate": 2.2370000000000004e-06, "loss": 0.0108, "step": 38816 }, { "epoch": 2.069467398837767, "grad_norm": 5.557567119598389, "learning_rate": 2.2368e-06, "loss": 0.0139, "step": 38817 }, { "epoch": 2.06952071226742, "grad_norm": 3.2810707092285156, "learning_rate": 2.2366000000000003e-06, "loss": 0.0328, "step": 38818 }, { "epoch": 2.069574025697073, "grad_norm": 1.8756980895996094, "learning_rate": 2.2364000000000003e-06, "loss": 0.007, "step": 38819 }, { "epoch": 2.069627339126726, "grad_norm": 1.2383092641830444, "learning_rate": 2.2362000000000003e-06, "loss": 0.0069, "step": 38820 }, { "epoch": 2.069680652556379, "grad_norm": 1.7761080265045166, "learning_rate": 2.2360000000000003e-06, "loss": 0.0083, "step": 38821 }, { "epoch": 2.069733965986032, "grad_norm": 2.5049657821655273, "learning_rate": 2.2358000000000003e-06, "loss": 0.0091, "step": 38822 }, { "epoch": 2.0697872794156846, "grad_norm": 2.464026927947998, "learning_rate": 2.2356000000000002e-06, "loss": 0.0201, "step": 38823 }, { "epoch": 2.069840592845338, "grad_norm": 3.702063798904419, "learning_rate": 2.2354000000000002e-06, "loss": 0.0493, "step": 38824 }, { "epoch": 2.0698939062749906, "grad_norm": 0.43382588028907776, "learning_rate": 2.2352000000000002e-06, "loss": 0.0026, "step": 38825 }, { "epoch": 2.069947219704644, "grad_norm": 1.9643694162368774, "learning_rate": 2.235e-06, "loss": 0.0183, "step": 38826 }, { "epoch": 2.0700005331342965, "grad_norm": 3.773726463317871, "learning_rate": 2.2348e-06, "loss": 0.013, "step": 38827 }, { "epoch": 2.0700538465639493, "grad_norm": 3.326908826828003, "learning_rate": 2.2346e-06, "loss": 0.0134, "step": 38828 }, { "epoch": 2.0701071599936025, "grad_norm": 1.8055170774459839, "learning_rate": 2.2344e-06, "loss": 0.0178, "step": 38829 }, { "epoch": 2.0701604734232553, "grad_norm": 2.027658462524414, "learning_rate": 2.2342e-06, "loss": 0.007, "step": 38830 }, { "epoch": 2.070213786852908, "grad_norm": 1.9419671297073364, "learning_rate": 2.234e-06, "loss": 0.0086, "step": 38831 }, { "epoch": 2.0702671002825612, "grad_norm": 1.5818883180618286, "learning_rate": 2.2338e-06, "loss": 0.0299, "step": 38832 }, { "epoch": 2.070320413712214, "grad_norm": 0.767284631729126, "learning_rate": 2.2336e-06, "loss": 0.0035, "step": 38833 }, { "epoch": 2.070373727141867, "grad_norm": 2.0884742736816406, "learning_rate": 2.2334e-06, "loss": 0.0074, "step": 38834 }, { "epoch": 2.07042704057152, "grad_norm": 1.1937562227249146, "learning_rate": 2.2332e-06, "loss": 0.028, "step": 38835 }, { "epoch": 2.0704803540011727, "grad_norm": 0.5730286240577698, "learning_rate": 2.233e-06, "loss": 0.0028, "step": 38836 }, { "epoch": 2.070533667430826, "grad_norm": 1.6545321941375732, "learning_rate": 2.2328e-06, "loss": 0.0181, "step": 38837 }, { "epoch": 2.0705869808604787, "grad_norm": 2.089771032333374, "learning_rate": 2.2326e-06, "loss": 0.0339, "step": 38838 }, { "epoch": 2.070640294290132, "grad_norm": 0.2880468964576721, "learning_rate": 2.2324e-06, "loss": 0.0029, "step": 38839 }, { "epoch": 2.0706936077197846, "grad_norm": 2.425244092941284, "learning_rate": 2.2322e-06, "loss": 0.03, "step": 38840 }, { "epoch": 2.0707469211494374, "grad_norm": 1.9018112421035767, "learning_rate": 2.2320000000000004e-06, "loss": 0.015, "step": 38841 }, { "epoch": 2.0708002345790906, "grad_norm": 2.787594795227051, "learning_rate": 2.2318e-06, "loss": 0.0621, "step": 38842 }, { "epoch": 2.0708535480087433, "grad_norm": 2.467219352722168, "learning_rate": 2.2316e-06, "loss": 0.0342, "step": 38843 }, { "epoch": 2.0709068614383965, "grad_norm": 0.5676456093788147, "learning_rate": 2.2314000000000004e-06, "loss": 0.0039, "step": 38844 }, { "epoch": 2.0709601748680493, "grad_norm": 0.8787720799446106, "learning_rate": 2.2312000000000003e-06, "loss": 0.0055, "step": 38845 }, { "epoch": 2.071013488297702, "grad_norm": 2.084754705429077, "learning_rate": 2.231e-06, "loss": 0.0326, "step": 38846 }, { "epoch": 2.0710668017273552, "grad_norm": 1.738493800163269, "learning_rate": 2.2308000000000003e-06, "loss": 0.0259, "step": 38847 }, { "epoch": 2.071120115157008, "grad_norm": 1.6457959413528442, "learning_rate": 2.2306000000000003e-06, "loss": 0.0259, "step": 38848 }, { "epoch": 2.071173428586661, "grad_norm": 1.7054873704910278, "learning_rate": 2.2304e-06, "loss": 0.0302, "step": 38849 }, { "epoch": 2.071226742016314, "grad_norm": 0.8700652122497559, "learning_rate": 2.2302000000000003e-06, "loss": 0.0065, "step": 38850 }, { "epoch": 2.0712800554459667, "grad_norm": 2.909635305404663, "learning_rate": 2.2300000000000002e-06, "loss": 0.0097, "step": 38851 }, { "epoch": 2.07133336887562, "grad_norm": 2.6973860263824463, "learning_rate": 2.2298000000000002e-06, "loss": 0.0375, "step": 38852 }, { "epoch": 2.0713866823052727, "grad_norm": 1.982970952987671, "learning_rate": 2.2296000000000002e-06, "loss": 0.0226, "step": 38853 }, { "epoch": 2.0714399957349254, "grad_norm": 3.7134015560150146, "learning_rate": 2.2294e-06, "loss": 0.0457, "step": 38854 }, { "epoch": 2.0714933091645786, "grad_norm": 5.641685485839844, "learning_rate": 2.2292e-06, "loss": 0.0138, "step": 38855 }, { "epoch": 2.0715466225942314, "grad_norm": 2.9199490547180176, "learning_rate": 2.229e-06, "loss": 0.0377, "step": 38856 }, { "epoch": 2.0715999360238846, "grad_norm": 5.669031143188477, "learning_rate": 2.2288e-06, "loss": 0.0528, "step": 38857 }, { "epoch": 2.0716532494535373, "grad_norm": 2.4405977725982666, "learning_rate": 2.2286e-06, "loss": 0.0214, "step": 38858 }, { "epoch": 2.07170656288319, "grad_norm": 1.9198859930038452, "learning_rate": 2.2284e-06, "loss": 0.0148, "step": 38859 }, { "epoch": 2.0717598763128433, "grad_norm": 1.9365416765213013, "learning_rate": 2.2282e-06, "loss": 0.0192, "step": 38860 }, { "epoch": 2.071813189742496, "grad_norm": 1.6508146524429321, "learning_rate": 2.228e-06, "loss": 0.0195, "step": 38861 }, { "epoch": 2.0718665031721493, "grad_norm": 2.2201218605041504, "learning_rate": 2.2278e-06, "loss": 0.0192, "step": 38862 }, { "epoch": 2.071919816601802, "grad_norm": 5.13815450668335, "learning_rate": 2.2276000000000005e-06, "loss": 0.014, "step": 38863 }, { "epoch": 2.071973130031455, "grad_norm": 2.5673422813415527, "learning_rate": 2.2274e-06, "loss": 0.0318, "step": 38864 }, { "epoch": 2.072026443461108, "grad_norm": 2.8973193168640137, "learning_rate": 2.2272e-06, "loss": 0.0191, "step": 38865 }, { "epoch": 2.0720797568907607, "grad_norm": 2.1136066913604736, "learning_rate": 2.2270000000000004e-06, "loss": 0.0123, "step": 38866 }, { "epoch": 2.0721330703204135, "grad_norm": 1.5301697254180908, "learning_rate": 2.2268e-06, "loss": 0.0102, "step": 38867 }, { "epoch": 2.0721863837500667, "grad_norm": 0.7887012958526611, "learning_rate": 2.2266e-06, "loss": 0.0066, "step": 38868 }, { "epoch": 2.0722396971797195, "grad_norm": 1.7859351634979248, "learning_rate": 2.2264000000000004e-06, "loss": 0.0122, "step": 38869 }, { "epoch": 2.0722930106093727, "grad_norm": 2.8669846057891846, "learning_rate": 2.2262000000000004e-06, "loss": 0.0255, "step": 38870 }, { "epoch": 2.0723463240390254, "grad_norm": 2.244755268096924, "learning_rate": 2.226e-06, "loss": 0.0094, "step": 38871 }, { "epoch": 2.072399637468678, "grad_norm": 2.6199092864990234, "learning_rate": 2.2258000000000004e-06, "loss": 0.0453, "step": 38872 }, { "epoch": 2.0724529508983314, "grad_norm": 2.8753974437713623, "learning_rate": 2.2256000000000003e-06, "loss": 0.0202, "step": 38873 }, { "epoch": 2.072506264327984, "grad_norm": 3.3162801265716553, "learning_rate": 2.2254e-06, "loss": 0.039, "step": 38874 }, { "epoch": 2.0725595777576373, "grad_norm": 5.687763690948486, "learning_rate": 2.2252000000000003e-06, "loss": 0.0452, "step": 38875 }, { "epoch": 2.07261289118729, "grad_norm": 1.158729910850525, "learning_rate": 2.2250000000000003e-06, "loss": 0.0052, "step": 38876 }, { "epoch": 2.072666204616943, "grad_norm": 0.9177940487861633, "learning_rate": 2.2248000000000003e-06, "loss": 0.0128, "step": 38877 }, { "epoch": 2.072719518046596, "grad_norm": 2.11665678024292, "learning_rate": 2.2246000000000003e-06, "loss": 0.0307, "step": 38878 }, { "epoch": 2.072772831476249, "grad_norm": 2.0438504219055176, "learning_rate": 2.2244000000000003e-06, "loss": 0.0194, "step": 38879 }, { "epoch": 2.072826144905902, "grad_norm": 0.8054399490356445, "learning_rate": 2.2242000000000002e-06, "loss": 0.0042, "step": 38880 }, { "epoch": 2.0728794583355548, "grad_norm": 3.974571466445923, "learning_rate": 2.2240000000000002e-06, "loss": 0.0272, "step": 38881 }, { "epoch": 2.0729327717652075, "grad_norm": 2.5661590099334717, "learning_rate": 2.2238e-06, "loss": 0.0195, "step": 38882 }, { "epoch": 2.0729860851948607, "grad_norm": 0.858903169631958, "learning_rate": 2.2236e-06, "loss": 0.0059, "step": 38883 }, { "epoch": 2.0730393986245135, "grad_norm": 1.5838712453842163, "learning_rate": 2.2234e-06, "loss": 0.0272, "step": 38884 }, { "epoch": 2.0730927120541662, "grad_norm": 1.4621086120605469, "learning_rate": 2.2232e-06, "loss": 0.0195, "step": 38885 }, { "epoch": 2.0731460254838194, "grad_norm": 2.0294463634490967, "learning_rate": 2.223e-06, "loss": 0.0081, "step": 38886 }, { "epoch": 2.073199338913472, "grad_norm": 1.2941182851791382, "learning_rate": 2.2228e-06, "loss": 0.005, "step": 38887 }, { "epoch": 2.0732526523431254, "grad_norm": 9.537055015563965, "learning_rate": 2.2226e-06, "loss": 0.0403, "step": 38888 }, { "epoch": 2.073305965772778, "grad_norm": 1.5599311590194702, "learning_rate": 2.2224e-06, "loss": 0.0286, "step": 38889 }, { "epoch": 2.073359279202431, "grad_norm": 2.5124874114990234, "learning_rate": 2.2222e-06, "loss": 0.0229, "step": 38890 }, { "epoch": 2.073412592632084, "grad_norm": 3.3376212120056152, "learning_rate": 2.222e-06, "loss": 0.0174, "step": 38891 }, { "epoch": 2.073465906061737, "grad_norm": 1.1293641328811646, "learning_rate": 2.2218e-06, "loss": 0.0102, "step": 38892 }, { "epoch": 2.07351921949139, "grad_norm": 3.7408130168914795, "learning_rate": 2.2216e-06, "loss": 0.0367, "step": 38893 }, { "epoch": 2.073572532921043, "grad_norm": 1.7038699388504028, "learning_rate": 2.2214e-06, "loss": 0.0207, "step": 38894 }, { "epoch": 2.0736258463506956, "grad_norm": 0.8355650901794434, "learning_rate": 2.2212000000000004e-06, "loss": 0.0218, "step": 38895 }, { "epoch": 2.073679159780349, "grad_norm": 1.831488013267517, "learning_rate": 2.221e-06, "loss": 0.0293, "step": 38896 }, { "epoch": 2.0737324732100015, "grad_norm": 1.0928611755371094, "learning_rate": 2.2208e-06, "loss": 0.0118, "step": 38897 }, { "epoch": 2.0737857866396547, "grad_norm": 2.8174490928649902, "learning_rate": 2.2206000000000004e-06, "loss": 0.0233, "step": 38898 }, { "epoch": 2.0738391000693075, "grad_norm": 0.4724922478199005, "learning_rate": 2.2204e-06, "loss": 0.0024, "step": 38899 }, { "epoch": 2.0738924134989603, "grad_norm": 1.6203007698059082, "learning_rate": 2.2202e-06, "loss": 0.0216, "step": 38900 }, { "epoch": 2.0739457269286135, "grad_norm": 2.0102500915527344, "learning_rate": 2.2200000000000003e-06, "loss": 0.0136, "step": 38901 }, { "epoch": 2.073999040358266, "grad_norm": 3.290252447128296, "learning_rate": 2.2198000000000003e-06, "loss": 0.027, "step": 38902 }, { "epoch": 2.074052353787919, "grad_norm": 1.54866361618042, "learning_rate": 2.2196e-06, "loss": 0.0108, "step": 38903 }, { "epoch": 2.074105667217572, "grad_norm": 1.775006651878357, "learning_rate": 2.2194000000000003e-06, "loss": 0.0143, "step": 38904 }, { "epoch": 2.074158980647225, "grad_norm": 3.2725698947906494, "learning_rate": 2.2192000000000003e-06, "loss": 0.0515, "step": 38905 }, { "epoch": 2.074212294076878, "grad_norm": 3.632943630218506, "learning_rate": 2.219e-06, "loss": 0.033, "step": 38906 }, { "epoch": 2.074265607506531, "grad_norm": 1.6566835641860962, "learning_rate": 2.2188000000000003e-06, "loss": 0.0122, "step": 38907 }, { "epoch": 2.0743189209361836, "grad_norm": 3.973841428756714, "learning_rate": 2.2186000000000002e-06, "loss": 0.0359, "step": 38908 }, { "epoch": 2.074372234365837, "grad_norm": 2.16861891746521, "learning_rate": 2.2184000000000002e-06, "loss": 0.0217, "step": 38909 }, { "epoch": 2.0744255477954896, "grad_norm": 0.8178170323371887, "learning_rate": 2.2182e-06, "loss": 0.0071, "step": 38910 }, { "epoch": 2.074478861225143, "grad_norm": 3.425414562225342, "learning_rate": 2.218e-06, "loss": 0.0168, "step": 38911 }, { "epoch": 2.0745321746547956, "grad_norm": 1.0748345851898193, "learning_rate": 2.2178e-06, "loss": 0.0194, "step": 38912 }, { "epoch": 2.0745854880844483, "grad_norm": 1.4247790575027466, "learning_rate": 2.2176e-06, "loss": 0.0056, "step": 38913 }, { "epoch": 2.0746388015141015, "grad_norm": 2.1527154445648193, "learning_rate": 2.2174e-06, "loss": 0.0095, "step": 38914 }, { "epoch": 2.0746921149437543, "grad_norm": 0.690284013748169, "learning_rate": 2.2172e-06, "loss": 0.0051, "step": 38915 }, { "epoch": 2.0747454283734075, "grad_norm": 2.178685426712036, "learning_rate": 2.217e-06, "loss": 0.0307, "step": 38916 }, { "epoch": 2.0747987418030602, "grad_norm": 3.6143298149108887, "learning_rate": 2.2168e-06, "loss": 0.0195, "step": 38917 }, { "epoch": 2.074852055232713, "grad_norm": 2.8281078338623047, "learning_rate": 2.2166e-06, "loss": 0.0292, "step": 38918 }, { "epoch": 2.074905368662366, "grad_norm": 1.535689115524292, "learning_rate": 2.2164e-06, "loss": 0.0174, "step": 38919 }, { "epoch": 2.074958682092019, "grad_norm": 0.6629682779312134, "learning_rate": 2.2162000000000005e-06, "loss": 0.0038, "step": 38920 }, { "epoch": 2.0750119955216717, "grad_norm": 2.425147294998169, "learning_rate": 2.216e-06, "loss": 0.0194, "step": 38921 }, { "epoch": 2.075065308951325, "grad_norm": 0.4625893235206604, "learning_rate": 2.2158e-06, "loss": 0.0024, "step": 38922 }, { "epoch": 2.0751186223809777, "grad_norm": 0.6842975616455078, "learning_rate": 2.2156000000000004e-06, "loss": 0.0118, "step": 38923 }, { "epoch": 2.075171935810631, "grad_norm": 2.1923727989196777, "learning_rate": 2.2154e-06, "loss": 0.0195, "step": 38924 }, { "epoch": 2.0752252492402836, "grad_norm": 1.8939355611801147, "learning_rate": 2.2152e-06, "loss": 0.0153, "step": 38925 }, { "epoch": 2.0752785626699364, "grad_norm": 2.6064703464508057, "learning_rate": 2.2150000000000004e-06, "loss": 0.0535, "step": 38926 }, { "epoch": 2.0753318760995896, "grad_norm": 1.4163037538528442, "learning_rate": 2.2148000000000004e-06, "loss": 0.0066, "step": 38927 }, { "epoch": 2.0753851895292423, "grad_norm": 2.574310541152954, "learning_rate": 2.2146e-06, "loss": 0.0131, "step": 38928 }, { "epoch": 2.0754385029588955, "grad_norm": 3.038853406906128, "learning_rate": 2.2144000000000003e-06, "loss": 0.0416, "step": 38929 }, { "epoch": 2.0754918163885483, "grad_norm": 4.834634780883789, "learning_rate": 2.2142000000000003e-06, "loss": 0.0362, "step": 38930 }, { "epoch": 2.075545129818201, "grad_norm": 2.8163869380950928, "learning_rate": 2.2140000000000003e-06, "loss": 0.0291, "step": 38931 }, { "epoch": 2.0755984432478543, "grad_norm": 1.6659009456634521, "learning_rate": 2.2138000000000003e-06, "loss": 0.0143, "step": 38932 }, { "epoch": 2.075651756677507, "grad_norm": 1.4836357831954956, "learning_rate": 2.2136000000000003e-06, "loss": 0.0161, "step": 38933 }, { "epoch": 2.07570507010716, "grad_norm": 2.2970194816589355, "learning_rate": 2.2134000000000003e-06, "loss": 0.034, "step": 38934 }, { "epoch": 2.075758383536813, "grad_norm": 1.6377817392349243, "learning_rate": 2.2132000000000003e-06, "loss": 0.0104, "step": 38935 }, { "epoch": 2.0758116969664657, "grad_norm": 2.3223958015441895, "learning_rate": 2.2130000000000002e-06, "loss": 0.058, "step": 38936 }, { "epoch": 2.075865010396119, "grad_norm": 2.1017916202545166, "learning_rate": 2.2128000000000002e-06, "loss": 0.0146, "step": 38937 }, { "epoch": 2.0759183238257717, "grad_norm": 1.725465178489685, "learning_rate": 2.2126000000000002e-06, "loss": 0.0127, "step": 38938 }, { "epoch": 2.0759716372554244, "grad_norm": 1.0635223388671875, "learning_rate": 2.2124e-06, "loss": 0.0267, "step": 38939 }, { "epoch": 2.0760249506850776, "grad_norm": 3.602086305618286, "learning_rate": 2.2122e-06, "loss": 0.0166, "step": 38940 }, { "epoch": 2.0760782641147304, "grad_norm": 1.8070728778839111, "learning_rate": 2.212e-06, "loss": 0.009, "step": 38941 }, { "epoch": 2.0761315775443836, "grad_norm": 0.8429811000823975, "learning_rate": 2.2118e-06, "loss": 0.007, "step": 38942 }, { "epoch": 2.0761848909740364, "grad_norm": 3.06537127494812, "learning_rate": 2.2116e-06, "loss": 0.0206, "step": 38943 }, { "epoch": 2.076238204403689, "grad_norm": 1.4947651624679565, "learning_rate": 2.2114e-06, "loss": 0.0126, "step": 38944 }, { "epoch": 2.0762915178333423, "grad_norm": 3.7738583087921143, "learning_rate": 2.2112e-06, "loss": 0.0277, "step": 38945 }, { "epoch": 2.076344831262995, "grad_norm": 2.50260329246521, "learning_rate": 2.211e-06, "loss": 0.0105, "step": 38946 }, { "epoch": 2.0763981446926483, "grad_norm": 7.790611267089844, "learning_rate": 2.2108e-06, "loss": 0.0308, "step": 38947 }, { "epoch": 2.076451458122301, "grad_norm": 1.6298514604568481, "learning_rate": 2.2106e-06, "loss": 0.0096, "step": 38948 }, { "epoch": 2.076504771551954, "grad_norm": 3.880521535873413, "learning_rate": 2.2104e-06, "loss": 0.0405, "step": 38949 }, { "epoch": 2.076558084981607, "grad_norm": 1.8057340383529663, "learning_rate": 2.2102e-06, "loss": 0.0132, "step": 38950 }, { "epoch": 2.0766113984112597, "grad_norm": 1.7142783403396606, "learning_rate": 2.21e-06, "loss": 0.0245, "step": 38951 }, { "epoch": 2.0766647118409125, "grad_norm": 1.4320850372314453, "learning_rate": 2.2098000000000004e-06, "loss": 0.0096, "step": 38952 }, { "epoch": 2.0767180252705657, "grad_norm": 1.2446564435958862, "learning_rate": 2.2096e-06, "loss": 0.0065, "step": 38953 }, { "epoch": 2.0767713387002185, "grad_norm": 1.4062517881393433, "learning_rate": 2.2094e-06, "loss": 0.0308, "step": 38954 }, { "epoch": 2.0768246521298717, "grad_norm": 1.8951002359390259, "learning_rate": 2.2092000000000004e-06, "loss": 0.0258, "step": 38955 }, { "epoch": 2.0768779655595244, "grad_norm": 1.9049855470657349, "learning_rate": 2.2090000000000004e-06, "loss": 0.0366, "step": 38956 }, { "epoch": 2.076931278989177, "grad_norm": 2.802492380142212, "learning_rate": 2.2088e-06, "loss": 0.0311, "step": 38957 }, { "epoch": 2.0769845924188304, "grad_norm": 2.3005831241607666, "learning_rate": 2.2086000000000003e-06, "loss": 0.0218, "step": 38958 }, { "epoch": 2.077037905848483, "grad_norm": 1.4149967432022095, "learning_rate": 2.2084000000000003e-06, "loss": 0.0095, "step": 38959 }, { "epoch": 2.0770912192781363, "grad_norm": 2.2513551712036133, "learning_rate": 2.2082e-06, "loss": 0.0121, "step": 38960 }, { "epoch": 2.077144532707789, "grad_norm": 0.5768921971321106, "learning_rate": 2.2080000000000003e-06, "loss": 0.0043, "step": 38961 }, { "epoch": 2.077197846137442, "grad_norm": 1.9238663911819458, "learning_rate": 2.2078000000000003e-06, "loss": 0.0098, "step": 38962 }, { "epoch": 2.077251159567095, "grad_norm": 1.804828405380249, "learning_rate": 2.2076000000000003e-06, "loss": 0.0241, "step": 38963 }, { "epoch": 2.077304472996748, "grad_norm": 1.5708214044570923, "learning_rate": 2.2074000000000002e-06, "loss": 0.0154, "step": 38964 }, { "epoch": 2.077357786426401, "grad_norm": 1.2912629842758179, "learning_rate": 2.2072000000000002e-06, "loss": 0.0172, "step": 38965 }, { "epoch": 2.0774110998560538, "grad_norm": 0.7099784016609192, "learning_rate": 2.2070000000000002e-06, "loss": 0.0055, "step": 38966 }, { "epoch": 2.0774644132857065, "grad_norm": 0.407972127199173, "learning_rate": 2.2068e-06, "loss": 0.0036, "step": 38967 }, { "epoch": 2.0775177267153597, "grad_norm": 0.7602831721305847, "learning_rate": 2.2066e-06, "loss": 0.0047, "step": 38968 }, { "epoch": 2.0775710401450125, "grad_norm": 3.2867941856384277, "learning_rate": 2.2064e-06, "loss": 0.0193, "step": 38969 }, { "epoch": 2.0776243535746657, "grad_norm": 2.494168996810913, "learning_rate": 2.2062e-06, "loss": 0.0232, "step": 38970 }, { "epoch": 2.0776776670043184, "grad_norm": 2.665914535522461, "learning_rate": 2.206e-06, "loss": 0.0366, "step": 38971 }, { "epoch": 2.077730980433971, "grad_norm": 0.5551722049713135, "learning_rate": 2.2058e-06, "loss": 0.0032, "step": 38972 }, { "epoch": 2.0777842938636244, "grad_norm": 1.308223843574524, "learning_rate": 2.2056e-06, "loss": 0.0277, "step": 38973 }, { "epoch": 2.077837607293277, "grad_norm": 2.8762271404266357, "learning_rate": 2.2054000000000005e-06, "loss": 0.025, "step": 38974 }, { "epoch": 2.07789092072293, "grad_norm": 1.8573215007781982, "learning_rate": 2.2052e-06, "loss": 0.0264, "step": 38975 }, { "epoch": 2.077944234152583, "grad_norm": 1.4740400314331055, "learning_rate": 2.205e-06, "loss": 0.0179, "step": 38976 }, { "epoch": 2.077997547582236, "grad_norm": 1.6986411809921265, "learning_rate": 2.2048000000000005e-06, "loss": 0.0121, "step": 38977 }, { "epoch": 2.078050861011889, "grad_norm": 1.6880857944488525, "learning_rate": 2.2046e-06, "loss": 0.0102, "step": 38978 }, { "epoch": 2.078104174441542, "grad_norm": 1.27792227268219, "learning_rate": 2.2044e-06, "loss": 0.0062, "step": 38979 }, { "epoch": 2.0781574878711946, "grad_norm": 1.466854453086853, "learning_rate": 2.2042000000000004e-06, "loss": 0.0204, "step": 38980 }, { "epoch": 2.078210801300848, "grad_norm": 1.3266565799713135, "learning_rate": 2.2040000000000004e-06, "loss": 0.0106, "step": 38981 }, { "epoch": 2.0782641147305005, "grad_norm": 2.6428635120391846, "learning_rate": 2.2038e-06, "loss": 0.0282, "step": 38982 }, { "epoch": 2.0783174281601537, "grad_norm": 2.6094610691070557, "learning_rate": 2.2036000000000004e-06, "loss": 0.0112, "step": 38983 }, { "epoch": 2.0783707415898065, "grad_norm": 3.248086929321289, "learning_rate": 2.2034000000000004e-06, "loss": 0.0107, "step": 38984 }, { "epoch": 2.0784240550194593, "grad_norm": 1.1076316833496094, "learning_rate": 2.2032e-06, "loss": 0.0192, "step": 38985 }, { "epoch": 2.0784773684491125, "grad_norm": 1.7505090236663818, "learning_rate": 2.2030000000000003e-06, "loss": 0.0128, "step": 38986 }, { "epoch": 2.078530681878765, "grad_norm": 1.7043718099594116, "learning_rate": 2.2028000000000003e-06, "loss": 0.0317, "step": 38987 }, { "epoch": 2.078583995308418, "grad_norm": 2.255617618560791, "learning_rate": 2.2026000000000003e-06, "loss": 0.0151, "step": 38988 }, { "epoch": 2.078637308738071, "grad_norm": 5.737717151641846, "learning_rate": 2.2024000000000003e-06, "loss": 0.019, "step": 38989 }, { "epoch": 2.078690622167724, "grad_norm": 1.0552966594696045, "learning_rate": 2.2022000000000003e-06, "loss": 0.0137, "step": 38990 }, { "epoch": 2.078743935597377, "grad_norm": 1.5615519285202026, "learning_rate": 2.2020000000000003e-06, "loss": 0.0291, "step": 38991 }, { "epoch": 2.07879724902703, "grad_norm": 0.270175963640213, "learning_rate": 2.2018000000000002e-06, "loss": 0.0025, "step": 38992 }, { "epoch": 2.0788505624566826, "grad_norm": 1.1897454261779785, "learning_rate": 2.2016000000000002e-06, "loss": 0.0069, "step": 38993 }, { "epoch": 2.078903875886336, "grad_norm": 1.4521111249923706, "learning_rate": 2.2014000000000002e-06, "loss": 0.0207, "step": 38994 }, { "epoch": 2.0789571893159886, "grad_norm": 0.5532042384147644, "learning_rate": 2.2012e-06, "loss": 0.0046, "step": 38995 }, { "epoch": 2.079010502745642, "grad_norm": 1.665635585784912, "learning_rate": 2.201e-06, "loss": 0.0208, "step": 38996 }, { "epoch": 2.0790638161752946, "grad_norm": 1.0790237188339233, "learning_rate": 2.2008e-06, "loss": 0.0132, "step": 38997 }, { "epoch": 2.0791171296049473, "grad_norm": 1.7817410230636597, "learning_rate": 2.2006e-06, "loss": 0.0098, "step": 38998 }, { "epoch": 2.0791704430346005, "grad_norm": 1.5909326076507568, "learning_rate": 2.2004e-06, "loss": 0.0182, "step": 38999 }, { "epoch": 2.0792237564642533, "grad_norm": 1.643315076828003, "learning_rate": 2.2002e-06, "loss": 0.0271, "step": 39000 }, { "epoch": 2.0792770698939065, "grad_norm": 2.644810199737549, "learning_rate": 2.2e-06, "loss": 0.0268, "step": 39001 }, { "epoch": 2.0793303833235592, "grad_norm": 0.8213669061660767, "learning_rate": 2.1998e-06, "loss": 0.0053, "step": 39002 }, { "epoch": 2.079383696753212, "grad_norm": 1.1846944093704224, "learning_rate": 2.1996e-06, "loss": 0.013, "step": 39003 }, { "epoch": 2.079437010182865, "grad_norm": 2.0658533573150635, "learning_rate": 2.1994e-06, "loss": 0.0219, "step": 39004 }, { "epoch": 2.079490323612518, "grad_norm": 0.26915299892425537, "learning_rate": 2.1992e-06, "loss": 0.0023, "step": 39005 }, { "epoch": 2.0795436370421707, "grad_norm": 1.0460914373397827, "learning_rate": 2.1990000000000005e-06, "loss": 0.019, "step": 39006 }, { "epoch": 2.079596950471824, "grad_norm": 1.8226432800292969, "learning_rate": 2.1988e-06, "loss": 0.0147, "step": 39007 }, { "epoch": 2.0796502639014767, "grad_norm": 3.4763901233673096, "learning_rate": 2.1986e-06, "loss": 0.0295, "step": 39008 }, { "epoch": 2.07970357733113, "grad_norm": 1.7730063199996948, "learning_rate": 2.1984000000000004e-06, "loss": 0.0094, "step": 39009 }, { "epoch": 2.0797568907607826, "grad_norm": 1.355242371559143, "learning_rate": 2.1982e-06, "loss": 0.0086, "step": 39010 }, { "epoch": 2.0798102041904354, "grad_norm": 0.6925601959228516, "learning_rate": 2.198e-06, "loss": 0.0039, "step": 39011 }, { "epoch": 2.0798635176200886, "grad_norm": 0.9306300282478333, "learning_rate": 2.1978000000000004e-06, "loss": 0.0046, "step": 39012 }, { "epoch": 2.0799168310497413, "grad_norm": 2.1278347969055176, "learning_rate": 2.1976000000000004e-06, "loss": 0.0075, "step": 39013 }, { "epoch": 2.0799701444793945, "grad_norm": 1.2788310050964355, "learning_rate": 2.1974e-06, "loss": 0.0189, "step": 39014 }, { "epoch": 2.0800234579090473, "grad_norm": 0.8747391700744629, "learning_rate": 2.1972000000000003e-06, "loss": 0.0176, "step": 39015 }, { "epoch": 2.0800767713387, "grad_norm": 3.4309744834899902, "learning_rate": 2.1970000000000003e-06, "loss": 0.0221, "step": 39016 }, { "epoch": 2.0801300847683533, "grad_norm": 1.64047110080719, "learning_rate": 2.1968e-06, "loss": 0.0087, "step": 39017 }, { "epoch": 2.080183398198006, "grad_norm": 2.7662763595581055, "learning_rate": 2.1966000000000003e-06, "loss": 0.0428, "step": 39018 }, { "epoch": 2.080236711627659, "grad_norm": 2.425097942352295, "learning_rate": 2.1964000000000003e-06, "loss": 0.0416, "step": 39019 }, { "epoch": 2.080290025057312, "grad_norm": 2.1067116260528564, "learning_rate": 2.1962000000000003e-06, "loss": 0.0214, "step": 39020 }, { "epoch": 2.0803433384869647, "grad_norm": 0.6895468235015869, "learning_rate": 2.1960000000000002e-06, "loss": 0.0043, "step": 39021 }, { "epoch": 2.080396651916618, "grad_norm": 2.3198626041412354, "learning_rate": 2.1958000000000002e-06, "loss": 0.0278, "step": 39022 }, { "epoch": 2.0804499653462707, "grad_norm": 2.368915557861328, "learning_rate": 2.1956e-06, "loss": 0.0293, "step": 39023 }, { "epoch": 2.0805032787759234, "grad_norm": 1.8207037448883057, "learning_rate": 2.1954e-06, "loss": 0.031, "step": 39024 }, { "epoch": 2.0805565922055766, "grad_norm": 1.7926592826843262, "learning_rate": 2.1952e-06, "loss": 0.011, "step": 39025 }, { "epoch": 2.0806099056352294, "grad_norm": 1.92436683177948, "learning_rate": 2.195e-06, "loss": 0.0254, "step": 39026 }, { "epoch": 2.0806632190648826, "grad_norm": 1.7536497116088867, "learning_rate": 2.1948e-06, "loss": 0.0095, "step": 39027 }, { "epoch": 2.0807165324945354, "grad_norm": 1.5489262342453003, "learning_rate": 2.1946e-06, "loss": 0.0226, "step": 39028 }, { "epoch": 2.080769845924188, "grad_norm": 1.858919382095337, "learning_rate": 2.1944e-06, "loss": 0.0114, "step": 39029 }, { "epoch": 2.0808231593538413, "grad_norm": 2.3569834232330322, "learning_rate": 2.1942e-06, "loss": 0.0142, "step": 39030 }, { "epoch": 2.080876472783494, "grad_norm": 1.6125026941299438, "learning_rate": 2.194e-06, "loss": 0.0228, "step": 39031 }, { "epoch": 2.0809297862131473, "grad_norm": 1.7545381784439087, "learning_rate": 2.1938e-06, "loss": 0.0272, "step": 39032 }, { "epoch": 2.0809830996428, "grad_norm": 1.786052942276001, "learning_rate": 2.1936e-06, "loss": 0.0109, "step": 39033 }, { "epoch": 2.081036413072453, "grad_norm": 0.9976466298103333, "learning_rate": 2.1934e-06, "loss": 0.0065, "step": 39034 }, { "epoch": 2.081089726502106, "grad_norm": 1.286190152168274, "learning_rate": 2.1932e-06, "loss": 0.0126, "step": 39035 }, { "epoch": 2.0811430399317588, "grad_norm": 1.526261568069458, "learning_rate": 2.193e-06, "loss": 0.0068, "step": 39036 }, { "epoch": 2.0811963533614115, "grad_norm": 0.9567044973373413, "learning_rate": 2.1928e-06, "loss": 0.0178, "step": 39037 }, { "epoch": 2.0812496667910647, "grad_norm": 0.7329753041267395, "learning_rate": 2.1926000000000004e-06, "loss": 0.0057, "step": 39038 }, { "epoch": 2.0813029802207175, "grad_norm": 1.270891547203064, "learning_rate": 2.1924e-06, "loss": 0.0096, "step": 39039 }, { "epoch": 2.0813562936503707, "grad_norm": 2.0389139652252197, "learning_rate": 2.1922e-06, "loss": 0.0084, "step": 39040 }, { "epoch": 2.0814096070800234, "grad_norm": 2.0477631092071533, "learning_rate": 2.1920000000000004e-06, "loss": 0.019, "step": 39041 }, { "epoch": 2.081462920509676, "grad_norm": 0.8978902101516724, "learning_rate": 2.1918000000000003e-06, "loss": 0.0065, "step": 39042 }, { "epoch": 2.0815162339393294, "grad_norm": 1.2466753721237183, "learning_rate": 2.1916e-06, "loss": 0.008, "step": 39043 }, { "epoch": 2.081569547368982, "grad_norm": 0.4563218951225281, "learning_rate": 2.1914000000000003e-06, "loss": 0.0027, "step": 39044 }, { "epoch": 2.0816228607986353, "grad_norm": 3.3400824069976807, "learning_rate": 2.1912000000000003e-06, "loss": 0.0107, "step": 39045 }, { "epoch": 2.081676174228288, "grad_norm": 2.012322187423706, "learning_rate": 2.191e-06, "loss": 0.0392, "step": 39046 }, { "epoch": 2.081729487657941, "grad_norm": 3.1707656383514404, "learning_rate": 2.1908000000000003e-06, "loss": 0.0663, "step": 39047 }, { "epoch": 2.081782801087594, "grad_norm": 1.6060277223587036, "learning_rate": 2.1906000000000003e-06, "loss": 0.0087, "step": 39048 }, { "epoch": 2.081836114517247, "grad_norm": 1.6437091827392578, "learning_rate": 2.1904000000000002e-06, "loss": 0.0135, "step": 39049 }, { "epoch": 2.0818894279469, "grad_norm": 1.262094259262085, "learning_rate": 2.1902000000000002e-06, "loss": 0.0087, "step": 39050 }, { "epoch": 2.0819427413765528, "grad_norm": 2.3982982635498047, "learning_rate": 2.19e-06, "loss": 0.04, "step": 39051 }, { "epoch": 2.0819960548062055, "grad_norm": 0.43114331364631653, "learning_rate": 2.1898e-06, "loss": 0.0031, "step": 39052 }, { "epoch": 2.0820493682358587, "grad_norm": 2.6317148208618164, "learning_rate": 2.1896e-06, "loss": 0.0107, "step": 39053 }, { "epoch": 2.0821026816655115, "grad_norm": 2.1137266159057617, "learning_rate": 2.1894e-06, "loss": 0.0271, "step": 39054 }, { "epoch": 2.0821559950951647, "grad_norm": 1.3587435483932495, "learning_rate": 2.1892e-06, "loss": 0.0111, "step": 39055 }, { "epoch": 2.0822093085248174, "grad_norm": 0.5958709716796875, "learning_rate": 2.189e-06, "loss": 0.0063, "step": 39056 }, { "epoch": 2.08226262195447, "grad_norm": 1.1642918586730957, "learning_rate": 2.1888e-06, "loss": 0.0052, "step": 39057 }, { "epoch": 2.0823159353841234, "grad_norm": 1.0635203123092651, "learning_rate": 2.1886e-06, "loss": 0.0098, "step": 39058 }, { "epoch": 2.082369248813776, "grad_norm": 1.4005719423294067, "learning_rate": 2.1884e-06, "loss": 0.0066, "step": 39059 }, { "epoch": 2.082422562243429, "grad_norm": 2.5444138050079346, "learning_rate": 2.1882e-06, "loss": 0.0254, "step": 39060 }, { "epoch": 2.082475875673082, "grad_norm": 0.9678284525871277, "learning_rate": 2.188e-06, "loss": 0.0239, "step": 39061 }, { "epoch": 2.082529189102735, "grad_norm": 2.311984062194824, "learning_rate": 2.1878e-06, "loss": 0.0339, "step": 39062 }, { "epoch": 2.082582502532388, "grad_norm": 2.5128257274627686, "learning_rate": 2.1876000000000005e-06, "loss": 0.0278, "step": 39063 }, { "epoch": 2.082635815962041, "grad_norm": 1.3499869108200073, "learning_rate": 2.1874e-06, "loss": 0.0162, "step": 39064 }, { "epoch": 2.0826891293916936, "grad_norm": 2.6486706733703613, "learning_rate": 2.1872e-06, "loss": 0.0488, "step": 39065 }, { "epoch": 2.082742442821347, "grad_norm": 2.6871497631073, "learning_rate": 2.1870000000000004e-06, "loss": 0.0198, "step": 39066 }, { "epoch": 2.0827957562509996, "grad_norm": 1.7645689249038696, "learning_rate": 2.1868000000000004e-06, "loss": 0.0094, "step": 39067 }, { "epoch": 2.0828490696806528, "grad_norm": 1.7609996795654297, "learning_rate": 2.1866e-06, "loss": 0.0196, "step": 39068 }, { "epoch": 2.0829023831103055, "grad_norm": 2.1413729190826416, "learning_rate": 2.1864000000000004e-06, "loss": 0.016, "step": 39069 }, { "epoch": 2.0829556965399583, "grad_norm": 2.29948091506958, "learning_rate": 2.1862000000000003e-06, "loss": 0.0098, "step": 39070 }, { "epoch": 2.0830090099696115, "grad_norm": 1.206167459487915, "learning_rate": 2.186e-06, "loss": 0.0064, "step": 39071 }, { "epoch": 2.0830623233992642, "grad_norm": 1.4540499448776245, "learning_rate": 2.1858000000000003e-06, "loss": 0.0087, "step": 39072 }, { "epoch": 2.083115636828917, "grad_norm": 0.6905199289321899, "learning_rate": 2.1856000000000003e-06, "loss": 0.0051, "step": 39073 }, { "epoch": 2.08316895025857, "grad_norm": 1.9528253078460693, "learning_rate": 2.1854000000000003e-06, "loss": 0.0088, "step": 39074 }, { "epoch": 2.083222263688223, "grad_norm": 3.6677069664001465, "learning_rate": 2.1852000000000003e-06, "loss": 0.0146, "step": 39075 }, { "epoch": 2.083275577117876, "grad_norm": 1.8765130043029785, "learning_rate": 2.1850000000000003e-06, "loss": 0.0088, "step": 39076 }, { "epoch": 2.083328890547529, "grad_norm": 3.237093210220337, "learning_rate": 2.1848000000000002e-06, "loss": 0.0182, "step": 39077 }, { "epoch": 2.0833822039771817, "grad_norm": 2.760429859161377, "learning_rate": 2.1846000000000002e-06, "loss": 0.0133, "step": 39078 }, { "epoch": 2.083435517406835, "grad_norm": 0.6697821021080017, "learning_rate": 2.1844e-06, "loss": 0.0044, "step": 39079 }, { "epoch": 2.0834888308364876, "grad_norm": 2.0519795417785645, "learning_rate": 2.1842e-06, "loss": 0.0179, "step": 39080 }, { "epoch": 2.083542144266141, "grad_norm": 1.0984165668487549, "learning_rate": 2.184e-06, "loss": 0.0168, "step": 39081 }, { "epoch": 2.0835954576957936, "grad_norm": 1.9551044702529907, "learning_rate": 2.1838e-06, "loss": 0.0077, "step": 39082 }, { "epoch": 2.0836487711254463, "grad_norm": 1.4503161907196045, "learning_rate": 2.1836e-06, "loss": 0.0212, "step": 39083 }, { "epoch": 2.0837020845550995, "grad_norm": 1.3887813091278076, "learning_rate": 2.1834e-06, "loss": 0.0374, "step": 39084 }, { "epoch": 2.0837553979847523, "grad_norm": 1.1798546314239502, "learning_rate": 2.1832e-06, "loss": 0.0165, "step": 39085 }, { "epoch": 2.0838087114144055, "grad_norm": 3.086531400680542, "learning_rate": 2.183e-06, "loss": 0.0172, "step": 39086 }, { "epoch": 2.0838620248440582, "grad_norm": 0.6746364831924438, "learning_rate": 2.1828e-06, "loss": 0.018, "step": 39087 }, { "epoch": 2.083915338273711, "grad_norm": 2.2709357738494873, "learning_rate": 2.1826e-06, "loss": 0.0202, "step": 39088 }, { "epoch": 2.083968651703364, "grad_norm": 1.4441988468170166, "learning_rate": 2.1824e-06, "loss": 0.0223, "step": 39089 }, { "epoch": 2.084021965133017, "grad_norm": 4.250414848327637, "learning_rate": 2.1822e-06, "loss": 0.0635, "step": 39090 }, { "epoch": 2.08407527856267, "grad_norm": 1.0234243869781494, "learning_rate": 2.182e-06, "loss": 0.0211, "step": 39091 }, { "epoch": 2.084128591992323, "grad_norm": 0.7372839450836182, "learning_rate": 2.1818000000000004e-06, "loss": 0.0042, "step": 39092 }, { "epoch": 2.0841819054219757, "grad_norm": 3.9216911792755127, "learning_rate": 2.1816e-06, "loss": 0.0341, "step": 39093 }, { "epoch": 2.084235218851629, "grad_norm": 2.4081132411956787, "learning_rate": 2.1814e-06, "loss": 0.0162, "step": 39094 }, { "epoch": 2.0842885322812816, "grad_norm": 0.37671494483947754, "learning_rate": 2.1812000000000004e-06, "loss": 0.0037, "step": 39095 }, { "epoch": 2.0843418457109344, "grad_norm": 0.3743284046649933, "learning_rate": 2.181e-06, "loss": 0.0025, "step": 39096 }, { "epoch": 2.0843951591405876, "grad_norm": 2.0087006092071533, "learning_rate": 2.1808e-06, "loss": 0.0107, "step": 39097 }, { "epoch": 2.0844484725702404, "grad_norm": 1.279192328453064, "learning_rate": 2.1806000000000003e-06, "loss": 0.008, "step": 39098 }, { "epoch": 2.0845017859998936, "grad_norm": 1.269357681274414, "learning_rate": 2.1804000000000003e-06, "loss": 0.0107, "step": 39099 }, { "epoch": 2.0845550994295463, "grad_norm": 1.7681653499603271, "learning_rate": 2.1802e-06, "loss": 0.0207, "step": 39100 }, { "epoch": 2.084608412859199, "grad_norm": 2.518588066101074, "learning_rate": 2.1800000000000003e-06, "loss": 0.0488, "step": 39101 }, { "epoch": 2.0846617262888523, "grad_norm": 2.72509765625, "learning_rate": 2.1798000000000003e-06, "loss": 0.0476, "step": 39102 }, { "epoch": 2.084715039718505, "grad_norm": 1.3022135496139526, "learning_rate": 2.1796e-06, "loss": 0.0185, "step": 39103 }, { "epoch": 2.0847683531481582, "grad_norm": 1.8255481719970703, "learning_rate": 2.1794000000000003e-06, "loss": 0.0445, "step": 39104 }, { "epoch": 2.084821666577811, "grad_norm": 1.6139050722122192, "learning_rate": 2.1792000000000002e-06, "loss": 0.0159, "step": 39105 }, { "epoch": 2.0848749800074637, "grad_norm": 0.9126744866371155, "learning_rate": 2.1790000000000002e-06, "loss": 0.0049, "step": 39106 }, { "epoch": 2.084928293437117, "grad_norm": 2.2712552547454834, "learning_rate": 2.1788000000000002e-06, "loss": 0.0117, "step": 39107 }, { "epoch": 2.0849816068667697, "grad_norm": 1.2016888856887817, "learning_rate": 2.1786e-06, "loss": 0.0077, "step": 39108 }, { "epoch": 2.0850349202964225, "grad_norm": 1.9486628770828247, "learning_rate": 2.1784e-06, "loss": 0.0237, "step": 39109 }, { "epoch": 2.0850882337260757, "grad_norm": 1.968415379524231, "learning_rate": 2.1782e-06, "loss": 0.0078, "step": 39110 }, { "epoch": 2.0851415471557284, "grad_norm": 0.6573827862739563, "learning_rate": 2.178e-06, "loss": 0.0056, "step": 39111 }, { "epoch": 2.0851948605853816, "grad_norm": 1.7227731943130493, "learning_rate": 2.1778e-06, "loss": 0.0088, "step": 39112 }, { "epoch": 2.0852481740150344, "grad_norm": 0.7396296262741089, "learning_rate": 2.1776e-06, "loss": 0.0049, "step": 39113 }, { "epoch": 2.085301487444687, "grad_norm": 1.5115859508514404, "learning_rate": 2.1774e-06, "loss": 0.0086, "step": 39114 }, { "epoch": 2.0853548008743403, "grad_norm": 2.481001853942871, "learning_rate": 2.1772e-06, "loss": 0.0222, "step": 39115 }, { "epoch": 2.085408114303993, "grad_norm": 2.4769725799560547, "learning_rate": 2.177e-06, "loss": 0.0392, "step": 39116 }, { "epoch": 2.0854614277336463, "grad_norm": 1.7331193685531616, "learning_rate": 2.1768000000000005e-06, "loss": 0.0104, "step": 39117 }, { "epoch": 2.085514741163299, "grad_norm": 2.118579864501953, "learning_rate": 2.1766e-06, "loss": 0.0297, "step": 39118 }, { "epoch": 2.085568054592952, "grad_norm": 1.1003667116165161, "learning_rate": 2.1764e-06, "loss": 0.0068, "step": 39119 }, { "epoch": 2.085621368022605, "grad_norm": 0.1709260195493698, "learning_rate": 2.1762000000000004e-06, "loss": 0.0019, "step": 39120 }, { "epoch": 2.0856746814522578, "grad_norm": 2.181716203689575, "learning_rate": 2.176e-06, "loss": 0.0234, "step": 39121 }, { "epoch": 2.085727994881911, "grad_norm": 0.9712232351303101, "learning_rate": 2.1758e-06, "loss": 0.0045, "step": 39122 }, { "epoch": 2.0857813083115637, "grad_norm": 4.604849815368652, "learning_rate": 2.1756000000000004e-06, "loss": 0.0237, "step": 39123 }, { "epoch": 2.0858346217412165, "grad_norm": 1.9299908876419067, "learning_rate": 2.1754000000000004e-06, "loss": 0.0269, "step": 39124 }, { "epoch": 2.0858879351708697, "grad_norm": 1.652956485748291, "learning_rate": 2.1752e-06, "loss": 0.0112, "step": 39125 }, { "epoch": 2.0859412486005224, "grad_norm": 1.364985466003418, "learning_rate": 2.1750000000000004e-06, "loss": 0.0097, "step": 39126 }, { "epoch": 2.085994562030175, "grad_norm": 0.5184404850006104, "learning_rate": 2.1748000000000003e-06, "loss": 0.0047, "step": 39127 }, { "epoch": 2.0860478754598284, "grad_norm": 3.2062411308288574, "learning_rate": 2.1746e-06, "loss": 0.0331, "step": 39128 }, { "epoch": 2.086101188889481, "grad_norm": 1.9231221675872803, "learning_rate": 2.1744000000000003e-06, "loss": 0.0213, "step": 39129 }, { "epoch": 2.0861545023191344, "grad_norm": 1.7342824935913086, "learning_rate": 2.1742000000000003e-06, "loss": 0.0108, "step": 39130 }, { "epoch": 2.086207815748787, "grad_norm": 4.125068187713623, "learning_rate": 2.1740000000000003e-06, "loss": 0.0276, "step": 39131 }, { "epoch": 2.08626112917844, "grad_norm": 2.4776923656463623, "learning_rate": 2.1738000000000003e-06, "loss": 0.0138, "step": 39132 }, { "epoch": 2.086314442608093, "grad_norm": 0.9290500283241272, "learning_rate": 2.1736000000000002e-06, "loss": 0.0058, "step": 39133 }, { "epoch": 2.086367756037746, "grad_norm": 3.231071949005127, "learning_rate": 2.1734000000000002e-06, "loss": 0.0391, "step": 39134 }, { "epoch": 2.086421069467399, "grad_norm": 2.2968521118164062, "learning_rate": 2.1732000000000002e-06, "loss": 0.0244, "step": 39135 }, { "epoch": 2.086474382897052, "grad_norm": 0.47278645634651184, "learning_rate": 2.173e-06, "loss": 0.003, "step": 39136 }, { "epoch": 2.0865276963267045, "grad_norm": 2.5015618801116943, "learning_rate": 2.1728e-06, "loss": 0.0254, "step": 39137 }, { "epoch": 2.0865810097563577, "grad_norm": 1.8659664392471313, "learning_rate": 2.1726e-06, "loss": 0.0253, "step": 39138 }, { "epoch": 2.0866343231860105, "grad_norm": 2.062878131866455, "learning_rate": 2.1724e-06, "loss": 0.0306, "step": 39139 }, { "epoch": 2.0866876366156637, "grad_norm": 1.9160254001617432, "learning_rate": 2.1722e-06, "loss": 0.04, "step": 39140 }, { "epoch": 2.0867409500453165, "grad_norm": 4.441188812255859, "learning_rate": 2.172e-06, "loss": 0.0209, "step": 39141 }, { "epoch": 2.086794263474969, "grad_norm": 1.341722011566162, "learning_rate": 2.1718e-06, "loss": 0.0157, "step": 39142 }, { "epoch": 2.0868475769046224, "grad_norm": 3.8904054164886475, "learning_rate": 2.1716e-06, "loss": 0.0383, "step": 39143 }, { "epoch": 2.086900890334275, "grad_norm": 1.7478294372558594, "learning_rate": 2.1714e-06, "loss": 0.0283, "step": 39144 }, { "epoch": 2.086954203763928, "grad_norm": 0.720578134059906, "learning_rate": 2.1712e-06, "loss": 0.0185, "step": 39145 }, { "epoch": 2.087007517193581, "grad_norm": 1.4515810012817383, "learning_rate": 2.171e-06, "loss": 0.0093, "step": 39146 }, { "epoch": 2.087060830623234, "grad_norm": 4.133269309997559, "learning_rate": 2.1708e-06, "loss": 0.012, "step": 39147 }, { "epoch": 2.087114144052887, "grad_norm": 1.3748860359191895, "learning_rate": 2.1706e-06, "loss": 0.0234, "step": 39148 }, { "epoch": 2.08716745748254, "grad_norm": 1.6235955953598022, "learning_rate": 2.1704000000000004e-06, "loss": 0.0259, "step": 39149 }, { "epoch": 2.0872207709121926, "grad_norm": 0.6871331334114075, "learning_rate": 2.1702e-06, "loss": 0.0033, "step": 39150 }, { "epoch": 2.087274084341846, "grad_norm": 3.3520381450653076, "learning_rate": 2.17e-06, "loss": 0.0261, "step": 39151 }, { "epoch": 2.0873273977714986, "grad_norm": 1.3879525661468506, "learning_rate": 2.1698000000000004e-06, "loss": 0.019, "step": 39152 }, { "epoch": 2.0873807112011518, "grad_norm": 0.34016087651252747, "learning_rate": 2.1696e-06, "loss": 0.0034, "step": 39153 }, { "epoch": 2.0874340246308045, "grad_norm": 2.1060996055603027, "learning_rate": 2.1694e-06, "loss": 0.015, "step": 39154 }, { "epoch": 2.0874873380604573, "grad_norm": 1.1113488674163818, "learning_rate": 2.1692000000000003e-06, "loss": 0.0108, "step": 39155 }, { "epoch": 2.0875406514901105, "grad_norm": 2.3595235347747803, "learning_rate": 2.1690000000000003e-06, "loss": 0.0178, "step": 39156 }, { "epoch": 2.0875939649197632, "grad_norm": 1.515660047531128, "learning_rate": 2.1688e-06, "loss": 0.0122, "step": 39157 }, { "epoch": 2.087647278349416, "grad_norm": 1.350821614265442, "learning_rate": 2.1686000000000003e-06, "loss": 0.0201, "step": 39158 }, { "epoch": 2.087700591779069, "grad_norm": 1.250709056854248, "learning_rate": 2.1684000000000003e-06, "loss": 0.0206, "step": 39159 }, { "epoch": 2.087753905208722, "grad_norm": 2.0971322059631348, "learning_rate": 2.1682000000000003e-06, "loss": 0.0089, "step": 39160 }, { "epoch": 2.087807218638375, "grad_norm": 2.4126806259155273, "learning_rate": 2.1680000000000002e-06, "loss": 0.0313, "step": 39161 }, { "epoch": 2.087860532068028, "grad_norm": 1.7415857315063477, "learning_rate": 2.1678000000000002e-06, "loss": 0.0218, "step": 39162 }, { "epoch": 2.0879138454976807, "grad_norm": 2.4268851280212402, "learning_rate": 2.1676000000000002e-06, "loss": 0.0158, "step": 39163 }, { "epoch": 2.087967158927334, "grad_norm": 2.5964441299438477, "learning_rate": 2.1674e-06, "loss": 0.0198, "step": 39164 }, { "epoch": 2.0880204723569866, "grad_norm": 0.2755192816257477, "learning_rate": 2.1672e-06, "loss": 0.0021, "step": 39165 }, { "epoch": 2.08807378578664, "grad_norm": 1.442358374595642, "learning_rate": 2.167e-06, "loss": 0.0077, "step": 39166 }, { "epoch": 2.0881270992162926, "grad_norm": 1.5886071920394897, "learning_rate": 2.1668e-06, "loss": 0.0083, "step": 39167 }, { "epoch": 2.0881804126459453, "grad_norm": 3.478754758834839, "learning_rate": 2.1666e-06, "loss": 0.0381, "step": 39168 }, { "epoch": 2.0882337260755985, "grad_norm": 1.8262625932693481, "learning_rate": 2.1664e-06, "loss": 0.0074, "step": 39169 }, { "epoch": 2.0882870395052513, "grad_norm": 1.8647315502166748, "learning_rate": 2.1662e-06, "loss": 0.0133, "step": 39170 }, { "epoch": 2.0883403529349045, "grad_norm": 2.798953056335449, "learning_rate": 2.166e-06, "loss": 0.0309, "step": 39171 }, { "epoch": 2.0883936663645573, "grad_norm": 3.326289653778076, "learning_rate": 2.1658e-06, "loss": 0.0509, "step": 39172 }, { "epoch": 2.08844697979421, "grad_norm": 1.2820935249328613, "learning_rate": 2.1656e-06, "loss": 0.0189, "step": 39173 }, { "epoch": 2.088500293223863, "grad_norm": 2.758427619934082, "learning_rate": 2.1654000000000005e-06, "loss": 0.016, "step": 39174 }, { "epoch": 2.088553606653516, "grad_norm": 2.4655017852783203, "learning_rate": 2.1652e-06, "loss": 0.0224, "step": 39175 }, { "epoch": 2.088606920083169, "grad_norm": 1.8591536283493042, "learning_rate": 2.165e-06, "loss": 0.0123, "step": 39176 }, { "epoch": 2.088660233512822, "grad_norm": 2.8620831966400146, "learning_rate": 2.1648000000000004e-06, "loss": 0.0281, "step": 39177 }, { "epoch": 2.0887135469424747, "grad_norm": 1.48341965675354, "learning_rate": 2.1646000000000004e-06, "loss": 0.0088, "step": 39178 }, { "epoch": 2.088766860372128, "grad_norm": 0.46925488114356995, "learning_rate": 2.1644e-06, "loss": 0.0036, "step": 39179 }, { "epoch": 2.0888201738017806, "grad_norm": 3.022827386856079, "learning_rate": 2.1642000000000004e-06, "loss": 0.0341, "step": 39180 }, { "epoch": 2.0888734872314334, "grad_norm": 2.9845473766326904, "learning_rate": 2.1640000000000004e-06, "loss": 0.0188, "step": 39181 }, { "epoch": 2.0889268006610866, "grad_norm": 1.7585794925689697, "learning_rate": 2.1638e-06, "loss": 0.0293, "step": 39182 }, { "epoch": 2.0889801140907394, "grad_norm": 2.6627914905548096, "learning_rate": 2.1636000000000003e-06, "loss": 0.0212, "step": 39183 }, { "epoch": 2.0890334275203926, "grad_norm": 1.750381350517273, "learning_rate": 2.1634000000000003e-06, "loss": 0.0143, "step": 39184 }, { "epoch": 2.0890867409500453, "grad_norm": 1.0230231285095215, "learning_rate": 2.1632000000000003e-06, "loss": 0.0051, "step": 39185 }, { "epoch": 2.089140054379698, "grad_norm": 3.54624342918396, "learning_rate": 2.1630000000000003e-06, "loss": 0.0156, "step": 39186 }, { "epoch": 2.0891933678093513, "grad_norm": 2.0747599601745605, "learning_rate": 2.1628000000000003e-06, "loss": 0.012, "step": 39187 }, { "epoch": 2.089246681239004, "grad_norm": 0.7669370770454407, "learning_rate": 2.1626000000000003e-06, "loss": 0.0062, "step": 39188 }, { "epoch": 2.0892999946686572, "grad_norm": 0.48804572224617004, "learning_rate": 2.1624000000000003e-06, "loss": 0.0037, "step": 39189 }, { "epoch": 2.08935330809831, "grad_norm": 2.404310464859009, "learning_rate": 2.1622000000000002e-06, "loss": 0.0327, "step": 39190 }, { "epoch": 2.0894066215279627, "grad_norm": 0.8268876671791077, "learning_rate": 2.1620000000000002e-06, "loss": 0.0049, "step": 39191 }, { "epoch": 2.089459934957616, "grad_norm": 2.257473945617676, "learning_rate": 2.1618e-06, "loss": 0.0199, "step": 39192 }, { "epoch": 2.0895132483872687, "grad_norm": 1.2764945030212402, "learning_rate": 2.1616e-06, "loss": 0.0068, "step": 39193 }, { "epoch": 2.0895665618169215, "grad_norm": 2.0303962230682373, "learning_rate": 2.1614e-06, "loss": 0.0153, "step": 39194 }, { "epoch": 2.0896198752465747, "grad_norm": 0.7660520076751709, "learning_rate": 2.1612e-06, "loss": 0.0024, "step": 39195 }, { "epoch": 2.0896731886762274, "grad_norm": 2.850706100463867, "learning_rate": 2.161e-06, "loss": 0.049, "step": 39196 }, { "epoch": 2.0897265021058806, "grad_norm": 1.7234686613082886, "learning_rate": 2.1608e-06, "loss": 0.0049, "step": 39197 }, { "epoch": 2.0897798155355334, "grad_norm": 1.7683813571929932, "learning_rate": 2.1606e-06, "loss": 0.0098, "step": 39198 }, { "epoch": 2.089833128965186, "grad_norm": 0.625254213809967, "learning_rate": 2.1604e-06, "loss": 0.0037, "step": 39199 }, { "epoch": 2.0898864423948393, "grad_norm": 3.8171303272247314, "learning_rate": 2.1602e-06, "loss": 0.0281, "step": 39200 }, { "epoch": 2.089939755824492, "grad_norm": 0.8525005578994751, "learning_rate": 2.16e-06, "loss": 0.0048, "step": 39201 }, { "epoch": 2.0899930692541453, "grad_norm": 1.104617714881897, "learning_rate": 2.1598e-06, "loss": 0.0025, "step": 39202 }, { "epoch": 2.090046382683798, "grad_norm": 2.523144483566284, "learning_rate": 2.1596000000000005e-06, "loss": 0.023, "step": 39203 }, { "epoch": 2.090099696113451, "grad_norm": 1.4771820306777954, "learning_rate": 2.1594e-06, "loss": 0.0132, "step": 39204 }, { "epoch": 2.090153009543104, "grad_norm": 4.297720909118652, "learning_rate": 2.1592e-06, "loss": 0.022, "step": 39205 }, { "epoch": 2.0902063229727568, "grad_norm": 0.4712233245372772, "learning_rate": 2.1590000000000004e-06, "loss": 0.0034, "step": 39206 }, { "epoch": 2.09025963640241, "grad_norm": 2.1187429428100586, "learning_rate": 2.1588e-06, "loss": 0.0057, "step": 39207 }, { "epoch": 2.0903129498320627, "grad_norm": 1.4156115055084229, "learning_rate": 2.1586e-06, "loss": 0.0155, "step": 39208 }, { "epoch": 2.0903662632617155, "grad_norm": 1.7959949970245361, "learning_rate": 2.1584000000000004e-06, "loss": 0.0081, "step": 39209 }, { "epoch": 2.0904195766913687, "grad_norm": 3.2220256328582764, "learning_rate": 2.1582000000000004e-06, "loss": 0.0424, "step": 39210 }, { "epoch": 2.0904728901210214, "grad_norm": 1.9259892702102661, "learning_rate": 2.158e-06, "loss": 0.0178, "step": 39211 }, { "epoch": 2.090526203550674, "grad_norm": 4.556501388549805, "learning_rate": 2.1578000000000003e-06, "loss": 0.0256, "step": 39212 }, { "epoch": 2.0905795169803274, "grad_norm": 3.602337121963501, "learning_rate": 2.1576000000000003e-06, "loss": 0.0191, "step": 39213 }, { "epoch": 2.09063283040998, "grad_norm": 2.6328182220458984, "learning_rate": 2.1574e-06, "loss": 0.0461, "step": 39214 }, { "epoch": 2.0906861438396334, "grad_norm": 1.3506336212158203, "learning_rate": 2.1572000000000003e-06, "loss": 0.0131, "step": 39215 }, { "epoch": 2.090739457269286, "grad_norm": 1.6426984071731567, "learning_rate": 2.1570000000000003e-06, "loss": 0.0185, "step": 39216 }, { "epoch": 2.090792770698939, "grad_norm": 0.28716960549354553, "learning_rate": 2.1568000000000003e-06, "loss": 0.0026, "step": 39217 }, { "epoch": 2.090846084128592, "grad_norm": 2.772390127182007, "learning_rate": 2.1566000000000002e-06, "loss": 0.0304, "step": 39218 }, { "epoch": 2.090899397558245, "grad_norm": 1.9205914735794067, "learning_rate": 2.1564000000000002e-06, "loss": 0.0306, "step": 39219 }, { "epoch": 2.090952710987898, "grad_norm": 3.814406633377075, "learning_rate": 2.1562e-06, "loss": 0.0408, "step": 39220 }, { "epoch": 2.091006024417551, "grad_norm": 3.0218374729156494, "learning_rate": 2.156e-06, "loss": 0.0228, "step": 39221 }, { "epoch": 2.0910593378472035, "grad_norm": 1.0179548263549805, "learning_rate": 2.1558e-06, "loss": 0.0034, "step": 39222 }, { "epoch": 2.0911126512768567, "grad_norm": 1.2092440128326416, "learning_rate": 2.1556e-06, "loss": 0.012, "step": 39223 }, { "epoch": 2.0911659647065095, "grad_norm": 2.0451345443725586, "learning_rate": 2.1554e-06, "loss": 0.0123, "step": 39224 }, { "epoch": 2.0912192781361627, "grad_norm": 0.7497991919517517, "learning_rate": 2.1552e-06, "loss": 0.0166, "step": 39225 }, { "epoch": 2.0912725915658155, "grad_norm": 1.8336143493652344, "learning_rate": 2.155e-06, "loss": 0.01, "step": 39226 }, { "epoch": 2.091325904995468, "grad_norm": 2.2475333213806152, "learning_rate": 2.1548e-06, "loss": 0.0148, "step": 39227 }, { "epoch": 2.0913792184251214, "grad_norm": 2.1068198680877686, "learning_rate": 2.1546000000000005e-06, "loss": 0.0215, "step": 39228 }, { "epoch": 2.091432531854774, "grad_norm": 2.118330955505371, "learning_rate": 2.1544e-06, "loss": 0.0229, "step": 39229 }, { "epoch": 2.091485845284427, "grad_norm": 1.0425359010696411, "learning_rate": 2.1542e-06, "loss": 0.012, "step": 39230 }, { "epoch": 2.09153915871408, "grad_norm": 3.3852474689483643, "learning_rate": 2.1540000000000005e-06, "loss": 0.0146, "step": 39231 }, { "epoch": 2.091592472143733, "grad_norm": 1.7395484447479248, "learning_rate": 2.1538e-06, "loss": 0.0271, "step": 39232 }, { "epoch": 2.091645785573386, "grad_norm": 1.2195326089859009, "learning_rate": 2.1536e-06, "loss": 0.0097, "step": 39233 }, { "epoch": 2.091699099003039, "grad_norm": 2.7565853595733643, "learning_rate": 2.1534000000000004e-06, "loss": 0.0416, "step": 39234 }, { "epoch": 2.0917524124326916, "grad_norm": 0.3765687346458435, "learning_rate": 2.1532000000000004e-06, "loss": 0.0037, "step": 39235 }, { "epoch": 2.091805725862345, "grad_norm": 1.0113458633422852, "learning_rate": 2.153e-06, "loss": 0.0194, "step": 39236 }, { "epoch": 2.0918590392919976, "grad_norm": 1.2966769933700562, "learning_rate": 2.1528000000000004e-06, "loss": 0.0147, "step": 39237 }, { "epoch": 2.0919123527216508, "grad_norm": 0.4379764795303345, "learning_rate": 2.1526000000000004e-06, "loss": 0.0032, "step": 39238 }, { "epoch": 2.0919656661513035, "grad_norm": 1.4084336757659912, "learning_rate": 2.1524e-06, "loss": 0.013, "step": 39239 }, { "epoch": 2.0920189795809563, "grad_norm": 1.2498823404312134, "learning_rate": 2.1522000000000003e-06, "loss": 0.0177, "step": 39240 }, { "epoch": 2.0920722930106095, "grad_norm": 1.044883370399475, "learning_rate": 2.1520000000000003e-06, "loss": 0.0079, "step": 39241 }, { "epoch": 2.0921256064402622, "grad_norm": 1.3356616497039795, "learning_rate": 2.1518000000000003e-06, "loss": 0.0293, "step": 39242 }, { "epoch": 2.0921789198699154, "grad_norm": 0.8096807599067688, "learning_rate": 2.1516000000000003e-06, "loss": 0.0064, "step": 39243 }, { "epoch": 2.092232233299568, "grad_norm": 1.0757992267608643, "learning_rate": 2.1514000000000003e-06, "loss": 0.0077, "step": 39244 }, { "epoch": 2.092285546729221, "grad_norm": 3.164180278778076, "learning_rate": 2.1512000000000003e-06, "loss": 0.0377, "step": 39245 }, { "epoch": 2.092338860158874, "grad_norm": 0.9225243330001831, "learning_rate": 2.1510000000000002e-06, "loss": 0.02, "step": 39246 }, { "epoch": 2.092392173588527, "grad_norm": 2.25066876411438, "learning_rate": 2.1508000000000002e-06, "loss": 0.0208, "step": 39247 }, { "epoch": 2.0924454870181797, "grad_norm": 2.02485990524292, "learning_rate": 2.1506000000000002e-06, "loss": 0.01, "step": 39248 }, { "epoch": 2.092498800447833, "grad_norm": 2.846543550491333, "learning_rate": 2.1504e-06, "loss": 0.0425, "step": 39249 }, { "epoch": 2.0925521138774856, "grad_norm": 0.8443077206611633, "learning_rate": 2.1502e-06, "loss": 0.0164, "step": 39250 }, { "epoch": 2.092605427307139, "grad_norm": 1.6208860874176025, "learning_rate": 2.15e-06, "loss": 0.0129, "step": 39251 }, { "epoch": 2.0926587407367916, "grad_norm": 2.2498159408569336, "learning_rate": 2.1498e-06, "loss": 0.0237, "step": 39252 }, { "epoch": 2.0927120541664443, "grad_norm": 2.298790216445923, "learning_rate": 2.1496e-06, "loss": 0.0119, "step": 39253 }, { "epoch": 2.0927653675960975, "grad_norm": 1.2251478433609009, "learning_rate": 2.1494e-06, "loss": 0.0177, "step": 39254 }, { "epoch": 2.0928186810257503, "grad_norm": 1.235692024230957, "learning_rate": 2.1492e-06, "loss": 0.0195, "step": 39255 }, { "epoch": 2.0928719944554035, "grad_norm": 3.06805419921875, "learning_rate": 2.149e-06, "loss": 0.0245, "step": 39256 }, { "epoch": 2.0929253078850563, "grad_norm": 1.3417158126831055, "learning_rate": 2.1488e-06, "loss": 0.0139, "step": 39257 }, { "epoch": 2.092978621314709, "grad_norm": 0.9194629788398743, "learning_rate": 2.1486e-06, "loss": 0.0046, "step": 39258 }, { "epoch": 2.093031934744362, "grad_norm": 2.6584970951080322, "learning_rate": 2.1484e-06, "loss": 0.0219, "step": 39259 }, { "epoch": 2.093085248174015, "grad_norm": 1.9754070043563843, "learning_rate": 2.1482000000000005e-06, "loss": 0.0155, "step": 39260 }, { "epoch": 2.093138561603668, "grad_norm": 1.2384090423583984, "learning_rate": 2.148e-06, "loss": 0.0068, "step": 39261 }, { "epoch": 2.093191875033321, "grad_norm": 1.3949439525604248, "learning_rate": 2.1478e-06, "loss": 0.0321, "step": 39262 }, { "epoch": 2.0932451884629737, "grad_norm": 2.1907315254211426, "learning_rate": 2.1476000000000004e-06, "loss": 0.0315, "step": 39263 }, { "epoch": 2.093298501892627, "grad_norm": 1.2100472450256348, "learning_rate": 2.1474e-06, "loss": 0.0232, "step": 39264 }, { "epoch": 2.0933518153222797, "grad_norm": 1.2943028211593628, "learning_rate": 2.1472e-06, "loss": 0.02, "step": 39265 }, { "epoch": 2.0934051287519324, "grad_norm": 2.021174907684326, "learning_rate": 2.1470000000000004e-06, "loss": 0.0142, "step": 39266 }, { "epoch": 2.0934584421815856, "grad_norm": 1.2291934490203857, "learning_rate": 2.1468000000000004e-06, "loss": 0.0183, "step": 39267 }, { "epoch": 2.0935117556112384, "grad_norm": 0.8084448575973511, "learning_rate": 2.1466e-06, "loss": 0.0189, "step": 39268 }, { "epoch": 2.0935650690408916, "grad_norm": 1.4323288202285767, "learning_rate": 2.1464000000000003e-06, "loss": 0.0122, "step": 39269 }, { "epoch": 2.0936183824705443, "grad_norm": 1.4905287027359009, "learning_rate": 2.1462000000000003e-06, "loss": 0.0129, "step": 39270 }, { "epoch": 2.093671695900197, "grad_norm": 2.8645741939544678, "learning_rate": 2.1460000000000003e-06, "loss": 0.0328, "step": 39271 }, { "epoch": 2.0937250093298503, "grad_norm": 2.1885805130004883, "learning_rate": 2.1458000000000003e-06, "loss": 0.0319, "step": 39272 }, { "epoch": 2.093778322759503, "grad_norm": 1.7822413444519043, "learning_rate": 2.1456000000000003e-06, "loss": 0.0051, "step": 39273 }, { "epoch": 2.0938316361891562, "grad_norm": 1.3670731782913208, "learning_rate": 2.1454000000000002e-06, "loss": 0.0068, "step": 39274 }, { "epoch": 2.093884949618809, "grad_norm": 0.4483480453491211, "learning_rate": 2.1452000000000002e-06, "loss": 0.0041, "step": 39275 }, { "epoch": 2.0939382630484618, "grad_norm": 0.9272089004516602, "learning_rate": 2.1450000000000002e-06, "loss": 0.0085, "step": 39276 }, { "epoch": 2.093991576478115, "grad_norm": 1.153762936592102, "learning_rate": 2.1448e-06, "loss": 0.0091, "step": 39277 }, { "epoch": 2.0940448899077677, "grad_norm": 2.4201343059539795, "learning_rate": 2.1446e-06, "loss": 0.0177, "step": 39278 }, { "epoch": 2.0940982033374205, "grad_norm": 0.8234313726425171, "learning_rate": 2.1444e-06, "loss": 0.0038, "step": 39279 }, { "epoch": 2.0941515167670737, "grad_norm": 1.324253797531128, "learning_rate": 2.1442e-06, "loss": 0.0109, "step": 39280 }, { "epoch": 2.0942048301967264, "grad_norm": 1.1596499681472778, "learning_rate": 2.144e-06, "loss": 0.0056, "step": 39281 }, { "epoch": 2.0942581436263796, "grad_norm": 1.0732556581497192, "learning_rate": 2.1438e-06, "loss": 0.0071, "step": 39282 }, { "epoch": 2.0943114570560324, "grad_norm": 1.9445147514343262, "learning_rate": 2.1436e-06, "loss": 0.0205, "step": 39283 }, { "epoch": 2.094364770485685, "grad_norm": 2.333237648010254, "learning_rate": 2.1434e-06, "loss": 0.0082, "step": 39284 }, { "epoch": 2.0944180839153383, "grad_norm": 3.0592076778411865, "learning_rate": 2.1432e-06, "loss": 0.0283, "step": 39285 }, { "epoch": 2.094471397344991, "grad_norm": 1.0747841596603394, "learning_rate": 2.143e-06, "loss": 0.0237, "step": 39286 }, { "epoch": 2.0945247107746443, "grad_norm": 1.0934356451034546, "learning_rate": 2.1428e-06, "loss": 0.0176, "step": 39287 }, { "epoch": 2.094578024204297, "grad_norm": 1.0487351417541504, "learning_rate": 2.1426e-06, "loss": 0.0148, "step": 39288 }, { "epoch": 2.09463133763395, "grad_norm": 0.6248292326927185, "learning_rate": 2.1424000000000004e-06, "loss": 0.0043, "step": 39289 }, { "epoch": 2.094684651063603, "grad_norm": 0.9988123774528503, "learning_rate": 2.1422e-06, "loss": 0.0048, "step": 39290 }, { "epoch": 2.0947379644932558, "grad_norm": 2.344385862350464, "learning_rate": 2.142e-06, "loss": 0.0213, "step": 39291 }, { "epoch": 2.094791277922909, "grad_norm": 1.6806153059005737, "learning_rate": 2.1418000000000004e-06, "loss": 0.0137, "step": 39292 }, { "epoch": 2.0948445913525617, "grad_norm": 1.395947813987732, "learning_rate": 2.1416e-06, "loss": 0.0054, "step": 39293 }, { "epoch": 2.0948979047822145, "grad_norm": 3.0742030143737793, "learning_rate": 2.1414e-06, "loss": 0.008, "step": 39294 }, { "epoch": 2.0949512182118677, "grad_norm": 1.1533523797988892, "learning_rate": 2.1412000000000004e-06, "loss": 0.0102, "step": 39295 }, { "epoch": 2.0950045316415205, "grad_norm": 1.5695148706436157, "learning_rate": 2.1410000000000003e-06, "loss": 0.0109, "step": 39296 }, { "epoch": 2.0950578450711737, "grad_norm": 1.6689772605895996, "learning_rate": 2.1408e-06, "loss": 0.0162, "step": 39297 }, { "epoch": 2.0951111585008264, "grad_norm": 1.6267236471176147, "learning_rate": 2.1406000000000003e-06, "loss": 0.0228, "step": 39298 }, { "epoch": 2.095164471930479, "grad_norm": 1.7415564060211182, "learning_rate": 2.1404000000000003e-06, "loss": 0.0092, "step": 39299 }, { "epoch": 2.0952177853601324, "grad_norm": 0.8453873991966248, "learning_rate": 2.1402e-06, "loss": 0.0037, "step": 39300 }, { "epoch": 2.095271098789785, "grad_norm": 3.71354341506958, "learning_rate": 2.1400000000000003e-06, "loss": 0.0346, "step": 39301 }, { "epoch": 2.095324412219438, "grad_norm": 2.7080278396606445, "learning_rate": 2.1398000000000002e-06, "loss": 0.0163, "step": 39302 }, { "epoch": 2.095377725649091, "grad_norm": 3.4670183658599854, "learning_rate": 2.1396000000000002e-06, "loss": 0.009, "step": 39303 }, { "epoch": 2.095431039078744, "grad_norm": 1.6512717008590698, "learning_rate": 2.1394000000000002e-06, "loss": 0.028, "step": 39304 }, { "epoch": 2.095484352508397, "grad_norm": 2.564373731613159, "learning_rate": 2.1392e-06, "loss": 0.0123, "step": 39305 }, { "epoch": 2.09553766593805, "grad_norm": 2.8033885955810547, "learning_rate": 2.139e-06, "loss": 0.0483, "step": 39306 }, { "epoch": 2.0955909793677026, "grad_norm": 0.7097347974777222, "learning_rate": 2.1388e-06, "loss": 0.005, "step": 39307 }, { "epoch": 2.0956442927973558, "grad_norm": 3.7348575592041016, "learning_rate": 2.1386e-06, "loss": 0.0303, "step": 39308 }, { "epoch": 2.0956976062270085, "grad_norm": 1.8893455266952515, "learning_rate": 2.1384e-06, "loss": 0.0429, "step": 39309 }, { "epoch": 2.0957509196566617, "grad_norm": 0.35053491592407227, "learning_rate": 2.1382e-06, "loss": 0.0028, "step": 39310 }, { "epoch": 2.0958042330863145, "grad_norm": 2.0500006675720215, "learning_rate": 2.138e-06, "loss": 0.03, "step": 39311 }, { "epoch": 2.0958575465159672, "grad_norm": 3.4600048065185547, "learning_rate": 2.1378e-06, "loss": 0.0101, "step": 39312 }, { "epoch": 2.0959108599456204, "grad_norm": 1.70639967918396, "learning_rate": 2.1376e-06, "loss": 0.0075, "step": 39313 }, { "epoch": 2.095964173375273, "grad_norm": 2.125974416732788, "learning_rate": 2.1374000000000005e-06, "loss": 0.0262, "step": 39314 }, { "epoch": 2.096017486804926, "grad_norm": 1.7440475225448608, "learning_rate": 2.1372e-06, "loss": 0.0199, "step": 39315 }, { "epoch": 2.096070800234579, "grad_norm": 2.9834089279174805, "learning_rate": 2.137e-06, "loss": 0.0307, "step": 39316 }, { "epoch": 2.096124113664232, "grad_norm": 0.4701244533061981, "learning_rate": 2.1368000000000004e-06, "loss": 0.003, "step": 39317 }, { "epoch": 2.096177427093885, "grad_norm": 1.6471437215805054, "learning_rate": 2.1366e-06, "loss": 0.0156, "step": 39318 }, { "epoch": 2.096230740523538, "grad_norm": 2.9067561626434326, "learning_rate": 2.1364e-06, "loss": 0.015, "step": 39319 }, { "epoch": 2.0962840539531906, "grad_norm": 0.9131983518600464, "learning_rate": 2.1362000000000004e-06, "loss": 0.0131, "step": 39320 }, { "epoch": 2.096337367382844, "grad_norm": 2.064176082611084, "learning_rate": 2.1360000000000004e-06, "loss": 0.0244, "step": 39321 }, { "epoch": 2.0963906808124966, "grad_norm": 1.7688815593719482, "learning_rate": 2.1358e-06, "loss": 0.0295, "step": 39322 }, { "epoch": 2.0964439942421498, "grad_norm": 0.8223583102226257, "learning_rate": 2.1356000000000004e-06, "loss": 0.004, "step": 39323 }, { "epoch": 2.0964973076718025, "grad_norm": 2.1595375537872314, "learning_rate": 2.1354000000000003e-06, "loss": 0.0295, "step": 39324 }, { "epoch": 2.0965506211014553, "grad_norm": 1.8295036554336548, "learning_rate": 2.1352e-06, "loss": 0.0376, "step": 39325 }, { "epoch": 2.0966039345311085, "grad_norm": 3.4396347999572754, "learning_rate": 2.1350000000000003e-06, "loss": 0.0362, "step": 39326 }, { "epoch": 2.0966572479607613, "grad_norm": 5.304730415344238, "learning_rate": 2.1348000000000003e-06, "loss": 0.0546, "step": 39327 }, { "epoch": 2.0967105613904145, "grad_norm": 1.6107511520385742, "learning_rate": 2.1346000000000003e-06, "loss": 0.0216, "step": 39328 }, { "epoch": 2.096763874820067, "grad_norm": 1.5970927476882935, "learning_rate": 2.1344000000000003e-06, "loss": 0.0188, "step": 39329 }, { "epoch": 2.09681718824972, "grad_norm": 1.5668742656707764, "learning_rate": 2.1342000000000003e-06, "loss": 0.0262, "step": 39330 }, { "epoch": 2.096870501679373, "grad_norm": 2.9244766235351562, "learning_rate": 2.1340000000000002e-06, "loss": 0.0162, "step": 39331 }, { "epoch": 2.096923815109026, "grad_norm": 1.3814691305160522, "learning_rate": 2.1338000000000002e-06, "loss": 0.007, "step": 39332 }, { "epoch": 2.0969771285386787, "grad_norm": 2.9130678176879883, "learning_rate": 2.1336e-06, "loss": 0.0227, "step": 39333 }, { "epoch": 2.097030441968332, "grad_norm": 4.495689868927002, "learning_rate": 2.1334e-06, "loss": 0.0198, "step": 39334 }, { "epoch": 2.0970837553979846, "grad_norm": 0.31151118874549866, "learning_rate": 2.1332e-06, "loss": 0.003, "step": 39335 }, { "epoch": 2.097137068827638, "grad_norm": 0.6318862438201904, "learning_rate": 2.133e-06, "loss": 0.0053, "step": 39336 }, { "epoch": 2.0971903822572906, "grad_norm": 4.040657997131348, "learning_rate": 2.1328e-06, "loss": 0.0205, "step": 39337 }, { "epoch": 2.0972436956869434, "grad_norm": 0.9830033779144287, "learning_rate": 2.1326e-06, "loss": 0.0106, "step": 39338 }, { "epoch": 2.0972970091165966, "grad_norm": 2.150846004486084, "learning_rate": 2.1324e-06, "loss": 0.0228, "step": 39339 }, { "epoch": 2.0973503225462493, "grad_norm": 0.8282246589660645, "learning_rate": 2.1322e-06, "loss": 0.0173, "step": 39340 }, { "epoch": 2.0974036359759025, "grad_norm": 0.9322766661643982, "learning_rate": 2.132e-06, "loss": 0.0139, "step": 39341 }, { "epoch": 2.0974569494055553, "grad_norm": 1.1354435682296753, "learning_rate": 2.1318e-06, "loss": 0.0211, "step": 39342 }, { "epoch": 2.097510262835208, "grad_norm": 0.5427150726318359, "learning_rate": 2.1316e-06, "loss": 0.0144, "step": 39343 }, { "epoch": 2.0975635762648612, "grad_norm": 0.6820847988128662, "learning_rate": 2.1314e-06, "loss": 0.017, "step": 39344 }, { "epoch": 2.097616889694514, "grad_norm": 1.7443835735321045, "learning_rate": 2.1312e-06, "loss": 0.0243, "step": 39345 }, { "epoch": 2.097670203124167, "grad_norm": 0.14226993918418884, "learning_rate": 2.1310000000000004e-06, "loss": 0.0016, "step": 39346 }, { "epoch": 2.09772351655382, "grad_norm": 1.960073709487915, "learning_rate": 2.1308e-06, "loss": 0.0169, "step": 39347 }, { "epoch": 2.0977768299834727, "grad_norm": 2.422208309173584, "learning_rate": 2.1306e-06, "loss": 0.0275, "step": 39348 }, { "epoch": 2.097830143413126, "grad_norm": 2.8427159786224365, "learning_rate": 2.1304000000000004e-06, "loss": 0.0219, "step": 39349 }, { "epoch": 2.0978834568427787, "grad_norm": 0.10670637339353561, "learning_rate": 2.1302e-06, "loss": 0.0014, "step": 39350 }, { "epoch": 2.0979367702724314, "grad_norm": 1.6193163394927979, "learning_rate": 2.13e-06, "loss": 0.0072, "step": 39351 }, { "epoch": 2.0979900837020846, "grad_norm": 1.4267734289169312, "learning_rate": 2.1298000000000003e-06, "loss": 0.0103, "step": 39352 }, { "epoch": 2.0980433971317374, "grad_norm": 2.3790462017059326, "learning_rate": 2.1296000000000003e-06, "loss": 0.0206, "step": 39353 }, { "epoch": 2.0980967105613906, "grad_norm": 2.9860146045684814, "learning_rate": 2.1294e-06, "loss": 0.0098, "step": 39354 }, { "epoch": 2.0981500239910433, "grad_norm": 0.2674933671951294, "learning_rate": 2.1292000000000003e-06, "loss": 0.0027, "step": 39355 }, { "epoch": 2.098203337420696, "grad_norm": 3.3921303749084473, "learning_rate": 2.1290000000000003e-06, "loss": 0.0424, "step": 39356 }, { "epoch": 2.0982566508503493, "grad_norm": 0.9867553114891052, "learning_rate": 2.1288000000000003e-06, "loss": 0.0054, "step": 39357 }, { "epoch": 2.098309964280002, "grad_norm": 1.631212592124939, "learning_rate": 2.1286000000000003e-06, "loss": 0.0112, "step": 39358 }, { "epoch": 2.0983632777096553, "grad_norm": 1.9155421257019043, "learning_rate": 2.1284000000000002e-06, "loss": 0.0159, "step": 39359 }, { "epoch": 2.098416591139308, "grad_norm": 2.323737144470215, "learning_rate": 2.1282000000000002e-06, "loss": 0.0164, "step": 39360 }, { "epoch": 2.0984699045689608, "grad_norm": 2.338679552078247, "learning_rate": 2.128e-06, "loss": 0.0238, "step": 39361 }, { "epoch": 2.098523217998614, "grad_norm": 2.432971715927124, "learning_rate": 2.1278e-06, "loss": 0.0235, "step": 39362 }, { "epoch": 2.0985765314282667, "grad_norm": 2.776688575744629, "learning_rate": 2.1276e-06, "loss": 0.0304, "step": 39363 }, { "epoch": 2.09862984485792, "grad_norm": 1.6171274185180664, "learning_rate": 2.1274e-06, "loss": 0.0179, "step": 39364 }, { "epoch": 2.0986831582875727, "grad_norm": 3.023564577102661, "learning_rate": 2.1272e-06, "loss": 0.032, "step": 39365 }, { "epoch": 2.0987364717172254, "grad_norm": 2.573920249938965, "learning_rate": 2.127e-06, "loss": 0.0199, "step": 39366 }, { "epoch": 2.0987897851468786, "grad_norm": 0.8475965857505798, "learning_rate": 2.1268e-06, "loss": 0.0146, "step": 39367 }, { "epoch": 2.0988430985765314, "grad_norm": 1.3696495294570923, "learning_rate": 2.1266e-06, "loss": 0.019, "step": 39368 }, { "epoch": 2.098896412006184, "grad_norm": 1.7374261617660522, "learning_rate": 2.1264e-06, "loss": 0.0283, "step": 39369 }, { "epoch": 2.0989497254358374, "grad_norm": 3.2758285999298096, "learning_rate": 2.1262e-06, "loss": 0.0197, "step": 39370 }, { "epoch": 2.09900303886549, "grad_norm": 2.5222349166870117, "learning_rate": 2.1260000000000005e-06, "loss": 0.0188, "step": 39371 }, { "epoch": 2.0990563522951433, "grad_norm": 3.3879668712615967, "learning_rate": 2.1258e-06, "loss": 0.0246, "step": 39372 }, { "epoch": 2.099109665724796, "grad_norm": 1.3533008098602295, "learning_rate": 2.1256e-06, "loss": 0.0239, "step": 39373 }, { "epoch": 2.099162979154449, "grad_norm": 3.863480806350708, "learning_rate": 2.1254000000000004e-06, "loss": 0.0137, "step": 39374 }, { "epoch": 2.099216292584102, "grad_norm": 1.2868449687957764, "learning_rate": 2.1252e-06, "loss": 0.0137, "step": 39375 }, { "epoch": 2.099269606013755, "grad_norm": 0.3916742205619812, "learning_rate": 2.125e-06, "loss": 0.0027, "step": 39376 }, { "epoch": 2.099322919443408, "grad_norm": 1.3483235836029053, "learning_rate": 2.1248000000000004e-06, "loss": 0.0062, "step": 39377 }, { "epoch": 2.0993762328730607, "grad_norm": 0.7822124361991882, "learning_rate": 2.1246000000000004e-06, "loss": 0.005, "step": 39378 }, { "epoch": 2.0994295463027135, "grad_norm": 2.5698137283325195, "learning_rate": 2.1244e-06, "loss": 0.0328, "step": 39379 }, { "epoch": 2.0994828597323667, "grad_norm": 1.073114275932312, "learning_rate": 2.1242000000000003e-06, "loss": 0.0192, "step": 39380 }, { "epoch": 2.0995361731620195, "grad_norm": 3.1952626705169678, "learning_rate": 2.1240000000000003e-06, "loss": 0.0173, "step": 39381 }, { "epoch": 2.0995894865916727, "grad_norm": 2.702928066253662, "learning_rate": 2.1238000000000003e-06, "loss": 0.0096, "step": 39382 }, { "epoch": 2.0996428000213254, "grad_norm": 3.532999038696289, "learning_rate": 2.1236000000000003e-06, "loss": 0.038, "step": 39383 }, { "epoch": 2.099696113450978, "grad_norm": 0.2884455919265747, "learning_rate": 2.1234000000000003e-06, "loss": 0.0017, "step": 39384 }, { "epoch": 2.0997494268806314, "grad_norm": 4.044517993927002, "learning_rate": 2.1232000000000003e-06, "loss": 0.0229, "step": 39385 }, { "epoch": 2.099802740310284, "grad_norm": 1.4592807292938232, "learning_rate": 2.1230000000000003e-06, "loss": 0.0234, "step": 39386 }, { "epoch": 2.099856053739937, "grad_norm": 2.980006456375122, "learning_rate": 2.1228000000000002e-06, "loss": 0.0366, "step": 39387 }, { "epoch": 2.09990936716959, "grad_norm": 0.477436363697052, "learning_rate": 2.1226000000000002e-06, "loss": 0.0041, "step": 39388 }, { "epoch": 2.099962680599243, "grad_norm": 2.2536823749542236, "learning_rate": 2.1224e-06, "loss": 0.0372, "step": 39389 }, { "epoch": 2.100015994028896, "grad_norm": 1.7802938222885132, "learning_rate": 2.1222e-06, "loss": 0.0151, "step": 39390 }, { "epoch": 2.100069307458549, "grad_norm": 0.6300953030586243, "learning_rate": 2.122e-06, "loss": 0.0052, "step": 39391 }, { "epoch": 2.1001226208882016, "grad_norm": 1.848598837852478, "learning_rate": 2.1218e-06, "loss": 0.007, "step": 39392 }, { "epoch": 2.1001759343178548, "grad_norm": 1.6921707391738892, "learning_rate": 2.1216e-06, "loss": 0.0281, "step": 39393 }, { "epoch": 2.1002292477475075, "grad_norm": 1.2463260889053345, "learning_rate": 2.1214e-06, "loss": 0.0075, "step": 39394 }, { "epoch": 2.1002825611771607, "grad_norm": 1.951413631439209, "learning_rate": 2.1212e-06, "loss": 0.0115, "step": 39395 }, { "epoch": 2.1003358746068135, "grad_norm": 1.453255295753479, "learning_rate": 2.121e-06, "loss": 0.0086, "step": 39396 }, { "epoch": 2.1003891880364662, "grad_norm": 3.8518295288085938, "learning_rate": 2.1208e-06, "loss": 0.0478, "step": 39397 }, { "epoch": 2.1004425014661194, "grad_norm": 1.035737156867981, "learning_rate": 2.1206e-06, "loss": 0.0087, "step": 39398 }, { "epoch": 2.100495814895772, "grad_norm": 1.1452405452728271, "learning_rate": 2.1204e-06, "loss": 0.0132, "step": 39399 }, { "epoch": 2.100549128325425, "grad_norm": 1.3706609010696411, "learning_rate": 2.1202e-06, "loss": 0.0227, "step": 39400 }, { "epoch": 2.100602441755078, "grad_norm": 1.9198051691055298, "learning_rate": 2.12e-06, "loss": 0.017, "step": 39401 }, { "epoch": 2.100655755184731, "grad_norm": 1.556583046913147, "learning_rate": 2.1198e-06, "loss": 0.0171, "step": 39402 }, { "epoch": 2.100709068614384, "grad_norm": 0.41372618079185486, "learning_rate": 2.1196000000000004e-06, "loss": 0.0032, "step": 39403 }, { "epoch": 2.100762382044037, "grad_norm": 1.1786028146743774, "learning_rate": 2.1194e-06, "loss": 0.0061, "step": 39404 }, { "epoch": 2.1008156954736896, "grad_norm": 1.7022911310195923, "learning_rate": 2.1192e-06, "loss": 0.0127, "step": 39405 }, { "epoch": 2.100869008903343, "grad_norm": 1.6134692430496216, "learning_rate": 2.1190000000000004e-06, "loss": 0.0055, "step": 39406 }, { "epoch": 2.1009223223329956, "grad_norm": 2.5285403728485107, "learning_rate": 2.1188000000000004e-06, "loss": 0.0253, "step": 39407 }, { "epoch": 2.100975635762649, "grad_norm": 0.8752005100250244, "learning_rate": 2.1186e-06, "loss": 0.0246, "step": 39408 }, { "epoch": 2.1010289491923015, "grad_norm": 1.8402472734451294, "learning_rate": 2.1184000000000003e-06, "loss": 0.0153, "step": 39409 }, { "epoch": 2.1010822626219543, "grad_norm": 2.433885335922241, "learning_rate": 2.1182000000000003e-06, "loss": 0.0224, "step": 39410 }, { "epoch": 2.1011355760516075, "grad_norm": 2.018278121948242, "learning_rate": 2.118e-06, "loss": 0.0194, "step": 39411 }, { "epoch": 2.1011888894812603, "grad_norm": 0.7467679977416992, "learning_rate": 2.1178000000000003e-06, "loss": 0.006, "step": 39412 }, { "epoch": 2.1012422029109135, "grad_norm": 0.9993798732757568, "learning_rate": 2.1176000000000003e-06, "loss": 0.0108, "step": 39413 }, { "epoch": 2.101295516340566, "grad_norm": 1.213995337486267, "learning_rate": 2.1174000000000003e-06, "loss": 0.006, "step": 39414 }, { "epoch": 2.101348829770219, "grad_norm": 1.7586886882781982, "learning_rate": 2.1172000000000002e-06, "loss": 0.015, "step": 39415 }, { "epoch": 2.101402143199872, "grad_norm": 1.35385000705719, "learning_rate": 2.1170000000000002e-06, "loss": 0.006, "step": 39416 }, { "epoch": 2.101455456629525, "grad_norm": 2.064030170440674, "learning_rate": 2.1168000000000002e-06, "loss": 0.0122, "step": 39417 }, { "epoch": 2.101508770059178, "grad_norm": 0.6862253546714783, "learning_rate": 2.1166e-06, "loss": 0.0045, "step": 39418 }, { "epoch": 2.101562083488831, "grad_norm": 2.014636754989624, "learning_rate": 2.1164e-06, "loss": 0.0351, "step": 39419 }, { "epoch": 2.1016153969184836, "grad_norm": 1.3088654279708862, "learning_rate": 2.1162e-06, "loss": 0.0152, "step": 39420 }, { "epoch": 2.101668710348137, "grad_norm": 1.9945321083068848, "learning_rate": 2.116e-06, "loss": 0.028, "step": 39421 }, { "epoch": 2.1017220237777896, "grad_norm": 1.097687840461731, "learning_rate": 2.1158e-06, "loss": 0.0063, "step": 39422 }, { "epoch": 2.1017753372074424, "grad_norm": 2.3884670734405518, "learning_rate": 2.1156e-06, "loss": 0.0102, "step": 39423 }, { "epoch": 2.1018286506370956, "grad_norm": 2.9565131664276123, "learning_rate": 2.1154e-06, "loss": 0.0275, "step": 39424 }, { "epoch": 2.1018819640667483, "grad_norm": 2.361612558364868, "learning_rate": 2.1152000000000005e-06, "loss": 0.0137, "step": 39425 }, { "epoch": 2.1019352774964015, "grad_norm": 4.357141017913818, "learning_rate": 2.115e-06, "loss": 0.0332, "step": 39426 }, { "epoch": 2.1019885909260543, "grad_norm": 4.700377941131592, "learning_rate": 2.1148e-06, "loss": 0.0244, "step": 39427 }, { "epoch": 2.102041904355707, "grad_norm": 2.052687644958496, "learning_rate": 2.1146000000000005e-06, "loss": 0.0135, "step": 39428 }, { "epoch": 2.1020952177853602, "grad_norm": 1.8486894369125366, "learning_rate": 2.1144e-06, "loss": 0.0313, "step": 39429 }, { "epoch": 2.102148531215013, "grad_norm": 0.6515182256698608, "learning_rate": 2.1142e-06, "loss": 0.0055, "step": 39430 }, { "epoch": 2.102201844644666, "grad_norm": 2.5685617923736572, "learning_rate": 2.1140000000000004e-06, "loss": 0.0198, "step": 39431 }, { "epoch": 2.102255158074319, "grad_norm": 2.030106544494629, "learning_rate": 2.1138000000000004e-06, "loss": 0.0196, "step": 39432 }, { "epoch": 2.1023084715039717, "grad_norm": 3.046729326248169, "learning_rate": 2.1136e-06, "loss": 0.0262, "step": 39433 }, { "epoch": 2.102361784933625, "grad_norm": 0.6104692816734314, "learning_rate": 2.1134000000000004e-06, "loss": 0.0049, "step": 39434 }, { "epoch": 2.1024150983632777, "grad_norm": 1.8431425094604492, "learning_rate": 2.1132000000000004e-06, "loss": 0.0098, "step": 39435 }, { "epoch": 2.1024684117929304, "grad_norm": 2.1313910484313965, "learning_rate": 2.113e-06, "loss": 0.0062, "step": 39436 }, { "epoch": 2.1025217252225836, "grad_norm": 1.8609042167663574, "learning_rate": 2.1128000000000003e-06, "loss": 0.0155, "step": 39437 }, { "epoch": 2.1025750386522364, "grad_norm": 0.5785331726074219, "learning_rate": 2.1126000000000003e-06, "loss": 0.0043, "step": 39438 }, { "epoch": 2.1026283520818896, "grad_norm": 2.1854681968688965, "learning_rate": 2.1124000000000003e-06, "loss": 0.0192, "step": 39439 }, { "epoch": 2.1026816655115423, "grad_norm": 0.32761964201927185, "learning_rate": 2.1122000000000003e-06, "loss": 0.003, "step": 39440 }, { "epoch": 2.102734978941195, "grad_norm": 1.3119298219680786, "learning_rate": 2.1120000000000003e-06, "loss": 0.0211, "step": 39441 }, { "epoch": 2.1027882923708483, "grad_norm": 5.230432987213135, "learning_rate": 2.1118000000000003e-06, "loss": 0.0145, "step": 39442 }, { "epoch": 2.102841605800501, "grad_norm": 1.9788968563079834, "learning_rate": 2.1116000000000002e-06, "loss": 0.0111, "step": 39443 }, { "epoch": 2.1028949192301543, "grad_norm": 2.3851451873779297, "learning_rate": 2.1114000000000002e-06, "loss": 0.0074, "step": 39444 }, { "epoch": 2.102948232659807, "grad_norm": 1.6724355220794678, "learning_rate": 2.1112000000000002e-06, "loss": 0.0385, "step": 39445 }, { "epoch": 2.1030015460894598, "grad_norm": 2.62528657913208, "learning_rate": 2.111e-06, "loss": 0.0224, "step": 39446 }, { "epoch": 2.103054859519113, "grad_norm": 1.760179877281189, "learning_rate": 2.1108e-06, "loss": 0.0212, "step": 39447 }, { "epoch": 2.1031081729487657, "grad_norm": 1.999282717704773, "learning_rate": 2.1106e-06, "loss": 0.0378, "step": 39448 }, { "epoch": 2.103161486378419, "grad_norm": 0.6385349631309509, "learning_rate": 2.1104e-06, "loss": 0.0031, "step": 39449 }, { "epoch": 2.1032147998080717, "grad_norm": 0.3727358281612396, "learning_rate": 2.1102e-06, "loss": 0.0031, "step": 39450 }, { "epoch": 2.1032681132377244, "grad_norm": 2.2964465618133545, "learning_rate": 2.11e-06, "loss": 0.015, "step": 39451 }, { "epoch": 2.1033214266673776, "grad_norm": 0.40433311462402344, "learning_rate": 2.1098e-06, "loss": 0.0037, "step": 39452 }, { "epoch": 2.1033747400970304, "grad_norm": 0.8680659532546997, "learning_rate": 2.1096e-06, "loss": 0.0061, "step": 39453 }, { "epoch": 2.103428053526683, "grad_norm": 0.7815715670585632, "learning_rate": 2.1094e-06, "loss": 0.0053, "step": 39454 }, { "epoch": 2.1034813669563364, "grad_norm": 1.2212036848068237, "learning_rate": 2.1092e-06, "loss": 0.0108, "step": 39455 }, { "epoch": 2.103534680385989, "grad_norm": 2.320657968521118, "learning_rate": 2.109e-06, "loss": 0.0187, "step": 39456 }, { "epoch": 2.1035879938156423, "grad_norm": 2.7706780433654785, "learning_rate": 2.1088000000000005e-06, "loss": 0.0194, "step": 39457 }, { "epoch": 2.103641307245295, "grad_norm": 0.32664093375205994, "learning_rate": 2.1086e-06, "loss": 0.0024, "step": 39458 }, { "epoch": 2.103694620674948, "grad_norm": 0.39043867588043213, "learning_rate": 2.1084e-06, "loss": 0.003, "step": 39459 }, { "epoch": 2.103747934104601, "grad_norm": 1.6602942943572998, "learning_rate": 2.1082000000000004e-06, "loss": 0.0118, "step": 39460 }, { "epoch": 2.103801247534254, "grad_norm": 1.9339560270309448, "learning_rate": 2.108e-06, "loss": 0.0224, "step": 39461 }, { "epoch": 2.103854560963907, "grad_norm": 0.5133334994316101, "learning_rate": 2.1078e-06, "loss": 0.0039, "step": 39462 }, { "epoch": 2.1039078743935598, "grad_norm": 3.6714282035827637, "learning_rate": 2.1076000000000004e-06, "loss": 0.0172, "step": 39463 }, { "epoch": 2.1039611878232125, "grad_norm": 2.644230365753174, "learning_rate": 2.1074000000000004e-06, "loss": 0.0311, "step": 39464 }, { "epoch": 2.1040145012528657, "grad_norm": 0.8132414221763611, "learning_rate": 2.1072e-06, "loss": 0.0259, "step": 39465 }, { "epoch": 2.1040678146825185, "grad_norm": 1.9154417514801025, "learning_rate": 2.1070000000000003e-06, "loss": 0.01, "step": 39466 }, { "epoch": 2.1041211281121717, "grad_norm": 2.1992299556732178, "learning_rate": 2.1068000000000003e-06, "loss": 0.0283, "step": 39467 }, { "epoch": 2.1041744415418244, "grad_norm": 2.416297435760498, "learning_rate": 2.1066e-06, "loss": 0.014, "step": 39468 }, { "epoch": 2.104227754971477, "grad_norm": 0.3322329819202423, "learning_rate": 2.1064000000000003e-06, "loss": 0.0025, "step": 39469 }, { "epoch": 2.1042810684011304, "grad_norm": 2.3845667839050293, "learning_rate": 2.1062000000000003e-06, "loss": 0.0218, "step": 39470 }, { "epoch": 2.104334381830783, "grad_norm": 1.1335020065307617, "learning_rate": 2.1060000000000002e-06, "loss": 0.013, "step": 39471 }, { "epoch": 2.104387695260436, "grad_norm": 3.1262617111206055, "learning_rate": 2.1058000000000002e-06, "loss": 0.0583, "step": 39472 }, { "epoch": 2.104441008690089, "grad_norm": 2.1623222827911377, "learning_rate": 2.1056000000000002e-06, "loss": 0.0102, "step": 39473 }, { "epoch": 2.104494322119742, "grad_norm": 1.630517601966858, "learning_rate": 2.1054e-06, "loss": 0.0063, "step": 39474 }, { "epoch": 2.104547635549395, "grad_norm": 1.0266191959381104, "learning_rate": 2.1052e-06, "loss": 0.0234, "step": 39475 }, { "epoch": 2.104600948979048, "grad_norm": 1.5531450510025024, "learning_rate": 2.105e-06, "loss": 0.0097, "step": 39476 }, { "epoch": 2.1046542624087006, "grad_norm": 0.5332766175270081, "learning_rate": 2.1048e-06, "loss": 0.0017, "step": 39477 }, { "epoch": 2.1047075758383538, "grad_norm": 2.2392756938934326, "learning_rate": 2.1046e-06, "loss": 0.016, "step": 39478 }, { "epoch": 2.1047608892680065, "grad_norm": 1.8804137706756592, "learning_rate": 2.1044e-06, "loss": 0.018, "step": 39479 }, { "epoch": 2.1048142026976597, "grad_norm": 0.9846604466438293, "learning_rate": 2.1042e-06, "loss": 0.007, "step": 39480 }, { "epoch": 2.1048675161273125, "grad_norm": 1.188033938407898, "learning_rate": 2.104e-06, "loss": 0.0083, "step": 39481 }, { "epoch": 2.1049208295569652, "grad_norm": 0.25391867756843567, "learning_rate": 2.1038e-06, "loss": 0.0027, "step": 39482 }, { "epoch": 2.1049741429866184, "grad_norm": 2.4475412368774414, "learning_rate": 2.1036e-06, "loss": 0.0235, "step": 39483 }, { "epoch": 2.105027456416271, "grad_norm": 1.5928270816802979, "learning_rate": 2.1034e-06, "loss": 0.0163, "step": 39484 }, { "epoch": 2.105080769845924, "grad_norm": 2.008700132369995, "learning_rate": 2.1032e-06, "loss": 0.0268, "step": 39485 }, { "epoch": 2.105134083275577, "grad_norm": 1.4023563861846924, "learning_rate": 2.103e-06, "loss": 0.0081, "step": 39486 }, { "epoch": 2.10518739670523, "grad_norm": 1.0600636005401611, "learning_rate": 2.1028e-06, "loss": 0.0223, "step": 39487 }, { "epoch": 2.105240710134883, "grad_norm": 2.725856065750122, "learning_rate": 2.1026e-06, "loss": 0.0362, "step": 39488 }, { "epoch": 2.105294023564536, "grad_norm": 1.564601182937622, "learning_rate": 2.1024000000000004e-06, "loss": 0.007, "step": 39489 }, { "epoch": 2.1053473369941886, "grad_norm": 2.1493444442749023, "learning_rate": 2.1022e-06, "loss": 0.0239, "step": 39490 }, { "epoch": 2.105400650423842, "grad_norm": 1.6902270317077637, "learning_rate": 2.102e-06, "loss": 0.026, "step": 39491 }, { "epoch": 2.1054539638534946, "grad_norm": 1.25713050365448, "learning_rate": 2.1018000000000004e-06, "loss": 0.0123, "step": 39492 }, { "epoch": 2.105507277283148, "grad_norm": 1.5959362983703613, "learning_rate": 2.1016000000000003e-06, "loss": 0.0111, "step": 39493 }, { "epoch": 2.1055605907128006, "grad_norm": 2.1596572399139404, "learning_rate": 2.1014e-06, "loss": 0.0203, "step": 39494 }, { "epoch": 2.1056139041424533, "grad_norm": 2.490804433822632, "learning_rate": 2.1012000000000003e-06, "loss": 0.0084, "step": 39495 }, { "epoch": 2.1056672175721065, "grad_norm": 1.4773114919662476, "learning_rate": 2.1010000000000003e-06, "loss": 0.03, "step": 39496 }, { "epoch": 2.1057205310017593, "grad_norm": 3.2279772758483887, "learning_rate": 2.1008e-06, "loss": 0.01, "step": 39497 }, { "epoch": 2.1057738444314125, "grad_norm": 1.905547022819519, "learning_rate": 2.1006000000000003e-06, "loss": 0.0066, "step": 39498 }, { "epoch": 2.1058271578610652, "grad_norm": 2.377023458480835, "learning_rate": 2.1004000000000003e-06, "loss": 0.012, "step": 39499 }, { "epoch": 2.105880471290718, "grad_norm": 0.411310076713562, "learning_rate": 2.1002000000000002e-06, "loss": 0.0028, "step": 39500 }, { "epoch": 2.105933784720371, "grad_norm": 1.8105170726776123, "learning_rate": 2.1000000000000002e-06, "loss": 0.0127, "step": 39501 }, { "epoch": 2.105987098150024, "grad_norm": 1.8520578145980835, "learning_rate": 2.0998e-06, "loss": 0.0152, "step": 39502 }, { "epoch": 2.106040411579677, "grad_norm": 1.9545701742172241, "learning_rate": 2.0996e-06, "loss": 0.0201, "step": 39503 }, { "epoch": 2.10609372500933, "grad_norm": 1.8052496910095215, "learning_rate": 2.0994e-06, "loss": 0.0123, "step": 39504 }, { "epoch": 2.1061470384389827, "grad_norm": 2.3658206462860107, "learning_rate": 2.0992e-06, "loss": 0.0213, "step": 39505 }, { "epoch": 2.106200351868636, "grad_norm": 0.3724503815174103, "learning_rate": 2.099e-06, "loss": 0.0032, "step": 39506 }, { "epoch": 2.1062536652982886, "grad_norm": 2.72464656829834, "learning_rate": 2.0988e-06, "loss": 0.039, "step": 39507 }, { "epoch": 2.1063069787279414, "grad_norm": 2.089345932006836, "learning_rate": 2.0986e-06, "loss": 0.0267, "step": 39508 }, { "epoch": 2.1063602921575946, "grad_norm": 2.5452311038970947, "learning_rate": 2.0984e-06, "loss": 0.0143, "step": 39509 }, { "epoch": 2.1064136055872473, "grad_norm": 1.2226200103759766, "learning_rate": 2.0982e-06, "loss": 0.0118, "step": 39510 }, { "epoch": 2.1064669190169005, "grad_norm": 2.374525785446167, "learning_rate": 2.098e-06, "loss": 0.011, "step": 39511 }, { "epoch": 2.1065202324465533, "grad_norm": 0.7443493604660034, "learning_rate": 2.0978e-06, "loss": 0.0029, "step": 39512 }, { "epoch": 2.106573545876206, "grad_norm": 0.6588371992111206, "learning_rate": 2.0976e-06, "loss": 0.0048, "step": 39513 }, { "epoch": 2.1066268593058592, "grad_norm": 0.989273726940155, "learning_rate": 2.0974000000000005e-06, "loss": 0.026, "step": 39514 }, { "epoch": 2.106680172735512, "grad_norm": 0.9881563782691956, "learning_rate": 2.0972e-06, "loss": 0.0053, "step": 39515 }, { "epoch": 2.106733486165165, "grad_norm": 0.3522433638572693, "learning_rate": 2.097e-06, "loss": 0.0027, "step": 39516 }, { "epoch": 2.106786799594818, "grad_norm": 2.625790596008301, "learning_rate": 2.0968000000000004e-06, "loss": 0.0161, "step": 39517 }, { "epoch": 2.1068401130244707, "grad_norm": 1.1898363828659058, "learning_rate": 2.0966000000000004e-06, "loss": 0.0282, "step": 39518 }, { "epoch": 2.106893426454124, "grad_norm": 1.3737508058547974, "learning_rate": 2.0964e-06, "loss": 0.0201, "step": 39519 }, { "epoch": 2.1069467398837767, "grad_norm": 1.0464448928833008, "learning_rate": 2.0962000000000004e-06, "loss": 0.0047, "step": 39520 }, { "epoch": 2.1070000533134294, "grad_norm": 0.7500625848770142, "learning_rate": 2.0960000000000003e-06, "loss": 0.0058, "step": 39521 }, { "epoch": 2.1070533667430826, "grad_norm": 1.4956517219543457, "learning_rate": 2.0958e-06, "loss": 0.0086, "step": 39522 }, { "epoch": 2.1071066801727354, "grad_norm": 1.453570008277893, "learning_rate": 2.0956000000000003e-06, "loss": 0.0105, "step": 39523 }, { "epoch": 2.1071599936023886, "grad_norm": 0.8315523266792297, "learning_rate": 2.0954000000000003e-06, "loss": 0.0151, "step": 39524 }, { "epoch": 2.1072133070320413, "grad_norm": 1.8578966856002808, "learning_rate": 2.0952000000000003e-06, "loss": 0.0252, "step": 39525 }, { "epoch": 2.107266620461694, "grad_norm": 0.7024213075637817, "learning_rate": 2.0950000000000003e-06, "loss": 0.0035, "step": 39526 }, { "epoch": 2.1073199338913473, "grad_norm": 1.6573466062545776, "learning_rate": 2.0948000000000003e-06, "loss": 0.0277, "step": 39527 }, { "epoch": 2.107373247321, "grad_norm": 2.4351155757904053, "learning_rate": 2.0946000000000002e-06, "loss": 0.0235, "step": 39528 }, { "epoch": 2.1074265607506533, "grad_norm": 1.733396053314209, "learning_rate": 2.0944000000000002e-06, "loss": 0.0224, "step": 39529 }, { "epoch": 2.107479874180306, "grad_norm": 3.646749258041382, "learning_rate": 2.0942e-06, "loss": 0.0557, "step": 39530 }, { "epoch": 2.107533187609959, "grad_norm": 1.543701171875, "learning_rate": 2.094e-06, "loss": 0.0258, "step": 39531 }, { "epoch": 2.107586501039612, "grad_norm": 2.3411128520965576, "learning_rate": 2.0938e-06, "loss": 0.0105, "step": 39532 }, { "epoch": 2.1076398144692647, "grad_norm": 4.330833911895752, "learning_rate": 2.0936e-06, "loss": 0.0566, "step": 39533 }, { "epoch": 2.107693127898918, "grad_norm": 2.506836414337158, "learning_rate": 2.0934e-06, "loss": 0.0227, "step": 39534 }, { "epoch": 2.1077464413285707, "grad_norm": 1.6011933088302612, "learning_rate": 2.0932e-06, "loss": 0.0068, "step": 39535 }, { "epoch": 2.1077997547582235, "grad_norm": 2.124159097671509, "learning_rate": 2.093e-06, "loss": 0.0175, "step": 39536 }, { "epoch": 2.1078530681878767, "grad_norm": 1.9855453968048096, "learning_rate": 2.0928e-06, "loss": 0.0387, "step": 39537 }, { "epoch": 2.1079063816175294, "grad_norm": 1.4146274328231812, "learning_rate": 2.0926e-06, "loss": 0.0087, "step": 39538 }, { "epoch": 2.1079596950471826, "grad_norm": 1.7829803228378296, "learning_rate": 2.0924e-06, "loss": 0.0076, "step": 39539 }, { "epoch": 2.1080130084768354, "grad_norm": 1.273594617843628, "learning_rate": 2.0922e-06, "loss": 0.0052, "step": 39540 }, { "epoch": 2.108066321906488, "grad_norm": 1.4379382133483887, "learning_rate": 2.092e-06, "loss": 0.0297, "step": 39541 }, { "epoch": 2.1081196353361413, "grad_norm": 1.3334325551986694, "learning_rate": 2.0918e-06, "loss": 0.0284, "step": 39542 }, { "epoch": 2.108172948765794, "grad_norm": 3.2284862995147705, "learning_rate": 2.0916000000000004e-06, "loss": 0.0112, "step": 39543 }, { "epoch": 2.108226262195447, "grad_norm": 1.5594233274459839, "learning_rate": 2.0914e-06, "loss": 0.0238, "step": 39544 }, { "epoch": 2.1082795756251, "grad_norm": 1.0923562049865723, "learning_rate": 2.0912e-06, "loss": 0.0054, "step": 39545 }, { "epoch": 2.108332889054753, "grad_norm": 1.5535216331481934, "learning_rate": 2.0910000000000004e-06, "loss": 0.02, "step": 39546 }, { "epoch": 2.108386202484406, "grad_norm": 1.0546693801879883, "learning_rate": 2.0908e-06, "loss": 0.0069, "step": 39547 }, { "epoch": 2.1084395159140588, "grad_norm": 1.844226598739624, "learning_rate": 2.0906e-06, "loss": 0.0115, "step": 39548 }, { "epoch": 2.1084928293437115, "grad_norm": 1.3871153593063354, "learning_rate": 2.0904000000000003e-06, "loss": 0.013, "step": 39549 }, { "epoch": 2.1085461427733647, "grad_norm": 0.9698294401168823, "learning_rate": 2.0902000000000003e-06, "loss": 0.0072, "step": 39550 }, { "epoch": 2.1085994562030175, "grad_norm": 0.726223349571228, "learning_rate": 2.09e-06, "loss": 0.0036, "step": 39551 }, { "epoch": 2.1086527696326707, "grad_norm": 1.9518407583236694, "learning_rate": 2.0898000000000003e-06, "loss": 0.0167, "step": 39552 }, { "epoch": 2.1087060830623234, "grad_norm": 1.5345510244369507, "learning_rate": 2.0896000000000003e-06, "loss": 0.0341, "step": 39553 }, { "epoch": 2.108759396491976, "grad_norm": 2.053393602371216, "learning_rate": 2.0894e-06, "loss": 0.0194, "step": 39554 }, { "epoch": 2.1088127099216294, "grad_norm": 1.2210426330566406, "learning_rate": 2.0892000000000003e-06, "loss": 0.0085, "step": 39555 }, { "epoch": 2.108866023351282, "grad_norm": 1.9970059394836426, "learning_rate": 2.0890000000000002e-06, "loss": 0.0185, "step": 39556 }, { "epoch": 2.108919336780935, "grad_norm": 1.4595664739608765, "learning_rate": 2.0888000000000002e-06, "loss": 0.0199, "step": 39557 }, { "epoch": 2.108972650210588, "grad_norm": 1.2442569732666016, "learning_rate": 2.0886000000000002e-06, "loss": 0.0267, "step": 39558 }, { "epoch": 2.109025963640241, "grad_norm": 2.6156952381134033, "learning_rate": 2.0884e-06, "loss": 0.0287, "step": 39559 }, { "epoch": 2.109079277069894, "grad_norm": 4.24021577835083, "learning_rate": 2.0882e-06, "loss": 0.0173, "step": 39560 }, { "epoch": 2.109132590499547, "grad_norm": 3.228461980819702, "learning_rate": 2.088e-06, "loss": 0.0138, "step": 39561 }, { "epoch": 2.1091859039291996, "grad_norm": 1.2762444019317627, "learning_rate": 2.0878e-06, "loss": 0.0159, "step": 39562 }, { "epoch": 2.109239217358853, "grad_norm": 5.610987663269043, "learning_rate": 2.0876e-06, "loss": 0.0221, "step": 39563 }, { "epoch": 2.1092925307885055, "grad_norm": 3.439600706100464, "learning_rate": 2.0874e-06, "loss": 0.0591, "step": 39564 }, { "epoch": 2.1093458442181587, "grad_norm": 2.639578104019165, "learning_rate": 2.0872e-06, "loss": 0.0147, "step": 39565 }, { "epoch": 2.1093991576478115, "grad_norm": 1.1386185884475708, "learning_rate": 2.087e-06, "loss": 0.0076, "step": 39566 }, { "epoch": 2.1094524710774643, "grad_norm": 0.7441633939743042, "learning_rate": 2.0868e-06, "loss": 0.0059, "step": 39567 }, { "epoch": 2.1095057845071175, "grad_norm": 1.5614523887634277, "learning_rate": 2.0866000000000005e-06, "loss": 0.0173, "step": 39568 }, { "epoch": 2.10955909793677, "grad_norm": 2.842555284500122, "learning_rate": 2.0864e-06, "loss": 0.0248, "step": 39569 }, { "epoch": 2.1096124113664234, "grad_norm": 2.604543447494507, "learning_rate": 2.0862e-06, "loss": 0.0162, "step": 39570 }, { "epoch": 2.109665724796076, "grad_norm": 0.7010834217071533, "learning_rate": 2.0860000000000004e-06, "loss": 0.004, "step": 39571 }, { "epoch": 2.109719038225729, "grad_norm": 1.565604567527771, "learning_rate": 2.0858e-06, "loss": 0.0163, "step": 39572 }, { "epoch": 2.109772351655382, "grad_norm": 1.788639783859253, "learning_rate": 2.0856e-06, "loss": 0.0085, "step": 39573 }, { "epoch": 2.109825665085035, "grad_norm": 0.8380115032196045, "learning_rate": 2.0854000000000004e-06, "loss": 0.0182, "step": 39574 }, { "epoch": 2.1098789785146876, "grad_norm": 2.269343137741089, "learning_rate": 2.0852000000000004e-06, "loss": 0.0164, "step": 39575 }, { "epoch": 2.109932291944341, "grad_norm": 2.124800443649292, "learning_rate": 2.085e-06, "loss": 0.0174, "step": 39576 }, { "epoch": 2.1099856053739936, "grad_norm": 1.3164244890213013, "learning_rate": 2.0848000000000004e-06, "loss": 0.0078, "step": 39577 }, { "epoch": 2.110038918803647, "grad_norm": 1.4822827577590942, "learning_rate": 2.0846000000000003e-06, "loss": 0.0176, "step": 39578 }, { "epoch": 2.1100922322332996, "grad_norm": 3.3451621532440186, "learning_rate": 2.0844e-06, "loss": 0.0445, "step": 39579 }, { "epoch": 2.1101455456629523, "grad_norm": 3.1753664016723633, "learning_rate": 2.0842000000000003e-06, "loss": 0.0276, "step": 39580 }, { "epoch": 2.1101988590926055, "grad_norm": 0.9363002777099609, "learning_rate": 2.0840000000000003e-06, "loss": 0.0074, "step": 39581 }, { "epoch": 2.1102521725222583, "grad_norm": 1.3259435892105103, "learning_rate": 2.0838000000000003e-06, "loss": 0.0215, "step": 39582 }, { "epoch": 2.1103054859519115, "grad_norm": 1.9162931442260742, "learning_rate": 2.0836000000000003e-06, "loss": 0.0094, "step": 39583 }, { "epoch": 2.1103587993815642, "grad_norm": 1.0994731187820435, "learning_rate": 2.0834000000000002e-06, "loss": 0.007, "step": 39584 }, { "epoch": 2.110412112811217, "grad_norm": 0.4721364676952362, "learning_rate": 2.0832000000000002e-06, "loss": 0.0031, "step": 39585 }, { "epoch": 2.11046542624087, "grad_norm": 1.4939967393875122, "learning_rate": 2.0830000000000002e-06, "loss": 0.0246, "step": 39586 }, { "epoch": 2.110518739670523, "grad_norm": 1.1175642013549805, "learning_rate": 2.0828e-06, "loss": 0.0198, "step": 39587 }, { "epoch": 2.110572053100176, "grad_norm": 2.4934520721435547, "learning_rate": 2.0826e-06, "loss": 0.0351, "step": 39588 }, { "epoch": 2.110625366529829, "grad_norm": 3.843104362487793, "learning_rate": 2.0824e-06, "loss": 0.071, "step": 39589 }, { "epoch": 2.1106786799594817, "grad_norm": 1.7922457456588745, "learning_rate": 2.0822e-06, "loss": 0.0123, "step": 39590 }, { "epoch": 2.110731993389135, "grad_norm": 2.636256217956543, "learning_rate": 2.082e-06, "loss": 0.0334, "step": 39591 }, { "epoch": 2.1107853068187876, "grad_norm": 3.8162038326263428, "learning_rate": 2.0818e-06, "loss": 0.0241, "step": 39592 }, { "epoch": 2.1108386202484404, "grad_norm": 3.930710554122925, "learning_rate": 2.0816e-06, "loss": 0.0116, "step": 39593 }, { "epoch": 2.1108919336780936, "grad_norm": 2.789339780807495, "learning_rate": 2.0814e-06, "loss": 0.0197, "step": 39594 }, { "epoch": 2.1109452471077463, "grad_norm": 6.3529133796691895, "learning_rate": 2.0812e-06, "loss": 0.0421, "step": 39595 }, { "epoch": 2.1109985605373995, "grad_norm": 1.225299596786499, "learning_rate": 2.081e-06, "loss": 0.0052, "step": 39596 }, { "epoch": 2.1110518739670523, "grad_norm": 1.6514657735824585, "learning_rate": 2.0808e-06, "loss": 0.022, "step": 39597 }, { "epoch": 2.111105187396705, "grad_norm": 1.212822675704956, "learning_rate": 2.0806e-06, "loss": 0.0163, "step": 39598 }, { "epoch": 2.1111585008263583, "grad_norm": 2.0417885780334473, "learning_rate": 2.0804e-06, "loss": 0.0316, "step": 39599 }, { "epoch": 2.111211814256011, "grad_norm": 0.6692584156990051, "learning_rate": 2.0802000000000004e-06, "loss": 0.0048, "step": 39600 }, { "epoch": 2.111265127685664, "grad_norm": 0.6941445469856262, "learning_rate": 2.08e-06, "loss": 0.0037, "step": 39601 }, { "epoch": 2.111318441115317, "grad_norm": 1.3930065631866455, "learning_rate": 2.0798e-06, "loss": 0.0113, "step": 39602 }, { "epoch": 2.1113717545449697, "grad_norm": 2.301713705062866, "learning_rate": 2.0796000000000004e-06, "loss": 0.0195, "step": 39603 }, { "epoch": 2.111425067974623, "grad_norm": 2.243894338607788, "learning_rate": 2.0794000000000004e-06, "loss": 0.0123, "step": 39604 }, { "epoch": 2.1114783814042757, "grad_norm": 2.8852713108062744, "learning_rate": 2.0792e-06, "loss": 0.0153, "step": 39605 }, { "epoch": 2.1115316948339284, "grad_norm": 1.0767940282821655, "learning_rate": 2.0790000000000003e-06, "loss": 0.0268, "step": 39606 }, { "epoch": 2.1115850082635816, "grad_norm": 1.3319034576416016, "learning_rate": 2.0788000000000003e-06, "loss": 0.0254, "step": 39607 }, { "epoch": 2.1116383216932344, "grad_norm": 2.0795130729675293, "learning_rate": 2.0786e-06, "loss": 0.0227, "step": 39608 }, { "epoch": 2.1116916351228876, "grad_norm": 1.7775096893310547, "learning_rate": 2.0784000000000003e-06, "loss": 0.0094, "step": 39609 }, { "epoch": 2.1117449485525404, "grad_norm": 2.0789616107940674, "learning_rate": 2.0782000000000003e-06, "loss": 0.0068, "step": 39610 }, { "epoch": 2.111798261982193, "grad_norm": 0.8954402208328247, "learning_rate": 2.0780000000000003e-06, "loss": 0.0031, "step": 39611 }, { "epoch": 2.1118515754118463, "grad_norm": 2.7839298248291016, "learning_rate": 2.0778000000000002e-06, "loss": 0.0567, "step": 39612 }, { "epoch": 2.111904888841499, "grad_norm": 3.3038299083709717, "learning_rate": 2.0776000000000002e-06, "loss": 0.0353, "step": 39613 }, { "epoch": 2.1119582022711523, "grad_norm": 1.336780309677124, "learning_rate": 2.0774000000000002e-06, "loss": 0.0148, "step": 39614 }, { "epoch": 2.112011515700805, "grad_norm": 1.1352698802947998, "learning_rate": 2.0772e-06, "loss": 0.0132, "step": 39615 }, { "epoch": 2.112064829130458, "grad_norm": 1.177215814590454, "learning_rate": 2.077e-06, "loss": 0.0073, "step": 39616 }, { "epoch": 2.112118142560111, "grad_norm": 2.79194712638855, "learning_rate": 2.0768e-06, "loss": 0.0376, "step": 39617 }, { "epoch": 2.1121714559897637, "grad_norm": 1.062960147857666, "learning_rate": 2.0766e-06, "loss": 0.0105, "step": 39618 }, { "epoch": 2.112224769419417, "grad_norm": 1.840169072151184, "learning_rate": 2.0764e-06, "loss": 0.01, "step": 39619 }, { "epoch": 2.1122780828490697, "grad_norm": 1.5783566236495972, "learning_rate": 2.0762e-06, "loss": 0.0346, "step": 39620 }, { "epoch": 2.1123313962787225, "grad_norm": 1.1065248250961304, "learning_rate": 2.076e-06, "loss": 0.0106, "step": 39621 }, { "epoch": 2.1123847097083757, "grad_norm": 1.1413915157318115, "learning_rate": 2.0758e-06, "loss": 0.0178, "step": 39622 }, { "epoch": 2.1124380231380284, "grad_norm": 2.98954439163208, "learning_rate": 2.0756e-06, "loss": 0.0634, "step": 39623 }, { "epoch": 2.1124913365676816, "grad_norm": 1.500654935836792, "learning_rate": 2.0754e-06, "loss": 0.0119, "step": 39624 }, { "epoch": 2.1125446499973344, "grad_norm": 1.977470874786377, "learning_rate": 2.0752000000000005e-06, "loss": 0.0339, "step": 39625 }, { "epoch": 2.112597963426987, "grad_norm": 2.0528311729431152, "learning_rate": 2.075e-06, "loss": 0.0222, "step": 39626 }, { "epoch": 2.1126512768566403, "grad_norm": 1.2448254823684692, "learning_rate": 2.0748e-06, "loss": 0.0083, "step": 39627 }, { "epoch": 2.112704590286293, "grad_norm": 3.862414836883545, "learning_rate": 2.0746000000000004e-06, "loss": 0.047, "step": 39628 }, { "epoch": 2.112757903715946, "grad_norm": 1.889695405960083, "learning_rate": 2.0744000000000004e-06, "loss": 0.0244, "step": 39629 }, { "epoch": 2.112811217145599, "grad_norm": 1.144483208656311, "learning_rate": 2.0742e-06, "loss": 0.0122, "step": 39630 }, { "epoch": 2.112864530575252, "grad_norm": 1.094602346420288, "learning_rate": 2.0740000000000004e-06, "loss": 0.0149, "step": 39631 }, { "epoch": 2.112917844004905, "grad_norm": 2.7690072059631348, "learning_rate": 2.0738000000000004e-06, "loss": 0.0214, "step": 39632 }, { "epoch": 2.1129711574345578, "grad_norm": 2.176278829574585, "learning_rate": 2.0736e-06, "loss": 0.0245, "step": 39633 }, { "epoch": 2.1130244708642105, "grad_norm": 1.4360610246658325, "learning_rate": 2.0734000000000003e-06, "loss": 0.0139, "step": 39634 }, { "epoch": 2.1130777842938637, "grad_norm": 1.3315649032592773, "learning_rate": 2.0732000000000003e-06, "loss": 0.0132, "step": 39635 }, { "epoch": 2.1131310977235165, "grad_norm": 1.8367044925689697, "learning_rate": 2.0730000000000003e-06, "loss": 0.0152, "step": 39636 }, { "epoch": 2.1131844111531697, "grad_norm": 1.6167641878128052, "learning_rate": 2.0728000000000003e-06, "loss": 0.0056, "step": 39637 }, { "epoch": 2.1132377245828224, "grad_norm": 3.703282356262207, "learning_rate": 2.0726000000000003e-06, "loss": 0.014, "step": 39638 }, { "epoch": 2.113291038012475, "grad_norm": 3.7326314449310303, "learning_rate": 2.0724000000000003e-06, "loss": 0.0169, "step": 39639 }, { "epoch": 2.1133443514421284, "grad_norm": 1.7031277418136597, "learning_rate": 2.0722000000000003e-06, "loss": 0.0063, "step": 39640 }, { "epoch": 2.113397664871781, "grad_norm": 1.9284031391143799, "learning_rate": 2.0720000000000002e-06, "loss": 0.0453, "step": 39641 }, { "epoch": 2.113450978301434, "grad_norm": 3.3768527507781982, "learning_rate": 2.0718000000000002e-06, "loss": 0.0212, "step": 39642 }, { "epoch": 2.113504291731087, "grad_norm": 1.0048465728759766, "learning_rate": 2.0716e-06, "loss": 0.0049, "step": 39643 }, { "epoch": 2.11355760516074, "grad_norm": 1.107133150100708, "learning_rate": 2.0714e-06, "loss": 0.0044, "step": 39644 }, { "epoch": 2.113610918590393, "grad_norm": 0.7691963911056519, "learning_rate": 2.0712e-06, "loss": 0.006, "step": 39645 }, { "epoch": 2.113664232020046, "grad_norm": 0.5331312417984009, "learning_rate": 2.071e-06, "loss": 0.0037, "step": 39646 }, { "epoch": 2.1137175454496986, "grad_norm": 0.3259829580783844, "learning_rate": 2.0708e-06, "loss": 0.0035, "step": 39647 }, { "epoch": 2.113770858879352, "grad_norm": 2.1531693935394287, "learning_rate": 2.0706e-06, "loss": 0.0252, "step": 39648 }, { "epoch": 2.1138241723090045, "grad_norm": 2.2063193321228027, "learning_rate": 2.0704e-06, "loss": 0.0252, "step": 39649 }, { "epoch": 2.1138774857386577, "grad_norm": 2.43247652053833, "learning_rate": 2.0702e-06, "loss": 0.0194, "step": 39650 }, { "epoch": 2.1139307991683105, "grad_norm": 1.6558940410614014, "learning_rate": 2.07e-06, "loss": 0.0105, "step": 39651 }, { "epoch": 2.1139841125979633, "grad_norm": 1.5852371454238892, "learning_rate": 2.0698e-06, "loss": 0.0085, "step": 39652 }, { "epoch": 2.1140374260276165, "grad_norm": 1.2768183946609497, "learning_rate": 2.0696e-06, "loss": 0.0134, "step": 39653 }, { "epoch": 2.114090739457269, "grad_norm": 1.839147686958313, "learning_rate": 2.0694000000000005e-06, "loss": 0.0281, "step": 39654 }, { "epoch": 2.1141440528869224, "grad_norm": 1.998150110244751, "learning_rate": 2.0692e-06, "loss": 0.0456, "step": 39655 }, { "epoch": 2.114197366316575, "grad_norm": 3.957292079925537, "learning_rate": 2.069e-06, "loss": 0.0362, "step": 39656 }, { "epoch": 2.114250679746228, "grad_norm": 2.4011240005493164, "learning_rate": 2.0688000000000004e-06, "loss": 0.0257, "step": 39657 }, { "epoch": 2.114303993175881, "grad_norm": 2.011035203933716, "learning_rate": 2.0686e-06, "loss": 0.0278, "step": 39658 }, { "epoch": 2.114357306605534, "grad_norm": 1.8909579515457153, "learning_rate": 2.0684e-06, "loss": 0.0176, "step": 39659 }, { "epoch": 2.1144106200351866, "grad_norm": 2.1065187454223633, "learning_rate": 2.0682000000000004e-06, "loss": 0.0459, "step": 39660 }, { "epoch": 2.11446393346484, "grad_norm": 2.1834371089935303, "learning_rate": 2.0680000000000004e-06, "loss": 0.0189, "step": 39661 }, { "epoch": 2.1145172468944926, "grad_norm": 1.1078321933746338, "learning_rate": 2.0678e-06, "loss": 0.0107, "step": 39662 }, { "epoch": 2.114570560324146, "grad_norm": 1.5108733177185059, "learning_rate": 2.0676000000000003e-06, "loss": 0.0149, "step": 39663 }, { "epoch": 2.1146238737537986, "grad_norm": 1.4452524185180664, "learning_rate": 2.0674000000000003e-06, "loss": 0.0089, "step": 39664 }, { "epoch": 2.1146771871834513, "grad_norm": 2.2202863693237305, "learning_rate": 2.0672e-06, "loss": 0.0299, "step": 39665 }, { "epoch": 2.1147305006131045, "grad_norm": 1.9929745197296143, "learning_rate": 2.0670000000000003e-06, "loss": 0.032, "step": 39666 }, { "epoch": 2.1147838140427573, "grad_norm": 1.8765718936920166, "learning_rate": 2.0668000000000003e-06, "loss": 0.0207, "step": 39667 }, { "epoch": 2.1148371274724105, "grad_norm": 0.4718950390815735, "learning_rate": 2.0666000000000003e-06, "loss": 0.004, "step": 39668 }, { "epoch": 2.1148904409020632, "grad_norm": 1.2898483276367188, "learning_rate": 2.0664000000000002e-06, "loss": 0.0076, "step": 39669 }, { "epoch": 2.114943754331716, "grad_norm": 1.804019808769226, "learning_rate": 2.0662000000000002e-06, "loss": 0.0176, "step": 39670 }, { "epoch": 2.114997067761369, "grad_norm": 1.6610472202301025, "learning_rate": 2.066e-06, "loss": 0.0251, "step": 39671 }, { "epoch": 2.115050381191022, "grad_norm": 3.314235210418701, "learning_rate": 2.0658e-06, "loss": 0.0211, "step": 39672 }, { "epoch": 2.115103694620675, "grad_norm": 1.5200586318969727, "learning_rate": 2.0656e-06, "loss": 0.0116, "step": 39673 }, { "epoch": 2.115157008050328, "grad_norm": 2.3463900089263916, "learning_rate": 2.0654e-06, "loss": 0.0314, "step": 39674 }, { "epoch": 2.1152103214799807, "grad_norm": 4.4424967765808105, "learning_rate": 2.0652e-06, "loss": 0.0204, "step": 39675 }, { "epoch": 2.115263634909634, "grad_norm": 2.988919734954834, "learning_rate": 2.065e-06, "loss": 0.0127, "step": 39676 }, { "epoch": 2.1153169483392866, "grad_norm": 1.5731326341629028, "learning_rate": 2.0648e-06, "loss": 0.014, "step": 39677 }, { "epoch": 2.1153702617689394, "grad_norm": 0.353259414434433, "learning_rate": 2.0646e-06, "loss": 0.0032, "step": 39678 }, { "epoch": 2.1154235751985926, "grad_norm": 2.7299387454986572, "learning_rate": 2.0644e-06, "loss": 0.0598, "step": 39679 }, { "epoch": 2.1154768886282453, "grad_norm": 2.8719165325164795, "learning_rate": 2.0642e-06, "loss": 0.025, "step": 39680 }, { "epoch": 2.1155302020578985, "grad_norm": 1.4330285787582397, "learning_rate": 2.064e-06, "loss": 0.015, "step": 39681 }, { "epoch": 2.1155835154875513, "grad_norm": 2.9832046031951904, "learning_rate": 2.0638e-06, "loss": 0.0218, "step": 39682 }, { "epoch": 2.115636828917204, "grad_norm": 2.018246650695801, "learning_rate": 2.0636e-06, "loss": 0.0068, "step": 39683 }, { "epoch": 2.1156901423468573, "grad_norm": 0.9492295384407043, "learning_rate": 2.0634e-06, "loss": 0.0057, "step": 39684 }, { "epoch": 2.11574345577651, "grad_norm": 1.8520466089248657, "learning_rate": 2.0632e-06, "loss": 0.0107, "step": 39685 }, { "epoch": 2.115796769206163, "grad_norm": 0.4633627235889435, "learning_rate": 2.0630000000000004e-06, "loss": 0.0033, "step": 39686 }, { "epoch": 2.115850082635816, "grad_norm": 0.42657986283302307, "learning_rate": 2.0628e-06, "loss": 0.0034, "step": 39687 }, { "epoch": 2.1159033960654687, "grad_norm": 1.98308527469635, "learning_rate": 2.0626e-06, "loss": 0.0144, "step": 39688 }, { "epoch": 2.115956709495122, "grad_norm": 0.8092917799949646, "learning_rate": 2.0624000000000004e-06, "loss": 0.0069, "step": 39689 }, { "epoch": 2.1160100229247747, "grad_norm": 1.4602214097976685, "learning_rate": 2.0622e-06, "loss": 0.0205, "step": 39690 }, { "epoch": 2.116063336354428, "grad_norm": 1.830894112586975, "learning_rate": 2.062e-06, "loss": 0.0439, "step": 39691 }, { "epoch": 2.1161166497840806, "grad_norm": 1.7377190589904785, "learning_rate": 2.0618000000000003e-06, "loss": 0.0085, "step": 39692 }, { "epoch": 2.1161699632137334, "grad_norm": 3.3329546451568604, "learning_rate": 2.0616000000000003e-06, "loss": 0.0482, "step": 39693 }, { "epoch": 2.1162232766433866, "grad_norm": 1.6032525300979614, "learning_rate": 2.0614e-06, "loss": 0.0065, "step": 39694 }, { "epoch": 2.1162765900730394, "grad_norm": 2.161083221435547, "learning_rate": 2.0612000000000003e-06, "loss": 0.0086, "step": 39695 }, { "epoch": 2.116329903502692, "grad_norm": 2.6132218837738037, "learning_rate": 2.0610000000000003e-06, "loss": 0.0158, "step": 39696 }, { "epoch": 2.1163832169323453, "grad_norm": 1.8740493059158325, "learning_rate": 2.0608000000000002e-06, "loss": 0.0193, "step": 39697 }, { "epoch": 2.116436530361998, "grad_norm": 1.191148281097412, "learning_rate": 2.0606000000000002e-06, "loss": 0.0134, "step": 39698 }, { "epoch": 2.1164898437916513, "grad_norm": 1.7207967042922974, "learning_rate": 2.0604000000000002e-06, "loss": 0.0285, "step": 39699 }, { "epoch": 2.116543157221304, "grad_norm": 2.101783275604248, "learning_rate": 2.0602e-06, "loss": 0.038, "step": 39700 }, { "epoch": 2.116596470650957, "grad_norm": 1.3164032697677612, "learning_rate": 2.06e-06, "loss": 0.0066, "step": 39701 }, { "epoch": 2.11664978408061, "grad_norm": 4.508659839630127, "learning_rate": 2.0598e-06, "loss": 0.0129, "step": 39702 }, { "epoch": 2.1167030975102628, "grad_norm": 3.960716485977173, "learning_rate": 2.0596e-06, "loss": 0.0252, "step": 39703 }, { "epoch": 2.116756410939916, "grad_norm": 1.8770843744277954, "learning_rate": 2.0594e-06, "loss": 0.0107, "step": 39704 }, { "epoch": 2.1168097243695687, "grad_norm": 1.6352324485778809, "learning_rate": 2.0592e-06, "loss": 0.0102, "step": 39705 }, { "epoch": 2.1168630377992215, "grad_norm": 1.6188175678253174, "learning_rate": 2.059e-06, "loss": 0.0329, "step": 39706 }, { "epoch": 2.1169163512288747, "grad_norm": 2.591048240661621, "learning_rate": 2.0588e-06, "loss": 0.0262, "step": 39707 }, { "epoch": 2.1169696646585274, "grad_norm": 0.8796894550323486, "learning_rate": 2.0586e-06, "loss": 0.0047, "step": 39708 }, { "epoch": 2.1170229780881806, "grad_norm": 1.0416696071624756, "learning_rate": 2.0584e-06, "loss": 0.0059, "step": 39709 }, { "epoch": 2.1170762915178334, "grad_norm": 1.5267993211746216, "learning_rate": 2.0582e-06, "loss": 0.01, "step": 39710 }, { "epoch": 2.117129604947486, "grad_norm": 2.742716073989868, "learning_rate": 2.0580000000000005e-06, "loss": 0.0295, "step": 39711 }, { "epoch": 2.1171829183771393, "grad_norm": 1.032636284828186, "learning_rate": 2.0578e-06, "loss": 0.0057, "step": 39712 }, { "epoch": 2.117236231806792, "grad_norm": 2.828198194503784, "learning_rate": 2.0576e-06, "loss": 0.0198, "step": 39713 }, { "epoch": 2.117289545236445, "grad_norm": 2.19320011138916, "learning_rate": 2.0574000000000004e-06, "loss": 0.0081, "step": 39714 }, { "epoch": 2.117342858666098, "grad_norm": 2.2708077430725098, "learning_rate": 2.0572e-06, "loss": 0.02, "step": 39715 }, { "epoch": 2.117396172095751, "grad_norm": 1.4660724401474, "learning_rate": 2.057e-06, "loss": 0.0116, "step": 39716 }, { "epoch": 2.117449485525404, "grad_norm": 1.5706920623779297, "learning_rate": 2.0568000000000004e-06, "loss": 0.0113, "step": 39717 }, { "epoch": 2.1175027989550568, "grad_norm": 2.1239449977874756, "learning_rate": 2.0566000000000004e-06, "loss": 0.0245, "step": 39718 }, { "epoch": 2.1175561123847095, "grad_norm": 0.8207659721374512, "learning_rate": 2.0564e-06, "loss": 0.0037, "step": 39719 }, { "epoch": 2.1176094258143627, "grad_norm": 3.085012674331665, "learning_rate": 2.0562000000000003e-06, "loss": 0.0241, "step": 39720 }, { "epoch": 2.1176627392440155, "grad_norm": 1.6294463872909546, "learning_rate": 2.0560000000000003e-06, "loss": 0.0085, "step": 39721 }, { "epoch": 2.1177160526736687, "grad_norm": 1.1100311279296875, "learning_rate": 2.0558000000000003e-06, "loss": 0.0065, "step": 39722 }, { "epoch": 2.1177693661033214, "grad_norm": 2.0711586475372314, "learning_rate": 2.0556000000000003e-06, "loss": 0.0216, "step": 39723 }, { "epoch": 2.117822679532974, "grad_norm": 2.4517829418182373, "learning_rate": 2.0554000000000003e-06, "loss": 0.0191, "step": 39724 }, { "epoch": 2.1178759929626274, "grad_norm": 1.2508093118667603, "learning_rate": 2.0552000000000002e-06, "loss": 0.0195, "step": 39725 }, { "epoch": 2.11792930639228, "grad_norm": 1.6171480417251587, "learning_rate": 2.0550000000000002e-06, "loss": 0.0275, "step": 39726 }, { "epoch": 2.117982619821933, "grad_norm": 2.532503604888916, "learning_rate": 2.0548000000000002e-06, "loss": 0.0259, "step": 39727 }, { "epoch": 2.118035933251586, "grad_norm": 1.490613579750061, "learning_rate": 2.0546e-06, "loss": 0.0075, "step": 39728 }, { "epoch": 2.118089246681239, "grad_norm": 1.6399027109146118, "learning_rate": 2.0544e-06, "loss": 0.0226, "step": 39729 }, { "epoch": 2.118142560110892, "grad_norm": 2.1038291454315186, "learning_rate": 2.0542e-06, "loss": 0.0276, "step": 39730 }, { "epoch": 2.118195873540545, "grad_norm": 1.220019817352295, "learning_rate": 2.054e-06, "loss": 0.0094, "step": 39731 }, { "epoch": 2.1182491869701976, "grad_norm": 1.8556140661239624, "learning_rate": 2.0538e-06, "loss": 0.0193, "step": 39732 }, { "epoch": 2.118302500399851, "grad_norm": 4.588583469390869, "learning_rate": 2.0536e-06, "loss": 0.0328, "step": 39733 }, { "epoch": 2.1183558138295036, "grad_norm": 2.5095415115356445, "learning_rate": 2.0534e-06, "loss": 0.0173, "step": 39734 }, { "epoch": 2.1184091272591568, "grad_norm": 1.523413062095642, "learning_rate": 2.0532e-06, "loss": 0.0097, "step": 39735 }, { "epoch": 2.1184624406888095, "grad_norm": 2.0831425189971924, "learning_rate": 2.053e-06, "loss": 0.0122, "step": 39736 }, { "epoch": 2.1185157541184623, "grad_norm": 0.7433385848999023, "learning_rate": 2.0528e-06, "loss": 0.0046, "step": 39737 }, { "epoch": 2.1185690675481155, "grad_norm": 2.3499886989593506, "learning_rate": 2.0526e-06, "loss": 0.0254, "step": 39738 }, { "epoch": 2.1186223809777682, "grad_norm": 3.096163272857666, "learning_rate": 2.0524e-06, "loss": 0.0121, "step": 39739 }, { "epoch": 2.1186756944074214, "grad_norm": 0.5009805560112, "learning_rate": 2.0522000000000004e-06, "loss": 0.0035, "step": 39740 }, { "epoch": 2.118729007837074, "grad_norm": 2.919185161590576, "learning_rate": 2.052e-06, "loss": 0.0179, "step": 39741 }, { "epoch": 2.118782321266727, "grad_norm": 3.0912668704986572, "learning_rate": 2.0518e-06, "loss": 0.022, "step": 39742 }, { "epoch": 2.11883563469638, "grad_norm": 2.128054141998291, "learning_rate": 2.0516000000000004e-06, "loss": 0.0177, "step": 39743 }, { "epoch": 2.118888948126033, "grad_norm": 1.8320540189743042, "learning_rate": 2.0514e-06, "loss": 0.0306, "step": 39744 }, { "epoch": 2.118942261555686, "grad_norm": 2.341696262359619, "learning_rate": 2.0512e-06, "loss": 0.0076, "step": 39745 }, { "epoch": 2.118995574985339, "grad_norm": 2.9858672618865967, "learning_rate": 2.0510000000000004e-06, "loss": 0.0126, "step": 39746 }, { "epoch": 2.1190488884149916, "grad_norm": 1.3448928594589233, "learning_rate": 2.0508000000000003e-06, "loss": 0.0287, "step": 39747 }, { "epoch": 2.119102201844645, "grad_norm": 2.0028419494628906, "learning_rate": 2.0506e-06, "loss": 0.0186, "step": 39748 }, { "epoch": 2.1191555152742976, "grad_norm": 1.000360369682312, "learning_rate": 2.0504000000000003e-06, "loss": 0.0154, "step": 39749 }, { "epoch": 2.1192088287039503, "grad_norm": 2.2050747871398926, "learning_rate": 2.0502000000000003e-06, "loss": 0.0247, "step": 39750 }, { "epoch": 2.1192621421336035, "grad_norm": 1.236518144607544, "learning_rate": 2.05e-06, "loss": 0.0066, "step": 39751 }, { "epoch": 2.1193154555632563, "grad_norm": 2.411846876144409, "learning_rate": 2.0498000000000003e-06, "loss": 0.0413, "step": 39752 }, { "epoch": 2.1193687689929095, "grad_norm": 1.27252995967865, "learning_rate": 2.0496000000000002e-06, "loss": 0.0061, "step": 39753 }, { "epoch": 2.1194220824225622, "grad_norm": 2.087787389755249, "learning_rate": 2.0494000000000002e-06, "loss": 0.0222, "step": 39754 }, { "epoch": 2.119475395852215, "grad_norm": 0.6276354193687439, "learning_rate": 2.0492000000000002e-06, "loss": 0.0047, "step": 39755 }, { "epoch": 2.119528709281868, "grad_norm": 2.9284090995788574, "learning_rate": 2.049e-06, "loss": 0.0124, "step": 39756 }, { "epoch": 2.119582022711521, "grad_norm": 1.3348912000656128, "learning_rate": 2.0488e-06, "loss": 0.0188, "step": 39757 }, { "epoch": 2.119635336141174, "grad_norm": 0.9234607219696045, "learning_rate": 2.0486e-06, "loss": 0.023, "step": 39758 }, { "epoch": 2.119688649570827, "grad_norm": 1.6489185094833374, "learning_rate": 2.0484e-06, "loss": 0.0282, "step": 39759 }, { "epoch": 2.1197419630004797, "grad_norm": 0.5210909247398376, "learning_rate": 2.0482e-06, "loss": 0.0032, "step": 39760 }, { "epoch": 2.119795276430133, "grad_norm": 1.680479884147644, "learning_rate": 2.048e-06, "loss": 0.0144, "step": 39761 }, { "epoch": 2.1198485898597856, "grad_norm": 0.40143629908561707, "learning_rate": 2.0478e-06, "loss": 0.003, "step": 39762 }, { "epoch": 2.1199019032894384, "grad_norm": 2.1841609477996826, "learning_rate": 2.0476e-06, "loss": 0.0103, "step": 39763 }, { "epoch": 2.1199552167190916, "grad_norm": 0.2747127413749695, "learning_rate": 2.0474e-06, "loss": 0.0028, "step": 39764 }, { "epoch": 2.1200085301487444, "grad_norm": 3.1764912605285645, "learning_rate": 2.0472000000000005e-06, "loss": 0.0283, "step": 39765 }, { "epoch": 2.1200618435783976, "grad_norm": 4.070827007293701, "learning_rate": 2.047e-06, "loss": 0.0285, "step": 39766 }, { "epoch": 2.1201151570080503, "grad_norm": 1.143523931503296, "learning_rate": 2.0468e-06, "loss": 0.0216, "step": 39767 }, { "epoch": 2.120168470437703, "grad_norm": 0.7578750848770142, "learning_rate": 2.0466000000000004e-06, "loss": 0.0124, "step": 39768 }, { "epoch": 2.1202217838673563, "grad_norm": 0.7922539114952087, "learning_rate": 2.0464e-06, "loss": 0.0046, "step": 39769 }, { "epoch": 2.120275097297009, "grad_norm": 1.2314000129699707, "learning_rate": 2.0462e-06, "loss": 0.0084, "step": 39770 }, { "epoch": 2.1203284107266622, "grad_norm": 1.6066747903823853, "learning_rate": 2.0460000000000004e-06, "loss": 0.0084, "step": 39771 }, { "epoch": 2.120381724156315, "grad_norm": 3.0017800331115723, "learning_rate": 2.0458000000000004e-06, "loss": 0.0098, "step": 39772 }, { "epoch": 2.1204350375859677, "grad_norm": 1.20035719871521, "learning_rate": 2.0456e-06, "loss": 0.0061, "step": 39773 }, { "epoch": 2.120488351015621, "grad_norm": 1.4554628133773804, "learning_rate": 2.0454000000000004e-06, "loss": 0.0129, "step": 39774 }, { "epoch": 2.1205416644452737, "grad_norm": 0.8584291338920593, "learning_rate": 2.0452000000000003e-06, "loss": 0.0065, "step": 39775 }, { "epoch": 2.120594977874927, "grad_norm": 2.2360990047454834, "learning_rate": 2.045e-06, "loss": 0.0152, "step": 39776 }, { "epoch": 2.1206482913045797, "grad_norm": 1.390960454940796, "learning_rate": 2.0448000000000003e-06, "loss": 0.0132, "step": 39777 }, { "epoch": 2.1207016047342324, "grad_norm": 2.2988693714141846, "learning_rate": 2.0446000000000003e-06, "loss": 0.0102, "step": 39778 }, { "epoch": 2.1207549181638856, "grad_norm": 1.902919888496399, "learning_rate": 2.0444000000000003e-06, "loss": 0.0363, "step": 39779 }, { "epoch": 2.1208082315935384, "grad_norm": 2.897200584411621, "learning_rate": 2.0442000000000003e-06, "loss": 0.0141, "step": 39780 }, { "epoch": 2.120861545023191, "grad_norm": 1.7973183393478394, "learning_rate": 2.0440000000000003e-06, "loss": 0.0112, "step": 39781 }, { "epoch": 2.1209148584528443, "grad_norm": 2.8673031330108643, "learning_rate": 2.0438000000000002e-06, "loss": 0.0264, "step": 39782 }, { "epoch": 2.120968171882497, "grad_norm": 1.3565335273742676, "learning_rate": 2.0436000000000002e-06, "loss": 0.0067, "step": 39783 }, { "epoch": 2.1210214853121503, "grad_norm": 0.47328394651412964, "learning_rate": 2.0434e-06, "loss": 0.0034, "step": 39784 }, { "epoch": 2.121074798741803, "grad_norm": 2.4037203788757324, "learning_rate": 2.0432e-06, "loss": 0.0171, "step": 39785 }, { "epoch": 2.121128112171456, "grad_norm": 0.44592124223709106, "learning_rate": 2.043e-06, "loss": 0.0034, "step": 39786 }, { "epoch": 2.121181425601109, "grad_norm": 2.1493163108825684, "learning_rate": 2.0428e-06, "loss": 0.0233, "step": 39787 }, { "epoch": 2.1212347390307618, "grad_norm": 3.0311334133148193, "learning_rate": 2.0426e-06, "loss": 0.0165, "step": 39788 }, { "epoch": 2.121288052460415, "grad_norm": 1.2400072813034058, "learning_rate": 2.0424e-06, "loss": 0.0064, "step": 39789 }, { "epoch": 2.1213413658900677, "grad_norm": 1.3009121417999268, "learning_rate": 2.0422e-06, "loss": 0.0067, "step": 39790 }, { "epoch": 2.1213946793197205, "grad_norm": 2.021165132522583, "learning_rate": 2.042e-06, "loss": 0.0104, "step": 39791 }, { "epoch": 2.1214479927493737, "grad_norm": 0.8739237189292908, "learning_rate": 2.0418e-06, "loss": 0.0056, "step": 39792 }, { "epoch": 2.1215013061790264, "grad_norm": 2.506718397140503, "learning_rate": 2.0416e-06, "loss": 0.0274, "step": 39793 }, { "epoch": 2.1215546196086796, "grad_norm": 1.9332541227340698, "learning_rate": 2.0414e-06, "loss": 0.0119, "step": 39794 }, { "epoch": 2.1216079330383324, "grad_norm": 2.7909035682678223, "learning_rate": 2.0412e-06, "loss": 0.0103, "step": 39795 }, { "epoch": 2.121661246467985, "grad_norm": 1.7444462776184082, "learning_rate": 2.041e-06, "loss": 0.0133, "step": 39796 }, { "epoch": 2.1217145598976384, "grad_norm": 1.3876487016677856, "learning_rate": 2.0408000000000004e-06, "loss": 0.0189, "step": 39797 }, { "epoch": 2.121767873327291, "grad_norm": 1.461654782295227, "learning_rate": 2.0406e-06, "loss": 0.0058, "step": 39798 }, { "epoch": 2.121821186756944, "grad_norm": 0.5099493861198425, "learning_rate": 2.0404e-06, "loss": 0.0038, "step": 39799 }, { "epoch": 2.121874500186597, "grad_norm": 0.9740104079246521, "learning_rate": 2.0402000000000004e-06, "loss": 0.0207, "step": 39800 }, { "epoch": 2.12192781361625, "grad_norm": 3.9681150913238525, "learning_rate": 2.04e-06, "loss": 0.0156, "step": 39801 }, { "epoch": 2.121981127045903, "grad_norm": 0.9716079831123352, "learning_rate": 2.0398e-06, "loss": 0.0042, "step": 39802 }, { "epoch": 2.122034440475556, "grad_norm": 1.018754005432129, "learning_rate": 2.0396000000000003e-06, "loss": 0.0115, "step": 39803 }, { "epoch": 2.1220877539052085, "grad_norm": 1.3043850660324097, "learning_rate": 2.0394000000000003e-06, "loss": 0.0047, "step": 39804 }, { "epoch": 2.1221410673348617, "grad_norm": 2.0611913204193115, "learning_rate": 2.0392e-06, "loss": 0.0359, "step": 39805 }, { "epoch": 2.1221943807645145, "grad_norm": 2.937513589859009, "learning_rate": 2.0390000000000003e-06, "loss": 0.0298, "step": 39806 }, { "epoch": 2.1222476941941677, "grad_norm": 7.1139750480651855, "learning_rate": 2.0388000000000003e-06, "loss": 0.0684, "step": 39807 }, { "epoch": 2.1223010076238205, "grad_norm": 1.1305875778198242, "learning_rate": 2.0386000000000003e-06, "loss": 0.0053, "step": 39808 }, { "epoch": 2.122354321053473, "grad_norm": 3.40766978263855, "learning_rate": 2.0384000000000003e-06, "loss": 0.0133, "step": 39809 }, { "epoch": 2.1224076344831264, "grad_norm": 2.6070499420166016, "learning_rate": 2.0382000000000002e-06, "loss": 0.0316, "step": 39810 }, { "epoch": 2.122460947912779, "grad_norm": 1.589826226234436, "learning_rate": 2.0380000000000002e-06, "loss": 0.0298, "step": 39811 }, { "epoch": 2.1225142613424324, "grad_norm": 1.7024667263031006, "learning_rate": 2.0378e-06, "loss": 0.0049, "step": 39812 }, { "epoch": 2.122567574772085, "grad_norm": 2.4888367652893066, "learning_rate": 2.0376e-06, "loss": 0.0387, "step": 39813 }, { "epoch": 2.122620888201738, "grad_norm": 3.100909948348999, "learning_rate": 2.0374e-06, "loss": 0.0159, "step": 39814 }, { "epoch": 2.122674201631391, "grad_norm": 1.0825027227401733, "learning_rate": 2.0372e-06, "loss": 0.0152, "step": 39815 }, { "epoch": 2.122727515061044, "grad_norm": 1.18077552318573, "learning_rate": 2.037e-06, "loss": 0.0055, "step": 39816 }, { "epoch": 2.1227808284906966, "grad_norm": 2.8323638439178467, "learning_rate": 2.0368e-06, "loss": 0.0131, "step": 39817 }, { "epoch": 2.12283414192035, "grad_norm": 0.9288493990898132, "learning_rate": 2.0366e-06, "loss": 0.0057, "step": 39818 }, { "epoch": 2.1228874553500026, "grad_norm": 2.763458251953125, "learning_rate": 2.0364e-06, "loss": 0.0114, "step": 39819 }, { "epoch": 2.1229407687796558, "grad_norm": 4.028753757476807, "learning_rate": 2.0362e-06, "loss": 0.0411, "step": 39820 }, { "epoch": 2.1229940822093085, "grad_norm": 3.285329818725586, "learning_rate": 2.036e-06, "loss": 0.0137, "step": 39821 }, { "epoch": 2.1230473956389613, "grad_norm": 1.6476517915725708, "learning_rate": 2.0358000000000005e-06, "loss": 0.0111, "step": 39822 }, { "epoch": 2.1231007090686145, "grad_norm": 1.8087457418441772, "learning_rate": 2.0356e-06, "loss": 0.0278, "step": 39823 }, { "epoch": 2.1231540224982672, "grad_norm": 1.843942642211914, "learning_rate": 2.0354e-06, "loss": 0.0157, "step": 39824 }, { "epoch": 2.1232073359279204, "grad_norm": 2.6422035694122314, "learning_rate": 2.0352000000000004e-06, "loss": 0.0364, "step": 39825 }, { "epoch": 2.123260649357573, "grad_norm": 0.5271517634391785, "learning_rate": 2.035e-06, "loss": 0.0033, "step": 39826 }, { "epoch": 2.123313962787226, "grad_norm": 2.367441415786743, "learning_rate": 2.0348e-06, "loss": 0.0586, "step": 39827 }, { "epoch": 2.123367276216879, "grad_norm": 3.385010242462158, "learning_rate": 2.0346000000000004e-06, "loss": 0.0553, "step": 39828 }, { "epoch": 2.123420589646532, "grad_norm": 1.065362811088562, "learning_rate": 2.0344000000000004e-06, "loss": 0.0209, "step": 39829 }, { "epoch": 2.123473903076185, "grad_norm": 3.94331955909729, "learning_rate": 2.0342e-06, "loss": 0.022, "step": 39830 }, { "epoch": 2.123527216505838, "grad_norm": 1.9211434125900269, "learning_rate": 2.0340000000000003e-06, "loss": 0.0085, "step": 39831 }, { "epoch": 2.1235805299354906, "grad_norm": 1.5868592262268066, "learning_rate": 2.0338000000000003e-06, "loss": 0.0238, "step": 39832 }, { "epoch": 2.123633843365144, "grad_norm": 0.47165241837501526, "learning_rate": 2.0336000000000003e-06, "loss": 0.0039, "step": 39833 }, { "epoch": 2.1236871567947966, "grad_norm": 2.2039544582366943, "learning_rate": 2.0334000000000003e-06, "loss": 0.0174, "step": 39834 }, { "epoch": 2.1237404702244493, "grad_norm": 0.8717697858810425, "learning_rate": 2.0332000000000003e-06, "loss": 0.011, "step": 39835 }, { "epoch": 2.1237937836541025, "grad_norm": 0.5601540207862854, "learning_rate": 2.0330000000000003e-06, "loss": 0.0038, "step": 39836 }, { "epoch": 2.1238470970837553, "grad_norm": 2.867027997970581, "learning_rate": 2.0328000000000003e-06, "loss": 0.046, "step": 39837 }, { "epoch": 2.1239004105134085, "grad_norm": 1.0890438556671143, "learning_rate": 2.0326000000000002e-06, "loss": 0.0062, "step": 39838 }, { "epoch": 2.1239537239430613, "grad_norm": 2.486207962036133, "learning_rate": 2.0324000000000002e-06, "loss": 0.0379, "step": 39839 }, { "epoch": 2.124007037372714, "grad_norm": 1.713868498802185, "learning_rate": 2.0322e-06, "loss": 0.0121, "step": 39840 }, { "epoch": 2.124060350802367, "grad_norm": 1.2009391784667969, "learning_rate": 2.032e-06, "loss": 0.0048, "step": 39841 }, { "epoch": 2.12411366423202, "grad_norm": 1.0397098064422607, "learning_rate": 2.0318e-06, "loss": 0.0062, "step": 39842 }, { "epoch": 2.124166977661673, "grad_norm": 2.113973379135132, "learning_rate": 2.0316e-06, "loss": 0.0068, "step": 39843 }, { "epoch": 2.124220291091326, "grad_norm": 0.9540079832077026, "learning_rate": 2.0314e-06, "loss": 0.02, "step": 39844 }, { "epoch": 2.1242736045209787, "grad_norm": 1.6060198545455933, "learning_rate": 2.0312e-06, "loss": 0.0166, "step": 39845 }, { "epoch": 2.124326917950632, "grad_norm": 0.6638191342353821, "learning_rate": 2.031e-06, "loss": 0.0144, "step": 39846 }, { "epoch": 2.1243802313802846, "grad_norm": 1.239250659942627, "learning_rate": 2.0308e-06, "loss": 0.0088, "step": 39847 }, { "epoch": 2.1244335448099374, "grad_norm": 2.553734302520752, "learning_rate": 2.0306e-06, "loss": 0.0271, "step": 39848 }, { "epoch": 2.1244868582395906, "grad_norm": 1.514175534248352, "learning_rate": 2.0304e-06, "loss": 0.0173, "step": 39849 }, { "epoch": 2.1245401716692434, "grad_norm": 1.4787328243255615, "learning_rate": 2.0302e-06, "loss": 0.0083, "step": 39850 }, { "epoch": 2.1245934850988966, "grad_norm": 1.9335159063339233, "learning_rate": 2.0300000000000005e-06, "loss": 0.0405, "step": 39851 }, { "epoch": 2.1246467985285493, "grad_norm": 2.0221261978149414, "learning_rate": 2.0298e-06, "loss": 0.0145, "step": 39852 }, { "epoch": 2.124700111958202, "grad_norm": 0.38887012004852295, "learning_rate": 2.0296e-06, "loss": 0.003, "step": 39853 }, { "epoch": 2.1247534253878553, "grad_norm": 0.6231024265289307, "learning_rate": 2.0294000000000004e-06, "loss": 0.0049, "step": 39854 }, { "epoch": 2.124806738817508, "grad_norm": 1.1626888513565063, "learning_rate": 2.0292e-06, "loss": 0.0179, "step": 39855 }, { "epoch": 2.1248600522471612, "grad_norm": 2.5386452674865723, "learning_rate": 2.029e-06, "loss": 0.0203, "step": 39856 }, { "epoch": 2.124913365676814, "grad_norm": 1.2457441091537476, "learning_rate": 2.0288000000000004e-06, "loss": 0.0057, "step": 39857 }, { "epoch": 2.1249666791064667, "grad_norm": 2.4214112758636475, "learning_rate": 2.0286000000000004e-06, "loss": 0.0222, "step": 39858 }, { "epoch": 2.12501999253612, "grad_norm": 1.2544159889221191, "learning_rate": 2.0284e-06, "loss": 0.0107, "step": 39859 }, { "epoch": 2.1250733059657727, "grad_norm": 1.9056340456008911, "learning_rate": 2.0282000000000003e-06, "loss": 0.0101, "step": 39860 }, { "epoch": 2.125126619395426, "grad_norm": 1.6702228784561157, "learning_rate": 2.0280000000000003e-06, "loss": 0.0173, "step": 39861 }, { "epoch": 2.1251799328250787, "grad_norm": 1.7106751203536987, "learning_rate": 2.0278e-06, "loss": 0.0264, "step": 39862 }, { "epoch": 2.1252332462547314, "grad_norm": 3.3412632942199707, "learning_rate": 2.0276000000000003e-06, "loss": 0.0145, "step": 39863 }, { "epoch": 2.1252865596843846, "grad_norm": 1.3450504541397095, "learning_rate": 2.0274000000000003e-06, "loss": 0.0105, "step": 39864 }, { "epoch": 2.1253398731140374, "grad_norm": 1.7099635601043701, "learning_rate": 2.0272000000000003e-06, "loss": 0.0122, "step": 39865 }, { "epoch": 2.1253931865436906, "grad_norm": 3.1520440578460693, "learning_rate": 2.0270000000000002e-06, "loss": 0.0104, "step": 39866 }, { "epoch": 2.1254464999733433, "grad_norm": 2.900559902191162, "learning_rate": 2.0268000000000002e-06, "loss": 0.0168, "step": 39867 }, { "epoch": 2.125499813402996, "grad_norm": 2.956359386444092, "learning_rate": 2.0266000000000002e-06, "loss": 0.0222, "step": 39868 }, { "epoch": 2.1255531268326493, "grad_norm": 0.5301321744918823, "learning_rate": 2.0264e-06, "loss": 0.0042, "step": 39869 }, { "epoch": 2.125606440262302, "grad_norm": 1.6291718482971191, "learning_rate": 2.0262e-06, "loss": 0.0256, "step": 39870 }, { "epoch": 2.125659753691955, "grad_norm": 1.2675567865371704, "learning_rate": 2.026e-06, "loss": 0.0086, "step": 39871 }, { "epoch": 2.125713067121608, "grad_norm": 2.3888039588928223, "learning_rate": 2.0258e-06, "loss": 0.0146, "step": 39872 }, { "epoch": 2.1257663805512608, "grad_norm": 2.185974359512329, "learning_rate": 2.0256e-06, "loss": 0.0149, "step": 39873 }, { "epoch": 2.125819693980914, "grad_norm": 1.9310996532440186, "learning_rate": 2.0254e-06, "loss": 0.0051, "step": 39874 }, { "epoch": 2.1258730074105667, "grad_norm": 2.05765962600708, "learning_rate": 2.0252e-06, "loss": 0.0162, "step": 39875 }, { "epoch": 2.1259263208402195, "grad_norm": 2.605459213256836, "learning_rate": 2.025e-06, "loss": 0.0179, "step": 39876 }, { "epoch": 2.1259796342698727, "grad_norm": 0.6703032851219177, "learning_rate": 2.0248e-06, "loss": 0.0044, "step": 39877 }, { "epoch": 2.1260329476995254, "grad_norm": 3.7636196613311768, "learning_rate": 2.0246e-06, "loss": 0.0617, "step": 39878 }, { "epoch": 2.1260862611291786, "grad_norm": 2.7315845489501953, "learning_rate": 2.0244e-06, "loss": 0.0127, "step": 39879 }, { "epoch": 2.1261395745588314, "grad_norm": 0.9045307040214539, "learning_rate": 2.0242e-06, "loss": 0.006, "step": 39880 }, { "epoch": 2.126192887988484, "grad_norm": 1.6426666975021362, "learning_rate": 2.024e-06, "loss": 0.0062, "step": 39881 }, { "epoch": 2.1262462014181374, "grad_norm": 2.97296404838562, "learning_rate": 2.0238e-06, "loss": 0.0271, "step": 39882 }, { "epoch": 2.12629951484779, "grad_norm": 0.7795112133026123, "learning_rate": 2.0236000000000004e-06, "loss": 0.0057, "step": 39883 }, { "epoch": 2.126352828277443, "grad_norm": 1.190691590309143, "learning_rate": 2.0234e-06, "loss": 0.0211, "step": 39884 }, { "epoch": 2.126406141707096, "grad_norm": 1.3176506757736206, "learning_rate": 2.0232e-06, "loss": 0.0067, "step": 39885 }, { "epoch": 2.126459455136749, "grad_norm": 1.9604883193969727, "learning_rate": 2.0230000000000004e-06, "loss": 0.0167, "step": 39886 }, { "epoch": 2.126512768566402, "grad_norm": 1.4181562662124634, "learning_rate": 2.0228e-06, "loss": 0.0082, "step": 39887 }, { "epoch": 2.126566081996055, "grad_norm": 0.536670446395874, "learning_rate": 2.0226e-06, "loss": 0.0037, "step": 39888 }, { "epoch": 2.1266193954257075, "grad_norm": 2.450207471847534, "learning_rate": 2.0224000000000003e-06, "loss": 0.033, "step": 39889 }, { "epoch": 2.1266727088553607, "grad_norm": 1.8384323120117188, "learning_rate": 2.0222000000000003e-06, "loss": 0.0088, "step": 39890 }, { "epoch": 2.1267260222850135, "grad_norm": 1.7436386346817017, "learning_rate": 2.022e-06, "loss": 0.0135, "step": 39891 }, { "epoch": 2.1267793357146667, "grad_norm": 1.6415278911590576, "learning_rate": 2.0218000000000003e-06, "loss": 0.0261, "step": 39892 }, { "epoch": 2.1268326491443195, "grad_norm": 2.3653781414031982, "learning_rate": 2.0216000000000003e-06, "loss": 0.0123, "step": 39893 }, { "epoch": 2.126885962573972, "grad_norm": 2.3550057411193848, "learning_rate": 2.0214e-06, "loss": 0.0089, "step": 39894 }, { "epoch": 2.1269392760036254, "grad_norm": 1.850205421447754, "learning_rate": 2.0212000000000002e-06, "loss": 0.0071, "step": 39895 }, { "epoch": 2.126992589433278, "grad_norm": 0.6228213906288147, "learning_rate": 2.0210000000000002e-06, "loss": 0.0048, "step": 39896 }, { "epoch": 2.127045902862931, "grad_norm": 1.6072369813919067, "learning_rate": 2.0208e-06, "loss": 0.0048, "step": 39897 }, { "epoch": 2.127099216292584, "grad_norm": 2.0337250232696533, "learning_rate": 2.0206e-06, "loss": 0.0173, "step": 39898 }, { "epoch": 2.127152529722237, "grad_norm": 2.492602586746216, "learning_rate": 2.0204e-06, "loss": 0.0237, "step": 39899 }, { "epoch": 2.12720584315189, "grad_norm": 7.004340171813965, "learning_rate": 2.0202e-06, "loss": 0.0371, "step": 39900 }, { "epoch": 2.127259156581543, "grad_norm": 2.7366867065429688, "learning_rate": 2.02e-06, "loss": 0.0139, "step": 39901 }, { "epoch": 2.127312470011196, "grad_norm": 0.964083194732666, "learning_rate": 2.0198e-06, "loss": 0.0154, "step": 39902 }, { "epoch": 2.127365783440849, "grad_norm": 1.346705436706543, "learning_rate": 2.0196e-06, "loss": 0.0064, "step": 39903 }, { "epoch": 2.1274190968705016, "grad_norm": 0.42534729838371277, "learning_rate": 2.0194e-06, "loss": 0.0029, "step": 39904 }, { "epoch": 2.1274724103001548, "grad_norm": 0.44921624660491943, "learning_rate": 2.0192e-06, "loss": 0.0028, "step": 39905 }, { "epoch": 2.1275257237298075, "grad_norm": 2.102107286453247, "learning_rate": 2.019e-06, "loss": 0.0311, "step": 39906 }, { "epoch": 2.1275790371594603, "grad_norm": 1.579546332359314, "learning_rate": 2.0188e-06, "loss": 0.0218, "step": 39907 }, { "epoch": 2.1276323505891135, "grad_norm": 1.7705833911895752, "learning_rate": 2.0186000000000005e-06, "loss": 0.0108, "step": 39908 }, { "epoch": 2.1276856640187662, "grad_norm": 1.815023422241211, "learning_rate": 2.0184e-06, "loss": 0.0305, "step": 39909 }, { "epoch": 2.1277389774484194, "grad_norm": 2.120638370513916, "learning_rate": 2.0182e-06, "loss": 0.0184, "step": 39910 }, { "epoch": 2.127792290878072, "grad_norm": 1.3418959379196167, "learning_rate": 2.0180000000000004e-06, "loss": 0.0059, "step": 39911 }, { "epoch": 2.127845604307725, "grad_norm": 0.7033968567848206, "learning_rate": 2.0178e-06, "loss": 0.0036, "step": 39912 }, { "epoch": 2.127898917737378, "grad_norm": 2.2149951457977295, "learning_rate": 2.0176e-06, "loss": 0.0166, "step": 39913 }, { "epoch": 2.127952231167031, "grad_norm": 1.583870768547058, "learning_rate": 2.0174000000000004e-06, "loss": 0.0078, "step": 39914 }, { "epoch": 2.128005544596684, "grad_norm": 2.8800885677337646, "learning_rate": 2.0172000000000004e-06, "loss": 0.0388, "step": 39915 }, { "epoch": 2.128058858026337, "grad_norm": 2.0063796043395996, "learning_rate": 2.017e-06, "loss": 0.0173, "step": 39916 }, { "epoch": 2.1281121714559896, "grad_norm": 2.213977813720703, "learning_rate": 2.0168000000000003e-06, "loss": 0.0291, "step": 39917 }, { "epoch": 2.128165484885643, "grad_norm": 1.0115211009979248, "learning_rate": 2.0166000000000003e-06, "loss": 0.0043, "step": 39918 }, { "epoch": 2.1282187983152956, "grad_norm": 0.889341413974762, "learning_rate": 2.0164000000000003e-06, "loss": 0.0089, "step": 39919 }, { "epoch": 2.1282721117449483, "grad_norm": 2.2274959087371826, "learning_rate": 2.0162000000000003e-06, "loss": 0.0176, "step": 39920 }, { "epoch": 2.1283254251746015, "grad_norm": 1.5882831811904907, "learning_rate": 2.0160000000000003e-06, "loss": 0.0252, "step": 39921 }, { "epoch": 2.1283787386042543, "grad_norm": 7.538758754730225, "learning_rate": 2.0158000000000002e-06, "loss": 0.0149, "step": 39922 }, { "epoch": 2.1284320520339075, "grad_norm": 2.4579110145568848, "learning_rate": 2.0156000000000002e-06, "loss": 0.0146, "step": 39923 }, { "epoch": 2.1284853654635603, "grad_norm": 2.040642738342285, "learning_rate": 2.0154000000000002e-06, "loss": 0.0112, "step": 39924 }, { "epoch": 2.128538678893213, "grad_norm": 1.0083537101745605, "learning_rate": 2.0152e-06, "loss": 0.0139, "step": 39925 }, { "epoch": 2.1285919923228662, "grad_norm": 2.4602785110473633, "learning_rate": 2.015e-06, "loss": 0.0132, "step": 39926 }, { "epoch": 2.128645305752519, "grad_norm": 7.369668483734131, "learning_rate": 2.0148e-06, "loss": 0.0208, "step": 39927 }, { "epoch": 2.128698619182172, "grad_norm": 1.9149045944213867, "learning_rate": 2.0146e-06, "loss": 0.0317, "step": 39928 }, { "epoch": 2.128751932611825, "grad_norm": 2.2157444953918457, "learning_rate": 2.0144e-06, "loss": 0.0173, "step": 39929 }, { "epoch": 2.1288052460414777, "grad_norm": 2.4877450466156006, "learning_rate": 2.0142e-06, "loss": 0.0054, "step": 39930 }, { "epoch": 2.128858559471131, "grad_norm": 1.2567541599273682, "learning_rate": 2.014e-06, "loss": 0.0051, "step": 39931 }, { "epoch": 2.1289118729007837, "grad_norm": 1.3721753358840942, "learning_rate": 2.0138e-06, "loss": 0.0288, "step": 39932 }, { "epoch": 2.1289651863304364, "grad_norm": 1.839057445526123, "learning_rate": 2.0136e-06, "loss": 0.0284, "step": 39933 }, { "epoch": 2.1290184997600896, "grad_norm": 1.2617933750152588, "learning_rate": 2.0134e-06, "loss": 0.0046, "step": 39934 }, { "epoch": 2.1290718131897424, "grad_norm": 1.1464654207229614, "learning_rate": 2.0132e-06, "loss": 0.0056, "step": 39935 }, { "epoch": 2.1291251266193956, "grad_norm": 3.270763635635376, "learning_rate": 2.013e-06, "loss": 0.0354, "step": 39936 }, { "epoch": 2.1291784400490483, "grad_norm": 1.7556477785110474, "learning_rate": 2.0128e-06, "loss": 0.0232, "step": 39937 }, { "epoch": 2.129231753478701, "grad_norm": 1.5887531042099, "learning_rate": 2.0126e-06, "loss": 0.0187, "step": 39938 }, { "epoch": 2.1292850669083543, "grad_norm": 0.7639947533607483, "learning_rate": 2.0124e-06, "loss": 0.0043, "step": 39939 }, { "epoch": 2.129338380338007, "grad_norm": 1.9177411794662476, "learning_rate": 2.0122000000000004e-06, "loss": 0.0375, "step": 39940 }, { "epoch": 2.1293916937676602, "grad_norm": 4.35740327835083, "learning_rate": 2.012e-06, "loss": 0.0599, "step": 39941 }, { "epoch": 2.129445007197313, "grad_norm": 2.9526543617248535, "learning_rate": 2.0118e-06, "loss": 0.0305, "step": 39942 }, { "epoch": 2.1294983206269658, "grad_norm": 0.5657024383544922, "learning_rate": 2.0116000000000004e-06, "loss": 0.003, "step": 39943 }, { "epoch": 2.129551634056619, "grad_norm": 2.421110153198242, "learning_rate": 2.0114000000000003e-06, "loss": 0.0257, "step": 39944 }, { "epoch": 2.1296049474862717, "grad_norm": 1.3382707834243774, "learning_rate": 2.0112e-06, "loss": 0.0142, "step": 39945 }, { "epoch": 2.129658260915925, "grad_norm": 2.9311375617980957, "learning_rate": 2.0110000000000003e-06, "loss": 0.0395, "step": 39946 }, { "epoch": 2.1297115743455777, "grad_norm": 2.5458152294158936, "learning_rate": 2.0108000000000003e-06, "loss": 0.0297, "step": 39947 }, { "epoch": 2.1297648877752304, "grad_norm": 1.8419510126113892, "learning_rate": 2.0106e-06, "loss": 0.0128, "step": 39948 }, { "epoch": 2.1298182012048836, "grad_norm": 2.47037410736084, "learning_rate": 2.0104000000000003e-06, "loss": 0.0106, "step": 39949 }, { "epoch": 2.1298715146345364, "grad_norm": 1.3813161849975586, "learning_rate": 2.0102000000000003e-06, "loss": 0.0062, "step": 39950 }, { "epoch": 2.1299248280641896, "grad_norm": 2.8797459602355957, "learning_rate": 2.0100000000000002e-06, "loss": 0.0102, "step": 39951 }, { "epoch": 2.1299781414938423, "grad_norm": 3.3124499320983887, "learning_rate": 2.0098000000000002e-06, "loss": 0.0126, "step": 39952 }, { "epoch": 2.130031454923495, "grad_norm": 0.7865492701530457, "learning_rate": 2.0096e-06, "loss": 0.0051, "step": 39953 }, { "epoch": 2.1300847683531483, "grad_norm": 2.5514354705810547, "learning_rate": 2.0094e-06, "loss": 0.0305, "step": 39954 }, { "epoch": 2.130138081782801, "grad_norm": 1.6934727430343628, "learning_rate": 2.0092e-06, "loss": 0.0372, "step": 39955 }, { "epoch": 2.130191395212454, "grad_norm": 2.9281105995178223, "learning_rate": 2.009e-06, "loss": 0.0136, "step": 39956 }, { "epoch": 2.130244708642107, "grad_norm": 4.27692985534668, "learning_rate": 2.0088e-06, "loss": 0.0168, "step": 39957 }, { "epoch": 2.13029802207176, "grad_norm": 2.0794599056243896, "learning_rate": 2.0086e-06, "loss": 0.031, "step": 39958 }, { "epoch": 2.130351335501413, "grad_norm": 0.813459038734436, "learning_rate": 2.0084e-06, "loss": 0.0088, "step": 39959 }, { "epoch": 2.1304046489310657, "grad_norm": 2.690154552459717, "learning_rate": 2.0082e-06, "loss": 0.0098, "step": 39960 }, { "epoch": 2.1304579623607185, "grad_norm": 0.9838732481002808, "learning_rate": 2.008e-06, "loss": 0.0063, "step": 39961 }, { "epoch": 2.1305112757903717, "grad_norm": 2.028332471847534, "learning_rate": 2.0078e-06, "loss": 0.0349, "step": 39962 }, { "epoch": 2.1305645892200245, "grad_norm": 2.101790189743042, "learning_rate": 2.0076e-06, "loss": 0.014, "step": 39963 }, { "epoch": 2.1306179026496777, "grad_norm": 3.5137779712677, "learning_rate": 2.0074e-06, "loss": 0.049, "step": 39964 }, { "epoch": 2.1306712160793304, "grad_norm": 0.9826393723487854, "learning_rate": 2.0072000000000005e-06, "loss": 0.0113, "step": 39965 }, { "epoch": 2.130724529508983, "grad_norm": 1.0033302307128906, "learning_rate": 2.007e-06, "loss": 0.0041, "step": 39966 }, { "epoch": 2.1307778429386364, "grad_norm": 1.2938002347946167, "learning_rate": 2.0068e-06, "loss": 0.0152, "step": 39967 }, { "epoch": 2.130831156368289, "grad_norm": 2.329342842102051, "learning_rate": 2.0066000000000004e-06, "loss": 0.0244, "step": 39968 }, { "epoch": 2.130884469797942, "grad_norm": 2.2621219158172607, "learning_rate": 2.0064000000000004e-06, "loss": 0.0183, "step": 39969 }, { "epoch": 2.130937783227595, "grad_norm": 1.1384611129760742, "learning_rate": 2.0062e-06, "loss": 0.0061, "step": 39970 }, { "epoch": 2.130991096657248, "grad_norm": 2.238301992416382, "learning_rate": 2.0060000000000004e-06, "loss": 0.0465, "step": 39971 }, { "epoch": 2.131044410086901, "grad_norm": 0.2101110816001892, "learning_rate": 2.0058000000000003e-06, "loss": 0.0017, "step": 39972 }, { "epoch": 2.131097723516554, "grad_norm": 1.2066664695739746, "learning_rate": 2.0056e-06, "loss": 0.0143, "step": 39973 }, { "epoch": 2.1311510369462066, "grad_norm": 3.029343366622925, "learning_rate": 2.0054000000000003e-06, "loss": 0.0152, "step": 39974 }, { "epoch": 2.1312043503758598, "grad_norm": 4.880726337432861, "learning_rate": 2.0052000000000003e-06, "loss": 0.0479, "step": 39975 }, { "epoch": 2.1312576638055125, "grad_norm": 3.3873648643493652, "learning_rate": 2.0050000000000003e-06, "loss": 0.0394, "step": 39976 }, { "epoch": 2.1313109772351657, "grad_norm": 3.412370443344116, "learning_rate": 2.0048000000000003e-06, "loss": 0.0328, "step": 39977 }, { "epoch": 2.1313642906648185, "grad_norm": 1.0033117532730103, "learning_rate": 2.0046000000000003e-06, "loss": 0.0161, "step": 39978 }, { "epoch": 2.1314176040944712, "grad_norm": 2.9721670150756836, "learning_rate": 2.0044000000000002e-06, "loss": 0.0214, "step": 39979 }, { "epoch": 2.1314709175241244, "grad_norm": 0.7864738702774048, "learning_rate": 2.0042000000000002e-06, "loss": 0.0042, "step": 39980 }, { "epoch": 2.131524230953777, "grad_norm": 1.1165789365768433, "learning_rate": 2.004e-06, "loss": 0.0131, "step": 39981 }, { "epoch": 2.1315775443834304, "grad_norm": 2.777207136154175, "learning_rate": 2.0038e-06, "loss": 0.048, "step": 39982 }, { "epoch": 2.131630857813083, "grad_norm": 1.561686396598816, "learning_rate": 2.0036e-06, "loss": 0.0079, "step": 39983 }, { "epoch": 2.131684171242736, "grad_norm": 1.872887134552002, "learning_rate": 2.0034e-06, "loss": 0.0109, "step": 39984 }, { "epoch": 2.131737484672389, "grad_norm": 0.8818327784538269, "learning_rate": 2.0032e-06, "loss": 0.0051, "step": 39985 }, { "epoch": 2.131790798102042, "grad_norm": 2.3486790657043457, "learning_rate": 2.003e-06, "loss": 0.01, "step": 39986 }, { "epoch": 2.131844111531695, "grad_norm": 2.0442397594451904, "learning_rate": 2.0028e-06, "loss": 0.0237, "step": 39987 }, { "epoch": 2.131897424961348, "grad_norm": 3.803142547607422, "learning_rate": 2.0026e-06, "loss": 0.0541, "step": 39988 }, { "epoch": 2.1319507383910006, "grad_norm": 0.6343413591384888, "learning_rate": 2.0024e-06, "loss": 0.0047, "step": 39989 }, { "epoch": 2.132004051820654, "grad_norm": 2.7595248222351074, "learning_rate": 2.0022e-06, "loss": 0.0137, "step": 39990 }, { "epoch": 2.1320573652503065, "grad_norm": 2.6754069328308105, "learning_rate": 2.002e-06, "loss": 0.0179, "step": 39991 }, { "epoch": 2.1321106786799593, "grad_norm": 5.542099475860596, "learning_rate": 2.0018e-06, "loss": 0.0165, "step": 39992 }, { "epoch": 2.1321639921096125, "grad_norm": 2.612057685852051, "learning_rate": 2.0016e-06, "loss": 0.01, "step": 39993 }, { "epoch": 2.1322173055392653, "grad_norm": 1.4440191984176636, "learning_rate": 2.0014000000000004e-06, "loss": 0.017, "step": 39994 }, { "epoch": 2.1322706189689185, "grad_norm": 6.388282299041748, "learning_rate": 2.0012e-06, "loss": 0.0451, "step": 39995 }, { "epoch": 2.132323932398571, "grad_norm": 1.6149249076843262, "learning_rate": 2.001e-06, "loss": 0.0152, "step": 39996 }, { "epoch": 2.132377245828224, "grad_norm": 1.7728254795074463, "learning_rate": 2.0008000000000004e-06, "loss": 0.0225, "step": 39997 }, { "epoch": 2.132430559257877, "grad_norm": 1.3644028902053833, "learning_rate": 2.0006e-06, "loss": 0.0066, "step": 39998 }, { "epoch": 2.13248387268753, "grad_norm": 1.146478533744812, "learning_rate": 2.0004e-06, "loss": 0.0141, "step": 39999 }, { "epoch": 2.132537186117183, "grad_norm": 0.7752593755722046, "learning_rate": 2.0002000000000003e-06, "loss": 0.0048, "step": 40000 }, { "epoch": 2.132590499546836, "grad_norm": 0.577925443649292, "learning_rate": 2.0000000000000003e-06, "loss": 0.005, "step": 40001 }, { "epoch": 2.1326438129764886, "grad_norm": 2.9547667503356934, "learning_rate": 1.9998e-06, "loss": 0.0051, "step": 40002 }, { "epoch": 2.132697126406142, "grad_norm": 1.3128674030303955, "learning_rate": 1.9996000000000003e-06, "loss": 0.0106, "step": 40003 }, { "epoch": 2.1327504398357946, "grad_norm": 2.199998378753662, "learning_rate": 1.9994000000000003e-06, "loss": 0.0083, "step": 40004 }, { "epoch": 2.1328037532654474, "grad_norm": 0.9480785727500916, "learning_rate": 1.9992e-06, "loss": 0.0067, "step": 40005 }, { "epoch": 2.1328570666951006, "grad_norm": 2.10797381401062, "learning_rate": 1.9990000000000003e-06, "loss": 0.0166, "step": 40006 }, { "epoch": 2.1329103801247533, "grad_norm": 1.6448062658309937, "learning_rate": 1.9988000000000002e-06, "loss": 0.0071, "step": 40007 }, { "epoch": 2.1329636935544065, "grad_norm": 2.4517529010772705, "learning_rate": 1.9986000000000002e-06, "loss": 0.0262, "step": 40008 }, { "epoch": 2.1330170069840593, "grad_norm": 1.2939677238464355, "learning_rate": 1.9984000000000002e-06, "loss": 0.0064, "step": 40009 }, { "epoch": 2.133070320413712, "grad_norm": 1.7216325998306274, "learning_rate": 1.9982e-06, "loss": 0.0068, "step": 40010 }, { "epoch": 2.1331236338433652, "grad_norm": 2.7553188800811768, "learning_rate": 1.998e-06, "loss": 0.0238, "step": 40011 }, { "epoch": 2.133176947273018, "grad_norm": 1.7894898653030396, "learning_rate": 1.9978e-06, "loss": 0.0165, "step": 40012 }, { "epoch": 2.133230260702671, "grad_norm": 3.1609103679656982, "learning_rate": 1.9976e-06, "loss": 0.0128, "step": 40013 }, { "epoch": 2.133283574132324, "grad_norm": 5.427053451538086, "learning_rate": 1.9974e-06, "loss": 0.0441, "step": 40014 }, { "epoch": 2.1333368875619767, "grad_norm": 0.7784215211868286, "learning_rate": 1.9972e-06, "loss": 0.0047, "step": 40015 }, { "epoch": 2.13339020099163, "grad_norm": 2.1850786209106445, "learning_rate": 1.997e-06, "loss": 0.0247, "step": 40016 }, { "epoch": 2.1334435144212827, "grad_norm": 3.443788766860962, "learning_rate": 1.9968e-06, "loss": 0.0561, "step": 40017 }, { "epoch": 2.1334968278509354, "grad_norm": 0.7250094413757324, "learning_rate": 1.9966e-06, "loss": 0.0048, "step": 40018 }, { "epoch": 2.1335501412805886, "grad_norm": 1.7992860078811646, "learning_rate": 1.9964000000000005e-06, "loss": 0.0099, "step": 40019 }, { "epoch": 2.1336034547102414, "grad_norm": 1.2553292512893677, "learning_rate": 1.9962e-06, "loss": 0.0105, "step": 40020 }, { "epoch": 2.1336567681398946, "grad_norm": 3.460536479949951, "learning_rate": 1.996e-06, "loss": 0.0479, "step": 40021 }, { "epoch": 2.1337100815695473, "grad_norm": 7.684599876403809, "learning_rate": 1.9958000000000004e-06, "loss": 0.0424, "step": 40022 }, { "epoch": 2.1337633949992, "grad_norm": 2.0111982822418213, "learning_rate": 1.9956e-06, "loss": 0.0184, "step": 40023 }, { "epoch": 2.1338167084288533, "grad_norm": 2.257079839706421, "learning_rate": 1.9954e-06, "loss": 0.0112, "step": 40024 }, { "epoch": 2.133870021858506, "grad_norm": 2.412986993789673, "learning_rate": 1.9952000000000004e-06, "loss": 0.0122, "step": 40025 }, { "epoch": 2.1339233352881593, "grad_norm": 0.3839367926120758, "learning_rate": 1.9950000000000004e-06, "loss": 0.0027, "step": 40026 }, { "epoch": 2.133976648717812, "grad_norm": 1.2915081977844238, "learning_rate": 1.9948e-06, "loss": 0.0136, "step": 40027 }, { "epoch": 2.1340299621474648, "grad_norm": 3.103205680847168, "learning_rate": 1.9946000000000004e-06, "loss": 0.0301, "step": 40028 }, { "epoch": 2.134083275577118, "grad_norm": 1.2338836193084717, "learning_rate": 1.9944000000000003e-06, "loss": 0.0075, "step": 40029 }, { "epoch": 2.1341365890067707, "grad_norm": 4.0556488037109375, "learning_rate": 1.9942e-06, "loss": 0.036, "step": 40030 }, { "epoch": 2.134189902436424, "grad_norm": 0.7577974200248718, "learning_rate": 1.9940000000000003e-06, "loss": 0.0059, "step": 40031 }, { "epoch": 2.1342432158660767, "grad_norm": 2.21541690826416, "learning_rate": 1.9938000000000003e-06, "loss": 0.0154, "step": 40032 }, { "epoch": 2.1342965292957294, "grad_norm": 1.6472262144088745, "learning_rate": 1.9936000000000003e-06, "loss": 0.0317, "step": 40033 }, { "epoch": 2.1343498427253826, "grad_norm": 2.219743490219116, "learning_rate": 1.9934000000000003e-06, "loss": 0.0137, "step": 40034 }, { "epoch": 2.1344031561550354, "grad_norm": 1.873724102973938, "learning_rate": 1.9932000000000002e-06, "loss": 0.0154, "step": 40035 }, { "epoch": 2.1344564695846886, "grad_norm": 1.2826400995254517, "learning_rate": 1.9930000000000002e-06, "loss": 0.0197, "step": 40036 }, { "epoch": 2.1345097830143414, "grad_norm": 1.1207677125930786, "learning_rate": 1.9928000000000002e-06, "loss": 0.0191, "step": 40037 }, { "epoch": 2.134563096443994, "grad_norm": 2.561897039413452, "learning_rate": 1.9926e-06, "loss": 0.0249, "step": 40038 }, { "epoch": 2.1346164098736473, "grad_norm": 1.0545088052749634, "learning_rate": 1.9924e-06, "loss": 0.0089, "step": 40039 }, { "epoch": 2.1346697233033, "grad_norm": 1.957793116569519, "learning_rate": 1.9922e-06, "loss": 0.0139, "step": 40040 }, { "epoch": 2.134723036732953, "grad_norm": 2.794663429260254, "learning_rate": 1.992e-06, "loss": 0.0363, "step": 40041 }, { "epoch": 2.134776350162606, "grad_norm": 3.0506796836853027, "learning_rate": 1.9918e-06, "loss": 0.0353, "step": 40042 }, { "epoch": 2.134829663592259, "grad_norm": 0.16342215240001678, "learning_rate": 1.9916e-06, "loss": 0.0017, "step": 40043 }, { "epoch": 2.134882977021912, "grad_norm": 1.5505728721618652, "learning_rate": 1.9914e-06, "loss": 0.0075, "step": 40044 }, { "epoch": 2.1349362904515647, "grad_norm": 1.072821855545044, "learning_rate": 1.9912e-06, "loss": 0.0124, "step": 40045 }, { "epoch": 2.1349896038812175, "grad_norm": 2.090380907058716, "learning_rate": 1.991e-06, "loss": 0.0072, "step": 40046 }, { "epoch": 2.1350429173108707, "grad_norm": 2.939000368118286, "learning_rate": 1.9908e-06, "loss": 0.0135, "step": 40047 }, { "epoch": 2.1350962307405235, "grad_norm": 0.690161943435669, "learning_rate": 1.9906e-06, "loss": 0.0038, "step": 40048 }, { "epoch": 2.1351495441701767, "grad_norm": 2.778026819229126, "learning_rate": 1.9904e-06, "loss": 0.0417, "step": 40049 }, { "epoch": 2.1352028575998294, "grad_norm": 2.805135726928711, "learning_rate": 1.9902e-06, "loss": 0.0309, "step": 40050 }, { "epoch": 2.135256171029482, "grad_norm": 1.850648045539856, "learning_rate": 1.9900000000000004e-06, "loss": 0.0179, "step": 40051 }, { "epoch": 2.1353094844591354, "grad_norm": 0.8429234027862549, "learning_rate": 1.9898e-06, "loss": 0.0082, "step": 40052 }, { "epoch": 2.135362797888788, "grad_norm": 1.609202265739441, "learning_rate": 1.9896e-06, "loss": 0.0078, "step": 40053 }, { "epoch": 2.135416111318441, "grad_norm": 1.745398998260498, "learning_rate": 1.9894000000000004e-06, "loss": 0.0087, "step": 40054 }, { "epoch": 2.135469424748094, "grad_norm": 2.9736006259918213, "learning_rate": 1.9892000000000004e-06, "loss": 0.0439, "step": 40055 }, { "epoch": 2.135522738177747, "grad_norm": 4.724802494049072, "learning_rate": 1.989e-06, "loss": 0.0484, "step": 40056 }, { "epoch": 2.1355760516074, "grad_norm": 1.736883521080017, "learning_rate": 1.9888000000000003e-06, "loss": 0.0275, "step": 40057 }, { "epoch": 2.135629365037053, "grad_norm": 3.988236427307129, "learning_rate": 1.9886000000000003e-06, "loss": 0.0307, "step": 40058 }, { "epoch": 2.1356826784667056, "grad_norm": 2.1326050758361816, "learning_rate": 1.9884e-06, "loss": 0.0098, "step": 40059 }, { "epoch": 2.1357359918963588, "grad_norm": 3.1244430541992188, "learning_rate": 1.9882000000000003e-06, "loss": 0.0342, "step": 40060 }, { "epoch": 2.1357893053260115, "grad_norm": 0.9913356304168701, "learning_rate": 1.9880000000000003e-06, "loss": 0.0059, "step": 40061 }, { "epoch": 2.1358426187556647, "grad_norm": 3.258791208267212, "learning_rate": 1.9878000000000003e-06, "loss": 0.024, "step": 40062 }, { "epoch": 2.1358959321853175, "grad_norm": 3.0556981563568115, "learning_rate": 1.9876000000000002e-06, "loss": 0.0146, "step": 40063 }, { "epoch": 2.1359492456149702, "grad_norm": 2.7068474292755127, "learning_rate": 1.9874000000000002e-06, "loss": 0.0257, "step": 40064 }, { "epoch": 2.1360025590446234, "grad_norm": 1.7523345947265625, "learning_rate": 1.9872000000000002e-06, "loss": 0.0143, "step": 40065 }, { "epoch": 2.136055872474276, "grad_norm": 1.9997283220291138, "learning_rate": 1.987e-06, "loss": 0.0118, "step": 40066 }, { "epoch": 2.1361091859039294, "grad_norm": 2.802938938140869, "learning_rate": 1.9868e-06, "loss": 0.0565, "step": 40067 }, { "epoch": 2.136162499333582, "grad_norm": 0.8813467621803284, "learning_rate": 1.9866e-06, "loss": 0.0192, "step": 40068 }, { "epoch": 2.136215812763235, "grad_norm": 0.7677504420280457, "learning_rate": 1.9864e-06, "loss": 0.0052, "step": 40069 }, { "epoch": 2.136269126192888, "grad_norm": 1.6219650506973267, "learning_rate": 1.9862e-06, "loss": 0.0235, "step": 40070 }, { "epoch": 2.136322439622541, "grad_norm": 1.520566463470459, "learning_rate": 1.986e-06, "loss": 0.0135, "step": 40071 }, { "epoch": 2.136375753052194, "grad_norm": 1.415183663368225, "learning_rate": 1.9858e-06, "loss": 0.0313, "step": 40072 }, { "epoch": 2.136429066481847, "grad_norm": 2.756312608718872, "learning_rate": 1.9856e-06, "loss": 0.0171, "step": 40073 }, { "epoch": 2.1364823799114996, "grad_norm": 2.604440689086914, "learning_rate": 1.9854e-06, "loss": 0.0095, "step": 40074 }, { "epoch": 2.136535693341153, "grad_norm": 3.1389553546905518, "learning_rate": 1.9852e-06, "loss": 0.049, "step": 40075 }, { "epoch": 2.1365890067708055, "grad_norm": 1.5858500003814697, "learning_rate": 1.985e-06, "loss": 0.0098, "step": 40076 }, { "epoch": 2.1366423202004583, "grad_norm": 2.0866782665252686, "learning_rate": 1.9848e-06, "loss": 0.0263, "step": 40077 }, { "epoch": 2.1366956336301115, "grad_norm": 0.6174424290657043, "learning_rate": 1.9846e-06, "loss": 0.0042, "step": 40078 }, { "epoch": 2.1367489470597643, "grad_norm": 1.8200163841247559, "learning_rate": 1.9844e-06, "loss": 0.0105, "step": 40079 }, { "epoch": 2.1368022604894175, "grad_norm": 0.9717234969139099, "learning_rate": 1.9842000000000004e-06, "loss": 0.0048, "step": 40080 }, { "epoch": 2.13685557391907, "grad_norm": 4.955150127410889, "learning_rate": 1.984e-06, "loss": 0.0151, "step": 40081 }, { "epoch": 2.136908887348723, "grad_norm": 3.0783088207244873, "learning_rate": 1.9838e-06, "loss": 0.0106, "step": 40082 }, { "epoch": 2.136962200778376, "grad_norm": 3.652181386947632, "learning_rate": 1.9836000000000004e-06, "loss": 0.0466, "step": 40083 }, { "epoch": 2.137015514208029, "grad_norm": 1.2300482988357544, "learning_rate": 1.9834e-06, "loss": 0.0183, "step": 40084 }, { "epoch": 2.137068827637682, "grad_norm": 0.553268551826477, "learning_rate": 1.9832e-06, "loss": 0.0039, "step": 40085 }, { "epoch": 2.137122141067335, "grad_norm": 2.3196299076080322, "learning_rate": 1.9830000000000003e-06, "loss": 0.0213, "step": 40086 }, { "epoch": 2.1371754544969876, "grad_norm": 1.3161120414733887, "learning_rate": 1.9828000000000003e-06, "loss": 0.0274, "step": 40087 }, { "epoch": 2.137228767926641, "grad_norm": 3.024841070175171, "learning_rate": 1.9826e-06, "loss": 0.0235, "step": 40088 }, { "epoch": 2.1372820813562936, "grad_norm": 1.190203309059143, "learning_rate": 1.9824000000000003e-06, "loss": 0.0234, "step": 40089 }, { "epoch": 2.1373353947859464, "grad_norm": 1.9357941150665283, "learning_rate": 1.9822000000000003e-06, "loss": 0.0173, "step": 40090 }, { "epoch": 2.1373887082155996, "grad_norm": 1.9651665687561035, "learning_rate": 1.982e-06, "loss": 0.0058, "step": 40091 }, { "epoch": 2.1374420216452523, "grad_norm": 0.5647801160812378, "learning_rate": 1.9818000000000002e-06, "loss": 0.0038, "step": 40092 }, { "epoch": 2.1374953350749055, "grad_norm": 2.407186508178711, "learning_rate": 1.9816000000000002e-06, "loss": 0.0197, "step": 40093 }, { "epoch": 2.1375486485045583, "grad_norm": 0.4418419301509857, "learning_rate": 1.9814e-06, "loss": 0.0038, "step": 40094 }, { "epoch": 2.137601961934211, "grad_norm": 1.1973581314086914, "learning_rate": 1.9812e-06, "loss": 0.0168, "step": 40095 }, { "epoch": 2.1376552753638642, "grad_norm": 1.3473817110061646, "learning_rate": 1.981e-06, "loss": 0.0257, "step": 40096 }, { "epoch": 2.137708588793517, "grad_norm": 2.380002737045288, "learning_rate": 1.9808e-06, "loss": 0.0373, "step": 40097 }, { "epoch": 2.13776190222317, "grad_norm": 0.7826101183891296, "learning_rate": 1.9806e-06, "loss": 0.0211, "step": 40098 }, { "epoch": 2.137815215652823, "grad_norm": 0.8147566318511963, "learning_rate": 1.9804e-06, "loss": 0.0048, "step": 40099 }, { "epoch": 2.1378685290824757, "grad_norm": 1.97555410861969, "learning_rate": 1.9802e-06, "loss": 0.0235, "step": 40100 }, { "epoch": 2.137921842512129, "grad_norm": 2.626727819442749, "learning_rate": 1.98e-06, "loss": 0.013, "step": 40101 }, { "epoch": 2.1379751559417817, "grad_norm": 1.7839287519454956, "learning_rate": 1.9798e-06, "loss": 0.0074, "step": 40102 }, { "epoch": 2.138028469371435, "grad_norm": 1.3598943948745728, "learning_rate": 1.9796e-06, "loss": 0.0096, "step": 40103 }, { "epoch": 2.1380817828010876, "grad_norm": 1.7376540899276733, "learning_rate": 1.9794e-06, "loss": 0.0102, "step": 40104 }, { "epoch": 2.1381350962307404, "grad_norm": 2.1580159664154053, "learning_rate": 1.9792000000000005e-06, "loss": 0.0217, "step": 40105 }, { "epoch": 2.1381884096603936, "grad_norm": 1.3030478954315186, "learning_rate": 1.979e-06, "loss": 0.0065, "step": 40106 }, { "epoch": 2.1382417230900463, "grad_norm": 1.3126858472824097, "learning_rate": 1.9788e-06, "loss": 0.0115, "step": 40107 }, { "epoch": 2.1382950365196995, "grad_norm": 1.9041355848312378, "learning_rate": 1.9786000000000004e-06, "loss": 0.0125, "step": 40108 }, { "epoch": 2.1383483499493523, "grad_norm": 0.7808413505554199, "learning_rate": 1.9784e-06, "loss": 0.0065, "step": 40109 }, { "epoch": 2.138401663379005, "grad_norm": 1.423986792564392, "learning_rate": 1.9782e-06, "loss": 0.0082, "step": 40110 }, { "epoch": 2.1384549768086583, "grad_norm": 0.2444084733724594, "learning_rate": 1.9780000000000004e-06, "loss": 0.0018, "step": 40111 }, { "epoch": 2.138508290238311, "grad_norm": 1.0583311319351196, "learning_rate": 1.9778000000000004e-06, "loss": 0.0193, "step": 40112 }, { "epoch": 2.1385616036679638, "grad_norm": 2.004896879196167, "learning_rate": 1.9776e-06, "loss": 0.038, "step": 40113 }, { "epoch": 2.138614917097617, "grad_norm": 0.32157084345817566, "learning_rate": 1.9774000000000003e-06, "loss": 0.0019, "step": 40114 }, { "epoch": 2.1386682305272697, "grad_norm": 1.9670937061309814, "learning_rate": 1.9772000000000003e-06, "loss": 0.0141, "step": 40115 }, { "epoch": 2.138721543956923, "grad_norm": 2.038752555847168, "learning_rate": 1.977e-06, "loss": 0.0126, "step": 40116 }, { "epoch": 2.1387748573865757, "grad_norm": 2.5365586280822754, "learning_rate": 1.9768000000000003e-06, "loss": 0.0205, "step": 40117 }, { "epoch": 2.1388281708162284, "grad_norm": 1.59890615940094, "learning_rate": 1.9766000000000003e-06, "loss": 0.0262, "step": 40118 }, { "epoch": 2.1388814842458816, "grad_norm": 1.3837361335754395, "learning_rate": 1.9764000000000003e-06, "loss": 0.0133, "step": 40119 }, { "epoch": 2.1389347976755344, "grad_norm": 3.643855094909668, "learning_rate": 1.9762000000000002e-06, "loss": 0.0176, "step": 40120 }, { "epoch": 2.1389881111051876, "grad_norm": 0.6685653328895569, "learning_rate": 1.9760000000000002e-06, "loss": 0.0039, "step": 40121 }, { "epoch": 2.1390414245348404, "grad_norm": 1.1920828819274902, "learning_rate": 1.9758e-06, "loss": 0.0133, "step": 40122 }, { "epoch": 2.139094737964493, "grad_norm": 1.5262017250061035, "learning_rate": 1.9756e-06, "loss": 0.0082, "step": 40123 }, { "epoch": 2.1391480513941463, "grad_norm": 1.796929121017456, "learning_rate": 1.9754e-06, "loss": 0.0358, "step": 40124 }, { "epoch": 2.139201364823799, "grad_norm": 1.8042798042297363, "learning_rate": 1.9752e-06, "loss": 0.0351, "step": 40125 }, { "epoch": 2.139254678253452, "grad_norm": 2.8004283905029297, "learning_rate": 1.975e-06, "loss": 0.0328, "step": 40126 }, { "epoch": 2.139307991683105, "grad_norm": 2.2643399238586426, "learning_rate": 1.9748e-06, "loss": 0.0276, "step": 40127 }, { "epoch": 2.139361305112758, "grad_norm": 2.682769536972046, "learning_rate": 1.9746e-06, "loss": 0.0443, "step": 40128 }, { "epoch": 2.139414618542411, "grad_norm": 2.3676981925964355, "learning_rate": 1.9744e-06, "loss": 0.0385, "step": 40129 }, { "epoch": 2.1394679319720638, "grad_norm": 1.3352431058883667, "learning_rate": 1.9742e-06, "loss": 0.0102, "step": 40130 }, { "epoch": 2.1395212454017165, "grad_norm": 1.0273057222366333, "learning_rate": 1.974e-06, "loss": 0.0068, "step": 40131 }, { "epoch": 2.1395745588313697, "grad_norm": 0.5419008135795593, "learning_rate": 1.9738e-06, "loss": 0.0045, "step": 40132 }, { "epoch": 2.1396278722610225, "grad_norm": 0.7810271978378296, "learning_rate": 1.9736e-06, "loss": 0.0065, "step": 40133 }, { "epoch": 2.1396811856906757, "grad_norm": 3.3488047122955322, "learning_rate": 1.9734e-06, "loss": 0.0138, "step": 40134 }, { "epoch": 2.1397344991203284, "grad_norm": 0.5529882907867432, "learning_rate": 1.9732e-06, "loss": 0.0046, "step": 40135 }, { "epoch": 2.139787812549981, "grad_norm": 1.0706439018249512, "learning_rate": 1.973e-06, "loss": 0.0054, "step": 40136 }, { "epoch": 2.1398411259796344, "grad_norm": 0.9593403339385986, "learning_rate": 1.9728000000000004e-06, "loss": 0.0036, "step": 40137 }, { "epoch": 2.139894439409287, "grad_norm": 1.0360239744186401, "learning_rate": 1.9726e-06, "loss": 0.0152, "step": 40138 }, { "epoch": 2.13994775283894, "grad_norm": 0.9845490455627441, "learning_rate": 1.9724e-06, "loss": 0.0041, "step": 40139 }, { "epoch": 2.140001066268593, "grad_norm": 2.149693250656128, "learning_rate": 1.9722000000000004e-06, "loss": 0.0168, "step": 40140 }, { "epoch": 2.140054379698246, "grad_norm": 1.1668429374694824, "learning_rate": 1.972e-06, "loss": 0.0077, "step": 40141 }, { "epoch": 2.140107693127899, "grad_norm": 2.1777944564819336, "learning_rate": 1.9718e-06, "loss": 0.0155, "step": 40142 }, { "epoch": 2.140161006557552, "grad_norm": 2.1630642414093018, "learning_rate": 1.9716000000000003e-06, "loss": 0.0092, "step": 40143 }, { "epoch": 2.1402143199872046, "grad_norm": 1.4276509284973145, "learning_rate": 1.9714000000000003e-06, "loss": 0.0201, "step": 40144 }, { "epoch": 2.1402676334168578, "grad_norm": 2.6717352867126465, "learning_rate": 1.9712e-06, "loss": 0.0315, "step": 40145 }, { "epoch": 2.1403209468465105, "grad_norm": 3.9620440006256104, "learning_rate": 1.9710000000000003e-06, "loss": 0.0279, "step": 40146 }, { "epoch": 2.1403742602761637, "grad_norm": 2.0323429107666016, "learning_rate": 1.9708000000000003e-06, "loss": 0.0333, "step": 40147 }, { "epoch": 2.1404275737058165, "grad_norm": 2.2963058948516846, "learning_rate": 1.9706000000000002e-06, "loss": 0.0162, "step": 40148 }, { "epoch": 2.1404808871354692, "grad_norm": 2.2182462215423584, "learning_rate": 1.9704000000000002e-06, "loss": 0.0157, "step": 40149 }, { "epoch": 2.1405342005651224, "grad_norm": 1.8618055582046509, "learning_rate": 1.9702e-06, "loss": 0.0089, "step": 40150 }, { "epoch": 2.140587513994775, "grad_norm": 1.22939133644104, "learning_rate": 1.97e-06, "loss": 0.0059, "step": 40151 }, { "epoch": 2.1406408274244284, "grad_norm": 4.0538787841796875, "learning_rate": 1.9698e-06, "loss": 0.03, "step": 40152 }, { "epoch": 2.140694140854081, "grad_norm": 2.6493794918060303, "learning_rate": 1.9696e-06, "loss": 0.0409, "step": 40153 }, { "epoch": 2.140747454283734, "grad_norm": 2.1107757091522217, "learning_rate": 1.9694e-06, "loss": 0.0249, "step": 40154 }, { "epoch": 2.140800767713387, "grad_norm": 1.6399626731872559, "learning_rate": 1.9692e-06, "loss": 0.0111, "step": 40155 }, { "epoch": 2.14085408114304, "grad_norm": 0.8280246257781982, "learning_rate": 1.969e-06, "loss": 0.0048, "step": 40156 }, { "epoch": 2.140907394572693, "grad_norm": 0.6875426769256592, "learning_rate": 1.9688e-06, "loss": 0.0045, "step": 40157 }, { "epoch": 2.140960708002346, "grad_norm": 1.726595163345337, "learning_rate": 1.9686e-06, "loss": 0.0104, "step": 40158 }, { "epoch": 2.1410140214319986, "grad_norm": 1.1545466184616089, "learning_rate": 1.9684e-06, "loss": 0.0109, "step": 40159 }, { "epoch": 2.141067334861652, "grad_norm": 4.224684715270996, "learning_rate": 1.9682e-06, "loss": 0.0207, "step": 40160 }, { "epoch": 2.1411206482913046, "grad_norm": 2.933568239212036, "learning_rate": 1.968e-06, "loss": 0.0116, "step": 40161 }, { "epoch": 2.1411739617209573, "grad_norm": 2.4497809410095215, "learning_rate": 1.9678000000000005e-06, "loss": 0.0262, "step": 40162 }, { "epoch": 2.1412272751506105, "grad_norm": 1.275977373123169, "learning_rate": 1.9676e-06, "loss": 0.0202, "step": 40163 }, { "epoch": 2.1412805885802633, "grad_norm": 2.320432424545288, "learning_rate": 1.9674e-06, "loss": 0.0245, "step": 40164 }, { "epoch": 2.1413339020099165, "grad_norm": 1.305731177330017, "learning_rate": 1.9672000000000004e-06, "loss": 0.0086, "step": 40165 }, { "epoch": 2.1413872154395692, "grad_norm": 1.0253825187683105, "learning_rate": 1.9670000000000004e-06, "loss": 0.0059, "step": 40166 }, { "epoch": 2.141440528869222, "grad_norm": 0.6931650042533875, "learning_rate": 1.9668e-06, "loss": 0.0062, "step": 40167 }, { "epoch": 2.141493842298875, "grad_norm": 2.8837873935699463, "learning_rate": 1.9666000000000004e-06, "loss": 0.0161, "step": 40168 }, { "epoch": 2.141547155728528, "grad_norm": 1.7287886142730713, "learning_rate": 1.9664000000000003e-06, "loss": 0.0239, "step": 40169 }, { "epoch": 2.141600469158181, "grad_norm": 2.8126208782196045, "learning_rate": 1.9662e-06, "loss": 0.0437, "step": 40170 }, { "epoch": 2.141653782587834, "grad_norm": 2.131453514099121, "learning_rate": 1.9660000000000003e-06, "loss": 0.0156, "step": 40171 }, { "epoch": 2.1417070960174867, "grad_norm": 1.2366559505462646, "learning_rate": 1.9658000000000003e-06, "loss": 0.022, "step": 40172 }, { "epoch": 2.14176040944714, "grad_norm": 2.2794299125671387, "learning_rate": 1.9656000000000003e-06, "loss": 0.0112, "step": 40173 }, { "epoch": 2.1418137228767926, "grad_norm": 0.7911899089813232, "learning_rate": 1.9654000000000003e-06, "loss": 0.004, "step": 40174 }, { "epoch": 2.1418670363064454, "grad_norm": 1.1608234643936157, "learning_rate": 1.9652000000000003e-06, "loss": 0.0142, "step": 40175 }, { "epoch": 2.1419203497360986, "grad_norm": 1.0318503379821777, "learning_rate": 1.9650000000000002e-06, "loss": 0.0161, "step": 40176 }, { "epoch": 2.1419736631657513, "grad_norm": 1.631406307220459, "learning_rate": 1.9648000000000002e-06, "loss": 0.0099, "step": 40177 }, { "epoch": 2.1420269765954045, "grad_norm": 3.283510684967041, "learning_rate": 1.9646000000000002e-06, "loss": 0.0136, "step": 40178 }, { "epoch": 2.1420802900250573, "grad_norm": 1.9127912521362305, "learning_rate": 1.9644e-06, "loss": 0.0103, "step": 40179 }, { "epoch": 2.14213360345471, "grad_norm": 1.5173571109771729, "learning_rate": 1.9642e-06, "loss": 0.0068, "step": 40180 }, { "epoch": 2.1421869168843632, "grad_norm": 4.554232597351074, "learning_rate": 1.964e-06, "loss": 0.0477, "step": 40181 }, { "epoch": 2.142240230314016, "grad_norm": 2.024526357650757, "learning_rate": 1.9638e-06, "loss": 0.0153, "step": 40182 }, { "epoch": 2.142293543743669, "grad_norm": 1.402023196220398, "learning_rate": 1.9636e-06, "loss": 0.0094, "step": 40183 }, { "epoch": 2.142346857173322, "grad_norm": 0.49068933725357056, "learning_rate": 1.9634e-06, "loss": 0.0037, "step": 40184 }, { "epoch": 2.1424001706029747, "grad_norm": 2.946765422821045, "learning_rate": 1.9632e-06, "loss": 0.027, "step": 40185 }, { "epoch": 2.142453484032628, "grad_norm": 0.4570797383785248, "learning_rate": 1.963e-06, "loss": 0.0028, "step": 40186 }, { "epoch": 2.1425067974622807, "grad_norm": 1.5883978605270386, "learning_rate": 1.9628e-06, "loss": 0.0277, "step": 40187 }, { "epoch": 2.142560110891934, "grad_norm": 1.470471978187561, "learning_rate": 1.9626e-06, "loss": 0.0155, "step": 40188 }, { "epoch": 2.1426134243215866, "grad_norm": 3.1613690853118896, "learning_rate": 1.9624e-06, "loss": 0.0293, "step": 40189 }, { "epoch": 2.1426667377512394, "grad_norm": 1.703860878944397, "learning_rate": 1.9622e-06, "loss": 0.0066, "step": 40190 }, { "epoch": 2.1427200511808926, "grad_norm": 1.3134583234786987, "learning_rate": 1.9620000000000004e-06, "loss": 0.0047, "step": 40191 }, { "epoch": 2.1427733646105454, "grad_norm": 4.185140609741211, "learning_rate": 1.9618e-06, "loss": 0.0387, "step": 40192 }, { "epoch": 2.1428266780401986, "grad_norm": 0.30128049850463867, "learning_rate": 1.9616e-06, "loss": 0.0021, "step": 40193 }, { "epoch": 2.1428799914698513, "grad_norm": 0.8649441599845886, "learning_rate": 1.9614000000000004e-06, "loss": 0.0038, "step": 40194 }, { "epoch": 2.142933304899504, "grad_norm": 1.6182602643966675, "learning_rate": 1.9612e-06, "loss": 0.0098, "step": 40195 }, { "epoch": 2.1429866183291573, "grad_norm": 0.4655195474624634, "learning_rate": 1.961e-06, "loss": 0.0034, "step": 40196 }, { "epoch": 2.14303993175881, "grad_norm": 2.107189416885376, "learning_rate": 1.9608000000000004e-06, "loss": 0.0246, "step": 40197 }, { "epoch": 2.143093245188463, "grad_norm": 0.5742480754852295, "learning_rate": 1.9606000000000003e-06, "loss": 0.0024, "step": 40198 }, { "epoch": 2.143146558618116, "grad_norm": 2.0500144958496094, "learning_rate": 1.9604e-06, "loss": 0.0129, "step": 40199 }, { "epoch": 2.1431998720477687, "grad_norm": 1.5277096033096313, "learning_rate": 1.9602000000000003e-06, "loss": 0.0133, "step": 40200 }, { "epoch": 2.143253185477422, "grad_norm": 1.792137622833252, "learning_rate": 1.9600000000000003e-06, "loss": 0.0085, "step": 40201 }, { "epoch": 2.1433064989070747, "grad_norm": 1.7395566701889038, "learning_rate": 1.9598e-06, "loss": 0.0053, "step": 40202 }, { "epoch": 2.1433598123367275, "grad_norm": 1.508637547492981, "learning_rate": 1.9596000000000003e-06, "loss": 0.0324, "step": 40203 }, { "epoch": 2.1434131257663807, "grad_norm": 3.1034324169158936, "learning_rate": 1.9594000000000002e-06, "loss": 0.0372, "step": 40204 }, { "epoch": 2.1434664391960334, "grad_norm": 1.2886921167373657, "learning_rate": 1.9592000000000002e-06, "loss": 0.015, "step": 40205 }, { "epoch": 2.1435197526256866, "grad_norm": 3.123805046081543, "learning_rate": 1.9590000000000002e-06, "loss": 0.0121, "step": 40206 }, { "epoch": 2.1435730660553394, "grad_norm": 2.9749438762664795, "learning_rate": 1.9588e-06, "loss": 0.0178, "step": 40207 }, { "epoch": 2.143626379484992, "grad_norm": 0.5250218510627747, "learning_rate": 1.9586e-06, "loss": 0.0024, "step": 40208 }, { "epoch": 2.1436796929146453, "grad_norm": 0.8150937557220459, "learning_rate": 1.9584e-06, "loss": 0.0034, "step": 40209 }, { "epoch": 2.143733006344298, "grad_norm": 1.4915701150894165, "learning_rate": 1.9582e-06, "loss": 0.0382, "step": 40210 }, { "epoch": 2.143786319773951, "grad_norm": 1.4201756715774536, "learning_rate": 1.958e-06, "loss": 0.0184, "step": 40211 }, { "epoch": 2.143839633203604, "grad_norm": 1.1245290040969849, "learning_rate": 1.9578e-06, "loss": 0.0168, "step": 40212 }, { "epoch": 2.143892946633257, "grad_norm": 1.3081169128417969, "learning_rate": 1.9576e-06, "loss": 0.0166, "step": 40213 }, { "epoch": 2.14394626006291, "grad_norm": 2.9543280601501465, "learning_rate": 1.9574e-06, "loss": 0.0545, "step": 40214 }, { "epoch": 2.1439995734925628, "grad_norm": 1.6917387247085571, "learning_rate": 1.9572e-06, "loss": 0.0143, "step": 40215 }, { "epoch": 2.1440528869222155, "grad_norm": 0.876976490020752, "learning_rate": 1.9570000000000005e-06, "loss": 0.0052, "step": 40216 }, { "epoch": 2.1441062003518687, "grad_norm": 2.878061532974243, "learning_rate": 1.9568e-06, "loss": 0.0342, "step": 40217 }, { "epoch": 2.1441595137815215, "grad_norm": 4.309149265289307, "learning_rate": 1.9566e-06, "loss": 0.0499, "step": 40218 }, { "epoch": 2.1442128272111747, "grad_norm": 2.7806315422058105, "learning_rate": 1.9564000000000004e-06, "loss": 0.0222, "step": 40219 }, { "epoch": 2.1442661406408274, "grad_norm": 1.1245076656341553, "learning_rate": 1.9562e-06, "loss": 0.0103, "step": 40220 }, { "epoch": 2.14431945407048, "grad_norm": 1.2536886930465698, "learning_rate": 1.956e-06, "loss": 0.0095, "step": 40221 }, { "epoch": 2.1443727675001334, "grad_norm": 1.8979952335357666, "learning_rate": 1.9558000000000004e-06, "loss": 0.025, "step": 40222 }, { "epoch": 2.144426080929786, "grad_norm": 0.8767828345298767, "learning_rate": 1.9556000000000004e-06, "loss": 0.0049, "step": 40223 }, { "epoch": 2.1444793943594394, "grad_norm": 1.5328688621520996, "learning_rate": 1.9554e-06, "loss": 0.0227, "step": 40224 }, { "epoch": 2.144532707789092, "grad_norm": 2.591608762741089, "learning_rate": 1.9552000000000004e-06, "loss": 0.0306, "step": 40225 }, { "epoch": 2.144586021218745, "grad_norm": 3.1871440410614014, "learning_rate": 1.9550000000000003e-06, "loss": 0.0548, "step": 40226 }, { "epoch": 2.144639334648398, "grad_norm": 2.3192429542541504, "learning_rate": 1.9548e-06, "loss": 0.0443, "step": 40227 }, { "epoch": 2.144692648078051, "grad_norm": 1.01776123046875, "learning_rate": 1.9546000000000003e-06, "loss": 0.0055, "step": 40228 }, { "epoch": 2.144745961507704, "grad_norm": 1.9258074760437012, "learning_rate": 1.9544000000000003e-06, "loss": 0.0261, "step": 40229 }, { "epoch": 2.144799274937357, "grad_norm": 1.2979872226715088, "learning_rate": 1.9542000000000003e-06, "loss": 0.0133, "step": 40230 }, { "epoch": 2.1448525883670095, "grad_norm": 1.8210662603378296, "learning_rate": 1.9540000000000003e-06, "loss": 0.0384, "step": 40231 }, { "epoch": 2.1449059017966627, "grad_norm": 0.5808349847793579, "learning_rate": 1.9538000000000003e-06, "loss": 0.0044, "step": 40232 }, { "epoch": 2.1449592152263155, "grad_norm": 2.865511894226074, "learning_rate": 1.9536000000000002e-06, "loss": 0.0157, "step": 40233 }, { "epoch": 2.1450125286559683, "grad_norm": 2.227393865585327, "learning_rate": 1.9534000000000002e-06, "loss": 0.0322, "step": 40234 }, { "epoch": 2.1450658420856215, "grad_norm": 0.6898784041404724, "learning_rate": 1.9532e-06, "loss": 0.0048, "step": 40235 }, { "epoch": 2.145119155515274, "grad_norm": 4.95951509475708, "learning_rate": 1.953e-06, "loss": 0.0369, "step": 40236 }, { "epoch": 2.1451724689449274, "grad_norm": 2.450284242630005, "learning_rate": 1.9528e-06, "loss": 0.0155, "step": 40237 }, { "epoch": 2.14522578237458, "grad_norm": 2.5240347385406494, "learning_rate": 1.9526e-06, "loss": 0.0096, "step": 40238 }, { "epoch": 2.145279095804233, "grad_norm": 1.773432731628418, "learning_rate": 1.9524e-06, "loss": 0.0106, "step": 40239 }, { "epoch": 2.145332409233886, "grad_norm": 2.8043322563171387, "learning_rate": 1.9522e-06, "loss": 0.0202, "step": 40240 }, { "epoch": 2.145385722663539, "grad_norm": 2.4466731548309326, "learning_rate": 1.952e-06, "loss": 0.0213, "step": 40241 }, { "epoch": 2.145439036093192, "grad_norm": 2.560239315032959, "learning_rate": 1.9518e-06, "loss": 0.0134, "step": 40242 }, { "epoch": 2.145492349522845, "grad_norm": 1.2971532344818115, "learning_rate": 1.9516e-06, "loss": 0.0092, "step": 40243 }, { "epoch": 2.1455456629524976, "grad_norm": 3.6151740550994873, "learning_rate": 1.9514e-06, "loss": 0.023, "step": 40244 }, { "epoch": 2.145598976382151, "grad_norm": 3.077108383178711, "learning_rate": 1.9512e-06, "loss": 0.013, "step": 40245 }, { "epoch": 2.1456522898118036, "grad_norm": 2.4157817363739014, "learning_rate": 1.951e-06, "loss": 0.011, "step": 40246 }, { "epoch": 2.1457056032414563, "grad_norm": 3.620709180831909, "learning_rate": 1.9508e-06, "loss": 0.0126, "step": 40247 }, { "epoch": 2.1457589166711095, "grad_norm": 4.839817523956299, "learning_rate": 1.9506000000000004e-06, "loss": 0.0166, "step": 40248 }, { "epoch": 2.1458122301007623, "grad_norm": 1.13947594165802, "learning_rate": 1.9504e-06, "loss": 0.0161, "step": 40249 }, { "epoch": 2.1458655435304155, "grad_norm": 0.3826868534088135, "learning_rate": 1.9502e-06, "loss": 0.0033, "step": 40250 }, { "epoch": 2.1459188569600682, "grad_norm": 2.203265905380249, "learning_rate": 1.9500000000000004e-06, "loss": 0.0186, "step": 40251 }, { "epoch": 2.145972170389721, "grad_norm": 0.628868818283081, "learning_rate": 1.9498e-06, "loss": 0.0035, "step": 40252 }, { "epoch": 2.146025483819374, "grad_norm": 1.8289622068405151, "learning_rate": 1.9496e-06, "loss": 0.0147, "step": 40253 }, { "epoch": 2.146078797249027, "grad_norm": 1.0671995878219604, "learning_rate": 1.9494000000000003e-06, "loss": 0.006, "step": 40254 }, { "epoch": 2.14613211067868, "grad_norm": 3.197465419769287, "learning_rate": 1.9492000000000003e-06, "loss": 0.0495, "step": 40255 }, { "epoch": 2.146185424108333, "grad_norm": 3.2572131156921387, "learning_rate": 1.949e-06, "loss": 0.0318, "step": 40256 }, { "epoch": 2.1462387375379857, "grad_norm": 2.6372902393341064, "learning_rate": 1.9488000000000003e-06, "loss": 0.0313, "step": 40257 }, { "epoch": 2.146292050967639, "grad_norm": 1.9834908246994019, "learning_rate": 1.9486000000000003e-06, "loss": 0.0131, "step": 40258 }, { "epoch": 2.1463453643972916, "grad_norm": 0.9536247253417969, "learning_rate": 1.9484000000000003e-06, "loss": 0.0173, "step": 40259 }, { "epoch": 2.1463986778269444, "grad_norm": 1.8229925632476807, "learning_rate": 1.9482000000000003e-06, "loss": 0.0309, "step": 40260 }, { "epoch": 2.1464519912565976, "grad_norm": 2.045806884765625, "learning_rate": 1.9480000000000002e-06, "loss": 0.0269, "step": 40261 }, { "epoch": 2.1465053046862503, "grad_norm": 1.9486747980117798, "learning_rate": 1.9478000000000002e-06, "loss": 0.0199, "step": 40262 }, { "epoch": 2.1465586181159035, "grad_norm": 1.7115734815597534, "learning_rate": 1.9476e-06, "loss": 0.0068, "step": 40263 }, { "epoch": 2.1466119315455563, "grad_norm": 1.645031452178955, "learning_rate": 1.9474e-06, "loss": 0.0304, "step": 40264 }, { "epoch": 2.146665244975209, "grad_norm": 1.954023838043213, "learning_rate": 1.9472e-06, "loss": 0.0302, "step": 40265 }, { "epoch": 2.1467185584048623, "grad_norm": 1.812574863433838, "learning_rate": 1.947e-06, "loss": 0.0314, "step": 40266 }, { "epoch": 2.146771871834515, "grad_norm": 1.313823938369751, "learning_rate": 1.9468e-06, "loss": 0.0126, "step": 40267 }, { "epoch": 2.146825185264168, "grad_norm": 2.5308687686920166, "learning_rate": 1.9466e-06, "loss": 0.0207, "step": 40268 }, { "epoch": 2.146878498693821, "grad_norm": 0.9077953696250916, "learning_rate": 1.9464e-06, "loss": 0.0053, "step": 40269 }, { "epoch": 2.1469318121234737, "grad_norm": 0.2214956432580948, "learning_rate": 1.9462e-06, "loss": 0.0021, "step": 40270 }, { "epoch": 2.146985125553127, "grad_norm": 1.0380640029907227, "learning_rate": 1.946e-06, "loss": 0.0027, "step": 40271 }, { "epoch": 2.1470384389827797, "grad_norm": 2.0090689659118652, "learning_rate": 1.9458e-06, "loss": 0.0187, "step": 40272 }, { "epoch": 2.147091752412433, "grad_norm": 2.1018640995025635, "learning_rate": 1.9456e-06, "loss": 0.0106, "step": 40273 }, { "epoch": 2.1471450658420856, "grad_norm": 1.3193566799163818, "learning_rate": 1.9454e-06, "loss": 0.0078, "step": 40274 }, { "epoch": 2.1471983792717384, "grad_norm": 1.4942806959152222, "learning_rate": 1.9452e-06, "loss": 0.0107, "step": 40275 }, { "epoch": 2.1472516927013916, "grad_norm": 2.1890978813171387, "learning_rate": 1.945e-06, "loss": 0.0401, "step": 40276 }, { "epoch": 2.1473050061310444, "grad_norm": 1.2762243747711182, "learning_rate": 1.9448e-06, "loss": 0.0105, "step": 40277 }, { "epoch": 2.1473583195606976, "grad_norm": 1.2024645805358887, "learning_rate": 1.9446e-06, "loss": 0.0086, "step": 40278 }, { "epoch": 2.1474116329903503, "grad_norm": 1.905907154083252, "learning_rate": 1.9444e-06, "loss": 0.0155, "step": 40279 }, { "epoch": 2.147464946420003, "grad_norm": 1.2487530708312988, "learning_rate": 1.9442000000000004e-06, "loss": 0.0054, "step": 40280 }, { "epoch": 2.1475182598496563, "grad_norm": 1.4877212047576904, "learning_rate": 1.944e-06, "loss": 0.0258, "step": 40281 }, { "epoch": 2.147571573279309, "grad_norm": 1.4380828142166138, "learning_rate": 1.9438e-06, "loss": 0.0078, "step": 40282 }, { "epoch": 2.147624886708962, "grad_norm": 0.804117739200592, "learning_rate": 1.9436000000000003e-06, "loss": 0.0059, "step": 40283 }, { "epoch": 2.147678200138615, "grad_norm": 2.6285629272460938, "learning_rate": 1.9434000000000003e-06, "loss": 0.0203, "step": 40284 }, { "epoch": 2.1477315135682677, "grad_norm": 2.2202658653259277, "learning_rate": 1.9432e-06, "loss": 0.0282, "step": 40285 }, { "epoch": 2.147784826997921, "grad_norm": 0.26746439933776855, "learning_rate": 1.9430000000000003e-06, "loss": 0.0026, "step": 40286 }, { "epoch": 2.1478381404275737, "grad_norm": 1.771459937095642, "learning_rate": 1.9428000000000003e-06, "loss": 0.0228, "step": 40287 }, { "epoch": 2.1478914538572265, "grad_norm": 0.4873391389846802, "learning_rate": 1.9426e-06, "loss": 0.0037, "step": 40288 }, { "epoch": 2.1479447672868797, "grad_norm": 0.753482460975647, "learning_rate": 1.9424000000000002e-06, "loss": 0.0044, "step": 40289 }, { "epoch": 2.1479980807165324, "grad_norm": 2.2503674030303955, "learning_rate": 1.9422000000000002e-06, "loss": 0.0369, "step": 40290 }, { "epoch": 2.1480513941461856, "grad_norm": 3.754192352294922, "learning_rate": 1.942e-06, "loss": 0.0213, "step": 40291 }, { "epoch": 2.1481047075758384, "grad_norm": 1.5717886686325073, "learning_rate": 1.9418e-06, "loss": 0.0106, "step": 40292 }, { "epoch": 2.148158021005491, "grad_norm": 3.1124074459075928, "learning_rate": 1.9416e-06, "loss": 0.0335, "step": 40293 }, { "epoch": 2.1482113344351443, "grad_norm": 0.38975146412849426, "learning_rate": 1.9414e-06, "loss": 0.0034, "step": 40294 }, { "epoch": 2.148264647864797, "grad_norm": 2.634955644607544, "learning_rate": 1.9412e-06, "loss": 0.0168, "step": 40295 }, { "epoch": 2.14831796129445, "grad_norm": 2.3473901748657227, "learning_rate": 1.941e-06, "loss": 0.0143, "step": 40296 }, { "epoch": 2.148371274724103, "grad_norm": 1.7788606882095337, "learning_rate": 1.9408e-06, "loss": 0.0154, "step": 40297 }, { "epoch": 2.148424588153756, "grad_norm": 0.4111616611480713, "learning_rate": 1.9406e-06, "loss": 0.0038, "step": 40298 }, { "epoch": 2.148477901583409, "grad_norm": 1.8198952674865723, "learning_rate": 1.9404e-06, "loss": 0.0078, "step": 40299 }, { "epoch": 2.1485312150130618, "grad_norm": 1.5599948167800903, "learning_rate": 1.9402e-06, "loss": 0.0181, "step": 40300 }, { "epoch": 2.1485845284427145, "grad_norm": 2.204451322555542, "learning_rate": 1.94e-06, "loss": 0.0165, "step": 40301 }, { "epoch": 2.1486378418723677, "grad_norm": 2.155280828475952, "learning_rate": 1.9398000000000005e-06, "loss": 0.0121, "step": 40302 }, { "epoch": 2.1486911553020205, "grad_norm": 5.0513081550598145, "learning_rate": 1.9396e-06, "loss": 0.0262, "step": 40303 }, { "epoch": 2.1487444687316737, "grad_norm": 2.8825764656066895, "learning_rate": 1.9394e-06, "loss": 0.0346, "step": 40304 }, { "epoch": 2.1487977821613264, "grad_norm": 1.4101322889328003, "learning_rate": 1.9392000000000004e-06, "loss": 0.027, "step": 40305 }, { "epoch": 2.148851095590979, "grad_norm": 1.4832621812820435, "learning_rate": 1.939e-06, "loss": 0.0212, "step": 40306 }, { "epoch": 2.1489044090206324, "grad_norm": 1.1797236204147339, "learning_rate": 1.9388e-06, "loss": 0.0087, "step": 40307 }, { "epoch": 2.148957722450285, "grad_norm": 2.553056478500366, "learning_rate": 1.9386000000000004e-06, "loss": 0.0174, "step": 40308 }, { "epoch": 2.1490110358799384, "grad_norm": 0.3614031672477722, "learning_rate": 1.9384000000000004e-06, "loss": 0.0029, "step": 40309 }, { "epoch": 2.149064349309591, "grad_norm": 2.0981686115264893, "learning_rate": 1.9382e-06, "loss": 0.0179, "step": 40310 }, { "epoch": 2.149117662739244, "grad_norm": 1.7191689014434814, "learning_rate": 1.9380000000000003e-06, "loss": 0.0061, "step": 40311 }, { "epoch": 2.149170976168897, "grad_norm": 1.4350422620773315, "learning_rate": 1.9378000000000003e-06, "loss": 0.0047, "step": 40312 }, { "epoch": 2.14922428959855, "grad_norm": 3.1464149951934814, "learning_rate": 1.9376e-06, "loss": 0.0263, "step": 40313 }, { "epoch": 2.149277603028203, "grad_norm": 1.1760257482528687, "learning_rate": 1.9374000000000003e-06, "loss": 0.0181, "step": 40314 }, { "epoch": 2.149330916457856, "grad_norm": 2.267742395401001, "learning_rate": 1.9372000000000003e-06, "loss": 0.0272, "step": 40315 }, { "epoch": 2.1493842298875085, "grad_norm": 2.6285529136657715, "learning_rate": 1.9370000000000003e-06, "loss": 0.0143, "step": 40316 }, { "epoch": 2.1494375433171617, "grad_norm": 3.57287859916687, "learning_rate": 1.9368000000000002e-06, "loss": 0.0396, "step": 40317 }, { "epoch": 2.1494908567468145, "grad_norm": 1.1622194051742554, "learning_rate": 1.9366000000000002e-06, "loss": 0.0047, "step": 40318 }, { "epoch": 2.1495441701764673, "grad_norm": 4.263663291931152, "learning_rate": 1.9364000000000002e-06, "loss": 0.0336, "step": 40319 }, { "epoch": 2.1495974836061205, "grad_norm": 2.3239634037017822, "learning_rate": 1.9362e-06, "loss": 0.0157, "step": 40320 }, { "epoch": 2.149650797035773, "grad_norm": 0.9978475570678711, "learning_rate": 1.936e-06, "loss": 0.0045, "step": 40321 }, { "epoch": 2.1497041104654264, "grad_norm": 1.4367989301681519, "learning_rate": 1.9358e-06, "loss": 0.0103, "step": 40322 }, { "epoch": 2.149757423895079, "grad_norm": 2.769960880279541, "learning_rate": 1.9356e-06, "loss": 0.0271, "step": 40323 }, { "epoch": 2.149810737324732, "grad_norm": 0.41684120893478394, "learning_rate": 1.9354e-06, "loss": 0.0029, "step": 40324 }, { "epoch": 2.149864050754385, "grad_norm": 1.978963851928711, "learning_rate": 1.9352e-06, "loss": 0.008, "step": 40325 }, { "epoch": 2.149917364184038, "grad_norm": 1.55314302444458, "learning_rate": 1.935e-06, "loss": 0.0066, "step": 40326 }, { "epoch": 2.149970677613691, "grad_norm": 2.033240795135498, "learning_rate": 1.9348e-06, "loss": 0.0094, "step": 40327 }, { "epoch": 2.150023991043344, "grad_norm": 4.055741310119629, "learning_rate": 1.9346e-06, "loss": 0.0495, "step": 40328 }, { "epoch": 2.1500773044729966, "grad_norm": 2.2959563732147217, "learning_rate": 1.9344e-06, "loss": 0.0265, "step": 40329 }, { "epoch": 2.15013061790265, "grad_norm": 0.5474575757980347, "learning_rate": 1.9342e-06, "loss": 0.0033, "step": 40330 }, { "epoch": 2.1501839313323026, "grad_norm": 2.121295213699341, "learning_rate": 1.934e-06, "loss": 0.0107, "step": 40331 }, { "epoch": 2.1502372447619553, "grad_norm": 1.9926915168762207, "learning_rate": 1.9338e-06, "loss": 0.011, "step": 40332 }, { "epoch": 2.1502905581916085, "grad_norm": 2.9769442081451416, "learning_rate": 1.9336e-06, "loss": 0.0237, "step": 40333 }, { "epoch": 2.1503438716212613, "grad_norm": 1.0640987157821655, "learning_rate": 1.9334000000000004e-06, "loss": 0.0169, "step": 40334 }, { "epoch": 2.1503971850509145, "grad_norm": 1.8421932458877563, "learning_rate": 1.9332e-06, "loss": 0.0146, "step": 40335 }, { "epoch": 2.1504504984805672, "grad_norm": 0.388719767332077, "learning_rate": 1.933e-06, "loss": 0.0035, "step": 40336 }, { "epoch": 2.15050381191022, "grad_norm": 1.692324161529541, "learning_rate": 1.9328000000000004e-06, "loss": 0.0235, "step": 40337 }, { "epoch": 2.150557125339873, "grad_norm": 2.9214279651641846, "learning_rate": 1.9326e-06, "loss": 0.0227, "step": 40338 }, { "epoch": 2.150610438769526, "grad_norm": 1.5339523553848267, "learning_rate": 1.9324e-06, "loss": 0.0078, "step": 40339 }, { "epoch": 2.150663752199179, "grad_norm": 1.6393029689788818, "learning_rate": 1.9322000000000003e-06, "loss": 0.0052, "step": 40340 }, { "epoch": 2.150717065628832, "grad_norm": 1.4699403047561646, "learning_rate": 1.9320000000000003e-06, "loss": 0.006, "step": 40341 }, { "epoch": 2.1507703790584847, "grad_norm": 1.6556081771850586, "learning_rate": 1.9318e-06, "loss": 0.0085, "step": 40342 }, { "epoch": 2.150823692488138, "grad_norm": 0.21478600800037384, "learning_rate": 1.9316000000000003e-06, "loss": 0.0016, "step": 40343 }, { "epoch": 2.1508770059177906, "grad_norm": 1.4619735479354858, "learning_rate": 1.9314000000000003e-06, "loss": 0.0052, "step": 40344 }, { "epoch": 2.1509303193474434, "grad_norm": 3.4979913234710693, "learning_rate": 1.9312000000000002e-06, "loss": 0.0258, "step": 40345 }, { "epoch": 2.1509836327770966, "grad_norm": 7.631587505340576, "learning_rate": 1.9310000000000002e-06, "loss": 0.0135, "step": 40346 }, { "epoch": 2.1510369462067493, "grad_norm": 0.6923505067825317, "learning_rate": 1.9308000000000002e-06, "loss": 0.0033, "step": 40347 }, { "epoch": 2.1510902596364025, "grad_norm": 2.031815767288208, "learning_rate": 1.9306e-06, "loss": 0.0399, "step": 40348 }, { "epoch": 2.1511435730660553, "grad_norm": 1.3988683223724365, "learning_rate": 1.9304e-06, "loss": 0.0215, "step": 40349 }, { "epoch": 2.1511968864957085, "grad_norm": 3.9723379611968994, "learning_rate": 1.9302e-06, "loss": 0.0305, "step": 40350 }, { "epoch": 2.1512501999253613, "grad_norm": 2.9437332153320312, "learning_rate": 1.93e-06, "loss": 0.0228, "step": 40351 }, { "epoch": 2.151303513355014, "grad_norm": 0.615360677242279, "learning_rate": 1.9298e-06, "loss": 0.0045, "step": 40352 }, { "epoch": 2.151356826784667, "grad_norm": 0.9172725081443787, "learning_rate": 1.9296e-06, "loss": 0.0042, "step": 40353 }, { "epoch": 2.15141014021432, "grad_norm": 3.0439231395721436, "learning_rate": 1.9294e-06, "loss": 0.0449, "step": 40354 }, { "epoch": 2.1514634536439727, "grad_norm": 1.0282295942306519, "learning_rate": 1.9292e-06, "loss": 0.0033, "step": 40355 }, { "epoch": 2.151516767073626, "grad_norm": 1.9735116958618164, "learning_rate": 1.929e-06, "loss": 0.0148, "step": 40356 }, { "epoch": 2.1515700805032787, "grad_norm": 2.3371400833129883, "learning_rate": 1.9288e-06, "loss": 0.0271, "step": 40357 }, { "epoch": 2.151623393932932, "grad_norm": 1.1775473356246948, "learning_rate": 1.9286e-06, "loss": 0.0065, "step": 40358 }, { "epoch": 2.1516767073625847, "grad_norm": 2.5384583473205566, "learning_rate": 1.9284000000000005e-06, "loss": 0.0216, "step": 40359 }, { "epoch": 2.1517300207922374, "grad_norm": 0.5012519359588623, "learning_rate": 1.9282e-06, "loss": 0.0037, "step": 40360 }, { "epoch": 2.1517833342218906, "grad_norm": 2.064552068710327, "learning_rate": 1.928e-06, "loss": 0.0212, "step": 40361 }, { "epoch": 2.1518366476515434, "grad_norm": 1.8308184146881104, "learning_rate": 1.9278000000000004e-06, "loss": 0.0365, "step": 40362 }, { "epoch": 2.1518899610811966, "grad_norm": 2.588918924331665, "learning_rate": 1.9276e-06, "loss": 0.0411, "step": 40363 }, { "epoch": 2.1519432745108493, "grad_norm": 1.0797454118728638, "learning_rate": 1.9274e-06, "loss": 0.0197, "step": 40364 }, { "epoch": 2.151996587940502, "grad_norm": 2.167684316635132, "learning_rate": 1.9272000000000004e-06, "loss": 0.0064, "step": 40365 }, { "epoch": 2.1520499013701553, "grad_norm": 2.038456678390503, "learning_rate": 1.9270000000000004e-06, "loss": 0.0163, "step": 40366 }, { "epoch": 2.152103214799808, "grad_norm": 1.5470157861709595, "learning_rate": 1.9268e-06, "loss": 0.0246, "step": 40367 }, { "epoch": 2.152156528229461, "grad_norm": 1.0035799741744995, "learning_rate": 1.9266000000000003e-06, "loss": 0.0052, "step": 40368 }, { "epoch": 2.152209841659114, "grad_norm": 4.4753241539001465, "learning_rate": 1.9264000000000003e-06, "loss": 0.0294, "step": 40369 }, { "epoch": 2.1522631550887668, "grad_norm": 1.2346241474151611, "learning_rate": 1.9262000000000003e-06, "loss": 0.0074, "step": 40370 }, { "epoch": 2.15231646851842, "grad_norm": 2.1715643405914307, "learning_rate": 1.9260000000000003e-06, "loss": 0.0084, "step": 40371 }, { "epoch": 2.1523697819480727, "grad_norm": 1.4686148166656494, "learning_rate": 1.9258000000000003e-06, "loss": 0.0355, "step": 40372 }, { "epoch": 2.1524230953777255, "grad_norm": 1.4335216283798218, "learning_rate": 1.9256000000000002e-06, "loss": 0.0277, "step": 40373 }, { "epoch": 2.1524764088073787, "grad_norm": 3.0318946838378906, "learning_rate": 1.9254000000000002e-06, "loss": 0.0169, "step": 40374 }, { "epoch": 2.1525297222370314, "grad_norm": 1.5341086387634277, "learning_rate": 1.9252000000000002e-06, "loss": 0.015, "step": 40375 }, { "epoch": 2.1525830356666846, "grad_norm": 3.0439765453338623, "learning_rate": 1.925e-06, "loss": 0.0108, "step": 40376 }, { "epoch": 2.1526363490963374, "grad_norm": 1.7887009382247925, "learning_rate": 1.9248e-06, "loss": 0.0339, "step": 40377 }, { "epoch": 2.15268966252599, "grad_norm": 1.9293550252914429, "learning_rate": 1.9246e-06, "loss": 0.0221, "step": 40378 }, { "epoch": 2.1527429759556433, "grad_norm": 1.8297237157821655, "learning_rate": 1.9244e-06, "loss": 0.0122, "step": 40379 }, { "epoch": 2.152796289385296, "grad_norm": 3.8369505405426025, "learning_rate": 1.9242e-06, "loss": 0.0192, "step": 40380 }, { "epoch": 2.152849602814949, "grad_norm": 1.1102890968322754, "learning_rate": 1.924e-06, "loss": 0.0068, "step": 40381 }, { "epoch": 2.152902916244602, "grad_norm": 0.9722704291343689, "learning_rate": 1.9238e-06, "loss": 0.0056, "step": 40382 }, { "epoch": 2.152956229674255, "grad_norm": 4.158807277679443, "learning_rate": 1.9236e-06, "loss": 0.015, "step": 40383 }, { "epoch": 2.153009543103908, "grad_norm": 1.811377763748169, "learning_rate": 1.9234e-06, "loss": 0.0064, "step": 40384 }, { "epoch": 2.1530628565335608, "grad_norm": 1.9714295864105225, "learning_rate": 1.9232e-06, "loss": 0.0211, "step": 40385 }, { "epoch": 2.1531161699632135, "grad_norm": 0.8441112041473389, "learning_rate": 1.923e-06, "loss": 0.0049, "step": 40386 }, { "epoch": 2.1531694833928667, "grad_norm": 2.221315622329712, "learning_rate": 1.9228e-06, "loss": 0.0084, "step": 40387 }, { "epoch": 2.1532227968225195, "grad_norm": 3.671503782272339, "learning_rate": 1.9226e-06, "loss": 0.0275, "step": 40388 }, { "epoch": 2.1532761102521727, "grad_norm": 1.446800947189331, "learning_rate": 1.9224e-06, "loss": 0.0098, "step": 40389 }, { "epoch": 2.1533294236818254, "grad_norm": 1.614288091659546, "learning_rate": 1.9222e-06, "loss": 0.0139, "step": 40390 }, { "epoch": 2.153382737111478, "grad_norm": 1.5243507623672485, "learning_rate": 1.9220000000000004e-06, "loss": 0.0147, "step": 40391 }, { "epoch": 2.1534360505411314, "grad_norm": 2.827465534210205, "learning_rate": 1.9218e-06, "loss": 0.0299, "step": 40392 }, { "epoch": 2.153489363970784, "grad_norm": 1.9730174541473389, "learning_rate": 1.9216e-06, "loss": 0.0261, "step": 40393 }, { "epoch": 2.1535426774004374, "grad_norm": 1.8906522989273071, "learning_rate": 1.9214000000000004e-06, "loss": 0.0116, "step": 40394 }, { "epoch": 2.15359599083009, "grad_norm": 2.81305193901062, "learning_rate": 1.9212000000000003e-06, "loss": 0.014, "step": 40395 }, { "epoch": 2.153649304259743, "grad_norm": 0.7616772651672363, "learning_rate": 1.921e-06, "loss": 0.0026, "step": 40396 }, { "epoch": 2.153702617689396, "grad_norm": 2.2562825679779053, "learning_rate": 1.9208000000000003e-06, "loss": 0.0225, "step": 40397 }, { "epoch": 2.153755931119049, "grad_norm": 1.8249421119689941, "learning_rate": 1.9206000000000003e-06, "loss": 0.0123, "step": 40398 }, { "epoch": 2.153809244548702, "grad_norm": 2.684232234954834, "learning_rate": 1.9204e-06, "loss": 0.0402, "step": 40399 }, { "epoch": 2.153862557978355, "grad_norm": 1.7178800106048584, "learning_rate": 1.9202000000000003e-06, "loss": 0.0204, "step": 40400 }, { "epoch": 2.1539158714080076, "grad_norm": 1.255725383758545, "learning_rate": 1.9200000000000003e-06, "loss": 0.0148, "step": 40401 }, { "epoch": 2.1539691848376608, "grad_norm": 1.2313884496688843, "learning_rate": 1.9198000000000002e-06, "loss": 0.0037, "step": 40402 }, { "epoch": 2.1540224982673135, "grad_norm": 0.49091836810112, "learning_rate": 1.9196000000000002e-06, "loss": 0.0039, "step": 40403 }, { "epoch": 2.1540758116969663, "grad_norm": 1.4135279655456543, "learning_rate": 1.9194e-06, "loss": 0.0275, "step": 40404 }, { "epoch": 2.1541291251266195, "grad_norm": 1.7192976474761963, "learning_rate": 1.9192e-06, "loss": 0.024, "step": 40405 }, { "epoch": 2.1541824385562722, "grad_norm": 1.434640884399414, "learning_rate": 1.919e-06, "loss": 0.011, "step": 40406 }, { "epoch": 2.1542357519859254, "grad_norm": 3.2675888538360596, "learning_rate": 1.9188e-06, "loss": 0.0311, "step": 40407 }, { "epoch": 2.154289065415578, "grad_norm": 2.0267927646636963, "learning_rate": 1.9186e-06, "loss": 0.0178, "step": 40408 }, { "epoch": 2.154342378845231, "grad_norm": 4.19080924987793, "learning_rate": 1.9184e-06, "loss": 0.0325, "step": 40409 }, { "epoch": 2.154395692274884, "grad_norm": 4.041271209716797, "learning_rate": 1.9182e-06, "loss": 0.0402, "step": 40410 }, { "epoch": 2.154449005704537, "grad_norm": 4.692131996154785, "learning_rate": 1.918e-06, "loss": 0.0197, "step": 40411 }, { "epoch": 2.15450231913419, "grad_norm": 4.8867669105529785, "learning_rate": 1.9178e-06, "loss": 0.0394, "step": 40412 }, { "epoch": 2.154555632563843, "grad_norm": 2.2532663345336914, "learning_rate": 1.9176000000000005e-06, "loss": 0.028, "step": 40413 }, { "epoch": 2.1546089459934956, "grad_norm": 1.0981355905532837, "learning_rate": 1.9174e-06, "loss": 0.0051, "step": 40414 }, { "epoch": 2.154662259423149, "grad_norm": 3.1296286582946777, "learning_rate": 1.9172e-06, "loss": 0.0428, "step": 40415 }, { "epoch": 2.1547155728528016, "grad_norm": 1.033059000968933, "learning_rate": 1.9170000000000005e-06, "loss": 0.0061, "step": 40416 }, { "epoch": 2.1547688862824543, "grad_norm": 1.5248111486434937, "learning_rate": 1.9168e-06, "loss": 0.0113, "step": 40417 }, { "epoch": 2.1548221997121075, "grad_norm": 2.2056026458740234, "learning_rate": 1.9166e-06, "loss": 0.0237, "step": 40418 }, { "epoch": 2.1548755131417603, "grad_norm": 1.5197144746780396, "learning_rate": 1.9164000000000004e-06, "loss": 0.0294, "step": 40419 }, { "epoch": 2.1549288265714135, "grad_norm": 1.7563679218292236, "learning_rate": 1.9162000000000004e-06, "loss": 0.0061, "step": 40420 }, { "epoch": 2.1549821400010662, "grad_norm": 0.541877031326294, "learning_rate": 1.916e-06, "loss": 0.0049, "step": 40421 }, { "epoch": 2.155035453430719, "grad_norm": 3.0676417350769043, "learning_rate": 1.9158000000000004e-06, "loss": 0.0153, "step": 40422 }, { "epoch": 2.155088766860372, "grad_norm": 1.5568565130233765, "learning_rate": 1.9156000000000003e-06, "loss": 0.0085, "step": 40423 }, { "epoch": 2.155142080290025, "grad_norm": 2.4386837482452393, "learning_rate": 1.9154e-06, "loss": 0.0134, "step": 40424 }, { "epoch": 2.155195393719678, "grad_norm": 0.9487823247909546, "learning_rate": 1.9152000000000003e-06, "loss": 0.0059, "step": 40425 }, { "epoch": 2.155248707149331, "grad_norm": 2.370920419692993, "learning_rate": 1.9150000000000003e-06, "loss": 0.0349, "step": 40426 }, { "epoch": 2.1553020205789837, "grad_norm": 1.1293175220489502, "learning_rate": 1.9148000000000003e-06, "loss": 0.0049, "step": 40427 }, { "epoch": 2.155355334008637, "grad_norm": 2.8077924251556396, "learning_rate": 1.9146000000000003e-06, "loss": 0.0262, "step": 40428 }, { "epoch": 2.1554086474382896, "grad_norm": 1.5698531866073608, "learning_rate": 1.9144000000000003e-06, "loss": 0.0241, "step": 40429 }, { "epoch": 2.155461960867943, "grad_norm": 1.8041859865188599, "learning_rate": 1.9142000000000002e-06, "loss": 0.0101, "step": 40430 }, { "epoch": 2.1555152742975956, "grad_norm": 1.815682053565979, "learning_rate": 1.9140000000000002e-06, "loss": 0.018, "step": 40431 }, { "epoch": 2.1555685877272484, "grad_norm": 3.3060197830200195, "learning_rate": 1.9138e-06, "loss": 0.0233, "step": 40432 }, { "epoch": 2.1556219011569016, "grad_norm": 1.4235923290252686, "learning_rate": 1.9136e-06, "loss": 0.0098, "step": 40433 }, { "epoch": 2.1556752145865543, "grad_norm": 2.2183685302734375, "learning_rate": 1.9134e-06, "loss": 0.0159, "step": 40434 }, { "epoch": 2.1557285280162075, "grad_norm": 3.4976236820220947, "learning_rate": 1.9132e-06, "loss": 0.0209, "step": 40435 }, { "epoch": 2.1557818414458603, "grad_norm": 1.0607333183288574, "learning_rate": 1.913e-06, "loss": 0.0189, "step": 40436 }, { "epoch": 2.155835154875513, "grad_norm": 3.3800742626190186, "learning_rate": 1.9128e-06, "loss": 0.0285, "step": 40437 }, { "epoch": 2.1558884683051662, "grad_norm": 0.7315539717674255, "learning_rate": 1.9126e-06, "loss": 0.0035, "step": 40438 }, { "epoch": 2.155941781734819, "grad_norm": 1.1683207750320435, "learning_rate": 1.9124e-06, "loss": 0.0149, "step": 40439 }, { "epoch": 2.1559950951644717, "grad_norm": 1.5888079404830933, "learning_rate": 1.9122e-06, "loss": 0.0108, "step": 40440 }, { "epoch": 2.156048408594125, "grad_norm": 0.8709409832954407, "learning_rate": 1.912e-06, "loss": 0.0055, "step": 40441 }, { "epoch": 2.1561017220237777, "grad_norm": 2.0059430599212646, "learning_rate": 1.9118e-06, "loss": 0.0099, "step": 40442 }, { "epoch": 2.156155035453431, "grad_norm": 0.6825864911079407, "learning_rate": 1.9116e-06, "loss": 0.0047, "step": 40443 }, { "epoch": 2.1562083488830837, "grad_norm": 1.8695940971374512, "learning_rate": 1.9114e-06, "loss": 0.0292, "step": 40444 }, { "epoch": 2.1562616623127364, "grad_norm": 1.994273066520691, "learning_rate": 1.9112000000000004e-06, "loss": 0.0133, "step": 40445 }, { "epoch": 2.1563149757423896, "grad_norm": 1.9693171977996826, "learning_rate": 1.911e-06, "loss": 0.0063, "step": 40446 }, { "epoch": 2.1563682891720424, "grad_norm": 1.7735930681228638, "learning_rate": 1.9108e-06, "loss": 0.0201, "step": 40447 }, { "epoch": 2.1564216026016956, "grad_norm": 0.7480677366256714, "learning_rate": 1.9106000000000004e-06, "loss": 0.0267, "step": 40448 }, { "epoch": 2.1564749160313483, "grad_norm": 1.151269555091858, "learning_rate": 1.9104e-06, "loss": 0.0058, "step": 40449 }, { "epoch": 2.156528229461001, "grad_norm": 2.459278106689453, "learning_rate": 1.9102e-06, "loss": 0.0186, "step": 40450 }, { "epoch": 2.1565815428906543, "grad_norm": 1.9305458068847656, "learning_rate": 1.9100000000000003e-06, "loss": 0.04, "step": 40451 }, { "epoch": 2.156634856320307, "grad_norm": 2.545153856277466, "learning_rate": 1.9098000000000003e-06, "loss": 0.0087, "step": 40452 }, { "epoch": 2.15668816974996, "grad_norm": 1.827141523361206, "learning_rate": 1.9096e-06, "loss": 0.0119, "step": 40453 }, { "epoch": 2.156741483179613, "grad_norm": 2.6289610862731934, "learning_rate": 1.9094000000000003e-06, "loss": 0.0279, "step": 40454 }, { "epoch": 2.1567947966092658, "grad_norm": 3.3626716136932373, "learning_rate": 1.9092000000000003e-06, "loss": 0.0126, "step": 40455 }, { "epoch": 2.156848110038919, "grad_norm": 1.7013124227523804, "learning_rate": 1.909e-06, "loss": 0.0066, "step": 40456 }, { "epoch": 2.1569014234685717, "grad_norm": 2.0680017471313477, "learning_rate": 1.9088000000000003e-06, "loss": 0.0369, "step": 40457 }, { "epoch": 2.1569547368982245, "grad_norm": 1.1373735666275024, "learning_rate": 1.9086000000000002e-06, "loss": 0.0049, "step": 40458 }, { "epoch": 2.1570080503278777, "grad_norm": 2.024352550506592, "learning_rate": 1.9084000000000002e-06, "loss": 0.01, "step": 40459 }, { "epoch": 2.1570613637575304, "grad_norm": 1.2445311546325684, "learning_rate": 1.9082000000000002e-06, "loss": 0.01, "step": 40460 }, { "epoch": 2.1571146771871836, "grad_norm": 0.9951903820037842, "learning_rate": 1.908e-06, "loss": 0.0057, "step": 40461 }, { "epoch": 2.1571679906168364, "grad_norm": 1.4446943998336792, "learning_rate": 1.9078e-06, "loss": 0.0075, "step": 40462 }, { "epoch": 2.157221304046489, "grad_norm": 0.9333270788192749, "learning_rate": 1.9076e-06, "loss": 0.0033, "step": 40463 }, { "epoch": 2.1572746174761424, "grad_norm": 2.152873992919922, "learning_rate": 1.9074e-06, "loss": 0.0262, "step": 40464 }, { "epoch": 2.157327930905795, "grad_norm": 1.1810842752456665, "learning_rate": 1.9072000000000001e-06, "loss": 0.0123, "step": 40465 }, { "epoch": 2.157381244335448, "grad_norm": 1.3238170146942139, "learning_rate": 1.9070000000000003e-06, "loss": 0.0234, "step": 40466 }, { "epoch": 2.157434557765101, "grad_norm": 1.7561591863632202, "learning_rate": 1.9068000000000001e-06, "loss": 0.0278, "step": 40467 }, { "epoch": 2.157487871194754, "grad_norm": 2.6401665210723877, "learning_rate": 1.9066e-06, "loss": 0.0282, "step": 40468 }, { "epoch": 2.157541184624407, "grad_norm": 0.8923553824424744, "learning_rate": 1.9064000000000003e-06, "loss": 0.0055, "step": 40469 }, { "epoch": 2.15759449805406, "grad_norm": 1.029319405555725, "learning_rate": 1.9062000000000003e-06, "loss": 0.018, "step": 40470 }, { "epoch": 2.1576478114837125, "grad_norm": 3.11232852935791, "learning_rate": 1.906e-06, "loss": 0.0276, "step": 40471 }, { "epoch": 2.1577011249133657, "grad_norm": 1.597498893737793, "learning_rate": 1.9058000000000002e-06, "loss": 0.0182, "step": 40472 }, { "epoch": 2.1577544383430185, "grad_norm": 1.3175439834594727, "learning_rate": 1.9056000000000002e-06, "loss": 0.0162, "step": 40473 }, { "epoch": 2.1578077517726717, "grad_norm": 5.550133228302002, "learning_rate": 1.9054e-06, "loss": 0.0218, "step": 40474 }, { "epoch": 2.1578610652023245, "grad_norm": 1.8630499839782715, "learning_rate": 1.9052000000000002e-06, "loss": 0.0275, "step": 40475 }, { "epoch": 2.157914378631977, "grad_norm": 3.2913436889648438, "learning_rate": 1.9050000000000002e-06, "loss": 0.0313, "step": 40476 }, { "epoch": 2.1579676920616304, "grad_norm": 0.8056800961494446, "learning_rate": 1.9048000000000002e-06, "loss": 0.0051, "step": 40477 }, { "epoch": 2.158021005491283, "grad_norm": 1.487030267715454, "learning_rate": 1.9046000000000002e-06, "loss": 0.0156, "step": 40478 }, { "epoch": 2.1580743189209364, "grad_norm": 2.78621768951416, "learning_rate": 1.9044000000000001e-06, "loss": 0.0342, "step": 40479 }, { "epoch": 2.158127632350589, "grad_norm": 2.0810065269470215, "learning_rate": 1.9042000000000001e-06, "loss": 0.0155, "step": 40480 }, { "epoch": 2.158180945780242, "grad_norm": 1.1224873065948486, "learning_rate": 1.9040000000000003e-06, "loss": 0.0067, "step": 40481 }, { "epoch": 2.158234259209895, "grad_norm": 1.4135560989379883, "learning_rate": 1.9038e-06, "loss": 0.0051, "step": 40482 }, { "epoch": 2.158287572639548, "grad_norm": 1.7106986045837402, "learning_rate": 1.9036e-06, "loss": 0.0082, "step": 40483 }, { "epoch": 2.158340886069201, "grad_norm": 2.9889113903045654, "learning_rate": 1.9034000000000003e-06, "loss": 0.0225, "step": 40484 }, { "epoch": 2.158394199498854, "grad_norm": 5.946094036102295, "learning_rate": 1.9032e-06, "loss": 0.0163, "step": 40485 }, { "epoch": 2.1584475129285066, "grad_norm": 1.3329366445541382, "learning_rate": 1.903e-06, "loss": 0.0086, "step": 40486 }, { "epoch": 2.1585008263581598, "grad_norm": 4.07602596282959, "learning_rate": 1.9028000000000002e-06, "loss": 0.0155, "step": 40487 }, { "epoch": 2.1585541397878125, "grad_norm": 2.7845568656921387, "learning_rate": 1.9026000000000002e-06, "loss": 0.0295, "step": 40488 }, { "epoch": 2.1586074532174653, "grad_norm": 1.6158636808395386, "learning_rate": 1.9024e-06, "loss": 0.0184, "step": 40489 }, { "epoch": 2.1586607666471185, "grad_norm": 0.7456588745117188, "learning_rate": 1.9022000000000002e-06, "loss": 0.0044, "step": 40490 }, { "epoch": 2.1587140800767712, "grad_norm": 2.4106295108795166, "learning_rate": 1.9020000000000002e-06, "loss": 0.0323, "step": 40491 }, { "epoch": 2.1587673935064244, "grad_norm": 1.4544116258621216, "learning_rate": 1.9018e-06, "loss": 0.0072, "step": 40492 }, { "epoch": 2.158820706936077, "grad_norm": 0.6262178421020508, "learning_rate": 1.9016000000000001e-06, "loss": 0.0035, "step": 40493 }, { "epoch": 2.15887402036573, "grad_norm": 1.4637504816055298, "learning_rate": 1.9014000000000001e-06, "loss": 0.0204, "step": 40494 }, { "epoch": 2.158927333795383, "grad_norm": 1.9951525926589966, "learning_rate": 1.9012000000000003e-06, "loss": 0.0084, "step": 40495 }, { "epoch": 2.158980647225036, "grad_norm": 0.9338102340698242, "learning_rate": 1.901e-06, "loss": 0.0161, "step": 40496 }, { "epoch": 2.159033960654689, "grad_norm": 2.7225570678710938, "learning_rate": 1.9008e-06, "loss": 0.0174, "step": 40497 }, { "epoch": 2.159087274084342, "grad_norm": 2.4431607723236084, "learning_rate": 1.9006000000000003e-06, "loss": 0.0241, "step": 40498 }, { "epoch": 2.1591405875139946, "grad_norm": 1.565171241760254, "learning_rate": 1.9004e-06, "loss": 0.006, "step": 40499 }, { "epoch": 2.159193900943648, "grad_norm": 0.8082761168479919, "learning_rate": 1.9002e-06, "loss": 0.0163, "step": 40500 }, { "epoch": 2.1592472143733006, "grad_norm": 0.8224857449531555, "learning_rate": 1.9000000000000002e-06, "loss": 0.0165, "step": 40501 }, { "epoch": 2.1593005278029533, "grad_norm": 1.8531259298324585, "learning_rate": 1.8998000000000002e-06, "loss": 0.02, "step": 40502 }, { "epoch": 2.1593538412326065, "grad_norm": 3.124002695083618, "learning_rate": 1.8996e-06, "loss": 0.0313, "step": 40503 }, { "epoch": 2.1594071546622593, "grad_norm": 1.1910123825073242, "learning_rate": 1.8994000000000002e-06, "loss": 0.0071, "step": 40504 }, { "epoch": 2.1594604680919125, "grad_norm": 0.8245786428451538, "learning_rate": 1.8992000000000002e-06, "loss": 0.0187, "step": 40505 }, { "epoch": 2.1595137815215653, "grad_norm": 1.4130263328552246, "learning_rate": 1.8990000000000004e-06, "loss": 0.0149, "step": 40506 }, { "epoch": 2.159567094951218, "grad_norm": 0.9403741359710693, "learning_rate": 1.8988000000000001e-06, "loss": 0.0045, "step": 40507 }, { "epoch": 2.159620408380871, "grad_norm": 1.6136326789855957, "learning_rate": 1.8986000000000001e-06, "loss": 0.0305, "step": 40508 }, { "epoch": 2.159673721810524, "grad_norm": 2.455772876739502, "learning_rate": 1.8984000000000003e-06, "loss": 0.017, "step": 40509 }, { "epoch": 2.159727035240177, "grad_norm": 1.615566372871399, "learning_rate": 1.8982e-06, "loss": 0.0059, "step": 40510 }, { "epoch": 2.15978034866983, "grad_norm": 1.5583148002624512, "learning_rate": 1.898e-06, "loss": 0.0117, "step": 40511 }, { "epoch": 2.1598336620994827, "grad_norm": 1.859688639640808, "learning_rate": 1.8978000000000003e-06, "loss": 0.0086, "step": 40512 }, { "epoch": 2.159886975529136, "grad_norm": 2.0066559314727783, "learning_rate": 1.8976000000000003e-06, "loss": 0.0172, "step": 40513 }, { "epoch": 2.1599402889587886, "grad_norm": 2.4774880409240723, "learning_rate": 1.8974e-06, "loss": 0.0299, "step": 40514 }, { "epoch": 2.159993602388442, "grad_norm": 1.010831594467163, "learning_rate": 1.8972000000000002e-06, "loss": 0.0063, "step": 40515 }, { "epoch": 2.1600469158180946, "grad_norm": 3.4257237911224365, "learning_rate": 1.8970000000000002e-06, "loss": 0.0284, "step": 40516 }, { "epoch": 2.1601002292477474, "grad_norm": 0.79473876953125, "learning_rate": 1.8968e-06, "loss": 0.0161, "step": 40517 }, { "epoch": 2.1601535426774006, "grad_norm": 1.3047153949737549, "learning_rate": 1.8966000000000002e-06, "loss": 0.018, "step": 40518 }, { "epoch": 2.1602068561070533, "grad_norm": 2.9652490615844727, "learning_rate": 1.8964000000000002e-06, "loss": 0.0115, "step": 40519 }, { "epoch": 2.1602601695367065, "grad_norm": 3.479304313659668, "learning_rate": 1.8962000000000002e-06, "loss": 0.0162, "step": 40520 }, { "epoch": 2.1603134829663593, "grad_norm": 2.609926700592041, "learning_rate": 1.8960000000000001e-06, "loss": 0.0334, "step": 40521 }, { "epoch": 2.160366796396012, "grad_norm": 3.677929162979126, "learning_rate": 1.8958000000000001e-06, "loss": 0.0284, "step": 40522 }, { "epoch": 2.1604201098256652, "grad_norm": 1.435593843460083, "learning_rate": 1.8956000000000001e-06, "loss": 0.014, "step": 40523 }, { "epoch": 2.160473423255318, "grad_norm": 0.45259010791778564, "learning_rate": 1.8954e-06, "loss": 0.0043, "step": 40524 }, { "epoch": 2.1605267366849707, "grad_norm": 0.6283987164497375, "learning_rate": 1.8952e-06, "loss": 0.0041, "step": 40525 }, { "epoch": 2.160580050114624, "grad_norm": 1.7249361276626587, "learning_rate": 1.895e-06, "loss": 0.0109, "step": 40526 }, { "epoch": 2.1606333635442767, "grad_norm": 1.6009700298309326, "learning_rate": 1.8948000000000003e-06, "loss": 0.0088, "step": 40527 }, { "epoch": 2.16068667697393, "grad_norm": 1.9727457761764526, "learning_rate": 1.8946e-06, "loss": 0.0077, "step": 40528 }, { "epoch": 2.1607399904035827, "grad_norm": 1.1327359676361084, "learning_rate": 1.8944e-06, "loss": 0.0062, "step": 40529 }, { "epoch": 2.1607933038332354, "grad_norm": 1.6212795972824097, "learning_rate": 1.8942000000000002e-06, "loss": 0.0243, "step": 40530 }, { "epoch": 2.1608466172628886, "grad_norm": 2.349498987197876, "learning_rate": 1.8940000000000002e-06, "loss": 0.0175, "step": 40531 }, { "epoch": 2.1608999306925414, "grad_norm": 0.9767021536827087, "learning_rate": 1.8938e-06, "loss": 0.005, "step": 40532 }, { "epoch": 2.1609532441221946, "grad_norm": 1.599921703338623, "learning_rate": 1.8936000000000002e-06, "loss": 0.014, "step": 40533 }, { "epoch": 2.1610065575518473, "grad_norm": 4.658049583435059, "learning_rate": 1.8934000000000002e-06, "loss": 0.0413, "step": 40534 }, { "epoch": 2.1610598709815, "grad_norm": 5.415356159210205, "learning_rate": 1.8932e-06, "loss": 0.0273, "step": 40535 }, { "epoch": 2.1611131844111533, "grad_norm": 1.728122591972351, "learning_rate": 1.8930000000000001e-06, "loss": 0.0254, "step": 40536 }, { "epoch": 2.161166497840806, "grad_norm": 1.5204498767852783, "learning_rate": 1.8928000000000001e-06, "loss": 0.0184, "step": 40537 }, { "epoch": 2.161219811270459, "grad_norm": 2.2387969493865967, "learning_rate": 1.8926000000000003e-06, "loss": 0.0207, "step": 40538 }, { "epoch": 2.161273124700112, "grad_norm": 2.3371846675872803, "learning_rate": 1.8924e-06, "loss": 0.0096, "step": 40539 }, { "epoch": 2.1613264381297648, "grad_norm": 1.4257760047912598, "learning_rate": 1.8922e-06, "loss": 0.0123, "step": 40540 }, { "epoch": 2.161379751559418, "grad_norm": 1.4551416635513306, "learning_rate": 1.8920000000000003e-06, "loss": 0.0088, "step": 40541 }, { "epoch": 2.1614330649890707, "grad_norm": 1.6336827278137207, "learning_rate": 1.8918e-06, "loss": 0.023, "step": 40542 }, { "epoch": 2.1614863784187235, "grad_norm": 2.183957576751709, "learning_rate": 1.8916e-06, "loss": 0.0081, "step": 40543 }, { "epoch": 2.1615396918483767, "grad_norm": 2.347952365875244, "learning_rate": 1.8914000000000002e-06, "loss": 0.0159, "step": 40544 }, { "epoch": 2.1615930052780294, "grad_norm": 3.895906448364258, "learning_rate": 1.8912000000000002e-06, "loss": 0.0337, "step": 40545 }, { "epoch": 2.1616463187076826, "grad_norm": 1.7776867151260376, "learning_rate": 1.891e-06, "loss": 0.0197, "step": 40546 }, { "epoch": 2.1616996321373354, "grad_norm": 1.6118242740631104, "learning_rate": 1.8908000000000002e-06, "loss": 0.0191, "step": 40547 }, { "epoch": 2.161752945566988, "grad_norm": 2.375831365585327, "learning_rate": 1.8906000000000002e-06, "loss": 0.0162, "step": 40548 }, { "epoch": 2.1618062589966414, "grad_norm": 2.4267578125, "learning_rate": 1.8904000000000004e-06, "loss": 0.0578, "step": 40549 }, { "epoch": 2.161859572426294, "grad_norm": 0.8531606197357178, "learning_rate": 1.8902000000000001e-06, "loss": 0.0207, "step": 40550 }, { "epoch": 2.1619128858559473, "grad_norm": 0.7246048450469971, "learning_rate": 1.8900000000000001e-06, "loss": 0.0052, "step": 40551 }, { "epoch": 2.1619661992856, "grad_norm": 0.28690844774246216, "learning_rate": 1.8898000000000003e-06, "loss": 0.0019, "step": 40552 }, { "epoch": 2.162019512715253, "grad_norm": 2.2649073600769043, "learning_rate": 1.8896e-06, "loss": 0.0377, "step": 40553 }, { "epoch": 2.162072826144906, "grad_norm": 2.4413089752197266, "learning_rate": 1.8894e-06, "loss": 0.0189, "step": 40554 }, { "epoch": 2.162126139574559, "grad_norm": 1.8379169702529907, "learning_rate": 1.8892000000000003e-06, "loss": 0.0376, "step": 40555 }, { "epoch": 2.162179453004212, "grad_norm": 1.0343540906906128, "learning_rate": 1.8890000000000003e-06, "loss": 0.0139, "step": 40556 }, { "epoch": 2.1622327664338648, "grad_norm": 2.8153815269470215, "learning_rate": 1.8888e-06, "loss": 0.0153, "step": 40557 }, { "epoch": 2.1622860798635175, "grad_norm": 2.370490550994873, "learning_rate": 1.8886000000000002e-06, "loss": 0.054, "step": 40558 }, { "epoch": 2.1623393932931707, "grad_norm": 2.3166329860687256, "learning_rate": 1.8884000000000002e-06, "loss": 0.0139, "step": 40559 }, { "epoch": 2.1623927067228235, "grad_norm": 2.555893659591675, "learning_rate": 1.8882e-06, "loss": 0.0239, "step": 40560 }, { "epoch": 2.1624460201524762, "grad_norm": 0.995573103427887, "learning_rate": 1.8880000000000002e-06, "loss": 0.0147, "step": 40561 }, { "epoch": 2.1624993335821294, "grad_norm": 1.5017917156219482, "learning_rate": 1.8878000000000002e-06, "loss": 0.017, "step": 40562 }, { "epoch": 2.162552647011782, "grad_norm": 0.8679901361465454, "learning_rate": 1.8876000000000004e-06, "loss": 0.006, "step": 40563 }, { "epoch": 2.1626059604414354, "grad_norm": 1.8195186853408813, "learning_rate": 1.8874000000000001e-06, "loss": 0.0134, "step": 40564 }, { "epoch": 2.162659273871088, "grad_norm": 1.0380045175552368, "learning_rate": 1.8872000000000001e-06, "loss": 0.0241, "step": 40565 }, { "epoch": 2.162712587300741, "grad_norm": 1.7825344800949097, "learning_rate": 1.8870000000000003e-06, "loss": 0.0123, "step": 40566 }, { "epoch": 2.162765900730394, "grad_norm": 3.4275436401367188, "learning_rate": 1.8868e-06, "loss": 0.0207, "step": 40567 }, { "epoch": 2.162819214160047, "grad_norm": 2.1705565452575684, "learning_rate": 1.8866e-06, "loss": 0.0221, "step": 40568 }, { "epoch": 2.1628725275897, "grad_norm": 2.3221940994262695, "learning_rate": 1.8864000000000003e-06, "loss": 0.0084, "step": 40569 }, { "epoch": 2.162925841019353, "grad_norm": 2.4202816486358643, "learning_rate": 1.8862000000000003e-06, "loss": 0.0218, "step": 40570 }, { "epoch": 2.1629791544490056, "grad_norm": 1.0201841592788696, "learning_rate": 1.886e-06, "loss": 0.0058, "step": 40571 }, { "epoch": 2.1630324678786588, "grad_norm": 1.9146987199783325, "learning_rate": 1.8858000000000002e-06, "loss": 0.0083, "step": 40572 }, { "epoch": 2.1630857813083115, "grad_norm": 0.6446922421455383, "learning_rate": 1.8856000000000002e-06, "loss": 0.0048, "step": 40573 }, { "epoch": 2.1631390947379643, "grad_norm": 1.6977348327636719, "learning_rate": 1.8854000000000002e-06, "loss": 0.0154, "step": 40574 }, { "epoch": 2.1631924081676175, "grad_norm": 2.2293803691864014, "learning_rate": 1.8852000000000002e-06, "loss": 0.0256, "step": 40575 }, { "epoch": 2.1632457215972702, "grad_norm": 1.7178970575332642, "learning_rate": 1.8850000000000002e-06, "loss": 0.0247, "step": 40576 }, { "epoch": 2.1632990350269234, "grad_norm": 0.9673940539360046, "learning_rate": 1.8848000000000001e-06, "loss": 0.0054, "step": 40577 }, { "epoch": 2.163352348456576, "grad_norm": 0.6439740657806396, "learning_rate": 1.8846000000000001e-06, "loss": 0.0025, "step": 40578 }, { "epoch": 2.163405661886229, "grad_norm": 2.0732226371765137, "learning_rate": 1.8844000000000001e-06, "loss": 0.019, "step": 40579 }, { "epoch": 2.163458975315882, "grad_norm": 2.742215871810913, "learning_rate": 1.8842e-06, "loss": 0.0291, "step": 40580 }, { "epoch": 2.163512288745535, "grad_norm": 0.2832857370376587, "learning_rate": 1.8840000000000003e-06, "loss": 0.0028, "step": 40581 }, { "epoch": 2.163565602175188, "grad_norm": 1.933584451675415, "learning_rate": 1.8838e-06, "loss": 0.0195, "step": 40582 }, { "epoch": 2.163618915604841, "grad_norm": 1.5874168872833252, "learning_rate": 1.8836e-06, "loss": 0.0192, "step": 40583 }, { "epoch": 2.1636722290344936, "grad_norm": 0.17073021829128265, "learning_rate": 1.8834000000000003e-06, "loss": 0.002, "step": 40584 }, { "epoch": 2.163725542464147, "grad_norm": 1.208240270614624, "learning_rate": 1.8832e-06, "loss": 0.0117, "step": 40585 }, { "epoch": 2.1637788558937996, "grad_norm": 0.3742320239543915, "learning_rate": 1.883e-06, "loss": 0.0031, "step": 40586 }, { "epoch": 2.1638321693234523, "grad_norm": 2.461310386657715, "learning_rate": 1.8828000000000002e-06, "loss": 0.0264, "step": 40587 }, { "epoch": 2.1638854827531055, "grad_norm": 0.24816259741783142, "learning_rate": 1.8826000000000002e-06, "loss": 0.002, "step": 40588 }, { "epoch": 2.1639387961827583, "grad_norm": 1.9578286409378052, "learning_rate": 1.8824e-06, "loss": 0.0164, "step": 40589 }, { "epoch": 2.1639921096124115, "grad_norm": 2.019052505493164, "learning_rate": 1.8822000000000002e-06, "loss": 0.0258, "step": 40590 }, { "epoch": 2.1640454230420643, "grad_norm": 1.6477131843566895, "learning_rate": 1.8820000000000001e-06, "loss": 0.0048, "step": 40591 }, { "epoch": 2.164098736471717, "grad_norm": 0.1134350523352623, "learning_rate": 1.8818e-06, "loss": 0.0014, "step": 40592 }, { "epoch": 2.1641520499013702, "grad_norm": 2.3136959075927734, "learning_rate": 1.8816000000000001e-06, "loss": 0.0436, "step": 40593 }, { "epoch": 2.164205363331023, "grad_norm": 2.3958323001861572, "learning_rate": 1.8814e-06, "loss": 0.0307, "step": 40594 }, { "epoch": 2.164258676760676, "grad_norm": 1.2411998510360718, "learning_rate": 1.8812000000000003e-06, "loss": 0.0117, "step": 40595 }, { "epoch": 2.164311990190329, "grad_norm": 3.241442918777466, "learning_rate": 1.881e-06, "loss": 0.0319, "step": 40596 }, { "epoch": 2.1643653036199817, "grad_norm": 2.8714957237243652, "learning_rate": 1.8808e-06, "loss": 0.0272, "step": 40597 }, { "epoch": 2.164418617049635, "grad_norm": 0.33747413754463196, "learning_rate": 1.8806000000000003e-06, "loss": 0.0033, "step": 40598 }, { "epoch": 2.1644719304792877, "grad_norm": 1.345648169517517, "learning_rate": 1.8804000000000002e-06, "loss": 0.0221, "step": 40599 }, { "epoch": 2.164525243908941, "grad_norm": 1.6972748041152954, "learning_rate": 1.8802e-06, "loss": 0.0229, "step": 40600 }, { "epoch": 2.1645785573385936, "grad_norm": 1.6562803983688354, "learning_rate": 1.8800000000000002e-06, "loss": 0.0164, "step": 40601 }, { "epoch": 2.1646318707682464, "grad_norm": 2.4888086318969727, "learning_rate": 1.8798000000000002e-06, "loss": 0.0236, "step": 40602 }, { "epoch": 2.1646851841978996, "grad_norm": 1.2085984945297241, "learning_rate": 1.8796e-06, "loss": 0.008, "step": 40603 }, { "epoch": 2.1647384976275523, "grad_norm": 1.5098564624786377, "learning_rate": 1.8794000000000002e-06, "loss": 0.0079, "step": 40604 }, { "epoch": 2.1647918110572055, "grad_norm": 3.481386184692383, "learning_rate": 1.8792000000000002e-06, "loss": 0.037, "step": 40605 }, { "epoch": 2.1648451244868583, "grad_norm": 1.3591588735580444, "learning_rate": 1.8790000000000003e-06, "loss": 0.0172, "step": 40606 }, { "epoch": 2.164898437916511, "grad_norm": 1.6439836025238037, "learning_rate": 1.8788000000000001e-06, "loss": 0.0195, "step": 40607 }, { "epoch": 2.1649517513461642, "grad_norm": 3.6388611793518066, "learning_rate": 1.8786000000000001e-06, "loss": 0.0167, "step": 40608 }, { "epoch": 2.165005064775817, "grad_norm": 1.5035388469696045, "learning_rate": 1.8784000000000003e-06, "loss": 0.0104, "step": 40609 }, { "epoch": 2.1650583782054698, "grad_norm": 1.4936772584915161, "learning_rate": 1.8782e-06, "loss": 0.01, "step": 40610 }, { "epoch": 2.165111691635123, "grad_norm": 2.4719550609588623, "learning_rate": 1.878e-06, "loss": 0.0257, "step": 40611 }, { "epoch": 2.1651650050647757, "grad_norm": 1.1631189584732056, "learning_rate": 1.8778000000000003e-06, "loss": 0.0104, "step": 40612 }, { "epoch": 2.165218318494429, "grad_norm": 0.8267998695373535, "learning_rate": 1.8776000000000002e-06, "loss": 0.0041, "step": 40613 }, { "epoch": 2.1652716319240817, "grad_norm": 2.2133634090423584, "learning_rate": 1.8774e-06, "loss": 0.0345, "step": 40614 }, { "epoch": 2.1653249453537344, "grad_norm": 1.3754388093948364, "learning_rate": 1.8772000000000002e-06, "loss": 0.0146, "step": 40615 }, { "epoch": 2.1653782587833876, "grad_norm": 0.623180627822876, "learning_rate": 1.8770000000000002e-06, "loss": 0.0046, "step": 40616 }, { "epoch": 2.1654315722130404, "grad_norm": 1.1399281024932861, "learning_rate": 1.8768000000000002e-06, "loss": 0.0214, "step": 40617 }, { "epoch": 2.1654848856426936, "grad_norm": 1.3995345830917358, "learning_rate": 1.8766000000000002e-06, "loss": 0.0255, "step": 40618 }, { "epoch": 2.1655381990723463, "grad_norm": 2.6236424446105957, "learning_rate": 1.8764000000000002e-06, "loss": 0.0158, "step": 40619 }, { "epoch": 2.165591512501999, "grad_norm": 2.3775124549865723, "learning_rate": 1.8762000000000001e-06, "loss": 0.0205, "step": 40620 }, { "epoch": 2.1656448259316523, "grad_norm": 3.056838035583496, "learning_rate": 1.8760000000000001e-06, "loss": 0.038, "step": 40621 }, { "epoch": 2.165698139361305, "grad_norm": 0.962711751461029, "learning_rate": 1.8758000000000001e-06, "loss": 0.0051, "step": 40622 }, { "epoch": 2.165751452790958, "grad_norm": 1.8742272853851318, "learning_rate": 1.8756e-06, "loss": 0.0097, "step": 40623 }, { "epoch": 2.165804766220611, "grad_norm": 0.5599890947341919, "learning_rate": 1.8754000000000003e-06, "loss": 0.0033, "step": 40624 }, { "epoch": 2.165858079650264, "grad_norm": 0.30998802185058594, "learning_rate": 1.8752e-06, "loss": 0.0027, "step": 40625 }, { "epoch": 2.165911393079917, "grad_norm": 1.52316415309906, "learning_rate": 1.8750000000000003e-06, "loss": 0.0266, "step": 40626 }, { "epoch": 2.1659647065095697, "grad_norm": 3.089409351348877, "learning_rate": 1.8748000000000002e-06, "loss": 0.0405, "step": 40627 }, { "epoch": 2.1660180199392225, "grad_norm": 0.6818354725837708, "learning_rate": 1.8746e-06, "loss": 0.0046, "step": 40628 }, { "epoch": 2.1660713333688757, "grad_norm": 2.324357032775879, "learning_rate": 1.8744000000000002e-06, "loss": 0.0116, "step": 40629 }, { "epoch": 2.1661246467985285, "grad_norm": 2.210615873336792, "learning_rate": 1.8742000000000002e-06, "loss": 0.0302, "step": 40630 }, { "epoch": 2.1661779602281817, "grad_norm": 1.5592385530471802, "learning_rate": 1.8740000000000002e-06, "loss": 0.0051, "step": 40631 }, { "epoch": 2.1662312736578344, "grad_norm": 0.16837377846240997, "learning_rate": 1.8738000000000002e-06, "loss": 0.0018, "step": 40632 }, { "epoch": 2.166284587087487, "grad_norm": 3.1998610496520996, "learning_rate": 1.8736000000000002e-06, "loss": 0.0576, "step": 40633 }, { "epoch": 2.1663379005171404, "grad_norm": 0.5033723711967468, "learning_rate": 1.8734000000000001e-06, "loss": 0.0045, "step": 40634 }, { "epoch": 2.166391213946793, "grad_norm": 2.7172224521636963, "learning_rate": 1.8732000000000001e-06, "loss": 0.0164, "step": 40635 }, { "epoch": 2.1664445273764463, "grad_norm": 1.3043301105499268, "learning_rate": 1.8730000000000001e-06, "loss": 0.0085, "step": 40636 }, { "epoch": 2.166497840806099, "grad_norm": 2.895073890686035, "learning_rate": 1.8728e-06, "loss": 0.0193, "step": 40637 }, { "epoch": 2.166551154235752, "grad_norm": 2.3789563179016113, "learning_rate": 1.8726000000000003e-06, "loss": 0.0172, "step": 40638 }, { "epoch": 2.166604467665405, "grad_norm": 2.048818588256836, "learning_rate": 1.8724e-06, "loss": 0.0167, "step": 40639 }, { "epoch": 2.166657781095058, "grad_norm": 0.48077815771102905, "learning_rate": 1.8722e-06, "loss": 0.0043, "step": 40640 }, { "epoch": 2.166711094524711, "grad_norm": 1.0403698682785034, "learning_rate": 1.8720000000000002e-06, "loss": 0.005, "step": 40641 }, { "epoch": 2.1667644079543638, "grad_norm": 1.332328200340271, "learning_rate": 1.8718000000000002e-06, "loss": 0.0057, "step": 40642 }, { "epoch": 2.1668177213840165, "grad_norm": 3.837527275085449, "learning_rate": 1.8716e-06, "loss": 0.0399, "step": 40643 }, { "epoch": 2.1668710348136697, "grad_norm": 3.154189348220825, "learning_rate": 1.8714000000000002e-06, "loss": 0.0551, "step": 40644 }, { "epoch": 2.1669243482433225, "grad_norm": 0.8888877630233765, "learning_rate": 1.8712000000000002e-06, "loss": 0.0142, "step": 40645 }, { "epoch": 2.1669776616729752, "grad_norm": 2.27699875831604, "learning_rate": 1.871e-06, "loss": 0.0201, "step": 40646 }, { "epoch": 2.1670309751026284, "grad_norm": 0.6551096439361572, "learning_rate": 1.8708000000000002e-06, "loss": 0.0042, "step": 40647 }, { "epoch": 2.167084288532281, "grad_norm": 2.8592307567596436, "learning_rate": 1.8706000000000001e-06, "loss": 0.0156, "step": 40648 }, { "epoch": 2.1671376019619344, "grad_norm": 1.4521572589874268, "learning_rate": 1.8704000000000003e-06, "loss": 0.0181, "step": 40649 }, { "epoch": 2.167190915391587, "grad_norm": 2.413409471511841, "learning_rate": 1.8702000000000001e-06, "loss": 0.0264, "step": 40650 }, { "epoch": 2.16724422882124, "grad_norm": 1.5312801599502563, "learning_rate": 1.87e-06, "loss": 0.0236, "step": 40651 }, { "epoch": 2.167297542250893, "grad_norm": 1.7425222396850586, "learning_rate": 1.8698000000000003e-06, "loss": 0.01, "step": 40652 }, { "epoch": 2.167350855680546, "grad_norm": 1.240043044090271, "learning_rate": 1.8696e-06, "loss": 0.006, "step": 40653 }, { "epoch": 2.167404169110199, "grad_norm": 3.658660650253296, "learning_rate": 1.8694e-06, "loss": 0.0154, "step": 40654 }, { "epoch": 2.167457482539852, "grad_norm": 3.3619887828826904, "learning_rate": 1.8692000000000002e-06, "loss": 0.0205, "step": 40655 }, { "epoch": 2.1675107959695046, "grad_norm": 1.219708800315857, "learning_rate": 1.8690000000000002e-06, "loss": 0.0076, "step": 40656 }, { "epoch": 2.167564109399158, "grad_norm": 0.9314088225364685, "learning_rate": 1.8688e-06, "loss": 0.0049, "step": 40657 }, { "epoch": 2.1676174228288105, "grad_norm": 2.9448630809783936, "learning_rate": 1.8686000000000002e-06, "loss": 0.0299, "step": 40658 }, { "epoch": 2.1676707362584633, "grad_norm": 0.44144049286842346, "learning_rate": 1.8684000000000002e-06, "loss": 0.0036, "step": 40659 }, { "epoch": 2.1677240496881165, "grad_norm": 3.758985996246338, "learning_rate": 1.8682000000000004e-06, "loss": 0.0276, "step": 40660 }, { "epoch": 2.1677773631177693, "grad_norm": 3.0450165271759033, "learning_rate": 1.8680000000000002e-06, "loss": 0.0193, "step": 40661 }, { "epoch": 2.1678306765474225, "grad_norm": 1.4807268381118774, "learning_rate": 1.8678000000000001e-06, "loss": 0.0248, "step": 40662 }, { "epoch": 2.167883989977075, "grad_norm": 2.0429766178131104, "learning_rate": 1.8676000000000003e-06, "loss": 0.0113, "step": 40663 }, { "epoch": 2.167937303406728, "grad_norm": 1.8280011415481567, "learning_rate": 1.8674000000000001e-06, "loss": 0.0099, "step": 40664 }, { "epoch": 2.167990616836381, "grad_norm": 2.1987435817718506, "learning_rate": 1.8672e-06, "loss": 0.023, "step": 40665 }, { "epoch": 2.168043930266034, "grad_norm": 1.994188904762268, "learning_rate": 1.8670000000000003e-06, "loss": 0.0088, "step": 40666 }, { "epoch": 2.168097243695687, "grad_norm": 1.552182912826538, "learning_rate": 1.8668000000000003e-06, "loss": 0.0082, "step": 40667 }, { "epoch": 2.16815055712534, "grad_norm": 0.7800897359848022, "learning_rate": 1.8666e-06, "loss": 0.0045, "step": 40668 }, { "epoch": 2.1682038705549926, "grad_norm": 2.3637020587921143, "learning_rate": 1.8664000000000002e-06, "loss": 0.0135, "step": 40669 }, { "epoch": 2.168257183984646, "grad_norm": 2.0432069301605225, "learning_rate": 1.8662000000000002e-06, "loss": 0.0138, "step": 40670 }, { "epoch": 2.1683104974142986, "grad_norm": 1.610632300376892, "learning_rate": 1.866e-06, "loss": 0.0239, "step": 40671 }, { "epoch": 2.1683638108439514, "grad_norm": 0.8582860231399536, "learning_rate": 1.8658000000000002e-06, "loss": 0.0037, "step": 40672 }, { "epoch": 2.1684171242736046, "grad_norm": 1.7635868787765503, "learning_rate": 1.8656000000000002e-06, "loss": 0.017, "step": 40673 }, { "epoch": 2.1684704377032573, "grad_norm": 2.0926897525787354, "learning_rate": 1.8654000000000002e-06, "loss": 0.0302, "step": 40674 }, { "epoch": 2.1685237511329105, "grad_norm": 1.0434644222259521, "learning_rate": 1.8652000000000002e-06, "loss": 0.0041, "step": 40675 }, { "epoch": 2.1685770645625633, "grad_norm": 0.5047233700752258, "learning_rate": 1.8650000000000001e-06, "loss": 0.0036, "step": 40676 }, { "epoch": 2.1686303779922165, "grad_norm": 2.6797618865966797, "learning_rate": 1.8648000000000001e-06, "loss": 0.0495, "step": 40677 }, { "epoch": 2.1686836914218692, "grad_norm": 1.3032817840576172, "learning_rate": 1.8646000000000001e-06, "loss": 0.0084, "step": 40678 }, { "epoch": 2.168737004851522, "grad_norm": 2.159275531768799, "learning_rate": 1.8644e-06, "loss": 0.0185, "step": 40679 }, { "epoch": 2.168790318281175, "grad_norm": 1.7298755645751953, "learning_rate": 1.8642e-06, "loss": 0.0039, "step": 40680 }, { "epoch": 2.168843631710828, "grad_norm": 0.4005388617515564, "learning_rate": 1.8640000000000003e-06, "loss": 0.0022, "step": 40681 }, { "epoch": 2.1688969451404807, "grad_norm": 2.144284725189209, "learning_rate": 1.8638e-06, "loss": 0.0505, "step": 40682 }, { "epoch": 2.168950258570134, "grad_norm": 1.810990810394287, "learning_rate": 1.8636e-06, "loss": 0.0222, "step": 40683 }, { "epoch": 2.1690035719997867, "grad_norm": 1.971907138824463, "learning_rate": 1.8634000000000002e-06, "loss": 0.0096, "step": 40684 }, { "epoch": 2.16905688542944, "grad_norm": 0.7964106202125549, "learning_rate": 1.8632000000000002e-06, "loss": 0.003, "step": 40685 }, { "epoch": 2.1691101988590926, "grad_norm": 1.0582550764083862, "learning_rate": 1.863e-06, "loss": 0.0117, "step": 40686 }, { "epoch": 2.1691635122887454, "grad_norm": 2.5852737426757812, "learning_rate": 1.8628000000000002e-06, "loss": 0.0305, "step": 40687 }, { "epoch": 2.1692168257183986, "grad_norm": 1.834898591041565, "learning_rate": 1.8626000000000002e-06, "loss": 0.0245, "step": 40688 }, { "epoch": 2.1692701391480513, "grad_norm": 1.4410322904586792, "learning_rate": 1.8624e-06, "loss": 0.0088, "step": 40689 }, { "epoch": 2.1693234525777045, "grad_norm": 2.247009515762329, "learning_rate": 1.8622000000000001e-06, "loss": 0.0169, "step": 40690 }, { "epoch": 2.1693767660073573, "grad_norm": 1.1297661066055298, "learning_rate": 1.8620000000000001e-06, "loss": 0.0058, "step": 40691 }, { "epoch": 2.16943007943701, "grad_norm": 0.6584681272506714, "learning_rate": 1.8618000000000003e-06, "loss": 0.005, "step": 40692 }, { "epoch": 2.1694833928666633, "grad_norm": 2.044952154159546, "learning_rate": 1.8616e-06, "loss": 0.0128, "step": 40693 }, { "epoch": 2.169536706296316, "grad_norm": 1.772401213645935, "learning_rate": 1.8614e-06, "loss": 0.0263, "step": 40694 }, { "epoch": 2.1695900197259688, "grad_norm": 1.9395653009414673, "learning_rate": 1.8612000000000003e-06, "loss": 0.0076, "step": 40695 }, { "epoch": 2.169643333155622, "grad_norm": 1.8314896821975708, "learning_rate": 1.861e-06, "loss": 0.0307, "step": 40696 }, { "epoch": 2.1696966465852747, "grad_norm": 2.2282164096832275, "learning_rate": 1.8608e-06, "loss": 0.01, "step": 40697 }, { "epoch": 2.169749960014928, "grad_norm": 3.3668766021728516, "learning_rate": 1.8606000000000002e-06, "loss": 0.0126, "step": 40698 }, { "epoch": 2.1698032734445807, "grad_norm": 2.2311596870422363, "learning_rate": 1.8604000000000002e-06, "loss": 0.0274, "step": 40699 }, { "epoch": 2.1698565868742334, "grad_norm": 3.3969764709472656, "learning_rate": 1.8602e-06, "loss": 0.0117, "step": 40700 }, { "epoch": 2.1699099003038866, "grad_norm": 1.8310493230819702, "learning_rate": 1.8600000000000002e-06, "loss": 0.0172, "step": 40701 }, { "epoch": 2.1699632137335394, "grad_norm": 2.351670503616333, "learning_rate": 1.8598000000000002e-06, "loss": 0.0383, "step": 40702 }, { "epoch": 2.1700165271631926, "grad_norm": 2.2481744289398193, "learning_rate": 1.8596e-06, "loss": 0.0529, "step": 40703 }, { "epoch": 2.1700698405928454, "grad_norm": 1.8966634273529053, "learning_rate": 1.8594000000000001e-06, "loss": 0.0146, "step": 40704 }, { "epoch": 2.170123154022498, "grad_norm": 1.4015498161315918, "learning_rate": 1.8592000000000001e-06, "loss": 0.0123, "step": 40705 }, { "epoch": 2.1701764674521513, "grad_norm": 1.9841487407684326, "learning_rate": 1.8590000000000003e-06, "loss": 0.0069, "step": 40706 }, { "epoch": 2.170229780881804, "grad_norm": 1.113501787185669, "learning_rate": 1.8588e-06, "loss": 0.017, "step": 40707 }, { "epoch": 2.170283094311457, "grad_norm": 1.3448071479797363, "learning_rate": 1.8586e-06, "loss": 0.0111, "step": 40708 }, { "epoch": 2.17033640774111, "grad_norm": 1.631957769393921, "learning_rate": 1.8584000000000003e-06, "loss": 0.0302, "step": 40709 }, { "epoch": 2.170389721170763, "grad_norm": 2.8042478561401367, "learning_rate": 1.8582000000000003e-06, "loss": 0.0131, "step": 40710 }, { "epoch": 2.170443034600416, "grad_norm": 0.8020713329315186, "learning_rate": 1.858e-06, "loss": 0.0052, "step": 40711 }, { "epoch": 2.1704963480300687, "grad_norm": 1.4216336011886597, "learning_rate": 1.8578000000000002e-06, "loss": 0.0098, "step": 40712 }, { "epoch": 2.1705496614597215, "grad_norm": 3.4775021076202393, "learning_rate": 1.8576000000000002e-06, "loss": 0.0167, "step": 40713 }, { "epoch": 2.1706029748893747, "grad_norm": 1.9582085609436035, "learning_rate": 1.8574e-06, "loss": 0.0143, "step": 40714 }, { "epoch": 2.1706562883190275, "grad_norm": 0.4211878478527069, "learning_rate": 1.8572000000000002e-06, "loss": 0.004, "step": 40715 }, { "epoch": 2.1707096017486807, "grad_norm": 1.1291371583938599, "learning_rate": 1.8570000000000002e-06, "loss": 0.0156, "step": 40716 }, { "epoch": 2.1707629151783334, "grad_norm": 2.2187132835388184, "learning_rate": 1.8568000000000004e-06, "loss": 0.0245, "step": 40717 }, { "epoch": 2.170816228607986, "grad_norm": 0.48088932037353516, "learning_rate": 1.8566000000000001e-06, "loss": 0.0041, "step": 40718 }, { "epoch": 2.1708695420376394, "grad_norm": 5.975640773773193, "learning_rate": 1.8564000000000001e-06, "loss": 0.0137, "step": 40719 }, { "epoch": 2.170922855467292, "grad_norm": 1.6286959648132324, "learning_rate": 1.8562000000000003e-06, "loss": 0.0088, "step": 40720 }, { "epoch": 2.1709761688969453, "grad_norm": 0.5711736679077148, "learning_rate": 1.856e-06, "loss": 0.0029, "step": 40721 }, { "epoch": 2.171029482326598, "grad_norm": 1.7122517824172974, "learning_rate": 1.8558e-06, "loss": 0.0276, "step": 40722 }, { "epoch": 2.171082795756251, "grad_norm": 1.558638334274292, "learning_rate": 1.8556000000000003e-06, "loss": 0.0082, "step": 40723 }, { "epoch": 2.171136109185904, "grad_norm": 2.195974349975586, "learning_rate": 1.8554000000000003e-06, "loss": 0.0365, "step": 40724 }, { "epoch": 2.171189422615557, "grad_norm": 1.9001771211624146, "learning_rate": 1.8552e-06, "loss": 0.0087, "step": 40725 }, { "epoch": 2.17124273604521, "grad_norm": 0.6027767062187195, "learning_rate": 1.8550000000000002e-06, "loss": 0.0043, "step": 40726 }, { "epoch": 2.1712960494748628, "grad_norm": 2.3750782012939453, "learning_rate": 1.8548000000000002e-06, "loss": 0.0181, "step": 40727 }, { "epoch": 2.1713493629045155, "grad_norm": 2.188483238220215, "learning_rate": 1.8546000000000002e-06, "loss": 0.0436, "step": 40728 }, { "epoch": 2.1714026763341687, "grad_norm": 2.13529896736145, "learning_rate": 1.8544000000000002e-06, "loss": 0.0213, "step": 40729 }, { "epoch": 2.1714559897638215, "grad_norm": 1.6033220291137695, "learning_rate": 1.8542000000000002e-06, "loss": 0.0145, "step": 40730 }, { "epoch": 2.1715093031934742, "grad_norm": 2.771897077560425, "learning_rate": 1.8540000000000002e-06, "loss": 0.0378, "step": 40731 }, { "epoch": 2.1715626166231274, "grad_norm": 1.4977542161941528, "learning_rate": 1.8538000000000001e-06, "loss": 0.0244, "step": 40732 }, { "epoch": 2.17161593005278, "grad_norm": 1.9777107238769531, "learning_rate": 1.8536000000000001e-06, "loss": 0.0114, "step": 40733 }, { "epoch": 2.1716692434824334, "grad_norm": 2.1094629764556885, "learning_rate": 1.8534000000000001e-06, "loss": 0.0121, "step": 40734 }, { "epoch": 2.171722556912086, "grad_norm": 3.1452865600585938, "learning_rate": 1.8532000000000003e-06, "loss": 0.0298, "step": 40735 }, { "epoch": 2.171775870341739, "grad_norm": 2.066859722137451, "learning_rate": 1.853e-06, "loss": 0.0065, "step": 40736 }, { "epoch": 2.171829183771392, "grad_norm": 0.5823577046394348, "learning_rate": 1.8528e-06, "loss": 0.0051, "step": 40737 }, { "epoch": 2.171882497201045, "grad_norm": 1.5407590866088867, "learning_rate": 1.8526000000000003e-06, "loss": 0.0128, "step": 40738 }, { "epoch": 2.171935810630698, "grad_norm": 2.512791872024536, "learning_rate": 1.8524e-06, "loss": 0.0232, "step": 40739 }, { "epoch": 2.171989124060351, "grad_norm": 2.79590106010437, "learning_rate": 1.8522e-06, "loss": 0.0264, "step": 40740 }, { "epoch": 2.1720424374900036, "grad_norm": 1.5060786008834839, "learning_rate": 1.8520000000000002e-06, "loss": 0.0155, "step": 40741 }, { "epoch": 2.172095750919657, "grad_norm": 1.8591458797454834, "learning_rate": 1.8518000000000002e-06, "loss": 0.0194, "step": 40742 }, { "epoch": 2.1721490643493095, "grad_norm": 1.053790807723999, "learning_rate": 1.8516e-06, "loss": 0.0145, "step": 40743 }, { "epoch": 2.1722023777789623, "grad_norm": 0.5477633476257324, "learning_rate": 1.8514000000000002e-06, "loss": 0.0049, "step": 40744 }, { "epoch": 2.1722556912086155, "grad_norm": 2.004117012023926, "learning_rate": 1.8512000000000002e-06, "loss": 0.0273, "step": 40745 }, { "epoch": 2.1723090046382683, "grad_norm": 1.6056362390518188, "learning_rate": 1.851e-06, "loss": 0.0155, "step": 40746 }, { "epoch": 2.1723623180679215, "grad_norm": 1.9980523586273193, "learning_rate": 1.8508000000000001e-06, "loss": 0.0137, "step": 40747 }, { "epoch": 2.172415631497574, "grad_norm": 1.9895116090774536, "learning_rate": 1.8506000000000001e-06, "loss": 0.024, "step": 40748 }, { "epoch": 2.172468944927227, "grad_norm": 2.030531644821167, "learning_rate": 1.8504000000000003e-06, "loss": 0.0096, "step": 40749 }, { "epoch": 2.17252225835688, "grad_norm": 2.0250205993652344, "learning_rate": 1.8502e-06, "loss": 0.0171, "step": 40750 }, { "epoch": 2.172575571786533, "grad_norm": 1.4125909805297852, "learning_rate": 1.85e-06, "loss": 0.0235, "step": 40751 }, { "epoch": 2.172628885216186, "grad_norm": 0.5539407134056091, "learning_rate": 1.8498000000000003e-06, "loss": 0.0037, "step": 40752 }, { "epoch": 2.172682198645839, "grad_norm": 1.3109430074691772, "learning_rate": 1.8496000000000003e-06, "loss": 0.0136, "step": 40753 }, { "epoch": 2.1727355120754916, "grad_norm": 2.91921067237854, "learning_rate": 1.8494e-06, "loss": 0.0193, "step": 40754 }, { "epoch": 2.172788825505145, "grad_norm": 1.5104782581329346, "learning_rate": 1.8492000000000002e-06, "loss": 0.0089, "step": 40755 }, { "epoch": 2.1728421389347976, "grad_norm": 2.554842472076416, "learning_rate": 1.8490000000000002e-06, "loss": 0.032, "step": 40756 }, { "epoch": 2.172895452364451, "grad_norm": 5.439613342285156, "learning_rate": 1.8488e-06, "loss": 0.0218, "step": 40757 }, { "epoch": 2.1729487657941036, "grad_norm": 1.9088038206100464, "learning_rate": 1.8486000000000002e-06, "loss": 0.0067, "step": 40758 }, { "epoch": 2.1730020792237563, "grad_norm": 1.957077145576477, "learning_rate": 1.8484000000000002e-06, "loss": 0.0203, "step": 40759 }, { "epoch": 2.1730553926534095, "grad_norm": 4.037933826446533, "learning_rate": 1.8482000000000004e-06, "loss": 0.0198, "step": 40760 }, { "epoch": 2.1731087060830623, "grad_norm": 1.6284574270248413, "learning_rate": 1.8480000000000001e-06, "loss": 0.0098, "step": 40761 }, { "epoch": 2.1731620195127155, "grad_norm": 1.5285604000091553, "learning_rate": 1.8478000000000001e-06, "loss": 0.0071, "step": 40762 }, { "epoch": 2.1732153329423682, "grad_norm": 1.9747085571289062, "learning_rate": 1.8476000000000003e-06, "loss": 0.0151, "step": 40763 }, { "epoch": 2.173268646372021, "grad_norm": 1.7941828966140747, "learning_rate": 1.8474e-06, "loss": 0.0617, "step": 40764 }, { "epoch": 2.173321959801674, "grad_norm": 1.2159274816513062, "learning_rate": 1.8472e-06, "loss": 0.0047, "step": 40765 }, { "epoch": 2.173375273231327, "grad_norm": 1.8275411128997803, "learning_rate": 1.8470000000000003e-06, "loss": 0.0352, "step": 40766 }, { "epoch": 2.1734285866609797, "grad_norm": 2.333963632583618, "learning_rate": 1.8468000000000003e-06, "loss": 0.0203, "step": 40767 }, { "epoch": 2.173481900090633, "grad_norm": 0.8293726444244385, "learning_rate": 1.8466e-06, "loss": 0.0047, "step": 40768 }, { "epoch": 2.1735352135202857, "grad_norm": 2.8911542892456055, "learning_rate": 1.8464000000000002e-06, "loss": 0.0206, "step": 40769 }, { "epoch": 2.173588526949939, "grad_norm": 1.411900281906128, "learning_rate": 1.8462000000000002e-06, "loss": 0.0061, "step": 40770 }, { "epoch": 2.1736418403795916, "grad_norm": 3.303661823272705, "learning_rate": 1.846e-06, "loss": 0.0413, "step": 40771 }, { "epoch": 2.1736951538092444, "grad_norm": 1.4585219621658325, "learning_rate": 1.8458000000000002e-06, "loss": 0.0103, "step": 40772 }, { "epoch": 2.1737484672388976, "grad_norm": 2.282369613647461, "learning_rate": 1.8456000000000002e-06, "loss": 0.0236, "step": 40773 }, { "epoch": 2.1738017806685503, "grad_norm": 1.5611201524734497, "learning_rate": 1.8454000000000002e-06, "loss": 0.0242, "step": 40774 }, { "epoch": 2.1738550940982035, "grad_norm": 4.082587242126465, "learning_rate": 1.8452000000000001e-06, "loss": 0.0191, "step": 40775 }, { "epoch": 2.1739084075278563, "grad_norm": 1.2607014179229736, "learning_rate": 1.8450000000000001e-06, "loss": 0.0108, "step": 40776 }, { "epoch": 2.173961720957509, "grad_norm": 1.5514461994171143, "learning_rate": 1.8448000000000001e-06, "loss": 0.0143, "step": 40777 }, { "epoch": 2.1740150343871623, "grad_norm": 1.1278597116470337, "learning_rate": 1.8446000000000003e-06, "loss": 0.0068, "step": 40778 }, { "epoch": 2.174068347816815, "grad_norm": 1.685031533241272, "learning_rate": 1.8444e-06, "loss": 0.023, "step": 40779 }, { "epoch": 2.1741216612464678, "grad_norm": 0.53192138671875, "learning_rate": 1.8442e-06, "loss": 0.004, "step": 40780 }, { "epoch": 2.174174974676121, "grad_norm": 1.8458305597305298, "learning_rate": 1.8440000000000003e-06, "loss": 0.02, "step": 40781 }, { "epoch": 2.1742282881057737, "grad_norm": 1.6658356189727783, "learning_rate": 1.8438e-06, "loss": 0.0164, "step": 40782 }, { "epoch": 2.174281601535427, "grad_norm": 4.399128437042236, "learning_rate": 1.8436e-06, "loss": 0.0143, "step": 40783 }, { "epoch": 2.1743349149650797, "grad_norm": 1.5658272504806519, "learning_rate": 1.8434000000000002e-06, "loss": 0.0331, "step": 40784 }, { "epoch": 2.1743882283947324, "grad_norm": 1.6769777536392212, "learning_rate": 1.8432000000000002e-06, "loss": 0.0378, "step": 40785 }, { "epoch": 2.1744415418243856, "grad_norm": 3.1036367416381836, "learning_rate": 1.843e-06, "loss": 0.0217, "step": 40786 }, { "epoch": 2.1744948552540384, "grad_norm": 2.2611255645751953, "learning_rate": 1.8428000000000002e-06, "loss": 0.0123, "step": 40787 }, { "epoch": 2.1745481686836916, "grad_norm": 1.4279111623764038, "learning_rate": 1.8426000000000002e-06, "loss": 0.0179, "step": 40788 }, { "epoch": 2.1746014821133444, "grad_norm": 1.3199127912521362, "learning_rate": 1.8424e-06, "loss": 0.0128, "step": 40789 }, { "epoch": 2.174654795542997, "grad_norm": 0.20459598302841187, "learning_rate": 1.8422000000000001e-06, "loss": 0.0023, "step": 40790 }, { "epoch": 2.1747081089726503, "grad_norm": 0.9571322202682495, "learning_rate": 1.8420000000000001e-06, "loss": 0.0058, "step": 40791 }, { "epoch": 2.174761422402303, "grad_norm": 0.951255738735199, "learning_rate": 1.8418000000000003e-06, "loss": 0.0141, "step": 40792 }, { "epoch": 2.174814735831956, "grad_norm": 5.922459125518799, "learning_rate": 1.8416e-06, "loss": 0.0188, "step": 40793 }, { "epoch": 2.174868049261609, "grad_norm": 1.3098971843719482, "learning_rate": 1.8414e-06, "loss": 0.006, "step": 40794 }, { "epoch": 2.174921362691262, "grad_norm": 1.6101094484329224, "learning_rate": 1.8412000000000003e-06, "loss": 0.0259, "step": 40795 }, { "epoch": 2.174974676120915, "grad_norm": 1.0932856798171997, "learning_rate": 1.8410000000000002e-06, "loss": 0.0153, "step": 40796 }, { "epoch": 2.1750279895505678, "grad_norm": 1.4601036310195923, "learning_rate": 1.8408e-06, "loss": 0.0165, "step": 40797 }, { "epoch": 2.1750813029802205, "grad_norm": 1.2617067098617554, "learning_rate": 1.8406000000000002e-06, "loss": 0.0149, "step": 40798 }, { "epoch": 2.1751346164098737, "grad_norm": 0.9864876866340637, "learning_rate": 1.8404000000000002e-06, "loss": 0.003, "step": 40799 }, { "epoch": 2.1751879298395265, "grad_norm": 1.1160420179367065, "learning_rate": 1.8402e-06, "loss": 0.0052, "step": 40800 }, { "epoch": 2.1752412432691797, "grad_norm": 1.7579538822174072, "learning_rate": 1.8400000000000002e-06, "loss": 0.016, "step": 40801 }, { "epoch": 2.1752945566988324, "grad_norm": 2.9971537590026855, "learning_rate": 1.8398000000000002e-06, "loss": 0.0183, "step": 40802 }, { "epoch": 2.175347870128485, "grad_norm": 2.2826309204101562, "learning_rate": 1.8396000000000004e-06, "loss": 0.0235, "step": 40803 }, { "epoch": 2.1754011835581384, "grad_norm": 2.6579325199127197, "learning_rate": 1.8394000000000001e-06, "loss": 0.0329, "step": 40804 }, { "epoch": 2.175454496987791, "grad_norm": 0.6682888269424438, "learning_rate": 1.8392000000000001e-06, "loss": 0.004, "step": 40805 }, { "epoch": 2.1755078104174443, "grad_norm": 1.265434980392456, "learning_rate": 1.8390000000000003e-06, "loss": 0.0094, "step": 40806 }, { "epoch": 2.175561123847097, "grad_norm": 1.307687520980835, "learning_rate": 1.8388e-06, "loss": 0.0175, "step": 40807 }, { "epoch": 2.17561443727675, "grad_norm": 1.6999772787094116, "learning_rate": 1.8386e-06, "loss": 0.0261, "step": 40808 }, { "epoch": 2.175667750706403, "grad_norm": 0.7705154418945312, "learning_rate": 1.8384000000000003e-06, "loss": 0.019, "step": 40809 }, { "epoch": 2.175721064136056, "grad_norm": 1.4082958698272705, "learning_rate": 1.8382000000000002e-06, "loss": 0.0132, "step": 40810 }, { "epoch": 2.175774377565709, "grad_norm": 1.7883872985839844, "learning_rate": 1.838e-06, "loss": 0.0221, "step": 40811 }, { "epoch": 2.1758276909953618, "grad_norm": 0.693389356136322, "learning_rate": 1.8378000000000002e-06, "loss": 0.0165, "step": 40812 }, { "epoch": 2.1758810044250145, "grad_norm": 2.002448320388794, "learning_rate": 1.8376000000000002e-06, "loss": 0.0128, "step": 40813 }, { "epoch": 2.1759343178546677, "grad_norm": 1.7093855142593384, "learning_rate": 1.8374e-06, "loss": 0.0237, "step": 40814 }, { "epoch": 2.1759876312843205, "grad_norm": 1.638129472732544, "learning_rate": 1.8372000000000002e-06, "loss": 0.0233, "step": 40815 }, { "epoch": 2.1760409447139732, "grad_norm": 1.0812143087387085, "learning_rate": 1.8370000000000002e-06, "loss": 0.0147, "step": 40816 }, { "epoch": 2.1760942581436264, "grad_norm": 3.7240335941314697, "learning_rate": 1.8368000000000004e-06, "loss": 0.0312, "step": 40817 }, { "epoch": 2.176147571573279, "grad_norm": 1.045767068862915, "learning_rate": 1.8366000000000001e-06, "loss": 0.0161, "step": 40818 }, { "epoch": 2.1762008850029324, "grad_norm": 3.1992897987365723, "learning_rate": 1.8364000000000001e-06, "loss": 0.0332, "step": 40819 }, { "epoch": 2.176254198432585, "grad_norm": 2.8380844593048096, "learning_rate": 1.8362000000000003e-06, "loss": 0.0358, "step": 40820 }, { "epoch": 2.176307511862238, "grad_norm": 1.6427894830703735, "learning_rate": 1.8360000000000003e-06, "loss": 0.0088, "step": 40821 }, { "epoch": 2.176360825291891, "grad_norm": 2.1210765838623047, "learning_rate": 1.8358e-06, "loss": 0.0204, "step": 40822 }, { "epoch": 2.176414138721544, "grad_norm": 1.7108275890350342, "learning_rate": 1.8356000000000003e-06, "loss": 0.0234, "step": 40823 }, { "epoch": 2.176467452151197, "grad_norm": 2.102374315261841, "learning_rate": 1.8354000000000002e-06, "loss": 0.0194, "step": 40824 }, { "epoch": 2.17652076558085, "grad_norm": 1.5268303155899048, "learning_rate": 1.8352e-06, "loss": 0.0094, "step": 40825 }, { "epoch": 2.1765740790105026, "grad_norm": 0.37912020087242126, "learning_rate": 1.8350000000000002e-06, "loss": 0.0033, "step": 40826 }, { "epoch": 2.176627392440156, "grad_norm": 1.2289925813674927, "learning_rate": 1.8348000000000002e-06, "loss": 0.0154, "step": 40827 }, { "epoch": 2.1766807058698086, "grad_norm": 2.0306529998779297, "learning_rate": 1.8346000000000002e-06, "loss": 0.0212, "step": 40828 }, { "epoch": 2.1767340192994613, "grad_norm": 0.32602646946907043, "learning_rate": 1.8344000000000002e-06, "loss": 0.0028, "step": 40829 }, { "epoch": 2.1767873327291145, "grad_norm": 0.8925120234489441, "learning_rate": 1.8342000000000002e-06, "loss": 0.0043, "step": 40830 }, { "epoch": 2.1768406461587673, "grad_norm": 0.7002934813499451, "learning_rate": 1.8340000000000001e-06, "loss": 0.0038, "step": 40831 }, { "epoch": 2.1768939595884205, "grad_norm": 2.089912176132202, "learning_rate": 1.8338000000000001e-06, "loss": 0.0139, "step": 40832 }, { "epoch": 2.1769472730180732, "grad_norm": 0.9659368991851807, "learning_rate": 1.8336000000000001e-06, "loss": 0.007, "step": 40833 }, { "epoch": 2.177000586447726, "grad_norm": 4.516622066497803, "learning_rate": 1.8334e-06, "loss": 0.0087, "step": 40834 }, { "epoch": 2.177053899877379, "grad_norm": 1.0615226030349731, "learning_rate": 1.8332000000000003e-06, "loss": 0.0076, "step": 40835 }, { "epoch": 2.177107213307032, "grad_norm": 0.4104011058807373, "learning_rate": 1.833e-06, "loss": 0.0032, "step": 40836 }, { "epoch": 2.177160526736685, "grad_norm": 1.6355115175247192, "learning_rate": 1.8328e-06, "loss": 0.0255, "step": 40837 }, { "epoch": 2.177213840166338, "grad_norm": 1.56121826171875, "learning_rate": 1.8326000000000003e-06, "loss": 0.0134, "step": 40838 }, { "epoch": 2.1772671535959907, "grad_norm": 2.0209949016571045, "learning_rate": 1.8324e-06, "loss": 0.0423, "step": 40839 }, { "epoch": 2.177320467025644, "grad_norm": 0.29688218235969543, "learning_rate": 1.8322e-06, "loss": 0.0025, "step": 40840 }, { "epoch": 2.1773737804552966, "grad_norm": 1.0703120231628418, "learning_rate": 1.8320000000000002e-06, "loss": 0.0183, "step": 40841 }, { "epoch": 2.17742709388495, "grad_norm": 0.8671690225601196, "learning_rate": 1.8318000000000002e-06, "loss": 0.0048, "step": 40842 }, { "epoch": 2.1774804073146026, "grad_norm": 2.6692233085632324, "learning_rate": 1.8316e-06, "loss": 0.0314, "step": 40843 }, { "epoch": 2.1775337207442553, "grad_norm": 2.412403106689453, "learning_rate": 1.8314000000000002e-06, "loss": 0.0287, "step": 40844 }, { "epoch": 2.1775870341739085, "grad_norm": 1.2232881784439087, "learning_rate": 1.8312000000000001e-06, "loss": 0.0179, "step": 40845 }, { "epoch": 2.1776403476035613, "grad_norm": 2.3126049041748047, "learning_rate": 1.8310000000000003e-06, "loss": 0.0408, "step": 40846 }, { "epoch": 2.1776936610332145, "grad_norm": 0.9656651616096497, "learning_rate": 1.8308000000000001e-06, "loss": 0.0205, "step": 40847 }, { "epoch": 2.1777469744628672, "grad_norm": 2.8673648834228516, "learning_rate": 1.8306e-06, "loss": 0.0492, "step": 40848 }, { "epoch": 2.17780028789252, "grad_norm": 0.9723789691925049, "learning_rate": 1.8304000000000003e-06, "loss": 0.0072, "step": 40849 }, { "epoch": 2.177853601322173, "grad_norm": 2.4707846641540527, "learning_rate": 1.8302e-06, "loss": 0.0073, "step": 40850 }, { "epoch": 2.177906914751826, "grad_norm": 4.912806034088135, "learning_rate": 1.83e-06, "loss": 0.0134, "step": 40851 }, { "epoch": 2.1779602281814787, "grad_norm": 1.3766567707061768, "learning_rate": 1.8298000000000003e-06, "loss": 0.007, "step": 40852 }, { "epoch": 2.178013541611132, "grad_norm": 2.6340596675872803, "learning_rate": 1.8296000000000002e-06, "loss": 0.0304, "step": 40853 }, { "epoch": 2.1780668550407847, "grad_norm": 3.143735885620117, "learning_rate": 1.8294e-06, "loss": 0.0178, "step": 40854 }, { "epoch": 2.178120168470438, "grad_norm": 1.3307409286499023, "learning_rate": 1.8292000000000002e-06, "loss": 0.014, "step": 40855 }, { "epoch": 2.1781734819000906, "grad_norm": 0.566805899143219, "learning_rate": 1.8290000000000002e-06, "loss": 0.0043, "step": 40856 }, { "epoch": 2.1782267953297434, "grad_norm": 3.01320743560791, "learning_rate": 1.8288e-06, "loss": 0.0305, "step": 40857 }, { "epoch": 2.1782801087593966, "grad_norm": 2.0615522861480713, "learning_rate": 1.8286000000000002e-06, "loss": 0.0084, "step": 40858 }, { "epoch": 2.1783334221890494, "grad_norm": 1.5765712261199951, "learning_rate": 1.8284000000000001e-06, "loss": 0.0318, "step": 40859 }, { "epoch": 2.1783867356187026, "grad_norm": 1.721794605255127, "learning_rate": 1.8282000000000003e-06, "loss": 0.034, "step": 40860 }, { "epoch": 2.1784400490483553, "grad_norm": 2.5030128955841064, "learning_rate": 1.8280000000000001e-06, "loss": 0.0245, "step": 40861 }, { "epoch": 2.178493362478008, "grad_norm": 2.2867860794067383, "learning_rate": 1.8278e-06, "loss": 0.0084, "step": 40862 }, { "epoch": 2.1785466759076613, "grad_norm": 1.8225276470184326, "learning_rate": 1.8276000000000003e-06, "loss": 0.0378, "step": 40863 }, { "epoch": 2.178599989337314, "grad_norm": 0.5727112293243408, "learning_rate": 1.8274000000000003e-06, "loss": 0.0042, "step": 40864 }, { "epoch": 2.178653302766967, "grad_norm": 0.5426995158195496, "learning_rate": 1.8272e-06, "loss": 0.004, "step": 40865 }, { "epoch": 2.17870661619662, "grad_norm": 2.7563400268554688, "learning_rate": 1.8270000000000003e-06, "loss": 0.0259, "step": 40866 }, { "epoch": 2.1787599296262727, "grad_norm": 0.40386638045310974, "learning_rate": 1.8268000000000002e-06, "loss": 0.0024, "step": 40867 }, { "epoch": 2.178813243055926, "grad_norm": 0.9953258037567139, "learning_rate": 1.8266e-06, "loss": 0.004, "step": 40868 }, { "epoch": 2.1788665564855787, "grad_norm": 1.4274044036865234, "learning_rate": 1.8264000000000002e-06, "loss": 0.0083, "step": 40869 }, { "epoch": 2.1789198699152315, "grad_norm": 1.441877841949463, "learning_rate": 1.8262000000000002e-06, "loss": 0.0089, "step": 40870 }, { "epoch": 2.1789731833448847, "grad_norm": 1.6736210584640503, "learning_rate": 1.8260000000000002e-06, "loss": 0.009, "step": 40871 }, { "epoch": 2.1790264967745374, "grad_norm": 2.345717668533325, "learning_rate": 1.8258000000000002e-06, "loss": 0.0207, "step": 40872 }, { "epoch": 2.1790798102041906, "grad_norm": 1.5774180889129639, "learning_rate": 1.8256000000000001e-06, "loss": 0.0085, "step": 40873 }, { "epoch": 2.1791331236338434, "grad_norm": 1.3075225353240967, "learning_rate": 1.8254000000000001e-06, "loss": 0.0256, "step": 40874 }, { "epoch": 2.179186437063496, "grad_norm": 1.0900064706802368, "learning_rate": 1.8252000000000001e-06, "loss": 0.0068, "step": 40875 }, { "epoch": 2.1792397504931493, "grad_norm": 2.013645648956299, "learning_rate": 1.825e-06, "loss": 0.0222, "step": 40876 }, { "epoch": 2.179293063922802, "grad_norm": 1.6217949390411377, "learning_rate": 1.8248e-06, "loss": 0.0288, "step": 40877 }, { "epoch": 2.1793463773524553, "grad_norm": 1.6865041255950928, "learning_rate": 1.8246000000000003e-06, "loss": 0.0174, "step": 40878 }, { "epoch": 2.179399690782108, "grad_norm": 1.3759864568710327, "learning_rate": 1.8244e-06, "loss": 0.0117, "step": 40879 }, { "epoch": 2.179453004211761, "grad_norm": 2.1242191791534424, "learning_rate": 1.8242e-06, "loss": 0.0076, "step": 40880 }, { "epoch": 2.179506317641414, "grad_norm": 1.7776589393615723, "learning_rate": 1.8240000000000002e-06, "loss": 0.0179, "step": 40881 }, { "epoch": 2.1795596310710668, "grad_norm": 1.2322359085083008, "learning_rate": 1.8238e-06, "loss": 0.0077, "step": 40882 }, { "epoch": 2.17961294450072, "grad_norm": 4.268925666809082, "learning_rate": 1.8236e-06, "loss": 0.0241, "step": 40883 }, { "epoch": 2.1796662579303727, "grad_norm": 1.6706689596176147, "learning_rate": 1.8234000000000002e-06, "loss": 0.0056, "step": 40884 }, { "epoch": 2.1797195713600255, "grad_norm": 1.2575678825378418, "learning_rate": 1.8232000000000002e-06, "loss": 0.0272, "step": 40885 }, { "epoch": 2.1797728847896787, "grad_norm": 1.6976984739303589, "learning_rate": 1.823e-06, "loss": 0.0054, "step": 40886 }, { "epoch": 2.1798261982193314, "grad_norm": 2.9958014488220215, "learning_rate": 1.8228000000000001e-06, "loss": 0.0258, "step": 40887 }, { "epoch": 2.179879511648984, "grad_norm": 1.8285725116729736, "learning_rate": 1.8226000000000001e-06, "loss": 0.0209, "step": 40888 }, { "epoch": 2.1799328250786374, "grad_norm": 0.910434365272522, "learning_rate": 1.8224000000000003e-06, "loss": 0.0047, "step": 40889 }, { "epoch": 2.17998613850829, "grad_norm": 3.958484172821045, "learning_rate": 1.8222e-06, "loss": 0.0172, "step": 40890 }, { "epoch": 2.1800394519379434, "grad_norm": 0.3451923131942749, "learning_rate": 1.822e-06, "loss": 0.0028, "step": 40891 }, { "epoch": 2.180092765367596, "grad_norm": 2.382845640182495, "learning_rate": 1.8218000000000003e-06, "loss": 0.0241, "step": 40892 }, { "epoch": 2.180146078797249, "grad_norm": 3.637192726135254, "learning_rate": 1.8216e-06, "loss": 0.0144, "step": 40893 }, { "epoch": 2.180199392226902, "grad_norm": 0.9772976040840149, "learning_rate": 1.8214e-06, "loss": 0.0166, "step": 40894 }, { "epoch": 2.180252705656555, "grad_norm": 1.690968632698059, "learning_rate": 1.8212000000000002e-06, "loss": 0.0324, "step": 40895 }, { "epoch": 2.180306019086208, "grad_norm": 2.8522119522094727, "learning_rate": 1.8210000000000002e-06, "loss": 0.0251, "step": 40896 }, { "epoch": 2.180359332515861, "grad_norm": 1.4497911930084229, "learning_rate": 1.8208e-06, "loss": 0.0087, "step": 40897 }, { "epoch": 2.1804126459455135, "grad_norm": 0.6479260325431824, "learning_rate": 1.8206000000000002e-06, "loss": 0.0038, "step": 40898 }, { "epoch": 2.1804659593751667, "grad_norm": 1.901938557624817, "learning_rate": 1.8204000000000002e-06, "loss": 0.0284, "step": 40899 }, { "epoch": 2.1805192728048195, "grad_norm": 2.3181562423706055, "learning_rate": 1.8202e-06, "loss": 0.036, "step": 40900 }, { "epoch": 2.1805725862344723, "grad_norm": 1.9888544082641602, "learning_rate": 1.8200000000000002e-06, "loss": 0.0131, "step": 40901 }, { "epoch": 2.1806258996641255, "grad_norm": 1.3149515390396118, "learning_rate": 1.8198000000000001e-06, "loss": 0.0307, "step": 40902 }, { "epoch": 2.180679213093778, "grad_norm": 0.7562174201011658, "learning_rate": 1.8196000000000003e-06, "loss": 0.0041, "step": 40903 }, { "epoch": 2.1807325265234314, "grad_norm": 2.61712908744812, "learning_rate": 1.8194e-06, "loss": 0.0326, "step": 40904 }, { "epoch": 2.180785839953084, "grad_norm": 0.9478219747543335, "learning_rate": 1.8192e-06, "loss": 0.0052, "step": 40905 }, { "epoch": 2.180839153382737, "grad_norm": 1.1484509706497192, "learning_rate": 1.8190000000000003e-06, "loss": 0.0149, "step": 40906 }, { "epoch": 2.18089246681239, "grad_norm": 0.6236213445663452, "learning_rate": 1.8188e-06, "loss": 0.004, "step": 40907 }, { "epoch": 2.180945780242043, "grad_norm": 1.240849256515503, "learning_rate": 1.8186e-06, "loss": 0.0143, "step": 40908 }, { "epoch": 2.180999093671696, "grad_norm": 1.090516209602356, "learning_rate": 1.8184000000000002e-06, "loss": 0.0159, "step": 40909 }, { "epoch": 2.181052407101349, "grad_norm": 1.1396180391311646, "learning_rate": 1.8182000000000002e-06, "loss": 0.0048, "step": 40910 }, { "epoch": 2.1811057205310016, "grad_norm": 2.4548237323760986, "learning_rate": 1.818e-06, "loss": 0.0083, "step": 40911 }, { "epoch": 2.181159033960655, "grad_norm": 0.7036688923835754, "learning_rate": 1.8178000000000002e-06, "loss": 0.0036, "step": 40912 }, { "epoch": 2.1812123473903076, "grad_norm": 1.3746641874313354, "learning_rate": 1.8176000000000002e-06, "loss": 0.0287, "step": 40913 }, { "epoch": 2.1812656608199603, "grad_norm": 2.4727540016174316, "learning_rate": 1.8174000000000004e-06, "loss": 0.0107, "step": 40914 }, { "epoch": 2.1813189742496135, "grad_norm": 0.8313015103340149, "learning_rate": 1.8172000000000002e-06, "loss": 0.0038, "step": 40915 }, { "epoch": 2.1813722876792663, "grad_norm": 2.2155511379241943, "learning_rate": 1.8170000000000001e-06, "loss": 0.0166, "step": 40916 }, { "epoch": 2.1814256011089195, "grad_norm": 2.6071243286132812, "learning_rate": 1.8168000000000003e-06, "loss": 0.0146, "step": 40917 }, { "epoch": 2.1814789145385722, "grad_norm": 2.659231424331665, "learning_rate": 1.8166000000000001e-06, "loss": 0.0403, "step": 40918 }, { "epoch": 2.181532227968225, "grad_norm": 2.661902904510498, "learning_rate": 1.8164e-06, "loss": 0.0068, "step": 40919 }, { "epoch": 2.181585541397878, "grad_norm": 0.15545174479484558, "learning_rate": 1.8162000000000003e-06, "loss": 0.0015, "step": 40920 }, { "epoch": 2.181638854827531, "grad_norm": 1.629865288734436, "learning_rate": 1.8160000000000003e-06, "loss": 0.0094, "step": 40921 }, { "epoch": 2.181692168257184, "grad_norm": 2.949695587158203, "learning_rate": 1.8158e-06, "loss": 0.0143, "step": 40922 }, { "epoch": 2.181745481686837, "grad_norm": 0.522390604019165, "learning_rate": 1.8156000000000002e-06, "loss": 0.0035, "step": 40923 }, { "epoch": 2.1817987951164897, "grad_norm": 1.0412423610687256, "learning_rate": 1.8154000000000002e-06, "loss": 0.0164, "step": 40924 }, { "epoch": 2.181852108546143, "grad_norm": 0.5143808722496033, "learning_rate": 1.8152e-06, "loss": 0.0045, "step": 40925 }, { "epoch": 2.1819054219757956, "grad_norm": 2.1915602684020996, "learning_rate": 1.8150000000000002e-06, "loss": 0.0223, "step": 40926 }, { "epoch": 2.181958735405449, "grad_norm": 1.886526107788086, "learning_rate": 1.8148000000000002e-06, "loss": 0.0098, "step": 40927 }, { "epoch": 2.1820120488351016, "grad_norm": 1.3217538595199585, "learning_rate": 1.8146000000000002e-06, "loss": 0.0054, "step": 40928 }, { "epoch": 2.1820653622647543, "grad_norm": 1.2420592308044434, "learning_rate": 1.8144000000000002e-06, "loss": 0.0112, "step": 40929 }, { "epoch": 2.1821186756944075, "grad_norm": 2.9957211017608643, "learning_rate": 1.8142000000000001e-06, "loss": 0.0081, "step": 40930 }, { "epoch": 2.1821719891240603, "grad_norm": 1.611504316329956, "learning_rate": 1.8140000000000001e-06, "loss": 0.0062, "step": 40931 }, { "epoch": 2.1822253025537135, "grad_norm": 1.782680869102478, "learning_rate": 1.8138000000000003e-06, "loss": 0.0156, "step": 40932 }, { "epoch": 2.1822786159833663, "grad_norm": 0.8572654128074646, "learning_rate": 1.8136e-06, "loss": 0.0038, "step": 40933 }, { "epoch": 2.182331929413019, "grad_norm": 0.33484524488449097, "learning_rate": 1.8134e-06, "loss": 0.0029, "step": 40934 }, { "epoch": 2.182385242842672, "grad_norm": 2.588947057723999, "learning_rate": 1.8132000000000003e-06, "loss": 0.0133, "step": 40935 }, { "epoch": 2.182438556272325, "grad_norm": 1.8390570878982544, "learning_rate": 1.813e-06, "loss": 0.0113, "step": 40936 }, { "epoch": 2.1824918697019777, "grad_norm": 5.276271343231201, "learning_rate": 1.8128e-06, "loss": 0.0294, "step": 40937 }, { "epoch": 2.182545183131631, "grad_norm": 1.8588955402374268, "learning_rate": 1.8126000000000002e-06, "loss": 0.0246, "step": 40938 }, { "epoch": 2.1825984965612837, "grad_norm": 3.7568864822387695, "learning_rate": 1.8124000000000002e-06, "loss": 0.0148, "step": 40939 }, { "epoch": 2.182651809990937, "grad_norm": 2.057476758956909, "learning_rate": 1.8122e-06, "loss": 0.0075, "step": 40940 }, { "epoch": 2.1827051234205896, "grad_norm": 1.7225720882415771, "learning_rate": 1.8120000000000002e-06, "loss": 0.006, "step": 40941 }, { "epoch": 2.1827584368502424, "grad_norm": 0.35868915915489197, "learning_rate": 1.8118000000000002e-06, "loss": 0.0022, "step": 40942 }, { "epoch": 2.1828117502798956, "grad_norm": 1.7548328638076782, "learning_rate": 1.8116e-06, "loss": 0.016, "step": 40943 }, { "epoch": 2.1828650637095484, "grad_norm": 1.4752792119979858, "learning_rate": 1.8114000000000001e-06, "loss": 0.0072, "step": 40944 }, { "epoch": 2.1829183771392016, "grad_norm": 6.493565082550049, "learning_rate": 1.8112000000000001e-06, "loss": 0.0182, "step": 40945 }, { "epoch": 2.1829716905688543, "grad_norm": 1.9223911762237549, "learning_rate": 1.8110000000000003e-06, "loss": 0.0127, "step": 40946 }, { "epoch": 2.183025003998507, "grad_norm": 6.628808498382568, "learning_rate": 1.8108e-06, "loss": 0.036, "step": 40947 }, { "epoch": 2.1830783174281603, "grad_norm": 5.317120552062988, "learning_rate": 1.8106e-06, "loss": 0.0183, "step": 40948 }, { "epoch": 2.183131630857813, "grad_norm": 1.1062201261520386, "learning_rate": 1.8104000000000003e-06, "loss": 0.0063, "step": 40949 }, { "epoch": 2.183184944287466, "grad_norm": 4.050532817840576, "learning_rate": 1.8102e-06, "loss": 0.0212, "step": 40950 }, { "epoch": 2.183238257717119, "grad_norm": 2.9207077026367188, "learning_rate": 1.81e-06, "loss": 0.0289, "step": 40951 }, { "epoch": 2.1832915711467717, "grad_norm": 2.7773354053497314, "learning_rate": 1.8098000000000002e-06, "loss": 0.0165, "step": 40952 }, { "epoch": 2.183344884576425, "grad_norm": 3.8000078201293945, "learning_rate": 1.8096000000000002e-06, "loss": 0.0241, "step": 40953 }, { "epoch": 2.1833981980060777, "grad_norm": 5.2910075187683105, "learning_rate": 1.8094e-06, "loss": 0.0153, "step": 40954 }, { "epoch": 2.1834515114357305, "grad_norm": 3.48569655418396, "learning_rate": 1.8092000000000002e-06, "loss": 0.0116, "step": 40955 }, { "epoch": 2.1835048248653837, "grad_norm": 1.9413297176361084, "learning_rate": 1.8090000000000002e-06, "loss": 0.0211, "step": 40956 }, { "epoch": 2.1835581382950364, "grad_norm": 3.5095012187957764, "learning_rate": 1.8088000000000004e-06, "loss": 0.0226, "step": 40957 }, { "epoch": 2.1836114517246896, "grad_norm": 2.899566888809204, "learning_rate": 1.8086000000000001e-06, "loss": 0.0486, "step": 40958 }, { "epoch": 2.1836647651543424, "grad_norm": 0.27463170886039734, "learning_rate": 1.8084000000000001e-06, "loss": 0.002, "step": 40959 }, { "epoch": 2.183718078583995, "grad_norm": 1.5835151672363281, "learning_rate": 1.8082000000000003e-06, "loss": 0.0187, "step": 40960 }, { "epoch": 2.1837713920136483, "grad_norm": 1.536820411682129, "learning_rate": 1.808e-06, "loss": 0.0184, "step": 40961 }, { "epoch": 2.183824705443301, "grad_norm": 1.9596244096755981, "learning_rate": 1.8078e-06, "loss": 0.016, "step": 40962 }, { "epoch": 2.1838780188729543, "grad_norm": 1.8966425657272339, "learning_rate": 1.8076000000000003e-06, "loss": 0.0198, "step": 40963 }, { "epoch": 2.183931332302607, "grad_norm": 2.3526082038879395, "learning_rate": 1.8074000000000003e-06, "loss": 0.0287, "step": 40964 }, { "epoch": 2.18398464573226, "grad_norm": 3.0100057125091553, "learning_rate": 1.8072e-06, "loss": 0.0222, "step": 40965 }, { "epoch": 2.184037959161913, "grad_norm": 0.7210564017295837, "learning_rate": 1.8070000000000002e-06, "loss": 0.0037, "step": 40966 }, { "epoch": 2.1840912725915658, "grad_norm": 2.9031171798706055, "learning_rate": 1.8068000000000002e-06, "loss": 0.0352, "step": 40967 }, { "epoch": 2.184144586021219, "grad_norm": 0.9301746487617493, "learning_rate": 1.8066e-06, "loss": 0.0183, "step": 40968 }, { "epoch": 2.1841978994508717, "grad_norm": 1.1457651853561401, "learning_rate": 1.8064000000000002e-06, "loss": 0.0051, "step": 40969 }, { "epoch": 2.1842512128805245, "grad_norm": 1.3830726146697998, "learning_rate": 1.8062000000000002e-06, "loss": 0.0244, "step": 40970 }, { "epoch": 2.1843045263101777, "grad_norm": 1.5572987794876099, "learning_rate": 1.8060000000000002e-06, "loss": 0.0111, "step": 40971 }, { "epoch": 2.1843578397398304, "grad_norm": 0.4824226498603821, "learning_rate": 1.8058000000000001e-06, "loss": 0.0028, "step": 40972 }, { "epoch": 2.184411153169483, "grad_norm": 0.7574489116668701, "learning_rate": 1.8056000000000001e-06, "loss": 0.0046, "step": 40973 }, { "epoch": 2.1844644665991364, "grad_norm": 1.982011318206787, "learning_rate": 1.8054000000000001e-06, "loss": 0.0188, "step": 40974 }, { "epoch": 2.184517780028789, "grad_norm": 2.59246826171875, "learning_rate": 1.8052000000000003e-06, "loss": 0.0225, "step": 40975 }, { "epoch": 2.1845710934584424, "grad_norm": 0.6622364521026611, "learning_rate": 1.805e-06, "loss": 0.0036, "step": 40976 }, { "epoch": 2.184624406888095, "grad_norm": 0.29744958877563477, "learning_rate": 1.8048e-06, "loss": 0.0024, "step": 40977 }, { "epoch": 2.184677720317748, "grad_norm": 1.2074209451675415, "learning_rate": 1.8046000000000003e-06, "loss": 0.0052, "step": 40978 }, { "epoch": 2.184731033747401, "grad_norm": 1.7084529399871826, "learning_rate": 1.8044e-06, "loss": 0.02, "step": 40979 }, { "epoch": 2.184784347177054, "grad_norm": 1.7479161024093628, "learning_rate": 1.8042e-06, "loss": 0.0091, "step": 40980 }, { "epoch": 2.184837660606707, "grad_norm": 2.994197368621826, "learning_rate": 1.8040000000000002e-06, "loss": 0.0374, "step": 40981 }, { "epoch": 2.18489097403636, "grad_norm": 2.1224350929260254, "learning_rate": 1.8038000000000002e-06, "loss": 0.0106, "step": 40982 }, { "epoch": 2.1849442874660125, "grad_norm": 3.249011754989624, "learning_rate": 1.8036e-06, "loss": 0.0142, "step": 40983 }, { "epoch": 2.1849976008956657, "grad_norm": 2.1495277881622314, "learning_rate": 1.8034000000000002e-06, "loss": 0.0281, "step": 40984 }, { "epoch": 2.1850509143253185, "grad_norm": 3.1944539546966553, "learning_rate": 1.8032000000000002e-06, "loss": 0.0181, "step": 40985 }, { "epoch": 2.1851042277549713, "grad_norm": 2.6314187049865723, "learning_rate": 1.803e-06, "loss": 0.0128, "step": 40986 }, { "epoch": 2.1851575411846245, "grad_norm": 1.0145961046218872, "learning_rate": 1.8028000000000001e-06, "loss": 0.0055, "step": 40987 }, { "epoch": 2.185210854614277, "grad_norm": 3.7441771030426025, "learning_rate": 1.8026000000000001e-06, "loss": 0.0231, "step": 40988 }, { "epoch": 2.1852641680439304, "grad_norm": 0.8140287399291992, "learning_rate": 1.8024000000000003e-06, "loss": 0.0151, "step": 40989 }, { "epoch": 2.185317481473583, "grad_norm": 0.850058376789093, "learning_rate": 1.8022e-06, "loss": 0.006, "step": 40990 }, { "epoch": 2.185370794903236, "grad_norm": 1.244997262954712, "learning_rate": 1.802e-06, "loss": 0.0052, "step": 40991 }, { "epoch": 2.185424108332889, "grad_norm": 2.6335606575012207, "learning_rate": 1.8018000000000003e-06, "loss": 0.0177, "step": 40992 }, { "epoch": 2.185477421762542, "grad_norm": 1.6997541189193726, "learning_rate": 1.8016e-06, "loss": 0.0226, "step": 40993 }, { "epoch": 2.185530735192195, "grad_norm": 2.4950950145721436, "learning_rate": 1.8014e-06, "loss": 0.0284, "step": 40994 }, { "epoch": 2.185584048621848, "grad_norm": 1.5019910335540771, "learning_rate": 1.8012000000000002e-06, "loss": 0.0101, "step": 40995 }, { "epoch": 2.1856373620515006, "grad_norm": 5.245087146759033, "learning_rate": 1.8010000000000002e-06, "loss": 0.0268, "step": 40996 }, { "epoch": 2.185690675481154, "grad_norm": 0.6883662343025208, "learning_rate": 1.8008e-06, "loss": 0.0045, "step": 40997 }, { "epoch": 2.1857439889108066, "grad_norm": 4.046586513519287, "learning_rate": 1.8006000000000002e-06, "loss": 0.0551, "step": 40998 }, { "epoch": 2.1857973023404598, "grad_norm": 2.3524160385131836, "learning_rate": 1.8004000000000002e-06, "loss": 0.0413, "step": 40999 }, { "epoch": 2.1858506157701125, "grad_norm": 2.0735671520233154, "learning_rate": 1.8002000000000004e-06, "loss": 0.0211, "step": 41000 }, { "epoch": 2.1859039291997653, "grad_norm": 6.5423150062561035, "learning_rate": 1.8000000000000001e-06, "loss": 0.0423, "step": 41001 }, { "epoch": 2.1859572426294185, "grad_norm": 0.9055016040802002, "learning_rate": 1.7998000000000001e-06, "loss": 0.0049, "step": 41002 }, { "epoch": 2.1860105560590712, "grad_norm": 2.0953545570373535, "learning_rate": 1.7996000000000003e-06, "loss": 0.0201, "step": 41003 }, { "epoch": 2.1860638694887244, "grad_norm": 1.9927363395690918, "learning_rate": 1.7994e-06, "loss": 0.0251, "step": 41004 }, { "epoch": 2.186117182918377, "grad_norm": 1.0304352045059204, "learning_rate": 1.7992e-06, "loss": 0.0036, "step": 41005 }, { "epoch": 2.18617049634803, "grad_norm": 1.9691355228424072, "learning_rate": 1.7990000000000003e-06, "loss": 0.028, "step": 41006 }, { "epoch": 2.186223809777683, "grad_norm": 3.9519925117492676, "learning_rate": 1.7988000000000003e-06, "loss": 0.037, "step": 41007 }, { "epoch": 2.186277123207336, "grad_norm": 1.9147660732269287, "learning_rate": 1.7986e-06, "loss": 0.0079, "step": 41008 }, { "epoch": 2.1863304366369887, "grad_norm": 3.0811140537261963, "learning_rate": 1.7984000000000002e-06, "loss": 0.009, "step": 41009 }, { "epoch": 2.186383750066642, "grad_norm": 2.3348677158355713, "learning_rate": 1.7982000000000002e-06, "loss": 0.0103, "step": 41010 }, { "epoch": 2.1864370634962946, "grad_norm": 7.128660202026367, "learning_rate": 1.798e-06, "loss": 0.0339, "step": 41011 }, { "epoch": 2.186490376925948, "grad_norm": 1.0622832775115967, "learning_rate": 1.7978000000000002e-06, "loss": 0.0067, "step": 41012 }, { "epoch": 2.1865436903556006, "grad_norm": 1.8405994176864624, "learning_rate": 1.7976000000000002e-06, "loss": 0.0488, "step": 41013 }, { "epoch": 2.1865970037852533, "grad_norm": 1.921665072441101, "learning_rate": 1.7974000000000004e-06, "loss": 0.0609, "step": 41014 }, { "epoch": 2.1866503172149065, "grad_norm": 2.8286261558532715, "learning_rate": 1.7972000000000001e-06, "loss": 0.0342, "step": 41015 }, { "epoch": 2.1867036306445593, "grad_norm": 0.2569833993911743, "learning_rate": 1.7970000000000001e-06, "loss": 0.0023, "step": 41016 }, { "epoch": 2.1867569440742125, "grad_norm": 0.8197492957115173, "learning_rate": 1.7968000000000003e-06, "loss": 0.006, "step": 41017 }, { "epoch": 2.1868102575038653, "grad_norm": 3.223634958267212, "learning_rate": 1.7966e-06, "loss": 0.0361, "step": 41018 }, { "epoch": 2.186863570933518, "grad_norm": 2.3827946186065674, "learning_rate": 1.7964e-06, "loss": 0.0126, "step": 41019 }, { "epoch": 2.186916884363171, "grad_norm": 2.044532537460327, "learning_rate": 1.7962000000000003e-06, "loss": 0.0255, "step": 41020 }, { "epoch": 2.186970197792824, "grad_norm": 0.38967761397361755, "learning_rate": 1.7960000000000003e-06, "loss": 0.0031, "step": 41021 }, { "epoch": 2.1870235112224767, "grad_norm": 0.975648820400238, "learning_rate": 1.7958e-06, "loss": 0.0047, "step": 41022 }, { "epoch": 2.18707682465213, "grad_norm": 3.0461039543151855, "learning_rate": 1.7956000000000002e-06, "loss": 0.0366, "step": 41023 }, { "epoch": 2.1871301380817827, "grad_norm": 2.168407440185547, "learning_rate": 1.7954000000000002e-06, "loss": 0.0108, "step": 41024 }, { "epoch": 2.187183451511436, "grad_norm": 3.197753667831421, "learning_rate": 1.7952000000000002e-06, "loss": 0.0198, "step": 41025 }, { "epoch": 2.1872367649410887, "grad_norm": 1.4517552852630615, "learning_rate": 1.7950000000000002e-06, "loss": 0.0074, "step": 41026 }, { "epoch": 2.1872900783707414, "grad_norm": 0.4592623710632324, "learning_rate": 1.7948000000000002e-06, "loss": 0.0037, "step": 41027 }, { "epoch": 2.1873433918003946, "grad_norm": 0.9976081848144531, "learning_rate": 1.7946000000000001e-06, "loss": 0.0043, "step": 41028 }, { "epoch": 2.1873967052300474, "grad_norm": 1.9838558435440063, "learning_rate": 1.7944000000000001e-06, "loss": 0.0101, "step": 41029 }, { "epoch": 2.1874500186597006, "grad_norm": 2.191509246826172, "learning_rate": 1.7942000000000001e-06, "loss": 0.0312, "step": 41030 }, { "epoch": 2.1875033320893533, "grad_norm": 3.095243215560913, "learning_rate": 1.794e-06, "loss": 0.0246, "step": 41031 }, { "epoch": 2.187556645519006, "grad_norm": 2.0086541175842285, "learning_rate": 1.7938000000000003e-06, "loss": 0.0136, "step": 41032 }, { "epoch": 2.1876099589486593, "grad_norm": 4.8782958984375, "learning_rate": 1.7936e-06, "loss": 0.0122, "step": 41033 }, { "epoch": 2.187663272378312, "grad_norm": 1.4328198432922363, "learning_rate": 1.7934e-06, "loss": 0.0091, "step": 41034 }, { "epoch": 2.187716585807965, "grad_norm": 1.4610652923583984, "learning_rate": 1.7932000000000003e-06, "loss": 0.0122, "step": 41035 }, { "epoch": 2.187769899237618, "grad_norm": 1.5536173582077026, "learning_rate": 1.793e-06, "loss": 0.0262, "step": 41036 }, { "epoch": 2.1878232126672708, "grad_norm": 1.4966111183166504, "learning_rate": 1.7928e-06, "loss": 0.0072, "step": 41037 }, { "epoch": 2.187876526096924, "grad_norm": 2.1774871349334717, "learning_rate": 1.7926000000000002e-06, "loss": 0.03, "step": 41038 }, { "epoch": 2.1879298395265767, "grad_norm": 1.428421974182129, "learning_rate": 1.7924000000000002e-06, "loss": 0.0211, "step": 41039 }, { "epoch": 2.1879831529562295, "grad_norm": 1.9494415521621704, "learning_rate": 1.7922e-06, "loss": 0.049, "step": 41040 }, { "epoch": 2.1880364663858827, "grad_norm": 2.48465895652771, "learning_rate": 1.7920000000000002e-06, "loss": 0.02, "step": 41041 }, { "epoch": 2.1880897798155354, "grad_norm": 2.157653570175171, "learning_rate": 1.7918000000000001e-06, "loss": 0.0336, "step": 41042 }, { "epoch": 2.1881430932451886, "grad_norm": 0.8514837622642517, "learning_rate": 1.7916000000000003e-06, "loss": 0.0045, "step": 41043 }, { "epoch": 2.1881964066748414, "grad_norm": 1.791255235671997, "learning_rate": 1.7914000000000001e-06, "loss": 0.0173, "step": 41044 }, { "epoch": 2.188249720104494, "grad_norm": 2.8908891677856445, "learning_rate": 1.7912e-06, "loss": 0.0256, "step": 41045 }, { "epoch": 2.1883030335341473, "grad_norm": 1.264454960823059, "learning_rate": 1.7910000000000003e-06, "loss": 0.0093, "step": 41046 }, { "epoch": 2.1883563469638, "grad_norm": 1.5925010442733765, "learning_rate": 1.7908e-06, "loss": 0.0083, "step": 41047 }, { "epoch": 2.1884096603934533, "grad_norm": 2.852769613265991, "learning_rate": 1.7906e-06, "loss": 0.0245, "step": 41048 }, { "epoch": 2.188462973823106, "grad_norm": 2.2124016284942627, "learning_rate": 1.7904000000000003e-06, "loss": 0.022, "step": 41049 }, { "epoch": 2.188516287252759, "grad_norm": 1.2905348539352417, "learning_rate": 1.7902000000000002e-06, "loss": 0.0183, "step": 41050 }, { "epoch": 2.188569600682412, "grad_norm": 1.1904953718185425, "learning_rate": 1.79e-06, "loss": 0.015, "step": 41051 }, { "epoch": 2.1886229141120648, "grad_norm": 0.6848445534706116, "learning_rate": 1.7898000000000002e-06, "loss": 0.0039, "step": 41052 }, { "epoch": 2.188676227541718, "grad_norm": 1.864640712738037, "learning_rate": 1.7896000000000002e-06, "loss": 0.0097, "step": 41053 }, { "epoch": 2.1887295409713707, "grad_norm": 0.9370583295822144, "learning_rate": 1.7894e-06, "loss": 0.0135, "step": 41054 }, { "epoch": 2.1887828544010235, "grad_norm": 1.573201298713684, "learning_rate": 1.7892000000000002e-06, "loss": 0.0112, "step": 41055 }, { "epoch": 2.1888361678306767, "grad_norm": 3.5740973949432373, "learning_rate": 1.7890000000000002e-06, "loss": 0.0191, "step": 41056 }, { "epoch": 2.1888894812603295, "grad_norm": 2.2108051776885986, "learning_rate": 1.7888000000000003e-06, "loss": 0.0181, "step": 41057 }, { "epoch": 2.188942794689982, "grad_norm": 3.337371349334717, "learning_rate": 1.7886000000000001e-06, "loss": 0.0553, "step": 41058 }, { "epoch": 2.1889961081196354, "grad_norm": 1.717271327972412, "learning_rate": 1.7884e-06, "loss": 0.0183, "step": 41059 }, { "epoch": 2.189049421549288, "grad_norm": 1.5664052963256836, "learning_rate": 1.7882000000000003e-06, "loss": 0.0318, "step": 41060 }, { "epoch": 2.1891027349789414, "grad_norm": 2.22389554977417, "learning_rate": 1.788e-06, "loss": 0.0097, "step": 41061 }, { "epoch": 2.189156048408594, "grad_norm": 1.6017539501190186, "learning_rate": 1.7878e-06, "loss": 0.0088, "step": 41062 }, { "epoch": 2.189209361838247, "grad_norm": 1.3560218811035156, "learning_rate": 1.7876000000000003e-06, "loss": 0.0063, "step": 41063 }, { "epoch": 2.1892626752679, "grad_norm": 1.9737893342971802, "learning_rate": 1.7874000000000002e-06, "loss": 0.022, "step": 41064 }, { "epoch": 2.189315988697553, "grad_norm": 1.3396921157836914, "learning_rate": 1.7872e-06, "loss": 0.0103, "step": 41065 }, { "epoch": 2.189369302127206, "grad_norm": 0.6961163878440857, "learning_rate": 1.7870000000000002e-06, "loss": 0.0042, "step": 41066 }, { "epoch": 2.189422615556859, "grad_norm": 1.3575068712234497, "learning_rate": 1.7868000000000002e-06, "loss": 0.0113, "step": 41067 }, { "epoch": 2.1894759289865116, "grad_norm": 1.7117035388946533, "learning_rate": 1.7866000000000002e-06, "loss": 0.0156, "step": 41068 }, { "epoch": 2.1895292424161648, "grad_norm": 1.9431545734405518, "learning_rate": 1.7864000000000002e-06, "loss": 0.0227, "step": 41069 }, { "epoch": 2.1895825558458175, "grad_norm": 1.9100675582885742, "learning_rate": 1.7862000000000002e-06, "loss": 0.0197, "step": 41070 }, { "epoch": 2.1896358692754703, "grad_norm": 0.9356868267059326, "learning_rate": 1.7860000000000001e-06, "loss": 0.0129, "step": 41071 }, { "epoch": 2.1896891827051235, "grad_norm": 1.1786022186279297, "learning_rate": 1.7858000000000001e-06, "loss": 0.0145, "step": 41072 }, { "epoch": 2.1897424961347762, "grad_norm": 2.213867664337158, "learning_rate": 1.7856000000000001e-06, "loss": 0.0246, "step": 41073 }, { "epoch": 2.1897958095644294, "grad_norm": 1.8265153169631958, "learning_rate": 1.7854e-06, "loss": 0.0194, "step": 41074 }, { "epoch": 2.189849122994082, "grad_norm": 0.21659903228282928, "learning_rate": 1.7852000000000003e-06, "loss": 0.0016, "step": 41075 }, { "epoch": 2.189902436423735, "grad_norm": 1.4652396440505981, "learning_rate": 1.785e-06, "loss": 0.0078, "step": 41076 }, { "epoch": 2.189955749853388, "grad_norm": 1.1515083312988281, "learning_rate": 1.7848e-06, "loss": 0.0276, "step": 41077 }, { "epoch": 2.190009063283041, "grad_norm": 0.9906592965126038, "learning_rate": 1.7846000000000002e-06, "loss": 0.0057, "step": 41078 }, { "epoch": 2.190062376712694, "grad_norm": 2.6650118827819824, "learning_rate": 1.7844e-06, "loss": 0.0278, "step": 41079 }, { "epoch": 2.190115690142347, "grad_norm": 2.112351655960083, "learning_rate": 1.7842e-06, "loss": 0.0149, "step": 41080 }, { "epoch": 2.1901690035719996, "grad_norm": 3.7715811729431152, "learning_rate": 1.7840000000000002e-06, "loss": 0.0165, "step": 41081 }, { "epoch": 2.190222317001653, "grad_norm": 1.867218255996704, "learning_rate": 1.7838000000000002e-06, "loss": 0.0344, "step": 41082 }, { "epoch": 2.1902756304313056, "grad_norm": 0.6626426577568054, "learning_rate": 1.7836e-06, "loss": 0.0046, "step": 41083 }, { "epoch": 2.1903289438609588, "grad_norm": 1.9573935270309448, "learning_rate": 1.7834000000000002e-06, "loss": 0.03, "step": 41084 }, { "epoch": 2.1903822572906115, "grad_norm": 2.395737409591675, "learning_rate": 1.7832000000000001e-06, "loss": 0.0373, "step": 41085 }, { "epoch": 2.1904355707202643, "grad_norm": 0.7996866703033447, "learning_rate": 1.783e-06, "loss": 0.0051, "step": 41086 }, { "epoch": 2.1904888841499175, "grad_norm": 2.7155041694641113, "learning_rate": 1.7828000000000001e-06, "loss": 0.0182, "step": 41087 }, { "epoch": 2.1905421975795702, "grad_norm": 1.5489628314971924, "learning_rate": 1.7826e-06, "loss": 0.0078, "step": 41088 }, { "epoch": 2.1905955110092235, "grad_norm": 1.9013363122940063, "learning_rate": 1.7824000000000003e-06, "loss": 0.0156, "step": 41089 }, { "epoch": 2.190648824438876, "grad_norm": 1.587142825126648, "learning_rate": 1.7822e-06, "loss": 0.022, "step": 41090 }, { "epoch": 2.190702137868529, "grad_norm": 0.961311936378479, "learning_rate": 1.782e-06, "loss": 0.0066, "step": 41091 }, { "epoch": 2.190755451298182, "grad_norm": 0.7208766341209412, "learning_rate": 1.7818000000000002e-06, "loss": 0.0043, "step": 41092 }, { "epoch": 2.190808764727835, "grad_norm": 1.034173846244812, "learning_rate": 1.7816000000000002e-06, "loss": 0.0046, "step": 41093 }, { "epoch": 2.1908620781574877, "grad_norm": 3.02356219291687, "learning_rate": 1.7814e-06, "loss": 0.0376, "step": 41094 }, { "epoch": 2.190915391587141, "grad_norm": 1.7472153902053833, "learning_rate": 1.7812000000000002e-06, "loss": 0.009, "step": 41095 }, { "epoch": 2.1909687050167936, "grad_norm": 1.8418372869491577, "learning_rate": 1.7810000000000002e-06, "loss": 0.023, "step": 41096 }, { "epoch": 2.191022018446447, "grad_norm": 1.7525050640106201, "learning_rate": 1.7808e-06, "loss": 0.0317, "step": 41097 }, { "epoch": 2.1910753318760996, "grad_norm": 1.7288249731063843, "learning_rate": 1.7806000000000002e-06, "loss": 0.0156, "step": 41098 }, { "epoch": 2.1911286453057524, "grad_norm": 1.6331816911697388, "learning_rate": 1.7804000000000001e-06, "loss": 0.0068, "step": 41099 }, { "epoch": 2.1911819587354056, "grad_norm": 1.1945146322250366, "learning_rate": 1.7802000000000003e-06, "loss": 0.0124, "step": 41100 }, { "epoch": 2.1912352721650583, "grad_norm": 1.6020874977111816, "learning_rate": 1.7800000000000001e-06, "loss": 0.0256, "step": 41101 }, { "epoch": 2.1912885855947115, "grad_norm": 1.1582587957382202, "learning_rate": 1.7798e-06, "loss": 0.0195, "step": 41102 }, { "epoch": 2.1913418990243643, "grad_norm": 7.270473480224609, "learning_rate": 1.7796000000000003e-06, "loss": 0.0217, "step": 41103 }, { "epoch": 2.191395212454017, "grad_norm": 2.137205123901367, "learning_rate": 1.7794e-06, "loss": 0.0068, "step": 41104 }, { "epoch": 2.1914485258836702, "grad_norm": 1.7202436923980713, "learning_rate": 1.7792e-06, "loss": 0.0135, "step": 41105 }, { "epoch": 2.191501839313323, "grad_norm": 0.3708192706108093, "learning_rate": 1.7790000000000002e-06, "loss": 0.0024, "step": 41106 }, { "epoch": 2.1915551527429757, "grad_norm": 3.202016830444336, "learning_rate": 1.7788000000000002e-06, "loss": 0.0366, "step": 41107 }, { "epoch": 2.191608466172629, "grad_norm": 1.1903737783432007, "learning_rate": 1.7786e-06, "loss": 0.0146, "step": 41108 }, { "epoch": 2.1916617796022817, "grad_norm": 0.3387521505355835, "learning_rate": 1.7784000000000002e-06, "loss": 0.002, "step": 41109 }, { "epoch": 2.191715093031935, "grad_norm": 2.3853976726531982, "learning_rate": 1.7782000000000002e-06, "loss": 0.0127, "step": 41110 }, { "epoch": 2.1917684064615877, "grad_norm": 2.2130088806152344, "learning_rate": 1.7780000000000004e-06, "loss": 0.0089, "step": 41111 }, { "epoch": 2.1918217198912404, "grad_norm": 1.878933072090149, "learning_rate": 1.7778000000000002e-06, "loss": 0.0045, "step": 41112 }, { "epoch": 2.1918750333208936, "grad_norm": 0.610611081123352, "learning_rate": 1.7776000000000001e-06, "loss": 0.0044, "step": 41113 }, { "epoch": 2.1919283467505464, "grad_norm": 2.4288861751556396, "learning_rate": 1.7774000000000003e-06, "loss": 0.0122, "step": 41114 }, { "epoch": 2.1919816601801996, "grad_norm": 5.33041524887085, "learning_rate": 1.7772000000000001e-06, "loss": 0.0717, "step": 41115 }, { "epoch": 2.1920349736098523, "grad_norm": 0.5524731874465942, "learning_rate": 1.777e-06, "loss": 0.0035, "step": 41116 }, { "epoch": 2.192088287039505, "grad_norm": 0.6983784437179565, "learning_rate": 1.7768000000000003e-06, "loss": 0.0029, "step": 41117 }, { "epoch": 2.1921416004691583, "grad_norm": 1.84722900390625, "learning_rate": 1.7766000000000003e-06, "loss": 0.0157, "step": 41118 }, { "epoch": 2.192194913898811, "grad_norm": 1.7430452108383179, "learning_rate": 1.7764e-06, "loss": 0.0086, "step": 41119 }, { "epoch": 2.192248227328464, "grad_norm": 0.6969243884086609, "learning_rate": 1.7762000000000002e-06, "loss": 0.0027, "step": 41120 }, { "epoch": 2.192301540758117, "grad_norm": 1.3608912229537964, "learning_rate": 1.7760000000000002e-06, "loss": 0.0067, "step": 41121 }, { "epoch": 2.1923548541877698, "grad_norm": 1.6957225799560547, "learning_rate": 1.7758e-06, "loss": 0.0086, "step": 41122 }, { "epoch": 2.192408167617423, "grad_norm": 1.194181203842163, "learning_rate": 1.7756000000000002e-06, "loss": 0.012, "step": 41123 }, { "epoch": 2.1924614810470757, "grad_norm": 2.3656840324401855, "learning_rate": 1.7754000000000002e-06, "loss": 0.028, "step": 41124 }, { "epoch": 2.192514794476729, "grad_norm": 1.5732775926589966, "learning_rate": 1.7752000000000002e-06, "loss": 0.0085, "step": 41125 }, { "epoch": 2.1925681079063817, "grad_norm": 1.958668828010559, "learning_rate": 1.7750000000000002e-06, "loss": 0.0059, "step": 41126 }, { "epoch": 2.1926214213360344, "grad_norm": 0.9585924744606018, "learning_rate": 1.7748000000000001e-06, "loss": 0.0032, "step": 41127 }, { "epoch": 2.1926747347656876, "grad_norm": 0.5888788104057312, "learning_rate": 1.7746000000000001e-06, "loss": 0.0035, "step": 41128 }, { "epoch": 2.1927280481953404, "grad_norm": 0.8050803542137146, "learning_rate": 1.7744000000000001e-06, "loss": 0.0186, "step": 41129 }, { "epoch": 2.192781361624993, "grad_norm": 1.047584056854248, "learning_rate": 1.7742e-06, "loss": 0.0071, "step": 41130 }, { "epoch": 2.1928346750546464, "grad_norm": 1.6805105209350586, "learning_rate": 1.774e-06, "loss": 0.0101, "step": 41131 }, { "epoch": 2.192887988484299, "grad_norm": 2.9815139770507812, "learning_rate": 1.7738000000000003e-06, "loss": 0.0377, "step": 41132 }, { "epoch": 2.1929413019139523, "grad_norm": 2.8742129802703857, "learning_rate": 1.7736e-06, "loss": 0.0431, "step": 41133 }, { "epoch": 2.192994615343605, "grad_norm": 1.3307645320892334, "learning_rate": 1.7734e-06, "loss": 0.0124, "step": 41134 }, { "epoch": 2.193047928773258, "grad_norm": 2.600464105606079, "learning_rate": 1.7732000000000002e-06, "loss": 0.0336, "step": 41135 }, { "epoch": 2.193101242202911, "grad_norm": 2.118828058242798, "learning_rate": 1.7730000000000002e-06, "loss": 0.0275, "step": 41136 }, { "epoch": 2.193154555632564, "grad_norm": 1.1270074844360352, "learning_rate": 1.7728e-06, "loss": 0.0049, "step": 41137 }, { "epoch": 2.193207869062217, "grad_norm": 1.776335597038269, "learning_rate": 1.7726000000000002e-06, "loss": 0.0258, "step": 41138 }, { "epoch": 2.1932611824918697, "grad_norm": 1.5205787420272827, "learning_rate": 1.7724000000000002e-06, "loss": 0.0197, "step": 41139 }, { "epoch": 2.1933144959215225, "grad_norm": 1.4297511577606201, "learning_rate": 1.7722e-06, "loss": 0.0241, "step": 41140 }, { "epoch": 2.1933678093511757, "grad_norm": 3.8690240383148193, "learning_rate": 1.7720000000000001e-06, "loss": 0.0354, "step": 41141 }, { "epoch": 2.1934211227808285, "grad_norm": 1.009119987487793, "learning_rate": 1.7718000000000001e-06, "loss": 0.0119, "step": 41142 }, { "epoch": 2.193474436210481, "grad_norm": 1.2888582944869995, "learning_rate": 1.7716000000000003e-06, "loss": 0.0213, "step": 41143 }, { "epoch": 2.1935277496401344, "grad_norm": 1.3137502670288086, "learning_rate": 1.7714e-06, "loss": 0.0062, "step": 41144 }, { "epoch": 2.193581063069787, "grad_norm": 4.357871055603027, "learning_rate": 1.7712e-06, "loss": 0.0234, "step": 41145 }, { "epoch": 2.1936343764994404, "grad_norm": 2.195235013961792, "learning_rate": 1.7710000000000003e-06, "loss": 0.0235, "step": 41146 }, { "epoch": 2.193687689929093, "grad_norm": 1.3584753274917603, "learning_rate": 1.7708e-06, "loss": 0.0062, "step": 41147 }, { "epoch": 2.193741003358746, "grad_norm": 0.4307219684123993, "learning_rate": 1.7706e-06, "loss": 0.0037, "step": 41148 }, { "epoch": 2.193794316788399, "grad_norm": 1.6472212076187134, "learning_rate": 1.7704000000000002e-06, "loss": 0.0077, "step": 41149 }, { "epoch": 2.193847630218052, "grad_norm": 1.5717761516571045, "learning_rate": 1.7702000000000002e-06, "loss": 0.0112, "step": 41150 }, { "epoch": 2.193900943647705, "grad_norm": 6.168513774871826, "learning_rate": 1.77e-06, "loss": 0.0303, "step": 41151 }, { "epoch": 2.193954257077358, "grad_norm": 2.319553852081299, "learning_rate": 1.7698000000000002e-06, "loss": 0.0475, "step": 41152 }, { "epoch": 2.1940075705070106, "grad_norm": 1.124845266342163, "learning_rate": 1.7696000000000002e-06, "loss": 0.0175, "step": 41153 }, { "epoch": 2.1940608839366638, "grad_norm": 2.914266586303711, "learning_rate": 1.7694e-06, "loss": 0.0188, "step": 41154 }, { "epoch": 2.1941141973663165, "grad_norm": 2.622053384780884, "learning_rate": 1.7692000000000001e-06, "loss": 0.018, "step": 41155 }, { "epoch": 2.1941675107959693, "grad_norm": 0.7440510988235474, "learning_rate": 1.7690000000000001e-06, "loss": 0.0051, "step": 41156 }, { "epoch": 2.1942208242256225, "grad_norm": 2.854039192199707, "learning_rate": 1.7688000000000003e-06, "loss": 0.0675, "step": 41157 }, { "epoch": 2.1942741376552752, "grad_norm": 0.37813329696655273, "learning_rate": 1.7686e-06, "loss": 0.0024, "step": 41158 }, { "epoch": 2.1943274510849284, "grad_norm": 1.676882028579712, "learning_rate": 1.7684e-06, "loss": 0.0073, "step": 41159 }, { "epoch": 2.194380764514581, "grad_norm": 0.5178917646408081, "learning_rate": 1.7682000000000003e-06, "loss": 0.0037, "step": 41160 }, { "epoch": 2.194434077944234, "grad_norm": 1.4203858375549316, "learning_rate": 1.7680000000000003e-06, "loss": 0.0103, "step": 41161 }, { "epoch": 2.194487391373887, "grad_norm": 2.0017623901367188, "learning_rate": 1.7678e-06, "loss": 0.0096, "step": 41162 }, { "epoch": 2.19454070480354, "grad_norm": 1.1870654821395874, "learning_rate": 1.7676000000000002e-06, "loss": 0.0048, "step": 41163 }, { "epoch": 2.194594018233193, "grad_norm": 0.8381151556968689, "learning_rate": 1.7674000000000002e-06, "loss": 0.0053, "step": 41164 }, { "epoch": 2.194647331662846, "grad_norm": 2.825430154800415, "learning_rate": 1.7672e-06, "loss": 0.0258, "step": 41165 }, { "epoch": 2.1947006450924986, "grad_norm": 3.4184532165527344, "learning_rate": 1.7670000000000002e-06, "loss": 0.0111, "step": 41166 }, { "epoch": 2.194753958522152, "grad_norm": 2.2188620567321777, "learning_rate": 1.7668000000000002e-06, "loss": 0.0423, "step": 41167 }, { "epoch": 2.1948072719518046, "grad_norm": 2.350297689437866, "learning_rate": 1.7666000000000002e-06, "loss": 0.0163, "step": 41168 }, { "epoch": 2.194860585381458, "grad_norm": 0.2466381937265396, "learning_rate": 1.7664000000000001e-06, "loss": 0.0023, "step": 41169 }, { "epoch": 2.1949138988111105, "grad_norm": 1.8958196640014648, "learning_rate": 1.7662000000000001e-06, "loss": 0.0127, "step": 41170 }, { "epoch": 2.1949672122407633, "grad_norm": 0.5981401205062866, "learning_rate": 1.7660000000000001e-06, "loss": 0.0036, "step": 41171 }, { "epoch": 2.1950205256704165, "grad_norm": 1.876796841621399, "learning_rate": 1.7658e-06, "loss": 0.015, "step": 41172 }, { "epoch": 2.1950738391000693, "grad_norm": 1.3732104301452637, "learning_rate": 1.7656e-06, "loss": 0.0127, "step": 41173 }, { "epoch": 2.1951271525297225, "grad_norm": 1.077743411064148, "learning_rate": 1.7654e-06, "loss": 0.0287, "step": 41174 }, { "epoch": 2.195180465959375, "grad_norm": 3.3554277420043945, "learning_rate": 1.7652000000000003e-06, "loss": 0.0322, "step": 41175 }, { "epoch": 2.195233779389028, "grad_norm": 3.0328028202056885, "learning_rate": 1.765e-06, "loss": 0.0271, "step": 41176 }, { "epoch": 2.195287092818681, "grad_norm": 2.32253360748291, "learning_rate": 1.7648e-06, "loss": 0.0399, "step": 41177 }, { "epoch": 2.195340406248334, "grad_norm": 1.9815620183944702, "learning_rate": 1.7646000000000002e-06, "loss": 0.0343, "step": 41178 }, { "epoch": 2.1953937196779867, "grad_norm": 2.81514835357666, "learning_rate": 1.7644000000000002e-06, "loss": 0.0212, "step": 41179 }, { "epoch": 2.19544703310764, "grad_norm": 0.5690876245498657, "learning_rate": 1.7642e-06, "loss": 0.0036, "step": 41180 }, { "epoch": 2.1955003465372926, "grad_norm": 2.2130045890808105, "learning_rate": 1.7640000000000002e-06, "loss": 0.0217, "step": 41181 }, { "epoch": 2.195553659966946, "grad_norm": 2.303375720977783, "learning_rate": 1.7638000000000002e-06, "loss": 0.0267, "step": 41182 }, { "epoch": 2.1956069733965986, "grad_norm": 2.0418291091918945, "learning_rate": 1.7636e-06, "loss": 0.0187, "step": 41183 }, { "epoch": 2.1956602868262514, "grad_norm": 2.001450300216675, "learning_rate": 1.7634000000000001e-06, "loss": 0.0228, "step": 41184 }, { "epoch": 2.1957136002559046, "grad_norm": 0.5962427854537964, "learning_rate": 1.7632000000000001e-06, "loss": 0.005, "step": 41185 }, { "epoch": 2.1957669136855573, "grad_norm": 3.1093661785125732, "learning_rate": 1.7630000000000003e-06, "loss": 0.038, "step": 41186 }, { "epoch": 2.1958202271152105, "grad_norm": 3.907322406768799, "learning_rate": 1.7628e-06, "loss": 0.0172, "step": 41187 }, { "epoch": 2.1958735405448633, "grad_norm": 1.941418170928955, "learning_rate": 1.7626e-06, "loss": 0.0251, "step": 41188 }, { "epoch": 2.195926853974516, "grad_norm": 2.9431214332580566, "learning_rate": 1.7624000000000003e-06, "loss": 0.0197, "step": 41189 }, { "epoch": 2.1959801674041692, "grad_norm": 1.9337632656097412, "learning_rate": 1.7622e-06, "loss": 0.0175, "step": 41190 }, { "epoch": 2.196033480833822, "grad_norm": 3.4122166633605957, "learning_rate": 1.762e-06, "loss": 0.0263, "step": 41191 }, { "epoch": 2.1960867942634748, "grad_norm": 1.071877360343933, "learning_rate": 1.7618000000000002e-06, "loss": 0.0073, "step": 41192 }, { "epoch": 2.196140107693128, "grad_norm": 2.963221549987793, "learning_rate": 1.7616000000000002e-06, "loss": 0.017, "step": 41193 }, { "epoch": 2.1961934211227807, "grad_norm": 4.582078456878662, "learning_rate": 1.7614e-06, "loss": 0.0255, "step": 41194 }, { "epoch": 2.196246734552434, "grad_norm": 4.285765647888184, "learning_rate": 1.7612000000000002e-06, "loss": 0.0232, "step": 41195 }, { "epoch": 2.1963000479820867, "grad_norm": 1.5599043369293213, "learning_rate": 1.7610000000000002e-06, "loss": 0.0127, "step": 41196 }, { "epoch": 2.1963533614117394, "grad_norm": 1.5823338031768799, "learning_rate": 1.7608e-06, "loss": 0.0065, "step": 41197 }, { "epoch": 2.1964066748413926, "grad_norm": 1.8428866863250732, "learning_rate": 1.7606000000000001e-06, "loss": 0.0297, "step": 41198 }, { "epoch": 2.1964599882710454, "grad_norm": 2.565598964691162, "learning_rate": 1.7604000000000001e-06, "loss": 0.0121, "step": 41199 }, { "epoch": 2.1965133017006986, "grad_norm": 1.4017335176467896, "learning_rate": 1.7602000000000003e-06, "loss": 0.0068, "step": 41200 }, { "epoch": 2.1965666151303513, "grad_norm": 1.626359462738037, "learning_rate": 1.76e-06, "loss": 0.0248, "step": 41201 }, { "epoch": 2.196619928560004, "grad_norm": 1.1700365543365479, "learning_rate": 1.7598e-06, "loss": 0.0129, "step": 41202 }, { "epoch": 2.1966732419896573, "grad_norm": 1.3913987874984741, "learning_rate": 1.7596000000000003e-06, "loss": 0.0145, "step": 41203 }, { "epoch": 2.19672655541931, "grad_norm": 1.7063168287277222, "learning_rate": 1.7594000000000003e-06, "loss": 0.0147, "step": 41204 }, { "epoch": 2.1967798688489633, "grad_norm": 2.0928876399993896, "learning_rate": 1.7592e-06, "loss": 0.0185, "step": 41205 }, { "epoch": 2.196833182278616, "grad_norm": 2.2284247875213623, "learning_rate": 1.7590000000000002e-06, "loss": 0.0254, "step": 41206 }, { "epoch": 2.1968864957082688, "grad_norm": 1.9683834314346313, "learning_rate": 1.7588000000000002e-06, "loss": 0.0256, "step": 41207 }, { "epoch": 2.196939809137922, "grad_norm": 2.8059732913970947, "learning_rate": 1.7586e-06, "loss": 0.0058, "step": 41208 }, { "epoch": 2.1969931225675747, "grad_norm": 2.159313917160034, "learning_rate": 1.7584000000000002e-06, "loss": 0.011, "step": 41209 }, { "epoch": 2.197046435997228, "grad_norm": 1.4560703039169312, "learning_rate": 1.7582000000000002e-06, "loss": 0.0266, "step": 41210 }, { "epoch": 2.1970997494268807, "grad_norm": 1.9606239795684814, "learning_rate": 1.7580000000000004e-06, "loss": 0.0105, "step": 41211 }, { "epoch": 2.1971530628565334, "grad_norm": 3.174992561340332, "learning_rate": 1.7578000000000001e-06, "loss": 0.024, "step": 41212 }, { "epoch": 2.1972063762861866, "grad_norm": 3.1877715587615967, "learning_rate": 1.7576000000000001e-06, "loss": 0.0187, "step": 41213 }, { "epoch": 2.1972596897158394, "grad_norm": 1.003922462463379, "learning_rate": 1.7574000000000003e-06, "loss": 0.0218, "step": 41214 }, { "epoch": 2.197313003145492, "grad_norm": 1.6621850728988647, "learning_rate": 1.7572e-06, "loss": 0.0064, "step": 41215 }, { "epoch": 2.1973663165751454, "grad_norm": 1.2071938514709473, "learning_rate": 1.757e-06, "loss": 0.0077, "step": 41216 }, { "epoch": 2.197419630004798, "grad_norm": 1.2091199159622192, "learning_rate": 1.7568000000000003e-06, "loss": 0.0334, "step": 41217 }, { "epoch": 2.1974729434344513, "grad_norm": 2.5148234367370605, "learning_rate": 1.7566000000000003e-06, "loss": 0.0335, "step": 41218 }, { "epoch": 2.197526256864104, "grad_norm": 1.3215266466140747, "learning_rate": 1.7564e-06, "loss": 0.0178, "step": 41219 }, { "epoch": 2.197579570293757, "grad_norm": 3.1632585525512695, "learning_rate": 1.7562000000000002e-06, "loss": 0.0478, "step": 41220 }, { "epoch": 2.19763288372341, "grad_norm": 1.5704736709594727, "learning_rate": 1.7560000000000002e-06, "loss": 0.0216, "step": 41221 }, { "epoch": 2.197686197153063, "grad_norm": 0.925225019454956, "learning_rate": 1.7558000000000002e-06, "loss": 0.0056, "step": 41222 }, { "epoch": 2.197739510582716, "grad_norm": 0.9366655945777893, "learning_rate": 1.7556000000000002e-06, "loss": 0.0042, "step": 41223 }, { "epoch": 2.1977928240123688, "grad_norm": 1.5820707082748413, "learning_rate": 1.7554000000000002e-06, "loss": 0.0153, "step": 41224 }, { "epoch": 2.1978461374420215, "grad_norm": 1.28378427028656, "learning_rate": 1.7552000000000002e-06, "loss": 0.0069, "step": 41225 }, { "epoch": 2.1978994508716747, "grad_norm": 1.7039117813110352, "learning_rate": 1.7550000000000001e-06, "loss": 0.0313, "step": 41226 }, { "epoch": 2.1979527643013275, "grad_norm": 1.6205798387527466, "learning_rate": 1.7548000000000001e-06, "loss": 0.0171, "step": 41227 }, { "epoch": 2.1980060777309802, "grad_norm": 1.1731735467910767, "learning_rate": 1.7546000000000001e-06, "loss": 0.0091, "step": 41228 }, { "epoch": 2.1980593911606334, "grad_norm": 1.6358212232589722, "learning_rate": 1.7544000000000003e-06, "loss": 0.0223, "step": 41229 }, { "epoch": 2.198112704590286, "grad_norm": 1.3627129793167114, "learning_rate": 1.7542e-06, "loss": 0.0074, "step": 41230 }, { "epoch": 2.1981660180199394, "grad_norm": 0.6215364933013916, "learning_rate": 1.754e-06, "loss": 0.0053, "step": 41231 }, { "epoch": 2.198219331449592, "grad_norm": 2.4718017578125, "learning_rate": 1.7538000000000003e-06, "loss": 0.0251, "step": 41232 }, { "epoch": 2.198272644879245, "grad_norm": 1.8325296640396118, "learning_rate": 1.7536e-06, "loss": 0.0093, "step": 41233 }, { "epoch": 2.198325958308898, "grad_norm": 2.0324316024780273, "learning_rate": 1.7534e-06, "loss": 0.0183, "step": 41234 }, { "epoch": 2.198379271738551, "grad_norm": 2.607103109359741, "learning_rate": 1.7532000000000002e-06, "loss": 0.0299, "step": 41235 }, { "epoch": 2.198432585168204, "grad_norm": 1.2554618120193481, "learning_rate": 1.7530000000000002e-06, "loss": 0.0139, "step": 41236 }, { "epoch": 2.198485898597857, "grad_norm": 0.6177704930305481, "learning_rate": 1.7528e-06, "loss": 0.0055, "step": 41237 }, { "epoch": 2.1985392120275096, "grad_norm": 2.6550686359405518, "learning_rate": 1.7526000000000002e-06, "loss": 0.0324, "step": 41238 }, { "epoch": 2.1985925254571628, "grad_norm": 1.0785257816314697, "learning_rate": 1.7524000000000002e-06, "loss": 0.0175, "step": 41239 }, { "epoch": 2.1986458388868155, "grad_norm": 1.0933893918991089, "learning_rate": 1.7522e-06, "loss": 0.017, "step": 41240 }, { "epoch": 2.1986991523164683, "grad_norm": 1.3791824579238892, "learning_rate": 1.7520000000000001e-06, "loss": 0.0175, "step": 41241 }, { "epoch": 2.1987524657461215, "grad_norm": 1.6280523538589478, "learning_rate": 1.7518000000000001e-06, "loss": 0.0182, "step": 41242 }, { "epoch": 2.1988057791757742, "grad_norm": 1.012305498123169, "learning_rate": 1.7516000000000003e-06, "loss": 0.0201, "step": 41243 }, { "epoch": 2.1988590926054274, "grad_norm": 1.5294575691223145, "learning_rate": 1.7514e-06, "loss": 0.0256, "step": 41244 }, { "epoch": 2.19891240603508, "grad_norm": 1.5373972654342651, "learning_rate": 1.7512e-06, "loss": 0.0163, "step": 41245 }, { "epoch": 2.198965719464733, "grad_norm": 0.9043763279914856, "learning_rate": 1.7510000000000003e-06, "loss": 0.0065, "step": 41246 }, { "epoch": 2.199019032894386, "grad_norm": 1.2323250770568848, "learning_rate": 1.7508000000000002e-06, "loss": 0.0051, "step": 41247 }, { "epoch": 2.199072346324039, "grad_norm": 2.936572551727295, "learning_rate": 1.7506e-06, "loss": 0.0462, "step": 41248 }, { "epoch": 2.199125659753692, "grad_norm": 0.5091372132301331, "learning_rate": 1.7504000000000002e-06, "loss": 0.0035, "step": 41249 }, { "epoch": 2.199178973183345, "grad_norm": 1.693778157234192, "learning_rate": 1.7502000000000002e-06, "loss": 0.0111, "step": 41250 }, { "epoch": 2.1992322866129976, "grad_norm": 1.4761438369750977, "learning_rate": 1.75e-06, "loss": 0.0187, "step": 41251 }, { "epoch": 2.199285600042651, "grad_norm": 1.2224524021148682, "learning_rate": 1.7498000000000002e-06, "loss": 0.0081, "step": 41252 }, { "epoch": 2.1993389134723036, "grad_norm": 2.2840189933776855, "learning_rate": 1.7496000000000002e-06, "loss": 0.0077, "step": 41253 }, { "epoch": 2.199392226901957, "grad_norm": 1.3721084594726562, "learning_rate": 1.7494000000000004e-06, "loss": 0.0068, "step": 41254 }, { "epoch": 2.1994455403316095, "grad_norm": 0.9418070912361145, "learning_rate": 1.7492000000000001e-06, "loss": 0.0046, "step": 41255 }, { "epoch": 2.1994988537612623, "grad_norm": 1.5296119451522827, "learning_rate": 1.7490000000000001e-06, "loss": 0.0081, "step": 41256 }, { "epoch": 2.1995521671909155, "grad_norm": 1.6308040618896484, "learning_rate": 1.7488000000000003e-06, "loss": 0.0087, "step": 41257 }, { "epoch": 2.1996054806205683, "grad_norm": 1.8229420185089111, "learning_rate": 1.7486e-06, "loss": 0.0227, "step": 41258 }, { "epoch": 2.1996587940502215, "grad_norm": 1.630861520767212, "learning_rate": 1.7484e-06, "loss": 0.0253, "step": 41259 }, { "epoch": 2.1997121074798742, "grad_norm": 5.070693016052246, "learning_rate": 1.7482000000000003e-06, "loss": 0.0309, "step": 41260 }, { "epoch": 2.199765420909527, "grad_norm": 0.7045930624008179, "learning_rate": 1.7480000000000002e-06, "loss": 0.0191, "step": 41261 }, { "epoch": 2.19981873433918, "grad_norm": 1.2726341485977173, "learning_rate": 1.7478e-06, "loss": 0.0074, "step": 41262 }, { "epoch": 2.199872047768833, "grad_norm": 1.9707399606704712, "learning_rate": 1.7476000000000002e-06, "loss": 0.0152, "step": 41263 }, { "epoch": 2.1999253611984857, "grad_norm": 0.9657717943191528, "learning_rate": 1.7474000000000002e-06, "loss": 0.0135, "step": 41264 }, { "epoch": 2.199978674628139, "grad_norm": 1.5277503728866577, "learning_rate": 1.7472e-06, "loss": 0.0276, "step": 41265 }, { "epoch": 2.2000319880577917, "grad_norm": 1.1676744222640991, "learning_rate": 1.7470000000000002e-06, "loss": 0.0063, "step": 41266 }, { "epoch": 2.200085301487445, "grad_norm": 1.0331846475601196, "learning_rate": 1.7468000000000002e-06, "loss": 0.0158, "step": 41267 }, { "epoch": 2.2001386149170976, "grad_norm": 1.5439192056655884, "learning_rate": 1.7466000000000001e-06, "loss": 0.01, "step": 41268 }, { "epoch": 2.2001919283467504, "grad_norm": 0.7902451157569885, "learning_rate": 1.7464000000000001e-06, "loss": 0.0173, "step": 41269 }, { "epoch": 2.2002452417764036, "grad_norm": 2.3145968914031982, "learning_rate": 1.7462000000000001e-06, "loss": 0.0287, "step": 41270 }, { "epoch": 2.2002985552060563, "grad_norm": 1.8459950685501099, "learning_rate": 1.746e-06, "loss": 0.0408, "step": 41271 }, { "epoch": 2.2003518686357095, "grad_norm": 1.4287692308425903, "learning_rate": 1.7458000000000003e-06, "loss": 0.0193, "step": 41272 }, { "epoch": 2.2004051820653623, "grad_norm": 1.155497670173645, "learning_rate": 1.7456e-06, "loss": 0.0119, "step": 41273 }, { "epoch": 2.200458495495015, "grad_norm": 1.3716992139816284, "learning_rate": 1.7454e-06, "loss": 0.0344, "step": 41274 }, { "epoch": 2.2005118089246682, "grad_norm": 0.9259241819381714, "learning_rate": 1.7452000000000002e-06, "loss": 0.0053, "step": 41275 }, { "epoch": 2.200565122354321, "grad_norm": 1.6337438821792603, "learning_rate": 1.745e-06, "loss": 0.008, "step": 41276 }, { "epoch": 2.2006184357839738, "grad_norm": 2.598942279815674, "learning_rate": 1.7448e-06, "loss": 0.0335, "step": 41277 }, { "epoch": 2.200671749213627, "grad_norm": 0.7054129838943481, "learning_rate": 1.7446000000000002e-06, "loss": 0.0051, "step": 41278 }, { "epoch": 2.2007250626432797, "grad_norm": 1.5251514911651611, "learning_rate": 1.7444000000000002e-06, "loss": 0.0131, "step": 41279 }, { "epoch": 2.200778376072933, "grad_norm": 2.822007894515991, "learning_rate": 1.7442e-06, "loss": 0.0317, "step": 41280 }, { "epoch": 2.2008316895025857, "grad_norm": 1.791922688484192, "learning_rate": 1.7440000000000002e-06, "loss": 0.023, "step": 41281 }, { "epoch": 2.2008850029322384, "grad_norm": 2.502220869064331, "learning_rate": 1.7438000000000001e-06, "loss": 0.0182, "step": 41282 }, { "epoch": 2.2009383163618916, "grad_norm": 2.3128244876861572, "learning_rate": 1.7436e-06, "loss": 0.0289, "step": 41283 }, { "epoch": 2.2009916297915444, "grad_norm": 1.8418546915054321, "learning_rate": 1.7434000000000001e-06, "loss": 0.0205, "step": 41284 }, { "epoch": 2.2010449432211976, "grad_norm": 2.9765994548797607, "learning_rate": 1.7432e-06, "loss": 0.0429, "step": 41285 }, { "epoch": 2.2010982566508503, "grad_norm": 2.212357521057129, "learning_rate": 1.7430000000000003e-06, "loss": 0.0314, "step": 41286 }, { "epoch": 2.201151570080503, "grad_norm": 1.6118297576904297, "learning_rate": 1.7428e-06, "loss": 0.0199, "step": 41287 }, { "epoch": 2.2012048835101563, "grad_norm": 2.699230432510376, "learning_rate": 1.7426e-06, "loss": 0.0315, "step": 41288 }, { "epoch": 2.201258196939809, "grad_norm": 2.262148141860962, "learning_rate": 1.7424000000000002e-06, "loss": 0.0438, "step": 41289 }, { "epoch": 2.2013115103694623, "grad_norm": 1.4873168468475342, "learning_rate": 1.7422000000000002e-06, "loss": 0.0205, "step": 41290 }, { "epoch": 2.201364823799115, "grad_norm": 2.1246445178985596, "learning_rate": 1.742e-06, "loss": 0.0247, "step": 41291 }, { "epoch": 2.201418137228768, "grad_norm": 3.2495503425598145, "learning_rate": 1.7418000000000002e-06, "loss": 0.0365, "step": 41292 }, { "epoch": 2.201471450658421, "grad_norm": 2.5531766414642334, "learning_rate": 1.7416000000000002e-06, "loss": 0.024, "step": 41293 }, { "epoch": 2.2015247640880737, "grad_norm": 1.2782714366912842, "learning_rate": 1.7414e-06, "loss": 0.0083, "step": 41294 }, { "epoch": 2.201578077517727, "grad_norm": 1.6950732469558716, "learning_rate": 1.7412000000000002e-06, "loss": 0.0509, "step": 41295 }, { "epoch": 2.2016313909473797, "grad_norm": 2.9682223796844482, "learning_rate": 1.7410000000000001e-06, "loss": 0.017, "step": 41296 }, { "epoch": 2.2016847043770325, "grad_norm": 1.3031768798828125, "learning_rate": 1.7408000000000003e-06, "loss": 0.0061, "step": 41297 }, { "epoch": 2.2017380178066857, "grad_norm": 3.1499245166778564, "learning_rate": 1.7406000000000001e-06, "loss": 0.01, "step": 41298 }, { "epoch": 2.2017913312363384, "grad_norm": 1.6452434062957764, "learning_rate": 1.7404e-06, "loss": 0.0276, "step": 41299 }, { "epoch": 2.201844644665991, "grad_norm": 2.5222949981689453, "learning_rate": 1.7402000000000003e-06, "loss": 0.0488, "step": 41300 }, { "epoch": 2.2018979580956444, "grad_norm": 1.0385373830795288, "learning_rate": 1.74e-06, "loss": 0.0082, "step": 41301 }, { "epoch": 2.201951271525297, "grad_norm": 2.4662413597106934, "learning_rate": 1.7398e-06, "loss": 0.0303, "step": 41302 }, { "epoch": 2.2020045849549503, "grad_norm": 0.8532686829566956, "learning_rate": 1.7396000000000003e-06, "loss": 0.006, "step": 41303 }, { "epoch": 2.202057898384603, "grad_norm": 0.8805186152458191, "learning_rate": 1.7394000000000002e-06, "loss": 0.0244, "step": 41304 }, { "epoch": 2.202111211814256, "grad_norm": 0.3404771387577057, "learning_rate": 1.7392e-06, "loss": 0.003, "step": 41305 }, { "epoch": 2.202164525243909, "grad_norm": 1.5276286602020264, "learning_rate": 1.7390000000000002e-06, "loss": 0.0211, "step": 41306 }, { "epoch": 2.202217838673562, "grad_norm": 4.752659320831299, "learning_rate": 1.7388000000000002e-06, "loss": 0.0506, "step": 41307 }, { "epoch": 2.202271152103215, "grad_norm": 1.1632689237594604, "learning_rate": 1.7386e-06, "loss": 0.0252, "step": 41308 }, { "epoch": 2.2023244655328678, "grad_norm": 0.6373146772384644, "learning_rate": 1.7384000000000002e-06, "loss": 0.0045, "step": 41309 }, { "epoch": 2.2023777789625205, "grad_norm": 2.019468069076538, "learning_rate": 1.7382000000000001e-06, "loss": 0.0208, "step": 41310 }, { "epoch": 2.2024310923921737, "grad_norm": 2.3267812728881836, "learning_rate": 1.7380000000000003e-06, "loss": 0.0132, "step": 41311 }, { "epoch": 2.2024844058218265, "grad_norm": 1.795928716659546, "learning_rate": 1.7378000000000001e-06, "loss": 0.0409, "step": 41312 }, { "epoch": 2.2025377192514792, "grad_norm": 0.506161093711853, "learning_rate": 1.7376e-06, "loss": 0.0029, "step": 41313 }, { "epoch": 2.2025910326811324, "grad_norm": 2.6129307746887207, "learning_rate": 1.7374000000000003e-06, "loss": 0.02, "step": 41314 }, { "epoch": 2.202644346110785, "grad_norm": 1.7677688598632812, "learning_rate": 1.7372000000000003e-06, "loss": 0.0155, "step": 41315 }, { "epoch": 2.2026976595404384, "grad_norm": 2.031975269317627, "learning_rate": 1.737e-06, "loss": 0.0073, "step": 41316 }, { "epoch": 2.202750972970091, "grad_norm": 1.332695484161377, "learning_rate": 1.7368000000000003e-06, "loss": 0.0136, "step": 41317 }, { "epoch": 2.202804286399744, "grad_norm": 1.9675637483596802, "learning_rate": 1.7366000000000002e-06, "loss": 0.0122, "step": 41318 }, { "epoch": 2.202857599829397, "grad_norm": 1.9236361980438232, "learning_rate": 1.7364e-06, "loss": 0.0085, "step": 41319 }, { "epoch": 2.20291091325905, "grad_norm": 0.915346622467041, "learning_rate": 1.7362000000000002e-06, "loss": 0.0056, "step": 41320 }, { "epoch": 2.202964226688703, "grad_norm": 1.3193238973617554, "learning_rate": 1.7360000000000002e-06, "loss": 0.0094, "step": 41321 }, { "epoch": 2.203017540118356, "grad_norm": 1.6459918022155762, "learning_rate": 1.7358000000000002e-06, "loss": 0.01, "step": 41322 }, { "epoch": 2.2030708535480086, "grad_norm": 1.5668810606002808, "learning_rate": 1.7356000000000002e-06, "loss": 0.01, "step": 41323 }, { "epoch": 2.203124166977662, "grad_norm": 1.8587369918823242, "learning_rate": 1.7354000000000001e-06, "loss": 0.0441, "step": 41324 }, { "epoch": 2.2031774804073145, "grad_norm": 1.0903878211975098, "learning_rate": 1.7352000000000001e-06, "loss": 0.0085, "step": 41325 }, { "epoch": 2.2032307938369677, "grad_norm": 2.4211409091949463, "learning_rate": 1.7350000000000001e-06, "loss": 0.016, "step": 41326 }, { "epoch": 2.2032841072666205, "grad_norm": 1.4911953210830688, "learning_rate": 1.7348e-06, "loss": 0.007, "step": 41327 }, { "epoch": 2.2033374206962733, "grad_norm": 0.62468022108078, "learning_rate": 1.7346e-06, "loss": 0.0035, "step": 41328 }, { "epoch": 2.2033907341259265, "grad_norm": 1.3118479251861572, "learning_rate": 1.7344000000000003e-06, "loss": 0.0169, "step": 41329 }, { "epoch": 2.203444047555579, "grad_norm": 0.42838194966316223, "learning_rate": 1.7342e-06, "loss": 0.0037, "step": 41330 }, { "epoch": 2.2034973609852324, "grad_norm": 1.2975842952728271, "learning_rate": 1.734e-06, "loss": 0.0065, "step": 41331 }, { "epoch": 2.203550674414885, "grad_norm": 1.432831048965454, "learning_rate": 1.7338000000000002e-06, "loss": 0.0237, "step": 41332 }, { "epoch": 2.203603987844538, "grad_norm": 1.191985845565796, "learning_rate": 1.7336e-06, "loss": 0.0087, "step": 41333 }, { "epoch": 2.203657301274191, "grad_norm": 1.746330976486206, "learning_rate": 1.7334e-06, "loss": 0.0081, "step": 41334 }, { "epoch": 2.203710614703844, "grad_norm": 3.8318376541137695, "learning_rate": 1.7332000000000002e-06, "loss": 0.0216, "step": 41335 }, { "epoch": 2.2037639281334966, "grad_norm": 3.0515027046203613, "learning_rate": 1.7330000000000002e-06, "loss": 0.0449, "step": 41336 }, { "epoch": 2.20381724156315, "grad_norm": 2.180480480194092, "learning_rate": 1.7328e-06, "loss": 0.0198, "step": 41337 }, { "epoch": 2.2038705549928026, "grad_norm": 1.6176358461380005, "learning_rate": 1.7326000000000001e-06, "loss": 0.0097, "step": 41338 }, { "epoch": 2.203923868422456, "grad_norm": 2.840512752532959, "learning_rate": 1.7324000000000001e-06, "loss": 0.062, "step": 41339 }, { "epoch": 2.2039771818521086, "grad_norm": 3.075572967529297, "learning_rate": 1.7322000000000003e-06, "loss": 0.0424, "step": 41340 }, { "epoch": 2.2040304952817613, "grad_norm": 1.8126955032348633, "learning_rate": 1.732e-06, "loss": 0.026, "step": 41341 }, { "epoch": 2.2040838087114145, "grad_norm": 3.553811550140381, "learning_rate": 1.7318e-06, "loss": 0.0329, "step": 41342 }, { "epoch": 2.2041371221410673, "grad_norm": 0.7178412675857544, "learning_rate": 1.7316000000000003e-06, "loss": 0.0041, "step": 41343 }, { "epoch": 2.2041904355707205, "grad_norm": 2.992525815963745, "learning_rate": 1.7314e-06, "loss": 0.0315, "step": 41344 }, { "epoch": 2.2042437490003732, "grad_norm": 4.791736125946045, "learning_rate": 1.7312e-06, "loss": 0.0287, "step": 41345 }, { "epoch": 2.204297062430026, "grad_norm": 0.47236859798431396, "learning_rate": 1.7310000000000002e-06, "loss": 0.0046, "step": 41346 }, { "epoch": 2.204350375859679, "grad_norm": 1.2592029571533203, "learning_rate": 1.7308000000000002e-06, "loss": 0.0064, "step": 41347 }, { "epoch": 2.204403689289332, "grad_norm": 2.496610164642334, "learning_rate": 1.7306e-06, "loss": 0.0119, "step": 41348 }, { "epoch": 2.2044570027189847, "grad_norm": 0.6183760166168213, "learning_rate": 1.7304000000000002e-06, "loss": 0.004, "step": 41349 }, { "epoch": 2.204510316148638, "grad_norm": 1.4037339687347412, "learning_rate": 1.7302000000000002e-06, "loss": 0.0076, "step": 41350 }, { "epoch": 2.2045636295782907, "grad_norm": 2.423168659210205, "learning_rate": 1.73e-06, "loss": 0.0154, "step": 41351 }, { "epoch": 2.204616943007944, "grad_norm": 2.5701897144317627, "learning_rate": 1.7298000000000002e-06, "loss": 0.0255, "step": 41352 }, { "epoch": 2.2046702564375966, "grad_norm": 1.1387853622436523, "learning_rate": 1.7296000000000001e-06, "loss": 0.0078, "step": 41353 }, { "epoch": 2.2047235698672494, "grad_norm": 1.6266494989395142, "learning_rate": 1.7294000000000003e-06, "loss": 0.0103, "step": 41354 }, { "epoch": 2.2047768832969026, "grad_norm": 2.013681173324585, "learning_rate": 1.7292e-06, "loss": 0.0344, "step": 41355 }, { "epoch": 2.2048301967265553, "grad_norm": 1.0348249673843384, "learning_rate": 1.729e-06, "loss": 0.0065, "step": 41356 }, { "epoch": 2.2048835101562085, "grad_norm": 2.394376277923584, "learning_rate": 1.7288000000000003e-06, "loss": 0.0199, "step": 41357 }, { "epoch": 2.2049368235858613, "grad_norm": 2.4562878608703613, "learning_rate": 1.7286000000000003e-06, "loss": 0.0112, "step": 41358 }, { "epoch": 2.204990137015514, "grad_norm": 2.036105155944824, "learning_rate": 1.7284e-06, "loss": 0.016, "step": 41359 }, { "epoch": 2.2050434504451673, "grad_norm": 2.6426022052764893, "learning_rate": 1.7282000000000002e-06, "loss": 0.0422, "step": 41360 }, { "epoch": 2.20509676387482, "grad_norm": 2.384661912918091, "learning_rate": 1.7280000000000002e-06, "loss": 0.0492, "step": 41361 }, { "epoch": 2.2051500773044728, "grad_norm": 4.872062683105469, "learning_rate": 1.7278e-06, "loss": 0.0403, "step": 41362 }, { "epoch": 2.205203390734126, "grad_norm": 0.6676648855209351, "learning_rate": 1.7276000000000002e-06, "loss": 0.0041, "step": 41363 }, { "epoch": 2.2052567041637787, "grad_norm": 2.026165246963501, "learning_rate": 1.7274000000000002e-06, "loss": 0.0151, "step": 41364 }, { "epoch": 2.205310017593432, "grad_norm": 1.1431609392166138, "learning_rate": 1.7272000000000002e-06, "loss": 0.0065, "step": 41365 }, { "epoch": 2.2053633310230847, "grad_norm": 2.5243353843688965, "learning_rate": 1.7270000000000002e-06, "loss": 0.0187, "step": 41366 }, { "epoch": 2.2054166444527374, "grad_norm": 3.4659643173217773, "learning_rate": 1.7268000000000001e-06, "loss": 0.0095, "step": 41367 }, { "epoch": 2.2054699578823906, "grad_norm": 1.5278854370117188, "learning_rate": 1.7266000000000001e-06, "loss": 0.0159, "step": 41368 }, { "epoch": 2.2055232713120434, "grad_norm": 2.1213064193725586, "learning_rate": 1.7264000000000001e-06, "loss": 0.0254, "step": 41369 }, { "epoch": 2.2055765847416966, "grad_norm": 2.0653152465820312, "learning_rate": 1.7262e-06, "loss": 0.0135, "step": 41370 }, { "epoch": 2.2056298981713494, "grad_norm": 4.031669616699219, "learning_rate": 1.726e-06, "loss": 0.0296, "step": 41371 }, { "epoch": 2.205683211601002, "grad_norm": 1.6762934923171997, "learning_rate": 1.7258000000000003e-06, "loss": 0.0125, "step": 41372 }, { "epoch": 2.2057365250306553, "grad_norm": 0.9439535140991211, "learning_rate": 1.7256e-06, "loss": 0.012, "step": 41373 }, { "epoch": 2.205789838460308, "grad_norm": 0.605820894241333, "learning_rate": 1.7254e-06, "loss": 0.0039, "step": 41374 }, { "epoch": 2.2058431518899613, "grad_norm": 2.34201717376709, "learning_rate": 1.7252000000000002e-06, "loss": 0.008, "step": 41375 }, { "epoch": 2.205896465319614, "grad_norm": 3.147017478942871, "learning_rate": 1.725e-06, "loss": 0.0161, "step": 41376 }, { "epoch": 2.205949778749267, "grad_norm": 2.4119455814361572, "learning_rate": 1.7248e-06, "loss": 0.0235, "step": 41377 }, { "epoch": 2.20600309217892, "grad_norm": 1.7057234048843384, "learning_rate": 1.7246000000000002e-06, "loss": 0.0078, "step": 41378 }, { "epoch": 2.2060564056085727, "grad_norm": 0.5591159462928772, "learning_rate": 1.7244000000000002e-06, "loss": 0.0038, "step": 41379 }, { "epoch": 2.206109719038226, "grad_norm": 0.2607176601886749, "learning_rate": 1.7242e-06, "loss": 0.0024, "step": 41380 }, { "epoch": 2.2061630324678787, "grad_norm": 2.4992172718048096, "learning_rate": 1.7240000000000001e-06, "loss": 0.0195, "step": 41381 }, { "epoch": 2.2062163458975315, "grad_norm": 2.089114189147949, "learning_rate": 1.7238000000000001e-06, "loss": 0.0171, "step": 41382 }, { "epoch": 2.2062696593271847, "grad_norm": 1.2131130695343018, "learning_rate": 1.7236000000000003e-06, "loss": 0.0273, "step": 41383 }, { "epoch": 2.2063229727568374, "grad_norm": 1.9291536808013916, "learning_rate": 1.7234e-06, "loss": 0.0079, "step": 41384 }, { "epoch": 2.20637628618649, "grad_norm": 1.307854175567627, "learning_rate": 1.7232e-06, "loss": 0.0177, "step": 41385 }, { "epoch": 2.2064295996161434, "grad_norm": 2.5967612266540527, "learning_rate": 1.7230000000000003e-06, "loss": 0.0432, "step": 41386 }, { "epoch": 2.206482913045796, "grad_norm": 0.4085598289966583, "learning_rate": 1.7228e-06, "loss": 0.0031, "step": 41387 }, { "epoch": 2.2065362264754493, "grad_norm": 2.2195169925689697, "learning_rate": 1.7226e-06, "loss": 0.0148, "step": 41388 }, { "epoch": 2.206589539905102, "grad_norm": 0.6232171654701233, "learning_rate": 1.7224000000000002e-06, "loss": 0.004, "step": 41389 }, { "epoch": 2.206642853334755, "grad_norm": 1.057215929031372, "learning_rate": 1.7222000000000002e-06, "loss": 0.0105, "step": 41390 }, { "epoch": 2.206696166764408, "grad_norm": 4.1122541427612305, "learning_rate": 1.722e-06, "loss": 0.0349, "step": 41391 }, { "epoch": 2.206749480194061, "grad_norm": 2.691208839416504, "learning_rate": 1.7218000000000002e-06, "loss": 0.0311, "step": 41392 }, { "epoch": 2.206802793623714, "grad_norm": 1.2530072927474976, "learning_rate": 1.7216000000000002e-06, "loss": 0.0248, "step": 41393 }, { "epoch": 2.2068561070533668, "grad_norm": 0.5976417064666748, "learning_rate": 1.7214e-06, "loss": 0.0043, "step": 41394 }, { "epoch": 2.2069094204830195, "grad_norm": 2.126283884048462, "learning_rate": 1.7212000000000001e-06, "loss": 0.0178, "step": 41395 }, { "epoch": 2.2069627339126727, "grad_norm": 0.794234573841095, "learning_rate": 1.7210000000000001e-06, "loss": 0.0037, "step": 41396 }, { "epoch": 2.2070160473423255, "grad_norm": 2.3105247020721436, "learning_rate": 1.7208000000000003e-06, "loss": 0.0396, "step": 41397 }, { "epoch": 2.2070693607719782, "grad_norm": 1.446905255317688, "learning_rate": 1.7206e-06, "loss": 0.0216, "step": 41398 }, { "epoch": 2.2071226742016314, "grad_norm": 3.7074785232543945, "learning_rate": 1.7204e-06, "loss": 0.0117, "step": 41399 }, { "epoch": 2.207175987631284, "grad_norm": 2.0931684970855713, "learning_rate": 1.7202000000000003e-06, "loss": 0.0216, "step": 41400 }, { "epoch": 2.2072293010609374, "grad_norm": 1.5468415021896362, "learning_rate": 1.72e-06, "loss": 0.028, "step": 41401 }, { "epoch": 2.20728261449059, "grad_norm": 2.501227378845215, "learning_rate": 1.7198e-06, "loss": 0.033, "step": 41402 }, { "epoch": 2.207335927920243, "grad_norm": 0.8855366706848145, "learning_rate": 1.7196000000000002e-06, "loss": 0.0051, "step": 41403 }, { "epoch": 2.207389241349896, "grad_norm": 1.6806806325912476, "learning_rate": 1.7194000000000002e-06, "loss": 0.0089, "step": 41404 }, { "epoch": 2.207442554779549, "grad_norm": 0.9455738663673401, "learning_rate": 1.7192e-06, "loss": 0.0064, "step": 41405 }, { "epoch": 2.207495868209202, "grad_norm": 0.533456563949585, "learning_rate": 1.7190000000000002e-06, "loss": 0.0036, "step": 41406 }, { "epoch": 2.207549181638855, "grad_norm": 2.2077674865722656, "learning_rate": 1.7188000000000002e-06, "loss": 0.0147, "step": 41407 }, { "epoch": 2.2076024950685076, "grad_norm": 1.6942734718322754, "learning_rate": 1.7186000000000004e-06, "loss": 0.0218, "step": 41408 }, { "epoch": 2.207655808498161, "grad_norm": 1.8901524543762207, "learning_rate": 1.7184000000000001e-06, "loss": 0.018, "step": 41409 }, { "epoch": 2.2077091219278135, "grad_norm": 0.657600998878479, "learning_rate": 1.7182000000000001e-06, "loss": 0.0052, "step": 41410 }, { "epoch": 2.2077624353574667, "grad_norm": 0.5418014526367188, "learning_rate": 1.7180000000000003e-06, "loss": 0.0037, "step": 41411 }, { "epoch": 2.2078157487871195, "grad_norm": 2.337820529937744, "learning_rate": 1.7178e-06, "loss": 0.0175, "step": 41412 }, { "epoch": 2.2078690622167723, "grad_norm": 1.98553466796875, "learning_rate": 1.7176e-06, "loss": 0.0252, "step": 41413 }, { "epoch": 2.2079223756464255, "grad_norm": 0.9764372706413269, "learning_rate": 1.7174000000000003e-06, "loss": 0.0097, "step": 41414 }, { "epoch": 2.207975689076078, "grad_norm": 2.7887370586395264, "learning_rate": 1.7172000000000003e-06, "loss": 0.0351, "step": 41415 }, { "epoch": 2.2080290025057314, "grad_norm": 1.2730703353881836, "learning_rate": 1.717e-06, "loss": 0.0263, "step": 41416 }, { "epoch": 2.208082315935384, "grad_norm": 3.4116153717041016, "learning_rate": 1.7168000000000002e-06, "loss": 0.0484, "step": 41417 }, { "epoch": 2.208135629365037, "grad_norm": 1.2538224458694458, "learning_rate": 1.7166000000000002e-06, "loss": 0.0072, "step": 41418 }, { "epoch": 2.20818894279469, "grad_norm": 1.3324891328811646, "learning_rate": 1.7164e-06, "loss": 0.0188, "step": 41419 }, { "epoch": 2.208242256224343, "grad_norm": 2.7444722652435303, "learning_rate": 1.7162000000000002e-06, "loss": 0.0186, "step": 41420 }, { "epoch": 2.2082955696539956, "grad_norm": 0.8680633902549744, "learning_rate": 1.7160000000000002e-06, "loss": 0.0062, "step": 41421 }, { "epoch": 2.208348883083649, "grad_norm": 1.0648818016052246, "learning_rate": 1.7158000000000002e-06, "loss": 0.0077, "step": 41422 }, { "epoch": 2.2084021965133016, "grad_norm": 1.544540524482727, "learning_rate": 1.7156000000000001e-06, "loss": 0.0178, "step": 41423 }, { "epoch": 2.208455509942955, "grad_norm": 1.3036137819290161, "learning_rate": 1.7154000000000001e-06, "loss": 0.0087, "step": 41424 }, { "epoch": 2.2085088233726076, "grad_norm": 2.259500741958618, "learning_rate": 1.7152000000000001e-06, "loss": 0.0102, "step": 41425 }, { "epoch": 2.2085621368022603, "grad_norm": 1.0925464630126953, "learning_rate": 1.7150000000000003e-06, "loss": 0.0064, "step": 41426 }, { "epoch": 2.2086154502319135, "grad_norm": 3.145228147506714, "learning_rate": 1.7148e-06, "loss": 0.0367, "step": 41427 }, { "epoch": 2.2086687636615663, "grad_norm": 1.9484572410583496, "learning_rate": 1.7146e-06, "loss": 0.0171, "step": 41428 }, { "epoch": 2.2087220770912195, "grad_norm": 2.5207481384277344, "learning_rate": 1.7144000000000003e-06, "loss": 0.0351, "step": 41429 }, { "epoch": 2.2087753905208722, "grad_norm": 1.9107657670974731, "learning_rate": 1.7142e-06, "loss": 0.0206, "step": 41430 }, { "epoch": 2.208828703950525, "grad_norm": 2.0355007648468018, "learning_rate": 1.714e-06, "loss": 0.0238, "step": 41431 }, { "epoch": 2.208882017380178, "grad_norm": 1.8517951965332031, "learning_rate": 1.7138000000000002e-06, "loss": 0.013, "step": 41432 }, { "epoch": 2.208935330809831, "grad_norm": 4.927596569061279, "learning_rate": 1.7136000000000002e-06, "loss": 0.0262, "step": 41433 }, { "epoch": 2.2089886442394837, "grad_norm": 0.3779834806919098, "learning_rate": 1.7134e-06, "loss": 0.0031, "step": 41434 }, { "epoch": 2.209041957669137, "grad_norm": 2.814392328262329, "learning_rate": 1.7132000000000002e-06, "loss": 0.0276, "step": 41435 }, { "epoch": 2.2090952710987897, "grad_norm": 1.8425086736679077, "learning_rate": 1.7130000000000002e-06, "loss": 0.0257, "step": 41436 }, { "epoch": 2.209148584528443, "grad_norm": 3.602712392807007, "learning_rate": 1.7128e-06, "loss": 0.0287, "step": 41437 }, { "epoch": 2.2092018979580956, "grad_norm": 2.1892290115356445, "learning_rate": 1.7126000000000001e-06, "loss": 0.0158, "step": 41438 }, { "epoch": 2.2092552113877484, "grad_norm": 1.6458181142807007, "learning_rate": 1.7124000000000001e-06, "loss": 0.0288, "step": 41439 }, { "epoch": 2.2093085248174016, "grad_norm": 1.6247862577438354, "learning_rate": 1.7122000000000003e-06, "loss": 0.0362, "step": 41440 }, { "epoch": 2.2093618382470543, "grad_norm": 2.5638532638549805, "learning_rate": 1.712e-06, "loss": 0.0174, "step": 41441 }, { "epoch": 2.2094151516767075, "grad_norm": 2.7940680980682373, "learning_rate": 1.7118e-06, "loss": 0.0148, "step": 41442 }, { "epoch": 2.2094684651063603, "grad_norm": 2.2329843044281006, "learning_rate": 1.7116000000000003e-06, "loss": 0.0219, "step": 41443 }, { "epoch": 2.209521778536013, "grad_norm": 1.3690180778503418, "learning_rate": 1.7114e-06, "loss": 0.0066, "step": 41444 }, { "epoch": 2.2095750919656663, "grad_norm": 1.2904587984085083, "learning_rate": 1.7112e-06, "loss": 0.0123, "step": 41445 }, { "epoch": 2.209628405395319, "grad_norm": 1.2785403728485107, "learning_rate": 1.7110000000000002e-06, "loss": 0.0124, "step": 41446 }, { "epoch": 2.209681718824972, "grad_norm": 1.774044394493103, "learning_rate": 1.7108000000000002e-06, "loss": 0.0166, "step": 41447 }, { "epoch": 2.209735032254625, "grad_norm": 1.87959623336792, "learning_rate": 1.7106e-06, "loss": 0.0275, "step": 41448 }, { "epoch": 2.2097883456842777, "grad_norm": 2.032958507537842, "learning_rate": 1.7104000000000002e-06, "loss": 0.0266, "step": 41449 }, { "epoch": 2.209841659113931, "grad_norm": 1.8196163177490234, "learning_rate": 1.7102000000000002e-06, "loss": 0.0172, "step": 41450 }, { "epoch": 2.2098949725435837, "grad_norm": 0.999852180480957, "learning_rate": 1.7100000000000004e-06, "loss": 0.0171, "step": 41451 }, { "epoch": 2.209948285973237, "grad_norm": 2.0195348262786865, "learning_rate": 1.7098000000000001e-06, "loss": 0.0329, "step": 41452 }, { "epoch": 2.2100015994028896, "grad_norm": 1.3953843116760254, "learning_rate": 1.7096000000000001e-06, "loss": 0.0097, "step": 41453 }, { "epoch": 2.2100549128325424, "grad_norm": 2.4839072227478027, "learning_rate": 1.7094000000000003e-06, "loss": 0.0217, "step": 41454 }, { "epoch": 2.2101082262621956, "grad_norm": 2.412686586380005, "learning_rate": 1.7092e-06, "loss": 0.0307, "step": 41455 }, { "epoch": 2.2101615396918484, "grad_norm": 1.013908863067627, "learning_rate": 1.709e-06, "loss": 0.0085, "step": 41456 }, { "epoch": 2.210214853121501, "grad_norm": 1.3155790567398071, "learning_rate": 1.7088000000000003e-06, "loss": 0.0065, "step": 41457 }, { "epoch": 2.2102681665511543, "grad_norm": 2.4615468978881836, "learning_rate": 1.7086000000000003e-06, "loss": 0.0204, "step": 41458 }, { "epoch": 2.210321479980807, "grad_norm": 1.7013094425201416, "learning_rate": 1.7084e-06, "loss": 0.0085, "step": 41459 }, { "epoch": 2.2103747934104603, "grad_norm": 3.5465590953826904, "learning_rate": 1.7082000000000002e-06, "loss": 0.0191, "step": 41460 }, { "epoch": 2.210428106840113, "grad_norm": 1.8160961866378784, "learning_rate": 1.7080000000000002e-06, "loss": 0.0124, "step": 41461 }, { "epoch": 2.210481420269766, "grad_norm": 2.113166332244873, "learning_rate": 1.7078e-06, "loss": 0.0146, "step": 41462 }, { "epoch": 2.210534733699419, "grad_norm": 2.9867472648620605, "learning_rate": 1.7076000000000002e-06, "loss": 0.0173, "step": 41463 }, { "epoch": 2.2105880471290718, "grad_norm": 2.9853641986846924, "learning_rate": 1.7074000000000002e-06, "loss": 0.025, "step": 41464 }, { "epoch": 2.210641360558725, "grad_norm": 2.037886619567871, "learning_rate": 1.7072000000000001e-06, "loss": 0.013, "step": 41465 }, { "epoch": 2.2106946739883777, "grad_norm": 1.9202061891555786, "learning_rate": 1.7070000000000001e-06, "loss": 0.0323, "step": 41466 }, { "epoch": 2.2107479874180305, "grad_norm": 1.8325634002685547, "learning_rate": 1.7068000000000001e-06, "loss": 0.0105, "step": 41467 }, { "epoch": 2.2108013008476837, "grad_norm": 2.191751480102539, "learning_rate": 1.7066e-06, "loss": 0.026, "step": 41468 }, { "epoch": 2.2108546142773364, "grad_norm": 2.925827980041504, "learning_rate": 1.7064e-06, "loss": 0.0153, "step": 41469 }, { "epoch": 2.210907927706989, "grad_norm": 2.3404083251953125, "learning_rate": 1.7062e-06, "loss": 0.0355, "step": 41470 }, { "epoch": 2.2109612411366424, "grad_norm": 1.0446161031723022, "learning_rate": 1.706e-06, "loss": 0.0124, "step": 41471 }, { "epoch": 2.211014554566295, "grad_norm": 2.1265552043914795, "learning_rate": 1.7058000000000003e-06, "loss": 0.0118, "step": 41472 }, { "epoch": 2.2110678679959483, "grad_norm": 1.6605987548828125, "learning_rate": 1.7056e-06, "loss": 0.0215, "step": 41473 }, { "epoch": 2.211121181425601, "grad_norm": 0.22406119108200073, "learning_rate": 1.7054e-06, "loss": 0.0021, "step": 41474 }, { "epoch": 2.211174494855254, "grad_norm": 3.2463161945343018, "learning_rate": 1.7052000000000002e-06, "loss": 0.0336, "step": 41475 }, { "epoch": 2.211227808284907, "grad_norm": 1.1506582498550415, "learning_rate": 1.7050000000000002e-06, "loss": 0.0148, "step": 41476 }, { "epoch": 2.21128112171456, "grad_norm": 1.6053568124771118, "learning_rate": 1.7048e-06, "loss": 0.007, "step": 41477 }, { "epoch": 2.211334435144213, "grad_norm": 1.2715996503829956, "learning_rate": 1.7046000000000002e-06, "loss": 0.0148, "step": 41478 }, { "epoch": 2.2113877485738658, "grad_norm": 0.6569584012031555, "learning_rate": 1.7044000000000001e-06, "loss": 0.0144, "step": 41479 }, { "epoch": 2.2114410620035185, "grad_norm": 1.152169108390808, "learning_rate": 1.7042e-06, "loss": 0.006, "step": 41480 }, { "epoch": 2.2114943754331717, "grad_norm": 2.0804286003112793, "learning_rate": 1.7040000000000001e-06, "loss": 0.0193, "step": 41481 }, { "epoch": 2.2115476888628245, "grad_norm": 2.666949510574341, "learning_rate": 1.7038e-06, "loss": 0.0064, "step": 41482 }, { "epoch": 2.2116010022924772, "grad_norm": 3.6204001903533936, "learning_rate": 1.7036000000000003e-06, "loss": 0.0162, "step": 41483 }, { "epoch": 2.2116543157221304, "grad_norm": 3.2426936626434326, "learning_rate": 1.7034e-06, "loss": 0.0172, "step": 41484 }, { "epoch": 2.211707629151783, "grad_norm": 2.5271201133728027, "learning_rate": 1.7032e-06, "loss": 0.0201, "step": 41485 }, { "epoch": 2.2117609425814364, "grad_norm": 0.9318689107894897, "learning_rate": 1.7030000000000003e-06, "loss": 0.0054, "step": 41486 }, { "epoch": 2.211814256011089, "grad_norm": 1.08235502243042, "learning_rate": 1.7028e-06, "loss": 0.0087, "step": 41487 }, { "epoch": 2.211867569440742, "grad_norm": 3.4162161350250244, "learning_rate": 1.7026e-06, "loss": 0.0425, "step": 41488 }, { "epoch": 2.211920882870395, "grad_norm": 1.4516080617904663, "learning_rate": 1.7024000000000002e-06, "loss": 0.0227, "step": 41489 }, { "epoch": 2.211974196300048, "grad_norm": 0.9480779767036438, "learning_rate": 1.7022000000000002e-06, "loss": 0.0057, "step": 41490 }, { "epoch": 2.212027509729701, "grad_norm": 1.260452389717102, "learning_rate": 1.702e-06, "loss": 0.0174, "step": 41491 }, { "epoch": 2.212080823159354, "grad_norm": 3.3651998043060303, "learning_rate": 1.7018000000000002e-06, "loss": 0.0209, "step": 41492 }, { "epoch": 2.2121341365890066, "grad_norm": 1.6457006931304932, "learning_rate": 1.7016000000000001e-06, "loss": 0.0294, "step": 41493 }, { "epoch": 2.21218745001866, "grad_norm": 2.0768191814422607, "learning_rate": 1.7014000000000003e-06, "loss": 0.01, "step": 41494 }, { "epoch": 2.2122407634483126, "grad_norm": 1.1068756580352783, "learning_rate": 1.7012000000000001e-06, "loss": 0.0141, "step": 41495 }, { "epoch": 2.2122940768779658, "grad_norm": 1.8616055250167847, "learning_rate": 1.701e-06, "loss": 0.0332, "step": 41496 }, { "epoch": 2.2123473903076185, "grad_norm": 2.1761157512664795, "learning_rate": 1.7008000000000003e-06, "loss": 0.0189, "step": 41497 }, { "epoch": 2.2124007037372713, "grad_norm": 2.4390835762023926, "learning_rate": 1.7006e-06, "loss": 0.0133, "step": 41498 }, { "epoch": 2.2124540171669245, "grad_norm": 1.8552051782608032, "learning_rate": 1.7004e-06, "loss": 0.0329, "step": 41499 }, { "epoch": 2.2125073305965772, "grad_norm": 1.0354317426681519, "learning_rate": 1.7002000000000003e-06, "loss": 0.0059, "step": 41500 }, { "epoch": 2.2125606440262304, "grad_norm": 1.3127979040145874, "learning_rate": 1.7000000000000002e-06, "loss": 0.013, "step": 41501 }, { "epoch": 2.212613957455883, "grad_norm": 1.811837911605835, "learning_rate": 1.6998e-06, "loss": 0.032, "step": 41502 }, { "epoch": 2.212667270885536, "grad_norm": 2.973051071166992, "learning_rate": 1.6996000000000002e-06, "loss": 0.0133, "step": 41503 }, { "epoch": 2.212720584315189, "grad_norm": 2.8868560791015625, "learning_rate": 1.6994000000000002e-06, "loss": 0.0243, "step": 41504 }, { "epoch": 2.212773897744842, "grad_norm": 3.2156436443328857, "learning_rate": 1.6992e-06, "loss": 0.0139, "step": 41505 }, { "epoch": 2.2128272111744947, "grad_norm": 0.48691415786743164, "learning_rate": 1.6990000000000002e-06, "loss": 0.003, "step": 41506 }, { "epoch": 2.212880524604148, "grad_norm": 2.07627272605896, "learning_rate": 1.6988000000000002e-06, "loss": 0.0223, "step": 41507 }, { "epoch": 2.2129338380338006, "grad_norm": 3.5869572162628174, "learning_rate": 1.6986000000000003e-06, "loss": 0.0281, "step": 41508 }, { "epoch": 2.212987151463454, "grad_norm": 2.316957950592041, "learning_rate": 1.6984000000000001e-06, "loss": 0.0352, "step": 41509 }, { "epoch": 2.2130404648931066, "grad_norm": 1.5737898349761963, "learning_rate": 1.6982e-06, "loss": 0.027, "step": 41510 }, { "epoch": 2.2130937783227593, "grad_norm": 0.42944130301475525, "learning_rate": 1.6980000000000003e-06, "loss": 0.0034, "step": 41511 }, { "epoch": 2.2131470917524125, "grad_norm": 2.314039707183838, "learning_rate": 1.6978e-06, "loss": 0.0113, "step": 41512 }, { "epoch": 2.2132004051820653, "grad_norm": 2.1585958003997803, "learning_rate": 1.6976e-06, "loss": 0.0283, "step": 41513 }, { "epoch": 2.2132537186117185, "grad_norm": 1.665061354637146, "learning_rate": 1.6974000000000003e-06, "loss": 0.0067, "step": 41514 }, { "epoch": 2.2133070320413712, "grad_norm": 1.4953147172927856, "learning_rate": 1.6972000000000002e-06, "loss": 0.0134, "step": 41515 }, { "epoch": 2.213360345471024, "grad_norm": 1.0976582765579224, "learning_rate": 1.697e-06, "loss": 0.0079, "step": 41516 }, { "epoch": 2.213413658900677, "grad_norm": 1.6863654851913452, "learning_rate": 1.6968000000000002e-06, "loss": 0.0148, "step": 41517 }, { "epoch": 2.21346697233033, "grad_norm": 0.4102518558502197, "learning_rate": 1.6966000000000002e-06, "loss": 0.003, "step": 41518 }, { "epoch": 2.2135202857599827, "grad_norm": 3.452150821685791, "learning_rate": 1.6964000000000002e-06, "loss": 0.0173, "step": 41519 }, { "epoch": 2.213573599189636, "grad_norm": 3.468489408493042, "learning_rate": 1.6962000000000002e-06, "loss": 0.0335, "step": 41520 }, { "epoch": 2.2136269126192887, "grad_norm": 0.8866876363754272, "learning_rate": 1.6960000000000002e-06, "loss": 0.0055, "step": 41521 }, { "epoch": 2.213680226048942, "grad_norm": 0.6432567834854126, "learning_rate": 1.6958000000000001e-06, "loss": 0.0038, "step": 41522 }, { "epoch": 2.2137335394785946, "grad_norm": 2.3215208053588867, "learning_rate": 1.6956000000000001e-06, "loss": 0.0376, "step": 41523 }, { "epoch": 2.2137868529082474, "grad_norm": 0.9323928952217102, "learning_rate": 1.6954000000000001e-06, "loss": 0.0087, "step": 41524 }, { "epoch": 2.2138401663379006, "grad_norm": 0.2815984785556793, "learning_rate": 1.6952e-06, "loss": 0.0024, "step": 41525 }, { "epoch": 2.2138934797675534, "grad_norm": 2.395268678665161, "learning_rate": 1.6950000000000003e-06, "loss": 0.0194, "step": 41526 }, { "epoch": 2.2139467931972066, "grad_norm": 1.0279428958892822, "learning_rate": 1.6948e-06, "loss": 0.0211, "step": 41527 }, { "epoch": 2.2140001066268593, "grad_norm": 2.3935577869415283, "learning_rate": 1.6946e-06, "loss": 0.0362, "step": 41528 }, { "epoch": 2.214053420056512, "grad_norm": 1.3379757404327393, "learning_rate": 1.6944000000000002e-06, "loss": 0.0107, "step": 41529 }, { "epoch": 2.2141067334861653, "grad_norm": 1.1013636589050293, "learning_rate": 1.6942e-06, "loss": 0.0091, "step": 41530 }, { "epoch": 2.214160046915818, "grad_norm": 0.5784342885017395, "learning_rate": 1.694e-06, "loss": 0.0038, "step": 41531 }, { "epoch": 2.2142133603454712, "grad_norm": 1.4888137578964233, "learning_rate": 1.6938000000000002e-06, "loss": 0.016, "step": 41532 }, { "epoch": 2.214266673775124, "grad_norm": 3.69608736038208, "learning_rate": 1.6936000000000002e-06, "loss": 0.0322, "step": 41533 }, { "epoch": 2.2143199872047767, "grad_norm": 0.4596143066883087, "learning_rate": 1.6934e-06, "loss": 0.0032, "step": 41534 }, { "epoch": 2.21437330063443, "grad_norm": 0.504290759563446, "learning_rate": 1.6932000000000002e-06, "loss": 0.0044, "step": 41535 }, { "epoch": 2.2144266140640827, "grad_norm": 0.9812433123588562, "learning_rate": 1.6930000000000001e-06, "loss": 0.0223, "step": 41536 }, { "epoch": 2.214479927493736, "grad_norm": 1.4500658512115479, "learning_rate": 1.6928000000000003e-06, "loss": 0.0098, "step": 41537 }, { "epoch": 2.2145332409233887, "grad_norm": 2.243381977081299, "learning_rate": 1.6926000000000001e-06, "loss": 0.0111, "step": 41538 }, { "epoch": 2.2145865543530414, "grad_norm": 1.8908132314682007, "learning_rate": 1.6924e-06, "loss": 0.01, "step": 41539 }, { "epoch": 2.2146398677826946, "grad_norm": 2.877671718597412, "learning_rate": 1.6922000000000003e-06, "loss": 0.0444, "step": 41540 }, { "epoch": 2.2146931812123474, "grad_norm": 0.6471024751663208, "learning_rate": 1.692e-06, "loss": 0.0043, "step": 41541 }, { "epoch": 2.214746494642, "grad_norm": 1.643757939338684, "learning_rate": 1.6918e-06, "loss": 0.0208, "step": 41542 }, { "epoch": 2.2147998080716533, "grad_norm": 2.4029922485351562, "learning_rate": 1.6916000000000002e-06, "loss": 0.0412, "step": 41543 }, { "epoch": 2.214853121501306, "grad_norm": 1.6843503713607788, "learning_rate": 1.6914000000000002e-06, "loss": 0.0285, "step": 41544 }, { "epoch": 2.2149064349309593, "grad_norm": 1.2305681705474854, "learning_rate": 1.6912e-06, "loss": 0.0083, "step": 41545 }, { "epoch": 2.214959748360612, "grad_norm": 1.6188799142837524, "learning_rate": 1.6910000000000002e-06, "loss": 0.039, "step": 41546 }, { "epoch": 2.215013061790265, "grad_norm": 0.42796945571899414, "learning_rate": 1.6908000000000002e-06, "loss": 0.0032, "step": 41547 }, { "epoch": 2.215066375219918, "grad_norm": 1.0075913667678833, "learning_rate": 1.6906e-06, "loss": 0.0077, "step": 41548 }, { "epoch": 2.2151196886495708, "grad_norm": 1.796396255493164, "learning_rate": 1.6904000000000002e-06, "loss": 0.0223, "step": 41549 }, { "epoch": 2.215173002079224, "grad_norm": 1.9353543519973755, "learning_rate": 1.6902000000000001e-06, "loss": 0.0252, "step": 41550 }, { "epoch": 2.2152263155088767, "grad_norm": 1.4267356395721436, "learning_rate": 1.6900000000000003e-06, "loss": 0.0143, "step": 41551 }, { "epoch": 2.2152796289385295, "grad_norm": 2.727646589279175, "learning_rate": 1.6898000000000001e-06, "loss": 0.0236, "step": 41552 }, { "epoch": 2.2153329423681827, "grad_norm": 1.972429633140564, "learning_rate": 1.6896e-06, "loss": 0.029, "step": 41553 }, { "epoch": 2.2153862557978354, "grad_norm": 0.760536253452301, "learning_rate": 1.6894000000000003e-06, "loss": 0.0045, "step": 41554 }, { "epoch": 2.215439569227488, "grad_norm": 1.7078953981399536, "learning_rate": 1.6892e-06, "loss": 0.026, "step": 41555 }, { "epoch": 2.2154928826571414, "grad_norm": 2.045912265777588, "learning_rate": 1.689e-06, "loss": 0.0131, "step": 41556 }, { "epoch": 2.215546196086794, "grad_norm": 1.080684781074524, "learning_rate": 1.6888000000000002e-06, "loss": 0.0206, "step": 41557 }, { "epoch": 2.2155995095164474, "grad_norm": 2.5616631507873535, "learning_rate": 1.6886000000000002e-06, "loss": 0.0216, "step": 41558 }, { "epoch": 2.2156528229461, "grad_norm": 0.8873342871665955, "learning_rate": 1.6884e-06, "loss": 0.0054, "step": 41559 }, { "epoch": 2.215706136375753, "grad_norm": 0.7120257616043091, "learning_rate": 1.6882000000000002e-06, "loss": 0.0035, "step": 41560 }, { "epoch": 2.215759449805406, "grad_norm": 1.6451798677444458, "learning_rate": 1.6880000000000002e-06, "loss": 0.0199, "step": 41561 }, { "epoch": 2.215812763235059, "grad_norm": 2.107799530029297, "learning_rate": 1.6878000000000002e-06, "loss": 0.0508, "step": 41562 }, { "epoch": 2.215866076664712, "grad_norm": 3.499079942703247, "learning_rate": 1.6876000000000002e-06, "loss": 0.0604, "step": 41563 }, { "epoch": 2.215919390094365, "grad_norm": 2.1275763511657715, "learning_rate": 1.6874000000000001e-06, "loss": 0.0165, "step": 41564 }, { "epoch": 2.2159727035240175, "grad_norm": 2.5116426944732666, "learning_rate": 1.6872000000000001e-06, "loss": 0.0145, "step": 41565 }, { "epoch": 2.2160260169536707, "grad_norm": 0.5269268155097961, "learning_rate": 1.6870000000000001e-06, "loss": 0.0038, "step": 41566 }, { "epoch": 2.2160793303833235, "grad_norm": 1.525245189666748, "learning_rate": 1.6868e-06, "loss": 0.018, "step": 41567 }, { "epoch": 2.2161326438129763, "grad_norm": 2.8604111671447754, "learning_rate": 1.6866e-06, "loss": 0.0153, "step": 41568 }, { "epoch": 2.2161859572426295, "grad_norm": 2.5052950382232666, "learning_rate": 1.6864000000000003e-06, "loss": 0.0346, "step": 41569 }, { "epoch": 2.216239270672282, "grad_norm": 1.3762199878692627, "learning_rate": 1.6862e-06, "loss": 0.0196, "step": 41570 }, { "epoch": 2.2162925841019354, "grad_norm": 2.031235456466675, "learning_rate": 1.686e-06, "loss": 0.0094, "step": 41571 }, { "epoch": 2.216345897531588, "grad_norm": 1.3531625270843506, "learning_rate": 1.6858000000000002e-06, "loss": 0.0192, "step": 41572 }, { "epoch": 2.216399210961241, "grad_norm": 2.297708749771118, "learning_rate": 1.6856e-06, "loss": 0.0386, "step": 41573 }, { "epoch": 2.216452524390894, "grad_norm": 1.58872652053833, "learning_rate": 1.6854e-06, "loss": 0.0106, "step": 41574 }, { "epoch": 2.216505837820547, "grad_norm": 1.9376611709594727, "learning_rate": 1.6852000000000002e-06, "loss": 0.0121, "step": 41575 }, { "epoch": 2.2165591512502, "grad_norm": 2.1833653450012207, "learning_rate": 1.6850000000000002e-06, "loss": 0.0129, "step": 41576 }, { "epoch": 2.216612464679853, "grad_norm": 2.484652280807495, "learning_rate": 1.6848e-06, "loss": 0.0149, "step": 41577 }, { "epoch": 2.2166657781095056, "grad_norm": 2.0120651721954346, "learning_rate": 1.6846000000000001e-06, "loss": 0.0095, "step": 41578 }, { "epoch": 2.216719091539159, "grad_norm": 1.675461769104004, "learning_rate": 1.6844000000000001e-06, "loss": 0.0098, "step": 41579 }, { "epoch": 2.2167724049688116, "grad_norm": 2.1206295490264893, "learning_rate": 1.6842e-06, "loss": 0.012, "step": 41580 }, { "epoch": 2.2168257183984648, "grad_norm": 1.9493366479873657, "learning_rate": 1.684e-06, "loss": 0.0396, "step": 41581 }, { "epoch": 2.2168790318281175, "grad_norm": 1.0565273761749268, "learning_rate": 1.6838e-06, "loss": 0.0204, "step": 41582 }, { "epoch": 2.2169323452577703, "grad_norm": 2.1342551708221436, "learning_rate": 1.6836000000000003e-06, "loss": 0.043, "step": 41583 }, { "epoch": 2.2169856586874235, "grad_norm": 2.424258232116699, "learning_rate": 1.6834e-06, "loss": 0.0176, "step": 41584 }, { "epoch": 2.2170389721170762, "grad_norm": 2.013766050338745, "learning_rate": 1.6832e-06, "loss": 0.0165, "step": 41585 }, { "epoch": 2.2170922855467294, "grad_norm": 0.9181773066520691, "learning_rate": 1.6830000000000002e-06, "loss": 0.0186, "step": 41586 }, { "epoch": 2.217145598976382, "grad_norm": 1.6845840215682983, "learning_rate": 1.6828000000000002e-06, "loss": 0.0128, "step": 41587 }, { "epoch": 2.217198912406035, "grad_norm": 2.3444254398345947, "learning_rate": 1.6826e-06, "loss": 0.0225, "step": 41588 }, { "epoch": 2.217252225835688, "grad_norm": 1.293277382850647, "learning_rate": 1.6824000000000002e-06, "loss": 0.0113, "step": 41589 }, { "epoch": 2.217305539265341, "grad_norm": 1.0713293552398682, "learning_rate": 1.6822000000000002e-06, "loss": 0.0124, "step": 41590 }, { "epoch": 2.2173588526949937, "grad_norm": 1.9341357946395874, "learning_rate": 1.682e-06, "loss": 0.023, "step": 41591 }, { "epoch": 2.217412166124647, "grad_norm": 2.0524399280548096, "learning_rate": 1.6818000000000001e-06, "loss": 0.0158, "step": 41592 }, { "epoch": 2.2174654795542996, "grad_norm": 2.9619364738464355, "learning_rate": 1.6816000000000001e-06, "loss": 0.0369, "step": 41593 }, { "epoch": 2.217518792983953, "grad_norm": 1.896454930305481, "learning_rate": 1.6814000000000003e-06, "loss": 0.0189, "step": 41594 }, { "epoch": 2.2175721064136056, "grad_norm": 2.0192549228668213, "learning_rate": 1.6812e-06, "loss": 0.0254, "step": 41595 }, { "epoch": 2.2176254198432583, "grad_norm": 1.9747035503387451, "learning_rate": 1.681e-06, "loss": 0.0167, "step": 41596 }, { "epoch": 2.2176787332729115, "grad_norm": 4.170199394226074, "learning_rate": 1.6808000000000003e-06, "loss": 0.0179, "step": 41597 }, { "epoch": 2.2177320467025643, "grad_norm": 2.2113945484161377, "learning_rate": 1.6806e-06, "loss": 0.0101, "step": 41598 }, { "epoch": 2.2177853601322175, "grad_norm": 2.127769947052002, "learning_rate": 1.6804e-06, "loss": 0.024, "step": 41599 }, { "epoch": 2.2178386735618703, "grad_norm": 0.9824264049530029, "learning_rate": 1.6802000000000002e-06, "loss": 0.0051, "step": 41600 }, { "epoch": 2.217891986991523, "grad_norm": 3.92069149017334, "learning_rate": 1.6800000000000002e-06, "loss": 0.0169, "step": 41601 }, { "epoch": 2.217945300421176, "grad_norm": 1.2089163064956665, "learning_rate": 1.6798e-06, "loss": 0.0199, "step": 41602 }, { "epoch": 2.217998613850829, "grad_norm": 1.5086969137191772, "learning_rate": 1.6796000000000002e-06, "loss": 0.0069, "step": 41603 }, { "epoch": 2.2180519272804817, "grad_norm": 5.373552322387695, "learning_rate": 1.6794000000000002e-06, "loss": 0.0379, "step": 41604 }, { "epoch": 2.218105240710135, "grad_norm": 1.2897697687149048, "learning_rate": 1.6792000000000004e-06, "loss": 0.0211, "step": 41605 }, { "epoch": 2.2181585541397877, "grad_norm": 0.6710918545722961, "learning_rate": 1.6790000000000001e-06, "loss": 0.0048, "step": 41606 }, { "epoch": 2.218211867569441, "grad_norm": 1.1127986907958984, "learning_rate": 1.6788000000000001e-06, "loss": 0.0051, "step": 41607 }, { "epoch": 2.2182651809990936, "grad_norm": 1.127632975578308, "learning_rate": 1.6786000000000003e-06, "loss": 0.0235, "step": 41608 }, { "epoch": 2.2183184944287464, "grad_norm": 3.351869821548462, "learning_rate": 1.6784e-06, "loss": 0.0667, "step": 41609 }, { "epoch": 2.2183718078583996, "grad_norm": 2.0422818660736084, "learning_rate": 1.6782e-06, "loss": 0.0183, "step": 41610 }, { "epoch": 2.2184251212880524, "grad_norm": 0.6899304389953613, "learning_rate": 1.6780000000000003e-06, "loss": 0.0024, "step": 41611 }, { "epoch": 2.2184784347177056, "grad_norm": 1.7122156620025635, "learning_rate": 1.6778000000000003e-06, "loss": 0.0156, "step": 41612 }, { "epoch": 2.2185317481473583, "grad_norm": 2.045664072036743, "learning_rate": 1.6776e-06, "loss": 0.0097, "step": 41613 }, { "epoch": 2.218585061577011, "grad_norm": 1.193079948425293, "learning_rate": 1.6774000000000002e-06, "loss": 0.0183, "step": 41614 }, { "epoch": 2.2186383750066643, "grad_norm": 1.2092891931533813, "learning_rate": 1.6772000000000002e-06, "loss": 0.0131, "step": 41615 }, { "epoch": 2.218691688436317, "grad_norm": 3.0094645023345947, "learning_rate": 1.677e-06, "loss": 0.049, "step": 41616 }, { "epoch": 2.2187450018659702, "grad_norm": 1.1154108047485352, "learning_rate": 1.6768000000000002e-06, "loss": 0.0169, "step": 41617 }, { "epoch": 2.218798315295623, "grad_norm": 1.6993563175201416, "learning_rate": 1.6766000000000002e-06, "loss": 0.0207, "step": 41618 }, { "epoch": 2.2188516287252757, "grad_norm": 1.4058277606964111, "learning_rate": 1.6764000000000002e-06, "loss": 0.0087, "step": 41619 }, { "epoch": 2.218904942154929, "grad_norm": 2.896756887435913, "learning_rate": 1.6762000000000001e-06, "loss": 0.027, "step": 41620 }, { "epoch": 2.2189582555845817, "grad_norm": 1.0859532356262207, "learning_rate": 1.6760000000000001e-06, "loss": 0.0048, "step": 41621 }, { "epoch": 2.219011569014235, "grad_norm": 1.6299329996109009, "learning_rate": 1.6758000000000001e-06, "loss": 0.0169, "step": 41622 }, { "epoch": 2.2190648824438877, "grad_norm": 2.755478858947754, "learning_rate": 1.6756e-06, "loss": 0.0276, "step": 41623 }, { "epoch": 2.2191181958735404, "grad_norm": 1.2924827337265015, "learning_rate": 1.6754e-06, "loss": 0.0168, "step": 41624 }, { "epoch": 2.2191715093031936, "grad_norm": 1.7931143045425415, "learning_rate": 1.6752e-06, "loss": 0.0265, "step": 41625 }, { "epoch": 2.2192248227328464, "grad_norm": 1.6145720481872559, "learning_rate": 1.6750000000000003e-06, "loss": 0.0096, "step": 41626 }, { "epoch": 2.219278136162499, "grad_norm": 1.0455950498580933, "learning_rate": 1.6748e-06, "loss": 0.0056, "step": 41627 }, { "epoch": 2.2193314495921523, "grad_norm": 1.8699008226394653, "learning_rate": 1.6746e-06, "loss": 0.0178, "step": 41628 }, { "epoch": 2.219384763021805, "grad_norm": 3.07065486907959, "learning_rate": 1.6744000000000002e-06, "loss": 0.009, "step": 41629 }, { "epoch": 2.2194380764514583, "grad_norm": 1.5763949155807495, "learning_rate": 1.6742000000000002e-06, "loss": 0.0092, "step": 41630 }, { "epoch": 2.219491389881111, "grad_norm": 1.0337530374526978, "learning_rate": 1.674e-06, "loss": 0.0073, "step": 41631 }, { "epoch": 2.219544703310764, "grad_norm": 0.5598121285438538, "learning_rate": 1.6738000000000002e-06, "loss": 0.0032, "step": 41632 }, { "epoch": 2.219598016740417, "grad_norm": 2.788639783859253, "learning_rate": 1.6736000000000002e-06, "loss": 0.0089, "step": 41633 }, { "epoch": 2.2196513301700698, "grad_norm": 0.5995733141899109, "learning_rate": 1.6734e-06, "loss": 0.0046, "step": 41634 }, { "epoch": 2.219704643599723, "grad_norm": 3.3638951778411865, "learning_rate": 1.6732000000000001e-06, "loss": 0.0097, "step": 41635 }, { "epoch": 2.2197579570293757, "grad_norm": 1.6470859050750732, "learning_rate": 1.6730000000000001e-06, "loss": 0.0103, "step": 41636 }, { "epoch": 2.2198112704590285, "grad_norm": 0.5743711590766907, "learning_rate": 1.6728000000000003e-06, "loss": 0.0043, "step": 41637 }, { "epoch": 2.2198645838886817, "grad_norm": 2.0954928398132324, "learning_rate": 1.6726e-06, "loss": 0.0097, "step": 41638 }, { "epoch": 2.2199178973183344, "grad_norm": 2.6317572593688965, "learning_rate": 1.6724e-06, "loss": 0.0296, "step": 41639 }, { "epoch": 2.219971210747987, "grad_norm": 6.790988445281982, "learning_rate": 1.6722000000000003e-06, "loss": 0.023, "step": 41640 }, { "epoch": 2.2200245241776404, "grad_norm": 2.0722742080688477, "learning_rate": 1.672e-06, "loss": 0.0479, "step": 41641 }, { "epoch": 2.220077837607293, "grad_norm": 1.4785151481628418, "learning_rate": 1.6718e-06, "loss": 0.0316, "step": 41642 }, { "epoch": 2.2201311510369464, "grad_norm": 0.36032867431640625, "learning_rate": 1.6716000000000002e-06, "loss": 0.0032, "step": 41643 }, { "epoch": 2.220184464466599, "grad_norm": 1.0476362705230713, "learning_rate": 1.6714000000000002e-06, "loss": 0.0062, "step": 41644 }, { "epoch": 2.220237777896252, "grad_norm": 1.106322169303894, "learning_rate": 1.6712e-06, "loss": 0.0103, "step": 41645 }, { "epoch": 2.220291091325905, "grad_norm": 5.3689866065979, "learning_rate": 1.6710000000000002e-06, "loss": 0.0265, "step": 41646 }, { "epoch": 2.220344404755558, "grad_norm": 1.238510251045227, "learning_rate": 1.6708000000000002e-06, "loss": 0.0067, "step": 41647 }, { "epoch": 2.220397718185211, "grad_norm": 1.5667030811309814, "learning_rate": 1.6706e-06, "loss": 0.0072, "step": 41648 }, { "epoch": 2.220451031614864, "grad_norm": 3.3771393299102783, "learning_rate": 1.6704000000000001e-06, "loss": 0.038, "step": 41649 }, { "epoch": 2.2205043450445165, "grad_norm": 0.8756487965583801, "learning_rate": 1.6702000000000001e-06, "loss": 0.005, "step": 41650 }, { "epoch": 2.2205576584741697, "grad_norm": 0.7391825318336487, "learning_rate": 1.6700000000000003e-06, "loss": 0.0148, "step": 41651 }, { "epoch": 2.2206109719038225, "grad_norm": 0.24057170748710632, "learning_rate": 1.6698e-06, "loss": 0.0024, "step": 41652 }, { "epoch": 2.2206642853334757, "grad_norm": 2.6092369556427, "learning_rate": 1.6696e-06, "loss": 0.0162, "step": 41653 }, { "epoch": 2.2207175987631285, "grad_norm": 2.2512075901031494, "learning_rate": 1.6694000000000003e-06, "loss": 0.0439, "step": 41654 }, { "epoch": 2.220770912192781, "grad_norm": 1.8808834552764893, "learning_rate": 1.6692000000000003e-06, "loss": 0.021, "step": 41655 }, { "epoch": 2.2208242256224344, "grad_norm": 0.19426976144313812, "learning_rate": 1.669e-06, "loss": 0.002, "step": 41656 }, { "epoch": 2.220877539052087, "grad_norm": 1.0208181142807007, "learning_rate": 1.6688000000000002e-06, "loss": 0.0053, "step": 41657 }, { "epoch": 2.2209308524817404, "grad_norm": 1.6732558012008667, "learning_rate": 1.6686000000000002e-06, "loss": 0.0107, "step": 41658 }, { "epoch": 2.220984165911393, "grad_norm": 0.37949883937835693, "learning_rate": 1.6684e-06, "loss": 0.0031, "step": 41659 }, { "epoch": 2.221037479341046, "grad_norm": 2.9227356910705566, "learning_rate": 1.6682000000000002e-06, "loss": 0.0212, "step": 41660 }, { "epoch": 2.221090792770699, "grad_norm": 2.3989834785461426, "learning_rate": 1.6680000000000002e-06, "loss": 0.0085, "step": 41661 }, { "epoch": 2.221144106200352, "grad_norm": 0.8760321736335754, "learning_rate": 1.6678000000000002e-06, "loss": 0.0032, "step": 41662 }, { "epoch": 2.2211974196300046, "grad_norm": 1.4184921979904175, "learning_rate": 1.6676000000000001e-06, "loss": 0.0359, "step": 41663 }, { "epoch": 2.221250733059658, "grad_norm": 2.3046987056732178, "learning_rate": 1.6674000000000001e-06, "loss": 0.0106, "step": 41664 }, { "epoch": 2.2213040464893106, "grad_norm": 2.279933452606201, "learning_rate": 1.6672e-06, "loss": 0.0437, "step": 41665 }, { "epoch": 2.2213573599189638, "grad_norm": 0.40334823727607727, "learning_rate": 1.667e-06, "loss": 0.003, "step": 41666 }, { "epoch": 2.2214106733486165, "grad_norm": 0.6609383225440979, "learning_rate": 1.6668e-06, "loss": 0.0048, "step": 41667 }, { "epoch": 2.2214639867782693, "grad_norm": 2.060703992843628, "learning_rate": 1.6666e-06, "loss": 0.0132, "step": 41668 }, { "epoch": 2.2215173002079225, "grad_norm": 1.7150706052780151, "learning_rate": 1.6664000000000003e-06, "loss": 0.0153, "step": 41669 }, { "epoch": 2.2215706136375752, "grad_norm": 1.6138949394226074, "learning_rate": 1.6662e-06, "loss": 0.041, "step": 41670 }, { "epoch": 2.2216239270672284, "grad_norm": 2.966663360595703, "learning_rate": 1.666e-06, "loss": 0.025, "step": 41671 }, { "epoch": 2.221677240496881, "grad_norm": 3.4260189533233643, "learning_rate": 1.6658000000000002e-06, "loss": 0.0307, "step": 41672 }, { "epoch": 2.221730553926534, "grad_norm": 2.484407901763916, "learning_rate": 1.6656000000000002e-06, "loss": 0.0286, "step": 41673 }, { "epoch": 2.221783867356187, "grad_norm": 1.0591846704483032, "learning_rate": 1.6654e-06, "loss": 0.0066, "step": 41674 }, { "epoch": 2.22183718078584, "grad_norm": 2.6609442234039307, "learning_rate": 1.6652000000000002e-06, "loss": 0.0492, "step": 41675 }, { "epoch": 2.2218904942154927, "grad_norm": 0.5628983974456787, "learning_rate": 1.6650000000000002e-06, "loss": 0.0043, "step": 41676 }, { "epoch": 2.221943807645146, "grad_norm": 4.4673357009887695, "learning_rate": 1.6648e-06, "loss": 0.0472, "step": 41677 }, { "epoch": 2.2219971210747986, "grad_norm": 3.13905930519104, "learning_rate": 1.6646000000000001e-06, "loss": 0.0275, "step": 41678 }, { "epoch": 2.222050434504452, "grad_norm": 1.3033840656280518, "learning_rate": 1.6644000000000001e-06, "loss": 0.0097, "step": 41679 }, { "epoch": 2.2221037479341046, "grad_norm": 3.192092180252075, "learning_rate": 1.6642000000000003e-06, "loss": 0.0271, "step": 41680 }, { "epoch": 2.2221570613637573, "grad_norm": 2.1827640533447266, "learning_rate": 1.664e-06, "loss": 0.0196, "step": 41681 }, { "epoch": 2.2222103747934105, "grad_norm": 4.558976173400879, "learning_rate": 1.6638e-06, "loss": 0.0153, "step": 41682 }, { "epoch": 2.2222636882230633, "grad_norm": 2.1966660022735596, "learning_rate": 1.6636000000000003e-06, "loss": 0.0234, "step": 41683 }, { "epoch": 2.2223170016527165, "grad_norm": 2.245882272720337, "learning_rate": 1.6634e-06, "loss": 0.0146, "step": 41684 }, { "epoch": 2.2223703150823693, "grad_norm": 2.430680990219116, "learning_rate": 1.6632e-06, "loss": 0.0105, "step": 41685 }, { "epoch": 2.222423628512022, "grad_norm": 1.4430656433105469, "learning_rate": 1.6630000000000002e-06, "loss": 0.0237, "step": 41686 }, { "epoch": 2.222476941941675, "grad_norm": 0.4660923182964325, "learning_rate": 1.6628000000000002e-06, "loss": 0.0037, "step": 41687 }, { "epoch": 2.222530255371328, "grad_norm": 1.7434896230697632, "learning_rate": 1.6626e-06, "loss": 0.0251, "step": 41688 }, { "epoch": 2.2225835688009807, "grad_norm": 0.2134580910205841, "learning_rate": 1.6624000000000002e-06, "loss": 0.0025, "step": 41689 }, { "epoch": 2.222636882230634, "grad_norm": 1.0232133865356445, "learning_rate": 1.6622000000000002e-06, "loss": 0.0242, "step": 41690 }, { "epoch": 2.2226901956602867, "grad_norm": 1.8826841115951538, "learning_rate": 1.662e-06, "loss": 0.0179, "step": 41691 }, { "epoch": 2.22274350908994, "grad_norm": 1.5459297895431519, "learning_rate": 1.6618000000000001e-06, "loss": 0.0184, "step": 41692 }, { "epoch": 2.2227968225195927, "grad_norm": 1.9393212795257568, "learning_rate": 1.6616000000000001e-06, "loss": 0.0148, "step": 41693 }, { "epoch": 2.2228501359492454, "grad_norm": 2.1331369876861572, "learning_rate": 1.6614000000000003e-06, "loss": 0.0185, "step": 41694 }, { "epoch": 2.2229034493788986, "grad_norm": 0.9214309453964233, "learning_rate": 1.6612e-06, "loss": 0.0073, "step": 41695 }, { "epoch": 2.2229567628085514, "grad_norm": 0.8451663255691528, "learning_rate": 1.661e-06, "loss": 0.0138, "step": 41696 }, { "epoch": 2.2230100762382046, "grad_norm": 3.0197038650512695, "learning_rate": 1.6608000000000003e-06, "loss": 0.0377, "step": 41697 }, { "epoch": 2.2230633896678573, "grad_norm": 2.759091854095459, "learning_rate": 1.6606000000000002e-06, "loss": 0.0142, "step": 41698 }, { "epoch": 2.22311670309751, "grad_norm": 0.2964855432510376, "learning_rate": 1.6604e-06, "loss": 0.0026, "step": 41699 }, { "epoch": 2.2231700165271633, "grad_norm": 0.8441362380981445, "learning_rate": 1.6602000000000002e-06, "loss": 0.0043, "step": 41700 }, { "epoch": 2.223223329956816, "grad_norm": 0.3708479106426239, "learning_rate": 1.6600000000000002e-06, "loss": 0.0034, "step": 41701 }, { "epoch": 2.2232766433864692, "grad_norm": 2.0555620193481445, "learning_rate": 1.6598e-06, "loss": 0.0267, "step": 41702 }, { "epoch": 2.223329956816122, "grad_norm": 1.7456282377243042, "learning_rate": 1.6596000000000002e-06, "loss": 0.024, "step": 41703 }, { "epoch": 2.2233832702457748, "grad_norm": 0.9913156032562256, "learning_rate": 1.6594000000000002e-06, "loss": 0.006, "step": 41704 }, { "epoch": 2.223436583675428, "grad_norm": 2.7120211124420166, "learning_rate": 1.6592000000000004e-06, "loss": 0.0159, "step": 41705 }, { "epoch": 2.2234898971050807, "grad_norm": 3.1363677978515625, "learning_rate": 1.6590000000000001e-06, "loss": 0.0254, "step": 41706 }, { "epoch": 2.223543210534734, "grad_norm": 1.587157130241394, "learning_rate": 1.6588000000000001e-06, "loss": 0.009, "step": 41707 }, { "epoch": 2.2235965239643867, "grad_norm": 1.6374120712280273, "learning_rate": 1.6586000000000003e-06, "loss": 0.0079, "step": 41708 }, { "epoch": 2.2236498373940394, "grad_norm": 1.2643331289291382, "learning_rate": 1.6584e-06, "loss": 0.0069, "step": 41709 }, { "epoch": 2.2237031508236926, "grad_norm": 1.5856996774673462, "learning_rate": 1.6582e-06, "loss": 0.0115, "step": 41710 }, { "epoch": 2.2237564642533454, "grad_norm": 1.2314380407333374, "learning_rate": 1.6580000000000003e-06, "loss": 0.0077, "step": 41711 }, { "epoch": 2.223809777682998, "grad_norm": 0.8966167569160461, "learning_rate": 1.6578000000000002e-06, "loss": 0.023, "step": 41712 }, { "epoch": 2.2238630911126513, "grad_norm": 0.8998309969902039, "learning_rate": 1.6576e-06, "loss": 0.0065, "step": 41713 }, { "epoch": 2.223916404542304, "grad_norm": 1.8835110664367676, "learning_rate": 1.6574000000000002e-06, "loss": 0.0051, "step": 41714 }, { "epoch": 2.2239697179719573, "grad_norm": 3.236811399459839, "learning_rate": 1.6572000000000002e-06, "loss": 0.044, "step": 41715 }, { "epoch": 2.22402303140161, "grad_norm": 1.5131489038467407, "learning_rate": 1.657e-06, "loss": 0.0283, "step": 41716 }, { "epoch": 2.224076344831263, "grad_norm": 0.668475329875946, "learning_rate": 1.6568000000000002e-06, "loss": 0.0034, "step": 41717 }, { "epoch": 2.224129658260916, "grad_norm": 0.49294513463974, "learning_rate": 1.6566000000000002e-06, "loss": 0.0041, "step": 41718 }, { "epoch": 2.2241829716905688, "grad_norm": 0.3112989068031311, "learning_rate": 1.6564000000000001e-06, "loss": 0.0026, "step": 41719 }, { "epoch": 2.224236285120222, "grad_norm": 1.2157256603240967, "learning_rate": 1.6562000000000001e-06, "loss": 0.0074, "step": 41720 }, { "epoch": 2.2242895985498747, "grad_norm": 1.8509271144866943, "learning_rate": 1.6560000000000001e-06, "loss": 0.0172, "step": 41721 }, { "epoch": 2.2243429119795275, "grad_norm": 1.8021446466445923, "learning_rate": 1.6558e-06, "loss": 0.0077, "step": 41722 }, { "epoch": 2.2243962254091807, "grad_norm": 1.8827937841415405, "learning_rate": 1.6556000000000003e-06, "loss": 0.0109, "step": 41723 }, { "epoch": 2.2244495388388335, "grad_norm": 1.4283171892166138, "learning_rate": 1.6554e-06, "loss": 0.0074, "step": 41724 }, { "epoch": 2.224502852268486, "grad_norm": 1.2478636503219604, "learning_rate": 1.6552e-06, "loss": 0.034, "step": 41725 }, { "epoch": 2.2245561656981394, "grad_norm": 0.9146846532821655, "learning_rate": 1.6550000000000002e-06, "loss": 0.0074, "step": 41726 }, { "epoch": 2.224609479127792, "grad_norm": 0.583021879196167, "learning_rate": 1.6548e-06, "loss": 0.0042, "step": 41727 }, { "epoch": 2.2246627925574454, "grad_norm": 2.684252977371216, "learning_rate": 1.6546e-06, "loss": 0.0075, "step": 41728 }, { "epoch": 2.224716105987098, "grad_norm": 1.3398327827453613, "learning_rate": 1.6544000000000002e-06, "loss": 0.0135, "step": 41729 }, { "epoch": 2.224769419416751, "grad_norm": 4.039025783538818, "learning_rate": 1.6542000000000002e-06, "loss": 0.0357, "step": 41730 }, { "epoch": 2.224822732846404, "grad_norm": 1.0914448499679565, "learning_rate": 1.654e-06, "loss": 0.0176, "step": 41731 }, { "epoch": 2.224876046276057, "grad_norm": 0.946871817111969, "learning_rate": 1.6538000000000002e-06, "loss": 0.0067, "step": 41732 }, { "epoch": 2.22492935970571, "grad_norm": 0.8961591124534607, "learning_rate": 1.6536000000000001e-06, "loss": 0.0334, "step": 41733 }, { "epoch": 2.224982673135363, "grad_norm": 2.5904386043548584, "learning_rate": 1.6534e-06, "loss": 0.0261, "step": 41734 }, { "epoch": 2.2250359865650156, "grad_norm": 3.2248458862304688, "learning_rate": 1.6532000000000001e-06, "loss": 0.0251, "step": 41735 }, { "epoch": 2.2250892999946688, "grad_norm": 2.5777652263641357, "learning_rate": 1.653e-06, "loss": 0.0133, "step": 41736 }, { "epoch": 2.2251426134243215, "grad_norm": 3.2810370922088623, "learning_rate": 1.6528000000000003e-06, "loss": 0.0261, "step": 41737 }, { "epoch": 2.2251959268539747, "grad_norm": 2.0318453311920166, "learning_rate": 1.6526e-06, "loss": 0.0325, "step": 41738 }, { "epoch": 2.2252492402836275, "grad_norm": 1.8491442203521729, "learning_rate": 1.6524e-06, "loss": 0.0075, "step": 41739 }, { "epoch": 2.2253025537132802, "grad_norm": 1.4445524215698242, "learning_rate": 1.6522000000000002e-06, "loss": 0.0185, "step": 41740 }, { "epoch": 2.2253558671429334, "grad_norm": 0.8044378757476807, "learning_rate": 1.6520000000000002e-06, "loss": 0.0039, "step": 41741 }, { "epoch": 2.225409180572586, "grad_norm": 1.4320653676986694, "learning_rate": 1.6518e-06, "loss": 0.035, "step": 41742 }, { "epoch": 2.2254624940022394, "grad_norm": 1.8638174533843994, "learning_rate": 1.6516000000000002e-06, "loss": 0.0095, "step": 41743 }, { "epoch": 2.225515807431892, "grad_norm": 1.5791255235671997, "learning_rate": 1.6514000000000002e-06, "loss": 0.0148, "step": 41744 }, { "epoch": 2.225569120861545, "grad_norm": 0.5201253294944763, "learning_rate": 1.6512e-06, "loss": 0.0045, "step": 41745 }, { "epoch": 2.225622434291198, "grad_norm": 0.9841662049293518, "learning_rate": 1.6510000000000002e-06, "loss": 0.0191, "step": 41746 }, { "epoch": 2.225675747720851, "grad_norm": 1.644737958908081, "learning_rate": 1.6508000000000001e-06, "loss": 0.0088, "step": 41747 }, { "epoch": 2.2257290611505036, "grad_norm": 3.7401905059814453, "learning_rate": 1.6506000000000003e-06, "loss": 0.0138, "step": 41748 }, { "epoch": 2.225782374580157, "grad_norm": 0.881545901298523, "learning_rate": 1.6504000000000001e-06, "loss": 0.019, "step": 41749 }, { "epoch": 2.2258356880098096, "grad_norm": 2.166534900665283, "learning_rate": 1.6502e-06, "loss": 0.0114, "step": 41750 }, { "epoch": 2.2258890014394628, "grad_norm": 0.8828220367431641, "learning_rate": 1.6500000000000003e-06, "loss": 0.0041, "step": 41751 }, { "epoch": 2.2259423148691155, "grad_norm": 1.1732920408248901, "learning_rate": 1.6498e-06, "loss": 0.0113, "step": 41752 }, { "epoch": 2.2259956282987683, "grad_norm": 0.5298563241958618, "learning_rate": 1.6496e-06, "loss": 0.0032, "step": 41753 }, { "epoch": 2.2260489417284215, "grad_norm": 0.7243072390556335, "learning_rate": 1.6494000000000003e-06, "loss": 0.0038, "step": 41754 }, { "epoch": 2.2261022551580743, "grad_norm": 1.2164239883422852, "learning_rate": 1.6492000000000002e-06, "loss": 0.0059, "step": 41755 }, { "epoch": 2.2261555685877275, "grad_norm": 0.8349757194519043, "learning_rate": 1.649e-06, "loss": 0.004, "step": 41756 }, { "epoch": 2.22620888201738, "grad_norm": 2.367597818374634, "learning_rate": 1.6488000000000002e-06, "loss": 0.0137, "step": 41757 }, { "epoch": 2.226262195447033, "grad_norm": 3.9487993717193604, "learning_rate": 1.6486000000000002e-06, "loss": 0.0465, "step": 41758 }, { "epoch": 2.226315508876686, "grad_norm": 0.49626627564430237, "learning_rate": 1.6484e-06, "loss": 0.0027, "step": 41759 }, { "epoch": 2.226368822306339, "grad_norm": 1.9888719320297241, "learning_rate": 1.6482000000000002e-06, "loss": 0.0392, "step": 41760 }, { "epoch": 2.2264221357359917, "grad_norm": 2.878484010696411, "learning_rate": 1.6480000000000001e-06, "loss": 0.0124, "step": 41761 }, { "epoch": 2.226475449165645, "grad_norm": 1.6912199258804321, "learning_rate": 1.6478000000000001e-06, "loss": 0.0194, "step": 41762 }, { "epoch": 2.2265287625952976, "grad_norm": 2.1097424030303955, "learning_rate": 1.6476000000000001e-06, "loss": 0.0071, "step": 41763 }, { "epoch": 2.226582076024951, "grad_norm": 3.209862232208252, "learning_rate": 1.6474e-06, "loss": 0.0496, "step": 41764 }, { "epoch": 2.2266353894546036, "grad_norm": 2.410963535308838, "learning_rate": 1.6472e-06, "loss": 0.0278, "step": 41765 }, { "epoch": 2.2266887028842564, "grad_norm": 0.7085644006729126, "learning_rate": 1.6470000000000003e-06, "loss": 0.0046, "step": 41766 }, { "epoch": 2.2267420163139096, "grad_norm": 1.430387020111084, "learning_rate": 1.6468e-06, "loss": 0.023, "step": 41767 }, { "epoch": 2.2267953297435623, "grad_norm": 0.20903775095939636, "learning_rate": 1.6466e-06, "loss": 0.0021, "step": 41768 }, { "epoch": 2.2268486431732155, "grad_norm": 3.0417706966400146, "learning_rate": 1.6464000000000002e-06, "loss": 0.0284, "step": 41769 }, { "epoch": 2.2269019566028683, "grad_norm": 1.2614843845367432, "learning_rate": 1.6462e-06, "loss": 0.0185, "step": 41770 }, { "epoch": 2.226955270032521, "grad_norm": 1.5671303272247314, "learning_rate": 1.646e-06, "loss": 0.0065, "step": 41771 }, { "epoch": 2.2270085834621742, "grad_norm": 0.9933038353919983, "learning_rate": 1.6458000000000002e-06, "loss": 0.007, "step": 41772 }, { "epoch": 2.227061896891827, "grad_norm": 0.8101380467414856, "learning_rate": 1.6456000000000002e-06, "loss": 0.0244, "step": 41773 }, { "epoch": 2.22711521032148, "grad_norm": 3.435828685760498, "learning_rate": 1.6454e-06, "loss": 0.021, "step": 41774 }, { "epoch": 2.227168523751133, "grad_norm": 0.6414815783500671, "learning_rate": 1.6452000000000001e-06, "loss": 0.0032, "step": 41775 }, { "epoch": 2.2272218371807857, "grad_norm": 2.385909080505371, "learning_rate": 1.6450000000000001e-06, "loss": 0.0291, "step": 41776 }, { "epoch": 2.227275150610439, "grad_norm": 1.788446068763733, "learning_rate": 1.6448e-06, "loss": 0.0259, "step": 41777 }, { "epoch": 2.2273284640400917, "grad_norm": 2.535491466522217, "learning_rate": 1.6446e-06, "loss": 0.0292, "step": 41778 }, { "epoch": 2.227381777469745, "grad_norm": 1.0872416496276855, "learning_rate": 1.6444e-06, "loss": 0.0184, "step": 41779 }, { "epoch": 2.2274350908993976, "grad_norm": 1.637590765953064, "learning_rate": 1.6442000000000003e-06, "loss": 0.0085, "step": 41780 }, { "epoch": 2.2274884043290504, "grad_norm": 2.574451446533203, "learning_rate": 1.644e-06, "loss": 0.0106, "step": 41781 }, { "epoch": 2.2275417177587036, "grad_norm": 1.3667711019515991, "learning_rate": 1.6438e-06, "loss": 0.0064, "step": 41782 }, { "epoch": 2.2275950311883563, "grad_norm": 0.2687733471393585, "learning_rate": 1.6436000000000002e-06, "loss": 0.0024, "step": 41783 }, { "epoch": 2.227648344618009, "grad_norm": 1.7930309772491455, "learning_rate": 1.6434000000000002e-06, "loss": 0.0173, "step": 41784 }, { "epoch": 2.2277016580476623, "grad_norm": 1.3697726726531982, "learning_rate": 1.6432e-06, "loss": 0.0091, "step": 41785 }, { "epoch": 2.227754971477315, "grad_norm": 1.2003103494644165, "learning_rate": 1.6430000000000002e-06, "loss": 0.0141, "step": 41786 }, { "epoch": 2.2278082849069683, "grad_norm": 0.35942336916923523, "learning_rate": 1.6428000000000002e-06, "loss": 0.0027, "step": 41787 }, { "epoch": 2.227861598336621, "grad_norm": 0.625652015209198, "learning_rate": 1.6426e-06, "loss": 0.0049, "step": 41788 }, { "epoch": 2.2279149117662738, "grad_norm": 1.010745644569397, "learning_rate": 1.6424000000000001e-06, "loss": 0.0107, "step": 41789 }, { "epoch": 2.227968225195927, "grad_norm": 0.9660789370536804, "learning_rate": 1.6422000000000001e-06, "loss": 0.0037, "step": 41790 }, { "epoch": 2.2280215386255797, "grad_norm": 2.650669813156128, "learning_rate": 1.6420000000000003e-06, "loss": 0.0158, "step": 41791 }, { "epoch": 2.228074852055233, "grad_norm": 1.5365686416625977, "learning_rate": 1.6418e-06, "loss": 0.0162, "step": 41792 }, { "epoch": 2.2281281654848857, "grad_norm": 2.0466551780700684, "learning_rate": 1.6416e-06, "loss": 0.0238, "step": 41793 }, { "epoch": 2.2281814789145384, "grad_norm": 1.7829474210739136, "learning_rate": 1.6414000000000003e-06, "loss": 0.0176, "step": 41794 }, { "epoch": 2.2282347923441916, "grad_norm": 1.5081534385681152, "learning_rate": 1.6412e-06, "loss": 0.0166, "step": 41795 }, { "epoch": 2.2282881057738444, "grad_norm": 2.6964051723480225, "learning_rate": 1.641e-06, "loss": 0.047, "step": 41796 }, { "epoch": 2.228341419203497, "grad_norm": 1.9782133102416992, "learning_rate": 1.6408000000000002e-06, "loss": 0.0275, "step": 41797 }, { "epoch": 2.2283947326331504, "grad_norm": 2.5920217037200928, "learning_rate": 1.6406000000000002e-06, "loss": 0.0101, "step": 41798 }, { "epoch": 2.228448046062803, "grad_norm": 2.676304817199707, "learning_rate": 1.6404e-06, "loss": 0.0267, "step": 41799 }, { "epoch": 2.2285013594924563, "grad_norm": 0.8508829474449158, "learning_rate": 1.6402000000000002e-06, "loss": 0.0051, "step": 41800 }, { "epoch": 2.228554672922109, "grad_norm": 3.8329904079437256, "learning_rate": 1.6400000000000002e-06, "loss": 0.0302, "step": 41801 }, { "epoch": 2.228607986351762, "grad_norm": 1.884806513786316, "learning_rate": 1.6398e-06, "loss": 0.0095, "step": 41802 }, { "epoch": 2.228661299781415, "grad_norm": 3.7630646228790283, "learning_rate": 1.6396000000000001e-06, "loss": 0.0241, "step": 41803 }, { "epoch": 2.228714613211068, "grad_norm": 0.6154676079750061, "learning_rate": 1.6394000000000001e-06, "loss": 0.0046, "step": 41804 }, { "epoch": 2.228767926640721, "grad_norm": 2.2187674045562744, "learning_rate": 1.6392000000000003e-06, "loss": 0.0171, "step": 41805 }, { "epoch": 2.2288212400703737, "grad_norm": 1.2232558727264404, "learning_rate": 1.639e-06, "loss": 0.0059, "step": 41806 }, { "epoch": 2.2288745535000265, "grad_norm": 2.8323912620544434, "learning_rate": 1.6388e-06, "loss": 0.02, "step": 41807 }, { "epoch": 2.2289278669296797, "grad_norm": 3.6949663162231445, "learning_rate": 1.6386000000000003e-06, "loss": 0.0086, "step": 41808 }, { "epoch": 2.2289811803593325, "grad_norm": 1.261889934539795, "learning_rate": 1.6384000000000003e-06, "loss": 0.015, "step": 41809 }, { "epoch": 2.229034493788985, "grad_norm": 0.18609559535980225, "learning_rate": 1.6382e-06, "loss": 0.0018, "step": 41810 }, { "epoch": 2.2290878072186384, "grad_norm": 1.3732631206512451, "learning_rate": 1.6380000000000002e-06, "loss": 0.0161, "step": 41811 }, { "epoch": 2.229141120648291, "grad_norm": 0.9500543475151062, "learning_rate": 1.6378000000000002e-06, "loss": 0.0161, "step": 41812 }, { "epoch": 2.2291944340779444, "grad_norm": 1.8688318729400635, "learning_rate": 1.6376e-06, "loss": 0.0279, "step": 41813 }, { "epoch": 2.229247747507597, "grad_norm": 1.1071043014526367, "learning_rate": 1.6374000000000002e-06, "loss": 0.0132, "step": 41814 }, { "epoch": 2.22930106093725, "grad_norm": 2.855118989944458, "learning_rate": 1.6372000000000002e-06, "loss": 0.0188, "step": 41815 }, { "epoch": 2.229354374366903, "grad_norm": 1.9637484550476074, "learning_rate": 1.6370000000000002e-06, "loss": 0.0062, "step": 41816 }, { "epoch": 2.229407687796556, "grad_norm": 0.7398316264152527, "learning_rate": 1.6368000000000002e-06, "loss": 0.0043, "step": 41817 }, { "epoch": 2.229461001226209, "grad_norm": 0.36730948090553284, "learning_rate": 1.6366000000000001e-06, "loss": 0.0022, "step": 41818 }, { "epoch": 2.229514314655862, "grad_norm": 1.052314043045044, "learning_rate": 1.6364000000000001e-06, "loss": 0.0238, "step": 41819 }, { "epoch": 2.2295676280855146, "grad_norm": 2.4857001304626465, "learning_rate": 1.6362000000000001e-06, "loss": 0.0358, "step": 41820 }, { "epoch": 2.2296209415151678, "grad_norm": 1.7138055562973022, "learning_rate": 1.636e-06, "loss": 0.0102, "step": 41821 }, { "epoch": 2.2296742549448205, "grad_norm": 1.9455052614212036, "learning_rate": 1.6358e-06, "loss": 0.0198, "step": 41822 }, { "epoch": 2.2297275683744737, "grad_norm": 0.36187902092933655, "learning_rate": 1.6356000000000003e-06, "loss": 0.0033, "step": 41823 }, { "epoch": 2.2297808818041265, "grad_norm": 0.4296847879886627, "learning_rate": 1.6354e-06, "loss": 0.0032, "step": 41824 }, { "epoch": 2.2298341952337792, "grad_norm": 1.3036763668060303, "learning_rate": 1.6352e-06, "loss": 0.0174, "step": 41825 }, { "epoch": 2.2298875086634324, "grad_norm": 3.401182174682617, "learning_rate": 1.6350000000000002e-06, "loss": 0.0177, "step": 41826 }, { "epoch": 2.229940822093085, "grad_norm": 0.7613146901130676, "learning_rate": 1.6348e-06, "loss": 0.004, "step": 41827 }, { "epoch": 2.2299941355227384, "grad_norm": 2.4762277603149414, "learning_rate": 1.6346e-06, "loss": 0.0459, "step": 41828 }, { "epoch": 2.230047448952391, "grad_norm": 1.6508731842041016, "learning_rate": 1.6344000000000002e-06, "loss": 0.0211, "step": 41829 }, { "epoch": 2.230100762382044, "grad_norm": 2.0445396900177, "learning_rate": 1.6342000000000002e-06, "loss": 0.0103, "step": 41830 }, { "epoch": 2.230154075811697, "grad_norm": 1.7163856029510498, "learning_rate": 1.634e-06, "loss": 0.004, "step": 41831 }, { "epoch": 2.23020738924135, "grad_norm": 3.3554463386535645, "learning_rate": 1.6338000000000001e-06, "loss": 0.0245, "step": 41832 }, { "epoch": 2.2302607026710026, "grad_norm": 2.081531286239624, "learning_rate": 1.6336000000000001e-06, "loss": 0.009, "step": 41833 }, { "epoch": 2.230314016100656, "grad_norm": 0.983267605304718, "learning_rate": 1.6334000000000003e-06, "loss": 0.022, "step": 41834 }, { "epoch": 2.2303673295303086, "grad_norm": 3.4981212615966797, "learning_rate": 1.6332e-06, "loss": 0.058, "step": 41835 }, { "epoch": 2.230420642959962, "grad_norm": 2.8974664211273193, "learning_rate": 1.633e-06, "loss": 0.0258, "step": 41836 }, { "epoch": 2.2304739563896145, "grad_norm": 1.3558610677719116, "learning_rate": 1.6328000000000003e-06, "loss": 0.0059, "step": 41837 }, { "epoch": 2.2305272698192673, "grad_norm": 1.878665566444397, "learning_rate": 1.6326e-06, "loss": 0.0045, "step": 41838 }, { "epoch": 2.2305805832489205, "grad_norm": 3.4729955196380615, "learning_rate": 1.6324e-06, "loss": 0.0189, "step": 41839 }, { "epoch": 2.2306338966785733, "grad_norm": 3.170485734939575, "learning_rate": 1.6322000000000002e-06, "loss": 0.0317, "step": 41840 }, { "epoch": 2.2306872101082265, "grad_norm": 3.040262222290039, "learning_rate": 1.6320000000000002e-06, "loss": 0.0095, "step": 41841 }, { "epoch": 2.230740523537879, "grad_norm": 2.058870792388916, "learning_rate": 1.6318e-06, "loss": 0.012, "step": 41842 }, { "epoch": 2.230793836967532, "grad_norm": 2.8180243968963623, "learning_rate": 1.6316000000000002e-06, "loss": 0.0243, "step": 41843 }, { "epoch": 2.230847150397185, "grad_norm": 2.537078857421875, "learning_rate": 1.6314000000000002e-06, "loss": 0.0134, "step": 41844 }, { "epoch": 2.230900463826838, "grad_norm": 2.0981814861297607, "learning_rate": 1.6312e-06, "loss": 0.0169, "step": 41845 }, { "epoch": 2.2309537772564907, "grad_norm": 2.459535837173462, "learning_rate": 1.6310000000000001e-06, "loss": 0.0074, "step": 41846 }, { "epoch": 2.231007090686144, "grad_norm": 0.915341317653656, "learning_rate": 1.6308000000000001e-06, "loss": 0.0127, "step": 41847 }, { "epoch": 2.2310604041157966, "grad_norm": 0.739393949508667, "learning_rate": 1.6306000000000003e-06, "loss": 0.0043, "step": 41848 }, { "epoch": 2.23111371754545, "grad_norm": 2.3537774085998535, "learning_rate": 1.6304e-06, "loss": 0.0258, "step": 41849 }, { "epoch": 2.2311670309751026, "grad_norm": 1.0956021547317505, "learning_rate": 1.6302e-06, "loss": 0.0137, "step": 41850 }, { "epoch": 2.2312203444047554, "grad_norm": 3.0508229732513428, "learning_rate": 1.6300000000000003e-06, "loss": 0.012, "step": 41851 }, { "epoch": 2.2312736578344086, "grad_norm": 1.3771281242370605, "learning_rate": 1.6298000000000003e-06, "loss": 0.0147, "step": 41852 }, { "epoch": 2.2313269712640613, "grad_norm": 2.623432159423828, "learning_rate": 1.6296e-06, "loss": 0.0419, "step": 41853 }, { "epoch": 2.2313802846937145, "grad_norm": 1.101251244544983, "learning_rate": 1.6294000000000002e-06, "loss": 0.0173, "step": 41854 }, { "epoch": 2.2314335981233673, "grad_norm": 2.1922316551208496, "learning_rate": 1.6292000000000002e-06, "loss": 0.028, "step": 41855 }, { "epoch": 2.23148691155302, "grad_norm": 1.1223597526550293, "learning_rate": 1.629e-06, "loss": 0.0029, "step": 41856 }, { "epoch": 2.2315402249826732, "grad_norm": 1.7850782871246338, "learning_rate": 1.6288000000000002e-06, "loss": 0.0266, "step": 41857 }, { "epoch": 2.231593538412326, "grad_norm": 1.2384033203125, "learning_rate": 1.6286000000000002e-06, "loss": 0.0069, "step": 41858 }, { "epoch": 2.231646851841979, "grad_norm": 2.3508782386779785, "learning_rate": 1.6284000000000002e-06, "loss": 0.0061, "step": 41859 }, { "epoch": 2.231700165271632, "grad_norm": 3.0330448150634766, "learning_rate": 1.6282000000000001e-06, "loss": 0.0155, "step": 41860 }, { "epoch": 2.2317534787012847, "grad_norm": 1.6429352760314941, "learning_rate": 1.6280000000000001e-06, "loss": 0.0171, "step": 41861 }, { "epoch": 2.231806792130938, "grad_norm": 0.6056172847747803, "learning_rate": 1.6278000000000001e-06, "loss": 0.0043, "step": 41862 }, { "epoch": 2.2318601055605907, "grad_norm": 0.45864349603652954, "learning_rate": 1.6276e-06, "loss": 0.0029, "step": 41863 }, { "epoch": 2.231913418990244, "grad_norm": 2.3959362506866455, "learning_rate": 1.6274e-06, "loss": 0.0148, "step": 41864 }, { "epoch": 2.2319667324198966, "grad_norm": 0.5036793351173401, "learning_rate": 1.6272e-06, "loss": 0.0028, "step": 41865 }, { "epoch": 2.2320200458495494, "grad_norm": 0.5756040215492249, "learning_rate": 1.6270000000000003e-06, "loss": 0.0037, "step": 41866 }, { "epoch": 2.2320733592792026, "grad_norm": 2.397770881652832, "learning_rate": 1.6268e-06, "loss": 0.0407, "step": 41867 }, { "epoch": 2.2321266727088553, "grad_norm": 1.6740663051605225, "learning_rate": 1.6266e-06, "loss": 0.0114, "step": 41868 }, { "epoch": 2.232179986138508, "grad_norm": 1.8289047479629517, "learning_rate": 1.6264000000000002e-06, "loss": 0.0129, "step": 41869 }, { "epoch": 2.2322332995681613, "grad_norm": 2.325288772583008, "learning_rate": 1.6262e-06, "loss": 0.0336, "step": 41870 }, { "epoch": 2.232286612997814, "grad_norm": 3.2820522785186768, "learning_rate": 1.626e-06, "loss": 0.038, "step": 41871 }, { "epoch": 2.2323399264274673, "grad_norm": 0.6580925583839417, "learning_rate": 1.6258000000000002e-06, "loss": 0.0038, "step": 41872 }, { "epoch": 2.23239323985712, "grad_norm": 2.6547231674194336, "learning_rate": 1.6256000000000002e-06, "loss": 0.0216, "step": 41873 }, { "epoch": 2.2324465532867728, "grad_norm": 1.537940502166748, "learning_rate": 1.6254e-06, "loss": 0.0082, "step": 41874 }, { "epoch": 2.232499866716426, "grad_norm": 1.3537442684173584, "learning_rate": 1.6252000000000001e-06, "loss": 0.0056, "step": 41875 }, { "epoch": 2.2325531801460787, "grad_norm": 0.8273703455924988, "learning_rate": 1.6250000000000001e-06, "loss": 0.0041, "step": 41876 }, { "epoch": 2.232606493575732, "grad_norm": 1.04044771194458, "learning_rate": 1.6248000000000003e-06, "loss": 0.007, "step": 41877 }, { "epoch": 2.2326598070053847, "grad_norm": 1.0025464296340942, "learning_rate": 1.6246e-06, "loss": 0.0042, "step": 41878 }, { "epoch": 2.2327131204350374, "grad_norm": 2.146709680557251, "learning_rate": 1.6244e-06, "loss": 0.0218, "step": 41879 }, { "epoch": 2.2327664338646906, "grad_norm": 1.3187379837036133, "learning_rate": 1.6242000000000003e-06, "loss": 0.0073, "step": 41880 }, { "epoch": 2.2328197472943434, "grad_norm": 1.8641846179962158, "learning_rate": 1.624e-06, "loss": 0.0126, "step": 41881 }, { "epoch": 2.232873060723996, "grad_norm": 2.094977855682373, "learning_rate": 1.6238e-06, "loss": 0.0218, "step": 41882 }, { "epoch": 2.2329263741536494, "grad_norm": 1.5603007078170776, "learning_rate": 1.6236000000000002e-06, "loss": 0.0097, "step": 41883 }, { "epoch": 2.232979687583302, "grad_norm": 2.157228946685791, "learning_rate": 1.6234000000000002e-06, "loss": 0.0115, "step": 41884 }, { "epoch": 2.2330330010129553, "grad_norm": 2.5831735134124756, "learning_rate": 1.6232e-06, "loss": 0.0125, "step": 41885 }, { "epoch": 2.233086314442608, "grad_norm": 2.1055185794830322, "learning_rate": 1.6230000000000002e-06, "loss": 0.0168, "step": 41886 }, { "epoch": 2.233139627872261, "grad_norm": 1.6968744993209839, "learning_rate": 1.6228000000000002e-06, "loss": 0.0231, "step": 41887 }, { "epoch": 2.233192941301914, "grad_norm": 1.4830565452575684, "learning_rate": 1.6226e-06, "loss": 0.0086, "step": 41888 }, { "epoch": 2.233246254731567, "grad_norm": 4.53548526763916, "learning_rate": 1.6224000000000001e-06, "loss": 0.0179, "step": 41889 }, { "epoch": 2.23329956816122, "grad_norm": 2.681248903274536, "learning_rate": 1.6222000000000001e-06, "loss": 0.0224, "step": 41890 }, { "epoch": 2.2333528815908728, "grad_norm": 1.6971911191940308, "learning_rate": 1.6220000000000003e-06, "loss": 0.0097, "step": 41891 }, { "epoch": 2.2334061950205255, "grad_norm": 0.770961344242096, "learning_rate": 1.6218e-06, "loss": 0.0148, "step": 41892 }, { "epoch": 2.2334595084501787, "grad_norm": 2.649909734725952, "learning_rate": 1.6216e-06, "loss": 0.022, "step": 41893 }, { "epoch": 2.2335128218798315, "grad_norm": 5.081359386444092, "learning_rate": 1.6214000000000003e-06, "loss": 0.0363, "step": 41894 }, { "epoch": 2.2335661353094842, "grad_norm": 0.2932298183441162, "learning_rate": 1.6212e-06, "loss": 0.002, "step": 41895 }, { "epoch": 2.2336194487391374, "grad_norm": 1.8380250930786133, "learning_rate": 1.621e-06, "loss": 0.0208, "step": 41896 }, { "epoch": 2.23367276216879, "grad_norm": 1.0198898315429688, "learning_rate": 1.6208000000000002e-06, "loss": 0.0133, "step": 41897 }, { "epoch": 2.2337260755984434, "grad_norm": 3.574134349822998, "learning_rate": 1.6206000000000002e-06, "loss": 0.034, "step": 41898 }, { "epoch": 2.233779389028096, "grad_norm": 1.8740795850753784, "learning_rate": 1.6204e-06, "loss": 0.0092, "step": 41899 }, { "epoch": 2.2338327024577493, "grad_norm": 2.298919439315796, "learning_rate": 1.6202000000000002e-06, "loss": 0.0112, "step": 41900 }, { "epoch": 2.233886015887402, "grad_norm": 2.6858808994293213, "learning_rate": 1.6200000000000002e-06, "loss": 0.0324, "step": 41901 }, { "epoch": 2.233939329317055, "grad_norm": 1.7420356273651123, "learning_rate": 1.6198000000000004e-06, "loss": 0.0079, "step": 41902 }, { "epoch": 2.233992642746708, "grad_norm": 3.449486494064331, "learning_rate": 1.6196000000000001e-06, "loss": 0.0295, "step": 41903 }, { "epoch": 2.234045956176361, "grad_norm": 6.70751428604126, "learning_rate": 1.6194000000000001e-06, "loss": 0.0274, "step": 41904 }, { "epoch": 2.2340992696060136, "grad_norm": 2.0037810802459717, "learning_rate": 1.6192000000000003e-06, "loss": 0.008, "step": 41905 }, { "epoch": 2.2341525830356668, "grad_norm": 4.134212493896484, "learning_rate": 1.619e-06, "loss": 0.0179, "step": 41906 }, { "epoch": 2.2342058964653195, "grad_norm": 3.3000857830047607, "learning_rate": 1.6188e-06, "loss": 0.0118, "step": 41907 }, { "epoch": 2.2342592098949727, "grad_norm": 1.8949692249298096, "learning_rate": 1.6186000000000003e-06, "loss": 0.0246, "step": 41908 }, { "epoch": 2.2343125233246255, "grad_norm": 1.6779881715774536, "learning_rate": 1.6184000000000003e-06, "loss": 0.0099, "step": 41909 }, { "epoch": 2.2343658367542782, "grad_norm": 0.4132797122001648, "learning_rate": 1.6182e-06, "loss": 0.0028, "step": 41910 }, { "epoch": 2.2344191501839314, "grad_norm": 1.6191679239273071, "learning_rate": 1.6180000000000002e-06, "loss": 0.0177, "step": 41911 }, { "epoch": 2.234472463613584, "grad_norm": 0.5169010758399963, "learning_rate": 1.6178000000000002e-06, "loss": 0.0034, "step": 41912 }, { "epoch": 2.2345257770432374, "grad_norm": 4.568844318389893, "learning_rate": 1.6176e-06, "loss": 0.0284, "step": 41913 }, { "epoch": 2.23457909047289, "grad_norm": 2.2834317684173584, "learning_rate": 1.6174000000000002e-06, "loss": 0.0074, "step": 41914 }, { "epoch": 2.234632403902543, "grad_norm": 0.8349266052246094, "learning_rate": 1.6172000000000002e-06, "loss": 0.0151, "step": 41915 }, { "epoch": 2.234685717332196, "grad_norm": 0.8971216082572937, "learning_rate": 1.6170000000000001e-06, "loss": 0.0043, "step": 41916 }, { "epoch": 2.234739030761849, "grad_norm": 0.6381049752235413, "learning_rate": 1.6168000000000001e-06, "loss": 0.0038, "step": 41917 }, { "epoch": 2.2347923441915016, "grad_norm": 0.6365686655044556, "learning_rate": 1.6166000000000001e-06, "loss": 0.0035, "step": 41918 }, { "epoch": 2.234845657621155, "grad_norm": 2.0600268840789795, "learning_rate": 1.6164e-06, "loss": 0.0058, "step": 41919 }, { "epoch": 2.2348989710508076, "grad_norm": 1.3442537784576416, "learning_rate": 1.6162000000000003e-06, "loss": 0.0203, "step": 41920 }, { "epoch": 2.234952284480461, "grad_norm": 1.7854474782943726, "learning_rate": 1.616e-06, "loss": 0.0225, "step": 41921 }, { "epoch": 2.2350055979101136, "grad_norm": 0.8301660418510437, "learning_rate": 1.6158e-06, "loss": 0.0034, "step": 41922 }, { "epoch": 2.2350589113397663, "grad_norm": 3.9284279346466064, "learning_rate": 1.6156000000000003e-06, "loss": 0.0483, "step": 41923 }, { "epoch": 2.2351122247694195, "grad_norm": 2.141810894012451, "learning_rate": 1.6154e-06, "loss": 0.0085, "step": 41924 }, { "epoch": 2.2351655381990723, "grad_norm": 1.4858890771865845, "learning_rate": 1.6152e-06, "loss": 0.037, "step": 41925 }, { "epoch": 2.2352188516287255, "grad_norm": 4.278793811798096, "learning_rate": 1.6150000000000002e-06, "loss": 0.0313, "step": 41926 }, { "epoch": 2.2352721650583782, "grad_norm": 1.9517083168029785, "learning_rate": 1.6148000000000002e-06, "loss": 0.0064, "step": 41927 }, { "epoch": 2.235325478488031, "grad_norm": 1.0546737909317017, "learning_rate": 1.6146e-06, "loss": 0.0202, "step": 41928 }, { "epoch": 2.235378791917684, "grad_norm": 3.3233959674835205, "learning_rate": 1.6144000000000002e-06, "loss": 0.0197, "step": 41929 }, { "epoch": 2.235432105347337, "grad_norm": 1.688593864440918, "learning_rate": 1.6142000000000001e-06, "loss": 0.0099, "step": 41930 }, { "epoch": 2.2354854187769897, "grad_norm": 1.7942309379577637, "learning_rate": 1.614e-06, "loss": 0.0055, "step": 41931 }, { "epoch": 2.235538732206643, "grad_norm": 0.7337415218353271, "learning_rate": 1.6138000000000001e-06, "loss": 0.0049, "step": 41932 }, { "epoch": 2.2355920456362957, "grad_norm": 3.604151964187622, "learning_rate": 1.6136e-06, "loss": 0.0466, "step": 41933 }, { "epoch": 2.235645359065949, "grad_norm": 1.9428625106811523, "learning_rate": 1.6134000000000003e-06, "loss": 0.0204, "step": 41934 }, { "epoch": 2.2356986724956016, "grad_norm": 2.49853515625, "learning_rate": 1.6132e-06, "loss": 0.0266, "step": 41935 }, { "epoch": 2.2357519859252544, "grad_norm": 0.5825157165527344, "learning_rate": 1.613e-06, "loss": 0.0029, "step": 41936 }, { "epoch": 2.2358052993549076, "grad_norm": 1.9570715427398682, "learning_rate": 1.6128000000000003e-06, "loss": 0.0319, "step": 41937 }, { "epoch": 2.2358586127845603, "grad_norm": 2.5688672065734863, "learning_rate": 1.6126e-06, "loss": 0.0484, "step": 41938 }, { "epoch": 2.2359119262142135, "grad_norm": 2.6949212551116943, "learning_rate": 1.6124e-06, "loss": 0.023, "step": 41939 }, { "epoch": 2.2359652396438663, "grad_norm": 2.7951982021331787, "learning_rate": 1.6122000000000002e-06, "loss": 0.0183, "step": 41940 }, { "epoch": 2.236018553073519, "grad_norm": 0.37438222765922546, "learning_rate": 1.6120000000000002e-06, "loss": 0.0033, "step": 41941 }, { "epoch": 2.2360718665031722, "grad_norm": 1.5748413801193237, "learning_rate": 1.6118e-06, "loss": 0.0092, "step": 41942 }, { "epoch": 2.236125179932825, "grad_norm": 1.6123251914978027, "learning_rate": 1.6116000000000002e-06, "loss": 0.0114, "step": 41943 }, { "epoch": 2.236178493362478, "grad_norm": 1.2795275449752808, "learning_rate": 1.6114000000000001e-06, "loss": 0.0116, "step": 41944 }, { "epoch": 2.236231806792131, "grad_norm": 0.7174407839775085, "learning_rate": 1.6112000000000003e-06, "loss": 0.0028, "step": 41945 }, { "epoch": 2.2362851202217837, "grad_norm": 2.2152578830718994, "learning_rate": 1.6110000000000001e-06, "loss": 0.0125, "step": 41946 }, { "epoch": 2.236338433651437, "grad_norm": 3.316762685775757, "learning_rate": 1.6108e-06, "loss": 0.0324, "step": 41947 }, { "epoch": 2.2363917470810897, "grad_norm": 1.6320337057113647, "learning_rate": 1.6106000000000003e-06, "loss": 0.01, "step": 41948 }, { "epoch": 2.236445060510743, "grad_norm": 1.8317056894302368, "learning_rate": 1.6104e-06, "loss": 0.0099, "step": 41949 }, { "epoch": 2.2364983739403956, "grad_norm": 3.31467866897583, "learning_rate": 1.6102e-06, "loss": 0.058, "step": 41950 }, { "epoch": 2.2365516873700484, "grad_norm": 1.5184524059295654, "learning_rate": 1.6100000000000003e-06, "loss": 0.0115, "step": 41951 }, { "epoch": 2.2366050007997016, "grad_norm": 2.9966485500335693, "learning_rate": 1.6098000000000002e-06, "loss": 0.0366, "step": 41952 }, { "epoch": 2.2366583142293543, "grad_norm": 3.003617286682129, "learning_rate": 1.6096e-06, "loss": 0.0308, "step": 41953 }, { "epoch": 2.236711627659007, "grad_norm": 0.4712131917476654, "learning_rate": 1.6094000000000002e-06, "loss": 0.0036, "step": 41954 }, { "epoch": 2.2367649410886603, "grad_norm": 1.5250507593154907, "learning_rate": 1.6092000000000002e-06, "loss": 0.0209, "step": 41955 }, { "epoch": 2.236818254518313, "grad_norm": 2.4802403450012207, "learning_rate": 1.609e-06, "loss": 0.011, "step": 41956 }, { "epoch": 2.2368715679479663, "grad_norm": 0.8918319940567017, "learning_rate": 1.6088000000000002e-06, "loss": 0.0044, "step": 41957 }, { "epoch": 2.236924881377619, "grad_norm": 0.986539363861084, "learning_rate": 1.6086000000000002e-06, "loss": 0.0035, "step": 41958 }, { "epoch": 2.236978194807272, "grad_norm": 1.8215296268463135, "learning_rate": 1.6084000000000001e-06, "loss": 0.0138, "step": 41959 }, { "epoch": 2.237031508236925, "grad_norm": 1.322819709777832, "learning_rate": 1.6082000000000001e-06, "loss": 0.0121, "step": 41960 }, { "epoch": 2.2370848216665777, "grad_norm": 1.7610697746276855, "learning_rate": 1.608e-06, "loss": 0.0152, "step": 41961 }, { "epoch": 2.237138135096231, "grad_norm": 3.3703627586364746, "learning_rate": 1.6078e-06, "loss": 0.0352, "step": 41962 }, { "epoch": 2.2371914485258837, "grad_norm": 1.853493571281433, "learning_rate": 1.6076e-06, "loss": 0.0341, "step": 41963 }, { "epoch": 2.2372447619555365, "grad_norm": 0.8637627959251404, "learning_rate": 1.6074e-06, "loss": 0.006, "step": 41964 }, { "epoch": 2.2372980753851897, "grad_norm": 1.616815447807312, "learning_rate": 1.6072e-06, "loss": 0.0084, "step": 41965 }, { "epoch": 2.2373513888148424, "grad_norm": 0.864193856716156, "learning_rate": 1.6070000000000002e-06, "loss": 0.0153, "step": 41966 }, { "epoch": 2.237404702244495, "grad_norm": 1.8530036211013794, "learning_rate": 1.6068e-06, "loss": 0.029, "step": 41967 }, { "epoch": 2.2374580156741484, "grad_norm": 1.6288368701934814, "learning_rate": 1.6066e-06, "loss": 0.0122, "step": 41968 }, { "epoch": 2.237511329103801, "grad_norm": 1.2213894128799438, "learning_rate": 1.6064000000000002e-06, "loss": 0.005, "step": 41969 }, { "epoch": 2.2375646425334543, "grad_norm": 1.148602843284607, "learning_rate": 1.6062000000000002e-06, "loss": 0.0073, "step": 41970 }, { "epoch": 2.237617955963107, "grad_norm": 2.0918262004852295, "learning_rate": 1.606e-06, "loss": 0.0136, "step": 41971 }, { "epoch": 2.23767126939276, "grad_norm": 3.941986083984375, "learning_rate": 1.6058000000000002e-06, "loss": 0.0637, "step": 41972 }, { "epoch": 2.237724582822413, "grad_norm": 0.18219409883022308, "learning_rate": 1.6056000000000001e-06, "loss": 0.0012, "step": 41973 }, { "epoch": 2.237777896252066, "grad_norm": 0.7309382557868958, "learning_rate": 1.6054e-06, "loss": 0.0048, "step": 41974 }, { "epoch": 2.237831209681719, "grad_norm": 1.1113098859786987, "learning_rate": 1.6052000000000001e-06, "loss": 0.0173, "step": 41975 }, { "epoch": 2.2378845231113718, "grad_norm": 1.991645097732544, "learning_rate": 1.605e-06, "loss": 0.0099, "step": 41976 }, { "epoch": 2.2379378365410245, "grad_norm": 0.6665138602256775, "learning_rate": 1.6048000000000003e-06, "loss": 0.0043, "step": 41977 }, { "epoch": 2.2379911499706777, "grad_norm": 1.6563607454299927, "learning_rate": 1.6046e-06, "loss": 0.0115, "step": 41978 }, { "epoch": 2.2380444634003305, "grad_norm": 2.692629814147949, "learning_rate": 1.6044e-06, "loss": 0.0267, "step": 41979 }, { "epoch": 2.2380977768299837, "grad_norm": 1.4112905263900757, "learning_rate": 1.6042000000000002e-06, "loss": 0.0139, "step": 41980 }, { "epoch": 2.2381510902596364, "grad_norm": 2.266892194747925, "learning_rate": 1.604e-06, "loss": 0.0075, "step": 41981 }, { "epoch": 2.238204403689289, "grad_norm": 2.225560188293457, "learning_rate": 1.6038e-06, "loss": 0.0098, "step": 41982 }, { "epoch": 2.2382577171189424, "grad_norm": 3.5805773735046387, "learning_rate": 1.6036000000000002e-06, "loss": 0.0359, "step": 41983 }, { "epoch": 2.238311030548595, "grad_norm": 0.7500020265579224, "learning_rate": 1.6034000000000002e-06, "loss": 0.004, "step": 41984 }, { "epoch": 2.2383643439782483, "grad_norm": 4.23984956741333, "learning_rate": 1.6032e-06, "loss": 0.0106, "step": 41985 }, { "epoch": 2.238417657407901, "grad_norm": 2.552532911300659, "learning_rate": 1.6030000000000002e-06, "loss": 0.0408, "step": 41986 }, { "epoch": 2.238470970837554, "grad_norm": 1.9181435108184814, "learning_rate": 1.6028000000000001e-06, "loss": 0.0101, "step": 41987 }, { "epoch": 2.238524284267207, "grad_norm": 2.114469289779663, "learning_rate": 1.6026000000000003e-06, "loss": 0.0285, "step": 41988 }, { "epoch": 2.23857759769686, "grad_norm": 0.8511232733726501, "learning_rate": 1.6024000000000001e-06, "loss": 0.0046, "step": 41989 }, { "epoch": 2.2386309111265126, "grad_norm": 1.7575844526290894, "learning_rate": 1.6022e-06, "loss": 0.0108, "step": 41990 }, { "epoch": 2.238684224556166, "grad_norm": 0.6367936730384827, "learning_rate": 1.6020000000000003e-06, "loss": 0.0045, "step": 41991 }, { "epoch": 2.2387375379858185, "grad_norm": 2.5929930210113525, "learning_rate": 1.6018e-06, "loss": 0.0105, "step": 41992 }, { "epoch": 2.2387908514154717, "grad_norm": 1.467470407485962, "learning_rate": 1.6016e-06, "loss": 0.0062, "step": 41993 }, { "epoch": 2.2388441648451245, "grad_norm": 0.56947261095047, "learning_rate": 1.6014000000000002e-06, "loss": 0.0033, "step": 41994 }, { "epoch": 2.2388974782747773, "grad_norm": 4.201329231262207, "learning_rate": 1.6012000000000002e-06, "loss": 0.0186, "step": 41995 }, { "epoch": 2.2389507917044305, "grad_norm": 1.2118600606918335, "learning_rate": 1.601e-06, "loss": 0.0213, "step": 41996 }, { "epoch": 2.239004105134083, "grad_norm": 0.7079046964645386, "learning_rate": 1.6008000000000002e-06, "loss": 0.0043, "step": 41997 }, { "epoch": 2.2390574185637364, "grad_norm": 4.1598801612854, "learning_rate": 1.6006000000000002e-06, "loss": 0.0241, "step": 41998 }, { "epoch": 2.239110731993389, "grad_norm": 0.3468797206878662, "learning_rate": 1.6004e-06, "loss": 0.0025, "step": 41999 }, { "epoch": 2.239164045423042, "grad_norm": 0.2667420506477356, "learning_rate": 1.6002000000000002e-06, "loss": 0.0022, "step": 42000 }, { "epoch": 2.239217358852695, "grad_norm": 0.5648603439331055, "learning_rate": 1.6000000000000001e-06, "loss": 0.0034, "step": 42001 }, { "epoch": 2.239270672282348, "grad_norm": 1.9605093002319336, "learning_rate": 1.5998000000000003e-06, "loss": 0.0164, "step": 42002 }, { "epoch": 2.2393239857120006, "grad_norm": 2.1882243156433105, "learning_rate": 1.5996000000000001e-06, "loss": 0.0234, "step": 42003 }, { "epoch": 2.239377299141654, "grad_norm": 1.4657318592071533, "learning_rate": 1.5994e-06, "loss": 0.0072, "step": 42004 }, { "epoch": 2.2394306125713066, "grad_norm": 2.570521116256714, "learning_rate": 1.5992000000000003e-06, "loss": 0.0517, "step": 42005 }, { "epoch": 2.23948392600096, "grad_norm": 1.4099053144454956, "learning_rate": 1.599e-06, "loss": 0.0063, "step": 42006 }, { "epoch": 2.2395372394306126, "grad_norm": 1.5016889572143555, "learning_rate": 1.5988e-06, "loss": 0.0054, "step": 42007 }, { "epoch": 2.2395905528602653, "grad_norm": 3.9351119995117188, "learning_rate": 1.5986000000000002e-06, "loss": 0.0369, "step": 42008 }, { "epoch": 2.2396438662899185, "grad_norm": 1.4293198585510254, "learning_rate": 1.5984000000000002e-06, "loss": 0.0224, "step": 42009 }, { "epoch": 2.2396971797195713, "grad_norm": 0.8303660750389099, "learning_rate": 1.5982e-06, "loss": 0.0038, "step": 42010 }, { "epoch": 2.2397504931492245, "grad_norm": 1.797218680381775, "learning_rate": 1.5980000000000002e-06, "loss": 0.0179, "step": 42011 }, { "epoch": 2.2398038065788772, "grad_norm": 1.2502970695495605, "learning_rate": 1.5978000000000002e-06, "loss": 0.0269, "step": 42012 }, { "epoch": 2.23985712000853, "grad_norm": 0.09295811504125595, "learning_rate": 1.5976000000000002e-06, "loss": 0.0009, "step": 42013 }, { "epoch": 2.239910433438183, "grad_norm": 1.4228118658065796, "learning_rate": 1.5974000000000002e-06, "loss": 0.0172, "step": 42014 }, { "epoch": 2.239963746867836, "grad_norm": 0.9566429257392883, "learning_rate": 1.5972000000000001e-06, "loss": 0.0195, "step": 42015 }, { "epoch": 2.2400170602974887, "grad_norm": 1.7568004131317139, "learning_rate": 1.5970000000000001e-06, "loss": 0.0108, "step": 42016 }, { "epoch": 2.240070373727142, "grad_norm": 2.5802977085113525, "learning_rate": 1.5968000000000001e-06, "loss": 0.0285, "step": 42017 }, { "epoch": 2.2401236871567947, "grad_norm": 1.3933037519454956, "learning_rate": 1.5966e-06, "loss": 0.0091, "step": 42018 }, { "epoch": 2.240177000586448, "grad_norm": 1.043737769126892, "learning_rate": 1.5964e-06, "loss": 0.0036, "step": 42019 }, { "epoch": 2.2402303140161006, "grad_norm": 1.2958236932754517, "learning_rate": 1.5962000000000003e-06, "loss": 0.0154, "step": 42020 }, { "epoch": 2.2402836274457534, "grad_norm": 3.909088611602783, "learning_rate": 1.596e-06, "loss": 0.0415, "step": 42021 }, { "epoch": 2.2403369408754066, "grad_norm": 1.8352982997894287, "learning_rate": 1.5958e-06, "loss": 0.0132, "step": 42022 }, { "epoch": 2.2403902543050593, "grad_norm": 2.5709338188171387, "learning_rate": 1.5956000000000002e-06, "loss": 0.0158, "step": 42023 }, { "epoch": 2.2404435677347125, "grad_norm": 1.478769302368164, "learning_rate": 1.5954e-06, "loss": 0.0171, "step": 42024 }, { "epoch": 2.2404968811643653, "grad_norm": 1.915891170501709, "learning_rate": 1.5952e-06, "loss": 0.0083, "step": 42025 }, { "epoch": 2.240550194594018, "grad_norm": 3.287003755569458, "learning_rate": 1.5950000000000002e-06, "loss": 0.0229, "step": 42026 }, { "epoch": 2.2406035080236713, "grad_norm": 1.7387561798095703, "learning_rate": 1.5948000000000002e-06, "loss": 0.011, "step": 42027 }, { "epoch": 2.240656821453324, "grad_norm": 1.779929280281067, "learning_rate": 1.5946e-06, "loss": 0.045, "step": 42028 }, { "epoch": 2.240710134882977, "grad_norm": 1.5564666986465454, "learning_rate": 1.5944000000000001e-06, "loss": 0.0083, "step": 42029 }, { "epoch": 2.24076344831263, "grad_norm": 2.146954298019409, "learning_rate": 1.5942000000000001e-06, "loss": 0.0198, "step": 42030 }, { "epoch": 2.2408167617422827, "grad_norm": 1.6114760637283325, "learning_rate": 1.594e-06, "loss": 0.0182, "step": 42031 }, { "epoch": 2.240870075171936, "grad_norm": 0.41457274556159973, "learning_rate": 1.5938e-06, "loss": 0.0034, "step": 42032 }, { "epoch": 2.2409233886015887, "grad_norm": 2.737699031829834, "learning_rate": 1.5936e-06, "loss": 0.0299, "step": 42033 }, { "epoch": 2.240976702031242, "grad_norm": 2.124586343765259, "learning_rate": 1.5934000000000003e-06, "loss": 0.0362, "step": 42034 }, { "epoch": 2.2410300154608946, "grad_norm": 4.089888572692871, "learning_rate": 1.5932e-06, "loss": 0.0146, "step": 42035 }, { "epoch": 2.2410833288905474, "grad_norm": 3.720832586288452, "learning_rate": 1.593e-06, "loss": 0.0183, "step": 42036 }, { "epoch": 2.2411366423202006, "grad_norm": 1.3056011199951172, "learning_rate": 1.5928000000000002e-06, "loss": 0.0072, "step": 42037 }, { "epoch": 2.2411899557498534, "grad_norm": 1.0967979431152344, "learning_rate": 1.5926000000000002e-06, "loss": 0.0074, "step": 42038 }, { "epoch": 2.241243269179506, "grad_norm": 2.2224230766296387, "learning_rate": 1.5924e-06, "loss": 0.0072, "step": 42039 }, { "epoch": 2.2412965826091593, "grad_norm": 2.031148910522461, "learning_rate": 1.5922000000000002e-06, "loss": 0.0095, "step": 42040 }, { "epoch": 2.241349896038812, "grad_norm": 1.754636526107788, "learning_rate": 1.5920000000000002e-06, "loss": 0.0149, "step": 42041 }, { "epoch": 2.2414032094684653, "grad_norm": 2.059816360473633, "learning_rate": 1.5918e-06, "loss": 0.0097, "step": 42042 }, { "epoch": 2.241456522898118, "grad_norm": 1.693772315979004, "learning_rate": 1.5916000000000001e-06, "loss": 0.0188, "step": 42043 }, { "epoch": 2.241509836327771, "grad_norm": 2.0812060832977295, "learning_rate": 1.5914000000000001e-06, "loss": 0.0237, "step": 42044 }, { "epoch": 2.241563149757424, "grad_norm": 0.6363409757614136, "learning_rate": 1.5912000000000003e-06, "loss": 0.0024, "step": 42045 }, { "epoch": 2.2416164631870767, "grad_norm": 1.646777629852295, "learning_rate": 1.591e-06, "loss": 0.0136, "step": 42046 }, { "epoch": 2.24166977661673, "grad_norm": 1.6265780925750732, "learning_rate": 1.5908e-06, "loss": 0.0095, "step": 42047 }, { "epoch": 2.2417230900463827, "grad_norm": 1.4680877923965454, "learning_rate": 1.5906000000000003e-06, "loss": 0.0166, "step": 42048 }, { "epoch": 2.2417764034760355, "grad_norm": 2.1006529331207275, "learning_rate": 1.5904e-06, "loss": 0.0185, "step": 42049 }, { "epoch": 2.2418297169056887, "grad_norm": 0.7049522399902344, "learning_rate": 1.5902e-06, "loss": 0.0039, "step": 42050 }, { "epoch": 2.2418830303353414, "grad_norm": 0.587510347366333, "learning_rate": 1.5900000000000002e-06, "loss": 0.0041, "step": 42051 }, { "epoch": 2.241936343764994, "grad_norm": 2.2211103439331055, "learning_rate": 1.5898000000000002e-06, "loss": 0.0236, "step": 42052 }, { "epoch": 2.2419896571946474, "grad_norm": 0.5682880878448486, "learning_rate": 1.5896e-06, "loss": 0.0032, "step": 42053 }, { "epoch": 2.2420429706243, "grad_norm": 2.096374750137329, "learning_rate": 1.5894000000000002e-06, "loss": 0.0124, "step": 42054 }, { "epoch": 2.2420962840539533, "grad_norm": 1.5685062408447266, "learning_rate": 1.5892000000000002e-06, "loss": 0.0215, "step": 42055 }, { "epoch": 2.242149597483606, "grad_norm": 4.512634754180908, "learning_rate": 1.5890000000000002e-06, "loss": 0.0326, "step": 42056 }, { "epoch": 2.242202910913259, "grad_norm": 0.3919779360294342, "learning_rate": 1.5888000000000001e-06, "loss": 0.0032, "step": 42057 }, { "epoch": 2.242256224342912, "grad_norm": 2.7341701984405518, "learning_rate": 1.5886000000000001e-06, "loss": 0.0263, "step": 42058 }, { "epoch": 2.242309537772565, "grad_norm": 3.8661599159240723, "learning_rate": 1.5884000000000001e-06, "loss": 0.0483, "step": 42059 }, { "epoch": 2.242362851202218, "grad_norm": 1.7988907098770142, "learning_rate": 1.5882e-06, "loss": 0.0197, "step": 42060 }, { "epoch": 2.2424161646318708, "grad_norm": 1.2419987916946411, "learning_rate": 1.588e-06, "loss": 0.0155, "step": 42061 }, { "epoch": 2.2424694780615235, "grad_norm": 2.440033197402954, "learning_rate": 1.5878e-06, "loss": 0.0129, "step": 42062 }, { "epoch": 2.2425227914911767, "grad_norm": 0.6140822768211365, "learning_rate": 1.5876000000000003e-06, "loss": 0.0039, "step": 42063 }, { "epoch": 2.2425761049208295, "grad_norm": 1.832713007926941, "learning_rate": 1.5874e-06, "loss": 0.028, "step": 42064 }, { "epoch": 2.2426294183504827, "grad_norm": 1.341601014137268, "learning_rate": 1.5872e-06, "loss": 0.0159, "step": 42065 }, { "epoch": 2.2426827317801354, "grad_norm": 2.361215829849243, "learning_rate": 1.5870000000000002e-06, "loss": 0.0239, "step": 42066 }, { "epoch": 2.242736045209788, "grad_norm": 2.0460546016693115, "learning_rate": 1.5868e-06, "loss": 0.0328, "step": 42067 }, { "epoch": 2.2427893586394414, "grad_norm": 2.039524555206299, "learning_rate": 1.5866e-06, "loss": 0.0127, "step": 42068 }, { "epoch": 2.242842672069094, "grad_norm": 6.300121784210205, "learning_rate": 1.5864000000000002e-06, "loss": 0.0325, "step": 42069 }, { "epoch": 2.2428959854987474, "grad_norm": 1.4817845821380615, "learning_rate": 1.5862000000000002e-06, "loss": 0.0256, "step": 42070 }, { "epoch": 2.2429492989284, "grad_norm": 1.5585061311721802, "learning_rate": 1.586e-06, "loss": 0.0151, "step": 42071 }, { "epoch": 2.243002612358053, "grad_norm": 2.174060344696045, "learning_rate": 1.5858000000000001e-06, "loss": 0.0072, "step": 42072 }, { "epoch": 2.243055925787706, "grad_norm": 1.7073564529418945, "learning_rate": 1.5856000000000001e-06, "loss": 0.0085, "step": 42073 }, { "epoch": 2.243109239217359, "grad_norm": 1.5235035419464111, "learning_rate": 1.5853999999999999e-06, "loss": 0.0191, "step": 42074 }, { "epoch": 2.2431625526470116, "grad_norm": 2.2141311168670654, "learning_rate": 1.5852e-06, "loss": 0.0201, "step": 42075 }, { "epoch": 2.243215866076665, "grad_norm": 0.7848996520042419, "learning_rate": 1.585e-06, "loss": 0.006, "step": 42076 }, { "epoch": 2.2432691795063175, "grad_norm": 1.2275527715682983, "learning_rate": 1.5848000000000003e-06, "loss": 0.0329, "step": 42077 }, { "epoch": 2.2433224929359707, "grad_norm": 0.9651703834533691, "learning_rate": 1.5846e-06, "loss": 0.005, "step": 42078 }, { "epoch": 2.2433758063656235, "grad_norm": 1.610280156135559, "learning_rate": 1.5844e-06, "loss": 0.0149, "step": 42079 }, { "epoch": 2.2434291197952763, "grad_norm": 1.3052414655685425, "learning_rate": 1.5842000000000002e-06, "loss": 0.0126, "step": 42080 }, { "epoch": 2.2434824332249295, "grad_norm": 2.588083505630493, "learning_rate": 1.5840000000000002e-06, "loss": 0.0083, "step": 42081 }, { "epoch": 2.243535746654582, "grad_norm": 1.547460913658142, "learning_rate": 1.5838e-06, "loss": 0.0145, "step": 42082 }, { "epoch": 2.2435890600842354, "grad_norm": 1.436737060546875, "learning_rate": 1.5836000000000002e-06, "loss": 0.008, "step": 42083 }, { "epoch": 2.243642373513888, "grad_norm": 1.6524572372436523, "learning_rate": 1.5834000000000002e-06, "loss": 0.0268, "step": 42084 }, { "epoch": 2.243695686943541, "grad_norm": 2.539428472518921, "learning_rate": 1.5832e-06, "loss": 0.0127, "step": 42085 }, { "epoch": 2.243749000373194, "grad_norm": 1.045501947402954, "learning_rate": 1.5830000000000001e-06, "loss": 0.0149, "step": 42086 }, { "epoch": 2.243802313802847, "grad_norm": 1.7286572456359863, "learning_rate": 1.5828000000000001e-06, "loss": 0.0097, "step": 42087 }, { "epoch": 2.2438556272324996, "grad_norm": 1.329111099243164, "learning_rate": 1.5826000000000003e-06, "loss": 0.0172, "step": 42088 }, { "epoch": 2.243908940662153, "grad_norm": 1.7555065155029297, "learning_rate": 1.5824e-06, "loss": 0.004, "step": 42089 }, { "epoch": 2.2439622540918056, "grad_norm": 2.4372918605804443, "learning_rate": 1.5822e-06, "loss": 0.0149, "step": 42090 }, { "epoch": 2.244015567521459, "grad_norm": 0.2690314054489136, "learning_rate": 1.5820000000000003e-06, "loss": 0.0021, "step": 42091 }, { "epoch": 2.2440688809511116, "grad_norm": 2.837434768676758, "learning_rate": 1.5818e-06, "loss": 0.0242, "step": 42092 }, { "epoch": 2.2441221943807643, "grad_norm": 3.3550515174865723, "learning_rate": 1.5816e-06, "loss": 0.0552, "step": 42093 }, { "epoch": 2.2441755078104175, "grad_norm": 2.1418707370758057, "learning_rate": 1.5814000000000002e-06, "loss": 0.0083, "step": 42094 }, { "epoch": 2.2442288212400703, "grad_norm": 1.8324991464614868, "learning_rate": 1.5812000000000002e-06, "loss": 0.0088, "step": 42095 }, { "epoch": 2.2442821346697235, "grad_norm": 1.7709895372390747, "learning_rate": 1.581e-06, "loss": 0.0124, "step": 42096 }, { "epoch": 2.2443354480993762, "grad_norm": 0.8754526972770691, "learning_rate": 1.5808000000000002e-06, "loss": 0.0097, "step": 42097 }, { "epoch": 2.244388761529029, "grad_norm": 3.7881383895874023, "learning_rate": 1.5806000000000002e-06, "loss": 0.0308, "step": 42098 }, { "epoch": 2.244442074958682, "grad_norm": 1.8110917806625366, "learning_rate": 1.5804000000000004e-06, "loss": 0.0168, "step": 42099 }, { "epoch": 2.244495388388335, "grad_norm": 1.4603835344314575, "learning_rate": 1.5802000000000001e-06, "loss": 0.008, "step": 42100 }, { "epoch": 2.244548701817988, "grad_norm": 1.366919994354248, "learning_rate": 1.5800000000000001e-06, "loss": 0.0081, "step": 42101 }, { "epoch": 2.244602015247641, "grad_norm": 2.1105732917785645, "learning_rate": 1.5798000000000003e-06, "loss": 0.013, "step": 42102 }, { "epoch": 2.2446553286772937, "grad_norm": 1.806217908859253, "learning_rate": 1.5796e-06, "loss": 0.0073, "step": 42103 }, { "epoch": 2.244708642106947, "grad_norm": 5.73006010055542, "learning_rate": 1.5794e-06, "loss": 0.0163, "step": 42104 }, { "epoch": 2.2447619555365996, "grad_norm": 1.8379931449890137, "learning_rate": 1.5792000000000003e-06, "loss": 0.0092, "step": 42105 }, { "epoch": 2.244815268966253, "grad_norm": 0.7456946969032288, "learning_rate": 1.5790000000000003e-06, "loss": 0.0052, "step": 42106 }, { "epoch": 2.2448685823959056, "grad_norm": 1.7572885751724243, "learning_rate": 1.5788e-06, "loss": 0.0076, "step": 42107 }, { "epoch": 2.2449218958255583, "grad_norm": 2.1569926738739014, "learning_rate": 1.5786000000000002e-06, "loss": 0.0146, "step": 42108 }, { "epoch": 2.2449752092552115, "grad_norm": 1.2051725387573242, "learning_rate": 1.5784000000000002e-06, "loss": 0.0113, "step": 42109 }, { "epoch": 2.2450285226848643, "grad_norm": 1.1568272113800049, "learning_rate": 1.5782e-06, "loss": 0.0216, "step": 42110 }, { "epoch": 2.245081836114517, "grad_norm": 3.826162338256836, "learning_rate": 1.5780000000000002e-06, "loss": 0.0124, "step": 42111 }, { "epoch": 2.2451351495441703, "grad_norm": 2.8894927501678467, "learning_rate": 1.5778000000000002e-06, "loss": 0.0357, "step": 42112 }, { "epoch": 2.245188462973823, "grad_norm": 1.5988694429397583, "learning_rate": 1.5776000000000002e-06, "loss": 0.0169, "step": 42113 }, { "epoch": 2.245241776403476, "grad_norm": 0.9261529445648193, "learning_rate": 1.5774000000000001e-06, "loss": 0.0213, "step": 42114 }, { "epoch": 2.245295089833129, "grad_norm": 1.4814740419387817, "learning_rate": 1.5772000000000001e-06, "loss": 0.0318, "step": 42115 }, { "epoch": 2.2453484032627817, "grad_norm": 2.1041266918182373, "learning_rate": 1.577e-06, "loss": 0.0357, "step": 42116 }, { "epoch": 2.245401716692435, "grad_norm": 0.3431673049926758, "learning_rate": 1.5768e-06, "loss": 0.0029, "step": 42117 }, { "epoch": 2.2454550301220877, "grad_norm": 1.6939867734909058, "learning_rate": 1.5766e-06, "loss": 0.0238, "step": 42118 }, { "epoch": 2.245508343551741, "grad_norm": 2.147382974624634, "learning_rate": 1.5764e-06, "loss": 0.0077, "step": 42119 }, { "epoch": 2.2455616569813937, "grad_norm": 1.721557855606079, "learning_rate": 1.5762000000000003e-06, "loss": 0.0097, "step": 42120 }, { "epoch": 2.2456149704110464, "grad_norm": 3.6774380207061768, "learning_rate": 1.576e-06, "loss": 0.0128, "step": 42121 }, { "epoch": 2.2456682838406996, "grad_norm": 0.30731016397476196, "learning_rate": 1.5758e-06, "loss": 0.0028, "step": 42122 }, { "epoch": 2.2457215972703524, "grad_norm": 1.312330961227417, "learning_rate": 1.5756000000000002e-06, "loss": 0.0078, "step": 42123 }, { "epoch": 2.245774910700005, "grad_norm": 2.884758234024048, "learning_rate": 1.5754000000000002e-06, "loss": 0.036, "step": 42124 }, { "epoch": 2.2458282241296583, "grad_norm": 0.7862004637718201, "learning_rate": 1.5752e-06, "loss": 0.0145, "step": 42125 }, { "epoch": 2.245881537559311, "grad_norm": 1.7806288003921509, "learning_rate": 1.5750000000000002e-06, "loss": 0.0131, "step": 42126 }, { "epoch": 2.2459348509889643, "grad_norm": 0.7597764134407043, "learning_rate": 1.5748000000000002e-06, "loss": 0.0041, "step": 42127 }, { "epoch": 2.245988164418617, "grad_norm": 0.548667311668396, "learning_rate": 1.5746e-06, "loss": 0.0045, "step": 42128 }, { "epoch": 2.24604147784827, "grad_norm": 3.1920382976531982, "learning_rate": 1.5744000000000001e-06, "loss": 0.0122, "step": 42129 }, { "epoch": 2.246094791277923, "grad_norm": 1.8239563703536987, "learning_rate": 1.5742000000000001e-06, "loss": 0.0067, "step": 42130 }, { "epoch": 2.2461481047075758, "grad_norm": 2.2890970706939697, "learning_rate": 1.5740000000000003e-06, "loss": 0.0453, "step": 42131 }, { "epoch": 2.246201418137229, "grad_norm": 0.460729718208313, "learning_rate": 1.5738e-06, "loss": 0.0026, "step": 42132 }, { "epoch": 2.2462547315668817, "grad_norm": 2.044482469558716, "learning_rate": 1.5736e-06, "loss": 0.0248, "step": 42133 }, { "epoch": 2.2463080449965345, "grad_norm": 1.070099949836731, "learning_rate": 1.5734000000000003e-06, "loss": 0.0122, "step": 42134 }, { "epoch": 2.2463613584261877, "grad_norm": 1.4289718866348267, "learning_rate": 1.5732e-06, "loss": 0.0046, "step": 42135 }, { "epoch": 2.2464146718558404, "grad_norm": 1.880313515663147, "learning_rate": 1.573e-06, "loss": 0.013, "step": 42136 }, { "epoch": 2.246467985285493, "grad_norm": 1.056769847869873, "learning_rate": 1.5728000000000002e-06, "loss": 0.0276, "step": 42137 }, { "epoch": 2.2465212987151464, "grad_norm": 2.4792463779449463, "learning_rate": 1.5726000000000002e-06, "loss": 0.0198, "step": 42138 }, { "epoch": 2.246574612144799, "grad_norm": 1.883193850517273, "learning_rate": 1.5724e-06, "loss": 0.0252, "step": 42139 }, { "epoch": 2.2466279255744523, "grad_norm": 3.4624063968658447, "learning_rate": 1.5722000000000002e-06, "loss": 0.0326, "step": 42140 }, { "epoch": 2.246681239004105, "grad_norm": 3.3092663288116455, "learning_rate": 1.5720000000000002e-06, "loss": 0.0359, "step": 42141 }, { "epoch": 2.246734552433758, "grad_norm": 1.1582790613174438, "learning_rate": 1.5718e-06, "loss": 0.0051, "step": 42142 }, { "epoch": 2.246787865863411, "grad_norm": 4.2567267417907715, "learning_rate": 1.5716000000000001e-06, "loss": 0.0296, "step": 42143 }, { "epoch": 2.246841179293064, "grad_norm": 1.4990030527114868, "learning_rate": 1.5714000000000001e-06, "loss": 0.0299, "step": 42144 }, { "epoch": 2.246894492722717, "grad_norm": 0.540710985660553, "learning_rate": 1.5712000000000003e-06, "loss": 0.004, "step": 42145 }, { "epoch": 2.2469478061523698, "grad_norm": 5.262446403503418, "learning_rate": 1.571e-06, "loss": 0.0438, "step": 42146 }, { "epoch": 2.2470011195820225, "grad_norm": 2.829061508178711, "learning_rate": 1.5708e-06, "loss": 0.0125, "step": 42147 }, { "epoch": 2.2470544330116757, "grad_norm": 2.1392078399658203, "learning_rate": 1.5706000000000003e-06, "loss": 0.0155, "step": 42148 }, { "epoch": 2.2471077464413285, "grad_norm": 1.9791877269744873, "learning_rate": 1.5704000000000002e-06, "loss": 0.0324, "step": 42149 }, { "epoch": 2.2471610598709817, "grad_norm": 1.7548861503601074, "learning_rate": 1.5702e-06, "loss": 0.02, "step": 42150 }, { "epoch": 2.2472143733006344, "grad_norm": 0.6891607046127319, "learning_rate": 1.5700000000000002e-06, "loss": 0.0041, "step": 42151 }, { "epoch": 2.247267686730287, "grad_norm": 2.377821683883667, "learning_rate": 1.5698000000000002e-06, "loss": 0.0227, "step": 42152 }, { "epoch": 2.2473210001599404, "grad_norm": 3.111135244369507, "learning_rate": 1.5696e-06, "loss": 0.0237, "step": 42153 }, { "epoch": 2.247374313589593, "grad_norm": 1.8628984689712524, "learning_rate": 1.5694000000000002e-06, "loss": 0.0139, "step": 42154 }, { "epoch": 2.2474276270192464, "grad_norm": 0.6173938512802124, "learning_rate": 1.5692000000000002e-06, "loss": 0.0045, "step": 42155 }, { "epoch": 2.247480940448899, "grad_norm": 0.8211716413497925, "learning_rate": 1.5690000000000001e-06, "loss": 0.0176, "step": 42156 }, { "epoch": 2.247534253878552, "grad_norm": 0.8140172958374023, "learning_rate": 1.5688000000000001e-06, "loss": 0.0114, "step": 42157 }, { "epoch": 2.247587567308205, "grad_norm": 2.2539544105529785, "learning_rate": 1.5686000000000001e-06, "loss": 0.0285, "step": 42158 }, { "epoch": 2.247640880737858, "grad_norm": 1.5911580324172974, "learning_rate": 1.5684e-06, "loss": 0.0091, "step": 42159 }, { "epoch": 2.2476941941675106, "grad_norm": 1.341840147972107, "learning_rate": 1.5682e-06, "loss": 0.0112, "step": 42160 }, { "epoch": 2.247747507597164, "grad_norm": 1.2240992784500122, "learning_rate": 1.568e-06, "loss": 0.0171, "step": 42161 }, { "epoch": 2.2478008210268166, "grad_norm": 1.9905723333358765, "learning_rate": 1.5678e-06, "loss": 0.0151, "step": 42162 }, { "epoch": 2.2478541344564698, "grad_norm": 1.479699730873108, "learning_rate": 1.5676000000000002e-06, "loss": 0.0149, "step": 42163 }, { "epoch": 2.2479074478861225, "grad_norm": 0.5576313138008118, "learning_rate": 1.5674e-06, "loss": 0.0047, "step": 42164 }, { "epoch": 2.2479607613157753, "grad_norm": 1.2318611145019531, "learning_rate": 1.5672e-06, "loss": 0.0098, "step": 42165 }, { "epoch": 2.2480140747454285, "grad_norm": 3.7329163551330566, "learning_rate": 1.5670000000000002e-06, "loss": 0.0386, "step": 42166 }, { "epoch": 2.2480673881750812, "grad_norm": 2.0631606578826904, "learning_rate": 1.5668000000000002e-06, "loss": 0.0251, "step": 42167 }, { "epoch": 2.2481207016047344, "grad_norm": 2.470308780670166, "learning_rate": 1.5666e-06, "loss": 0.0081, "step": 42168 }, { "epoch": 2.248174015034387, "grad_norm": 1.5166805982589722, "learning_rate": 1.5664000000000002e-06, "loss": 0.0179, "step": 42169 }, { "epoch": 2.24822732846404, "grad_norm": 1.7197564840316772, "learning_rate": 1.5662000000000001e-06, "loss": 0.0061, "step": 42170 }, { "epoch": 2.248280641893693, "grad_norm": 2.031031608581543, "learning_rate": 1.566e-06, "loss": 0.0137, "step": 42171 }, { "epoch": 2.248333955323346, "grad_norm": 1.0215184688568115, "learning_rate": 1.5658000000000001e-06, "loss": 0.0166, "step": 42172 }, { "epoch": 2.2483872687529987, "grad_norm": 0.8453013896942139, "learning_rate": 1.5656e-06, "loss": 0.0038, "step": 42173 }, { "epoch": 2.248440582182652, "grad_norm": 1.9374793767929077, "learning_rate": 1.5654000000000003e-06, "loss": 0.0265, "step": 42174 }, { "epoch": 2.2484938956123046, "grad_norm": 1.6709001064300537, "learning_rate": 1.5652e-06, "loss": 0.0066, "step": 42175 }, { "epoch": 2.248547209041958, "grad_norm": 1.5674645900726318, "learning_rate": 1.565e-06, "loss": 0.0234, "step": 42176 }, { "epoch": 2.2486005224716106, "grad_norm": 2.113095760345459, "learning_rate": 1.5648000000000002e-06, "loss": 0.0193, "step": 42177 }, { "epoch": 2.2486538359012633, "grad_norm": 0.8733501434326172, "learning_rate": 1.5646e-06, "loss": 0.0051, "step": 42178 }, { "epoch": 2.2487071493309165, "grad_norm": 2.6802661418914795, "learning_rate": 1.5644e-06, "loss": 0.0246, "step": 42179 }, { "epoch": 2.2487604627605693, "grad_norm": 1.967543125152588, "learning_rate": 1.5642000000000002e-06, "loss": 0.0171, "step": 42180 }, { "epoch": 2.2488137761902225, "grad_norm": 0.5473439693450928, "learning_rate": 1.5640000000000002e-06, "loss": 0.0035, "step": 42181 }, { "epoch": 2.2488670896198752, "grad_norm": 2.1827313899993896, "learning_rate": 1.5638e-06, "loss": 0.0294, "step": 42182 }, { "epoch": 2.248920403049528, "grad_norm": 1.5134657621383667, "learning_rate": 1.5636000000000002e-06, "loss": 0.0177, "step": 42183 }, { "epoch": 2.248973716479181, "grad_norm": 1.1513184309005737, "learning_rate": 1.5634000000000001e-06, "loss": 0.0234, "step": 42184 }, { "epoch": 2.249027029908834, "grad_norm": 2.0095365047454834, "learning_rate": 1.5632e-06, "loss": 0.0195, "step": 42185 }, { "epoch": 2.249080343338487, "grad_norm": 3.1763086318969727, "learning_rate": 1.5630000000000001e-06, "loss": 0.0641, "step": 42186 }, { "epoch": 2.24913365676814, "grad_norm": 0.8712440133094788, "learning_rate": 1.5628e-06, "loss": 0.0046, "step": 42187 }, { "epoch": 2.2491869701977927, "grad_norm": 1.3457465171813965, "learning_rate": 1.5626000000000003e-06, "loss": 0.0353, "step": 42188 }, { "epoch": 2.249240283627446, "grad_norm": 1.226542353630066, "learning_rate": 1.5624e-06, "loss": 0.0046, "step": 42189 }, { "epoch": 2.2492935970570986, "grad_norm": 0.7089195251464844, "learning_rate": 1.5622e-06, "loss": 0.0057, "step": 42190 }, { "epoch": 2.249346910486752, "grad_norm": 1.5248527526855469, "learning_rate": 1.5620000000000002e-06, "loss": 0.025, "step": 42191 }, { "epoch": 2.2494002239164046, "grad_norm": 0.8134058117866516, "learning_rate": 1.5618000000000002e-06, "loss": 0.0047, "step": 42192 }, { "epoch": 2.2494535373460574, "grad_norm": 0.6295433044433594, "learning_rate": 1.5616e-06, "loss": 0.0032, "step": 42193 }, { "epoch": 2.2495068507757106, "grad_norm": 1.2182022333145142, "learning_rate": 1.5614000000000002e-06, "loss": 0.0053, "step": 42194 }, { "epoch": 2.2495601642053633, "grad_norm": 1.4146538972854614, "learning_rate": 1.5612000000000002e-06, "loss": 0.0055, "step": 42195 }, { "epoch": 2.249613477635016, "grad_norm": 2.150374412536621, "learning_rate": 1.561e-06, "loss": 0.0184, "step": 42196 }, { "epoch": 2.2496667910646693, "grad_norm": 1.8978121280670166, "learning_rate": 1.5608000000000002e-06, "loss": 0.0128, "step": 42197 }, { "epoch": 2.249720104494322, "grad_norm": 0.2758711874485016, "learning_rate": 1.5606000000000001e-06, "loss": 0.0028, "step": 42198 }, { "epoch": 2.2497734179239752, "grad_norm": 2.273129463195801, "learning_rate": 1.5604000000000003e-06, "loss": 0.0285, "step": 42199 }, { "epoch": 2.249826731353628, "grad_norm": 2.5638961791992188, "learning_rate": 1.5602000000000001e-06, "loss": 0.0355, "step": 42200 }, { "epoch": 2.2498800447832807, "grad_norm": 1.1882857084274292, "learning_rate": 1.56e-06, "loss": 0.0163, "step": 42201 }, { "epoch": 2.249933358212934, "grad_norm": 1.456405758857727, "learning_rate": 1.5598000000000003e-06, "loss": 0.0082, "step": 42202 }, { "epoch": 2.2499866716425867, "grad_norm": 1.6333057880401611, "learning_rate": 1.5596e-06, "loss": 0.0102, "step": 42203 }, { "epoch": 2.25003998507224, "grad_norm": 2.2969486713409424, "learning_rate": 1.5594e-06, "loss": 0.024, "step": 42204 }, { "epoch": 2.2500932985018927, "grad_norm": 2.5041980743408203, "learning_rate": 1.5592000000000003e-06, "loss": 0.0214, "step": 42205 }, { "epoch": 2.2501466119315454, "grad_norm": 0.1589576005935669, "learning_rate": 1.5590000000000002e-06, "loss": 0.0016, "step": 42206 }, { "epoch": 2.2501999253611986, "grad_norm": 0.28751665353775024, "learning_rate": 1.5588e-06, "loss": 0.0018, "step": 42207 }, { "epoch": 2.2502532387908514, "grad_norm": 5.607679843902588, "learning_rate": 1.5586000000000002e-06, "loss": 0.0284, "step": 42208 }, { "epoch": 2.250306552220504, "grad_norm": 1.7925519943237305, "learning_rate": 1.5584000000000002e-06, "loss": 0.0242, "step": 42209 }, { "epoch": 2.2503598656501573, "grad_norm": 1.3921689987182617, "learning_rate": 1.5582e-06, "loss": 0.0218, "step": 42210 }, { "epoch": 2.25041317907981, "grad_norm": 1.851466417312622, "learning_rate": 1.5580000000000002e-06, "loss": 0.0193, "step": 42211 }, { "epoch": 2.2504664925094633, "grad_norm": 1.399103045463562, "learning_rate": 1.5578000000000001e-06, "loss": 0.0136, "step": 42212 }, { "epoch": 2.250519805939116, "grad_norm": 1.6184734106063843, "learning_rate": 1.5576000000000001e-06, "loss": 0.0081, "step": 42213 }, { "epoch": 2.250573119368769, "grad_norm": 1.156875729560852, "learning_rate": 1.5574000000000001e-06, "loss": 0.0225, "step": 42214 }, { "epoch": 2.250626432798422, "grad_norm": 1.3388850688934326, "learning_rate": 1.5572e-06, "loss": 0.0073, "step": 42215 }, { "epoch": 2.2506797462280748, "grad_norm": 0.9039285182952881, "learning_rate": 1.557e-06, "loss": 0.007, "step": 42216 }, { "epoch": 2.250733059657728, "grad_norm": 2.5999252796173096, "learning_rate": 1.5568000000000003e-06, "loss": 0.0222, "step": 42217 }, { "epoch": 2.2507863730873807, "grad_norm": 1.1748816967010498, "learning_rate": 1.5566e-06, "loss": 0.0073, "step": 42218 }, { "epoch": 2.2508396865170335, "grad_norm": 1.3470734357833862, "learning_rate": 1.5564e-06, "loss": 0.0069, "step": 42219 }, { "epoch": 2.2508929999466867, "grad_norm": 0.6344759464263916, "learning_rate": 1.5562000000000002e-06, "loss": 0.0055, "step": 42220 }, { "epoch": 2.2509463133763394, "grad_norm": 2.66341233253479, "learning_rate": 1.556e-06, "loss": 0.0142, "step": 42221 }, { "epoch": 2.250999626805992, "grad_norm": 1.9128270149230957, "learning_rate": 1.5558e-06, "loss": 0.022, "step": 42222 }, { "epoch": 2.2510529402356454, "grad_norm": 0.975682258605957, "learning_rate": 1.5556000000000002e-06, "loss": 0.005, "step": 42223 }, { "epoch": 2.251106253665298, "grad_norm": 2.006540536880493, "learning_rate": 1.5554000000000002e-06, "loss": 0.0076, "step": 42224 }, { "epoch": 2.2511595670949514, "grad_norm": 1.0705095529556274, "learning_rate": 1.5552e-06, "loss": 0.024, "step": 42225 }, { "epoch": 2.251212880524604, "grad_norm": 3.1316452026367188, "learning_rate": 1.5550000000000001e-06, "loss": 0.0146, "step": 42226 }, { "epoch": 2.2512661939542573, "grad_norm": 0.5551312565803528, "learning_rate": 1.5548000000000001e-06, "loss": 0.0035, "step": 42227 }, { "epoch": 2.25131950738391, "grad_norm": 1.1359682083129883, "learning_rate": 1.5546e-06, "loss": 0.0165, "step": 42228 }, { "epoch": 2.251372820813563, "grad_norm": 2.896028757095337, "learning_rate": 1.5544e-06, "loss": 0.0479, "step": 42229 }, { "epoch": 2.251426134243216, "grad_norm": 0.8950355648994446, "learning_rate": 1.5542e-06, "loss": 0.0201, "step": 42230 }, { "epoch": 2.251479447672869, "grad_norm": 0.41833966970443726, "learning_rate": 1.5540000000000003e-06, "loss": 0.003, "step": 42231 }, { "epoch": 2.2515327611025215, "grad_norm": 2.916175603866577, "learning_rate": 1.5538e-06, "loss": 0.0236, "step": 42232 }, { "epoch": 2.2515860745321747, "grad_norm": 1.2389928102493286, "learning_rate": 1.5536e-06, "loss": 0.0108, "step": 42233 }, { "epoch": 2.2516393879618275, "grad_norm": 1.1761531829833984, "learning_rate": 1.5534000000000002e-06, "loss": 0.0044, "step": 42234 }, { "epoch": 2.2516927013914807, "grad_norm": 1.3278805017471313, "learning_rate": 1.5532000000000002e-06, "loss": 0.0316, "step": 42235 }, { "epoch": 2.2517460148211335, "grad_norm": 1.7276153564453125, "learning_rate": 1.553e-06, "loss": 0.0183, "step": 42236 }, { "epoch": 2.251799328250786, "grad_norm": 1.2729225158691406, "learning_rate": 1.5528000000000002e-06, "loss": 0.0075, "step": 42237 }, { "epoch": 2.2518526416804394, "grad_norm": 0.5065354108810425, "learning_rate": 1.5526000000000002e-06, "loss": 0.0038, "step": 42238 }, { "epoch": 2.251905955110092, "grad_norm": 0.3385589122772217, "learning_rate": 1.5524e-06, "loss": 0.0031, "step": 42239 }, { "epoch": 2.2519592685397454, "grad_norm": 2.2909319400787354, "learning_rate": 1.5522000000000001e-06, "loss": 0.026, "step": 42240 }, { "epoch": 2.252012581969398, "grad_norm": 1.9498385190963745, "learning_rate": 1.5520000000000001e-06, "loss": 0.017, "step": 42241 }, { "epoch": 2.252065895399051, "grad_norm": 1.0456594228744507, "learning_rate": 1.5518000000000003e-06, "loss": 0.0111, "step": 42242 }, { "epoch": 2.252119208828704, "grad_norm": 0.6481245756149292, "learning_rate": 1.5516e-06, "loss": 0.0037, "step": 42243 }, { "epoch": 2.252172522258357, "grad_norm": 3.05700945854187, "learning_rate": 1.5514e-06, "loss": 0.0091, "step": 42244 }, { "epoch": 2.2522258356880096, "grad_norm": 3.4635798931121826, "learning_rate": 1.5512000000000003e-06, "loss": 0.0751, "step": 42245 }, { "epoch": 2.252279149117663, "grad_norm": 2.1937854290008545, "learning_rate": 1.551e-06, "loss": 0.0128, "step": 42246 }, { "epoch": 2.2523324625473156, "grad_norm": 1.4960402250289917, "learning_rate": 1.5508e-06, "loss": 0.0172, "step": 42247 }, { "epoch": 2.2523857759769688, "grad_norm": 2.0598034858703613, "learning_rate": 1.5506000000000002e-06, "loss": 0.0221, "step": 42248 }, { "epoch": 2.2524390894066215, "grad_norm": 1.392255187034607, "learning_rate": 1.5504000000000002e-06, "loss": 0.0171, "step": 42249 }, { "epoch": 2.2524924028362743, "grad_norm": 1.6561344861984253, "learning_rate": 1.5502e-06, "loss": 0.0176, "step": 42250 }, { "epoch": 2.2525457162659275, "grad_norm": 2.5467708110809326, "learning_rate": 1.5500000000000002e-06, "loss": 0.0078, "step": 42251 }, { "epoch": 2.2525990296955802, "grad_norm": 3.396291971206665, "learning_rate": 1.5498000000000002e-06, "loss": 0.0106, "step": 42252 }, { "epoch": 2.2526523431252334, "grad_norm": 2.5357155799865723, "learning_rate": 1.5496e-06, "loss": 0.0052, "step": 42253 }, { "epoch": 2.252705656554886, "grad_norm": 0.968488335609436, "learning_rate": 1.5494000000000001e-06, "loss": 0.0066, "step": 42254 }, { "epoch": 2.252758969984539, "grad_norm": 3.0944314002990723, "learning_rate": 1.5492000000000001e-06, "loss": 0.0246, "step": 42255 }, { "epoch": 2.252812283414192, "grad_norm": 0.7450485825538635, "learning_rate": 1.5490000000000003e-06, "loss": 0.0044, "step": 42256 }, { "epoch": 2.252865596843845, "grad_norm": 0.27685296535491943, "learning_rate": 1.5488e-06, "loss": 0.0023, "step": 42257 }, { "epoch": 2.2529189102734977, "grad_norm": 1.1613764762878418, "learning_rate": 1.5486e-06, "loss": 0.0049, "step": 42258 }, { "epoch": 2.252972223703151, "grad_norm": 2.1621434688568115, "learning_rate": 1.5484000000000003e-06, "loss": 0.0128, "step": 42259 }, { "epoch": 2.2530255371328036, "grad_norm": 1.4671220779418945, "learning_rate": 1.5482000000000003e-06, "loss": 0.0082, "step": 42260 }, { "epoch": 2.253078850562457, "grad_norm": 0.6076477766036987, "learning_rate": 1.548e-06, "loss": 0.0038, "step": 42261 }, { "epoch": 2.2531321639921096, "grad_norm": 1.1858285665512085, "learning_rate": 1.5478000000000002e-06, "loss": 0.0061, "step": 42262 }, { "epoch": 2.253185477421763, "grad_norm": 3.1814160346984863, "learning_rate": 1.5476000000000002e-06, "loss": 0.0321, "step": 42263 }, { "epoch": 2.2532387908514155, "grad_norm": 5.226125240325928, "learning_rate": 1.5474e-06, "loss": 0.0323, "step": 42264 }, { "epoch": 2.2532921042810683, "grad_norm": 2.211333751678467, "learning_rate": 1.5472000000000002e-06, "loss": 0.0172, "step": 42265 }, { "epoch": 2.2533454177107215, "grad_norm": 0.9358463883399963, "learning_rate": 1.5470000000000002e-06, "loss": 0.0058, "step": 42266 }, { "epoch": 2.2533987311403743, "grad_norm": 1.673394799232483, "learning_rate": 1.5468000000000002e-06, "loss": 0.0131, "step": 42267 }, { "epoch": 2.253452044570027, "grad_norm": 0.6800279021263123, "learning_rate": 1.5466000000000002e-06, "loss": 0.0049, "step": 42268 }, { "epoch": 2.25350535799968, "grad_norm": 2.0258758068084717, "learning_rate": 1.5464000000000001e-06, "loss": 0.0166, "step": 42269 }, { "epoch": 2.253558671429333, "grad_norm": 2.0562798976898193, "learning_rate": 1.5462000000000001e-06, "loss": 0.0292, "step": 42270 }, { "epoch": 2.253611984858986, "grad_norm": 4.1273064613342285, "learning_rate": 1.546e-06, "loss": 0.0265, "step": 42271 }, { "epoch": 2.253665298288639, "grad_norm": 5.514099597930908, "learning_rate": 1.5458e-06, "loss": 0.0385, "step": 42272 }, { "epoch": 2.2537186117182917, "grad_norm": 2.0349340438842773, "learning_rate": 1.5456e-06, "loss": 0.0127, "step": 42273 }, { "epoch": 2.253771925147945, "grad_norm": 1.2274397611618042, "learning_rate": 1.5454000000000003e-06, "loss": 0.0251, "step": 42274 }, { "epoch": 2.2538252385775976, "grad_norm": 1.4655593633651733, "learning_rate": 1.5452e-06, "loss": 0.0432, "step": 42275 }, { "epoch": 2.253878552007251, "grad_norm": 2.328641176223755, "learning_rate": 1.545e-06, "loss": 0.018, "step": 42276 }, { "epoch": 2.2539318654369036, "grad_norm": 1.571895956993103, "learning_rate": 1.5448000000000002e-06, "loss": 0.0191, "step": 42277 }, { "epoch": 2.2539851788665564, "grad_norm": 2.3610734939575195, "learning_rate": 1.5446e-06, "loss": 0.0136, "step": 42278 }, { "epoch": 2.2540384922962096, "grad_norm": 3.5523204803466797, "learning_rate": 1.5444e-06, "loss": 0.0143, "step": 42279 }, { "epoch": 2.2540918057258623, "grad_norm": 2.0084474086761475, "learning_rate": 1.5442000000000002e-06, "loss": 0.0218, "step": 42280 }, { "epoch": 2.254145119155515, "grad_norm": 2.4266228675842285, "learning_rate": 1.5440000000000002e-06, "loss": 0.0254, "step": 42281 }, { "epoch": 2.2541984325851683, "grad_norm": 0.554890513420105, "learning_rate": 1.5438e-06, "loss": 0.0036, "step": 42282 }, { "epoch": 2.254251746014821, "grad_norm": 1.364255666732788, "learning_rate": 1.5436000000000001e-06, "loss": 0.0318, "step": 42283 }, { "epoch": 2.2543050594444742, "grad_norm": 0.9812746047973633, "learning_rate": 1.5434000000000001e-06, "loss": 0.0078, "step": 42284 }, { "epoch": 2.254358372874127, "grad_norm": 1.1183594465255737, "learning_rate": 1.5432000000000003e-06, "loss": 0.0066, "step": 42285 }, { "epoch": 2.2544116863037797, "grad_norm": 2.3157360553741455, "learning_rate": 1.543e-06, "loss": 0.0308, "step": 42286 }, { "epoch": 2.254464999733433, "grad_norm": 1.5660755634307861, "learning_rate": 1.5428e-06, "loss": 0.0169, "step": 42287 }, { "epoch": 2.2545183131630857, "grad_norm": 1.8472307920455933, "learning_rate": 1.5426000000000003e-06, "loss": 0.0349, "step": 42288 }, { "epoch": 2.254571626592739, "grad_norm": 1.4903686046600342, "learning_rate": 1.5424e-06, "loss": 0.021, "step": 42289 }, { "epoch": 2.2546249400223917, "grad_norm": 3.8034589290618896, "learning_rate": 1.5422e-06, "loss": 0.0325, "step": 42290 }, { "epoch": 2.2546782534520444, "grad_norm": 0.4162273705005646, "learning_rate": 1.5420000000000002e-06, "loss": 0.0034, "step": 42291 }, { "epoch": 2.2547315668816976, "grad_norm": 1.054686188697815, "learning_rate": 1.5418000000000002e-06, "loss": 0.0064, "step": 42292 }, { "epoch": 2.2547848803113504, "grad_norm": 1.325745701789856, "learning_rate": 1.5416e-06, "loss": 0.0154, "step": 42293 }, { "epoch": 2.254838193741003, "grad_norm": 2.672339916229248, "learning_rate": 1.5414000000000002e-06, "loss": 0.0213, "step": 42294 }, { "epoch": 2.2548915071706563, "grad_norm": 1.9905731678009033, "learning_rate": 1.5412000000000002e-06, "loss": 0.0295, "step": 42295 }, { "epoch": 2.254944820600309, "grad_norm": 3.2286601066589355, "learning_rate": 1.541e-06, "loss": 0.0219, "step": 42296 }, { "epoch": 2.2549981340299623, "grad_norm": 1.7223937511444092, "learning_rate": 1.5408000000000001e-06, "loss": 0.02, "step": 42297 }, { "epoch": 2.255051447459615, "grad_norm": 1.529020071029663, "learning_rate": 1.5406000000000001e-06, "loss": 0.0207, "step": 42298 }, { "epoch": 2.2551047608892683, "grad_norm": 1.5025004148483276, "learning_rate": 1.5404000000000003e-06, "loss": 0.009, "step": 42299 }, { "epoch": 2.255158074318921, "grad_norm": 1.8833651542663574, "learning_rate": 1.5402e-06, "loss": 0.0141, "step": 42300 }, { "epoch": 2.2552113877485738, "grad_norm": 2.8778674602508545, "learning_rate": 1.54e-06, "loss": 0.0222, "step": 42301 }, { "epoch": 2.255264701178227, "grad_norm": 1.5693055391311646, "learning_rate": 1.5398000000000003e-06, "loss": 0.0101, "step": 42302 }, { "epoch": 2.2553180146078797, "grad_norm": 2.4693639278411865, "learning_rate": 1.5396000000000003e-06, "loss": 0.0156, "step": 42303 }, { "epoch": 2.2553713280375325, "grad_norm": 0.8896510601043701, "learning_rate": 1.5394e-06, "loss": 0.0045, "step": 42304 }, { "epoch": 2.2554246414671857, "grad_norm": 1.3098403215408325, "learning_rate": 1.5392000000000002e-06, "loss": 0.0072, "step": 42305 }, { "epoch": 2.2554779548968384, "grad_norm": 2.102405548095703, "learning_rate": 1.5390000000000002e-06, "loss": 0.0099, "step": 42306 }, { "epoch": 2.255531268326491, "grad_norm": 1.9336146116256714, "learning_rate": 1.5388e-06, "loss": 0.0068, "step": 42307 }, { "epoch": 2.2555845817561444, "grad_norm": 1.3664467334747314, "learning_rate": 1.5386000000000002e-06, "loss": 0.0194, "step": 42308 }, { "epoch": 2.255637895185797, "grad_norm": 0.8983915448188782, "learning_rate": 1.5384000000000002e-06, "loss": 0.0047, "step": 42309 }, { "epoch": 2.2556912086154504, "grad_norm": 1.467081904411316, "learning_rate": 1.5382000000000002e-06, "loss": 0.0109, "step": 42310 }, { "epoch": 2.255744522045103, "grad_norm": 2.362687587738037, "learning_rate": 1.5380000000000001e-06, "loss": 0.0205, "step": 42311 }, { "epoch": 2.2557978354747563, "grad_norm": 1.5602045059204102, "learning_rate": 1.5378000000000001e-06, "loss": 0.0103, "step": 42312 }, { "epoch": 2.255851148904409, "grad_norm": 3.001732349395752, "learning_rate": 1.5376000000000001e-06, "loss": 0.0255, "step": 42313 }, { "epoch": 2.255904462334062, "grad_norm": 0.94907146692276, "learning_rate": 1.5374e-06, "loss": 0.0167, "step": 42314 }, { "epoch": 2.255957775763715, "grad_norm": 1.6488909721374512, "learning_rate": 1.5372e-06, "loss": 0.0269, "step": 42315 }, { "epoch": 2.256011089193368, "grad_norm": 1.9710040092468262, "learning_rate": 1.537e-06, "loss": 0.016, "step": 42316 }, { "epoch": 2.2560644026230205, "grad_norm": 1.4466785192489624, "learning_rate": 1.5368000000000003e-06, "loss": 0.0174, "step": 42317 }, { "epoch": 2.2561177160526737, "grad_norm": 4.155156135559082, "learning_rate": 1.5366e-06, "loss": 0.0318, "step": 42318 }, { "epoch": 2.2561710294823265, "grad_norm": 1.3488010168075562, "learning_rate": 1.5364e-06, "loss": 0.0174, "step": 42319 }, { "epoch": 2.2562243429119797, "grad_norm": 2.4206297397613525, "learning_rate": 1.5362000000000002e-06, "loss": 0.0244, "step": 42320 }, { "epoch": 2.2562776563416325, "grad_norm": 0.8397015333175659, "learning_rate": 1.536e-06, "loss": 0.0054, "step": 42321 }, { "epoch": 2.256330969771285, "grad_norm": 1.8854609727859497, "learning_rate": 1.5358e-06, "loss": 0.0137, "step": 42322 }, { "epoch": 2.2563842832009384, "grad_norm": 0.8458443284034729, "learning_rate": 1.5356000000000002e-06, "loss": 0.0049, "step": 42323 }, { "epoch": 2.256437596630591, "grad_norm": 1.2152447700500488, "learning_rate": 1.5354000000000002e-06, "loss": 0.0059, "step": 42324 }, { "epoch": 2.2564909100602444, "grad_norm": 2.5262136459350586, "learning_rate": 1.5352e-06, "loss": 0.013, "step": 42325 }, { "epoch": 2.256544223489897, "grad_norm": 4.065900802612305, "learning_rate": 1.5350000000000001e-06, "loss": 0.0317, "step": 42326 }, { "epoch": 2.25659753691955, "grad_norm": 1.726730227470398, "learning_rate": 1.5348000000000001e-06, "loss": 0.0081, "step": 42327 }, { "epoch": 2.256650850349203, "grad_norm": 1.6275147199630737, "learning_rate": 1.5346000000000003e-06, "loss": 0.0181, "step": 42328 }, { "epoch": 2.256704163778856, "grad_norm": 2.1561946868896484, "learning_rate": 1.5344e-06, "loss": 0.0118, "step": 42329 }, { "epoch": 2.2567574772085086, "grad_norm": 1.1883617639541626, "learning_rate": 1.5342e-06, "loss": 0.006, "step": 42330 }, { "epoch": 2.256810790638162, "grad_norm": 1.8201643228530884, "learning_rate": 1.5340000000000003e-06, "loss": 0.0234, "step": 42331 }, { "epoch": 2.2568641040678146, "grad_norm": 2.2435054779052734, "learning_rate": 1.5338e-06, "loss": 0.0235, "step": 42332 }, { "epoch": 2.2569174174974678, "grad_norm": 1.4827041625976562, "learning_rate": 1.5336e-06, "loss": 0.0186, "step": 42333 }, { "epoch": 2.2569707309271205, "grad_norm": 1.122299075126648, "learning_rate": 1.5334000000000002e-06, "loss": 0.0045, "step": 42334 }, { "epoch": 2.2570240443567733, "grad_norm": 2.515732765197754, "learning_rate": 1.5332000000000002e-06, "loss": 0.0214, "step": 42335 }, { "epoch": 2.2570773577864265, "grad_norm": 1.0570636987686157, "learning_rate": 1.533e-06, "loss": 0.0127, "step": 42336 }, { "epoch": 2.2571306712160792, "grad_norm": 0.7262169718742371, "learning_rate": 1.5328000000000002e-06, "loss": 0.0211, "step": 42337 }, { "epoch": 2.2571839846457324, "grad_norm": 0.7242742776870728, "learning_rate": 1.5326000000000002e-06, "loss": 0.0039, "step": 42338 }, { "epoch": 2.257237298075385, "grad_norm": 3.6447079181671143, "learning_rate": 1.5324e-06, "loss": 0.0236, "step": 42339 }, { "epoch": 2.257290611505038, "grad_norm": 2.6851284503936768, "learning_rate": 1.5322000000000001e-06, "loss": 0.0179, "step": 42340 }, { "epoch": 2.257343924934691, "grad_norm": 0.3433149456977844, "learning_rate": 1.5320000000000001e-06, "loss": 0.0026, "step": 42341 }, { "epoch": 2.257397238364344, "grad_norm": 1.0399117469787598, "learning_rate": 1.5318000000000003e-06, "loss": 0.0041, "step": 42342 }, { "epoch": 2.2574505517939967, "grad_norm": 1.7545019388198853, "learning_rate": 1.5316e-06, "loss": 0.0293, "step": 42343 }, { "epoch": 2.25750386522365, "grad_norm": 2.6877291202545166, "learning_rate": 1.5314e-06, "loss": 0.0241, "step": 42344 }, { "epoch": 2.2575571786533026, "grad_norm": 1.6147338151931763, "learning_rate": 1.5312000000000003e-06, "loss": 0.0053, "step": 42345 }, { "epoch": 2.257610492082956, "grad_norm": 2.1797420978546143, "learning_rate": 1.5310000000000002e-06, "loss": 0.0109, "step": 42346 }, { "epoch": 2.2576638055126086, "grad_norm": 3.082681179046631, "learning_rate": 1.5308e-06, "loss": 0.032, "step": 42347 }, { "epoch": 2.257717118942262, "grad_norm": 1.337297797203064, "learning_rate": 1.5306000000000002e-06, "loss": 0.0041, "step": 42348 }, { "epoch": 2.2577704323719145, "grad_norm": 2.413212776184082, "learning_rate": 1.5304000000000002e-06, "loss": 0.0176, "step": 42349 }, { "epoch": 2.2578237458015673, "grad_norm": 0.84288090467453, "learning_rate": 1.5302e-06, "loss": 0.003, "step": 42350 }, { "epoch": 2.2578770592312205, "grad_norm": 1.4484813213348389, "learning_rate": 1.5300000000000002e-06, "loss": 0.0057, "step": 42351 }, { "epoch": 2.2579303726608733, "grad_norm": 1.0447673797607422, "learning_rate": 1.5298000000000002e-06, "loss": 0.0033, "step": 42352 }, { "epoch": 2.257983686090526, "grad_norm": 2.8307063579559326, "learning_rate": 1.5296000000000004e-06, "loss": 0.0167, "step": 42353 }, { "epoch": 2.2580369995201792, "grad_norm": 1.7064982652664185, "learning_rate": 1.5294000000000001e-06, "loss": 0.0133, "step": 42354 }, { "epoch": 2.258090312949832, "grad_norm": 0.47927823662757874, "learning_rate": 1.5292000000000001e-06, "loss": 0.003, "step": 42355 }, { "epoch": 2.258143626379485, "grad_norm": 1.1353613138198853, "learning_rate": 1.5290000000000003e-06, "loss": 0.0134, "step": 42356 }, { "epoch": 2.258196939809138, "grad_norm": 2.7024974822998047, "learning_rate": 1.5288e-06, "loss": 0.049, "step": 42357 }, { "epoch": 2.2582502532387907, "grad_norm": 0.32560864090919495, "learning_rate": 1.5286e-06, "loss": 0.003, "step": 42358 }, { "epoch": 2.258303566668444, "grad_norm": 0.638116717338562, "learning_rate": 1.5284000000000003e-06, "loss": 0.0074, "step": 42359 }, { "epoch": 2.2583568800980967, "grad_norm": 1.4894871711730957, "learning_rate": 1.5282000000000003e-06, "loss": 0.0126, "step": 42360 }, { "epoch": 2.25841019352775, "grad_norm": 3.5180575847625732, "learning_rate": 1.528e-06, "loss": 0.0277, "step": 42361 }, { "epoch": 2.2584635069574026, "grad_norm": 3.1042263507843018, "learning_rate": 1.5278000000000002e-06, "loss": 0.041, "step": 42362 }, { "epoch": 2.2585168203870554, "grad_norm": 2.609158992767334, "learning_rate": 1.5276000000000002e-06, "loss": 0.0168, "step": 42363 }, { "epoch": 2.2585701338167086, "grad_norm": 3.1982619762420654, "learning_rate": 1.5274e-06, "loss": 0.0169, "step": 42364 }, { "epoch": 2.2586234472463613, "grad_norm": 1.0690370798110962, "learning_rate": 1.5272000000000002e-06, "loss": 0.0056, "step": 42365 }, { "epoch": 2.258676760676014, "grad_norm": 2.299602508544922, "learning_rate": 1.5270000000000002e-06, "loss": 0.0098, "step": 42366 }, { "epoch": 2.2587300741056673, "grad_norm": 1.8799992799758911, "learning_rate": 1.5268000000000001e-06, "loss": 0.0239, "step": 42367 }, { "epoch": 2.25878338753532, "grad_norm": 4.554165363311768, "learning_rate": 1.5266000000000001e-06, "loss": 0.0219, "step": 42368 }, { "epoch": 2.2588367009649732, "grad_norm": 3.263998031616211, "learning_rate": 1.5264000000000001e-06, "loss": 0.0282, "step": 42369 }, { "epoch": 2.258890014394626, "grad_norm": 2.2235107421875, "learning_rate": 1.5262e-06, "loss": 0.0096, "step": 42370 }, { "epoch": 2.2589433278242788, "grad_norm": 0.6729739308357239, "learning_rate": 1.5260000000000003e-06, "loss": 0.0042, "step": 42371 }, { "epoch": 2.258996641253932, "grad_norm": 0.8488525748252869, "learning_rate": 1.5258e-06, "loss": 0.0037, "step": 42372 }, { "epoch": 2.2590499546835847, "grad_norm": 2.052431106567383, "learning_rate": 1.5256e-06, "loss": 0.0093, "step": 42373 }, { "epoch": 2.259103268113238, "grad_norm": 0.7879768013954163, "learning_rate": 1.5254000000000003e-06, "loss": 0.0148, "step": 42374 }, { "epoch": 2.2591565815428907, "grad_norm": 1.19237220287323, "learning_rate": 1.5252e-06, "loss": 0.009, "step": 42375 }, { "epoch": 2.2592098949725434, "grad_norm": 1.0387593507766724, "learning_rate": 1.525e-06, "loss": 0.0133, "step": 42376 }, { "epoch": 2.2592632084021966, "grad_norm": 1.6862090826034546, "learning_rate": 1.5248000000000002e-06, "loss": 0.0083, "step": 42377 }, { "epoch": 2.2593165218318494, "grad_norm": 0.8618777394294739, "learning_rate": 1.5246000000000002e-06, "loss": 0.0233, "step": 42378 }, { "epoch": 2.259369835261502, "grad_norm": 0.23656481504440308, "learning_rate": 1.5244e-06, "loss": 0.0018, "step": 42379 }, { "epoch": 2.2594231486911553, "grad_norm": 0.19484901428222656, "learning_rate": 1.5242000000000002e-06, "loss": 0.0013, "step": 42380 }, { "epoch": 2.259476462120808, "grad_norm": 0.8278285264968872, "learning_rate": 1.5240000000000001e-06, "loss": 0.0036, "step": 42381 }, { "epoch": 2.2595297755504613, "grad_norm": 1.0318293571472168, "learning_rate": 1.5238e-06, "loss": 0.0074, "step": 42382 }, { "epoch": 2.259583088980114, "grad_norm": 1.5233429670333862, "learning_rate": 1.5236000000000001e-06, "loss": 0.0058, "step": 42383 }, { "epoch": 2.2596364024097673, "grad_norm": 0.340782105922699, "learning_rate": 1.5234e-06, "loss": 0.0023, "step": 42384 }, { "epoch": 2.25968971583942, "grad_norm": 0.6173714399337769, "learning_rate": 1.5232000000000003e-06, "loss": 0.0029, "step": 42385 }, { "epoch": 2.259743029269073, "grad_norm": 1.4144798517227173, "learning_rate": 1.523e-06, "loss": 0.0084, "step": 42386 }, { "epoch": 2.259796342698726, "grad_norm": 6.018331527709961, "learning_rate": 1.5228e-06, "loss": 0.0188, "step": 42387 }, { "epoch": 2.2598496561283787, "grad_norm": 0.22469139099121094, "learning_rate": 1.5226000000000003e-06, "loss": 0.0016, "step": 42388 }, { "epoch": 2.2599029695580315, "grad_norm": 1.5198825597763062, "learning_rate": 1.5224e-06, "loss": 0.0157, "step": 42389 }, { "epoch": 2.2599562829876847, "grad_norm": 2.2717504501342773, "learning_rate": 1.5222e-06, "loss": 0.0264, "step": 42390 }, { "epoch": 2.2600095964173375, "grad_norm": 6.60758638381958, "learning_rate": 1.5220000000000002e-06, "loss": 0.0112, "step": 42391 }, { "epoch": 2.26006290984699, "grad_norm": 3.211346387863159, "learning_rate": 1.5218000000000002e-06, "loss": 0.037, "step": 42392 }, { "epoch": 2.2601162232766434, "grad_norm": 0.20093593001365662, "learning_rate": 1.5216e-06, "loss": 0.002, "step": 42393 }, { "epoch": 2.260169536706296, "grad_norm": 0.1506974995136261, "learning_rate": 1.5214000000000002e-06, "loss": 0.0018, "step": 42394 }, { "epoch": 2.2602228501359494, "grad_norm": 1.9609559774398804, "learning_rate": 1.5212000000000001e-06, "loss": 0.0139, "step": 42395 }, { "epoch": 2.260276163565602, "grad_norm": 1.6677641868591309, "learning_rate": 1.5210000000000003e-06, "loss": 0.007, "step": 42396 }, { "epoch": 2.2603294769952553, "grad_norm": 4.1556901931762695, "learning_rate": 1.5208000000000001e-06, "loss": 0.0269, "step": 42397 }, { "epoch": 2.260382790424908, "grad_norm": 1.5703492164611816, "learning_rate": 1.5206e-06, "loss": 0.007, "step": 42398 }, { "epoch": 2.260436103854561, "grad_norm": 2.2357189655303955, "learning_rate": 1.5204000000000003e-06, "loss": 0.0082, "step": 42399 }, { "epoch": 2.260489417284214, "grad_norm": 3.444676160812378, "learning_rate": 1.5202e-06, "loss": 0.0528, "step": 42400 }, { "epoch": 2.260542730713867, "grad_norm": 2.0241267681121826, "learning_rate": 1.52e-06, "loss": 0.0348, "step": 42401 }, { "epoch": 2.2605960441435196, "grad_norm": 2.473297119140625, "learning_rate": 1.5198000000000003e-06, "loss": 0.0183, "step": 42402 }, { "epoch": 2.2606493575731728, "grad_norm": 1.3764619827270508, "learning_rate": 1.5196000000000002e-06, "loss": 0.0157, "step": 42403 }, { "epoch": 2.2607026710028255, "grad_norm": 1.6771339178085327, "learning_rate": 1.5194e-06, "loss": 0.0251, "step": 42404 }, { "epoch": 2.2607559844324787, "grad_norm": 1.3402153253555298, "learning_rate": 1.5192000000000002e-06, "loss": 0.0264, "step": 42405 }, { "epoch": 2.2608092978621315, "grad_norm": 3.457124710083008, "learning_rate": 1.5190000000000002e-06, "loss": 0.0357, "step": 42406 }, { "epoch": 2.2608626112917842, "grad_norm": 2.6852779388427734, "learning_rate": 1.5188e-06, "loss": 0.0211, "step": 42407 }, { "epoch": 2.2609159247214374, "grad_norm": 0.9131448268890381, "learning_rate": 1.5186000000000002e-06, "loss": 0.0187, "step": 42408 }, { "epoch": 2.26096923815109, "grad_norm": 2.344043016433716, "learning_rate": 1.5184000000000001e-06, "loss": 0.0289, "step": 42409 }, { "epoch": 2.2610225515807434, "grad_norm": 1.0226870775222778, "learning_rate": 1.5182000000000001e-06, "loss": 0.0076, "step": 42410 }, { "epoch": 2.261075865010396, "grad_norm": 1.461721420288086, "learning_rate": 1.5180000000000001e-06, "loss": 0.0185, "step": 42411 }, { "epoch": 2.261129178440049, "grad_norm": 2.7135987281799316, "learning_rate": 1.5178e-06, "loss": 0.0356, "step": 42412 }, { "epoch": 2.261182491869702, "grad_norm": 2.9992949962615967, "learning_rate": 1.5176e-06, "loss": 0.0187, "step": 42413 }, { "epoch": 2.261235805299355, "grad_norm": 0.8282670378684998, "learning_rate": 1.5174000000000003e-06, "loss": 0.0114, "step": 42414 }, { "epoch": 2.2612891187290076, "grad_norm": 3.215981960296631, "learning_rate": 1.5172e-06, "loss": 0.0327, "step": 42415 }, { "epoch": 2.261342432158661, "grad_norm": 2.5265414714813232, "learning_rate": 1.517e-06, "loss": 0.0195, "step": 42416 }, { "epoch": 2.2613957455883136, "grad_norm": 1.1464383602142334, "learning_rate": 1.5168000000000002e-06, "loss": 0.0051, "step": 42417 }, { "epoch": 2.261449059017967, "grad_norm": 1.1876627206802368, "learning_rate": 1.5166e-06, "loss": 0.011, "step": 42418 }, { "epoch": 2.2615023724476195, "grad_norm": 5.605227947235107, "learning_rate": 1.5164e-06, "loss": 0.0292, "step": 42419 }, { "epoch": 2.2615556858772723, "grad_norm": 1.9975978136062622, "learning_rate": 1.5162000000000002e-06, "loss": 0.0169, "step": 42420 }, { "epoch": 2.2616089993069255, "grad_norm": 2.5541274547576904, "learning_rate": 1.5160000000000002e-06, "loss": 0.0188, "step": 42421 }, { "epoch": 2.2616623127365783, "grad_norm": 2.602548360824585, "learning_rate": 1.5158e-06, "loss": 0.0256, "step": 42422 }, { "epoch": 2.2617156261662315, "grad_norm": 2.0515570640563965, "learning_rate": 1.5156000000000002e-06, "loss": 0.0233, "step": 42423 }, { "epoch": 2.261768939595884, "grad_norm": 1.5738539695739746, "learning_rate": 1.5154000000000001e-06, "loss": 0.0237, "step": 42424 }, { "epoch": 2.261822253025537, "grad_norm": 1.0614120960235596, "learning_rate": 1.5152e-06, "loss": 0.0187, "step": 42425 }, { "epoch": 2.26187556645519, "grad_norm": 1.818217158317566, "learning_rate": 1.5150000000000001e-06, "loss": 0.0183, "step": 42426 }, { "epoch": 2.261928879884843, "grad_norm": 3.5771989822387695, "learning_rate": 1.5148e-06, "loss": 0.0334, "step": 42427 }, { "epoch": 2.2619821933144957, "grad_norm": 1.3784613609313965, "learning_rate": 1.5146000000000003e-06, "loss": 0.0089, "step": 42428 }, { "epoch": 2.262035506744149, "grad_norm": 1.66776442527771, "learning_rate": 1.5144e-06, "loss": 0.0071, "step": 42429 }, { "epoch": 2.2620888201738016, "grad_norm": 1.8903323411941528, "learning_rate": 1.5142e-06, "loss": 0.0136, "step": 42430 }, { "epoch": 2.262142133603455, "grad_norm": 1.233573079109192, "learning_rate": 1.5140000000000002e-06, "loss": 0.0166, "step": 42431 }, { "epoch": 2.2621954470331076, "grad_norm": 2.292231321334839, "learning_rate": 1.5138e-06, "loss": 0.0203, "step": 42432 }, { "epoch": 2.262248760462761, "grad_norm": 1.1857192516326904, "learning_rate": 1.5136e-06, "loss": 0.022, "step": 42433 }, { "epoch": 2.2623020738924136, "grad_norm": 1.9942779541015625, "learning_rate": 1.5134000000000002e-06, "loss": 0.0215, "step": 42434 }, { "epoch": 2.2623553873220663, "grad_norm": 1.8603729009628296, "learning_rate": 1.5132000000000002e-06, "loss": 0.0269, "step": 42435 }, { "epoch": 2.2624087007517195, "grad_norm": 1.9587602615356445, "learning_rate": 1.513e-06, "loss": 0.0116, "step": 42436 }, { "epoch": 2.2624620141813723, "grad_norm": 1.5541902780532837, "learning_rate": 1.5128000000000002e-06, "loss": 0.0126, "step": 42437 }, { "epoch": 2.262515327611025, "grad_norm": 0.5452035069465637, "learning_rate": 1.5126000000000001e-06, "loss": 0.0033, "step": 42438 }, { "epoch": 2.2625686410406782, "grad_norm": 2.4634039402008057, "learning_rate": 1.5124000000000003e-06, "loss": 0.0469, "step": 42439 }, { "epoch": 2.262621954470331, "grad_norm": 0.8097689747810364, "learning_rate": 1.5122000000000001e-06, "loss": 0.0052, "step": 42440 }, { "epoch": 2.262675267899984, "grad_norm": 1.7429662942886353, "learning_rate": 1.512e-06, "loss": 0.0149, "step": 42441 }, { "epoch": 2.262728581329637, "grad_norm": 1.6519354581832886, "learning_rate": 1.5118000000000003e-06, "loss": 0.0085, "step": 42442 }, { "epoch": 2.2627818947592897, "grad_norm": 2.61609148979187, "learning_rate": 1.5116e-06, "loss": 0.025, "step": 42443 }, { "epoch": 2.262835208188943, "grad_norm": 1.4489201307296753, "learning_rate": 1.5114e-06, "loss": 0.009, "step": 42444 }, { "epoch": 2.2628885216185957, "grad_norm": 0.22263483703136444, "learning_rate": 1.5112000000000002e-06, "loss": 0.0022, "step": 42445 }, { "epoch": 2.262941835048249, "grad_norm": 1.8671339750289917, "learning_rate": 1.5110000000000002e-06, "loss": 0.011, "step": 42446 }, { "epoch": 2.2629951484779016, "grad_norm": 0.2752005457878113, "learning_rate": 1.5108e-06, "loss": 0.0025, "step": 42447 }, { "epoch": 2.2630484619075544, "grad_norm": 1.8706310987472534, "learning_rate": 1.5106000000000002e-06, "loss": 0.0194, "step": 42448 }, { "epoch": 2.2631017753372076, "grad_norm": 1.0968313217163086, "learning_rate": 1.5104000000000002e-06, "loss": 0.007, "step": 42449 }, { "epoch": 2.2631550887668603, "grad_norm": 2.6749746799468994, "learning_rate": 1.5102e-06, "loss": 0.0269, "step": 42450 }, { "epoch": 2.263208402196513, "grad_norm": 2.5887441635131836, "learning_rate": 1.5100000000000002e-06, "loss": 0.0225, "step": 42451 }, { "epoch": 2.2632617156261663, "grad_norm": 0.7996627688407898, "learning_rate": 1.5098000000000001e-06, "loss": 0.004, "step": 42452 }, { "epoch": 2.263315029055819, "grad_norm": 0.6115342974662781, "learning_rate": 1.5096000000000003e-06, "loss": 0.004, "step": 42453 }, { "epoch": 2.2633683424854723, "grad_norm": 1.8575938940048218, "learning_rate": 1.5094000000000001e-06, "loss": 0.0338, "step": 42454 }, { "epoch": 2.263421655915125, "grad_norm": 2.094212293624878, "learning_rate": 1.5092e-06, "loss": 0.0083, "step": 42455 }, { "epoch": 2.2634749693447778, "grad_norm": 0.7698478102684021, "learning_rate": 1.5090000000000003e-06, "loss": 0.0051, "step": 42456 }, { "epoch": 2.263528282774431, "grad_norm": 2.1399309635162354, "learning_rate": 1.5088e-06, "loss": 0.0329, "step": 42457 }, { "epoch": 2.2635815962040837, "grad_norm": 1.2298752069473267, "learning_rate": 1.5086e-06, "loss": 0.0072, "step": 42458 }, { "epoch": 2.263634909633737, "grad_norm": 1.136330008506775, "learning_rate": 1.5084000000000002e-06, "loss": 0.0053, "step": 42459 }, { "epoch": 2.2636882230633897, "grad_norm": 1.443495750427246, "learning_rate": 1.5082000000000002e-06, "loss": 0.0184, "step": 42460 }, { "epoch": 2.2637415364930424, "grad_norm": 2.4145615100860596, "learning_rate": 1.508e-06, "loss": 0.0164, "step": 42461 }, { "epoch": 2.2637948499226956, "grad_norm": 1.4982032775878906, "learning_rate": 1.5078000000000002e-06, "loss": 0.0161, "step": 42462 }, { "epoch": 2.2638481633523484, "grad_norm": 2.137197732925415, "learning_rate": 1.5076000000000002e-06, "loss": 0.0305, "step": 42463 }, { "epoch": 2.263901476782001, "grad_norm": 1.6286617517471313, "learning_rate": 1.5074000000000002e-06, "loss": 0.0156, "step": 42464 }, { "epoch": 2.2639547902116544, "grad_norm": 0.8709290623664856, "learning_rate": 1.5072000000000002e-06, "loss": 0.0123, "step": 42465 }, { "epoch": 2.264008103641307, "grad_norm": 2.6352150440216064, "learning_rate": 1.5070000000000001e-06, "loss": 0.0196, "step": 42466 }, { "epoch": 2.2640614170709603, "grad_norm": 1.3708990812301636, "learning_rate": 1.5068000000000001e-06, "loss": 0.0038, "step": 42467 }, { "epoch": 2.264114730500613, "grad_norm": 0.6714587211608887, "learning_rate": 1.5066000000000001e-06, "loss": 0.0037, "step": 42468 }, { "epoch": 2.2641680439302663, "grad_norm": 3.6896915435791016, "learning_rate": 1.5064e-06, "loss": 0.0136, "step": 42469 }, { "epoch": 2.264221357359919, "grad_norm": 1.5841294527053833, "learning_rate": 1.5062e-06, "loss": 0.0091, "step": 42470 }, { "epoch": 2.264274670789572, "grad_norm": 2.3568291664123535, "learning_rate": 1.5060000000000003e-06, "loss": 0.0171, "step": 42471 }, { "epoch": 2.264327984219225, "grad_norm": 2.2197105884552, "learning_rate": 1.5058e-06, "loss": 0.0321, "step": 42472 }, { "epoch": 2.2643812976488777, "grad_norm": 1.5376746654510498, "learning_rate": 1.5056e-06, "loss": 0.0114, "step": 42473 }, { "epoch": 2.2644346110785305, "grad_norm": 2.1514618396759033, "learning_rate": 1.5054000000000002e-06, "loss": 0.019, "step": 42474 }, { "epoch": 2.2644879245081837, "grad_norm": 2.683568239212036, "learning_rate": 1.5052e-06, "loss": 0.012, "step": 42475 }, { "epoch": 2.2645412379378365, "grad_norm": 1.2425237894058228, "learning_rate": 1.505e-06, "loss": 0.0051, "step": 42476 }, { "epoch": 2.2645945513674897, "grad_norm": 3.067537784576416, "learning_rate": 1.5048000000000002e-06, "loss": 0.0295, "step": 42477 }, { "epoch": 2.2646478647971424, "grad_norm": 1.0361461639404297, "learning_rate": 1.5046000000000002e-06, "loss": 0.0058, "step": 42478 }, { "epoch": 2.264701178226795, "grad_norm": 2.661848306655884, "learning_rate": 1.5044e-06, "loss": 0.0338, "step": 42479 }, { "epoch": 2.2647544916564484, "grad_norm": 1.0224370956420898, "learning_rate": 1.5042000000000001e-06, "loss": 0.0054, "step": 42480 }, { "epoch": 2.264807805086101, "grad_norm": 1.6714593172073364, "learning_rate": 1.5040000000000001e-06, "loss": 0.0125, "step": 42481 }, { "epoch": 2.2648611185157543, "grad_norm": 1.4619741439819336, "learning_rate": 1.5038000000000003e-06, "loss": 0.0326, "step": 42482 }, { "epoch": 2.264914431945407, "grad_norm": 2.204695224761963, "learning_rate": 1.5036e-06, "loss": 0.0106, "step": 42483 }, { "epoch": 2.26496774537506, "grad_norm": 0.6938298940658569, "learning_rate": 1.5034e-06, "loss": 0.0034, "step": 42484 }, { "epoch": 2.265021058804713, "grad_norm": 1.3089489936828613, "learning_rate": 1.5032000000000003e-06, "loss": 0.0157, "step": 42485 }, { "epoch": 2.265074372234366, "grad_norm": 0.5024866461753845, "learning_rate": 1.503e-06, "loss": 0.0036, "step": 42486 }, { "epoch": 2.2651276856640186, "grad_norm": 0.9893465042114258, "learning_rate": 1.5028e-06, "loss": 0.0198, "step": 42487 }, { "epoch": 2.2651809990936718, "grad_norm": 1.362903356552124, "learning_rate": 1.5026000000000002e-06, "loss": 0.0075, "step": 42488 }, { "epoch": 2.2652343125233245, "grad_norm": 1.9406359195709229, "learning_rate": 1.5024000000000002e-06, "loss": 0.0078, "step": 42489 }, { "epoch": 2.2652876259529777, "grad_norm": 1.3844811916351318, "learning_rate": 1.5022e-06, "loss": 0.0224, "step": 42490 }, { "epoch": 2.2653409393826305, "grad_norm": 2.6272711753845215, "learning_rate": 1.5020000000000002e-06, "loss": 0.0149, "step": 42491 }, { "epoch": 2.2653942528122832, "grad_norm": 1.8316352367401123, "learning_rate": 1.5018000000000002e-06, "loss": 0.0204, "step": 42492 }, { "epoch": 2.2654475662419364, "grad_norm": 0.6687511801719666, "learning_rate": 1.5016e-06, "loss": 0.0035, "step": 42493 }, { "epoch": 2.265500879671589, "grad_norm": 1.9852131605148315, "learning_rate": 1.5014000000000001e-06, "loss": 0.0073, "step": 42494 }, { "epoch": 2.2655541931012424, "grad_norm": 1.580655813217163, "learning_rate": 1.5012000000000001e-06, "loss": 0.0137, "step": 42495 }, { "epoch": 2.265607506530895, "grad_norm": 2.642543077468872, "learning_rate": 1.5010000000000003e-06, "loss": 0.0439, "step": 42496 }, { "epoch": 2.265660819960548, "grad_norm": 2.1997814178466797, "learning_rate": 1.5008e-06, "loss": 0.0179, "step": 42497 }, { "epoch": 2.265714133390201, "grad_norm": 0.8347665071487427, "learning_rate": 1.5006e-06, "loss": 0.005, "step": 42498 }, { "epoch": 2.265767446819854, "grad_norm": 2.6631956100463867, "learning_rate": 1.5004000000000003e-06, "loss": 0.0254, "step": 42499 }, { "epoch": 2.2658207602495066, "grad_norm": 1.4866125583648682, "learning_rate": 1.5002e-06, "loss": 0.0149, "step": 42500 }, { "epoch": 2.26587407367916, "grad_norm": 2.7310352325439453, "learning_rate": 1.5e-06, "loss": 0.0304, "step": 42501 }, { "epoch": 2.2659273871088126, "grad_norm": 2.4193263053894043, "learning_rate": 1.4998000000000002e-06, "loss": 0.0156, "step": 42502 }, { "epoch": 2.265980700538466, "grad_norm": 1.413541316986084, "learning_rate": 1.4996000000000002e-06, "loss": 0.0142, "step": 42503 }, { "epoch": 2.2660340139681185, "grad_norm": 1.4500906467437744, "learning_rate": 1.4994e-06, "loss": 0.0096, "step": 42504 }, { "epoch": 2.2660873273977717, "grad_norm": 1.5406712293624878, "learning_rate": 1.4992000000000002e-06, "loss": 0.009, "step": 42505 }, { "epoch": 2.2661406408274245, "grad_norm": 3.3411035537719727, "learning_rate": 1.4990000000000002e-06, "loss": 0.0199, "step": 42506 }, { "epoch": 2.2661939542570773, "grad_norm": 1.1452128887176514, "learning_rate": 1.4988000000000002e-06, "loss": 0.0046, "step": 42507 }, { "epoch": 2.2662472676867305, "grad_norm": 1.704734206199646, "learning_rate": 1.4986000000000001e-06, "loss": 0.0378, "step": 42508 }, { "epoch": 2.266300581116383, "grad_norm": 1.0672162771224976, "learning_rate": 1.4984000000000001e-06, "loss": 0.006, "step": 42509 }, { "epoch": 2.266353894546036, "grad_norm": 3.3751306533813477, "learning_rate": 1.4982000000000001e-06, "loss": 0.024, "step": 42510 }, { "epoch": 2.266407207975689, "grad_norm": 2.03218936920166, "learning_rate": 1.498e-06, "loss": 0.0092, "step": 42511 }, { "epoch": 2.266460521405342, "grad_norm": 2.0238535404205322, "learning_rate": 1.4978e-06, "loss": 0.0086, "step": 42512 }, { "epoch": 2.2665138348349947, "grad_norm": 1.33046293258667, "learning_rate": 1.4976e-06, "loss": 0.0086, "step": 42513 }, { "epoch": 2.266567148264648, "grad_norm": 2.5261166095733643, "learning_rate": 1.4974000000000003e-06, "loss": 0.0152, "step": 42514 }, { "epoch": 2.2666204616943006, "grad_norm": 0.8443194627761841, "learning_rate": 1.4972e-06, "loss": 0.0053, "step": 42515 }, { "epoch": 2.266673775123954, "grad_norm": 3.207616090774536, "learning_rate": 1.497e-06, "loss": 0.0198, "step": 42516 }, { "epoch": 2.2667270885536066, "grad_norm": 1.091308832168579, "learning_rate": 1.4968000000000002e-06, "loss": 0.0074, "step": 42517 }, { "epoch": 2.26678040198326, "grad_norm": 4.048964977264404, "learning_rate": 1.4966e-06, "loss": 0.0265, "step": 42518 }, { "epoch": 2.2668337154129126, "grad_norm": 2.764267683029175, "learning_rate": 1.4964e-06, "loss": 0.0388, "step": 42519 }, { "epoch": 2.2668870288425653, "grad_norm": 1.2588695287704468, "learning_rate": 1.4962000000000002e-06, "loss": 0.024, "step": 42520 }, { "epoch": 2.2669403422722185, "grad_norm": 2.646465301513672, "learning_rate": 1.4960000000000002e-06, "loss": 0.0122, "step": 42521 }, { "epoch": 2.2669936557018713, "grad_norm": 0.5185524225234985, "learning_rate": 1.4958e-06, "loss": 0.0026, "step": 42522 }, { "epoch": 2.267046969131524, "grad_norm": 4.6628642082214355, "learning_rate": 1.4956000000000001e-06, "loss": 0.0239, "step": 42523 }, { "epoch": 2.2671002825611772, "grad_norm": 2.655071258544922, "learning_rate": 1.4954000000000001e-06, "loss": 0.0381, "step": 42524 }, { "epoch": 2.26715359599083, "grad_norm": 1.7343987226486206, "learning_rate": 1.4951999999999999e-06, "loss": 0.0251, "step": 42525 }, { "epoch": 2.267206909420483, "grad_norm": 1.9427332878112793, "learning_rate": 1.495e-06, "loss": 0.0276, "step": 42526 }, { "epoch": 2.267260222850136, "grad_norm": 1.9744231700897217, "learning_rate": 1.4948e-06, "loss": 0.0184, "step": 42527 }, { "epoch": 2.2673135362797887, "grad_norm": 2.0749764442443848, "learning_rate": 1.4946000000000003e-06, "loss": 0.0099, "step": 42528 }, { "epoch": 2.267366849709442, "grad_norm": 2.182145595550537, "learning_rate": 1.4944e-06, "loss": 0.025, "step": 42529 }, { "epoch": 2.2674201631390947, "grad_norm": 1.3892704248428345, "learning_rate": 1.4942e-06, "loss": 0.0057, "step": 42530 }, { "epoch": 2.267473476568748, "grad_norm": 5.417149543762207, "learning_rate": 1.4940000000000002e-06, "loss": 0.0112, "step": 42531 }, { "epoch": 2.2675267899984006, "grad_norm": 4.487936973571777, "learning_rate": 1.4938000000000002e-06, "loss": 0.0138, "step": 42532 }, { "epoch": 2.2675801034280534, "grad_norm": 2.017625331878662, "learning_rate": 1.4936e-06, "loss": 0.0147, "step": 42533 }, { "epoch": 2.2676334168577066, "grad_norm": 1.5302629470825195, "learning_rate": 1.4934000000000002e-06, "loss": 0.0087, "step": 42534 }, { "epoch": 2.2676867302873593, "grad_norm": 2.025880813598633, "learning_rate": 1.4932000000000002e-06, "loss": 0.0095, "step": 42535 }, { "epoch": 2.267740043717012, "grad_norm": 2.920927047729492, "learning_rate": 1.493e-06, "loss": 0.0254, "step": 42536 }, { "epoch": 2.2677933571466653, "grad_norm": 2.420910358428955, "learning_rate": 1.4928000000000001e-06, "loss": 0.0377, "step": 42537 }, { "epoch": 2.267846670576318, "grad_norm": 2.0681917667388916, "learning_rate": 1.4926000000000001e-06, "loss": 0.0223, "step": 42538 }, { "epoch": 2.2678999840059713, "grad_norm": 0.4197160303592682, "learning_rate": 1.4924000000000003e-06, "loss": 0.0024, "step": 42539 }, { "epoch": 2.267953297435624, "grad_norm": 2.296321153640747, "learning_rate": 1.4922e-06, "loss": 0.0315, "step": 42540 }, { "epoch": 2.2680066108652768, "grad_norm": 1.7379646301269531, "learning_rate": 1.492e-06, "loss": 0.0405, "step": 42541 }, { "epoch": 2.26805992429493, "grad_norm": 2.974199056625366, "learning_rate": 1.4918000000000003e-06, "loss": 0.0421, "step": 42542 }, { "epoch": 2.2681132377245827, "grad_norm": 1.9415929317474365, "learning_rate": 1.4916e-06, "loss": 0.0229, "step": 42543 }, { "epoch": 2.268166551154236, "grad_norm": 1.1325324773788452, "learning_rate": 1.4914e-06, "loss": 0.0196, "step": 42544 }, { "epoch": 2.2682198645838887, "grad_norm": 2.5026726722717285, "learning_rate": 1.4912000000000002e-06, "loss": 0.014, "step": 42545 }, { "epoch": 2.2682731780135414, "grad_norm": 0.9174381494522095, "learning_rate": 1.4910000000000002e-06, "loss": 0.0161, "step": 42546 }, { "epoch": 2.2683264914431946, "grad_norm": 0.39840057492256165, "learning_rate": 1.4908e-06, "loss": 0.0027, "step": 42547 }, { "epoch": 2.2683798048728474, "grad_norm": 0.5880402326583862, "learning_rate": 1.4906000000000002e-06, "loss": 0.0039, "step": 42548 }, { "epoch": 2.2684331183025, "grad_norm": 0.9337099194526672, "learning_rate": 1.4904000000000002e-06, "loss": 0.0047, "step": 42549 }, { "epoch": 2.2684864317321534, "grad_norm": 3.626124620437622, "learning_rate": 1.4902000000000004e-06, "loss": 0.0344, "step": 42550 }, { "epoch": 2.268539745161806, "grad_norm": 1.0120155811309814, "learning_rate": 1.4900000000000001e-06, "loss": 0.0058, "step": 42551 }, { "epoch": 2.2685930585914593, "grad_norm": 0.71587735414505, "learning_rate": 1.4898000000000001e-06, "loss": 0.0199, "step": 42552 }, { "epoch": 2.268646372021112, "grad_norm": 1.5215544700622559, "learning_rate": 1.4896000000000003e-06, "loss": 0.0147, "step": 42553 }, { "epoch": 2.2686996854507653, "grad_norm": 2.082691192626953, "learning_rate": 1.4894e-06, "loss": 0.0182, "step": 42554 }, { "epoch": 2.268752998880418, "grad_norm": 1.529769778251648, "learning_rate": 1.4892e-06, "loss": 0.0175, "step": 42555 }, { "epoch": 2.268806312310071, "grad_norm": 0.9956609010696411, "learning_rate": 1.4890000000000003e-06, "loss": 0.0054, "step": 42556 }, { "epoch": 2.268859625739724, "grad_norm": 2.090250015258789, "learning_rate": 1.4888000000000003e-06, "loss": 0.0145, "step": 42557 }, { "epoch": 2.2689129391693768, "grad_norm": 1.3905718326568604, "learning_rate": 1.4886e-06, "loss": 0.0049, "step": 42558 }, { "epoch": 2.2689662525990295, "grad_norm": 1.3510514497756958, "learning_rate": 1.4884000000000002e-06, "loss": 0.0117, "step": 42559 }, { "epoch": 2.2690195660286827, "grad_norm": 4.587832927703857, "learning_rate": 1.4882000000000002e-06, "loss": 0.0631, "step": 42560 }, { "epoch": 2.2690728794583355, "grad_norm": 3.1976709365844727, "learning_rate": 1.488e-06, "loss": 0.0285, "step": 42561 }, { "epoch": 2.2691261928879887, "grad_norm": 1.3878093957901, "learning_rate": 1.4878000000000002e-06, "loss": 0.0079, "step": 42562 }, { "epoch": 2.2691795063176414, "grad_norm": 0.8984552025794983, "learning_rate": 1.4876000000000002e-06, "loss": 0.0036, "step": 42563 }, { "epoch": 2.269232819747294, "grad_norm": 2.6216928958892822, "learning_rate": 1.4874000000000002e-06, "loss": 0.0339, "step": 42564 }, { "epoch": 2.2692861331769474, "grad_norm": 2.4181997776031494, "learning_rate": 1.4872000000000001e-06, "loss": 0.0183, "step": 42565 }, { "epoch": 2.2693394466066, "grad_norm": 1.5514200925827026, "learning_rate": 1.4870000000000001e-06, "loss": 0.0242, "step": 42566 }, { "epoch": 2.2693927600362533, "grad_norm": 1.5410007238388062, "learning_rate": 1.4868e-06, "loss": 0.0085, "step": 42567 }, { "epoch": 2.269446073465906, "grad_norm": 1.894464135169983, "learning_rate": 1.4866e-06, "loss": 0.0205, "step": 42568 }, { "epoch": 2.269499386895559, "grad_norm": 2.5768072605133057, "learning_rate": 1.4864e-06, "loss": 0.0318, "step": 42569 }, { "epoch": 2.269552700325212, "grad_norm": 1.9771487712860107, "learning_rate": 1.4862e-06, "loss": 0.0136, "step": 42570 }, { "epoch": 2.269606013754865, "grad_norm": 1.9118491411209106, "learning_rate": 1.4860000000000003e-06, "loss": 0.0094, "step": 42571 }, { "epoch": 2.2696593271845176, "grad_norm": 4.264279842376709, "learning_rate": 1.4858e-06, "loss": 0.0382, "step": 42572 }, { "epoch": 2.2697126406141708, "grad_norm": 1.5160350799560547, "learning_rate": 1.4856e-06, "loss": 0.0223, "step": 42573 }, { "epoch": 2.2697659540438235, "grad_norm": 1.289719820022583, "learning_rate": 1.4854000000000002e-06, "loss": 0.0252, "step": 42574 }, { "epoch": 2.2698192674734767, "grad_norm": 1.5757946968078613, "learning_rate": 1.4852000000000002e-06, "loss": 0.0075, "step": 42575 }, { "epoch": 2.2698725809031295, "grad_norm": 1.2677870988845825, "learning_rate": 1.485e-06, "loss": 0.0145, "step": 42576 }, { "epoch": 2.2699258943327822, "grad_norm": 2.635870933532715, "learning_rate": 1.4848000000000002e-06, "loss": 0.0207, "step": 42577 }, { "epoch": 2.2699792077624354, "grad_norm": 2.8767762184143066, "learning_rate": 1.4846000000000002e-06, "loss": 0.0102, "step": 42578 }, { "epoch": 2.270032521192088, "grad_norm": 3.8516504764556885, "learning_rate": 1.4844e-06, "loss": 0.0163, "step": 42579 }, { "epoch": 2.2700858346217414, "grad_norm": 0.5982677340507507, "learning_rate": 1.4842000000000001e-06, "loss": 0.0033, "step": 42580 }, { "epoch": 2.270139148051394, "grad_norm": 2.851895809173584, "learning_rate": 1.4840000000000001e-06, "loss": 0.0296, "step": 42581 }, { "epoch": 2.270192461481047, "grad_norm": 3.7874488830566406, "learning_rate": 1.4838000000000003e-06, "loss": 0.016, "step": 42582 }, { "epoch": 2.2702457749107, "grad_norm": 1.147917628288269, "learning_rate": 1.4836e-06, "loss": 0.0083, "step": 42583 }, { "epoch": 2.270299088340353, "grad_norm": 1.8055410385131836, "learning_rate": 1.4834e-06, "loss": 0.0094, "step": 42584 }, { "epoch": 2.2703524017700056, "grad_norm": 2.8353636264801025, "learning_rate": 1.4832000000000003e-06, "loss": 0.0158, "step": 42585 }, { "epoch": 2.270405715199659, "grad_norm": 2.071204662322998, "learning_rate": 1.483e-06, "loss": 0.0238, "step": 42586 }, { "epoch": 2.2704590286293116, "grad_norm": 0.16815778613090515, "learning_rate": 1.4828e-06, "loss": 0.0016, "step": 42587 }, { "epoch": 2.270512342058965, "grad_norm": 3.199268341064453, "learning_rate": 1.4826000000000002e-06, "loss": 0.0146, "step": 42588 }, { "epoch": 2.2705656554886176, "grad_norm": 1.7236131429672241, "learning_rate": 1.4824000000000002e-06, "loss": 0.0054, "step": 42589 }, { "epoch": 2.2706189689182708, "grad_norm": 1.2437433004379272, "learning_rate": 1.4822e-06, "loss": 0.005, "step": 42590 }, { "epoch": 2.2706722823479235, "grad_norm": 2.5297229290008545, "learning_rate": 1.4820000000000002e-06, "loss": 0.014, "step": 42591 }, { "epoch": 2.2707255957775763, "grad_norm": 3.636580228805542, "learning_rate": 1.4818000000000002e-06, "loss": 0.0299, "step": 42592 }, { "epoch": 2.2707789092072295, "grad_norm": 0.6168013215065002, "learning_rate": 1.4816e-06, "loss": 0.0033, "step": 42593 }, { "epoch": 2.2708322226368822, "grad_norm": 2.992079019546509, "learning_rate": 1.4814000000000001e-06, "loss": 0.0107, "step": 42594 }, { "epoch": 2.270885536066535, "grad_norm": 1.7198389768600464, "learning_rate": 1.4812000000000001e-06, "loss": 0.0375, "step": 42595 }, { "epoch": 2.270938849496188, "grad_norm": 0.9581426978111267, "learning_rate": 1.4810000000000003e-06, "loss": 0.0147, "step": 42596 }, { "epoch": 2.270992162925841, "grad_norm": 2.269103527069092, "learning_rate": 1.4808e-06, "loss": 0.0157, "step": 42597 }, { "epoch": 2.271045476355494, "grad_norm": 1.2221146821975708, "learning_rate": 1.4806e-06, "loss": 0.0199, "step": 42598 }, { "epoch": 2.271098789785147, "grad_norm": 2.41292667388916, "learning_rate": 1.4804000000000003e-06, "loss": 0.0101, "step": 42599 }, { "epoch": 2.2711521032147997, "grad_norm": 3.913982391357422, "learning_rate": 1.4802000000000002e-06, "loss": 0.0294, "step": 42600 }, { "epoch": 2.271205416644453, "grad_norm": 1.335458517074585, "learning_rate": 1.48e-06, "loss": 0.0042, "step": 42601 }, { "epoch": 2.2712587300741056, "grad_norm": 2.823514461517334, "learning_rate": 1.4798000000000002e-06, "loss": 0.0133, "step": 42602 }, { "epoch": 2.271312043503759, "grad_norm": 0.3761447072029114, "learning_rate": 1.4796000000000002e-06, "loss": 0.0024, "step": 42603 }, { "epoch": 2.2713653569334116, "grad_norm": 2.006916046142578, "learning_rate": 1.4794e-06, "loss": 0.0106, "step": 42604 }, { "epoch": 2.2714186703630643, "grad_norm": 1.8919429779052734, "learning_rate": 1.4792000000000002e-06, "loss": 0.0111, "step": 42605 }, { "epoch": 2.2714719837927175, "grad_norm": 1.349056363105774, "learning_rate": 1.4790000000000002e-06, "loss": 0.019, "step": 42606 }, { "epoch": 2.2715252972223703, "grad_norm": 3.8795642852783203, "learning_rate": 1.4788000000000001e-06, "loss": 0.0136, "step": 42607 }, { "epoch": 2.271578610652023, "grad_norm": 0.6296195387840271, "learning_rate": 1.4786000000000001e-06, "loss": 0.0037, "step": 42608 }, { "epoch": 2.2716319240816762, "grad_norm": 1.7569836378097534, "learning_rate": 1.4784000000000001e-06, "loss": 0.0199, "step": 42609 }, { "epoch": 2.271685237511329, "grad_norm": 1.1538357734680176, "learning_rate": 1.4782e-06, "loss": 0.0157, "step": 42610 }, { "epoch": 2.271738550940982, "grad_norm": 1.4430181980133057, "learning_rate": 1.478e-06, "loss": 0.0053, "step": 42611 }, { "epoch": 2.271791864370635, "grad_norm": 1.9872914552688599, "learning_rate": 1.4778e-06, "loss": 0.0054, "step": 42612 }, { "epoch": 2.2718451778002877, "grad_norm": 2.831969976425171, "learning_rate": 1.4776e-06, "loss": 0.0286, "step": 42613 }, { "epoch": 2.271898491229941, "grad_norm": 1.0514460802078247, "learning_rate": 1.4774000000000002e-06, "loss": 0.009, "step": 42614 }, { "epoch": 2.2719518046595937, "grad_norm": 2.295804262161255, "learning_rate": 1.4772e-06, "loss": 0.0248, "step": 42615 }, { "epoch": 2.272005118089247, "grad_norm": 2.647519826889038, "learning_rate": 1.477e-06, "loss": 0.028, "step": 42616 }, { "epoch": 2.2720584315188996, "grad_norm": 1.274855375289917, "learning_rate": 1.4768000000000002e-06, "loss": 0.0054, "step": 42617 }, { "epoch": 2.2721117449485524, "grad_norm": 1.2896655797958374, "learning_rate": 1.4766000000000002e-06, "loss": 0.0116, "step": 42618 }, { "epoch": 2.2721650583782056, "grad_norm": 1.3112356662750244, "learning_rate": 1.4764e-06, "loss": 0.0225, "step": 42619 }, { "epoch": 2.2722183718078584, "grad_norm": 1.337545394897461, "learning_rate": 1.4762000000000002e-06, "loss": 0.0088, "step": 42620 }, { "epoch": 2.272271685237511, "grad_norm": 1.1303417682647705, "learning_rate": 1.4760000000000001e-06, "loss": 0.0292, "step": 42621 }, { "epoch": 2.2723249986671643, "grad_norm": 1.3110275268554688, "learning_rate": 1.4758e-06, "loss": 0.0037, "step": 42622 }, { "epoch": 2.272378312096817, "grad_norm": 1.011997938156128, "learning_rate": 1.4756000000000001e-06, "loss": 0.0052, "step": 42623 }, { "epoch": 2.2724316255264703, "grad_norm": 0.5326142311096191, "learning_rate": 1.4754e-06, "loss": 0.0038, "step": 42624 }, { "epoch": 2.272484938956123, "grad_norm": 1.5593897104263306, "learning_rate": 1.4752000000000003e-06, "loss": 0.0067, "step": 42625 }, { "epoch": 2.2725382523857762, "grad_norm": 1.669877052307129, "learning_rate": 1.475e-06, "loss": 0.0244, "step": 42626 }, { "epoch": 2.272591565815429, "grad_norm": 0.9388971328735352, "learning_rate": 1.4748e-06, "loss": 0.0213, "step": 42627 }, { "epoch": 2.2726448792450817, "grad_norm": 1.7242439985275269, "learning_rate": 1.4746000000000002e-06, "loss": 0.0125, "step": 42628 }, { "epoch": 2.272698192674735, "grad_norm": 1.8663934469223022, "learning_rate": 1.4744e-06, "loss": 0.0323, "step": 42629 }, { "epoch": 2.2727515061043877, "grad_norm": 0.8722928762435913, "learning_rate": 1.4742e-06, "loss": 0.0043, "step": 42630 }, { "epoch": 2.2728048195340405, "grad_norm": 2.09482741355896, "learning_rate": 1.4740000000000002e-06, "loss": 0.0075, "step": 42631 }, { "epoch": 2.2728581329636937, "grad_norm": 1.562280297279358, "learning_rate": 1.4738000000000002e-06, "loss": 0.0051, "step": 42632 }, { "epoch": 2.2729114463933464, "grad_norm": 1.1217435598373413, "learning_rate": 1.4736e-06, "loss": 0.0215, "step": 42633 }, { "epoch": 2.272964759822999, "grad_norm": 0.6776745319366455, "learning_rate": 1.4734000000000002e-06, "loss": 0.0053, "step": 42634 }, { "epoch": 2.2730180732526524, "grad_norm": 1.9310392141342163, "learning_rate": 1.4732000000000001e-06, "loss": 0.0115, "step": 42635 }, { "epoch": 2.273071386682305, "grad_norm": 0.5321447849273682, "learning_rate": 1.473e-06, "loss": 0.0031, "step": 42636 }, { "epoch": 2.2731247001119583, "grad_norm": 1.9045677185058594, "learning_rate": 1.4728000000000001e-06, "loss": 0.0164, "step": 42637 }, { "epoch": 2.273178013541611, "grad_norm": 0.17757277190685272, "learning_rate": 1.4726e-06, "loss": 0.0015, "step": 42638 }, { "epoch": 2.2732313269712643, "grad_norm": 1.4950510263442993, "learning_rate": 1.4724000000000003e-06, "loss": 0.0078, "step": 42639 }, { "epoch": 2.273284640400917, "grad_norm": 2.232867956161499, "learning_rate": 1.4722e-06, "loss": 0.0239, "step": 42640 }, { "epoch": 2.27333795383057, "grad_norm": 2.615978240966797, "learning_rate": 1.472e-06, "loss": 0.0049, "step": 42641 }, { "epoch": 2.273391267260223, "grad_norm": 1.344852328300476, "learning_rate": 1.4718000000000002e-06, "loss": 0.0094, "step": 42642 }, { "epoch": 2.2734445806898758, "grad_norm": 2.049384593963623, "learning_rate": 1.4716000000000002e-06, "loss": 0.0054, "step": 42643 }, { "epoch": 2.2734978941195285, "grad_norm": 0.5366068482398987, "learning_rate": 1.4714e-06, "loss": 0.0047, "step": 42644 }, { "epoch": 2.2735512075491817, "grad_norm": 3.565919876098633, "learning_rate": 1.4712000000000002e-06, "loss": 0.057, "step": 42645 }, { "epoch": 2.2736045209788345, "grad_norm": 4.726454734802246, "learning_rate": 1.4710000000000002e-06, "loss": 0.0738, "step": 42646 }, { "epoch": 2.2736578344084877, "grad_norm": 3.098273515701294, "learning_rate": 1.4708e-06, "loss": 0.0275, "step": 42647 }, { "epoch": 2.2737111478381404, "grad_norm": 1.8298884630203247, "learning_rate": 1.4706000000000002e-06, "loss": 0.0067, "step": 42648 }, { "epoch": 2.273764461267793, "grad_norm": 3.9746034145355225, "learning_rate": 1.4704000000000001e-06, "loss": 0.0124, "step": 42649 }, { "epoch": 2.2738177746974464, "grad_norm": 3.2359042167663574, "learning_rate": 1.4702000000000003e-06, "loss": 0.0168, "step": 42650 }, { "epoch": 2.273871088127099, "grad_norm": 1.4727784395217896, "learning_rate": 1.4700000000000001e-06, "loss": 0.0365, "step": 42651 }, { "epoch": 2.2739244015567524, "grad_norm": 4.074866771697998, "learning_rate": 1.4698e-06, "loss": 0.0138, "step": 42652 }, { "epoch": 2.273977714986405, "grad_norm": 0.8838263750076294, "learning_rate": 1.4696000000000003e-06, "loss": 0.0054, "step": 42653 }, { "epoch": 2.274031028416058, "grad_norm": 3.499938726425171, "learning_rate": 1.4694e-06, "loss": 0.035, "step": 42654 }, { "epoch": 2.274084341845711, "grad_norm": 1.275910496711731, "learning_rate": 1.4692e-06, "loss": 0.0252, "step": 42655 }, { "epoch": 2.274137655275364, "grad_norm": 3.2979047298431396, "learning_rate": 1.4690000000000003e-06, "loss": 0.0123, "step": 42656 }, { "epoch": 2.2741909687050166, "grad_norm": 0.7814009189605713, "learning_rate": 1.4688000000000002e-06, "loss": 0.0215, "step": 42657 }, { "epoch": 2.27424428213467, "grad_norm": 0.3985075354576111, "learning_rate": 1.4686e-06, "loss": 0.0025, "step": 42658 }, { "epoch": 2.2742975955643225, "grad_norm": 0.9178749918937683, "learning_rate": 1.4684000000000002e-06, "loss": 0.0052, "step": 42659 }, { "epoch": 2.2743509089939757, "grad_norm": 1.543724775314331, "learning_rate": 1.4682000000000002e-06, "loss": 0.0196, "step": 42660 }, { "epoch": 2.2744042224236285, "grad_norm": 0.8021383881568909, "learning_rate": 1.4680000000000002e-06, "loss": 0.0181, "step": 42661 }, { "epoch": 2.2744575358532813, "grad_norm": 2.8998587131500244, "learning_rate": 1.4678000000000002e-06, "loss": 0.0156, "step": 42662 }, { "epoch": 2.2745108492829345, "grad_norm": 2.0981476306915283, "learning_rate": 1.4676000000000001e-06, "loss": 0.0222, "step": 42663 }, { "epoch": 2.274564162712587, "grad_norm": 2.0230605602264404, "learning_rate": 1.4674000000000001e-06, "loss": 0.0094, "step": 42664 }, { "epoch": 2.2746174761422404, "grad_norm": 2.151905059814453, "learning_rate": 1.4672000000000001e-06, "loss": 0.0258, "step": 42665 }, { "epoch": 2.274670789571893, "grad_norm": 1.3311406373977661, "learning_rate": 1.467e-06, "loss": 0.0172, "step": 42666 }, { "epoch": 2.274724103001546, "grad_norm": 1.3176981210708618, "learning_rate": 1.4668e-06, "loss": 0.0135, "step": 42667 }, { "epoch": 2.274777416431199, "grad_norm": 1.8418387174606323, "learning_rate": 1.4666000000000003e-06, "loss": 0.0258, "step": 42668 }, { "epoch": 2.274830729860852, "grad_norm": 2.083151340484619, "learning_rate": 1.4664e-06, "loss": 0.0092, "step": 42669 }, { "epoch": 2.2748840432905046, "grad_norm": 2.556647300720215, "learning_rate": 1.4662e-06, "loss": 0.0155, "step": 42670 }, { "epoch": 2.274937356720158, "grad_norm": 2.2599761486053467, "learning_rate": 1.4660000000000002e-06, "loss": 0.032, "step": 42671 }, { "epoch": 2.2749906701498106, "grad_norm": 1.405068278312683, "learning_rate": 1.4658e-06, "loss": 0.0089, "step": 42672 }, { "epoch": 2.275043983579464, "grad_norm": 1.964606523513794, "learning_rate": 1.4656e-06, "loss": 0.0227, "step": 42673 }, { "epoch": 2.2750972970091166, "grad_norm": 3.192455291748047, "learning_rate": 1.4654000000000002e-06, "loss": 0.0279, "step": 42674 }, { "epoch": 2.2751506104387698, "grad_norm": 2.292210578918457, "learning_rate": 1.4652000000000002e-06, "loss": 0.0381, "step": 42675 }, { "epoch": 2.2752039238684225, "grad_norm": 2.833359718322754, "learning_rate": 1.465e-06, "loss": 0.0377, "step": 42676 }, { "epoch": 2.2752572372980753, "grad_norm": 2.4884543418884277, "learning_rate": 1.4648000000000001e-06, "loss": 0.0111, "step": 42677 }, { "epoch": 2.2753105507277285, "grad_norm": 2.3801236152648926, "learning_rate": 1.4646000000000001e-06, "loss": 0.012, "step": 42678 }, { "epoch": 2.2753638641573812, "grad_norm": 2.770137071609497, "learning_rate": 1.4644e-06, "loss": 0.0256, "step": 42679 }, { "epoch": 2.275417177587034, "grad_norm": 2.5826191902160645, "learning_rate": 1.4642e-06, "loss": 0.0114, "step": 42680 }, { "epoch": 2.275470491016687, "grad_norm": 2.7726144790649414, "learning_rate": 1.464e-06, "loss": 0.0236, "step": 42681 }, { "epoch": 2.27552380444634, "grad_norm": 2.891145944595337, "learning_rate": 1.4638000000000003e-06, "loss": 0.0201, "step": 42682 }, { "epoch": 2.275577117875993, "grad_norm": 2.4653706550598145, "learning_rate": 1.4636e-06, "loss": 0.0182, "step": 42683 }, { "epoch": 2.275630431305646, "grad_norm": 5.036764621734619, "learning_rate": 1.4634e-06, "loss": 0.0277, "step": 42684 }, { "epoch": 2.2756837447352987, "grad_norm": 2.273397922515869, "learning_rate": 1.4632000000000002e-06, "loss": 0.0207, "step": 42685 }, { "epoch": 2.275737058164952, "grad_norm": 0.3866940438747406, "learning_rate": 1.4630000000000002e-06, "loss": 0.0034, "step": 42686 }, { "epoch": 2.2757903715946046, "grad_norm": 5.133682727813721, "learning_rate": 1.4628e-06, "loss": 0.0556, "step": 42687 }, { "epoch": 2.275843685024258, "grad_norm": 2.081390142440796, "learning_rate": 1.4626000000000002e-06, "loss": 0.0147, "step": 42688 }, { "epoch": 2.2758969984539106, "grad_norm": 1.496201992034912, "learning_rate": 1.4624000000000002e-06, "loss": 0.0079, "step": 42689 }, { "epoch": 2.2759503118835633, "grad_norm": 1.3889312744140625, "learning_rate": 1.4622e-06, "loss": 0.0105, "step": 42690 }, { "epoch": 2.2760036253132165, "grad_norm": 2.5869905948638916, "learning_rate": 1.4620000000000001e-06, "loss": 0.0297, "step": 42691 }, { "epoch": 2.2760569387428693, "grad_norm": 3.3947958946228027, "learning_rate": 1.4618000000000001e-06, "loss": 0.0186, "step": 42692 }, { "epoch": 2.276110252172522, "grad_norm": 2.3770647048950195, "learning_rate": 1.4616000000000003e-06, "loss": 0.01, "step": 42693 }, { "epoch": 2.2761635656021753, "grad_norm": 2.7503178119659424, "learning_rate": 1.4614e-06, "loss": 0.0269, "step": 42694 }, { "epoch": 2.276216879031828, "grad_norm": 1.5360969305038452, "learning_rate": 1.4612e-06, "loss": 0.013, "step": 42695 }, { "epoch": 2.276270192461481, "grad_norm": 1.964253306388855, "learning_rate": 1.4610000000000003e-06, "loss": 0.0093, "step": 42696 }, { "epoch": 2.276323505891134, "grad_norm": 2.6744964122772217, "learning_rate": 1.4608e-06, "loss": 0.0311, "step": 42697 }, { "epoch": 2.2763768193207867, "grad_norm": 3.061164140701294, "learning_rate": 1.4606e-06, "loss": 0.0387, "step": 42698 }, { "epoch": 2.27643013275044, "grad_norm": 1.3902581930160522, "learning_rate": 1.4604000000000002e-06, "loss": 0.0131, "step": 42699 }, { "epoch": 2.2764834461800927, "grad_norm": 1.9587489366531372, "learning_rate": 1.4602000000000002e-06, "loss": 0.0084, "step": 42700 }, { "epoch": 2.276536759609746, "grad_norm": 2.142143964767456, "learning_rate": 1.46e-06, "loss": 0.0074, "step": 42701 }, { "epoch": 2.2765900730393986, "grad_norm": 0.9549461603164673, "learning_rate": 1.4598000000000002e-06, "loss": 0.0137, "step": 42702 }, { "epoch": 2.2766433864690514, "grad_norm": 0.250085711479187, "learning_rate": 1.4596000000000002e-06, "loss": 0.0019, "step": 42703 }, { "epoch": 2.2766966998987046, "grad_norm": 1.3212676048278809, "learning_rate": 1.4594e-06, "loss": 0.0091, "step": 42704 }, { "epoch": 2.2767500133283574, "grad_norm": 2.0742485523223877, "learning_rate": 1.4592000000000001e-06, "loss": 0.0273, "step": 42705 }, { "epoch": 2.27680332675801, "grad_norm": 2.389962911605835, "learning_rate": 1.4590000000000001e-06, "loss": 0.0114, "step": 42706 }, { "epoch": 2.2768566401876633, "grad_norm": 0.7081251740455627, "learning_rate": 1.4588000000000001e-06, "loss": 0.0046, "step": 42707 }, { "epoch": 2.276909953617316, "grad_norm": 3.4742820262908936, "learning_rate": 1.4586e-06, "loss": 0.0509, "step": 42708 }, { "epoch": 2.2769632670469693, "grad_norm": 1.6859183311462402, "learning_rate": 1.4584e-06, "loss": 0.012, "step": 42709 }, { "epoch": 2.277016580476622, "grad_norm": 1.006985068321228, "learning_rate": 1.4582e-06, "loss": 0.0068, "step": 42710 }, { "epoch": 2.2770698939062752, "grad_norm": 1.1424208879470825, "learning_rate": 1.4580000000000003e-06, "loss": 0.0164, "step": 42711 }, { "epoch": 2.277123207335928, "grad_norm": 1.2526413202285767, "learning_rate": 1.4578e-06, "loss": 0.017, "step": 42712 }, { "epoch": 2.2771765207655807, "grad_norm": 0.5939788818359375, "learning_rate": 1.4576e-06, "loss": 0.0037, "step": 42713 }, { "epoch": 2.277229834195234, "grad_norm": 2.6866304874420166, "learning_rate": 1.4574000000000002e-06, "loss": 0.0366, "step": 42714 }, { "epoch": 2.2772831476248867, "grad_norm": 1.2612009048461914, "learning_rate": 1.4572e-06, "loss": 0.0266, "step": 42715 }, { "epoch": 2.2773364610545395, "grad_norm": 0.17960716784000397, "learning_rate": 1.457e-06, "loss": 0.0017, "step": 42716 }, { "epoch": 2.2773897744841927, "grad_norm": 0.9457234144210815, "learning_rate": 1.4568000000000002e-06, "loss": 0.0083, "step": 42717 }, { "epoch": 2.2774430879138454, "grad_norm": 1.892581582069397, "learning_rate": 1.4566000000000002e-06, "loss": 0.0206, "step": 42718 }, { "epoch": 2.277496401343498, "grad_norm": 0.41972553730010986, "learning_rate": 1.4564e-06, "loss": 0.0028, "step": 42719 }, { "epoch": 2.2775497147731514, "grad_norm": 2.5494489669799805, "learning_rate": 1.4562000000000001e-06, "loss": 0.0094, "step": 42720 }, { "epoch": 2.277603028202804, "grad_norm": 1.7596774101257324, "learning_rate": 1.4560000000000001e-06, "loss": 0.0119, "step": 42721 }, { "epoch": 2.2776563416324573, "grad_norm": 2.4870433807373047, "learning_rate": 1.4557999999999999e-06, "loss": 0.0247, "step": 42722 }, { "epoch": 2.27770965506211, "grad_norm": 1.4802252054214478, "learning_rate": 1.4556e-06, "loss": 0.0094, "step": 42723 }, { "epoch": 2.2777629684917633, "grad_norm": 1.0628111362457275, "learning_rate": 1.4554e-06, "loss": 0.005, "step": 42724 }, { "epoch": 2.277816281921416, "grad_norm": 2.6932830810546875, "learning_rate": 1.4552000000000003e-06, "loss": 0.0344, "step": 42725 }, { "epoch": 2.277869595351069, "grad_norm": 1.762463092803955, "learning_rate": 1.455e-06, "loss": 0.0142, "step": 42726 }, { "epoch": 2.277922908780722, "grad_norm": 2.1967568397521973, "learning_rate": 1.4548e-06, "loss": 0.0188, "step": 42727 }, { "epoch": 2.2779762222103748, "grad_norm": 4.378480434417725, "learning_rate": 1.4546000000000002e-06, "loss": 0.0284, "step": 42728 }, { "epoch": 2.2780295356400275, "grad_norm": 2.2794384956359863, "learning_rate": 1.4544000000000002e-06, "loss": 0.0218, "step": 42729 }, { "epoch": 2.2780828490696807, "grad_norm": 1.663376808166504, "learning_rate": 1.4542e-06, "loss": 0.007, "step": 42730 }, { "epoch": 2.2781361624993335, "grad_norm": 1.5409765243530273, "learning_rate": 1.4540000000000002e-06, "loss": 0.0106, "step": 42731 }, { "epoch": 2.2781894759289867, "grad_norm": 0.7717739939689636, "learning_rate": 1.4538000000000002e-06, "loss": 0.0111, "step": 42732 }, { "epoch": 2.2782427893586394, "grad_norm": 2.589798927307129, "learning_rate": 1.4536e-06, "loss": 0.0251, "step": 42733 }, { "epoch": 2.278296102788292, "grad_norm": 1.2798422574996948, "learning_rate": 1.4534000000000001e-06, "loss": 0.0083, "step": 42734 }, { "epoch": 2.2783494162179454, "grad_norm": 1.1392728090286255, "learning_rate": 1.4532000000000001e-06, "loss": 0.0057, "step": 42735 }, { "epoch": 2.278402729647598, "grad_norm": 1.0625596046447754, "learning_rate": 1.4530000000000003e-06, "loss": 0.0137, "step": 42736 }, { "epoch": 2.2784560430772514, "grad_norm": 1.8599436283111572, "learning_rate": 1.4528e-06, "loss": 0.0558, "step": 42737 }, { "epoch": 2.278509356506904, "grad_norm": 1.2981070280075073, "learning_rate": 1.4526e-06, "loss": 0.0111, "step": 42738 }, { "epoch": 2.278562669936557, "grad_norm": 2.44206166267395, "learning_rate": 1.4524000000000003e-06, "loss": 0.0657, "step": 42739 }, { "epoch": 2.27861598336621, "grad_norm": 1.337857961654663, "learning_rate": 1.4522e-06, "loss": 0.0137, "step": 42740 }, { "epoch": 2.278669296795863, "grad_norm": 1.2105188369750977, "learning_rate": 1.452e-06, "loss": 0.0064, "step": 42741 }, { "epoch": 2.2787226102255156, "grad_norm": 2.9844894409179688, "learning_rate": 1.4518000000000002e-06, "loss": 0.0289, "step": 42742 }, { "epoch": 2.278775923655169, "grad_norm": 3.009120225906372, "learning_rate": 1.4516000000000002e-06, "loss": 0.011, "step": 42743 }, { "epoch": 2.2788292370848215, "grad_norm": 7.957141399383545, "learning_rate": 1.4514e-06, "loss": 0.0249, "step": 42744 }, { "epoch": 2.2788825505144747, "grad_norm": 2.619657039642334, "learning_rate": 1.4512000000000002e-06, "loss": 0.007, "step": 42745 }, { "epoch": 2.2789358639441275, "grad_norm": 2.429124116897583, "learning_rate": 1.4510000000000002e-06, "loss": 0.0128, "step": 42746 }, { "epoch": 2.2789891773737803, "grad_norm": 1.2793619632720947, "learning_rate": 1.4508e-06, "loss": 0.0089, "step": 42747 }, { "epoch": 2.2790424908034335, "grad_norm": 0.5550663471221924, "learning_rate": 1.4506000000000001e-06, "loss": 0.0029, "step": 42748 }, { "epoch": 2.279095804233086, "grad_norm": 1.497056484222412, "learning_rate": 1.4504000000000001e-06, "loss": 0.0057, "step": 42749 }, { "epoch": 2.2791491176627394, "grad_norm": 1.4889702796936035, "learning_rate": 1.4502000000000003e-06, "loss": 0.0083, "step": 42750 }, { "epoch": 2.279202431092392, "grad_norm": 2.5674097537994385, "learning_rate": 1.45e-06, "loss": 0.0087, "step": 42751 }, { "epoch": 2.279255744522045, "grad_norm": 1.621248722076416, "learning_rate": 1.4498e-06, "loss": 0.015, "step": 42752 }, { "epoch": 2.279309057951698, "grad_norm": 3.517608165740967, "learning_rate": 1.4496000000000003e-06, "loss": 0.0194, "step": 42753 }, { "epoch": 2.279362371381351, "grad_norm": 2.6489477157592773, "learning_rate": 1.4494000000000003e-06, "loss": 0.0125, "step": 42754 }, { "epoch": 2.2794156848110037, "grad_norm": 1.435071587562561, "learning_rate": 1.4492e-06, "loss": 0.0229, "step": 42755 }, { "epoch": 2.279468998240657, "grad_norm": 2.1629092693328857, "learning_rate": 1.4490000000000002e-06, "loss": 0.0072, "step": 42756 }, { "epoch": 2.2795223116703096, "grad_norm": 0.15244260430335999, "learning_rate": 1.4488000000000002e-06, "loss": 0.0017, "step": 42757 }, { "epoch": 2.279575625099963, "grad_norm": 2.4425203800201416, "learning_rate": 1.4486e-06, "loss": 0.0212, "step": 42758 }, { "epoch": 2.2796289385296156, "grad_norm": 0.3015570342540741, "learning_rate": 1.4484000000000002e-06, "loss": 0.0026, "step": 42759 }, { "epoch": 2.2796822519592688, "grad_norm": 1.8004575967788696, "learning_rate": 1.4482000000000002e-06, "loss": 0.0202, "step": 42760 }, { "epoch": 2.2797355653889215, "grad_norm": 1.4995607137680054, "learning_rate": 1.4480000000000002e-06, "loss": 0.0347, "step": 42761 }, { "epoch": 2.2797888788185743, "grad_norm": 1.535516381263733, "learning_rate": 1.4478000000000001e-06, "loss": 0.0082, "step": 42762 }, { "epoch": 2.2798421922482275, "grad_norm": 2.9495060443878174, "learning_rate": 1.4476000000000001e-06, "loss": 0.0258, "step": 42763 }, { "epoch": 2.2798955056778802, "grad_norm": 1.2386642694473267, "learning_rate": 1.4474000000000001e-06, "loss": 0.0044, "step": 42764 }, { "epoch": 2.279948819107533, "grad_norm": 2.0365428924560547, "learning_rate": 1.4472e-06, "loss": 0.0133, "step": 42765 }, { "epoch": 2.280002132537186, "grad_norm": 3.5178260803222656, "learning_rate": 1.447e-06, "loss": 0.0259, "step": 42766 }, { "epoch": 2.280055445966839, "grad_norm": 3.0502305030822754, "learning_rate": 1.4468e-06, "loss": 0.0309, "step": 42767 }, { "epoch": 2.280108759396492, "grad_norm": 2.7135114669799805, "learning_rate": 1.4466000000000003e-06, "loss": 0.0165, "step": 42768 }, { "epoch": 2.280162072826145, "grad_norm": 2.588900089263916, "learning_rate": 1.4464e-06, "loss": 0.0164, "step": 42769 }, { "epoch": 2.2802153862557977, "grad_norm": 2.3847928047180176, "learning_rate": 1.4462e-06, "loss": 0.0276, "step": 42770 }, { "epoch": 2.280268699685451, "grad_norm": 0.48716622591018677, "learning_rate": 1.4460000000000002e-06, "loss": 0.0043, "step": 42771 }, { "epoch": 2.2803220131151036, "grad_norm": 1.2109776735305786, "learning_rate": 1.4458e-06, "loss": 0.0173, "step": 42772 }, { "epoch": 2.280375326544757, "grad_norm": 0.39734625816345215, "learning_rate": 1.4456e-06, "loss": 0.0031, "step": 42773 }, { "epoch": 2.2804286399744096, "grad_norm": 0.1551520973443985, "learning_rate": 1.4454000000000002e-06, "loss": 0.0013, "step": 42774 }, { "epoch": 2.2804819534040623, "grad_norm": 0.6321040391921997, "learning_rate": 1.4452000000000002e-06, "loss": 0.0047, "step": 42775 }, { "epoch": 2.2805352668337155, "grad_norm": 3.0073721408843994, "learning_rate": 1.445e-06, "loss": 0.0324, "step": 42776 }, { "epoch": 2.2805885802633683, "grad_norm": 2.25113582611084, "learning_rate": 1.4448000000000001e-06, "loss": 0.0076, "step": 42777 }, { "epoch": 2.280641893693021, "grad_norm": 0.942946195602417, "learning_rate": 1.4446000000000001e-06, "loss": 0.0162, "step": 42778 }, { "epoch": 2.2806952071226743, "grad_norm": 1.3649797439575195, "learning_rate": 1.4444000000000003e-06, "loss": 0.0197, "step": 42779 }, { "epoch": 2.280748520552327, "grad_norm": 0.742684543132782, "learning_rate": 1.4442e-06, "loss": 0.0051, "step": 42780 }, { "epoch": 2.28080183398198, "grad_norm": 3.4785549640655518, "learning_rate": 1.444e-06, "loss": 0.0244, "step": 42781 }, { "epoch": 2.280855147411633, "grad_norm": 1.7394568920135498, "learning_rate": 1.4438000000000003e-06, "loss": 0.0204, "step": 42782 }, { "epoch": 2.2809084608412857, "grad_norm": 1.957727074623108, "learning_rate": 1.4436e-06, "loss": 0.026, "step": 42783 }, { "epoch": 2.280961774270939, "grad_norm": 2.4125802516937256, "learning_rate": 1.4434e-06, "loss": 0.0192, "step": 42784 }, { "epoch": 2.2810150877005917, "grad_norm": 1.5283738374710083, "learning_rate": 1.4432000000000002e-06, "loss": 0.0312, "step": 42785 }, { "epoch": 2.281068401130245, "grad_norm": 1.9448587894439697, "learning_rate": 1.4430000000000002e-06, "loss": 0.0408, "step": 42786 }, { "epoch": 2.2811217145598977, "grad_norm": 2.2629621028900146, "learning_rate": 1.4428e-06, "loss": 0.0092, "step": 42787 }, { "epoch": 2.2811750279895504, "grad_norm": 3.571798801422119, "learning_rate": 1.4426000000000002e-06, "loss": 0.0335, "step": 42788 }, { "epoch": 2.2812283414192036, "grad_norm": 2.6834683418273926, "learning_rate": 1.4424000000000002e-06, "loss": 0.0127, "step": 42789 }, { "epoch": 2.2812816548488564, "grad_norm": 1.1812431812286377, "learning_rate": 1.4422e-06, "loss": 0.007, "step": 42790 }, { "epoch": 2.281334968278509, "grad_norm": 0.14895984530448914, "learning_rate": 1.4420000000000001e-06, "loss": 0.001, "step": 42791 }, { "epoch": 2.2813882817081623, "grad_norm": 2.1684625148773193, "learning_rate": 1.4418000000000001e-06, "loss": 0.0286, "step": 42792 }, { "epoch": 2.281441595137815, "grad_norm": 0.8282457590103149, "learning_rate": 1.4416000000000003e-06, "loss": 0.0053, "step": 42793 }, { "epoch": 2.2814949085674683, "grad_norm": 1.2647625207901, "learning_rate": 1.4414e-06, "loss": 0.0155, "step": 42794 }, { "epoch": 2.281548221997121, "grad_norm": 1.506118655204773, "learning_rate": 1.4412e-06, "loss": 0.0097, "step": 42795 }, { "epoch": 2.2816015354267742, "grad_norm": 1.0601719617843628, "learning_rate": 1.4410000000000003e-06, "loss": 0.0049, "step": 42796 }, { "epoch": 2.281654848856427, "grad_norm": 1.7815072536468506, "learning_rate": 1.4408000000000002e-06, "loss": 0.0093, "step": 42797 }, { "epoch": 2.2817081622860798, "grad_norm": 2.4060146808624268, "learning_rate": 1.4406e-06, "loss": 0.0117, "step": 42798 }, { "epoch": 2.281761475715733, "grad_norm": 1.5680804252624512, "learning_rate": 1.4404000000000002e-06, "loss": 0.0203, "step": 42799 }, { "epoch": 2.2818147891453857, "grad_norm": 3.5659618377685547, "learning_rate": 1.4402000000000002e-06, "loss": 0.0264, "step": 42800 }, { "epoch": 2.2818681025750385, "grad_norm": 0.6569046378135681, "learning_rate": 1.44e-06, "loss": 0.0042, "step": 42801 }, { "epoch": 2.2819214160046917, "grad_norm": 2.4283831119537354, "learning_rate": 1.4398000000000002e-06, "loss": 0.03, "step": 42802 }, { "epoch": 2.2819747294343444, "grad_norm": 1.611729621887207, "learning_rate": 1.4396000000000002e-06, "loss": 0.0056, "step": 42803 }, { "epoch": 2.2820280428639976, "grad_norm": 1.3724117279052734, "learning_rate": 1.4394000000000001e-06, "loss": 0.0143, "step": 42804 }, { "epoch": 2.2820813562936504, "grad_norm": 3.6246063709259033, "learning_rate": 1.4392000000000001e-06, "loss": 0.0194, "step": 42805 }, { "epoch": 2.282134669723303, "grad_norm": 1.0043550729751587, "learning_rate": 1.4390000000000001e-06, "loss": 0.0128, "step": 42806 }, { "epoch": 2.2821879831529563, "grad_norm": 2.7206051349639893, "learning_rate": 1.4388e-06, "loss": 0.0317, "step": 42807 }, { "epoch": 2.282241296582609, "grad_norm": 3.436638355255127, "learning_rate": 1.4386e-06, "loss": 0.01, "step": 42808 }, { "epoch": 2.2822946100122623, "grad_norm": 1.527808666229248, "learning_rate": 1.4384e-06, "loss": 0.0083, "step": 42809 }, { "epoch": 2.282347923441915, "grad_norm": 2.6886627674102783, "learning_rate": 1.4382e-06, "loss": 0.0089, "step": 42810 }, { "epoch": 2.282401236871568, "grad_norm": 2.8874619007110596, "learning_rate": 1.4380000000000003e-06, "loss": 0.0107, "step": 42811 }, { "epoch": 2.282454550301221, "grad_norm": 2.9452688694000244, "learning_rate": 1.4378e-06, "loss": 0.0427, "step": 42812 }, { "epoch": 2.2825078637308738, "grad_norm": 1.3312126398086548, "learning_rate": 1.4376e-06, "loss": 0.0095, "step": 42813 }, { "epoch": 2.2825611771605265, "grad_norm": 4.290884971618652, "learning_rate": 1.4374000000000002e-06, "loss": 0.036, "step": 42814 }, { "epoch": 2.2826144905901797, "grad_norm": 3.248358726501465, "learning_rate": 1.4372e-06, "loss": 0.0421, "step": 42815 }, { "epoch": 2.2826678040198325, "grad_norm": 1.0494651794433594, "learning_rate": 1.437e-06, "loss": 0.0044, "step": 42816 }, { "epoch": 2.2827211174494857, "grad_norm": 1.595193862915039, "learning_rate": 1.4368000000000002e-06, "loss": 0.0131, "step": 42817 }, { "epoch": 2.2827744308791384, "grad_norm": 2.2211873531341553, "learning_rate": 1.4366000000000001e-06, "loss": 0.0166, "step": 42818 }, { "epoch": 2.282827744308791, "grad_norm": 2.8737130165100098, "learning_rate": 1.4364e-06, "loss": 0.0133, "step": 42819 }, { "epoch": 2.2828810577384444, "grad_norm": 1.5319435596466064, "learning_rate": 1.4362000000000001e-06, "loss": 0.0071, "step": 42820 }, { "epoch": 2.282934371168097, "grad_norm": 1.6017721891403198, "learning_rate": 1.436e-06, "loss": 0.026, "step": 42821 }, { "epoch": 2.2829876845977504, "grad_norm": 1.2956465482711792, "learning_rate": 1.4358000000000003e-06, "loss": 0.0062, "step": 42822 }, { "epoch": 2.283040998027403, "grad_norm": 1.3522733449935913, "learning_rate": 1.4356e-06, "loss": 0.0106, "step": 42823 }, { "epoch": 2.283094311457056, "grad_norm": 0.7457008361816406, "learning_rate": 1.4354e-06, "loss": 0.0108, "step": 42824 }, { "epoch": 2.283147624886709, "grad_norm": 1.8196009397506714, "learning_rate": 1.4352000000000003e-06, "loss": 0.0212, "step": 42825 }, { "epoch": 2.283200938316362, "grad_norm": 1.1659162044525146, "learning_rate": 1.435e-06, "loss": 0.0067, "step": 42826 }, { "epoch": 2.2832542517460146, "grad_norm": 3.1466989517211914, "learning_rate": 1.4348e-06, "loss": 0.0094, "step": 42827 }, { "epoch": 2.283307565175668, "grad_norm": 1.3460745811462402, "learning_rate": 1.4346000000000002e-06, "loss": 0.0082, "step": 42828 }, { "epoch": 2.2833608786053206, "grad_norm": 3.104135513305664, "learning_rate": 1.4344000000000002e-06, "loss": 0.0474, "step": 42829 }, { "epoch": 2.2834141920349738, "grad_norm": 1.2110161781311035, "learning_rate": 1.4342e-06, "loss": 0.0237, "step": 42830 }, { "epoch": 2.2834675054646265, "grad_norm": 1.3428564071655273, "learning_rate": 1.4340000000000002e-06, "loss": 0.0143, "step": 42831 }, { "epoch": 2.2835208188942797, "grad_norm": 2.1262195110321045, "learning_rate": 1.4338000000000001e-06, "loss": 0.0136, "step": 42832 }, { "epoch": 2.2835741323239325, "grad_norm": 1.3099379539489746, "learning_rate": 1.4336e-06, "loss": 0.0133, "step": 42833 }, { "epoch": 2.2836274457535852, "grad_norm": 3.169288158416748, "learning_rate": 1.4334000000000001e-06, "loss": 0.0294, "step": 42834 }, { "epoch": 2.2836807591832384, "grad_norm": 1.9539191722869873, "learning_rate": 1.4332e-06, "loss": 0.0151, "step": 42835 }, { "epoch": 2.283734072612891, "grad_norm": 2.832972526550293, "learning_rate": 1.4330000000000003e-06, "loss": 0.0172, "step": 42836 }, { "epoch": 2.283787386042544, "grad_norm": 2.0264322757720947, "learning_rate": 1.4328e-06, "loss": 0.0291, "step": 42837 }, { "epoch": 2.283840699472197, "grad_norm": 0.7161661386489868, "learning_rate": 1.4326e-06, "loss": 0.0046, "step": 42838 }, { "epoch": 2.28389401290185, "grad_norm": 1.356386423110962, "learning_rate": 1.4324000000000003e-06, "loss": 0.0209, "step": 42839 }, { "epoch": 2.2839473263315027, "grad_norm": 1.6639524698257446, "learning_rate": 1.4322e-06, "loss": 0.0304, "step": 42840 }, { "epoch": 2.284000639761156, "grad_norm": 1.4780421257019043, "learning_rate": 1.432e-06, "loss": 0.0096, "step": 42841 }, { "epoch": 2.2840539531908086, "grad_norm": 1.2200496196746826, "learning_rate": 1.4318000000000002e-06, "loss": 0.0279, "step": 42842 }, { "epoch": 2.284107266620462, "grad_norm": 2.206942081451416, "learning_rate": 1.4316000000000002e-06, "loss": 0.022, "step": 42843 }, { "epoch": 2.2841605800501146, "grad_norm": 1.2598215341567993, "learning_rate": 1.4314e-06, "loss": 0.0071, "step": 42844 }, { "epoch": 2.2842138934797678, "grad_norm": 1.1280165910720825, "learning_rate": 1.4312000000000002e-06, "loss": 0.0034, "step": 42845 }, { "epoch": 2.2842672069094205, "grad_norm": 1.443692684173584, "learning_rate": 1.4310000000000001e-06, "loss": 0.0073, "step": 42846 }, { "epoch": 2.2843205203390733, "grad_norm": 2.5008082389831543, "learning_rate": 1.4308000000000003e-06, "loss": 0.0274, "step": 42847 }, { "epoch": 2.2843738337687265, "grad_norm": 2.4657375812530518, "learning_rate": 1.4306000000000001e-06, "loss": 0.0287, "step": 42848 }, { "epoch": 2.2844271471983792, "grad_norm": 1.5435678958892822, "learning_rate": 1.4304e-06, "loss": 0.0241, "step": 42849 }, { "epoch": 2.284480460628032, "grad_norm": 3.2162744998931885, "learning_rate": 1.4302000000000003e-06, "loss": 0.0232, "step": 42850 }, { "epoch": 2.284533774057685, "grad_norm": 5.003544330596924, "learning_rate": 1.43e-06, "loss": 0.0277, "step": 42851 }, { "epoch": 2.284587087487338, "grad_norm": 2.163255453109741, "learning_rate": 1.4298e-06, "loss": 0.0323, "step": 42852 }, { "epoch": 2.284640400916991, "grad_norm": 3.7777130603790283, "learning_rate": 1.4296000000000003e-06, "loss": 0.0183, "step": 42853 }, { "epoch": 2.284693714346644, "grad_norm": 1.6397770643234253, "learning_rate": 1.4294000000000002e-06, "loss": 0.0378, "step": 42854 }, { "epoch": 2.2847470277762967, "grad_norm": 2.1975953578948975, "learning_rate": 1.4292e-06, "loss": 0.0347, "step": 42855 }, { "epoch": 2.28480034120595, "grad_norm": 0.9894364476203918, "learning_rate": 1.4290000000000002e-06, "loss": 0.0089, "step": 42856 }, { "epoch": 2.2848536546356026, "grad_norm": 2.172854423522949, "learning_rate": 1.4288000000000002e-06, "loss": 0.0141, "step": 42857 }, { "epoch": 2.284906968065256, "grad_norm": 1.3728001117706299, "learning_rate": 1.4286e-06, "loss": 0.0044, "step": 42858 }, { "epoch": 2.2849602814949086, "grad_norm": 2.719336748123169, "learning_rate": 1.4284000000000002e-06, "loss": 0.0068, "step": 42859 }, { "epoch": 2.2850135949245614, "grad_norm": 2.432077407836914, "learning_rate": 1.4282000000000001e-06, "loss": 0.0175, "step": 42860 }, { "epoch": 2.2850669083542146, "grad_norm": 1.9195705652236938, "learning_rate": 1.4280000000000001e-06, "loss": 0.0193, "step": 42861 }, { "epoch": 2.2851202217838673, "grad_norm": 0.295282244682312, "learning_rate": 1.4278000000000001e-06, "loss": 0.0023, "step": 42862 }, { "epoch": 2.28517353521352, "grad_norm": 3.8844926357269287, "learning_rate": 1.4276e-06, "loss": 0.0344, "step": 42863 }, { "epoch": 2.2852268486431733, "grad_norm": 1.3771979808807373, "learning_rate": 1.4274e-06, "loss": 0.0067, "step": 42864 }, { "epoch": 2.285280162072826, "grad_norm": 1.568978190422058, "learning_rate": 1.4272000000000003e-06, "loss": 0.0059, "step": 42865 }, { "epoch": 2.2853334755024792, "grad_norm": 1.929975986480713, "learning_rate": 1.427e-06, "loss": 0.02, "step": 42866 }, { "epoch": 2.285386788932132, "grad_norm": 2.067117929458618, "learning_rate": 1.4268e-06, "loss": 0.0047, "step": 42867 }, { "epoch": 2.2854401023617847, "grad_norm": 1.948457956314087, "learning_rate": 1.4266000000000002e-06, "loss": 0.0307, "step": 42868 }, { "epoch": 2.285493415791438, "grad_norm": 2.5013527870178223, "learning_rate": 1.4264e-06, "loss": 0.0112, "step": 42869 }, { "epoch": 2.2855467292210907, "grad_norm": 1.7126779556274414, "learning_rate": 1.4262e-06, "loss": 0.0161, "step": 42870 }, { "epoch": 2.285600042650744, "grad_norm": 1.8746007680892944, "learning_rate": 1.4260000000000002e-06, "loss": 0.0257, "step": 42871 }, { "epoch": 2.2856533560803967, "grad_norm": 0.850609302520752, "learning_rate": 1.4258000000000002e-06, "loss": 0.0047, "step": 42872 }, { "epoch": 2.2857066695100494, "grad_norm": 1.2064489126205444, "learning_rate": 1.4256e-06, "loss": 0.0093, "step": 42873 }, { "epoch": 2.2857599829397026, "grad_norm": 1.5909777879714966, "learning_rate": 1.4254000000000002e-06, "loss": 0.0142, "step": 42874 }, { "epoch": 2.2858132963693554, "grad_norm": 2.175797700881958, "learning_rate": 1.4252000000000001e-06, "loss": 0.0231, "step": 42875 }, { "epoch": 2.285866609799008, "grad_norm": 2.5534815788269043, "learning_rate": 1.425e-06, "loss": 0.035, "step": 42876 }, { "epoch": 2.2859199232286613, "grad_norm": 3.33400297164917, "learning_rate": 1.4248e-06, "loss": 0.0184, "step": 42877 }, { "epoch": 2.285973236658314, "grad_norm": 0.8762684464454651, "learning_rate": 1.4246e-06, "loss": 0.006, "step": 42878 }, { "epoch": 2.2860265500879673, "grad_norm": 2.6868274211883545, "learning_rate": 1.4244000000000003e-06, "loss": 0.0135, "step": 42879 }, { "epoch": 2.28607986351762, "grad_norm": 0.9628812074661255, "learning_rate": 1.4242e-06, "loss": 0.0142, "step": 42880 }, { "epoch": 2.2861331769472732, "grad_norm": 0.6676083207130432, "learning_rate": 1.424e-06, "loss": 0.0055, "step": 42881 }, { "epoch": 2.286186490376926, "grad_norm": 1.0899667739868164, "learning_rate": 1.4238000000000002e-06, "loss": 0.0043, "step": 42882 }, { "epoch": 2.2862398038065788, "grad_norm": 3.450998067855835, "learning_rate": 1.4236e-06, "loss": 0.0153, "step": 42883 }, { "epoch": 2.286293117236232, "grad_norm": 0.3053646385669708, "learning_rate": 1.4234e-06, "loss": 0.0026, "step": 42884 }, { "epoch": 2.2863464306658847, "grad_norm": 1.894066572189331, "learning_rate": 1.4232000000000002e-06, "loss": 0.034, "step": 42885 }, { "epoch": 2.2863997440955375, "grad_norm": 1.7706555128097534, "learning_rate": 1.4230000000000002e-06, "loss": 0.0244, "step": 42886 }, { "epoch": 2.2864530575251907, "grad_norm": 0.6444425582885742, "learning_rate": 1.4228e-06, "loss": 0.0045, "step": 42887 }, { "epoch": 2.2865063709548434, "grad_norm": 0.4826182425022125, "learning_rate": 1.4226000000000002e-06, "loss": 0.0035, "step": 42888 }, { "epoch": 2.2865596843844966, "grad_norm": 3.796572685241699, "learning_rate": 1.4224000000000001e-06, "loss": 0.0186, "step": 42889 }, { "epoch": 2.2866129978141494, "grad_norm": 2.4061596393585205, "learning_rate": 1.4222000000000003e-06, "loss": 0.0211, "step": 42890 }, { "epoch": 2.286666311243802, "grad_norm": 0.8908309936523438, "learning_rate": 1.4220000000000001e-06, "loss": 0.0046, "step": 42891 }, { "epoch": 2.2867196246734554, "grad_norm": 2.485499620437622, "learning_rate": 1.4218e-06, "loss": 0.0355, "step": 42892 }, { "epoch": 2.286772938103108, "grad_norm": 2.6787362098693848, "learning_rate": 1.4216000000000003e-06, "loss": 0.0316, "step": 42893 }, { "epoch": 2.2868262515327613, "grad_norm": 0.825925350189209, "learning_rate": 1.4214e-06, "loss": 0.0042, "step": 42894 }, { "epoch": 2.286879564962414, "grad_norm": 2.9016034603118896, "learning_rate": 1.4212e-06, "loss": 0.0301, "step": 42895 }, { "epoch": 2.286932878392067, "grad_norm": 1.4615999460220337, "learning_rate": 1.4210000000000002e-06, "loss": 0.0086, "step": 42896 }, { "epoch": 2.28698619182172, "grad_norm": 0.9538934826850891, "learning_rate": 1.4208000000000002e-06, "loss": 0.0042, "step": 42897 }, { "epoch": 2.287039505251373, "grad_norm": 2.2410550117492676, "learning_rate": 1.4206e-06, "loss": 0.011, "step": 42898 }, { "epoch": 2.2870928186810255, "grad_norm": 2.1782045364379883, "learning_rate": 1.4204000000000002e-06, "loss": 0.0123, "step": 42899 }, { "epoch": 2.2871461321106787, "grad_norm": 1.5282115936279297, "learning_rate": 1.4202000000000002e-06, "loss": 0.0082, "step": 42900 }, { "epoch": 2.2871994455403315, "grad_norm": 0.8212168216705322, "learning_rate": 1.42e-06, "loss": 0.0056, "step": 42901 }, { "epoch": 2.2872527589699847, "grad_norm": 2.6367735862731934, "learning_rate": 1.4198000000000002e-06, "loss": 0.0275, "step": 42902 }, { "epoch": 2.2873060723996375, "grad_norm": 1.4955596923828125, "learning_rate": 1.4196000000000001e-06, "loss": 0.0291, "step": 42903 }, { "epoch": 2.28735938582929, "grad_norm": 2.5714781284332275, "learning_rate": 1.4194000000000001e-06, "loss": 0.0258, "step": 42904 }, { "epoch": 2.2874126992589434, "grad_norm": 0.4563911259174347, "learning_rate": 1.4192000000000001e-06, "loss": 0.0038, "step": 42905 }, { "epoch": 2.287466012688596, "grad_norm": 2.1364150047302246, "learning_rate": 1.419e-06, "loss": 0.04, "step": 42906 }, { "epoch": 2.2875193261182494, "grad_norm": 2.8693885803222656, "learning_rate": 1.4188e-06, "loss": 0.0264, "step": 42907 }, { "epoch": 2.287572639547902, "grad_norm": 2.4689290523529053, "learning_rate": 1.4186000000000003e-06, "loss": 0.0148, "step": 42908 }, { "epoch": 2.287625952977555, "grad_norm": 2.193023681640625, "learning_rate": 1.4184e-06, "loss": 0.012, "step": 42909 }, { "epoch": 2.287679266407208, "grad_norm": 0.7797554135322571, "learning_rate": 1.4182e-06, "loss": 0.0042, "step": 42910 }, { "epoch": 2.287732579836861, "grad_norm": 1.743664264678955, "learning_rate": 1.4180000000000002e-06, "loss": 0.0234, "step": 42911 }, { "epoch": 2.2877858932665136, "grad_norm": 0.9496095776557922, "learning_rate": 1.4178e-06, "loss": 0.0179, "step": 42912 }, { "epoch": 2.287839206696167, "grad_norm": 0.4256369173526764, "learning_rate": 1.4176e-06, "loss": 0.0029, "step": 42913 }, { "epoch": 2.2878925201258196, "grad_norm": 3.374945640563965, "learning_rate": 1.4174000000000002e-06, "loss": 0.0179, "step": 42914 }, { "epoch": 2.2879458335554728, "grad_norm": 0.31918099522590637, "learning_rate": 1.4172000000000002e-06, "loss": 0.002, "step": 42915 }, { "epoch": 2.2879991469851255, "grad_norm": 1.712298035621643, "learning_rate": 1.417e-06, "loss": 0.008, "step": 42916 }, { "epoch": 2.2880524604147787, "grad_norm": 1.1778538227081299, "learning_rate": 1.4168000000000001e-06, "loss": 0.0216, "step": 42917 }, { "epoch": 2.2881057738444315, "grad_norm": 1.9411511421203613, "learning_rate": 1.4166000000000001e-06, "loss": 0.0125, "step": 42918 }, { "epoch": 2.2881590872740842, "grad_norm": 2.505478858947754, "learning_rate": 1.4164e-06, "loss": 0.0315, "step": 42919 }, { "epoch": 2.2882124007037374, "grad_norm": 2.998624563217163, "learning_rate": 1.4162e-06, "loss": 0.0228, "step": 42920 }, { "epoch": 2.28826571413339, "grad_norm": 2.014975070953369, "learning_rate": 1.416e-06, "loss": 0.0331, "step": 42921 }, { "epoch": 2.288319027563043, "grad_norm": 1.484606385231018, "learning_rate": 1.4158000000000003e-06, "loss": 0.0048, "step": 42922 }, { "epoch": 2.288372340992696, "grad_norm": 1.6105228662490845, "learning_rate": 1.4156e-06, "loss": 0.0174, "step": 42923 }, { "epoch": 2.288425654422349, "grad_norm": 1.8340156078338623, "learning_rate": 1.4154e-06, "loss": 0.0081, "step": 42924 }, { "epoch": 2.288478967852002, "grad_norm": 2.8152668476104736, "learning_rate": 1.4152000000000002e-06, "loss": 0.0266, "step": 42925 }, { "epoch": 2.288532281281655, "grad_norm": 1.5955328941345215, "learning_rate": 1.415e-06, "loss": 0.0085, "step": 42926 }, { "epoch": 2.2885855947113076, "grad_norm": 4.592845439910889, "learning_rate": 1.4148e-06, "loss": 0.0637, "step": 42927 }, { "epoch": 2.288638908140961, "grad_norm": 2.5886800289154053, "learning_rate": 1.4146000000000002e-06, "loss": 0.0183, "step": 42928 }, { "epoch": 2.2886922215706136, "grad_norm": 2.4053521156311035, "learning_rate": 1.4144000000000002e-06, "loss": 0.0179, "step": 42929 }, { "epoch": 2.288745535000267, "grad_norm": 0.47943612933158875, "learning_rate": 1.4142e-06, "loss": 0.0044, "step": 42930 }, { "epoch": 2.2887988484299195, "grad_norm": 6.334766864776611, "learning_rate": 1.4140000000000001e-06, "loss": 0.0536, "step": 42931 }, { "epoch": 2.2888521618595723, "grad_norm": 1.6209439039230347, "learning_rate": 1.4138000000000001e-06, "loss": 0.0226, "step": 42932 }, { "epoch": 2.2889054752892255, "grad_norm": 1.4871675968170166, "learning_rate": 1.4136000000000003e-06, "loss": 0.0134, "step": 42933 }, { "epoch": 2.2889587887188783, "grad_norm": 0.9870641231536865, "learning_rate": 1.4134e-06, "loss": 0.0058, "step": 42934 }, { "epoch": 2.289012102148531, "grad_norm": 3.5894157886505127, "learning_rate": 1.4132e-06, "loss": 0.0367, "step": 42935 }, { "epoch": 2.289065415578184, "grad_norm": 2.9852051734924316, "learning_rate": 1.4130000000000003e-06, "loss": 0.0118, "step": 42936 }, { "epoch": 2.289118729007837, "grad_norm": 6.1430487632751465, "learning_rate": 1.4128e-06, "loss": 0.0507, "step": 42937 }, { "epoch": 2.28917204243749, "grad_norm": 2.4188222885131836, "learning_rate": 1.4126e-06, "loss": 0.0171, "step": 42938 }, { "epoch": 2.289225355867143, "grad_norm": 2.2617342472076416, "learning_rate": 1.4124000000000002e-06, "loss": 0.0112, "step": 42939 }, { "epoch": 2.2892786692967957, "grad_norm": 2.7555878162384033, "learning_rate": 1.4122000000000002e-06, "loss": 0.028, "step": 42940 }, { "epoch": 2.289331982726449, "grad_norm": 6.605544567108154, "learning_rate": 1.412e-06, "loss": 0.0336, "step": 42941 }, { "epoch": 2.2893852961561016, "grad_norm": 2.786080837249756, "learning_rate": 1.4118000000000002e-06, "loss": 0.0195, "step": 42942 }, { "epoch": 2.289438609585755, "grad_norm": 1.1670242547988892, "learning_rate": 1.4116000000000002e-06, "loss": 0.0052, "step": 42943 }, { "epoch": 2.2894919230154076, "grad_norm": 2.4956893920898438, "learning_rate": 1.4114e-06, "loss": 0.0094, "step": 42944 }, { "epoch": 2.2895452364450604, "grad_norm": 0.3850478231906891, "learning_rate": 1.4112000000000001e-06, "loss": 0.0028, "step": 42945 }, { "epoch": 2.2895985498747136, "grad_norm": 1.7310075759887695, "learning_rate": 1.4110000000000001e-06, "loss": 0.0192, "step": 42946 }, { "epoch": 2.2896518633043663, "grad_norm": 2.530405282974243, "learning_rate": 1.4108000000000003e-06, "loss": 0.0262, "step": 42947 }, { "epoch": 2.289705176734019, "grad_norm": 2.295778274536133, "learning_rate": 1.4106e-06, "loss": 0.0204, "step": 42948 }, { "epoch": 2.2897584901636723, "grad_norm": 3.699767827987671, "learning_rate": 1.4104e-06, "loss": 0.0415, "step": 42949 }, { "epoch": 2.289811803593325, "grad_norm": 2.3621654510498047, "learning_rate": 1.4102000000000003e-06, "loss": 0.0244, "step": 42950 }, { "epoch": 2.2898651170229782, "grad_norm": 0.5256118774414062, "learning_rate": 1.41e-06, "loss": 0.0042, "step": 42951 }, { "epoch": 2.289918430452631, "grad_norm": 1.619814157485962, "learning_rate": 1.4098e-06, "loss": 0.0282, "step": 42952 }, { "epoch": 2.289971743882284, "grad_norm": 2.613128662109375, "learning_rate": 1.4096000000000002e-06, "loss": 0.0631, "step": 42953 }, { "epoch": 2.290025057311937, "grad_norm": 1.4907147884368896, "learning_rate": 1.4094000000000002e-06, "loss": 0.0079, "step": 42954 }, { "epoch": 2.2900783707415897, "grad_norm": 1.8359180688858032, "learning_rate": 1.4092e-06, "loss": 0.0166, "step": 42955 }, { "epoch": 2.290131684171243, "grad_norm": 1.7918201684951782, "learning_rate": 1.4090000000000002e-06, "loss": 0.0297, "step": 42956 }, { "epoch": 2.2901849976008957, "grad_norm": 1.151585340499878, "learning_rate": 1.4088000000000002e-06, "loss": 0.0201, "step": 42957 }, { "epoch": 2.2902383110305484, "grad_norm": 2.9511799812316895, "learning_rate": 1.4086000000000002e-06, "loss": 0.0104, "step": 42958 }, { "epoch": 2.2902916244602016, "grad_norm": 2.8647005558013916, "learning_rate": 1.4084000000000001e-06, "loss": 0.0134, "step": 42959 }, { "epoch": 2.2903449378898544, "grad_norm": 3.6895220279693604, "learning_rate": 1.4082000000000001e-06, "loss": 0.0155, "step": 42960 }, { "epoch": 2.290398251319507, "grad_norm": 1.5086696147918701, "learning_rate": 1.4080000000000001e-06, "loss": 0.0428, "step": 42961 }, { "epoch": 2.2904515647491603, "grad_norm": 1.3140572309494019, "learning_rate": 1.4078e-06, "loss": 0.0066, "step": 42962 }, { "epoch": 2.290504878178813, "grad_norm": 1.2225711345672607, "learning_rate": 1.4076e-06, "loss": 0.0092, "step": 42963 }, { "epoch": 2.2905581916084663, "grad_norm": 4.140297889709473, "learning_rate": 1.4074e-06, "loss": 0.0485, "step": 42964 }, { "epoch": 2.290611505038119, "grad_norm": 1.2246695756912231, "learning_rate": 1.4072000000000003e-06, "loss": 0.008, "step": 42965 }, { "epoch": 2.2906648184677723, "grad_norm": 0.9121987819671631, "learning_rate": 1.407e-06, "loss": 0.017, "step": 42966 }, { "epoch": 2.290718131897425, "grad_norm": 1.6736007928848267, "learning_rate": 1.4068e-06, "loss": 0.0069, "step": 42967 }, { "epoch": 2.2907714453270778, "grad_norm": 3.1619668006896973, "learning_rate": 1.4066000000000002e-06, "loss": 0.0187, "step": 42968 }, { "epoch": 2.290824758756731, "grad_norm": 2.2446956634521484, "learning_rate": 1.4064e-06, "loss": 0.0271, "step": 42969 }, { "epoch": 2.2908780721863837, "grad_norm": 3.3106040954589844, "learning_rate": 1.4062e-06, "loss": 0.0201, "step": 42970 }, { "epoch": 2.2909313856160365, "grad_norm": 3.5374910831451416, "learning_rate": 1.4060000000000002e-06, "loss": 0.0315, "step": 42971 }, { "epoch": 2.2909846990456897, "grad_norm": 3.3565869331359863, "learning_rate": 1.4058000000000002e-06, "loss": 0.0237, "step": 42972 }, { "epoch": 2.2910380124753424, "grad_norm": 2.7949368953704834, "learning_rate": 1.4056e-06, "loss": 0.0265, "step": 42973 }, { "epoch": 2.2910913259049956, "grad_norm": 1.1043291091918945, "learning_rate": 1.4054000000000001e-06, "loss": 0.0069, "step": 42974 }, { "epoch": 2.2911446393346484, "grad_norm": 1.766359806060791, "learning_rate": 1.4052000000000001e-06, "loss": 0.0185, "step": 42975 }, { "epoch": 2.291197952764301, "grad_norm": 1.7835571765899658, "learning_rate": 1.4050000000000003e-06, "loss": 0.0067, "step": 42976 }, { "epoch": 2.2912512661939544, "grad_norm": 1.4518318176269531, "learning_rate": 1.4048e-06, "loss": 0.0054, "step": 42977 }, { "epoch": 2.291304579623607, "grad_norm": 2.5746824741363525, "learning_rate": 1.4046e-06, "loss": 0.0195, "step": 42978 }, { "epoch": 2.2913578930532603, "grad_norm": 1.561463713645935, "learning_rate": 1.4044000000000003e-06, "loss": 0.0063, "step": 42979 }, { "epoch": 2.291411206482913, "grad_norm": 1.8250432014465332, "learning_rate": 1.4042e-06, "loss": 0.0067, "step": 42980 }, { "epoch": 2.291464519912566, "grad_norm": 0.7410463094711304, "learning_rate": 1.404e-06, "loss": 0.0091, "step": 42981 }, { "epoch": 2.291517833342219, "grad_norm": 1.5441724061965942, "learning_rate": 1.4038000000000002e-06, "loss": 0.0088, "step": 42982 }, { "epoch": 2.291571146771872, "grad_norm": 1.0019117593765259, "learning_rate": 1.4036000000000002e-06, "loss": 0.0063, "step": 42983 }, { "epoch": 2.2916244602015245, "grad_norm": 2.3254597187042236, "learning_rate": 1.4034e-06, "loss": 0.0203, "step": 42984 }, { "epoch": 2.2916777736311778, "grad_norm": 1.8645844459533691, "learning_rate": 1.4032000000000002e-06, "loss": 0.0282, "step": 42985 }, { "epoch": 2.2917310870608305, "grad_norm": 0.5632519721984863, "learning_rate": 1.4030000000000002e-06, "loss": 0.0051, "step": 42986 }, { "epoch": 2.2917844004904837, "grad_norm": 2.9945623874664307, "learning_rate": 1.4028e-06, "loss": 0.0456, "step": 42987 }, { "epoch": 2.2918377139201365, "grad_norm": 0.5501413345336914, "learning_rate": 1.4026000000000001e-06, "loss": 0.003, "step": 42988 }, { "epoch": 2.2918910273497892, "grad_norm": 1.7661179304122925, "learning_rate": 1.4024000000000001e-06, "loss": 0.0212, "step": 42989 }, { "epoch": 2.2919443407794424, "grad_norm": 0.9940290451049805, "learning_rate": 1.4022000000000003e-06, "loss": 0.0063, "step": 42990 }, { "epoch": 2.291997654209095, "grad_norm": 1.1162294149398804, "learning_rate": 1.402e-06, "loss": 0.0061, "step": 42991 }, { "epoch": 2.2920509676387484, "grad_norm": 1.4221477508544922, "learning_rate": 1.4018e-06, "loss": 0.0082, "step": 42992 }, { "epoch": 2.292104281068401, "grad_norm": 1.9819371700286865, "learning_rate": 1.4016000000000003e-06, "loss": 0.01, "step": 42993 }, { "epoch": 2.292157594498054, "grad_norm": 1.5124585628509521, "learning_rate": 1.4014e-06, "loss": 0.008, "step": 42994 }, { "epoch": 2.292210907927707, "grad_norm": 2.196737289428711, "learning_rate": 1.4012e-06, "loss": 0.0208, "step": 42995 }, { "epoch": 2.29226422135736, "grad_norm": 1.2405608892440796, "learning_rate": 1.4010000000000002e-06, "loss": 0.0114, "step": 42996 }, { "epoch": 2.2923175347870126, "grad_norm": 2.3511087894439697, "learning_rate": 1.4008000000000002e-06, "loss": 0.0199, "step": 42997 }, { "epoch": 2.292370848216666, "grad_norm": 0.4084823727607727, "learning_rate": 1.4006e-06, "loss": 0.0032, "step": 42998 }, { "epoch": 2.2924241616463186, "grad_norm": 2.020131826400757, "learning_rate": 1.4004000000000002e-06, "loss": 0.0368, "step": 42999 }, { "epoch": 2.2924774750759718, "grad_norm": 1.4467930793762207, "learning_rate": 1.4002000000000002e-06, "loss": 0.0441, "step": 43000 }, { "epoch": 2.2925307885056245, "grad_norm": 1.0256956815719604, "learning_rate": 1.4000000000000001e-06, "loss": 0.0044, "step": 43001 }, { "epoch": 2.2925841019352777, "grad_norm": 3.139962911605835, "learning_rate": 1.3998000000000001e-06, "loss": 0.0258, "step": 43002 }, { "epoch": 2.2926374153649305, "grad_norm": 1.896178126335144, "learning_rate": 1.3996000000000001e-06, "loss": 0.0212, "step": 43003 }, { "epoch": 2.2926907287945832, "grad_norm": 1.3821313381195068, "learning_rate": 1.3994e-06, "loss": 0.0061, "step": 43004 }, { "epoch": 2.2927440422242364, "grad_norm": 0.8314201235771179, "learning_rate": 1.3992e-06, "loss": 0.0037, "step": 43005 }, { "epoch": 2.292797355653889, "grad_norm": 3.070521831512451, "learning_rate": 1.399e-06, "loss": 0.0079, "step": 43006 }, { "epoch": 2.292850669083542, "grad_norm": 2.209728956222534, "learning_rate": 1.3988e-06, "loss": 0.0294, "step": 43007 }, { "epoch": 2.292903982513195, "grad_norm": 2.2308263778686523, "learning_rate": 1.3986000000000003e-06, "loss": 0.0133, "step": 43008 }, { "epoch": 2.292957295942848, "grad_norm": 4.370617389678955, "learning_rate": 1.3984e-06, "loss": 0.0113, "step": 43009 }, { "epoch": 2.293010609372501, "grad_norm": 1.4753432273864746, "learning_rate": 1.3982e-06, "loss": 0.0166, "step": 43010 }, { "epoch": 2.293063922802154, "grad_norm": 2.45261812210083, "learning_rate": 1.3980000000000002e-06, "loss": 0.0186, "step": 43011 }, { "epoch": 2.2931172362318066, "grad_norm": 3.3227057456970215, "learning_rate": 1.3978e-06, "loss": 0.0322, "step": 43012 }, { "epoch": 2.29317054966146, "grad_norm": 0.7512257695198059, "learning_rate": 1.3976e-06, "loss": 0.0049, "step": 43013 }, { "epoch": 2.2932238630911126, "grad_norm": 3.480482339859009, "learning_rate": 1.3974000000000002e-06, "loss": 0.0293, "step": 43014 }, { "epoch": 2.293277176520766, "grad_norm": 2.858271598815918, "learning_rate": 1.3972000000000001e-06, "loss": 0.0138, "step": 43015 }, { "epoch": 2.2933304899504185, "grad_norm": 0.6019482016563416, "learning_rate": 1.397e-06, "loss": 0.0038, "step": 43016 }, { "epoch": 2.2933838033800713, "grad_norm": 1.0787360668182373, "learning_rate": 1.3968000000000001e-06, "loss": 0.017, "step": 43017 }, { "epoch": 2.2934371168097245, "grad_norm": 1.1875364780426025, "learning_rate": 1.3966e-06, "loss": 0.0061, "step": 43018 }, { "epoch": 2.2934904302393773, "grad_norm": 1.45945405960083, "learning_rate": 1.3963999999999999e-06, "loss": 0.0293, "step": 43019 }, { "epoch": 2.29354374366903, "grad_norm": 3.590235471725464, "learning_rate": 1.3962e-06, "loss": 0.0202, "step": 43020 }, { "epoch": 2.2935970570986832, "grad_norm": 1.6206105947494507, "learning_rate": 1.396e-06, "loss": 0.0108, "step": 43021 }, { "epoch": 2.293650370528336, "grad_norm": 1.0303689241409302, "learning_rate": 1.3958000000000003e-06, "loss": 0.0044, "step": 43022 }, { "epoch": 2.293703683957989, "grad_norm": 1.3362349271774292, "learning_rate": 1.3956e-06, "loss": 0.0186, "step": 43023 }, { "epoch": 2.293756997387642, "grad_norm": 0.29795485734939575, "learning_rate": 1.3954e-06, "loss": 0.0024, "step": 43024 }, { "epoch": 2.2938103108172947, "grad_norm": 0.7363037467002869, "learning_rate": 1.3952000000000002e-06, "loss": 0.0051, "step": 43025 }, { "epoch": 2.293863624246948, "grad_norm": 1.2579765319824219, "learning_rate": 1.3950000000000002e-06, "loss": 0.0079, "step": 43026 }, { "epoch": 2.2939169376766007, "grad_norm": 1.2627965211868286, "learning_rate": 1.3948e-06, "loss": 0.0103, "step": 43027 }, { "epoch": 2.293970251106254, "grad_norm": 1.5783814191818237, "learning_rate": 1.3946000000000002e-06, "loss": 0.024, "step": 43028 }, { "epoch": 2.2940235645359066, "grad_norm": 1.3553314208984375, "learning_rate": 1.3944000000000002e-06, "loss": 0.0056, "step": 43029 }, { "epoch": 2.2940768779655594, "grad_norm": 2.690778970718384, "learning_rate": 1.3942e-06, "loss": 0.0421, "step": 43030 }, { "epoch": 2.2941301913952126, "grad_norm": 2.5473504066467285, "learning_rate": 1.3940000000000001e-06, "loss": 0.041, "step": 43031 }, { "epoch": 2.2941835048248653, "grad_norm": 0.7222505211830139, "learning_rate": 1.3938000000000001e-06, "loss": 0.0051, "step": 43032 }, { "epoch": 2.294236818254518, "grad_norm": 3.4108924865722656, "learning_rate": 1.3936000000000003e-06, "loss": 0.0368, "step": 43033 }, { "epoch": 2.2942901316841713, "grad_norm": 1.7835330963134766, "learning_rate": 1.3934e-06, "loss": 0.0122, "step": 43034 }, { "epoch": 2.294343445113824, "grad_norm": 1.1314104795455933, "learning_rate": 1.3932e-06, "loss": 0.0124, "step": 43035 }, { "epoch": 2.2943967585434772, "grad_norm": 1.6110420227050781, "learning_rate": 1.3930000000000003e-06, "loss": 0.011, "step": 43036 }, { "epoch": 2.29445007197313, "grad_norm": 2.6382248401641846, "learning_rate": 1.3928e-06, "loss": 0.0154, "step": 43037 }, { "epoch": 2.294503385402783, "grad_norm": 0.4975278675556183, "learning_rate": 1.3926e-06, "loss": 0.0038, "step": 43038 }, { "epoch": 2.294556698832436, "grad_norm": 0.8177931308746338, "learning_rate": 1.3924000000000002e-06, "loss": 0.0053, "step": 43039 }, { "epoch": 2.2946100122620887, "grad_norm": 1.3525310754776, "learning_rate": 1.3922000000000002e-06, "loss": 0.0229, "step": 43040 }, { "epoch": 2.294663325691742, "grad_norm": 3.2376515865325928, "learning_rate": 1.392e-06, "loss": 0.023, "step": 43041 }, { "epoch": 2.2947166391213947, "grad_norm": 3.2129058837890625, "learning_rate": 1.3918000000000002e-06, "loss": 0.0157, "step": 43042 }, { "epoch": 2.2947699525510474, "grad_norm": 1.2792069911956787, "learning_rate": 1.3916000000000002e-06, "loss": 0.0146, "step": 43043 }, { "epoch": 2.2948232659807006, "grad_norm": 0.9217740893363953, "learning_rate": 1.3914000000000003e-06, "loss": 0.0135, "step": 43044 }, { "epoch": 2.2948765794103534, "grad_norm": 1.3833272457122803, "learning_rate": 1.3912000000000001e-06, "loss": 0.0192, "step": 43045 }, { "epoch": 2.2949298928400066, "grad_norm": 3.986917018890381, "learning_rate": 1.3910000000000001e-06, "loss": 0.0338, "step": 43046 }, { "epoch": 2.2949832062696593, "grad_norm": 0.5685141086578369, "learning_rate": 1.3908000000000003e-06, "loss": 0.0035, "step": 43047 }, { "epoch": 2.295036519699312, "grad_norm": 0.41357046365737915, "learning_rate": 1.3906e-06, "loss": 0.0028, "step": 43048 }, { "epoch": 2.2950898331289653, "grad_norm": 2.342967987060547, "learning_rate": 1.3904e-06, "loss": 0.0145, "step": 43049 }, { "epoch": 2.295143146558618, "grad_norm": 2.6323609352111816, "learning_rate": 1.3902000000000003e-06, "loss": 0.0337, "step": 43050 }, { "epoch": 2.2951964599882713, "grad_norm": 0.5744507908821106, "learning_rate": 1.3900000000000002e-06, "loss": 0.0035, "step": 43051 }, { "epoch": 2.295249773417924, "grad_norm": 1.8598871231079102, "learning_rate": 1.3898e-06, "loss": 0.0198, "step": 43052 }, { "epoch": 2.295303086847577, "grad_norm": 1.582411527633667, "learning_rate": 1.3896000000000002e-06, "loss": 0.0298, "step": 43053 }, { "epoch": 2.29535640027723, "grad_norm": 2.2377004623413086, "learning_rate": 1.3894000000000002e-06, "loss": 0.0113, "step": 43054 }, { "epoch": 2.2954097137068827, "grad_norm": 1.7052147388458252, "learning_rate": 1.3892e-06, "loss": 0.0104, "step": 43055 }, { "epoch": 2.2954630271365355, "grad_norm": 2.220958948135376, "learning_rate": 1.3890000000000002e-06, "loss": 0.0194, "step": 43056 }, { "epoch": 2.2955163405661887, "grad_norm": 3.099303722381592, "learning_rate": 1.3888000000000002e-06, "loss": 0.0124, "step": 43057 }, { "epoch": 2.2955696539958415, "grad_norm": 2.0735361576080322, "learning_rate": 1.3886000000000001e-06, "loss": 0.0385, "step": 43058 }, { "epoch": 2.2956229674254947, "grad_norm": 2.3101747035980225, "learning_rate": 1.3884000000000001e-06, "loss": 0.0087, "step": 43059 }, { "epoch": 2.2956762808551474, "grad_norm": 1.2560306787490845, "learning_rate": 1.3882000000000001e-06, "loss": 0.0069, "step": 43060 }, { "epoch": 2.2957295942848, "grad_norm": 2.068030834197998, "learning_rate": 1.388e-06, "loss": 0.0111, "step": 43061 }, { "epoch": 2.2957829077144534, "grad_norm": 1.0878469944000244, "learning_rate": 1.3878e-06, "loss": 0.0049, "step": 43062 }, { "epoch": 2.295836221144106, "grad_norm": 3.295664072036743, "learning_rate": 1.3876e-06, "loss": 0.0296, "step": 43063 }, { "epoch": 2.2958895345737593, "grad_norm": 2.866471529006958, "learning_rate": 1.3874e-06, "loss": 0.0128, "step": 43064 }, { "epoch": 2.295942848003412, "grad_norm": 1.1257587671279907, "learning_rate": 1.3872000000000002e-06, "loss": 0.0071, "step": 43065 }, { "epoch": 2.295996161433065, "grad_norm": 4.207496643066406, "learning_rate": 1.387e-06, "loss": 0.0172, "step": 43066 }, { "epoch": 2.296049474862718, "grad_norm": 1.093913197517395, "learning_rate": 1.3868e-06, "loss": 0.0089, "step": 43067 }, { "epoch": 2.296102788292371, "grad_norm": 2.9658071994781494, "learning_rate": 1.3866000000000002e-06, "loss": 0.0082, "step": 43068 }, { "epoch": 2.2961561017220236, "grad_norm": 2.0891904830932617, "learning_rate": 1.3864000000000002e-06, "loss": 0.0271, "step": 43069 }, { "epoch": 2.2962094151516768, "grad_norm": 1.221486210823059, "learning_rate": 1.3862e-06, "loss": 0.0054, "step": 43070 }, { "epoch": 2.2962627285813295, "grad_norm": 2.478111505508423, "learning_rate": 1.3860000000000002e-06, "loss": 0.0104, "step": 43071 }, { "epoch": 2.2963160420109827, "grad_norm": 1.0276328325271606, "learning_rate": 1.3858000000000001e-06, "loss": 0.0238, "step": 43072 }, { "epoch": 2.2963693554406355, "grad_norm": 3.2303309440612793, "learning_rate": 1.3856e-06, "loss": 0.0155, "step": 43073 }, { "epoch": 2.2964226688702887, "grad_norm": 1.5072262287139893, "learning_rate": 1.3854000000000001e-06, "loss": 0.0081, "step": 43074 }, { "epoch": 2.2964759822999414, "grad_norm": 0.980491578578949, "learning_rate": 1.3852e-06, "loss": 0.0188, "step": 43075 }, { "epoch": 2.296529295729594, "grad_norm": 1.5967202186584473, "learning_rate": 1.3850000000000003e-06, "loss": 0.0073, "step": 43076 }, { "epoch": 2.2965826091592474, "grad_norm": 0.15360699594020844, "learning_rate": 1.3848e-06, "loss": 0.0015, "step": 43077 }, { "epoch": 2.2966359225889, "grad_norm": 1.7771114110946655, "learning_rate": 1.3846e-06, "loss": 0.0075, "step": 43078 }, { "epoch": 2.296689236018553, "grad_norm": 1.4001376628875732, "learning_rate": 1.3844000000000002e-06, "loss": 0.0236, "step": 43079 }, { "epoch": 2.296742549448206, "grad_norm": 2.3052573204040527, "learning_rate": 1.3842e-06, "loss": 0.0186, "step": 43080 }, { "epoch": 2.296795862877859, "grad_norm": 1.590958595275879, "learning_rate": 1.384e-06, "loss": 0.0152, "step": 43081 }, { "epoch": 2.2968491763075116, "grad_norm": 2.1241092681884766, "learning_rate": 1.3838000000000002e-06, "loss": 0.0268, "step": 43082 }, { "epoch": 2.296902489737165, "grad_norm": 1.790144443511963, "learning_rate": 1.3836000000000002e-06, "loss": 0.0311, "step": 43083 }, { "epoch": 2.2969558031668176, "grad_norm": 1.7968651056289673, "learning_rate": 1.3834e-06, "loss": 0.0108, "step": 43084 }, { "epoch": 2.297009116596471, "grad_norm": 2.126171350479126, "learning_rate": 1.3832000000000002e-06, "loss": 0.0312, "step": 43085 }, { "epoch": 2.2970624300261235, "grad_norm": 2.7143287658691406, "learning_rate": 1.3830000000000001e-06, "loss": 0.0103, "step": 43086 }, { "epoch": 2.2971157434557767, "grad_norm": 2.272289514541626, "learning_rate": 1.3828e-06, "loss": 0.0107, "step": 43087 }, { "epoch": 2.2971690568854295, "grad_norm": 0.38077595829963684, "learning_rate": 1.3826000000000001e-06, "loss": 0.0022, "step": 43088 }, { "epoch": 2.2972223703150823, "grad_norm": 2.2946667671203613, "learning_rate": 1.3824e-06, "loss": 0.0343, "step": 43089 }, { "epoch": 2.2972756837447355, "grad_norm": 3.06734299659729, "learning_rate": 1.3822000000000003e-06, "loss": 0.0261, "step": 43090 }, { "epoch": 2.297328997174388, "grad_norm": 0.34479641914367676, "learning_rate": 1.382e-06, "loss": 0.0023, "step": 43091 }, { "epoch": 2.297382310604041, "grad_norm": 1.8314146995544434, "learning_rate": 1.3818e-06, "loss": 0.0183, "step": 43092 }, { "epoch": 2.297435624033694, "grad_norm": 3.6808953285217285, "learning_rate": 1.3816000000000002e-06, "loss": 0.0346, "step": 43093 }, { "epoch": 2.297488937463347, "grad_norm": 1.4151842594146729, "learning_rate": 1.3814000000000002e-06, "loss": 0.0066, "step": 43094 }, { "epoch": 2.297542250893, "grad_norm": 0.7228200435638428, "learning_rate": 1.3812e-06, "loss": 0.0039, "step": 43095 }, { "epoch": 2.297595564322653, "grad_norm": 3.421745538711548, "learning_rate": 1.3810000000000002e-06, "loss": 0.0275, "step": 43096 }, { "epoch": 2.2976488777523056, "grad_norm": 4.0643463134765625, "learning_rate": 1.3808000000000002e-06, "loss": 0.0131, "step": 43097 }, { "epoch": 2.297702191181959, "grad_norm": 2.1650123596191406, "learning_rate": 1.3806e-06, "loss": 0.0097, "step": 43098 }, { "epoch": 2.2977555046116116, "grad_norm": 0.8073241114616394, "learning_rate": 1.3804000000000002e-06, "loss": 0.0043, "step": 43099 }, { "epoch": 2.297808818041265, "grad_norm": 4.442052364349365, "learning_rate": 1.3802000000000001e-06, "loss": 0.0179, "step": 43100 }, { "epoch": 2.2978621314709176, "grad_norm": 0.7896131873130798, "learning_rate": 1.3800000000000001e-06, "loss": 0.0209, "step": 43101 }, { "epoch": 2.2979154449005703, "grad_norm": 3.4670510292053223, "learning_rate": 1.3798000000000001e-06, "loss": 0.0305, "step": 43102 }, { "epoch": 2.2979687583302235, "grad_norm": 0.3806595504283905, "learning_rate": 1.3796e-06, "loss": 0.0027, "step": 43103 }, { "epoch": 2.2980220717598763, "grad_norm": 0.39381569623947144, "learning_rate": 1.3794e-06, "loss": 0.0037, "step": 43104 }, { "epoch": 2.298075385189529, "grad_norm": 1.7003800868988037, "learning_rate": 1.3792e-06, "loss": 0.0095, "step": 43105 }, { "epoch": 2.2981286986191822, "grad_norm": 2.989227533340454, "learning_rate": 1.379e-06, "loss": 0.0173, "step": 43106 }, { "epoch": 2.298182012048835, "grad_norm": 2.73292875289917, "learning_rate": 1.3788e-06, "loss": 0.0124, "step": 43107 }, { "epoch": 2.298235325478488, "grad_norm": 2.105961322784424, "learning_rate": 1.3786000000000002e-06, "loss": 0.0122, "step": 43108 }, { "epoch": 2.298288638908141, "grad_norm": 0.27470067143440247, "learning_rate": 1.3784e-06, "loss": 0.0024, "step": 43109 }, { "epoch": 2.2983419523377937, "grad_norm": 2.471585750579834, "learning_rate": 1.3782e-06, "loss": 0.0151, "step": 43110 }, { "epoch": 2.298395265767447, "grad_norm": 2.6295690536499023, "learning_rate": 1.3780000000000002e-06, "loss": 0.0271, "step": 43111 }, { "epoch": 2.2984485791970997, "grad_norm": 0.735231876373291, "learning_rate": 1.3778000000000002e-06, "loss": 0.0047, "step": 43112 }, { "epoch": 2.298501892626753, "grad_norm": 0.2064962089061737, "learning_rate": 1.3776e-06, "loss": 0.0014, "step": 43113 }, { "epoch": 2.2985552060564056, "grad_norm": 2.968017339706421, "learning_rate": 1.3774000000000001e-06, "loss": 0.0108, "step": 43114 }, { "epoch": 2.2986085194860584, "grad_norm": 1.2555789947509766, "learning_rate": 1.3772000000000001e-06, "loss": 0.0057, "step": 43115 }, { "epoch": 2.2986618329157116, "grad_norm": 1.110840916633606, "learning_rate": 1.377e-06, "loss": 0.0063, "step": 43116 }, { "epoch": 2.2987151463453643, "grad_norm": 1.9535348415374756, "learning_rate": 1.3768e-06, "loss": 0.0175, "step": 43117 }, { "epoch": 2.298768459775017, "grad_norm": 2.8558480739593506, "learning_rate": 1.3766e-06, "loss": 0.0555, "step": 43118 }, { "epoch": 2.2988217732046703, "grad_norm": 1.173828363418579, "learning_rate": 1.3764000000000003e-06, "loss": 0.0168, "step": 43119 }, { "epoch": 2.298875086634323, "grad_norm": 0.475222647190094, "learning_rate": 1.3762e-06, "loss": 0.0025, "step": 43120 }, { "epoch": 2.2989284000639763, "grad_norm": 1.9054988622665405, "learning_rate": 1.376e-06, "loss": 0.0087, "step": 43121 }, { "epoch": 2.298981713493629, "grad_norm": 3.2694203853607178, "learning_rate": 1.3758000000000002e-06, "loss": 0.0171, "step": 43122 }, { "epoch": 2.299035026923282, "grad_norm": 1.443287968635559, "learning_rate": 1.3756e-06, "loss": 0.0358, "step": 43123 }, { "epoch": 2.299088340352935, "grad_norm": 2.3912084102630615, "learning_rate": 1.3754e-06, "loss": 0.0194, "step": 43124 }, { "epoch": 2.2991416537825877, "grad_norm": 2.8190982341766357, "learning_rate": 1.3752000000000002e-06, "loss": 0.0402, "step": 43125 }, { "epoch": 2.299194967212241, "grad_norm": 0.21685124933719635, "learning_rate": 1.3750000000000002e-06, "loss": 0.0022, "step": 43126 }, { "epoch": 2.2992482806418937, "grad_norm": 0.9779834747314453, "learning_rate": 1.3748e-06, "loss": 0.0127, "step": 43127 }, { "epoch": 2.2993015940715464, "grad_norm": 2.70001220703125, "learning_rate": 1.3746000000000001e-06, "loss": 0.0249, "step": 43128 }, { "epoch": 2.2993549075011996, "grad_norm": 0.9992350339889526, "learning_rate": 1.3744000000000001e-06, "loss": 0.0088, "step": 43129 }, { "epoch": 2.2994082209308524, "grad_norm": 2.455005168914795, "learning_rate": 1.3742e-06, "loss": 0.0192, "step": 43130 }, { "epoch": 2.2994615343605056, "grad_norm": 2.0515692234039307, "learning_rate": 1.374e-06, "loss": 0.0071, "step": 43131 }, { "epoch": 2.2995148477901584, "grad_norm": 2.4877500534057617, "learning_rate": 1.3738e-06, "loss": 0.0096, "step": 43132 }, { "epoch": 2.299568161219811, "grad_norm": 1.6895694732666016, "learning_rate": 1.3736000000000003e-06, "loss": 0.031, "step": 43133 }, { "epoch": 2.2996214746494643, "grad_norm": 1.7197948694229126, "learning_rate": 1.3734e-06, "loss": 0.012, "step": 43134 }, { "epoch": 2.299674788079117, "grad_norm": 1.4015530347824097, "learning_rate": 1.3732e-06, "loss": 0.0074, "step": 43135 }, { "epoch": 2.2997281015087703, "grad_norm": 0.2984199523925781, "learning_rate": 1.3730000000000002e-06, "loss": 0.0027, "step": 43136 }, { "epoch": 2.299781414938423, "grad_norm": 1.3277548551559448, "learning_rate": 1.3728000000000002e-06, "loss": 0.0188, "step": 43137 }, { "epoch": 2.299834728368076, "grad_norm": 3.2514724731445312, "learning_rate": 1.3726e-06, "loss": 0.0115, "step": 43138 }, { "epoch": 2.299888041797729, "grad_norm": 2.7009122371673584, "learning_rate": 1.3724000000000002e-06, "loss": 0.0292, "step": 43139 }, { "epoch": 2.2999413552273817, "grad_norm": 1.3918423652648926, "learning_rate": 1.3722000000000002e-06, "loss": 0.0062, "step": 43140 }, { "epoch": 2.2999946686570345, "grad_norm": 2.409726619720459, "learning_rate": 1.372e-06, "loss": 0.0372, "step": 43141 }, { "epoch": 2.3000479820866877, "grad_norm": 0.3425595760345459, "learning_rate": 1.3718000000000001e-06, "loss": 0.0024, "step": 43142 }, { "epoch": 2.3001012955163405, "grad_norm": 1.0840049982070923, "learning_rate": 1.3716000000000001e-06, "loss": 0.0086, "step": 43143 }, { "epoch": 2.3001546089459937, "grad_norm": 2.3432159423828125, "learning_rate": 1.3714000000000003e-06, "loss": 0.021, "step": 43144 }, { "epoch": 2.3002079223756464, "grad_norm": 0.6543183922767639, "learning_rate": 1.3712e-06, "loss": 0.0053, "step": 43145 }, { "epoch": 2.300261235805299, "grad_norm": 1.8983402252197266, "learning_rate": 1.371e-06, "loss": 0.0273, "step": 43146 }, { "epoch": 2.3003145492349524, "grad_norm": 2.692026138305664, "learning_rate": 1.3708000000000003e-06, "loss": 0.0363, "step": 43147 }, { "epoch": 2.300367862664605, "grad_norm": 0.21965627372264862, "learning_rate": 1.3706e-06, "loss": 0.0016, "step": 43148 }, { "epoch": 2.3004211760942583, "grad_norm": 2.4510133266448975, "learning_rate": 1.3704e-06, "loss": 0.0113, "step": 43149 }, { "epoch": 2.300474489523911, "grad_norm": 2.1404378414154053, "learning_rate": 1.3702000000000002e-06, "loss": 0.0061, "step": 43150 }, { "epoch": 2.300527802953564, "grad_norm": 2.8534348011016846, "learning_rate": 1.3700000000000002e-06, "loss": 0.0162, "step": 43151 }, { "epoch": 2.300581116383217, "grad_norm": 0.6506591439247131, "learning_rate": 1.3698e-06, "loss": 0.0031, "step": 43152 }, { "epoch": 2.30063442981287, "grad_norm": 1.158190369606018, "learning_rate": 1.3696000000000002e-06, "loss": 0.0186, "step": 43153 }, { "epoch": 2.3006877432425226, "grad_norm": 2.8349053859710693, "learning_rate": 1.3694000000000002e-06, "loss": 0.0386, "step": 43154 }, { "epoch": 2.3007410566721758, "grad_norm": 1.670240044593811, "learning_rate": 1.3692e-06, "loss": 0.0072, "step": 43155 }, { "epoch": 2.3007943701018285, "grad_norm": 1.520485520362854, "learning_rate": 1.3690000000000001e-06, "loss": 0.0144, "step": 43156 }, { "epoch": 2.3008476835314817, "grad_norm": 2.679640054702759, "learning_rate": 1.3688000000000001e-06, "loss": 0.0065, "step": 43157 }, { "epoch": 2.3009009969611345, "grad_norm": 1.2948402166366577, "learning_rate": 1.3686000000000001e-06, "loss": 0.0058, "step": 43158 }, { "epoch": 2.3009543103907877, "grad_norm": 1.986317753791809, "learning_rate": 1.3684e-06, "loss": 0.0295, "step": 43159 }, { "epoch": 2.3010076238204404, "grad_norm": 3.235990047454834, "learning_rate": 1.3682e-06, "loss": 0.0216, "step": 43160 }, { "epoch": 2.301060937250093, "grad_norm": 2.9184799194335938, "learning_rate": 1.368e-06, "loss": 0.0273, "step": 43161 }, { "epoch": 2.3011142506797464, "grad_norm": 0.9167550206184387, "learning_rate": 1.3678000000000003e-06, "loss": 0.0034, "step": 43162 }, { "epoch": 2.301167564109399, "grad_norm": 2.6424341201782227, "learning_rate": 1.3676e-06, "loss": 0.0186, "step": 43163 }, { "epoch": 2.301220877539052, "grad_norm": 4.267983913421631, "learning_rate": 1.3674e-06, "loss": 0.0149, "step": 43164 }, { "epoch": 2.301274190968705, "grad_norm": 0.8308346271514893, "learning_rate": 1.3672000000000002e-06, "loss": 0.0051, "step": 43165 }, { "epoch": 2.301327504398358, "grad_norm": 0.9946979880332947, "learning_rate": 1.367e-06, "loss": 0.006, "step": 43166 }, { "epoch": 2.3013808178280106, "grad_norm": 1.780824899673462, "learning_rate": 1.3668e-06, "loss": 0.0128, "step": 43167 }, { "epoch": 2.301434131257664, "grad_norm": 0.843480110168457, "learning_rate": 1.3666000000000002e-06, "loss": 0.0043, "step": 43168 }, { "epoch": 2.3014874446873166, "grad_norm": 3.7552549839019775, "learning_rate": 1.3664000000000002e-06, "loss": 0.0226, "step": 43169 }, { "epoch": 2.30154075811697, "grad_norm": 0.8034151792526245, "learning_rate": 1.3662e-06, "loss": 0.0041, "step": 43170 }, { "epoch": 2.3015940715466225, "grad_norm": 1.1557693481445312, "learning_rate": 1.3660000000000001e-06, "loss": 0.0207, "step": 43171 }, { "epoch": 2.3016473849762757, "grad_norm": 1.7027419805526733, "learning_rate": 1.3658000000000001e-06, "loss": 0.0358, "step": 43172 }, { "epoch": 2.3017006984059285, "grad_norm": 2.824110984802246, "learning_rate": 1.3655999999999999e-06, "loss": 0.0107, "step": 43173 }, { "epoch": 2.3017540118355813, "grad_norm": 1.8192557096481323, "learning_rate": 1.3654e-06, "loss": 0.0299, "step": 43174 }, { "epoch": 2.3018073252652345, "grad_norm": 2.4061574935913086, "learning_rate": 1.3652e-06, "loss": 0.0355, "step": 43175 }, { "epoch": 2.301860638694887, "grad_norm": 1.5394002199172974, "learning_rate": 1.3650000000000003e-06, "loss": 0.0271, "step": 43176 }, { "epoch": 2.30191395212454, "grad_norm": 0.40771278738975525, "learning_rate": 1.3648e-06, "loss": 0.0035, "step": 43177 }, { "epoch": 2.301967265554193, "grad_norm": 1.135955572128296, "learning_rate": 1.3646e-06, "loss": 0.0048, "step": 43178 }, { "epoch": 2.302020578983846, "grad_norm": 1.6870906352996826, "learning_rate": 1.3644000000000002e-06, "loss": 0.0156, "step": 43179 }, { "epoch": 2.302073892413499, "grad_norm": 1.0902202129364014, "learning_rate": 1.3642000000000002e-06, "loss": 0.0055, "step": 43180 }, { "epoch": 2.302127205843152, "grad_norm": 2.348506212234497, "learning_rate": 1.364e-06, "loss": 0.0214, "step": 43181 }, { "epoch": 2.3021805192728046, "grad_norm": 2.2126073837280273, "learning_rate": 1.3638000000000002e-06, "loss": 0.0118, "step": 43182 }, { "epoch": 2.302233832702458, "grad_norm": 3.6693360805511475, "learning_rate": 1.3636000000000002e-06, "loss": 0.0216, "step": 43183 }, { "epoch": 2.3022871461321106, "grad_norm": 0.784934937953949, "learning_rate": 1.3634e-06, "loss": 0.0054, "step": 43184 }, { "epoch": 2.302340459561764, "grad_norm": 1.931273102760315, "learning_rate": 1.3632000000000001e-06, "loss": 0.0289, "step": 43185 }, { "epoch": 2.3023937729914166, "grad_norm": 1.0233253240585327, "learning_rate": 1.3630000000000001e-06, "loss": 0.0041, "step": 43186 }, { "epoch": 2.3024470864210693, "grad_norm": 2.3573193550109863, "learning_rate": 1.3628000000000003e-06, "loss": 0.0181, "step": 43187 }, { "epoch": 2.3025003998507225, "grad_norm": 1.1621443033218384, "learning_rate": 1.3626e-06, "loss": 0.0056, "step": 43188 }, { "epoch": 2.3025537132803753, "grad_norm": 2.8068528175354004, "learning_rate": 1.3624e-06, "loss": 0.0357, "step": 43189 }, { "epoch": 2.302607026710028, "grad_norm": 3.606860637664795, "learning_rate": 1.3622000000000003e-06, "loss": 0.0472, "step": 43190 }, { "epoch": 2.3026603401396812, "grad_norm": 1.0914448499679565, "learning_rate": 1.362e-06, "loss": 0.0194, "step": 43191 }, { "epoch": 2.302713653569334, "grad_norm": 0.9333827495574951, "learning_rate": 1.3618e-06, "loss": 0.017, "step": 43192 }, { "epoch": 2.302766966998987, "grad_norm": 1.6129006147384644, "learning_rate": 1.3616000000000002e-06, "loss": 0.0126, "step": 43193 }, { "epoch": 2.30282028042864, "grad_norm": 2.0240848064422607, "learning_rate": 1.3614000000000002e-06, "loss": 0.0102, "step": 43194 }, { "epoch": 2.3028735938582927, "grad_norm": 1.4426692724227905, "learning_rate": 1.3612e-06, "loss": 0.0234, "step": 43195 }, { "epoch": 2.302926907287946, "grad_norm": 2.51773738861084, "learning_rate": 1.3610000000000002e-06, "loss": 0.0232, "step": 43196 }, { "epoch": 2.3029802207175987, "grad_norm": 0.8553028106689453, "learning_rate": 1.3608000000000002e-06, "loss": 0.0048, "step": 43197 }, { "epoch": 2.303033534147252, "grad_norm": 0.6056851148605347, "learning_rate": 1.3606e-06, "loss": 0.0039, "step": 43198 }, { "epoch": 2.3030868475769046, "grad_norm": 3.1300370693206787, "learning_rate": 1.3604000000000001e-06, "loss": 0.0238, "step": 43199 }, { "epoch": 2.3031401610065574, "grad_norm": 2.136228561401367, "learning_rate": 1.3602000000000001e-06, "loss": 0.0076, "step": 43200 }, { "epoch": 2.3031934744362106, "grad_norm": 1.6850093603134155, "learning_rate": 1.3600000000000001e-06, "loss": 0.008, "step": 43201 }, { "epoch": 2.3032467878658633, "grad_norm": 1.2097578048706055, "learning_rate": 1.3598e-06, "loss": 0.0235, "step": 43202 }, { "epoch": 2.303300101295516, "grad_norm": 1.6249011754989624, "learning_rate": 1.3596e-06, "loss": 0.0166, "step": 43203 }, { "epoch": 2.3033534147251693, "grad_norm": 0.8241895437240601, "learning_rate": 1.3594e-06, "loss": 0.005, "step": 43204 }, { "epoch": 2.303406728154822, "grad_norm": 2.031442880630493, "learning_rate": 1.3592000000000003e-06, "loss": 0.0228, "step": 43205 }, { "epoch": 2.3034600415844753, "grad_norm": 1.5700973272323608, "learning_rate": 1.359e-06, "loss": 0.0206, "step": 43206 }, { "epoch": 2.303513355014128, "grad_norm": 0.7340115308761597, "learning_rate": 1.3588e-06, "loss": 0.0047, "step": 43207 }, { "epoch": 2.303566668443781, "grad_norm": 1.0173550844192505, "learning_rate": 1.3586000000000002e-06, "loss": 0.014, "step": 43208 }, { "epoch": 2.303619981873434, "grad_norm": 1.0603318214416504, "learning_rate": 1.3584e-06, "loss": 0.0164, "step": 43209 }, { "epoch": 2.3036732953030867, "grad_norm": 0.36927810311317444, "learning_rate": 1.3582e-06, "loss": 0.0016, "step": 43210 }, { "epoch": 2.30372660873274, "grad_norm": 2.8847670555114746, "learning_rate": 1.3580000000000002e-06, "loss": 0.0194, "step": 43211 }, { "epoch": 2.3037799221623927, "grad_norm": 1.5731916427612305, "learning_rate": 1.3578000000000002e-06, "loss": 0.0172, "step": 43212 }, { "epoch": 2.3038332355920454, "grad_norm": 4.9864349365234375, "learning_rate": 1.3576e-06, "loss": 0.0105, "step": 43213 }, { "epoch": 2.3038865490216986, "grad_norm": 0.6130604147911072, "learning_rate": 1.3574000000000001e-06, "loss": 0.0034, "step": 43214 }, { "epoch": 2.3039398624513514, "grad_norm": 0.8604447841644287, "learning_rate": 1.3572000000000001e-06, "loss": 0.0043, "step": 43215 }, { "epoch": 2.3039931758810046, "grad_norm": 2.6196858882904053, "learning_rate": 1.3569999999999999e-06, "loss": 0.0112, "step": 43216 }, { "epoch": 2.3040464893106574, "grad_norm": 0.17618782818317413, "learning_rate": 1.3568e-06, "loss": 0.0016, "step": 43217 }, { "epoch": 2.30409980274031, "grad_norm": 0.5980393886566162, "learning_rate": 1.3566e-06, "loss": 0.0039, "step": 43218 }, { "epoch": 2.3041531161699633, "grad_norm": 1.0115793943405151, "learning_rate": 1.3564000000000003e-06, "loss": 0.0066, "step": 43219 }, { "epoch": 2.304206429599616, "grad_norm": 0.5873084664344788, "learning_rate": 1.3562e-06, "loss": 0.0037, "step": 43220 }, { "epoch": 2.3042597430292693, "grad_norm": 2.2538819313049316, "learning_rate": 1.356e-06, "loss": 0.0136, "step": 43221 }, { "epoch": 2.304313056458922, "grad_norm": 2.4605143070220947, "learning_rate": 1.3558000000000002e-06, "loss": 0.0081, "step": 43222 }, { "epoch": 2.304366369888575, "grad_norm": 2.4741034507751465, "learning_rate": 1.3556000000000002e-06, "loss": 0.0307, "step": 43223 }, { "epoch": 2.304419683318228, "grad_norm": 0.48005151748657227, "learning_rate": 1.3554e-06, "loss": 0.0026, "step": 43224 }, { "epoch": 2.3044729967478808, "grad_norm": 1.4866715669631958, "learning_rate": 1.3552000000000002e-06, "loss": 0.023, "step": 43225 }, { "epoch": 2.3045263101775335, "grad_norm": 2.4368629455566406, "learning_rate": 1.3550000000000002e-06, "loss": 0.0168, "step": 43226 }, { "epoch": 2.3045796236071867, "grad_norm": 2.444730281829834, "learning_rate": 1.3548e-06, "loss": 0.0214, "step": 43227 }, { "epoch": 2.3046329370368395, "grad_norm": 2.0046017169952393, "learning_rate": 1.3546000000000001e-06, "loss": 0.021, "step": 43228 }, { "epoch": 2.3046862504664927, "grad_norm": 2.1718151569366455, "learning_rate": 1.3544000000000001e-06, "loss": 0.0195, "step": 43229 }, { "epoch": 2.3047395638961454, "grad_norm": 2.198667526245117, "learning_rate": 1.3542000000000003e-06, "loss": 0.0085, "step": 43230 }, { "epoch": 2.304792877325798, "grad_norm": 3.1101973056793213, "learning_rate": 1.354e-06, "loss": 0.0414, "step": 43231 }, { "epoch": 2.3048461907554514, "grad_norm": 1.8541889190673828, "learning_rate": 1.3538e-06, "loss": 0.046, "step": 43232 }, { "epoch": 2.304899504185104, "grad_norm": 1.0480151176452637, "learning_rate": 1.3536000000000003e-06, "loss": 0.0083, "step": 43233 }, { "epoch": 2.3049528176147573, "grad_norm": 1.6515848636627197, "learning_rate": 1.3534e-06, "loss": 0.0179, "step": 43234 }, { "epoch": 2.30500613104441, "grad_norm": 1.6468316316604614, "learning_rate": 1.3532e-06, "loss": 0.0095, "step": 43235 }, { "epoch": 2.305059444474063, "grad_norm": 1.9373328685760498, "learning_rate": 1.3530000000000002e-06, "loss": 0.0072, "step": 43236 }, { "epoch": 2.305112757903716, "grad_norm": 3.233595371246338, "learning_rate": 1.3528000000000002e-06, "loss": 0.052, "step": 43237 }, { "epoch": 2.305166071333369, "grad_norm": 1.6068525314331055, "learning_rate": 1.3526e-06, "loss": 0.008, "step": 43238 }, { "epoch": 2.3052193847630216, "grad_norm": 0.8875523805618286, "learning_rate": 1.3524000000000002e-06, "loss": 0.0038, "step": 43239 }, { "epoch": 2.3052726981926748, "grad_norm": 2.9073214530944824, "learning_rate": 1.3522000000000002e-06, "loss": 0.0228, "step": 43240 }, { "epoch": 2.3053260116223275, "grad_norm": 1.5102170705795288, "learning_rate": 1.352e-06, "loss": 0.0094, "step": 43241 }, { "epoch": 2.3053793250519807, "grad_norm": 1.3604881763458252, "learning_rate": 1.3518000000000001e-06, "loss": 0.0143, "step": 43242 }, { "epoch": 2.3054326384816335, "grad_norm": 2.6758203506469727, "learning_rate": 1.3516000000000001e-06, "loss": 0.0333, "step": 43243 }, { "epoch": 2.3054859519112867, "grad_norm": 1.5333137512207031, "learning_rate": 1.3514000000000003e-06, "loss": 0.0216, "step": 43244 }, { "epoch": 2.3055392653409394, "grad_norm": 0.7000307440757751, "learning_rate": 1.3512e-06, "loss": 0.0038, "step": 43245 }, { "epoch": 2.305592578770592, "grad_norm": 2.826630115509033, "learning_rate": 1.351e-06, "loss": 0.0221, "step": 43246 }, { "epoch": 2.3056458922002454, "grad_norm": 1.8896420001983643, "learning_rate": 1.3508000000000003e-06, "loss": 0.0088, "step": 43247 }, { "epoch": 2.305699205629898, "grad_norm": 3.2417423725128174, "learning_rate": 1.3506000000000002e-06, "loss": 0.029, "step": 43248 }, { "epoch": 2.305752519059551, "grad_norm": 1.0184866189956665, "learning_rate": 1.3504e-06, "loss": 0.0059, "step": 43249 }, { "epoch": 2.305805832489204, "grad_norm": 2.470048666000366, "learning_rate": 1.3502000000000002e-06, "loss": 0.0406, "step": 43250 }, { "epoch": 2.305859145918857, "grad_norm": 1.2708919048309326, "learning_rate": 1.3500000000000002e-06, "loss": 0.0251, "step": 43251 }, { "epoch": 2.30591245934851, "grad_norm": 1.1482852697372437, "learning_rate": 1.3498e-06, "loss": 0.0059, "step": 43252 }, { "epoch": 2.305965772778163, "grad_norm": 1.1643176078796387, "learning_rate": 1.3496000000000002e-06, "loss": 0.0466, "step": 43253 }, { "epoch": 2.3060190862078156, "grad_norm": 2.658064365386963, "learning_rate": 1.3494000000000002e-06, "loss": 0.0295, "step": 43254 }, { "epoch": 2.306072399637469, "grad_norm": 1.1415276527404785, "learning_rate": 1.3492000000000001e-06, "loss": 0.0121, "step": 43255 }, { "epoch": 2.3061257130671216, "grad_norm": 0.6879205703735352, "learning_rate": 1.3490000000000001e-06, "loss": 0.0037, "step": 43256 }, { "epoch": 2.3061790264967748, "grad_norm": 1.6898295879364014, "learning_rate": 1.3488000000000001e-06, "loss": 0.0332, "step": 43257 }, { "epoch": 2.3062323399264275, "grad_norm": 2.2068965435028076, "learning_rate": 1.3486e-06, "loss": 0.0235, "step": 43258 }, { "epoch": 2.3062856533560803, "grad_norm": 1.3872616291046143, "learning_rate": 1.3484e-06, "loss": 0.0054, "step": 43259 }, { "epoch": 2.3063389667857335, "grad_norm": 1.6917959451675415, "learning_rate": 1.3482e-06, "loss": 0.0077, "step": 43260 }, { "epoch": 2.3063922802153862, "grad_norm": 1.0044658184051514, "learning_rate": 1.348e-06, "loss": 0.0064, "step": 43261 }, { "epoch": 2.306445593645039, "grad_norm": 1.3220096826553345, "learning_rate": 1.3478000000000003e-06, "loss": 0.0105, "step": 43262 }, { "epoch": 2.306498907074692, "grad_norm": 1.5130127668380737, "learning_rate": 1.3476e-06, "loss": 0.0175, "step": 43263 }, { "epoch": 2.306552220504345, "grad_norm": 2.4035487174987793, "learning_rate": 1.3474e-06, "loss": 0.0082, "step": 43264 }, { "epoch": 2.306605533933998, "grad_norm": 2.6213717460632324, "learning_rate": 1.3472000000000002e-06, "loss": 0.0278, "step": 43265 }, { "epoch": 2.306658847363651, "grad_norm": 2.149839401245117, "learning_rate": 1.347e-06, "loss": 0.0245, "step": 43266 }, { "epoch": 2.3067121607933037, "grad_norm": 2.127084493637085, "learning_rate": 1.3468e-06, "loss": 0.0117, "step": 43267 }, { "epoch": 2.306765474222957, "grad_norm": 1.413090467453003, "learning_rate": 1.3466000000000002e-06, "loss": 0.0163, "step": 43268 }, { "epoch": 2.3068187876526096, "grad_norm": 1.8121806383132935, "learning_rate": 1.3464000000000001e-06, "loss": 0.0198, "step": 43269 }, { "epoch": 2.306872101082263, "grad_norm": 1.5430134534835815, "learning_rate": 1.3462e-06, "loss": 0.0064, "step": 43270 }, { "epoch": 2.3069254145119156, "grad_norm": 2.340404987335205, "learning_rate": 1.3460000000000001e-06, "loss": 0.0253, "step": 43271 }, { "epoch": 2.3069787279415683, "grad_norm": 1.3897161483764648, "learning_rate": 1.3458e-06, "loss": 0.0087, "step": 43272 }, { "epoch": 2.3070320413712215, "grad_norm": 3.3957717418670654, "learning_rate": 1.3456000000000003e-06, "loss": 0.0305, "step": 43273 }, { "epoch": 2.3070853548008743, "grad_norm": 2.8795928955078125, "learning_rate": 1.3454e-06, "loss": 0.0338, "step": 43274 }, { "epoch": 2.307138668230527, "grad_norm": 0.5011530518531799, "learning_rate": 1.3452e-06, "loss": 0.0044, "step": 43275 }, { "epoch": 2.3071919816601802, "grad_norm": 4.053744316101074, "learning_rate": 1.3450000000000003e-06, "loss": 0.0397, "step": 43276 }, { "epoch": 2.307245295089833, "grad_norm": 1.0723425149917603, "learning_rate": 1.3448e-06, "loss": 0.0158, "step": 43277 }, { "epoch": 2.307298608519486, "grad_norm": 3.5433223247528076, "learning_rate": 1.3446e-06, "loss": 0.0147, "step": 43278 }, { "epoch": 2.307351921949139, "grad_norm": 1.7201862335205078, "learning_rate": 1.3444000000000002e-06, "loss": 0.0127, "step": 43279 }, { "epoch": 2.307405235378792, "grad_norm": 1.968964695930481, "learning_rate": 1.3442000000000002e-06, "loss": 0.0055, "step": 43280 }, { "epoch": 2.307458548808445, "grad_norm": 2.458888292312622, "learning_rate": 1.344e-06, "loss": 0.0405, "step": 43281 }, { "epoch": 2.3075118622380977, "grad_norm": 2.1368401050567627, "learning_rate": 1.3438000000000002e-06, "loss": 0.0136, "step": 43282 }, { "epoch": 2.307565175667751, "grad_norm": 1.145608901977539, "learning_rate": 1.3436000000000001e-06, "loss": 0.0198, "step": 43283 }, { "epoch": 2.3076184890974036, "grad_norm": 2.181893825531006, "learning_rate": 1.3434e-06, "loss": 0.0166, "step": 43284 }, { "epoch": 2.3076718025270564, "grad_norm": 2.5153167247772217, "learning_rate": 1.3432000000000001e-06, "loss": 0.0341, "step": 43285 }, { "epoch": 2.3077251159567096, "grad_norm": 3.5151076316833496, "learning_rate": 1.343e-06, "loss": 0.0392, "step": 43286 }, { "epoch": 2.3077784293863624, "grad_norm": 1.2076395750045776, "learning_rate": 1.3428000000000003e-06, "loss": 0.0094, "step": 43287 }, { "epoch": 2.307831742816015, "grad_norm": 1.6798049211502075, "learning_rate": 1.3426e-06, "loss": 0.0127, "step": 43288 }, { "epoch": 2.3078850562456683, "grad_norm": 3.245612382888794, "learning_rate": 1.3424e-06, "loss": 0.0559, "step": 43289 }, { "epoch": 2.307938369675321, "grad_norm": 0.3706546425819397, "learning_rate": 1.3422000000000003e-06, "loss": 0.0025, "step": 43290 }, { "epoch": 2.3079916831049743, "grad_norm": 1.568099021911621, "learning_rate": 1.3420000000000002e-06, "loss": 0.0278, "step": 43291 }, { "epoch": 2.308044996534627, "grad_norm": 2.9585258960723877, "learning_rate": 1.3418e-06, "loss": 0.0345, "step": 43292 }, { "epoch": 2.3080983099642802, "grad_norm": 0.7953624725341797, "learning_rate": 1.3416000000000002e-06, "loss": 0.004, "step": 43293 }, { "epoch": 2.308151623393933, "grad_norm": 1.2467999458312988, "learning_rate": 1.3414000000000002e-06, "loss": 0.0197, "step": 43294 }, { "epoch": 2.3082049368235857, "grad_norm": 2.56150484085083, "learning_rate": 1.3412e-06, "loss": 0.0197, "step": 43295 }, { "epoch": 2.308258250253239, "grad_norm": 2.0454273223876953, "learning_rate": 1.3410000000000002e-06, "loss": 0.0111, "step": 43296 }, { "epoch": 2.3083115636828917, "grad_norm": 2.6378896236419678, "learning_rate": 1.3408000000000001e-06, "loss": 0.0109, "step": 43297 }, { "epoch": 2.3083648771125445, "grad_norm": 1.9960814714431763, "learning_rate": 1.3406000000000001e-06, "loss": 0.0156, "step": 43298 }, { "epoch": 2.3084181905421977, "grad_norm": 3.1742122173309326, "learning_rate": 1.3404000000000001e-06, "loss": 0.0376, "step": 43299 }, { "epoch": 2.3084715039718504, "grad_norm": 4.412993431091309, "learning_rate": 1.3402e-06, "loss": 0.0257, "step": 43300 }, { "epoch": 2.3085248174015036, "grad_norm": 2.4888484477996826, "learning_rate": 1.34e-06, "loss": 0.0449, "step": 43301 }, { "epoch": 2.3085781308311564, "grad_norm": 2.748901128768921, "learning_rate": 1.3398e-06, "loss": 0.0244, "step": 43302 }, { "epoch": 2.308631444260809, "grad_norm": 1.7816413640975952, "learning_rate": 1.3396e-06, "loss": 0.0114, "step": 43303 }, { "epoch": 2.3086847576904623, "grad_norm": 1.5137585401535034, "learning_rate": 1.3394e-06, "loss": 0.0089, "step": 43304 }, { "epoch": 2.308738071120115, "grad_norm": 3.954117774963379, "learning_rate": 1.3392000000000002e-06, "loss": 0.0425, "step": 43305 }, { "epoch": 2.3087913845497683, "grad_norm": 1.2247592210769653, "learning_rate": 1.339e-06, "loss": 0.022, "step": 43306 }, { "epoch": 2.308844697979421, "grad_norm": 2.1094820499420166, "learning_rate": 1.3388e-06, "loss": 0.0274, "step": 43307 }, { "epoch": 2.308898011409074, "grad_norm": 2.805006742477417, "learning_rate": 1.3386000000000002e-06, "loss": 0.0381, "step": 43308 }, { "epoch": 2.308951324838727, "grad_norm": 0.8840765357017517, "learning_rate": 1.3384e-06, "loss": 0.0058, "step": 43309 }, { "epoch": 2.3090046382683798, "grad_norm": 1.9681634902954102, "learning_rate": 1.3382e-06, "loss": 0.0067, "step": 43310 }, { "epoch": 2.3090579516980325, "grad_norm": 1.0328134298324585, "learning_rate": 1.3380000000000001e-06, "loss": 0.0185, "step": 43311 }, { "epoch": 2.3091112651276857, "grad_norm": 1.2628570795059204, "learning_rate": 1.3378000000000001e-06, "loss": 0.0327, "step": 43312 }, { "epoch": 2.3091645785573385, "grad_norm": 2.261094808578491, "learning_rate": 1.3376e-06, "loss": 0.0149, "step": 43313 }, { "epoch": 2.3092178919869917, "grad_norm": 2.141845941543579, "learning_rate": 1.3374e-06, "loss": 0.0361, "step": 43314 }, { "epoch": 2.3092712054166444, "grad_norm": 1.430497646331787, "learning_rate": 1.3372e-06, "loss": 0.0133, "step": 43315 }, { "epoch": 2.309324518846297, "grad_norm": 1.3035310506820679, "learning_rate": 1.3370000000000003e-06, "loss": 0.0161, "step": 43316 }, { "epoch": 2.3093778322759504, "grad_norm": 3.6935243606567383, "learning_rate": 1.3368e-06, "loss": 0.0325, "step": 43317 }, { "epoch": 2.309431145705603, "grad_norm": 2.377488136291504, "learning_rate": 1.3366e-06, "loss": 0.0094, "step": 43318 }, { "epoch": 2.3094844591352564, "grad_norm": 1.8301500082015991, "learning_rate": 1.3364000000000002e-06, "loss": 0.0101, "step": 43319 }, { "epoch": 2.309537772564909, "grad_norm": 1.8207448720932007, "learning_rate": 1.3362e-06, "loss": 0.0177, "step": 43320 }, { "epoch": 2.309591085994562, "grad_norm": 1.4742106199264526, "learning_rate": 1.336e-06, "loss": 0.0219, "step": 43321 }, { "epoch": 2.309644399424215, "grad_norm": 3.4720280170440674, "learning_rate": 1.3358000000000002e-06, "loss": 0.0207, "step": 43322 }, { "epoch": 2.309697712853868, "grad_norm": 0.7615867853164673, "learning_rate": 1.3356000000000002e-06, "loss": 0.0046, "step": 43323 }, { "epoch": 2.3097510262835206, "grad_norm": 1.3870601654052734, "learning_rate": 1.3354e-06, "loss": 0.0173, "step": 43324 }, { "epoch": 2.309804339713174, "grad_norm": 2.0444436073303223, "learning_rate": 1.3352000000000002e-06, "loss": 0.0151, "step": 43325 }, { "epoch": 2.3098576531428265, "grad_norm": 0.9655904769897461, "learning_rate": 1.3350000000000001e-06, "loss": 0.0053, "step": 43326 }, { "epoch": 2.3099109665724797, "grad_norm": 1.9752837419509888, "learning_rate": 1.3348e-06, "loss": 0.0129, "step": 43327 }, { "epoch": 2.3099642800021325, "grad_norm": 1.8261816501617432, "learning_rate": 1.3346e-06, "loss": 0.0242, "step": 43328 }, { "epoch": 2.3100175934317857, "grad_norm": 1.9102507829666138, "learning_rate": 1.3344e-06, "loss": 0.0333, "step": 43329 }, { "epoch": 2.3100709068614385, "grad_norm": 1.2047549486160278, "learning_rate": 1.3342000000000003e-06, "loss": 0.0176, "step": 43330 }, { "epoch": 2.310124220291091, "grad_norm": 2.2964439392089844, "learning_rate": 1.334e-06, "loss": 0.0278, "step": 43331 }, { "epoch": 2.3101775337207444, "grad_norm": 2.791846990585327, "learning_rate": 1.3338e-06, "loss": 0.0268, "step": 43332 }, { "epoch": 2.310230847150397, "grad_norm": 0.6171053051948547, "learning_rate": 1.3336000000000002e-06, "loss": 0.0043, "step": 43333 }, { "epoch": 2.31028416058005, "grad_norm": 1.7265678644180298, "learning_rate": 1.3334e-06, "loss": 0.0328, "step": 43334 }, { "epoch": 2.310337474009703, "grad_norm": 2.6306238174438477, "learning_rate": 1.3332e-06, "loss": 0.0153, "step": 43335 }, { "epoch": 2.310390787439356, "grad_norm": 2.091200113296509, "learning_rate": 1.3330000000000002e-06, "loss": 0.0099, "step": 43336 }, { "epoch": 2.310444100869009, "grad_norm": 1.239433765411377, "learning_rate": 1.3328000000000002e-06, "loss": 0.0085, "step": 43337 }, { "epoch": 2.310497414298662, "grad_norm": 6.044960975646973, "learning_rate": 1.3326e-06, "loss": 0.0223, "step": 43338 }, { "epoch": 2.3105507277283146, "grad_norm": 1.325769066810608, "learning_rate": 1.3324000000000002e-06, "loss": 0.0183, "step": 43339 }, { "epoch": 2.310604041157968, "grad_norm": 2.6540744304656982, "learning_rate": 1.3322000000000001e-06, "loss": 0.0696, "step": 43340 }, { "epoch": 2.3106573545876206, "grad_norm": 1.3449476957321167, "learning_rate": 1.3320000000000003e-06, "loss": 0.0175, "step": 43341 }, { "epoch": 2.3107106680172738, "grad_norm": 1.6003670692443848, "learning_rate": 1.3318000000000001e-06, "loss": 0.0208, "step": 43342 }, { "epoch": 2.3107639814469265, "grad_norm": 2.10005521774292, "learning_rate": 1.3316e-06, "loss": 0.0144, "step": 43343 }, { "epoch": 2.3108172948765793, "grad_norm": 2.7627360820770264, "learning_rate": 1.3314000000000003e-06, "loss": 0.0357, "step": 43344 }, { "epoch": 2.3108706083062325, "grad_norm": 4.8705902099609375, "learning_rate": 1.3312e-06, "loss": 0.0345, "step": 43345 }, { "epoch": 2.3109239217358852, "grad_norm": 1.4234992265701294, "learning_rate": 1.331e-06, "loss": 0.026, "step": 43346 }, { "epoch": 2.310977235165538, "grad_norm": 1.1356867551803589, "learning_rate": 1.3308000000000002e-06, "loss": 0.0083, "step": 43347 }, { "epoch": 2.311030548595191, "grad_norm": 3.814155101776123, "learning_rate": 1.3306000000000002e-06, "loss": 0.0282, "step": 43348 }, { "epoch": 2.311083862024844, "grad_norm": 2.9928138256073, "learning_rate": 1.3304e-06, "loss": 0.0137, "step": 43349 }, { "epoch": 2.311137175454497, "grad_norm": 1.7313027381896973, "learning_rate": 1.3302000000000002e-06, "loss": 0.0254, "step": 43350 }, { "epoch": 2.31119048888415, "grad_norm": 3.720046043395996, "learning_rate": 1.3300000000000002e-06, "loss": 0.0216, "step": 43351 }, { "epoch": 2.3112438023138027, "grad_norm": 3.473422050476074, "learning_rate": 1.3298e-06, "loss": 0.0361, "step": 43352 }, { "epoch": 2.311297115743456, "grad_norm": 1.673012137413025, "learning_rate": 1.3296000000000002e-06, "loss": 0.0138, "step": 43353 }, { "epoch": 2.3113504291731086, "grad_norm": 1.9911394119262695, "learning_rate": 1.3294000000000001e-06, "loss": 0.0221, "step": 43354 }, { "epoch": 2.311403742602762, "grad_norm": 1.9611884355545044, "learning_rate": 1.3292000000000001e-06, "loss": 0.0333, "step": 43355 }, { "epoch": 2.3114570560324146, "grad_norm": 1.4857721328735352, "learning_rate": 1.3290000000000001e-06, "loss": 0.0246, "step": 43356 }, { "epoch": 2.3115103694620673, "grad_norm": 0.719785213470459, "learning_rate": 1.3288e-06, "loss": 0.0051, "step": 43357 }, { "epoch": 2.3115636828917205, "grad_norm": 1.353447675704956, "learning_rate": 1.3286e-06, "loss": 0.0187, "step": 43358 }, { "epoch": 2.3116169963213733, "grad_norm": 1.7494512796401978, "learning_rate": 1.3284000000000003e-06, "loss": 0.0293, "step": 43359 }, { "epoch": 2.311670309751026, "grad_norm": 3.369168758392334, "learning_rate": 1.3282e-06, "loss": 0.0171, "step": 43360 }, { "epoch": 2.3117236231806793, "grad_norm": 3.016803503036499, "learning_rate": 1.328e-06, "loss": 0.0116, "step": 43361 }, { "epoch": 2.311776936610332, "grad_norm": 1.0379822254180908, "learning_rate": 1.3278000000000002e-06, "loss": 0.0156, "step": 43362 }, { "epoch": 2.311830250039985, "grad_norm": 2.0526630878448486, "learning_rate": 1.3276e-06, "loss": 0.0252, "step": 43363 }, { "epoch": 2.311883563469638, "grad_norm": 0.6717032194137573, "learning_rate": 1.3274e-06, "loss": 0.0042, "step": 43364 }, { "epoch": 2.311936876899291, "grad_norm": 0.9251124262809753, "learning_rate": 1.3272000000000002e-06, "loss": 0.0054, "step": 43365 }, { "epoch": 2.311990190328944, "grad_norm": 2.4693045616149902, "learning_rate": 1.3270000000000002e-06, "loss": 0.0319, "step": 43366 }, { "epoch": 2.3120435037585967, "grad_norm": 1.3643872737884521, "learning_rate": 1.3268e-06, "loss": 0.0146, "step": 43367 }, { "epoch": 2.31209681718825, "grad_norm": 1.3352361917495728, "learning_rate": 1.3266000000000001e-06, "loss": 0.0217, "step": 43368 }, { "epoch": 2.3121501306179026, "grad_norm": 1.7322291135787964, "learning_rate": 1.3264000000000001e-06, "loss": 0.0092, "step": 43369 }, { "epoch": 2.3122034440475554, "grad_norm": 1.1067224740982056, "learning_rate": 1.3262e-06, "loss": 0.0145, "step": 43370 }, { "epoch": 2.3122567574772086, "grad_norm": 2.9676849842071533, "learning_rate": 1.326e-06, "loss": 0.0139, "step": 43371 }, { "epoch": 2.3123100709068614, "grad_norm": 2.149998426437378, "learning_rate": 1.3258e-06, "loss": 0.0387, "step": 43372 }, { "epoch": 2.3123633843365146, "grad_norm": 1.150708556175232, "learning_rate": 1.3256000000000003e-06, "loss": 0.0211, "step": 43373 }, { "epoch": 2.3124166977661673, "grad_norm": 0.7812538743019104, "learning_rate": 1.3254e-06, "loss": 0.0057, "step": 43374 }, { "epoch": 2.31247001119582, "grad_norm": 1.6881389617919922, "learning_rate": 1.3252e-06, "loss": 0.0117, "step": 43375 }, { "epoch": 2.3125233246254733, "grad_norm": 1.6914604902267456, "learning_rate": 1.3250000000000002e-06, "loss": 0.0081, "step": 43376 }, { "epoch": 2.312576638055126, "grad_norm": 1.725168228149414, "learning_rate": 1.3248e-06, "loss": 0.0087, "step": 43377 }, { "epoch": 2.3126299514847792, "grad_norm": 1.9224344491958618, "learning_rate": 1.3246e-06, "loss": 0.0209, "step": 43378 }, { "epoch": 2.312683264914432, "grad_norm": 0.5787254571914673, "learning_rate": 1.3244000000000002e-06, "loss": 0.0047, "step": 43379 }, { "epoch": 2.3127365783440847, "grad_norm": 3.476780652999878, "learning_rate": 1.3242000000000002e-06, "loss": 0.0479, "step": 43380 }, { "epoch": 2.312789891773738, "grad_norm": 1.7103379964828491, "learning_rate": 1.324e-06, "loss": 0.0089, "step": 43381 }, { "epoch": 2.3128432052033907, "grad_norm": 1.9274611473083496, "learning_rate": 1.3238000000000001e-06, "loss": 0.0124, "step": 43382 }, { "epoch": 2.3128965186330435, "grad_norm": 2.3693296909332275, "learning_rate": 1.3236000000000001e-06, "loss": 0.021, "step": 43383 }, { "epoch": 2.3129498320626967, "grad_norm": 1.224789023399353, "learning_rate": 1.3234000000000003e-06, "loss": 0.0342, "step": 43384 }, { "epoch": 2.3130031454923494, "grad_norm": 3.4210946559906006, "learning_rate": 1.3232e-06, "loss": 0.0153, "step": 43385 }, { "epoch": 2.3130564589220026, "grad_norm": 3.1755363941192627, "learning_rate": 1.323e-06, "loss": 0.0193, "step": 43386 }, { "epoch": 2.3131097723516554, "grad_norm": 1.7925225496292114, "learning_rate": 1.3228000000000003e-06, "loss": 0.0171, "step": 43387 }, { "epoch": 2.313163085781308, "grad_norm": 2.059678316116333, "learning_rate": 1.3226e-06, "loss": 0.0219, "step": 43388 }, { "epoch": 2.3132163992109613, "grad_norm": 6.607131004333496, "learning_rate": 1.3224e-06, "loss": 0.0257, "step": 43389 }, { "epoch": 2.313269712640614, "grad_norm": 1.133270025253296, "learning_rate": 1.3222000000000002e-06, "loss": 0.005, "step": 43390 }, { "epoch": 2.3133230260702673, "grad_norm": 2.399017572402954, "learning_rate": 1.3220000000000002e-06, "loss": 0.0168, "step": 43391 }, { "epoch": 2.31337633949992, "grad_norm": 2.9284703731536865, "learning_rate": 1.3218e-06, "loss": 0.0265, "step": 43392 }, { "epoch": 2.313429652929573, "grad_norm": 0.1975877732038498, "learning_rate": 1.3216000000000002e-06, "loss": 0.0017, "step": 43393 }, { "epoch": 2.313482966359226, "grad_norm": 0.41552668809890747, "learning_rate": 1.3214000000000002e-06, "loss": 0.0047, "step": 43394 }, { "epoch": 2.3135362797888788, "grad_norm": 1.7832518815994263, "learning_rate": 1.3212e-06, "loss": 0.0057, "step": 43395 }, { "epoch": 2.3135895932185315, "grad_norm": 1.6564364433288574, "learning_rate": 1.3210000000000001e-06, "loss": 0.0225, "step": 43396 }, { "epoch": 2.3136429066481847, "grad_norm": 2.448089599609375, "learning_rate": 1.3208000000000001e-06, "loss": 0.0123, "step": 43397 }, { "epoch": 2.3136962200778375, "grad_norm": 1.187282681465149, "learning_rate": 1.3206000000000001e-06, "loss": 0.0106, "step": 43398 }, { "epoch": 2.3137495335074907, "grad_norm": 3.32045316696167, "learning_rate": 1.3204e-06, "loss": 0.0107, "step": 43399 }, { "epoch": 2.3138028469371434, "grad_norm": 0.9727886915206909, "learning_rate": 1.3202e-06, "loss": 0.0077, "step": 43400 }, { "epoch": 2.3138561603667966, "grad_norm": 1.176596760749817, "learning_rate": 1.32e-06, "loss": 0.0059, "step": 43401 }, { "epoch": 2.3139094737964494, "grad_norm": 1.6444499492645264, "learning_rate": 1.3198e-06, "loss": 0.018, "step": 43402 }, { "epoch": 2.313962787226102, "grad_norm": 1.8096897602081299, "learning_rate": 1.3196e-06, "loss": 0.0231, "step": 43403 }, { "epoch": 2.3140161006557554, "grad_norm": 1.1555540561676025, "learning_rate": 1.3194e-06, "loss": 0.0055, "step": 43404 }, { "epoch": 2.314069414085408, "grad_norm": 0.5718342065811157, "learning_rate": 1.3192000000000002e-06, "loss": 0.0044, "step": 43405 }, { "epoch": 2.314122727515061, "grad_norm": 1.0038851499557495, "learning_rate": 1.319e-06, "loss": 0.0057, "step": 43406 }, { "epoch": 2.314176040944714, "grad_norm": 1.019069790840149, "learning_rate": 1.3188e-06, "loss": 0.0062, "step": 43407 }, { "epoch": 2.314229354374367, "grad_norm": 2.071485757827759, "learning_rate": 1.3186000000000002e-06, "loss": 0.0096, "step": 43408 }, { "epoch": 2.3142826678040196, "grad_norm": 0.850874662399292, "learning_rate": 1.3184000000000002e-06, "loss": 0.0057, "step": 43409 }, { "epoch": 2.314335981233673, "grad_norm": 2.6122450828552246, "learning_rate": 1.3182e-06, "loss": 0.0431, "step": 43410 }, { "epoch": 2.3143892946633255, "grad_norm": 2.2748842239379883, "learning_rate": 1.3180000000000001e-06, "loss": 0.0404, "step": 43411 }, { "epoch": 2.3144426080929787, "grad_norm": 1.6579279899597168, "learning_rate": 1.3178000000000001e-06, "loss": 0.0086, "step": 43412 }, { "epoch": 2.3144959215226315, "grad_norm": 2.765503168106079, "learning_rate": 1.3175999999999999e-06, "loss": 0.0325, "step": 43413 }, { "epoch": 2.3145492349522847, "grad_norm": 1.7518401145935059, "learning_rate": 1.3174e-06, "loss": 0.0062, "step": 43414 }, { "epoch": 2.3146025483819375, "grad_norm": 0.1969212144613266, "learning_rate": 1.3172e-06, "loss": 0.0021, "step": 43415 }, { "epoch": 2.31465586181159, "grad_norm": 2.4144763946533203, "learning_rate": 1.3170000000000003e-06, "loss": 0.0175, "step": 43416 }, { "epoch": 2.3147091752412434, "grad_norm": 0.8616211414337158, "learning_rate": 1.3168e-06, "loss": 0.0038, "step": 43417 }, { "epoch": 2.314762488670896, "grad_norm": 3.798168182373047, "learning_rate": 1.3166e-06, "loss": 0.0499, "step": 43418 }, { "epoch": 2.314815802100549, "grad_norm": 0.8173025846481323, "learning_rate": 1.3164000000000002e-06, "loss": 0.0044, "step": 43419 }, { "epoch": 2.314869115530202, "grad_norm": 1.4493671655654907, "learning_rate": 1.3162e-06, "loss": 0.0125, "step": 43420 }, { "epoch": 2.314922428959855, "grad_norm": 1.1455146074295044, "learning_rate": 1.316e-06, "loss": 0.0055, "step": 43421 }, { "epoch": 2.314975742389508, "grad_norm": 2.975301742553711, "learning_rate": 1.3158000000000002e-06, "loss": 0.0125, "step": 43422 }, { "epoch": 2.315029055819161, "grad_norm": 1.5227687358856201, "learning_rate": 1.3156000000000002e-06, "loss": 0.0094, "step": 43423 }, { "epoch": 2.3150823692488136, "grad_norm": 1.725144863128662, "learning_rate": 1.3154e-06, "loss": 0.0105, "step": 43424 }, { "epoch": 2.315135682678467, "grad_norm": 2.487076997756958, "learning_rate": 1.3152000000000001e-06, "loss": 0.0317, "step": 43425 }, { "epoch": 2.3151889961081196, "grad_norm": 1.6403117179870605, "learning_rate": 1.3150000000000001e-06, "loss": 0.0157, "step": 43426 }, { "epoch": 2.3152423095377728, "grad_norm": 2.823413610458374, "learning_rate": 1.3148000000000003e-06, "loss": 0.014, "step": 43427 }, { "epoch": 2.3152956229674255, "grad_norm": 2.255488872528076, "learning_rate": 1.3146e-06, "loss": 0.0134, "step": 43428 }, { "epoch": 2.3153489363970783, "grad_norm": 1.5450056791305542, "learning_rate": 1.3144e-06, "loss": 0.0046, "step": 43429 }, { "epoch": 2.3154022498267315, "grad_norm": 1.4201141595840454, "learning_rate": 1.3142000000000003e-06, "loss": 0.0056, "step": 43430 }, { "epoch": 2.3154555632563842, "grad_norm": 2.74615740776062, "learning_rate": 1.314e-06, "loss": 0.0256, "step": 43431 }, { "epoch": 2.315508876686037, "grad_norm": 1.7640162706375122, "learning_rate": 1.3138e-06, "loss": 0.0098, "step": 43432 }, { "epoch": 2.31556219011569, "grad_norm": 1.9356127977371216, "learning_rate": 1.3136000000000002e-06, "loss": 0.0388, "step": 43433 }, { "epoch": 2.315615503545343, "grad_norm": 2.509514808654785, "learning_rate": 1.3134000000000002e-06, "loss": 0.0109, "step": 43434 }, { "epoch": 2.315668816974996, "grad_norm": 0.8647364974021912, "learning_rate": 1.3132e-06, "loss": 0.0052, "step": 43435 }, { "epoch": 2.315722130404649, "grad_norm": 2.0425572395324707, "learning_rate": 1.3130000000000002e-06, "loss": 0.0129, "step": 43436 }, { "epoch": 2.3157754438343017, "grad_norm": 6.5009846687316895, "learning_rate": 1.3128000000000002e-06, "loss": 0.0251, "step": 43437 }, { "epoch": 2.315828757263955, "grad_norm": 0.424228698015213, "learning_rate": 1.3126e-06, "loss": 0.0036, "step": 43438 }, { "epoch": 2.3158820706936076, "grad_norm": 2.5192019939422607, "learning_rate": 1.3124000000000001e-06, "loss": 0.0207, "step": 43439 }, { "epoch": 2.315935384123261, "grad_norm": 0.30933189392089844, "learning_rate": 1.3122000000000001e-06, "loss": 0.0023, "step": 43440 }, { "epoch": 2.3159886975529136, "grad_norm": 2.3223462104797363, "learning_rate": 1.3120000000000003e-06, "loss": 0.0154, "step": 43441 }, { "epoch": 2.3160420109825663, "grad_norm": 2.000448226928711, "learning_rate": 1.3118e-06, "loss": 0.0181, "step": 43442 }, { "epoch": 2.3160953244122195, "grad_norm": 1.8315562009811401, "learning_rate": 1.3116e-06, "loss": 0.0386, "step": 43443 }, { "epoch": 2.3161486378418723, "grad_norm": 0.5518746972084045, "learning_rate": 1.3114000000000003e-06, "loss": 0.0041, "step": 43444 }, { "epoch": 2.316201951271525, "grad_norm": 1.8696274757385254, "learning_rate": 1.3112e-06, "loss": 0.0257, "step": 43445 }, { "epoch": 2.3162552647011783, "grad_norm": 0.49888452887535095, "learning_rate": 1.311e-06, "loss": 0.0028, "step": 43446 }, { "epoch": 2.316308578130831, "grad_norm": 1.4575523138046265, "learning_rate": 1.3108000000000002e-06, "loss": 0.0076, "step": 43447 }, { "epoch": 2.316361891560484, "grad_norm": 0.27991238236427307, "learning_rate": 1.3106000000000002e-06, "loss": 0.0023, "step": 43448 }, { "epoch": 2.316415204990137, "grad_norm": 1.654538869857788, "learning_rate": 1.3104e-06, "loss": 0.0203, "step": 43449 }, { "epoch": 2.31646851841979, "grad_norm": 3.6671078205108643, "learning_rate": 1.3102000000000002e-06, "loss": 0.0351, "step": 43450 }, { "epoch": 2.316521831849443, "grad_norm": 2.7677559852600098, "learning_rate": 1.3100000000000002e-06, "loss": 0.0105, "step": 43451 }, { "epoch": 2.3165751452790957, "grad_norm": 2.227365016937256, "learning_rate": 1.3098000000000001e-06, "loss": 0.0123, "step": 43452 }, { "epoch": 2.316628458708749, "grad_norm": 3.3391990661621094, "learning_rate": 1.3096000000000001e-06, "loss": 0.0373, "step": 43453 }, { "epoch": 2.3166817721384017, "grad_norm": 1.7081387042999268, "learning_rate": 1.3094000000000001e-06, "loss": 0.0101, "step": 43454 }, { "epoch": 2.3167350855680544, "grad_norm": 1.9527788162231445, "learning_rate": 1.3092e-06, "loss": 0.0167, "step": 43455 }, { "epoch": 2.3167883989977076, "grad_norm": 0.6511872410774231, "learning_rate": 1.309e-06, "loss": 0.0048, "step": 43456 }, { "epoch": 2.3168417124273604, "grad_norm": 1.2163846492767334, "learning_rate": 1.3088e-06, "loss": 0.017, "step": 43457 }, { "epoch": 2.3168950258570136, "grad_norm": 2.0298914909362793, "learning_rate": 1.3086e-06, "loss": 0.0145, "step": 43458 }, { "epoch": 2.3169483392866663, "grad_norm": 1.5201081037521362, "learning_rate": 1.3084000000000003e-06, "loss": 0.0141, "step": 43459 }, { "epoch": 2.317001652716319, "grad_norm": 1.4107121229171753, "learning_rate": 1.3082e-06, "loss": 0.0122, "step": 43460 }, { "epoch": 2.3170549661459723, "grad_norm": 3.6189334392547607, "learning_rate": 1.308e-06, "loss": 0.0114, "step": 43461 }, { "epoch": 2.317108279575625, "grad_norm": 2.611156940460205, "learning_rate": 1.3078000000000002e-06, "loss": 0.0224, "step": 43462 }, { "epoch": 2.3171615930052782, "grad_norm": 1.554909110069275, "learning_rate": 1.3076e-06, "loss": 0.0248, "step": 43463 }, { "epoch": 2.317214906434931, "grad_norm": 1.0156117677688599, "learning_rate": 1.3074e-06, "loss": 0.0196, "step": 43464 }, { "epoch": 2.3172682198645838, "grad_norm": 1.5910731554031372, "learning_rate": 1.3072000000000002e-06, "loss": 0.0217, "step": 43465 }, { "epoch": 2.317321533294237, "grad_norm": 1.5244537591934204, "learning_rate": 1.3070000000000001e-06, "loss": 0.0205, "step": 43466 }, { "epoch": 2.3173748467238897, "grad_norm": 1.0072382688522339, "learning_rate": 1.3068e-06, "loss": 0.0076, "step": 43467 }, { "epoch": 2.3174281601535425, "grad_norm": 0.8375078439712524, "learning_rate": 1.3066000000000001e-06, "loss": 0.0052, "step": 43468 }, { "epoch": 2.3174814735831957, "grad_norm": 1.6528120040893555, "learning_rate": 1.3064e-06, "loss": 0.009, "step": 43469 }, { "epoch": 2.3175347870128484, "grad_norm": 1.7742223739624023, "learning_rate": 1.3062000000000003e-06, "loss": 0.0069, "step": 43470 }, { "epoch": 2.3175881004425016, "grad_norm": 2.338059186935425, "learning_rate": 1.306e-06, "loss": 0.0186, "step": 43471 }, { "epoch": 2.3176414138721544, "grad_norm": 0.8610942363739014, "learning_rate": 1.3058e-06, "loss": 0.005, "step": 43472 }, { "epoch": 2.317694727301807, "grad_norm": 2.3647451400756836, "learning_rate": 1.3056000000000003e-06, "loss": 0.0273, "step": 43473 }, { "epoch": 2.3177480407314603, "grad_norm": 0.774143397808075, "learning_rate": 1.3054e-06, "loss": 0.0031, "step": 43474 }, { "epoch": 2.317801354161113, "grad_norm": 1.133844017982483, "learning_rate": 1.3052e-06, "loss": 0.0166, "step": 43475 }, { "epoch": 2.3178546675907663, "grad_norm": 2.16607666015625, "learning_rate": 1.3050000000000002e-06, "loss": 0.0084, "step": 43476 }, { "epoch": 2.317907981020419, "grad_norm": 2.104233980178833, "learning_rate": 1.3048000000000002e-06, "loss": 0.0181, "step": 43477 }, { "epoch": 2.317961294450072, "grad_norm": 3.0140292644500732, "learning_rate": 1.3046e-06, "loss": 0.0372, "step": 43478 }, { "epoch": 2.318014607879725, "grad_norm": 2.193704605102539, "learning_rate": 1.3044000000000002e-06, "loss": 0.0126, "step": 43479 }, { "epoch": 2.3180679213093778, "grad_norm": 1.4705379009246826, "learning_rate": 1.3042000000000002e-06, "loss": 0.0157, "step": 43480 }, { "epoch": 2.3181212347390305, "grad_norm": 0.30396437644958496, "learning_rate": 1.304e-06, "loss": 0.0022, "step": 43481 }, { "epoch": 2.3181745481686837, "grad_norm": 2.1070172786712646, "learning_rate": 1.3038000000000001e-06, "loss": 0.0224, "step": 43482 }, { "epoch": 2.3182278615983365, "grad_norm": 2.6666386127471924, "learning_rate": 1.3036e-06, "loss": 0.0189, "step": 43483 }, { "epoch": 2.3182811750279897, "grad_norm": 1.5939486026763916, "learning_rate": 1.3034000000000003e-06, "loss": 0.015, "step": 43484 }, { "epoch": 2.3183344884576425, "grad_norm": 1.8232899904251099, "learning_rate": 1.3032e-06, "loss": 0.009, "step": 43485 }, { "epoch": 2.3183878018872957, "grad_norm": 1.4294036626815796, "learning_rate": 1.303e-06, "loss": 0.0084, "step": 43486 }, { "epoch": 2.3184411153169484, "grad_norm": 1.7746425867080688, "learning_rate": 1.3028000000000003e-06, "loss": 0.0168, "step": 43487 }, { "epoch": 2.318494428746601, "grad_norm": 0.7189205884933472, "learning_rate": 1.3026e-06, "loss": 0.0024, "step": 43488 }, { "epoch": 2.3185477421762544, "grad_norm": 1.7567449808120728, "learning_rate": 1.3024e-06, "loss": 0.0099, "step": 43489 }, { "epoch": 2.318601055605907, "grad_norm": 1.4731532335281372, "learning_rate": 1.3022000000000002e-06, "loss": 0.0133, "step": 43490 }, { "epoch": 2.31865436903556, "grad_norm": 1.0918782949447632, "learning_rate": 1.3020000000000002e-06, "loss": 0.0165, "step": 43491 }, { "epoch": 2.318707682465213, "grad_norm": 3.1790711879730225, "learning_rate": 1.3018e-06, "loss": 0.0358, "step": 43492 }, { "epoch": 2.318760995894866, "grad_norm": 0.6469853520393372, "learning_rate": 1.3016000000000002e-06, "loss": 0.0022, "step": 43493 }, { "epoch": 2.3188143093245186, "grad_norm": 2.9550163745880127, "learning_rate": 1.3014000000000002e-06, "loss": 0.0218, "step": 43494 }, { "epoch": 2.318867622754172, "grad_norm": 1.576723575592041, "learning_rate": 1.3012000000000001e-06, "loss": 0.0186, "step": 43495 }, { "epoch": 2.3189209361838246, "grad_norm": 2.226588487625122, "learning_rate": 1.3010000000000001e-06, "loss": 0.011, "step": 43496 }, { "epoch": 2.3189742496134778, "grad_norm": 1.6085410118103027, "learning_rate": 1.3008000000000001e-06, "loss": 0.0158, "step": 43497 }, { "epoch": 2.3190275630431305, "grad_norm": 4.351844310760498, "learning_rate": 1.3006e-06, "loss": 0.0105, "step": 43498 }, { "epoch": 2.3190808764727837, "grad_norm": 3.2939369678497314, "learning_rate": 1.3004e-06, "loss": 0.0089, "step": 43499 }, { "epoch": 2.3191341899024365, "grad_norm": 3.042743444442749, "learning_rate": 1.3002e-06, "loss": 0.0127, "step": 43500 }, { "epoch": 2.3191875033320892, "grad_norm": 1.2843925952911377, "learning_rate": 1.3e-06, "loss": 0.0142, "step": 43501 }, { "epoch": 2.3192408167617424, "grad_norm": 3.0862534046173096, "learning_rate": 1.2998000000000002e-06, "loss": 0.014, "step": 43502 }, { "epoch": 2.319294130191395, "grad_norm": 3.0972416400909424, "learning_rate": 1.2996e-06, "loss": 0.0242, "step": 43503 }, { "epoch": 2.319347443621048, "grad_norm": 3.2099618911743164, "learning_rate": 1.2994e-06, "loss": 0.0102, "step": 43504 }, { "epoch": 2.319400757050701, "grad_norm": 1.941436529159546, "learning_rate": 1.2992000000000002e-06, "loss": 0.0131, "step": 43505 }, { "epoch": 2.319454070480354, "grad_norm": 1.7324591875076294, "learning_rate": 1.299e-06, "loss": 0.0047, "step": 43506 }, { "epoch": 2.319507383910007, "grad_norm": 4.011951923370361, "learning_rate": 1.2988e-06, "loss": 0.0257, "step": 43507 }, { "epoch": 2.31956069733966, "grad_norm": 0.36263906955718994, "learning_rate": 1.2986000000000002e-06, "loss": 0.0032, "step": 43508 }, { "epoch": 2.3196140107693126, "grad_norm": 1.8845621347427368, "learning_rate": 1.2984000000000001e-06, "loss": 0.0349, "step": 43509 }, { "epoch": 2.319667324198966, "grad_norm": 3.773651599884033, "learning_rate": 1.2982e-06, "loss": 0.0433, "step": 43510 }, { "epoch": 2.3197206376286186, "grad_norm": 1.4755442142486572, "learning_rate": 1.2980000000000001e-06, "loss": 0.005, "step": 43511 }, { "epoch": 2.3197739510582718, "grad_norm": 2.137556791305542, "learning_rate": 1.2978e-06, "loss": 0.0112, "step": 43512 }, { "epoch": 2.3198272644879245, "grad_norm": 2.1610565185546875, "learning_rate": 1.2975999999999999e-06, "loss": 0.009, "step": 43513 }, { "epoch": 2.3198805779175773, "grad_norm": 1.22422194480896, "learning_rate": 1.2974e-06, "loss": 0.0135, "step": 43514 }, { "epoch": 2.3199338913472305, "grad_norm": 2.6209025382995605, "learning_rate": 1.2972e-06, "loss": 0.009, "step": 43515 }, { "epoch": 2.3199872047768832, "grad_norm": 0.5641321539878845, "learning_rate": 1.2970000000000002e-06, "loss": 0.0036, "step": 43516 }, { "epoch": 2.320040518206536, "grad_norm": 1.5229570865631104, "learning_rate": 1.2968e-06, "loss": 0.0157, "step": 43517 }, { "epoch": 2.320093831636189, "grad_norm": 0.7862324118614197, "learning_rate": 1.2966e-06, "loss": 0.0032, "step": 43518 }, { "epoch": 2.320147145065842, "grad_norm": 3.591665029525757, "learning_rate": 1.2964000000000002e-06, "loss": 0.0394, "step": 43519 }, { "epoch": 2.320200458495495, "grad_norm": 1.1658954620361328, "learning_rate": 1.2962000000000002e-06, "loss": 0.0048, "step": 43520 }, { "epoch": 2.320253771925148, "grad_norm": 1.485339641571045, "learning_rate": 1.296e-06, "loss": 0.0178, "step": 43521 }, { "epoch": 2.3203070853548007, "grad_norm": 1.48660409450531, "learning_rate": 1.2958000000000002e-06, "loss": 0.0249, "step": 43522 }, { "epoch": 2.320360398784454, "grad_norm": 2.8652007579803467, "learning_rate": 1.2956000000000001e-06, "loss": 0.0426, "step": 43523 }, { "epoch": 2.3204137122141066, "grad_norm": 2.2389678955078125, "learning_rate": 1.2954e-06, "loss": 0.0072, "step": 43524 }, { "epoch": 2.32046702564376, "grad_norm": 1.683054804801941, "learning_rate": 1.2952000000000001e-06, "loss": 0.0059, "step": 43525 }, { "epoch": 2.3205203390734126, "grad_norm": 2.198420286178589, "learning_rate": 1.295e-06, "loss": 0.0272, "step": 43526 }, { "epoch": 2.3205736525030654, "grad_norm": 0.650722324848175, "learning_rate": 1.2948000000000003e-06, "loss": 0.0031, "step": 43527 }, { "epoch": 2.3206269659327186, "grad_norm": 1.5701643228530884, "learning_rate": 1.2946e-06, "loss": 0.009, "step": 43528 }, { "epoch": 2.3206802793623713, "grad_norm": 0.3664165139198303, "learning_rate": 1.2944e-06, "loss": 0.0023, "step": 43529 }, { "epoch": 2.320733592792024, "grad_norm": 2.230764389038086, "learning_rate": 1.2942000000000002e-06, "loss": 0.0429, "step": 43530 }, { "epoch": 2.3207869062216773, "grad_norm": 2.392536163330078, "learning_rate": 1.294e-06, "loss": 0.0062, "step": 43531 }, { "epoch": 2.32084021965133, "grad_norm": 1.3974016904830933, "learning_rate": 1.2938e-06, "loss": 0.0169, "step": 43532 }, { "epoch": 2.3208935330809832, "grad_norm": 0.5508905649185181, "learning_rate": 1.2936000000000002e-06, "loss": 0.0028, "step": 43533 }, { "epoch": 2.320946846510636, "grad_norm": 2.977797269821167, "learning_rate": 1.2934000000000002e-06, "loss": 0.0134, "step": 43534 }, { "epoch": 2.321000159940289, "grad_norm": 2.1508686542510986, "learning_rate": 1.2932e-06, "loss": 0.0277, "step": 43535 }, { "epoch": 2.321053473369942, "grad_norm": 0.3235495984554291, "learning_rate": 1.2930000000000002e-06, "loss": 0.0031, "step": 43536 }, { "epoch": 2.3211067867995947, "grad_norm": 0.9486847519874573, "learning_rate": 1.2928000000000001e-06, "loss": 0.0037, "step": 43537 }, { "epoch": 2.321160100229248, "grad_norm": 1.5227388143539429, "learning_rate": 1.2926000000000003e-06, "loss": 0.0174, "step": 43538 }, { "epoch": 2.3212134136589007, "grad_norm": 1.389535903930664, "learning_rate": 1.2924000000000001e-06, "loss": 0.0235, "step": 43539 }, { "epoch": 2.3212667270885534, "grad_norm": 1.5001773834228516, "learning_rate": 1.2922e-06, "loss": 0.0358, "step": 43540 }, { "epoch": 2.3213200405182066, "grad_norm": 2.935506582260132, "learning_rate": 1.2920000000000003e-06, "loss": 0.0148, "step": 43541 }, { "epoch": 2.3213733539478594, "grad_norm": 1.4426209926605225, "learning_rate": 1.2918e-06, "loss": 0.019, "step": 43542 }, { "epoch": 2.3214266673775126, "grad_norm": 3.3207192420959473, "learning_rate": 1.2916e-06, "loss": 0.0146, "step": 43543 }, { "epoch": 2.3214799808071653, "grad_norm": 1.4105780124664307, "learning_rate": 1.2914000000000002e-06, "loss": 0.0159, "step": 43544 }, { "epoch": 2.321533294236818, "grad_norm": 1.2751679420471191, "learning_rate": 1.2912000000000002e-06, "loss": 0.0187, "step": 43545 }, { "epoch": 2.3215866076664713, "grad_norm": 2.6455070972442627, "learning_rate": 1.291e-06, "loss": 0.0157, "step": 43546 }, { "epoch": 2.321639921096124, "grad_norm": 1.6887218952178955, "learning_rate": 1.2908000000000002e-06, "loss": 0.013, "step": 43547 }, { "epoch": 2.3216932345257772, "grad_norm": 1.9181110858917236, "learning_rate": 1.2906000000000002e-06, "loss": 0.0178, "step": 43548 }, { "epoch": 2.32174654795543, "grad_norm": 1.6224051713943481, "learning_rate": 1.2904e-06, "loss": 0.018, "step": 43549 }, { "epoch": 2.3217998613850828, "grad_norm": 2.032653331756592, "learning_rate": 1.2902000000000002e-06, "loss": 0.0054, "step": 43550 }, { "epoch": 2.321853174814736, "grad_norm": 1.5808802843093872, "learning_rate": 1.2900000000000001e-06, "loss": 0.024, "step": 43551 }, { "epoch": 2.3219064882443887, "grad_norm": 3.244389295578003, "learning_rate": 1.2898000000000001e-06, "loss": 0.0227, "step": 43552 }, { "epoch": 2.3219598016740415, "grad_norm": 2.0328543186187744, "learning_rate": 1.2896000000000001e-06, "loss": 0.0219, "step": 43553 }, { "epoch": 2.3220131151036947, "grad_norm": 1.7510490417480469, "learning_rate": 1.2894e-06, "loss": 0.0283, "step": 43554 }, { "epoch": 2.3220664285333474, "grad_norm": 0.49324679374694824, "learning_rate": 1.2892e-06, "loss": 0.0037, "step": 43555 }, { "epoch": 2.3221197419630006, "grad_norm": 1.9353506565093994, "learning_rate": 1.289e-06, "loss": 0.0154, "step": 43556 }, { "epoch": 2.3221730553926534, "grad_norm": 2.403560161590576, "learning_rate": 1.2888e-06, "loss": 0.0436, "step": 43557 }, { "epoch": 2.322226368822306, "grad_norm": 2.03232479095459, "learning_rate": 1.2886e-06, "loss": 0.0073, "step": 43558 }, { "epoch": 2.3222796822519594, "grad_norm": 2.049389600753784, "learning_rate": 1.2884000000000002e-06, "loss": 0.0083, "step": 43559 }, { "epoch": 2.322332995681612, "grad_norm": 2.4755136966705322, "learning_rate": 1.2882e-06, "loss": 0.0127, "step": 43560 }, { "epoch": 2.3223863091112653, "grad_norm": 2.043426036834717, "learning_rate": 1.288e-06, "loss": 0.0189, "step": 43561 }, { "epoch": 2.322439622540918, "grad_norm": 3.707374095916748, "learning_rate": 1.2878000000000002e-06, "loss": 0.0412, "step": 43562 }, { "epoch": 2.322492935970571, "grad_norm": 1.6377233266830444, "learning_rate": 1.2876000000000002e-06, "loss": 0.0083, "step": 43563 }, { "epoch": 2.322546249400224, "grad_norm": 1.978525996208191, "learning_rate": 1.2874e-06, "loss": 0.0142, "step": 43564 }, { "epoch": 2.322599562829877, "grad_norm": 5.5388007164001465, "learning_rate": 1.2872000000000001e-06, "loss": 0.0311, "step": 43565 }, { "epoch": 2.3226528762595295, "grad_norm": 0.8755733370780945, "learning_rate": 1.2870000000000001e-06, "loss": 0.0056, "step": 43566 }, { "epoch": 2.3227061896891827, "grad_norm": 3.649876594543457, "learning_rate": 1.2868e-06, "loss": 0.0487, "step": 43567 }, { "epoch": 2.3227595031188355, "grad_norm": 1.057908296585083, "learning_rate": 1.2866e-06, "loss": 0.0087, "step": 43568 }, { "epoch": 2.3228128165484887, "grad_norm": 1.0648508071899414, "learning_rate": 1.2864e-06, "loss": 0.0165, "step": 43569 }, { "epoch": 2.3228661299781415, "grad_norm": 2.743915319442749, "learning_rate": 1.2862000000000003e-06, "loss": 0.0171, "step": 43570 }, { "epoch": 2.3229194434077947, "grad_norm": 2.44254732131958, "learning_rate": 1.286e-06, "loss": 0.0091, "step": 43571 }, { "epoch": 2.3229727568374474, "grad_norm": 2.272616147994995, "learning_rate": 1.2858e-06, "loss": 0.0267, "step": 43572 }, { "epoch": 2.3230260702671, "grad_norm": 0.9185313582420349, "learning_rate": 1.2856000000000002e-06, "loss": 0.0217, "step": 43573 }, { "epoch": 2.3230793836967534, "grad_norm": 1.2234232425689697, "learning_rate": 1.2854e-06, "loss": 0.0063, "step": 43574 }, { "epoch": 2.323132697126406, "grad_norm": 1.791839599609375, "learning_rate": 1.2852e-06, "loss": 0.0069, "step": 43575 }, { "epoch": 2.323186010556059, "grad_norm": 0.870944619178772, "learning_rate": 1.2850000000000002e-06, "loss": 0.02, "step": 43576 }, { "epoch": 2.323239323985712, "grad_norm": 0.3967830240726471, "learning_rate": 1.2848000000000002e-06, "loss": 0.0027, "step": 43577 }, { "epoch": 2.323292637415365, "grad_norm": 1.8258814811706543, "learning_rate": 1.2846e-06, "loss": 0.0204, "step": 43578 }, { "epoch": 2.323345950845018, "grad_norm": 0.8040035367012024, "learning_rate": 1.2844000000000001e-06, "loss": 0.0044, "step": 43579 }, { "epoch": 2.323399264274671, "grad_norm": 1.230995535850525, "learning_rate": 1.2842000000000001e-06, "loss": 0.0125, "step": 43580 }, { "epoch": 2.3234525777043236, "grad_norm": 5.519005298614502, "learning_rate": 1.284e-06, "loss": 0.0327, "step": 43581 }, { "epoch": 2.3235058911339768, "grad_norm": 2.0231544971466064, "learning_rate": 1.2838e-06, "loss": 0.0101, "step": 43582 }, { "epoch": 2.3235592045636295, "grad_norm": 1.9218863248825073, "learning_rate": 1.2836e-06, "loss": 0.0132, "step": 43583 }, { "epoch": 2.3236125179932827, "grad_norm": 4.371109962463379, "learning_rate": 1.2834000000000003e-06, "loss": 0.0382, "step": 43584 }, { "epoch": 2.3236658314229355, "grad_norm": 2.1080212593078613, "learning_rate": 1.2832e-06, "loss": 0.0241, "step": 43585 }, { "epoch": 2.3237191448525882, "grad_norm": 3.6761813163757324, "learning_rate": 1.283e-06, "loss": 0.0608, "step": 43586 }, { "epoch": 2.3237724582822414, "grad_norm": 1.685940146446228, "learning_rate": 1.2828000000000002e-06, "loss": 0.0226, "step": 43587 }, { "epoch": 2.323825771711894, "grad_norm": 2.216718912124634, "learning_rate": 1.2826000000000002e-06, "loss": 0.0282, "step": 43588 }, { "epoch": 2.323879085141547, "grad_norm": 2.2029976844787598, "learning_rate": 1.2824e-06, "loss": 0.0162, "step": 43589 }, { "epoch": 2.3239323985712, "grad_norm": 1.6173679828643799, "learning_rate": 1.2822000000000002e-06, "loss": 0.0086, "step": 43590 }, { "epoch": 2.323985712000853, "grad_norm": 1.2183212041854858, "learning_rate": 1.2820000000000002e-06, "loss": 0.007, "step": 43591 }, { "epoch": 2.324039025430506, "grad_norm": 0.3164638876914978, "learning_rate": 1.2818e-06, "loss": 0.0032, "step": 43592 }, { "epoch": 2.324092338860159, "grad_norm": 0.8802191019058228, "learning_rate": 1.2816000000000001e-06, "loss": 0.0067, "step": 43593 }, { "epoch": 2.3241456522898116, "grad_norm": 2.204468250274658, "learning_rate": 1.2814000000000001e-06, "loss": 0.0237, "step": 43594 }, { "epoch": 2.324198965719465, "grad_norm": 2.322488307952881, "learning_rate": 1.2812000000000001e-06, "loss": 0.0315, "step": 43595 }, { "epoch": 2.3242522791491176, "grad_norm": 0.21943315863609314, "learning_rate": 1.281e-06, "loss": 0.002, "step": 43596 }, { "epoch": 2.324305592578771, "grad_norm": 2.13338303565979, "learning_rate": 1.2808e-06, "loss": 0.0142, "step": 43597 }, { "epoch": 2.3243589060084235, "grad_norm": 2.8398375511169434, "learning_rate": 1.2806e-06, "loss": 0.0288, "step": 43598 }, { "epoch": 2.3244122194380763, "grad_norm": 1.8879197835922241, "learning_rate": 1.2804e-06, "loss": 0.0257, "step": 43599 }, { "epoch": 2.3244655328677295, "grad_norm": 2.4657034873962402, "learning_rate": 1.2802e-06, "loss": 0.0181, "step": 43600 }, { "epoch": 2.3245188462973823, "grad_norm": 1.9413317441940308, "learning_rate": 1.28e-06, "loss": 0.0179, "step": 43601 }, { "epoch": 2.324572159727035, "grad_norm": 0.9053436517715454, "learning_rate": 1.2798000000000002e-06, "loss": 0.0042, "step": 43602 }, { "epoch": 2.324625473156688, "grad_norm": 2.2833454608917236, "learning_rate": 1.2796e-06, "loss": 0.0273, "step": 43603 }, { "epoch": 2.324678786586341, "grad_norm": 2.045595645904541, "learning_rate": 1.2794e-06, "loss": 0.0183, "step": 43604 }, { "epoch": 2.324732100015994, "grad_norm": 4.934189796447754, "learning_rate": 1.2792000000000002e-06, "loss": 0.0397, "step": 43605 }, { "epoch": 2.324785413445647, "grad_norm": 1.198433518409729, "learning_rate": 1.2790000000000002e-06, "loss": 0.0078, "step": 43606 }, { "epoch": 2.3248387268753, "grad_norm": 1.9773293733596802, "learning_rate": 1.2788e-06, "loss": 0.0271, "step": 43607 }, { "epoch": 2.324892040304953, "grad_norm": 1.014660120010376, "learning_rate": 1.2786000000000001e-06, "loss": 0.0072, "step": 43608 }, { "epoch": 2.3249453537346056, "grad_norm": 0.7538238167762756, "learning_rate": 1.2784000000000001e-06, "loss": 0.0041, "step": 43609 }, { "epoch": 2.324998667164259, "grad_norm": 1.8688459396362305, "learning_rate": 1.2781999999999999e-06, "loss": 0.0094, "step": 43610 }, { "epoch": 2.3250519805939116, "grad_norm": 1.64100182056427, "learning_rate": 1.278e-06, "loss": 0.0207, "step": 43611 }, { "epoch": 2.3251052940235644, "grad_norm": 0.6650810241699219, "learning_rate": 1.2778e-06, "loss": 0.0041, "step": 43612 }, { "epoch": 2.3251586074532176, "grad_norm": 1.2912514209747314, "learning_rate": 1.2776000000000003e-06, "loss": 0.0211, "step": 43613 }, { "epoch": 2.3252119208828703, "grad_norm": 2.3153011798858643, "learning_rate": 1.2774e-06, "loss": 0.0379, "step": 43614 }, { "epoch": 2.325265234312523, "grad_norm": 2.3095412254333496, "learning_rate": 1.2772e-06, "loss": 0.0268, "step": 43615 }, { "epoch": 2.3253185477421763, "grad_norm": 1.9698753356933594, "learning_rate": 1.2770000000000002e-06, "loss": 0.0233, "step": 43616 }, { "epoch": 2.325371861171829, "grad_norm": 1.541129231452942, "learning_rate": 1.2768e-06, "loss": 0.0191, "step": 43617 }, { "epoch": 2.3254251746014822, "grad_norm": 1.8933403491973877, "learning_rate": 1.2766e-06, "loss": 0.0389, "step": 43618 }, { "epoch": 2.325478488031135, "grad_norm": 1.6317943334579468, "learning_rate": 1.2764000000000002e-06, "loss": 0.0149, "step": 43619 }, { "epoch": 2.325531801460788, "grad_norm": 1.597365379333496, "learning_rate": 1.2762000000000002e-06, "loss": 0.0072, "step": 43620 }, { "epoch": 2.325585114890441, "grad_norm": 3.863844156265259, "learning_rate": 1.276e-06, "loss": 0.0451, "step": 43621 }, { "epoch": 2.3256384283200937, "grad_norm": 2.759798526763916, "learning_rate": 1.2758000000000001e-06, "loss": 0.0213, "step": 43622 }, { "epoch": 2.325691741749747, "grad_norm": 2.1960251331329346, "learning_rate": 1.2756000000000001e-06, "loss": 0.023, "step": 43623 }, { "epoch": 2.3257450551793997, "grad_norm": 3.0018820762634277, "learning_rate": 1.2753999999999999e-06, "loss": 0.0245, "step": 43624 }, { "epoch": 2.3257983686090524, "grad_norm": 1.974308967590332, "learning_rate": 1.2752e-06, "loss": 0.0453, "step": 43625 }, { "epoch": 2.3258516820387056, "grad_norm": 3.3748960494995117, "learning_rate": 1.275e-06, "loss": 0.0281, "step": 43626 }, { "epoch": 2.3259049954683584, "grad_norm": 2.938563585281372, "learning_rate": 1.2748000000000003e-06, "loss": 0.026, "step": 43627 }, { "epoch": 2.3259583088980116, "grad_norm": 2.978076934814453, "learning_rate": 1.2746e-06, "loss": 0.026, "step": 43628 }, { "epoch": 2.3260116223276643, "grad_norm": 3.315248727798462, "learning_rate": 1.2744e-06, "loss": 0.0328, "step": 43629 }, { "epoch": 2.326064935757317, "grad_norm": 1.2696800231933594, "learning_rate": 1.2742000000000002e-06, "loss": 0.012, "step": 43630 }, { "epoch": 2.3261182491869703, "grad_norm": 1.2334402799606323, "learning_rate": 1.2740000000000002e-06, "loss": 0.0264, "step": 43631 }, { "epoch": 2.326171562616623, "grad_norm": 1.4806331396102905, "learning_rate": 1.2738e-06, "loss": 0.0127, "step": 43632 }, { "epoch": 2.3262248760462763, "grad_norm": 1.777443528175354, "learning_rate": 1.2736000000000002e-06, "loss": 0.0222, "step": 43633 }, { "epoch": 2.326278189475929, "grad_norm": 2.0153255462646484, "learning_rate": 1.2734000000000002e-06, "loss": 0.0193, "step": 43634 }, { "epoch": 2.3263315029055818, "grad_norm": 2.803208589553833, "learning_rate": 1.2732e-06, "loss": 0.0408, "step": 43635 }, { "epoch": 2.326384816335235, "grad_norm": 1.7292754650115967, "learning_rate": 1.2730000000000001e-06, "loss": 0.0153, "step": 43636 }, { "epoch": 2.3264381297648877, "grad_norm": 1.075570821762085, "learning_rate": 1.2728000000000001e-06, "loss": 0.0071, "step": 43637 }, { "epoch": 2.3264914431945405, "grad_norm": 1.8211655616760254, "learning_rate": 1.2726000000000003e-06, "loss": 0.0167, "step": 43638 }, { "epoch": 2.3265447566241937, "grad_norm": 1.9691433906555176, "learning_rate": 1.2724e-06, "loss": 0.0227, "step": 43639 }, { "epoch": 2.3265980700538464, "grad_norm": 1.6012649536132812, "learning_rate": 1.2722e-06, "loss": 0.022, "step": 43640 }, { "epoch": 2.3266513834834996, "grad_norm": 1.8548682928085327, "learning_rate": 1.2720000000000003e-06, "loss": 0.0086, "step": 43641 }, { "epoch": 2.3267046969131524, "grad_norm": 4.734177112579346, "learning_rate": 1.2718e-06, "loss": 0.0283, "step": 43642 }, { "epoch": 2.326758010342805, "grad_norm": 2.30871319770813, "learning_rate": 1.2716e-06, "loss": 0.0101, "step": 43643 }, { "epoch": 2.3268113237724584, "grad_norm": 1.4373705387115479, "learning_rate": 1.2714000000000002e-06, "loss": 0.0082, "step": 43644 }, { "epoch": 2.326864637202111, "grad_norm": 1.7778058052062988, "learning_rate": 1.2712000000000002e-06, "loss": 0.0096, "step": 43645 }, { "epoch": 2.3269179506317643, "grad_norm": 1.4364211559295654, "learning_rate": 1.271e-06, "loss": 0.0256, "step": 43646 }, { "epoch": 2.326971264061417, "grad_norm": 0.9425971508026123, "learning_rate": 1.2708000000000002e-06, "loss": 0.0056, "step": 43647 }, { "epoch": 2.32702457749107, "grad_norm": 1.805943250656128, "learning_rate": 1.2706000000000002e-06, "loss": 0.0203, "step": 43648 }, { "epoch": 2.327077890920723, "grad_norm": 1.640057921409607, "learning_rate": 1.2704e-06, "loss": 0.0083, "step": 43649 }, { "epoch": 2.327131204350376, "grad_norm": 2.0850791931152344, "learning_rate": 1.2702000000000001e-06, "loss": 0.0136, "step": 43650 }, { "epoch": 2.3271845177800285, "grad_norm": 1.6838953495025635, "learning_rate": 1.2700000000000001e-06, "loss": 0.0141, "step": 43651 }, { "epoch": 2.3272378312096818, "grad_norm": 5.060072422027588, "learning_rate": 1.2698000000000001e-06, "loss": 0.0271, "step": 43652 }, { "epoch": 2.3272911446393345, "grad_norm": 2.1170766353607178, "learning_rate": 1.2696e-06, "loss": 0.0517, "step": 43653 }, { "epoch": 2.3273444580689877, "grad_norm": 1.6227854490280151, "learning_rate": 1.2694e-06, "loss": 0.0082, "step": 43654 }, { "epoch": 2.3273977714986405, "grad_norm": 0.47050267457962036, "learning_rate": 1.2692e-06, "loss": 0.0025, "step": 43655 }, { "epoch": 2.3274510849282937, "grad_norm": 4.302446365356445, "learning_rate": 1.2690000000000003e-06, "loss": 0.0168, "step": 43656 }, { "epoch": 2.3275043983579464, "grad_norm": 2.598337173461914, "learning_rate": 1.2688e-06, "loss": 0.0093, "step": 43657 }, { "epoch": 2.327557711787599, "grad_norm": 1.1975582838058472, "learning_rate": 1.2686e-06, "loss": 0.0088, "step": 43658 }, { "epoch": 2.3276110252172524, "grad_norm": 1.3210755586624146, "learning_rate": 1.2684000000000002e-06, "loss": 0.0205, "step": 43659 }, { "epoch": 2.327664338646905, "grad_norm": 0.734747588634491, "learning_rate": 1.2682e-06, "loss": 0.0047, "step": 43660 }, { "epoch": 2.327717652076558, "grad_norm": 1.4565625190734863, "learning_rate": 1.268e-06, "loss": 0.0231, "step": 43661 }, { "epoch": 2.327770965506211, "grad_norm": 1.1411067247390747, "learning_rate": 1.2678000000000002e-06, "loss": 0.0055, "step": 43662 }, { "epoch": 2.327824278935864, "grad_norm": 3.527033805847168, "learning_rate": 1.2676000000000002e-06, "loss": 0.0133, "step": 43663 }, { "epoch": 2.327877592365517, "grad_norm": 3.7914395332336426, "learning_rate": 1.2674e-06, "loss": 0.0113, "step": 43664 }, { "epoch": 2.32793090579517, "grad_norm": 1.0563677549362183, "learning_rate": 1.2672000000000001e-06, "loss": 0.0045, "step": 43665 }, { "epoch": 2.3279842192248226, "grad_norm": 3.9027905464172363, "learning_rate": 1.2670000000000001e-06, "loss": 0.0172, "step": 43666 }, { "epoch": 2.3280375326544758, "grad_norm": 1.0378410816192627, "learning_rate": 1.2667999999999999e-06, "loss": 0.0129, "step": 43667 }, { "epoch": 2.3280908460841285, "grad_norm": 2.6665990352630615, "learning_rate": 1.2666e-06, "loss": 0.0119, "step": 43668 }, { "epoch": 2.3281441595137817, "grad_norm": 2.0180447101593018, "learning_rate": 1.2664e-06, "loss": 0.0142, "step": 43669 }, { "epoch": 2.3281974729434345, "grad_norm": 1.2177962064743042, "learning_rate": 1.2662000000000003e-06, "loss": 0.0053, "step": 43670 }, { "epoch": 2.3282507863730872, "grad_norm": 2.648578643798828, "learning_rate": 1.266e-06, "loss": 0.0238, "step": 43671 }, { "epoch": 2.3283040998027404, "grad_norm": 3.200377941131592, "learning_rate": 1.2658e-06, "loss": 0.0242, "step": 43672 }, { "epoch": 2.328357413232393, "grad_norm": 2.02268385887146, "learning_rate": 1.2656000000000002e-06, "loss": 0.0058, "step": 43673 }, { "epoch": 2.328410726662046, "grad_norm": 1.7515274286270142, "learning_rate": 1.2654000000000002e-06, "loss": 0.0081, "step": 43674 }, { "epoch": 2.328464040091699, "grad_norm": 2.0866522789001465, "learning_rate": 1.2652e-06, "loss": 0.0476, "step": 43675 }, { "epoch": 2.328517353521352, "grad_norm": 0.44058483839035034, "learning_rate": 1.2650000000000002e-06, "loss": 0.0033, "step": 43676 }, { "epoch": 2.328570666951005, "grad_norm": 2.007110118865967, "learning_rate": 1.2648000000000002e-06, "loss": 0.0252, "step": 43677 }, { "epoch": 2.328623980380658, "grad_norm": 1.7269457578659058, "learning_rate": 1.2646e-06, "loss": 0.0264, "step": 43678 }, { "epoch": 2.3286772938103106, "grad_norm": 1.812474250793457, "learning_rate": 1.2644000000000001e-06, "loss": 0.007, "step": 43679 }, { "epoch": 2.328730607239964, "grad_norm": 2.2444908618927, "learning_rate": 1.2642000000000001e-06, "loss": 0.0223, "step": 43680 }, { "epoch": 2.3287839206696166, "grad_norm": 2.1294867992401123, "learning_rate": 1.2640000000000003e-06, "loss": 0.0306, "step": 43681 }, { "epoch": 2.32883723409927, "grad_norm": 1.7358614206314087, "learning_rate": 1.2638e-06, "loss": 0.0098, "step": 43682 }, { "epoch": 2.3288905475289225, "grad_norm": 1.3129366636276245, "learning_rate": 1.2636e-06, "loss": 0.0125, "step": 43683 }, { "epoch": 2.3289438609585753, "grad_norm": 2.0369231700897217, "learning_rate": 1.2634000000000003e-06, "loss": 0.0148, "step": 43684 }, { "epoch": 2.3289971743882285, "grad_norm": 2.1020331382751465, "learning_rate": 1.2632e-06, "loss": 0.0076, "step": 43685 }, { "epoch": 2.3290504878178813, "grad_norm": 3.625401496887207, "learning_rate": 1.263e-06, "loss": 0.0511, "step": 43686 }, { "epoch": 2.329103801247534, "grad_norm": 2.654003858566284, "learning_rate": 1.2628000000000002e-06, "loss": 0.0146, "step": 43687 }, { "epoch": 2.3291571146771872, "grad_norm": 3.164841890335083, "learning_rate": 1.2626000000000002e-06, "loss": 0.0356, "step": 43688 }, { "epoch": 2.32921042810684, "grad_norm": 1.3143287897109985, "learning_rate": 1.2624e-06, "loss": 0.0072, "step": 43689 }, { "epoch": 2.329263741536493, "grad_norm": 1.8980505466461182, "learning_rate": 1.2622000000000002e-06, "loss": 0.0133, "step": 43690 }, { "epoch": 2.329317054966146, "grad_norm": 0.8564536571502686, "learning_rate": 1.2620000000000002e-06, "loss": 0.0059, "step": 43691 }, { "epoch": 2.329370368395799, "grad_norm": 3.008389949798584, "learning_rate": 1.2618e-06, "loss": 0.0149, "step": 43692 }, { "epoch": 2.329423681825452, "grad_norm": 1.7473291158676147, "learning_rate": 1.2616000000000001e-06, "loss": 0.0225, "step": 43693 }, { "epoch": 2.3294769952551047, "grad_norm": 3.284482717514038, "learning_rate": 1.2614000000000001e-06, "loss": 0.011, "step": 43694 }, { "epoch": 2.329530308684758, "grad_norm": 1.7233235836029053, "learning_rate": 1.2612e-06, "loss": 0.0071, "step": 43695 }, { "epoch": 2.3295836221144106, "grad_norm": 1.745926856994629, "learning_rate": 1.261e-06, "loss": 0.0139, "step": 43696 }, { "epoch": 2.3296369355440634, "grad_norm": 2.6863834857940674, "learning_rate": 1.2608e-06, "loss": 0.0138, "step": 43697 }, { "epoch": 2.3296902489737166, "grad_norm": 1.364943504333496, "learning_rate": 1.2606e-06, "loss": 0.0098, "step": 43698 }, { "epoch": 2.3297435624033693, "grad_norm": 0.7389947175979614, "learning_rate": 1.2604000000000002e-06, "loss": 0.0052, "step": 43699 }, { "epoch": 2.3297968758330225, "grad_norm": 2.348815679550171, "learning_rate": 1.2602e-06, "loss": 0.0072, "step": 43700 }, { "epoch": 2.3298501892626753, "grad_norm": 1.7541677951812744, "learning_rate": 1.26e-06, "loss": 0.0078, "step": 43701 }, { "epoch": 2.329903502692328, "grad_norm": 1.6816071271896362, "learning_rate": 1.2598000000000002e-06, "loss": 0.0229, "step": 43702 }, { "epoch": 2.3299568161219812, "grad_norm": 0.5518516302108765, "learning_rate": 1.2596e-06, "loss": 0.0038, "step": 43703 }, { "epoch": 2.330010129551634, "grad_norm": 2.4794609546661377, "learning_rate": 1.2594e-06, "loss": 0.0104, "step": 43704 }, { "epoch": 2.330063442981287, "grad_norm": 2.3817739486694336, "learning_rate": 1.2592000000000002e-06, "loss": 0.013, "step": 43705 }, { "epoch": 2.33011675641094, "grad_norm": 2.3265812397003174, "learning_rate": 1.2590000000000001e-06, "loss": 0.0093, "step": 43706 }, { "epoch": 2.3301700698405927, "grad_norm": 2.056905746459961, "learning_rate": 1.2588e-06, "loss": 0.012, "step": 43707 }, { "epoch": 2.330223383270246, "grad_norm": 1.2991795539855957, "learning_rate": 1.2586000000000001e-06, "loss": 0.005, "step": 43708 }, { "epoch": 2.3302766966998987, "grad_norm": 1.656368374824524, "learning_rate": 1.2584e-06, "loss": 0.0168, "step": 43709 }, { "epoch": 2.3303300101295514, "grad_norm": 1.4962332248687744, "learning_rate": 1.2581999999999999e-06, "loss": 0.0176, "step": 43710 }, { "epoch": 2.3303833235592046, "grad_norm": 0.6582872271537781, "learning_rate": 1.258e-06, "loss": 0.004, "step": 43711 }, { "epoch": 2.3304366369888574, "grad_norm": 3.010676383972168, "learning_rate": 1.2578e-06, "loss": 0.0184, "step": 43712 }, { "epoch": 2.3304899504185106, "grad_norm": 0.4568788707256317, "learning_rate": 1.2576000000000002e-06, "loss": 0.0035, "step": 43713 }, { "epoch": 2.3305432638481633, "grad_norm": 3.577634811401367, "learning_rate": 1.2574e-06, "loss": 0.0097, "step": 43714 }, { "epoch": 2.330596577277816, "grad_norm": 1.9897651672363281, "learning_rate": 1.2572e-06, "loss": 0.0148, "step": 43715 }, { "epoch": 2.3306498907074693, "grad_norm": 1.058644413948059, "learning_rate": 1.2570000000000002e-06, "loss": 0.0143, "step": 43716 }, { "epoch": 2.330703204137122, "grad_norm": 1.9850314855575562, "learning_rate": 1.2568e-06, "loss": 0.0104, "step": 43717 }, { "epoch": 2.3307565175667753, "grad_norm": 2.7134313583374023, "learning_rate": 1.2566e-06, "loss": 0.0115, "step": 43718 }, { "epoch": 2.330809830996428, "grad_norm": 2.6215431690216064, "learning_rate": 1.2564000000000002e-06, "loss": 0.0288, "step": 43719 }, { "epoch": 2.330863144426081, "grad_norm": 0.3175802230834961, "learning_rate": 1.2562000000000001e-06, "loss": 0.0018, "step": 43720 }, { "epoch": 2.330916457855734, "grad_norm": 0.5345120429992676, "learning_rate": 1.256e-06, "loss": 0.0029, "step": 43721 }, { "epoch": 2.3309697712853867, "grad_norm": 3.1613872051239014, "learning_rate": 1.2558000000000001e-06, "loss": 0.0243, "step": 43722 }, { "epoch": 2.3310230847150395, "grad_norm": 1.2734488248825073, "learning_rate": 1.2556e-06, "loss": 0.0058, "step": 43723 }, { "epoch": 2.3310763981446927, "grad_norm": 0.2954772710800171, "learning_rate": 1.2554000000000003e-06, "loss": 0.0018, "step": 43724 }, { "epoch": 2.3311297115743455, "grad_norm": 0.6347736120223999, "learning_rate": 1.2552e-06, "loss": 0.004, "step": 43725 }, { "epoch": 2.3311830250039987, "grad_norm": 2.121786117553711, "learning_rate": 1.255e-06, "loss": 0.0087, "step": 43726 }, { "epoch": 2.3312363384336514, "grad_norm": 0.9998679161071777, "learning_rate": 1.2548000000000003e-06, "loss": 0.0043, "step": 43727 }, { "epoch": 2.3312896518633046, "grad_norm": 1.1317322254180908, "learning_rate": 1.2546e-06, "loss": 0.0102, "step": 43728 }, { "epoch": 2.3313429652929574, "grad_norm": 1.4896016120910645, "learning_rate": 1.2544e-06, "loss": 0.0151, "step": 43729 }, { "epoch": 2.33139627872261, "grad_norm": 2.4870765209198, "learning_rate": 1.2542000000000002e-06, "loss": 0.0044, "step": 43730 }, { "epoch": 2.3314495921522633, "grad_norm": 1.2367194890975952, "learning_rate": 1.2540000000000002e-06, "loss": 0.0126, "step": 43731 }, { "epoch": 2.331502905581916, "grad_norm": 1.7594181299209595, "learning_rate": 1.2538e-06, "loss": 0.0268, "step": 43732 }, { "epoch": 2.331556219011569, "grad_norm": 2.06913161277771, "learning_rate": 1.2536000000000002e-06, "loss": 0.0148, "step": 43733 }, { "epoch": 2.331609532441222, "grad_norm": 3.000822067260742, "learning_rate": 1.2534000000000001e-06, "loss": 0.0166, "step": 43734 }, { "epoch": 2.331662845870875, "grad_norm": 1.5827875137329102, "learning_rate": 1.2532e-06, "loss": 0.0138, "step": 43735 }, { "epoch": 2.3317161593005276, "grad_norm": 0.8621706366539001, "learning_rate": 1.2530000000000001e-06, "loss": 0.017, "step": 43736 }, { "epoch": 2.3317694727301808, "grad_norm": 1.4778902530670166, "learning_rate": 1.2528e-06, "loss": 0.0135, "step": 43737 }, { "epoch": 2.3318227861598335, "grad_norm": 2.4378812313079834, "learning_rate": 1.2526000000000003e-06, "loss": 0.0333, "step": 43738 }, { "epoch": 2.3318760995894867, "grad_norm": 0.8431481719017029, "learning_rate": 1.2524e-06, "loss": 0.0035, "step": 43739 }, { "epoch": 2.3319294130191395, "grad_norm": 0.9069947600364685, "learning_rate": 1.2522e-06, "loss": 0.0107, "step": 43740 }, { "epoch": 2.3319827264487927, "grad_norm": 1.054978847503662, "learning_rate": 1.2520000000000003e-06, "loss": 0.0051, "step": 43741 }, { "epoch": 2.3320360398784454, "grad_norm": 1.744032382965088, "learning_rate": 1.2518000000000002e-06, "loss": 0.0102, "step": 43742 }, { "epoch": 2.332089353308098, "grad_norm": 1.0096231698989868, "learning_rate": 1.2516e-06, "loss": 0.0055, "step": 43743 }, { "epoch": 2.3321426667377514, "grad_norm": 2.0797901153564453, "learning_rate": 1.2514000000000002e-06, "loss": 0.0348, "step": 43744 }, { "epoch": 2.332195980167404, "grad_norm": 1.1190102100372314, "learning_rate": 1.2512000000000002e-06, "loss": 0.0047, "step": 43745 }, { "epoch": 2.332249293597057, "grad_norm": 1.9692680835723877, "learning_rate": 1.251e-06, "loss": 0.022, "step": 43746 }, { "epoch": 2.33230260702671, "grad_norm": 1.2497005462646484, "learning_rate": 1.2508000000000002e-06, "loss": 0.0046, "step": 43747 }, { "epoch": 2.332355920456363, "grad_norm": 5.723013401031494, "learning_rate": 1.2506000000000001e-06, "loss": 0.028, "step": 43748 }, { "epoch": 2.332409233886016, "grad_norm": 1.5587587356567383, "learning_rate": 1.2504000000000001e-06, "loss": 0.0199, "step": 43749 }, { "epoch": 2.332462547315669, "grad_norm": 0.2762022316455841, "learning_rate": 1.2502000000000001e-06, "loss": 0.0021, "step": 43750 }, { "epoch": 2.3325158607453216, "grad_norm": 3.2125394344329834, "learning_rate": 1.25e-06, "loss": 0.0234, "step": 43751 }, { "epoch": 2.332569174174975, "grad_norm": 1.0295566320419312, "learning_rate": 1.2498e-06, "loss": 0.0055, "step": 43752 }, { "epoch": 2.3326224876046275, "grad_norm": 1.2122385501861572, "learning_rate": 1.2496e-06, "loss": 0.005, "step": 43753 }, { "epoch": 2.3326758010342807, "grad_norm": 1.9718960523605347, "learning_rate": 1.2494e-06, "loss": 0.0252, "step": 43754 }, { "epoch": 2.3327291144639335, "grad_norm": 0.2958509027957916, "learning_rate": 1.2492e-06, "loss": 0.002, "step": 43755 }, { "epoch": 2.3327824278935863, "grad_norm": 0.9500394463539124, "learning_rate": 1.249e-06, "loss": 0.004, "step": 43756 }, { "epoch": 2.3328357413232395, "grad_norm": 1.0490050315856934, "learning_rate": 1.2488000000000002e-06, "loss": 0.005, "step": 43757 }, { "epoch": 2.332889054752892, "grad_norm": 4.135334491729736, "learning_rate": 1.2486e-06, "loss": 0.0128, "step": 43758 }, { "epoch": 2.332942368182545, "grad_norm": 1.7638720273971558, "learning_rate": 1.2484000000000002e-06, "loss": 0.0343, "step": 43759 }, { "epoch": 2.332995681612198, "grad_norm": 1.8320136070251465, "learning_rate": 1.2482000000000002e-06, "loss": 0.0072, "step": 43760 }, { "epoch": 2.333048995041851, "grad_norm": 3.7991480827331543, "learning_rate": 1.248e-06, "loss": 0.027, "step": 43761 }, { "epoch": 2.333102308471504, "grad_norm": 1.8860933780670166, "learning_rate": 1.2478000000000001e-06, "loss": 0.0258, "step": 43762 }, { "epoch": 2.333155621901157, "grad_norm": 2.4518349170684814, "learning_rate": 1.2476000000000001e-06, "loss": 0.0117, "step": 43763 }, { "epoch": 2.3332089353308096, "grad_norm": 1.7310538291931152, "learning_rate": 1.2474000000000001e-06, "loss": 0.0126, "step": 43764 }, { "epoch": 2.333262248760463, "grad_norm": 2.6278412342071533, "learning_rate": 1.2472e-06, "loss": 0.0342, "step": 43765 }, { "epoch": 2.3333155621901156, "grad_norm": 2.391876220703125, "learning_rate": 1.247e-06, "loss": 0.0206, "step": 43766 }, { "epoch": 2.333368875619769, "grad_norm": 1.9510051012039185, "learning_rate": 1.2468e-06, "loss": 0.0076, "step": 43767 }, { "epoch": 2.3334221890494216, "grad_norm": 3.2498068809509277, "learning_rate": 1.2466000000000003e-06, "loss": 0.0128, "step": 43768 }, { "epoch": 2.3334755024790743, "grad_norm": 1.2239408493041992, "learning_rate": 1.2464e-06, "loss": 0.0242, "step": 43769 }, { "epoch": 2.3335288159087275, "grad_norm": 1.2217230796813965, "learning_rate": 1.2462e-06, "loss": 0.0049, "step": 43770 }, { "epoch": 2.3335821293383803, "grad_norm": 2.778669834136963, "learning_rate": 1.2460000000000002e-06, "loss": 0.0099, "step": 43771 }, { "epoch": 2.333635442768033, "grad_norm": 2.270085096359253, "learning_rate": 1.2458e-06, "loss": 0.0245, "step": 43772 }, { "epoch": 2.3336887561976862, "grad_norm": 1.570649266242981, "learning_rate": 1.2456000000000002e-06, "loss": 0.0066, "step": 43773 }, { "epoch": 2.333742069627339, "grad_norm": 2.23190975189209, "learning_rate": 1.2454000000000002e-06, "loss": 0.0333, "step": 43774 }, { "epoch": 2.333795383056992, "grad_norm": 2.1785850524902344, "learning_rate": 1.2452000000000002e-06, "loss": 0.009, "step": 43775 }, { "epoch": 2.333848696486645, "grad_norm": 0.6880914568901062, "learning_rate": 1.2450000000000002e-06, "loss": 0.0141, "step": 43776 }, { "epoch": 2.333902009916298, "grad_norm": 1.2683837413787842, "learning_rate": 1.2448000000000001e-06, "loss": 0.0198, "step": 43777 }, { "epoch": 2.333955323345951, "grad_norm": 1.718886375427246, "learning_rate": 1.2446000000000001e-06, "loss": 0.0206, "step": 43778 }, { "epoch": 2.3340086367756037, "grad_norm": 1.380864143371582, "learning_rate": 1.2444e-06, "loss": 0.0186, "step": 43779 }, { "epoch": 2.334061950205257, "grad_norm": 1.1585758924484253, "learning_rate": 1.2442e-06, "loss": 0.012, "step": 43780 }, { "epoch": 2.3341152636349096, "grad_norm": 1.1709202527999878, "learning_rate": 1.244e-06, "loss": 0.0159, "step": 43781 }, { "epoch": 2.3341685770645624, "grad_norm": 3.2240421772003174, "learning_rate": 1.2438e-06, "loss": 0.0152, "step": 43782 }, { "epoch": 2.3342218904942156, "grad_norm": 0.5171347260475159, "learning_rate": 1.2436e-06, "loss": 0.0025, "step": 43783 }, { "epoch": 2.3342752039238683, "grad_norm": 2.876188039779663, "learning_rate": 1.2434000000000002e-06, "loss": 0.0121, "step": 43784 }, { "epoch": 2.3343285173535215, "grad_norm": 1.9639018774032593, "learning_rate": 1.2432e-06, "loss": 0.0059, "step": 43785 }, { "epoch": 2.3343818307831743, "grad_norm": 3.571943759918213, "learning_rate": 1.243e-06, "loss": 0.0093, "step": 43786 }, { "epoch": 2.334435144212827, "grad_norm": 2.211766242980957, "learning_rate": 1.2428000000000002e-06, "loss": 0.0257, "step": 43787 }, { "epoch": 2.3344884576424803, "grad_norm": 2.8230302333831787, "learning_rate": 1.2426e-06, "loss": 0.02, "step": 43788 }, { "epoch": 2.334541771072133, "grad_norm": 2.208155632019043, "learning_rate": 1.2424000000000002e-06, "loss": 0.0182, "step": 43789 }, { "epoch": 2.334595084501786, "grad_norm": 0.4156947433948517, "learning_rate": 1.2422000000000002e-06, "loss": 0.0029, "step": 43790 }, { "epoch": 2.334648397931439, "grad_norm": 2.2367796897888184, "learning_rate": 1.2420000000000001e-06, "loss": 0.0066, "step": 43791 }, { "epoch": 2.3347017113610917, "grad_norm": 1.216089129447937, "learning_rate": 1.2418000000000001e-06, "loss": 0.0143, "step": 43792 }, { "epoch": 2.334755024790745, "grad_norm": 1.4267972707748413, "learning_rate": 1.2416000000000001e-06, "loss": 0.0155, "step": 43793 }, { "epoch": 2.3348083382203977, "grad_norm": 3.640587091445923, "learning_rate": 1.2414e-06, "loss": 0.0101, "step": 43794 }, { "epoch": 2.3348616516500504, "grad_norm": 1.3840889930725098, "learning_rate": 1.2412e-06, "loss": 0.0051, "step": 43795 }, { "epoch": 2.3349149650797036, "grad_norm": 1.2886183261871338, "learning_rate": 1.241e-06, "loss": 0.0257, "step": 43796 }, { "epoch": 2.3349682785093564, "grad_norm": 2.1297414302825928, "learning_rate": 1.2408e-06, "loss": 0.0047, "step": 43797 }, { "epoch": 2.3350215919390096, "grad_norm": 2.0219430923461914, "learning_rate": 1.2406000000000002e-06, "loss": 0.0132, "step": 43798 }, { "epoch": 2.3350749053686624, "grad_norm": 1.1943163871765137, "learning_rate": 1.2404e-06, "loss": 0.0135, "step": 43799 }, { "epoch": 2.335128218798315, "grad_norm": 0.7058718204498291, "learning_rate": 1.2402000000000002e-06, "loss": 0.0038, "step": 43800 }, { "epoch": 2.3351815322279683, "grad_norm": 1.9466545581817627, "learning_rate": 1.2400000000000002e-06, "loss": 0.0328, "step": 43801 }, { "epoch": 2.335234845657621, "grad_norm": 2.231611967086792, "learning_rate": 1.2398000000000002e-06, "loss": 0.0211, "step": 43802 }, { "epoch": 2.3352881590872743, "grad_norm": 1.1025570631027222, "learning_rate": 1.2396000000000002e-06, "loss": 0.004, "step": 43803 }, { "epoch": 2.335341472516927, "grad_norm": 0.6586016416549683, "learning_rate": 1.2394000000000002e-06, "loss": 0.0025, "step": 43804 }, { "epoch": 2.33539478594658, "grad_norm": 0.770999550819397, "learning_rate": 1.2392000000000001e-06, "loss": 0.0032, "step": 43805 }, { "epoch": 2.335448099376233, "grad_norm": 1.3274420499801636, "learning_rate": 1.2390000000000001e-06, "loss": 0.0289, "step": 43806 }, { "epoch": 2.3355014128058857, "grad_norm": 2.0851643085479736, "learning_rate": 1.2388000000000001e-06, "loss": 0.0059, "step": 43807 }, { "epoch": 2.3355547262355385, "grad_norm": 2.627174139022827, "learning_rate": 1.2386e-06, "loss": 0.0216, "step": 43808 }, { "epoch": 2.3356080396651917, "grad_norm": 0.3662394881248474, "learning_rate": 1.2384e-06, "loss": 0.0026, "step": 43809 }, { "epoch": 2.3356613530948445, "grad_norm": 0.4086248576641083, "learning_rate": 1.2382e-06, "loss": 0.0029, "step": 43810 }, { "epoch": 2.3357146665244977, "grad_norm": 1.452857255935669, "learning_rate": 1.238e-06, "loss": 0.018, "step": 43811 }, { "epoch": 2.3357679799541504, "grad_norm": 2.0207059383392334, "learning_rate": 1.2378e-06, "loss": 0.0165, "step": 43812 }, { "epoch": 2.3358212933838036, "grad_norm": 1.4090663194656372, "learning_rate": 1.2376e-06, "loss": 0.0116, "step": 43813 }, { "epoch": 2.3358746068134564, "grad_norm": 2.5395407676696777, "learning_rate": 1.2374000000000002e-06, "loss": 0.0075, "step": 43814 }, { "epoch": 2.335927920243109, "grad_norm": 1.159838080406189, "learning_rate": 1.2372e-06, "loss": 0.0072, "step": 43815 }, { "epoch": 2.3359812336727623, "grad_norm": 1.9907102584838867, "learning_rate": 1.2370000000000002e-06, "loss": 0.0356, "step": 43816 }, { "epoch": 2.336034547102415, "grad_norm": 1.0175657272338867, "learning_rate": 1.2368000000000002e-06, "loss": 0.0046, "step": 43817 }, { "epoch": 2.336087860532068, "grad_norm": 2.9038310050964355, "learning_rate": 1.2366000000000002e-06, "loss": 0.0309, "step": 43818 }, { "epoch": 2.336141173961721, "grad_norm": 0.9547075629234314, "learning_rate": 1.2364000000000001e-06, "loss": 0.0035, "step": 43819 }, { "epoch": 2.336194487391374, "grad_norm": 1.80412757396698, "learning_rate": 1.2362000000000001e-06, "loss": 0.0101, "step": 43820 }, { "epoch": 2.336247800821027, "grad_norm": 2.5204715728759766, "learning_rate": 1.2360000000000001e-06, "loss": 0.0336, "step": 43821 }, { "epoch": 2.3363011142506798, "grad_norm": 2.5563814640045166, "learning_rate": 1.2358e-06, "loss": 0.0091, "step": 43822 }, { "epoch": 2.3363544276803325, "grad_norm": 1.8376929759979248, "learning_rate": 1.2356e-06, "loss": 0.0323, "step": 43823 }, { "epoch": 2.3364077411099857, "grad_norm": 1.3768590688705444, "learning_rate": 1.2354e-06, "loss": 0.0302, "step": 43824 }, { "epoch": 2.3364610545396385, "grad_norm": 2.295778512954712, "learning_rate": 1.2352e-06, "loss": 0.0089, "step": 43825 }, { "epoch": 2.3365143679692917, "grad_norm": 1.2357994318008423, "learning_rate": 1.235e-06, "loss": 0.005, "step": 43826 }, { "epoch": 2.3365676813989444, "grad_norm": 1.62209951877594, "learning_rate": 1.2348000000000002e-06, "loss": 0.0402, "step": 43827 }, { "epoch": 2.336620994828597, "grad_norm": 1.301649570465088, "learning_rate": 1.2346e-06, "loss": 0.024, "step": 43828 }, { "epoch": 2.3366743082582504, "grad_norm": 1.0463546514511108, "learning_rate": 1.2344e-06, "loss": 0.0031, "step": 43829 }, { "epoch": 2.336727621687903, "grad_norm": 2.009377956390381, "learning_rate": 1.2342000000000002e-06, "loss": 0.0263, "step": 43830 }, { "epoch": 2.336780935117556, "grad_norm": 0.6075669527053833, "learning_rate": 1.234e-06, "loss": 0.0039, "step": 43831 }, { "epoch": 2.336834248547209, "grad_norm": 4.289801120758057, "learning_rate": 1.2338000000000002e-06, "loss": 0.0278, "step": 43832 }, { "epoch": 2.336887561976862, "grad_norm": 3.4982261657714844, "learning_rate": 1.2336000000000001e-06, "loss": 0.0131, "step": 43833 }, { "epoch": 2.336940875406515, "grad_norm": 2.5093936920166016, "learning_rate": 1.2334000000000001e-06, "loss": 0.0441, "step": 43834 }, { "epoch": 2.336994188836168, "grad_norm": 1.6656979322433472, "learning_rate": 1.2332000000000001e-06, "loss": 0.0105, "step": 43835 }, { "epoch": 2.3370475022658206, "grad_norm": 0.31557705998420715, "learning_rate": 1.233e-06, "loss": 0.0028, "step": 43836 }, { "epoch": 2.337100815695474, "grad_norm": 0.35455459356307983, "learning_rate": 1.2328e-06, "loss": 0.0028, "step": 43837 }, { "epoch": 2.3371541291251265, "grad_norm": 0.562671422958374, "learning_rate": 1.2326e-06, "loss": 0.0041, "step": 43838 }, { "epoch": 2.3372074425547797, "grad_norm": 0.26198846101760864, "learning_rate": 1.2324e-06, "loss": 0.0015, "step": 43839 }, { "epoch": 2.3372607559844325, "grad_norm": 2.5714733600616455, "learning_rate": 1.2322e-06, "loss": 0.0263, "step": 43840 }, { "epoch": 2.3373140694140853, "grad_norm": 2.644674777984619, "learning_rate": 1.2320000000000002e-06, "loss": 0.0359, "step": 43841 }, { "epoch": 2.3373673828437385, "grad_norm": 1.3051248788833618, "learning_rate": 1.2318e-06, "loss": 0.0247, "step": 43842 }, { "epoch": 2.337420696273391, "grad_norm": 0.7508469223976135, "learning_rate": 1.2316000000000002e-06, "loss": 0.0024, "step": 43843 }, { "epoch": 2.337474009703044, "grad_norm": 1.9030215740203857, "learning_rate": 1.2314000000000002e-06, "loss": 0.0155, "step": 43844 }, { "epoch": 2.337527323132697, "grad_norm": 3.47782039642334, "learning_rate": 1.2312e-06, "loss": 0.0129, "step": 43845 }, { "epoch": 2.33758063656235, "grad_norm": 2.307098388671875, "learning_rate": 1.2310000000000002e-06, "loss": 0.0053, "step": 43846 }, { "epoch": 2.337633949992003, "grad_norm": 1.830268383026123, "learning_rate": 1.2308000000000001e-06, "loss": 0.0132, "step": 43847 }, { "epoch": 2.337687263421656, "grad_norm": 1.0827748775482178, "learning_rate": 1.2306000000000001e-06, "loss": 0.0066, "step": 43848 }, { "epoch": 2.337740576851309, "grad_norm": 1.7358381748199463, "learning_rate": 1.2304000000000001e-06, "loss": 0.0139, "step": 43849 }, { "epoch": 2.337793890280962, "grad_norm": 2.2456107139587402, "learning_rate": 1.2302e-06, "loss": 0.0068, "step": 43850 }, { "epoch": 2.3378472037106146, "grad_norm": 1.7166695594787598, "learning_rate": 1.23e-06, "loss": 0.0132, "step": 43851 }, { "epoch": 2.337900517140268, "grad_norm": 1.0508335828781128, "learning_rate": 1.2298e-06, "loss": 0.0052, "step": 43852 }, { "epoch": 2.3379538305699206, "grad_norm": 1.9050589799880981, "learning_rate": 1.2296e-06, "loss": 0.018, "step": 43853 }, { "epoch": 2.3380071439995733, "grad_norm": 2.8332107067108154, "learning_rate": 1.2294e-06, "loss": 0.0161, "step": 43854 }, { "epoch": 2.3380604574292265, "grad_norm": 1.6423118114471436, "learning_rate": 1.2292e-06, "loss": 0.007, "step": 43855 }, { "epoch": 2.3381137708588793, "grad_norm": 2.4595260620117188, "learning_rate": 1.229e-06, "loss": 0.0215, "step": 43856 }, { "epoch": 2.338167084288532, "grad_norm": 2.9862592220306396, "learning_rate": 1.2288000000000002e-06, "loss": 0.0183, "step": 43857 }, { "epoch": 2.3382203977181852, "grad_norm": 0.6210223436355591, "learning_rate": 1.2286e-06, "loss": 0.003, "step": 43858 }, { "epoch": 2.338273711147838, "grad_norm": 1.6626553535461426, "learning_rate": 1.2284000000000002e-06, "loss": 0.0231, "step": 43859 }, { "epoch": 2.338327024577491, "grad_norm": 3.2416350841522217, "learning_rate": 1.2282000000000002e-06, "loss": 0.0364, "step": 43860 }, { "epoch": 2.338380338007144, "grad_norm": 2.137570858001709, "learning_rate": 1.2280000000000001e-06, "loss": 0.0261, "step": 43861 }, { "epoch": 2.338433651436797, "grad_norm": 2.0326569080352783, "learning_rate": 1.2278000000000001e-06, "loss": 0.0199, "step": 43862 }, { "epoch": 2.33848696486645, "grad_norm": 0.4857430160045624, "learning_rate": 1.2276000000000001e-06, "loss": 0.0025, "step": 43863 }, { "epoch": 2.3385402782961027, "grad_norm": 2.6114742755889893, "learning_rate": 1.2274e-06, "loss": 0.0192, "step": 43864 }, { "epoch": 2.338593591725756, "grad_norm": 1.1918202638626099, "learning_rate": 1.2272e-06, "loss": 0.0149, "step": 43865 }, { "epoch": 2.3386469051554086, "grad_norm": 0.9063137173652649, "learning_rate": 1.227e-06, "loss": 0.0058, "step": 43866 }, { "epoch": 2.3387002185850614, "grad_norm": 2.399069309234619, "learning_rate": 1.2268e-06, "loss": 0.0115, "step": 43867 }, { "epoch": 2.3387535320147146, "grad_norm": 2.565399169921875, "learning_rate": 1.2266000000000003e-06, "loss": 0.0211, "step": 43868 }, { "epoch": 2.3388068454443673, "grad_norm": 1.7328300476074219, "learning_rate": 1.2264e-06, "loss": 0.0241, "step": 43869 }, { "epoch": 2.3388601588740205, "grad_norm": 3.2151718139648438, "learning_rate": 1.2262000000000002e-06, "loss": 0.0176, "step": 43870 }, { "epoch": 2.3389134723036733, "grad_norm": 4.197755336761475, "learning_rate": 1.2260000000000002e-06, "loss": 0.0216, "step": 43871 }, { "epoch": 2.338966785733326, "grad_norm": 1.2983179092407227, "learning_rate": 1.2258e-06, "loss": 0.0058, "step": 43872 }, { "epoch": 2.3390200991629793, "grad_norm": 2.68894624710083, "learning_rate": 1.2256000000000002e-06, "loss": 0.0247, "step": 43873 }, { "epoch": 2.339073412592632, "grad_norm": 0.949803352355957, "learning_rate": 1.2254000000000002e-06, "loss": 0.0045, "step": 43874 }, { "epoch": 2.339126726022285, "grad_norm": 1.0738598108291626, "learning_rate": 1.2252000000000001e-06, "loss": 0.0047, "step": 43875 }, { "epoch": 2.339180039451938, "grad_norm": 2.271395444869995, "learning_rate": 1.2250000000000001e-06, "loss": 0.007, "step": 43876 }, { "epoch": 2.3392333528815907, "grad_norm": 0.9960248470306396, "learning_rate": 1.2248000000000001e-06, "loss": 0.0043, "step": 43877 }, { "epoch": 2.339286666311244, "grad_norm": 1.5386466979980469, "learning_rate": 1.2246e-06, "loss": 0.0254, "step": 43878 }, { "epoch": 2.3393399797408967, "grad_norm": 1.8968608379364014, "learning_rate": 1.2244e-06, "loss": 0.0286, "step": 43879 }, { "epoch": 2.3393932931705494, "grad_norm": 0.9618691802024841, "learning_rate": 1.2242e-06, "loss": 0.0175, "step": 43880 }, { "epoch": 2.3394466066002026, "grad_norm": 2.4880213737487793, "learning_rate": 1.224e-06, "loss": 0.0418, "step": 43881 }, { "epoch": 2.3394999200298554, "grad_norm": 1.8191016912460327, "learning_rate": 1.2238e-06, "loss": 0.0181, "step": 43882 }, { "epoch": 2.3395532334595086, "grad_norm": 0.296529620885849, "learning_rate": 1.2236e-06, "loss": 0.0022, "step": 43883 }, { "epoch": 2.3396065468891614, "grad_norm": 1.1780681610107422, "learning_rate": 1.2234000000000002e-06, "loss": 0.0207, "step": 43884 }, { "epoch": 2.339659860318814, "grad_norm": 0.9062765836715698, "learning_rate": 1.2232e-06, "loss": 0.0043, "step": 43885 }, { "epoch": 2.3397131737484673, "grad_norm": 1.814488410949707, "learning_rate": 1.2230000000000002e-06, "loss": 0.0244, "step": 43886 }, { "epoch": 2.33976648717812, "grad_norm": 1.997882604598999, "learning_rate": 1.2228000000000002e-06, "loss": 0.0103, "step": 43887 }, { "epoch": 2.3398198006077733, "grad_norm": 2.579681873321533, "learning_rate": 1.2226e-06, "loss": 0.0135, "step": 43888 }, { "epoch": 2.339873114037426, "grad_norm": 0.5787954926490784, "learning_rate": 1.2224000000000001e-06, "loss": 0.0028, "step": 43889 }, { "epoch": 2.339926427467079, "grad_norm": 2.895934581756592, "learning_rate": 1.2222000000000001e-06, "loss": 0.0119, "step": 43890 }, { "epoch": 2.339979740896732, "grad_norm": 2.934858560562134, "learning_rate": 1.2220000000000001e-06, "loss": 0.0155, "step": 43891 }, { "epoch": 2.3400330543263848, "grad_norm": 3.445706605911255, "learning_rate": 1.2218e-06, "loss": 0.0432, "step": 43892 }, { "epoch": 2.3400863677560375, "grad_norm": 0.43387699127197266, "learning_rate": 1.2216e-06, "loss": 0.0027, "step": 43893 }, { "epoch": 2.3401396811856907, "grad_norm": 0.5221739411354065, "learning_rate": 1.2214e-06, "loss": 0.0023, "step": 43894 }, { "epoch": 2.3401929946153435, "grad_norm": 1.1822367906570435, "learning_rate": 1.2212000000000003e-06, "loss": 0.0094, "step": 43895 }, { "epoch": 2.3402463080449967, "grad_norm": 2.073289394378662, "learning_rate": 1.221e-06, "loss": 0.0125, "step": 43896 }, { "epoch": 2.3402996214746494, "grad_norm": 1.2406061887741089, "learning_rate": 1.2208e-06, "loss": 0.0175, "step": 43897 }, { "epoch": 2.3403529349043026, "grad_norm": 1.2382919788360596, "learning_rate": 1.2206000000000002e-06, "loss": 0.0066, "step": 43898 }, { "epoch": 2.3404062483339554, "grad_norm": 1.9934072494506836, "learning_rate": 1.2204e-06, "loss": 0.0297, "step": 43899 }, { "epoch": 2.340459561763608, "grad_norm": 1.684237003326416, "learning_rate": 1.2202000000000002e-06, "loss": 0.022, "step": 43900 }, { "epoch": 2.3405128751932613, "grad_norm": 0.6960976719856262, "learning_rate": 1.2200000000000002e-06, "loss": 0.0026, "step": 43901 }, { "epoch": 2.340566188622914, "grad_norm": 2.3804187774658203, "learning_rate": 1.2198000000000002e-06, "loss": 0.0063, "step": 43902 }, { "epoch": 2.340619502052567, "grad_norm": 0.3102000057697296, "learning_rate": 1.2196000000000001e-06, "loss": 0.0027, "step": 43903 }, { "epoch": 2.34067281548222, "grad_norm": 1.489495873451233, "learning_rate": 1.2194000000000001e-06, "loss": 0.0121, "step": 43904 }, { "epoch": 2.340726128911873, "grad_norm": 2.6016156673431396, "learning_rate": 1.2192000000000001e-06, "loss": 0.0415, "step": 43905 }, { "epoch": 2.340779442341526, "grad_norm": 0.8808154463768005, "learning_rate": 1.219e-06, "loss": 0.0053, "step": 43906 }, { "epoch": 2.3408327557711788, "grad_norm": 2.446500539779663, "learning_rate": 1.2188e-06, "loss": 0.0257, "step": 43907 }, { "epoch": 2.3408860692008315, "grad_norm": 1.1225699186325073, "learning_rate": 1.2186e-06, "loss": 0.0054, "step": 43908 }, { "epoch": 2.3409393826304847, "grad_norm": 1.0370583534240723, "learning_rate": 1.2184e-06, "loss": 0.0208, "step": 43909 }, { "epoch": 2.3409926960601375, "grad_norm": 1.8194791078567505, "learning_rate": 1.2182e-06, "loss": 0.0033, "step": 43910 }, { "epoch": 2.3410460094897907, "grad_norm": 2.90775990486145, "learning_rate": 1.2180000000000002e-06, "loss": 0.0449, "step": 43911 }, { "epoch": 2.3410993229194434, "grad_norm": 1.562654733657837, "learning_rate": 1.2178e-06, "loss": 0.0067, "step": 43912 }, { "epoch": 2.341152636349096, "grad_norm": 8.322787284851074, "learning_rate": 1.2176e-06, "loss": 0.0377, "step": 43913 }, { "epoch": 2.3412059497787494, "grad_norm": 3.113607406616211, "learning_rate": 1.2174000000000002e-06, "loss": 0.0103, "step": 43914 }, { "epoch": 2.341259263208402, "grad_norm": 1.886609673500061, "learning_rate": 1.2172e-06, "loss": 0.0201, "step": 43915 }, { "epoch": 2.341312576638055, "grad_norm": 2.4699900150299072, "learning_rate": 1.2170000000000002e-06, "loss": 0.0122, "step": 43916 }, { "epoch": 2.341365890067708, "grad_norm": 1.5434372425079346, "learning_rate": 1.2168000000000001e-06, "loss": 0.0285, "step": 43917 }, { "epoch": 2.341419203497361, "grad_norm": 1.2461347579956055, "learning_rate": 1.2166000000000001e-06, "loss": 0.008, "step": 43918 }, { "epoch": 2.341472516927014, "grad_norm": 3.0491559505462646, "learning_rate": 1.2164000000000001e-06, "loss": 0.0105, "step": 43919 }, { "epoch": 2.341525830356667, "grad_norm": 1.6801540851593018, "learning_rate": 1.2162e-06, "loss": 0.0232, "step": 43920 }, { "epoch": 2.3415791437863196, "grad_norm": 4.468519687652588, "learning_rate": 1.216e-06, "loss": 0.0151, "step": 43921 }, { "epoch": 2.341632457215973, "grad_norm": 0.5982885360717773, "learning_rate": 1.2158e-06, "loss": 0.0043, "step": 43922 }, { "epoch": 2.3416857706456256, "grad_norm": 1.8674277067184448, "learning_rate": 1.2156e-06, "loss": 0.0246, "step": 43923 }, { "epoch": 2.3417390840752788, "grad_norm": 0.8953061699867249, "learning_rate": 1.2154e-06, "loss": 0.0059, "step": 43924 }, { "epoch": 2.3417923975049315, "grad_norm": 2.3222413063049316, "learning_rate": 1.2152e-06, "loss": 0.0208, "step": 43925 }, { "epoch": 2.3418457109345843, "grad_norm": 1.7267463207244873, "learning_rate": 1.215e-06, "loss": 0.0319, "step": 43926 }, { "epoch": 2.3418990243642375, "grad_norm": 1.171684980392456, "learning_rate": 1.2148000000000002e-06, "loss": 0.0066, "step": 43927 }, { "epoch": 2.3419523377938902, "grad_norm": 0.7627180814743042, "learning_rate": 1.2146e-06, "loss": 0.0154, "step": 43928 }, { "epoch": 2.342005651223543, "grad_norm": 0.5739158987998962, "learning_rate": 1.2144000000000002e-06, "loss": 0.0033, "step": 43929 }, { "epoch": 2.342058964653196, "grad_norm": 1.5012186765670776, "learning_rate": 1.2142000000000002e-06, "loss": 0.0093, "step": 43930 }, { "epoch": 2.342112278082849, "grad_norm": 3.368833303451538, "learning_rate": 1.214e-06, "loss": 0.0441, "step": 43931 }, { "epoch": 2.342165591512502, "grad_norm": 1.95073401927948, "learning_rate": 1.2138000000000001e-06, "loss": 0.0278, "step": 43932 }, { "epoch": 2.342218904942155, "grad_norm": 2.242744207382202, "learning_rate": 1.2136000000000001e-06, "loss": 0.0221, "step": 43933 }, { "epoch": 2.342272218371808, "grad_norm": 2.892380475997925, "learning_rate": 1.2134e-06, "loss": 0.0075, "step": 43934 }, { "epoch": 2.342325531801461, "grad_norm": 0.6071662902832031, "learning_rate": 1.2132e-06, "loss": 0.0034, "step": 43935 }, { "epoch": 2.3423788452311136, "grad_norm": 2.2381038665771484, "learning_rate": 1.213e-06, "loss": 0.008, "step": 43936 }, { "epoch": 2.342432158660767, "grad_norm": 1.0427125692367554, "learning_rate": 1.2128e-06, "loss": 0.0155, "step": 43937 }, { "epoch": 2.3424854720904196, "grad_norm": 2.2329001426696777, "learning_rate": 1.2126000000000003e-06, "loss": 0.0099, "step": 43938 }, { "epoch": 2.3425387855200723, "grad_norm": 2.671304225921631, "learning_rate": 1.2124e-06, "loss": 0.0173, "step": 43939 }, { "epoch": 2.3425920989497255, "grad_norm": 2.3582074642181396, "learning_rate": 1.2122e-06, "loss": 0.0072, "step": 43940 }, { "epoch": 2.3426454123793783, "grad_norm": 0.9445396661758423, "learning_rate": 1.2120000000000002e-06, "loss": 0.006, "step": 43941 }, { "epoch": 2.342698725809031, "grad_norm": 2.033818244934082, "learning_rate": 1.2118e-06, "loss": 0.0057, "step": 43942 }, { "epoch": 2.3427520392386842, "grad_norm": 1.0763627290725708, "learning_rate": 1.2116000000000002e-06, "loss": 0.0059, "step": 43943 }, { "epoch": 2.342805352668337, "grad_norm": 1.672942876815796, "learning_rate": 1.2114000000000002e-06, "loss": 0.0063, "step": 43944 }, { "epoch": 2.34285866609799, "grad_norm": 1.1348143815994263, "learning_rate": 1.2112000000000002e-06, "loss": 0.0047, "step": 43945 }, { "epoch": 2.342911979527643, "grad_norm": 2.3890836238861084, "learning_rate": 1.2110000000000001e-06, "loss": 0.0086, "step": 43946 }, { "epoch": 2.342965292957296, "grad_norm": 1.6161638498306274, "learning_rate": 1.2108000000000001e-06, "loss": 0.038, "step": 43947 }, { "epoch": 2.343018606386949, "grad_norm": 0.6080054044723511, "learning_rate": 1.2106000000000001e-06, "loss": 0.0039, "step": 43948 }, { "epoch": 2.3430719198166017, "grad_norm": 0.8994560837745667, "learning_rate": 1.2104e-06, "loss": 0.0042, "step": 43949 }, { "epoch": 2.343125233246255, "grad_norm": 1.2688285112380981, "learning_rate": 1.2102e-06, "loss": 0.0178, "step": 43950 }, { "epoch": 2.3431785466759076, "grad_norm": 1.7869353294372559, "learning_rate": 1.21e-06, "loss": 0.0095, "step": 43951 }, { "epoch": 2.3432318601055604, "grad_norm": 2.398810625076294, "learning_rate": 1.2098e-06, "loss": 0.0276, "step": 43952 }, { "epoch": 2.3432851735352136, "grad_norm": 0.5910571813583374, "learning_rate": 1.2096e-06, "loss": 0.0027, "step": 43953 }, { "epoch": 2.3433384869648664, "grad_norm": 0.5571306943893433, "learning_rate": 1.2094000000000002e-06, "loss": 0.0025, "step": 43954 }, { "epoch": 2.3433918003945196, "grad_norm": 1.7486510276794434, "learning_rate": 1.2092e-06, "loss": 0.0303, "step": 43955 }, { "epoch": 2.3434451138241723, "grad_norm": 1.8096401691436768, "learning_rate": 1.209e-06, "loss": 0.0148, "step": 43956 }, { "epoch": 2.343498427253825, "grad_norm": 1.12352454662323, "learning_rate": 1.2088000000000002e-06, "loss": 0.0161, "step": 43957 }, { "epoch": 2.3435517406834783, "grad_norm": 3.34688663482666, "learning_rate": 1.2086e-06, "loss": 0.0241, "step": 43958 }, { "epoch": 2.343605054113131, "grad_norm": 2.185438632965088, "learning_rate": 1.2084000000000002e-06, "loss": 0.0366, "step": 43959 }, { "epoch": 2.3436583675427842, "grad_norm": 2.1378345489501953, "learning_rate": 1.2082000000000001e-06, "loss": 0.0167, "step": 43960 }, { "epoch": 2.343711680972437, "grad_norm": 1.9594390392303467, "learning_rate": 1.2080000000000001e-06, "loss": 0.0054, "step": 43961 }, { "epoch": 2.3437649944020897, "grad_norm": 3.3273282051086426, "learning_rate": 1.2078000000000001e-06, "loss": 0.016, "step": 43962 }, { "epoch": 2.343818307831743, "grad_norm": 0.4418815076351166, "learning_rate": 1.2076e-06, "loss": 0.0023, "step": 43963 }, { "epoch": 2.3438716212613957, "grad_norm": 0.5532600283622742, "learning_rate": 1.2074e-06, "loss": 0.0044, "step": 43964 }, { "epoch": 2.3439249346910485, "grad_norm": 1.5044877529144287, "learning_rate": 1.2072e-06, "loss": 0.0384, "step": 43965 }, { "epoch": 2.3439782481207017, "grad_norm": 1.1062556505203247, "learning_rate": 1.207e-06, "loss": 0.005, "step": 43966 }, { "epoch": 2.3440315615503544, "grad_norm": 3.6480071544647217, "learning_rate": 1.2068e-06, "loss": 0.0104, "step": 43967 }, { "epoch": 2.3440848749800076, "grad_norm": 1.6721456050872803, "learning_rate": 1.2066000000000002e-06, "loss": 0.0121, "step": 43968 }, { "epoch": 2.3441381884096604, "grad_norm": 2.4409430027008057, "learning_rate": 1.2064e-06, "loss": 0.0247, "step": 43969 }, { "epoch": 2.344191501839313, "grad_norm": 0.4375966787338257, "learning_rate": 1.2062000000000002e-06, "loss": 0.0029, "step": 43970 }, { "epoch": 2.3442448152689663, "grad_norm": 2.634814739227295, "learning_rate": 1.2060000000000002e-06, "loss": 0.0138, "step": 43971 }, { "epoch": 2.344298128698619, "grad_norm": 3.7641048431396484, "learning_rate": 1.2058000000000002e-06, "loss": 0.0197, "step": 43972 }, { "epoch": 2.3443514421282723, "grad_norm": 2.1541266441345215, "learning_rate": 1.2056000000000002e-06, "loss": 0.0175, "step": 43973 }, { "epoch": 2.344404755557925, "grad_norm": 0.49060946702957153, "learning_rate": 1.2054000000000001e-06, "loss": 0.003, "step": 43974 }, { "epoch": 2.344458068987578, "grad_norm": 3.497340440750122, "learning_rate": 1.2052000000000001e-06, "loss": 0.0203, "step": 43975 }, { "epoch": 2.344511382417231, "grad_norm": 1.3809232711791992, "learning_rate": 1.2050000000000001e-06, "loss": 0.0065, "step": 43976 }, { "epoch": 2.3445646958468838, "grad_norm": 2.297370672225952, "learning_rate": 1.2048e-06, "loss": 0.0225, "step": 43977 }, { "epoch": 2.3446180092765365, "grad_norm": 0.7216565012931824, "learning_rate": 1.2046e-06, "loss": 0.0034, "step": 43978 }, { "epoch": 2.3446713227061897, "grad_norm": 2.167318820953369, "learning_rate": 1.2044e-06, "loss": 0.0284, "step": 43979 }, { "epoch": 2.3447246361358425, "grad_norm": 1.6745208501815796, "learning_rate": 1.2042e-06, "loss": 0.0071, "step": 43980 }, { "epoch": 2.3447779495654957, "grad_norm": 1.68770170211792, "learning_rate": 1.204e-06, "loss": 0.0084, "step": 43981 }, { "epoch": 2.3448312629951484, "grad_norm": 0.17184416949748993, "learning_rate": 1.2038e-06, "loss": 0.0018, "step": 43982 }, { "epoch": 2.3448845764248016, "grad_norm": 1.4542326927185059, "learning_rate": 1.2036e-06, "loss": 0.0045, "step": 43983 }, { "epoch": 2.3449378898544544, "grad_norm": 4.297450065612793, "learning_rate": 1.2034000000000002e-06, "loss": 0.044, "step": 43984 }, { "epoch": 2.344991203284107, "grad_norm": 3.74568510055542, "learning_rate": 1.2032e-06, "loss": 0.0118, "step": 43985 }, { "epoch": 2.3450445167137604, "grad_norm": 0.5342652797698975, "learning_rate": 1.2030000000000002e-06, "loss": 0.0026, "step": 43986 }, { "epoch": 2.345097830143413, "grad_norm": 2.8384718894958496, "learning_rate": 1.2028000000000002e-06, "loss": 0.0338, "step": 43987 }, { "epoch": 2.345151143573066, "grad_norm": 3.2675366401672363, "learning_rate": 1.2026000000000001e-06, "loss": 0.0142, "step": 43988 }, { "epoch": 2.345204457002719, "grad_norm": 1.9854532480239868, "learning_rate": 1.2024000000000001e-06, "loss": 0.0156, "step": 43989 }, { "epoch": 2.345257770432372, "grad_norm": 2.108065366744995, "learning_rate": 1.2022000000000001e-06, "loss": 0.0147, "step": 43990 }, { "epoch": 2.345311083862025, "grad_norm": 0.5916610956192017, "learning_rate": 1.202e-06, "loss": 0.0033, "step": 43991 }, { "epoch": 2.345364397291678, "grad_norm": 0.5590217113494873, "learning_rate": 1.2018e-06, "loss": 0.0033, "step": 43992 }, { "epoch": 2.3454177107213305, "grad_norm": 0.2932922840118408, "learning_rate": 1.2016e-06, "loss": 0.0025, "step": 43993 }, { "epoch": 2.3454710241509837, "grad_norm": 0.8642897009849548, "learning_rate": 1.2014e-06, "loss": 0.0044, "step": 43994 }, { "epoch": 2.3455243375806365, "grad_norm": 0.25167009234428406, "learning_rate": 1.2012000000000002e-06, "loss": 0.002, "step": 43995 }, { "epoch": 2.3455776510102897, "grad_norm": 3.1844139099121094, "learning_rate": 1.201e-06, "loss": 0.0092, "step": 43996 }, { "epoch": 2.3456309644399425, "grad_norm": 1.1596735715866089, "learning_rate": 1.2008000000000002e-06, "loss": 0.015, "step": 43997 }, { "epoch": 2.345684277869595, "grad_norm": 2.0860698223114014, "learning_rate": 1.2006000000000002e-06, "loss": 0.0308, "step": 43998 }, { "epoch": 2.3457375912992484, "grad_norm": 0.6105355620384216, "learning_rate": 1.2004e-06, "loss": 0.0031, "step": 43999 }, { "epoch": 2.345790904728901, "grad_norm": 1.620004415512085, "learning_rate": 1.2002000000000002e-06, "loss": 0.0177, "step": 44000 }, { "epoch": 2.345844218158554, "grad_norm": 1.085404872894287, "learning_rate": 1.2000000000000002e-06, "loss": 0.0048, "step": 44001 }, { "epoch": 2.345897531588207, "grad_norm": 1.87409245967865, "learning_rate": 1.1998000000000001e-06, "loss": 0.0126, "step": 44002 }, { "epoch": 2.34595084501786, "grad_norm": 5.326430320739746, "learning_rate": 1.1996000000000001e-06, "loss": 0.0442, "step": 44003 }, { "epoch": 2.346004158447513, "grad_norm": 1.6810877323150635, "learning_rate": 1.1994000000000001e-06, "loss": 0.0065, "step": 44004 }, { "epoch": 2.346057471877166, "grad_norm": 1.8362547159194946, "learning_rate": 1.1992e-06, "loss": 0.0118, "step": 44005 }, { "epoch": 2.3461107853068186, "grad_norm": 2.2357594966888428, "learning_rate": 1.199e-06, "loss": 0.0099, "step": 44006 }, { "epoch": 2.346164098736472, "grad_norm": 3.1668004989624023, "learning_rate": 1.1988e-06, "loss": 0.0335, "step": 44007 }, { "epoch": 2.3462174121661246, "grad_norm": 1.767503023147583, "learning_rate": 1.1986e-06, "loss": 0.0087, "step": 44008 }, { "epoch": 2.3462707255957778, "grad_norm": 1.411038875579834, "learning_rate": 1.1984e-06, "loss": 0.0059, "step": 44009 }, { "epoch": 2.3463240390254305, "grad_norm": 2.193150281906128, "learning_rate": 1.1982e-06, "loss": 0.0389, "step": 44010 }, { "epoch": 2.3463773524550833, "grad_norm": 1.4066263437271118, "learning_rate": 1.1980000000000002e-06, "loss": 0.017, "step": 44011 }, { "epoch": 2.3464306658847365, "grad_norm": 2.199817180633545, "learning_rate": 1.1978e-06, "loss": 0.0084, "step": 44012 }, { "epoch": 2.3464839793143892, "grad_norm": 2.8282346725463867, "learning_rate": 1.1976000000000002e-06, "loss": 0.032, "step": 44013 }, { "epoch": 2.346537292744042, "grad_norm": 4.602030277252197, "learning_rate": 1.1974000000000002e-06, "loss": 0.0489, "step": 44014 }, { "epoch": 2.346590606173695, "grad_norm": 1.0184754133224487, "learning_rate": 1.1972e-06, "loss": 0.0035, "step": 44015 }, { "epoch": 2.346643919603348, "grad_norm": 0.5845865607261658, "learning_rate": 1.1970000000000001e-06, "loss": 0.0023, "step": 44016 }, { "epoch": 2.346697233033001, "grad_norm": 2.252108335494995, "learning_rate": 1.1968000000000001e-06, "loss": 0.0075, "step": 44017 }, { "epoch": 2.346750546462654, "grad_norm": 3.4688713550567627, "learning_rate": 1.1966000000000001e-06, "loss": 0.0422, "step": 44018 }, { "epoch": 2.346803859892307, "grad_norm": 0.3391869068145752, "learning_rate": 1.1964e-06, "loss": 0.002, "step": 44019 }, { "epoch": 2.34685717332196, "grad_norm": 0.47171589732170105, "learning_rate": 1.1962e-06, "loss": 0.0035, "step": 44020 }, { "epoch": 2.3469104867516126, "grad_norm": 0.9336928129196167, "learning_rate": 1.196e-06, "loss": 0.0045, "step": 44021 }, { "epoch": 2.346963800181266, "grad_norm": 2.039275646209717, "learning_rate": 1.1958e-06, "loss": 0.017, "step": 44022 }, { "epoch": 2.3470171136109186, "grad_norm": 1.7129297256469727, "learning_rate": 1.1956e-06, "loss": 0.0188, "step": 44023 }, { "epoch": 2.3470704270405713, "grad_norm": 1.7743500471115112, "learning_rate": 1.1954e-06, "loss": 0.035, "step": 44024 }, { "epoch": 2.3471237404702245, "grad_norm": 0.1078752875328064, "learning_rate": 1.1952e-06, "loss": 0.0008, "step": 44025 }, { "epoch": 2.3471770538998773, "grad_norm": 3.6027209758758545, "learning_rate": 1.195e-06, "loss": 0.0284, "step": 44026 }, { "epoch": 2.3472303673295305, "grad_norm": 1.6059608459472656, "learning_rate": 1.1948000000000002e-06, "loss": 0.0386, "step": 44027 }, { "epoch": 2.3472836807591833, "grad_norm": 1.9247174263000488, "learning_rate": 1.1946e-06, "loss": 0.0222, "step": 44028 }, { "epoch": 2.347336994188836, "grad_norm": 0.9574758410453796, "learning_rate": 1.1944000000000002e-06, "loss": 0.0061, "step": 44029 }, { "epoch": 2.347390307618489, "grad_norm": 1.986143708229065, "learning_rate": 1.1942000000000001e-06, "loss": 0.0094, "step": 44030 }, { "epoch": 2.347443621048142, "grad_norm": 1.4657827615737915, "learning_rate": 1.1940000000000001e-06, "loss": 0.0149, "step": 44031 }, { "epoch": 2.347496934477795, "grad_norm": 3.008657693862915, "learning_rate": 1.1938000000000001e-06, "loss": 0.027, "step": 44032 }, { "epoch": 2.347550247907448, "grad_norm": 1.5694960355758667, "learning_rate": 1.1936e-06, "loss": 0.0258, "step": 44033 }, { "epoch": 2.3476035613371007, "grad_norm": 2.7855968475341797, "learning_rate": 1.1934e-06, "loss": 0.012, "step": 44034 }, { "epoch": 2.347656874766754, "grad_norm": 1.0640466213226318, "learning_rate": 1.1932e-06, "loss": 0.0141, "step": 44035 }, { "epoch": 2.3477101881964066, "grad_norm": 2.291036367416382, "learning_rate": 1.193e-06, "loss": 0.0208, "step": 44036 }, { "epoch": 2.3477635016260594, "grad_norm": 2.1323189735412598, "learning_rate": 1.1928e-06, "loss": 0.0106, "step": 44037 }, { "epoch": 2.3478168150557126, "grad_norm": 1.6768170595169067, "learning_rate": 1.1926000000000002e-06, "loss": 0.0074, "step": 44038 }, { "epoch": 2.3478701284853654, "grad_norm": 3.6452245712280273, "learning_rate": 1.1924e-06, "loss": 0.0309, "step": 44039 }, { "epoch": 2.3479234419150186, "grad_norm": 2.2088379859924316, "learning_rate": 1.1922000000000002e-06, "loss": 0.009, "step": 44040 }, { "epoch": 2.3479767553446713, "grad_norm": 2.0899658203125, "learning_rate": 1.1920000000000002e-06, "loss": 0.0214, "step": 44041 }, { "epoch": 2.348030068774324, "grad_norm": 1.3899506330490112, "learning_rate": 1.1918e-06, "loss": 0.0188, "step": 44042 }, { "epoch": 2.3480833822039773, "grad_norm": 1.7831445932388306, "learning_rate": 1.1916000000000002e-06, "loss": 0.011, "step": 44043 }, { "epoch": 2.34813669563363, "grad_norm": 0.7217293381690979, "learning_rate": 1.1914000000000001e-06, "loss": 0.0046, "step": 44044 }, { "epoch": 2.3481900090632832, "grad_norm": 2.2061643600463867, "learning_rate": 1.1912000000000001e-06, "loss": 0.0062, "step": 44045 }, { "epoch": 2.348243322492936, "grad_norm": 2.5832130908966064, "learning_rate": 1.1910000000000001e-06, "loss": 0.0091, "step": 44046 }, { "epoch": 2.3482966359225887, "grad_norm": 1.5890381336212158, "learning_rate": 1.1908e-06, "loss": 0.0163, "step": 44047 }, { "epoch": 2.348349949352242, "grad_norm": 0.7878063321113586, "learning_rate": 1.1906e-06, "loss": 0.0052, "step": 44048 }, { "epoch": 2.3484032627818947, "grad_norm": 4.858794212341309, "learning_rate": 1.1904e-06, "loss": 0.0371, "step": 44049 }, { "epoch": 2.3484565762115475, "grad_norm": 1.6571780443191528, "learning_rate": 1.1902e-06, "loss": 0.0108, "step": 44050 }, { "epoch": 2.3485098896412007, "grad_norm": 2.4718992710113525, "learning_rate": 1.19e-06, "loss": 0.0333, "step": 44051 }, { "epoch": 2.3485632030708534, "grad_norm": 3.937162399291992, "learning_rate": 1.1898e-06, "loss": 0.0265, "step": 44052 }, { "epoch": 2.3486165165005066, "grad_norm": 1.2694334983825684, "learning_rate": 1.1896e-06, "loss": 0.0054, "step": 44053 }, { "epoch": 2.3486698299301594, "grad_norm": 1.0636842250823975, "learning_rate": 1.1894000000000002e-06, "loss": 0.011, "step": 44054 }, { "epoch": 2.3487231433598126, "grad_norm": 4.261735916137695, "learning_rate": 1.1892e-06, "loss": 0.0121, "step": 44055 }, { "epoch": 2.3487764567894653, "grad_norm": 1.2874704599380493, "learning_rate": 1.1890000000000002e-06, "loss": 0.0182, "step": 44056 }, { "epoch": 2.348829770219118, "grad_norm": 1.2117152214050293, "learning_rate": 1.1888000000000002e-06, "loss": 0.0053, "step": 44057 }, { "epoch": 2.3488830836487713, "grad_norm": 1.5808401107788086, "learning_rate": 1.1886e-06, "loss": 0.0198, "step": 44058 }, { "epoch": 2.348936397078424, "grad_norm": 1.2825297117233276, "learning_rate": 1.1884000000000001e-06, "loss": 0.0198, "step": 44059 }, { "epoch": 2.348989710508077, "grad_norm": 2.4045212268829346, "learning_rate": 1.1882000000000001e-06, "loss": 0.0206, "step": 44060 }, { "epoch": 2.34904302393773, "grad_norm": 2.3530540466308594, "learning_rate": 1.188e-06, "loss": 0.0468, "step": 44061 }, { "epoch": 2.3490963373673828, "grad_norm": 0.09117665886878967, "learning_rate": 1.1878e-06, "loss": 0.001, "step": 44062 }, { "epoch": 2.3491496507970355, "grad_norm": 0.2759879231452942, "learning_rate": 1.1876e-06, "loss": 0.0024, "step": 44063 }, { "epoch": 2.3492029642266887, "grad_norm": 3.0024280548095703, "learning_rate": 1.1874e-06, "loss": 0.0447, "step": 44064 }, { "epoch": 2.3492562776563415, "grad_norm": 0.6528860926628113, "learning_rate": 1.1872000000000003e-06, "loss": 0.0037, "step": 44065 }, { "epoch": 2.3493095910859947, "grad_norm": 1.990181565284729, "learning_rate": 1.187e-06, "loss": 0.0259, "step": 44066 }, { "epoch": 2.3493629045156474, "grad_norm": 1.7694226503372192, "learning_rate": 1.1868e-06, "loss": 0.0171, "step": 44067 }, { "epoch": 2.3494162179453006, "grad_norm": 2.905207872390747, "learning_rate": 1.1866000000000002e-06, "loss": 0.0167, "step": 44068 }, { "epoch": 2.3494695313749534, "grad_norm": 0.97943514585495, "learning_rate": 1.1864e-06, "loss": 0.0193, "step": 44069 }, { "epoch": 2.349522844804606, "grad_norm": 0.1838756799697876, "learning_rate": 1.1862000000000002e-06, "loss": 0.0015, "step": 44070 }, { "epoch": 2.3495761582342594, "grad_norm": 1.7508786916732788, "learning_rate": 1.1860000000000002e-06, "loss": 0.01, "step": 44071 }, { "epoch": 2.349629471663912, "grad_norm": 1.2155735492706299, "learning_rate": 1.1858000000000001e-06, "loss": 0.0122, "step": 44072 }, { "epoch": 2.349682785093565, "grad_norm": 0.2536737620830536, "learning_rate": 1.1856000000000001e-06, "loss": 0.0017, "step": 44073 }, { "epoch": 2.349736098523218, "grad_norm": 1.1383451223373413, "learning_rate": 1.1854000000000001e-06, "loss": 0.0043, "step": 44074 }, { "epoch": 2.349789411952871, "grad_norm": 1.9608631134033203, "learning_rate": 1.1852e-06, "loss": 0.0335, "step": 44075 }, { "epoch": 2.349842725382524, "grad_norm": 1.549010992050171, "learning_rate": 1.185e-06, "loss": 0.0049, "step": 44076 }, { "epoch": 2.349896038812177, "grad_norm": 3.234269142150879, "learning_rate": 1.1848e-06, "loss": 0.0304, "step": 44077 }, { "epoch": 2.3499493522418295, "grad_norm": 2.2079696655273438, "learning_rate": 1.1846e-06, "loss": 0.0318, "step": 44078 }, { "epoch": 2.3500026656714827, "grad_norm": 0.40087956190109253, "learning_rate": 1.1844e-06, "loss": 0.0021, "step": 44079 }, { "epoch": 2.3500559791011355, "grad_norm": 2.7405033111572266, "learning_rate": 1.1842e-06, "loss": 0.0388, "step": 44080 }, { "epoch": 2.3501092925307887, "grad_norm": 1.1149520874023438, "learning_rate": 1.1840000000000002e-06, "loss": 0.0057, "step": 44081 }, { "epoch": 2.3501626059604415, "grad_norm": 1.8225187063217163, "learning_rate": 1.1838e-06, "loss": 0.0124, "step": 44082 }, { "epoch": 2.350215919390094, "grad_norm": 0.38180291652679443, "learning_rate": 1.1836000000000002e-06, "loss": 0.0033, "step": 44083 }, { "epoch": 2.3502692328197474, "grad_norm": 0.9096300005912781, "learning_rate": 1.1834000000000002e-06, "loss": 0.0168, "step": 44084 }, { "epoch": 2.3503225462494, "grad_norm": 0.9172008633613586, "learning_rate": 1.1832e-06, "loss": 0.0103, "step": 44085 }, { "epoch": 2.350375859679053, "grad_norm": 0.3359275758266449, "learning_rate": 1.1830000000000002e-06, "loss": 0.0026, "step": 44086 }, { "epoch": 2.350429173108706, "grad_norm": 0.5481169819831848, "learning_rate": 1.1828000000000001e-06, "loss": 0.0032, "step": 44087 }, { "epoch": 2.350482486538359, "grad_norm": 0.6811811923980713, "learning_rate": 1.1826000000000001e-06, "loss": 0.0026, "step": 44088 }, { "epoch": 2.350535799968012, "grad_norm": 1.3973747491836548, "learning_rate": 1.1824e-06, "loss": 0.0105, "step": 44089 }, { "epoch": 2.350589113397665, "grad_norm": 1.32598876953125, "learning_rate": 1.1822e-06, "loss": 0.004, "step": 44090 }, { "epoch": 2.3506424268273176, "grad_norm": 1.591068983078003, "learning_rate": 1.182e-06, "loss": 0.0244, "step": 44091 }, { "epoch": 2.350695740256971, "grad_norm": 0.3456033170223236, "learning_rate": 1.1818e-06, "loss": 0.0025, "step": 44092 }, { "epoch": 2.3507490536866236, "grad_norm": 2.7570948600769043, "learning_rate": 1.1816e-06, "loss": 0.0199, "step": 44093 }, { "epoch": 2.3508023671162768, "grad_norm": 1.4081251621246338, "learning_rate": 1.1814e-06, "loss": 0.0073, "step": 44094 }, { "epoch": 2.3508556805459295, "grad_norm": 1.6770853996276855, "learning_rate": 1.1812000000000002e-06, "loss": 0.0242, "step": 44095 }, { "epoch": 2.3509089939755823, "grad_norm": 1.3102600574493408, "learning_rate": 1.181e-06, "loss": 0.0142, "step": 44096 }, { "epoch": 2.3509623074052355, "grad_norm": 2.4780611991882324, "learning_rate": 1.1808000000000002e-06, "loss": 0.041, "step": 44097 }, { "epoch": 2.3510156208348882, "grad_norm": 1.6146944761276245, "learning_rate": 1.1806000000000002e-06, "loss": 0.0086, "step": 44098 }, { "epoch": 2.351068934264541, "grad_norm": 1.1190643310546875, "learning_rate": 1.1804000000000002e-06, "loss": 0.0322, "step": 44099 }, { "epoch": 2.351122247694194, "grad_norm": 0.9422294497489929, "learning_rate": 1.1802000000000002e-06, "loss": 0.0063, "step": 44100 }, { "epoch": 2.351175561123847, "grad_norm": 2.5447278022766113, "learning_rate": 1.1800000000000001e-06, "loss": 0.0254, "step": 44101 }, { "epoch": 2.3512288745535, "grad_norm": 1.0918041467666626, "learning_rate": 1.1798000000000001e-06, "loss": 0.0222, "step": 44102 }, { "epoch": 2.351282187983153, "grad_norm": 1.216268539428711, "learning_rate": 1.1796000000000001e-06, "loss": 0.0057, "step": 44103 }, { "epoch": 2.351335501412806, "grad_norm": 1.2264899015426636, "learning_rate": 1.1794e-06, "loss": 0.0064, "step": 44104 }, { "epoch": 2.351388814842459, "grad_norm": 1.24940824508667, "learning_rate": 1.1792e-06, "loss": 0.0181, "step": 44105 }, { "epoch": 2.3514421282721116, "grad_norm": 0.4010467231273651, "learning_rate": 1.179e-06, "loss": 0.0024, "step": 44106 }, { "epoch": 2.351495441701765, "grad_norm": 0.6778977513313293, "learning_rate": 1.1788e-06, "loss": 0.0033, "step": 44107 }, { "epoch": 2.3515487551314176, "grad_norm": 0.4231516420841217, "learning_rate": 1.1786000000000002e-06, "loss": 0.0029, "step": 44108 }, { "epoch": 2.3516020685610703, "grad_norm": 2.284151315689087, "learning_rate": 1.1784e-06, "loss": 0.027, "step": 44109 }, { "epoch": 2.3516553819907235, "grad_norm": 1.1655073165893555, "learning_rate": 1.1782e-06, "loss": 0.0067, "step": 44110 }, { "epoch": 2.3517086954203763, "grad_norm": 1.9715455770492554, "learning_rate": 1.1780000000000002e-06, "loss": 0.0205, "step": 44111 }, { "epoch": 2.3517620088500295, "grad_norm": 2.21628737449646, "learning_rate": 1.1778e-06, "loss": 0.0134, "step": 44112 }, { "epoch": 2.3518153222796823, "grad_norm": 2.3799242973327637, "learning_rate": 1.1776000000000002e-06, "loss": 0.0631, "step": 44113 }, { "epoch": 2.351868635709335, "grad_norm": 0.5749064087867737, "learning_rate": 1.1774000000000002e-06, "loss": 0.0032, "step": 44114 }, { "epoch": 2.351921949138988, "grad_norm": 2.3141841888427734, "learning_rate": 1.1772000000000001e-06, "loss": 0.0111, "step": 44115 }, { "epoch": 2.351975262568641, "grad_norm": 3.5069241523742676, "learning_rate": 1.1770000000000001e-06, "loss": 0.0138, "step": 44116 }, { "epoch": 2.352028575998294, "grad_norm": 0.3908647894859314, "learning_rate": 1.1768000000000001e-06, "loss": 0.0019, "step": 44117 }, { "epoch": 2.352081889427947, "grad_norm": 2.4342780113220215, "learning_rate": 1.1766e-06, "loss": 0.0101, "step": 44118 }, { "epoch": 2.3521352028575997, "grad_norm": 2.020660638809204, "learning_rate": 1.1764e-06, "loss": 0.0095, "step": 44119 }, { "epoch": 2.352188516287253, "grad_norm": 2.694549560546875, "learning_rate": 1.1762e-06, "loss": 0.0121, "step": 44120 }, { "epoch": 2.3522418297169057, "grad_norm": 2.4872655868530273, "learning_rate": 1.176e-06, "loss": 0.0289, "step": 44121 }, { "epoch": 2.3522951431465584, "grad_norm": 2.4538867473602295, "learning_rate": 1.1758e-06, "loss": 0.0065, "step": 44122 }, { "epoch": 2.3523484565762116, "grad_norm": 1.0643337965011597, "learning_rate": 1.1756e-06, "loss": 0.02, "step": 44123 }, { "epoch": 2.3524017700058644, "grad_norm": 3.068737506866455, "learning_rate": 1.1754000000000002e-06, "loss": 0.042, "step": 44124 }, { "epoch": 2.3524550834355176, "grad_norm": 1.4782097339630127, "learning_rate": 1.1752e-06, "loss": 0.0356, "step": 44125 }, { "epoch": 2.3525083968651703, "grad_norm": 0.5959217548370361, "learning_rate": 1.175e-06, "loss": 0.0034, "step": 44126 }, { "epoch": 2.352561710294823, "grad_norm": 2.1554672718048096, "learning_rate": 1.1748000000000002e-06, "loss": 0.0051, "step": 44127 }, { "epoch": 2.3526150237244763, "grad_norm": 2.379028797149658, "learning_rate": 1.1746e-06, "loss": 0.0714, "step": 44128 }, { "epoch": 2.352668337154129, "grad_norm": 1.347025990486145, "learning_rate": 1.1744000000000001e-06, "loss": 0.0058, "step": 44129 }, { "epoch": 2.3527216505837822, "grad_norm": 1.6202465295791626, "learning_rate": 1.1742000000000001e-06, "loss": 0.0164, "step": 44130 }, { "epoch": 2.352774964013435, "grad_norm": 2.0691986083984375, "learning_rate": 1.1740000000000001e-06, "loss": 0.0425, "step": 44131 }, { "epoch": 2.3528282774430878, "grad_norm": 0.5151306390762329, "learning_rate": 1.1738e-06, "loss": 0.002, "step": 44132 }, { "epoch": 2.352881590872741, "grad_norm": 0.250271737575531, "learning_rate": 1.1736e-06, "loss": 0.0023, "step": 44133 }, { "epoch": 2.3529349043023937, "grad_norm": 1.866952657699585, "learning_rate": 1.1734e-06, "loss": 0.0302, "step": 44134 }, { "epoch": 2.3529882177320465, "grad_norm": 2.697467565536499, "learning_rate": 1.1732e-06, "loss": 0.0318, "step": 44135 }, { "epoch": 2.3530415311616997, "grad_norm": 0.39603492617607117, "learning_rate": 1.173e-06, "loss": 0.0037, "step": 44136 }, { "epoch": 2.3530948445913524, "grad_norm": 1.7836196422576904, "learning_rate": 1.1728e-06, "loss": 0.0218, "step": 44137 }, { "epoch": 2.3531481580210056, "grad_norm": 2.6603500843048096, "learning_rate": 1.1726000000000002e-06, "loss": 0.0207, "step": 44138 }, { "epoch": 2.3532014714506584, "grad_norm": 3.1082069873809814, "learning_rate": 1.1724e-06, "loss": 0.0111, "step": 44139 }, { "epoch": 2.3532547848803116, "grad_norm": 1.9113901853561401, "learning_rate": 1.1722000000000002e-06, "loss": 0.0271, "step": 44140 }, { "epoch": 2.3533080983099643, "grad_norm": 2.638256788253784, "learning_rate": 1.1720000000000002e-06, "loss": 0.0327, "step": 44141 }, { "epoch": 2.353361411739617, "grad_norm": 1.682645320892334, "learning_rate": 1.1718000000000002e-06, "loss": 0.0051, "step": 44142 }, { "epoch": 2.3534147251692703, "grad_norm": 3.642559766769409, "learning_rate": 1.1716000000000001e-06, "loss": 0.022, "step": 44143 }, { "epoch": 2.353468038598923, "grad_norm": 1.3001612424850464, "learning_rate": 1.1714000000000001e-06, "loss": 0.0138, "step": 44144 }, { "epoch": 2.353521352028576, "grad_norm": 3.1645617485046387, "learning_rate": 1.1712000000000001e-06, "loss": 0.0209, "step": 44145 }, { "epoch": 2.353574665458229, "grad_norm": 4.323349475860596, "learning_rate": 1.171e-06, "loss": 0.0573, "step": 44146 }, { "epoch": 2.3536279788878818, "grad_norm": 1.1927330493927002, "learning_rate": 1.1708e-06, "loss": 0.0056, "step": 44147 }, { "epoch": 2.353681292317535, "grad_norm": 0.902839720249176, "learning_rate": 1.1706e-06, "loss": 0.0088, "step": 44148 }, { "epoch": 2.3537346057471877, "grad_norm": 1.0306079387664795, "learning_rate": 1.1704e-06, "loss": 0.0042, "step": 44149 }, { "epoch": 2.3537879191768405, "grad_norm": 2.5007576942443848, "learning_rate": 1.1702e-06, "loss": 0.0353, "step": 44150 }, { "epoch": 2.3538412326064937, "grad_norm": 1.1632388830184937, "learning_rate": 1.1700000000000002e-06, "loss": 0.0161, "step": 44151 }, { "epoch": 2.3538945460361465, "grad_norm": 3.008984088897705, "learning_rate": 1.1698e-06, "loss": 0.047, "step": 44152 }, { "epoch": 2.3539478594657997, "grad_norm": 0.7249323129653931, "learning_rate": 1.1696e-06, "loss": 0.0026, "step": 44153 }, { "epoch": 2.3540011728954524, "grad_norm": 1.4823088645935059, "learning_rate": 1.1694000000000002e-06, "loss": 0.0154, "step": 44154 }, { "epoch": 2.354054486325105, "grad_norm": 1.9265382289886475, "learning_rate": 1.1692e-06, "loss": 0.0089, "step": 44155 }, { "epoch": 2.3541077997547584, "grad_norm": 3.972888946533203, "learning_rate": 1.1690000000000002e-06, "loss": 0.0349, "step": 44156 }, { "epoch": 2.354161113184411, "grad_norm": 1.6239639520645142, "learning_rate": 1.1688000000000001e-06, "loss": 0.008, "step": 44157 }, { "epoch": 2.354214426614064, "grad_norm": 5.012071132659912, "learning_rate": 1.1686000000000001e-06, "loss": 0.0224, "step": 44158 }, { "epoch": 2.354267740043717, "grad_norm": 1.8535261154174805, "learning_rate": 1.1684000000000001e-06, "loss": 0.0215, "step": 44159 }, { "epoch": 2.35432105347337, "grad_norm": 1.120051383972168, "learning_rate": 1.1682e-06, "loss": 0.0215, "step": 44160 }, { "epoch": 2.354374366903023, "grad_norm": 0.5193961262702942, "learning_rate": 1.168e-06, "loss": 0.004, "step": 44161 }, { "epoch": 2.354427680332676, "grad_norm": 3.881528377532959, "learning_rate": 1.1678e-06, "loss": 0.0125, "step": 44162 }, { "epoch": 2.3544809937623286, "grad_norm": 2.1434030532836914, "learning_rate": 1.1676e-06, "loss": 0.0137, "step": 44163 }, { "epoch": 2.3545343071919818, "grad_norm": 3.446441888809204, "learning_rate": 1.1674e-06, "loss": 0.0196, "step": 44164 }, { "epoch": 2.3545876206216345, "grad_norm": 2.4246456623077393, "learning_rate": 1.1672000000000002e-06, "loss": 0.0124, "step": 44165 }, { "epoch": 2.3546409340512877, "grad_norm": 3.382849931716919, "learning_rate": 1.167e-06, "loss": 0.0176, "step": 44166 }, { "epoch": 2.3546942474809405, "grad_norm": 0.9111109972000122, "learning_rate": 1.1668000000000002e-06, "loss": 0.0057, "step": 44167 }, { "epoch": 2.3547475609105932, "grad_norm": 1.9061936140060425, "learning_rate": 1.1666000000000002e-06, "loss": 0.0478, "step": 44168 }, { "epoch": 2.3548008743402464, "grad_norm": 0.998729407787323, "learning_rate": 1.1664e-06, "loss": 0.0168, "step": 44169 }, { "epoch": 2.354854187769899, "grad_norm": 0.9524508714675903, "learning_rate": 1.1662000000000002e-06, "loss": 0.0072, "step": 44170 }, { "epoch": 2.354907501199552, "grad_norm": 2.160585880279541, "learning_rate": 1.1660000000000001e-06, "loss": 0.0153, "step": 44171 }, { "epoch": 2.354960814629205, "grad_norm": 1.1073654890060425, "learning_rate": 1.1658000000000001e-06, "loss": 0.006, "step": 44172 }, { "epoch": 2.355014128058858, "grad_norm": 3.819469451904297, "learning_rate": 1.1656000000000001e-06, "loss": 0.0262, "step": 44173 }, { "epoch": 2.355067441488511, "grad_norm": 0.3232875466346741, "learning_rate": 1.1654e-06, "loss": 0.0022, "step": 44174 }, { "epoch": 2.355120754918164, "grad_norm": 1.1512502431869507, "learning_rate": 1.1652e-06, "loss": 0.0135, "step": 44175 }, { "epoch": 2.355174068347817, "grad_norm": 1.3667161464691162, "learning_rate": 1.165e-06, "loss": 0.0203, "step": 44176 }, { "epoch": 2.35522738177747, "grad_norm": 0.651545524597168, "learning_rate": 1.1648e-06, "loss": 0.0036, "step": 44177 }, { "epoch": 2.3552806952071226, "grad_norm": 1.7065495252609253, "learning_rate": 1.1646e-06, "loss": 0.0093, "step": 44178 }, { "epoch": 2.3553340086367758, "grad_norm": 3.739732027053833, "learning_rate": 1.1644e-06, "loss": 0.0096, "step": 44179 }, { "epoch": 2.3553873220664285, "grad_norm": 1.9886494874954224, "learning_rate": 1.1642e-06, "loss": 0.0243, "step": 44180 }, { "epoch": 2.3554406354960813, "grad_norm": 1.5370670557022095, "learning_rate": 1.1640000000000002e-06, "loss": 0.008, "step": 44181 }, { "epoch": 2.3554939489257345, "grad_norm": 2.6948459148406982, "learning_rate": 1.1638e-06, "loss": 0.018, "step": 44182 }, { "epoch": 2.3555472623553873, "grad_norm": 1.8396952152252197, "learning_rate": 1.1636000000000002e-06, "loss": 0.0254, "step": 44183 }, { "epoch": 2.35560057578504, "grad_norm": 2.368076801300049, "learning_rate": 1.1634000000000002e-06, "loss": 0.0108, "step": 44184 }, { "epoch": 2.355653889214693, "grad_norm": 0.603251576423645, "learning_rate": 1.1632000000000001e-06, "loss": 0.0126, "step": 44185 }, { "epoch": 2.355707202644346, "grad_norm": 1.7003793716430664, "learning_rate": 1.1630000000000001e-06, "loss": 0.0164, "step": 44186 }, { "epoch": 2.355760516073999, "grad_norm": 0.3936142325401306, "learning_rate": 1.1628000000000001e-06, "loss": 0.0036, "step": 44187 }, { "epoch": 2.355813829503652, "grad_norm": 2.051558494567871, "learning_rate": 1.1626e-06, "loss": 0.026, "step": 44188 }, { "epoch": 2.355867142933305, "grad_norm": 0.7139763832092285, "learning_rate": 1.1624e-06, "loss": 0.0048, "step": 44189 }, { "epoch": 2.355920456362958, "grad_norm": 1.4653613567352295, "learning_rate": 1.1622e-06, "loss": 0.0234, "step": 44190 }, { "epoch": 2.3559737697926106, "grad_norm": 1.090904951095581, "learning_rate": 1.162e-06, "loss": 0.0058, "step": 44191 }, { "epoch": 2.356027083222264, "grad_norm": 2.3000099658966064, "learning_rate": 1.1618000000000003e-06, "loss": 0.0096, "step": 44192 }, { "epoch": 2.3560803966519166, "grad_norm": 1.3018255233764648, "learning_rate": 1.1616e-06, "loss": 0.0223, "step": 44193 }, { "epoch": 2.3561337100815694, "grad_norm": 1.4967280626296997, "learning_rate": 1.1614e-06, "loss": 0.0382, "step": 44194 }, { "epoch": 2.3561870235112226, "grad_norm": 1.4048625230789185, "learning_rate": 1.1612000000000002e-06, "loss": 0.0063, "step": 44195 }, { "epoch": 2.3562403369408753, "grad_norm": 0.43994763493537903, "learning_rate": 1.161e-06, "loss": 0.0043, "step": 44196 }, { "epoch": 2.3562936503705285, "grad_norm": 1.8871729373931885, "learning_rate": 1.1608000000000002e-06, "loss": 0.0199, "step": 44197 }, { "epoch": 2.3563469638001813, "grad_norm": 0.5728090405464172, "learning_rate": 1.1606000000000002e-06, "loss": 0.0033, "step": 44198 }, { "epoch": 2.356400277229834, "grad_norm": 1.0880138874053955, "learning_rate": 1.1604000000000001e-06, "loss": 0.0143, "step": 44199 }, { "epoch": 2.3564535906594872, "grad_norm": 1.0596957206726074, "learning_rate": 1.1602000000000001e-06, "loss": 0.012, "step": 44200 }, { "epoch": 2.35650690408914, "grad_norm": 1.9763190746307373, "learning_rate": 1.1600000000000001e-06, "loss": 0.0351, "step": 44201 }, { "epoch": 2.356560217518793, "grad_norm": 1.5167639255523682, "learning_rate": 1.1598e-06, "loss": 0.0119, "step": 44202 }, { "epoch": 2.356613530948446, "grad_norm": 2.3190066814422607, "learning_rate": 1.1596e-06, "loss": 0.0226, "step": 44203 }, { "epoch": 2.3566668443780987, "grad_norm": 1.5603742599487305, "learning_rate": 1.1594e-06, "loss": 0.0365, "step": 44204 }, { "epoch": 2.356720157807752, "grad_norm": 2.8580219745635986, "learning_rate": 1.1592e-06, "loss": 0.0174, "step": 44205 }, { "epoch": 2.3567734712374047, "grad_norm": 2.2545502185821533, "learning_rate": 1.159e-06, "loss": 0.0255, "step": 44206 }, { "epoch": 2.3568267846670574, "grad_norm": 0.845611572265625, "learning_rate": 1.1588e-06, "loss": 0.0032, "step": 44207 }, { "epoch": 2.3568800980967106, "grad_norm": 1.1301542520523071, "learning_rate": 1.1586000000000002e-06, "loss": 0.0085, "step": 44208 }, { "epoch": 2.3569334115263634, "grad_norm": 1.0402593612670898, "learning_rate": 1.1584e-06, "loss": 0.005, "step": 44209 }, { "epoch": 2.3569867249560166, "grad_norm": 1.407626748085022, "learning_rate": 1.1582000000000002e-06, "loss": 0.0217, "step": 44210 }, { "epoch": 2.3570400383856693, "grad_norm": 2.042722463607788, "learning_rate": 1.1580000000000002e-06, "loss": 0.0132, "step": 44211 }, { "epoch": 2.357093351815322, "grad_norm": 4.111589431762695, "learning_rate": 1.1578e-06, "loss": 0.0369, "step": 44212 }, { "epoch": 2.3571466652449753, "grad_norm": 1.722867727279663, "learning_rate": 1.1576000000000001e-06, "loss": 0.0098, "step": 44213 }, { "epoch": 2.357199978674628, "grad_norm": 0.37965211272239685, "learning_rate": 1.1574000000000001e-06, "loss": 0.0037, "step": 44214 }, { "epoch": 2.3572532921042813, "grad_norm": 1.2127902507781982, "learning_rate": 1.1572000000000001e-06, "loss": 0.0051, "step": 44215 }, { "epoch": 2.357306605533934, "grad_norm": 0.9831783771514893, "learning_rate": 1.157e-06, "loss": 0.0136, "step": 44216 }, { "epoch": 2.3573599189635868, "grad_norm": 2.9023947715759277, "learning_rate": 1.1568e-06, "loss": 0.0174, "step": 44217 }, { "epoch": 2.35741323239324, "grad_norm": 4.144817352294922, "learning_rate": 1.1566e-06, "loss": 0.008, "step": 44218 }, { "epoch": 2.3574665458228927, "grad_norm": 3.427516222000122, "learning_rate": 1.1564e-06, "loss": 0.0194, "step": 44219 }, { "epoch": 2.3575198592525455, "grad_norm": 1.6940710544586182, "learning_rate": 1.1562e-06, "loss": 0.0115, "step": 44220 }, { "epoch": 2.3575731726821987, "grad_norm": 3.098142385482788, "learning_rate": 1.156e-06, "loss": 0.0131, "step": 44221 }, { "epoch": 2.3576264861118514, "grad_norm": 1.1579281091690063, "learning_rate": 1.1558e-06, "loss": 0.0108, "step": 44222 }, { "epoch": 2.3576797995415046, "grad_norm": 0.5913028717041016, "learning_rate": 1.1556e-06, "loss": 0.0037, "step": 44223 }, { "epoch": 2.3577331129711574, "grad_norm": 0.9023627638816833, "learning_rate": 1.1554000000000002e-06, "loss": 0.0045, "step": 44224 }, { "epoch": 2.3577864264008106, "grad_norm": 1.8552933931350708, "learning_rate": 1.1552e-06, "loss": 0.018, "step": 44225 }, { "epoch": 2.3578397398304634, "grad_norm": 3.0950260162353516, "learning_rate": 1.1550000000000002e-06, "loss": 0.0143, "step": 44226 }, { "epoch": 2.357893053260116, "grad_norm": 0.7506491541862488, "learning_rate": 1.1548000000000001e-06, "loss": 0.005, "step": 44227 }, { "epoch": 2.3579463666897693, "grad_norm": 0.92960125207901, "learning_rate": 1.1546e-06, "loss": 0.0065, "step": 44228 }, { "epoch": 2.357999680119422, "grad_norm": 2.5401830673217773, "learning_rate": 1.1544000000000001e-06, "loss": 0.0113, "step": 44229 }, { "epoch": 2.358052993549075, "grad_norm": 1.105617880821228, "learning_rate": 1.1542e-06, "loss": 0.0068, "step": 44230 }, { "epoch": 2.358106306978728, "grad_norm": 1.2349218130111694, "learning_rate": 1.154e-06, "loss": 0.0059, "step": 44231 }, { "epoch": 2.358159620408381, "grad_norm": 4.05049991607666, "learning_rate": 1.1538e-06, "loss": 0.0204, "step": 44232 }, { "epoch": 2.358212933838034, "grad_norm": 2.532545804977417, "learning_rate": 1.1536e-06, "loss": 0.0219, "step": 44233 }, { "epoch": 2.3582662472676867, "grad_norm": 2.5070090293884277, "learning_rate": 1.1534e-06, "loss": 0.0302, "step": 44234 }, { "epoch": 2.3583195606973395, "grad_norm": 1.9284029006958008, "learning_rate": 1.1532000000000002e-06, "loss": 0.0234, "step": 44235 }, { "epoch": 2.3583728741269927, "grad_norm": 1.6242094039916992, "learning_rate": 1.153e-06, "loss": 0.0117, "step": 44236 }, { "epoch": 2.3584261875566455, "grad_norm": 2.0193252563476562, "learning_rate": 1.1528e-06, "loss": 0.0495, "step": 44237 }, { "epoch": 2.3584795009862987, "grad_norm": 1.588234305381775, "learning_rate": 1.1526000000000002e-06, "loss": 0.0089, "step": 44238 }, { "epoch": 2.3585328144159514, "grad_norm": 2.970280408859253, "learning_rate": 1.1524e-06, "loss": 0.0187, "step": 44239 }, { "epoch": 2.358586127845604, "grad_norm": 2.1556334495544434, "learning_rate": 1.1522000000000002e-06, "loss": 0.0233, "step": 44240 }, { "epoch": 2.3586394412752574, "grad_norm": 2.9142134189605713, "learning_rate": 1.1520000000000002e-06, "loss": 0.0132, "step": 44241 }, { "epoch": 2.35869275470491, "grad_norm": 1.840510368347168, "learning_rate": 1.1518000000000001e-06, "loss": 0.007, "step": 44242 }, { "epoch": 2.358746068134563, "grad_norm": 0.37000134587287903, "learning_rate": 1.1516000000000001e-06, "loss": 0.0028, "step": 44243 }, { "epoch": 2.358799381564216, "grad_norm": 0.5176233053207397, "learning_rate": 1.1514000000000001e-06, "loss": 0.003, "step": 44244 }, { "epoch": 2.358852694993869, "grad_norm": 3.801314353942871, "learning_rate": 1.1512e-06, "loss": 0.0383, "step": 44245 }, { "epoch": 2.358906008423522, "grad_norm": 3.11545729637146, "learning_rate": 1.151e-06, "loss": 0.032, "step": 44246 }, { "epoch": 2.358959321853175, "grad_norm": 3.261651039123535, "learning_rate": 1.1508e-06, "loss": 0.0224, "step": 44247 }, { "epoch": 2.3590126352828276, "grad_norm": 2.05460524559021, "learning_rate": 1.1506e-06, "loss": 0.016, "step": 44248 }, { "epoch": 2.3590659487124808, "grad_norm": 1.1040444374084473, "learning_rate": 1.1504e-06, "loss": 0.009, "step": 44249 }, { "epoch": 2.3591192621421335, "grad_norm": 1.6234208345413208, "learning_rate": 1.1502e-06, "loss": 0.0092, "step": 44250 }, { "epoch": 2.3591725755717867, "grad_norm": 2.442542314529419, "learning_rate": 1.1500000000000002e-06, "loss": 0.007, "step": 44251 }, { "epoch": 2.3592258890014395, "grad_norm": 3.6606907844543457, "learning_rate": 1.1498e-06, "loss": 0.0154, "step": 44252 }, { "epoch": 2.3592792024310922, "grad_norm": 2.1046571731567383, "learning_rate": 1.1496000000000002e-06, "loss": 0.0145, "step": 44253 }, { "epoch": 2.3593325158607454, "grad_norm": 1.5485986471176147, "learning_rate": 1.1494000000000002e-06, "loss": 0.0177, "step": 44254 }, { "epoch": 2.359385829290398, "grad_norm": 3.5026731491088867, "learning_rate": 1.1492e-06, "loss": 0.0283, "step": 44255 }, { "epoch": 2.359439142720051, "grad_norm": 3.914763927459717, "learning_rate": 1.1490000000000001e-06, "loss": 0.0085, "step": 44256 }, { "epoch": 2.359492456149704, "grad_norm": 1.557383418083191, "learning_rate": 1.1488000000000001e-06, "loss": 0.0142, "step": 44257 }, { "epoch": 2.359545769579357, "grad_norm": 2.7261147499084473, "learning_rate": 1.1486000000000001e-06, "loss": 0.0106, "step": 44258 }, { "epoch": 2.35959908300901, "grad_norm": 0.900176465511322, "learning_rate": 1.1484e-06, "loss": 0.0043, "step": 44259 }, { "epoch": 2.359652396438663, "grad_norm": 1.584146499633789, "learning_rate": 1.1482e-06, "loss": 0.0079, "step": 44260 }, { "epoch": 2.359705709868316, "grad_norm": 1.2779611349105835, "learning_rate": 1.148e-06, "loss": 0.0053, "step": 44261 }, { "epoch": 2.359759023297969, "grad_norm": 0.5904104709625244, "learning_rate": 1.1478e-06, "loss": 0.0102, "step": 44262 }, { "epoch": 2.3598123367276216, "grad_norm": 1.5536384582519531, "learning_rate": 1.1476e-06, "loss": 0.0159, "step": 44263 }, { "epoch": 2.359865650157275, "grad_norm": 3.249962329864502, "learning_rate": 1.1474e-06, "loss": 0.069, "step": 44264 }, { "epoch": 2.3599189635869275, "grad_norm": 2.0688846111297607, "learning_rate": 1.1472000000000002e-06, "loss": 0.0258, "step": 44265 }, { "epoch": 2.3599722770165803, "grad_norm": 2.924933433532715, "learning_rate": 1.147e-06, "loss": 0.0343, "step": 44266 }, { "epoch": 2.3600255904462335, "grad_norm": 1.5654746294021606, "learning_rate": 1.1468000000000002e-06, "loss": 0.0083, "step": 44267 }, { "epoch": 2.3600789038758863, "grad_norm": 0.9805282950401306, "learning_rate": 1.1466000000000002e-06, "loss": 0.005, "step": 44268 }, { "epoch": 2.3601322173055395, "grad_norm": 0.6454764604568481, "learning_rate": 1.1464000000000002e-06, "loss": 0.0029, "step": 44269 }, { "epoch": 2.360185530735192, "grad_norm": 4.802034378051758, "learning_rate": 1.1462000000000001e-06, "loss": 0.0297, "step": 44270 }, { "epoch": 2.360238844164845, "grad_norm": 1.427256464958191, "learning_rate": 1.1460000000000001e-06, "loss": 0.0205, "step": 44271 }, { "epoch": 2.360292157594498, "grad_norm": 1.004605770111084, "learning_rate": 1.1458000000000001e-06, "loss": 0.0238, "step": 44272 }, { "epoch": 2.360345471024151, "grad_norm": 7.093649387359619, "learning_rate": 1.1456e-06, "loss": 0.0509, "step": 44273 }, { "epoch": 2.360398784453804, "grad_norm": 1.2681853771209717, "learning_rate": 1.1454e-06, "loss": 0.0071, "step": 44274 }, { "epoch": 2.360452097883457, "grad_norm": 2.8820972442626953, "learning_rate": 1.1452e-06, "loss": 0.0134, "step": 44275 }, { "epoch": 2.3605054113131096, "grad_norm": 4.797036647796631, "learning_rate": 1.145e-06, "loss": 0.0335, "step": 44276 }, { "epoch": 2.360558724742763, "grad_norm": 1.0870211124420166, "learning_rate": 1.1448e-06, "loss": 0.0034, "step": 44277 }, { "epoch": 2.3606120381724156, "grad_norm": 1.9820406436920166, "learning_rate": 1.1446000000000002e-06, "loss": 0.0135, "step": 44278 }, { "epoch": 2.3606653516020684, "grad_norm": 0.9151648879051208, "learning_rate": 1.1444e-06, "loss": 0.0053, "step": 44279 }, { "epoch": 2.3607186650317216, "grad_norm": 1.3360615968704224, "learning_rate": 1.1442e-06, "loss": 0.0252, "step": 44280 }, { "epoch": 2.3607719784613743, "grad_norm": 2.774799346923828, "learning_rate": 1.1440000000000002e-06, "loss": 0.0113, "step": 44281 }, { "epoch": 2.3608252918910275, "grad_norm": 2.0452706813812256, "learning_rate": 1.1438e-06, "loss": 0.0242, "step": 44282 }, { "epoch": 2.3608786053206803, "grad_norm": 0.3771800100803375, "learning_rate": 1.1436000000000002e-06, "loss": 0.0026, "step": 44283 }, { "epoch": 2.360931918750333, "grad_norm": 3.1187641620635986, "learning_rate": 1.1434000000000001e-06, "loss": 0.0279, "step": 44284 }, { "epoch": 2.3609852321799862, "grad_norm": 3.186309814453125, "learning_rate": 1.1432000000000001e-06, "loss": 0.0527, "step": 44285 }, { "epoch": 2.361038545609639, "grad_norm": 2.517596483230591, "learning_rate": 1.1430000000000001e-06, "loss": 0.0061, "step": 44286 }, { "epoch": 2.361091859039292, "grad_norm": 1.3115259408950806, "learning_rate": 1.1428e-06, "loss": 0.0166, "step": 44287 }, { "epoch": 2.361145172468945, "grad_norm": 2.48077130317688, "learning_rate": 1.1426e-06, "loss": 0.0131, "step": 44288 }, { "epoch": 2.3611984858985977, "grad_norm": 1.5315217971801758, "learning_rate": 1.1424e-06, "loss": 0.0169, "step": 44289 }, { "epoch": 2.361251799328251, "grad_norm": 1.0686489343643188, "learning_rate": 1.1422e-06, "loss": 0.0106, "step": 44290 }, { "epoch": 2.3613051127579037, "grad_norm": 0.38808587193489075, "learning_rate": 1.142e-06, "loss": 0.0033, "step": 44291 }, { "epoch": 2.3613584261875564, "grad_norm": 0.6344529986381531, "learning_rate": 1.1418000000000002e-06, "loss": 0.0053, "step": 44292 }, { "epoch": 2.3614117396172096, "grad_norm": 0.6086764335632324, "learning_rate": 1.1416e-06, "loss": 0.0027, "step": 44293 }, { "epoch": 2.3614650530468624, "grad_norm": 4.301617622375488, "learning_rate": 1.1414000000000002e-06, "loss": 0.0291, "step": 44294 }, { "epoch": 2.3615183664765156, "grad_norm": 0.9093194007873535, "learning_rate": 1.1412000000000002e-06, "loss": 0.0165, "step": 44295 }, { "epoch": 2.3615716799061683, "grad_norm": 1.8685559034347534, "learning_rate": 1.141e-06, "loss": 0.0067, "step": 44296 }, { "epoch": 2.3616249933358215, "grad_norm": 0.28943556547164917, "learning_rate": 1.1408000000000002e-06, "loss": 0.003, "step": 44297 }, { "epoch": 2.3616783067654743, "grad_norm": 1.7259578704833984, "learning_rate": 1.1406000000000001e-06, "loss": 0.0178, "step": 44298 }, { "epoch": 2.361731620195127, "grad_norm": 0.8042137622833252, "learning_rate": 1.1404000000000001e-06, "loss": 0.0152, "step": 44299 }, { "epoch": 2.3617849336247803, "grad_norm": 0.4571409523487091, "learning_rate": 1.1402000000000001e-06, "loss": 0.0026, "step": 44300 }, { "epoch": 2.361838247054433, "grad_norm": 2.9316771030426025, "learning_rate": 1.14e-06, "loss": 0.0221, "step": 44301 }, { "epoch": 2.3618915604840858, "grad_norm": 1.5038403272628784, "learning_rate": 1.1398e-06, "loss": 0.0159, "step": 44302 }, { "epoch": 2.361944873913739, "grad_norm": 2.771942138671875, "learning_rate": 1.1396e-06, "loss": 0.048, "step": 44303 }, { "epoch": 2.3619981873433917, "grad_norm": 1.5992894172668457, "learning_rate": 1.1394e-06, "loss": 0.029, "step": 44304 }, { "epoch": 2.3620515007730445, "grad_norm": 2.44146728515625, "learning_rate": 1.1392e-06, "loss": 0.0253, "step": 44305 }, { "epoch": 2.3621048142026977, "grad_norm": 2.818148374557495, "learning_rate": 1.139e-06, "loss": 0.0402, "step": 44306 }, { "epoch": 2.3621581276323504, "grad_norm": 0.5420661568641663, "learning_rate": 1.1388e-06, "loss": 0.003, "step": 44307 }, { "epoch": 2.3622114410620036, "grad_norm": 0.7341570258140564, "learning_rate": 1.1386000000000002e-06, "loss": 0.0052, "step": 44308 }, { "epoch": 2.3622647544916564, "grad_norm": 1.4182237386703491, "learning_rate": 1.1384e-06, "loss": 0.0153, "step": 44309 }, { "epoch": 2.3623180679213096, "grad_norm": 2.253636360168457, "learning_rate": 1.1382000000000002e-06, "loss": 0.0094, "step": 44310 }, { "epoch": 2.3623713813509624, "grad_norm": 1.490199089050293, "learning_rate": 1.1380000000000002e-06, "loss": 0.0125, "step": 44311 }, { "epoch": 2.362424694780615, "grad_norm": 0.6142252087593079, "learning_rate": 1.1378000000000001e-06, "loss": 0.0054, "step": 44312 }, { "epoch": 2.3624780082102683, "grad_norm": 0.7673885226249695, "learning_rate": 1.1376000000000001e-06, "loss": 0.0043, "step": 44313 }, { "epoch": 2.362531321639921, "grad_norm": 2.1065316200256348, "learning_rate": 1.1374000000000001e-06, "loss": 0.0108, "step": 44314 }, { "epoch": 2.362584635069574, "grad_norm": 1.7237738370895386, "learning_rate": 1.1372e-06, "loss": 0.0071, "step": 44315 }, { "epoch": 2.362637948499227, "grad_norm": 1.7695108652114868, "learning_rate": 1.137e-06, "loss": 0.0147, "step": 44316 }, { "epoch": 2.36269126192888, "grad_norm": 1.0815216302871704, "learning_rate": 1.1368e-06, "loss": 0.0233, "step": 44317 }, { "epoch": 2.362744575358533, "grad_norm": 1.585371732711792, "learning_rate": 1.1366e-06, "loss": 0.0069, "step": 44318 }, { "epoch": 2.3627978887881858, "grad_norm": 2.402132272720337, "learning_rate": 1.1364e-06, "loss": 0.024, "step": 44319 }, { "epoch": 2.3628512022178385, "grad_norm": 2.317556381225586, "learning_rate": 1.1362e-06, "loss": 0.0308, "step": 44320 }, { "epoch": 2.3629045156474917, "grad_norm": 1.538506269454956, "learning_rate": 1.1360000000000002e-06, "loss": 0.0232, "step": 44321 }, { "epoch": 2.3629578290771445, "grad_norm": 1.4690868854522705, "learning_rate": 1.1358e-06, "loss": 0.0182, "step": 44322 }, { "epoch": 2.3630111425067977, "grad_norm": 1.9070440530776978, "learning_rate": 1.1356e-06, "loss": 0.0157, "step": 44323 }, { "epoch": 2.3630644559364504, "grad_norm": 1.1506761312484741, "learning_rate": 1.1354000000000002e-06, "loss": 0.0053, "step": 44324 }, { "epoch": 2.363117769366103, "grad_norm": 1.5904618501663208, "learning_rate": 1.1352e-06, "loss": 0.0087, "step": 44325 }, { "epoch": 2.3631710827957564, "grad_norm": 0.5697722434997559, "learning_rate": 1.1350000000000001e-06, "loss": 0.0033, "step": 44326 }, { "epoch": 2.363224396225409, "grad_norm": 1.0874905586242676, "learning_rate": 1.1348000000000001e-06, "loss": 0.0237, "step": 44327 }, { "epoch": 2.363277709655062, "grad_norm": 0.7455185055732727, "learning_rate": 1.1346000000000001e-06, "loss": 0.0206, "step": 44328 }, { "epoch": 2.363331023084715, "grad_norm": 2.7867140769958496, "learning_rate": 1.1344e-06, "loss": 0.0238, "step": 44329 }, { "epoch": 2.363384336514368, "grad_norm": 1.1039531230926514, "learning_rate": 1.1342e-06, "loss": 0.008, "step": 44330 }, { "epoch": 2.363437649944021, "grad_norm": 1.1681429147720337, "learning_rate": 1.134e-06, "loss": 0.0166, "step": 44331 }, { "epoch": 2.363490963373674, "grad_norm": 1.3435136079788208, "learning_rate": 1.1338e-06, "loss": 0.0116, "step": 44332 }, { "epoch": 2.3635442768033266, "grad_norm": 2.066213846206665, "learning_rate": 1.1336e-06, "loss": 0.0075, "step": 44333 }, { "epoch": 2.3635975902329798, "grad_norm": 2.654165029525757, "learning_rate": 1.1334e-06, "loss": 0.0311, "step": 44334 }, { "epoch": 2.3636509036626325, "grad_norm": 1.5702229738235474, "learning_rate": 1.1332000000000002e-06, "loss": 0.0158, "step": 44335 }, { "epoch": 2.3637042170922857, "grad_norm": 0.6296096444129944, "learning_rate": 1.133e-06, "loss": 0.0042, "step": 44336 }, { "epoch": 2.3637575305219385, "grad_norm": 0.7953614592552185, "learning_rate": 1.1328000000000002e-06, "loss": 0.007, "step": 44337 }, { "epoch": 2.3638108439515912, "grad_norm": 1.9428454637527466, "learning_rate": 1.1326000000000002e-06, "loss": 0.0084, "step": 44338 }, { "epoch": 2.3638641573812444, "grad_norm": 0.5823218822479248, "learning_rate": 1.1324e-06, "loss": 0.004, "step": 44339 }, { "epoch": 2.363917470810897, "grad_norm": 1.0151000022888184, "learning_rate": 1.1322000000000001e-06, "loss": 0.0178, "step": 44340 }, { "epoch": 2.36397078424055, "grad_norm": 1.4338818788528442, "learning_rate": 1.1320000000000001e-06, "loss": 0.0079, "step": 44341 }, { "epoch": 2.364024097670203, "grad_norm": 2.153296709060669, "learning_rate": 1.1318000000000001e-06, "loss": 0.0225, "step": 44342 }, { "epoch": 2.364077411099856, "grad_norm": 0.4590167999267578, "learning_rate": 1.1316e-06, "loss": 0.0025, "step": 44343 }, { "epoch": 2.364130724529509, "grad_norm": 1.2455319166183472, "learning_rate": 1.1314e-06, "loss": 0.014, "step": 44344 }, { "epoch": 2.364184037959162, "grad_norm": 1.1478626728057861, "learning_rate": 1.1312e-06, "loss": 0.0055, "step": 44345 }, { "epoch": 2.364237351388815, "grad_norm": 1.9925507307052612, "learning_rate": 1.131e-06, "loss": 0.0088, "step": 44346 }, { "epoch": 2.364290664818468, "grad_norm": 0.8292906284332275, "learning_rate": 1.1308e-06, "loss": 0.0042, "step": 44347 }, { "epoch": 2.3643439782481206, "grad_norm": 2.018754243850708, "learning_rate": 1.1306e-06, "loss": 0.0188, "step": 44348 }, { "epoch": 2.364397291677774, "grad_norm": 3.2897558212280273, "learning_rate": 1.1304e-06, "loss": 0.0398, "step": 44349 }, { "epoch": 2.3644506051074266, "grad_norm": 1.2566184997558594, "learning_rate": 1.1302e-06, "loss": 0.0059, "step": 44350 }, { "epoch": 2.3645039185370793, "grad_norm": 2.190415859222412, "learning_rate": 1.1300000000000002e-06, "loss": 0.0221, "step": 44351 }, { "epoch": 2.3645572319667325, "grad_norm": 1.1092015504837036, "learning_rate": 1.1298e-06, "loss": 0.0067, "step": 44352 }, { "epoch": 2.3646105453963853, "grad_norm": 1.3101122379302979, "learning_rate": 1.1296000000000002e-06, "loss": 0.0057, "step": 44353 }, { "epoch": 2.3646638588260385, "grad_norm": 2.159005641937256, "learning_rate": 1.1294000000000001e-06, "loss": 0.0294, "step": 44354 }, { "epoch": 2.3647171722556912, "grad_norm": 0.8205670714378357, "learning_rate": 1.1292000000000001e-06, "loss": 0.0039, "step": 44355 }, { "epoch": 2.364770485685344, "grad_norm": 0.8290070295333862, "learning_rate": 1.1290000000000001e-06, "loss": 0.0056, "step": 44356 }, { "epoch": 2.364823799114997, "grad_norm": 1.0073431730270386, "learning_rate": 1.1288e-06, "loss": 0.0056, "step": 44357 }, { "epoch": 2.36487711254465, "grad_norm": 1.9431936740875244, "learning_rate": 1.1286e-06, "loss": 0.0081, "step": 44358 }, { "epoch": 2.364930425974303, "grad_norm": 1.8420312404632568, "learning_rate": 1.1284e-06, "loss": 0.0061, "step": 44359 }, { "epoch": 2.364983739403956, "grad_norm": 1.5351417064666748, "learning_rate": 1.1282e-06, "loss": 0.0157, "step": 44360 }, { "epoch": 2.3650370528336087, "grad_norm": 1.8405545949935913, "learning_rate": 1.128e-06, "loss": 0.018, "step": 44361 }, { "epoch": 2.365090366263262, "grad_norm": 1.856433629989624, "learning_rate": 1.1278000000000002e-06, "loss": 0.0205, "step": 44362 }, { "epoch": 2.3651436796929146, "grad_norm": 1.6240605115890503, "learning_rate": 1.1276e-06, "loss": 0.0188, "step": 44363 }, { "epoch": 2.3651969931225674, "grad_norm": 0.7125946879386902, "learning_rate": 1.1274000000000002e-06, "loss": 0.0032, "step": 44364 }, { "epoch": 2.3652503065522206, "grad_norm": 0.8656815886497498, "learning_rate": 1.1272000000000002e-06, "loss": 0.0164, "step": 44365 }, { "epoch": 2.3653036199818733, "grad_norm": 2.808861017227173, "learning_rate": 1.127e-06, "loss": 0.0407, "step": 44366 }, { "epoch": 2.3653569334115265, "grad_norm": 0.5624793171882629, "learning_rate": 1.1268000000000002e-06, "loss": 0.0036, "step": 44367 }, { "epoch": 2.3654102468411793, "grad_norm": 1.5645997524261475, "learning_rate": 1.1266000000000001e-06, "loss": 0.0105, "step": 44368 }, { "epoch": 2.365463560270832, "grad_norm": 2.638704776763916, "learning_rate": 1.1264000000000001e-06, "loss": 0.0286, "step": 44369 }, { "epoch": 2.3655168737004852, "grad_norm": 1.5115638971328735, "learning_rate": 1.1262000000000001e-06, "loss": 0.0055, "step": 44370 }, { "epoch": 2.365570187130138, "grad_norm": 0.25939512252807617, "learning_rate": 1.126e-06, "loss": 0.0021, "step": 44371 }, { "epoch": 2.365623500559791, "grad_norm": 0.5867149233818054, "learning_rate": 1.1258e-06, "loss": 0.0035, "step": 44372 }, { "epoch": 2.365676813989444, "grad_norm": 1.4820951223373413, "learning_rate": 1.1256e-06, "loss": 0.0139, "step": 44373 }, { "epoch": 2.3657301274190967, "grad_norm": 1.1667969226837158, "learning_rate": 1.1254e-06, "loss": 0.0048, "step": 44374 }, { "epoch": 2.36578344084875, "grad_norm": 1.1019814014434814, "learning_rate": 1.1252e-06, "loss": 0.0128, "step": 44375 }, { "epoch": 2.3658367542784027, "grad_norm": 2.326852798461914, "learning_rate": 1.125e-06, "loss": 0.0097, "step": 44376 }, { "epoch": 2.3658900677080554, "grad_norm": 1.416431188583374, "learning_rate": 1.1248e-06, "loss": 0.0277, "step": 44377 }, { "epoch": 2.3659433811377086, "grad_norm": 1.2809635400772095, "learning_rate": 1.1246000000000002e-06, "loss": 0.0066, "step": 44378 }, { "epoch": 2.3659966945673614, "grad_norm": 2.265781879425049, "learning_rate": 1.1244e-06, "loss": 0.0068, "step": 44379 }, { "epoch": 2.3660500079970146, "grad_norm": 1.6492737531661987, "learning_rate": 1.1242000000000002e-06, "loss": 0.0295, "step": 44380 }, { "epoch": 2.3661033214266673, "grad_norm": 2.0564124584198, "learning_rate": 1.1240000000000002e-06, "loss": 0.039, "step": 44381 }, { "epoch": 2.3661566348563206, "grad_norm": 0.3585917055606842, "learning_rate": 1.1238e-06, "loss": 0.0021, "step": 44382 }, { "epoch": 2.3662099482859733, "grad_norm": 0.9316765666007996, "learning_rate": 1.1236000000000001e-06, "loss": 0.0148, "step": 44383 }, { "epoch": 2.366263261715626, "grad_norm": 2.2631261348724365, "learning_rate": 1.1234000000000001e-06, "loss": 0.0198, "step": 44384 }, { "epoch": 2.3663165751452793, "grad_norm": 2.579958200454712, "learning_rate": 1.1232e-06, "loss": 0.0118, "step": 44385 }, { "epoch": 2.366369888574932, "grad_norm": 1.1738966703414917, "learning_rate": 1.123e-06, "loss": 0.0177, "step": 44386 }, { "epoch": 2.366423202004585, "grad_norm": 3.1986701488494873, "learning_rate": 1.1228e-06, "loss": 0.02, "step": 44387 }, { "epoch": 2.366476515434238, "grad_norm": 1.3853590488433838, "learning_rate": 1.1226e-06, "loss": 0.0231, "step": 44388 }, { "epoch": 2.3665298288638907, "grad_norm": 0.32331162691116333, "learning_rate": 1.1224000000000003e-06, "loss": 0.0019, "step": 44389 }, { "epoch": 2.3665831422935435, "grad_norm": 1.4861725568771362, "learning_rate": 1.1222e-06, "loss": 0.0187, "step": 44390 }, { "epoch": 2.3666364557231967, "grad_norm": 0.4748087525367737, "learning_rate": 1.122e-06, "loss": 0.0022, "step": 44391 }, { "epoch": 2.3666897691528495, "grad_norm": 1.7020843029022217, "learning_rate": 1.1218000000000002e-06, "loss": 0.0147, "step": 44392 }, { "epoch": 2.3667430825825027, "grad_norm": 1.9113558530807495, "learning_rate": 1.1216e-06, "loss": 0.0159, "step": 44393 }, { "epoch": 2.3667963960121554, "grad_norm": 0.9939989447593689, "learning_rate": 1.1214000000000002e-06, "loss": 0.007, "step": 44394 }, { "epoch": 2.3668497094418086, "grad_norm": 2.213792085647583, "learning_rate": 1.1212000000000002e-06, "loss": 0.0101, "step": 44395 }, { "epoch": 2.3669030228714614, "grad_norm": 2.6857454776763916, "learning_rate": 1.1210000000000002e-06, "loss": 0.0246, "step": 44396 }, { "epoch": 2.366956336301114, "grad_norm": 2.0689496994018555, "learning_rate": 1.1208000000000001e-06, "loss": 0.0082, "step": 44397 }, { "epoch": 2.3670096497307673, "grad_norm": 0.22262760996818542, "learning_rate": 1.1206000000000001e-06, "loss": 0.0019, "step": 44398 }, { "epoch": 2.36706296316042, "grad_norm": 2.118152618408203, "learning_rate": 1.1204000000000001e-06, "loss": 0.0223, "step": 44399 }, { "epoch": 2.367116276590073, "grad_norm": 3.7560598850250244, "learning_rate": 1.1202e-06, "loss": 0.0357, "step": 44400 }, { "epoch": 2.367169590019726, "grad_norm": 1.5962293148040771, "learning_rate": 1.12e-06, "loss": 0.0073, "step": 44401 }, { "epoch": 2.367222903449379, "grad_norm": 1.3931537866592407, "learning_rate": 1.1198e-06, "loss": 0.0045, "step": 44402 }, { "epoch": 2.367276216879032, "grad_norm": 1.7327558994293213, "learning_rate": 1.1196e-06, "loss": 0.023, "step": 44403 }, { "epoch": 2.3673295303086848, "grad_norm": 1.1155614852905273, "learning_rate": 1.1194e-06, "loss": 0.004, "step": 44404 }, { "epoch": 2.3673828437383375, "grad_norm": 5.4776105880737305, "learning_rate": 1.1192000000000002e-06, "loss": 0.0508, "step": 44405 }, { "epoch": 2.3674361571679907, "grad_norm": 1.8223457336425781, "learning_rate": 1.119e-06, "loss": 0.0063, "step": 44406 }, { "epoch": 2.3674894705976435, "grad_norm": 0.9832481741905212, "learning_rate": 1.1188e-06, "loss": 0.0047, "step": 44407 }, { "epoch": 2.3675427840272967, "grad_norm": 1.172210454940796, "learning_rate": 1.1186000000000002e-06, "loss": 0.0124, "step": 44408 }, { "epoch": 2.3675960974569494, "grad_norm": 1.9489514827728271, "learning_rate": 1.1184e-06, "loss": 0.0424, "step": 44409 }, { "epoch": 2.367649410886602, "grad_norm": 1.7499865293502808, "learning_rate": 1.1182000000000002e-06, "loss": 0.0085, "step": 44410 }, { "epoch": 2.3677027243162554, "grad_norm": 3.5712532997131348, "learning_rate": 1.1180000000000001e-06, "loss": 0.0286, "step": 44411 }, { "epoch": 2.367756037745908, "grad_norm": 1.3295986652374268, "learning_rate": 1.1178000000000001e-06, "loss": 0.0082, "step": 44412 }, { "epoch": 2.367809351175561, "grad_norm": 1.4668511152267456, "learning_rate": 1.1176000000000001e-06, "loss": 0.0058, "step": 44413 }, { "epoch": 2.367862664605214, "grad_norm": 1.325163722038269, "learning_rate": 1.1174e-06, "loss": 0.02, "step": 44414 }, { "epoch": 2.367915978034867, "grad_norm": 0.6821867227554321, "learning_rate": 1.1172e-06, "loss": 0.0038, "step": 44415 }, { "epoch": 2.36796929146452, "grad_norm": 1.7140215635299683, "learning_rate": 1.117e-06, "loss": 0.0139, "step": 44416 }, { "epoch": 2.368022604894173, "grad_norm": 2.9278650283813477, "learning_rate": 1.1168e-06, "loss": 0.0158, "step": 44417 }, { "epoch": 2.3680759183238256, "grad_norm": 3.8175058364868164, "learning_rate": 1.1166e-06, "loss": 0.0291, "step": 44418 }, { "epoch": 2.368129231753479, "grad_norm": 0.3111705183982849, "learning_rate": 1.1164e-06, "loss": 0.0021, "step": 44419 }, { "epoch": 2.3681825451831315, "grad_norm": 2.7694356441497803, "learning_rate": 1.1162e-06, "loss": 0.0389, "step": 44420 }, { "epoch": 2.3682358586127847, "grad_norm": 0.9201681613922119, "learning_rate": 1.1160000000000002e-06, "loss": 0.0164, "step": 44421 }, { "epoch": 2.3682891720424375, "grad_norm": 3.1521925926208496, "learning_rate": 1.1158e-06, "loss": 0.0271, "step": 44422 }, { "epoch": 2.3683424854720903, "grad_norm": 2.5521345138549805, "learning_rate": 1.1156000000000002e-06, "loss": 0.0286, "step": 44423 }, { "epoch": 2.3683957989017435, "grad_norm": 3.4585297107696533, "learning_rate": 1.1154000000000002e-06, "loss": 0.017, "step": 44424 }, { "epoch": 2.368449112331396, "grad_norm": 1.9501100778579712, "learning_rate": 1.1152e-06, "loss": 0.0395, "step": 44425 }, { "epoch": 2.368502425761049, "grad_norm": 4.163830280303955, "learning_rate": 1.1150000000000001e-06, "loss": 0.0361, "step": 44426 }, { "epoch": 2.368555739190702, "grad_norm": 5.9610209465026855, "learning_rate": 1.1148000000000001e-06, "loss": 0.0414, "step": 44427 }, { "epoch": 2.368609052620355, "grad_norm": 1.4031810760498047, "learning_rate": 1.1146e-06, "loss": 0.0053, "step": 44428 }, { "epoch": 2.368662366050008, "grad_norm": 0.9604203104972839, "learning_rate": 1.1144e-06, "loss": 0.0185, "step": 44429 }, { "epoch": 2.368715679479661, "grad_norm": 1.1636408567428589, "learning_rate": 1.1142e-06, "loss": 0.0046, "step": 44430 }, { "epoch": 2.368768992909314, "grad_norm": 1.0396440029144287, "learning_rate": 1.114e-06, "loss": 0.0054, "step": 44431 }, { "epoch": 2.368822306338967, "grad_norm": 1.9525551795959473, "learning_rate": 1.1138000000000002e-06, "loss": 0.0186, "step": 44432 }, { "epoch": 2.3688756197686196, "grad_norm": 1.7511005401611328, "learning_rate": 1.1136e-06, "loss": 0.0112, "step": 44433 }, { "epoch": 2.368928933198273, "grad_norm": 0.7045235633850098, "learning_rate": 1.1134e-06, "loss": 0.005, "step": 44434 }, { "epoch": 2.3689822466279256, "grad_norm": 0.8599106073379517, "learning_rate": 1.1132000000000002e-06, "loss": 0.0148, "step": 44435 }, { "epoch": 2.3690355600575783, "grad_norm": 2.3189542293548584, "learning_rate": 1.113e-06, "loss": 0.0097, "step": 44436 }, { "epoch": 2.3690888734872315, "grad_norm": 2.0717718601226807, "learning_rate": 1.1128000000000002e-06, "loss": 0.016, "step": 44437 }, { "epoch": 2.3691421869168843, "grad_norm": 0.956932544708252, "learning_rate": 1.1126000000000002e-06, "loss": 0.0224, "step": 44438 }, { "epoch": 2.3691955003465375, "grad_norm": 2.7260167598724365, "learning_rate": 1.1124000000000001e-06, "loss": 0.0122, "step": 44439 }, { "epoch": 2.3692488137761902, "grad_norm": 0.8114615082740784, "learning_rate": 1.1122000000000001e-06, "loss": 0.004, "step": 44440 }, { "epoch": 2.369302127205843, "grad_norm": 4.153814315795898, "learning_rate": 1.1120000000000001e-06, "loss": 0.0123, "step": 44441 }, { "epoch": 2.369355440635496, "grad_norm": 1.4605941772460938, "learning_rate": 1.1118e-06, "loss": 0.0114, "step": 44442 }, { "epoch": 2.369408754065149, "grad_norm": 0.6083248853683472, "learning_rate": 1.1116e-06, "loss": 0.0029, "step": 44443 }, { "epoch": 2.369462067494802, "grad_norm": 4.449282646179199, "learning_rate": 1.1114e-06, "loss": 0.0191, "step": 44444 }, { "epoch": 2.369515380924455, "grad_norm": 2.8030595779418945, "learning_rate": 1.1112e-06, "loss": 0.0276, "step": 44445 }, { "epoch": 2.3695686943541077, "grad_norm": 0.9708607792854309, "learning_rate": 1.111e-06, "loss": 0.0052, "step": 44446 }, { "epoch": 2.369622007783761, "grad_norm": 1.3729206323623657, "learning_rate": 1.1108e-06, "loss": 0.0216, "step": 44447 }, { "epoch": 2.3696753212134136, "grad_norm": 1.3256535530090332, "learning_rate": 1.1106000000000002e-06, "loss": 0.0136, "step": 44448 }, { "epoch": 2.3697286346430664, "grad_norm": 1.722109317779541, "learning_rate": 1.1104e-06, "loss": 0.0254, "step": 44449 }, { "epoch": 2.3697819480727196, "grad_norm": 1.6380943059921265, "learning_rate": 1.1102e-06, "loss": 0.025, "step": 44450 }, { "epoch": 2.3698352615023723, "grad_norm": 3.609773635864258, "learning_rate": 1.1100000000000002e-06, "loss": 0.022, "step": 44451 }, { "epoch": 2.3698885749320255, "grad_norm": 2.3699235916137695, "learning_rate": 1.1098e-06, "loss": 0.0063, "step": 44452 }, { "epoch": 2.3699418883616783, "grad_norm": 0.949959933757782, "learning_rate": 1.1096000000000001e-06, "loss": 0.004, "step": 44453 }, { "epoch": 2.369995201791331, "grad_norm": 3.3908655643463135, "learning_rate": 1.1094000000000001e-06, "loss": 0.032, "step": 44454 }, { "epoch": 2.3700485152209843, "grad_norm": 1.4029674530029297, "learning_rate": 1.1092000000000001e-06, "loss": 0.0065, "step": 44455 }, { "epoch": 2.370101828650637, "grad_norm": 1.080833077430725, "learning_rate": 1.109e-06, "loss": 0.0101, "step": 44456 }, { "epoch": 2.37015514208029, "grad_norm": 1.736145257949829, "learning_rate": 1.1088e-06, "loss": 0.0182, "step": 44457 }, { "epoch": 2.370208455509943, "grad_norm": 0.7797763347625732, "learning_rate": 1.1086e-06, "loss": 0.0048, "step": 44458 }, { "epoch": 2.3702617689395957, "grad_norm": 2.5331664085388184, "learning_rate": 1.1084e-06, "loss": 0.0144, "step": 44459 }, { "epoch": 2.370315082369249, "grad_norm": 1.4363899230957031, "learning_rate": 1.1082e-06, "loss": 0.0126, "step": 44460 }, { "epoch": 2.3703683957989017, "grad_norm": 1.6058367490768433, "learning_rate": 1.108e-06, "loss": 0.0053, "step": 44461 }, { "epoch": 2.3704217092285544, "grad_norm": 2.083517551422119, "learning_rate": 1.1078000000000002e-06, "loss": 0.0129, "step": 44462 }, { "epoch": 2.3704750226582076, "grad_norm": 5.014946460723877, "learning_rate": 1.1076e-06, "loss": 0.0138, "step": 44463 }, { "epoch": 2.3705283360878604, "grad_norm": 0.2992210388183594, "learning_rate": 1.1074000000000002e-06, "loss": 0.0026, "step": 44464 }, { "epoch": 2.3705816495175136, "grad_norm": 1.8255319595336914, "learning_rate": 1.1072000000000002e-06, "loss": 0.0159, "step": 44465 }, { "epoch": 2.3706349629471664, "grad_norm": 1.2790237665176392, "learning_rate": 1.1070000000000002e-06, "loss": 0.0056, "step": 44466 }, { "epoch": 2.3706882763768196, "grad_norm": 0.35845404863357544, "learning_rate": 1.1068000000000001e-06, "loss": 0.0024, "step": 44467 }, { "epoch": 2.3707415898064723, "grad_norm": 1.2029916048049927, "learning_rate": 1.1066000000000001e-06, "loss": 0.0145, "step": 44468 }, { "epoch": 2.370794903236125, "grad_norm": 3.7862462997436523, "learning_rate": 1.1064000000000001e-06, "loss": 0.0166, "step": 44469 }, { "epoch": 2.3708482166657783, "grad_norm": 1.0521444082260132, "learning_rate": 1.1062e-06, "loss": 0.0032, "step": 44470 }, { "epoch": 2.370901530095431, "grad_norm": 0.8730363845825195, "learning_rate": 1.106e-06, "loss": 0.0204, "step": 44471 }, { "epoch": 2.370954843525084, "grad_norm": 1.2477107048034668, "learning_rate": 1.1058e-06, "loss": 0.0038, "step": 44472 }, { "epoch": 2.371008156954737, "grad_norm": 1.9921971559524536, "learning_rate": 1.1056e-06, "loss": 0.009, "step": 44473 }, { "epoch": 2.3710614703843897, "grad_norm": 1.3737694025039673, "learning_rate": 1.1054e-06, "loss": 0.0146, "step": 44474 }, { "epoch": 2.371114783814043, "grad_norm": 0.9978036880493164, "learning_rate": 1.1052e-06, "loss": 0.0266, "step": 44475 }, { "epoch": 2.3711680972436957, "grad_norm": 0.9437569379806519, "learning_rate": 1.105e-06, "loss": 0.0046, "step": 44476 }, { "epoch": 2.3712214106733485, "grad_norm": 2.711090326309204, "learning_rate": 1.1048e-06, "loss": 0.023, "step": 44477 }, { "epoch": 2.3712747241030017, "grad_norm": 1.1837100982666016, "learning_rate": 1.1046000000000002e-06, "loss": 0.0187, "step": 44478 }, { "epoch": 2.3713280375326544, "grad_norm": 2.379296064376831, "learning_rate": 1.1044e-06, "loss": 0.0278, "step": 44479 }, { "epoch": 2.3713813509623076, "grad_norm": 1.510156273841858, "learning_rate": 1.1042000000000002e-06, "loss": 0.0339, "step": 44480 }, { "epoch": 2.3714346643919604, "grad_norm": 0.5641108155250549, "learning_rate": 1.1040000000000001e-06, "loss": 0.0047, "step": 44481 }, { "epoch": 2.371487977821613, "grad_norm": 3.3519203662872314, "learning_rate": 1.1038000000000001e-06, "loss": 0.0083, "step": 44482 }, { "epoch": 2.3715412912512663, "grad_norm": 0.6263019442558289, "learning_rate": 1.1036000000000001e-06, "loss": 0.0034, "step": 44483 }, { "epoch": 2.371594604680919, "grad_norm": 1.7639803886413574, "learning_rate": 1.1034e-06, "loss": 0.0056, "step": 44484 }, { "epoch": 2.371647918110572, "grad_norm": 0.3881690204143524, "learning_rate": 1.1032e-06, "loss": 0.0025, "step": 44485 }, { "epoch": 2.371701231540225, "grad_norm": 0.43099212646484375, "learning_rate": 1.103e-06, "loss": 0.002, "step": 44486 }, { "epoch": 2.371754544969878, "grad_norm": 3.814636707305908, "learning_rate": 1.1028e-06, "loss": 0.0533, "step": 44487 }, { "epoch": 2.371807858399531, "grad_norm": 2.168898105621338, "learning_rate": 1.1026e-06, "loss": 0.0098, "step": 44488 }, { "epoch": 2.3718611718291838, "grad_norm": 4.081218242645264, "learning_rate": 1.1024000000000002e-06, "loss": 0.0229, "step": 44489 }, { "epoch": 2.3719144852588365, "grad_norm": 2.9411003589630127, "learning_rate": 1.1022e-06, "loss": 0.0563, "step": 44490 }, { "epoch": 2.3719677986884897, "grad_norm": 1.3698655366897583, "learning_rate": 1.1020000000000002e-06, "loss": 0.0372, "step": 44491 }, { "epoch": 2.3720211121181425, "grad_norm": 1.1444612741470337, "learning_rate": 1.1018000000000002e-06, "loss": 0.0038, "step": 44492 }, { "epoch": 2.3720744255477957, "grad_norm": 0.42734643816947937, "learning_rate": 1.1016e-06, "loss": 0.0029, "step": 44493 }, { "epoch": 2.3721277389774484, "grad_norm": 0.3474801480770111, "learning_rate": 1.1014000000000002e-06, "loss": 0.0018, "step": 44494 }, { "epoch": 2.372181052407101, "grad_norm": 3.8102641105651855, "learning_rate": 1.1012000000000001e-06, "loss": 0.0259, "step": 44495 }, { "epoch": 2.3722343658367544, "grad_norm": 0.2796826958656311, "learning_rate": 1.1010000000000001e-06, "loss": 0.0022, "step": 44496 }, { "epoch": 2.372287679266407, "grad_norm": 0.4979270100593567, "learning_rate": 1.1008000000000001e-06, "loss": 0.0032, "step": 44497 }, { "epoch": 2.37234099269606, "grad_norm": 2.3734896183013916, "learning_rate": 1.1006e-06, "loss": 0.024, "step": 44498 }, { "epoch": 2.372394306125713, "grad_norm": 2.30940318107605, "learning_rate": 1.1004e-06, "loss": 0.0193, "step": 44499 }, { "epoch": 2.372447619555366, "grad_norm": 0.20131556689739227, "learning_rate": 1.1002e-06, "loss": 0.0018, "step": 44500 }, { "epoch": 2.372500932985019, "grad_norm": 1.1711586713790894, "learning_rate": 1.1e-06, "loss": 0.0138, "step": 44501 }, { "epoch": 2.372554246414672, "grad_norm": 1.8551863431930542, "learning_rate": 1.0998e-06, "loss": 0.0307, "step": 44502 }, { "epoch": 2.372607559844325, "grad_norm": 1.8575738668441772, "learning_rate": 1.0996e-06, "loss": 0.0153, "step": 44503 }, { "epoch": 2.372660873273978, "grad_norm": 1.2117094993591309, "learning_rate": 1.0994e-06, "loss": 0.0043, "step": 44504 }, { "epoch": 2.3727141867036305, "grad_norm": 1.9873147010803223, "learning_rate": 1.0992000000000002e-06, "loss": 0.0075, "step": 44505 }, { "epoch": 2.3727675001332837, "grad_norm": 1.785513162612915, "learning_rate": 1.099e-06, "loss": 0.0333, "step": 44506 }, { "epoch": 2.3728208135629365, "grad_norm": 2.380584716796875, "learning_rate": 1.0988000000000002e-06, "loss": 0.0219, "step": 44507 }, { "epoch": 2.3728741269925893, "grad_norm": 1.9206600189208984, "learning_rate": 1.0986000000000002e-06, "loss": 0.0059, "step": 44508 }, { "epoch": 2.3729274404222425, "grad_norm": 2.018038272857666, "learning_rate": 1.0984e-06, "loss": 0.0241, "step": 44509 }, { "epoch": 2.372980753851895, "grad_norm": 2.0252554416656494, "learning_rate": 1.0982000000000001e-06, "loss": 0.0108, "step": 44510 }, { "epoch": 2.373034067281548, "grad_norm": 0.8064329028129578, "learning_rate": 1.0980000000000001e-06, "loss": 0.0113, "step": 44511 }, { "epoch": 2.373087380711201, "grad_norm": 0.57124924659729, "learning_rate": 1.0978e-06, "loss": 0.0034, "step": 44512 }, { "epoch": 2.373140694140854, "grad_norm": 1.7031879425048828, "learning_rate": 1.0976e-06, "loss": 0.0163, "step": 44513 }, { "epoch": 2.373194007570507, "grad_norm": 3.584995746612549, "learning_rate": 1.0974e-06, "loss": 0.0371, "step": 44514 }, { "epoch": 2.37324732100016, "grad_norm": 1.6722761392593384, "learning_rate": 1.0972e-06, "loss": 0.0083, "step": 44515 }, { "epoch": 2.373300634429813, "grad_norm": 2.7616539001464844, "learning_rate": 1.097e-06, "loss": 0.0193, "step": 44516 }, { "epoch": 2.373353947859466, "grad_norm": 0.5030500292778015, "learning_rate": 1.0968e-06, "loss": 0.0038, "step": 44517 }, { "epoch": 2.3734072612891186, "grad_norm": 0.554559588432312, "learning_rate": 1.0966e-06, "loss": 0.0038, "step": 44518 }, { "epoch": 2.373460574718772, "grad_norm": 4.603774070739746, "learning_rate": 1.0964e-06, "loss": 0.0091, "step": 44519 }, { "epoch": 2.3735138881484246, "grad_norm": 1.9760491847991943, "learning_rate": 1.0962e-06, "loss": 0.0127, "step": 44520 }, { "epoch": 2.3735672015780773, "grad_norm": 4.876843452453613, "learning_rate": 1.0960000000000002e-06, "loss": 0.0132, "step": 44521 }, { "epoch": 2.3736205150077305, "grad_norm": 0.5188691020011902, "learning_rate": 1.0958e-06, "loss": 0.0029, "step": 44522 }, { "epoch": 2.3736738284373833, "grad_norm": 1.130699634552002, "learning_rate": 1.0956000000000001e-06, "loss": 0.0171, "step": 44523 }, { "epoch": 2.3737271418670365, "grad_norm": 1.9968485832214355, "learning_rate": 1.0954000000000001e-06, "loss": 0.0303, "step": 44524 }, { "epoch": 2.3737804552966892, "grad_norm": 0.4227517247200012, "learning_rate": 1.0952000000000001e-06, "loss": 0.0038, "step": 44525 }, { "epoch": 2.373833768726342, "grad_norm": 2.692735195159912, "learning_rate": 1.095e-06, "loss": 0.0125, "step": 44526 }, { "epoch": 2.373887082155995, "grad_norm": 1.51261305809021, "learning_rate": 1.0948e-06, "loss": 0.0147, "step": 44527 }, { "epoch": 2.373940395585648, "grad_norm": 3.1073241233825684, "learning_rate": 1.0946e-06, "loss": 0.0167, "step": 44528 }, { "epoch": 2.373993709015301, "grad_norm": 1.6665120124816895, "learning_rate": 1.0944e-06, "loss": 0.0119, "step": 44529 }, { "epoch": 2.374047022444954, "grad_norm": 2.148223400115967, "learning_rate": 1.0942e-06, "loss": 0.0101, "step": 44530 }, { "epoch": 2.3741003358746067, "grad_norm": 1.1520780324935913, "learning_rate": 1.094e-06, "loss": 0.0173, "step": 44531 }, { "epoch": 2.37415364930426, "grad_norm": 0.8817326426506042, "learning_rate": 1.0938000000000002e-06, "loss": 0.0045, "step": 44532 }, { "epoch": 2.3742069627339126, "grad_norm": 1.4202619791030884, "learning_rate": 1.0936e-06, "loss": 0.0172, "step": 44533 }, { "epoch": 2.3742602761635654, "grad_norm": 1.5310101509094238, "learning_rate": 1.0934000000000002e-06, "loss": 0.0189, "step": 44534 }, { "epoch": 2.3743135895932186, "grad_norm": 1.2702378034591675, "learning_rate": 1.0932000000000002e-06, "loss": 0.0135, "step": 44535 }, { "epoch": 2.3743669030228713, "grad_norm": 1.8049900531768799, "learning_rate": 1.093e-06, "loss": 0.0071, "step": 44536 }, { "epoch": 2.3744202164525245, "grad_norm": 3.9664745330810547, "learning_rate": 1.0928000000000002e-06, "loss": 0.0428, "step": 44537 }, { "epoch": 2.3744735298821773, "grad_norm": 1.643657922744751, "learning_rate": 1.0926000000000001e-06, "loss": 0.0121, "step": 44538 }, { "epoch": 2.37452684331183, "grad_norm": 3.394462823867798, "learning_rate": 1.0924000000000001e-06, "loss": 0.0366, "step": 44539 }, { "epoch": 2.3745801567414833, "grad_norm": 1.5804108381271362, "learning_rate": 1.0922e-06, "loss": 0.0125, "step": 44540 }, { "epoch": 2.374633470171136, "grad_norm": 1.4120441675186157, "learning_rate": 1.092e-06, "loss": 0.0163, "step": 44541 }, { "epoch": 2.374686783600789, "grad_norm": 0.9479020833969116, "learning_rate": 1.0918e-06, "loss": 0.0062, "step": 44542 }, { "epoch": 2.374740097030442, "grad_norm": 2.0456109046936035, "learning_rate": 1.0916e-06, "loss": 0.0299, "step": 44543 }, { "epoch": 2.3747934104600947, "grad_norm": 0.4856194853782654, "learning_rate": 1.0914e-06, "loss": 0.0035, "step": 44544 }, { "epoch": 2.374846723889748, "grad_norm": 2.2994863986968994, "learning_rate": 1.0912e-06, "loss": 0.0183, "step": 44545 }, { "epoch": 2.3749000373194007, "grad_norm": 1.7955596446990967, "learning_rate": 1.091e-06, "loss": 0.0165, "step": 44546 }, { "epoch": 2.3749533507490534, "grad_norm": 1.3130390644073486, "learning_rate": 1.0908e-06, "loss": 0.006, "step": 44547 }, { "epoch": 2.3750066641787067, "grad_norm": 1.2537944316864014, "learning_rate": 1.0906000000000002e-06, "loss": 0.0043, "step": 44548 }, { "epoch": 2.3750599776083594, "grad_norm": 2.0005757808685303, "learning_rate": 1.0904e-06, "loss": 0.0189, "step": 44549 }, { "epoch": 2.3751132910380126, "grad_norm": 0.7029759287834167, "learning_rate": 1.0902000000000002e-06, "loss": 0.0035, "step": 44550 }, { "epoch": 2.3751666044676654, "grad_norm": 2.0198683738708496, "learning_rate": 1.0900000000000002e-06, "loss": 0.0263, "step": 44551 }, { "epoch": 2.3752199178973186, "grad_norm": 2.181112051010132, "learning_rate": 1.0898e-06, "loss": 0.0289, "step": 44552 }, { "epoch": 2.3752732313269713, "grad_norm": 1.6638987064361572, "learning_rate": 1.0896000000000001e-06, "loss": 0.0122, "step": 44553 }, { "epoch": 2.375326544756624, "grad_norm": 2.2215535640716553, "learning_rate": 1.0894000000000001e-06, "loss": 0.0108, "step": 44554 }, { "epoch": 2.3753798581862773, "grad_norm": 2.4459903240203857, "learning_rate": 1.0892e-06, "loss": 0.0162, "step": 44555 }, { "epoch": 2.37543317161593, "grad_norm": 4.352564334869385, "learning_rate": 1.089e-06, "loss": 0.0146, "step": 44556 }, { "epoch": 2.375486485045583, "grad_norm": 1.3116925954818726, "learning_rate": 1.0888e-06, "loss": 0.0079, "step": 44557 }, { "epoch": 2.375539798475236, "grad_norm": 0.5971168279647827, "learning_rate": 1.0886e-06, "loss": 0.0036, "step": 44558 }, { "epoch": 2.3755931119048888, "grad_norm": 1.2829945087432861, "learning_rate": 1.0884000000000002e-06, "loss": 0.007, "step": 44559 }, { "epoch": 2.375646425334542, "grad_norm": 2.6581618785858154, "learning_rate": 1.0882e-06, "loss": 0.0226, "step": 44560 }, { "epoch": 2.3756997387641947, "grad_norm": 1.0670429468154907, "learning_rate": 1.088e-06, "loss": 0.0057, "step": 44561 }, { "epoch": 2.3757530521938475, "grad_norm": 1.6178078651428223, "learning_rate": 1.0878000000000002e-06, "loss": 0.0071, "step": 44562 }, { "epoch": 2.3758063656235007, "grad_norm": 2.427248001098633, "learning_rate": 1.0876e-06, "loss": 0.0318, "step": 44563 }, { "epoch": 2.3758596790531534, "grad_norm": 1.502212405204773, "learning_rate": 1.0874000000000002e-06, "loss": 0.021, "step": 44564 }, { "epoch": 2.3759129924828066, "grad_norm": 1.3531053066253662, "learning_rate": 1.0872000000000002e-06, "loss": 0.0185, "step": 44565 }, { "epoch": 2.3759663059124594, "grad_norm": 4.261025905609131, "learning_rate": 1.0870000000000001e-06, "loss": 0.0326, "step": 44566 }, { "epoch": 2.376019619342112, "grad_norm": 2.558181047439575, "learning_rate": 1.0868000000000001e-06, "loss": 0.0116, "step": 44567 }, { "epoch": 2.3760729327717653, "grad_norm": 1.1362848281860352, "learning_rate": 1.0866000000000001e-06, "loss": 0.0065, "step": 44568 }, { "epoch": 2.376126246201418, "grad_norm": 4.803584098815918, "learning_rate": 1.0864e-06, "loss": 0.0242, "step": 44569 }, { "epoch": 2.376179559631071, "grad_norm": 2.0863394737243652, "learning_rate": 1.0862e-06, "loss": 0.0194, "step": 44570 }, { "epoch": 2.376232873060724, "grad_norm": 3.3731272220611572, "learning_rate": 1.086e-06, "loss": 0.0112, "step": 44571 }, { "epoch": 2.376286186490377, "grad_norm": 3.3428738117218018, "learning_rate": 1.0858e-06, "loss": 0.0223, "step": 44572 }, { "epoch": 2.37633949992003, "grad_norm": 2.9562184810638428, "learning_rate": 1.0856e-06, "loss": 0.013, "step": 44573 }, { "epoch": 2.3763928133496828, "grad_norm": 1.6153204441070557, "learning_rate": 1.0854e-06, "loss": 0.0317, "step": 44574 }, { "epoch": 2.3764461267793355, "grad_norm": 1.3701540231704712, "learning_rate": 1.0852000000000002e-06, "loss": 0.0035, "step": 44575 }, { "epoch": 2.3764994402089887, "grad_norm": 1.0223305225372314, "learning_rate": 1.085e-06, "loss": 0.0047, "step": 44576 }, { "epoch": 2.3765527536386415, "grad_norm": 2.9302287101745605, "learning_rate": 1.0848e-06, "loss": 0.037, "step": 44577 }, { "epoch": 2.3766060670682947, "grad_norm": 2.472456455230713, "learning_rate": 1.0846000000000002e-06, "loss": 0.0098, "step": 44578 }, { "epoch": 2.3766593804979474, "grad_norm": 1.82285475730896, "learning_rate": 1.0844e-06, "loss": 0.0184, "step": 44579 }, { "epoch": 2.3767126939276, "grad_norm": 2.4096269607543945, "learning_rate": 1.0842000000000001e-06, "loss": 0.0131, "step": 44580 }, { "epoch": 2.3767660073572534, "grad_norm": 0.6546390056610107, "learning_rate": 1.0840000000000001e-06, "loss": 0.0032, "step": 44581 }, { "epoch": 2.376819320786906, "grad_norm": 0.47281840443611145, "learning_rate": 1.0838000000000001e-06, "loss": 0.0035, "step": 44582 }, { "epoch": 2.376872634216559, "grad_norm": 1.1869412660598755, "learning_rate": 1.0836e-06, "loss": 0.0064, "step": 44583 }, { "epoch": 2.376925947646212, "grad_norm": 0.9556449055671692, "learning_rate": 1.0834e-06, "loss": 0.0203, "step": 44584 }, { "epoch": 2.376979261075865, "grad_norm": 0.25812169909477234, "learning_rate": 1.0832e-06, "loss": 0.0015, "step": 44585 }, { "epoch": 2.377032574505518, "grad_norm": 2.1825039386749268, "learning_rate": 1.083e-06, "loss": 0.0392, "step": 44586 }, { "epoch": 2.377085887935171, "grad_norm": 2.4797401428222656, "learning_rate": 1.0828e-06, "loss": 0.0232, "step": 44587 }, { "epoch": 2.377139201364824, "grad_norm": 1.1466144323349, "learning_rate": 1.0826e-06, "loss": 0.0041, "step": 44588 }, { "epoch": 2.377192514794477, "grad_norm": 1.3146836757659912, "learning_rate": 1.0824000000000002e-06, "loss": 0.0133, "step": 44589 }, { "epoch": 2.3772458282241296, "grad_norm": 2.3093783855438232, "learning_rate": 1.0822e-06, "loss": 0.0083, "step": 44590 }, { "epoch": 2.3772991416537828, "grad_norm": 2.599018096923828, "learning_rate": 1.0820000000000002e-06, "loss": 0.0185, "step": 44591 }, { "epoch": 2.3773524550834355, "grad_norm": 3.224787950515747, "learning_rate": 1.0818000000000002e-06, "loss": 0.0308, "step": 44592 }, { "epoch": 2.3774057685130883, "grad_norm": 0.5125934481620789, "learning_rate": 1.0816000000000002e-06, "loss": 0.003, "step": 44593 }, { "epoch": 2.3774590819427415, "grad_norm": 2.1036622524261475, "learning_rate": 1.0814000000000001e-06, "loss": 0.0139, "step": 44594 }, { "epoch": 2.3775123953723942, "grad_norm": 3.5370137691497803, "learning_rate": 1.0812000000000001e-06, "loss": 0.0407, "step": 44595 }, { "epoch": 2.3775657088020474, "grad_norm": 2.2535905838012695, "learning_rate": 1.0810000000000001e-06, "loss": 0.0225, "step": 44596 }, { "epoch": 2.3776190222317, "grad_norm": 1.483273983001709, "learning_rate": 1.0808e-06, "loss": 0.0155, "step": 44597 }, { "epoch": 2.377672335661353, "grad_norm": 1.4292019605636597, "learning_rate": 1.0806e-06, "loss": 0.0101, "step": 44598 }, { "epoch": 2.377725649091006, "grad_norm": 1.0891056060791016, "learning_rate": 1.0804e-06, "loss": 0.0059, "step": 44599 }, { "epoch": 2.377778962520659, "grad_norm": 2.231689453125, "learning_rate": 1.0802e-06, "loss": 0.0177, "step": 44600 }, { "epoch": 2.377832275950312, "grad_norm": 3.86791729927063, "learning_rate": 1.08e-06, "loss": 0.0286, "step": 44601 }, { "epoch": 2.377885589379965, "grad_norm": 3.319929838180542, "learning_rate": 1.0798000000000002e-06, "loss": 0.0133, "step": 44602 }, { "epoch": 2.3779389028096176, "grad_norm": 2.0930089950561523, "learning_rate": 1.0796e-06, "loss": 0.033, "step": 44603 }, { "epoch": 2.377992216239271, "grad_norm": 0.6833306550979614, "learning_rate": 1.0794e-06, "loss": 0.0053, "step": 44604 }, { "epoch": 2.3780455296689236, "grad_norm": 1.3229769468307495, "learning_rate": 1.0792000000000002e-06, "loss": 0.0026, "step": 44605 }, { "epoch": 2.3780988430985763, "grad_norm": 2.4272713661193848, "learning_rate": 1.079e-06, "loss": 0.0151, "step": 44606 }, { "epoch": 2.3781521565282295, "grad_norm": 2.3785808086395264, "learning_rate": 1.0788000000000002e-06, "loss": 0.0265, "step": 44607 }, { "epoch": 2.3782054699578823, "grad_norm": 1.7678743600845337, "learning_rate": 1.0786000000000001e-06, "loss": 0.0087, "step": 44608 }, { "epoch": 2.3782587833875355, "grad_norm": 2.4247288703918457, "learning_rate": 1.0784000000000001e-06, "loss": 0.0093, "step": 44609 }, { "epoch": 2.3783120968171882, "grad_norm": 2.425623655319214, "learning_rate": 1.0782000000000001e-06, "loss": 0.0183, "step": 44610 }, { "epoch": 2.378365410246841, "grad_norm": 1.4149078130722046, "learning_rate": 1.078e-06, "loss": 0.0052, "step": 44611 }, { "epoch": 2.378418723676494, "grad_norm": 1.8473687171936035, "learning_rate": 1.0778e-06, "loss": 0.0077, "step": 44612 }, { "epoch": 2.378472037106147, "grad_norm": 1.930033564567566, "learning_rate": 1.0776e-06, "loss": 0.0278, "step": 44613 }, { "epoch": 2.3785253505358, "grad_norm": 1.2126007080078125, "learning_rate": 1.0774e-06, "loss": 0.0073, "step": 44614 }, { "epoch": 2.378578663965453, "grad_norm": 0.6781184673309326, "learning_rate": 1.0772e-06, "loss": 0.0033, "step": 44615 }, { "epoch": 2.3786319773951057, "grad_norm": 0.5786640048027039, "learning_rate": 1.0770000000000002e-06, "loss": 0.004, "step": 44616 }, { "epoch": 2.378685290824759, "grad_norm": 1.5892459154129028, "learning_rate": 1.0768e-06, "loss": 0.0168, "step": 44617 }, { "epoch": 2.3787386042544116, "grad_norm": 2.1018693447113037, "learning_rate": 1.0766000000000002e-06, "loss": 0.0097, "step": 44618 }, { "epoch": 2.3787919176840644, "grad_norm": 1.3736339807510376, "learning_rate": 1.0764000000000002e-06, "loss": 0.0196, "step": 44619 }, { "epoch": 2.3788452311137176, "grad_norm": 2.72453236579895, "learning_rate": 1.0762e-06, "loss": 0.0314, "step": 44620 }, { "epoch": 2.3788985445433704, "grad_norm": 2.2105138301849365, "learning_rate": 1.0760000000000002e-06, "loss": 0.0128, "step": 44621 }, { "epoch": 2.3789518579730236, "grad_norm": 2.9772207736968994, "learning_rate": 1.0758000000000001e-06, "loss": 0.0401, "step": 44622 }, { "epoch": 2.3790051714026763, "grad_norm": 0.31594571471214294, "learning_rate": 1.0756000000000001e-06, "loss": 0.002, "step": 44623 }, { "epoch": 2.3790584848323295, "grad_norm": 2.165024995803833, "learning_rate": 1.0754000000000001e-06, "loss": 0.0317, "step": 44624 }, { "epoch": 2.3791117982619823, "grad_norm": 1.1376378536224365, "learning_rate": 1.0752e-06, "loss": 0.0053, "step": 44625 }, { "epoch": 2.379165111691635, "grad_norm": 0.7516162991523743, "learning_rate": 1.075e-06, "loss": 0.0045, "step": 44626 }, { "epoch": 2.3792184251212882, "grad_norm": 0.3252486288547516, "learning_rate": 1.0748e-06, "loss": 0.0029, "step": 44627 }, { "epoch": 2.379271738550941, "grad_norm": 0.9201695919036865, "learning_rate": 1.0746e-06, "loss": 0.0194, "step": 44628 }, { "epoch": 2.3793250519805937, "grad_norm": 3.5344345569610596, "learning_rate": 1.0744e-06, "loss": 0.0361, "step": 44629 }, { "epoch": 2.379378365410247, "grad_norm": 1.9380978345870972, "learning_rate": 1.0742e-06, "loss": 0.0083, "step": 44630 }, { "epoch": 2.3794316788398997, "grad_norm": 2.193138599395752, "learning_rate": 1.074e-06, "loss": 0.022, "step": 44631 }, { "epoch": 2.3794849922695525, "grad_norm": 1.0931181907653809, "learning_rate": 1.0738000000000002e-06, "loss": 0.0273, "step": 44632 }, { "epoch": 2.3795383056992057, "grad_norm": 1.2046500444412231, "learning_rate": 1.0736e-06, "loss": 0.0167, "step": 44633 }, { "epoch": 2.3795916191288584, "grad_norm": 1.549221396446228, "learning_rate": 1.0734000000000002e-06, "loss": 0.0115, "step": 44634 }, { "epoch": 2.3796449325585116, "grad_norm": 1.5477898120880127, "learning_rate": 1.0732000000000002e-06, "loss": 0.0168, "step": 44635 }, { "epoch": 2.3796982459881644, "grad_norm": 1.0458877086639404, "learning_rate": 1.0730000000000001e-06, "loss": 0.0059, "step": 44636 }, { "epoch": 2.3797515594178176, "grad_norm": 1.8530620336532593, "learning_rate": 1.0728000000000001e-06, "loss": 0.0198, "step": 44637 }, { "epoch": 2.3798048728474703, "grad_norm": 1.2968988418579102, "learning_rate": 1.0726000000000001e-06, "loss": 0.0052, "step": 44638 }, { "epoch": 2.379858186277123, "grad_norm": 1.7512015104293823, "learning_rate": 1.0724e-06, "loss": 0.01, "step": 44639 }, { "epoch": 2.3799114997067763, "grad_norm": 1.184727430343628, "learning_rate": 1.0722e-06, "loss": 0.007, "step": 44640 }, { "epoch": 2.379964813136429, "grad_norm": 2.00217604637146, "learning_rate": 1.072e-06, "loss": 0.0081, "step": 44641 }, { "epoch": 2.380018126566082, "grad_norm": 3.91650652885437, "learning_rate": 1.0718e-06, "loss": 0.0602, "step": 44642 }, { "epoch": 2.380071439995735, "grad_norm": 0.9221994876861572, "learning_rate": 1.0716e-06, "loss": 0.0037, "step": 44643 }, { "epoch": 2.3801247534253878, "grad_norm": 0.2804133892059326, "learning_rate": 1.0714e-06, "loss": 0.0027, "step": 44644 }, { "epoch": 2.380178066855041, "grad_norm": 0.9569138884544373, "learning_rate": 1.0712000000000002e-06, "loss": 0.0097, "step": 44645 }, { "epoch": 2.3802313802846937, "grad_norm": 1.5626434087753296, "learning_rate": 1.071e-06, "loss": 0.0096, "step": 44646 }, { "epoch": 2.3802846937143465, "grad_norm": 0.30181312561035156, "learning_rate": 1.0708e-06, "loss": 0.0023, "step": 44647 }, { "epoch": 2.3803380071439997, "grad_norm": 1.4485318660736084, "learning_rate": 1.0706000000000002e-06, "loss": 0.0118, "step": 44648 }, { "epoch": 2.3803913205736524, "grad_norm": 0.540149450302124, "learning_rate": 1.0704e-06, "loss": 0.0035, "step": 44649 }, { "epoch": 2.3804446340033056, "grad_norm": 1.5379364490509033, "learning_rate": 1.0702000000000001e-06, "loss": 0.0059, "step": 44650 }, { "epoch": 2.3804979474329584, "grad_norm": 0.7839004993438721, "learning_rate": 1.0700000000000001e-06, "loss": 0.0051, "step": 44651 }, { "epoch": 2.380551260862611, "grad_norm": 2.108335494995117, "learning_rate": 1.0698000000000001e-06, "loss": 0.0239, "step": 44652 }, { "epoch": 2.3806045742922644, "grad_norm": 0.8170751929283142, "learning_rate": 1.0696e-06, "loss": 0.0038, "step": 44653 }, { "epoch": 2.380657887721917, "grad_norm": 6.374882221221924, "learning_rate": 1.0694e-06, "loss": 0.0349, "step": 44654 }, { "epoch": 2.38071120115157, "grad_norm": 0.3415520191192627, "learning_rate": 1.0692e-06, "loss": 0.0015, "step": 44655 }, { "epoch": 2.380764514581223, "grad_norm": 1.956735610961914, "learning_rate": 1.069e-06, "loss": 0.0208, "step": 44656 }, { "epoch": 2.380817828010876, "grad_norm": 1.0394304990768433, "learning_rate": 1.0688e-06, "loss": 0.0051, "step": 44657 }, { "epoch": 2.380871141440529, "grad_norm": 0.3955654203891754, "learning_rate": 1.0686e-06, "loss": 0.0031, "step": 44658 }, { "epoch": 2.380924454870182, "grad_norm": 1.4399409294128418, "learning_rate": 1.0684000000000002e-06, "loss": 0.0196, "step": 44659 }, { "epoch": 2.3809777682998345, "grad_norm": 1.482423186302185, "learning_rate": 1.0682e-06, "loss": 0.0196, "step": 44660 }, { "epoch": 2.3810310817294877, "grad_norm": 2.0997068881988525, "learning_rate": 1.0680000000000002e-06, "loss": 0.0173, "step": 44661 }, { "epoch": 2.3810843951591405, "grad_norm": 1.5327568054199219, "learning_rate": 1.0678000000000002e-06, "loss": 0.0106, "step": 44662 }, { "epoch": 2.3811377085887937, "grad_norm": 2.6475141048431396, "learning_rate": 1.0676e-06, "loss": 0.0064, "step": 44663 }, { "epoch": 2.3811910220184465, "grad_norm": 2.037999391555786, "learning_rate": 1.0674000000000001e-06, "loss": 0.0201, "step": 44664 }, { "epoch": 2.381244335448099, "grad_norm": 1.9722999334335327, "learning_rate": 1.0672000000000001e-06, "loss": 0.0313, "step": 44665 }, { "epoch": 2.3812976488777524, "grad_norm": 3.3443381786346436, "learning_rate": 1.0670000000000001e-06, "loss": 0.0106, "step": 44666 }, { "epoch": 2.381350962307405, "grad_norm": 1.9693315029144287, "learning_rate": 1.0668e-06, "loss": 0.0093, "step": 44667 }, { "epoch": 2.381404275737058, "grad_norm": 1.3521828651428223, "learning_rate": 1.0666e-06, "loss": 0.021, "step": 44668 }, { "epoch": 2.381457589166711, "grad_norm": 1.8608134984970093, "learning_rate": 1.0664e-06, "loss": 0.0156, "step": 44669 }, { "epoch": 2.381510902596364, "grad_norm": 0.6564356088638306, "learning_rate": 1.0662e-06, "loss": 0.0032, "step": 44670 }, { "epoch": 2.381564216026017, "grad_norm": 3.0183990001678467, "learning_rate": 1.066e-06, "loss": 0.0346, "step": 44671 }, { "epoch": 2.38161752945567, "grad_norm": 2.405726909637451, "learning_rate": 1.0658e-06, "loss": 0.0091, "step": 44672 }, { "epoch": 2.381670842885323, "grad_norm": 1.5684233903884888, "learning_rate": 1.0656e-06, "loss": 0.007, "step": 44673 }, { "epoch": 2.381724156314976, "grad_norm": 0.6072782278060913, "learning_rate": 1.0654e-06, "loss": 0.0035, "step": 44674 }, { "epoch": 2.3817774697446286, "grad_norm": 2.5400166511535645, "learning_rate": 1.0652000000000002e-06, "loss": 0.0145, "step": 44675 }, { "epoch": 2.3818307831742818, "grad_norm": 2.2483246326446533, "learning_rate": 1.065e-06, "loss": 0.0181, "step": 44676 }, { "epoch": 2.3818840966039345, "grad_norm": 2.811690092086792, "learning_rate": 1.0648000000000002e-06, "loss": 0.0209, "step": 44677 }, { "epoch": 2.3819374100335873, "grad_norm": 0.7233427166938782, "learning_rate": 1.0646000000000001e-06, "loss": 0.0039, "step": 44678 }, { "epoch": 2.3819907234632405, "grad_norm": 1.577397108078003, "learning_rate": 1.0644000000000001e-06, "loss": 0.0138, "step": 44679 }, { "epoch": 2.3820440368928932, "grad_norm": 0.8237941265106201, "learning_rate": 1.0642000000000001e-06, "loss": 0.0051, "step": 44680 }, { "epoch": 2.3820973503225464, "grad_norm": 0.8049858212471008, "learning_rate": 1.064e-06, "loss": 0.0054, "step": 44681 }, { "epoch": 2.382150663752199, "grad_norm": 3.8664798736572266, "learning_rate": 1.0638e-06, "loss": 0.0136, "step": 44682 }, { "epoch": 2.382203977181852, "grad_norm": 1.9410957098007202, "learning_rate": 1.0636e-06, "loss": 0.0236, "step": 44683 }, { "epoch": 2.382257290611505, "grad_norm": 4.002501487731934, "learning_rate": 1.0634e-06, "loss": 0.0176, "step": 44684 }, { "epoch": 2.382310604041158, "grad_norm": 1.1872836351394653, "learning_rate": 1.0632e-06, "loss": 0.0052, "step": 44685 }, { "epoch": 2.382363917470811, "grad_norm": 1.1723068952560425, "learning_rate": 1.0630000000000002e-06, "loss": 0.0043, "step": 44686 }, { "epoch": 2.382417230900464, "grad_norm": 1.7012135982513428, "learning_rate": 1.0628e-06, "loss": 0.0111, "step": 44687 }, { "epoch": 2.3824705443301166, "grad_norm": 2.137240171432495, "learning_rate": 1.0626e-06, "loss": 0.016, "step": 44688 }, { "epoch": 2.38252385775977, "grad_norm": 1.4984440803527832, "learning_rate": 1.0624000000000002e-06, "loss": 0.0171, "step": 44689 }, { "epoch": 2.3825771711894226, "grad_norm": 2.3548641204833984, "learning_rate": 1.0622e-06, "loss": 0.0379, "step": 44690 }, { "epoch": 2.3826304846190753, "grad_norm": 1.008152961730957, "learning_rate": 1.0620000000000002e-06, "loss": 0.0049, "step": 44691 }, { "epoch": 2.3826837980487285, "grad_norm": 4.106822490692139, "learning_rate": 1.0618000000000002e-06, "loss": 0.0203, "step": 44692 }, { "epoch": 2.3827371114783813, "grad_norm": 0.5146475434303284, "learning_rate": 1.0616000000000001e-06, "loss": 0.004, "step": 44693 }, { "epoch": 2.3827904249080345, "grad_norm": 1.8207155466079712, "learning_rate": 1.0614000000000001e-06, "loss": 0.0172, "step": 44694 }, { "epoch": 2.3828437383376873, "grad_norm": 1.8098993301391602, "learning_rate": 1.0612e-06, "loss": 0.0211, "step": 44695 }, { "epoch": 2.38289705176734, "grad_norm": 1.7126215696334839, "learning_rate": 1.061e-06, "loss": 0.026, "step": 44696 }, { "epoch": 2.382950365196993, "grad_norm": 0.8922473192214966, "learning_rate": 1.0608e-06, "loss": 0.005, "step": 44697 }, { "epoch": 2.383003678626646, "grad_norm": 1.6388661861419678, "learning_rate": 1.0606e-06, "loss": 0.0234, "step": 44698 }, { "epoch": 2.383056992056299, "grad_norm": 1.2545158863067627, "learning_rate": 1.0604e-06, "loss": 0.0043, "step": 44699 }, { "epoch": 2.383110305485952, "grad_norm": 1.6293829679489136, "learning_rate": 1.0602e-06, "loss": 0.0098, "step": 44700 }, { "epoch": 2.3831636189156047, "grad_norm": 1.2220088243484497, "learning_rate": 1.06e-06, "loss": 0.0215, "step": 44701 }, { "epoch": 2.383216932345258, "grad_norm": 4.291385650634766, "learning_rate": 1.0598000000000002e-06, "loss": 0.0161, "step": 44702 }, { "epoch": 2.3832702457749106, "grad_norm": 1.230115294456482, "learning_rate": 1.0596e-06, "loss": 0.0194, "step": 44703 }, { "epoch": 2.3833235592045634, "grad_norm": 1.9029842615127563, "learning_rate": 1.0594000000000002e-06, "loss": 0.0328, "step": 44704 }, { "epoch": 2.3833768726342166, "grad_norm": 2.8213324546813965, "learning_rate": 1.0592000000000002e-06, "loss": 0.0265, "step": 44705 }, { "epoch": 2.3834301860638694, "grad_norm": 1.1105046272277832, "learning_rate": 1.059e-06, "loss": 0.0045, "step": 44706 }, { "epoch": 2.3834834994935226, "grad_norm": 0.6872832775115967, "learning_rate": 1.0588000000000001e-06, "loss": 0.0033, "step": 44707 }, { "epoch": 2.3835368129231753, "grad_norm": 3.173396110534668, "learning_rate": 1.0586000000000001e-06, "loss": 0.0302, "step": 44708 }, { "epoch": 2.3835901263528285, "grad_norm": 0.18258553743362427, "learning_rate": 1.0584000000000001e-06, "loss": 0.0013, "step": 44709 }, { "epoch": 2.3836434397824813, "grad_norm": 0.7229796051979065, "learning_rate": 1.0582e-06, "loss": 0.0041, "step": 44710 }, { "epoch": 2.383696753212134, "grad_norm": 4.005563735961914, "learning_rate": 1.058e-06, "loss": 0.0201, "step": 44711 }, { "epoch": 2.3837500666417872, "grad_norm": 1.4558155536651611, "learning_rate": 1.0578e-06, "loss": 0.0123, "step": 44712 }, { "epoch": 2.38380338007144, "grad_norm": 2.0135905742645264, "learning_rate": 1.0576000000000003e-06, "loss": 0.0192, "step": 44713 }, { "epoch": 2.3838566935010927, "grad_norm": 1.075613021850586, "learning_rate": 1.0574e-06, "loss": 0.0177, "step": 44714 }, { "epoch": 2.383910006930746, "grad_norm": 1.5632777214050293, "learning_rate": 1.0572e-06, "loss": 0.0243, "step": 44715 }, { "epoch": 2.3839633203603987, "grad_norm": 2.5684571266174316, "learning_rate": 1.0570000000000002e-06, "loss": 0.0127, "step": 44716 }, { "epoch": 2.3840166337900515, "grad_norm": 0.36165711283683777, "learning_rate": 1.0568e-06, "loss": 0.0016, "step": 44717 }, { "epoch": 2.3840699472197047, "grad_norm": 2.2120676040649414, "learning_rate": 1.0566000000000002e-06, "loss": 0.0221, "step": 44718 }, { "epoch": 2.3841232606493574, "grad_norm": 1.7445942163467407, "learning_rate": 1.0564000000000002e-06, "loss": 0.0161, "step": 44719 }, { "epoch": 2.3841765740790106, "grad_norm": 1.6724380254745483, "learning_rate": 1.0562000000000002e-06, "loss": 0.0091, "step": 44720 }, { "epoch": 2.3842298875086634, "grad_norm": 1.6673489809036255, "learning_rate": 1.0560000000000001e-06, "loss": 0.0046, "step": 44721 }, { "epoch": 2.3842832009383166, "grad_norm": 0.4524153769016266, "learning_rate": 1.0558000000000001e-06, "loss": 0.0027, "step": 44722 }, { "epoch": 2.3843365143679693, "grad_norm": 2.478346347808838, "learning_rate": 1.0556000000000001e-06, "loss": 0.0063, "step": 44723 }, { "epoch": 2.384389827797622, "grad_norm": 2.951514482498169, "learning_rate": 1.0554e-06, "loss": 0.0136, "step": 44724 }, { "epoch": 2.3844431412272753, "grad_norm": 2.2128469944000244, "learning_rate": 1.0552e-06, "loss": 0.0125, "step": 44725 }, { "epoch": 2.384496454656928, "grad_norm": 1.7409067153930664, "learning_rate": 1.055e-06, "loss": 0.0101, "step": 44726 }, { "epoch": 2.384549768086581, "grad_norm": 1.7869386672973633, "learning_rate": 1.0548e-06, "loss": 0.0092, "step": 44727 }, { "epoch": 2.384603081516234, "grad_norm": 1.4907740354537964, "learning_rate": 1.0546e-06, "loss": 0.0111, "step": 44728 }, { "epoch": 2.3846563949458868, "grad_norm": 3.3080673217773438, "learning_rate": 1.0544000000000002e-06, "loss": 0.0216, "step": 44729 }, { "epoch": 2.38470970837554, "grad_norm": 0.771244466304779, "learning_rate": 1.0542e-06, "loss": 0.0052, "step": 44730 }, { "epoch": 2.3847630218051927, "grad_norm": 4.695533275604248, "learning_rate": 1.054e-06, "loss": 0.0104, "step": 44731 }, { "epoch": 2.3848163352348455, "grad_norm": 0.3985271155834198, "learning_rate": 1.0538000000000002e-06, "loss": 0.0021, "step": 44732 }, { "epoch": 2.3848696486644987, "grad_norm": 1.2398943901062012, "learning_rate": 1.0536e-06, "loss": 0.0138, "step": 44733 }, { "epoch": 2.3849229620941514, "grad_norm": 2.424138307571411, "learning_rate": 1.0534000000000002e-06, "loss": 0.0076, "step": 44734 }, { "epoch": 2.3849762755238046, "grad_norm": 4.039971351623535, "learning_rate": 1.0532000000000001e-06, "loss": 0.0245, "step": 44735 }, { "epoch": 2.3850295889534574, "grad_norm": 1.4477894306182861, "learning_rate": 1.0530000000000001e-06, "loss": 0.0234, "step": 44736 }, { "epoch": 2.38508290238311, "grad_norm": 2.0537123680114746, "learning_rate": 1.0528000000000001e-06, "loss": 0.0095, "step": 44737 }, { "epoch": 2.3851362158127634, "grad_norm": 3.2641406059265137, "learning_rate": 1.0526e-06, "loss": 0.0324, "step": 44738 }, { "epoch": 2.385189529242416, "grad_norm": 2.203022003173828, "learning_rate": 1.0524e-06, "loss": 0.023, "step": 44739 }, { "epoch": 2.385242842672069, "grad_norm": 2.0300164222717285, "learning_rate": 1.0522e-06, "loss": 0.0069, "step": 44740 }, { "epoch": 2.385296156101722, "grad_norm": 2.7132036685943604, "learning_rate": 1.052e-06, "loss": 0.0101, "step": 44741 }, { "epoch": 2.385349469531375, "grad_norm": 2.5541152954101562, "learning_rate": 1.0518e-06, "loss": 0.0366, "step": 44742 }, { "epoch": 2.385402782961028, "grad_norm": 1.9061291217803955, "learning_rate": 1.0516e-06, "loss": 0.0084, "step": 44743 }, { "epoch": 2.385456096390681, "grad_norm": 1.0659538507461548, "learning_rate": 1.0514e-06, "loss": 0.0238, "step": 44744 }, { "epoch": 2.3855094098203335, "grad_norm": 1.506654143333435, "learning_rate": 1.0512000000000002e-06, "loss": 0.019, "step": 44745 }, { "epoch": 2.3855627232499867, "grad_norm": 0.9217444658279419, "learning_rate": 1.051e-06, "loss": 0.019, "step": 44746 }, { "epoch": 2.3856160366796395, "grad_norm": 1.9543142318725586, "learning_rate": 1.0508000000000002e-06, "loss": 0.0133, "step": 44747 }, { "epoch": 2.3856693501092927, "grad_norm": 1.435106635093689, "learning_rate": 1.0506000000000002e-06, "loss": 0.0069, "step": 44748 }, { "epoch": 2.3857226635389455, "grad_norm": 1.1193461418151855, "learning_rate": 1.0504e-06, "loss": 0.0098, "step": 44749 }, { "epoch": 2.3857759769685982, "grad_norm": 2.7574286460876465, "learning_rate": 1.0502000000000001e-06, "loss": 0.0194, "step": 44750 }, { "epoch": 2.3858292903982514, "grad_norm": 2.1042089462280273, "learning_rate": 1.0500000000000001e-06, "loss": 0.0083, "step": 44751 }, { "epoch": 2.385882603827904, "grad_norm": 0.3897371292114258, "learning_rate": 1.0498e-06, "loss": 0.0029, "step": 44752 }, { "epoch": 2.385935917257557, "grad_norm": 2.9274580478668213, "learning_rate": 1.0496e-06, "loss": 0.019, "step": 44753 }, { "epoch": 2.38598923068721, "grad_norm": 2.06884503364563, "learning_rate": 1.0494e-06, "loss": 0.0247, "step": 44754 }, { "epoch": 2.386042544116863, "grad_norm": 0.7182199358940125, "learning_rate": 1.0492e-06, "loss": 0.003, "step": 44755 }, { "epoch": 2.386095857546516, "grad_norm": 2.380845308303833, "learning_rate": 1.049e-06, "loss": 0.0247, "step": 44756 }, { "epoch": 2.386149170976169, "grad_norm": 3.1208102703094482, "learning_rate": 1.0488e-06, "loss": 0.0172, "step": 44757 }, { "epoch": 2.386202484405822, "grad_norm": 1.171879529953003, "learning_rate": 1.0486e-06, "loss": 0.0046, "step": 44758 }, { "epoch": 2.386255797835475, "grad_norm": 4.055698871612549, "learning_rate": 1.0484000000000002e-06, "loss": 0.0374, "step": 44759 }, { "epoch": 2.3863091112651276, "grad_norm": 1.4538406133651733, "learning_rate": 1.0482e-06, "loss": 0.0084, "step": 44760 }, { "epoch": 2.3863624246947808, "grad_norm": 2.707669496536255, "learning_rate": 1.0480000000000002e-06, "loss": 0.0191, "step": 44761 }, { "epoch": 2.3864157381244335, "grad_norm": 4.927882194519043, "learning_rate": 1.0478000000000002e-06, "loss": 0.0398, "step": 44762 }, { "epoch": 2.3864690515540863, "grad_norm": 2.9941375255584717, "learning_rate": 1.0476000000000001e-06, "loss": 0.028, "step": 44763 }, { "epoch": 2.3865223649837395, "grad_norm": 2.148954391479492, "learning_rate": 1.0474000000000001e-06, "loss": 0.0288, "step": 44764 }, { "epoch": 2.3865756784133922, "grad_norm": 5.000340461730957, "learning_rate": 1.0472000000000001e-06, "loss": 0.0313, "step": 44765 }, { "epoch": 2.3866289918430454, "grad_norm": 1.9887877702713013, "learning_rate": 1.047e-06, "loss": 0.0384, "step": 44766 }, { "epoch": 2.386682305272698, "grad_norm": 1.399692177772522, "learning_rate": 1.0468e-06, "loss": 0.0097, "step": 44767 }, { "epoch": 2.386735618702351, "grad_norm": 0.45930856466293335, "learning_rate": 1.0466e-06, "loss": 0.0032, "step": 44768 }, { "epoch": 2.386788932132004, "grad_norm": 2.522279977798462, "learning_rate": 1.0464e-06, "loss": 0.01, "step": 44769 }, { "epoch": 2.386842245561657, "grad_norm": 2.134760856628418, "learning_rate": 1.0462e-06, "loss": 0.0062, "step": 44770 }, { "epoch": 2.38689555899131, "grad_norm": 0.9642691612243652, "learning_rate": 1.046e-06, "loss": 0.0058, "step": 44771 }, { "epoch": 2.386948872420963, "grad_norm": 2.360161781311035, "learning_rate": 1.0458000000000002e-06, "loss": 0.026, "step": 44772 }, { "epoch": 2.3870021858506156, "grad_norm": 1.0716243982315063, "learning_rate": 1.0456e-06, "loss": 0.0123, "step": 44773 }, { "epoch": 2.387055499280269, "grad_norm": 1.4014410972595215, "learning_rate": 1.0454e-06, "loss": 0.0068, "step": 44774 }, { "epoch": 2.3871088127099216, "grad_norm": 2.2462127208709717, "learning_rate": 1.0452000000000002e-06, "loss": 0.0322, "step": 44775 }, { "epoch": 2.3871621261395743, "grad_norm": 2.085085868835449, "learning_rate": 1.045e-06, "loss": 0.0424, "step": 44776 }, { "epoch": 2.3872154395692275, "grad_norm": 1.7846635580062866, "learning_rate": 1.0448000000000001e-06, "loss": 0.0127, "step": 44777 }, { "epoch": 2.3872687529988803, "grad_norm": 2.4451165199279785, "learning_rate": 1.0446000000000001e-06, "loss": 0.0246, "step": 44778 }, { "epoch": 2.3873220664285335, "grad_norm": 2.549020528793335, "learning_rate": 1.0444000000000001e-06, "loss": 0.0193, "step": 44779 }, { "epoch": 2.3873753798581863, "grad_norm": 1.7636909484863281, "learning_rate": 1.0442e-06, "loss": 0.0156, "step": 44780 }, { "epoch": 2.387428693287839, "grad_norm": 3.3975048065185547, "learning_rate": 1.044e-06, "loss": 0.0118, "step": 44781 }, { "epoch": 2.3874820067174922, "grad_norm": 3.068166494369507, "learning_rate": 1.0438e-06, "loss": 0.0086, "step": 44782 }, { "epoch": 2.387535320147145, "grad_norm": 3.531297206878662, "learning_rate": 1.0436e-06, "loss": 0.0252, "step": 44783 }, { "epoch": 2.387588633576798, "grad_norm": 3.22544002532959, "learning_rate": 1.0434e-06, "loss": 0.0297, "step": 44784 }, { "epoch": 2.387641947006451, "grad_norm": 0.20800864696502686, "learning_rate": 1.0432e-06, "loss": 0.0015, "step": 44785 }, { "epoch": 2.3876952604361037, "grad_norm": 2.6036980152130127, "learning_rate": 1.0430000000000002e-06, "loss": 0.0183, "step": 44786 }, { "epoch": 2.387748573865757, "grad_norm": 1.8987840414047241, "learning_rate": 1.0428e-06, "loss": 0.0245, "step": 44787 }, { "epoch": 2.3878018872954097, "grad_norm": 2.1469814777374268, "learning_rate": 1.0426000000000002e-06, "loss": 0.0214, "step": 44788 }, { "epoch": 2.3878552007250624, "grad_norm": 3.1610982418060303, "learning_rate": 1.0424000000000002e-06, "loss": 0.0215, "step": 44789 }, { "epoch": 2.3879085141547156, "grad_norm": 1.5153905153274536, "learning_rate": 1.0422e-06, "loss": 0.0047, "step": 44790 }, { "epoch": 2.3879618275843684, "grad_norm": 2.1030006408691406, "learning_rate": 1.0420000000000001e-06, "loss": 0.0108, "step": 44791 }, { "epoch": 2.3880151410140216, "grad_norm": 0.3839362859725952, "learning_rate": 1.0418000000000001e-06, "loss": 0.0032, "step": 44792 }, { "epoch": 2.3880684544436743, "grad_norm": 1.4913614988327026, "learning_rate": 1.0416000000000001e-06, "loss": 0.0073, "step": 44793 }, { "epoch": 2.3881217678733275, "grad_norm": 3.2758524417877197, "learning_rate": 1.0414e-06, "loss": 0.0434, "step": 44794 }, { "epoch": 2.3881750813029803, "grad_norm": 2.018054246902466, "learning_rate": 1.0412e-06, "loss": 0.0137, "step": 44795 }, { "epoch": 2.388228394732633, "grad_norm": 1.159124493598938, "learning_rate": 1.041e-06, "loss": 0.0194, "step": 44796 }, { "epoch": 2.3882817081622862, "grad_norm": 3.0573458671569824, "learning_rate": 1.0408e-06, "loss": 0.027, "step": 44797 }, { "epoch": 2.388335021591939, "grad_norm": 4.18773889541626, "learning_rate": 1.0406e-06, "loss": 0.0214, "step": 44798 }, { "epoch": 2.3883883350215918, "grad_norm": 3.592693328857422, "learning_rate": 1.0404e-06, "loss": 0.0283, "step": 44799 }, { "epoch": 2.388441648451245, "grad_norm": 3.9855432510375977, "learning_rate": 1.0402e-06, "loss": 0.0121, "step": 44800 }, { "epoch": 2.3884949618808977, "grad_norm": 3.2844269275665283, "learning_rate": 1.04e-06, "loss": 0.0169, "step": 44801 }, { "epoch": 2.388548275310551, "grad_norm": 0.8664588928222656, "learning_rate": 1.0398000000000002e-06, "loss": 0.003, "step": 44802 }, { "epoch": 2.3886015887402037, "grad_norm": 2.887800931930542, "learning_rate": 1.0396e-06, "loss": 0.0257, "step": 44803 }, { "epoch": 2.3886549021698564, "grad_norm": 2.5328426361083984, "learning_rate": 1.0394000000000002e-06, "loss": 0.0082, "step": 44804 }, { "epoch": 2.3887082155995096, "grad_norm": 1.909977912902832, "learning_rate": 1.0392000000000001e-06, "loss": 0.0112, "step": 44805 }, { "epoch": 2.3887615290291624, "grad_norm": 2.150390625, "learning_rate": 1.0390000000000001e-06, "loss": 0.0284, "step": 44806 }, { "epoch": 2.3888148424588156, "grad_norm": 1.932078242301941, "learning_rate": 1.0388000000000001e-06, "loss": 0.0114, "step": 44807 }, { "epoch": 2.3888681558884683, "grad_norm": 1.6666507720947266, "learning_rate": 1.0386e-06, "loss": 0.016, "step": 44808 }, { "epoch": 2.388921469318121, "grad_norm": 1.0472445487976074, "learning_rate": 1.0384e-06, "loss": 0.0225, "step": 44809 }, { "epoch": 2.3889747827477743, "grad_norm": 1.1690030097961426, "learning_rate": 1.0382e-06, "loss": 0.0046, "step": 44810 }, { "epoch": 2.389028096177427, "grad_norm": 0.6951078772544861, "learning_rate": 1.038e-06, "loss": 0.0031, "step": 44811 }, { "epoch": 2.38908140960708, "grad_norm": 2.97988224029541, "learning_rate": 1.0378e-06, "loss": 0.0227, "step": 44812 }, { "epoch": 2.389134723036733, "grad_norm": 0.8842719793319702, "learning_rate": 1.0376000000000002e-06, "loss": 0.0053, "step": 44813 }, { "epoch": 2.389188036466386, "grad_norm": 1.324401617050171, "learning_rate": 1.0374e-06, "loss": 0.0058, "step": 44814 }, { "epoch": 2.389241349896039, "grad_norm": 0.8297832012176514, "learning_rate": 1.0372000000000002e-06, "loss": 0.0047, "step": 44815 }, { "epoch": 2.3892946633256917, "grad_norm": 1.3105090856552124, "learning_rate": 1.0370000000000002e-06, "loss": 0.0166, "step": 44816 }, { "epoch": 2.3893479767553445, "grad_norm": 0.35291436314582825, "learning_rate": 1.0368e-06, "loss": 0.002, "step": 44817 }, { "epoch": 2.3894012901849977, "grad_norm": 1.4834511280059814, "learning_rate": 1.0366000000000002e-06, "loss": 0.0174, "step": 44818 }, { "epoch": 2.3894546036146505, "grad_norm": 0.5319483280181885, "learning_rate": 1.0364000000000001e-06, "loss": 0.004, "step": 44819 }, { "epoch": 2.3895079170443037, "grad_norm": 1.7842720746994019, "learning_rate": 1.0362000000000001e-06, "loss": 0.0102, "step": 44820 }, { "epoch": 2.3895612304739564, "grad_norm": 2.2303056716918945, "learning_rate": 1.0360000000000001e-06, "loss": 0.0106, "step": 44821 }, { "epoch": 2.389614543903609, "grad_norm": 4.029085159301758, "learning_rate": 1.0358e-06, "loss": 0.0092, "step": 44822 }, { "epoch": 2.3896678573332624, "grad_norm": 1.1158068180084229, "learning_rate": 1.0356e-06, "loss": 0.0195, "step": 44823 }, { "epoch": 2.389721170762915, "grad_norm": 1.7848354578018188, "learning_rate": 1.0354e-06, "loss": 0.0174, "step": 44824 }, { "epoch": 2.389774484192568, "grad_norm": 2.4693922996520996, "learning_rate": 1.0352e-06, "loss": 0.0155, "step": 44825 }, { "epoch": 2.389827797622221, "grad_norm": 0.2578273117542267, "learning_rate": 1.035e-06, "loss": 0.0014, "step": 44826 }, { "epoch": 2.389881111051874, "grad_norm": 1.821163296699524, "learning_rate": 1.0348e-06, "loss": 0.0064, "step": 44827 }, { "epoch": 2.389934424481527, "grad_norm": 1.1580265760421753, "learning_rate": 1.0346e-06, "loss": 0.0057, "step": 44828 }, { "epoch": 2.38998773791118, "grad_norm": 2.1236350536346436, "learning_rate": 1.0344000000000002e-06, "loss": 0.0305, "step": 44829 }, { "epoch": 2.390041051340833, "grad_norm": 2.277153491973877, "learning_rate": 1.0342e-06, "loss": 0.0319, "step": 44830 }, { "epoch": 2.3900943647704858, "grad_norm": 2.6110332012176514, "learning_rate": 1.0340000000000002e-06, "loss": 0.0132, "step": 44831 }, { "epoch": 2.3901476782001385, "grad_norm": 1.9636170864105225, "learning_rate": 1.0338000000000002e-06, "loss": 0.031, "step": 44832 }, { "epoch": 2.3902009916297917, "grad_norm": 3.4987049102783203, "learning_rate": 1.0336e-06, "loss": 0.041, "step": 44833 }, { "epoch": 2.3902543050594445, "grad_norm": 1.9308910369873047, "learning_rate": 1.0334000000000001e-06, "loss": 0.026, "step": 44834 }, { "epoch": 2.3903076184890972, "grad_norm": 2.1397366523742676, "learning_rate": 1.0332000000000001e-06, "loss": 0.0266, "step": 44835 }, { "epoch": 2.3903609319187504, "grad_norm": 1.3102864027023315, "learning_rate": 1.033e-06, "loss": 0.0168, "step": 44836 }, { "epoch": 2.390414245348403, "grad_norm": 0.6126887202262878, "learning_rate": 1.0328e-06, "loss": 0.0028, "step": 44837 }, { "epoch": 2.390467558778056, "grad_norm": 1.6589512825012207, "learning_rate": 1.0326e-06, "loss": 0.0312, "step": 44838 }, { "epoch": 2.390520872207709, "grad_norm": 2.131427764892578, "learning_rate": 1.0324e-06, "loss": 0.0191, "step": 44839 }, { "epoch": 2.390574185637362, "grad_norm": 0.5091785788536072, "learning_rate": 1.0322e-06, "loss": 0.0031, "step": 44840 }, { "epoch": 2.390627499067015, "grad_norm": 1.5058245658874512, "learning_rate": 1.032e-06, "loss": 0.0071, "step": 44841 }, { "epoch": 2.390680812496668, "grad_norm": 1.7176587581634521, "learning_rate": 1.0318e-06, "loss": 0.0206, "step": 44842 }, { "epoch": 2.390734125926321, "grad_norm": 2.124258518218994, "learning_rate": 1.0316e-06, "loss": 0.0097, "step": 44843 }, { "epoch": 2.390787439355974, "grad_norm": 1.5078731775283813, "learning_rate": 1.0314e-06, "loss": 0.0176, "step": 44844 }, { "epoch": 2.3908407527856266, "grad_norm": 4.072239875793457, "learning_rate": 1.0312000000000002e-06, "loss": 0.0249, "step": 44845 }, { "epoch": 2.39089406621528, "grad_norm": 2.274994134902954, "learning_rate": 1.031e-06, "loss": 0.0163, "step": 44846 }, { "epoch": 2.3909473796449325, "grad_norm": 2.1191134452819824, "learning_rate": 1.0308000000000002e-06, "loss": 0.0076, "step": 44847 }, { "epoch": 2.3910006930745853, "grad_norm": 0.5878374576568604, "learning_rate": 1.0306000000000001e-06, "loss": 0.0034, "step": 44848 }, { "epoch": 2.3910540065042385, "grad_norm": 1.2452811002731323, "learning_rate": 1.0304000000000001e-06, "loss": 0.0232, "step": 44849 }, { "epoch": 2.3911073199338913, "grad_norm": 1.7822785377502441, "learning_rate": 1.0302000000000001e-06, "loss": 0.0063, "step": 44850 }, { "epoch": 2.3911606333635445, "grad_norm": 0.12532174587249756, "learning_rate": 1.03e-06, "loss": 0.0014, "step": 44851 }, { "epoch": 2.391213946793197, "grad_norm": 1.0966547727584839, "learning_rate": 1.0298e-06, "loss": 0.0035, "step": 44852 }, { "epoch": 2.39126726022285, "grad_norm": 1.9517871141433716, "learning_rate": 1.0296e-06, "loss": 0.0071, "step": 44853 }, { "epoch": 2.391320573652503, "grad_norm": 1.5459414720535278, "learning_rate": 1.0294e-06, "loss": 0.0157, "step": 44854 }, { "epoch": 2.391373887082156, "grad_norm": 1.245949387550354, "learning_rate": 1.0292e-06, "loss": 0.0098, "step": 44855 }, { "epoch": 2.391427200511809, "grad_norm": 0.7913864850997925, "learning_rate": 1.0290000000000002e-06, "loss": 0.0042, "step": 44856 }, { "epoch": 2.391480513941462, "grad_norm": 2.469839096069336, "learning_rate": 1.0288e-06, "loss": 0.0413, "step": 44857 }, { "epoch": 2.3915338273711146, "grad_norm": 1.7636772394180298, "learning_rate": 1.0286e-06, "loss": 0.037, "step": 44858 }, { "epoch": 2.391587140800768, "grad_norm": 1.3717402219772339, "learning_rate": 1.0284000000000002e-06, "loss": 0.0128, "step": 44859 }, { "epoch": 2.3916404542304206, "grad_norm": 4.857411861419678, "learning_rate": 1.0282e-06, "loss": 0.0204, "step": 44860 }, { "epoch": 2.3916937676600734, "grad_norm": 3.643449068069458, "learning_rate": 1.0280000000000002e-06, "loss": 0.0231, "step": 44861 }, { "epoch": 2.3917470810897266, "grad_norm": 2.571531057357788, "learning_rate": 1.0278000000000001e-06, "loss": 0.0164, "step": 44862 }, { "epoch": 2.3918003945193793, "grad_norm": 2.691174030303955, "learning_rate": 1.0276000000000001e-06, "loss": 0.0305, "step": 44863 }, { "epoch": 2.3918537079490325, "grad_norm": 0.5188009738922119, "learning_rate": 1.0274000000000001e-06, "loss": 0.0033, "step": 44864 }, { "epoch": 2.3919070213786853, "grad_norm": 2.228001356124878, "learning_rate": 1.0272e-06, "loss": 0.028, "step": 44865 }, { "epoch": 2.391960334808338, "grad_norm": 3.939460039138794, "learning_rate": 1.027e-06, "loss": 0.0479, "step": 44866 }, { "epoch": 2.3920136482379912, "grad_norm": 1.9017770290374756, "learning_rate": 1.0268e-06, "loss": 0.0057, "step": 44867 }, { "epoch": 2.392066961667644, "grad_norm": 0.3747263550758362, "learning_rate": 1.0266e-06, "loss": 0.0025, "step": 44868 }, { "epoch": 2.392120275097297, "grad_norm": 0.7847859859466553, "learning_rate": 1.0264e-06, "loss": 0.0042, "step": 44869 }, { "epoch": 2.39217358852695, "grad_norm": 1.4188364744186401, "learning_rate": 1.0262e-06, "loss": 0.0112, "step": 44870 }, { "epoch": 2.3922269019566027, "grad_norm": 2.477945566177368, "learning_rate": 1.026e-06, "loss": 0.0113, "step": 44871 }, { "epoch": 2.392280215386256, "grad_norm": 3.576448440551758, "learning_rate": 1.0258000000000002e-06, "loss": 0.017, "step": 44872 }, { "epoch": 2.3923335288159087, "grad_norm": 4.254547119140625, "learning_rate": 1.0256e-06, "loss": 0.0251, "step": 44873 }, { "epoch": 2.3923868422455614, "grad_norm": 4.013380527496338, "learning_rate": 1.0254000000000002e-06, "loss": 0.0359, "step": 44874 }, { "epoch": 2.3924401556752146, "grad_norm": 2.2990763187408447, "learning_rate": 1.0252000000000002e-06, "loss": 0.0099, "step": 44875 }, { "epoch": 2.3924934691048674, "grad_norm": 2.434044122695923, "learning_rate": 1.025e-06, "loss": 0.0072, "step": 44876 }, { "epoch": 2.3925467825345206, "grad_norm": 4.249757289886475, "learning_rate": 1.0248000000000001e-06, "loss": 0.0216, "step": 44877 }, { "epoch": 2.3926000959641733, "grad_norm": 1.1160519123077393, "learning_rate": 1.0246000000000001e-06, "loss": 0.0204, "step": 44878 }, { "epoch": 2.3926534093938265, "grad_norm": 0.9897511601448059, "learning_rate": 1.0244e-06, "loss": 0.0191, "step": 44879 }, { "epoch": 2.3927067228234793, "grad_norm": 1.0099811553955078, "learning_rate": 1.0242e-06, "loss": 0.0162, "step": 44880 }, { "epoch": 2.392760036253132, "grad_norm": 2.4635233879089355, "learning_rate": 1.024e-06, "loss": 0.0168, "step": 44881 }, { "epoch": 2.3928133496827853, "grad_norm": 1.435030460357666, "learning_rate": 1.0238e-06, "loss": 0.0072, "step": 44882 }, { "epoch": 2.392866663112438, "grad_norm": 2.6509108543395996, "learning_rate": 1.0236000000000002e-06, "loss": 0.0193, "step": 44883 }, { "epoch": 2.3929199765420908, "grad_norm": 0.9343199133872986, "learning_rate": 1.0234e-06, "loss": 0.0241, "step": 44884 }, { "epoch": 2.392973289971744, "grad_norm": 1.508347511291504, "learning_rate": 1.0232e-06, "loss": 0.0192, "step": 44885 }, { "epoch": 2.3930266034013967, "grad_norm": 1.520639419555664, "learning_rate": 1.0230000000000002e-06, "loss": 0.0223, "step": 44886 }, { "epoch": 2.39307991683105, "grad_norm": 1.1979116201400757, "learning_rate": 1.0228e-06, "loss": 0.005, "step": 44887 }, { "epoch": 2.3931332302607027, "grad_norm": 3.660590887069702, "learning_rate": 1.0226000000000002e-06, "loss": 0.0533, "step": 44888 }, { "epoch": 2.3931865436903554, "grad_norm": 1.556870460510254, "learning_rate": 1.0224000000000002e-06, "loss": 0.039, "step": 44889 }, { "epoch": 2.3932398571200086, "grad_norm": 2.3938815593719482, "learning_rate": 1.0222000000000001e-06, "loss": 0.0207, "step": 44890 }, { "epoch": 2.3932931705496614, "grad_norm": 2.2623817920684814, "learning_rate": 1.0220000000000001e-06, "loss": 0.019, "step": 44891 }, { "epoch": 2.3933464839793146, "grad_norm": 1.586559772491455, "learning_rate": 1.0218000000000001e-06, "loss": 0.0127, "step": 44892 }, { "epoch": 2.3933997974089674, "grad_norm": 3.191725492477417, "learning_rate": 1.0216e-06, "loss": 0.0115, "step": 44893 }, { "epoch": 2.39345311083862, "grad_norm": 1.8837577104568481, "learning_rate": 1.0214e-06, "loss": 0.0194, "step": 44894 }, { "epoch": 2.3935064242682733, "grad_norm": 6.872825622558594, "learning_rate": 1.0212e-06, "loss": 0.0406, "step": 44895 }, { "epoch": 2.393559737697926, "grad_norm": 1.1372605562210083, "learning_rate": 1.021e-06, "loss": 0.0064, "step": 44896 }, { "epoch": 2.393613051127579, "grad_norm": 1.0498870611190796, "learning_rate": 1.0208e-06, "loss": 0.0067, "step": 44897 }, { "epoch": 2.393666364557232, "grad_norm": 3.68021559715271, "learning_rate": 1.0206e-06, "loss": 0.0224, "step": 44898 }, { "epoch": 2.393719677986885, "grad_norm": 2.001387119293213, "learning_rate": 1.0204000000000002e-06, "loss": 0.0245, "step": 44899 }, { "epoch": 2.393772991416538, "grad_norm": 0.20605160295963287, "learning_rate": 1.0202e-06, "loss": 0.0016, "step": 44900 }, { "epoch": 2.3938263048461907, "grad_norm": 0.5025820732116699, "learning_rate": 1.02e-06, "loss": 0.0037, "step": 44901 }, { "epoch": 2.3938796182758435, "grad_norm": 3.6150412559509277, "learning_rate": 1.0198000000000002e-06, "loss": 0.0418, "step": 44902 }, { "epoch": 2.3939329317054967, "grad_norm": 0.8854438662528992, "learning_rate": 1.0196e-06, "loss": 0.0048, "step": 44903 }, { "epoch": 2.3939862451351495, "grad_norm": 1.2050185203552246, "learning_rate": 1.0194000000000001e-06, "loss": 0.0093, "step": 44904 }, { "epoch": 2.3940395585648027, "grad_norm": 1.2491470575332642, "learning_rate": 1.0192000000000001e-06, "loss": 0.0034, "step": 44905 }, { "epoch": 2.3940928719944554, "grad_norm": 1.51555335521698, "learning_rate": 1.0190000000000001e-06, "loss": 0.0122, "step": 44906 }, { "epoch": 2.394146185424108, "grad_norm": 1.4044301509857178, "learning_rate": 1.0188e-06, "loss": 0.0031, "step": 44907 }, { "epoch": 2.3941994988537614, "grad_norm": 3.1382429599761963, "learning_rate": 1.0186e-06, "loss": 0.039, "step": 44908 }, { "epoch": 2.394252812283414, "grad_norm": 1.5659379959106445, "learning_rate": 1.0184e-06, "loss": 0.0202, "step": 44909 }, { "epoch": 2.394306125713067, "grad_norm": 4.559643268585205, "learning_rate": 1.0182e-06, "loss": 0.0224, "step": 44910 }, { "epoch": 2.39435943914272, "grad_norm": 0.6266883611679077, "learning_rate": 1.018e-06, "loss": 0.003, "step": 44911 }, { "epoch": 2.394412752572373, "grad_norm": 1.707719326019287, "learning_rate": 1.0178e-06, "loss": 0.0045, "step": 44912 }, { "epoch": 2.394466066002026, "grad_norm": 2.604090452194214, "learning_rate": 1.0176000000000002e-06, "loss": 0.0239, "step": 44913 }, { "epoch": 2.394519379431679, "grad_norm": 2.1454215049743652, "learning_rate": 1.0174e-06, "loss": 0.0098, "step": 44914 }, { "epoch": 2.394572692861332, "grad_norm": 0.46400606632232666, "learning_rate": 1.0172000000000002e-06, "loss": 0.0034, "step": 44915 }, { "epoch": 2.3946260062909848, "grad_norm": 0.8225280046463013, "learning_rate": 1.0170000000000002e-06, "loss": 0.0025, "step": 44916 }, { "epoch": 2.3946793197206375, "grad_norm": 1.8924275636672974, "learning_rate": 1.0168000000000002e-06, "loss": 0.0121, "step": 44917 }, { "epoch": 2.3947326331502907, "grad_norm": 2.4412689208984375, "learning_rate": 1.0166000000000001e-06, "loss": 0.0076, "step": 44918 }, { "epoch": 2.3947859465799435, "grad_norm": 2.0848746299743652, "learning_rate": 1.0164000000000001e-06, "loss": 0.0157, "step": 44919 }, { "epoch": 2.3948392600095962, "grad_norm": 1.3953595161437988, "learning_rate": 1.0162000000000001e-06, "loss": 0.0064, "step": 44920 }, { "epoch": 2.3948925734392494, "grad_norm": 1.7516083717346191, "learning_rate": 1.016e-06, "loss": 0.026, "step": 44921 }, { "epoch": 2.394945886868902, "grad_norm": 2.537445545196533, "learning_rate": 1.0158e-06, "loss": 0.008, "step": 44922 }, { "epoch": 2.3949992002985554, "grad_norm": 4.248513698577881, "learning_rate": 1.0156e-06, "loss": 0.0605, "step": 44923 }, { "epoch": 2.395052513728208, "grad_norm": 0.5172839164733887, "learning_rate": 1.0154e-06, "loss": 0.0036, "step": 44924 }, { "epoch": 2.395105827157861, "grad_norm": 1.262786865234375, "learning_rate": 1.0152e-06, "loss": 0.0075, "step": 44925 }, { "epoch": 2.395159140587514, "grad_norm": 1.6988584995269775, "learning_rate": 1.0150000000000002e-06, "loss": 0.0202, "step": 44926 }, { "epoch": 2.395212454017167, "grad_norm": 1.2027523517608643, "learning_rate": 1.0148e-06, "loss": 0.0121, "step": 44927 }, { "epoch": 2.39526576744682, "grad_norm": 1.3864721059799194, "learning_rate": 1.0146e-06, "loss": 0.0145, "step": 44928 }, { "epoch": 2.395319080876473, "grad_norm": 2.7586655616760254, "learning_rate": 1.0144000000000002e-06, "loss": 0.0407, "step": 44929 }, { "epoch": 2.3953723943061256, "grad_norm": 0.837934136390686, "learning_rate": 1.0142e-06, "loss": 0.0036, "step": 44930 }, { "epoch": 2.395425707735779, "grad_norm": 0.5767902731895447, "learning_rate": 1.0140000000000002e-06, "loss": 0.0032, "step": 44931 }, { "epoch": 2.3954790211654315, "grad_norm": 0.27032989263534546, "learning_rate": 1.0138000000000001e-06, "loss": 0.0025, "step": 44932 }, { "epoch": 2.3955323345950843, "grad_norm": 2.5479884147644043, "learning_rate": 1.0136000000000001e-06, "loss": 0.0426, "step": 44933 }, { "epoch": 2.3955856480247375, "grad_norm": 0.6222538948059082, "learning_rate": 1.0134000000000001e-06, "loss": 0.004, "step": 44934 }, { "epoch": 2.3956389614543903, "grad_norm": 0.46575549244880676, "learning_rate": 1.0132e-06, "loss": 0.0024, "step": 44935 }, { "epoch": 2.3956922748840435, "grad_norm": 1.442706823348999, "learning_rate": 1.013e-06, "loss": 0.0162, "step": 44936 }, { "epoch": 2.395745588313696, "grad_norm": 0.9338293671607971, "learning_rate": 1.0128e-06, "loss": 0.0039, "step": 44937 }, { "epoch": 2.395798901743349, "grad_norm": 0.9214280247688293, "learning_rate": 1.0126e-06, "loss": 0.0051, "step": 44938 }, { "epoch": 2.395852215173002, "grad_norm": 3.404010772705078, "learning_rate": 1.0124e-06, "loss": 0.0498, "step": 44939 }, { "epoch": 2.395905528602655, "grad_norm": 0.47127771377563477, "learning_rate": 1.0122e-06, "loss": 0.0038, "step": 44940 }, { "epoch": 2.395958842032308, "grad_norm": 2.2879514694213867, "learning_rate": 1.012e-06, "loss": 0.01, "step": 44941 }, { "epoch": 2.396012155461961, "grad_norm": 1.9042894840240479, "learning_rate": 1.0118000000000002e-06, "loss": 0.0243, "step": 44942 }, { "epoch": 2.3960654688916136, "grad_norm": 2.3157098293304443, "learning_rate": 1.0116e-06, "loss": 0.0309, "step": 44943 }, { "epoch": 2.396118782321267, "grad_norm": 3.344525098800659, "learning_rate": 1.0114e-06, "loss": 0.0321, "step": 44944 }, { "epoch": 2.3961720957509196, "grad_norm": 1.7504175901412964, "learning_rate": 1.0112000000000002e-06, "loss": 0.0263, "step": 44945 }, { "epoch": 2.3962254091805724, "grad_norm": 0.7652817964553833, "learning_rate": 1.011e-06, "loss": 0.0048, "step": 44946 }, { "epoch": 2.3962787226102256, "grad_norm": 0.5313849449157715, "learning_rate": 1.0108000000000001e-06, "loss": 0.0033, "step": 44947 }, { "epoch": 2.3963320360398783, "grad_norm": 0.8290113210678101, "learning_rate": 1.0106000000000001e-06, "loss": 0.0128, "step": 44948 }, { "epoch": 2.3963853494695315, "grad_norm": 2.161625623703003, "learning_rate": 1.0104e-06, "loss": 0.0125, "step": 44949 }, { "epoch": 2.3964386628991843, "grad_norm": 0.400226891040802, "learning_rate": 1.0102e-06, "loss": 0.0031, "step": 44950 }, { "epoch": 2.3964919763288375, "grad_norm": 2.4446518421173096, "learning_rate": 1.01e-06, "loss": 0.0222, "step": 44951 }, { "epoch": 2.3965452897584902, "grad_norm": 1.1581478118896484, "learning_rate": 1.0098e-06, "loss": 0.0073, "step": 44952 }, { "epoch": 2.396598603188143, "grad_norm": 1.9155733585357666, "learning_rate": 1.0096e-06, "loss": 0.0218, "step": 44953 }, { "epoch": 2.396651916617796, "grad_norm": 1.4690563678741455, "learning_rate": 1.0094e-06, "loss": 0.0115, "step": 44954 }, { "epoch": 2.396705230047449, "grad_norm": 0.4846719801425934, "learning_rate": 1.0092e-06, "loss": 0.0029, "step": 44955 }, { "epoch": 2.3967585434771017, "grad_norm": 3.184445381164551, "learning_rate": 1.0090000000000002e-06, "loss": 0.0164, "step": 44956 }, { "epoch": 2.396811856906755, "grad_norm": 2.1011645793914795, "learning_rate": 1.0088e-06, "loss": 0.0298, "step": 44957 }, { "epoch": 2.3968651703364077, "grad_norm": 0.8421128392219543, "learning_rate": 1.0086000000000002e-06, "loss": 0.0047, "step": 44958 }, { "epoch": 2.3969184837660604, "grad_norm": 2.6363158226013184, "learning_rate": 1.0084000000000002e-06, "loss": 0.0266, "step": 44959 }, { "epoch": 2.3969717971957136, "grad_norm": 0.5556217432022095, "learning_rate": 1.0082000000000001e-06, "loss": 0.0029, "step": 44960 }, { "epoch": 2.3970251106253664, "grad_norm": 3.2671470642089844, "learning_rate": 1.0080000000000001e-06, "loss": 0.0236, "step": 44961 }, { "epoch": 2.3970784240550196, "grad_norm": 2.2056806087493896, "learning_rate": 1.0078000000000001e-06, "loss": 0.0277, "step": 44962 }, { "epoch": 2.3971317374846723, "grad_norm": 1.1006938219070435, "learning_rate": 1.0076e-06, "loss": 0.0115, "step": 44963 }, { "epoch": 2.3971850509143255, "grad_norm": 0.3130171000957489, "learning_rate": 1.0074e-06, "loss": 0.0032, "step": 44964 }, { "epoch": 2.3972383643439783, "grad_norm": 1.0564297437667847, "learning_rate": 1.0072e-06, "loss": 0.0035, "step": 44965 }, { "epoch": 2.397291677773631, "grad_norm": 1.177966833114624, "learning_rate": 1.007e-06, "loss": 0.0052, "step": 44966 }, { "epoch": 2.3973449912032843, "grad_norm": 1.6668781042099, "learning_rate": 1.0068e-06, "loss": 0.0115, "step": 44967 }, { "epoch": 2.397398304632937, "grad_norm": 2.510197639465332, "learning_rate": 1.0066e-06, "loss": 0.0144, "step": 44968 }, { "epoch": 2.3974516180625898, "grad_norm": 2.9654290676116943, "learning_rate": 1.0064e-06, "loss": 0.0159, "step": 44969 }, { "epoch": 2.397504931492243, "grad_norm": 0.9421195983886719, "learning_rate": 1.0062e-06, "loss": 0.0048, "step": 44970 }, { "epoch": 2.3975582449218957, "grad_norm": 0.75399249792099, "learning_rate": 1.006e-06, "loss": 0.0044, "step": 44971 }, { "epoch": 2.397611558351549, "grad_norm": 1.4943000078201294, "learning_rate": 1.0058000000000002e-06, "loss": 0.0232, "step": 44972 }, { "epoch": 2.3976648717812017, "grad_norm": 0.7515789866447449, "learning_rate": 1.0056e-06, "loss": 0.0038, "step": 44973 }, { "epoch": 2.3977181852108544, "grad_norm": 0.3420051038265228, "learning_rate": 1.0054000000000001e-06, "loss": 0.0028, "step": 44974 }, { "epoch": 2.3977714986405076, "grad_norm": 0.5419795513153076, "learning_rate": 1.0052000000000001e-06, "loss": 0.003, "step": 44975 }, { "epoch": 2.3978248120701604, "grad_norm": 2.6376397609710693, "learning_rate": 1.0050000000000001e-06, "loss": 0.0184, "step": 44976 }, { "epoch": 2.3978781254998136, "grad_norm": 0.8847474455833435, "learning_rate": 1.0048e-06, "loss": 0.0186, "step": 44977 }, { "epoch": 2.3979314389294664, "grad_norm": 1.721327543258667, "learning_rate": 1.0046e-06, "loss": 0.0074, "step": 44978 }, { "epoch": 2.397984752359119, "grad_norm": 0.6782434582710266, "learning_rate": 1.0044e-06, "loss": 0.0033, "step": 44979 }, { "epoch": 2.3980380657887723, "grad_norm": 1.4513789415359497, "learning_rate": 1.0042e-06, "loss": 0.0153, "step": 44980 }, { "epoch": 2.398091379218425, "grad_norm": 1.7436070442199707, "learning_rate": 1.004e-06, "loss": 0.0224, "step": 44981 }, { "epoch": 2.398144692648078, "grad_norm": 1.118526577949524, "learning_rate": 1.0038e-06, "loss": 0.0035, "step": 44982 }, { "epoch": 2.398198006077731, "grad_norm": 2.39663028717041, "learning_rate": 1.0036000000000002e-06, "loss": 0.0138, "step": 44983 }, { "epoch": 2.398251319507384, "grad_norm": 1.5101866722106934, "learning_rate": 1.0034e-06, "loss": 0.0392, "step": 44984 }, { "epoch": 2.398304632937037, "grad_norm": 0.6065256595611572, "learning_rate": 1.0032000000000002e-06, "loss": 0.0038, "step": 44985 }, { "epoch": 2.3983579463666898, "grad_norm": 1.794878363609314, "learning_rate": 1.0030000000000002e-06, "loss": 0.0135, "step": 44986 }, { "epoch": 2.3984112597963425, "grad_norm": 0.6363685727119446, "learning_rate": 1.0028e-06, "loss": 0.0037, "step": 44987 }, { "epoch": 2.3984645732259957, "grad_norm": 1.2883576154708862, "learning_rate": 1.0026000000000002e-06, "loss": 0.0049, "step": 44988 }, { "epoch": 2.3985178866556485, "grad_norm": 2.6028778553009033, "learning_rate": 1.0024000000000001e-06, "loss": 0.0222, "step": 44989 }, { "epoch": 2.3985712000853017, "grad_norm": 1.3709927797317505, "learning_rate": 1.0022000000000001e-06, "loss": 0.0055, "step": 44990 }, { "epoch": 2.3986245135149544, "grad_norm": 0.7115373611450195, "learning_rate": 1.002e-06, "loss": 0.0049, "step": 44991 }, { "epoch": 2.398677826944607, "grad_norm": 1.1186684370040894, "learning_rate": 1.0018e-06, "loss": 0.0085, "step": 44992 }, { "epoch": 2.3987311403742604, "grad_norm": 2.3080053329467773, "learning_rate": 1.0016e-06, "loss": 0.0174, "step": 44993 }, { "epoch": 2.398784453803913, "grad_norm": 0.8706672191619873, "learning_rate": 1.0014e-06, "loss": 0.0066, "step": 44994 }, { "epoch": 2.398837767233566, "grad_norm": 3.70609188079834, "learning_rate": 1.0012e-06, "loss": 0.0279, "step": 44995 }, { "epoch": 2.398891080663219, "grad_norm": 1.6773115396499634, "learning_rate": 1.001e-06, "loss": 0.0061, "step": 44996 }, { "epoch": 2.398944394092872, "grad_norm": 2.7245795726776123, "learning_rate": 1.0008e-06, "loss": 0.0219, "step": 44997 }, { "epoch": 2.398997707522525, "grad_norm": 3.460249423980713, "learning_rate": 1.0006e-06, "loss": 0.0136, "step": 44998 }, { "epoch": 2.399051020952178, "grad_norm": 1.1060285568237305, "learning_rate": 1.0004000000000002e-06, "loss": 0.0242, "step": 44999 }, { "epoch": 2.399104334381831, "grad_norm": 2.4252400398254395, "learning_rate": 1.0002e-06, "loss": 0.0266, "step": 45000 }, { "epoch": 2.3991576478114838, "grad_norm": 2.075594425201416, "learning_rate": 1.0000000000000002e-06, "loss": 0.0162, "step": 45001 }, { "epoch": 2.3992109612411365, "grad_norm": 1.4559217691421509, "learning_rate": 9.998000000000002e-07, "loss": 0.0099, "step": 45002 }, { "epoch": 2.3992642746707897, "grad_norm": 2.8708157539367676, "learning_rate": 9.996e-07, "loss": 0.0265, "step": 45003 }, { "epoch": 2.3993175881004425, "grad_norm": 1.8151592016220093, "learning_rate": 9.994000000000001e-07, "loss": 0.0229, "step": 45004 }, { "epoch": 2.3993709015300952, "grad_norm": 2.507796049118042, "learning_rate": 9.992000000000001e-07, "loss": 0.0362, "step": 45005 }, { "epoch": 2.3994242149597484, "grad_norm": 3.535911798477173, "learning_rate": 9.99e-07, "loss": 0.0202, "step": 45006 }, { "epoch": 2.399477528389401, "grad_norm": 1.3026888370513916, "learning_rate": 9.988e-07, "loss": 0.0157, "step": 45007 }, { "epoch": 2.3995308418190544, "grad_norm": 0.4282212555408478, "learning_rate": 9.986e-07, "loss": 0.0036, "step": 45008 }, { "epoch": 2.399584155248707, "grad_norm": 1.5020133256912231, "learning_rate": 9.984e-07, "loss": 0.0199, "step": 45009 }, { "epoch": 2.39963746867836, "grad_norm": 0.33901649713516235, "learning_rate": 9.982000000000002e-07, "loss": 0.0025, "step": 45010 }, { "epoch": 2.399690782108013, "grad_norm": 1.6031665802001953, "learning_rate": 9.98e-07, "loss": 0.029, "step": 45011 }, { "epoch": 2.399744095537666, "grad_norm": 0.8080718517303467, "learning_rate": 9.978e-07, "loss": 0.0054, "step": 45012 }, { "epoch": 2.399797408967319, "grad_norm": 1.7853877544403076, "learning_rate": 9.976000000000002e-07, "loss": 0.018, "step": 45013 }, { "epoch": 2.399850722396972, "grad_norm": 5.5255560874938965, "learning_rate": 9.974e-07, "loss": 0.0185, "step": 45014 }, { "epoch": 2.3999040358266246, "grad_norm": 1.1490529775619507, "learning_rate": 9.972000000000002e-07, "loss": 0.0117, "step": 45015 }, { "epoch": 2.399957349256278, "grad_norm": 2.353980779647827, "learning_rate": 9.970000000000002e-07, "loss": 0.0087, "step": 45016 }, { "epoch": 2.4000106626859306, "grad_norm": 2.381059408187866, "learning_rate": 9.968000000000001e-07, "loss": 0.0255, "step": 45017 }, { "epoch": 2.4000639761155833, "grad_norm": 1.559401273727417, "learning_rate": 9.966000000000001e-07, "loss": 0.0257, "step": 45018 }, { "epoch": 2.4001172895452365, "grad_norm": 0.5684689879417419, "learning_rate": 9.964000000000001e-07, "loss": 0.0034, "step": 45019 }, { "epoch": 2.4001706029748893, "grad_norm": 2.1923201084136963, "learning_rate": 9.962e-07, "loss": 0.015, "step": 45020 }, { "epoch": 2.4002239164045425, "grad_norm": 1.5722894668579102, "learning_rate": 9.96e-07, "loss": 0.0065, "step": 45021 }, { "epoch": 2.4002772298341952, "grad_norm": 3.1535661220550537, "learning_rate": 9.958e-07, "loss": 0.019, "step": 45022 }, { "epoch": 2.400330543263848, "grad_norm": 0.7938998937606812, "learning_rate": 9.956e-07, "loss": 0.0138, "step": 45023 }, { "epoch": 2.400383856693501, "grad_norm": 0.7155497670173645, "learning_rate": 9.954e-07, "loss": 0.0045, "step": 45024 }, { "epoch": 2.400437170123154, "grad_norm": 2.4365391731262207, "learning_rate": 9.952e-07, "loss": 0.0423, "step": 45025 }, { "epoch": 2.400490483552807, "grad_norm": 0.953214704990387, "learning_rate": 9.950000000000002e-07, "loss": 0.006, "step": 45026 }, { "epoch": 2.40054379698246, "grad_norm": 1.199417233467102, "learning_rate": 9.948e-07, "loss": 0.003, "step": 45027 }, { "epoch": 2.4005971104121127, "grad_norm": 0.5629693865776062, "learning_rate": 9.946000000000002e-07, "loss": 0.0035, "step": 45028 }, { "epoch": 2.400650423841766, "grad_norm": 0.5201163291931152, "learning_rate": 9.944000000000002e-07, "loss": 0.0044, "step": 45029 }, { "epoch": 2.4007037372714186, "grad_norm": 2.0633461475372314, "learning_rate": 9.942e-07, "loss": 0.012, "step": 45030 }, { "epoch": 2.4007570507010714, "grad_norm": 1.9608478546142578, "learning_rate": 9.940000000000001e-07, "loss": 0.0125, "step": 45031 }, { "epoch": 2.4008103641307246, "grad_norm": 1.787036657333374, "learning_rate": 9.938000000000001e-07, "loss": 0.0191, "step": 45032 }, { "epoch": 2.4008636775603773, "grad_norm": 1.1324647665023804, "learning_rate": 9.936000000000001e-07, "loss": 0.0188, "step": 45033 }, { "epoch": 2.4009169909900305, "grad_norm": 3.4008634090423584, "learning_rate": 9.934e-07, "loss": 0.012, "step": 45034 }, { "epoch": 2.4009703044196833, "grad_norm": 0.989151656627655, "learning_rate": 9.932e-07, "loss": 0.004, "step": 45035 }, { "epoch": 2.4010236178493365, "grad_norm": 2.4852242469787598, "learning_rate": 9.93e-07, "loss": 0.0349, "step": 45036 }, { "epoch": 2.4010769312789892, "grad_norm": 2.0743112564086914, "learning_rate": 9.928e-07, "loss": 0.0297, "step": 45037 }, { "epoch": 2.401130244708642, "grad_norm": 2.3648128509521484, "learning_rate": 9.926e-07, "loss": 0.0181, "step": 45038 }, { "epoch": 2.401183558138295, "grad_norm": 0.8056357502937317, "learning_rate": 9.924e-07, "loss": 0.0036, "step": 45039 }, { "epoch": 2.401236871567948, "grad_norm": 0.48636481165885925, "learning_rate": 9.922e-07, "loss": 0.002, "step": 45040 }, { "epoch": 2.4012901849976007, "grad_norm": 1.043413758277893, "learning_rate": 9.92e-07, "loss": 0.0132, "step": 45041 }, { "epoch": 2.401343498427254, "grad_norm": 3.810075044631958, "learning_rate": 9.918000000000002e-07, "loss": 0.0274, "step": 45042 }, { "epoch": 2.4013968118569067, "grad_norm": 3.104173183441162, "learning_rate": 9.916e-07, "loss": 0.0355, "step": 45043 }, { "epoch": 2.40145012528656, "grad_norm": 2.903606414794922, "learning_rate": 9.914000000000002e-07, "loss": 0.0241, "step": 45044 }, { "epoch": 2.4015034387162126, "grad_norm": 1.5444989204406738, "learning_rate": 9.912000000000001e-07, "loss": 0.0075, "step": 45045 }, { "epoch": 2.4015567521458654, "grad_norm": 0.5237071514129639, "learning_rate": 9.91e-07, "loss": 0.0036, "step": 45046 }, { "epoch": 2.4016100655755186, "grad_norm": 0.33674541115760803, "learning_rate": 9.908000000000001e-07, "loss": 0.0023, "step": 45047 }, { "epoch": 2.4016633790051714, "grad_norm": 1.2533082962036133, "learning_rate": 9.906e-07, "loss": 0.0134, "step": 45048 }, { "epoch": 2.4017166924348246, "grad_norm": 3.5003910064697266, "learning_rate": 9.904e-07, "loss": 0.016, "step": 45049 }, { "epoch": 2.4017700058644773, "grad_norm": 1.0111335515975952, "learning_rate": 9.902e-07, "loss": 0.0063, "step": 45050 }, { "epoch": 2.40182331929413, "grad_norm": 3.093738079071045, "learning_rate": 9.9e-07, "loss": 0.0268, "step": 45051 }, { "epoch": 2.4018766327237833, "grad_norm": 1.4029899835586548, "learning_rate": 9.898e-07, "loss": 0.0068, "step": 45052 }, { "epoch": 2.401929946153436, "grad_norm": 2.5239648818969727, "learning_rate": 9.896000000000002e-07, "loss": 0.0167, "step": 45053 }, { "epoch": 2.401983259583089, "grad_norm": 1.917294979095459, "learning_rate": 9.894e-07, "loss": 0.0134, "step": 45054 }, { "epoch": 2.402036573012742, "grad_norm": 1.9000015258789062, "learning_rate": 9.892e-07, "loss": 0.012, "step": 45055 }, { "epoch": 2.4020898864423947, "grad_norm": 0.777618944644928, "learning_rate": 9.890000000000002e-07, "loss": 0.0129, "step": 45056 }, { "epoch": 2.402143199872048, "grad_norm": 1.9715746641159058, "learning_rate": 9.888e-07, "loss": 0.0086, "step": 45057 }, { "epoch": 2.4021965133017007, "grad_norm": 2.109785318374634, "learning_rate": 9.886000000000002e-07, "loss": 0.009, "step": 45058 }, { "epoch": 2.4022498267313535, "grad_norm": 1.2580246925354004, "learning_rate": 9.884000000000001e-07, "loss": 0.0182, "step": 45059 }, { "epoch": 2.4023031401610067, "grad_norm": 1.8601268529891968, "learning_rate": 9.882000000000001e-07, "loss": 0.0183, "step": 45060 }, { "epoch": 2.4023564535906594, "grad_norm": 1.1762962341308594, "learning_rate": 9.880000000000001e-07, "loss": 0.0039, "step": 45061 }, { "epoch": 2.4024097670203126, "grad_norm": 3.0656356811523438, "learning_rate": 9.878e-07, "loss": 0.0192, "step": 45062 }, { "epoch": 2.4024630804499654, "grad_norm": 1.233252763748169, "learning_rate": 9.876e-07, "loss": 0.02, "step": 45063 }, { "epoch": 2.402516393879618, "grad_norm": 2.5625035762786865, "learning_rate": 9.874e-07, "loss": 0.0093, "step": 45064 }, { "epoch": 2.4025697073092713, "grad_norm": 2.4032137393951416, "learning_rate": 9.872e-07, "loss": 0.008, "step": 45065 }, { "epoch": 2.402623020738924, "grad_norm": 3.791506052017212, "learning_rate": 9.87e-07, "loss": 0.0172, "step": 45066 }, { "epoch": 2.402676334168577, "grad_norm": 1.259185552597046, "learning_rate": 9.868e-07, "loss": 0.0044, "step": 45067 }, { "epoch": 2.40272964759823, "grad_norm": 2.7615413665771484, "learning_rate": 9.866e-07, "loss": 0.0191, "step": 45068 }, { "epoch": 2.402782961027883, "grad_norm": 0.6048566102981567, "learning_rate": 9.864000000000002e-07, "loss": 0.0045, "step": 45069 }, { "epoch": 2.402836274457536, "grad_norm": 2.3898348808288574, "learning_rate": 9.862e-07, "loss": 0.0152, "step": 45070 }, { "epoch": 2.4028895878871888, "grad_norm": 0.8581053018569946, "learning_rate": 9.86e-07, "loss": 0.0043, "step": 45071 }, { "epoch": 2.402942901316842, "grad_norm": 1.8460973501205444, "learning_rate": 9.858000000000002e-07, "loss": 0.0045, "step": 45072 }, { "epoch": 2.4029962147464947, "grad_norm": 2.1729485988616943, "learning_rate": 9.856e-07, "loss": 0.0205, "step": 45073 }, { "epoch": 2.4030495281761475, "grad_norm": 4.1522345542907715, "learning_rate": 9.854000000000001e-07, "loss": 0.0091, "step": 45074 }, { "epoch": 2.4031028416058007, "grad_norm": 2.3215339183807373, "learning_rate": 9.852000000000001e-07, "loss": 0.0097, "step": 45075 }, { "epoch": 2.4031561550354534, "grad_norm": 1.8626761436462402, "learning_rate": 9.85e-07, "loss": 0.0299, "step": 45076 }, { "epoch": 2.403209468465106, "grad_norm": 6.08624792098999, "learning_rate": 9.848e-07, "loss": 0.0229, "step": 45077 }, { "epoch": 2.4032627818947594, "grad_norm": 2.1833341121673584, "learning_rate": 9.846e-07, "loss": 0.0087, "step": 45078 }, { "epoch": 2.403316095324412, "grad_norm": 3.2385783195495605, "learning_rate": 9.844e-07, "loss": 0.0125, "step": 45079 }, { "epoch": 2.403369408754065, "grad_norm": 0.964889407157898, "learning_rate": 9.842e-07, "loss": 0.01, "step": 45080 }, { "epoch": 2.403422722183718, "grad_norm": 0.4334339499473572, "learning_rate": 9.84e-07, "loss": 0.0038, "step": 45081 }, { "epoch": 2.403476035613371, "grad_norm": 3.51607084274292, "learning_rate": 9.838e-07, "loss": 0.0272, "step": 45082 }, { "epoch": 2.403529349043024, "grad_norm": 1.8640822172164917, "learning_rate": 9.836000000000002e-07, "loss": 0.0314, "step": 45083 }, { "epoch": 2.403582662472677, "grad_norm": 0.18843136727809906, "learning_rate": 9.834e-07, "loss": 0.0016, "step": 45084 }, { "epoch": 2.40363597590233, "grad_norm": 4.066010475158691, "learning_rate": 9.832000000000002e-07, "loss": 0.0408, "step": 45085 }, { "epoch": 2.403689289331983, "grad_norm": 2.0595695972442627, "learning_rate": 9.830000000000002e-07, "loss": 0.0151, "step": 45086 }, { "epoch": 2.4037426027616355, "grad_norm": 2.784238338470459, "learning_rate": 9.828000000000001e-07, "loss": 0.0336, "step": 45087 }, { "epoch": 2.4037959161912887, "grad_norm": 2.756342649459839, "learning_rate": 9.826000000000001e-07, "loss": 0.015, "step": 45088 }, { "epoch": 2.4038492296209415, "grad_norm": 3.851661205291748, "learning_rate": 9.824000000000001e-07, "loss": 0.0338, "step": 45089 }, { "epoch": 2.4039025430505943, "grad_norm": 0.8670977354049683, "learning_rate": 9.822e-07, "loss": 0.0026, "step": 45090 }, { "epoch": 2.4039558564802475, "grad_norm": 7.216765880584717, "learning_rate": 9.82e-07, "loss": 0.0148, "step": 45091 }, { "epoch": 2.4040091699099, "grad_norm": 2.758089303970337, "learning_rate": 9.818e-07, "loss": 0.0188, "step": 45092 }, { "epoch": 2.4040624833395534, "grad_norm": 2.1580567359924316, "learning_rate": 9.816e-07, "loss": 0.0236, "step": 45093 }, { "epoch": 2.404115796769206, "grad_norm": 1.6654455661773682, "learning_rate": 9.814e-07, "loss": 0.0136, "step": 45094 }, { "epoch": 2.404169110198859, "grad_norm": 1.163480520248413, "learning_rate": 9.812e-07, "loss": 0.0249, "step": 45095 }, { "epoch": 2.404222423628512, "grad_norm": 1.419898509979248, "learning_rate": 9.810000000000002e-07, "loss": 0.0122, "step": 45096 }, { "epoch": 2.404275737058165, "grad_norm": 2.7476422786712646, "learning_rate": 9.808e-07, "loss": 0.0131, "step": 45097 }, { "epoch": 2.404329050487818, "grad_norm": 0.7565276622772217, "learning_rate": 9.806e-07, "loss": 0.0025, "step": 45098 }, { "epoch": 2.404382363917471, "grad_norm": 1.4283349514007568, "learning_rate": 9.804000000000002e-07, "loss": 0.0063, "step": 45099 }, { "epoch": 2.4044356773471236, "grad_norm": 2.854776382446289, "learning_rate": 9.802e-07, "loss": 0.0231, "step": 45100 }, { "epoch": 2.404488990776777, "grad_norm": 1.2897416353225708, "learning_rate": 9.800000000000001e-07, "loss": 0.0206, "step": 45101 }, { "epoch": 2.4045423042064296, "grad_norm": 0.5430787205696106, "learning_rate": 9.798000000000001e-07, "loss": 0.0028, "step": 45102 }, { "epoch": 2.4045956176360823, "grad_norm": 2.358159065246582, "learning_rate": 9.796000000000001e-07, "loss": 0.0293, "step": 45103 }, { "epoch": 2.4046489310657355, "grad_norm": 1.861142635345459, "learning_rate": 9.794e-07, "loss": 0.0092, "step": 45104 }, { "epoch": 2.4047022444953883, "grad_norm": 1.821101188659668, "learning_rate": 9.792e-07, "loss": 0.0219, "step": 45105 }, { "epoch": 2.4047555579250415, "grad_norm": 7.295811176300049, "learning_rate": 9.79e-07, "loss": 0.0325, "step": 45106 }, { "epoch": 2.4048088713546942, "grad_norm": 1.339065670967102, "learning_rate": 9.788e-07, "loss": 0.0131, "step": 45107 }, { "epoch": 2.404862184784347, "grad_norm": 2.2732596397399902, "learning_rate": 9.786e-07, "loss": 0.0068, "step": 45108 }, { "epoch": 2.404915498214, "grad_norm": 1.2701393365859985, "learning_rate": 9.784e-07, "loss": 0.0113, "step": 45109 }, { "epoch": 2.404968811643653, "grad_norm": 0.5840885639190674, "learning_rate": 9.782000000000002e-07, "loss": 0.0043, "step": 45110 }, { "epoch": 2.405022125073306, "grad_norm": 1.8764369487762451, "learning_rate": 9.78e-07, "loss": 0.0198, "step": 45111 }, { "epoch": 2.405075438502959, "grad_norm": 1.328943133354187, "learning_rate": 9.778000000000002e-07, "loss": 0.0156, "step": 45112 }, { "epoch": 2.4051287519326117, "grad_norm": 0.35955333709716797, "learning_rate": 9.776000000000002e-07, "loss": 0.0022, "step": 45113 }, { "epoch": 2.405182065362265, "grad_norm": 2.5238070487976074, "learning_rate": 9.774e-07, "loss": 0.0189, "step": 45114 }, { "epoch": 2.4052353787919176, "grad_norm": 2.629091501235962, "learning_rate": 9.772000000000001e-07, "loss": 0.0318, "step": 45115 }, { "epoch": 2.4052886922215704, "grad_norm": 3.0574920177459717, "learning_rate": 9.770000000000001e-07, "loss": 0.0235, "step": 45116 }, { "epoch": 2.4053420056512236, "grad_norm": 1.427022933959961, "learning_rate": 9.768000000000001e-07, "loss": 0.0411, "step": 45117 }, { "epoch": 2.4053953190808763, "grad_norm": 0.7081210613250732, "learning_rate": 9.766e-07, "loss": 0.0041, "step": 45118 }, { "epoch": 2.4054486325105295, "grad_norm": 1.371536135673523, "learning_rate": 9.764e-07, "loss": 0.0246, "step": 45119 }, { "epoch": 2.4055019459401823, "grad_norm": 0.1726221740245819, "learning_rate": 9.762e-07, "loss": 0.0019, "step": 45120 }, { "epoch": 2.4055552593698355, "grad_norm": 1.2194292545318604, "learning_rate": 9.76e-07, "loss": 0.0056, "step": 45121 }, { "epoch": 2.4056085727994883, "grad_norm": 1.3620473146438599, "learning_rate": 9.758e-07, "loss": 0.014, "step": 45122 }, { "epoch": 2.405661886229141, "grad_norm": 1.8012912273406982, "learning_rate": 9.756e-07, "loss": 0.009, "step": 45123 }, { "epoch": 2.405715199658794, "grad_norm": 2.598013162612915, "learning_rate": 9.754e-07, "loss": 0.0133, "step": 45124 }, { "epoch": 2.405768513088447, "grad_norm": 1.24245285987854, "learning_rate": 9.752e-07, "loss": 0.005, "step": 45125 }, { "epoch": 2.4058218265180997, "grad_norm": 3.6608827114105225, "learning_rate": 9.750000000000002e-07, "loss": 0.0267, "step": 45126 }, { "epoch": 2.405875139947753, "grad_norm": 2.155442476272583, "learning_rate": 9.748e-07, "loss": 0.0475, "step": 45127 }, { "epoch": 2.4059284533774057, "grad_norm": 2.5607221126556396, "learning_rate": 9.746000000000002e-07, "loss": 0.0106, "step": 45128 }, { "epoch": 2.405981766807059, "grad_norm": 3.0732243061065674, "learning_rate": 9.744000000000001e-07, "loss": 0.014, "step": 45129 }, { "epoch": 2.4060350802367116, "grad_norm": 1.4324952363967896, "learning_rate": 9.742000000000001e-07, "loss": 0.024, "step": 45130 }, { "epoch": 2.4060883936663644, "grad_norm": 1.6756036281585693, "learning_rate": 9.740000000000001e-07, "loss": 0.0071, "step": 45131 }, { "epoch": 2.4061417070960176, "grad_norm": 4.304277420043945, "learning_rate": 9.738e-07, "loss": 0.0132, "step": 45132 }, { "epoch": 2.4061950205256704, "grad_norm": 1.2114923000335693, "learning_rate": 9.736e-07, "loss": 0.0079, "step": 45133 }, { "epoch": 2.4062483339553236, "grad_norm": 0.7128974199295044, "learning_rate": 9.734e-07, "loss": 0.0041, "step": 45134 }, { "epoch": 2.4063016473849763, "grad_norm": 2.556849956512451, "learning_rate": 9.732e-07, "loss": 0.0124, "step": 45135 }, { "epoch": 2.406354960814629, "grad_norm": 1.0178216695785522, "learning_rate": 9.73e-07, "loss": 0.005, "step": 45136 }, { "epoch": 2.4064082742442823, "grad_norm": 2.7342395782470703, "learning_rate": 9.728e-07, "loss": 0.0293, "step": 45137 }, { "epoch": 2.406461587673935, "grad_norm": 3.126386880874634, "learning_rate": 9.726e-07, "loss": 0.0286, "step": 45138 }, { "epoch": 2.406514901103588, "grad_norm": 0.9741333723068237, "learning_rate": 9.724e-07, "loss": 0.0049, "step": 45139 }, { "epoch": 2.406568214533241, "grad_norm": 2.4408841133117676, "learning_rate": 9.722e-07, "loss": 0.0258, "step": 45140 }, { "epoch": 2.4066215279628937, "grad_norm": 1.6388020515441895, "learning_rate": 9.72e-07, "loss": 0.0067, "step": 45141 }, { "epoch": 2.406674841392547, "grad_norm": 1.0795881748199463, "learning_rate": 9.718000000000002e-07, "loss": 0.0057, "step": 45142 }, { "epoch": 2.4067281548221997, "grad_norm": 2.251995801925659, "learning_rate": 9.716e-07, "loss": 0.0135, "step": 45143 }, { "epoch": 2.4067814682518525, "grad_norm": 2.540748119354248, "learning_rate": 9.714000000000001e-07, "loss": 0.0311, "step": 45144 }, { "epoch": 2.4068347816815057, "grad_norm": 0.316541463136673, "learning_rate": 9.712000000000001e-07, "loss": 0.0023, "step": 45145 }, { "epoch": 2.4068880951111584, "grad_norm": 0.7228207588195801, "learning_rate": 9.71e-07, "loss": 0.006, "step": 45146 }, { "epoch": 2.4069414085408116, "grad_norm": 2.8857688903808594, "learning_rate": 9.708e-07, "loss": 0.0075, "step": 45147 }, { "epoch": 2.4069947219704644, "grad_norm": 1.2331666946411133, "learning_rate": 9.706e-07, "loss": 0.0074, "step": 45148 }, { "epoch": 2.407048035400117, "grad_norm": 1.1802873611450195, "learning_rate": 9.704e-07, "loss": 0.0102, "step": 45149 }, { "epoch": 2.4071013488297703, "grad_norm": 2.1337294578552246, "learning_rate": 9.702e-07, "loss": 0.0269, "step": 45150 }, { "epoch": 2.407154662259423, "grad_norm": 0.5515615940093994, "learning_rate": 9.7e-07, "loss": 0.0034, "step": 45151 }, { "epoch": 2.407207975689076, "grad_norm": 2.0241167545318604, "learning_rate": 9.698e-07, "loss": 0.0143, "step": 45152 }, { "epoch": 2.407261289118729, "grad_norm": 1.3108283281326294, "learning_rate": 9.696000000000002e-07, "loss": 0.0256, "step": 45153 }, { "epoch": 2.407314602548382, "grad_norm": 1.112736463546753, "learning_rate": 9.694e-07, "loss": 0.0252, "step": 45154 }, { "epoch": 2.407367915978035, "grad_norm": 1.1588804721832275, "learning_rate": 9.692000000000002e-07, "loss": 0.0218, "step": 45155 }, { "epoch": 2.4074212294076878, "grad_norm": 0.7842468619346619, "learning_rate": 9.690000000000002e-07, "loss": 0.005, "step": 45156 }, { "epoch": 2.407474542837341, "grad_norm": 2.29286789894104, "learning_rate": 9.688e-07, "loss": 0.015, "step": 45157 }, { "epoch": 2.4075278562669937, "grad_norm": 2.0029678344726562, "learning_rate": 9.686000000000001e-07, "loss": 0.0136, "step": 45158 }, { "epoch": 2.4075811696966465, "grad_norm": 1.3592816591262817, "learning_rate": 9.684000000000001e-07, "loss": 0.0067, "step": 45159 }, { "epoch": 2.4076344831262997, "grad_norm": 1.9458460807800293, "learning_rate": 9.682000000000001e-07, "loss": 0.0087, "step": 45160 }, { "epoch": 2.4076877965559524, "grad_norm": 2.8794379234313965, "learning_rate": 9.68e-07, "loss": 0.0124, "step": 45161 }, { "epoch": 2.407741109985605, "grad_norm": 0.4496467411518097, "learning_rate": 9.678e-07, "loss": 0.0029, "step": 45162 }, { "epoch": 2.4077944234152584, "grad_norm": 2.708313465118408, "learning_rate": 9.676e-07, "loss": 0.0116, "step": 45163 }, { "epoch": 2.407847736844911, "grad_norm": 4.757959842681885, "learning_rate": 9.674e-07, "loss": 0.0086, "step": 45164 }, { "epoch": 2.407901050274564, "grad_norm": 0.4445818364620209, "learning_rate": 9.672e-07, "loss": 0.0017, "step": 45165 }, { "epoch": 2.407954363704217, "grad_norm": 0.2551145553588867, "learning_rate": 9.67e-07, "loss": 0.0018, "step": 45166 }, { "epoch": 2.40800767713387, "grad_norm": 1.2759758234024048, "learning_rate": 9.668e-07, "loss": 0.0035, "step": 45167 }, { "epoch": 2.408060990563523, "grad_norm": 3.0817270278930664, "learning_rate": 9.666e-07, "loss": 0.0247, "step": 45168 }, { "epoch": 2.408114303993176, "grad_norm": 0.4557449519634247, "learning_rate": 9.664000000000002e-07, "loss": 0.0033, "step": 45169 }, { "epoch": 2.408167617422829, "grad_norm": 0.1797635704278946, "learning_rate": 9.662e-07, "loss": 0.0016, "step": 45170 }, { "epoch": 2.408220930852482, "grad_norm": 0.5265568494796753, "learning_rate": 9.660000000000002e-07, "loss": 0.0037, "step": 45171 }, { "epoch": 2.4082742442821345, "grad_norm": 2.3096818923950195, "learning_rate": 9.658000000000001e-07, "loss": 0.0072, "step": 45172 }, { "epoch": 2.4083275577117877, "grad_norm": 0.30262652039527893, "learning_rate": 9.656000000000001e-07, "loss": 0.0021, "step": 45173 }, { "epoch": 2.4083808711414405, "grad_norm": 2.8254454135894775, "learning_rate": 9.654000000000001e-07, "loss": 0.0289, "step": 45174 }, { "epoch": 2.4084341845710933, "grad_norm": 1.2521413564682007, "learning_rate": 9.652e-07, "loss": 0.0043, "step": 45175 }, { "epoch": 2.4084874980007465, "grad_norm": 1.4601730108261108, "learning_rate": 9.65e-07, "loss": 0.0055, "step": 45176 }, { "epoch": 2.408540811430399, "grad_norm": 1.5250009298324585, "learning_rate": 9.648e-07, "loss": 0.0283, "step": 45177 }, { "epoch": 2.4085941248600524, "grad_norm": 1.5788078308105469, "learning_rate": 9.646e-07, "loss": 0.0035, "step": 45178 }, { "epoch": 2.408647438289705, "grad_norm": 1.5051296949386597, "learning_rate": 9.644e-07, "loss": 0.0197, "step": 45179 }, { "epoch": 2.408700751719358, "grad_norm": 0.23742415010929108, "learning_rate": 9.642000000000002e-07, "loss": 0.0018, "step": 45180 }, { "epoch": 2.408754065149011, "grad_norm": 0.2957673966884613, "learning_rate": 9.64e-07, "loss": 0.0018, "step": 45181 }, { "epoch": 2.408807378578664, "grad_norm": 2.8193020820617676, "learning_rate": 9.638e-07, "loss": 0.0206, "step": 45182 }, { "epoch": 2.408860692008317, "grad_norm": 5.665653705596924, "learning_rate": 9.636000000000002e-07, "loss": 0.0405, "step": 45183 }, { "epoch": 2.40891400543797, "grad_norm": 0.8422719836235046, "learning_rate": 9.634e-07, "loss": 0.0035, "step": 45184 }, { "epoch": 2.4089673188676226, "grad_norm": 2.0180604457855225, "learning_rate": 9.632000000000002e-07, "loss": 0.0076, "step": 45185 }, { "epoch": 2.409020632297276, "grad_norm": 3.1492903232574463, "learning_rate": 9.630000000000001e-07, "loss": 0.0407, "step": 45186 }, { "epoch": 2.4090739457269286, "grad_norm": 2.6571338176727295, "learning_rate": 9.628000000000001e-07, "loss": 0.0106, "step": 45187 }, { "epoch": 2.4091272591565813, "grad_norm": 2.5136091709136963, "learning_rate": 9.626000000000001e-07, "loss": 0.0231, "step": 45188 }, { "epoch": 2.4091805725862345, "grad_norm": 3.121964454650879, "learning_rate": 9.624e-07, "loss": 0.0139, "step": 45189 }, { "epoch": 2.4092338860158873, "grad_norm": 3.864135980606079, "learning_rate": 9.622e-07, "loss": 0.0211, "step": 45190 }, { "epoch": 2.4092871994455405, "grad_norm": 2.53355073928833, "learning_rate": 9.62e-07, "loss": 0.0375, "step": 45191 }, { "epoch": 2.4093405128751932, "grad_norm": 2.6608593463897705, "learning_rate": 9.618e-07, "loss": 0.0125, "step": 45192 }, { "epoch": 2.409393826304846, "grad_norm": 0.6236881017684937, "learning_rate": 9.616e-07, "loss": 0.0043, "step": 45193 }, { "epoch": 2.409447139734499, "grad_norm": 0.5971977710723877, "learning_rate": 9.614e-07, "loss": 0.0048, "step": 45194 }, { "epoch": 2.409500453164152, "grad_norm": 1.4314823150634766, "learning_rate": 9.612e-07, "loss": 0.0097, "step": 45195 }, { "epoch": 2.409553766593805, "grad_norm": 2.597992420196533, "learning_rate": 9.610000000000002e-07, "loss": 0.0148, "step": 45196 }, { "epoch": 2.409607080023458, "grad_norm": 1.3408151865005493, "learning_rate": 9.608e-07, "loss": 0.0167, "step": 45197 }, { "epoch": 2.4096603934531107, "grad_norm": 2.97556209564209, "learning_rate": 9.606000000000002e-07, "loss": 0.024, "step": 45198 }, { "epoch": 2.409713706882764, "grad_norm": 2.9100372791290283, "learning_rate": 9.604000000000002e-07, "loss": 0.0146, "step": 45199 }, { "epoch": 2.4097670203124166, "grad_norm": 2.829998016357422, "learning_rate": 9.602e-07, "loss": 0.0186, "step": 45200 }, { "epoch": 2.4098203337420694, "grad_norm": 2.936384916305542, "learning_rate": 9.600000000000001e-07, "loss": 0.0377, "step": 45201 }, { "epoch": 2.4098736471717226, "grad_norm": 1.5994669198989868, "learning_rate": 9.598000000000001e-07, "loss": 0.0088, "step": 45202 }, { "epoch": 2.4099269606013753, "grad_norm": 1.9355978965759277, "learning_rate": 9.596e-07, "loss": 0.0234, "step": 45203 }, { "epoch": 2.4099802740310285, "grad_norm": 3.036733627319336, "learning_rate": 9.594e-07, "loss": 0.0448, "step": 45204 }, { "epoch": 2.4100335874606813, "grad_norm": 1.4618816375732422, "learning_rate": 9.592e-07, "loss": 0.0179, "step": 45205 }, { "epoch": 2.4100869008903345, "grad_norm": 5.688957691192627, "learning_rate": 9.59e-07, "loss": 0.0231, "step": 45206 }, { "epoch": 2.4101402143199873, "grad_norm": 0.664959728717804, "learning_rate": 9.588000000000002e-07, "loss": 0.0026, "step": 45207 }, { "epoch": 2.41019352774964, "grad_norm": 3.58302640914917, "learning_rate": 9.586e-07, "loss": 0.0128, "step": 45208 }, { "epoch": 2.410246841179293, "grad_norm": 1.0048640966415405, "learning_rate": 9.584e-07, "loss": 0.0048, "step": 45209 }, { "epoch": 2.410300154608946, "grad_norm": 1.743295431137085, "learning_rate": 9.582000000000002e-07, "loss": 0.0164, "step": 45210 }, { "epoch": 2.4103534680385987, "grad_norm": 1.483794927597046, "learning_rate": 9.58e-07, "loss": 0.0072, "step": 45211 }, { "epoch": 2.410406781468252, "grad_norm": 2.537182569503784, "learning_rate": 9.578000000000002e-07, "loss": 0.0166, "step": 45212 }, { "epoch": 2.4104600948979047, "grad_norm": 1.4973986148834229, "learning_rate": 9.576000000000002e-07, "loss": 0.0073, "step": 45213 }, { "epoch": 2.410513408327558, "grad_norm": 3.3678834438323975, "learning_rate": 9.574000000000001e-07, "loss": 0.0292, "step": 45214 }, { "epoch": 2.4105667217572107, "grad_norm": 1.679626226425171, "learning_rate": 9.572000000000001e-07, "loss": 0.0039, "step": 45215 }, { "epoch": 2.4106200351868634, "grad_norm": 2.68833065032959, "learning_rate": 9.570000000000001e-07, "loss": 0.0146, "step": 45216 }, { "epoch": 2.4106733486165166, "grad_norm": 0.8530656099319458, "learning_rate": 9.568e-07, "loss": 0.0037, "step": 45217 }, { "epoch": 2.4107266620461694, "grad_norm": 1.7020461559295654, "learning_rate": 9.566e-07, "loss": 0.008, "step": 45218 }, { "epoch": 2.4107799754758226, "grad_norm": 1.9448314905166626, "learning_rate": 9.564e-07, "loss": 0.0084, "step": 45219 }, { "epoch": 2.4108332889054753, "grad_norm": 3.693591356277466, "learning_rate": 9.562e-07, "loss": 0.0355, "step": 45220 }, { "epoch": 2.410886602335128, "grad_norm": 1.7464860677719116, "learning_rate": 9.56e-07, "loss": 0.0161, "step": 45221 }, { "epoch": 2.4109399157647813, "grad_norm": 4.1122660636901855, "learning_rate": 9.558e-07, "loss": 0.0488, "step": 45222 }, { "epoch": 2.410993229194434, "grad_norm": 2.7872817516326904, "learning_rate": 9.556000000000002e-07, "loss": 0.0265, "step": 45223 }, { "epoch": 2.411046542624087, "grad_norm": 0.8455019593238831, "learning_rate": 9.554e-07, "loss": 0.0035, "step": 45224 }, { "epoch": 2.41109985605374, "grad_norm": 0.34324222803115845, "learning_rate": 9.552e-07, "loss": 0.002, "step": 45225 }, { "epoch": 2.4111531694833928, "grad_norm": 3.2194738388061523, "learning_rate": 9.550000000000002e-07, "loss": 0.0141, "step": 45226 }, { "epoch": 2.411206482913046, "grad_norm": 1.4801149368286133, "learning_rate": 9.548e-07, "loss": 0.0176, "step": 45227 }, { "epoch": 2.4112597963426987, "grad_norm": 2.8661954402923584, "learning_rate": 9.546000000000001e-07, "loss": 0.0312, "step": 45228 }, { "epoch": 2.4113131097723515, "grad_norm": 2.155651569366455, "learning_rate": 9.544000000000001e-07, "loss": 0.015, "step": 45229 }, { "epoch": 2.4113664232020047, "grad_norm": 2.6443533897399902, "learning_rate": 9.542000000000001e-07, "loss": 0.0056, "step": 45230 }, { "epoch": 2.4114197366316574, "grad_norm": 0.4716908633708954, "learning_rate": 9.54e-07, "loss": 0.0029, "step": 45231 }, { "epoch": 2.4114730500613106, "grad_norm": 2.8403375148773193, "learning_rate": 9.538e-07, "loss": 0.0306, "step": 45232 }, { "epoch": 2.4115263634909634, "grad_norm": 1.7082550525665283, "learning_rate": 9.536000000000001e-07, "loss": 0.0257, "step": 45233 }, { "epoch": 2.411579676920616, "grad_norm": 2.402578115463257, "learning_rate": 9.534000000000001e-07, "loss": 0.0231, "step": 45234 }, { "epoch": 2.4116329903502693, "grad_norm": 4.2970194816589355, "learning_rate": 9.532000000000001e-07, "loss": 0.0202, "step": 45235 }, { "epoch": 2.411686303779922, "grad_norm": 2.3542683124542236, "learning_rate": 9.53e-07, "loss": 0.0131, "step": 45236 }, { "epoch": 2.411739617209575, "grad_norm": 3.4410388469696045, "learning_rate": 9.528000000000001e-07, "loss": 0.0368, "step": 45237 }, { "epoch": 2.411792930639228, "grad_norm": 2.4884254932403564, "learning_rate": 9.526000000000001e-07, "loss": 0.0138, "step": 45238 }, { "epoch": 2.411846244068881, "grad_norm": 3.7171130180358887, "learning_rate": 9.524000000000001e-07, "loss": 0.0294, "step": 45239 }, { "epoch": 2.411899557498534, "grad_norm": 2.171638011932373, "learning_rate": 9.522000000000001e-07, "loss": 0.0065, "step": 45240 }, { "epoch": 2.4119528709281868, "grad_norm": 2.8092398643493652, "learning_rate": 9.520000000000002e-07, "loss": 0.0143, "step": 45241 }, { "epoch": 2.41200618435784, "grad_norm": 2.708953380584717, "learning_rate": 9.518e-07, "loss": 0.009, "step": 45242 }, { "epoch": 2.4120594977874927, "grad_norm": 1.4223296642303467, "learning_rate": 9.516e-07, "loss": 0.0062, "step": 45243 }, { "epoch": 2.4121128112171455, "grad_norm": 1.818894863128662, "learning_rate": 9.514000000000001e-07, "loss": 0.0244, "step": 45244 }, { "epoch": 2.4121661246467987, "grad_norm": 1.188918948173523, "learning_rate": 9.512e-07, "loss": 0.0052, "step": 45245 }, { "epoch": 2.4122194380764514, "grad_norm": 1.4807318449020386, "learning_rate": 9.510000000000001e-07, "loss": 0.0136, "step": 45246 }, { "epoch": 2.412272751506104, "grad_norm": 2.332740306854248, "learning_rate": 9.508000000000001e-07, "loss": 0.0337, "step": 45247 }, { "epoch": 2.4123260649357574, "grad_norm": 1.8077489137649536, "learning_rate": 9.506000000000002e-07, "loss": 0.0117, "step": 45248 }, { "epoch": 2.41237937836541, "grad_norm": 0.5507253408432007, "learning_rate": 9.504e-07, "loss": 0.0046, "step": 45249 }, { "epoch": 2.4124326917950634, "grad_norm": 1.9705771207809448, "learning_rate": 9.502e-07, "loss": 0.009, "step": 45250 }, { "epoch": 2.412486005224716, "grad_norm": 0.7492890357971191, "learning_rate": 9.500000000000001e-07, "loss": 0.005, "step": 45251 }, { "epoch": 2.412539318654369, "grad_norm": 3.5082905292510986, "learning_rate": 9.498e-07, "loss": 0.0127, "step": 45252 }, { "epoch": 2.412592632084022, "grad_norm": 1.9308257102966309, "learning_rate": 9.496000000000001e-07, "loss": 0.0228, "step": 45253 }, { "epoch": 2.412645945513675, "grad_norm": 1.792661190032959, "learning_rate": 9.494000000000001e-07, "loss": 0.0062, "step": 45254 }, { "epoch": 2.412699258943328, "grad_norm": 1.7425992488861084, "learning_rate": 9.492000000000002e-07, "loss": 0.0087, "step": 45255 }, { "epoch": 2.412752572372981, "grad_norm": 1.5452189445495605, "learning_rate": 9.49e-07, "loss": 0.0069, "step": 45256 }, { "epoch": 2.4128058858026336, "grad_norm": 1.0990757942199707, "learning_rate": 9.488000000000001e-07, "loss": 0.0212, "step": 45257 }, { "epoch": 2.4128591992322868, "grad_norm": 1.3992891311645508, "learning_rate": 9.486000000000001e-07, "loss": 0.0078, "step": 45258 }, { "epoch": 2.4129125126619395, "grad_norm": 0.7523952126502991, "learning_rate": 9.484e-07, "loss": 0.004, "step": 45259 }, { "epoch": 2.4129658260915923, "grad_norm": 0.5048986077308655, "learning_rate": 9.482000000000001e-07, "loss": 0.0036, "step": 45260 }, { "epoch": 2.4130191395212455, "grad_norm": 4.247135639190674, "learning_rate": 9.480000000000001e-07, "loss": 0.0423, "step": 45261 }, { "epoch": 2.4130724529508982, "grad_norm": 0.5575157999992371, "learning_rate": 9.478000000000001e-07, "loss": 0.0033, "step": 45262 }, { "epoch": 2.4131257663805514, "grad_norm": 2.995809555053711, "learning_rate": 9.476e-07, "loss": 0.0199, "step": 45263 }, { "epoch": 2.413179079810204, "grad_norm": 0.8276483416557312, "learning_rate": 9.474000000000001e-07, "loss": 0.0045, "step": 45264 }, { "epoch": 2.413232393239857, "grad_norm": 1.4418628215789795, "learning_rate": 9.472e-07, "loss": 0.0312, "step": 45265 }, { "epoch": 2.41328570666951, "grad_norm": 2.426769733428955, "learning_rate": 9.470000000000001e-07, "loss": 0.0323, "step": 45266 }, { "epoch": 2.413339020099163, "grad_norm": 1.5859569311141968, "learning_rate": 9.468000000000001e-07, "loss": 0.018, "step": 45267 }, { "epoch": 2.413392333528816, "grad_norm": 1.4939029216766357, "learning_rate": 9.466e-07, "loss": 0.0113, "step": 45268 }, { "epoch": 2.413445646958469, "grad_norm": 0.6788206696510315, "learning_rate": 9.464000000000001e-07, "loss": 0.0154, "step": 45269 }, { "epoch": 2.4134989603881216, "grad_norm": 2.094460964202881, "learning_rate": 9.462e-07, "loss": 0.0173, "step": 45270 }, { "epoch": 2.413552273817775, "grad_norm": 2.686180353164673, "learning_rate": 9.460000000000001e-07, "loss": 0.0186, "step": 45271 }, { "epoch": 2.4136055872474276, "grad_norm": 2.2404446601867676, "learning_rate": 9.458e-07, "loss": 0.0316, "step": 45272 }, { "epoch": 2.4136589006770803, "grad_norm": 2.504542589187622, "learning_rate": 9.456000000000001e-07, "loss": 0.0065, "step": 45273 }, { "epoch": 2.4137122141067335, "grad_norm": 0.9758428931236267, "learning_rate": 9.454000000000001e-07, "loss": 0.0043, "step": 45274 }, { "epoch": 2.4137655275363863, "grad_norm": 1.482324481010437, "learning_rate": 9.452000000000002e-07, "loss": 0.0063, "step": 45275 }, { "epoch": 2.4138188409660395, "grad_norm": 3.7929062843322754, "learning_rate": 9.450000000000001e-07, "loss": 0.0562, "step": 45276 }, { "epoch": 2.4138721543956922, "grad_norm": 1.1894093751907349, "learning_rate": 9.448e-07, "loss": 0.0089, "step": 45277 }, { "epoch": 2.4139254678253455, "grad_norm": 2.1346192359924316, "learning_rate": 9.446000000000001e-07, "loss": 0.0102, "step": 45278 }, { "epoch": 2.413978781254998, "grad_norm": 1.3011006116867065, "learning_rate": 9.444e-07, "loss": 0.0204, "step": 45279 }, { "epoch": 2.414032094684651, "grad_norm": 3.247126340866089, "learning_rate": 9.442000000000001e-07, "loss": 0.0302, "step": 45280 }, { "epoch": 2.414085408114304, "grad_norm": 0.797428548336029, "learning_rate": 9.440000000000001e-07, "loss": 0.003, "step": 45281 }, { "epoch": 2.414138721543957, "grad_norm": 1.702547550201416, "learning_rate": 9.438000000000002e-07, "loss": 0.0168, "step": 45282 }, { "epoch": 2.4141920349736097, "grad_norm": 0.9212111234664917, "learning_rate": 9.436000000000001e-07, "loss": 0.0042, "step": 45283 }, { "epoch": 2.414245348403263, "grad_norm": 1.9469839334487915, "learning_rate": 9.434e-07, "loss": 0.0128, "step": 45284 }, { "epoch": 2.4142986618329156, "grad_norm": 3.865403413772583, "learning_rate": 9.432000000000001e-07, "loss": 0.0089, "step": 45285 }, { "epoch": 2.4143519752625684, "grad_norm": 2.0681662559509277, "learning_rate": 9.43e-07, "loss": 0.0256, "step": 45286 }, { "epoch": 2.4144052886922216, "grad_norm": 2.5406858921051025, "learning_rate": 9.428000000000001e-07, "loss": 0.0419, "step": 45287 }, { "epoch": 2.4144586021218744, "grad_norm": 3.1998445987701416, "learning_rate": 9.426000000000001e-07, "loss": 0.0205, "step": 45288 }, { "epoch": 2.4145119155515276, "grad_norm": 2.173259973526001, "learning_rate": 9.424000000000001e-07, "loss": 0.0221, "step": 45289 }, { "epoch": 2.4145652289811803, "grad_norm": 1.0859042406082153, "learning_rate": 9.422000000000001e-07, "loss": 0.0174, "step": 45290 }, { "epoch": 2.4146185424108335, "grad_norm": 3.2163503170013428, "learning_rate": 9.420000000000002e-07, "loss": 0.0162, "step": 45291 }, { "epoch": 2.4146718558404863, "grad_norm": 3.816274404525757, "learning_rate": 9.418e-07, "loss": 0.0141, "step": 45292 }, { "epoch": 2.414725169270139, "grad_norm": 0.7422002553939819, "learning_rate": 9.416e-07, "loss": 0.0038, "step": 45293 }, { "epoch": 2.4147784826997922, "grad_norm": 3.2169368267059326, "learning_rate": 9.414000000000001e-07, "loss": 0.0091, "step": 45294 }, { "epoch": 2.414831796129445, "grad_norm": 1.5156186819076538, "learning_rate": 9.412e-07, "loss": 0.0164, "step": 45295 }, { "epoch": 2.4148851095590977, "grad_norm": 2.217759609222412, "learning_rate": 9.410000000000001e-07, "loss": 0.0218, "step": 45296 }, { "epoch": 2.414938422988751, "grad_norm": 2.0855154991149902, "learning_rate": 9.408000000000001e-07, "loss": 0.0319, "step": 45297 }, { "epoch": 2.4149917364184037, "grad_norm": 1.7466261386871338, "learning_rate": 9.406000000000002e-07, "loss": 0.0235, "step": 45298 }, { "epoch": 2.415045049848057, "grad_norm": 0.6351469159126282, "learning_rate": 9.404e-07, "loss": 0.0045, "step": 45299 }, { "epoch": 2.4150983632777097, "grad_norm": 0.4174700677394867, "learning_rate": 9.402000000000001e-07, "loss": 0.003, "step": 45300 }, { "epoch": 2.4151516767073624, "grad_norm": 0.9770643711090088, "learning_rate": 9.400000000000001e-07, "loss": 0.0122, "step": 45301 }, { "epoch": 2.4152049901370156, "grad_norm": 0.6092689037322998, "learning_rate": 9.398e-07, "loss": 0.0037, "step": 45302 }, { "epoch": 2.4152583035666684, "grad_norm": 1.8882167339324951, "learning_rate": 9.396000000000001e-07, "loss": 0.0071, "step": 45303 }, { "epoch": 2.4153116169963216, "grad_norm": 1.4347224235534668, "learning_rate": 9.394000000000001e-07, "loss": 0.0163, "step": 45304 }, { "epoch": 2.4153649304259743, "grad_norm": 1.679768681526184, "learning_rate": 9.392000000000002e-07, "loss": 0.0224, "step": 45305 }, { "epoch": 2.415418243855627, "grad_norm": 1.8236963748931885, "learning_rate": 9.39e-07, "loss": 0.0234, "step": 45306 }, { "epoch": 2.4154715572852803, "grad_norm": 2.3740158081054688, "learning_rate": 9.388000000000001e-07, "loss": 0.0069, "step": 45307 }, { "epoch": 2.415524870714933, "grad_norm": 1.1543580293655396, "learning_rate": 9.386000000000001e-07, "loss": 0.0043, "step": 45308 }, { "epoch": 2.415578184144586, "grad_norm": 2.158018112182617, "learning_rate": 9.384000000000001e-07, "loss": 0.0163, "step": 45309 }, { "epoch": 2.415631497574239, "grad_norm": 2.0282328128814697, "learning_rate": 9.382000000000001e-07, "loss": 0.008, "step": 45310 }, { "epoch": 2.4156848110038918, "grad_norm": 3.3212130069732666, "learning_rate": 9.380000000000001e-07, "loss": 0.0336, "step": 45311 }, { "epoch": 2.415738124433545, "grad_norm": 0.41365545988082886, "learning_rate": 9.378e-07, "loss": 0.0038, "step": 45312 }, { "epoch": 2.4157914378631977, "grad_norm": 2.819282054901123, "learning_rate": 9.376e-07, "loss": 0.0399, "step": 45313 }, { "epoch": 2.4158447512928505, "grad_norm": 1.6067252159118652, "learning_rate": 9.374000000000001e-07, "loss": 0.0152, "step": 45314 }, { "epoch": 2.4158980647225037, "grad_norm": 0.35847294330596924, "learning_rate": 9.372000000000001e-07, "loss": 0.0027, "step": 45315 }, { "epoch": 2.4159513781521564, "grad_norm": 1.2890701293945312, "learning_rate": 9.370000000000001e-07, "loss": 0.0063, "step": 45316 }, { "epoch": 2.4160046915818096, "grad_norm": 0.4220968782901764, "learning_rate": 9.368000000000001e-07, "loss": 0.0032, "step": 45317 }, { "epoch": 2.4160580050114624, "grad_norm": 2.5547730922698975, "learning_rate": 9.366000000000001e-07, "loss": 0.0104, "step": 45318 }, { "epoch": 2.416111318441115, "grad_norm": 0.5228657722473145, "learning_rate": 9.364e-07, "loss": 0.0032, "step": 45319 }, { "epoch": 2.4161646318707684, "grad_norm": 2.1187212467193604, "learning_rate": 9.362e-07, "loss": 0.013, "step": 45320 }, { "epoch": 2.416217945300421, "grad_norm": 2.2797834873199463, "learning_rate": 9.360000000000001e-07, "loss": 0.0074, "step": 45321 }, { "epoch": 2.416271258730074, "grad_norm": 1.0022070407867432, "learning_rate": 9.358e-07, "loss": 0.004, "step": 45322 }, { "epoch": 2.416324572159727, "grad_norm": 1.061892032623291, "learning_rate": 9.356000000000001e-07, "loss": 0.0038, "step": 45323 }, { "epoch": 2.41637788558938, "grad_norm": 0.8684917092323303, "learning_rate": 9.354000000000001e-07, "loss": 0.0179, "step": 45324 }, { "epoch": 2.416431199019033, "grad_norm": 2.192291736602783, "learning_rate": 9.352000000000002e-07, "loss": 0.0236, "step": 45325 }, { "epoch": 2.416484512448686, "grad_norm": 2.398143768310547, "learning_rate": 9.35e-07, "loss": 0.0162, "step": 45326 }, { "epoch": 2.416537825878339, "grad_norm": 0.08757315576076508, "learning_rate": 9.348e-07, "loss": 0.001, "step": 45327 }, { "epoch": 2.4165911393079917, "grad_norm": 2.1595256328582764, "learning_rate": 9.346000000000001e-07, "loss": 0.0377, "step": 45328 }, { "epoch": 2.4166444527376445, "grad_norm": 1.327369213104248, "learning_rate": 9.344e-07, "loss": 0.0209, "step": 45329 }, { "epoch": 2.4166977661672977, "grad_norm": 0.7361290454864502, "learning_rate": 9.342000000000001e-07, "loss": 0.012, "step": 45330 }, { "epoch": 2.4167510795969505, "grad_norm": 3.736212730407715, "learning_rate": 9.340000000000001e-07, "loss": 0.0264, "step": 45331 }, { "epoch": 2.416804393026603, "grad_norm": 1.6471686363220215, "learning_rate": 9.338000000000002e-07, "loss": 0.0126, "step": 45332 }, { "epoch": 2.4168577064562564, "grad_norm": 0.23791718482971191, "learning_rate": 9.336e-07, "loss": 0.0019, "step": 45333 }, { "epoch": 2.416911019885909, "grad_norm": 2.010740041732788, "learning_rate": 9.334000000000001e-07, "loss": 0.0132, "step": 45334 }, { "epoch": 2.4169643333155624, "grad_norm": 1.6731045246124268, "learning_rate": 9.332000000000001e-07, "loss": 0.008, "step": 45335 }, { "epoch": 2.417017646745215, "grad_norm": 0.3502562940120697, "learning_rate": 9.33e-07, "loss": 0.0021, "step": 45336 }, { "epoch": 2.417070960174868, "grad_norm": 1.4693297147750854, "learning_rate": 9.328000000000001e-07, "loss": 0.005, "step": 45337 }, { "epoch": 2.417124273604521, "grad_norm": 1.898730993270874, "learning_rate": 9.326000000000001e-07, "loss": 0.0072, "step": 45338 }, { "epoch": 2.417177587034174, "grad_norm": 3.763181447982788, "learning_rate": 9.324000000000001e-07, "loss": 0.043, "step": 45339 }, { "epoch": 2.417230900463827, "grad_norm": 1.233512043952942, "learning_rate": 9.322e-07, "loss": 0.0047, "step": 45340 }, { "epoch": 2.41728421389348, "grad_norm": 2.099243640899658, "learning_rate": 9.320000000000001e-07, "loss": 0.0292, "step": 45341 }, { "epoch": 2.4173375273231326, "grad_norm": 3.2576451301574707, "learning_rate": 9.318e-07, "loss": 0.0259, "step": 45342 }, { "epoch": 2.4173908407527858, "grad_norm": 1.1236927509307861, "learning_rate": 9.316000000000001e-07, "loss": 0.0103, "step": 45343 }, { "epoch": 2.4174441541824385, "grad_norm": 3.442702293395996, "learning_rate": 9.314000000000001e-07, "loss": 0.0207, "step": 45344 }, { "epoch": 2.4174974676120913, "grad_norm": 0.7546215653419495, "learning_rate": 9.312e-07, "loss": 0.0043, "step": 45345 }, { "epoch": 2.4175507810417445, "grad_norm": 2.304060935974121, "learning_rate": 9.310000000000001e-07, "loss": 0.0328, "step": 45346 }, { "epoch": 2.4176040944713972, "grad_norm": 2.18900728225708, "learning_rate": 9.308e-07, "loss": 0.0244, "step": 45347 }, { "epoch": 2.4176574079010504, "grad_norm": 1.0342861413955688, "learning_rate": 9.306000000000001e-07, "loss": 0.0215, "step": 45348 }, { "epoch": 2.417710721330703, "grad_norm": 1.3956235647201538, "learning_rate": 9.304e-07, "loss": 0.008, "step": 45349 }, { "epoch": 2.417764034760356, "grad_norm": 1.59397292137146, "learning_rate": 9.302000000000001e-07, "loss": 0.0217, "step": 45350 }, { "epoch": 2.417817348190009, "grad_norm": 1.85288667678833, "learning_rate": 9.300000000000001e-07, "loss": 0.032, "step": 45351 }, { "epoch": 2.417870661619662, "grad_norm": 0.3783857524394989, "learning_rate": 9.298e-07, "loss": 0.0028, "step": 45352 }, { "epoch": 2.417923975049315, "grad_norm": 1.3882149457931519, "learning_rate": 9.296000000000001e-07, "loss": 0.006, "step": 45353 }, { "epoch": 2.417977288478968, "grad_norm": 1.9795944690704346, "learning_rate": 9.294e-07, "loss": 0.0166, "step": 45354 }, { "epoch": 2.4180306019086206, "grad_norm": 0.4349474608898163, "learning_rate": 9.292000000000001e-07, "loss": 0.0039, "step": 45355 }, { "epoch": 2.418083915338274, "grad_norm": 2.2243382930755615, "learning_rate": 9.29e-07, "loss": 0.0287, "step": 45356 }, { "epoch": 2.4181372287679266, "grad_norm": 2.8781418800354004, "learning_rate": 9.288000000000001e-07, "loss": 0.0165, "step": 45357 }, { "epoch": 2.4181905421975793, "grad_norm": 3.422485828399658, "learning_rate": 9.286000000000001e-07, "loss": 0.0197, "step": 45358 }, { "epoch": 2.4182438556272325, "grad_norm": 1.7720030546188354, "learning_rate": 9.284000000000002e-07, "loss": 0.0076, "step": 45359 }, { "epoch": 2.4182971690568853, "grad_norm": 1.232328176498413, "learning_rate": 9.282000000000001e-07, "loss": 0.023, "step": 45360 }, { "epoch": 2.4183504824865385, "grad_norm": 2.8310348987579346, "learning_rate": 9.28e-07, "loss": 0.0522, "step": 45361 }, { "epoch": 2.4184037959161913, "grad_norm": 3.4309675693511963, "learning_rate": 9.278000000000001e-07, "loss": 0.017, "step": 45362 }, { "epoch": 2.4184571093458445, "grad_norm": 1.4446967840194702, "learning_rate": 9.276e-07, "loss": 0.019, "step": 45363 }, { "epoch": 2.418510422775497, "grad_norm": 3.1259539127349854, "learning_rate": 9.274000000000001e-07, "loss": 0.0166, "step": 45364 }, { "epoch": 2.41856373620515, "grad_norm": 1.0803829431533813, "learning_rate": 9.272000000000001e-07, "loss": 0.0058, "step": 45365 }, { "epoch": 2.418617049634803, "grad_norm": 1.2209978103637695, "learning_rate": 9.270000000000001e-07, "loss": 0.0077, "step": 45366 }, { "epoch": 2.418670363064456, "grad_norm": 2.009086847305298, "learning_rate": 9.268000000000001e-07, "loss": 0.0082, "step": 45367 }, { "epoch": 2.4187236764941087, "grad_norm": 1.5911227464675903, "learning_rate": 9.266000000000002e-07, "loss": 0.0074, "step": 45368 }, { "epoch": 2.418776989923762, "grad_norm": 1.1633176803588867, "learning_rate": 9.264e-07, "loss": 0.0052, "step": 45369 }, { "epoch": 2.4188303033534146, "grad_norm": 1.493405818939209, "learning_rate": 9.262e-07, "loss": 0.0095, "step": 45370 }, { "epoch": 2.418883616783068, "grad_norm": 2.583542823791504, "learning_rate": 9.260000000000001e-07, "loss": 0.0097, "step": 45371 }, { "epoch": 2.4189369302127206, "grad_norm": 2.5722806453704834, "learning_rate": 9.258e-07, "loss": 0.0171, "step": 45372 }, { "epoch": 2.4189902436423734, "grad_norm": 3.1153643131256104, "learning_rate": 9.256000000000001e-07, "loss": 0.0149, "step": 45373 }, { "epoch": 2.4190435570720266, "grad_norm": 1.6138949394226074, "learning_rate": 9.254000000000001e-07, "loss": 0.0192, "step": 45374 }, { "epoch": 2.4190968705016793, "grad_norm": 0.9283450245857239, "learning_rate": 9.252000000000002e-07, "loss": 0.0048, "step": 45375 }, { "epoch": 2.4191501839313325, "grad_norm": 2.493786334991455, "learning_rate": 9.25e-07, "loss": 0.0387, "step": 45376 }, { "epoch": 2.4192034973609853, "grad_norm": 2.260759115219116, "learning_rate": 9.248000000000001e-07, "loss": 0.0187, "step": 45377 }, { "epoch": 2.419256810790638, "grad_norm": 2.4360926151275635, "learning_rate": 9.246000000000001e-07, "loss": 0.0193, "step": 45378 }, { "epoch": 2.4193101242202912, "grad_norm": 3.00793719291687, "learning_rate": 9.244e-07, "loss": 0.0321, "step": 45379 }, { "epoch": 2.419363437649944, "grad_norm": 2.7142646312713623, "learning_rate": 9.242000000000001e-07, "loss": 0.0119, "step": 45380 }, { "epoch": 2.4194167510795968, "grad_norm": 1.8254246711730957, "learning_rate": 9.240000000000001e-07, "loss": 0.0345, "step": 45381 }, { "epoch": 2.41947006450925, "grad_norm": 1.4081164598464966, "learning_rate": 9.238000000000002e-07, "loss": 0.0048, "step": 45382 }, { "epoch": 2.4195233779389027, "grad_norm": 2.081021308898926, "learning_rate": 9.236e-07, "loss": 0.0254, "step": 45383 }, { "epoch": 2.419576691368556, "grad_norm": 1.6099237203598022, "learning_rate": 9.234000000000001e-07, "loss": 0.0045, "step": 45384 }, { "epoch": 2.4196300047982087, "grad_norm": 2.5914032459259033, "learning_rate": 9.232000000000001e-07, "loss": 0.0125, "step": 45385 }, { "epoch": 2.4196833182278614, "grad_norm": 3.185713052749634, "learning_rate": 9.23e-07, "loss": 0.0096, "step": 45386 }, { "epoch": 2.4197366316575146, "grad_norm": 0.7940409779548645, "learning_rate": 9.228000000000001e-07, "loss": 0.0033, "step": 45387 }, { "epoch": 2.4197899450871674, "grad_norm": 3.5909101963043213, "learning_rate": 9.226000000000001e-07, "loss": 0.0251, "step": 45388 }, { "epoch": 2.4198432585168206, "grad_norm": 1.4805275201797485, "learning_rate": 9.224000000000001e-07, "loss": 0.0179, "step": 45389 }, { "epoch": 2.4198965719464733, "grad_norm": 2.940762519836426, "learning_rate": 9.222e-07, "loss": 0.0568, "step": 45390 }, { "epoch": 2.419949885376126, "grad_norm": 2.16127347946167, "learning_rate": 9.220000000000001e-07, "loss": 0.0213, "step": 45391 }, { "epoch": 2.4200031988057793, "grad_norm": 2.256962537765503, "learning_rate": 9.218e-07, "loss": 0.0208, "step": 45392 }, { "epoch": 2.420056512235432, "grad_norm": 0.2194453775882721, "learning_rate": 9.216000000000001e-07, "loss": 0.0017, "step": 45393 }, { "epoch": 2.420109825665085, "grad_norm": 2.284832715988159, "learning_rate": 9.214000000000001e-07, "loss": 0.0123, "step": 45394 }, { "epoch": 2.420163139094738, "grad_norm": 1.6523147821426392, "learning_rate": 9.212e-07, "loss": 0.0176, "step": 45395 }, { "epoch": 2.4202164525243908, "grad_norm": 3.1767866611480713, "learning_rate": 9.210000000000001e-07, "loss": 0.0114, "step": 45396 }, { "epoch": 2.420269765954044, "grad_norm": 0.5886709690093994, "learning_rate": 9.208e-07, "loss": 0.0043, "step": 45397 }, { "epoch": 2.4203230793836967, "grad_norm": 3.73399019241333, "learning_rate": 9.206000000000001e-07, "loss": 0.051, "step": 45398 }, { "epoch": 2.42037639281335, "grad_norm": 1.9372997283935547, "learning_rate": 9.204e-07, "loss": 0.0173, "step": 45399 }, { "epoch": 2.4204297062430027, "grad_norm": 1.761927843093872, "learning_rate": 9.202000000000001e-07, "loss": 0.0078, "step": 45400 }, { "epoch": 2.4204830196726554, "grad_norm": 2.937368392944336, "learning_rate": 9.200000000000001e-07, "loss": 0.0407, "step": 45401 }, { "epoch": 2.4205363331023086, "grad_norm": 1.263704538345337, "learning_rate": 9.198000000000002e-07, "loss": 0.0235, "step": 45402 }, { "epoch": 2.4205896465319614, "grad_norm": 2.898155689239502, "learning_rate": 9.196000000000001e-07, "loss": 0.0209, "step": 45403 }, { "epoch": 2.420642959961614, "grad_norm": 1.6150157451629639, "learning_rate": 9.194e-07, "loss": 0.0177, "step": 45404 }, { "epoch": 2.4206962733912674, "grad_norm": 2.7961838245391846, "learning_rate": 9.192000000000001e-07, "loss": 0.025, "step": 45405 }, { "epoch": 2.42074958682092, "grad_norm": 4.079680919647217, "learning_rate": 9.19e-07, "loss": 0.0453, "step": 45406 }, { "epoch": 2.420802900250573, "grad_norm": 2.890711545944214, "learning_rate": 9.188000000000001e-07, "loss": 0.0125, "step": 45407 }, { "epoch": 2.420856213680226, "grad_norm": 3.8219611644744873, "learning_rate": 9.186000000000001e-07, "loss": 0.0211, "step": 45408 }, { "epoch": 2.420909527109879, "grad_norm": 0.337082177400589, "learning_rate": 9.184000000000002e-07, "loss": 0.0023, "step": 45409 }, { "epoch": 2.420962840539532, "grad_norm": 0.5223451852798462, "learning_rate": 9.182000000000001e-07, "loss": 0.0038, "step": 45410 }, { "epoch": 2.421016153969185, "grad_norm": 2.5089526176452637, "learning_rate": 9.180000000000001e-07, "loss": 0.0236, "step": 45411 }, { "epoch": 2.421069467398838, "grad_norm": 0.6410806775093079, "learning_rate": 9.178000000000001e-07, "loss": 0.0025, "step": 45412 }, { "epoch": 2.4211227808284908, "grad_norm": 3.292623519897461, "learning_rate": 9.176e-07, "loss": 0.0184, "step": 45413 }, { "epoch": 2.4211760942581435, "grad_norm": 1.9844741821289062, "learning_rate": 9.174000000000001e-07, "loss": 0.0223, "step": 45414 }, { "epoch": 2.4212294076877967, "grad_norm": 1.476318120956421, "learning_rate": 9.172000000000001e-07, "loss": 0.0349, "step": 45415 }, { "epoch": 2.4212827211174495, "grad_norm": 2.6195013523101807, "learning_rate": 9.170000000000001e-07, "loss": 0.0186, "step": 45416 }, { "epoch": 2.4213360345471022, "grad_norm": 1.9528814554214478, "learning_rate": 9.168000000000001e-07, "loss": 0.0195, "step": 45417 }, { "epoch": 2.4213893479767554, "grad_norm": 0.23902982473373413, "learning_rate": 9.166000000000001e-07, "loss": 0.0025, "step": 45418 }, { "epoch": 2.421442661406408, "grad_norm": 0.9306818246841431, "learning_rate": 9.164e-07, "loss": 0.0047, "step": 45419 }, { "epoch": 2.4214959748360614, "grad_norm": 2.763798713684082, "learning_rate": 9.162e-07, "loss": 0.0102, "step": 45420 }, { "epoch": 2.421549288265714, "grad_norm": 2.5045225620269775, "learning_rate": 9.160000000000001e-07, "loss": 0.0143, "step": 45421 }, { "epoch": 2.421602601695367, "grad_norm": 1.8418580293655396, "learning_rate": 9.158e-07, "loss": 0.0282, "step": 45422 }, { "epoch": 2.42165591512502, "grad_norm": 2.5863897800445557, "learning_rate": 9.156000000000001e-07, "loss": 0.0086, "step": 45423 }, { "epoch": 2.421709228554673, "grad_norm": 0.49933257699012756, "learning_rate": 9.154000000000001e-07, "loss": 0.0034, "step": 45424 }, { "epoch": 2.421762541984326, "grad_norm": 1.7061858177185059, "learning_rate": 9.152000000000001e-07, "loss": 0.0128, "step": 45425 }, { "epoch": 2.421815855413979, "grad_norm": 0.28233790397644043, "learning_rate": 9.15e-07, "loss": 0.0019, "step": 45426 }, { "epoch": 2.4218691688436316, "grad_norm": 0.6943986415863037, "learning_rate": 9.148000000000001e-07, "loss": 0.0029, "step": 45427 }, { "epoch": 2.4219224822732848, "grad_norm": 0.8794722557067871, "learning_rate": 9.146000000000001e-07, "loss": 0.006, "step": 45428 }, { "epoch": 2.4219757957029375, "grad_norm": 1.68172025680542, "learning_rate": 9.144e-07, "loss": 0.0112, "step": 45429 }, { "epoch": 2.4220291091325903, "grad_norm": 1.153418779373169, "learning_rate": 9.142000000000001e-07, "loss": 0.0045, "step": 45430 }, { "epoch": 2.4220824225622435, "grad_norm": 2.42179012298584, "learning_rate": 9.140000000000001e-07, "loss": 0.0094, "step": 45431 }, { "epoch": 2.4221357359918962, "grad_norm": 1.8218410015106201, "learning_rate": 9.138000000000001e-07, "loss": 0.0155, "step": 45432 }, { "epoch": 2.4221890494215494, "grad_norm": 2.552813768386841, "learning_rate": 9.136e-07, "loss": 0.0084, "step": 45433 }, { "epoch": 2.422242362851202, "grad_norm": 1.5905699729919434, "learning_rate": 9.134000000000001e-07, "loss": 0.0179, "step": 45434 }, { "epoch": 2.422295676280855, "grad_norm": 2.1245973110198975, "learning_rate": 9.132000000000001e-07, "loss": 0.0131, "step": 45435 }, { "epoch": 2.422348989710508, "grad_norm": 1.5693711042404175, "learning_rate": 9.130000000000001e-07, "loss": 0.0159, "step": 45436 }, { "epoch": 2.422402303140161, "grad_norm": 2.3550643920898438, "learning_rate": 9.128000000000001e-07, "loss": 0.0289, "step": 45437 }, { "epoch": 2.422455616569814, "grad_norm": 2.7013633251190186, "learning_rate": 9.126000000000001e-07, "loss": 0.0605, "step": 45438 }, { "epoch": 2.422508929999467, "grad_norm": 3.919973373413086, "learning_rate": 9.124e-07, "loss": 0.0273, "step": 45439 }, { "epoch": 2.4225622434291196, "grad_norm": 1.8485480546951294, "learning_rate": 9.122e-07, "loss": 0.0056, "step": 45440 }, { "epoch": 2.422615556858773, "grad_norm": 2.1231658458709717, "learning_rate": 9.120000000000001e-07, "loss": 0.0085, "step": 45441 }, { "epoch": 2.4226688702884256, "grad_norm": 1.2238492965698242, "learning_rate": 9.118e-07, "loss": 0.0206, "step": 45442 }, { "epoch": 2.4227221837180783, "grad_norm": 2.435429573059082, "learning_rate": 9.116000000000001e-07, "loss": 0.0083, "step": 45443 }, { "epoch": 2.4227754971477315, "grad_norm": 0.6274316310882568, "learning_rate": 9.114000000000001e-07, "loss": 0.0032, "step": 45444 }, { "epoch": 2.4228288105773843, "grad_norm": 2.5099353790283203, "learning_rate": 9.112000000000002e-07, "loss": 0.0287, "step": 45445 }, { "epoch": 2.4228821240070375, "grad_norm": 1.2067837715148926, "learning_rate": 9.11e-07, "loss": 0.0168, "step": 45446 }, { "epoch": 2.4229354374366903, "grad_norm": 1.0280160903930664, "learning_rate": 9.108e-07, "loss": 0.0114, "step": 45447 }, { "epoch": 2.4229887508663435, "grad_norm": 2.977222204208374, "learning_rate": 9.106000000000001e-07, "loss": 0.0141, "step": 45448 }, { "epoch": 2.4230420642959962, "grad_norm": 0.41739827394485474, "learning_rate": 9.104e-07, "loss": 0.0027, "step": 45449 }, { "epoch": 2.423095377725649, "grad_norm": 1.6348904371261597, "learning_rate": 9.102000000000001e-07, "loss": 0.013, "step": 45450 }, { "epoch": 2.423148691155302, "grad_norm": 1.1001328229904175, "learning_rate": 9.100000000000001e-07, "loss": 0.0057, "step": 45451 }, { "epoch": 2.423202004584955, "grad_norm": 1.9968085289001465, "learning_rate": 9.098000000000002e-07, "loss": 0.0343, "step": 45452 }, { "epoch": 2.4232553180146077, "grad_norm": 2.4768426418304443, "learning_rate": 9.096e-07, "loss": 0.0186, "step": 45453 }, { "epoch": 2.423308631444261, "grad_norm": 1.2424226999282837, "learning_rate": 9.094e-07, "loss": 0.0079, "step": 45454 }, { "epoch": 2.4233619448739137, "grad_norm": 1.5119998455047607, "learning_rate": 9.092000000000001e-07, "loss": 0.0097, "step": 45455 }, { "epoch": 2.423415258303567, "grad_norm": 2.6716816425323486, "learning_rate": 9.09e-07, "loss": 0.0194, "step": 45456 }, { "epoch": 2.4234685717332196, "grad_norm": 3.3133721351623535, "learning_rate": 9.088000000000001e-07, "loss": 0.0151, "step": 45457 }, { "epoch": 2.4235218851628724, "grad_norm": 3.446927309036255, "learning_rate": 9.086000000000001e-07, "loss": 0.0155, "step": 45458 }, { "epoch": 2.4235751985925256, "grad_norm": 2.5527844429016113, "learning_rate": 9.084000000000002e-07, "loss": 0.0577, "step": 45459 }, { "epoch": 2.4236285120221783, "grad_norm": 1.9401018619537354, "learning_rate": 9.082e-07, "loss": 0.0096, "step": 45460 }, { "epoch": 2.4236818254518315, "grad_norm": 0.9530086517333984, "learning_rate": 9.080000000000001e-07, "loss": 0.015, "step": 45461 }, { "epoch": 2.4237351388814843, "grad_norm": 4.201262950897217, "learning_rate": 9.078000000000001e-07, "loss": 0.0254, "step": 45462 }, { "epoch": 2.423788452311137, "grad_norm": 1.3995835781097412, "learning_rate": 9.076e-07, "loss": 0.0057, "step": 45463 }, { "epoch": 2.4238417657407902, "grad_norm": 3.0254242420196533, "learning_rate": 9.074000000000001e-07, "loss": 0.0109, "step": 45464 }, { "epoch": 2.423895079170443, "grad_norm": 0.44642844796180725, "learning_rate": 9.072000000000001e-07, "loss": 0.0032, "step": 45465 }, { "epoch": 2.4239483926000958, "grad_norm": 0.4916349947452545, "learning_rate": 9.070000000000001e-07, "loss": 0.0033, "step": 45466 }, { "epoch": 2.424001706029749, "grad_norm": 1.4768147468566895, "learning_rate": 9.068e-07, "loss": 0.0387, "step": 45467 }, { "epoch": 2.4240550194594017, "grad_norm": 2.5646064281463623, "learning_rate": 9.066000000000001e-07, "loss": 0.0074, "step": 45468 }, { "epoch": 2.424108332889055, "grad_norm": 1.5557005405426025, "learning_rate": 9.064e-07, "loss": 0.0166, "step": 45469 }, { "epoch": 2.4241616463187077, "grad_norm": 1.2510219812393188, "learning_rate": 9.062000000000001e-07, "loss": 0.0035, "step": 45470 }, { "epoch": 2.4242149597483604, "grad_norm": 2.244825839996338, "learning_rate": 9.060000000000001e-07, "loss": 0.0124, "step": 45471 }, { "epoch": 2.4242682731780136, "grad_norm": 0.48139357566833496, "learning_rate": 9.058e-07, "loss": 0.0034, "step": 45472 }, { "epoch": 2.4243215866076664, "grad_norm": 1.7480313777923584, "learning_rate": 9.056000000000001e-07, "loss": 0.0066, "step": 45473 }, { "epoch": 2.4243749000373196, "grad_norm": 1.5566447973251343, "learning_rate": 9.054e-07, "loss": 0.0043, "step": 45474 }, { "epoch": 2.4244282134669723, "grad_norm": 1.9247386455535889, "learning_rate": 9.052000000000001e-07, "loss": 0.0285, "step": 45475 }, { "epoch": 2.424481526896625, "grad_norm": 0.7700898051261902, "learning_rate": 9.05e-07, "loss": 0.0037, "step": 45476 }, { "epoch": 2.4245348403262783, "grad_norm": 0.2838910222053528, "learning_rate": 9.048000000000001e-07, "loss": 0.0018, "step": 45477 }, { "epoch": 2.424588153755931, "grad_norm": 0.5596769452095032, "learning_rate": 9.046000000000001e-07, "loss": 0.0032, "step": 45478 }, { "epoch": 2.424641467185584, "grad_norm": 3.0487148761749268, "learning_rate": 9.044000000000002e-07, "loss": 0.0197, "step": 45479 }, { "epoch": 2.424694780615237, "grad_norm": 1.4710428714752197, "learning_rate": 9.042000000000001e-07, "loss": 0.0146, "step": 45480 }, { "epoch": 2.42474809404489, "grad_norm": 0.9754325747489929, "learning_rate": 9.04e-07, "loss": 0.004, "step": 45481 }, { "epoch": 2.424801407474543, "grad_norm": 0.40853676199913025, "learning_rate": 9.038000000000001e-07, "loss": 0.003, "step": 45482 }, { "epoch": 2.4248547209041957, "grad_norm": 1.8468660116195679, "learning_rate": 9.036e-07, "loss": 0.0048, "step": 45483 }, { "epoch": 2.424908034333849, "grad_norm": 2.0169694423675537, "learning_rate": 9.034000000000001e-07, "loss": 0.0206, "step": 45484 }, { "epoch": 2.4249613477635017, "grad_norm": 3.5259029865264893, "learning_rate": 9.032000000000001e-07, "loss": 0.0126, "step": 45485 }, { "epoch": 2.4250146611931545, "grad_norm": 0.8909129500389099, "learning_rate": 9.030000000000001e-07, "loss": 0.0042, "step": 45486 }, { "epoch": 2.4250679746228077, "grad_norm": 4.302690505981445, "learning_rate": 9.028000000000001e-07, "loss": 0.0212, "step": 45487 }, { "epoch": 2.4251212880524604, "grad_norm": 1.3915034532546997, "learning_rate": 9.026000000000002e-07, "loss": 0.0042, "step": 45488 }, { "epoch": 2.425174601482113, "grad_norm": 1.8965651988983154, "learning_rate": 9.024e-07, "loss": 0.0234, "step": 45489 }, { "epoch": 2.4252279149117664, "grad_norm": 1.584325909614563, "learning_rate": 9.022e-07, "loss": 0.0061, "step": 45490 }, { "epoch": 2.425281228341419, "grad_norm": 2.6924688816070557, "learning_rate": 9.020000000000001e-07, "loss": 0.0325, "step": 45491 }, { "epoch": 2.425334541771072, "grad_norm": 2.268291473388672, "learning_rate": 9.018e-07, "loss": 0.0088, "step": 45492 }, { "epoch": 2.425387855200725, "grad_norm": 4.059277534484863, "learning_rate": 9.016000000000001e-07, "loss": 0.0172, "step": 45493 }, { "epoch": 2.425441168630378, "grad_norm": 1.115224003791809, "learning_rate": 9.014000000000001e-07, "loss": 0.0035, "step": 45494 }, { "epoch": 2.425494482060031, "grad_norm": 2.320753335952759, "learning_rate": 9.012000000000002e-07, "loss": 0.0337, "step": 45495 }, { "epoch": 2.425547795489684, "grad_norm": 3.7858190536499023, "learning_rate": 9.01e-07, "loss": 0.0126, "step": 45496 }, { "epoch": 2.425601108919337, "grad_norm": 3.2571494579315186, "learning_rate": 9.008e-07, "loss": 0.028, "step": 45497 }, { "epoch": 2.4256544223489898, "grad_norm": 0.4563162922859192, "learning_rate": 9.006000000000001e-07, "loss": 0.0034, "step": 45498 }, { "epoch": 2.4257077357786425, "grad_norm": 1.9704105854034424, "learning_rate": 9.004e-07, "loss": 0.0058, "step": 45499 }, { "epoch": 2.4257610492082957, "grad_norm": 1.3077081441879272, "learning_rate": 9.002000000000001e-07, "loss": 0.0057, "step": 45500 }, { "epoch": 2.4258143626379485, "grad_norm": 7.050397872924805, "learning_rate": 9.000000000000001e-07, "loss": 0.0264, "step": 45501 }, { "epoch": 2.4258676760676012, "grad_norm": 2.5051674842834473, "learning_rate": 8.998000000000002e-07, "loss": 0.0411, "step": 45502 }, { "epoch": 2.4259209894972544, "grad_norm": 3.3789806365966797, "learning_rate": 8.996e-07, "loss": 0.0229, "step": 45503 }, { "epoch": 2.425974302926907, "grad_norm": 2.246652841567993, "learning_rate": 8.994000000000001e-07, "loss": 0.01, "step": 45504 }, { "epoch": 2.4260276163565604, "grad_norm": 2.6362359523773193, "learning_rate": 8.992000000000001e-07, "loss": 0.0102, "step": 45505 }, { "epoch": 2.426080929786213, "grad_norm": 3.8260819911956787, "learning_rate": 8.99e-07, "loss": 0.0321, "step": 45506 }, { "epoch": 2.426134243215866, "grad_norm": 1.199911117553711, "learning_rate": 8.988000000000001e-07, "loss": 0.0064, "step": 45507 }, { "epoch": 2.426187556645519, "grad_norm": 1.7285892963409424, "learning_rate": 8.986000000000001e-07, "loss": 0.0174, "step": 45508 }, { "epoch": 2.426240870075172, "grad_norm": 2.1288809776306152, "learning_rate": 8.984000000000002e-07, "loss": 0.0046, "step": 45509 }, { "epoch": 2.426294183504825, "grad_norm": 2.5499587059020996, "learning_rate": 8.982e-07, "loss": 0.0304, "step": 45510 }, { "epoch": 2.426347496934478, "grad_norm": 1.0734654664993286, "learning_rate": 8.980000000000001e-07, "loss": 0.0204, "step": 45511 }, { "epoch": 2.4264008103641306, "grad_norm": 1.0761569738388062, "learning_rate": 8.978000000000001e-07, "loss": 0.0046, "step": 45512 }, { "epoch": 2.426454123793784, "grad_norm": 3.992555856704712, "learning_rate": 8.976000000000001e-07, "loss": 0.0345, "step": 45513 }, { "epoch": 2.4265074372234365, "grad_norm": 0.5653207302093506, "learning_rate": 8.974000000000001e-07, "loss": 0.0036, "step": 45514 }, { "epoch": 2.4265607506530893, "grad_norm": 2.6943464279174805, "learning_rate": 8.972000000000001e-07, "loss": 0.0294, "step": 45515 }, { "epoch": 2.4266140640827425, "grad_norm": 2.3045687675476074, "learning_rate": 8.97e-07, "loss": 0.0233, "step": 45516 }, { "epoch": 2.4266673775123953, "grad_norm": 2.2213099002838135, "learning_rate": 8.968e-07, "loss": 0.0237, "step": 45517 }, { "epoch": 2.4267206909420485, "grad_norm": 0.3743756115436554, "learning_rate": 8.966000000000001e-07, "loss": 0.0032, "step": 45518 }, { "epoch": 2.426774004371701, "grad_norm": 1.8785220384597778, "learning_rate": 8.964e-07, "loss": 0.0164, "step": 45519 }, { "epoch": 2.426827317801354, "grad_norm": 3.4601330757141113, "learning_rate": 8.962000000000001e-07, "loss": 0.0234, "step": 45520 }, { "epoch": 2.426880631231007, "grad_norm": 0.3789564073085785, "learning_rate": 8.960000000000001e-07, "loss": 0.0027, "step": 45521 }, { "epoch": 2.42693394466066, "grad_norm": 1.4175307750701904, "learning_rate": 8.958000000000002e-07, "loss": 0.0063, "step": 45522 }, { "epoch": 2.426987258090313, "grad_norm": 0.6766833662986755, "learning_rate": 8.956e-07, "loss": 0.0045, "step": 45523 }, { "epoch": 2.427040571519966, "grad_norm": 2.8555595874786377, "learning_rate": 8.954e-07, "loss": 0.0145, "step": 45524 }, { "epoch": 2.4270938849496186, "grad_norm": 2.6244826316833496, "learning_rate": 8.952000000000001e-07, "loss": 0.02, "step": 45525 }, { "epoch": 2.427147198379272, "grad_norm": 1.269573450088501, "learning_rate": 8.95e-07, "loss": 0.0057, "step": 45526 }, { "epoch": 2.4272005118089246, "grad_norm": 3.9244046211242676, "learning_rate": 8.948000000000001e-07, "loss": 0.0113, "step": 45527 }, { "epoch": 2.4272538252385774, "grad_norm": 1.1114009618759155, "learning_rate": 8.946000000000001e-07, "loss": 0.0153, "step": 45528 }, { "epoch": 2.4273071386682306, "grad_norm": 3.2003278732299805, "learning_rate": 8.944000000000002e-07, "loss": 0.0314, "step": 45529 }, { "epoch": 2.4273604520978833, "grad_norm": 1.7138350009918213, "learning_rate": 8.942e-07, "loss": 0.0067, "step": 45530 }, { "epoch": 2.4274137655275365, "grad_norm": 2.059972047805786, "learning_rate": 8.94e-07, "loss": 0.017, "step": 45531 }, { "epoch": 2.4274670789571893, "grad_norm": 2.168020248413086, "learning_rate": 8.938000000000001e-07, "loss": 0.0116, "step": 45532 }, { "epoch": 2.4275203923868425, "grad_norm": 2.6604418754577637, "learning_rate": 8.936e-07, "loss": 0.0422, "step": 45533 }, { "epoch": 2.4275737058164952, "grad_norm": 0.6731951236724854, "learning_rate": 8.934000000000001e-07, "loss": 0.0043, "step": 45534 }, { "epoch": 2.427627019246148, "grad_norm": 1.6503932476043701, "learning_rate": 8.932000000000001e-07, "loss": 0.0246, "step": 45535 }, { "epoch": 2.427680332675801, "grad_norm": 0.5112258195877075, "learning_rate": 8.930000000000001e-07, "loss": 0.0033, "step": 45536 }, { "epoch": 2.427733646105454, "grad_norm": 2.4119601249694824, "learning_rate": 8.928000000000001e-07, "loss": 0.0131, "step": 45537 }, { "epoch": 2.4277869595351067, "grad_norm": 0.9853209853172302, "learning_rate": 8.926000000000001e-07, "loss": 0.005, "step": 45538 }, { "epoch": 2.42784027296476, "grad_norm": 1.7705209255218506, "learning_rate": 8.924e-07, "loss": 0.0072, "step": 45539 }, { "epoch": 2.4278935863944127, "grad_norm": 1.448124885559082, "learning_rate": 8.922e-07, "loss": 0.0378, "step": 45540 }, { "epoch": 2.427946899824066, "grad_norm": 0.5377407073974609, "learning_rate": 8.920000000000001e-07, "loss": 0.0038, "step": 45541 }, { "epoch": 2.4280002132537186, "grad_norm": 2.1144487857818604, "learning_rate": 8.918e-07, "loss": 0.0081, "step": 45542 }, { "epoch": 2.4280535266833714, "grad_norm": 1.838454008102417, "learning_rate": 8.916000000000001e-07, "loss": 0.0529, "step": 45543 }, { "epoch": 2.4281068401130246, "grad_norm": 3.4813642501831055, "learning_rate": 8.914000000000001e-07, "loss": 0.0144, "step": 45544 }, { "epoch": 2.4281601535426773, "grad_norm": 1.8389992713928223, "learning_rate": 8.912000000000001e-07, "loss": 0.0192, "step": 45545 }, { "epoch": 2.4282134669723305, "grad_norm": 1.7458922863006592, "learning_rate": 8.91e-07, "loss": 0.0281, "step": 45546 }, { "epoch": 2.4282667804019833, "grad_norm": 2.013932466506958, "learning_rate": 8.908000000000001e-07, "loss": 0.0215, "step": 45547 }, { "epoch": 2.428320093831636, "grad_norm": 0.984100341796875, "learning_rate": 8.906000000000001e-07, "loss": 0.005, "step": 45548 }, { "epoch": 2.4283734072612893, "grad_norm": 0.3954363465309143, "learning_rate": 8.904e-07, "loss": 0.0032, "step": 45549 }, { "epoch": 2.428426720690942, "grad_norm": 2.65478777885437, "learning_rate": 8.902000000000001e-07, "loss": 0.0208, "step": 45550 }, { "epoch": 2.4284800341205948, "grad_norm": 4.0119948387146, "learning_rate": 8.900000000000001e-07, "loss": 0.0218, "step": 45551 }, { "epoch": 2.428533347550248, "grad_norm": 1.2844728231430054, "learning_rate": 8.898000000000001e-07, "loss": 0.0138, "step": 45552 }, { "epoch": 2.4285866609799007, "grad_norm": 1.3800878524780273, "learning_rate": 8.896e-07, "loss": 0.0168, "step": 45553 }, { "epoch": 2.428639974409554, "grad_norm": 1.6136740446090698, "learning_rate": 8.894000000000001e-07, "loss": 0.0087, "step": 45554 }, { "epoch": 2.4286932878392067, "grad_norm": 2.131594657897949, "learning_rate": 8.892000000000001e-07, "loss": 0.0106, "step": 45555 }, { "epoch": 2.4287466012688594, "grad_norm": 1.192731261253357, "learning_rate": 8.890000000000002e-07, "loss": 0.0056, "step": 45556 }, { "epoch": 2.4287999146985126, "grad_norm": 2.071744441986084, "learning_rate": 8.888000000000001e-07, "loss": 0.0264, "step": 45557 }, { "epoch": 2.4288532281281654, "grad_norm": 0.4443241357803345, "learning_rate": 8.886000000000001e-07, "loss": 0.0025, "step": 45558 }, { "epoch": 2.4289065415578186, "grad_norm": 1.1145141124725342, "learning_rate": 8.884000000000001e-07, "loss": 0.0132, "step": 45559 }, { "epoch": 2.4289598549874714, "grad_norm": 2.3338727951049805, "learning_rate": 8.882e-07, "loss": 0.0077, "step": 45560 }, { "epoch": 2.429013168417124, "grad_norm": 3.3833281993865967, "learning_rate": 8.880000000000001e-07, "loss": 0.0344, "step": 45561 }, { "epoch": 2.4290664818467773, "grad_norm": 3.72332501411438, "learning_rate": 8.878000000000001e-07, "loss": 0.0226, "step": 45562 }, { "epoch": 2.42911979527643, "grad_norm": 1.0472909212112427, "learning_rate": 8.876000000000001e-07, "loss": 0.0046, "step": 45563 }, { "epoch": 2.429173108706083, "grad_norm": 1.4711664915084839, "learning_rate": 8.874000000000001e-07, "loss": 0.0077, "step": 45564 }, { "epoch": 2.429226422135736, "grad_norm": 1.9747047424316406, "learning_rate": 8.872000000000001e-07, "loss": 0.0147, "step": 45565 }, { "epoch": 2.429279735565389, "grad_norm": 1.0873382091522217, "learning_rate": 8.87e-07, "loss": 0.0065, "step": 45566 }, { "epoch": 2.429333048995042, "grad_norm": 1.7315870523452759, "learning_rate": 8.868e-07, "loss": 0.0066, "step": 45567 }, { "epoch": 2.4293863624246947, "grad_norm": 3.1324455738067627, "learning_rate": 8.866000000000001e-07, "loss": 0.0232, "step": 45568 }, { "epoch": 2.429439675854348, "grad_norm": 2.71431827545166, "learning_rate": 8.864e-07, "loss": 0.0131, "step": 45569 }, { "epoch": 2.4294929892840007, "grad_norm": 0.5972163081169128, "learning_rate": 8.862000000000001e-07, "loss": 0.0038, "step": 45570 }, { "epoch": 2.4295463027136535, "grad_norm": 1.2623919248580933, "learning_rate": 8.860000000000001e-07, "loss": 0.005, "step": 45571 }, { "epoch": 2.4295996161433067, "grad_norm": 0.8865551948547363, "learning_rate": 8.858000000000002e-07, "loss": 0.0039, "step": 45572 }, { "epoch": 2.4296529295729594, "grad_norm": 1.6388171911239624, "learning_rate": 8.856e-07, "loss": 0.0236, "step": 45573 }, { "epoch": 2.429706243002612, "grad_norm": 1.6599504947662354, "learning_rate": 8.854e-07, "loss": 0.0164, "step": 45574 }, { "epoch": 2.4297595564322654, "grad_norm": 0.9910685420036316, "learning_rate": 8.852000000000001e-07, "loss": 0.0048, "step": 45575 }, { "epoch": 2.429812869861918, "grad_norm": 3.666407346725464, "learning_rate": 8.85e-07, "loss": 0.0276, "step": 45576 }, { "epoch": 2.4298661832915713, "grad_norm": 1.8969712257385254, "learning_rate": 8.848000000000001e-07, "loss": 0.0252, "step": 45577 }, { "epoch": 2.429919496721224, "grad_norm": 1.805423378944397, "learning_rate": 8.846000000000001e-07, "loss": 0.0187, "step": 45578 }, { "epoch": 2.429972810150877, "grad_norm": 2.5411736965179443, "learning_rate": 8.844000000000002e-07, "loss": 0.016, "step": 45579 }, { "epoch": 2.43002612358053, "grad_norm": 0.7112393975257874, "learning_rate": 8.842e-07, "loss": 0.003, "step": 45580 }, { "epoch": 2.430079437010183, "grad_norm": 0.31451842188835144, "learning_rate": 8.840000000000001e-07, "loss": 0.0024, "step": 45581 }, { "epoch": 2.430132750439836, "grad_norm": 0.9513154029846191, "learning_rate": 8.838000000000001e-07, "loss": 0.0046, "step": 45582 }, { "epoch": 2.4301860638694888, "grad_norm": 2.7566614151000977, "learning_rate": 8.836e-07, "loss": 0.0285, "step": 45583 }, { "epoch": 2.4302393772991415, "grad_norm": 2.4077577590942383, "learning_rate": 8.834000000000001e-07, "loss": 0.0279, "step": 45584 }, { "epoch": 2.4302926907287947, "grad_norm": 2.9350485801696777, "learning_rate": 8.832000000000001e-07, "loss": 0.0431, "step": 45585 }, { "epoch": 2.4303460041584475, "grad_norm": 3.053807497024536, "learning_rate": 8.830000000000001e-07, "loss": 0.0178, "step": 45586 }, { "epoch": 2.4303993175881002, "grad_norm": 2.4128100872039795, "learning_rate": 8.828e-07, "loss": 0.015, "step": 45587 }, { "epoch": 2.4304526310177534, "grad_norm": 2.600651741027832, "learning_rate": 8.826000000000001e-07, "loss": 0.0271, "step": 45588 }, { "epoch": 2.430505944447406, "grad_norm": 1.8585411310195923, "learning_rate": 8.824e-07, "loss": 0.0246, "step": 45589 }, { "epoch": 2.4305592578770594, "grad_norm": 1.5218347311019897, "learning_rate": 8.822000000000001e-07, "loss": 0.0071, "step": 45590 }, { "epoch": 2.430612571306712, "grad_norm": 1.539328932762146, "learning_rate": 8.820000000000001e-07, "loss": 0.01, "step": 45591 }, { "epoch": 2.430665884736365, "grad_norm": 2.2627978324890137, "learning_rate": 8.818e-07, "loss": 0.0356, "step": 45592 }, { "epoch": 2.430719198166018, "grad_norm": 1.4488344192504883, "learning_rate": 8.816000000000001e-07, "loss": 0.0074, "step": 45593 }, { "epoch": 2.430772511595671, "grad_norm": 3.442343235015869, "learning_rate": 8.814e-07, "loss": 0.0173, "step": 45594 }, { "epoch": 2.430825825025324, "grad_norm": 2.6908209323883057, "learning_rate": 8.812000000000001e-07, "loss": 0.034, "step": 45595 }, { "epoch": 2.430879138454977, "grad_norm": 2.935591697692871, "learning_rate": 8.81e-07, "loss": 0.018, "step": 45596 }, { "epoch": 2.4309324518846296, "grad_norm": 1.1949864625930786, "learning_rate": 8.808000000000001e-07, "loss": 0.0215, "step": 45597 }, { "epoch": 2.430985765314283, "grad_norm": 3.7077713012695312, "learning_rate": 8.806000000000001e-07, "loss": 0.0221, "step": 45598 }, { "epoch": 2.4310390787439355, "grad_norm": 2.0544304847717285, "learning_rate": 8.804e-07, "loss": 0.0244, "step": 45599 }, { "epoch": 2.4310923921735883, "grad_norm": 1.8068140745162964, "learning_rate": 8.802000000000001e-07, "loss": 0.0241, "step": 45600 }, { "epoch": 2.4311457056032415, "grad_norm": 1.0784006118774414, "learning_rate": 8.8e-07, "loss": 0.005, "step": 45601 }, { "epoch": 2.4311990190328943, "grad_norm": 0.652929961681366, "learning_rate": 8.798000000000001e-07, "loss": 0.0037, "step": 45602 }, { "epoch": 2.4312523324625475, "grad_norm": 1.4021360874176025, "learning_rate": 8.796e-07, "loss": 0.0069, "step": 45603 }, { "epoch": 2.4313056458922, "grad_norm": 0.7819417119026184, "learning_rate": 8.794000000000001e-07, "loss": 0.0035, "step": 45604 }, { "epoch": 2.4313589593218534, "grad_norm": 1.3785836696624756, "learning_rate": 8.792000000000001e-07, "loss": 0.0195, "step": 45605 }, { "epoch": 2.431412272751506, "grad_norm": 2.310702085494995, "learning_rate": 8.790000000000002e-07, "loss": 0.0239, "step": 45606 }, { "epoch": 2.431465586181159, "grad_norm": 0.5849858522415161, "learning_rate": 8.788000000000001e-07, "loss": 0.0031, "step": 45607 }, { "epoch": 2.431518899610812, "grad_norm": 0.5816038250923157, "learning_rate": 8.786e-07, "loss": 0.0029, "step": 45608 }, { "epoch": 2.431572213040465, "grad_norm": 3.146510124206543, "learning_rate": 8.784000000000001e-07, "loss": 0.0248, "step": 45609 }, { "epoch": 2.4316255264701176, "grad_norm": 0.5779361724853516, "learning_rate": 8.782e-07, "loss": 0.0036, "step": 45610 }, { "epoch": 2.431678839899771, "grad_norm": 1.891624927520752, "learning_rate": 8.780000000000001e-07, "loss": 0.005, "step": 45611 }, { "epoch": 2.4317321533294236, "grad_norm": 1.0070531368255615, "learning_rate": 8.778000000000001e-07, "loss": 0.0041, "step": 45612 }, { "epoch": 2.4317854667590764, "grad_norm": 1.4644593000411987, "learning_rate": 8.776000000000001e-07, "loss": 0.0085, "step": 45613 }, { "epoch": 2.4318387801887296, "grad_norm": 1.4153989553451538, "learning_rate": 8.774000000000001e-07, "loss": 0.0071, "step": 45614 }, { "epoch": 2.4318920936183823, "grad_norm": 0.17301088571548462, "learning_rate": 8.772000000000002e-07, "loss": 0.0018, "step": 45615 }, { "epoch": 2.4319454070480355, "grad_norm": 2.631692886352539, "learning_rate": 8.77e-07, "loss": 0.0308, "step": 45616 }, { "epoch": 2.4319987204776883, "grad_norm": 1.49496591091156, "learning_rate": 8.768e-07, "loss": 0.0065, "step": 45617 }, { "epoch": 2.4320520339073415, "grad_norm": 0.5670074224472046, "learning_rate": 8.766000000000001e-07, "loss": 0.0045, "step": 45618 }, { "epoch": 2.4321053473369942, "grad_norm": 0.5387188196182251, "learning_rate": 8.764e-07, "loss": 0.0031, "step": 45619 }, { "epoch": 2.432158660766647, "grad_norm": 1.4469484090805054, "learning_rate": 8.762000000000001e-07, "loss": 0.0416, "step": 45620 }, { "epoch": 2.4322119741963, "grad_norm": 2.4616148471832275, "learning_rate": 8.760000000000001e-07, "loss": 0.0397, "step": 45621 }, { "epoch": 2.432265287625953, "grad_norm": 4.010857105255127, "learning_rate": 8.758000000000002e-07, "loss": 0.03, "step": 45622 }, { "epoch": 2.4323186010556057, "grad_norm": 1.3786216974258423, "learning_rate": 8.756e-07, "loss": 0.0118, "step": 45623 }, { "epoch": 2.432371914485259, "grad_norm": 3.555722713470459, "learning_rate": 8.754000000000001e-07, "loss": 0.0165, "step": 45624 }, { "epoch": 2.4324252279149117, "grad_norm": 2.892925262451172, "learning_rate": 8.752000000000001e-07, "loss": 0.0249, "step": 45625 }, { "epoch": 2.432478541344565, "grad_norm": 2.781036615371704, "learning_rate": 8.75e-07, "loss": 0.0135, "step": 45626 }, { "epoch": 2.4325318547742176, "grad_norm": 2.8304266929626465, "learning_rate": 8.748000000000001e-07, "loss": 0.0339, "step": 45627 }, { "epoch": 2.4325851682038704, "grad_norm": 3.3553082942962646, "learning_rate": 8.746000000000001e-07, "loss": 0.0166, "step": 45628 }, { "epoch": 2.4326384816335236, "grad_norm": 0.8618316054344177, "learning_rate": 8.744000000000002e-07, "loss": 0.0038, "step": 45629 }, { "epoch": 2.4326917950631763, "grad_norm": 1.5750616788864136, "learning_rate": 8.742e-07, "loss": 0.0176, "step": 45630 }, { "epoch": 2.4327451084928295, "grad_norm": 1.4141173362731934, "learning_rate": 8.740000000000001e-07, "loss": 0.0091, "step": 45631 }, { "epoch": 2.4327984219224823, "grad_norm": 1.3236390352249146, "learning_rate": 8.738000000000001e-07, "loss": 0.0059, "step": 45632 }, { "epoch": 2.432851735352135, "grad_norm": 1.3311347961425781, "learning_rate": 8.736e-07, "loss": 0.0045, "step": 45633 }, { "epoch": 2.4329050487817883, "grad_norm": 3.6549060344696045, "learning_rate": 8.734000000000001e-07, "loss": 0.027, "step": 45634 }, { "epoch": 2.432958362211441, "grad_norm": 2.6109068393707275, "learning_rate": 8.732000000000001e-07, "loss": 0.0168, "step": 45635 }, { "epoch": 2.4330116756410938, "grad_norm": 2.287137508392334, "learning_rate": 8.73e-07, "loss": 0.0113, "step": 45636 }, { "epoch": 2.433064989070747, "grad_norm": 0.7772205471992493, "learning_rate": 8.728e-07, "loss": 0.0039, "step": 45637 }, { "epoch": 2.4331183025003997, "grad_norm": 2.120922327041626, "learning_rate": 8.726000000000001e-07, "loss": 0.0335, "step": 45638 }, { "epoch": 2.433171615930053, "grad_norm": 1.6896265745162964, "learning_rate": 8.724e-07, "loss": 0.0072, "step": 45639 }, { "epoch": 2.4332249293597057, "grad_norm": 2.5099101066589355, "learning_rate": 8.722000000000001e-07, "loss": 0.0426, "step": 45640 }, { "epoch": 2.4332782427893584, "grad_norm": 0.2255651354789734, "learning_rate": 8.720000000000001e-07, "loss": 0.0018, "step": 45641 }, { "epoch": 2.4333315562190116, "grad_norm": 2.663499593734741, "learning_rate": 8.718e-07, "loss": 0.0285, "step": 45642 }, { "epoch": 2.4333848696486644, "grad_norm": 2.670248031616211, "learning_rate": 8.716e-07, "loss": 0.0184, "step": 45643 }, { "epoch": 2.4334381830783176, "grad_norm": 2.4357526302337646, "learning_rate": 8.714e-07, "loss": 0.013, "step": 45644 }, { "epoch": 2.4334914965079704, "grad_norm": 3.209324598312378, "learning_rate": 8.712000000000001e-07, "loss": 0.0163, "step": 45645 }, { "epoch": 2.433544809937623, "grad_norm": 0.48515111207962036, "learning_rate": 8.71e-07, "loss": 0.0026, "step": 45646 }, { "epoch": 2.4335981233672763, "grad_norm": 1.9469691514968872, "learning_rate": 8.708000000000001e-07, "loss": 0.0073, "step": 45647 }, { "epoch": 2.433651436796929, "grad_norm": 4.548505783081055, "learning_rate": 8.706000000000001e-07, "loss": 0.0182, "step": 45648 }, { "epoch": 2.433704750226582, "grad_norm": 3.005986452102661, "learning_rate": 8.704000000000002e-07, "loss": 0.0105, "step": 45649 }, { "epoch": 2.433758063656235, "grad_norm": 1.628517985343933, "learning_rate": 8.702e-07, "loss": 0.0073, "step": 45650 }, { "epoch": 2.433811377085888, "grad_norm": 1.3136347532272339, "learning_rate": 8.7e-07, "loss": 0.005, "step": 45651 }, { "epoch": 2.433864690515541, "grad_norm": 2.431609630584717, "learning_rate": 8.698000000000001e-07, "loss": 0.0093, "step": 45652 }, { "epoch": 2.4339180039451938, "grad_norm": 1.9906525611877441, "learning_rate": 8.696e-07, "loss": 0.0077, "step": 45653 }, { "epoch": 2.433971317374847, "grad_norm": 2.2654497623443604, "learning_rate": 8.694000000000001e-07, "loss": 0.0095, "step": 45654 }, { "epoch": 2.4340246308044997, "grad_norm": 2.2888829708099365, "learning_rate": 8.692000000000001e-07, "loss": 0.0083, "step": 45655 }, { "epoch": 2.4340779442341525, "grad_norm": 3.152245044708252, "learning_rate": 8.690000000000002e-07, "loss": 0.0264, "step": 45656 }, { "epoch": 2.4341312576638057, "grad_norm": 2.0316922664642334, "learning_rate": 8.688e-07, "loss": 0.0161, "step": 45657 }, { "epoch": 2.4341845710934584, "grad_norm": 3.986025333404541, "learning_rate": 8.686000000000001e-07, "loss": 0.0538, "step": 45658 }, { "epoch": 2.434237884523111, "grad_norm": 2.2391517162323, "learning_rate": 8.684000000000001e-07, "loss": 0.0261, "step": 45659 }, { "epoch": 2.4342911979527644, "grad_norm": 1.52866792678833, "learning_rate": 8.682e-07, "loss": 0.0079, "step": 45660 }, { "epoch": 2.434344511382417, "grad_norm": 1.2065964937210083, "learning_rate": 8.680000000000001e-07, "loss": 0.013, "step": 45661 }, { "epoch": 2.4343978248120703, "grad_norm": 0.41838526725769043, "learning_rate": 8.678000000000001e-07, "loss": 0.0027, "step": 45662 }, { "epoch": 2.434451138241723, "grad_norm": 2.557220220565796, "learning_rate": 8.676000000000001e-07, "loss": 0.0244, "step": 45663 }, { "epoch": 2.434504451671376, "grad_norm": 1.974205493927002, "learning_rate": 8.674e-07, "loss": 0.0268, "step": 45664 }, { "epoch": 2.434557765101029, "grad_norm": 0.6942281723022461, "learning_rate": 8.672000000000001e-07, "loss": 0.0054, "step": 45665 }, { "epoch": 2.434611078530682, "grad_norm": 1.4157733917236328, "learning_rate": 8.67e-07, "loss": 0.006, "step": 45666 }, { "epoch": 2.434664391960335, "grad_norm": 1.1212904453277588, "learning_rate": 8.668e-07, "loss": 0.0045, "step": 45667 }, { "epoch": 2.4347177053899878, "grad_norm": 2.360158920288086, "learning_rate": 8.666000000000001e-07, "loss": 0.0216, "step": 45668 }, { "epoch": 2.4347710188196405, "grad_norm": 1.264405608177185, "learning_rate": 8.664e-07, "loss": 0.007, "step": 45669 }, { "epoch": 2.4348243322492937, "grad_norm": 3.6006131172180176, "learning_rate": 8.662000000000001e-07, "loss": 0.0213, "step": 45670 }, { "epoch": 2.4348776456789465, "grad_norm": 1.0325742959976196, "learning_rate": 8.66e-07, "loss": 0.0173, "step": 45671 }, { "epoch": 2.4349309591085992, "grad_norm": 1.1792606115341187, "learning_rate": 8.658000000000001e-07, "loss": 0.0051, "step": 45672 }, { "epoch": 2.4349842725382524, "grad_norm": 2.7106053829193115, "learning_rate": 8.656e-07, "loss": 0.0329, "step": 45673 }, { "epoch": 2.435037585967905, "grad_norm": 1.5988248586654663, "learning_rate": 8.654000000000001e-07, "loss": 0.0118, "step": 45674 }, { "epoch": 2.4350908993975584, "grad_norm": 3.2338404655456543, "learning_rate": 8.652000000000001e-07, "loss": 0.0126, "step": 45675 }, { "epoch": 2.435144212827211, "grad_norm": 2.852527618408203, "learning_rate": 8.65e-07, "loss": 0.009, "step": 45676 }, { "epoch": 2.435197526256864, "grad_norm": 1.5311825275421143, "learning_rate": 8.648000000000001e-07, "loss": 0.017, "step": 45677 }, { "epoch": 2.435250839686517, "grad_norm": 2.67724871635437, "learning_rate": 8.646e-07, "loss": 0.0202, "step": 45678 }, { "epoch": 2.43530415311617, "grad_norm": 2.4289121627807617, "learning_rate": 8.644000000000001e-07, "loss": 0.0113, "step": 45679 }, { "epoch": 2.435357466545823, "grad_norm": 0.671467125415802, "learning_rate": 8.642e-07, "loss": 0.0042, "step": 45680 }, { "epoch": 2.435410779975476, "grad_norm": 3.738840103149414, "learning_rate": 8.640000000000001e-07, "loss": 0.0132, "step": 45681 }, { "epoch": 2.4354640934051286, "grad_norm": 2.6069259643554688, "learning_rate": 8.638000000000001e-07, "loss": 0.0037, "step": 45682 }, { "epoch": 2.435517406834782, "grad_norm": 0.8981006741523743, "learning_rate": 8.636000000000001e-07, "loss": 0.003, "step": 45683 }, { "epoch": 2.4355707202644346, "grad_norm": 1.944217562675476, "learning_rate": 8.634000000000001e-07, "loss": 0.0302, "step": 45684 }, { "epoch": 2.4356240336940873, "grad_norm": 1.7518153190612793, "learning_rate": 8.632000000000001e-07, "loss": 0.0214, "step": 45685 }, { "epoch": 2.4356773471237405, "grad_norm": 2.9346675872802734, "learning_rate": 8.63e-07, "loss": 0.0278, "step": 45686 }, { "epoch": 2.4357306605533933, "grad_norm": 0.5109113454818726, "learning_rate": 8.628e-07, "loss": 0.0038, "step": 45687 }, { "epoch": 2.4357839739830465, "grad_norm": 2.5231010913848877, "learning_rate": 8.626000000000001e-07, "loss": 0.0176, "step": 45688 }, { "epoch": 2.4358372874126992, "grad_norm": 1.853053092956543, "learning_rate": 8.624e-07, "loss": 0.0081, "step": 45689 }, { "epoch": 2.4358906008423524, "grad_norm": 0.7333195805549622, "learning_rate": 8.622000000000001e-07, "loss": 0.0041, "step": 45690 }, { "epoch": 2.435943914272005, "grad_norm": 1.9812432527542114, "learning_rate": 8.620000000000001e-07, "loss": 0.0104, "step": 45691 }, { "epoch": 2.435997227701658, "grad_norm": 3.0611064434051514, "learning_rate": 8.618000000000002e-07, "loss": 0.0275, "step": 45692 }, { "epoch": 2.436050541131311, "grad_norm": 1.1093971729278564, "learning_rate": 8.616e-07, "loss": 0.0147, "step": 45693 }, { "epoch": 2.436103854560964, "grad_norm": 0.5095944404602051, "learning_rate": 8.614e-07, "loss": 0.0027, "step": 45694 }, { "epoch": 2.4361571679906167, "grad_norm": 1.131364107131958, "learning_rate": 8.612000000000001e-07, "loss": 0.0167, "step": 45695 }, { "epoch": 2.43621048142027, "grad_norm": 2.3846395015716553, "learning_rate": 8.61e-07, "loss": 0.0149, "step": 45696 }, { "epoch": 2.4362637948499226, "grad_norm": 0.9279190897941589, "learning_rate": 8.608000000000001e-07, "loss": 0.0033, "step": 45697 }, { "epoch": 2.436317108279576, "grad_norm": 2.4940309524536133, "learning_rate": 8.606000000000001e-07, "loss": 0.0144, "step": 45698 }, { "epoch": 2.4363704217092286, "grad_norm": 1.357147216796875, "learning_rate": 8.604000000000002e-07, "loss": 0.0158, "step": 45699 }, { "epoch": 2.4364237351388813, "grad_norm": 2.5666401386260986, "learning_rate": 8.602e-07, "loss": 0.0142, "step": 45700 }, { "epoch": 2.4364770485685345, "grad_norm": 3.078037738800049, "learning_rate": 8.6e-07, "loss": 0.014, "step": 45701 }, { "epoch": 2.4365303619981873, "grad_norm": 3.5319623947143555, "learning_rate": 8.598000000000001e-07, "loss": 0.0255, "step": 45702 }, { "epoch": 2.4365836754278405, "grad_norm": 2.0257809162139893, "learning_rate": 8.596e-07, "loss": 0.0237, "step": 45703 }, { "epoch": 2.4366369888574932, "grad_norm": 3.1166651248931885, "learning_rate": 8.594000000000001e-07, "loss": 0.0558, "step": 45704 }, { "epoch": 2.436690302287146, "grad_norm": 1.9421271085739136, "learning_rate": 8.592000000000001e-07, "loss": 0.0059, "step": 45705 }, { "epoch": 2.436743615716799, "grad_norm": 6.977597236633301, "learning_rate": 8.590000000000002e-07, "loss": 0.0149, "step": 45706 }, { "epoch": 2.436796929146452, "grad_norm": 1.2633248567581177, "learning_rate": 8.588e-07, "loss": 0.0076, "step": 45707 }, { "epoch": 2.4368502425761047, "grad_norm": 2.8754079341888428, "learning_rate": 8.586000000000001e-07, "loss": 0.0238, "step": 45708 }, { "epoch": 2.436903556005758, "grad_norm": 1.8350653648376465, "learning_rate": 8.584000000000001e-07, "loss": 0.0256, "step": 45709 }, { "epoch": 2.4369568694354107, "grad_norm": 2.481192111968994, "learning_rate": 8.582e-07, "loss": 0.018, "step": 45710 }, { "epoch": 2.437010182865064, "grad_norm": 0.6779799461364746, "learning_rate": 8.580000000000001e-07, "loss": 0.0035, "step": 45711 }, { "epoch": 2.4370634962947166, "grad_norm": 2.0935616493225098, "learning_rate": 8.578000000000001e-07, "loss": 0.0075, "step": 45712 }, { "epoch": 2.4371168097243694, "grad_norm": 2.8372578620910645, "learning_rate": 8.576000000000001e-07, "loss": 0.0158, "step": 45713 }, { "epoch": 2.4371701231540226, "grad_norm": 1.2900081872940063, "learning_rate": 8.574e-07, "loss": 0.0093, "step": 45714 }, { "epoch": 2.4372234365836754, "grad_norm": 0.6469518542289734, "learning_rate": 8.572000000000001e-07, "loss": 0.0031, "step": 45715 }, { "epoch": 2.4372767500133286, "grad_norm": 1.6094084978103638, "learning_rate": 8.57e-07, "loss": 0.01, "step": 45716 }, { "epoch": 2.4373300634429813, "grad_norm": 2.832017183303833, "learning_rate": 8.568000000000001e-07, "loss": 0.0115, "step": 45717 }, { "epoch": 2.437383376872634, "grad_norm": 1.2618095874786377, "learning_rate": 8.566000000000001e-07, "loss": 0.0116, "step": 45718 }, { "epoch": 2.4374366903022873, "grad_norm": 3.6255030632019043, "learning_rate": 8.564e-07, "loss": 0.0152, "step": 45719 }, { "epoch": 2.43749000373194, "grad_norm": 0.7516041398048401, "learning_rate": 8.562000000000001e-07, "loss": 0.0048, "step": 45720 }, { "epoch": 2.437543317161593, "grad_norm": 2.7030203342437744, "learning_rate": 8.56e-07, "loss": 0.0199, "step": 45721 }, { "epoch": 2.437596630591246, "grad_norm": 3.2751004695892334, "learning_rate": 8.558000000000001e-07, "loss": 0.0355, "step": 45722 }, { "epoch": 2.4376499440208987, "grad_norm": 2.7970387935638428, "learning_rate": 8.556e-07, "loss": 0.0131, "step": 45723 }, { "epoch": 2.437703257450552, "grad_norm": 2.605879783630371, "learning_rate": 8.554000000000001e-07, "loss": 0.0082, "step": 45724 }, { "epoch": 2.4377565708802047, "grad_norm": 4.454489231109619, "learning_rate": 8.552000000000001e-07, "loss": 0.0105, "step": 45725 }, { "epoch": 2.437809884309858, "grad_norm": 2.1698663234710693, "learning_rate": 8.550000000000002e-07, "loss": 0.0129, "step": 45726 }, { "epoch": 2.4378631977395107, "grad_norm": 1.130936861038208, "learning_rate": 8.548000000000001e-07, "loss": 0.0158, "step": 45727 }, { "epoch": 2.4379165111691634, "grad_norm": 0.5985724329948425, "learning_rate": 8.546e-07, "loss": 0.0043, "step": 45728 }, { "epoch": 2.4379698245988166, "grad_norm": 2.392385959625244, "learning_rate": 8.544000000000001e-07, "loss": 0.0096, "step": 45729 }, { "epoch": 2.4380231380284694, "grad_norm": 2.788344144821167, "learning_rate": 8.542e-07, "loss": 0.0229, "step": 45730 }, { "epoch": 2.438076451458122, "grad_norm": 2.752124309539795, "learning_rate": 8.540000000000001e-07, "loss": 0.0259, "step": 45731 }, { "epoch": 2.4381297648877753, "grad_norm": 1.6744518280029297, "learning_rate": 8.538000000000001e-07, "loss": 0.0203, "step": 45732 }, { "epoch": 2.438183078317428, "grad_norm": 1.0157325267791748, "learning_rate": 8.536000000000001e-07, "loss": 0.0045, "step": 45733 }, { "epoch": 2.438236391747081, "grad_norm": 1.8659402132034302, "learning_rate": 8.534000000000001e-07, "loss": 0.0058, "step": 45734 }, { "epoch": 2.438289705176734, "grad_norm": 2.450526237487793, "learning_rate": 8.532e-07, "loss": 0.0283, "step": 45735 }, { "epoch": 2.438343018606387, "grad_norm": 1.5029999017715454, "learning_rate": 8.53e-07, "loss": 0.0288, "step": 45736 }, { "epoch": 2.43839633203604, "grad_norm": 4.760225772857666, "learning_rate": 8.528e-07, "loss": 0.0414, "step": 45737 }, { "epoch": 2.4384496454656928, "grad_norm": 4.100423812866211, "learning_rate": 8.526000000000001e-07, "loss": 0.0284, "step": 45738 }, { "epoch": 2.438502958895346, "grad_norm": 1.1366877555847168, "learning_rate": 8.524e-07, "loss": 0.0153, "step": 45739 }, { "epoch": 2.4385562723249987, "grad_norm": 5.589506149291992, "learning_rate": 8.522000000000001e-07, "loss": 0.0423, "step": 45740 }, { "epoch": 2.4386095857546515, "grad_norm": 2.2850606441497803, "learning_rate": 8.520000000000001e-07, "loss": 0.0271, "step": 45741 }, { "epoch": 2.4386628991843047, "grad_norm": 2.515882730484009, "learning_rate": 8.518000000000001e-07, "loss": 0.0081, "step": 45742 }, { "epoch": 2.4387162126139574, "grad_norm": 1.9810794591903687, "learning_rate": 8.516e-07, "loss": 0.0104, "step": 45743 }, { "epoch": 2.43876952604361, "grad_norm": 2.565088987350464, "learning_rate": 8.514e-07, "loss": 0.0375, "step": 45744 }, { "epoch": 2.4388228394732634, "grad_norm": 2.4503190517425537, "learning_rate": 8.512000000000001e-07, "loss": 0.0126, "step": 45745 }, { "epoch": 2.438876152902916, "grad_norm": 0.8087418675422668, "learning_rate": 8.51e-07, "loss": 0.0038, "step": 45746 }, { "epoch": 2.4389294663325694, "grad_norm": 0.976557731628418, "learning_rate": 8.508000000000001e-07, "loss": 0.0041, "step": 45747 }, { "epoch": 2.438982779762222, "grad_norm": 3.7483909130096436, "learning_rate": 8.506000000000001e-07, "loss": 0.0139, "step": 45748 }, { "epoch": 2.439036093191875, "grad_norm": 1.8910448551177979, "learning_rate": 8.504000000000002e-07, "loss": 0.01, "step": 45749 }, { "epoch": 2.439089406621528, "grad_norm": 1.9222321510314941, "learning_rate": 8.502e-07, "loss": 0.0278, "step": 45750 }, { "epoch": 2.439142720051181, "grad_norm": 1.9263622760772705, "learning_rate": 8.500000000000001e-07, "loss": 0.0051, "step": 45751 }, { "epoch": 2.439196033480834, "grad_norm": 1.530660629272461, "learning_rate": 8.498000000000001e-07, "loss": 0.0086, "step": 45752 }, { "epoch": 2.439249346910487, "grad_norm": 1.6859924793243408, "learning_rate": 8.496e-07, "loss": 0.0075, "step": 45753 }, { "epoch": 2.4393026603401395, "grad_norm": 1.5181801319122314, "learning_rate": 8.494000000000001e-07, "loss": 0.007, "step": 45754 }, { "epoch": 2.4393559737697927, "grad_norm": 1.5598561763763428, "learning_rate": 8.492000000000001e-07, "loss": 0.0187, "step": 45755 }, { "epoch": 2.4394092871994455, "grad_norm": 5.673757076263428, "learning_rate": 8.490000000000002e-07, "loss": 0.0173, "step": 45756 }, { "epoch": 2.4394626006290983, "grad_norm": 2.761341094970703, "learning_rate": 8.488e-07, "loss": 0.0091, "step": 45757 }, { "epoch": 2.4395159140587515, "grad_norm": 2.8937411308288574, "learning_rate": 8.486000000000001e-07, "loss": 0.0089, "step": 45758 }, { "epoch": 2.439569227488404, "grad_norm": 1.6656075716018677, "learning_rate": 8.484000000000001e-07, "loss": 0.0152, "step": 45759 }, { "epoch": 2.4396225409180574, "grad_norm": 0.870725154876709, "learning_rate": 8.482000000000001e-07, "loss": 0.0029, "step": 45760 }, { "epoch": 2.43967585434771, "grad_norm": 0.8414081931114197, "learning_rate": 8.480000000000001e-07, "loss": 0.0166, "step": 45761 }, { "epoch": 2.439729167777363, "grad_norm": 2.921102285385132, "learning_rate": 8.478000000000001e-07, "loss": 0.0436, "step": 45762 }, { "epoch": 2.439782481207016, "grad_norm": 2.352405071258545, "learning_rate": 8.476e-07, "loss": 0.0367, "step": 45763 }, { "epoch": 2.439835794636669, "grad_norm": 1.8410240411758423, "learning_rate": 8.474e-07, "loss": 0.0175, "step": 45764 }, { "epoch": 2.439889108066322, "grad_norm": 0.21296744048595428, "learning_rate": 8.472000000000001e-07, "loss": 0.0019, "step": 45765 }, { "epoch": 2.439942421495975, "grad_norm": 0.5739901661872864, "learning_rate": 8.47e-07, "loss": 0.0026, "step": 45766 }, { "epoch": 2.4399957349256276, "grad_norm": 2.085891008377075, "learning_rate": 8.468000000000001e-07, "loss": 0.0092, "step": 45767 }, { "epoch": 2.440049048355281, "grad_norm": 0.9823064208030701, "learning_rate": 8.466000000000001e-07, "loss": 0.005, "step": 45768 }, { "epoch": 2.4401023617849336, "grad_norm": 1.1449600458145142, "learning_rate": 8.464000000000002e-07, "loss": 0.005, "step": 45769 }, { "epoch": 2.4401556752145863, "grad_norm": 2.8810739517211914, "learning_rate": 8.462e-07, "loss": 0.04, "step": 45770 }, { "epoch": 2.4402089886442395, "grad_norm": 3.6586270332336426, "learning_rate": 8.46e-07, "loss": 0.0158, "step": 45771 }, { "epoch": 2.4402623020738923, "grad_norm": 2.0240602493286133, "learning_rate": 8.458000000000001e-07, "loss": 0.0126, "step": 45772 }, { "epoch": 2.4403156155035455, "grad_norm": 1.8618419170379639, "learning_rate": 8.456e-07, "loss": 0.017, "step": 45773 }, { "epoch": 2.4403689289331982, "grad_norm": 1.8583430051803589, "learning_rate": 8.454000000000001e-07, "loss": 0.0254, "step": 45774 }, { "epoch": 2.4404222423628514, "grad_norm": 0.27137431502342224, "learning_rate": 8.452000000000001e-07, "loss": 0.0018, "step": 45775 }, { "epoch": 2.440475555792504, "grad_norm": 1.8310757875442505, "learning_rate": 8.450000000000002e-07, "loss": 0.0095, "step": 45776 }, { "epoch": 2.440528869222157, "grad_norm": 1.6984119415283203, "learning_rate": 8.448e-07, "loss": 0.0242, "step": 45777 }, { "epoch": 2.44058218265181, "grad_norm": 1.040451169013977, "learning_rate": 8.446e-07, "loss": 0.0147, "step": 45778 }, { "epoch": 2.440635496081463, "grad_norm": 1.927978515625, "learning_rate": 8.444000000000001e-07, "loss": 0.0169, "step": 45779 }, { "epoch": 2.4406888095111157, "grad_norm": 0.9009097814559937, "learning_rate": 8.442e-07, "loss": 0.0045, "step": 45780 }, { "epoch": 2.440742122940769, "grad_norm": 1.3101508617401123, "learning_rate": 8.440000000000001e-07, "loss": 0.0064, "step": 45781 }, { "epoch": 2.4407954363704216, "grad_norm": 4.227065563201904, "learning_rate": 8.438000000000001e-07, "loss": 0.0622, "step": 45782 }, { "epoch": 2.440848749800075, "grad_norm": 4.35919189453125, "learning_rate": 8.436000000000001e-07, "loss": 0.0181, "step": 45783 }, { "epoch": 2.4409020632297276, "grad_norm": 3.7097246646881104, "learning_rate": 8.434e-07, "loss": 0.0443, "step": 45784 }, { "epoch": 2.4409553766593803, "grad_norm": 4.106740474700928, "learning_rate": 8.432000000000001e-07, "loss": 0.0723, "step": 45785 }, { "epoch": 2.4410086900890335, "grad_norm": 0.34185758233070374, "learning_rate": 8.43e-07, "loss": 0.0026, "step": 45786 }, { "epoch": 2.4410620035186863, "grad_norm": 3.3467838764190674, "learning_rate": 8.428e-07, "loss": 0.0498, "step": 45787 }, { "epoch": 2.4411153169483395, "grad_norm": 2.7942872047424316, "learning_rate": 8.426000000000001e-07, "loss": 0.0242, "step": 45788 }, { "epoch": 2.4411686303779923, "grad_norm": 1.265904426574707, "learning_rate": 8.424e-07, "loss": 0.0052, "step": 45789 }, { "epoch": 2.441221943807645, "grad_norm": 1.9717020988464355, "learning_rate": 8.422000000000001e-07, "loss": 0.0153, "step": 45790 }, { "epoch": 2.441275257237298, "grad_norm": 1.4616330862045288, "learning_rate": 8.42e-07, "loss": 0.0085, "step": 45791 }, { "epoch": 2.441328570666951, "grad_norm": 1.620596170425415, "learning_rate": 8.418000000000001e-07, "loss": 0.0239, "step": 45792 }, { "epoch": 2.4413818840966037, "grad_norm": 2.255622625350952, "learning_rate": 8.416e-07, "loss": 0.0332, "step": 45793 }, { "epoch": 2.441435197526257, "grad_norm": 1.4069359302520752, "learning_rate": 8.414000000000001e-07, "loss": 0.0267, "step": 45794 }, { "epoch": 2.4414885109559097, "grad_norm": 0.3938119113445282, "learning_rate": 8.412000000000001e-07, "loss": 0.004, "step": 45795 }, { "epoch": 2.441541824385563, "grad_norm": 3.982816696166992, "learning_rate": 8.41e-07, "loss": 0.0152, "step": 45796 }, { "epoch": 2.4415951378152156, "grad_norm": 0.9756181240081787, "learning_rate": 8.408000000000001e-07, "loss": 0.004, "step": 45797 }, { "epoch": 2.4416484512448684, "grad_norm": 0.6635347604751587, "learning_rate": 8.406e-07, "loss": 0.0035, "step": 45798 }, { "epoch": 2.4417017646745216, "grad_norm": 2.1599771976470947, "learning_rate": 8.404000000000001e-07, "loss": 0.0091, "step": 45799 }, { "epoch": 2.4417550781041744, "grad_norm": 1.6546906232833862, "learning_rate": 8.402e-07, "loss": 0.0173, "step": 45800 }, { "epoch": 2.4418083915338276, "grad_norm": 1.912757396697998, "learning_rate": 8.400000000000001e-07, "loss": 0.013, "step": 45801 }, { "epoch": 2.4418617049634803, "grad_norm": 1.9939502477645874, "learning_rate": 8.398000000000001e-07, "loss": 0.0261, "step": 45802 }, { "epoch": 2.441915018393133, "grad_norm": 3.6887941360473633, "learning_rate": 8.396000000000002e-07, "loss": 0.0217, "step": 45803 }, { "epoch": 2.4419683318227863, "grad_norm": 2.356794595718384, "learning_rate": 8.394000000000001e-07, "loss": 0.0272, "step": 45804 }, { "epoch": 2.442021645252439, "grad_norm": 2.6085762977600098, "learning_rate": 8.392e-07, "loss": 0.0342, "step": 45805 }, { "epoch": 2.442074958682092, "grad_norm": 1.6105360984802246, "learning_rate": 8.390000000000001e-07, "loss": 0.0072, "step": 45806 }, { "epoch": 2.442128272111745, "grad_norm": 0.928422749042511, "learning_rate": 8.388e-07, "loss": 0.0052, "step": 45807 }, { "epoch": 2.4421815855413977, "grad_norm": 2.514598846435547, "learning_rate": 8.386000000000001e-07, "loss": 0.0188, "step": 45808 }, { "epoch": 2.442234898971051, "grad_norm": 2.9508628845214844, "learning_rate": 8.384000000000001e-07, "loss": 0.0079, "step": 45809 }, { "epoch": 2.4422882124007037, "grad_norm": 2.6754491329193115, "learning_rate": 8.382000000000001e-07, "loss": 0.0199, "step": 45810 }, { "epoch": 2.442341525830357, "grad_norm": 2.123779535293579, "learning_rate": 8.380000000000001e-07, "loss": 0.0401, "step": 45811 }, { "epoch": 2.4423948392600097, "grad_norm": 1.517686128616333, "learning_rate": 8.378e-07, "loss": 0.026, "step": 45812 }, { "epoch": 2.4424481526896624, "grad_norm": 1.810166835784912, "learning_rate": 8.376e-07, "loss": 0.0129, "step": 45813 }, { "epoch": 2.4425014661193156, "grad_norm": 1.2246588468551636, "learning_rate": 8.374e-07, "loss": 0.0056, "step": 45814 }, { "epoch": 2.4425547795489684, "grad_norm": 2.378750801086426, "learning_rate": 8.372000000000001e-07, "loss": 0.0121, "step": 45815 }, { "epoch": 2.442608092978621, "grad_norm": 2.063141345977783, "learning_rate": 8.37e-07, "loss": 0.0246, "step": 45816 }, { "epoch": 2.4426614064082743, "grad_norm": 2.3839659690856934, "learning_rate": 8.368000000000001e-07, "loss": 0.009, "step": 45817 }, { "epoch": 2.442714719837927, "grad_norm": 1.3834477663040161, "learning_rate": 8.366000000000001e-07, "loss": 0.0069, "step": 45818 }, { "epoch": 2.4427680332675803, "grad_norm": 2.380624771118164, "learning_rate": 8.364000000000002e-07, "loss": 0.0448, "step": 45819 }, { "epoch": 2.442821346697233, "grad_norm": 3.2643871307373047, "learning_rate": 8.362e-07, "loss": 0.0135, "step": 45820 }, { "epoch": 2.442874660126886, "grad_norm": 2.231419324874878, "learning_rate": 8.36e-07, "loss": 0.0075, "step": 45821 }, { "epoch": 2.442927973556539, "grad_norm": 1.3417352437973022, "learning_rate": 8.358000000000001e-07, "loss": 0.0051, "step": 45822 }, { "epoch": 2.4429812869861918, "grad_norm": 2.238607168197632, "learning_rate": 8.356e-07, "loss": 0.0251, "step": 45823 }, { "epoch": 2.443034600415845, "grad_norm": 2.881364583969116, "learning_rate": 8.354000000000001e-07, "loss": 0.0088, "step": 45824 }, { "epoch": 2.4430879138454977, "grad_norm": 1.36591637134552, "learning_rate": 8.352000000000001e-07, "loss": 0.0121, "step": 45825 }, { "epoch": 2.4431412272751505, "grad_norm": 1.4745763540267944, "learning_rate": 8.350000000000002e-07, "loss": 0.0069, "step": 45826 }, { "epoch": 2.4431945407048037, "grad_norm": 4.183753490447998, "learning_rate": 8.348e-07, "loss": 0.017, "step": 45827 }, { "epoch": 2.4432478541344564, "grad_norm": 2.5327157974243164, "learning_rate": 8.346000000000001e-07, "loss": 0.0187, "step": 45828 }, { "epoch": 2.443301167564109, "grad_norm": 2.1745400428771973, "learning_rate": 8.344000000000001e-07, "loss": 0.0185, "step": 45829 }, { "epoch": 2.4433544809937624, "grad_norm": 2.571502923965454, "learning_rate": 8.342e-07, "loss": 0.0174, "step": 45830 }, { "epoch": 2.443407794423415, "grad_norm": 2.228797435760498, "learning_rate": 8.340000000000001e-07, "loss": 0.0068, "step": 45831 }, { "epoch": 2.4434611078530684, "grad_norm": 2.2365851402282715, "learning_rate": 8.338000000000001e-07, "loss": 0.0156, "step": 45832 }, { "epoch": 2.443514421282721, "grad_norm": 0.5026811957359314, "learning_rate": 8.336e-07, "loss": 0.0031, "step": 45833 }, { "epoch": 2.443567734712374, "grad_norm": 3.3522632122039795, "learning_rate": 8.334e-07, "loss": 0.0475, "step": 45834 }, { "epoch": 2.443621048142027, "grad_norm": 4.536945819854736, "learning_rate": 8.332000000000001e-07, "loss": 0.0255, "step": 45835 }, { "epoch": 2.44367436157168, "grad_norm": 1.4198436737060547, "learning_rate": 8.33e-07, "loss": 0.0064, "step": 45836 }, { "epoch": 2.443727675001333, "grad_norm": 1.6016019582748413, "learning_rate": 8.328000000000001e-07, "loss": 0.0055, "step": 45837 }, { "epoch": 2.443780988430986, "grad_norm": 2.3238134384155273, "learning_rate": 8.326000000000001e-07, "loss": 0.0372, "step": 45838 }, { "epoch": 2.4438343018606385, "grad_norm": 2.64481782913208, "learning_rate": 8.324e-07, "loss": 0.0333, "step": 45839 }, { "epoch": 2.4438876152902917, "grad_norm": 2.2944560050964355, "learning_rate": 8.322000000000001e-07, "loss": 0.0105, "step": 45840 }, { "epoch": 2.4439409287199445, "grad_norm": 1.5297267436981201, "learning_rate": 8.32e-07, "loss": 0.0097, "step": 45841 }, { "epoch": 2.4439942421495973, "grad_norm": 0.4753509759902954, "learning_rate": 8.318000000000001e-07, "loss": 0.0036, "step": 45842 }, { "epoch": 2.4440475555792505, "grad_norm": 2.0349557399749756, "learning_rate": 8.316e-07, "loss": 0.0218, "step": 45843 }, { "epoch": 2.444100869008903, "grad_norm": 0.3038578927516937, "learning_rate": 8.314000000000001e-07, "loss": 0.0024, "step": 45844 }, { "epoch": 2.4441541824385564, "grad_norm": 1.3233023881912231, "learning_rate": 8.312000000000001e-07, "loss": 0.0057, "step": 45845 }, { "epoch": 2.444207495868209, "grad_norm": 0.587500810623169, "learning_rate": 8.31e-07, "loss": 0.0036, "step": 45846 }, { "epoch": 2.4442608092978624, "grad_norm": 3.179603338241577, "learning_rate": 8.308000000000001e-07, "loss": 0.022, "step": 45847 }, { "epoch": 2.444314122727515, "grad_norm": 1.2256391048431396, "learning_rate": 8.306e-07, "loss": 0.0296, "step": 45848 }, { "epoch": 2.444367436157168, "grad_norm": 2.601545810699463, "learning_rate": 8.304000000000001e-07, "loss": 0.0237, "step": 45849 }, { "epoch": 2.444420749586821, "grad_norm": 1.8663921356201172, "learning_rate": 8.302e-07, "loss": 0.0313, "step": 45850 }, { "epoch": 2.444474063016474, "grad_norm": 3.7889034748077393, "learning_rate": 8.300000000000001e-07, "loss": 0.0357, "step": 45851 }, { "epoch": 2.4445273764461266, "grad_norm": 2.1855201721191406, "learning_rate": 8.298000000000001e-07, "loss": 0.0299, "step": 45852 }, { "epoch": 2.44458068987578, "grad_norm": 1.050491452217102, "learning_rate": 8.296000000000002e-07, "loss": 0.0073, "step": 45853 }, { "epoch": 2.4446340033054326, "grad_norm": 2.4766061305999756, "learning_rate": 8.294000000000001e-07, "loss": 0.0128, "step": 45854 }, { "epoch": 2.4446873167350853, "grad_norm": 1.910925030708313, "learning_rate": 8.292e-07, "loss": 0.0184, "step": 45855 }, { "epoch": 2.4447406301647385, "grad_norm": 0.7776221036911011, "learning_rate": 8.290000000000001e-07, "loss": 0.0044, "step": 45856 }, { "epoch": 2.4447939435943913, "grad_norm": 1.545714020729065, "learning_rate": 8.288e-07, "loss": 0.0306, "step": 45857 }, { "epoch": 2.4448472570240445, "grad_norm": 1.790840983390808, "learning_rate": 8.286000000000001e-07, "loss": 0.0067, "step": 45858 }, { "epoch": 2.4449005704536972, "grad_norm": 2.220198631286621, "learning_rate": 8.284000000000001e-07, "loss": 0.0129, "step": 45859 }, { "epoch": 2.4449538838833504, "grad_norm": 4.312711715698242, "learning_rate": 8.282000000000001e-07, "loss": 0.0183, "step": 45860 }, { "epoch": 2.445007197313003, "grad_norm": 2.3100287914276123, "learning_rate": 8.280000000000001e-07, "loss": 0.0422, "step": 45861 }, { "epoch": 2.445060510742656, "grad_norm": 2.3681445121765137, "learning_rate": 8.278000000000001e-07, "loss": 0.0108, "step": 45862 }, { "epoch": 2.445113824172309, "grad_norm": 0.9297805428504944, "learning_rate": 8.276e-07, "loss": 0.0171, "step": 45863 }, { "epoch": 2.445167137601962, "grad_norm": 0.42489027976989746, "learning_rate": 8.274e-07, "loss": 0.0031, "step": 45864 }, { "epoch": 2.4452204510316147, "grad_norm": 1.680010199546814, "learning_rate": 8.272000000000001e-07, "loss": 0.0082, "step": 45865 }, { "epoch": 2.445273764461268, "grad_norm": 0.9186164140701294, "learning_rate": 8.27e-07, "loss": 0.0043, "step": 45866 }, { "epoch": 2.4453270778909206, "grad_norm": 0.8789371848106384, "learning_rate": 8.268000000000001e-07, "loss": 0.0048, "step": 45867 }, { "epoch": 2.445380391320574, "grad_norm": 1.2310558557510376, "learning_rate": 8.266000000000001e-07, "loss": 0.0087, "step": 45868 }, { "epoch": 2.4454337047502266, "grad_norm": 0.2600707709789276, "learning_rate": 8.264000000000001e-07, "loss": 0.0023, "step": 45869 }, { "epoch": 2.4454870181798793, "grad_norm": 0.9253358840942383, "learning_rate": 8.262e-07, "loss": 0.0039, "step": 45870 }, { "epoch": 2.4455403316095325, "grad_norm": 0.6764793395996094, "learning_rate": 8.260000000000001e-07, "loss": 0.0044, "step": 45871 }, { "epoch": 2.4455936450391853, "grad_norm": 2.891324996948242, "learning_rate": 8.258000000000001e-07, "loss": 0.0139, "step": 45872 }, { "epoch": 2.4456469584688385, "grad_norm": 2.0726981163024902, "learning_rate": 8.256e-07, "loss": 0.011, "step": 45873 }, { "epoch": 2.4457002718984913, "grad_norm": 1.9099464416503906, "learning_rate": 8.254000000000001e-07, "loss": 0.0145, "step": 45874 }, { "epoch": 2.445753585328144, "grad_norm": 0.44384920597076416, "learning_rate": 8.252000000000001e-07, "loss": 0.0027, "step": 45875 }, { "epoch": 2.445806898757797, "grad_norm": 1.664335012435913, "learning_rate": 8.250000000000001e-07, "loss": 0.0054, "step": 45876 }, { "epoch": 2.44586021218745, "grad_norm": 3.630988597869873, "learning_rate": 8.248e-07, "loss": 0.027, "step": 45877 }, { "epoch": 2.4459135256171027, "grad_norm": 1.6246075630187988, "learning_rate": 8.246000000000001e-07, "loss": 0.018, "step": 45878 }, { "epoch": 2.445966839046756, "grad_norm": 2.4367451667785645, "learning_rate": 8.244000000000001e-07, "loss": 0.0131, "step": 45879 }, { "epoch": 2.4460201524764087, "grad_norm": 2.3853793144226074, "learning_rate": 8.242e-07, "loss": 0.0118, "step": 45880 }, { "epoch": 2.446073465906062, "grad_norm": 1.5300437211990356, "learning_rate": 8.240000000000001e-07, "loss": 0.0079, "step": 45881 }, { "epoch": 2.4461267793357147, "grad_norm": 1.3705992698669434, "learning_rate": 8.238000000000001e-07, "loss": 0.0043, "step": 45882 }, { "epoch": 2.4461800927653674, "grad_norm": 3.1236155033111572, "learning_rate": 8.236e-07, "loss": 0.0357, "step": 45883 }, { "epoch": 2.4462334061950206, "grad_norm": 0.9818372130393982, "learning_rate": 8.234e-07, "loss": 0.0051, "step": 45884 }, { "epoch": 2.4462867196246734, "grad_norm": 1.3918377161026, "learning_rate": 8.232000000000001e-07, "loss": 0.0059, "step": 45885 }, { "epoch": 2.4463400330543266, "grad_norm": 3.170189380645752, "learning_rate": 8.23e-07, "loss": 0.0227, "step": 45886 }, { "epoch": 2.4463933464839793, "grad_norm": 1.4594696760177612, "learning_rate": 8.228000000000001e-07, "loss": 0.0257, "step": 45887 }, { "epoch": 2.446446659913632, "grad_norm": 0.9981916546821594, "learning_rate": 8.226000000000001e-07, "loss": 0.023, "step": 45888 }, { "epoch": 2.4464999733432853, "grad_norm": 0.3908741772174835, "learning_rate": 8.224e-07, "loss": 0.0028, "step": 45889 }, { "epoch": 2.446553286772938, "grad_norm": 2.592923879623413, "learning_rate": 8.222e-07, "loss": 0.0141, "step": 45890 }, { "epoch": 2.446606600202591, "grad_norm": 1.4227018356323242, "learning_rate": 8.22e-07, "loss": 0.0099, "step": 45891 }, { "epoch": 2.446659913632244, "grad_norm": 1.7426928281784058, "learning_rate": 8.218000000000001e-07, "loss": 0.0336, "step": 45892 }, { "epoch": 2.4467132270618968, "grad_norm": 1.8387582302093506, "learning_rate": 8.216e-07, "loss": 0.0143, "step": 45893 }, { "epoch": 2.44676654049155, "grad_norm": 2.253801107406616, "learning_rate": 8.214000000000001e-07, "loss": 0.0162, "step": 45894 }, { "epoch": 2.4468198539212027, "grad_norm": 1.3400338888168335, "learning_rate": 8.212000000000001e-07, "loss": 0.0133, "step": 45895 }, { "epoch": 2.446873167350856, "grad_norm": 2.151719093322754, "learning_rate": 8.210000000000002e-07, "loss": 0.018, "step": 45896 }, { "epoch": 2.4469264807805087, "grad_norm": 2.024937629699707, "learning_rate": 8.208e-07, "loss": 0.0157, "step": 45897 }, { "epoch": 2.4469797942101614, "grad_norm": 1.4744352102279663, "learning_rate": 8.206e-07, "loss": 0.0069, "step": 45898 }, { "epoch": 2.4470331076398146, "grad_norm": 1.5753358602523804, "learning_rate": 8.204000000000001e-07, "loss": 0.0135, "step": 45899 }, { "epoch": 2.4470864210694674, "grad_norm": 1.6540350914001465, "learning_rate": 8.202e-07, "loss": 0.0142, "step": 45900 }, { "epoch": 2.44713973449912, "grad_norm": 0.9150387644767761, "learning_rate": 8.200000000000001e-07, "loss": 0.0035, "step": 45901 }, { "epoch": 2.4471930479287733, "grad_norm": 0.7040997743606567, "learning_rate": 8.198000000000001e-07, "loss": 0.0046, "step": 45902 }, { "epoch": 2.447246361358426, "grad_norm": 1.170931100845337, "learning_rate": 8.196000000000002e-07, "loss": 0.0049, "step": 45903 }, { "epoch": 2.4472996747880793, "grad_norm": 1.247614860534668, "learning_rate": 8.194e-07, "loss": 0.0099, "step": 45904 }, { "epoch": 2.447352988217732, "grad_norm": 1.2481979131698608, "learning_rate": 8.192000000000001e-07, "loss": 0.019, "step": 45905 }, { "epoch": 2.447406301647385, "grad_norm": 1.9313721656799316, "learning_rate": 8.190000000000001e-07, "loss": 0.0265, "step": 45906 }, { "epoch": 2.447459615077038, "grad_norm": 4.017854690551758, "learning_rate": 8.188e-07, "loss": 0.0271, "step": 45907 }, { "epoch": 2.4475129285066908, "grad_norm": 1.4752459526062012, "learning_rate": 8.186000000000001e-07, "loss": 0.0144, "step": 45908 }, { "epoch": 2.447566241936344, "grad_norm": 2.562509775161743, "learning_rate": 8.184000000000001e-07, "loss": 0.0307, "step": 45909 }, { "epoch": 2.4476195553659967, "grad_norm": 1.743679404258728, "learning_rate": 8.182000000000001e-07, "loss": 0.0104, "step": 45910 }, { "epoch": 2.4476728687956495, "grad_norm": 1.3904818296432495, "learning_rate": 8.18e-07, "loss": 0.0246, "step": 45911 }, { "epoch": 2.4477261822253027, "grad_norm": 1.301944375038147, "learning_rate": 8.178000000000001e-07, "loss": 0.0192, "step": 45912 }, { "epoch": 2.4477794956549555, "grad_norm": 2.724595069885254, "learning_rate": 8.176e-07, "loss": 0.0173, "step": 45913 }, { "epoch": 2.447832809084608, "grad_norm": 2.372849941253662, "learning_rate": 8.174e-07, "loss": 0.0276, "step": 45914 }, { "epoch": 2.4478861225142614, "grad_norm": 1.7122877836227417, "learning_rate": 8.172000000000001e-07, "loss": 0.008, "step": 45915 }, { "epoch": 2.447939435943914, "grad_norm": 2.7458009719848633, "learning_rate": 8.17e-07, "loss": 0.0289, "step": 45916 }, { "epoch": 2.4479927493735674, "grad_norm": 1.2880219221115112, "learning_rate": 8.168000000000001e-07, "loss": 0.0236, "step": 45917 }, { "epoch": 2.44804606280322, "grad_norm": 1.9011343717575073, "learning_rate": 8.166e-07, "loss": 0.006, "step": 45918 }, { "epoch": 2.448099376232873, "grad_norm": 1.2663248777389526, "learning_rate": 8.164000000000001e-07, "loss": 0.0064, "step": 45919 }, { "epoch": 2.448152689662526, "grad_norm": 2.608316421508789, "learning_rate": 8.162e-07, "loss": 0.0201, "step": 45920 }, { "epoch": 2.448206003092179, "grad_norm": 0.9906054735183716, "learning_rate": 8.160000000000001e-07, "loss": 0.0068, "step": 45921 }, { "epoch": 2.448259316521832, "grad_norm": 2.1328084468841553, "learning_rate": 8.158000000000001e-07, "loss": 0.0079, "step": 45922 }, { "epoch": 2.448312629951485, "grad_norm": 1.4167811870574951, "learning_rate": 8.156e-07, "loss": 0.007, "step": 45923 }, { "epoch": 2.4483659433811376, "grad_norm": 2.457643747329712, "learning_rate": 8.154000000000001e-07, "loss": 0.0134, "step": 45924 }, { "epoch": 2.4484192568107908, "grad_norm": 1.6190919876098633, "learning_rate": 8.152e-07, "loss": 0.0121, "step": 45925 }, { "epoch": 2.4484725702404435, "grad_norm": 0.47563982009887695, "learning_rate": 8.150000000000001e-07, "loss": 0.0023, "step": 45926 }, { "epoch": 2.4485258836700963, "grad_norm": 1.4310824871063232, "learning_rate": 8.148e-07, "loss": 0.0159, "step": 45927 }, { "epoch": 2.4485791970997495, "grad_norm": 0.5506475567817688, "learning_rate": 8.146000000000001e-07, "loss": 0.0033, "step": 45928 }, { "epoch": 2.4486325105294022, "grad_norm": 0.9047490954399109, "learning_rate": 8.144000000000001e-07, "loss": 0.0059, "step": 45929 }, { "epoch": 2.4486858239590554, "grad_norm": 3.9257280826568604, "learning_rate": 8.142000000000001e-07, "loss": 0.026, "step": 45930 }, { "epoch": 2.448739137388708, "grad_norm": 0.6224234104156494, "learning_rate": 8.140000000000001e-07, "loss": 0.0036, "step": 45931 }, { "epoch": 2.4487924508183614, "grad_norm": 0.7344306111335754, "learning_rate": 8.138e-07, "loss": 0.0044, "step": 45932 }, { "epoch": 2.448845764248014, "grad_norm": 1.3866170644760132, "learning_rate": 8.136e-07, "loss": 0.0151, "step": 45933 }, { "epoch": 2.448899077677667, "grad_norm": 2.0035762786865234, "learning_rate": 8.134e-07, "loss": 0.0386, "step": 45934 }, { "epoch": 2.44895239110732, "grad_norm": 1.4435415267944336, "learning_rate": 8.132000000000001e-07, "loss": 0.0083, "step": 45935 }, { "epoch": 2.449005704536973, "grad_norm": 0.6097514033317566, "learning_rate": 8.13e-07, "loss": 0.003, "step": 45936 }, { "epoch": 2.4490590179666256, "grad_norm": 1.3169472217559814, "learning_rate": 8.128000000000001e-07, "loss": 0.0335, "step": 45937 }, { "epoch": 2.449112331396279, "grad_norm": 3.6819355487823486, "learning_rate": 8.126000000000001e-07, "loss": 0.038, "step": 45938 }, { "epoch": 2.4491656448259316, "grad_norm": 3.024650812149048, "learning_rate": 8.124000000000002e-07, "loss": 0.0158, "step": 45939 }, { "epoch": 2.4492189582555843, "grad_norm": 2.7285349369049072, "learning_rate": 8.122e-07, "loss": 0.0065, "step": 45940 }, { "epoch": 2.4492722716852375, "grad_norm": 2.848613739013672, "learning_rate": 8.12e-07, "loss": 0.0325, "step": 45941 }, { "epoch": 2.4493255851148903, "grad_norm": 1.9349539279937744, "learning_rate": 8.118000000000001e-07, "loss": 0.0199, "step": 45942 }, { "epoch": 2.4493788985445435, "grad_norm": 1.9848288297653198, "learning_rate": 8.116e-07, "loss": 0.0084, "step": 45943 }, { "epoch": 2.4494322119741962, "grad_norm": 2.434006452560425, "learning_rate": 8.114000000000001e-07, "loss": 0.0514, "step": 45944 }, { "epoch": 2.4494855254038495, "grad_norm": 1.7020388841629028, "learning_rate": 8.112000000000001e-07, "loss": 0.0044, "step": 45945 }, { "epoch": 2.449538838833502, "grad_norm": 2.7820310592651367, "learning_rate": 8.110000000000002e-07, "loss": 0.0231, "step": 45946 }, { "epoch": 2.449592152263155, "grad_norm": 0.26719844341278076, "learning_rate": 8.108e-07, "loss": 0.0019, "step": 45947 }, { "epoch": 2.449645465692808, "grad_norm": 1.950448989868164, "learning_rate": 8.106e-07, "loss": 0.0321, "step": 45948 }, { "epoch": 2.449698779122461, "grad_norm": 2.062730073928833, "learning_rate": 8.104000000000001e-07, "loss": 0.0284, "step": 45949 }, { "epoch": 2.4497520925521137, "grad_norm": 0.7870271801948547, "learning_rate": 8.102e-07, "loss": 0.0056, "step": 45950 }, { "epoch": 2.449805405981767, "grad_norm": 2.3046135902404785, "learning_rate": 8.100000000000001e-07, "loss": 0.0312, "step": 45951 }, { "epoch": 2.4498587194114196, "grad_norm": 0.5057769417762756, "learning_rate": 8.098000000000001e-07, "loss": 0.0034, "step": 45952 }, { "epoch": 2.449912032841073, "grad_norm": 1.0954474210739136, "learning_rate": 8.096000000000002e-07, "loss": 0.0055, "step": 45953 }, { "epoch": 2.4499653462707256, "grad_norm": 1.977003574371338, "learning_rate": 8.094e-07, "loss": 0.0097, "step": 45954 }, { "epoch": 2.4500186597003784, "grad_norm": 0.3653460741043091, "learning_rate": 8.092000000000001e-07, "loss": 0.0027, "step": 45955 }, { "epoch": 2.4500719731300316, "grad_norm": 3.8913118839263916, "learning_rate": 8.090000000000001e-07, "loss": 0.0437, "step": 45956 }, { "epoch": 2.4501252865596843, "grad_norm": 5.3199849128723145, "learning_rate": 8.088e-07, "loss": 0.0256, "step": 45957 }, { "epoch": 2.4501785999893375, "grad_norm": 1.3719630241394043, "learning_rate": 8.086000000000001e-07, "loss": 0.0077, "step": 45958 }, { "epoch": 2.4502319134189903, "grad_norm": 2.599761962890625, "learning_rate": 8.084000000000001e-07, "loss": 0.0191, "step": 45959 }, { "epoch": 2.450285226848643, "grad_norm": 2.9022462368011475, "learning_rate": 8.082e-07, "loss": 0.0175, "step": 45960 }, { "epoch": 2.4503385402782962, "grad_norm": 1.877143383026123, "learning_rate": 8.08e-07, "loss": 0.009, "step": 45961 }, { "epoch": 2.450391853707949, "grad_norm": 1.052018642425537, "learning_rate": 8.078000000000001e-07, "loss": 0.0077, "step": 45962 }, { "epoch": 2.4504451671376017, "grad_norm": 0.5500254034996033, "learning_rate": 8.076e-07, "loss": 0.004, "step": 45963 }, { "epoch": 2.450498480567255, "grad_norm": 2.1920652389526367, "learning_rate": 8.074000000000001e-07, "loss": 0.0102, "step": 45964 }, { "epoch": 2.4505517939969077, "grad_norm": 1.2093759775161743, "learning_rate": 8.072000000000001e-07, "loss": 0.0054, "step": 45965 }, { "epoch": 2.450605107426561, "grad_norm": 1.9943745136260986, "learning_rate": 8.07e-07, "loss": 0.0263, "step": 45966 }, { "epoch": 2.4506584208562137, "grad_norm": 1.9712512493133545, "learning_rate": 8.068e-07, "loss": 0.0342, "step": 45967 }, { "epoch": 2.4507117342858664, "grad_norm": 1.7514516115188599, "learning_rate": 8.066e-07, "loss": 0.0099, "step": 45968 }, { "epoch": 2.4507650477155196, "grad_norm": 1.8469579219818115, "learning_rate": 8.064000000000001e-07, "loss": 0.0082, "step": 45969 }, { "epoch": 2.4508183611451724, "grad_norm": 0.6980924010276794, "learning_rate": 8.062e-07, "loss": 0.004, "step": 45970 }, { "epoch": 2.4508716745748256, "grad_norm": 3.154317855834961, "learning_rate": 8.060000000000001e-07, "loss": 0.0217, "step": 45971 }, { "epoch": 2.4509249880044783, "grad_norm": 2.1738996505737305, "learning_rate": 8.058000000000001e-07, "loss": 0.0263, "step": 45972 }, { "epoch": 2.450978301434131, "grad_norm": 2.5061757564544678, "learning_rate": 8.056000000000002e-07, "loss": 0.0486, "step": 45973 }, { "epoch": 2.4510316148637843, "grad_norm": 2.6930336952209473, "learning_rate": 8.054e-07, "loss": 0.0181, "step": 45974 }, { "epoch": 2.451084928293437, "grad_norm": 0.8328229188919067, "learning_rate": 8.052e-07, "loss": 0.016, "step": 45975 }, { "epoch": 2.45113824172309, "grad_norm": 2.5776076316833496, "learning_rate": 8.050000000000001e-07, "loss": 0.0137, "step": 45976 }, { "epoch": 2.451191555152743, "grad_norm": 1.813628077507019, "learning_rate": 8.048e-07, "loss": 0.0243, "step": 45977 }, { "epoch": 2.4512448685823958, "grad_norm": 4.428990840911865, "learning_rate": 8.046000000000001e-07, "loss": 0.0311, "step": 45978 }, { "epoch": 2.451298182012049, "grad_norm": 1.361158013343811, "learning_rate": 8.044000000000001e-07, "loss": 0.006, "step": 45979 }, { "epoch": 2.4513514954417017, "grad_norm": 2.287792444229126, "learning_rate": 8.042000000000001e-07, "loss": 0.0213, "step": 45980 }, { "epoch": 2.451404808871355, "grad_norm": 1.8638911247253418, "learning_rate": 8.04e-07, "loss": 0.0201, "step": 45981 }, { "epoch": 2.4514581223010077, "grad_norm": 1.024778962135315, "learning_rate": 8.038e-07, "loss": 0.0044, "step": 45982 }, { "epoch": 2.4515114357306604, "grad_norm": 2.7333791255950928, "learning_rate": 8.036e-07, "loss": 0.0092, "step": 45983 }, { "epoch": 2.4515647491603136, "grad_norm": 2.330059051513672, "learning_rate": 8.034e-07, "loss": 0.0394, "step": 45984 }, { "epoch": 2.4516180625899664, "grad_norm": 2.4256367683410645, "learning_rate": 8.032000000000001e-07, "loss": 0.0133, "step": 45985 }, { "epoch": 2.451671376019619, "grad_norm": 4.218086242675781, "learning_rate": 8.03e-07, "loss": 0.0281, "step": 45986 }, { "epoch": 2.4517246894492724, "grad_norm": 1.5419776439666748, "learning_rate": 8.028000000000001e-07, "loss": 0.0065, "step": 45987 }, { "epoch": 2.451778002878925, "grad_norm": 0.6843166947364807, "learning_rate": 8.026000000000001e-07, "loss": 0.004, "step": 45988 }, { "epoch": 2.4518313163085783, "grad_norm": 2.062328577041626, "learning_rate": 8.024000000000001e-07, "loss": 0.0353, "step": 45989 }, { "epoch": 2.451884629738231, "grad_norm": 2.2987887859344482, "learning_rate": 8.022e-07, "loss": 0.0215, "step": 45990 }, { "epoch": 2.451937943167884, "grad_norm": 1.2119145393371582, "learning_rate": 8.02e-07, "loss": 0.0159, "step": 45991 }, { "epoch": 2.451991256597537, "grad_norm": 2.154279947280884, "learning_rate": 8.018000000000001e-07, "loss": 0.0172, "step": 45992 }, { "epoch": 2.45204457002719, "grad_norm": 2.732501983642578, "learning_rate": 8.016e-07, "loss": 0.0661, "step": 45993 }, { "epoch": 2.452097883456843, "grad_norm": 0.6525807976722717, "learning_rate": 8.014000000000001e-07, "loss": 0.003, "step": 45994 }, { "epoch": 2.4521511968864957, "grad_norm": 0.7549439668655396, "learning_rate": 8.012000000000001e-07, "loss": 0.0039, "step": 45995 }, { "epoch": 2.4522045103161485, "grad_norm": 1.7552577257156372, "learning_rate": 8.010000000000001e-07, "loss": 0.0046, "step": 45996 }, { "epoch": 2.4522578237458017, "grad_norm": 2.232938528060913, "learning_rate": 8.008e-07, "loss": 0.027, "step": 45997 }, { "epoch": 2.4523111371754545, "grad_norm": 1.459765911102295, "learning_rate": 8.006000000000001e-07, "loss": 0.0316, "step": 45998 }, { "epoch": 2.452364450605107, "grad_norm": 3.610825777053833, "learning_rate": 8.004000000000001e-07, "loss": 0.0151, "step": 45999 }, { "epoch": 2.4524177640347604, "grad_norm": 3.104811668395996, "learning_rate": 8.002e-07, "loss": 0.0282, "step": 46000 }, { "epoch": 2.452471077464413, "grad_norm": 0.759528636932373, "learning_rate": 8.000000000000001e-07, "loss": 0.0052, "step": 46001 }, { "epoch": 2.4525243908940664, "grad_norm": 0.8957908153533936, "learning_rate": 7.998000000000001e-07, "loss": 0.0034, "step": 46002 }, { "epoch": 2.452577704323719, "grad_norm": 0.16587698459625244, "learning_rate": 7.996000000000001e-07, "loss": 0.0016, "step": 46003 }, { "epoch": 2.452631017753372, "grad_norm": 1.306414008140564, "learning_rate": 7.994e-07, "loss": 0.0056, "step": 46004 }, { "epoch": 2.452684331183025, "grad_norm": 3.783555746078491, "learning_rate": 7.992000000000001e-07, "loss": 0.0273, "step": 46005 }, { "epoch": 2.452737644612678, "grad_norm": 3.1456172466278076, "learning_rate": 7.990000000000001e-07, "loss": 0.0112, "step": 46006 }, { "epoch": 2.452790958042331, "grad_norm": 2.360659122467041, "learning_rate": 7.988000000000001e-07, "loss": 0.0077, "step": 46007 }, { "epoch": 2.452844271471984, "grad_norm": 1.3544288873672485, "learning_rate": 7.986000000000001e-07, "loss": 0.0123, "step": 46008 }, { "epoch": 2.4528975849016366, "grad_norm": 0.7742226123809814, "learning_rate": 7.984000000000001e-07, "loss": 0.0036, "step": 46009 }, { "epoch": 2.4529508983312898, "grad_norm": 0.45992255210876465, "learning_rate": 7.982e-07, "loss": 0.003, "step": 46010 }, { "epoch": 2.4530042117609425, "grad_norm": 1.5080803632736206, "learning_rate": 7.98e-07, "loss": 0.0156, "step": 46011 }, { "epoch": 2.4530575251905953, "grad_norm": 2.170633316040039, "learning_rate": 7.978000000000001e-07, "loss": 0.011, "step": 46012 }, { "epoch": 2.4531108386202485, "grad_norm": 0.9656623601913452, "learning_rate": 7.976e-07, "loss": 0.0045, "step": 46013 }, { "epoch": 2.4531641520499012, "grad_norm": 1.4611759185791016, "learning_rate": 7.974000000000001e-07, "loss": 0.0225, "step": 46014 }, { "epoch": 2.4532174654795544, "grad_norm": 2.656543016433716, "learning_rate": 7.972000000000001e-07, "loss": 0.0169, "step": 46015 }, { "epoch": 2.453270778909207, "grad_norm": 2.1192984580993652, "learning_rate": 7.97e-07, "loss": 0.0178, "step": 46016 }, { "epoch": 2.4533240923388604, "grad_norm": 1.9811038970947266, "learning_rate": 7.968e-07, "loss": 0.0144, "step": 46017 }, { "epoch": 2.453377405768513, "grad_norm": 2.430703639984131, "learning_rate": 7.966e-07, "loss": 0.0344, "step": 46018 }, { "epoch": 2.453430719198166, "grad_norm": 1.8577269315719604, "learning_rate": 7.964000000000001e-07, "loss": 0.0169, "step": 46019 }, { "epoch": 2.453484032627819, "grad_norm": 1.9029455184936523, "learning_rate": 7.962e-07, "loss": 0.0176, "step": 46020 }, { "epoch": 2.453537346057472, "grad_norm": 1.5675204992294312, "learning_rate": 7.960000000000001e-07, "loss": 0.0119, "step": 46021 }, { "epoch": 2.4535906594871246, "grad_norm": 1.63960862159729, "learning_rate": 7.958000000000001e-07, "loss": 0.0243, "step": 46022 }, { "epoch": 2.453643972916778, "grad_norm": 2.1378190517425537, "learning_rate": 7.956000000000002e-07, "loss": 0.0178, "step": 46023 }, { "epoch": 2.4536972863464306, "grad_norm": 1.3377809524536133, "learning_rate": 7.954e-07, "loss": 0.0216, "step": 46024 }, { "epoch": 2.453750599776084, "grad_norm": 1.775688886642456, "learning_rate": 7.952e-07, "loss": 0.0094, "step": 46025 }, { "epoch": 2.4538039132057365, "grad_norm": 1.5906952619552612, "learning_rate": 7.950000000000001e-07, "loss": 0.0088, "step": 46026 }, { "epoch": 2.4538572266353893, "grad_norm": 1.3670119047164917, "learning_rate": 7.948e-07, "loss": 0.0173, "step": 46027 }, { "epoch": 2.4539105400650425, "grad_norm": 2.202972173690796, "learning_rate": 7.946000000000001e-07, "loss": 0.0248, "step": 46028 }, { "epoch": 2.4539638534946953, "grad_norm": 0.342668741941452, "learning_rate": 7.944000000000001e-07, "loss": 0.0022, "step": 46029 }, { "epoch": 2.4540171669243485, "grad_norm": 3.224334239959717, "learning_rate": 7.942000000000001e-07, "loss": 0.0322, "step": 46030 }, { "epoch": 2.454070480354001, "grad_norm": 2.308887243270874, "learning_rate": 7.94e-07, "loss": 0.0091, "step": 46031 }, { "epoch": 2.454123793783654, "grad_norm": 3.3284103870391846, "learning_rate": 7.938000000000001e-07, "loss": 0.0265, "step": 46032 }, { "epoch": 2.454177107213307, "grad_norm": 1.9133248329162598, "learning_rate": 7.936e-07, "loss": 0.0181, "step": 46033 }, { "epoch": 2.45423042064296, "grad_norm": 1.922734260559082, "learning_rate": 7.934e-07, "loss": 0.0236, "step": 46034 }, { "epoch": 2.4542837340726127, "grad_norm": 2.9455180168151855, "learning_rate": 7.932000000000001e-07, "loss": 0.0248, "step": 46035 }, { "epoch": 2.454337047502266, "grad_norm": 2.8443994522094727, "learning_rate": 7.93e-07, "loss": 0.019, "step": 46036 }, { "epoch": 2.4543903609319186, "grad_norm": 1.619304895401001, "learning_rate": 7.928000000000001e-07, "loss": 0.0117, "step": 46037 }, { "epoch": 2.454443674361572, "grad_norm": 1.1378514766693115, "learning_rate": 7.926e-07, "loss": 0.0097, "step": 46038 }, { "epoch": 2.4544969877912246, "grad_norm": 4.430093765258789, "learning_rate": 7.924000000000001e-07, "loss": 0.0144, "step": 46039 }, { "epoch": 2.4545503012208774, "grad_norm": 1.4551130533218384, "learning_rate": 7.922e-07, "loss": 0.0212, "step": 46040 }, { "epoch": 2.4546036146505306, "grad_norm": 2.2111129760742188, "learning_rate": 7.920000000000001e-07, "loss": 0.008, "step": 46041 }, { "epoch": 2.4546569280801833, "grad_norm": 2.490614652633667, "learning_rate": 7.918000000000001e-07, "loss": 0.0238, "step": 46042 }, { "epoch": 2.4547102415098365, "grad_norm": 1.4244517087936401, "learning_rate": 7.916e-07, "loss": 0.0193, "step": 46043 }, { "epoch": 2.4547635549394893, "grad_norm": 2.5716686248779297, "learning_rate": 7.914000000000001e-07, "loss": 0.0211, "step": 46044 }, { "epoch": 2.454816868369142, "grad_norm": 2.363908290863037, "learning_rate": 7.912e-07, "loss": 0.0156, "step": 46045 }, { "epoch": 2.4548701817987952, "grad_norm": 3.5229861736297607, "learning_rate": 7.910000000000001e-07, "loss": 0.0139, "step": 46046 }, { "epoch": 2.454923495228448, "grad_norm": 1.1559072732925415, "learning_rate": 7.908e-07, "loss": 0.006, "step": 46047 }, { "epoch": 2.4549768086581008, "grad_norm": 2.086793899536133, "learning_rate": 7.906000000000001e-07, "loss": 0.0092, "step": 46048 }, { "epoch": 2.455030122087754, "grad_norm": 1.7788846492767334, "learning_rate": 7.904000000000001e-07, "loss": 0.0092, "step": 46049 }, { "epoch": 2.4550834355174067, "grad_norm": 1.9932446479797363, "learning_rate": 7.902000000000002e-07, "loss": 0.04, "step": 46050 }, { "epoch": 2.45513674894706, "grad_norm": 1.6780891418457031, "learning_rate": 7.900000000000001e-07, "loss": 0.0105, "step": 46051 }, { "epoch": 2.4551900623767127, "grad_norm": 1.440168023109436, "learning_rate": 7.898e-07, "loss": 0.0113, "step": 46052 }, { "epoch": 2.455243375806366, "grad_norm": 1.1414382457733154, "learning_rate": 7.896000000000001e-07, "loss": 0.0056, "step": 46053 }, { "epoch": 2.4552966892360186, "grad_norm": 2.7857675552368164, "learning_rate": 7.894e-07, "loss": 0.0387, "step": 46054 }, { "epoch": 2.4553500026656714, "grad_norm": 0.3017219603061676, "learning_rate": 7.892000000000001e-07, "loss": 0.002, "step": 46055 }, { "epoch": 2.4554033160953246, "grad_norm": 1.8955947160720825, "learning_rate": 7.890000000000001e-07, "loss": 0.0181, "step": 46056 }, { "epoch": 2.4554566295249773, "grad_norm": 3.261324644088745, "learning_rate": 7.888000000000001e-07, "loss": 0.0192, "step": 46057 }, { "epoch": 2.45550994295463, "grad_norm": 0.8624382019042969, "learning_rate": 7.886000000000001e-07, "loss": 0.0042, "step": 46058 }, { "epoch": 2.4555632563842833, "grad_norm": 4.504915714263916, "learning_rate": 7.884e-07, "loss": 0.0198, "step": 46059 }, { "epoch": 2.455616569813936, "grad_norm": 0.2422034591436386, "learning_rate": 7.882e-07, "loss": 0.002, "step": 46060 }, { "epoch": 2.455669883243589, "grad_norm": 0.8347589373588562, "learning_rate": 7.88e-07, "loss": 0.0036, "step": 46061 }, { "epoch": 2.455723196673242, "grad_norm": 1.9799941778182983, "learning_rate": 7.878000000000001e-07, "loss": 0.0294, "step": 46062 }, { "epoch": 2.4557765101028948, "grad_norm": 3.8774919509887695, "learning_rate": 7.876e-07, "loss": 0.0249, "step": 46063 }, { "epoch": 2.455829823532548, "grad_norm": 1.8533458709716797, "learning_rate": 7.874000000000001e-07, "loss": 0.0184, "step": 46064 }, { "epoch": 2.4558831369622007, "grad_norm": 1.4796338081359863, "learning_rate": 7.872000000000001e-07, "loss": 0.0066, "step": 46065 }, { "epoch": 2.455936450391854, "grad_norm": 2.298698902130127, "learning_rate": 7.870000000000002e-07, "loss": 0.0108, "step": 46066 }, { "epoch": 2.4559897638215067, "grad_norm": 2.2798585891723633, "learning_rate": 7.868e-07, "loss": 0.0166, "step": 46067 }, { "epoch": 2.4560430772511594, "grad_norm": 2.349137544631958, "learning_rate": 7.866e-07, "loss": 0.0237, "step": 46068 }, { "epoch": 2.4560963906808126, "grad_norm": 2.345313310623169, "learning_rate": 7.864000000000001e-07, "loss": 0.0174, "step": 46069 }, { "epoch": 2.4561497041104654, "grad_norm": 2.9746484756469727, "learning_rate": 7.862e-07, "loss": 0.0309, "step": 46070 }, { "epoch": 2.456203017540118, "grad_norm": 1.171988606452942, "learning_rate": 7.860000000000001e-07, "loss": 0.006, "step": 46071 }, { "epoch": 2.4562563309697714, "grad_norm": 2.318510055541992, "learning_rate": 7.858000000000001e-07, "loss": 0.0118, "step": 46072 }, { "epoch": 2.456309644399424, "grad_norm": 2.1186985969543457, "learning_rate": 7.856000000000002e-07, "loss": 0.0215, "step": 46073 }, { "epoch": 2.4563629578290773, "grad_norm": 0.9058212637901306, "learning_rate": 7.854e-07, "loss": 0.0198, "step": 46074 }, { "epoch": 2.45641627125873, "grad_norm": 1.132935643196106, "learning_rate": 7.852000000000001e-07, "loss": 0.017, "step": 46075 }, { "epoch": 2.456469584688383, "grad_norm": 3.3201310634613037, "learning_rate": 7.850000000000001e-07, "loss": 0.0091, "step": 46076 }, { "epoch": 2.456522898118036, "grad_norm": 1.8982189893722534, "learning_rate": 7.848e-07, "loss": 0.0206, "step": 46077 }, { "epoch": 2.456576211547689, "grad_norm": 1.9766021966934204, "learning_rate": 7.846000000000001e-07, "loss": 0.0101, "step": 46078 }, { "epoch": 2.456629524977342, "grad_norm": 0.729710578918457, "learning_rate": 7.844000000000001e-07, "loss": 0.006, "step": 46079 }, { "epoch": 2.4566828384069948, "grad_norm": 1.24501633644104, "learning_rate": 7.842e-07, "loss": 0.0228, "step": 46080 }, { "epoch": 2.4567361518366475, "grad_norm": 2.213883638381958, "learning_rate": 7.84e-07, "loss": 0.0143, "step": 46081 }, { "epoch": 2.4567894652663007, "grad_norm": 0.8404591083526611, "learning_rate": 7.838000000000001e-07, "loss": 0.0049, "step": 46082 }, { "epoch": 2.4568427786959535, "grad_norm": 2.5545177459716797, "learning_rate": 7.836e-07, "loss": 0.029, "step": 46083 }, { "epoch": 2.4568960921256062, "grad_norm": 3.368797540664673, "learning_rate": 7.834000000000001e-07, "loss": 0.0205, "step": 46084 }, { "epoch": 2.4569494055552594, "grad_norm": 1.6169679164886475, "learning_rate": 7.832000000000001e-07, "loss": 0.0231, "step": 46085 }, { "epoch": 2.457002718984912, "grad_norm": 1.8071459531784058, "learning_rate": 7.83e-07, "loss": 0.0157, "step": 46086 }, { "epoch": 2.4570560324145654, "grad_norm": 2.5757803916931152, "learning_rate": 7.828e-07, "loss": 0.0261, "step": 46087 }, { "epoch": 2.457109345844218, "grad_norm": 2.575294256210327, "learning_rate": 7.826e-07, "loss": 0.0319, "step": 46088 }, { "epoch": 2.457162659273871, "grad_norm": 1.3673081398010254, "learning_rate": 7.824000000000001e-07, "loss": 0.0063, "step": 46089 }, { "epoch": 2.457215972703524, "grad_norm": 1.122412085533142, "learning_rate": 7.822e-07, "loss": 0.005, "step": 46090 }, { "epoch": 2.457269286133177, "grad_norm": 2.9507839679718018, "learning_rate": 7.820000000000001e-07, "loss": 0.0146, "step": 46091 }, { "epoch": 2.45732259956283, "grad_norm": 2.497375726699829, "learning_rate": 7.818000000000001e-07, "loss": 0.0299, "step": 46092 }, { "epoch": 2.457375912992483, "grad_norm": 1.2778629064559937, "learning_rate": 7.816e-07, "loss": 0.0186, "step": 46093 }, { "epoch": 2.4574292264221356, "grad_norm": 0.460807204246521, "learning_rate": 7.814e-07, "loss": 0.0038, "step": 46094 }, { "epoch": 2.4574825398517888, "grad_norm": 1.6585530042648315, "learning_rate": 7.812e-07, "loss": 0.0358, "step": 46095 }, { "epoch": 2.4575358532814415, "grad_norm": 1.4098926782608032, "learning_rate": 7.810000000000001e-07, "loss": 0.0039, "step": 46096 }, { "epoch": 2.4575891667110943, "grad_norm": 3.612670660018921, "learning_rate": 7.808e-07, "loss": 0.0291, "step": 46097 }, { "epoch": 2.4576424801407475, "grad_norm": 1.5818400382995605, "learning_rate": 7.806000000000001e-07, "loss": 0.0353, "step": 46098 }, { "epoch": 2.4576957935704002, "grad_norm": 3.7811055183410645, "learning_rate": 7.804000000000001e-07, "loss": 0.0424, "step": 46099 }, { "epoch": 2.4577491070000534, "grad_norm": 2.161301851272583, "learning_rate": 7.802000000000002e-07, "loss": 0.0082, "step": 46100 }, { "epoch": 2.457802420429706, "grad_norm": 2.21555757522583, "learning_rate": 7.8e-07, "loss": 0.0103, "step": 46101 }, { "epoch": 2.4578557338593594, "grad_norm": 2.866664171218872, "learning_rate": 7.798e-07, "loss": 0.0308, "step": 46102 }, { "epoch": 2.457909047289012, "grad_norm": 1.2974014282226562, "learning_rate": 7.796000000000001e-07, "loss": 0.016, "step": 46103 }, { "epoch": 2.457962360718665, "grad_norm": 1.6663943529129028, "learning_rate": 7.794e-07, "loss": 0.0085, "step": 46104 }, { "epoch": 2.458015674148318, "grad_norm": 1.982438087463379, "learning_rate": 7.792000000000001e-07, "loss": 0.013, "step": 46105 }, { "epoch": 2.458068987577971, "grad_norm": 3.1242494583129883, "learning_rate": 7.790000000000001e-07, "loss": 0.028, "step": 46106 }, { "epoch": 2.4581223010076236, "grad_norm": 1.8067957162857056, "learning_rate": 7.788000000000001e-07, "loss": 0.0408, "step": 46107 }, { "epoch": 2.458175614437277, "grad_norm": 1.278053641319275, "learning_rate": 7.786e-07, "loss": 0.0076, "step": 46108 }, { "epoch": 2.4582289278669296, "grad_norm": 2.7958455085754395, "learning_rate": 7.784000000000001e-07, "loss": 0.0154, "step": 46109 }, { "epoch": 2.458282241296583, "grad_norm": 0.912510335445404, "learning_rate": 7.782e-07, "loss": 0.0048, "step": 46110 }, { "epoch": 2.4583355547262356, "grad_norm": 1.5225692987442017, "learning_rate": 7.78e-07, "loss": 0.0065, "step": 46111 }, { "epoch": 2.4583888681558883, "grad_norm": 2.0483522415161133, "learning_rate": 7.778000000000001e-07, "loss": 0.007, "step": 46112 }, { "epoch": 2.4584421815855415, "grad_norm": 2.9357805252075195, "learning_rate": 7.776e-07, "loss": 0.028, "step": 46113 }, { "epoch": 2.4584954950151943, "grad_norm": 1.0770901441574097, "learning_rate": 7.774000000000001e-07, "loss": 0.0061, "step": 46114 }, { "epoch": 2.4585488084448475, "grad_norm": 1.4403231143951416, "learning_rate": 7.772e-07, "loss": 0.0226, "step": 46115 }, { "epoch": 2.4586021218745002, "grad_norm": 2.908454179763794, "learning_rate": 7.770000000000001e-07, "loss": 0.0114, "step": 46116 }, { "epoch": 2.458655435304153, "grad_norm": 5.309262752532959, "learning_rate": 7.768e-07, "loss": 0.0255, "step": 46117 }, { "epoch": 2.458708748733806, "grad_norm": 0.25873538851737976, "learning_rate": 7.766000000000001e-07, "loss": 0.0018, "step": 46118 }, { "epoch": 2.458762062163459, "grad_norm": 0.9114510416984558, "learning_rate": 7.764000000000001e-07, "loss": 0.0138, "step": 46119 }, { "epoch": 2.4588153755931117, "grad_norm": 0.7365323901176453, "learning_rate": 7.762e-07, "loss": 0.0036, "step": 46120 }, { "epoch": 2.458868689022765, "grad_norm": 1.1066596508026123, "learning_rate": 7.760000000000001e-07, "loss": 0.0163, "step": 46121 }, { "epoch": 2.4589220024524177, "grad_norm": 2.137758493423462, "learning_rate": 7.758e-07, "loss": 0.013, "step": 46122 }, { "epoch": 2.458975315882071, "grad_norm": 1.4199618101119995, "learning_rate": 7.756000000000001e-07, "loss": 0.0075, "step": 46123 }, { "epoch": 2.4590286293117236, "grad_norm": 1.1812148094177246, "learning_rate": 7.754e-07, "loss": 0.0034, "step": 46124 }, { "epoch": 2.4590819427413764, "grad_norm": 0.6774179935455322, "learning_rate": 7.752000000000001e-07, "loss": 0.0028, "step": 46125 }, { "epoch": 2.4591352561710296, "grad_norm": 1.8362300395965576, "learning_rate": 7.750000000000001e-07, "loss": 0.0065, "step": 46126 }, { "epoch": 2.4591885696006823, "grad_norm": 2.553231954574585, "learning_rate": 7.748e-07, "loss": 0.0124, "step": 46127 }, { "epoch": 2.4592418830303355, "grad_norm": 0.9144281148910522, "learning_rate": 7.746000000000001e-07, "loss": 0.0047, "step": 46128 }, { "epoch": 2.4592951964599883, "grad_norm": 1.7555313110351562, "learning_rate": 7.744e-07, "loss": 0.021, "step": 46129 }, { "epoch": 2.459348509889641, "grad_norm": 3.921332359313965, "learning_rate": 7.742000000000001e-07, "loss": 0.0079, "step": 46130 }, { "epoch": 2.4594018233192942, "grad_norm": 2.7607808113098145, "learning_rate": 7.74e-07, "loss": 0.0097, "step": 46131 }, { "epoch": 2.459455136748947, "grad_norm": 1.945570707321167, "learning_rate": 7.738000000000001e-07, "loss": 0.0284, "step": 46132 }, { "epoch": 2.4595084501785998, "grad_norm": 3.2580556869506836, "learning_rate": 7.736000000000001e-07, "loss": 0.0494, "step": 46133 }, { "epoch": 2.459561763608253, "grad_norm": 1.5109864473342896, "learning_rate": 7.734000000000001e-07, "loss": 0.0143, "step": 46134 }, { "epoch": 2.4596150770379057, "grad_norm": 2.004523515701294, "learning_rate": 7.732000000000001e-07, "loss": 0.022, "step": 46135 }, { "epoch": 2.459668390467559, "grad_norm": 0.5087413191795349, "learning_rate": 7.73e-07, "loss": 0.0028, "step": 46136 }, { "epoch": 2.4597217038972117, "grad_norm": 0.7524468898773193, "learning_rate": 7.728e-07, "loss": 0.0046, "step": 46137 }, { "epoch": 2.459775017326865, "grad_norm": 3.4181251525878906, "learning_rate": 7.726e-07, "loss": 0.0352, "step": 46138 }, { "epoch": 2.4598283307565176, "grad_norm": 0.8099033832550049, "learning_rate": 7.724000000000001e-07, "loss": 0.0033, "step": 46139 }, { "epoch": 2.4598816441861704, "grad_norm": 3.656846761703491, "learning_rate": 7.722e-07, "loss": 0.0392, "step": 46140 }, { "epoch": 2.4599349576158236, "grad_norm": 2.1403777599334717, "learning_rate": 7.720000000000001e-07, "loss": 0.0136, "step": 46141 }, { "epoch": 2.4599882710454763, "grad_norm": 1.1926630735397339, "learning_rate": 7.718000000000001e-07, "loss": 0.0111, "step": 46142 }, { "epoch": 2.460041584475129, "grad_norm": 0.7357276678085327, "learning_rate": 7.716000000000002e-07, "loss": 0.0037, "step": 46143 }, { "epoch": 2.4600948979047823, "grad_norm": 2.638744354248047, "learning_rate": 7.714e-07, "loss": 0.0127, "step": 46144 }, { "epoch": 2.460148211334435, "grad_norm": 1.885063648223877, "learning_rate": 7.712e-07, "loss": 0.0116, "step": 46145 }, { "epoch": 2.4602015247640883, "grad_norm": 0.7949973940849304, "learning_rate": 7.710000000000001e-07, "loss": 0.0036, "step": 46146 }, { "epoch": 2.460254838193741, "grad_norm": 0.966781497001648, "learning_rate": 7.708e-07, "loss": 0.005, "step": 46147 }, { "epoch": 2.460308151623394, "grad_norm": 0.31016209721565247, "learning_rate": 7.706000000000001e-07, "loss": 0.0018, "step": 46148 }, { "epoch": 2.460361465053047, "grad_norm": 0.5121656656265259, "learning_rate": 7.704000000000001e-07, "loss": 0.0027, "step": 46149 }, { "epoch": 2.4604147784826997, "grad_norm": 2.292332410812378, "learning_rate": 7.702000000000002e-07, "loss": 0.0175, "step": 46150 }, { "epoch": 2.460468091912353, "grad_norm": 2.1134791374206543, "learning_rate": 7.7e-07, "loss": 0.0169, "step": 46151 }, { "epoch": 2.4605214053420057, "grad_norm": 1.172715663909912, "learning_rate": 7.698000000000001e-07, "loss": 0.0065, "step": 46152 }, { "epoch": 2.4605747187716585, "grad_norm": 0.46269094944000244, "learning_rate": 7.696000000000001e-07, "loss": 0.0028, "step": 46153 }, { "epoch": 2.4606280322013117, "grad_norm": 0.5687275528907776, "learning_rate": 7.694e-07, "loss": 0.004, "step": 46154 }, { "epoch": 2.4606813456309644, "grad_norm": 2.177793025970459, "learning_rate": 7.692000000000001e-07, "loss": 0.0092, "step": 46155 }, { "epoch": 2.460734659060617, "grad_norm": 2.208683967590332, "learning_rate": 7.690000000000001e-07, "loss": 0.0144, "step": 46156 }, { "epoch": 2.4607879724902704, "grad_norm": 4.430401802062988, "learning_rate": 7.688000000000001e-07, "loss": 0.0208, "step": 46157 }, { "epoch": 2.460841285919923, "grad_norm": 1.4038472175598145, "learning_rate": 7.686e-07, "loss": 0.0177, "step": 46158 }, { "epoch": 2.4608945993495763, "grad_norm": 1.9480011463165283, "learning_rate": 7.684000000000001e-07, "loss": 0.0178, "step": 46159 }, { "epoch": 2.460947912779229, "grad_norm": 1.8167256116867065, "learning_rate": 7.682e-07, "loss": 0.0185, "step": 46160 }, { "epoch": 2.461001226208882, "grad_norm": 1.7849534749984741, "learning_rate": 7.68e-07, "loss": 0.0124, "step": 46161 }, { "epoch": 2.461054539638535, "grad_norm": 1.8164302110671997, "learning_rate": 7.678000000000001e-07, "loss": 0.0242, "step": 46162 }, { "epoch": 2.461107853068188, "grad_norm": 0.7952022552490234, "learning_rate": 7.676e-07, "loss": 0.0047, "step": 46163 }, { "epoch": 2.461161166497841, "grad_norm": 0.9837464094161987, "learning_rate": 7.674000000000001e-07, "loss": 0.0063, "step": 46164 }, { "epoch": 2.4612144799274938, "grad_norm": 0.8235899806022644, "learning_rate": 7.672e-07, "loss": 0.0051, "step": 46165 }, { "epoch": 2.4612677933571465, "grad_norm": 4.196481704711914, "learning_rate": 7.670000000000001e-07, "loss": 0.0123, "step": 46166 }, { "epoch": 2.4613211067867997, "grad_norm": 1.7036031484603882, "learning_rate": 7.668e-07, "loss": 0.0102, "step": 46167 }, { "epoch": 2.4613744202164525, "grad_norm": 1.8368912935256958, "learning_rate": 7.666000000000001e-07, "loss": 0.0087, "step": 46168 }, { "epoch": 2.4614277336461052, "grad_norm": 1.8744765520095825, "learning_rate": 7.664000000000001e-07, "loss": 0.0085, "step": 46169 }, { "epoch": 2.4614810470757584, "grad_norm": 0.7655724883079529, "learning_rate": 7.662e-07, "loss": 0.0055, "step": 46170 }, { "epoch": 2.461534360505411, "grad_norm": 2.4372708797454834, "learning_rate": 7.660000000000001e-07, "loss": 0.0407, "step": 46171 }, { "epoch": 2.4615876739350644, "grad_norm": 1.3059611320495605, "learning_rate": 7.658e-07, "loss": 0.0199, "step": 46172 }, { "epoch": 2.461640987364717, "grad_norm": 1.154376745223999, "learning_rate": 7.656000000000001e-07, "loss": 0.0042, "step": 46173 }, { "epoch": 2.4616943007943703, "grad_norm": 1.063750147819519, "learning_rate": 7.654e-07, "loss": 0.0036, "step": 46174 }, { "epoch": 2.461747614224023, "grad_norm": 2.280646562576294, "learning_rate": 7.652000000000001e-07, "loss": 0.0105, "step": 46175 }, { "epoch": 2.461800927653676, "grad_norm": 1.9349254369735718, "learning_rate": 7.650000000000001e-07, "loss": 0.0122, "step": 46176 }, { "epoch": 2.461854241083329, "grad_norm": 1.3409572839736938, "learning_rate": 7.648000000000002e-07, "loss": 0.0126, "step": 46177 }, { "epoch": 2.461907554512982, "grad_norm": 2.508594512939453, "learning_rate": 7.646000000000001e-07, "loss": 0.0273, "step": 46178 }, { "epoch": 2.4619608679426346, "grad_norm": 0.5799762010574341, "learning_rate": 7.644e-07, "loss": 0.0026, "step": 46179 }, { "epoch": 2.462014181372288, "grad_norm": 1.0367786884307861, "learning_rate": 7.642000000000001e-07, "loss": 0.0148, "step": 46180 }, { "epoch": 2.4620674948019405, "grad_norm": 4.122138977050781, "learning_rate": 7.64e-07, "loss": 0.042, "step": 46181 }, { "epoch": 2.4621208082315933, "grad_norm": 0.7544744610786438, "learning_rate": 7.638000000000001e-07, "loss": 0.0028, "step": 46182 }, { "epoch": 2.4621741216612465, "grad_norm": 0.20738540589809418, "learning_rate": 7.636000000000001e-07, "loss": 0.0014, "step": 46183 }, { "epoch": 2.4622274350908993, "grad_norm": 1.9879051446914673, "learning_rate": 7.634000000000001e-07, "loss": 0.0247, "step": 46184 }, { "epoch": 2.4622807485205525, "grad_norm": 1.1273068189620972, "learning_rate": 7.632000000000001e-07, "loss": 0.0157, "step": 46185 }, { "epoch": 2.462334061950205, "grad_norm": 1.956529140472412, "learning_rate": 7.630000000000001e-07, "loss": 0.0105, "step": 46186 }, { "epoch": 2.4623873753798584, "grad_norm": 2.708296537399292, "learning_rate": 7.628e-07, "loss": 0.0141, "step": 46187 }, { "epoch": 2.462440688809511, "grad_norm": 1.9461312294006348, "learning_rate": 7.626e-07, "loss": 0.0105, "step": 46188 }, { "epoch": 2.462494002239164, "grad_norm": 2.0434317588806152, "learning_rate": 7.624000000000001e-07, "loss": 0.0113, "step": 46189 }, { "epoch": 2.462547315668817, "grad_norm": 1.9103271961212158, "learning_rate": 7.622e-07, "loss": 0.0134, "step": 46190 }, { "epoch": 2.46260062909847, "grad_norm": 1.8678197860717773, "learning_rate": 7.620000000000001e-07, "loss": 0.0217, "step": 46191 }, { "epoch": 2.4626539425281226, "grad_norm": 1.6618022918701172, "learning_rate": 7.618000000000001e-07, "loss": 0.0208, "step": 46192 }, { "epoch": 2.462707255957776, "grad_norm": 0.804538905620575, "learning_rate": 7.616000000000001e-07, "loss": 0.0043, "step": 46193 }, { "epoch": 2.4627605693874286, "grad_norm": 0.8570404052734375, "learning_rate": 7.614e-07, "loss": 0.0041, "step": 46194 }, { "epoch": 2.462813882817082, "grad_norm": 1.7221496105194092, "learning_rate": 7.612e-07, "loss": 0.0059, "step": 46195 }, { "epoch": 2.4628671962467346, "grad_norm": 1.3875325918197632, "learning_rate": 7.610000000000001e-07, "loss": 0.0054, "step": 46196 }, { "epoch": 2.4629205096763873, "grad_norm": 3.2737576961517334, "learning_rate": 7.608e-07, "loss": 0.022, "step": 46197 }, { "epoch": 2.4629738231060405, "grad_norm": 2.3383283615112305, "learning_rate": 7.606000000000001e-07, "loss": 0.0281, "step": 46198 }, { "epoch": 2.4630271365356933, "grad_norm": 0.6847879886627197, "learning_rate": 7.604000000000001e-07, "loss": 0.0032, "step": 46199 }, { "epoch": 2.4630804499653465, "grad_norm": 1.253106951713562, "learning_rate": 7.602000000000002e-07, "loss": 0.0057, "step": 46200 }, { "epoch": 2.4631337633949992, "grad_norm": 1.8910918235778809, "learning_rate": 7.6e-07, "loss": 0.0075, "step": 46201 }, { "epoch": 2.463187076824652, "grad_norm": 1.1504231691360474, "learning_rate": 7.598000000000001e-07, "loss": 0.0167, "step": 46202 }, { "epoch": 2.463240390254305, "grad_norm": 2.996708393096924, "learning_rate": 7.596000000000001e-07, "loss": 0.0189, "step": 46203 }, { "epoch": 2.463293703683958, "grad_norm": 2.5596911907196045, "learning_rate": 7.594e-07, "loss": 0.0138, "step": 46204 }, { "epoch": 2.4633470171136107, "grad_norm": 2.169685125350952, "learning_rate": 7.592000000000001e-07, "loss": 0.0078, "step": 46205 }, { "epoch": 2.463400330543264, "grad_norm": 3.137336254119873, "learning_rate": 7.590000000000001e-07, "loss": 0.0149, "step": 46206 }, { "epoch": 2.4634536439729167, "grad_norm": 0.6193313598632812, "learning_rate": 7.588e-07, "loss": 0.005, "step": 46207 }, { "epoch": 2.46350695740257, "grad_norm": 1.2242718935012817, "learning_rate": 7.586e-07, "loss": 0.0067, "step": 46208 }, { "epoch": 2.4635602708322226, "grad_norm": 1.8564759492874146, "learning_rate": 7.584000000000001e-07, "loss": 0.0042, "step": 46209 }, { "epoch": 2.4636135842618754, "grad_norm": 1.0030323266983032, "learning_rate": 7.582e-07, "loss": 0.0127, "step": 46210 }, { "epoch": 2.4636668976915286, "grad_norm": 1.207637071609497, "learning_rate": 7.580000000000001e-07, "loss": 0.018, "step": 46211 }, { "epoch": 2.4637202111211813, "grad_norm": 1.3270939588546753, "learning_rate": 7.578000000000001e-07, "loss": 0.0083, "step": 46212 }, { "epoch": 2.4637735245508345, "grad_norm": 1.6222268342971802, "learning_rate": 7.576e-07, "loss": 0.0068, "step": 46213 }, { "epoch": 2.4638268379804873, "grad_norm": 4.340734481811523, "learning_rate": 7.574e-07, "loss": 0.037, "step": 46214 }, { "epoch": 2.46388015141014, "grad_norm": 2.8173506259918213, "learning_rate": 7.572e-07, "loss": 0.0133, "step": 46215 }, { "epoch": 2.4639334648397933, "grad_norm": 1.6818350553512573, "learning_rate": 7.570000000000001e-07, "loss": 0.0195, "step": 46216 }, { "epoch": 2.463986778269446, "grad_norm": 2.26632022857666, "learning_rate": 7.568e-07, "loss": 0.0323, "step": 46217 }, { "epoch": 2.4640400916990988, "grad_norm": 4.139883041381836, "learning_rate": 7.566000000000001e-07, "loss": 0.0435, "step": 46218 }, { "epoch": 2.464093405128752, "grad_norm": 1.5124746561050415, "learning_rate": 7.564000000000001e-07, "loss": 0.0094, "step": 46219 }, { "epoch": 2.4641467185584047, "grad_norm": 0.796504557132721, "learning_rate": 7.562000000000002e-07, "loss": 0.01, "step": 46220 }, { "epoch": 2.464200031988058, "grad_norm": 1.9761215448379517, "learning_rate": 7.56e-07, "loss": 0.0171, "step": 46221 }, { "epoch": 2.4642533454177107, "grad_norm": 0.6658323407173157, "learning_rate": 7.558e-07, "loss": 0.0036, "step": 46222 }, { "epoch": 2.464306658847364, "grad_norm": 4.0521240234375, "learning_rate": 7.556000000000001e-07, "loss": 0.0236, "step": 46223 }, { "epoch": 2.4643599722770166, "grad_norm": 2.154775857925415, "learning_rate": 7.554e-07, "loss": 0.0099, "step": 46224 }, { "epoch": 2.4644132857066694, "grad_norm": 1.8144025802612305, "learning_rate": 7.552000000000001e-07, "loss": 0.0248, "step": 46225 }, { "epoch": 2.4644665991363226, "grad_norm": 0.7493017315864563, "learning_rate": 7.550000000000001e-07, "loss": 0.0137, "step": 46226 }, { "epoch": 2.4645199125659754, "grad_norm": 0.7227615118026733, "learning_rate": 7.548000000000002e-07, "loss": 0.0041, "step": 46227 }, { "epoch": 2.464573225995628, "grad_norm": 2.077601671218872, "learning_rate": 7.546e-07, "loss": 0.006, "step": 46228 }, { "epoch": 2.4646265394252813, "grad_norm": 2.1315622329711914, "learning_rate": 7.544e-07, "loss": 0.0338, "step": 46229 }, { "epoch": 2.464679852854934, "grad_norm": 1.4628846645355225, "learning_rate": 7.542000000000001e-07, "loss": 0.0092, "step": 46230 }, { "epoch": 2.4647331662845873, "grad_norm": 2.5391876697540283, "learning_rate": 7.54e-07, "loss": 0.0124, "step": 46231 }, { "epoch": 2.46478647971424, "grad_norm": 1.7740672826766968, "learning_rate": 7.538000000000001e-07, "loss": 0.0046, "step": 46232 }, { "epoch": 2.464839793143893, "grad_norm": 0.7921581268310547, "learning_rate": 7.536000000000001e-07, "loss": 0.0044, "step": 46233 }, { "epoch": 2.464893106573546, "grad_norm": 1.9351741075515747, "learning_rate": 7.534000000000001e-07, "loss": 0.0116, "step": 46234 }, { "epoch": 2.4649464200031987, "grad_norm": 2.7287702560424805, "learning_rate": 7.532e-07, "loss": 0.0074, "step": 46235 }, { "epoch": 2.464999733432852, "grad_norm": 2.0209739208221436, "learning_rate": 7.530000000000001e-07, "loss": 0.0296, "step": 46236 }, { "epoch": 2.4650530468625047, "grad_norm": 1.533187985420227, "learning_rate": 7.528e-07, "loss": 0.0074, "step": 46237 }, { "epoch": 2.4651063602921575, "grad_norm": 3.296187400817871, "learning_rate": 7.526e-07, "loss": 0.0092, "step": 46238 }, { "epoch": 2.4651596737218107, "grad_norm": 1.9318206310272217, "learning_rate": 7.524000000000001e-07, "loss": 0.0092, "step": 46239 }, { "epoch": 2.4652129871514634, "grad_norm": 1.0802102088928223, "learning_rate": 7.522e-07, "loss": 0.0052, "step": 46240 }, { "epoch": 2.465266300581116, "grad_norm": 3.9186244010925293, "learning_rate": 7.520000000000001e-07, "loss": 0.0296, "step": 46241 }, { "epoch": 2.4653196140107694, "grad_norm": 1.4755182266235352, "learning_rate": 7.518e-07, "loss": 0.0182, "step": 46242 }, { "epoch": 2.465372927440422, "grad_norm": 2.537510633468628, "learning_rate": 7.516000000000001e-07, "loss": 0.0439, "step": 46243 }, { "epoch": 2.4654262408700753, "grad_norm": 0.8771805763244629, "learning_rate": 7.514e-07, "loss": 0.0066, "step": 46244 }, { "epoch": 2.465479554299728, "grad_norm": 3.9824533462524414, "learning_rate": 7.512000000000001e-07, "loss": 0.0223, "step": 46245 }, { "epoch": 2.465532867729381, "grad_norm": 0.8816482424736023, "learning_rate": 7.510000000000001e-07, "loss": 0.0047, "step": 46246 }, { "epoch": 2.465586181159034, "grad_norm": 3.2211618423461914, "learning_rate": 7.508e-07, "loss": 0.0115, "step": 46247 }, { "epoch": 2.465639494588687, "grad_norm": 0.7156670093536377, "learning_rate": 7.506000000000001e-07, "loss": 0.0035, "step": 46248 }, { "epoch": 2.46569280801834, "grad_norm": 2.6491751670837402, "learning_rate": 7.504e-07, "loss": 0.0307, "step": 46249 }, { "epoch": 2.4657461214479928, "grad_norm": 2.0763652324676514, "learning_rate": 7.502000000000001e-07, "loss": 0.0096, "step": 46250 }, { "epoch": 2.4657994348776455, "grad_norm": 1.1995662450790405, "learning_rate": 7.5e-07, "loss": 0.0187, "step": 46251 }, { "epoch": 2.4658527483072987, "grad_norm": 0.8821271657943726, "learning_rate": 7.498000000000001e-07, "loss": 0.0049, "step": 46252 }, { "epoch": 2.4659060617369515, "grad_norm": 1.2968131303787231, "learning_rate": 7.496000000000001e-07, "loss": 0.0228, "step": 46253 }, { "epoch": 2.4659593751666042, "grad_norm": 1.5562001466751099, "learning_rate": 7.494000000000001e-07, "loss": 0.0171, "step": 46254 }, { "epoch": 2.4660126885962574, "grad_norm": 2.828214645385742, "learning_rate": 7.492000000000001e-07, "loss": 0.0262, "step": 46255 }, { "epoch": 2.46606600202591, "grad_norm": 0.31431424617767334, "learning_rate": 7.49e-07, "loss": 0.0022, "step": 46256 }, { "epoch": 2.4661193154555634, "grad_norm": 0.5916180610656738, "learning_rate": 7.488e-07, "loss": 0.0039, "step": 46257 }, { "epoch": 2.466172628885216, "grad_norm": 3.471158027648926, "learning_rate": 7.486e-07, "loss": 0.0396, "step": 46258 }, { "epoch": 2.4662259423148694, "grad_norm": 0.37009236216545105, "learning_rate": 7.484000000000001e-07, "loss": 0.0026, "step": 46259 }, { "epoch": 2.466279255744522, "grad_norm": 2.090639352798462, "learning_rate": 7.482e-07, "loss": 0.0202, "step": 46260 }, { "epoch": 2.466332569174175, "grad_norm": 2.233327865600586, "learning_rate": 7.480000000000001e-07, "loss": 0.023, "step": 46261 }, { "epoch": 2.466385882603828, "grad_norm": 2.514677047729492, "learning_rate": 7.478000000000001e-07, "loss": 0.0303, "step": 46262 }, { "epoch": 2.466439196033481, "grad_norm": 3.61950945854187, "learning_rate": 7.475999999999999e-07, "loss": 0.0494, "step": 46263 }, { "epoch": 2.4664925094631336, "grad_norm": 1.5663703680038452, "learning_rate": 7.474e-07, "loss": 0.0094, "step": 46264 }, { "epoch": 2.466545822892787, "grad_norm": 2.7215793132781982, "learning_rate": 7.472e-07, "loss": 0.0153, "step": 46265 }, { "epoch": 2.4665991363224395, "grad_norm": 5.373642921447754, "learning_rate": 7.470000000000001e-07, "loss": 0.0256, "step": 46266 }, { "epoch": 2.4666524497520927, "grad_norm": 3.5422515869140625, "learning_rate": 7.468e-07, "loss": 0.0222, "step": 46267 }, { "epoch": 2.4667057631817455, "grad_norm": 1.934258222579956, "learning_rate": 7.466000000000001e-07, "loss": 0.0062, "step": 46268 }, { "epoch": 2.4667590766113983, "grad_norm": 2.9840340614318848, "learning_rate": 7.464000000000001e-07, "loss": 0.0199, "step": 46269 }, { "epoch": 2.4668123900410515, "grad_norm": 1.5909767150878906, "learning_rate": 7.462000000000002e-07, "loss": 0.0122, "step": 46270 }, { "epoch": 2.466865703470704, "grad_norm": 1.8312047719955444, "learning_rate": 7.46e-07, "loss": 0.0068, "step": 46271 }, { "epoch": 2.4669190169003574, "grad_norm": 3.990905284881592, "learning_rate": 7.458e-07, "loss": 0.0434, "step": 46272 }, { "epoch": 2.46697233033001, "grad_norm": 1.4874428510665894, "learning_rate": 7.456000000000001e-07, "loss": 0.0272, "step": 46273 }, { "epoch": 2.467025643759663, "grad_norm": 2.6218559741973877, "learning_rate": 7.454e-07, "loss": 0.025, "step": 46274 }, { "epoch": 2.467078957189316, "grad_norm": 3.416828155517578, "learning_rate": 7.452000000000001e-07, "loss": 0.0173, "step": 46275 }, { "epoch": 2.467132270618969, "grad_norm": 2.1440927982330322, "learning_rate": 7.450000000000001e-07, "loss": 0.0309, "step": 46276 }, { "epoch": 2.4671855840486216, "grad_norm": 1.8572391271591187, "learning_rate": 7.448000000000002e-07, "loss": 0.0221, "step": 46277 }, { "epoch": 2.467238897478275, "grad_norm": 2.2954444885253906, "learning_rate": 7.446e-07, "loss": 0.0081, "step": 46278 }, { "epoch": 2.4672922109079276, "grad_norm": 0.8951566815376282, "learning_rate": 7.444000000000001e-07, "loss": 0.0046, "step": 46279 }, { "epoch": 2.467345524337581, "grad_norm": 1.4537335634231567, "learning_rate": 7.442000000000001e-07, "loss": 0.0048, "step": 46280 }, { "epoch": 2.4673988377672336, "grad_norm": 0.8337725400924683, "learning_rate": 7.44e-07, "loss": 0.0044, "step": 46281 }, { "epoch": 2.4674521511968863, "grad_norm": 0.5493561625480652, "learning_rate": 7.438000000000001e-07, "loss": 0.0035, "step": 46282 }, { "epoch": 2.4675054646265395, "grad_norm": 1.2596523761749268, "learning_rate": 7.436000000000001e-07, "loss": 0.0231, "step": 46283 }, { "epoch": 2.4675587780561923, "grad_norm": 1.5021589994430542, "learning_rate": 7.434e-07, "loss": 0.008, "step": 46284 }, { "epoch": 2.4676120914858455, "grad_norm": 1.4745326042175293, "learning_rate": 7.432e-07, "loss": 0.0167, "step": 46285 }, { "epoch": 2.4676654049154982, "grad_norm": 1.8373819589614868, "learning_rate": 7.430000000000001e-07, "loss": 0.0116, "step": 46286 }, { "epoch": 2.467718718345151, "grad_norm": 2.1085619926452637, "learning_rate": 7.428e-07, "loss": 0.0104, "step": 46287 }, { "epoch": 2.467772031774804, "grad_norm": 1.6388638019561768, "learning_rate": 7.426000000000001e-07, "loss": 0.0122, "step": 46288 }, { "epoch": 2.467825345204457, "grad_norm": 0.9897454380989075, "learning_rate": 7.424000000000001e-07, "loss": 0.0155, "step": 46289 }, { "epoch": 2.4678786586341097, "grad_norm": 1.3847553730010986, "learning_rate": 7.422e-07, "loss": 0.0302, "step": 46290 }, { "epoch": 2.467931972063763, "grad_norm": 3.914109945297241, "learning_rate": 7.420000000000001e-07, "loss": 0.0507, "step": 46291 }, { "epoch": 2.4679852854934157, "grad_norm": 1.094016432762146, "learning_rate": 7.418e-07, "loss": 0.0051, "step": 46292 }, { "epoch": 2.468038598923069, "grad_norm": 1.6257137060165405, "learning_rate": 7.416000000000001e-07, "loss": 0.0211, "step": 46293 }, { "epoch": 2.4680919123527216, "grad_norm": 1.5267632007598877, "learning_rate": 7.414e-07, "loss": 0.009, "step": 46294 }, { "epoch": 2.468145225782375, "grad_norm": 2.4592320919036865, "learning_rate": 7.412000000000001e-07, "loss": 0.0259, "step": 46295 }, { "epoch": 2.4681985392120276, "grad_norm": 2.2136647701263428, "learning_rate": 7.410000000000001e-07, "loss": 0.0186, "step": 46296 }, { "epoch": 2.4682518526416803, "grad_norm": 2.863318681716919, "learning_rate": 7.408e-07, "loss": 0.008, "step": 46297 }, { "epoch": 2.4683051660713335, "grad_norm": 1.5845210552215576, "learning_rate": 7.406000000000001e-07, "loss": 0.0281, "step": 46298 }, { "epoch": 2.4683584795009863, "grad_norm": 1.57350754737854, "learning_rate": 7.404e-07, "loss": 0.0203, "step": 46299 }, { "epoch": 2.468411792930639, "grad_norm": 1.493154764175415, "learning_rate": 7.402000000000001e-07, "loss": 0.0031, "step": 46300 }, { "epoch": 2.4684651063602923, "grad_norm": 2.5898797512054443, "learning_rate": 7.4e-07, "loss": 0.0127, "step": 46301 }, { "epoch": 2.468518419789945, "grad_norm": 1.3750249147415161, "learning_rate": 7.398000000000001e-07, "loss": 0.0144, "step": 46302 }, { "epoch": 2.4685717332195978, "grad_norm": 2.0141854286193848, "learning_rate": 7.396000000000001e-07, "loss": 0.0205, "step": 46303 }, { "epoch": 2.468625046649251, "grad_norm": 1.3469700813293457, "learning_rate": 7.394000000000001e-07, "loss": 0.005, "step": 46304 }, { "epoch": 2.4686783600789037, "grad_norm": 1.7514296770095825, "learning_rate": 7.392000000000001e-07, "loss": 0.0319, "step": 46305 }, { "epoch": 2.468731673508557, "grad_norm": 1.7343026399612427, "learning_rate": 7.39e-07, "loss": 0.0105, "step": 46306 }, { "epoch": 2.4687849869382097, "grad_norm": 4.224492073059082, "learning_rate": 7.388e-07, "loss": 0.0465, "step": 46307 }, { "epoch": 2.468838300367863, "grad_norm": 3.5447800159454346, "learning_rate": 7.386e-07, "loss": 0.0221, "step": 46308 }, { "epoch": 2.4688916137975156, "grad_norm": 1.7544628381729126, "learning_rate": 7.384000000000001e-07, "loss": 0.0048, "step": 46309 }, { "epoch": 2.4689449272271684, "grad_norm": 1.914516568183899, "learning_rate": 7.382e-07, "loss": 0.0185, "step": 46310 }, { "epoch": 2.4689982406568216, "grad_norm": 1.8369548320770264, "learning_rate": 7.380000000000001e-07, "loss": 0.0065, "step": 46311 }, { "epoch": 2.4690515540864744, "grad_norm": 2.0405688285827637, "learning_rate": 7.378000000000001e-07, "loss": 0.0059, "step": 46312 }, { "epoch": 2.469104867516127, "grad_norm": 3.2859699726104736, "learning_rate": 7.376000000000001e-07, "loss": 0.0172, "step": 46313 }, { "epoch": 2.4691581809457803, "grad_norm": 0.9669773578643799, "learning_rate": 7.374e-07, "loss": 0.0054, "step": 46314 }, { "epoch": 2.469211494375433, "grad_norm": 1.7715116739273071, "learning_rate": 7.372e-07, "loss": 0.0072, "step": 46315 }, { "epoch": 2.4692648078050863, "grad_norm": 2.000256061553955, "learning_rate": 7.370000000000001e-07, "loss": 0.0237, "step": 46316 }, { "epoch": 2.469318121234739, "grad_norm": 0.9177915453910828, "learning_rate": 7.368e-07, "loss": 0.0048, "step": 46317 }, { "epoch": 2.469371434664392, "grad_norm": 2.2332684993743896, "learning_rate": 7.366000000000001e-07, "loss": 0.0116, "step": 46318 }, { "epoch": 2.469424748094045, "grad_norm": 2.9238979816436768, "learning_rate": 7.364000000000001e-07, "loss": 0.015, "step": 46319 }, { "epoch": 2.4694780615236978, "grad_norm": 3.1997385025024414, "learning_rate": 7.362000000000001e-07, "loss": 0.0214, "step": 46320 }, { "epoch": 2.469531374953351, "grad_norm": 1.4086122512817383, "learning_rate": 7.36e-07, "loss": 0.0211, "step": 46321 }, { "epoch": 2.4695846883830037, "grad_norm": 3.612297534942627, "learning_rate": 7.358000000000001e-07, "loss": 0.0176, "step": 46322 }, { "epoch": 2.4696380018126565, "grad_norm": 2.0586559772491455, "learning_rate": 7.356000000000001e-07, "loss": 0.0272, "step": 46323 }, { "epoch": 2.4696913152423097, "grad_norm": 3.8040199279785156, "learning_rate": 7.354e-07, "loss": 0.0202, "step": 46324 }, { "epoch": 2.4697446286719624, "grad_norm": 0.7829592227935791, "learning_rate": 7.352000000000001e-07, "loss": 0.0033, "step": 46325 }, { "epoch": 2.469797942101615, "grad_norm": 2.8642404079437256, "learning_rate": 7.350000000000001e-07, "loss": 0.0194, "step": 46326 }, { "epoch": 2.4698512555312684, "grad_norm": 0.6867783665657043, "learning_rate": 7.348000000000001e-07, "loss": 0.0037, "step": 46327 }, { "epoch": 2.469904568960921, "grad_norm": 1.8441951274871826, "learning_rate": 7.346e-07, "loss": 0.0101, "step": 46328 }, { "epoch": 2.4699578823905743, "grad_norm": 3.3426201343536377, "learning_rate": 7.344000000000001e-07, "loss": 0.0373, "step": 46329 }, { "epoch": 2.470011195820227, "grad_norm": 1.9945167303085327, "learning_rate": 7.342000000000001e-07, "loss": 0.0202, "step": 46330 }, { "epoch": 2.47006450924988, "grad_norm": 0.9649631381034851, "learning_rate": 7.340000000000001e-07, "loss": 0.0247, "step": 46331 }, { "epoch": 2.470117822679533, "grad_norm": 3.4945642948150635, "learning_rate": 7.338000000000001e-07, "loss": 0.0117, "step": 46332 }, { "epoch": 2.470171136109186, "grad_norm": 1.9163700342178345, "learning_rate": 7.336000000000001e-07, "loss": 0.0201, "step": 46333 }, { "epoch": 2.470224449538839, "grad_norm": 0.5142902731895447, "learning_rate": 7.334e-07, "loss": 0.0021, "step": 46334 }, { "epoch": 2.4702777629684918, "grad_norm": 1.3676855564117432, "learning_rate": 7.332e-07, "loss": 0.004, "step": 46335 }, { "epoch": 2.4703310763981445, "grad_norm": 1.740301489830017, "learning_rate": 7.330000000000001e-07, "loss": 0.032, "step": 46336 }, { "epoch": 2.4703843898277977, "grad_norm": 1.9245573282241821, "learning_rate": 7.328e-07, "loss": 0.0132, "step": 46337 }, { "epoch": 2.4704377032574505, "grad_norm": 1.6200566291809082, "learning_rate": 7.326000000000001e-07, "loss": 0.0149, "step": 46338 }, { "epoch": 2.4704910166871032, "grad_norm": 0.47276705503463745, "learning_rate": 7.324000000000001e-07, "loss": 0.0024, "step": 46339 }, { "epoch": 2.4705443301167564, "grad_norm": 2.653773069381714, "learning_rate": 7.322e-07, "loss": 0.021, "step": 46340 }, { "epoch": 2.470597643546409, "grad_norm": 2.0303256511688232, "learning_rate": 7.32e-07, "loss": 0.0175, "step": 46341 }, { "epoch": 2.4706509569760624, "grad_norm": 1.0167289972305298, "learning_rate": 7.318e-07, "loss": 0.004, "step": 46342 }, { "epoch": 2.470704270405715, "grad_norm": 2.6967740058898926, "learning_rate": 7.316000000000001e-07, "loss": 0.0376, "step": 46343 }, { "epoch": 2.4707575838353684, "grad_norm": 1.8089152574539185, "learning_rate": 7.314e-07, "loss": 0.0136, "step": 46344 }, { "epoch": 2.470810897265021, "grad_norm": 1.0531445741653442, "learning_rate": 7.312000000000001e-07, "loss": 0.0168, "step": 46345 }, { "epoch": 2.470864210694674, "grad_norm": 2.0927622318267822, "learning_rate": 7.310000000000001e-07, "loss": 0.0072, "step": 46346 }, { "epoch": 2.470917524124327, "grad_norm": 2.5011138916015625, "learning_rate": 7.308000000000002e-07, "loss": 0.0369, "step": 46347 }, { "epoch": 2.47097083755398, "grad_norm": 2.6540799140930176, "learning_rate": 7.306e-07, "loss": 0.0096, "step": 46348 }, { "epoch": 2.4710241509836326, "grad_norm": 1.6357120275497437, "learning_rate": 7.304e-07, "loss": 0.0109, "step": 46349 }, { "epoch": 2.471077464413286, "grad_norm": 3.0525593757629395, "learning_rate": 7.302000000000001e-07, "loss": 0.0402, "step": 46350 }, { "epoch": 2.4711307778429386, "grad_norm": 1.8050471544265747, "learning_rate": 7.3e-07, "loss": 0.0189, "step": 46351 }, { "epoch": 2.4711840912725918, "grad_norm": 0.8926911950111389, "learning_rate": 7.298000000000001e-07, "loss": 0.004, "step": 46352 }, { "epoch": 2.4712374047022445, "grad_norm": 0.1937684714794159, "learning_rate": 7.296000000000001e-07, "loss": 0.002, "step": 46353 }, { "epoch": 2.4712907181318973, "grad_norm": 1.2698572874069214, "learning_rate": 7.294000000000001e-07, "loss": 0.0064, "step": 46354 }, { "epoch": 2.4713440315615505, "grad_norm": 1.0379831790924072, "learning_rate": 7.292e-07, "loss": 0.0218, "step": 46355 }, { "epoch": 2.4713973449912032, "grad_norm": 0.9992457628250122, "learning_rate": 7.290000000000001e-07, "loss": 0.0044, "step": 46356 }, { "epoch": 2.4714506584208564, "grad_norm": 0.424012154340744, "learning_rate": 7.288e-07, "loss": 0.0033, "step": 46357 }, { "epoch": 2.471503971850509, "grad_norm": 1.3905209302902222, "learning_rate": 7.286e-07, "loss": 0.0103, "step": 46358 }, { "epoch": 2.471557285280162, "grad_norm": 1.259318470954895, "learning_rate": 7.284000000000001e-07, "loss": 0.0183, "step": 46359 }, { "epoch": 2.471610598709815, "grad_norm": 0.4766353666782379, "learning_rate": 7.282e-07, "loss": 0.003, "step": 46360 }, { "epoch": 2.471663912139468, "grad_norm": 2.1158487796783447, "learning_rate": 7.280000000000001e-07, "loss": 0.0093, "step": 46361 }, { "epoch": 2.4717172255691207, "grad_norm": 0.38831377029418945, "learning_rate": 7.278e-07, "loss": 0.0035, "step": 46362 }, { "epoch": 2.471770538998774, "grad_norm": 1.8717243671417236, "learning_rate": 7.276000000000001e-07, "loss": 0.0357, "step": 46363 }, { "epoch": 2.4718238524284266, "grad_norm": 2.5628626346588135, "learning_rate": 7.274e-07, "loss": 0.0152, "step": 46364 }, { "epoch": 2.47187716585808, "grad_norm": 1.7807163000106812, "learning_rate": 7.272000000000001e-07, "loss": 0.0307, "step": 46365 }, { "epoch": 2.4719304792877326, "grad_norm": 2.29376482963562, "learning_rate": 7.270000000000001e-07, "loss": 0.0297, "step": 46366 }, { "epoch": 2.4719837927173853, "grad_norm": 0.46432197093963623, "learning_rate": 7.268e-07, "loss": 0.0019, "step": 46367 }, { "epoch": 2.4720371061470385, "grad_norm": 0.685528576374054, "learning_rate": 7.266000000000001e-07, "loss": 0.0056, "step": 46368 }, { "epoch": 2.4720904195766913, "grad_norm": 1.5081578493118286, "learning_rate": 7.264e-07, "loss": 0.0116, "step": 46369 }, { "epoch": 2.4721437330063445, "grad_norm": 1.227105975151062, "learning_rate": 7.262000000000001e-07, "loss": 0.0142, "step": 46370 }, { "epoch": 2.4721970464359972, "grad_norm": 2.048200845718384, "learning_rate": 7.26e-07, "loss": 0.0253, "step": 46371 }, { "epoch": 2.47225035986565, "grad_norm": 3.076099395751953, "learning_rate": 7.258000000000001e-07, "loss": 0.0341, "step": 46372 }, { "epoch": 2.472303673295303, "grad_norm": 0.5356056094169617, "learning_rate": 7.256000000000001e-07, "loss": 0.0035, "step": 46373 }, { "epoch": 2.472356986724956, "grad_norm": 1.8397163152694702, "learning_rate": 7.254e-07, "loss": 0.0214, "step": 46374 }, { "epoch": 2.4724103001546087, "grad_norm": 3.199357509613037, "learning_rate": 7.252000000000001e-07, "loss": 0.0506, "step": 46375 }, { "epoch": 2.472463613584262, "grad_norm": 1.6426392793655396, "learning_rate": 7.25e-07, "loss": 0.0243, "step": 46376 }, { "epoch": 2.4725169270139147, "grad_norm": 1.6912480592727661, "learning_rate": 7.248000000000001e-07, "loss": 0.0075, "step": 46377 }, { "epoch": 2.472570240443568, "grad_norm": 0.3816417455673218, "learning_rate": 7.246e-07, "loss": 0.0027, "step": 46378 }, { "epoch": 2.4726235538732206, "grad_norm": 1.3369132280349731, "learning_rate": 7.244000000000001e-07, "loss": 0.0056, "step": 46379 }, { "epoch": 2.472676867302874, "grad_norm": 1.2566239833831787, "learning_rate": 7.242000000000001e-07, "loss": 0.0044, "step": 46380 }, { "epoch": 2.4727301807325266, "grad_norm": 1.402452826499939, "learning_rate": 7.240000000000001e-07, "loss": 0.0125, "step": 46381 }, { "epoch": 2.4727834941621794, "grad_norm": 0.6327972412109375, "learning_rate": 7.238000000000001e-07, "loss": 0.0036, "step": 46382 }, { "epoch": 2.4728368075918326, "grad_norm": 2.7883048057556152, "learning_rate": 7.236e-07, "loss": 0.011, "step": 46383 }, { "epoch": 2.4728901210214853, "grad_norm": 1.84060537815094, "learning_rate": 7.234e-07, "loss": 0.0189, "step": 46384 }, { "epoch": 2.472943434451138, "grad_norm": 3.3312761783599854, "learning_rate": 7.232e-07, "loss": 0.0644, "step": 46385 }, { "epoch": 2.4729967478807913, "grad_norm": 1.6200392246246338, "learning_rate": 7.230000000000001e-07, "loss": 0.0096, "step": 46386 }, { "epoch": 2.473050061310444, "grad_norm": 3.9974522590637207, "learning_rate": 7.228e-07, "loss": 0.0407, "step": 46387 }, { "epoch": 2.473103374740097, "grad_norm": 1.5752131938934326, "learning_rate": 7.226000000000001e-07, "loss": 0.0075, "step": 46388 }, { "epoch": 2.47315668816975, "grad_norm": 0.8831936120986938, "learning_rate": 7.224000000000001e-07, "loss": 0.0053, "step": 46389 }, { "epoch": 2.4732100015994027, "grad_norm": 2.509172201156616, "learning_rate": 7.222000000000002e-07, "loss": 0.0104, "step": 46390 }, { "epoch": 2.473263315029056, "grad_norm": 0.6601106524467468, "learning_rate": 7.22e-07, "loss": 0.003, "step": 46391 }, { "epoch": 2.4733166284587087, "grad_norm": 2.8329074382781982, "learning_rate": 7.218e-07, "loss": 0.0103, "step": 46392 }, { "epoch": 2.473369941888362, "grad_norm": 1.908889889717102, "learning_rate": 7.216000000000001e-07, "loss": 0.0269, "step": 46393 }, { "epoch": 2.4734232553180147, "grad_norm": 1.2654798030853271, "learning_rate": 7.214e-07, "loss": 0.0084, "step": 46394 }, { "epoch": 2.4734765687476674, "grad_norm": 0.639187753200531, "learning_rate": 7.212000000000001e-07, "loss": 0.0034, "step": 46395 }, { "epoch": 2.4735298821773206, "grad_norm": 1.46015465259552, "learning_rate": 7.210000000000001e-07, "loss": 0.0044, "step": 46396 }, { "epoch": 2.4735831956069734, "grad_norm": 2.089931011199951, "learning_rate": 7.208000000000002e-07, "loss": 0.0267, "step": 46397 }, { "epoch": 2.473636509036626, "grad_norm": 2.370636463165283, "learning_rate": 7.206e-07, "loss": 0.0301, "step": 46398 }, { "epoch": 2.4736898224662793, "grad_norm": 0.5866274833679199, "learning_rate": 7.204000000000001e-07, "loss": 0.0026, "step": 46399 }, { "epoch": 2.473743135895932, "grad_norm": 2.0403683185577393, "learning_rate": 7.202000000000001e-07, "loss": 0.0247, "step": 46400 }, { "epoch": 2.4737964493255853, "grad_norm": 1.2865761518478394, "learning_rate": 7.2e-07, "loss": 0.008, "step": 46401 }, { "epoch": 2.473849762755238, "grad_norm": 2.2320237159729004, "learning_rate": 7.198000000000001e-07, "loss": 0.0084, "step": 46402 }, { "epoch": 2.473903076184891, "grad_norm": 1.6214956045150757, "learning_rate": 7.196000000000001e-07, "loss": 0.0135, "step": 46403 }, { "epoch": 2.473956389614544, "grad_norm": 2.810842514038086, "learning_rate": 7.194e-07, "loss": 0.0289, "step": 46404 }, { "epoch": 2.4740097030441968, "grad_norm": 2.4841108322143555, "learning_rate": 7.192e-07, "loss": 0.0091, "step": 46405 }, { "epoch": 2.47406301647385, "grad_norm": 1.0377802848815918, "learning_rate": 7.190000000000001e-07, "loss": 0.0141, "step": 46406 }, { "epoch": 2.4741163299035027, "grad_norm": 1.9397156238555908, "learning_rate": 7.188e-07, "loss": 0.0329, "step": 46407 }, { "epoch": 2.4741696433331555, "grad_norm": 2.3674120903015137, "learning_rate": 7.186e-07, "loss": 0.0172, "step": 46408 }, { "epoch": 2.4742229567628087, "grad_norm": 1.8740228414535522, "learning_rate": 7.184000000000001e-07, "loss": 0.0146, "step": 46409 }, { "epoch": 2.4742762701924614, "grad_norm": 1.7624627351760864, "learning_rate": 7.182e-07, "loss": 0.0085, "step": 46410 }, { "epoch": 2.474329583622114, "grad_norm": 1.756652593612671, "learning_rate": 7.18e-07, "loss": 0.0138, "step": 46411 }, { "epoch": 2.4743828970517674, "grad_norm": 0.8915852308273315, "learning_rate": 7.178e-07, "loss": 0.0053, "step": 46412 }, { "epoch": 2.47443621048142, "grad_norm": 1.4491307735443115, "learning_rate": 7.176000000000001e-07, "loss": 0.0042, "step": 46413 }, { "epoch": 2.4744895239110734, "grad_norm": 1.967143177986145, "learning_rate": 7.174e-07, "loss": 0.0072, "step": 46414 }, { "epoch": 2.474542837340726, "grad_norm": 1.674034833908081, "learning_rate": 7.172000000000001e-07, "loss": 0.0046, "step": 46415 }, { "epoch": 2.474596150770379, "grad_norm": 2.0549309253692627, "learning_rate": 7.170000000000001e-07, "loss": 0.0072, "step": 46416 }, { "epoch": 2.474649464200032, "grad_norm": 1.6343967914581299, "learning_rate": 7.168e-07, "loss": 0.008, "step": 46417 }, { "epoch": 2.474702777629685, "grad_norm": 2.298816442489624, "learning_rate": 7.166e-07, "loss": 0.0174, "step": 46418 }, { "epoch": 2.474756091059338, "grad_norm": 1.4629994630813599, "learning_rate": 7.164e-07, "loss": 0.0049, "step": 46419 }, { "epoch": 2.474809404488991, "grad_norm": 0.4804946184158325, "learning_rate": 7.162000000000001e-07, "loss": 0.0036, "step": 46420 }, { "epoch": 2.4748627179186435, "grad_norm": 5.742250442504883, "learning_rate": 7.16e-07, "loss": 0.0276, "step": 46421 }, { "epoch": 2.4749160313482967, "grad_norm": 2.5730812549591064, "learning_rate": 7.158000000000001e-07, "loss": 0.0184, "step": 46422 }, { "epoch": 2.4749693447779495, "grad_norm": 3.033595085144043, "learning_rate": 7.156000000000001e-07, "loss": 0.0202, "step": 46423 }, { "epoch": 2.4750226582076023, "grad_norm": 1.0832599401474, "learning_rate": 7.154000000000002e-07, "loss": 0.0053, "step": 46424 }, { "epoch": 2.4750759716372555, "grad_norm": 2.511754274368286, "learning_rate": 7.152e-07, "loss": 0.005, "step": 46425 }, { "epoch": 2.475129285066908, "grad_norm": 0.48322591185569763, "learning_rate": 7.15e-07, "loss": 0.0029, "step": 46426 }, { "epoch": 2.4751825984965614, "grad_norm": 2.748107671737671, "learning_rate": 7.148000000000001e-07, "loss": 0.021, "step": 46427 }, { "epoch": 2.475235911926214, "grad_norm": 2.4323325157165527, "learning_rate": 7.146e-07, "loss": 0.0427, "step": 46428 }, { "epoch": 2.4752892253558674, "grad_norm": 1.8837945461273193, "learning_rate": 7.144000000000001e-07, "loss": 0.022, "step": 46429 }, { "epoch": 2.47534253878552, "grad_norm": 1.1858937740325928, "learning_rate": 7.142000000000001e-07, "loss": 0.0146, "step": 46430 }, { "epoch": 2.475395852215173, "grad_norm": 1.9534802436828613, "learning_rate": 7.140000000000001e-07, "loss": 0.02, "step": 46431 }, { "epoch": 2.475449165644826, "grad_norm": 1.614254117012024, "learning_rate": 7.138e-07, "loss": 0.0086, "step": 46432 }, { "epoch": 2.475502479074479, "grad_norm": 2.8435721397399902, "learning_rate": 7.136000000000001e-07, "loss": 0.0801, "step": 46433 }, { "epoch": 2.4755557925041316, "grad_norm": 2.7222323417663574, "learning_rate": 7.134e-07, "loss": 0.0286, "step": 46434 }, { "epoch": 2.475609105933785, "grad_norm": 2.2945809364318848, "learning_rate": 7.132e-07, "loss": 0.0215, "step": 46435 }, { "epoch": 2.4756624193634376, "grad_norm": 1.8401069641113281, "learning_rate": 7.130000000000001e-07, "loss": 0.0156, "step": 46436 }, { "epoch": 2.4757157327930908, "grad_norm": 3.057188034057617, "learning_rate": 7.128e-07, "loss": 0.0161, "step": 46437 }, { "epoch": 2.4757690462227435, "grad_norm": 0.5851011276245117, "learning_rate": 7.126000000000001e-07, "loss": 0.004, "step": 46438 }, { "epoch": 2.4758223596523963, "grad_norm": 1.0328398942947388, "learning_rate": 7.124e-07, "loss": 0.0032, "step": 46439 }, { "epoch": 2.4758756730820495, "grad_norm": 1.0668225288391113, "learning_rate": 7.122000000000001e-07, "loss": 0.0044, "step": 46440 }, { "epoch": 2.4759289865117022, "grad_norm": 2.8734960556030273, "learning_rate": 7.12e-07, "loss": 0.0336, "step": 46441 }, { "epoch": 2.4759822999413554, "grad_norm": 1.1586068868637085, "learning_rate": 7.118e-07, "loss": 0.0077, "step": 46442 }, { "epoch": 2.476035613371008, "grad_norm": 7.152093410491943, "learning_rate": 7.116000000000001e-07, "loss": 0.0504, "step": 46443 }, { "epoch": 2.476088926800661, "grad_norm": 0.29749539494514465, "learning_rate": 7.114e-07, "loss": 0.0022, "step": 46444 }, { "epoch": 2.476142240230314, "grad_norm": 3.0662105083465576, "learning_rate": 7.112000000000001e-07, "loss": 0.0187, "step": 46445 }, { "epoch": 2.476195553659967, "grad_norm": 1.0154533386230469, "learning_rate": 7.110000000000001e-07, "loss": 0.0047, "step": 46446 }, { "epoch": 2.4762488670896197, "grad_norm": 0.6492154598236084, "learning_rate": 7.108000000000001e-07, "loss": 0.0027, "step": 46447 }, { "epoch": 2.476302180519273, "grad_norm": 1.4193452596664429, "learning_rate": 7.106e-07, "loss": 0.0118, "step": 46448 }, { "epoch": 2.4763554939489256, "grad_norm": 1.3978395462036133, "learning_rate": 7.104000000000001e-07, "loss": 0.0096, "step": 46449 }, { "epoch": 2.476408807378579, "grad_norm": 2.046180248260498, "learning_rate": 7.102000000000001e-07, "loss": 0.019, "step": 46450 }, { "epoch": 2.4764621208082316, "grad_norm": 1.8904690742492676, "learning_rate": 7.1e-07, "loss": 0.0148, "step": 46451 }, { "epoch": 2.4765154342378843, "grad_norm": 4.66056489944458, "learning_rate": 7.098000000000001e-07, "loss": 0.0141, "step": 46452 }, { "epoch": 2.4765687476675375, "grad_norm": 1.467930793762207, "learning_rate": 7.096000000000001e-07, "loss": 0.0094, "step": 46453 }, { "epoch": 2.4766220610971903, "grad_norm": 1.531760573387146, "learning_rate": 7.094e-07, "loss": 0.0069, "step": 46454 }, { "epoch": 2.4766753745268435, "grad_norm": 1.7820420265197754, "learning_rate": 7.092e-07, "loss": 0.0093, "step": 46455 }, { "epoch": 2.4767286879564963, "grad_norm": 1.2414156198501587, "learning_rate": 7.090000000000001e-07, "loss": 0.0058, "step": 46456 }, { "epoch": 2.476782001386149, "grad_norm": 0.25081202387809753, "learning_rate": 7.088e-07, "loss": 0.0023, "step": 46457 }, { "epoch": 2.476835314815802, "grad_norm": 2.3409130573272705, "learning_rate": 7.086000000000001e-07, "loss": 0.0385, "step": 46458 }, { "epoch": 2.476888628245455, "grad_norm": 2.9330525398254395, "learning_rate": 7.084000000000001e-07, "loss": 0.0133, "step": 46459 }, { "epoch": 2.4769419416751077, "grad_norm": 2.163477897644043, "learning_rate": 7.082e-07, "loss": 0.0228, "step": 46460 }, { "epoch": 2.476995255104761, "grad_norm": 2.0577409267425537, "learning_rate": 7.08e-07, "loss": 0.01, "step": 46461 }, { "epoch": 2.4770485685344137, "grad_norm": 1.55344820022583, "learning_rate": 7.078e-07, "loss": 0.0336, "step": 46462 }, { "epoch": 2.477101881964067, "grad_norm": 1.447783350944519, "learning_rate": 7.076000000000001e-07, "loss": 0.0302, "step": 46463 }, { "epoch": 2.4771551953937196, "grad_norm": 1.4191030263900757, "learning_rate": 7.074e-07, "loss": 0.0136, "step": 46464 }, { "epoch": 2.477208508823373, "grad_norm": 1.8763890266418457, "learning_rate": 7.072000000000001e-07, "loss": 0.0083, "step": 46465 }, { "epoch": 2.4772618222530256, "grad_norm": 1.175520658493042, "learning_rate": 7.070000000000001e-07, "loss": 0.007, "step": 46466 }, { "epoch": 2.4773151356826784, "grad_norm": 2.7688307762145996, "learning_rate": 7.068000000000002e-07, "loss": 0.0137, "step": 46467 }, { "epoch": 2.4773684491123316, "grad_norm": 0.39029660820961, "learning_rate": 7.066e-07, "loss": 0.0032, "step": 46468 }, { "epoch": 2.4774217625419843, "grad_norm": 1.6325322389602661, "learning_rate": 7.064e-07, "loss": 0.0367, "step": 46469 }, { "epoch": 2.477475075971637, "grad_norm": 1.6481939554214478, "learning_rate": 7.062000000000001e-07, "loss": 0.0061, "step": 46470 }, { "epoch": 2.4775283894012903, "grad_norm": 1.3024119138717651, "learning_rate": 7.06e-07, "loss": 0.0083, "step": 46471 }, { "epoch": 2.477581702830943, "grad_norm": 1.5437017679214478, "learning_rate": 7.058000000000001e-07, "loss": 0.018, "step": 46472 }, { "epoch": 2.4776350162605962, "grad_norm": 3.1725852489471436, "learning_rate": 7.056000000000001e-07, "loss": 0.0434, "step": 46473 }, { "epoch": 2.477688329690249, "grad_norm": 0.8896492719650269, "learning_rate": 7.054000000000002e-07, "loss": 0.0048, "step": 46474 }, { "epoch": 2.4777416431199017, "grad_norm": 0.9856719374656677, "learning_rate": 7.052e-07, "loss": 0.0128, "step": 46475 }, { "epoch": 2.477794956549555, "grad_norm": 1.6437852382659912, "learning_rate": 7.05e-07, "loss": 0.0298, "step": 46476 }, { "epoch": 2.4778482699792077, "grad_norm": 1.1134852170944214, "learning_rate": 7.048000000000001e-07, "loss": 0.0052, "step": 46477 }, { "epoch": 2.477901583408861, "grad_norm": 3.3180935382843018, "learning_rate": 7.046e-07, "loss": 0.0235, "step": 46478 }, { "epoch": 2.4779548968385137, "grad_norm": 1.5908994674682617, "learning_rate": 7.044000000000001e-07, "loss": 0.0073, "step": 46479 }, { "epoch": 2.4780082102681664, "grad_norm": 1.8340696096420288, "learning_rate": 7.042000000000001e-07, "loss": 0.0382, "step": 46480 }, { "epoch": 2.4780615236978196, "grad_norm": 2.691918134689331, "learning_rate": 7.040000000000001e-07, "loss": 0.0154, "step": 46481 }, { "epoch": 2.4781148371274724, "grad_norm": 2.1105008125305176, "learning_rate": 7.038e-07, "loss": 0.0175, "step": 46482 }, { "epoch": 2.478168150557125, "grad_norm": 1.841587781906128, "learning_rate": 7.036000000000001e-07, "loss": 0.0143, "step": 46483 }, { "epoch": 2.4782214639867783, "grad_norm": 0.8922198414802551, "learning_rate": 7.034e-07, "loss": 0.0037, "step": 46484 }, { "epoch": 2.478274777416431, "grad_norm": 0.860573947429657, "learning_rate": 7.032e-07, "loss": 0.0052, "step": 46485 }, { "epoch": 2.4783280908460843, "grad_norm": 1.0389906167984009, "learning_rate": 7.030000000000001e-07, "loss": 0.0042, "step": 46486 }, { "epoch": 2.478381404275737, "grad_norm": 1.7086992263793945, "learning_rate": 7.028e-07, "loss": 0.018, "step": 46487 }, { "epoch": 2.47843471770539, "grad_norm": 3.041468381881714, "learning_rate": 7.026000000000001e-07, "loss": 0.0143, "step": 46488 }, { "epoch": 2.478488031135043, "grad_norm": 3.302903175354004, "learning_rate": 7.024e-07, "loss": 0.0175, "step": 46489 }, { "epoch": 2.4785413445646958, "grad_norm": 2.3061022758483887, "learning_rate": 7.022000000000001e-07, "loss": 0.0066, "step": 46490 }, { "epoch": 2.478594657994349, "grad_norm": 5.047746658325195, "learning_rate": 7.02e-07, "loss": 0.0333, "step": 46491 }, { "epoch": 2.4786479714240017, "grad_norm": 2.2846884727478027, "learning_rate": 7.018000000000001e-07, "loss": 0.0291, "step": 46492 }, { "epoch": 2.4787012848536545, "grad_norm": 0.38270121812820435, "learning_rate": 7.016000000000001e-07, "loss": 0.0035, "step": 46493 }, { "epoch": 2.4787545982833077, "grad_norm": 0.8759312629699707, "learning_rate": 7.014e-07, "loss": 0.0051, "step": 46494 }, { "epoch": 2.4788079117129604, "grad_norm": 0.6804826855659485, "learning_rate": 7.012000000000001e-07, "loss": 0.0058, "step": 46495 }, { "epoch": 2.478861225142613, "grad_norm": 4.793531894683838, "learning_rate": 7.01e-07, "loss": 0.0113, "step": 46496 }, { "epoch": 2.4789145385722664, "grad_norm": 2.31044602394104, "learning_rate": 7.008000000000001e-07, "loss": 0.0122, "step": 46497 }, { "epoch": 2.478967852001919, "grad_norm": 0.7976034879684448, "learning_rate": 7.006e-07, "loss": 0.0038, "step": 46498 }, { "epoch": 2.4790211654315724, "grad_norm": 0.9831572771072388, "learning_rate": 7.004000000000001e-07, "loss": 0.0058, "step": 46499 }, { "epoch": 2.479074478861225, "grad_norm": 3.246860980987549, "learning_rate": 7.002000000000001e-07, "loss": 0.0253, "step": 46500 }, { "epoch": 2.4791277922908783, "grad_norm": 1.6670256853103638, "learning_rate": 7.000000000000001e-07, "loss": 0.0244, "step": 46501 }, { "epoch": 2.479181105720531, "grad_norm": 1.7437463998794556, "learning_rate": 6.998000000000001e-07, "loss": 0.0115, "step": 46502 }, { "epoch": 2.479234419150184, "grad_norm": 1.8866021633148193, "learning_rate": 6.996e-07, "loss": 0.0089, "step": 46503 }, { "epoch": 2.479287732579837, "grad_norm": 2.2394049167633057, "learning_rate": 6.994e-07, "loss": 0.0226, "step": 46504 }, { "epoch": 2.47934104600949, "grad_norm": 1.6448088884353638, "learning_rate": 6.992e-07, "loss": 0.0202, "step": 46505 }, { "epoch": 2.4793943594391425, "grad_norm": 1.4472017288208008, "learning_rate": 6.990000000000001e-07, "loss": 0.0045, "step": 46506 }, { "epoch": 2.4794476728687957, "grad_norm": 3.258389472961426, "learning_rate": 6.988e-07, "loss": 0.0305, "step": 46507 }, { "epoch": 2.4795009862984485, "grad_norm": 1.4288116693496704, "learning_rate": 6.986000000000001e-07, "loss": 0.0078, "step": 46508 }, { "epoch": 2.4795542997281013, "grad_norm": 0.6863000988960266, "learning_rate": 6.984000000000001e-07, "loss": 0.0032, "step": 46509 }, { "epoch": 2.4796076131577545, "grad_norm": 3.1836884021759033, "learning_rate": 6.981999999999999e-07, "loss": 0.0116, "step": 46510 }, { "epoch": 2.479660926587407, "grad_norm": 3.5707435607910156, "learning_rate": 6.98e-07, "loss": 0.033, "step": 46511 }, { "epoch": 2.4797142400170604, "grad_norm": 1.312575340270996, "learning_rate": 6.978e-07, "loss": 0.0186, "step": 46512 }, { "epoch": 2.479767553446713, "grad_norm": 2.3650832176208496, "learning_rate": 6.976000000000001e-07, "loss": 0.0123, "step": 46513 }, { "epoch": 2.4798208668763664, "grad_norm": 1.334906816482544, "learning_rate": 6.974e-07, "loss": 0.0071, "step": 46514 }, { "epoch": 2.479874180306019, "grad_norm": 1.340522050857544, "learning_rate": 6.972000000000001e-07, "loss": 0.0089, "step": 46515 }, { "epoch": 2.479927493735672, "grad_norm": 5.640796661376953, "learning_rate": 6.970000000000001e-07, "loss": 0.0292, "step": 46516 }, { "epoch": 2.479980807165325, "grad_norm": 3.534213066101074, "learning_rate": 6.968000000000002e-07, "loss": 0.0128, "step": 46517 }, { "epoch": 2.480034120594978, "grad_norm": 2.846454620361328, "learning_rate": 6.966e-07, "loss": 0.031, "step": 46518 }, { "epoch": 2.4800874340246306, "grad_norm": 5.073848247528076, "learning_rate": 6.964e-07, "loss": 0.0091, "step": 46519 }, { "epoch": 2.480140747454284, "grad_norm": 2.855722665786743, "learning_rate": 6.962000000000001e-07, "loss": 0.0518, "step": 46520 }, { "epoch": 2.4801940608839366, "grad_norm": 2.9252467155456543, "learning_rate": 6.96e-07, "loss": 0.0156, "step": 46521 }, { "epoch": 2.4802473743135898, "grad_norm": 1.5099138021469116, "learning_rate": 6.958000000000001e-07, "loss": 0.015, "step": 46522 }, { "epoch": 2.4803006877432425, "grad_norm": 2.185468912124634, "learning_rate": 6.956000000000001e-07, "loss": 0.0147, "step": 46523 }, { "epoch": 2.4803540011728953, "grad_norm": 1.9046177864074707, "learning_rate": 6.954000000000002e-07, "loss": 0.0193, "step": 46524 }, { "epoch": 2.4804073146025485, "grad_norm": 1.972630262374878, "learning_rate": 6.952e-07, "loss": 0.0323, "step": 46525 }, { "epoch": 2.4804606280322012, "grad_norm": 1.4577327966690063, "learning_rate": 6.950000000000001e-07, "loss": 0.0098, "step": 46526 }, { "epoch": 2.4805139414618544, "grad_norm": 2.317761182785034, "learning_rate": 6.948000000000001e-07, "loss": 0.0208, "step": 46527 }, { "epoch": 2.480567254891507, "grad_norm": 2.9740171432495117, "learning_rate": 6.946e-07, "loss": 0.0232, "step": 46528 }, { "epoch": 2.48062056832116, "grad_norm": 1.9213119745254517, "learning_rate": 6.944000000000001e-07, "loss": 0.0191, "step": 46529 }, { "epoch": 2.480673881750813, "grad_norm": 3.303311586380005, "learning_rate": 6.942000000000001e-07, "loss": 0.0208, "step": 46530 }, { "epoch": 2.480727195180466, "grad_norm": 1.7947402000427246, "learning_rate": 6.94e-07, "loss": 0.0065, "step": 46531 }, { "epoch": 2.4807805086101187, "grad_norm": 2.271090507507324, "learning_rate": 6.938e-07, "loss": 0.0177, "step": 46532 }, { "epoch": 2.480833822039772, "grad_norm": 0.955517590045929, "learning_rate": 6.936000000000001e-07, "loss": 0.0044, "step": 46533 }, { "epoch": 2.4808871354694246, "grad_norm": 0.7731850147247314, "learning_rate": 6.934e-07, "loss": 0.0035, "step": 46534 }, { "epoch": 2.480940448899078, "grad_norm": 1.5807161331176758, "learning_rate": 6.932000000000001e-07, "loss": 0.0188, "step": 46535 }, { "epoch": 2.4809937623287306, "grad_norm": 0.6646716594696045, "learning_rate": 6.930000000000001e-07, "loss": 0.0037, "step": 46536 }, { "epoch": 2.4810470757583833, "grad_norm": 1.3411448001861572, "learning_rate": 6.928e-07, "loss": 0.0069, "step": 46537 }, { "epoch": 2.4811003891880365, "grad_norm": 1.3145296573638916, "learning_rate": 6.926e-07, "loss": 0.008, "step": 46538 }, { "epoch": 2.4811537026176893, "grad_norm": 1.359208106994629, "learning_rate": 6.924e-07, "loss": 0.0066, "step": 46539 }, { "epoch": 2.4812070160473425, "grad_norm": 0.9725591540336609, "learning_rate": 6.922000000000001e-07, "loss": 0.0046, "step": 46540 }, { "epoch": 2.4812603294769953, "grad_norm": 2.2468488216400146, "learning_rate": 6.92e-07, "loss": 0.0181, "step": 46541 }, { "epoch": 2.481313642906648, "grad_norm": 3.135267496109009, "learning_rate": 6.918000000000001e-07, "loss": 0.0099, "step": 46542 }, { "epoch": 2.481366956336301, "grad_norm": 2.7344884872436523, "learning_rate": 6.916000000000001e-07, "loss": 0.0272, "step": 46543 }, { "epoch": 2.481420269765954, "grad_norm": 0.9681714773178101, "learning_rate": 6.914e-07, "loss": 0.016, "step": 46544 }, { "epoch": 2.4814735831956067, "grad_norm": 1.0200029611587524, "learning_rate": 6.912e-07, "loss": 0.0274, "step": 46545 }, { "epoch": 2.48152689662526, "grad_norm": 1.00283944606781, "learning_rate": 6.91e-07, "loss": 0.0139, "step": 46546 }, { "epoch": 2.4815802100549127, "grad_norm": 1.7752676010131836, "learning_rate": 6.908000000000001e-07, "loss": 0.0212, "step": 46547 }, { "epoch": 2.481633523484566, "grad_norm": 2.941006660461426, "learning_rate": 6.906e-07, "loss": 0.0245, "step": 46548 }, { "epoch": 2.4816868369142187, "grad_norm": 0.7896276116371155, "learning_rate": 6.904000000000001e-07, "loss": 0.0039, "step": 46549 }, { "epoch": 2.481740150343872, "grad_norm": 0.577214777469635, "learning_rate": 6.902000000000001e-07, "loss": 0.0029, "step": 46550 }, { "epoch": 2.4817934637735246, "grad_norm": 0.7898439764976501, "learning_rate": 6.900000000000001e-07, "loss": 0.0039, "step": 46551 }, { "epoch": 2.4818467772031774, "grad_norm": 1.038501262664795, "learning_rate": 6.898e-07, "loss": 0.0164, "step": 46552 }, { "epoch": 2.4819000906328306, "grad_norm": 2.235936403274536, "learning_rate": 6.896e-07, "loss": 0.0283, "step": 46553 }, { "epoch": 2.4819534040624833, "grad_norm": 0.8667845726013184, "learning_rate": 6.894e-07, "loss": 0.0205, "step": 46554 }, { "epoch": 2.482006717492136, "grad_norm": 0.8775776028633118, "learning_rate": 6.892e-07, "loss": 0.0031, "step": 46555 }, { "epoch": 2.4820600309217893, "grad_norm": 2.2632741928100586, "learning_rate": 6.890000000000001e-07, "loss": 0.008, "step": 46556 }, { "epoch": 2.482113344351442, "grad_norm": 2.6162338256835938, "learning_rate": 6.888e-07, "loss": 0.0135, "step": 46557 }, { "epoch": 2.4821666577810952, "grad_norm": 0.3043668270111084, "learning_rate": 6.886000000000001e-07, "loss": 0.0027, "step": 46558 }, { "epoch": 2.482219971210748, "grad_norm": 1.5470963716506958, "learning_rate": 6.884e-07, "loss": 0.0253, "step": 46559 }, { "epoch": 2.4822732846404008, "grad_norm": 2.111443042755127, "learning_rate": 6.882000000000001e-07, "loss": 0.0158, "step": 46560 }, { "epoch": 2.482326598070054, "grad_norm": 5.2744221687316895, "learning_rate": 6.88e-07, "loss": 0.0144, "step": 46561 }, { "epoch": 2.4823799114997067, "grad_norm": 1.5332030057907104, "learning_rate": 6.878e-07, "loss": 0.0102, "step": 46562 }, { "epoch": 2.48243322492936, "grad_norm": 1.6272671222686768, "learning_rate": 6.876000000000001e-07, "loss": 0.0063, "step": 46563 }, { "epoch": 2.4824865383590127, "grad_norm": 0.43438026309013367, "learning_rate": 6.874e-07, "loss": 0.0032, "step": 46564 }, { "epoch": 2.4825398517886654, "grad_norm": 1.4639781713485718, "learning_rate": 6.872000000000001e-07, "loss": 0.0136, "step": 46565 }, { "epoch": 2.4825931652183186, "grad_norm": 3.023132085800171, "learning_rate": 6.87e-07, "loss": 0.0271, "step": 46566 }, { "epoch": 2.4826464786479714, "grad_norm": 2.7326834201812744, "learning_rate": 6.868000000000001e-07, "loss": 0.0131, "step": 46567 }, { "epoch": 2.482699792077624, "grad_norm": 1.6207036972045898, "learning_rate": 6.866e-07, "loss": 0.038, "step": 46568 }, { "epoch": 2.4827531055072773, "grad_norm": 2.8320772647857666, "learning_rate": 6.864000000000001e-07, "loss": 0.0418, "step": 46569 }, { "epoch": 2.48280641893693, "grad_norm": 2.766558885574341, "learning_rate": 6.862000000000001e-07, "loss": 0.0274, "step": 46570 }, { "epoch": 2.4828597323665833, "grad_norm": 2.4136674404144287, "learning_rate": 6.86e-07, "loss": 0.016, "step": 46571 }, { "epoch": 2.482913045796236, "grad_norm": 1.6551097631454468, "learning_rate": 6.858000000000001e-07, "loss": 0.012, "step": 46572 }, { "epoch": 2.482966359225889, "grad_norm": 3.3449933528900146, "learning_rate": 6.856e-07, "loss": 0.0229, "step": 46573 }, { "epoch": 2.483019672655542, "grad_norm": 3.306692361831665, "learning_rate": 6.854000000000001e-07, "loss": 0.029, "step": 46574 }, { "epoch": 2.4830729860851948, "grad_norm": 3.2155022621154785, "learning_rate": 6.852e-07, "loss": 0.0279, "step": 46575 }, { "epoch": 2.483126299514848, "grad_norm": 1.7933377027511597, "learning_rate": 6.850000000000001e-07, "loss": 0.0183, "step": 46576 }, { "epoch": 2.4831796129445007, "grad_norm": 1.7544593811035156, "learning_rate": 6.848000000000001e-07, "loss": 0.023, "step": 46577 }, { "epoch": 2.4832329263741535, "grad_norm": 2.883685827255249, "learning_rate": 6.846e-07, "loss": 0.0301, "step": 46578 }, { "epoch": 2.4832862398038067, "grad_norm": 0.5759098529815674, "learning_rate": 6.844000000000001e-07, "loss": 0.0033, "step": 46579 }, { "epoch": 2.4833395532334595, "grad_norm": 1.4355934858322144, "learning_rate": 6.842e-07, "loss": 0.0072, "step": 46580 }, { "epoch": 2.483392866663112, "grad_norm": 2.4544458389282227, "learning_rate": 6.84e-07, "loss": 0.0118, "step": 46581 }, { "epoch": 2.4834461800927654, "grad_norm": 1.0419199466705322, "learning_rate": 6.838e-07, "loss": 0.0064, "step": 46582 }, { "epoch": 2.483499493522418, "grad_norm": 2.4750173091888428, "learning_rate": 6.836000000000001e-07, "loss": 0.0092, "step": 46583 }, { "epoch": 2.4835528069520714, "grad_norm": 2.4460790157318115, "learning_rate": 6.834e-07, "loss": 0.0223, "step": 46584 }, { "epoch": 2.483606120381724, "grad_norm": 0.7235485315322876, "learning_rate": 6.832000000000001e-07, "loss": 0.003, "step": 46585 }, { "epoch": 2.4836594338113773, "grad_norm": 1.976474404335022, "learning_rate": 6.830000000000001e-07, "loss": 0.0228, "step": 46586 }, { "epoch": 2.48371274724103, "grad_norm": 2.074079751968384, "learning_rate": 6.827999999999999e-07, "loss": 0.0344, "step": 46587 }, { "epoch": 2.483766060670683, "grad_norm": 1.875996470451355, "learning_rate": 6.826e-07, "loss": 0.0095, "step": 46588 }, { "epoch": 2.483819374100336, "grad_norm": 2.417571544647217, "learning_rate": 6.824e-07, "loss": 0.0171, "step": 46589 }, { "epoch": 2.483872687529989, "grad_norm": 1.0375216007232666, "learning_rate": 6.822000000000001e-07, "loss": 0.0038, "step": 46590 }, { "epoch": 2.4839260009596416, "grad_norm": 2.86968731880188, "learning_rate": 6.82e-07, "loss": 0.0146, "step": 46591 }, { "epoch": 2.4839793143892948, "grad_norm": 2.541170597076416, "learning_rate": 6.818000000000001e-07, "loss": 0.0103, "step": 46592 }, { "epoch": 2.4840326278189475, "grad_norm": 1.4301953315734863, "learning_rate": 6.816000000000001e-07, "loss": 0.0189, "step": 46593 }, { "epoch": 2.4840859412486007, "grad_norm": 5.30139684677124, "learning_rate": 6.814000000000002e-07, "loss": 0.0495, "step": 46594 }, { "epoch": 2.4841392546782535, "grad_norm": 2.052868604660034, "learning_rate": 6.812e-07, "loss": 0.0179, "step": 46595 }, { "epoch": 2.4841925681079062, "grad_norm": 2.7313895225524902, "learning_rate": 6.81e-07, "loss": 0.0072, "step": 46596 }, { "epoch": 2.4842458815375594, "grad_norm": 3.8763082027435303, "learning_rate": 6.808000000000001e-07, "loss": 0.0139, "step": 46597 }, { "epoch": 2.484299194967212, "grad_norm": 0.8876359462738037, "learning_rate": 6.806e-07, "loss": 0.0155, "step": 46598 }, { "epoch": 2.4843525083968654, "grad_norm": 2.5713491439819336, "learning_rate": 6.804000000000001e-07, "loss": 0.0258, "step": 46599 }, { "epoch": 2.484405821826518, "grad_norm": 2.100735902786255, "learning_rate": 6.802000000000001e-07, "loss": 0.0324, "step": 46600 }, { "epoch": 2.484459135256171, "grad_norm": 3.357168674468994, "learning_rate": 6.800000000000001e-07, "loss": 0.0278, "step": 46601 }, { "epoch": 2.484512448685824, "grad_norm": 1.6678767204284668, "learning_rate": 6.798e-07, "loss": 0.0144, "step": 46602 }, { "epoch": 2.484565762115477, "grad_norm": 1.5995301008224487, "learning_rate": 6.796000000000001e-07, "loss": 0.0219, "step": 46603 }, { "epoch": 2.4846190755451296, "grad_norm": 2.764641761779785, "learning_rate": 6.794e-07, "loss": 0.0208, "step": 46604 }, { "epoch": 2.484672388974783, "grad_norm": 1.4577150344848633, "learning_rate": 6.792e-07, "loss": 0.0161, "step": 46605 }, { "epoch": 2.4847257024044356, "grad_norm": 0.27372753620147705, "learning_rate": 6.790000000000001e-07, "loss": 0.0022, "step": 46606 }, { "epoch": 2.4847790158340888, "grad_norm": 1.3232189416885376, "learning_rate": 6.788e-07, "loss": 0.0268, "step": 46607 }, { "epoch": 2.4848323292637415, "grad_norm": 0.48614150285720825, "learning_rate": 6.786000000000001e-07, "loss": 0.0033, "step": 46608 }, { "epoch": 2.4848856426933943, "grad_norm": 1.617021083831787, "learning_rate": 6.784e-07, "loss": 0.0227, "step": 46609 }, { "epoch": 2.4849389561230475, "grad_norm": 1.7588555812835693, "learning_rate": 6.782000000000001e-07, "loss": 0.0083, "step": 46610 }, { "epoch": 2.4849922695527003, "grad_norm": 5.268894672393799, "learning_rate": 6.78e-07, "loss": 0.0478, "step": 46611 }, { "epoch": 2.4850455829823535, "grad_norm": 1.0290157794952393, "learning_rate": 6.778000000000001e-07, "loss": 0.0032, "step": 46612 }, { "epoch": 2.485098896412006, "grad_norm": 2.498535394668579, "learning_rate": 6.776000000000001e-07, "loss": 0.0263, "step": 46613 }, { "epoch": 2.485152209841659, "grad_norm": 1.9619615077972412, "learning_rate": 6.774e-07, "loss": 0.0088, "step": 46614 }, { "epoch": 2.485205523271312, "grad_norm": 0.6099340915679932, "learning_rate": 6.772000000000001e-07, "loss": 0.0028, "step": 46615 }, { "epoch": 2.485258836700965, "grad_norm": 2.2517712116241455, "learning_rate": 6.77e-07, "loss": 0.0095, "step": 46616 }, { "epoch": 2.4853121501306177, "grad_norm": 1.2576712369918823, "learning_rate": 6.768000000000001e-07, "loss": 0.0051, "step": 46617 }, { "epoch": 2.485365463560271, "grad_norm": 2.169430732727051, "learning_rate": 6.766e-07, "loss": 0.0094, "step": 46618 }, { "epoch": 2.4854187769899236, "grad_norm": 2.3625402450561523, "learning_rate": 6.764000000000001e-07, "loss": 0.0306, "step": 46619 }, { "epoch": 2.485472090419577, "grad_norm": 4.265292167663574, "learning_rate": 6.762000000000001e-07, "loss": 0.0158, "step": 46620 }, { "epoch": 2.4855254038492296, "grad_norm": 1.2243930101394653, "learning_rate": 6.76e-07, "loss": 0.0058, "step": 46621 }, { "epoch": 2.485578717278883, "grad_norm": 0.8897668123245239, "learning_rate": 6.758000000000001e-07, "loss": 0.0031, "step": 46622 }, { "epoch": 2.4856320307085356, "grad_norm": 2.334094285964966, "learning_rate": 6.756e-07, "loss": 0.019, "step": 46623 }, { "epoch": 2.4856853441381883, "grad_norm": 3.072572946548462, "learning_rate": 6.754000000000001e-07, "loss": 0.0248, "step": 46624 }, { "epoch": 2.4857386575678415, "grad_norm": 2.3376119136810303, "learning_rate": 6.752e-07, "loss": 0.0315, "step": 46625 }, { "epoch": 2.4857919709974943, "grad_norm": 1.2482990026474, "learning_rate": 6.750000000000001e-07, "loss": 0.0079, "step": 46626 }, { "epoch": 2.485845284427147, "grad_norm": 2.134295701980591, "learning_rate": 6.748000000000001e-07, "loss": 0.0233, "step": 46627 }, { "epoch": 2.4858985978568002, "grad_norm": 0.9410929083824158, "learning_rate": 6.746000000000001e-07, "loss": 0.0053, "step": 46628 }, { "epoch": 2.485951911286453, "grad_norm": 1.6329145431518555, "learning_rate": 6.744000000000001e-07, "loss": 0.0111, "step": 46629 }, { "epoch": 2.4860052247161057, "grad_norm": 2.2459194660186768, "learning_rate": 6.742e-07, "loss": 0.012, "step": 46630 }, { "epoch": 2.486058538145759, "grad_norm": 2.8379437923431396, "learning_rate": 6.74e-07, "loss": 0.0156, "step": 46631 }, { "epoch": 2.4861118515754117, "grad_norm": 1.6304758787155151, "learning_rate": 6.738e-07, "loss": 0.013, "step": 46632 }, { "epoch": 2.486165165005065, "grad_norm": 1.6267765760421753, "learning_rate": 6.736000000000001e-07, "loss": 0.0334, "step": 46633 }, { "epoch": 2.4862184784347177, "grad_norm": 3.180558443069458, "learning_rate": 6.734e-07, "loss": 0.0136, "step": 46634 }, { "epoch": 2.486271791864371, "grad_norm": 1.063597321510315, "learning_rate": 6.732000000000001e-07, "loss": 0.0044, "step": 46635 }, { "epoch": 2.4863251052940236, "grad_norm": 0.7368273138999939, "learning_rate": 6.730000000000001e-07, "loss": 0.0149, "step": 46636 }, { "epoch": 2.4863784187236764, "grad_norm": 1.23737370967865, "learning_rate": 6.728000000000001e-07, "loss": 0.0057, "step": 46637 }, { "epoch": 2.4864317321533296, "grad_norm": 2.7820851802825928, "learning_rate": 6.726e-07, "loss": 0.0307, "step": 46638 }, { "epoch": 2.4864850455829823, "grad_norm": 1.0411348342895508, "learning_rate": 6.724e-07, "loss": 0.0248, "step": 46639 }, { "epoch": 2.486538359012635, "grad_norm": 1.1412733793258667, "learning_rate": 6.722000000000001e-07, "loss": 0.0063, "step": 46640 }, { "epoch": 2.4865916724422883, "grad_norm": 2.3776168823242188, "learning_rate": 6.72e-07, "loss": 0.0179, "step": 46641 }, { "epoch": 2.486644985871941, "grad_norm": 1.5240575075149536, "learning_rate": 6.718000000000001e-07, "loss": 0.0209, "step": 46642 }, { "epoch": 2.4866982993015943, "grad_norm": 1.924958348274231, "learning_rate": 6.716000000000001e-07, "loss": 0.0179, "step": 46643 }, { "epoch": 2.486751612731247, "grad_norm": 1.918637752532959, "learning_rate": 6.714000000000001e-07, "loss": 0.0282, "step": 46644 }, { "epoch": 2.4868049261608998, "grad_norm": 1.471547245979309, "learning_rate": 6.712e-07, "loss": 0.0105, "step": 46645 }, { "epoch": 2.486858239590553, "grad_norm": 2.0960099697113037, "learning_rate": 6.710000000000001e-07, "loss": 0.0072, "step": 46646 }, { "epoch": 2.4869115530202057, "grad_norm": 2.2142670154571533, "learning_rate": 6.708000000000001e-07, "loss": 0.012, "step": 46647 }, { "epoch": 2.486964866449859, "grad_norm": 2.381441116333008, "learning_rate": 6.706e-07, "loss": 0.0333, "step": 46648 }, { "epoch": 2.4870181798795117, "grad_norm": 1.6374529600143433, "learning_rate": 6.704000000000001e-07, "loss": 0.0107, "step": 46649 }, { "epoch": 2.4870714933091644, "grad_norm": 3.360219717025757, "learning_rate": 6.702000000000001e-07, "loss": 0.0144, "step": 46650 }, { "epoch": 2.4871248067388176, "grad_norm": 2.5920467376708984, "learning_rate": 6.7e-07, "loss": 0.0071, "step": 46651 }, { "epoch": 2.4871781201684704, "grad_norm": 0.9151296615600586, "learning_rate": 6.698e-07, "loss": 0.0036, "step": 46652 }, { "epoch": 2.487231433598123, "grad_norm": 3.882763385772705, "learning_rate": 6.696000000000001e-07, "loss": 0.033, "step": 46653 }, { "epoch": 2.4872847470277764, "grad_norm": 0.4218907058238983, "learning_rate": 6.694e-07, "loss": 0.002, "step": 46654 }, { "epoch": 2.487338060457429, "grad_norm": 1.8042380809783936, "learning_rate": 6.692e-07, "loss": 0.0237, "step": 46655 }, { "epoch": 2.4873913738870823, "grad_norm": 1.5934728384017944, "learning_rate": 6.690000000000001e-07, "loss": 0.0097, "step": 46656 }, { "epoch": 2.487444687316735, "grad_norm": 1.4059184789657593, "learning_rate": 6.688e-07, "loss": 0.0066, "step": 46657 }, { "epoch": 2.487498000746388, "grad_norm": 1.8716819286346436, "learning_rate": 6.686e-07, "loss": 0.0197, "step": 46658 }, { "epoch": 2.487551314176041, "grad_norm": 2.3717517852783203, "learning_rate": 6.684e-07, "loss": 0.0171, "step": 46659 }, { "epoch": 2.487604627605694, "grad_norm": 2.6560986042022705, "learning_rate": 6.682000000000001e-07, "loss": 0.0154, "step": 46660 }, { "epoch": 2.487657941035347, "grad_norm": 2.0137038230895996, "learning_rate": 6.68e-07, "loss": 0.0109, "step": 46661 }, { "epoch": 2.4877112544649997, "grad_norm": 1.584187388420105, "learning_rate": 6.678000000000001e-07, "loss": 0.0211, "step": 46662 }, { "epoch": 2.4877645678946525, "grad_norm": 1.5137752294540405, "learning_rate": 6.676000000000001e-07, "loss": 0.0043, "step": 46663 }, { "epoch": 2.4878178813243057, "grad_norm": 1.1067309379577637, "learning_rate": 6.674e-07, "loss": 0.0112, "step": 46664 }, { "epoch": 2.4878711947539585, "grad_norm": 0.9173750281333923, "learning_rate": 6.672e-07, "loss": 0.0037, "step": 46665 }, { "epoch": 2.487924508183611, "grad_norm": 2.0328500270843506, "learning_rate": 6.67e-07, "loss": 0.0096, "step": 46666 }, { "epoch": 2.4879778216132644, "grad_norm": 0.46731653809547424, "learning_rate": 6.668000000000001e-07, "loss": 0.0036, "step": 46667 }, { "epoch": 2.488031135042917, "grad_norm": 1.4286150932312012, "learning_rate": 6.666e-07, "loss": 0.0063, "step": 46668 }, { "epoch": 2.4880844484725704, "grad_norm": 2.5821118354797363, "learning_rate": 6.664000000000001e-07, "loss": 0.01, "step": 46669 }, { "epoch": 2.488137761902223, "grad_norm": 1.1420928239822388, "learning_rate": 6.662000000000001e-07, "loss": 0.005, "step": 46670 }, { "epoch": 2.4881910753318763, "grad_norm": 2.2370550632476807, "learning_rate": 6.660000000000002e-07, "loss": 0.0094, "step": 46671 }, { "epoch": 2.488244388761529, "grad_norm": 2.715982675552368, "learning_rate": 6.658e-07, "loss": 0.0117, "step": 46672 }, { "epoch": 2.488297702191182, "grad_norm": 2.6997017860412598, "learning_rate": 6.656e-07, "loss": 0.0497, "step": 46673 }, { "epoch": 2.488351015620835, "grad_norm": 3.49084210395813, "learning_rate": 6.654000000000001e-07, "loss": 0.0185, "step": 46674 }, { "epoch": 2.488404329050488, "grad_norm": 1.1748950481414795, "learning_rate": 6.652e-07, "loss": 0.0199, "step": 46675 }, { "epoch": 2.4884576424801406, "grad_norm": 1.6923414468765259, "learning_rate": 6.650000000000001e-07, "loss": 0.0245, "step": 46676 }, { "epoch": 2.4885109559097938, "grad_norm": 1.912817358970642, "learning_rate": 6.648000000000001e-07, "loss": 0.0079, "step": 46677 }, { "epoch": 2.4885642693394465, "grad_norm": 1.0559669733047485, "learning_rate": 6.646000000000001e-07, "loss": 0.0108, "step": 46678 }, { "epoch": 2.4886175827690997, "grad_norm": 3.418524742126465, "learning_rate": 6.644e-07, "loss": 0.0094, "step": 46679 }, { "epoch": 2.4886708961987525, "grad_norm": 2.584967851638794, "learning_rate": 6.642000000000001e-07, "loss": 0.0474, "step": 46680 }, { "epoch": 2.4887242096284052, "grad_norm": 1.135600209236145, "learning_rate": 6.64e-07, "loss": 0.0065, "step": 46681 }, { "epoch": 2.4887775230580584, "grad_norm": 2.0219128131866455, "learning_rate": 6.638e-07, "loss": 0.0365, "step": 46682 }, { "epoch": 2.488830836487711, "grad_norm": 1.488013505935669, "learning_rate": 6.636000000000001e-07, "loss": 0.0119, "step": 46683 }, { "epoch": 2.4888841499173644, "grad_norm": 3.5577902793884277, "learning_rate": 6.634e-07, "loss": 0.0331, "step": 46684 }, { "epoch": 2.488937463347017, "grad_norm": 3.037587881088257, "learning_rate": 6.632000000000001e-07, "loss": 0.0094, "step": 46685 }, { "epoch": 2.48899077677667, "grad_norm": 2.103754997253418, "learning_rate": 6.63e-07, "loss": 0.011, "step": 46686 }, { "epoch": 2.489044090206323, "grad_norm": 1.2421492338180542, "learning_rate": 6.628000000000001e-07, "loss": 0.0046, "step": 46687 }, { "epoch": 2.489097403635976, "grad_norm": 0.8927409648895264, "learning_rate": 6.626e-07, "loss": 0.0143, "step": 46688 }, { "epoch": 2.4891507170656286, "grad_norm": 1.6281015872955322, "learning_rate": 6.624e-07, "loss": 0.0097, "step": 46689 }, { "epoch": 2.489204030495282, "grad_norm": 1.272857904434204, "learning_rate": 6.622000000000001e-07, "loss": 0.0062, "step": 46690 }, { "epoch": 2.4892573439249346, "grad_norm": 4.126591205596924, "learning_rate": 6.62e-07, "loss": 0.0165, "step": 46691 }, { "epoch": 2.489310657354588, "grad_norm": 3.8455636501312256, "learning_rate": 6.618000000000001e-07, "loss": 0.031, "step": 46692 }, { "epoch": 2.4893639707842405, "grad_norm": 2.062732219696045, "learning_rate": 6.616e-07, "loss": 0.0155, "step": 46693 }, { "epoch": 2.4894172842138933, "grad_norm": 0.22785139083862305, "learning_rate": 6.614000000000001e-07, "loss": 0.0018, "step": 46694 }, { "epoch": 2.4894705976435465, "grad_norm": 2.3741307258605957, "learning_rate": 6.612e-07, "loss": 0.019, "step": 46695 }, { "epoch": 2.4895239110731993, "grad_norm": 4.150012016296387, "learning_rate": 6.610000000000001e-07, "loss": 0.0104, "step": 46696 }, { "epoch": 2.4895772245028525, "grad_norm": 2.5171656608581543, "learning_rate": 6.608000000000001e-07, "loss": 0.0072, "step": 46697 }, { "epoch": 2.489630537932505, "grad_norm": 0.6876300573348999, "learning_rate": 6.606e-07, "loss": 0.0039, "step": 46698 }, { "epoch": 2.489683851362158, "grad_norm": 1.234810709953308, "learning_rate": 6.604000000000001e-07, "loss": 0.0087, "step": 46699 }, { "epoch": 2.489737164791811, "grad_norm": 1.8876466751098633, "learning_rate": 6.602e-07, "loss": 0.0297, "step": 46700 }, { "epoch": 2.489790478221464, "grad_norm": 0.32509395480155945, "learning_rate": 6.6e-07, "loss": 0.0026, "step": 46701 }, { "epoch": 2.4898437916511167, "grad_norm": 0.3922688663005829, "learning_rate": 6.598e-07, "loss": 0.0031, "step": 46702 }, { "epoch": 2.48989710508077, "grad_norm": 0.5291170477867126, "learning_rate": 6.596000000000001e-07, "loss": 0.0038, "step": 46703 }, { "epoch": 2.4899504185104226, "grad_norm": 0.9286863803863525, "learning_rate": 6.594e-07, "loss": 0.0035, "step": 46704 }, { "epoch": 2.490003731940076, "grad_norm": 1.1176408529281616, "learning_rate": 6.592000000000001e-07, "loss": 0.0046, "step": 46705 }, { "epoch": 2.4900570453697286, "grad_norm": 1.9311350584030151, "learning_rate": 6.590000000000001e-07, "loss": 0.0074, "step": 46706 }, { "epoch": 2.490110358799382, "grad_norm": 1.0062227249145508, "learning_rate": 6.587999999999999e-07, "loss": 0.0044, "step": 46707 }, { "epoch": 2.4901636722290346, "grad_norm": 0.7137212753295898, "learning_rate": 6.586e-07, "loss": 0.0038, "step": 46708 }, { "epoch": 2.4902169856586873, "grad_norm": 1.1235229969024658, "learning_rate": 6.584e-07, "loss": 0.0059, "step": 46709 }, { "epoch": 2.4902702990883405, "grad_norm": 2.33259654045105, "learning_rate": 6.582000000000001e-07, "loss": 0.0303, "step": 46710 }, { "epoch": 2.4903236125179933, "grad_norm": 1.3153020143508911, "learning_rate": 6.58e-07, "loss": 0.0139, "step": 46711 }, { "epoch": 2.490376925947646, "grad_norm": 1.1629067659378052, "learning_rate": 6.578000000000001e-07, "loss": 0.0147, "step": 46712 }, { "epoch": 2.4904302393772992, "grad_norm": 1.1358022689819336, "learning_rate": 6.576000000000001e-07, "loss": 0.0048, "step": 46713 }, { "epoch": 2.490483552806952, "grad_norm": 1.788927435874939, "learning_rate": 6.574000000000002e-07, "loss": 0.007, "step": 46714 }, { "epoch": 2.4905368662366048, "grad_norm": 3.6568551063537598, "learning_rate": 6.572e-07, "loss": 0.0122, "step": 46715 }, { "epoch": 2.490590179666258, "grad_norm": 1.8205703496932983, "learning_rate": 6.57e-07, "loss": 0.0136, "step": 46716 }, { "epoch": 2.4906434930959107, "grad_norm": 6.760446071624756, "learning_rate": 6.568000000000001e-07, "loss": 0.039, "step": 46717 }, { "epoch": 2.490696806525564, "grad_norm": 1.5073411464691162, "learning_rate": 6.566e-07, "loss": 0.0219, "step": 46718 }, { "epoch": 2.4907501199552167, "grad_norm": 1.4566060304641724, "learning_rate": 6.564000000000001e-07, "loss": 0.0357, "step": 46719 }, { "epoch": 2.49080343338487, "grad_norm": 0.5155893564224243, "learning_rate": 6.562000000000001e-07, "loss": 0.0032, "step": 46720 }, { "epoch": 2.4908567468145226, "grad_norm": 1.188285231590271, "learning_rate": 6.560000000000002e-07, "loss": 0.0166, "step": 46721 }, { "epoch": 2.4909100602441754, "grad_norm": 2.3120384216308594, "learning_rate": 6.558e-07, "loss": 0.011, "step": 46722 }, { "epoch": 2.4909633736738286, "grad_norm": 1.114660382270813, "learning_rate": 6.556e-07, "loss": 0.004, "step": 46723 }, { "epoch": 2.4910166871034813, "grad_norm": 5.782220363616943, "learning_rate": 6.554000000000001e-07, "loss": 0.0463, "step": 46724 }, { "epoch": 2.491070000533134, "grad_norm": 2.268641710281372, "learning_rate": 6.552e-07, "loss": 0.0232, "step": 46725 }, { "epoch": 2.4911233139627873, "grad_norm": 1.2577394247055054, "learning_rate": 6.550000000000001e-07, "loss": 0.0164, "step": 46726 }, { "epoch": 2.49117662739244, "grad_norm": 3.989724636077881, "learning_rate": 6.548000000000001e-07, "loss": 0.0294, "step": 46727 }, { "epoch": 2.4912299408220933, "grad_norm": 1.2225092649459839, "learning_rate": 6.546e-07, "loss": 0.01, "step": 46728 }, { "epoch": 2.491283254251746, "grad_norm": 2.422363519668579, "learning_rate": 6.544e-07, "loss": 0.0089, "step": 46729 }, { "epoch": 2.4913365676813988, "grad_norm": 1.5064908266067505, "learning_rate": 6.542000000000001e-07, "loss": 0.008, "step": 46730 }, { "epoch": 2.491389881111052, "grad_norm": 1.5809811353683472, "learning_rate": 6.54e-07, "loss": 0.0261, "step": 46731 }, { "epoch": 2.4914431945407047, "grad_norm": 2.8615174293518066, "learning_rate": 6.538e-07, "loss": 0.0204, "step": 46732 }, { "epoch": 2.491496507970358, "grad_norm": 0.4569736123085022, "learning_rate": 6.536000000000001e-07, "loss": 0.0035, "step": 46733 }, { "epoch": 2.4915498214000107, "grad_norm": 1.5898970365524292, "learning_rate": 6.534e-07, "loss": 0.0282, "step": 46734 }, { "epoch": 2.4916031348296634, "grad_norm": 3.01291823387146, "learning_rate": 6.532e-07, "loss": 0.0214, "step": 46735 }, { "epoch": 2.4916564482593166, "grad_norm": 1.5456427335739136, "learning_rate": 6.53e-07, "loss": 0.0225, "step": 46736 }, { "epoch": 2.4917097616889694, "grad_norm": 0.6043679714202881, "learning_rate": 6.528000000000001e-07, "loss": 0.0047, "step": 46737 }, { "epoch": 2.491763075118622, "grad_norm": 2.524242639541626, "learning_rate": 6.526e-07, "loss": 0.0297, "step": 46738 }, { "epoch": 2.4918163885482754, "grad_norm": 3.3593623638153076, "learning_rate": 6.524000000000001e-07, "loss": 0.0215, "step": 46739 }, { "epoch": 2.491869701977928, "grad_norm": 1.130704402923584, "learning_rate": 6.522000000000001e-07, "loss": 0.0034, "step": 46740 }, { "epoch": 2.4919230154075813, "grad_norm": 1.4722211360931396, "learning_rate": 6.52e-07, "loss": 0.0203, "step": 46741 }, { "epoch": 2.491976328837234, "grad_norm": 5.967278480529785, "learning_rate": 6.518e-07, "loss": 0.0145, "step": 46742 }, { "epoch": 2.492029642266887, "grad_norm": 1.6674429178237915, "learning_rate": 6.516e-07, "loss": 0.0146, "step": 46743 }, { "epoch": 2.49208295569654, "grad_norm": 0.8833093643188477, "learning_rate": 6.514000000000001e-07, "loss": 0.0064, "step": 46744 }, { "epoch": 2.492136269126193, "grad_norm": 1.39242684841156, "learning_rate": 6.512e-07, "loss": 0.0119, "step": 46745 }, { "epoch": 2.492189582555846, "grad_norm": 2.1059601306915283, "learning_rate": 6.510000000000001e-07, "loss": 0.0104, "step": 46746 }, { "epoch": 2.4922428959854988, "grad_norm": 0.5693378448486328, "learning_rate": 6.508000000000001e-07, "loss": 0.0032, "step": 46747 }, { "epoch": 2.4922962094151515, "grad_norm": 0.5075734853744507, "learning_rate": 6.506000000000001e-07, "loss": 0.0035, "step": 46748 }, { "epoch": 2.4923495228448047, "grad_norm": 1.1027641296386719, "learning_rate": 6.504000000000001e-07, "loss": 0.0256, "step": 46749 }, { "epoch": 2.4924028362744575, "grad_norm": 1.2068495750427246, "learning_rate": 6.502e-07, "loss": 0.0059, "step": 46750 }, { "epoch": 2.4924561497041102, "grad_norm": 2.052668333053589, "learning_rate": 6.5e-07, "loss": 0.0087, "step": 46751 }, { "epoch": 2.4925094631337634, "grad_norm": 2.4874777793884277, "learning_rate": 6.498e-07, "loss": 0.0302, "step": 46752 }, { "epoch": 2.492562776563416, "grad_norm": 2.151350498199463, "learning_rate": 6.496000000000001e-07, "loss": 0.0148, "step": 46753 }, { "epoch": 2.4926160899930694, "grad_norm": 1.992051362991333, "learning_rate": 6.494e-07, "loss": 0.0167, "step": 46754 }, { "epoch": 2.492669403422722, "grad_norm": 1.7252566814422607, "learning_rate": 6.492000000000001e-07, "loss": 0.0243, "step": 46755 }, { "epoch": 2.4927227168523753, "grad_norm": 3.000730037689209, "learning_rate": 6.490000000000001e-07, "loss": 0.0481, "step": 46756 }, { "epoch": 2.492776030282028, "grad_norm": 3.6166133880615234, "learning_rate": 6.487999999999999e-07, "loss": 0.0224, "step": 46757 }, { "epoch": 2.492829343711681, "grad_norm": 1.7655164003372192, "learning_rate": 6.486e-07, "loss": 0.0167, "step": 46758 }, { "epoch": 2.492882657141334, "grad_norm": 2.520202875137329, "learning_rate": 6.484e-07, "loss": 0.0204, "step": 46759 }, { "epoch": 2.492935970570987, "grad_norm": 2.8396804332733154, "learning_rate": 6.482000000000001e-07, "loss": 0.0303, "step": 46760 }, { "epoch": 2.4929892840006396, "grad_norm": 0.46899718046188354, "learning_rate": 6.48e-07, "loss": 0.0034, "step": 46761 }, { "epoch": 2.4930425974302928, "grad_norm": 0.8205611705780029, "learning_rate": 6.478000000000001e-07, "loss": 0.0051, "step": 46762 }, { "epoch": 2.4930959108599455, "grad_norm": 0.9792314171791077, "learning_rate": 6.476000000000001e-07, "loss": 0.0071, "step": 46763 }, { "epoch": 2.4931492242895987, "grad_norm": 0.7645521759986877, "learning_rate": 6.474000000000001e-07, "loss": 0.0038, "step": 46764 }, { "epoch": 2.4932025377192515, "grad_norm": 2.914888620376587, "learning_rate": 6.472e-07, "loss": 0.0129, "step": 46765 }, { "epoch": 2.4932558511489042, "grad_norm": 3.1568715572357178, "learning_rate": 6.47e-07, "loss": 0.0233, "step": 46766 }, { "epoch": 2.4933091645785574, "grad_norm": 2.972971200942993, "learning_rate": 6.468000000000001e-07, "loss": 0.0094, "step": 46767 }, { "epoch": 2.49336247800821, "grad_norm": 1.7595452070236206, "learning_rate": 6.466e-07, "loss": 0.0168, "step": 46768 }, { "epoch": 2.4934157914378634, "grad_norm": 0.2093208134174347, "learning_rate": 6.464000000000001e-07, "loss": 0.0012, "step": 46769 }, { "epoch": 2.493469104867516, "grad_norm": 2.780904769897461, "learning_rate": 6.462000000000001e-07, "loss": 0.0229, "step": 46770 }, { "epoch": 2.493522418297169, "grad_norm": 1.2853457927703857, "learning_rate": 6.460000000000001e-07, "loss": 0.0143, "step": 46771 }, { "epoch": 2.493575731726822, "grad_norm": 2.362809419631958, "learning_rate": 6.458e-07, "loss": 0.005, "step": 46772 }, { "epoch": 2.493629045156475, "grad_norm": 4.110498428344727, "learning_rate": 6.456000000000001e-07, "loss": 0.0093, "step": 46773 }, { "epoch": 2.4936823585861276, "grad_norm": 3.2070741653442383, "learning_rate": 6.454000000000001e-07, "loss": 0.0118, "step": 46774 }, { "epoch": 2.493735672015781, "grad_norm": 0.6882081031799316, "learning_rate": 6.452e-07, "loss": 0.0029, "step": 46775 }, { "epoch": 2.4937889854454336, "grad_norm": 5.579166889190674, "learning_rate": 6.450000000000001e-07, "loss": 0.0352, "step": 46776 }, { "epoch": 2.493842298875087, "grad_norm": 5.844295978546143, "learning_rate": 6.448000000000001e-07, "loss": 0.0333, "step": 46777 }, { "epoch": 2.4938956123047396, "grad_norm": 2.4400200843811035, "learning_rate": 6.446e-07, "loss": 0.044, "step": 46778 }, { "epoch": 2.4939489257343923, "grad_norm": 2.2712955474853516, "learning_rate": 6.444e-07, "loss": 0.0087, "step": 46779 }, { "epoch": 2.4940022391640455, "grad_norm": 3.1664702892303467, "learning_rate": 6.442000000000001e-07, "loss": 0.0227, "step": 46780 }, { "epoch": 2.4940555525936983, "grad_norm": 2.85101318359375, "learning_rate": 6.44e-07, "loss": 0.0303, "step": 46781 }, { "epoch": 2.4941088660233515, "grad_norm": 1.293071985244751, "learning_rate": 6.438000000000001e-07, "loss": 0.0046, "step": 46782 }, { "epoch": 2.4941621794530042, "grad_norm": 1.4741939306259155, "learning_rate": 6.436000000000001e-07, "loss": 0.0245, "step": 46783 }, { "epoch": 2.494215492882657, "grad_norm": 1.1676716804504395, "learning_rate": 6.434e-07, "loss": 0.0222, "step": 46784 }, { "epoch": 2.49426880631231, "grad_norm": 1.294392466545105, "learning_rate": 6.432e-07, "loss": 0.0052, "step": 46785 }, { "epoch": 2.494322119741963, "grad_norm": 0.8878882527351379, "learning_rate": 6.43e-07, "loss": 0.0036, "step": 46786 }, { "epoch": 2.4943754331716157, "grad_norm": 1.0273371934890747, "learning_rate": 6.428000000000001e-07, "loss": 0.0126, "step": 46787 }, { "epoch": 2.494428746601269, "grad_norm": 5.482708930969238, "learning_rate": 6.426e-07, "loss": 0.0322, "step": 46788 }, { "epoch": 2.4944820600309217, "grad_norm": 2.669553518295288, "learning_rate": 6.424000000000001e-07, "loss": 0.0165, "step": 46789 }, { "epoch": 2.494535373460575, "grad_norm": 3.309567451477051, "learning_rate": 6.422000000000001e-07, "loss": 0.0184, "step": 46790 }, { "epoch": 2.4945886868902276, "grad_norm": 1.4072494506835938, "learning_rate": 6.42e-07, "loss": 0.0073, "step": 46791 }, { "epoch": 2.494642000319881, "grad_norm": 2.147162914276123, "learning_rate": 6.418e-07, "loss": 0.0128, "step": 46792 }, { "epoch": 2.4946953137495336, "grad_norm": 1.0589487552642822, "learning_rate": 6.416e-07, "loss": 0.0033, "step": 46793 }, { "epoch": 2.4947486271791863, "grad_norm": 2.0371832847595215, "learning_rate": 6.414000000000001e-07, "loss": 0.007, "step": 46794 }, { "epoch": 2.4948019406088395, "grad_norm": 1.7470917701721191, "learning_rate": 6.412e-07, "loss": 0.0269, "step": 46795 }, { "epoch": 2.4948552540384923, "grad_norm": 2.670008659362793, "learning_rate": 6.410000000000001e-07, "loss": 0.0338, "step": 46796 }, { "epoch": 2.494908567468145, "grad_norm": 0.7955076694488525, "learning_rate": 6.408000000000001e-07, "loss": 0.004, "step": 46797 }, { "epoch": 2.4949618808977982, "grad_norm": 1.2519382238388062, "learning_rate": 6.406000000000001e-07, "loss": 0.0143, "step": 46798 }, { "epoch": 2.495015194327451, "grad_norm": 1.2629473209381104, "learning_rate": 6.404e-07, "loss": 0.0203, "step": 46799 }, { "epoch": 2.495068507757104, "grad_norm": 2.4943108558654785, "learning_rate": 6.402e-07, "loss": 0.0097, "step": 46800 }, { "epoch": 2.495121821186757, "grad_norm": 0.9207639098167419, "learning_rate": 6.4e-07, "loss": 0.0175, "step": 46801 }, { "epoch": 2.4951751346164097, "grad_norm": 4.073456764221191, "learning_rate": 6.398e-07, "loss": 0.0291, "step": 46802 }, { "epoch": 2.495228448046063, "grad_norm": 2.3289904594421387, "learning_rate": 6.396000000000001e-07, "loss": 0.0086, "step": 46803 }, { "epoch": 2.4952817614757157, "grad_norm": 1.6632755994796753, "learning_rate": 6.394e-07, "loss": 0.0139, "step": 46804 }, { "epoch": 2.495335074905369, "grad_norm": 3.3619632720947266, "learning_rate": 6.392000000000001e-07, "loss": 0.0445, "step": 46805 }, { "epoch": 2.4953883883350216, "grad_norm": 2.744833469390869, "learning_rate": 6.39e-07, "loss": 0.0149, "step": 46806 }, { "epoch": 2.4954417017646744, "grad_norm": 3.2274060249328613, "learning_rate": 6.388000000000001e-07, "loss": 0.0098, "step": 46807 }, { "epoch": 2.4954950151943276, "grad_norm": 0.9688887000083923, "learning_rate": 6.386e-07, "loss": 0.0053, "step": 46808 }, { "epoch": 2.4955483286239803, "grad_norm": 4.475805282592773, "learning_rate": 6.384e-07, "loss": 0.0096, "step": 46809 }, { "epoch": 2.495601642053633, "grad_norm": 1.3360780477523804, "learning_rate": 6.382000000000001e-07, "loss": 0.0055, "step": 46810 }, { "epoch": 2.4956549554832863, "grad_norm": 1.8388621807098389, "learning_rate": 6.38e-07, "loss": 0.0236, "step": 46811 }, { "epoch": 2.495708268912939, "grad_norm": 2.861848831176758, "learning_rate": 6.378000000000001e-07, "loss": 0.0247, "step": 46812 }, { "epoch": 2.4957615823425923, "grad_norm": 2.0199742317199707, "learning_rate": 6.376e-07, "loss": 0.0046, "step": 46813 }, { "epoch": 2.495814895772245, "grad_norm": 1.1748237609863281, "learning_rate": 6.374000000000001e-07, "loss": 0.0194, "step": 46814 }, { "epoch": 2.495868209201898, "grad_norm": 0.5085280537605286, "learning_rate": 6.372e-07, "loss": 0.003, "step": 46815 }, { "epoch": 2.495921522631551, "grad_norm": 0.5096064805984497, "learning_rate": 6.370000000000001e-07, "loss": 0.0028, "step": 46816 }, { "epoch": 2.4959748360612037, "grad_norm": 3.1354665756225586, "learning_rate": 6.368000000000001e-07, "loss": 0.0503, "step": 46817 }, { "epoch": 2.496028149490857, "grad_norm": 2.296229600906372, "learning_rate": 6.366e-07, "loss": 0.0082, "step": 46818 }, { "epoch": 2.4960814629205097, "grad_norm": 2.2458982467651367, "learning_rate": 6.364000000000001e-07, "loss": 0.009, "step": 46819 }, { "epoch": 2.4961347763501625, "grad_norm": 1.2261962890625, "learning_rate": 6.362e-07, "loss": 0.0066, "step": 46820 }, { "epoch": 2.4961880897798157, "grad_norm": 0.353201299905777, "learning_rate": 6.360000000000001e-07, "loss": 0.0017, "step": 46821 }, { "epoch": 2.4962414032094684, "grad_norm": 1.0056872367858887, "learning_rate": 6.358e-07, "loss": 0.0072, "step": 46822 }, { "epoch": 2.496294716639121, "grad_norm": 1.9386279582977295, "learning_rate": 6.356000000000001e-07, "loss": 0.0135, "step": 46823 }, { "epoch": 2.4963480300687744, "grad_norm": 2.7733566761016846, "learning_rate": 6.354000000000001e-07, "loss": 0.0108, "step": 46824 }, { "epoch": 2.496401343498427, "grad_norm": 0.6773371696472168, "learning_rate": 6.352e-07, "loss": 0.0033, "step": 46825 }, { "epoch": 2.4964546569280803, "grad_norm": 2.2624523639678955, "learning_rate": 6.350000000000001e-07, "loss": 0.0284, "step": 46826 }, { "epoch": 2.496507970357733, "grad_norm": 3.65881609916687, "learning_rate": 6.348e-07, "loss": 0.0097, "step": 46827 }, { "epoch": 2.4965612837873863, "grad_norm": 1.5932354927062988, "learning_rate": 6.346e-07, "loss": 0.0288, "step": 46828 }, { "epoch": 2.496614597217039, "grad_norm": 0.2551114559173584, "learning_rate": 6.344e-07, "loss": 0.0023, "step": 46829 }, { "epoch": 2.496667910646692, "grad_norm": 1.1389946937561035, "learning_rate": 6.342000000000001e-07, "loss": 0.0042, "step": 46830 }, { "epoch": 2.496721224076345, "grad_norm": 1.3832091093063354, "learning_rate": 6.34e-07, "loss": 0.0044, "step": 46831 }, { "epoch": 2.4967745375059978, "grad_norm": 0.47618407011032104, "learning_rate": 6.338000000000001e-07, "loss": 0.0031, "step": 46832 }, { "epoch": 2.4968278509356505, "grad_norm": 0.2907220721244812, "learning_rate": 6.336000000000001e-07, "loss": 0.0018, "step": 46833 }, { "epoch": 2.4968811643653037, "grad_norm": 4.052336692810059, "learning_rate": 6.333999999999999e-07, "loss": 0.0557, "step": 46834 }, { "epoch": 2.4969344777949565, "grad_norm": 2.8542704582214355, "learning_rate": 6.332e-07, "loss": 0.0351, "step": 46835 }, { "epoch": 2.4969877912246092, "grad_norm": 3.5749244689941406, "learning_rate": 6.33e-07, "loss": 0.0229, "step": 46836 }, { "epoch": 2.4970411046542624, "grad_norm": 0.6283722519874573, "learning_rate": 6.328000000000001e-07, "loss": 0.0023, "step": 46837 }, { "epoch": 2.497094418083915, "grad_norm": 1.7813427448272705, "learning_rate": 6.326e-07, "loss": 0.0075, "step": 46838 }, { "epoch": 2.4971477315135684, "grad_norm": 1.6856744289398193, "learning_rate": 6.324000000000001e-07, "loss": 0.0284, "step": 46839 }, { "epoch": 2.497201044943221, "grad_norm": 0.3718869090080261, "learning_rate": 6.322000000000001e-07, "loss": 0.003, "step": 46840 }, { "epoch": 2.4972543583728743, "grad_norm": 4.22083854675293, "learning_rate": 6.320000000000002e-07, "loss": 0.0452, "step": 46841 }, { "epoch": 2.497307671802527, "grad_norm": 3.74775767326355, "learning_rate": 6.318e-07, "loss": 0.0461, "step": 46842 }, { "epoch": 2.49736098523218, "grad_norm": 1.3150447607040405, "learning_rate": 6.316e-07, "loss": 0.0222, "step": 46843 }, { "epoch": 2.497414298661833, "grad_norm": 2.0297234058380127, "learning_rate": 6.314000000000001e-07, "loss": 0.013, "step": 46844 }, { "epoch": 2.497467612091486, "grad_norm": 0.5108199715614319, "learning_rate": 6.312e-07, "loss": 0.0032, "step": 46845 }, { "epoch": 2.4975209255211386, "grad_norm": 1.8528655767440796, "learning_rate": 6.310000000000001e-07, "loss": 0.01, "step": 46846 }, { "epoch": 2.497574238950792, "grad_norm": 3.442420721054077, "learning_rate": 6.308000000000001e-07, "loss": 0.0177, "step": 46847 }, { "epoch": 2.4976275523804445, "grad_norm": 1.6681550741195679, "learning_rate": 6.306e-07, "loss": 0.0152, "step": 46848 }, { "epoch": 2.4976808658100977, "grad_norm": 2.3474655151367188, "learning_rate": 6.304e-07, "loss": 0.0263, "step": 46849 }, { "epoch": 2.4977341792397505, "grad_norm": 1.4603053331375122, "learning_rate": 6.302000000000001e-07, "loss": 0.0133, "step": 46850 }, { "epoch": 2.4977874926694033, "grad_norm": 4.159486770629883, "learning_rate": 6.3e-07, "loss": 0.0295, "step": 46851 }, { "epoch": 2.4978408060990565, "grad_norm": 1.9976054430007935, "learning_rate": 6.298e-07, "loss": 0.0271, "step": 46852 }, { "epoch": 2.497894119528709, "grad_norm": 3.112185478210449, "learning_rate": 6.296000000000001e-07, "loss": 0.0314, "step": 46853 }, { "epoch": 2.4979474329583624, "grad_norm": 3.8282647132873535, "learning_rate": 6.294e-07, "loss": 0.037, "step": 46854 }, { "epoch": 2.498000746388015, "grad_norm": 0.16387301683425903, "learning_rate": 6.292e-07, "loss": 0.0014, "step": 46855 }, { "epoch": 2.498054059817668, "grad_norm": 2.630169630050659, "learning_rate": 6.29e-07, "loss": 0.0228, "step": 46856 }, { "epoch": 2.498107373247321, "grad_norm": 4.419291019439697, "learning_rate": 6.288000000000001e-07, "loss": 0.0213, "step": 46857 }, { "epoch": 2.498160686676974, "grad_norm": 1.1508560180664062, "learning_rate": 6.286e-07, "loss": 0.0068, "step": 46858 }, { "epoch": 2.4982140001066266, "grad_norm": 3.766331911087036, "learning_rate": 6.284e-07, "loss": 0.0325, "step": 46859 }, { "epoch": 2.49826731353628, "grad_norm": 0.8346496820449829, "learning_rate": 6.282000000000001e-07, "loss": 0.0046, "step": 46860 }, { "epoch": 2.4983206269659326, "grad_norm": 3.2412471771240234, "learning_rate": 6.28e-07, "loss": 0.0123, "step": 46861 }, { "epoch": 2.498373940395586, "grad_norm": 1.2870203256607056, "learning_rate": 6.278e-07, "loss": 0.0182, "step": 46862 }, { "epoch": 2.4984272538252386, "grad_norm": 0.715732991695404, "learning_rate": 6.276e-07, "loss": 0.0039, "step": 46863 }, { "epoch": 2.4984805672548913, "grad_norm": 0.33731234073638916, "learning_rate": 6.274000000000001e-07, "loss": 0.0027, "step": 46864 }, { "epoch": 2.4985338806845445, "grad_norm": 2.0080058574676514, "learning_rate": 6.272e-07, "loss": 0.0087, "step": 46865 }, { "epoch": 2.4985871941141973, "grad_norm": 3.606983184814453, "learning_rate": 6.270000000000001e-07, "loss": 0.0306, "step": 46866 }, { "epoch": 2.4986405075438505, "grad_norm": 2.429178237915039, "learning_rate": 6.268000000000001e-07, "loss": 0.0261, "step": 46867 }, { "epoch": 2.4986938209735032, "grad_norm": 2.3638503551483154, "learning_rate": 6.266e-07, "loss": 0.0054, "step": 46868 }, { "epoch": 2.498747134403156, "grad_norm": 1.3569800853729248, "learning_rate": 6.264e-07, "loss": 0.0121, "step": 46869 }, { "epoch": 2.498800447832809, "grad_norm": 3.2779948711395264, "learning_rate": 6.262e-07, "loss": 0.05, "step": 46870 }, { "epoch": 2.498853761262462, "grad_norm": 0.9140782356262207, "learning_rate": 6.260000000000001e-07, "loss": 0.0048, "step": 46871 }, { "epoch": 2.4989070746921147, "grad_norm": 1.72176194190979, "learning_rate": 6.258e-07, "loss": 0.0173, "step": 46872 }, { "epoch": 2.498960388121768, "grad_norm": 2.3515114784240723, "learning_rate": 6.256000000000001e-07, "loss": 0.0107, "step": 46873 }, { "epoch": 2.4990137015514207, "grad_norm": 1.4278098344802856, "learning_rate": 6.254000000000001e-07, "loss": 0.0203, "step": 46874 }, { "epoch": 2.499067014981074, "grad_norm": 0.6462703943252563, "learning_rate": 6.252000000000001e-07, "loss": 0.003, "step": 46875 }, { "epoch": 2.4991203284107266, "grad_norm": 2.302863836288452, "learning_rate": 6.25e-07, "loss": 0.0139, "step": 46876 }, { "epoch": 2.49917364184038, "grad_norm": 1.9630151987075806, "learning_rate": 6.248e-07, "loss": 0.0191, "step": 46877 }, { "epoch": 2.4992269552700326, "grad_norm": 1.3904000520706177, "learning_rate": 6.246e-07, "loss": 0.0034, "step": 46878 }, { "epoch": 2.4992802686996853, "grad_norm": 1.8023278713226318, "learning_rate": 6.244000000000001e-07, "loss": 0.0056, "step": 46879 }, { "epoch": 2.4993335821293385, "grad_norm": 1.1671464443206787, "learning_rate": 6.242000000000001e-07, "loss": 0.0057, "step": 46880 }, { "epoch": 2.4993868955589913, "grad_norm": 1.114749550819397, "learning_rate": 6.24e-07, "loss": 0.0126, "step": 46881 }, { "epoch": 2.499440208988644, "grad_norm": 0.5149563550949097, "learning_rate": 6.238000000000001e-07, "loss": 0.0027, "step": 46882 }, { "epoch": 2.4994935224182973, "grad_norm": 2.3123762607574463, "learning_rate": 6.236e-07, "loss": 0.0134, "step": 46883 }, { "epoch": 2.49954683584795, "grad_norm": 2.008976459503174, "learning_rate": 6.234e-07, "loss": 0.0069, "step": 46884 }, { "epoch": 2.499600149277603, "grad_norm": 1.2954570055007935, "learning_rate": 6.232e-07, "loss": 0.0138, "step": 46885 }, { "epoch": 2.499653462707256, "grad_norm": 1.7426880598068237, "learning_rate": 6.230000000000001e-07, "loss": 0.0162, "step": 46886 }, { "epoch": 2.4997067761369087, "grad_norm": 2.48966646194458, "learning_rate": 6.228000000000001e-07, "loss": 0.0145, "step": 46887 }, { "epoch": 2.499760089566562, "grad_norm": 6.784890174865723, "learning_rate": 6.226000000000001e-07, "loss": 0.0249, "step": 46888 }, { "epoch": 2.4998134029962147, "grad_norm": 1.076140284538269, "learning_rate": 6.224000000000001e-07, "loss": 0.0055, "step": 46889 }, { "epoch": 2.499866716425868, "grad_norm": 0.8045108914375305, "learning_rate": 6.222e-07, "loss": 0.0044, "step": 46890 }, { "epoch": 2.4999200298555206, "grad_norm": 1.588821291923523, "learning_rate": 6.22e-07, "loss": 0.0151, "step": 46891 }, { "epoch": 2.4999733432851734, "grad_norm": 2.8615877628326416, "learning_rate": 6.218e-07, "loss": 0.0361, "step": 46892 }, { "epoch": 2.5000266567148266, "grad_norm": 0.7363959550857544, "learning_rate": 6.216e-07, "loss": 0.004, "step": 46893 }, { "epoch": 2.5000799701444794, "grad_norm": 2.6428844928741455, "learning_rate": 6.214000000000001e-07, "loss": 0.0162, "step": 46894 }, { "epoch": 2.500133283574132, "grad_norm": 1.5841714143753052, "learning_rate": 6.212000000000001e-07, "loss": 0.0084, "step": 46895 }, { "epoch": 2.5001865970037853, "grad_norm": 3.2329976558685303, "learning_rate": 6.210000000000001e-07, "loss": 0.029, "step": 46896 }, { "epoch": 2.500239910433438, "grad_norm": 2.143474578857422, "learning_rate": 6.208000000000001e-07, "loss": 0.0057, "step": 46897 }, { "epoch": 2.5002932238630913, "grad_norm": 0.6337712407112122, "learning_rate": 6.206e-07, "loss": 0.0026, "step": 46898 }, { "epoch": 2.500346537292744, "grad_norm": 1.1568777561187744, "learning_rate": 6.204e-07, "loss": 0.005, "step": 46899 }, { "epoch": 2.5003998507223972, "grad_norm": 2.953664779663086, "learning_rate": 6.202e-07, "loss": 0.0121, "step": 46900 }, { "epoch": 2.50045316415205, "grad_norm": 1.5402435064315796, "learning_rate": 6.200000000000001e-07, "loss": 0.0068, "step": 46901 }, { "epoch": 2.5005064775817027, "grad_norm": 0.7202178239822388, "learning_rate": 6.198000000000001e-07, "loss": 0.0046, "step": 46902 }, { "epoch": 2.500559791011356, "grad_norm": 1.3391106128692627, "learning_rate": 6.196000000000001e-07, "loss": 0.0039, "step": 46903 }, { "epoch": 2.5006131044410087, "grad_norm": 2.958827495574951, "learning_rate": 6.194000000000001e-07, "loss": 0.0097, "step": 46904 }, { "epoch": 2.5006664178706615, "grad_norm": 2.787574052810669, "learning_rate": 6.192e-07, "loss": 0.0247, "step": 46905 }, { "epoch": 2.5007197313003147, "grad_norm": 1.4047991037368774, "learning_rate": 6.19e-07, "loss": 0.005, "step": 46906 }, { "epoch": 2.5007730447299674, "grad_norm": 1.7602485418319702, "learning_rate": 6.188e-07, "loss": 0.0156, "step": 46907 }, { "epoch": 2.50082635815962, "grad_norm": 2.2766213417053223, "learning_rate": 6.186e-07, "loss": 0.0272, "step": 46908 }, { "epoch": 2.5008796715892734, "grad_norm": 2.138256549835205, "learning_rate": 6.184000000000001e-07, "loss": 0.0343, "step": 46909 }, { "epoch": 2.500932985018926, "grad_norm": 2.5586166381835938, "learning_rate": 6.182000000000001e-07, "loss": 0.0144, "step": 46910 }, { "epoch": 2.5009862984485793, "grad_norm": 3.693674087524414, "learning_rate": 6.180000000000001e-07, "loss": 0.0137, "step": 46911 }, { "epoch": 2.501039611878232, "grad_norm": 2.514059066772461, "learning_rate": 6.178e-07, "loss": 0.0101, "step": 46912 }, { "epoch": 2.5010929253078853, "grad_norm": 2.2499170303344727, "learning_rate": 6.176e-07, "loss": 0.0117, "step": 46913 }, { "epoch": 2.501146238737538, "grad_norm": 1.6911861896514893, "learning_rate": 6.174000000000001e-07, "loss": 0.006, "step": 46914 }, { "epoch": 2.501199552167191, "grad_norm": 4.220880031585693, "learning_rate": 6.172e-07, "loss": 0.0198, "step": 46915 }, { "epoch": 2.501252865596844, "grad_norm": 2.354104518890381, "learning_rate": 6.17e-07, "loss": 0.0302, "step": 46916 }, { "epoch": 2.5013061790264968, "grad_norm": 0.7361446619033813, "learning_rate": 6.168000000000001e-07, "loss": 0.0035, "step": 46917 }, { "epoch": 2.5013594924561495, "grad_norm": 3.0650413036346436, "learning_rate": 6.166000000000001e-07, "loss": 0.0251, "step": 46918 }, { "epoch": 2.5014128058858027, "grad_norm": 1.764281988143921, "learning_rate": 6.164e-07, "loss": 0.0123, "step": 46919 }, { "epoch": 2.5014661193154555, "grad_norm": 1.8418965339660645, "learning_rate": 6.162e-07, "loss": 0.0243, "step": 46920 }, { "epoch": 2.5015194327451082, "grad_norm": 0.4830515384674072, "learning_rate": 6.160000000000001e-07, "loss": 0.0024, "step": 46921 }, { "epoch": 2.5015727461747614, "grad_norm": 0.4590415060520172, "learning_rate": 6.158000000000001e-07, "loss": 0.0038, "step": 46922 }, { "epoch": 2.501626059604414, "grad_norm": 1.6980212926864624, "learning_rate": 6.156e-07, "loss": 0.0086, "step": 46923 }, { "epoch": 2.5016793730340674, "grad_norm": 1.9763532876968384, "learning_rate": 6.154000000000001e-07, "loss": 0.0273, "step": 46924 }, { "epoch": 2.50173268646372, "grad_norm": 1.185054063796997, "learning_rate": 6.152000000000001e-07, "loss": 0.005, "step": 46925 }, { "epoch": 2.5017859998933734, "grad_norm": 2.2890818119049072, "learning_rate": 6.15e-07, "loss": 0.0271, "step": 46926 }, { "epoch": 2.501839313323026, "grad_norm": 4.371461868286133, "learning_rate": 6.148e-07, "loss": 0.0101, "step": 46927 }, { "epoch": 2.501892626752679, "grad_norm": 3.3331668376922607, "learning_rate": 6.146e-07, "loss": 0.0171, "step": 46928 }, { "epoch": 2.501945940182332, "grad_norm": 2.366722345352173, "learning_rate": 6.144000000000001e-07, "loss": 0.0247, "step": 46929 }, { "epoch": 2.501999253611985, "grad_norm": 1.8543674945831299, "learning_rate": 6.142000000000001e-07, "loss": 0.0245, "step": 46930 }, { "epoch": 2.5020525670416376, "grad_norm": 2.6253764629364014, "learning_rate": 6.140000000000001e-07, "loss": 0.0082, "step": 46931 }, { "epoch": 2.502105880471291, "grad_norm": 1.6284414529800415, "learning_rate": 6.138000000000001e-07, "loss": 0.0191, "step": 46932 }, { "epoch": 2.5021591939009435, "grad_norm": 2.457526683807373, "learning_rate": 6.136e-07, "loss": 0.0147, "step": 46933 }, { "epoch": 2.5022125073305963, "grad_norm": 0.555541455745697, "learning_rate": 6.134e-07, "loss": 0.0035, "step": 46934 }, { "epoch": 2.5022658207602495, "grad_norm": 3.9928109645843506, "learning_rate": 6.132e-07, "loss": 0.021, "step": 46935 }, { "epoch": 2.5023191341899027, "grad_norm": 0.2611038088798523, "learning_rate": 6.130000000000001e-07, "loss": 0.0019, "step": 46936 }, { "epoch": 2.5023724476195555, "grad_norm": 0.22978925704956055, "learning_rate": 6.128000000000001e-07, "loss": 0.0018, "step": 46937 }, { "epoch": 2.502425761049208, "grad_norm": 3.7613372802734375, "learning_rate": 6.126000000000001e-07, "loss": 0.0117, "step": 46938 }, { "epoch": 2.5024790744788614, "grad_norm": 0.7326259016990662, "learning_rate": 6.124000000000001e-07, "loss": 0.0042, "step": 46939 }, { "epoch": 2.502532387908514, "grad_norm": 3.033604860305786, "learning_rate": 6.122e-07, "loss": 0.0509, "step": 46940 }, { "epoch": 2.502585701338167, "grad_norm": 1.7699640989303589, "learning_rate": 6.12e-07, "loss": 0.0148, "step": 46941 }, { "epoch": 2.50263901476782, "grad_norm": 1.489880084991455, "learning_rate": 6.118e-07, "loss": 0.0201, "step": 46942 }, { "epoch": 2.502692328197473, "grad_norm": 1.4308258295059204, "learning_rate": 6.116e-07, "loss": 0.0094, "step": 46943 }, { "epoch": 2.5027456416271257, "grad_norm": 1.011192798614502, "learning_rate": 6.114000000000001e-07, "loss": 0.0035, "step": 46944 }, { "epoch": 2.502798955056779, "grad_norm": 1.325061321258545, "learning_rate": 6.112000000000001e-07, "loss": 0.0105, "step": 46945 }, { "epoch": 2.5028522684864316, "grad_norm": 3.311985731124878, "learning_rate": 6.110000000000001e-07, "loss": 0.0304, "step": 46946 }, { "epoch": 2.502905581916085, "grad_norm": 2.9738309383392334, "learning_rate": 6.108e-07, "loss": 0.0108, "step": 46947 }, { "epoch": 2.5029588953457376, "grad_norm": 3.1076736450195312, "learning_rate": 6.106000000000001e-07, "loss": 0.0227, "step": 46948 }, { "epoch": 2.5030122087753908, "grad_norm": 1.3158893585205078, "learning_rate": 6.104e-07, "loss": 0.0054, "step": 46949 }, { "epoch": 2.5030655222050435, "grad_norm": 0.776106595993042, "learning_rate": 6.102e-07, "loss": 0.0048, "step": 46950 }, { "epoch": 2.5031188356346963, "grad_norm": 0.2016540765762329, "learning_rate": 6.100000000000001e-07, "loss": 0.0016, "step": 46951 }, { "epoch": 2.5031721490643495, "grad_norm": 3.8891119956970215, "learning_rate": 6.098000000000001e-07, "loss": 0.0202, "step": 46952 }, { "epoch": 2.5032254624940022, "grad_norm": 2.0834736824035645, "learning_rate": 6.096000000000001e-07, "loss": 0.0339, "step": 46953 }, { "epoch": 2.503278775923655, "grad_norm": 1.1073236465454102, "learning_rate": 6.094e-07, "loss": 0.0083, "step": 46954 }, { "epoch": 2.503332089353308, "grad_norm": 2.3976407051086426, "learning_rate": 6.092e-07, "loss": 0.02, "step": 46955 }, { "epoch": 2.503385402782961, "grad_norm": 0.23156245052814484, "learning_rate": 6.090000000000001e-07, "loss": 0.0015, "step": 46956 }, { "epoch": 2.5034387162126137, "grad_norm": 1.3022645711898804, "learning_rate": 6.088e-07, "loss": 0.0205, "step": 46957 }, { "epoch": 2.503492029642267, "grad_norm": 1.3992549180984497, "learning_rate": 6.086e-07, "loss": 0.007, "step": 46958 }, { "epoch": 2.5035453430719197, "grad_norm": 2.0359723567962646, "learning_rate": 6.084000000000001e-07, "loss": 0.0089, "step": 46959 }, { "epoch": 2.503598656501573, "grad_norm": 0.46178749203681946, "learning_rate": 6.082000000000001e-07, "loss": 0.0022, "step": 46960 }, { "epoch": 2.5036519699312256, "grad_norm": 1.925350308418274, "learning_rate": 6.08e-07, "loss": 0.0251, "step": 46961 }, { "epoch": 2.503705283360879, "grad_norm": 2.581657886505127, "learning_rate": 6.078e-07, "loss": 0.0108, "step": 46962 }, { "epoch": 2.5037585967905316, "grad_norm": 1.1158019304275513, "learning_rate": 6.076e-07, "loss": 0.0059, "step": 46963 }, { "epoch": 2.5038119102201843, "grad_norm": 1.912835955619812, "learning_rate": 6.074000000000001e-07, "loss": 0.0097, "step": 46964 }, { "epoch": 2.5038652236498375, "grad_norm": 0.33256369829177856, "learning_rate": 6.072000000000001e-07, "loss": 0.0025, "step": 46965 }, { "epoch": 2.5039185370794903, "grad_norm": 0.39104148745536804, "learning_rate": 6.07e-07, "loss": 0.0026, "step": 46966 }, { "epoch": 2.503971850509143, "grad_norm": 1.9089480638504028, "learning_rate": 6.068000000000001e-07, "loss": 0.0192, "step": 46967 }, { "epoch": 2.5040251639387963, "grad_norm": 2.45414400100708, "learning_rate": 6.066e-07, "loss": 0.0068, "step": 46968 }, { "epoch": 2.504078477368449, "grad_norm": 2.2097437381744385, "learning_rate": 6.064e-07, "loss": 0.0241, "step": 46969 }, { "epoch": 2.5041317907981018, "grad_norm": 4.244110107421875, "learning_rate": 6.062e-07, "loss": 0.0135, "step": 46970 }, { "epoch": 2.504185104227755, "grad_norm": 1.2956924438476562, "learning_rate": 6.060000000000001e-07, "loss": 0.0108, "step": 46971 }, { "epoch": 2.504238417657408, "grad_norm": 3.001112699508667, "learning_rate": 6.058000000000001e-07, "loss": 0.0483, "step": 46972 }, { "epoch": 2.504291731087061, "grad_norm": 0.502249538898468, "learning_rate": 6.056000000000001e-07, "loss": 0.0033, "step": 46973 }, { "epoch": 2.5043450445167137, "grad_norm": 2.1224119663238525, "learning_rate": 6.054000000000001e-07, "loss": 0.022, "step": 46974 }, { "epoch": 2.504398357946367, "grad_norm": 1.1420515775680542, "learning_rate": 6.052e-07, "loss": 0.0199, "step": 46975 }, { "epoch": 2.5044516713760197, "grad_norm": 1.4322001934051514, "learning_rate": 6.05e-07, "loss": 0.0164, "step": 46976 }, { "epoch": 2.5045049848056724, "grad_norm": 1.7453397512435913, "learning_rate": 6.048e-07, "loss": 0.0252, "step": 46977 }, { "epoch": 2.5045582982353256, "grad_norm": 2.0762219429016113, "learning_rate": 6.046e-07, "loss": 0.0313, "step": 46978 }, { "epoch": 2.5046116116649784, "grad_norm": 2.776071071624756, "learning_rate": 6.044000000000001e-07, "loss": 0.0096, "step": 46979 }, { "epoch": 2.504664925094631, "grad_norm": 1.1243754625320435, "learning_rate": 6.042000000000001e-07, "loss": 0.0042, "step": 46980 }, { "epoch": 2.5047182385242843, "grad_norm": 2.7780327796936035, "learning_rate": 6.040000000000001e-07, "loss": 0.0192, "step": 46981 }, { "epoch": 2.504771551953937, "grad_norm": 0.8549156785011292, "learning_rate": 6.038e-07, "loss": 0.0039, "step": 46982 }, { "epoch": 2.5048248653835903, "grad_norm": 0.9236465692520142, "learning_rate": 6.036e-07, "loss": 0.0045, "step": 46983 }, { "epoch": 2.504878178813243, "grad_norm": 1.3314483165740967, "learning_rate": 6.034e-07, "loss": 0.0197, "step": 46984 }, { "epoch": 2.5049314922428962, "grad_norm": 1.294171929359436, "learning_rate": 6.032e-07, "loss": 0.0039, "step": 46985 }, { "epoch": 2.504984805672549, "grad_norm": 1.7762435674667358, "learning_rate": 6.030000000000001e-07, "loss": 0.0088, "step": 46986 }, { "epoch": 2.5050381191022018, "grad_norm": 1.77945077419281, "learning_rate": 6.028000000000001e-07, "loss": 0.0372, "step": 46987 }, { "epoch": 2.505091432531855, "grad_norm": 2.3462979793548584, "learning_rate": 6.026000000000001e-07, "loss": 0.012, "step": 46988 }, { "epoch": 2.5051447459615077, "grad_norm": 0.7751457691192627, "learning_rate": 6.024e-07, "loss": 0.0048, "step": 46989 }, { "epoch": 2.5051980593911605, "grad_norm": 1.5069844722747803, "learning_rate": 6.022e-07, "loss": 0.0126, "step": 46990 }, { "epoch": 2.5052513728208137, "grad_norm": 1.638269066810608, "learning_rate": 6.02e-07, "loss": 0.0189, "step": 46991 }, { "epoch": 2.5053046862504664, "grad_norm": 0.5051426291465759, "learning_rate": 6.018e-07, "loss": 0.0029, "step": 46992 }, { "epoch": 2.505357999680119, "grad_norm": 0.24898304045200348, "learning_rate": 6.016e-07, "loss": 0.0019, "step": 46993 }, { "epoch": 2.5054113131097724, "grad_norm": 1.4683150053024292, "learning_rate": 6.014000000000001e-07, "loss": 0.0177, "step": 46994 }, { "epoch": 2.505464626539425, "grad_norm": 2.063427448272705, "learning_rate": 6.012000000000001e-07, "loss": 0.0138, "step": 46995 }, { "epoch": 2.5055179399690783, "grad_norm": 1.223504900932312, "learning_rate": 6.01e-07, "loss": 0.0123, "step": 46996 }, { "epoch": 2.505571253398731, "grad_norm": 1.2760652303695679, "learning_rate": 6.008e-07, "loss": 0.0119, "step": 46997 }, { "epoch": 2.5056245668283843, "grad_norm": 0.37437936663627625, "learning_rate": 6.006000000000001e-07, "loss": 0.0027, "step": 46998 }, { "epoch": 2.505677880258037, "grad_norm": 1.4704850912094116, "learning_rate": 6.004000000000001e-07, "loss": 0.0043, "step": 46999 }, { "epoch": 2.50573119368769, "grad_norm": 2.96686053276062, "learning_rate": 6.002e-07, "loss": 0.0156, "step": 47000 }, { "epoch": 2.505784507117343, "grad_norm": 2.891573429107666, "learning_rate": 6.000000000000001e-07, "loss": 0.0107, "step": 47001 }, { "epoch": 2.5058378205469958, "grad_norm": 1.2506457567214966, "learning_rate": 5.998000000000001e-07, "loss": 0.0208, "step": 47002 }, { "epoch": 2.5058911339766485, "grad_norm": 1.76739501953125, "learning_rate": 5.996e-07, "loss": 0.0193, "step": 47003 }, { "epoch": 2.5059444474063017, "grad_norm": 1.7024413347244263, "learning_rate": 5.994e-07, "loss": 0.0098, "step": 47004 }, { "epoch": 2.5059977608359545, "grad_norm": 1.7392350435256958, "learning_rate": 5.992e-07, "loss": 0.0137, "step": 47005 }, { "epoch": 2.5060510742656072, "grad_norm": 1.0951259136199951, "learning_rate": 5.990000000000001e-07, "loss": 0.0125, "step": 47006 }, { "epoch": 2.5061043876952604, "grad_norm": 2.0252461433410645, "learning_rate": 5.988000000000001e-07, "loss": 0.0128, "step": 47007 }, { "epoch": 2.506157701124913, "grad_norm": 1.725550889968872, "learning_rate": 5.986e-07, "loss": 0.0197, "step": 47008 }, { "epoch": 2.5062110145545664, "grad_norm": 1.603255033493042, "learning_rate": 5.984000000000001e-07, "loss": 0.0224, "step": 47009 }, { "epoch": 2.506264327984219, "grad_norm": 0.7459386587142944, "learning_rate": 5.982e-07, "loss": 0.0154, "step": 47010 }, { "epoch": 2.5063176414138724, "grad_norm": 3.0288095474243164, "learning_rate": 5.98e-07, "loss": 0.0213, "step": 47011 }, { "epoch": 2.506370954843525, "grad_norm": 2.373703718185425, "learning_rate": 5.978e-07, "loss": 0.0104, "step": 47012 }, { "epoch": 2.506424268273178, "grad_norm": 2.109341621398926, "learning_rate": 5.976e-07, "loss": 0.0438, "step": 47013 }, { "epoch": 2.506477581702831, "grad_norm": 1.4306813478469849, "learning_rate": 5.974000000000001e-07, "loss": 0.0107, "step": 47014 }, { "epoch": 2.506530895132484, "grad_norm": 0.29136428236961365, "learning_rate": 5.972000000000001e-07, "loss": 0.0024, "step": 47015 }, { "epoch": 2.5065842085621366, "grad_norm": 1.8303109407424927, "learning_rate": 5.970000000000001e-07, "loss": 0.0205, "step": 47016 }, { "epoch": 2.50663752199179, "grad_norm": 1.9128328561782837, "learning_rate": 5.968e-07, "loss": 0.0199, "step": 47017 }, { "epoch": 2.5066908354214426, "grad_norm": 1.6619194746017456, "learning_rate": 5.966e-07, "loss": 0.0143, "step": 47018 }, { "epoch": 2.5067441488510958, "grad_norm": 0.8542488217353821, "learning_rate": 5.964e-07, "loss": 0.005, "step": 47019 }, { "epoch": 2.5067974622807485, "grad_norm": 2.2328531742095947, "learning_rate": 5.962e-07, "loss": 0.0258, "step": 47020 }, { "epoch": 2.5068507757104017, "grad_norm": 5.499993801116943, "learning_rate": 5.960000000000001e-07, "loss": 0.0195, "step": 47021 }, { "epoch": 2.5069040891400545, "grad_norm": 3.1255078315734863, "learning_rate": 5.958000000000001e-07, "loss": 0.0155, "step": 47022 }, { "epoch": 2.5069574025697072, "grad_norm": 0.325086772441864, "learning_rate": 5.956000000000001e-07, "loss": 0.0025, "step": 47023 }, { "epoch": 2.5070107159993604, "grad_norm": 3.4470221996307373, "learning_rate": 5.954e-07, "loss": 0.0132, "step": 47024 }, { "epoch": 2.507064029429013, "grad_norm": 0.744124710559845, "learning_rate": 5.952e-07, "loss": 0.0037, "step": 47025 }, { "epoch": 2.507117342858666, "grad_norm": 1.957880973815918, "learning_rate": 5.95e-07, "loss": 0.0196, "step": 47026 }, { "epoch": 2.507170656288319, "grad_norm": 0.6691879630088806, "learning_rate": 5.948e-07, "loss": 0.003, "step": 47027 }, { "epoch": 2.507223969717972, "grad_norm": 1.2418198585510254, "learning_rate": 5.946e-07, "loss": 0.0183, "step": 47028 }, { "epoch": 2.5072772831476247, "grad_norm": 1.873812198638916, "learning_rate": 5.944000000000001e-07, "loss": 0.0196, "step": 47029 }, { "epoch": 2.507330596577278, "grad_norm": 0.17243708670139313, "learning_rate": 5.942000000000001e-07, "loss": 0.0014, "step": 47030 }, { "epoch": 2.5073839100069306, "grad_norm": 2.239065408706665, "learning_rate": 5.94e-07, "loss": 0.0066, "step": 47031 }, { "epoch": 2.507437223436584, "grad_norm": 4.233496189117432, "learning_rate": 5.938e-07, "loss": 0.0338, "step": 47032 }, { "epoch": 2.5074905368662366, "grad_norm": 3.197481155395508, "learning_rate": 5.936000000000001e-07, "loss": 0.0146, "step": 47033 }, { "epoch": 2.5075438502958898, "grad_norm": 3.7550015449523926, "learning_rate": 5.934e-07, "loss": 0.0216, "step": 47034 }, { "epoch": 2.5075971637255425, "grad_norm": 4.873031139373779, "learning_rate": 5.932e-07, "loss": 0.0205, "step": 47035 }, { "epoch": 2.5076504771551953, "grad_norm": 2.1488230228424072, "learning_rate": 5.930000000000001e-07, "loss": 0.0056, "step": 47036 }, { "epoch": 2.5077037905848485, "grad_norm": 0.4235033094882965, "learning_rate": 5.928000000000001e-07, "loss": 0.0027, "step": 47037 }, { "epoch": 2.5077571040145012, "grad_norm": 0.442108690738678, "learning_rate": 5.926e-07, "loss": 0.003, "step": 47038 }, { "epoch": 2.507810417444154, "grad_norm": 1.7536156177520752, "learning_rate": 5.924e-07, "loss": 0.0231, "step": 47039 }, { "epoch": 2.507863730873807, "grad_norm": 1.837849736213684, "learning_rate": 5.922e-07, "loss": 0.0243, "step": 47040 }, { "epoch": 2.50791704430346, "grad_norm": 2.399219512939453, "learning_rate": 5.920000000000001e-07, "loss": 0.0116, "step": 47041 }, { "epoch": 2.5079703577331127, "grad_norm": 3.9393668174743652, "learning_rate": 5.918000000000001e-07, "loss": 0.022, "step": 47042 }, { "epoch": 2.508023671162766, "grad_norm": 2.0992825031280518, "learning_rate": 5.916e-07, "loss": 0.0164, "step": 47043 }, { "epoch": 2.5080769845924187, "grad_norm": 4.700223445892334, "learning_rate": 5.914000000000001e-07, "loss": 0.0158, "step": 47044 }, { "epoch": 2.508130298022072, "grad_norm": 1.2042131423950195, "learning_rate": 5.912e-07, "loss": 0.0054, "step": 47045 }, { "epoch": 2.5081836114517246, "grad_norm": 1.3123395442962646, "learning_rate": 5.91e-07, "loss": 0.0441, "step": 47046 }, { "epoch": 2.508236924881378, "grad_norm": 1.6488313674926758, "learning_rate": 5.908e-07, "loss": 0.0245, "step": 47047 }, { "epoch": 2.5082902383110306, "grad_norm": 1.0483875274658203, "learning_rate": 5.906000000000001e-07, "loss": 0.0069, "step": 47048 }, { "epoch": 2.5083435517406834, "grad_norm": 1.5355769395828247, "learning_rate": 5.904000000000001e-07, "loss": 0.0268, "step": 47049 }, { "epoch": 2.5083968651703366, "grad_norm": 3.304298162460327, "learning_rate": 5.902000000000001e-07, "loss": 0.0125, "step": 47050 }, { "epoch": 2.5084501785999893, "grad_norm": 1.6301071643829346, "learning_rate": 5.900000000000001e-07, "loss": 0.0155, "step": 47051 }, { "epoch": 2.508503492029642, "grad_norm": 1.3962410688400269, "learning_rate": 5.898000000000001e-07, "loss": 0.0065, "step": 47052 }, { "epoch": 2.5085568054592953, "grad_norm": 2.9655659198760986, "learning_rate": 5.896e-07, "loss": 0.0434, "step": 47053 }, { "epoch": 2.508610118888948, "grad_norm": 1.3417398929595947, "learning_rate": 5.894e-07, "loss": 0.0219, "step": 47054 }, { "epoch": 2.508663432318601, "grad_norm": 0.7339754104614258, "learning_rate": 5.892e-07, "loss": 0.0041, "step": 47055 }, { "epoch": 2.508716745748254, "grad_norm": 0.4187068045139313, "learning_rate": 5.890000000000001e-07, "loss": 0.0029, "step": 47056 }, { "epoch": 2.508770059177907, "grad_norm": 1.6620162725448608, "learning_rate": 5.888000000000001e-07, "loss": 0.0126, "step": 47057 }, { "epoch": 2.50882337260756, "grad_norm": 2.3327527046203613, "learning_rate": 5.886000000000001e-07, "loss": 0.0175, "step": 47058 }, { "epoch": 2.5088766860372127, "grad_norm": 3.432455062866211, "learning_rate": 5.884000000000001e-07, "loss": 0.0339, "step": 47059 }, { "epoch": 2.508929999466866, "grad_norm": 1.8233304023742676, "learning_rate": 5.882e-07, "loss": 0.0206, "step": 47060 }, { "epoch": 2.5089833128965187, "grad_norm": 2.752878189086914, "learning_rate": 5.88e-07, "loss": 0.0284, "step": 47061 }, { "epoch": 2.5090366263261714, "grad_norm": 2.2436318397521973, "learning_rate": 5.878e-07, "loss": 0.0128, "step": 47062 }, { "epoch": 2.5090899397558246, "grad_norm": 2.7333364486694336, "learning_rate": 5.876e-07, "loss": 0.0216, "step": 47063 }, { "epoch": 2.5091432531854774, "grad_norm": 1.2903509140014648, "learning_rate": 5.874000000000001e-07, "loss": 0.0277, "step": 47064 }, { "epoch": 2.50919656661513, "grad_norm": 2.5335185527801514, "learning_rate": 5.872000000000001e-07, "loss": 0.0334, "step": 47065 }, { "epoch": 2.5092498800447833, "grad_norm": 0.8104720711708069, "learning_rate": 5.870000000000001e-07, "loss": 0.0053, "step": 47066 }, { "epoch": 2.509303193474436, "grad_norm": 2.912451982498169, "learning_rate": 5.868e-07, "loss": 0.018, "step": 47067 }, { "epoch": 2.5093565069040893, "grad_norm": 2.1035642623901367, "learning_rate": 5.866e-07, "loss": 0.007, "step": 47068 }, { "epoch": 2.509409820333742, "grad_norm": 2.5300962924957275, "learning_rate": 5.864e-07, "loss": 0.0179, "step": 47069 }, { "epoch": 2.5094631337633952, "grad_norm": 3.1468770503997803, "learning_rate": 5.862e-07, "loss": 0.0119, "step": 47070 }, { "epoch": 2.509516447193048, "grad_norm": 1.4450880289077759, "learning_rate": 5.860000000000001e-07, "loss": 0.0165, "step": 47071 }, { "epoch": 2.5095697606227008, "grad_norm": 1.3032422065734863, "learning_rate": 5.858000000000001e-07, "loss": 0.0247, "step": 47072 }, { "epoch": 2.509623074052354, "grad_norm": 3.227654218673706, "learning_rate": 5.856000000000001e-07, "loss": 0.0245, "step": 47073 }, { "epoch": 2.5096763874820067, "grad_norm": 2.341646194458008, "learning_rate": 5.854e-07, "loss": 0.0097, "step": 47074 }, { "epoch": 2.5097297009116595, "grad_norm": 2.260349988937378, "learning_rate": 5.852e-07, "loss": 0.0294, "step": 47075 }, { "epoch": 2.5097830143413127, "grad_norm": 1.5627460479736328, "learning_rate": 5.850000000000001e-07, "loss": 0.0339, "step": 47076 }, { "epoch": 2.5098363277709654, "grad_norm": 1.756572961807251, "learning_rate": 5.848e-07, "loss": 0.0063, "step": 47077 }, { "epoch": 2.509889641200618, "grad_norm": 1.6976932287216187, "learning_rate": 5.846e-07, "loss": 0.0061, "step": 47078 }, { "epoch": 2.5099429546302714, "grad_norm": 4.544451713562012, "learning_rate": 5.844000000000001e-07, "loss": 0.0195, "step": 47079 }, { "epoch": 2.509996268059924, "grad_norm": 0.44692856073379517, "learning_rate": 5.842000000000001e-07, "loss": 0.0038, "step": 47080 }, { "epoch": 2.5100495814895774, "grad_norm": 1.5941983461380005, "learning_rate": 5.84e-07, "loss": 0.0077, "step": 47081 }, { "epoch": 2.51010289491923, "grad_norm": 2.15624737739563, "learning_rate": 5.838e-07, "loss": 0.0161, "step": 47082 }, { "epoch": 2.5101562083488833, "grad_norm": 0.5317786335945129, "learning_rate": 5.836000000000001e-07, "loss": 0.0035, "step": 47083 }, { "epoch": 2.510209521778536, "grad_norm": 1.188553810119629, "learning_rate": 5.834000000000001e-07, "loss": 0.0204, "step": 47084 }, { "epoch": 2.510262835208189, "grad_norm": 2.103013277053833, "learning_rate": 5.832e-07, "loss": 0.0107, "step": 47085 }, { "epoch": 2.510316148637842, "grad_norm": 2.333454132080078, "learning_rate": 5.830000000000001e-07, "loss": 0.0107, "step": 47086 }, { "epoch": 2.510369462067495, "grad_norm": 4.248865127563477, "learning_rate": 5.828000000000001e-07, "loss": 0.0203, "step": 47087 }, { "epoch": 2.5104227754971475, "grad_norm": 2.2719011306762695, "learning_rate": 5.826e-07, "loss": 0.0104, "step": 47088 }, { "epoch": 2.5104760889268007, "grad_norm": 1.7471998929977417, "learning_rate": 5.824e-07, "loss": 0.0156, "step": 47089 }, { "epoch": 2.5105294023564535, "grad_norm": 2.3939032554626465, "learning_rate": 5.822e-07, "loss": 0.0078, "step": 47090 }, { "epoch": 2.5105827157861063, "grad_norm": 3.2416253089904785, "learning_rate": 5.820000000000001e-07, "loss": 0.0216, "step": 47091 }, { "epoch": 2.5106360292157595, "grad_norm": 1.7054423093795776, "learning_rate": 5.818000000000001e-07, "loss": 0.0117, "step": 47092 }, { "epoch": 2.510689342645412, "grad_norm": 1.6495033502578735, "learning_rate": 5.816000000000001e-07, "loss": 0.0214, "step": 47093 }, { "epoch": 2.5107426560750654, "grad_norm": 1.2885555028915405, "learning_rate": 5.814000000000001e-07, "loss": 0.0049, "step": 47094 }, { "epoch": 2.510795969504718, "grad_norm": 3.260359287261963, "learning_rate": 5.812e-07, "loss": 0.0307, "step": 47095 }, { "epoch": 2.5108492829343714, "grad_norm": 1.3428090810775757, "learning_rate": 5.81e-07, "loss": 0.004, "step": 47096 }, { "epoch": 2.510902596364024, "grad_norm": 0.6261933445930481, "learning_rate": 5.808e-07, "loss": 0.0025, "step": 47097 }, { "epoch": 2.510955909793677, "grad_norm": 1.807984709739685, "learning_rate": 5.806000000000001e-07, "loss": 0.0045, "step": 47098 }, { "epoch": 2.51100922322333, "grad_norm": 1.8499377965927124, "learning_rate": 5.804000000000001e-07, "loss": 0.026, "step": 47099 }, { "epoch": 2.511062536652983, "grad_norm": 0.48215317726135254, "learning_rate": 5.802000000000001e-07, "loss": 0.0018, "step": 47100 }, { "epoch": 2.5111158500826356, "grad_norm": 2.6109964847564697, "learning_rate": 5.800000000000001e-07, "loss": 0.0309, "step": 47101 }, { "epoch": 2.511169163512289, "grad_norm": 1.9494308233261108, "learning_rate": 5.798e-07, "loss": 0.008, "step": 47102 }, { "epoch": 2.5112224769419416, "grad_norm": 2.1961302757263184, "learning_rate": 5.796e-07, "loss": 0.0196, "step": 47103 }, { "epoch": 2.5112757903715948, "grad_norm": 0.5325579047203064, "learning_rate": 5.794e-07, "loss": 0.0022, "step": 47104 }, { "epoch": 2.5113291038012475, "grad_norm": 2.63078236579895, "learning_rate": 5.792e-07, "loss": 0.0238, "step": 47105 }, { "epoch": 2.5113824172309007, "grad_norm": 1.1152983903884888, "learning_rate": 5.790000000000001e-07, "loss": 0.0126, "step": 47106 }, { "epoch": 2.5114357306605535, "grad_norm": 0.8739365339279175, "learning_rate": 5.788000000000001e-07, "loss": 0.0136, "step": 47107 }, { "epoch": 2.5114890440902062, "grad_norm": 2.9705169200897217, "learning_rate": 5.786000000000001e-07, "loss": 0.0085, "step": 47108 }, { "epoch": 2.5115423575198594, "grad_norm": 1.124824047088623, "learning_rate": 5.784e-07, "loss": 0.0054, "step": 47109 }, { "epoch": 2.511595670949512, "grad_norm": 2.4584903717041016, "learning_rate": 5.782e-07, "loss": 0.0088, "step": 47110 }, { "epoch": 2.511648984379165, "grad_norm": 1.7824451923370361, "learning_rate": 5.78e-07, "loss": 0.0048, "step": 47111 }, { "epoch": 2.511702297808818, "grad_norm": 3.555093288421631, "learning_rate": 5.778e-07, "loss": 0.035, "step": 47112 }, { "epoch": 2.511755611238471, "grad_norm": 2.439812660217285, "learning_rate": 5.776e-07, "loss": 0.0319, "step": 47113 }, { "epoch": 2.5118089246681237, "grad_norm": 2.235297918319702, "learning_rate": 5.774000000000001e-07, "loss": 0.0177, "step": 47114 }, { "epoch": 2.511862238097777, "grad_norm": 2.858085870742798, "learning_rate": 5.772000000000001e-07, "loss": 0.0263, "step": 47115 }, { "epoch": 2.5119155515274296, "grad_norm": 2.8148226737976074, "learning_rate": 5.77e-07, "loss": 0.0276, "step": 47116 }, { "epoch": 2.511968864957083, "grad_norm": 2.0206592082977295, "learning_rate": 5.768e-07, "loss": 0.0279, "step": 47117 }, { "epoch": 2.5120221783867356, "grad_norm": 4.786863327026367, "learning_rate": 5.766000000000001e-07, "loss": 0.0254, "step": 47118 }, { "epoch": 2.512075491816389, "grad_norm": 1.8339780569076538, "learning_rate": 5.764e-07, "loss": 0.0199, "step": 47119 }, { "epoch": 2.5121288052460415, "grad_norm": 0.47894173860549927, "learning_rate": 5.762e-07, "loss": 0.0029, "step": 47120 }, { "epoch": 2.5121821186756943, "grad_norm": 1.2129424810409546, "learning_rate": 5.760000000000001e-07, "loss": 0.0038, "step": 47121 }, { "epoch": 2.5122354321053475, "grad_norm": 1.349984884262085, "learning_rate": 5.758000000000001e-07, "loss": 0.0153, "step": 47122 }, { "epoch": 2.5122887455350003, "grad_norm": 1.9128470420837402, "learning_rate": 5.756e-07, "loss": 0.0115, "step": 47123 }, { "epoch": 2.512342058964653, "grad_norm": 1.432266116142273, "learning_rate": 5.754e-07, "loss": 0.0115, "step": 47124 }, { "epoch": 2.512395372394306, "grad_norm": 2.7433736324310303, "learning_rate": 5.752e-07, "loss": 0.0097, "step": 47125 }, { "epoch": 2.512448685823959, "grad_norm": 1.3779799938201904, "learning_rate": 5.750000000000001e-07, "loss": 0.0065, "step": 47126 }, { "epoch": 2.5125019992536117, "grad_norm": 1.9481714963912964, "learning_rate": 5.748000000000001e-07, "loss": 0.0222, "step": 47127 }, { "epoch": 2.512555312683265, "grad_norm": 2.7864999771118164, "learning_rate": 5.746e-07, "loss": 0.0098, "step": 47128 }, { "epoch": 2.5126086261129177, "grad_norm": 1.902393102645874, "learning_rate": 5.744000000000001e-07, "loss": 0.0078, "step": 47129 }, { "epoch": 2.512661939542571, "grad_norm": 1.9980416297912598, "learning_rate": 5.742e-07, "loss": 0.0147, "step": 47130 }, { "epoch": 2.5127152529722236, "grad_norm": 1.4168211221694946, "learning_rate": 5.74e-07, "loss": 0.0066, "step": 47131 }, { "epoch": 2.512768566401877, "grad_norm": 1.1341606378555298, "learning_rate": 5.738e-07, "loss": 0.0192, "step": 47132 }, { "epoch": 2.5128218798315296, "grad_norm": 2.9620840549468994, "learning_rate": 5.736000000000001e-07, "loss": 0.0068, "step": 47133 }, { "epoch": 2.5128751932611824, "grad_norm": 0.404485821723938, "learning_rate": 5.734000000000001e-07, "loss": 0.0028, "step": 47134 }, { "epoch": 2.5129285066908356, "grad_norm": 1.3182470798492432, "learning_rate": 5.732000000000001e-07, "loss": 0.0375, "step": 47135 }, { "epoch": 2.5129818201204883, "grad_norm": 4.870708465576172, "learning_rate": 5.730000000000001e-07, "loss": 0.0161, "step": 47136 }, { "epoch": 2.513035133550141, "grad_norm": 1.9908397197723389, "learning_rate": 5.728e-07, "loss": 0.0114, "step": 47137 }, { "epoch": 2.5130884469797943, "grad_norm": 1.8591063022613525, "learning_rate": 5.726e-07, "loss": 0.0231, "step": 47138 }, { "epoch": 2.513141760409447, "grad_norm": 1.4343631267547607, "learning_rate": 5.724e-07, "loss": 0.0301, "step": 47139 }, { "epoch": 2.5131950738391002, "grad_norm": 0.7367507815361023, "learning_rate": 5.722e-07, "loss": 0.0035, "step": 47140 }, { "epoch": 2.513248387268753, "grad_norm": 4.036891460418701, "learning_rate": 5.720000000000001e-07, "loss": 0.0267, "step": 47141 }, { "epoch": 2.513301700698406, "grad_norm": 1.4856274127960205, "learning_rate": 5.718000000000001e-07, "loss": 0.0166, "step": 47142 }, { "epoch": 2.513355014128059, "grad_norm": 2.630333185195923, "learning_rate": 5.716000000000001e-07, "loss": 0.022, "step": 47143 }, { "epoch": 2.5134083275577117, "grad_norm": 3.985081911087036, "learning_rate": 5.714e-07, "loss": 0.0212, "step": 47144 }, { "epoch": 2.513461640987365, "grad_norm": 2.471632242202759, "learning_rate": 5.712e-07, "loss": 0.0136, "step": 47145 }, { "epoch": 2.5135149544170177, "grad_norm": 0.6069352626800537, "learning_rate": 5.71e-07, "loss": 0.0029, "step": 47146 }, { "epoch": 2.5135682678466704, "grad_norm": 3.017214059829712, "learning_rate": 5.708e-07, "loss": 0.0307, "step": 47147 }, { "epoch": 2.5136215812763236, "grad_norm": 1.3015562295913696, "learning_rate": 5.706000000000001e-07, "loss": 0.0086, "step": 47148 }, { "epoch": 2.5136748947059764, "grad_norm": 3.5982818603515625, "learning_rate": 5.704000000000001e-07, "loss": 0.0444, "step": 47149 }, { "epoch": 2.513728208135629, "grad_norm": 0.8352086544036865, "learning_rate": 5.702000000000001e-07, "loss": 0.0114, "step": 47150 }, { "epoch": 2.5137815215652823, "grad_norm": 0.5710079073905945, "learning_rate": 5.7e-07, "loss": 0.0023, "step": 47151 }, { "epoch": 2.513834834994935, "grad_norm": 4.642932891845703, "learning_rate": 5.698e-07, "loss": 0.0315, "step": 47152 }, { "epoch": 2.5138881484245883, "grad_norm": 2.4515199661254883, "learning_rate": 5.696e-07, "loss": 0.0194, "step": 47153 }, { "epoch": 2.513941461854241, "grad_norm": 1.0019694566726685, "learning_rate": 5.694e-07, "loss": 0.0144, "step": 47154 }, { "epoch": 2.5139947752838943, "grad_norm": 1.5531891584396362, "learning_rate": 5.692e-07, "loss": 0.0193, "step": 47155 }, { "epoch": 2.514048088713547, "grad_norm": 3.11324405670166, "learning_rate": 5.690000000000001e-07, "loss": 0.011, "step": 47156 }, { "epoch": 2.5141014021431998, "grad_norm": 2.2201485633850098, "learning_rate": 5.688000000000001e-07, "loss": 0.0173, "step": 47157 }, { "epoch": 2.514154715572853, "grad_norm": 1.9060524702072144, "learning_rate": 5.686e-07, "loss": 0.02, "step": 47158 }, { "epoch": 2.5142080290025057, "grad_norm": 0.930907666683197, "learning_rate": 5.684e-07, "loss": 0.0044, "step": 47159 }, { "epoch": 2.5142613424321585, "grad_norm": 0.7154898047447205, "learning_rate": 5.682e-07, "loss": 0.0032, "step": 47160 }, { "epoch": 2.5143146558618117, "grad_norm": 2.0406746864318848, "learning_rate": 5.680000000000001e-07, "loss": 0.0077, "step": 47161 }, { "epoch": 2.5143679692914644, "grad_norm": 3.240736246109009, "learning_rate": 5.678e-07, "loss": 0.031, "step": 47162 }, { "epoch": 2.514421282721117, "grad_norm": 1.4788320064544678, "learning_rate": 5.676e-07, "loss": 0.005, "step": 47163 }, { "epoch": 2.5144745961507704, "grad_norm": 2.238077402114868, "learning_rate": 5.674000000000001e-07, "loss": 0.0059, "step": 47164 }, { "epoch": 2.514527909580423, "grad_norm": 2.132966995239258, "learning_rate": 5.672e-07, "loss": 0.0179, "step": 47165 }, { "epoch": 2.5145812230100764, "grad_norm": 1.2822681665420532, "learning_rate": 5.67e-07, "loss": 0.008, "step": 47166 }, { "epoch": 2.514634536439729, "grad_norm": 0.6878233551979065, "learning_rate": 5.668e-07, "loss": 0.0052, "step": 47167 }, { "epoch": 2.5146878498693823, "grad_norm": 2.540224552154541, "learning_rate": 5.666000000000001e-07, "loss": 0.0169, "step": 47168 }, { "epoch": 2.514741163299035, "grad_norm": 1.7962627410888672, "learning_rate": 5.664000000000001e-07, "loss": 0.0081, "step": 47169 }, { "epoch": 2.514794476728688, "grad_norm": 7.488855361938477, "learning_rate": 5.662e-07, "loss": 0.0207, "step": 47170 }, { "epoch": 2.514847790158341, "grad_norm": 0.9000612497329712, "learning_rate": 5.660000000000001e-07, "loss": 0.0132, "step": 47171 }, { "epoch": 2.514901103587994, "grad_norm": 2.0067873001098633, "learning_rate": 5.658e-07, "loss": 0.018, "step": 47172 }, { "epoch": 2.5149544170176465, "grad_norm": 0.708561897277832, "learning_rate": 5.656e-07, "loss": 0.0034, "step": 47173 }, { "epoch": 2.5150077304472997, "grad_norm": 0.6693531274795532, "learning_rate": 5.654e-07, "loss": 0.0026, "step": 47174 }, { "epoch": 2.5150610438769525, "grad_norm": 2.5385384559631348, "learning_rate": 5.652e-07, "loss": 0.0092, "step": 47175 }, { "epoch": 2.5151143573066053, "grad_norm": 0.6048274040222168, "learning_rate": 5.650000000000001e-07, "loss": 0.0032, "step": 47176 }, { "epoch": 2.5151676707362585, "grad_norm": 4.825777530670166, "learning_rate": 5.648000000000001e-07, "loss": 0.0652, "step": 47177 }, { "epoch": 2.5152209841659117, "grad_norm": 2.7964391708374023, "learning_rate": 5.646000000000001e-07, "loss": 0.0665, "step": 47178 }, { "epoch": 2.5152742975955644, "grad_norm": 2.3788323402404785, "learning_rate": 5.644e-07, "loss": 0.0146, "step": 47179 }, { "epoch": 2.515327611025217, "grad_norm": 2.3740084171295166, "learning_rate": 5.642e-07, "loss": 0.0204, "step": 47180 }, { "epoch": 2.5153809244548704, "grad_norm": 3.4853827953338623, "learning_rate": 5.64e-07, "loss": 0.0172, "step": 47181 }, { "epoch": 2.515434237884523, "grad_norm": 1.1947290897369385, "learning_rate": 5.638e-07, "loss": 0.0192, "step": 47182 }, { "epoch": 2.515487551314176, "grad_norm": 1.384272575378418, "learning_rate": 5.636000000000001e-07, "loss": 0.014, "step": 47183 }, { "epoch": 2.515540864743829, "grad_norm": 0.4345620572566986, "learning_rate": 5.634000000000001e-07, "loss": 0.0027, "step": 47184 }, { "epoch": 2.515594178173482, "grad_norm": 2.7702841758728027, "learning_rate": 5.632000000000001e-07, "loss": 0.0215, "step": 47185 }, { "epoch": 2.5156474916031346, "grad_norm": 2.797985792160034, "learning_rate": 5.63e-07, "loss": 0.0301, "step": 47186 }, { "epoch": 2.515700805032788, "grad_norm": 2.1583690643310547, "learning_rate": 5.628e-07, "loss": 0.0298, "step": 47187 }, { "epoch": 2.5157541184624406, "grad_norm": 5.0633320808410645, "learning_rate": 5.626e-07, "loss": 0.0185, "step": 47188 }, { "epoch": 2.5158074318920938, "grad_norm": 1.5706143379211426, "learning_rate": 5.624e-07, "loss": 0.0063, "step": 47189 }, { "epoch": 2.5158607453217465, "grad_norm": 1.781104326248169, "learning_rate": 5.622e-07, "loss": 0.0066, "step": 47190 }, { "epoch": 2.5159140587513997, "grad_norm": 2.2588939666748047, "learning_rate": 5.620000000000001e-07, "loss": 0.005, "step": 47191 }, { "epoch": 2.5159673721810525, "grad_norm": 1.4963189363479614, "learning_rate": 5.618000000000001e-07, "loss": 0.0078, "step": 47192 }, { "epoch": 2.5160206856107052, "grad_norm": 2.163759231567383, "learning_rate": 5.616e-07, "loss": 0.0261, "step": 47193 }, { "epoch": 2.5160739990403584, "grad_norm": 0.257295161485672, "learning_rate": 5.614e-07, "loss": 0.0019, "step": 47194 }, { "epoch": 2.516127312470011, "grad_norm": 1.90854012966156, "learning_rate": 5.612000000000001e-07, "loss": 0.0251, "step": 47195 }, { "epoch": 2.516180625899664, "grad_norm": 0.24730266630649567, "learning_rate": 5.61e-07, "loss": 0.0022, "step": 47196 }, { "epoch": 2.516233939329317, "grad_norm": 1.1169798374176025, "learning_rate": 5.608e-07, "loss": 0.0046, "step": 47197 }, { "epoch": 2.51628725275897, "grad_norm": 1.5403040647506714, "learning_rate": 5.606000000000001e-07, "loss": 0.0082, "step": 47198 }, { "epoch": 2.5163405661886227, "grad_norm": 0.8562338352203369, "learning_rate": 5.604000000000001e-07, "loss": 0.0023, "step": 47199 }, { "epoch": 2.516393879618276, "grad_norm": 3.0286190509796143, "learning_rate": 5.602000000000001e-07, "loss": 0.0605, "step": 47200 }, { "epoch": 2.5164471930479286, "grad_norm": 2.721825361251831, "learning_rate": 5.6e-07, "loss": 0.0208, "step": 47201 }, { "epoch": 2.516500506477582, "grad_norm": 1.2513055801391602, "learning_rate": 5.598e-07, "loss": 0.0236, "step": 47202 }, { "epoch": 2.5165538199072346, "grad_norm": 3.225604295730591, "learning_rate": 5.596000000000001e-07, "loss": 0.0213, "step": 47203 }, { "epoch": 2.516607133336888, "grad_norm": 2.098783493041992, "learning_rate": 5.594e-07, "loss": 0.0076, "step": 47204 }, { "epoch": 2.5166604467665405, "grad_norm": 0.941315233707428, "learning_rate": 5.592e-07, "loss": 0.013, "step": 47205 }, { "epoch": 2.5167137601961933, "grad_norm": 0.8630419373512268, "learning_rate": 5.590000000000001e-07, "loss": 0.0044, "step": 47206 }, { "epoch": 2.5167670736258465, "grad_norm": 1.7939661741256714, "learning_rate": 5.588000000000001e-07, "loss": 0.0149, "step": 47207 }, { "epoch": 2.5168203870554993, "grad_norm": 1.9733244180679321, "learning_rate": 5.586e-07, "loss": 0.0133, "step": 47208 }, { "epoch": 2.516873700485152, "grad_norm": 4.790864944458008, "learning_rate": 5.584e-07, "loss": 0.0084, "step": 47209 }, { "epoch": 2.5169270139148052, "grad_norm": 2.5125715732574463, "learning_rate": 5.582e-07, "loss": 0.0235, "step": 47210 }, { "epoch": 2.516980327344458, "grad_norm": 1.8820663690567017, "learning_rate": 5.580000000000001e-07, "loss": 0.007, "step": 47211 }, { "epoch": 2.5170336407741107, "grad_norm": 1.555181860923767, "learning_rate": 5.578000000000001e-07, "loss": 0.0114, "step": 47212 }, { "epoch": 2.517086954203764, "grad_norm": 3.028104305267334, "learning_rate": 5.576e-07, "loss": 0.0249, "step": 47213 }, { "epoch": 2.5171402676334167, "grad_norm": 1.6363528966903687, "learning_rate": 5.574000000000001e-07, "loss": 0.0076, "step": 47214 }, { "epoch": 2.51719358106307, "grad_norm": 2.366478443145752, "learning_rate": 5.572e-07, "loss": 0.0224, "step": 47215 }, { "epoch": 2.5172468944927227, "grad_norm": 1.708439588546753, "learning_rate": 5.57e-07, "loss": 0.0196, "step": 47216 }, { "epoch": 2.517300207922376, "grad_norm": 2.313514471054077, "learning_rate": 5.568e-07, "loss": 0.0194, "step": 47217 }, { "epoch": 2.5173535213520286, "grad_norm": 1.484915018081665, "learning_rate": 5.566000000000001e-07, "loss": 0.0308, "step": 47218 }, { "epoch": 2.5174068347816814, "grad_norm": 0.6633838415145874, "learning_rate": 5.564000000000001e-07, "loss": 0.0035, "step": 47219 }, { "epoch": 2.5174601482113346, "grad_norm": 1.0030790567398071, "learning_rate": 5.562000000000001e-07, "loss": 0.0117, "step": 47220 }, { "epoch": 2.5175134616409873, "grad_norm": 1.5174729824066162, "learning_rate": 5.560000000000001e-07, "loss": 0.0035, "step": 47221 }, { "epoch": 2.51756677507064, "grad_norm": 0.2828216254711151, "learning_rate": 5.558e-07, "loss": 0.0018, "step": 47222 }, { "epoch": 2.5176200885002933, "grad_norm": 0.5635886788368225, "learning_rate": 5.556e-07, "loss": 0.0038, "step": 47223 }, { "epoch": 2.517673401929946, "grad_norm": 1.7821176052093506, "learning_rate": 5.554e-07, "loss": 0.0299, "step": 47224 }, { "epoch": 2.5177267153595992, "grad_norm": 0.6836138963699341, "learning_rate": 5.552e-07, "loss": 0.0043, "step": 47225 }, { "epoch": 2.517780028789252, "grad_norm": 1.258741855621338, "learning_rate": 5.550000000000001e-07, "loss": 0.0058, "step": 47226 }, { "epoch": 2.517833342218905, "grad_norm": 1.4242912530899048, "learning_rate": 5.548000000000001e-07, "loss": 0.0107, "step": 47227 }, { "epoch": 2.517886655648558, "grad_norm": 1.4446734189987183, "learning_rate": 5.546000000000001e-07, "loss": 0.0112, "step": 47228 }, { "epoch": 2.5179399690782107, "grad_norm": 1.6360232830047607, "learning_rate": 5.544e-07, "loss": 0.0159, "step": 47229 }, { "epoch": 2.517993282507864, "grad_norm": 1.9720274209976196, "learning_rate": 5.542e-07, "loss": 0.0062, "step": 47230 }, { "epoch": 2.5180465959375167, "grad_norm": 1.4215768575668335, "learning_rate": 5.54e-07, "loss": 0.0066, "step": 47231 }, { "epoch": 2.5180999093671694, "grad_norm": 0.40617799758911133, "learning_rate": 5.538e-07, "loss": 0.0031, "step": 47232 }, { "epoch": 2.5181532227968226, "grad_norm": 1.167582631111145, "learning_rate": 5.536000000000001e-07, "loss": 0.0041, "step": 47233 }, { "epoch": 2.5182065362264754, "grad_norm": 3.0687859058380127, "learning_rate": 5.534000000000001e-07, "loss": 0.0448, "step": 47234 }, { "epoch": 2.518259849656128, "grad_norm": 2.6609578132629395, "learning_rate": 5.532000000000001e-07, "loss": 0.0056, "step": 47235 }, { "epoch": 2.5183131630857813, "grad_norm": 2.1387243270874023, "learning_rate": 5.53e-07, "loss": 0.0366, "step": 47236 }, { "epoch": 2.518366476515434, "grad_norm": 0.6762412190437317, "learning_rate": 5.528e-07, "loss": 0.0035, "step": 47237 }, { "epoch": 2.5184197899450873, "grad_norm": 3.669772148132324, "learning_rate": 5.526e-07, "loss": 0.0181, "step": 47238 }, { "epoch": 2.51847310337474, "grad_norm": 1.5980538129806519, "learning_rate": 5.524e-07, "loss": 0.024, "step": 47239 }, { "epoch": 2.5185264168043933, "grad_norm": 2.019294500350952, "learning_rate": 5.522e-07, "loss": 0.0111, "step": 47240 }, { "epoch": 2.518579730234046, "grad_norm": 1.4604874849319458, "learning_rate": 5.520000000000001e-07, "loss": 0.01, "step": 47241 }, { "epoch": 2.518633043663699, "grad_norm": 1.3056203126907349, "learning_rate": 5.518000000000001e-07, "loss": 0.0115, "step": 47242 }, { "epoch": 2.518686357093352, "grad_norm": 2.29197096824646, "learning_rate": 5.516e-07, "loss": 0.0191, "step": 47243 }, { "epoch": 2.5187396705230047, "grad_norm": 0.5781240463256836, "learning_rate": 5.514e-07, "loss": 0.0032, "step": 47244 }, { "epoch": 2.5187929839526575, "grad_norm": 3.75390625, "learning_rate": 5.512000000000001e-07, "loss": 0.023, "step": 47245 }, { "epoch": 2.5188462973823107, "grad_norm": 0.9568055868148804, "learning_rate": 5.510000000000001e-07, "loss": 0.0045, "step": 47246 }, { "epoch": 2.5188996108119635, "grad_norm": 0.6578405499458313, "learning_rate": 5.508e-07, "loss": 0.0034, "step": 47247 }, { "epoch": 2.518952924241616, "grad_norm": 0.5126574635505676, "learning_rate": 5.506000000000001e-07, "loss": 0.0034, "step": 47248 }, { "epoch": 2.5190062376712694, "grad_norm": 1.5898123979568481, "learning_rate": 5.504000000000001e-07, "loss": 0.0229, "step": 47249 }, { "epoch": 2.519059551100922, "grad_norm": 1.065748691558838, "learning_rate": 5.502e-07, "loss": 0.006, "step": 47250 }, { "epoch": 2.5191128645305754, "grad_norm": 2.104437828063965, "learning_rate": 5.5e-07, "loss": 0.0173, "step": 47251 }, { "epoch": 2.519166177960228, "grad_norm": 1.944136142730713, "learning_rate": 5.498e-07, "loss": 0.0126, "step": 47252 }, { "epoch": 2.5192194913898813, "grad_norm": 0.2528137266635895, "learning_rate": 5.496000000000001e-07, "loss": 0.0015, "step": 47253 }, { "epoch": 2.519272804819534, "grad_norm": 4.30657434463501, "learning_rate": 5.494000000000001e-07, "loss": 0.0225, "step": 47254 }, { "epoch": 2.519326118249187, "grad_norm": 2.0591859817504883, "learning_rate": 5.492e-07, "loss": 0.0097, "step": 47255 }, { "epoch": 2.51937943167884, "grad_norm": 1.2169595956802368, "learning_rate": 5.490000000000001e-07, "loss": 0.0113, "step": 47256 }, { "epoch": 2.519432745108493, "grad_norm": 2.4963886737823486, "learning_rate": 5.488e-07, "loss": 0.0201, "step": 47257 }, { "epoch": 2.5194860585381456, "grad_norm": 1.436026692390442, "learning_rate": 5.486e-07, "loss": 0.0175, "step": 47258 }, { "epoch": 2.5195393719677988, "grad_norm": 0.30975961685180664, "learning_rate": 5.484e-07, "loss": 0.0018, "step": 47259 }, { "epoch": 2.5195926853974515, "grad_norm": 2.3165390491485596, "learning_rate": 5.482e-07, "loss": 0.0123, "step": 47260 }, { "epoch": 2.5196459988271043, "grad_norm": 2.1935007572174072, "learning_rate": 5.480000000000001e-07, "loss": 0.0394, "step": 47261 }, { "epoch": 2.5196993122567575, "grad_norm": 2.326280355453491, "learning_rate": 5.478000000000001e-07, "loss": 0.0235, "step": 47262 }, { "epoch": 2.5197526256864107, "grad_norm": 2.1073200702667236, "learning_rate": 5.476000000000001e-07, "loss": 0.0093, "step": 47263 }, { "epoch": 2.5198059391160634, "grad_norm": 1.621354103088379, "learning_rate": 5.474e-07, "loss": 0.0334, "step": 47264 }, { "epoch": 2.519859252545716, "grad_norm": 2.1133275032043457, "learning_rate": 5.472e-07, "loss": 0.0134, "step": 47265 }, { "epoch": 2.5199125659753694, "grad_norm": 3.3586201667785645, "learning_rate": 5.47e-07, "loss": 0.0117, "step": 47266 }, { "epoch": 2.519965879405022, "grad_norm": 3.5092148780822754, "learning_rate": 5.468e-07, "loss": 0.0156, "step": 47267 }, { "epoch": 2.520019192834675, "grad_norm": 5.151117324829102, "learning_rate": 5.466000000000001e-07, "loss": 0.0167, "step": 47268 }, { "epoch": 2.520072506264328, "grad_norm": 1.6114627122879028, "learning_rate": 5.464000000000001e-07, "loss": 0.0329, "step": 47269 }, { "epoch": 2.520125819693981, "grad_norm": 0.9735040664672852, "learning_rate": 5.462000000000001e-07, "loss": 0.0041, "step": 47270 }, { "epoch": 2.5201791331236336, "grad_norm": 4.4156317710876465, "learning_rate": 5.46e-07, "loss": 0.031, "step": 47271 }, { "epoch": 2.520232446553287, "grad_norm": 0.6476696729660034, "learning_rate": 5.458e-07, "loss": 0.0042, "step": 47272 }, { "epoch": 2.5202857599829396, "grad_norm": 3.360534906387329, "learning_rate": 5.456e-07, "loss": 0.0131, "step": 47273 }, { "epoch": 2.520339073412593, "grad_norm": 1.9688985347747803, "learning_rate": 5.454e-07, "loss": 0.0085, "step": 47274 }, { "epoch": 2.5203923868422455, "grad_norm": 2.070239782333374, "learning_rate": 5.452e-07, "loss": 0.0069, "step": 47275 }, { "epoch": 2.5204457002718987, "grad_norm": 1.2452815771102905, "learning_rate": 5.450000000000001e-07, "loss": 0.0067, "step": 47276 }, { "epoch": 2.5204990137015515, "grad_norm": 0.39208316802978516, "learning_rate": 5.448000000000001e-07, "loss": 0.003, "step": 47277 }, { "epoch": 2.5205523271312043, "grad_norm": 0.9742641448974609, "learning_rate": 5.446e-07, "loss": 0.0052, "step": 47278 }, { "epoch": 2.5206056405608575, "grad_norm": 2.842172861099243, "learning_rate": 5.444e-07, "loss": 0.0267, "step": 47279 }, { "epoch": 2.52065895399051, "grad_norm": 0.521766185760498, "learning_rate": 5.442000000000001e-07, "loss": 0.0042, "step": 47280 }, { "epoch": 2.520712267420163, "grad_norm": 0.8915992975234985, "learning_rate": 5.44e-07, "loss": 0.005, "step": 47281 }, { "epoch": 2.520765580849816, "grad_norm": 2.9784979820251465, "learning_rate": 5.438e-07, "loss": 0.0184, "step": 47282 }, { "epoch": 2.520818894279469, "grad_norm": 0.8256347179412842, "learning_rate": 5.436000000000001e-07, "loss": 0.0041, "step": 47283 }, { "epoch": 2.5208722077091217, "grad_norm": 2.1074156761169434, "learning_rate": 5.434000000000001e-07, "loss": 0.0281, "step": 47284 }, { "epoch": 2.520925521138775, "grad_norm": 1.1441210508346558, "learning_rate": 5.432e-07, "loss": 0.0034, "step": 47285 }, { "epoch": 2.5209788345684276, "grad_norm": 1.743050456047058, "learning_rate": 5.43e-07, "loss": 0.0209, "step": 47286 }, { "epoch": 2.521032147998081, "grad_norm": 3.0298259258270264, "learning_rate": 5.428e-07, "loss": 0.0303, "step": 47287 }, { "epoch": 2.5210854614277336, "grad_norm": 1.5804272890090942, "learning_rate": 5.426000000000001e-07, "loss": 0.0117, "step": 47288 }, { "epoch": 2.521138774857387, "grad_norm": 2.3980748653411865, "learning_rate": 5.424e-07, "loss": 0.0297, "step": 47289 }, { "epoch": 2.5211920882870396, "grad_norm": 0.3745330572128296, "learning_rate": 5.422e-07, "loss": 0.0023, "step": 47290 }, { "epoch": 2.5212454017166923, "grad_norm": 1.8656831979751587, "learning_rate": 5.420000000000001e-07, "loss": 0.0153, "step": 47291 }, { "epoch": 2.5212987151463455, "grad_norm": 0.9897047281265259, "learning_rate": 5.418e-07, "loss": 0.0033, "step": 47292 }, { "epoch": 2.5213520285759983, "grad_norm": 2.6616017818450928, "learning_rate": 5.416e-07, "loss": 0.0187, "step": 47293 }, { "epoch": 2.521405342005651, "grad_norm": 1.0302960872650146, "learning_rate": 5.414e-07, "loss": 0.0249, "step": 47294 }, { "epoch": 2.5214586554353042, "grad_norm": 1.7847894430160522, "learning_rate": 5.412000000000001e-07, "loss": 0.006, "step": 47295 }, { "epoch": 2.521511968864957, "grad_norm": 2.447984218597412, "learning_rate": 5.410000000000001e-07, "loss": 0.0113, "step": 47296 }, { "epoch": 2.5215652822946097, "grad_norm": 0.9908000230789185, "learning_rate": 5.408000000000001e-07, "loss": 0.0185, "step": 47297 }, { "epoch": 2.521618595724263, "grad_norm": 4.551881790161133, "learning_rate": 5.406000000000001e-07, "loss": 0.0193, "step": 47298 }, { "epoch": 2.521671909153916, "grad_norm": 0.5856178402900696, "learning_rate": 5.404e-07, "loss": 0.0028, "step": 47299 }, { "epoch": 2.521725222583569, "grad_norm": 0.23726238310337067, "learning_rate": 5.402e-07, "loss": 0.0016, "step": 47300 }, { "epoch": 2.5217785360132217, "grad_norm": 0.7061827778816223, "learning_rate": 5.4e-07, "loss": 0.0038, "step": 47301 }, { "epoch": 2.521831849442875, "grad_norm": 2.7529609203338623, "learning_rate": 5.398e-07, "loss": 0.0146, "step": 47302 }, { "epoch": 2.5218851628725276, "grad_norm": 1.3684710264205933, "learning_rate": 5.396000000000001e-07, "loss": 0.0086, "step": 47303 }, { "epoch": 2.5219384763021804, "grad_norm": 2.46150541305542, "learning_rate": 5.394000000000001e-07, "loss": 0.0143, "step": 47304 }, { "epoch": 2.5219917897318336, "grad_norm": 0.23502963781356812, "learning_rate": 5.392000000000001e-07, "loss": 0.0018, "step": 47305 }, { "epoch": 2.5220451031614863, "grad_norm": 3.4477639198303223, "learning_rate": 5.39e-07, "loss": 0.0503, "step": 47306 }, { "epoch": 2.522098416591139, "grad_norm": 2.4759557247161865, "learning_rate": 5.388e-07, "loss": 0.0242, "step": 47307 }, { "epoch": 2.5221517300207923, "grad_norm": 1.7248765230178833, "learning_rate": 5.386e-07, "loss": 0.0124, "step": 47308 }, { "epoch": 2.522205043450445, "grad_norm": 0.7520515322685242, "learning_rate": 5.384e-07, "loss": 0.0029, "step": 47309 }, { "epoch": 2.5222583568800983, "grad_norm": 1.226998209953308, "learning_rate": 5.382000000000001e-07, "loss": 0.0203, "step": 47310 }, { "epoch": 2.522311670309751, "grad_norm": 0.3665927052497864, "learning_rate": 5.380000000000001e-07, "loss": 0.0028, "step": 47311 }, { "epoch": 2.522364983739404, "grad_norm": 2.736365556716919, "learning_rate": 5.378000000000001e-07, "loss": 0.017, "step": 47312 }, { "epoch": 2.522418297169057, "grad_norm": 1.4251675605773926, "learning_rate": 5.376e-07, "loss": 0.0081, "step": 47313 }, { "epoch": 2.5224716105987097, "grad_norm": 1.0440669059753418, "learning_rate": 5.374e-07, "loss": 0.0153, "step": 47314 }, { "epoch": 2.522524924028363, "grad_norm": 1.052337884902954, "learning_rate": 5.372e-07, "loss": 0.0037, "step": 47315 }, { "epoch": 2.5225782374580157, "grad_norm": 2.035454750061035, "learning_rate": 5.37e-07, "loss": 0.0154, "step": 47316 }, { "epoch": 2.5226315508876684, "grad_norm": 1.537280797958374, "learning_rate": 5.368e-07, "loss": 0.012, "step": 47317 }, { "epoch": 2.5226848643173216, "grad_norm": 1.8550492525100708, "learning_rate": 5.366000000000001e-07, "loss": 0.0049, "step": 47318 }, { "epoch": 2.5227381777469744, "grad_norm": 1.7068572044372559, "learning_rate": 5.364000000000001e-07, "loss": 0.0253, "step": 47319 }, { "epoch": 2.522791491176627, "grad_norm": 1.3057185411453247, "learning_rate": 5.362e-07, "loss": 0.0159, "step": 47320 }, { "epoch": 2.5228448046062804, "grad_norm": 1.9086247682571411, "learning_rate": 5.36e-07, "loss": 0.0261, "step": 47321 }, { "epoch": 2.522898118035933, "grad_norm": 1.1165821552276611, "learning_rate": 5.358e-07, "loss": 0.0048, "step": 47322 }, { "epoch": 2.5229514314655863, "grad_norm": 1.8429443836212158, "learning_rate": 5.356000000000001e-07, "loss": 0.0205, "step": 47323 }, { "epoch": 2.523004744895239, "grad_norm": 3.9193475246429443, "learning_rate": 5.354e-07, "loss": 0.0356, "step": 47324 }, { "epoch": 2.5230580583248923, "grad_norm": 1.3092864751815796, "learning_rate": 5.352e-07, "loss": 0.005, "step": 47325 }, { "epoch": 2.523111371754545, "grad_norm": 3.5574400424957275, "learning_rate": 5.350000000000001e-07, "loss": 0.0304, "step": 47326 }, { "epoch": 2.523164685184198, "grad_norm": 0.2368975728750229, "learning_rate": 5.348e-07, "loss": 0.0013, "step": 47327 }, { "epoch": 2.523217998613851, "grad_norm": 3.1451687812805176, "learning_rate": 5.346e-07, "loss": 0.0279, "step": 47328 }, { "epoch": 2.5232713120435037, "grad_norm": 2.811711549758911, "learning_rate": 5.344e-07, "loss": 0.0119, "step": 47329 }, { "epoch": 2.5233246254731565, "grad_norm": 0.6334101557731628, "learning_rate": 5.342000000000001e-07, "loss": 0.0036, "step": 47330 }, { "epoch": 2.5233779389028097, "grad_norm": 1.8523246049880981, "learning_rate": 5.340000000000001e-07, "loss": 0.0231, "step": 47331 }, { "epoch": 2.5234312523324625, "grad_norm": 3.250446081161499, "learning_rate": 5.338e-07, "loss": 0.0228, "step": 47332 }, { "epoch": 2.523484565762115, "grad_norm": 1.957196593284607, "learning_rate": 5.336000000000001e-07, "loss": 0.0124, "step": 47333 }, { "epoch": 2.5235378791917684, "grad_norm": 0.48177653551101685, "learning_rate": 5.334e-07, "loss": 0.0026, "step": 47334 }, { "epoch": 2.523591192621421, "grad_norm": 2.1403679847717285, "learning_rate": 5.332e-07, "loss": 0.0139, "step": 47335 }, { "epoch": 2.5236445060510744, "grad_norm": 0.7911403775215149, "learning_rate": 5.33e-07, "loss": 0.0042, "step": 47336 }, { "epoch": 2.523697819480727, "grad_norm": 1.0623674392700195, "learning_rate": 5.328e-07, "loss": 0.0201, "step": 47337 }, { "epoch": 2.5237511329103803, "grad_norm": 3.042226791381836, "learning_rate": 5.326000000000001e-07, "loss": 0.0049, "step": 47338 }, { "epoch": 2.523804446340033, "grad_norm": 0.7349934577941895, "learning_rate": 5.324000000000001e-07, "loss": 0.0044, "step": 47339 }, { "epoch": 2.523857759769686, "grad_norm": 0.8077946901321411, "learning_rate": 5.322000000000001e-07, "loss": 0.0033, "step": 47340 }, { "epoch": 2.523911073199339, "grad_norm": 0.8410366177558899, "learning_rate": 5.32e-07, "loss": 0.0035, "step": 47341 }, { "epoch": 2.523964386628992, "grad_norm": 0.6949256062507629, "learning_rate": 5.318e-07, "loss": 0.0035, "step": 47342 }, { "epoch": 2.5240177000586446, "grad_norm": 1.3105591535568237, "learning_rate": 5.316e-07, "loss": 0.0052, "step": 47343 }, { "epoch": 2.5240710134882978, "grad_norm": 2.0052452087402344, "learning_rate": 5.314e-07, "loss": 0.0187, "step": 47344 }, { "epoch": 2.5241243269179505, "grad_norm": 2.1187286376953125, "learning_rate": 5.312000000000001e-07, "loss": 0.0118, "step": 47345 }, { "epoch": 2.5241776403476037, "grad_norm": 1.8047268390655518, "learning_rate": 5.310000000000001e-07, "loss": 0.0178, "step": 47346 }, { "epoch": 2.5242309537772565, "grad_norm": 1.9887944459915161, "learning_rate": 5.308000000000001e-07, "loss": 0.0053, "step": 47347 }, { "epoch": 2.5242842672069097, "grad_norm": 2.374126434326172, "learning_rate": 5.306e-07, "loss": 0.0234, "step": 47348 }, { "epoch": 2.5243375806365624, "grad_norm": 3.5943212509155273, "learning_rate": 5.304e-07, "loss": 0.0142, "step": 47349 }, { "epoch": 2.524390894066215, "grad_norm": 1.2206546068191528, "learning_rate": 5.302e-07, "loss": 0.0057, "step": 47350 }, { "epoch": 2.5244442074958684, "grad_norm": 1.796204686164856, "learning_rate": 5.3e-07, "loss": 0.0195, "step": 47351 }, { "epoch": 2.524497520925521, "grad_norm": 4.287190914154053, "learning_rate": 5.298e-07, "loss": 0.0547, "step": 47352 }, { "epoch": 2.524550834355174, "grad_norm": 1.9457383155822754, "learning_rate": 5.296000000000001e-07, "loss": 0.0325, "step": 47353 }, { "epoch": 2.524604147784827, "grad_norm": 0.319132536649704, "learning_rate": 5.294000000000001e-07, "loss": 0.0021, "step": 47354 }, { "epoch": 2.52465746121448, "grad_norm": 1.914422869682312, "learning_rate": 5.292000000000001e-07, "loss": 0.0102, "step": 47355 }, { "epoch": 2.5247107746441326, "grad_norm": 2.307872772216797, "learning_rate": 5.29e-07, "loss": 0.0331, "step": 47356 }, { "epoch": 2.524764088073786, "grad_norm": 0.5134007334709167, "learning_rate": 5.288000000000001e-07, "loss": 0.0029, "step": 47357 }, { "epoch": 2.5248174015034386, "grad_norm": 1.6630676984786987, "learning_rate": 5.286e-07, "loss": 0.0062, "step": 47358 }, { "epoch": 2.524870714933092, "grad_norm": 2.0638492107391357, "learning_rate": 5.284e-07, "loss": 0.031, "step": 47359 }, { "epoch": 2.5249240283627445, "grad_norm": 0.6679086685180664, "learning_rate": 5.282000000000001e-07, "loss": 0.003, "step": 47360 }, { "epoch": 2.5249773417923977, "grad_norm": 1.8940006494522095, "learning_rate": 5.280000000000001e-07, "loss": 0.0078, "step": 47361 }, { "epoch": 2.5250306552220505, "grad_norm": 2.8509156703948975, "learning_rate": 5.278000000000001e-07, "loss": 0.0179, "step": 47362 }, { "epoch": 2.5250839686517033, "grad_norm": 0.5000818371772766, "learning_rate": 5.276e-07, "loss": 0.003, "step": 47363 }, { "epoch": 2.5251372820813565, "grad_norm": 2.290893077850342, "learning_rate": 5.274e-07, "loss": 0.0246, "step": 47364 }, { "epoch": 2.525190595511009, "grad_norm": 1.0649040937423706, "learning_rate": 5.272000000000001e-07, "loss": 0.0324, "step": 47365 }, { "epoch": 2.525243908940662, "grad_norm": 1.910025715827942, "learning_rate": 5.27e-07, "loss": 0.0222, "step": 47366 }, { "epoch": 2.525297222370315, "grad_norm": 2.1078615188598633, "learning_rate": 5.268e-07, "loss": 0.0072, "step": 47367 }, { "epoch": 2.525350535799968, "grad_norm": 2.5010111331939697, "learning_rate": 5.266000000000001e-07, "loss": 0.0116, "step": 47368 }, { "epoch": 2.5254038492296207, "grad_norm": 1.8392118215560913, "learning_rate": 5.264000000000001e-07, "loss": 0.0531, "step": 47369 }, { "epoch": 2.525457162659274, "grad_norm": 2.380131721496582, "learning_rate": 5.262e-07, "loss": 0.0124, "step": 47370 }, { "epoch": 2.5255104760889266, "grad_norm": 0.836277425289154, "learning_rate": 5.26e-07, "loss": 0.005, "step": 47371 }, { "epoch": 2.52556378951858, "grad_norm": 0.45317554473876953, "learning_rate": 5.258e-07, "loss": 0.0025, "step": 47372 }, { "epoch": 2.5256171029482326, "grad_norm": 0.8912420868873596, "learning_rate": 5.256000000000001e-07, "loss": 0.0036, "step": 47373 }, { "epoch": 2.525670416377886, "grad_norm": 0.6743747591972351, "learning_rate": 5.254000000000001e-07, "loss": 0.0037, "step": 47374 }, { "epoch": 2.5257237298075386, "grad_norm": 0.9550691843032837, "learning_rate": 5.252e-07, "loss": 0.0061, "step": 47375 }, { "epoch": 2.5257770432371913, "grad_norm": 1.1791324615478516, "learning_rate": 5.250000000000001e-07, "loss": 0.0038, "step": 47376 }, { "epoch": 2.5258303566668445, "grad_norm": 0.9014402031898499, "learning_rate": 5.248e-07, "loss": 0.0051, "step": 47377 }, { "epoch": 2.5258836700964973, "grad_norm": 2.857949733734131, "learning_rate": 5.246e-07, "loss": 0.0128, "step": 47378 }, { "epoch": 2.52593698352615, "grad_norm": 4.894914627075195, "learning_rate": 5.244e-07, "loss": 0.0574, "step": 47379 }, { "epoch": 2.5259902969558032, "grad_norm": 0.7124335765838623, "learning_rate": 5.242000000000001e-07, "loss": 0.0107, "step": 47380 }, { "epoch": 2.526043610385456, "grad_norm": 2.27983021736145, "learning_rate": 5.240000000000001e-07, "loss": 0.0235, "step": 47381 }, { "epoch": 2.5260969238151088, "grad_norm": 0.9555439352989197, "learning_rate": 5.238000000000001e-07, "loss": 0.004, "step": 47382 }, { "epoch": 2.526150237244762, "grad_norm": 0.9665311574935913, "learning_rate": 5.236000000000001e-07, "loss": 0.0155, "step": 47383 }, { "epoch": 2.526203550674415, "grad_norm": 1.666740894317627, "learning_rate": 5.234e-07, "loss": 0.0192, "step": 47384 }, { "epoch": 2.526256864104068, "grad_norm": 1.67937433719635, "learning_rate": 5.232e-07, "loss": 0.0134, "step": 47385 }, { "epoch": 2.5263101775337207, "grad_norm": 4.478055953979492, "learning_rate": 5.23e-07, "loss": 0.0261, "step": 47386 }, { "epoch": 2.526363490963374, "grad_norm": 4.680293560028076, "learning_rate": 5.228e-07, "loss": 0.0678, "step": 47387 }, { "epoch": 2.5264168043930266, "grad_norm": 0.3761564791202545, "learning_rate": 5.226000000000001e-07, "loss": 0.0024, "step": 47388 }, { "epoch": 2.5264701178226794, "grad_norm": 2.3626222610473633, "learning_rate": 5.224000000000001e-07, "loss": 0.0092, "step": 47389 }, { "epoch": 2.5265234312523326, "grad_norm": 1.3833895921707153, "learning_rate": 5.222000000000001e-07, "loss": 0.0047, "step": 47390 }, { "epoch": 2.5265767446819853, "grad_norm": 0.3653643727302551, "learning_rate": 5.22e-07, "loss": 0.0028, "step": 47391 }, { "epoch": 2.526630058111638, "grad_norm": 1.8452410697937012, "learning_rate": 5.218e-07, "loss": 0.0082, "step": 47392 }, { "epoch": 2.5266833715412913, "grad_norm": 0.34627822041511536, "learning_rate": 5.216e-07, "loss": 0.0018, "step": 47393 }, { "epoch": 2.526736684970944, "grad_norm": 2.8297910690307617, "learning_rate": 5.214e-07, "loss": 0.0398, "step": 47394 }, { "epoch": 2.5267899984005973, "grad_norm": 0.7957346439361572, "learning_rate": 5.212000000000001e-07, "loss": 0.0031, "step": 47395 }, { "epoch": 2.52684331183025, "grad_norm": 1.702621340751648, "learning_rate": 5.210000000000001e-07, "loss": 0.0067, "step": 47396 }, { "epoch": 2.526896625259903, "grad_norm": 0.4892084300518036, "learning_rate": 5.208000000000001e-07, "loss": 0.003, "step": 47397 }, { "epoch": 2.526949938689556, "grad_norm": 0.5160293579101562, "learning_rate": 5.206e-07, "loss": 0.0024, "step": 47398 }, { "epoch": 2.5270032521192087, "grad_norm": 2.954624652862549, "learning_rate": 5.204e-07, "loss": 0.0093, "step": 47399 }, { "epoch": 2.527056565548862, "grad_norm": 3.046109199523926, "learning_rate": 5.202e-07, "loss": 0.0344, "step": 47400 }, { "epoch": 2.5271098789785147, "grad_norm": 2.003899574279785, "learning_rate": 5.2e-07, "loss": 0.0052, "step": 47401 }, { "epoch": 2.5271631924081674, "grad_norm": 1.0663419961929321, "learning_rate": 5.198e-07, "loss": 0.0053, "step": 47402 }, { "epoch": 2.5272165058378206, "grad_norm": 4.823736190795898, "learning_rate": 5.196000000000001e-07, "loss": 0.0268, "step": 47403 }, { "epoch": 2.5272698192674734, "grad_norm": 3.318405866622925, "learning_rate": 5.194000000000001e-07, "loss": 0.0218, "step": 47404 }, { "epoch": 2.527323132697126, "grad_norm": 0.8677063584327698, "learning_rate": 5.192e-07, "loss": 0.0053, "step": 47405 }, { "epoch": 2.5273764461267794, "grad_norm": 0.7641687989234924, "learning_rate": 5.19e-07, "loss": 0.002, "step": 47406 }, { "epoch": 2.527429759556432, "grad_norm": 0.4761599600315094, "learning_rate": 5.188000000000001e-07, "loss": 0.0046, "step": 47407 }, { "epoch": 2.5274830729860853, "grad_norm": 3.1803205013275146, "learning_rate": 5.186000000000001e-07, "loss": 0.0278, "step": 47408 }, { "epoch": 2.527536386415738, "grad_norm": 1.4737019538879395, "learning_rate": 5.184e-07, "loss": 0.0047, "step": 47409 }, { "epoch": 2.5275896998453913, "grad_norm": 3.3817546367645264, "learning_rate": 5.182000000000001e-07, "loss": 0.0175, "step": 47410 }, { "epoch": 2.527643013275044, "grad_norm": 3.045983076095581, "learning_rate": 5.180000000000001e-07, "loss": 0.0234, "step": 47411 }, { "epoch": 2.527696326704697, "grad_norm": 1.4942986965179443, "learning_rate": 5.178e-07, "loss": 0.0082, "step": 47412 }, { "epoch": 2.52774964013435, "grad_norm": 1.290006160736084, "learning_rate": 5.176e-07, "loss": 0.0066, "step": 47413 }, { "epoch": 2.5278029535640028, "grad_norm": 2.1174826622009277, "learning_rate": 5.174e-07, "loss": 0.0118, "step": 47414 }, { "epoch": 2.5278562669936555, "grad_norm": 4.143703937530518, "learning_rate": 5.172000000000001e-07, "loss": 0.0203, "step": 47415 }, { "epoch": 2.5279095804233087, "grad_norm": 2.3037476539611816, "learning_rate": 5.170000000000001e-07, "loss": 0.0216, "step": 47416 }, { "epoch": 2.5279628938529615, "grad_norm": 2.2049198150634766, "learning_rate": 5.168e-07, "loss": 0.0389, "step": 47417 }, { "epoch": 2.5280162072826142, "grad_norm": 4.498595714569092, "learning_rate": 5.166000000000001e-07, "loss": 0.0497, "step": 47418 }, { "epoch": 2.5280695207122674, "grad_norm": 2.0643413066864014, "learning_rate": 5.164e-07, "loss": 0.0135, "step": 47419 }, { "epoch": 2.52812283414192, "grad_norm": 1.386599063873291, "learning_rate": 5.162e-07, "loss": 0.0179, "step": 47420 }, { "epoch": 2.5281761475715734, "grad_norm": 0.412423312664032, "learning_rate": 5.16e-07, "loss": 0.0027, "step": 47421 }, { "epoch": 2.528229461001226, "grad_norm": 1.2374382019042969, "learning_rate": 5.158e-07, "loss": 0.0145, "step": 47422 }, { "epoch": 2.5282827744308793, "grad_norm": 0.8955402374267578, "learning_rate": 5.156000000000001e-07, "loss": 0.003, "step": 47423 }, { "epoch": 2.528336087860532, "grad_norm": 2.4045870304107666, "learning_rate": 5.154000000000001e-07, "loss": 0.0189, "step": 47424 }, { "epoch": 2.528389401290185, "grad_norm": 2.5576424598693848, "learning_rate": 5.152000000000001e-07, "loss": 0.0119, "step": 47425 }, { "epoch": 2.528442714719838, "grad_norm": 3.689507484436035, "learning_rate": 5.15e-07, "loss": 0.0375, "step": 47426 }, { "epoch": 2.528496028149491, "grad_norm": 0.5933546423912048, "learning_rate": 5.148e-07, "loss": 0.003, "step": 47427 }, { "epoch": 2.5285493415791436, "grad_norm": 3.1713616847991943, "learning_rate": 5.146e-07, "loss": 0.0234, "step": 47428 }, { "epoch": 2.5286026550087968, "grad_norm": 3.7074480056762695, "learning_rate": 5.144e-07, "loss": 0.0397, "step": 47429 }, { "epoch": 2.5286559684384495, "grad_norm": 1.1521683931350708, "learning_rate": 5.142000000000001e-07, "loss": 0.0041, "step": 47430 }, { "epoch": 2.5287092818681027, "grad_norm": 0.8573311567306519, "learning_rate": 5.140000000000001e-07, "loss": 0.0031, "step": 47431 }, { "epoch": 2.5287625952977555, "grad_norm": 2.291943311691284, "learning_rate": 5.138000000000001e-07, "loss": 0.0217, "step": 47432 }, { "epoch": 2.5288159087274087, "grad_norm": 1.763201117515564, "learning_rate": 5.136e-07, "loss": 0.0074, "step": 47433 }, { "epoch": 2.5288692221570614, "grad_norm": 2.7618370056152344, "learning_rate": 5.134e-07, "loss": 0.0192, "step": 47434 }, { "epoch": 2.528922535586714, "grad_norm": 1.1538490056991577, "learning_rate": 5.132e-07, "loss": 0.0058, "step": 47435 }, { "epoch": 2.5289758490163674, "grad_norm": 1.164236068725586, "learning_rate": 5.13e-07, "loss": 0.0055, "step": 47436 }, { "epoch": 2.52902916244602, "grad_norm": 2.158386468887329, "learning_rate": 5.128e-07, "loss": 0.0308, "step": 47437 }, { "epoch": 2.529082475875673, "grad_norm": 2.220273494720459, "learning_rate": 5.126000000000001e-07, "loss": 0.0101, "step": 47438 }, { "epoch": 2.529135789305326, "grad_norm": 2.4530930519104004, "learning_rate": 5.124000000000001e-07, "loss": 0.0134, "step": 47439 }, { "epoch": 2.529189102734979, "grad_norm": 0.8415656685829163, "learning_rate": 5.122e-07, "loss": 0.0263, "step": 47440 }, { "epoch": 2.5292424161646316, "grad_norm": 2.9150002002716064, "learning_rate": 5.12e-07, "loss": 0.0137, "step": 47441 }, { "epoch": 2.529295729594285, "grad_norm": 1.4918431043624878, "learning_rate": 5.118000000000001e-07, "loss": 0.0133, "step": 47442 }, { "epoch": 2.5293490430239376, "grad_norm": 1.3893531560897827, "learning_rate": 5.116e-07, "loss": 0.0061, "step": 47443 }, { "epoch": 2.529402356453591, "grad_norm": 0.7334054112434387, "learning_rate": 5.114e-07, "loss": 0.0183, "step": 47444 }, { "epoch": 2.5294556698832436, "grad_norm": 1.0737885236740112, "learning_rate": 5.112000000000001e-07, "loss": 0.004, "step": 47445 }, { "epoch": 2.5295089833128968, "grad_norm": 1.2468489408493042, "learning_rate": 5.110000000000001e-07, "loss": 0.0125, "step": 47446 }, { "epoch": 2.5295622967425495, "grad_norm": 1.522523283958435, "learning_rate": 5.108e-07, "loss": 0.005, "step": 47447 }, { "epoch": 2.5296156101722023, "grad_norm": 3.7358193397521973, "learning_rate": 5.106e-07, "loss": 0.024, "step": 47448 }, { "epoch": 2.5296689236018555, "grad_norm": 0.47448843717575073, "learning_rate": 5.104e-07, "loss": 0.0021, "step": 47449 }, { "epoch": 2.5297222370315082, "grad_norm": 0.6477904319763184, "learning_rate": 5.102000000000001e-07, "loss": 0.0029, "step": 47450 }, { "epoch": 2.529775550461161, "grad_norm": 1.0857619047164917, "learning_rate": 5.1e-07, "loss": 0.0057, "step": 47451 }, { "epoch": 2.529828863890814, "grad_norm": 1.4188175201416016, "learning_rate": 5.098e-07, "loss": 0.0315, "step": 47452 }, { "epoch": 2.529882177320467, "grad_norm": 2.759662628173828, "learning_rate": 5.096000000000001e-07, "loss": 0.0278, "step": 47453 }, { "epoch": 2.5299354907501197, "grad_norm": 1.3455421924591064, "learning_rate": 5.094e-07, "loss": 0.0204, "step": 47454 }, { "epoch": 2.529988804179773, "grad_norm": 1.7112327814102173, "learning_rate": 5.092e-07, "loss": 0.0063, "step": 47455 }, { "epoch": 2.5300421176094257, "grad_norm": 2.5298073291778564, "learning_rate": 5.09e-07, "loss": 0.0252, "step": 47456 }, { "epoch": 2.530095431039079, "grad_norm": 0.3433935344219208, "learning_rate": 5.088000000000001e-07, "loss": 0.0024, "step": 47457 }, { "epoch": 2.5301487444687316, "grad_norm": 1.9976094961166382, "learning_rate": 5.086000000000001e-07, "loss": 0.0113, "step": 47458 }, { "epoch": 2.530202057898385, "grad_norm": 1.2100563049316406, "learning_rate": 5.084000000000001e-07, "loss": 0.0095, "step": 47459 }, { "epoch": 2.5302553713280376, "grad_norm": 1.3913027048110962, "learning_rate": 5.082000000000001e-07, "loss": 0.0067, "step": 47460 }, { "epoch": 2.5303086847576903, "grad_norm": 1.4195096492767334, "learning_rate": 5.08e-07, "loss": 0.0058, "step": 47461 }, { "epoch": 2.5303619981873435, "grad_norm": 0.3310295641422272, "learning_rate": 5.078e-07, "loss": 0.0025, "step": 47462 }, { "epoch": 2.5304153116169963, "grad_norm": 0.7396126389503479, "learning_rate": 5.076e-07, "loss": 0.0029, "step": 47463 }, { "epoch": 2.530468625046649, "grad_norm": 1.1640818119049072, "learning_rate": 5.074e-07, "loss": 0.0039, "step": 47464 }, { "epoch": 2.5305219384763022, "grad_norm": 1.7931488752365112, "learning_rate": 5.072000000000001e-07, "loss": 0.0111, "step": 47465 }, { "epoch": 2.530575251905955, "grad_norm": 4.010115146636963, "learning_rate": 5.070000000000001e-07, "loss": 0.0074, "step": 47466 }, { "epoch": 2.530628565335608, "grad_norm": 1.1986746788024902, "learning_rate": 5.068000000000001e-07, "loss": 0.0063, "step": 47467 }, { "epoch": 2.530681878765261, "grad_norm": 3.7030460834503174, "learning_rate": 5.066e-07, "loss": 0.0065, "step": 47468 }, { "epoch": 2.530735192194914, "grad_norm": 0.4310397207736969, "learning_rate": 5.064e-07, "loss": 0.0036, "step": 47469 }, { "epoch": 2.530788505624567, "grad_norm": 2.6750741004943848, "learning_rate": 5.062e-07, "loss": 0.0077, "step": 47470 }, { "epoch": 2.5308418190542197, "grad_norm": 0.49867331981658936, "learning_rate": 5.06e-07, "loss": 0.0023, "step": 47471 }, { "epoch": 2.530895132483873, "grad_norm": 1.0881472826004028, "learning_rate": 5.058e-07, "loss": 0.0049, "step": 47472 }, { "epoch": 2.5309484459135256, "grad_norm": 0.6532757878303528, "learning_rate": 5.056000000000001e-07, "loss": 0.0042, "step": 47473 }, { "epoch": 2.5310017593431784, "grad_norm": 1.755696415901184, "learning_rate": 5.054000000000001e-07, "loss": 0.0163, "step": 47474 }, { "epoch": 2.5310550727728316, "grad_norm": 2.170050621032715, "learning_rate": 5.052e-07, "loss": 0.0226, "step": 47475 }, { "epoch": 2.5311083862024844, "grad_norm": 2.2440593242645264, "learning_rate": 5.05e-07, "loss": 0.0077, "step": 47476 }, { "epoch": 2.531161699632137, "grad_norm": 1.6705257892608643, "learning_rate": 5.048e-07, "loss": 0.0311, "step": 47477 }, { "epoch": 2.5312150130617903, "grad_norm": 2.1070945262908936, "learning_rate": 5.046e-07, "loss": 0.0071, "step": 47478 }, { "epoch": 2.531268326491443, "grad_norm": 2.659654140472412, "learning_rate": 5.044e-07, "loss": 0.0373, "step": 47479 }, { "epoch": 2.5313216399210963, "grad_norm": 0.9863331913948059, "learning_rate": 5.042000000000001e-07, "loss": 0.0217, "step": 47480 }, { "epoch": 2.531374953350749, "grad_norm": 0.8613151907920837, "learning_rate": 5.040000000000001e-07, "loss": 0.0033, "step": 47481 }, { "epoch": 2.5314282667804022, "grad_norm": 2.65708065032959, "learning_rate": 5.038e-07, "loss": 0.0357, "step": 47482 }, { "epoch": 2.531481580210055, "grad_norm": 0.8627607226371765, "learning_rate": 5.036e-07, "loss": 0.0181, "step": 47483 }, { "epoch": 2.5315348936397077, "grad_norm": 1.1923754215240479, "learning_rate": 5.034e-07, "loss": 0.0134, "step": 47484 }, { "epoch": 2.531588207069361, "grad_norm": 3.811929225921631, "learning_rate": 5.032e-07, "loss": 0.0367, "step": 47485 }, { "epoch": 2.5316415204990137, "grad_norm": 4.949131488800049, "learning_rate": 5.03e-07, "loss": 0.034, "step": 47486 }, { "epoch": 2.5316948339286665, "grad_norm": 1.9268275499343872, "learning_rate": 5.028e-07, "loss": 0.0341, "step": 47487 }, { "epoch": 2.5317481473583197, "grad_norm": 1.0957597494125366, "learning_rate": 5.026000000000001e-07, "loss": 0.021, "step": 47488 }, { "epoch": 2.5318014607879724, "grad_norm": 0.38066864013671875, "learning_rate": 5.024e-07, "loss": 0.0026, "step": 47489 }, { "epoch": 2.531854774217625, "grad_norm": 4.787647247314453, "learning_rate": 5.022e-07, "loss": 0.0082, "step": 47490 }, { "epoch": 2.5319080876472784, "grad_norm": 0.5703204870223999, "learning_rate": 5.02e-07, "loss": 0.0035, "step": 47491 }, { "epoch": 2.531961401076931, "grad_norm": 0.9198541045188904, "learning_rate": 5.018000000000001e-07, "loss": 0.0051, "step": 47492 }, { "epoch": 2.5320147145065843, "grad_norm": 2.1455156803131104, "learning_rate": 5.016000000000001e-07, "loss": 0.0085, "step": 47493 }, { "epoch": 2.532068027936237, "grad_norm": 0.8775084018707275, "learning_rate": 5.014e-07, "loss": 0.0145, "step": 47494 }, { "epoch": 2.5321213413658903, "grad_norm": 3.2470967769622803, "learning_rate": 5.012000000000001e-07, "loss": 0.0128, "step": 47495 }, { "epoch": 2.532174654795543, "grad_norm": 1.2404382228851318, "learning_rate": 5.01e-07, "loss": 0.0048, "step": 47496 }, { "epoch": 2.532227968225196, "grad_norm": 2.500920295715332, "learning_rate": 5.008e-07, "loss": 0.0172, "step": 47497 }, { "epoch": 2.532281281654849, "grad_norm": 0.39866286516189575, "learning_rate": 5.006e-07, "loss": 0.0021, "step": 47498 }, { "epoch": 2.5323345950845018, "grad_norm": 2.3437752723693848, "learning_rate": 5.004e-07, "loss": 0.008, "step": 47499 }, { "epoch": 2.5323879085141545, "grad_norm": 2.092822313308716, "learning_rate": 5.002000000000001e-07, "loss": 0.0076, "step": 47500 }, { "epoch": 2.5324412219438077, "grad_norm": 2.2345011234283447, "learning_rate": 5.000000000000001e-07, "loss": 0.0251, "step": 47501 }, { "epoch": 2.5324945353734605, "grad_norm": 2.1473939418792725, "learning_rate": 4.998e-07, "loss": 0.0235, "step": 47502 }, { "epoch": 2.5325478488031132, "grad_norm": 1.2762808799743652, "learning_rate": 4.996000000000001e-07, "loss": 0.0189, "step": 47503 }, { "epoch": 2.5326011622327664, "grad_norm": 1.0450488328933716, "learning_rate": 4.994e-07, "loss": 0.0043, "step": 47504 }, { "epoch": 2.5326544756624196, "grad_norm": 3.345799207687378, "learning_rate": 4.992e-07, "loss": 0.0148, "step": 47505 }, { "epoch": 2.5327077890920724, "grad_norm": 1.5578210353851318, "learning_rate": 4.99e-07, "loss": 0.0059, "step": 47506 }, { "epoch": 2.532761102521725, "grad_norm": 1.6571203470230103, "learning_rate": 4.988000000000001e-07, "loss": 0.0045, "step": 47507 }, { "epoch": 2.5328144159513784, "grad_norm": 3.4416651725769043, "learning_rate": 4.986000000000001e-07, "loss": 0.0374, "step": 47508 }, { "epoch": 2.532867729381031, "grad_norm": 1.7577769756317139, "learning_rate": 4.984000000000001e-07, "loss": 0.0043, "step": 47509 }, { "epoch": 2.532921042810684, "grad_norm": 2.456150770187378, "learning_rate": 4.982000000000001e-07, "loss": 0.0051, "step": 47510 }, { "epoch": 2.532974356240337, "grad_norm": 1.9797375202178955, "learning_rate": 4.98e-07, "loss": 0.007, "step": 47511 }, { "epoch": 2.53302766966999, "grad_norm": 3.2635321617126465, "learning_rate": 4.978e-07, "loss": 0.031, "step": 47512 }, { "epoch": 2.5330809830996426, "grad_norm": 3.725837469100952, "learning_rate": 4.976e-07, "loss": 0.0169, "step": 47513 }, { "epoch": 2.533134296529296, "grad_norm": 0.2702386677265167, "learning_rate": 4.974e-07, "loss": 0.0021, "step": 47514 }, { "epoch": 2.5331876099589485, "grad_norm": 0.7109014987945557, "learning_rate": 4.972000000000001e-07, "loss": 0.0233, "step": 47515 }, { "epoch": 2.5332409233886017, "grad_norm": 1.2709306478500366, "learning_rate": 4.970000000000001e-07, "loss": 0.004, "step": 47516 }, { "epoch": 2.5332942368182545, "grad_norm": 1.5384411811828613, "learning_rate": 4.968000000000001e-07, "loss": 0.0226, "step": 47517 }, { "epoch": 2.5333475502479077, "grad_norm": 0.5801690220832825, "learning_rate": 4.966e-07, "loss": 0.0021, "step": 47518 }, { "epoch": 2.5334008636775605, "grad_norm": 1.7395604848861694, "learning_rate": 4.964e-07, "loss": 0.0113, "step": 47519 }, { "epoch": 2.533454177107213, "grad_norm": 1.2663146257400513, "learning_rate": 4.962e-07, "loss": 0.003, "step": 47520 }, { "epoch": 2.5335074905368664, "grad_norm": 1.3579201698303223, "learning_rate": 4.96e-07, "loss": 0.014, "step": 47521 }, { "epoch": 2.533560803966519, "grad_norm": 1.5073184967041016, "learning_rate": 4.958e-07, "loss": 0.0033, "step": 47522 }, { "epoch": 2.533614117396172, "grad_norm": 3.9286460876464844, "learning_rate": 4.956000000000001e-07, "loss": 0.0268, "step": 47523 }, { "epoch": 2.533667430825825, "grad_norm": 1.8488538265228271, "learning_rate": 4.954000000000001e-07, "loss": 0.0215, "step": 47524 }, { "epoch": 2.533720744255478, "grad_norm": 1.8618032932281494, "learning_rate": 4.952e-07, "loss": 0.0102, "step": 47525 }, { "epoch": 2.5337740576851306, "grad_norm": 5.437044143676758, "learning_rate": 4.95e-07, "loss": 0.0546, "step": 47526 }, { "epoch": 2.533827371114784, "grad_norm": 2.9805569648742676, "learning_rate": 4.948000000000001e-07, "loss": 0.0348, "step": 47527 }, { "epoch": 2.5338806845444366, "grad_norm": 1.115978479385376, "learning_rate": 4.946e-07, "loss": 0.0169, "step": 47528 }, { "epoch": 2.53393399797409, "grad_norm": 1.3317795991897583, "learning_rate": 4.944e-07, "loss": 0.004, "step": 47529 }, { "epoch": 2.5339873114037426, "grad_norm": 1.4386858940124512, "learning_rate": 4.942000000000001e-07, "loss": 0.014, "step": 47530 }, { "epoch": 2.5340406248333958, "grad_norm": 3.918642997741699, "learning_rate": 4.940000000000001e-07, "loss": 0.0091, "step": 47531 }, { "epoch": 2.5340939382630485, "grad_norm": 1.8125945329666138, "learning_rate": 4.938e-07, "loss": 0.012, "step": 47532 }, { "epoch": 2.5341472516927013, "grad_norm": 1.4608412981033325, "learning_rate": 4.936e-07, "loss": 0.0074, "step": 47533 }, { "epoch": 2.5342005651223545, "grad_norm": 5.065800189971924, "learning_rate": 4.934e-07, "loss": 0.0114, "step": 47534 }, { "epoch": 2.5342538785520072, "grad_norm": 1.6692427396774292, "learning_rate": 4.932000000000001e-07, "loss": 0.0295, "step": 47535 }, { "epoch": 2.53430719198166, "grad_norm": 2.9477734565734863, "learning_rate": 4.93e-07, "loss": 0.0064, "step": 47536 }, { "epoch": 2.534360505411313, "grad_norm": 2.2578513622283936, "learning_rate": 4.928e-07, "loss": 0.0148, "step": 47537 }, { "epoch": 2.534413818840966, "grad_norm": 1.5755566358566284, "learning_rate": 4.926000000000001e-07, "loss": 0.0069, "step": 47538 }, { "epoch": 2.5344671322706187, "grad_norm": 2.359954833984375, "learning_rate": 4.924e-07, "loss": 0.018, "step": 47539 }, { "epoch": 2.534520445700272, "grad_norm": 1.6046656370162964, "learning_rate": 4.922e-07, "loss": 0.0511, "step": 47540 }, { "epoch": 2.5345737591299247, "grad_norm": 3.339003086090088, "learning_rate": 4.92e-07, "loss": 0.0498, "step": 47541 }, { "epoch": 2.534627072559578, "grad_norm": 3.0677175521850586, "learning_rate": 4.918000000000001e-07, "loss": 0.0162, "step": 47542 }, { "epoch": 2.5346803859892306, "grad_norm": 4.497359275817871, "learning_rate": 4.916000000000001e-07, "loss": 0.0292, "step": 47543 }, { "epoch": 2.534733699418884, "grad_norm": 1.7251423597335815, "learning_rate": 4.914000000000001e-07, "loss": 0.0127, "step": 47544 }, { "epoch": 2.5347870128485366, "grad_norm": 1.1698884963989258, "learning_rate": 4.912000000000001e-07, "loss": 0.0054, "step": 47545 }, { "epoch": 2.5348403262781893, "grad_norm": 1.5497605800628662, "learning_rate": 4.91e-07, "loss": 0.0187, "step": 47546 }, { "epoch": 2.5348936397078425, "grad_norm": 1.7729156017303467, "learning_rate": 4.908e-07, "loss": 0.026, "step": 47547 }, { "epoch": 2.5349469531374953, "grad_norm": 2.070100784301758, "learning_rate": 4.906e-07, "loss": 0.0093, "step": 47548 }, { "epoch": 2.535000266567148, "grad_norm": 4.039140224456787, "learning_rate": 4.904e-07, "loss": 0.0343, "step": 47549 }, { "epoch": 2.5350535799968013, "grad_norm": 0.738798975944519, "learning_rate": 4.902000000000001e-07, "loss": 0.0039, "step": 47550 }, { "epoch": 2.535106893426454, "grad_norm": 3.267831802368164, "learning_rate": 4.900000000000001e-07, "loss": 0.0118, "step": 47551 }, { "epoch": 2.535160206856107, "grad_norm": 2.644399881362915, "learning_rate": 4.898000000000001e-07, "loss": 0.0059, "step": 47552 }, { "epoch": 2.53521352028576, "grad_norm": 0.7906655669212341, "learning_rate": 4.896e-07, "loss": 0.0035, "step": 47553 }, { "epoch": 2.535266833715413, "grad_norm": 1.9927321672439575, "learning_rate": 4.894e-07, "loss": 0.0089, "step": 47554 }, { "epoch": 2.535320147145066, "grad_norm": 5.652426719665527, "learning_rate": 4.892e-07, "loss": 0.0223, "step": 47555 }, { "epoch": 2.5353734605747187, "grad_norm": 2.807961940765381, "learning_rate": 4.89e-07, "loss": 0.0067, "step": 47556 }, { "epoch": 2.535426774004372, "grad_norm": 3.0754804611206055, "learning_rate": 4.888000000000001e-07, "loss": 0.0294, "step": 47557 }, { "epoch": 2.5354800874340246, "grad_norm": 0.7383769750595093, "learning_rate": 4.886000000000001e-07, "loss": 0.0035, "step": 47558 }, { "epoch": 2.5355334008636774, "grad_norm": 4.310047626495361, "learning_rate": 4.884000000000001e-07, "loss": 0.0076, "step": 47559 }, { "epoch": 2.5355867142933306, "grad_norm": 1.4247790575027466, "learning_rate": 4.882e-07, "loss": 0.0047, "step": 47560 }, { "epoch": 2.5356400277229834, "grad_norm": 1.1490862369537354, "learning_rate": 4.88e-07, "loss": 0.0075, "step": 47561 }, { "epoch": 2.535693341152636, "grad_norm": 2.686654567718506, "learning_rate": 4.878e-07, "loss": 0.0067, "step": 47562 }, { "epoch": 2.5357466545822893, "grad_norm": 2.746338129043579, "learning_rate": 4.876e-07, "loss": 0.0285, "step": 47563 }, { "epoch": 2.535799968011942, "grad_norm": 2.03048038482666, "learning_rate": 4.874e-07, "loss": 0.0182, "step": 47564 }, { "epoch": 2.5358532814415953, "grad_norm": 5.298123836517334, "learning_rate": 4.872000000000001e-07, "loss": 0.0536, "step": 47565 }, { "epoch": 2.535906594871248, "grad_norm": 4.00961971282959, "learning_rate": 4.870000000000001e-07, "loss": 0.0529, "step": 47566 }, { "epoch": 2.5359599083009012, "grad_norm": 2.175311803817749, "learning_rate": 4.868e-07, "loss": 0.0106, "step": 47567 }, { "epoch": 2.536013221730554, "grad_norm": 2.879390239715576, "learning_rate": 4.866e-07, "loss": 0.0299, "step": 47568 }, { "epoch": 2.5360665351602067, "grad_norm": 1.2126929759979248, "learning_rate": 4.864e-07, "loss": 0.0056, "step": 47569 }, { "epoch": 2.53611984858986, "grad_norm": 1.8801935911178589, "learning_rate": 4.862e-07, "loss": 0.0066, "step": 47570 }, { "epoch": 2.5361731620195127, "grad_norm": 2.331650495529175, "learning_rate": 4.86e-07, "loss": 0.0097, "step": 47571 }, { "epoch": 2.5362264754491655, "grad_norm": 1.429396629333496, "learning_rate": 4.858e-07, "loss": 0.0102, "step": 47572 }, { "epoch": 2.5362797888788187, "grad_norm": 1.9665093421936035, "learning_rate": 4.856000000000001e-07, "loss": 0.0189, "step": 47573 }, { "epoch": 2.5363331023084714, "grad_norm": 3.333664894104004, "learning_rate": 4.854e-07, "loss": 0.0093, "step": 47574 }, { "epoch": 2.536386415738124, "grad_norm": 0.9065684676170349, "learning_rate": 4.852e-07, "loss": 0.0054, "step": 47575 }, { "epoch": 2.5364397291677774, "grad_norm": 1.1310396194458008, "learning_rate": 4.85e-07, "loss": 0.0146, "step": 47576 }, { "epoch": 2.53649304259743, "grad_norm": 3.8359951972961426, "learning_rate": 4.848000000000001e-07, "loss": 0.0282, "step": 47577 }, { "epoch": 2.5365463560270833, "grad_norm": 2.5196492671966553, "learning_rate": 4.846000000000001e-07, "loss": 0.0136, "step": 47578 }, { "epoch": 2.536599669456736, "grad_norm": 0.9193636178970337, "learning_rate": 4.844e-07, "loss": 0.0041, "step": 47579 }, { "epoch": 2.5366529828863893, "grad_norm": 2.4216487407684326, "learning_rate": 4.842000000000001e-07, "loss": 0.008, "step": 47580 }, { "epoch": 2.536706296316042, "grad_norm": 2.070155382156372, "learning_rate": 4.84e-07, "loss": 0.0099, "step": 47581 }, { "epoch": 2.536759609745695, "grad_norm": 2.3529748916625977, "learning_rate": 4.838e-07, "loss": 0.0154, "step": 47582 }, { "epoch": 2.536812923175348, "grad_norm": 2.128011703491211, "learning_rate": 4.836e-07, "loss": 0.0167, "step": 47583 }, { "epoch": 2.5368662366050008, "grad_norm": 2.2894697189331055, "learning_rate": 4.834e-07, "loss": 0.021, "step": 47584 }, { "epoch": 2.5369195500346535, "grad_norm": 2.095256805419922, "learning_rate": 4.832000000000001e-07, "loss": 0.0058, "step": 47585 }, { "epoch": 2.5369728634643067, "grad_norm": 0.659389078617096, "learning_rate": 4.830000000000001e-07, "loss": 0.0039, "step": 47586 }, { "epoch": 2.5370261768939595, "grad_norm": 1.8147213459014893, "learning_rate": 4.828000000000001e-07, "loss": 0.0119, "step": 47587 }, { "epoch": 2.5370794903236122, "grad_norm": 1.902220606803894, "learning_rate": 4.826e-07, "loss": 0.0171, "step": 47588 }, { "epoch": 2.5371328037532654, "grad_norm": 1.5998581647872925, "learning_rate": 4.824e-07, "loss": 0.0053, "step": 47589 }, { "epoch": 2.5371861171829186, "grad_norm": 0.4577929377555847, "learning_rate": 4.822e-07, "loss": 0.0034, "step": 47590 }, { "epoch": 2.5372394306125714, "grad_norm": 2.915519952774048, "learning_rate": 4.82e-07, "loss": 0.0104, "step": 47591 }, { "epoch": 2.537292744042224, "grad_norm": 1.5852534770965576, "learning_rate": 4.818000000000001e-07, "loss": 0.0109, "step": 47592 }, { "epoch": 2.5373460574718774, "grad_norm": 4.363394260406494, "learning_rate": 4.816000000000001e-07, "loss": 0.0267, "step": 47593 }, { "epoch": 2.53739937090153, "grad_norm": 2.282017707824707, "learning_rate": 4.814000000000001e-07, "loss": 0.034, "step": 47594 }, { "epoch": 2.537452684331183, "grad_norm": 1.4517656564712524, "learning_rate": 4.812e-07, "loss": 0.009, "step": 47595 }, { "epoch": 2.537505997760836, "grad_norm": 3.1360747814178467, "learning_rate": 4.81e-07, "loss": 0.0287, "step": 47596 }, { "epoch": 2.537559311190489, "grad_norm": 2.182758092880249, "learning_rate": 4.808e-07, "loss": 0.0298, "step": 47597 }, { "epoch": 2.5376126246201416, "grad_norm": 2.171971321105957, "learning_rate": 4.806e-07, "loss": 0.0166, "step": 47598 }, { "epoch": 2.537665938049795, "grad_norm": 0.8039624094963074, "learning_rate": 4.804e-07, "loss": 0.0037, "step": 47599 }, { "epoch": 2.5377192514794475, "grad_norm": 2.0686347484588623, "learning_rate": 4.802000000000001e-07, "loss": 0.0067, "step": 47600 }, { "epoch": 2.5377725649091007, "grad_norm": 1.1210016012191772, "learning_rate": 4.800000000000001e-07, "loss": 0.0235, "step": 47601 }, { "epoch": 2.5378258783387535, "grad_norm": 2.581900119781494, "learning_rate": 4.798e-07, "loss": 0.0317, "step": 47602 }, { "epoch": 2.5378791917684067, "grad_norm": 1.4956235885620117, "learning_rate": 4.796e-07, "loss": 0.006, "step": 47603 }, { "epoch": 2.5379325051980595, "grad_norm": 2.454921245574951, "learning_rate": 4.794000000000001e-07, "loss": 0.013, "step": 47604 }, { "epoch": 2.537985818627712, "grad_norm": 1.5535818338394165, "learning_rate": 4.792e-07, "loss": 0.0095, "step": 47605 }, { "epoch": 2.5380391320573654, "grad_norm": 2.230985403060913, "learning_rate": 4.79e-07, "loss": 0.02, "step": 47606 }, { "epoch": 2.538092445487018, "grad_norm": 2.5754079818725586, "learning_rate": 4.788000000000001e-07, "loss": 0.012, "step": 47607 }, { "epoch": 2.538145758916671, "grad_norm": 2.705206871032715, "learning_rate": 4.786000000000001e-07, "loss": 0.0157, "step": 47608 }, { "epoch": 2.538199072346324, "grad_norm": 2.8224849700927734, "learning_rate": 4.784e-07, "loss": 0.022, "step": 47609 }, { "epoch": 2.538252385775977, "grad_norm": 0.16341380774974823, "learning_rate": 4.782e-07, "loss": 0.0013, "step": 47610 }, { "epoch": 2.5383056992056297, "grad_norm": 1.1931298971176147, "learning_rate": 4.78e-07, "loss": 0.0039, "step": 47611 }, { "epoch": 2.538359012635283, "grad_norm": 0.6600654125213623, "learning_rate": 4.778000000000001e-07, "loss": 0.0031, "step": 47612 }, { "epoch": 2.5384123260649356, "grad_norm": 2.9185283184051514, "learning_rate": 4.776e-07, "loss": 0.0094, "step": 47613 }, { "epoch": 2.538465639494589, "grad_norm": 0.9778263568878174, "learning_rate": 4.774e-07, "loss": 0.0049, "step": 47614 }, { "epoch": 2.5385189529242416, "grad_norm": 0.7356770634651184, "learning_rate": 4.772000000000001e-07, "loss": 0.0032, "step": 47615 }, { "epoch": 2.5385722663538948, "grad_norm": 0.9706537127494812, "learning_rate": 4.77e-07, "loss": 0.0041, "step": 47616 }, { "epoch": 2.5386255797835475, "grad_norm": 0.18814992904663086, "learning_rate": 4.7680000000000003e-07, "loss": 0.0018, "step": 47617 }, { "epoch": 2.5386788932132003, "grad_norm": 1.6792128086090088, "learning_rate": 4.7660000000000007e-07, "loss": 0.0073, "step": 47618 }, { "epoch": 2.5387322066428535, "grad_norm": 3.2601444721221924, "learning_rate": 4.7640000000000006e-07, "loss": 0.0241, "step": 47619 }, { "epoch": 2.5387855200725062, "grad_norm": 3.816331148147583, "learning_rate": 4.7620000000000004e-07, "loss": 0.0164, "step": 47620 }, { "epoch": 2.538838833502159, "grad_norm": 2.636916160583496, "learning_rate": 4.760000000000001e-07, "loss": 0.0181, "step": 47621 }, { "epoch": 2.538892146931812, "grad_norm": 1.6813123226165771, "learning_rate": 4.758e-07, "loss": 0.0098, "step": 47622 }, { "epoch": 2.538945460361465, "grad_norm": 1.757585048675537, "learning_rate": 4.756e-07, "loss": 0.0264, "step": 47623 }, { "epoch": 2.5389987737911177, "grad_norm": 1.312408208847046, "learning_rate": 4.7540000000000004e-07, "loss": 0.018, "step": 47624 }, { "epoch": 2.539052087220771, "grad_norm": 0.3489212393760681, "learning_rate": 4.752e-07, "loss": 0.0028, "step": 47625 }, { "epoch": 2.539105400650424, "grad_norm": 2.881181240081787, "learning_rate": 4.7500000000000006e-07, "loss": 0.0206, "step": 47626 }, { "epoch": 2.539158714080077, "grad_norm": 1.9981048107147217, "learning_rate": 4.7480000000000004e-07, "loss": 0.025, "step": 47627 }, { "epoch": 2.5392120275097296, "grad_norm": 1.228882074356079, "learning_rate": 4.746000000000001e-07, "loss": 0.004, "step": 47628 }, { "epoch": 2.539265340939383, "grad_norm": 1.194770336151123, "learning_rate": 4.7440000000000007e-07, "loss": 0.0044, "step": 47629 }, { "epoch": 2.5393186543690356, "grad_norm": 1.997272253036499, "learning_rate": 4.742e-07, "loss": 0.0054, "step": 47630 }, { "epoch": 2.5393719677986883, "grad_norm": 3.2821481227874756, "learning_rate": 4.7400000000000004e-07, "loss": 0.0193, "step": 47631 }, { "epoch": 2.5394252812283415, "grad_norm": 0.9688006043434143, "learning_rate": 4.738e-07, "loss": 0.0052, "step": 47632 }, { "epoch": 2.5394785946579943, "grad_norm": 2.1854517459869385, "learning_rate": 4.736e-07, "loss": 0.0205, "step": 47633 }, { "epoch": 2.539531908087647, "grad_norm": 1.7602821588516235, "learning_rate": 4.7340000000000004e-07, "loss": 0.0228, "step": 47634 }, { "epoch": 2.5395852215173003, "grad_norm": 1.634019136428833, "learning_rate": 4.7320000000000003e-07, "loss": 0.011, "step": 47635 }, { "epoch": 2.539638534946953, "grad_norm": 4.110219478607178, "learning_rate": 4.7300000000000007e-07, "loss": 0.0152, "step": 47636 }, { "epoch": 2.539691848376606, "grad_norm": 2.38834285736084, "learning_rate": 4.7280000000000005e-07, "loss": 0.0219, "step": 47637 }, { "epoch": 2.539745161806259, "grad_norm": 0.8730568885803223, "learning_rate": 4.726000000000001e-07, "loss": 0.0039, "step": 47638 }, { "epoch": 2.539798475235912, "grad_norm": 1.0095257759094238, "learning_rate": 4.724e-07, "loss": 0.005, "step": 47639 }, { "epoch": 2.539851788665565, "grad_norm": 3.3133866786956787, "learning_rate": 4.722e-07, "loss": 0.0175, "step": 47640 }, { "epoch": 2.5399051020952177, "grad_norm": 2.0284388065338135, "learning_rate": 4.7200000000000004e-07, "loss": 0.007, "step": 47641 }, { "epoch": 2.539958415524871, "grad_norm": 0.5559196472167969, "learning_rate": 4.7180000000000003e-07, "loss": 0.0019, "step": 47642 }, { "epoch": 2.5400117289545237, "grad_norm": 4.405308246612549, "learning_rate": 4.7160000000000007e-07, "loss": 0.0426, "step": 47643 }, { "epoch": 2.5400650423841764, "grad_norm": 2.857553720474243, "learning_rate": 4.7140000000000005e-07, "loss": 0.025, "step": 47644 }, { "epoch": 2.5401183558138296, "grad_norm": 2.4297935962677, "learning_rate": 4.7120000000000004e-07, "loss": 0.0108, "step": 47645 }, { "epoch": 2.5401716692434824, "grad_norm": 1.7586326599121094, "learning_rate": 4.710000000000001e-07, "loss": 0.0066, "step": 47646 }, { "epoch": 2.540224982673135, "grad_norm": 1.7466354370117188, "learning_rate": 4.708e-07, "loss": 0.0178, "step": 47647 }, { "epoch": 2.5402782961027883, "grad_norm": 5.488219738006592, "learning_rate": 4.706e-07, "loss": 0.0494, "step": 47648 }, { "epoch": 2.540331609532441, "grad_norm": 0.1019759550690651, "learning_rate": 4.7040000000000003e-07, "loss": 0.0011, "step": 47649 }, { "epoch": 2.5403849229620943, "grad_norm": 1.609941840171814, "learning_rate": 4.702e-07, "loss": 0.0265, "step": 47650 }, { "epoch": 2.540438236391747, "grad_norm": 0.8397917747497559, "learning_rate": 4.7000000000000005e-07, "loss": 0.0029, "step": 47651 }, { "epoch": 2.5404915498214002, "grad_norm": 1.363777756690979, "learning_rate": 4.6980000000000004e-07, "loss": 0.0178, "step": 47652 }, { "epoch": 2.540544863251053, "grad_norm": 2.93900990486145, "learning_rate": 4.696000000000001e-07, "loss": 0.0302, "step": 47653 }, { "epoch": 2.5405981766807058, "grad_norm": 1.5105023384094238, "learning_rate": 4.6940000000000006e-07, "loss": 0.0362, "step": 47654 }, { "epoch": 2.540651490110359, "grad_norm": 0.46657103300094604, "learning_rate": 4.6920000000000005e-07, "loss": 0.003, "step": 47655 }, { "epoch": 2.5407048035400117, "grad_norm": 2.988450527191162, "learning_rate": 4.6900000000000003e-07, "loss": 0.0284, "step": 47656 }, { "epoch": 2.5407581169696645, "grad_norm": 1.4634490013122559, "learning_rate": 4.688e-07, "loss": 0.0155, "step": 47657 }, { "epoch": 2.5408114303993177, "grad_norm": 3.6033129692077637, "learning_rate": 4.6860000000000005e-07, "loss": 0.0187, "step": 47658 }, { "epoch": 2.5408647438289704, "grad_norm": 2.0279598236083984, "learning_rate": 4.6840000000000004e-07, "loss": 0.01, "step": 47659 }, { "epoch": 2.540918057258623, "grad_norm": 3.56797194480896, "learning_rate": 4.682e-07, "loss": 0.0244, "step": 47660 }, { "epoch": 2.5409713706882764, "grad_norm": 2.9061450958251953, "learning_rate": 4.6800000000000006e-07, "loss": 0.0053, "step": 47661 }, { "epoch": 2.541024684117929, "grad_norm": 2.1969780921936035, "learning_rate": 4.6780000000000005e-07, "loss": 0.0266, "step": 47662 }, { "epoch": 2.5410779975475823, "grad_norm": 1.275802731513977, "learning_rate": 4.676000000000001e-07, "loss": 0.0054, "step": 47663 }, { "epoch": 2.541131310977235, "grad_norm": 2.122405767440796, "learning_rate": 4.674e-07, "loss": 0.0066, "step": 47664 }, { "epoch": 2.5411846244068883, "grad_norm": 3.022303342819214, "learning_rate": 4.672e-07, "loss": 0.0126, "step": 47665 }, { "epoch": 2.541237937836541, "grad_norm": 1.8856242895126343, "learning_rate": 4.6700000000000004e-07, "loss": 0.0068, "step": 47666 }, { "epoch": 2.541291251266194, "grad_norm": 1.47051203250885, "learning_rate": 4.668e-07, "loss": 0.0085, "step": 47667 }, { "epoch": 2.541344564695847, "grad_norm": 4.683034896850586, "learning_rate": 4.6660000000000006e-07, "loss": 0.0178, "step": 47668 }, { "epoch": 2.5413978781254998, "grad_norm": 1.5687518119812012, "learning_rate": 4.6640000000000005e-07, "loss": 0.018, "step": 47669 }, { "epoch": 2.5414511915551525, "grad_norm": 1.3065733909606934, "learning_rate": 4.6620000000000003e-07, "loss": 0.0169, "step": 47670 }, { "epoch": 2.5415045049848057, "grad_norm": 2.317551851272583, "learning_rate": 4.6600000000000007e-07, "loss": 0.0094, "step": 47671 }, { "epoch": 2.5415578184144585, "grad_norm": 3.325316905975342, "learning_rate": 4.6580000000000005e-07, "loss": 0.0076, "step": 47672 }, { "epoch": 2.5416111318441117, "grad_norm": 1.2836745977401733, "learning_rate": 4.656e-07, "loss": 0.0053, "step": 47673 }, { "epoch": 2.5416644452737644, "grad_norm": 2.17567777633667, "learning_rate": 4.654e-07, "loss": 0.0058, "step": 47674 }, { "epoch": 2.5417177587034177, "grad_norm": 3.861416816711426, "learning_rate": 4.652e-07, "loss": 0.0175, "step": 47675 }, { "epoch": 2.5417710721330704, "grad_norm": 2.1418492794036865, "learning_rate": 4.6500000000000005e-07, "loss": 0.0264, "step": 47676 }, { "epoch": 2.541824385562723, "grad_norm": 0.44129303097724915, "learning_rate": 4.6480000000000003e-07, "loss": 0.0028, "step": 47677 }, { "epoch": 2.5418776989923764, "grad_norm": 3.5835556983947754, "learning_rate": 4.6460000000000007e-07, "loss": 0.0278, "step": 47678 }, { "epoch": 2.541931012422029, "grad_norm": 1.6606388092041016, "learning_rate": 4.6440000000000006e-07, "loss": 0.0048, "step": 47679 }, { "epoch": 2.541984325851682, "grad_norm": 1.7750238180160522, "learning_rate": 4.642000000000001e-07, "loss": 0.0063, "step": 47680 }, { "epoch": 2.542037639281335, "grad_norm": 2.2534406185150146, "learning_rate": 4.64e-07, "loss": 0.0194, "step": 47681 }, { "epoch": 2.542090952710988, "grad_norm": 3.936413049697876, "learning_rate": 4.638e-07, "loss": 0.0093, "step": 47682 }, { "epoch": 2.5421442661406406, "grad_norm": 0.8706550598144531, "learning_rate": 4.6360000000000005e-07, "loss": 0.0077, "step": 47683 }, { "epoch": 2.542197579570294, "grad_norm": 1.2259665727615356, "learning_rate": 4.6340000000000003e-07, "loss": 0.0141, "step": 47684 }, { "epoch": 2.5422508929999466, "grad_norm": 1.0431573390960693, "learning_rate": 4.632e-07, "loss": 0.0201, "step": 47685 }, { "epoch": 2.5423042064295998, "grad_norm": 2.4520533084869385, "learning_rate": 4.6300000000000006e-07, "loss": 0.034, "step": 47686 }, { "epoch": 2.5423575198592525, "grad_norm": 6.913516044616699, "learning_rate": 4.6280000000000004e-07, "loss": 0.0099, "step": 47687 }, { "epoch": 2.5424108332889057, "grad_norm": 2.6355347633361816, "learning_rate": 4.626000000000001e-07, "loss": 0.0292, "step": 47688 }, { "epoch": 2.5424641467185585, "grad_norm": 4.51577091217041, "learning_rate": 4.6240000000000006e-07, "loss": 0.0313, "step": 47689 }, { "epoch": 2.5425174601482112, "grad_norm": 1.1975008249282837, "learning_rate": 4.622e-07, "loss": 0.0179, "step": 47690 }, { "epoch": 2.5425707735778644, "grad_norm": 1.7489745616912842, "learning_rate": 4.6200000000000003e-07, "loss": 0.0101, "step": 47691 }, { "epoch": 2.542624087007517, "grad_norm": 1.8297010660171509, "learning_rate": 4.618e-07, "loss": 0.0182, "step": 47692 }, { "epoch": 2.54267740043717, "grad_norm": 0.37942254543304443, "learning_rate": 4.6160000000000006e-07, "loss": 0.0024, "step": 47693 }, { "epoch": 2.542730713866823, "grad_norm": 0.8491824269294739, "learning_rate": 4.6140000000000004e-07, "loss": 0.0234, "step": 47694 }, { "epoch": 2.542784027296476, "grad_norm": 1.7426027059555054, "learning_rate": 4.6120000000000003e-07, "loss": 0.033, "step": 47695 }, { "epoch": 2.5428373407261287, "grad_norm": 1.9947211742401123, "learning_rate": 4.6100000000000006e-07, "loss": 0.0238, "step": 47696 }, { "epoch": 2.542890654155782, "grad_norm": 1.1331195831298828, "learning_rate": 4.6080000000000005e-07, "loss": 0.0193, "step": 47697 }, { "epoch": 2.5429439675854346, "grad_norm": 0.3419945538043976, "learning_rate": 4.606e-07, "loss": 0.0023, "step": 47698 }, { "epoch": 2.542997281015088, "grad_norm": 1.6473103761672974, "learning_rate": 4.604e-07, "loss": 0.0051, "step": 47699 }, { "epoch": 2.5430505944447406, "grad_norm": 2.1573917865753174, "learning_rate": 4.602e-07, "loss": 0.0203, "step": 47700 }, { "epoch": 2.5431039078743938, "grad_norm": 2.343142509460449, "learning_rate": 4.6000000000000004e-07, "loss": 0.0129, "step": 47701 }, { "epoch": 2.5431572213040465, "grad_norm": 3.405202627182007, "learning_rate": 4.5980000000000003e-07, "loss": 0.0252, "step": 47702 }, { "epoch": 2.5432105347336993, "grad_norm": 2.145442247390747, "learning_rate": 4.5960000000000007e-07, "loss": 0.0079, "step": 47703 }, { "epoch": 2.5432638481633525, "grad_norm": 1.2203004360198975, "learning_rate": 4.5940000000000005e-07, "loss": 0.0062, "step": 47704 }, { "epoch": 2.5433171615930052, "grad_norm": 3.4545135498046875, "learning_rate": 4.592000000000001e-07, "loss": 0.0131, "step": 47705 }, { "epoch": 2.543370475022658, "grad_norm": 5.389636516571045, "learning_rate": 4.5900000000000007e-07, "loss": 0.0378, "step": 47706 }, { "epoch": 2.543423788452311, "grad_norm": 2.479316234588623, "learning_rate": 4.588e-07, "loss": 0.02, "step": 47707 }, { "epoch": 2.543477101881964, "grad_norm": 1.793570637702942, "learning_rate": 4.5860000000000004e-07, "loss": 0.0174, "step": 47708 }, { "epoch": 2.5435304153116167, "grad_norm": 2.47070050239563, "learning_rate": 4.5840000000000003e-07, "loss": 0.0111, "step": 47709 }, { "epoch": 2.54358372874127, "grad_norm": 0.8994465470314026, "learning_rate": 4.582e-07, "loss": 0.0045, "step": 47710 }, { "epoch": 2.543637042170923, "grad_norm": 1.945868730545044, "learning_rate": 4.5800000000000005e-07, "loss": 0.048, "step": 47711 }, { "epoch": 2.543690355600576, "grad_norm": 3.1501386165618896, "learning_rate": 4.5780000000000004e-07, "loss": 0.0319, "step": 47712 }, { "epoch": 2.5437436690302286, "grad_norm": 3.45973539352417, "learning_rate": 4.576000000000001e-07, "loss": 0.0359, "step": 47713 }, { "epoch": 2.543796982459882, "grad_norm": 2.4122581481933594, "learning_rate": 4.5740000000000006e-07, "loss": 0.0176, "step": 47714 }, { "epoch": 2.5438502958895346, "grad_norm": 1.2387168407440186, "learning_rate": 4.572e-07, "loss": 0.0039, "step": 47715 }, { "epoch": 2.5439036093191874, "grad_norm": 1.7730677127838135, "learning_rate": 4.5700000000000003e-07, "loss": 0.0103, "step": 47716 }, { "epoch": 2.5439569227488406, "grad_norm": 1.5979880094528198, "learning_rate": 4.568e-07, "loss": 0.0054, "step": 47717 }, { "epoch": 2.5440102361784933, "grad_norm": 0.4894346296787262, "learning_rate": 4.5660000000000005e-07, "loss": 0.0032, "step": 47718 }, { "epoch": 2.544063549608146, "grad_norm": 0.7658076286315918, "learning_rate": 4.5640000000000004e-07, "loss": 0.0046, "step": 47719 }, { "epoch": 2.5441168630377993, "grad_norm": 3.313627004623413, "learning_rate": 4.562e-07, "loss": 0.0408, "step": 47720 }, { "epoch": 2.544170176467452, "grad_norm": 1.9211698770523071, "learning_rate": 4.5600000000000006e-07, "loss": 0.0089, "step": 47721 }, { "epoch": 2.5442234898971052, "grad_norm": 2.0215418338775635, "learning_rate": 4.5580000000000004e-07, "loss": 0.0227, "step": 47722 }, { "epoch": 2.544276803326758, "grad_norm": 0.5537534356117249, "learning_rate": 4.556000000000001e-07, "loss": 0.0028, "step": 47723 }, { "epoch": 2.544330116756411, "grad_norm": 1.940824270248413, "learning_rate": 4.554e-07, "loss": 0.0147, "step": 47724 }, { "epoch": 2.544383430186064, "grad_norm": 1.9384907484054565, "learning_rate": 4.552e-07, "loss": 0.0207, "step": 47725 }, { "epoch": 2.5444367436157167, "grad_norm": 1.39254629611969, "learning_rate": 4.5500000000000004e-07, "loss": 0.0194, "step": 47726 }, { "epoch": 2.54449005704537, "grad_norm": 3.006072998046875, "learning_rate": 4.548e-07, "loss": 0.0241, "step": 47727 }, { "epoch": 2.5445433704750227, "grad_norm": 1.5749074220657349, "learning_rate": 4.5460000000000006e-07, "loss": 0.0191, "step": 47728 }, { "epoch": 2.5445966839046754, "grad_norm": 0.4578636586666107, "learning_rate": 4.5440000000000005e-07, "loss": 0.0028, "step": 47729 }, { "epoch": 2.5446499973343286, "grad_norm": 2.1136996746063232, "learning_rate": 4.542000000000001e-07, "loss": 0.0062, "step": 47730 }, { "epoch": 2.5447033107639814, "grad_norm": 2.853282928466797, "learning_rate": 4.5400000000000007e-07, "loss": 0.0191, "step": 47731 }, { "epoch": 2.544756624193634, "grad_norm": 2.822514533996582, "learning_rate": 4.538e-07, "loss": 0.0301, "step": 47732 }, { "epoch": 2.5448099376232873, "grad_norm": 2.4875848293304443, "learning_rate": 4.5360000000000004e-07, "loss": 0.0145, "step": 47733 }, { "epoch": 2.54486325105294, "grad_norm": 1.656322956085205, "learning_rate": 4.534e-07, "loss": 0.007, "step": 47734 }, { "epoch": 2.5449165644825933, "grad_norm": 2.479294538497925, "learning_rate": 4.532e-07, "loss": 0.0288, "step": 47735 }, { "epoch": 2.544969877912246, "grad_norm": 1.528845191001892, "learning_rate": 4.5300000000000005e-07, "loss": 0.0045, "step": 47736 }, { "epoch": 2.5450231913418992, "grad_norm": 2.775294780731201, "learning_rate": 4.5280000000000003e-07, "loss": 0.0262, "step": 47737 }, { "epoch": 2.545076504771552, "grad_norm": 1.2736316919326782, "learning_rate": 4.5260000000000007e-07, "loss": 0.0247, "step": 47738 }, { "epoch": 2.5451298182012048, "grad_norm": 4.39465856552124, "learning_rate": 4.5240000000000005e-07, "loss": 0.0212, "step": 47739 }, { "epoch": 2.545183131630858, "grad_norm": 3.4768292903900146, "learning_rate": 4.522000000000001e-07, "loss": 0.061, "step": 47740 }, { "epoch": 2.5452364450605107, "grad_norm": 0.37916257977485657, "learning_rate": 4.52e-07, "loss": 0.0033, "step": 47741 }, { "epoch": 2.5452897584901635, "grad_norm": 0.7397108674049377, "learning_rate": 4.518e-07, "loss": 0.0038, "step": 47742 }, { "epoch": 2.5453430719198167, "grad_norm": 2.508143186569214, "learning_rate": 4.5160000000000005e-07, "loss": 0.0072, "step": 47743 }, { "epoch": 2.5453963853494694, "grad_norm": 2.7986977100372314, "learning_rate": 4.5140000000000003e-07, "loss": 0.0282, "step": 47744 }, { "epoch": 2.545449698779122, "grad_norm": 0.3548837900161743, "learning_rate": 4.512e-07, "loss": 0.0017, "step": 47745 }, { "epoch": 2.5455030122087754, "grad_norm": 0.9260558485984802, "learning_rate": 4.5100000000000005e-07, "loss": 0.0058, "step": 47746 }, { "epoch": 2.5455563256384286, "grad_norm": 0.7325119376182556, "learning_rate": 4.5080000000000004e-07, "loss": 0.0041, "step": 47747 }, { "epoch": 2.5456096390680814, "grad_norm": 2.254256248474121, "learning_rate": 4.506000000000001e-07, "loss": 0.0364, "step": 47748 }, { "epoch": 2.545662952497734, "grad_norm": 1.914865493774414, "learning_rate": 4.504e-07, "loss": 0.028, "step": 47749 }, { "epoch": 2.5457162659273873, "grad_norm": 1.6287591457366943, "learning_rate": 4.502e-07, "loss": 0.004, "step": 47750 }, { "epoch": 2.54576957935704, "grad_norm": 2.5586278438568115, "learning_rate": 4.5000000000000003e-07, "loss": 0.03, "step": 47751 }, { "epoch": 2.545822892786693, "grad_norm": 1.662358283996582, "learning_rate": 4.498e-07, "loss": 0.0054, "step": 47752 }, { "epoch": 2.545876206216346, "grad_norm": 1.7646223306655884, "learning_rate": 4.4960000000000006e-07, "loss": 0.0363, "step": 47753 }, { "epoch": 2.545929519645999, "grad_norm": 2.8535757064819336, "learning_rate": 4.4940000000000004e-07, "loss": 0.0112, "step": 47754 }, { "epoch": 2.5459828330756515, "grad_norm": 3.4155540466308594, "learning_rate": 4.492000000000001e-07, "loss": 0.043, "step": 47755 }, { "epoch": 2.5460361465053047, "grad_norm": 1.5146301984786987, "learning_rate": 4.4900000000000006e-07, "loss": 0.0162, "step": 47756 }, { "epoch": 2.5460894599349575, "grad_norm": 2.5188307762145996, "learning_rate": 4.4880000000000005e-07, "loss": 0.0155, "step": 47757 }, { "epoch": 2.5461427733646107, "grad_norm": 0.29540905356407166, "learning_rate": 4.4860000000000003e-07, "loss": 0.002, "step": 47758 }, { "epoch": 2.5461960867942635, "grad_norm": 0.47300153970718384, "learning_rate": 4.484e-07, "loss": 0.002, "step": 47759 }, { "epoch": 2.5462494002239167, "grad_norm": 1.2069189548492432, "learning_rate": 4.482e-07, "loss": 0.0057, "step": 47760 }, { "epoch": 2.5463027136535694, "grad_norm": 1.686629295349121, "learning_rate": 4.4800000000000004e-07, "loss": 0.0127, "step": 47761 }, { "epoch": 2.546356027083222, "grad_norm": 2.676424264907837, "learning_rate": 4.478e-07, "loss": 0.0262, "step": 47762 }, { "epoch": 2.5464093405128754, "grad_norm": 0.8144016265869141, "learning_rate": 4.4760000000000006e-07, "loss": 0.0044, "step": 47763 }, { "epoch": 2.546462653942528, "grad_norm": 3.619300365447998, "learning_rate": 4.4740000000000005e-07, "loss": 0.0313, "step": 47764 }, { "epoch": 2.546515967372181, "grad_norm": 1.5255703926086426, "learning_rate": 4.472000000000001e-07, "loss": 0.0057, "step": 47765 }, { "epoch": 2.546569280801834, "grad_norm": 4.530970096588135, "learning_rate": 4.47e-07, "loss": 0.0225, "step": 47766 }, { "epoch": 2.546622594231487, "grad_norm": 2.53985857963562, "learning_rate": 4.468e-07, "loss": 0.0074, "step": 47767 }, { "epoch": 2.5466759076611396, "grad_norm": 0.39991673827171326, "learning_rate": 4.4660000000000004e-07, "loss": 0.0032, "step": 47768 }, { "epoch": 2.546729221090793, "grad_norm": 0.4798215925693512, "learning_rate": 4.4640000000000003e-07, "loss": 0.0029, "step": 47769 }, { "epoch": 2.5467825345204456, "grad_norm": 0.28630077838897705, "learning_rate": 4.462e-07, "loss": 0.0027, "step": 47770 }, { "epoch": 2.5468358479500988, "grad_norm": 2.4841725826263428, "learning_rate": 4.4600000000000005e-07, "loss": 0.0125, "step": 47771 }, { "epoch": 2.5468891613797515, "grad_norm": 1.6443870067596436, "learning_rate": 4.4580000000000003e-07, "loss": 0.0154, "step": 47772 }, { "epoch": 2.5469424748094047, "grad_norm": 1.87037193775177, "learning_rate": 4.4560000000000007e-07, "loss": 0.0369, "step": 47773 }, { "epoch": 2.5469957882390575, "grad_norm": 1.4992578029632568, "learning_rate": 4.4540000000000006e-07, "loss": 0.0068, "step": 47774 }, { "epoch": 2.5470491016687102, "grad_norm": 1.644770622253418, "learning_rate": 4.452e-07, "loss": 0.0201, "step": 47775 }, { "epoch": 2.5471024150983634, "grad_norm": 0.6368350982666016, "learning_rate": 4.4500000000000003e-07, "loss": 0.0043, "step": 47776 }, { "epoch": 2.547155728528016, "grad_norm": 1.2288330793380737, "learning_rate": 4.448e-07, "loss": 0.0057, "step": 47777 }, { "epoch": 2.547209041957669, "grad_norm": 2.182053327560425, "learning_rate": 4.4460000000000005e-07, "loss": 0.017, "step": 47778 }, { "epoch": 2.547262355387322, "grad_norm": 3.3039369583129883, "learning_rate": 4.4440000000000004e-07, "loss": 0.0255, "step": 47779 }, { "epoch": 2.547315668816975, "grad_norm": 1.0630379915237427, "learning_rate": 4.4420000000000007e-07, "loss": 0.0268, "step": 47780 }, { "epoch": 2.5473689822466277, "grad_norm": 4.128617286682129, "learning_rate": 4.4400000000000006e-07, "loss": 0.0479, "step": 47781 }, { "epoch": 2.547422295676281, "grad_norm": 1.1552250385284424, "learning_rate": 4.4380000000000004e-07, "loss": 0.0171, "step": 47782 }, { "epoch": 2.5474756091059336, "grad_norm": 3.0497984886169434, "learning_rate": 4.4360000000000003e-07, "loss": 0.0158, "step": 47783 }, { "epoch": 2.547528922535587, "grad_norm": 4.104601860046387, "learning_rate": 4.434e-07, "loss": 0.0306, "step": 47784 }, { "epoch": 2.5475822359652396, "grad_norm": 1.9357999563217163, "learning_rate": 4.432e-07, "loss": 0.0082, "step": 47785 }, { "epoch": 2.547635549394893, "grad_norm": 1.3643168210983276, "learning_rate": 4.4300000000000004e-07, "loss": 0.0146, "step": 47786 }, { "epoch": 2.5476888628245455, "grad_norm": 3.123520612716675, "learning_rate": 4.428e-07, "loss": 0.0191, "step": 47787 }, { "epoch": 2.5477421762541983, "grad_norm": 3.9442105293273926, "learning_rate": 4.4260000000000006e-07, "loss": 0.0224, "step": 47788 }, { "epoch": 2.5477954896838515, "grad_norm": 0.9808018803596497, "learning_rate": 4.4240000000000004e-07, "loss": 0.0034, "step": 47789 }, { "epoch": 2.5478488031135043, "grad_norm": 1.5303494930267334, "learning_rate": 4.422000000000001e-07, "loss": 0.0074, "step": 47790 }, { "epoch": 2.547902116543157, "grad_norm": 1.3660436868667603, "learning_rate": 4.4200000000000007e-07, "loss": 0.0126, "step": 47791 }, { "epoch": 2.54795542997281, "grad_norm": 2.1761832237243652, "learning_rate": 4.418e-07, "loss": 0.006, "step": 47792 }, { "epoch": 2.548008743402463, "grad_norm": 2.348388671875, "learning_rate": 4.4160000000000004e-07, "loss": 0.0073, "step": 47793 }, { "epoch": 2.548062056832116, "grad_norm": 1.7760915756225586, "learning_rate": 4.414e-07, "loss": 0.0105, "step": 47794 }, { "epoch": 2.548115370261769, "grad_norm": 0.9622431993484497, "learning_rate": 4.412e-07, "loss": 0.0102, "step": 47795 }, { "epoch": 2.548168683691422, "grad_norm": 2.6726796627044678, "learning_rate": 4.4100000000000004e-07, "loss": 0.0131, "step": 47796 }, { "epoch": 2.548221997121075, "grad_norm": 2.296802043914795, "learning_rate": 4.4080000000000003e-07, "loss": 0.0211, "step": 47797 }, { "epoch": 2.5482753105507276, "grad_norm": 1.839048147201538, "learning_rate": 4.4060000000000007e-07, "loss": 0.0066, "step": 47798 }, { "epoch": 2.548328623980381, "grad_norm": 1.9900193214416504, "learning_rate": 4.4040000000000005e-07, "loss": 0.0354, "step": 47799 }, { "epoch": 2.5483819374100336, "grad_norm": 0.66555255651474, "learning_rate": 4.402e-07, "loss": 0.0034, "step": 47800 }, { "epoch": 2.5484352508396864, "grad_norm": 1.8141928911209106, "learning_rate": 4.4e-07, "loss": 0.0404, "step": 47801 }, { "epoch": 2.5484885642693396, "grad_norm": 1.180220365524292, "learning_rate": 4.398e-07, "loss": 0.0233, "step": 47802 }, { "epoch": 2.5485418776989923, "grad_norm": 1.1907435655593872, "learning_rate": 4.3960000000000005e-07, "loss": 0.0048, "step": 47803 }, { "epoch": 2.548595191128645, "grad_norm": 1.9234613180160522, "learning_rate": 4.3940000000000003e-07, "loss": 0.0152, "step": 47804 }, { "epoch": 2.5486485045582983, "grad_norm": 1.531722068786621, "learning_rate": 4.3920000000000007e-07, "loss": 0.0073, "step": 47805 }, { "epoch": 2.548701817987951, "grad_norm": 3.4434311389923096, "learning_rate": 4.3900000000000005e-07, "loss": 0.0119, "step": 47806 }, { "epoch": 2.5487551314176042, "grad_norm": 3.682953357696533, "learning_rate": 4.3880000000000004e-07, "loss": 0.06, "step": 47807 }, { "epoch": 2.548808444847257, "grad_norm": 3.709439754486084, "learning_rate": 4.386000000000001e-07, "loss": 0.0342, "step": 47808 }, { "epoch": 2.54886175827691, "grad_norm": 2.343496322631836, "learning_rate": 4.384e-07, "loss": 0.0065, "step": 47809 }, { "epoch": 2.548915071706563, "grad_norm": 1.0540308952331543, "learning_rate": 4.382e-07, "loss": 0.0042, "step": 47810 }, { "epoch": 2.5489683851362157, "grad_norm": 2.0247437953948975, "learning_rate": 4.3800000000000003e-07, "loss": 0.0262, "step": 47811 }, { "epoch": 2.549021698565869, "grad_norm": 2.2277417182922363, "learning_rate": 4.378e-07, "loss": 0.0113, "step": 47812 }, { "epoch": 2.5490750119955217, "grad_norm": 2.0998916625976562, "learning_rate": 4.3760000000000005e-07, "loss": 0.0258, "step": 47813 }, { "epoch": 2.5491283254251744, "grad_norm": 3.4971673488616943, "learning_rate": 4.3740000000000004e-07, "loss": 0.0103, "step": 47814 }, { "epoch": 2.5491816388548276, "grad_norm": 1.5732437372207642, "learning_rate": 4.372000000000001e-07, "loss": 0.0062, "step": 47815 }, { "epoch": 2.5492349522844804, "grad_norm": 2.222966194152832, "learning_rate": 4.3700000000000006e-07, "loss": 0.015, "step": 47816 }, { "epoch": 2.549288265714133, "grad_norm": 1.295350432395935, "learning_rate": 4.368e-07, "loss": 0.0053, "step": 47817 }, { "epoch": 2.5493415791437863, "grad_norm": 3.7398464679718018, "learning_rate": 4.3660000000000003e-07, "loss": 0.0199, "step": 47818 }, { "epoch": 2.549394892573439, "grad_norm": 6.989306926727295, "learning_rate": 4.364e-07, "loss": 0.0125, "step": 47819 }, { "epoch": 2.5494482060030923, "grad_norm": 1.859286904335022, "learning_rate": 4.362e-07, "loss": 0.0124, "step": 47820 }, { "epoch": 2.549501519432745, "grad_norm": 0.6848067045211792, "learning_rate": 4.3600000000000004e-07, "loss": 0.0033, "step": 47821 }, { "epoch": 2.5495548328623983, "grad_norm": 1.677626132965088, "learning_rate": 4.358e-07, "loss": 0.014, "step": 47822 }, { "epoch": 2.549608146292051, "grad_norm": 2.13671875, "learning_rate": 4.3560000000000006e-07, "loss": 0.0257, "step": 47823 }, { "epoch": 2.5496614597217038, "grad_norm": 2.152472734451294, "learning_rate": 4.3540000000000005e-07, "loss": 0.0161, "step": 47824 }, { "epoch": 2.549714773151357, "grad_norm": 2.0042972564697266, "learning_rate": 4.352000000000001e-07, "loss": 0.0083, "step": 47825 }, { "epoch": 2.5497680865810097, "grad_norm": 6.419985771179199, "learning_rate": 4.35e-07, "loss": 0.0146, "step": 47826 }, { "epoch": 2.5498214000106625, "grad_norm": 0.9646424055099487, "learning_rate": 4.348e-07, "loss": 0.0042, "step": 47827 }, { "epoch": 2.5498747134403157, "grad_norm": 1.8249298334121704, "learning_rate": 4.3460000000000004e-07, "loss": 0.0162, "step": 47828 }, { "epoch": 2.5499280268699684, "grad_norm": 3.6527771949768066, "learning_rate": 4.344e-07, "loss": 0.0322, "step": 47829 }, { "epoch": 2.549981340299621, "grad_norm": 4.279304027557373, "learning_rate": 4.3420000000000006e-07, "loss": 0.0115, "step": 47830 }, { "epoch": 2.5500346537292744, "grad_norm": 1.2204172611236572, "learning_rate": 4.3400000000000005e-07, "loss": 0.006, "step": 47831 }, { "epoch": 2.5500879671589276, "grad_norm": 1.617079496383667, "learning_rate": 4.3380000000000003e-07, "loss": 0.0046, "step": 47832 }, { "epoch": 2.5501412805885804, "grad_norm": 2.34661865234375, "learning_rate": 4.3360000000000007e-07, "loss": 0.0075, "step": 47833 }, { "epoch": 2.550194594018233, "grad_norm": 0.9641081690788269, "learning_rate": 4.334e-07, "loss": 0.0053, "step": 47834 }, { "epoch": 2.5502479074478863, "grad_norm": 5.639466762542725, "learning_rate": 4.332e-07, "loss": 0.0503, "step": 47835 }, { "epoch": 2.550301220877539, "grad_norm": 1.3598202466964722, "learning_rate": 4.33e-07, "loss": 0.0307, "step": 47836 }, { "epoch": 2.550354534307192, "grad_norm": 1.0073403120040894, "learning_rate": 4.328e-07, "loss": 0.0032, "step": 47837 }, { "epoch": 2.550407847736845, "grad_norm": 0.7566428184509277, "learning_rate": 4.3260000000000005e-07, "loss": 0.0042, "step": 47838 }, { "epoch": 2.550461161166498, "grad_norm": 2.2628021240234375, "learning_rate": 4.3240000000000003e-07, "loss": 0.0365, "step": 47839 }, { "epoch": 2.5505144745961505, "grad_norm": 1.3190497159957886, "learning_rate": 4.3220000000000007e-07, "loss": 0.0036, "step": 47840 }, { "epoch": 2.5505677880258038, "grad_norm": 0.40525272488594055, "learning_rate": 4.3200000000000006e-07, "loss": 0.0033, "step": 47841 }, { "epoch": 2.5506211014554565, "grad_norm": 0.9428344964981079, "learning_rate": 4.3180000000000004e-07, "loss": 0.005, "step": 47842 }, { "epoch": 2.5506744148851097, "grad_norm": 3.4497382640838623, "learning_rate": 4.3160000000000003e-07, "loss": 0.0393, "step": 47843 }, { "epoch": 2.5507277283147625, "grad_norm": 1.8143390417099, "learning_rate": 4.314e-07, "loss": 0.0135, "step": 47844 }, { "epoch": 2.5507810417444157, "grad_norm": 1.6942166090011597, "learning_rate": 4.312e-07, "loss": 0.0418, "step": 47845 }, { "epoch": 2.5508343551740684, "grad_norm": 3.4497857093811035, "learning_rate": 4.3100000000000003e-07, "loss": 0.0146, "step": 47846 }, { "epoch": 2.550887668603721, "grad_norm": 1.4143675565719604, "learning_rate": 4.308e-07, "loss": 0.0069, "step": 47847 }, { "epoch": 2.5509409820333744, "grad_norm": 3.2655868530273438, "learning_rate": 4.3060000000000006e-07, "loss": 0.0101, "step": 47848 }, { "epoch": 2.550994295463027, "grad_norm": 1.8859432935714722, "learning_rate": 4.3040000000000004e-07, "loss": 0.0253, "step": 47849 }, { "epoch": 2.55104760889268, "grad_norm": 3.0430800914764404, "learning_rate": 4.302000000000001e-07, "loss": 0.0256, "step": 47850 }, { "epoch": 2.551100922322333, "grad_norm": 1.0928350687026978, "learning_rate": 4.3e-07, "loss": 0.0049, "step": 47851 }, { "epoch": 2.551154235751986, "grad_norm": 0.9533293843269348, "learning_rate": 4.298e-07, "loss": 0.0034, "step": 47852 }, { "epoch": 2.5512075491816386, "grad_norm": 0.6923473477363586, "learning_rate": 4.2960000000000004e-07, "loss": 0.0035, "step": 47853 }, { "epoch": 2.551260862611292, "grad_norm": 1.4836434125900269, "learning_rate": 4.294e-07, "loss": 0.0045, "step": 47854 }, { "epoch": 2.5513141760409446, "grad_norm": 1.8047773838043213, "learning_rate": 4.2920000000000006e-07, "loss": 0.0209, "step": 47855 }, { "epoch": 2.5513674894705978, "grad_norm": 1.6910734176635742, "learning_rate": 4.2900000000000004e-07, "loss": 0.0181, "step": 47856 }, { "epoch": 2.5514208029002505, "grad_norm": 0.2341901957988739, "learning_rate": 4.2880000000000003e-07, "loss": 0.0013, "step": 47857 }, { "epoch": 2.5514741163299037, "grad_norm": 0.5340784788131714, "learning_rate": 4.2860000000000007e-07, "loss": 0.0045, "step": 47858 }, { "epoch": 2.5515274297595565, "grad_norm": 2.395979642868042, "learning_rate": 4.2840000000000005e-07, "loss": 0.0248, "step": 47859 }, { "epoch": 2.5515807431892092, "grad_norm": 2.8987011909484863, "learning_rate": 4.282e-07, "loss": 0.0238, "step": 47860 }, { "epoch": 2.5516340566188624, "grad_norm": 1.5048342943191528, "learning_rate": 4.28e-07, "loss": 0.0168, "step": 47861 }, { "epoch": 2.551687370048515, "grad_norm": 2.2820072174072266, "learning_rate": 4.278e-07, "loss": 0.0216, "step": 47862 }, { "epoch": 2.551740683478168, "grad_norm": 1.7978549003601074, "learning_rate": 4.2760000000000004e-07, "loss": 0.0424, "step": 47863 }, { "epoch": 2.551793996907821, "grad_norm": 3.784086227416992, "learning_rate": 4.2740000000000003e-07, "loss": 0.0396, "step": 47864 }, { "epoch": 2.551847310337474, "grad_norm": 1.1751888990402222, "learning_rate": 4.2720000000000007e-07, "loss": 0.0177, "step": 47865 }, { "epoch": 2.5519006237671267, "grad_norm": 0.6031985282897949, "learning_rate": 4.2700000000000005e-07, "loss": 0.0034, "step": 47866 }, { "epoch": 2.55195393719678, "grad_norm": 1.6918245553970337, "learning_rate": 4.2680000000000004e-07, "loss": 0.008, "step": 47867 }, { "epoch": 2.5520072506264326, "grad_norm": 1.329985499382019, "learning_rate": 4.266e-07, "loss": 0.0063, "step": 47868 }, { "epoch": 2.552060564056086, "grad_norm": 1.0860662460327148, "learning_rate": 4.264e-07, "loss": 0.0138, "step": 47869 }, { "epoch": 2.5521138774857386, "grad_norm": 0.6116555333137512, "learning_rate": 4.262e-07, "loss": 0.004, "step": 47870 }, { "epoch": 2.552167190915392, "grad_norm": 2.8669180870056152, "learning_rate": 4.2600000000000003e-07, "loss": 0.0268, "step": 47871 }, { "epoch": 2.5522205043450445, "grad_norm": 3.5283634662628174, "learning_rate": 4.258e-07, "loss": 0.0319, "step": 47872 }, { "epoch": 2.5522738177746973, "grad_norm": 1.2924901247024536, "learning_rate": 4.2560000000000005e-07, "loss": 0.0043, "step": 47873 }, { "epoch": 2.5523271312043505, "grad_norm": 1.5777373313903809, "learning_rate": 4.2540000000000004e-07, "loss": 0.0044, "step": 47874 }, { "epoch": 2.5523804446340033, "grad_norm": 2.7698662281036377, "learning_rate": 4.252000000000001e-07, "loss": 0.0072, "step": 47875 }, { "epoch": 2.552433758063656, "grad_norm": 0.9719195365905762, "learning_rate": 4.2500000000000006e-07, "loss": 0.006, "step": 47876 }, { "epoch": 2.5524870714933092, "grad_norm": 0.6578080654144287, "learning_rate": 4.248e-07, "loss": 0.0156, "step": 47877 }, { "epoch": 2.552540384922962, "grad_norm": 1.9491595029830933, "learning_rate": 4.2460000000000003e-07, "loss": 0.0065, "step": 47878 }, { "epoch": 2.552593698352615, "grad_norm": 2.4900963306427, "learning_rate": 4.244e-07, "loss": 0.0081, "step": 47879 }, { "epoch": 2.552647011782268, "grad_norm": 1.5981725454330444, "learning_rate": 4.2420000000000005e-07, "loss": 0.0081, "step": 47880 }, { "epoch": 2.552700325211921, "grad_norm": 0.7788522243499756, "learning_rate": 4.2400000000000004e-07, "loss": 0.0037, "step": 47881 }, { "epoch": 2.552753638641574, "grad_norm": 1.1580151319503784, "learning_rate": 4.238e-07, "loss": 0.0064, "step": 47882 }, { "epoch": 2.5528069520712267, "grad_norm": 1.197054386138916, "learning_rate": 4.2360000000000006e-07, "loss": 0.0066, "step": 47883 }, { "epoch": 2.55286026550088, "grad_norm": 1.5257833003997803, "learning_rate": 4.2340000000000005e-07, "loss": 0.0166, "step": 47884 }, { "epoch": 2.5529135789305326, "grad_norm": 1.7801544666290283, "learning_rate": 4.232000000000001e-07, "loss": 0.0247, "step": 47885 }, { "epoch": 2.5529668923601854, "grad_norm": 2.2875232696533203, "learning_rate": 4.23e-07, "loss": 0.0085, "step": 47886 }, { "epoch": 2.5530202057898386, "grad_norm": 2.7613298892974854, "learning_rate": 4.228e-07, "loss": 0.0186, "step": 47887 }, { "epoch": 2.5530735192194913, "grad_norm": 1.2507669925689697, "learning_rate": 4.2260000000000004e-07, "loss": 0.0043, "step": 47888 }, { "epoch": 2.553126832649144, "grad_norm": 2.499542713165283, "learning_rate": 4.224e-07, "loss": 0.0152, "step": 47889 }, { "epoch": 2.5531801460787973, "grad_norm": 1.7256540060043335, "learning_rate": 4.2220000000000006e-07, "loss": 0.0174, "step": 47890 }, { "epoch": 2.55323345950845, "grad_norm": 0.9409584999084473, "learning_rate": 4.2200000000000005e-07, "loss": 0.0215, "step": 47891 }, { "epoch": 2.5532867729381032, "grad_norm": 1.8344264030456543, "learning_rate": 4.2180000000000003e-07, "loss": 0.0099, "step": 47892 }, { "epoch": 2.553340086367756, "grad_norm": 2.3869681358337402, "learning_rate": 4.2160000000000007e-07, "loss": 0.0049, "step": 47893 }, { "epoch": 2.553393399797409, "grad_norm": 2.2104651927948, "learning_rate": 4.214e-07, "loss": 0.0112, "step": 47894 }, { "epoch": 2.553446713227062, "grad_norm": 2.648564338684082, "learning_rate": 4.212e-07, "loss": 0.0229, "step": 47895 }, { "epoch": 2.5535000266567147, "grad_norm": 1.9870924949645996, "learning_rate": 4.21e-07, "loss": 0.0058, "step": 47896 }, { "epoch": 2.553553340086368, "grad_norm": 1.405495285987854, "learning_rate": 4.208e-07, "loss": 0.0174, "step": 47897 }, { "epoch": 2.5536066535160207, "grad_norm": 0.8480596542358398, "learning_rate": 4.2060000000000005e-07, "loss": 0.0039, "step": 47898 }, { "epoch": 2.5536599669456734, "grad_norm": 2.476592540740967, "learning_rate": 4.2040000000000003e-07, "loss": 0.0167, "step": 47899 }, { "epoch": 2.5537132803753266, "grad_norm": 1.0344791412353516, "learning_rate": 4.2020000000000007e-07, "loss": 0.0175, "step": 47900 }, { "epoch": 2.5537665938049794, "grad_norm": 1.0114120244979858, "learning_rate": 4.2000000000000006e-07, "loss": 0.0077, "step": 47901 }, { "epoch": 2.553819907234632, "grad_norm": 3.0801198482513428, "learning_rate": 4.198000000000001e-07, "loss": 0.0239, "step": 47902 }, { "epoch": 2.5538732206642853, "grad_norm": 4.673127174377441, "learning_rate": 4.196e-07, "loss": 0.0348, "step": 47903 }, { "epoch": 2.553926534093938, "grad_norm": 1.7319755554199219, "learning_rate": 4.194e-07, "loss": 0.0128, "step": 47904 }, { "epoch": 2.5539798475235913, "grad_norm": 1.4919782876968384, "learning_rate": 4.1920000000000005e-07, "loss": 0.0169, "step": 47905 }, { "epoch": 2.554033160953244, "grad_norm": 2.818978786468506, "learning_rate": 4.1900000000000003e-07, "loss": 0.0218, "step": 47906 }, { "epoch": 2.5540864743828973, "grad_norm": 3.2379562854766846, "learning_rate": 4.188e-07, "loss": 0.0243, "step": 47907 }, { "epoch": 2.55413978781255, "grad_norm": 2.3967349529266357, "learning_rate": 4.1860000000000006e-07, "loss": 0.0257, "step": 47908 }, { "epoch": 2.554193101242203, "grad_norm": 0.9520402550697327, "learning_rate": 4.1840000000000004e-07, "loss": 0.0094, "step": 47909 }, { "epoch": 2.554246414671856, "grad_norm": 1.2077487707138062, "learning_rate": 4.182000000000001e-07, "loss": 0.0047, "step": 47910 }, { "epoch": 2.5542997281015087, "grad_norm": 0.7340608239173889, "learning_rate": 4.18e-07, "loss": 0.0041, "step": 47911 }, { "epoch": 2.5543530415311615, "grad_norm": 2.2696692943573, "learning_rate": 4.178e-07, "loss": 0.0063, "step": 47912 }, { "epoch": 2.5544063549608147, "grad_norm": 0.25112950801849365, "learning_rate": 4.1760000000000003e-07, "loss": 0.0018, "step": 47913 }, { "epoch": 2.5544596683904675, "grad_norm": 3.582594394683838, "learning_rate": 4.174e-07, "loss": 0.011, "step": 47914 }, { "epoch": 2.5545129818201207, "grad_norm": 0.9127772450447083, "learning_rate": 4.1720000000000006e-07, "loss": 0.0113, "step": 47915 }, { "epoch": 2.5545662952497734, "grad_norm": 0.9465378522872925, "learning_rate": 4.1700000000000004e-07, "loss": 0.0042, "step": 47916 }, { "epoch": 2.5546196086794266, "grad_norm": 1.7165024280548096, "learning_rate": 4.168e-07, "loss": 0.0201, "step": 47917 }, { "epoch": 2.5546729221090794, "grad_norm": 2.2610230445861816, "learning_rate": 4.1660000000000006e-07, "loss": 0.0124, "step": 47918 }, { "epoch": 2.554726235538732, "grad_norm": 0.27597320079803467, "learning_rate": 4.1640000000000005e-07, "loss": 0.0026, "step": 47919 }, { "epoch": 2.5547795489683853, "grad_norm": 2.010702133178711, "learning_rate": 4.162e-07, "loss": 0.0077, "step": 47920 }, { "epoch": 2.554832862398038, "grad_norm": 3.430314064025879, "learning_rate": 4.16e-07, "loss": 0.0517, "step": 47921 }, { "epoch": 2.554886175827691, "grad_norm": 2.1901473999023438, "learning_rate": 4.158e-07, "loss": 0.0216, "step": 47922 }, { "epoch": 2.554939489257344, "grad_norm": 2.377124071121216, "learning_rate": 4.1560000000000004e-07, "loss": 0.0066, "step": 47923 }, { "epoch": 2.554992802686997, "grad_norm": 2.577221155166626, "learning_rate": 4.1540000000000003e-07, "loss": 0.0148, "step": 47924 }, { "epoch": 2.5550461161166496, "grad_norm": 1.1135934591293335, "learning_rate": 4.1520000000000007e-07, "loss": 0.0039, "step": 47925 }, { "epoch": 2.5550994295463028, "grad_norm": 2.5383706092834473, "learning_rate": 4.1500000000000005e-07, "loss": 0.0192, "step": 47926 }, { "epoch": 2.5551527429759555, "grad_norm": 1.5035651922225952, "learning_rate": 4.148000000000001e-07, "loss": 0.0241, "step": 47927 }, { "epoch": 2.5552060564056087, "grad_norm": 1.238616704940796, "learning_rate": 4.146e-07, "loss": 0.0057, "step": 47928 }, { "epoch": 2.5552593698352615, "grad_norm": 1.324580430984497, "learning_rate": 4.144e-07, "loss": 0.0076, "step": 47929 }, { "epoch": 2.5553126832649147, "grad_norm": 1.3060189485549927, "learning_rate": 4.1420000000000004e-07, "loss": 0.0153, "step": 47930 }, { "epoch": 2.5553659966945674, "grad_norm": 0.2897924780845642, "learning_rate": 4.1400000000000003e-07, "loss": 0.002, "step": 47931 }, { "epoch": 2.55541931012422, "grad_norm": 2.096630811691284, "learning_rate": 4.138e-07, "loss": 0.0068, "step": 47932 }, { "epoch": 2.5554726235538734, "grad_norm": 2.2278671264648438, "learning_rate": 4.1360000000000005e-07, "loss": 0.0147, "step": 47933 }, { "epoch": 2.555525936983526, "grad_norm": 1.463441252708435, "learning_rate": 4.1340000000000004e-07, "loss": 0.0069, "step": 47934 }, { "epoch": 2.555579250413179, "grad_norm": 1.9154294729232788, "learning_rate": 4.132000000000001e-07, "loss": 0.0405, "step": 47935 }, { "epoch": 2.555632563842832, "grad_norm": 0.9854407906532288, "learning_rate": 4.1300000000000006e-07, "loss": 0.0046, "step": 47936 }, { "epoch": 2.555685877272485, "grad_norm": 7.090266704559326, "learning_rate": 4.128e-07, "loss": 0.0193, "step": 47937 }, { "epoch": 2.5557391907021376, "grad_norm": 1.367998719215393, "learning_rate": 4.1260000000000003e-07, "loss": 0.0111, "step": 47938 }, { "epoch": 2.555792504131791, "grad_norm": 2.9689958095550537, "learning_rate": 4.124e-07, "loss": 0.0328, "step": 47939 }, { "epoch": 2.5558458175614436, "grad_norm": 0.6560521125793457, "learning_rate": 4.1220000000000005e-07, "loss": 0.0039, "step": 47940 }, { "epoch": 2.555899130991097, "grad_norm": 3.71730637550354, "learning_rate": 4.1200000000000004e-07, "loss": 0.0267, "step": 47941 }, { "epoch": 2.5559524444207495, "grad_norm": 2.3436522483825684, "learning_rate": 4.118e-07, "loss": 0.0133, "step": 47942 }, { "epoch": 2.5560057578504027, "grad_norm": 3.931349754333496, "learning_rate": 4.1160000000000006e-07, "loss": 0.0223, "step": 47943 }, { "epoch": 2.5560590712800555, "grad_norm": 2.090956449508667, "learning_rate": 4.1140000000000004e-07, "loss": 0.009, "step": 47944 }, { "epoch": 2.5561123847097083, "grad_norm": 0.7174983620643616, "learning_rate": 4.112e-07, "loss": 0.0042, "step": 47945 }, { "epoch": 2.5561656981393615, "grad_norm": 0.508765697479248, "learning_rate": 4.11e-07, "loss": 0.003, "step": 47946 }, { "epoch": 2.556219011569014, "grad_norm": 3.085456132888794, "learning_rate": 4.108e-07, "loss": 0.0198, "step": 47947 }, { "epoch": 2.556272324998667, "grad_norm": 2.0343377590179443, "learning_rate": 4.1060000000000004e-07, "loss": 0.0114, "step": 47948 }, { "epoch": 2.55632563842832, "grad_norm": 1.5494521856307983, "learning_rate": 4.104e-07, "loss": 0.0097, "step": 47949 }, { "epoch": 2.556378951857973, "grad_norm": 2.1145808696746826, "learning_rate": 4.1020000000000006e-07, "loss": 0.008, "step": 47950 }, { "epoch": 2.5564322652876257, "grad_norm": 2.486707925796509, "learning_rate": 4.1000000000000004e-07, "loss": 0.0233, "step": 47951 }, { "epoch": 2.556485578717279, "grad_norm": 1.6136479377746582, "learning_rate": 4.098000000000001e-07, "loss": 0.0061, "step": 47952 }, { "epoch": 2.556538892146932, "grad_norm": 0.9892939329147339, "learning_rate": 4.0960000000000007e-07, "loss": 0.0167, "step": 47953 }, { "epoch": 2.556592205576585, "grad_norm": 1.6809511184692383, "learning_rate": 4.094e-07, "loss": 0.0211, "step": 47954 }, { "epoch": 2.5566455190062376, "grad_norm": 2.2298572063446045, "learning_rate": 4.0920000000000004e-07, "loss": 0.0446, "step": 47955 }, { "epoch": 2.556698832435891, "grad_norm": 1.3486055135726929, "learning_rate": 4.09e-07, "loss": 0.0162, "step": 47956 }, { "epoch": 2.5567521458655436, "grad_norm": 1.9351415634155273, "learning_rate": 4.088e-07, "loss": 0.0194, "step": 47957 }, { "epoch": 2.5568054592951963, "grad_norm": 4.645509719848633, "learning_rate": 4.0860000000000005e-07, "loss": 0.0149, "step": 47958 }, { "epoch": 2.5568587727248495, "grad_norm": 1.1009458303451538, "learning_rate": 4.0840000000000003e-07, "loss": 0.0175, "step": 47959 }, { "epoch": 2.5569120861545023, "grad_norm": 2.1214680671691895, "learning_rate": 4.0820000000000007e-07, "loss": 0.0086, "step": 47960 }, { "epoch": 2.556965399584155, "grad_norm": 5.9466938972473145, "learning_rate": 4.0800000000000005e-07, "loss": 0.0145, "step": 47961 }, { "epoch": 2.5570187130138082, "grad_norm": 1.4945595264434814, "learning_rate": 4.078e-07, "loss": 0.0062, "step": 47962 }, { "epoch": 2.557072026443461, "grad_norm": 2.373403310775757, "learning_rate": 4.076e-07, "loss": 0.0152, "step": 47963 }, { "epoch": 2.557125339873114, "grad_norm": 3.66170597076416, "learning_rate": 4.074e-07, "loss": 0.0387, "step": 47964 }, { "epoch": 2.557178653302767, "grad_norm": 0.8929103016853333, "learning_rate": 4.0720000000000005e-07, "loss": 0.0046, "step": 47965 }, { "epoch": 2.55723196673242, "grad_norm": 1.096721887588501, "learning_rate": 4.0700000000000003e-07, "loss": 0.0063, "step": 47966 }, { "epoch": 2.557285280162073, "grad_norm": 0.8856016397476196, "learning_rate": 4.068e-07, "loss": 0.0039, "step": 47967 }, { "epoch": 2.5573385935917257, "grad_norm": 1.8169057369232178, "learning_rate": 4.0660000000000005e-07, "loss": 0.0146, "step": 47968 }, { "epoch": 2.557391907021379, "grad_norm": 2.412428140640259, "learning_rate": 4.0640000000000004e-07, "loss": 0.017, "step": 47969 }, { "epoch": 2.5574452204510316, "grad_norm": 1.6443167924880981, "learning_rate": 4.062000000000001e-07, "loss": 0.007, "step": 47970 }, { "epoch": 2.5574985338806844, "grad_norm": 2.703338861465454, "learning_rate": 4.06e-07, "loss": 0.0246, "step": 47971 }, { "epoch": 2.5575518473103376, "grad_norm": 1.2446863651275635, "learning_rate": 4.058e-07, "loss": 0.0045, "step": 47972 }, { "epoch": 2.5576051607399903, "grad_norm": 1.513219952583313, "learning_rate": 4.0560000000000003e-07, "loss": 0.0061, "step": 47973 }, { "epoch": 2.557658474169643, "grad_norm": 1.6194977760314941, "learning_rate": 4.054e-07, "loss": 0.0046, "step": 47974 }, { "epoch": 2.5577117875992963, "grad_norm": 2.457304000854492, "learning_rate": 4.0520000000000005e-07, "loss": 0.0237, "step": 47975 }, { "epoch": 2.557765101028949, "grad_norm": 0.606483519077301, "learning_rate": 4.0500000000000004e-07, "loss": 0.0036, "step": 47976 }, { "epoch": 2.5578184144586023, "grad_norm": 2.8952348232269287, "learning_rate": 4.048000000000001e-07, "loss": 0.0245, "step": 47977 }, { "epoch": 2.557871727888255, "grad_norm": 2.174882411956787, "learning_rate": 4.0460000000000006e-07, "loss": 0.0275, "step": 47978 }, { "epoch": 2.557925041317908, "grad_norm": 1.671486735343933, "learning_rate": 4.044e-07, "loss": 0.0121, "step": 47979 }, { "epoch": 2.557978354747561, "grad_norm": 2.394913673400879, "learning_rate": 4.0420000000000003e-07, "loss": 0.0306, "step": 47980 }, { "epoch": 2.5580316681772137, "grad_norm": 2.3939483165740967, "learning_rate": 4.04e-07, "loss": 0.0112, "step": 47981 }, { "epoch": 2.558084981606867, "grad_norm": 1.7587119340896606, "learning_rate": 4.038e-07, "loss": 0.0208, "step": 47982 }, { "epoch": 2.5581382950365197, "grad_norm": 1.351385474205017, "learning_rate": 4.0360000000000004e-07, "loss": 0.0173, "step": 47983 }, { "epoch": 2.5581916084661724, "grad_norm": 0.767029345035553, "learning_rate": 4.034e-07, "loss": 0.0036, "step": 47984 }, { "epoch": 2.5582449218958256, "grad_norm": 1.6899539232254028, "learning_rate": 4.0320000000000006e-07, "loss": 0.0121, "step": 47985 }, { "epoch": 2.5582982353254784, "grad_norm": 1.8938884735107422, "learning_rate": 4.0300000000000005e-07, "loss": 0.0252, "step": 47986 }, { "epoch": 2.558351548755131, "grad_norm": 0.6746195554733276, "learning_rate": 4.028000000000001e-07, "loss": 0.0041, "step": 47987 }, { "epoch": 2.5584048621847844, "grad_norm": 0.2211066335439682, "learning_rate": 4.026e-07, "loss": 0.002, "step": 47988 }, { "epoch": 2.558458175614437, "grad_norm": 0.3065727651119232, "learning_rate": 4.024e-07, "loss": 0.0028, "step": 47989 }, { "epoch": 2.5585114890440903, "grad_norm": 3.3498423099517822, "learning_rate": 4.0220000000000004e-07, "loss": 0.0104, "step": 47990 }, { "epoch": 2.558564802473743, "grad_norm": 1.297186017036438, "learning_rate": 4.02e-07, "loss": 0.0109, "step": 47991 }, { "epoch": 2.5586181159033963, "grad_norm": 1.329786777496338, "learning_rate": 4.018e-07, "loss": 0.0048, "step": 47992 }, { "epoch": 2.558671429333049, "grad_norm": 2.97986102104187, "learning_rate": 4.0160000000000005e-07, "loss": 0.0217, "step": 47993 }, { "epoch": 2.558724742762702, "grad_norm": 1.4102901220321655, "learning_rate": 4.0140000000000003e-07, "loss": 0.0289, "step": 47994 }, { "epoch": 2.558778056192355, "grad_norm": 1.845394253730774, "learning_rate": 4.0120000000000007e-07, "loss": 0.0051, "step": 47995 }, { "epoch": 2.5588313696220077, "grad_norm": 0.7550430297851562, "learning_rate": 4.01e-07, "loss": 0.0037, "step": 47996 }, { "epoch": 2.5588846830516605, "grad_norm": 2.70967698097229, "learning_rate": 4.008e-07, "loss": 0.0166, "step": 47997 }, { "epoch": 2.5589379964813137, "grad_norm": 3.351541519165039, "learning_rate": 4.0060000000000003e-07, "loss": 0.0526, "step": 47998 }, { "epoch": 2.5589913099109665, "grad_norm": 0.6885789632797241, "learning_rate": 4.004e-07, "loss": 0.004, "step": 47999 }, { "epoch": 2.5590446233406197, "grad_norm": 2.89007830619812, "learning_rate": 4.0020000000000005e-07, "loss": 0.0137, "step": 48000 }, { "epoch": 2.5590979367702724, "grad_norm": 1.4675202369689941, "learning_rate": 4.0000000000000003e-07, "loss": 0.0082, "step": 48001 }, { "epoch": 2.5591512501999256, "grad_norm": 0.5430654287338257, "learning_rate": 3.9980000000000007e-07, "loss": 0.002, "step": 48002 }, { "epoch": 2.5592045636295784, "grad_norm": 1.4572169780731201, "learning_rate": 3.9960000000000006e-07, "loss": 0.0062, "step": 48003 }, { "epoch": 2.559257877059231, "grad_norm": 0.45681726932525635, "learning_rate": 3.9940000000000004e-07, "loss": 0.0026, "step": 48004 }, { "epoch": 2.5593111904888843, "grad_norm": 1.7500399351119995, "learning_rate": 3.9920000000000003e-07, "loss": 0.0427, "step": 48005 }, { "epoch": 2.559364503918537, "grad_norm": 1.9881393909454346, "learning_rate": 3.99e-07, "loss": 0.0304, "step": 48006 }, { "epoch": 2.55941781734819, "grad_norm": 2.70967173576355, "learning_rate": 3.988e-07, "loss": 0.0176, "step": 48007 }, { "epoch": 2.559471130777843, "grad_norm": 3.2997498512268066, "learning_rate": 3.9860000000000004e-07, "loss": 0.0118, "step": 48008 }, { "epoch": 2.559524444207496, "grad_norm": 1.362173318862915, "learning_rate": 3.984e-07, "loss": 0.0096, "step": 48009 }, { "epoch": 2.5595777576371486, "grad_norm": 2.3393516540527344, "learning_rate": 3.9820000000000006e-07, "loss": 0.0045, "step": 48010 }, { "epoch": 2.5596310710668018, "grad_norm": 2.697523832321167, "learning_rate": 3.9800000000000004e-07, "loss": 0.0163, "step": 48011 }, { "epoch": 2.5596843844964545, "grad_norm": 0.3103892207145691, "learning_rate": 3.978000000000001e-07, "loss": 0.0022, "step": 48012 }, { "epoch": 2.5597376979261077, "grad_norm": 0.5916504263877869, "learning_rate": 3.976e-07, "loss": 0.0037, "step": 48013 }, { "epoch": 2.5597910113557605, "grad_norm": 4.485052108764648, "learning_rate": 3.974e-07, "loss": 0.0261, "step": 48014 }, { "epoch": 2.5598443247854137, "grad_norm": 3.180445671081543, "learning_rate": 3.9720000000000004e-07, "loss": 0.0177, "step": 48015 }, { "epoch": 2.5598976382150664, "grad_norm": 2.748785972595215, "learning_rate": 3.97e-07, "loss": 0.0245, "step": 48016 }, { "epoch": 2.559950951644719, "grad_norm": 2.8939290046691895, "learning_rate": 3.968e-07, "loss": 0.0162, "step": 48017 }, { "epoch": 2.5600042650743724, "grad_norm": 0.7172154188156128, "learning_rate": 3.9660000000000004e-07, "loss": 0.0032, "step": 48018 }, { "epoch": 2.560057578504025, "grad_norm": 0.8939428329467773, "learning_rate": 3.9640000000000003e-07, "loss": 0.0057, "step": 48019 }, { "epoch": 2.560110891933678, "grad_norm": 3.2806694507598877, "learning_rate": 3.9620000000000007e-07, "loss": 0.0286, "step": 48020 }, { "epoch": 2.560164205363331, "grad_norm": 1.4254077672958374, "learning_rate": 3.9600000000000005e-07, "loss": 0.0043, "step": 48021 }, { "epoch": 2.560217518792984, "grad_norm": 2.81138277053833, "learning_rate": 3.958e-07, "loss": 0.0095, "step": 48022 }, { "epoch": 2.5602708322226366, "grad_norm": 0.9209804534912109, "learning_rate": 3.956e-07, "loss": 0.0039, "step": 48023 }, { "epoch": 2.56032414565229, "grad_norm": 1.0063395500183105, "learning_rate": 3.954e-07, "loss": 0.0049, "step": 48024 }, { "epoch": 2.5603774590819426, "grad_norm": 2.1472015380859375, "learning_rate": 3.9520000000000004e-07, "loss": 0.0123, "step": 48025 }, { "epoch": 2.560430772511596, "grad_norm": 1.201871633529663, "learning_rate": 3.9500000000000003e-07, "loss": 0.0051, "step": 48026 }, { "epoch": 2.5604840859412485, "grad_norm": 0.37147045135498047, "learning_rate": 3.9480000000000007e-07, "loss": 0.0025, "step": 48027 }, { "epoch": 2.5605373993709017, "grad_norm": 1.0979256629943848, "learning_rate": 3.9460000000000005e-07, "loss": 0.004, "step": 48028 }, { "epoch": 2.5605907128005545, "grad_norm": 3.0657694339752197, "learning_rate": 3.9440000000000004e-07, "loss": 0.0259, "step": 48029 }, { "epoch": 2.5606440262302073, "grad_norm": 3.9086790084838867, "learning_rate": 3.942e-07, "loss": 0.016, "step": 48030 }, { "epoch": 2.5606973396598605, "grad_norm": 2.6864750385284424, "learning_rate": 3.94e-07, "loss": 0.0121, "step": 48031 }, { "epoch": 2.560750653089513, "grad_norm": 1.549331784248352, "learning_rate": 3.938e-07, "loss": 0.0129, "step": 48032 }, { "epoch": 2.560803966519166, "grad_norm": 1.7639254331588745, "learning_rate": 3.9360000000000003e-07, "loss": 0.0359, "step": 48033 }, { "epoch": 2.560857279948819, "grad_norm": 0.3739221692085266, "learning_rate": 3.934e-07, "loss": 0.0023, "step": 48034 }, { "epoch": 2.560910593378472, "grad_norm": 2.287381887435913, "learning_rate": 3.9320000000000005e-07, "loss": 0.0336, "step": 48035 }, { "epoch": 2.5609639068081247, "grad_norm": 1.0214011669158936, "learning_rate": 3.9300000000000004e-07, "loss": 0.0068, "step": 48036 }, { "epoch": 2.561017220237778, "grad_norm": 2.726090669631958, "learning_rate": 3.928000000000001e-07, "loss": 0.0129, "step": 48037 }, { "epoch": 2.561070533667431, "grad_norm": 3.3477137088775635, "learning_rate": 3.9260000000000006e-07, "loss": 0.0162, "step": 48038 }, { "epoch": 2.561123847097084, "grad_norm": 4.40657377243042, "learning_rate": 3.924e-07, "loss": 0.0182, "step": 48039 }, { "epoch": 2.5611771605267366, "grad_norm": 3.2644951343536377, "learning_rate": 3.9220000000000003e-07, "loss": 0.0441, "step": 48040 }, { "epoch": 2.56123047395639, "grad_norm": 1.6173018217086792, "learning_rate": 3.92e-07, "loss": 0.0052, "step": 48041 }, { "epoch": 2.5612837873860426, "grad_norm": 0.4043349325656891, "learning_rate": 3.918e-07, "loss": 0.0031, "step": 48042 }, { "epoch": 2.5613371008156953, "grad_norm": 3.6984102725982666, "learning_rate": 3.9160000000000004e-07, "loss": 0.0214, "step": 48043 }, { "epoch": 2.5613904142453485, "grad_norm": 0.9134652614593506, "learning_rate": 3.914e-07, "loss": 0.0041, "step": 48044 }, { "epoch": 2.5614437276750013, "grad_norm": 5.538528919219971, "learning_rate": 3.9120000000000006e-07, "loss": 0.0111, "step": 48045 }, { "epoch": 2.561497041104654, "grad_norm": 1.5417579412460327, "learning_rate": 3.9100000000000005e-07, "loss": 0.0191, "step": 48046 }, { "epoch": 2.5615503545343072, "grad_norm": 4.697894096374512, "learning_rate": 3.908e-07, "loss": 0.0415, "step": 48047 }, { "epoch": 2.56160366796396, "grad_norm": 3.5060110092163086, "learning_rate": 3.906e-07, "loss": 0.0335, "step": 48048 }, { "epoch": 2.561656981393613, "grad_norm": 0.3833691477775574, "learning_rate": 3.904e-07, "loss": 0.0024, "step": 48049 }, { "epoch": 2.561710294823266, "grad_norm": 1.7922852039337158, "learning_rate": 3.9020000000000004e-07, "loss": 0.0182, "step": 48050 }, { "epoch": 2.561763608252919, "grad_norm": 2.5443480014801025, "learning_rate": 3.9e-07, "loss": 0.0175, "step": 48051 }, { "epoch": 2.561816921682572, "grad_norm": 2.287022829055786, "learning_rate": 3.8980000000000006e-07, "loss": 0.0202, "step": 48052 }, { "epoch": 2.5618702351122247, "grad_norm": 2.2084107398986816, "learning_rate": 3.8960000000000005e-07, "loss": 0.0334, "step": 48053 }, { "epoch": 2.561923548541878, "grad_norm": 0.8251879215240479, "learning_rate": 3.8940000000000003e-07, "loss": 0.0037, "step": 48054 }, { "epoch": 2.5619768619715306, "grad_norm": 0.7513657808303833, "learning_rate": 3.8920000000000007e-07, "loss": 0.0039, "step": 48055 }, { "epoch": 2.5620301754011834, "grad_norm": 1.3555808067321777, "learning_rate": 3.89e-07, "loss": 0.0044, "step": 48056 }, { "epoch": 2.5620834888308366, "grad_norm": 3.5508675575256348, "learning_rate": 3.888e-07, "loss": 0.0496, "step": 48057 }, { "epoch": 2.5621368022604893, "grad_norm": 1.706338882446289, "learning_rate": 3.886e-07, "loss": 0.0054, "step": 48058 }, { "epoch": 2.562190115690142, "grad_norm": 1.270740270614624, "learning_rate": 3.884e-07, "loss": 0.0191, "step": 48059 }, { "epoch": 2.5622434291197953, "grad_norm": 1.1340670585632324, "learning_rate": 3.8820000000000005e-07, "loss": 0.0123, "step": 48060 }, { "epoch": 2.562296742549448, "grad_norm": 1.1124601364135742, "learning_rate": 3.8800000000000003e-07, "loss": 0.0158, "step": 48061 }, { "epoch": 2.5623500559791013, "grad_norm": 2.157517194747925, "learning_rate": 3.8780000000000007e-07, "loss": 0.0238, "step": 48062 }, { "epoch": 2.562403369408754, "grad_norm": 2.157148599624634, "learning_rate": 3.8760000000000006e-07, "loss": 0.0192, "step": 48063 }, { "epoch": 2.562456682838407, "grad_norm": 1.0571542978286743, "learning_rate": 3.874e-07, "loss": 0.0242, "step": 48064 }, { "epoch": 2.56250999626806, "grad_norm": 3.1001675128936768, "learning_rate": 3.872e-07, "loss": 0.0058, "step": 48065 }, { "epoch": 2.5625633096977127, "grad_norm": 1.7086985111236572, "learning_rate": 3.87e-07, "loss": 0.0038, "step": 48066 }, { "epoch": 2.562616623127366, "grad_norm": 1.6405303478240967, "learning_rate": 3.8680000000000005e-07, "loss": 0.0071, "step": 48067 }, { "epoch": 2.5626699365570187, "grad_norm": 4.410006523132324, "learning_rate": 3.8660000000000003e-07, "loss": 0.0261, "step": 48068 }, { "epoch": 2.5627232499866714, "grad_norm": 2.1059317588806152, "learning_rate": 3.864e-07, "loss": 0.0062, "step": 48069 }, { "epoch": 2.5627765634163246, "grad_norm": 1.789363145828247, "learning_rate": 3.8620000000000006e-07, "loss": 0.0266, "step": 48070 }, { "epoch": 2.5628298768459774, "grad_norm": 2.685483694076538, "learning_rate": 3.8600000000000004e-07, "loss": 0.0164, "step": 48071 }, { "epoch": 2.56288319027563, "grad_norm": 1.9126135110855103, "learning_rate": 3.858000000000001e-07, "loss": 0.0207, "step": 48072 }, { "epoch": 2.5629365037052834, "grad_norm": 2.098419666290283, "learning_rate": 3.856e-07, "loss": 0.018, "step": 48073 }, { "epoch": 2.5629898171349366, "grad_norm": 1.6846237182617188, "learning_rate": 3.854e-07, "loss": 0.0138, "step": 48074 }, { "epoch": 2.5630431305645893, "grad_norm": 2.4242143630981445, "learning_rate": 3.8520000000000003e-07, "loss": 0.0292, "step": 48075 }, { "epoch": 2.563096443994242, "grad_norm": 2.881652593612671, "learning_rate": 3.85e-07, "loss": 0.0146, "step": 48076 }, { "epoch": 2.5631497574238953, "grad_norm": 1.2539454698562622, "learning_rate": 3.8480000000000006e-07, "loss": 0.0041, "step": 48077 }, { "epoch": 2.563203070853548, "grad_norm": 1.4435797929763794, "learning_rate": 3.8460000000000004e-07, "loss": 0.0131, "step": 48078 }, { "epoch": 2.563256384283201, "grad_norm": 1.2742300033569336, "learning_rate": 3.8440000000000003e-07, "loss": 0.0247, "step": 48079 }, { "epoch": 2.563309697712854, "grad_norm": 0.3874853551387787, "learning_rate": 3.8420000000000007e-07, "loss": 0.002, "step": 48080 }, { "epoch": 2.5633630111425068, "grad_norm": 1.0124610662460327, "learning_rate": 3.84e-07, "loss": 0.0225, "step": 48081 }, { "epoch": 2.5634163245721595, "grad_norm": 1.6619269847869873, "learning_rate": 3.838e-07, "loss": 0.0044, "step": 48082 }, { "epoch": 2.5634696380018127, "grad_norm": 1.649740219116211, "learning_rate": 3.836e-07, "loss": 0.0155, "step": 48083 }, { "epoch": 2.5635229514314655, "grad_norm": 0.4364335536956787, "learning_rate": 3.834e-07, "loss": 0.0033, "step": 48084 }, { "epoch": 2.5635762648611187, "grad_norm": 0.734277606010437, "learning_rate": 3.8320000000000004e-07, "loss": 0.0031, "step": 48085 }, { "epoch": 2.5636295782907714, "grad_norm": 2.065688133239746, "learning_rate": 3.8300000000000003e-07, "loss": 0.0182, "step": 48086 }, { "epoch": 2.5636828917204246, "grad_norm": 2.0300559997558594, "learning_rate": 3.8280000000000007e-07, "loss": 0.0153, "step": 48087 }, { "epoch": 2.5637362051500774, "grad_norm": 1.2366125583648682, "learning_rate": 3.8260000000000005e-07, "loss": 0.0231, "step": 48088 }, { "epoch": 2.56378951857973, "grad_norm": 4.847385406494141, "learning_rate": 3.824000000000001e-07, "loss": 0.0178, "step": 48089 }, { "epoch": 2.5638428320093833, "grad_norm": 3.3428852558135986, "learning_rate": 3.822e-07, "loss": 0.0706, "step": 48090 }, { "epoch": 2.563896145439036, "grad_norm": 1.9938050508499146, "learning_rate": 3.82e-07, "loss": 0.0137, "step": 48091 }, { "epoch": 2.563949458868689, "grad_norm": 1.1619572639465332, "learning_rate": 3.8180000000000004e-07, "loss": 0.0058, "step": 48092 }, { "epoch": 2.564002772298342, "grad_norm": 1.4028408527374268, "learning_rate": 3.8160000000000003e-07, "loss": 0.0051, "step": 48093 }, { "epoch": 2.564056085727995, "grad_norm": 2.661450147628784, "learning_rate": 3.814e-07, "loss": 0.0224, "step": 48094 }, { "epoch": 2.5641093991576476, "grad_norm": 1.5854369401931763, "learning_rate": 3.8120000000000005e-07, "loss": 0.0053, "step": 48095 }, { "epoch": 2.5641627125873008, "grad_norm": 1.098229169845581, "learning_rate": 3.8100000000000004e-07, "loss": 0.003, "step": 48096 }, { "epoch": 2.5642160260169535, "grad_norm": 3.8161838054656982, "learning_rate": 3.808000000000001e-07, "loss": 0.0205, "step": 48097 }, { "epoch": 2.5642693394466067, "grad_norm": 5.124579906463623, "learning_rate": 3.806e-07, "loss": 0.0203, "step": 48098 }, { "epoch": 2.5643226528762595, "grad_norm": 2.1189396381378174, "learning_rate": 3.804e-07, "loss": 0.0079, "step": 48099 }, { "epoch": 2.5643759663059127, "grad_norm": 1.210033893585205, "learning_rate": 3.8020000000000003e-07, "loss": 0.0204, "step": 48100 }, { "epoch": 2.5644292797355654, "grad_norm": 1.1330455541610718, "learning_rate": 3.8e-07, "loss": 0.0139, "step": 48101 }, { "epoch": 2.564482593165218, "grad_norm": 3.4879415035247803, "learning_rate": 3.7980000000000005e-07, "loss": 0.0352, "step": 48102 }, { "epoch": 2.5645359065948714, "grad_norm": 3.3695225715637207, "learning_rate": 3.7960000000000004e-07, "loss": 0.0189, "step": 48103 }, { "epoch": 2.564589220024524, "grad_norm": 2.323529005050659, "learning_rate": 3.794e-07, "loss": 0.008, "step": 48104 }, { "epoch": 2.564642533454177, "grad_norm": 1.1347401142120361, "learning_rate": 3.7920000000000006e-07, "loss": 0.011, "step": 48105 }, { "epoch": 2.56469584688383, "grad_norm": 1.1137868165969849, "learning_rate": 3.7900000000000005e-07, "loss": 0.0049, "step": 48106 }, { "epoch": 2.564749160313483, "grad_norm": 4.144832134246826, "learning_rate": 3.788e-07, "loss": 0.0331, "step": 48107 }, { "epoch": 2.5648024737431356, "grad_norm": 0.9222760796546936, "learning_rate": 3.786e-07, "loss": 0.0039, "step": 48108 }, { "epoch": 2.564855787172789, "grad_norm": 3.018543243408203, "learning_rate": 3.784e-07, "loss": 0.0153, "step": 48109 }, { "epoch": 2.5649091006024416, "grad_norm": 3.421815872192383, "learning_rate": 3.7820000000000004e-07, "loss": 0.0206, "step": 48110 }, { "epoch": 2.564962414032095, "grad_norm": 0.4691932797431946, "learning_rate": 3.78e-07, "loss": 0.0025, "step": 48111 }, { "epoch": 2.5650157274617476, "grad_norm": 2.7757315635681152, "learning_rate": 3.7780000000000006e-07, "loss": 0.0205, "step": 48112 }, { "epoch": 2.5650690408914008, "grad_norm": 2.606884479522705, "learning_rate": 3.7760000000000005e-07, "loss": 0.0269, "step": 48113 }, { "epoch": 2.5651223543210535, "grad_norm": 1.7866735458374023, "learning_rate": 3.774000000000001e-07, "loss": 0.006, "step": 48114 }, { "epoch": 2.5651756677507063, "grad_norm": 1.5513030290603638, "learning_rate": 3.772e-07, "loss": 0.0047, "step": 48115 }, { "epoch": 2.5652289811803595, "grad_norm": 1.521532654762268, "learning_rate": 3.77e-07, "loss": 0.0052, "step": 48116 }, { "epoch": 2.5652822946100122, "grad_norm": 0.5114048719406128, "learning_rate": 3.7680000000000004e-07, "loss": 0.0021, "step": 48117 }, { "epoch": 2.565335608039665, "grad_norm": 2.0129261016845703, "learning_rate": 3.766e-07, "loss": 0.0037, "step": 48118 }, { "epoch": 2.565388921469318, "grad_norm": 1.4542815685272217, "learning_rate": 3.764e-07, "loss": 0.0196, "step": 48119 }, { "epoch": 2.565442234898971, "grad_norm": 0.5636411905288696, "learning_rate": 3.7620000000000005e-07, "loss": 0.0041, "step": 48120 }, { "epoch": 2.565495548328624, "grad_norm": 2.683201313018799, "learning_rate": 3.7600000000000003e-07, "loss": 0.0454, "step": 48121 }, { "epoch": 2.565548861758277, "grad_norm": 1.0220389366149902, "learning_rate": 3.7580000000000007e-07, "loss": 0.0052, "step": 48122 }, { "epoch": 2.56560217518793, "grad_norm": 4.056957244873047, "learning_rate": 3.7560000000000005e-07, "loss": 0.0234, "step": 48123 }, { "epoch": 2.565655488617583, "grad_norm": 1.4334907531738281, "learning_rate": 3.754e-07, "loss": 0.0154, "step": 48124 }, { "epoch": 2.5657088020472356, "grad_norm": 0.5843020677566528, "learning_rate": 3.752e-07, "loss": 0.0037, "step": 48125 }, { "epoch": 2.565762115476889, "grad_norm": 1.1819020509719849, "learning_rate": 3.75e-07, "loss": 0.005, "step": 48126 }, { "epoch": 2.5658154289065416, "grad_norm": 0.8199328780174255, "learning_rate": 3.7480000000000005e-07, "loss": 0.0039, "step": 48127 }, { "epoch": 2.5658687423361943, "grad_norm": 1.6234021186828613, "learning_rate": 3.7460000000000003e-07, "loss": 0.0102, "step": 48128 }, { "epoch": 2.5659220557658475, "grad_norm": 2.0470123291015625, "learning_rate": 3.744e-07, "loss": 0.0133, "step": 48129 }, { "epoch": 2.5659753691955003, "grad_norm": 1.3841335773468018, "learning_rate": 3.7420000000000006e-07, "loss": 0.0087, "step": 48130 }, { "epoch": 2.566028682625153, "grad_norm": 0.6149640679359436, "learning_rate": 3.7400000000000004e-07, "loss": 0.003, "step": 48131 }, { "epoch": 2.5660819960548062, "grad_norm": 2.2516441345214844, "learning_rate": 3.7379999999999997e-07, "loss": 0.0346, "step": 48132 }, { "epoch": 2.566135309484459, "grad_norm": 1.7020039558410645, "learning_rate": 3.736e-07, "loss": 0.0059, "step": 48133 }, { "epoch": 2.566188622914112, "grad_norm": 1.0818654298782349, "learning_rate": 3.734e-07, "loss": 0.0058, "step": 48134 }, { "epoch": 2.566241936343765, "grad_norm": 1.4319096803665161, "learning_rate": 3.7320000000000003e-07, "loss": 0.0203, "step": 48135 }, { "epoch": 2.566295249773418, "grad_norm": 4.149701118469238, "learning_rate": 3.73e-07, "loss": 0.0333, "step": 48136 }, { "epoch": 2.566348563203071, "grad_norm": 2.5902884006500244, "learning_rate": 3.7280000000000006e-07, "loss": 0.0117, "step": 48137 }, { "epoch": 2.5664018766327237, "grad_norm": 1.4374734163284302, "learning_rate": 3.7260000000000004e-07, "loss": 0.0134, "step": 48138 }, { "epoch": 2.566455190062377, "grad_norm": 2.2249886989593506, "learning_rate": 3.724000000000001e-07, "loss": 0.0102, "step": 48139 }, { "epoch": 2.5665085034920296, "grad_norm": 1.6395436525344849, "learning_rate": 3.7220000000000006e-07, "loss": 0.0128, "step": 48140 }, { "epoch": 2.5665618169216824, "grad_norm": 1.6591095924377441, "learning_rate": 3.72e-07, "loss": 0.0048, "step": 48141 }, { "epoch": 2.5666151303513356, "grad_norm": 0.1127200722694397, "learning_rate": 3.7180000000000003e-07, "loss": 0.0011, "step": 48142 }, { "epoch": 2.5666684437809884, "grad_norm": 2.906775712966919, "learning_rate": 3.716e-07, "loss": 0.0196, "step": 48143 }, { "epoch": 2.566721757210641, "grad_norm": 0.29480794072151184, "learning_rate": 3.714e-07, "loss": 0.0025, "step": 48144 }, { "epoch": 2.5667750706402943, "grad_norm": 1.5431925058364868, "learning_rate": 3.7120000000000004e-07, "loss": 0.0125, "step": 48145 }, { "epoch": 2.566828384069947, "grad_norm": 0.4136243164539337, "learning_rate": 3.7100000000000003e-07, "loss": 0.0024, "step": 48146 }, { "epoch": 2.5668816974996003, "grad_norm": 1.0446546077728271, "learning_rate": 3.7080000000000006e-07, "loss": 0.0047, "step": 48147 }, { "epoch": 2.566935010929253, "grad_norm": 1.4782480001449585, "learning_rate": 3.7060000000000005e-07, "loss": 0.0057, "step": 48148 }, { "epoch": 2.5669883243589062, "grad_norm": 1.9625164270401, "learning_rate": 3.704e-07, "loss": 0.0045, "step": 48149 }, { "epoch": 2.567041637788559, "grad_norm": 2.562525510787964, "learning_rate": 3.702e-07, "loss": 0.0127, "step": 48150 }, { "epoch": 2.5670949512182117, "grad_norm": 1.2368992567062378, "learning_rate": 3.7e-07, "loss": 0.0048, "step": 48151 }, { "epoch": 2.567148264647865, "grad_norm": 0.8327459692955017, "learning_rate": 3.6980000000000004e-07, "loss": 0.0288, "step": 48152 }, { "epoch": 2.5672015780775177, "grad_norm": 2.186685562133789, "learning_rate": 3.6960000000000003e-07, "loss": 0.0165, "step": 48153 }, { "epoch": 2.5672548915071705, "grad_norm": 4.197482585906982, "learning_rate": 3.694e-07, "loss": 0.0174, "step": 48154 }, { "epoch": 2.5673082049368237, "grad_norm": 1.6707154512405396, "learning_rate": 3.6920000000000005e-07, "loss": 0.0264, "step": 48155 }, { "epoch": 2.5673615183664764, "grad_norm": 1.0270220041275024, "learning_rate": 3.6900000000000004e-07, "loss": 0.0046, "step": 48156 }, { "epoch": 2.567414831796129, "grad_norm": 2.4838244915008545, "learning_rate": 3.6880000000000007e-07, "loss": 0.0089, "step": 48157 }, { "epoch": 2.5674681452257824, "grad_norm": 3.130549192428589, "learning_rate": 3.686e-07, "loss": 0.0403, "step": 48158 }, { "epoch": 2.5675214586554356, "grad_norm": 3.8881278038024902, "learning_rate": 3.684e-07, "loss": 0.01, "step": 48159 }, { "epoch": 2.5675747720850883, "grad_norm": 1.2213698625564575, "learning_rate": 3.6820000000000003e-07, "loss": 0.0174, "step": 48160 }, { "epoch": 2.567628085514741, "grad_norm": 0.812633752822876, "learning_rate": 3.68e-07, "loss": 0.0034, "step": 48161 }, { "epoch": 2.5676813989443943, "grad_norm": 2.0234386920928955, "learning_rate": 3.6780000000000005e-07, "loss": 0.0174, "step": 48162 }, { "epoch": 2.567734712374047, "grad_norm": 3.315462589263916, "learning_rate": 3.6760000000000004e-07, "loss": 0.0255, "step": 48163 }, { "epoch": 2.5677880258037, "grad_norm": 1.8491061925888062, "learning_rate": 3.674000000000001e-07, "loss": 0.0122, "step": 48164 }, { "epoch": 2.567841339233353, "grad_norm": 2.592491388320923, "learning_rate": 3.6720000000000006e-07, "loss": 0.0191, "step": 48165 }, { "epoch": 2.5678946526630058, "grad_norm": 1.074908971786499, "learning_rate": 3.6700000000000004e-07, "loss": 0.0051, "step": 48166 }, { "epoch": 2.5679479660926585, "grad_norm": 3.0008795261383057, "learning_rate": 3.6680000000000003e-07, "loss": 0.0395, "step": 48167 }, { "epoch": 2.5680012795223117, "grad_norm": 1.65403413772583, "learning_rate": 3.666e-07, "loss": 0.0184, "step": 48168 }, { "epoch": 2.5680545929519645, "grad_norm": 2.1124820709228516, "learning_rate": 3.664e-07, "loss": 0.0408, "step": 48169 }, { "epoch": 2.5681079063816177, "grad_norm": 1.5079692602157593, "learning_rate": 3.6620000000000004e-07, "loss": 0.0044, "step": 48170 }, { "epoch": 2.5681612198112704, "grad_norm": 2.701469659805298, "learning_rate": 3.66e-07, "loss": 0.0241, "step": 48171 }, { "epoch": 2.5682145332409236, "grad_norm": 2.4163455963134766, "learning_rate": 3.6580000000000006e-07, "loss": 0.0159, "step": 48172 }, { "epoch": 2.5682678466705764, "grad_norm": 0.9682287573814392, "learning_rate": 3.6560000000000004e-07, "loss": 0.0055, "step": 48173 }, { "epoch": 2.568321160100229, "grad_norm": 2.05122447013855, "learning_rate": 3.654000000000001e-07, "loss": 0.0049, "step": 48174 }, { "epoch": 2.5683744735298824, "grad_norm": 1.3976693153381348, "learning_rate": 3.652e-07, "loss": 0.0245, "step": 48175 }, { "epoch": 2.568427786959535, "grad_norm": 2.1874587535858154, "learning_rate": 3.65e-07, "loss": 0.0149, "step": 48176 }, { "epoch": 2.568481100389188, "grad_norm": 0.7744835019111633, "learning_rate": 3.6480000000000004e-07, "loss": 0.0047, "step": 48177 }, { "epoch": 2.568534413818841, "grad_norm": 0.5941694378852844, "learning_rate": 3.646e-07, "loss": 0.0035, "step": 48178 }, { "epoch": 2.568587727248494, "grad_norm": 1.0412843227386475, "learning_rate": 3.644e-07, "loss": 0.0039, "step": 48179 }, { "epoch": 2.5686410406781466, "grad_norm": 2.9896979331970215, "learning_rate": 3.6420000000000005e-07, "loss": 0.0119, "step": 48180 }, { "epoch": 2.5686943541078, "grad_norm": 0.2932267487049103, "learning_rate": 3.6400000000000003e-07, "loss": 0.0021, "step": 48181 }, { "epoch": 2.5687476675374525, "grad_norm": 1.2785102128982544, "learning_rate": 3.6380000000000007e-07, "loss": 0.0088, "step": 48182 }, { "epoch": 2.5688009809671057, "grad_norm": 1.1163626909255981, "learning_rate": 3.6360000000000005e-07, "loss": 0.0054, "step": 48183 }, { "epoch": 2.5688542943967585, "grad_norm": 0.8466250896453857, "learning_rate": 3.634e-07, "loss": 0.0046, "step": 48184 }, { "epoch": 2.5689076078264117, "grad_norm": 0.8495730757713318, "learning_rate": 3.632e-07, "loss": 0.0048, "step": 48185 }, { "epoch": 2.5689609212560645, "grad_norm": 1.7497411966323853, "learning_rate": 3.63e-07, "loss": 0.0063, "step": 48186 }, { "epoch": 2.569014234685717, "grad_norm": 1.8752232789993286, "learning_rate": 3.6280000000000005e-07, "loss": 0.0084, "step": 48187 }, { "epoch": 2.5690675481153704, "grad_norm": 1.7570703029632568, "learning_rate": 3.6260000000000003e-07, "loss": 0.0055, "step": 48188 }, { "epoch": 2.569120861545023, "grad_norm": 0.547116756439209, "learning_rate": 3.6240000000000007e-07, "loss": 0.0034, "step": 48189 }, { "epoch": 2.569174174974676, "grad_norm": 1.6633589267730713, "learning_rate": 3.6220000000000005e-07, "loss": 0.0095, "step": 48190 }, { "epoch": 2.569227488404329, "grad_norm": 0.32151171565055847, "learning_rate": 3.6200000000000004e-07, "loss": 0.0026, "step": 48191 }, { "epoch": 2.569280801833982, "grad_norm": 1.8685482740402222, "learning_rate": 3.618e-07, "loss": 0.0144, "step": 48192 }, { "epoch": 2.5693341152636346, "grad_norm": 1.7149548530578613, "learning_rate": 3.616e-07, "loss": 0.0052, "step": 48193 }, { "epoch": 2.569387428693288, "grad_norm": 3.9606242179870605, "learning_rate": 3.614e-07, "loss": 0.0317, "step": 48194 }, { "epoch": 2.569440742122941, "grad_norm": 0.7570644021034241, "learning_rate": 3.6120000000000003e-07, "loss": 0.0037, "step": 48195 }, { "epoch": 2.569494055552594, "grad_norm": 3.0042996406555176, "learning_rate": 3.61e-07, "loss": 0.0293, "step": 48196 }, { "epoch": 2.5695473689822466, "grad_norm": 3.032712936401367, "learning_rate": 3.6080000000000005e-07, "loss": 0.01, "step": 48197 }, { "epoch": 2.5696006824118998, "grad_norm": 1.203971028327942, "learning_rate": 3.6060000000000004e-07, "loss": 0.0056, "step": 48198 }, { "epoch": 2.5696539958415525, "grad_norm": 4.117821216583252, "learning_rate": 3.604000000000001e-07, "loss": 0.0224, "step": 48199 }, { "epoch": 2.5697073092712053, "grad_norm": 0.78094881772995, "learning_rate": 3.6020000000000006e-07, "loss": 0.0037, "step": 48200 }, { "epoch": 2.5697606227008585, "grad_norm": 0.9569716453552246, "learning_rate": 3.6e-07, "loss": 0.0047, "step": 48201 }, { "epoch": 2.5698139361305112, "grad_norm": 2.067619800567627, "learning_rate": 3.5980000000000003e-07, "loss": 0.0223, "step": 48202 }, { "epoch": 2.569867249560164, "grad_norm": 3.021934747695923, "learning_rate": 3.596e-07, "loss": 0.0421, "step": 48203 }, { "epoch": 2.569920562989817, "grad_norm": 1.349708080291748, "learning_rate": 3.594e-07, "loss": 0.0048, "step": 48204 }, { "epoch": 2.56997387641947, "grad_norm": 1.9285908937454224, "learning_rate": 3.5920000000000004e-07, "loss": 0.024, "step": 48205 }, { "epoch": 2.570027189849123, "grad_norm": 0.5998952984809875, "learning_rate": 3.59e-07, "loss": 0.0035, "step": 48206 }, { "epoch": 2.570080503278776, "grad_norm": 0.5463844537734985, "learning_rate": 3.5880000000000006e-07, "loss": 0.0025, "step": 48207 }, { "epoch": 2.570133816708429, "grad_norm": 0.562846839427948, "learning_rate": 3.5860000000000005e-07, "loss": 0.0022, "step": 48208 }, { "epoch": 2.570187130138082, "grad_norm": 1.2512527704238892, "learning_rate": 3.584e-07, "loss": 0.004, "step": 48209 }, { "epoch": 2.5702404435677346, "grad_norm": 3.323910713195801, "learning_rate": 3.582e-07, "loss": 0.027, "step": 48210 }, { "epoch": 2.570293756997388, "grad_norm": 2.063106060028076, "learning_rate": 3.58e-07, "loss": 0.018, "step": 48211 }, { "epoch": 2.5703470704270406, "grad_norm": 0.9464350342750549, "learning_rate": 3.5780000000000004e-07, "loss": 0.0034, "step": 48212 }, { "epoch": 2.5704003838566933, "grad_norm": 0.23587529361248016, "learning_rate": 3.576e-07, "loss": 0.0019, "step": 48213 }, { "epoch": 2.5704536972863465, "grad_norm": 2.4778530597686768, "learning_rate": 3.5740000000000006e-07, "loss": 0.0092, "step": 48214 }, { "epoch": 2.5705070107159993, "grad_norm": 1.568571925163269, "learning_rate": 3.5720000000000005e-07, "loss": 0.0362, "step": 48215 }, { "epoch": 2.570560324145652, "grad_norm": 0.1713334023952484, "learning_rate": 3.5700000000000003e-07, "loss": 0.0015, "step": 48216 }, { "epoch": 2.5706136375753053, "grad_norm": 4.357478618621826, "learning_rate": 3.5680000000000007e-07, "loss": 0.0266, "step": 48217 }, { "epoch": 2.570666951004958, "grad_norm": 0.5551767349243164, "learning_rate": 3.566e-07, "loss": 0.0034, "step": 48218 }, { "epoch": 2.570720264434611, "grad_norm": 0.3325323462486267, "learning_rate": 3.564e-07, "loss": 0.0027, "step": 48219 }, { "epoch": 2.570773577864264, "grad_norm": 3.722193479537964, "learning_rate": 3.562e-07, "loss": 0.0079, "step": 48220 }, { "epoch": 2.570826891293917, "grad_norm": 1.66444993019104, "learning_rate": 3.56e-07, "loss": 0.0139, "step": 48221 }, { "epoch": 2.57088020472357, "grad_norm": 1.8280525207519531, "learning_rate": 3.5580000000000005e-07, "loss": 0.0135, "step": 48222 }, { "epoch": 2.5709335181532227, "grad_norm": 0.7790080308914185, "learning_rate": 3.5560000000000003e-07, "loss": 0.0036, "step": 48223 }, { "epoch": 2.570986831582876, "grad_norm": 0.46815457940101624, "learning_rate": 3.5540000000000007e-07, "loss": 0.0023, "step": 48224 }, { "epoch": 2.5710401450125286, "grad_norm": 0.8464075922966003, "learning_rate": 3.5520000000000006e-07, "loss": 0.0039, "step": 48225 }, { "epoch": 2.5710934584421814, "grad_norm": 0.41603419184684753, "learning_rate": 3.55e-07, "loss": 0.003, "step": 48226 }, { "epoch": 2.5711467718718346, "grad_norm": 0.45962846279144287, "learning_rate": 3.5480000000000003e-07, "loss": 0.0033, "step": 48227 }, { "epoch": 2.5712000853014874, "grad_norm": 2.4714245796203613, "learning_rate": 3.546e-07, "loss": 0.0448, "step": 48228 }, { "epoch": 2.57125339873114, "grad_norm": 1.9750112295150757, "learning_rate": 3.544e-07, "loss": 0.0093, "step": 48229 }, { "epoch": 2.5713067121607933, "grad_norm": 3.527859926223755, "learning_rate": 3.5420000000000004e-07, "loss": 0.0154, "step": 48230 }, { "epoch": 2.571360025590446, "grad_norm": 2.584505796432495, "learning_rate": 3.54e-07, "loss": 0.0225, "step": 48231 }, { "epoch": 2.5714133390200993, "grad_norm": 0.42667055130004883, "learning_rate": 3.5380000000000006e-07, "loss": 0.0027, "step": 48232 }, { "epoch": 2.571466652449752, "grad_norm": 0.7278598546981812, "learning_rate": 3.5360000000000004e-07, "loss": 0.0033, "step": 48233 }, { "epoch": 2.5715199658794052, "grad_norm": 1.1405231952667236, "learning_rate": 3.534000000000001e-07, "loss": 0.028, "step": 48234 }, { "epoch": 2.571573279309058, "grad_norm": 1.7765663862228394, "learning_rate": 3.532e-07, "loss": 0.0263, "step": 48235 }, { "epoch": 2.5716265927387107, "grad_norm": 1.4928312301635742, "learning_rate": 3.53e-07, "loss": 0.0038, "step": 48236 }, { "epoch": 2.571679906168364, "grad_norm": 0.8117392063140869, "learning_rate": 3.5280000000000004e-07, "loss": 0.0047, "step": 48237 }, { "epoch": 2.5717332195980167, "grad_norm": 0.8327388167381287, "learning_rate": 3.526e-07, "loss": 0.0039, "step": 48238 }, { "epoch": 2.5717865330276695, "grad_norm": 0.4738354980945587, "learning_rate": 3.5240000000000006e-07, "loss": 0.0022, "step": 48239 }, { "epoch": 2.5718398464573227, "grad_norm": 2.0561208724975586, "learning_rate": 3.5220000000000004e-07, "loss": 0.0173, "step": 48240 }, { "epoch": 2.5718931598869754, "grad_norm": 3.329364061355591, "learning_rate": 3.5200000000000003e-07, "loss": 0.0135, "step": 48241 }, { "epoch": 2.5719464733166286, "grad_norm": 1.4467324018478394, "learning_rate": 3.5180000000000007e-07, "loss": 0.0055, "step": 48242 }, { "epoch": 2.5719997867462814, "grad_norm": 0.35840046405792236, "learning_rate": 3.516e-07, "loss": 0.002, "step": 48243 }, { "epoch": 2.5720531001759346, "grad_norm": 1.030481219291687, "learning_rate": 3.514e-07, "loss": 0.0046, "step": 48244 }, { "epoch": 2.5721064136055873, "grad_norm": 2.0878775119781494, "learning_rate": 3.512e-07, "loss": 0.0077, "step": 48245 }, { "epoch": 2.57215972703524, "grad_norm": 0.8666468262672424, "learning_rate": 3.51e-07, "loss": 0.0025, "step": 48246 }, { "epoch": 2.5722130404648933, "grad_norm": 1.7244248390197754, "learning_rate": 3.5080000000000004e-07, "loss": 0.0044, "step": 48247 }, { "epoch": 2.572266353894546, "grad_norm": 2.285083293914795, "learning_rate": 3.5060000000000003e-07, "loss": 0.0127, "step": 48248 }, { "epoch": 2.572319667324199, "grad_norm": 1.0047701597213745, "learning_rate": 3.5040000000000007e-07, "loss": 0.0136, "step": 48249 }, { "epoch": 2.572372980753852, "grad_norm": 1.1423877477645874, "learning_rate": 3.5020000000000005e-07, "loss": 0.0045, "step": 48250 }, { "epoch": 2.5724262941835048, "grad_norm": 1.5748646259307861, "learning_rate": 3.5000000000000004e-07, "loss": 0.0049, "step": 48251 }, { "epoch": 2.5724796076131575, "grad_norm": 1.1798655986785889, "learning_rate": 3.498e-07, "loss": 0.0118, "step": 48252 }, { "epoch": 2.5725329210428107, "grad_norm": 0.5348600149154663, "learning_rate": 3.496e-07, "loss": 0.0032, "step": 48253 }, { "epoch": 2.5725862344724635, "grad_norm": 1.1863596439361572, "learning_rate": 3.494e-07, "loss": 0.0057, "step": 48254 }, { "epoch": 2.5726395479021167, "grad_norm": 1.8677889108657837, "learning_rate": 3.4920000000000003e-07, "loss": 0.017, "step": 48255 }, { "epoch": 2.5726928613317694, "grad_norm": 1.3198307752609253, "learning_rate": 3.49e-07, "loss": 0.0047, "step": 48256 }, { "epoch": 2.5727461747614226, "grad_norm": 2.2963194847106934, "learning_rate": 3.4880000000000005e-07, "loss": 0.0051, "step": 48257 }, { "epoch": 2.5727994881910754, "grad_norm": 0.730066180229187, "learning_rate": 3.4860000000000004e-07, "loss": 0.0024, "step": 48258 }, { "epoch": 2.572852801620728, "grad_norm": 2.28210711479187, "learning_rate": 3.484000000000001e-07, "loss": 0.0092, "step": 48259 }, { "epoch": 2.5729061150503814, "grad_norm": 1.7324553728103638, "learning_rate": 3.482e-07, "loss": 0.0207, "step": 48260 }, { "epoch": 2.572959428480034, "grad_norm": 2.6175477504730225, "learning_rate": 3.48e-07, "loss": 0.0319, "step": 48261 }, { "epoch": 2.573012741909687, "grad_norm": 3.3868567943573, "learning_rate": 3.4780000000000003e-07, "loss": 0.0465, "step": 48262 }, { "epoch": 2.57306605533934, "grad_norm": 1.5616408586502075, "learning_rate": 3.476e-07, "loss": 0.0135, "step": 48263 }, { "epoch": 2.573119368768993, "grad_norm": 1.7193607091903687, "learning_rate": 3.4740000000000005e-07, "loss": 0.0159, "step": 48264 }, { "epoch": 2.5731726821986456, "grad_norm": 1.7181957960128784, "learning_rate": 3.4720000000000004e-07, "loss": 0.0172, "step": 48265 }, { "epoch": 2.573225995628299, "grad_norm": 0.39980700612068176, "learning_rate": 3.47e-07, "loss": 0.0024, "step": 48266 }, { "epoch": 2.5732793090579515, "grad_norm": 2.44980525970459, "learning_rate": 3.4680000000000006e-07, "loss": 0.0162, "step": 48267 }, { "epoch": 2.5733326224876047, "grad_norm": 2.23443865776062, "learning_rate": 3.4660000000000005e-07, "loss": 0.0178, "step": 48268 }, { "epoch": 2.5733859359172575, "grad_norm": 0.6669009923934937, "learning_rate": 3.464e-07, "loss": 0.0032, "step": 48269 }, { "epoch": 2.5734392493469107, "grad_norm": 4.186395168304443, "learning_rate": 3.462e-07, "loss": 0.0113, "step": 48270 }, { "epoch": 2.5734925627765635, "grad_norm": 2.9586269855499268, "learning_rate": 3.46e-07, "loss": 0.0271, "step": 48271 }, { "epoch": 2.573545876206216, "grad_norm": 2.9384634494781494, "learning_rate": 3.4580000000000004e-07, "loss": 0.0128, "step": 48272 }, { "epoch": 2.5735991896358694, "grad_norm": 0.921293318271637, "learning_rate": 3.456e-07, "loss": 0.0245, "step": 48273 }, { "epoch": 2.573652503065522, "grad_norm": 1.632109522819519, "learning_rate": 3.4540000000000006e-07, "loss": 0.0044, "step": 48274 }, { "epoch": 2.573705816495175, "grad_norm": 0.3707902431488037, "learning_rate": 3.4520000000000005e-07, "loss": 0.0027, "step": 48275 }, { "epoch": 2.573759129924828, "grad_norm": 3.2092158794403076, "learning_rate": 3.4500000000000003e-07, "loss": 0.0199, "step": 48276 }, { "epoch": 2.573812443354481, "grad_norm": 1.9716087579727173, "learning_rate": 3.448e-07, "loss": 0.0145, "step": 48277 }, { "epoch": 2.5738657567841337, "grad_norm": 0.4710506796836853, "learning_rate": 3.446e-07, "loss": 0.0023, "step": 48278 }, { "epoch": 2.573919070213787, "grad_norm": 2.1889138221740723, "learning_rate": 3.444e-07, "loss": 0.0238, "step": 48279 }, { "epoch": 2.57397238364344, "grad_norm": 0.9861194491386414, "learning_rate": 3.442e-07, "loss": 0.0037, "step": 48280 }, { "epoch": 2.574025697073093, "grad_norm": 2.2854490280151367, "learning_rate": 3.44e-07, "loss": 0.0107, "step": 48281 }, { "epoch": 2.5740790105027456, "grad_norm": 0.23689979314804077, "learning_rate": 3.4380000000000005e-07, "loss": 0.0021, "step": 48282 }, { "epoch": 2.5741323239323988, "grad_norm": 1.2076067924499512, "learning_rate": 3.4360000000000003e-07, "loss": 0.0069, "step": 48283 }, { "epoch": 2.5741856373620515, "grad_norm": 1.9467389583587646, "learning_rate": 3.4340000000000007e-07, "loss": 0.0375, "step": 48284 }, { "epoch": 2.5742389507917043, "grad_norm": 1.673540711402893, "learning_rate": 3.4320000000000006e-07, "loss": 0.0064, "step": 48285 }, { "epoch": 2.5742922642213575, "grad_norm": 2.7560112476348877, "learning_rate": 3.43e-07, "loss": 0.014, "step": 48286 }, { "epoch": 2.5743455776510102, "grad_norm": 2.2278153896331787, "learning_rate": 3.428e-07, "loss": 0.0386, "step": 48287 }, { "epoch": 2.574398891080663, "grad_norm": 1.1390511989593506, "learning_rate": 3.426e-07, "loss": 0.0114, "step": 48288 }, { "epoch": 2.574452204510316, "grad_norm": 0.7752126455307007, "learning_rate": 3.4240000000000005e-07, "loss": 0.0031, "step": 48289 }, { "epoch": 2.574505517939969, "grad_norm": 2.4970970153808594, "learning_rate": 3.4220000000000003e-07, "loss": 0.0148, "step": 48290 }, { "epoch": 2.574558831369622, "grad_norm": 3.129915714263916, "learning_rate": 3.42e-07, "loss": 0.0212, "step": 48291 }, { "epoch": 2.574612144799275, "grad_norm": 1.8541051149368286, "learning_rate": 3.4180000000000006e-07, "loss": 0.024, "step": 48292 }, { "epoch": 2.574665458228928, "grad_norm": 0.7869991064071655, "learning_rate": 3.4160000000000004e-07, "loss": 0.0037, "step": 48293 }, { "epoch": 2.574718771658581, "grad_norm": 0.35477542877197266, "learning_rate": 3.4139999999999997e-07, "loss": 0.0018, "step": 48294 }, { "epoch": 2.5747720850882336, "grad_norm": 1.7134546041488647, "learning_rate": 3.412e-07, "loss": 0.0137, "step": 48295 }, { "epoch": 2.574825398517887, "grad_norm": 1.608248233795166, "learning_rate": 3.41e-07, "loss": 0.0349, "step": 48296 }, { "epoch": 2.5748787119475396, "grad_norm": 3.5356504917144775, "learning_rate": 3.4080000000000003e-07, "loss": 0.0213, "step": 48297 }, { "epoch": 2.5749320253771923, "grad_norm": 2.5847220420837402, "learning_rate": 3.406e-07, "loss": 0.0144, "step": 48298 }, { "epoch": 2.5749853388068455, "grad_norm": 1.6345025300979614, "learning_rate": 3.4040000000000006e-07, "loss": 0.0188, "step": 48299 }, { "epoch": 2.5750386522364983, "grad_norm": 0.41852396726608276, "learning_rate": 3.4020000000000004e-07, "loss": 0.0031, "step": 48300 }, { "epoch": 2.575091965666151, "grad_norm": 2.5057551860809326, "learning_rate": 3.4000000000000003e-07, "loss": 0.0111, "step": 48301 }, { "epoch": 2.5751452790958043, "grad_norm": 3.191195011138916, "learning_rate": 3.3980000000000006e-07, "loss": 0.0482, "step": 48302 }, { "epoch": 2.575198592525457, "grad_norm": 3.6759414672851562, "learning_rate": 3.396e-07, "loss": 0.016, "step": 48303 }, { "epoch": 2.57525190595511, "grad_norm": 1.3271541595458984, "learning_rate": 3.394e-07, "loss": 0.0139, "step": 48304 }, { "epoch": 2.575305219384763, "grad_norm": 2.2130823135375977, "learning_rate": 3.392e-07, "loss": 0.0179, "step": 48305 }, { "epoch": 2.575358532814416, "grad_norm": 4.155598163604736, "learning_rate": 3.39e-07, "loss": 0.0087, "step": 48306 }, { "epoch": 2.575411846244069, "grad_norm": 2.034605026245117, "learning_rate": 3.3880000000000004e-07, "loss": 0.0292, "step": 48307 }, { "epoch": 2.5754651596737217, "grad_norm": 1.3235018253326416, "learning_rate": 3.3860000000000003e-07, "loss": 0.0194, "step": 48308 }, { "epoch": 2.575518473103375, "grad_norm": 2.0619957447052, "learning_rate": 3.3840000000000007e-07, "loss": 0.0162, "step": 48309 }, { "epoch": 2.5755717865330277, "grad_norm": 2.942077398300171, "learning_rate": 3.3820000000000005e-07, "loss": 0.0144, "step": 48310 }, { "epoch": 2.5756250999626804, "grad_norm": 1.605288028717041, "learning_rate": 3.38e-07, "loss": 0.0067, "step": 48311 }, { "epoch": 2.5756784133923336, "grad_norm": 1.4747637510299683, "learning_rate": 3.378e-07, "loss": 0.0262, "step": 48312 }, { "epoch": 2.5757317268219864, "grad_norm": 1.9842578172683716, "learning_rate": 3.376e-07, "loss": 0.0235, "step": 48313 }, { "epoch": 2.575785040251639, "grad_norm": 5.284940719604492, "learning_rate": 3.3740000000000004e-07, "loss": 0.0233, "step": 48314 }, { "epoch": 2.5758383536812923, "grad_norm": 1.739775538444519, "learning_rate": 3.3720000000000003e-07, "loss": 0.0049, "step": 48315 }, { "epoch": 2.575891667110945, "grad_norm": 0.7914032936096191, "learning_rate": 3.37e-07, "loss": 0.0037, "step": 48316 }, { "epoch": 2.5759449805405983, "grad_norm": 0.4157031774520874, "learning_rate": 3.3680000000000005e-07, "loss": 0.0026, "step": 48317 }, { "epoch": 2.575998293970251, "grad_norm": 1.757004737854004, "learning_rate": 3.3660000000000004e-07, "loss": 0.0173, "step": 48318 }, { "epoch": 2.5760516073999042, "grad_norm": 1.4438072443008423, "learning_rate": 3.364000000000001e-07, "loss": 0.018, "step": 48319 }, { "epoch": 2.576104920829557, "grad_norm": 4.395983695983887, "learning_rate": 3.362e-07, "loss": 0.013, "step": 48320 }, { "epoch": 2.5761582342592098, "grad_norm": 3.5342772006988525, "learning_rate": 3.36e-07, "loss": 0.0127, "step": 48321 }, { "epoch": 2.576211547688863, "grad_norm": 0.611584484577179, "learning_rate": 3.3580000000000003e-07, "loss": 0.0033, "step": 48322 }, { "epoch": 2.5762648611185157, "grad_norm": 1.4833261966705322, "learning_rate": 3.356e-07, "loss": 0.0214, "step": 48323 }, { "epoch": 2.5763181745481685, "grad_norm": 3.419541358947754, "learning_rate": 3.3540000000000005e-07, "loss": 0.0188, "step": 48324 }, { "epoch": 2.5763714879778217, "grad_norm": 0.47072479128837585, "learning_rate": 3.3520000000000004e-07, "loss": 0.0027, "step": 48325 }, { "epoch": 2.5764248014074744, "grad_norm": 2.107738971710205, "learning_rate": 3.35e-07, "loss": 0.0082, "step": 48326 }, { "epoch": 2.5764781148371276, "grad_norm": 1.442888855934143, "learning_rate": 3.3480000000000006e-07, "loss": 0.0296, "step": 48327 }, { "epoch": 2.5765314282667804, "grad_norm": 1.9607304334640503, "learning_rate": 3.346e-07, "loss": 0.0112, "step": 48328 }, { "epoch": 2.5765847416964336, "grad_norm": 2.3171679973602295, "learning_rate": 3.344e-07, "loss": 0.0269, "step": 48329 }, { "epoch": 2.5766380551260863, "grad_norm": 1.5834461450576782, "learning_rate": 3.342e-07, "loss": 0.0178, "step": 48330 }, { "epoch": 2.576691368555739, "grad_norm": 2.270747423171997, "learning_rate": 3.34e-07, "loss": 0.026, "step": 48331 }, { "epoch": 2.5767446819853923, "grad_norm": 0.29314154386520386, "learning_rate": 3.3380000000000004e-07, "loss": 0.002, "step": 48332 }, { "epoch": 2.576797995415045, "grad_norm": 1.7865995168685913, "learning_rate": 3.336e-07, "loss": 0.0052, "step": 48333 }, { "epoch": 2.576851308844698, "grad_norm": 2.3719682693481445, "learning_rate": 3.3340000000000006e-07, "loss": 0.0311, "step": 48334 }, { "epoch": 2.576904622274351, "grad_norm": 2.4698269367218018, "learning_rate": 3.3320000000000005e-07, "loss": 0.0375, "step": 48335 }, { "epoch": 2.5769579357040038, "grad_norm": 1.8216553926467896, "learning_rate": 3.330000000000001e-07, "loss": 0.0093, "step": 48336 }, { "epoch": 2.5770112491336565, "grad_norm": 3.894132614135742, "learning_rate": 3.328e-07, "loss": 0.0244, "step": 48337 }, { "epoch": 2.5770645625633097, "grad_norm": 7.602561950683594, "learning_rate": 3.326e-07, "loss": 0.012, "step": 48338 }, { "epoch": 2.5771178759929625, "grad_norm": 2.3594281673431396, "learning_rate": 3.3240000000000004e-07, "loss": 0.0246, "step": 48339 }, { "epoch": 2.5771711894226157, "grad_norm": 0.9031506776809692, "learning_rate": 3.322e-07, "loss": 0.0043, "step": 48340 }, { "epoch": 2.5772245028522685, "grad_norm": 0.46528688073158264, "learning_rate": 3.32e-07, "loss": 0.003, "step": 48341 }, { "epoch": 2.5772778162819217, "grad_norm": 0.9385698437690735, "learning_rate": 3.3180000000000005e-07, "loss": 0.0058, "step": 48342 }, { "epoch": 2.5773311297115744, "grad_norm": 0.45871272683143616, "learning_rate": 3.3160000000000003e-07, "loss": 0.0033, "step": 48343 }, { "epoch": 2.577384443141227, "grad_norm": 2.1942334175109863, "learning_rate": 3.3140000000000007e-07, "loss": 0.0096, "step": 48344 }, { "epoch": 2.5774377565708804, "grad_norm": 1.5339208841323853, "learning_rate": 3.312e-07, "loss": 0.0268, "step": 48345 }, { "epoch": 2.577491070000533, "grad_norm": 2.9727580547332764, "learning_rate": 3.31e-07, "loss": 0.0224, "step": 48346 }, { "epoch": 2.577544383430186, "grad_norm": 1.3136786222457886, "learning_rate": 3.308e-07, "loss": 0.0081, "step": 48347 }, { "epoch": 2.577597696859839, "grad_norm": 0.594822883605957, "learning_rate": 3.306e-07, "loss": 0.0038, "step": 48348 }, { "epoch": 2.577651010289492, "grad_norm": 3.5732131004333496, "learning_rate": 3.3040000000000005e-07, "loss": 0.0119, "step": 48349 }, { "epoch": 2.5777043237191446, "grad_norm": 1.0321873426437378, "learning_rate": 3.3020000000000003e-07, "loss": 0.0045, "step": 48350 }, { "epoch": 2.577757637148798, "grad_norm": 2.5664284229278564, "learning_rate": 3.3e-07, "loss": 0.0116, "step": 48351 }, { "epoch": 2.5778109505784506, "grad_norm": 0.3922238051891327, "learning_rate": 3.2980000000000005e-07, "loss": 0.0027, "step": 48352 }, { "epoch": 2.5778642640081038, "grad_norm": 1.987027883529663, "learning_rate": 3.2960000000000004e-07, "loss": 0.0127, "step": 48353 }, { "epoch": 2.5779175774377565, "grad_norm": 1.136177897453308, "learning_rate": 3.2939999999999997e-07, "loss": 0.0047, "step": 48354 }, { "epoch": 2.5779708908674097, "grad_norm": 1.6288625001907349, "learning_rate": 3.292e-07, "loss": 0.0057, "step": 48355 }, { "epoch": 2.5780242042970625, "grad_norm": 1.1305468082427979, "learning_rate": 3.29e-07, "loss": 0.0044, "step": 48356 }, { "epoch": 2.5780775177267152, "grad_norm": 0.6920828819274902, "learning_rate": 3.2880000000000003e-07, "loss": 0.0028, "step": 48357 }, { "epoch": 2.5781308311563684, "grad_norm": 3.053311586380005, "learning_rate": 3.286e-07, "loss": 0.0089, "step": 48358 }, { "epoch": 2.578184144586021, "grad_norm": 2.279418468475342, "learning_rate": 3.2840000000000006e-07, "loss": 0.0084, "step": 48359 }, { "epoch": 2.578237458015674, "grad_norm": 3.71390700340271, "learning_rate": 3.2820000000000004e-07, "loss": 0.0193, "step": 48360 }, { "epoch": 2.578290771445327, "grad_norm": 1.4120657444000244, "learning_rate": 3.280000000000001e-07, "loss": 0.0224, "step": 48361 }, { "epoch": 2.57834408487498, "grad_norm": 3.3210718631744385, "learning_rate": 3.278e-07, "loss": 0.0378, "step": 48362 }, { "epoch": 2.578397398304633, "grad_norm": 1.2315458059310913, "learning_rate": 3.276e-07, "loss": 0.019, "step": 48363 }, { "epoch": 2.578450711734286, "grad_norm": 1.0297210216522217, "learning_rate": 3.2740000000000003e-07, "loss": 0.016, "step": 48364 }, { "epoch": 2.578504025163939, "grad_norm": 1.2224427461624146, "learning_rate": 3.272e-07, "loss": 0.0064, "step": 48365 }, { "epoch": 2.578557338593592, "grad_norm": 0.7979171872138977, "learning_rate": 3.27e-07, "loss": 0.003, "step": 48366 }, { "epoch": 2.5786106520232446, "grad_norm": 3.5311312675476074, "learning_rate": 3.2680000000000004e-07, "loss": 0.0239, "step": 48367 }, { "epoch": 2.5786639654528978, "grad_norm": 1.2064177989959717, "learning_rate": 3.266e-07, "loss": 0.0189, "step": 48368 }, { "epoch": 2.5787172788825505, "grad_norm": 1.1589984893798828, "learning_rate": 3.2640000000000006e-07, "loss": 0.0171, "step": 48369 }, { "epoch": 2.5787705923122033, "grad_norm": 2.0489132404327393, "learning_rate": 3.2620000000000005e-07, "loss": 0.0067, "step": 48370 }, { "epoch": 2.5788239057418565, "grad_norm": 2.3494691848754883, "learning_rate": 3.26e-07, "loss": 0.0155, "step": 48371 }, { "epoch": 2.5788772191715092, "grad_norm": 0.6561926007270813, "learning_rate": 3.258e-07, "loss": 0.0037, "step": 48372 }, { "epoch": 2.578930532601162, "grad_norm": 0.36600261926651, "learning_rate": 3.256e-07, "loss": 0.0029, "step": 48373 }, { "epoch": 2.578983846030815, "grad_norm": 1.8347504138946533, "learning_rate": 3.2540000000000004e-07, "loss": 0.0199, "step": 48374 }, { "epoch": 2.579037159460468, "grad_norm": 2.0109665393829346, "learning_rate": 3.2520000000000003e-07, "loss": 0.0092, "step": 48375 }, { "epoch": 2.579090472890121, "grad_norm": 2.3563077449798584, "learning_rate": 3.25e-07, "loss": 0.015, "step": 48376 }, { "epoch": 2.579143786319774, "grad_norm": 2.470041275024414, "learning_rate": 3.2480000000000005e-07, "loss": 0.0073, "step": 48377 }, { "epoch": 2.579197099749427, "grad_norm": 1.7233585119247437, "learning_rate": 3.2460000000000003e-07, "loss": 0.0289, "step": 48378 }, { "epoch": 2.57925041317908, "grad_norm": 0.9702701568603516, "learning_rate": 3.2439999999999997e-07, "loss": 0.0045, "step": 48379 }, { "epoch": 2.5793037266087326, "grad_norm": 2.8523592948913574, "learning_rate": 3.242e-07, "loss": 0.0071, "step": 48380 }, { "epoch": 2.579357040038386, "grad_norm": 0.9726341366767883, "learning_rate": 3.24e-07, "loss": 0.0179, "step": 48381 }, { "epoch": 2.5794103534680386, "grad_norm": 1.1395939588546753, "learning_rate": 3.2380000000000003e-07, "loss": 0.0058, "step": 48382 }, { "epoch": 2.5794636668976914, "grad_norm": 1.270240068435669, "learning_rate": 3.236e-07, "loss": 0.0156, "step": 48383 }, { "epoch": 2.5795169803273446, "grad_norm": 0.5545084476470947, "learning_rate": 3.2340000000000005e-07, "loss": 0.0038, "step": 48384 }, { "epoch": 2.5795702937569973, "grad_norm": 1.9781838655471802, "learning_rate": 3.2320000000000004e-07, "loss": 0.0085, "step": 48385 }, { "epoch": 2.57962360718665, "grad_norm": 5.10848331451416, "learning_rate": 3.2300000000000007e-07, "loss": 0.022, "step": 48386 }, { "epoch": 2.5796769206163033, "grad_norm": 1.9213736057281494, "learning_rate": 3.2280000000000006e-07, "loss": 0.0361, "step": 48387 }, { "epoch": 2.579730234045956, "grad_norm": 0.47455981373786926, "learning_rate": 3.226e-07, "loss": 0.0033, "step": 48388 }, { "epoch": 2.5797835474756092, "grad_norm": 1.8513792753219604, "learning_rate": 3.2240000000000003e-07, "loss": 0.0048, "step": 48389 }, { "epoch": 2.579836860905262, "grad_norm": 0.6839842796325684, "learning_rate": 3.222e-07, "loss": 0.004, "step": 48390 }, { "epoch": 2.579890174334915, "grad_norm": 0.4945944547653198, "learning_rate": 3.22e-07, "loss": 0.0035, "step": 48391 }, { "epoch": 2.579943487764568, "grad_norm": 0.8211157917976379, "learning_rate": 3.2180000000000004e-07, "loss": 0.0055, "step": 48392 }, { "epoch": 2.5799968011942207, "grad_norm": 2.8530068397521973, "learning_rate": 3.216e-07, "loss": 0.0402, "step": 48393 }, { "epoch": 2.580050114623874, "grad_norm": 0.9026092290878296, "learning_rate": 3.2140000000000006e-07, "loss": 0.0039, "step": 48394 }, { "epoch": 2.5801034280535267, "grad_norm": 2.479924440383911, "learning_rate": 3.2120000000000004e-07, "loss": 0.0055, "step": 48395 }, { "epoch": 2.5801567414831794, "grad_norm": 0.7681522965431213, "learning_rate": 3.21e-07, "loss": 0.0034, "step": 48396 }, { "epoch": 2.5802100549128326, "grad_norm": 3.7931344509124756, "learning_rate": 3.208e-07, "loss": 0.0142, "step": 48397 }, { "epoch": 2.5802633683424854, "grad_norm": 0.34237051010131836, "learning_rate": 3.206e-07, "loss": 0.0024, "step": 48398 }, { "epoch": 2.580316681772138, "grad_norm": 1.5987493991851807, "learning_rate": 3.2040000000000004e-07, "loss": 0.0074, "step": 48399 }, { "epoch": 2.5803699952017913, "grad_norm": 4.789516925811768, "learning_rate": 3.202e-07, "loss": 0.028, "step": 48400 }, { "epoch": 2.5804233086314445, "grad_norm": 2.0395452976226807, "learning_rate": 3.2e-07, "loss": 0.024, "step": 48401 }, { "epoch": 2.5804766220610973, "grad_norm": 1.6883808374404907, "learning_rate": 3.1980000000000004e-07, "loss": 0.0256, "step": 48402 }, { "epoch": 2.58052993549075, "grad_norm": 2.1692373752593994, "learning_rate": 3.1960000000000003e-07, "loss": 0.0078, "step": 48403 }, { "epoch": 2.5805832489204032, "grad_norm": 0.3977440893650055, "learning_rate": 3.1940000000000007e-07, "loss": 0.0026, "step": 48404 }, { "epoch": 2.580636562350056, "grad_norm": 0.3955863118171692, "learning_rate": 3.192e-07, "loss": 0.0026, "step": 48405 }, { "epoch": 2.5806898757797088, "grad_norm": 4.057639122009277, "learning_rate": 3.19e-07, "loss": 0.056, "step": 48406 }, { "epoch": 2.580743189209362, "grad_norm": 1.928501009941101, "learning_rate": 3.188e-07, "loss": 0.0169, "step": 48407 }, { "epoch": 2.5807965026390147, "grad_norm": 3.3375532627105713, "learning_rate": 3.186e-07, "loss": 0.0107, "step": 48408 }, { "epoch": 2.5808498160686675, "grad_norm": 0.6975454092025757, "learning_rate": 3.1840000000000005e-07, "loss": 0.0127, "step": 48409 }, { "epoch": 2.5809031294983207, "grad_norm": 2.394857406616211, "learning_rate": 3.1820000000000003e-07, "loss": 0.0246, "step": 48410 }, { "epoch": 2.5809564429279734, "grad_norm": 0.5858908891677856, "learning_rate": 3.1800000000000007e-07, "loss": 0.0024, "step": 48411 }, { "epoch": 2.5810097563576266, "grad_norm": 1.2026125192642212, "learning_rate": 3.1780000000000005e-07, "loss": 0.0136, "step": 48412 }, { "epoch": 2.5810630697872794, "grad_norm": 1.1821484565734863, "learning_rate": 3.176e-07, "loss": 0.0108, "step": 48413 }, { "epoch": 2.5811163832169326, "grad_norm": 2.428499221801758, "learning_rate": 3.174e-07, "loss": 0.0124, "step": 48414 }, { "epoch": 2.5811696966465854, "grad_norm": 1.6390880346298218, "learning_rate": 3.172e-07, "loss": 0.0072, "step": 48415 }, { "epoch": 2.581223010076238, "grad_norm": 2.2324318885803223, "learning_rate": 3.17e-07, "loss": 0.0186, "step": 48416 }, { "epoch": 2.5812763235058913, "grad_norm": 2.827998161315918, "learning_rate": 3.1680000000000003e-07, "loss": 0.0323, "step": 48417 }, { "epoch": 2.581329636935544, "grad_norm": 2.164266586303711, "learning_rate": 3.166e-07, "loss": 0.0193, "step": 48418 }, { "epoch": 2.581382950365197, "grad_norm": 1.0741016864776611, "learning_rate": 3.1640000000000005e-07, "loss": 0.0058, "step": 48419 }, { "epoch": 2.58143626379485, "grad_norm": 1.7439206838607788, "learning_rate": 3.1620000000000004e-07, "loss": 0.0495, "step": 48420 }, { "epoch": 2.581489577224503, "grad_norm": 2.336794376373291, "learning_rate": 3.160000000000001e-07, "loss": 0.0077, "step": 48421 }, { "epoch": 2.5815428906541555, "grad_norm": 2.3539540767669678, "learning_rate": 3.158e-07, "loss": 0.0098, "step": 48422 }, { "epoch": 2.5815962040838087, "grad_norm": 0.6867042779922485, "learning_rate": 3.156e-07, "loss": 0.0031, "step": 48423 }, { "epoch": 2.5816495175134615, "grad_norm": 1.384049654006958, "learning_rate": 3.1540000000000003e-07, "loss": 0.0182, "step": 48424 }, { "epoch": 2.5817028309431147, "grad_norm": 3.866499662399292, "learning_rate": 3.152e-07, "loss": 0.055, "step": 48425 }, { "epoch": 2.5817561443727675, "grad_norm": 0.5982341766357422, "learning_rate": 3.15e-07, "loss": 0.0027, "step": 48426 }, { "epoch": 2.5818094578024207, "grad_norm": 0.7677086591720581, "learning_rate": 3.1480000000000004e-07, "loss": 0.0034, "step": 48427 }, { "epoch": 2.5818627712320734, "grad_norm": 5.2595534324646, "learning_rate": 3.146e-07, "loss": 0.0297, "step": 48428 }, { "epoch": 2.581916084661726, "grad_norm": 1.4543813467025757, "learning_rate": 3.1440000000000006e-07, "loss": 0.0208, "step": 48429 }, { "epoch": 2.5819693980913794, "grad_norm": 1.0728566646575928, "learning_rate": 3.142e-07, "loss": 0.0051, "step": 48430 }, { "epoch": 2.582022711521032, "grad_norm": 1.6260069608688354, "learning_rate": 3.14e-07, "loss": 0.0105, "step": 48431 }, { "epoch": 2.582076024950685, "grad_norm": 1.674338936805725, "learning_rate": 3.138e-07, "loss": 0.0156, "step": 48432 }, { "epoch": 2.582129338380338, "grad_norm": 2.4582626819610596, "learning_rate": 3.136e-07, "loss": 0.024, "step": 48433 }, { "epoch": 2.582182651809991, "grad_norm": 1.262990117073059, "learning_rate": 3.1340000000000004e-07, "loss": 0.0052, "step": 48434 }, { "epoch": 2.5822359652396436, "grad_norm": 1.7476786375045776, "learning_rate": 3.132e-07, "loss": 0.015, "step": 48435 }, { "epoch": 2.582289278669297, "grad_norm": 1.8439159393310547, "learning_rate": 3.1300000000000006e-07, "loss": 0.0073, "step": 48436 }, { "epoch": 2.5823425920989496, "grad_norm": 0.9717403054237366, "learning_rate": 3.1280000000000005e-07, "loss": 0.0059, "step": 48437 }, { "epoch": 2.5823959055286028, "grad_norm": 3.4234910011291504, "learning_rate": 3.1260000000000003e-07, "loss": 0.0251, "step": 48438 }, { "epoch": 2.5824492189582555, "grad_norm": 1.4046467542648315, "learning_rate": 3.124e-07, "loss": 0.0068, "step": 48439 }, { "epoch": 2.5825025323879087, "grad_norm": 2.6919500827789307, "learning_rate": 3.1220000000000006e-07, "loss": 0.0236, "step": 48440 }, { "epoch": 2.5825558458175615, "grad_norm": 1.2688151597976685, "learning_rate": 3.12e-07, "loss": 0.0042, "step": 48441 }, { "epoch": 2.5826091592472142, "grad_norm": 2.909864902496338, "learning_rate": 3.118e-07, "loss": 0.018, "step": 48442 }, { "epoch": 2.5826624726768674, "grad_norm": 1.462206244468689, "learning_rate": 3.116e-07, "loss": 0.0061, "step": 48443 }, { "epoch": 2.58271578610652, "grad_norm": 1.6967413425445557, "learning_rate": 3.1140000000000005e-07, "loss": 0.0099, "step": 48444 }, { "epoch": 2.582769099536173, "grad_norm": 1.087459683418274, "learning_rate": 3.1120000000000003e-07, "loss": 0.0054, "step": 48445 }, { "epoch": 2.582822412965826, "grad_norm": 2.712226152420044, "learning_rate": 3.11e-07, "loss": 0.0155, "step": 48446 }, { "epoch": 2.582875726395479, "grad_norm": 1.5066527128219604, "learning_rate": 3.108e-07, "loss": 0.0088, "step": 48447 }, { "epoch": 2.582929039825132, "grad_norm": 2.4121146202087402, "learning_rate": 3.1060000000000004e-07, "loss": 0.0411, "step": 48448 }, { "epoch": 2.582982353254785, "grad_norm": 1.7356443405151367, "learning_rate": 3.1040000000000003e-07, "loss": 0.0265, "step": 48449 }, { "epoch": 2.583035666684438, "grad_norm": 0.6057618856430054, "learning_rate": 3.102e-07, "loss": 0.0029, "step": 48450 }, { "epoch": 2.583088980114091, "grad_norm": 1.591772437095642, "learning_rate": 3.1000000000000005e-07, "loss": 0.009, "step": 48451 }, { "epoch": 2.5831422935437436, "grad_norm": 1.5132625102996826, "learning_rate": 3.0980000000000003e-07, "loss": 0.0233, "step": 48452 }, { "epoch": 2.583195606973397, "grad_norm": 3.030543565750122, "learning_rate": 3.096e-07, "loss": 0.0192, "step": 48453 }, { "epoch": 2.5832489204030495, "grad_norm": 1.1785194873809814, "learning_rate": 3.094e-07, "loss": 0.0045, "step": 48454 }, { "epoch": 2.5833022338327023, "grad_norm": 3.1551225185394287, "learning_rate": 3.0920000000000004e-07, "loss": 0.0197, "step": 48455 }, { "epoch": 2.5833555472623555, "grad_norm": 1.5535169839859009, "learning_rate": 3.0900000000000003e-07, "loss": 0.0154, "step": 48456 }, { "epoch": 2.5834088606920083, "grad_norm": 0.44109636545181274, "learning_rate": 3.088e-07, "loss": 0.0027, "step": 48457 }, { "epoch": 2.583462174121661, "grad_norm": 1.723304033279419, "learning_rate": 3.086e-07, "loss": 0.0057, "step": 48458 }, { "epoch": 2.583515487551314, "grad_norm": 0.3848963677883148, "learning_rate": 3.0840000000000004e-07, "loss": 0.0026, "step": 48459 }, { "epoch": 2.583568800980967, "grad_norm": 0.48803651332855225, "learning_rate": 3.082e-07, "loss": 0.003, "step": 48460 }, { "epoch": 2.58362211441062, "grad_norm": 0.5627217292785645, "learning_rate": 3.0800000000000006e-07, "loss": 0.0035, "step": 48461 }, { "epoch": 2.583675427840273, "grad_norm": 2.0047178268432617, "learning_rate": 3.078e-07, "loss": 0.0141, "step": 48462 }, { "epoch": 2.583728741269926, "grad_norm": 0.36583641171455383, "learning_rate": 3.0760000000000003e-07, "loss": 0.0037, "step": 48463 }, { "epoch": 2.583782054699579, "grad_norm": 1.1784106492996216, "learning_rate": 3.074e-07, "loss": 0.0044, "step": 48464 }, { "epoch": 2.5838353681292316, "grad_norm": 1.2847846746444702, "learning_rate": 3.0720000000000005e-07, "loss": 0.0047, "step": 48465 }, { "epoch": 2.583888681558885, "grad_norm": 1.1869693994522095, "learning_rate": 3.0700000000000004e-07, "loss": 0.013, "step": 48466 }, { "epoch": 2.5839419949885376, "grad_norm": 1.9473966360092163, "learning_rate": 3.068e-07, "loss": 0.04, "step": 48467 }, { "epoch": 2.5839953084181904, "grad_norm": 3.455012559890747, "learning_rate": 3.066e-07, "loss": 0.0199, "step": 48468 }, { "epoch": 2.5840486218478436, "grad_norm": 1.4081963300704956, "learning_rate": 3.0640000000000004e-07, "loss": 0.0081, "step": 48469 }, { "epoch": 2.5841019352774963, "grad_norm": 1.2083301544189453, "learning_rate": 3.0620000000000003e-07, "loss": 0.0088, "step": 48470 }, { "epoch": 2.584155248707149, "grad_norm": 0.13669589161872864, "learning_rate": 3.06e-07, "loss": 0.0012, "step": 48471 }, { "epoch": 2.5842085621368023, "grad_norm": 1.6294044256210327, "learning_rate": 3.058e-07, "loss": 0.0069, "step": 48472 }, { "epoch": 2.584261875566455, "grad_norm": 0.2905650734901428, "learning_rate": 3.0560000000000004e-07, "loss": 0.0025, "step": 48473 }, { "epoch": 2.5843151889961082, "grad_norm": 2.8614230155944824, "learning_rate": 3.054e-07, "loss": 0.0231, "step": 48474 }, { "epoch": 2.584368502425761, "grad_norm": 1.8129520416259766, "learning_rate": 3.052e-07, "loss": 0.0155, "step": 48475 }, { "epoch": 2.584421815855414, "grad_norm": 1.4317373037338257, "learning_rate": 3.0500000000000004e-07, "loss": 0.0089, "step": 48476 }, { "epoch": 2.584475129285067, "grad_norm": 4.4081854820251465, "learning_rate": 3.0480000000000003e-07, "loss": 0.0126, "step": 48477 }, { "epoch": 2.5845284427147197, "grad_norm": 2.5908761024475098, "learning_rate": 3.046e-07, "loss": 0.0271, "step": 48478 }, { "epoch": 2.584581756144373, "grad_norm": 2.0547146797180176, "learning_rate": 3.044e-07, "loss": 0.0104, "step": 48479 }, { "epoch": 2.5846350695740257, "grad_norm": 3.492135524749756, "learning_rate": 3.0420000000000004e-07, "loss": 0.0384, "step": 48480 }, { "epoch": 2.5846883830036784, "grad_norm": 2.7362756729125977, "learning_rate": 3.04e-07, "loss": 0.0112, "step": 48481 }, { "epoch": 2.5847416964333316, "grad_norm": 0.6489982008934021, "learning_rate": 3.038e-07, "loss": 0.0029, "step": 48482 }, { "epoch": 2.5847950098629844, "grad_norm": 1.591353416442871, "learning_rate": 3.0360000000000005e-07, "loss": 0.0081, "step": 48483 }, { "epoch": 2.584848323292637, "grad_norm": 1.711456537246704, "learning_rate": 3.0340000000000003e-07, "loss": 0.019, "step": 48484 }, { "epoch": 2.5849016367222903, "grad_norm": 1.8465778827667236, "learning_rate": 3.032e-07, "loss": 0.0228, "step": 48485 }, { "epoch": 2.5849549501519435, "grad_norm": 0.8393083214759827, "learning_rate": 3.0300000000000005e-07, "loss": 0.0038, "step": 48486 }, { "epoch": 2.5850082635815963, "grad_norm": 3.6700565814971924, "learning_rate": 3.0280000000000004e-07, "loss": 0.0235, "step": 48487 }, { "epoch": 2.585061577011249, "grad_norm": 2.131153106689453, "learning_rate": 3.026e-07, "loss": 0.0064, "step": 48488 }, { "epoch": 2.5851148904409023, "grad_norm": 1.8895808458328247, "learning_rate": 3.024e-07, "loss": 0.0044, "step": 48489 }, { "epoch": 2.585168203870555, "grad_norm": 2.7981255054473877, "learning_rate": 3.0220000000000005e-07, "loss": 0.0153, "step": 48490 }, { "epoch": 2.5852215173002078, "grad_norm": 2.1080214977264404, "learning_rate": 3.0200000000000003e-07, "loss": 0.0134, "step": 48491 }, { "epoch": 2.585274830729861, "grad_norm": 1.3434501886367798, "learning_rate": 3.018e-07, "loss": 0.0155, "step": 48492 }, { "epoch": 2.5853281441595137, "grad_norm": 0.43745073676109314, "learning_rate": 3.016e-07, "loss": 0.0025, "step": 48493 }, { "epoch": 2.5853814575891665, "grad_norm": 1.4948724508285522, "learning_rate": 3.0140000000000004e-07, "loss": 0.0198, "step": 48494 }, { "epoch": 2.5854347710188197, "grad_norm": 1.0461360216140747, "learning_rate": 3.012e-07, "loss": 0.005, "step": 48495 }, { "epoch": 2.5854880844484724, "grad_norm": 1.9147056341171265, "learning_rate": 3.01e-07, "loss": 0.0116, "step": 48496 }, { "epoch": 2.5855413978781256, "grad_norm": 0.4290503263473511, "learning_rate": 3.008e-07, "loss": 0.0035, "step": 48497 }, { "epoch": 2.5855947113077784, "grad_norm": 0.6189311146736145, "learning_rate": 3.0060000000000003e-07, "loss": 0.0024, "step": 48498 }, { "epoch": 2.5856480247374316, "grad_norm": 1.5087828636169434, "learning_rate": 3.004e-07, "loss": 0.0078, "step": 48499 }, { "epoch": 2.5857013381670844, "grad_norm": 1.5083287954330444, "learning_rate": 3.0020000000000005e-07, "loss": 0.0057, "step": 48500 }, { "epoch": 2.585754651596737, "grad_norm": 1.000732421875, "learning_rate": 3.0000000000000004e-07, "loss": 0.0125, "step": 48501 }, { "epoch": 2.5858079650263903, "grad_norm": 3.5560789108276367, "learning_rate": 2.998e-07, "loss": 0.014, "step": 48502 }, { "epoch": 2.585861278456043, "grad_norm": 2.3714122772216797, "learning_rate": 2.996e-07, "loss": 0.0148, "step": 48503 }, { "epoch": 2.585914591885696, "grad_norm": 1.6681352853775024, "learning_rate": 2.9940000000000005e-07, "loss": 0.0198, "step": 48504 }, { "epoch": 2.585967905315349, "grad_norm": 0.47179514169692993, "learning_rate": 2.9920000000000003e-07, "loss": 0.0035, "step": 48505 }, { "epoch": 2.586021218745002, "grad_norm": 0.37857308983802795, "learning_rate": 2.99e-07, "loss": 0.0032, "step": 48506 }, { "epoch": 2.5860745321746545, "grad_norm": 1.4260746240615845, "learning_rate": 2.988e-07, "loss": 0.0259, "step": 48507 }, { "epoch": 2.5861278456043078, "grad_norm": 1.2785117626190186, "learning_rate": 2.9860000000000004e-07, "loss": 0.004, "step": 48508 }, { "epoch": 2.5861811590339605, "grad_norm": 0.6381672024726868, "learning_rate": 2.984e-07, "loss": 0.0023, "step": 48509 }, { "epoch": 2.5862344724636137, "grad_norm": 3.5599493980407715, "learning_rate": 2.982e-07, "loss": 0.0533, "step": 48510 }, { "epoch": 2.5862877858932665, "grad_norm": 1.8456089496612549, "learning_rate": 2.9800000000000005e-07, "loss": 0.0068, "step": 48511 }, { "epoch": 2.5863410993229197, "grad_norm": 2.052953004837036, "learning_rate": 2.9780000000000003e-07, "loss": 0.0222, "step": 48512 }, { "epoch": 2.5863944127525724, "grad_norm": 2.444826126098633, "learning_rate": 2.976e-07, "loss": 0.0081, "step": 48513 }, { "epoch": 2.586447726182225, "grad_norm": 2.4704642295837402, "learning_rate": 2.974e-07, "loss": 0.0218, "step": 48514 }, { "epoch": 2.5865010396118784, "grad_norm": 1.7276397943496704, "learning_rate": 2.9720000000000004e-07, "loss": 0.0142, "step": 48515 }, { "epoch": 2.586554353041531, "grad_norm": 2.035918951034546, "learning_rate": 2.97e-07, "loss": 0.0128, "step": 48516 }, { "epoch": 2.586607666471184, "grad_norm": 1.223703145980835, "learning_rate": 2.9680000000000006e-07, "loss": 0.0148, "step": 48517 }, { "epoch": 2.586660979900837, "grad_norm": 0.5149627327919006, "learning_rate": 2.966e-07, "loss": 0.0037, "step": 48518 }, { "epoch": 2.58671429333049, "grad_norm": 1.346470594406128, "learning_rate": 2.9640000000000003e-07, "loss": 0.0062, "step": 48519 }, { "epoch": 2.5867676067601426, "grad_norm": 4.187541961669922, "learning_rate": 2.962e-07, "loss": 0.0167, "step": 48520 }, { "epoch": 2.586820920189796, "grad_norm": 1.4957387447357178, "learning_rate": 2.9600000000000006e-07, "loss": 0.0051, "step": 48521 }, { "epoch": 2.586874233619449, "grad_norm": 3.014558792114258, "learning_rate": 2.958e-07, "loss": 0.009, "step": 48522 }, { "epoch": 2.5869275470491018, "grad_norm": 0.42457547783851624, "learning_rate": 2.956e-07, "loss": 0.0023, "step": 48523 }, { "epoch": 2.5869808604787545, "grad_norm": 0.4096372723579407, "learning_rate": 2.954e-07, "loss": 0.0018, "step": 48524 }, { "epoch": 2.5870341739084077, "grad_norm": 0.5482465028762817, "learning_rate": 2.9520000000000005e-07, "loss": 0.0032, "step": 48525 }, { "epoch": 2.5870874873380605, "grad_norm": 3.6825616359710693, "learning_rate": 2.9500000000000003e-07, "loss": 0.0187, "step": 48526 }, { "epoch": 2.5871408007677132, "grad_norm": 1.4265238046646118, "learning_rate": 2.948e-07, "loss": 0.0048, "step": 48527 }, { "epoch": 2.5871941141973664, "grad_norm": 1.4864959716796875, "learning_rate": 2.946e-07, "loss": 0.0102, "step": 48528 }, { "epoch": 2.587247427627019, "grad_norm": 1.8959866762161255, "learning_rate": 2.9440000000000004e-07, "loss": 0.0075, "step": 48529 }, { "epoch": 2.587300741056672, "grad_norm": 4.64055871963501, "learning_rate": 2.9420000000000003e-07, "loss": 0.0265, "step": 48530 }, { "epoch": 2.587354054486325, "grad_norm": 3.48169207572937, "learning_rate": 2.94e-07, "loss": 0.0109, "step": 48531 }, { "epoch": 2.587407367915978, "grad_norm": 0.8840554356575012, "learning_rate": 2.938e-07, "loss": 0.0037, "step": 48532 }, { "epoch": 2.587460681345631, "grad_norm": 0.5199778079986572, "learning_rate": 2.9360000000000003e-07, "loss": 0.0032, "step": 48533 }, { "epoch": 2.587513994775284, "grad_norm": 0.8212583661079407, "learning_rate": 2.934e-07, "loss": 0.0054, "step": 48534 }, { "epoch": 2.587567308204937, "grad_norm": 2.126737594604492, "learning_rate": 2.932e-07, "loss": 0.0244, "step": 48535 }, { "epoch": 2.58762062163459, "grad_norm": 1.0410953760147095, "learning_rate": 2.9300000000000004e-07, "loss": 0.0047, "step": 48536 }, { "epoch": 2.5876739350642426, "grad_norm": 2.581765651702881, "learning_rate": 2.9280000000000003e-07, "loss": 0.0194, "step": 48537 }, { "epoch": 2.587727248493896, "grad_norm": 3.024414300918579, "learning_rate": 2.926e-07, "loss": 0.0392, "step": 48538 }, { "epoch": 2.5877805619235486, "grad_norm": 2.6252553462982178, "learning_rate": 2.924e-07, "loss": 0.0223, "step": 48539 }, { "epoch": 2.5878338753532013, "grad_norm": 0.3132718801498413, "learning_rate": 2.9220000000000004e-07, "loss": 0.0016, "step": 48540 }, { "epoch": 2.5878871887828545, "grad_norm": 1.1290746927261353, "learning_rate": 2.92e-07, "loss": 0.0047, "step": 48541 }, { "epoch": 2.5879405022125073, "grad_norm": 2.191040277481079, "learning_rate": 2.9180000000000006e-07, "loss": 0.0165, "step": 48542 }, { "epoch": 2.58799381564216, "grad_norm": 5.09122896194458, "learning_rate": 2.916e-07, "loss": 0.0261, "step": 48543 }, { "epoch": 2.5880471290718132, "grad_norm": 0.6978916525840759, "learning_rate": 2.9140000000000003e-07, "loss": 0.0027, "step": 48544 }, { "epoch": 2.588100442501466, "grad_norm": 1.868773341178894, "learning_rate": 2.912e-07, "loss": 0.0072, "step": 48545 }, { "epoch": 2.588153755931119, "grad_norm": 3.1643855571746826, "learning_rate": 2.9100000000000005e-07, "loss": 0.0235, "step": 48546 }, { "epoch": 2.588207069360772, "grad_norm": 2.6344211101531982, "learning_rate": 2.9080000000000004e-07, "loss": 0.0332, "step": 48547 }, { "epoch": 2.588260382790425, "grad_norm": 2.690605640411377, "learning_rate": 2.906e-07, "loss": 0.0195, "step": 48548 }, { "epoch": 2.588313696220078, "grad_norm": 1.195212483406067, "learning_rate": 2.904e-07, "loss": 0.0159, "step": 48549 }, { "epoch": 2.5883670096497307, "grad_norm": 1.2246462106704712, "learning_rate": 2.9020000000000004e-07, "loss": 0.0064, "step": 48550 }, { "epoch": 2.588420323079384, "grad_norm": 2.488252878189087, "learning_rate": 2.9000000000000003e-07, "loss": 0.0087, "step": 48551 }, { "epoch": 2.5884736365090366, "grad_norm": 0.7891266942024231, "learning_rate": 2.898e-07, "loss": 0.0117, "step": 48552 }, { "epoch": 2.5885269499386894, "grad_norm": 1.307755947113037, "learning_rate": 2.896e-07, "loss": 0.0055, "step": 48553 }, { "epoch": 2.5885802633683426, "grad_norm": 2.3671517372131348, "learning_rate": 2.8940000000000004e-07, "loss": 0.0245, "step": 48554 }, { "epoch": 2.5886335767979953, "grad_norm": 1.1837563514709473, "learning_rate": 2.892e-07, "loss": 0.0039, "step": 48555 }, { "epoch": 2.588686890227648, "grad_norm": 2.2055838108062744, "learning_rate": 2.89e-07, "loss": 0.0284, "step": 48556 }, { "epoch": 2.5887402036573013, "grad_norm": 3.590794324874878, "learning_rate": 2.888e-07, "loss": 0.0336, "step": 48557 }, { "epoch": 2.588793517086954, "grad_norm": 0.9983136057853699, "learning_rate": 2.8860000000000003e-07, "loss": 0.0142, "step": 48558 }, { "epoch": 2.5888468305166072, "grad_norm": 1.682196021080017, "learning_rate": 2.884e-07, "loss": 0.0061, "step": 48559 }, { "epoch": 2.58890014394626, "grad_norm": 1.8207930326461792, "learning_rate": 2.882e-07, "loss": 0.0053, "step": 48560 }, { "epoch": 2.588953457375913, "grad_norm": 1.3576079607009888, "learning_rate": 2.8800000000000004e-07, "loss": 0.0055, "step": 48561 }, { "epoch": 2.589006770805566, "grad_norm": 2.014188528060913, "learning_rate": 2.878e-07, "loss": 0.0059, "step": 48562 }, { "epoch": 2.5890600842352187, "grad_norm": 1.9422866106033325, "learning_rate": 2.876e-07, "loss": 0.0083, "step": 48563 }, { "epoch": 2.589113397664872, "grad_norm": 4.044260025024414, "learning_rate": 2.8740000000000005e-07, "loss": 0.0339, "step": 48564 }, { "epoch": 2.5891667110945247, "grad_norm": 2.13720440864563, "learning_rate": 2.8720000000000003e-07, "loss": 0.0256, "step": 48565 }, { "epoch": 2.5892200245241774, "grad_norm": 0.9838641881942749, "learning_rate": 2.87e-07, "loss": 0.0175, "step": 48566 }, { "epoch": 2.5892733379538306, "grad_norm": 2.526561975479126, "learning_rate": 2.8680000000000005e-07, "loss": 0.0325, "step": 48567 }, { "epoch": 2.5893266513834834, "grad_norm": 3.245419502258301, "learning_rate": 2.8660000000000004e-07, "loss": 0.0262, "step": 48568 }, { "epoch": 2.5893799648131366, "grad_norm": 4.216637134552002, "learning_rate": 2.864e-07, "loss": 0.0113, "step": 48569 }, { "epoch": 2.5894332782427893, "grad_norm": 5.2701191902160645, "learning_rate": 2.862e-07, "loss": 0.0294, "step": 48570 }, { "epoch": 2.5894865916724426, "grad_norm": 2.186769485473633, "learning_rate": 2.8600000000000005e-07, "loss": 0.0084, "step": 48571 }, { "epoch": 2.5895399051020953, "grad_norm": 0.4721936285495758, "learning_rate": 2.8580000000000003e-07, "loss": 0.0034, "step": 48572 }, { "epoch": 2.589593218531748, "grad_norm": 1.337819218635559, "learning_rate": 2.856e-07, "loss": 0.0159, "step": 48573 }, { "epoch": 2.5896465319614013, "grad_norm": 2.1600470542907715, "learning_rate": 2.854e-07, "loss": 0.0039, "step": 48574 }, { "epoch": 2.589699845391054, "grad_norm": 0.7504911422729492, "learning_rate": 2.8520000000000004e-07, "loss": 0.0035, "step": 48575 }, { "epoch": 2.589753158820707, "grad_norm": 5.903887748718262, "learning_rate": 2.85e-07, "loss": 0.0272, "step": 48576 }, { "epoch": 2.58980647225036, "grad_norm": 2.1835861206054688, "learning_rate": 2.848e-07, "loss": 0.0123, "step": 48577 }, { "epoch": 2.5898597856800127, "grad_norm": 2.466848373413086, "learning_rate": 2.846e-07, "loss": 0.0112, "step": 48578 }, { "epoch": 2.5899130991096655, "grad_norm": 5.20820951461792, "learning_rate": 2.8440000000000003e-07, "loss": 0.026, "step": 48579 }, { "epoch": 2.5899664125393187, "grad_norm": 0.6792345643043518, "learning_rate": 2.842e-07, "loss": 0.0042, "step": 48580 }, { "epoch": 2.5900197259689715, "grad_norm": 2.102832794189453, "learning_rate": 2.8400000000000005e-07, "loss": 0.012, "step": 48581 }, { "epoch": 2.5900730393986247, "grad_norm": 4.349808216094971, "learning_rate": 2.838e-07, "loss": 0.0222, "step": 48582 }, { "epoch": 2.5901263528282774, "grad_norm": 0.4072800576686859, "learning_rate": 2.836e-07, "loss": 0.0023, "step": 48583 }, { "epoch": 2.5901796662579306, "grad_norm": 0.456508606672287, "learning_rate": 2.834e-07, "loss": 0.0033, "step": 48584 }, { "epoch": 2.5902329796875834, "grad_norm": 2.2017018795013428, "learning_rate": 2.8320000000000005e-07, "loss": 0.0299, "step": 48585 }, { "epoch": 2.590286293117236, "grad_norm": 1.8363149166107178, "learning_rate": 2.8300000000000003e-07, "loss": 0.0223, "step": 48586 }, { "epoch": 2.5903396065468893, "grad_norm": 1.477005958557129, "learning_rate": 2.828e-07, "loss": 0.0078, "step": 48587 }, { "epoch": 2.590392919976542, "grad_norm": 3.3805971145629883, "learning_rate": 2.826e-07, "loss": 0.0208, "step": 48588 }, { "epoch": 2.590446233406195, "grad_norm": 2.3849904537200928, "learning_rate": 2.8240000000000004e-07, "loss": 0.0082, "step": 48589 }, { "epoch": 2.590499546835848, "grad_norm": 1.4906983375549316, "learning_rate": 2.822e-07, "loss": 0.0052, "step": 48590 }, { "epoch": 2.590552860265501, "grad_norm": 1.9441479444503784, "learning_rate": 2.82e-07, "loss": 0.0208, "step": 48591 }, { "epoch": 2.5906061736951536, "grad_norm": 2.948754072189331, "learning_rate": 2.8180000000000005e-07, "loss": 0.0221, "step": 48592 }, { "epoch": 2.5906594871248068, "grad_norm": 0.16238664090633392, "learning_rate": 2.8160000000000003e-07, "loss": 0.0015, "step": 48593 }, { "epoch": 2.5907128005544595, "grad_norm": 2.4278039932250977, "learning_rate": 2.814e-07, "loss": 0.043, "step": 48594 }, { "epoch": 2.5907661139841127, "grad_norm": 0.20828112959861755, "learning_rate": 2.812e-07, "loss": 0.0017, "step": 48595 }, { "epoch": 2.5908194274137655, "grad_norm": 1.1559252738952637, "learning_rate": 2.8100000000000004e-07, "loss": 0.0125, "step": 48596 }, { "epoch": 2.5908727408434187, "grad_norm": 0.21043112874031067, "learning_rate": 2.808e-07, "loss": 0.0018, "step": 48597 }, { "epoch": 2.5909260542730714, "grad_norm": 3.1295406818389893, "learning_rate": 2.8060000000000006e-07, "loss": 0.01, "step": 48598 }, { "epoch": 2.590979367702724, "grad_norm": 3.4406023025512695, "learning_rate": 2.804e-07, "loss": 0.0125, "step": 48599 }, { "epoch": 2.5910326811323774, "grad_norm": 1.0161045789718628, "learning_rate": 2.8020000000000003e-07, "loss": 0.0141, "step": 48600 }, { "epoch": 2.59108599456203, "grad_norm": 2.399064302444458, "learning_rate": 2.8e-07, "loss": 0.0053, "step": 48601 }, { "epoch": 2.591139307991683, "grad_norm": 0.8105255961418152, "learning_rate": 2.7980000000000006e-07, "loss": 0.003, "step": 48602 }, { "epoch": 2.591192621421336, "grad_norm": 0.4959075450897217, "learning_rate": 2.796e-07, "loss": 0.0029, "step": 48603 }, { "epoch": 2.591245934850989, "grad_norm": 3.0534627437591553, "learning_rate": 2.7940000000000003e-07, "loss": 0.0183, "step": 48604 }, { "epoch": 2.5912992482806416, "grad_norm": 3.252366065979004, "learning_rate": 2.792e-07, "loss": 0.0368, "step": 48605 }, { "epoch": 2.591352561710295, "grad_norm": 6.047362804412842, "learning_rate": 2.7900000000000005e-07, "loss": 0.0191, "step": 48606 }, { "epoch": 2.591405875139948, "grad_norm": 1.1328935623168945, "learning_rate": 2.788e-07, "loss": 0.0044, "step": 48607 }, { "epoch": 2.591459188569601, "grad_norm": 0.9480907320976257, "learning_rate": 2.786e-07, "loss": 0.0126, "step": 48608 }, { "epoch": 2.5915125019992535, "grad_norm": 3.569380283355713, "learning_rate": 2.784e-07, "loss": 0.0219, "step": 48609 }, { "epoch": 2.5915658154289067, "grad_norm": 3.548128366470337, "learning_rate": 2.7820000000000004e-07, "loss": 0.0114, "step": 48610 }, { "epoch": 2.5916191288585595, "grad_norm": 0.37433290481567383, "learning_rate": 2.7800000000000003e-07, "loss": 0.0019, "step": 48611 }, { "epoch": 2.5916724422882123, "grad_norm": 2.6193363666534424, "learning_rate": 2.778e-07, "loss": 0.0204, "step": 48612 }, { "epoch": 2.5917257557178655, "grad_norm": 1.6697781085968018, "learning_rate": 2.776e-07, "loss": 0.0192, "step": 48613 }, { "epoch": 2.591779069147518, "grad_norm": 1.7817516326904297, "learning_rate": 2.7740000000000004e-07, "loss": 0.0034, "step": 48614 }, { "epoch": 2.591832382577171, "grad_norm": 0.6635306477546692, "learning_rate": 2.772e-07, "loss": 0.0036, "step": 48615 }, { "epoch": 2.591885696006824, "grad_norm": 4.464115619659424, "learning_rate": 2.77e-07, "loss": 0.0296, "step": 48616 }, { "epoch": 2.591939009436477, "grad_norm": 1.4558238983154297, "learning_rate": 2.7680000000000004e-07, "loss": 0.0045, "step": 48617 }, { "epoch": 2.59199232286613, "grad_norm": 0.3471145033836365, "learning_rate": 2.7660000000000003e-07, "loss": 0.0024, "step": 48618 }, { "epoch": 2.592045636295783, "grad_norm": 3.885798215866089, "learning_rate": 2.764e-07, "loss": 0.0297, "step": 48619 }, { "epoch": 2.592098949725436, "grad_norm": 0.448225736618042, "learning_rate": 2.762e-07, "loss": 0.0023, "step": 48620 }, { "epoch": 2.592152263155089, "grad_norm": 2.6537904739379883, "learning_rate": 2.7600000000000004e-07, "loss": 0.0092, "step": 48621 }, { "epoch": 2.5922055765847416, "grad_norm": 2.2147912979125977, "learning_rate": 2.758e-07, "loss": 0.0231, "step": 48622 }, { "epoch": 2.592258890014395, "grad_norm": 2.848884105682373, "learning_rate": 2.7560000000000006e-07, "loss": 0.0278, "step": 48623 }, { "epoch": 2.5923122034440476, "grad_norm": 2.0725433826446533, "learning_rate": 2.754e-07, "loss": 0.024, "step": 48624 }, { "epoch": 2.5923655168737003, "grad_norm": 2.382833480834961, "learning_rate": 2.7520000000000003e-07, "loss": 0.0209, "step": 48625 }, { "epoch": 2.5924188303033535, "grad_norm": 0.2622615694999695, "learning_rate": 2.75e-07, "loss": 0.0021, "step": 48626 }, { "epoch": 2.5924721437330063, "grad_norm": 0.2587231695652008, "learning_rate": 2.7480000000000005e-07, "loss": 0.0019, "step": 48627 }, { "epoch": 2.592525457162659, "grad_norm": 3.905933141708374, "learning_rate": 2.746e-07, "loss": 0.0239, "step": 48628 }, { "epoch": 2.5925787705923122, "grad_norm": 2.7642853260040283, "learning_rate": 2.744e-07, "loss": 0.0082, "step": 48629 }, { "epoch": 2.592632084021965, "grad_norm": 1.0155644416809082, "learning_rate": 2.742e-07, "loss": 0.017, "step": 48630 }, { "epoch": 2.592685397451618, "grad_norm": 0.9958248734474182, "learning_rate": 2.7400000000000004e-07, "loss": 0.0048, "step": 48631 }, { "epoch": 2.592738710881271, "grad_norm": 2.3395779132843018, "learning_rate": 2.7380000000000003e-07, "loss": 0.0082, "step": 48632 }, { "epoch": 2.592792024310924, "grad_norm": 1.3171055316925049, "learning_rate": 2.736e-07, "loss": 0.0164, "step": 48633 }, { "epoch": 2.592845337740577, "grad_norm": 1.3742550611495972, "learning_rate": 2.734e-07, "loss": 0.0059, "step": 48634 }, { "epoch": 2.5928986511702297, "grad_norm": 0.9591172933578491, "learning_rate": 2.7320000000000004e-07, "loss": 0.003, "step": 48635 }, { "epoch": 2.592951964599883, "grad_norm": 2.05849027633667, "learning_rate": 2.73e-07, "loss": 0.0067, "step": 48636 }, { "epoch": 2.5930052780295356, "grad_norm": 3.351065158843994, "learning_rate": 2.728e-07, "loss": 0.0095, "step": 48637 }, { "epoch": 2.5930585914591884, "grad_norm": 1.1027145385742188, "learning_rate": 2.726e-07, "loss": 0.0162, "step": 48638 }, { "epoch": 2.5931119048888416, "grad_norm": 1.1450221538543701, "learning_rate": 2.7240000000000003e-07, "loss": 0.0179, "step": 48639 }, { "epoch": 2.5931652183184943, "grad_norm": 1.283324122428894, "learning_rate": 2.722e-07, "loss": 0.0223, "step": 48640 }, { "epoch": 2.593218531748147, "grad_norm": 2.2337188720703125, "learning_rate": 2.72e-07, "loss": 0.0211, "step": 48641 }, { "epoch": 2.5932718451778003, "grad_norm": 1.2251310348510742, "learning_rate": 2.7180000000000004e-07, "loss": 0.0059, "step": 48642 }, { "epoch": 2.593325158607453, "grad_norm": 1.2866066694259644, "learning_rate": 2.716e-07, "loss": 0.0123, "step": 48643 }, { "epoch": 2.5933784720371063, "grad_norm": 1.9486587047576904, "learning_rate": 2.714e-07, "loss": 0.0054, "step": 48644 }, { "epoch": 2.593431785466759, "grad_norm": 2.380152702331543, "learning_rate": 2.712e-07, "loss": 0.0181, "step": 48645 }, { "epoch": 2.593485098896412, "grad_norm": 2.443638563156128, "learning_rate": 2.7100000000000003e-07, "loss": 0.0185, "step": 48646 }, { "epoch": 2.593538412326065, "grad_norm": 1.9960401058197021, "learning_rate": 2.708e-07, "loss": 0.0342, "step": 48647 }, { "epoch": 2.5935917257557177, "grad_norm": 1.3017773628234863, "learning_rate": 2.7060000000000005e-07, "loss": 0.0047, "step": 48648 }, { "epoch": 2.593645039185371, "grad_norm": 0.2888004779815674, "learning_rate": 2.7040000000000004e-07, "loss": 0.0022, "step": 48649 }, { "epoch": 2.5936983526150237, "grad_norm": 1.129865288734436, "learning_rate": 2.702e-07, "loss": 0.0167, "step": 48650 }, { "epoch": 2.5937516660446764, "grad_norm": 0.3539053797721863, "learning_rate": 2.7e-07, "loss": 0.0024, "step": 48651 }, { "epoch": 2.5938049794743296, "grad_norm": 1.2148789167404175, "learning_rate": 2.6980000000000005e-07, "loss": 0.0186, "step": 48652 }, { "epoch": 2.5938582929039824, "grad_norm": 2.263688325881958, "learning_rate": 2.6960000000000003e-07, "loss": 0.044, "step": 48653 }, { "epoch": 2.5939116063336356, "grad_norm": 3.0717597007751465, "learning_rate": 2.694e-07, "loss": 0.0198, "step": 48654 }, { "epoch": 2.5939649197632884, "grad_norm": 1.3639187812805176, "learning_rate": 2.692e-07, "loss": 0.0119, "step": 48655 }, { "epoch": 2.5940182331929416, "grad_norm": 4.577805042266846, "learning_rate": 2.6900000000000004e-07, "loss": 0.0306, "step": 48656 }, { "epoch": 2.5940715466225943, "grad_norm": 2.559053897857666, "learning_rate": 2.688e-07, "loss": 0.0293, "step": 48657 }, { "epoch": 2.594124860052247, "grad_norm": 1.94752836227417, "learning_rate": 2.686e-07, "loss": 0.0064, "step": 48658 }, { "epoch": 2.5941781734819003, "grad_norm": 3.9239323139190674, "learning_rate": 2.684e-07, "loss": 0.0274, "step": 48659 }, { "epoch": 2.594231486911553, "grad_norm": 2.2172977924346924, "learning_rate": 2.6820000000000003e-07, "loss": 0.0104, "step": 48660 }, { "epoch": 2.594284800341206, "grad_norm": 3.2511181831359863, "learning_rate": 2.68e-07, "loss": 0.0229, "step": 48661 }, { "epoch": 2.594338113770859, "grad_norm": 1.5637893676757812, "learning_rate": 2.6780000000000006e-07, "loss": 0.0058, "step": 48662 }, { "epoch": 2.5943914272005117, "grad_norm": 0.5522946715354919, "learning_rate": 2.676e-07, "loss": 0.0036, "step": 48663 }, { "epoch": 2.5944447406301645, "grad_norm": 4.173927307128906, "learning_rate": 2.674e-07, "loss": 0.0194, "step": 48664 }, { "epoch": 2.5944980540598177, "grad_norm": 2.5206804275512695, "learning_rate": 2.672e-07, "loss": 0.0288, "step": 48665 }, { "epoch": 2.5945513674894705, "grad_norm": 2.371504306793213, "learning_rate": 2.6700000000000005e-07, "loss": 0.0123, "step": 48666 }, { "epoch": 2.5946046809191237, "grad_norm": 0.7364323139190674, "learning_rate": 2.6680000000000003e-07, "loss": 0.0031, "step": 48667 }, { "epoch": 2.5946579943487764, "grad_norm": 1.9294884204864502, "learning_rate": 2.666e-07, "loss": 0.0304, "step": 48668 }, { "epoch": 2.5947113077784296, "grad_norm": 2.494826078414917, "learning_rate": 2.664e-07, "loss": 0.0279, "step": 48669 }, { "epoch": 2.5947646212080824, "grad_norm": 0.8352546095848083, "learning_rate": 2.6620000000000004e-07, "loss": 0.0078, "step": 48670 }, { "epoch": 2.594817934637735, "grad_norm": 3.4341704845428467, "learning_rate": 2.66e-07, "loss": 0.0117, "step": 48671 }, { "epoch": 2.5948712480673883, "grad_norm": 0.592136025428772, "learning_rate": 2.658e-07, "loss": 0.0034, "step": 48672 }, { "epoch": 2.594924561497041, "grad_norm": 0.2425759881734848, "learning_rate": 2.6560000000000005e-07, "loss": 0.0016, "step": 48673 }, { "epoch": 2.594977874926694, "grad_norm": 6.928555488586426, "learning_rate": 2.6540000000000003e-07, "loss": 0.0412, "step": 48674 }, { "epoch": 2.595031188356347, "grad_norm": 0.6730223298072815, "learning_rate": 2.652e-07, "loss": 0.0032, "step": 48675 }, { "epoch": 2.595084501786, "grad_norm": 1.022878885269165, "learning_rate": 2.65e-07, "loss": 0.0042, "step": 48676 }, { "epoch": 2.5951378152156526, "grad_norm": 1.6535824537277222, "learning_rate": 2.6480000000000004e-07, "loss": 0.0149, "step": 48677 }, { "epoch": 2.5951911286453058, "grad_norm": 1.9192932844161987, "learning_rate": 2.6460000000000003e-07, "loss": 0.0283, "step": 48678 }, { "epoch": 2.5952444420749585, "grad_norm": 1.4385662078857422, "learning_rate": 2.6440000000000006e-07, "loss": 0.0123, "step": 48679 }, { "epoch": 2.5952977555046117, "grad_norm": 0.8760129809379578, "learning_rate": 2.642e-07, "loss": 0.0035, "step": 48680 }, { "epoch": 2.5953510689342645, "grad_norm": 3.492652177810669, "learning_rate": 2.6400000000000003e-07, "loss": 0.0245, "step": 48681 }, { "epoch": 2.5954043823639177, "grad_norm": 2.264819860458374, "learning_rate": 2.638e-07, "loss": 0.006, "step": 48682 }, { "epoch": 2.5954576957935704, "grad_norm": 1.4830046892166138, "learning_rate": 2.6360000000000006e-07, "loss": 0.006, "step": 48683 }, { "epoch": 2.595511009223223, "grad_norm": 2.532668113708496, "learning_rate": 2.634e-07, "loss": 0.0237, "step": 48684 }, { "epoch": 2.5955643226528764, "grad_norm": 1.8426624536514282, "learning_rate": 2.6320000000000003e-07, "loss": 0.0085, "step": 48685 }, { "epoch": 2.595617636082529, "grad_norm": 1.0046299695968628, "learning_rate": 2.63e-07, "loss": 0.0043, "step": 48686 }, { "epoch": 2.595670949512182, "grad_norm": 3.319242238998413, "learning_rate": 2.6280000000000005e-07, "loss": 0.0242, "step": 48687 }, { "epoch": 2.595724262941835, "grad_norm": 0.5644171833992004, "learning_rate": 2.626e-07, "loss": 0.0024, "step": 48688 }, { "epoch": 2.595777576371488, "grad_norm": 2.970395088195801, "learning_rate": 2.624e-07, "loss": 0.0178, "step": 48689 }, { "epoch": 2.595830889801141, "grad_norm": 1.1846842765808105, "learning_rate": 2.622e-07, "loss": 0.0196, "step": 48690 }, { "epoch": 2.595884203230794, "grad_norm": 2.4596996307373047, "learning_rate": 2.6200000000000004e-07, "loss": 0.0417, "step": 48691 }, { "epoch": 2.595937516660447, "grad_norm": 1.703985333442688, "learning_rate": 2.6180000000000003e-07, "loss": 0.0046, "step": 48692 }, { "epoch": 2.5959908300901, "grad_norm": 3.2426140308380127, "learning_rate": 2.616e-07, "loss": 0.0314, "step": 48693 }, { "epoch": 2.5960441435197525, "grad_norm": 2.9349327087402344, "learning_rate": 2.614e-07, "loss": 0.0065, "step": 48694 }, { "epoch": 2.5960974569494057, "grad_norm": 1.91054105758667, "learning_rate": 2.6120000000000004e-07, "loss": 0.0061, "step": 48695 }, { "epoch": 2.5961507703790585, "grad_norm": 1.3952640295028687, "learning_rate": 2.61e-07, "loss": 0.0141, "step": 48696 }, { "epoch": 2.5962040838087113, "grad_norm": 0.8011325597763062, "learning_rate": 2.608e-07, "loss": 0.0031, "step": 48697 }, { "epoch": 2.5962573972383645, "grad_norm": 3.8046295642852783, "learning_rate": 2.6060000000000004e-07, "loss": 0.0398, "step": 48698 }, { "epoch": 2.596310710668017, "grad_norm": 1.9592010974884033, "learning_rate": 2.6040000000000003e-07, "loss": 0.0118, "step": 48699 }, { "epoch": 2.59636402409767, "grad_norm": 1.1396700143814087, "learning_rate": 2.602e-07, "loss": 0.0069, "step": 48700 }, { "epoch": 2.596417337527323, "grad_norm": 2.2479655742645264, "learning_rate": 2.6e-07, "loss": 0.04, "step": 48701 }, { "epoch": 2.596470650956976, "grad_norm": 2.589749574661255, "learning_rate": 2.5980000000000004e-07, "loss": 0.0418, "step": 48702 }, { "epoch": 2.596523964386629, "grad_norm": 1.9385217428207397, "learning_rate": 2.596e-07, "loss": 0.0219, "step": 48703 }, { "epoch": 2.596577277816282, "grad_norm": 0.7606748938560486, "learning_rate": 2.5940000000000006e-07, "loss": 0.0042, "step": 48704 }, { "epoch": 2.596630591245935, "grad_norm": 0.8754756450653076, "learning_rate": 2.592e-07, "loss": 0.0038, "step": 48705 }, { "epoch": 2.596683904675588, "grad_norm": 0.4441355764865875, "learning_rate": 2.5900000000000003e-07, "loss": 0.0023, "step": 48706 }, { "epoch": 2.5967372181052406, "grad_norm": 1.3128931522369385, "learning_rate": 2.588e-07, "loss": 0.007, "step": 48707 }, { "epoch": 2.596790531534894, "grad_norm": 2.2275869846343994, "learning_rate": 2.5860000000000005e-07, "loss": 0.0136, "step": 48708 }, { "epoch": 2.5968438449645466, "grad_norm": 1.6820963621139526, "learning_rate": 2.584e-07, "loss": 0.0378, "step": 48709 }, { "epoch": 2.5968971583941993, "grad_norm": 1.4970452785491943, "learning_rate": 2.582e-07, "loss": 0.0206, "step": 48710 }, { "epoch": 2.5969504718238525, "grad_norm": 1.472435474395752, "learning_rate": 2.58e-07, "loss": 0.0045, "step": 48711 }, { "epoch": 2.5970037852535053, "grad_norm": 0.9328974485397339, "learning_rate": 2.5780000000000005e-07, "loss": 0.0042, "step": 48712 }, { "epoch": 2.597057098683158, "grad_norm": 1.3680392503738403, "learning_rate": 2.5760000000000003e-07, "loss": 0.0066, "step": 48713 }, { "epoch": 2.5971104121128112, "grad_norm": 1.5825093984603882, "learning_rate": 2.574e-07, "loss": 0.0112, "step": 48714 }, { "epoch": 2.597163725542464, "grad_norm": 3.1858479976654053, "learning_rate": 2.572e-07, "loss": 0.0176, "step": 48715 }, { "epoch": 2.597217038972117, "grad_norm": 0.9448770880699158, "learning_rate": 2.5700000000000004e-07, "loss": 0.0053, "step": 48716 }, { "epoch": 2.59727035240177, "grad_norm": 0.6209081411361694, "learning_rate": 2.568e-07, "loss": 0.0035, "step": 48717 }, { "epoch": 2.597323665831423, "grad_norm": 1.6905720233917236, "learning_rate": 2.566e-07, "loss": 0.0063, "step": 48718 }, { "epoch": 2.597376979261076, "grad_norm": 2.7353689670562744, "learning_rate": 2.564e-07, "loss": 0.0246, "step": 48719 }, { "epoch": 2.5974302926907287, "grad_norm": 2.6379287242889404, "learning_rate": 2.5620000000000003e-07, "loss": 0.0224, "step": 48720 }, { "epoch": 2.597483606120382, "grad_norm": 2.5716564655303955, "learning_rate": 2.56e-07, "loss": 0.0092, "step": 48721 }, { "epoch": 2.5975369195500346, "grad_norm": 2.6419336795806885, "learning_rate": 2.558e-07, "loss": 0.0366, "step": 48722 }, { "epoch": 2.5975902329796874, "grad_norm": 2.1414847373962402, "learning_rate": 2.5560000000000004e-07, "loss": 0.0208, "step": 48723 }, { "epoch": 2.5976435464093406, "grad_norm": 2.1151139736175537, "learning_rate": 2.554e-07, "loss": 0.0149, "step": 48724 }, { "epoch": 2.5976968598389933, "grad_norm": 0.7583122253417969, "learning_rate": 2.552e-07, "loss": 0.0026, "step": 48725 }, { "epoch": 2.597750173268646, "grad_norm": 0.9231866002082825, "learning_rate": 2.55e-07, "loss": 0.0045, "step": 48726 }, { "epoch": 2.5978034866982993, "grad_norm": 2.596336603164673, "learning_rate": 2.5480000000000003e-07, "loss": 0.0296, "step": 48727 }, { "epoch": 2.5978568001279525, "grad_norm": 1.5647317171096802, "learning_rate": 2.546e-07, "loss": 0.0226, "step": 48728 }, { "epoch": 2.5979101135576053, "grad_norm": 1.0792114734649658, "learning_rate": 2.5440000000000005e-07, "loss": 0.0049, "step": 48729 }, { "epoch": 2.597963426987258, "grad_norm": 0.435084730386734, "learning_rate": 2.5420000000000004e-07, "loss": 0.0025, "step": 48730 }, { "epoch": 2.598016740416911, "grad_norm": 0.7761592864990234, "learning_rate": 2.54e-07, "loss": 0.003, "step": 48731 }, { "epoch": 2.598070053846564, "grad_norm": 2.6864993572235107, "learning_rate": 2.538e-07, "loss": 0.0205, "step": 48732 }, { "epoch": 2.5981233672762167, "grad_norm": 0.6794412136077881, "learning_rate": 2.5360000000000005e-07, "loss": 0.0033, "step": 48733 }, { "epoch": 2.59817668070587, "grad_norm": 2.242910146713257, "learning_rate": 2.5340000000000003e-07, "loss": 0.0387, "step": 48734 }, { "epoch": 2.5982299941355227, "grad_norm": 1.2111256122589111, "learning_rate": 2.532e-07, "loss": 0.0184, "step": 48735 }, { "epoch": 2.5982833075651754, "grad_norm": 2.260178804397583, "learning_rate": 2.53e-07, "loss": 0.0306, "step": 48736 }, { "epoch": 2.5983366209948286, "grad_norm": 1.3959722518920898, "learning_rate": 2.5280000000000004e-07, "loss": 0.0165, "step": 48737 }, { "epoch": 2.5983899344244814, "grad_norm": 1.4086514711380005, "learning_rate": 2.526e-07, "loss": 0.02, "step": 48738 }, { "epoch": 2.5984432478541346, "grad_norm": 0.8754789233207703, "learning_rate": 2.524e-07, "loss": 0.0035, "step": 48739 }, { "epoch": 2.5984965612837874, "grad_norm": 3.096888303756714, "learning_rate": 2.522e-07, "loss": 0.0159, "step": 48740 }, { "epoch": 2.5985498747134406, "grad_norm": 2.0498342514038086, "learning_rate": 2.5200000000000003e-07, "loss": 0.0125, "step": 48741 }, { "epoch": 2.5986031881430933, "grad_norm": 1.9003217220306396, "learning_rate": 2.518e-07, "loss": 0.0108, "step": 48742 }, { "epoch": 2.598656501572746, "grad_norm": 4.101003646850586, "learning_rate": 2.516e-07, "loss": 0.0384, "step": 48743 }, { "epoch": 2.5987098150023993, "grad_norm": 2.811976909637451, "learning_rate": 2.514e-07, "loss": 0.033, "step": 48744 }, { "epoch": 2.598763128432052, "grad_norm": 3.051272392272949, "learning_rate": 2.512e-07, "loss": 0.0215, "step": 48745 }, { "epoch": 2.598816441861705, "grad_norm": 4.64715051651001, "learning_rate": 2.51e-07, "loss": 0.0311, "step": 48746 }, { "epoch": 2.598869755291358, "grad_norm": 2.667240619659424, "learning_rate": 2.5080000000000005e-07, "loss": 0.0127, "step": 48747 }, { "epoch": 2.5989230687210108, "grad_norm": 1.676138997077942, "learning_rate": 2.5060000000000003e-07, "loss": 0.0076, "step": 48748 }, { "epoch": 2.5989763821506635, "grad_norm": 1.4700050354003906, "learning_rate": 2.504e-07, "loss": 0.0134, "step": 48749 }, { "epoch": 2.5990296955803167, "grad_norm": 5.120362758636475, "learning_rate": 2.502e-07, "loss": 0.0397, "step": 48750 }, { "epoch": 2.5990830090099695, "grad_norm": 1.8403451442718506, "learning_rate": 2.5000000000000004e-07, "loss": 0.0053, "step": 48751 }, { "epoch": 2.5991363224396227, "grad_norm": 1.9606444835662842, "learning_rate": 2.4980000000000003e-07, "loss": 0.0188, "step": 48752 }, { "epoch": 2.5991896358692754, "grad_norm": 0.8829551339149475, "learning_rate": 2.496e-07, "loss": 0.0044, "step": 48753 }, { "epoch": 2.5992429492989286, "grad_norm": 2.920013904571533, "learning_rate": 2.4940000000000005e-07, "loss": 0.0209, "step": 48754 }, { "epoch": 2.5992962627285814, "grad_norm": 3.6565725803375244, "learning_rate": 2.4920000000000003e-07, "loss": 0.0204, "step": 48755 }, { "epoch": 2.599349576158234, "grad_norm": 1.5035796165466309, "learning_rate": 2.49e-07, "loss": 0.0146, "step": 48756 }, { "epoch": 2.5994028895878873, "grad_norm": 1.4024533033370972, "learning_rate": 2.488e-07, "loss": 0.006, "step": 48757 }, { "epoch": 2.59945620301754, "grad_norm": 1.589835524559021, "learning_rate": 2.4860000000000004e-07, "loss": 0.0038, "step": 48758 }, { "epoch": 2.599509516447193, "grad_norm": 1.4018926620483398, "learning_rate": 2.4840000000000003e-07, "loss": 0.0193, "step": 48759 }, { "epoch": 2.599562829876846, "grad_norm": 1.776227355003357, "learning_rate": 2.482e-07, "loss": 0.0067, "step": 48760 }, { "epoch": 2.599616143306499, "grad_norm": 0.8870022892951965, "learning_rate": 2.48e-07, "loss": 0.0043, "step": 48761 }, { "epoch": 2.5996694567361516, "grad_norm": 0.8644792437553406, "learning_rate": 2.4780000000000004e-07, "loss": 0.0055, "step": 48762 }, { "epoch": 2.5997227701658048, "grad_norm": 2.543436288833618, "learning_rate": 2.476e-07, "loss": 0.0252, "step": 48763 }, { "epoch": 2.5997760835954575, "grad_norm": 2.46020245552063, "learning_rate": 2.4740000000000006e-07, "loss": 0.0404, "step": 48764 }, { "epoch": 2.5998293970251107, "grad_norm": 0.3902952969074249, "learning_rate": 2.472e-07, "loss": 0.0022, "step": 48765 }, { "epoch": 2.5998827104547635, "grad_norm": 2.8104684352874756, "learning_rate": 2.4700000000000003e-07, "loss": 0.0112, "step": 48766 }, { "epoch": 2.5999360238844167, "grad_norm": 0.6490763425827026, "learning_rate": 2.468e-07, "loss": 0.0036, "step": 48767 }, { "epoch": 2.5999893373140694, "grad_norm": 1.9712120294570923, "learning_rate": 2.4660000000000005e-07, "loss": 0.0197, "step": 48768 }, { "epoch": 2.600042650743722, "grad_norm": 1.0285236835479736, "learning_rate": 2.464e-07, "loss": 0.0043, "step": 48769 }, { "epoch": 2.6000959641733754, "grad_norm": 1.5972847938537598, "learning_rate": 2.462e-07, "loss": 0.0336, "step": 48770 }, { "epoch": 2.600149277603028, "grad_norm": 3.0030784606933594, "learning_rate": 2.46e-07, "loss": 0.0383, "step": 48771 }, { "epoch": 2.600202591032681, "grad_norm": 1.6377243995666504, "learning_rate": 2.4580000000000004e-07, "loss": 0.0066, "step": 48772 }, { "epoch": 2.600255904462334, "grad_norm": 4.074860095977783, "learning_rate": 2.4560000000000003e-07, "loss": 0.018, "step": 48773 }, { "epoch": 2.600309217891987, "grad_norm": 4.259678363800049, "learning_rate": 2.454e-07, "loss": 0.0225, "step": 48774 }, { "epoch": 2.60036253132164, "grad_norm": 0.7976800203323364, "learning_rate": 2.452e-07, "loss": 0.0048, "step": 48775 }, { "epoch": 2.600415844751293, "grad_norm": 3.544297933578491, "learning_rate": 2.4500000000000004e-07, "loss": 0.0285, "step": 48776 }, { "epoch": 2.600469158180946, "grad_norm": 3.115769624710083, "learning_rate": 2.448e-07, "loss": 0.012, "step": 48777 }, { "epoch": 2.600522471610599, "grad_norm": 2.194002389907837, "learning_rate": 2.446e-07, "loss": 0.0145, "step": 48778 }, { "epoch": 2.6005757850402516, "grad_norm": 2.6954662799835205, "learning_rate": 2.4440000000000004e-07, "loss": 0.0281, "step": 48779 }, { "epoch": 2.6006290984699048, "grad_norm": 0.747298002243042, "learning_rate": 2.4420000000000003e-07, "loss": 0.0047, "step": 48780 }, { "epoch": 2.6006824118995575, "grad_norm": 0.3237394690513611, "learning_rate": 2.44e-07, "loss": 0.0022, "step": 48781 }, { "epoch": 2.6007357253292103, "grad_norm": 0.6638980507850647, "learning_rate": 2.438e-07, "loss": 0.0039, "step": 48782 }, { "epoch": 2.6007890387588635, "grad_norm": 2.706298828125, "learning_rate": 2.4360000000000004e-07, "loss": 0.0103, "step": 48783 }, { "epoch": 2.6008423521885162, "grad_norm": 2.2368829250335693, "learning_rate": 2.434e-07, "loss": 0.0225, "step": 48784 }, { "epoch": 2.600895665618169, "grad_norm": 2.262611150741577, "learning_rate": 2.432e-07, "loss": 0.0114, "step": 48785 }, { "epoch": 2.600948979047822, "grad_norm": 2.9660332202911377, "learning_rate": 2.43e-07, "loss": 0.0129, "step": 48786 }, { "epoch": 2.601002292477475, "grad_norm": 1.9072794914245605, "learning_rate": 2.4280000000000003e-07, "loss": 0.0072, "step": 48787 }, { "epoch": 2.601055605907128, "grad_norm": 3.632603406906128, "learning_rate": 2.426e-07, "loss": 0.0339, "step": 48788 }, { "epoch": 2.601108919336781, "grad_norm": 1.466886043548584, "learning_rate": 2.4240000000000005e-07, "loss": 0.0165, "step": 48789 }, { "epoch": 2.601162232766434, "grad_norm": 0.3599761724472046, "learning_rate": 2.422e-07, "loss": 0.0023, "step": 48790 }, { "epoch": 2.601215546196087, "grad_norm": 0.71617591381073, "learning_rate": 2.42e-07, "loss": 0.0048, "step": 48791 }, { "epoch": 2.6012688596257396, "grad_norm": 1.3402621746063232, "learning_rate": 2.418e-07, "loss": 0.0031, "step": 48792 }, { "epoch": 2.601322173055393, "grad_norm": 1.1363385915756226, "learning_rate": 2.4160000000000005e-07, "loss": 0.003, "step": 48793 }, { "epoch": 2.6013754864850456, "grad_norm": 3.347578763961792, "learning_rate": 2.4140000000000003e-07, "loss": 0.0219, "step": 48794 }, { "epoch": 2.6014287999146983, "grad_norm": 2.8911476135253906, "learning_rate": 2.412e-07, "loss": 0.0065, "step": 48795 }, { "epoch": 2.6014821133443515, "grad_norm": 1.6994311809539795, "learning_rate": 2.41e-07, "loss": 0.0076, "step": 48796 }, { "epoch": 2.6015354267740043, "grad_norm": 1.1746536493301392, "learning_rate": 2.4080000000000004e-07, "loss": 0.0135, "step": 48797 }, { "epoch": 2.601588740203657, "grad_norm": 1.639472484588623, "learning_rate": 2.406e-07, "loss": 0.0064, "step": 48798 }, { "epoch": 2.6016420536333102, "grad_norm": 2.3804521560668945, "learning_rate": 2.404e-07, "loss": 0.0084, "step": 48799 }, { "epoch": 2.601695367062963, "grad_norm": 0.9300336837768555, "learning_rate": 2.402e-07, "loss": 0.0091, "step": 48800 }, { "epoch": 2.601748680492616, "grad_norm": 2.8859260082244873, "learning_rate": 2.4000000000000003e-07, "loss": 0.0127, "step": 48801 }, { "epoch": 2.601801993922269, "grad_norm": 3.1021311283111572, "learning_rate": 2.398e-07, "loss": 0.0269, "step": 48802 }, { "epoch": 2.601855307351922, "grad_norm": 1.7669315338134766, "learning_rate": 2.396e-07, "loss": 0.0148, "step": 48803 }, { "epoch": 2.601908620781575, "grad_norm": 1.3130433559417725, "learning_rate": 2.3940000000000004e-07, "loss": 0.0065, "step": 48804 }, { "epoch": 2.6019619342112277, "grad_norm": 1.399693489074707, "learning_rate": 2.392e-07, "loss": 0.0116, "step": 48805 }, { "epoch": 2.602015247640881, "grad_norm": 1.3443856239318848, "learning_rate": 2.39e-07, "loss": 0.0079, "step": 48806 }, { "epoch": 2.6020685610705336, "grad_norm": 1.486335039138794, "learning_rate": 2.388e-07, "loss": 0.0041, "step": 48807 }, { "epoch": 2.6021218745001864, "grad_norm": 1.553075909614563, "learning_rate": 2.3860000000000003e-07, "loss": 0.0078, "step": 48808 }, { "epoch": 2.6021751879298396, "grad_norm": 0.5142871141433716, "learning_rate": 2.3840000000000002e-07, "loss": 0.0033, "step": 48809 }, { "epoch": 2.6022285013594924, "grad_norm": 1.9411165714263916, "learning_rate": 2.3820000000000003e-07, "loss": 0.009, "step": 48810 }, { "epoch": 2.602281814789145, "grad_norm": 1.6731511354446411, "learning_rate": 2.3800000000000004e-07, "loss": 0.013, "step": 48811 }, { "epoch": 2.6023351282187983, "grad_norm": 2.6570918560028076, "learning_rate": 2.378e-07, "loss": 0.0748, "step": 48812 }, { "epoch": 2.6023884416484515, "grad_norm": 2.5296597480773926, "learning_rate": 2.376e-07, "loss": 0.0099, "step": 48813 }, { "epoch": 2.6024417550781043, "grad_norm": 2.1540307998657227, "learning_rate": 2.3740000000000002e-07, "loss": 0.0119, "step": 48814 }, { "epoch": 2.602495068507757, "grad_norm": 1.3846497535705566, "learning_rate": 2.3720000000000003e-07, "loss": 0.0145, "step": 48815 }, { "epoch": 2.6025483819374102, "grad_norm": 2.8242883682250977, "learning_rate": 2.3700000000000002e-07, "loss": 0.021, "step": 48816 }, { "epoch": 2.602601695367063, "grad_norm": 4.583283424377441, "learning_rate": 2.368e-07, "loss": 0.012, "step": 48817 }, { "epoch": 2.6026550087967157, "grad_norm": 1.2755507230758667, "learning_rate": 2.3660000000000001e-07, "loss": 0.0065, "step": 48818 }, { "epoch": 2.602708322226369, "grad_norm": 2.2742412090301514, "learning_rate": 2.3640000000000003e-07, "loss": 0.0148, "step": 48819 }, { "epoch": 2.6027616356560217, "grad_norm": 2.9467105865478516, "learning_rate": 2.362e-07, "loss": 0.0308, "step": 48820 }, { "epoch": 2.6028149490856745, "grad_norm": 1.072126865386963, "learning_rate": 2.3600000000000002e-07, "loss": 0.0085, "step": 48821 }, { "epoch": 2.6028682625153277, "grad_norm": 2.8477532863616943, "learning_rate": 2.3580000000000003e-07, "loss": 0.0234, "step": 48822 }, { "epoch": 2.6029215759449804, "grad_norm": 2.1260862350463867, "learning_rate": 2.3560000000000002e-07, "loss": 0.0227, "step": 48823 }, { "epoch": 2.6029748893746336, "grad_norm": 2.6635327339172363, "learning_rate": 2.354e-07, "loss": 0.0371, "step": 48824 }, { "epoch": 2.6030282028042864, "grad_norm": 3.342971086502075, "learning_rate": 2.3520000000000001e-07, "loss": 0.0094, "step": 48825 }, { "epoch": 2.6030815162339396, "grad_norm": 2.631420135498047, "learning_rate": 2.3500000000000003e-07, "loss": 0.0345, "step": 48826 }, { "epoch": 2.6031348296635923, "grad_norm": 0.7157492637634277, "learning_rate": 2.3480000000000004e-07, "loss": 0.0036, "step": 48827 }, { "epoch": 2.603188143093245, "grad_norm": 1.4124338626861572, "learning_rate": 2.3460000000000002e-07, "loss": 0.0187, "step": 48828 }, { "epoch": 2.6032414565228983, "grad_norm": 1.255738615989685, "learning_rate": 2.344e-07, "loss": 0.0138, "step": 48829 }, { "epoch": 2.603294769952551, "grad_norm": 1.8055264949798584, "learning_rate": 2.3420000000000002e-07, "loss": 0.0266, "step": 48830 }, { "epoch": 2.603348083382204, "grad_norm": 2.092047691345215, "learning_rate": 2.3400000000000003e-07, "loss": 0.0065, "step": 48831 }, { "epoch": 2.603401396811857, "grad_norm": 1.6092733144760132, "learning_rate": 2.3380000000000004e-07, "loss": 0.0254, "step": 48832 }, { "epoch": 2.6034547102415098, "grad_norm": 2.812103271484375, "learning_rate": 2.336e-07, "loss": 0.0256, "step": 48833 }, { "epoch": 2.6035080236711625, "grad_norm": 1.1256040334701538, "learning_rate": 2.334e-07, "loss": 0.0143, "step": 48834 }, { "epoch": 2.6035613371008157, "grad_norm": 3.0465681552886963, "learning_rate": 2.3320000000000002e-07, "loss": 0.0256, "step": 48835 }, { "epoch": 2.6036146505304685, "grad_norm": 0.8733263611793518, "learning_rate": 2.3300000000000003e-07, "loss": 0.0035, "step": 48836 }, { "epoch": 2.6036679639601217, "grad_norm": 1.9146195650100708, "learning_rate": 2.328e-07, "loss": 0.0256, "step": 48837 }, { "epoch": 2.6037212773897744, "grad_norm": 2.444013833999634, "learning_rate": 2.326e-07, "loss": 0.0186, "step": 48838 }, { "epoch": 2.6037745908194276, "grad_norm": 0.7390512228012085, "learning_rate": 2.3240000000000002e-07, "loss": 0.0121, "step": 48839 }, { "epoch": 2.6038279042490804, "grad_norm": 0.8455252051353455, "learning_rate": 2.3220000000000003e-07, "loss": 0.0038, "step": 48840 }, { "epoch": 2.603881217678733, "grad_norm": 2.3971519470214844, "learning_rate": 2.32e-07, "loss": 0.0077, "step": 48841 }, { "epoch": 2.6039345311083864, "grad_norm": 2.51745343208313, "learning_rate": 2.3180000000000002e-07, "loss": 0.0132, "step": 48842 }, { "epoch": 2.603987844538039, "grad_norm": 2.5348246097564697, "learning_rate": 2.316e-07, "loss": 0.0757, "step": 48843 }, { "epoch": 2.604041157967692, "grad_norm": 1.855600357055664, "learning_rate": 2.3140000000000002e-07, "loss": 0.0157, "step": 48844 }, { "epoch": 2.604094471397345, "grad_norm": 2.038860559463501, "learning_rate": 2.3120000000000003e-07, "loss": 0.0458, "step": 48845 }, { "epoch": 2.604147784826998, "grad_norm": 0.7113609910011292, "learning_rate": 2.3100000000000002e-07, "loss": 0.0039, "step": 48846 }, { "epoch": 2.6042010982566506, "grad_norm": 1.8785490989685059, "learning_rate": 2.3080000000000003e-07, "loss": 0.0053, "step": 48847 }, { "epoch": 2.604254411686304, "grad_norm": 2.909405469894409, "learning_rate": 2.3060000000000001e-07, "loss": 0.014, "step": 48848 }, { "epoch": 2.604307725115957, "grad_norm": 1.430053472518921, "learning_rate": 2.3040000000000002e-07, "loss": 0.0112, "step": 48849 }, { "epoch": 2.6043610385456097, "grad_norm": 0.991596519947052, "learning_rate": 2.302e-07, "loss": 0.0064, "step": 48850 }, { "epoch": 2.6044143519752625, "grad_norm": 0.903976321220398, "learning_rate": 2.3000000000000002e-07, "loss": 0.0068, "step": 48851 }, { "epoch": 2.6044676654049157, "grad_norm": 4.695570468902588, "learning_rate": 2.2980000000000003e-07, "loss": 0.0097, "step": 48852 }, { "epoch": 2.6045209788345685, "grad_norm": 3.1621811389923096, "learning_rate": 2.2960000000000004e-07, "loss": 0.0086, "step": 48853 }, { "epoch": 2.604574292264221, "grad_norm": 1.238122582435608, "learning_rate": 2.294e-07, "loss": 0.0053, "step": 48854 }, { "epoch": 2.6046276056938744, "grad_norm": 0.5542114973068237, "learning_rate": 2.2920000000000001e-07, "loss": 0.0029, "step": 48855 }, { "epoch": 2.604680919123527, "grad_norm": 0.3792307674884796, "learning_rate": 2.2900000000000003e-07, "loss": 0.0017, "step": 48856 }, { "epoch": 2.60473423255318, "grad_norm": 1.441705346107483, "learning_rate": 2.2880000000000004e-07, "loss": 0.0094, "step": 48857 }, { "epoch": 2.604787545982833, "grad_norm": 2.77291202545166, "learning_rate": 2.286e-07, "loss": 0.0234, "step": 48858 }, { "epoch": 2.604840859412486, "grad_norm": 2.110562324523926, "learning_rate": 2.284e-07, "loss": 0.0103, "step": 48859 }, { "epoch": 2.604894172842139, "grad_norm": 1.3814888000488281, "learning_rate": 2.2820000000000002e-07, "loss": 0.0088, "step": 48860 }, { "epoch": 2.604947486271792, "grad_norm": 1.8195271492004395, "learning_rate": 2.2800000000000003e-07, "loss": 0.0169, "step": 48861 }, { "epoch": 2.605000799701445, "grad_norm": 0.7767014503479004, "learning_rate": 2.2780000000000004e-07, "loss": 0.0047, "step": 48862 }, { "epoch": 2.605054113131098, "grad_norm": 2.716299295425415, "learning_rate": 2.276e-07, "loss": 0.0237, "step": 48863 }, { "epoch": 2.6051074265607506, "grad_norm": 2.543548822402954, "learning_rate": 2.274e-07, "loss": 0.0154, "step": 48864 }, { "epoch": 2.6051607399904038, "grad_norm": 3.7655372619628906, "learning_rate": 2.2720000000000002e-07, "loss": 0.0438, "step": 48865 }, { "epoch": 2.6052140534200565, "grad_norm": 0.2608421742916107, "learning_rate": 2.2700000000000003e-07, "loss": 0.0017, "step": 48866 }, { "epoch": 2.6052673668497093, "grad_norm": 4.01753044128418, "learning_rate": 2.2680000000000002e-07, "loss": 0.0114, "step": 48867 }, { "epoch": 2.6053206802793625, "grad_norm": 1.709409236907959, "learning_rate": 2.266e-07, "loss": 0.0157, "step": 48868 }, { "epoch": 2.6053739937090152, "grad_norm": 0.6259187459945679, "learning_rate": 2.2640000000000002e-07, "loss": 0.0031, "step": 48869 }, { "epoch": 2.605427307138668, "grad_norm": 2.862612009048462, "learning_rate": 2.2620000000000003e-07, "loss": 0.0118, "step": 48870 }, { "epoch": 2.605480620568321, "grad_norm": 2.008742570877075, "learning_rate": 2.26e-07, "loss": 0.0239, "step": 48871 }, { "epoch": 2.605533933997974, "grad_norm": 1.048323631286621, "learning_rate": 2.2580000000000002e-07, "loss": 0.0202, "step": 48872 }, { "epoch": 2.605587247427627, "grad_norm": 3.6602299213409424, "learning_rate": 2.256e-07, "loss": 0.0141, "step": 48873 }, { "epoch": 2.60564056085728, "grad_norm": 2.2968361377716064, "learning_rate": 2.2540000000000002e-07, "loss": 0.0137, "step": 48874 }, { "epoch": 2.605693874286933, "grad_norm": 2.0169589519500732, "learning_rate": 2.252e-07, "loss": 0.0122, "step": 48875 }, { "epoch": 2.605747187716586, "grad_norm": 0.9791198372840881, "learning_rate": 2.2500000000000002e-07, "loss": 0.0246, "step": 48876 }, { "epoch": 2.6058005011462386, "grad_norm": 0.8542923927307129, "learning_rate": 2.2480000000000003e-07, "loss": 0.0037, "step": 48877 }, { "epoch": 2.605853814575892, "grad_norm": 1.2593423128128052, "learning_rate": 2.2460000000000004e-07, "loss": 0.0353, "step": 48878 }, { "epoch": 2.6059071280055446, "grad_norm": 3.806586980819702, "learning_rate": 2.2440000000000002e-07, "loss": 0.0302, "step": 48879 }, { "epoch": 2.6059604414351973, "grad_norm": 1.6080175638198853, "learning_rate": 2.242e-07, "loss": 0.0214, "step": 48880 }, { "epoch": 2.6060137548648505, "grad_norm": 3.1929774284362793, "learning_rate": 2.2400000000000002e-07, "loss": 0.0299, "step": 48881 }, { "epoch": 2.6060670682945033, "grad_norm": 1.5015606880187988, "learning_rate": 2.2380000000000003e-07, "loss": 0.0215, "step": 48882 }, { "epoch": 2.606120381724156, "grad_norm": 6.010352611541748, "learning_rate": 2.2360000000000004e-07, "loss": 0.0273, "step": 48883 }, { "epoch": 2.6061736951538093, "grad_norm": 2.2565786838531494, "learning_rate": 2.234e-07, "loss": 0.0202, "step": 48884 }, { "epoch": 2.606227008583462, "grad_norm": 1.0942285060882568, "learning_rate": 2.2320000000000001e-07, "loss": 0.0092, "step": 48885 }, { "epoch": 2.606280322013115, "grad_norm": 1.473874807357788, "learning_rate": 2.2300000000000002e-07, "loss": 0.0073, "step": 48886 }, { "epoch": 2.606333635442768, "grad_norm": 2.227444887161255, "learning_rate": 2.2280000000000004e-07, "loss": 0.0238, "step": 48887 }, { "epoch": 2.606386948872421, "grad_norm": 1.6353871822357178, "learning_rate": 2.226e-07, "loss": 0.0379, "step": 48888 }, { "epoch": 2.606440262302074, "grad_norm": 4.231416702270508, "learning_rate": 2.224e-07, "loss": 0.0119, "step": 48889 }, { "epoch": 2.6064935757317267, "grad_norm": 1.7963873147964478, "learning_rate": 2.2220000000000002e-07, "loss": 0.0189, "step": 48890 }, { "epoch": 2.60654688916138, "grad_norm": 0.6537153720855713, "learning_rate": 2.2200000000000003e-07, "loss": 0.0031, "step": 48891 }, { "epoch": 2.6066002025910326, "grad_norm": 2.5658485889434814, "learning_rate": 2.2180000000000001e-07, "loss": 0.0304, "step": 48892 }, { "epoch": 2.6066535160206854, "grad_norm": 2.440255641937256, "learning_rate": 2.216e-07, "loss": 0.0075, "step": 48893 }, { "epoch": 2.6067068294503386, "grad_norm": 2.25766921043396, "learning_rate": 2.214e-07, "loss": 0.0411, "step": 48894 }, { "epoch": 2.6067601428799914, "grad_norm": 1.4299591779708862, "learning_rate": 2.2120000000000002e-07, "loss": 0.0048, "step": 48895 }, { "epoch": 2.6068134563096446, "grad_norm": 0.3754385709762573, "learning_rate": 2.2100000000000003e-07, "loss": 0.0025, "step": 48896 }, { "epoch": 2.6068667697392973, "grad_norm": 0.7001315951347351, "learning_rate": 2.2080000000000002e-07, "loss": 0.0022, "step": 48897 }, { "epoch": 2.6069200831689505, "grad_norm": 1.3278485536575317, "learning_rate": 2.206e-07, "loss": 0.0252, "step": 48898 }, { "epoch": 2.6069733965986033, "grad_norm": 0.6706788539886475, "learning_rate": 2.2040000000000001e-07, "loss": 0.0039, "step": 48899 }, { "epoch": 2.607026710028256, "grad_norm": 0.6742667555809021, "learning_rate": 2.2020000000000003e-07, "loss": 0.0049, "step": 48900 }, { "epoch": 2.6070800234579092, "grad_norm": 1.4397070407867432, "learning_rate": 2.2e-07, "loss": 0.0067, "step": 48901 }, { "epoch": 2.607133336887562, "grad_norm": 2.179561138153076, "learning_rate": 2.1980000000000002e-07, "loss": 0.0138, "step": 48902 }, { "epoch": 2.6071866503172147, "grad_norm": 0.8267459273338318, "learning_rate": 2.1960000000000003e-07, "loss": 0.0035, "step": 48903 }, { "epoch": 2.607239963746868, "grad_norm": 2.4706339836120605, "learning_rate": 2.1940000000000002e-07, "loss": 0.0225, "step": 48904 }, { "epoch": 2.6072932771765207, "grad_norm": 0.502983570098877, "learning_rate": 2.192e-07, "loss": 0.0037, "step": 48905 }, { "epoch": 2.6073465906061735, "grad_norm": 0.9515155553817749, "learning_rate": 2.1900000000000002e-07, "loss": 0.0039, "step": 48906 }, { "epoch": 2.6073999040358267, "grad_norm": 0.5603885054588318, "learning_rate": 2.1880000000000003e-07, "loss": 0.0027, "step": 48907 }, { "epoch": 2.6074532174654794, "grad_norm": 0.32636383175849915, "learning_rate": 2.1860000000000004e-07, "loss": 0.0028, "step": 48908 }, { "epoch": 2.6075065308951326, "grad_norm": 1.5339053869247437, "learning_rate": 2.184e-07, "loss": 0.0125, "step": 48909 }, { "epoch": 2.6075598443247854, "grad_norm": 0.946159303188324, "learning_rate": 2.182e-07, "loss": 0.0044, "step": 48910 }, { "epoch": 2.6076131577544386, "grad_norm": 1.42025887966156, "learning_rate": 2.1800000000000002e-07, "loss": 0.0219, "step": 48911 }, { "epoch": 2.6076664711840913, "grad_norm": 2.3047738075256348, "learning_rate": 2.1780000000000003e-07, "loss": 0.0212, "step": 48912 }, { "epoch": 2.607719784613744, "grad_norm": 3.0651535987854004, "learning_rate": 2.1760000000000004e-07, "loss": 0.0276, "step": 48913 }, { "epoch": 2.6077730980433973, "grad_norm": 2.6458473205566406, "learning_rate": 2.174e-07, "loss": 0.0349, "step": 48914 }, { "epoch": 2.60782641147305, "grad_norm": 2.0230889320373535, "learning_rate": 2.172e-07, "loss": 0.0096, "step": 48915 }, { "epoch": 2.607879724902703, "grad_norm": 1.0950429439544678, "learning_rate": 2.1700000000000002e-07, "loss": 0.0175, "step": 48916 }, { "epoch": 2.607933038332356, "grad_norm": 2.3288965225219727, "learning_rate": 2.1680000000000004e-07, "loss": 0.009, "step": 48917 }, { "epoch": 2.6079863517620088, "grad_norm": 1.866447925567627, "learning_rate": 2.166e-07, "loss": 0.0118, "step": 48918 }, { "epoch": 2.6080396651916615, "grad_norm": 0.7496653199195862, "learning_rate": 2.164e-07, "loss": 0.0047, "step": 48919 }, { "epoch": 2.6080929786213147, "grad_norm": 0.6087117791175842, "learning_rate": 2.1620000000000002e-07, "loss": 0.003, "step": 48920 }, { "epoch": 2.6081462920509675, "grad_norm": 0.23008345067501068, "learning_rate": 2.1600000000000003e-07, "loss": 0.0015, "step": 48921 }, { "epoch": 2.6081996054806207, "grad_norm": 0.6002259254455566, "learning_rate": 2.1580000000000001e-07, "loss": 0.0032, "step": 48922 }, { "epoch": 2.6082529189102734, "grad_norm": 2.1434059143066406, "learning_rate": 2.156e-07, "loss": 0.0146, "step": 48923 }, { "epoch": 2.6083062323399266, "grad_norm": 1.2310296297073364, "learning_rate": 2.154e-07, "loss": 0.0215, "step": 48924 }, { "epoch": 2.6083595457695794, "grad_norm": 3.562746047973633, "learning_rate": 2.1520000000000002e-07, "loss": 0.0248, "step": 48925 }, { "epoch": 2.608412859199232, "grad_norm": 1.9745970964431763, "learning_rate": 2.15e-07, "loss": 0.0083, "step": 48926 }, { "epoch": 2.6084661726288854, "grad_norm": 1.6384730339050293, "learning_rate": 2.1480000000000002e-07, "loss": 0.0184, "step": 48927 }, { "epoch": 2.608519486058538, "grad_norm": 0.9112149477005005, "learning_rate": 2.1460000000000003e-07, "loss": 0.0047, "step": 48928 }, { "epoch": 2.608572799488191, "grad_norm": 0.4544999301433563, "learning_rate": 2.1440000000000001e-07, "loss": 0.0025, "step": 48929 }, { "epoch": 2.608626112917844, "grad_norm": 0.5206016898155212, "learning_rate": 2.1420000000000003e-07, "loss": 0.0035, "step": 48930 }, { "epoch": 2.608679426347497, "grad_norm": 1.5395565032958984, "learning_rate": 2.14e-07, "loss": 0.0168, "step": 48931 }, { "epoch": 2.6087327397771496, "grad_norm": 2.0736570358276367, "learning_rate": 2.1380000000000002e-07, "loss": 0.016, "step": 48932 }, { "epoch": 2.608786053206803, "grad_norm": 1.7365362644195557, "learning_rate": 2.1360000000000003e-07, "loss": 0.0086, "step": 48933 }, { "epoch": 2.608839366636456, "grad_norm": 1.982184886932373, "learning_rate": 2.1340000000000002e-07, "loss": 0.0211, "step": 48934 }, { "epoch": 2.6088926800661087, "grad_norm": 2.5727570056915283, "learning_rate": 2.132e-07, "loss": 0.0076, "step": 48935 }, { "epoch": 2.6089459934957615, "grad_norm": 1.268180012702942, "learning_rate": 2.1300000000000001e-07, "loss": 0.005, "step": 48936 }, { "epoch": 2.6089993069254147, "grad_norm": 1.759893774986267, "learning_rate": 2.1280000000000003e-07, "loss": 0.0083, "step": 48937 }, { "epoch": 2.6090526203550675, "grad_norm": 0.15813298523426056, "learning_rate": 2.1260000000000004e-07, "loss": 0.0011, "step": 48938 }, { "epoch": 2.60910593378472, "grad_norm": 3.5282976627349854, "learning_rate": 2.124e-07, "loss": 0.0339, "step": 48939 }, { "epoch": 2.6091592472143734, "grad_norm": 0.6463561654090881, "learning_rate": 2.122e-07, "loss": 0.0035, "step": 48940 }, { "epoch": 2.609212560644026, "grad_norm": 2.3907647132873535, "learning_rate": 2.1200000000000002e-07, "loss": 0.0074, "step": 48941 }, { "epoch": 2.609265874073679, "grad_norm": 0.47427624464035034, "learning_rate": 2.1180000000000003e-07, "loss": 0.003, "step": 48942 }, { "epoch": 2.609319187503332, "grad_norm": 1.7954314947128296, "learning_rate": 2.1160000000000004e-07, "loss": 0.0321, "step": 48943 }, { "epoch": 2.609372500932985, "grad_norm": 1.6373587846755981, "learning_rate": 2.114e-07, "loss": 0.016, "step": 48944 }, { "epoch": 2.609425814362638, "grad_norm": 2.778813362121582, "learning_rate": 2.112e-07, "loss": 0.0136, "step": 48945 }, { "epoch": 2.609479127792291, "grad_norm": 0.91783607006073, "learning_rate": 2.1100000000000002e-07, "loss": 0.0133, "step": 48946 }, { "epoch": 2.609532441221944, "grad_norm": 1.1029671430587769, "learning_rate": 2.1080000000000003e-07, "loss": 0.0059, "step": 48947 }, { "epoch": 2.609585754651597, "grad_norm": 2.4207427501678467, "learning_rate": 2.106e-07, "loss": 0.0237, "step": 48948 }, { "epoch": 2.6096390680812496, "grad_norm": 2.844714879989624, "learning_rate": 2.104e-07, "loss": 0.0115, "step": 48949 }, { "epoch": 2.6096923815109028, "grad_norm": 2.5158183574676514, "learning_rate": 2.1020000000000002e-07, "loss": 0.0254, "step": 48950 }, { "epoch": 2.6097456949405555, "grad_norm": 1.437864899635315, "learning_rate": 2.1000000000000003e-07, "loss": 0.005, "step": 48951 }, { "epoch": 2.6097990083702083, "grad_norm": 1.7615896463394165, "learning_rate": 2.098e-07, "loss": 0.0192, "step": 48952 }, { "epoch": 2.6098523217998615, "grad_norm": 2.1837317943573, "learning_rate": 2.0960000000000002e-07, "loss": 0.0077, "step": 48953 }, { "epoch": 2.6099056352295142, "grad_norm": 1.9041211605072021, "learning_rate": 2.094e-07, "loss": 0.0205, "step": 48954 }, { "epoch": 2.609958948659167, "grad_norm": 0.39319974184036255, "learning_rate": 2.0920000000000002e-07, "loss": 0.0026, "step": 48955 }, { "epoch": 2.61001226208882, "grad_norm": 3.628502130508423, "learning_rate": 2.09e-07, "loss": 0.0205, "step": 48956 }, { "epoch": 2.610065575518473, "grad_norm": 2.044023275375366, "learning_rate": 2.0880000000000002e-07, "loss": 0.0316, "step": 48957 }, { "epoch": 2.610118888948126, "grad_norm": 0.5511320233345032, "learning_rate": 2.0860000000000003e-07, "loss": 0.004, "step": 48958 }, { "epoch": 2.610172202377779, "grad_norm": 4.007858753204346, "learning_rate": 2.084e-07, "loss": 0.0412, "step": 48959 }, { "epoch": 2.610225515807432, "grad_norm": 0.7792408466339111, "learning_rate": 2.0820000000000002e-07, "loss": 0.004, "step": 48960 }, { "epoch": 2.610278829237085, "grad_norm": 2.3018264770507812, "learning_rate": 2.08e-07, "loss": 0.0288, "step": 48961 }, { "epoch": 2.6103321426667376, "grad_norm": 0.6885779500007629, "learning_rate": 2.0780000000000002e-07, "loss": 0.0034, "step": 48962 }, { "epoch": 2.610385456096391, "grad_norm": 2.767751455307007, "learning_rate": 2.0760000000000003e-07, "loss": 0.0395, "step": 48963 }, { "epoch": 2.6104387695260436, "grad_norm": 1.0771111249923706, "learning_rate": 2.0740000000000004e-07, "loss": 0.0218, "step": 48964 }, { "epoch": 2.6104920829556963, "grad_norm": 3.6478021144866943, "learning_rate": 2.072e-07, "loss": 0.0453, "step": 48965 }, { "epoch": 2.6105453963853495, "grad_norm": 0.5845074653625488, "learning_rate": 2.0700000000000001e-07, "loss": 0.0037, "step": 48966 }, { "epoch": 2.6105987098150023, "grad_norm": 1.7431386709213257, "learning_rate": 2.0680000000000003e-07, "loss": 0.0068, "step": 48967 }, { "epoch": 2.610652023244655, "grad_norm": 0.19027164578437805, "learning_rate": 2.0660000000000004e-07, "loss": 0.0009, "step": 48968 }, { "epoch": 2.6107053366743083, "grad_norm": 0.3373221158981323, "learning_rate": 2.064e-07, "loss": 0.0024, "step": 48969 }, { "epoch": 2.6107586501039615, "grad_norm": 0.882949709892273, "learning_rate": 2.062e-07, "loss": 0.0038, "step": 48970 }, { "epoch": 2.610811963533614, "grad_norm": 1.2643630504608154, "learning_rate": 2.0600000000000002e-07, "loss": 0.0037, "step": 48971 }, { "epoch": 2.610865276963267, "grad_norm": 0.1999671459197998, "learning_rate": 2.0580000000000003e-07, "loss": 0.0017, "step": 48972 }, { "epoch": 2.61091859039292, "grad_norm": 2.4452948570251465, "learning_rate": 2.056e-07, "loss": 0.0215, "step": 48973 }, { "epoch": 2.610971903822573, "grad_norm": 1.5259085893630981, "learning_rate": 2.054e-07, "loss": 0.0099, "step": 48974 }, { "epoch": 2.6110252172522257, "grad_norm": 2.4783921241760254, "learning_rate": 2.052e-07, "loss": 0.0313, "step": 48975 }, { "epoch": 2.611078530681879, "grad_norm": 0.39125490188598633, "learning_rate": 2.0500000000000002e-07, "loss": 0.0025, "step": 48976 }, { "epoch": 2.6111318441115317, "grad_norm": 1.4663574695587158, "learning_rate": 2.0480000000000003e-07, "loss": 0.0104, "step": 48977 }, { "epoch": 2.6111851575411844, "grad_norm": 2.7174649238586426, "learning_rate": 2.0460000000000002e-07, "loss": 0.016, "step": 48978 }, { "epoch": 2.6112384709708376, "grad_norm": 0.7214241027832031, "learning_rate": 2.044e-07, "loss": 0.0043, "step": 48979 }, { "epoch": 2.6112917844004904, "grad_norm": 1.2474830150604248, "learning_rate": 2.0420000000000002e-07, "loss": 0.0184, "step": 48980 }, { "epoch": 2.6113450978301436, "grad_norm": 2.711040735244751, "learning_rate": 2.0400000000000003e-07, "loss": 0.034, "step": 48981 }, { "epoch": 2.6113984112597963, "grad_norm": 0.7717283368110657, "learning_rate": 2.038e-07, "loss": 0.0054, "step": 48982 }, { "epoch": 2.6114517246894495, "grad_norm": 1.632201910018921, "learning_rate": 2.0360000000000002e-07, "loss": 0.0155, "step": 48983 }, { "epoch": 2.6115050381191023, "grad_norm": 0.3672259747982025, "learning_rate": 2.034e-07, "loss": 0.0023, "step": 48984 }, { "epoch": 2.611558351548755, "grad_norm": 0.9367513060569763, "learning_rate": 2.0320000000000002e-07, "loss": 0.0033, "step": 48985 }, { "epoch": 2.6116116649784082, "grad_norm": 3.2038462162017822, "learning_rate": 2.03e-07, "loss": 0.0142, "step": 48986 }, { "epoch": 2.611664978408061, "grad_norm": 3.946559429168701, "learning_rate": 2.0280000000000002e-07, "loss": 0.0169, "step": 48987 }, { "epoch": 2.6117182918377138, "grad_norm": 1.8299285173416138, "learning_rate": 2.0260000000000003e-07, "loss": 0.0134, "step": 48988 }, { "epoch": 2.611771605267367, "grad_norm": 4.3830695152282715, "learning_rate": 2.0240000000000004e-07, "loss": 0.0483, "step": 48989 }, { "epoch": 2.6118249186970197, "grad_norm": 1.1743476390838623, "learning_rate": 2.022e-07, "loss": 0.0082, "step": 48990 }, { "epoch": 2.6118782321266725, "grad_norm": 2.7984752655029297, "learning_rate": 2.02e-07, "loss": 0.0287, "step": 48991 }, { "epoch": 2.6119315455563257, "grad_norm": 0.5847910642623901, "learning_rate": 2.0180000000000002e-07, "loss": 0.0033, "step": 48992 }, { "epoch": 2.6119848589859784, "grad_norm": 2.4944992065429688, "learning_rate": 2.0160000000000003e-07, "loss": 0.0157, "step": 48993 }, { "epoch": 2.6120381724156316, "grad_norm": 6.411859035491943, "learning_rate": 2.0140000000000004e-07, "loss": 0.0344, "step": 48994 }, { "epoch": 2.6120914858452844, "grad_norm": 1.8186544179916382, "learning_rate": 2.012e-07, "loss": 0.0213, "step": 48995 }, { "epoch": 2.6121447992749376, "grad_norm": 0.9449599981307983, "learning_rate": 2.01e-07, "loss": 0.0041, "step": 48996 }, { "epoch": 2.6121981127045903, "grad_norm": 1.8590978384017944, "learning_rate": 2.0080000000000002e-07, "loss": 0.0104, "step": 48997 }, { "epoch": 2.612251426134243, "grad_norm": 2.3972668647766113, "learning_rate": 2.0060000000000004e-07, "loss": 0.0128, "step": 48998 }, { "epoch": 2.6123047395638963, "grad_norm": 1.586806297302246, "learning_rate": 2.004e-07, "loss": 0.0073, "step": 48999 }, { "epoch": 2.612358052993549, "grad_norm": 2.250777244567871, "learning_rate": 2.002e-07, "loss": 0.0175, "step": 49000 }, { "epoch": 2.612411366423202, "grad_norm": 0.8377165794372559, "learning_rate": 2.0000000000000002e-07, "loss": 0.0137, "step": 49001 }, { "epoch": 2.612464679852855, "grad_norm": 0.5753382444381714, "learning_rate": 1.9980000000000003e-07, "loss": 0.0032, "step": 49002 }, { "epoch": 2.6125179932825078, "grad_norm": 1.785187005996704, "learning_rate": 1.9960000000000001e-07, "loss": 0.0056, "step": 49003 }, { "epoch": 2.6125713067121605, "grad_norm": 1.5534875392913818, "learning_rate": 1.994e-07, "loss": 0.0093, "step": 49004 }, { "epoch": 2.6126246201418137, "grad_norm": 3.2083024978637695, "learning_rate": 1.992e-07, "loss": 0.0317, "step": 49005 }, { "epoch": 2.6126779335714665, "grad_norm": 5.347265720367432, "learning_rate": 1.9900000000000002e-07, "loss": 0.0318, "step": 49006 }, { "epoch": 2.6127312470011197, "grad_norm": 1.3384867906570435, "learning_rate": 1.988e-07, "loss": 0.018, "step": 49007 }, { "epoch": 2.6127845604307725, "grad_norm": 2.728405714035034, "learning_rate": 1.9860000000000002e-07, "loss": 0.0116, "step": 49008 }, { "epoch": 2.6128378738604257, "grad_norm": 2.803731679916382, "learning_rate": 1.984e-07, "loss": 0.0289, "step": 49009 }, { "epoch": 2.6128911872900784, "grad_norm": 1.2372431755065918, "learning_rate": 1.9820000000000001e-07, "loss": 0.0054, "step": 49010 }, { "epoch": 2.612944500719731, "grad_norm": 2.655306100845337, "learning_rate": 1.9800000000000003e-07, "loss": 0.0329, "step": 49011 }, { "epoch": 2.6129978141493844, "grad_norm": 1.6300321817398071, "learning_rate": 1.978e-07, "loss": 0.0108, "step": 49012 }, { "epoch": 2.613051127579037, "grad_norm": 1.2905431985855103, "learning_rate": 1.9760000000000002e-07, "loss": 0.0136, "step": 49013 }, { "epoch": 2.61310444100869, "grad_norm": 0.5241079330444336, "learning_rate": 1.9740000000000003e-07, "loss": 0.0036, "step": 49014 }, { "epoch": 2.613157754438343, "grad_norm": 1.0768721103668213, "learning_rate": 1.9720000000000002e-07, "loss": 0.0189, "step": 49015 }, { "epoch": 2.613211067867996, "grad_norm": 0.9853094220161438, "learning_rate": 1.97e-07, "loss": 0.0037, "step": 49016 }, { "epoch": 2.613264381297649, "grad_norm": 0.4091348350048065, "learning_rate": 1.9680000000000002e-07, "loss": 0.0025, "step": 49017 }, { "epoch": 2.613317694727302, "grad_norm": 1.0534495115280151, "learning_rate": 1.9660000000000003e-07, "loss": 0.0078, "step": 49018 }, { "epoch": 2.613371008156955, "grad_norm": 0.6470345258712769, "learning_rate": 1.9640000000000004e-07, "loss": 0.0043, "step": 49019 }, { "epoch": 2.6134243215866078, "grad_norm": 2.9040446281433105, "learning_rate": 1.962e-07, "loss": 0.0217, "step": 49020 }, { "epoch": 2.6134776350162605, "grad_norm": 0.30821493268013, "learning_rate": 1.96e-07, "loss": 0.0024, "step": 49021 }, { "epoch": 2.6135309484459137, "grad_norm": 2.5426766872406006, "learning_rate": 1.9580000000000002e-07, "loss": 0.0524, "step": 49022 }, { "epoch": 2.6135842618755665, "grad_norm": 1.5670784711837769, "learning_rate": 1.9560000000000003e-07, "loss": 0.0345, "step": 49023 }, { "epoch": 2.6136375753052192, "grad_norm": 2.1166324615478516, "learning_rate": 1.954e-07, "loss": 0.0208, "step": 49024 }, { "epoch": 2.6136908887348724, "grad_norm": 2.084502935409546, "learning_rate": 1.952e-07, "loss": 0.0151, "step": 49025 }, { "epoch": 2.613744202164525, "grad_norm": 0.9052750468254089, "learning_rate": 1.95e-07, "loss": 0.0042, "step": 49026 }, { "epoch": 2.613797515594178, "grad_norm": 1.762133240699768, "learning_rate": 1.9480000000000002e-07, "loss": 0.0165, "step": 49027 }, { "epoch": 2.613850829023831, "grad_norm": 1.7196426391601562, "learning_rate": 1.9460000000000004e-07, "loss": 0.0221, "step": 49028 }, { "epoch": 2.613904142453484, "grad_norm": 0.29589158296585083, "learning_rate": 1.944e-07, "loss": 0.0026, "step": 49029 }, { "epoch": 2.613957455883137, "grad_norm": 0.28732216358184814, "learning_rate": 1.942e-07, "loss": 0.0024, "step": 49030 }, { "epoch": 2.61401076931279, "grad_norm": 0.2010296881198883, "learning_rate": 1.9400000000000002e-07, "loss": 0.0015, "step": 49031 }, { "epoch": 2.614064082742443, "grad_norm": 1.5530740022659302, "learning_rate": 1.9380000000000003e-07, "loss": 0.0043, "step": 49032 }, { "epoch": 2.614117396172096, "grad_norm": 3.3288938999176025, "learning_rate": 1.936e-07, "loss": 0.0114, "step": 49033 }, { "epoch": 2.6141707096017486, "grad_norm": 1.9429512023925781, "learning_rate": 1.9340000000000002e-07, "loss": 0.0045, "step": 49034 }, { "epoch": 2.6142240230314018, "grad_norm": 1.939253330230713, "learning_rate": 1.932e-07, "loss": 0.01, "step": 49035 }, { "epoch": 2.6142773364610545, "grad_norm": 2.1249167919158936, "learning_rate": 1.9300000000000002e-07, "loss": 0.0221, "step": 49036 }, { "epoch": 2.6143306498907073, "grad_norm": 1.15097177028656, "learning_rate": 1.928e-07, "loss": 0.0126, "step": 49037 }, { "epoch": 2.6143839633203605, "grad_norm": 0.36193743348121643, "learning_rate": 1.9260000000000002e-07, "loss": 0.002, "step": 49038 }, { "epoch": 2.6144372767500133, "grad_norm": 2.1654810905456543, "learning_rate": 1.9240000000000003e-07, "loss": 0.0046, "step": 49039 }, { "epoch": 2.614490590179666, "grad_norm": 2.661527395248413, "learning_rate": 1.9220000000000001e-07, "loss": 0.0169, "step": 49040 }, { "epoch": 2.614543903609319, "grad_norm": 3.4950687885284424, "learning_rate": 1.92e-07, "loss": 0.0249, "step": 49041 }, { "epoch": 2.614597217038972, "grad_norm": 2.386721611022949, "learning_rate": 1.918e-07, "loss": 0.0087, "step": 49042 }, { "epoch": 2.614650530468625, "grad_norm": 2.9750845432281494, "learning_rate": 1.9160000000000002e-07, "loss": 0.0081, "step": 49043 }, { "epoch": 2.614703843898278, "grad_norm": 0.4161934554576874, "learning_rate": 1.9140000000000003e-07, "loss": 0.003, "step": 49044 }, { "epoch": 2.614757157327931, "grad_norm": 0.2731388211250305, "learning_rate": 1.9120000000000004e-07, "loss": 0.0022, "step": 49045 }, { "epoch": 2.614810470757584, "grad_norm": 1.3890899419784546, "learning_rate": 1.91e-07, "loss": 0.0392, "step": 49046 }, { "epoch": 2.6148637841872366, "grad_norm": 3.040358304977417, "learning_rate": 1.9080000000000001e-07, "loss": 0.0209, "step": 49047 }, { "epoch": 2.61491709761689, "grad_norm": 3.0784010887145996, "learning_rate": 1.9060000000000003e-07, "loss": 0.0167, "step": 49048 }, { "epoch": 2.6149704110465426, "grad_norm": 0.702597975730896, "learning_rate": 1.9040000000000004e-07, "loss": 0.0046, "step": 49049 }, { "epoch": 2.6150237244761954, "grad_norm": 1.4393056631088257, "learning_rate": 1.902e-07, "loss": 0.0172, "step": 49050 }, { "epoch": 2.6150770379058486, "grad_norm": 1.8986750841140747, "learning_rate": 1.9e-07, "loss": 0.0296, "step": 49051 }, { "epoch": 2.6151303513355013, "grad_norm": 1.4017168283462524, "learning_rate": 1.8980000000000002e-07, "loss": 0.0228, "step": 49052 }, { "epoch": 2.615183664765154, "grad_norm": 1.6133443117141724, "learning_rate": 1.8960000000000003e-07, "loss": 0.0168, "step": 49053 }, { "epoch": 2.6152369781948073, "grad_norm": 0.9576521515846252, "learning_rate": 1.894e-07, "loss": 0.0032, "step": 49054 }, { "epoch": 2.6152902916244605, "grad_norm": 1.5033726692199707, "learning_rate": 1.892e-07, "loss": 0.0203, "step": 49055 }, { "epoch": 2.6153436050541132, "grad_norm": 0.6215567588806152, "learning_rate": 1.89e-07, "loss": 0.0033, "step": 49056 }, { "epoch": 2.615396918483766, "grad_norm": 2.0863125324249268, "learning_rate": 1.8880000000000002e-07, "loss": 0.0095, "step": 49057 }, { "epoch": 2.615450231913419, "grad_norm": 2.45218825340271, "learning_rate": 1.886e-07, "loss": 0.0251, "step": 49058 }, { "epoch": 2.615503545343072, "grad_norm": 0.2550164461135864, "learning_rate": 1.8840000000000002e-07, "loss": 0.0017, "step": 49059 }, { "epoch": 2.6155568587727247, "grad_norm": 2.791292190551758, "learning_rate": 1.882e-07, "loss": 0.0309, "step": 49060 }, { "epoch": 2.615610172202378, "grad_norm": 0.5791634917259216, "learning_rate": 1.8800000000000002e-07, "loss": 0.0037, "step": 49061 }, { "epoch": 2.6156634856320307, "grad_norm": 1.7837659120559692, "learning_rate": 1.8780000000000003e-07, "loss": 0.0135, "step": 49062 }, { "epoch": 2.6157167990616834, "grad_norm": 1.116592288017273, "learning_rate": 1.876e-07, "loss": 0.0049, "step": 49063 }, { "epoch": 2.6157701124913366, "grad_norm": 1.8737417459487915, "learning_rate": 1.8740000000000002e-07, "loss": 0.0056, "step": 49064 }, { "epoch": 2.6158234259209894, "grad_norm": 1.987776279449463, "learning_rate": 1.872e-07, "loss": 0.0167, "step": 49065 }, { "epoch": 2.6158767393506426, "grad_norm": 1.824602484703064, "learning_rate": 1.8700000000000002e-07, "loss": 0.0045, "step": 49066 }, { "epoch": 2.6159300527802953, "grad_norm": 2.2028043270111084, "learning_rate": 1.868e-07, "loss": 0.0106, "step": 49067 }, { "epoch": 2.6159833662099485, "grad_norm": 1.7910653352737427, "learning_rate": 1.8660000000000002e-07, "loss": 0.0081, "step": 49068 }, { "epoch": 2.6160366796396013, "grad_norm": 2.121872901916504, "learning_rate": 1.8640000000000003e-07, "loss": 0.0099, "step": 49069 }, { "epoch": 2.616089993069254, "grad_norm": 0.5492656230926514, "learning_rate": 1.8620000000000004e-07, "loss": 0.0036, "step": 49070 }, { "epoch": 2.6161433064989073, "grad_norm": 1.9999371767044067, "learning_rate": 1.86e-07, "loss": 0.0299, "step": 49071 }, { "epoch": 2.61619661992856, "grad_norm": 1.3173191547393799, "learning_rate": 1.858e-07, "loss": 0.0052, "step": 49072 }, { "epoch": 2.6162499333582128, "grad_norm": 0.8069007396697998, "learning_rate": 1.8560000000000002e-07, "loss": 0.0038, "step": 49073 }, { "epoch": 2.616303246787866, "grad_norm": 3.299421787261963, "learning_rate": 1.8540000000000003e-07, "loss": 0.0465, "step": 49074 }, { "epoch": 2.6163565602175187, "grad_norm": 1.5164477825164795, "learning_rate": 1.852e-07, "loss": 0.018, "step": 49075 }, { "epoch": 2.6164098736471715, "grad_norm": 3.2475197315216064, "learning_rate": 1.85e-07, "loss": 0.0162, "step": 49076 }, { "epoch": 2.6164631870768247, "grad_norm": 1.264593482017517, "learning_rate": 1.8480000000000001e-07, "loss": 0.0038, "step": 49077 }, { "epoch": 2.6165165005064774, "grad_norm": 0.8806896209716797, "learning_rate": 1.8460000000000003e-07, "loss": 0.0115, "step": 49078 }, { "epoch": 2.6165698139361306, "grad_norm": 1.549216866493225, "learning_rate": 1.8440000000000004e-07, "loss": 0.0065, "step": 49079 }, { "epoch": 2.6166231273657834, "grad_norm": 2.7558624744415283, "learning_rate": 1.842e-07, "loss": 0.0206, "step": 49080 }, { "epoch": 2.6166764407954366, "grad_norm": 2.4359560012817383, "learning_rate": 1.84e-07, "loss": 0.0093, "step": 49081 }, { "epoch": 2.6167297542250894, "grad_norm": 2.000047206878662, "learning_rate": 1.8380000000000002e-07, "loss": 0.0334, "step": 49082 }, { "epoch": 2.616783067654742, "grad_norm": 0.7304573655128479, "learning_rate": 1.8360000000000003e-07, "loss": 0.0033, "step": 49083 }, { "epoch": 2.6168363810843953, "grad_norm": 3.3798022270202637, "learning_rate": 1.8340000000000001e-07, "loss": 0.0141, "step": 49084 }, { "epoch": 2.616889694514048, "grad_norm": 4.177741527557373, "learning_rate": 1.832e-07, "loss": 0.0092, "step": 49085 }, { "epoch": 2.616943007943701, "grad_norm": 4.754517078399658, "learning_rate": 1.83e-07, "loss": 0.0167, "step": 49086 }, { "epoch": 2.616996321373354, "grad_norm": 0.48379209637641907, "learning_rate": 1.8280000000000002e-07, "loss": 0.0022, "step": 49087 }, { "epoch": 2.617049634803007, "grad_norm": 0.989300012588501, "learning_rate": 1.826e-07, "loss": 0.0215, "step": 49088 }, { "epoch": 2.6171029482326595, "grad_norm": 1.9189707040786743, "learning_rate": 1.8240000000000002e-07, "loss": 0.0334, "step": 49089 }, { "epoch": 2.6171562616623127, "grad_norm": 1.9812171459197998, "learning_rate": 1.822e-07, "loss": 0.0083, "step": 49090 }, { "epoch": 2.6172095750919655, "grad_norm": 1.9531193971633911, "learning_rate": 1.8200000000000002e-07, "loss": 0.0056, "step": 49091 }, { "epoch": 2.6172628885216187, "grad_norm": 2.598361015319824, "learning_rate": 1.8180000000000003e-07, "loss": 0.0059, "step": 49092 }, { "epoch": 2.6173162019512715, "grad_norm": 0.5309008955955505, "learning_rate": 1.816e-07, "loss": 0.0031, "step": 49093 }, { "epoch": 2.6173695153809247, "grad_norm": 2.2496232986450195, "learning_rate": 1.8140000000000002e-07, "loss": 0.0086, "step": 49094 }, { "epoch": 2.6174228288105774, "grad_norm": 1.8048845529556274, "learning_rate": 1.8120000000000003e-07, "loss": 0.0143, "step": 49095 }, { "epoch": 2.61747614224023, "grad_norm": 1.5205594301223755, "learning_rate": 1.8100000000000002e-07, "loss": 0.0113, "step": 49096 }, { "epoch": 2.6175294556698834, "grad_norm": 1.1398541927337646, "learning_rate": 1.808e-07, "loss": 0.0185, "step": 49097 }, { "epoch": 2.617582769099536, "grad_norm": 2.1494739055633545, "learning_rate": 1.8060000000000002e-07, "loss": 0.0102, "step": 49098 }, { "epoch": 2.617636082529189, "grad_norm": 2.866232395172119, "learning_rate": 1.8040000000000003e-07, "loss": 0.0129, "step": 49099 }, { "epoch": 2.617689395958842, "grad_norm": 0.35745248198509216, "learning_rate": 1.8020000000000004e-07, "loss": 0.0018, "step": 49100 }, { "epoch": 2.617742709388495, "grad_norm": 1.0973687171936035, "learning_rate": 1.8e-07, "loss": 0.0034, "step": 49101 }, { "epoch": 2.617796022818148, "grad_norm": 0.9755684733390808, "learning_rate": 1.798e-07, "loss": 0.0037, "step": 49102 }, { "epoch": 2.617849336247801, "grad_norm": 1.2355899810791016, "learning_rate": 1.7960000000000002e-07, "loss": 0.0061, "step": 49103 }, { "epoch": 2.617902649677454, "grad_norm": 2.3329851627349854, "learning_rate": 1.7940000000000003e-07, "loss": 0.0325, "step": 49104 }, { "epoch": 2.6179559631071068, "grad_norm": 2.3511900901794434, "learning_rate": 1.792e-07, "loss": 0.0269, "step": 49105 }, { "epoch": 2.6180092765367595, "grad_norm": 1.869105577468872, "learning_rate": 1.79e-07, "loss": 0.0078, "step": 49106 }, { "epoch": 2.6180625899664127, "grad_norm": 1.1730724573135376, "learning_rate": 1.788e-07, "loss": 0.0197, "step": 49107 }, { "epoch": 2.6181159033960655, "grad_norm": 0.6955938339233398, "learning_rate": 1.7860000000000002e-07, "loss": 0.0039, "step": 49108 }, { "epoch": 2.6181692168257182, "grad_norm": 1.0890963077545166, "learning_rate": 1.7840000000000004e-07, "loss": 0.0035, "step": 49109 }, { "epoch": 2.6182225302553714, "grad_norm": 0.4785134196281433, "learning_rate": 1.782e-07, "loss": 0.0034, "step": 49110 }, { "epoch": 2.618275843685024, "grad_norm": 1.7923791408538818, "learning_rate": 1.78e-07, "loss": 0.0116, "step": 49111 }, { "epoch": 2.618329157114677, "grad_norm": 2.3641891479492188, "learning_rate": 1.7780000000000002e-07, "loss": 0.0104, "step": 49112 }, { "epoch": 2.61838247054433, "grad_norm": 1.3730239868164062, "learning_rate": 1.7760000000000003e-07, "loss": 0.0041, "step": 49113 }, { "epoch": 2.618435783973983, "grad_norm": 1.6064010858535767, "learning_rate": 1.7740000000000001e-07, "loss": 0.017, "step": 49114 }, { "epoch": 2.618489097403636, "grad_norm": 2.1838784217834473, "learning_rate": 1.772e-07, "loss": 0.0379, "step": 49115 }, { "epoch": 2.618542410833289, "grad_norm": 3.3649110794067383, "learning_rate": 1.77e-07, "loss": 0.0149, "step": 49116 }, { "epoch": 2.618595724262942, "grad_norm": 1.5575535297393799, "learning_rate": 1.7680000000000002e-07, "loss": 0.009, "step": 49117 }, { "epoch": 2.618649037692595, "grad_norm": 2.605922222137451, "learning_rate": 1.766e-07, "loss": 0.0244, "step": 49118 }, { "epoch": 2.6187023511222476, "grad_norm": 1.472673773765564, "learning_rate": 1.7640000000000002e-07, "loss": 0.0053, "step": 49119 }, { "epoch": 2.618755664551901, "grad_norm": 2.326983690261841, "learning_rate": 1.7620000000000003e-07, "loss": 0.0084, "step": 49120 }, { "epoch": 2.6188089779815535, "grad_norm": 0.8127269148826599, "learning_rate": 1.7600000000000001e-07, "loss": 0.0051, "step": 49121 }, { "epoch": 2.6188622914112063, "grad_norm": 2.190603017807007, "learning_rate": 1.758e-07, "loss": 0.0436, "step": 49122 }, { "epoch": 2.6189156048408595, "grad_norm": 1.4620299339294434, "learning_rate": 1.756e-07, "loss": 0.0096, "step": 49123 }, { "epoch": 2.6189689182705123, "grad_norm": 2.261439323425293, "learning_rate": 1.7540000000000002e-07, "loss": 0.0085, "step": 49124 }, { "epoch": 2.619022231700165, "grad_norm": 2.0238125324249268, "learning_rate": 1.7520000000000003e-07, "loss": 0.0234, "step": 49125 }, { "epoch": 2.619075545129818, "grad_norm": 2.5335395336151123, "learning_rate": 1.7500000000000002e-07, "loss": 0.0277, "step": 49126 }, { "epoch": 2.619128858559471, "grad_norm": 2.604128360748291, "learning_rate": 1.748e-07, "loss": 0.0177, "step": 49127 }, { "epoch": 2.619182171989124, "grad_norm": 1.3175522089004517, "learning_rate": 1.7460000000000001e-07, "loss": 0.031, "step": 49128 }, { "epoch": 2.619235485418777, "grad_norm": 1.750754714012146, "learning_rate": 1.7440000000000003e-07, "loss": 0.0167, "step": 49129 }, { "epoch": 2.61928879884843, "grad_norm": 0.23591284453868866, "learning_rate": 1.7420000000000004e-07, "loss": 0.0016, "step": 49130 }, { "epoch": 2.619342112278083, "grad_norm": 2.004366159439087, "learning_rate": 1.74e-07, "loss": 0.0067, "step": 49131 }, { "epoch": 2.6193954257077356, "grad_norm": 3.7749855518341064, "learning_rate": 1.738e-07, "loss": 0.0144, "step": 49132 }, { "epoch": 2.619448739137389, "grad_norm": 2.4745352268218994, "learning_rate": 1.7360000000000002e-07, "loss": 0.0117, "step": 49133 }, { "epoch": 2.6195020525670416, "grad_norm": 3.879798412322998, "learning_rate": 1.7340000000000003e-07, "loss": 0.0233, "step": 49134 }, { "epoch": 2.6195553659966944, "grad_norm": 1.8978573083877563, "learning_rate": 1.732e-07, "loss": 0.008, "step": 49135 }, { "epoch": 2.6196086794263476, "grad_norm": 1.0614296197891235, "learning_rate": 1.73e-07, "loss": 0.0054, "step": 49136 }, { "epoch": 2.6196619928560003, "grad_norm": 2.594660520553589, "learning_rate": 1.728e-07, "loss": 0.015, "step": 49137 }, { "epoch": 2.6197153062856535, "grad_norm": 1.2217228412628174, "learning_rate": 1.7260000000000002e-07, "loss": 0.0131, "step": 49138 }, { "epoch": 2.6197686197153063, "grad_norm": 3.1800127029418945, "learning_rate": 1.724e-07, "loss": 0.0461, "step": 49139 }, { "epoch": 2.6198219331449595, "grad_norm": 3.7104651927948, "learning_rate": 1.722e-07, "loss": 0.0089, "step": 49140 }, { "epoch": 2.6198752465746122, "grad_norm": 1.7646654844284058, "learning_rate": 1.72e-07, "loss": 0.0118, "step": 49141 }, { "epoch": 2.619928560004265, "grad_norm": 2.9396848678588867, "learning_rate": 1.7180000000000002e-07, "loss": 0.0155, "step": 49142 }, { "epoch": 2.619981873433918, "grad_norm": 1.0519118309020996, "learning_rate": 1.7160000000000003e-07, "loss": 0.005, "step": 49143 }, { "epoch": 2.620035186863571, "grad_norm": 0.7810713648796082, "learning_rate": 1.714e-07, "loss": 0.0039, "step": 49144 }, { "epoch": 2.6200885002932237, "grad_norm": 0.4913504123687744, "learning_rate": 1.7120000000000002e-07, "loss": 0.0024, "step": 49145 }, { "epoch": 2.620141813722877, "grad_norm": 2.116525888442993, "learning_rate": 1.71e-07, "loss": 0.0153, "step": 49146 }, { "epoch": 2.6201951271525297, "grad_norm": 1.3764876127243042, "learning_rate": 1.7080000000000002e-07, "loss": 0.0059, "step": 49147 }, { "epoch": 2.6202484405821824, "grad_norm": 0.6788245439529419, "learning_rate": 1.706e-07, "loss": 0.0033, "step": 49148 }, { "epoch": 2.6203017540118356, "grad_norm": 1.6966344118118286, "learning_rate": 1.7040000000000002e-07, "loss": 0.0044, "step": 49149 }, { "epoch": 2.6203550674414884, "grad_norm": 2.3409082889556885, "learning_rate": 1.7020000000000003e-07, "loss": 0.0264, "step": 49150 }, { "epoch": 2.6204083808711416, "grad_norm": 1.833857536315918, "learning_rate": 1.7000000000000001e-07, "loss": 0.031, "step": 49151 }, { "epoch": 2.6204616943007943, "grad_norm": 2.3382649421691895, "learning_rate": 1.698e-07, "loss": 0.0145, "step": 49152 }, { "epoch": 2.6205150077304475, "grad_norm": 1.503515362739563, "learning_rate": 1.696e-07, "loss": 0.0069, "step": 49153 }, { "epoch": 2.6205683211601003, "grad_norm": 1.9851375818252563, "learning_rate": 1.6940000000000002e-07, "loss": 0.0235, "step": 49154 }, { "epoch": 2.620621634589753, "grad_norm": 2.0066514015197754, "learning_rate": 1.6920000000000003e-07, "loss": 0.0088, "step": 49155 }, { "epoch": 2.6206749480194063, "grad_norm": 1.6722891330718994, "learning_rate": 1.69e-07, "loss": 0.0091, "step": 49156 }, { "epoch": 2.620728261449059, "grad_norm": 2.0047574043273926, "learning_rate": 1.688e-07, "loss": 0.0134, "step": 49157 }, { "epoch": 2.6207815748787118, "grad_norm": 1.5435826778411865, "learning_rate": 1.6860000000000001e-07, "loss": 0.0053, "step": 49158 }, { "epoch": 2.620834888308365, "grad_norm": 4.66132116317749, "learning_rate": 1.6840000000000003e-07, "loss": 0.0285, "step": 49159 }, { "epoch": 2.6208882017380177, "grad_norm": 1.0293381214141846, "learning_rate": 1.6820000000000004e-07, "loss": 0.021, "step": 49160 }, { "epoch": 2.6209415151676705, "grad_norm": 1.401812195777893, "learning_rate": 1.68e-07, "loss": 0.0039, "step": 49161 }, { "epoch": 2.6209948285973237, "grad_norm": 2.2807626724243164, "learning_rate": 1.678e-07, "loss": 0.0098, "step": 49162 }, { "epoch": 2.6210481420269764, "grad_norm": 3.5278191566467285, "learning_rate": 1.6760000000000002e-07, "loss": 0.0187, "step": 49163 }, { "epoch": 2.6211014554566296, "grad_norm": 1.2810699939727783, "learning_rate": 1.6740000000000003e-07, "loss": 0.0042, "step": 49164 }, { "epoch": 2.6211547688862824, "grad_norm": 2.8840997219085693, "learning_rate": 1.672e-07, "loss": 0.0195, "step": 49165 }, { "epoch": 2.6212080823159356, "grad_norm": 1.7940123081207275, "learning_rate": 1.67e-07, "loss": 0.0064, "step": 49166 }, { "epoch": 2.6212613957455884, "grad_norm": 1.3264394998550415, "learning_rate": 1.668e-07, "loss": 0.0094, "step": 49167 }, { "epoch": 2.621314709175241, "grad_norm": 0.9077911376953125, "learning_rate": 1.6660000000000002e-07, "loss": 0.0045, "step": 49168 }, { "epoch": 2.6213680226048943, "grad_norm": 2.2539889812469482, "learning_rate": 1.664e-07, "loss": 0.0169, "step": 49169 }, { "epoch": 2.621421336034547, "grad_norm": 2.4749999046325684, "learning_rate": 1.6620000000000002e-07, "loss": 0.0274, "step": 49170 }, { "epoch": 2.6214746494642, "grad_norm": 1.3690550327301025, "learning_rate": 1.66e-07, "loss": 0.0035, "step": 49171 }, { "epoch": 2.621527962893853, "grad_norm": 1.5256156921386719, "learning_rate": 1.6580000000000002e-07, "loss": 0.0388, "step": 49172 }, { "epoch": 2.621581276323506, "grad_norm": 1.514730453491211, "learning_rate": 1.656e-07, "loss": 0.0128, "step": 49173 }, { "epoch": 2.6216345897531586, "grad_norm": 2.5593526363372803, "learning_rate": 1.654e-07, "loss": 0.0063, "step": 49174 }, { "epoch": 2.6216879031828118, "grad_norm": 5.805052757263184, "learning_rate": 1.6520000000000002e-07, "loss": 0.0251, "step": 49175 }, { "epoch": 2.621741216612465, "grad_norm": 3.0631210803985596, "learning_rate": 1.65e-07, "loss": 0.0073, "step": 49176 }, { "epoch": 2.6217945300421177, "grad_norm": 1.5823378562927246, "learning_rate": 1.6480000000000002e-07, "loss": 0.0066, "step": 49177 }, { "epoch": 2.6218478434717705, "grad_norm": 0.37319159507751465, "learning_rate": 1.646e-07, "loss": 0.0017, "step": 49178 }, { "epoch": 2.6219011569014237, "grad_norm": 2.808638572692871, "learning_rate": 1.6440000000000002e-07, "loss": 0.0166, "step": 49179 }, { "epoch": 2.6219544703310764, "grad_norm": 2.910392999649048, "learning_rate": 1.6420000000000003e-07, "loss": 0.0321, "step": 49180 }, { "epoch": 2.622007783760729, "grad_norm": 2.6408071517944336, "learning_rate": 1.6400000000000004e-07, "loss": 0.0359, "step": 49181 }, { "epoch": 2.6220610971903824, "grad_norm": 1.8751885890960693, "learning_rate": 1.638e-07, "loss": 0.0084, "step": 49182 }, { "epoch": 2.622114410620035, "grad_norm": 1.4227869510650635, "learning_rate": 1.636e-07, "loss": 0.0152, "step": 49183 }, { "epoch": 2.622167724049688, "grad_norm": 1.7175912857055664, "learning_rate": 1.6340000000000002e-07, "loss": 0.0053, "step": 49184 }, { "epoch": 2.622221037479341, "grad_norm": 1.8358166217803955, "learning_rate": 1.6320000000000003e-07, "loss": 0.0071, "step": 49185 }, { "epoch": 2.622274350908994, "grad_norm": 3.130763292312622, "learning_rate": 1.63e-07, "loss": 0.0088, "step": 49186 }, { "epoch": 2.622327664338647, "grad_norm": 2.712024688720703, "learning_rate": 1.628e-07, "loss": 0.0275, "step": 49187 }, { "epoch": 2.6223809777683, "grad_norm": 1.5986665487289429, "learning_rate": 1.6260000000000001e-07, "loss": 0.0228, "step": 49188 }, { "epoch": 2.622434291197953, "grad_norm": 2.207080125808716, "learning_rate": 1.6240000000000002e-07, "loss": 0.0114, "step": 49189 }, { "epoch": 2.6224876046276058, "grad_norm": 2.115293502807617, "learning_rate": 1.6219999999999998e-07, "loss": 0.0171, "step": 49190 }, { "epoch": 2.6225409180572585, "grad_norm": 2.0332558155059814, "learning_rate": 1.62e-07, "loss": 0.0087, "step": 49191 }, { "epoch": 2.6225942314869117, "grad_norm": 2.0806992053985596, "learning_rate": 1.618e-07, "loss": 0.0426, "step": 49192 }, { "epoch": 2.6226475449165645, "grad_norm": 1.4664143323898315, "learning_rate": 1.6160000000000002e-07, "loss": 0.0278, "step": 49193 }, { "epoch": 2.6227008583462172, "grad_norm": 1.8582571744918823, "learning_rate": 1.6140000000000003e-07, "loss": 0.0174, "step": 49194 }, { "epoch": 2.6227541717758704, "grad_norm": 1.4642821550369263, "learning_rate": 1.6120000000000001e-07, "loss": 0.0049, "step": 49195 }, { "epoch": 2.622807485205523, "grad_norm": 1.859261155128479, "learning_rate": 1.61e-07, "loss": 0.0248, "step": 49196 }, { "epoch": 2.622860798635176, "grad_norm": 0.609086811542511, "learning_rate": 1.608e-07, "loss": 0.0033, "step": 49197 }, { "epoch": 2.622914112064829, "grad_norm": 0.9985894560813904, "learning_rate": 1.6060000000000002e-07, "loss": 0.0047, "step": 49198 }, { "epoch": 2.622967425494482, "grad_norm": 0.5347952246665955, "learning_rate": 1.604e-07, "loss": 0.0028, "step": 49199 }, { "epoch": 2.623020738924135, "grad_norm": 0.37227460741996765, "learning_rate": 1.6020000000000002e-07, "loss": 0.0026, "step": 49200 }, { "epoch": 2.623074052353788, "grad_norm": 3.17454195022583, "learning_rate": 1.6e-07, "loss": 0.0288, "step": 49201 }, { "epoch": 2.623127365783441, "grad_norm": 0.6092604994773865, "learning_rate": 1.5980000000000001e-07, "loss": 0.0031, "step": 49202 }, { "epoch": 2.623180679213094, "grad_norm": 0.48756077885627747, "learning_rate": 1.596e-07, "loss": 0.0021, "step": 49203 }, { "epoch": 2.6232339926427466, "grad_norm": 1.0996670722961426, "learning_rate": 1.594e-07, "loss": 0.0205, "step": 49204 }, { "epoch": 2.6232873060724, "grad_norm": 0.7913315296173096, "learning_rate": 1.5920000000000002e-07, "loss": 0.0036, "step": 49205 }, { "epoch": 2.6233406195020526, "grad_norm": 3.715085983276367, "learning_rate": 1.5900000000000003e-07, "loss": 0.0433, "step": 49206 }, { "epoch": 2.6233939329317053, "grad_norm": 1.9057410955429077, "learning_rate": 1.588e-07, "loss": 0.0152, "step": 49207 }, { "epoch": 2.6234472463613585, "grad_norm": 2.1643569469451904, "learning_rate": 1.586e-07, "loss": 0.0186, "step": 49208 }, { "epoch": 2.6235005597910113, "grad_norm": 1.1418471336364746, "learning_rate": 1.5840000000000002e-07, "loss": 0.0149, "step": 49209 }, { "epoch": 2.623553873220664, "grad_norm": 2.1899077892303467, "learning_rate": 1.5820000000000003e-07, "loss": 0.0081, "step": 49210 }, { "epoch": 2.6236071866503172, "grad_norm": 0.20903034508228302, "learning_rate": 1.5800000000000004e-07, "loss": 0.0016, "step": 49211 }, { "epoch": 2.62366050007997, "grad_norm": 0.8894374370574951, "learning_rate": 1.578e-07, "loss": 0.0141, "step": 49212 }, { "epoch": 2.623713813509623, "grad_norm": 2.3338589668273926, "learning_rate": 1.576e-07, "loss": 0.0163, "step": 49213 }, { "epoch": 2.623767126939276, "grad_norm": 3.0887293815612793, "learning_rate": 1.5740000000000002e-07, "loss": 0.0115, "step": 49214 }, { "epoch": 2.623820440368929, "grad_norm": 0.5978005528450012, "learning_rate": 1.5720000000000003e-07, "loss": 0.0021, "step": 49215 }, { "epoch": 2.623873753798582, "grad_norm": 2.41513729095459, "learning_rate": 1.57e-07, "loss": 0.0129, "step": 49216 }, { "epoch": 2.6239270672282347, "grad_norm": 2.3510773181915283, "learning_rate": 1.568e-07, "loss": 0.0422, "step": 49217 }, { "epoch": 2.623980380657888, "grad_norm": 1.583777666091919, "learning_rate": 1.566e-07, "loss": 0.0204, "step": 49218 }, { "epoch": 2.6240336940875406, "grad_norm": 7.360172271728516, "learning_rate": 1.5640000000000002e-07, "loss": 0.0164, "step": 49219 }, { "epoch": 2.6240870075171934, "grad_norm": 5.999633312225342, "learning_rate": 1.562e-07, "loss": 0.0257, "step": 49220 }, { "epoch": 2.6241403209468466, "grad_norm": 2.0809476375579834, "learning_rate": 1.56e-07, "loss": 0.0294, "step": 49221 }, { "epoch": 2.6241936343764993, "grad_norm": 2.4287827014923096, "learning_rate": 1.558e-07, "loss": 0.0222, "step": 49222 }, { "epoch": 2.6242469478061525, "grad_norm": 2.938619613647461, "learning_rate": 1.5560000000000002e-07, "loss": 0.0108, "step": 49223 }, { "epoch": 2.6243002612358053, "grad_norm": 0.45103269815444946, "learning_rate": 1.554e-07, "loss": 0.0024, "step": 49224 }, { "epoch": 2.6243535746654585, "grad_norm": 3.874626874923706, "learning_rate": 1.5520000000000001e-07, "loss": 0.0425, "step": 49225 }, { "epoch": 2.6244068880951112, "grad_norm": 0.4927818179130554, "learning_rate": 1.5500000000000002e-07, "loss": 0.0027, "step": 49226 }, { "epoch": 2.624460201524764, "grad_norm": 0.565451979637146, "learning_rate": 1.548e-07, "loss": 0.0104, "step": 49227 }, { "epoch": 2.624513514954417, "grad_norm": 5.397843837738037, "learning_rate": 1.5460000000000002e-07, "loss": 0.0451, "step": 49228 }, { "epoch": 2.62456682838407, "grad_norm": 2.8457274436950684, "learning_rate": 1.544e-07, "loss": 0.0148, "step": 49229 }, { "epoch": 2.6246201418137227, "grad_norm": 2.4640603065490723, "learning_rate": 1.5420000000000002e-07, "loss": 0.0083, "step": 49230 }, { "epoch": 2.624673455243376, "grad_norm": 1.5365501642227173, "learning_rate": 1.5400000000000003e-07, "loss": 0.0088, "step": 49231 }, { "epoch": 2.6247267686730287, "grad_norm": 2.2216970920562744, "learning_rate": 1.5380000000000001e-07, "loss": 0.0217, "step": 49232 }, { "epoch": 2.6247800821026814, "grad_norm": 2.2084715366363525, "learning_rate": 1.5360000000000003e-07, "loss": 0.0209, "step": 49233 }, { "epoch": 2.6248333955323346, "grad_norm": 0.3443472683429718, "learning_rate": 1.534e-07, "loss": 0.0019, "step": 49234 }, { "epoch": 2.6248867089619874, "grad_norm": 3.336394786834717, "learning_rate": 1.5320000000000002e-07, "loss": 0.0084, "step": 49235 }, { "epoch": 2.6249400223916406, "grad_norm": 1.5014028549194336, "learning_rate": 1.53e-07, "loss": 0.0057, "step": 49236 }, { "epoch": 2.6249933358212933, "grad_norm": 0.6402746438980103, "learning_rate": 1.5280000000000002e-07, "loss": 0.0041, "step": 49237 }, { "epoch": 2.6250466492509466, "grad_norm": 0.8882389068603516, "learning_rate": 1.526e-07, "loss": 0.0041, "step": 49238 }, { "epoch": 2.6250999626805993, "grad_norm": 1.072142481803894, "learning_rate": 1.5240000000000001e-07, "loss": 0.0165, "step": 49239 }, { "epoch": 2.625153276110252, "grad_norm": 1.4858931303024292, "learning_rate": 1.522e-07, "loss": 0.0055, "step": 49240 }, { "epoch": 2.6252065895399053, "grad_norm": 2.5923051834106445, "learning_rate": 1.52e-07, "loss": 0.0331, "step": 49241 }, { "epoch": 2.625259902969558, "grad_norm": 1.3166992664337158, "learning_rate": 1.5180000000000002e-07, "loss": 0.0054, "step": 49242 }, { "epoch": 2.625313216399211, "grad_norm": 1.341378927230835, "learning_rate": 1.516e-07, "loss": 0.0197, "step": 49243 }, { "epoch": 2.625366529828864, "grad_norm": 0.598665177822113, "learning_rate": 1.5140000000000002e-07, "loss": 0.003, "step": 49244 }, { "epoch": 2.6254198432585167, "grad_norm": 1.3859738111495972, "learning_rate": 1.512e-07, "loss": 0.0159, "step": 49245 }, { "epoch": 2.6254731566881695, "grad_norm": 1.9108184576034546, "learning_rate": 1.5100000000000002e-07, "loss": 0.0303, "step": 49246 }, { "epoch": 2.6255264701178227, "grad_norm": 2.313112735748291, "learning_rate": 1.508e-07, "loss": 0.0066, "step": 49247 }, { "epoch": 2.6255797835474755, "grad_norm": 0.9101638197898865, "learning_rate": 1.506e-07, "loss": 0.0183, "step": 49248 }, { "epoch": 2.6256330969771287, "grad_norm": 0.8350411653518677, "learning_rate": 1.504e-07, "loss": 0.0125, "step": 49249 }, { "epoch": 2.6256864104067814, "grad_norm": 0.23638001084327698, "learning_rate": 1.502e-07, "loss": 0.0022, "step": 49250 }, { "epoch": 2.6257397238364346, "grad_norm": 4.454234600067139, "learning_rate": 1.5000000000000002e-07, "loss": 0.0136, "step": 49251 }, { "epoch": 2.6257930372660874, "grad_norm": 0.7615798115730286, "learning_rate": 1.498e-07, "loss": 0.0028, "step": 49252 }, { "epoch": 2.62584635069574, "grad_norm": 2.4828505516052246, "learning_rate": 1.4960000000000002e-07, "loss": 0.0106, "step": 49253 }, { "epoch": 2.6258996641253933, "grad_norm": 1.6676924228668213, "learning_rate": 1.494e-07, "loss": 0.0123, "step": 49254 }, { "epoch": 2.625952977555046, "grad_norm": 0.37890079617500305, "learning_rate": 1.492e-07, "loss": 0.0024, "step": 49255 }, { "epoch": 2.626006290984699, "grad_norm": 1.856616497039795, "learning_rate": 1.4900000000000002e-07, "loss": 0.0118, "step": 49256 }, { "epoch": 2.626059604414352, "grad_norm": 2.4434754848480225, "learning_rate": 1.488e-07, "loss": 0.0129, "step": 49257 }, { "epoch": 2.626112917844005, "grad_norm": 0.9900218844413757, "learning_rate": 1.4860000000000002e-07, "loss": 0.0187, "step": 49258 }, { "epoch": 2.6261662312736576, "grad_norm": 0.5647398829460144, "learning_rate": 1.4840000000000003e-07, "loss": 0.0041, "step": 49259 }, { "epoch": 2.6262195447033108, "grad_norm": 2.1451234817504883, "learning_rate": 1.4820000000000002e-07, "loss": 0.0124, "step": 49260 }, { "epoch": 2.626272858132964, "grad_norm": 3.2517590522766113, "learning_rate": 1.4800000000000003e-07, "loss": 0.0165, "step": 49261 }, { "epoch": 2.6263261715626167, "grad_norm": 1.586700439453125, "learning_rate": 1.478e-07, "loss": 0.0146, "step": 49262 }, { "epoch": 2.6263794849922695, "grad_norm": 0.9401281476020813, "learning_rate": 1.4760000000000002e-07, "loss": 0.0129, "step": 49263 }, { "epoch": 2.6264327984219227, "grad_norm": 2.663627862930298, "learning_rate": 1.474e-07, "loss": 0.0269, "step": 49264 }, { "epoch": 2.6264861118515754, "grad_norm": 2.2936654090881348, "learning_rate": 1.4720000000000002e-07, "loss": 0.0271, "step": 49265 }, { "epoch": 2.626539425281228, "grad_norm": 3.654893398284912, "learning_rate": 1.47e-07, "loss": 0.0202, "step": 49266 }, { "epoch": 2.6265927387108814, "grad_norm": 1.552981972694397, "learning_rate": 1.4680000000000002e-07, "loss": 0.0055, "step": 49267 }, { "epoch": 2.626646052140534, "grad_norm": 1.2391070127487183, "learning_rate": 1.466e-07, "loss": 0.0192, "step": 49268 }, { "epoch": 2.626699365570187, "grad_norm": 2.3023440837860107, "learning_rate": 1.4640000000000001e-07, "loss": 0.0318, "step": 49269 }, { "epoch": 2.62675267899984, "grad_norm": 2.1114301681518555, "learning_rate": 1.462e-07, "loss": 0.0403, "step": 49270 }, { "epoch": 2.626805992429493, "grad_norm": 2.0319786071777344, "learning_rate": 1.46e-07, "loss": 0.0179, "step": 49271 }, { "epoch": 2.626859305859146, "grad_norm": 3.9898107051849365, "learning_rate": 1.458e-07, "loss": 0.0119, "step": 49272 }, { "epoch": 2.626912619288799, "grad_norm": 2.334200382232666, "learning_rate": 1.456e-07, "loss": 0.0252, "step": 49273 }, { "epoch": 2.626965932718452, "grad_norm": 1.5242993831634521, "learning_rate": 1.4540000000000002e-07, "loss": 0.0311, "step": 49274 }, { "epoch": 2.627019246148105, "grad_norm": 0.7502093315124512, "learning_rate": 1.452e-07, "loss": 0.0131, "step": 49275 }, { "epoch": 2.6270725595777575, "grad_norm": 2.6891977787017822, "learning_rate": 1.4500000000000001e-07, "loss": 0.0318, "step": 49276 }, { "epoch": 2.6271258730074107, "grad_norm": 1.5928953886032104, "learning_rate": 1.448e-07, "loss": 0.007, "step": 49277 }, { "epoch": 2.6271791864370635, "grad_norm": 1.55851411819458, "learning_rate": 1.446e-07, "loss": 0.0068, "step": 49278 }, { "epoch": 2.6272324998667163, "grad_norm": 2.738189220428467, "learning_rate": 1.444e-07, "loss": 0.0184, "step": 49279 }, { "epoch": 2.6272858132963695, "grad_norm": 2.809091806411743, "learning_rate": 1.442e-07, "loss": 0.0372, "step": 49280 }, { "epoch": 2.627339126726022, "grad_norm": 1.7669546604156494, "learning_rate": 1.4400000000000002e-07, "loss": 0.033, "step": 49281 }, { "epoch": 2.627392440155675, "grad_norm": 1.7255531549453735, "learning_rate": 1.438e-07, "loss": 0.017, "step": 49282 }, { "epoch": 2.627445753585328, "grad_norm": 2.702711343765259, "learning_rate": 1.4360000000000002e-07, "loss": 0.0052, "step": 49283 }, { "epoch": 2.627499067014981, "grad_norm": 3.5934689044952393, "learning_rate": 1.4340000000000003e-07, "loss": 0.0179, "step": 49284 }, { "epoch": 2.627552380444634, "grad_norm": 0.8415267467498779, "learning_rate": 1.432e-07, "loss": 0.0214, "step": 49285 }, { "epoch": 2.627605693874287, "grad_norm": 2.2690486907958984, "learning_rate": 1.4300000000000002e-07, "loss": 0.0058, "step": 49286 }, { "epoch": 2.62765900730394, "grad_norm": 2.124873161315918, "learning_rate": 1.428e-07, "loss": 0.0103, "step": 49287 }, { "epoch": 2.627712320733593, "grad_norm": 3.3726696968078613, "learning_rate": 1.4260000000000002e-07, "loss": 0.0139, "step": 49288 }, { "epoch": 2.6277656341632456, "grad_norm": 5.870667934417725, "learning_rate": 1.424e-07, "loss": 0.0163, "step": 49289 }, { "epoch": 2.627818947592899, "grad_norm": 2.1151630878448486, "learning_rate": 1.4220000000000002e-07, "loss": 0.0158, "step": 49290 }, { "epoch": 2.6278722610225516, "grad_norm": 2.816582202911377, "learning_rate": 1.4200000000000003e-07, "loss": 0.0158, "step": 49291 }, { "epoch": 2.6279255744522043, "grad_norm": 0.8704649806022644, "learning_rate": 1.418e-07, "loss": 0.0055, "step": 49292 }, { "epoch": 2.6279788878818575, "grad_norm": 3.4172348976135254, "learning_rate": 1.4160000000000002e-07, "loss": 0.0094, "step": 49293 }, { "epoch": 2.6280322013115103, "grad_norm": 1.2181415557861328, "learning_rate": 1.414e-07, "loss": 0.0145, "step": 49294 }, { "epoch": 2.628085514741163, "grad_norm": 0.6717051863670349, "learning_rate": 1.4120000000000002e-07, "loss": 0.0029, "step": 49295 }, { "epoch": 2.6281388281708162, "grad_norm": 1.5522561073303223, "learning_rate": 1.41e-07, "loss": 0.0119, "step": 49296 }, { "epoch": 2.6281921416004694, "grad_norm": 2.8376667499542236, "learning_rate": 1.4080000000000002e-07, "loss": 0.0287, "step": 49297 }, { "epoch": 2.628245455030122, "grad_norm": 0.29669296741485596, "learning_rate": 1.406e-07, "loss": 0.0025, "step": 49298 }, { "epoch": 2.628298768459775, "grad_norm": 1.1601781845092773, "learning_rate": 1.404e-07, "loss": 0.0281, "step": 49299 }, { "epoch": 2.628352081889428, "grad_norm": 0.30635809898376465, "learning_rate": 1.402e-07, "loss": 0.0024, "step": 49300 }, { "epoch": 2.628405395319081, "grad_norm": 3.196683645248413, "learning_rate": 1.4e-07, "loss": 0.0528, "step": 49301 }, { "epoch": 2.6284587087487337, "grad_norm": 2.0180649757385254, "learning_rate": 1.398e-07, "loss": 0.034, "step": 49302 }, { "epoch": 2.628512022178387, "grad_norm": 2.7650861740112305, "learning_rate": 1.396e-07, "loss": 0.0125, "step": 49303 }, { "epoch": 2.6285653356080396, "grad_norm": 0.7656557559967041, "learning_rate": 1.394e-07, "loss": 0.0038, "step": 49304 }, { "epoch": 2.6286186490376924, "grad_norm": 0.133967325091362, "learning_rate": 1.392e-07, "loss": 0.0013, "step": 49305 }, { "epoch": 2.6286719624673456, "grad_norm": 1.2587018013000488, "learning_rate": 1.3900000000000001e-07, "loss": 0.0093, "step": 49306 }, { "epoch": 2.6287252758969983, "grad_norm": 1.846186876296997, "learning_rate": 1.388e-07, "loss": 0.0141, "step": 49307 }, { "epoch": 2.6287785893266515, "grad_norm": 0.5297234058380127, "learning_rate": 1.386e-07, "loss": 0.0026, "step": 49308 }, { "epoch": 2.6288319027563043, "grad_norm": 1.4012490510940552, "learning_rate": 1.3840000000000002e-07, "loss": 0.0045, "step": 49309 }, { "epoch": 2.6288852161859575, "grad_norm": 3.0876238346099854, "learning_rate": 1.382e-07, "loss": 0.0208, "step": 49310 }, { "epoch": 2.6289385296156103, "grad_norm": 0.5665060877799988, "learning_rate": 1.3800000000000002e-07, "loss": 0.003, "step": 49311 }, { "epoch": 2.628991843045263, "grad_norm": 2.3035647869110107, "learning_rate": 1.3780000000000003e-07, "loss": 0.0088, "step": 49312 }, { "epoch": 2.629045156474916, "grad_norm": 2.66296124458313, "learning_rate": 1.3760000000000001e-07, "loss": 0.0042, "step": 49313 }, { "epoch": 2.629098469904569, "grad_norm": 3.0177810192108154, "learning_rate": 1.3740000000000003e-07, "loss": 0.0237, "step": 49314 }, { "epoch": 2.6291517833342217, "grad_norm": 0.923149049282074, "learning_rate": 1.372e-07, "loss": 0.0039, "step": 49315 }, { "epoch": 2.629205096763875, "grad_norm": 1.0552351474761963, "learning_rate": 1.3700000000000002e-07, "loss": 0.0049, "step": 49316 }, { "epoch": 2.6292584101935277, "grad_norm": 1.9202696084976196, "learning_rate": 1.368e-07, "loss": 0.0113, "step": 49317 }, { "epoch": 2.6293117236231804, "grad_norm": 2.8075146675109863, "learning_rate": 1.3660000000000002e-07, "loss": 0.032, "step": 49318 }, { "epoch": 2.6293650370528336, "grad_norm": 2.415663003921509, "learning_rate": 1.364e-07, "loss": 0.038, "step": 49319 }, { "epoch": 2.6294183504824864, "grad_norm": 1.9763898849487305, "learning_rate": 1.3620000000000002e-07, "loss": 0.0066, "step": 49320 }, { "epoch": 2.6294716639121396, "grad_norm": 2.620323657989502, "learning_rate": 1.36e-07, "loss": 0.0358, "step": 49321 }, { "epoch": 2.6295249773417924, "grad_norm": 2.7100870609283447, "learning_rate": 1.358e-07, "loss": 0.0084, "step": 49322 }, { "epoch": 2.6295782907714456, "grad_norm": 2.406977415084839, "learning_rate": 1.356e-07, "loss": 0.0139, "step": 49323 }, { "epoch": 2.6296316042010983, "grad_norm": 6.971536159515381, "learning_rate": 1.354e-07, "loss": 0.0248, "step": 49324 }, { "epoch": 2.629684917630751, "grad_norm": 2.5471432209014893, "learning_rate": 1.3520000000000002e-07, "loss": 0.0224, "step": 49325 }, { "epoch": 2.6297382310604043, "grad_norm": 0.7106161713600159, "learning_rate": 1.35e-07, "loss": 0.0042, "step": 49326 }, { "epoch": 2.629791544490057, "grad_norm": 1.5571987628936768, "learning_rate": 1.3480000000000002e-07, "loss": 0.0056, "step": 49327 }, { "epoch": 2.62984485791971, "grad_norm": 4.703600883483887, "learning_rate": 1.346e-07, "loss": 0.041, "step": 49328 }, { "epoch": 2.629898171349363, "grad_norm": 1.7519268989562988, "learning_rate": 1.344e-07, "loss": 0.0229, "step": 49329 }, { "epoch": 2.6299514847790157, "grad_norm": 1.6896008253097534, "learning_rate": 1.342e-07, "loss": 0.024, "step": 49330 }, { "epoch": 2.6300047982086685, "grad_norm": 1.610032081604004, "learning_rate": 1.34e-07, "loss": 0.0213, "step": 49331 }, { "epoch": 2.6300581116383217, "grad_norm": 2.4249796867370605, "learning_rate": 1.338e-07, "loss": 0.0228, "step": 49332 }, { "epoch": 2.6301114250679745, "grad_norm": 1.3787884712219238, "learning_rate": 1.336e-07, "loss": 0.0043, "step": 49333 }, { "epoch": 2.6301647384976277, "grad_norm": 0.8654738068580627, "learning_rate": 1.3340000000000002e-07, "loss": 0.0031, "step": 49334 }, { "epoch": 2.6302180519272804, "grad_norm": 2.1652300357818604, "learning_rate": 1.332e-07, "loss": 0.0228, "step": 49335 }, { "epoch": 2.6302713653569336, "grad_norm": 0.537559986114502, "learning_rate": 1.33e-07, "loss": 0.0032, "step": 49336 }, { "epoch": 2.6303246787865864, "grad_norm": 2.882295608520508, "learning_rate": 1.3280000000000002e-07, "loss": 0.0336, "step": 49337 }, { "epoch": 2.630377992216239, "grad_norm": 0.7315914630889893, "learning_rate": 1.326e-07, "loss": 0.0143, "step": 49338 }, { "epoch": 2.6304313056458923, "grad_norm": 2.6711432933807373, "learning_rate": 1.3240000000000002e-07, "loss": 0.0329, "step": 49339 }, { "epoch": 2.630484619075545, "grad_norm": 1.596030354499817, "learning_rate": 1.3220000000000003e-07, "loss": 0.0041, "step": 49340 }, { "epoch": 2.630537932505198, "grad_norm": 2.710845708847046, "learning_rate": 1.3200000000000002e-07, "loss": 0.0093, "step": 49341 }, { "epoch": 2.630591245934851, "grad_norm": 2.3955159187316895, "learning_rate": 1.3180000000000003e-07, "loss": 0.0147, "step": 49342 }, { "epoch": 2.630644559364504, "grad_norm": 0.45748841762542725, "learning_rate": 1.3160000000000001e-07, "loss": 0.0027, "step": 49343 }, { "epoch": 2.630697872794157, "grad_norm": 0.46045854687690735, "learning_rate": 1.3140000000000003e-07, "loss": 0.0026, "step": 49344 }, { "epoch": 2.6307511862238098, "grad_norm": 1.3624175786972046, "learning_rate": 1.312e-07, "loss": 0.0157, "step": 49345 }, { "epoch": 2.630804499653463, "grad_norm": 3.600266456604004, "learning_rate": 1.3100000000000002e-07, "loss": 0.0207, "step": 49346 }, { "epoch": 2.6308578130831157, "grad_norm": 2.3378212451934814, "learning_rate": 1.308e-07, "loss": 0.0071, "step": 49347 }, { "epoch": 2.6309111265127685, "grad_norm": 1.5287059545516968, "learning_rate": 1.3060000000000002e-07, "loss": 0.0146, "step": 49348 }, { "epoch": 2.6309644399424217, "grad_norm": 0.5393342971801758, "learning_rate": 1.304e-07, "loss": 0.003, "step": 49349 }, { "epoch": 2.6310177533720744, "grad_norm": 2.9411633014678955, "learning_rate": 1.3020000000000001e-07, "loss": 0.0232, "step": 49350 }, { "epoch": 2.631071066801727, "grad_norm": 0.5111469626426697, "learning_rate": 1.3e-07, "loss": 0.0028, "step": 49351 }, { "epoch": 2.6311243802313804, "grad_norm": 2.1168556213378906, "learning_rate": 1.298e-07, "loss": 0.0076, "step": 49352 }, { "epoch": 2.631177693661033, "grad_norm": 1.6125102043151855, "learning_rate": 1.296e-07, "loss": 0.0111, "step": 49353 }, { "epoch": 2.631231007090686, "grad_norm": 0.5166861414909363, "learning_rate": 1.294e-07, "loss": 0.0023, "step": 49354 }, { "epoch": 2.631284320520339, "grad_norm": 3.0099687576293945, "learning_rate": 1.292e-07, "loss": 0.0241, "step": 49355 }, { "epoch": 2.631337633949992, "grad_norm": 2.268307685852051, "learning_rate": 1.29e-07, "loss": 0.01, "step": 49356 }, { "epoch": 2.631390947379645, "grad_norm": 2.599600315093994, "learning_rate": 1.2880000000000002e-07, "loss": 0.019, "step": 49357 }, { "epoch": 2.631444260809298, "grad_norm": 0.4731353521347046, "learning_rate": 1.286e-07, "loss": 0.0026, "step": 49358 }, { "epoch": 2.631497574238951, "grad_norm": 2.1545467376708984, "learning_rate": 1.284e-07, "loss": 0.0285, "step": 49359 }, { "epoch": 2.631550887668604, "grad_norm": 0.865919828414917, "learning_rate": 1.282e-07, "loss": 0.0054, "step": 49360 }, { "epoch": 2.6316042010982565, "grad_norm": 2.5890495777130127, "learning_rate": 1.28e-07, "loss": 0.0267, "step": 49361 }, { "epoch": 2.6316575145279097, "grad_norm": 1.9377007484436035, "learning_rate": 1.2780000000000002e-07, "loss": 0.022, "step": 49362 }, { "epoch": 2.6317108279575625, "grad_norm": 1.686999797821045, "learning_rate": 1.276e-07, "loss": 0.0218, "step": 49363 }, { "epoch": 2.6317641413872153, "grad_norm": 2.123272180557251, "learning_rate": 1.2740000000000002e-07, "loss": 0.0165, "step": 49364 }, { "epoch": 2.6318174548168685, "grad_norm": 1.4537968635559082, "learning_rate": 1.2720000000000003e-07, "loss": 0.0152, "step": 49365 }, { "epoch": 2.631870768246521, "grad_norm": 3.3730905055999756, "learning_rate": 1.27e-07, "loss": 0.0198, "step": 49366 }, { "epoch": 2.631924081676174, "grad_norm": 0.2385915070772171, "learning_rate": 1.2680000000000002e-07, "loss": 0.0015, "step": 49367 }, { "epoch": 2.631977395105827, "grad_norm": 2.3807694911956787, "learning_rate": 1.266e-07, "loss": 0.0193, "step": 49368 }, { "epoch": 2.63203070853548, "grad_norm": 3.4400408267974854, "learning_rate": 1.2640000000000002e-07, "loss": 0.04, "step": 49369 }, { "epoch": 2.632084021965133, "grad_norm": 1.5237090587615967, "learning_rate": 1.262e-07, "loss": 0.0065, "step": 49370 }, { "epoch": 2.632137335394786, "grad_norm": 2.2892801761627197, "learning_rate": 1.2600000000000002e-07, "loss": 0.0136, "step": 49371 }, { "epoch": 2.632190648824439, "grad_norm": 2.000596761703491, "learning_rate": 1.258e-07, "loss": 0.0133, "step": 49372 }, { "epoch": 2.632243962254092, "grad_norm": 4.514322757720947, "learning_rate": 1.256e-07, "loss": 0.0241, "step": 49373 }, { "epoch": 2.6322972756837446, "grad_norm": 1.5358566045761108, "learning_rate": 1.2540000000000002e-07, "loss": 0.021, "step": 49374 }, { "epoch": 2.632350589113398, "grad_norm": 2.4696929454803467, "learning_rate": 1.252e-07, "loss": 0.0316, "step": 49375 }, { "epoch": 2.6324039025430506, "grad_norm": 2.4479048252105713, "learning_rate": 1.2500000000000002e-07, "loss": 0.0149, "step": 49376 }, { "epoch": 2.6324572159727033, "grad_norm": 3.7170536518096924, "learning_rate": 1.248e-07, "loss": 0.0126, "step": 49377 }, { "epoch": 2.6325105294023565, "grad_norm": 3.058307647705078, "learning_rate": 1.2460000000000002e-07, "loss": 0.0318, "step": 49378 }, { "epoch": 2.6325638428320093, "grad_norm": 1.9577187299728394, "learning_rate": 1.244e-07, "loss": 0.0159, "step": 49379 }, { "epoch": 2.632617156261662, "grad_norm": 1.5244189500808716, "learning_rate": 1.2420000000000001e-07, "loss": 0.0043, "step": 49380 }, { "epoch": 2.6326704696913152, "grad_norm": 3.6918156147003174, "learning_rate": 1.24e-07, "loss": 0.0391, "step": 49381 }, { "epoch": 2.6327237831209684, "grad_norm": 0.800146758556366, "learning_rate": 1.238e-07, "loss": 0.0025, "step": 49382 }, { "epoch": 2.632777096550621, "grad_norm": 2.784468412399292, "learning_rate": 1.236e-07, "loss": 0.0323, "step": 49383 }, { "epoch": 2.632830409980274, "grad_norm": 2.870279312133789, "learning_rate": 1.234e-07, "loss": 0.0127, "step": 49384 }, { "epoch": 2.632883723409927, "grad_norm": 2.2685205936431885, "learning_rate": 1.232e-07, "loss": 0.0436, "step": 49385 }, { "epoch": 2.63293703683958, "grad_norm": 1.9806544780731201, "learning_rate": 1.23e-07, "loss": 0.0096, "step": 49386 }, { "epoch": 2.6329903502692327, "grad_norm": 2.5962765216827393, "learning_rate": 1.2280000000000001e-07, "loss": 0.0118, "step": 49387 }, { "epoch": 2.633043663698886, "grad_norm": 1.5930825471878052, "learning_rate": 1.226e-07, "loss": 0.0041, "step": 49388 }, { "epoch": 2.6330969771285386, "grad_norm": 0.8926764726638794, "learning_rate": 1.224e-07, "loss": 0.0043, "step": 49389 }, { "epoch": 2.6331502905581914, "grad_norm": 3.4786033630371094, "learning_rate": 1.2220000000000002e-07, "loss": 0.0416, "step": 49390 }, { "epoch": 2.6332036039878446, "grad_norm": 1.5175209045410156, "learning_rate": 1.22e-07, "loss": 0.004, "step": 49391 }, { "epoch": 2.6332569174174973, "grad_norm": 1.3637669086456299, "learning_rate": 1.2180000000000002e-07, "loss": 0.0063, "step": 49392 }, { "epoch": 2.6333102308471505, "grad_norm": 1.0516330003738403, "learning_rate": 1.216e-07, "loss": 0.0041, "step": 49393 }, { "epoch": 2.6333635442768033, "grad_norm": 2.15863299369812, "learning_rate": 1.2140000000000002e-07, "loss": 0.0077, "step": 49394 }, { "epoch": 2.6334168577064565, "grad_norm": 2.7729618549346924, "learning_rate": 1.2120000000000003e-07, "loss": 0.0083, "step": 49395 }, { "epoch": 2.6334701711361093, "grad_norm": 1.09587824344635, "learning_rate": 1.21e-07, "loss": 0.0039, "step": 49396 }, { "epoch": 2.633523484565762, "grad_norm": 1.505029559135437, "learning_rate": 1.2080000000000002e-07, "loss": 0.0173, "step": 49397 }, { "epoch": 2.633576797995415, "grad_norm": 0.5413234233856201, "learning_rate": 1.206e-07, "loss": 0.003, "step": 49398 }, { "epoch": 2.633630111425068, "grad_norm": 2.937140464782715, "learning_rate": 1.2040000000000002e-07, "loss": 0.0122, "step": 49399 }, { "epoch": 2.6336834248547207, "grad_norm": 2.8308889865875244, "learning_rate": 1.202e-07, "loss": 0.0273, "step": 49400 }, { "epoch": 2.633736738284374, "grad_norm": 1.4457546472549438, "learning_rate": 1.2000000000000002e-07, "loss": 0.0232, "step": 49401 }, { "epoch": 2.6337900517140267, "grad_norm": 1.6840323209762573, "learning_rate": 1.198e-07, "loss": 0.0086, "step": 49402 }, { "epoch": 2.6338433651436794, "grad_norm": 3.8263638019561768, "learning_rate": 1.196e-07, "loss": 0.0226, "step": 49403 }, { "epoch": 2.6338966785733327, "grad_norm": 1.0092651844024658, "learning_rate": 1.194e-07, "loss": 0.0046, "step": 49404 }, { "epoch": 2.6339499920029854, "grad_norm": 1.6562482118606567, "learning_rate": 1.1920000000000001e-07, "loss": 0.0095, "step": 49405 }, { "epoch": 2.6340033054326386, "grad_norm": 1.8624026775360107, "learning_rate": 1.1900000000000002e-07, "loss": 0.0054, "step": 49406 }, { "epoch": 2.6340566188622914, "grad_norm": 1.435376763343811, "learning_rate": 1.188e-07, "loss": 0.0063, "step": 49407 }, { "epoch": 2.6341099322919446, "grad_norm": 3.891598701477051, "learning_rate": 1.1860000000000002e-07, "loss": 0.0448, "step": 49408 }, { "epoch": 2.6341632457215973, "grad_norm": 0.7307546734809875, "learning_rate": 1.184e-07, "loss": 0.0032, "step": 49409 }, { "epoch": 2.63421655915125, "grad_norm": 1.2983211278915405, "learning_rate": 1.1820000000000001e-07, "loss": 0.0056, "step": 49410 }, { "epoch": 2.6342698725809033, "grad_norm": 2.8396499156951904, "learning_rate": 1.1800000000000001e-07, "loss": 0.0112, "step": 49411 }, { "epoch": 2.634323186010556, "grad_norm": 1.8866057395935059, "learning_rate": 1.1780000000000001e-07, "loss": 0.0065, "step": 49412 }, { "epoch": 2.634376499440209, "grad_norm": 0.49862363934516907, "learning_rate": 1.1760000000000001e-07, "loss": 0.003, "step": 49413 }, { "epoch": 2.634429812869862, "grad_norm": 1.2983973026275635, "learning_rate": 1.1740000000000002e-07, "loss": 0.0312, "step": 49414 }, { "epoch": 2.6344831262995148, "grad_norm": 0.2928939759731293, "learning_rate": 1.172e-07, "loss": 0.002, "step": 49415 }, { "epoch": 2.6345364397291675, "grad_norm": 0.335172176361084, "learning_rate": 1.1700000000000002e-07, "loss": 0.0022, "step": 49416 }, { "epoch": 2.6345897531588207, "grad_norm": 2.9678421020507812, "learning_rate": 1.168e-07, "loss": 0.0303, "step": 49417 }, { "epoch": 2.6346430665884735, "grad_norm": 1.587515115737915, "learning_rate": 1.1660000000000001e-07, "loss": 0.0295, "step": 49418 }, { "epoch": 2.6346963800181267, "grad_norm": 1.971150517463684, "learning_rate": 1.164e-07, "loss": 0.0241, "step": 49419 }, { "epoch": 2.6347496934477794, "grad_norm": 1.4836704730987549, "learning_rate": 1.1620000000000001e-07, "loss": 0.0059, "step": 49420 }, { "epoch": 2.6348030068774326, "grad_norm": 1.3455733060836792, "learning_rate": 1.16e-07, "loss": 0.0123, "step": 49421 }, { "epoch": 2.6348563203070854, "grad_norm": 0.9296677112579346, "learning_rate": 1.158e-07, "loss": 0.0041, "step": 49422 }, { "epoch": 2.634909633736738, "grad_norm": 2.444763422012329, "learning_rate": 1.1560000000000002e-07, "loss": 0.0334, "step": 49423 }, { "epoch": 2.6349629471663913, "grad_norm": 1.6693968772888184, "learning_rate": 1.1540000000000001e-07, "loss": 0.0177, "step": 49424 }, { "epoch": 2.635016260596044, "grad_norm": 3.46250057220459, "learning_rate": 1.1520000000000001e-07, "loss": 0.0133, "step": 49425 }, { "epoch": 2.635069574025697, "grad_norm": 1.5207477807998657, "learning_rate": 1.1500000000000001e-07, "loss": 0.0212, "step": 49426 }, { "epoch": 2.63512288745535, "grad_norm": 2.1885390281677246, "learning_rate": 1.1480000000000002e-07, "loss": 0.0125, "step": 49427 }, { "epoch": 2.635176200885003, "grad_norm": 2.800454616546631, "learning_rate": 1.1460000000000001e-07, "loss": 0.0227, "step": 49428 }, { "epoch": 2.635229514314656, "grad_norm": 0.4429011642932892, "learning_rate": 1.1440000000000002e-07, "loss": 0.0031, "step": 49429 }, { "epoch": 2.6352828277443088, "grad_norm": 2.7468650341033936, "learning_rate": 1.142e-07, "loss": 0.0296, "step": 49430 }, { "epoch": 2.635336141173962, "grad_norm": 1.606066346168518, "learning_rate": 1.1400000000000001e-07, "loss": 0.0088, "step": 49431 }, { "epoch": 2.6353894546036147, "grad_norm": 1.6977505683898926, "learning_rate": 1.138e-07, "loss": 0.0175, "step": 49432 }, { "epoch": 2.6354427680332675, "grad_norm": 0.4085375964641571, "learning_rate": 1.1360000000000001e-07, "loss": 0.0026, "step": 49433 }, { "epoch": 2.6354960814629207, "grad_norm": 1.433520793914795, "learning_rate": 1.1340000000000001e-07, "loss": 0.0079, "step": 49434 }, { "epoch": 2.6355493948925734, "grad_norm": 2.453979730606079, "learning_rate": 1.1320000000000001e-07, "loss": 0.0111, "step": 49435 }, { "epoch": 2.635602708322226, "grad_norm": 3.979308605194092, "learning_rate": 1.13e-07, "loss": 0.0257, "step": 49436 }, { "epoch": 2.6356560217518794, "grad_norm": 0.4854576587677002, "learning_rate": 1.128e-07, "loss": 0.0031, "step": 49437 }, { "epoch": 2.635709335181532, "grad_norm": 1.0380125045776367, "learning_rate": 1.126e-07, "loss": 0.0188, "step": 49438 }, { "epoch": 2.635762648611185, "grad_norm": 1.3764785528182983, "learning_rate": 1.1240000000000001e-07, "loss": 0.0051, "step": 49439 }, { "epoch": 2.635815962040838, "grad_norm": 2.383491277694702, "learning_rate": 1.1220000000000001e-07, "loss": 0.0185, "step": 49440 }, { "epoch": 2.635869275470491, "grad_norm": 0.6008782982826233, "learning_rate": 1.1200000000000001e-07, "loss": 0.0041, "step": 49441 }, { "epoch": 2.635922588900144, "grad_norm": 0.9057796597480774, "learning_rate": 1.1180000000000002e-07, "loss": 0.0041, "step": 49442 }, { "epoch": 2.635975902329797, "grad_norm": 2.1445000171661377, "learning_rate": 1.1160000000000001e-07, "loss": 0.0068, "step": 49443 }, { "epoch": 2.63602921575945, "grad_norm": 1.708949327468872, "learning_rate": 1.1140000000000002e-07, "loss": 0.0078, "step": 49444 }, { "epoch": 2.636082529189103, "grad_norm": 3.9596571922302246, "learning_rate": 1.112e-07, "loss": 0.0311, "step": 49445 }, { "epoch": 2.6361358426187556, "grad_norm": 0.2341899275779724, "learning_rate": 1.1100000000000001e-07, "loss": 0.0016, "step": 49446 }, { "epoch": 2.6361891560484088, "grad_norm": 2.0302047729492188, "learning_rate": 1.108e-07, "loss": 0.0191, "step": 49447 }, { "epoch": 2.6362424694780615, "grad_norm": 0.8914574384689331, "learning_rate": 1.1060000000000001e-07, "loss": 0.0049, "step": 49448 }, { "epoch": 2.6362957829077143, "grad_norm": 0.5075214505195618, "learning_rate": 1.1040000000000001e-07, "loss": 0.0035, "step": 49449 }, { "epoch": 2.6363490963373675, "grad_norm": 2.1894521713256836, "learning_rate": 1.1020000000000001e-07, "loss": 0.0165, "step": 49450 }, { "epoch": 2.6364024097670202, "grad_norm": 2.4421653747558594, "learning_rate": 1.1e-07, "loss": 0.0106, "step": 49451 }, { "epoch": 2.636455723196673, "grad_norm": 0.5216584801673889, "learning_rate": 1.0980000000000002e-07, "loss": 0.0027, "step": 49452 }, { "epoch": 2.636509036626326, "grad_norm": 3.674201726913452, "learning_rate": 1.096e-07, "loss": 0.0188, "step": 49453 }, { "epoch": 2.636562350055979, "grad_norm": 2.3610799312591553, "learning_rate": 1.0940000000000001e-07, "loss": 0.0161, "step": 49454 }, { "epoch": 2.636615663485632, "grad_norm": 0.9662057161331177, "learning_rate": 1.092e-07, "loss": 0.0035, "step": 49455 }, { "epoch": 2.636668976915285, "grad_norm": 1.553252935409546, "learning_rate": 1.0900000000000001e-07, "loss": 0.0073, "step": 49456 }, { "epoch": 2.636722290344938, "grad_norm": 2.888615846633911, "learning_rate": 1.0880000000000002e-07, "loss": 0.0335, "step": 49457 }, { "epoch": 2.636775603774591, "grad_norm": 2.390968084335327, "learning_rate": 1.086e-07, "loss": 0.0091, "step": 49458 }, { "epoch": 2.6368289172042436, "grad_norm": 1.50387704372406, "learning_rate": 1.0840000000000002e-07, "loss": 0.0054, "step": 49459 }, { "epoch": 2.636882230633897, "grad_norm": 2.714426040649414, "learning_rate": 1.082e-07, "loss": 0.0138, "step": 49460 }, { "epoch": 2.6369355440635496, "grad_norm": 2.261935234069824, "learning_rate": 1.0800000000000001e-07, "loss": 0.0076, "step": 49461 }, { "epoch": 2.6369888574932023, "grad_norm": 1.4705054759979248, "learning_rate": 1.078e-07, "loss": 0.0076, "step": 49462 }, { "epoch": 2.6370421709228555, "grad_norm": 1.5295131206512451, "learning_rate": 1.0760000000000001e-07, "loss": 0.0112, "step": 49463 }, { "epoch": 2.6370954843525083, "grad_norm": 2.090634822845459, "learning_rate": 1.0740000000000001e-07, "loss": 0.0064, "step": 49464 }, { "epoch": 2.6371487977821615, "grad_norm": 4.362428665161133, "learning_rate": 1.0720000000000001e-07, "loss": 0.0303, "step": 49465 }, { "epoch": 2.6372021112118142, "grad_norm": 1.5732556581497192, "learning_rate": 1.07e-07, "loss": 0.0149, "step": 49466 }, { "epoch": 2.6372554246414674, "grad_norm": 1.3645731210708618, "learning_rate": 1.0680000000000002e-07, "loss": 0.0162, "step": 49467 }, { "epoch": 2.63730873807112, "grad_norm": 1.5689630508422852, "learning_rate": 1.066e-07, "loss": 0.0053, "step": 49468 }, { "epoch": 2.637362051500773, "grad_norm": 2.3319551944732666, "learning_rate": 1.0640000000000001e-07, "loss": 0.0125, "step": 49469 }, { "epoch": 2.637415364930426, "grad_norm": 2.0849339962005615, "learning_rate": 1.062e-07, "loss": 0.0461, "step": 49470 }, { "epoch": 2.637468678360079, "grad_norm": 2.8924880027770996, "learning_rate": 1.0600000000000001e-07, "loss": 0.0239, "step": 49471 }, { "epoch": 2.6375219917897317, "grad_norm": 0.8871474266052246, "learning_rate": 1.0580000000000002e-07, "loss": 0.0036, "step": 49472 }, { "epoch": 2.637575305219385, "grad_norm": 1.5497056245803833, "learning_rate": 1.056e-07, "loss": 0.0079, "step": 49473 }, { "epoch": 2.6376286186490376, "grad_norm": 1.4190620183944702, "learning_rate": 1.0540000000000002e-07, "loss": 0.0045, "step": 49474 }, { "epoch": 2.6376819320786904, "grad_norm": 0.7524065375328064, "learning_rate": 1.052e-07, "loss": 0.0048, "step": 49475 }, { "epoch": 2.6377352455083436, "grad_norm": 2.63533091545105, "learning_rate": 1.0500000000000001e-07, "loss": 0.0439, "step": 49476 }, { "epoch": 2.6377885589379964, "grad_norm": 0.6001874208450317, "learning_rate": 1.0480000000000001e-07, "loss": 0.0029, "step": 49477 }, { "epoch": 2.6378418723676496, "grad_norm": 0.6050419211387634, "learning_rate": 1.0460000000000001e-07, "loss": 0.0034, "step": 49478 }, { "epoch": 2.6378951857973023, "grad_norm": 1.3803280591964722, "learning_rate": 1.0440000000000001e-07, "loss": 0.026, "step": 49479 }, { "epoch": 2.6379484992269555, "grad_norm": 2.5930256843566895, "learning_rate": 1.042e-07, "loss": 0.0229, "step": 49480 }, { "epoch": 2.6380018126566083, "grad_norm": 2.8287439346313477, "learning_rate": 1.04e-07, "loss": 0.0227, "step": 49481 }, { "epoch": 2.638055126086261, "grad_norm": 3.2021148204803467, "learning_rate": 1.0380000000000002e-07, "loss": 0.0327, "step": 49482 }, { "epoch": 2.6381084395159142, "grad_norm": 1.5683965682983398, "learning_rate": 1.036e-07, "loss": 0.0207, "step": 49483 }, { "epoch": 2.638161752945567, "grad_norm": 1.4231173992156982, "learning_rate": 1.0340000000000001e-07, "loss": 0.0119, "step": 49484 }, { "epoch": 2.6382150663752197, "grad_norm": 1.2176439762115479, "learning_rate": 1.032e-07, "loss": 0.004, "step": 49485 }, { "epoch": 2.638268379804873, "grad_norm": 1.452090859413147, "learning_rate": 1.0300000000000001e-07, "loss": 0.0062, "step": 49486 }, { "epoch": 2.6383216932345257, "grad_norm": 1.8572694063186646, "learning_rate": 1.028e-07, "loss": 0.0068, "step": 49487 }, { "epoch": 2.6383750066641785, "grad_norm": 2.3380613327026367, "learning_rate": 1.026e-07, "loss": 0.0201, "step": 49488 }, { "epoch": 2.6384283200938317, "grad_norm": 2.592223644256592, "learning_rate": 1.0240000000000002e-07, "loss": 0.0161, "step": 49489 }, { "epoch": 2.6384816335234844, "grad_norm": 2.8660449981689453, "learning_rate": 1.022e-07, "loss": 0.0048, "step": 49490 }, { "epoch": 2.6385349469531376, "grad_norm": 1.5422931909561157, "learning_rate": 1.0200000000000001e-07, "loss": 0.0269, "step": 49491 }, { "epoch": 2.6385882603827904, "grad_norm": 2.133866786956787, "learning_rate": 1.0180000000000001e-07, "loss": 0.0322, "step": 49492 }, { "epoch": 2.6386415738124436, "grad_norm": 1.2529585361480713, "learning_rate": 1.0160000000000001e-07, "loss": 0.0077, "step": 49493 }, { "epoch": 2.6386948872420963, "grad_norm": 3.368555784225464, "learning_rate": 1.0140000000000001e-07, "loss": 0.0159, "step": 49494 }, { "epoch": 2.638748200671749, "grad_norm": 0.5731239914894104, "learning_rate": 1.0120000000000002e-07, "loss": 0.0034, "step": 49495 }, { "epoch": 2.6388015141014023, "grad_norm": 2.600478410720825, "learning_rate": 1.01e-07, "loss": 0.0312, "step": 49496 }, { "epoch": 2.638854827531055, "grad_norm": 1.234089970588684, "learning_rate": 1.0080000000000002e-07, "loss": 0.0053, "step": 49497 }, { "epoch": 2.638908140960708, "grad_norm": 3.497906446456909, "learning_rate": 1.006e-07, "loss": 0.0252, "step": 49498 }, { "epoch": 2.638961454390361, "grad_norm": 1.495631456375122, "learning_rate": 1.0040000000000001e-07, "loss": 0.0078, "step": 49499 }, { "epoch": 2.6390147678200138, "grad_norm": 2.2365124225616455, "learning_rate": 1.002e-07, "loss": 0.0236, "step": 49500 }, { "epoch": 2.6390680812496665, "grad_norm": 2.5468337535858154, "learning_rate": 1.0000000000000001e-07, "loss": 0.0084, "step": 49501 }, { "epoch": 2.6391213946793197, "grad_norm": 2.320632219314575, "learning_rate": 9.980000000000001e-08, "loss": 0.0211, "step": 49502 }, { "epoch": 2.639174708108973, "grad_norm": 3.0025343894958496, "learning_rate": 9.96e-08, "loss": 0.0162, "step": 49503 }, { "epoch": 2.6392280215386257, "grad_norm": 2.0020670890808105, "learning_rate": 9.94e-08, "loss": 0.0065, "step": 49504 }, { "epoch": 2.6392813349682784, "grad_norm": 3.0797829627990723, "learning_rate": 9.92e-08, "loss": 0.0099, "step": 49505 }, { "epoch": 2.6393346483979316, "grad_norm": 2.0632364749908447, "learning_rate": 9.900000000000001e-08, "loss": 0.0274, "step": 49506 }, { "epoch": 2.6393879618275844, "grad_norm": 1.637829303741455, "learning_rate": 9.880000000000001e-08, "loss": 0.0135, "step": 49507 }, { "epoch": 2.639441275257237, "grad_norm": 0.34008604288101196, "learning_rate": 9.860000000000001e-08, "loss": 0.0019, "step": 49508 }, { "epoch": 2.6394945886868904, "grad_norm": 1.6294854879379272, "learning_rate": 9.840000000000001e-08, "loss": 0.0321, "step": 49509 }, { "epoch": 2.639547902116543, "grad_norm": 1.8924089670181274, "learning_rate": 9.820000000000002e-08, "loss": 0.0221, "step": 49510 }, { "epoch": 2.639601215546196, "grad_norm": 0.6415215134620667, "learning_rate": 9.8e-08, "loss": 0.0043, "step": 49511 }, { "epoch": 2.639654528975849, "grad_norm": 1.6618614196777344, "learning_rate": 9.780000000000002e-08, "loss": 0.0103, "step": 49512 }, { "epoch": 2.639707842405502, "grad_norm": 2.7142624855041504, "learning_rate": 9.76e-08, "loss": 0.0129, "step": 49513 }, { "epoch": 2.639761155835155, "grad_norm": 1.5860371589660645, "learning_rate": 9.740000000000001e-08, "loss": 0.0253, "step": 49514 }, { "epoch": 2.639814469264808, "grad_norm": 1.3133121728897095, "learning_rate": 9.72e-08, "loss": 0.0061, "step": 49515 }, { "epoch": 2.639867782694461, "grad_norm": 1.630157232284546, "learning_rate": 9.700000000000001e-08, "loss": 0.005, "step": 49516 }, { "epoch": 2.6399210961241137, "grad_norm": 1.3892964124679565, "learning_rate": 9.68e-08, "loss": 0.0034, "step": 49517 }, { "epoch": 2.6399744095537665, "grad_norm": 1.0237077474594116, "learning_rate": 9.66e-08, "loss": 0.005, "step": 49518 }, { "epoch": 2.6400277229834197, "grad_norm": 0.7853339314460754, "learning_rate": 9.64e-08, "loss": 0.0037, "step": 49519 }, { "epoch": 2.6400810364130725, "grad_norm": 0.9221968054771423, "learning_rate": 9.620000000000001e-08, "loss": 0.004, "step": 49520 }, { "epoch": 2.640134349842725, "grad_norm": 1.3172744512557983, "learning_rate": 9.6e-08, "loss": 0.0029, "step": 49521 }, { "epoch": 2.6401876632723784, "grad_norm": 0.6345984935760498, "learning_rate": 9.580000000000001e-08, "loss": 0.0039, "step": 49522 }, { "epoch": 2.640240976702031, "grad_norm": 0.5612360239028931, "learning_rate": 9.560000000000002e-08, "loss": 0.0023, "step": 49523 }, { "epoch": 2.640294290131684, "grad_norm": 0.3641924262046814, "learning_rate": 9.540000000000001e-08, "loss": 0.002, "step": 49524 }, { "epoch": 2.640347603561337, "grad_norm": 0.9724694490432739, "learning_rate": 9.520000000000002e-08, "loss": 0.0129, "step": 49525 }, { "epoch": 2.64040091699099, "grad_norm": 1.8943724632263184, "learning_rate": 9.5e-08, "loss": 0.0219, "step": 49526 }, { "epoch": 2.640454230420643, "grad_norm": 1.0784012079238892, "learning_rate": 9.480000000000002e-08, "loss": 0.0147, "step": 49527 }, { "epoch": 2.640507543850296, "grad_norm": 2.963063955307007, "learning_rate": 9.46e-08, "loss": 0.0683, "step": 49528 }, { "epoch": 2.640560857279949, "grad_norm": 0.28727105259895325, "learning_rate": 9.440000000000001e-08, "loss": 0.002, "step": 49529 }, { "epoch": 2.640614170709602, "grad_norm": 2.377403736114502, "learning_rate": 9.420000000000001e-08, "loss": 0.0137, "step": 49530 }, { "epoch": 2.6406674841392546, "grad_norm": 1.8525029420852661, "learning_rate": 9.400000000000001e-08, "loss": 0.0235, "step": 49531 }, { "epoch": 2.6407207975689078, "grad_norm": 0.937746524810791, "learning_rate": 9.38e-08, "loss": 0.0048, "step": 49532 }, { "epoch": 2.6407741109985605, "grad_norm": 2.5595388412475586, "learning_rate": 9.36e-08, "loss": 0.0208, "step": 49533 }, { "epoch": 2.6408274244282133, "grad_norm": 3.430758237838745, "learning_rate": 9.34e-08, "loss": 0.0388, "step": 49534 }, { "epoch": 2.6408807378578665, "grad_norm": 0.8810322880744934, "learning_rate": 9.320000000000001e-08, "loss": 0.0046, "step": 49535 }, { "epoch": 2.6409340512875192, "grad_norm": 2.257143974304199, "learning_rate": 9.3e-08, "loss": 0.0072, "step": 49536 }, { "epoch": 2.640987364717172, "grad_norm": 2.506528615951538, "learning_rate": 9.280000000000001e-08, "loss": 0.0652, "step": 49537 }, { "epoch": 2.641040678146825, "grad_norm": 3.204519510269165, "learning_rate": 9.26e-08, "loss": 0.0123, "step": 49538 }, { "epoch": 2.641093991576478, "grad_norm": 0.1855166107416153, "learning_rate": 9.240000000000001e-08, "loss": 0.0013, "step": 49539 }, { "epoch": 2.641147305006131, "grad_norm": 1.372956395149231, "learning_rate": 9.220000000000002e-08, "loss": 0.0112, "step": 49540 }, { "epoch": 2.641200618435784, "grad_norm": 1.4452853202819824, "learning_rate": 9.2e-08, "loss": 0.0316, "step": 49541 }, { "epoch": 2.641253931865437, "grad_norm": 3.202653408050537, "learning_rate": 9.180000000000001e-08, "loss": 0.0099, "step": 49542 }, { "epoch": 2.64130724529509, "grad_norm": 0.34257009625434875, "learning_rate": 9.16e-08, "loss": 0.002, "step": 49543 }, { "epoch": 2.6413605587247426, "grad_norm": 2.000922441482544, "learning_rate": 9.140000000000001e-08, "loss": 0.0224, "step": 49544 }, { "epoch": 2.641413872154396, "grad_norm": 2.2842307090759277, "learning_rate": 9.120000000000001e-08, "loss": 0.036, "step": 49545 }, { "epoch": 2.6414671855840486, "grad_norm": 1.5913764238357544, "learning_rate": 9.100000000000001e-08, "loss": 0.0063, "step": 49546 }, { "epoch": 2.6415204990137013, "grad_norm": 2.867466926574707, "learning_rate": 9.08e-08, "loss": 0.0347, "step": 49547 }, { "epoch": 2.6415738124433545, "grad_norm": 0.6641877293586731, "learning_rate": 9.060000000000002e-08, "loss": 0.0043, "step": 49548 }, { "epoch": 2.6416271258730073, "grad_norm": 0.6519944071769714, "learning_rate": 9.04e-08, "loss": 0.0037, "step": 49549 }, { "epoch": 2.6416804393026605, "grad_norm": 0.48233306407928467, "learning_rate": 9.020000000000001e-08, "loss": 0.003, "step": 49550 }, { "epoch": 2.6417337527323133, "grad_norm": 2.0485756397247314, "learning_rate": 9e-08, "loss": 0.0096, "step": 49551 }, { "epoch": 2.6417870661619665, "grad_norm": 0.18663477897644043, "learning_rate": 8.980000000000001e-08, "loss": 0.0012, "step": 49552 }, { "epoch": 2.641840379591619, "grad_norm": 3.9430882930755615, "learning_rate": 8.96e-08, "loss": 0.0176, "step": 49553 }, { "epoch": 2.641893693021272, "grad_norm": 0.5404067635536194, "learning_rate": 8.94e-08, "loss": 0.0026, "step": 49554 }, { "epoch": 2.641947006450925, "grad_norm": 1.6770879030227661, "learning_rate": 8.920000000000002e-08, "loss": 0.0075, "step": 49555 }, { "epoch": 2.642000319880578, "grad_norm": 1.8912311792373657, "learning_rate": 8.9e-08, "loss": 0.0263, "step": 49556 }, { "epoch": 2.6420536333102307, "grad_norm": 1.6189045906066895, "learning_rate": 8.880000000000001e-08, "loss": 0.0132, "step": 49557 }, { "epoch": 2.642106946739884, "grad_norm": 0.4633142352104187, "learning_rate": 8.86e-08, "loss": 0.0025, "step": 49558 }, { "epoch": 2.6421602601695366, "grad_norm": 3.772193193435669, "learning_rate": 8.840000000000001e-08, "loss": 0.026, "step": 49559 }, { "epoch": 2.6422135735991894, "grad_norm": 2.1708171367645264, "learning_rate": 8.820000000000001e-08, "loss": 0.0302, "step": 49560 }, { "epoch": 2.6422668870288426, "grad_norm": 3.180845260620117, "learning_rate": 8.800000000000001e-08, "loss": 0.03, "step": 49561 }, { "epoch": 2.6423202004584954, "grad_norm": 4.306434154510498, "learning_rate": 8.78e-08, "loss": 0.0092, "step": 49562 }, { "epoch": 2.6423735138881486, "grad_norm": 2.4753453731536865, "learning_rate": 8.760000000000002e-08, "loss": 0.0168, "step": 49563 }, { "epoch": 2.6424268273178013, "grad_norm": 1.6622965335845947, "learning_rate": 8.74e-08, "loss": 0.0131, "step": 49564 }, { "epoch": 2.6424801407474545, "grad_norm": 0.7390051484107971, "learning_rate": 8.720000000000001e-08, "loss": 0.0207, "step": 49565 }, { "epoch": 2.6425334541771073, "grad_norm": 1.1023908853530884, "learning_rate": 8.7e-08, "loss": 0.0074, "step": 49566 }, { "epoch": 2.64258676760676, "grad_norm": 1.366127371788025, "learning_rate": 8.680000000000001e-08, "loss": 0.0153, "step": 49567 }, { "epoch": 2.6426400810364132, "grad_norm": 2.4617815017700195, "learning_rate": 8.66e-08, "loss": 0.0124, "step": 49568 }, { "epoch": 2.642693394466066, "grad_norm": 3.165876865386963, "learning_rate": 8.64e-08, "loss": 0.0164, "step": 49569 }, { "epoch": 2.6427467078957187, "grad_norm": 0.6079636216163635, "learning_rate": 8.62e-08, "loss": 0.0018, "step": 49570 }, { "epoch": 2.642800021325372, "grad_norm": 1.5186657905578613, "learning_rate": 8.6e-08, "loss": 0.0052, "step": 49571 }, { "epoch": 2.6428533347550247, "grad_norm": 0.6990574598312378, "learning_rate": 8.580000000000001e-08, "loss": 0.0033, "step": 49572 }, { "epoch": 2.6429066481846775, "grad_norm": 2.240194320678711, "learning_rate": 8.560000000000001e-08, "loss": 0.0182, "step": 49573 }, { "epoch": 2.6429599616143307, "grad_norm": 0.7405518293380737, "learning_rate": 8.540000000000001e-08, "loss": 0.004, "step": 49574 }, { "epoch": 2.6430132750439834, "grad_norm": 2.428567886352539, "learning_rate": 8.520000000000001e-08, "loss": 0.0166, "step": 49575 }, { "epoch": 2.6430665884736366, "grad_norm": 2.935606002807617, "learning_rate": 8.500000000000001e-08, "loss": 0.0366, "step": 49576 }, { "epoch": 2.6431199019032894, "grad_norm": 3.372041702270508, "learning_rate": 8.48e-08, "loss": 0.012, "step": 49577 }, { "epoch": 2.6431732153329426, "grad_norm": 2.0841219425201416, "learning_rate": 8.460000000000002e-08, "loss": 0.0427, "step": 49578 }, { "epoch": 2.6432265287625953, "grad_norm": 1.9007244110107422, "learning_rate": 8.44e-08, "loss": 0.0177, "step": 49579 }, { "epoch": 2.643279842192248, "grad_norm": 2.4103806018829346, "learning_rate": 8.420000000000001e-08, "loss": 0.0147, "step": 49580 }, { "epoch": 2.6433331556219013, "grad_norm": 1.6307893991470337, "learning_rate": 8.4e-08, "loss": 0.0185, "step": 49581 }, { "epoch": 2.643386469051554, "grad_norm": 2.3606553077697754, "learning_rate": 8.380000000000001e-08, "loss": 0.0095, "step": 49582 }, { "epoch": 2.643439782481207, "grad_norm": 1.2391108274459839, "learning_rate": 8.36e-08, "loss": 0.0098, "step": 49583 }, { "epoch": 2.64349309591086, "grad_norm": 2.2266836166381836, "learning_rate": 8.34e-08, "loss": 0.0049, "step": 49584 }, { "epoch": 2.6435464093405128, "grad_norm": 1.4733723402023315, "learning_rate": 8.32e-08, "loss": 0.0196, "step": 49585 }, { "epoch": 2.6435997227701655, "grad_norm": 3.520200490951538, "learning_rate": 8.3e-08, "loss": 0.0287, "step": 49586 }, { "epoch": 2.6436530361998187, "grad_norm": 0.8383808135986328, "learning_rate": 8.28e-08, "loss": 0.0055, "step": 49587 }, { "epoch": 2.643706349629472, "grad_norm": 2.554544687271118, "learning_rate": 8.260000000000001e-08, "loss": 0.0185, "step": 49588 }, { "epoch": 2.6437596630591247, "grad_norm": 0.5056750178337097, "learning_rate": 8.240000000000001e-08, "loss": 0.0034, "step": 49589 }, { "epoch": 2.6438129764887774, "grad_norm": 1.8998401165008545, "learning_rate": 8.220000000000001e-08, "loss": 0.0098, "step": 49590 }, { "epoch": 2.6438662899184306, "grad_norm": 0.9973134398460388, "learning_rate": 8.200000000000002e-08, "loss": 0.0139, "step": 49591 }, { "epoch": 2.6439196033480834, "grad_norm": 3.1690642833709717, "learning_rate": 8.18e-08, "loss": 0.0204, "step": 49592 }, { "epoch": 2.643972916777736, "grad_norm": 0.935312807559967, "learning_rate": 8.160000000000002e-08, "loss": 0.0131, "step": 49593 }, { "epoch": 2.6440262302073894, "grad_norm": 2.509761333465576, "learning_rate": 8.14e-08, "loss": 0.0176, "step": 49594 }, { "epoch": 2.644079543637042, "grad_norm": 2.5727648735046387, "learning_rate": 8.120000000000001e-08, "loss": 0.0071, "step": 49595 }, { "epoch": 2.644132857066695, "grad_norm": 1.544312596321106, "learning_rate": 8.1e-08, "loss": 0.028, "step": 49596 }, { "epoch": 2.644186170496348, "grad_norm": 0.9635212421417236, "learning_rate": 8.080000000000001e-08, "loss": 0.0024, "step": 49597 }, { "epoch": 2.644239483926001, "grad_norm": 1.6889461278915405, "learning_rate": 8.060000000000001e-08, "loss": 0.026, "step": 49598 }, { "epoch": 2.644292797355654, "grad_norm": 1.4590362310409546, "learning_rate": 8.04e-08, "loss": 0.0054, "step": 49599 }, { "epoch": 2.644346110785307, "grad_norm": 2.6519668102264404, "learning_rate": 8.02e-08, "loss": 0.0227, "step": 49600 }, { "epoch": 2.64439942421496, "grad_norm": 2.6975467205047607, "learning_rate": 8e-08, "loss": 0.0225, "step": 49601 }, { "epoch": 2.6444527376446127, "grad_norm": 0.5738686919212341, "learning_rate": 7.98e-08, "loss": 0.0024, "step": 49602 }, { "epoch": 2.6445060510742655, "grad_norm": 3.4863603115081787, "learning_rate": 7.960000000000001e-08, "loss": 0.0376, "step": 49603 }, { "epoch": 2.6445593645039187, "grad_norm": 1.8138532638549805, "learning_rate": 7.94e-08, "loss": 0.0081, "step": 49604 }, { "epoch": 2.6446126779335715, "grad_norm": 2.0920963287353516, "learning_rate": 7.920000000000001e-08, "loss": 0.0433, "step": 49605 }, { "epoch": 2.6446659913632242, "grad_norm": 1.1487185955047607, "learning_rate": 7.900000000000002e-08, "loss": 0.0043, "step": 49606 }, { "epoch": 2.6447193047928774, "grad_norm": 1.6691703796386719, "learning_rate": 7.88e-08, "loss": 0.007, "step": 49607 }, { "epoch": 2.64477261822253, "grad_norm": 2.0610032081604004, "learning_rate": 7.860000000000002e-08, "loss": 0.0079, "step": 49608 }, { "epoch": 2.644825931652183, "grad_norm": 1.9330168962478638, "learning_rate": 7.84e-08, "loss": 0.0068, "step": 49609 }, { "epoch": 2.644879245081836, "grad_norm": 2.0558340549468994, "learning_rate": 7.820000000000001e-08, "loss": 0.0045, "step": 49610 }, { "epoch": 2.644932558511489, "grad_norm": 0.4618418514728546, "learning_rate": 7.8e-08, "loss": 0.0022, "step": 49611 }, { "epoch": 2.644985871941142, "grad_norm": 1.9597111940383911, "learning_rate": 7.780000000000001e-08, "loss": 0.0228, "step": 49612 }, { "epoch": 2.645039185370795, "grad_norm": 1.7346714735031128, "learning_rate": 7.760000000000001e-08, "loss": 0.0167, "step": 49613 }, { "epoch": 2.645092498800448, "grad_norm": 6.996209144592285, "learning_rate": 7.74e-08, "loss": 0.0309, "step": 49614 }, { "epoch": 2.645145812230101, "grad_norm": 0.7322323322296143, "learning_rate": 7.72e-08, "loss": 0.005, "step": 49615 }, { "epoch": 2.6451991256597536, "grad_norm": 0.595831573009491, "learning_rate": 7.700000000000001e-08, "loss": 0.0026, "step": 49616 }, { "epoch": 2.6452524390894068, "grad_norm": 1.9180577993392944, "learning_rate": 7.680000000000001e-08, "loss": 0.0112, "step": 49617 }, { "epoch": 2.6453057525190595, "grad_norm": 2.8106675148010254, "learning_rate": 7.660000000000001e-08, "loss": 0.0093, "step": 49618 }, { "epoch": 2.6453590659487123, "grad_norm": 0.7362390756607056, "learning_rate": 7.640000000000001e-08, "loss": 0.0034, "step": 49619 }, { "epoch": 2.6454123793783655, "grad_norm": 0.7766059041023254, "learning_rate": 7.620000000000001e-08, "loss": 0.0038, "step": 49620 }, { "epoch": 2.6454656928080182, "grad_norm": 1.7345366477966309, "learning_rate": 7.6e-08, "loss": 0.0056, "step": 49621 }, { "epoch": 2.645519006237671, "grad_norm": 2.654282331466675, "learning_rate": 7.58e-08, "loss": 0.0227, "step": 49622 }, { "epoch": 2.645572319667324, "grad_norm": 1.9512205123901367, "learning_rate": 7.56e-08, "loss": 0.0037, "step": 49623 }, { "epoch": 2.6456256330969774, "grad_norm": 1.901073932647705, "learning_rate": 7.54e-08, "loss": 0.0077, "step": 49624 }, { "epoch": 2.64567894652663, "grad_norm": 1.183899998664856, "learning_rate": 7.52e-08, "loss": 0.005, "step": 49625 }, { "epoch": 2.645732259956283, "grad_norm": 3.0343446731567383, "learning_rate": 7.500000000000001e-08, "loss": 0.0088, "step": 49626 }, { "epoch": 2.645785573385936, "grad_norm": 0.4193364381790161, "learning_rate": 7.480000000000001e-08, "loss": 0.0033, "step": 49627 }, { "epoch": 2.645838886815589, "grad_norm": 4.159707069396973, "learning_rate": 7.46e-08, "loss": 0.0146, "step": 49628 }, { "epoch": 2.6458922002452416, "grad_norm": 2.0661845207214355, "learning_rate": 7.44e-08, "loss": 0.0062, "step": 49629 }, { "epoch": 2.645945513674895, "grad_norm": 1.3771835565567017, "learning_rate": 7.420000000000002e-08, "loss": 0.006, "step": 49630 }, { "epoch": 2.6459988271045476, "grad_norm": 0.2801481783390045, "learning_rate": 7.400000000000001e-08, "loss": 0.002, "step": 49631 }, { "epoch": 2.6460521405342003, "grad_norm": 1.4485424757003784, "learning_rate": 7.380000000000001e-08, "loss": 0.0053, "step": 49632 }, { "epoch": 2.6461054539638535, "grad_norm": 3.7202084064483643, "learning_rate": 7.360000000000001e-08, "loss": 0.0372, "step": 49633 }, { "epoch": 2.6461587673935063, "grad_norm": 1.470365285873413, "learning_rate": 7.340000000000001e-08, "loss": 0.0101, "step": 49634 }, { "epoch": 2.6462120808231595, "grad_norm": 1.5941671133041382, "learning_rate": 7.320000000000001e-08, "loss": 0.0062, "step": 49635 }, { "epoch": 2.6462653942528123, "grad_norm": 0.5151599049568176, "learning_rate": 7.3e-08, "loss": 0.0026, "step": 49636 }, { "epoch": 2.6463187076824655, "grad_norm": 0.7130787968635559, "learning_rate": 7.28e-08, "loss": 0.015, "step": 49637 }, { "epoch": 2.6463720211121182, "grad_norm": 1.5329968929290771, "learning_rate": 7.26e-08, "loss": 0.0211, "step": 49638 }, { "epoch": 2.646425334541771, "grad_norm": 2.969862699508667, "learning_rate": 7.24e-08, "loss": 0.0518, "step": 49639 }, { "epoch": 2.646478647971424, "grad_norm": 3.5694050788879395, "learning_rate": 7.22e-08, "loss": 0.0179, "step": 49640 }, { "epoch": 2.646531961401077, "grad_norm": 2.885937213897705, "learning_rate": 7.200000000000001e-08, "loss": 0.0075, "step": 49641 }, { "epoch": 2.6465852748307297, "grad_norm": 2.707010269165039, "learning_rate": 7.180000000000001e-08, "loss": 0.029, "step": 49642 }, { "epoch": 2.646638588260383, "grad_norm": 0.9841248989105225, "learning_rate": 7.16e-08, "loss": 0.0031, "step": 49643 }, { "epoch": 2.6466919016900357, "grad_norm": 1.5664935111999512, "learning_rate": 7.14e-08, "loss": 0.0166, "step": 49644 }, { "epoch": 2.6467452151196884, "grad_norm": 3.2126989364624023, "learning_rate": 7.12e-08, "loss": 0.0466, "step": 49645 }, { "epoch": 2.6467985285493416, "grad_norm": 0.3350505232810974, "learning_rate": 7.100000000000001e-08, "loss": 0.0031, "step": 49646 }, { "epoch": 2.6468518419789944, "grad_norm": 2.8876776695251465, "learning_rate": 7.080000000000001e-08, "loss": 0.0101, "step": 49647 }, { "epoch": 2.6469051554086476, "grad_norm": 3.9644036293029785, "learning_rate": 7.060000000000001e-08, "loss": 0.044, "step": 49648 }, { "epoch": 2.6469584688383003, "grad_norm": 1.9972469806671143, "learning_rate": 7.040000000000001e-08, "loss": 0.0127, "step": 49649 }, { "epoch": 2.6470117822679535, "grad_norm": 4.696502208709717, "learning_rate": 7.02e-08, "loss": 0.0351, "step": 49650 }, { "epoch": 2.6470650956976063, "grad_norm": 2.2654590606689453, "learning_rate": 7e-08, "loss": 0.031, "step": 49651 }, { "epoch": 2.647118409127259, "grad_norm": 0.536803126335144, "learning_rate": 6.98e-08, "loss": 0.0035, "step": 49652 }, { "epoch": 2.6471717225569122, "grad_norm": 2.456357479095459, "learning_rate": 6.96e-08, "loss": 0.0292, "step": 49653 }, { "epoch": 2.647225035986565, "grad_norm": 1.469448208808899, "learning_rate": 6.94e-08, "loss": 0.0052, "step": 49654 }, { "epoch": 2.6472783494162178, "grad_norm": 0.3741863965988159, "learning_rate": 6.920000000000001e-08, "loss": 0.0025, "step": 49655 }, { "epoch": 2.647331662845871, "grad_norm": 0.7851146459579468, "learning_rate": 6.900000000000001e-08, "loss": 0.004, "step": 49656 }, { "epoch": 2.6473849762755237, "grad_norm": 1.8158286809921265, "learning_rate": 6.880000000000001e-08, "loss": 0.0287, "step": 49657 }, { "epoch": 2.6474382897051765, "grad_norm": 5.608353137969971, "learning_rate": 6.86e-08, "loss": 0.0405, "step": 49658 }, { "epoch": 2.6474916031348297, "grad_norm": 1.601722240447998, "learning_rate": 6.84e-08, "loss": 0.0254, "step": 49659 }, { "epoch": 2.6475449165644824, "grad_norm": 1.7187483310699463, "learning_rate": 6.82e-08, "loss": 0.0134, "step": 49660 }, { "epoch": 2.6475982299941356, "grad_norm": 2.5128390789031982, "learning_rate": 6.8e-08, "loss": 0.0214, "step": 49661 }, { "epoch": 2.6476515434237884, "grad_norm": 0.3742966949939728, "learning_rate": 6.78e-08, "loss": 0.0024, "step": 49662 }, { "epoch": 2.6477048568534416, "grad_norm": 0.7592261433601379, "learning_rate": 6.760000000000001e-08, "loss": 0.0046, "step": 49663 }, { "epoch": 2.6477581702830943, "grad_norm": 3.2047669887542725, "learning_rate": 6.740000000000001e-08, "loss": 0.0216, "step": 49664 }, { "epoch": 2.647811483712747, "grad_norm": 3.5131313800811768, "learning_rate": 6.72e-08, "loss": 0.0107, "step": 49665 }, { "epoch": 2.6478647971424003, "grad_norm": 1.8897547721862793, "learning_rate": 6.7e-08, "loss": 0.0103, "step": 49666 }, { "epoch": 2.647918110572053, "grad_norm": 1.7589832544326782, "learning_rate": 6.68e-08, "loss": 0.0222, "step": 49667 }, { "epoch": 2.647971424001706, "grad_norm": 1.8860080242156982, "learning_rate": 6.66e-08, "loss": 0.0148, "step": 49668 }, { "epoch": 2.648024737431359, "grad_norm": 0.8729761242866516, "learning_rate": 6.640000000000001e-08, "loss": 0.0025, "step": 49669 }, { "epoch": 2.648078050861012, "grad_norm": 1.4345115423202515, "learning_rate": 6.620000000000001e-08, "loss": 0.0104, "step": 49670 }, { "epoch": 2.648131364290665, "grad_norm": 2.383085250854492, "learning_rate": 6.600000000000001e-08, "loss": 0.0157, "step": 49671 }, { "epoch": 2.6481846777203177, "grad_norm": 1.0176219940185547, "learning_rate": 6.580000000000001e-08, "loss": 0.0157, "step": 49672 }, { "epoch": 2.648237991149971, "grad_norm": 2.557608127593994, "learning_rate": 6.56e-08, "loss": 0.0175, "step": 49673 }, { "epoch": 2.6482913045796237, "grad_norm": 0.8231086730957031, "learning_rate": 6.54e-08, "loss": 0.0043, "step": 49674 }, { "epoch": 2.6483446180092765, "grad_norm": 1.428279161453247, "learning_rate": 6.52e-08, "loss": 0.0167, "step": 49675 }, { "epoch": 2.6483979314389297, "grad_norm": 1.4261536598205566, "learning_rate": 6.5e-08, "loss": 0.0054, "step": 49676 }, { "epoch": 2.6484512448685824, "grad_norm": 0.9305543899536133, "learning_rate": 6.48e-08, "loss": 0.0039, "step": 49677 }, { "epoch": 2.648504558298235, "grad_norm": 1.327480435371399, "learning_rate": 6.46e-08, "loss": 0.0054, "step": 49678 }, { "epoch": 2.6485578717278884, "grad_norm": 3.3929190635681152, "learning_rate": 6.440000000000001e-08, "loss": 0.0339, "step": 49679 }, { "epoch": 2.648611185157541, "grad_norm": 1.34490966796875, "learning_rate": 6.42e-08, "loss": 0.008, "step": 49680 }, { "epoch": 2.648664498587194, "grad_norm": 3.396810293197632, "learning_rate": 6.4e-08, "loss": 0.021, "step": 49681 }, { "epoch": 2.648717812016847, "grad_norm": 2.5658018589019775, "learning_rate": 6.38e-08, "loss": 0.0182, "step": 49682 }, { "epoch": 2.6487711254465, "grad_norm": 1.5097302198410034, "learning_rate": 6.360000000000001e-08, "loss": 0.0446, "step": 49683 }, { "epoch": 2.648824438876153, "grad_norm": 1.617821455001831, "learning_rate": 6.340000000000001e-08, "loss": 0.0047, "step": 49684 }, { "epoch": 2.648877752305806, "grad_norm": 1.7227814197540283, "learning_rate": 6.320000000000001e-08, "loss": 0.0302, "step": 49685 }, { "epoch": 2.648931065735459, "grad_norm": 0.456744909286499, "learning_rate": 6.300000000000001e-08, "loss": 0.0028, "step": 49686 }, { "epoch": 2.6489843791651118, "grad_norm": 3.104570150375366, "learning_rate": 6.28e-08, "loss": 0.0473, "step": 49687 }, { "epoch": 2.6490376925947645, "grad_norm": 1.5437493324279785, "learning_rate": 6.26e-08, "loss": 0.0088, "step": 49688 }, { "epoch": 2.6490910060244177, "grad_norm": 2.146101951599121, "learning_rate": 6.24e-08, "loss": 0.0116, "step": 49689 }, { "epoch": 2.6491443194540705, "grad_norm": 0.39394593238830566, "learning_rate": 6.22e-08, "loss": 0.0025, "step": 49690 }, { "epoch": 2.6491976328837232, "grad_norm": 2.5941319465637207, "learning_rate": 6.2e-08, "loss": 0.0087, "step": 49691 }, { "epoch": 2.6492509463133764, "grad_norm": 0.46472927927970886, "learning_rate": 6.18e-08, "loss": 0.0028, "step": 49692 }, { "epoch": 2.649304259743029, "grad_norm": 0.9866666793823242, "learning_rate": 6.16e-08, "loss": 0.0051, "step": 49693 }, { "epoch": 2.649357573172682, "grad_norm": 0.18723627924919128, "learning_rate": 6.140000000000001e-08, "loss": 0.0015, "step": 49694 }, { "epoch": 2.649410886602335, "grad_norm": 0.8355230689048767, "learning_rate": 6.12e-08, "loss": 0.0142, "step": 49695 }, { "epoch": 2.649464200031988, "grad_norm": 1.8564573526382446, "learning_rate": 6.1e-08, "loss": 0.0193, "step": 49696 }, { "epoch": 2.649517513461641, "grad_norm": 0.49646061658859253, "learning_rate": 6.08e-08, "loss": 0.0028, "step": 49697 }, { "epoch": 2.649570826891294, "grad_norm": 0.9415380954742432, "learning_rate": 6.060000000000001e-08, "loss": 0.005, "step": 49698 }, { "epoch": 2.649624140320947, "grad_norm": 3.2079484462738037, "learning_rate": 6.040000000000001e-08, "loss": 0.0181, "step": 49699 }, { "epoch": 2.6496774537506, "grad_norm": 1.0068477392196655, "learning_rate": 6.020000000000001e-08, "loss": 0.0216, "step": 49700 }, { "epoch": 2.6497307671802526, "grad_norm": 1.7275687456130981, "learning_rate": 6.000000000000001e-08, "loss": 0.0133, "step": 49701 }, { "epoch": 2.649784080609906, "grad_norm": 0.85152667760849, "learning_rate": 5.98e-08, "loss": 0.0043, "step": 49702 }, { "epoch": 2.6498373940395585, "grad_norm": 2.623415470123291, "learning_rate": 5.9600000000000004e-08, "loss": 0.0302, "step": 49703 }, { "epoch": 2.6498907074692113, "grad_norm": 1.786412000656128, "learning_rate": 5.94e-08, "loss": 0.0072, "step": 49704 }, { "epoch": 2.6499440208988645, "grad_norm": 2.047252893447876, "learning_rate": 5.92e-08, "loss": 0.0107, "step": 49705 }, { "epoch": 2.6499973343285173, "grad_norm": 2.2529964447021484, "learning_rate": 5.9000000000000006e-08, "loss": 0.0065, "step": 49706 }, { "epoch": 2.65005064775817, "grad_norm": 1.853628158569336, "learning_rate": 5.8800000000000004e-08, "loss": 0.012, "step": 49707 }, { "epoch": 2.650103961187823, "grad_norm": 1.0337564945220947, "learning_rate": 5.86e-08, "loss": 0.0045, "step": 49708 }, { "epoch": 2.6501572746174764, "grad_norm": 0.16354261338710785, "learning_rate": 5.84e-08, "loss": 0.0013, "step": 49709 }, { "epoch": 2.650210588047129, "grad_norm": 0.9698092341423035, "learning_rate": 5.82e-08, "loss": 0.0038, "step": 49710 }, { "epoch": 2.650263901476782, "grad_norm": 3.231078863143921, "learning_rate": 5.8e-08, "loss": 0.0507, "step": 49711 }, { "epoch": 2.650317214906435, "grad_norm": 2.3551647663116455, "learning_rate": 5.780000000000001e-08, "loss": 0.0112, "step": 49712 }, { "epoch": 2.650370528336088, "grad_norm": 2.4816670417785645, "learning_rate": 5.7600000000000006e-08, "loss": 0.0104, "step": 49713 }, { "epoch": 2.6504238417657406, "grad_norm": 2.767824172973633, "learning_rate": 5.740000000000001e-08, "loss": 0.0163, "step": 49714 }, { "epoch": 2.650477155195394, "grad_norm": 2.5950093269348145, "learning_rate": 5.720000000000001e-08, "loss": 0.0221, "step": 49715 }, { "epoch": 2.6505304686250466, "grad_norm": 0.4833373725414276, "learning_rate": 5.700000000000001e-08, "loss": 0.0038, "step": 49716 }, { "epoch": 2.6505837820546994, "grad_norm": 1.1137274503707886, "learning_rate": 5.6800000000000006e-08, "loss": 0.0075, "step": 49717 }, { "epoch": 2.6506370954843526, "grad_norm": 1.542528748512268, "learning_rate": 5.6600000000000004e-08, "loss": 0.0196, "step": 49718 }, { "epoch": 2.6506904089140053, "grad_norm": 1.1414026021957397, "learning_rate": 5.64e-08, "loss": 0.0129, "step": 49719 }, { "epoch": 2.6507437223436585, "grad_norm": 2.7578399181365967, "learning_rate": 5.620000000000001e-08, "loss": 0.0129, "step": 49720 }, { "epoch": 2.6507970357733113, "grad_norm": 0.6518396735191345, "learning_rate": 5.6000000000000005e-08, "loss": 0.0037, "step": 49721 }, { "epoch": 2.6508503492029645, "grad_norm": 0.3764795958995819, "learning_rate": 5.5800000000000003e-08, "loss": 0.0024, "step": 49722 }, { "epoch": 2.6509036626326172, "grad_norm": 3.366509437561035, "learning_rate": 5.56e-08, "loss": 0.0638, "step": 49723 }, { "epoch": 2.65095697606227, "grad_norm": 1.1391834020614624, "learning_rate": 5.54e-08, "loss": 0.0048, "step": 49724 }, { "epoch": 2.651010289491923, "grad_norm": 1.549199104309082, "learning_rate": 5.5200000000000005e-08, "loss": 0.0073, "step": 49725 }, { "epoch": 2.651063602921576, "grad_norm": 1.757617712020874, "learning_rate": 5.5e-08, "loss": 0.0081, "step": 49726 }, { "epoch": 2.6511169163512287, "grad_norm": 0.7843894362449646, "learning_rate": 5.48e-08, "loss": 0.0035, "step": 49727 }, { "epoch": 2.651170229780882, "grad_norm": 3.5981075763702393, "learning_rate": 5.46e-08, "loss": 0.0392, "step": 49728 }, { "epoch": 2.6512235432105347, "grad_norm": 1.3037097454071045, "learning_rate": 5.440000000000001e-08, "loss": 0.0052, "step": 49729 }, { "epoch": 2.6512768566401874, "grad_norm": 0.9160434007644653, "learning_rate": 5.420000000000001e-08, "loss": 0.013, "step": 49730 }, { "epoch": 2.6513301700698406, "grad_norm": 3.0593414306640625, "learning_rate": 5.400000000000001e-08, "loss": 0.0414, "step": 49731 }, { "epoch": 2.6513834834994934, "grad_norm": 0.9476640820503235, "learning_rate": 5.3800000000000005e-08, "loss": 0.0103, "step": 49732 }, { "epoch": 2.6514367969291466, "grad_norm": 1.7897837162017822, "learning_rate": 5.3600000000000004e-08, "loss": 0.0209, "step": 49733 }, { "epoch": 2.6514901103587993, "grad_norm": 0.544548511505127, "learning_rate": 5.340000000000001e-08, "loss": 0.0025, "step": 49734 }, { "epoch": 2.6515434237884525, "grad_norm": 1.212451457977295, "learning_rate": 5.3200000000000007e-08, "loss": 0.0064, "step": 49735 }, { "epoch": 2.6515967372181053, "grad_norm": 1.0996930599212646, "learning_rate": 5.3000000000000005e-08, "loss": 0.0184, "step": 49736 }, { "epoch": 2.651650050647758, "grad_norm": 1.84458327293396, "learning_rate": 5.28e-08, "loss": 0.0105, "step": 49737 }, { "epoch": 2.6517033640774113, "grad_norm": 1.4185969829559326, "learning_rate": 5.26e-08, "loss": 0.0232, "step": 49738 }, { "epoch": 2.651756677507064, "grad_norm": 0.21442830562591553, "learning_rate": 5.2400000000000006e-08, "loss": 0.0019, "step": 49739 }, { "epoch": 2.6518099909367168, "grad_norm": 1.821413516998291, "learning_rate": 5.2200000000000004e-08, "loss": 0.0096, "step": 49740 }, { "epoch": 2.65186330436637, "grad_norm": 1.8191457986831665, "learning_rate": 5.2e-08, "loss": 0.0327, "step": 49741 }, { "epoch": 2.6519166177960227, "grad_norm": 0.36855047941207886, "learning_rate": 5.18e-08, "loss": 0.0026, "step": 49742 }, { "epoch": 2.6519699312256755, "grad_norm": 2.0414235591888428, "learning_rate": 5.16e-08, "loss": 0.0185, "step": 49743 }, { "epoch": 2.6520232446553287, "grad_norm": 2.371323347091675, "learning_rate": 5.14e-08, "loss": 0.0155, "step": 49744 }, { "epoch": 2.652076558084982, "grad_norm": 0.38343724608421326, "learning_rate": 5.120000000000001e-08, "loss": 0.0018, "step": 49745 }, { "epoch": 2.6521298715146346, "grad_norm": 0.5074772238731384, "learning_rate": 5.100000000000001e-08, "loss": 0.0034, "step": 49746 }, { "epoch": 2.6521831849442874, "grad_norm": 2.324537992477417, "learning_rate": 5.0800000000000005e-08, "loss": 0.0078, "step": 49747 }, { "epoch": 2.6522364983739406, "grad_norm": 0.9182280898094177, "learning_rate": 5.060000000000001e-08, "loss": 0.0032, "step": 49748 }, { "epoch": 2.6522898118035934, "grad_norm": 3.2633399963378906, "learning_rate": 5.040000000000001e-08, "loss": 0.0373, "step": 49749 }, { "epoch": 2.652343125233246, "grad_norm": 1.8597004413604736, "learning_rate": 5.0200000000000006e-08, "loss": 0.0311, "step": 49750 }, { "epoch": 2.6523964386628993, "grad_norm": 0.551064133644104, "learning_rate": 5.0000000000000004e-08, "loss": 0.0032, "step": 49751 }, { "epoch": 2.652449752092552, "grad_norm": 1.4957455396652222, "learning_rate": 4.98e-08, "loss": 0.0054, "step": 49752 }, { "epoch": 2.652503065522205, "grad_norm": 1.7570241689682007, "learning_rate": 4.96e-08, "loss": 0.0091, "step": 49753 }, { "epoch": 2.652556378951858, "grad_norm": 1.0328826904296875, "learning_rate": 4.9400000000000006e-08, "loss": 0.0043, "step": 49754 }, { "epoch": 2.652609692381511, "grad_norm": 1.5370749235153198, "learning_rate": 4.9200000000000004e-08, "loss": 0.0083, "step": 49755 }, { "epoch": 2.652663005811164, "grad_norm": 3.1333534717559814, "learning_rate": 4.9e-08, "loss": 0.0095, "step": 49756 }, { "epoch": 2.6527163192408167, "grad_norm": 2.2927987575531006, "learning_rate": 4.88e-08, "loss": 0.0064, "step": 49757 }, { "epoch": 2.65276963267047, "grad_norm": 1.9326931238174438, "learning_rate": 4.86e-08, "loss": 0.0169, "step": 49758 }, { "epoch": 2.6528229461001227, "grad_norm": 0.30507025122642517, "learning_rate": 4.84e-08, "loss": 0.0025, "step": 49759 }, { "epoch": 2.6528762595297755, "grad_norm": 2.747969627380371, "learning_rate": 4.82e-08, "loss": 0.0243, "step": 49760 }, { "epoch": 2.6529295729594287, "grad_norm": 1.5695085525512695, "learning_rate": 4.8e-08, "loss": 0.0052, "step": 49761 }, { "epoch": 2.6529828863890814, "grad_norm": 1.0743021965026855, "learning_rate": 4.780000000000001e-08, "loss": 0.0036, "step": 49762 }, { "epoch": 2.653036199818734, "grad_norm": 1.7388625144958496, "learning_rate": 4.760000000000001e-08, "loss": 0.0106, "step": 49763 }, { "epoch": 2.6530895132483874, "grad_norm": 1.3076140880584717, "learning_rate": 4.740000000000001e-08, "loss": 0.0186, "step": 49764 }, { "epoch": 2.65314282667804, "grad_norm": 2.4185190200805664, "learning_rate": 4.7200000000000006e-08, "loss": 0.0236, "step": 49765 }, { "epoch": 2.653196140107693, "grad_norm": 1.5511935949325562, "learning_rate": 4.7000000000000004e-08, "loss": 0.006, "step": 49766 }, { "epoch": 2.653249453537346, "grad_norm": 2.0899951457977295, "learning_rate": 4.68e-08, "loss": 0.0191, "step": 49767 }, { "epoch": 2.653302766966999, "grad_norm": 3.1691815853118896, "learning_rate": 4.660000000000001e-08, "loss": 0.0204, "step": 49768 }, { "epoch": 2.653356080396652, "grad_norm": 0.9316370487213135, "learning_rate": 4.6400000000000005e-08, "loss": 0.0036, "step": 49769 }, { "epoch": 2.653409393826305, "grad_norm": 1.275388240814209, "learning_rate": 4.6200000000000003e-08, "loss": 0.0048, "step": 49770 }, { "epoch": 2.653462707255958, "grad_norm": 0.6793678998947144, "learning_rate": 4.6e-08, "loss": 0.0042, "step": 49771 }, { "epoch": 2.6535160206856108, "grad_norm": 0.7003395557403564, "learning_rate": 4.58e-08, "loss": 0.0267, "step": 49772 }, { "epoch": 2.6535693341152635, "grad_norm": 3.7557430267333984, "learning_rate": 4.5600000000000005e-08, "loss": 0.0399, "step": 49773 }, { "epoch": 2.6536226475449167, "grad_norm": 1.7765326499938965, "learning_rate": 4.54e-08, "loss": 0.0121, "step": 49774 }, { "epoch": 2.6536759609745695, "grad_norm": 0.9574971199035645, "learning_rate": 4.52e-08, "loss": 0.0039, "step": 49775 }, { "epoch": 2.6537292744042222, "grad_norm": 2.8417558670043945, "learning_rate": 4.5e-08, "loss": 0.0207, "step": 49776 }, { "epoch": 2.6537825878338754, "grad_norm": 1.7912888526916504, "learning_rate": 4.48e-08, "loss": 0.0197, "step": 49777 }, { "epoch": 2.653835901263528, "grad_norm": 1.7011414766311646, "learning_rate": 4.460000000000001e-08, "loss": 0.0054, "step": 49778 }, { "epoch": 2.653889214693181, "grad_norm": 1.9873257875442505, "learning_rate": 4.440000000000001e-08, "loss": 0.0206, "step": 49779 }, { "epoch": 2.653942528122834, "grad_norm": 3.0330100059509277, "learning_rate": 4.4200000000000005e-08, "loss": 0.0212, "step": 49780 }, { "epoch": 2.653995841552487, "grad_norm": 1.955085277557373, "learning_rate": 4.4000000000000004e-08, "loss": 0.0133, "step": 49781 }, { "epoch": 2.65404915498214, "grad_norm": 1.8598861694335938, "learning_rate": 4.380000000000001e-08, "loss": 0.0391, "step": 49782 }, { "epoch": 2.654102468411793, "grad_norm": 4.591716289520264, "learning_rate": 4.3600000000000007e-08, "loss": 0.0201, "step": 49783 }, { "epoch": 2.654155781841446, "grad_norm": 0.45236334204673767, "learning_rate": 4.3400000000000005e-08, "loss": 0.003, "step": 49784 }, { "epoch": 2.654209095271099, "grad_norm": 4.0196213722229, "learning_rate": 4.32e-08, "loss": 0.0172, "step": 49785 }, { "epoch": 2.6542624087007516, "grad_norm": 2.8725996017456055, "learning_rate": 4.3e-08, "loss": 0.0098, "step": 49786 }, { "epoch": 2.654315722130405, "grad_norm": 1.557096004486084, "learning_rate": 4.2800000000000006e-08, "loss": 0.0251, "step": 49787 }, { "epoch": 2.6543690355600575, "grad_norm": 0.44058945775032043, "learning_rate": 4.2600000000000004e-08, "loss": 0.003, "step": 49788 }, { "epoch": 2.6544223489897103, "grad_norm": 2.073042392730713, "learning_rate": 4.24e-08, "loss": 0.0295, "step": 49789 }, { "epoch": 2.6544756624193635, "grad_norm": 3.092839479446411, "learning_rate": 4.22e-08, "loss": 0.011, "step": 49790 }, { "epoch": 2.6545289758490163, "grad_norm": 0.274240106344223, "learning_rate": 4.2e-08, "loss": 0.0018, "step": 49791 }, { "epoch": 2.6545822892786695, "grad_norm": 0.7246448993682861, "learning_rate": 4.18e-08, "loss": 0.0047, "step": 49792 }, { "epoch": 2.654635602708322, "grad_norm": 1.7059659957885742, "learning_rate": 4.16e-08, "loss": 0.0086, "step": 49793 }, { "epoch": 2.6546889161379754, "grad_norm": 1.9626606702804565, "learning_rate": 4.14e-08, "loss": 0.0365, "step": 49794 }, { "epoch": 2.654742229567628, "grad_norm": 0.2946910858154297, "learning_rate": 4.1200000000000005e-08, "loss": 0.0021, "step": 49795 }, { "epoch": 2.654795542997281, "grad_norm": 1.930120825767517, "learning_rate": 4.100000000000001e-08, "loss": 0.0123, "step": 49796 }, { "epoch": 2.654848856426934, "grad_norm": 0.3007947504520416, "learning_rate": 4.080000000000001e-08, "loss": 0.0015, "step": 49797 }, { "epoch": 2.654902169856587, "grad_norm": 1.2315001487731934, "learning_rate": 4.0600000000000006e-08, "loss": 0.012, "step": 49798 }, { "epoch": 2.6549554832862396, "grad_norm": 2.343097686767578, "learning_rate": 4.0400000000000004e-08, "loss": 0.0227, "step": 49799 }, { "epoch": 2.655008796715893, "grad_norm": 2.081575870513916, "learning_rate": 4.02e-08, "loss": 0.0187, "step": 49800 }, { "epoch": 2.6550621101455456, "grad_norm": 1.7947955131530762, "learning_rate": 4e-08, "loss": 0.0111, "step": 49801 }, { "epoch": 2.6551154235751984, "grad_norm": 1.0926028490066528, "learning_rate": 3.9800000000000006e-08, "loss": 0.005, "step": 49802 }, { "epoch": 2.6551687370048516, "grad_norm": 2.3916537761688232, "learning_rate": 3.9600000000000004e-08, "loss": 0.0079, "step": 49803 }, { "epoch": 2.6552220504345043, "grad_norm": 1.624894618988037, "learning_rate": 3.94e-08, "loss": 0.011, "step": 49804 }, { "epoch": 2.6552753638641575, "grad_norm": 0.6547437906265259, "learning_rate": 3.92e-08, "loss": 0.003, "step": 49805 }, { "epoch": 2.6553286772938103, "grad_norm": 3.7592742443084717, "learning_rate": 3.9e-08, "loss": 0.0252, "step": 49806 }, { "epoch": 2.6553819907234635, "grad_norm": 0.7066628336906433, "learning_rate": 3.8800000000000003e-08, "loss": 0.0199, "step": 49807 }, { "epoch": 2.6554353041531162, "grad_norm": 2.0938642024993896, "learning_rate": 3.86e-08, "loss": 0.0188, "step": 49808 }, { "epoch": 2.655488617582769, "grad_norm": 4.024927139282227, "learning_rate": 3.8400000000000006e-08, "loss": 0.0204, "step": 49809 }, { "epoch": 2.655541931012422, "grad_norm": 2.0872535705566406, "learning_rate": 3.8200000000000005e-08, "loss": 0.0156, "step": 49810 }, { "epoch": 2.655595244442075, "grad_norm": 2.3055260181427, "learning_rate": 3.8e-08, "loss": 0.0276, "step": 49811 }, { "epoch": 2.6556485578717277, "grad_norm": 2.4286704063415527, "learning_rate": 3.78e-08, "loss": 0.0069, "step": 49812 }, { "epoch": 2.655701871301381, "grad_norm": 2.976705312728882, "learning_rate": 3.76e-08, "loss": 0.0186, "step": 49813 }, { "epoch": 2.6557551847310337, "grad_norm": 2.185342788696289, "learning_rate": 3.7400000000000004e-08, "loss": 0.0592, "step": 49814 }, { "epoch": 2.6558084981606864, "grad_norm": 0.8381252884864807, "learning_rate": 3.72e-08, "loss": 0.0033, "step": 49815 }, { "epoch": 2.6558618115903396, "grad_norm": 3.1191132068634033, "learning_rate": 3.700000000000001e-08, "loss": 0.0256, "step": 49816 }, { "epoch": 2.6559151250199924, "grad_norm": 1.7986748218536377, "learning_rate": 3.6800000000000005e-08, "loss": 0.0078, "step": 49817 }, { "epoch": 2.6559684384496456, "grad_norm": 2.635158061981201, "learning_rate": 3.6600000000000003e-08, "loss": 0.0044, "step": 49818 }, { "epoch": 2.6560217518792983, "grad_norm": 4.698808193206787, "learning_rate": 3.64e-08, "loss": 0.0331, "step": 49819 }, { "epoch": 2.6560750653089515, "grad_norm": 1.8742696046829224, "learning_rate": 3.62e-08, "loss": 0.0081, "step": 49820 }, { "epoch": 2.6561283787386043, "grad_norm": 3.026315450668335, "learning_rate": 3.6000000000000005e-08, "loss": 0.0346, "step": 49821 }, { "epoch": 2.656181692168257, "grad_norm": 2.8685858249664307, "learning_rate": 3.58e-08, "loss": 0.0188, "step": 49822 }, { "epoch": 2.6562350055979103, "grad_norm": 2.907221555709839, "learning_rate": 3.56e-08, "loss": 0.0235, "step": 49823 }, { "epoch": 2.656288319027563, "grad_norm": 1.7816622257232666, "learning_rate": 3.5400000000000006e-08, "loss": 0.0183, "step": 49824 }, { "epoch": 2.6563416324572158, "grad_norm": 1.1143028736114502, "learning_rate": 3.5200000000000004e-08, "loss": 0.0207, "step": 49825 }, { "epoch": 2.656394945886869, "grad_norm": 2.0587310791015625, "learning_rate": 3.5e-08, "loss": 0.0228, "step": 49826 }, { "epoch": 2.6564482593165217, "grad_norm": 1.6393659114837646, "learning_rate": 3.48e-08, "loss": 0.0278, "step": 49827 }, { "epoch": 2.6565015727461745, "grad_norm": 1.92619788646698, "learning_rate": 3.4600000000000005e-08, "loss": 0.0097, "step": 49828 }, { "epoch": 2.6565548861758277, "grad_norm": 2.544741630554199, "learning_rate": 3.4400000000000004e-08, "loss": 0.0274, "step": 49829 }, { "epoch": 2.656608199605481, "grad_norm": 3.154496908187866, "learning_rate": 3.42e-08, "loss": 0.0174, "step": 49830 }, { "epoch": 2.6566615130351336, "grad_norm": 2.0963377952575684, "learning_rate": 3.4e-08, "loss": 0.0098, "step": 49831 }, { "epoch": 2.6567148264647864, "grad_norm": 0.43025752902030945, "learning_rate": 3.3800000000000005e-08, "loss": 0.0027, "step": 49832 }, { "epoch": 2.6567681398944396, "grad_norm": 1.9318864345550537, "learning_rate": 3.36e-08, "loss": 0.0146, "step": 49833 }, { "epoch": 2.6568214533240924, "grad_norm": 1.5796294212341309, "learning_rate": 3.34e-08, "loss": 0.0185, "step": 49834 }, { "epoch": 2.656874766753745, "grad_norm": 0.9803500175476074, "learning_rate": 3.3200000000000006e-08, "loss": 0.0048, "step": 49835 }, { "epoch": 2.6569280801833983, "grad_norm": 2.1961498260498047, "learning_rate": 3.3000000000000004e-08, "loss": 0.0185, "step": 49836 }, { "epoch": 2.656981393613051, "grad_norm": 1.269211769104004, "learning_rate": 3.28e-08, "loss": 0.0086, "step": 49837 }, { "epoch": 2.657034707042704, "grad_norm": 1.7973229885101318, "learning_rate": 3.26e-08, "loss": 0.0101, "step": 49838 }, { "epoch": 2.657088020472357, "grad_norm": 2.2026915550231934, "learning_rate": 3.24e-08, "loss": 0.0353, "step": 49839 }, { "epoch": 2.65714133390201, "grad_norm": 0.6739957928657532, "learning_rate": 3.2200000000000004e-08, "loss": 0.0031, "step": 49840 }, { "epoch": 2.657194647331663, "grad_norm": 0.34431731700897217, "learning_rate": 3.2e-08, "loss": 0.0021, "step": 49841 }, { "epoch": 2.6572479607613158, "grad_norm": 1.9353429079055786, "learning_rate": 3.180000000000001e-08, "loss": 0.0279, "step": 49842 }, { "epoch": 2.657301274190969, "grad_norm": 0.3830093741416931, "learning_rate": 3.1600000000000005e-08, "loss": 0.0031, "step": 49843 }, { "epoch": 2.6573545876206217, "grad_norm": 1.792048692703247, "learning_rate": 3.14e-08, "loss": 0.0091, "step": 49844 }, { "epoch": 2.6574079010502745, "grad_norm": 2.464871406555176, "learning_rate": 3.12e-08, "loss": 0.0132, "step": 49845 }, { "epoch": 2.6574612144799277, "grad_norm": 2.2712273597717285, "learning_rate": 3.1e-08, "loss": 0.0169, "step": 49846 }, { "epoch": 2.6575145279095804, "grad_norm": 2.7853329181671143, "learning_rate": 3.08e-08, "loss": 0.0239, "step": 49847 }, { "epoch": 2.657567841339233, "grad_norm": 0.9723155498504639, "learning_rate": 3.06e-08, "loss": 0.021, "step": 49848 }, { "epoch": 2.6576211547688864, "grad_norm": 1.5486345291137695, "learning_rate": 3.04e-08, "loss": 0.0085, "step": 49849 }, { "epoch": 2.657674468198539, "grad_norm": 2.3137476444244385, "learning_rate": 3.0200000000000006e-08, "loss": 0.0112, "step": 49850 }, { "epoch": 2.657727781628192, "grad_norm": 1.8280106782913208, "learning_rate": 3.0000000000000004e-08, "loss": 0.0126, "step": 49851 }, { "epoch": 2.657781095057845, "grad_norm": 1.8764152526855469, "learning_rate": 2.9800000000000002e-08, "loss": 0.0268, "step": 49852 }, { "epoch": 2.657834408487498, "grad_norm": 2.9710934162139893, "learning_rate": 2.96e-08, "loss": 0.0162, "step": 49853 }, { "epoch": 2.657887721917151, "grad_norm": 3.8870127201080322, "learning_rate": 2.9400000000000002e-08, "loss": 0.032, "step": 49854 }, { "epoch": 2.657941035346804, "grad_norm": 3.0926718711853027, "learning_rate": 2.92e-08, "loss": 0.0402, "step": 49855 }, { "epoch": 2.657994348776457, "grad_norm": 1.2606881856918335, "learning_rate": 2.9e-08, "loss": 0.0174, "step": 49856 }, { "epoch": 2.6580476622061098, "grad_norm": 0.9994840621948242, "learning_rate": 2.8800000000000003e-08, "loss": 0.0041, "step": 49857 }, { "epoch": 2.6581009756357625, "grad_norm": 2.219310998916626, "learning_rate": 2.8600000000000005e-08, "loss": 0.0105, "step": 49858 }, { "epoch": 2.6581542890654157, "grad_norm": 1.7617671489715576, "learning_rate": 2.8400000000000003e-08, "loss": 0.0065, "step": 49859 }, { "epoch": 2.6582076024950685, "grad_norm": 2.732056140899658, "learning_rate": 2.82e-08, "loss": 0.0106, "step": 49860 }, { "epoch": 2.6582609159247212, "grad_norm": 1.9339998960494995, "learning_rate": 2.8000000000000003e-08, "loss": 0.0038, "step": 49861 }, { "epoch": 2.6583142293543744, "grad_norm": 1.8481144905090332, "learning_rate": 2.78e-08, "loss": 0.0133, "step": 49862 }, { "epoch": 2.658367542784027, "grad_norm": 2.6768341064453125, "learning_rate": 2.7600000000000002e-08, "loss": 0.0423, "step": 49863 }, { "epoch": 2.65842085621368, "grad_norm": 1.3810579776763916, "learning_rate": 2.74e-08, "loss": 0.0117, "step": 49864 }, { "epoch": 2.658474169643333, "grad_norm": 0.7379642724990845, "learning_rate": 2.7200000000000005e-08, "loss": 0.0032, "step": 49865 }, { "epoch": 2.658527483072986, "grad_norm": 0.8965467810630798, "learning_rate": 2.7000000000000004e-08, "loss": 0.0053, "step": 49866 }, { "epoch": 2.658580796502639, "grad_norm": 3.590054988861084, "learning_rate": 2.6800000000000002e-08, "loss": 0.0099, "step": 49867 }, { "epoch": 2.658634109932292, "grad_norm": 3.8694679737091064, "learning_rate": 2.6600000000000003e-08, "loss": 0.0119, "step": 49868 }, { "epoch": 2.658687423361945, "grad_norm": 1.756988286972046, "learning_rate": 2.64e-08, "loss": 0.0135, "step": 49869 }, { "epoch": 2.658740736791598, "grad_norm": 0.9513427019119263, "learning_rate": 2.6200000000000003e-08, "loss": 0.0128, "step": 49870 }, { "epoch": 2.6587940502212506, "grad_norm": 2.697439670562744, "learning_rate": 2.6e-08, "loss": 0.0093, "step": 49871 }, { "epoch": 2.658847363650904, "grad_norm": 1.7905967235565186, "learning_rate": 2.58e-08, "loss": 0.0307, "step": 49872 }, { "epoch": 2.6589006770805566, "grad_norm": 1.8912932872772217, "learning_rate": 2.5600000000000004e-08, "loss": 0.0264, "step": 49873 }, { "epoch": 2.6589539905102093, "grad_norm": 2.6295273303985596, "learning_rate": 2.5400000000000002e-08, "loss": 0.016, "step": 49874 }, { "epoch": 2.6590073039398625, "grad_norm": 1.7677489519119263, "learning_rate": 2.5200000000000004e-08, "loss": 0.024, "step": 49875 }, { "epoch": 2.6590606173695153, "grad_norm": 2.2519354820251465, "learning_rate": 2.5000000000000002e-08, "loss": 0.0227, "step": 49876 }, { "epoch": 2.6591139307991685, "grad_norm": 3.279651403427124, "learning_rate": 2.48e-08, "loss": 0.0135, "step": 49877 }, { "epoch": 2.6591672442288212, "grad_norm": 1.3052880764007568, "learning_rate": 2.4600000000000002e-08, "loss": 0.0048, "step": 49878 }, { "epoch": 2.6592205576584744, "grad_norm": 1.0817312002182007, "learning_rate": 2.44e-08, "loss": 0.0159, "step": 49879 }, { "epoch": 2.659273871088127, "grad_norm": 1.6758629083633423, "learning_rate": 2.42e-08, "loss": 0.0077, "step": 49880 }, { "epoch": 2.65932718451778, "grad_norm": 2.1148321628570557, "learning_rate": 2.4e-08, "loss": 0.0198, "step": 49881 }, { "epoch": 2.659380497947433, "grad_norm": 1.5803906917572021, "learning_rate": 2.3800000000000005e-08, "loss": 0.0073, "step": 49882 }, { "epoch": 2.659433811377086, "grad_norm": 2.586437225341797, "learning_rate": 2.3600000000000003e-08, "loss": 0.011, "step": 49883 }, { "epoch": 2.6594871248067387, "grad_norm": 2.28393292427063, "learning_rate": 2.34e-08, "loss": 0.0184, "step": 49884 }, { "epoch": 2.659540438236392, "grad_norm": 1.8609453439712524, "learning_rate": 2.3200000000000003e-08, "loss": 0.0145, "step": 49885 }, { "epoch": 2.6595937516660446, "grad_norm": 2.922792911529541, "learning_rate": 2.3e-08, "loss": 0.0233, "step": 49886 }, { "epoch": 2.6596470650956974, "grad_norm": 1.2933813333511353, "learning_rate": 2.2800000000000002e-08, "loss": 0.0045, "step": 49887 }, { "epoch": 2.6597003785253506, "grad_norm": 3.5854222774505615, "learning_rate": 2.26e-08, "loss": 0.0367, "step": 49888 }, { "epoch": 2.6597536919550033, "grad_norm": 2.2966835498809814, "learning_rate": 2.24e-08, "loss": 0.0091, "step": 49889 }, { "epoch": 2.6598070053846565, "grad_norm": 0.5832660794258118, "learning_rate": 2.2200000000000004e-08, "loss": 0.0039, "step": 49890 }, { "epoch": 2.6598603188143093, "grad_norm": 2.4619429111480713, "learning_rate": 2.2000000000000002e-08, "loss": 0.0151, "step": 49891 }, { "epoch": 2.6599136322439625, "grad_norm": 3.38191819190979, "learning_rate": 2.1800000000000003e-08, "loss": 0.0161, "step": 49892 }, { "epoch": 2.6599669456736152, "grad_norm": 2.3234148025512695, "learning_rate": 2.16e-08, "loss": 0.0399, "step": 49893 }, { "epoch": 2.660020259103268, "grad_norm": 0.737583339214325, "learning_rate": 2.1400000000000003e-08, "loss": 0.0042, "step": 49894 }, { "epoch": 2.660073572532921, "grad_norm": 2.0563132762908936, "learning_rate": 2.12e-08, "loss": 0.0106, "step": 49895 }, { "epoch": 2.660126885962574, "grad_norm": 1.7886792421340942, "learning_rate": 2.1e-08, "loss": 0.0164, "step": 49896 }, { "epoch": 2.6601801993922267, "grad_norm": 0.32860901951789856, "learning_rate": 2.08e-08, "loss": 0.0027, "step": 49897 }, { "epoch": 2.66023351282188, "grad_norm": 0.9809861779212952, "learning_rate": 2.0600000000000002e-08, "loss": 0.0024, "step": 49898 }, { "epoch": 2.6602868262515327, "grad_norm": 1.5803278684616089, "learning_rate": 2.0400000000000004e-08, "loss": 0.0092, "step": 49899 }, { "epoch": 2.6603401396811854, "grad_norm": 2.977670907974243, "learning_rate": 2.0200000000000002e-08, "loss": 0.0319, "step": 49900 }, { "epoch": 2.6603934531108386, "grad_norm": 2.073704719543457, "learning_rate": 2e-08, "loss": 0.0097, "step": 49901 }, { "epoch": 2.6604467665404914, "grad_norm": 3.261608600616455, "learning_rate": 1.9800000000000002e-08, "loss": 0.0115, "step": 49902 }, { "epoch": 2.6605000799701446, "grad_norm": 0.2064923644065857, "learning_rate": 1.96e-08, "loss": 0.0015, "step": 49903 }, { "epoch": 2.6605533933997974, "grad_norm": 2.931612730026245, "learning_rate": 1.9400000000000002e-08, "loss": 0.0145, "step": 49904 }, { "epoch": 2.6606067068294506, "grad_norm": 2.1603622436523438, "learning_rate": 1.9200000000000003e-08, "loss": 0.0198, "step": 49905 }, { "epoch": 2.6606600202591033, "grad_norm": 1.0563260316848755, "learning_rate": 1.9e-08, "loss": 0.0151, "step": 49906 }, { "epoch": 2.660713333688756, "grad_norm": 3.51300048828125, "learning_rate": 1.88e-08, "loss": 0.0301, "step": 49907 }, { "epoch": 2.6607666471184093, "grad_norm": 2.18955135345459, "learning_rate": 1.86e-08, "loss": 0.0301, "step": 49908 }, { "epoch": 2.660819960548062, "grad_norm": 0.6479572057723999, "learning_rate": 1.8400000000000003e-08, "loss": 0.0034, "step": 49909 }, { "epoch": 2.660873273977715, "grad_norm": 2.517388105392456, "learning_rate": 1.82e-08, "loss": 0.0156, "step": 49910 }, { "epoch": 2.660926587407368, "grad_norm": 1.3060590028762817, "learning_rate": 1.8000000000000002e-08, "loss": 0.0102, "step": 49911 }, { "epoch": 2.6609799008370207, "grad_norm": 1.3195825815200806, "learning_rate": 1.78e-08, "loss": 0.0204, "step": 49912 }, { "epoch": 2.661033214266674, "grad_norm": 1.5820525884628296, "learning_rate": 1.7600000000000002e-08, "loss": 0.0067, "step": 49913 }, { "epoch": 2.6610865276963267, "grad_norm": 4.559638500213623, "learning_rate": 1.74e-08, "loss": 0.0151, "step": 49914 }, { "epoch": 2.66113984112598, "grad_norm": 1.4692270755767822, "learning_rate": 1.7200000000000002e-08, "loss": 0.008, "step": 49915 }, { "epoch": 2.6611931545556327, "grad_norm": 2.620957851409912, "learning_rate": 1.7e-08, "loss": 0.0283, "step": 49916 }, { "epoch": 2.6612464679852854, "grad_norm": 2.1340904235839844, "learning_rate": 1.68e-08, "loss": 0.0091, "step": 49917 }, { "epoch": 2.6612997814149386, "grad_norm": 4.470603942871094, "learning_rate": 1.6600000000000003e-08, "loss": 0.0237, "step": 49918 }, { "epoch": 2.6613530948445914, "grad_norm": 1.2994343042373657, "learning_rate": 1.64e-08, "loss": 0.0248, "step": 49919 }, { "epoch": 2.661406408274244, "grad_norm": 2.8400769233703613, "learning_rate": 1.62e-08, "loss": 0.031, "step": 49920 }, { "epoch": 2.6614597217038973, "grad_norm": 2.961599349975586, "learning_rate": 1.6e-08, "loss": 0.029, "step": 49921 }, { "epoch": 2.66151303513355, "grad_norm": 4.761732578277588, "learning_rate": 1.5800000000000003e-08, "loss": 0.0285, "step": 49922 }, { "epoch": 2.661566348563203, "grad_norm": 1.2244739532470703, "learning_rate": 1.56e-08, "loss": 0.0156, "step": 49923 }, { "epoch": 2.661619661992856, "grad_norm": 2.778346538543701, "learning_rate": 1.54e-08, "loss": 0.0353, "step": 49924 }, { "epoch": 2.661672975422509, "grad_norm": 2.030794858932495, "learning_rate": 1.52e-08, "loss": 0.0221, "step": 49925 }, { "epoch": 2.661726288852162, "grad_norm": 2.947838068008423, "learning_rate": 1.5000000000000002e-08, "loss": 0.0281, "step": 49926 }, { "epoch": 2.6617796022818148, "grad_norm": 0.35921424627304077, "learning_rate": 1.48e-08, "loss": 0.0023, "step": 49927 }, { "epoch": 2.661832915711468, "grad_norm": 1.7172385454177856, "learning_rate": 1.46e-08, "loss": 0.0325, "step": 49928 }, { "epoch": 2.6618862291411207, "grad_norm": 1.6730631589889526, "learning_rate": 1.4400000000000002e-08, "loss": 0.0044, "step": 49929 }, { "epoch": 2.6619395425707735, "grad_norm": 0.5593414306640625, "learning_rate": 1.4200000000000001e-08, "loss": 0.0029, "step": 49930 }, { "epoch": 2.6619928560004267, "grad_norm": 1.829106330871582, "learning_rate": 1.4000000000000001e-08, "loss": 0.0195, "step": 49931 }, { "epoch": 2.6620461694300794, "grad_norm": 3.9283385276794434, "learning_rate": 1.3800000000000001e-08, "loss": 0.0115, "step": 49932 }, { "epoch": 2.662099482859732, "grad_norm": 2.079293727874756, "learning_rate": 1.3600000000000003e-08, "loss": 0.0094, "step": 49933 }, { "epoch": 2.6621527962893854, "grad_norm": 0.982408881187439, "learning_rate": 1.3400000000000001e-08, "loss": 0.004, "step": 49934 }, { "epoch": 2.662206109719038, "grad_norm": 1.8881973028182983, "learning_rate": 1.32e-08, "loss": 0.0135, "step": 49935 }, { "epoch": 2.662259423148691, "grad_norm": 0.6026322245597839, "learning_rate": 1.3e-08, "loss": 0.0031, "step": 49936 }, { "epoch": 2.662312736578344, "grad_norm": 1.2003090381622314, "learning_rate": 1.2800000000000002e-08, "loss": 0.0148, "step": 49937 }, { "epoch": 2.662366050007997, "grad_norm": 1.0409928560256958, "learning_rate": 1.2600000000000002e-08, "loss": 0.0043, "step": 49938 }, { "epoch": 2.66241936343765, "grad_norm": 2.7160987854003906, "learning_rate": 1.24e-08, "loss": 0.0314, "step": 49939 }, { "epoch": 2.662472676867303, "grad_norm": 0.34180551767349243, "learning_rate": 1.22e-08, "loss": 0.0022, "step": 49940 }, { "epoch": 2.662525990296956, "grad_norm": 3.200100898742676, "learning_rate": 1.2e-08, "loss": 0.0117, "step": 49941 }, { "epoch": 2.662579303726609, "grad_norm": 0.20015950500965118, "learning_rate": 1.1800000000000001e-08, "loss": 0.0014, "step": 49942 }, { "epoch": 2.6626326171562615, "grad_norm": 11.648344039916992, "learning_rate": 1.1600000000000001e-08, "loss": 0.0524, "step": 49943 }, { "epoch": 2.6626859305859147, "grad_norm": 2.282156467437744, "learning_rate": 1.1400000000000001e-08, "loss": 0.0106, "step": 49944 }, { "epoch": 2.6627392440155675, "grad_norm": 1.304456353187561, "learning_rate": 1.12e-08, "loss": 0.0053, "step": 49945 }, { "epoch": 2.6627925574452203, "grad_norm": 2.916203260421753, "learning_rate": 1.1000000000000001e-08, "loss": 0.0152, "step": 49946 }, { "epoch": 2.6628458708748735, "grad_norm": 2.373947858810425, "learning_rate": 1.08e-08, "loss": 0.0113, "step": 49947 }, { "epoch": 2.662899184304526, "grad_norm": 1.2057745456695557, "learning_rate": 1.06e-08, "loss": 0.0033, "step": 49948 }, { "epoch": 2.662952497734179, "grad_norm": 3.0297493934631348, "learning_rate": 1.04e-08, "loss": 0.0169, "step": 49949 }, { "epoch": 2.663005811163832, "grad_norm": 1.3801378011703491, "learning_rate": 1.0200000000000002e-08, "loss": 0.0086, "step": 49950 }, { "epoch": 2.6630591245934854, "grad_norm": 0.7373265027999878, "learning_rate": 1e-08, "loss": 0.004, "step": 49951 }, { "epoch": 2.663112438023138, "grad_norm": 2.9034955501556396, "learning_rate": 9.8e-09, "loss": 0.0184, "step": 49952 }, { "epoch": 2.663165751452791, "grad_norm": 0.6398720145225525, "learning_rate": 9.600000000000002e-09, "loss": 0.003, "step": 49953 }, { "epoch": 2.663219064882444, "grad_norm": 2.042240858078003, "learning_rate": 9.4e-09, "loss": 0.0484, "step": 49954 }, { "epoch": 2.663272378312097, "grad_norm": 1.6410157680511475, "learning_rate": 9.200000000000001e-09, "loss": 0.0177, "step": 49955 }, { "epoch": 2.6633256917417496, "grad_norm": 2.8226826190948486, "learning_rate": 9.000000000000001e-09, "loss": 0.0151, "step": 49956 }, { "epoch": 2.663379005171403, "grad_norm": 0.6560187935829163, "learning_rate": 8.800000000000001e-09, "loss": 0.0033, "step": 49957 }, { "epoch": 2.6634323186010556, "grad_norm": 2.3083746433258057, "learning_rate": 8.600000000000001e-09, "loss": 0.01, "step": 49958 }, { "epoch": 2.6634856320307083, "grad_norm": 2.869934320449829, "learning_rate": 8.4e-09, "loss": 0.0121, "step": 49959 }, { "epoch": 2.6635389454603615, "grad_norm": 1.8125190734863281, "learning_rate": 8.2e-09, "loss": 0.0169, "step": 49960 }, { "epoch": 2.6635922588900143, "grad_norm": 3.837536573410034, "learning_rate": 8e-09, "loss": 0.0498, "step": 49961 }, { "epoch": 2.6636455723196675, "grad_norm": 2.1472620964050293, "learning_rate": 7.8e-09, "loss": 0.0282, "step": 49962 }, { "epoch": 2.6636988857493202, "grad_norm": 4.022251129150391, "learning_rate": 7.6e-09, "loss": 0.0111, "step": 49963 }, { "epoch": 2.6637521991789734, "grad_norm": 2.0808160305023193, "learning_rate": 7.4e-09, "loss": 0.0103, "step": 49964 }, { "epoch": 2.663805512608626, "grad_norm": 2.4892823696136475, "learning_rate": 7.200000000000001e-09, "loss": 0.0301, "step": 49965 }, { "epoch": 2.663858826038279, "grad_norm": 1.910728931427002, "learning_rate": 7.000000000000001e-09, "loss": 0.0182, "step": 49966 }, { "epoch": 2.663912139467932, "grad_norm": 2.2611892223358154, "learning_rate": 6.800000000000001e-09, "loss": 0.006, "step": 49967 }, { "epoch": 2.663965452897585, "grad_norm": 1.1347203254699707, "learning_rate": 6.6e-09, "loss": 0.0058, "step": 49968 }, { "epoch": 2.6640187663272377, "grad_norm": 1.4754565954208374, "learning_rate": 6.400000000000001e-09, "loss": 0.038, "step": 49969 }, { "epoch": 2.664072079756891, "grad_norm": 2.4157347679138184, "learning_rate": 6.2e-09, "loss": 0.0112, "step": 49970 }, { "epoch": 2.6641253931865436, "grad_norm": 1.7981213331222534, "learning_rate": 6e-09, "loss": 0.0063, "step": 49971 }, { "epoch": 2.6641787066161964, "grad_norm": 1.5078598260879517, "learning_rate": 5.800000000000001e-09, "loss": 0.021, "step": 49972 }, { "epoch": 2.6642320200458496, "grad_norm": 3.4331016540527344, "learning_rate": 5.6e-09, "loss": 0.0124, "step": 49973 }, { "epoch": 2.6642853334755023, "grad_norm": 0.6492955088615417, "learning_rate": 5.4e-09, "loss": 0.0034, "step": 49974 }, { "epoch": 2.6643386469051555, "grad_norm": 0.9699780941009521, "learning_rate": 5.2e-09, "loss": 0.0059, "step": 49975 }, { "epoch": 2.6643919603348083, "grad_norm": 2.005124092102051, "learning_rate": 5e-09, "loss": 0.0097, "step": 49976 }, { "epoch": 2.6644452737644615, "grad_norm": 1.742114782333374, "learning_rate": 4.800000000000001e-09, "loss": 0.0174, "step": 49977 }, { "epoch": 2.6644985871941143, "grad_norm": 1.6519957780838013, "learning_rate": 4.600000000000001e-09, "loss": 0.0329, "step": 49978 }, { "epoch": 2.664551900623767, "grad_norm": 3.6024391651153564, "learning_rate": 4.4000000000000005e-09, "loss": 0.0176, "step": 49979 }, { "epoch": 2.66460521405342, "grad_norm": 1.5352458953857422, "learning_rate": 4.2e-09, "loss": 0.0082, "step": 49980 }, { "epoch": 2.664658527483073, "grad_norm": 1.374563455581665, "learning_rate": 4e-09, "loss": 0.0054, "step": 49981 }, { "epoch": 2.6647118409127257, "grad_norm": 3.126561403274536, "learning_rate": 3.8e-09, "loss": 0.0189, "step": 49982 }, { "epoch": 2.664765154342379, "grad_norm": 1.26680326461792, "learning_rate": 3.6000000000000004e-09, "loss": 0.0038, "step": 49983 }, { "epoch": 2.6648184677720317, "grad_norm": 1.178152084350586, "learning_rate": 3.4000000000000007e-09, "loss": 0.0162, "step": 49984 }, { "epoch": 2.6648717812016844, "grad_norm": 2.168060302734375, "learning_rate": 3.2000000000000005e-09, "loss": 0.019, "step": 49985 }, { "epoch": 2.6649250946313376, "grad_norm": 1.6592470407485962, "learning_rate": 3e-09, "loss": 0.0069, "step": 49986 }, { "epoch": 2.6649784080609904, "grad_norm": 2.2173938751220703, "learning_rate": 2.8e-09, "loss": 0.0149, "step": 49987 }, { "epoch": 2.6650317214906436, "grad_norm": 2.4177839756011963, "learning_rate": 2.6e-09, "loss": 0.0083, "step": 49988 }, { "epoch": 2.6650850349202964, "grad_norm": 2.0348262786865234, "learning_rate": 2.4000000000000004e-09, "loss": 0.0137, "step": 49989 }, { "epoch": 2.6651383483499496, "grad_norm": 2.7200472354888916, "learning_rate": 2.2000000000000003e-09, "loss": 0.0269, "step": 49990 }, { "epoch": 2.6651916617796023, "grad_norm": 2.27258563041687, "learning_rate": 2e-09, "loss": 0.0133, "step": 49991 }, { "epoch": 2.665244975209255, "grad_norm": 0.7623713612556458, "learning_rate": 1.8000000000000002e-09, "loss": 0.003, "step": 49992 }, { "epoch": 2.6652982886389083, "grad_norm": 2.5488498210906982, "learning_rate": 1.6000000000000003e-09, "loss": 0.0138, "step": 49993 }, { "epoch": 2.665351602068561, "grad_norm": 1.551347255706787, "learning_rate": 1.4e-09, "loss": 0.0373, "step": 49994 }, { "epoch": 2.665404915498214, "grad_norm": 3.035823345184326, "learning_rate": 1.2000000000000002e-09, "loss": 0.0224, "step": 49995 }, { "epoch": 2.665458228927867, "grad_norm": 1.1327728033065796, "learning_rate": 1e-09, "loss": 0.0105, "step": 49996 }, { "epoch": 2.6655115423575197, "grad_norm": 3.155060291290283, "learning_rate": 8.000000000000001e-10, "loss": 0.0139, "step": 49997 }, { "epoch": 2.665564855787173, "grad_norm": 1.6610770225524902, "learning_rate": 6.000000000000001e-10, "loss": 0.0065, "step": 49998 }, { "epoch": 2.6656181692168257, "grad_norm": 0.5912570357322693, "learning_rate": 4.0000000000000007e-10, "loss": 0.0029, "step": 49999 }, { "epoch": 2.665671482646479, "grad_norm": 2.506667137145996, "learning_rate": 2.0000000000000003e-10, "loss": 0.0436, "step": 50000 } ], "logging_steps": 1.0, "max_steps": 50000, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 2000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 1.6317608074194348e+20, "train_batch_size": 8, "trial_name": null, "trial_params": null }